# HARNESSING THE PARTICIPATION OF DENDRITIC CELLS IN IMMUNITY AND TOLERANCE

EDITED BY : Silvia Beatriz Boscardin, Diana Dudziak, Daniela Santoro Rosa and Christian Muenz PUBLISHED IN : Frontiers in Immunology

#### Frontiers eBook Copyright Statement

The copyright in the text of individual articles in this eBook is the property of their respective authors or their respective institutions or funders. The copyright in graphics and images within each article may be subject to copyright of other parties. In both cases this is subject to a license granted to Frontiers. The compilation of articles constituting this eBook is the property of Frontiers.

Each article within this eBook, and the eBook itself, are published under the most recent version of the Creative Commons CC-BY licence. The version current at the date of publication of this eBook is CC-BY 4.0. If the CC-BY licence is updated, the licence granted by Frontiers is automatically updated to the new version.

When exercising any right under the CC-BY licence, Frontiers must be attributed as the original publisher of the article or eBook, as applicable.

Authors have the responsibility of ensuring that any graphics or other materials which are the property of others may be included in the CC-BY licence, but this should be checked before relying on the CC-BY licence to reproduce those materials. Any copyright notices relating to those materials must be complied with.

Copyright and source acknowledgement notices may not be removed and must be displayed in any copy, derivative work or partial copy which includes the elements in question.

All copyright, and all rights therein, are protected by national and international copyright laws. The above represents a summary only. For further information please read Frontiers' Conditions for Website Use and Copyright Statement, and the applicable CC-BY licence.

ISSN 1664-8714 ISBN 978-2-88966-201-2 DOI 10.3389/978-2-88966-201-2

#### About Frontiers

Frontiers is more than just an open-access publisher of scholarly articles: it is a pioneering approach to the world of academia, radically improving the way scholarly research is managed. The grand vision of Frontiers is a world where all people have an equal opportunity to seek, share and generate knowledge. Frontiers provides immediate and permanent online open access to all its publications, but this alone is not enough to realize our grand goals.

#### Frontiers Journal Series

The Frontiers Journal Series is a multi-tier and interdisciplinary set of open-access, online journals, promising a paradigm shift from the current review, selection and dissemination processes in academic publishing. All Frontiers journals are driven by researchers for researchers; therefore, they constitute a service to the scholarly community. At the same time, the Frontiers Journal Series operates on a revolutionary invention, the tiered publishing system, initially addressing specific communities of scholars, and gradually climbing up to broader public understanding, thus serving the interests of the lay society, too.

#### Dedication to Quality

Each Frontiers article is a landmark of the highest quality, thanks to genuinely collaborative interactions between authors and review editors, who include some of the world's best academicians. Research must be certified by peers before entering a stream of knowledge that may eventually reach the public - and shape society; therefore, Frontiers only applies the most rigorous and unbiased reviews.

Frontiers revolutionizes research publishing by freely delivering the most outstanding research, evaluated with no bias from both the academic and social point of view. By applying the most advanced information technologies, Frontiers is catapulting scholarly publishing into a new generation.

#### What are Frontiers Research Topics?

Frontiers Research Topics are very popular trademarks of the Frontiers Journals Series: they are collections of at least ten articles, all centered on a particular subject. With their unique mix of varied contributions from Original Research to Review Articles, Frontiers Research Topics unify the most influential researchers, the latest key findings and historical advances in a hot research area! Find out more on how to host your own Frontiers Research Topic or contribute to one as an author by contacting the Frontiers Editorial Office: researchtopics@frontiersin.org

# HARNESSING THE PARTICIPATION OF DENDRITIC CELLS IN IMMUNITY AND TOLERANCE

Topic Editors:

Silvia Beatriz Boscardin, University of São Paulo, Brazil Diana Dudziak, Universitatsklinikum Erlangen, Germany Daniela Santoro Rosa, Federal University of São Paulo, Brazil Christian Muenz, University of Zurich, Switzerland

Citation: Boscardin, S. B., Dudziak, D., Rosa, D. S., Muenz, C., eds. (2020). Harnessing the Participation of Dendritic Cells in Immunity and Tolerance. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88966-201-2

# Table of Contents

*07 Editorial: Harnessing the Participation of Dendritic Cells in Immunity and Tolerance*

Silvia Beatriz Boscardin, Diana Dudziak, Christian Münz and Daniela Santoro Rosa

*11 NH36 and F3 Antigen-Primed Dendritic Cells Show Preserved Migrating Capabilities and CCR7 Expression and F3 is Effective in Immunotherapy of Visceral Leishmaniasis*

Dirlei Nico, Fernanda Martins Almeida, Juliana Maria Motta, Fellipe Soares dos Santos Cardoso, Celio Geraldo Freire-de-Lima, Leonardo Freire-de-Lima, Paula Melo de Luca, Ana Maria Blanco Martinez, Alexandre Morrot and Clarisa Beatriz Palatnik-de-Sousa

*29 Langerin+ CD8*a*+ Dendritic Cells Drive Early CD8+ T Cell Activation and IL-12 Production During Systemic Bacterial Infection*

Kelly A. Prendergast, Naomi J. Daniels, Troels R. Petersen, Ian F. Hermans and Joanna R. Kirman

*39 Introduction of Human Flt3-L and GM-CSF into Humanized Mice Enhances the Reconstitution and Maturation of Myeloid Dendritic Cells and the Development of Foxp3+CD4+ T Cells*

Ryutaro Iwabuchi, Shota Ikeno, Mie Kobayashi-Ishihara, Haruko Takeyama, Manabu Ato, Yasuko Tsunetsugu-Yokota and Kazutaka Terahara

*57 Dendritic Cells Actively Limit Interleukin-10 Production Under Inflammatory Conditions* via *DC-SCRIPT and Dual-Specificity Phosphatase 4*

Jonas Nørskov Søndergaard, Simon J. van Heeringen, Maaike W. G. Looman, Chunling Tang, Vassilis Triantis, Pauline Louche, Eva M. Janssen-Megens, Anieta M. Sieuwerts, John W. M. Martens, Colin Logie, Hendrik G. Stunnenberg, Marleen Ansems and Gosse J. Adema


Laura Antonio-Herrera, Oscar Badillo-Godinez, Oscar Medina-Contreras, Araceli Tepale-Segura, Alberto García-Lozano, Lourdes Gutierrez-Xicotencatl, Gloria Soldevila, Fernando R. Esquivel-Guadarrama, Juliana Idoyaga and Laura C. Bonifaz

*110 Intestinal CD103+CD11b+ cDC2 Conventional Dendritic Cells are Required for Primary CD4+ T and B Cell Responses to Soluble Flagellin* Adriana Flores-Langarica, Charlotte Cook, Katarzyna Müller Luda,

Emma K. Persson, Jennifer L. Marshall, Nonantzin Beristain-Covarrubias,

Juan Carlos Yam-Puc, Madelene Dahlgren, Jenny J. Persson, Satoshi Uematsu, Shizuo Akira, Ian R. Henderson,

Bengt Johansson Lindbom, William Agace and Adam F. Cunningham


Raphaël Mattiuz, Christian Wohn, Sonia Ghilas, Marc Ambrosini, Yannick O. Alexandre, Cindy Sanchez, Anissa Fries, Thien-Phong Vu Manh, Bernard Malissen, Marc Dalod and Karine Crozat


Arthur L. Kroczek, Evelyn Hartung, Stephanie Gurka, Martina Becker, Nele Reeg, Hans W. Mages, Sebastian Voigt, Christian Freund and Richard A. Kroczek

*229 Using Dendritic Cell-Based Immunotherapy to Treat HIV: How Can This Strategy be Improved?*

Laís Teodoro da Silva, Bruna Tereso Santillo, Alexandre de Almeida, Alberto Jose da Silva Duarte and Telma Miyuki Oshiro


Sebastian Montealegre and Peter M. van Endert

*261 Understanding the Functional Properties of Neonatal Dendritic Cells: A Doorway to Enhance Vaccine Effectiveness?*

Nikos E. Papaioannou, Maria Pasztoi and Barbara U. Schraml

*269 Human Dendritic Cells: Their Heterogeneity and Clinical Application Potential in Cancer Immunotherapy*

Thiago A. Patente, Mariana P. Pinho, Aline A. Oliveira, Gabriela C. M. Evangelista, Patrícia C. Bergami-Santos and José A. M. Barbuto

#### *287 Generation of an Oncolytic Herpes Simplex Virus 1 Expressing Human MelanA*

Jan B. Boscheinen, Sabrina Thomann, David M. Knipe, Neal DeLuca, Beatrice Schuler-Thurner, Stefanie Gross, Jan Dörrie, Niels Schaft, Christian Bach, Anette Rohrhofer, Melanie Werner-Klein, Barbara Schmidt and Philipp Schuster


Jean-Charles Cancel, Karine Crozat, Marc Dalod and Raphaël Mattiuz

*360 Molecular Aspects of Dendritic Cell Activation in Leishmaniasis: An Immunobiological View*

Rafael Tibúrcio, Sara Nunes, Ivanéia Nunes, Mariana Rosa Ampuero, Icaro Bonyek Silva, Reinan Lima, Natalia Machado Tavares and Cláudia Brodskyn

*374 Neoadjuvant Radiochemotherapy Significantly Alters the Phenotype of Plasmacytoid Dendritic Cells and 6-Sulfo LacNAc+ Monocytes in Rectal Cancer*

Felix Wagner, Ulrike Hölig, Friederike Wilczkowski, Ioana Plesca, Ulrich Sommer, Rebekka Wehner, Maximilian Kießler, Armin Jarosch, Katharina Flecke, Maia Arsova, Antje Tunger, Andreas Bogner, Christoph Reißfelder, Jürgen Weitz, Knut Schäkel, Esther G. C. Troost, Mechthild Krause, Gunnar Folprecht, Martin Bornhäuser, Michael P. Bachmann, Daniela Aust, Gustavo Baretton and Marc Schmitz


Ronald A. Backer, Nathalie Diener and Björn E. Clausen

*440 Poly(I:C) Potentiates T Cell Immunity to a Dendritic Cell Targeted HIV-Multiepitope Vaccine*

Juliana de Souza Apostólico, Victória Alves Santos Lunardelli, Marcio Massao Yamamoto, Edecio Cunha-Neto, Silvia Beatriz Boscardin and Daniela Santoro Rosa


Manuela Schönfeld, Ulla Knackmuss, Parul Chandorkar, Paul Hörtnagl, Thomas John Hope, Arnaud Moris, Rosa Bellmann-Weiler, Cornelia Lass-Flörl, Wilfried Posch and Doris Wilflingseder

*490 What Makes a pDC: Recent Advances in Understanding Plasmacytoid DC Development and Heterogeneity*

Andrea Musumeci, Konstantin Lutz, Elena Winheim and Anne Barbara Krug

*497 Flagellin/NLRC4 Pathway Rescues NLRP3-Inflammasome Defect in Dendritic Cells From HIV-Infected Patients: Perspective for New Adjuvant in Immunocompromised Individuals*

Edione Cristina dos Reis, Vinícius Nunes Cordeiro Leal, Jaíne Lima da Silva Soares, Fernanda Pereira Fernandes, Dhêmerson Souza de Lima, Bruna Cunha de Alencar and Alessandra Pontillo

# Editorial: Harnessing the Participation of Dendritic Cells in Immunity and Tolerance

#### Silvia Beatriz Boscardin<sup>1</sup> \*, Diana Dudziak 2,3,4,5, Christian Münz <sup>6</sup> and Daniela Santoro Rosa<sup>7</sup>

<sup>1</sup> Department of Parasitology, Institute of Biomedical Sciences, University of São Paulo, São Paulo, Brazil, <sup>2</sup> Laboratory of Dendritic Cell Biology, Department of Dermatology, Friedrich-Alexander University of Erlangen-Nürnberg (FAU), University Hospital Erlangen, Erlangen, Germany, <sup>3</sup> Medical Immunology Campus Erlangen, Erlangen, Germany, <sup>4</sup> Deutsches Zentrum Immuntherapie (DZI), Erlangen, Germany, <sup>5</sup> Comprehensive Cancer Center Erlangen-European Metropolitan Area of Nuremberg (CCC ER-EMN), Erlangen, Germany, <sup>6</sup> Viral Immunobiology, Institute of Experimental Immunology, University of Zürich, Zurich, Switzerland, <sup>7</sup> Department of Microbiology, Immunology and Parasitology, Federal University of São Paulo, São Paulo, Brazil

#### Keywords: dendritic cell, DC subtypes, antigen presentation, dendritic cell vaccine therapy, dendritic cell activation

#### **Editorial on the Research Topic**

#### **Harnessing the Participation of Dendritic Cells in Immunity and Tolerance**

Edited and reviewed by: Florent Ginhoux,

Singapore Immunology Network (A∗STAR), Singapore

\*Correspondence: Silvia Beatriz Boscardin sbboscardin@usp.br

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 17 August 2020 Accepted: 31 August 2020 Published: 07 October 2020

#### Citation:

Boscardin SB, Dudziak D, Münz C and Rosa DS (2020) Editorial: Harnessing the Participation of Dendritic Cells in Immunity and Tolerance. Front. Immunol. 11:595841. doi: 10.3389/fimmu.2020.595841 Dendritic cells (DCs) are cells of the innate immune system directly associated with the instruction and regulation of the adaptive immune response, thus bridging the innate and adaptive immune systems (1). DCs are the main antigen presenting cells (APCs), specialized in naive T cell priming into effector T cells. DCs recognize, internalize, process, and present antigens complexed to class I and II major histocompatibility complex molecules (MHCs), providing the three signals necessary for efficient T cell activation (2). Thus, DCs are essential for the induction of immune responses mediated by CD4<sup>+</sup> and CD8<sup>+</sup> T cells and, directly or indirectly, by B cells (3, 4).

However, DCs are not a homogeneous cell line, on the contrary, there are several subtypes that can be classified according to membrane markers and/or function. Basically, they can be classified into plasmacytoid and classical/conventional DCs (5, 6). Plasmacytoid DCs (pDCs) are specialized in the expression of type I interferons (IFN), responding quickly and efficiently to viral infections (7). Recent advances in the knowledge of their ontogeny, both in humans and in mice, are carefully reviewed by Musumeci et al., while Ali et al. review in detail the role of type I IFN production by these cells, but also by other cell types. Recent data indicate that several other cell types play a prominent role in the production of type I IFN, depending on the pathogen causing infection. Classical DCs (cDCs), on the other hand, are mainly associated with antigen presentation and T cell instruction. Most often, cDCs are subdivided into different subtypes based on the expression of various surface markers. However, a new classification according to their ontogeny has been proposed, as the functional identity of these cells is determined by their differentiation process (6, 8). A detailed review of the two main subtypes of cDCs (cDC1s and cDC2s), as well as their markers and location in different tissues, is presented by Backer et al. which, in addition, explore in great detail the origin, location, and role of splenic CD8α <sup>+</sup>Langerin<sup>+</sup> cDC1s in the context of systemic infections and immunotherapy. Complementing the previous review, Prendergast et al.

**7**

report that CD8α <sup>+</sup>Langerin<sup>+</sup> cDC1s play an essential role in sepsis control (in a mouse model using BCG administered iv) as they activate CD8<sup>+</sup> T lymphocytes and IL-12 production, which appear to be essential for the initial control of systemic bacterial infections. The role of cDC1s in anti-tumor immunity has not been fully clarified and Cancel et al. explore the latest literature on the subject providing evidence, mainly in animal models, that cDC1s have an important role in priming and/or sustaining the activation of both, T lymphocytes and NK cells. Human studies also point to a strong association between typical cDC1s signatures and a better prognosis for some types of tumors. The role of cDC1s in cancer, as well as the cross-talk between these cells and other DC subtypes and cell types, is discussed in detail by Noubade et al. who argue that understanding these interactions is important for the design of more effective DCbased cancer immunotherapies.

In this topic, the role of other cDC subtypes in different inflammatory, infectious diseases, and cancer was also explored. Tiburcio et al. analyze the interaction between DCs and Leishmania parasites, trying to elucidate the role of DCs in establishing the immune response in leishmaniasis and how this parasite subverts this response to establish itself. Furthermore, Nico et al. showed that a vaccine using Leishmania (L.) donovani antigens was able to prevent the dysfunctional migration of DCs seen in this animal model, and induce a protective response against this parasite. In the case of bacterial diseases, Richardson et al. demonstrated that dendritic cells derived from human monocytes (moDCs) are modulated by peptides derived from Staphylococcus aureus becoming more tolerogenic and consequently inducing regulatory T lymphocytes. The use of these peptides in vaccine strategies against autoimmune diseases appears very promising. Another molecule, derived from gram-negative bacteria, that has also been explored as an adjuvant in vaccination protocols, is flagellin. Flores-Langarica et al. demonstrate that lamina propria CD103+CD11b<sup>+</sup> cDC2s respond to flagellin and are essential for the induction of T and B lymphocytes in the mucosa, especially inducing a Th2 type response. The effect of flagellin was further explored by Dos Reis et al. when moDCs from HIV-infected patients were treated with this compound. The treatment with flagellin suggested that this molecule was able to activate the inflammasome NAIP/NLRC4 and thereby promote the activation of moDCs derived from HIV<sup>+</sup> patients. Still in the context of HIV infection, Schonfeld et al. found that the infection of moDCs with the sexually transmitted bacteria Chlamydia trachomatis followed by HIV infection was able to increase viral infection and to reduce the activation of cytotoxic T lymphocytes.

The role of different DC subtypes has been reviewed in the context of autoimmune diseases and pulmonary arterial hypertension (van Uden et al.). Although several DC subtypes were found to be involved, the exact functional role of each one, or of their interactions, in the development of these pathologies needs to be further defined. In fact, the crosstalk between different DC subtypes or with other cell types is extremely important, both in the steady state and in the context of inflammation. Grabowska et al. dedicate a detailed review on the interaction between CD169<sup>+</sup> macrophages and DCs in the contexts of immunity and tolerance, presenting evidence that suggests that these cells greatly interact with each other, and that the cytokines produced by macrophages have an important impact on the activation of adaptive responses initiated by DCs against different pathogens. On the other hand, Wagner et al. evaluate the role of pDCs and 6-sulfo LacNAc-expressing monocytes (slanMo) in rectal cancer after neoadjuvant radiochemotherapy, and demonstrate that the benefits of this therapy in this particular cancer type may be related to increased infiltration of activated pDCs expressing IFNα and cytotoxic CD8<sup>+</sup> T lymphocytes. Complementing these results, Ahmad et al. not only revise the role of slanMo in cancer, but also in other pathologies.

Most of the studies carried out to better characterize DCs and their functions and interactions have exploited either animal models or adult human samples. However, in recent years, different groups were dedicated to study the functional differences between the DC subtypes found in neonates and in adult individuals. Papaioannou et al. review such efforts and speculate on strategies that could be applied to promote neonatal immunity.

The capacity of DCs and other APCs in sensing the environment in which they are situated allows them to respond, or not, to different external stimuli. The ability of DCs and other APCs to respond to different sugars has been highlighted for the induction of immunity as well as tolerance. Lubbers et al. present a detailed review of how signaling through receptors that bind sialic acid is important in inducing immune tolerance, and how this knowledge could be utilized to develop strategies to treat autoimmune diseases or allergies. Another pathway capable of suppressing DC functions is that of adenosine (Ado), a metabolite of extracellular ATP (which acts as a danger signal in the context of inflammation). Silva-Vilches et al. present a careful review of how extracellular Ado suppresses the functions of DCs previously activated by the presence of extracellular ATP. Another important contribution was made by Han et al. who suggested that deficiency in Fas signaling in DCs increases allergy by inducing pulmonary inflammation mediated by the activation of a potent Th2 response. Other articles have explored not the signaling pathways, but rather the role of different transcription factors in DC activity. Sondergaard et al. and Tel-Karthaus et al. evaluate the role of DC-SCRIPT and Nur77 transcription factors, respectively, in human moDCs, demonstrating that both control the activation and presentation capacity of these cells, thereby modulating the activation of immune responses. The transcription factor Zeb1, on the other hand, seems to mediate the production of IL-6, IL-10, and IL-12 by cDC1, thus inducing a pro-inflammatory response (Smita et al.).

The success of DCs in antigen presentation and T cell priming can be partly explained by their remarkable ability to efficiently present internalized exogenous antigens in MHC I molecules. This phenomenon is known as cross-presentation. Through this process, DCs (especially the cDC1 subtype) are able to activate CD8<sup>+</sup> T lymphocytes that play an important role during the course of the immune response, especially in viral infections and tumors. Three reviews discuss our current knowledge on the pathways that lead to efficient cross-presentation by DCs, from the acquisition of the exogenous antigen to its processing and peptide loading in the MHC I. Embgenbroich and Burgdorf discuss the two main cross-presentation pathways (the endosome-to-cytosol and the vacuolar pathways), while Gros and Amigorena emphasize what is currently known about the mechanisms for antigen export to the cytosol. Finally, Montealegre and van Endert debate in great detail the differences in the recycling pathways between DCs and non-antigen presenting cells. Despite some controversies in the field, these authors elegantly present an overview of the state-of-the-art of this very important phenomenon in DC biology.

Their incredible ability to modulate immune responses transforms DCs into ideal targets for manipulation. Once it was demonstrated that functional DCs could be obtained from peripheral blood monocytes (moDCs) (9), the next logical step was to produce these cells, load them with target antigens, induce maturation and administer them as vaccines. The first attempts to use autologous DCs loaded with antigens were made in patients with B-cell lymphomas more than two decades ago (10). Patente et al. review the clinical application of this technology, especially in cancer patients. In complement Huber et al., not only elaborate in detail the results obtained with the use of autologous moDCs in the treatment of cancer, but also present the potential use of the other subtypes of cDCs in this setting. On the other hand, da Silva et al. review the progress obtained with the use of moDCs for the treatment of HIV infection. Despite the various advances in the use of autologous moDCs as vaccines, the fact that these preparations are laborious, timeconsuming and extremely expensive have led to research aimed at directing antigens directly to the different DC subtypes in vivo. Several strategies were then devised and promising results have been obtained in animal models. Kroczek et al. describe the use of the chemokine (C motif) ligand 1 (XCL1) to target the antigen to its receptor (XCR1) in cDC1s, while Koerner et al. review the use of poly (D, L-lactide-co-glycolide) microspheres (PLGA MS) in directing antigens to DCs. Antigen targeting using monoclonal antibodies (mAbs) against DC surface receptors fused to proteins of interest has also been studied as a promising way to deliver the antigen of interest to these cells. Apostolico et al. and Antonio-Herrera et al. explore the use of different adjuvants in combination with the administration of chimeric mAbs to improve the induced immune response. Differently, Zaneti et al. direct the antigen of interest through a DNA vaccine that encodes a single-chain Fv antibody designed to be secreted

#### REFERENCES


and direct the fused protein to cDC1s. In all cases, activation of the immune system was observed.

The use of animal models to study DCs ontogeny or the effect of ablation of a specific subtype on the development of the immune response can be very useful. In this context, Iwabuchi et al. created a model of humanized mice that allowed them to study the effect of fms-related tyrosine kinase 3 ligand (Flt3- L) and granulocyte-macrophage colony-stimulating factor (GM-CSF) cytokines on the development of human DC subtypes, Moreover, Mattiuz et al. were able to generate mouse strains capable of specifically depleting cDC1s. Further studies using these models will be important to determine cDC1 function in different settings.

In conclusion, this topic presents articles that contribute to a broader understanding not only of DC biology, but also of the processes involved in the many functions that these cells play during steady-state or infections/inflammation.

#### AUTHOR CONTRIBUTIONS

SB wrote the text. SB, DD, CM, and DR reviewed and approved the final version of this Editorial. All authors contributed to the article and approved the submitted version.

#### FUNDING

This work was supported by the São Paulo Research Foundation (FAPESP, grant numbers 2018/07142-9 and 2017/17471-7), the Brazilian National Research Council (CNPq, grant number 472509/2011-0), and the Coordination for the Improvement of Higher Level Personnel (CAPES, Finance Code 001) to SB and DR, by Cancer Research Switzerland (KFS-4091-02-2017 and KFS-4962-02-2020), KFSP-PrecisionMS and HMZ ImmunoTargET of the University of Zurich, the Coronavirusfonds UZH, the Cancer Research Center Zurich, the Vontobel Foundation, the Baugarten Foundation, the Sobek Foundation, the Swiss Vaccine Research Institute, Roche, Novartis and the Swiss National Science Foundation (310030B\_182827 and CRSII5\_180323) to CM, and by the German Research Foundation (DFG) (CRC1181- TPA7, DU548/5-1), Agence Nationale de la Recherche (ANR) and the DFG (DU548/6-1), the Interdisziplinäres Zentrum für Klinische Forschung (IZKF) (IZKF-A80) and the Bavarian State of Ministry of Science and Art, Bayresq.Net (IRIS) to DD. SB and DR received fellowships from CNPq.


**Conflict of Interest:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2020 Boscardin, Dudziak, Münz and Rosa. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

#### *Edited by:*

*Nahid Ali, Indian Institute of Chemical Biology, India*

#### *Reviewed by:*

*Sima Rafati, Pasteur Institute of Iran, Iran Anita S. Iyer, Harvard Medical School, United States*

#### *\*Correspondence:*

*Clarisa Beatriz Palatnik-de-Sousa immgcpa@micro.ufrj.br*

#### *Specialty section:*

*This article was submitted to Vaccines and Molecular Therapeutics, a section of the journal Frontiers in Immunology*

*Received: 03 August 2017 Accepted: 18 April 2018 Published: 07 May 2018*

#### *Citation:*

*Nico D, Martins Almeida F, Maria Motta J, Soares dos Santos Cardoso F, Freire-de-Lima CG, Freire-de-Lima L, de Luca PM, Maria Blanco Martinez A, Morrot A and Palatnik-de-Sousa CB (2018) NH36 and F3 Antigen-Primed Dendritic Cells Show Preserved Migrating Capabilities and CCR7 Expression and F3 Is Effective in Immunotherapy of Visceral Leishmaniasis. Front. Immunol. 9:967. doi: 10.3389/fimmu.2018.00967*

*Dirlei Nico1 , Fernanda Martins Almeida2,3, Juliana Maria Motta4 , Fellipe Soares dos Santos Cardoso2 , Celio Geraldo Freire-de-Lima5 , Leonardo Freire-de-Lima6 , Paula Melo de Luca7 , Ana Maria Blanco Martinez2 , Alexandre Morrot7,8 and Clarisa Beatriz Palatnik-de-Sousa1,9\**

*1Departamento de Microbiologia Geral, Instituto de Microbiologia Paulo de Góes, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 2Programa de Pós Graduação em Anatomia Patológica, HUCFF, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 3Programa de Graduação de Histologia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 4Programa de Glicobiologia, Instituto de Bioquímica Médica Leopoldo De Meis, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 5Programa de Imunobiologia, Instituto de Biofísica Carlos Chagas Filho, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 6Programa de Medicina Regenerativa, Instituto de Biofísica Carlos Chagas Filho, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 7 Laboratório de Imunoparasitologia, Instituto Oswaldo Cruz (IOC), Rio de Janeiro, Brazil, 8Centro de Pesquisas em Tuberculose, Faculdade de Medicina, Universidade Federal do Rio de Janeiro, Rio de Janeiro, Brazil, 9 Instituto Nacional de Ciência e Tecnologia de Investigação em Imunologia, São Paulo, Brazil*

Physical contact between dendritic cells (DCs) and T cell lymphocytes is necessary to trigger the immune cell response. CCL19 and CCL21 chemokines bind to the CCR7 receptor of mature DCs, and of T cells and regulate DCs migration to the white pulp (wp) of the spleen, where they encounter lymphocytes. In visceral leishmaniasis (VL), cellular immunosuppression is mediated by impaired DC migration due to the decreased chemokine secretion by endothelium and to the reduced DCs CCR7 expression. The *Leishmania (L.) donovani* nucleoside hydrolase NH36 and its C-terminal domain, the F3 peptide are prominent antigens in the generation of preventive immunity to VL. We assessed whether these vaccines could prevent the migrating defect of DCs by restoring the expression of CCR7 receptors. C57Bl6 mice were vaccinated with NH36 and F3 and challenged with *L. (L.) infantum chagasi*. The F3 vaccine induced a 100% of survival and a long-lasting immune protection with an earlier CD4+Th1 response, with secretion of higher IFN-γ and TNF-α/IL-10 ratios, and higher frequencies of CD4+ T cells secreting IL-2+, TNF-α+, or IFN-γ+, or a combination of two or the three cytokines (IL-2+TNFα+IFN-γ+). The CD8+ T cell response was promoted earlier by the NH36-vaccine, and later by the F3-vaccine. Maximal number of F3-primed DCs migrated *in vitro* in response to CCL19 and showed a high expression of CCR7 receptors (26.06%). Anti-CCR7 antibody treatment inhibited DCs migration *in vitro* (90%) and increased parasite load *in vivo*. When transferred into 28-day-infected mice, only 8% of DCs from infected, 59%

**11**

of DCs from NH36-vaccinated, and 84% of DCs from F3-vaccinated mice migrated to the wp. Consequently, immunotherapy of infected mice with F3-primed DCs only, promoted increases in corporal weight and reductions of spleen and liver parasite loads and relative weights. Our findings indicate that vaccination with F3-vaccine preserves the maturation, migration properties and CCR7 expression of DCs, which are essential processes for the generation of cell-mediated immunity. The F3 vaccine is more potent in reversing the migration defect that occurs in VL and, therefore, more efficient in immunotherapy of VL.

Keywords: visceral leishmaniasis, dendritic cells defective migration, CCR7 expression, nucleoside hydrolase, NH36, F3 domain, *Leishmania donovani*, *Leishmania infantum chagasi*

#### INTRODUCTION

Leishmaniasis is still considered one of the most neglected diseases in the world (1). Approximately, 350 million people are at risk, and about two million new cases are registered annually (2). More than 20 *Leishmania* species are involved in the transmission of Leishmaniasis. The parasites are transferred to humans by hematophagous phlebotomine sandflies (3). *Leishmania (L.) donovani*, *L. (L.) infantum*, and *Leishmania (L.) infantum chagasi* are the agents of visceral leishmaniasis (VL). While the disease is anthroponotic in India and East Africa, in the Americas, North Africa, Asia, and the Mediterranean, VL is a canid zoonosis (4). Clinical signs of human VL, include fever, malaise, anorexia, cachexia, hypergammaglobulinemia, hepato- and splenomegaly, anemia, and progressive suppression of the cellular immune response. Currently, the annual incidence reaches 400 thousands cases and 30 thousands deaths worldwide (5). Ninety percent of VL cases are registered in India, Ethiopia, South Sudan, Bangladesh, Sudan, and Brazil. Although the VL control programs in South-East Asia are reducing the human incidence of the disease, and the number of VL cases declined in Bangladesh, India, and Nepal (3), recurrent outbreaks of VL in and Sudan, Kenya, Ethiopia, and South Sudan are raising concern.

The development of the cellular immune response requires that T cell lymphocytes make contact with dendritic cells (DCs) in the spleen and lymph nodes (6, 7). When the spleen is chronically infected with *Leishmania* parasites, the structural design of the B cell follicles and the marginal zone (MZ) are disrupted (8–10). This disorder determines an insufficient antigen presentation to T cells. In fact, splenic T cells and DCs move from the MZ to the periarteriolar lymphoid sheath (PALS), where there are increased concentrations of chemokines (11). CCL19 and CCL21 chemokines produced by endothelium venules (12) bind to CCR7 receptor. These chemokines are attractants to T cell naïve lymphocytes, mature DCs, and a subset of memory T cells (12, 13). Therefore, although the number of splenic DCs increases after *L. (L.) donovani* infection (11), they fail to migrate to PALS, due to the reduced chemokine secretion by PALS, and to the inhibition of CCR7 expression on DCs (11). This spatial separation of T lymphocytes and DCs impedes their physical contact and is one of the reasons of the suppression of cellular immune response in VL (11).

Confirming the results obtained by Ato et al. (11) who studied the *L. (L.) donovani* infections, we recently demonstrated that mice chronically infected with *L. (L.) infantum chagasi*

also show an increased spleen relative weight, correlated to a DCs hyperplasia, and to an increased spleen parasite load (14). In contrast, mice vaccinated with the nucleoside hydrolase (NH36) recombinant antigen, or with its C-terminal moiety (F3) and saponin, showed a strong reduction in spleen parasite load and prevented the hyperplasia of spleen DCs and an increase in spleen relative weight (14).

NH36 is a promising vaccine antigen, which protects mice and dogs from VL infection (15–18). NH36 domains and epitopes are recognized by PBMC of subclinical and cured human patients from Brazil (19) and from Spain (20). F3 holds the most important NH36-epitopes for antibodies and MHC class II receptors of mice and induces a CD4<sup>+</sup> T-cell-mediated protection against VL, correlated with an enhanced TNF-α and strong decrease of IL-10 secretion (15, 21).

No human vaccine against VL has been licensed until now and chemotherapy shows toxicity and failure issues (22, 23). Leishmaniasis remains as one of the main parasitic diseases of major impact on humanity and the search for isolated, combined, or alternative therapies that are safe, effective, and easily administered for the treatment of VL remains a promising target study. In this investigation, we take a step further in the study of the efficacy of the NH36 and F3 vaccines against murine VL and demonstrate that, besides preventing the increase of the numbers of DCs, they also prevent the migration dysfunction of DCs by restoring their CCR7 receptor expression. Additionally, we show that immunotherapy of infected mice with DCs derived from animals vaccinated with the F3 domain is a potential tool to assist in the treatment of the VL.

#### MATERIALS AND METHODS

#### Ethical Statement and Biosafety Measures

Protocol design of the experiments was approved by the Comissão de Ética no Uso de Animais of the Universidade Federal do Rio de Janeiro (CEUA protocol IMPPG040-07/16), in agreement with Brazilian laws for animal safety and the guidelines of National Institute of Health (15). Animals were maintained at the Instituto de Microbiologia Paulo de Góes, which is part of Universidade Federal do Rio de Janeiro (UFRJ) facilities, given water and food *ad libitum*, with a 12 h light/dark cycles and controlled temperature (15). We aimed to reduce any animal suffering to a minimum.

In this investigation, we worked with genetically modified *Escherichia coli* BL21 and DH5 strains cloned with the pET28b plasmid expressing the NH36 and F3 recombinant proteins. These bacterial clones used in this investigation are considered OGM risk level 1 (CQB 0108/99 IMPG-UFRJ) because they are not associated with disease in adults. We also manipulated nongenetically modified *Leishmania* parasites for which the CTNBio biosafety level is 2. Following the Brazilian Ministry of Health Biosafety regulations for Biomedical and Microbiology laboratories we performed the following biosafety measures: use of apron, gloves, masks, and protective glasses, decontamination of infected biological material and animal cases before washing and limited access to, security block and signalizing of the risk area. Additionally, we used laminar flow, aseptic chambers, automatic pipetters, and reservoirs with sterilizing solutions. On the day of the experiments with the bacterial clones, no other manipulation of microorganisms occurred. All the non-disposable material and surgery tools are autoclaved. A separated manipulation room with an ultraviolet and exhaust chamber was used when necessary. Carcasses were incinerated.

# Recombinant NH36 Antigen and F3 Domain

NH36 is composed of 314 amino acids. The NH36 sequence is deposited in SWISS-PROT (accession code Q8WQX2), EMBL (AY007193), GenBank™ and DDJB (AAG02281.1) data bases (15, 20). F3 is composed of the amino acids 199–314 of NH36. The sequences of the NH36 and F3 were cloned in the pET28b plasmid system, expressed in *E. coli* Bl21DE3 cells and purified by affinity chromatography (Ni-NTA, Qiagen), as described before (15, 20). The presence or absence of LPS was confirmed using the LAL QCL-1000 kit (Lonza). The levels of LPS were lower than the sensitivity range of the Limulus amebocyte lysate test, which is 0.1–1.0 EU/ml. Therefore, there was no need for endotoxin removal.

# Vaccination and Challenge With *L. (L.) infantum chagasi*

Eight-week-old C57BL/6 female mice, obtained from the Fiocruz-Cecal facilities (Rio de Janeiro, Brazil) were vaccinated subcutaneously, with three doses of 100 µg of NH36 or F3 formulated with 100 µg of saponin (SIGMA), at weekly intervals (15). Control mice received only saline. On week 4, mice were challenged with an intravenous injection in the tail vain of 3 × 107 *L. (L.) infantum chagasi* amastigotes (strain IOC-L 3324) isolated from infected hamsters' as described before (24, 25). On day 28 after infection, euthanasia was performed and the variation in total body weight and relative liver/body weight were recorded (15). The parasite load in the livers was determined by microscopic observation of Giemsa-stained impression smears. The parasite burden was assessed as LDU values (number of amastigotes per 1,000 of organ cell nuclei/mg of organ weight) (11, 15, 16, 18, 25). Alternatively, the parasite burden was assessed by a limiting dilution assay (LDA) of the aseptically removed liver fragments suspended in a 1/5 serial dilution in Schneider's medium, and incubated at 26°C. Promastigotes present in the last well containing visible parasites were quantified in a hemocytometer (21). Two-independent experiments were performed with *n* = 7 mice per treatment in each treatment.

#### Assessment of the Intradermal Response to Leishmanial Antigen (IDR)

IDR against *L. (L.) donovani* lysate was measured in the footpads on day 7 after immunization and on day 25 after challenge (15, 26). The right hind footpads were injected intradermally with 107 freeze–thawed *L. (L.) donovani* promastigotes at the stationary phase (15, 26–28). Before, and at 24 and 48 h after injection the swelling was assessed with a Mitutoyo apparatus. Each animal received only 0.1 ml saline in the left hind footpad as control. Values of the saline control were subtracted from the reaction due to *Leishmania* antigen, at each measurement (15, 26). Twoindependent experiments were performed with *n* = 6–7 C57BL/6 mice per treatment in each experiment.

# DCs Isolation and Migration Tests *In Vitro* and *In Vivo*

Dendritic cells was isolated from spleens of vaccinated mice, 28 days after challenge, following the instructions of the magnetic beads manufacturer (Miltenyi Biotec, USA). Briefly, spleens were collected, shredded, and incubated with collagenase, at 1 mg/ml, and DNAse (Sigma) at 20 µg/ml concentration (29), respectively, for 20 min, at room temperature (Figure S1 in Supplementary Material). After that, the spleens were macerated and the splenocytes passed through a cell mesh. Also an ACK solution was used for erythrocyte lysis. Magnetic microbeads conjugated with anti-mouse CD11c were added to the preparation and then incubated for an additional 30 min. After that, the DCs were purified using MACS magnetic columns (Miltenyi Biotec, Germany) (11) and labeled with anti-mouse CD11c antibody conjugated to fluorescein isothiocyanate (FITC), clone N418 (eBiosciense, USA). Cells were additionally incubated with allophycocyanin (APC)-conjugated to anti-CCR7 (eBioscense) (two experiments with *n* = 5 mice per treatment for each experiment) (Figure S1 in Supplementary Material). The cytometry analysis was performed using a FACSCalibur apparatus (Becton Dickinson).

The *in vitro* migration assays used DCs purified from C57BL/6 naïve, *L. (L.) infantum chagasi* infected and NH36 or F3 vaccinated and infected mice. The DCs (5 × 105 per well) were suspended in RPMI containing 1% fetal calf serum and placed into the upper chamber of Transwell inserts (5 µm pore size, Coaster, Corning). The inserts were further positioned in plates containing RPMI, with or without the addition of 100 ηM CCL19 (R&D Systems). After 2 h of incubation at 37°C, the cells in the lower wells of the inserts were collected and counted in a hemocytometer chamber. For the purity analysis, cells which migrated were stained with anti-mouse CD11c antibody, and analyzed by flow cytometry. Two-independent experiments were performed, each one of them with *n* = 5–6 mice per treatment. In order to prove that the DCs migration is due to the sensitivity of CCR7 receptors to the CCL19 chemokine gradient, blocking experiments were performed by incubating the DCs obtained from infected or F3sap-vaccinated mice with anti-CCR7 antibody (e Bioscience) at 1/50 dilution for 30 min at 4°C, before the migration assay in transwell plates, as described.

The migration assays *in vivo* used DCs purified from spleens of *L. (L.) infantum chagasi* infected and NH36 or F3-vaccinated, or from infected mice. Isolation and purification of DCs were performed using magnetic beads as described for the *in vitro* assays (Figure S1 in Supplementary Material). Additionally, the DCs were stained with two drops/ml (6 µg/ml) of Hoescht 33342 (Life Technologies), for 20 min at 37°C, protected from light. After that, 106 labeled DCs were injected intravenously into the infected receptors, on day 28 after infection, as described by Ato et al. (11). Spleens of receptors were removed 24 h after the DCs transfer and immersed in paraformaldehyde, followed by cryoprotection in an increasing sucrose gradient (10, 20, and 30%). Cryostat sections (16 µm) were performed and mounted in Fluoromount and observed under a confocal microscope (LeicaTCS SP5). Sections from three mice of each treatment were chosen randomly and analyzed for the presence of labeled DCs in the white pulp (wp) or in the red pulp (rp) or in the MZ. Approximately 1,000 stained DCs were recorded and the differences between the proportions detected in the wp, and the rp, were analyzed by the Fischer exact test for comparison of proportions (https://www.graphpad.com/quickcalcs/contingency2/). Two-independent experiments were performed, each one of them with *n* = 3 receptor mice per treatment.

#### DCs Immunotherapy

Splenic DCs from *L. (L.) infantum chagasi*-infected mice*,* and from NH36 or F3-vaccinated and challenged mice, were purified on day 28 after infection. The total amount of purified DCs obtained from the spleens from the different experimental groups varied from 1 to 2.6 × 106 . For the immunotherapy experiments, 106 DCs of each group of mice were injected intravenously into *L. (L.) infantum chagasi*-infected C57BL/6 recipients on day 28 post infection. The parasite loads in spleen and livers, the spleen and liver/corporal relative weights, and the variation in body weights were recorded in the recipient mice 7 days later by LDU. Two-independent experiments were performed with *n* = 5 mice per treatment in each experiment. As a control, blocking experiments to prove that migration of DC is critical for the protection in mice were performed using DCs obtained from normal, infected, and NH36, and F3-vaccinated and infected mice, that were preincubated with anti-CCR7 antibody and further transferred to infected receptors.

# Cytokine-Secretion Assay in Vaccinated Mice

Spleens were aseptically removed and splenocyte suspensions were prepared as described before (21). Purified DCs were also obtained. Cells were distributed into 96-well plate (106 cell/well) and incubated with 5 µg/ml NH36 or lysate of stationary phase *L. (L.) infantum chagasi* promastigotes, or with no addition for 72 h *in vitro,* at 37°C with 5% CO2. After that, supernatants were harvested and assayed using the Mouse IFN gamma, TNF-α, and IL10 ELISA Ready-SET-Go! (e-Biosciences, USA). The sensitivity of the assay was established with a range of 10–1,000 pg/ml for TNF-α, 100–1,000 pg/ml for, and of 15–2,000 pg/ml for IFN-γ. Reactions were developed using biotinylated anti-cytokine antibodies, streptavidin (SAv-HRP) enzymatic reagent, and TMB (Zymed, USA). Absorbances were monitored in a BioRad ELISA reader at 655 ηm. Two-independent experiments (*n* = 5 mice per treatment in each experiment) were performed.

# Intracellular Cytokine Staining in Vaccinated Mice

The analysis of the T cell response used splenocytes stimulated *in vitro* with 5 µg/ml NH36 or lysate of stationary phase *L. (L.) infantum chagasi* promastigotes or with no addition for 24 h at 37°C with 5% CO2, as described (21), and with incubation with Brefeldin A (SIGMA). After that, the splenocytes were stained with rat anti-mouse-CD4FITC (clone GK1.5) and -CD8FITC (clone 53–6.7) monoclonal antibodies (eBioscience), fixed, washed, treated with 0.5% saponin-FACS buffer, and stained with IFNγAPC, IL-2-PerCP-Cy5.5, and TNF-αPE monoclonal antibodies (eBioscience). A total of 100,000 events were acquired. Gating for CD4<sup>+</sup> ad CD8<sup>+</sup> T cells was performed using a FACSCalibur apparatus (Becton Dickinson). Data analysis was performed with the Flow-jo program (Treestar, USA). Two-independent experiments were performed with *n* = 5 mice per treatment in each experiment.

# Statistical Analysis

Means were compared using the Kruskal–Wallis and Mann– Whitney non-parametrical tests. Correlation coefficients were established by the Pearson's two-tailed correlation test (GraphPad Prism 6 software) as described (21). Also compared the survival distribution of individuals in infected controls and vaccinated and challenged mice were compared using the Log-rank (Mantel– Cox) and the Gehan–Breslow–Wilcoxon tests (GraphPad Prism 6 software).

# RESULTS

#### Vaccination With F3 and NH36 Prevents Dysfunctional DCs Migration and Decreased CCR7 Expression

In the present investigation, we aimed to assess whether DCs from vaccinated mice preserve their correct migrating capabilities in response to CCL19 chemokine, and if this fact is related to their normal sustained expression of the CCR7 receptor.

As a source of DCs we used normal, infected, and NH36 and F3-vaccinated and challenged mice. Initially, we confirmed that the vaccine-induced protection in the these mice by assessing the achievement of enhanced IDR responses, reduced liver parasite loads and liver relative weights, and sustained corporal weights (**Figure 1**). In fact, vaccination with both, the F3 and the NH36 vaccines, significantly increased the IDR after complete immunization (**Figures 1A,C**), and after challenge (**Figures 1B,D**) above the levels detected in normal control mice, at all tested times (24 and 48 h after injection). Remarkably, after infection, the F3 vaccine was 74 and 79% stronger than controls, at 24 and

48 h, respectively and 27% stronger (*p* < 0.0001) than the NH36 vaccine (**Figures 1B,D**).

IDR values correlated negatively with the liver/body relative weight after immunization (*p* < 0.0001, *R* = −0.5768, *R*<sup>2</sup> = 0.3327 at 24 h and *p*< 0.0001, *R*=−0.5833, *R*2= 0.3402 at 48 h after injection) and after challenge (*p* < 0.0001, *R* = −0.5989, *R*<sup>2</sup> = 0.3587 at 24 h and *p* < 0.0001, *R* = −0. 6334, *R*<sup>2</sup> = 0.4012 at 48 h after injection).

Notably, both vaccines strongly reduced the liver LDU values compared to those of infected mice (*p* < 0.0001) (**Figure 1E**). Compared to infected controls, F3 vaccine produced a 98.8% reduction, and the NH36 vaccine, a 97.6% of reduction of the parasite load (*p* = 0.0318) (**Figure 1E**). F3 vaccine reduced the LDUs values by 50% in comparison to the NH36 vaccine. The reduction of the parasite load determined by both vaccines was also confirmed by the LDA method (**Figure 1F**). The F3 vaccine induced 98% (*p* < 0.079), and the NH36 vaccine only 88% reduction (*p*< 0.0079) (*p*< 0.05) of the number of promastigote in liver culture. Additionally, the F3 vaccine reduced the promastigotes counts by 87% (*p* < 0.05) in comparison to the NH36 vaccine (**Figure 1F**). The LDU and LDA values for the liver parasite loads were positively correlated (*p* < 0.026, *R* = 0.496, *R*<sup>2</sup> = 0.2464).

The increases in IDR after infection were strong correlates of protection regarding the liver LDU values. In fact, LDU values were negatively correlated with IDR after immunization, at 24 h (*p* < 0.0001, *R* = –0.7875, *R*<sup>2</sup> = 0.6202) and at 48 h (*p* < 0.0001, *R* = −0.7674, *R*<sup>2</sup> = 0.5890) and after challenge, at 24 h (*p* < 0.0001, *R* = −0.7796, *R*<sup>2</sup> = 0.6077) and at 48 h (*p* < 0.0001, *R* = −0.7543, *R*<sup>2</sup> = 0.5690).

Increases in liver/body relative weight were also significantly higher in infected control animals than in normal uninfected controls and in F3 and NH36-vaccinated and challenged mice (*p* < 0.0001 for all comparisons) (**Figure 1G**). In addition, only infected controls lose corporal weight (**Figure 1H**) in comparison to normal, F3 and NH36-vaccinated mice (*p* < 0.0001 for all comparisons). Vaccinated and normal mice showed similar gain in corporal weight. Besides, IDR was positively correlated to corporal weight gain after immunization (*p* = 0.0009, *R* = 0.5055, *R*<sup>2</sup> = 0.2556, at 24 h and *P* = 0.0002, *R* = 0.5504, *R*<sup>2</sup> = 0.3030, at 48 h after injection) and after challenge (*p* < 0.0001, *R* = 0.5888, *R*<sup>2</sup> = 0.3467, at 24 h and *p* < 0.0001, *R* = 0.6532, *R*<sup>2</sup> = 0.4267, at 48 h after injection).

Once the generation of a protective response was confirmed, we further studied the migrating capabilities of DCs obtained from these vaccinated, naïve controls, and infected mice on day 28 after challenge. In **Figure 2**, we represent the results as boxes and whiskers. The whiskers show the maximal and minimum values and the top and bottom of the bars show, respectively, the 75th and 25th percentiles. The 75th percentile is the value below which 75% of observations in a group of observations fall. **Figure 2A** shows that the number of total splenocytes isolated was, as expected, higher in infected mice than in the two vaccinated groups. FACS cytometry analysis of the affinity chromatography purified CD11c<sup>+</sup> DCs preparations disclosed 90–100% rates of purity. We further observed that the number of DCs purified from total splenocytes was higher in infected and F3-vaccinated, than in normal mice (**Figure 2B**).

Additionally, none of the DCs migrated toward the medium containing no additions (**Figure 2C**). In contrast, while DCs of normal controls and of F3 and NH36-vaccinated mice migrated toward the CCL19 gradient, DCs of infected animals did not (*p* < 0.008 for all comparisons). The maximal number of migrated DCs was 11,496 for the F3, and 9,152 for the NH36 vaccine (**Figure 2C**). Accordingly, 75% of the migrated DCs count of the F3 vaccine group fall below 8,596, while 75% of the NH36 vaccine counts fall only below 6,130 DCs counts. Therefore, although the differences in the generation of the DC migrating capabilities of the two vaccines were not statistically significant, the maximal values and 75th percentiles suggest the superiority of the F3 vaccine. As a blocking control, we incubated DCs obtained from either infected or F3sap-vaccinated and challenged mice, with anti-CCR7 antibody, in the presence of CCL19 stimulus (**Figure 2D**). The anti-CCR7 antibody blocked 90% of the *in vitro* migrating capabilities of the DCs obtained from the F3sap-vaccinated mice.

Our results indicate that protective immunity against the F3 and NH36 antigens, not only restores the DCs migrating dysfunction of mice infected with *L. (L.) infantum chagasi,* but it also potentiates this capability above the levels found in normal naïve mice (**Figure 2**). In agreement with the parasitological results, which indicated that the F3 vaccine was the most protective, the superiority of the F3 vaccine is reinforced by the finding of a higher number of purified DCs, higher maximal counts of migrated DCs, and strong inhibition of these migrating capabilities by treatment with anti-CCR7 antibody.

In VL, abnormal DC migration is partially due to the reduced expression of their CCR7 receptor, which is sensitive to the CCL19 chemokine. In order to evaluate whether the restored DCs migrating capabilities of vaccinated mice were related to a preserved CCR7 expression on DCs. With that aim, we analyzed the CCR7 protein expression on the surface of purified splenic DCs using APC-conjugated anti-CCR7 antibody and flow cytometry analysis. Based on our findings (**Figure 3**), vaccination with F3sap and NH36 upregulated the expression of CCR7 receptor on DCs. CCR7 expression was low in normal mice (1.56%) (**Figure 3A**), extremely reduced in infected mice (0.37%) (**Figure 3B**), and much stronger in animals treated with the F3 vaccine (26.06%) (**Figure 3C**) than in those vaccinated with NH36 (5.29%) (**Figure 3D**).

Therefore, vaccination with F3 and NH36 stimulates IDR, protects mice from clinical VL, restores and enhances DCs migrating capabilities probably due to the increase in CCR7 expression on DCs. Our findings of the reduced liver parasite load, increased IDR and enhancement of the CCR7 expression highlight the stronger immunogenic effect of the F3 vaccine.

#### Immunotherapy With DCs

In order to confirm the DC migration *in vivo*, DCs from infected, or NH36- or F3-vaccinated and *L. (L.) infantum chagasi*-challenged mice, were stained with Hoescht 33342 *in vitro,* and further injected into infected mice. Twenty four hours after injection, the spleens of receptors were removed and frozen, and the transferred DCs were observed by fluorescence microscopy in the wp or rp of the frozen sections (**Figure 4**). Fluorescent DCs were counted

in a total of 19–21 sections of each treatment and representative images are shown in **Figures 4A–C**. Additionally, the records are represented as proportions of DCs from donors distributed either in the white or rps of infected receptor spleens. On day 28 of the infection of recipients, DCs from infected donors remained mostly in the rp (92%) (**Figure 4A**) and, only minor proportions migrated to the MZ of the wp (8%; *p* < 0.001). DCs from NH36-vaccinated mice are equally distributed between the white (59%) and the rp (41%; *p* = 0.1144) (**Figure 4B**). Remarkably, 84% of DCs from F3-vaccinated mice preferentially migrated to the wp and its MZ (**Figure 4C**), while only 16% remained in the rp (*p* < 0.0001).

as disclosed by Mann–Whitney non-parametrical test.

In order to study the potential use of F3 and NH36 antigenprimed DCs in the immunotherapy of VL we transferred DCs from infected, and from vaccinated and challenged mice, into recipient mice infected with *L. (L.) infantum chagasi* 28 days before. Seven days after the DCs transfer, the impact on the parasitological and clinical cure was assessed. While infected animals lost 1.54 g of corporal weight, the weight loss was 79% lower in mice that received F3-primed DCs (0.32 g), and 91% lower in mice treated with NH36-primed DCs (0.14 g) (**Figure 5A**). Additionally, only the F3-primed DCs were able to alter the impact of the infection on mice, by reducing the parasite load in the spleen (*p* < 0.007) (**Figure 5B**), the spleen/body relative weight (**Figure 5C**), the liver/body relative weight (*p* < 0.05) (**Figure 5D**), and the liver LDU values (*p* < 0.05) (**Figure 5E**). Therefore, the results of immunotherapy with DCs correlate with the results of the preventive vaccination, thus demonstrating that the F3 vaccine is more potent than the NH36 vaccine. The immunotherapeutic effect of F3-primed DCs (**Figure 5F**), but not of the NH36-primed DCs (not shown) was blocked by pre-incubation with anti-CCR7 antibody, and a 43 and 52% increased parasite load was observed in spleens and livers, respectively of the recipient mice. These results indicate that the increased migrating capabilities of DCs from F3-vaccinated mice, both *in vitro* and *in vivo* (**Figures 2D**, **4C** and **5F**) are related to the enhanced CCR7 receptor expression (**Figure 3**), which contributes to the cure of VL.

#### Evolution of the Cellular Immune Response of Vaccinated DC Donor Mice Secreted Cytokines and Survival

We assessed the profile of IFN-γ, TNF-α, and IL-10 secretion along the time, in the supernatants of splenocytes of mice vaccinated with F3 or NH36 vaccines and further challenged (**Figure 6**). We monitored the cytokine secretion of splenocytes in response to NH36 or to the *L. (L.) infantum chagasi* lysate antigen. On day 15 after challenge, when the parasite load was maximal in livers (**Figure 6C**), the F3-vaccine enhanced the IFNγ and TNF-α secretion in response to NH36, five times more than the NH36 vaccine (**Figures 6A,D**). The F3-vaccine also enhanced by a factor of 1.4–5, the respective IFN-γ and TNF-α levels when compared to infected controls (**Figures 6A,D**). On the other hand, IL-10 secretion was reduced more by the F3 than by the NH36 vaccine (**Figure 6G**). The IFN-γ/IL-10 and TNF-α/IL-10 ratios (**Figure 6I**), that were also higher for the F3 vaccine (10 and 2.9, respectively) than for the NH36 vaccine (6.5 and 1.7, respectively) suggest the superiority of the F3sap vaccine in the induction of a probable Th1 response. The *Leishmania* lysate promoted a similar response with superiority of the F3 vaccine although with a lower IFN-γ on day 15 and higher IL-10 secretion on day 28 (**Figures 6B,G**). Globally, the pro-inflammatory cytokine response of splenocytes was maximal (**Figures 6A,B,D,E**) since the early infection, when vaccine protection was already registered in spleens (**Figure 6F**). Levels of IFN-γ and TNF-α decreased until day 28, when, in contrast, IL-10 secretion was maximal. This pike in IL-10 secretion is coincident with the maximal parasite load detected in spleens. However, the vaccine efficacy was also long-lasting until day 28, when the parasite load was higher in spleens. In fact, while 90% of the parasite load reduction was induced by the F3 vaccine on day 15, protection was maximal (94%) on day 28 after infection (**Figure 6F**).The survival Kaplan–Meier curve analysis represented in **Figure 7**, confirmed that protection generated by the F3sap vaccine is long-lasting. In fact, while all mice from the infected control group died between day 29 and day 40 after infection, a 100% of mice vaccinated with F3sap survived until euthanasia on day 45, confirming the longevity of the vaccine immune response. In contrast, 80% of NH36-vaccinated mice survived until day 29 and only 40% of them were alive on day 45 (**Figure 7**).We also studied the cytokine response *in vitro*, in the supernatants of DCs obtained from infected, and vaccinated and infected DC donors, on day 28, before transfer to infected mice (**Figure 8**). Although the total splenocyte cytokine response was lower on day 28, than on day 15 (**Figures 6A,B,D,E,G,H**), the DCs of F3-vaccinated mice secreted enhanced levels of IFN-γ, in response to NH36 and lysate (**Figure 8A**) and of TNF-α, in response to lysate only (**Figure 8B**). No significant differences in IL-10 levels were observed in response to the antigens (**Figure 8C**).

#### Intracellular Expression of Cytokines

The Figure S2 in Supplementary Material, summarized the strategy used for the analysis of multifunctional T cell response

using a four-color flow cytometry panel to simultaneously analyze multiple cytokines at the single-cell level in splenocytes cultures. After analyzing the total production of each cytokine by CD4<sup>+</sup> or CD8<sup>+</sup> cells, the Boolean gates (or combinatorial analysis) tool of the FlowJo program was used to determine the frequencies of the possible seven combinations of cytokineproducing cells.

In correlation with the cytokine secretion to supernatants by splenocytes (**Figure 6**), the intracellular expression of cytokines by CD4<sup>+</sup> T lymphocytes in response to NH36, was more intense on day 15 than on day 28 after infection (**Figure 9**). On both days, the F3sap vaccine was superior to the NH36 formulation enhancing all types of CD4<sup>+</sup> T cells secreting one cytokine (IL-2<sup>+</sup>, TNF-α+, or IFN-γ+) (**Figures 9A,D,G**), the combination of two (IL-2<sup>+</sup>TNF-α+, TNF-α+IFN-γ+, or IL-2<sup>+</sup>IFN-γ+) (**Figures 9B,E,H**), or the three cytokines simultaneously (IL-2<sup>+</sup>TNF-α+IFN-γ+) (**Figure 9C**). The only exception was detected in CD4<sup>+</sup>TNF-α+ (**Figure 9D**) and CD4<sup>+</sup>TNF-α+IFN-γ+ T cells (**Figure 9H**) which were increased more, in the infected controls than in the F3-vaccinated mice, only on day 15.

Additionally, the global proportions of CD4<sup>+</sup> T lymphocytes were lower in infected, than in normal mice (**Figure 9F**), although the F3 vaccine sustained higher CD4<sup>+</sup> T cells counts than the NH36 vaccine, even despite the advancement of spleen infection on day 28 (**Figure 6F**).

In contrast to the IFN-γ and TNF-α splenocyte secretion to supernatants (**Figure 6**), which were slightly higher in response to NH36 than to the lysate (**Figure 6**), the intracellular expression of cytokines in CD4<sup>+</sup> T cells was similar, in response to both antigens (**Figure 9**; Figure S3 in Supplementary Material). In fact, the *Leishmania* lysate stimulus also increased the frequencies of CD4<sup>+</sup>-secreting T cells, more on day 15 than on day 28 after

infection. The F3 vaccine was stronger than the NH36 vaccine for all types of CD4<sup>+</sup> T cells (Figure S3 in Supplementary Material). Also, and as detected after stimulation with NH36, the frequencies of CD4<sup>+</sup>TNF-α+ and CD4<sup>+</sup>TNF-α+IFN-α T cells were still higher in infected controls and only increased by the F3-vaccine, on day 28 after infection (Figures S3D,E in Supplementary Material).

Regarding the cytotoxic immunity in response to NH36, the NH36-vaccine was the strongest enhancer of the proportions of CD8<sup>+</sup>IL-2<sup>+</sup> and CD8<sup>+</sup>TNF-α+IFN-α+ on day 15 (**Figures 10A,E**) and of CD8<sup>+</sup>IL-2<sup>+</sup>TNF-α+IFN-γ+ lymphocytes on day 28 (**Figure 10C**). In contrast, on day 28, the F3-vaccine was the most potent enhancer of the frequencies of CD8<sup>+</sup>IL-2<sup>+</sup> (**Figure 10A**), IFN-γ+ (**Figure 10G**), TNF-α+IL-2<sup>+</sup> (**Figure 10B**), TNF-α+IFN-γ+ (**Figure 10E**), and IL-2<sup>+</sup>IFN-γ+-secreting T cells (**Figure 10H**). The F3-vaccine superiority for TNF-α+IL-2<sup>+</sup> enhancement was already evident on day 15 (**Figure 10B**).

Additionally, when stimulated with lysate stronger frequencies of CD8<sup>+</sup>IL-2<sup>+</sup>, CD8<sup>+</sup>TNF-α+IL-2<sup>+</sup>, CD8<sup>+</sup>TNF-α+IFN-γ+, and CD8<sup>+</sup>IL-2<sup>+</sup>TNF-α+IFN-γ+ T cells were observed, on day 15 (Figure S4 in Supplementary Material) in mice vaccinated with the NH36 vaccine. In contrast, the F3-vaccine was superior for the proportions of CD8<sup>+</sup>IFN-γ+ and CD8<sup>+</sup>IL-2<sup>+</sup>IFN-γ+ T cells, both on days 15 and 28, and of CD8<sup>+</sup>IL-2<sup>+</sup> and CD8<sup>+</sup>IL-2<sup>+</sup>IFN-γ+ T cells, on day 28. Similar to what was detected for CD4<sup>+</sup> T cell whole frequencies (**Figure 9F**) the CD8<sup>+</sup> T cell proportions were reduced in all challenged mice, but in this case, no preventive effective was induced by any of the vaccines (Figure S4F in Supplementary Material).

# DISCUSSION

Although no human vaccine is available for human VL yet, there are four veterinary vaccines that have been developed for prophylaxis of canine VL (30–33). FML, a complex glycoprotein antigen of *L. (L.) donovani* and saponin, are the components of Leishmune®, the first licensed canine vaccine against leishmaniasis (33–35). The Leishmune® dog vaccination has decreased the incidence of canine and human leishmaniasis in endemic areas (33). Leishmune® is a transmission blocking vaccine (27, 34) and vaccinated dogs remained non-infective to insect vectors (35). NH36 is the main antigen of FML (27, 36) and F3 is its C-terminal domain (15). The F3-vaccine induced a significantly higher decrease in parasite load (95%) and splenomegaly (49%) in C57Bl6 mice infected with *L. (L.) infantum chagasi* (14) than

the NH36 vaccine, which in contrast, reduced the parasite load and relative weight by 87 and 39%, respectively (14). Both vaccines, however, prevented, to a similar extent, the increase in total counts of DCs and severe splenomegaly due to *L. (L.) infantum chagasi* infection (14), both of which were also previously observed during infections with *L. (L) donovani* in BALB/c and C57BL/6 mice (11).

In this investigation, we demonstrate that both NH36 and F3 vaccines increased the IDR, strongly reduced the parasite load in livers and the liver/body relative weight and promoted a gain in corporal weight. We also showed that IDR was also a strong correlate of protection against VL (15). Noteworthy, F3 was more potent than the NH36-vaccine in IDR after *L. (L.) infantum chagasi* challenge and also in reducing the liver parasite load of C57Bl6 mice. The importance of NH36 in immune prevention against VL has been extensively proven in mice (15, 16, 18, 21), dogs (17), and suggested for humans (19, 20, 37). Nonetheless, we also demonstrated that the F3 domain is the responsible for the CD4<sup>+</sup> T cell-mediated protection induced by NH36 (15). In fact, F3 was 37% more powerful than the NH36 cognate protein in the increase of antibodies, frequencies of CD4<sup>+</sup> T lymphocytes, levels of secreted IFN-γ, and ratios of CD4<sup>+</sup> and CD8<sup>+</sup> T lymphocytes producing IFN-γ and IL-10 (15) and holds the most potent antibody (15, 38) and MHC class II restricted epitopes (15, 21). In that investigation (15), IDR and ratios of TNFα/IL-10 producing CD4+ T cells were strong correlates of protection. Further, *in vivo* depletion with anti-CD4 and anti-CD8 monoclonal antibodies confirmed protection, which

was also, long-lasting. The F3 domain was also responsible for the CD4<sup>+</sup> T cell-mediated protection against mice infection by *L. (L.) amazonensis* (21).

however, only detected by the Log-rank Mantel–Cox (*p* = 0.0277).

In this investigation, our results confirm the superior efficacy of the F3-vaccine in inducing a Th1 response against *L. (L.) infantum chagasi* infection in C57Bl6 mice, which includes the generation of long-lasting protection. The F3 vaccine induced an earlier protective T cell response (day 15) with multifunctional CD4+IL-2<sup>+</sup>TNF-α+IFN-γ+ T cells, while the NH36 vaccine promoted a CD8<sup>+</sup>IL-2<sup>+</sup>TNF-α+IFN-γ+ T cell response, which started later (day 28). Besides the multifunctional T cells, the F3 vaccine also increased the proportions of CD4<sup>+</sup> T cell secreting IL-2 or IL-2 and TNF-α, and the NH36 vaccine enhanced the frequencies of CD8<sup>+</sup> T cell secreting IL-2 or IL-2 and TNF-α. These cells have been considered as a reservoir of memory T cells that have also effector potential and that can, together with the multifunctional T cells, promote optimal protection (39).

Additionally, the response of CD4<sup>+</sup>-secreting T cells was slightly higher for the NH36 than to the lysate stimulus, indicating that the NH36 is an important and predominant antigen of *Leishmania* promastigotes. The cytotoxic response, in contrast, is more stimulated by the lysate than by the NH36 antigen, and stronger in mice vaccinated with the NH36, than with the F3 vaccine. These results suggest that the CD8 response is directed to epitopes of NH36, which are not located in the F3 domain. As a matter of fact, the most important epitope for MHC class I molecules of NH36 is the YPPEFKTKL epitope of the F1 domain, which was also shown to be responsible for IL-10 secretion (21). In this way, this F1 domain-epitope contributes to the lowest IFNγ and TNF-α/IL-10 ratios found in supernatants of splenocytes of NH36-vaccinated mice.

F3sap, and NH36sap vaccinated and infected in response to NH36 or *Leishmania (L.) infantum chagasi* lysate antigens, on day 28 after infection. Bars represent the mean + SE values of two-independent experiments (*n* = 5 mice per treatment in each experiment). Asterisks and horizontal lines show significant differences between treatments as disclosed by Mann–Whitney non-parametrical test.

Some of the evidence gathered in this investigation pointed out the superiority of the F3 vaccine, which correlated with the parasitological results. Among them, we can consider the findings of: (1) higher numbers of purified DCs in F3-vaccinated mice; (2) higher maximal counts of migrated DCs; (3) stronger inhibition of these migrating capabilities *in vitro* and *in vivo* by treatment with anti-CCR7 antibody, and (4) higher expression of CCR7, detected on DCs of vaccinated mice. This shows why, the F3 vaccine protects mice from the defective migration of DCs. The higher expression of CCR7 receptors on DCs and the stronger migration of DCs to the splenic wp, induced by the F3-vaccine, explain the

treatment in each experiment). Asterisks and horizontal lines show significant differences between treatments as disclosed by Mann–Whitney non-parametrical test.

triggering of the most potent CD4-Th1 immune response. This immune response determines the cure of VL, as evidenced by the significant reduction of spleen and liver parasite load and spleen and liver/relative weights, the long-lasting immunity and 100% of survival. On the other hand, NH36-primed DCs, which showed lower expression of CCR7 receptors and consequently lower migration capabilities (11), did not show an effective cure of VL. However, the direct effect of protection offered by F3 vaccination *via* DC-mediated mechanism against other *Leishmania* species, agents of visceral or cutaneous leishmaniasis, has not yet been investigated and this is a limitation.

The CCR7 expression on the surfaces of DCs is mandatory to ensure their correct movement toward lymph nodes. The geographical encounter of DCs and naive T lymphocytes occurs in secondary lymphoid organs (40). This important event is compromised in VL. The main mechanism that promotes the defective localization of DCs is dependent on the IL-10-mediated inhibition of the CCR7 expression. IL-10 reduced migration of DCs by 47% and decreased the DCs CCR7 expression, in *L. (L.) donovani*-infected mice (11). Treatment with anti-IL-10 monoclonal antibody even restored DC migration (11). We recently showed in BALB/c mice that, in contrast to the high IL-10 secretion observed in infected animals, F3-vaccinated and challenged mice showed absolutely no secretion of IL-10 (15, 21). In fact, we described that the FRYPRPKHCHTQVA and KFWCLVIDALKRIG CD4-epitopes of F3, promoted the secretion of TNF-α, but not of IL-10 (21). In the present investigation, secretion of IL-10 was present in infected animals, but was more reduced by the F3 than by the NH36 vaccine, as observed by their respective IFN-γ/IL-10 and TNF-α/IL-10 ratios, which characterize a Th1 response against the infection. These results can partially explain the sustained high CCR7 expression on DCs of thr F3-vaccinated mice and their correct migration capabilities.

In VL, the protective response to VL is compromised by the functional impediment of DCs, which no longer move toward the areas of secondary lymphoid organs, where T cells are located (11). This defect restricts the initial specific-T cell response to *Leishmania* infection (41). Immunization with NH36 and F3 were able to restore the functional impairment of CD11c<sup>+</sup> DCs by inducing CCR7/CCL19-mediated responses and protective immunity in VL. We explored the possible use of mature DCs obtained from mice previously immunized with the C-terminal F3 or the NH36 antigens, to recover protective responses in

treatment in each experiment). Asterisks and horizontal lines show significant differences between treatments as disclosed by Mann–Whitney non-parametrical test.

*L. (L.) infantum chagasi-*infected mice. Our results indicate a clear immunotherapeutic effect in mice receiving the F3 antigenprimed DCs.

The work of Ato et al. (11), showed that normal naïve DCs preserve their migration capabilities and, in this way, help in immunotherapy of VL. However, they used DCs pre-pulsed with LPS and/or antigen to obtain a reduction of the spleen parasite load (11). We now showed that DCs of F3-vaccinated mice enhanced by 58% the migrating capabilities of DCs from normal mice *in vitro,* and without antigen or LPS pre-incubation, and also determined a 65% of reduction of spleen parasite load *in vivo*. While only 70% of DCs of normal naïve mice migrate to the wp of *L (L.) donovani*-infected mice (11), we demonstrated that 84% of DCs from F3-vaccinated mice preferentially migrated to the wp of *L. (L.) infantum chagasi*-infected mice. These effects explain the F3-vaccine usefulness and its higher efficacy in immunotherapy of VL, long-lasting preventive immune response and the 100% survival.

NH36 is a recombinant protein that was first described as the main native glycoprotein GP36 of the FML extract (42). The GP36-glycidic moiety is composed of short chains of 4-O-mannopyranose alternating with 3-O and 4-O-substituted fucopyranose residues (27) and its antigenicity is abolished by treatment with sodium m-periodate (27, 42). We, and others, further identified and cloned the NH36 gene and described its peptide sequence (36, 43). NH36 recombinant protein is a very strong and specific protein diagnostic antigen for human and canine VL (36, 44) and is recognized by mice antibodies (15, 21, 26). We proved that vaccination with the NH36 recombinant protein or DNA protect mice from visceral (15, 16, 18) and cutaneous leishmaniasis (16, 21, 26, 45), and controls canine VL (17). NH36 protein sequence, obtained in *E. coli* is now considered worldwide, as a very potent *Leishmania* antigen and a good candidate for an animal and human vaccine against leishmaniasis (15–21, 26, 37, 38, 45). Recently, the NH36 sequence was analyzed and four N-glycosylation sites represented by asparagine (Q) were found (46), but only two of them of high predictive value. As expected, the further cloning in *Pichia pastoris* enhanced the expressed protein yield but also, apparent molecular weight of NH36, suggesting the presence of high mannose chains (46). However, in order to avoid the glycoside moiety interference to the peptide antigenicity, the asparagine residues were mutated, and still, the un-glycosylated NH36 remained strongly antigenic (46).

Although the GP36 antigen in its native form, exhibits a glycidic moiety (27), it is the NH36 protein that is explored as an extremely successful antigen, and is the tool developed for effective vaccination (15–21, 26, 37, 38, 45). We showed that F3 is the strongest NH36 domain to be used in vaccination and, supporting our strategy, none of the four potential sites of glycosylation of NH36 are located in the F3 domain (amino acids N39, N77, N89, and N189) (46). Therefore, there would be no influence of any carbohydrate native moiety in the native form of the F3 antigen.

Trying to assess which are the receptors for F3 on DCs we previously studied the potential contribution of G-protein-coupled kinin receptors (B2R) in the protective immunity against mice VL induced by the F3-saponin vaccine (47). B2R<sup>−</sup>/<sup>−</sup> and wild type C57BL/6 controls (B2R<sup>+</sup>/<sup>+</sup>) were vaccinated with F3 and saponin, challenged with amastigotes of *Leishmania (L.) infantum chagasi* and euthanized 30 days later. The IDR to leishmanial from B2R<sup>−</sup>/<sup>−</sup> vaccinated mice was lower than in wild type controls. Additionally, a significant decrease of 44.7% of spleen relative weight was noted in vaccinated B2R<sup>+</sup>/<sup>+</sup>, but not in vaccinated B2R<sup>−</sup>/<sup>−</sup> mice, indicating that the B2R of kinin, present on the surface of DCs cells has a partial contribution to the protection against splenomegaly, induced by the F3 vaccine (47). The G protein-coupled bradykinin type 2 receptor (B2R) may couple different classes of G proteins and simultaneously initiate different signal chains, which have extensive cross-talk. In addition, the B2R receptor can generate mitogenic signals that involve the mitogen-activated protein kinases and transactivation of receptor tyrosine kinases (48). Therefore, the signaling pathway activated after recognition of F3 deserves further studies.

Additionally, MHC class II receptors on the surface of DCs of mice are potential targets of recognition of the epitopes of NH36. We recently described the induction of a mixed Th1/Th2 immunity in response to the FMLQILDFYTKVYE of F3, and in contrast, the generation of a main Th1 response, with a predominant TNF-α production and low IL-10 secretion induced by two final CD4-predicted epitopes of F3, FRYPRPKHCHTQVA, and KFWCLVIDALKRIG (21). The FRYPRPKHCHTQVA epitope is also the more potent enhancer of the CD4+TNF-α+, -IFN-γ+, -TNFα+IL-2+, -TNF-α+IFN-γ+, and -IFN-γ+IL-2+ T cell proportions, confirming its capability to raise a specific Th1 response. Additionally, the multifunctional IL-2<sup>+</sup>TNF-α+IFN-γ+-secreting CD4<sup>+</sup> T-cells were raised only, in response to the FRYPRPKHCHTQVA and FMLQILDFYTKVYE (21).

NH36 is a strong phylogenetic marker of the genus *Leishmania*, which exhibits high identity of its amino acid sequence in all studied species of *Leishmania*. Hence, vaccination with NH36 in recombinant protein or DNA forms induced strong prophylactic effect against mice VL due to *L. (L.) donovani* and *L. (L.) infantum chagasi* (15, 16) and tegumentary leishmaniasis due to *L. (L.) amazonensis* (15, 21) and *L (L.) mexicana* (16).

However, recent advances in vaccinology demonstrate that vaccine-induced protection could be enhanced by the identification of the main domains or epitopes of the whole protein. In fact, vaccines that contain the short peptides, which represent the immunogenic epitopes, are able to optimize and even exceed the protective potential induced by the whole cognate protein (15, 21, 26, 49). These short peptide vaccines can also induce universal T cell responses, which are related to many human HLA-DR allotypes and to diverse mice strains (50, 51). This is the rationale of the development of T-epitope vaccines and it was also the guide for the development of the NH36 vaccines (15, 21, 26).

In fact, although 1 M of F3 (13,100 g/l) represents only 38% of 1 M of NH36 (34,240 g/l), its sequence is the most immunogenic in the NH36 molecule. The three CD4<sup>+</sup> T cell epitopes of NH36: FMLQILDFYTKVYE (1,810.13 g/l), FRYPRPKHCHTQVA (1,740.01 g/l), and KFWCLVIDALKRIG (1,620 g/l) correspond to a 42 amino acid sequence totally located in F3. Taken together, 1 M of these 42 amino acids (5,180 g/l) constitutes 39% of the sequence of the F3 peptide (13,100 KDa), which is composed of the last 115 amino acids of NH36, but represents only 15% of the sequence of the whole NH36 protein (34,240 KDa), which is composed of 314 amino acids. Therefore, although the most potent epitopes are also present in NH36, in the F3 domain they are 2.6 times more concentrated. That is why the F3 vaccine exceeds the protective potential of the NH36 whole protein. The superiority of the F3 peptide domains over the NH36 vaccine in prophylaxis was previously observed in several parasitological and immunological variables before and it was calculated according to the following equation = (F3-NH36/ F3) values × 100 = protective effect increment (15, 26). This is a concept related to the purification yield of an active peptide by biochemical techniques. As expected, using that calculation we found a 36% average stronger protective response induced by the F3 vaccine against mice infection by *L. (L.) infantum chagasi* (15) and a 40% enhanced response against challenge by *L. (L.) amazonensis* (26). Accordingly, in this investigation, the F3 vaccine promoted an increment of 80% in the expression of CCR7, of 42% in the migration of DCs to the wp, and of 54% in the reduction of the spleen parasite load after immunotherapy induced by the F3 vaccine.

Although it is true that F3 holds the most important epitopes of NH36, the adaptive immune response starts with the interaction of DCs with lymphocytes, and the reasons why the F3 vaccine induced a higher expression of CCR7 and stronger migration of DCs than NH36 are not yet, fully understood.

F3 epitopes not only interact more than the rest of the NH36 epitopes with the G-protein-coupled kinin receptors (BR2) (47), but they are also more expressed more by the MHC class II receptors (15, 21, 26) and induce a higher expression of CCR7. However, the CCR7 expression alone does not guarantee the correct DC migration toward CCL19 or CCL21 gradients (52). A second signal, mediated by prostaglandin E2 (PGE2), is responsible for DC migration. The PGE2–EP4 axis is involved not only in migration, but also in upregulation of co-stimulatory molecules and increased T cells activation (52), and thus is, crucial for the development of the immune response. In contrast, activation of the liver X receptor (LXR)α interferes with CCR7 expression and migration of DCs resulting in a reduced immune response. PGE2 was recently demonstrated to downregulate LXRα expression in *ex vivo* DCs, by enhancing CCR7 expression and migration of LXR-activated DCs (52). These facts indicate that the F3 peptide and its epitopes, more than the whole NH36 antigen, not only promote the upregulated expression of CCR7, but also probably the PGE2 signaling and stimulation of DCs, which triggers the Th1 immune response against *Leishmania (L.) infantum chagasi*. This hypothesis deserves further studies.

Our findings suggest that DCs can be used in efficacious antigen transportation in vaccination protocols against VL (41). Other infection models using *Mycobacterium tuberculosis* (53), *Chlamydia trachomatis* (54), and other infectious diseases and cancer (55) have shown that immunizations performed with DCs previously incubated with the antigen, as natural adjuvant, can achieve the generation of protective pathogen-specific T cell responses. Although the use of DCs in immunotherapeutic protocols has been considered a promising tool to induce effective immunity against VL (41), the limitations of using DC base therapeutic immunization and prevention have also been extensively discussed (56, 57).

The main contribution of our investigation is the description of the enhancement of CCR7 expression and the migrating DCs capabilities generated by the F3 vaccine. This vaccine property might be an important trigger of the Th1 response and of an immunotherapeutic effect against VL.

#### ETHICS STATEMENT

All animals studies followed the guidelines set by the National Institutes of Health, USA, and the Institutional Animal Care and Use Committee approved the animal protocols (Comissão de Ética no Uso de Animais da Universidade Federal do Rio de Janeiro, CEUA protocol IMPPG040-07/16). All animal experimentation was performed in accordance with the terms of the Brazilian guidelines for the animal welfare regulations. Animals were kept at the Instituto de Microbiologia Paulo de Góes, da Universidade Federal do Rio de Janeiro (UFRJ) facilities, with controlled temperature, 12 h light/dark cycles and given water and feed *ad libitum*. We made all efforts in order to minimize animal suffering.

#### AUTHOR CONTRIBUTIONS

DN, AM, FA, and JM conducted the experiments. DN, AM, CF-d-L, LF-d-L, and FC acquired data. DN, FA, PL, CP-d-S analyzed data. CP-d-S, AM, and AMBM designed research studies. DN and CP-d-S wrote the manuscript. All authors have read and approved the final manuscript.

#### ACKNOWLEDGMENTS

The authors would like to acknowledge Dr. Maria Alice Fusco for kindly providing C57BL/6 mice from the Instituto de Pesquisas Biomédicas – Hospital Naval Marcílio Dias and Professor Leonardo R. Andrade of the Programa de Histologia, Instituto de Ciências Biomédicas, Universidade Federal do Rio de Janeiro, for

#### REFERENCES

1. Yamey G. The world's most neglected diseases. Ignored by the pharmaceutical industry and by public-private partnerships. *BMJ* (2002) 325:176–7. doi:10.1136/bmj.325.7357.176

kindly giving the NucBlue™ Life Ready probes™ Reagent. We also thank David Straker for language review.

#### FUNDING

This work was supported by: Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPQ) [Fellowships, 310977/2014-2 to CP-d-S, 310797/2015-2 to AM and grant 404400/2012-4 to CP-d-S, AM]; by Fundação Carlos Chagas de Amparo à Pesquisa do Estado de Rio de Janeiro (FAPERJ) [grant E-26-201.583/2014, E-26-102957/2011 and E-26/111.682/2013 to CP-d-S and fellowships E-26/102415/2010 and E-26/201747/2015 to DN].

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.00967/ full#supplementary-material.

Figure S1 | Schematic representation of dendritic cells (DCs) purification method and transwell experiment. DCs of mice vaccinated with F3 and challenged with *Leishmania (L.) infantum chagasi* migrate from the upper to the lower chamber of a transwell plate, in response to the CCL19 chemokine gradient. In contrast, DCs from unvaccinated infected mice are unresponsive to CCL19 and do not migrate.

Figure S2 | Strategy for the analysis of multifunctional T cell response using a four-color flow cytometry panel to simultaneously analyze multiple cytokines at the single-cell level in splenocytes cultures. After single cells selection (FSC-A × FSC-H), lymphocytes from a representative mouse immunized with the F3 vaccine were selected according to a FSC-A versus SSC-A dot plot, followed by CD4<sup>+</sup> (A) or CD8<sup>+</sup> (B) gating. Afterward, CD4+ or CD8+ T-cell phenotypes were plotted against each cytokine individually: interferon-γ (IFN-γ), tumor necrosis factor-α (TNF-α), and interleukin (IL)-2. Boolean gating was performed to generate the frequencies of the possible seven combinations of cytokineproducing CD4+ or CD8+ cells using FlowJo V10 software. In this demonstrative figure, we show the results obtained with splenocytes from an animal immunized with the F3 vaccine, 15 days after infection, stimulated "*in vitro*" with the NH36 recombinant protein.

Figure S3 | Cytokines expressed by CD4+ T lymphocytes in response to *Leishmania (L.) infantum chagasi* lysate. Effect of the F3 and NH36-vaccines on the frequencies of CD4+IL-2<sup>+</sup> (A), TNF-α+ (D), IFN-γ+ (G), TNF-α+IL-2<sup>+</sup> (B), TNF-α+IFN-γ+ (E), IL-2+IFN-γ+ (H), and IL-2+TNF-α+IFN-γ+-secreting T cells (C) in response to the promastigote lysate, on day 15 and 28 post challenge. The total CD4+ T cell frequencies are also represented (F). Bars represent the mean + SE values of two-independent experiments (*n* = 5 mice per treatment in each experiment). Asterisks and horizontal lines show significant differences between treatments as disclosed by Mann–Whitney non-parametrical test.

Figure S4 | Cytokines expressed by CD8+ T lymphocytes to *Leishmania (L.) infantum chagasi* lysate. Effect of the F3 and NH36-vaccines on the frequencies of CD8+IL-2<sup>+</sup> (A), TNF-α+ (D), IFN-γ+ (G), TNF-α+IL-2<sup>+</sup> (B), TNF-α+IFN-γ+ (E), IL-2+IFN-γ+ (H), and IL-2+TNF-α+IFN-γ+-secreting T cells (C) in response to the promastigote lysate, on day 15 and 28 post challenge. The total CD8<sup>+</sup> T cell frequencies are also represented (F). Bars represent the mean + SE values of two-independent experiments (*n* = 5 mice per treatment in each experiment). Asterisks and horizontal lines show significant differences between treatments as disclosed by Mann–Whitney non-parametrical test.


**Conflict of Interest Statement:** DN and CP-d-S are the inventors of the patent file PI1015788-3 (INPI Brazil). AM, FA, FC, AMBM, JM, CF-d-L, PL, and LF-d-L, declare no conflict of interest.

*Copyright © 2018 Nico, Martins Almeida, Maria Motta, Soares dos Santos Cardoso, Freire-de-Lima, Freire-de-Lima, de Luca, Maria Blanco Martinez, Morrot and Palatnik-de-Sousa. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

*Kelly A. Prendergast 1,2†, Naomi J. Daniels3 , Troels R. Petersen1†, Ian F. Hermans 1,2,4 and Joanna R. Kirman3,4\**

*1Malaghan Institute of Medical Research, Wellington, New Zealand, 2School of Biological Sciences, Victoria University of Wellington, Wellington, New Zealand, 3Department of Microbiology and Immunology, University of Otago, Dunedin, New Zealand, 4Maurice Wilkins Centre for Molecular Biodiscovery, Auckland, New Zealand*

#### *Edited by:*

*Silvia Beatriz Boscardin, Universidade de São Paulo, Brazil*

*Reviewed by:* 

*Katsuaki Hoshino, Kagawa University, Japan Björn E. Clausen, Johannes Gutenberg-Universität Mainz, Germany*

*\*Correspondence:*

*Joanna R. Kirman jo.kirman@otago.ac.nz*

#### *†Present address:*

*Kelly A. Prendergast, Garvan Institute of Medical Research, Darlinghurst, NSW, Australia; Troels R. Petersen, Royal Society Te Aparangi, Wellington, New Zealand*

#### *Specialty section:*

*This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology*

*Received: 08 December 2017 Accepted: 17 April 2018 Published: 07 May 2018*

#### *Citation:*

*Prendergast KA, Daniels NJ, Petersen TR, Hermans IF and Kirman JR (2018) Langerin+ CD8α<sup>+</sup> Dendritic Cells Drive Early CD8+ T Cell Activation and IL-12 Production During Systemic Bacterial Infection. Front. Immunol. 9:953. doi: 10.3389/fimmu.2018.00953*

Bloodstream infections induce considerable morbidity, high mortality, and represent a significant burden of cost in health care; however, our understanding of the immune response to bacteremia is incomplete. Langerin+ CD8α+ dendritic cells (DCs), residing in the marginal zone of the murine spleen, have the capacity to cross-prime CD8+ T cells and produce IL-12, both of which are important components of antimicrobial immunity. Accordingly, we hypothesized that this DC subset may be a key promoter of adaptive immune responses to blood-borne bacterial infections. Utilizing mice that express the diphtheria toxin receptor under control of the langerin promoter, we investigated the impact of depleting langerin+ CD8α+ DCs in a murine model of intravenous infection with *Mycobacterium bovis* bacille Calmette–Guerin (BCG). In the absence of langerin<sup>+</sup> CD8α+ DCs, the immune response to blood-borne BCG infection was diminished: bacterial numbers in the spleen increased, serum IL-12p40 decreased, and delayed CD8<sup>+</sup> T cell activation, proliferation, and IFN-γ production was evident. Our data revealed that langerin+ CD8α+ DCs play a pivotal role in initiating CD8+ T cell responses and IL-12 production in response to bacteremia and may influence the early control of systemic bacterial infections.

Keywords: langerin, dendritic cell, diphtheria toxin, bacille Calmette–Guerin, systemic infection

#### INTRODUCTION

Bloodstream infections are commonly associated with significant morbidity and mortality, with substantial costs to health-care systems (1). Comprehensive knowledge of the essential constituents of an effective immune response to blood-borne bacterial exposure is, therefore, critical for reducing the associated burden to human health. The spleen is an important site for the induction of adaptive immunity. With extensive vasculature and associated lymphoid tissue, blood-borne bacteria or associated particulate components such as immune complexes are efficiently captured by a network of resident antigen-presenting cells (APCs) that include macrophages, B cells, and dendritic cells (DCs) (2). The acquisition of immunogenic elements and subsequent ligation of pathogen recognition receptors, primarily in the marginal zone of the spleen, drives activation of APCs, with some of these cells then moving to the white pulp to initiate adaptive immune responses. The precise roles of different APCs in promoting immune responses to infectious agents, however, have not been fully elucidated.

Splenic DCs can be broadly classified into conventional or plasmacytoid DCs, with conventional DCs further subdivided into three distinct populations: CD4+ DCs, CD8α+ DCs, and double negative DCs (3, 4). The CD8α+ DC population as a whole are reportedly efficient at cross-presenting antigen to CD8<sup>+</sup> T cells (5, 6) and are major producers of IL-12, which is known to play a key role in promoting differentiation of IFN-γ-producing T cells (7, 8). However, despite their localization and ability to promote both IL-12 and CD8<sup>+</sup> T cell responses, little is known about the importance of CD8α+ DCs in mediating immune responses to systemic bacterial infections.

Confounding the situation, however, it is recognized that there is heterogeneity in function between subpopulations of CD8α+ DCs. A subset of CD8α+ DCs expressing CX3CR1 has been identified, which lack the ability to produce IL-12 and cross-prime CD8<sup>+</sup> T cells and have rearranged immunoglobulin genes—a feature more related to plasmacytoid DCs (9). The CD8α+ DC population can also be subdivided on the basis of expression of the c-type lectin receptor langerin (CD207). The langerin<sup>+</sup> CD8α+ DCs in steady state are predominantly localized in the marginal zone (10, 11), a prime location for sampling bloodstream constituents. They have been shown to take up dying cells from the circulation and move to the T cell areas of the spleen to induce tolerance to acquired cell-associated antigens, a process thought to be involved in maintaining self-tolerance (11). However, in response to protein/adjuvant recognition, it was predominantly this langerin-expressing subset of CD8α+ DCs that produced IL-12 in the spleen, and these cells were critical for the priming of potent CD8<sup>+</sup> T cell responses to circulating antigens (12). Given these characteristics, we hypothesized that the langerin<sup>+</sup> CD8α+ DC subset may be involved in inducing protective responses to systemic infections. To address this hypothesis, we used knockin mice expressing diphtheria toxin receptor (DTR) under the control of the langerin promoter (*lang*-DTREGFP mice) so that diphtheria toxin (DT) could be used to transiently deplete langerin-expressing DCs during blood-borne bacterial infection.

Bloodstream infections can be caused by a wide range of pathogenic microorganisms. *Escherichia coli*, *Staphylococcus aureus*, and *Streptococcus pneumoniae* are the most common causes of community-acquired bloodstream infection (13); however, mycobacterial species are also an important cause of bloodstream infection, particularly in immune-suppressed individuals (14–16). Since studies in mice depleted of CD11c<sup>+</sup> DCs identified a crucial role for splenic DCs in mediating protective adaptive immunity after *Mycobacterium tuberculosis* (*Mtb*) infection (17), we chose to utilize a murine model of intravenous mycobacterial exposure.

To date, there is little information on which subgroups of DCs are important to the antimycobacterial response (18). However, as mice lacking either IFN-γ or IL-12p40 are highly susceptible to infection with *Mtb* (19–23), we considered it likely that the IL-12 producing capabilities of langerin+ CD8α+ DCs would contribute to control of a systemic mycobacterial infection. In addition, the ability of langerin<sup>+</sup> CD8α+ DCs to cross-prime CD8<sup>+</sup> T cells may be important in the context of mycobacterial infection as studies have shown that antigen-specific CD8+ T cells proliferate rapidly and contribute to immunity in the antimycobacterial response (21–24).

We report herein that during intravenous *Mycobacterium bovis* bacille Calmette–Guerin (BCG) infection, the depletion of langerin<sup>+</sup> CD8α+ DCs led to a diminished immune response, with decreased serum IL-12p40 and delayed CD8+ T cell activation, proliferation, and IFN-γ production during infection. An increase in the bacterial burden in the spleen was also evident. These findings suggest that langerin<sup>+</sup> CD8α+ DCs may play an important role in the response to blood-borne bacterial infection.

#### MATERIALS AND METHODS

#### Mice

Male *lang*-DTREGFP (24), *lang*-EGFP (24), and C57BL/6J mice were bred and housed in the Biomedical Research Unit at the Malaghan Institute of Medical Research. Male *lang*-DTREGFP mice crossed with *lang*-EGFP mice (*lang*-DTREGFP × *lang*-EGFP) were used to better visualize GFP expression on langerin<sup>+</sup> cells. OT-I and OT-II mice were crossed with B6.SJL-Ptprca Pep3b / BoyJArc congenic mice to enable cell tracking through the congenic marker CD45.1. All mice were housed under specific pathogen-free conditions. All experiments were undertaken within the provisions of the Animal Welfare Act (1999) of New Zealand and approved by the Victoria University of Wellington Animal Ethics Committee.

#### Mycobacteria

*Mycobacterium bovis* BCG Pasteur strain 1173P2 was grown at 37°C in Dubos broth (Difco, BD Diagnostics Systems, Sparks, MD, USA), supplemented with 10% Middlebrook oleic acid-albumindextrose-catalase (OADC) (Difco), until mid log phase and stored at −80°C in 0.05% PBS Tween80. For recombinant BCG-OVA (25) (a gift from Dr. James Triccas, University of Sydney, NSW, Australia), 50 µg/mL hygromycin (Roche, Manheim, Germany) was added. Before use, defrosted BCG stocks were sonicated briefly prior to dilution in PBS. BCG Pasteur and rBCG-OVA were injected intravenously (i.v.) in the lateral tail vein at 105 CFU per mouse.

# Depletion of Langerin**<sup>+</sup>** CD8**α+** DCs In Vivo

*Lang*-DTREGFP mice were injected i.p. with 350 ng DT (Sigma-Aldrich, St. Louis, MO, USA) every 2 days for the period of time indicated in each experiment, commencing 2 days before BCG infection.

#### Determination of Bacterial Burden

Spleens and livers were homogenized in PBS with 0.5% Tween80 (Sigma-Aldrich) and serial dilutions were plated on 7H11 agar (Difco) supplemented with 10% OADC, 25 mg carbenicillin and 100,000 U polymyxin B (Gibco, Invitrogen, Auckland, New Zealand). Plates were incubated at 37°C and bacterial counts performed after 2–3 weeks growth.

#### Tissue Preparation

Spleens were digested in Liberase TL/DNAse I (Roche) in IMDM (Gibco, Life Technologies) for 30 min at 37°C. Cells were passed through a 70 µm strainer and red blood cells were lysed (Qiagen, MD, USA) before live cells were counted by Trypan blue exclusion.

#### Flow Cytometry

Cells were blocked with anti-CD32/16 (clone 2.4G2, produced in-house) and stained with surface antibodies as indicated; CD8-PE (53-6.7), CD62L-APC (MEL-14), IFNγ-PE Cy7 (XMG1.2), CD11b APC CY7 (M1-70), and GR-1 FITC (RB6-8C5) from BD Pharmingen; B220-A647 (RA3-6B2), CD3-PE Cy7 (145-2C11), CD8-A700 (53-6.7), CD11c-eFluor450 (N418), CD44-PE Cy7 (IM7), CD45.1-PE (A20), Vα2-APC (B20.1) from eBioscience, CD11c-PE Cy7 (N418) from BioLegend. A viability dye, live/ dead fixable blue (Invitrogen), was included before fixing cells with formalin (Sigma-Aldrich). Samples were collected on an LSRII SORP (BD, San Jose, CA, USA). Data were analyzed using FlowJo version 9.6 (Tree Star, Ashland, OR, USA).

#### Adoptive Transfer of Carboxyflourescein Diacetate Succinimidyl Ester (CFSE)- Labeled OT-I and OT-II Cells

Spleens and lymph nodes from OT-I × B6 or OT-II × B6 mice were pooled and labeled with 2.5 µM CFSE (Molecular Probes, Eugene, OR, USA) at 37°C for 10 min. On day −1, 5 × 106 cells were transferred intravenously to *lang*-DTREGFP recipient mice. On day 0, mice received 105 CFU rBCG-OVA intravenously, and spleens were harvested at indicated time points to assess the CFSE proliferation profile by flow cytometry.

# *In Vitro* Re-Stimulation of OT-I Cells

Seven days after rBCG-OVA infection of OT-I transfer recipients, splenocytes were cultured with 1 µg/mL OVA257–265 (SIINFEKL) peptide (GenScript Corporation, Piscataway, NJ, USA) and 2 µg/mL anti-CD28 (clone 37.51, produced in-house) for 6 h at 37°C in complete IMDM (Gibco, Life Technologies), which contained 5% FCS (PAA Laboratories GmbH, Pasching, Austria), 1,000 µg/mL penicillin/streptomycin, 2 mM Glutamax, and 2-Mercaptoethanol (all Gibco, Invitrogen). 2 µM monensin (Sigma-Aldrich) was added for the last 4 h of incubation. Cells were fixed with formalin containing 4% formaldehyde (Sigma-Aldrich) and permeabilized with 0.1% Saponin buffer (Sigma-Aldrich) before being stained for intracellular IFN-γ, which was measured by flow cytometry.

# ELISA

Blood was collected at indicated time points from the lateral tail vein and left overnight to clot. The serum was separated by centrifugation and frozen at −20°C. IL-12p40 and IFN-γ ELISAs were performed following the manufacturer's instructions (BD OptEIA) and the plate was read using a Versamax plate reader (Molecular Devices).

#### Statistics

Bar graphs show mean + SEM error bars. For graphs displaying CFU (log10), the geometric mean + 95% CI is shown. Statistical significance was determined by one-way ANOVA with the Tukey posttest or Kruskal–Wallis test as indicated; significance within groups was determined by two-way ANOVA with the Bonferroni posttest. Graphpad Prism 5 software (Graphpad Software Inc., San Diego, CA, USA) was used for all analyses.

# RESULTS

#### Serum IL-12p40 Is Decreased, and Splenic Bacterial Burden Increased, in the Absence of Langerin**<sup>+</sup>** CD8**α+** DCs in BCG-Infected Mice

To determine if splenic langerin<sup>+</sup> CD8α+ DCs were required for control of systemic BCG infection, we used *lang-*DTREGFP mice (referred to as Lang-DTR mice), which allowed depletion of langerin-expressing cells with DT during the course of infection. Multiple doses of DT were well tolerated and resulted in effective depletion of langerin<sup>+</sup> CD8α+ DCs in the spleen [**Figures 1A,B**; (26)]. Depletion of langerin<sup>−</sup> CD8α+ DCs was not evident (Figure S1 in Supplementary Material). Langerin+ CD8α+ DCs repopulated the spleen within 2–3 days of the final DT treatment (26).

To determine the effect of early depletion of splenic langerin<sup>+</sup> CD8α+ DC on control of systemic blood-borne infection, the bacterial burden in the spleen was assessed over a 10-week period. BCG-infected, Lang-DTR mice were treated with DT from day −2 until day 6 of infection and spleens were harvested at 1, 3, 4, 6, and 10 weeks and then cultured to determine bacterial counts.

In langerin<sup>+</sup> CD8α+ DC-depleted mice, a significant increase in spleen bacterial burden was evident 1 week after infection, compared to non-depleted mice, and this difference was maintained until 3 weeks after infection when spleen bacterial counts peaked (**Figures 1C,D**). In the later stage of infection, from 4 weeks onward, the difference in splenic bacterial burden between depleted and non-depleted mice was insignificant; by 10 weeks after infection, there was no discernible difference (**Figure 1C**). Extended depletion of langerin<sup>+</sup> CD8α+ DCs (DT-treatment sustained until the 3-week peak of bacterial burden) had no further effect on the bacterial load in the spleen (Figure S2 in Supplementary Material).

Depletion of langerin<sup>+</sup> CD8α+ DCs resulted in a small, but statistically insignificant, increase in the bacterial load of the liver compared to non-depleted mice (**Figure 1D**). DT treatment of C57BL/6 mice did not result in an increase in bacterial load in the liver or spleen (**Figure 1D**), confirming that the depletion of langerin<sup>+</sup> CD8α+ DCs, rather than the DT treatment itself, was the cause of the increased bacterial burden during systemic BCG infection. These data may suggest the role of langerin<sup>+</sup> CD8α+ DCs in the control of blood-borne bacterial infection is particularly important in, and potentially limited to, the spleen.

It has been reported in the CD11c-DTR mouse model that neutrophilia occurred after DT treatment in both naïve and bacterially infected mice, suggesting that neutrophilia was induced by DT itself (27). In the present study, an influx of neutrophils into the spleen was also observed early after BCG infection; however, this was significantly higher in DT-treated BCG-infected mice compared to DT-treated uninfected mice, suggesting that neutrophil influx was a response to BCG infection rather than DT treatment (Figure S3 in Supplementary Material). Indeed, uninfected mice treated with DT did not exhibit splenic neutrophilia.

3 weeks after BCG infection (*n* = 10 mice per group). The results are representative of two independent experiments. \**p* < 0.05, \*\*\**p* < 0.001, NS, not significant. (E) Mice were infected with BCG i.v. on day 0. At the times indicated, mice were tail-bled and serum IL-12p40 was measured by ELISA (*n* = 4–8 mice per group). \**p* < 0.05, \*\**p* < 0.01, \*\*\**p* < 0.001, two-way ANOVA between DT and PBS-treated groups, \*\*\**p* < 0.001, one-way ANOVA between uninfected and 3-week PBS-treated groups.

IL-12p40 is essential for immune control of mycobacterial infections (19–21), and as such, the concentration in serum was measured during systemic BCG infection in the presence or absence of langerin<sup>+</sup> CD8α+ DCs. Interestingly, the concentration of IL-12p40 was significantly lower in the serum of uninfected mice that were depleted of langerin<sup>+</sup> CD8α+ DCs (**Figure 1E**), compared to non-depleted mice, suggesting that these DCs contribute to basal IL-12p40 production. After a transient increase in IL-12p40 in both DT-treated and non-depleted mice 6 h after infection, no significant increase in IL-12p40 above basal levels was measured until 3 weeks after infection, at which point the serum concentration was significantly higher in BCGinfected, non-depleted mice compared to uninfected controls; the depletion of langerin<sup>+</sup> CD8α+ DCs resulted in reduced IL-12p40 Prendergast et al. CD8α**<sup>+</sup>** DCs in Bloodstream Infection

production at this time point. Importantly, in C57BL/6 controls, DT treatment did not affect the serum concentration of IL-12p40, which was significantly increased 3 weeks after BCG infection.

To determine if the decreased IL-12p40 levels in infected langerin<sup>+</sup> CD8α+ DC-depleted mice resulted in reduced IFN-γ production, serum IFN-γ was measured, although typically Th1 IFN-γ responses do not develop until 4 weeks after BCG infection (28). As anticipated, IFN-γ was undetectable in the serum early after BCG infection (weeks 1, 2, and 3; data not shown); however, at 4 weeks post BCG infection, when it was detected, the depletion of langerin<sup>+</sup> CD8α+ DCs had no effect on IFN-γ levels in the serum (Figure S4 in Supplementary Material).

# Proliferation of OVA-Specific CD8**+** T Cells in Response to Recombinant BCG-OVA Infection Is Delayed, and Activation Diminished, in the Absence of Langerin**<sup>+</sup>** CD8**α+** DCs

CD8α+ DCs are reportedly the most efficient splenic DC population for cross-presentation of antigen to CD8<sup>+</sup> T cells (5, 6). It is also known that mice deficient in CD8<sup>+</sup> T cells are impaired in their ability to control mycobacterial infections (29–31). Therefore, the proliferation of CD8<sup>+</sup> T cells was assessed in the context of CD8α+ DC depletion. CFSE-labeled transgenic OVAspecific CD8<sup>+</sup> T cells (OT-I cells) were adoptively transferred into Lang-DTR mice, followed by i.v. infection the next day with recombinant BCG that expressed the model antigen ovalbumin (rBCG-OVA). Control animals were infected with non-recombinant BCG, or were left uninfected (OT-I only controls). Animals were then treated with DT every 2 days for the first week of infection, or continuously until the end of the experiment (cont) as indicated, to deplete langerin-expressing cells.

As expected (25), 1 week after rBCG-OVA infection, OT-I cells in the non-depleted recipient mice (rBCG-OVA + PBS) revealed diluted CFSE expression, indicative of proliferation (**Figure 2A**). In contrast, there was little proliferation at this time point in mice depleted of langerin+ CD8α+ DCs. Two weeks after infection, however, OT-I proliferation was diminished in DT-treated mice, and by 3 weeks after infection, OT-I proliferation occurred regardless of whether langerin<sup>+</sup> CD8α+ DCs had been depleted (data not shown). Mice treated continuously with DT for the duration of the experiment did not have significantly different OT-I proliferation than mice depleted for just the first week of infection. These data suggest that depletion of langerin<sup>+</sup> CD8α+ DCs delayed rather than prevented OT-I activation.

Infection with wild-type BCG did not induce significant OT-I proliferation compared to uninfected controls (OT-I only) at any time point assessed, confirming that the T cell activation was antigen-specific. The overall number of OT-I cells, and the percentage of proliferating cells, was significantly lower in the spleen of langerin<sup>+</sup> CD8α+ DC-depleted mice 1 week after rBCG-OVA infection, as well as at 2 weeks after infection in terms of OT-I number, compared to non-depleted mice (**Figure 2B**). Three weeks after infection, however, OT-I cells in the spleen had increased significantly in response to rBCG-OVA infection in both depleted and non-depleted mice, compared to uninfected controls.

The apparently diminished CD8<sup>+</sup> T cell response in the absence of langerin<sup>+</sup> CD8α+ DCs was also reflected in the activation status and IFN-γ production of OT-I cells. One week after rBCG-OVA infection, OT-I cells in langerin+ CD8α+ DC-depleted mice displayed higher CD62L and lower CD44 compared to OT-I cells in non-depleted mice (**Figure 2C**), indicative of reduced activation. In addition, in OT-I cells isolated from mice depleted of langerin<sup>+</sup> CD8α+ DC during infection, the proportion of IFN-γ+ OT-I cells in response to *in vitro* stimulation with OVA257–265 peptide was diminished compared to non-depleted mice (**Figure 2D**). Together, these data suggest that langerin<sup>+</sup> CD8α+ DCs are important for early CD8<sup>+</sup> T cell activation and function after BCG infection.

In analogous experiments, using adoptively transferred CFSElabeled transgenic OVA-specific CD4<sup>+</sup> T cells (OT-II cells), we did not observe any effect of langerin<sup>+</sup> CD8α+ DC depletion on OT-II cell proliferation or the total number of OT-II cells in the spleen (**Figure 3**). As such, additional effects of depletion on these cells were not investigated further in this study.

#### DISCUSSION

Langerin<sup>+</sup> CD8α+ DCs, resident in the marginal zone of the spleen, are localized for effective sampling of the blood; however, to the best of our knowledge, the role of these DCs in protection against systemic bacterial infection is yet to be reported. Here, we show that *in vivo* depletion of langerin<sup>+</sup> CD8α+ DCs during intravenous BCG infection resulted in decreased IL-12p40 in the serum and a delay in antigen-specific CD8<sup>+</sup> T cell proliferation, with reduced activation and IFN-γ production associated. An increased bacterial burden in the spleen was also evident. Together, these results suggest that langerin<sup>+</sup> CD8α+ DCs may play an important role in the immune response against blood-borne bacterial infection.

In similar experimental contexts, several *in vivo* DC-depletion studies have demonstrated protective antimicrobial DC-mediated responses. Depletion of CD11c+ DCs during *S. aureus* bloodstream infection resulted in decreased serum IL-12, concomitant with an increased bacterial load (32). Similarly, bacterial numbers were increased after depletion of CD11c+ DCs during intravenous *Mtb* infection (17). Depletion of dermal langerin<sup>+</sup> DCs in the context of *Leishmania major* infection resulted in a significant reduction in CD8<sup>+</sup> T cell proliferation, with no effect on CD4<sup>+</sup> T cell responses; however, no impact on parasite clearance was demonstrated (33).

By contrast, a number of reports have shown that depletion of DCs enhances, rather than impairs, host defense to the infective organism. A study of *Listeria monocytogenes* infection revealed a direct role for CD8α+ DCs in promoting bacterial disease. In the absence of CD8α+ DCs, bacteria were unable to traffic into the periarteriolar sheath and remained trapped in the marginal zone of the spleen, suggesting that the DCs facilitated entry of these microorganisms into the spleen. Not surprisingly, the absence of CD8α+ DCs during *L. monocytogenes* infection resulted in a reduced bacterial load compared to wild-type animals (34). Depletion of CD11c<sup>+</sup> DCs prior to *Yersinia enterocolitica* infection also led to the surprising finding that animal survival increased; this was discovered to be due to neutrophil accumulation in the spleen following DC depletion, rather than a direct effect of the

Figure 2 | OT-I T cell proliferation in response to rBCG-OVA infection is delayed, and activation diminished, in the absence of langerin+ CD8α+ dendritic cells (DCs). Lang-diphtheria toxin receptor mice received 5 × 106 carboxyflourescein diacetate succinimidyl ester (CFSE)-labeled splenocytes from OT-I × B6 donor mice 1 day before i.v. infection with rBCG-OVA or BCG. One group of mice were not infected (OT-I only group). Mice infected with rBCG-OVA were treated with 350 ng diphtheria toxin (DT) i.p. (or PBS as a control) starting on day −2, every 2 days for the first week of infection, or until the end of the experiment as denoted. At the indicated time points, mice were culled and spleens removed. The CFSE profile of splenic OT-I cells was assessed by flow cytometry. (A) Representative flow cytometry plots of the CFSE profile after gating on CD8+ CD45.1+ Vα2+ OT-I cells; flow plots show the proliferating cells in mice depleted with DT for 1 week. (B) The number of OT-I cells, and the number of proliferating OT-I cells within the spleen are shown, for mice receiving 1 week or continuous DT treatment (cont). (C) One week after rBCG-OVA infection, the median fluorescent intensity (MFI) of CD62L and CD44 expressed on OT-I cells was measured by flow cytometry. (D) One week after rBCG-OVA infection, splenocytes were cultured with 1 µg/mL OVA257–265 peptide for 6 h. Intracellular IFN-γ production by OT-I cells was measured by flow cytometry (*n* = 3–6 mice per group). \**p* < 0.05, \*\**p* < 0.01, \*\*\**p* < 0.001, NS, not significant, one-way ANOVA. The results are representative of two independent experiments.

DCs (35). It appears, therefore, that while DCs are key players in the antibacterial response to blood-borne bacteria, specific DC subsets may be protective or detrimental to the immune response, depending on the infecting bacterial species.

The results are representative of two independent experiments.

The effect of depletion of CD8α+ DCs during systemic mycobacterial infection has not been reported, to date. In terms of the langerin<sup>+</sup> CD8α+ DC subset, we have previously shown the functional specialization of these DCs *in vitro* (12), and *in vivo* (36) in protein/adjuvant models, particularly, with respect to enhanced IL-12 production. CD8α+ DCs, as a whole population, were superior to CD8α− DCs in their ability to induce protective responses against BCG infection in an adoptive transfer setting (37), and have also been implicated as transient producers of IL-12p40 5 h after i.v. BCG exposure (38). In light of these findings, it could be expected that langerin<sup>+</sup> CD8α+ DCs would be important for IL-12p40 production in the context of BCG infection; however, as our data show a transient increase of IL-12p40 6 h after infection in both DT-treated and non-depleted mice, it suggests that perhaps langerin<sup>−</sup> CD8α+ DCs may be the early producers of IL-12p40, before langerin<sup>+</sup> CD8α+ DCs are implicated by 3 weeks after infection.

The role of CD8α+ DCs has also been examined in *Batf3*<sup>−</sup>/<sup>−</sup> mice. These mice lack the Batf3 transcription factor and, reportedly, as a consequence are deficient in CD8α+ DCs (39). However, as we and others have recently discovered, CD8α+ DCs do develop in these mice when bred on a C57BL/6 background under certain, as yet undefined, conditions (36, 40). We suggest that langerin<sup>−</sup> CX3CR1<sup>−</sup> CD8α+ cells that remain in *Batf3*<sup>−</sup>/<sup>−</sup> mice are in fact precursors for the mature langerin+ population. Moreover, others have shown that infection with *Mtb*, *Toxoplasma gondii* or *L. monocytogenes* causes restoration of fully functional CD8α+ cDCs in *Batf3*−/− mice, attributed to compensatory pathways involving the related transcription factors Batf and Batf2 (41). Therefore, the findings reported in this depletion study provide important novel insights into the role of langerin<sup>+</sup> CD8α+ DCs in the immune response to blood-borne mycobacteria.

While we acknowledge that DT treatment also depletes Langerhans cells and langerin<sup>+</sup> dermal DCs (24, 42), these cells do not have access to the spleen. Therefore, they likely play no role in the protection observed in non-depleted, BCG-infected mice. Furthermore, when comparing DT-treated Lang-DTR mice and splenectomized mice, CD8<sup>+</sup> T cell activation in response to antigen and synthetic NKT cell ligand exposure is severely compromised in both models (43), indicating that the presence of the splenic langerin-expressing cells are crucial for promoting effective CD8<sup>+</sup> T cell responses to circulating antigens. However, we cannot entirely discount the possibility that DT treatment led to a decrease in early killing efficiency by macrophages or DCs due to the uptake of apoptotic cells. It is also possible that the observed impact of depleting langerin-expressing cells in the context of BCG infection could have resulted from a reduction in a functional activity that can be attributed to the CD8α+ DC population as a whole. In this situation, it is possible the effect of depleting langerin-expressing cells could, therefore, simply reflect a large reduction [up to 60% (26)] of total CD8α+ DCs in spleens of these mice.

Interestingly, our work revealed that the depletion of langerin<sup>+</sup> CD8α+ DCs lead to reduced CD8+ T cell activation and proliferation during the first week of BCG infection; however, 2 and 3 weeks after infection, CD8<sup>+</sup> T cells proliferated in both depleted and non-depleted mice. As this effect was similarly apparent in mice treated with DT for either the first week or the full duration of the experiment, this suggested that langerin<sup>+</sup> CD8α+ DCs primarily drive CD8<sup>+</sup> T cell proliferation early after BCG infection, and may be redundant after this time. It is possible that other subsets of cross-presenting DCs, such as langerin<sup>−</sup> XCR1<sup>+</sup> DCs, were responsible for stimulating CD8<sup>+</sup> T cells during these later time points, as these cells would not have been depleted in the langerin-DTR mice. In contrast, an increase in the spleen bacterial load was detected whether DT treatment was continued throughout the experiment or only administered for the first week after infection (in the latter case, DCs would have reconstituted during the second week). This suggests the role of langerin<sup>+</sup> CD8α+ DCs in antimycobacterial control may be of particular significance in the first week after infection, and the importance of this early effect on the ensuing immune response is maintained throughout the bacterial proliferation phase.

Although significant insights into the beneficial role of langerin+ CD8α+ DCs in blood-borne bacterial infection are presented in this study, the mechanism of protection was not fully elucidated in this work. Interestingly, numerous reports have shown that *Mtb*-infected mice deficient in IL-12 or CD8<sup>+</sup> T cells have increased bacterial burdens after 3–4 or 4–6 weeks of infection, respectively (20, 29, 31, 44). Therefore, in concordance with our findings, the IL-12 production and CD8<sup>+</sup> T cell stimulation of langerin<sup>+</sup> CD8α+ DCs may represent important aspects of the antibacterial response.

In this study, we report a previously unobserved role for langerin<sup>+</sup> CD8α+ DCs during the initiation of the immune response against systemic mycobacterial infection. The functional heterogeneity of CD8α+ DCs has been underappreciated in the literature to date; thus, the data presented herein provide important insights specifically related to the langerin<sup>+</sup> DC subset. These significant findings provide a platform for further investigations to more conclusively determine the mechanism of protective influence of langerin+ CD8α+ DCs, and whether this extends to other bloodborne bacterial infections.

#### ETHICS STATEMENT

All experiments were undertaken within the provisions of the Animal Welfare Act (1999) of New Zealand and approved by the Victoria University of Wellington Animal Ethics Committee.

# AUTHOR CONTRIBUTIONS

KP planned and performed all experimental work, analyzed and prepared figures, and assisted with writing of the manuscript; ND assisted with figure preparation and wrote the manuscript in consultation with KP and JK; TP assisted with experimental conception and analysis; IH assisted with experimental conception and planning; JK was in charge of overall direction and planning of the project, as well as experimental conception and manuscript preparation.

## ACKNOWLEDGMENTS

We would like to thank Dr. Lindsay Ancelet, Ms. Fenella Rich, and Ms. Clare Burn for technical assistance, the Malaghan Institute Biomedical Research Unit for animal husbandry, and the Cell Technology Suit for flow cytometry support. This work was supported by the Marsden Fund, New Zealand.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.00953/ full#supplementary-material.

Figure S1 | Diphtheria toxin (DT) treatment mediates specific depletion of langerin+ CD8α+ DCs. Mice were treated with 350 ng DT i.p. (or PBS as a control) every 2 days from day −2 to day 6 after bacille Calmette–Guerin (BCG) infection. On day 7 after BCG infection, GFP expression in the spleens of lang-diphtheria toxin receptor × lang-EGFP mice was assessed by flow cytometry after gating on live CD3− B220− CD11c+ cells (*n* = 5 mice per group), as per Figure 1A. (A) Gating strategy for langerin+ CD8α+ and langerin- CD8α+ DC populations. (B) Bar graphs show the mean + SEM percentage of langerin<sup>+</sup> CD8α+ and langerin− CD8α+ DCs in mice treated with DT or PBS. NS, not significant, \*\*\*\**p* < 0.0001, one-way ANOVA. The results are representative of two pooled experiments.

Figure S2 | Extended depletion of langerin+ CD8α+ DCs had no further effect on the bacterial load in the spleen. Mice were treated with 350 ng diphtheria toxin (DT) i.p. (or PBS as a control) every 2 days from day −2 to day 6 after bacille Calmette–Guerin (BCG) infection, or until 3 weeks after BCG infection. On day 0, all groups of mice were infected with BCG i.v. At 3 weeks after BCG infection, mice were culled and spleens removed, homogenized, and plated on 7H11 agar (*n* = 10–11 mice per group). Colonies were counted after 2–3 weeks. Graph shows the geometric mean of spleen bacterial CFU + 95% CI for mice treated with DT for 1 or 3 weeks, or PBS control. NS, not significant, \*\**p* < 0.01, one-way ANOVA. The results are representative of two independent experiments.

Figure S3 | Diphtheria toxin (DT) treatment itself did not induce neutrophilia Lang-diphtheria toxin receptor mice were treated with 350 ng DT i.p. (or PBS as a control) on day −1. Bacille Calmette–Guerin (BCG) infection was carried out on day 0, and mice were culled and spleens removed 2 or 24 h after infection; uninfected mice were culled at the 24 h time point. (A) Representative flow cytometry plots of neutrophil proportions in BCG infected and uninfected mice, treated with DT or PBS. (B) The percentage of neutrophils (CD11b+ GR1high) in the spleens of BCG infected or uninfected mice, treated with DT or PBS is shown (*n* = 3–4 mice per group). NS, not significant, \**p* < 0.05, \*\**p* < 0.01, \*\*\**p* < 0.001, one-way ANOVA.

Figure S4 | Depletion of langerin+ CD8α+ DCs had no effect on IFN-γ levels in the serum 4 weeks after bacille Calmette–Guerin (BCG) infection Lang-diphtheria toxin receptor or C57BL/6 mice were infected with 105 CFU BCG i.v. on day 0 and treated with 350 ng diphtheria toxin i.p. (or PBS as a control) starting on day −2 and continuing every 2 days for 1 week. Four weeks after BCG infection, mice were tail-bled and serum IFN-γ levels were measured by ELISA (*n* = 10 mice per group). NS, not significant, Kruskal–Wallis test. This result is representative of two independent experiments.

# REFERENCES


antigen-specific IFN-gamma responses if IL-12p70 is available. *J Immunol* (2005) 175(2):788–95. doi:10.4049/jimmunol.175.2.788


mycobacterial challenge infection. *PLoS One* (2010) 5(2):e9281. doi:10.1371/ journal.pone.0009281


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Prendergast, Daniels, Petersen, Hermans and Kirman. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

*, Haruko Takeyama2*

*,* 

#### *Edited by:*

*Diana Dudziak, Hautklinik, Universitätsklinikum Erlangen, Germany*

#### *Reviewed by:*

*Sayuri Yamazaki, Nagoya City University, Japan Kristen J. Radford, The University of Queensland, Australia Stella E. Autenrieth, Universität Tübingen, Germany*

> *\*Correspondence: Kazutaka Terahara tera@nih.go.jp*

#### *Specialty section:*

*This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology*

> *Received: 14 January 2018 Accepted: 26 April 2018 Published: 28 May 2018*

#### *Citation:*

*Iwabuchi R, Ikeno S, Kobayashi-Ishihara M, Takeyama H, Ato M, Tsunetsugu-Yokota Y and Terahara K (2018) Introduction of Human Flt3-L and GM-CSF into Humanized Mice Enhances the Reconstitution and Maturation of Myeloid Dendritic Cells and the Development of Foxp3+CD4+ T Cells. Front. Immunol. 9:1042. doi: 10.3389/fimmu.2018.01042*

*Manabu Ato1,3, Yasuko Tsunetsugu-Yokota1,4 and Kazutaka Terahara1 \* 1Department of Immunology, National Institute of Infectious Diseases, Tokyo, Japan, 2Department of Life Science and* 

*Ryutaro Iwabuchi1,2, Shota Ikeno1,2, Mie Kobayashi-Ishihara1*

*Medical Bioscience, Waseda University, Tokyo, Japan, 3Department of Mycobacteriology, Leprosy Research Center, National Institute of Infectious Diseases, Tokyo, Japan, 4Department of Medical Technology, School of Human Sciences, Tokyo University of Technology, Tokyo, Japan*

Two cytokines, fms-related tyrosine kinase 3 ligand (Flt3-L) and granulocyte-macrophage colony-stimulating factor (GM-CSF) are considered to be the essential regulators of dendritic cell (DC) development *in vivo*. However, the combined effect of Flt3-L and GM-CSF on human DCs has not been evaluated *in vivo*. In this study, we, therefore, aimed at evaluating this using a humanized mouse model. Humanized non-obese diabetic/SCID/Jak3null (hNOJ) mice were constructed by transplanting hematopoietic stem cells from human umbilical cord blood into newborn NOJ mice, and *in vivo* transfection (IVT) was performed by hydrodynamic injection-mediated gene delivery using plasmids encoding human Flt3-L and GM-CSF. Following IVT, Flt3-L and GM-CSF were successfully induced in hNOJ mice. At 10 days post-IVT, we found, in the spleen, that treatment with both Flt3-L and GM-CSF enhanced the reconstitution of two myeloid DC subsets, CD14−CD1c+ conventional DCs (cDCs) and CD14−CD141+ cDCs, in addition to CD14+ monocyte-like cells expressing CD1c and/or CD141. GM-CSF alone had less effect on the reconstitution of these myeloid cell populations. By contrast, none of the cytokine treatments enhanced CD123+ plasmacytoid DC (pDC) reconstitution. Regardless of the reconstitution levels, three cell populations (CD1c+ myeloid cells, CD141+ myeloid cells, and pDCs) could be matured by treatment with cytokines, in terms of upregulation of CD40, CD80, CD86, and CD184/CXCR4 and downregulation of CD195/CCR5. In particular, GM-CSF contributed to upregulation of CD80 in all these cell populations. Interestingly, we further observed that Foxp3+ cells within splenic CD4<sup>+</sup> T cells were significantly increased in the presence of GM-CSF. Foxp3+ T cells could be subdivided into two subpopulations, CD45RA−Foxp3hi and CD45RA−Foxp3lo T cells. Whereas CD45RA−Foxp3hi T cells were increased only after treatment with GM-CSF alone, CD45RA−Foxp3lo T cells were increased only after treatment with both Flt3-L and

**39**

GM-CSF. Treatment with Flt3-L alone had no effect on the number of Foxp3+ T cells. The correlation analysis demonstrated that the development of these Foxp3+ subpopulations was associated with the maturation status of DC(-like) cells. Taken together, this study provides a platform for studying the *in vivo* effect of Flt3-L and GM-CSF on human DCs and regulatory T cells.

Keywords: humanized mice, dendritic cells, cytokines, Flt3-L, GM-CSF, T cells, Foxp3

#### INTRODUCTION

Dendritic cells (DCs) play a pivotal role in maintaining the immune responses (1, 2). DCs comprise multiple subsets with distinct functions but can be broadly classified into two major subsets, myeloid DCs [including classical/conventional DCs (cDCs), monocyte-derived DCs (MoDCs), and Langerhans cells] and plasmacytoid DCs (pDCs), on the basis of ontogeny (3–11). Two cytokines, fms-related tyrosine kinase 3 ligand (Flt3-L) and granulocyte-macrophage colony-stimulating factor (GM-CSF), are considered to be the essential regulators of DC development *in vivo*: Flt3-L supports the development of cDCs and pDCs derived from bone marrow (BM) progenitors, while GM-CSF contributes to the development of MoDCs as well as inflammation-induced myeloid DCs (3, 4, 10, 12, 13). One study using knock-out mice showed that combined deficiency of Flt3-L and GM-CSF, rather than a single deficiency of either cytokine, massively reduced not only DCs in the periphery but also monocyte-macrophage DC progenitors and further downstream common DC progenitors in the BM, indicating the concerted action of Flt3-L and GM-CSF on DC homeostasis *in vivo* (13). Cytokines, such as IL-3, IL-4, IL-15, TNF-α, and TGF-β are selectively responsible for the development and maturation of specific DC subsets, which affects the type of immune response that ultimately develops (6, 7). However, in humans, the effect of Flt3-L and GM-CSF singly or in combination in the absence of any other cytokine on the development of DCs remains to be evaluated *in vivo*.

Humanized mice, which are reconstituted with human immune cells, provide an opportunity to study human hematopoiesis *in vivo*. Recent advances in the development of humanized mice have been achieved by using second-generation immunodeficient mouse strains, such as non-obese diabetic (NOD)/SCID/IL2Rγnull (NSG or NOG), NOD/Rag1null/IL2Rγnull (NRG), and BALB/c/ Rag2null/IL2Rγnull (BRG) mice, in all of which the IL-2 receptor common γ-chain is defective, preventing host B, T, and NK cell development, allowing for efficient xenotransplantation (14–16). In addition, xenotransplantation of human hematopoietic stem cells (HSCs) instead of human peripheral blood mononuclear cells into immunodeficient mice enables long-term and multilineage human hematopoiesis (17–21). Our research group has also developed a humanized mouse model using human HSC-transplanted NOD/SCID/Jak3null (NOJ) mice (22–24). NOJ mice, which have an identical phenotype as NSG and NOG mice due to the deficiency of IL-2 downstream molecule Jak3, were developed as an alternative recipient mouse strain for humanization (25). However, some issues with these humanized mouse models still need to be overcome. One of these issues is the limited biologic cross-reactivity of cytokines between mice and humans, which leads not only to insufficient development of human hematopoietic cells, especially myeloid-lineage cells, but also to insufficient human innate and adaptive immunity (14, 15, 26, 27). To overcome this, several approaches for introducing human cytokines have been proposed, including the development of genetically engineered mouse strains, administration of recombinant proteins, and hydrodynamic injection of cytokine gene-encoding plasmids (14, 16, 27). For example, treatment of humanized NSG mice with human GM-CSF and IL-4 by hydrodynamic injection of plasmids successfully enhances DC reconstitution and induced antigen-specific immune responses (28). In addition, humanized NSG-SGM3 mice, into which human stem cell factor, GM-CSF, and IL-3 genes are genetically introduced, display increased human myeloid cells (specifically myeloid DCs) (29). However, it should be noticed that additional cytokines can also affect other immune cell populations, including T cells, and non-physiological exposure to cytokines during T cell development can influence the cell populations that are generated (16, 29). Indeed, human GM-CSF and IL-4-introduced humanized NSG mice accelerated the maturation/activation of CD4<sup>+</sup> T cells (28), and humanized NSG-SGM3 mice displayed skewed development of Foxp3<sup>+</sup> regulatory T cells (Tregs) within the CD4<sup>+</sup> T cell population (29).

In this study, we aimed at evaluating the effect of Flt3-L and GM-CSF singly or in combination in the absence of any other cytokine on the reconstitution and maturation of human DCs *in vivo* using a humanized mouse model. Our humanized NOJ (hNOJ) mice were rather beneficial than other genetically engineered humanized mouse models, in terms of evaluating the effect of exogenous human cytokines. In order to exogenously introduce human Flt3-L and GM-CSF into hNOJ mice, we used the hydrodynamic gene delivery technique, since this is a simple and efficient method to express cytokines in mice (28, 30, 31). The reconstitution and maturation of systemic human DC subsets in hNOJ mice were evaluated following expression of these cytokines *in vivo*. T cell populations were also evaluated in terms of the induction of Foxp3<sup>+</sup> Treg(-like) cells and the development status (naïve/memory) of other T cell subsets.

#### MATERIALS AND METHODS

#### Construction of hNOJ Mice

Human HSCs were isolated from umbilical cord blood using the CD133 MicroBead Kit (Miltenyi Biotec, Tokyo, Japan). Humanized NOJ mice were constructed as described previously (23, 25), with minor modifications. In brief, freshly isolated human HSCs (1–1.5 × 105 cells) were transplanted into the livers of non-irradiated NOJ mice (≤2 days old). Approximately 20 µl of peripheral blood was periodically obtained from the facial vein to determine the extent of chimerism [the percentage of human CD45 (hCD45)<sup>+</sup> cells within total peripheral blood cells]. The individual mice used in this study are listed in Table S1 in Supplementary Material with information on chimerism and the HSC donor ID number. It should be noted that the development of T cells is delayed compared with that of myeloid cells and B cells, and at least 12 weeks is required to see substantial development of T cells in the periphery after transplantation of HSCs into hNOJ mice (23), as in other humanized mouse models (17, 21). Therefore, 15- to 17-week-old hNOJ mice (old mice) were generally used, except for in the experiments, in which 4-week-old hNOJ mice (young mice) were used. All mice were maintained under specific pathogen-free conditions in the animal facility at the National Institute of Infectious Diseases (NIID).

#### IVT of Human Flt3-L and GM-CSF by Hydrodynamic Gene Delivery in hNOJ Mice

The open reading frames for the genes encoding human Flt3-L and GM-CSF (GenBank: NM\_001459.3 and NM\_000758.3, respectively) were subcloned separately into the pEF-BOS-bsr plasmid (32). Plasmid DNA was purified using the NucleoBond Xtra Maxi EF Kit (Macherey-Nagel, Düren, Germany). For hydrodynamic gene delivery, hNOJ mice were intravenously injected with 50 µg of each plasmid in TransIT-QR Hydrodynamic Delivery Solution (Mirus, Madison, WI, USA) within 4 s using a 27-gauge needle. As a control, 50 µg of the empty vector pEF-BOS-bsr plasmid was administered.

#### Measurement of Plasma Cytokines

For this analysis, peripheral blood was collected from the tail vein of hNOJ mice. Human Flt3-L and GM-CSF in the plasma of hNOJ mice was determined by cytometric bead array (CBA) (33). Plasma samples were serially diluted with the Assay Diluent supplied in the BD CBA Human Soluble Protein Master Buffer Kit (BD Biosciences, San Diego, CA, USA). The Human GM-CSF Flex Set (BD Biosciences) was used to measure circulating human GM-CSF. To measure circulating human Flt3-L, an anti-human Flt3-L monoclonal capture antibody (40416; R&D Systems, Minneapolis, MN, USA) was conjugated to beads (Functional Bead A9; BD Biosciences) that have a distinct fluorescent intensity from the human GM-CSF capture bead using the Functional Bead Conjugation Buffer Set (BD Biosciences). A biotinylated anti-human Flt3-L polyclonal antibody (R&D Systems) was used as a detection antibody and was treated with phycoerythrin (PE) conjugated streptavidin (BioLegend, San Diego, CA, USA). Data were collected on a FACSCanto II (BD Biosciences) and analyzed using FCAP Array Software v3.0 (Soft Flow Inc., St. Louis Park, MN, USA) based on the fluorescent intensity of PE. Standard curves were set using recombinant human Flt3-L (R&D Systems) and GM-CSF (supplied in the Human GM-CSF Flex Set). The limit of detection of both human Flt3-L and GM-CSF was <0.01 ng/ml.

### Cell Preparation

Cells were prepared from the peripheral blood, spleen, and BM of hNOJ mice. The peripheral blood was collected from the tail vein before the initiation of IVT and was treated with EDTA-2Na at a final concentration of 5 mM. Splenocytes were prepared at 10 days post-IVT using the Spleen Dissociation Kit mouse (Miltenyi Biotec) and the gentleMACS Dissociator (Miltenyi Biotec). BM cells were prepared by flushing the femurs and tibias of naïve hNOJ mice and hNOJ mice at 10 days post-IVT. Peripheral blood and cells isolated from the spleen or BM were treated with ACK buffer (0.15 M NH4Cl, 1 mM KHCO3, and 0.1 mM EDTA-2Na; pH 7.2−7.4) for 3 and 1 min, respectively, at RT to lyse the red blood cells and then suspended in staining buffer (PBS containing 2% fetal bovine serum and 0.01% sodium azide). For myeloid cell phenotyping by flow cytometry, cells were suspended in PBS containing 0.5% bovine serum albumin and 5 mM EDTA-2Na.

#### Flow Cytometry for Human Leukocytes

The fluorochrome-conjugated monoclonal antibodies used are listed in **Table 1**. All monoclonal antibodies except one that is specific for mouse CD45 (mCD45) were specific for human antigens. An FcR Blocking Reagent (Miltenyi Biotec) was used to prevent non-specific binding of monoclonal antibodies. The Live/Dead Fixable Dead Cell Stain Kit (ThermoFisher Scientific, Waltham, MA, USA) was used for staining dead cells, which were gated out during analysis. All of the cells collected from the peripheral blood samples and 0.5–1 × 106 cells from the BM and spleen were stained with the mixture of fluorochrome-conjugated antibodies, FcR blocking reagent, and the Live/Dead reagent in staining buffer for 30 min on ice. Intracellular staining for Foxp3 was performed for 1 h on ice using the Foxp3 Staining Buffer Set (eBioscience/ThermoFisher Scientific) after cell surface staining. Data from all cells stained were collected on a FACSCanto II or FACSAria III (BD Biosciences). For determining the absolute cell numbers, 20 µl of peripheral blood or 1/1,000 of the cells obtained from the BM and spleen were stained with antibodies specific for human CD45 (hCD45) and mCD45 and the Live/Dead reagent in Trucount tubes (BD Biosciences) for 30 min at RT. After staining, cells were resuspended in buffer (ACK buffer for peripheral blood samples, and staining buffer for BM and spleen samples), and the cells were subjected to flow cytometry without washing. Data were collected on a FACSCanto II until 3 × 104 of the reference beads in the Trucount tubes were acquired. All data were analyzed using FACSDiva software (BD Biosciences) or FlowJo software (LLC, Ashland, OR, USA). The absolute numbers of each cell population were calculated based on the percentages within live CD45<sup>+</sup> cells.

#### Microscopic Analysis

Cells isolated on a FACSAria III were cytospun onto glass slides and stained with the May–Grünwald Stain Solution (Wako Pure Chemical Industries, Osaka, Japan) and Giemsa Stain Solution (Wako Pure Chemical Industries). Slides were visualized on a CKX41 microscope (Olympus, Tokyo, Japan), and images were captured with a Macromax camera (Goko Camera, Kanagawa, Japan).



*a Human CD45. bMouse CD45.*

*c Allophycocyanin. dPeridinin–chlorophyll protein. e Fluorescein isothiocyanate. f Allophycocyanin-cyanin 7. g Phycoerythrin-cyanin 7. h Phycoerythrin-cyanin 5.5. i Brea, CA, USA. j San Diego, CA, USA.*

#### Statistical Analysis

*T*-tests (unpaired, parametric) and Mann–Whitney *U*-tests (unpaired, non-parametric) were used for two-group comparisons. For multiple comparisons, data were analyzed by one-way ANOVA (unpaired, parametric) followed by Holm–Sidak's test (parametric), or by Kruskal–Wallis test (unpaired, non-parametric) followed by Dunn's multiple comparisons test (non-parametric). The Spearman's rank correlation coefficient (non-parametric) was used for correlation analyses. The variance of the data being compared was determined by F-test for two-group comparisons and by the Brown–Forsythe test for multiple comparisons. When the variance was significant, a non-parametric test was used. GraphPad Prism version 6 (GraphPad Software, CA, USA) was used for all analyses. A *P*-value of <0.05 was considered statistically significant.

#### RESULTS

#### Successful Induction of Human Flt3-L and GM-CSF by IVT

Humanized NOJ mice at the steady state (day 0) showed undetectable levels of human Flt3-L and GM-CSF in the plasma (<0.01 ng/ml; **Figure 1**). When hNOJ mice were injected with the Flt3-L-expressing plasmid (Group F), the GM-CSF-expressing plasmid (Group G), or both plasmids (Group F + G), the corresponding cytokines could be induced within 3 days of IVT. The mean concentrations of plasma Flt3-L at day 3 post-IVT were 2,533 and 2,762 ng/ml in Group F and Group F + G, respectively, and those of plasma GM-CSF were 4.5 and 7.6 ng/ml in Group G and Group F + G, respectively. The levels of these cytokines gradually decreased with time, but were detectable for at least 10 days post-IVT. No significant differences in the concentration of either Flt3-L or GM-CSF between the IVT groups were observed at any time. These results indicate that IVT by hydrodynamic injection-mediated gene delivery is a useful method to transiently introduce human cytokines into hNOJ mice.

#### Characterization of Putative DC Populations in hNOJ Mice at the Steady State

Human DCs are well characterized on the basis of cell surface markers including CD1c (for cDC2), CD141 (for cDC1), and CD123 (for pDC) (7, 8, 11, 34–36). However, some monocytelike DC subsets such as "inflammatory" DCs can express CD14 (8, 34). Therefore, although we used cell surface markers of CD1c, CD141, and CD123 to distinguish each putative DC population, we did not deplete CD14<sup>+</sup> cells during cell preparation. Cells at the steady state were prepared from the BM and spleen specimens from naïve hNOJ mice or hNOJ mice that were injected with the empty vector pEF-BOS-bsr plasmid (these mice are referred to as Group E hereafter). To distinguish putative DC populations, CD45<sup>+</sup>CD3<sup>−</sup>CD19<sup>−</sup> cells were divided into CD123<sup>+</sup>CD33<sup>+</sup>/<sup>−</sup> population (Population 3) and CD123<sup>+</sup>/<sup>−</sup>CD33<sup>+</sup> myeloid cells. CD123<sup>+</sup>/<sup>−</sup>CD33<sup>+</sup> myeloid cells were subdivided into CD1c<sup>+</sup> population (Population 1) and CD141<sup>+</sup> population (Population 2) (**Figure 2A**). May–Grünwald and Giemsa staining revealed the typical morphologies for each DC subset (20, 35): Population 1 and Population 2 exhibited a similar morphology to cDCs, with a less round shape and multilobulated nuclei, whereas Population 3 exhibited a similar morphology to pDCs, with a round shape and indented nuclei (**Figure 2B**). We next analyzed the expression profiles of subset-associated markers (HLA-DR, CD11c, CD303, CD4, CD11b, and CD14) (**Figure 2C**). All three putative DC populations expressed HLA-DR, a defining feature of antigen-presenting cells (37). As with all of DC subsets in humans (7, 35, 36, 38), all three populations in hNOJ mice expressed CD4. CD11c and CD303 are used as distinctive markers for human cDCs and pDCs, respectively (7–9, 11, 35, 36), and myeloid cell populations (Population 1 and Population 2) and Population 3

in hNOJ mice could clearly be distinguished by these markers. Population 1 and Population 2, but not Population 3, in hNOJ mice expressed CD11b and CD14, though the expression of CD11b was more evident in Population 1 than in Population 2. These results indicate that Population 3 in hNOJ mice was phenotypically identical to pDCs in human blood. By contrast, Population 1 and Population 2 were heterogeneous, consisting of CD14<sup>−</sup> genuine cDCs and CD14<sup>+</sup> monocyte-like cells.

#### Assessment of CD14**+** Monocyte-Like Cells in Putative DC Populations Following Cytokine Induction

We next asked how much CD14<sup>+</sup> monocyte-like cells were included in each putative DC population in hNOJ mice following cytokine induction. On flow cytometry, it should be noted that CD1c and CD141 double-positive cells were often observed within myeloid cell compartments, especially in the presence of Flt3-L (Figure S1 in Supplementary Material), and these cells were counted as CD141<sup>+</sup> cells, as described elsewhere (39). An obvious finding was that CD14+ monocyte-like cells were significantly enriched in Population 1 in the BM after treatment with GM-CSF alone (**Figure 3A**, upper left). Although Population 3 in the BM also involved increased CD14<sup>+</sup> monocyte-like cells when Flt3-L was induced, the percentage of CD14<sup>+</sup> monocyte-like cells was minimal (<5%) (**Figure 3A**, upper right). Apart from these cases, the ratios of CD14+ monocyte-like cells in every cytokineinduced IVT group were similar or decreased compared with those at the steady state in the BM and spleen (Group E). We further compared the level of CD14 expression among Population 1, Population 2, and CD1c<sup>−</sup>CD141<sup>−</sup> myeloid cells. Notably, whereas CD1c<sup>−</sup>CD141<sup>−</sup> myeloid cells included cells expressing higher level of CD14, the level of CD14 expression in Population 1 and Population 2 in the BM and spleen was low or intermediate in any cytokine-induced IVT group (**Figure 3B**). Since three types (classical, intermediate, and non-classical) of monocytes are defined in human blood and CD14 expression level of classical and intermediate monocytes is higher than that of non-classical monocytes (40), CD14<sup>+</sup> monocyte-like cells within Population 1 and Population 2 would be separated from classical and intermediate monocytes.

#### Enhanced Reconstitution of DCs Following Cytokine Induction

Given that CD14<sup>+</sup> monocyte-like cells substantially existed in any IVT group (**Figure 3**), each putative DC population was subdivided into CD14<sup>−</sup> DCs and CD14<sup>+</sup> monocyte-like cells to assess the effect of Flt3-L and GM-CSF on the reconstitution of each cell population in hNOJ mice. The absolute cell numbers and the percentages of each cell population were measured in the BM and spleen, and the reconstitution levels were compared among the IVT groups (**Figure 4**). This percentage would be informative if there was individual variability in the reconstitution levels of human leukocytes in the BM and spleen prior to IVT; however, chimerism, as determined by the percentage of hCD45<sup>+</sup> cells in the peripheral blood population at the initiation of IVT, was not significantly different among the IVT groups (Figure S2 in Supplementary Material).

In the BM (**Figure 4A**), GM-CSF seemed to be responsible for the increased reconstitution of Population 1 and, surprisingly, Population 3, but not Population 2. In other words, the absolute cell numbers and/or the percentages of Population 1 and Population 3 were increased in the presence of GM-CSF. However, the increased reconstitution of Population 1 after treatment with GM-CSF alone could be attributed to that of CD14<sup>+</sup> monocyte-like cells, but not CD14<sup>−</sup>CD1c<sup>+</sup> cDCs. Indeed, CD14<sup>+</sup> monocyte-like cells were significantly enriched in Population 1 after treatment with GM-CSF alone (**Figure 3A**). On the other hand, Population 3 after treatment with GM-CSF alone involved the increased reconstitution of both CD14<sup>−</sup> pDCs and CD14<sup>+</sup> monocyte-like cells, though the contamination of CD14<sup>+</sup> monocyte-like cells into Population 3 was a negligible level (<4%) (**Figure 3A**). In contrast to GM-CSF, Flt3-L was required for the increased reconstitution of Population 2, and this could be substantially attributed to that of CD14<sup>−</sup>CD141<sup>+</sup> cDCs. Interestingly,

whereas an additive effect of Flt3-L and GM-CSF was observed on the reconstitution of both CD14− cDCs and CD14+ monocytelike cells in Population 1 and Population 2, this was not true for pDC (Population 3) reconstitution.

The reconstitution profiles in the spleen (**Figure 4B**) differed from what was observed in the BM, likely due to egression/ immigration and/or the site-specific milieu responsible for cell maintenance. An obvious difference between the two organs was observed with respect to Population 1: Flt3-L increased the reconstitution of this population including both CD14<sup>−</sup> cDCs and CD14<sup>+</sup> monocyte-like cells in the spleen, but not the BM. In addition, few pDCs (Population 3) were expanded in the spleen in any IVT group.

#### Assessment of the *In Vivo* Effect of Flt3-L on the Reconstitution of pDCs Using Young hNOJ Mice

Whereas Ding et al. showed that treatment with Flt3-L contributes to robust expansion of pDCs as well as CD1c<sup>+</sup> cDCs and CD141<sup>+</sup> cDCs in the BM and spleen of humanized NOD/SCID mice (39), in our study, pDCs (Population 3) were not expanded by treatment with Flt3-L (**Figure 4**). Since Ding et al. treated mice with the cytokine earlier at 4 weeks after HSC transplantation (39), we evaluated the *in vivo* effect of Flt3-L in younger hNOJ mice. Four-week-old hNOJ mice were injected with either the Flt3-L-expressing plasmid (Group yF) or the empty vector (Group yE). Both pDCs (Population 3) and Population 1 significantly expanded in the BM and spleen in response to treatment with Flt3-L, while Population 2 did not (**Figure 5**). Interestingly, as shown in the previous experiment (**Figure 4**), an inverse pattern of expansion had been observed between CD141<sup>+</sup> myeloid cells and pDCs. These results suggest that unknown age-related factors are involved in the differential developmental regulation of CD141<sup>+</sup> cDCs and pDCs.

### Comparison of BM Hematopoietic Progenitor Populations Between the Young and Old hNOJ Mice

We further investigated the populations of BM hematopoietic progenitors in the young and old hNOJ mice that were injected

with the empty vector at 4 or 16 weeks of age, respectively. According to previous reports (41–43), hematopoietic progenitors within hCD45<sup>+</sup>CD34<sup>+</sup> BM cells were divided into four populations in this study: CD38<sup>−</sup>CD45RA<sup>−</sup> HSCs/multipotent progenitors (MPPs), CD38<sup>−</sup>CD45RA<sup>+</sup>CD116<sup>−</sup> multi-lymphoid progenitors (MLPs)/common lymphoid progenitors (CLPs), CD38<sup>+</sup>CD45RA<sup>−</sup>CD123lo common myeloid progenitors (CMPs), and CD38+CD45RA+CD123lo granulocyte-macrophage progenitors (GMPs) (**Figures 6A,B**). When the frequencies of these populations were compared between the young and old hNOJ mice (**Figure 6C**), the old hNOJ mice showed higher frequencies of myeloid-lineage progenitors (CMPs and GMPs) than the young hNOJ mice. Although the old hNOJ mice tended to have a higher frequency of HSCs/MPPs than the young hNOJ mice, this difference was not significant. By contrast, there was a similar frequency of lymphoid-lineage progenitors (MLPs/CLPs) in the young and old hNOJ mice. Furthermore, when CD135/ Flt3 expression on these hematopoietic progenitors was compared between the young and old hNOJ mice (**Figures 6D,E**), there was a higher frequency of CD135/Flt3<sup>+</sup> HSCs/MPPs in the young hNOJ mice than in the old. By contrast, CD135/Flt3<sup>+</sup> MLPs/CLPs were more abundant in the old hNOJ mice, while there were similar frequencies of CD135/Flt3<sup>+</sup> CMPs and GMPs between the two groups. Although it remains unclear why the

*in vivo* effect of Flt3-L differed according to age, these findings may partly explain the age-related differences in hNOJ mice with respect to the sensitivity to Flt3-L.

#### Enhanced Maturation of Putative DC Populations Following Cytokine Induction

We next examined whether treatment with cytokines affected the maturation status of each putative DC population in hNOJ mice. Since upregulation of CD40, CD80, CD86, and CD184/CXCR4 and downregulation of CD195/CCR5 are associated with DC maturation (44, 45), the expression of these markers was examined on splenocytes by flow cytometry. Because it was difficult to clearly distinguish positive and negative populations using some markers, and in some cases, background levels of fluorescence (staining with the isotype control) were different among the IVT groups (**Figure 7A**), the normalized mean fluorescence intensity (nMFI) was used as a quantitative measure of the expression of each marker (nMFI = test marker MFI/isotype MFI) (**Figure 7B**). Population-specific differences in the expression of maturation-associated markers were observed in the different IVT groups. Details are described below.

#### Population 1 (CD1c**+**)

CD40 was highly expressed even at the steady state (Group E), while the expression of CD184/CXCR4 was upregulated by cytokine treatment. The expression of CD80 was significantly upregulated by treatment with both Flt3-L and GM-CSF (Group F + G). Although the expression of CD86 was relatively low compared with CD80, a significant upregulation was also observed in Group F + G. CD195/CCR5, like CD40, was substantially expressed at the steady state (Group E), and downregulation was observed in response to treatment with GM-CSF alone (Group G).

#### Population 2 (CD141**+**)

The expression patterns of CD40 and CD184/CXCR4 in Population 2 were similar to those in Population 1. However, potent upregulation of CD80 was observed only after treatment with GM-CSF alone (Group G), and CD86 was not upregulated by any cytokine treatment. Furthermore, although CD195/ CCR5 expression was lower at the steady state (Group E) than in Population 1 and Population 3, downregulation of CD195/CCR5 was observed in Group G, as with Population 1.

#### Population 3 (CD123**+**)

In contrast to Population 1 and Population 2, the expression of CD40 was upregulated by treatment with Flt3-L and/or GM-CSF. CD80 was hardly expressed at the steady state, but was also upregulated by GM-CSF alone (Group G) or in combination with Flt3-L (Group F + G). CD86 expression was significantly increased only after treatment with the combination of Flt3-L and GM-CSF (Group F + G). Whereas treatment with the combination of Flt3-L and GM-CSF upregulated the expression of CD40, CD80, and CD86, this was not the case for CD184/ CXCR4 expression: upregulated expression of CD184/CXCR4 was observed only in response to single treatment with either Flt3-L (Group F) or GM-CSF (Group G). CD195/CCR5 was substantially expressed at the steady state. Although its expression might be downregulated in response to single treatment with GM-CSF (Group G), the level was not significant.

Figure 6 | Composition of hematopoietic progenitors in hNOJ mice. The young (4-week-old) and old (16-week-old) hNOJ mice were subjected to *in vivo* transfection (IVT) with the empty vector plasmid. At 10 days post-IVT, bone marrow (BM) cells were analyzed by flow cytometry. (A) Schematic view of human dendritic cell hematopoiesis. (B) Identification of BM hematopoietic progenitors in hNOJ mice: hematopoietic stem cell (HSC)/MPP (CD34+CD38−CD45RA−), MLP/CLP (CD34+CD38−CD45RA+CD116−), CMP (CD34+CD38+CD45RA−CD123lo), and GMP (CD34+CD38+CD45RA+CD123lo). (C) Comparison of the frequency of each hematopoietic progenitor population between the young and old hNOJ mice. Data are the individual values (young: *n* = 5, old: *n* = 4). Significant differences (\**P* < 0.05) were determined by the Mann–Whitney *U* test. (D) Representative histograms of CD135/Flt3 and isotype staining. (E) Comparison of the frequency of CD135/Flt3+ cells within each hematopoietic progenitor population between the young and old hNOJ mice. Significant differences (\**P* < 0.05) were determined by an unpaired *t*-test.

Collectively, our data demonstrate that the maturation of each population was enhanced by treatment with cytokines irrespective of the level of reconstitution. Specifically, combined treatment with Flt3-L and GM-CSF resulted in increased expression of the essential co-stimulatory molecules B7-1 (CD80) and B7-2 (CD86) by Population 1 and Population 3 in hNOJ mice. However, the influence of CD14<sup>+</sup> monocyte-like cells that were substantially included in Population 1 and Population 2 should be reminded.

#### Altered T Cell Subpopulations Following Cytokine Induction

A humanized mouse model using the NSG-SGM3 strain, in which human stem cell factor, GM-CSF, and IL-3 are expressed, displayed not only increased reconstitution of human myeloid DCs but also skewed development of Foxp3<sup>+</sup> Tregs (29). Therefore, we extended our flow cytometric analysis to the detection of Foxp3+CD4+ T cells in the spleen. We further subdivided Foxp3<sup>+</sup>CD4<sup>+</sup> T cells into three subpopulations: resting Tregs (CD45RA<sup>+</sup>Foxp3lo), activated Tregs (CD45RA−Foxp3hi), and non-Tregs (CD45RA<sup>−</sup>Foxp3lo) (**Figure 8A**), as previously reported (46). Although the percentages of Foxp3<sup>+</sup>CD4<sup>+</sup> T cells were increased in the presence of GM-CSF (Groups G and F + G), there were differences in the subpopulations of Foxp3<sup>+</sup> cells within each group (**Figures 8A,B**). Whereas CD45RA<sup>−</sup>Foxp3lo non-Tregs were only significantly increased in Group F + G (22.5 ± 6.7% of CD4<sup>+</sup> T cells, **Figure 8B**, middle panel), CD45RA<sup>−</sup> Foxp3hi activated Tregs were only significantly increased in Group G (27.8 ± 2.5% of CD4<sup>+</sup> T cells, **Figure 8B**, right panel). CD45RA<sup>+</sup>Foxp3lo resting Tregs were rarely observed in any of the IVT groups (**Figure 8B**, left panel).

Furthermore, to examine the differentiation status of the other T cell subsets (Foxp3<sup>−</sup> conventional CD4<sup>+</sup> T cells and CD8<sup>+</sup> T cells), the T cells were divided into three subpopulations based on the expression patterns of CD45RA and CD27, as previously reported (47): naïve (CD45RA<sup>+</sup>CD27<sup>+</sup>), central memory (CM; CD45RA<sup>−</sup>CD27<sup>+</sup>), and effector memory (EM; CD45RA<sup>−</sup>CD27<sup>−</sup>) populations (**Figure 8C**, upper panels). There was a large amount of variability in the percentages of each subpopulation, and no significant changes were noted among the IVT groups (**Figure 8C**, lower panels). Nevertheless, CD8<sup>+</sup> T cells consisted of substantial proportions of naïve and CM cells in all IVT groups (**Figure 8C**, lower right panel).

#### Correlation Between the Maturation Status of Putative DC Populations and the Development of Foxp3**+**CD4**+** T Cells Following Cytokine Induction

Our data indicated that treatment with GM-CSF alone preferentially contributed to the enhanced development of CD45RA<sup>−</sup>Foxp3hi activated Tregs in hNOJ mice (**Figure 8B**). However, GM-CSF alone had less impact on the reconstitution of putative DC populations including both CD14− cDCs and CD14+ monocytelike cells in the spleen (**Figure 4**). We, therefore, investigated the relationship between the maturation status of putative DC populations and the development of CD45RA<sup>−</sup>Foxp3hi activated Tregs, and found that the expression of CD80 and CD86 as well as CD184/CXCR4 in every putative DC population positively correlated with the percentage of CD45RA<sup>−</sup>Foxp3hi activated Tregs (**Figure 9**). These results indicate that the development of CD45RA<sup>−</sup>Foxp3hi activated Tregs was associated with DC(-like cell) maturation, as characterized by the expression of costimulatory molecules (CD80 and CD86). Furthermore, when the percentages of CD45RA<sup>−</sup>Foxp3lo non-Tregs were compared, a positive correlation was observed with the expression of some maturation-associated markers. However, it was only in Population 1 that the levels of CD80, CD86, and CD184/CXCR4 all correlated with the levels of Foxp3<sup>+</sup> non-Tregs (Figure S3 in Supplementary Material), suggesting that the development of CD45RA<sup>−</sup>Foxp3lo non-Tregs might be influenced by Population 1 rather than Population 2 and Population 3.

#### DISCUSSION

In this study, we asked how human Flt3-L and GM-CSF affected the reconstitution and maturation of DCs and the cellularity of T cells *in vivo* using hNOJ mice. We show that IVT by hydrodynamic injection-mediated gene delivery is a useful method to transiently introduce these cytokines into hNOJ mice. When comparing the concentrations of Flt3-L and GM-CSF induced by IVT, the Flt3-L concentration was much higher than the GM-CSF concentration even though the same expression vector (containing the same EF-1α promoter) was used to induce both cytokines. However, even the GM-CSF concentration in hNOJ mice was supraphysiological, since Flt3-L and GM-CSF levels are barely detectable in the circulation [Flt3-L: <100 pg/ ml (48) and GM-CSF: <10 pg/ml (49)] in humans at the steady state. Furthermore, the Flt3-L and GM-CSF concentration was almost 100 times higher than in humanized NSG mice injected with Flt3-L DNA (31) or in humanized NSG-SGM3 mice in which GM-CSF is stably expressed (29). Although both the Flt3-L and GM-CSF concentrations were substantially decreased at 10 days post-IVT, we conducted *ex vivo* analyses of DCs and T cells at that time because the cytokine levels were still detectable and because the experiments by Chen et al. were conducted at 9 days post-injection of plasmids, when the concentrations of the introduced cytokines were also substantially decreased (31).

In hNOJ mice, three human DC subsets (CD1c<sup>+</sup> cDCs, CD141<sup>+</sup> cDCs, and pDCs) were reconstituted, as observed in other humanized mice (20, 50). Further phenotyping on the basis of the expression of HLA-DR, CD11c, CD303, CD4, and CD11b confirmed that all of these DC subsets were phenotypically similar to their equivalents in human blood. However, it should be noted that the pDC population might include pre-cDCs, since pre-cDCs and pDCs have similar expression of CD33 and CD123, as reported recently by See et al. (51), though this pDC population was not increased in the spleen when CD1c<sup>+</sup> cDCs or CD141<sup>+</sup> cDCs were increased. In addition to these DC subsets, we found that CD14<sup>+</sup> monocyte-like cells substantially expressed CD1c and/or CD141 even at the steady state and that they were certainly increased after treatment with both Flt3-L and GM-CSF. The level of CD14 expression by CD14<sup>+</sup> monocyte-like cells was

cytometric analysis for T cell phenotyping. (A) A representative gating strategy for resting regulatory T cells (Tregs) (CD45RA+Foxp3lo), non-Tregs (CD45RA−Foxp3lo), and activated Tregs (CD45RA−Foxp3hi) within Foxp3+CD4+ T cells. The histogram shows higher expression of Foxp3 in Group G. (B) Comparison of the percentages of Foxp3+CD4+ T cell subsets among the IVT groups. Significant differences (\**P* < 0.05, \*\*\**P* < 0.001) were determined by one-way ANOVA followed by the Holm– Sidak's multiple comparisons test (Group E: *n* = 10, Group F: *n* = 6, Group G: *n* = 4, Group F + G: *n* = 10). (C) A representative gating strategy for conventional CD4+ T cells and CD8+ T cells and differentiation stages of each T cell population. Significant difference (\**P* < 0.05) was determined by the Kruskal–Wallis test followed by the Dunn's multiple comparisons test (Group E: *n* = 10, Group F: *n* = 6, Group G: *n* = 4, Group F + G: *n* = 10).

(Population 2), and CD123+ population (Population 3) (Figure 7B), and the percentages of CD45RA−Foxp3hi activated Tregs (Figure 8B) were plotted (total *n* = 25, consisting of Group E: *n* = 9; Group F: *n* = 6; Group G: *n* = 4; Group F + G: *n* = 6). The Spearman's rank correlation coefficient was used for statistical analysis.

low or intermediate, suggesting a similar phenotype of nonclassical monocytes (40). By contrast, it is known that putative MoDCs such as CD14<sup>+</sup> DCs and inflammatory DCs can express CD1c and CD141 (8, 34, 52). Interestingly, a recent single-cell RNA-seq analysis demonstrated that in human blood one of the CD1c<sup>+</sup> DC subsets, "inflammatory" CD1c<sup>+</sup> DCs (also designated CD1c<sup>+</sup> B DCs), do not express CD14 on their cell surface but do express CD14 mRNA (53). Furthermore, in another humanized mouse model, it has been shown that CD1c<sup>+</sup> cDCs in the BM contain both "non-inflammatory" (CD1c+ A) and "inflammatory" (CD1c<sup>+</sup> B) DC subsets and that CD1c<sup>+</sup> B DC-associated inflammatory markers including CD14 mRNA are upregulated after *in vivo* activation with TLR ligands, poly I:C, and/or R848 (50). Therefore, it is possible that cDCs, especially CD1c<sup>+</sup> cDCs, could express CD14, depending on the tissue milieu. However, in this study, it was difficult to define whether CD14<sup>+</sup> monocyte-like cells were categorized into monocyte, MoDC, or CD1c<sup>+</sup> B DC populations. New technologies developed in recent years such as single-cell RNA-seq or CyTOF would be helpful for characterizing CD14<sup>+</sup> monocyte-like cells in hNOJ mice.

Nevertheless, we show here the effect of *in vivo* expression of Flt3-L and GM-CSF on the reconstitution of CD1c<sup>+</sup> cDCs, CD141+ cDCs, and pDCs. Although the effect on the DC reconstitution varied across the subsets and organs investigated, the introduction of both Flt3-L and GM-CSF reliably resulted in myeloid DC-rich hNOJ mice. On the other hand, pDCs failed to expand in the spleen in response to any of the cytokine treatments studied, despite the expected effect of Flt3-L on pDC reconstitution *in vivo* (12, 39) and despite the possible pre-cDCs within this population (51). Although it has been reported that GM-CSF impairs Flt3-L-induced pDC generation from BM progenitors in mice *in vitro* (54), pDCs failed to expand in hNOJ mice even in response to Flt3-L alone, indicating that this effect was independent of GM-CSF. However, pDCs did expand in the young hNOJ mice treated with Flt3-L alone though the possible contamination of pre-cDCs could not be excluded. These findings also suggest another possibility that the composition of hematopoietic progenitors might be affected by aging, i.e., by the amount of time after HSC transplantation. Human cDCs and pDCs arise independently of lineage commitment, in contrast to murine DC hematopoiesis (41, 55). With respect to myeloid-lineage progenitors (CMPs and GMPs), the frequency of CD135/Flt3<sup>+</sup> cells was independent of aging in hNOJ mice. By contrast, higher frequencies of CD135/Flt3<sup>+</sup> lymphoid progenitors (MLPs and CLPs) were detected in the old hNOJ mice, despite the fact that pDCs did not expand in response to Flt3-L treatment in these hNOJ mice. This might be in part due to the absence of human IL-3 in hNOJ mice, since IL-3 is required for the generation and survival of pDCs (56, 57). Interestingly, we found that when CD1c<sup>+</sup> cDCs were expanded by treatment with certain cytokines, either CD141<sup>+</sup> cDCs (or myeloid cells) or pDCs were expanded, but not both. However, these findings should be required close attention, since *in vitro* culture of human CD34<sup>+</sup> HSCs with Flt3-L generates CLEC9A<sup>+</sup> DCs, but they lack CD141 expression (58). Since it has been demonstrated recently that CLEC9A is a perfect discriminative surface marker for cDC1 (53), this marker should be helpful in the future study. Although the developmental regulation of each DC subset by Flt3-L and GM-CSF in hNOJ mice remains unclear, age-related unknown factors might be involved in the underlying mechanisms.

Dendritic cells play an essential role in the induction of not only immunity but also tolerance, and the maturation of DCs is considered to be crucial for the induction of T cell immunity (1). However, it has been suggested that treatment with Flt3-L alone may not be sufficient to generate fully functional DCs (10). Indeed, this study demonstrated that the effect of Flt3-L alone on the maturation in hNOJ mice was limited to, for example, CD184/CXCR4 upregulation on CD1c<sup>+</sup> myeloid cells and pDCs. By contrast, GM-CSF alone or in combination with Flt3-L upregulated the expression of CD80, one of the co-stimulatory molecules, in all DC(-like) subsets. Strikingly, splenic Foxp3<sup>+</sup>CD4<sup>+</sup> T cells preferentially expanded in hNOJ mice in the presence of GM-CSF. This finding is in agreement with earlier studies in mouse models: humanized NSG-SGM3 mice, in which human GM-CSF is stably expressed, showed skewed development of human Foxp3<sup>+</sup> Tregs (29), and NOD mice, an animal model for type 1 diabetes, showed expansion of mouse Foxp3<sup>+</sup> Tregs after treatment with mouse GM-CSF but not with mouse Flt3-L (59).

Human Foxp3+CD4+ T cells can be subdivided into three subpopulations on the basis of the expression of CD45RA and Foxp3: CD45RA<sup>+</sup>Foxp3lo resting Tregs, CD45RA−Foxp3hi activated Tregs, and CD45RA<sup>−</sup>Foxp3lo non-Tregs (46). Whereas CD45RA+Foxp3lo resting Tregs and CD45RA−Foxp3hi activated Tregs are suppressive, CD45RA<sup>−</sup>Foxp3lo non-Tregs are not suppressive, and are the highest producers of IL-17 among whole CD4<sup>+</sup> T cells, suggestive of a T helper (Th)17 phenotype (46). Foxp3<sup>+</sup>CD4<sup>+</sup> T cells in hNOJ mice consisted primarily of CD45RA<sup>−</sup>Foxp3hi activated Tregs and CD45RA<sup>−</sup>Foxp3lo non-Tregs. Remarkably, the developmental regulation of the two subpopulations differed, as CD45RA<sup>−</sup>Foxp3hi activated Tregs were only increased after treatment with GM-CSF alone, whereas CD45RA<sup>−</sup>Foxp3lo non-Tregs were only increased after treatment with both Flt3-L and GM-CSF. This differential regulation by Flt3-L has not been addressed elsewhere. Because the increased Foxp3<sup>+</sup>CD4<sup>+</sup> T cells in hNOJ mice were distinguishable from naturally occurring Tregs, which have a CD45RA<sup>+</sup> naïve phenotype (46, 60, 61), these cells had not recently migrated from the thymus but presumably had expanded in the spleen after interacting with antigen-presenting cells. It is not likely that GM-CSF acted directly on the T cells, since T cells in humanized NSG mice (29) and in humans (62) do not express the receptor for GM-CSF. It has been suggested that DCs, especially those expressing MHC II and co-stimulatory molecules (CD80 and CD86), play a major role in the development of Tregs (1, 63, 64). Our correlation analysis demonstrated that the development of CD45RA<sup>−</sup>Foxp3hi activated Tregs was associated with the maturation status of all putative DC populations, particularly with respect to the expression of the B7 family co-stimulatory molecules (CD80 and CD86) by all putative DC populations. Interestingly, GM-CSF-treated human CD1c<sup>+</sup> cDCs (65) and their equivalents in mice (66) can induce Tregs, suggesting a unique role for GM-CSF in the modulation of CD1c<sup>+</sup> cDCs. Although whether GM-CSF can induce tolerogenicity in CD141<sup>+</sup> cDCs and pDCs remains unknown, both DC subsets can induce Tregs under certain conditions (67, 68). Further characterization regarding tolerogenicity and which subsets of DCs are directly involved in the skewed development of CD45RA−Foxp3hi activated Tregs in hNOJ mice should be undertaken in the future.

A recent study by Minoda et al. showed that human CD1c<sup>+</sup> cDCs and CD141<sup>+</sup> cDCs reconstituted in humanized NSG-A2 mice, into which HLA-A2 is genetically introduced, are functionally equivalent to mouse CD11b<sup>+</sup> cDCs that promote Th2 and Th17 responses and mouse CD8<sup>+</sup> cDCs that promote Th1 and CD8+ T cell responses, respectively (50). Interestingly, our correlation analysis suggests that the development of CD45RA<sup>−</sup>Foxp3lo non-Tregs is associated with CD1c<sup>+</sup> myeloid cells, but not CD141<sup>+</sup> myeloid cells in terms of CD80 and CD86 expressions. Chen et al. demonstrated that induction of GM-CSF and IL-4 in humanized NSG mice resulted in T cell activation and differentiation toward a CD45RA<sup>−</sup> memory phenotype (28). In particular, these humanized mice could induce antigen-specific CD4<sup>+</sup> T cell responses, including the secretion of IFN-γ and IL-4, following immunization with tetanus toxoid (28), suggesting that Th1 and Th2 development can occur in humanized mice under suitable conditions. Whether hNOJ mice could induce antigen-specific T cell responses as well as immunity or tolerance in the cytokine setting tested in this study remains to be investigated.

In conclusion, this study could provide a platform for understanding the development of human DCs and Tregs *in vivo*. Furthermore, this study sheds light on the methodology of using conventionally available second-generation immunodeficient mice expressing certain human cytokines *in vivo*. However, it should be noted that the induced cytokine concentrations are transient and unphysiological in this system. Further improvement could be achieved using human cytokine knock-in immunodeficient mice.

#### ETHICS STATEMENT

Human umbilical cord blood was donated by the Japanese Red Cross Society Kanto-Koshinetsu Block Blood Center (Tokyo, Japan), Sugiura Women's Clinic (Tokyo, Japan), and Fukuda

#### REFERENCES


Hospital (Kumamoto, Japan) after receiving written informed consent. The use of human umbilical cord blood was approved by the Medical Research Ethics Committee of the NIID for the use of human subjects (Tokyo, Japan) (protocol numbers 500 and 585). All mice were treated in accordance with the guidelines approved by the Institutional Animal Care and Use Committee of the NIID (protocol numbers 114035 and 215005).

#### AUTHOR CONTRIBUTIONS

Study design: KT. Data curation: RI, SI, MK-I, YT-Y, and KT. Acquisition of data: RI and KT. Analysis and interpretation of data: RI, SI, and KT. Validation: RI, SI, MK-I, HT, MA, YT-Y, and KT. Writing the original manuscript: RI and KT. Review and/or revision of the manuscript: MA and YT-Y.

#### ACKNOWLEDGMENTS

We thank the Japanese Red Cross Society Kanto-Koshinetsu Block Blood Center, Dr. K. Sugiura (Sugiura Women's Clinic, Tokyo, Japan), and Dr. K. Matsui (Fukuda Hospital, Kumamoto, Japan) for donating human umbilical cord blood. We also thank Dr. S. Okada (Kumamoto University, Kumamoto, Japan) for providing NOJ mice and K. Okano and R. Iwaki (NIID, Tokyo, Japan) for technical support.

#### FUNDING

This work was supported by JSPS KAKENHI under Grant Number JP17K08800 (KT), ViiV healthcare Japan Research Grant 2015 (KT), and AMED under Grant Numbers JP18fk0410003 (KT) and JP17fk0410305 (YT-Y).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.01042/ full#supplementary-material.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Iwabuchi, Ikeno, Kobayashi-Ishihara, Takeyama, Ato, Tsunetsugu-Yokota and Terahara. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

*Edited by:* 

*Christian Muenz, Universität Zürich, Switzerland*

#### *Reviewed by:*

*Veronika Lukacs-Kornek, Saarland University, Germany Elodie Segura, Institut Curie, France*

#### *\*Correspondence:*

*Marleen Ansems marleen.ansems@radboudumc.nl; Gosse J. Adema gosse.adema@radboudumc.nl*

#### *†Present address:*

*Jonas Nørskov Søndergaard, Department of Microbiology, Tumor and Cell Biology, Science for Life Laboratory, Karolinska Institutet, Stockholm, Sweden; Vassilis Triantis, FrieslandCampina, Amersfoort, Netherlands*

#### *Specialty section:*

*This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology*

*Received: 19 April 2018 Accepted: 07 June 2018 Published: 22 June 2018*

#### *Citation:*

*Søndergaard JN, van Heeringen SJ, Looman MWG, Tang C, Triantis V, Louche P, Janssen-Megens EM, Sieuwerts AM, Martens JWM, Logie C, Stunnenberg HG, Ansems M and Adema GJ (2018) Dendritic Cells Actively Limit Interleukin-10 Production Under Inflammatory Conditions via DC-SCRIPT and Dual-Specificity Phosphatase 4. Front. Immunol. 9:1420. doi: 10.3389/fimmu.2018.01420*

# Dendritic cells actively limit interleukin-10 Production Under inflammatory conditions *via* Dc-scriPT and Dual-specificity Phosphatase 4

*Jonas Nørskov Søndergaard1†, Simon J. van Heeringen2 , Maaike W. G. Looman1 , Chunling Tang1 , Vassilis Triantis 1†, Pauline Louche1 , Eva M. Janssen-Megens <sup>3</sup> , Anieta M. Sieuwerts <sup>4</sup> , John W. M. Martens <sup>4</sup> , Colin Logie3 , Hendrik G. Stunnenberg3 , Marleen Ansems <sup>1</sup> \* and Gosse J. Adema1 \**

*1Radiotherapy & OncoImmunology Laboratory, Department of Radiation Oncology, Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, Nijmegen, Netherlands, 2Department of Molecular Developmental Biology, Faculty of Science, Radboud Institute for Molecular Life Sciences, Radboud University, Nijmegen, Netherlands, 3Department of Molecular Biology, Faculties of Science and Medicine, Radboud Institute for Molecular Life Sciences, Radboud University, Nijmegen, Netherlands, 4Department of Medical Oncology and Cancer Genomics Netherlands, Erasmus MC Cancer Institute, Erasmus University Medical Center, Rotterdam, Netherlands*

Dendritic cell (DC)-based immunotherapy makes use of the DC's ability to direct the adaptive immune response toward activation or inhibition. DCs perform this immune orchestration in part by secretion of selected cytokines. The most potent anti-inflammatory cytokine interleukin-10 (IL-10) is under tight regulation, as it needs to be predominantly expressed during the resolution phase of the immune response. Currently it is not clear whether there is active suppression of IL-10 by DCs at the initial pro-inflammatory stage of the immune response. Previously, knockdown of the DC-specific transcription factor DC-SCRIPT has been demonstrated to mediate an extensive increase in IL-10 production upon encounter with pro-inflammatory immune stimuli. Here, we explored how DC-SCRIPT contributes to IL-10 suppression under pro-inflammatory conditions by applying chromatin immunoprecipitation sequencing analysis of DC-SCRIPT and the epigenetic marks H3K4me3 and H3K27ac in human DCs. The data showed binding of DC-SCRIPT to a GA-rich motif at H3K27ac-marked genomic enhancers that associated with genes encoding MAPK dual-specificity phosphatases (DUSPs). Functional studies revealed that upon knockdown of DC-SCRIPT, human DCs express much less DUSP4 and exhibit increased phosphorylation of the three major MAPKs (ERK, JNK, and p38). Enhanced ERK signaling in DC-SCRIPT-knockdown-DCs led to higher production of IL-10, which was reverted by rescuing DUSP4 expression. Finally, DC-SCRIPTknockdown-DCs induced less IFN-γ and increased IL-10 production in naïve T cells, indicative for a more anti-inflammatory phenotype. In conclusion, we have delineated a new mechanism by which DC-SCRIPT allows DCs to limit IL-10 production under inflammatory conditions and potentiate pro-inflammatory Th1 responses. These insights may be exploited to improve DC-based immunotherapies.

Keywords: ZNF366, ERK, chromatin immunoprecipitation sequencing, dual-specificity phosphatase 4, dendritic cells, MAPK, DC-SCRIPT, IL-10

# INTRODUCTION

Immunotherapy has gained great success in recent years due to the success of checkpoint inhibitors targeting the PD-1/PD-L1 or CTLA-4 pathways (1–4). These checkpoint inhibitors act on T cells, and it is widely accepted that T cells are quintessential in the fight against cancer. However, T cells are not capable of eliciting an anti-tumor response by themselves, because without the right instructions, the T cells will remain naïve or become tolerogenic (5). The only cell type capable of efficiently educating naïve T cells is the dendritic cell (DC). We have come very far in our understanding of the role of cell surface receptors and secreted molecules involved in immune regulation by DCs, which has been exploited in therapies against cancer with promising results (6, 7). By contrast, information regarding intracellular regulatory circuits in DCs is far scarcer, and could possess potential for future therapeutic strategies as well.

DCs sample their environment, and sense foreign molecules using pattern-recognition receptors of which the toll-like receptors (TLRs) are the best known (8). Activation of DCs through TLR-triggering is needed for upregulation of antigen-presenting and co-stimulatory molecules, and production of cytokines that direct the T cell response toward activation (9). Thus, DCs hold the potential to induce either an immune-activating or immune-dampening response (10). Interleukin-10 (IL-10) has been demonstrated to play a crucial role in this process, by inhibiting development of pro-inflammatory Th1 cells and promoting development of anti-inflammatory Tr1 cells (11). Besides being involved in T cell education, several studies also reported on downregulation of MHC class II expression on antigen-presenting cells (APCs) and of MHC class I on tumor cells in cancer patients with elevated IL-10 levels in serum or tumors (12–18). Although high IL-10 production in cancer patients is associated with a poor prognosis, some recent reports have suggested that situations may exist in which elevated levels of IL-10 in cancer patients may also have beneficial effects *via* dampening the chronic inflammation in tumor microenvironments and by stimulation of cytotoxicity of already activated CD8+ T cells (19).

In the context of APCs, we have learned a great deal about how IL-10 is upregulated *via* activation of the MAP kinase ERK and the transcription factor NF-κB during the resolution phase of an immune response (20). Whether IL-10 expression is actively kept in check in DCs during the initial inflammatory phase of the immune response is currently unknown (20). Dendritic cell-specific transcript (DC-SCRIPT, *ZNF366*) is a transcription factor uniquely expressed by DCs in the immune system (21–23). DC-SCRIPT has been shown to have a complex collection of functions, including nuclear receptor co-regulatory activity (24–29), cell cycle regulation (30), NF-κB activity modulation (31), and mediating induction of T cell proliferation and IFN-γ production (23). Most strikingly, DC-SCRIPT knockdown leads to a massive increase in IL-10 production after pro-inflammatory TLR-triggering (23, 27, 31), suggesting that DC-SCRIPT may participate in active suppression of IL-10. In this study. we explored how DC-SCRIPT contributes to the active suppression of IL-10 production in human activated monocyte-derived DCs.

# RESULTS

#### DC-SCRIPT Binds Genomic Enhancers Near MAPK Phosphatases

In order to delineate the molecular mechanism for how DC-SCRIPT regulates IL-10 expression in DCs under inflammatory conditions, we conducted a chromatin immunoprecipitation sequencing (ChIP-Seq) analysis using a DC-SCRIPT specific antibody on human monocyte-derived DCs. Antibodies specific for the epigenetic marks histone H3 lysine 4 tri-methylation [H3K4me3; associated with promoters of active genes (32)] and H3K27 acetylation (ac) [present at active enhancers and promoters (33)] were taken along in the same DC samples (**Figure 1**; GEO accession number: GSE78923). To account for donor variation and dynamics, three donors were assayed, using immature, 1 h, and 24 h TLR ligand R848 (Resiquimod)-stimulated DCs. Only DC-SCRIPT binding sites consistently present in all three donors in at least one of the assayed time points were used for subsequent analysis, yielding a total of 10,833 DC-SCRIPT binding sites in the human genome. Clustering of the DC-SCRIPT DNA binding sites with the genomic H3K27ac and H3K4me3 histone marks in DCs displayed a 37% overlap (**Figure 1A**, cluster 2–6). Out of these, 1,462 DC-SCRIPT binding sites overlapped with the promoter mark H3K4me3 (**Figure 1A**, clusters 2–4 and **Figure 1B**, *top*), and 2,550 DC-SCRIPT binding sites overlapped with histone marks characteristic for enhancers (high H3K27ac, low/no H3K4me3, clusters 5–6 **Figures 1A,B**, *middle*). Promoterassociated DC-SCRIPT binding sites are henceforth referred to as PA-SC binding sites and enhancer-associated DC-SCRIPT binding sites referred to as EA-SC binding sites. The DC-SCRIPT binding sites that did not co-localize with either of these marks (**Figures 1A,B**, *bottom*) were left out from the current analysis. To further characterize the sequence content of the genomic locations where DC-SCRIPT binds, a *de novo* motif analysis was performed on the DC-SCRIPT ChIP-Seq dataset using GimmeMotifs (34). This unguided comparison of the DNA sequence under all the DC-SCRIPT binding sites, yielded a GA-rich DC-SCRIPT binding motif (**Figure 1C**). To validate the motif in a different and independent assay, we also performed *in vitro* cyclic amplification and selection of targets (CAST) in a cell-free system (**Figure 1D**). The CAST motif successfully validated the *de novo* motif with the two independently generated motifs being 78% similar to each other [assayed by MAST in MEME (35), Figure S1 in Supplementary Material]. The *de novo* and the *in vitro* motif were found in up to 44% of the EA-SC binding sites and 38% of the PA-SC binding sites, respectively (Figure S1 in Supplementary Material).

Previously, we have shown that in the absence of DC-SCRIPT there is increased NF-κB binding to the *il10* enhancer (31), suggesting that DC-SCRIPT may also bind there. Surprisingly, there was no DC-SCRIPT binding site in the *il10* promoter or any *il10*-associated enhancer that could explain the effect of DC-SCRIPT expression on IL-10 production. Genomic Regions Enrichment of Annotations Tool [GREAT (36)] was then used to get more insight into the pathways that DC-SCRIPT may regulate to affect IL-10 expression. The gene ontology (GO)

FIGURE 1 | Genome-wide mapping of DC-SCRIPT binding sites in human dendritic cells (DCs). Chromatin immunoprecipitation sequencing (ChIP-Seq) of human immature or R848 (toll-like receptors 7/8 ligand)-activated DCs using DC-SCRIPT, H3K4me3, or H3K27ac Abs (*n* = 3). (A) Heatmap of *k*-means clustering analysis (*k* = 6, Euclidean distance) of DC-SCRIPT (blue), H3K4me3 (green), and H3K27ac (purple) in 10-kb windows around DC-SCRIPT peak summits. Clusters 2–4 are merged into an enhancer-associated cluster; clusters 5–6 are merged into a promoter-associated cluster. The rows correspond to the peaks; the *x*-axis shows the position relative to the peak center. The intensity of the color represents the number of reads in 100-bp windows. (B) Representative UCSC Genome Browser tracks of a PA-SC binding site (top), an EA-SC binding site (middle), and a no mark binding site (bottom). The tracks from top to bottom show the gene annotation, the ChIP-Seq signal for DC-SCRIPT (blue), H3K4me3 (green), H3K27ac (purple), and input (gray). (C) *de novo* motif generated from DC-SCRIPT ChIP-Seq binding sites. (D) *In vitro* motif identified by cyclic amplification and selection of targets. (E) Genomic Regions Enrichment of Annotations Tool analysis of all DC-SCRIPTbinding sites containing the GA-rich motif. The table contains the top 10 most significant terms in the gene ontology category molecular function. See also Figure S1 and Tables S1 and S2 in Supplementary Material.

biological process term was dominated by immune regulatory processes (Table S1 in Supplementary Material), reinforcing the previously demonstrated role of DC-SCRIPT in immune regulation (23, 31). Interestingly, the molecular function GO showed that DC-SCRIPT binds in the vicinity of MAPK phosphatase genes (**Figure 1E**; Table S2 in Supplementary Material). MAPK phosphatases are a subgroup of the dual-specificity phosphatases (DUSPs) and are responsible for the dephosphorylation of MAPKs (37). Given that the MAPK ERK previously has been associated with IL-10 production (20), DC-SCRIPT may therefore potentially regulate DUSPs to modulate IL-10 production.

#### DC-SCRIPT Modulates the MAPK Pathway

The GREAT analysis showed that 4 out of 14 MAPK DUSPs were associated with a total of 9 EA-SC binding sites located on average 246 kb (range: 3–859 kb) from the transcription start site. In order to validate their expression in relation to DC-SCRIPT, we knocked down DC-SCRIPT using siRNA specific to DC-SCRIPT (SC-KD-DCs) or control non-targeting siRNA (Ctrl-DCs) (**Figure 2A**). The expression of the four DUSP genes that had DC-SCRIPT binding sites associated (DUSP1, DUSP4, DUSP5, and DUSP10) were assayed by RT-qPCR at different time points after R848 stimulation (**Figures 2B–E**). As a control, the wellknown ERK phosphatase DUSP6 (38), that in this study did not have any DC-SCRIPT binding sites associated, was also assayed (**Figure 2F**). All five assayed DUSPs were upregulated after R848 stimulation, but only DUSP4 and DUSP6 displayed a significant difference in expression upon DC-SCRIPT knockdown. DUSP4 was consistently higher expressed in Ctrl-DCs, while DUSP6 was consistently higher expressed in SC-KD-DCs. Since the main role of DUSPs is to inhibit MAPK signaling, we next assayed if DC-SCRIPT silencing has an impact on the signaling of the MAPKs. The levels of phosphorylation of the three major MAPKs: ERK, JNK, and p38 were assayed by western blotting (WB)

dendritic cells (DCs). (B–F) R848-stimulated Ctrl-DC and SC-KD-DC time course for dual-specificity phosphatase (DUSP)1, 4, 5, 6, and 10 (*n* = 4, mean + SEM). (G–J) Phosphorylation and total protein expression of ERK, JNK, and p38 in Ctrl-DCs and SC-KD-DCs after stimulation with R848 (*n* = 5, mean + SEM). (K,L) DUSP6 and DUSP4 expression in Ctrl-DCs and SC-KD-DCs pretreated with UO126 (MEK inhibitor) or vehicle control for 1 h, followed by R848 stimulation (*n* = 4, mean + SEM). Statistics: Student's paired *t* test, Ctrl-DCs were compared to SC-KD-DCs at the same time point. \**p* < 0.05, \*\**p* < 0.01. See also Figure S2 in Supplementary Material.

(**Figures 2G–J**) and flow cytometry (Figure S2 in Supplementary Material). Stimulation with R848 led to a peak in phosphorylation of all three kinases after 30 min in both SC-KD-DCs and Ctrl-DCs. DC-SCRIPT knockdown significantly enhanced the phosphorylation of all three kinases, with the most potent increase in ERK, and only a minor change of phosphorylation on p38. As ERK signaling has previously been demonstrated to upregulate expression of DUSPs (39, 40), we pretreated the cells with a MEK inhibitor prior to stimulation, to investigate whether ERK signaling was responsible for the observed effect on DUSP4 and DUSP6 expression (**Figures 2K,L**). Strikingly, pretreatment with the inhibitor diminished DUSP6 upregulation, while DUSP4 remained unaltered. This suggests that DUSP4 is regulated directly by DC-SCRIPT, whereas the effect on DUSP6 is rather part of an ERK signaling feedback loop.

#### DC-SCRIPT Enhances DUSP4 Expression

The DC-SCRIPT binding site associated with the DUSP4 gene was further validated by ChIP-qPCR in seven additional independent donors (Figure S3 in Supplementary Material). The binding site displayed enhancer characteristics (high H3K27ac and low H3K4me3 binding) and was located 285kB downstream of the DUSP4 transcription start site (**Figure 3A**). Both the *de novo* and the *in vitro* motif could be found toward the center of the binding site (**Figure 3B**). Intriguingly, this location falls within a local topological associated domain (TAD) previously found in all 19 different assayed cell lines and 13 different tissues, including immune cells (**Figure 3C**). Together this would suggest that this DC-SCRIPT binding position is an enhancer for DUSP4. To further evaluate this, a luciferase vector containing the DUSP4 promoter and the genomic DNA underlying the DUSP4 EA-SC-binding site was generated. For comparison, a control vector with a piece of genomic DNA of similar size, without the motif, and located between the DUSP4 TSS and

the DUSP4 EA-SC binding site was employed. Interestingly, the genomic DNA underlying the DUSP4 EA-SC genomic DNA lead to a 8.5-fold increase in luciferase activity (**Figure 3D**). Co-transfection with DC-SCRIPT further increased the luciferase signal in a dose-dependent manner (1.8-fold compared to no DC-SCRIPT). To confirm the impact of DC-SCRIPT expression on DUSP4 at the functional level, DUSP4 protein expression was determined in immature and R848-stimulated Ctrl-DCs and SC-KD-DCs (**Figures 3E,F**). In accordance with the mRNA expression, DUSP4 protein levels were significantly higher in Ctrl-DCs relative to DC-SCRIPT silenced DCs both at the immature state and after stimulation. Altogether, these data therefore indicate that the DUSP4 gene-associated EA-SC binding site works as an enhancer, and that DC-SCRIPT expression impacts the DUSP4 protein level.

#### The DC-SCRIPT-Induced Phenotype Is Mediated by DUSP4

The most dominant immunological phenotype reported for DC-SCRIPT in DCs so far is the major increase in IL-10 production following siRNA-mediated knockdown of DC-SCRIPT (23, 31). To assess whether the observed change in DUSP4 expression level could be responsible for this effect on IL-10 production, we evaluated Ctrl-DCs and SC-KD-DCs ability to produce IL-10 after MAPK inhibition and DUSP4 overexpression (**Figure 4**). While inhibition of JNK did not affect IL-10 secretion, both inhibition of the upstream kinase of ERK (MEK) and p38 lead to a major decrease in IL-10 (**Figure 4A**).

DUSP4 has previously been demonstrated to have a high affinity for dephosphorylating ERK (44), and could therefore explain the enhanced ERK phosphorylation after DC-SCRIPT knockdown. To confirm this in human DCs, we rescued DUSP4 expression by introducing DUSP4/GFP or GFP in SC-KD-DCs using Ctrl-DCs as controls and determined the level of phosphorylated ERK (**Figures 4B,C**; Figure S4 in Supplementary Material). Strikingly, rescuing DUSP4 expression changed phosphorylation of ERK in the SC-KD-DCs back to wild-type levels. As DC-SCRIPT knockdown also had an effect on phosphorylation of p38, the effect of DUSP4 on p38 phosphorylation was also assayed under the same settings (**Figures 4D,E**). Interestingly, like ERK, phosphorylation of p38 was also reduced to wildtype levels after rescuing DUSP4 in DC-SCRIPT knockdown DCs. Finally, rescuing DUSP4 expression in SC-KD-DCs also normalized levels of IL-10 production (similar level as Ctrl-DC) (**Figure 4F**).

#### DC-SCRIPT Expression in DCs Skews Naïve T Cells Toward Th1 Immune Activation

To determine the impact of DC-SCRIPT expression on the cytokine polarization of naïve CD4+ T cells, we co-cultured SC-KD-DCs or Ctrl-DCs with allogeneic naïve CD4+ T cells and assayed their cytokine production. Interestingly, the responding T cells produced less IFN-γ and more IL-10 upon incubation with SC-KD-DCs (**Figure 5A**). In line with these data, the expression of T cell subset restricted transcription factors revealed a significant decrease in T-bet (Th1 TF) expressing T cells, while the number of GATA3 (Th2 TF) and RORgT (Th17 TF) expressing T cells remained similar (**Figure 5B**).

Altogether, these data demonstrate that DC-SCRIPT-mediated enhancement of DUSP4 expression is responsible for restricting ERK and p38 signaling and subsequent IL-10 production by professional antigen-presenting DCs under inflammatory conditions (**Figure 6**), which in turn limits pro-inflammatory CD4+ T cell polarization.

#### DISCUSSION

Dendritic cells are the sentinels of the immune system, playing a decisive role in the balance between immunogenic and tolerogenic immune responses (10), however, the molecular mechanism of how DCs control production of the anti-inflammatory cytokine IL-10 in an inflammatory setting is still unclear. Here, we show that DC-SCRIPT, a DC-specific transcription factor in the immune system, binds to GA-rich sequences in enhancerand promoter-associated DNA regions for many immune-related genes, including an enhancer for the MAPK phosphatase DUSP4. Moreover, we show that DC-SCRIPT knockdown limits DUSP4 expression resulting in an increase in the activity of the MAPK signaling pathway and subsequent IL-10 production by DCs. In addition, the knockdown of DC-SCRIPT also hindered differentiation of naïve CD4+ T cells into pro-inflammatory T cells. Altogether, these data show a novel mechanism by which professional antigen-presenting DCs limit IL-10 production, a pathway that needs to be tightly controlled to induce protective immune responses (20).

Many different immune cells can produce IL-10 and during cancer development, this has mainly been considered detrimental for successful eradication of the tumor cells due to IL-10's immunosuppressive capabilities on DCs (45) and T cells (46). Previously, we have demonstrated that DC-SCRIPT affects TLRmediated IL-10 production in human DCs (23), and described that enhanced IL-10 production in SC-KD-DCs was partly caused by altered post-translational modifications of NF-κBp65 leading to increased binding of NF-κBp65 to an IL-10 enhancer element (31). In our current ChIP-Seq data, we did not find a DC-SCRIPT binding site at the same position as for NF-κBp65 in the enhancer of the IL-10 gene, i.e., it does not seem that DC-SCRIPT is part of the NF-κBp65 DNA binding complex. In line with these observations, knockdown of DC-SCRIPT affected the phosphorylation status of NF-κB (31), and with the current data linking DC-SCRIPT to phosphatases, it may suggest that DC-SCRIPT affects NF-κB activation indirectly. In support of this, a downstream kinase (MSK1) of the MAPKs ERK and p38 has been demonstrated to mediate phosphorylation of NF-κBp65 (47), and ERK has been shown to play a critical role for IL-10 production in DCs (20).

In the current work, we found that DC-SCRIPT knockdown leads to higher DUSP4 expression and that DC-SCRIPT binds a GA-rich DNA sequence with enhancer abilities within a local DUSP4-TAD. Using luciferase assays, we found that the largest enhancing effect was gained by cloning the enhancer close to the DUSP4 promoter, while co-transfecting DC-SCRIPT only

arbitrary unit. See also Figure S4 in Supplementary Material.

led to a subtle (but significant) further increase in the signal. These data could indicate that one of the roles of DC-SCRIPT in enhancing DUSP4 expression would be to bring the DUSP4 enhancer in close proximity to the promoter. Currently, only little is known about the functional role of DUSP4 in DCs. Gene expression analysis has revealed that immune cells express up to 17 different DUSPs to various extent, but only one or two DUSPs showed high expression in any single cell type (37). DUSP4 was predominantly expressed in DCs, and furthermore, high DUSP4 expression could distinguish DCs from the closely related monocyte-derived macrophages (37). Interestingly, accumulating evidence suggests that cell type-specific transcription factors often regulate cell type-specific genes *via* binding to enhancers (48).

by ELISA (*n* = 12 + SEM). Statistics: Student's paired *t* test. \*\**p* < 0.01, \*\*\**p* < 0.001.

One of DCs main function in the immune system is to educate T cells toward pro-inflammatory or anti-inflammatory responses. We observed that knockdown of DC-SCRIPT in DCs limited the DCs capability to induce pro-inflammatory CD4+ T cells when polarized from naïve cells. These T cells produced less IFN-γ and more IL-10, and in line with these data, the number of T-bet expressing T cells was also reduced. Surprisingly, even though DC-SCRIPT knockdown leads to increased production of IL-10, and IL-10 is a potent inducer of anti-inflammatory Tr1 cells (11), preliminary experiments did not show any skewing of naïve T cells into Tr1 cells (data not shown). As DC-SCRIPT levels affect a DC's capacity to induce inflammatory Th1 responses, one could consider monitoring DC-SCRIPT expression in DCs used in vaccination studies.

In summary, the transcription factor DC-SCRIPT binds regulatory DNA sequences linked to genes involved in the immune system and the MAPK pathway, including MAPK phosphatases. We identify regulation of expression of the MAPK phosphatase DUSP4 as the mode of action through which DC-SCRIPT restricts the expression of the crucial immuneinhibitory cytokine IL-10 in DCs. These data help to delineate the mechanisms that govern DCs unique molecular function, and its central role in controlling immune responses. As DC immunotherapy is a promising approach to treat cancer patients, specifically targeting tumor cells and having only few side effects (49, 50), much research has focused on which patternrecognition receptor ligands or cytokine-cocktails would generate a DC-phenotype with a favorable cytokine profile, often focusing on high IL-12 and low IL-10 levels (51–53). The current data showing IL-10 being actively inhibited by DC-SCRIPT-induced phosphatases imply that maturing DCs with a combination of immune-activating adjuvants and either kinase inhibitors or

FIGURE 6 | Model of DC-SCRIPT mediated control of IL-10 production in DCs. In the presence of DC-SCRIPT (left figure), DC-SCRIPT binds an enhancer (E) for DUSP4 *via* a GA-rich motif. This leads to enhanced DUSP4 expression, which limits ERK signaling, and subsequent IL-10 production under inflammatory conditions. By contrast, in the absence of DC-SCRIPT (right figure), DUSP4 is only expressed at a low level, thereby enabling higher ERK signaling leading to an increase in IL-10 production. X and Y are potential protein partners in the enhancer complex.

phosphatase inducers may be beneficial to improve DC-based immunotherapy.

# MATERIALS AND METHODS

### Generation of Human Monocyte-Derived Dendritic Cells (moDCs)

Human moDCs were generated from PBMCs as described previously (54). Buffy coats were obtained from healthy volunteers (Sanquin, Nijmegen, The Netherlands) after informed consent and according to institutional guidelines. Plastic-adhered monocytes were cultured for a total of 6 days in RPMI 1640 medium (Life Technologies) supplemented with 1% ultra-glutamine (Cambrex), 0.5% antibiotic–antimycotic (Invitrogen), 10% (v/v) FCS (Greiner Bio-one), IL-4 (300 U/mL), and GM-CSF (450 U/mL) (both from Cellgenix). On day 3, moDCs were supplemented with new IL-4 (300 U/mL) and GM-CSF (450 U/mL).

## Small Interfering RNA (siRNA)-Mediated Knockdown

On day 3–4 of DC differentiation, cells were harvested and subjected to electroporation. For DC-SCRIPT silencing, a 23-nt custom ZNF366 siRNA termed SC38 targeting the DC-SCRIPT gene at position 2,349–2,369 was used (Thermo Scientific). siRNA ON-TARGETplus Non-Targeting siRNA#1 (Thermo Scientific) was used as control. Cells were washed twice in PBS and once in OptiMEM without phenol red (Invitrogen). A total of 10 µg siRNA was transferred to a 4-mm cuvette (Bio-Rad), and 10 × 106 DCs were added in 200 µL OptiMEM and incubated for 3 min before being pulsed with an exponential decay pulse at 300 V, 150 μF, in a Genepulser Xcell (Bio-Rad), as previously described (55). Immediately after electroporation, the cells were transferred to pre-heated (37°C) phenol red-free RPMI 1640 culture medium supplemented with 1% ultra-glutamine, 10% (v/v) FCS, IL-4 (300 U/mL), and GM-CSF (450 U/mL).

#### Stimulations and Inhibitors

Immature DCs were stimulated with 4 µg/mL R848 (Axxora). In some experiments, DCs were pre-treated 1–2 h with one of the following inhibitors: 4 µM UO126 (MEK1/2, LC Laboratories), 2.5 µM SB203580 (p38, LC laboratories), 5 µM SP600125 (JNK, Tocris Bioscience).

#### RNA Isolation, Reverse Transcription, and Quantitative PCR

Total RNA was isolated from cells using Trizol (Ambion). RNA quantity and purity were determined on a NanoDrop spectrophotometer. RNA was treated with DNase I (amplification grade; Invitrogen) and reverse transcribed into cDNA by using random hexamers and Moloney murine leukemia virus reverse transcriptase (Invitrogen). mRNA levels for the genes of interest were determined with a CFX96 sequence detection system (Bio-Rad) with SYBR Green (Roche) as the fluorophore and gene-specific oligonucleotide primers. Primers used are as follows (forward, reverse): DC-SCRIPT (*ZNF366*): (5′-AAGCATGGAGTCATGGAG-3, 5′-TTC TGAGAGAGGTCAAAGG-3′), *PGK1*: (5′-CAAGAAGTATGCT GAGGCTGTCA-3, 5′-CAAATACCCCCACAGGACCAT-3′), *DU SP4*: (5′-AGTGGAAGATAACCACAAGG-3, 5′-GCTTAACGA ACTCGAAGG-3′), *DUSP6*: (5′-GATCACTGGAGCCAAAAC-3, 5′-CAAGCAATGTACCAAGACAC-3′), *DUSP1*: (5′-AGTAC CCCACTCTACGATCAGG-3, 5′-GAAGCGTGATACGCACTG C-3′), *DUSP5*: (5′-TGTCGTCCTCACCTCGCTA-3, 5′-GGGCT CTCTCACTCTCAATCTTC-3′), *DUSP10*: (5′-TTTGAAGAGG CTTTTGAGTT-3, 5′-GGGAGATAATTGGTCGTTT-3′). Reaction mixtures and program conditions were used as recommended by the manufacturer (Bio-Rad). Quantitative PCR data were analyzed with the CFX Manager software (Bio-Rad) and checked for correct amplification and dissociation of the products. mRNA levels of the genes of interest were normalized to mRNA levels of the housekeeping genes *GAPDH* or *PGK1* and were calculated according to the cycle threshold method (56).

# ELISA

Secreted IL-10 was measured using the human IL-10 ready-setgo kit (eBioscience) in the supernatants of 16–24 h-stimulated DCs or T cells. Secreted IFN-γ was measured using IFN-γ monoclonal antibodies; coating clone 2G1, detection Ab-biotin, clone XMG1.2 (both Thermo Fisher).

# Western Blotting

Monocyte-derived dendritic cells were lysed in a concentration of 106 /100 μL 4°C cold lysis buffer consisting of 62.5 mM Tris (pH 6.8, Sigma-Aldrich), 1% SDS (Invitrogen), and freshly added complete protease inhibitor cocktail (Roche), 1 mM PMSF (Sigma-Aldrich), and for phosphorylation-specific western blot: phosphatase inhibitors 1 mM Na3VO4 (Sigma-Aldrich) and 10 mM NaF (Merck). Cell lysates were mixed 1:4 with sample buffer containing 5% glycerol (Invitrogen), 6% SDS, 125 mM Tris–HCl (pH 6.8), 0.1 mg/mL bromophenol blue (Gebr. Schmid), and 10% 2-ME (Sigma-Aldrich), heated at 95°C for 5 min, and then cooled on ice. The proteins were resolved by electrophoresis on a 10% polyacrylamide gel (ratio of acrylamide to bisacrylamide, 37.5:1) and transferred to Protran nitrocellulose transfer membranes (Amersham) at 4°C. The following Abs were used for staining in TBS with 0.1% tween 20 (TBST) and 5% BSA (all from Cell Signaling Technology): rabbit anti-DUSP4 (1:1,000 dilution, clone: D9A5), rabbit anti-ERK (1:1,000 dilution, cat: 9102), rabbit anti-pERK (Thr202/Tyr204) (1:2,000 dilution, clone: D13.14.4E), rabbit anti-JNK (1:1,000 dilution, cat: 9252), rabbit anti-p38 (1:2,000 dilution, clone: D13E1), and rabbit anti-p-p38 (Thr180/Tyr182) (1:1,000 dilution, clone: D3F9). Goat anti-DC-SCRIPT were used for staining in PBS with 0.1% tween 20 and 1% skimmed milk powder (Campina) and 3% BSA (1:600 dilution, cat: AF4707, R&D Systems). Mouse anti-p-JNK (Thr183/Tyr185) were used for staining in TBST with 2% skimmed milk powder and 2% BSA (1:250 dilution, clone: G9, Cell Signaling Technology). Membranes were blocked for 1 h at room temperature, stained overnight with primary Abs [including rabbit anti-actin (1:2,000 dilution, clone: 20–33, Sigma-Aldrich) or rat anti-tubulin (1:2,000 dilution, clone: YOL1/34, Novus Biologicals)], and stained for 1 h at room temperature with corresponding secondary Abs (diluted 1:5,000): goat anti-mouse IgG IRDye 800CW, goat anti-rabbit IgG IRDye 800CW, donkey anti-goat IgG IRDye 800CW, donkey anti-rabbit IgG IRDye 680 (all Li-cor Biosciences), and goat antirat Alexa Fluor 680 (Invitrogen). Membranes were scanned by using an Odyssey Infrared Imaging System (Li-cor Biosciences).

#### Flow Cytometry of DCs

For phosphorylation-specific flow cytometry, paraformaldehyde (Merck) was added directly to the culture medium at a final concentration of 1.6% at the end of the stimulation. Cells were fixed for 10 min at room temperature. Subsequently, culture medium was aspirated, resuspended in 100% ice-cold MeOH (Boom) and incubated at −20°C overnight, followed by extensive washing (4×) in PBS with 1% BSA, 0.05% NaN3 (Merck). Cells were blocked and stained in PBS with 1% BSA, 0.05% NaN3 (Merck), and 2% human serum (Sanquin), using 1:200 diluted pERK, 1:400 diluted p-p38 Abs, or isotype control (rabbit IgG, Jackson Immuno Research). Cells were stained with primary Ab/isotype control for 45 min on ice, followed by 30 min on ice with a secondary PE-Cy5.5 goat anti-rabbit IgG Ab (Invitrogen, cat# L42018). Data were acquired on an FACSCyan (Beckman Coulter). Isotype controls gave a staining intensity similar as unstimulated moDCs, indicating that blocking conditions were sufficient to avoid unspecific staining (data not shown). Viability was measured using a fixable viability dye eFluor780 (eBioscience), following the manufacturers instructions. Data were analyzed using FlowJo (Treestar).

#### Chromatin Immunoprecipitation Sequencing

Chromatin immunoprecipitation was performed as previously described (31), using 20 µg goat-α-DC-SCRIPT Ab (R&D Systems). 10 ng of input or ChIP-enriched DNA was end-paired using T4 DNA polymerase, *E. coli* DNA Pol I large fragment (Klenow polymerase, New England Biolabs), and T4 polynucleotide kinase (New England Biolabs), followed by purification using the QIAquick PCR purification kit (Qiagen). Subsequently, DNA was dA-tailed using the Klenow fragment (3′ to 5′ exo minus, New England Biolabs), followed by purification using the MinElute Reaction Cleanup kit (Qiagen). Next, DNA was ligated to multiplex NEXTflex adapters (Bioo Scientific). IP and input DNA were purified by the MinElute reaction Cleanup kit, and amplified by PCR using the KAPA HiFi HotStart ReadyMix PCR kit (KAPA Biosystems) with the following program: 45 s at 98°C for initial denaturation, 15 s at 98°C, 30 s at 65°C, 30 s at 72°C for four cycles, followed by 1 min at 72°C for final extension. Removal of excess adaptors and selection of 300 bp bands was done using 2% E-Gel SizeSelect Agarose Gels (Invitrogen). Adapter-modified DNA fragments were enriched by PCR using the KAPA HiFi HotStart ReadyMix PCR kit for 8–10 cycles with the aforementioned program. To get rid of the 120 bp adapter dimer, the PCR product was purified using Ampure beads (Beckman Coulter). Libraries were sequenced on the Illumina HiSeq 2000. H3K4me3 and H3K27ac antibodies were extensively characterized1 and used for ChIP according to standard BLUEPRINT protocols.1

#### ChIP-Seq Data Processing

Peaks were called using the algorithm MACS2 version 2.0.10.20120913 (57) with default settings. In order to account for donor variation and dynamics, three donors were assayed, using immature, 1 h-, and 24 h-stimulated DCs. Peaks present in all three donors were determined using intersect with BEDTools version 2.20.1 (58), and only DC-SCRIPT peaks present in all three donors were used for subsequent analysis. All ChIP-Seq data have been submitted to the GEO database (accession number: GSE78923) *k*-Means clustering (*k* = 6, Euclidean distance) and heatmaps were generated using Fluff (59). *De novo* motif analysis was done using GimmeMotifs version 0.8.6 (34). The motif was trimmed to remove low information content containing bases.

#### Cyclic Amplification and Selection of Targets

Human DC-SCRIPT was cloned in the pCATCH vector (60), as BamHI-XbaI inserts. *In vitro* transcription/translation was performed with the TNT T7 Quick Coupled Transcription/ Translation System (Promega) according to the manufacturer's recommendations, using 1 µg of DNA as input. Transcription/ translation took place at 30°C for 90 min. 10% of the reaction was tested by western blot analysis to verify protein production, while 20% was used in each CAST round.

Oligo-nucleotides carrying defined ends and a 21-nt region of degeneracy (5′-GCCTCCATGGACGAATTCTGT-(N)21-AGCG GATCCCGCATATGACCG-3′) and PCR primers (forward: 5′- G CCTCCATGGACGAATTCTGT-3′ and reverse: 5′-CGGTCAT ATGCGGGATCCGCT-3′) were used during CAST. As a first step, double-stranded oligo-nucleotides were prepared as follows. 8.5 µg of the degenerative nucleotides were mixed with 4.3 µg of the reverse primer in 50 µL of Tris–HCl (100 mM, pH 8) and heated at 80°C, then cooled down slowly to 4°C. 2 µL of the hybridized oligo-nucleotides were used together with 2 U of the Klenow fragment of DNA polymerase I (37°C for 1 h) to create dsDNA. dsDNA was precipitated and used in the first round of CAST. Each CAST round was performed in binding buffer containing 30 mM HEPES pH 7.4, 100 mM NaCl, 0.01% NP40, 0.01 mg/mL BSA, 0.05 mM ZnSO4, 2 mM MgCl2, 0.6 mM PMSF, and 10% glycerol. In brief, 500 µL of binding buffer were mixed with 20% *in vitro* transcribed/translated proteins and DNA and incubated for 30 min at 4°C. Then 10 µL of Protein G beads and 3 µg of mouse M2 anti-FLAG mAb (Sigma) were added and incubated overnight. Precipitated dsDNA was used for the first round of CAST, or 80% of the PCR reaction for the subsequent rounds. After the binding reaction, Protein G beads were washed twice with 600 µL of binding buffer and resuspended in 20 µL of 5 mM EDTA pH 8 for 10 min at 90°C. Beads were pelleted and supernatant was used for PCR. 20% of the PCR reaction was tested on gel to verify DNA precipitation and amplification by CAST.

PCR reactions for CAST were performed using 100 ng of forward and reverse primer, 0.5 mM of dNTPs, 5 mM of MgCl2, and 2.5 U of Taq polymerase with 58°C as an annealing temperature. The number of PCR cycles was kept to a minimum, i.e., 15 cycles, in the first two rounds. Minimal PCR amplification

<sup>1</sup>www.blueprint-epigenome.eu (Accessed: April 18, 2018).

helped to reduce the amplification of non-specific oligonucleotides and the formation of hetero-duplexes that resulted from the re-annealing of products that were mismatched in the 21-bp central-region. After the third round, however, 20 cycles of PCR ensured good amplification and abundance of specific oligo-nucleotides. Sequences from four rounds of CAST were used as input for MEME (35) to generate a consensus sequence for DC-SCRIPT.

#### Genomic Regions Enrichment of Annotations Tool

Genomic Regions Enrichment of Annotations Tool analysis was done in version 3.0.0 as previously described (36), with DC-SCRIPT binding sites with an H3K4me3 and/or H3K27ac histone mark, and containing the GA-rich motif. Default settings were employed, i.e., basal plus extension: proximal: 5.0 kb upstream, 1.0 kb downstream, plus distal: up to 1,000.0 kb. Statistical significance is based on false discovery rate (cutoff: 0.05). Displayed data contains minimum three genes in each GO.

#### Luciferase Assays

The DNA sequence underlying the ChIP-Seq-identified DUSP4 EA-SC-binding site was cloned into a pGL4.10 luciferase vector (Promega) behind a DUSP4 promoter. For comparison, a control pGL4.10 vector with the DUSP4 promoter and a piece of genomic DNA of similar size, and located between the DUSP4 TSS and the DUSP4 EA-SC binding site was also generated. These vectors were transfected into HEK293 [ATCC, tested to be mycoplasma free using mycoalert mycoplasma detection kit (Lonza), and used between passages 2–15 after thawing] together with a renilla control vector (pRL-TK, Promega) and increasing amounts of a pCATCH-DC-SCRIPT expression vector. HEK293s were plated 24 h before transfection using metafectene. Cells were harvested after 24 h, and cell lysates were analyzed for luminescence according to the manufacturer's protocol (Dual Luciferase Reporter assay, Promega) using a Victor3 luminometer (PerkinElmer). Relative light units were calculated after correction for transfection efficiency based on the activity of the cotransfected pRL-TK.

#### Overexpression of DUSP4

To generate a DUSP4/GFP expression vector, DUSP4 (NM\_001394) was cloned into the expression vector pEGFP-N3 (Clontech, Mountain View, CA, USA), using the restriction enzyme sites BglII and BamHI. As a control, the empty pEGFP-N3 vector was used. Immature SC-KD-DCs and Ctrl-DCs were harvested on day 6 and electroporated using the Neon transfection system (Invitrogen) according to the manufacturer's instructions. Briefly, 106 DCs were mixed with 5 µg DNA, and electroporated with two pulses of 1,000 V for 40 ms. Subsequently, DCs were seeded in microtiter plates and rested for 5–6 h until GFP was visible. The cells were subsequently stimulated and used for functional assays.

#### Naïve T Cell Polarization

Naïve T cells were isolated from buffy coats using magneticassociated cell sorting, and negative selection, by depleting cells expressing CD8a, CD14, CD15, CD16, CD19, CD36, CD56, CD132, TcRγ/δ, and CD235a (CD4+ T Cell isolation kit, human, Miltenyi), and CD45RO [anti-CD45RO-PE (DAKO) plus anti-PE microbeads (Miltenyi)]. Purity was checked using CD3-FITC (BD), CD4-PE-Cy7 (BioLegend), CD45RA-APC-Cy7 (BioLegend), and CD45RO-PE and determined to be >97% of CD3+ CD4+ CD45RO− CD45RA+ T cells.

SC-KD-DCs or Ctrl-DCs were stimulated for 16 h with R848, washed and counted, before co-culturing with naïve CD4+ T cells in a ratio of 5,000:20,000 DC:T. The super antigen SEB (Sigma-Aldrich) was added at 10 pg/mL. After 5, 7, and 9 days of co-culture, the T cells were split 1:2 and recombinant human IL-2 were added at a final concentration of 20 U/mL. On day 11, the T cells are in a resting state, which can be seen in a light microscope by the T cell clusters falling apart and cells are rounded. The resting T cells were assayed for intracellular transcription factor expression using the cytofix/cytoperm kit (BD). Prior to permeabilization cells were stained with fixable viability dye eFluor780 (eBioscience). Antibodies used for staining were: anti-human Gata-3-Alexa Fluor488, anti-human RORgamma(t)-APC, and anti-human T-bet-PE (all eBioscience).

For cytokine production, T cells were harvested, counted, and 100,000 cells re-plated in a 96-well round bottom plate, followed by addition of 100,000 anti CD3/CD28 beads (Gibco). Supernatant was harvested after 24 h and assayed by ELISA. Data presented in the figure consists of two T cell donors and six DC donors.

## AVAILABILITY OF DATA

The datasets generated during the current study are available in the GEO repository, under accession number: GSE78923.

#### ETHICS STATEMENT

Buffy coats were obtained from healthy volunteers (Sanquin, Nijmegen, The Netherlands) after informed consent in accordance with the Declaration of Helsinki. The study was approved by the Institutional Review Board of the Radboud University Nijmegen Medical Center, Commissie Mensgebonden Onderzoek.

# AUTHOR CONTRIBUTIONS

Conceptualization: JS, MA, and GA; methodology: JS, SH, MA, and GA; software: SH; formal analysis: JS and SH; investigation: JS, ML, CT, VT, PL, EJ-M, AS, and MA; writing—original draft: JS, SH, MA, and GA; writing—review and editing: JS, SH, ML, CT, VT, PL, EJ-M, AS, JM, CL, HS, MA, and GA; visualization: JS and SH; funding acquisition: JS, CT, JM, MA, and GA.

#### ACKNOWLEDGMENTS

MA is recipient of NWO-Veni 91615093 and a long-term fellowship (BUIT 2012-5347) from the Dutch Cancer Society. SH is supported by NWO, the Netherlands Organisation of Scientific Research (NWO-ALW grant 863.12.002). JM and AS received funding through Cancer Genomics Netherlands (CGC.nl) and a grant from the Netherlands Organization for Scientific Research (NWO).

#### FUNDING

This work was supported by grant KUN2011-5229 and KUN2009- 4402 from the Dutch Cancer Society (PI GA), and a grant from the Chinese Scholarship Council to CT.

#### REFERENCES


### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.01420/ full#supplementary-material.

TABLE S1 | GREAT analysis, GO Biological Process.

TABLE S2 | GREAT analysis, GO Molecular Function.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Søndergaard, van Heeringen, Looman, Tang, Triantis, Louche, Janssen-Megens, Sieuwerts, Martens, Logie, Stunnenberg, Ansems and Adema. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Current Concepts of Antigen Cross-Presentation

#### *Maria Embgenbroich and Sven Burgdorf\**

*Life and Medical Sciences (LIMES) Institute, University of Bonn, Bonn, Germany*

Dendritic cells have the ability to efficiently present internalized antigens on major histocompatibility complex (MHC) I molecules. This process is termed cross-presentation and is important role in the generation of an immune response against viruses and tumors, after vaccinations or in the induction of immune tolerance. The molecular mechanisms enabling cross-presentation have been topic of intense debate since many years. However, a clear view on these mechanisms remains difficult, partially due to important remaining questions, controversial results and discussions. Here, we give an overview of the current concepts of antigen cross-presentation and focus on a description of the major cross-presentation pathways, the role of retarded antigen degradation for efficient cross-presentation, the dislocation of antigens from endosomal compartment into the cytosol, the reverse transport of proteasome-derived peptides for loading on MHC I and the translocation of the cross-presentation machinery from the ER to endosomes. We try to highlight recent advances, discuss some of the controversial data and point out some of the major open questions in the field.

#### *Edited by:*

*Diana Dudziak, Universitätsklinikum Erlangen, Germany*

#### *Reviewed by:*

*Steffen Jung, Weizmann Institute of Science, Israel Joke M. M. Den Haan, VU University Medical Center, Netherlands*

> *\*Correspondence: Sven Burgdorf burgdorf@uni-bonn.de*

#### *Specialty section:*

*This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology*

*Received: 22 May 2018 Accepted: 04 July 2018 Published: 16 July 2018*

#### *Citation:*

*Embgenbroich M and Burgdorf S (2018) Current Concepts of Antigen Cross-Presentation. Front. Immunol. 9:1643. doi: 10.3389/fimmu.2018.01643*

Keywords: dendritic cells, cross-presentation, antigen processing, endosomes, antigen dislocation

#### INTRODUCTION

Dendritic cells (DCs) scan the peripheral tissue for antigens. Upon their recognition, antigens are internalized and the DCs activated and migrate toward the draining lymph node, where they can induce an adaptive immune response (1). In order to do so, they need to process the internalized antigens and load antigen-derived peptides on major histocompatibility complex (MHC) molecules. Peptides loaded onto MHC II molecules can be recognized by antigen-specific CD4<sup>+</sup> T helper cells. Similarly, peptides loaded on MHC I molecules can be recognized by antigen-specific CD8<sup>+</sup> T cells, leading to their proliferation and the activation of their cytotoxic capacities.

The presentation of internalized antigens on MHC I molecules is a process termed crosspresentation. Efficient cross-presentation has been shown to be crucial in, e.g., the induction of an adaptive immune response against tumors and viruses that do not infect DCs directly and in the induction of peripheral tolerance (2–5).

The molecular mechanisms that regulate classical antigen presentation on MHC II molecules and cross-presentation, however, have been shown to be quite divers. For MHC II-restricted presentation, internalized antigens are degraded in endo/lysosomal compartments by proteases such as cathepsins. Newly synthesized MHC II molecules, which are stabilized by binding to the invariant chain (Ii), are transported from the ER toward this compartment, where Ii is degraded by lysosomal proteases, resulting in the binding of only a small peptide fragment (CLIP) to MHC II. Subsequently, CLIP is replaced by antigen-derived peptides by the chaperon HLA-DM (6).

In contrast to MHC II-restricted presentation, the molecular mechanisms regulating crosspresentation are less understood and in part discussed controversially. There seems to be a whole variety of pathways leading to antigen cross-presentation and, despite intensive investigations, the molecular mechanisms and individual contribution of each pathway are rather unclear.

In this review, we try to describe some of the recent advances in cross-presentation, focusing on the major cross-presentation pathways and highlighting some of the controversial observations in the field.

#### CROSS-PRESENTING DC SUBSETS

Although many cells are able to present extracellular antigens on MHC I, DCs are considered to be the most prominent and most relevant cross-presenting cells.

In general, DCs are subdivided into conventional DCs (cDCs) and plasmacytoid DCs (pDCs). cDCs are further classified into cDC1 and cDC2 (7). In mice and human, cDC1 are characterized by the expression of the chemokine receptor XCR1 (7–9) and their development relies on the expression of the transcription factors IRF8 and Batf3 (10–12), whereas the development of cDC2 is mainly regulated by IRF4 (7, 13). Additionally, murine cDC1 express either CD8 (in lymphoid tissues) or CD103 (in non-lymphoid tissues), whereas human cDC1 are characterized by the expression of BDCA-3 (CD141) (8, 14–17).

The cDC1 are generally considered to be potent crosspresenting DCs *in vivo*. Accordingly, in murine lymphoid tissue, soluble and cell-associated OVA are cross-presented by resident CD8+ DCs (18–20), whereas soluble and cell-associated antigens in lung (21, 22), intestine, and skin (23–25) are cross-presented by migratory CD103+ DCs. Further functional properties of cDC1 are the uptake of apoptotic cells *via* Clec9A/DNGR1 (26–29) and the responsiveness to TLR3 stimulation (30).

cDC1 express high levels of MHC I pathway genes (31), show high intra-endosomal reactive oxygen species (ROS) production and low acidification in endosomes (32, 33), all features of efficient cross-presentation (see below). They express the small GTPase Rac2, which enables the assembly of the NAPDH oxidase complex NADPH oxidase 2 (NOX2), resulting in enhanced ROS production and active alkalization of endosomes (33). Additionally, cDC1 show only marginal expression levels of the C-type lectin Siglec-G, a potent inhibitor of NOX2 (34).

However, the cDC1 are not the only cross-presenting DC population. Many other DC subpopulations, including cDC2, have been shown to cross-present as well (35–39). For human DCs, it even has been demonstrated that BDCA3<sup>+</sup> (cDC1s), BDCA1<sup>+</sup> (cDC2s), and even pDCs all bear intrinsic capacities to cross-present extracellular antigens (40). The exact role of different cDC1 and cDC2 subpopulations in cross-presentation is, therefore, under debate and, especially since functional data on the physiological role of human DC subsets in cross-presentation is hard to obtain, future experiments will have to shed light on this question.

Although pDCs have been shown to be able to cross-present antigens (41), their role in cross-presentation *in vivo* is questionable, especially since their depletion did not affect crosspresentation and clearance of viral antigens (42).

### MAJOR PATHWAYS OF ANTIGEN CROSS-PRESENTATION

Intensive research has clearly shown that there are a wide variety of mechanisms by which peptides derived from extracellular antigens can be presented on MHC I molecules. In general, there are two main cross-presentation pathways: the vacuolar pathway and the endosome-to-cytosol pathway (**Figure 1**). In the vacuolar pathway, antigen processing and loading onto MHC I molecules occurs within the endo/lysosomal compartment. After internalization, antigens are degraded by lysosomal proteases and antigen-derived peptides are loaded onto MHC class I molecules there. The lysosomal protease Cathepsin S has been demonstrated to play a crucial role in antigen degradation for the vacuolar pathway (43). In the endosome-to-cytosol pathway, internalized antigens need to be transported from the endosomal compartment into the cytosol, where they are degraded by the proteasome (44–46). Derived peptides are subsequently transported by the transporter associated with antigen processing (TAP) into the ER or back into the antigen-containing endosomes, where they can be loaded onto MHC class I (44, 45, 47–49). Although substantial evidence points out that some antigens indeed can be cross-presented independent of proteasomal degradation and TAP-mediated peptide transport by the vacuolar pathway (43, 50–53), most cross-presentation studies report of cross-presentation *via* the endosome-to-cytosol pathway. The dependency of crosspresentation on proteasomal degradation seems logical, since the functional outcome of cross-presentation is the activation of antigen-specific cytotoxic T cells. After migration toward to site of infection, these T cells are fully equipped to kill potential target cells, like virus-infected cells or tumor cells. In order to become functionally active, T cells must recognize the same epitope presented on MHC I by the target cells. Importantly, MHC I-loaded peptides on target cells do not emerge from cross-presentation but rather are the result of direct (classical) MHC I-restricted presentation of endogenous antigens, in which peptides are generated by the proteasome. Since it is hard to assume that for all antigens, the proteasome and lysosomal proteases generate exactly the same epitopes, the dependency of cross-presentation on proteasomal degradation for at least a substantial part of the antigens might circumvent this problem. Accordingly, DCs deficient in the LMP7 subunit of the immunoproteasome are impaired in cross-presentation *in vitro* and *in vivo* (46). However, it needs to be mentioned that very few information about the *in vivo* significance of the vacuolar vs. endosome-to-cytosol pathway is available, pointing out that future experiments are needed to further investigate the relative importance of both pathways *in vivo*.

#### DELAYED ANTIGEN DEGRADATION AND ITS ROLE IN CROSS-PRESENTATION

Over the last years, it has become clear that intra-endosomal antigen stability critically regulates cross-presentation, which efficiency is negatively affected by rapid lysosomal degradation of

endosome-to-cytosol pathway, antigens are transported into the cytosol for proteasomal degradation. Afterward, antigen-derived peptides are transported back into the endosomes (soluble and particular antigens) or into the ER (particular antigens) *via* TAP. There, they are trimmed by IRAP (endosomes) or ERAP (ER) and loaded onto MHC I. They cross-presentation machinery might be translocated toward endosomes *via* Sec22b.

internalized antigens (54). Lysosomal maturation and activation of lysosomal proteases is fine-tuned by the transcription factor TFEB, an important regulator of cross-presentation (55).

It is generally assumed that rapid antigen degradation quickly destroys a large amount of epitopes before they can be processed properly and loaded onto MHC I molecules (56, 57). Additionally, peptide-loaded MHC I molecules have a limited life span at the cell membrane (58–60). In order to enable T cell activation after migration toward the draining lymph node, however, prolonged cross-presentation seems to be essential. Therefore, limited antigen degradation in antigen-presenting cells might be a mechanism to generate a kind of intracellular antigen depot, from where continuous antigen processing and presentation might ensure the presence of peptide-loaded MHC I molecules over a longer period of time (61). Such intracellular antigen storage depots have also been shown in human monocytes, which accumulate long-peptide antigens for over 5 days in non-lysosomal compartments, where day are protected from rapid degradation (62).

Since DCs are the most efficient cross-presenting cells, these cells possess several mechanisms by which they can actively prevent rapid lysosomal antigen degradation.

First, it was demonstrated that DCs express lower levels of lysosomal proteases (63) and display a reduced velocity of endosome maturation (64) compared to other immune cells. Expression of asparagine endopeptidase and Cathepsins L, S, D, and B in phagosomes of DCs was clearly reduced compared to macrophages, resulting in impaired phagolysosomal degradation and prolonged antigen stability after internalization by DCs (63). The delivery of lysosomal proteases toward phagosomes was even further reduced after stimulation of DCs with LPS (64).

Second, in DCs, an active alkalization of endosomes prevents pH-dependent activation of lysosomal proteases. During lysosome maturation, protons are transported into the luminal space by the V-ATPase, leading to the activation of pHdependent lysosomal proteases. Reduced V-ATPase activity in DCs might contribute to prevent a rapid drop in pH after antigen internalization (65). Additionally, DCs seem to have the unique capacity to alkalize their endosomes by the recruitment the NOX2 toward the endosomal membrane (32, 66). There, NOX2 can mediate the generation of ROS, which in turn capture protons to build hydrogen peroxide (**Figure 1**). Proton trapping by ROS causes an active alkalization, impairing pH-dependent activation of lysosomal proteases, which in turn prevents rapid antigen degradation and stimulates cross-presentation (32). The recruitment of NOX2 toward endosomes is mediated by Rab27a (67).

Third, DCs express endocytosis receptors that specifically target non-degradative endosomal compartments. Previous studies have demonstrated that the endocytosis receptor used to internalize an antigen critically determines its intracellular routing and degradation (68). A previous study from our group demonstrated that antigens internalized by fluid phase pinocytosis or scavenger receptor-mediated endocytosis are rapidly targeted toward lysosomes, where they are efficiently degraded by lysosomal proteases, resulting in poor cross-presentation (68). However, if the same DCs simultaneously internalized the same antigen by the mannose receptor, it was targeted toward a distinct pool of early endosomes, which did not undergo rapid fusion with lysosomes and in which antigens were protected from lysosomal degradation, resulting in efficient cross-presentation of MR-internalized antigens (68). Although the role of the MR in *in vivo* cross-presentation has been discussed controversially (69, 70), it now is clear that CD103<sup>+</sup> DCs in liver and lung use this receptor for cross-presentation of, e.g., viral antigens (71). A correlation between antigen targeting into early endosomes clearly distinct from lysosomes and cross-presentation efficiency has also been confirmed in human DCs. Also in these cells, MR-mediated internalization resulted in its routing into early endosomes, retarded degradation and efficient cross-presentation, whereas uptake by DEC205 lead to antigen targeting into lysosomes, rapid lysosomal degradation, and hence poor cross-presentation (56). Interestingly, attenuating lysosomal degradation was sufficient to rescue the cross-presentation of DEC-205-internalized antigens (56), highlighting again the importance of intra-endosomal antigen stability for efficient cross-presentation. Additionally, the targeted region of the endocytosis receptor might also play a role in antigen degradation and presentation, adding even an additional degree of complexity. Figdor and colleagues demonstrated that antigen targeting toward the carbohydrate recognition domain of DC-SIGN delivers antigens to lysosomal compartments, resulting in rapid degradation and poor crosspresentation, whereas targeting the neck region of DC-SIGN causes antigen delivery in early endosomal compartments clearly distinct from lysosomes, causing prolonged stability and efficient cross-presentation (72, 73).

#### ANTIGEN TRANSLOCATION INTO THE CYTOSOL AS CRITICAL STEP IN ANTIGEN CROSS-PRESENTATION

After being internalized into a non-degradative endosomal compartment, antigens need to be processed before they can be loaded onto MHC I. In the endosome-to-cytosol pathway, internalized antigens, therefore, need to be transported across the endosomal membrane into the cytosol for proteasomal degradation. Although this is a key step in antigen cross-presentation and significant efforts have been made to shed light on this process, the underlying mechanisms mediating such intracellular antigen transport are still topic of debate.

In general, if DCs enable access of endosomal antigens to the cytosol, this must be a process, which is controlled very tightly. Uncontrolled lysosome leakage would lead to the cytosolic release of Cathepsins, which in turn would activate the NLRP3 inflammasome (74) and result in pyroptosis, an inflammatory form of cell death (75). To avoid this, total lysosomal content cannot just be released into the cytosol in an uncontrolled fashion. Accordingly, antigens need to be unfolded (76) and disulfide bridges need to be reduced by the γ-interferon-inducible lysosomal thiol reductase GILT (77) before efficient translocation and hence cross-presentation can take place. This supports the idea that antigen translocation is highly regulated, might involve dislocation through a transmembrane pore complex, and is presumably not the result of simple lysosome leakage.

It is generally assumed that members of the ER-associated degradation (ERAD) machinery contribute in enabling antigen dislocation for cross-presentation. First indirect indications for a role of ERAD in this process came from observations describing the presence of ERAD components in the phagosomal membrane (47, 48) and from experiments using the ERAD inhibitor Exotoxin A, which specifically represses crosspresentation (49, 78).

First direct evidence for the involvement of the ERAD machinery came from the Cresswell group, who demonstrated an important role of the AAA ATPase p97 in antigen dislocation (49). Whereas expression of a dominant-negative p97 mutant in DCs represses cross-presentation, the addition of purified wild-type p97 but not the dominant-negative mutant to purified phagosomes enhanced antigen translocation (49, 79–81), indicating that p97 indeed might provide the energy to pull endosomal antigens into the cytosol.

The identification of a dedicated translocon, which actually functions as a transmembrane pore complex to enable antigen dislocation across the endosomal membrane, has been (and still is) by far more difficult. One putative candidate, which has been proposed to mediate antigen dislocation into the cytosol over a decade ago, is the ERAD member Sec61 (49, 78), a trimeric protein whose downregulation has been shown to inhibit antigen translocation and cross-presentation (79, 82). However, since Sec61 plays an important role in the dislocation of proteins at the ER membrane, like, e.g., the dislocation of MHC I molecules themselves (83), it is very hard to distinguish endosome-specific effects of Sec61 from general effects at the ER. In an attempt to solve this problem, we generated Sec61-specific intracellular antibody (intrabody), which we fused to an ER retention signal (84), leading to the trapping of Sec61 in the ER and preventing its transport toward endosomes (82). By this means, we could demonstrate that the transport of Sec61 toward endosomes indeed is essential for antigen dislocation and cross-presentation. Additionally, we could demonstrate that the expression of the intrabody did not alter overall Sec61 expression and did not affected the ERAD-mediated dislocation of MHC I, TCR, CD3δ and the split venus protein at the ER membrane. This points out that ERAD activity at the ER remained unaltered by the expression of the intrabody. These data suggest that Sec61 indeed might serve as a translocon for cross-presentation (**Figure 1**). However, it cannot be formally excluded that the translation of another pore complex at the ER membrane is changed by manipulating intracellular Sec61 transport or that another putative pore complex is translocated toward endosomes in a complex with Sec61, being influences by intrabody expression. Additionally, Grotzke et al. demonstrated that a chemical inhibitor of Sec61, mycolactone, does not seem to influence antigen dislocation from the cytosol (85). This inhibitor has been shown to directly bind Sec61 and targets proteins that are co-translationally imported in a Sec61-mediated fashion into the ER toward proteasomal degradation (86, 87). However, whether mycolactone could possibly affect Sec61-mediated protein dislocation from the ER into the cytosol is not clear, especially since such proteins are generally ubiquitinated and targeted for proteasomal degradation also in the absence of mycolactone (80). Indeed, mycolactone was shown to have no influence on ERAD (86) and also Grotzke et al. demonstrated that mycolactone does not affect protein retranslocation from the ER into the cytosol. Whether this is due to a missing role of Sec61 in this process or to specific properties of the inhibitor needs to be determined. Especially since addition of Exotoxin A, an inhibitor that blocks Sec61 channel openings (78, 88), clearly affects antigen translocation and cross-presentation (49, 82), there seems to be a need of information on the exact working mechanism of these inhibitors and on the role of Sec61 in dislocation from the ER to finally clear a potential role of Sec61 on cross-presentation.

In addition to antigen dislocation through a pore complex, a recent study postulated that lipid peroxidation in DCs might play a crucial role in antigen transport into the cytoplasm (89). Here, the authors proposed that the specific recruitment of NOX2 might cause lipid peroxidation in endosomes. As mentioned above, NOX2 captures protons to generate hydrogen peroxide, preventing rapid acidification of the endosome. Lipid peroxidation caused by such hydrogen peroxide was suggested to result in leakiness of the endosomal membrane and hence, antigen access into the cytosol and enhanced cross-presentation. However, it remains unclear how the antigen-presenting cell in this case would prevent inflammasome-induced cell death caused by unspecific release of cathepsins. Additionally, the necessity for endosomal antigens to be unfolded (76) and reduced by GILT (77) cannot be explained by simple leackage of the endosomal membrane. Therefore, the significance of such a pathway in crosspresentation *in vivo* remains to be elucidated.

# TRANSPORT OF PROTEASOME-DERIVED PEPTIDES FOR LOADING ONTO MHC I

After being transported into the cytosol, internalized antigens are degraded by the proteasome. Subsequently, antigen-derived peptides can be transported through the TAP transporter into the ER or alternatively, by endosomal TAP, back into the endosomal compartments (44, 45, 47, 48, 90). There, peptides are trimmed into a suited size for loading onto MHC I molecules. Such trimming can occur *via* the peptidases ERAP (in the ER) or IRAP (in endosomes) (**Figure 1**) (91). Presentation of peptides derived from soluble antigens is mainly ERAP-independent *in vitro* and *in vivo* (92), but rather occurs in endosomes after transport by endosomal TAP and IRAP-mediated peptide trimming (90, 92). Proteasome-derived peptides derived from particulate antigens, however, can be transferred into both the ER and endosomes, where they are trimmed by ERAP or IRAP, respectively, and loaded onto MHC I (91). These underlying mechanisms for these differences are unknown.

Recently, it was demonstrated that, in addition to antigen translocation through endosomal TAP, some peptides might enter endosomes in an energy-consuming but TAP-independent fashion (93), pointing out the possibility of additional (unknown) transporters involved in peptide transport into endosomes for cross-presentation. Additionally, since TAP-independency was often used to demonstrate cross-presentation *via* the vacuolar pathway, there is the possibility that at least in part of these studies, antigens might have entered the endosome *via* the endosometo-cytosol pathway, using alternative peptide transporters.

After peptide reimport into the endosomes, they can be loaded onto MHC I molecules. In general, there are two basic possibilities how MHC I molecules can enter the endosome. First, newly synthesized MHC I molecules could be transported from the ER to the endosomes and used for peptide loading in cross-presentation. Second, MHC I from the cell surface (that are already loaded with peptides) could be transported toward endosomes during endocytosis events. The Blander group demonstrated that for particulate antigens, MHC I molecules used for cross-presentation mainly originated from the cell membrane and were translocated into an endosomal recycling compartment in a Rab11a-dependent fashion (**Figure 1**) (94). From these organelles, MHC I molecules can be transported toward phagosomes, a process that is mediated by the SNARE protein SNAP23 and critically depends on MyD88 signaling (94). It remains unclear, however, whether peptide exchange on recycling MHC I molecules requires the help of additional chaperon proteins (similar to the function of HLA-DM in MHC II-restricted presentation) or can occur after simple weakening of the peptide–MHC I binding in endosomes. Additionally, it has been demonstrated that for cross-presentation of elongated peptides, a substantial part of the used MHC I molecules are newly synthesized molecules recruited from the ER (95). In this case, it needs to be determined whether such MHC I molecules are loaded with peptides in the ER and undergo peptide exchange in acidic endosomes, or whether the transport of the entire peptide loading complex, which stabilizes unbound MHC I molecules and assists in peptide binding, to the endosomes is required for cross-presentation. Despite the presence of several members of the peptide loading machinery in endosomes (45, 47, 48), a functional relevance of these proteins in cross-presentation is missing.

#### TRANSPORT OF ER COMPONENTS TO ENDOSOMES

As described above, efficient cross-presentation requires the transport of ER proteins toward endosomes. The exact mechanisms, by which this transfer occurs, are not completely understood and in part contradictory data complicate a clear view on this process.

Since it is known that endosomes during their maturation directly interact with the ER to exchange a wide variety of molecules (96), such ER-endosome membrane contact sites would offer an easy explanation for the transfer of ER proteins toward endosomes. However, it has been proposed by Amigorena and Savina that the transport of cross-presentation components toward endosomes takes place from the ER-golgi intermediate compartment (ERGIC) (97). Membrane fusion between the ERGIC and the phagosomes has been postulated to be mediated by the SNARE proteins Sec22b (in the ERGIC) and syntaxin 4 (in the phagosome). Accordingly, shRNA-mediated downregulation of Sec22b resulted in impaired recruitment of ER proteins toward phagosomes, decreased antigen translocation into the cytosol, and hence reduced cross-presentation (94, 97). These observations support a critical role of the ERAD machinery in antigen dislocation into the cytosol for cross-presentation as described above. However, a recent study by Reddy and colleagues demonstrated that severe off target effects of the used shRNA might have caused the observed influence on cross-presentation (98), questioning the role of Sec22b in cross-presentation. Since such off target effects of shRNA molecules can be circumvented by the generation of Sec22b-deficient mice, one could expect that the use of conditional knockout mice would shed light on the situation and would clearly indicate whether Sec22b is indeed involved in cross-presentation. Mice bearing a conditional knockout of Sec22b in CD11c<sup>+</sup> DCs were generated by both the Reddy and the Amigorena group. Strikingly, whereas Reddy et al. reported complete independency of cross-presentation on Sec22b (98), Amigorena et al. showed a clear impairment of cross-presentation in Sec22b-knockout DCs, hence drawing completely opposite conclusions (99). Both groups used partially different *in vitro* and *in vivo* systems to substantiate their findings, but since also opposite effects of Sec22b on crosspresentation using the same cells (BM-DCs and splenic DCs) and the same antigens (soluble and bead-bound OVA) were observed, these contractionary results cannot be explained by different experimental setups only (100). Therefore, the exact role of Sec22b and ERGIC-mediated transport of ER proteins needs to be confirmed.

The recruitment of MHC I molecules toward antigen-containing phagosomes was shown to be induced by TLR ligands. TLR-induced and MyD88-dependent signaling resulted in the activation of IKK2, which phosphorylates SNAP23, mediating fusion events between phagosomes and MHC I-containing recycling endosomes (94). Also the transport of other ER proteins toward endosomes has been shown to be stimulated by TLR ligands (82, 90). Using flow cytometric analysis of individual endosomes (101), have demonstrated before that low amounts of Sec61 are present in endosomes also in the absence of TLR ligands, and that a clear recruitment of Sec61 toward antigencontaining endosomes was induced by LPS (82). Since it is very unlikely that Sec61 is also recruited *via* recycling endosomes, distinct mechanisms might come into play for the transport of these molecules.

One of these mechanisms might rely on the uncoordinated 93 homolog B1 (UNC93B1), which is activated by TLR triggering and mediates the transport of TLRs from the ER toward endosomes (102–104). Interestingly, UNC91B1 has been demonstrated to be critically involved in cross-presentation (105). Although a putative role of UNC93B has also been discussed controversially (106), it now becomes clear that an essential role of UNC93B1 in cross-presentation is based on its interaction with the store-operated-Ca2<sup>+</sup>-entry regulator STIM1. UNC93B1 has been shown to be essential for oligomerization of hence activation of STIM1, which in turn alters local Calcium signaling regulating phago/endosome fusion events (107, 108). Ablation of UNC93B1 impairs antigen translocation into the cytosol and cross-presentation (107). Interestingly, antigen dislocation into the cytosol was impaired despite reduced endosomal antigen degradation, which is generally assumed to increase antigen export from the endosomes. Since UNC93B1 upon TLR stimulation mediates TLR transport from the ER toward endosomes, it, therefore, is thinkable that ER members of the cross-presentation machinery are transported from the ER toward endosomes in a similar UNC93B1-dependent fashion.

Additionally, DC activation by TLR ligands can have other effects on the cross-presentation machinery independent of ER to endosome transport, like the prevention of phagosome fusion with lysosomes and concomitant antigen stabilization (57) or increases in antigen internalization (109).

#### ALTERNATIVE CROSS-PRESENTATION PATHWAYS

In all cross-presentation pathways described above, crosspresented antigens entered the DC *via* endocytosis. However, there are some reports indicating that also distinct mechanisms can lead to cross-presentation.

One of these mechanisms is the transport of pre-processed antigens (peptides) from a donor cell to a DC. Such transport can occur *via* direct cell–cell contact, mediated by gap junctions (110, 111). After gap junction-mediated transport from one cell to another, antigen-derived peptides can enter the normal MHC I presentation pathway. Interestingly, the donor cell does not need to be an antigen-presenting cell, offering the possibility that DCs can obtain such peptides directly from infected cells. Infection of melanoma cells with *Salmonella* has been demonstrated to increase the expression of Connexin 43, an important gap junction protein, enabling efficient gap junction-mediated peptide transfer from the infected cell to the DC and hence efficient cross-presentation (112). However, given the limited stability of intracellular peptides (113), the physiological significance of such peptide transfer in cross-presentation remains unclear.

Another alternative cross-presentation pathway is termed cross-dressing, which generally implies that the cross-presenting DC becomes an MHC I molecule, which has already been loaded with an antigen-derived peptide, transferred from a donor cell (114, 115). Similar to gap junction-mediated peptide transfer, such donor cell does not necessarily need to be an antigen-presenting cell, suggesting that DCs can derive peptide-loaded MHC I molecules directly from infected cells or even apoptotic cells. The transfer of loaded MHC I molecules is thought to be mediated by cell–cell contact rather than secretory vesicles (114, 116) and overcomes the need of intracellular antigen processing within the DC. Cross-dressing has been shown to occur *in vivo* (114, 116) and cross-dressed DCs have been shown to activate memory T cells after viral infection (116). Remarkably, in this study, the activation of naive T cells did not depend on cross-dressing (116), offering the possibility that different cross-presentation pathways might be responsible for the activation of different T cell populations or for T cell activation under specific conditions. However, the exact physiological relevance of cross-dressing and especially its contribution compared to the other crosspresentation pathways in specific situations, however, remains to be elucidated.

#### REFERENCES


#### CONCLUSION

Despite intensive research over the last decades, several questions regarding the molecular mechanisms of cross-presentation remain unsolved. How are antigens translocated into the cytosol? How are ER components recruited toward endosomes? What is the role of Sec22b and TLR ligands in this process? And probably most important: which of all these proposed mechanisms holds true *in vivo*? Are different cross-presentation pathways used *in vivo* by distinct cell types or antigens (e.g., particulate vs. soluble), or under different physiological conditions? Without any doubt, the elucidation of the molecular mechanisms underlying cross-presentation *in vivo* bears a high intrinsic potential to optimize various vaccination strategies. Therefore, future investigations will be required to shed more light into the exact pathways of cross-presentation and to solve remaining controversies. The publication of clearly contradicting data might suggest the need for common protocols to perform cross-presentation experiments, in particular in regard to cell culture procedures to generate the often used BM-DCs.

#### AUTHOR CONTRIBUTIONS

ME and SB designed the article and wrote the manuscript.

#### FUNDING

This work was supported by the German Research Foundation consortium SFB704 project A24.


through phagosomal destabilization. *Nat Immunol* (2008) 9:847–56. doi:10.1038/ni.1631


efficient antigen cross-presentation in dendritic cells. *Nat Commun* (2017) 8:1640. doi:10.1038/s41467-017-01601-5


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Embgenbroich and Burgdorf. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# Nuclear Receptor Nur77 Deficiency Alters Dendritic Cell Function

*Nina Tel-Karthaus1 , Esther D. Kers-Rebel <sup>1</sup> , Maaike W. Looman1 , Hiroshi Ichinose2 , Carlie J. de Vries <sup>3</sup> and Marleen Ansems1 \**

*1Department of Radiation Oncology, Radiotherapy & OncoImmunology Laboratory, Radboud Institute for Molecular Life Sciences, Radboud University Medical Center, Nijmegen, Netherlands, 2School of Life Science and Technology, Tokyo Institute of Technology, Yokohama, Japan, 3Department of Medical Biochemistry, Academic Medical Center, Amsterdam Cardiovascular Sciences, Amsterdam, Netherlands*

#### *Edited by:*

*Silvia Beatriz Boscardin, Universidade de São Paulo, Brazil*

#### *Reviewed by:*

*Dennis Qing Wang, Zhujiang Hospital of Southern Medical University, China Amanda Jane Gibson, Royal Veterinary College, United Kingdom Marc Poirot, Institut National de la Santé et de la Recherche Médicale (INSERM), France*

*\*Correspondence:*

*Marleen Ansems marleen.ansems@radboudumc.nl*

#### *Specialty section:*

*This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology*

> *Received: 01 May 2018 Accepted: 20 July 2018 Published: 03 August 2018*

#### *Citation:*

*Tel-Karthaus N, Kers-Rebel ED, Looman MW, Ichinose H, de Vries CJ and Ansems M (2018) Nuclear Receptor Nur77 Deficiency Alters Dendritic Cell Function. Front. Immunol. 9:1797. doi: 10.3389/fimmu.2018.01797*

Dendritic cells (DCs) are the professional antigen-presenting cells of the immune system. Proper function of DCs is crucial to elicit an effective immune response against pathogens and to induce antitumor immunity. Different members of the nuclear receptor (NR) family of transcription factors have been reported to affect proper function of immune cells. Nur77 is a member of the NR4A subfamily of orphan NRs that is expressed and has a function within the immune system. We now show that Nur77 is expressed in different murine DCs subsets *in vitro* and *ex vivo*, in human monocyte-derived DCs (moDCs) and in freshly isolated human BDCA1+ DCs, but its expression is dispensable for DC development in the spleen and lymph nodes. We show, by siRNA-mediated knockdown of Nur77 in human moDCs and by using Nur77−/− murine DCs, that Nur77-deficient DCs have enhanced inflammatory responses leading to increased T cell proliferation. Treatment of human moDCs with 6-mercaptopurine, an activator of Nur77, leads to diminished DC activation resulting in an impaired capacity to induce IFNγ production by allogeneic T cells. Altogether, our data show a yet unexplored role for Nur77 in modifying the activation status of murine and human DCs. Ultimately, targeting Nur77 may prove to be efficacious in boosting or diminishing the activation status of DCs and may lead to the development of improved DC-based immunotherapies in, respectively, cancer treatment or treatment of autoimmune diseases.

Keywords: dendritic cells, dendritic cell-based immunotherapy, nuclear receptors, NR4A, Nur77

# INTRODUCTION

Dendritic cells (DCs) are professional antigen-presenting cells. An important function of DCs is to instruct T cells to elicit immunity or tolerance (1, 2). Many factors contribute to the way DCs are shaped to elicit this function. Important factors are the type of pathogens that DCs encounter, such as bacteria or viruses, but also different microenvironmental factors in the tissues they reside in play a crucial role. DCs can be subdivided into classical or conventional DCs (cDC), interferonproducing plasmacytoid DCs (pDC), and monocyte-derived DCs (moDC) each with their own specialized function (3–5). Because of their crucial role in the immune system, different subsets of DCs are exploited in immune therapy (6–15). So far, treatment success is limited and functional knowledge on how DCs initiate and stably steer antitumor responses *in vivo* is important (13–15). Identification of transcription factors that control DC function in both immunity and tolerance is

**80**

highly relevant, as these factors may serve as targets to modulate DC activity and function for the development of more successful DC-based immunotherapies.

Different members of the nuclear receptor (NR) family of transcription factors and their ligands have been shown to affect immune cells, including DCs (16–20). NRs are ligand inducible transcription factors having among others, steroid hormones or cellular metabolites as ligands. Several members have been well studied and were shown to play an immune modulatory role in DCs. Another group of NRs are so called "orphan" NRs for which no natural ligand has been identified yet, and the existence of ligands is disputed. The NR4A subfamily of orphan receptors comprises three members, namely, Nur77 (NR4A1/TR3/ NGFI-B), Nurr1 (NR4A2/NOT/TINUR), and NOR-1 (NR4A3/ TEC/MINOR). Their activity appears to be primarily regulated at the expression level. The expression of the NR4As can be induced by a diverse range of signals, including fatty acids, stress, growth factors, cytokines, peptide hormones, and physical stimuli (21). Hallmark of this subfamily is to respond quickly to such changes in cellular environments and regulate gene expression in a ligandindependent manner.

Members of this subfamily have been shown to be involved in a wide variety of pathological conditions. They have been shown to be dysregulated in multiple cancer types and promote or suppress tumors depending on specific cellular and tissue context, subcellular localization, external stimuli, protein–protein interactions, and post-translational modifications in cancer cells [reviewed in Ref. (22)]. In addition, there is also increasing evidence that the NR4As play a role in neurodegenerative disorders such as Alzheimer's and Parkinson's disease by contributing to neuronal cell death *via* modulating mitochondrial function and ER stress by controlling intracellular levels of ROS and Ca2<sup>+</sup> and regulating cellular autophagy (23–26). Also in autoimmune-driven central nervous system (CNS) inflammation, the NR4A NRs have been shown to play an important role (27, 28).

NR4A receptors have emerged to play an important role within the immune balance by transcriptional regulation of cytokines and growth factors in macrophages (29, 30). In addition, they have been shown to be involved in the negative selection of self-reactive T cell clones in the thymus (31, 32) and are essential for thymic regulatory T cell development (33). Studies in Nur77<sup>−</sup>/<sup>−</sup> mice imply that Nur77 functions as a master regulator in the differentiation and survival of Ly-6C<sup>−</sup> monocytes (34, 35). Ly-6C<sup>+</sup> and Ly-6C<sup>−</sup> monocytes that do express Nur77 do not develop into moDCs (36). Thus, Nur77 expression is not required for the development into moDCs but is for differentiation of Ly-6C<sup>+</sup> monocytes into Ly-6C<sup>−</sup> "patrolling" monocytes (34, 36). Moreover, Nur77 has been shown to be involved in the polarization of macrophages toward an inflammatory phenotype important in atherosclerosis (37, 38).

We and others have recently reported expression of Nur77, Nurr1, and NOR-1 in murine DCs (39–43). Nurr1 has been shown to restrict the immunogenicity of bone marrow derived DCs (BMDCs) (43) and NOR-1 leads to activation-induced cell death in DCs (39), is important in DC migration (42), and is involved in TLR-mediated activation and gene expression of DCs (44). However, so far, the role of Nur77 expression in DCs remains elusive. We here set out to assess the expression kinetics and function of Nur77 in multiple subsets of murine and human DCs and its subsequent effect on inducing T cell activation, revealing a function as activation modulator for Nur77 in DCs. Knowledge regarding the possibilities in altering the activation status of DCs may prove to be beneficial in improving DC-based vaccination strategies.

#### MATERIALS AND METHODS

#### Mice

6- to 16-week-old C57BL/6J and Balb/C mice (Charles River), Nur77<sup>−</sup>/− mice (45) on a C57BL/6 background, and Nur77GFP mice [016607; C57BL/6-Tg(Nr4a1-EGFP/cre)820Khog/J; Jackson Laboratory] were housed under specific pathogen-free conditions in individually ventilated cage units at the Central Animal Laboratory (Nijmegen, The Netherlands). Standard laboratory chow and sterile drinking water were provided *ad libitum*. All animal experiments were approved by the Radboud University's Animal Welfare Body (*Instantie voor Dierenwelzijn IvD*) and the Animal Experiment Committee (*DierExperimentenCommissie, RUDEC*) that is recognized by the CCD (Central Authority for Scientific Procedures on Animals). The experiments were performed according to institutional, national, and European guidelines as stipulated in the *Wet op de dierproeven* and in the *Dierproevenbesluit*.

#### *In Vitro* Generation of Murine DCs

DCs were generated from murine BM isolated from the femur/ tibia of the mice. To obtain pDCs and cDCs, cells were cultured for 8–10 days (37°C, 10% CO2) in RPMI 1640 supplemented with 10% fetal calf serum (Gibco-BRL Life Technologies), 0.5% antibiotic–antimycotic (Gibco/Invitrogen), 1% ultra-glutamine (Lonza), 50 µM β-mercaptoethanol (Sigma-Aldrich), and 200 ng/ml human rFlt3L (PeproTech). Pure cell populations were isolated by labeling single cell suspensions with anti-SiglecH-FITC (eBiosciences) and anti-CD11c-APC antibodies for pDCs and cDCs, respectively. pDCs were positively sorted with anti-FITC microbeads, the negative fraction was subjected to positive selection with anti-APC microbeads (both Miltenyi Biotec, Germany) to obtain cDCs as described previously (40). CD103<sup>+</sup> murine DCs were generated by culturing BM cells in RPMI 1640 supplemented with 10% FCS, 0.5% antibiotic–antimycotic, 1% ultra-glutamine, 50 µM β-mercaptoethanol, 5 ng/ml mGM-CSF, and 200 ng/ml human rFlt3L, fresh medium was added at day 6, and cells were replated in fresh medium at day 9. Cells were harvested and used for experiments at day 14. The purity of the isolated DC subsets was ensured by flow cytometry.

#### Tumor Induction

The transgenic cell line 9464D was derived from spontaneous tumors from TH-MYCN transgenic mice on C57BL/6 background and were a kind gift from Dr. Orentas (NIH, Bethesda, MD, USA). 9464D cells were cultured in DMEM containing 10% fetal calf serum, 1% non-essential amino acids, 0.5% antibiotic– antimycotic, and 50 µM β-mercaptoethanol. For induction of tumors, 1 × 10e6 9464D cells were injected s.c. in 100 µl PBS on the right flank of the mice. Tumor growth was measured every 3–4 days using calipers. Spleen and lymph nodes (LNs) of the mice were taken when the tumor was more than 5 mm in diameter.

# Flow Cytometry

To obtain single cells for flow cytometric staining, murine spleen was passaged over a 100 µm cell strainer, and murine LNs were incubated in serum-free medium containing collagenase (Worthington) and DNAseI (Roche), later supplemented with 1 mM EDTA. *In vitro* generated human and murine DCs, and *ex vivo* isolated murine spleen and LN cells were stained using standard antibody staining protocols with antibodies listed in Table S1 in Supplementary Material. Cell viability was assessed by staining with fixable viability dye eFluor™ 450 (eBioscience). Samples were acquired on a FACS Verse (BD Bioscience), and data were analyzed with FlowJo software (Tree Star).

#### ELISA

Human and mouse IL-6, TNFα, IL-12p70, and human IFNγ present in the supernatant of DC cultures was measured using the ELISA kit (Thermo Fisher) according to the manufacturers protocol.

# Murine Type I IFN Bioassay

Type I IFN activity in the supernatant of murine pDCs was measured using L929 cells transfected with an interferon-sensitive luciferase construct (ISRE-L929) (46) with reference to a recombinant mouse IFN-β standard (Sigma). In short, pDC culture supernatants were added to ISRE-L929 IFN reporter cells and incubated for 4–6 h. Then, the cells were lysed in Passive Lysis Buffer (Promega), mixed with firefly luciferin substrate (Promega), and measured on a Victor3 Luminometer.

### Mixed Leukocyte Reaction (MLR) Murine DCs

After 16–24 h of stimulation with 1 µg/ml CpGB (1668, Sigma-Aldrich) or 4 µg/ml R848, pDCs, cDCs, or CD103<sup>+</sup> DCs were washed and co-incubated with carboxyfluorescein diacetate succinimidyl ester (CFSE)-labeled allogeneic BalB/C T cells. T cells were isolated using the T cell isolation kit (EasySep). The cells were co-incubated for 3 days in round-bottom 96-well cluster plates (Corning). T cell proliferation was measured by CFSE dilution by FACS.

# Generation of Human DCs

DCs were generated from cells isolated from buffy coats obtained from healthy volunteers (Sanquin, Nijmegen, The Netherlands) after written informed consent as per the Declaration of Helsinki. The study was approved by the Institutional Review Board of the Radboud University Nijmegen Medical Center, Commissie Mensgebonden Onderzoek. Peripheral blood mononuclear cells (PBMCs) were purified *via* Ficoll density gradient centrifugation (Lucron Bioproducts). moDCs were cultured as described previously (47). In short, plastic-adherent monocytes were cultured for 6 days in RPMI 1640 medium with 1% ultra-glutamine, 0.5% antibiotic–antimycotic, 10% (v/v) fetal calf serum, 300 U/ml IL4, and 450 U/ml GM-CSF (both Cellgenix). IL4 and GM-CSF were added again at day 3. To obtain fresh human myeloid dendritic cells (mDCs), CD14<sup>+</sup> cells were depleted from the PBMCs followed by BDCA1<sup>+</sup> DC isolation using the CD1c (BDCA1)<sup>+</sup> Dendritic Cell Isolation Kit (Miltenyi Biotec). Purity of the freshly isolated mDCs was ensured by flow cytometry.

#### Small Interfering RNA-Mediated Knockdown

For Nur77 silencing in human moDCs, the ON-TARGETplus SMARTpool NR4A1 (Dharmacon) containing four different Nur77 targeting siRNA oligos each 19 nt long was used. The irrelevant siRNA ON-TARGETplus Non-Targeting siRNA#1 (Dharmacon) was used as control. moDCs were electroporated at day 4 as described before (47). Electroporated DCs were stimulated with 1 µg/ml LPS (Sigma) or 4 µg/ml R848 (Enzo Life Sciences) at day 6. Supernatant was taken 24 h later.

### RNA Isolation and Quantitative PCR

Total RNA was isolated and cDNA was synthesized as described before (47). mRNA levels for the genes of interest were determined with a CFX96 sequence detection system (Bio-Rad) using the Faststart SYBR green mastermix (Roche) with SYBR Green as the fluorophore and gene-specific oligonucleotide primers. The primers for human porphobilinogen deaminase (PBGD), IL-6, TNFα, and IL-12 (47) and murine PBGD, TLR7, and TLR9 (40) were described previously. Other primers used (forward and reverse) are listed in Table S2 in Supplementary Material. Reaction mixtures and program conditions were used that were recommended by the manufacturer (Bio-Rad). Quantitative PCR data were analyzed with the CFX Manager V1.6.541.1028 software (Bio-Rad) and checked for correct amplification and dissociation of the products. As we described previously for human and murine DCs, mRNA levels of the genes of interest were normalized to mRNA levels of the housekeeping gene PBGD (19, 20, 40, 47, 48) and were calculated according to the cycle threshold method (49).

#### Human MLR

Human day 6 moDCs were pretreated with 1 or 10 µM 6-mercaptopurine (6-MP) (Sigma) or vehicle control (DMSO) for 8 h, before o/n stimulation with 4 µg/ml R848. At day 7, the medium was replaced with fresh DC medium, and allogeneic peripheral blood lymphocytes were added to the DCs, in a ratio of 1:10 (DCs:T cells) and cocultured for 144 h. Supernatant was taken for IFNγ measurements.

#### Statistical Analysis

In each experiment, at least three mice or human donors were used to be able to perform statistical testing. Each legend contains the information of the number of mice or human donors used including the statistics that was used to calculate significance. Statistical testing was performed using GraphPad Prism 5 software (GraphPad, La Jolla, CA, USA). A *P* < 0.05 was considered significant.

# RESULTS

# Nur77 Expression and Function in Murine DCs

As NR4A NRs are typical early response genes induced upon stimulation (29), we tested Nur77 expression in cDCs and pDCs 3 h after stimulation with a combination of the TLR7/8 ligand R848 and the TLR9 ligand CpG. DCs were differentiated from murine BM *in vitro* with FLT3L as this reflects physiologic DC development and gives rise to a mixture of both cDCs and pDCs (50). pDCs were detected as CD11cposB220posSiglecHpos and cDCs were defined as CD11cposB220negSiglecHneg and were sorted and stimulated as described before (40) (Figure S1A in Supplementary Material). In agreement with its classification as early response gene, Nur77 mRNA levels were strongly upregulated after 3 h of stimulation in murine pDCs as well as cDCs compared with freshly sorted cells (0 h) (**Figure 1A**). To further assess Nur77 expression kinetics, we used BM cells from transgenic Nur77 reporter mice, where the induction of the Nur77 promoter drives GFP expression (Nur77GFP) (51). In line with its mRNA expression, cDCs and to a lesser extend pDCs up regulate Nur77GFP already after 3 h of stimulation with CpG (**Figure 1B**). In addition to FLT3L-derived BMDCs, we tested Nur77GFP in BMDCs differentiated into Batf3-dependent CD103<sup>+</sup> DCs (CD11cposB220negCD103pos) when cultured with GM-CSF and FLT3L (52) (for gating strategy see Figure S1B in Supplementary Material). In CD103<sup>+</sup> DCs, there is also already prominent expression of Nur77GFP after 3 h stimulation with CpG (**Figure 1B**). Our data further show that the expression in cDCs was highest after CpG and LPS stimulation, whereas the expression was less pronounced in response to R848. pDC and CD103<sup>+</sup> DCs revealed highest expression of Nur77GFP after stimulation with CpG, compared with LPS and R848 (**Figure 1C**). These data indicate that in different types of *in vitro* generated DCs, Nur77 expression is quickly induced upon stimulation with inflammatory ligands, and that the expression in response to TLR-specific agonists varies in different DC subsets.

#### Nur77 Does Not Have a Major Impact on the Development of Murine DCs in Spleen and LNs

To test whether Nur77 expression is required for the development of DCs, we investigated the presence of different DC subsets in spleen and LNs of WT and Nur77<sup>−</sup>/<sup>−</sup> mice. We observed a small but significant increase in the percentage of total CD11chiMHCIIhi DCs and in CD11b+ DCs (CD11chi MHCIIhiSirpαposCD24negCD115negCD4pos) of the spleen of Nur77<sup>−</sup>/<sup>−</sup> mice relative to WT mice (**Figure 2A**; Figure S2A in Supplementary Material). The number of CD8α+ spleen DCs (CD11chiMHCIIhiSirpαnegCD24posFLT3pos) was similar between WT and Nur77<sup>−</sup>/<sup>−</sup> mice. Also in the LNs, the presence of resident

Figure 1 | Nur77 expression level in *in vitro* generated murine dendritic cells (DCs). (A) Plasmacytoid DCs (pDCs) and conventional DCs (cDCs) were sorted from FLT3L bone marrow cultures pooled from three mice per experiment and immediately lysed for RNA isolation (0 h) or stimulated for 3 h with a combination of R848 and CpG. mRNA expression levels of Nur77 were detected by qPCR analysis. Data shown are the mean of three independent experiments ± SEM, two-tailed unpaired *t*-test: \**P* < 0.05; \*\**P* < 0.01; and \*\*\**P* < 0.001. (B) *In vitro* generated cDCs, pDCs, or CD103+ DCs from Nur77GFP or control mice were stimulated for the indicated times with CpG, and GFP expression was determined by flow cytometry. Shown are the representative data of three mice. (C) Quantification of Nur77GFP expression in cDC, pDC, or CD103+ DC stimulated with CpG, LPS, or R848 for 0 or 3 h, presented as the geometric mean (MFI) ± SEM, two-way ANOVA with Sidak's multiple comparisons test (*n* = 3). *n.s.*, not significant; \**P* < 0.05; \*\**P* < 0.01; \*\*\**P* < 0.001; and \*\*\*\**P* < 0.0001.

(CD11chiMHCII<sup>+</sup>) and migratory DCs (CD11c<sup>+</sup>MHCIIhi) was comparable (**Figure 2B**; Figure S2B in Supplementary Material). Also in a transplantable autologous TH-MYCN 9464D mouse model of neuroblastoma (53), we did not observe differences in the presence of the different subsets of DCs in the spleen or (non)draining LN (Figures S3A,B in Supplementary Material). These data indicate that Nur77 is dispensable for DC development. Next, we tested the expression level of Nur77 in different DC subsets by analyzing DCs from the spleen and LNs from transgenic Nur77 reporter mice that express GFP upon activation of the Nur77 promoter. Nur77GFP was clearly expressed in CD11b<sup>+</sup> spleen DCs. The expression in CD8α+ spleen DCs was less well defined and consisted of a population expressing Nur77GFP at a very low level and a population expressing Nur77GFP to a similar level as the CD11b<sup>+</sup> DCs (**Figures 3A,B**). Resident DCs of inguinal and axillary LN expressed clear levels of Nur77GFP, in contrast to significantly lower expression in migratory DCs of these LN (**Figures 3C,D**). Mice bearing a neuroblastoma tumor showed a similar Nur77GFP expression pattern in DCs (Figures S3C,D in Supplementary Material) as in naïve mice. These data indicate that Nur77 expression does not have a major impact on the development and presence of different DC subsets in the spleen and LNs and that Nur77 is most abundantly expressed in CD11b<sup>+</sup> spleen DCs and resident DCs of different LNs.

#### Nur77-Deficient Murine DCs Have Altered Cytokine Production and T Cell Stimulatory Capacity

To assess the functional role of Nur77 in different murine DC subsets, we investigated cytokine production by murine Nur77<sup>−</sup>/<sup>−</sup> BMDCs after stimulation with different inflammatory stimuli. We found that Nur77<sup>−</sup>/<sup>−</sup> cDCs produced significantly more IL-6, TNFα, and IL-12 upon CpG and R848 stimulation (**Figure 4A**). Nur77<sup>−</sup>/<sup>−</sup> pDCs showed increased production of IL-6 and IL-12 upon R848 stimulation, whereas TNFα production was not affected. After stimulation with CpG, type I IFN production was much higher in Nur77<sup>−</sup>/<sup>−</sup> pDCs compared with WT pDCs (**Figure 4B**). CD103<sup>+</sup> DCs showed a stronger response to CpG than to R848, revealing increased production of IL-6, TNFα, and IL-12 (**Figure 4C**). To rule out the possibility that the increase in cytokine production was (partly) mediated by enhanced TLR expression, we profiled TLR7 and TLR9 expression in these cells.

TLR7 and TLR9 expression was similar in WT and Nur77<sup>−</sup>/<sup>−</sup> cDCs, whereas TLR7 expression was reduced in Nur77<sup>−</sup>/<sup>−</sup> pDCs (Figure S4 in Supplementary Material). In addition to cytokine production, we investigated the T-cell stimulatory capacities for Nur77-deficient DCs. To this end, control, CpG, or R848 stimulated DCs were added to an allogeneic MLR. All Nur77 deficient DC subsets were significantly more potent in inducing T cell proliferation than WT DCs upon stimulation with CpG (**Figure 4D**). These data indicate that Nur77 deficiency in DCs leads to enhanced cytokine production and subsequent increased T cell proliferation.

#### Nur77 Expression and Function in Human DCs

In addition to defining its expression and function in murine DCs we profiled Nur77 mRNA expression in human moDCs after stimulation with LPS and R848. In accordance with murine DCs, human moDCs quickly upregulated Nur77 mRNA expression and the expression remained stable for 24 h after stimulation with either LPS or R848 (**Figure 5A**). To investigate Nur77 expression in freshly isolated BDCA1<sup>+</sup> blood myeloid DCs, purified BDCA1<sup>+</sup> DCs were stimulated for different time periods with LPS or R848 (**Figure 5B**). Compared with moDCs, freshly isolated BDCA1<sup>+</sup> DCs had much higher expression levels of Nur77 expression under resting conditions. Stimulation with R848 led to a further increase of Nur77 expression, which diminished to lower levels 16 h after stimulation. These data indicate that in different subsets of human DCs Nur77 is expressed with varying expression levels.

#### Human Nur77-Modified DCs Have Altered Cytokine Production and T Cell Stimulatory Capacity

To test Nur77 function in human DCs, we silenced Nur77 expression in moDCs using a siRNA smartpool. Nur77 expression in moDCs decreased by 60–70% using siNur77 compared with control siRNA (siCTRL) (**Figure 6A**). These siNur77 targeted DCs had increased mRNA and protein expression of IL-6 and TNFα compared with siCTRL-treated DCs (**Figures 6B,C**), especially after R848 stimulation. Nur77-deficient DCs also showed enhanced IL-12 protein production. Profiling of TLR4, TLR7, and TLR8 expression (Figure S5 in Supplementary Material), revealed no change in TLR expression, indicating that the effect on cytokine production is not mediated *via* altered TLR expression. As NR4A family members have been reported to crosstalk with the NF-κB pathway (54), we investigated whether the enhanced cytokine production was dependent on NF-κB signaling. Blocking NF-κB signaling with the NF-κB inhibitor BAY11- 7082 inhibited IL-6 and TNFα production in siNur77 DCs and siCTRL DCs to the same level (**Figure 6D**), indicating that the enhanced expression of IL-6 and TNFα was indeed dependent on NF-κB signaling. We next determined the expression of CD40, CD86, and CCR7 in siNur77 DCs. While siNur77 and siCTRL DCs show similar expression of the co-stimulatory markers CD40 and CD86, a significantly lower percentage of CCR7<sup>+</sup> DCs were present in siNur77 DCs (**Figure 6E**). To further substantiate these data, we treated DCs with 6-MP, an activator of Nur77 (55–59). Treating DCs with 6-MP before stimulation with R848, led to a dose-dependent decrease of IL-6 and IL-12 production, while TNFα levels were not altered (**Figure 6F**). No effect of 6-MP on cell viability could be detected (Figure S6 in Supplementary Material). In line with decreased IL-6 and IL-12 production, DCs pretreated with 6-MP were less capable of inducing IFNγ production by T cells in an allogeneic MLR (**Figure 6G**). These data show that human Nur77-modified moDCs have altered NF-κB-dependent inflammatory responses that are important in inducing T cell activation.

#### DISCUSSION

Nuclear receptors have been shown to play a critical role in immune cell function, including members of the NR4A subgroup. However, the expression and function of Nur77 in different DC subsets has not been studied so far. We now show that Nur77 is expressed in different human as well as murine DC subsets. Its expression is rapidly upregulated upon stimulation with different TLR ligands. Deficiency of Nur77 leads to enhanced NF-κB dependent cytokine production and T cell stimulatory capacity of DCs, while stimulation with the Nur77 activator 6-MP limits cytokine production by DCs and its capacity to stimulate allogeneic T cells.

Nur77 expression has been shown to be essential in the differentiation and survival of Ly-6C<sup>−</sup> monocytes (34, 35), in the polarization of macrophages (37, 38, 60) and in the function and negative selection of T cells (31, 32). This NR is also expressed in infiltrating monocytes and monocyte-derived macrophages of the CNS that are important in experimental autoimmune encephalomyelitis (28) and in patrolling monocytes that control metastasis to the lung (61). We now show, in line with its classification as early response gene, that Nur77 expression is quickly upregulated in different human and murine DC subsets after stimulation with distinct TLR ligands *in vitro*. However, the expression levels in the different DC subsets and level of response toward diverse stimuli vary. We also found Nur77 expression in different subsets of DCs

(moDCs) were electroporated with a smartpool siRNA targeting Nur77 (siNur77) or a control siRNA (siCTRL). At day 6, cells were stimulated for 8 h with LPS, and Nur77 mRNA expression was detected by qPCR analysis (A), cytokine mRNA expression after 8 h of stimulation was measured by qPCR analysis (B), and cytokine levels were measured 24 h after stimulation with ELISA. (C) Electroporated moDCs were pretreated with Bay11-7082 and then stimulated with R848 for 24 h. Cytokine levels were measured by ELISA. (D) CD40, CD86, and CCR7 expression was determined by FACS analysis. (E) moDCs were pretreated with 6-mercaptopurine (6-MP) and then stimulated with TLRL for 24 h. Cytokine levels were measured by ELISA. (F) moDCs were pretreated with 6-MP and subsequently stimulated with R848, T cell stimulatory capacity was measured in an allogeneic mixed leukocyte reaction by measuring IFNγ production by ELISA (G). Data shown are the mean ± SEM. Two-way ANOVA with Bonferroni posttest (*n* = 3–12 independent donors): \**P* < 0.05; \*\**P* < 0.01; and \*\*\**P* < 0.001.

in the spleen and LNs directly in naïve and in tumor-bearing mice *ex vivo*. Expression was more pronounced in the CD11b<sup>+</sup> DCs of the spleen compared with CD8α+ DCs and higher in the resident than in the migratory DCs of the LNs. Previously, it has been shown that Nur77 is not required for the differentiation of Ly-6Chi monocytes into moDCs (36). We now also show that Nur77 deficiency does not have a major impact on the presence of different DC subsets in the spleen and LN at steady state conditions as well as mice bearing a neuroblastoma tumor. This confirms that in contrast to its expression in Ly-6C<sup>−</sup> monocytes, Nur77 expression is dispensable for the development of spleen and LN DCs.

Although most studies have reported that Nur77 has an antiinflammatory role in monocytes and macrophages (37, 38, 62), it has been shown that its overexpression in murine macrophages can lead to a pro-inflammatory response (63). Our data point towards an anti-inflammatory role in human and murine DC subsets. Nur77 deficiency in DCs leads to enhanced production of IL-6, TNFα, and IL-12 and subsequent enhanced T cell proliferation, while Nur77 activation leads to reduced IL-6 and IL-12 production and reduced T cell activation. It has been hypothesized that Nur77 acts to resolve inflammation in macrophages (38, 64) and based on our data we now suggest a similar role for Nur77 in DCs.

All NR4A family members, including Nur77, have been shown to modulate immune cell function *via* crosstalk to NF-κB (30, 38, 54, 65). Our data show that also in human DCs, Nur77 affects cytokine production by modulating the NF-κB pathway. It has been shown that Nur77 can affect the NF-κB pathway signaling in numerous ways (38, 63, 65–68). Besides modulating phosphorylation of p65 Ser536 and Ser529 in macrophages (38, 69), Nur77 has also been shown to directly interact with the p65 subunit of NF-κB (65, 66) and block p65 binding to DNA (65). Moreover, Nur77 can regulate TRAF6 auto-ubiquitination (67), important for NF-κB signal transduction (70–72). Future studies should reveal which mechanism underlies Nur77-mediated modulation of NF-κB signaling in DCs and whether different DC subsets or different inflammatory conditions involve specific ways of regulating NF-κB signaling.

While Nur77-deficient DCs show enhanced inflammatory responses, pretreating human DCs with 6-MP led to reduced inflammatory responses and a diminished capacity to induce IFNγ production by T cells in an allogeneic MLR. 6-MP is a nucleic acid analog and has been shown to enhance Nur77 transcriptional activity (55–59). Currently, it is being applied as an immunosuppressive drug for the treatment of several chronic inflammatory diseases such as inflammatory bowel disease, systemic lupus erythematosus, acute lymphoblastic leukemia of childhood, inflammatory myopathies, and rheumatoid arthritis and to prevent acute rejection in organ transplant patients (73–75). It has been shown that besides activating Nur77 (55, 56, 76) 6-MP can also activate the NR4A members Nurr1 (77) and NOR-1 (76) and inhibit the GTPase proteins Rac1 and Rac2 (78, 79). Therefore, the effect observed in moDCs may be a combined effect of 6-MP on the function of either of these proteins. In addition to 6-MP many other pharmacological compounds have been generated to modulate Nur77 function. Among them are different C-DIMs [synthetic 1,1-bis(3′-indolyl)-1-(substituted phenyl)methane analogs] (80), cytosporone B and its structural analogs (81, 82), and TMPA (ethyl 2-[2,3,4-trimethoxy-6- (1-octanoyl)phenyl]acetate) (83). They have been shown to regulate Nur77 function by modulating Nur77-dependent transactivation, influencing its expression levels, inducing nuclear export of Nur77 or affecting binding to other proteins (80–88). Many of these compounds have, as also shown for 6-MP, also Nur77-independent actions (85, 89, 90). In cancer cells, neuronal cells, as well as different immune cells, it has been shown that Nur77 function depends on tissue context, subcellular localization, external stimuli, protein–protein interactions, or post-translational modifications (22–26, 31, 32, 34, 35, 37, 38, 60). How Nur77 function in DCs is exactly regulated upon specific immune stimuli and whether that is different in different DC subsets is currently unknown. Future studies should aim at fully elucidating whether specific stimuli in different subsets of DCs and under specific (pathological) conditions affect Nur77 activation and thereby modulate DC function. More knowledge regarding the exact mechanism(s) of Nur77 activation in DCs will help to choose the best pharmacological compound targeting specific actions of Nur77 in DCs. This will not only be important in optimizing current DC-based immunotherapies but also when more generally targeting Nur77 in different cell types and pathological conditions.

Interestingly, in tumor cells, the natural steroid Dendrogenin A has been shown to stimulate expression of Nur77 *via* binding to LXRβ and induce lethal autophagy (91, 92), opening up new perspectives for cancer treatment (93). Moreover, it has been shown that Dendrogenin A, in addition to inducing growth control and improve overall survival in mice, also induces immune cell infiltration, including DCs, in the tumor (94). As LXR has been shown to affect DC differentiation, maturation and migration (95–101), it is tempting to speculate that part of these effects are mediated *via* regulation of Nur77 expression, especially when DCs are stimulated with Dendrogenin A.

One striking observation is that the percentage of CCR7 expressing human DCs was decreased in siNur77 treated moDCs. Interestingly, another member of the NR4A subfamily, NOR-1, has been shown to affect CCR7-dependent murine CD103<sup>+</sup> DC migration from tissues to LNs *in vivo* (42). Nevertheless, we did not observe a similar effect on CCR7 expression in *in vitro* cultured murine CD103<sup>+</sup> DCs (data not shown). In agreement with Park et al., we did not find differences in the number of migratory murine DCs present in the LN in Nur77<sup>−</sup>/<sup>−</sup> mice compared with WT mice, suggesting a less pronounced role for Nur77 in CCR7-dependent DC migration in mice. However, since NR4A family members are highly homologous proteins and can have redundant functions (102–104), it is also possible that the absence of Nur77 is compensated by NOR-1 in murine DCs.

Given that Nur77 modifies DC function with altered inflammatory responses, Nur77 may be an interesting therapeutic target to either boost or diminish the activation status of DCs in DC-based vaccination strategies in cancer or treatment of autoimmune diseases, respectively.

#### ETHICS STATEMENT

All animal experiments were approved by the Radboud University's Animal Welfare Body (AWB) (*Instantie voor Dierenwelzijn IvD*) and the Animal Experiment Committee (*DierExperimentenCommissie, RUDEC*) that is recognized by the CCD (Central Authority for Scientific Procedures on Animals). The experiments were performed according to institutional, national, and European guidelines as stipulated in the *Wet op de dierproeven* (WOD) and in the *Dierproevenbesluit*. All experiments involving human material were carried out after obtaining written informed consent from all subjects as per the Declaration of Helsinki. The study was approved by the Institutional Review Board of the Radboud University Nijmegen Medical Center, Commissie Mensgebonden Onderzoek.

# AUTHOR CONTRIBUTIONS

NT-K and MA planned and performed experiments. EK-R and ML performed experiments. HI generated and provided mice. NT-K, CV, and MA contributed to the interpretation of the data. MA wrote the manuscript. NT-K, EK-R, ML, HI, and CV contributed to the review of the manuscript. MA designed the study.

# REFERENCES


#### ACKNOWLEDGMENTS

We are grateful for the constructive discussions and critically reading of the manuscript to Prof. G. J. Adema. This work is supported by project number 016.156.093 of the Netherlands Organisation for Scientific Research (NWO-Veni) (to MA). NT-K is recipient of a PhD grant from the Radboud University Nijmegen Medical Centre. MA is a recipient of a long-term fellowship (BUIT 2012-5347) from the Dutch Cancer Society.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.01797/ full#supplementary-material.


and regulator of TLR-induced cytokine production. *J Immunol* (2012) 189(1):138–45. doi:10.4049/jimmunol.1103709


104. Tontonoz P, Cortez-Toledo O, Wroblewski K, Hong C, Lim L, Carranza R, et al. The orphan nuclear receptor Nur77 is a determinant of myofiber size and muscle mass in mice. *Mol Cell Biol* (2015) 35(7):1125–38. doi:10.1128/MCB.00715-14

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

*Copyright © 2018 Tel-Karthaus, Kers-Rebel, Looman, Ichinose, de Vries and Ansems. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.*

# The Nontoxic Cholera B Subunit Is a Potent Adjuvant for Intradermal DC-Targeted Vaccination

Laura Antonio-Herrera1,2, Oscar Badillo-Godinez <sup>3</sup> , Oscar Medina-Contreras <sup>4</sup> , Araceli Tepale-Segura<sup>1</sup> , Alberto García-Lozano<sup>1</sup> , Lourdes Gutierrez-Xicotencatl <sup>3</sup> , Gloria Soldevila<sup>5</sup> , Fernando R. Esquivel-Guadarrama<sup>6</sup> , Juliana Idoyaga<sup>7</sup> \* and Laura C. Bonifaz <sup>1</sup> \*

#### Edited by:

Silvia Beatriz Boscardin, Universidade de São Paulo, Brazil

#### Reviewed by:

Luis C.S. Ferreira, Universidade de São Paulo, Brazil Jesus Hernandez, Centro de Investigación en Alimentación y Desarrollo (CIAD), Mexico Adriana Flores-Langarica, University of Birmingham, United Kingdom

#### \*Correspondence:

Juliana Idoyaga jidoyaga@stanford.edu Laura C. Bonifaz labonifaz@yahoo.com

#### Specialty section:

This article was submitted to Vaccines and Molecular Therapeutics, a section of the journal Frontiers in Immunology

> Received: 24 May 2018 Accepted: 06 September 2018 Published: 27 September 2018

#### Citation:

Antonio-Herrera L, Badillo-Godinez O, Medina-Contreras O, Tepale-Segura A, García-Lozano A, Gutierrez-Xicotencatl L, Soldevila G, Esquivel-Guadarrama FR, Idoyaga J and Bonifaz LC (2018) The Nontoxic Cholera B Subunit Is a Potent Adjuvant for Intradermal DC-Targeted Vaccination. Front. Immunol. 9:2212. doi: 10.3389/fimmu.2018.02212 <sup>1</sup> Hospital de Especialidades, Centro Médico Nacional Siglo XXI, Instituto Mexicano del Seguro Social, Unidad de Investigación Médica en Inmunoquímica, Mexico City, Mexico, <sup>2</sup> Universidad Nacional Autónoma de México, Mexico City, Mexico, <sup>3</sup> Centro de Investigación Sobre Enfermedades Infecciosas, Instituto Nacional de Salud Pública, SS, Cuernavaca, Mexico, <sup>4</sup> Immunology and Proteomics Laboratory, Mexico Children's Hospital "Federico Gómez", Mexico City, Mexico, <sup>5</sup> Departamento de Inmunología, Instituto de Investigaciones Biomédicas, Universidad Nacional Autónoma de México, Mexico City, Mexico, <sup>6</sup> Laboratorio de Inmunología Viral, Facultad de Medicina, UAEM, Cuernavaca, Mexico, <sup>7</sup> Department of Microbiology and Immunology, Stanford University School of Medicine, Stanford, CA, United States

CD4<sup>+</sup> T cells are major players in the immune response against several diseases; including AIDS, leishmaniasis, tuberculosis, influenza and cancer. Their activation has been successfully achieved by administering antigen coupled with antibodies, against DC-specific receptors in combination with adjuvants. Unfortunately, most of the adjuvants used so far in experimental models are unsuitable for human use. Therefore, human DC-targeted vaccination awaits the description of potent, yet nontoxic adjuvants. The nontoxic cholera B subunit (CTB) can be safely used in humans and it has the potential to activate CD4<sup>+</sup> T cell responses. However, it remains unclear whether CTB can promote DC activation and can act as an adjuvant for DC-targeted antigens. Here, we evaluated the CTB's capacity to activate DCs and CD4<sup>+</sup> T cell responses, and to generate long-lasting protective immunity. Intradermal (i.d.) administration of CTB promoted late and prolonged activation and accumulation of skin and lymphoid-resident DCs. When CTB was co-administered with anti-DEC205-OVA, it promoted CD4<sup>+</sup> T cell expansion, differentiation, and infiltration to peripheral nonlymphoid tissues, i.e., the skin, lungs and intestine. Indeed, CTB promoted a polyfunctional CD4<sup>+</sup> T cell response, including the priming of Th1 and Th17 cells, as well as resident memory T (RM) cell differentiation in peripheral nonlymphoid tissues. It is worth noting that CTB together with a DC-targeted antigen promoted local and systemic protection against experimental melanoma and murine rotavirus. We conclude that CTB administered i.d. can be used as an adjuvant to DC-targeted antigens for the induction of broad CD4<sup>+</sup> T cell responses as well as for promoting long-lasting protective immunity.

Keywords: anti-DEC205, CTB, adjuvant, skin, memory, T cells, dendritic cells

# INTRODUCTION

Formulation of successful subunit vaccines requires the optimal combination of antigen and adjuvant to ensure the development of long-lasting protective immunity. Expected responses should include the development of memory CD4<sup>+</sup> T cells, which play a major role in protecting against a myriad of pathogens (1–3) and against tumors (4, 5). To achieve this goal, delivering antigens via monoclonal antibodies (mAbs) targeting DCs, in combination with strong adjuvants, is one of the most promising strategies.

The administration of anti-DEC205-antigen mAbs can increase the efficiency of MHC-II antigen presentation relative to soluble antigen by 300-fold (4, 6). In combination with strong adjuvants, e.g., Poly IC, anti-CD40 mAbs, CpG, and flagellin (4, 7–9), it induces T helper (Th) cell differentiation and it mediates long-lasting immunity against experimental melanoma, malaria and influenza (4, 7, 10, 11). Moreover, DC-targeted vaccination can induce polyfunctional memory CD4<sup>+</sup> T cells that produce IFNγ, TNFα and IL-2 (7, 9). Therefore, DC-targeted vaccination serves as a powerful strategy to promote protective CD4<sup>+</sup> T cell responses.

Unfortunately, due to their toxicity, the adjuvants mentioned above are not approved for human use. Only synthetic derivatives, such as AS04 and phosphorothioate-backbone CpG adjuvants are undergoing trials with humans (12). However, these synthetic derivatives have shown adverse effects in murine models including splenomegaly, lymphoid follicle destruction and immunosuppression (12), which make them less promising for human use. Therefore, there is a need to identify adjuvants, which can be co-administered with DC-targeted antigens, for the induction of protective CD4<sup>+</sup> T cell responses in humans.

CTB has been proven to be safe for human use as an adjuvant (13–15). Its use has been approved for the killed whole-cell monovalent vaccine (WC-rBS) against cholera in humans, which has only induced mild adverse effects in a few individuals, and it has been safe for and well-tolerated by immunocompromized subjects (16). Unfortunately, the capacity of CTB to activate DCs is controversial. Some in vitro studies using bone marrow-derived DCs (BMDCs) and macrophages (BMDM) show that CTB can promote expression of TLRs, CD86 and production of IL-5, IL-12p70, IL-6, IL-10, IL-3, G-CSF, MIP-2 and eotaxin, as well as it can activate the NFkB pathway (17, 18). In contrast, other studies suggest that CTB does not induce the activation of ex vivo DCs (19–21). Therefore, it is necessary to evaluate the capacity of CTB to activate DCs in vivo.

Still, several reports have shown that CTB can be used as a strong adjuvant. When admixed or conjugated with pathogen derived antigens, it can promote the generation of long-lived CD4<sup>+</sup> T cells. Such responses mediate systemic immunity against several pathogens, including the influenza virus (22), Helicobacter pylori (23), Streptococcus pneumoniae (24), Bordetella pertussis (25), and Francisella tularensis (26). Furthermore, we have previously demonstrated that i.d. administration of soluble antigens in combination with CTB promotes CD4<sup>+</sup> T cell activation and differentiation of Th1 and Th17 cells (27). However, CTB adjuvant's capacity has never been tested with DC-targeted antigens administered i.d. Here, we asked whether CTB co-administration with anti-DEC205-antigen mAbs could induce DC activation and consequently promote long-lasting and protective CD4<sup>+</sup> T cell responses.

#### MATERIALS AND METHODS

#### Mice

WT C57BL/6 mice and transgenic mice expressing green fluorescent protein (GFP) under the major histocompatibility complex class II molecule promoter were obtained from Unidad de Medicina Experimental, UNAM animal facility. BALB/c mice were obtained from INSP, SS animal facility. OT-II CD45.1 mice were obtained from Instituto de Investigaciones Biomédicas, UNAM animal facility. All animal experiments were performed following the Institutional Ethics Committee and the Mexican national regulations on animal care and experimentation. Experiments with DO11.10 Thy1.1<sup>+</sup> mice were performed at the Department of Microbiology and Immunology of the School of Medicine, at Stanford University, following institutional guidelines. Mice were sex (male or female)- and age (7–10 weeks)-matched.

# CD4<sup>+</sup> T Cell Enrichment

Skin-draining lymph nodes (SDLN), spleen, and mesenteric lymph nodes were collected from OT-II CD45.1<sup>+</sup> or DO11 Thy1.1<sup>+</sup> mice, placed in RPMI medium (Gibco) supplemented with 5% fetal bovine serum (FBS) (HyClone), 300µg/mL glutamine (Gibco) and 100 U/mL penicillin/100µg/mL streptomycin (Biowest), and mashed separately to obtain cell suspensions. Red blood cells were lysed with RBC lysis buffer (Biolegend). Both LN and spleen suspensions were incubated for 30 min on ice with homemade rat hybridoma supernatants against CD8 (2.43), B cells (B220), MHCII-expressing cells (TIB120), and macrophages (F4/80). Next, cells were washed, suspended in supplemented RPMI and poured into petri dishes previously coated with rat anti-IgG (ThermoFisher) for 40 min at 4 ◦C. Non-adherent cells were recovered, washed and suspended in PBS for injection through the retro orbital vein.

#### Cell Transfer and Immunization

Congenic mice received 4.5–5 × 10<sup>6</sup> CD4<sup>+</sup> T cells intravenously (i.v.). After 24 h, anesthetized mice were immunized i.d. in both ears (or in the right flank for melanoma and viral challenge experiments) with 1 µg of anti-DEC205-OVA (containing ∼0.5 µg of OVA protein), 1 µg of a control mAb-OVA without receptor affinity or 3–30 µg of soluble unconjugated OVA in the presence or absence of 10 µg of CTB (Sigma-Aldrich). For proliferation experiments mice received 4.5–5 × 10<sup>6</sup> CFSElabeled CD4<sup>+</sup> T cells 24 h before i.d. administration of 1 µg of anti-DEC205-OVA or 1, 3, or 10 µg of soluble unconjugated OVA. For prime/boost experiments, mice were immunized i.d. in both ears with 1 µg of anti-DEC205-OVA or 3 µg of soluble unconjugated OVA plus 10 µg of CTB. After 15 days, mice

**Abbreviations:** CTB, cholera B subunit; i.d., intradermal; i.v., intravenous; s.c., subcutaneous; DC, dendritic cell; Ab, antibody; mAb, monoclonal antibody; SDLN, skin draining lymph node; T RM, resident memory T cell; T CM, central memory T cell; T EM, effector memory T cell.

received i.p. 1 µg of anti-DEC205-OVA or 3 µg of soluble unconjugated OVA.

#### Tissue Processing

At 3 or 7 days post-immunization, mice were sacrificed to collect SDLN and skin. SDLN were enzymatically digested with 0.25 mg/mL Liberase TL (Roche) and 0.125 mg/mL DNAse (Roche) for 25 min at 37◦C. Skin cell suspensions were also obtained by enzymatic digestion with 0.25 mg/mL Liberase TL and 0.125 mg/mL DNAse for 45 min at 37◦C, then chopped with scissors and incubated under the same conditions with constant shaking. Next, enzymatic digestion was stopped by adding 0.5µM EDTA, and cell suspensions were filtered through a 70µm strainer (Corning), followed by the addition of 0.125 mg/mL DNAse. Finally, cells were washed, counted, stained and/or re-stimulated as needed.

To obtain cells from the lungs, mice were sacrificed 7 days post-immunization. Lungs were rinsed with water to remove excess blood, placed into polypropylene tubes and chopped into small pieces to digest with 0.25 mg/mL Liberase TL (Roche) and 0.125 mg/mL DNAse (Roche) for 1 h at 37◦C with constant shaking. Next, enzymatic digestion was stopped by adding 0.5µM EDTA, and cell suspensions were filtered through a 70µm strainer (Corning), followed by the addition of 0.125 mg/mL DNAse. Next, cells were lysed with the RBC lysis buffer (Biolegend). Finally, cells were washed, counted and stained.

Isolation of intestinal cells was performed as previously described elsewhere (28). Briefly: intestines were removed and carefully cleaned off their mesentery lymph nodes and Peyer's patches were excised. Intestines were opened longitudinally, washed off fecal contents, cut into pieces 0.5 cm in length, and subjected to two sequential 20-min incubations in HBSS with 5% FCS and 2 mM EDTA at 37◦C with agitation to remove epithelial cells. After each incubation step, media containing epithelial cells and debris were discarded. The remaining tissue was minced and incubated for 20 min in HBSS with 5% FCS, 1 mg/ml collagenase IV and 40 U/ml DNase I at 37◦C in agitation. Cell suspensions were collected and passed through a 100-µm strainer and pelleted by centrifugation at 300 g. Cells were counted and divided for in vitro re-stimulation and cell surface staining.

#### In vitro Re-stimulation

Cells were resuspended in RPMI medium supplemented with 10% FBS, 300µg/mL glutamine, 100 U/mL penicillin/100µg/mL streptomycin, 110µg/mL sodium pyruvate and 10µM βmercaptoethanol. SDLN cells were incubated for 48 h with OVA peptide 323–339 (in vivogen), followed by cell stimulation cocktail plus protein transport inhibitor, added according to the manufacturer's instructions (eBioscience), and cells were incubated for an additional 4 h at 37◦C. Cells from the skin and intestine were only re-stimulated with cell cocktail stimulation plus protein transport inhibitor for 4 h without OVA.

#### Flow Cytometry

To allow for counting, cells were stained with anti-CD45-PECy7 (Biolegend) and DAPI (ThermoFisher), immediately mixed with CountBright absolute counting beads (ThermoFisher), acquired for flow cytometry. Cell surface staining was performed first by blocking Fc receptors (supernatant of 2.4G2 hybridoma against CD16/32) and then by staining using the following antibodies: anti-CD45-APC (Biolegend) or -PECy7 (Biolegend), anti-CD4-APC-Cy7 (Biolegend), anti-TCRVβ5.1, 5.2-PECy7 (Biolegend) or anti-Vα2-FITC (eBioscience), anti-CD45.2-Percp-Cy5.5 (Biolegend) or anti-CD45.1-Percp-Cy5.5 (Biolegend), anti-CD69-PE (ebioscience), and anti-CCR7-FITC (Biolegend). LIVE/DEAD Fixable Aqua (Thermofisher) staining was included. For DC analysis the following Abs were used: anti-CD45-APC (Biolegend), anti-Ter119- Percp-Cy5.5, anti-CD3-Percp-Cy5.5, anti-CD19-PercpCy5.5, anti-CD44b-Percp-Cy5.5, anti-MHCII-FITC (Biologend), and CD86-PE (eBioscience). To achieve intracellular staining, cell surface staining was first performed, followed by fixation and permeabilization using the intracellular fixation and permeabilization buffer set (Thermofisher), according to the manufacturer's instructions. To stain cytokine and transcription factors, the True-Nuclear transcription factor buffer set (Biolegend) was used according to the manufacturer's instructions. Intracellular staining included anti-IL-17-PE (BD Bioscience), anti-IFNγ-APC (Biolegend), anti-T-bet-BV421 (BD Biosciences), or anti-RORγT-APC (Thermofisher). Cells were acquired in a BD FACSCanto II or BD LSRFortessa cytometer (Becton, Dickinson and company). Data were analyzed with FlowJo software (Tree Star, Inc.).

#### Melanoma Challenge

Mice were transferred with OT-II CD45.1<sup>+</sup> cells 24 h before i.d. immunization with 1 µg of anti-DEC205-OVA or with 3 µg of soluble untargeted OVA ± 10 µg of CTB. After 30 days, mice received 2.5 × 10<sup>5</sup> MO4 cells subcutaneously (s.c.) in the right flank and then they were monitored for 21 days for survival. Alternatively, C57BL/6 naive mice were challenged i.v. in the tail vein 30 days after immunization to induce metastatic nodules in the lungs. For some experiments, anti-DEC205-OVAvaccinated mice received i.p. 250 µg of anti-CD4 Ab (GK1.5, in house) or isotype control Ab (eBRG1, in house) as follows: 1 day before MO4 inoculation, on the day of MO4 inoculation and every 3 days after MO4 inoculation, up to day 12. Sixteen days after MO4 inoculation, mice were sacrificed and lungs were harvested for metastatic nodule count as described elsewhere (29). Briefly: lungs were rinsed with water to remove excess blood and bleached with Feket's solution, and metastatic nodules were counted under a stereoscope (Leica Microsystems). Uncountable nodules were reported as >250.

#### Viral Challenge

BALB/c mice were immunized i.d. in the right flank with 23 µg of anti-DEC205-VP6 (corresponding to 1.5 µg of VP6) or with 3 µg of in vitro synthetized soluble untargeted VP6 (produced from the murine rotavirus Ew in vitro with the Rapid Translation System, Roche), in the presence of 10 µg of CTB. After 20 days, mice were orally challenged with 1 × 10<sup>4</sup> focus forming units of murine RV EDIMWT as described elsewhere (30). For prime/boost experiments, mice were i.d. immunized with anti-DEC205-VP6 or 3 µg of VP6 plus 10 µg of CTB and, after 15 days, mice received i.p. anti-DEC205-VP6 or VP6 (same dose as before). For CD4<sup>+</sup> T cell depletion experiments, mice immunized with anti-DEC205-VP6 received either 250 µg of anti-CD4 Ab (GK1.5, in house) or isotype control Ab (eBRG1, in house) as follows: 3 days before the viral challenge, on the day of the challenge and 3 days after the challenge. Seven days after boost, mice were orally challenged with 1 × 10<sup>4</sup> focus forming units of murine RV EDIMWT. Stool samples were collected daily for 8 days and kept at −20◦C for further analysis of viral load by sandwich ELISA. Protection against infection was calculated as % protection = 100% – [area under the curve of the experimental group (Absorbance at 405 nm)/area under the curve of the control group (Absorbance at 405 nm)] × 100%. This represents a decrease in the quantity of rotavirus antigen shed after immunization, relative to control mice, during the 8 days after the challenge.

#### ELISA

Viral load in the stool was determined by sandwich ELISA, as described elsewhere (30). Briefly: diluted stool samples were poured into 96-well plates (Costar) previously coated with a goat polyclonal antibody (Ab) against different strains of RV (in house). After 2 h at 37◦C, plates were washed, and a rabbit polyclonal Ab against RV RRV was added. After 1 h at 37◦C, plates were washed and a PA-conjugated goat anti-rabbit IgG (Zymed) was added, which was incubated for 1 h at 37◦C. Finally, after washing, the substrate (p-nitrophenyl phosphate, disodium; Sigma) was added, and plates were developed for 30–45 min at 37◦C. The absorbance at 405 nm was read with a 96-well plate reader (BIO-TEK Instruments, Burlington, VT).

# DC Activation

GFP-MHC-II mice received 10 µg of CTB or PBS i.d. in the ear. After 12, 24 or 72 h, epidermal sheets were obtained, stained with anti-CD86-PE (eBioscience), mounted with VectaShield (Vector Laboratories) and sealed. The images were obtained with a Leica TCS SP8x Confocal Microscope (Wetzlar, Germany) and analyzed with Leica Application Suite Advanced Fluorescent Lite software (Leica Microsystems, Mannheim, Germany). Alternatively, C57BL/6 mice received 10 µg of CTB or PBS i.d. in the ear. After 24, 72 h, or 7 days, mice were sacrificed to collect SDLN and skin. Tissues were processed and stained to be analyzed by flow cytometry.

#### Statistics

Statistical analysis was performed using Prism 6.0 (GraphPad Software Inc., La Jolla, CA, USA). Statistical significance was calculated when comparing two groups, using unpaired twotailed Student's t-test. For comparison of more than two groups, one-way or two-way ANOVA with Tukey's multiple comparison test was used. A P-value <0.05 was considered significant.

# RESULTS

#### CTB Induces Late and Prolonged Activation and Accumulation of SDLN and Skin DCs

We first aimed to determine whether CTB could induce in vivo activation of DCs. To this end, epidermal sheets of GFP-MHC-II mice were obtained at 12, 24, or 72 h after i.d. administration of CTB; followed by staining with fluorescent Ab specific for CD86. Using confocal microscopy, we observed co-expression of CD86 by epidermal MHC-II<sup>+</sup> cells, only after 72 h, and at no earlier time (**Supplementary Figure 1**). Next, we characterized skin DCs as viable CD45+Lineage−CD11c+MHC-II<sup>+</sup> cells by multiparametric flow cytometry (**Supplementary Figure 2A**). We confirmed that CTB induces in vivo activation of DCs after 72 h by overexpression of CD86 (**Figure 1A**) and, interestingly, their accumulation in the inoculation site as well. It was striking that both the activation and accumulation were sustained 7 days after the i.d. administration of CTB (**Figure 1A**).

Next, we asked whether CTB could induce accumulation of activated DCs in the SDLN. To answer this question, we analyzed SDLN cells by multiparametric flow cytometry, which allowed us to discriminate between migrating (CD11c+MHC-IIhi) and resident (CD11c+MHC-IIlow) DCs (**Supplementary Figure 2B**). Seventy-two hours after its administration, CTB induced the accumulation of migratory DCs in the SDLN, which displayed an increased expression of CD86 compared to the PBS control (**Figure 1B**). The accumulation of migrating DCs with an activated phenotype dropped after 72 h. However, it was still higher than the PBS control after 7 days. Interestingly, CTB also induced an increased expression of CD86 on resident DCs as well as their accumulation after 7 days (**Figure 1C**). It is worth noting that the accumulation and activation of DCs took place only at the inoculation site and the draining lymph node, as we did not find either effect on a distal organ, i.e., the mesenteric lymph nodes (**Supplementary Figure 2C**).

As a whole, our results demonstrate that skin administration of CTB acts as a potent stimulus to induce late and prolonged accumulation and activation of lymphoid-resident and skin DCs.

#### CTB Co-administration With a DC-Targeted or Soluble Antigen Promotes Expansion and Differential Activation of CD4<sup>+</sup> T Cells

To study the development of antigen specific CD4<sup>+</sup> T cell responses we used a DC-targeted OVA antigen and, for comparison, soluble OVA antigen. After 3 days, we observed a 20-fold increase in the proliferation of CD4<sup>+</sup> T cells after the i.d. inoculation of 1 µg of anti-DEC205-OVA, compared to 10 µg of soluble OVA (**Supplementary Figure 3B**). Furthermore, cells undergoing the last rounds of proliferation showed downregulation of CD69, which was more pronounced in cells from anti-DEC205-OVA-inoculated mice (**Figure 2A**). CD69 is rapidly activated after TCR engagement, but it decreases as T cells divide (31, 32). Even so, similar numbers of OVA-specific CD4<sup>+</sup> T cells were found in the SDLN of mice administered with 3 µg of soluble OVA or with 1 µg of anti-DEC205-OVA (**Figure 2B**).

Next, we evaluated the outcome of CTB co-administration in T cell activation. Three days post-immunization, cells from mice administered with CTB plus a DC-targeted antigen remained low for CD69 expression; while a soluble antigen admixed with CTB resulted in higher expression of CD69 (**Figure 2C**). The

Graphs depicting the percentage, absolute cell numbers of DCs and geometric median fluorescence intensity (MFI) of CD86 on DCs in the skin. Mean ± SD, N = 4–6, data pooled from two independent experiments. One-way ANOVA with Tukey's multiple comparisons test (\*P < 0.05, \*\*P < 0.005, \*\*\*P < 0.0005, \*\*\*\*P < 0.0001). (B) Migratory and (C) resident DCs from the SDLN were gated as in Supplementary Figure 2B. Graphs of percentage, total numbers of DCs and geometric MFI of CD86 on DCs. Mean ± SD, N = 4–6, data pooled from two independent experiments. One-way ANOVA with Tukey's multiple comparisons test (\*P < 0.05, \*\*P < 0.005, \*\*\*P < 0.0005, \*\*\*\*P < 0.0001).

expression of CD69 promotes retention of T cells in the lymph node; while its deregulation allows cells to migrate to distal peripheral tissues (31, 32). Thus, similar to others (6, 33), our data suggest the possibility of systemic dissemination of CD4<sup>+</sup> T cells after DC-targeted antigen inoculation.

After 7 days, we observed a significant effect on T cell expansion, as CTB co-administered with a DC-targeted antigen promoted larger numbers of OVA-specific CD4<sup>+</sup> T cells (**Figure 2D**). This result was dependent on the antigen being targeted to DCs, since the administration of the isotype Ab conjugated with OVA, with or without CTB, did not promote expansion (**Figure 2D**; **Supplementary Figure 3B**). CTB co-administration with soluble OVA promoted larger accumulation of CD4<sup>+</sup> T cells in the SDLN as compared with the DC-targeted OVA group (**Figure 2D**), and it was consistent with a higher expression of CD69.

We next asked if CTB could promote the migration of antigenspecific CD4<sup>+</sup> T cells to the inoculation site. After 7 days of i.d. immunization, we observed a large infiltration of OT-II CD45.1<sup>+</sup> cells in the skin, which was promoted by the co-administration of CTB and not by the antigen alone (**Figure 2E**). Strikingly, higher numbers of OVA-specific T cells were observed in the skin of mice immunized with soluble OVA along with CTB compared to the DC-targeted vaccination group.

All together, these data demonstrate that CTB can be used as a strong adjuvant with a DC-targeted or soluble antigen to promote local expansion of antigen-specific CD4<sup>+</sup> T cells in the SDLN, and to induce their efficient migration to the inoculation site (i.e., skin). Remarkably, our data suggest that a DC-targeted

adoptively transferred with OT-II CD45.1<sup>+</sup> cells, 24 h later they were immunized i.d. in both ears, as indicated, and 3 or 7 days later, they were sacrificed for SDLN and skin harvesting. (A) Representative dot plot of CFSE dilution and CD69 expression by SDLN OT-II cells 3 days after inoculation of anti-DEC205-OVA or soluble OVA and (B) total numbers of OT-II cells. (C) Geometric median fluorescence intensity (MFI) of CD69 by OT-II cells 3 days after anti-DEC205-OVA or soluble OVA ± CTB's i.d. administration. Mean ± SD, N = 4–6, data pooled from four independent experiments. One-way ANOVA with Turkey's multiple comparisons test (ns, P > 0.05, \*P < 0.05, \*\*P < 0.005, \*\*\*P < 0.0005). (D) Representative dot plots and total number of SDLN OT-II cells 7 days after anti-DEC205-OVA or soluble OVA ± CTB's i.d. administration. Mean ± SD, N = 5–8 data pooled from four independent experiments. One-way ANOVA with Tukey's multiple comparisons test (\*P < 0.05, \*\*\*\*P < 0.0001). Transferred cells recovered from the SDLN were identified as viable CD4+CD45.2−TCRVβ 5.1, 5.2<sup>+</sup> T cells (Supplementary Figure 3A). (E) Representative dot plot and total numbers of migrating OT-II cells identified as viable CD45+CD4+CD45.2−TCRVβ 5.1, 5.2<sup>+</sup> (Supplementary Figure 3C). Mean ± SD, N = 4–6, data pooled from four independent experiments. One-way ANOVA with Tukey's multiple comparisons test (\*\*P < 0.005, \*\*\*\*P < 0.0001).

antigen induces differential activation of CD4<sup>+</sup> T cells, which might impact their differentiation and, possibly, the differential anatomical localization of CD4<sup>+</sup> T cells after DC-targeted or soluble antigen immunization.

#### CTB Promotes a Combined Th1/Th17 Response When Co-administered With a DC-Targeted Antigen

We next asked whether CTB admixed with a DC-targeted antigen or a soluble antigen could promote the differentiation of CD4<sup>+</sup> T cells into Th1 or Th17 cells. At day 7 post-immunization, we observed antigen-specific IFNγ <sup>+</sup> cells in the SDLN, induced by the administration of CTB in combination with a DC-targeted antigen or a soluble antigen (**Supplementary Figure 4A**; **Figure 3A**). Remarkably, only DC-targeted vaccination promoted significant differentiation of IL-17<sup>+</sup> CD4<sup>+</sup> T cells (**Supplementary Figure 4A**; **Figure 3A**). These results were confirmed in the DO11.10 model (**Supplementary Figure 5A**), which is prone to Th2 and Treg responses. Moreover, IL-17<sup>+</sup> and IFNγ <sup>+</sup> cells expressed the transcription factors RORγt and T-bet, respectively (**Supplementary Figure 4A**). Thus, DC-targeted vaccination promoted a combined Th1/Th17 response in the SDLN, in contrast to soluble antigen, which induced mainly Th1 responses (**Figure 3B**).

We then analyzed skin-infiltrating T cells. Immunization with either DC-targeted OVA or soluble OVA together with CTB induced a similar percentage of Th1 CD4<sup>+</sup> T cells (**Figure 3C**). However, DC-targeted OVA + CTB induced a higher frequency of and absolute cell numbers of Th17, compared to soluble OVA + CTB (**Supplementary Figure 4B**; **Figure 3C**). Indeed, we confirmed that DC-targeted OVA + CTB promote a combined Th1/Th17 response in the skin, while immunization with the soluble OVA + CTB promotes a skewed Th1 response by calculating the Th1/Th17 ratio (**Figure 3D**). Similarly, we also observed great infiltration of Th17 cells and almost no Foxp3<sup>+</sup> regulatory T cell differentiation in the skin of BALB/c mice transferred with DO11.10 cells after DC-targeted OVA + CTB administration (**Supplementary Figures 5B,C**).

As a whole, our results demonstrate that CTB, in combination with a DC-targeted antigen, promotes a combined Th1 and Th17 response, while soluble antigen vaccination promotes a skewed Th1 response.

#### Antigen Targeting to DCs Along With CTB Promotes CD4<sup>+</sup> T RM Cell Differentiation in the Skin

We next aimed to dissect the memory response induced by a DCtargeted antigen or a soluble antigen in combination with CTB. We first characterized the circulating and re-circulating memory of the CD4<sup>+</sup> T cell pool in the SDLN of immunized mice. CD4<sup>+</sup> T cells were classified as central memory (T CM) T cells or effector memory (T EM) T cells, according to their expression of CD44 and CD62L. The CTB's co-administration promoted increased differentiation of both T CM and T EM antigen-specific CD4<sup>+</sup> T cells in the SDLN, in combination with a DC-targeted or soluble antigen (**Figure 4A**).

Next, we studied the differentiation of skin-resident memory CD4<sup>+</sup> T cells [T RM; CD69+CCR7<sup>−</sup> (34)] after immunization. At the effector stage, a fraction of T cells that migrate to nonlymphoid organs acquire the expression of CD69 just upon their arrival to these sites (35), which can give rise to a smaller population of long-lived T RM cells (36). Accordingly, 7 days post i.d. immunization, we found that ∼30% of OT-II cells were CD69+CCR7<sup>−</sup> cells after DC-targeted OVA + CTB and, surprisingly, only ∼15% after soluble OVA + CTB immunization (**Figure 4B**). Furthermore, 30 days post-immunization, most of the OVA-specific CD4<sup>+</sup> T cells from the skin of DC-targeted OVA + CTB mice were CD69<sup>+</sup> (**Figure 4C**). Interestingly, a DCtargeted antigen was more efficient at generating long-lived T RM cells, even in comparison with a high dose of soluble OVA (30 µg of OVA, which is ∼60 times more than the amount of OVA contained in 1 µg of anti-DEC205-OVA; **Figure 4C**).

All together, our findings show that CTB can be used to enhance the differentiation of central and effector memory CD4<sup>+</sup> T cells, and that its combination with an antigen targeted to DCs efficiently promotes the differentiation of skin CD4<sup>+</sup> T RM cells.

#### Intradermal Immunization With CTB Along With a DC-Targeted Antigen Provides Local and Systemic Long-Lasting Immunity

The fact that the CTB's i.d. co-administration with a DCtargeted antigen promoted CD4<sup>+</sup> T cell activation, Th1/Th17 differentiation and migration to the skin, as well as CD4<sup>+</sup> TRM cell differentiation, prompted us to investigate whether this immunization strategy could translate into protective longterm immunity. Thus, we first made use of the subcutaneous OVA-expressing melanoma model (**Figure 5A**). We found that i.d. immunization with DC-targeted OVA or soluble OVA in combination with CTB promoted local protection against a subcutaneous challenge with an OVA-expressing melanoma (**Figure 5A**).

To evaluate if the CTB's co-administration with a DC-targeted antigen could elicit systemic activation of T cells, mice vaccinated i.d. were i.v. challenged with MO4 cells. Mice immunized with DC-targeted OVA developed ∼5 times fewer metastatic nodules than control mice and superior systemic protection (∼3 times less metastatic nodules) than mice immunized with soluble OVA + CTB (**Figure 5B**). Therefore, these data demonstrate that in comparison with the soluble antigen, CTB co-administered with a DC-targeted antigen can provide superior systemic immunity against melanoma. Interestingly, antigen specific CD4<sup>+</sup> T cells could be found in the lungs after i.d. priming, which were slightly increased after DC-targeted vaccination (**Supplementary Figure 4C**). However, the administration of an anti-CD4 Ab 30 days after priming, and prior to i.v. melanoma challenge, did not affect protection (**Figure 5B**). Nevertheless, our results show that the immune response induced by a single i.d. dose of CTB co-administered with a DC-targeted antigen provides long-term local and systemic immunity, and, as importantly, the infiltration of CD4<sup>+</sup> T cells in distal tissues.

IFNγ <sup>+</sup> and IL-17<sup>+</sup> OT-II cells (identified as in Figure 2A). (B) Ratio of SDLN Th17/Th1 cells. Mean ± SD, N = 6–8, data pooled from two independent experiments. Unpaired T-test (ns, P > 0.05, \*P < 0.05, \*\*P < 0.005). Skin cell suspensions were stimulated with cell cocktail stimulation + protein transport inhibitor for 4 h. (C) Graphs of percentage and total numbers of skin IFNγ <sup>+</sup> and IL-17<sup>+</sup> of OT-II cells (identified as in Figure 2B). (D) Ratio of skin Th17/Th1 cells. Mean ± SD, N = 6–8, data pooled from three independent experiments. Unpaired T-test (ns, P > 0.05, \*\*P < 0.005, \*\*\*P = 0.0001).

#### A DC Targeted Antigen Along With CTB Induces Infiltration of Polyfunctional CD4<sup>+</sup> T Cells in the Intestine and Provides CD4<sup>+</sup> T Cell Dependent Protection Against Rotavirus

Next, we asked whether the CTB's co-administration with a DC-targeted antigen could induce CD4<sup>+</sup> T cell responses in another distal tissue, i.e., the intestine. Indeed, very few cells were found in the intestine after i.d. immunization; however, DC-targeted vaccination promoted superior infiltration of OVAspecific CD4<sup>+</sup> T cells, as compared with the soluble antigen immunization (**Figures 6A,B**). Furthermore, a higher percentage and number of cells from the intestines of the DC-targeted vaccination group expressed the T RM marker CD69 (**Figure 6C**; **Supplementary Figure 4D**).

To evaluate whether i.d. DC-targeted vaccination could provide protection in the intestine we made use of a murine

cells (identified as in Figure 2A), and graphs of the percentage of each population 7 days post-immunization. Mean ± SD, N = 4–6, data pooled from two independent experiments. One-way ANOVA with Tukey's multiple comparisons test (\*P < 0.05, \*\*P < 0.005, \*\*\*P < 0.0005). (B) Representative contour plots and a graph showing percentages of CD69+CCR7<sup>−</sup> OT-II CD45.1<sup>+</sup> cells (identified as in Figure 2B) from the inoculation site 7 days post-immunization. Mean ± SD, N = 5–6, data pooled from three independent experiments. (C) CD45.1<sup>+</sup> mice received i.v. OT-II CD45.2<sup>+</sup> cells and 1 day later were inoculated with 1 µg of anti-DEC205-OVA or with 30 µg of OVA, both in combination with CTB. Representative contour plots and a graph showing percentages of CD69<sup>+</sup> OT-II CD45.2<sup>+</sup> cells 30 days post-immunization. Mean ± SD, N = 3–5 data pooled from two independent experiments. Unpaired T-test.

graph of metastatic nodules per lung, 16 days after challenge. Mean ± SD, N = 5–10, data pooled from two independent experiments. One-way ANOVA with Tukey's multiple comparisons test (ns, P > 0.05, \*\*\*\*P < 0.0001).

rotavirus model. Rotavirus infection is mostly limited to the small intestine; therefore, the immune response is highly compartmentalized (37). Thus, we made use of a VP6-based vaccine model. VP6 is a highly conserved antigen among different strains of rotavirus (38), and it has been shown to promote protective immunity when targeted to DCs in the presence of Poly IC (30). Furthermore, protection against murine rotavirus, in models of soluble VP6 immunization, is dependent on CD4<sup>+</sup> T cells (39, 40). Thus, mice were i.d. administered with anti-DEC205−VP6+CTB or soluble VP6+CTB, 20 days before the challenge with oral rotavirus. Only antigen targeting immunization provided intestinal protection (∼10%), while soluble immunization did not provide protection against the viral challenge (**Figure 6D**). Therefore, our results suggest that the immune response elicited by a single dose of i.d. DC-targeted antigen admixed with CTB provides partial long-term immunity in the intestine.

The development of partial protection after a single i.d. dose of a DC-targeted antigen could have been due to poor infiltration of functional T cells in the intestine. Therefore, we asked whether a prime/boost immunization scheme could expand the specific CD4<sup>+</sup> T cells. To answer this question, mice were i.d. immunized with a DC-targeted antigen or a soluble antigen admixed with CTB. Fifteen days later, mice received, i.p. the targeted or soluble antigen. After 5 days, we observed a greater expansion of antigen-specific CD4<sup>+</sup> T cells in the intestine after DC-targeted prime/boost, compared to the soluble antigen prime/boost group (**Figure 7A**). In addition, DC-targeted prime/boost promoted the

expansion of IL-17<sup>+</sup> CD4<sup>+</sup> T cells, that can also produce other cytokines like IFNγ and/or TNFα, in contrast to soluble antigen immunization (**Figure 7B**).

The above results prompted us to discern whether the prime/boost immunization strategy could improve protection in the murine rotavirus model. To this end, mice received anti-DEC205-VP6 or soluble VP6 admixed with CTB, via the i.d. route; 15 days later they received i.p. anti-DEC-VP6 or VP6 only. After 7 days, mice were orally challenged with rotavirus (**Figure 7C**). Four days after the challenge, the viral

FIGURE 7 | Intradermal prime/i.p. boost immunization with a DC-targeted antigen + CTB induces functional CD4<sup>+</sup> T cells in the intestine and provides CD4<sup>+</sup> T cell dependent protection against rotavirus. C57BL6 mice were adoptively transferred with OT-II CD45.1<sup>+</sup> cells 24 h before i.d. anti-DEC205-OVA or soluble OVA with CTB. Fifteen days later, immune mice received i.p. anti-DEC205-OVA or soluble OVA and after 5 days, mice were sacrificed, and intestines were collected. (A) Cells were gated as viable CD45+CD4+CD45.1<sup>+</sup> cells to calculate percentage and total number of transferred cells present in the intestine. Mean ± SD, N = 6 per group, data pooled from two independent experiments. Unpaired T-test (\*P < 0.05). (B) Freshly isolated cells were stimulated 4 h with cell cocktail stimulation + protein transport inhibitor. Graphs of percentage and total numbers of CD4+CD45.1<sup>+</sup> cytokine producing cells (gated as in B). Boolean combinations were calculated using FlowJo software. Mean ± SD, N = 6 per group, data pooled from two independent experiments. Two-way ANOVA with Bonferroni's multiple comparison test (\*\*P = 0.0017, \*\*\*\*P < 0.0001). (C) Strategy followed for oral viral challenge with murine rotavirus after i.d. immunizations and i.p. boost. Mice immunized with anti-DEC-VP6+CTB received i.p. anti-CD4 or the control isotype Ab, before, during and after the viral challenge. (D) Stool samples were collected every day up to day 8 and viral load was determined by sandwich ELISA. (E) Percentage of protection relative to control (vehicle) mice, calculated as area under the curve (From D). Mean ± SD, N = 5–8 per group, data pooled from two independent experiments. Two-way ANOVA with Tukey's multiple comparisons test.

load dramatically dropped in stool samples from the DCtargeted vaccination group (**Figure 7D**). This meant ∼60% protection against infection relative to naïve mice (**Figure 7E**). Protection relied on the antigen being targeted to DCs, since the isotype Ab conjugated with VP6 and admixed with CTB only provided partial protection (∼15%). Protection was significantly dampened when CD4<sup>+</sup> T cells were depleted by the administration of anti-CD4 antibody. On the other hand, soluble antigen vaccination provided only partial protection against infection (∼15%; **Figures 7D,E**).

Collectively, our results show that i.d. administration of DCtargeted antigens admixed with CTB promotes the infiltration of polyfunctional CD4<sup>+</sup> T cells in the intestine. It is important to point out that our data suggest that this response provides longterm immunity against a pathogen whose clearance is partially dependent on CD4<sup>+</sup> T cells.

#### DISCUSSION

Immunization strategies that confer broad long-lasting immunity mediated by CD4<sup>+</sup> T cells are fundamental to eradicate modern pandemics. To achieve this goal, mAbs targeting antigen to DEC205<sup>+</sup> DCs, in combination with maturation stimuli, is one of the most promising strategies. Here we have demonstrated that DC-targeted antigens admixed with CTB promote the development of long-lasting systemic protective CD4<sup>+</sup> T cell responses.

Successful DC-targeted vaccination requires DC stimulation by strong adjuvants, which ultimately promotes T cell responses. Therefore, we studied the activation and accumulation of DCs following the CTB's i.d. administration. It took 72 h to observe both DC activation and accumulation in the skin; in contrast, other adjuvants (i.e., LPS, CpG, flagellin and the complete cholera toxin) can induce local activation as soon as 6 to 24 h (9, 41–45). Differences could be related to the receptors engaged by CTB on the DCs (17, 18). The late activation of skin DCs was also seen in the SDLN, where activated migratory DCs accumulated 72 h after CTB inoculation. These findings could explain why others have failed at demonstrating activation and accumulation of DCs in draining lymph nodes 2–24 h following CTB administration (44, 46). Therefore, while other adjuvants can promote rapid activation and accumulation of DCs, our results indicate that CTB induces late activation and accumulation of skin DCs.

Interestingly, the accumulation and the activated phenotype of DCs were still observed after 7 days, in both the skin and the SDLN. Similar observations have been reported after the administration of CpG, alum or the MF59 oil-in-water emulsion (47), which induced accumulation in the muscle of MHC-II<sup>+</sup> cells up to 4 days after inoculation. The same phenomenon was true for resident lymph node DCs following CTB administration. These findings suggest that CTB can stimulate various populations of DCs for a prolonged time, which could potentially lead to sustained and diverse DC-T cell interactions. Noticeably, the late accumulation of activated skin and lymph node-DCs correlated with the priming of CD4<sup>+</sup> T cell responses observed at day 7, following the CTB's coadministration with antigen. Together, our findings shed light on the CTB's controversial ability to activate DCs in vivo.

Antigen targeting to DEC205<sup>+</sup> DCs is a promising system to promote CD4<sup>+</sup> T cell responses (4, 6). Indeed, i.d administration of anti-DEC205-OVA increased the efficiency of antigen presentation relative to the soluble OVA. It was not, however, as large as reported by previous publications that used the s.c. or i.p. routes. Because the SDLN are very close to the inoculation site, i.d. administration of very small quantities of soluble antigen can efficiently promote CD4<sup>+</sup> T cell proliferation, in contrast with the s.c or i.p. routes (27). Furthermore, i.d. administration of anti-DEC205-OVA clearly induced a different activation of CD4<sup>+</sup> T cells as compared with soluble OVA. Not only did it induce cells to proliferate more, but it also induced a marked downregulation of CD69, which is necessary for T cells' egress to the periphery (31, 32). In this regard, the soluble antigen along with CTB promoted a localized CD4<sup>+</sup> T cell response, while a DC-targeted antigen admixed with CTB induced systemic CD4<sup>+</sup> T cell responses. Considering that we did not observe DC activation in distal sites, our results suggest that following i.d. DC-targeted vaccination; the priming occurs in the SDLN, and then, activated CD4<sup>+</sup> T cells migrate to infiltrate the site of inoculation and, remarkably, other peripheral tissues. Therefore, our results suggest that the priming induced by a DC-targeted antigen admixed with CTB promotes unique systemic CD4<sup>+</sup> T cell responses.

Indeed, a DC-targeted antigen along with CTB induced a combined and systemic Th1/Th17 response. In contrast, soluble antigen immunization promoted a skewed localized Th1 response, which is similar to that observed when using CTB as an adjuvant linked with antigens or admixed with pathogen derived antigens (22–24, 26). Furthermore, the CTB's combination with a DC-targeted antigen promoted the differentiation of polyfunctional Th cells. It has been documented that anti-DEC205-antigen Abs admixed with experimental adjuvants i.e., CpG oligonucleotides, flagellin (9) and Poly IC (7)—induce differentiation of polyfunctional CD4<sup>+</sup> T cells that produce IFNγ, TNFα, and IL2. However, none of these adjuvants are able to induce Th17 differentiation (48–50). In our model, the superior induction of Th17 cells seemed to depend on both the adjuvant and the antigen being directly delivered to DEC205<sup>+</sup> DCs, since soluble antigen vaccination induced IL-17<sup>+</sup> antigen specific CD4<sup>+</sup> T cells only marginally. To our knowledge, this is the first report showing induction of systemic polyfunctional CD4<sup>+</sup> T cell responses that include IL-17<sup>+</sup> cells after antigen targeting to DEC205<sup>+</sup> DCs by genetically engineered mAbs admixed with CTB.

We also demonstrate that a DC-targeted antigen admixed with CTB efficiently promotes the generation of memory CD4<sup>+</sup> T cells, something that has not been extensively explored after performing DC-targeted vaccination. Here, using cell surface markers, we found in the SDLN the presence of circulating and re-circulating memory CD4<sup>+</sup> T cells after using a DCtargeted antigen or a soluble antigen admixed with CTB. Strikingly, the high infiltration of CD4<sup>+</sup> T cells after soluble antigen immunization did not translate into more T RM differentiation. In contrast, DC-targeted vaccination induced superior differentiation of CD4<sup>+</sup> T RM cells at the site of inoculation and, more importantly, at a distal nonlymphoid tissue, i.e., the intestine. This is similar to what has been observed in studies inoculating recombinant vaccinia virus expressing OVA through skin scarification, which induces the differentiation of protective T RM cells in the skin and lungs (51–53). Thus far, there are only a couple of publications reporting CD8<sup>+</sup> T RM cell differentiation after immunization with anti-DEC205 antigen mAbs, using LPS (54) or Poly IC (55) as adjuvants. However, none of them have shown the presence of T RM cells in distal sites after local vaccination. Our observations suggest that antigen targeting to DEC205<sup>+</sup> DCs, in combination with CTB, is an effective strategy to promote systemic differentiation of CD4<sup>+</sup> T RM cells. This is of particular relevance in light of recent studies, pointing to T RM cells as essential players against several infections (34, 36, 56) and melanoma (53, 55) protection.

Following this line, DC-targeted and soluble antigen vaccination provided similar long-term protection against subcutaneous melanoma. This could be related to the protective capacity of both circulating and T RM cells against melanoma (53). However, we found that DC-targeted vaccination provided superior systemic protection against pulmonary tumor growth. Although we found antigen specific CD4<sup>+</sup> T cells in the lungs of immune mice, the administration of a neutralizing anti-CD4 Ab during the memory phase did not abrogate protection against i.v. melanoma. This is contrary to melanoma studies in CD4 knockout mice, where protection is partially dampened (4). Therefore, we cannot completely rule out the participation of CD4<sup>+</sup> T cells in the priming of protective CD8<sup>+</sup> T cell responses against i.v. melanoma. In this regard, our results suggest that protection in the lungs could be primarily mediated by memory CD8<sup>+</sup> T cells after DC targeted vaccination using CTB as adjuvant. This idea is supported by the fact that CD8<sup>+</sup> T cells are efficiently activated by anti-DEC205 Abs (4, 6) and by antigens linked to CTB (57, 58). Since priming occurred in the SDLN, our findings suggest that DC-targeted vaccination using CTB as adjuvant can be used as an efficient immunization strategy to provide systemic long-term immunity against melanoma.

Interestingly, DC-targeted vaccination-induced systemic CD4<sup>+</sup> T cell responses translated into protection in the intestine. This could have been mediated by the T RM and polyfunctional CD4<sup>+</sup> T cells found in the intestine after DCtargeted vaccination. However, a single immunization induced only small numbers of T cells in the intestine and partial protection. Since protective immunity correlates with high numbers of functional cells infiltrating the site of infection, we took advantage of the ability of anti-DEC205 Abs to disseminate systemically (6) to successfully expand the antigen specific CD4<sup>+</sup> T cells in the intestine through a DC-targeted antigen + CTB i.d. prime/DC-targeted antigen i.p. boost. Remarkably, this strategy promoted higher numbers of IL-17<sup>+</sup> CD4<sup>+</sup> T cells to be present in the intestine, as well as polyfunctional CD4<sup>+</sup> T cells. Furthermore, the prime/boost scheme dramatically improved protection against the oral viral challenge, but only when the antigen was targeted to DCs. Moreover, the protection observed was superior than the one reported by s.c. administration of the same antibody in the presence of Poly IC, which was related to the development of Th1 responses (30). Also, protection in our model was partially dependent on CD4<sup>+</sup> T cells, according with the CD4 blockade experiments. However, we cannot exclude the participation of CD8<sup>+</sup> T cells. These findings indicate that DC-targeted antigens admixed with CTB promote infiltration of the intestine with functional CD4<sup>+</sup> T cells capable of mediating protection against pathogens with intestine tropism.

Our results extend the advantages of immunization with antigens targeted to DEC205<sup>+</sup> DCs with mAbs in combination with strong adjuvants (CTB) to induce high quality systemic immune responses that translate into protection. We propose that a DC-targeted antigen can be co-administered with CTB i.d.; a suitable novel combination with potential human use, for the generation of protective, systemic and long-lasting Th17 CD4<sup>+</sup> and polyfunctional responses, which, importantly, are characterized by CD4<sup>+</sup> T RM cells. Furthermore, this immunization strategy could be used to fight infections and tumors.

#### ETHICS STATEMENT

This study was carried out following the recommendations of the Institutional Ethics Committee and the Comité Local de Investigación en Salud, Protocol number R-2015-785-023. All procedures for animals were approved by the Animal Ethics Committee of the Faculty of Medicine at UNAM, and they followed the Mexican Official Guide (NOM 062-ZOO-1999) for the care and use of laboratory animals.

# AUTHOR CONTRIBUTIONS

LA-H performed the majority of the experiments, interpreted the data and drafted the manuscript. OB-G performed the murine rotavirus protection assay. OM-C processed intestine samples. AT-S helped with skin DCs analysis. AT-S and AG-L performed confocal microscopy experiments. FE-G and LG-X contributed to design the murine rotavirus experiments. GS provided transgenic OT-II CD45.1<sup>+</sup> mice and helped with the interpretation of data. JI helped to train the first author in experimental techniques, provided reagents necessary for the study (i.e., anti-DEC-OVA mAb), helped with the design of the study and the interpretation of results, and revised the manuscript. LB conceived and directed the project and revised the manuscript. All the authors reviewed the manuscript critically.

# FUNDING

This study was funded by the Instituto Mexicano del Seguro Social (IMSS) R-2015-785-023 FIS/IMSS/PROT/G151435 and by Consejo Nacional de Ciencia y Tecnología (CONACyT) CB-2010-01157018 (to LB) 180441 (to FE-G) in Mexico, and by the National Institute of Arthritis, Musculoskeletal and Skin Diseases of the US National Institutes of Health (R00 AR062595 to JI) and the National Cancer Institute of the US National Institutes of Health (R01 CA219994 to JI). LA-H is a doctoral student from Programa de Doctorado en Ciencias Biomédicas at UNAM and she received a fellowship (275768) from Consejo Nacional de Ciencia y Tcenología (CONACYT) in Mexico, and a fellowship by the Coordinación de Investigación en Salud: Apoyo de movilidad internacional, programa de cooperación internacional. OB-G is a posdoctoral fellow from Programa Nacional para Estudios de Posdoctorado (CVU 348201) from CONACyT.

#### ACKNOWLEDGMENTS

We would like to thank the staff of the animal facility at the Experimental Medicine Unit, Faculty of Medicine,

#### REFERENCES


UNAM, and at the INSP, SS, for providing expert animal care. We also thank the members of Bonifaz' laboratory and of Idoyaga Lab for their advice and protocols. Finally, we extend our gratitude to the Flow Cytometry core facility, the Coordinación de Investigación en Salud, the CMN S XXI, and the IMSS, for their instrumental and technical support.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2018.02212/full#supplementary-material


tularensis in the presence versus absence of cholera toxin B and inactivated F. tularensis. J Immunol. (2009) 182:4899–909. doi: 10.4049/jimmunol.0803242


after oral administration of cholera toxin. J Immunol. (2004) 173:5103–11. doi: 10.4049/jimmunol.173.8.5103


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

The reviewer LF and handling Editor declared their shared affiliation.

Copyright © 2018 Antonio-Herrera, Badillo-Godinez, Medina-Contreras, Tepale-Segura, García-Lozano, Gutierrez-Xicotencatl, Soldevila, Esquivel-Guadarrama, Idoyaga and Bonifaz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Intestinal CD103+CD11b<sup>+</sup> cDC2 Conventional Dendritic Cells Are Required for Primary CD4<sup>+</sup> T and B Cell Responses to Soluble Flagellin

Adriana Flores-Langarica<sup>1</sup> \*, Charlotte Cook <sup>2</sup> , Katarzyna Müller Luda<sup>3</sup> , Emma K. Persson<sup>4</sup> , Jennifer L. Marshall <sup>5</sup> , Nonantzin Beristain-Covarrubias <sup>1</sup> , Juan Carlos Yam-Puc<sup>1</sup> , Madelene Dahlgren<sup>3</sup> , Jenny J. Persson<sup>3</sup> , Satoshi Uematsu6,7, Shizuo Akira<sup>8</sup> , Ian R. Henderson<sup>2</sup> , Bengt Johansson Lindbom3,9, William Agace3,9 and Adam F. Cunningham<sup>1</sup>

#### Edited by:

Silvia Beatriz Boscardin, Universidade de São Paulo, Brazil

#### Reviewed by:

Henning Lauterbach, Bavarian Nordic, Germany Kiwook Kim, Washington University School of Medicine in St. Louis, United States

> \*Correspondence: Adriana Flores-Langarica

a.floreslangarica@bham.ac.uk

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 19 July 2018 Accepted: 28 September 2018 Published: 17 October 2018

#### Citation:

Flores-Langarica A, Cook C, Müller Luda K, Persson EK, Marshall JL, Beristain-Covarrubias N, Yam-Puc JC, Dahlgren M, Persson JJ, Uematsu S, Akira S, Henderson IR, Lindbom BJ, Agace W and Cunningham AF (2018) Intestinal CD103+CD11b<sup>+</sup> cDC2 Conventional Dendritic Cells Are Required for Primary CD4<sup>+</sup> T and B Cell Responses to Soluble Flagellin. Front. Immunol. 9:2409. doi: 10.3389/fimmu.2018.02409 1 Institute of Immunology and Immunotherapy, College of Medical and Dental Sciences, University of Birmingham, Birmingham, United Kingdom, <sup>2</sup> Institute of Microbiology and Infection, College of Medical and Dental Sciences, University of Birmingham, Birmingham, United Kingdom, <sup>3</sup> Immunology Section, Department of Experimental Medical Science, Lund University, Lund, Sweden, <sup>4</sup> VIB-Ugent Center for Inflammation Research, Ghent, Belgium, <sup>5</sup> Institute of Inflammation and Ageing, College of Medical and Dental Sciences, University of Birmingham, Birmingham, United Kingdom, <sup>6</sup> International Research and Development Centre for Mucosal Vaccine, Institute for Medical Science, The University of Tokyo, Tokyo, Japan, <sup>7</sup> Department of Immunology and Genomics, Osaka City University Graduate School of Medicine, Osaka, Japan, <sup>8</sup> World Premier International Immunology Frontier Research Centre, Osaka University, Suita, Japan, <sup>9</sup> Section of Biology and Chemistry, Department for Micro- and Nanotechnology, Technical University of Denmark, Kongens Lyngby, Denmark

Systemic immunization with soluble flagellin (sFliC) from Salmonella Typhimurium induces mucosal responses, offering potential as an adjuvant platform for vaccines. Moreover, this engagement of mucosal immunity is necessary for optimal systemic immunity, demonstrating an interaction between these two semi-autonomous immune systems. Although TLR5 and CD103+CD11b<sup>+</sup> cDC2 contribute to this process, the relationship between these is unclear in the early activation of CD4<sup>+</sup> T cells and the development of antigen-specific B cell responses. In this work, we use TLR5-deficient mice and CD11c-cre.Irf4fl/fl mice (which have reduced numbers of cDC2, particularly intestinal CD103+CD11b<sup>+</sup> cDCs), to address these points by studying the responses concurrently in the spleen and the mesenteric lymph nodes (MLN). We show that CD103+CD11b<sup>+</sup> cDC2 respond rapidly and accumulate in the MLN after immunization with sFliC in a TLR5-dependent manner. Furthermore, we identify that whilst CD103+CD11b<sup>+</sup> cDC2 are essential for the induction of primary T and B cell responses in the mucosa, they do not play such a central role for the induction of these responses in the spleen. Additionally, we show the involvement of CD103+CD11b<sup>+</sup> cDC2 in the induction of Th2-associated responses. CD11c-cre.Irf4fl/fl mice showed a reduced primary FliC-specific Th2-associated IgG1 responses, but enhanced Th1-associated IgG2c responses. These data expand our current understanding of the mucosal immune responses promoted by sFliC and highlights the potential of this adjuvant for vaccine usage by taking advantage of the functionality of mucosal CD103+CD11b<sup>+</sup> cDC2.

Keywords: flagellin, mucosa, immune response, dendritic cells, cDC2

# INTRODUCTION

The systemic and mucosal immune systems are semiautonomous and engaging systemic immunity does not necessarily induce immunity in mucosal sites. Engaging the two immune systems concurrently could potentially enhance the benefits of vaccination, as most vaccines are administered through subcutaneous (s.c.) or intra-muscular injection. One antigen that can induce both mucosal and systemic immunity concurrently after intraperitoneal (i.p.) or s.c. immunization is purified, soluble flagellin (sFliC) from Salmonella Typhimurium (1–3). This 51 kDa bacterial motility protein is the only known ligand for TLR5 (4). Moreover, flagellin is an immunodominant antigen that can induce robust innate and adaptive immune responses, which can also be protective (5–7). These properties, alongside its potential as an adjuvant, mean flagellin is the focus of multiple vaccine strategies in livestock and in humans (8–12). The antigenic environment in which flagellin is encountered influences the type of immune response induced to this protein. When surface-localized on the bacterium, the antigen-specific response is Th1-reflecting, whereas to purified flagellin the response is significantly more Th2-like, including the induction of FliC-specific IgG1 (13, 14).

Conventional dendritic cells (cDC) are key initiators and modulators of adaptive immune responses and as such targeting cDC directly is an approach to enhance responses to vaccines (15, 16). cDCs can be classified into two major subsets; cDC1 that are require the transcription factors IRF8, BATF3, and ID2, and cDC2 that development is independent of these transcription factors, importantly some them require the transcription factor IRF4 for their survival and function. This classification is particularly important since it allows the identification of cDCs equivalents across tissues and even across species (17, 18). In the intestinal mucosa several sub populations of cDC can be found, CD103+CD11b−, CD103+CD11b+, and CD103−CD11b<sup>+</sup> cDC. The first corresponds to cDC1 and the latter two to cDC2. Each of these subsets plays key, non-redundant roles in controlling immune homeostasis in the intestinal mucosa (19–21).

In vivo studies have shown that by 24 h after i.p. or s.c. immunization with sFliC, T cell priming is established in multiple sites concurrently, including the mesenteric lymph node (MLN), spleen and peripheral lymph nodes (1). Analysis of cDCs shows that exclusively in the MLN, there is a rapid TLR5-dependent accumulation of CD103<sup>+</sup> cDC post sFliC-immunization (1). Moreover, using Cd11c-cre.Irf4fl/fl mice, which have diminished numbers of CD103+CD11b<sup>+</sup> cDCs in the small intestine lamina propria and a 90% reduction of this population in the MLN, we showed that this subset was essential for the induction of adaptive immune responses in the MLN, while splenic cDC2 play only a partial role. For clarity, CD103+CD11b<sup>+</sup> cDCs will be referred to throughout as CD103+cDC2 (3). This indicates that i.p., immunization with sFliC can bridge both systemic and mucosal immune systems through the targeting of a single mucosal cDC subset.

Our previous work examining the role of CD103+cDC2 in regulating the response to sFliC focused on the long-term antibody response in vivo using the Cd11c-cre.Irf4fl/fl mice. This necessitated the use of a prime-boost system and did not focus on the primary T and B cell responses. Whilst all elements of the response were lost in the MLN when mucosal CD103+cDC2 were reduced, some features of the anti-FliC response were retained in the spleen. This could be because some T and B cell responses were generated in the MLN shortly after immunization, which could lead to the generation of memory T and B cell responses that contribute to the responses observed after secondary immunization. Alternatively, it could be that cDC2 and cDC1 contributed differentially to the anti-sFliC response in the MLN and spleen. Therefore, we examine here the development of the anti-sFliC response in the first days after immunization to characterize the relationship between cDC2 and TLR5 and the early induction of IgG switching.

#### MATERIAL AND METHODS

#### Mice

Cd11c-cre.Irf4fl/<sup>f</sup> (19) and NAIP5−/<sup>−</sup> mice were maintained at the Biomedical Center at Lund University. Specific pathogenfree 8 week C57BL/6 mice were purchased from Harlan Sprague-Dawley. TLR5−/<sup>−</sup> mice were maintained in-house at the Biomedical Service Unit at the University of Birmingham. Littermates or age matched mice were used for all experiments. All animal procedures were carried out in strict accordance with the Lund/Malmö Animal Ethics Committee, the University of Birmingham Ethics Committee and were covered under the UK Home Office Project license 30/2850.

#### Antigen Preparation and Immunization

sFliC was generated as described (22), a his-tagged recombinant protein and purified by nickel affinity chromatography and immunoprecipitation with a FliC-specific monoclonal. Mice were immunized i.p. with 20 µg recombinant sFliC for 24 h or 7 days as indicated.

#### Cell Isolation and Flow Cytometry

Single cell suspensions from spleens and MLNs were generated by mechanical disruption. When evaluating cDCs, enzymatic digestion was performed using collagenase VIII digestion (400 U/ml; 25 min; 37◦C). Cells were processed for flow cytometry using previously described procedures (1). Data acquisition was performed on a LSRII (BD Bioscience) or a CyAn ADP (Beckman Coulter) and analyzed using FlowJo software 9.8.2. (Tree Star). The following FITC-conjugated antibodies were used, CD3 (145-2C11), B220 (RA3-6B2), and NK1,1 (PK136; all from eBioscience). The following PE-conjugated antibodies were used; CD103 (M290) and CD62L (MEL-14; both from eBioscience). CD11c (N418), CD44 (IM7), and CD95 (MFL3; all from eBioscience) were PE-Cy7-conjugated. CD11b (M1/70), CD4 (RM4-5, both BD Biosciences), TCRβ (H57-597, Biolegend), GL7 (eBioscience) were PB-conjugated. MHC-II (M5/114.15.2) and

**Abbreviations:** Ab, antibody; cDCs, conventional dendritic cells; GC, germinal center; i. p., intraperitoneal; MLN, mesenteric lymph node; s.c., subcutaneous; sFliC, soluble flagellin.

Streptavidin (eBioscience) APC-conjugated where used. B220 (RA3-6B2), CD11b (MI-70), CD11c (N418) and NK1.1 (PK136; all from eBioscience) were Alexa700-conjugated. TCRβ (H57- 597, Biolegend) was APC Cy7-conjugated. CD8α (5H10) from Invitrogen was used PO-conjugated.

cDCs were gated as Lin−[CD3,B220,NK1.1,GR1]CD11c+MHChi cells, splenic cDC1 were defined as CD8α <sup>+</sup> and cDC2 as CD11b+CD4<sup>+</sup> cells, mucosal cDC1 were defined as CD103+CD11b<sup>−</sup> and cDC2 as CD103+CD11b<sup>+</sup> cells. Activated CD4<sup>+</sup> T cells were gated as CD3+CD4+CD44+CD62L−. Germinal center (GC) B cells were defined as TCRβ <sup>−</sup>CD19+GL7+CD95<sup>+</sup> cells.

#### Immunohistochemistry and Confocal Microscopy

Immunohistochemistry was performed as described previously (13). Cryosections (6µm) were incubated with primary unlabelled Abs for 45 min at RT before addition of either HRP-conjugated or biotin-conjugated secondary antibodies. FliC-binding cells were identified as described (1) using soluble biotinylated FliC. Subsequently streptavidin ABComplex alkaline phosphatase (Dako) was used. Signal was detected using diaminobenzidine for HRP activity and naphthol AS-MX phosphate with Fast Blue salt and levamisole for alkaline phosphatase activity. Images were acquired using a Leica microscope DM6000 using 10x and 20x objectives. Quantification of sFliC+IgG1<sup>+</sup> and sFliC+IgG2c<sup>+</sup> was performed in two independent experiments, each with 4 mice per group. A total of 10 random fields were evaluated per slide.

Immunofluorescence was performed on frozen sections. Staining was performed in PBS containing 10% FCS, 0.1% sodium azide and sections were mounted in 2.5% 1,4- Diazabicyclo(2,2,2)octane (pH 8.6) in 90% glycerol in PBS. After incubation with primary Abs (1 h, room temperature), secondary Abs were added (30 min; room temperature). Images were acquired using the Fluorescence Zeiss Axio Scan.Z1, image analysis was performed with Zen, 2012 blue edition.

The following Abs were used for immunofluorescence: CD11c, Dec205, CD103. DCIR2 was used biotinylated. CD11b (eBioscience) and streptavidin (Jackson Immunoresearch) were used A488 conjugated. And AMCA-conjugated anti-IgM (Jackson Immunoresearch) was used. The following Abs were used for immunohistochemistry: IgD, IgG1, and IgG2c. sFliC and PNA (Vector) biotinylated were used.

#### In vitro Restimulation for the Detection of Antigen-Specific CD4<sup>+</sup> T Cells

Antigen-specific CD4<sup>+</sup> T cells were detected by their expression of CD154 post-restimulation as previously described (23). In brief, 6 × 10<sup>6</sup> cells (from spleen or MLN) were cultured in the presence of sFliC (5µg/ml), anti-CD40 (1C10 2µg/ml), biotinylated anti-CD154 (MR1 5µg/ml) and anti-Fcγ receptor (2.4G2 50µg/ml) for 48 h. Control wells included cells cultured without antigen. The expression of CD154 was evaluated by staining the cells with Streptavidin-APC and gating was done on Lin−(CD19/CD11b/NK1.1/CD11c)TCRβ <sup>+</sup>CD4+CD62L−.

# Sflic-Specific ELISA

ELISA plates were coated with 5µg/ml of sFliC (24 h at 4◦C) and blocked with 1% BSA overnight at 4◦C. Serum was diluted 1:100 in PBS−0.05% Tween, and was further diluted stepwise. Plates were incubated for 1 h at 37◦C. Bound antibodies were detected using alkaline phosphatase conjugated, goat anti-mouse IgG, IgG1, and IgG2c Abs (Southern Biotech). Alkaline phosphatase activity was detected using Sigma-Fast p-nitrophenylphosphate (Sigma Aldrich). Relative reciprocal titers were calculated by measuring the dilution at which the serum reached a defined OD405.

#### Statistics

For statistics we used the non-parametric Mann-Whitney sum of ranks test or two-way ANOVA as appropriate using the GraphPad Prism software (GraphPad).

### RESULTS

### The Accumulation of CD103+cDC2 in the MLN After sFliC Immunization Is TLR5-Dependent

Intraperitoneal immunization of sFliC induces a rapid MyD88 dependent accumulation of intestinal-derived CD103+cDC2 in the MLN (1, 3). To analyse the role of TLR5 in this response, WT and TLR5−/<sup>−</sup> mice were immunized with sFliC and the cDC response analyzed by flow cytometry and in situ by immunofluorescence 24 h post-immunization. In the MLN of WT mice, immunization with sFliC resulted in an increase in the frequency and absolute numbers of CD103+cDC2, which was abrogated in TLR5−/<sup>−</sup> mice, no significant change in absolute numbers or frequency was observed for cDC1 or CD103−cDC2 (**Figure 1A**). Immunofluorescence microscopy showed that CD103+cDC2 were mainly located in the T zone, before and after immunization, although they were less abundant in the absence of immunization. The high zoom-insets confirmed that CD103+cDC2 in the T zone express CD11c, CD103, and CD11b. TLR5 is the extracellular receptor for sFliC, however there is an intracellular pathway for flagellin detection controlled by NAIP5 (24). To evaluate the contribution of this alternative pathway we evaluated the CD103+cDC2 in NAIP5−/<sup>−</sup> mice, we observed a similar accumulation of those cells in comparison with the WT after sFliC immunization, strongly suggesting that TLR5 exclusively controls the response to sFliC by CD103+cDC2 in the MLN (**Figure 1B**).

In contrast to the MLN, in the spleen there was no change in the frequency and absolute numbers of cDC2 (identified as CD11b<sup>+</sup> cDC) of WT mice after immunization with sFliC (**Figure 1C**). Furthermore, no difference was observed in the location of cDC2 (DCIR2+) which remain primarily localized in the bridging channels between the red and white pulps (25). Loss of TLR5 did not alter the numbers of cDC2 before or after immunization, nor the distribution of these cells within the spleen (**Figure 1B**). Therefore, immunization with sFliC results

and cDCs (Lin−MHC-IIhiCD11chi) were evaluated 24 h later, alongside non-immunized (N.I.) mice. (A) MLN representative flow cytometry plots (including percentages) of cDC1 (Lin−MHC-IIhiCD11chiCD11b−CD103+), CD103+cDC2s (Lin−MHC-IIhiCD11chiCD11b+CD103+) and CD103−cDC2 (Lin−MHC-IIhiCD11chiCD103−) are shown with adjacent graphs of absolute numbers. Representative photomicrographs of MLN sections stained for CD11c; blue, CD103; red, CD11b; green, and IgM; white (scale bar = 200µm) are shown (top right). Zoom-in insets (white boxes) show single staining and a merge of CD11c, CD103, and CD11b (scale bar = 20µm). (Continued)

FIGURE 1 | T, T zone; B, B zone. (B) WT or NAIP5−/<sup>−</sup> mice were immunized i.p., with sFliC and absolute numbers of MLN CD103+cDC2s were evaluated 24 h later, alongside non-immunized (N.I.) mice. (C) Wild-type (WT) or TLR5−/<sup>−</sup> mice were immunized i.p. with sFliC and splenic cDCs (Lin−MHC-IIhiCD11chi) were evaluated 24 h later, alongside non-immunized (N.I.) mice. Representative flow cytometry plots (including percentages) of cDC2s (Lin−MHC-IIhiCD11chiCD11b+) are shown with adjacent graphs of absolute numbers. Representative photomicrographs of spleen sections stained for CD11c; blue, Dec205; green, DCIR2; red, and IgM; white (scale bar = 100µm). Zoom-in insets (white boxes) show the differential location of cDC1s (Dec205+) in the T zone and cDC2s (DCIR2+) in the bridging channels. Data shown as mean+s.d. of 4 mice and are representative of 3 independent experiments. \*\*P < 0.001, by two-way analysis of variance (ANOVA) N.S., not significant.

in the selective accumulation of CD103+cDC2 in the MLN, but not in the spleen in a TLR5 dependent manner.

### CD103+cDC2 Are Essential for T Cell Priming in the MLN After Immunization With sFliC

To assess the contribution of cDC2 to T cell priming after immunization with sFliC, we assessed responses in Cd11ccre.Irf4fl/fl mice. These mice lack IRF4 in cells that express CD11c, resulting in a 50% reduction of CD103+CD11b<sup>+</sup> cDCs in the small intestine lamina propria and a 90% reduction in the MLN (19). **Figure 2A** shows the significant reduction of MLN CD103+CD11b<sup>+</sup> cDCs in Cd11c-cre.Irf4fl/fl mice in comparison with the Irf4fl/fl mice. Furthermore, we also show the expected reduction of splenic cDC2 in these mice (**Figure 2B**). The frequency and number of activated CD4<sup>+</sup> T cells in the MLN of Cd11c-cre.Irf4fl/fl mice 7 days after FliC immunization, was lower compared to Irf4fl/fl mice and similar to non-immunized mice (**Figure 2C**). In contrast, absolute numbers of activated CD4<sup>+</sup> T cells in the spleen were similar between Cd11c-cre.Irf4fl/fl and Irf4fl/fl mice (**Figure 2D**). To examine the endogenous antigenspecific T cell response we performed an in vitro re-stimulation essay that uses the transient expression of CD154 to identify antigen-specific CD4<sup>+</sup> T cells (23). In the MLN from immunized Cd11c-cre.Irf4fl/fl mice, the frequency and number of CD154<sup>+</sup> sFliC-specific CD4<sup>+</sup> T cells was significantly lower than those in immunized Irf4fl/fl mice and similar to levels observed in from non-immunized mice (**Figure 2E**). In contrast, in the spleen an increase of CD154<sup>+</sup> sFliC-specific CD4<sup>+</sup> T cells was observed

FIGURE 3 | The generation of primary B cell responses to sFliC in the MLN, but not the spleen, are dependent upon CD103+cDC2. Irf4fl/fl or Cd11c-cre.Irf4fl/fl mice were either non-immunized (N.I.) sFliC-immunized and GC B cells (TCRβ <sup>−</sup>CD19+GL7+CD95+) were evaluated 7 days later. (A) MLN and (B) spleen representative flow cytometry plots (percentages) and absolute numbers (graphs) of GC B cells. Data are mean+s.d. (n = 4 mice/group) representative experiment of 3 performed. \*\*P < 0.001; \*P < 0.05, by two-way analysis of variance (ANOVA), NS, not significant. (C) Serum anti-sFliC IgG, IgG1, and IgG2c evaluated by enzyme-linked immunosorbent assay (ELISA). Data shown as mean + s.d. (n = 12 mice/group) and shows three independent experiments pooled together. \*P < 0.05, two-way analysis of variance (ANOVA), N.S., not significant.

in the Cd11c-cre.Irf4fl/fl mice compared to the non-immunized mice but it was significantly lower compared to Irf4fl/fl mice (**Figure 2F**). Collectively, these results demonstrate that the early T cell response induced to sFliC is dependent on mucosal CD103+cDC2s, however in the spleen IRF4 expression by cDCs only impacts partially on T cell priming.

# cDC2 Influence the Extent and Direction of IgG Switching

In order to address how the B cell response to sFliC is affected in the Cd11c-cre.Irf4fl/fl mice we analyzed GC B cells by flow cytometry 7-days post-immunization. In the MLN, Cd11ccre.Irf4fl/fl mice showed no increase in the number of GC B cells in comparison to non-immunized and Irf4fl/fl sFliC-immunized mice (**Figure 3A**). In contrast, in the spleen, immunized Cd11ccre.Irf4fl/fl and Irf4fl/fl mice displayed a similar increase in GC B cell numbers (**Figure 3B**). To analyse the FliC-specific Ab response in more detail, serum Ab titres were evaluated by ELISA. Total sFliC-specific IgG titers were reduced in Cd11ccre.Irf4fl/fl mice in comparison with Irf4fl/fl mice (**Figure 3C**). In mice, sFliC induces some Th2-associated features including antigen-specific IgG1 (13, 14). FliC-specific IgG1 was detected in sFliC-immunized Irf4fl/fl mice, but was absent in Cd11ccre.Irf4fl/fl mice (**Figure 3D**). This was unexpected since some sFliC-specific IgG was detected in Cd11c-cre.Irf4fl/fl mice and so titers of Th1-associated IgG2c were assessed. This isotype was detected exclusively in sFliC-immunized Cd11c-cre.Irf4fl/fl mice (**Figure 3E**), suggesting that cDC2 contribute to IgG1 switching.

We hypothesized that the serum IgG2c derived from spleen. To analyze this possibility in situ, we performed immunohistochemistry on serial sections from the MLN and spleen of sFliC-immunized Irf4fl/fl and Cd11c-cre.Irf4fl/fl mice. GCs were identified as follicular areas that bind PNA and sFliC-specific cells were identified by using biotinylated sFliC in conjunction with either anti-IgG1 or anti-IgG2c Abs. In the MLN, sFliC-specific cells were exclusively observed in Irf4fl/fl mice and were IgG1<sup>+</sup> (**Figure 4A**). In contrast, sFliC-binding cells were found in the spleens of both Irf4fl/fl and Cd11ccre.Irf4fl/fl mice whilst in the Irf4fl/fl mice, the sFliC-specific cells were IgG1<sup>+</sup> and IgG2c−. In contrast, in Cd11c-cre.Irf4fl/fl mice the sFliC-specific cells were exclusively IgG2c<sup>+</sup> (**Figure 4B**). Collectively, these results show that cDC2s play a role in the polarization of the Ab response in vivo.

# DISCUSSION

We have previously shown that sFliC can drive a long-term mucosal adaptive response after i.p. immunization (1, 3). Furthermore, we also have shown that s.c. and i.p. immunization both induce similar cDC and IgA responses in the MLN, suggesting that our observations are not dependent of the route of immunization, but due to the intrinsic properties of sFliC. When addressing the prime-boost immune response in Cd11c-cre.Irf4fl/fl mice the immune response to sFliC in the spleen was reduced, but not abrogated, suggesting the possibility that memory cells could contribute to the response. To address this possibility we studied the primary immune response to sFliC and show that the primary T and B cell responses to sFliC in the MLN are completely dependent on CD103+cDC2, while that in the spleen is only partially dependent on cDC2.

Mucosal CD103+cDC2s are probably more efficient at driving responses after sFliC immunization because of their high

photomicrographs of serial sections from MLN and spleen stained for: PNA-binding cells (blue) and IgD-expressing cells (brown) (first column) or sFliC-binding cells (blue) and IgD-expressing cells (brown) (second column), scale bar = 200µm. The third and fourth columns show zoom-in insets (black-boxed areas) stained to detect sFliC-binding cells and IgG1 and IgG2c respectively (scale bar = 50µm). T, T zone; B, B zone. (B) Quantification of sFliC+IgG1<sup>+</sup> cells and sFliC+IgG2c<sup>+</sup> cells in the MLN and spleen. A total of 10 random fields were evaluated per slide. Data shown as mean + s.d. (n = 8 mice pooled from two independent experiments). \*\*\*P < 0.0001, by Mann-Whitney. N.D. non-detected.

expression levels of TLR5 in comparison to splenic cDCs (26, 27). TLR5 can play an additional role in enhancing antigen capture and presentation through MHC-II and this is not MyD88 dependent (28). Moreover, ligation of TLR5 itself will lead to an upregulation of co-stimulatory molecules and cytokine expression, in a MyD88-dependent manner (4, 14). This is likely to be the mechanism that mediates the accumulation of CD103+cDC2 into the MLN after immunization with sFliC (1, 3). Furthermore, splenic cDCs are able to respond rapidly after sFliC immunization, possibly through activation by a bystander effect (22, 29). Therefore, in the spleen there may be less of a selective advantage for one subset to capture sFliC over another, meaning that both cDC1 and cDC2 are potentially able to present antigen to CD4<sup>+</sup> T cells and initiate priming. In support of this idea ex vivo data using sorted, in vivo loaded, cDCs showed that both splenic cDC1 and cDC2 are able to mediate to T cell priming. Additionally, when flagellin is used as an adjuvant in studies using DEC205 and 33D1 to target splenic cDC1 and cDC2 cells, it shows that the presence of flagellin enhances the capacity of both cDC subsets to mediate T cell proliferation (30). This demonstrates that flagellin can promote responses in both cDC subsets, which indirectly supports our findings. In contrast to this, in the intestinal mucosa only CD103+cDC2 mediate T cell priming (3). Therefore, despite having a similar ontogeny, mucosal CD103+cDC2 and splenic cDC2 show differences in their capacity to capture sFliC and this difference may account for why cDC2 play such a dominant role in driving T cell responses to FliC in the MLN but not the spleen.

After primary immunization of WT mice with sFliC, there is a robust GC response in the spleen, but a limited extrafollicular plasma cell response (13, 22). The predominant antigen-specific IgG isotype detected after immunization with sFliC in the serum is IgG1, associated to a Th2-like response (13, 14). Importantly, in the Cd11c-cre.Irf4fl/fl mice there was no significant increase in numbers of GC B cells in the MLN. In contrast, in the spleen, there was a normal GC response, which suggested that a B cell response developed in the spleen but not in the MLN when CD103+cDC2 were reduced. Nevertheless, there were lower serum total IgG titers in the CD11-cre.Irf4fl/fl mice and an abrogated IgG response in the BM (3). One interpretation of this is that in the primary response to sFliC, the antibody response generated in the MLN is a significant, if not predominant, contributor, to the serum total FliC-specific IgG pool.

A more detailed analysis of the IgG response showed that there was a difference in the predominant IgG isotype induced between immunized Cd11c-cre.Irf4fl/fl and Irf4fl/fl mice. Inducing the appropriate IgG isotype is important as the distinct IgG isotypes can influence the level of protection afforded by vaccination (31) or against different pathogens (32). Surprisingly, the residual FliC-specific IgG response observed in Cd11c-cre.Irf4fl/fl mice was not of the IgG1 isotype, but instead was of the IgG2c isotype. Immunohistochemistry showed that the FliC-specific IgG2c was being produced locally in the spleen by cells proximal to GC (**Figure 3**). This suggests that although a B cell response is maintained in the spleens of the Cd11c-cre.Irf4fl/fl mice, this B cell response is substantially different qualitatively. Further work is needed to identify what B cell associated factors, such as BAFF or APRIL, cDC1 and cDC2 produce that may contribute to this and whether these differ between cDCs from different anatomical sites. In mice, there is a partial association between

#### REFERENCES


the direction of the T helper response and IgG isotype switching and so these findings may suggest that cDC2 contribute to Th2 associated responses. An association between cDC2 and Th2 polarization has been described previously in the context of infection or in atopic asthma models (21, 33–35). Moreover, we are working toward developing strategies to conjugate sFliC to different antigens and evaluate if these features that sFliC is able to promote as an adjuvant can be transferred to clinically relevant antigens. Our data helps inform on the relative merits of targeting specific DC populations in vaccination strategies.

#### AUTHOR CONTRIBUTIONS

Conceptualization: AF-L, AC, and WA. Methodology: AF-L, CC, NB-C, JY-P, IH, MD, and BL. Investigation: AF-L, KM, EP, JM, NB-C, and JY-P. Resources: JP, SU, and SA. Writing—original draft: AF-L. Writing, review, and editing: AF-L, AC, and WA. Funding acquisition: AF-L, AC, and WA.

#### FUNDING

This work was supported by grants from the BBSRC UK (BB/L009986/1) to AC, U21 staff fellowship program and Wellcome Trust ISSF Mobility grant to AF-L, and grants from the Danish Council for Independent Research (Sapere Aude III) and the Swedish Medical Research Council to WA.

#### ACKNOWLEDGMENTS

The authors would like to thank Dr. K. Kotarsky and A. Selberg for animal typing and husbandry (Lund University) and Ian Ricketts at the Biomedical Service Unit at the University of Birmingham.


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Flores-Langarica, Cook, Müller Luda, Persson, Marshall, Beristain-Covarrubias, Yam-Puc, Dahlgren, Persson, Uematsu, Akira, Henderson, Lindbom, Agace and Cunningham. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# CD169<sup>+</sup> Macrophages Capture and Dendritic Cells Instruct: The Interplay of the Gatekeeper and the General of the Immune System

Joanna Grabowska† , Miguel A. Lopez-Venegas † , Alsya J. Affandi and Joke M. M. den Haan\*

Department of Molecular Cell Biology and Immunology, Amsterdam University Medical Center, Cancer Center Amsterdam, Amsterdam Infection and Immunity Institute, Vrije Universiteit Amsterdam, Amsterdam, Netherlands

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

E. Ashley Moseman, National Institute of Neurological Disorders and Stroke (NINDS), United States Reinhard Obst, Ludwig-Maximilians-Universität München, Germany

> \*Correspondence: Joke M. M. den Haan j.denhaan@vumc.nl

†These authors share first authorship

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 26 July 2018 Accepted: 05 October 2018 Published: 26 October 2018

#### Citation:

Grabowska J, Lopez-Venegas MA, Affandi AJ and den Haan JMM (2018) CD169<sup>+</sup> Macrophages Capture and Dendritic Cells Instruct: The Interplay of the Gatekeeper and the General of the Immune System. Front. Immunol. 9:2472. doi: 10.3389/fimmu.2018.02472 Since the seminal discovery of dendritic cells (DCs) by Steinman and Cohn in 1973, there has been an ongoing debate to what extent macrophages and DCs are related and perform different functions. The current view is that macrophages and DCs originate from different lineages and that only DCs have the capacity to initiate adaptive immunity. Nevertheless, as we will discuss in this review, lymphoid tissue resident CD169<sup>+</sup> macrophages have been shown to act in concert with DCs to promote or suppress adaptive immune responses for pathogens and self-antigens, respectively. Accordingly, we propose a functional alliance between CD169<sup>+</sup> macrophages and DCs in which a division of tasks is established. CD169<sup>+</sup> macrophages are responsible for the capture of pathogens and are frequently the first cell type infected and thereby provide a confined source of antigen. Subsequently, cross-presenting DCs interact with these antigen-containing CD169<sup>+</sup> macrophages, pick up antigens and activate T cells. The cross-priming of T cells by DCs is enhanced by the localized production of type I interferons (IFN-I) derived from CD169<sup>+</sup> macrophages and plasmacytoid DCs (pDCs) that induces DC maturation. The interaction between CD169<sup>+</sup> macrophages and DCs appears not only to be essential for immune responses against pathogens, but also plays a role in the induction of self-tolerance and immune responses against cancer. In this review we will discuss the studies that demonstrate the collaboration between CD169<sup>+</sup> macrophages and DCs in adaptive immunity.

Keywords: CD169, siglec-1, sialoadhesin, macrophages, dendritic cells, T cell, antigen, cross-presentation

# INTRODUCTION

While the first recognized characteristic of macrophages was their excellent capacity to phagocytose, dendritic cells (DCs) were acknowledged for their superior ability to stimulate naïve T cell responses. However, ever since tissue macrophages and DCs showed overlapping expression of several markers and were both generated from monocytes in in vitro models, it has been debated whether these cell types were closely related and had equivalent functions. The introduction of unbiased single cell multi-parameter analyses on the protein and RNA level, and the

**119**

generation of cell-type specific and inducible genetically modified mouse models has enabled a new understanding of the generation and functions of both macrophages and DCs, and has even led to a new nomenclature (1). The current view is that the two cell types have very different functions in the immune system. However, this viewpoint potentially overlooks functional collaborations between the two cell types. In this review we will focus on the interactions between lymphoid tissue resident CD169<sup>+</sup> macrophages and DCs and how these support the activation of adaptive immune responses.

#### DCs and Macrophages Are Different Cell Types With Different Functions

The generation of macrophages is dependent on the growth factor M-CSF and occurs in three waves [reviewed by (2, 3)]. First, during early embryonic development, yolk sac-derived progenitors seed several peripheral tissues, such as the brain and the epidermis. A second wave of progenitors derive from the fetal liver and seed lungs and liver. These two types of macrophages are characterized by high expression of F4/80 and in general reconstitute autonomously. Additionally, they are thought to have a long half-life and exhibit local proliferation. After birth, monocytes develop from hematopoietic stem cells in the bone marrow and tissues, such as the intestines and the skin that continuously receive monocytes to generate macrophages. The latter macrophages generally express low levels of F4/80.

Macrophages form a very heterogeneous population of cells and their diversity in phenotype and function is a reflection of the variety of the tissues in which they reside [reviewed by (4, 5)]. They are best known for their capacity to phagocytose and eliminate pathogens and to alarm the immune system. In addition to this important function in immunosurveillance, they are essential for the clearance of apoptotic cells and suppression of (auto) immune responses and mediate resolution of inflammatory responses and tissue repair. Furthermore, depending on their tissue of residence, macrophages have important specialized functions in development, homeostasis and metabolism [discussed in more detail in (4, 6)]. The general view is that macrophages exert their functions locally in the tissues and that in steady state tissue resident macrophages do not migrate to secondary lymph nodes to activate naïve T cells. This latter function is attributed to DCs that also reside in tissues, but upon pathogen recognition, upregulate CCR7 and travel to the lymphoid organs. However, upon inflammation monocytederived macrophages or DCs may also acquire the capacity to travel to the lymph nodes and stimulate T cells, which is a matter that has to be further clarified (7).

Currently, three types of DCs are being recognized [reviewed by (8, 9)]. Conventional or classical DCs (cDCs) are continuously generated in the bone marrow and require Flt3L for their generation. Pre-cDCs seed the tissues and the lymphoid organs and have a half-life of 5–7 days. Upon activation and upregulation of CCR7, tissue cDCs migrate to the lymph nodes and can activate T cells. Within cDCs two subsets can be identified. The cDC1 is more specialized in the uptake of dying cells, crosspresentation and activation of CD8<sup>+</sup> T cells, while cDC2 has a more important role in CD4<sup>+</sup> T cell activation and B cell responses. The generation of these two subsets is dependent on different transcription factors. While cDC1 requires Batf3, Id2 and Irf8, cDC2 development depends on Irf4 and RelB and requires additional Notch2 and vitamin A signals (10). With regard to the surface phenotype, cDC1 can be identified by XCR1 and CLEC9A, and additionally by CD8α in lymphoid organs and by CD103 in peripheral tissues. On the other hand, Sirpα, CD11b and CD4 expression marks the cDC2 subset. Next to cDCs, pDCs form another class of DCs that also develop in a Flt3L-dependent manner. This lineage splits from the cDC lineage before the separation in cDC1 and cDC2. They can be identified by CD123, BDCA2, and BDCA4 in humans and by high expression of BST2 and Siglec-H and by low expression of CD11c and B220 in the mouse. Recent studies have indicated further heterogeneity in CD123-expressing pDCs (11, 12). While early studies indicate that pDCs can take up antigens and stimulate T cells upon activation, recent studies suggest that very pure pDC populations only produce IFN-I and are not able to activate T cells unless they are pre-treated with CD40L and IL-3 (13). This suggests a limited function for pDC in T cell activation.

Next to these two Flt3L-dependent DC subsets, DCs can differentiate from monocytes during inflammatory conditions (7). The function of these DCs in the regulation of adaptive immune responses remains to be elucidated.

# Antigen Cross-Presentation by Macrophages and Dendritic Cells

Both macrophages and DCs process antigens via the classical endogenous and exogenous pathways and present these on their MHC class I and II molecules, respectively, but they differ in their capacity to cross-present exogenous antigens in MHC class I and to cross-prime CD8<sup>+</sup> T cells. Cross-presentation was first described in 1976 by Bevan as the process in which CD8<sup>+</sup> T cell responses were initiated against donor antigens restricted by recipient MHC molecules (14). This process is thought to be essential in the activation of anti-viral and anti-tumor specific CD8<sup>+</sup> T cell responses. While a number of studies have shown that exogenous antigens can be cross-presented by different cell types including macrophages (15), the mouse cDC1 subset exhibits a higher capacity to cross-present and is especially equipped for the uptake of dead cells and the cross-presentation of cell-associated antigens (16–19). However, depending on the antigen and activation stimuli, mouse cDC2 and several human DC subsets are also able to cross-present (17, 20). There are two main routes of antigen processing exist that leads to crosspresentation. In the cytosolic route, antigens are transported from the endosomal/phagosomal pathway to the cytoplasm and this pathway depends on proteasomes and TAP. In the vacuolar route, antigens are degraded in the endosomal/phagosomal pathway and bind to recycling MHC class I molecules. This pathway relies on the activity of cathepsin S. DCs mainly utilize the cytosolic route, while macrophages and monocytederived DCs have been shown to use the vacuolar route of cross-presentation (21, 22). Recent studies have identified a number of molecules involved in vesicular trafficking that play a role in cross-presentation [see for more details reviews (15, 23, 24)]. One of the important factors for cross-presentation is the rate of antigen degradation. Macrophages are more proteolytically active than DCs, which impairs their capacity to cross-present (25). DCs prevent the acidification of their phagosomes and thereby inhibit proteolysis by the activity of NADPH oxidase 2 (NOX2) enzyme [reviewed in (26)]. The NOX2 enzyme may also contribute to the translocation of antigens to the cytosol by disrupting the phagosomal membrane. The longer preservation of antigens in DCs and stronger phagosome-cytotosol translocation compared to macrophages may be responsible for the more prominent role of DCs in cross-priming.

# Generation of CD169<sup>+</sup> Macrophages and Their Innate Functions

Macrophages expressing high levels of CD169, also known as Siglec-1 or sialoadheasin, constitute a minor macrophage population present in lymphoid tissues (27, 28). While several macrophage populations in tissues have low levels of CD169, which can be upregulated upon exposure to IFN-I, this lymphoid resident population has a very high constitutive expression of CD169. CD169<sup>+</sup> macrophages are situated on top of B cell follicles bordering the marginal sinus in the spleen and the subcapsular sinus (SCS) in the lymph nodes and are also known as metallophilic marginal zone macrophages and SCS macrophages, respectively. The presence of B cells is necessary for the generation of CD169<sup>+</sup> macrophages, which is mediated by their production of LTα1β2 (29, 30). In addition, they require RANK, LXR, and M-CSF signals and their survival is further promoted by TNF-α (31–34). Currently it is unclear which precursor gives rise to CD169<sup>+</sup> macrophages, although their low level of F4/80 expression would suggest that they are not derived from yolk sac precursors. After elimination, they are repopulated from monocytes (34).

The strategic position of CD169<sup>+</sup> macrophages at the entry site of lymphoid tissues determines their function. CD169<sup>+</sup> macrophages are the first cell type in the spleen and lymph nodes to bind particulate antigens and pathogens and they function as a filter to remove foreign particles from the lymph fluid and blood. When these cells are deleted by clodronate liposomes in an experimental setting, pathogens can disseminate to other organs as has been demonstrated for several viral, bacterial and parasitic infections (35–39). This particular observation coined the term "gatekeeper" to describe CD169<sup>+</sup> macrophages. This first line of defense, capturing invading viruses and limiting their spread to other organs, is not only mediated via the physical binding and capture of pathogens. CD169<sup>+</sup> macrophages also exert their protective functions by the production of cytokines, such as IFN-I, IL-1, and IL-18. This cytokine secretion not only prevents subsequent infection of other cells and activates innate lymphocytes that help to contain the early infection (40–42), but also acts on DCs and stimulate adaptive immune responses.

# Model Systems to Study CD169<sup>+</sup> Macrophages

Due to their low abundance and sensitivity to manipulation, CD169<sup>+</sup> macrophages are quite an enigmatic and technically challenging subset to study. Although it is feasible to extract these cells from spleen or lymph nodes by combination of mechanistic dissociation and enzymatic digestion, they rapidly die and form apoptotic blebs that bind to interacting cells (41, 43, 44). This feature greatly hampers the purification of CD169<sup>+</sup> macrophages using fluorescence-activated cell sorting (FACS) for in vitro analysis. Unfortunately, available in vitro models do not offer a satisfactory method to investigate this macrophage population. In vitro cultured macrophages can be treated with IFN-α, which induces CD169 expression on the cell surface, but it is not clear whether these cells exhibit other characteristics of the CD169<sup>+</sup> macrophages present in vivo. Most studies investigating CD169<sup>+</sup> macrophages take advantage of cell ablation tools, either chemical using clodronate liposomes or genetic using diphtheria toxin receptor (DTR) systems. Despite representing a very effective method for transient depletion of macrophages, clodronate liposomes lack specificity. This apoptosis-inducing agent is toxic for all phagocytosing cells including monocytes and DCs (45). Noteworthy, the treatment with clodronate liposomes affects the anatomy of the surrounding tissue and induces off-target effects on B cells (46). In comparison to clodronate liposomes, DTR-mediated cell ablation allows for conditional and targeted depletion of a cell subset engineered to express DTR. CD11c-DTR and CD169-DTR are two DTR transgenic mouse strains that deplete CD169<sup>+</sup> macrophages (47, 48). Although the CD11c-DTR model mainly depletes cells with high expression of CD11c, thus DCs, it does not spare macrophages that express low levels of this DC marker (49). The CD169-DTR model, on the other hand provides a more specific approach to study CD169<sup>+</sup> macrophages, leaving the DC population unaffected. The only other population affected by DT treatment in CD169-DTR model, are SIGN-R1<sup>+</sup> marginal zone macrophages that express low levels of CD169 (47). Similarly, the LXR-α KO lack both CD169<sup>+</sup> and SIGNR1<sup>+</sup> splenic marginal zone macrophage subsets (34). More recently, CD169-Cre mice have been generated and when crossed to the ROSA26-YFP mice generate reporter mice (50). The CD169-Cre mice will allow the generation of CD169-specific conditional KO mice and is therefore expected to provide a wealth of new insights for this macrophage population.

# CD169<sup>+</sup> MACROPHAGES AND IFN-I PRODUCTION

Upon encounter with pathogens, such as viruses, CD169<sup>+</sup> macrophages regulate pathogen spread and induce immune responses by producing IFN-I. IFN-I consist of a single IFN-β and several subtypes of IFN-α, that signal through IFN-I receptor, a shared receptor expressed in almost all cell types (51). The importance of IFN-I signaling is 2-fold: (1) IFN-I can induce intracellular antiviral responses to suppress viral replication in the infected cells (52), and (2) IFN-I can regulate both innate and adaptive immune responses that are required to clear pathogens. However, depending on the type of pathogen, the outcome of IFN-I actions can play both protective and detrimental roles to the host (53, 54).

# Viral Infection of CD169<sup>+</sup> Macrophages Results in IFN-I Production

CD169<sup>+</sup> macrophages rapidly produce IFN-I after infection and thereby restrict the spread of a variety of viruses including mouse cytomegaloviruses (CMV), herpesvirus, and lymphocytic choriomeningitis virus (LCMV) (55–58). Several studies using neurotropic vesicular stomatitis virus (VSV) infection show that IFN-I signaling is necessary for the survival of the mice. Upon VSV infection, IFN-I was shown to be largely produced by CD169<sup>+</sup> macrophages and this prevented VSV from entering the central nervous system (38). Similarly, during experimental infection with recombinant modified vaccinia virus Ankara (MVA), CD169<sup>+</sup> macrophages were found to be the main IFN-I producers (59). In this model, CD169<sup>+</sup> macrophages recruited and activated NK cells upon MVA infection, which was dependent on the production of IFN-I by CD169<sup>+</sup> macrophages. Additionally, MVA infection also induced inflammasome activation by CD169<sup>+</sup> macrophages that led to pyroptotic cell death, cytokine burst, and recruitment of inflammatory cells (60).

# CD169<sup>+</sup> Macrophages Recruit and Prime IFN-I Production by pDCs

Next to CD169<sup>+</sup> macrophages, pDCs are well-known for their capacity to produce IFN-I. They express TLR7 and TLR9 and high basal levels of IRF7 that allows them to detect intracellular nucleic acids and to produce IFN-α immediately upon encounter with pathogens (61). pDCs are located mainly in the lymphoid organs, such as bone marrow, spleen, and lymph nodes, but not in non-lymphoid tissues. In steady state, pDCs can be found in the T cell zone and peri-follicular area of the lymph node. Upon infection with pathogens, such as VSV, pDCs migrate to the SCS and medulla, areas rich in CD169<sup>+</sup> macrophages (38). pDCs were reported to account for half of the IFN-I produced upon VSV infection, which was dependent on the presence of CD169<sup>+</sup> macrophages. In a another study, the migration of pDCs to SCS was shown to be mediated by CXCR3, chemokine receptor of CXCL9, CXCL10, amongst others (62). It was suggested that viral particles from the infected CD169<sup>+</sup> macrophages could activate these migrating pDCs. However, the direct interaction between SCS CD169<sup>+</sup> macrophages and pDCs and its consequences are still unclear.

In a malaria infection model, pDCs accounted for the majority of IFN-I produced which led to lethal outcomes of infected mice (63). Spaulding et al. reported that after infection with malaria, CD169<sup>+</sup> macrophages sustained prolonged interaction with pDCs in the bone marrow and primed them to produce IFN-I. Thus, this study provides evidence of an active interaction between CD169<sup>+</sup> macrophages and pDCs that may also occur in other lymphoid organs.

However, pDC-derived IFN-I may be dispensable in some situations. In a study that exploited an MCMV footpad infection model, pDC depletion using αBST2 antibodies led to an increase in MCMV escape from SCS and spread to other tissues (55). Nevertheless, this effect was moderate when compared to blocking IFN-I using anti-IFN-I receptor antibodies. In another MCMV model where MCMV was administered intraperitoneally, depletion of pDCs also resulted in an increase of viral spread and dissemination, but only when a low dose was used (64). pDCs were also demonstrated dispensable for survival of the mice upon infection with VSV and Plasmodium (38, 63). Thus, upon pathogen encounter by CD169<sup>+</sup> macrophages, pDCs are recruited to amplify IFN-I signaling, however this is not always essential for pathogen clearance or mice survival. Nevertheless, pDCderived IFN-I may still contribute to other aspects of immune responses.

#### IFN-I Augments cDCs to Initiate Adaptive Immune Responses

The initiation of adaptive immune response by cDCs involves multiple mechanisms including antigen presentation, costimulatory/inhibitory molecules, and immunomodulation by cytokines. Next to its role in inhibiting viral replication, IFN-I has been demonstrated to augment NK cell function, B cell isotype switching, and T cell survival and activation (65). IFN-I is also critical for the function of cDCs to fully activate naïve T cells as it stimulates the expression of co-stimulatory molecules, enhances responses to TLR-ligands and increases antigen presentation capacity (66–69). cDC1, in particular, require the presence of IFN-I for antigen cross-presentation and subsequent CD8<sup>+</sup> T cell activation (70). Several reports have demonstrated IFN-activated cDC1 to be important for generating CD8<sup>+</sup> T cell responses against tumor or viral infections (71–73). In fact, IFN-I signaling induced by viruses could enhance the development of CD8<sup>+</sup> T cell-mediated anti-tumor responses as a vaccination strategy (70, 74, 75).

Studies have been performed to identify the source of IFN-I required for the maturation of cDCs. In a vaccination system using tumor protein antigen and an iNKT cell ligand α-GalCer, splenic pDCs produced high amounts of IFN-I (76). Importantly, prior to cDC1 trafficking to the white pulp for T cell stimulation, pDCs were found to cluster with cDC1s in the CD169<sup>+</sup> macrophage-rich marginal zone and red pulp area of the spleen. It was further shown that abolishing IFN-I signaling in CD11c<sup>+</sup> cells led to an impaired memory T cell formation. This was in line with a previous study, where pDCs were reported to promote the generation and survival of antigen-specific CD8<sup>+</sup> T cells upon VSV infection (64). More recently, Brewitz and colleagues have demonstrated pDC-derived IFN-I to be important for CD8<sup>+</sup> T cell activation by cDC1 when mice were exposed to MVA (62). After MVA infection, pDCs, cDC1s, and CD8<sup>+</sup> T cells formed superclusters in the interfollicular area of the lymph node. This event was required for CD8<sup>+</sup> T cell responses. Additionally, in a vaccination strategy using TLR7 agonist as an adjuvant, pDC-derived IFN-I was crucial for in vivo CD8<sup>+</sup> T cell killing (77). These observations suggest an important cross-talk between IFN-I-producing pDCs and CD8<sup>+</sup> XCR1<sup>+</sup> cDC1 for an optimal CD8<sup>+</sup> T cell activation in vaccination or viral infection.

The effect of IFN-I derived from CD169<sup>+</sup> macrophages and pDCs on the function of cDCs is not limited to CD8<sup>+</sup> T cell activation. Upon infection with S. mansoni eggs, IFN-I was needed for an optimal cDC activation, migration and induction of Th2 immune responses in vivo (78). In a DC-targeting vaccination using HIV gag-protein and poly(I:C) as an adjuvant, CD4<sup>+</sup> Th1 responses were abolished upon interference with IFN-I signaling (79). Next to T cells, IFN-I signaling on DCs could also mediate B cell function including antibody production, isotype switching and the development of T follicular helper cells (80, 81). Thus, IFN-I stimulated DCs have an enhanced capacity to activate both humoral and cell-mediated adaptive immune responses.

A similar priming effect of IFN-I on cross-presentation has also been shown in human DCs (82, 83). In humans, the level of IFN-I is highly elevated and has been suggested to contribute to the break of tolerance in many autoimmune diseases (84). For example in systemic lupus erythematosus (SLE), the increased level of IFN-I produced by pDCs directly induced cDCs maturation and CD4<sup>+</sup> T cell activation (85). In psoriasis, pDCderived IFN-I was sufficient to drive T cells infiltration and psoriatic plaque lesion formation (86). Interestingly, the numbers of CD169-expressing monocytes/macrophages were increased in the circulation and affected tissues of patients with systemic sclerosis and multiple sclerosis (87, 88). More investigation is needed to clarify the intricate cross-talk of CD169<sup>+</sup> macrophage and pDC-derived IFN-I, cDC1, and T cell immunity in human diseases.

#### Suppressive Effects of IFN-I

Of note, the role of IFN-I during an infection is largely context-dependent and can also result in immunosuppression. A sustained IFN-I production can lead to increase of IL-10 and a higher expression of PD-L1. In a model of a persistent infection using LCMV strain Docile, upregulation of PD-L1 expression by CD169<sup>+</sup> macrophages was important to promote CD8<sup>+</sup> T cell exhaustion and prevented lethal immunopathology (58). The increased expression of PD-L1 in CD169<sup>+</sup> macrophages was also observed in infection model with other LCMV strains (89). In addition, chronic infection with LCMV led to a sustained IFN-I production that prevented mice from mounting immune responses to a secondary infection by VSV (90). This was due to a reduced viral replication in CD169<sup>+</sup> macrophages and subsequent impaired antigen presentation and lack of adaptive immune responses, rather than immunosuppression. However, using a model of E. coli-induced septic shock and subsequent systemic challenge with ovalbumin (OVA)-containing viruses, Schwandt et al. demonstrated that mice with sepsis had reduced antigen-specific CD8<sup>+</sup> T cell responses. This suppression was mediated by macrophage-derived IFN-I that hampered cDC1 function to activate CD8<sup>+</sup> T cells (91). Together these studies indicate that during chronic infections IFN-I production by CD169<sup>+</sup> macrophages inhibits activation of immune responses toward secondary infections.

In conclusion, the production of IFN-I by CD169<sup>+</sup> macrophages, potentially amplified by pDC-derived IFN-I, can strongly stimulate cDC function and the activation of immune responses, but may also result in immunosuppression.

### CD169<sup>+</sup> MACROPHAGES EFFICIENTLY CAPTURE PATHOGENS AND MEDIATE ANTIGEN TRANSFER

Their strategic location in spleen and in lymph nodes endows CD169<sup>+</sup> macrophages with the capacity to capture blood- and lymph-borne pathogens. In fact, CD169<sup>+</sup> macrophages appear to be extremely efficient in this process, as showed by multiple groups using various infection models (37, 38, 40, 92–96). Having acquired viral antigens, CD169<sup>+</sup> macrophages were reported to transfer antigen to DCs and B cells mainly contributing to the infection control but also to virus dissemination in some cases.

# CD169<sup>+</sup> Macrophages Enable Containment of Viral Infection and Localized Production of Antigen

The role of CD169<sup>+</sup> macrophages as efficient gatekeepers has been demonstrated in a large number of viral infections, such as adenovirus, vaccinia virus, West Nile virus, and VSV (37, 92, 97). Additionally, experiments with human immunodeficiency virus (HIV) and murine leukemia virus (MLV) models confirmed prompt and potent virus capture by these gatekeeping macrophages (93). Deletion of splenic CD169<sup>+</sup> macrophages was reported to cause rapid dissemination of LCMV and herpes virus infection (35, 98). Along the same line, local depletion of SCS macrophages resulted in higher viral titers in the spleen and other organs providing direct evidence for the protective role of CD169<sup>+</sup> macrophages in systemic viral spread (37, 38, 40) (99). This clearly demonstrated the importance of CD169<sup>+</sup> macrophages in infection containment.

Paradoxically, CD169<sup>+</sup> macrophages can also support virus replication (33, 38, 99). Enforced virus replication within CD169<sup>+</sup> macrophages endowed them with the distinct feature of being a source of viral antigen that facilitated activation of adaptive immune responses. Accordingly, increased expression of inhibitory protein Usp18 rendered splenic CD169<sup>+</sup> macrophages unresponsive to IFN-I. As a consequence, enhanced cytopathic VSV replication in these cells was facilitated (94). CD169<sup>+</sup> macrophage-mediated VSV replication mediated a strong VSV-neutralizing antibody response that rescued infected animals. Positive correlation between viral replication in CD169<sup>+</sup> macrophages and protective adaptive immune responses was also shown in LCMV infection (100).

# CD169<sup>+</sup> Macrophages Transfer Antigens to DCs in Viral Infections

Apart from effective viral capture and containment of the infection, CD169<sup>+</sup> macrophages have been previously reported to directly present particulate antigens, immune complexes as well as intact virus particles to non-cognate and cognate B cells (37, 101–103). This process was shown to stimulate germinal center responses and production of high affinity antibodies (103, 104). While in these initial studies that used clodronate liposomes, B cells were still activated in the absence of SCS macrophages (38), a recent study indicated that absence of SCS macrophages led to defective B cell responses (105). This process of intact virus presentation to B cells by CD169<sup>+</sup> macrophages has also been implicated in trans-infection of B cells, contributing to the virus dissemination rather than to the virus containment (discussed in more detail in section CD169 as a viral receptor that mediates virus capture and trans-infection).

Despite robust evidence proving the importance of CD169<sup>+</sup> macrophages in the induction of anti-viral B cell responses, their role in the activation of T cell responses is still being elucidated. While a number of studies demonstrate that CD169<sup>+</sup> macrophages are dispensable for T cell priming (35–38, 94, 97), interaction between CD169<sup>+</sup> macrophages and cDC1s has been shown to promote anti-viral T cell responses (44, 106, 107). The study by Backer et al. indicated that CD169<sup>+</sup> macrophages could transfer antigens to cDC1s for the stimulation of CTL responses (106). In line with this, Bernhard et al. showed that antigen transfer between CD169<sup>+</sup> macrophages and cDCs also occurred in adenoviral infection. Interestingly, CD169<sup>+</sup> macrophages were also able to directly present viral antigens to T cells bypassing the need for cDCs for T cell priming. While all epitopes, including low affinity peptides, were directly presented by CD169<sup>+</sup> macrophages, cDC1s only cross-presented high affinity T cell epitopes (107).

Recently, the collaboration between CD169<sup>+</sup> and cDC1s was investigated in more detail (44). This study revealed that the CD169 receptor enabled cell-cell contact with sialylated ligands on cDCs and thereby facilitated transfer of antigen to cDCs. In addition to mediating adhesion to DCs, CD169 has also been reported to support binding of innate-like lymphocytes and neutrophils (41, 108, 109). Remarkably, even upon disintegration, CD169<sup>+</sup> SCS macrophage cell-derived blebs are able to bind to IL-17 lymphocytes and NK cells (41, 43). Apparently, CD169 acts as an adhesion receptor that facilitates the interaction of CD169<sup>+</sup> macrophages with other innate immune cells.

Interestingly, in vivo blockade of CD169 receptor resulted in impaired MVA-specific, but not VSV-specific CD8<sup>+</sup> T cell responses (44). This observation could be explained by the dispensability of the cross-presentation process during certain viral infections, such as VSV in which DCs are likely to be directly infected (94). Specifically, KLRG-1low CD8<sup>+</sup> T cells with memory potential were negatively affected upon CD169 blocking in MVAinfected animals, indicating that CD169<sup>+</sup> macrophage-mediated antigen transfer to cDC1s might facilitate memory responses as well. In line with this, collaboration between splenic CD169<sup>+</sup> macrophages and cDC1s was important for activation of memory CD8<sup>+</sup> T cell responses in VSV infection (33).

Van Dinther et al. showed that CLEC9A/DNGR-1 expressed on cDC1 enhanced CD8<sup>+</sup> T cell cross-priming of antigens targeted to CD169<sup>+</sup> macrophages (44). CLEC9A/DNGR-1 binds to F-actin exposed on dying cells and while it does not increase antigen transfer, it enhances T cell responses toward cellassociated material and in viral infections (110–112). A number of studies have indicated the disappearance or death of CD169<sup>+</sup> macrophages induced by viral infection or other inflammatory agents (44, 60, 105). This suggests that upon infection, CD169<sup>+</sup> macrophages quickly die and thereby form a cellular substrate for antigen transfer by the cross-presenting cDC1. This process could be of particular importance in viral infections, such as MVA, that solely depend on cross-presentation as opposed to VSV where the virus directly infects DCs (44).

#### CD169 as a Viral Receptor That Mediates Virus Capture and Trans-infection

A decade ago, CD169 expressed on monocyte-derived DCs was found to promote HIV infection. This discovery brought a paradigm shift in the HIV field with CD169 replacing DC-SIGN as the main capture receptor responsible not only for HIV adhesion, but also for trans-infection (113–116). Following binding of CD169 to virus membrane-associated glycolipids (GM3), HIV-1 and CD169 were demonstrated to travel together to and accumulate at a non-lysosomal compartment. Consequently, the concentration of HIV-1 and CD169 at the so called infectious synapse enabled trans-infection of CD4<sup>+</sup> T cells (117). A similar trans-infection process was also shown to be important for henipavirus infection (118).

In a study that focused on MLV and HIV infection in vivo, CD169-mediated virus capture was also reported to occur via CD169 binding to gangliosides on the viral membrane (93). Interestingly, CD169<sup>+</sup> macrophages that had captured MLV, but were not infected themselves, were responsible for transinfection of permissive B cells which facilitated spread of the infection. Accordingly, considerably lower numbers of virusinfected cells were detected both in peripheral lymph nodes and spleen upon blocking of CD169 and in CD169-deficient mice. This clearly illustrated the importance of CD169 for effective virus dissemination. In line with this, MLV was also demonstrated to exploit CD169 expressed on primary mouse bone marrow macrophages for trans-infection of proliferating B cells (95). Apart from aforementioned retroviral models, a study performed in a porcine reproductive and respiratory syndrome virus (PRRSV) also experimentally addressed the role of CD169 in virus anchoring (119). The authors proved that the attachment of the virus was dependent on the sialic acid binding activity of the receptor that binds to sialylated viral glycoproteins on PRRSV.

While substantial evidence from retroviral studies validates CD169 as a viral receptor that is exploited by the pathogen for its dissemination, numerous studies in viral models have demonstrated the importance of CD169 expressing macrophages for the containment of viral infection and localized production of antigen. The latter suggests that these macrophages form a reservoir of viral antigen for transfer to cDC1. A small number of studies suggest that a similar process may take place in certain bacterial infections.

# CD169<sup>+</sup> Macrophages Efficiently Trap Bacteria and Allow Trans-infection of cDCs

Similar to what has been shown in viral infections, several studies using the Listeria monocytogenes (Lm) model confirmed CD169<sup>+</sup> macrophages as the initial cellular host that effectively traps the bacteria (36, 120–122). While as early as 2 h post-infection, the majority of Lm was detected within macrophages in the marginal zone, by 9 h CD11c<sup>+</sup> DCs were the main cell type carrying Lm (121). Two photon microscopy results showed clustering of Lm-specific T cells that associated with CD11c<sup>+</sup> DCs in periarteriolar lymphoid sheath (PALS), which was indicative of ongoing antigen presentation. At 24 h Lm-infection foci were mainly localized to PALS where Lm was shown to replicate extensively. Using a CD11c-DTR model that allows for CD11c depletion upon DT injection, the authors confirmed that Lm transport to the PALS and subsequent antigen presentation were dependent on the presence of cDCs. However, as mentioned already, CD11c-DTR model also abrogates CD11c- expressing CD169<sup>+</sup> macrophages. Therefore, only subsequent experiments performed in the Batf3−/<sup>−</sup> model, formally established the role of cross-presenting cDC1 in Lm delivery to the PALS (120, 122).

Recently, Perez et al. (122) also noted a shift in Lm distribution from CD169<sup>+</sup> macrophages to cDC1 over the course of infection and showed that CD169<sup>+</sup> macrophages mediate trans-infection of cDC1. Accordingly, while in wild type animals cDC1 formed clusters near Lm-infected CD169<sup>+</sup> macrophages in the marginal zone and efficiently delivered Lm to PALS, in CD169-DTR mice such clusters were not present and transport to PALS was impaired. Therefore, the presence of CD169<sup>+</sup> macrophages closely interacting with cDC1 promoted trans-infection and enabled subsequent Lm entry to the PALS.

Similar to viral infections, CD169<sup>+</sup> macrophages also control the spread of bacteria. Perez and colleagues reported increased bacterial titers in the spleen and blood of CD169-DTR mice, suggesting that these macrophages impede Lm replication and prevent Lm dissemination (122). Finally, using a CD169-DTR-Batf3−/<sup>−</sup> model, that allows for conditional depletion of CD169<sup>+</sup> macrophages in cDC1-deficient mice, it was demonstrated that rapid Lm capture and clearance secured by CD169<sup>+</sup> macrophages was instrumental for Lm control. Interestingly, the authors showed that cytosolic replication within CD169<sup>+</sup> macrophages due to phagosomal escape was necessary for recruitment of cDC1.

While cDC1s have been identified as replication- permissive cellular hosts for Lm, a recent study demonstrated that CD169<sup>+</sup> macrophages can have a similar role in pneumococcal septicaemia (123). Upon infecting CD169<sup>+</sup> macrophages, Streptococcus pneumoniae evaded phagosomal clearance, proliferated intracellularly and after causing cell lysis disseminated to the bloodstream. The authors concluded that intracellular replication within CD169<sup>+</sup> macrophages is crucial for resulting pneumococcal septicaemia.

Collectively, the findings from studies in bacterial infections, albeit almost exclusively performed in the Lm model, illustrate the importance of CD169<sup>+</sup> macrophages as the initial cellular host. By capturing the bacteria, CD169<sup>+</sup> macrophages initially mediate pathogen clearance and prevent systemic spread of the infection. However, they also serve as a bacterial reservoir that actually promotes propagation of the bacteria into the bloodstream at a later stage in the case of Streptococcus pneumonia or enable trans-infection of cDC1 by Lm. In addition to these two bacterial infections, the CD169 molecule has been shown to function as a receptor for bacterial uptake of pathogens rich in sialylated polysaccharides, such as Neisseria meningitidis, Campylobacter jejuni, and Trypanosoma cruzi (124–126). It remains to be established whether CD169<sup>+</sup> macrophages function as a bacterial and/or antigen reservoir in these infections.

### Uptake and Transfer of Apoptotic Cellular Material by CD169<sup>+</sup> Macrophages and the Implications for Tolerance and Cancer Immunity

The distinction between self and non-self is essential for the proper function of the immune system. Next to their essential role in initiating immune responses specific for pathogens, CD169<sup>+</sup> macrophages have also been shown to play a role in the induction of tolerance and anti-cancer immune responses.

# Role of CD169<sup>+</sup> Macrophages in Tolerance

Continuous and non-inflammatory removal of apoptotic cell material is essential for the maintenance of tolerance. Using a transfer model of apoptotic cells, cDC1 cells were specifically shown to take up and present these cell-associated antigens to CD8<sup>+</sup> T cells (16, 18, 19) and subsequently induce tolerance in the steady state (127). One of the first observations indicating a tolerogenic function for CD169<sup>+</sup> macrophages was made by Miyake et al., who generated CD169-DTR mice in which all marginal zone macrophages were eliminated upon injection with DT (47). Upon injection of apoptotic cells loaded with a fragment of the myelin oligodendrocyte glycoprotein peptide (MOG peptide), an accumulation of apoptotic cell content was observed in the marginal zone in wild type mice, which prevented the development of EAE. Depletion of marginal zone macrophages via DT administration in CD169-DTR mice resulted in a failure of induction of tolerance and a switch in the uptake of apoptotic cells from CD8<sup>+</sup> cDC1s to CD8<sup>−</sup> cDC2s (47).

Next to cDC2s, also red pulp macrophages, have been accounted for the defective uptake of apoptotic cells and the abrogation of tolerance in the absence of marginal zone macrophages. When marginal zone macrophages were depleted by means of clodronate liposomes, an accumulation of apoptotic cells was detected in F4/80<sup>+</sup> macrophages. This was correlated with the production of inflammatory cytokines and loss of tolerance induction (128).

In subsequent studies by McGaha et al. the interaction between CD169<sup>+</sup> macrophages and DCs was investigated. In their system, intravenous injection of apoptotic cells induced the expression of CCL22 on CD169<sup>+</sup> macrophages, which resulted in a coordinated clustering of CCR4-expressing cDC1s and regulatory T cells within the white pulp. The induction of tolerance was dependent on both CD169<sup>+</sup> macrophages and CCR4 (129). In contrast, another study reported that CCL22 is produced by the cDC1s upon injection with apoptotic cells, showing that the role of the cell type that produces CCL22 remains to be clarified (130). However, together these studies indicate that marginal zone CD169<sup>+</sup> macrophages and cDC1s are essential in the induction of tolerance via the uptake of apoptotic cells and suggest a functional collaboration in this process.

# Uptake of Tumor Cell Material and Exosomes by CD169<sup>+</sup> Macrophages Stimulate Anti-cancer Immunity

The previously discussed role of CD169<sup>+</sup> macrophages in mediating the removal of dying cells from the circulation to induce tolerance suggests that a similar process could potentially be involved in anti-tumor immunity. In this sense, a number of factors have been proposed to shift the balance of tolerance toward immunity. Whether DCs induce immunity is a contextdependent process, influenced by environmentally provided stimuli, stage and type of cell death as well as the location where it takes place (131, 132). An example of this has been provided by Lorenzi and colleagues, who demonstrated enhanced intracellular persistence of antigenic particles in cDC1 upon injection of tumor apoptotic cells in combination with IFN-I. After exposure to IFN-I, cDC1 not only contributed to the induction of OT-I proliferation, but also exhibited an enhanced lifespan and expression of co-stimulatory molecules (133). Since CD169<sup>+</sup> macrophages can produce high amounts of IFN-I, in combination with antigen this could provide the optimal stimulus for DCs to be able to cross-present cell-associated tumor antigens and to induce T cell activation.

However, the question remains whether CD169<sup>+</sup> macrophages have the capacity to cross-present tumor antigens autonomously. One of the first studies exploiting subcutaneously-injected dead cells showed these cells being transported throughout the lymphatic system to the lymph nodes, where SCS macrophages cross-presented dead cellassociated antigens to CD8<sup>+</sup> T cells. Mice that were lacking SCS macrophages at the moment of vaccination did not reject the tumors successfully (134). Interestingly, in this model the CD169<sup>+</sup> macrophages, and not cDC1, were thought to directly cross-prime CD8<sup>+</sup> T cells. This is reminiscent of the direct presentation of adenoviral antigens in the study of Bernhard et al., although the latter cannot be formally referred to as crosspresentation (107). Further studies are necessary to determine whether CD169<sup>+</sup> macrophages can cross-prime CD8<sup>+</sup> T cells independently or always require the collaboration with cDC1s.

In a model in which apoptotic cells were injected in vivo and induced CD4<sup>+</sup> T cell activation, again macrophages were shown to be the main cells involved in the uptake and in their absence or the absence of cDC1 the CD4<sup>+</sup> T cell activation was significantly decreased (135). Of note, an exosomal pathway was indicated to play a role in the cell-associated antigen transfer of macrophages to DCs. Exosomes are produced by many cell types and consist of small membrane vesicles that contain proteins, lipids, and nucleic acids. These vesicles can mediate transfer of such encapsulated molecules and thereby facilitate communication between cells (136). Exosomes have been found to be efficiently taken up by CD169<sup>+</sup> macrophages and cDC1 in the spleen (137). McLellan and colleagues demonstrated that exosomes can express high levels of α2,3-linked sialic acids and bind abundantly to CD169<sup>+</sup> macrophages in the spleen. Interestingly, CD169-deficient mice raised stronger CD8<sup>+</sup> T cell responses toward antigen-pulsed exosomes than wild type mice (138). A similar suppressive role of CD169<sup>+</sup> macrophages was observed in the T cell response toward tumor-derived apoptotic vesicles (139).

These studies suggest that CD169<sup>+</sup> macrophages scavenge exosomes and thereby prevent their uptake by other cell types. Proof of that concept was provided by Pittet and colleagues in a mice model bearing genetically modified B16F10 melanoma tumors. The authors observed that tumor-derived exosomes drained to the lymph node and bound to CD169<sup>+</sup> macrophages, which prevented the interaction with B cells that produce tumor promoting IgG. Elimination of CD169<sup>+</sup> macrophages by clodronate liposomes or by DT injection in the CD169-DTR mice promoted tumor growth. In the same study, melanoma-derived material was found in macrophages residing in the cancer-free sentinel lymph node of human biopsies, hinting to the potential relevance of these findings for human cancer research (140).

Several groups have reported association of the presence of CD169<sup>+</sup> macrophages in lymph nodes with good tumor prognosis in human. Ohnishi and colleagues showed a correlation between CD169<sup>+</sup> macrophages and CD8<sup>+</sup> T cell infiltration in colorectal cancer, improving overall survival rates. Furthermore, they observed co-localization of CD8<sup>+</sup> T cells and CD169<sup>+</sup> macrophages in regional lymph node section stainings (141). Similarly, more recent work from the same group demonstrated that the presence of CD169<sup>+</sup> macrophages in the lymph nodes was also correlated to CD8<sup>+</sup> T cell infiltration in malignant melanoma, endometrial carcinoma (where higher numbers of NK cells were also found), breast cancer and bladder cancer (142–145), all leading to a better prognosis and increased survival rates. Quite remarkably, in the study in malignant melanoma, IFN-α producing cells were detected around CD169<sup>+</sup> macrophages in the lymph node sinus area. Based on their morphology and marker expression, the authors hypothesized that the source of IFN-α, supporting the action of CD169<sup>+</sup> macrophages, could be CD68<sup>+</sup> macrophages and pDCs (142). Altogether, these studies present robust data illustrating the importance of CD169<sup>+</sup> macrophages in lymph nodes in proficient anti-tumor responses, characterized by a consistent CD8<sup>+</sup> T cell infiltration that benefits patient prognosis and survival. However, while CD169<sup>+</sup> macrophages where shown to co-localize with CD8<sup>+</sup> T cells, no direct evidence of antigen presentation by CD169<sup>+</sup> macrophages was provided at a functional level. Therefore, there might be room for other more specialized immune cells, such as cDC1 to cooperate in the process of T cell priming.

# Vaccination Strategies That Target to CD169<sup>+</sup> Macrophages

The presence of CD169<sup>+</sup> macrophages in lymph nodes draining different tumor types and their correlation with a better patient survival, their unique capacity to screen the lymphatic and blood circulation and, finally, their capacity to collaborate with DCs, all point to CD169<sup>+</sup> macrophages as appealing targets for the design of anti-cancer vaccines. Until now, several vaccination strategies targeting CD169<sup>+</sup> macrophages have been evaluated experimentally.

Due to their high specificity and the restricted expression pattern of CD169, monoclonal antibodies have been tested for antigen delivery to CD169<sup>+</sup> macrophages. Upon anti-CD169 specific antibody targeting of OVA, strong CTL responses were

particles are transferred via CD169, other components of bacteria and viruses can be transferred to cDC1s from the macrophages. Dead cells can stimulate cDC1s via CLEC9A expressed on the cDC1. The interaction between CD169<sup>+</sup> macrophages and cDC1s is dependent on binding of CD169 to sialic acid structures on cDC1s. (3) IFN-I priming: after encounter with bacteria, dead cells, or viruses, CD169<sup>+</sup> macrophages secrete IFN-I that is required for optimal activation of cDC1s and T cells. Subsequently, pDCs are recruited and their IFN-I production further amplifies the signal. (4) Trans-infection: in the case of HIV and MLV, CD169<sup>+</sup> macrophages can also mediate viral trans-infection to CD4 T cells and B cells.

generated mediated by antigen transfer to cDC1 (106). This effect was lost upon depletion of CD169<sup>+</sup> macrophages by the administration of clodronate liposomes and was shown to be mediated by BATF3-dependent cDC1s (44). Antibody-mediated targeting of OVA to CD169<sup>+</sup> macrophages also led to an isotypeswitched and high affinity antibody production due to germinal center activity. CD169<sup>+</sup> macrophages retained intact antigen on their surface for days and upregulated costimulatory molecules for B cell interaction upon activation (103). This feature of CD169<sup>+</sup> macrophages to retain intact molecules on their membrane has been correlated with low expression of proteolytic enzymes (104).

On a different note, Delputte et al. demonstrated that monoclonal antibodies against CD169 were not only binding, but also being efficiently internalized in a clathrin-dependent manner. Immunotoxins or antigens could be delivered to CD169<sup>+</sup> macrophages via antibody targeting, leading to killing of primary porcine macrophages and the generation of anti-HSA humoral responses, respectively (146, 147). It is not clear why certain studies report internalization and others longterm presence on the cell surface with antibody targeting. Both processes could occur simultaneously, but these divergent results could also be due to antibodies binding to different regions of the CD169 molecule.

In addition, liposomes have been used to target antigens to CD169<sup>+</sup> macrophages. Chen and colleagues generated OVAcontaining liposomes decorated with 3′ -BPCNeuAc, a synthetic ligand of CD169, and showed that targeting of IFN-α stimulated bone marrow-derived mouse macrophages with 3′ -BPCNeuAcliposomes induced OVA specific T cell proliferation (148). Moreover, the same authors could also accomplish activation of iNKT cells by including the lipid antigen αGalCer in the 3′ - BPCNeuAc-liposomes (149). CD169<sup>+</sup> macrophages seem well equipped in stimulating NKT cells via CD1d, which subsequently help B cell responses (42, 150). Liposomes with the endogenous ligand for CD169, ganglioside GM3, have also been shown to bind to CD169<sup>+</sup> monocyte-derived DCs (151). These studies indicate that also liposomal strategies could be employed to target antigens and activating agents to CD169<sup>+</sup> macrophages.

#### CONCLUDING REMARKS

In recent years a considerable number of studies have focused on the role that CD169<sup>+</sup> macrophages play in the SCS of the lymph node and the marginal zone of the spleen (summarized in **Figure 1**). These studies, as discussed in this review, point to CD169<sup>+</sup> macrophages as the main cell type to capture viruses, bacteria, dead cells and exosomes from the lymph fluid and the blood. This filtering capacity prevents further dissemination and enables a localized contained production of antigen that is efficiently transferred to DCs and B cells for the activation of adaptive immune responses. The collaboration between CD169<sup>+</sup> macrophages and cDC1s is especially important in the activation of CD8<sup>+</sup> T cell responses toward viral or tumor antigens. In this context, IFN-I derived from CD169<sup>+</sup> macrophages and pDCs plays a crucial role for an appropriate cDC1 activation. However, a number of pathogens have exploited this pathway and utilize CD169<sup>+</sup> macrophages as a niche to replicate and to mediate trans-infection of other cell types. In the coming years, the role of the human equivalent of this cell type will hopefully be elucidated and the development of treatment strategies to boost or down-regulate immune responses via the actions of the CD169<sup>+</sup> macrophages may well be expected.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### REFERENCES


#### FUNDING

This work was supported by grants from the Dutch Cancer Society (VU2013-5940 and VU2016-10449) a grant from the VUmc CCA (grant 2015-5-22) to JdH and a EU research framework programme grant (H2020-MSCA-ITN-2014-ETN-642870) to DC4U/MLV.

#### ACKNOWLEDGMENTS

The authors thank D. van Dinther for critical review of the manuscript.


vivo depletion of dendritic cells. Clin Exp Immunol. (2005) 141:398–404. doi: 10.1111/j.1365-2249.2005.02868.x


modulating antigen survival, endocytic routing, and processing. Blood (2012) 119:1407–17. doi: 10.1182/blood-2011-06-363564


macrophages impairs B cell responses to secondary infection. Science (2015) 347:667–72. doi: 10.1126/science.aaa1300


macrophages serves as a reservoir for septicaemia. Nat Microbiol. (2018) 3:600–610. doi: 10.1038/s41564-018-0147-1


nanoparticles targeting sialoadhesin/CD169. PLoS ONE (2012) 7:e39039. doi: 10.1371/journal.pone.0039039


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Grabowska, Lopez-Venegas, Affandi and den Haan. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# ATP and Its Metabolite Adenosine as Regulators of Dendritic Cell Activity

#### Cinthia Silva-Vilches, Sabine Ring and Karsten Mahnke\*

Department of Dermatology, Ruprecht-Karls-University Heidelberg, University Hospital, Heidelberg, Germany

Adenosine (Ado) is a well-studied neurotransmitter, but it also exerts profound immune regulatory functions. Ado can (i) actively be released by various cells into the tissue environment and can (ii) be produced through the degradation of extracellular ATP by the concerted action of CD39 and CD73. In this sequence of events, the ectoenzyme CD39 degrades ATP into ADP and AMP, respectively, and CD73 catalyzes the last step leading to the production of Ado. Extracellular ATP acts as a "danger" signal and stimulates immune responses, i.e. by inflammasome activation. Its degradation product Ado on the other hand acts rather anti-inflammatory, as it down regulates functions of dendritic cells (DCs) and dampens T cell activation and cytokine secretion. Thus, the balance of proinflammatory ATP and anti-inflammatory Ado that is regulated by CD39+/CD73<sup>+</sup> immune cells, is important for decision making on whether tolerance or immunity ensues. DCs express both ectoenzymes, enabling them to produce Ado from extracellular ATP by activity of CD73 and CD39 and thus allow dampening of the proinflammatory activity of adjacent leukocytes in the tissue. On the other hand, as most DCs express at least one out of four so far known Ado receptors (AdoR), DC derived Ado can also act back onto the DCs in an autocrine manner. This leads to suppression of DC functions that are normally involved in stimulating immune responses. Moreover, ATP and Ado production thereof acts as "find me" signal that guides cellular interactions of leukocytes during immune responses. In this review we will state the means by which Ado producing DCs are able to suppress immune responses and how extracellular Ado conditions DCs for their tolerizing properties.

#### Edited by:

Christian Muenz, Universität Zürich, Switzerland

#### Reviewed by:

Sven Burgdorf, Universität Bonn, Germany Amanda S. MacLeod, Duke University, United States

#### \*Correspondence:

Karsten Mahnke karsten.mahnke @med.uni-heidelberg.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 31 July 2018 Accepted: 19 October 2018 Published: 09 November 2018

#### Citation:

Silva-Vilches C, Ring S and Mahnke K (2018) ATP and Its Metabolite Adenosine as Regulators of Dendritic Cell Activity. Front. Immunol. 9:2581. doi: 10.3389/fimmu.2018.02581

Keywords: dendritic cells, adenosine, CD73, tolerance, ATP

#### ADENOSINE TRIPHOSPHATE (ATP) IN PERIPHERAL TISSUES

The chemical family of purines comprises of heterocyclic aromatic organic compounds, consisting of a pyrimidine ring fused to an imidazole ring. It comprehends biologically active molecules such as Adenosine-triphosphate (ATP) and its degradation product adenosine (Ado). ATP is widely known as an energy carrier within cells, but it can also be released from cells into the environment by cell membrane channels (gap junctions, pannexin channels) or specialized transporters (**Figure 1**) (1–4). Once located in the intercellular space, ATP transmits signals to other cells by engaging P2 receptors. P2 receptors can be divided into P2X and P2Y subtypes, which comprise different members as indicated by numbers, e.g., P2X<sup>1</sup> to P2X<sup>7</sup> and P2Y1, P2Y2, P2Y11. While all P2X receptors bind ATP, only the P2Y1, P2Y2, and P2Y<sup>11</sup> receptors are engaged by ATP. The mode of action of P2X and P2Y receptors differs also and can be described as ionotropic for P2X receptors, or metabotropic G-protein coupled in case of P2Y types. The P2X<sup>7</sup> receptor is a well-studied example and serves as prototypic ATP receptor in many investigations. P2X receptors often form multimeric complexes that upon engagement open a pore for cations such as Na+, Ca2+, or K<sup>+</sup> (5). This ion flux will then induce further intracellular signaling events. The most important pathway triggered by P2X receptors involves activation of the NLRP3 inflammasome, leading to caspase-1 activation, which in turn activates interleukin (IL-) 1β and IL-18, two important pro inflammatory cytokines. But this is only one well studied example. In particular the transmembrane flux of Ca2<sup>+</sup> ions can trigger multiple signaling events in cells involving mitogen activated kinases (MAPK), protein kinase C (PKC) and calmodulin. Therefore, many more effects of ATP induced signaling in leukocytes have been described These comprise the activation of T cells, (6–8), the release of IL-6, TNF (9, 10), prostaglandin (11), CXCL8, CCL2, CCL3 (12, 13) and metalloproteinase 9 (14), just to name a few [comprehensive list in Zimmermann. (15)]. The P2Y<sup>1</sup> receptor, which is binds ATP in rodents and the P2Y<sup>2</sup> receptor act via Gq coupled receptors and phospholipase C. Downstream, the second messengers inositol 1,4,5-triphosphate (IP3) that signals further via intracellular Ca2<sup>+</sup> levels and diacylglycerol (DAC), which activates PKC, are produced. This rather general activation scheme illustrates the diverse groups of effects that can be induced by P2Y receptor engagement. Indeed, involvement of P2Y receptors in regulating

hormone release and CNS activity has been documented in many instances. Beyond that, P2Y receptors are expressed by neutrophils, monocytes and T cells, indicating a role for immune regulation as well.

Due to the potent immune stimulatory actions of ATP, the extracellular concentrations are kept in check by enzymatic digestion of ATP. ATP is degraded fast within tissues, making it difficult to investigate its controlled release in defined organs in vivo. However, as skin is assessable for manipulation and measurement of ATP (16) and harbors several phenotypically distinct DC subtypes (17), it may be an organ of choice for investigating purine mediated signaling in vivo. At first, under non-inflammatory conditions the initial differentiation of skin keratinocytes (KCs) is guided by ATP. Upon binding of ATP the intracellular calcium levels rise gradually (as KCs express different subsets of ATP-specific P2X receptors depending on the layer), inducing the differentiation of the KCs (18, 19). Even the terminal differentiation and subsequent apoptosis of KC in the junction between stratum granulosum and stratum corneum seems to be dependent on ATP. Here, extensive colocalization of P2X<sup>7</sup> receptors with caspase-3 is evident (20), suggesting induction of cell death by ATP. This is corroborated by in vitro data, showing that prolonged engagement of P2X<sup>7</sup> receptors leads to extended pore-opening enabling even macromolecules of up to 900 Da to travel into cells, leading to induction of

caspase-dependent cell death (21). Beyond serving as messenger involved in skin differentiation, ATP has also clear functions as a danger molecule. Due to its function as activator of the NLRP3 inflammasome, ATP is involved in triggering skin allograft rejection. Here it has been shown that ATP is released by host cells in response to transplantation leading to IL-18 production and Th1 responses. Moreover, the skin may "use" ATP even to alert the peripheral immune system, as monocytes during acute rejection of transplants exhibited higher expression of P2X<sup>7</sup> receptors (22). Skin, as opposed to most other organs, is exposed to UV irradiation. This causes DNA damage, which produces a special set of danger signals. In response to UV irradiation, ATP is released by KCs triggering activation and release of IL-17 by dendritic epidermal γδ T cells (23). Once activated, γδ T cells can release ATP by themselves, leading to an autocrine activation loop maintained by P2X<sup>4</sup> receptors (24). Functionally this sustained production of IL-17 is of importance for limiting adverse effects of UV, as it upregulates genes necessary for DNAdamage repair, such as TNF-related weak inducer of apoptosis (TWEAK) and the growth arrest gene GADD45 (23). Therefore, in case of UV induced cancers, therapeutic enhancement of extracellular ATP may offer a way for treatment.

Also in chronically diseased skin the distribution of ATP and its receptors change. For instance, in psoriatic plaques P2X<sup>7</sup> receptors were found to be upregulated in the basal cell layer, suggesting that activation of KCs is facilitated by ATP (25). ATP is indeed elevated under pathological conditions, as it can be released by IFNγ activated and/or dying leukocytes and KCs (26, 27). Moreover, early results demonstrated defective hydrolysis of ATP in the psoriatic epidermis, leading to accumulation of extracellular ATP in the diseased skin, which supports the notion that ATP is profoundly involved in development of psoriasis (28). These early studies were recently confirmed by Killeen et al. (29), showing in the dermis of psoriatic lesions in a skin explant model elevated expression of P2X<sup>7</sup> receptors as compared to healthy skin. This increased P2X<sup>7</sup> signaling lead also to a phenotype of skin-DCs that predominantly induced Th17 cells, which are the main drivers of psoriasis. Finally, the elevated ATP concentrations in skin can also activate neutrophils, which in conjunction with IL-23, form a local inflammatory circuit maintaining psoriasiform dermatitis in mice (30). Therefore, increased levels of ATP together with enhanced expression of ATP receptors seem to be involved in maintaining an inflammatory environment in psoriatic skin.

On the other hand counter regulatory mechanisms directly related to the degradation product of ATP, i.e., Ado, have been described too. For instance, chronically stimulated epidermal KCs have an altered expression pattern of different Ado receptor (AdoR) types, with the rather pro-proliferative acting A2A receptor upregulated and reduced expression of the inhibitory A2B receptor (31). These and other observations led to investigations that utilize topical application of AdoR agonists for the treatment of psoriasis. Indeed, engagement of the AdoR A<sup>3</sup> leads to reduced production of IL-17 and IL-23 in KCs of psoriatic patients, inducing amelioration of the disease (32, 33). Therefore, several drugs acting as agonist for different types of AdoR are currently used in clinical trials of skin- and other inflammatory diseases (34, 35). But not only in inflammatory diseases ATP plays a role, it is also important for induction of acute inflammation in skin. Weber et al. have shown that skin DCs without functioning P2X<sup>7</sup> receptors are unable to sensitize T cell responses, indicating a role for directed ATP release as mediator of innate immune reactions (16). At the same time it became clear that haptens only act as trigger for hypersensitivity reactions when they induce release of ATP. Therefore, even experimental attempts were made to predict the "allergic potential" of chemicals by their ability to induce ATP release in KC cultures (36).

#### ATP as Substrate for Adenosine Production

A major degradation product of ATP is Ado, which can be generated intracellularly as well as extracellularly. Ado derives from the dephosphorylation of ATP, catalyzed by different enzymes: the ectonucleoside triphosphate diphosphohydrolase 1 (CD39) and the ecto-5'-nucleotidase (CD73) (37, 38). Both enzymes act sequentially in degrading extracellular ATP to adenosine. In a first step CD39 converts ATP to adenosinedi-phosphate and adenosine-mono-phosphate. In a second step the action of CD73 clips off the last remaining phosphate group, producing Ado (39). Ado can be released by nucleoside transporters from the cytoplasm of cells (4), however, the extracellular degradation of ATP by CD39 and CD73 is thought to provide the major pathway for regulating extracellular Ado concentrations. Its degradation is accomplished by adenosine deaminase (ADA), which exists in intra- as well as extracellular forms (40, 41). Extracellular ADA can bind to CD26 (42). Thus, similar to ATP and ADP, Ado can be degraded to inosine by cell membrane bound enzymes. In summary, the regulated destruction of extracellular ATP to Ado by enzymatic digestions offers cells a possibility to shape the tissue environment from a pro-inflammatory (high concentrations of free ATP) to a rather immunosuppressive (elevated levels of Ado) ambiance (43). As DCs express CD39 and/or CD73 as well as AdoR, they actively participate in immune responses affected by Ado (**Figure 1**).

# Regulation of Extracellular Ado and ATP Concentrations

In light of the opposing functions of the two mutually transformable signaling molecules ATP (activating) and Ado (suppressing) on immune reactions, their temporal/spatial distribution in tissues or along the plasma membranes of cells is of importance. Cells will presumably integrate activating (ATP) and suppressive (Ado) signaling pathways rendering a "final" outcome. Therefore, the half live as well as the diffusion speed through tissues is a critical factor determining the effects of ATP/Ado signaling. Real "in tissue" data of the distribution of extracellular ATP or Ado, respectively, are hardly available. However, contents in body fluids or organ cultures can be measured. For instance, in dog as well as human plasma Ado is only stable for a few seconds (44), making it a "short range" molecule. This rapid degradation may be useful to prevent a generalized immune suppression and it further prevents Ado from reaching the central nervous system, where it acts as neurotransmitter (45) and elevated levels may therefore disturb

nerve functions. Moreover, a short half-life makes Ado a more defined tool for cellular communication. Because only cells that harbor CD73 on their surfaces are able to produce sufficient amounts of Ado that then acts locally by engaging AdoR of adjacent cells. This mechanism may in particular of importance for tolerance induction, as Ado production by CD73 expressing DCs is required during the intimate DC:T cell priming process in order to render T cells tolerant (own unpublished results). Finally, to regulate Ado concentrations in relation to ATP not only the half-life is important, also the regulation of expression of the Ado producing ectoenzyme CD73 provides a means to fine tune the extracellular Ado content. During ischemic preconditioning expression of CD73 is induced within 30 min (46), greatly enhancing the extracellular Ado concentration in tissues and thereby overcoming the degradation by ADA.

For ATP biosensors are available (47) making is more feasible to monitor extracellular ATP content in cell culture settings. The reported half live of ATP varies from up 2–20 min depending on the organ and the methods used (48–51). Of note, in the immune system ATP actions are rather fast, as neutrophils show a burst of ATP release for only 5 s after being stimulated with fMLP (52). However, these data are once more obtained in in vitro culture systems, which differ from the in situ situation, but after all these data give an impression on the speed and range of ATP or Ado signaling. It provides evidence that Ado may not act "cytokinelike" with distribution via the blood stream and exerting action(s) in tissues far from its origin.

#### EFFECTS OF Ado ON DCs

#### Expression of Ado Receptors by DCs

Four Ado Receptors (AdoR) are known so far (A1, A2A, A2B, and A3). Structurally they all belong to G-protein-coupled-receptors (GPCRs), but their intracellular signaling differs (**Figure 2**). In general the A<sup>2</sup> receptor types are Gαs-protein coupled receptors, with the A2B receptor additionally signaling via Gαq. In cells an activated G protein complex forms at the inner leaflet of the cell membrane after Ado engagement, which leads to activation of the adenylate cyclase (AC) and to rising cAMP levels (in case of Gαs). As a consequence protein kinase A (PKA) is activated as secondary effector. On a molecular level this can directly be counteracted by engagement of A<sup>1</sup> or A<sup>3</sup> AdoR, which signal via Gαi/Gα<sup>q</sup> complexes. Among them, the Gαi/<sup>o</sup> complex inhibits AC activity and thus dampens A<sup>2</sup> mediated signaling. The main signal transduction of A1 and A3 receptors downstream of G proteins is mediated by phospholipase C induced secondary messengers that ultimately leads to increased Ca2<sup>+</sup> levels and PKC activation. Thus, a different secondary effector is induced by A<sup>1</sup> and A<sup>3</sup> AdoR, resulting in activation of different sets of genes. But nevertheless, even here a crosstalk with the A2B receptors is possible, as A2B AdoR via its coupling to Gα<sup>q</sup> can feed into the PLC mediated pathway and support A<sup>1</sup> and A<sup>3</sup> AdoR signaling (53, 54).

Many reports show expression of all four subtypes of AdoR by DCs in varying degrees (55, 56). However, the levels of expression and their distribution among defined subset of DC remain uncertain. When analysing the available data on AdoR expression by DCs at a glance it becomes clear that AdoR expression correlates with the maturation status of DC. Human immature DC express A<sup>1</sup> and A<sup>3</sup> AdoR, which after engagement activate and recruit DCs to inflammatory sites (57). Upon maturation A<sup>2</sup> AdoR emerge in DCs, now triggering rather inhibitory effects such as reduced secretion of IL-6, IL-12, and IFNγ (58). Here, differential expression of AdoR by DCs serves the purpose of regulating inflammatory processes. I.e., in the beginning of an insult, immature DCs are rendered active and are recruited to the inflammatory site whereas later A2-type AdoR expression limits over boarding inflammatory reactions. However, with several ways of cross talk between AdoR (as described above), differential expression by different cell types as well as varying affinities for purines, it is nearly impossible to assign one defined effect on cell physiology to the sole action of one AdoR or to one ligand in vivo. But in vitro studies can at least give insight into general pathways modulated by Ado.

#### Effects of Ado on Functions of DCs

Despite the fact that four different AdoR can activate different pathways at the same time that may have opposite effects on immune cell activation, many reports unequivocally demonstrate immune suppressive actions of Ado on DCs. In particular cAMP elevating AdoR A2A and A2B mediate rather inhibitory functions in DCs (53). For instance, after stimulation of respective AdoR in vitro, human DCs downmodulate secretion of IL-12 and TNFα. The cells expressed low amounts of MHC class II and were functionally impaired in stimulating proliferation of allogenic T cells. Further parameters of DC activation such as CXCL10, CCL2 and CCL12 secretion were also downregulated by Ado (56, 59–62). All of these features are indicators for a less mature phenotype of DC, which can be regarded as a tolerogenic type of DC (63).

In an even broader context a CD73<sup>+</sup> cellular environment may be important to keep DC in "steady state" condition. In vivo genetic ablation of CD73 in mice leads to enhanced inflammatory reactions in a contact hypersensitivity model that is driven by increased migration of skin DCs to peripheral lymph nodes (64). Moreover, when analyzing the expression of T cell costimulatory molecules by different DC subsets after application of the hapten TNCB, we found increased expression of CD86 in subsets of skin DCs in CD73 deficient as compared to control mice. These data are further corroborated by findings using stimulation or blockade of Ado deaminase (ADA), an enzyme that is crucial for degradation of extracellular Ado. ADA is expressed by DCs during ongoing inflammation to degrade CD73 derived Ado and to maintain their hyper-reactive state (65). In contrast, in absence of ADA Ado levels in cellular environments are increased, as a consequence tolerogenic functions of DCs are enhanced (60). Moreover, addition of ADA to DC:T cell cultures, which leads to depletion of Ado from the cellular environment, enhanced priming of effector T cells and suppressed induction of Treg (66). In aggregate, adequate levels of extracellular Ado in peripheral tissues may be of importance to prevent overshooting DC activity

AC and thus cAMP is suppressed by A1 and A3 AdoR engagement, which themselves signal via phospholipase C (PLC) and proteinkinase C (PKC). However, raising Ca2<sup>+</sup> levels, which transmit a P2X<sup>7</sup> derived signal are blocked by A1 AdoR. Finally, A2B AdoR can augment signals derived from A1 and A3 as it stimulates Ca2<sup>+</sup> mediated PKC activation also.

and to maintain their "steady state," which has been shown to be crucial for the tolerogenic function of DCs (67).

But beyond the mere prevention of DC maturation by Ado, the DC phenotype may be impacted in more fundamental ways. For example engagement of A<sup>2</sup> AdoR in DCs enables them to actively suppress immune reactions. The mechanisms include the stimulation of IL-10 secretion or the upregulation of T cell inhibitory molecules such as B7H1, resulting in tolerant T cells as their proper activation by DCs is impaired (62, 68, 69).

Even "imprinting" tolerogenic functions in DCs has been attributed to AdoR engagement. Li et al. (70) were able to attenuate acute kidney injury by infusing DCs pretreated ex vivo with A2A AdoR agonists. This phenotype of Ado tolerogenic DCs was stable for more than a week and its action in vivo relies on impeding NKT cell activation by a so far unknown mechanism. AdoR expression can also be intrinsically upregulated by already immunosuppressive DC subtypes to bolster their immunoregulatory functions. For instance, in a tolerogenic pediatric DC subtype, IL-10 is upregulated after Fc receptor mediated stimulation along with increased expression of the A2A AdoR, which after Engagement further augments their IL-10 production (71). Thus, A2A AdoR expression helps to reinforce the immunosuppressive capacity of the DCs.

# SIGNALING OF AdoR IN DCs

### The Molecular Mechanisms of cAMP in DCs

The main intracellular suppressive pathways triggered by A<sup>2</sup> AdoR types involve cAMP as a second messenger. Roughly, both A2-type AdoR elevate cAMP levels by activating Adenylyl Cylase (AC). Further downstream cAMP signals via PKA that regulates gene transcription via NF-κB, HIF-1α and CREB. In addition, A2B AdoR also acts on PLC, inducing raising intracellular Ca2<sup>+</sup> levels.

In a recent transcriptomic approach performed in bone marrow derived DCs (72), elevated activity of AC was connected to both, inhibition of AKT signaling and to activation of PKA (**Figure 3**). PKA has relevance for host defense capacities, as inhibition of Salt induced kinases (SIK) by cAMP-activated and PKA-mediated phosphorylation was shown to suppress secretion of the pro-inflammatory cytokines IL-6, IL-12, and TNFα by DCs and macrophages (73–75). Moreover, one of the SIK targets is the CREB-regulated transcription coactivator 3 (CRTC3) that can be phosphorylated at several serine residues. Phosphorylation of CRTC3 is inhibited by the cAMP-activated PKA, leading to translocation of the

concomitantly activate Phospholipase C (PLC), thereby triggering elevated levels of free Calcium (Ca2+). PKC mediated activation of NF-κB related gene expression follows. cAMP can suppress AKT activation leading to reduced mTORC activity either directly or via PRAS40. Without proper mTORC activity 4E-BP1 complex will terminate translation of proteins widely necessary for activation of cells. Additionally cAMP can signal via PKA, leading to hyper phosphorylation of Salt induced kinase (SIK) 2, allowing the phosphorylated transcription factor CRTC3 to cluster with CREB and to initiate translation of IL-10. Besides, further interaction of PKA with transcription factors such as NF-κB, HIF and CREB can induce "inhibitory" gene expression. A PKA independent pathway is mediated by EPAC, an enzyme that activates RAP via GTP binding, leading to profound changes in cytoskeleton and migration of DCs. As a result inhibitory DC:Treg clusters are formed and the immunological synapse may be changed in a tolerogenic fashion.

non-phosphorylated CRTC3 into the nucleus, where interaction with activated CREB upregulate IL-10 gene transcription (74).

In parallel to PKA activation, AKT activity is downregulated by elevated cAMP levels, promoting mTOR inhibition via PRAS40 (76). As a result, the downstream effectors of mTOR involved in the synthesis of cellular proteins, such as 4E-BP1 are hypophosphorylated. In this state, 4E-BP1 forms complexes with eukaryotic translation initiation factors and prevents translation (77). mTOR signaling regulation by AdoR driven cAMP content in DC may act as an important regulator of the antibacterial inflammatory response in monocytes, macrophages and primary dendritic cells (78, 79).

# Effects of AdoR Triggered cAMP Levels on Phenotype and Function

Despite that AdoR triggered cAMP elevation has multiple molecular targets the overall effect is obvious, as several reports show clear induction of an immunocompromised and tolerogenic phenotype of DC by cAMP. This is indicated by reduced secretion of proinflammatory cytokines, reduced expression of MHC class-II but elevated secretion of IL-10. Also the capacity of DCs to prime CD8<sup>+</sup> T cells in vitro was impaired in DCs with elevated intracellular levels of cAMP after induction by Ado or defined AdoR agonists such as 5′ -N-Ethylcarboxamidoadenosine (62, 68, 80– 83). In turn, cAMP can feed back on AdoR expression. For example, high levels of cAMP induced by agents that trigger Gs-protein coupled receptors, upregulates expression of A<sup>2</sup> AdoR in PC12 tumor cells (84). This cycle may therefore vigorously enhance Ado mediated suppressive effects in cells, as cAMP triggered upregulation of AdoR provides a means that leads to an even more sustained cAMP production.

To further delineate the possible cAMP effects that are mediated by AdoR engagement, one can artificially raise the cAMP content in DCs with Cholera toxin to mimic A<sup>2</sup> AdoR triggering. This leads yet to another subtype of tolerogenic DCs, i.e., DCs that express both isoforms of the tolerogenic molecule cytotoxic T lymphocyte antigen 2 (CTLA-2α and CTLA-2β) (85). These DCs resembled a semi-mature state and were able to promote TGFβ-dependent Foxp3<sup>+</sup> "induced" Treg conversion. Of note expression of CTLA-2 was critical for this function as genetic downregulation by siRNA reduced Treg conversion, while addition of recombinant CTLA-2α increased Treg conversion in vitro. Finally, when Lee et al. (81) investigated the role of DCs in priming of Th2 cells, they showed that deletion of genes that encode the GTP binding protein Gαs, leads to decreased cAMP signaling in DCs and provokes Th2 T cells with a prominent allergic phenotype. In contrast, increases in cAMP levels inhibited these responses. These findings imply that G protein-coupled receptors in DCs, such as A<sup>2</sup> AdoR, which are natural regulators of cAMP formation, can prevent Th2-mediated immunopathologies by rendering DCs unable to induce potent Th2 answers.

Another major pathway induced by rising cAMP levels, but independent from PKA, depends on the exchange protein EPAC. Upon cAMP mediated activation, EPAC catalyzes the GTP binding of RAP1, a major regulator of the cytoskeleton. Via this axis cAMP seems to affect cell motility, cell adhesion, chemotaxis and phagocytosis (86). For DCs in particular it has been shown that Ado released by Treg is responsible for attracting them (mediated by an EPAC-RAP dependent pathway), leading to formation of DC:Treg aggregates (87). In these aggregates DC undergo "tolerogenic instruction," as they start to produce IL-10, upregulate T cell inhibitory molecules and simultaneously downregulate expression of MHC class II molecules. Moreover, even the directed induction of DC:Treg clusters themselves may serve immunosuppressive functions, as Onishi et al. (88) have shown that Treg insolate effector T cells from proper activation by DCs by simply outcompeting them and keeping DCs in clusters.

### Priming of T Cells by DCs in Presence of Ado Is Altered

Despite the many well documented and long lasting effects of AdoR engagement on function of isolated DCs, the immediate presence of Ado during initial DC:T cell contact is crucially affecting the resulting immune response. For instance, in vitro engagement of A2A AdoR during the cognate MHC:peptide (as presented by DCs) T cell interaction leads to induction of T cell anergy and not to activation of T cells that normally ensues after DC:T cell interaction (89). This effect seems to be dependent on altered signaling in T cells, as reduced activation of the MAPK pathway was observed under these conditions. Ado:DC induced anergic T cells are not only refractory to restimulation, they also develop a CD25<sup>−</sup> LAG3<sup>+</sup> "regulatory" phenotype that actively prevents autoimmunity. Thus, the initial tolerogenic effects of Ado during antigen presentation by DCs will further

from cellular detritus, are present. ATP will be degraded immediately by CD39/CD73high Langerhans cells. (B) After injury high amounts of ATP are set free, which cannot be degraded effectively and overrule Ado production. This stimulates DCs via P2X7 receptors, leading to activation, to migration and to recruitment of different subsets of leukocytes. (C) When infection goes on, more CD39/CD73 expressing leukocytes are entering the tissue and P2X7 signaling ceases due to receptor desensitization and by counteraction of Ado. ATP is degraded to Ado which now prevails and terminates effector functions of various immune cells. Thus, immune homeostasis is reestablished.

be disseminated into tissues by these induced "regulatory" T cells.

As DCs can express CD73 themselves, production of extracellular Ado by DCs is conceivable and regulated expression of CD73 by DC subsets may one way to tune DC function for either tolerance (high CD73) or immunity (low CD73). Indeed, in a skin model for contact hypersensitivity application of the tolerogen 2,4-dinitrothiocyanobenzene (DNTB) rendered mice tolerant toward sensitization with the hapten 2,4 dinitrofluorobenzene (DNFB) (90). We found that induction of tolerance with DNTB was accompanied by increased expression of CD73 by skin migrating DCs and of note, in CD73 deficient animals tolerance induction by DNTB ceased (unpublished data). This underlines the importance of tissue derived Ado in governing DC functions under inflammatory conditions.

#### The Complex Regulation of ATP–Ado Signaling During Inflammation

As DCs can express all four AdoR, the ectonucleotidases CD39/CD73 as well as P2X<sup>7</sup> receptors, disentangling the ATP and Ado effects is very complex (91). It becomes even more complicated, as the different receptors transmit either stimulatory or suppressive signals, differ in their affinity for the respective ligands and are expressed to different degrees. The well investigated example of ATP induced chemotaxis of neutrophils gives an example how important the actual physical distribution of the different receptor molecules within a cell membrane is for their function. In neutrophils the chemotactic signal induced by fMLP is translated into ATP release by panx1. It will autocrinely act back on P2Y<sup>2</sup> receptors. At the same time stimulatory A<sup>3</sup> AdoR as well as CD39/CD73 are recruited to this part of the membrane, creating a local excitation circuit by activating PIP3, MAPK pathways and forming a "leading edge" for migration. A2A AdoR are excluded from this membrane site and are accumulating at the "trailing edge." At the same time Ado, produced at the "leading edge" by activity of CD39/CD73, diffuses to the "back" of the cell and engages A2A receptors. This signal is transmitted by means of cAMP–PKA activation and suppresses the activation of the cell locally. As a result neutrophils are polarized and find their way along chemotactic gradients (92–94). Altogether this was an elaborative effort of several research groups and similar investigation can be done for DCs too. Here we are just at the beginning, just investigating broad effects of ATP/Ado on DC migration and DC activation, without knowing how the individual pathways are interconnected at a molecular level.

Nevertheless, in a simplified scheme one can consider ATP as rather stimulatory and proinflammatory, and Ado (A2A and A2B receptors elevating cAMP) as being immune suppressive. In that sense CD39/CD73 expressing DCs are key cell for modulating homeostasis and inflammation and both receptor types (for ATP and Ado) are required to actually "measure" the degree of immune suppression or activation, respectively (**Figure 4**). Under non inflammatory conditions "steady state" DCs are patrolling through different tissues (95, 96) and sense only trace amounts of ATP, as tissues are intact and only limited amounts of extracellular ATP are produced, for instance by apoptotic cells. To maintain this homoeostatic status, high expression of CD39/CD73 ensures efficient degradation of ATP, preventing activation of the immune system. Examples are Langerhans cells in the epidermis that are highly positive for CD39 and degrade ATP effectively (37). Only when infection, tumor growth or trauma lead to elevated levels of extracellular ATP, the activating properties prevail, despite the fact that Ado receptors are expressed also. ATP simply outnumbers Ado effects. Subsets of immature peripheral DCs are recruited by ATP (58, 97) and an immune response is initiated. But counter regulatory mechanisms are initiated at the same time. For instance P2X<sup>7</sup> receptors become refractory to repeated stimulation by high ATP concentrations (37), making the DCs insensible to ATP mediated activation (58). Moreover, recruitment of regulatory T cells to inflammatory sites, which express high levels of CD39 and CD73, accelerates the degradation of ATP to Ado (98, 99). So the balance tips toward an Ado enriched ambiance that progressively exerts anti-inflammatory functions. More Ado means reduced proinflammatory functions of DCs (69, 70, 81, 100, 101), less migration of DCs from tissue to lymph nodes (64) and increased induction of regulatory T cells (60, 63, 87, 89). Thus, slowly immune homeostasis is reestablished.

#### CONCLUSION

The turnover of extracellular ATP to Ado by cell bound CD39 and CD73 offers a possibility to shape the tissue environment from an inflammatory (ATP high) to an immune suppressive habitat. DCs participate in this process as they (i) express ATP degrading enzymes CD39 and CD73 and (ii) harbor AdoR. Therefore, immunosuppressive effects of Ado can be mediated in two ways by DCs: First, DC derived Ado suppresses activation of T cells and fosters the induction of anergic and/or regulatory T cells during the cognate DC:T cell interactions. Secondly, Ado derived from adjacent cells act on DCs, preventing DC maturation and development of effector functions. These steady state DCs are considered tolerogenic. Thus, an Ado enriched tissue environment may be of importance to maintain the "steady state" of DCs to prevent autoimmunity.

# AUTHOR CONTRIBUTIONS

CS-V and SR contributed equally. SR prepared figures and CS-V performed experiments. All authors wrote the paper.

#### FUNDING

This work was supported by the DFG by a grant to the TR156, B03.

# REFERENCES


nucleoside diphosphokinase to extracellular nucleotide concentrations. J Biol Chem. (2000) 275:31061–8. doi: 10.1074/jbc.M003255200


adenosine: involvement of the A(2B) receptor. Eur J Immunol. (2008) 38:1610–20. doi: 10.1002/eji.200737781


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Silva-Vilches, Ring and Mahnke. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Importance of EMT Factor ZEB1 in cDC1 "MutuDC Line" Mediated Induction of Th1 Immune Response

Shuchi Smita1,2†, Abdul Ahad1,2†, Arup Ghosh1,3, Viplov K. Biswas 1,3, Marianna M. Koga<sup>4</sup> , Bhawna Gupta<sup>3</sup> , Hans Acha-Orbea<sup>4</sup> and Sunil K. Raghav 1,2,3 \*

1 Immuno-genomics and Systems Biology Laboratory, Institute of Life Sciences (ILS), Bhubaneswar, India, <sup>2</sup> Manipal Academy of Higher Education, Manipal, India, <sup>3</sup> Department of Biotechnology, Kalinga Institute of Industrial Technology (KIIT), Bhubaneswar, India, <sup>4</sup> Department of Biochemistry CIIL, University of Lausanne (UNIL), Epalinges, Switzerland

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Taiki Aoshi, Osaka University, Japan Elodie Segura, Institut Curie, France

#### \*Correspondence:

Sunil K. Raghav sunilraghav@ils.res.in; raghuvanshi2010@yahoo.co.in

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 25 July 2018 Accepted: 23 October 2018 Published: 13 November 2018

#### Citation:

Smita S, Ahad A, Ghosh A, Biswas VK, Koga MM, Gupta B, Acha-Orbea H and Raghav SK (2018) Importance of EMT Factor ZEB1 in cDC1 "MutuDC Line" Mediated Induction of Th1 Immune Response. Front. Immunol. 9:2604. doi: 10.3389/fimmu.2018.02604 The role of Epithelial to Mesenchymal Transition (EMT) factor Zeb1 is well defined in metastasis and cancer progression but it's importance in dendritic cells (DCs) is unexplored until now. For the first time we report here that Zeb1 controls immunogenic responses of CD8α <sup>+</sup> conventional Type-I (cDC1) DCs. We found that ZEB1 expression increases significantly after TLR9 stimulation and its depletion impairs activation, co-stimulation and secretion of important cytokines like IL-6, IL-10 and IL-12 in cDC1 MutuDC line. We further confirmed our findings in primary cDC1 DCs derived from bone marrow. Co-culture of these Zeb1 knock down (KD) DCs with OT-II CD4<sup>+</sup> T helper cells skewed their differentiation toward Th2 subtype. Moreover, adoptive transfer of activated Zeb1 KD DCs cleared intestinal worms in helminth infected mice by increasing Th2 responses in vivo. Integrative genomic analysis showed Zeb1 as an activator of immune response genes in cDC1 MutuDCs as compared to other pathway genes. In addition, differentially regulated genes in Zeb1 KD RNA-seq showed significant enrichment of Th2 activation pathways supporting our in vitro findings. Mechanistically, we showed that decreased IL-12 secreted by Zeb1 KD DCs is the plausible mechanism for increased Th2 differentiation. Collectively our data demonstrate that Zeb1 could be targeted in DCs to modulate T-cell mediated adaptive immune responses.

Keywords: ZEB1, cDC1 dendritic cells, integrative genomics, Th2 response, ChIP-seq, RNA-seq, helminth infection, immune modulation

# INTRODUCTION

Dendritic cells (DCs) are potent antigen presenting cells that play a pivotal role in developing immune responses as they govern both the initiation and polarization of adaptive immunity (1–5). The complex classification and nomenclature of DCs has now been refined into two levels, conventional or classical DCs comprising cDC1 (CD8α <sup>+</sup> and CD103+) and cDC2 (CD11b<sup>+</sup> and CD172a<sup>+</sup> DCs) depending on their distinct developmental pathways, and the plasmacytoid DCs (pDCs) (6–8). Genetic and functional studies have revealed that CD8α <sup>+</sup> (Lymphoid-resident DCs) and CD103<sup>+</sup> (Non-lymphoid tissue resident migratory DCs) DCs are specialized in antigen crosspresentation and polarization of Th cells into Th1 subset in response to stimulation via Toll Like receptor (TLR) ligands such as CpG for TLR9 and poly-IC for TLR3 (6, 7, 9–12). Upon pathogen encounter, DCs are activated leading to their maturation and migration toward secondary lymph nodes where they instruct T helper (Th) cell differentiation into different subtypes in a signal dependent manner (13–19). Among the many Th cell subsets, Th1 cells are critical for host defense against intracellular pathogens, while Th2 cells defend extracellular parasites (12, 15–17). Priming of Th cells toward Th1 requires inflammatory cytokine IL-12 whereas Th17 subtype depends on IL-6 and IL-23 cytokines produced by DCs (20, 21). In contrast, it has been widely accepted that DCs do not produce the Th2 speciation cytokines like IL-4 and IL-13. Therefore, decreased secretion of Th1 or Th17 promoting cytokines by DCs could induce the Th2 cell differentiation as a default outcome (22–24). Besides, considerable evidences suggest that DCs are required for optimal Th2 cell priming in vivo and expression of co-stimulatory molecules like OX40L or the Notch ligand Jagged-1 by DCs promotes Th2 cell priming (25, 26). On the other hand, it is explicitly known that cDC1 are prone to induce Th1 responses whereas cDC2 cells provide cooperative signal for Th2 responses where the IL-4 cytokine remains the key-determining factor for their polarization (27– 29). Interestingly, there are several reports showing upregulation of Th2 transcription factor GATA3 through IL-4 by activating STAT5 and STAT6 transcription factors (TFs), but few of them indicate that GATA3 expression can be independent of IL-4 as well (28, 30). Apart from signaling molecules, it has been reported that IRF4 depleted DCs are unable to induce Th2 differentiation (28, 31, 32), whereas increased KLF2 in DCs negatively regulates Th2 induction (33).

E-Box motif binding TF Zeb1 is a member of Zinc finger TF family, a known EMT master regulator. TGFβ signaling is one of the main mechanisms promoting EMT and is known to induce Zeb1 through SMAD signaling which in turn is well documented to repress E-cadherin (Cdh1) expression in epithelial cells (34, 35). The mir200 family members are predominantly present in epithelial cells and fine-tune the transcript expression of Zeb1 through feedback regulation (34, 36). In breast cancer cells, knock down of Zeb1 inhibits pro-inflammatory cytokines including IL-6 and IL-8 (37). Similarly, it has been widely reported that EMT in tumors is positively induced by inflammation (36, 38– 41). In contrast, Zeb1 has been reported to repress IL-2 by recruiting CTBP2 at its proximal promoter in T-cells irrespective of activation (42). There are reports suggesting higher expression of Zeb1 in migratory Langerhans cells, pertinent for their migration to secondary lymph nodes to present antigens to Th cells (43). This indicated that Zeb1 might be playing an important role in cDC1 axis of immune biology beyond just migratory properties. A forward genetic screen also revealed Zeb1 requirement for marginal zone of peritoneal B-1 B-cell development, T-cell development, germinal center formation, and memory B-cell responses (44). Though Zeb1 has been widely studied in cancer biology, few evidences with immunity and inflammation make it a potential candidate to look upon for its role in cDCs trajectory.

Here in this study, we investigated the role of Zeb1 in CD8α <sup>+</sup> cDC1 DCs and found it to be pertinent for their activation, co-stimulation and secretion of important immune response cytokines like IL-10 and IL-12. As a result, Zeb1 depleted DCs generated a strong Th2 phenotype ex vivo and in vivo, independent of IL-4 cytokine. Integrative genomic analysis demonstrated that Zeb1 has an indirect control on important cDC1 response cytokines and it does not act as a global repressor of immune response genes in DCs.

#### METHODS

#### Dendritic Cell (DC) Culture

Here in this study we have used CD8α <sup>+</sup> cDC1 MutuDC line recently developed by Prof. Hans Acha-Orbea's group. They have extensively characterized and compared these DC lines with primary CD8α <sup>+</sup> cDC1 DCs and reported that they perfectly mimic ex vivo immature CD8α <sup>+</sup> DCs isolated from spleen of C57BL/6 mice (9). The DCs were grown in IMDMglutamax (GIBCO) buffered with NaHCO<sup>3</sup> and supplemented with 8–10% heat inactivated FCS (tested for endotoxin toxicity toward DC cultures), 10 mM HEPES (GIBCO 15630), 50µM β-Mercaptoethanol (GIBCO 31350), and 50 U/mL of penicillin and 50µg/mL streptomycin (GIBCO 15070). The cells were maintained at 37◦C in a humidified incubator with 5% CO2. These DCs were dissociated with short incubation in nonenzymatic, 5 mM EDTA-based cell dissociation buffer (5 mM EDTA in 20 mM HEPES-PBS) at 37◦C.

For in vitro experiments, the DCs were plated in 6-well plates at a density of 5 × 10<sup>5</sup> cells/ml overnight. The cells were then challenged with different activation media containing TLR9 agonist CpG-B (Invivogen, cat no. tlrl-1826), TLR3 agonist pIC (Invivogen, cat no. tlrl-pic) and CpG+pIC for 2, 6, and 12 h. For performing RT-qPCR analysis the cells were washed in the plate once with PBS followed by addition of RNA-later (LBP) lysis buffer (Macherey-Nagel) for lysis of cells. The plates were then stored at −80 ◦ C until further RNA isolation and processing of samples.

#### Generation of Stable Zeb1 KD CD8α + MutuDCs

For generating stable Zeb1 knockdown and corresponding control DCs, lentiviral vector pLKO.1 (Sigma) containing three different Zeb1-specific shRNAs or control shRNA were used. Viral particles packaged with shRNA expressing transfer plasmids were produced in 293T cells using Cal-Phos (CaPO4) mammalian transfection kit (Clontech) according to an optimized protocol (45). 293T cells were transfected with transfer plasmids containing three different Zeb1 shRNAs or control shRNAs along with packaging plasmids (pCMVR8.74 and pMD2G). After 12–14 h the culture medium was replenished and supernatant containing viral particles were collected after 24 h in 50 ml conical tubes. Viral particle-containing culture supernatant was filtered through 0.45µm syringe filters (PES filters) and preserved at −80◦C in small aliquots. For transduction of shRNA containing viruses in CD8α <sup>+</sup> cDC1 MutuDC lines, the cells were plated at a density of 1.5 × 10<sup>5</sup> cells/well of 12 well plate followed by transduction with virus particles containing supernatant. The media was replaced with fresh media after 12 h of virus incubation with DCs followed by addition of 1µg/ml puromycin selection medium after 72 h of media replacement. The cells were puromycin selected for 2–3 weeks to get stable Zeb1 KD cells. The cells were also transduced with control shRNA-containing viruses to develop control cells for analysis comparisons. Efficiency of Zeb1 KD was quantified using Zeb1 gene specific primers by RT-qPCR (**Supplementary Table 8**). The shRNA that showed significant and maximum decrease in Zeb1 gene transcript levels compared to control transduced cells were used for further detailed study.

#### RNA Isolation and RT-qPCR

The cells preserved in LBP lysis buffer for RT-qPCR experiments were first taken out from −80◦C and thawed by placing the plates/tubes on ice. Total RNA was isolated using NucleoSpin RNA Plus kit (Machery-Nagel) according to manufacturer's protocol. RNA concentration was estimated by nanodrop (Thermo) and then 1 µg of total RNA was used to prepare cDNA using High capacity cDNA Reverse Transcriptase kit (Applied biosystems). Quantiative PCR was performed using SYBR Green master mix (Roche) and PCR amplification was monitored in real-time using LightCycler-480 Instrument (Roche). Primer oligonucleotides for qPCR were designed using universal probe library assay design system (Roche) and the primer pairs used are listed in **Supplementary Table 8**. Primers were optimized for linear and single product amplification by performing standard curve assays.

# Flow Cytometry (FACS)

Flow cytometric analyses of in vitro and ex vivo cultured cells were performed using well established protocol for FACS staining and analysis. For surface and intracellular (IC) staining 5 ∗ 10<sup>5</sup> and 1.5<sup>∗</sup> 10<sup>6</sup> cells were seeded respectively and stimulated with CpG, pIC and CpG + pIC for 12 h. After dissociation from plates, the cultured cells were washed with FACS buffer (3% FCS in PBS, 5 mM EDTA) followed by re-suspension in surface staining buffer. After washing, fluorochrome conjugated antibodies for proteins of interest were added to the cells as a cocktail (**Supplementary Table 8**). For intracellular (IC) staining of cytokines the cells were first fixed with 2% paraformaldehyde followed by permeabilisation using 1x permeabilisation buffer (eBiosciences). The fixed cells were then resuspended in intracellular staining buffer and stained with fluorochrome tagged antibodies for selected cytokines. For optimal staining the cells were incubated with antibodies for 30 min in dark at 4◦C. After incubation the cells were washed twice with FACS wash buffer and then acquired for differential expression analysis using LSRII fortessa flow cytometer (BD Biosciences). The acquired data was analyzed using FlowJo-X software (Treestar).

We used single color antibody stained cells as controls using pooled cells from all untreated and treated conditions for compensation and gating of positive population. In one of the biological replicates for immune profiling experiments, we used Florescence Minus One controls (FMO) to gate cells and for compensation. We found that single color stained and unstained negative control cells were giving similar results and therefore we didn't include FMOs in all our further experiments. From live cell population (high GFP positive cells) first we removed doublet cell population and then similar gates were employed for both control and Zeb1 KD DCs to observe any percentage cell population differences in surface markers and intracellular cytokines. Unstimulated DCs do not secrete cytokines therefore we used these cells stained with similar cocktail of antibodies for gating the cytokine positive cell population. In addition, we also analyzed for Median Florescence Intensity (MFI) shifts for each population in replicates to observe overall activation/costimulation markers and cytokines in control and Zeb1 depleted DCs.

### Bio-Plex Assay for Cytokine Quantitation From Cell Culture Supernatants

Bio-Plex assay (multiplex ELISA) was used to estimate the cytokine levels secreted in the cell culture supernatants of Zeb1 KD and control DCs after 12 h of CpG stimulation. After culture, the supernatants were stored at −80◦C in small aliquots until analysis. Cytokine levels were estimated using 23-plex-mouse cytokine assay kit following the vendor recommended protocol (Biorad).

### Generation of Bone Marrow Derived DCs (BMDCs) for ex-vivo Studies

Six to eight-week-old female C57BL/6 mice were killed by cervical dislocation and disinfected using 75% ethanol. The tibias and femurs were removed under sterile conditions, then soaked in RPMI-1640 medium supplemented with 10% FBS. Both ends of the bone were cut off with scissors, and the needle of a 1 mL syringe was inserted into the bone cavity to rinse the bone marrow out of the cavity into a sterile culture dish with RPMI-1640 medium (46). The cell suspension in the dish was collected and centrifuged at 350 g for 5 min, and the supernatant was discarded. The cell pellet was suspended with 1X RBC lysis buffer (Tonbo: TNB-4300) to lyse the RBCs and incubated for 5–10 min on ice. Cell clumps were then passed through a 70µm strainer to obtain single cell suspensions. The lysed cells were washed once with RPMI-1640, counted and used for differentiation into DCs.

We followed a well-established protocol for differentiation of BMDCs with slight modifications (47). The cells, suspended in RPMI-1640 medium supplemented with 10% FBS, were distributed into 6-well plates at a density of 1 × 10<sup>6</sup> cell/ml/well. Subsequently, 1µl/ml of FLT3L containing sera (derived from Flt3L expressing mice, gift from Hans Acha-Orbea, UNIL, Lausanne, Switzerland) was added into the medium. The cells were cultured at 37◦C in an incubator containing 5% CO<sup>2</sup> and left untouched for 5 days. On day 5, the suspended and loosely attached cells were collected, washed and counted. The cells were plated into 24-well plate for lentiviral transduction using concentrated viruses at a density of 0.4<sup>∗</sup> 10<sup>6</sup> cells/well for each Zeb1 shRNA and Control shRNA. After 72 h the cells were stimulated with CpG for 12 h and then immune-profiling was done at protein level using flow cytometry for observed markers that were found to be differentially regulated in CD8α <sup>+</sup> cDC1 cells in vitro. The antibodies used for staining were same as used for in vitro experiments.

# Co-culture of DCs With CD4<sup>+</sup> T Cells for Assessing T-Cell Proliferation and Differentiation

DC-T cell co-culture experiments were performed as described before (9). Naïve CD4<sup>+</sup> T cells were purified from spleen of TCR-transgenic OT-II mice using CD4<sup>+</sup> T cell isolation kit (EasySepTM Mouse CD4+T cells isolation Kit, Stem Cell Technologies). Zeb1 KD and control CD8α <sup>+</sup> cDC1 DCs were seeded at a density of 10,000 cells/well in round bottom 96 well plates followed by pulsing with OVA peptide (aa 323-339) and CpG for 2 h. After 2 h, purified OT-II T cells were added at the density of 100,000 cells/well (1:10 ratio) (48). Then T-cell proliferation and differentiation into distinct Th subtypes Th1, Th2, Th17 and Tregs were analyzed by FACS. Proliferation was measured using an amine based dye (eFluor 670). The rate of T-cell proliferation was inversely proportional to the Median Fluorescence Intensity (MFI) measured in FACS after 72 h of co-culture. For Th cell differentiation profiling after 96 h, the cocultured T cells were re-stimulated with PMA (10 ng/mL) and Ionomycin (500 ng/mL) and followed by Brefeldin-A (10µg/mL) treatment for 5 h to block the intracellular cytokines from being secreted. After 5 h, fluorochrome conjugated antibodies specific to different T cell subtypes were used to profile T cells into Th1 (Tbet and IFNγ), Th2 (GATA3, IL-13), Tregs (CD25, FoxP3) and Th17 (IL-17) (49). For gating effector T cells we used CD44 as a marker (see **Supplementary Table 8** for details of antibodies).

To confirm the default Th2 program recombinant IL-12 (5 ng/ml) and anti-IL4 (5µg/ml) was used in OT-II coculture experiments to confirm any perturbation in Th subtype differentiation using similar antibodies for various subtypes.

#### Chromatin Immuno-Precipitation (ChIP) for Zeb1

The ChIP for Zeb1 was performed according to the methods optimized previously by Raghav and Meyer's lab (50, 51). For ChIP assays, 30<sup>∗</sup> 10<sup>6</sup> CD8a<sup>+</sup> cDC1 MutuDCs were seeded in 15 cm<sup>2</sup> plates and prepared for ChIP by 10 min cross-linking with 1% formaldehyde (sigma) at room temperature followed by quenching using 2.5 M glycine (sigma) for 10 min. The plates were placed on ice and the cells were scraped and collected in 50 ml conical tubes. The cells were then washed three times using cold 1x PBS at 2,000 rpm for 10 min at 4◦C and the cell pellets were stored at −80◦C. At the day of the ChIP experiment, the cells were thawed on ice followed by lysis using nuclei extraction buffer (50 mM HEPES-NaOH pH 7.5, 140 mM NaCl, 1 mM EDTA pH 8.0, 10% glycerol, 0.5% NP-40, 0.25% Triton-X100) supplemented with protease and phosphatase inhibitors (Roche) for 10 min at 4◦C on rocker shaker. The prepared nuclei were then washed using protein extraction buffer (200 mMNaCl, 1 mM EDTA pH 8.0, 0.5 mM EGTA pH 8.0, 10 mM Tris-Cl pH 8.0) supplemented with a protease and phosphatase inhibitors (Roche) at room temperature for 10 min. Washed nuclei were resuspended in chromatin extraction buffer (1 mM EDTA pH 8.0, 0.5 mM EGTA pH 8.0, 10 mM Tris-Cl pH 8.0 and 1% TritonX-100) supplemented with protease and phosphatase inhibitors (Roche) and incubated for 20 min on ice for equilibration. The chromatin was fragmented using a Bioruptor (Diagenode) sonicator for 30 min using high amplitude and 30s ON & 30s OFF cycles to obtain 200-500 bp size fragments. A cooling unit was used to circulate the cold water during sonication to avoid de-crosslinking because of overheating. After sonication, chromatin length was checked in agarose gel. The fragmented chromatin was centrifuged at 10,000 rpm for 5 min and then clear supernatant was collected in 15 ml conical tubes. The DNA concentration of the chromatin was estimated using a Nano-Drop (Thermo) and the chromatin was diluted with ChIP dilution buffer (1 mM EDTA pH 8.0, 10 mM Tris-Cl pH 8.0 and 1% TritonX-100 containing protease and phosphatase inhibitors) to use 150µg/ml of chromatin for each IP. BSA and ssDNA (Salmon Sperm DNA) preblocked protein-A sepharose (80 µl/IP) beads were added to the samples on ice and incubated for 2 h to remove non-specific-binding chromatin. To the supernatant, 25 µl of rabbit polyclonal anti-Zeb1 (Santa Cruz H-102) were added to immunoprecipitate the chromatin complex at 4 ◦C overnight on rocker shaker. After the overnight incubation, 50 µl blocked beads were added to each sample and incubated for 2.5 h at 4◦C to pull down the respective antibody-chromatin complexes. The beads were then washed three times with low salt wash buffer (20 mM Tris-Cl pH 8.0, 150 mM NaCl, 2 mM EDTA pH 8.0, 0.1% SDS, 1% TritonX-100) followed by two washes with high salt wash buffer (20 mM Tris-Cl pH 8.0, 500 mM NaCl, 2 mM EDTA pH 8.0, 0.1% SDS, 1% TritonX- 100), lithium chloride wash buffer (10 mM Tris-Cl pH 8.0, 0.25 M LiCl, 1 mM EDTA pH 8.0, 1% NP-40, 1% sodium deoxycholate) and Tris-EDTA (TE) buffer (10 mM Tris-Cl pH 8.0, 1 mM EDTA pH 8.0). After removing the wash buffer completely, protein-bound chromatin complexes were eluted from beads for 30 min using elution buffer (100 mM NaHCO<sup>3</sup> and 1% SDS in milli-Q water). The eluted chromatin was then reverse crosslinked by incubating the eluted supernatant at 65◦C overnight on a heat block after adding 8 µl of 5 M NaCl. Next day DNA was purified from the reverse cross-linked chromatin by proteinase-K and RNase digestion followed by purification using PCR purification kit (Qiagen). The purified DNA was eluted in 40 µl of elution buffer.

#### ChIP-/RNA-seq Library Preparation for Next Generation Sequencing (NGS)

The RNA-seq library preparation was performed for Zeb1 KD and control cDC1 MutuDCs at 0, 6, and 12 h after CpG activation. As we did time kinetics we included two independent biological replicates to identify the Zeb1 depletion mediated global transcriptome changes. For RNA-seq library preparation 2 µg of total RNA was used to isolate mRNA through magnetic beads using mRNA isolation kit (PolyA mRNA isolation Module, NEB) followed by RNA-seq library preparation using mRNA library preparation kit (NEB) strictly following the vendor recommended protocol. After library preparation concentration of libraries were estimated using qubit 2.0 (Invitrogen) and the recommended fragmentation sizes were confirmed by Bioanalyzer (Agilent). For ChIP-seq library preparation, 30 µl of ChIP-DNA was processed for library preparation according to ChIP-seq library preparation recommended protocol (NEB). After library preparation and quality check using Bio-analyzer, the libraries were send to NGS service provider for Illumina sequencing using Hiseq-2500 instrument.

#### Western Blotting

Cells were collected in RIPA buffer (0.5 M EDTA, 1 M Tris-Cl pH7.5, 1 M NaCl, 200 mM PMSF, 10% NP-40, 10% SDS, 5% sodium deoxycholate, 1 M sodium orthovanadate and 1X Roche protease inhibitor) before and after CpG stimulation at different time points (0, 1, 2, 6, & 12 h). Cells were lysed completely by sonicating the samples in Bioruptor (Diagenode) for 10 min using high amplitude and 30s ON & 30s OFF cycles. Protein concentrations were measured in 96 well plate using BCA protein assay kit (BioRad).

### Adoptive Transfer of DCs in Helminth (Heligmosomoides polygyrus) Infection Mice Model

For DC Adoptive transfer experiments we took 6–8 week old female C57BL/6 mice and infected them with 200 infective L3 larvae/mice in PBS though oral gavage. Prof. Nicola Harris from EPFL (École Polytechnique Fédérale de Lausanne), Lausanne, Switzerland, provided the infective L3 larvae. The larvae were hatched from fecal charcoal cultures at day 7 after collection (52). After 7 days of infection, mice were treated with 100 µl of anti-CD8b antibody/mice followed by adoptive transfer of 10<sup>∗</sup> 10<sup>6</sup> CpG pulsed Zeb1 KD and control CD8α <sup>+</sup> cDC1 MutuDCs in sterile PBS intra-peritoneally (IP). Two booster doses of 5<sup>∗</sup> 10<sup>6</sup> cells pulsed with CpG were adoptively transferred consequtively after 48 h. After adoptive transfer of DCs the feces from infected animals were collected for worm eggs counting after every 24 h time period till day 31 using a well-optimized protocol. After observing a significant difference in egg count between Zeb1 KD DC treated and control animals, four mice from each group were sacrificed for detailed T cell profiling from mesenteric lymph nodes and the helminth worm counting from the intestine of the dissected animals. The intestines were longitudinally opened and flipped to count the worms and to take pictures. The T cell differentiation into Th1, Th2, Tregs, and Th17 subtypes was assessed using FACS as detailed above. This mouse experiment was performed following the institutional animal ethics guidelines after taking due approval from the institutional animal ethics committee at ILS, Bhubaneswar, India.

# Mixed Lymphocyte Reaction (MLR)

MLR setup was performed following well-documented protocol with slight modifications (49). The Zeb1 KD and control DCs were co-cultured with allogenic T cells from spleen of Balb/C mice. The Zeb1 KD and control CD8α <sup>+</sup> DCs were seeded at a density of 20,000 cells/well in a round bottom 96 well plates followed by challenge with CpG for 2 h. After 2 h, splenocytes containing mostly T cells from 6–8 week old Balb/C mice were added at the density of 200,000 cells/well (1:10 ratio). After 4 days (96 h) of co-culture, T-cell differentiation was assessed using same method as for OT-II co-culture.

### NGS ANALYSIS

### Quality Control & Preprocessing

The quality of RNA-seq and ChIP-seq reads from all sequencing experiments were determined using FastQC v0.11.5 (https:// www.bioinformatics.babraham.ac.uk/projects/fastqc/) tool. Reads having Phred Score(Q) <30 and over-represented sequences i.e. primer sequences, adapter sequences were removed using Trimmomatic (53) (http://www.usadellab.org/ cms/?page=trimmomatic) from both pairs. For binding events comparison, PU.1 (0 h) and Irf4 (0 and 2 h) ChIP-seq data of bone marrow-derived dendritic cell (BMDC) provided with LPS stimulation were accessed form GSE36104 (54).

#### RNA-seq Analysis

Filtered reads from transcriptome data were aligned with Tophat2 (55) using mm10 mouse genome assembly and Gencode M17 (GRCm38.p6) as reference transcript file. Rest of the alignment parameters remained same as Tophat2 is optimized for mammalian sequence alignment by default. Cufflinks v2.2.1 was used to calculate the abundance of transcripts from aligned files in terms of fragments per kilobase per million of reads (FPKM) and also generated the assembly file for differential expression using CuffDiff. All the assembly files from biological replicates were merged using Cuffmerge command from the same tool. Genes/transcripts form differential expression analysis having p < 0.05(significance) and q < 0.05 (false discovery rate) were classified as significant.

#### ChIP-seq Analysis

Alignment of ChIP-seq data was aligned with Bowtie2 (56) using mm10 mouse genome assembly as reference genome and—no-mixed,—no-discordant options to avoid unpaired read alignments. Aligned reads were deduplicated using Samtools (57) and randomly down-sampled to 22 million reads using Picard tool (https://broadinstitute.github.io/picard/). Peak calling was done using find Peaks tool form Homer suite (58) using a threshold of 4-fold change against input control and–factor option. Transcription factor motifs in −100/+100 region of the peaks were searched using findMotifsGenome.pl, and peaks were annotated to nearby gene using annotated Peaks.pl. Motifs obtained from both "De novo" and "Known Motifs" (having highest motif score) search having p < 10-20 were considered significant. Aligned reads from BMDC PU.1 and Irf4 ChIP-seq were compared with Zeb1 ChIP-seq data using SeqMiner (59) kmeans clustering. To validate the ChIP-seq results we randomly selected Zeb1 peaks found in ChIP-seq to confirm the enrichment using two independent ChIP-qPCR experiments.

#### Pathway Analysis

Ingenuity Pathway Analysis (IPA) software from Qiagen was used throughout the analysis process and p-value cutoff of 0.05 was considered significant. All the raw results of pathway analysis are attached as **Supplementary Files**.

#### Visualization

Gene expression scatter plot, pathway bar plots and other sequencing data representing bar graphs were generated using Ggplot2 (https://cran.r-project.org/web/packages/ggplot2/index. html) R package. The heatmap was generated using Complex Heatmap R package with row-wise clustering option (60).

#### RESULTS

#### EMT Factor Zeb1 KD Suppresses Activation, Co-stimulation and Cytokine Production in CD8α <sup>+</sup>cDC1 DCs

In this study, we have used a CD8α <sup>+</sup> cDC1 MutuDC line recently developed and well characterized by Fuertes Maracco and colleagues (9). They showed that this DC line mimic remarkably the primary lymphoid resident cDC1 responses isolated from spleen. While analyzing an unpublished RNA-seq data from these cDC1 cells, we found that Zeb1 transcript is constitutively expressed in cDC1 and upon TLR9 stimulation by CpG the expression increased significantly (**Supplementary Figures 1A,B**). To characterize the functional role of Zeb1 in cDC1, we generated a stable Zeb1 KD in cDC1 MutuDC line using lentiviral shRNA transduction followed by puromycin selection. We used three different shRNAs targeting different regions of Zeb1 transcript and found that two of them showed significant depletion of Zeb1 transcript (**Supplementary Table 8**). For downstream analyses we moved ahead with one shRNA sequence i.e., shRNA3, which showed higher Zeb1 depletion. We obtained 55–80% reduction of Zeb1 transcript in stable KD DCs and a concomitant decreased ZEB1 protein levels before and after 0.5, 2, 6, and 12 h after CpG activation as evident from western blotting analysis (**Figures 1A**). The western blot analysis demonstrated low levels of ZEB1 expression in unstimulated cells and a prominently increased expression within first few hours of CpG challenge that was decreased at later time points i.e., 6 and 12 h (**Figures 1A,B**).

After confirming the Zeb1 KD at transcript and protein level in stable KD DCs, we analyzed the impact of its depletion on DC activation and concomitant cytokine expression in CD8α + cDC1 MutuDCs. We performed detailed immune profiling of these stable Zeb1 KD DCs before and after 2, 6, and 12 h of CpG stimulation. It has been well established that Zeb1 directly represses Cdh1 gene and therefore first we analyzed the transcript expression of Cdh1 gene and found it to be significantly increased after Zeb1 depletion in cDC1 DCs confirming the impact of Zeb1 depletion (**Figure 1B**). Then we investigated the impact of Zeb1 KD on DC activation and co-stimulation along with expression of cytokine genes 12 h after CpG challenge using qPCR, flow cytometry (FACS) and multiplex ELISA (Bioplex) to profile the Zeb1 mediated immune-modulations. We found that Zeb1 KD DCs showed significantly reduced transcript expression of important DC response cytokines Il-10 and Il-27, while Il-12 (subunit p40) showed significant and sustained increase after 12 h of CpG stimulation in Zeb1 KD DCs (**Figure 1B**). Moreover, the Il-12p35 subunit of IL-12 cytokine showed significantly decreased expression after Zeb1 KD (**Figure 1B**). We did not observe any significant change in the mRNA expression of Cd80 and Cd86 activation markers upon Zeb1 KD before and after CpG activation (**Figure 1B**). FACS analysis showed significantly decreased expression of CD80 and CD86 in Zeb1 KD unstimulated DCs as compared to control cells, but no significant differences were observed after CpG stimulation (**Figure 1C**). The MFI analysis showed a significant increase in CD86 in CpG condition, whereas percent positive cells showed an insignificant increasing trend as CpG activation makes nearly 99–100 percent cells positive for CD86 (**Figure 1C**). The CD40 expression was unchanged in Zeb1 KD as compared to control DCs, whereas MHC-I and MHC-II percent positive cells showed a significant decrease in unstimulated condition (**Figures 1C,D**). In addition, the MFI shifts depicted a significant MHC-II decrease before and after CpG activation with a decreasing trend for MHCI (**Figures 1C,D**). Consequently the intracellular levels of IL-10 and IL-27 cytokines were significantly decreased in Zeb1 KD DCs after CpG activation (**Figures 2A,B**). Moreover, the MFI analysis (bar-plots and histograms) also showed similar trends (**Figures 1C,D**, **2A**). The cytokine IL-6 showed insignificant but decreasing trend (**Figure 2A**). Besides, the IL-12p40 levels were significantly increased in 12 h CpG activated Zeb1 KD DCs. At early time points (2 and 6 h) of CpG activation, we did not observe any significant change in the IL-12p40 expression (data not shown). Furthermore to estimate the secreted cytokine levels, we performed multiplex ELISA i.e., Bioplex analysis, which demonstrated a significant decrease in IL-6, IL-10, IL-12p70, and IFNγ cytokine levels in CpG activated Zeb1 KD DCs (**Figure 2B** and **Supplementary Figure 1F**). We observed significantly decreased levels of IL-12p70 in the culture supernatants of 12 h CpG activated Zeb1 KD DCs in contrast to a significant increase we found for IL-12p40, an important subunit of this cytokine (**Figure 2B**). This may be due to significantly decreased IL-12p35 subunit, which is an exclusive subunit of bioactive inflammatory cytokine IL-12. Other important DC markers like PDL1 and IL-27 also showed a significant decrease after Zeb1 depletion as compared to control DCs (**Figures 1C**, **2A**). These results showed that Zeb1 depletion suppressed DC activation and co-stimulation leading to decreased secretion of important DC response cytokines.

The cDC1 MutuDCs that we employed in our study express high levels of both TLR3 and TLR9 receptors (9). Therefore, we also treated Zeb1 KD and control DCs with TLR3 ligand pIC and CpG + pIC simultaneously to activate both TLR3 and TLR9 receptors together. We found that pIC resulted in weak activation of DCs as compared to CpG whereas simultaneous CpG + pIC activation resulted in strong activation as evident from CD80 and CD86 expression and synergistic expression of cytokines like IL-6, IL-10, and IL-12p40 (**Supplementary Figures 1C,D**). We found that Zeb1 depletion resulted in similar decrease in activation, co-stimulation and cytokine genes as we found after CpG activation. These results further confirmed that Zeb1 KD results in suboptimal activation of DCs leading to decreased expression of pro- and anti-inflammatory cytokines in cDC1 DCs, irrespective of any strong antigenic challenge.

To validate our in vitro findings, we generated primary cDC1 DCs from the bone marrow precursor cells using fmslike tyrosine kinase 3 ligand (FLT3L) supplemented medium (see Methods for details) (47). Bone marrow culture with FLT3L is the recent method that allow the generation of both cDCs

fold changes in the transcript expression of DC activation/co-stimulation markers and selected cytokine genes in Zeb1 KD DCs compared to control cells using

**150**

Frontiers in Immunology | www.frontiersin.org

(Continued)

FIGURE 1 | qPCR n = 3. (C) Scatter-plots showing the percentage positive cells for cell-surface expression of activation and co-stimulatory markers CD80, CD86, MHC-I, MHC-II, CD40, and PDL1 on Zeb1 KD cDC1 MutuDCs compared to control cells before and after 12 h CpG activation. The corresponding panels with bar plots depict the Median Florescence Intensity (MFI) shifts for the respective genes n = 6–12. (D) Representative histogram plots demonstrating the MFI shift observed for the activation/co-stimulatory markers and cytokines. p-values are calculated using two tailed unpaired student's t-test, error bars represent SEM. \*≤0.05, \*\*≤0.01, \*\*\*≤0.001.

and pDCs. In purview of published reports, which suggests that FLT3L derived DCs (FL-DCs) are CD11chiCD24hi and CD11blow, are putative of CD8α <sup>+</sup> cDC1 equivalent (47, 61), therefore we gated CD11c+CD24+MHCII<sup>+</sup> population in FACS to analyze the impact of Zeb1 transient depletion on cDC1 activation and cytokine expression. We found that transient KD of Zeb1 in primary cDC1 cells using Zeb1 shRNA reduced their activation, co-stimulation and production of cytokines like IL-6, IL-10, IL-12p40, and IL-27 as compared to control KD DCs (**Figures 3A–D** and **Supplementary Figure 2A**). In contrast to Zeb1 KD MutuDCs, IL-12p40 showed decreased expression in BMDCs. This could be due to transient Zeb1 KD, differential expression kinetics of IL-12p40 in BMDCs or due to cell heterogeneity of BMDCs. We also found that Zeb1 is expressed at similar levels in different primary DC subsets cDC1, cDC2, and pDCs isolated from spleen of FLT3L transgenic mice. Here we have focused on the role of Zeb1 in cDC1 DCs but it would be further intriguing to explore its importance in other DC subsets as well (**Supplementary Figure 2F**).

# OT-II T Helper Cells Co-cultured With Zeb1 KD DCs Enhanced Th2 Responses

As we found a consistent decrease in all the measured cytokines in Zeb1 KD DCs we were interested to identify if Zeb1 KD cDC1 MutuDCs would functionally interfere in T-helper (Th) cell differentiation. For the same, Zeb1 KD and control DCs were pulsed with OTII peptide with or without CpG for 2 h. Then, CD4<sup>+</sup> Th cells isolated from spleen of OT-II transgenic mice were labeled with a proliferation dye efluor 670 followed by coculture with these DCs for 72 h. Upon priming by DCs, the naïve T helper cells first undergo several rounds of clonal amplification and then polarization into various effector subtypes depending upon the DC responses (62, 63). We found that Zeb1 KD DCs induced higher antigen-specific CD4<sup>+</sup> Th cell proliferation compared to control DCs (**Figure 4A**). This was also observed when DCs had been previously stimulated with CpG, which induced even higher T cell proliferation after 3 days of coculture (**Figures 4A,B**). As we observed a decrease in activation and co-stimulation markers in Zeb1 KD DCs the increased proliferation of T cells was puzzling and therefore we looked into expression of IL-2 in cell culture supernatants of Zeb1 KD and control cells by Bio-plex. We found increased IL-2 cytokine after CpG activation in Zeb1 KD DC supernatants as compared to control cells (**Supplementary Figure 2E**). In addition, we looked into T cell differentiation profiles of these co-cultured Th cells. We found significantly increased differentiation of Zeb1 KD DCs primed Th cells toward Th2 subtype marked by increased number of GATA3 and GATA3+IL-13<sup>+</sup> expressing Th cells (**Figures 4C,D**). At the same time the Tbet+IFNγ <sup>+</sup> expressing Th1 cells were majorly decreased in CpG activated Zeb1 KD condition (**Supplementary Figures 2B,C**). We did not observe any significant difference in the Treg differentiation marker FoxP3 (**Supplementary Figure 2D**) In unstimulated conditions we did not found any significant increase in GATA3<sup>+</sup> alone or GATA3+IL-13<sup>+</sup> double positive Th cells (**Figures 4C,D**) and on the contrary the Tbet+IFNγ <sup>+</sup> population was increased (**Supplementary Figures 2B,C**)

Moreover we also performed allogeneic mixed lymphocytic reaction (MLR) assay to confirm the induction of Th2 responses by Zeb1 KD DCs (**Supplementary Figure 3A**). We found that the CD3+CD4+CD44<sup>+</sup> effector Th cells showed enhanced Th2 responses as evidenced by significantly increased GATA3 and IL-4 expression in T cells primed by CpG pulsed Zeb1 KD compared to control DCs (**Supplementary Figure 3B**). On the other hand, the Th1 subtype polarization marker Tbet and its signature cytokine IFNγ showed insignificant but decreasing trend (**Supplementary Figure 3C**). These analyses demonstrated that Zeb1 has the potential to modulate T helper cell differentiation toward Th2 subtype by modulating DC responses. We also looked into the impact of Zeb1 KD cDC1 on CD8<sup>+</sup> T cell function by gating them separately from CD4<sup>+</sup> Th cells (**Supplementary Figure 3D**). We found that the effector CD8<sup>+</sup> T cells generated in co-culture with Zeb1 KD DCs produced significantly less Granzyme and Perforin as compared to CD8<sup>+</sup> T cells cultured with control DCs (**Supplementary Figures 3D,E**) suggesting Zeb1 as important factor for inducing optimal T cell cytotoxic responses.

#### Adoptive Transfer of Zeb1 Depleted DCs Enhanced Helminth Clearance in Mice

After confirming that Zeb1 KD cDC1 MutuDCs enhanced Th2 cell development in vitro, we were interested in determining if these DCs were able to affect Th2 responses in an in vivo parasite infected animal model. Helminth (Heligmosomoides polygyrus) infection mice model is considered as one of the bestcharacterized disease models where it has been established that perturbation of Th cell subtype responses modulates the worm load in the intestine (64–67). Besides that the penetrance of pathogenesis is quite uniform in this disease model. Therefore, we performed adoptive transfer of CpG pulsed control and Zeb1 KD MutuDCs in H. polygyrus infected mice to identify its physiological impact on disease burden. The egg load in the feces were calculated starting from day 9 (D9) till 1 month after adoptive transfer of DCs to get an idea about the mature worm abundance in the intestine of infected animals. At D9, there were no eggs observed in both Zeb1 KD and control DC treated animals. This observation is consistent with the previous reports that worms get mature and move to intestinal lumen on the

FIGURE 3 | Zeb1 depletion in bone-marrow derived primary cDC1 DCs showed decreased DC activation, co-stimulation and cytokine secretion n = 6–8. (A) FACS contour-plot showing the gating strategy used to remove F4/80 positive macrophage population from the CD11c<sup>+</sup> DCs to analyze the impact of Zeb1 KD on CD11c+CD24+MHCII<sup>+</sup> DC population. High CD11c+CD24<sup>+</sup> double positive DCs were considered as cDC1 DCs for analysis. (B) Scatter-plot and representative contour plots depicting the percentage of MHCII positive cells in F4/80−CD11c+CD24<sup>+</sup> gated BMDCs treated with control and Zeb1 shRNA3 followed by 12 h CpG stimulation n = 6–8. (C) Scatter-plots and representative contour plots depicting the percentage positive cells for cell surface markers PDL1, CD86, and CD40 in F4/80−CD11c+CD24<sup>+</sup> gated BMDCs treated with control and Zeb1 shRNA3 followed by 12 h CpG stimulation n = 6–8. (D) Scatter-plots showing the percentage positive cells for intracellular cytokines IL-10, IL-12p40, IL-6, and IL-27 in F4/80−CD11c+CD24+MHCII<sup>+</sup> gated BMDCs treated with control and Zeb1 shRNA3 followed by 12 h CpG stimulation n = 6. p-values were calculated using two tailed unpaired student's t-test, error bars represent SEM. \*≤0.05, \*\*≤0.01, \*\*\*≤0.001.

cells for GATA3+IL-13<sup>+</sup> and single positive GATA3<sup>+</sup> cells in effector CD4+CD44<sup>+</sup> Th cell population n = 8. p-values are calculated using two tailed unpaired student's t-test, error bars represent SEM. \*≤0.05, \*\*\*≤0.001.

D10 and reproduce, resulting in release of eggs in feces (68). Interestingly at D10, we found that Zeb1 KD DC treated animals showed less egg counts in their feces compared to control DC treated animals even after injecting them with equal number of L3 stage larvae at D0 (**Figure 5A**). In addition, in Zeb1 KD DCs treated mice the intestines were almost clear of helminth infection as no or only few worms were found in the intestine depicted by the drastically reduced number of eggs on D31 (**Figure 5B**). Five animals from each group were dissected at D14 and D31 to estimate the worm load in the intestines. We found that animals treated with CpG activated Zeb1 KD DCs showed a significant decrease in intestinal helminth load as compared to control DC treated animals at D14 (**Figure 5C**). Besides, at D31 insignificant differences were observed in intestinal helminth count in animals treated with unstimulated Zeb1 KD and control DCs as clear from similar egg counts found at D31 (data not shown). However, the animals treated with CpG pulsed Zeb1 KD DCs were almost free of helminth infection in the intestine as compared to control DC treated mice at D31 (**Figure 5D**).

Furthermore to identify the impact of CD8α <sup>+</sup> cDC1 MutuDCs adoptive transfer on Th cell polarization in animals leading to the observed phenotype, we performed detailed Th cell subtype profiling from mesenteric lymph nodes (MLNs) of all the infected and treated animals at D14 and D31. We found that in CpG pulsed Zeb1 KD DC treated animals at D14 presented increased number of Th2 effector CD4+CD44<sup>+</sup> cells with significantly higher expression of GATA3 and IL-5 as compared to control DC treated animals (**Figure 5E**). The cytokine IL-13 also showed an insignificant but increasing trend (**Figure 5E**). We also observed mice that were treated with CpG pulsed Zeb1 depleted DCs had reduced Tbet and IFNγ positive cells in the MLNs. FoxP3, a Treg marker showed an increase in Zeb1 KD cells, which is well reported to be elevated during Th2 response (**Figure 5E**). At D31, we found a significant increase of IL-5 and IL-10. The cytokine IL-13 showed increasing trend, whereas IFNγ was significantly increased (**Supplementary Figure 4**), which could be the reason for increased Tregs. Increased Th2 cells in MLNs of helminth infected animals by treating animals with Zeb1 KD DCs strongly suggested that Zeb1 depletion in CD8α <sup>+</sup> cDC1 could potentiate Th2 responses in vivo, affecting helminth clearance.

#### Transcriptome Analysis of Zeb1 KD DCs Showed an Enrichment of Th2 Pathway

To understand mechanisms underlying the control of Zeb1 mediated DC responses we performed RNA-seq analysis of control and Zeb1 KD cDC1 MutuDCs at 0, 6, and 12 h after CpG stimulation. First we confirmed the significantly decreased transcript levels of Zeb1 and a concomitant significant increase in Cdh1 expression in RNA-seq datasets at all the analyzed time points (**Supplementary Table 1**). Downstream analysis and manual curation of the genes that were differentially expressed (corrected p-value ≤ 0.05) in Zeb1 KD DCs showed a significant down-regulation of cytokines like Il-6, Il-10, and Il-27 along with several C-type lectin receptors (CLRs) such as Clec1a, Clec4a1, Cleca7A (Dectin-1), Clec9a and Clec12a as compared to control cells (**Figures 6A,B** and **Supplementary Table 1**). It has been reported that CLRs present on DCs act in a pathogen/antigen dependent manner to control Th cell differentiation. Whereas one of the CLR i.e., DC-SIGN (Cd209c, Cd209f, Cd209g) showed significant and highest increase in 6 h as well as 12 h CpG activated Zeb1 KD DCs as compared to control cells (**Figures 6A–C** and **Supplementary Table 1**). We found that there were more upregulated genes after Zeb1 KD as compared to down-regulated ones at all the time points suggesting toward global repressive function of Zeb1 (**Figure 6D**). There were 229, 308, and 396 genes upregulated after Zeb1 KD at 0, 6 and 12 h time point, whereas 113, 212, and 234 genes were down regulated respectively (**Figure 6D** and **Supplementary Table 1**). Moreover, the major anti-inflammatory or tolerogenic cytokine Il-10 was also significantly decreased at both 6 and 12 h time points in Zeb1 depleted cells (**Figure 6C** and **Supplementary Table 2**). Ultimately to identify the biological pathways that were enriched for the genes differentially expressed in Zeb1 KD DCs after 6 and 12 h CpG activation, we performed Ingenuity pathway analysis (IPA). We found that at both the time points there was enrichment of "Dendritic cell maturation," "Receptors for bacterial/virus recognition," "Th2 pathways," "Th cell differentiation," and "STAT3 pathways" (**Figure 6E**, **Supplementary Figures 5A,B** and **Supplementary Table 3**).

#### Zeb1 Chip-seq Identified Its Direct and Indirect Target Genes in CD8α <sup>+</sup> cDC1 MutuDCs

We performed ChIP-seq for Zeb1 in unstimulated DCs to identify the genes that were directly bound and regulated by Zeb1. The major aim was to correlate the Zeb1 binding in ChIP-seq with RNA-seq to identify the genes that were directly controlled by Zeb1 in DCs. We found ∼3400 genomic regions bound by Zeb1 in unstimulated CD8α <sup>+</sup> cDC1 (**Supplementary Table 4**). Upon overlap of genes bound in ChIPseq with RNA-seq list we found 76, 130, and 141 genes to be directly regulated by Zeb1 (**Figure 7A**). Out of these genes 52, 85, and 95 were upregulated, whereas 24, 45 and 46 genes were down regulated at 0, 6, and 12 h respectively after Zeb1 KD (**Figure 7A**). To identify the Zeb1 bound genomic regions with respect to transcription start site (TSS), we did GREAT analysis and found that majority of Zeb1 bindings were distal (775 peaks in ±5 KB, 1576 peaks in > ±5 KB and < ±50 K region) and far from TSS (**Figure 7B**). Then, to identify if Zeb1 DNA binding motif is enriched on these bound peaks we did de novo motif analysis using HOMER and found 42% of bound regions showed canonical Zeb1 motif whereas other Zeb1 bound genomic regions showed significant enrichment of DNA motifs for IRF, ETS (PU.1), E2A and ETS-IRF (PU.1-IRF) TFs (**Figure 7C**). To experimentally validate it, we performed ChIPseq for PU.1, and the SeqMINER overlap with Zeb1 peaks showed that indeed PU.1 overlaps strongly at Zeb1 bound genomic regions (**Figure 7D**). We also overlapped publicly available IRF4 ChIP-seq data from unstimulated and 2 h LPS stimulated BMDCs with Zeb1 and found that IRF4 also showed similar percentage of overlap as predicted in our de novo motif analysis (**Figures 7C,D** and **Supplementary Table 5**). Moreover, Th1 and Th2 pathways are enriched for genes annotated to Zeb1-PU1- IRF4 overlapping genomic regions (**Supplementary Figure 5C**). Although Zeb1 is a known transcriptional repressor and there are more upregulated genes as compared to down regulated genes in RNA-seq data, we observed that most of the DC immune response genes were down regulated after Zeb1 depletion in our in vitro and ex vivo immune-profiling experiments. To understand it better, we divided all the Zeb1 KD differentially regulated genes into two groups based on gene ontology, i.e., immune response pathway and the other pathway genes and

(Continued)

FIGURE 5 | counting was started from D10 till D31. At day 9 there were no eggs detected in feces of animals n = 10. (B) Representative intestinal sections of helminth infected mice at D14 and D31 after helminth infection and treatment with CpG pulsed Zeb1 KD and control DC treated animals n = 5. (C) Intestinal worm load at day 14 in mice adoptively treated with CpG activated Zeb1 KD and control DCs n = 5. (D) Intestinal worm load in mice adoptively treated with CpG activated Zeb1 KD and control DCs at day 31 after infection n = 5. (E) Scatter-plots for Th1, Th2, and Treg markers like GATA3, IL-5, Tbet, IFNγ, IL-13, and FoxP3 from detailed immune profiling of CD4+CD44<sup>+</sup> effector Th cells isolated form mesenteric lymph nodes of helminth infected animals and treated with activated Zeb1 KD and control DCs. The corresponding panels also depict bar plots for MFI shifts for each marker in CD4+CD44<sup>+</sup> effector Th cell population n = 5. p-values are calculated using two tailed unpaired student's t-test, error bars represent SEM. \*≤0.05, \*\*≤0.01, \*\*\*≤0.001.

looked for any preferential Zeb1 bindings differences in these two groups. Surprisingly, the ratio of Zeb1 unbound vs. bound regulated genes was ≥2.5 for both the gene groups. Then, we analyzed the number of up- vs. down-regulated genes in immune response and other pathway group and interestingly we found that similar number of immune response genes were differentially regulated (ratio of ∼1.0 for up- vs. down genes) whereas for other pathways gene group there were more upregulated genes (ratio ≥ 1.6–2.5) (**Figures 7E,F** and **Supplementary Table 6**). This suggested that Zeb1 does not appear to be a global transcriptional repressor for immune response genes. Ultimately we analyzed the pathways enriched for the genes directly bound and regulated by Zeb1 using IPA. We found "T helper cell differentiation" and "Th2 pathway" to be significantly enriched for genes differentially regulated at 12 h time point (**Figure 7G** and **Supplementary Table 7**). It was interesting to find that even with such a low number of direct target genes we found T cell differentiation and Th2 pathway as highly enriched pathways.

#### Decreased IL-12 Cytokine Secretion by Zeb1 KD DCs Leads to Th2 Development

Though we found in our genomic analysis that CLRs were significantly down regulated along with IL-10, IL-6, and IL-12p70 cytokines, we were unable to pinpoint specific molecular mediator leading to the induction of Th2 responses by Zeb1 KD DCs. It has been reported extensively that decreased secretion of inflammatory IL-12p70 cytokine by DCs results in default polarization of T cells toward Th2 subtype (22, 28, 31, 69). Therefore we decided to confirm if Zeb1 KD DCs upon supplementation with recombinant IL-12 (rIL-12) containing medium decreased the polarization of Th cells toward Th2 with a concomitant increased Th1 subtype. We found that OT-II Th cells co-cultured with Zeb1 KD DCs supplemented with 5 ng/ml rIL-12 resulted into significantly increased IFNγ expressing Th1 cells comparable to control DCs (**Figure 8A** and **Supplementary Figure 6A**). On the contrary there was a significantly decreased expression of GATA3<sup>+</sup> Th2 cells in Zeb1 KD DCs after rIL-12 supplementation (**Figure 8B** and **Supplementary Figure 6B**). The MFI analysis of Tbet, Th1 inducing factor also demonstrated significant increase after rIL-12 addition in Zeb1 KD DCs whereas GATA3 showed insignificant (p = 0.09) but decreasing trend (**Figures 8C,D**). This lead us to conclude that decreased IL-12p70 secreted by activated Zeb1 KD DCs leads to default development of Th2 phenotype. As Il-12p35 is an exclusive subunit of bioactive IL-12 cytokine whereas Il-12p40 subunit is used as a dimerization partner for other cytokines, we suspected that the decreased IL-12 in Zeb1 KD DCs may be due to decreased Il-12p35. Besides, in our RNA-seq data as well we identified that FPKM of IL12p35 gene was ≥2 fold decreased in Zeb1 KD DCs as compared to control cells after 2 h CpG activation (**Supplementary Figure 5D**). Furthermore, we found IRF4 transcription factor to be significantly increased in Zeb1 KD DCs, which is well reported to repress inflammatory cytokines in DCs to increase Th2 development (**Supplementary Figures 6C,D**). Moreover, it would be difficult to exclude the effect of other molecules such as CLRs, Stat5b and other cytokines that are differentially regulated in Zeb1 KD DCs and are reported to impact Th cell polarization intoTh2 subtypes.

# DISCUSSION

Dendritic cells (DCs) are one of the major sentinels of the immune system that determines the fate of T cell dependent adaptive responses. Recently it has been proposed that altering the DC responses can be exploited to affect Th cell subtype development and hence the diseases phenotypes (70, 71). In this study, for the first time, we explored the role of EMT factor Zeb1 in DC function and found that Zeb1 depletion has immune-modulatory effects which skews the Th cells toward the Th2 subtype. Besides that, adoptive transfer of Zeb1 KD cDC1 MutuDCs clears helminth infection by inducing IL-13 and IL-5 secreting Th2 cells in MLNs of infected animals. Moreover, we identified that Zeb1 KD suppresses wide-variety of inflammatory response genes including several CLRs. We concluded that decreased secretion of Th1 inducing inflammatory cytokine IL-12 by Zeb1 KD cDC1 resulted into increased Th2 differentiation. Although the impact of other important factors that were differentially expressed in Zeb1 KD DCs cannot be ignored (24).

Zeb1 is a well-known EMT master regulator that induces mesenchymal properties in cancer cells by increasing N-cadherin and Vimentin levels making it more invasive and metastatic. It directly controls the expression of E-cadherin (Cdh1) by binding to its proximal promoter (72–74). We identified that Zeb1 decrease in our CD8a<sup>+</sup> DCs cDC1 correlated with the levels of Cdh1 and the cells appeared less migratory and smaller in size in vitro as compared to control DCs (data not shown). The impact on Cdh1 was so robust that in all our experiments including RNA-seq we used Cdh1 expression as an indicative of Zeb1 levels. It has been reported that decreased activation and co-stimulation of DCs marked by CD80, CD86, MHC-I and MHC-II resulted into suppression of cytokine secretion (21–23, 28, 31). Besides, these moderately activated

(Continued)

FIGURE 6 | KD cDC1 MutuDCs as compared to control DCs at 12 h after CpG activation n = 2. (C) Heat-map generated from a list of manually curated genes identified from differentially regulated gene list in Zeb1 KD RNA-seq data, which includes the genes that are reported to induce Th2 responses upon activation by DCs. Major DC response genes found to be predominantly regulated after Zeb1 KD are highlighted in bold. To generate this bar-graph Z-score was calculated for each gene from normalized FPKM values to demonstrate the differential regulation of genes in control and Zeb1 KD DCs at 0, 6, and 12 h after CpG activation. (D) Bar-plot depicting the number of genes differentially regulated in Zeb1 KD DCs at 0, 6, and 12 h CpG activation as compared to control DCs. Different color codes indicate the genes that are unique or similar in the list of differentially regulated genes at different time points (0, 6, and 12 h). (E) Bar-plot depicting the biological pathways that were significantly enriched for the genes that differentially regulated after Zeb1 KD in 6 and 12 h CpG activation as compared to control cells. Ingenuity Pathway Analysis (IPA) was used to perform pathway enrichment analysis.

DCs are either immune-suppressive leading to development of anergic T cells or immune-modulatory resulting into Th2 phenotype, which depends on the extent of DC activation (21, 75, 76). We observed a significantly decreased activation of DCs in Zeb1 KD DCs both before and after CpG activation. This resulted into decreased cytokine levels in our Zeb1 depleted DCs at both transcript and protein levels. Moreover we cannot exclude the weak TCR strength in this context with low CD80, CD86 and MHCII for triggering a Th2 phenotype.

We know that DCs educate naïve Th cells to differentiate into different subtypes (Th1, Th2, Th17 or Tregs) which depends on the state of DC activation and the extracellular milieu containing a cytokine cocktail majorly of IL-4, IL-10, IL-12p70, IL-6, TGFβ along with others (21, 77). The cDC1 DCs are reported to induce strong Th1 effector response as compared to their counterpart cDC2 DCs which have the inherent property of inducing Th2 responses (78). We observed constitutive expression of Zeb1 in DCs, which could be important to maintain the cDC1-mediated induction of Th1 cell differentiation. Th1 subtype is generated if there is higher concentration of inflammatory cytokine IL-12 in the milieu whereas increased IL-10 or TGFβ with IL-6 leads to Tregs or Th17 phenotype respectively (22, 28). Moreover, IL-12p40 is also reported to form dimer in the absence of IL-12p35 subunit, which is a potent inhibitor of IL-12p70 activity (79). In contrast the cytokine IL-4 is considered pertinent for Th2 subtype development but it is not secreted by DCs in general, therefore the mode of Th2 generation is considered as a default subtype in the absence or decreased secretion of IL-12 along with absence of other T cell differentiation modulators (21–23, 28, 30, 31, 77). In addition to cytokine milieu, the cell surface receptors like CLRs such as CLEC7a, CLEC4a, DC-SIGN are also reported to skew speciation of Th subtypes in a pathogen/antigen dependent manner (24, 80, 81). These reports supported the development of the Th2 phenotype by our Zeb1 KD cDC1 MutuDCs as we observed a significant decrease in Th1 and Th17 polarizing CLRs along with inflammatory cytokine IL-12p70 in Zeb1 KD DCs. It has also been reported that DC-SIGN interacts with ICAM-1 on T cells and enhances the activation and proliferation of T cells (82). In addition, the cytokine IL-27 that is reported to suppress Th2 differentiation was significantly decreased in our Zeb1 KD DCs (83). This further adds up as a plausible mechanism of Th2 polarization after Zeb1 depletion.

It has been extensively reported that Th subtype balance i.e., Th1 and Th2 controls the helminth H. polygyrus infection in animals. In Balb/C mice that are Th2 prone, the worms are cleared from intestine much faster i.e., 8 weeks as compared to in C57BL/6 animals where the inherent Th1 behavior sustains the worm infection levels for more than 15–20 weeks (66, 68). Therefore we performed adoptive transfer of activated Zeb1 KD cDC1 MutuDCs in C57BL/6 mice to identify if at physiological level Zeb1 KD DCs could skew the Th subtype toward Th2 and thereby leading to increased worm clearance from intestine. We found that treatment of the helminth infected mice with CpG pulsed Zeb1 KD cDC1 MutuDCs cleared the intestinal worms by D31. Even at D10 the egg counts were lower in Zeb1 KD DC treated mice, though we infected all the animals with equal number of larva at D0. We speculate that it was due to adoptive transfer of CpG pulsed Zeb1 KD DCs that secrete suboptimal inflammatory cytokines, at D7 after infection. Furthermore, we also observed that wild type CD8α <sup>+</sup> cDC1 treated animals showed 4- to 5 fold higher egg counts as compared to PBS treated animals due to their inherent property of inducing Th1 responses in vivo. Collectively, this depicted that ZEB1 expression in DCs is pertinent for initial activation of DCs leading to outburst of cytokines important for development of optimal inflammatory Th1 subtypes.

At the mechanistic level it has been demonstrated that Zeb1 acts as a global transcriptional repressor in pre-adipocytes, CD8<sup>+</sup> T cells and mesenchymal to epithelial transition (MET) process (84). In contrast to its role as transcriptional repressor there are studies indicating Zeb1 as a co-activator, for example, in complex with Yap1 it activates metastatic inducer genes (85). In addition, Gene Set Enrichment Analysis (GSEA) of Zeb1 depleted breast cancer cells showed its activating role in the expression of inflammatory response genes IL-6, IL-8, and IL-1α (37). Though our transcriptome analysis showed Zeb1 as a global repressor with higher number of upregulated genes at all the time points, we observed down-regulation of most of the immunogenic genes including CLRs, IL-6, IL-10, IL-12, and IL-27 in Zeb1 KD DCs as compared to control cells. We also showed here by integrating ChIP-seq and RNA-seq data that Zeb1 does not act as transcriptional repressor for immune response gene cluster as for other pathway genes group. It further substantiates and suggests that Zeb1 forms some differential complex to control the immune response genes or Zeb1 mostly regulates them indirectly at transcript level. Even in the cases of EMT and immune evasion it has been widely reported that EMT process coincides well with the increased inflammatory environment (86–88). It might be possible that Zeb1 upregulation during EMT also results in increased inflammation in the tumor microenvironment.

FIGURE 7 | Zeb1 ChIP-seq bound peaks by annotations tool GREAT showing the genomic locations of Zeb1 bound regulatory regions with respect to gene transcription start sites (TSS). (C) List of top de-novo motifs significantly enriched at the Zeb1 bound peaks in unstimulated DCs and their annotated TFs. Zeb1 motif was found to be the top highly enriched motif. (D) SeqMINER clustering demonstrating the overlap of Zeb1 bound peaks in unstimulated CD8α <sup>+</sup> cDC1 with PU.1 bound genomic regions in similar condition. The peaks were also overlapped with IRF4 ChIP-seq publicly available data of control and LPS stimulated primary BMDCs. (E) Heat-map demonstrating the immune response pathway genes that were differentially regulated after Zeb1 KD 0, 6, and 12 h after CpG activation as compared to control cells. The ChIP-seq binding was also overlapped to identify if Zeb1 directly regulates immune response genes. All the immune response genes that were differentially regulated (>2-fold up or down-regulated) are listed. The genes that were reported and are important in Th cell polarization through DCs are highlighted in bold. To generate this bar-graph Z-score was calculated for each gene from normalized FPKM values to demonstrate the differential regulation of genes in control and Zeb1 KD DCs at 0, 6, and 12 h after CpG activation. (F) Bar-plot depicting the differential regulated genes in Zeb1 KD DCs at 0, 6, and 12 h after CpG activation. The genes were classification into two different groups based on GO annotation i.e., immune response genes (immune) and other pathway genes (others). It was observed that in immune response gene group there is preferential upregulation of genes after Zeb1 KD whereas in the other group there was higher number of upregulated genes (>1.6- to 2.5-fold) compared to down-regulated ones. (G) Bar-plot showing the IPA pathways significantly enriched for the genes that were directly bound and regulated by Zeb1 at 12 h after CpG activation.

DC -T cell co-culture n = 8–10. (B) Scatter-plot depicting a significant decrease in percentage positive cells for Th2 marker GATA3<sup>+</sup> in Zeb1 KD DCs upon addition of recombinant IL-12 cytokine and anti-IL4 in the culture media during DC-T cell co-culture n = 8–10. (C) Bar-plot representing the MFI for Th1 marker Tbet<sup>+</sup> in Zeb1 KD DCs upon supplementation of rIL12 and anti-IL4 cytokine in the culture media during DC -T cell co-culture n = 8–10. (D) Bar-plot representing the MFI for Th2 marker GATA3<sup>+</sup> in Zeb1 KD DCs upon supplementation of rIL12 and anti-IL4 cytokine in the culture media during DC -T cell co-culture n = 8–10. p-values are calculated using two tailed unpaired student's t-test, error bars represent SEM. \*≤0.05, \*\*≤0.01, \*\*\*≤0.001).

We would like to conclude with a message that expression of TF Zeb1 is pertinent for CD8α <sup>+</sup> cDC1 activation leading to immunogenic response generation. It regulates activation and thereby secretion of cytokines by CD8α <sup>+</sup> cDC1 DCs that are pertinent to induce pathogen/signal specific T cell responses.

# ETHICS STATEMENT

All the animal experiments were carried following the guidelines of institutional animal ethical committee. The Committee for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA), under the guidelines approved by Institutional Animal Ethics Committee (IAEC).

#### AUTHOR CONTRIBUTIONS

SS, AA, BG, HA-O and SR designed the study. SS, AA, VB, MK performed the experiments SS, AA, BG, HA-O and SR did interpretation of the results. AG, SR performed the computational/bio-informatics analysis. SS, HA-O and SR wrote the manuscript. SS, AA, AG, BG, MK, HA-O and SR edited the manuscript.

#### FUNDING

Grant funds from SERB (EMR/2016/000717), DST-SNSF (DST/INT/SWISS/SNSF/P-47/2015), DBT Ramalingaswami fellowship, DBT (BT/PR15908/MED/12/725/2016). ILS provided intramural support and infrastructure.

#### ACKNOWLEDGMENTS

We thank Nicola Harris lab at EPFL, Lausanne for providing the helminth larvae. Christine Lavenchy and Vanessa Mack for excellent technical help. Thanks to Ton Rolink for providing the FLT3L transgenic mice. Atimukta Jha for editing the manuscript. SS and AG is supported by ILS fellowship. AA is supported by DBT-SRF. VB is supported by SERB grant.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2018.02604/full#supplementary-material

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Smita, Ahad, Ghosh, Biswas, Koga, Gupta, Acha-Orbea and Raghav. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Staphylococcus aureus PSM Peptides Modulate Human Monocyte-Derived Dendritic Cells to Prime Regulatory T Cells

Jennifer R. Richardson, Nicole S. Armbruster, Manina Günter, Jörg Henes and Stella E. Autenrieth\*

Department of Internal Medicine II, University of Tübingen, Tübingen, Germany

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Alexander Mildner, Helmholtz-Gemeinschaft Deutscher Forschungszentren (HZ), Germany Luigi Racioppi, Università degli Studi di Napoli Federico II, Italy

\*Correspondence:

Stella E. Autenrieth stella.autenrieth@ med.uni-tuebingen.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 27 July 2018 Accepted: 23 October 2018 Published: 13 November 2018

#### Citation:

Richardson JR, Armbruster NS, Günter M, Henes J and Autenrieth SE (2018) Staphylococcus aureus PSM Peptides Modulate Human Monocyte-Derived Dendritic Cells to Prime Regulatory T Cells. Front. Immunol. 9:2603. doi: 10.3389/fimmu.2018.02603 Staphylococcus aureus (Sa), as one of the major human pathogens, has very effective strategies to subvert the human immune system. Virulence of the emerging community-associated methicillin-resistant Sa (CA-MRSA) depends on the secretion of phenol-soluble modulin (PSM) peptide toxins e.g., by binding to and modulation of innate immune cells. Previously, by using mouse bone marrow-derived dendritic cells we demonstrated that PSMs in combination with various Toll-like receptor (TLR) ligands induce a tolerogenic DC phenotype (tDC) characterized by the production of IL-10 and impaired secretion of pro-inflammatory cytokines. Consequently, PSM-induced tDCs favored priming of CD4+CD25+FoxP3<sup>+</sup> Tregs with suppressor function while impairing the Th1 response. However, the relevance of these findings for the human system remained elusive. Here, we analyzed the impact of PSMα3 on the maturation, cytokine production, antigen uptake, and T cell stimulatory capacity of human monocyte-derived DCs (moDCs) treated simultaneously with either LPS (TLR4 ligand) or Sa cell lysate (TLR2 ligand). Herein, we demonstrate that PSMs indeed modulate human moDCs upon treatment with TLR2/4 ligands via multiple mechanisms, such as transient pore formation, impaired DC maturation, inhibited pro- and anti-inflammatory cytokine secretion, as well as reduced antigen uptake. As a result, the adaptive immune response was altered shown by an increased differentiation of naïve and even CD4<sup>+</sup> T cells from patients with Th1/Th17-induced diseases (spondyloarthritis and rheumatoid arthritis) into CD4+CD127−CD25hiCD45RA−FoxP3hi regulatory T cells (Tregs) with suppressor function. This Treg induction was mediated most predominantly by direct DC-T-cell interaction. Thus, PSMs from highly virulent Sa strains affect DC functions not only in the mouse, but also in the human system, thereby modulating the adaptive immune response and probably increasing the tolerance toward the bacteria. Moreover, PSMα3 might be a novel peptide for tolerogenic DC induction that may be used for DC vaccination strategies.

Keywords: monocyte-derived dendritic cells, Staphylococcus aureus, phenol-soluble modulins, immune tolerance, immunity, regulatory T cells

# INTRODUCTION

Dendritic cells (DCs) are specialized antigen presenting cells (APCs) able to prime naïve T cells thereby inducing a primary immune response and maintaining self-tolerance (1). Initially DCs occur in an immature state, specialized for antigen uptake with a high endocytic capability (2). The recognition of pathogens via germ-line encoded pattern-recognition receptors, like Toll-like Receptors (TLRs), leads to DC maturation (3). This event is characterized by the loss of their endocytic capacities and the upregulation of CCR7, co-stimulatory molecules, and HLA-DR, necessary for homing into the draining lymph node and T-cell priming (4). Furthermore, DCs secrete pro-inflammatory cytokines, such as TNF-α, IL-6 or IL-12, which recruit other immune cells for pathogen clearance and contribute to T helper cell (Th) differentiation (5). Apart from inducing an efficient immune response, DCs are also crucial for maintaining immune tolerance in the steady-state. Although the specific phenotype of so-called tolerogenic DCs (tDCs) and the molecular mechanism involved in tolerance induction by these cells are not entirely defined (6–9) they are characterized by an immature phenotype and produce high amounts of anti-inflammatory cytokines, e.g., IL-10 and TGF-ß, which possess critical immunoregulatory functions like controlling/regulating the production of proinflammatory cytokines. They have the potential to induce regulatory T cell (Treg) expansion thereby impairing effector T cell responses (8, 10–12).

Various pathogens and tumors can induce tDCs and subsequent Treg differentiation as immune escape strategy to impair clearance. This process is mediated by pathogenic products from e.g., C. albicans, S. mansoni and V. cholerae, which are partially used for the production of immunosuppressive drugs. These are widely used for therapy of autoimmune diseases or transplant rejections even though they have severe side effects by suppressing the entire host immune system (11, 13). Therefore, DC vaccination strategies by applying tDCs are an attractive alternative (8, 9, 13). Several clinical trials started to analyze the effect of tDCs as treatment option for patients with autoimmune disorders (8).

Phenol-soluble modulins (PSMs) are short amphipathic α-helical peptides, which are produced by highly virulent Staphylococci, such as community-associated Methicillinresistant Staphylococcus aureus (Sa) promoting, e.g., cell lysis thereby evading clearance by immune cells (14, 15). Two types of PSMs are distinguished according to their length: α-type PSMs (∼20–25 AA) and β-type PSMs (∼44 AA) (16). The PSMα peptides are the most potent PSMs regarding cytolysis and highly contribute to the virulence of Sa (16, 17). Own previous studies with mouse bone-marrow derived DCs (BM-DCs) showed that PSMα3 prime tDCs when co-incubated with various TLR ligands (TLRL), regardless which TLR was activated. Molecularly, this event is characterized by the increased activation of the p38-CREB pathway, which in consequence leads to diminished pro-inflammatory cytokine production but increased IL-10 secretion. These PSM-induced tDCs favored priming of CD4+CD25+FoxP3<sup>+</sup> Tregs with suppressor function (10, 12, 18). Thus, we hypothesized that PSMs of Sa likewise induce tDCs in the human system.

Herein, we show that PSMα3 penetrates and modulates human monocyte-derived DCs (moDCs) by altering the TLR2- or TLR4-induced maturation, inhibiting pro- and antiinflammatory cytokine production and reducing antigen uptake, but producing indolamin-2,3-dioxygenase (IDO). As a result, the frequency of CD4+CD127−CD25hiCD45RA−Foxp3hi Tregs is increased, while Th1 responses are diminished. Moreover, PSMα3-induced tDCs from healthy donors even enhanced differentiation of CD4<sup>+</sup> T cells from patients with Th17 associated autoimmune diseases to Tregs. Thus, PSMα3 might be a novel peptide for manipulating DCs to become tolerogenic for DC vaccination strategies.

# MATERIALS AND METHODS

#### Research Subjects

Buffy coats from healthy volunteers were obtained from the ZKT Tübingen GmbH. Fresh blood was obtained from healthy volunteers with informed consent. This was approved by the ethical review committee of the medical faculty of the Eberhard-Karls-University of Tübingen with the project number 633/2012BO2. Blood from patients with TH17-associated autoimmune diseases were obtained from the division of Rheumatology, Department of Internal Medicine II, University Hospital Tübingen. This was approved by the ethical review committee of the medical faculty of the Eberhard-Karls-University of Tübingen with the project number 046/2015BO2.

#### Reagents

Formylated PSM peptides (PSMα3, δ-Toxin) were synthesized at the Interfaculty Institute of Cell Biology, Department of Immunology, University of Tübingen. FITC-labeled PSMα2 was synthesized at the Group of Hubert Kalbacher, Interfaculty Institute of Biochemistry, University of Tübingen. Sa cell lysate (Sa lysate) containing lipopeptides and specifically activating TLR2 was prepared from a protein A-deficient Sa mutant strain (SA113) and provided by Andreas Peschel, Interfaculty Institute of Microbiology and Infection Medicine, University of Tübingen.

#### Isolation of Peripheral Blood Mononuclear Cells

Buffy coats or fresh blood was diluted with Dulbecco's PBS (Life Technologies) (Buffy Coats 1:7 blood: PBS; Fresh blood 1:1 blood: PBS). Peripheral blood mononuclear cells (PBMCs) were obtained by density gradient centrifugation at 2000 rpm for 20 min at room temperature with 35 mL cell suspension stacked on 15 mL Biocoll separation solution (Biochrom). The interphase containing the PBMCs was abstracted and washed twice with

**Abbreviations:** APC, antigen presenting cell; BM-DCs, bone-marrow derived DCs; CA-MRSA, community-associated methicillin-resistant Sa; DCs, dendritic cells; IDO, indolamin-2,3-dioxygenase; iTregs, induced regulatory T cells; LDH, Llactate dehydrogenase; moDCs, monocyte-derived DCs; PBMCs, peripheral blood mononuclear cells; PSM, phenol-soluble modulin peptides; Sa, Staphylococcus aureus; Sa lysate, Staphylococcus aureus cell lysate; tDCs, tolerogenic DCs; Th, T helper cell; TLR, Toll-like receptor; TLRL, Toll-like receptor ligand; Tregs, regulatory T cells.

PBS. PBMCs were further used to generate human moDCs and for the isolation of naïve CD4<sup>+</sup> T cells and CD4<sup>+</sup> T cells.

#### Generation of Human Monocyte-Derived DCs

PBMCs were plated in a tissue-treated 6-well plate (6 × 10<sup>6</sup> cells per well) in DC medium [RPMI1640 (Merck), 10% FBS (Sigma), 2 mM L-Glutamine (Life Technologies), 100 U/mL Penicillin-Streptomycin (Life Technologies), 1 × non-essential amino acids (Merck), 1 mM sodium pyruvate (Merck) and 50µM 2 mercaptoethanol (Roth)] and incubated for 1 h at 37◦C, 5% CO2. After that, wells were washed with medium and PBS discarding the non-adherent cells. 3 mL/well DC medium containing 50 ng IL-4 and 100 ng GM-CSF (both from Miltenyi) was added to the remaining cells. Cells were incubated for 6 d at 37◦C, 5% CO2. Cytokines were again added on day 2 and day 4. At day 6 the cells were used for the following experiments. The purity of the moDC culture was always >90% of leukocytes (**Figure S1**).

### Cytokine/Indolamin-2,3-Dioxygenase Production by moDCs

MoDCs (2.5 × 10<sup>5</sup> ) were seeded in a 96-well plate and treated with 3µg/mL Sa lysate or 100 ng/mL LPS in combination with or without PSMα3 (10µM). For some experiments moDCs were treated with different concentrations of PSMα3 or additionally stimulated with 100 ng/mL Pam2CSK4 (for TLR2/TLR6; InvivoGen), 1µg/mL Pam3CSK4 (for TLR1/TLR2; InvivoGen), 1µg/mL CpG ODN 2395 (for TLR9; InvivoGen), 5µg/mL Imiquimod (for TLR7; InvivoGen), 2µg/mL Flagellin (for TLR5; InvivoGen) or 10µg/mL LTA (for TLR2/TLR4; InvivoGen) in combination with or without PSMα3 (10µM). Supernatants were collected after 6 h, 24 h or 48 h and analyzed for TNF, IL-10, IL-12 and IDO production, respectively. Cytokines and IDO in the supernatants were determined by sandwich ELISA [eBioscience (TNF, IL-10, IL-12), R&D Systems (IDO)] according to the manufacturer's instructions.

#### Flow Cytometry

For moDC surface marker analysis of the costimulatory and inhibitory molecules, moDCs (2 × 10<sup>5</sup> ) were seeded in a 96-well plate and stimulated with 3µg/mL Sa lysate or 100 ng/mL LPS with or without PSMα3 (10µM) or PSMα3 alone for 6 h or 24 h. Cells were removed from the plate using Accutase (Sigma-Aldrich) and treated with IgG from human serum (1 µg of human IgG per 100,000 cells; Sigma-Aldrich) for 20 min at room temperature to avoid unspecific binding via Fc receptors. Cells were stained with ZombieAqua (BioLegend) to exclude dead cells and fluorochrome conjugated extracellular antibodies: HLA-DR BV650 (L243, BioLegend), CD11b BV510 (ICRF44, BioLegend), CD11c APC (MJ4-27G121, Miltenyi), CD11c PE-Cy7 (Bu15, BioLegend), CD40 FITC (5C3, eBioscience), CD80 PE-Cy7 (2D10, BioLegend), CD83 PE-Dazzle 594 (HB15e, BioLegend), CD86 BV605 (IT2.2, BioLegend), PD-L1 PE (29E.2A3, BioLegend), PD-L2 PE (MIH18), ILT3 PE (ZM4.1, BioLegend) for 20 min at 4◦C. FACS buffer [PBS containing 1% FBS, 2 mM EDTA (Merck) and 0.09% NaN<sup>3</sup> (Sigma-Aldrich)] was used for all incubations and washing steps. At least 50,000 cells were acquired using a LSR Fortessa flow cytometer (BD Biosciences) with the DIVA software (BD Biosciences) and were further analyzed using FlowJo 10.4.2 software (Tree Star).

# Phosphoflow

For the experiments analyzing phosphorylation of signaling cascades, moDCs (2 × 10<sup>5</sup> ) were seeded in a 96-well plate and treated for 1 h with 100 ng/mL LPS with or without PSMα3 (10µM) or PSMα3 alone. Cells were removed from the plate using Accutase and treated with IgG from human serum (1 µg of human IgG per 100,000 cells) for 20 min at room temperature to avoid unspecific binding via Fc receptors. Cells were stained with ZombieAqua (BioLegend) to exclude dead cells and fluorochrome conjugated extracellular antibodies: HLA-DR PE (L243; BD Biosciences) and CD11c APC/Cy7 (Bu15; BioLegend) for 20 min at 4◦C. To detect intracellular p-p38 and p-NF-κB cells were fixed with 2% paraformaldehyde (VWR) in PBS, permeabilized with 90% freezing methanol (Applichem) overnight and stained with the primary Abs to phospho-p38 MAPK (Thr180/Tyr182; clone 12F8) or phospho-NF-κB p65 (Ser536; clone: 93H1) (both from Cell Signaling) for 1 h in the dark at room temperature followed by DyLight649-conjugated AffiniPure Goat At-Rabbit IgG (Jackson ImmunoResearch) for 15 min at 4◦C. PBS with 0.5% BSA (Biomol) was used for incubation and washing steps of intracellular antibody stainings. At least 50,000 cells were acquired using a Canto-II (BD Biosciences) with DIVA software (BD Biosciences) and were further analyzed using the FlowJo 10.4.2 software (Tree Star).

### Measurement of Antigen Uptake by Flow Cytometry or Multispectral Imaging Flow Cytometry

MoDCs (5 × 10<sup>5</sup> ) were seeded in a 48-well plate and stimulated for 24 h with 3µg/mL Sa lysate or 100 ng/mL LPS with or without PSMα3 (10µM) or PSMα3 alone prior to the incubation with Ovalbumin (OVA)-AlexaFluor647 (5µg/mL, Invitrogen) together with PSMα2 FITC (0.5µM) for 30 min at 37◦C, 5% CO2. Unspecific binding of OVA/PSMα2 was assessed by incubating the cells on ice. Cells were washed twice with ice-cold PBS containing 2% FBS. Subsequently, cells were blocked and stained with ZombieAqua (BioLegend), HLA-DR BV650 (L243, BioLegend), HLA-DR APC-Cy7 (L243, BioLegend) and CD11c PE-Cy7 (Bu15, BioLegend) as described above and analyzed by flow cytometry or by multispectral imaging flow cytometry. For the latter, images of 10,000 living moDCs were acquired using the Image-Stream mkII (Amnis) with the INSPIRE instrument controller software. The data were analyzed using the IDEAS analysis software (Merck Millipore).

#### Lactate Dehydrogenase Release

MoDCs (2 × 10<sup>5</sup> ) were seeded in a 96-well plate and treated with Triton X100 (1%; Sigma-Aldrich), DMSO (2%, Fluka), PSMα2 (10µM), PSMα3 (2.5, 5, 7.5, and 10µM), δ-Toxin (10µM) or OVA (10 µg, Sigma-Aldrich) for 10 min at 37◦C, 5% CO2. Cell death was determined using 7-aminoactinomycin D (7-AAD, Biomol) staining and acquisition on a Canto II flow cytometer.

24 h (B,C) and analyzed by flow cytometry. MoDCs were characterized as living CD11c+HLA-DR<sup>+</sup> cells and the expression of the indicated costimulatory (A,B) and inhibitory molecules (C) was determined. The graphs show the mean fluorescence intensity of the respective marker expression as fold change of untreated cells. The graphs show n ≥ 3 independent experiments (mean ± SEM) performed in triplicates. Representative histogram overlays of HLA-DR and CD40 after 6 h (A), CD80, CD40 (B), and PD-L1 (C) after 24 h. \*p < 0.05, \*\*p < 0.005 or \*\*\*\*p < 0.0001, one-way ANOVA with Turkey's posttest or Kruskal-Wallis with Dunn's posttest.

Supernatants were used for the analysis of lactate dehydrogenase (LDH) release using the Cytotoxicity Detection Kit (Roche) according to the manufacturer's instructions. Absorbance was measured at 492 nm and 620 nm over a period of 1 h with an interval of 5 min using the Spark 10 M microplate reader (Tecan).

# T-Cell Assay

MoDCs (5 × 10<sup>4</sup> ) were seeded in a 96-well plate and stimulated with 3µg/mL Sa lysate or 100 ng/mL LPS with or without PSMα3 (10µM) or PSMα3 alone for 24 h. For some experiments moDCs were pre-treated with 200µM 1-Methyl-D-tryptophan (1-DMT) for 1 h prior to the stimulation. Human naïve CD4<sup>+</sup> T cells were isolated from PBMCs using the MojoSortTM Human CD4 Naïve T Cell Isolation Kit (BioLegend) according to the manufacturer's instructions. For the magnetic cell separation, a LS column (Miltenyi Biotech) was placed into the QuadroMACS Separator (Miltenyi Biotech) and rinsed with MACS buffer (PBS containing 0,5% BSA (Biomol) and 2 mM EDTA). The cell suspension was applied to the column, and the column was washed three times with 3 mL MACS buffer. The untouched naïve CD4<sup>+</sup> T cells were collected in the flow through. The purity of isolated (naïve) CD4<sup>+</sup> T cells was always ≈85% (**Figure S4**). The naïve CD4<sup>+</sup> T cells were labeled with CFSE (5µM, BioLegend) according to the manufacturer's instructions. 2 × 10<sup>5</sup> T cells diluted in 100 µL T cell medium [RPMI1640 (Merck), 10% FBS (Sigma), 2 mM L-Glutamine (Life Technologies), 100 U/mL penicillin-streptomycin (Life Technologies), 1 × non-essential amino acids (Merck), 1 mM sodium pyruvate (Merck), 10 mM HEPES (Biochrom) and 50µM 2-mercaptoethanol (Roth)] were added to the moDCs. To investigate whether secreted factors from DCs upon PSM-treatment mediate Treg priming, T cells were co-cultured with untreated moDCs adding conditioned medium from LPS or LPS + PSMα3 stimulated DCs. In a second assay moDCs treated as described above were splitted using Accutase (Sigma-Aldrich) for 5 min at room temperature and again sowed with either fresh or conditioned DC medium (TLRL or TLRL + PSMα3). In some conditions, DC were fixed with 1% paraformaldehyde for 10 min at 4◦C to address the impact of newly secreted factors on Treg priming by DCs. 3–4 d after co-culture T cells were blocked with IgG from human serum for 15 min at room temperature and subsequently stained with ZombieAqua, CD4 APC-Vio770 (REA623, Miltenyi), CD3 Pacific Blue (SK7, BioLegend), CD25 PE-Cy7 (BC96, eBiosciences), CD127 PE (eBioRDR5, eBiosciences) and CD45RA BV605 (HI100, BioLegend) for 20 min at 4◦C. For intracellular staining, cells were fixed and permeabilized with the Foxp3/ Transcription Factor Staining Buffer Set (eBiosciences), blocked and stained with CD4 APC-Vio770, CD3 Pacific Blue, FoxP3 AlexaFluor647 (259D, BioLegend), T-bet PE-Dazzle 594 (4B10, BioLegend), GATA3 PerCP-Cy5.5 (16E10A23, BioLegend) and RORγt BV650 (Q21- 559, BD Biosciences) for 45 min at 4◦C. FACS buffer was used for all incubations and washing steps for the extracellular staining, and 1 × permeabilization buffer (Foxp3/Transcription Factor Staining Buffer Set (eBiosciences) was used for all incubations and washing steps for the intracellular staining. At least 70,000 cells were acquired using an LSR Fortessa flow cytometer (BD Biosciences) with the DIVA software (BD Biosciences) and were further analyzed using FlowJo 10.4.2 software (Tree Star).

#### Autologous T-Cell Assay

CD14<sup>+</sup> cells from PBMCs of patients with TH17-associated autoimmune diseases were isolated by MACS using CD14 MicroBeads (Miltenyi Biotech) and plated in a tissue-treated 6 well plate (1.3 × 10<sup>6</sup> cells per well) in DC medium containing 50 ng IL-4 and 100 ng GM-CSF for 6 d to generate moDCs. The remaining CD14<sup>−</sup> cells were frozen at −80◦C in RPMI1640 supplemented with 20 % FBS and 10% DMSO. After 6 d moDCs (5 × 10<sup>4</sup> ) were seeded in a 96-well plate and stimulated with 100 ng/mL LPS with or without PSMα3 (10µM) for 24 h. The CD14<sup>−</sup> cells were thawed and used to isolate CD4<sup>+</sup> T cells by MACS with CD4 MicroBeads (Miltenyi Biotech). The CD4<sup>+</sup> T cells were labeled with CFSE (5µM, BioLegend) according to the manufacturer's instructions and 2 × 10<sup>5</sup> T cells diluted in 100 µL T cell medium were added to the moDCs. 3–4 days after coculture T cells were stained as above and iTregs were analyzed by flow cytometry using an LSR Fortessa flow cytometer (BD Biosciences) with the DIVA software (BD Biosciences) and were further analyzed using FlowJo 10.4.2 software (Tree Star).

# Cytokine Production in the moDC T Cell Co-culture

Fifty microliter cell culture supernatants from the T cell assay were taken on day 1, 2 and 3 and cytokine production from 15 µL was analyzed by performing bead-based immunoassays in a 96-well plate [LEGENDplex human B cell Panel (13-Plex) and LEGENDplex Free Active/Total TGF-β1 (BioLegend)] according to the manufacturer's instructions, using the Lyric flow cytometer with autosampler (BD Bioscience).

# T Cell Suppression Assay

MoDCs (2 × 10<sup>5</sup> ) were seeded in a 48-well plate and stimulated with 100ng/mL LPS and 10µM PSMα3 for 24 h. Human CD4<sup>+</sup> T cells were isolated from PBMCs using the human CD4 MicroBeads Kit (Miltenyi) according to the manufacturer's instructions using LS columns. 8 × 10<sup>5</sup> T cells were added to the stimulated moDCs and cultured for 4 d at 37◦C, 5% CO2. T cells were stained with CD4 APC-Vio770, CD25 PE-Cy7, CD127 PE and CD45RA PerCP (HI100, BioLegend) as described above and dead cells were excluded using DAPI (Sigma-Aldrich,16,7 ng/mL). Tregs were purified by FACS sorting using an ARIA IIu cell sorter (BD Bioscience), according to the surface molecule expression (CD4+CD127−CD25hiCD45RA−, see **Figure S7**). CD4<sup>+</sup> T cells isolated from PBMCs from a different donor were used as T effector (Teff) cells purified with CD4 MicroBeads Kit (Miltenyi). The Teffs were labeled with CFSE (5µM), and 8 × 10<sup>4</sup> cells were seeded in a 96- well plate in T cell medium together with the indicated numbers of sorted Tregs. For T cell activation of the Teffs Dynabeads (Human T-Activator CD3/CD28Proliferation; Gibco) were added according to the manufacturer's instructions. The proliferation of Teffs was assessed after 3 d by flow cytometry. Dead cells were excluded by staining the cells with ZombieAqua. 20,000 cells were acquired at the Canto II with the DIVA software (BD Biosciences) and were further analyzed using the proliferation tool in FlowJo 10.4.2 software (Tree Star).

#### Statistical Analysis

Statistical analysis was performed using GraphPad Prism 7.0a software (GraphPad, San Diego, CA). Statistical differences were determined using one-way ANOVA with Turkey's posttest in case data were normally distributed (Shapiro-Wilk normality test). Otherwise, data were analyzed using the Kruskal-Wallis nonparametric test. The differences were considered as

statistically significant if p < 0.05 (<sup>∗</sup> ), p < 0.005 (∗∗), p < 0.001 ( ∗∗∗ ), or p < 0.0001 (∗∗∗∗).

#### RESULTS

#### PSMs Modulate Surface Molecule Expression of TLR-Treated moDCs

The maturation of DCs is an essential event for successful T-cell activation comprising the upregulation of maturation markers (e.g., CD83 and HLA-DR), co-stimulatory (e.g., CD80, CD86, CD40) and co-inhibitory molecules (e.g., PD-L1, PD-L2 or ILT3). To investigate if the Sa-derived toxin PSMα3 has an impact on DC maturation, moDCs were treated with either the TLR2 ligand Sa lysate (12) or the TLR4 ligand LPS with or without PSMα3. Surface marker expression was analyzed by flow cytometry after 6 or 24 h (Gating see **Figure S1**). All surface molecules analyzed were increased upon TLRL treatment compared to untreated moDCs at 6 and 24 h with the exception of CD80 after 6 h (**Figure 1**) confirming DC maturation (mDCs – mature DCs). Simultaneous treatment of moDCs with TLRLs and PSMα3 for 6h revealed a tendency, but no significant increase of the maturation marker CD83 and CD86 compared to treatment with TLRLs alone (**Figure 1A**). In contrast, HLA-DR was significantly increased 6h (Sa p = 0.0081; LPS p = 0.0226) after TLRL treatment in combination with PSMα3 whereas CD40 was less expressed on mDCs treated with LPS in combination with PSMα3 compared to LPS treated cells after 6 h (LPS p < 0.0001) and 24 h (LPS p < 0.0049) (**Figures 1A,B**). Similarly, CD80

FIGURE 3 | OVA and PSMα2 by living CD11c+HLA-DR<sup>+</sup> moDCs was assessed by flow cytometry. The histogram overlays (left) and the bar graphs (right, collected data) show the mean fluorescence intensity of OVA-AlexaFluor647 (A) or PSMα2-FITC (B), in case of the collected data as fold change of unstimulated cells (n ≥ 5, performed in triplicates, mean ± SEM). (C) The localization of OVA-AlexaFluor647 and PSMα2-FITC in CD11c+HLA-DR<sup>+</sup> moDCs was analyzed by multispectral imaging flow cytometry (one representative experiment of n = 3 independent experiments; mean ± SEM). Representative bright field (BF) and fluorescence images of moDCs are shown for the different treatments. (D) MoDCs were either not treated (unst.) or incubated with 1% Triton X-100, 2% DMSO, PSMα3 (2.5µM, 5µM, 7.5µM or 10µM), 10µM PSMα2, 10µM δ-Toxin or 5µg/mL OVA for 10 min. The LDH release was determined in the cell culture supernatants (one representative experiment of n = 3 independent experiments; mean ± SEM). \*\*\*\*p < 0.0001, one-way ANOVA with Turkey's posttest or Kruskal-Wallis with Dunn's posttest.

up-regulation on TLR4-stimulated mDCs was impaired when the cells were treated with PSMα3 for 24 h (LPS p = 0.0135) (**Figure 1B**). The analysis of co-inhibitory molecule expression revealed no significant differences after 24 h; however, PSMα3 showed a tendency to prevent the upregulation of the coinhibitory molecule PD-L1 in TLR4-stimulated moDCs (LPS p = 0.0615) (**Figure 1C**). In summary, PSMα3 enhanced the early upregulation of HLA-DR on DCs but prevented that of the costimulatory molecules CD40 and CD80 especially upon TLR4 stimulation.

#### PSMs Impair Pro- and Anti-inflammatory Cytokine Secretion by TLR4-Treated moDCs

Stimulation of TLRs not only leads to DC maturation but also to the expression of cytokines to initiate an immune response. Previously, we showed that PSMα3 impaired the proinflammatory cytokine production triggered by various TLRLs in mouse BM-DCs, whereas it induced the expression of the anti-inflammatory cytokine IL-10 (12, 18). Therefore, cell culture supernatants from moDCs treated with Sa lysate or LPS with or without PSMα3 were analyzed after 6 h for TNF (**Figure 2A**) and after 24 h for IL-12 (**Figure 2B**), IL-10 (**Figure 2C**) or TNF production (**Figure S2A**). TLRL treatment led to an overall induction of cytokine secretion with the exception of IL-12 (**Figure 2B**) in TLR2-ligand treated mDCs (**Figure 2**). The production of TNF, IL-12 and IL-10 was impaired by PSMα3 in LPS treated mDCs (TNF 1.800 vs. 13.500 pg/ml, p = 0.0082; IL-12 550 vs. 4.500 pg/ml, p = 0.021; IL-10 1090 vs. 3850 pg/ml, p = 0.0080) (**Figure 2**) the latter in a concentration dependent manner (**Figure S2B**), but not in Sa lysate treated mDCs. The treatment of DCs with PSMα3 in combination with other TLRligands like Pam2CSK4, Pam3CSK4, LTA, Flagellin, CpG, and Imiquimod showed no significant differences in the production of TNF and IL-10 compared to TLR-L treatment alone (**Figure 2** and **Figure S2C**). In summary, PSMα3 inhibited pro- as well as anti-inflammatory cytokine production by LPS-treated mDCs.

# PSMs Impair NF-κB and p38 Phosphorylation in LPS-Treated moDCs

To investigate the signaling pathways involved in the cytokine modulation by PSMs, DCs were treated with LPS in the presence or absence of PSMα3 for 60 min and phosphorylation of NF-κB p65 (p-NF-κB) (**Figure 2D**) and p38 MAPK (pp38) (**Figure 2E**) was analyzed by flow cytometry. Treatment of DCs with PSMα3 did not affect p- NF-κB or p-p38 compared to untreated DCs (**Figures 2D,E** heatmap), whereas after LPS treatment a 2.5-fold and 2-fold increase of p- NF-κB and p-p38, respectively, was observed compared to untreated DCs (**Figures 2D,E**). DCs incubated with LPS combined with PSMα3 revealed a 2.0-fold and 1.5-fold increase of NF-κB or p38 phosphorylation (**Figures 2D,E**). Analyzing signaling pathways in moDCs from 5 donors revealed by trend a reduced, but not significant, NF-κB or p38 phosphorylation upon PSMα3 treatment. These data indicate that impaired phosphorylation of both NF-κB and p38 pathways by PSMα3 may account for impaired cytokine production as well as co-stimulatory molecule expression by moDCs upon LPS treatment.

#### PSMs Inhibit Antigen Uptake by moDCs

Antigen uptake is a pivotal task of DCs and necessary for T-cell activation. To elucidate possible effects of PSMs on this event, moDCs were treated with or without either Sa lysate or LPS in the presence or absence of PSMα3 for 24 h. Subsequently cells were incubated with the fluorescently-labeled (AlexaFluor647) model antigen OVA in combination with FITClabeled PSMα2 for 30 min. Immature DCs (iDCs – immature DCs) take up OVA mainly by clathrin-mediated endocytosis and macropinocytosis, whereas mature DCs retain antigen uptake by receptor-mediated endocytosis (12, 19–21). This effect was observed when moDCs matured after treatment with either Sa lysate or LPS by ∼68 or ∼63%, respectively (**Figure 3A**). OVA uptake was reduced by ∼48% after solely treatment with PSMα3 (p < 0.0001) (**Figure 3A**), showing that iDCs are affected in their task to take up antigen by PSMs. A trend to further impaired antigen uptake by TLRL and PSMα3 treated moDCs was observed (Sa p = 0.6603, LSP p = 0.9054) (**Figure 3A**). Incubation of iDCs with OVA on ice was used to exclude unspecific binding (**Figure 3A**) thereby preventing remodeling of the actin cytoskeleton, which is required for clathrin-mediated endocytosis and micropinocytosis (22). In conclusion, PSMα3 affects moDCs in their antigen uptake capacity.

#### PSMs Penetrate the Membrane of moDCs Via Transient Pore Formation

Previously, we and others showed that PSMs form transient pores into the cell membrane thereby entering the cytosol (10, 15). To address whether PSMs are internalized by human moDCs via mechanisms of antigen uptake or by pore formation iDCs were incubated with fluorescently-labeled PSMα2 as described above and analyzed by flow cytometry and multispectral imaging flow cytometry. Despite the uptake of OVA-AlexaFluor647, PSMα2- FITC was observed in moDCs regardless of their maturation

FIGURE 5 | PSMα3-treated moDCs increase the proliferation and frequency of iTregs. (A–C) MoDCs treated for 24 h with Sa lysate or LPS with or without PSMα3 or PSMα3 alone were co-cultured with CFSE-labeled naïve CD4<sup>+</sup> T cells for 3 d or 4 d. iTregs were analyzed by flow cytometry and characterized as CD4+CD127−CD25hiCD45RA−Foxp3hi cells (A). The frequency of iTregs from CD4<sup>+</sup> T cells (B) and the proliferation of iTregs (C) co-cultured with TLRL-only or in combination with PSMα3-treated mDCs was analyzed (n ≥ 5 performed in triplicates, mean ± SEM). Proliferation was compared to T cells (Continued) FIGURE 5 | cultured without DCs (gray dotted line). (D) Allogenic suppression assay of CFSE-labeled Teffs cultured in different ratios with or without sorted <sup>T</sup>regs (CD4+CD127−CD25hiCD45RA−, Figure S7) from DC T cell co-culture described in (A). Proliferation of the CFSE-labeled Teffs with or without activation via αCD3-beads was assessed after 3 d (n = 2 performed in triplicates; mean ± SEM). \*p < 0.05, \*\*p < 0.005, \*\*\*p < 0.001, or \*\*\*\*p < 0.0001, one-way ANOVA with Turkey's posttest or Kruskal-Wallis with Dunn's posttest.

status (**Figure 3B**). Moreover, PSMα2-FITC was also detected in moDCs incubated on ice but reduced by ∼54% (p = 0.0003), indicating that the uptake was not an active process requiring actin remodeling (**Figure 3B**). These observations were confirmed by multispectral imaging flow cytometry, showing PSMα2-FITC located at the membrane when actin-cytoskeleton rearrangement is blocked (ice control) as well as co-localized with OVA most likely in the phagosome when moDCs were incubated at 37◦C regardless of the treatment (**Figure 3C**).

Transient pore formation in the membrane of moDCs mediated by PSMα3 was demonstrated by measuring L-lactate dehydrogenase (LDH) in the supernatant of moDCs treated with PSMα2 various concentrations of PSMα3 (**Figure 3D**). Triton-X, which disrupts the membrane completely as well as δ-Toxin, which is known to form transient pores (23) were used as positive controls (**Figure 3D**). Neither treatment of moDCs with OVA, the peptide-dissolvent DMSO or medium alone had any effects on LDH release (**Figure 3D**), whereas PSMα3 significantly induced LDH release in a concentration-dependent manner. Further, flow cytometric analysis of the moDCs showed that apart from treatment with Triton-X, none of the reagents affected their viability (**Figure S3**). Together the data show that PSMs enter moDCs via transient pore formation without cytolytic effects.

#### Polarization of T Helper1 Cells Is Impaired by PSMs

DCs play a crucial role in the activation and polarization of T cells. Not only the direct interaction, but also the local cytokine milieu is important for the outcome and distinct transcription factors control the differentiation of the T cell subsets (24, 25). To assess the impact of PSMα3-treated moDCs on the polarization of T helper (Th) cell subsets, moDCs were treated with or without Sa lysate or LPS in combination with PSMα3 for 24 h. Subsequently, the cells were co-cultured with naive CD4<sup>+</sup> T cells for 3–4 d. Flow cytometry analysis of the Th subsets showed that TLRL-treated mDCs primed significantly more T-bet<sup>+</sup> Th1 cells whereas mDCs co-treated with PSMα3 prevented Th1 priming (Sa 11.2 vs. 7.8%, p < 0.0001; LPS 8.2 vs. 4.7%, p < 0.0001) (**Figures 4A,B**). In contrast, stimulation of iDCs with PSMα3 alone did not show any difference compared to untreated cells (**Figure 4B**), indicating that PSMα3 without TLRL has no influence on the priming capacity of moDCs. Analyzing Th2 cells by GATA3 expression after co-culture with moDCs treated with TLRL alone or in combination with PSMα3 revealed the same tendency as T-bet expression, but no significant differences (Sa 1.32 vs. 1.1%, p = 0.27; LPS 1.25 vs. 1.1%, p = 0.8) (**Figures 4A,B** and **Figure S6B**). However, no difference in the frequency of RORγt <sup>+</sup>CD4<sup>+</sup> T cells was observed (data not shown).

Cytokine secretion analysis of the major Th1, Th2, and Th17 cytokines in co-culture after 1–3 days showed that IFNγ secretion was completely prevented when TLR4 ligand-treated mDCs were co-treated with PSMα3 (LPS day 1 1.800 vs. 80 pg/ml, p = 0.5617; day 2 5.600 vs. 120 pg/ml, p = 0.0146, day 3 9.400 vs. 330 pg/ml, p < 0.0001), which in part, but not significant, was also seen for TLR2-stimulated mDCs (Sa day 1 60 vs. 40 pg/ml, p>0.999; day 2 6.500 vs. 3.000 pg/ml, p = 0.2534, day 3 7.500 vs. 5,800 pg/ml, p = 0.9105) (**Figure 4C**). Comparable results, but without significance, were observed for IL-17A secretion at day 2 and 3 by TLR4 ligand-treated mDCs together with PSMα3 (LPS day 2 10.8 vs. 5.1 pg/ml, p = 0.9658, day3 24.4 vs. 5.7 pg/ml, p = 0.4336). Similarly, no significant effects of PSMα3 were observed for IL-17A secretion after co-culture of CD4<sup>+</sup> T cells with mDCs treated with Sa lysate. Further, IL-4 secretion showed no difference after treatment with either Sa lysate or LPS, respectively (**Figure 4C**). However, the expression of the closely related cytokine IL-13 significantly increased 3 days after coculture of CD4<sup>+</sup> T cells with moDCs treated with TLRLs and PSMα3 (Sa 500 vs. 1.000 pg/ml, p = 0.0327; LPS 65 vs. 680 pg/ml, p = 0.0057) (**Figure 4C**). This Th cytokine expression is connected with the impaired production of pro-inflammatory cytokines in co-cultures of CD4<sup>+</sup> T cells with moDCs treated with TLRLs and PSMα3 (**Figures S5A–C**).

Thus, PSMα3-treated TLR2- and TLR4- stimulated mDCs decreased the frequency of T-bet<sup>+</sup> Th1 cells, as well as IFN-γ secretion.

#### PSMα3-Treated moDCs Induce Treg Priming Via Direct Cell Interaction and IDO Production

Previously, our group showed that PSMα3 primes mouse tDCs thereby enhancing the frequency and proliferation of Tregs (10, 12). Further, as PSMα3 treatment of moDCs attenuated the priming of Th1 cells, we investigated whether PSMα3 treated moDCs also increased Treg priming. Therefore, moDCs were treated with or without Sa lysate or LPS in combination with PSMα3 for 24 h. Next, the cells were co-cultured with naïve CFSE-labeled CD4<sup>+</sup> T cells for 3–4 d and analyzed by flow cytometry. Newly primed Tregs were characterized as CD4+CD127−CD25hiCD45RA−FoxP3hi induced Tregs [iTregs; (26)]. PSMα3 increased the frequency of iTregs upon co-culture with moDCs after TLR2 or TLR4 ligand treatment (Sa day 3 0.17 vs. 0.31%, p < 0.0033; Sa day 4 0.43 vs. 0.86%, p = 0.0054; LPS day 3 0.08 vs. 0.29%, p < 0.0001; LPS day 4 0.75 vs. 1.07%, p = 0.0204) (**Figures 5A,B**). Treating CD4<sup>+</sup> T cells with PSMα3 in combination with TLRLs without moDCs did not result in CD4<sup>+</sup> T cell activation or priming of iTregs (**Figure S6**).

Further, these iTregs showed a greater proliferation potential than the iTregs primed without PSMα3 (**Figure 5C**). To test the functionality of these iTregs, we sorted iTregs (CD4+CD127−CD25hiCD45RA−) after 4 d of co-culture with LPS + PSMα3-treated moDCs (**Figure S7**). Sorted iTregs were again cultured with CFSE-labeled CD4<sup>+</sup> T cells (Teff) in the presence of anti-CD3/CD28 coated beads to induce Teff proliferation for 4 d. Tregs significantly decreased Teff proliferation analyzed by CFSE-dilution in a concentrationdependent manner (1:1 reduction ∼38%, p = 0.0004; 1:2 reduction ∼25%, p = 0.0176; 1:4 reduction ∼12%, p = 0.2603) (**Figure 5D**), demonstrating that PSMα3-induced iTregs suppress Teff proliferation.

To investigate whether secreted factors from DCs upon PSM-treatment mediate iTreg priming, T cells were co-cultured with iDCs simultaneously adding medium from LPS or LPS with PSMα3-treated moDCs (conditioned medium) (**Figure 6A** condition 2). As described above, PSMα3 increased the frequency of iTregs upon co-culture of naïve CD4<sup>+</sup> T cells with moDCs (**Figure 6A** condition 1 and **Figure 6B**). A similar increase of iTregs was mediated by iDCs incubated with conditioned medium from LPS + PSMα3, whereas LPS conditioned medium had no effect (0.31% compared to 0.09%, respectively, p = 0.0757) **Figure 6A** condition 2 and **Figure 6B**). These observations show that PSM-induced secretion of soluble factors by mDCs modulate iDCs to prime iTregs.

In another assay DCs were washed 24 h after treatment with LPS or LPS + PSMα3 (mDCs) and further cultured for 3–4 days with CD4<sup>+</sup> T cells in fresh medium. These conditions revealed an even higher frequency of iTregs (0.52 or 1.32%, respectively; p < 0.0001) compared to mDCs without medium change (**Figure 6A** condition 3 and **Figure 6B**) and suggest that the interaction of mDCs with CD4<sup>+</sup> T cells as well as secreted factors produced by mDCs after interaction with T cells are important for iTreg priming.

To further address the impact of direct DC-T-cell interaction on Treg priming, mDCs were fixed and either cultured with naïve CD4<sup>+</sup> T cells in fresh or conditioned medium. Fixation of mDCs treated with or without PSMα3 led to low frequencies of iTregs after co-culture with naïve CD4<sup>+</sup> T cells in fresh medium (0.08% vs. 0.2%, p = 0.6038) (**Figure 6A** condition 4 and **Figure 6B**). Similar results were obtained by the addition of conditioned medium to this co-culture showing that mainly direct interaction of mDCs with naïve CD4<sup>+</sup> T cells is essential for iTreg induction. However, in this experimental setting we cannot exclude that soluble factors produced by mDCs after DC-T cell interaction are involved in iTreg priming. To address this issue, culture supernatants of mDCs with naïve CD4+T cells were analyzed over time for soluble factors, which were shown to be responsible for iTreg priming and proliferation, like TGF-β, IL-10, IL-2, CD40L and IDO. Neither, IL-10 (**Figure 6C**), TGF-ß (**Figure 6D**), CD40L nor IL-2 (**Figure S5** were increased in the co-culture of PSMα3-treated mDCs with naïve CD4<sup>+</sup> T cells compared to mDCs independently of the TLRL used. Moreover, T helper cell priming cytokines like IL-4, IL-6, IL-12p70, IL17A and IFN-γ were hardly detectable in the co-culture supernatants of LPS + PSMα3 treated mDCs with CD4<sup>+</sup> T cells (**Figure 4C** and **Figure S5**) as described above.

In contrast, secretion of the enzyme IDO, which is important for Treg differentiation from naïve T cells (8), was significantly increased by mDCs in response to PSMα3 treatment after 1 (11.000 vs. 1.440 pg/ml, p < 0.0001) and 2 days (42.000 vs. 33.800 pg/ml, p = 0.0006) compared to mDCs alone (**Figure 6E**). As previously described, LPS treatment alone also led to IDO secretion by moDCs (27). We addressed the impact of IDO on iTreg priming by mDCs using the specific IDO inhibitor 1-Methyl-D-Tryptophan (1-DMT) prior to treatment of iDCs with LPS or LPS + PSMα3. These mDCs were then co-cultured with CD4<sup>+</sup> T cells and the frequency of iTregs determined as described above. Surprisingly, IDO inhibition revealed increased frequencies of iTregs by 2 fold independently whether cells were treated with LPS + PSMα3 or the inhibitor alone (see **Figure S8**). No difference in the proliferation of CD4<sup>+</sup> T cells was observed upon IDO inhibition. Moreover, IDO inhibition had hardly any effect on the frequency of Th1, Th2 and Th17 cells in the culture (data not shown). Thus, the increased IDO production by PSMtreated DCs seems not to be essential for iTreg induction.

Together, these data indicate that PSMα3 modulates moDCs to prime predominantly Tregs via mechanisms involving mainly direct DC-T cell interaction in combination with DC-secreted yet unknown factors.

#### PSMα3-Treated moDCs Induce Tregs From CD4<sup>+</sup> T Cells of Patients With Autoimmune Diseases

In order to address whether PSMα3 can be used for therapeutic approaches by modulating moDCs for iTreg priming in a setting of T cell associated autoimmune diseases, moDCs from healthy donors were co-cultured after treatment with LPS or LPS + PSMα3 with CD4<sup>+</sup> T cells from patients with spondyloarthritis (**Figure 7A**) and rheumatoid arthritis (RA) (data not shown). These patients suffer from spondylitis/enthesitis of the spine or peripheral arthritis with pain, morning stiffness and consecutive ankylosing of the spine and/or joint destruction. These diseases display a high frequency of pro-inflammatory Th1 and Th17 cells. As shown for T cells from healthy donors, PSMα3 increased the frequency of iTregs upon co-culture with LPS-treated mDCs by > 10-fold compared to LPS treatment alone (p = 0.024), whereas no effects were observed for Th1, Th2, and Th17 priming of T cells from patients with spondylitis (**Figure 7A**). Moreover, similar, but not significant results (p = 0.0919) could be observed in an autologous setting using mDCs and CD4<sup>+</sup> T cells from patients with spondyloarthritis (**Figure 7B** and **Figure S9**). This shows that PSMα3 indeed modulates moDCs to prime iTregs even in allogenic and autologous disease settings, indicating its potential for DC therapy in chronic inflammatory diseases.

# DISCUSSION

There is a need for particular therapeutic approaches preventing or inhibiting immune activation in autoimmune diseases, allograft rejection, allergies, asthma and various forms of hypersensitivity. Current therapies, which mainly use nonspecific systemic immunosuppressants, are associated with severe side effects. Thus, ex vivo generated tDCs are an attractive alternative to enhance, maintain or restore immunological tolerance.

Here, we addressed the possibility of PSMα3, which is secreted by highly virulent CA-MRSA strains, to induce human tDCs as potential therapeutic for DC vaccination strategies. PSMs form transient pores into the membrane of neutrophils (15), and DCs thereby getting access to the cytosol, in the case of DCs without cell lysis (**Figure 3**) (10). Molecularly, PSMα3 enhanced the activation of the p38-CREB pathway upon TLR ligation, which in consequence diminished pro-inflammatory cytokine production but induced IL-10 secretion by mouse bone marrow-derived DCs (10). Here, we show that PSMα2 enters human moDCs via endocytosis and transient pore formation and is located in the cytosol as well as close to the membrane. Indeed, NF-κB as well as p38 activation are necessary for DC maturation including upregulation of CD80, CD86, and CD40, but also cytokine production (28–30). We believe that PSMα3 prevents TLRactivation either extracellularly and/or intracellularly thereby inducing tDCs. Direct extracellular interaction of PSMα1–3 with TLR4 was shown to prevent binding of HMGB1 to TLR4 and thus downstream activation of NF-κB (31). Similarly, we here show that PSMα3 prevents activation of NF-κB as well as p38 MAPK signaling. Whether PSMα3 also blocks binding of LPS to TLR4 remains to be shown. Moreover, our findings are in agreement with several studies showing that NF-κB inhibition favors an immature or tolerogenic DC phenotype, which stimulates the expansion of Foxp3 expressing regulatory T cells (32–35).

A hallmark of tDCs is their immature phenotype characterized by low surface levels of MHC class II and costimulatory molecules, such as CD86, CD40, CD54, and PD-L2, but increased expression of TLRs, chemokine receptors and PD-L1. In vivo studies showed that the tissue or even tumor microenvironment is important for regulating the development and function of tDCs (36). Although PSMα3 had little influence on TLR–induced upregulation of HLA-DR, CD83, CD86, and PD-L2, up-regulation of costimulatory molecules CD80, PD-L1, and CD40 was inhibited, preventing full DC-maturation. Likewise, cholera toxin in combination with LPS induced CD80 and CD86 but reduced CD40 and CD54 expression by DCs (37). Mechanistically, the inhibition of TNFα production, which among others regulates CD40 expression, upon synergistic treatment with TLRLs and PSMα3 could explain the impaired up-regulation of CD40 (38). This fits to the fact that the lack of CD40 expression on DCs was shown to be important for the generation of Tregs while suppressing primary immune responses (39). Moreover, some studies have shown that phenotypically mature DCs are also able to promote Tregs and act superior in activating their suppressor function (40, 41).

In addition, PSMα3 dramatically changed the cytokine secretion pattern of moDCs upon TLRL treatment by preventing the secretion of the pro-inflammatory cytokines TNF, IL-12, but also of anti-inflammatory IL-10, which is produced by DCs in response to TLR-stimulation. This is in contrast to the data observed by mouse bone-marrow derived DC, where IL-10 secretion is even increased upon PSM and TLRL treatment compared to TLRL treatment alone (12). However, PSM-treated moDCs still impair Th1 but promote Treg priming. Tolerogenic DCs are defined by their capacity to induce Tregs via production of anti-inflammatory molecules that may be secreted, membrane bound, or both. A variety of studies demonstrated the necessity of IL-10 secretion by tDCs for Treg induction (42–45) and for the maintenance of suppressive Tregs upon strong inflammatory

medium from LPS or LPS + PSMα3 treated DCs, with (3) washed and re-seeded mDCs together with new medium or (4) with fixed moDCs together with either new medium or conditioned medium. (B) The bar graphs show the frequency of iTregs from CD4<sup>+</sup> T cells analyzed by flow cytometry from the experiments described in (A) (one representative of n ≥ 3 independent experiments performed in triplicates). (C,D) Cell culture supernatants from the DC T cell co-culture (n ≥ 3 performed in triplicates; mean ± SEM) were analyzed for IL-10 (C) and TGF-β production (D) after 1 d, 2 d or 3 d by a bead-based immunoassay. (E) Cell culture supernatants (n = 2 performed in triplicates; mean ± SEM) from moDCs treated with Sa lysate or LPS with or without PSMα3 were analyzed for IDO production by sandwich ELISA. \*p < 0.05, \*\*\*p < 0.001 or \*\*\*\*p < 0.0001, one-way ANOVA with Turkey's posttest or Kruskal-Wallis with Dunn's posttest.

signals (12, 44, 46, 47). The data described herein raise the question whether IL-10 secretion is a hallmark of tDCs.

As T-cell differentiation is mainly controlled by cytokines mediating polarizing signals (48) we addressed other factors as mediators for Treg induction. IDO, an immune-regulatory enzyme, which is mainly expressed in DCs, was shown to modulate adaptive immune responses by promoting immunesuppression and tolerance (49–51). IDO expression in DCs is induced either by IFN-γ or by TLR activation via the non-canonical NF- κB pathway. IDO acts through tryptophan (TRP) depletion and production of TRP metabolites thereby inducing differentiation of new Tregs from naïve T cells (27, 51– 55). PSM treatment of moDCs resulted in an even enhanced IDO expression as compared to Sa lysate or LPS treatment alone. The latter was recently shown to induce IDO together with the transcription factor aryl-hydrocarbon receptor (AhR) (27). However, our results obtained with IDO inhibition by 1-DMT argue against a mechanistically role of IDO in PSMmediated Treg induction. Thus, it is tempting to speculate that PSMα3 modulated moDCs prime Tregs via mechanisms involving

or 4 d. The different T cell subsets were analyzed by flow cytometry. The graph shows the frequency of the T-bet<sup>+</sup> Th1, GATA3<sup>+</sup> Th2, RORγt <sup>+</sup> Th17 cells and CD127−CD25hiCD45RA−Foxp3hi iTregs from CD4<sup>+</sup> T cells (<sup>n</sup> <sup>=</sup> 6 patients; mean <sup>±</sup> SEM). \*<sup>p</sup> <sup>&</sup>lt; 0.05, one-way ANOVA with Turkey's posttest or Kruskal-Wallis with Dunn's posttest. (B) Autologous T-cell assay: MoDCs from spondyloarthritis patients were treated as described in (A) and co-cultured with CFSE-labeled CD4<sup>+</sup> T cells from the same patient. CD127−CD25hiCD45RA−Foxp3hi iTregs were analyzed by flow cytometry. The graph shows the frequency of iTregs as fold change from LPS-treated cells of 4 different spondyloarthritis patients. Data were analyzed by unpaired student's T-test.

predominantly direct DC-T cell interaction in combination with the absence of Th-priming cytokines and probably yet unknown DC-secreted factors.

The data from this study are a proof of concept of the potential use of Sa-derived PSMα3 to induce tolerogenic human DCs with the ability to prime iTregs. Especially, when using PSMα3 induced tDCs to prime iTregs in allogenic and autologous settings of CD4<sup>+</sup> T cells from spondylitis patients. It is believed that tolerogenic DCs may induce tolerance to the pathologic immune responses in a patient without affecting the immune defense against pathogens or tumors. There are two strategies to restore tolerance in autoimmunity: improve the induction and function of tolerogenic DC or generating tolerogenic DC in vitro for subsequent administration in vivo as cell therapy (13, 56, 57). Different immune-modulatory agents such as dexamethasone, vitamin D3, TNF or IL-10, but also pathogen-derived products have been used in order to modify the phenotype, cytokine profiles and activity of DCs (58–65). Pre-clinical models of arthritis (65), EAE (66), and type 1 diabetes [T1D; (67)] have demonstrated the efficacy of in vitro induced tolerogenic DCbased cell therapy.

There is a need for the complete understanding of the mechanisms that control tolerance and immunity in the context of the complexity and heterogeneity of autoimmune diseases, in which multiple cell types are affected and various genetic backgrounds are involved.

#### AUTHOR CONTRIBUTIONS

SA, JR, and NA contributed to the conception and design of the study. JR and MG performed the experiments and statistical analysis. SA and JR wrote the first draft of the manuscript. SA, JR, NA, MG, and JH wrote sections of the manuscript. All authors read and approved the submitted version of the manuscript.

#### FUNDING

This work was financed by the German Research Foundation grant SFB685-TP3 and the European Social Fund of Baden-Württemberg (Margarete von Wrangell Program) to SA.

#### ACKNOWLEDGMENTS

We thank Johannes Morschl for technical assistance and Peter Richardson for critical reading of the manuscript. Cell sorting and flow cytometry sample acquisition was done on shared instruments of the Flow Cytometry Core Facility Tübingen. Imaging flow cytometry sample acquisition, and data analysis was done on shared instruments of the Imaging Flow Cytometry Core Facility Tübingen. The Imaging Flow Cytometry Core Facility was supported by a grant from the Ministry of Science, Research and Arts of Baden Württemberg (Az.: SI-BW 01222-91) and the Deutsche Forschungsgemeinschaft DFG (German Research Foundation) (Az.: INST 2388/33-1).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2018.02603/full#supplementary-material

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Richardson, Armbruster, Günter, Henes and Autenrieth. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Current State of Dendritic Cell-Based Immunotherapy: Opportunities for in vitro Antigen Loading of Different DC Subsets?

Anne Huber <sup>1</sup> , Floris Dammeijer 1,2, Joachim G. J. V. Aerts 1,2 and Heleen Vroman1,2 \*

<sup>1</sup> Department of Pulmonary Medicine, Erasmus Medical Center, Rotterdam, Netherlands, <sup>2</sup> Erasmus Cancer Institute, Erasmus Medical Center, Rotterdam, Netherlands

Dendritic cell (DC) based cancer immunotherapy aims at the activation of the immune system, and in particular tumor-specific cytotoxic T lymphocytes (CTLs) to eradicate the tumor. DCs represent a heterogeneous cell population, including conventional DCs (cDCs), consisting of cDC1s, cDC2s, plasmacytoid DCs (pDCs), and monocyte-derived DCs (moDCs). These DC subsets differ both in ontogeny and functional properties, such as the capacity to induce CD4<sup>+</sup> and CD8<sup>+</sup> T-cell activation. MoDCs are most frequently used for vaccination purposes, based on technical aspects such as availability and in vitro expansion. However, whether moDCs are superior over other DC subsets in inducing anti-tumor immune responses, is unknown, and likely depends on tumor type and composition of the tumor microenvironment. In this review, we discuss cellular aspects essential for DC vaccination efficacy, and the most recent findings on different DC subsets that could be used for DC-based cancer immunotherapy. This can prove valuable for the future design of more effective DC vaccines by choosing different DC subsets, and sheds light on the working mechanism of DC immunotherapy.

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Katsuaki Sato, University of Miyazaki, Japan Richard A. Kroczek, Robert Koch Institute, Germany

> \*Correspondence: Heleen Vroman h.vroman@erasmusmc.nl

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 16 July 2018 Accepted: 14 November 2018 Published: 03 December 2018

#### Citation:

Huber A, Dammeijer F, Aerts JGJV and Vroman H (2018) Current State of Dendritic Cell-Based Immunotherapy: Opportunities for in vitro Antigen Loading of Different DC Subsets? Front. Immunol. 9:2804. doi: 10.3389/fimmu.2018.02804

Keywords: dendritic cell (DC), Immunotherapy, DC subsets, T cell responses, tumor immunology

# INTRODUCTION

The immune system is able to distinguish between self, non-self and eliminate damaged cells. Consequently, it has the potential to eradicate cancerous cells displaying mutated, or aberrantly expressed self-antigens. To avoid elimination by immune responses, tumors not only acquire the ability to prevent immune recognition, but also create an immunosuppressive environment and actively hijack immune cells to aid in tumor progression (1, 2). Re-activating the immune system to treat patients with cancer was already proposed at the end of the nineteenth century and cancer immunotherapy has further developed ever since (3–6). One type of immunotherapy is dendritic cell (DC) vaccination (7). DC vaccination makes use of autologous DCs loaded ex-vivo with specific tumor-associated antigens (TAAs) or whole tumor lysate to generate an immune response aiming for cancer-cell elimination. DC vaccination using ex-vivo generated monocyte-derived DCs (moDCs) in patients with cancer was first explored over two decades ago (8). Numerous clinical trials [over 200 (9)] have established the safety and ability of moDC vaccines to induce anti-tumor responses (10–12). More recently, also in vivo loading of DCs is being exploited (13–17). In this review, we will discuss the cellular aspects essential for DC vaccination efficacy, the potential of distinct DC subsets as sources for DC vaccination, and the implications for the future design of DC vaccines.

#### DENDRITIC CELLS

DCs play a crucial role in the immune system and link innate and adaptive immune responses (18–21). They arise from progenitor cells in the bone marrow and reside in peripheral tissues in an immature state. Immature DCs (iDCs) are specialized in antigen capturing, processing, and presentation. Upon appropriate stimulation mediated by inflammatory and pathogen-derived signals, iDCs undergo maturation. Mature DCs express co-stimulatory molecules, secrete cytokines, and migrate to lymphoid organs where they activate antigen-specific T-cells (22). Besides the presentation of exogenous antigens on MHC-II peptides, DCs are able to cross-present exogenously captured antigens on MHC I-associated peptides (23). Thereby, DCs can present TAAs to CD8<sup>+</sup> T-cells which makes them of particular interest for cancer immunotherapy (24).

DCs consist of developmentally and functionally distinct DC subsets. These include moDCs, conventional DCs—consisting of cDC1s and cDC2s—and plasmacytoid DCs (pDCs) (25– 27). While moDCs are derived from the common monocyte progenitors (cMoPs), cDCs, and pDCs arise from a common DC precursor (27–29). Each DC subset has specialized functions however, these are not exclusive and seem to depend on both location and environmental cues (30). In general, moDCs efficiently promote T-cell differentiation, but are poor inducers of CD4<sup>+</sup> T-cell proliferation (31). In contrast, moDCs can be powerful activators of tumor-specific CD8<sup>+</sup> T-cells (32). It is known that mature moDCs secrete chemokines and pro-inflammatory cytokines which are crucial to attract other immune cells and T-cells to the local environment (33). cDC1s are specialized in recognizing viral and intracellular antigens and are important for cytotoxic T-cell (CTL) responses, whereas cDC2s are particularly apt in priming CD4<sup>+</sup> T-cells (34). Depending on the experimental model, cDC2s induce T-helper (Th) 2 or Th17 responses (35, 36). pDCs are prominent producers of type I interferon in response to single-stranded RNA and double-stranded DNA upon e.g., viral infections, which is important for DC maturation and CD8<sup>+</sup> T-cell activation (34, 37). However, their antigen-presenting capacity is being questioned, especially as it was recently discovered that pDC characterized by CD123 expression and BDCA2 are contaminated by pre-cDCs (38, 39).

#### DC VACCINES

DC-based cancer immunotherapy depends on the crucial role that DCs play in inducing antigen-specific T-cell responses (40). In many tumors, immune responses are ineffective due to the immunosuppressive environment of the tumor and/or the lack of immunogenicity of the tumor (41, 42). In addition, the tumor microenvironment (TME) promotes exhaustion of effector CD8<sup>+</sup> T-cells (43). Some tumors are even able to hamper the recruitment of cDC1s, by downregulating CCL4 signaling upon constitutively active β-catenin signaling and thereby hamper priming and accumulation of tumor-infiltrating T-cells (44), indicating the importance of endogenous DCs for initiating antitumor immunity. DC vaccines aim to overcome the absence or malfunctioning of endogenous DCs by manipulating autologous DCs to enhance T-cell responses directed against the tumor.

Currently a wide range of procedures to generate autologous DCs exist using distinct sources, such as peripheral blood monocytes, naturally occuring DCs, or CD34<sup>+</sup> hematopoietic precursor cells mobilized from the bone marrow (10), enabling the generations of various DC subsets [such as moDCs, cDCs, or pDCs (45–47)]. In addition, different sources of TAAs [e.g., cancer cell line lysate, whole tumor lysate, or tumorassociated peptides (45, 48, 49)], as well as different antigenloading methods [such as pulsing, via viral vectors, or mRNA transfection (10)] are used to load DCs. Moreover, various maturation methods including cytokines, CD40 ligands, and TLR agonists (50) are known. Currently, there is a great effort made in improving existing DC vaccines and developing new ones. New approaches include genetically engineered DCs that express TAAs or display enhanced immunostimulatory properties or explore in vivo antigen loading of DCs with freshly released TAAs due to chemotherapy or immunogenic tumor-cell death (51–58).

#### GENERATION OF PATIENT-DERIVED DCs EX VIVO

Because DCs comprise <1% of peripheral blood mononuclear cells (PBMCs), one major challenge is the generation of sufficient numbers of DCs for vaccination purposes. Therefore, DC vaccination studies frequently used moDCs that can be generated ex-vivo in large numbers from purified monocytes that were consequently cultured with granulocyte-macrophage colony-stimulating factor (GM-CSF) and interleukin (IL)-4 (59). Recently, it was described that monocytes cultured with GM-CSF and IL-6, and activated with IFN-γ, give rise to a newly described mo-cDC1s population that has similarities to cDC1s (60). In addition, cDCs and pDCs can be generated from CD34<sup>+</sup> hematopoietic stem cells using fms-like tyrosine kinase 3 ligand (Flt3L) (61, 62).

The phenotype, function and ability to induce T-cell responses by in vitro generated DCs is highly dependent on the culture methods used (63). For instance, culturing human monocytes with CD137 protein generates DCs potent in inducing CD8<sup>+</sup> T-cells with superior lysing capabilities against cells infected with cancer-causing viruses (64, 65). Comparing different technologies for monocyte isolation demonstrated that isolation techniques can also influence the antitumor immunogenicity and cytokine production of the generated moDCs (66, 67). Furthermore, the cytokines and growth factors required for precursor-cell differentiation into DCs and subsequent activation influence DC function, and in consequence, the effectivity of DC vaccines (68–71).

#### LYMPH NODE HOMING OF VACCINATED DCs

To activate antigen-specific T-cell responses, DCs need to reach the lymph nodes (LNs) in order to present antigen to cognate TAA-specific T-cells. In order to optimize DC-trafficking to the LN, various injection routes and strategies have been explored. In a pre-clinical mouse study, different vaccination routes were compared to load DCs in vivo with naked antigenencoded RNA. Herein it was shown that only intra-nodal (i.n) vaccination induced potent expansion of antigen-specific Tcells resulting in prolonged survival, which was not observed upon intra-dermal (i.d.), subcutaneous, or near nodal vaccination (72), indicating the superiority of i.n. vaccination. However, in various clinical studies superior efficacy of i.n. vaccination was less clear. In one study, moDCs pulsed with three melanoma peptides were administered either i.d. or i.n. to 25 patients with metastatic melanoma. After i.d. administration, 4% of DCs migrated to the LNs, whereas migration upon i.n. injection varied between 0 and 56%. The total number of vaccinated moDCs in single LNs were 10- to 30-fold higher after i.n. administration than i.d. injection. However, surprisingly, there was no difference in the strength of the immune response evaluated by TAA-specific CD8<sup>+</sup> T-cells isolated from DTH reactions between the two administration routes (73). Another study in 54 patients with different types of HER2<sup>+</sup> breast cancer employed moDCs loaded with six HER2 MHC class II binding peptides injected intralesionally, i.n. or both. More than 80% of the patients had new or increased systemic anti-HER2 CD4<sup>+</sup> or CD8<sup>+</sup> T-cell responses and 32 patients had a HER2-specific CD4<sup>+</sup> T-cell response in the sentinel LN (SLN) after vaccination but these were not significantly different between the three administration routes (74). The large variation observed upon i.n. vaccination also stress the difficulty of i.n. vaccination over i.d. vaccination, and could indicate that accurate i.n. vaccination outperforms i.d. vaccination. It has also been shown that migration to the LNs upon i.d. vaccination can be improved by pre-treating the vaccination site with a potent recall antigen, as tetanus/diphtheria (Td) toxoid pretreatment. This improved DC migration to the LNs, progression free survival and overall survival in patients with glioblastoma (75). Strikingly, systemic TAA-specific immune responses and enhanced tumor CD8<sup>+</sup> T-cell infiltration were even observed upon intra-tumoral injection of DCs containing an vector expressing the CCL21 gene in 16 patients with advanced non-small cell lung carcinoma (NSCLC) (54).

Therefore, the superior route or site of injection is still unknown, as no differences were found in safety or antigenspecific immune responses upon either intradermal or -nodal injection (73, 74). These results further urge the need to compare DC vaccination efficacy between different administration routes.

#### EVALUATION OF EFFICACY OF EX VIVO GENERATED moDC VACCINES

As the molecular underpinnings of an effective DC-therapy induced T-cell response are still incompletely understood, it has been difficult to identify factors associated with therapeutic success. As the location and mechanism of T-cell immune responses initiated upon DC therapy is unknown, there is also no consensus how DC vaccination efficacy should be evaluated. One effort to generalize the monitoring of effectivity is by the Response Evaluation Criteria in Solid Tumors (RECIST) or by the more recently described modified RECIST, which enables categorization of patient responses into complete response, partial response, stable disease and progressive disease determined by the amount of tumor shrinkage of a given number of tumor lesions, disease progression, and assessment of pathological LNs (76, 77). Nevertheless, various studies monitored response differently and focused on either clinical responses (summarized in **Table 1**) or different aspects of the immune response. Moreover, most studies failed to find significant correlations of measured immune characteristics and clinical outcome.

A phase I clinical trial employed autologous tumor lysatepulsed moDCs in ten patients with malignant mesothelioma after chemotherapy. Clinical responses were evaluated by modified RECIST. In addition, efficacy of DC vaccination was determined by increased cytotoxicity of isolated PBMCs against tumor cells and higher percentages of CD8<sup>+</sup> T-cells expressing granzyme B, an indication for their capacity to lyse cells. After vaccination, four out of six patients showed increased cytotoxicity levels and granzyme B expressing CD8<sup>+</sup> T-cells increased in nine patients (45). In another phase I clinical trial in nine patients with mesothelioma using allogeneic tumor cell lysate-pulsed moDCs, tumor-specific T-cells could be detected in the majority of patients in a skin biopsy after a positive DTH skin test. In addition, radiographic responses (two partial responses and seven patients with stable disease), progression free survival (8.8 months) and overall survival [(OS) not reached] of the patients were monitored and analyzed according to modified RECIST criteria (78). During one study in 27 prostate cancer patients with rising serum prostate-specific antigen [(PSA); indication for biochemical relapse of prostate cancer] levels, kinetics of PSA was monitored and used to determine the efficacy of the vaccination with moDC pulsed with allogeneic tumor cell lysate. The median PSA doubling time (PSADT), which determines clinical outcome, increased from 5.67 to 18.85 months. In addition, the frequency of PSA-specific T-cells increased after vaccination and tumor-specific IgG antibodies could be detected in nine patients. However, these immune response characteristics did not significantly correlate with PSADT (48). A recent phase I clinical trial in patients with NSCLC employed moDCs pulsed with two TAAs, silenced with SOCS1, and stimulated with flagellin. Upon vaccination, regulatory T-cells (Tregs) decreased, and three patients had increased levels of IL-6 and/or TNFα, whereas IL-2, IL-4, IL-10, and IFNγ were unaffected. These observed immune responses did not correlate with the clinical response (49). Another phase II trial in 156 patients with hepatocellular carcinoma (no residual tumor after standard treatment) investigated DC-based adjuvant immunotherapy using triple TAA-pulsed moDCs. While recurrence-free survival (RFS) and OS were not different between the immunotherapy and control (no treatment) groups, immunotherapy increased TAA-reactive T-cell responses and IFNγ levels, whereas levels of serum TGF-β decreased. Nevertheless, this did not correlate to RFS. Interestingly, when radiofrequency ablation (RFA) patients were excluded in post-hoc analyses, immunotherapy did prolong RFS of non-RFA patients (79).



AML, acute myeloid leukemia; cDC2s, conventional DCs 2; CT, computed tomography; i.d, intra-dermal; i.n.,intra-nodal; i.v., intra-venously; moDCs, monocyte-derived DCs; MPM, malignant pleural mesothelioma; NR, no response; NSCLC, non-small cell lung carcinoma; PR, partial response; RECIST, Response Evaluation Criteria in Solid Tumors; OS, overall survival; pDCs, plasmacytoid DCs; PFS, progression-free survival; PSADT, prostate-specific antigen doubling time; RFS, recurrence-free survival; SD, stable disease; TAA, tumor-associated antigen; s.c. subcutaneous.

In contrast, a phase II study in 30 patients with acute myeloid leukemia could correlate long-term OS with higher numbers of circulating TAA-specific CD8<sup>+</sup> T-cells after therapy with moDCs electroporated with TAA-mRNA (80). Furthermore, a phase I/II clinical trial that studied the effectivity of DC vaccination in 62 patients with melanoma used moDCs loaded with 4 HLA class I peptides and 6 HLA class II peptides. DC vaccination increased the numbers of vaccines-specific

IFNγ-producing T-cells, whereas numbers of Tregs and myeloid derived suppressor cells (MDSCs) were unaltered. Surprisingly, IFNγ-producing T-cells did not correlate with OS, whereas the intensity of allergic vaccine-injection site reactions significantly correlated with OS. Furthermore, a maximal eosinophilic blood count (>250 per 100 µl blood) significantly improved survival specifically in tumor bearing melanoma patients (81). Another study in 42 patients with HER2<sup>+</sup> breast cancer, that used moDCs pulsed with six HER2 MHC class II binding peptides, could correlate pathologic complete response with the CD4<sup>+</sup> Th1 immune response in the sentinel LN, but in peripheral blood (74).

Overall, it seems that DC vaccination induced various immune responses, but most of the observed immunological responses do not reflect clinical responses (**Figure 1**). This could be due to the fact that most studies are phase I/II clinical trials in which safety and feasibility are the primary outcomes and not efficacy. Furthermore, this could be caused by the type and location of the immune response measured, as most studies focused on TAA-specific T-cells in peripheral blood. As DC vaccination initiates T-cell responses in the LNs and these TAA-specific T-cells exert their cytolytic function in the tumor, it would be more likely that immune responses in LNs or in the tumor predict OS better than immune responses measured in peripheral blood. This could be performed using a recently described method that can quantify tumor-specific CTLs in preclinical models at different sites (82).

Furthermore, it was shown in murine models that DC vaccines elicited cytotoxic and regulatory natural killer cell responses against tumors (83, 84). This stresses the necessity to investigate other cell subsets, besides T-cells, influenced by DC vaccines.

#### POTENTIAL OF NATURALLY OCCURRING DC SUBSETS FOR USE AS VACCINES

Despite the growing knowledge in DC immunobiology, the exact diversity and biology of T-cell responses generated by different DC vaccines is still poorly understood. The recent development of antibody-coated magnetic beads enables the isolation of natural occurring DC subsets directly from peripheral blood in considerate numbers. For example, more than 10 million pDCs or more than 27 million cDCs can be isolated from apheresis products (46, 47).

The first phase I/II clinical trials have been performed using naturally occurring DCs for DC therapy and have shown that this is safe and feasible (46, 85). One of the clinical trials that used naturally occurring cDC2s loaded with three TAAs in 14 melanoma patients showed that the presence of TAAspecific T-cells in peripheral blood and DTH tests correlated with progression-free survival in three patients (47). Another clinical study in 15 patients with metastatic melanoma used pDCs pulsed with three TAAs. Increased TAA-specific CD8<sup>+</sup> T-cell frequencies were measured in the blood of seven of the fifteen patients. Clinical outcome (PFS and OS) of patients treated with TAA-loaded pDCs was increased as compared to 72 matched control patients treated with chemotherapy (46).

Unfortunately, it is unknown whether naturally occurring DCs outperform cultured moDCs as source for DC therapy in patients, as clinical trials comparing different DC subsets as a source for DC therapy have not been performed. However, in mice, efficacy of different DC subsets for DC-therapy was compared. Herein, they found that moDCs in the tumor are superior in antigen uptake and processing but failed to induce efficient T-cell proliferation. MoDCs in the tumor even seemed to have immunosuppressive properties, as they inhibited T-cell proliferation by increased iNOS expression (86), however this is likely dependent on environmental cues, as cultured moDCs are highly immunogenic. Tumoral cDC1s were superior in stimulating naïve and previously activated CD8<sup>+</sup> T-cells, beneficial for tumors with abundant Tregs, whereas cDC2s purified from tumor were more efficient in CD4<sup>+</sup> T-cell stimulation and differentiation into Th17 cells, which was effective for tumors with abundant M2-oriented tumorpromoting tumor-associated macrophages (TAMs) (86, 87). In another study of melanoma mouse models, cDC1s but not cDC2s were shown to transport intact TAAs to TdLNs and cross-present them to CD8<sup>+</sup> T-cells (88). Whether these findings will be confirmed with ex vivo loading of natural occurring DCs remains to be determined, and is currently extensively studied.

#### IMPLICATIONS FOR FUTURE DESIGN OF DENDRITIC CELL VACCINES

The use of different natural occurring DC subsets for vaccination is promising and more studies directly comparing the various

#### REFERENCES


subsets are urgently needed. In addition, more research into the contribution of the DC subsets to the different aspects of antitumor immunity is required, as this can be beneficial for tumors with different composition of the TME.

It is known that different types of human solid tumors are infiltrated to various extents by different types of immune cells (89–91). The presence of these immune infiltrates even has prognostic value (92–94). Moreover, it might guide the choice of which DC type to employ for vaccination, as different DC subsets elicit differing T-cell responses against the tumor. Hence, identifying whether the immunosuppressive environment of the tumor consists Tregs or TAMs before treatment might help in choosing the right DC subset to induce the proper T-cell skewing.

Besides the direct (re)activation of tumor-specific Tcells, efforts are undertaken to combine DC vaccination with agents that can modulate the TME itself e.g., by immunotherapy, radiotherapy, or chemotherapy to act synergistically with DC vaccination, which can improve immunogenicity, T-cell infiltration, T-cell exhaustion, and overcome the immunosuppressive environment of the tumor (82, 95–98).

#### CONCLUSION REMARKS

Although DC vaccination has been optimized in recent years, a great potential for improvement still remains. More (pre)clinical studies investigating the working mechanisms underlying DC vaccine efficacy are required. Therein, a major focus should be laid on different DC (and other myeloid) subpopulations and their specialized contribution to antitumor immunity, as it is likely that different cancer types might need different DC therapeutic strategies.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.


prophylactic and therapeutic antitumoral immunity. Cancer Res. (2010) 70:9031–40. doi: 10.1158/0008-5472.CAN-10-0699


potent strategy for enhancing the antitumor effect of therapy composed of cyclophosphamide and dendritic cells. Front Immunol. (2017) 8:713. doi: 10.3389/fimmu.2017.00713

98. Inogés S, Tejada S, de Cerio AL-D, Pérez-Larraya JG, Espinós J, Idoate MA, et al. A phase II trial of autologous dendritic cell vaccination and radiochemotherapy following fluorescence-guided surgery in newly diagnosed glioblastoma patients. J Transl Med. (2017) 15:104. doi: 10.1186/s12967-017-1202-z

**Conflict of Interest Statement:** JA: speakers fee and consultancy Eli-Lilly, Boehringer Ingelheim, MSD, BMS, Astra Zeneca, Amphera, Roche; Stock owner Amphera b.v.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Huber, Dammeijer, Aerts and Vroman. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Novel Cre-Expressing Mouse Strains Permitting to Selectively Track and Edit Type 1 Conventional Dendritic Cells Facilitate Disentangling Their Complexity in vivo

Raphaël Mattiuz <sup>1</sup> , Christian Wohn<sup>1</sup> , Sonia Ghilas <sup>1</sup> , Marc Ambrosini <sup>1</sup> , Yannick O. Alexandre<sup>1</sup> , Cindy Sanchez <sup>1</sup> , Anissa Fries <sup>1</sup> , Thien-Phong Vu Manh<sup>1</sup> , Bernard Malissen1,2, Marc Dalod<sup>1</sup> \* and Karine Crozat <sup>1</sup> \*

<sup>1</sup> Centre d'Immunologie de Marseille-Luminy, Turing Center for Living Systems, CNRS, INSERM, Aix Marseille Univ, Marseille, France, <sup>2</sup> Centre d'Immunophénomique, Aix Marseille Univ, CNRS, INSERM, Marseille, France

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Natalio Garbi, Universität Bonn, Germany Christophe Jean Desmet, University of Liege, Belgium

#### \*Correspondence:

Marc Dalod dalod@ciml.univ-mrs.fr Karine Crozat crozat@ciml.univ-mrs.fr

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 29 July 2018 Accepted: 14 November 2018 Published: 04 December 2018

#### Citation:

Mattiuz R, Wohn C, Ghilas S, Ambrosini M, Alexandre YO, Sanchez C, Fries A, Vu Manh T-P, Malissen B, Dalod M and Crozat K (2018) Novel Cre-Expressing Mouse Strains Permitting to Selectively Track and Edit Type 1 Conventional Dendritic Cells Facilitate Disentangling Their Complexity in vivo. Front. Immunol. 9:2805. doi: 10.3389/fimmu.2018.02805 Type 1 conventional DCs (cDC1) excel in the cross-priming of CD8<sup>+</sup> T cells, which is crucial for orchestrating efficient immune responses against viruses or tumors. However, our understanding of their physiological functions and molecular regulation has been limited by the lack of proper mutant mouse models allowing their conditional genetic targeting. Because the Xcr1 and A530099j19rik (Karma/Gpr141b) genes belong to the core transcriptomic fingerprint of mouse cDC1, we used them to engineer two novel Cre-driver lines, the Xcr1Cre and KarmaCre mice, by knocking in an IRES-Cre expression cassette into their 3′ -UTR. We used genetic tracing to characterize the specificity and efficiency of these new models in several lymphoid and non-lymphoid tissues, and compared them to the Clec9aCre mouse model, which targets the immediate precursors of cDCs. Amongst the three Cre-driver mouse models examined, the Xcr1Cre model was the most efficient and specific for the fate mapping of all cDC1, regardless of the tissues examined. The KarmaCre model was rather specific for cDC1 when compared with the Clec9aCre mouse, but less efficient than the Xcr1Cre model. Unexpectedly, the Xcr1Cre model targeted a small fraction of CD4<sup>+</sup> T cells, and the KarmaCre model a significant proportion of mast cells in the skin. Importantly, the targeting specificity of these two mouse models was not changed upon inflammation. A high frequency of germline recombination was observed solely in the Xcr1Cre mouse model when both the Cre and the floxed alleles were brought by the same gamete irrespective of its gender. Xcr1, Karma, and Clec9a being differentially expressed within the cDC1 population, the three CRE-driver lines examined showed distinct recombination patterns in cDC1 phenotypic subsets. This advances our understanding of cDC1 subset heterogeneity and the differentiation trajectory of these cells. Therefore, to the best of our knowledge, upon informed use, the Xcr1Cre and KarmaCre mouse models represent the best tools currently reported to specifically and faithfully target cDC1 in vivo, both at steady state and upon inflammation. Future use of these mutant mouse models will undoubtedly boost our understanding of the biology of cDC1.

Keywords: dendritic cells, cDC1, XCR1, Gp141b, Karma, Clec9a, Cre, fate mapping

# INTRODUCTION

Dendritic cells (DCs) constitute a heterogeneous population of antigen presenting cells (APCs) which are instrumental for the orchestration of innate and adaptive immune responses. In mice and in humans, three distinct types of DCs differing in their phenotype, localization and functions populate all lymphoid and most non-lymphoid tissues at steady state. Plasmacytoid DCs (pDCs) are the major source of type I interferon (IFN) upon many viral infections. Conventional DCs (cDCs) consist of two populations, coined as type 1 and type 2 cDCs and which excel in the cross-priming of CD8<sup>+</sup> T cells or in the promotion of CD4<sup>+</sup> T cell and humoral immunity, respectively. The functions of cDCs and their molecular regulation have been studied in vivo by using a wealth of mouse models that enable their depletion or genetic manipulation, namely Cd11c (Itgax)hDTR (1) or Cd11c (Itgax)Cre (2, 3) and more recently the Zbtb46hDTR (4) or Zbtb46Cre (5). However, interpretation of the results obtained using those mice can be difficult due to the expression of Cd11c by many other cell types than cDCs and of Zbtb46 by committed erythroid progenitors and endothelial cell populations (6). Moreover, these mutant mouse models are not suited to study the respective functions of each of the two cDC types. This goal requires the use of refined mutant mouse models enabling specific targeting of either cDC1 or cDC2.

Constitutive (Batf3-KO mice) or conditional (ItgaxCre; Irf8fl/fl mice) genetic inactivation of transcription factors required for the differentiation of cDC1 allowed to study their specific functions in vivo (7, 8). However, interpretation of the results obtained with these models can be difficult because they are not targeting solely cDC1 (7, 9–11). Moreover, cDC1 are replenished in Batf3-KO mice under inflammatory conditions, due to expression of other Batf transcription factors that compensate for Batf3 loss (12). Finally, these models do not allow the editing of cDC1 genome, which would be a powerful method to decipher the molecular regulation of their functions. Hence, novel mutant mouse models are needed to reach this goal.

In all tissues with the exception of the intestine, cDC1 can be defined as CD24<sup>+</sup> SIRPα/CD172a<sup>−</sup> cDCs (13–15). In addition, lymphoid-tissue resident cDC1 express CD8α, whereas the cDC1 residing in the parenchyma of non-lymphoid tissues and their counterparts that have migrated in secondary lymphoid organs express CD103. CLEC9A, a C type lectin receptor that allows efficient cross-presentation by cDC1 of dying cell-associated antigens (16) has been identified as a good candidate to generate mice enabling selective targeting of cDC1 in vivo due to its selective expression in these cells and to a lesser extent in pDCs (17–20). However, a thorough analysis of mice expressing a Cre recombinase under the Clec9a promoter showed that Cre-driven recombination occurred not only in cDC1 and to some extent in pDCs, but also in cDC2, leading to the discovery that Clec9a is expressed in a progenitor cell common to both cDC types (21). Hence, the Clec9aCre mouse is not suitable for specific targeting of cDC1.

A major breakthrough in the field of cDC1 was the identification of XCR1 as a universal marker of all cDC1 regardless of their tissues of residency, and present in all the warm-blooded vertebrate species studied to date (22–27). Xcr1 encodes the chemokine receptor XCR1, which ligand XCL1 is strongly upregulated in natural killer (NK) cells, CD8<sup>+</sup> T cells and memory T cells upon activation in mice (24, 26, 28–31). Recently, a mouse model based on the expression of the Cre recombinase under the control of the Xcr1 promoter has been generated to specifically manipulate gene expression in cDC1. This mutant mouse model was engineered by replacing the single coding exon of Xcr1 by the Cre gene (32). This strategy assumes that the Xcr1 gene is haplosufficient. However, this hypothesis has to be tested considering that XCR1 promotes the crosstalk between cDC1 and NK cells or CD8<sup>+</sup> T lymphocytes, by facilitating their reciprocal recruitment and/or activation (24, 26, 29). Regardless of its potential limitation, this Xcr1tm4(cre)Ksho mouse model has been useful to decipher the role of cDC1 in intraepithelial T cell homeostasis in the intestine (32). However, to the best of our knowledge, it has not been used yet for conditional gene targeting of the cDC1 lineage. Besides Xcr1, the A530099j19rik gene (named Karma hereafter) has also been identified as selectively expressed in cDC1 by bulk transcriptomic analysis on immune cell subsets and organs (10, 33). The Karma gene encodes a protein with 7 transmembrane domains, likely corresponding to a G protein-coupled receptor, leading to its recent denomination as Gpr141b by the Mouse Genome Informatics. Recently, we generated the Karma knockin reporter/deleter mouse model, which expresses in the Karma locus a construct encoding both the fluorescent tandem dimer Tomato (tdTomato) and the human diphtheria toxin receptor (hDTR), allowing specific tracking and conditional depletion of cDC1 in vivo. Results obtained with this reporter mouse validated the Karma locus as highly reliable to functionally target cDC1 in vivo (33).

To match the unmet need of a mouse model allowing specific and efficient in vivo genetic manipulation of cDC1, we generated two novel Cre-driver lines, the Xcr1Cre and KarmaCre mice, by knocking in an IRES-Cre expression cassette in the 3′ -UTR of the Xcr1 or Karma locus, respectively. In this study, we used genetic tracing to characterize the specificity and efficiency of the Cre-mediated recombination in these new models at steady state and upon infection, and compared them with the Clec9aCre model. This study also advanced our understanding of the phenotypic heterogeneity of cDC1 with regard to their differentiation trajectory.

#### MATERIALS AND METHODS

#### Generation of cDC1 Targeting Cre Constructs and Mice

Xcr1Cre (B6-Xcr1tm1Ciphe) and KarmaCre (B6-Gpr141btm2Ciphe) mice were made according to a standard gene targeting approach in C57BL/6N-derived ES cells. They were constructed by inserting, through ET homologous recombination, a cassette containing the internal ribosome entry site (IRES) followed by a gene encoding the codon-improved version of Cre recombinase (34), into the 3′ -UTR of the Xcr1 or A530099j19rik/Gpr141b genes, 34 and 98 bp after the stop codon, respectively. These

mice were outcrossed for three generations with wild type (Wt) C57BL/6J mice purchased from Charles River Laboratories. All experiments were performed with sex-matched littermate mice at 6–12 weeks of age. Clec9aCre (Clec9atm2.1(icre)Crs) (21) knock-in mice [kindly provided by Caetano Reis e Sousa (The Francis Crick Institute, UK)], Karma-tdTomato-hDTR (Gp141btm1Ciphe) (33) were maintained on the C57BL/6J background. Rosa26lox−stop−lox−tdRFP (Gt(ROSA)26Sortm1Hjf ) mice in which expression of the tandem dimer Red Fluorescent Protein (tdRFP) is driven through the deletion of a "lox– stop–lox" sequence (35) were purchased from the Jackson Laboratory and maintained on the C57BL/6J background. Rosa26lox−stop−lox−DTA (Gt(ROSA)26Sortm1(DTA)Lky) mice in which expression of active domain of the diphtheria toxin (DTA) is driven through the deletion of a "lox–stop–lox" sequence (36) were obtained from Prs. David Voehringer and Richard M. Locksley, and maintained on the C57BL/6J background. Mice were bred and maintained in our specific pathogen—free animal facility. This study was carried out in accordance with institutional guidelines and with protocols approved by the Comité National de Réflexion Ethique sur l'Expérimentation Animale #14.

### Preparation of Cell Suspension From Blood and Tissues, and Analysis by Flow Cytometry

Splenocytes were prepared by infusing spleens with an enzymatic cocktail made of Collagenase D (1 mg/ml) and DNase I (70µg/ml, both Roche) in plain RPMI 1640, and further incubation for 25 min at 37◦C. Ice cold EDTA (2 mM) was added for additional 5 min. Cells were filtered through a 70-µm nylon sieve, and exposed to 0.155 M NH4Cl, 10 mM KHCO3, 0.127 M EDTA to lyse red blood cells. Liver and lungs were minced in an enzymatic cocktail (1 mg/ml of Collagenase D and 70µg/ml of DNase I), incubated for 25 min at 37◦C. Ice-cold EDTA (2 mM) was added for additional 5 min, then digested tissues were filtered through a 70µm nylon sieve (BD Falcon). Low-density cells were further enriched by centrifugation over a 1.069 g/ml density gradient (OptiPrep, Axis-Shield), washed and resuspended in PBS, EDTA 2 mM, 2% BSA, and red blood cells were lysed as detailed above. Cutaneous LNs (inguinal and axillar LNs) were cut into small pieces and digested for 25 min at 37◦C with a mixture of type II collagenase (Worthington Biochemical) and DNase I (Sigma-Aldrich) in plain RPMI 1640. The resulting cell suspension was treated with 5 mM EDTA and filtered through a 70µm nylon sieve (BD Falcon). For the skin, ears were split into a ventral and dorsal parts and incubated for 105 min at 37◦C in RPMI containing 0.25 mg/ml Liberase TL (Roche Diagnostic Corp.) and 0.5 mg/ml DNase I (Sigma Aldrich). Digested tissue was homogenized using Medicons and Medimachine (Becton Dickinson) to obtain homogenous cell suspensions. For skin mast cells, we used a protocol recently described (37). To test the germline recombination in blood cells, peripheral blood mononuclear cells (PBMCs) were enriched by centrifugation over a 1.077 g/ml density gradient (Ficoll-Paque Plus, GE Healthcare), washed and resuspended in PBS,

EDTA 2 mM, 2% BSA before staining. Staining of cells for flow cytometry started with a pre-incubation with 2.4G2 mAb to block unspecific binding to Fc-receptors. Staining with mAb (**Supplementay Table S1**) was then performed in PBS, 2% BSA, 2 mM EDTA for 25 min on ice. For exclusion of dead cells 4 ′ ,6-diamidino-2-phenylindole (DAPI, Thermo Fisher Scientific) was added 5 min before acquisition. Data were acquired on a LSRFortessa X-20 flow cytometer (BD Biosciences), and analyzed using FlowJo (Tree Star, Inc.).

#### Bone Marrow-Derived DC Differentiation

FLT3-L-BMDCs were generated as described (38) with some modifications. BM cell suspensions were prepared and red blood cells were lysed as detailed in the previous section. After washing in complete RPMI 1640 medium, cells were cultured at 3 × 10<sup>6</sup> cells/ml in 24 well plates, with 10% FBS, RPMI 1640 medium containing murine FLT3-L (in house supernatant from B16-Flt3l cells, used at 1/20 final) at 37◦C in 5% CO2. Four days after, half of the culture medium was replaced by fresh FLT3-L. Cells were harvested at indicated times for flow cytometry analysis after staining for CD11c, SiglecH, CD24, SIRPα, XCR1, and CD11b (**Supplementay Table S1**).

#### Microarray Data Generation and Analysis

DCs were generated in vitro from mouse BM FLT3-L cultures and sorting by flow cytometry to over 98% purity, as live, singlet, CD11c<sup>+</sup> cells that were SiglecH<sup>+</sup> for eq-pDCs, SIRPα <sup>−</sup>CD24high for eq-cDC1 and SIRPα <sup>+</sup>CD24−/low for eq-cDC2. Total RNA (50 ng) was used as starting material for each sample to synthesize biotinylated probes, using the NuGen protocol as described previously (39). Affymetrix Mouse Gene 1.0 ST raw. CEL files were analyzed in the R statistical environment (version 3.4.1). Data were RMA normalized using the oligo package and processed as described previously (40). Heatmaps of Log2 normalized expression values of selected genes were performed using the Morpheus website from the Broad Institute (https:// software.broadinstitute.org/morpheus/). Hierarchical clusterings were performed using the One-Pearson correlation as a metric and the average linkage as a clustering method for samples and genes, except for **Figure 5B** where the complete linkage method was used for the genes. The microarray data have been deposited in the GEO database under the series accession number GSE121859.

#### Mouse Cytomegalovirus Infection

Animals were infected intraperitoneally with 2 × 10<sup>5</sup> PFU of salivary gland-extracted MCMV Smith strain (3rd in vivo passage). Forty-Eight hours later, spleen and liver were harvested and prepared for flow cytometry analysis as described above.

#### Analysis of Germline Recombination

Four females and 2 males of each Xcr1Cre/wt; Rosa26tdRFP/wt and KarmaCre/wt; Rosa26tdRFP/wt genotype were backcrossed to C57BL/6J mice. Their progeny was genotyped and bled to analyse tdRFP expression in circulating T and B cells as a sign of germline recombination.

# RESULTS

#### Generation of New cDC1-Targeting Cre-Driver Lines

By comparative gene expression profiling, we and other have previously identified Xcr1 and a530099j19rik (Gpr141b/Karma) genes as specifically expressed by mouse cDC1 in different tissues throughout the body (10, 22–26, 31, 33). We used such unique gene expression profile to genetically target cDC1 in vivo by generating Xcr1Cre (**Figure 1A**) and KarmaCre (**Figure 1B**) knock-in mouse models. The insertion of an IRES-Cre cassette after the STOP codon of Xcr1 and Karma genes allows the translation of two separate proteins resulting in the expression of the Cre recombinase. Expression of endogenous XCR1 was not significantly altered in the Xcr1Cre mouse model (**Supplementary Figure S1A**, top).

#### The XCR1Cre Mouse Model Allows Selective and Efficient Recombination of loxP Sequences in Migratory and Resident cDC1 in All Tissues Examined

To determine the specificity of the Cre-induced recombination in Xcr1Cre (**Figure 1A**) and KarmaCre (**Figure 1B**) mice, we bred them with the Cre-reporter line Rosa26lox−stop−lox−tdRFP (named hereafter Rosa26tdRFP) (35), and analyzed the tdRFP expression pattern in immune cells of lymphoid organs [spleen and cutaneous lymph-nodes (CLNs)] (**Supplementary Figures S1A,B**) and non-lymphoid tissues (lungs, liver, and skin) (**Supplementary Figures S1C–E**). To define DC cell populations, we applied gating strategies adapted from (41). As a control Cre-driver line, we used the Clec9aCre mice (21) bred to Rosa26tdRFP. Regardless of the tissues examined, we found that all three mouse models achieved effective targeting of the cDC1-lineage, with Xcr1Cre , and Clec9aCre being the most efficient (**Figure 2**). In contrary to Clec9aCre, the Xcr1Cre and KarmaCre mouse models did not show any significant Cre activity, neither in pDCs nor in macrophages (**Figure 2**). However, Cre recombination (tdRFP signal) was detected in a fraction of other cell types. In Xcr1Cre/wt; Rosa26tdRFP/wt mice, a minute proportion (<1%) of CD4<sup>+</sup> T cells expressed tdRFP in the spleen, CLNs, lung and liver, which increased to a much higher fraction of CD4<sup>+</sup> T cells in the skin (8.4 ± 6.4%). The CD4<sup>+</sup> T cells harboring Xcr1-driven Cre recombination lacked detectable level of XCR1 (**Supplementary Figure S1E**, bottom). Hence, they likely derived from progenitors cells that transiently expressed Xcr1. In the skin, lungs and CLNs of KarmaCre/wt; Rosa26tdRFP/wt mice, a fraction of cDC2 had undergone recombination, although to a lesser extent than in Clec9aCre/wt; Rosa26tdRFP/wt mice where cDC2 were targeted in all tissues (**Figure 2**). Surprisingly, in the skin of KarmaCre/wt; Rosa26tdRFP/wt mice, a large proportion of mast cells (61.1 ± 12.8%) also expressed the tdRFP (**Figure 2**).

To further assess cDC1-targeting specificity in the three Cre-driver mouse strains, we analyzed the proportion of different immune cell types within the tdRFP<sup>+</sup> cells (**Figure 3**, **Supplementary Figure S2**). As expected on the basis of previous report (21), other cells than cDC1, in particular cDC2 and to some extent pDCs, represented the major fraction of the cells targeted in Clec9aCre mice (**Figure 3**). In contrast, in all examined organs, except for the CLNs, cDC1 represented the major fraction of the cells targeted in Xcr1Cre mice, CD4<sup>+</sup> T cells constituting the second most frequent cell types expressing tdRFP in these organs, and the most frequent in the CLNs. This reflects the higher numbers of CD4<sup>+</sup> T cells as compared to cDC1s in all these organs. Finally, cDC1 represented the major fraction of targeted cells in KarmaCre mice in all examined organs, except for the skin where 64.4 ± 5.8% of the tdRFP<sup>+</sup> cells were mast cells (**Figure 3**). We assessed the expression pattern of the Karma gene in mast cells and compared it to a variety of other immune cell types, using the database from the Immgen consortium. These results revealed that mast cells express similar levels of the Karma gene as cDC1 in all organs examined, trachea, tongue, esophagus, skin**,** and peritoneal cavity (**Supplementary Figure S3**), consistent with the efficient genetic tracing of mast cells in the skin of KarmaCre mice. Altogether, these results show that our novel Xcr1Cre and KarmaCre mouse models constitute the most reliable Cre-driver lines reported to date for selective and efficient in vivo targeting of cDC1, with the Xcr1Cre model performing the best. However, it should be noted that a small fraction of CD4<sup>+</sup> T cells is targeted in Xcr1Cr<sup>e</sup> mice and that mast cells are largely targeted in KarmaCre animals.

individual value, mean +/– SEM per group, and are from two pooled experiments with at least two mice per group.

#### The Use of Different cDC1-Specific Promoters for Driving Cre Expression Reveals Heterogeneity in the cDC1 Population Defined as CD24<sup>+</sup> SIRPα − cDCs

To try understanding the lower efficiency of the KarmaCre model for cDC1 targeting, as compared to the Xcr1Cre or Clec9aCre mice, we examined the expression pattern of the tdRFP reporter within the splenic cDC1 defined as CD24+SIRPα <sup>−</sup> cDCs (gated in the CD45+Lin−SiglecH−MerTK−CD64−CD11c+MHC-II+CD26<sup>+</sup> cells) (13). CD24+SIRPα <sup>−</sup> cDC1 can be split into 4 subsets according to their heterogeneous expression of CD8α and XCR1 (**Figure 4**). XCR1 expression was reported to correlate with a better crosspresentation by cDC1 (23, 42). The CD8α <sup>+</sup>XCR1<sup>+</sup> cDC1 subset was reported to be phenotypically and functionally homogenous (23, 42), and likely corresponds to full-fledged differentiated cDC1 endowed with a high

per group. Statistical analyses were performed using nonparametric Mann-Whitney tests in all experiments (\*\*, p < 0.01; n.s, non-significant).

crosspresentation activity (43). However, the three other cDC1 subsets, CD8α <sup>+</sup>XCR1−, CD8α <sup>−</sup>XCR1+, and CD8α <sup>−</sup>XCR1−, have not been extensively characterized. The CD8α <sup>−</sup>XCR1<sup>−</sup> cells may encompass pre-cDC1 (13, 44). The CD8α <sup>−</sup>XCR1<sup>+</sup> cells likely correspond to pre-terminally differentiated cDC1. The CD8α <sup>+</sup>XCR1<sup>−</sup> cells could correspond to the small fraction of homeostatically matured splenic cDC1 that have downregulated XCR1 expression, similarly to what occurs at steady state in the skin, the intestine or the thymus (25, 32, 39). It could also be possible that XCR1<sup>−</sup> cells encompass other cell types contaminating the SIRPα <sup>−</sup>CD24<sup>+</sup> cDC1 gate. However, the exclusion of CD3ε <sup>+</sup> and SiglecH<sup>+</sup> cells in our gating strategy ensured that the CD8α <sup>+</sup>XCR1<sup>−</sup> cDC1 subset was not contaminated by CD8<sup>+</sup> T cells nor CD8α <sup>+</sup> pDCs (45–47). Consistent with the early expression of Clec9a starting at the pre-DC stage (21), Clec9aCre model targeted efficiently all 4 subsets, regardless of XCR1 and CD8α acquisition, with more than 85% of the cells in each subset expressing tdRFP (**Figure 4**). Although the Xcr1Cre model was more efficient in targeting XCR1<sup>+</sup> cDC1 as initially expected, a significant Cre activity was also detected both in XCR1−CD8α <sup>−</sup> and XCR1−CD8α <sup>+</sup> cDC1

subsets, with 37% and 53% of tdRFP expression, respectively (**Figure 4**). This indicated that a significant proportion of these cells from these two subsets derived from XCR1-expressing precursors, consistent with the hypothesis that they respectively encompass pre-cDC1 and terminally matured cDC1. The KarmaCre model was effective in targeting the XCR1+CD8α + cDC1 subset, contrasting with no recombination detected in the XCR1−CD8α <sup>−</sup> subset and with only a weak Cre activity in the XCR1+CD8α <sup>−</sup> subset (**Figure 4**). The fraction of XCR1−CD8α + cDC1 targeted in KarmaCre mice (22.8 ± 6.1%) was lower than that of XCR1+CD8α <sup>+</sup> cDC1 (69.0 ± 2.1%). This suggest that most of XCR1−CD8α <sup>+</sup> cDC1 do not derive from XCR1+CD8α + cDC1 contrary to our expectation that the majority of the former cells correspond to an advanced maturation state of the latter ones. Altogether, the combined use of our fate mapping mouse models suggests consecutive expression of the corresponding genes along the differentiation of the cDC1 lineage in the spleen, with Clec9a expressed from the common cDC progenitor stage, Xcr1 likely starting at the pre-cDC1 stage and Karma turned on only at a later stage similarly to CD8α.

# Comparison of the Three Fate Mapping Mouse Strains Advances Our Understanding of the Differentiation Trajectory of cDC1

We further investigated to which extent our fate mapping mutant mouse models could help refining the differentiation trajectory of cDC1, using as a simple model bone marrow (BM) cells cultured with Fms-like tyrosine kinase 3 ligand (FLT3- L) (38). This model allows in vitro generation of three subsets of DCs, which are phenotypically and functionally equivalent to in vivo cDC1 (eq-cDC1), cDC2 (eq-cDC2) and pDCs (eqpDCs) (38, 48–50). To refine the differentiation trajectory of cDC1, we followed the acquisition of the tdRFP signal over time in FLT3-L-differentiated DCs generated from BM of our fate mapping mutant mice (**Supplementary Figure S4**, **Figure 5A**). Xcr1Cre and Clec9aCre models allowed efficient recombination in eq-cDC1 (56 vs. 90%) (**Figure 5A**). Interestingly, KarmaCre did not present any recombinase activity in any of the DC populations (**Figure 5A**). Consistently, whereas gene expression profiling of the eq-DC subsets generated in standard BM FLT3- L cultures confirmed their close homology to their in vivo counterparts isolated from the spleen (**Figure 5B**), it also showed that eq-cDC1 lacked expression of the Karma, Cd8a and Ly75 (Cd205) genes (**Figure 5B**, black arrows). This was confirmed using our previously published Karma reporter mouse model knocked-in for tdTomato in the 3′UTR of the Karma gene (33) (**Supplementary Figure S5A**). However, these eq-cDC1 acquired Karma and CD8α expression upon in vivo transfer (**Supplementary Figures S5B,C**). Karma was also expressed in eq-cDC1 differentiated from BM cells cultured with FLT3-L on feeder cells expressing the Notch ligand Delta-like 1 (**Figure 5C**, black arrows), similarly to what has been recently reported for CD8α and CD205 expression (43). Altogether, these results demonstrate a sequential expression of Clec9a, Xcr1 and Karma during cDC1 ontogeny, with Clec9a being induced early starting at the common cDC progenitor stage (21), then followed by Xcr1 which induction might be initiated already at the pre-cDC1 stage. Likewise to CD8α, Karma is acquired at a more advanced differentiation stage that is not reached under classical conditions of DC differentiation from BM in FLT3-L in vitro cultures but can be promoted by Notch signaling. This work thus significantly extends two recent studies showing that cDC1 derived in vitro from mouse or human hematopoietic precursors with a combination of cytokines and growth factors need additional signals to reach a terminal state of differentiation including acquisition of CD8α expression for mouse cDC1 (43, 51).

# The Cre Expression Under Xcr1 or Karma Promoters Remains cDC1-Specific Upon Infection-Induced Inflammation

The expression of many membrane proteins or transcription factors changes upon inflammation (12, 52). An important incentive for generating Xcr1Cre and KarmaCre models, was that the expression of the Xcr1 and Karma genes was specific for cDC1 both at steady state and under inflammatory conditions (53). To confirm this observation based on transcriptomic studies, we examined cDC1-targeting specificity of the KarmaCre and Xcr1Cre mouse models in an inflammatory context, namely systemic mouse cytomegalovirus (MCMV) infection, using the Rosa26tdRFP reporter as a read out. We adapted a gating strategy adapted from (52) to identify inflammatory DCs (InflDCs) in spleen and liver (**Supplementary Figure S6**). Although we could observe the appearance of InflDCs upon MCMV infection, tdRFP expression remained unchanged in infected animals as compared to control mice, being still essentially confined to the cDC1 population, both in Xcr1Cre/wt; Rosa26tdRFP/wt and KarmaCre/wt; Rosa26tdRFP/wt mice (**Supplementary Figure 6**). This demonstrated that both cDC1-targeting models are stable and allow excision of a floxed genomic sequence efficiently and largely selectively in cDC1 at steady state and upon inflammation.

### Germline Recombination of loxP Sequences Is Frequent in the Offspring of Xcr1Cre but Not KarmaCre Mice

Recombination of loxP-flanked genomic sequences in germ cells have been described in many Cre mouse models (54–56). To test whether germline recombination occurs in our cDC1-targeting Cre-driver lines, we backcrossed Xcr1Cre/wt; Rosa26tdRFP/wt and KarmaCre/wt; Rosa26tdRFP/wt mice to C57BL/6J mice, and analyzed their offspring for ubiquitous tdRFP expression, using blood T and B cells as a readout (**Figure 7A**). Total or partial germline recombination of the Rosa26tdRFP locus occurred in 95% of the offspring who had inherited one loxP-flanked allele from Xcr1Cre/wt; Rosa26tdRFP/wt male mice (**Figure 7B**). Germline recombination occurred with the same frequency irrespective of the segregation of the paternal Cre and loxPflanked alleles in the offspring, demonstrating that this process occurred during meiosis rather than in the embryo. No germline recombination was observed when both alleles were from maternal germ cells (**Figure 7B**). In ongoing crosses using the Xcr1Cre mouse model with different loxP-flanked mouse strains, we could also observe germline recombination in the progeny even when the floxed alleles were brought together with the Xcr1Cre allele by the maternal gamete (data not shown). This indicates that, contrary to what the results of the Rosa26tdRFP/wt backcross appears to suggest, off-target activity of the Cre recombinase in germline is not a gender effect. Additionally, the incidence of germline recombination depended on the loxPflanked allele (data not shown). No occurrence of germline recombination was detected so far for the KarmaCre model (**Figure 7B**). Therefore, the KarmaCre model might be more appropriate than the Xcr1Cre model to obtain rapidly mice in which cDC1 are inactivated for candidate genes, through conventional breeding strategies. Germline recombination in the offspring should however always be assessed for any novel loxP-flanked allele.

# DISCUSSION

Xcr1 and A530099j19rik (Karma/Gpr141b) genes code for the chemokine receptor XCR1 and for the putative G proteincoupled receptor Gpr141b, respectively, and are among the core

, KarmaCre/wt; Rosa26tdRFP/wt and Clec9aCre/wt; Rosa26tdRFP/wt BM cells cultured with FLT3-L. Eq-cDC2 were gated as

in vitro from Xcr1Cre/wt; Rosa26tdRFP/wt

(Continued)

FIGURE 5 | CD11c+MHC-II+CD24−SIRPα <sup>+</sup> cells, and eq-cDC1 as CD11c+ MHC-II+CD24+SIRPα <sup>−</sup> cells (Supplementay Figure S4). Data are shown for one experiment representative of two, with three mice per group. (B,C) Heatmaps display the expression profiles of archetypical genes previously shown to be selectively expressed in cDC1, cDC2, or pDC. (B) Gene expression across DC types either isolated from murine spleens (sp-pDC, sp-cDC1, and sp-cDC2), or derived in vitro in standard FLT3-L BM cultures (eq-pDC-FL, eq-cDC1-FL, and eq-cDC2-FL), as assessed with microarrays. (C) Gene expression patterns across cDC types derived in vitro from FLT3-L BM cultures under standard conditions (-FL) or on DL1-expression OP9 feeder cells (-FL-DL1), as assessed from public RNA-seq data (GEO accession number GSE110577).

MCMV infection. Cell population gating strategy detailed in Supplementay Figure S3. Data show one dot per individual value, mean +/- SEM per group, and are from two pooled experiments with at least three individuals per group of infected mice. Statistical analyses were performed using nonparametric Mann-Whitney tests when possible, and the difference between non infected and infected was non-significant in each cell population. NI, non-infected; InflDCs, inflammatory DCs.

gene signature specifically identifying mouse cDC1 throughout the organism (10, 22–25, 33). We have inserted an IRES-Cre cassette into the 3′ UTR of the Xcr1 and Karma coding exon to generate Xcr1Cre and KarmaCre mouse models, respectively. In this study, we have characterized the efficiency and specificity of Cre-mediated recombination in these novel mouse models at steady state and upon viral infection, comparing them to the Clec9aCre model. To the best of our knowledge, this study demonstrated that our novel Xcr1Cre and KarmaCre mouse models are the most trustful and robust for the genetic tracking and manipulation of cDC1 in vivo.

Amongst the three Cre-driver mouse models examined, Xcr1Cre model is the most efficient and specific for fate mapping all cDC1 regardless of the tissues examined. The KarmaCre model is rather specific for cDC1 when compared with Clec9aCre mouse, but much less efficient than the Xcr1Cre model. Unexpectedly, a fraction of CD4<sup>+</sup> T cells is labeled with tdRFP in the Xcr1Cre;Rosa26tdRFP/wt mouse (**Figure 2**) without expressing any detectable XCR1 at their cell surface (**Supplementary Figure S1E**). Further analysis need to be conducted to determine whether XCR1 was transiently turned on in the distant progenitors of these cells or on the contrary during their terminal differentiation. Interestingly, the proportion of Xcr1cre fate-mapped CD4<sup>+</sup> T cells was much higher in the skin than in the other organs examined, suggesting that these cells may be polarized toward specific functions and/or develop under instructive signals encountered preferentially in barrier organs. Further studies will be needed to test these hypotheses. In the

representative of each is shown. (B) Analysis of germline recombination in offspring from backcrosses of Xcr1Cre/wt; Rosa26tdRFP/wt and KarmaCre/wt; Rosa26tdRFP/wt mice of both sexes with wild-type (WT; C57BL/6J) mice. Germline recombination shown here occurred when both Cre and floxed alleles were of paternal origin. However, with other type of flox constructs, we regularly observed occurrences of germline recombination when both alleles were brought together, either by the father or by the mother.

skin of the KarmaCre; Rosa26tdRFP/wt mouse model, the vast majority of the tdRFP<sup>+</sup> cells were of mast cell origin (**Figure 3B**). Microarray data released recently by the Immgen consortium (https://www.immgen.org) show that mast cells from the skin, peritoneal cavity, trachea and esophagus express high level of the Karma gene (**Supplementary Figure S3**), confirming our observation. To the best of our knowledge, this is the first report of a gene that is selectively shared by both cDC1 and mast cells. The KarmaCre mice will therefore be of special interest to researchers aiming at genetically manipulating mast cells in tissues. In all tissues and in all mouse models examined, no tdRFP expression was detected in the CD45-negative cells present in cell suspensions (**Figure 2**). Although, we did not examine tdRFP expression in other non-hematopoietic cells, it is unlikely that some of these cell types would be targeted in Xcr1Cre or KarmaCre mice considering that the Xcr1 and Gp141b genes were not expressed outside of the hematopoietic system in all of the transcriptomic databases we queried.

To inactivate specifically a candidate gene in cDC1, both alleles of this candidate gene should be excised. This requires a breeding strategy in which one Cre allele and one floxed allele of the gene to be inactivated are brought by the same germ cells, where unexpected recombination could occur. We have tested the frequency of germline recombination for both the Xcr1Cre and KarmaCre models. Only the Xcr1Cre model showed adventitious Cre activity in germ cells resulting in progeny with recombined Rosa26tdRFP locus in all their cells. Interestingly, this was paternal inherited in this specific experimental setting. However, this may depend on the loxP-flanked construct, as we had events of germline recombination transmitted by females for other floxed genes than the Rosa26tdRFP reporter. Therefore, to reach specific recombination in cDC1 using the Xcr1Cre model, the Cre allele and the loxP-flanked allele should be inherited from different parents. We recommend breeding one parent homozygous for both the Cre allele and a null allele of the target gene, to another parent homozygous for the floxed allele of the target gene. Each investigator using the Xcr1Cre model should always test their progeny for unexpected off-target recombination. A recent publication strongly suggested to include, in each experimental procedure of publication using Cre mouse models, detailed procedures about the breeding strategies used and the method the investigators applied to detect any unexpected and unspecific recombination (56).

Our genetic tracing of cDC1 in vivo in the spleen (**Figure 4**), or in vitro in FLT3-L-differentiated BM-DC cultures (**Figure 5A**) revealed heterogeneity in the cDC1 population. The differential expression of Xcr1 and Karma genes within the cDC1 population qualifies Xcr1Cre and KarmaCre mouse models as powerful tools to describe further these cDC1 subsets in vivo. In vitro, BM-DCs derived from KarmaCre mice did not show any sign of Cre activity (**Figure 5A**), confirming that the Karma gene is not transcriptionally active in these cells as directly assessed through their gene expression profiling (**Figure 5B**), akin to Cd8a or Ly75 (Dec205) (43). We show here that expression of the Karma gene on cDC1 requires accessory signals, which can be provided upon in vivo transfer, or in vitro by Notch signaling likewise to what has been recently reported for Cd8a and Ly75 (43). Of note, in FLT3-L in vitro BM-DC culture, detection in eq-cDC1 of the activity of the Cre recombinase as readout by tdRFP expression seemed to be delayed over time as compared to cell surface acquisition of XCR1 (**Figure 5**), although the Cre and Xcr1 genes were expressed under the same promoter from one bi-cistronic mRNA. This might be explained by a delayed translation of the Cre gene as compared to Xcr1, or because efficient recombination of DNA by the Cre requires time. This latter case might especially apply to the Rosa26tdRFP reporter mouse line used in this study, because it was engineered as requiring two consecutive rounds of Cre-excision to generate detectable tdRFP signal, in order to limit any leaky transcription of the fluorescent reporter gene across the stopper at steady state (35). This contrasts to most reporter lines which require only one sequence of recombination to emit signal (57, 58) and must therefore require lower and/or less sustained Cre activity to allow recombination. Breeding the Xcr1Cre model with a Cre-reporter mouse which is easily recombined (57) might allow a better synchronization of XCR1 surface expression with Cre activity.

Our results advanced our understanding of the differentiation trajectory of cDC1, and validated the Xcr1Cre mouse model as a robust tool to inactivate genes selectively in cDC1 either in vivo or in vitro using BM-derived DC cultures. Future use of these mutant mouse models will undoubtedly boost the advancing of our understanding of the biology of cDC1.

## AUTHOR CONTRIBUTIONS

MD and KC: Conceptualization. RM, CW, BM, MD, and KC: Methodology, Validation, Formal Analysis. RM, CW, SG, MA, YA, CS, and AF: Investigation. T-PV, MD: bioinformatics analyses. BM: Mouse model construction supervision. RM, KC: Visualization. RM, KC, and MD: Writing. All authors: Editing. BM, MD, and KC: Project Supervision and Administration.

# FUNDING

This work was in part carried out in the frame of the Innate Immunocytes in Health and Disease (I2HD) collaborative project between CIML, AVIESAN, and SANOFI. It was supported by grants from the European Research Council under the European Community's Seventh Framework Program [FP7/2007–2013 grant agreement number 281225 to M. Dalod; FP7/2007-2013 grant no. 322465 (Integrate) to BM], from the Agence Nationale de la Recherche (ANR; XCR1-DirectingCells to KC), from the Fondation pour la Recherche Médicale (label Equipe FRM 2011, project number DEQ20110421284 to MD), which supported the KarmaCre mouse generation, from the Program Hubert Curien (PHC) Maïmonide-Israel 2017 project number 38155SJ (to MD) supported by the French Ministries of Foreign Affairs (MAEDI) and of Research and Higher Education (MENESR) and from the Fondation ARC pour la recherche sur le cancer (to KC). CIPHE is supported by the French Ministry of Research via PHENOMIN (French National Infrastructure for mouse Phenogenomics; ANR10-INBS-07 to BM). This work also benefited from institutional funding from CNRS and INSERM. RM was supported by doctoral fellowships from the Biotrail PhD program (Fondation A∗MIDEX). RM, CW, SG, and YA were supported by fellowships from Fondation ARC pour la recherche sur le cancer.

#### ACKNOWLEDGMENTS

We thank Frédéric Fiore and members of the Centre d'Immunophénomique (CIPHE) (Aix Marseille Univ, INSERM US012, CNRS UMS3367, Marseille, France) for generating the Xcr1Cre and KarmaCre mice, and for taking

#### REFERENCES


care of mouse breeding; Marilyn Boyron, the cytometry core, and animal house facilities (CIML, UMR7280, France) for critical technical assistance. The authors also thank Rebecca Gentek (Aix Marseille Univ, CNRS, INSERM, CIML, France) for helpful discussion and protocol sharing. This work benefited from data assembled by the ImmGen consortium.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2018.02805/full#supplementary-material


cross-presenting classical dendritic cells. Cell Rep. (2018) 23:3658–72.e6. doi: 10.1016/j.celrep.2018.05.068


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Mattiuz, Wohn, Ghilas, Ambrosini, Alexandre, Sanchez, Fries, Vu Manh, Malissen, Dalod and Crozat. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Modulation of Immune Tolerance via Siglec-Sialic Acid Interactions

Joyce Lübbers, Ernesto Rodríguez and Yvette van Kooyk\*

Molecular Cell Biology and Immunology, Amsterdam UMC, Vrije Universiteit Amsterdam, Cancer Center Amsterdam, Amsterdam Infection and Immunity Institute, Amsterdam, Netherlands

One of the key features of the immune system is its extraordinary capacity to discriminate between self and non-self and to respond accordingly. Several molecular interactions allow the induction of acquired immune responses when a foreign antigen is recognized, while others regulate the resolution of inflammation, or the induction of tolerance to self-antigens. Post-translational signatures, such as glycans that are part of proteins (glycoproteins) and lipids (glycolipids) of host cells or pathogens, are increasingly appreciated as key molecules in regulating immunity vs. tolerance. Glycans are sensed by glycan binding receptors expressed on immune cells, such as C-type lectin receptors (CLRs) and Sialic acid binding immunoglobulin type lectins (Siglecs), that respond to specific glycan signatures by triggering tolerogenic or immunogenic signaling pathways. Glycan signatures present on healthy tissue, inflamed and malignant tissue or pathogens provide signals for "self" or "non-self" recognition. In this review we will focus on sialic acids that serve as "self" molecular pattern ligands for Siglecs. We will emphasize on the function of Siglec-expressing mononuclear phagocytes as sensors for sialic acids in tissue homeostasis and describe how the sialic acid-Siglec axis is exploited by tumors and pathogens for the induction of immune tolerance. Furthermore, we highlight how the sialic acid-Siglec axis can be utilized for clinical applications to induce or inhibit immune tolerance.

Keywords: mononuclear phagocytes, dendritic cells, macrophages, Siglecs, tolerance, inflammation, sialic acid, cancer

# HIGHLIGHTS


# INTRODUCTION

The human mononuclear phagocyte network consists of monocytes, different subsets of macrophages (MQ) and Dendritic cells (DCs) depending on their origin and tissue micro-environment. In each microenvironment, differentiation is dictated by various components such as stromal cell compartment, presence of immune cells and the diversity

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Jan Lunemann, University of Zurich, Switzerland Sven Burgdorf, Universität Bonn, Germany

#### \*Correspondence:

Yvette van Kooyk Y.vanKooyk@vumc.nl

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 31 August 2018 Accepted: 14 November 2018 Published: 07 December 2018

#### Citation:

Lübbers J, Rodríguez E and van Kooyk Y (2018) Modulation of Immune Tolerance via Siglec-Sialic Acid Interactions. Front. Immunol. 9:2807. doi: 10.3389/fimmu.2018.02807 of chemokines and cytokines present (1). Moreover, mononuclear phagocytes are key instructors for inflammatory or tolerogenic programming of the immune system. The presence of MQ and DCs at multiple sites in the human body, like gut, lung, brain, oral mucosa, lymphnodes, spleen, skin and peripheral blood illustrates their importance in controlling immunity and tolerance (2–5). MQ are plastic cells that can polarize according to the signals they receive and this polarization is mainly described as classical activated M1, alternatively activated M2, or tumor-associated macrophages (TAM). The M1, depicted as pro-inflammatory cells, are induced by stimulating MQ with LPS and/or IFN-γ that produce IL-1, TNF-α and nitric oxide. On the other hand, M2 MQ are induced by stimulation with IL-4 and have anti-inflammatory and tissue repair properties, producing Il-10 and TGF-ß (6, 7). TAMs are found in the microenvironment of tumors, promoting tumor growth by among others release angiogenic factors like VEGF and EGF, attract regulatory T cells and inhibit effector T cells by the release of multiple cytokines and chemokines such as IL-10, TGF-ß, and CCL22 (8). DCs consists also of multiple subsets, were the conventional DCs (cDCs) and plasmacytoid DCs (pDCs) are the main populations in peripheral blood. The cDCs are the main antigen-presenting subset, able to present antigens to and activate antigen specific naïve CD4<sup>+</sup> and CD8<sup>+</sup> T cells and are able to secrete multiple proinflammatory and anti-inflammatory cytokines like IL12p70 and IL-10, respectively (9, 10). pDCs do not prime naïve T cells, however, there specialized function is the production of type I interferon (IFN-α/β) in response to viruses (11). Different cytokines like IFN-α, TNF-α and LPS can polarize cDCs into a more immunogenic state (12, 13), while other cytokines like IL-10 and TGF-ß induce tolerogenic cDCs that express checkpoint ligands like PD-L1 and produce the checkpoint molecule indoleamine 2,3-dioxygenase (IDO). In-vitro treatment of DCs with dexamethasone or vitamin D3 will also result in tolerogenic DCs (14). Functionally the main characteristics of MQ is their phagocytic capacity, while DCs are key in antigen presentation and stimulation of naïve T cells into antigen-specific effector T cells, however, some of these functions are not 100% restricted and are also shared between MQ and DCs.

In-vitro, human monocyte-derived DCs (moDCs) and monocyte-derived MQ (moMQ) can be generated from monocytes. Culturing monocytes with GM-CSF and IL-4 gives rise to moDCs, while culturing monocytes with M-CSF or GM-CSF alone creates moMQ (15, 16). moDCs and moMQ are often used as model systems for inflammatory DCs and MQ, respectively, as they are easily obtained in large numbers. moDCs are excellent antigen presenters, and able to induce antigen-specific CD4<sup>+</sup> and CD8<sup>+</sup> T cells, while culturing with IL-10 or TGF-β generates moDC prone to induce tolerance (4, 17–21). However, recent studies using mass cytometry as well as single cell RNA sequencing have revealed that moDCs are distinct from human peripheral blood and skin-derived DCs (2, 22).

Mononuclear phagocytes have an important function in maintenance of tissue homeostasis and the resolution of inflammation. They express multiple pattern recognition receptors (PRRs), like toll like receptors (TLR) and CLRs to recognize pathogen-associated molecular patterns (PAMPs), damaged self-antigens (DAMPs) or altered glycosylated selfantigens, such as tumor antigens (3, 23). The differentiation and maturation status of mononuclear phagocytes alters the expression levels of PRR (24, 25). CLRs is a large family of glycan-specific receptors that include, amongst others: DC-SIGN (CD209), Mannose receptor (MR, CD206), DEC-205 (CD205), Dectin-1, Macrophage galactose-type lectin (MGL, CD301) and Langerin (CD207) (26, 27). These CLRs are glycan-binding receptors, recognizing a wide variety of carbohydrate structures, like fucoses and mannoses found on host glycoproteins expressed by cells or pathogens or β-glucan structures that are only expressed on pathogens such as Aspergillus fumigatus and Saccharomyces cerevisiae (27–29). CLRs play an important role in the antigen uptake for processing and presentation of peptides on MHC class I and II, thereby stimulating antigen-specific T cell responses and T helper differentiation (27). Some CLRs, like Dectin-1, have the ability to directly modulate the DC or MQ phenotype and cytokine responses, while, other CLRs, like DC-SIGN and MGL are also highly expressed on tolerogenic DC/MQ and modulate TLR signaling through the acetylation of p65 and the induction of IL10 production (30–32).

Next to TLRs and CLRs, mononuclear phagocytes express Sialic acid binding immunoglobulin type lectins (Siglecs), that recognize sialic acids, a family of sugars with a nine-carbon sugar core structure derived from neuraminic acid, with the Nacetylneuraminic acid (Neu5Ac) being the main moiety present in humans (**Box 1** and **Figure 1**). Sialic acids are generally the last sugars added during the glycosylation process, thereby capping a diverse array of glycosylation structures (44, 45). Often, the presence of sialic acids functions as a self-associated molecular pattern (SAMP) and thus, Siglecs can serve as sensors for "self " (46). Most Siglecs possess an intracellular immunoreceptor tyrosine-based inhibition motif (ITIM) that induce strong inhibitory signaling when Siglecs bind sialic acids (47). Interestingly, both pathogens and tumor cells use enhanced expression of sialic acids as a mechanism to modify the immune system in their favor, illustrating that the sialic acid-Siglec axis is a key regulator in infection and cancer.

# SIGLECS

The human genome contains 14 different Siglecs, which can be divided into two groups based on their genetic homology among mammalian species. The first group is present in all mammals and consists of Siglec-1 (Sialoadhesin), Siglec-2 (CD22), Siglec-4 (MAG), and Siglec-15 (48–50). The second group consists of the CD33-related Siglecs that have evolved rapidly and therefore their repertoire differs between species. The CD33 related Siglecs are Siglec-3 (CD33),−5,−6,−7,−8,−9,−10,−11,−14, and −16 (51). Monocytes, moMQ and moDCs have largely the same Siglec profile (**Figure 2**), namely high expression of Siglec-3,−7,−9, low Siglec-10 expression and upon stimulation with IFN-α, also Siglec-1 (52–60) is expressed. In contrast, MQ have primarily expression of Siglec-1,−3,−8,−9,−11,−15, and−16 depending on their differentiation status (49, 52, 61, 62). cDCs express Siglec-3,−7, and−9, similar to moDCs, but in addition also express low levels of Siglec-2 and Siglec-15 (49, 63–67). pDCs are

#### Box 1 | Sialic acid.

Sialic acids are a family of sugars with nine carbons derived from neuraminic acid that are negatively charged. Humans are able to synthetize Neu5Ac (Figure 1A), while other mammals can also synthetize the structure N-glycolylneuraminic acid (Neu5Gc). A deletion in the gene encoding the enzyme CMAH (Cytidine monophosphate-N-acetylneuraminic acid hydroxylase) is the reason why humans cannot produce Neu5Gc (33).

Figure 1 | Sialic acids, linkages, and interactions. (A) chemical structure of sialic acids Neu5Ac and Neu5Gc. (B) α2,3; α2,6, and α2,8 linked sialic acids. (C) Trans and (D) Cis interactions of Siglecs with sialic acids.

#### Synthesis

The expression of sialylated glycans is the result of glycosylation related enzymes able to catalyse the addition or removal of a glycan to growing carbohydrate structures. The transfer of sialic acid motifs from an activated donor (CMP-NeuAc, Cytidine 5′ -MonoPhospho-N-AcetylNeuraminic acid) to underlying glycans that serve as acceptors, is performed by a group of enzymes called sialyltransferases. Humans express more than 20 different sialyltransferases, each differing in their tissue expression, substrate specificity and linkages produced (34). The synthesis of sialylated structures depends also on the presence of the donor, which is synthetized in the nucleus by the enzyme CMAS (CMP-Neu5Ac synthetase) and subsequently transported into the Golgi via the transporter SLC35A1 (33, 35). Sialic acid blocking glycomimetic: Ac53FaxNeu5Ac is a metabolic inhibitor of sialyltransferases that blocks the addition of sialic acids to the glycan backbone (36).

#### Sialic Acid Linkages

Sialic acids can be linked to the underlying glycan via different types of linkages, which affects their recognition by glycan-binding receptors, such as Siglecs. These linkages mainly have an alpha configuration and are defined by which carbon in the acceptor glycan is connected to the anomeric carbon in the Neu5Ac (carbon 2). When sialic acid is transferred to a different glycan, the bond can involve the carbon 3 or 6 in the acceptor rising to α2,3 or a α2,6 linkages, respectively, (33, 35) (Figure 2B). In poly-sialic acid structures, one Neu5Ac is added to a strain of sialic acids in an α2,8 linkage. The different Sialic acid linkages are depicted in the complementary figure to this box.

# Trans/Cis Interaction

Siglecs can interact with sialic acid on a different cell or protein/particulate (trans interaction) or with sialic acids present on the same cell that expresses the receptor (cis interaction), as depicted in the figure complementary with this box. An illustration of a trans interaction is α2,3 linked sialic acids expressed by lung epithelium under inflammatory conditions and Siglecs present on neutrophils (37, 38) (Figure 1C). An example of a cis interaction is α2,3 linked sialic acid present on the cell surface of moDCs, which bind to a Siglecs present on the same moDCs (39) (Figure 1D).

# Degradation

Specific glycosidases, called neuraminidases or sialidases, can hydrolyse the sialic acid from oligosaccharides. Present mainly in intracellular vesicles, these enzymes can be secreted, thereby changing the profile of sialylated structures present on the cell membrane. Their expression is dysregulated in many different types of cancer.

#### Sialic Acid Immune Modulation

Sialic acids can modulate the immune system in diverse ways through Siglecs, influence on antibody mediated clearance of pathogens and through complement. Sialylation of the antibody immunoglobulin A (IgA) interferes with the cell surface attachment of influenza A and mediates anti-viral activity of IgA (40). Sialic acids can also bind to complement regulator factor H and by this negatively regulate the complement alternative pathway (41–43).

different in their expression of Siglecs, as they express Siglec-1 and Siglec-5 (54, 68). The presence of the Siglecs on mononuclear phagocytes is based on their steady state situation, however, microenvironmental triggers that change the maturation status of the cell, may influence the loss or gain of the expression of Siglecs. Downregulation of Siglecs-7 and Siglec-9 expression on moDCs is observed after stimulating moDCs for 48 h with LPS, however, on moMQ Siglec expression is not changed upon LPS triggering (54). Clearly, further research on the regulation of Siglec expression during cellular maturation is needed. Siglecs are also present on other immune cells [nicely reviewed by MacAuley et al. (69)], such as B cells, basophils, neutrophils, and NK cells, with different expression patterns for every cell subset.

Sialic acids, the ligands for Siglecs receptors, are widely expressed as they are exposed on the outermost end of glycosylated structures of glycoproteins expressed on immune and other cells in the body, secreted glycoproteins in tissues and blood and on extracellular matrix in tissues (70, 71). It is the glycosylation machinery of the cells that determines the type of sialic acids to be added on the carbohydrate backbone to be expressed by the glycoprotein (**Box 1**, **Figure 1**). A Siglecexpressing immune cell can bind to sialic acids present on another cell or secreted glycoprotein and this is called a trans interaction (72) (**Figure 1**). Siglec receptors can also bind sialic acids exposed on the same cell, called a cis interaction. Moreover, Siglec receptors have different binding affinities for different linkage and modifications of sialic acids (see **Box 1** for more information about sialic acid). Most Siglecs have a preference for a particular sialic acid linkage, being either α2,3, α2,6, or α2,8-linked sialic acid but Siglecs may also show redundant specificity toward more linkages (52, 58).

# IMMUNE MODULATION THROUGH SIGLEC SIGNALING

The immune modulatory effect induced upon sialic acid binding to Siglec is regulated through downstream signaling pathways. Siglec-5 till Siglec-11, are the so-called inhibitory Siglecs, carrying ITIM and/or ITIM like motifs in their cytoplasmic domains, which can be phosphorylated by the Src family, thereby creating a binding site for the tyrosine phosphatases SHP-1 and SHP-2 (**Figure 3A**). Upon binding of SHP-1/2, de-phosphorylation of downstream targets can be achieved and ubiquitination, internalization, and phosphorylation of the receptor can be regulated (73, 74). The Src-mediated phosphorylation of ITIMs in Siglec-3 and possible also other ITIM-containing Siglecs can also lead to the binding of Cbl, a RING finger-containing E3 ligase, and suppressor of cytokine signaling 3 (SOCS3), resulting in the ubiquitination and protosomal degradation of Siglec-3. The same process also regulates the internalization and surface abundance of Siglec-3. SOCS proteins are upregulated by cytokines during inflammatory responses, leading to the loss of Siglec-3 and thereby higher proliferation of myeloid cells (75, 76) (**Figure 3A**). Signaling of different Siglecs through the binding of sialic acids or crosslinking via antibodies can lead to both an inflammatory or tolerogenic state in distinct mononuclear phagocytes. Antibodies against Siglec-3 and −7 inhibit the proliferation of myeloid cells (77) while monocytes

treated with Siglec-3 antibodies show increased production of the pro-inflammatory cytokines IL-1β, TNF-α, and IL-8 (78). These findings illustrate that crosslinking of Siglec-3 expressed on different myeloid cells induces opposite functional outcomes. Moreover, crosslinking Siglec-3 on monocytes via antibodies signals a pro-inflammatory effect, while cis binding of sialic acids to Siglec-3 represses IL-1β production by monocytes (78).

In contrast, Siglec-4 and Siglec-14 till Siglec-16 do not have an ITIM or ITIM like motif, but instead signal through the association of DNAX activation protein (DAP)12 and are therefore called activating Siglecs. DAP12 associates with these activating Siglecs through a positively charged lysine residue in the transmembrane domains and contain a cytosolic immunoreceptor tyrosine-based activation motifs (ITAM), which can recruit PI3K (**Figure 3B**) (49, 62). Furthermore, Siglec-14 can promote an inflammatory response by activating the MAPK pathway (46). The activating Siglecs most likely developed under evolutionary pressure, when pathogens adapted using the inhibitory Siglecs to circumvent the immune system, allowing sialic acids to also activate the immune system. There are a couple of paired Siglecs, consisting of an inhibitory and an activating Siglec, like Siglec-5 and Siglec-14 as well as Siglec-11 and Siglec-16. Polymorphisms in Siglec-5/14 have been described, whereby the Siglec-14 gene is deleted and the Siglec-5 gene is present under the Siglec-14 promotor. When monocytes from these individuals are challenged with LPS or with group B streptococcus (GBS) they produce less TNF-α than individuals that have normal Siglec-5 and Siglec-14 expression on their monocytes, indicating that Siglec-14 is tipping the balance toward the pro-inflammatory site when cells are confronted by pathogens (46, 61, 62).

Siglec-1 is a non-signaling Sigelc, that internalizes upon ligand binding (**Figure 3C**). Den Haan et al. showed in mice that antigen coupled to Siglec-1 antibodies targets Siglec-1 expressing marginal zone macrophages that transfer antigen to CD8<sup>+</sup> DC favoring effective antigen specific T cells to eradicate tumor growth (79, 80). Furthermore, it has been shown that Siglec-1 <sup>+</sup> MQ promote germinal center B cell responses upon Siglec-1 antibody targeting (81).

Siglecs can also exert their immune modulatory effects by altering TLR signaling. LPS stimulation of TLR4 induces CCR7 upregulation of moMQ which is inhibited by anti-Siglec-9 antibodies or knock down of expression of Siglec-9 (53, 82). G. Chen and colleagues revealed that TLR4 forms a complex with Siglec E (mouse homologue for human Siglec-7 and Siglec-9) in murine DCs and macrophages (83). This cis interaction between Siglec-E and TLR4 is likely mediated by sialic acids present on the TLR (83). The TLR-Siglec-E interaction, is abrogated by NEU1 (a lysosomal sialidase that cleaves sialic acids from their glycoprotein backbone), which is translocated to the cell membrane upon LPS stimulation.

moDCs treated with the Ac53FaxNeu5Ac (**Box 1**), showed reduced sialic acid expression and a lower threshold of TLR activation, leading to increased sensitivity and response to poly I:C (TLR 3 agonist) and LPS, as reflected by the induction of moDC maturation and cytokine production by moDCs (55). Furthermore, it has been reported that sialic acid removal from moDCs uncovers Siglecs from their cis binding sialic acid ligands and increases expression of the maturation markers CD80 and CD86 and the secretion of IL-12 (84). Also, the cross presentation of melanoma antigens gp100 by DCs to antigen-specific CD8<sup>+</sup> T cells was increased upon removal of sialic acid on moDC, illustrating that the presence of sialic acid constraints, that occupy Siglecs in cis, inhibits the effectiveness of moDC to induce immunity (84). Alternatively, targeting Siglecs with sialic acids or sialic acid mimetics in trans can modulate TLR signaling leading to a more tolerogenic DC phenotype. This illustrates that interference in the sialic acid-Siglec axis is central in the balance between immunity and tolerance.

#### SIALIC ACIDS USED BY PATHOGENS TO MODULATE IMMUNITY

The co-evolution of the immune system and pathogens has led to the acquisition of several strategies for pathogens to evade the immune system, which also includes the expression of sialylated glycans to induce tolerance. One of the most notable examples is Trypanosoma cruzi, a protist parasite responsible for Chagas disease. During its infective stage in vertebrates, called trypomastigote, T. cruzi expresses a unique enzyme called trans-sialidase that catalyses the reversible transference of sialic acid from host glycoconjugates to glycan structures on the surface of the parasite. By doing this, T. cruzi uses host glycans to mask its own antigens and to modulate anti-parasitic responses (85). Parasitic sialylated glycans can interact with Siglec-E [homolog of human Siglec 7 and 9, Siglec comparison between Mammalians was recently reviewed by Bornhöfft et al. (86)] in murine dendritic cells to suppress the production of the pro-inflammatory cytokine IL-12 (87). Moreover, the addition of sialic acid to the surface of the parasite results in a negatively charged coat that inhibits complement-mediated killing. Furthermore, thanks to the trans-sialidase activity, T. cruzi is also able to alter the sialylation status of CD8<sup>+</sup> T cells, dampening their capacity to induce an effective anti-parasitic immune response (88).

Interestingly, several pathogenic bacteria also use the sialic acid-Siglec axis to dampen the immune system in favor of their survival. Despite the fact that sialic acids are mainly restricted to vertebrates, some bacteria have acquired the ability to take sialic acids or sialylated structures from the host, to synthetize "mimic" structures or even perform de novo synthesis of sialic acids, giving them a survival advantage. For example, Siglec-5 and−9 on neutrophils can be triggered by glycoconjugates present in Pseudomonas aeruginosa or Group B streptococcus (GBS) serotypes Ia and III, thereby inhibiting their ability to respond to the bacteria. Moreover, sialylated glycans present in GBS are able to inhibit the complement system, by reducing deposition of C3b on their surface and, therefore, the generation of C5a and the membrane attack complex (89–91).

The presence of sialic acids in envelope glycoproteins of viruses also contributes to enhanced infection of the host. This is the case for the Human immunodeficiency virus (HIV) and the Porcine reproductive and respiratory syndrome virus (PRRSV), which can bind to Siglec-1 to promote trans infection (92– 94). Nevertheless, Siglec-1 ligands on GBS surface interact with Siglec-1 on marginal zone macrophages for the subsequent generation of anti-GBS immune responses (95).

Influenza A virus recognizes α2,3 and α2,6 linked sialic acids with its hemagglutinin (HA) glycoproteins to infect host cells. On the other hand, influenza A virus carries the neuraminidase (NA) glycoprotein that can cleave off sialic acids from cellular and viral glycoproteins that are expressed in infected cells and assembled in virions, to reduce HA causing aggregation of the virions to the cell surface. The HA and NA proteins are in perfect balance to warrant infection and to abolish detection by the immune system (96). Another example is the nontypeable Haemophilus influenzae (NTHi), which is also able to take up sialic acids through a tripartite ATP-independent periplasmic (TRAP) transporter. Incorporation of the sialic acids in the NTHi membrane protects it from serum-mediated killing (97).

Sialic acids are used by different pathogens to infect host cells and dampen the immune response. Knowledge on this mechanism can be exploited to design new therapeutic strategies in cancer or auto-immune diseases and asthma.

### SIALIC ACID—SIGLEC AXIS IN CANCER

Aberrant glycosylation of multiple cancers and its influence on cancer progression and metastasis are well-known. Increased sialylation, α2,3; α2,6, and α2,8 linked sialic acids, has been demonstrated in multiple tumor tissues like renal cell carcinoma, prostate cancer, colon cancer, breast cancer, head and neck squamous cell carcinoma and oral cancer (98–101). This aberrant sialylation can also be detected in serum serving as potential biomarkers for cancer detection, progression and treatment responses (99, 101–103) (**Figure 4**).

In a mouse model for melanoma, hyper sialylation of B16 melanoma cells leads to increased tumor growth, associated with an enhanced T regulatory/T effector balance and reduced NK cell activity within the tumor and secondary lymphoid organs (110). DCs that interacted and sampled sialylated antigens via Siglec-E (murine homologue of human Siglec-7 and Siglec-9) induced regulatory T cells and inhibited effector T cell function in-vivo. These findings revealed that tumor sialylation impedes T cell-mediated anti-tumor immune responses, while promoting tumor-associated regulatory T cells (110). Blocking the inhibitory effects of sialic acids with a sialic acid blocking glycomimetic (**Box 1**) in a B16-OVA mouse model revealed reduced tumor growth, enhanced tumor killing by ovalbumin specific CD8<sup>+</sup> T cells and inhibition of metastasis (106, 107) (**Figure 4C**).

In breast cancer a specific glycoform of transmembrane mucin 1, MUC1-T is sialylated, creating MUC1-sT (111, 112). The MUC1-sT can interact with Siglec-9 on monocytes and thereby induce secretion of IL-6, M-CSF and chemokines associated with tumor progression. Binding of MUC1-sT to Siglec-9 on macrophages induces a tumor-associated macrophage (TAM) phenotype, that inhibits CD8<sup>+</sup> T cell proliferation and results in the upregulation of IDO, CD163 and PD-L1 in-vivo (113, 114). Another specific mucin glycoform, called MUC2-sT, has been shown to increase apoptosis of immature moDCs (115). Together, this points toward a broad immunological suppression by tumor-produced sialylated mucins.

Antibodies against Siglecs are explored for the treatment of different cancer types. For Acute Lymphoblastic Lymphoma (ALL) the FDA approved Inotuzumab Ozogamicin (Besponsa <sup>R</sup> ), a monoclonal antibody against Siglec-2 coupled to the toxic agent calicheamicin is used. This antibody targets Siglec-2 positive Blymphoblasts and causes cell death of these cells through the toxic agent (**Figure 4A**). Trials with this antibody revealed that an enhanced number of patients reached complete remission and had an increased overall progression free survival. However, serious adverse effects were seen like myeloid suppression (104), which could be due to the presence of Siglec-2 on DC subsets. Another Siglec that is targeted for the treatment

induce apoptosis of Siglec-2-expressing acute lymphoblastic leukemia cells (104). (B) HER2 targeting with a sialidase coupled to the HER2 antibody or locally applied non-targeted sialidases/synthesis inhibitors. This decreases sialic acid expression, reduce T reg induction and induced T cell activation and initiates NK cell killing (105). (C) Sialic acid inhibitor P-3Fax-Neu5Ac inclusion in nanoparticles targeted to tumor cells inhibits the sialic acid expression on the tumor cells, thereby decreasing metastasis and increasing tumor cell killing (106, 107). (D) Sialylated antigens target DC to remove regulatory T cells (108). (E) Antigen-specific B cell apoptosis induction by STALs targeting Siglec-2 in combination with an antigen that inhibits B cell receptor signaling on B cells (109).

of acute myeloid leukemia (AML), is Siglec-3, using Siglec-3 antibody coupled to calicheamicin (116, 117). Hereby, the myeloid blasts that express Siglec-3 are targeted and this improved outcome in patients with relapsing disease as well as in elderly patients that were not eligible for extensive chemotherapy (116). Similar to the Siglec-2 antibody treatments, the Siglec-3 antibodies caused extensive adverse effects, probably due to the wide spread expression of Siglec-3 on (healthy) myeloid subsets. Therefore, it is of great imporatnce to have a complete and accurate overview of Siglec expression on immune cells. Other strategies to target Siglec-3 in AML include the use of CAR-T cells. Siglec-3 targeting CAR T-cells have shown to induce CD8<sup>+</sup> T cell degranulation against primary AML and AML cell lines in-vitro (118, 119). Although different CAR-T cells are already tested in the clinic for lymphoid leukemia (120), it is questionable whether Siglec3 targeting CAR-T cells have similar severe side effects as observed with Siglec-3 antibodies.

Instead of targeting the Siglecs using antibodies, modifying the phosphorylation status of Siglec-3 and Siglec-9, in particular dephosphorylation of the receptos, has shown to lead to increased immunity of moDCs, when treated with Dasatinib a SRC tryrosine kinase inhibitor that dephosphorylates Siglec-3 and Siglec-9 (121). Also, leukemia (BCR-ABL<sup>+</sup> AML) patients treated with Dasatinib, had a stronger CD8<sup>+</sup> T cell and NK cell response associated with long lasting remission (122). Another strategy to increase anti-tumor immunity through Siglecs has been developed by Xiao et al., where they target HER2 with a monoclonal antibody fused to a sialidase (105) (**Figure 4B**). This sialidase specifically cuts off the sialic acid ligands that are bound by Siglec-7 and Siglec-9 and thereby increase NK activity. In vitro these HER2 targeting antibodies fused to a sialidase, increased the NK cell mediated killing of HER2 positive tumor cells (105). As most breast cancer patients are HER2 positive, targeting of HER2 with this sialidase fused antibody could be an effective treatment strategy.

Most strategies that interfere with the sialic acids-Siglec Axis are developed for leukemic cells as they have high expression of Siglec-2 or Siglec-3 and are therefore easily targetable. Other cancer type treatments could also benefit from targeting Siglecs, blocking of Siglecs could abrogate the inhibitory effects on mononuclear phagocytes and lead to better migration and maturation of these cells, which subsequently stimulates tumorspecific T cell responses. Moreover, local removal of tumorassociated sialic acid may temporarily de-tolarize the tumor microenvironment and trigger immune activation at the tumor site. A combination with checkpoint inhibitors would than favor improved tumor eradication.

### SIALIC ACID—SIGLEC AXIS TO INDUCE TOLERANCE FOR ALLERGIES AND AUTO-IMMUNE DISEASES

While in cancer it is important to induce immunity, in allergies and auto-immune diseases, an overactive immune system needs to be restored by inducing tolerance. Exploring the potential of the Sialic acid-Siglec Axis is an alternative to induce tolerance in an antigen specific manner. Because immune inhibitory Siglecs are found on mononuclear phagocytes, strategies can be designed aimed to actively induce tolerance via targeting inhibitory Siglecs on mononuclear phagocytes.

Modification of antigens such as OVA or MOG peptides with α2,3 or α2,6 sialyl-lactose has shown to increase targeting of these antigens to Siglec E, the human Siglec 7 and Siglec-9 homologue, and alter DC function in mice. Both in-vitro and in-vivo experiments demonstrated that sialic acid modified antigens induced antigen specific T reg induction and inhibition of inflammatory effector cells when activated with LPS (108) (**Figure 4D**).

Also, Siglec-engaging tolerance-inducing antigenic liposomes (STALs) are employed, in which sialic acid decorated nanoparticles, or sialo-glycoproteins or Siglec antibody targeting are used for Siglec targeting to induce tolerance. STALs with the peanut allergen Ara h2 (Ah2) and a high affinity Siglec-2 ligand (modified α2,6 linked sialic acid) incorporated in the outer membrane have shown high binding affinity to the B cell receptor and Siglec-2 simultaneously and to prevent peanut allergy against the Ah2 allergen in mice (109) (**Figure 4E**). This is acclaimed to the forced interaction between the B cell receptor and Siglec-2, thereby inducing apoptosis of autoreactive B cells. Pang et al. used the same STALs and incorporated rapamycin in the STALs and thereby enhanced the tolerogenic capacity in mice, which was the result of increased phagocytosis of these STALs by macrophages and DCs (123).

Sialic acid mimetics, such as the modified sialic acid coupled to liposomes discussed above, comprise of natural sialic acids as a backbone and are modified at certain positions in the sialic acid structures to develop high affinity ligands for Siglecs (124– 126). Addition of hydrophobic groups at the C2 and C9 of α2,6 sialic acids results in a high affinity Siglec-2 ligands, which outcompeting the cis interaction between the Siglec and its ligand. As a result better binding, endoyctosis, and eventually apoptosis of targeted B cells is acquired (126, 127).

Nanoparticles decorated with α2,8 linked sialic acids were developed to target murine Siglec-E (homologue of human Siglec-7 and Siglec-9) on macrophages. This approach limited the pro-inflammatory cytokines production by LPS-treated MQ in-vitro. Subsequently, these nanoparticles were able to limit the inflammation and increase levels of IL-10 in serum in a mouse model for LPS-induced systemic inflammation. Similar results were seen with human moMQ, resulting in an antiinflammatory cytokine profile. In an ex-vivo human lung perfusion model the nanoparticles coated with α2,8 linked sialic acids reduced pulmonary oedema after LPS-induced injury (128).

Several studies have shown the importance of Siglec-sialic axes in auto-immune disease and allergy due to expression of Siglecs on other immune cells such as eosinophils and B-cells. Asthma is an eosinophil, expressing Siglec-8, mediated disease and it has been shown that polymorphisms in the SIGLEC8 gene are linked to the development of asthma (38, 129). Antibodies against Siglec-8 or the mouse homolog Siglec-F induce caspase and ROS dependent apoptosis of eosinophils (130, 131). Autoantibodies against Siglec-8 have been found in intravenous immunoglobulin preparations that are used in various chronic inflammatory disorders, although some cytotoxic effects are known (132). For asthma it would be beneficial to have specific Siglec-8 agonists to induce neutrophil apoptosis without the risk of side effects observed with intravenous immunoglobulin injections. Another example is the anti-Siglec-2 antibody (Epratuzumab) that has already been tested in seven clinical trials for the autoimmune disease systemic Lupus erythematosus (SLE). Although initial trials showed promising effects with reduced peripheral B cell numbers, the overall effect was not better than standard care for SLE (reviewed by D Geh) (133).

Most of these strategies are to date only tested in in-vitro or ex-vivo experiments and should be tested in in-vivo and clinical trials as they have great potential for future applications in the treatment of allergies and auto-immune diseases. It is also important to elucidate the Siglec expression on different human immune cell subsets in order to identify the potential risks on side-effects by targeting multiple Siglecs with one ligand.

### CONCLUDING REMARKS

The last decade researchers identified the enormous potential of the sialic acid—Siglec axis to induce wanted or unwanted immune tolerance in cancer, allergies or auto-immune diseases. Both Siglec targeting antibodies, sialic acid mimetics, or glycan modifying agents can be used to interfere in this process and open new area's in the design of novel therapies for cancer, allergy and auto-immune diseases.

Still several questions need to be answered related to a better understanding of the biology of the Siglec-sialic acid axis. Those relate to the signaling capacity that sialic acid impose on immune cells to modify its function toward tolerance induction or activation of immunity. Interesting research topics to be addressed are: How do the various Siglecs expressed on one cell communicate with each other What is the exact specificity of these receptors for sialic-acids? Does multivalency of sialic acids or Siglec receptors matter? Other intriguing questions to be solved are: Do we need to inhibit only one Siglec receptor or more Siglec receptors simultaneously on one cell to alter function? What is the relation to cis and trans interaction on Siglecsialic acid interactions? How important is the protein or lipid backbone on which the sialic acid is exposed? To answer these questions an urgent need for Siglec specific targeting molecules is needed, which can be sialic acid mimetics or Siglec specific antibodies.

As the Siglec—sialic axis plays a crucial role in tissue homeostasis and the resolution of inflammation, more studies are necessary to understand their involvement in these biomedical

#### REFERENCES


processes. A better understanding of its role in the resolution of inflammation is crucial for its application in the treatment of auto-immune diseases and allergies.

Moreover, both pathogens and tumors use the Siglecsialic acid axis in their own benefit. It is therefore of vital importance to design new methods to modify glycosylation at site of infection or tumor location. Several studies already touch upon the investigation of targeting specific sialidases to the tumor micro-environment to remove the sialic acid content involved in the induction of tolerance in the tumor microenvironment. To unleash the sialic acid imposed tolerance in the local tumor microenvironment may be combined with other immune checkpoint inhibitors to stimulate tumor immunity in a multilevel manner. Alternatively, presence of sialic acid signatures in the tumor microenvironment may serve as new biomarkers to define immune tolerizing signatures in individual tumors and response therapy prediction (99).

Future studies are of great importance to unveil the complex Siglec-sialic acid axis and will warrant new discoveries in clinical application strategies in cancer, allergy and auto-immune diseases.

# AUTHOR CONTRIBUTIONS

JL was involved in the writing, reading of literature, design and discussion of the figures. ER was involved in writing, critical reading of the manuscript and designing the figures. YvK was involved in the overall supervision of the review and editing of the manuscript.

# FUNDING

This study was supported by ERC/Advanced/ERC339977/ Glycotreat (JL and YvK) and the European Training Network IMMUNOSHAPE (ER).

# ACKNOWLEDGMENTS

We gratefully thank Sandra J. van Vliet and Juan J. García-Vallejo, Amsterdam, UMC, Vrije University Amsterdam (Amsterdam, The Netherlands) for critical reading and fruitful discussions.


for acute lymphoblastic leukemia. N Engl J Med. (2016) 375:740–53. doi: 10.1056/NEJMoa1509277


(Siglec-8) gene are associated with susceptibility to asthma. Eur J Hum Genet. (2010) 18:713–9. doi: 10.1038/ejhg.2009.239


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Lübbers, Rodríguez and van Kooyk. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Structure-Function Relationship of XCL1 Used for in vivo Targeting of Antigen Into XCR1<sup>+</sup> Dendritic Cells

Arthur L. Kroczek 1,2,3, Evelyn Hartung<sup>1</sup> , Stephanie Gurka<sup>1</sup> , Martina Becker <sup>1</sup> , Nele Reeg<sup>1</sup> , Hans W. Mages <sup>1</sup> , Sebastian Voigt <sup>4</sup> , Christian Freund<sup>2</sup> and Richard A. Kroczek <sup>1</sup> \*

<sup>1</sup> Molecular Immunology, Robert Koch-Institute, Berlin, Germany, <sup>2</sup> Protein Biochemistry, Institute for Biochemistry, Free University of Berlin, Berlin, Germany, <sup>3</sup> Institute of Biochemistry, Charité University Medicine Berlin, Berlin, Germany, <sup>4</sup> Virology, Robert Koch-Institute, Berlin, Germany

XCL1 is the ligand for XCR1, a chemokine receptor uniquely expressed on crosspresenting dendritic cells (DC) in mouse and man. We are interested in establishing therapeutic vaccines based on XCL1-mediated targeting of peptides or proteins into these DC. Therefore, we have functionally analyzed various XCL1 domains in highly relevant settings in vitro and in vivo. Murine XCL1 fused to ovalbumin (XCL1-OVA) was compared to an N-terminal deletion variant lacking the first seven N-terminal amino acids and to several C-terminal (deletion) variants. Binding studies with primary XCR1<sup>+</sup> DC revealed that the N-terminal region stabilizes the binding of XCL1 to its receptor, as is known for other chemokines. Deviating from the established paradigm for chemokines, the N-terminus does not contain critical elements for inducing chemotaxis. On the contrary, this region appears to limit the chemotactic action of XCL1 at higher concentrations. A participation of the XCL1 C-terminus in receptor binding or chemotaxis could be excluded in a series of experiments. Binding studies with apoptotic and necrotic XCR1-negative cells suggested a second function for XCL1: marking of stressed cells for uptake into cross-presenting DC. In vivo studies using CD8<sup>+</sup> T cell proliferation and cytotoxicity as readouts confirmed the critical role of the N-terminus for antigen targeting, and excluded any involvement of the C-terminus in the uptake, processing, and presentation of the fused OVA antigen. Together, these studies provide basic data on the function of the various XCL1 domains as well as relevant information on XCL1 as an antigen carrier in therapeutic vaccines.

Keywords: dendritic cells, XCR1, XCL1, cross-presentation, antigen targeting

#### INTRODUCTION

Murine XCL1 is a chemokine of 93 amino acids, and has been originally identified as lymphotactin by Kelner et al. (1), while human XCL1 was found by us (2), and by Yoshida et al. (3). Mature murine XCL1 exhibits a high degree of homology to human XCL1 (also 93 aa), with 61% identity and 84% similarity, and both homologs have an identical structure [for alignment please refer to Geyer et al. (4)]. The XCL1/XCR1 chemokine ligand-receptor pair exhibits some special structural and functional features.

#### Edited by:

Diana Dudziak, Hautklinik, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Kristen J. Radford, The University of Queensland, Australia Veronika Lukacs-Kornek, Saarland University, Germany

> \*Correspondence: Richard A. Kroczek kroczek@rki.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 31 July 2018 Accepted: 14 November 2018 Published: 13 December 2018

#### Citation:

Kroczek AL, Hartung E, Gurka S, Becker M, Reeg N, Mages HW, Voigt S, Freund C and Kroczek RA (2018) Structure-Function Relationship of XCL1 Used for in vivo Targeting of Antigen Into XCR1<sup>+</sup> Dendritic Cells. Front. Immunol. 9:2806. doi: 10.3389/fimmu.2018.02806

XCL1 is secreted by activated NK cells, activated Th1 polarized CD4<sup>+</sup> T cells, and activated CD8<sup>+</sup> T cells, and cosecreted with IFN-γ, MIP-1α, MIP-1β, and RANTES and is thus part of the Th1 immune defense (5, 6). XCR1 is the only receptor for XCL1, and XCL1 is the only ligand of this receptor (7, 8). Thus, this ligand/receptor pair is monogamous, a rare feature in the world of around 50 chemokines.

The receptor XCR1 is exclusively expressed on a subset of dendritic cells (DC), the "cross-presenting" DC (and not elsewhere in the body), in the mouse, the rat, and the human (4, 9–13). This narrow expression spectrum is another unusual feature of the XCL1/XCR1 axis. XCR1<sup>+</sup> DC, earlier designated as CD8<sup>+</sup> DC in the mouse, were demonstrated to be particularly efficient in the uptake of cells stressed by (intracellular) infection (14–16). Moreover, CD8<sup>+</sup> DC have consistently been shown to excel in antigen "cross-presentation," in which exogenous antigen is not presented in the context of MHC class II to CD4<sup>+</sup> T cells, but instead shunted into the MHC class I pathway of antigen presentation to CD8<sup>+</sup> T cells (17–19). Given the secretion profile of XCL1, XCR1<sup>+</sup> DC can be regarded as a DC population closely cooperating with NK cells, Th1-polarized CD4<sup>+</sup> T cells, and CD8<sup>+</sup> T cells in the surveillance of stressed/ transformed cells for "danger" (16). The cross-presenting XCR1<sup>+</sup> DC are now also commonly referred to as cDC1.

In the past, we have employed this highly specific expression of XCR1 to target antigens into cross-presenting DC in vivo. In these experiments, ovalbumin (OVA), recombinantly fused to the C-terminal portion of murine XCL1 ("XCL1-OVA"), was highly efficient in inducing antigen-specific CD8<sup>+</sup> T cell cytotoxicity, when compared to untargeted OVA (20). These experiments demonstrated that XCL1 can be employed as a carrier for therapeutic vaccines intended to elicit potent antigen-specific T cell cytotoxicity in vivo.

Because of this therapeutic potential, we are interested in the structure-function relationship of various domains of XCL1. Like classical chemokines, XCL1 has a free N-terminus of around 10 amino acids (aa), which is followed by a structured core domain of around 60 aa containing a three-stranded antiparallel beta-sheet and a C-terminal alpha-helix (classical "chemokine fold"). The C-terminal portion of XCL1 of around 20 aa is, typical for chemokines, again unstructured (21). The C-terminus is highly conserved between mouse, rat, and human XCL1, and of unknown function.

XCL1 is the only chemokine with one disulphide bridge, while all other chemokines stabilize their tertiary structure by two disulphide bridges. Kuloglu et al. (22) have demonstrated in vitro that due to the lack of this second disulphide bridge, XCL1 can assume at some more extreme conditions (45◦C, no salt) an alternative conformation (which is exceptional in the chemokine world), which could imply a second function. This unusual feature raised the question whether the various domains of XCL1 functionally differ from classical chemokines or whether XCL1 has more than one function.

To fully understand the usefulness of XCL1 as a vector system for protein vaccines, we set out to systematically test the contribution of XCL1-domains on receptor binding, its chemotactic function, and on antigen processing and presentation to CD8<sup>+</sup> T cells in vivo. To this end, N-terminal and C-terminal deletion variants of XCL1-OVA (which we have previously used for antigen targeting in vivo [Hartung et al. (20), see above] were generated. Further, we also replaced the entire C-terminal domain of XCL1 with the C-terminal domain of viral XCL1 (vXCL1), a rat cytomegalovirus-encoded XCL1 homolog, which we have recently identified and characterized (4). vXCL1, which can be assumed to interfere with the immune defense, has a fully intact chemotactic activity on cross-presenting DC and mainly differs in its C-terminal portion from its rat homolog. We thus utilized the viral C-terminus in order to determine whether this domain in some ways contributes to the function of XCL1.

#### MATERIALS AND METHODS

#### Cloning, Expression, and Purification of XCL1-OVA and Its Structural Variants

The DNA fragments coding for the various XCL1-OVA constructs with a C-terminal Strep-tag (IBA, Germany) were cloned into the drosophila expression vector pRmHa-3 (23) by standard procedures. XCL1-OVA encoding plasmids were electroporated together with the plasmid phshs.PURO into drosophila Schneider SL-3 cells (24) using a Bio-Rad Gene Pulser (450 V and 500 mF). The phshs.PURO plasmid (kindly provided by M. McKeown, Salk Institute) allows selection of positive transfectants by puromycin. Clones from limiting dilution cultures of transfected SL-3 cells were induced with 1 mM CuSO<sup>4</sup> and analyzed for high protein production using either XCL1- or Strep-tag-specific ELISA. Positive clones were expanded in Insect-XPRESS medium (Lonza) on a shaker platform (100 rpm) in normal air at 27◦C. XCL1-OVA proteins were purified from supernatants using Strep-Trap HP columns from GE Healthcare according to the manufacturer's instructions. Protein concentration was determined by measuring OD<sup>280</sup> using a Nanodrop ND-1000 (Thermo Scientific). LPS content in all protein samples was <0.5 EU/mg protein.

#### Mice

C57BL/6 mice (8–12 week old) were used for experiments and cell isolation, unless indicated otherwise. B6.XCR1-lacZ (The Jackson Laboratory) are XCR1-deficient mice in which the XCR1 gene has been replaced by the β-Gal reporter gene; these mice were fully backcrossed (>10x) onto the C57BL/6 background. OT-I TCR-transgenic mice were crossed onto the B6.PL background to allow identification of CD8<sup>+</sup> T cells using the CD90.1 marker. All mice were bred under specific pathogenfree conditions in the animal facility of the Federal Institute for Risk Assessment (Berlin, Germany). All animal experiments were performed according to state guidelines and approved by the local animal welfare committee.

**Abbreviations:** DC, dendritic cells; OVA, ovalbumin; MFI, mean fluorescence intensity.

# Cell Isolation

Splenocytes were obtained by mashing spleens through 70µm cell sieves into PBS, followed by erythrocyte lysis with ACK Buffer (155 mM NH4Cl, 10 mM KHCO3, 0.1 mM EDTA).

#### Chemotaxis

To obtain sufficient DC for chemotaxis assays, C57BL/6 mice were injected s.c. at 2 sites, each site with 1.5 × 10<sup>6</sup> B16 cells secreting Flt3 ligand (25) for 9 days. DC were then enriched by cutting spleens into small pieces followed by digestion with Collagenase D (500µg/ml) and DNase I (20µg/ml, both Roche) for 25 min at 37◦C in RPMI 1640 containing 2% FCS (low endotoxin, Biochrom); EDTA (10 mM) was added for additional 5 min and cells were filtered through a 70µm nylon sieve (BD Falcon). DC were further enriched by centrifugation over a 1.073 g/ml density gradient (NycoPrep, Axis-Shield). For chemotaxis assays, 1 × 10<sup>6</sup> DC (purity ∼70%) were suspended in 100 µl chemotaxis medium (RPMI 1,640, 1% BSA, 50 mM β-ME, 100 mg/ml penicillin/streptomycin) and placed into the upper chamber of a 24-well Transwell system (6.5 mm diameter, 5-µm pore polycarbonate membrane; Corning Costar). The lower chamber was filled with chemotaxis medium to which recombinant XCL1-OVA or the XCL1-variants were added. After incubation in 5% CO<sup>2</sup> for 2 h at 37◦C, the number of migrated DC was determined by counting cells in the lower chamber using a flow cytometer. DC were identified by staining for XCR1, CD8, CD11c, and MHC II after gating out cells expressing B220 and CD3. The percentage of migrated cells was calculated by dividing the number of cells in the lower chamber by the number of input cells (number migrated cells/number input cells × 100).

# In vivo Proliferation of OT-I T Cells

Recipient C57BL/6 mice were adoptively transferred with splenocytes containing 1 × 10<sup>6</sup> OT-I (CD8+) resting T cells (negative for CD25, CD69, and ICOS). For proliferation analysis, OT-I cells were labeled with 5µM CFSE (Invitrogen) before transfer and analyzed 48 h after immunization using the CFSE dilution assay. Detection of OT–I T cells after adoptive transfer was with mAb to CD8 and CD90.1 after gating out CD4<sup>+</sup> T cells and B220<sup>+</sup> cells.

#### In vivo Cytotoxicity Assay

Animals were immunized with the indicated amounts of either XCL1-OVA or the respective variants together with 3 µg of LPS (Sigma), which was mixed with the protein variants before injection i.v.. 6 days later, 10 × 10<sup>6</sup> syngeneic splenocytes were pulsed with SIINFEKL peptide (GenScript) and labeled with 10µM CFSE (CFSEhigh) in vitro, while 10 × 10<sup>6</sup> splenocytes were left unpulsed and labeled with 1µM CFSE (CFSElow). Both preparations were injected together i.v. into immunized and control animals, and the CFSE signal was determined by flow cytometry 18 h later. Specific lysis was calculated using the following formula: specific lysis (%) = 100–([CFSEhigh immunized/CFSElow immunized]/[CFSEhigh control/CFSElow control]) × 100.

#### Antibodies and Staining Reagents

Hybridomas producing mAb recognizing CD4 (clone YTS 191.1), CD8 (53–6.72), CD11c (N418), CD44 (IM7.8.1), CD45R (B220 clone RA3-6B2), CD62L (MEL-14), Ly6G/C (RB6-8C5), and MHC class II (M5/114.15.2) were obtained from ATCC, CD90.1 (OX-7) from ECACC. Mab to CD69 (H1.2F3) was from Biolegend, mAb PD-1 (J43) from eBioscience. Anti-CD3 (KT3) was generously provided by H. Savelkoul, anti-CD25 (2E4) by E. Shevach. Generation of anti-XCR1 mAb MARX10 (13) and anti-ICOS mAb [MIC-280 (26)] has been described before. Mab MTAC-311 detects the C-terminal part of murine XCL1 (unpublished antibody and data). Generation of XCL1-StrepTag is described in Hartung et al. (20). The non-agonistic mAb MARX10 (mouse IgG2b, in the recombinant version IgG1) does not block the binding of XCL1 to XCR1. OT–I T cells were identified with mAbs to Vα2-TCR (B20.1, eBioscience) and Vβ5- TCR (MR9-4, BD Biosciences). StrepMAB Immo conjugated to Oyster 645 was from IBA Lifesciences.

### Flow Cytometry

Antibodies were titrated for optimal signal-to-noise ratio. To block unspecific binding to Fc-receptors, cells were pre-incubated with 100µg/ml 2.4G2 mAb. Standard staining with mAb was in PBS, 0.25% BSA, 0.1% NaN<sup>3</sup> for 25 min on ice. For exclusion of dead cells 4′ ,6-diamidino-2-phenylindole (DAPI) was added 5 min before measurement. Doublets and autofluorescent cells were excluded from the analysis. Data were acquired on a LSRII cytometer (BD Biosciences), and analyzed using FlowJo (Tree Star Inc.). DC were defined as CD11c<sup>+</sup> MHC-II<sup>+</sup> Lin<sup>−</sup> cells.

#### Stressed Cells Assay

P3X63Ag8.653 myeloma cells (ATCC) were cultured at a density of 2 × 10<sup>6</sup> cell/ml in complete RMPI1640 medium. Some cells were exposed to heat shock (52◦C for 15 min) and thereafter cultured at a density of 2 × 10<sup>6</sup> cells/ml in a 6-well plate overnight at 37◦C and 5% CO2. Then, 0.5 × 10<sup>6</sup> cells were transferred into 24-well culture plate wells and 1 µg or 2 µg of wt XCL1-OVA or one of its variants were added for the last hour of culture. The cells were washed and stained with mAb StrepMAB Immo conjugated to Oyster 645 (to detect bound XCL1-OVA) and Annexin V-Cy5 (BD Pharmingen) in a binding buffer (10 mM Hepes, 140 mM NaCl, 2,5 mM CaCl2, 1% NaN3). After a further washing step, cells taken up in binding buffer and were analyzed by flow cytometry, DAPI was added just before analysis.

# RESULTS

### Generation of XCL1-OVA and the Structural Variants Del-N7, Del-C7, Del-C17, and vCterm

Various formats of XCL1 recombinantly fused to OVA were generated in order to test for the impact of the various domains (i) on the function of murine XCL1, (ii) on the ability of XCL1 to target the antigen OVA into XCR1<sup>+</sup> DC in vivo, and (iii) on the capacity of XCR1<sup>+</sup> DC to process and cross-present the attached antigen. The design, production, and in vivo targeting capacity and specificity of XCL1-OVA, used here as the standard for comparison, has been described earlier (20). Del-N7 XCL1 is a variant lacking the first seven N-terminal amino acids (aa) of XCL1, but is otherwise identical to XCL1-OVA (**Figure 1A**). Del-C7 XCL1 lacks the C-terminal 7 aa, Del-C17 the C-terminal 17 aa of XCL1; in both deletion variants a glycine-serine linker (GGGGS) was introduced C-terminally in order to (partially) compensate for any size/positional effects (**Figure 1B**). vCterm XCL1 is a variant in which the 17 C-terminal aa of murine XCL1 were replaced by the 20 C-terminal aa of a rat cytomegalovirusencoded XCL1 homolog (4). All constructs contained a C-terminal Strep-tag to allow detection of the bound protein variants to XCR1. The various constructs are represented schematically in **Figure 1A**. For clarity, the C-terminal sequences of the XCL1-variants are also shown (**Figure 1B**). The constructs were used to express the protein in Schneider cells.

#### The N-Terminus of XCL1 Is Critical for Binding to XCR1

The binding of XCL1-OVA and its structural variants to its receptor XCR1 was determined by incubating splenocytes with carefully titrated concentrations of each reagent for 25 min on ice, followed by washing. The bound protein variants on XCR1<sup>+</sup> DC were then detected using an anti-Strep-tag mAb and flow cytometry. Some of the results (incubation of the cells at 2.5, 0.625, 0.16, and 0.04 µg/ml) are represented in histograms in **Figure 2A**. **Figure 2B** summarizes experimental data points obtained with all concentrations of the respective protein variants. Small concentrations of XCL1-OVA (0.04µg/ml) sufficed to achieve substantial binding, and saturation was achieved at around 0.5µg/ml. The Del-C7 and Del-C17 structural variants showed a binding pattern comparable to wildtype XCL1-OVA (**Figures 2A,B**). In contrast, the Del-N7 variant only bound at high concentrations of protein. At 2.5µg/ml, the binding efficiency of Del-N7 was comparable to binding of XCL1-OVA at around 0.05µg/ml, and thus was diminished around 50-fold. Incubation with vCterm XCL1 resulted in clearly stronger binding signals, when compared to XCL1-OVA (**Figures 2A,B**).

To control for unspecific signals, the binding experiments were repeated with cells from XCR1-deficient (XCR1−/−) mice on the same C57BL/6 background. As can be seen from **Figure 2C**, all variants exhibited a similar XCR1-unspecific binding (only at this high concentration, data not shown), with the exception of the Del-C17 variant.

These experiments determined that the first seven N-terminal aa of XCL1 have a major influence on the binding of XCL1 to its receptor, which can be partially compensated at high protein concentrations by other structural elements of XCL1. At the same time, the experiments excluded any significant contribution of the C-terminal 17 aa to the binding of XCL1 to XCR1. Binding studies with XCR1−/<sup>−</sup> dendritic cells further demonstrated that all binding of XCL1 to the DC is mediated by XCR1; the unspecific signals obtained at higher protein concentrations (2.5 µg) are apparently mediated by the C-terminal portion of XCL1 and are most likely of no major relevance in vivo.

# Effects of the Structural Variants on Chemotaxis

In order to determine the functional capacity of the XCL1- OVA structural variants, DC were enriched from splenocytes and tested at various concentrations of the variants for chemotaxis in a transwell system. All of the cells which have migrated into the lower chamber were quantitatively analyzed using flow cytometry; therefore the intensity of the dot-plots truly represents the number of migrated cells. The DC in the input cell population were composed of around 70% of XCR1<sup>+</sup> DC and 30% XCR1<sup>−</sup> DC (**Figure 3A**, leftmost dotplot in upper panel). Virtually no migration into the lower chamber was observed in the medium control, while both XCR1<sup>+</sup> and XCR1<sup>−</sup> DC migrated equally well to the chemokine CCL21, which was used as a positive control (**Figure 3A**, upper panels). Essentially only XCR1<sup>+</sup> DC migrated in response to the various XCL1-OVA constructs, with virtually no response of T cells, B cells or other XCR1<sup>−</sup> cells (data not shown). The migration of XCR1<sup>+</sup> DC to various concentrations of the XCL1-OVA standard (1 ng/ml−10,000 ng/ml) exhibited the bell-shaped curve typical for chemokines, with maximal chemotactic activity at 100 ng/ml (**Figures 3A,B**). A similar pattern was observed for Del-C7, Del-C17, and also for vCterm (**Figures 3A,B**). With Del-N7, barely any chemotaxis was observed up to 100 ng/ml and even at 1,000 ng/ml the efficiency did not reach the maximum seen with wt XCL1-OVA. Interestingly, by further increasing the concentration of Del-N7 to 10,000 ng/ml in the lower chamber, Del-N7 was chemotactically active above the levels seen with optimal amounts (100 ng/ml) of wt XCL1-OVA (**Figures 3A,B**). This was a consistent phenomenon throughout the experiments.

The results obtained in this functional experiment were congruent with the previous binding studies. All XCL1- OVA versions exhibiting good binding also induced effective chemotaxis. Since the Del-C17 variant was similarly active compared to XCL1-OVA, it can be concluded that the C-terminal 17 aa of XCL1 do not participate in the induction of chemotaxis and thus must have other function(s). The positive functional data obtained with Del-N7 XCL1 at very high concentrations indicate that the core domain of XCL1 between aa 8 and 76 contains all necessary structural elements to induce chemotaxis. The first 7 N-terminal aa apparently play a major role in the stabilization of ligand binding to the receptor for induction of chemotaxis. Interestingly, for unknown reason, this N-terminal stretch of XCL1 seems also to limit the signaling at high ligand concentrations.

#### Binding of XCL1-OVA and Its Variants Del-N7, Del-C7, Del-C17, and vCterm to Apoptotic and Necrotic Cells

As outlined in the introduction, XCL1 is an integral part of the Th1-defense. Given the secretion of XCL1 by activated NK cells, Th1-polarized CD4<sup>+</sup> T cells, and by activated CD8<sup>+</sup> T cells, we tested the hypothesis that secreted XCL1 could "mark" stressed cells and thus facilitate their uptake by crosspresenting DC. To test this hypothesis, we examined the binding of wt XCL1-OVA and its structural variants to stressed

cells. To this end, P3X63Ag8.653 myeloma cells obtained from standard culture (and thus without stress signals) were doublestained with DAPI and Annexin V and arbitrarily subdivided into populations designated as "live" (Annexin V<sup>−</sup> DAPI−), "apoptotic" (Annexin V<sup>+</sup> DAPIlow), "necrotic" (Annexin V<sup>+</sup> DAPI+), and "dead" (Annexin V<sup>−</sup> DAPI+) (**Figure 4A**). While live and dead cells did not exhibit a strong binding of the various reagents, apoptotic and necrotic cells bound each reagent to a substantial degree, with a rather uniform staining pattern (**Figure 4B**, background staining with StrepMAB-Immo in gray). When the P3X cells were subjected to thermal stress (52◦C for 15 min, followed by overnight culture), again both necrotic and apoptotic cells bound the various reagents in an uniform fashion (**Figures 4C,D**). To determine which component(s) of the constructs was responsible for the observed binding, additional experiments were performed. Live, apoptotic, necrotic, and dead cells were reacted with XCL1 or with XCL1- StrepTag and any bound reagent detected with a mAb directed to murine XCL1 (**Figure 4E**). Only apoptotic and necrotic cells gave the characteristic signal pattern, after incubation with either XCL1 or XCL1-StrepTag. Since both reagents gave a staining pattern similar to the other constructs (**Figure 4B**), any significant binding of StrepTag or OVA to stressed cells could be excluded. Together, these experiments determined that XCL1 was responsible for binding to stressed cells. Furthermore, it could be concluded that the structured core region of XCL1 was responsible for the binding to stressed cells, with no obvious contribution of the free N-terminal or C-terminal regions. Altogether, the data were compatible with the notion that XCL1 marks stressed cells.

### Induction of CD8<sup>+</sup> T Cell Proliferation and Cytotoxic Capacity After in vivo Targeting of Antigen With XCL1-OVA and Its Variants Del-N7, Del-C7, Del-C17, and vCterm

In the next experiments we tested whether the various structural elements of XCL1 influence targeting of antigen in vivo. To this end, fluorescently labeled OT–I T cells were adoptively transferred into syngeneic C57BL/6 mice. One day later, mice were immunized i.v. with various amounts (0.1 µg, 0.3 µg, 1 µg) of wt XCL1-OVA, or alternatively with equal amounts of its variants Del-N7, Del-C7, Del-C17, and vCterm without adjuvant; PBS was used as negative control. After 48 h, spleens were removed and the proliferation of the OT–I T cells determined through dilution of the fluorescence signal using flow cytometry. In two independent experiments, Del-C17 induced higher proliferation of the OT–I CD8<sup>+</sup> T cells at lower dosages, possibly reflecting its higher binding capacity

FIGURE 2 | Binding of XCL1 and the structural variants Del-N7, Del-C7, Del-C17, and vCterm to XCR1 expressed on primary dendritic cells. Splenocytes were incubated with carefully titrated (0.04, 0.08, 0.16, 0.315, 0.625, 1.25, and 2.5µg/ml) concentrations of all XCL1-OVA protein variants (all concentrations are given based on the XCL1-component of the constructs) for 25 min on ice, washed, and the bound protein was detected on Ly6G/C<sup>−</sup> CD3<sup>−</sup> B220<sup>−</sup> CD8<sup>+</sup> CD11c<sup>+</sup> MHC-II<sup>+</sup> cells using an anti-Strep-tag mAb StrepMAB-Immo and flow cytometry (red histograms). Background staining, without pre-incubation, using StrepMAB-Immo is shown in gray. (A) Signals obtained with XCL1-variants at 0.04, 0.16, 0.625, and 2.5µg/ml. (B) Graphical representation of the mean fluorescence intensity (MFI) obtained in flow cytometry with all protein concentrations of the XCL1-OVA variants used. (C) Signals (shown for 0.625 and 2.5/µg/ml) obtained on CD8<sup>+</sup> DC lacking XCR1. Data are representative of 3 independent experiments.

centrifugation were placed in the upper chamber of a transwell system, the composition of input DC is shown in the leftmost dotplot in the upper pannel. Various concentrations of wt XCL1-OVA and the structural variants were established in the lower chamber and migration of cells was allowed for 2 h. All concentrations are given based on the XCL1-component of the constructs. Thereafter, all cells from the lower chambers were quantitatively analyzed by flow cytometry after staining for XCR1 (mAb MARX10 binds to XCR1 independent of XCL1) and CD8 (mAb 53–6.2). Therefore, the number of events in the dot plots directly represent the number of cells detected. The effect of the negative (medium) and the positive controls (CCL21) is shown in the upper pannels of dot plots. (B) Quantitave evaluation of migrated XCR1<sup>+</sup> DC expressed as percentage of input XCR1<sup>+</sup> DC of the experiment shown in (A). One experiment representative of 2 independent experiments (each independent experiment was done in duplicate on the same day and the data were combined (mean ± SEM).

(see **Figure 2**). vCterm was somewhat less effective, while Del-C7 was comparable to the wt XCL1-OVA standard regarding CD8<sup>+</sup> T cell proliferation (**Figures 5A,B**). Del-N7 gave at 1 µg a very subtle proliferation signal (**Figure 5B**). This experiment demonstrated a comparably effective targeting of XCL1-OVA in vivo into DC by all XCL1-OVA variants with comparable

FIGURE 4 | Binding of XCL1-OVA and its variants Del-N7, Del-C7, Del-C17, and vCterm to apoptotic and necrotic cells. P3X63Ag8.653 cells were either (A,B,E) cultured at standard conditions without stress, or (C,D) subjected to thermal stress (52◦C for 15 min, followed by culture overnight). For the last hour of culture, 1 µg of wt XCL1-OVA or one of its variants were added to the culture. For analysis, the cells were washed, and stained with DAPI and AnnexinV to subdivide the cells into "live" (Annexin−DAPI−), "apoptotic" (AnnexinV+DAPIlow), "necrotic" (Annexin+DAPI+), and "dead" (Annexin−DAPI+) cells. (A) Gating and (B) staining of cells without thermal stress, (C) gating and (D) staining of cells after thermal stress, using anti-StrepMAB-Immo for signal detection (red histograms); background staining with StrepMAB-Immo without any preincubation is shown in gray histograms. (E) P3X63Ag8.653 cells were cultured under identical conditions, without thermal stress. For the last hour of culture, 1 µg of wt XCL1 or XCL1-StrepTag were added to the culture. Washing of cells and gating with DAPI and AnnexinV was as described above. Signal was detected with mAb MTAC-311 specific for murine XCL1 (red histograms), background signals with MTAC-311, without any preincubation, are shown in gray. Concentrations of XLC1-variants are based on the XCL1-component of the respective construct.

binding capacity in vitro. This observation excluded a major effect of the C-terminal portion of XCL1 on OVA processing and presentation.

#### Induction of Cytotoxic Capacity After in vivo Targeting of Antigen With XCL1-OVA and Its Variants Del-N7, Del-C7, Del-C17, and vCterm

To further test the in vivo functional capacity of T cells induced by the various formats of targeted antigen, C57BL/6 mice were immunized i.v. with titrated amounts of wt XCL1-OVA and its variants, which were injected together with 3 µg LPS. On day 6, an in vivo cytotoxicity assay was performed. As can be seen from **Figure 6**, Del-C7, Del-C17, and vCterm were similarly effective in inducing cytotoxicity as XCL1-OVA. The Del-N7 variant, at 3.3 µg, also induced modest cytotoxic activity. The cytotoxicity results were thus congruent with the proliferation data of OT– I T cells obtained earlier and excluded a major effect of the C-terminal portion of XCL1 on the induction of cytotoxicity in vivo.

#### DISCUSSION

Structures of many chemokines have been solved by NMR and X-ray crystallography. These studies revealed that despite low sequence homology, chemokines adopt a remarkably conserved tertiary structure consisting of a disordered N-terminus of 6–10 aa, a structured core region (chemokine fold), and a disordered C-terminus of variable length (27, 28). From a great number of structure-function studies a general concept evolved which assumes an initial specific binding of the chemokine fold-domain to the N-terminus of the receptor. In a second step, this initial interaction is stabilized by a subsequent integration of the Nterminal disordered domain of the ligand into the orthosteric pocket of the receptor. Additional studies suggested that this Nterminal domain of the chemokine ligand has signaling capacity. For example, an N-terminal deletion variant of MCP-19−<sup>76</sup> bound to its receptor with similar strength compared to the wildtype version MCP-11−76, but had lost all chemotactic activity (29). Based on many experiments of this type, the disordered N-terminal region of chemokines is generally regarded as a key signaling domain (27, 28).

In our work, the N-terminal deletion variant Del-N7 (XCL18−93) reduced the binding of murine XCL1 to its receptor XCR1 approximately 50-fold. This reduction in binding was accompanied by a similar reduction in chemotaxis. The capacity of XCL1 to bind to its receptor was thus directly correlated to its chemotactic action. These data are compatible with a stabilization of the ligand-receptor interaction and thus with the general concept. However, we also made the surprising observation that at very high concentrations (approximately 100-fold of the presumed physiological concentration), this N-terminal deletion variant still exhibited fully preserved chemotactic action. Thus, with murine XCL1 there is no indication of an important signaling element in the N-terminal disordered domain which would be required for chemotaxis, as suggested by the general concept. Interestingly, at these supra-physiological concentrations (10,000 ng/ml) the N-terminal deletion variant consistently induced higher chemotaxis, when compared to the wild-type XCL1 at its optimal concentration (100 ng/ml). This observation indicates that the chemokine fold of XCL1 contains all necessary structures to induce chemotaxis. Finally, wild-type XCL1 at the same supra-physiological concentrations (10,000 ng/ml), exhibited largely reduced chemotaxis compared to its optimum at 100 ng/ml (as is typical for chemokines). This observation suggest that the disordered N-terminal region of XCL1 in some ways limits the functional activity of XCL1 at high concentrations.

When we analyzed the functional contribution of the Cterminal portion of murine XCL1, the deletion variants Del-C7 (XCL11−86) and Del-C17 (XCL11−76) did not show any functional effects on receptor binding or chemotaxis. Thus, we can exclude a major contribution of this region to the chemotactic function of murine XCL1. This conclusion clearly differs from the findings of a study utilizing human XCL11−<sup>72</sup> (30), where a complete loss of calcium activity was observed. However, this particular C-terminal deletion variant was 4 aa shorter than Del-C17 (XCL11−76) which was used in the present study. Our results are fully compatible with data on a series of C-terminally truncated variants of human XLC1 (1–68, 1–72, 1–78, and 1–85), which all elicited normal calcium signals in XCR1-transfected HEK-293 cells (31).

Regarding the structure-function relationship of murine XCL1, our data can be summarized as follows. The core domain of XCL1 contains all necessary structural elements to specifically bind to XCR1 and to elicit chemotaxis. This observation differs from the general concept for chemokines, which assumes a critical signal contribution of the N-terminal domain for chemotaxis (27, 28). The first 7 aa of the N-terminal domain stabilize the binding of XCL1 to its receptor and thus certainly optimize chemotaxis. At the same time, and apparently paradoxically, the first 7 aa appear to limit the chemotactic effect of supra-physiological concentrations of murine XCL1, suggesting some type of negative regulatory role of the Nterminus at high XCL1 concentrations. A contribution of the disordered C-terminus of XCL1 to chemotaxis can clearly be ruled out.

These conclusions were reached with binding and chemotaxis assays using primary cells expressing the native XCR1 receptor. This is in contrast to the very few studies on the structurefunction relationship of XCL1, which were conducted with XCR1-transfectants and mainly based on calcium mobilization studies.

Since XCL1 is secreted by activated NK cells, we pursued the hypothesis that in addition to its chemotactic function, XCL1 could "decorate" stressed cells recognized by NK cells. In such a scenario bound XCL1 could mark these stressed cells for phagocytosis by DC specialized for uptake of such cells (16). Using a murine myeloma line as an in vitro model system and also employing heat-shock experiments, we consistently observed binding of XCL1 to cells characterized as "necrotic" or "apoptotic," based on the use of Annexin V and DAPI. This binding was clearly mediated by the chemokine fold of XCL1, with no apparent contribution of the disordered Nterminal or C-terminal regions. It is unclear at present, to which structural elements present on apoptotic and necrotic cells XCL1 binds. Therefore, it remains to be determined whether this binding is specific or mediated by structural elements common to many chemokines, e.g., domains capable to mediate binding to glycosaminoglycans (present in the core domain of chemokines, also with XCL1). Preliminary studies with primary cells gave similar results as with the myeloma line, but turned out to be less reproducible, and therefore more work is needed to reach firm conclusions here. In particular, in vivo work will be required to generate essential data in order to support or reject the "decoration" hypothesis.

We are interested to use XCL1 as a carrier to transport proteins or peptides into cross-presenting DC. Therefore, all experiments were performed with XCL1 variants which were Cterminally fused to full-length OVA. We wanted to determine whether the various domains of XCL1 exert any influence on the targeting of the model protein to XCR1<sup>+</sup> DC in vivo. As independent and very sensitive readouts for CD8<sup>+</sup> T cell activation we used both induction of proliferation (response by adoptively transferred OT-I T cells) as well as induction of cytotoxic activity (by endogenous CD8<sup>+</sup> T cells). Since CD8<sup>+</sup> T cells in vivo become activated through crosspresentation of the soluble antigen OVA (17–19), we assume that these readouts measure cross-presentation of OVA-derived peptide SIINFEKL by XCR1<sup>+</sup> DC in vivo. They thus reflect the combined effects of antigen uptake, efficiency of antigen processing, and antigen presentation on the cell surface of the DC. Previous experiments which demonstrated that targeting of OVA using either XCL1 or an antibody directed to murine XCR1 drastically reduces the amount of antigen necessary to elicit CD8<sup>+</sup> T cell responses in vivo (20) support this assumption.

Several conclusions can be reached from our experiments regarding the use of XCL1 as antigen carrier. First, an intact N-terminus is required to efficiently target any peptide/protein cargo to XCR1<sup>+</sup> DC. Second, attachment of a relatively large protein cargo of around 40 kDa to the C-terminus does not sterically inhibit binding of XCL1 to its receptor. Third, attachment of protein cargo does not influence chemotaxis of XCR1<sup>+</sup> DC (chemotaxis was identical when compared with native XCL1 without OVA, data not shown). Since a chemotactic signal usually induces internalization of the chemokine receptor (27, 28), we assume that fusion of protein cargo to XCL1 does not influence the uptake of the protein into XCR1<sup>+</sup> DC. Fourth, the disordered C-terminus can be eliminated from XCL1, if necessary, when using protein cargo without any obvious deficits in antigen uptake and presentation. The last conclusion is supported by exchanging the natural C-terminus of XCL1 with the C-terminal domain of murid herpesvirus 8 encoded XCL1, which also did not show any changes in antigen presentation.

The conclusions on the capacity of XCL1 as an antigen carrier were reached with the model antigen OVA. Since this particular antigen is ideal in that it is highly soluble and shows little interaction with other proteins, there may be some limitations to the conclusions reached. Other proteins prone to binding to other structures in the body may not as efficiently be transported to XCR1<sup>+</sup> DC as OVA. Other cargo proteins may also interact with XCL1 in an intra- or intermolecular fashion. In spite of these potential limitations, our data clearly demonstrate the usefulness of XCL1 as a carrier to directly target large proteins or peptides to XCR1<sup>+</sup> cross-presenting DC. Such an antigen carrier system appears attractive for induction of antigen-specific cytotoxicity in antitumor therapeutic vaccines.

# ETHICS STATEMENT

This study was carried out in accordance with the recommendations of name of guidelines, name of LAGS Berlin. The protocol was approved by the LAGS Berlin.

# AUTHOR CONTRIBUTIONS

ALK performed all experiments and wrote parts of the manuscript. EH, NR, MB, and SG assisted in some of the biological experiments. HWM assisted in the molecular biology experiments. SV contributed information on the viral XCL1. CF and RAK designed the experiments. RAK wrote the manuscript.

# ACKNOWLEDGMENTS

ALK thanks Prof. Elke Krüger (Charité) for continuing interest and support.

# REFERENCES


**Conflict of Interest Statement:** RAK is inventor on a patent held by the Robert Koch-Institute on targeting of antigens via XCR1.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Kroczek, Hartung, Gurka, Becker, Reeg, Mages, Voigt, Freund and Kroczek. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Using Dendritic Cell-Based Immunotherapy to Treat HIV: How Can This Strategy be Improved?

Laís Teodoro da Silva, Bruna Tereso Santillo, Alexandre de Almeida, Alberto Jose da Silva Duarte and Telma Miyuki Oshiro\*

Laboratorio de Investigacao em Dermatologia e Imunodeficiencias, Hospital das Clinicas HCFMUSP, Faculdade de Medicina, Universidade de São Paulo, São Paulo, Brazil

Harnessing dendritic cells (DC) to treat HIV infection is considered a key strategy to improve anti-HIV treatment and promote the discovery of functional or sterilizing cures. Although this strategy represents a promising approach, the results of currently published trials suggest that opportunities to optimize its performance still exist. In addition to the genetic and clinical characteristics of patients, the efficacy of DC-based immunotherapy depends on the quality of the vaccine product, which is composed of precursor-derived DC and an antigen for pulsing. Here, we focus on some factors that can interfere with vaccine production and should thus be considered to improve DC-based immunotherapy for HIV infection.

#### Edited by:

Daniela Santoro Rosa, Federal University of São Paulo, Brazil

#### Reviewed by:

Juliana Maricato, Federal University of São Paulo, Brazil Irina Caminschi, Monash University, Australia

\*Correspondence:

Telma Miyuki Oshiro telma.oshiro@hc.fm.usp.br

#### Specialty section:

This article was submitted to Vaccines and Molecular Therapeutics, a section of the journal Frontiers in Immunology

> Received: 27 July 2018 Accepted: 04 December 2018 Published: 18 December 2018

#### Citation:

da Silva LT, Santillo BT, de Almeida A, Duarte AJdS and Oshiro TM (2018) Using Dendritic Cell-Based Immunotherapy to Treat HIV: How Can This Strategy be Improved? Front. Immunol. 9:2993. doi: 10.3389/fimmu.2018.02993 Keywords: dendritic cells, HIV, immunotherapy, therapeutic vaccine, clinical trial

# INTRODUCTION

Although antiretroviral therapy has deeply improved the quality of life of HIV-infected individuals, some problems must be overcome, such as viral resistance, drug toxicities, therapeutic failure, and lack of drug access to viral reservoirs (1–4), which impair treatment effectiveness and patient adherence and hinder the discovery of functional or sterilizing cures.

In this context, harnessing dendritic cells (DC) to treat HIV infection is a promising strategy that has been extensively studied in recent years (5–21). The rationale for using DC is based on their essential role in the immune system, priming a specific immune response (22, 23). This strategy was initially tested for cancer treatment (24) and then for infectious diseases (5, 25), and more recently, tolerogenic DC have been evaluated for treating autoimmune diseases (26, 27).

Particularly in HIV infection, DC are qualitatively and quantitatively impaired in the host. In fact, HIV is able to evade innate immune sensing by DC, leading to suboptimal maturation that results in a poor antiviral adaptive immune response (28). Thus, the administration of properly sensitized DC can drive the immune response to a specific target, improving the anti-HIV-specific response.

More recent approaches have proposed a strategy using DC to reactivate HIV reservoirs (29, 30) together with a potent antiretroviral drug, which could finally promote the discovery of a long-awaited sterilizing cure. This timely and paramount approach encourages further studies on this type of intervention.

While many studies have demonstrated the high potential of DC-based strategies to stimulate an anti-HIV immune response in vitro (31–33), a systematic review of currently published trials concluded that in general, clinical outcomes have been modest, and the expected success rates have not been achieved (34). This outcome suggests that the full potential of this technique has not yet been realized, and opportunities to improve the efficacy of this strategy remain.

In many clinical trials, the patients' vaccine responses are very heterogeneous. Among patients enrolled in the same study and treated with the same strategy, some patients have good clinical outcomes, while others do not present a vaccine response (6, 20), which could be due to the individual characteristics of each patient and/or differences in the final vaccine product.

In this context, the efficacy of DC-based immunotherapy depends mainly on two factors: (i) the general condition of the patient, which is determined by genetic factors and clinical status; and (ii) the vaccine product, generally composed of monocyte-derived DC (MoDC) and the antigen used to pulse them (**Figure 1**).

Host genetic and clinical determinants will not be addressed in the present review. Instead, this review will focus on some factors that can interfere with vaccine production and should be taken into account to improve DC-based immunotherapy for HIV infection.

#### CLINICAL TRIALS

Clinical trials performed thus far have been phase I or phase II trials enrolling from four up to fifty-two patients, who received from one to thirty million DC per dose (5–21). To date, four clinical trials recruited treatment-naïve (5, 6, 11) or untreated (13) HIV-1-infected subjects, while 13 other studies enrolled patients on combined antiretroviral therapy (cART) in which the drug treatment was either interrupted (7, 8, 10, 15–17, 19– 21) or not interrupted (9, 12, 14, 18) after patients received the immunization. Analytical treatment interruption is useful to evaluate the effects of DC immunization on viral replication (**Table 1**).

Overall, immunotherapy trials present high variability in terms of the protocol used to obtain DCs, the number of doses, patient profiles and the immunization route. In this regard, the only commonality between currently published DCbased HIV vaccines is that the DC used in all protocols have been derived from monocytes because they are easy to obtain (**Figure 1**). However, despite the variability in study design, DC immunotherapy has been shown to be well-tolerated and safe, with only minor and transient side effects, including fever (8, 9, 14), enlargement of local lymph nodes (8, 13, 16), mild local redness (14–16) and flu-like symptoms (7, 13, 16).

Clinical outcomes were also highly variable between studies. In some, decreased plasma viral loads were observed in HIVinfected vaccinated individuals, but specific immune responses were usually transitory (6, 7, 11, 13, 16, 17). When the effects of immunotherapy on activation markers were monitored, CD38, and human leukocyte antigen (HLA)–DR expression increased on T cells (16, 20). In addition, eight out of seventeen trials showed that some individuals exhibited HIV-specific T cell responses that were not associated with decreased viral load or virologic control (5, 8, 10, 12, 15, 19–21).

These different outcomes may have been influenced by the protocols used to generate the vaccine products and their quality (MoDC maturation cocktail, HIV antigen used to pulse MoDC, quantities of cells inoculated per dose and number of doses administered) as well as patients' individual characteristics (e.g., CD4<sup>+</sup> T cell nadir (35), HLA alleles (36), and polymorphisms in genes involved in immune modulation (37–41).

The combination of factors discussed above may have affected the immune responses of vaccinated patients, which could explain why some of these individuals did not respond to immunotherapy (40).

# CHALLENGES IN MODC PREPARATION

#### Cell Precursors

Myeloid DC can be detected at a reduced frequency in peripheral blood. For immunotherapeutic protocols in which large numbers of cells are required, DC may alternatively be differentiated from precursors, such as CD34<sup>+</sup> hematopoietic progenitor cells and CD14<sup>+</sup> monocytes present in peripheral blood (42, 43).

Considering that the number of peripheral blood DC is low and that differentiation techniques require complex generation methods, only a small number of clinical trials, all related to cancer, have been published using DC generated from CD34<sup>+</sup> cells (44, 45); thus, MoDC are the most commonly used cells in a wide range of clinical applications (5–21, 46).

Monocytes are highly plastic cells that can alter their phenotype according to signals present in the microenvironment; for example, they may differentiate into MoDC under inflammatory conditions (43). MoDC have a high capacity for antigen presentation and naive T lymphocyte stimulation, similar to DC generated from CD34<sup>+</sup> cells (47).

Three circulating monocyte subsets have been described in human blood: classical (CD14++CD16−), intermediate (CD14++CD16+), and non-classical (CD14+CD16++) (48). Increased numbers of inflammatory CD16<sup>+</sup> monocytes are found in HIV-infected individuals (49, 50) and can act as targets for HIV entry through the highly expressed CCR5 (an HIV co-receptor); these monocytes may be more permissive to productive HIV infection than other monocyte subtypes (51).

While MoDC generated from CD16<sup>+</sup> monocytes secrete increased amounts of TGF-β1, MoDC generated from CD16<sup>−</sup> monocytes produce more of the IL-12p70 cytokine (52). In this context, considering that all three monocyte subtypes can be differentiated into MoDC in vitro and that the MoDC generated have distinct phenotypic and functional abilities (53), selection of the monocytic precursor may substantially influence vaccine performance. In fact, it was shown that "CD16+" MoDC-stimulated T cells produce more IL-4 than lymphocytes co-cultured with MoDC obtained from CD16<sup>−</sup> monocytes; therefore, "CD16+" MoDC can polarize the naive T cell response toward the Th2 phenotype (53). In the context of anti-HIV therapy, obtaining MoDC that secrete IL-12p70 is desirable for inducing IFN-γ-producing T lymphocytes (Th1 profile) (54). In the future, developing a clinical-scale procedure to enrich the non-classical monocyte subset could be a promising option.

Another important point to consider is the technique used to acquire peripheral blood cells. To obtain a large number of monocytes, leukapheresis is first performed, followed by an additional step to isolate or enrich the monocyte population

achieve a sufficient immune response against HIV combined with viral load control. In general, these can include elements related to the individual patient (A), such as genetic factors, clinical status, and drug treatment (cART interruption or not after receiving the immunization). In addition, the range of antigens available to pulse DC is extensive, making it a challenge to choose the best one (B). The factors related to the vaccine product (B,C) are just as important, including the choice of appropriate DC precursors (CD34+ cells or monocytes) and their differentiation/activation protocols (e.g., standard DC, α-DC1, IFN-DC), while also taking into account the potential of DC to produce exosomes (considering their role in regulation of the immune response) (C). In this context, proper assembly of each individual gear could achieve viral infection control and make possible the "functional cure" (D).

[elutriation (55) or positive purification by CD14<sup>+</sup> microbeads or adherence to plastic (56)].

During leukapheresis, a higher centrifuge speed yields residual platelets (57), which may subsequently attach to the monocytes and induce the production of cytokines (IL-1α and TNF-α) (58), pre-activating monocytes that could impair their differentiation into DC after in vitro stimulation. In addition, if leukapheresis itself leads to platelet activation, HIV-infected patients, even those receiving cART, may exhibit basal activation of these blood cells (59–61), which interfere with monocyte functionality and induce DC activation in vitro, affecting DC-mediated T lymphocyte polarization (62, 63).

Furthermore, if the monocytes are obtained by plastic adherence, the adhesion capacity of the platelets can reduce their yield, which subsequently reduces the yield of MoDC.

Considering the factors discussed above, vaccine product quality can be directly influenced by the first stages of DC


Frontiers in Immunology | www.frontiersin.org

TABLE

1


**232**


cART, combine antiretroviral

 therapy; IV, intravenous;

 SC, subcutaneous;

 ID, intradermal.

acquisition, such as the leukapheresis process and the monocyte subsets used.

#### MoDC Differentiation/Activation Protocols

MoDC-based immunotherapy requires custom conditions for producing mature MoDC capable of stimulating an appropriate immune response. For this reason, protocols should be guided by factors that contribute to viability, migration, co-stimulatory molecule expression, cytokine secretion, antigen presentation and T cell stimulation (64).

Although IL-4 and GM-CSF are used for MoDC differentiation in most studies, different concentrations or cytokine arrangements have been used in clinical trials (16, 17, 20), resulting in different vaccine products with variable performance once these cells present high plasticity.

Another important point to consider is maturation/activation stimuli. Correct insight is fundamental because the product has the potential to "educate" MoDC behavior. The commonly used maturation cocktail for MoDC comprises the proinflammatory cytokines TNF-α, IL-1ß, and IL-6 combined with PGE2, which was established as the "gold standard" for MoDC maturation (the so-called "standard DC" or "sDC") (65). sDC upregulate major histocompatibility complex (MHC) class I and II molecules, costimulatory molecules and CCR7 but fail to induce IL-12p70 production, probably due to PGE2 (66–68). The removal of PGE2 from these cocktails generates MoDC with similar profiles but low CCR7 expression and subsequent decreased migration to the lymphoid organs (69). The combination of different cytokines induces distinct responses, reflecting the complexity involved in establishing an effective protocol. In fact, sDC (with or without PGE2) have been adopted in most MoDC-based HIV immunotherapy protocols (**Figure 1**).

To improve the performance of sDC, alternative strategies have been developed. For example, type I and II interferons have been used to supplement standard activation stimuli to obtain polarized DC, called alpha-type-I polarized DC (α-DC1), driving a potent Th1 response (54). Recently, α-DC1 were used in a clinical trial for the treatment of HIV-infected individuals after stimulation with autologous HIV-infected apoptotic cells (ApB-DC vaccine). Although safe and immunogenic, only a modest decrease in the HIV-1 RNA load set point was observed after vaccination, and this was not sustained after cART discontinuation. Suboptimal DC function, evidenced by low IL-12 production, was attributed to this modest outcome (20).

Recently, a cancer research group developed "selfdifferentiated myeloid-derived antigen-presenting cells reactive against tumors DC" (Smart-DC), which are generated via the genetic reprogramming of monocytes. For production, monocytes are transduced with lentiviral vectors co-expressing GM-CSF and IL-4 and a melanoma self-antigen, allowing their self-differentiation into DC, which express typical DC surface molecules and stimulate antigen-specific CTL responses (70). This interesting and innovative approach could be a potential option for future antiviral immunotherapy applications.

Another promising strategy is a preclinical evaluation of an mRNA-electroporated MoDC-based therapeutic vaccine against HIV-1-encoding activation signals (TriMix: CD40L + CD70 + caTLR4 -activated form of TLR4) combined with rationally selected antigen sequences of Gag, Pol, Vif and Nef (HTI—HIV T cell immunogen). In vitro assays demonstrated MoDC with appropriate maturation profiles and the ability to induce T cell responses, especially CD8<sup>+</sup> T cells (71).

Overall, these studies aim to reach the same goal of developing a protocol that can induce the best MoDC capable of eliciting a sufficiently potent immune response that is reproducible in vivo and also controls viral replication. Several combinations of differentiation and activation factors are available, but the search for an ideal MoDC continues, and a gold standard protocol for generating successful MoDC-based therapeutic vaccines for HIV-infected individuals has not been established.

#### Exosomes

Exosomes have emerged as potential modulators of the immune response in a DC-based immunotherapy context. As a type of extracellular vesicle, exosomes are endocytic-originating small particles (30-100 nm in diameter) composed of lipids that are released by cells into the extracellular environment by the fusion of internal multivesicular compartments. Exosomes participate in intercellular communication via the transfer of a variety of molecules, such as lipids, proteins, DNA, mRNA, and microRNA (72–74).

Many cells, such as neurons, tumors and immune cells, are capable of releasing exosomes. In particular, DC-derived exosomes can express class I and II MHCs, adhesion and costimulatory molecules, enabling their ability to directly activate CD8 and CD4 T cells (75–77).

Interestingly, in the context of HIV infection, the exosome dissemination pathway converges to capture and transfer HIV particles via mature DC, suggesting that HIV exploits this pathway to mediate T lymphocyte transinfection (78). Additionally, exosomes derived from HIV-infected DC can transmit HIV to T cells (79) and are capable of inducing the activation of resting primary CD4<sup>+</sup> T lymphocytes as well as reactivating the HIV-1 reservoir (80). These findings illustrate the close relationship between exosomes and HIV in the DC therapy context.

Considering that whole HIV particles are used in some DCbased immunotherapy protocols, the role of exosomes in DC performance should be considered. After pulsing DC with HIV, DC-derived exosomes were shown to induce the apoptosis of neighboring CD4 T lymphocytes, which has the potential to impair specific anti-HIV immune responses (81). In line with this, preliminary data suggested that exosomes may play a role in modulating the immune response during anti-HIV DC-based immunotherapy. Using gene expression analysis of an exosome marker, it was hypothesized that low exosomal release is more beneficial for DC-based immunotherapy responses than high exosomal release (82), which is an important point that should be considered when using whole viral particles to pulse DC.

Although systematic analyses have not considered the role of exosomes in anti-HIV DC-based clinical trial outcomes, studies have demonstrated a relevant function for these vesicles in immune response regulation. In the context of cancer research, if on one hand tumor-derived exosomes can directly activate specific immune responses and improve anti-cancer responses (83), on the other hand these exosomes can create an immunosuppressive pro-tumorigenic microenvironment, which allows the disease to progress (84). Similarly, in an anti-HIV context, this duality may also be present and should be considered in future trials.

#### ANTIGENS AND CHALLENGES IN DC-LOADING STRATEGIES

A fundamental aspect of DC-based immunotherapy is the selection of the antigen to be incorporated, a decision that must consider the safety and efficacy originating from the effects of the antigen's interaction with DC during the pathogenesis of infection in addition to knowledge of HIV structure.

Although whole particles (inactivated or attenuated) advantageously have greater epitope diversity, they have pathogenic potential due to the virus-cell interaction.

Attenuated viral particles represent one of the most potent known immunogens (85). Research on and development of an anti-HIV vaccine has previously incorporated attenuation procedures, but this approach was abandoned for safety reasons (86). With modern technologies, research examining this process has resumed, and such a strategy potentially represents another DC-based immunotherapy option (87, 88).

The deleterious effects arising from the interaction of viral particles with DC can be minimized by using killed whole HIV particles as an antigen. The consequences of such an approach will depend on the methodology used for chemical or heat inactivation (89–91), which may or may not alter the virus structure and will also influence the type of immune response induced (92).

When using HIV fragments as antigens, the risk of deleterious effects on the individual is lower (although persistent) relative to that posed by using the entire viral particle. However, an even more significant challenge associated with using HIV fragments is the selection of which one to use. The lack of immune protection correlates in HIV infection imposes an unprecedented degree of difficulty on this definition when determining the composition of the product for intervention (93, 94).

Groups have studied the immunogenic potential of DC transfected with mRNA encoding HIV proteins in vitro (71, 95, 96) and in vivo (12, 14, 15, 18, 19, 21). Some advantages of using RNA as an antigen include the absence of a biological risk of infection and the possibility of designing a sequence restricted to targeting MHC class I or II molecules, thus activating immune responses to CD8<sup>+</sup> T cells or both CD4<sup>+</sup> and CD8<sup>+</sup> T cells, respectively.

Another strategy is the use of the HIV fragments Gag, Tat, Rev, Nef and Vpr, which are commonly employed because they are more conserved than other proteins and are known to induce a T cell immune response (97). However, Nef interferes with DC functionality (98); for this reason, reduced quantities of Nef RNA are used in immunotherapeutic protocols. Similarly, the Vpr gene is truncated to remove its ability to impair DC expression of co-stimulatory molecules and production of the cytokine IL-12 (12, 99).

In clinical trials, both consensus viral sequences for a cohort of patients (14, 15, 18) or mRNA constructs personalized for each individual (12, 19, 21) have been used as immunogens. Although researchers have observed the induction of HIV-specific T cell responses in vivo as increases in T cell proliferation (12, 14) and enhancements in effector/memory CD8<sup>+</sup> T cell responses (15, 19, 21), there has been no sustained impact on patient viral load.

Another strategy is loading DC with DNA encoding HIV antigens. The use of DNA constructs as antigen vectors may overcome difficulties associated with MHC haplotype and peptide mismatches. Additionally, DNA can express antigens with natural posttranslational modifications (100). Moreover, DNA vaccines present several advantages, such as safety, potential to elicit both humoral and cellular immune responses and low cost (101). In this sense, plasmid DNA can be efficiently combined with DC to induce a specific immune response, as demonstrated in vitro (33, 102), representing a promising strategy to improve DC-based vaccines.

### SUMMARY

Overall, this review highlights some emerging factors that should be considered to improve the production of vaccines for anti-HIV DC-based immunotherapy protocols. Approaches to improve anti-HIV immunotherapy are complex and challenging. HIV infection is characterized by a chronic immune activation state, a consequence of intense immune stimulation and sustained inflammation, which promotes massive immune cell loss (103, 104). Given that immunotherapy aims to induce the patient's immune response, treatment requires an equilibrium between stimulating a specific immune response to fight the virus and limiting the immune activation state to avoid "adding fuel to the fire." Additionally, depending on the patient's clinical status, DC precursors as well as effector cells may be committed, which could interfere with the performance of vaccine products.

# AUTHOR CONTRIBUTIONS

TO conceived the study. AdA, BS, LdS and TO discussed, wrote, and edited the manuscript, and BS also contributed to figure construction. AD provided intellectual guidance. All authors have read and approved the final manuscript.

# FUNDING

This work was supported by the Sao Paulo Research Foundation—FAPESP, Brazil (grant number 2017/22131- 0 and 2016/25212-9). BS is a recipient of Coordenacao de Aperfeicoamento de Pessoal de Nível Superior—Brasil (CAPES)–Finance Code 001, and LdS is a recipient of FAPESP (grant number 2018/12460-0).

# REFERENCES


cells transfected with mRNA encoding HIV-1 gag and Nef: results of a randomized, placebo-controlled clinical trial. JAIDS (2016) 71:246–53. doi: 10.1097/qai.0000000000000852


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 da Silva, Santillo, de Almeida, Duarte and Oshiro. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Fas Signaling in Dendritic Cells Mediates Th2 Polarization in HDM-Induced Allergic Pulmonary Inflammation

Miaomiao Han1,2,3†, Ran Hu2†, Jingyu Ma<sup>2</sup> , Baohua Zhang<sup>2</sup> , Ce Chen<sup>4</sup> , Huabin Li <sup>1</sup> \*, Jun Yang<sup>3</sup> \* and Gonghua Huang2,4 \*

<sup>1</sup> Department of Otolaryngology-Head and Neck Surgery, Center for Allergic and Inflammatory Diseases, Affiliated Eye and ENT Hospital, Fudan University, Shanghai, China, <sup>2</sup> Shanghai Institute of Immunology, Shanghai Jiao Tong University School of Medicine, Shanghai, China, <sup>3</sup> Xin Hua Hospital Affiliated to Shanghai Jiao Tong University School of Medicine, Shanghai, China, <sup>4</sup> Guangdong Provincial Key Laboratory of Medical Molecular Diagnostics, Guangdong Medical University, Dongguan, China

#### Edited by:

Silvia Beatriz Boscardin, University of São Paulo, Brazil

#### Reviewed by:

Yusei Ohshima, University of Fukui, Japan Alexandre Castro Keller, Federal University of São Paulo, Brazil

#### \*Correspondence:

Huabin Li allergyli@163.com Jun Yang yangjun@xinhuamed.com.cn Gonghua Huang gonghua.huang@shsmu.edu.cn

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 17 August 2018 Accepted: 10 December 2018 Published: 21 December 2018

#### Citation:

Han M, Hu R, Ma J, Zhang B, Chen C, Li H, Yang J and Huang G (2018) Fas Signaling in Dendritic Cells Mediates Th2 Polarization in HDM-Induced Allergic Pulmonary Inflammation. Front. Immunol. 9:3045. doi: 10.3389/fimmu.2018.03045 Fas–Fas ligand (FasL) signaling plays an important role in the development of allergic inflammation, but the cellular and molecular mechanisms are still not well known. By using the bone marrow-derived dendritic cell (BMDC) transfer-induced pulmonary inflammation model, we found that house dust mite (HDM)-stimulated FAS-deficient BMDCs induced higher Th2-mediated allergic inflammation, associated with increased mucus production and eosinophilic inflammation. Moreover, FAS-deficient BMDCs promoted Th2 cell differentiation upon HDM stimulation in vitro. Compared to wild-type BMDCs, the Fas-deficient BMDCs had increased ERK activity and decreased IL-12 production upon HDM stimulation. Inhibition of ERK activity could largely increase IL-12 production, consequently restored the increased Th2 cytokine expression of OT-II CD4<sup>+</sup> T cells activated by Fas-deficient BMDCs. Thus, our results uncover an important role of DC-specific Fas signaling in Th2 differentiation and allergic inflammation, and modulation of Fas signaling in DCs may offer a useful strategy for the treatment of allergic inflammatory diseases.

Keywords: allergic inflammation, dendritic cells, Fas, house dust mite, Th2

# INTRODUCTION

Allergic inflammation has been generally considered as a T helper (Th) 2-mediated chronic immune response (1). Th2 cells produce effector cytokines such as IL-4, IL-5, and IL-13 to mediate the respiratory symptoms. Among these effector cytokines, IL-4 is involved in IgE synthesis and IL-5 can drive eosinophilia in lung tissue, while IL-13 contributes to mucus overproduction, airway hyper-responsiveness (AHR), goblet cell metaplasia and airway remodeling (1–3). House dust mite (HDM) has been reported to cause 50–85% of allergic asthmatic inflammation (4). The polarization of a Th2-mediated immune response to inhaled allergens (such as HDM) is determined by the status of dendritic cells (DCs). DCs promote Th2 differentiation through upregulation of the expression of several costimulatory molecules such as CD86, OX40L and polarization cytokines including IL-6, IL-10, and IL-33 (5–8), as well as Th2-cell-attracting chemokines, such as CCL17 and CCL22 (9). Although accumulating evidence suggests that DCs are sufficient and necessary to initiate Th2 responses, the underlying signaling mechanism for DCs to direct Th2 differentiation and function is still not wellunderstood.

Fas (CD95, also named APO-1) signaling is widely considered to mediate apoptosis upon binding to its ligand (FasL, also called "CD95L or APO-1L") or its agonist antibody (10). In an ovalbumin (OVA)-induced mouse asthma model, Fas-deficient mice have delayed resolution of airway hyperresponsiveness (AHR) compared to wild-type mice (11). Further study indicates that Fas deficiency in T cells contributes to the prolonged resolution of airway inflammation (12). Recent studies have shown that Fas–FasL interaction could also activate nonapoptotic pathways, such as Fas signaling leading to T cell activation, proliferation and differentiation (13) and promoting Th17 polarization and Th17-mediated autoimmunity (14). Fasdeficient mice sensitized with OVA increase the expression of IL-4, IL-5, and IL-13 compared to wild-type mice (15). Although DCs are pivotal in regulating T cell activation, proliferation, differentiation and allergic inflammation, the role of Fas signaling in DCs in driving Th2 differentiation and Th2-mediated allergic inflammation still need to be elucidated.

In the present study, we used the well-established model by adoptively transferring HDM-pulsed BMDCs to recipient mice to explore the role of FAS signaling in DCs in pulmonary inflammation. We found that Fas deficiency in DCs led to increased mucus production, eosinophilic inflammation and Th2 response in vivo. Fas-deficient BMDCs promoted the production of Th2-related cytokines such as IL-4 and IL-13. Further mechanistic study showed that DCs directed Th2 differentiation by modulating the Fas–ERK–IL-12 axis. Collectively, our results identify an important signaling mechanism of DC-mediated Th2 responses and modulation of Fas signaling in DCs might offer a useful strategy for the treatment of eosinophilic lung inflammatory diseases.

# MATERIALS AND METHODS

#### Mice

B6.MRL-Tnfrsf6lpr mice were from The Jackson Laboratory. C57BL/6 mice were from Shanghai SLAC Laboratory Animal Center (Shanghai, China). All mice were kept in a specific pathogen-free (SPF) barrier facility maintained by Shanghai Jiao Tong University School of Medicine. All the experimental mice were used at 6–10 weeks. Animal protocols were approved by Institutional Animal Care and Use Committee of Shanghai Jiao Tong University School of Medicine.

#### BMDC Culture

BMDCs were cultured as previously described (16). Briefly, bone marrow cells were collected by perfusing mouse femur and tibia. After red blood cell lysis, cells were cultured in RPMI 1640 (Invitrogen Corp.) supplemented with 10% fetal calf serum (FCS, Gibco), penicillin and streptomycin (Invitrogen), 2-mercaptoethanol (Sigma-Aldrich), supernatant from J5 cells (provided by Dr. Qibin Leng, Institut Pasteur of Shanghai, China) expressing GM-CSF (1:50) and 10 ng/ml IL-4 (R&D). On day 3, the entire medium was removed and replaced with fresh differentiation medium. On day 7, the cells were harvested for analyses. The purity of CD11c<sup>+</sup> BMDCs was >80%.

# BMDC Adoptive Transfer Experiment

In BMDC adoptive transfer experiment, wild-type or Fasdeficient BMDCs were pulsed with 50µg/ml HDM (Greer Laboratories, Lenoir, NC) for 12 h, washed, and then 1 × 10<sup>6</sup> HDM-pulsed BMDCs were administered intravenously into naïve C57BL/6 recipients. Recipients transferred with un-pulsed wild-type or Fas-deficient BMDCs as control. On day 10– 12, all the recipients were lightly anesthetized and challenged intranasally with 10 µg HDM in 40 µl PBS and the mice were sacrificed for analysis on day 13.

#### Antibodies and Flow Cytometry

Anti-mouse CD11c (N418), MHC-II (M5/114.15.2), CD11b (M1/70), Ly6G (RB6-8C5), siglec F (E50-2440), CD4 (RM4- 5), TCRβ (H57-597), IL-4 (11B11), IL-13 (eBio13A), IL-17A (eBio17B7), CD25 (PC61.5), CD44 (IM7), CD62L (MEL-14), CD69 (H1.2F3), Ki-67 (SolA15), IL-12p40 (C17.8), CD45 (30- F11), CD178 (MFL3) antibodies were obtained from eBioscience. Anti-mouse IL-5 (TRFK5) was obtained from BD Biosciences. Anti-mouse CCR3 (J073E5) was obtained from Biolegend. For surface staining, cells were stained with antibodies in PBS containing 1% FCS (Hyclone) on ice for 30 min. For intracellular staining, cells were stimulated with PMA (Sigma-Aldrich) and ionomycin (Sigma-Aldrich) for 5 h in the presence of Golgistop (BD Biosciences) before being stained according to the manufacturer's instructions (eBioscience). CD4+T cell proliferation was detected by anti-Ki-67 staining according to the manufacturer's instructions (eBioscience) 3 days after co-culture. Labeling OTII CD4<sup>+</sup> T cells with CFSE (carboxyfluorescein diacetate succinimidyl diester; Invitrogen), cell proliferation was detected by flow cytometry 4 days after co-culture. For cell apoptosis analysis, cells were stained with CaspACETM FITC-VAD-FMK in situ Marker (Promega). The samples were acquired on a FACSCantoII (BD) or LSRFortessaTM X-20 (BD) and analyzed with FlowJo software (Treestar).

#### Bronchoalveolar Lavage Fluid (BALF) Collection and Lung Mononuclear Cell Isolation

For BALF collection, lung tissues were lavaged with 1 ml cold PBS for 3 times and the supernatant was collected. Lung mononuclear cells were prepared as previously described (17). Briefly, lung tissues were removed, minced and digested with 1 mg/ml Collagenase IV (Life Technologies) in RPMI-1640 (Hyclone) with 5% FCS (Hyclone) for 45 min at 37◦C. Cells were enriched by using 38% Percoll gradient (GE Healthcare Life Sciences). Red blood cells were lyzed with ACK lysis buffer (R&D Systems). Cells were harvested for analyses.

#### Histology

Left lobe of lung tissues were removed from mice after BALF collection, fixed with 4% paraformaldehyde (PFA) at room temperature for 24 h and embedded in paraffin, cut into 5-µm sections for hematoxylin and eosin (HE) or periodic acid-Schiff (PAS) staining. The lung inflammation was blindly quantified using HE-stained sections according to the criteria previously published (18). The quantification of the goblet cell hyperplasia in the airway was done as previously described (19).

# RNA Isolation and Quantitative PCR (qPCR)

Total RNAs of lung tissues were isolated using Trizol regent (Invitrogen). Total RNAs of cells were isolated using RNeasy mini Kit (QIAGEN) according to the manufacturer's instructions. 1 µg total RNA was used for reverse transcription with PrimeScript RT Master Mix (TAKARA) according to the manufacturer's instructions in a total volume of 20 µl. qPCR was carried out with SYBR Green PCR Master Mix (Applied Biosystems) in a Vii7 Real-Time PCR system (Applied Biosystems). mRNA expression of genes was normalized to Hprt. The primers shown below were from Primerbank: Hprt, forward primer: TCAGTCAACGGGGGACATAAA, reverse primer: GGGGCT GTACTGCTTAACCAG; Il4, forward primer: GGTCTCAAC CC CCAGCTAGT, reverse primer: GCCGATGATCTCTCT CAAGTGAT; Il17a, forward primer: TCAGCGTGTCCAAAC ACTGAG, reverse primer: CGCCAAGGGAG TTAAAGACTT; Ifng, forward primer: GCCACGGCACAGTCATTGA, reverse primer: TGCTGATGGCCTGATTGTCTT; Il6, forward primer: CTGCAAGAGACTTCCATCCAG, reverse primer: AGTGGT ATAGACAGGTATGTTGG; Il12p35, forward primer: CAATCA CGC TACCTCCTCTTT, reverse primer: CAGCAGTGCAGG AATAATGTTTC; Il12p40, forward primer: GTCCTCAGAAGC TAACCATCTC, reverse primer: CCAGAGC CTATGACTCCA TGTC; Il10, forward primer: CTTACTGACTGGCATGAGGAT CA, reverse prime: GCAGCTCTAGGAGCATGTGG. The other primers were used as described, such as Il5, Il13 (20), Gata3 (21), Il9 (22), Tnfsf4 (8), Cd86 (23), Tslp (24).

#### Cell Stimulation and Culture

BMDCs were stimulated with HDM in the presence or absence of Fas agonistic antibody Jo2 (1µg/ml, BD) or Fas antagonistic antibody kp7-6 (1 mM, Merk) for 5 h for RNA analysis. For drug inhibitor treatments, cells were incubated with vehicle (DMSO) or U0126 (10µM) (from Calbiochem) for 0.5–1 h before adding other stimuli. For BMDC–T cell co-culture, BMDCs and flow cytometry-sorted naïve OT-II CD4<sup>+</sup> T cells (CD4+25−CD44−CD62L+, purity >99%) were mixed at a ratio of 1:10 in the presence of OVA323−<sup>339</sup> peptide and HDM, and then the CD4<sup>+</sup> T cells were harvested at 48 h for mRNA analysis or supernatant was harvested at 72 h for ELISA. For cytokine treatment, cultures were supplemented with 1 ng/ml IL-12p70 (R&D).

#### Protein Analysis

Concentrations of IL-4 and IL-13 were measured by ELISA according to the manufacturer's instructions (R&D; eBioscience). Read the OD values at 450 nm on the MultiSKAN GO microplate reader (Thermo). Immunoblot analysis was performed as described (25) with antibody to ERK phosphorylated at Thr202 and Tyr204, antibody to p38 phosphorylated at Thr180 and Tyr182, antibody to JNK phosphorylated at Thr183 and Tyr185 and antibody to ERK (all from Cell Signaling Technology), antibody to alpha tubulin (Proteintech).

# Statistical Analysis

All statistical analyses were performed by unpaired Student's ttests or ANOVA using GraphPad Prism software (version 5.0). P < 0.05 was considered significant. Results represent means ± SEM.

# RESULTS

#### Fas-Deficient BMDCs Enhance HDM-Induced Pulmonary Inflammation

To explore the role of Fas signaling in DCs in the regulation of HDM-induced allergic inflammation in mice, we used a BMDC adoptive transfer protocol to induce lung inflammation (**Figure 1A**). We transferred HDM-pulsed or un-pulsed wildtype or Fas-deficient BMDCs into naïve wild-type recipient mice. After HDM re-challenged, mice received HDM-pulsed BMDCs showed higher total cell number in the bronchoalveolar lavage (BAL) compared to mice received un-pulsed BMDCs (**Figure 1B**). A significantly increased total cell number was also observed in the BAL of recipients transferred with HDM-pulsed Fas-deficient BMDCs (**Figure 1B**). We also observed higher inflammatory cell infiltration and mucus production in lung tissues of recipients transferred with HDM-pulsed Fas-deficient BMDCs than those transferred with HDM-pulsed wild-type BMDCs (**Figures 1C,D**). Flow cytometry showed that a dramatically increased eosinophil infiltration both in the BAL and lung tissues of recipients transferred with HDMpulsed Fas-deficient BMDCs compared to those transferred with HDM-pulsed wild-type BMDCs (**Supplementary Figures 1A,B**). We also analyzed the inflammatory eosinophils (iEos) (CD45+Siglec FintCCR3+CD62L−) and resident eosinophils (rEos) (CD45+Siglec FintCCR3+CD62L+) in lung tissues (26). The cell number of iEos was increased in recipients transferred with HDM-pulsed Fas-deficient BMDCs compared to those transferred with HDM-pulsed wild-type BMDCs, but rEos was comparable between recipients transferred with HDM-pulsed wild-type BMDCs and those transferred with Fas-deficient BMDCs (**Figure 1E**). However, the neutrophil infiltration had no difference between the two groups (**Supplementary Figures 1C,D**). Together, these data indicate that HDM-pulsed Fas-deficient BMDCs can induce more severe allergic airway inflammation than HDM-pulsed wild-type BMDCs.

# Fas Signaling in BMDCs Does Not Affect CD4<sup>+</sup> T Cell Proliferation, Apoptosis, and Activation in vivo

In addition to a role of innate immune cells in the development of allergic inflammation, the adaptive immune system also play an important role in driving and sustaining this inflammation (27). A comparable CD4<sup>+</sup> T cell activation was observed in lung tissues of the recipients transferred with HDM-pulsed wild-type or Fas-deficient BMDCs (**Figure 2A**). We also

examined whether Fas-deficient in BMDCs could affect CD4<sup>+</sup> T cell proliferation or apoptosis in vivo, we performed Ki-67 and VAD staining assay, respectively. Our results showed that the proportion of Ki-67+CD4<sup>+</sup> T cells and the median fluorescence intensity (MFI) value of VAD+CD4<sup>+</sup> T cell in lung tissues and mediastinal lymph nodes (mLN) had no

differences between the recipients transferred with HDMpulsed wild-type BMDCs and those transferred with Fasdeficient BMDCs (**Figures 2B–E**). Taken together, these results show that Fas signaling in BMDCs is not required for CD4<sup>+</sup> T cell activation, proliferation and apoptosis upon HDM treatment.

# Fas-Deficient BMDCs Promote Th2 Responses Upon HDM Treatment

Given that there was no difference in the proliferation and apoptosis of CD4<sup>+</sup> T cells in the recipients transferred with HDM-pulsed wild-type or Fas-deficient BMDCs, we reasoned that the difference in inflammatory response might be due to the potential of these cells to produce inflammatory cytokines. Thus, we first measured cytokine expression in lung tissues of these two group mice. The qPCR results showed that the recipients transferred with HDM-pulsed Fas-deficient BMDCs had higher mRNA expression of Il4 (encoding IL-4), Il5 (encoding IL-5), and Il13 (encoding IL-13), but comparable expression of Ifng (encoding IFNγ) and Il17a (encoding IL-17A) in lung tissues than those transferred with HDM-pulsed wildtype BMDCs (**Figure 3A**). Although a comparable percentage and cell number of CD4<sup>+</sup> T cells was observed in lung tissues of the two HDM-pulsed groups (**Figure 3B**), intracellular staining showed that recipients transferred with HDM-pulsed Fas-deficient BMDCs had higher percentage of IL-4+, IL-5+, and IL-13+CD4<sup>+</sup> T cells in lung tissues than those transferred with wild-type BMDCs, along with higher cell number of IL-4 <sup>+</sup>, IL-5+, and IL-13+CD4<sup>+</sup> T cells (**Figures 3C,D**), whereas the percentage and cell number of IL-17+CD4<sup>+</sup> T cells were similar in the two groups (**Supplementary Figures 2A,B**). However, the percentage and cell number of IL-4+, IL-5+, and IL-13+CD4<sup>+</sup> T cells had no difference between the recipients transferred with un-pulsed wild-type and Fas-deficient BMDCs (**Supplementary Figures 2C,D**). Taken together, these results demonstrate that HDM-pulsed Fas-deficient BMDCs promote Th2 responses in vivo.

# Fas Signaling in BMDCs Instructs Th2 Cell Differentiation in vitro

To determine whether the increased Th2 response in vivo is due to the direct interaction between DCs and T cells, we cocultured wild-type or Fas-deficient BMDCs with naïve OT-II CD4<sup>+</sup> T cells in the presence of OVA323−<sup>339</sup> peptide with or without HDM. The polarization of CD4<sup>+</sup> T cells was determined by qPCR or ELISA, respectively. OT-II CD4<sup>+</sup> T cells activated with HDM-stimulated Fas-deficient BMDCs had higher IL-4 and IL-13 expression both in mRNA and in protein levels than those stimulated with wild-type BMDCs (**Figures 4A,B**). GATA3, the master transcription factor for Th2 cell differentiation, was also found increased in T cells activated by HDM-pulsed Fas-deficient BMDCs, while the expression of Il9 and Il10 was comparable (**Figure 4A**). The activation and proliferation of OT-II CD4<sup>+</sup> T cells activated by HDM-pulsed both wild-type and Fas-deficient BMDCs were comparable (**Figures 4C–E**). Collectively, these data indicate that Fas signaling mediates the direct crosstalk between BMDCs and Th2 cells upon HDM stimulation in vitro.

# Fas-Deficient BMDCs Promote Th2 Differentiation by Inhibiting IL-12 Expression

Next we explored the molecular mechanism by which Fas signaling in BMDCs to shape Th2 differentiation upon HDM stimulation. We stimulated wild-type BMDCs with Fas agonistic antibody Jo2, the expression of Cd86, Tslp, Il10, Il6, and Tnfsf4, which had been reported to regulate Th2 polarization, had no difference compared to control group

(**Figure 5A** and **Supplementary Figure 3A**). HDM stimulation could dramatically increase the expression of Il6 and Tnfsf4, while Fas agonistic antibody Jo2 did not affect the expression of these two genes (**Figure 5A**), indicating that FAS signaling is not required for Il6 and Tnfsf4 expression in DCs upon HDM stimulation. IL-12 has been reported to affect Th2 differentiation (28). Fas agonistic antibody Jo2 stimulation could not affect the expression of Il12p35 and Il12p40 in wild-type BMDCs without HDM stimulation (**Figure 5A**). Upon HDM stimulation, the expression of Il12p35 and Il12p40 was significantly increased compared with non-HDM stimulation, and Fas agonistic antibody Jo2 could further enhance the expression of Il12p35 and Il12p40 in mRNA level and IL-12p70 in protein level (**Figures 5A,B**). Given that HDM-stimulated BMDCs had increased FasL, Fas antagonistic antibody kp7-6-treated HDMpulsed wild-type BMDCs produced lower Il12p40 in mRNA level and IL-12p70 in protein level than those of BMDCs stimulated with HDM alone, but the expression of Il12p35 had no difference between HDM and HDM plus kp7-6 stimulated BMDCs (**Figures 5C,D** and **Supplementary Figure 3B**). Fasdeficient BMDCs had decreased expression of Il12p35 and Il12p40 compared to wild-type BMDCs upon HDM stimulation in the presence of Fas agonistic antibody Jo2 (**Figure 5E**). We next examined whether the lower expression of IL-12 in Fas-deficient BMDCs could contribute to the increased Th2 differentiation. We added exogenous IL-12 into the BMDC–T cell co-culture system and found that the IL-12 supplement could significantly decrease the expression of Th2-related cytokines, such as IL-4 and IL-13 in OT-II CD4<sup>+</sup> T cells activated by Fas-deficient BMDCs. The expression of IFNγ and IL-17 was comparable between wild-type and Fas-deficient BMDC activated T cells (**Figure 5F**). Altogether, these results indicate that Fas-deficient BMDCs promote Th2 differentiation through downregulation of IL-12 expression.

#### Fas Signaling Regulates IL-12 Expression by Modulation of ERK Activation in BMDCs

We next examined the downstream signaling of Fas involved in the regulation of IL-12 expression by analyzing the activation of p38, JNK, and ERK in BMDCs. We found that the phosphorylation of ERK was decreased in wild-type BMDCs treated with Fas agonistic antibody Jo2 and HDM compared to those treated with HDM alone (**Figure 6A**). Accordingly, the activation of ERK in Fas-deficient BMDCs was increased compared to that of wild-type BMDCs upon HDM and Fas agonistic antibody Jo2 stimulation (**Figure 6B**). To determine whether the increased ERK activation was contributed to the decreased IL-12 expression in Fas-deficient BMDCs, we treated Fas-deficient BMDCs with specific ERK inhibitor U0126, which resulted in a dramatically increased IL-12p70 expression in Fasdeficient BMDCs (**Figure 6C**). Consequently, U0126-treated Fasdeficient BMDCs completely restored the increased IL-4 and IL-13 expression in T cells to the level of T cells activated with wild-type BMDCs (**Figure 6D**). These results indicate that Fas signaling regulates IL-12 expression and Th2 differentiation through modulating the ERK activity in BMDCs.

# DISCUSSION

Numerous studies have reported that administration of allergenpulsed DCs is sufficient to induce airway inflammation by polarizing Th2 responses (3, 29–31). However, the mechanism of DCs to regulate Th2 cell differentiation is still unclear (32). In this study, we used a BMDC-transfer protocol to investigate the important role of DC-specific Fas signaling in the pathogenesis of HDM-induced Th2-mediated allergic inflammation. We found that HDM-pulsed Fas-deficient BMDCs could promote Th2 responses and allergic eosinophilic inflammation, without affecting T cell apoptosis and proliferation in the recipients. Our study identified a crucial role of Fas signaling in regulating IL-12 expression by modulating ERK activity in DCs to direct Th2 differentiation upon HDM stimulation, which may provide an attractive treatment strategy for allergic diseases.

DCs constitutively express non-canonical costimulatory molecule Fas (13), which is activated by FasL or Fas agonistic antibody Jo2 to induce an apoptotic signaling. Numerous studies have shown that activated Fas signaling in DCs induces the secretion of cytokines such as IL-1β and CXC or CC chemokines (13, 33, 34), which may play important roles in the recruitment, activation and proliferation of naïve T cells (14, 35). Fas– FasL interaction on T cells has been proposed to promote the differentiation of naïve T cells into functional T cells (14), but little is known about how the Fas signaling in DCs regulating Th2 differentiation. DCs could promote Th2 differentiation through upregulation signal 2 (such as CD86, OX40L) (7, 36, 37) or signal 3 (such as IL-6, IL-10,TSLP) (38–40). In this study, we found that Fas agonistic antibody Jo2 stimulation did not affect the expression of OX40L and IL-6. However, Jo2 stimulation increased IL-12 expression in BMDCs stimulated with HDM. Accordingly, the ablation of Fas in DCs largely reduced the expression of IL-12, which contributed to the increased Th2 differentiation. Our results identify a new mechanism by which DC regulating Th2 responses through modulation of IL-12 production during inflammation development.

ERK signaling has been shown to play important roles in Fas-mediated non-apoptotic function. Ligation of Fas agonistic antibody Jo2 with Fas on DCs can promote the activation of

FIGURE 5 | Fas-deficient BMDCs promote Th2 differentiation by inhibiting IL-12 expression. (A) Wild-type BMDCs were stimulated with Fas agonistic antibody Jo2, HDM, and HDM with Jo2 for 5 h, un-stimulated BMDCs as control. Expression of Il6, Tnfsf4, Il12p35, and Il12p40, was detected by qPCR and normalized to Hprt. (B) Wild-type BMDCs were stimulated with HDM in the presence or absence of Fas agonistic antibody Jo2 for 8 h and Golgistop was added into the system in the last 4 h. The MFI of IL-12 was detected by intracellular staining. (C) Wild-type BMDCs were stimulated with HDM in the presence or absence of Fas antagonistic antibody kp7-6 for 5 h, un-stimulated BMDCs as control. Expression Il12p35 and Il12p40 was detected by qPCR normalized to Hprt. (D) Wild-type BMDCs were stimulated with HDM in the presence or absence of Fas antagonistic antibody kp7-6 for 10 h and Golgistop was added into the system in the last 4 h, un-stimulated BMDCs as control. The MFI of IL-12 was detected by intracellular staining. (E) Wild-type and Fas-deficient BMDCs were stimulated with HDM and Fas agonistic antibody Jo2 for 5 h. mRNA expression of Il12p35 and Il12p40 was detected by qPCR and normalized to Hprt. (F) IL-12 (1 ng/ml) was added into the BMDC–OT-II CD4<sup>+</sup> T cell co-culture system. mRNA expression of Il4, Il13, Il17, and Ifng in CD4<sup>+</sup> T cells was detected by qPCR and normalized to Hprt. \*P < 0.05, \*\*P < 0.01, ns, not significant. Data are representative of two independent experiments with duplicate or triplicate wells per group (A–F). Student's t-tests (E), one-way ANOVA (A–D) or two-way ANOVA (F) were performed and data were presented as mean ± SEM.

ERK and subsequent IL-1β secretion (34). In the current study, we found that Jo2 stimulation could dramatically decrease the activity of ERK in the presence of HDM. This different role might be caused by the type of stimuli and the status of the BMDCs, and further study need to be explored the potential mechanism for this different regulation by ERK. ERK signaling could profoundly influence the immune response of T cells. ERK activity in CD4<sup>+</sup> T cells has a key role in Th2 cell polarization (41). Blocking MEK-ERK signaling effectively suppresses IL-12p40 production from Neospora caninum infected peritoneal

macrophages (42). In contrast, ERK signaling also has been reported to be an important negative regulator of IL-12 secretion in cigarette smoke extract (CSE) stimulated DCs (43). In this study, we found that a higher ERK activity could suppress IL-12 production in HDM-stimulated Fas-deficient BMDCs. Taken together, this study uncovers a specific role of Fas signaling in BMDCs in the regulation of Th2 differentiation and Th2 mediated allergic inflammation. Modulation of Fas signaling on DCs may provide a new strategy for treatment of allergic diseases.

# AUTHOR CONTRIBUTIONS

MH designed and performed the in vivo and cellular experiments and contributed to manuscript writing. RH performed the in vitro and cellular experiments. JM contributed mouse models. CC contributed to animal colony management. BZ contributed to western blot analysis. HL and JY provided reagents. GH designed experiments, analyzed the data, wrote the manuscript, and provided overall directions.

# FUNDING

This work was supported by the National Natural Science Foundation of China (81471528, 91642104, 31670897 to GH, and 81725004 to HL, 81600788 to MH), the Ministry of Science and Technology of China (973 Basic Science Project 2014CB541803 to GH), the Program for Professor of Special Appointment (Eastern Scholar) at Shanghai Institutions of Higher Learning (to GH).

# ACKNOWLEDGMENTS

We thank Dr. Qibin Leng (Institut Pasteur of Shanghai, China) for kindly providing the J5 cell line.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2018.03045/full#supplementary-material

# REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2018 Han, Hu, Ma, Zhang, Chen, Li, Yang and Huang. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Endocytic Recycling of MHC Class I Molecules in Non-professional Antigen Presenting and Dendritic Cells

#### Sebastian Montealegre1,2,3 \* and Peter M. van Endert 1,2,3 \*

1 Institut National de la Santé et de la Recherche Médicale, Unité 1151, Paris, France, <sup>2</sup> Université Paris Descartes, Faculté de Médecine, Paris, France, <sup>3</sup> Centre National de la Recherche Scientifique, UMR8253, Paris, France

Major histocompatibility complex class I (MHC I) molecules are glycoproteins that display peptide epitopes at the cell surface of nucleated cells for recognition by CD8<sup>+</sup> T cells. Like other cell surface receptors, MHC class I molecules are continuously removed from the surface followed by intracellular degradation or recycling to the cell surface, in a process likely involving active quality control the mechanism of which remains unknown. The molecular players and pathways involved in internalization and recycling have previously been studied in model cell lines such as HeLa. However, dendritic cells (DCs), which rely on a specialized endocytic machinery that confers them the unique ability to "cross"-present antigens acquired by internalization, may use distinct MHC I recycling pathways and quality control mechanisms. By providing MHC I molecules cross-presenting antigens, these pathways may play an important role in one of the key functions of DCs, priming of T cell responses against pathogens and tumors. In this review, we will focus on endocytic recycling of MHC I molecules in various experimental conditions and cell types. We discuss the organization of the recycling pathway in model cell lines compared to DCs, highlighting the differences in the recycling rates and pathways of MHC I molecules between various cell types, and their putative functional consequences. Reviewing the literature, we find that conclusive evidence for significant recycling of MHC I molecules in primary DCs has yet to be demonstrated. We conclude that endocytic trafficking of MHC class I in DCs remains poorly understood and should be further studied because of its likely role in antigen cross-presentation.

Keywords: major histocompatibility, endosome, dendritic cell, recycling, antigen presentation, cross-presentation, Arf6, Rab11

#### INTRODUCTION

MHC I molecules present pathogen, tumor and self-antigens to CD8+ T cells through the endogenous or direct and the exogenous or cross-presentation (1, 2) pathways. The spatio-temporal separation of these pathways (3) implies that intracellular transport of MHC-I molecules must be regulated, and that MHC I trafficking may vary according to cell type and particularly to the presence or absence of cross-presentation capacity. Mechanisms regulating trafficking likely are intertwined with mechanisms of quality control and act at various places in the cell: the endoplasmic reticulum (ER), the Golgi apparatus, the cell

#### Edited by:

Christian Muenz, University of Zurich, Switzerland

#### Reviewed by:

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France Sven Burgdorf, Universität Bonn, Germany

\*Correspondence:

Sebastian Montealegre sebastian.montealegre@inserm.fr Peter M. van Endert peter.van-endert@inserm.fr

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 06 September 2018 Accepted: 13 December 2018 Published: 07 January 2019

#### Citation:

Montealegre S and van Endert PM (2019) Endocytic Recycling of MHC Class I Molecules in Non-professional Antigen Presenting and Dendritic Cells. Front. Immunol. 9:3098. doi: 10.3389/fimmu.2018.03098

**250**

surface, and the endosomal system. A vast amount of information is available about the quality control steps selecting properly folded class I molecules in the secretory pathway (4–7). In contrast, the endocytic transport of class I molecules and the mechanisms of quality control in it are much less understood.

Analysis of endocytic trafficking is complicated by the fact that MHC I molecules exist in various forms presumably sensed by mechanisms of quality control, and that may follow distinct intracellular trafficking pathways: trimers made of a heavy chain, beta-2 microglobulin (β2m), and high affinity peptides; trimers made of heavy chain, β2m, and low affinity peptides; dimers without any bound peptides; and free heavy chains (FHC). For simplicity, we will refer to the trimers with high-affinity peptide as "fully conformed," and the other forms as "suboptimally loaded," unless otherwise specified. Distinguishing complexes with high and low affinity peptides is important, since the affinity of the peptide-MHC interaction is the first determinant of the lifetime of class I molecules at the cell surface (8–10). It also determines the dissociation of β2m, the binding of which acts as signal preventing degradation of class I complexes (9). The picture becomes even more complex considering that different class I allotypes have different half-lives at the cell surface (11). Thus, putative quality control mechanisms should sense correctly structural variants for more than 5,500 class I allotypes (12) to discriminate between degradation and recycling.

As we will describe in detail, a large amount of information about the recycling pathways followed by class I molecules has been obtained in HeLa cells and H-2L<sup>d</sup> -expressing L cell fibroblasts. Available data suggest quantitative and mechanistic differences relative to the speed and efficacy of class I recycling between model cell lines. In contrast, very limited information is available on MHC I trafficking in professional antigen presenting cells (pAPCs). It is ironic that recycling has mainly been studied in cell lines unable to cross-present, given that the likely biological role of recycling concerns crosspresentation in DCs priming T cell responses to tumors and pathogens.

While some discrepancies between published studies may be derived from methodological approaches, many will be due to variation between cell types studied. In this review we will not only emphasize differences between model cell lines and pAPCs, but also examine the methods that have been used to obtain quantitative data on recycling efficacy and kinetics. Moreover, we will highlight knowledge on trafficking of fully conformed and sub-optimally loaded class I molecules obtained studying non-immune, non-phagocytic cell lines. Finally, we will relate these observations to existing and lacking knowledge on MHC I trafficking in pAPCs and its role in antigen cross-presentation. As we discuss key data on endocytic trafficking and evidence for differential sorting of distinct MHC I conformers, it needs to be kept in mind that the molecular players and chaperones mediating quality control in this context remain unknown. We anticipate that identification of such players will be required to fully understand endocytic trafficking and recycling of class I molecules, and conclusively answer the questions discussed below.

# ENDOCYTOSIS IN A NUTSHELL

Proteins that are destined to be recycled to the cell surface need, by definition, first to be internalized from the cell surface. Internalization of cell surface components is a constitutive event in all cell types and important in nutrient uptake, signal transduction, cell adhesion, and in renewal and recycling of plasma membrane components, among others (2, 13). Internalization of membrane proteins requires the formation of endocytic vesicles delivering cargo to the cell. The formation of such vesicles can be mediated by clathrin, a protein forming a lattice around the newly generated vesicle in the form of triskelions (14), or can be clathrin-independent. Clathrin-mediated endocytosis (CME) and clathrin-independent endocytosis (CIE) are commonly distinguished as the two main routes of endocytosis (15, 16). Proteins that are internalized via CME typically harbor the motif Y-X-X-8 at their cytosolic tail, where 8 is any bulky hydrophobic amino acid and X is any amino acid. MHC I molecules do not possess such a motif, although a sequence in the cytoplasmic tail of HLA-A and B molecules has been postulated to represent a noncanonical motif for CME endocytosis (17). CIE pathways are named according to the morphology of the vesicle coat or cargo (e.g., caveolae, or lipid raft endocytosis), or to key intracellular proteins regulating trafficking (18). One of the latter CIE pathways is named for the small GTPase Arf6 and has been widely described as the mechanism of endocytosis of MHC I molecules in model cell lines. In this pathway, hydrolysis of Arf6-GTP is required twice, first after internalization of cargo (19) to change the phosphoinositide composition of the early endosomes and allow fusion with EEA-1 vesicles (13, 19), and a second time to promote recycling from tubular recycling endosomes (described below) (19, 20). Regardless of the mode of internalization, after 5 to 10 min, internalized proteins arrive to a shared station, the early sorting endosomes (21, 22). Typical markers of the sorting endosomes, with a luminal pH in the range of 6.3 to 6.8, are the GTPase Rab5 and its effector EEA-1 (23). From these vesicles, internalized cargo can be sorted to late endosomes and lysosomes for degradation or be re-directed to the cell surface using various routes of recycling. At least three recycling pathways have been described for model cell surface receptors and in model cell lines: direct recycling from the sorting endosomes to the cell surface ("fast recycling"), the route followed by the transferrin receptor; transport from sorting endosomes to the Trans Golgi Network (TGN) and then to the cell surface ("retrograde transport") (24); and transport from the sorting endosomes to the endocytic recycling compartment (ERC) and then to the cell surface ("slow recycling"). Fast recycling is in the order of 5–10 min (25), whereas slow recycling is in the order of 30–60 min (2). For an extensive review of the mechanistic regulation of each pathway we refer the reader to two excellent recent reviews (22, 26). For MHC I molecules, a fourth pathway from late endosomes to the surface (discussed below) has also been suggested. We will focus mainly on the pathway involving the ERC, the main route taken by recycling MHC I molecules in non-professional APCs according to literature data.

Recycling endosomes are dynamic, tubulo-vesicular structures of nearly neutral pH in charge of sorting and reexporting internalized membrane material (2, 26). The most widely used though by no means exclusive marker of recycling endosomes is the small GTPase Rab11. Rab11 localizes mostly to a perinuclear region that defines the ERC. Rab11 also localizes to the TGN (27) and to vesicular structures. Rab22 is another small GTPase involved in endosomal membrane trafficking (28) found both in the ERC and in tubular recycling endosomes (TRE) (29). Activation of Rab22 is required for the formation of TREs from the ERC, and inactivation is required for the final fusion of the tubules with the surface (29). Apart from Rab22, the biogenesis and maintenance of the TRE is in part regulated by the actin regulatory redox enzyme MICAL-L1(30), the family of Eps15 homology domain-containing 1-4 proteins (EHD-1 to 4) (31), and the small GTPases Arf6 (see above), Rab35, and Rab8a (32, 33). MICAL-L1 is a central protein required for the de novo generation of TREs (32) since it can bind directly phospholipids, and can form tubules in vitro and in vivo (34). MICAL-L1 serves as a hub for multiple proteins to regulate the formation of the TREs: it binds to EHD-1 via Rab8, as well as to Arf6 or Rab35 (33). Arf6 positively regulates recycling by aiding to localize Rab8 to the forming TREs, as well as by activating phospholipase D and PIP5 kinase, thereby providing the necessary lipids for the generation of recycling vesicles (35). Rab35, on the other hand, works as a negative regulator of TRE formation, by binding to MICAL-L1 and promoting GTP hydrolysis of Arf6 by the GTPase activating protein (GAP) ACAP2 (35, 36).

#### RECYCLING OF FULLY CONFORMED MHC CLASS I MOLECULES: METHODS AND EVIDENCE

Fully conformed and sub-optimally loaded MHC I molecules can be distinguished by monoclonal antibodies. We will first discuss work performed using antibodies recognizing the former category of class I molecules, which represents the vast majority of published studies (**Figure 1**). Early work from the groups of Watts and Jondal showed that upon internalization, class I molecules recycle to the cell surface (37, 38). Making use of a surface iodination assay in B lymphoblastoid cells, Reid and Watts were able to show that after accumulation of peptidebound class I molecules in intracellular compartments by incubation with the inhibitor primaquine for 30 min, removal of primaquine resulted in recycling of nearly all the internalized class I to the cell surface within 16 min. Using the TAPdeficient thymoma cell line RMA-S, Abdel-Mottal et al. loaded class I molecules with glycopeptides and antibodies against the glycopeptide, then allowed for internalization, removed the remaining cell-surface complexes by a protease, and finally found that, depending on the peptide sequence, 36 to 63% of the class I molecules bound to the glycopeptides recycled to the surface. The re-appearance of the class I molecules was sensitive to chloroquine and leupeptin, indicating trafficking of the complexes via endosomal compartments. Although these experiments were performed using the thymoma line RMA-S,

the findings of Abdel-Mottal et al. were a first indication that internalized peptides, and by extrapolation possibly antigens, might be loaded on recycling MHC I molecules in a "vacuolar" pathway.

Other work corroborated the conclusions of these early studies that class I molecules can recycle to the cell surface in model cell lines. The seminal work by Radakhrishna and Donaldson (20) showed for the first time the involvement of the small GTPase Arf6 in the recycling of MHC I molecules. As in most of the pertinent literature, the experimental system was HeLa cells, and the antibody used to detect class I was W6/32, which detects HLA heavy chains bound to β2m. In this system, class I localized to tubulo-vesicular structures decorated with Arf6. Moreover, in HeLa and Jurkat cells where the constitutively active mutant Arf6 Q67L was overexpressed, internalized class I molecules accumulated in PIP<sup>2</sup> rich endosomes, thereby preventing their further degradation or recycling (13, 39, 40). Furthermore, overexpression of an effector domain mutant of Arf6 (N48I) in HeLa cells decreased recycling of class I molecules by 60% relative to control cells (35).

Subsequent studies of class I recycling in HeLa cells showed that 10–15 min after internalization, class I molecules reached tubular structures, presumably the TRE, the formation of which required EHD-1 among other players (see above) (29, 41). The quantitative assay to examine the role of EHD-1 in the recycling of class I was a "CELISA": the authors seeded HeLa cells onto ELISA plates, incubated them with biotinylated MHC I antibodies at 37◦C for 5 min, washed out free antibodies, and incubated the cells at 37◦C for various time points. When EHD-1 was overexpressed, the number of cell surface class I molecules bound by biotinylated antibodies increased by 50% relative to the cell surface population present at the end of internalization (41).

The description of the role of Rab22 in endocytic recycling introduced the most widely used recycling assay so far (29). In a seminal study, HeLa cells were pulsed with mAb W6/32, allowed to internalize for 30 min at 37◦C, acid-stripped to remove remaining cell surface complexes, and then chased at 37◦C for various lengths of time. Quantification was based on the reappearance of class I at the cell surface in unpermeabilized cells, as detected with a secondary antibody, compared to the internal class I signal, which was detected by removing the recycled class I with a second acid treatment, permeabilization of the cells, and detection with a secondary antibody. The read-out used microscopy or flow cytometry. In untransfected cells, 30% of the internalized class I population recycled to the surface by 30 min (29). Overexpression of wt Rab22 reduced recycling by 50%, an inhibitory effect that became even more pronounced upon overexpression of the dominant negative Rab22 S19N and the constitutively active Rab22 Q64L mutants. Using the same experimental system, the authors found that overexpression of the dominant negative mutant Rab11 S25N reduced class I recycling by nearly 80% relative to control cells, confirming a role for Rab11 in class I recycling in HeLa cells.

More recently, using the same recycling assay and HeLa cells, it was shown that the enzyme diacylglycerol kinase alpha (DGKα) was required for formation of tubular recycling endosomes by interacting with MICAL-L1 and generating phosphatidic acid

(42). In turn, knockdown of DGKα delayed recycling of class I. Remarkably, after 30 min of internalization and 3 h of chase (recycling), the authors detected up to 40% recycled class I in wt Hela cells.

A putative alternative class I recycling pathway is mediated by Rab35, experimentally demonstrated in Cos-7 cells. Knockdown of Rab35 resulted in formation of enlarged EHD-1 negative endosomes (43). The authors proposed that Rab35 mediates "fast" direct recycling of class I from early endosomes to the cell surface, in a pathway distinct from the Rab22-Rab11 recycling axis. However, this conclusion was based on an assay that does not provide unequivocal evidence for recycling. The assay consisted in two 20-min incubations each followed by acid stripping to remove cell surface class I molecules. The first stripping removed MHC I not internalized after the initial 20-min pulse with an anti-MHC I antibody, and the second molecules "recycled" to surface after the second 20-min period. An increase in the number of labeled MHC-I molecules, detected by staining of permeabilized cells with a secondary antibody, was then interpreted as intracellular retention and lack of recycling. However, a role of Rab35 in degradation of internalized MHC-I molecules would equally well explain the increase by 60% of "retained" class I observed upon Rab35 knockdown.

Reviewing the different assays, some discrepancies are apparent, which may be biological or methodological. While the role of the different endocytic regulators is undisputed in the cell lines evaluated, the reported recycling kinetics of fully conformed class I molecules vary significantly from assay to assay and from cell line to cell line. The biochemical surface labeling assay (37) led to the conclusion that almost all fully conformed class I molecules that are internalized recycle very efficiently and fast. In contrast, the microscopy and FACSbased assays suggest that recycling of fully conformed class I recycling is slow and inefficient (**Table 1**). Perhaps the simplest explanation is that fully conformed class I can recycle via two different pathways, a fast one and a slow one. However, among the published pathways reviewed here, the "fastest" recycling pathway described implicates Rab35. In this pathway, recycling was detected after 20 min, which is already in the range of the slower recycling pathway dependent on the Arf6-Rab22-Rab11- MICAL-L1 axis. Thus, there is presently no conclusive evidence for class I recycling through a truly fast pathway returning for example the transferrin receptor to the surface. Variations between cell lines may also play a role. For example, in CHO cells, the Arf6 pathway plays a role in the recycling of the transferrin receptor, which is normally endocytosed via clathrin-mediated

#### TABLE 1 | Published data on MHC-I recycling.


The superscript letter indicates the allele of the gene indicated by the capital letter preceding it - no explanation required for any immunologist.

endocytosis (15), indicating that different recycling pathways might operate in different cells. Also, Rab22 is necessary for internalization of class I in Jurkat but not in HeLa cells (29, 40). As mentioned above, published data show remarkable variation with respect to the kinetics of class I recycling. Thus, in HeLa cells, the model system most frequently, recycling of W6/32 positive molecules ranges from 30 min to 3 h. At the same time, the authors of various publications agree on an estimated rate of 30–40% recycling class I molecules using the microscopybased assay. How can the same rate of recycling be obtained in such divergent time spans? One possible explanation would be a mechanism of quality control that keeps the number of recycling class I molecules on the cell surface at any time below a certain threshold. For example, fully conformed class I molecules reaching the cell surface using the Arf6-Rab22-Rab11-MICAL-L1 axis within 30 min could rapidly be internalized again. It is unknown how many rounds of recycling a single class I molecule can undergo. It is also unclear whether and to what extent a peptide exchange occurs during class I recycling in non-professional APCs.

Differences between class I allotypes might also lead to diversion into a different recycling pathway. In the secretory pathway, where the molecular players mediating quality control such as tapasin and calreticulin are well characterized, class I polymorphism is well known to affect quality control. For example, class I allotypes differing by a single amino acid, such as HLA-B<sup>∗</sup> 44:02 and B<sup>∗</sup> 44:05 or HLA-B<sup>∗</sup> 27:05 and B<sup>∗</sup> 27:09, differ greatly with respect to dependence on tapasin in order to acquire high affinity peptide ligands and leave the ER (48, 49). Similar differences might be revealed in endocytic quality control once the relevant chaperones will be identified. One candidate for such a role is the tapasin homolog TAPBPR, which can operate in a pH range of 6.0–7.2 and may therefore be able to mediate peptide exchange in endosomes. Interestingly, very recent data suggest that TAPBPR also acts in an MHC I allele-specific manner (50). In conclusion, the recycling pathways of fully conformed class I molecules in non-professional APCs still require clarification and additional investigation.

#### RECYCLING OF SUB-OPTIMALLY LOADED MHC CLASS I MOLECULES

Heavy chain-β2m empty dimers and FHC constitute a minor population out of the total pool of class I molecules present at the cell surface and can be identified by a number of monoclonal antibodies. The precise relative proportion of FHC and empty dimers among surface class I molecules are not known but likely vary according to the cell type and state as well as to the class I allotype considered. For example, among the two murine allotypes studied most frequently, H-2K<sup>b</sup> and H-2L<sup>d</sup> , cell surface L<sup>d</sup> is known to comprise a larger proportion of FHC due to its lower stability relative to K<sup>b</sup> . FHC have been found at the cell surface of T and B lymphocytes under inflammatory conditions (51, 52) and in β2m deficient cells at resting conditions (53). While the function of the FHC is starting to be elucidated and may reside mainly in their binding to particular KIR receptors on NK cells (54–58), there is limited mechanistic evidence about their endosomal regulation and recycling.

Recent results provide initial insight into how sub-optimally loaded dimers are internalized, recycled, and degraded. The best evidence comes from the work of Lucin and co-workers, who have systematically evaluated the constitutive internalization of the murine class I allotype H-2L<sup>d</sup> in their fully conformed and sub-optimally loaded forms (44, 59), as well as the early and late endosomal recycling of H-2L<sup>d</sup> molecules (45, 46). This allotype is particularly suited for studying the different class I conformers because of the availability of the monoclonal antibodies 30.5.7 and 64.3.7, originally produced by Hansen et al. with a wellstudied and exquisite specificity for the two conformers (60–62). As mentioned above, H-2L<sup>d</sup> is less stable than other allotypes, resulting in a higher proportion of sub-optimally loaded or empty molecules, adding another argument in support of studying trafficking of different class I conformers using L<sup>d</sup> as a model. The experimental system established by this group consists mainly, but not only, in L cell fibroblasts expressing L<sup>d</sup> , which constitutively express high levels of sub-optimally loaded dimers; monoclonal antibodies 30.5.7 and 64.3.7 distinguishing fully conformed class I trimers and β2m-bound heavy chains devoid of peptides, respectively; and a quantitative flow cytometry recycling assay based on the principle of the assay by Weigert et al. (29). Using these tools, they found that fully conformed internalized H-2L<sup>d</sup> molecules that had accumulated during 1 h in a Rab11+compartment, presumably the ERC (45, 46), recycled with an efficiency of 20–30 %, reaching a plateau at 30 min. In contrast, sub-optimally loaded H-2L<sup>d</sup> molecules were not detected in the Rab11<sup>+</sup> ERC and could not recycle (44). They extended these results to other class I allotypes, such as HLA-Cw6 and HLA-B7, and cell lines, obtaining similar recycling rates (45). Whether recycling of fully conformed H-2L<sup>d</sup> molecules was Arf6 dependent or not was not investigated. However, considering the recycling rates matching those observed in some of the HeLa-W6/32 recycling assays, and the involvement of the Rab11<sup>+</sup> ERC, it is likely that they followed the Arf6 pathway. Since recycling of sub-optimally loaded class I molecules was not detected using the conventional methods, they modified the assay by internalizing sub-optimally loaded molecules bound to 64.3.7 for 3 h instead of 1 h, and comparing the signal for internalized vs. total labeled class I by flow cytometry (46). Surprisingly, they were able to detect recycling sub-optimally loaded molecules that had passed through Rab7<sup>+</sup> late endosomes, with an efficiency of 15–20% after 30 min of chase.

So far, this is the sole quantitative evidence for recycling of sub-optimally loaded class I molecules, which may implicate a special pathway originating from late endosomes. In TAPdeficient fibroblasts pre-incubated at 26◦C bearing relatively large numbers of sub-optimally loaded class I molecules at the cell surface (63, 64), FHCs, but not dimers, are rapidly removed from the cell surface (9). It is conceivable that a late endosomal pathway provides recycling dimers at low temperature, which at physiological temperature are degraded upon dissociation of β2m (9). Importantly, the cellular compartment where class I molecules are sorted for a round of recycling using the ERCdependent recycling pathway or the late endosomal recycling pathway is not known. Whether these processes depend on Arf6, or whether they can be extrapolated to other cell lines or cell types, are other open questions.

# MHC CLASS I RECYCLING AND ANTIGEN PRESENTATION

While MHC I recycling has been subjected to some cell-biological scrutiny, there is surprisingly little published evidence on its role in antigen presentation. Given that the endocytic pathway plays a mandatory role in cross-presentation of internalized antigens, it may not surprise that the available literature concerns exclusively this pathway. Indeed, a role of recycling class I molecules in peptide loading of class I molecules in the ER appears little likely. However, class I molecules can also be loaded with endogenous peptides in post-ER compartments potentially accessible for recycling class I molecules. For example, Hsc70-coupled endogenous antigens can be processed through chaperone mediated autophagy (65), and HSV-1 antigens through a non-canonical pathway of macro-autophagy (66), both presumably implicating antigen degradation in endolysosomal compartments. Moreover, peptide fragments of endogenous transmembrane proteins can be produced and loaded in the same type of compartment (67). However, whether recycling MHC I molecules were responsible for presentation in these studies is not known.

As discussed above, there appears to exist significant variation regarding the pathways and type of class I molecules able to recycle in non-immune cells. The scenario in pAPCs, e.g., cells derived through differentiation from monocytes or from bone marrow precursors in vitro, or primary DCs obtained from mice or humans is even less understood. In DCs but probably also some macrophage types capable of crosspresentation, class I molecules need to have access to peptides from internalized antigens. This can occur in the perinuclear ER through a cytosolic pathway of cross-presentation, or in non-ER compartments, following either the ER-phagosome pathway or vacuolar pathways (2). In the two latter scenarios, crosspresenting pAPCs may require an alternative source of class I molecules independent of the secretory pathway. It is in this context that recycling class I molecules emerge as a candidate source of MHC I molecules provided to the ER-phagosome and the vacuolar pathway of cross-presentation (**Figure 2**).

One of the first indications of potential pathways operating in professional antigen presenting cells [for the subtypes of DCs, the reader is referred to other reviews (68, 69)], was curiously obtained in a melanoma cell line. Grommé et al. fractionated MelJuSo melanoma cells and identified HLA class I molecules in a fraction also containing acidic HLA class II loading compartments. A significant proportion of HLA class I-peptide complexes remained stable and could be immunoprecipitated at pH=5 using W6/32. This implied the existence of a compartment

potentially acidic enough to promote peptide exchange, but not enough to degrade class I molecules (70). More direct evidence for peptide exchange in presumably recycling MHC I molecules was provided by studies on TAP-deficient macrophages. Preincubation with peptide ligands stabilized a pool of class I molecules on these cells which then allowed for crosspresentation of a bacterial antigen through a vacuolar pathway, most likely involving peptide exchange in an acidic compartment (71). Coming back to a more physiologic setting, various groups have reported the existence of a post-ER compartment containing presumably fully conformed class I molecules in different types of (TAP-sufficient) DCs (72–74). A common characteristic of the latter studies is that upon incubation with primaquine, crosspresentation of soluble antigens is blocked (3, 75, 76). A possible interpretation is that class I molecules are located in a mildly acid endosomal compartment from which they are unable to recycle upon inhibition of acidification by primaquine.

Do class I molecules recycle in pAPCs? And if so, do they use the pathways described in non-immune cell lines? The recent studies from Nair-Gupta et al. and Cebrian et al. built on the knowledge gained from the studies of non-immune cell lines to identify players in the cross-presentation pathways. The former study characterized a post-ER Rab11+/VAMP8<sup>+</sup> compartment containing MHC I molecules in bone marrow-derived DCs (BM-DCs) and suggested that this compartment constituted an important source of cross-presenting MHC I molecules. Upon stimulation by TLR-2/4 ligands, class I molecules derived from the Rab11 compartment were recruited to phagosomes. Knockdown of Rab11 in BM-DCs had a profound detrimental effect on antigen cross-presentation. Remarkably, the authors showed that primary type 1 conventional DCs (cDC1), but not cDC2 harbor the class I/Rab11<sup>+</sup> compartment, suggesting that cDC2s may have a different source of class I molecules for cross-presentation (77). It is puzzling that pDCs, which also have the post-ER compartment (73) have shown to be less efficient than their cDC1 and cDC2 counterparts in antigen cross-presentation.

Consistent with observations made in HeLa cells, the study by Cebrian et al. (47) showed that Rab22 partially co-localizes with Rab11 and with fully conformed class I molecules in BM-DCs and in JAWS-II cells, an immortalized cell line derived from C57BL/6 BM-DCs lacking p53. A 50% knockdown of Rab22 was sufficient to abolish the post-ER compartment containing class I molecules, as well as to compromise antigen cross-presentation. Rab22 knockdown not only hampered cross-presentation of soluble OVA but also of OVA secreted by Toxoplasma gondii into the parasitophorous vacuole of parasite-infected DCs. This result corroborated the importance of a Rab11+ compartment containing MHC I molecules in cross-presentation, however it remained unclear whether the MHC I molecules in this compartment actually derived from the cell surface. To study recycling, Cebrian and associates used an assay similar to that used by Allaire to study the role of Rab35 (43), in which an increase in MHC I molecules "retained" intracellularly was interpreted as evidence of lack recycling. As noted above, an intracellular accumulation of internalized MHC I molecules can be due both to compromised recycling and to reduced degradation. Thus, the finding that Rab22 knockdown inhibited "disappearance" of internalized MHC I over a 40-min period almost completely may indicate a role in MHC I recycling and/or in routing to a degradative compartment. Thus, in our view, MHC I recycling in murine DCs remains to be demonstrated conclusively.

Our review of the literature, as well as our own unpublished observations, suggests that monitoring MHC I recycling in primary DC-like populations, be they differentiated in vitro from bone marrow or monocytes, or isolated as primary in vivo DC populations, remains challenging. Considering the specialization but also plasticity of DCs, as well as the variation already existing between non-immune cell lines, it would not be surprising to find distinct recycling rates and pathways in primary DC populations. The role of recycling MHC I molecules might also be limited to specific antigen types. As an example, in a study including an assay directly measuring internalized MHC I molecules reappearing at the cell surface, van den Eynde et al. recently showed that cross-presentation by human monocyte-derived DCs (mo-DCs) of a synthetic long peptide (an antigen type of interest for tumor vaccination) involved peptide exchange on MHC I molecules, however these molecules were nascent rather than recycling (78).

# OUTLOOK

Although the cited papers provide solid evidence for an important role of Rab11 and Rab22 in cross-presentation, important mechanistic issues remain unresolved. Identification of the Rab11/Rab22/VAMP8 compartment containing fully conformed class I molecules may suggest that the class I molecules detected originate from the cell surface, where they return to through recycling. However, as discussed above, there is no formal proof that this is the case. Several alternative scenarios could be considered. The TGN might feed this compartment with fully conformed class I molecules arriving through the secretory pathway, using, for example, Syntaxin 6<sup>+</sup> and Rab14<sup>+</sup> TGNderived vesicles (79, 80). Another hypothesis is the adaptation of the late endosomal recycling pathway described by Lucin and

coworkers, where lysosomes would communicate with recycling endosomes that are in physical proximity, thereby providing an environment where class I molecules are sorted for peptide exchange and routing into the recycling pathway, or degradation.

Another issue that has not been well studied in pAPCs is the formation of tubular recycling endosomes. In the work by Cebrian et al. (47), Di Puchio et al. (73), Zou et al. (74), Nair-Gupta et al. (77), Croce et al. (81), such tubules were not described. The endocytic system of DCs remodels upon stimulation by TLR ligands, such that resting DCs do not behave like an LPS stimulated DC (82, 83). In this context, it appears that the formation of elongated tubules originating from the ERC in moDCs requires TLR stimulation and the formation of the immunological synapse (84, 85). The formation of the tubules requires the presence of MICAL-L1 (86), as in HeLa cells. However, as opposed to HeLa cells, class I molecules have not been observed in such tubules in pAPCs. The observations reported so far have made use of specimens treated with fixation, which renders visualization of such tubules difficult (26). Live cell-imaging methods might reveal the presence of such tubules in resting dendritic cells. Whether Arf6, Rab8, Rab35, ACAP1, or other proteins are required for the formation of such tubules remains to be investigated.

It is important to highlight that BM-DCs and moDCs, which have served as a model to study cross-presentation (but not class I recycling so far) in vitro, might not reflect the pathways of class I recycling and cross-presentation operating in vivo (87, 88). Unfortunately, the typical cell biological and crosspresentation assays require a substantial number of cells that can be difficult to obtain for primary pAPC subsets. Since BM-DCs and moDCs reflect most closely the properties of inflammatory dendritic cells and macrophages found in vivo (89), and the latter ones use in some cases a purely vacuolar crosspresentation pathway (90), it is important to study recycling of class I molecules in primary cells. We anticipate that there will be significant variation and plasticity in the recycling pathways and rates of class I molecules in primary pAPC populations. Identifying the molecular machinery in charge of endocytic quality control will be essential in order to fully decipher MHC I endocytic trafficking, recycling and cross-presentation. Finally, the functional impact of MHC I recycling is almost entirely unexplored. Previous studies have been limited to examining presentation of the model antigen OVA to a CD8+ T cell line. The ultimate challenge for the field will be explore the impact of MHC I recycling in T cell responses to pathogens and tumors.

# AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

# ACKNOWLEDGMENTS

Work in the authors' laboratory was supported by grant 14-CE11- 0014 from the Agence Nationale de Recherche. We are grateful to Maria Camila Montealegre for assistance with graphic design.

# REFERENCES


MHC class I-restricted presentation. Immunol Rev. (1999) 172:131–52. doi: 10.1111/j.1600-065X.1999.tb01362.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Montealegre and van Endert. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Understanding the Functional Properties of Neonatal Dendritic Cells: A Doorway to Enhance Vaccine Effectiveness?

Nikos E. Papaioannou<sup>1</sup> , Maria Pasztoi <sup>1</sup> and Barbara U. Schraml 1,2 \*

<sup>1</sup> Biomedical Center, Institute for Cardiovascular Physiology and Pathophysiology, Ludwig-Maximilians-University Munich, Munich, Germany, <sup>2</sup> Walter-Brendel-Centre of Experimental Medicine, University Hospital, Ludwig-Maximilians-University Munich, Munich, Germany

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Joke M. M. Den Haan, VU University Medical Center, Netherlands Daniela Verthelyi, State Food and Drug Administration, China

#### \*Correspondence:

Barbara U. Schraml barbara.schraml@ med.uni-muenchen.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 29 August 2018 Accepted: 18 December 2018 Published: 10 January 2019

#### Citation:

Papaioannou NE, Pasztoi M and Schraml BU (2019) Understanding the Functional Properties of Neonatal Dendritic Cells: A Doorway to Enhance Vaccine Effectiveness? Front. Immunol. 9:3123. doi: 10.3389/fimmu.2018.03123 Increased susceptibility to infectious diseases is a hallmark of the neonatal period of life that is generally attributed to a relative immaturity of the immune system. Dendritic cells (DCs) are innate immune sentinels with vital roles in the initiation and orchestration of immune responses, thus, constituting a promising target for promoting neonatal immunity. However, as is the case for other immune cells, neonatal DCs have been suggested to be functionally immature compared to their adult counterparts. Here we review some of the unique aspects of neonatal DCs that shape immune responses in early life and speculate whether the functional properties of neonatal DCs could be exploited or manipulated to promote more effective vaccination in early life.

Keywords: dendritic cell (DC), DC subsets, early life immunity, vaccination, immune system development, T cell activation, innate immunity, DC targeting

# INTRODUCTION

Early life immune balance is essential for survival and establishment of healthy immunity in later life. The neonatal period in mammals represents a critical window, in which the immune system has to keep a fine balance between efficient pathogen defense and maintenance of tolerance against a continuous flood of commensal microbes and environmental antigens (1–4). Reduced inflammatory capacity is an inherent feature of early life immunity that has been attributed to an altered repertoire of immune cells, as well as a relative functional immaturity of immune cells in early compared to later life (2–5). For example, the T and B cell pools are not fully expanded at birth and are biased to generate T helper (Th) 2 type responses compared to a more Th1 type response in adults (1). Neonatal, but not adult, monocytes and neutrophils potently suppress T cell activation in vitro and therefore strongly resemble myeloid derived suppressor cells (6). Although neutrophil-like myeloid derived suppressor cells show microbicidal activity (6), the inflammationinduced trafficking of neutrophils, as well as their ability to form extracellular traps, are reduced in fetal and early life compared to adult (7–9). Fetal monocytes are transcriptionally distinct from their adult counterparts and fetal, as well as, neonatal monocytes show distinct responsiveness to inflammatory stimuli than adult monocytes (10–14). Macrophages first develop before birth and are thought to aid in tissue remodeling during development, whereas they acquire their full-blown immune functions with increasing age (11, 14). Microglia of the brain for instance gain an immune-related gene signature over time and in response to microbial signals (15). Additionally,

**261**

the existence of specific immune regulatory cells of erythroid origin in early life has been suggested to dampen inflammatory responses (4, 16).

As a result of these immune alterations, neonates exhibit an increased susceptibility to infections (2, 5). In humans, pathogens that are often asymptomatic in adults, such as Haemophilus influenzae type B, Bordetella pertussis and Streptococcus pneumoniae, account for the death of more than two million infants per year world-wide (17). Dendritic cells (DCs) have been implicated to promote immune responses to these pathogens in adults (18–21). It is possible to immunize infants under one year against these pathogens, but a single immunization does not necessarily provide immediate protection, or as in the case of H. influenzae type B, antibody titers may not persist (17). For other pathogens, such as rotavirus, immunization is first possible few weeks after birth leaving infants at risk of infection, when the disease is most severe (17, 22). The efficacy, success and challenges of vaccines in early life, as well as existing efforts to improve their effectiveness have recently been reviewed elsewhere (17). Here we focus on the functional differences between DCs in early and adult life. DCs sense the presence of pathogens or damage via so called pattern recognition receptors (PRRs) and initiate innate, as well as adaptive immune responses through cytokine production and antigen presentation (23–25). In their function as immune sentinels DCs have been extensively targeted to increase vaccine effectiveness and DC based vaccines hold promise in adults (26). However, as other immune cells, DCs in early life differ from their adult counterparts in phenotype and function, raising the question, whether targeting DCs could be used to elicit protective immunity in infants and increase vaccine effectiveness. Although most of the data discussed here derive from mouse studies, parallels likely exist in humans, as DC subsets and functions appear highly conserved across species (23, 25).

#### DENDRITIC CELLS DEVELOP AS FUNCTIONALLY DISTINCT SUBSETS

Among DCs, we distinguish two main functionally and developmentally distinct cell lineages. Conventional or classical DCs (cDCs) are remarkable activators of adaptive immune responses with a remarkable capacity to capture, process, and present antigens to T cells (23–25, 27). Plasmacytoid DCs (pDCs) on the other hand are critical for defense against viruses, because of their capacity to respond to viral antigens and secrete type I interferons (IFN) (28, 29). Most of our knowledge about the development of these cells is based on studies in adult mice. In adults, DCs have a short lifespan and rely on constant replenishment from bone marrow-derived hematopoietic stem cells (30, 31). cDCs and pDCs were long thought to derive from a common myeloid precursor, the so-called common dendritic cell progenitor (CDP) (32, 33). Within this progenitor fraction, expression of the C type lectin receptor CLEC9A/DNGR-1 distinguishes cells with cDCrestricted developmental potential (34). These cDC restricted CDPs further differentiate into pre-cDCs, which leave the bone marrow and seed lymphoid and non-lymphoid tissues (31, 35) where they differentiate into the two main cDC1 and cDC2 subsets in response to environmental cues (23–25, 27). Of note, the signals that regulate cDC differentiation in tissues remain poorly defined and recent studies indicate that the commitment of pre-cDCs toward cDC1 or cDC2 may already be imprinted in the bone marrow (36, 37). In contrast to cDCs, pDCs exit the bone marrow as fully differentiated cells (38) and only a fraction of pDCs appears to belong to the myeloid lineage, whereas the majority of pDCs arises from lymphoid progenitors (39).

In adults cDC1 and cDC2 are developmentally and functionally distinct cell subsets that can be distinguished based on their differential dependence on transcription factors (23–25, 27). While cDC1 rely on BATF3, ID2 and IRF8 for their development, cDC2 require the transcription factors IRF4, ZEB2, and RELB and are additionally influenced by Notch2 signaling and retinoic acid (23–25, 27). Although not yet investigated in detail, at least some of these developmental pathways are conserved with age, as cDC1-like cells are missing in spleen, mesenteric lymph node and intestinal lamina propria of neonatal BATF3-deficient mice (40, 41). Additionally, DCs in early and late life require FMS-like tyrosine kinase 3 ligand (FLT3L) for their development (42, 43). In adults cDC1 can be reliably identified across tissues by expression of XCR-1, DNGR-1, and CD205 (23–25, 27). In addition, CD8α and CD24 mark cDC1 in lymphoid tissues (23–25, 27). The integrin CD103 marks cDC1 in non-lymphoid tissues, although it is also expressed on a subset of intestinal cDC2 (23–25, 27). cDC2 on the other hand are marked by expression of CD11b, CD172a, and CLEC4A4 (23–25, 27). Functionally, cDC1 are exceptional activators of CD8<sup>+</sup> T cells, in part for their superior activity to cross-present cell associated antigens (23–25, 27). cDC1 are additionally dominant inducers of Th1 polarized immune responses due to their strong capacity to produce IL-12 (23–25, 27). In contrast, cDC2 are generally thought to be more efficient at activating CD4<sup>+</sup> T cell and inducing Th2- or Th17-biased effector responses (23, 24, 44, 45). Since cDC1 and cDC2 have unique functions in immunity and can be distinguished by expression of select surface markers, they are attractive targets for the manipulation of immune responses in adults (26).

# ALTERED SUBSET DISTRIBUTION OF DENDRITIC CELLS IN EARLY LIFE

Although DCs reportedly can be found in mice as early as embryonic day 17, the DC compartment of newborn mice is not fully developed and subject to dynamic age-dependent changes during development (46, 47). In mice the neonatal period includes the first 10 days after birth, which correlates to the first 28 days of life in humans (1, 17). However, it is important to note that in terms of immune development there is substantial temporal variation between mice and humans in early life (48). In murine neonates the DC compartment in lymphoid and non-lymphoid organs is much smaller than that of adults and hallmarked by distinct DC subset distribution (46, 47). The frequency of splenic cDCs at birth is about ten-fold lower than that of adult spleen and, similarly, neonatal splenic pDCs are seven-fold lower in terms of frequency compared to their adult counterparts (46, 47). This is also reflected in lower numbers of DCs in neonatal spleen, which is not fully developed at birth in terms of organ architecture and size (46, 47). By about 5 weeks of age, when the total splenic cellularity reaches adult levels, both cDC and pDC numbers also reach adult levels (46, 47). A relative scarcity of DCs in newborn mice is also evident in other organs, such as thymus, lymph nodes, lung and intestine (47, 49–53). Altered immune responses and infection susceptibility in early life could therefore simply be a by-product of low DC numbers. Administration of the DC growth factor FLT3L leads to a strong increase in DC numbers in neonatal mice and results in increased resistance to Listeria monocytogenes (L. monocytogenes) and herpes simplex virus 1 (HSV-1) (43). Similarly, administration of FLT3L significantly enhances resistance of neonates to the intestinal parasite Cryptosporidium parvum by increasing the number of intestinal CD103<sup>+</sup> cDCs, which include both cDC1 and a fraction of cDC2 (54).

In adults, the cDC compartment is dominated by cDC2, while in early life cDC1 appear to be the dominant cDC subtype in spleen and lymph nodes (46, 47, 51). In thymus, cDC1 remain the dominant cDC population also in adults, possibly owing to a unique requirement of this cell type in ensuring central T cell tolerance (55). A systematic analysis of DC subset distribution with age across non-lymphoid organs has not been performed, with one exception being the lung, where cDC1 outnumber cDC2 in neonates, but this relationship is inversed in adults (50). It is important to note that most of these studies relied on the use of surface markers to identify cDC subsets. Notably, within the first 6 days of life, cDC1 from spleen and mesenteric lymph node lack CD8α, although they do express CD24, CLEC9A/DNGR-1, and CD205 (40, 46, 47). Expression of XCR-1 on neonatal cDC1 has not been investigated. These data have led to the suggestion that CD8α <sup>−</sup> cDC1 may represent a progenitor of bona fide cDC1 (40) and indicate that the use of surface markers to define cDC subsets in early life needs to be approached with caution. A summary of surface markers expressed on neonatal and adult DCs can be found in **Figure 1**.

Why DC subset distribution differs between neonates and adults is unclear. It is possible that DC differentiation may be intrinsically programmed to generate a functionally adapted DC repertoire that meets the needs of immune responses in early life. However, age-specific changes in specific organ environments could alter DC subset development. cDC1 in neonatal mediastinal lymph nodes express lower levels of CD103 than cDC1 from adult mediastinal lymph nodes (49). This difference in CD103 expression has been attributed to the unique cytokine environment of the lung in early life, such as low expression of GM-CSF (49). Thus, the DC compartment of neonates differs from that of adults in terms of cell number, subset distribution and marker expression.

# DENDRITIC CELL FUNCTION IN NEONATES VS. ADULTS

It is well established that neonatal DCs in both mice and men are functionally distinct from DCs in adults, which has been suggested to represent a level of functional immaturity. In mice, early life cDCs produce lower levels of pro-inflammatory cytokines than their adult counterparts. In the first week of life, splenic cDC1 produce lower amounts of IL-12 in response to CpG or after in vivo poly I:C treatment (40, 47). Similarly, splenic cDC2 produce less IFN-γ after stimulation with IL-12 and IL-18 compared to their adult counterparts and pDCs produce less IFN-α after combined treatment with CpG, GM-CSF, IL-4 and IFN-γ than pDCs from 6-week old mice (47). Interestingly, when cDC1 from Balb/C rather than C57BL/6 mice were analyzed, CpG-induced IL-12 production did not differ between cDC1 from 7-day old and adult mice (46). In cDC1 from C57BL/6 mice CpG-induced IL-12 production in early life can be augmented by addition of GM-CSF, IFN-γ, and IL-4 to culture conditions, however, the level of IL-12 produced still does not reach that of adult cDC1 (40, 47). These data indicate that some pathogen sensing pathways are fully functional in early life cDCs and that cytokine production of neonatal cDCs may at least in part be augmented through the use of additional costimulatory signals, which in turn could be used to boost immune responses in early life.

A key property of DCs is their ability to activate naïve T cells. Early life cDCs express lower basal levels of major histocompatibility complex class II (MHCII) and costimulatory molecules compared to adult cDCs (43, 47, 49) but expression of these molecules can be induced, for instance upon CpG stimulation in vivo (46). Despite these differences, splenic cDC1 and cDC2 from 1-week-old mice are able to induce allogeneic CD4<sup>+</sup> T cell proliferation in vitro to a similar extent as the same subsets from 6-week old mice (47). In the first week of life, CD8α <sup>−</sup> cDC1 phagocytose L. monocytogenes and crosspresent L. monocytogenes-derived antigens to CD8<sup>+</sup> T cells as efficiently as adult cDC1 (40). However, while cDC1 from adults respond to this pathogen in a predominantly proinflammatory manner, neonatal cDC1 additionally produce the anti-inflammatory cytokine IL-10, which suppresses CD8<sup>+</sup> T cell activation (40). Accordingly, IL-10 blockade augments the antigen-specific CD8<sup>+</sup> T cell proliferation induced by neonatal CD8α <sup>−</sup> cDC1 in vitro (40). Whether this IL-10 production has an impact on pathogen burden in vivo has not been examined, but these results indicate that the response to L. monocytogenes is intrinsically different between cDC1 from neonatal and adult mice. Following infection with respiratory syncytial virus (RSV) 7-day-old mice show an altered CD8<sup>+</sup> T cell response to that is hallmarked by an epitope shift toward DbM187−195, rather than the KdM282−<sup>90</sup> epitope that is immunodominant in adults (49). This epitope shift can be partially rescued by administration of costimulatory signals in vivo (51), suggesting that lower levels of costimulatory molecules on cDCs contribute to the observed epitope bias. However, it is noteworthy, that cDC1 from 7 day-old RSV infected mice preferentially present DbM187−<sup>195</sup>


FIGURE 1 | Overview of typical surface markers expressed on cDC subsets from mice and humans in early and adult life. (+) marker is expressed, (–) marker is not expressed or is expressed by a small fraction of cells, n.d., not determined; (↓) lower expression; CB, cord blood; FT, fetal tissues; cDC, conventional dendritic cell; MHCII, major histocompatibility complex class II; CLEC, C-type lectin domain family; HLA, human leukocyte antigen.

epitopes (49, 51), indicating that early life cDCs may exhibit intrinsic differences in antigen processing. Notably, epitope bias may also be found in humans, as infants infected with RSV show age-dependent differences in antibody specificities (56). cDC1 are also required for generating antigen-specific CD8<sup>+</sup> T cell responses to rotavirus, a major cause of childhood gastroenteritis (41). Interestingly, neonatal Batf3-deficient mice have a stronger impairment in the antigen-specific CD8<sup>+</sup> T cell response to rotavirus than adults, yet, Batf3 deficiency delays viral clearance only minimally in neonates (41). These data suggest, that in vivo other immune mechanisms are put in place that compensate for a lack of cDC1 in this case.

Antigen exposure in early life can elicit both Th1 and Th2 responses (1, 57, 58), however generates a strong bias for Th2 recall responses later in life. This has been partially attributed to a T cell intrinsic Th2 bias in early life but also to the altered cytokine production of cDCs in the first weeks of life, such as low-level IL-12 production (59, 60). Immunization with OVA before postnatal day 6 leads to an upregulation of IL-13 receptor α (IL-13Rα) in antigen specific Th1 cells (59). During secondary exposure to OVA, IL-13Rα expression renders antigen-specific Th1 cells sensitive to IL-4-induced apoptosis, thus leading to Th2-biased recall responses (58, 59). Exogenous administration of IL-12 or adoptive transfer of IL-12 competent cDCs from adult mice reverses the Th2-biased recall response (59), indicating that the low level IL-12 production by cDCs in early life exerts lasting effects on immunity. Why functional differences between neonatal and adult DCs exist, is unclear but several studies suggest that the neonatal environment functionally imprints DCs. As an example, in the developing lung IL-33 produced by epithelial cells during alveolarization on postnatal day 14 suppresses the ability of pulmonary cDC2 to produce the Th1 cytokine IL-12 (50). IL-33 instead promotes OX40L expression, which in turn leads to a stronger ability of cDC2 to promote Th2-biased responses and allergy (50). Th17 responses can be mounted in neonatal mice, for instance after infection with Yersinia enterolytica (61), but early life Th17 responses are damped by T cell derived IL-4 (62). The role of neonatal DCs in this process has not been investigated in detail.

Functional differences between early life and adult DCs have also been observed in humans. cDC2 from fetal spleen respond differently to various PAMPs than adult splenic cDC2 and as in mice, the response in the fetus is marked by higher production of anti-inflammatory cytokines (63). Although human fetal splenic cDC2s can induce allogeneic T cell proliferation, they limit T-cell-derived TNF-α production in an arginase 2-dependent manner and promote differentiation of Foxp3<sup>+</sup> regulatory T cells (63). pDCs from cord blood exhibit a severe defect in IFN-I secretion upon TLR9 stimulation with CpG when compared to pDCs from peripheral blood in adults, whereas the cytokine response to influenza A or human immunodeficiency virus is similar (64, 65). Collectively, these data show that the response of neonatal DCs to pathogens differs from that of adults in many ways, however, some signaling pathways induce immune responses that are comparable to those in adults.

#### NEONATAL DENDRITIC CELLS AS TARGETS FOR VACCINATION?

The unique properties of neonatal cDCs discussed above likely contribute to the relative ineffectiveness of vaccines in early life **(Figure 2A)** and it will be important to determine to what extent tailoring vaccines to the properties of early life DCs can be used to boost immunity. Similarities in surface receptor expression exist between neonatal and adult cDC subsets in both mice and humans **(Figure 1)** and specific targeting of cDC subsets holds promise for immunization in adults (26). Subset specific targeting using selectively expressed surface receptors can induce protective immunity in murine neonates. Targeting OVA to cDC1 via antibodies directed against CLEC9A with poly I:C as adjuvant on postnatal day 3 efficiently protects murine neonates against lethal challenge with OVA-expressing L. monocytogenes in adulthood (40). Whether targeting cDC2 in early life can induce protective immunity remains to be investigated. The prominent capacity of cDC2 to migrate to draining lymph nodes from the lung (50) indicates that they may be potent targets for initiating T cell responses. But several functions of neonatal cDC2 have not been studied in detail and it is unclear whether they induce effector T cell responses and promote T follicular helper cell differentiation and concomitant antibody production as efficiently as their adult counterparts.

Synergistic use of TLR agonists greatly increases the Th1 polarizing capacity of human adult DCs (66) and thus the use of defined PRR agonists may be used to promote immunity in early life. A recent study showed that stimulation with multiple TLR agonists elicits stronger secretion of Th1 polarizing cytokines from total human cord blood mononuclear cells than stimulation with a single TLR agonist, however, in cord blood cDCs a single TLR agonist induced stronger pro-inflammatory cytokine production than a combinatorial treatment (67). Agonists of the stimulator of interferon genes (STING) induce expression of costimulatory molecules on neonatal bone marrow derived DCs in vitro and promote secretion of IFN-β (68). In vivo

prove promising in developing more efficient vaccination protocols tailored to early life. Through age-specific epitope selection, DC subset specific epitope delivery, the combinatorial use of select PRR agonists as adjuvants, as well as via the manipulation of the DC compartment using growth factors, increased protective humoral and cellular immunity may be promoted.

administration of STING agonists in alum in neonatal mice promotes germinal center formation, IFN-γ production by antigen specific T cells, as well as increased antibody titers (68). Taken together these data highlight that the correct choice of adjuvant, either alone or in combination, is important for the design of effective vaccine strategies (17).

Immunization critically depends on the pathogen epitope selected to be vaccinated against. In neonatal mice, T cell responses are directed against a distinct set of antigenic epitopes compared to adults (49, 51). Similarly, in human infants neutralizing antibodies against RSV are preferentially generated against a distinct array of viral epitopes than in adults (56). Thus, selection of age-specific epitopes may also serve as a strategy to foster early life immunity and may help to circumvent deprivation of antigen by pre-existing maternal antibodies (17). At the same time, expanding the DC compartment or specific DC subsets using growth factors may ultimately shift epitope bias and cytokine production. Understanding the functional and developmental properties of neonatal cDCs and how to manipulate them may potentially be used to increase the effectiveness of neonatal immunization beyond what is possible today **(Figure 2B)**. However, immune

#### REFERENCES


responses in early life are complex and vaccine effectiveness is influenced by a wide array of factors, including preexisting maternal antibodies (17). Therefore, further studies, especially in humans, are required to better understand the unique aspects of DC development and function in neonates and in early life, as well as the interplay of DCs with other components of the immune system, in order to fully capture whether DCs could be exploited to alter early life vaccination.

#### AUTHOR CONTRIBUTIONS

NP and BS wrote the manuscript with assistance from MP. NP designed the figures. All authors provided critical feedback throughout the writing process.

#### FUNDING

Work in the Schraml lab is funded by an ERC Starting Grant awarded to BS (ERC-2016-STG-715182) and by the German Research Foundation: Emmy Noether Grant: Schr 1444/1-1 and SFB914 project A11 and SFB1335 project 08.


state and the inflamed setting. Annu Rev Immunol. (2013) 31:563–604. doi: 10.1146/annurev-immunol-020711-074950


intracellular pathogens and efficiently control virus infections. J Exp Med. (2003) 197:575–84. doi: 10.1084/jem.20021900


chimerism in IL-4-deprived mice. Am J Transplant. (2012) 12:81–9. doi: 10.1111/j.1600-6143.2011.03778.x


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Papaioannou, Pasztoi and Schraml. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Human Dendritic Cells: Their Heterogeneity and Clinical Application Potential in Cancer Immunotherapy

Thiago A. Patente<sup>1</sup> , Mariana P. Pinho<sup>1</sup> , Aline A. Oliveira<sup>1</sup> , Gabriela C. M. Evangelista<sup>1</sup> , Patrícia C. Bergami-Santos <sup>1</sup> and José A. M. Barbuto1,2 \*

<sup>1</sup> Laboratory of Tumor Immunology, Department of Immunology, Institute of Biomedical Sciences, University of São Paulo, São Paulo, Brazil, <sup>2</sup> Discipline of Molecular Medicine, Department of Medicine, Faculty of Medicine, University of São Paulo, São Paulo, Brazil

Dendritic cells (DC) are professional antigen presenting cells, uniquely able to induce naïve T cell activation and effector differentiation. They are, likewise, involved in the induction and maintenance of immune tolerance in homeostatic conditions. Their phenotypic and functional heterogeneity points to their great plasticity and ability to modulate, according to their microenvironment, the acquired immune response and, at the same time, makes their precise classification complex and frequently subject to reviews and improvement. This review will present general aspects of the DC physiology and classification and will address their potential and actual uses in the management of human disease, more specifically cancer, as therapeutic and monitoring tools. New combination treatments with the participation of DC will be also discussed.

Keywords: human dendritic cells, DC, monocyte-derived dendritic cells, mo-DC, cancer vaccines, cancer combination therapies

#### INTRODUCTION

Identified in mouse spleen for their peculiar shape and capacity to activate naïve lymphocytes (1– 3), dendritic cells (DC) are considered the most efficient antigen presenting cells (APC) (3, 4), uniquely able to initiate, coordinate, and regulate adaptive immune responses. Though their ability to capture, process and present antigens is considered their main characteristic, their phenotypic heterogeneity is striking and very different consequences can come from their action. This review will present an overview of the main subpopulations of human DC described and will focus on their potential translational use.

#### OVERVIEW OF DENDRITIC CELLS IN THE IMMUNE SYSTEM PHYSIOLOGY

Human DC are identified by their high expression of major histocompatibility complex (MHC) class II molecules (MHC-II) and of CD11c, both of which are found on other cells, like lymphocytes, monocytes and macrophages (5–12). DC express many other molecules which allow their classification into various subtypes (**Table 1**). Although some of the DC subtypes were originally described as macrophages, DC and macrophages have distinct characteristics (13–15) and ontogeny, so that, currently, little doubt remains that they belong to distinct lineages (16–24).

#### Edited by:

Daniela Santoro Rosa, Federal University of São Paulo, Brazil

#### Reviewed by:

Susan Kovats, Oklahoma Medical Research Foundation, United States Laura Santambrogio, Albert Einstein College of Medicine, United States

> \*Correspondence: José A. M. Barbuto jbarbuto@icb.usp.br

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 31 August 2018 Accepted: 24 December 2018 Published: 21 January 2019

#### Citation:

Patente TA, Pinho MP, Oliveira AA, Evangelista GCM, Bergami-Santos PC and Barbuto JAM (2019) Human Dendritic Cells: Their Heterogeneity and Clinical Application Potential in Cancer Immunotherapy. Front. Immunol. 9:3176. doi: 10.3389/fimmu.2018.03176

DC can be found in practically all tissues, where they detect homeostatic imbalances and process antigens for presentation to T cells, establishing a link between innate and adaptive immune responses. Furthermore, DC can secrete cytokines and growth factors (25) that modify ongoing immune responses, and are influenced by their interactions with other immune cells, like natural killer (26–28) and innate lymphoid cells (ILCs) (29).

DC are found in two different functional states, "mature" and "immature". These are distinguished by many features, but the ability to activate antigen-specific naïve T cells in secondary lymphoid organs is the hallmark of mature DC (30–32). DC maturation is triggered by tissue homeostasis disturbances, detected by the recognition of pathogen-associated molecular patterns (PAMP) or damage-associated molecular patterns (DAMP) (33, 34) (**Figure 1**). Maturation turns on metabolic, cellular, and gene transcription programs allowing DC to migrate from peripheral tissues to T-dependent areas in secondary lymphoid organs, where T lymphocyte-activating antigen presentation may occur (35–40).

During maturation, DC lose adhesive structures, reorganize the cytoskeleton and increase their motility (41). DC maturation also leads to a decrease in their endocytic activity but increased expression of MHC-II and co-stimulatory molecules (42–44). Mature DC express higher levels of the chemokine receptor,

TABLE 1 | Main surface markers of human and mouse DC subtypes.


FIGURE 1 | Dendritic cells activation. Extracellular signals, such as PAMPs or DAMPs, trigger alterations on immature DCs culminating on significant changes on surface proteins, intracellular pathways and metabolic activity.

CCR7 (45–48) and secrete cytokines, essential for T-cell activation (42, 49–52). Thus, the interaction between mature DC and antigen-specific T cells is the trigger of antigen-specific immune responses (53, 54). When interacting with CD4+ T cells, DC may induce their differentiation into different T helper (Th) subsets (52) such as Th1 (55–60), Th2 (56, 57, 61, 62), Th17 (63–65), or other CD4+ T cell subtypes (66) (**Figure 2**). T cell differentiation in each subtype is a complex phenomenon, that can be influenced by the cytokines in the DC tissue of origin (67), their maturation state (42) and cause of tissue imbalance (68). However, this process is not completely elucidated, as, for example, the source of IL-4 during Th2 responses, which is discussed extensively elsewhere (69).

DC present a unique characteristic: the ability to perform cross-presentation (70–74). This phenomenon was described in 1976, by Bevan (75) and is defined as the presentation, in the context of class I MHC molecules (MHC-I), of antigens captured from the extracellular milieu. This feature allows DC to trigger responses against intracellular antigens from other cell types, thus providing means for the system to deal with threats that avoid professional APC (70, 76, 77) and, even, to prime CD8+ lymphocytes in the absence of CD4+ T cells (78, 79). Crosspresentation is involved also in the induction of tolerance to intracellular self-antigens that are not expressed by APC and, then, called, cross-tolerance (80, 81).

Before receiving maturation stimuli, DC are said to be in an "immature state." Immature DC are poor inducers of naïve lymphocyte effector responses, since they have low surface expression of co-stimulatory molecules, low expression of chemokine receptors, and do not release immunostimulatory cytokines (44, 82). These "immature" cells, though, are very efficient in antigen capture due to their high endocytic capacity, via receptor-mediated endocytosis, including lectin- (83–85);

FIGURE 2 | CD4+ T cell fate induced by dendritic cells. When in contact with DC, naïve CD4+ T cells can differentiate into a number of subtypes. Among them, are regulatory T cells (Treg) and T helper (Th) subsets, which include Th1, Th2, and Th17 cells. Each subtype expresses different transcription factors, which regulate the function and cytokine secretion pattern of the cells. The T cell fate decision is a complex phenomenon that heavily depends on the interaction of DC with the T cells and the cytokines present in the microenvironment.

Toll-like- (86–88), FC- and complement receptors (89) and macropinocytosis (84). Thus, immature DC act, indeed, as sentinels against invading pathogens (32, 90), but also as tissue scavengers, capturing apoptotic and necrotic cells (91).

This latter feature confers to immature DC an essential role in the induction and maintenance of immune tolerance (31, 92– 95). Apoptotic cells that arise in consequence of natural tissue turnover (96, 97) are internalized by DC but do not induce their maturation (31, 98–100). Thus, their antigens are presented to T cells without the activating co-stimulatory signals that a mature DC would deliver, resulting in T cell apoptosis (80, 101), anergy (102, 103) or development into regulatory T cells (104, 105).

These "tolerogenic DC" express less co-stimulatory molecules and proinflammatory cytokines, but upregulate the expression of inhibitory molecules (like PD-L1 and CTLA-4), secrete antiinflammatory cytokines (IL-10, for example) (102, 106–108) and are essential to prevent responses against healthy tissues (30, 31, 109–112). However, in some contexts, immature DC can be harmful to the body. It is known that DC that are unable to induce lymphocyte effector responses may contribute to the immune system's failure to fight infections (113, 114) or tumors (115–120). In these situations, DC, even after recognition of pathogens or other changes in microenvironment, fail to increase the co-stimulatory molecules required to activate T cells, thus allowing the disease to "escape" immune control.

Although many factors are recognized as contributing to drive DC maturation (100, 121, 122), the full set of such factors is not precisely defined, but involves a long series of transcriptional adaptations (119, 121, 123–125). The complexity and heterogeneity of these adaptations allows DC to translate effectively (most of the times) the pattern of homeostatic disturbance to interacting T lymphocytes, thus establishing DC as the main connector between innate and acquired mechanisms of immunity (43, 126).

#### HUMAN DENDRITIC CELL SUBPOPULATIONS AND MONOCYTE-DERIVED DENDRITIC CELLS

Dendritic cells can be divided into resident lymphoid tissue DC and migratory non-lymphoid tissue DC (16). Both are heterogeneous cell populations with different subsets that can be distinguished by phenotypic markers and genetic profile. The first identification of different DC subsets arose from the observation that CD8 expression occurred on some, but not all, mouse resident splenic and thymic DCs (127). While the identification of mouse DC subpopulations is well advanced (128, 129), mostly due to tissue accessibility, the same is not true for human DC, where most studies were performed only in peripheral blood or skin, in spite of recent data characterizing DC subpopulations in human lung (130) and intestine (131).

Recent efforts have been addressed to understand the ontogeny and function of human DC subsets, attempting to correlate well-defined murine subpopulations with those found in human peripheral blood (16, 128, 132). DC arise from a CD34<sup>+</sup> hematopoietic precursor that gives rise to myeloid (MP) and lymphoid (LP) precursors (**Figure 3**). MP differentiate into monocyte, macrophage and DC precursors (MDP), which will give rise to monocytes and to the common DC precursors (CDP). CDP can differentiate into plasmacytoid DC (pDC) or the preclassical DC (pre-cDC). Pre-cDC are the progenitors of the two major cDC subpopulations named cDC1 and cDC2 (14), which will be further discussed latter. Recent technologies, such as single cell RNAseq, are allowing a better characterization of DC ontogeny and the identification of DC subset precursors in peripheral blood (133), demonstrating that the commitment with a DC subset may be an early event, both in mice (134) and humans (135).

Curiously, in lymphohematopoietic tissue, such as spleen, thymus and blood, DC commitment to a subpopulation is mainly defined by ontogeny, while in non-lymphohematopoietic tissue, such as lung and skin, DC subpopulations are more influenced by signals derived from the microenvironment. This, once again, confirms that DC are a very plastic cell population that can shape its phenotype to the microenvironment and to homeostatic state of the tissue where it is located (136).

In blood, DC constitute a rare cell population that can be broadly divided into two subtypes (**Figure 4**): CD123+CD11c<sup>−</sup> DC, called plasmacytoid DC (pDC), and CD123−CD11c<sup>+</sup> cells, called classical DC or myeloid DC (cDC) (25, 128, 137). Dzionek et al. (138) identified three antigens called BDCA-2, BDCA-3, and BDCA-4 (Blood Dendritic Cell Antigens), which, together with BDCA-1 (CD1c), allowed the further discrimination of human blood DC subsets. cDC can be separated into cDC1 and cDC2 (139): cDC1 are characterized by the expression of BDCA-3 (CD141) and Clec9A, while cDC2 express CD1c. BDCA-2 (CD303) and−4 (CD304), on the other hand, together with CD123, characterize pDC.

It is noteworthy that recent genomic studies, with emphasis on the subpopulations of monocytes and DC, made it possible to align CD141<sup>+</sup> DC (cDC1) and CD1c<sup>+</sup> (cDC2) from human peripheral blood with the mouse CD8α <sup>+</sup>/CD103<sup>+</sup> and CD11b+DC, respectively (140, 141). This will allow the confirmation, or not, of the roles played by these subsets in murine immune responses also in humans.

#### cDC1

The human cDC1 subpopulation is present in blood and in lymphoid and non-lymphoid tissues (142). This subpopulation is characterized by the expression of CD141, the chemokine receptor XCR1, C-type lectin CLEC9A, the cell adhesion molecule CADM1, and is the counterpart of mouse CD8α <sup>+</sup>/CD103<sup>+</sup> cross-presenting DC subset (132, 142). cDC1 can be generated in vitro from CD34<sup>+</sup> progenitors after 21 days of culture with fms-like tyrosine kinase 3 ligand (Flt3L) and thrombopoietin (TPO) (143) or with Flt3L and murine bone marrow stromal cell lines (144). As mentioned above, this subpopulation of DC seems to be specially adapted to perform cross-presentation, a phenomenon that is associated with the expression of the chemokine receptor XCR1 (145). The main transcription factors (TF) shown to be essential for the generation of cDC1 are the basic leucine zipper transcriptional factor ATF-like 3 (BATF3) (146) and IFN-regulatory factor 8

pre-classical dendritic cells; pDC, plasmacytoid dendritic cells; cDCs, conventional dendritic cells; FLT3, Fms-Related Tyrosine Kinase 3.

(IRF8) (130). In mice, besides BATF3 (147) and IRF8 (148), gene knockout models pointed out to the role of two other TF: DNA binding protein inhibitor ID2 (149) and nuclear factor interleukin-3-regulated protein (NFIL3) (150), whose participation in the generation of human cDC1 needs yet to be demonstrated.

cDC1 prime CD8<sup>+</sup> T cells efficiently, what is important in anti-tumor and anti-virus immunity. However, the induction and modulation of an immune response is a very complex phenomenon that involves many cell interactions, including interactions among different DC subsets, as recently demonstrated in mice infected with modified vaccinia virus Ankara (MVA) (151). In this model, activated CD8<sup>+</sup> T cells recruit both pDC (via CCL3/CCL4) and cDC1 (via XCL1); type I interferons, (IFN-I) produced by pDC, act on cDC1 optimizing their maturation, costimulatory capacity and ability to cross-present viral antigens, thus leading to an effective antivirus response. cDC1 were also shown to be important for the antitumor activity induced by heat-inactivated MVA in murine melanoma and colon cancer models (152). Furthermore, both in mice and humans, cDC1 are found sparsely distributed along tumor margins (competing with tumor associated macrophages– TAM-for tumor antigens?) and their presence was important for the success of adoptively transferred cytotoxic T cells (CTL) (153) and for the delivery of tumor antigens to the draining lymph nodes, in a CCR7 dependent manner (154).

#### cDC2

cDC2 constitute a heterogeneous subset of DC that can be found in blood, lymphoid and non-lymphoid tissue (16, 142). SIRPα (CD172a) is expressed by cDC2 (both in humans and mice) (130) and, along with CD1c (humans) and CD11b (mice), characterizes this subpopulation (25, 132). Coherently with its heterogeneity, other markers are expressed by cDC2, according to their localization, as for example, CD1a in dermal and CD103 in gut cDC2 (25, 141). Like cDC1, cDC2 can also be differentiated from CD34<sup>+</sup> progenitors, after 21 days of culture with Flt3L and TPO (143) or with Flt3L and murine bone marrow stromal cell lines (144). More than one transcription factor is involved in cDC2 differentiation and IRF4 seems to be the master transcription factor (155), but other transcription factors are required. In mice, PU.1 (156), RelB (157) and recombining binding protein suppressor of hairless (RBP/J) (158) were shown to be associated with the differentiation of cDC2, and in humans, IRF8 (159).

Again, in accordance with their heterogeneity and innate plasticity (132), cDC2 have been show to induce Th1, Th2, and Th17 responses (160, 161). The puzzling heterogeneity of these cells is further illustrated by the recent description of two novel DC subtypes within the CD1c<sup>+</sup> subpopulation: DC2 and DC3. These two subpopulations diverged by the expression of CD32B and CD163/CD36. Functional experiments showed that both these cDC2 subtypes were potent stimulators of naïve T cell proliferation, but show a different pattern of cytokine secretion after stimulation with a series of toll like receptors (TLR) agonists (162).

In the immune system physiology, cDC2 seem to have many, but frequently, regulatory roles. These cells have been described as potent inducers of regulatory T cells in intestine (141), and as responsible for maintaining tolerance in the liver (163). Also, cDC2 have been described as the only DC subset able to produce retinoic acid upon stimulation with vitamin D3, thus stimulating CD4<sup>+</sup> naïve T cells to express gut-homing molecules and to produce Th2 cytokines (164).

# Plasmacytoid DC (pDC)

The pDC subpopulation is a subset of DC distinct from cDC, that arises directly from the CDP (while cDC arise from pre-DC precursor) (14). These cells are characterized by the secretion of high levels of IFN-α/β upon TLR7/9 stimulation, and are extremely important in viral infections (165). This subset of DC is phenotypically distinct in mice and humans. In mice, it is characterized as CD11cintCD11b−B220+SiglecH+CD317<sup>+</sup> while in humans it is characterized by the absence of expression of CD11c and the expression of CD123, CD303, and CD304 (25, 128, 132). In terms of transcription factors, on the other hand, both mouse and human pDC seem to depend on the same master transcription factor, E2.2 (25, 132, 166).

Since the secretion of IFN-α/β is the main feature of pDC, their association with viral infections is not surprising. The secretion of IFN-α/β by pDCs can be a consequence of direct viral infection [like in HIV infection, where the virus infects pDC via CD4, CCR5 and CXCR4 (167)], or from external stimuli. Indeed, human pDC were shown to secrete high levels of IFN-α/β in Aspergilus fumigatus infection in a Dectin-2-dependent manner (168).

In keeping with the other DC subpopulations heterogeneity, human pDC may be subdivided into two subpopulations, distinguished by the expression of CD2 (169). Both pDC subsets secrete IFN-α/β efficiently, but only the CD2hi subset secretes IL-12p40 and induces CD4<sup>+</sup> T cell proliferation. These data, however, may be in need of a second look. As mentioned before, single cell RNAseq analysis is providing new data and allowing better characterization of DC subpopulations. When this approach was used to study pDC subpopulations, a "contaminant" putative precursor of cDC (pre-cDC), characterized as CD123+CD33+CD303+CD304+CD2+, was identified. When these putative pre-cDC and "pure" pDC populations (characterized by the absence of CD2 and CD33 expression) were separated and stimulated, only pre-cDC were able to induce CD4<sup>+</sup> T cell proliferation and secrete IL-12p40 (135). This raises the possibility that many of the observed attributes of pDC, such as their ability to induce Th1 responses (170), to perform cross-presentation (171), to exhibit naïve T cell allostimulatory capacity (169) and expression of co-stimulatory (172) molecules might reflect the activity of this contaminating pre-cDC population.

Puzzling, as these data may seem, they illustrate quite well the plasticity of the cells "clustered" under the name of DC. They further suggest that attempts to classify strictly these cells may lead to more confusion than it is necessary to understand their role in responding to microenvironmental challenges, in shaping immune response patterns in the body and, eventually, in driving the immune response toward therapeutic goals in humans.

# Monocyte-Derived DC (mo-DC)

Much of the knowledge acquired in the past years about human DC biology was possible due to the methodology of in vitro deriving DC from CD34<sup>+</sup> precursors (stimulated with GM-CSF and TNF-α) (173) or from monocytes (stimulated with GM-CSF and IL-4) (174). Like cDC2, mo-DC depend on IRF4 for their differentiation (175). However, they do not seem to be an equivalent population, since they arise from different precursors (14).

In mice, the precursors used for in vitro generation of DC are extracted from the bone marrow. In the presence of GM-CSF, these precursors give rise to large number of cells that resemble tissue DC and are called bone marrow-derived dendritic cells (BMDC) (176). Helft et al. showed that BMDC comprise a heterogeneous population expressing both CD11c and MHCII. A CD11c+MHCIIint population seems to be more closely related to macrophages (hence, called GM-Macs), while the CD11c+MHCIIhigh population resembles DC and is, thus, called GM-DC. Addition of IL-4 to these cultures limits, but does not eliminate, the generation of GM-Macs (177). The heterogeneity of precursors and cell populations obtained in vitro fuels a vivid and complex discussion about the biological relevance of these cells (178–180).

It is still unclear to which subpopulation of DC, mo-DC are more closely related, but DC ontogeny data suggest that mo-DC are similar to the inflammatory DC (132). Not surprisingly, inflammatory DC is the designation of another heterogeneous subpopulation of DC, typically CD11chiMHCIIhi. One of the first reports of inflammatory DCs described a population of DC characterized by the production of TNF and iNOS, named Tip-DCs (181). Another study identified inflammatory DC in the skin of atopic dermatitis patients and named these cells inflammatory epidermal dendritic cells (IDECs), which were characterized by the expression of CD11c, CD206, CD1a, CD11b, CD209, FcεRI (182). Recently, another inflammatory DC population was described in the synovial fluid of rheumatoid arthritis patients and in the inflammatory ascites of untreated cancer patients. In this study, inflammatory DC were characterized as CD14+CD1c+SIRPα <sup>+</sup>CD206+FcεRI<sup>+</sup> and their gene signature (when compared to in vitro generated mo-DC, macrophages, cDC2, CD16<sup>+</sup> monocytes and CD14<sup>+</sup> monocytes) was more closely related to that of mo-DC, suggesting that inflammatory DC could be, indeed, the in vivo counterparts of mo-DC (183).

#### MO-DC AS A "WINDOW" TO IMMUNE SYSTEM EVALUATION IN CANCER PATIENTS

It has been known for a while that established tumors affect their microenvironment in ways that facilitate their persistence and progression. These local modifications include zones of hypoxia, altered pH, induction of angiogenesis (184), alterations of premRNA splicing in surrounding cells (185) and the recruitment of cells that facilitate tumor progression, such as tumor-associated macrophages (TAM) (186), immature DC (115), myeloid-derived suppressor cells (MDSC) (187) and regulatory T cells (188). However, mechanisms to avoid immune system surveillance and tumor progression (189) are not limited to the tumor site and, today, it is recognized that individuals with cancer present also systemic modifications to that effect as well (190). As discussed before, DC are a plastic and heterogeneous population and it should be expected that, among these systemic adaptations, some affect the various DC subpopulations, including mo-DC.

### Described Alterations in mo-DC of Cancer Patients

Various publications have described phenotypic and functional alterations in mo-DC from patients with different tumors (191–193). Our group demonstrated that mo-DC from breast cancer patients are poor stimulators of allogeneic T lymphocytes proliferation but are good inducers of regulatory T cells. These characteristics were observed both in immature and mature mo-DC and the regulatory T cell bias, though decreased by blocking of TGF-β, was not completely inhibited (192). Similar phenomena were also observed in patients with CLL, whose mo-DC expressed reduced levels of important molecules involved in antigen presentation and lymphocyte activation, such as HLA-DR, CD80, CD86, CD83, and CD40, and, coherently, were less effective in inducing proliferation of both CD4<sup>+</sup> and CD8<sup>+</sup> T cells. Furthermore, CD4<sup>+</sup> T lymphocytes co-cultivated with mo-DC from CLL patients presented reduced IFN-γ and IL-4 production, when compared to healthy donors (193). Further similar results were also observed in chronic myeloid leukemia (194), colorectal cancer (195), and cervical neoplasia (196). It is worth noting that dysfunctional and apoptosis prone mo-DC were also obtained from healthy donors, when their monocytes were exposed to tumor culture supernatants (197).

Although detected in cancer patients, the altered phenotype and functions of mo-DC could precede the emergence of the tumor and reflect an individual constitutional characteristic of the patients, which might be related or not to their disease. The follow up of cancer patients that present such alterations, however, suggests otherwise and indicate that, indeed, it is the presence of the tumor that affects the cells.

In a study of a chromophobe renal carcinoma patient, mo-DC obtained before surgery induced less allogeneic T cell proliferation and more regulatory T cells when compared to cells from healthy donors. Three months after surgery, yet, mo-DC from the patient exhibited functional properties similar to that of healthy controls, suggesting that the presence of the tumor was the cause of the biased mo-DC function in the patient (198). Another example of the transitory and, possibly, in this case, tumor-dependent functional bias of circulating cells has been described in a study with patients with obstructive jaundice. Monocytes from 53 patients with obstructive jaundice (44 due to cancer and 9 due to non-neoplastic diseases) were obtained before surgery and found unable to release H2O<sup>2</sup> upon stimulation, but this was progressively reversed after surgery (199). Yet, in another paper we described a patient with type 2-papillary renal cell carcinoma, whose mo-DC also presented functional biases. Though after the tumor was surgically removed, the patient's mo-DC already regained some activity, their T lymphocyte-stimulating activity reached healthy controls' levels only after the patient was submitted to treatment with a dendritic cell-based cancer vaccine (200).

Altogether, these data point out to the fact that circulating monocytes may reflect systemic effects of tumors in such a manner that their functional evaluation could become an effective tool to monitor disease progression and/or response to therapy.

#### Alterations in Circulating Subpopulations of DC in Cancer Patients

Circulating subsets of DC are also affected in cancer patients. Diminished numbers of total DC have been observed in melanoma patients; this was more intense in stage IV patients and, though it was more pronounced in the pDCs, it also occurred among cDC (201). In breast cancer patients, reductions in total circulating DC and in DC IL-12 production was also described. However, in these patients cDC were the culprit and not pDC (117). Circulating DC isolated from patients with CLL showed decreased expression of co-stimulatory molecules, lower ability to stimulate allogeneic T lymphocytes and did not secrete IL-12, but retained the ability to secrete IL-10 (202). A recent publication, evaluating the effects of different TLR-L in cDC1, cDC2, pDC, and monocytes from breast cancer patients showed that, upon stimulation with IFN-α, cDC2 and nonclassical monocytes (CD14−CD16+) exhibited reduced secretion of TNF-α (203).

These observations point out to systemic effects induced by tumors upon the immune-hematopoietic system and suggest that circulating cells are influenced and, possibly functionally handicapped to fight the tumor, even before actually infiltrating the tumor mass. These phenomena, added to our view and understanding of tumor biology, should allow the design of improved therapeutic approaches, even for those that do not specifically target the immune system.

#### Possible Mechanisms

It is quite evident, thus, that tumors promote local and systemic alterations in immune cells and substantial efforts have been made to identify possible mechanisms of how tumors promote these alterations and, most importantly, how to correct them.

Signal Transducer and Activator of Transcription 6 (STAT6) is an important molecule, induced by IL-4, in the process of mo-DC differentiation. STAT6 is naturally inhibited by the Suppressor Of Cytokine Signaling 5 (SOCS5), which, in turn, is up regulated by phosphorylated STAT3 in monocytes. In CLL patients, IL-10 induces the phosphorylation of STAT3, thus up regulating the expression of SOCS5. As a consequence, monocytes of CLL patients have impaired phosphorylation of STAT6 and its downstream genes, blocking their differentiation and maturation into functional mo-DC (193). However, mo-DC from healthy donors differentiated in the presence of lung cancer patients' sera, showed decreased STAT3 phosphorylation (204). Although apparently contradictory, these findings might reflect a difference in the monocytes of patients and healthy donors or a difference in the moment of analysis. If monocytes from patients and healthy donors differ, it would not be surprising that they would respond differently to the same stimuli. Likewise, the moment when STAT3 phosphorylation is analyzed may show quite different results. When monocytes from healthy donors were pre-treated with IL-10 and then stimulated with IL-4, an initial increase in STAT3 phosphorylation occurred during the first 72 h, but with the increasing SOCS5 expression, STAT3 (and STAT6) phosphorylation was downregulated (193).

The STAT3 pathway is activated also by IL-6, which, like IL-10, is found in higher concentration in patients sera (205). The impaired functions of DC have been, thus, also attributed to upregulation of IL-6-induced STAT3 activity, both in animal models (206) and humans (207)- these data were recently reviewed by Kitamura et al. (208). Offering a potential solution to these hard to reconcile data is the fact that STAT3 signaling induced by IL-6 seems to be modulated by SOCS in a different way than the IL-10-induced signaling, at least in human macrophages (209).

Undeniably, the available data, though suggesting possible pathways are not enough to elucidate the complex molecular mechanisms underlying DC dysfunction in patients.

#### DENDRITIC CELLS AS THERAPEUTIC INSTRUMENTS

The key concept of the cancer immunotherapy is that the manipulation of the immune system can achieve cancer control and, ideally, cure. The possibility of cancer immunotherapy was first shown by Coley, who used a mixture of bacterial toxins to treat patients with inoperable sarcomas (210). Since then, many studies have shown clinical benefit when using general immune system activators, such as bacterial products (211) and TLR agonists (212). The antitumor activity of these approaches, when it occurs, is attributed to the ability of these compounds to activate the immune system that, in turn, acquires the ability to kill tumor cells. Much of this effect was shown to be due to DC activation followed by the generation of T cell responses (213). Dendritic cells, as key activators of the adaptive immune response, would be expected to have a central role in inducing antitumor immune responses and the many functional deviations these cells show in cancer patients emphasize the relevant role they may, indeed, play in anti-tumor immune responses. In face of these data, it would be intuitive to exploit the immune activating potential of DC to induce antitumor responses in cancer patients. However, because of the difficulty of obtaining large numbers of these cells by non-invasive methods, therapeutic approaches using DC became possible only after methods for the in vitro generation of these cells were described (174).

# Use of Monocyte-Derived Dendritic Cells

mo-DC are able to present antigens in the context of both MHC class I (91) and class II molecules (214) and, hence, can be used to generate therapeutic cancer vaccines. When injected in humans, mo-DC can prime CD4+ and CD8+ T cells (215) and expand antigen-specific cytotoxic T cells, which can lead to regression of metastatic lesions in patients (216). Nevertheless, some argue that mo-DC, possibly due to a limited migration potential, might be insufficient to consistently induce effective immune responses in vivo (217). Contrastingly, Kuhn and co-workers have shown that successful therapy using immune-activating compounds was followed by the appearance of mo-DC in the draining lymph nodes of treated mice (218) and these cells were essential for the priming of CD8+ T cells and antitumor immunity (219).

Nonetheless, to be used as therapeutic instruments, mo-DC must be properly differentiated in vitro, induced to mature, loaded with tumor antigens, and, finally, administered to the patient (**Figure 5**). It is easy, thus, to realize the challenges that face the development of mo-DC-based vaccines. What are the markers of a "properly activated" DC? What is the "proper" response to be induced? What are the relevant tumor antigens? What is the best pathway for these cells to reach secondary lymphoid organs, where they should encounter tumor-specific T lymphocytes? Not surprisingly, each of the aforementioned steps diverges among the various clinical reported protocols, adding much complexity to the evaluation of the approach, but also a possible explanation for the large diversity in the reported efficiencies of such treatments.

To differentiate monocytes into dendritic cells, the cytokines IL-4 and GM-CSF are classically used (174). Most approaches use this protocol to obtain mo-DC, but other ways to differentiate monocytes into dendritic cells have been described and tested. mo-DC differentiated in the presence of GM-CSF and IFN-α, for example, secrete large amounts of pro-inflammatory cytokines, induce a IL-12p70-independent Th1 response (220) and have

given rise to cancer-specific CD8 responses, in phase I/II clinical trials (221). mo-DC differentiated in the presence of GM-CSF and IL-15, on the other hand, were better inducers of Th17 responses (222).

The lengthy culture time to achieve the differentiation of mo-DC (usually 5–7 days) is a limitation of the wide clinical use of these protocols. Thus, alternative protocols for mo-DC differentiation were developed. Dauer et al. have shown that monocytes cultured for 48-hours with IL-4 and GM-CSF already have characteristics of immature DC (223) and these, so called FastDC, prime tumor-antigen specific CD8 T cells as efficiently as conventional mo-DC (224). Another strategy is the transduction of monocytes with plasmids containing the genes of the cytokines, which, constitutively expressed, will lead to their differentiation into DC (225). The FDA-approved cancer vaccine, Sipuleucel-T (PROVENGE <sup>R</sup> ) uses a similar approach for mo-DC generation, in a protocol that only requires 3 days for manufacturing (226). This vaccine is approved for castrationresistant prostate cancer and consists of autologous PBMC incubated with a fusion protein containing both GM-CSF and PAP, a prostate-specific cancer-associated antigen (227).

The second step in vaccines generation consists of mo-DC activation, since differentiation generates immature cells. The maturation stimulus can come from a variety of molecules, including cytokines (TNF-α, IFN-γ), TLR agonists (LPS), agonistic recombinant proteins (CD40L) or maturation cocktails (228). However, the best conditions for mo-DC activation are still unclear. Activation with TNF-α, for example, has been implicated in the induction of mo-DC with impaired ability to secrete pro-inflammatory cytokines, which could even protect mice from autoimmunity (229). On the other hand, combinations of TLR agonists synergize to promote Th1 responses (230). Vopenkova et al. made a direct in vitro comparison of different maturation stimuli to induce tumorspecific T cells, showing that the highest response was achieved with the combination of IFN-γ and LPS (231). However, clinical effectiveness comparisons of different mo-DC formulations are still lacking.

Next, mo-DC need to be loaded with tumor antigens. For this, bulk tumor products or selected tumor antigens have been used. Tumor associated antigens (TAAs), recognized by T cells, are found in several tumors (232). Immunodominant synthetic peptides derived from TAAs have been tested and were able to induce clinical and immunological responses of the vaccinated patients (233). Also DNA molecules encoding TAA genes can be employed to load mo-DC, in which case, viral vectors, intrinsically able to activate DC (234), bring further advantage. It is noteworthy that, for all these methods, there is no need of tumor samples from the patient, which may be scarce. However, the use of single antigens has its drawback. Due to the cellular heterogeneity of tumors, they can escape from the immune response generated by the vaccine, through the selection of cells that do not express the immunizing antigens (235). Strategies that involve the induction of a polyantigenic response can be used to avoid this resistance, especially in melanoma, where this effect is frequently observed. Bulk tumor products may be used as a broad source of tumor antigens.

In addition to tumor lysates, living tumor cells, necrotic debris, apoptotic bodies and tumor-derived exosomes have been used (236). The type of antigenic source used, however, can interfere with the type of immune response obtained and it is impossible, today, to predict which would be the most appropriate antigenic source. For example, in mice, dendritic cells loaded with apoptotic tumor cells were show to induce better responses than tumor lysates, peptides or RNA (237), a finding that contradicts the many data showing that apoptotic cells captured by DC constitute a mechanism of immune tolerance induction.

Although several protocols of vaccination with mo-DC have been tested in clinical trials, only a few obtained relevant clinical responses, and most of them failed to reach the expected results (238). The lack of success in these approaches could be attributed to the functional alterations found in cancer patients mo-DC (239). The use of allogeneic mo-DC obtained from healthy histocompatible donors would be a strategy to bypass this problem, although limited by the need of a MHC-matched donor. Another approach is the use of dendritic cell-tumor cell hybrids. These fused cells express MHC molecules from both tumor and DC origin, forsaking the need of a MHCmatched donor to generate the mo-DC (240). They are also superior than the mixture of these cells, induce antitumor responses and clinical response in patients with advanced metastatic tumors (241). Regardless of the strategy, however, clinical responses to mo-DC-based vaccines are still beyond the desired. This suggests that it may be not enough to have an efficient antigen presentation to induce tumor regression, once it is established. Other compromises between the tumor and the immune system might still prevent an effective tumor-clearing immune response requiring the design of new approaches and, very likely, the combined targeting of different immunological pathways.

#### Targeting DC Subsets in vivo

More recently, a new modality of DC-based immunotherapy strategy is under development. With the better DC subsets characterization and the identification of specific surface markers for these subsets, it became possible to design strategies to deliver different molecules or "packages" to these cells in vivo (242). This would allow the selective delivery of antigens and/or immunostimulatory molecules to defined cell subtypes in vivo, preventing the costly and laborious ex vivo mo-DC generation.

Among the most studied DC-targeting antibodies are those specific for DEC205, CLEC9A, and CLEC12A. These C-type lectin receptors are expressed, in mice, by cDC1 and, the last two, also by pDC (243). Due to their cross-presentation ability, targeting to cDC1 seems to be a reasonable choice, which would favor a higher CD8+ T cell response.

Indeed, experimental settings targeting these molecules were able to induce T cell responses (244, 245) and regression of metastatic melanoma in mice (246). Interesting and well designed as this strategy may be, in humans this strategy is still restricted to in vitro studies (247) and awaits, urgently, translational research.

#### STRATEGIES TO IMPROVE THE CLINICAL EFFECTIVENESS OF MO-DC-BASED THERAPIES

Before specifically addressing the many current pathways for the improved translation of our knowledge of DC biology into clinical applications, it is worth mentioning that, though most of this effort is concentrated into the use of these cells to induce effector immune responses, it is only a matter of time till it becomes feasible to delineate DC-based strategies to treat conditions where the immune system went rogue and is causing autoimmunity, or where medical interventions require the limitation of immune responses, like organ transplantations.

That said, let us consider the strategies that may lead to enhanced immunogenic effects of mo-DC-based treatments.

## Approaches for the Improvement of DC-Based Treatments

Since mo-DC show deviant phenotypes in cancer (192) and are susceptible to negative modulation by different drugs, for example STAT5 inhibitors (248), the converse is also true and various approaches are under development to achieve the generation of "better" mo-DC.

The chemokine CXCL-4 is a powerful chemoattractant to monocytes and an important immunoregulator that has been shown to enhance the expression of MHC, CD86, and CD83 molecules by mo-DC of healthy donors, leading to more efficient antigen presentation, induction of CD4<sup>+</sup> and CD8<sup>+</sup> T cells proliferation and production of IFN-γ (249).

As mentioned before, IL-6 through the activation of STAT3 interferes with proper DC maturation and, indeed, in patients with colorectal cancer has been associated with poor CD4<sup>+</sup> T cells responses (207). Coherently, a phase-I study in ovarian cancer patients showed that, combined with chemotherapy, IL-6 blockade was safe and induced a series of positive modifications in immune parameters of the treated patients, including increases in IL-12, IL-1β, TNF-α, and IFN-γ secretion (250).

Besides targeting the negative regulators of DC activation, it is possible to overcome this phenomenon by changing the activating signals delivered to these cells. Following this line of research, a cocktail of inflammatory cytokines (TNF-α, IL-1β, poly I:C, IFN-α, and IFN-γ) has been tested for mo-DC maturation and was shown to increase their IL-12 production and their ability to prime melanoma-antigens-specific T cells in vitro (251). This mo-DC activating cocktail, in a vaccination study of 22 recurrent glioma patients, was associated with increases in serum type 1 cytokines and chemokines, tumorassociated antigens-specific T cell responses and clinical benefit in 9 patients (252).

Another approach is based on the use of adjuvants to boost the immune response. Among these, GM-CSF used in vaccines as GVAX (253) and STINGVAX (254) and, even TLR agonists (255), may be more effective for cell maturation. Other adjuvants could be listed, as for example, aluminum salts (256) (an inflammasome activator), and montanide (257) (an equivalent to incomplete Freund's adjuvant). Those adjuvants may boost responses due to physical effects upon antigens and cells, but also enhance DC activation. Nonetheless, the consideration of such a heterogeneous group of substances is enough to realize that adjuvant research is a rich field that may broaden the applicability and enhance the effectiveness of DC-based vaccination (258).

A different pathway to improve the effectiveness of DCbased therapy focuses on the selection of the immunizing antigens. In cancer, the mapping of a patient's set of neoantigens and use thereof would represent the epitome of personalized medicine. Though very tempting, this approach would still have its drawbacks, a significant one being the fact that not all tumors express immunogenic neoantigens (259), not to mention the cost that such strategy would impose on any health care system. Nevertheless, its feasibility and efficacy has already been demonstrated in an elegant study (260) where personalized vaccines were prepared for 6 melanoma patients. Whole-exome sequencing of their tumors allowed the identification of the mutated antigens from which a set of peptides was selected and synthesized so that they would be presented in the context of MHC-I. Four patients presented complete clinical responses to the vaccine alone and the other two, who had progressive disease after the vaccination, experienced complete responses after treatment with anti-PD-1. Curiously, in spite of the selection of MHC-I selective peptides, both CD4+ and CD8+ antigenspecific T cells were stimulated, with a predominance of CD4+ T cell responses. This observation illustrates very well how much "real life" immune responses still differ from our predictions.

Another ingenious strategy bypasses many of the known hurdles to exploit the immunogenic potential of DC. This approach aims to deliver RNA-containing nanoparticles systemically, which due to their lipid composition would be preferentially captured by DC and, then, release the RNA encoding the selected antigen(s) to be synthesized and presented. In a murine model, this approach lead, indeed, to DC maturation, IFN-α production and strong antigen-specific immune responses, which were effective in a series of tumor models (261). Accordingly, this strategy is under investigation in a clinical trial (NCT02410733) for patients with advanced melanoma.

#### Combination Treatments Including mo-DC

Chemotherapy and radiotherapy, together with surgery, still remain as the main pillars of cancer therapy. Since chemotherapy in general was formerly considered immunosuppressive, little attention was given to the fact that this is not always true. Indeed, some drugs might potentiate the anti-tumor immune response, by inducing the now recognized "immunogenic cell death" (262, 263). However, due to the frequently observed cancer patients' DC dysfunctions, the simple immunogenic death may not be enough to disrupt the tumor-favoring status of the immune response in patients. To achieve that, active immune interventions may be necessary to take advantage of the phenomenon. Indeed, a series of studies, both experimental and in humans, has been addressing this issue with promising results (264–266).

Radiation may also favor the induction of anti-tumor immune responses and, as with chemotherapy, there are plenty of data indicating a beneficial effect of its combination with cancer vaccines or other immune-stimulating strategies in different settings, including hepatocellular carcinoma (267), prostate cancer (268), lymphoma (269), and glioblastoma (270). Currently, the potential of such combinations are under scrutiny in a series of clinical trials for patients with such disparate diseases as anal (NCT01671488), lung (NCT01579188) and pancreatic cancer (NCT01072981) (271).

The disparity of the diseases mentioned at the previous paragraphs is a good indicator of the contrast between therapeutic strategies directed against the tumor cell and those targeting the immune system. Those that aim at the tumor cell will differ significantly from one tumor to the other, since each tumor has its own set of genetic changes and will respond differently to a given treatment. On the other hand, strategies that target the immune system, though still dealing with a very complex set of interactions, will face, very frequently, standard responses of the immune system to the perturbations caused by the presence of the tumor, regardless of the tumor's set of genetic mutations.

Actually, the realization of this scenario and the better understanding of the immune system and its interactions with tumors opened the way to a very attractive and successful approach for cancer immunotherapy: instead of targeting directly the tumor, one could target the immune regulatory mechanisms that allow a frequently immunogenic tumor to grow in an otherwise immunocompetent host. With this, the "checkpoint inhibitors era" started and achieved unprecedented good clinical results (272), leading to this 2018's award of the Nobel Prize in Medicine for James Allison and Tasuku Honjo for their work in this area.

However, after the initial excitement and even after the inclusion of other checkpoint inhibitors among the available armamentarium against cancer, it is necessary to appreciate that not all patients will respond to this approach, since it needs an existing response, kept in check and "waiting" to be released by the treatment. On the other hand, it is quite possible that the frequently unsatisfactory response to cancer vaccines is caused by the pre-existence or vaccine-induced activation of these same regulatory circuits. Hence, a coherent path to achieve better clinical results would be the combination of both immune modulating strategies. Indeed, experimental (273) and clinical data (274) suggest that this may be true. In the aforementioned clinical study, patients with advanced melanoma were treated with a combination of MART-1-peptide pulsed-DC and anti-CTLA-4 and the results indicated that the combination might be, indeed, more effective than either approach alone. Likewise, also in the PD-1/PD-L1-PD-L2 pathway (275, 276) the combination of DC vaccination with checkpoint inhibition may offer, at least theoretical, advantages.

A different set of combination treatments has been targeting immune modulatory enzymes. The enzyme indoleamine 2,3 dioxygenase (IDO) catalyzes the degradation of tryptophan contributing to tolerance induction by favoring regulatory T cell differentiation and reducing DC activity (277). IDO expression by DC is induced by inflammatory stimuli (278), but also by CTLA-4 and PD-1 (279). Accordingly, IDO inhibition has shown positive effects in murine models of pancreatic cancer (280) and a study combining IDO inhibitors with DC vaccines for breast cancer patients has completed recruitment (NCT01042535). Similarly, an inhibitor of BCR-ABL, SRC, c-KIT, PDGFR, and ephrin tyrosine kinases has shown synergistic effects with a DC vaccine in a mouse melanoma model (281) and this combination is the object of ongoing clinical trials in patients with melanoma (NCT01876212) and metastatic renal cells carcinoma (NCT02432846 phase II e NCT01582672 phase III). Arginase-1, an enzyme that regulates cell proliferation and is constitutively expressed by myeloid-derived suppressor cells (MDSC) (282) and cyclooxygenase-2 (COX2), are other two enzyme whose inhibition might have positive interactions with immunotherapeutic approaches, including those that exploit DC.

#### CONCLUDING REMARKS

Dendritic cells have a central role in the immune system homeostasis and are directly involved in defining the patterns of response the system develops when facing an antigenic challenge. Their normal function warrants protection against infections, possibly cancer, but also against autoimmunity

#### REFERENCES


and hypersensitivity reactions. The more is uncovered of the mechanisms that drive these cells to modulate the response in one way or another, the more tools will be available to direct the immune system to desired therapeutic outcomes. Today, much of the efforts and clinical results are focused into harnessing these cells to induce effector responses, mainly, but not only, in cancer. With the advancement of the understanding of their physiology and regulatory pathways, it is possible to predict their effective use in such opposing conditions as cancer and diabetes, with less untoward and more durable effects.

#### AUTHOR CONTRIBUTIONS

TP, MP, AO, and GE reviewed the literature and wrote the manuscript. PB-S revised the literature and the manuscript and JB designed, wrote and revised the manuscript.

### FUNDING

This work had the support of grants from the Sao Paulo Research Foundation-FAPESP (2014/25988-1; 2014/26437-9; 2105/03314- 1; 2016/01137-8), the Coordination for the Improvement of Higher Education Personnel-CAPES, and from National Council for Scientific and Technological Development-CNPq (308053/2017-6).

population is important for the development of autoimmunity. Blood (2011) 118:1305–15. doi: 10.1182/blood-2011-01-331462


of monocyte-derived macrophages. Immunity (2014) 41:465–77. doi: 10.1016/j.immuni.2014.08.006


programs of intestinal DC differentiation in humans and mice. Nat Immunol. (2013) 15:98–108. doi: 10.1038/ni.2768


of CD11c(+)MHCII(+) macrophages and dendritic cells. Immunity (2015) 42:1197–211. doi: 10.1016/j.immuni.2015.05.018


lymphoma regression: a phase I/II study. J Clin Oncol. (2010) 28:4324–32. doi: 10.1200/JCO.2010.28.9793


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Patente, Pinho, Oliveira, Evangelista, Bergami-Santos and Barbuto. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Generation of an Oncolytic Herpes Simplex Virus 1 Expressing Human MelanA

Jan B. Boscheinen<sup>1</sup> , Sabrina Thomann<sup>1</sup> , David M. Knipe<sup>2</sup> , Neal DeLuca<sup>3</sup> , Beatrice Schuler-Thurner <sup>4</sup> , Stefanie Gross <sup>4</sup> , Jan Dörrie<sup>4</sup> , Niels Schaft <sup>4</sup> , Christian Bach<sup>5</sup> , Anette Rohrhofer <sup>6</sup> , Melanie Werner-Klein<sup>7</sup> , Barbara Schmidt 1,6,8 \* and Philipp Schuster <sup>8</sup>

1 Institute of Clinical and Molecular Virology, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, <sup>2</sup> Department of Microbiology and Immunobiology, Harvard Medical School, Boston, MA, United States, <sup>3</sup> Department of Microbiology and Molecular Genetics, University of Pittsburgh School of Medicine, Pittsburgh, PA, United States, <sup>4</sup> Department of Dermatology, Universitätsklinikum Erlangen, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany, <sup>5</sup> Lab for Immunogenetics, Universitätsklinikum Erlangen, Erlangen, Germany, <sup>6</sup> Institute of Clinical Microbiology and Hygiene, University Hospital Regensburg, Regensburg, Germany, <sup>7</sup> Chair of Immunology, Regensburg Center for Interventional Immunology (RCI), University of Regensburg, Regensburg, Germany, <sup>8</sup> Institute of Medical Microbiology and Hygiene, University of Regensburg, Regensburg, Germany

#### Edited by: Silvia Beatriz Boscardin,

University of São Paulo, Brazil

#### Reviewed by:

Jurjen Tel, Eindhoven University of Technology, Netherlands Andrew Zloza, Rutgers Cancer Institute of New Jersey, United States Fabian Benencia, Ohio University, United States

> \*Correspondence: Barbara Schmidt barbara.schmidt@ukr.de

#### Specialty section:

This article was submitted to Vaccines and Molecular Therapeutics, a section of the journal Frontiers in Immunology

> Received: 24 August 2018 Accepted: 02 January 2019 Published: 22 January 2019

#### Citation:

Boscheinen JB, Thomann S, Knipe DM, DeLuca N, Schuler-Thurner B, Gross S, Dörrie J, Schaft N, Bach C, Rohrhofer A, Werner-Klein M, Schmidt B and Schuster P (2019) Generation of an Oncolytic Herpes Simplex Virus 1 Expressing Human MelanA. Front. Immunol. 10:2. doi: 10.3389/fimmu.2019.00002 Robust anti-tumor immunity requires innate as well as adaptive immune responses. We have shown that plasmacytoid dendritic cells develop killer cell-like activity in melanoma cell cocultures after exposure to the infectious but replication-deficient herpes simplex virus 1 (HSV-1) d106S. To combine this innate effect with an enhanced adaptive immune response, the gene encoding human MelanA/MART-1 was inserted into HSV-1 d106S via homologous recombination to increase direct expression of this tumor antigen. Infection of Vero cells using this recombinant virus confirmed MelanA expression by Western blotting, flow cytometry, and immunofluorescence. HSV-1 d106S-MelanA induced expression of the transgene in fibroblast and melanoma cell lines not naturally expressing MelanA. Infection of a melanoma cell line with CRISPR-Cas9-mediated knockout of MelanA confirmed de novo expression of the transgene in the viral context. Dependent on MelanA expression, infected fibroblast and melanoma cell lines induced degranulation of HLA-matched MelanA-specific CD8<sup>+</sup> T cells, followed by killing of infected cells. To study infection of immune cells, we exposed peripheral blood mononuclear cells and in vitro-differentiated macrophages to the parental HSV-1 d106S, resulting in expression of the transgene GFP in CD11c<sup>+</sup> cells and macrophages. These data provide evidence that the application of MelanA-encoding HSV-1 d106S could enhance adaptive immune responses and re-direct MelanA-specific CD8<sup>+</sup> T cells to tumor lesions, which have escaped adaptive immune responses via downregulation of their tumor antigen. Hence, HSV-1 d106S-MelanA harbors the potential to induce innate immune responses in conjunction with adaptive anti-tumor responses by CD8<sup>+</sup> T cells, which should be evaluated in further studies.

Keywords: vaccine, oncolytic viruses, malignant melanoma, herpes virus, cytotoxic T cell response

# INTRODUCTION

Based on the results of a large phase III trial (1), Talimogene laherparepvec (T-VEC) was recently approved as first oncolytic herpes virus 1 for the treatment of patients with stage IIIB, IIIC, or IVM1a malignant melanoma. This attenuated virus induces regression of injected or distant cutaneous, lymphatic, and visceral lesions (2). It preferentially replicates in tumor cells due to defects in the type I interferon pathway, which renders these cells more susceptible to virus replication (3). T-VEC encodes GM-CSF, which contributes to recruitment of antigenpresenting cells to the site of injection. Via lysis of melanoma cells and uptake into antigen-presenting cells, T-VEC enhances cross-presentation of tumor-associated antigens to T cells, which, in particular in combination with immune checkpoint inhibitors, induces strong anti-tumoral responses leading to significantly improved survival of the patients (4, 5).

T-VEC is one of several oncolytic herpes simplex viruses, which have made their way to the clinic. Amongst them are G207, HSV 1716, NV1020, and HF10, which have been used in phase I/II trials in glioma, glioblastoma, melanoma, neuroblastoma, breast, and pancreatic cancer (6). All these viruses are attenuated, but replication-competent. In addition to inactivation of the neurovirulence gene γ34.5 and the TAP-binding protein ICP47, the third generation oncolytic herpes virus 147 has mutated the ribonucleotide reductase ICP6, resulting in a more pronounced attenuation of the virus (7).

We have investigated the oncolytic effects of the HSV-1 d106S strain, which, in contrast to other oncolytic herpes viruses, is infectious but replication-deficient due to deletions of essential viral genes (8). HSV-1 d106S expresses GFP, which can be replaced by other genes of interest via homologous recombination. Using eleven different melanoma cell lines, we have shown that HSV-1 d106S is oncolytic, in particular if combined with plasmacytoid dendritic cells (PDC) (9). These cells are major producers of type I interferons (IFN) in the blood upon stimulation with herpes simplex or influenza viruses (10, 11). They surround and occasionally infiltrate primary melanoma lesions and sentinel lymph nodes (12–14).

Due to an immunosuppressive tumor microenvironment, infiltrating PDC are usually immature and tolerogenic, promoting regulatory immunity (15) and tumor progression (16, 17). Upon activation by Toll-like receptor (TLR) agonists, PDC induce a Th1-type immune response and contribute to T cell-mediated tumor regression. In this respect, we have shown that PDC develop strong killer-cell like activity against melanoma cells upon exposure to HSV-1 d106S (9). This was similarly observed by others using Toll-like receptor 7 and 9 agonists as well as viral vaccines for stimulation of PDC (18–26).

So far, evidence is accumulating that oncolytic herpes viruses are potent inducers of innate immune responses. Beyond that, robust anti-tumor immunity requires adaptive immune responses. Two recent proof-of-concept trials showed induction of strong (mostly CD4+) T cell responses with subsequent delay in reappearance of new metastases in melanoma patients via injection of minigenes coding for different neoantigen-derived peptides (27) or via vaccination using respective peptides in the context of adjuvant (28).

With oncolytic viruses, induction of adaptive immunity is currently based on the uptake and cross-presentation of tumor-specific antigens released from dying tumor cells. To enhance antigen presentation, we envisaged to replace the transgene GFP in HSV-1 d106S by the melanoma-associated antigen MelanA/MART-1. We hypothesized that the expression of a tumor antigen in the context of the oncolytic HSV-1 d106S may provoke CD8<sup>+</sup> T cell responses against melanoma cells, combining oncolytic effects of the virus with enhanced expression of melanoma-associated antigens. Hence, such an oncolytic virus may target both innate and adaptive immune responses.

# MATERIALS AND METHODS

#### Cloning of MelanA Into Transfer Plasmid pd27B

The transfer plasmid pd27B containing sequences homologous to HSV-1 d106S (8) and a MelanA expression plasmid (29) were propagated in Escherichia coli XL-1 Blue cells (Agilent, Böblingen, Germany) and isolated using the PureLink HiPure Plasmid Midiprep kit (Invitrogen/Life Technologies, Darmstadt, Germany). The coding sequence for MelanA was amplified using NheI-MelanA (5′ -TAGATAGCTAGCATGCCAAGAGAA GATGCTC-3′ ) and MelanA-XbaI (5′ -GTCCATTCTAGATTA AGGTGAATAAGGTGGTG-3′ ) (biomers.net, Ulm, Germany). The PCR product and pd27B were digested using NheI and XbaI (NEB, Frankfurt, Germany), followed by dephosphorylation of pd27B. Both products were purified using the QIAquick PCR purification kit (Qiagen, Hilden, Germany), ligated at room temperature overnight, and transformed into XL-1 Blue cells. Correct inserts were identified using T7-EEV-Prom (5′ - AAGGCTAGAGTACTTAATACGA-3′ ; Promega, Mannheim, Germany) with primers 5′ -CCGATGAGCAGTAAGACTC-3′ ; 5 ′ -AGTTGTGGTTTGTCCAAACTC-3′ ; 5′ -TGGATAAAAGTC TTCATGTTGG-3′ .

# Cultivation of HSV-1 d106S and HSV-1 d106S-MelanA

HSV-1 d106S is an infectious recombinant strain derived from the HSV-1 d106 virus (30). It expresses GFP under the control of a CMV promoter, has a restored susceptibility to aciclovir, and is replication-deficient due to deletions of essential viral genes and promoter regions (8). Complementing E11 cells providing ICP4 and ICP27/47 in trans were propagated in DMEM supplemented with 10% heat-inactivated FCS (Sigma-Aldrich, Munich, Germany), 90 U/ml streptomycin, 0.3 mg/ml glutamine, 200 U/ml penicillin, and periodic G418 selection (400µg/ml). Infected at 90% confluency (MOI 0.1), cells were harvested at 50– 60 h when they showed cytopathic effects but were still adherent. After three freeze-thaw cycles, cells were resuspended in DPBS. Supernatants were filtered through 0.45µm pores and stored at

**Abbreviations:** HSV, Herpes simplex virus; IFN, interferon; ko, knockout; MOI, multiplicity of infection; PBMC, peripheral blood mononuclear cells; PCR, polymerase chain reaction; PDC, plasmacytoid dendritic cells; p.i., post infection; sg, single guide.

−80◦C. The number of infectious HSV-1 particles was quantified using the 50% tissue culture infective dose (TCID50) according to the method of Reed and Munch.

### Isolation of HSV-1 d106S DNA

Viral DNA was prepared from nucleocapsids following a published protocol (31). Supernatants of infected cell cultures were loaded onto discontinuous OptiPrep gradients (Sigma-Aldrich) of 40% iodixanol overlayed with 20% iodixanol, and subjected to ultracentrifugation using SW41Ti Ultraclear tubes (Beckman Coulter, Krefeld, Germany) at 30,000 rpm for 2 h, without braking at 800 rpm. The visible whitish ring containing viral particles was harvested by side-puncture, transferred to a VTi65 ultraclear tube (Beckman Coulter), and filled with 30% iodixanol, forming a continuous gradient during ultracentrifugation at 55,000 rpm for 6 h. The visible ring was transferred to a SW41Ti ultraclear tube, filled with DPBS, and pelleted at 20,000 rpm for 90 min. Pellets were re-suspended in DPBS, filtered through 0.45µm pores, and digested using 10× Taq DNA polymerase buffer containing proteinase K (Sigma-Aldrich) and 0.1% (v/v) Tween 20 at 56◦C for 1 h. For all subsequent steps, shearing of viral DNA was minimized by cutting off pipet tips and gentle mixing of solutions. The digested pellet was transferred to a phase lock "light" gel tube (5 Prime, Hilden, Germany) and mixed with phenol-chloroformisoamylalcohol. After centrifugation at 2,000 rpm for 5 min, the aqueous phase was transferred to another phase lock gel tube, recapitulating the step described above. Traces of phenol were eliminated by chloroform extraction. The aqueous solution was precipitated with 7.5 M ammonium acetate and ice-cold ethanol at −80◦C overnight. DNA was pelleted at 15,000 rpm at 4◦C for 45 min, washed with 70% ethanol, and resuspended in TEbuffer. Purified DNA was not frozen to avoid double-strand breaks. Purity and integrity was checked using NanoDrop UVspectrophotometry and EcoRI-HF digestion (NEB).

#### Homologous Recombination

E11 cells were seeded into 6-well plates to obtain 90% confluency for transfection. pd27B-MelanA was linearized using SwaI (NEB), purified using the QiaQuick PCR purification kit, and mixed with HSV-1 d106S DNA at a ratio of 1:4 (w/w) in DMEM plus glutamine. The mixture was heated at 95◦C for 3 min, chilled on ice, and mixed with FuGENE HD transfection reagent (Roche, Mannheim, Germany). After incubation at room temperature for 30 min, the mixture was added to E11 cells. Cytopathic effects were identified after 2 days. Non-fluorescent viral plaques were purified using limiting dilution and further analyzed for evidence of homologous recombination.

#### Isolation and Cultivation of Cells

PBMC were isolated from EDTA-anticoagulated blood of healthy donors using standard Biocoll density gradient centrifugation (Biochrom AG, Berlin, Germany), as approved by the Ethical Committee of the Medical Faculty, Friedrich-Alexander-Universität Erlangen-Nürnberg (Ref. no. 3299). PBMC were cultivated in RPMI 1640 with supplements described above. For generation of macrophages, PBMC were seeded into Nunc Lab-Tek chamber slides (Thermo Fisher Scientific) and cultivated in the presence of 15% heat-inactivated autologous serum, removing non-adherent cells by trypsin after 3 days. At 10–14 days, macrophages were infected with wild type HSV-1 (32), HSV-1 166v (33), HSV-1 d106S, and HSV-1 d106S-MelanA. MRC-5 fibroblasts (ATCC <sup>R</sup> CCL-171TM) and melanoma cell lines (IGR-37, IGR-39, ARST-1, ICNI-5li, SK-MEL30, LIWE-7) were cultivated as described (9).

### CRISPR-Cas9 Knockout

The MelanA gene was knocked out from SK-MEL30 cells using CRISPR-Cas9 technology (Addgene, Cambridge, MA). Sequences of single guide (sg) RNAs were taken from the GECKO library (sgMelanA1: 5′ -GCACGGCCACTCTTACACCA-3′ ; sgMelanA2: 5′ -TTGAACTTACTCTTCAGCCG-3′ ) (34) and inserted into LentiCRISPRv2 puro (#52961) (35). Lentiviral stocks were produced from 293T cells transfected with plasmids LentiCRISPRv2 puro, psPAX2 (#12259), and pMD2.G (#12260).

#### HLA Typing

High resolution HLA-A, -B, and -C genotyping was performed using the HLA SBT S4 HLA class I kit (Protrans GmbH, Hockenheim, Germany) according to the manufacturer's instructions in full compliance with the HLA typing standards of the European Federation for Immunogenetics (EFI).

## MelanA-Specific T Cell Generation and Coculture

CD8<sup>+</sup> T cells were purified from PBMC of a HLA-A<sup>∗</sup> 02:01 positive donor using a CD8 cell isolation kit (Miltenyi Biotec, Bergisch-Gladbach, Germany) and stimulated using artificial antigen-presenting cells (36) loaded with MelanA/MART-1 27L26−<sup>34</sup> peptide (ELAGIGILTV, GenScript, distributed by Biozol, Eching, Germany). The coculture was carried out in M' medium (37) supplemented with 5% autologous plasma and 3% T cell growth factor (38), kindly provided by Mathias Oelke. On day 7 and weekly thereafter until week 4, T cells were restimulated. Purity was assessed using HLA-A ∗ 02:01/MART-127L26−<sup>35</sup> tetramers and found to be > 95%. Coculture with HLA-matched fibroblast and melanoma cell lines was carried out in the presence of Alexa 488-labeled CD107a (eBiosciences/ThermoFisher, Frankfurt, Germany) and Golgi blockers brefeldin A and monensin (Sigma-Aldrich/Merck; 1:1,000) for 4 h. Prior to coculture, melanoma and MRC5 cells were plated at 90% confluency, infected with the respective viruses (MOI 1) for 20 h or loaded with peptide for 1 h, washed, and subsequently overlaid with a total of 1.5 × 10<sup>5</sup> CD8<sup>+</sup> T cells in 96-well plates. After FcR blocking, cells were stained with fixable viability dye eFluor 506 (eBiosciences) and anti-CD8 APC/Cy7 (BioLegend, Koblenz, Germany), and, after permeabilization using the BD Cytofix/CytopermTM Kit, with anti-IFN-gamma PE-Cy7 or the respective isotype (eBiosciences).

#### FACS Analysis

PBMC were exposed to HSV-1 d106S and HSV-1 d106S-MelanA for 24 h, washed, and incubated with FcR blocking reagent at 4◦C for 10 min. Cell populations were stained at 4◦C for 20 min, using a published protocol (39) with antibodies to CD3 (Alexa Fluor700; clone UCHT1; BioLegend, London, UK), CD4 (PE-Cy7; clone RPA-T4; BioLegend), CD11c (PE-Cy5; clone B-ly6, BD Biosciences, Heidelberg, Germany), CD14 (APC; clone HCD14; BioLegend), CD19 (APC-H7; clone SJ25C1; BD Biosciences), CD56 (BV-605; clone NCAM16.2; BD Biosciences), and CD304 (PE; clone AD5-17F6; Miltenyi Biotec). Dead cells were stained using PacificBlue (Invitrogen/Life Technologies). Cells were collected using multiparameter LSR-II flow cytometer with FACSDiva software (BD Biosciences) and FCS Express 3 Software (De Novo Software, Los Angeles, CA, USA). MelanA expression was studied in infected cells, using the BD Cytofix/Cytoperm kit with FITC-conjugated (clone A103, Santa Cruz, Heidelberg, Germany) or unconjugated murine anti-MelanA (clone A103, Dako, Hamburg, Germany) and the respective isotype controls followed by Alexa Fluor 555-conjugated goat anti-mouse F(ab′ )2 (Invitrogen/Life Technologies).

#### Western Blot Analysis

Cells were lysed on ice for 30 min (50 mM TRIS, pH 8.0; 150 mM NaCl; 5 mM EDTA; 1% NP-40; 0.1 mM PMSF), heated in sample buffer containing SDS and ß-mercaptoethanol at 105◦C for 10 min, separated on a 10% polyacrylamide gel, and transferred to a PVDF membrane (Merck Millipore, Darmstadt, Germany). After blocking with 5% milk powder plus 0.4% Tween 20, samples were incubated with unconjugated MelanA antibody (1:750) at room temperature for 1.5 h or at 4◦C overnight, followed by HRP-conjugated rabbit polyclonal anti-mouse IgG (H+L) (DAKO Diagnostics GmbH, Hamburg; 1:1,000) at room temperature for 60 min. After adding ECL solution containing luminol (Sigma-Aldrich) for 1 min, luminescence was recorded using the Fujifilm LAS-1000 plus gel documentation system.

#### Immunofluorescence and Confocal Microscopy

E11/Vero cells and macrophages were infected in chamber slides using HSV-1 d106S and HSV-1 d106S-MelanA (MOI 10) for 16 h, and incubated in DPBS plus 0.3% Triton-X100 at 4◦C for 20 min. After blocking in DPBS with 1% BSA (NEB) and 5% FCS, cells were stained with unconjugated anti-MelanA (diluted 1:75 in DPBS plus 1% BSA) at room temperature for 1 h and Alexa Fluor 555-conjugated goat anti-mouse F(ab')<sup>2</sup> (Invitrogen/Life Technologies, diluted 1:500 in DPBS plus 1% BSA) for 30 min. Slides were washed with DPBS containing DAPI and covered with VectaShield (Vector Laboratories, distributed by Biozol). In some experiments, cell membranes were stained using Alexa Fluor 555-labeled wheat germ agglutinin (5µg/ml) (Life Technologies). Cells were analyzed using the DMI 6000B inverted microscope and the TCS SP5 laser scanning microscope equipped with the LAS-AF software (Leica Microsystems, Mannheim, Germany).

#### Cell Killing Assays

To investigate direct oncolytic effects of HSV-1 d106S and HSV-1 d106S-MelanA, 1 × 10<sup>4</sup> melanoma cells were infected with these viruses using different MOI (1 and 10). Cell viability was checked at day 1, 2, 3, and 4 p.i. using the MTT lysis assay according to the manufacturer's recommendations (Trevigen, R&D Systems, Nordenstadt, Germany). Oncolytic effects of MelanA-specific CD8<sup>+</sup> T-cells were studied in melanoma cells, which had been infected with HSV-1 d106S and HSV-1 d106S-MelanA (MOI 1) for 8 h, followed by CD8<sup>+</sup> T-cell coculture (ratio 1:8) for 16 h and subsequent MTT lysis assay. Peptide-loaded cells served as controls.

#### Statistics

In our statistical analysis, we used one-way ANOVA for multiple group comparisons with GraphPad Prism version 8. Two-sided p < 0.05 were considered significant.

# RESULTS

# Generation of HSV-1 d106S-MelanA

The infectious, but replication-deficient HSV-1 d106S expresses GFP under the control of a CMV promoter. To replace this transgene by MelanA, the coding sequence of MelanA was amplified from expression plasmid pcDNA3(+) MART-1 (29) (**Figure 1A**) and cloned into transfer plasmid pd27B (8). The clone used for homologous recombination was sequenced, revealing identity with the MelanA sequence in GenBank (accession no. NM\_005511). After infection of E11 cells with HSV-1 d106S (**Figure 1B**), viral DNA was isolated from nucleocapsids. Purity and integrity of DNA were confirmed by spectrometry and digestion with EcoRI, which revealed distinct bands (**Figure 1C**). Cotransfection of SwaI-linearized pd27B-MelanA with HSV-1 d106S DNA into E11 cells resulted in mostly GFP-expressing (**Figure 1D**, upper part) and a few non-fluorescent viral plaques (**Figure 1D**, lower part), which indicated homologous recombination with replacement of GFP in a minority of transfected cells.

#### Characterization of HSV-1 d106S-MelanA

Two non-fluorescent viral plaques were purified via limiting dilution and used to infect E11 cells. The coding sequence of MelanA was detected in cells infected with both clones, but not in cells exposed to HSV-1 d106S, while all three infections were positive for the housekeeping ß-glucuronidase gene (**Figure 2A**). Western blotting detected MelanA protein in E11 cells infected with the two non-fluorescing clones, but not with HSV-1 d106S, while ß-actin was present in all three infections (**Figure 2B**). In flow cytometry, E11 cells infected with HSV-1 d106S expressed GFP, while cells infected with the two putative MelanA-expressing clones showed red fluorescence after intracellular staining of MelanA (**Figure 2C**). Vero cells, which do not support productive HSV-1 d106S replication, also expressed GFP or MelanA upon infection (**Figure 2D**) with mostly nuclear and cytoplasmic expression of GFP and MelanA, respectively, as evident from confocal microscopy. Altogether, we obtained a recombined HSV-1 d106S strain expressing MelanA, further termed HSV-1 d106S-MelanA.

<sup>(40</sup> × magnification, 2.5 × zoom), respectively. Scale bars represent 50 and 10µm in immunofluorescence and confocal images, respectively.

#### Expression of MelanA in Human Fibroblast and Melanoma Cell Lines

Melanoma cells frequently express MelanA, which may be lost upon immune escape (40). Three of our melanoma cells lines expressed MelanA (IGR-37, ARST-1, SK-MEL30), while three others were negative (LIWE-7, IGR-39, ICNI-5li). We analyzed whether MelanA expression may be restored in the latter upon infection with HSV-1 d106S-MelanA (MOI 1). At 20h p.i., IGR-37 and ARST-1 cells still expressed MelanA (**Figure 3A**), while transgene expression was induced in LIWE-7, IGR-39, and ICNI-5li cells **(Figure 3B)**. Similarly, MelanA expression was induced in MRC-5 fibroblasts. MelanA protein expression was confirmed in IGR-37 and LIWE-7 cells using Western blotting (**Figure 3C**). These data indicated that melanoma cell lines which did not express MelanA per se could be induced to do so.

To exclude upregulation of endogenous MelanA in these cell lines, we used two CRISPR-Cas9 approaches targeting different regions of the MelanA gene (sgMelanA1, sgMelanA2) to knock out this gene in SK-MEL30 cells. Four weeks after lentiviral

transduction, MelanA was no longer detectable in 85% of cells. After single-cell sorting, two and five MelanA-negative cell clones were obtained for sgMelanA1 and sgMelanA2, respectively. Upon infection with HSV-1 d106S-MelanA, MelanA-negative cell clones (sgMelanA1-clone4, sgMelanA2-clone4) started to re-express MelanA, while MelanA expression was marginally downregulated in parental SK-MEL30 cells and a cell clone with ineffective MelanA knockout (sgMelanA1-clone1) (**Figure 3D**). These data indicated de novo expression of the transgene in the viral context.

# Presentation of MelanA in Human Fibroblast and Melanoma Cell Lines

In further experiments, we investigated whether expression of MelanA in infected cell lines was followed by presentation of MelanA peptides within the HLA-A context. To this end, we cocultured HLA-A<sup>∗</sup> 02:01-positive fibroblast (MRC-5) and melanoma (SK-MEL30) cell lines with HLA-A<sup>∗</sup> 02:01/MART-1 27L26−34-specific CD8<sup>+</sup> T cells. As expected, MelanA-expressing SK-MEL30 cells induced CD8<sup>+</sup> T cell activation after 4 h of coculture, as evident from degranulation (CD107a) (**Figure 4A)** and IFN-gamma (**Figure 4B**) production, while MelanA-negative MRC-5 cells failed to do so. Similar results were obtained after infection of cell lines using HSV-1 d106S, confirming that virus infection per se did not induce CD8<sup>+</sup> T cell activation. Upon infection of MRC-5 cells with HSV-1 d106S-MelanA, however, CD8<sup>+</sup> T cells showed enhanced surface exposure of CD107a (**Figure 4A**) and, at least to some extent, IFN-gamma production (**Figure 4B**). These results indicated processing of virus-encoded MelanA with presentation of the respective peptide in the context of HLA-A in these cells. As a positive control for CD8<sup>+</sup> T cell activation, MRC-5 and SK-MEL30 cells were exogenously loaded with saturating concentrations of the optimized MelanA/MART-1 27L26−<sup>34</sup> peptide.

To corroborate activation of CD8<sup>+</sup> T cells by virusencoded MelanA in melanoma cells, we investigated SK-MEL30 knockout cells. A MelanA-negative cell clone obtained using sgMelanA1 (sgMelanA1-clone4) did not activate HLA-A ∗ 02:01/MART-127L26−34-specific CD8<sup>+</sup> T cells, while HSV-1

fibroblast cell lines for 4 h. In contrast to SK-MEL30 cells, MelanA was not expressed by MRC-5 and SK-MEL30 knockout (ko) cells (sgMelanA1-clone4). Cell lines were infected with HSV-1 d106S or HSV-1 d106S-MelanA for 20 h prior to coculture or loaded with MelanA peptide MART-127L26−34. T cells were identified as viable CD8<sup>+</sup> cells after exclusion of doublets. (A) One representative experiment and (B) mean and standard error of four, five, and (C) three separate experiments for SK-MEL30, MRC-5, and SK-MEL30 ko cells, respectively. Percentages of CD107a- and IFNg-expressing cells were compared to mock using one-way ANOVA for multiple group comparisons; \*p < 0.05.

d106S-MelanA infection of this cell clone induced degranulation as evident from the detection of CD107a at the surface of CD8<sup>+</sup> T cells (**Figure 4A**). A similar increase in CD8<sup>+</sup> T cell degranulation was observed after infection of another MelanAnegative SK-MEL30 clone obtained using sgMelanA2 (data not shown). In independent experiments, significant CD8<sup>+</sup> T cell degranulation was induced by HSV-1 d106S-MelanA-infected MRC-5 compared to uninfected cells (0.2% vs. 3.2%, p = 0.03) (**Figure 4C**). A similar trend was observed in SK-MEL30 knockout cells (1.1% vs. 4.9%, p = 0.06). Altogether, fibroblast and melanoma cells were induced to express tumor antigen and present respective peptides to tumor antigen-specific HLAmatched CD8<sup>+</sup> T cells.

# Direct and CD8<sup>+</sup> T Cell-Mediated Oncolytic Effects of HSV-1 d106S-MelanA

To investigate direct effects of HSV-1 d106S and HSV-1 d106S-MelanA on tumor cell killing, SK-MEL30 wild type cells were infected using two different MOI. Oncolytic effects of both viruses on SK-MEL30 cells were comparable. Infection using a MOI of 10 resulted in a significantly stronger reduction of viability than infection using a MOI of 1 (p < 0.001 for d106S and p < 0.01 for d106S-MelanA at day 2 p.i.) (**Figure 5A**). MRC-5 cells were significantly less susceptible to this oncolytic effect (MOI 10) compared to SK-MEL30 cells at day 1 and 2 p.i. (p < 0.05).

In further experiments, we studied whether infection of MelanA-negative melanoma cells using HSV-1 d106S-MelanA

would contribute to the oncolytic effects of MelanA-specific CD8<sup>+</sup> T cells. SK-MEL30 wild type cells as well as sgMelanA1 clone1, which both expressed MelanA, were readily attacked, while MelanA-negative SK-MEL30 (sgMelanA1-clone 4) and MRC-5 cells remained mostly unaffected (**Figure 5B**). Exposure of all cell lines to MelanA peptide significantly increased cell death in comparison to untreated cells (p < 0.05). Notably, infection with HSV-1 d106S-MelanA significantly induced T cell-mediated killing of the MelanA-negative SK-MEL30 (sgMelanA1-clone4) and MRC-5 cells, and even enhanced lysis of the MelanA-expressing SK-MEL30 cell line (p < 0.05), whereas infection using HSV-1 d106S showed no effect. In sum, HSV-1 d106S-MelanA proved to be oncolytic via two effects: direct oncolysis of melanoma cells and induction of enhanced oncolytic activity by MelanA-specific CD8<sup>+</sup> T cells.

# Expression of the Transgene GFP in Human PBMC and Antigen-Presenting Cells

We have shown that fibroblast and melanoma cell lines can be induced to express MelanA upon HSV-1 d106S-MelanA infection. To find out whether the transgene can also be expressed in antigen-presenting cells, we studied the infection of PBMC obtained from healthy volunteers. Because GFP is more readily detected compared to MelanA, we used HSV-1 d106S and focused on monocytes, which can differentiate into antigen-presenting cells. However, CD14 was downregulated at 24 h p.i., as reported previously (41), which precluded proper identification of monocytes. Therefore, antigen-presenting cells including monocytes were labeled using CD11c, which remained expressed at the cell surface. Upon infection with wild type HSV-1, PBMC did not display green fluorescence, while GFP was detected in a proportion of cells exposed to HSV-1 v166 (33) **(Figure 6A)**. This virus codes for a VP22-GFP fusion protein, which is not only expressed but also secreted from infected cells. Therefore, cells with attached fluorescing viruses or VP22 cannot be discriminated from truly infected cells. In contrast, HSV-1 d106S expresses GFP under the control of a CMV promotor in infected cells only, and GFP is not incorporated into viral particles. This virus induced GFP expression in CD11c<sup>+</sup> PBMC, becoming more prominent with increasing MOI (**Figure 6A**). Individual cell populations were identified as T cells (CD3+), B cells (CD3<sup>−</sup> CD19+), NK cells (CD3<sup>−</sup> CD19<sup>−</sup> CD56+), and CD11c<sup>+</sup> cells (CD3<sup>−</sup> CD19<sup>−</sup> CD11c+) using a multicolor flow cytometry panel **(Supplementary Figure 1)**. Again, green

fluorescence was not detected upon infection with wild type HSV-1, but upon infection with HSV-1 v166 (CD11c<sup>+</sup> cells > CD56<sup>+</sup> NK cells = CD19<sup>+</sup> B cells > CD3<sup>+</sup> T cells) **(Figure 6B)**. With HSV-1 d106S, GFP was detected in CD11c<sup>+</sup> cells only, resulting in 22.1, 68.3, and 79.2% of cells infected at MOI of 1, 10, and 100, respectively **(Figure 6B)**.

In addition to primary CD11c<sup>+</sup> cells, we studied macrophages generated from PBMC of HSV-seronegative donors in the presence of autologous serum. Adherent cells were differentiated into macrophages, which, upon exposure to HSV-1 d106S, expressed GFP in confocal imaging (**Figure 6C**) and flow cytometry (**Figure 6D**) comparable to the extent observed in fibroblast and melanoma cell lines. Overall, these data indicated expression of virus-encoded GFP in CD11c<sup>+</sup> cells and in macrophages. We further sought to confirm expression of the transgene upon HSV-1 d106S-MelanA infection. However, we were not able to verify MelanA expression in PBMC and macrophages in any of the experimental settings (data not shown).

#### DISCUSSION

Oncolytic viruses infect and replicate in tumor tissues. Subsequent lysis of infected cells releases tumor-specific antigens, which are taken up by antigen-presenting cells and induce anti-tumor immune responses via cross-presentation to T cells (42). We sought to optimize the induction of adaptive immune responses by incorporation of a tumor antigen into the viral genome. For this purpose, we used the infectious but replication-deficient HSV-1 d106S, which exerts oncolytic activity in particular in combination with PDC (9), and replaced the transgene GFP by MelanA via homologous recombination. Using flow cytometry, Western blotting, and immunofluorescence, protein expression was confirmed in complementing E11 and Vero cells. Upon HSV-1 d106S-MelanA infection, we detected transgene expression in MelanA-negative fibroblast and melanoma cells, and in SK-MEL30 cells with specific knockout of the MelanA gene using CRISPR-Cas9 technology. These data confirmed de novo expression of MelanA in the viral context.

Subsequent coculture of infected melanoma and fibroblast cell lines with HLA-matched MelanA-specific CD8<sup>+</sup> T cells verified MelanA-specific activation, as evident from CD8<sup>+</sup> T cell degranulation upon induced MelanA expression. The infection of parental MelanA-expressing SK-MEL30 cells induced a slightly reduced degranulation of CD8<sup>+</sup> T cells, most likely due to the oncolytic activity of the virus on target melanoma cells. Notably, we observed an increase after HSV-1 d106S-MelanA infection of MelanA-negative cells. It has to be admitted, though, that the degree of IFN-gamma secretion in CD8<sup>+</sup> T cells was very low. This was not due to a functional limitation of CD8<sup>+</sup> T cells, as evident from the control using an optimized MelanA peptide. It may rather be the result of the limited MelanA expression induced upon HSV-1 d106S-MelanA infection, which was not significantly enhanced using a higher MOI (data not shown). The reason may be the efficient oncolytic activity of this replication-deficient virus, resulting in depletion of MelanAexpressing target cells (**Figure 5A**). Importantly, we observed enhanced CD8<sup>+</sup> T cell-mediated killing of MelanA-negative melanoma cells upon infection with HSV-1 d106S-MelanA, but not upon infection with HSV-1 d106S (**Figure 5B**). These data indicate that HSV-1 d106S-MelanA exerts direct and indirect oncolytic effects.

Altogether, our data confirmed that HSV-1 d106S-MelanA could re-express MelanA in melanoma cells which have escaped immune recognition via loss of tumor antigen expression. Loss of MelanA expression may be more frequent than previously thought, occurring in three of our 11 melanoma cell lines (**Figure 3** and data not shown). As a consequence, MelanAspecific CD8<sup>+</sup> T cells may be re-directed to infected tumor lesions, which will become re-accessible to this adaptive CD8<sup>+</sup> T cell response. In this case, an efficient oncolysis will be mediated by HSV-1 d106S-MelanA as well as by innate and adaptive immune cells. The apoptotic and necrotic tumor cells will serve as source for new tumor-associated antigens, which have evolved during tumor progression. In this respect, it has previously been shown that apoptotic cells, which were killed by infection with replication-deficient HSV, served as vaccines by pulsing DCs (43). Apoptotic debris will be phagocytosed by dendritic cells and cross-presented to T cells. In this respect, CD11c<sup>+</sup> cells and macrophages may also play an important role.

Our tumor vaccine may profit from incorporating other tumor antigens which are targets of cytotoxic CD8<sup>+</sup> T cells, like the MAGE-A family, tyrosinase, NY-ESO1, gp100 or neoantigens (44). With the incorporation of MelanA into HSV-1 d106S, however, viral stocks harbored slightly less infectious virions compared to the parental strain. The generation of infectious stocks may become increasingly challenging with the incorporation of additional tumor-associated antigens. The difficulty in inserting full-length sequences of tumor antigens may be overcome by introducing much smaller genomic information as minigenes coding for tumor antigenderived peptides (27). It may also be worth cloning the coding sequences of tumor antigens or peptides into T-VEC, which is fully replicative and thus more virulent than HSV-1 d106S-MelanA.

The minor virulence of the non-replicative HSV-1 d106S-MelanA in comparison to the replication-competent T-VEC strain may be advantageous for the infection of antigenpresenting cells: a reduced cytotoxicity may facilitate the presentation of tumor peptides in the context of HLA-ABC. For these purposes, we infected PBMC with HSV-1 d106S, showing GFP expression in CD11c<sup>+</sup> antigen-presenting cells, but not in other immune cells. Subsequently, we noticed expression of GFP in macrophages comparable to the extent of MelanA expression in infected MRC-5 cells. However, we were not able to detect MelanA expression in any of the immune cells investigated, which was unexpected because both transgenes are expressed from the same CMV promoter. So far, it is unclear whether expression of MelanA in antigen-presenting cells is too low to be detected reliably, or whether MelanA is proteasomally degraded and presented on HLA-ABC immediately after mRNA translation.

More importantly, HSV-1 d106S has been shown to induce CD8<sup>+</sup> T cell responses in vivo. To this end, studies in mice and monkeys (45–47) revealed that HSV-1 d106S can not only activate, but also induce virus-specific CD8<sup>+</sup> T cells. This de novo induction may be more difficult with tumor-associated antigens (with the exception of neoantigens), which, as autoantigens, need to overcome self-tolerance. De novo induction can occur via direct presentation of the tumor antigen synthesized in the cytosol or via indirect cross-presentation after endocytosis of the tumor antigen, export into the cytosol and proteasomal degradation, transport to the endoplasmic reticulum and loading on HLA-ABC. Whether the vaccine HSV-1 d106S-MelanA can induce expansion of MelanA-specific CD8<sup>+</sup> T cells, and if so, which of the two mechanisms hold true, needs to be evaluated in further studies. It would be particularly valuable to study these effects in vivo using suitable animal models. The immune stimulation following intratumoral injection of the oncolytic virus in vivo may enhance the CMV promotor activity and thus contribute to a more efficient transgene expression.

A further prospect of our research is the combination of oncolytic viruses with other anti-cancer approaches like checkpoint inhibitors, chemotherapy, targeted therapy, and radiation therapy (42, 48, 49). It may even be interesting to test oncolytic viruses in combination with tumor-specific peptides. These conditions may reduce the immune-inhibitory activities of tumors and help tumor antigen-specific CD8<sup>+</sup> T cells to gain access to the malignant lesion. In addition, a new generation of oncolytic herpes viruses has been designed, which is less virulent due to deletion of ICP6 in addition to inactivation of neurovirulence factor γ34.5 and antagonist of the host cell's transporter associated with antigen presentation, ICP47. This oncolytic herpes virus allows a broader applicability and is currently being tested in glioblastoma and prostate cancer patients (7).

For these reasons, our approach to develop an oncolytic herpes virus which augments antitumor responses by coding for a tumor antigen appears to be promising for further combination immunotherapies against malignant melanoma. It may also be promising for other tumors. This may be true in particular for tumors which are known to be infiltrated by PDC, like head and neck squamous cell carcinoma (50), and ovarian (51, 52) and breast cancer (53, 54). A tumor antigen-expressing HSV-1 d106S may target both PDC and myeloid dendritic cells, which cooperate in inducing effective anti-tumor T cell responses (55).

#### AUTHOR CONTRIBUTIONS

JB, MW-K, DK, BS, and PS conceived the experiments. JB, ST, AR, MW-K, BS, and PS performed the experiments, analyzed and interpreted the data. DK and ND provided the parental HSV-1 d106S virus, transfer plasmid, and complementing cell line. BS-T and SG provided melanoma cell lines. JD and NS contributed the MelanA expression plasmid. MW-K contributed MelanA-specific CD8<sup>+</sup> T cells to the project. CB performed HLA-typing of cell lines and donors, and BS and PS wrote the manuscript. All authors were involved in critically reading of the manuscript and approved the final version of the manuscript.

#### REFERENCES


#### FUNDING

This study was supported by graduate colleges 1071 (Viruses of the immune system; to JB) 1660 (Key signals of adaptive immune response; to ST), and the German Research Foundation (WE 4632/4-1 to MW-K), and NIH (grant AI 057552 to DK).

#### ACKNOWLEDGMENTS

The present work was performed by JB in fulfillment of the requirements for obtaining the degree Dr. med. We thank Bernhard Fleckenstein and Thomas Stamminger, Erlangen, and André Gessner, Regensburg, for support, and Valerie Bezler and Konstanze Kühn for excellent assistance. Christiane Heilingloh, Department of Dermatology, Universitätsklinikum Erlangen, and Frank Neipel and Michael Mach, Institute of Clinical and Molecular Virology, Erlangen, are acknowledged for help in establishing protocols. We thank Gillian Elliot and Peter O'Hare, Section of Virology, Faculty of Medicine, Imperial College London, London, UK, for their generous contribution of HSV-1 166v. Plasmids lentiCRISPRv2 puro, pMD2.G, and psPAX2 were kind gifts from Feng Zhang and Didier Trono.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.00002/full#supplementary-material

simplex virus type 1 and plasmacytoid dendritic cells against tumour cells. Immunology (2015) 146:327–38. doi: 10.1111/imm.12509


stage I/II melanoma. Cancer (2012) 118:2476–85. doi: 10.1002/cncr. 26511


**Conflict of Interest Statement:** DK is a co-inventor on patents held by Harvard University that includes claims on the use of replication-defective mutant viruses such as d106S vaccine vector and immunomodulatory agent.

The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Boscheinen, Thomann, Knipe, DeLuca, Schuler-Thurner, Gross, Dörrie, Schaft, Bach, Rohrhofer, Werner-Klein, Schmidt and Schuster. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Dendritic Cell Subsets and Effector Function in Idiopathic and Connective Tissue Disease-Associated Pulmonary Arterial Hypertension

Pulmonary arterial hypertension (PAH) is a cardiopulmonary disease characterized by an incurable condition of the pulmonary vasculature, leading to increased pulmonary

Denise van Uden, Karin Boomars and Mirjam Kool\*

Department of Pulmonary Medicine, Erasmus MC, Rotterdam, Netherlands

Edited by: Diana Dudziak, Universitätsklinikum Erlangen, Germany Reviewed by: Veronika Lukacs-Kornek, Saarland University, Germany Theresa T. Lu, Hospital for Special Surgery, United States \*Correspondence: Mirjam Kool m.kool@erasmusmc.nl vascular resistance, elevated pulmonary arterial pressure resulting in progressive right ventricular failure and ultimately death. PAH has different underlying causes. In approximately 30–40% of the patients no underlying risk factor or cause can be found, so-called idiopathic PAH (IPAH). Patients with an autoimmune connective tissue disease (CTD) can develop PAH [CTD-associated PAH (CTD-PAH)], suggesting a prominent role of immune cell activation in PAH pathophysiology. This is further supported by the presence of tertiary lymphoid organs (TLOs) near pulmonary blood vessels in IPAH and CTD-PAH. TLOs consist of myeloid cells, like monocytes and dendritic cells (DCs), T-cells, and B-cells. Next to their T-cell activating function, DCs are crucial for the preservation of TLOs. Multiple DC subsets can be found in steady state, such as conventional DCs (cDCs), including type 1 cDCs (cDC1s), and type 2 cDCs (cDC2s), AXL+Siglec6<sup>+</sup> DCs (AS-DCs), and plasmacytoid DCs (pDCs). Under inflammatory conditions monocytes can differentiate into monocyte-derived-DCs (mo-DCs). DC subset distribution and activation status play an important role in the pathobiology of autoimmune diseases and most

Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 28 August 2018 Accepted: 04 January 2019 Published: 22 January 2019

#### Citation:

van Uden D, Boomars K and Kool M (2019) Dendritic Cell Subsets and Effector Function in Idiopathic and Connective Tissue Disease-Associated Pulmonary Arterial Hypertension. Front. Immunol. 10:11. doi: 10.3389/fimmu.2019.00011 in DC function in IPAH and CTD-PAH to gain a better understanding of PAH pathology. Keywords: dendritic cell, dendritic cell subsets, pulmonary arterial hypertension, idiopathic pulmonary arterial hypertension, autoimmune disease, dendritic cell effector function, connective tissue disease

likely in the development of IPAH and CTD-PAH. DCs can contribute to pathology by activating T-cells (production of pro-inflammatory cytokines) and B-cells (pathogenic antibody secretion). In this review we therefore describe the latest knowledge about DC subset distribution, activation status, and effector functions, and polymorphisms involved

# INTRODUCTION PULMONARY ARTERIAL HYPERTENSION

Pulmonary arterial hypertension (PAH) is characterized by a mean pulmonary arterial pressure (PAP) of ≥25 mmHg at rest and a mean capillary wedge pressure of ≤15 mmHg (1). The high PAP causes hypertrophy of the right ventricle (RV) leading eventually to RV dilatation, heart failure, and ultimately death. Particularly small pulmonary arteries (PAs) and arterioles are affected. They show a thickened vascular wall and formation of plexiform lesions due to endothelial dysfunction and proliferation of all three cell layers, the endothelium, smooth muscle cells (SMC), and the adventitia (2).

PAH patients can be subdivided into groups based on associated conditions and risk factors. However, in a substantial proportion of PAH patients no cause or associated condition can be identified: idiopathic PAH (IPAH). In another subgroup of patients, PAH is associated with autoimmune diseases (AD) such as connective tissue disease (CTD). CTD includes systemic sclerosis (SSc), systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), and mixed connective tissue disease (MCTD). SSc is the most common AD associated with PAH, followed by SLE (3–6). PAH patients have a low 1-year survival rate: only 82% of SSc-PAH patients and 93% of IPAH patients are still alive after 1 year (6).

#### ROLE FOR IMMUNE ACTIVATION IN THE DEVELOPMENT OF PAH

The presence of PAH in a proportion of autoimmune patients suggests that activated immune cells (or their mediators) directly provoke pulmonary vascular remodeling. Local immune activation is also observed as tertiary lymphoid organs (TLOs or ectopic lymphoid structures) are present in the lungs of IPAH and CTD-associated PAH (CTD-PAH) patients (7, 8). TLOs are organized structures similar to lymph nodes (LNs), including distinct T-cell areas containing dendritic cells (DCs), organized B-cell follicles with germinal centers (GCs), high endothelial venules (HEV), and lymphatics. TLOs most likely develop due to long-lasting local immune activation and are considered a hallmark of chronic disease (9). In lungs of IPAH patients, TLOs are found in the vicinity of PAs, suggesting that they promote vascular remodeling (7). Not surprisingly, as TLOs are characteristic for ongoing/chronic immune activation, they are often found in target organs of several ADs. For instance, in SLE patients TLOs are present in the kidneys, and in SSc-PAH patients TLOs have even been found in the lungs (8, 10, 11). Even though the SSc-PAH patient group used in this study is small, it is conceivable that TLOs are present in the lungs of various CTD-PAH patients. In addition, it is very likely that immune activation in PAH patients will also occur in draining LNs.

During chronic antigenic stimulation, the lymphotoxin (LT)α1β2-LTβ receptor axes is crucial for development of TLOs (12), whereby lymphoid tissue inducer (LTi) cells interact with lymphoid tissue organizer (LTo) cells. Repeated DC injection in the lungs of mice, mimicking chronic activation, provokes TLO development (13). Activated DCs can produce chemokines which attract T-cells and B-cells (e.g., CCL19/21 and CXCL13, respectively), as well as T- and B-cell survival factors (e.g., interleukin (IL)-15 and BAFF/IL-6, respectively) (13–17). They furthermore secrete cytokines creating a pro-inflammatory milieu and promote innate and adaptive responses. This milieu can also induce post-translational modifications of proteins, altering self-antigens into new antigens which could provoke autoimmune responses as seen in SLE (18). Within TLOs and LNs, tissue-migrated DCs present antigens to naïve T-cells, inducing their activation and differentiation. The main T helper (Th)-cell subsets are Th1, Th2, Th17, follicular Th-cells (Tfh), and regulatory T-cells (Tregs). Within the GC reaction in TLOs and LNs, Tfh-cells provide help to B-cells by producing cytokines that induce class switching, survival, proliferation, and antibody production.

The role of DC subsets and their effector function in pathogenesis of IPAH, AD, and CTD-PAH will be discussed in this review and is shown in **Table 1**.

#### DENDRITIC CELLS IN IPAH, CTD-PAH, AND AD

DCs are equipped with pathogen recognition receptors (PRRs) like toll-like receptors (TLR) to sense their surroundings. Antigen recognition leads to DC activation and migration toward LNs. Activated DCs upregulate co-stimulatory molecules like CD86, produce pro-inflammatory cytokines, and present antigen to Tcells using major histocompatibility complex class-II (MHC-II). In TLOs, DCs are mature, indicated by high CD86 expression and IL-12 production (37). The maintenance of TLOs in two lung infection models, has been shown to be dependent on DCs as they disintegrate when DCs are ablated (13, 38). Furthermore, impaired DC migration due to defects in the CCR7-signaling, has been shown to lead to the formation of bronchus-associated lymphoid tissue (39).

Under steady state conditions, several DC subsets with unique functions can be identified (40, 41). Conventional DCs (cDCs), identified by CD11c, and HLA-DR expression in humans, are a major DC subset and can be divided in two subtypes, type 1 cDCs (cDC1s) and type 2 cDCs (cDC2s). cDC1s express IRF8 and CD141 and excel in cross presentation (42). IRF4 and CD1c classify cDC2s, which are potent inducers of Th-cell responses. Plasmacytoid DCs (pDCs) produce interferons (IFN) and do not express CD11c, but express HLA-DR and CD123. Recently, within this HLA-DR+CD123<sup>+</sup> population potent Th-cell inducers have been found, which

**Abbreviations:** PAH, Pulmonary arterial hypertension; PAP, pulmonary arterial pressure; RV, right ventricle; PAs, pulmonary arteries; SMC, smooth muscle cell; IPAH, idiopathic PAH; CTD, connective tissue diseases; CTD-PAH, CTDassociated PAH; AD, autoimmune disease; SSc, systemic sclerosis; SLE, systemic lupus erythematosus; RA, rheumatoid arthritis; MCTD, mixed connective tissue disease; SSc-PAH, Systemic sclerosis-PAH; TLOs, tertiary lymphoid organs; LNs, lymph nodes; DCs, dendritic cells; GCs, germinal centers; HEV, high endothelial venules; PH, pulmonary hypertensions; LT, lymphotoxin; LTi, lymphoid tissue inducers; LTo, lymphoid tissue organizer; Th, T helper; IL, interleukin; Tfh, follicular Th-cells; Tregs, regulatory T-cells; PRRs, pathogen recognition receptors; TLR, toll-like receptor; MHC-II, major histocompatibility complex class II; cDCs, conventional DCs; cDC1s, type 1 cDCs; cDC2s, type 2 cDCs; pDCs, Plasmacytoid DCs; IFN, interferons; AS–DCs, AXL+Siglec6<sup>+</sup> DCs; mo-DCs, monocyte-derived dendritic cells; BM, bone marrow; IGS, interferon gene signature; PBMCs, peripheral blood mononuclear cells; LPS, lipopolysaccharide; ECs, endothelial cells; IPF, idiopathic pulmonary fibrosis; SSc-PF, pulmonary fibrosis associated SSc; NF-kB, nuclear factor-kappa B.



<sup>a</sup>Graves disease and Hashimoto's thyroiditis, cDC, conventional dendritic cell; pDC, plasmacytoid dendritic cell; mo-DC, monocyte-derived-dendritic-cell; PAH, pulmonary arterial hypertension; IPAH, idiopathic pulmonary arterial hypertension; AD, autoimmune disease; CTD-PAH, connective tissue disease-associated PAH; SLE, systemic lupus erythematosus; SSc, systemic sclerosis; TLO, tertiary lymphoid organ; PAs, pulmonary arteries; TLR, toll-like receptor.

additionally express AXL and Siglec6 (AXL+Siglec6<sup>+</sup> (AS)- DCs) (43, 44). Under inflammatory conditions monocytes can differentiate into DCs, giving rise to monocyte-derived-DCs (mo-DCs).

#### Conventional Dendritic Cells

In IPAH patients, the proportion of circulating cDCs is decreased compared to controls (19). Numbers of circulating cDCs are also altered in several ADs associated with PAH. Both cDC1s and cDC2s are decreased in proportion and number in SLE patients compared to HCs, especially in patients with active disease (20– 23). The decrease in circulating cDCs in PAH could indicate an increased cDC migration toward lung TLOs (**Figure 1**). In support of this idea, DCs can be found in lung TLOs of IPAH patients and cDC numbers were increased in total lung cell suspensions of these patients (7, 27). In IPAH TLOs, DCs are found inside T-cell zones, suggesting that they promote T-cell activation. In patients with ADs, cDCs in TLOs show increased expression of costimulatory molecules and a cDC2 phenotype, since they express CD1c and CD11c (28). Alternatively, the reduction in circulating cDCs might also be caused by alterations in cDC viability or DC progenitors resulting in a decreased output of cDCs from the bone marrow.

In addition to DC or DC precursors entering the affected tissue from the blood circulation, DCs may accumulate in tissue and contribute to TLO formation as they fail to go to LNs (39). Upon activation, DCs upregulate CCR7. The CCR7 allows the DC to respond to CCL19 and CCL21 expressed by the lymphatic endothelial cells and to enter the lymphatic vessels to migrate to the draining LN. Both CCL19 and CCL21 are expressed by lymphatic vessels in IPAH patients, which could facilitate DC attraction (7). Strikingly, CCR7-deficient mice develop lung TLOs and signs of PH, perhaps due to DC retention in the lungs (39, 45). DCs, amongst other cells, can produce CCL20 and CXCL13, which attract T-cells, B-cells, and immature DCs. CCL20 and CXCL13 mRNA expression are increased in IPAH lungs compared to controls (7), contributing to TLO formation. However, the cell responsible for this increased expression in IPAH is yet unknown.

Research into cDC subset activation is still limited in PAH and ADs. In SSc patients, circulating cDC2s produce more IL-6, IL-10, and TNF-α after TLR2 and TLR4 stimulation (24, 25). These

cytokines appear to play a central role in the immunopathology of PAH, as IL-6 and IL-10 are increased in the serum of IPAH patients and correlate with mortality (46). Especially IL-6 appears to be a crucial cytokine in PAH pathobiology, as mice overexpressing IL-6 develop signs of PH, while IL-6-deficient mice do not develop PH after hypoxia (47, 48). At this time, a phase II trial using Tocilizumab, an IL-6 receptor antagonist, is conducted in PAH patients (49).

In conclusion, in both IPAH and ADs circulating cDC proportions are decreased possibly due to migration to target organs, where they can both initiate adaptive immune responses and maintain TLOs (**Figure 2B**). Currently, only little is known about cDC subset distribution and function in IPAH, CTD-PAH, and ADs.

#### Plasmacytoid Dendritic Cells

Plasmacytoid DCs are predominantly found in lymphoid tissues and blood in steady state conditions. During inflammation, pDCs home toward peripheral tissues, produce type I IFNs, and promote activation of immune cells. In IPAH lungs pDC numbers are enhanced and pDCs are specifically located around the pulmonary vessels, while circulating pDC numbers are unaltered (27). In contrast, in SLE and SSc patients, circulating pDC number and frequency are decreased compared to controls, which could be due to emigration into diseased tissues (22, 23, 29, 31). Indeed, pDCs are present in diseased organs of SSc patients (29). Several ADs are associated with the interferon gene signature (IGS), to which different cells contribute. pDCs are major contributors to the IGS through their production of type I IFNs. One of the most strongly upregulated genes in pDCs within the IGS is CXCL10 (50). Augmented serum CXCL10 levels are associated with PAH in SSc patients (51). Likewise, in IPAH patients, serum CXCL10 is elevated and even associated with poor RV function (52), suggesting the possibility of a prominent role for pDCs in disease immunopathology. Next to IFNs, pDCs are also large producers of CXCL4 in SSc (30). CXCL4 can induce an influx of CD45<sup>+</sup> cells in target tissues, perhaps leading to tissue remodeling and disease progression.

The associations of pDC with CTD-PAH and the increase in pDCs in lungs of IPAH patients suggest that type-I IFN and chemokine secretion by pDCs not only play an important role in several ADs, but also in CTD-PAH and IPAH pathology (**Figure 2A**).

# Monocytes and Monocyte-Derived DCs

Monocytes are precursors of mo-DCs that arise under inflammatory conditions (40). Monocytes are heterogeneous and can be divided into 3 subsets based on CD14 and CD16 expression (53, 54). Classical monocytes, also called inflammatory monocytes, express CD14 and can infiltrate tissues, produce pro-inflammatory cytokines, and differentiate into inflammatory macrophages. Classical monocytes express several PRRs and are superior in phagocytosis. Monocytes expressing both CD14 and CD16 are termed intermediate monocytes, can also produce pro-inflammatory cytokines (55) and are unique in their ability to produce reactive oxygen species. Their gene expression signature indicates their ability to present antigens and induce T-cell activation (56). Intermediate monocytes specifically promote pro-inflammatory Th17-cell responses, which also contribute to PAH development, as discussed below (55). Finally, non-classical monocytes, expressing CD16, are known to survey the endothelium for danger signals (54). They differentiate into tissue-resident macrophages in steady state or into anti-inflammatory macrophages during inflammation, to repair damaged tissues.

The number of non-classical monocytes is increased in SSc associated with PAH development, whereas there is no difference in the number of classical monocytes (34). The number of CTD-PAH patients in this study was very small, so this should be confirmed in a larger cohort. Increased numbers of CD14<sup>+</sup> cells, including classical/intermediate monocytes and macrophages, are observed around PAs of IPAH patients (36). Monocytes might be attracted to the PAs through their expression of CCR2 and CCR5 and an increased expression of their ligands CCL2 and CCL5 in lungs and serum of IPAH patients (57, 58). In SSc and CTD-PAH enhanced CCL2 is also observed in either skin or serum (59–61).

Strikingly, circulating monocytes of IPAH patients are hyporesponsive, as demonstrated by decreased cytokine production upon TLR4 stimulation (32). The local and/or systemic pro-inflammatory milieu in IPAH patients could provoke a feedback mechanism, resulting in hyporesponsive monocytes. However, the underlying mechanism is still unknown and further research is needed. In contrast to IPAH monocytes, monocytes from SSc-PAH patients are activated, as shown by their mRNA expression profile. This profile is even discriminative between SSc-PAH and SSc patients (33). Non-classical monocytes, expressing CXCL10, CXCL8, and CCL4 are involved in SSc pathology, and are found in increased numbers in SSc patients compared to controls (24).

Mo-DCs for in vitro assays, used to model and monitor human DC function, are commonly generated from monocytes. Contradictory results have been found using this model in IPAH. Decreased activation of monocytes together with lower T-cell stimulation (19), as well as a similar activation status with an increased Th-cell stimulatory capability have been observed (35). These opposite findings might be caused by the type of stimulation used to mature mo-DCs and different mo-DC:T-cell ratios in the T-cell stimulation assays.

Taken together, increased pulmonary expression of chemokines may attract monocytes to lungs of IPAH and CTD-PAH patients, where they become activated and alter their gene expression due to the pro-inflammatory environment. These altered monocytes may give rise to mo-DCs, which arise at places of inflammation and can induce T-cell activation (**Figure 2C**).

#### EFFECTOR FUNCTION OF DCS IN IPAH, CTD-PAH AND ADS

#### T-Cell Responses

DCs excel at antigen presentation to T-cells and together with their costimulatory molecule expression and cytokine production, they are pivotal for the succeeding T-cell response. Specifically, Th17-cells are implicated in the pathogenesis of many ADs and are observed inside mature TLOs of IPAH patients (7). Th17 differentiation from naïve Th-cells occurs in the presence of IL-1β, IL-6, and TGFβ (62), cytokines produced by activated DCs. Both IL-1β and IL-6 are elevated in serum of IPAH patients (46). Th17-cells are the main source of IL-17, IL-21, and IL-22. IL-21<sup>+</sup> cells are present in remodeled PAs of IPAH patients (63). In addition, IL-17 may affect structural remodeling observed in PAH, as IL-17 enhances fibroblast proliferation and collagen production in vitro (64). In SSc, IL-17 induces adhesion molecule expression and IL-1/chemokine production on endothelial cells (ECs) (65–67). Additionally, in IPAH PBMCs the IL-17 gene is hypo-methylated, indicating increased IL-17 transcription and supporting a possible role for Th17-cells in the pathology of IPAH (35). Indeed, IL-17 gene expression is enhanced in lungs of both IPAH and SSc-PAH compared to idiopathic pulmonary fibrosis (IPF) and pulmonary fibrosis associated SSc (SSc-PF) (68), this IL-17 may be expressed by cells in TLOs as well as in tissues outside of TLOs.

Furthermore, IL-23, also produced by DCs, stabilizes the phenotype of Th17-cells, but also promotes their proinflammatory potential (62). Th17-cells are also highly plastic cells and under the influence of IL-23 start co-expressing cytokines from the Th1-cell lineage. This leads to possibly pathogenic IFNγ-producing Th17-cells, also called Th17.1 cells. Enhanced expression of the IL-23 receptor on Th17(.1) cells might contribute to their pro-inflammatory pathogenic phenotype (62, 69, 70). IL-23 is increased in exhale breath condensate of SSc patients, so perhaps Th17 plasticity plays a role in SSc pathology (71). Furthermore, IFNγ, IL-12, and TNFα can induce plasticity toward Th17.1-cells (62). Both serum IL-12 and TNFα are enhanced in IPAH patients and mRNA transcripts of these cytokines were increased in lungs rats in a PH model (46, 72). IL-17/IFNγ-double producing Th-cells are observed within the arteries of atherosclerosis patients, where they provoke pro-inflammatory cytokine production (e.g., IL-6, CXCL10) by vascular SMCs (73). This feedback loop could also exist within PAH, since IL-6 is highly produced by pulmonary ECs of IPAH patients. In addition, IL-6 promotes SMC proliferation in a hypoxia-induced PH model (74, 75). Blocking of IL-6 signaling improved PH physiology in a hypoxiainduced PH mouse model and prevented accumulation of Th17 cells (63). IL-6 also converts Th17-cells into IL-17+ Tregs, which are less suppressive than conventional Tregs (76). In SSc, IL-17+ Tregs are observed in the circulation and possibly also in the skin, indicated by IL-17 and FoxP3 positivity (64, 65, 77). The balance between pro-inflammatory Th17-cells and antiinflammatory Tregs is crucial to control autoimmune features. IL-6 is a key cytokine in Th17/Treg balance, since TGF-β alone polarizes naïve Th-cells to Tregs, while TGF-β together with IL-6 induces Th17-cells (78). Active TGF-β signaling is very prominent in PAH and can be produced by different cells, like monocytes and DCs (79). However, whether DC-derived IL-6 plays a prominent role is unknown yet, as many cells can produce IL-6. In favor of a disturbed balance are the decreased number of Tregs observed in SLE, which correlates with disease severity (66). In CTD-PAH patients Th17-cells and Th17-related cytokines are elevated compared to AD patients without PAH (80). The disturbed Th17/Treg ratio even appears to correlate with PAH severity in APAH patients (80). This demonstrates that Th17-cells and Tregs are implicated not only in ADs but also in PAH (80).

Therefore, Th17 plasticity and Th17/Treg balance may contribute to ADs and PAH, potentially in part by modulating vascular remodeling.

#### Humoral Immune Response

Apart from their interaction with Th17-cells, DCs can induce (immature) Tfh-cells, which develop under the influence of IL-21, IL-6, IL-12, and IL-27 (78). In mature TLOs containing GCs, Tfhcells interact with B-cells, leading to either antibody-producing plasma cells or memory B-cells. There is clear evidence for B-cell dysregulation in IPAH and CTD-PAH (81, 82). In IPAH patients circulating B-cells have an increased expression of genes involved in inflammatory mechanisms, host defense, and endothelial dysfunction, suggesting increased activation of B-cells (82). Also numbers of circulating plasmablasts are elevated in IPAH patients (83). Anomalies in B-cell homeostasis were also observed in SSc-PAH patients, with increased circulating IgD+ B-cell proportions (81). Tfh-cell numbers crucially control the development of autoreactive B-cells, since an increase in Tfh-cell number can lead to increased autoantibody production (84, 85). In several ADs, Tfhcells are increased in blood and target organs (86–89). Serum IgG, IgM, and IgA antibodies are elevated in IPAH patients, and ECspecific IgA promotes cytokine production and upregulation of adhesion molecules (83, 90–92). IgG and IgM antibodies directed against EC-surface antigens are also found in ADs and CTD-PAH, being most prevalent in SSc-PAH patients, followed by IPAH patients and SSc patients without PAH (92). IgG antibodies in SSc and SLE were directed against microvascular ECs antigens, while IgG in SSc, IPAH, and CTD-PAH recognized microvascular dermal and lung EC antigens, and vascular SMCs (90, 91, 93– 95). Auto-reactive IgG provoked EC dysfunction, induced proinflammatory signals, and increased adhesiveness of T-cells to ECs, which also modulated migration and proliferation of SMC. These autoantibodies from SSc or CTD-PAH patients can directly cause signs of PH when injected into healthy mice (96). It is unknown where the autoantibodies found in IPAH and CTD-PAH patients are produced. TLO might be a likely location since Tfh-cells and B-cells, and perhaps antigens, are present in these TLOs. However, these autoantibodies can also be produced in the draining LNs.

In brief, pathogenic autoantibodies in CTD-PAH and IPAH might be produced by dysregulated B-cells that interact with Tfhcells in TLOs. These autoantibodies recognize protein epitopes expressed by ECs, leading to endothelial dysfunction and vascular remodeling. So far, the role of Tfh-cells in IPAH is unknown and further research is needed.

#### GENETICS

Increased activation of the immune system in PAH is also supported by different polymorphisms observed in genome wide association studies. A polymorphism in TLR2 of SSc patients is associated with PAH development (26). Functional analysis of mo-DCs and cDCs carrying the TLR2 polymorphism showed enhanced cytokine production, including IL-6, compared to control DCs. As discussed above, IL-6 plays a prominent role in PAH pathology. Strikingly, a decreased IL-6 serum level was observed in healthy individuals and patients with a single nucleotide polymorphism in the promotor region of the IL-6 gene, IL-6-572C/G, which correlated with decreased risk to develop IPAH (97). SNPs might not only be useful to determine disease susceptibility but also to determine disease onset or activity, as is seen for a specific SNP in TGFB gene in heritable PAH patients carrying a BMPR2 mutation (98). Another genetic association found in both PAH and SSc involving immune activation is a SNP in the TNFAIP3 gene (99). TNFAIP3 encodes for the ubiquitinating enzyme A20, which is crucial for down-regulation of the nuclear factorkappa B (NF-κB) signaling pathway and thereby cell activation (100). Macrophages, pulmonary arterial ECs, and pulmonary arterial SMCs in end-stage IPAH patients showed an increased

#### REFERENCES


expression in NF-κB (101), suggesting an important role for the NF-κB pathway in IPAH.

This demonstrates that several SNPs and genes that are involved in DC function are present in PAH patients.

#### FUTURE DIRECTIONS

In conclusion, different DC subsets are involved not only in the pathobiology of ADs but appear to play a role in the pathobiology of IPAH and CTD-PAH as well. However, the exact role of these DCs in PAH development has not been fully elucidated. The increasing knowledge on DC biology obtained by advanced immunological techniques has led to a more unified method to identify DC subsets and the discovery of new DC subsets. Determining the role of all currently known DC populations, including AS-DCs, as well as their specific functions may help to unravel the pathobiology of PAH. This might lead to new opportunities for therapies targeting specific DC subsets, their activation, and/or their effector function.

### AUTHOR CONTRIBUTIONS

DvU and MK wrote the manuscript. KB contributed to the review of the manuscript. All authors approved the manuscript for publication.

#### FUNDING

This work was supported by a grant of the Dutch Heart Foundation (2016T052).

#### ACKNOWLEDGMENTS

We would like to thank O.B.J. Corneth for critically reading the manuscript.


idiopathic pulmonary arterial hypertension. Eur Respir J. (2018) 51:1701214. doi: 10.1183/13993003.01214-2017


leukocyte infiltration and inflammation. Am J Physiol Lung Cell Mol Physiol. (2011) 301:L50–9. doi: 10.1152/ajplung.00048.2010


pulmonary arterial hypertension. Proc Natl Acad Sci USA (2015) 112:E2677– 86. doi: 10.1073/pnas.1424774112


with healthy controls: a cross sectional study. Respir Res. (2008) 9:20. doi: 10.1186/1465-9921-9-20


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 van Uden, Boomars and Kool. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Regulation of Antigen Export to the Cytosol During Cross-Presentation

Marine Gros and Sebastian Amigorena\*

INSERM U932, Institut Curie, Paris, France

Cross-priming refers to the induction of primary cytotoxic CD8<sup>+</sup> T cell responses to antigens that are not expressed in antigen presenting cells (APCs) responsible for T cell priming. Cross-priming is achieved through cross-presentation of exogenous antigens derived from tumors, extracellular pathogens or infected neighboring cells on Major Histocompatibility Complex (MHC) class I molecules. Despite extensive research efforts to understand the intracellular pathways involved in antigen cross-presentation, certain critical steps remain elusive and controversial. Here we review recent advances on antigen cross-presentation, focusing on the mechanisms involved in antigen export to the cytosol, a crucial step of this pathway.

Keywords: dendritic cells, cross-presentation, cytosolic antigen export, ERAD, endosomal leakage

#### INTRODUCTION

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Marianne Boes, Utrecht University, Netherlands Margarita Del Val, Spanish National Research Council (CSIC), Spain

> \*Correspondence: Sebastian Amigorena sebastian.amigorena@curie.fr

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 29 September 2018 Accepted: 09 January 2019 Published: 28 January 2019

#### Citation:

Gros M and Amigorena S (2019) Regulation of Antigen Export to the Cytosol During Cross-Presentation. Front. Immunol. 10:41. doi: 10.3389/fimmu.2019.00041 Dendritic cells (DCs) play a central role in immune homeostasis by linking innate sensing to adaptive immune responses. After sampling antigens in peripheral tissues, DCs mature and migrate to lymph nodes, where they initiate adaptive immune responses by presenting processed antigens in the context of Major Histocompatibility Complex (MHC) molecules to T cells. For a long time, the generally accepted paradigm supposed that exogenous antigens were exclusively presented via MHC-II molecules to CD4<sup>+</sup> T cells, while endogenous cytosolic antigens, derived from self or foreign proteins, were loaded on MHC-I, thereby leading to naïve cytotoxic CD8<sup>+</sup> T cell activation. Yet, this simple assumption failed to explain how cytotoxic immune responses could be mounted against pathogens that do not readily infect DCs. This apparent contradiction was resolved by the discovery of cross-presentation, a process enabling the delivery of exogenous antigens to the MHC-I pathway for cross-priming CD8<sup>+</sup> cytotoxic T cell responses (1, 2). Since its first description over forty years ago, our understanding of the sequence of events governing antigen cross-priming has extensively increased, leading to the description of two main pathways of antigen cross-presentation, referred to as "vacuolar" and "cytosolic." While the requirement for cross-presentation in the initiation of anti-tumor immune responses is now well established (3–7), its control and the precise intracellular routes involved remain incompletely understood and, for some parts, controversial.

Here, we review the most recent advances in the analysis of antigen cross-presentation in mouse (unless stated otherwise), with a particular emphasis on the advances in understanding of antigen export to the cytosol, a crucial, yet debated, step of the cytosolic pathway.

# PATHWAYS FOR ANTIGEN CROSS-PRESENTATION

In 1976, seminal work by M. Bevan showed that exogenous antigens could be presented on MHC-I molecules and prime cytotoxic immune responses, thereby unearthing a novel antigen presentation pathway that he called cross-priming (1, 2). However, the molecular mechanisms underlying cross-priming and "cross-presentation" remained elusive until the early nineties. At that time, several lines of evidence reported that cross-presentation of bacterial antigens [i.e., the 257-264 H-2K<sup>b</sup> -restricted epitope of ovalbumin (OVA) fused to E. coli Crl protein] was resistant to proteasome inhibitors (8) (suggesting lysosomal processing of the corresponding peptides), unaffected by brefeldin A (BFA) treatment (8–10) [arguing against a critical role for endoplasmic reticulum (ER)-Golgi transport] and most of the time, occurred independently from TAP, the transporter mediating peptide import into the ER (8, 11). These observations led to the first description of the "vacuolar pathway." After internalization, antigens remain confined in intracellular compartments, where they undergo lysosomal degradation, a process largely dependent on cathepsin S activity (12), and followed by loading onto post-Golgi MHC-I molecules.

Simultaneous studies with particulate, non-bacterial antigens (i.e., bead-bound OVA), showed that TAP1 deficiency in macrophages, as well as BFA treatment, abolished their ability to cross-present exogenous antigens, thereby suggesting that antigen-derived peptides must be transferred from the cytosol to the ER to bind newly synthesized MHC-I molecules (13). Additionally, cross-presentation was disrupted by proteasome inhibitors (13–16), consistent with a model in which antigens are delivered into the cytosol before proteasomal degradation and peptide import into the ER. This pathway, later termed the "cytosolic pathway," implies the export of antigens from endocytic compartments to the cytosol. The first experimental evidence of this crucial step was provided by the use of gelonin, a membrane-impermeant toxin that inactivates ribosomes when transferred to the cytosol. Macrophages phagocytosing gelonincoated beads displayed reduced protein synthesis, indicating export of bead-bound gelonin to the cytosol (13, 14).

The aforementioned pivotal studies used mouse macrophages as models of antigen-presenting cells (APCs). It later became clear that DCs, rather than macrophages, cross-present antigens and cross-prime cytotoxic immune responses efficiently (17, 18), by means of different properties of their phagocytic pathway, including lower degradation capacity (19). When considering DCs, these cells represent a series of ontogenically and functionally diverse populations. In mice, two main resident DC subsets are found in the spleen and lymph nodes, namely Batf3-dependent CD8<sup>+</sup> XCR1<sup>+</sup> DCs (DC1s) and IRF4 dependent CD8<sup>−</sup> CD11b<sup>+</sup> DCs (DC2s) [reviewed in (20)]. At steady state, DC1s cross-present cell-associated antigens more efficiently than their DC2 counterparts, a capacity first attributed to their increased ability to capture this type of antigen (21, 22). Later experiments showed that higher crosspresentation efficacy in mouse DC1s is intrinsic and unrelated to the route of antigen uptake (23, 24), thus contrasting with the FcγR-dependent optimization of cross-presentation observed in human DC1s (25). In mouse, surface receptors, including Clec9A/DNGR-1 (26–29) or mannose receptor (MR) (30), were proposed to preferentially deliver antigens to the crosspresentation pathway, most likely through delaying delivery of their cargoes to late endosomal and lysosomal degradative compartments. DC1s also bear specialized endocytic properties that reduce/delay acidification and degradation of endocytic cargo (19, 31).

Consistent with these in vitro observations, mice deficient for DC1s (5), or displaying cross-presentation-defective DCs (4, 6), fail to mount cytotoxic immune responses against tumors and to control tumor development, even after treatment with checkpoint blockers. Although DC1s are best suited for crosspresentation both in vitro and in vivo, DC2s' ability to crosspresent is increased by targeting antigens to DC2 specific receptors, such as FcγR (32) or DCIR2, in a stimulatory context (33), thus suggesting that both DC1 and DC2 are capable of cross-presenting antigens depending on the conditions.

The relative contributions of the cytosolic and vacuolar pathways to in vivo cross-presentation and cross-priming remain unclear. TAP dependency can potentially affect both pathways, as it impairs the exit of MHC-I molecules from the ER (34–37). Whether critical players in cross-presentation, such as Sec22b (4, 38, 39) or Rab43 (40), which are both required for effective cross-priming, are selectively involved in one or both pathways is unknown. The best available evidence for the cytosolic pathway being predominant in cross-priming comes from a study using mice defective for the immunoproteasome subunit LMP7. These mice show impaired cross-priming for an immunoproteasome-dependent H-Y epitope, supporting a critical role for proteasome-dependent processing, and therefore, for the cytosolic pathway in vivo (41). Since delivery of internalized antigens to the cytosol is very ineffective in most cell types, DCs might have developed specialized pathways to link these two subcellular compartments.

#### BIOLOGICAL PARAMETERS INFLUENCING ANTIGEN EXPORT TO THE CYTOSOL

#### Nature of Cytosolic Export-Competent Cells

By using gelonin activity or cytosolic fluorescence quantification as readouts, initial studies showed that inflammatory (14) or activated (16) mouse macrophages displayed a measurable ability to export bead-conjugated (14) or soluble (16) cargo into the cytosol. Further work revealed that soluble or complexed antigens also get access to the cytosol in steadystate bone-marrow derived DCs (BMDCs) or in a DC cell line, without prior activation (17, 18). Moreover, antigen export to the cytosol is more efficient in DCs than in macrophages, as illustrated by subcellular fractionation and subsequent western blotting (18). To assess whether DC subsets differ in their capacity to perform such transfer, Lin et al. developed a cytochrome c-based assay relying on the selective apoptosis of cells exporting exogenous cytochrome c into the cytosol (42). Only a fraction of DC1s showed susceptibility to cytochrome c-induced apoptosis, indicating a functional specialization for endosome to cytosol transport in these cells (42, 43). Notably, this cytochrome c-sensitive DC1 population strictly corresponds to the cohort of efficient crosspresenters, whereas cytochrome c-resistant DC1s cross-present antigens inefficiently and share other functional features with DC2s (42).

#### Nature of Antigens Exported to the Cytosol

Early microscopy observations showed that fluorescent (i.e., dextrans) or soluble (i.e., enzymatically active horseradish peroxidase: HRP) antigens gained access to the cytosol in DCs (17, 18). While 3–40 K dextrans are rapidly relocated to the cytosol, higher molecular mass dextrans (500–2,000 K) remain vacuolar (18), suggesting that antigen export to the cytosol is size-selective (18, 44). Particulate antigens, which are more efficiently cross-presented than soluble ones (14), often form large aggregates and therefore require dissociation before their translocation to the cytosol. Indeed, inhibition of vacuolar acidification abolishes the disaggregation of immune complexes and their subsequent cytosolic export (18), thus pointing to a crucial role ofslightly acidic endo/phagosomal pH in this process. While some degree of degradation might favor antigen export to the cytosol due to the size-restriction of transported antigens (18), high proteolytic activity, favored by acidic pH, could destroy MHC-I-binding epitopes. In this regard, regulation of endocytic pH is of crucial importance. In DCs' endocytic compartments, incomplete assembly of v-ATPase proton pump together with Rab27a-dependent recruitment of NOX2 jointly lead to active alkalinization of luminal pH (19, 45), thereby preserving antigens from detrimental excessive degradation (46).

#### Export to the Cytosol and DC Activation

Aside from putative intrinsic properties of DC1s, extrinsic signals, such as Toll-Like Receptor stimulation, influence antigen export to the cytosol. Indeed, short (3–5 h) lipopolysaccharide (LPS) stimulation of BMDCs increases the proportion of cells displaying exogenous HRP in their cytosol (47). A possible explanation for the observed LPS-mediated increase in antigen export may reside in the requirement for TRIF in this process (48). Until recently, absence of quantitative reliable antigen export assays based on endotoxin-free reagents impeded detailed analysis of the role of DC activation in antigen transport to the cytosol. Recently published export assays should overcome this limitation (49).

#### Kinetics of Antigen Export to the Cytosol in DCs

Kinetics studies showed that HRP appeared in BMDC cytosol only 15 min after internalization (17). Rapid egress suggests that antigens are exported from early endosomes (50), as supported by microscopy experiments (51) or by mathematical modeling (52). The latter predicts that 20 min after internalization, cytosolic export of yeast-derived antigen competes with degradation associated with maturation of the endocytic compartment. Thus, only a tiny fraction of, at least, non-complexed antigens released after this time point might contribute to cross-presentation. Cytosolic translocation of HRP immune complexes appears after 60 min, and reaches a plateau after 6 h (18). Similar findings were reported for cytosolic egress of antigens associated to beads (53, 54). Additionally, these two studies provided compelling evidence that ER-mediated delivery of MHC-I loading machinery to the phagosome rendered this compartment competent for cross-presentation (55) following TAP-mediated import of cytosolic peptides (53, 54). While the relative contributions of ER and plasma membrane to the formation of cross-presentationcompetent phagosomes remain debated (39, 56), Houde et al. postulated that an ER transporter, Sec61, might be involved in the translocation of antigens from the phagosomal lumen to the cytosol (53). This hypothesis was later experimentally supported by several studies detailed in the next section.

#### MOLECULAR MECHANISMS OF ANTIGEN EXPORT TO THE CYTOSOL

#### ERAD Transporter-Dependent Hypothesis

Existence of a transporter mediating antigen export to the cytosol naturally imposes conformational constraints on the translocated antigen. Indeed, antigens are unlikely to be transported in their native structure, considering the narrow diameter of known transporter pores, and are therefore expected to undergo an unfolding step before translocation. Supporting this idea, fixed OVA is less efficiently translocated into the cytosol than structurally flexible one (57). Moreover, during unfolding, reduction of disulfide bonds by GILT, a phagolysosomal thiol reductase constitutively expressed in APCs, is essential for cytosolic export of viral disulfide-rich antigens and subsequent cross-presentation (58) (**Figure 1**, left panel).

Although the requirement for protein unfolding suggests that antigens gain access to the cytosol through a transporter, the nature of the channel mediating this process remains controversial. Studies attempting to answer this question reported interactions between unfolded OVA and members of the ER-associated degradation (ERAD) machinery in ER-associated compartments (59), consistent with previous findings (53, 54). This observation led to the hypothesis that the ERAD machinery, mediating retro-translocation of misfolded proteins from the ER lumen to the cytosol, potentially through the trimeric Sec61 channel, could also operate from endocytic compartments during cross-presentation. The first functional insights into ERAD contributions to antigen export to the cytosol, came from studies using exotoxin A (ExoA), a bacterial toxin binding to the cytosolic N-terminal domain of Sec61α, and resulting in channel closure (60, 61). ExoA treatment reversed the ICP47-mediated inhibition of TAP, the latter resulting from the translocation of exogenously delivered ICP47 to the cytosol and its subsequent interaction with the cytosolic side of TAP (62). This finding, associated with the observed decrease in OVA cross-presentation following ExoA treatment (62, 63) or siRNA-mediated silencing of Sec61 (48, 59), strongly pointed to Sec61 being the channel controlling antigen export to the cytosol (**Figure 1**, left panel). In line with this hypothesis, the expression of the Sec61α, β and γ subunits is increased in DC1s, as compared to DC2s, correlating with their specific cross-presenting ability (64).

However, it has been extremely difficult to address the precise contribution of Sec61 in antigen cross-presentation and retrotranslocation from endo/phagosomes, as this channel also mediates co-translational import of proteins, including MHC-I, into the ER. To shed some light on this issue, Zehner et al. used a intrabody-based approach aiming to retain Sec61 in the ER and thereby prevent its recruitment to endocytic

compartments (48). Expression of the anti-Sec61 intrabody in BMDCs impairs antigen export to the cytosol and OVA crosspresentation, consistent with a role for Sec61 outside the ER, possibly in endosomes. Still, the involvement of Sec61 itself in ERAD-dependent retrotranslocation remains unclear and fraught with technical issues [reviewed in (65)]. Additionally, recent work has shown that sustained inhibition of Sec61 with a specific toxin, mycolactone, has no effect on antigen export to the cytosol, and indirectly reduces OVA cross-presentation through downregulation of other players in the pathway, including MHC-I (66). While Sec61 involvement in cytosolic antigen translocation needs further clarification, other ERAD components, such as Hrd1 and Derlin-1, might be alternative candidates.

Hrd1, an ER-resident ubiquitin ligase tagging ERAD substrates, exhibits six transmembrane domains, which is enough to form a channel (67, 68). siRNA-mediated depletion of Hrd1 in DCs results in decreased antigen export to the cytosol and cross-presentation, as well as impaired MHC-II presentation (48). These alterations in antigen presentation pathways might be caused by Hrd1 silencing-mediated ER stress and therefore require further investigation. On the other hand, the protease Derlin-1 (Der1), comprising four transmembrane domains, cannot form a channel but could possibly function as an accessory subunit of the export channel (69) by trapping ERAD substrates and rerouting them for cytosolic degradation (70). Yet, antigen cross-presentation is not perturbed by Der1 silencing in both murine BMDCs (48) and human monocyte-derived DCs (71), thus excluding a putative role for Der1 in antigen export to the cytosol.

To date, the best evidence available suggests that ERAD might control antigen transfer to the cytosol through the activity of the AAA ATPase p97. P97 forms an hexameric ring and is thought to provide the energy necessary for passage of proteins through the retrotranslocating channel (72). Exogenous addition of p97 to isolated phagosomes loaded with luciferase leads to luciferase release, whereas addition of a dominant negative version of p97 fails to do so (62), suggesting a role for the ATPase in antigen translocation from phagosomes (**Figure 1**, left panel). Along the same lines, human and mouse DCs silenced for p97 (59, 71) or expressing a dominant negative form of p97 (62), display impaired cross-presentation of MelanA and OVA antigens, respectively, whereas p97 overexpression enhances this pathway (73). P97 is recruited to endosomes following mannose receptor (MR)-poly-ubiquitination. This post-translational modification proves to be crucial for antigen export to the cytosol and OVA cross-presentation as expression of a mono-ubiquitinated form of the MR is sufficient to reduce both processes (73). Of note, MR poly-ubiquitination is triggered by OVA binding to the receptor, and is negatively regulated by the ESCRT (Endosomal Sorting Complex Required for Transport)-I protein TSG-101 (73).

Several studies investigating the role of p97 in antigen export to the cytosol used the luciferase enzyme to monitor this intracellular event (62, 74). Following unfolding in endocytic compartments and subsequent translocation into the cytosol, luciferase would need to be refolded to exert its functionality, a process likely mediated by the chaperone Hsp90. Indeed, cytosolic refolding of exogenous unfolded luciferase is compromised in Hsp90β-silenced human DCs or in DCs treated with the Hsp90 inhibitor radicicol (74). Furthermore, Hsp90α deficiency not only inhibits cross-presentation in mouse BMDCs, but also decreases cytosolic translocation of OVA, therefore implying that Hsp90 itself could mediate antigen transport to the cytosol (43, 57). Additionally, Hsp90 could protect the exported antigens from premature cytosolic degradation, before Hsp70 mediated targeting to the proteasome (57) (**Figure 1**, left panel).

The "transporter hypothesis" has, so far, garnered the most experimental support, as the main conduit for antigen export. However, it still raises important questions. Given the high degree of substrate selectivity during ERAD [reviewed in (65)], the use of a unique transporter translocating a wide variety of antigens seems unlikely. Moreover, this hypothesis fails to explain how large, non-protein molecules, such as dextrans, can be transferred to the cytosol in absence of ubiquitination, the latter being a prerequisite for ERAD-mediated translocation. Altogether, these observations do not exclude a role for ERAD in antigen export to the cytosol, but rather suggest the contribution of additional mechanisms.

#### Alternative Hypothesis: Rupture of the Antigen-Containing Compartment

The first descriptions of the cytosolic pathway for crosspresentation supposed that antigens could escape endocytic compartments through membrane rupture. This hypothesis, at that time termed "indigestion model," relies on the observation that large particles are more efficiently cross-presented than small ones, and could thus be responsible for phagosomal overload, leading to membrane disruption and efficient antigen leakage to the cytosol (14). Despite intensive use of this pathway for cytosolic delivery of antibodies or bioactive proteins conjugated with endosomolytic peptides (75–77), evidence of its contribution to cross-presentation were lacking, until recently.

#### ROS, Lipid Peroxidation, and Membrane Rupture

A recent study showed that following LPS stimulation, VAMP8 dependent NOX2 recruitment to BMDC endosomes resulted in Reactive Oxygen Species (ROS) production and subsequent endosomal lipid peroxidation (78, 79). This alteration of lipid structure disrupts endosomal membrane integrity, leading to antigen escape to the cytosol and OVA cross-presentation (78) (**Figure 1**, upper right panel). Interestingly, ROS production in endocytic compartments is intrinsically linked to cells' crosspresenting ability. Indeed, DCs show sustained and stronger ROS production than macrophages (19), the latter subset increasing phagosomal ROS production, as well as cross-presentation, only after activation (17, 19). Moreover, ROS generation is higher in DC1 than in DC2 phagosomes, thereby correlating with the enhanced ability of DC1s to cross-present antigens (31). Biophysical studies provided mechanistic insights into lipid peroxidation-dependent membrane rupture. Oxidized lipid-rich artificial bilayers show higher water permeability (80), as well as increased membrane curvature, associated with micellization and membrane destabilization (81).

#### Changes in Endolysosomal Membrane Lipid Composition

Aside from lipid peroxidation, enrichment in ceramide, and to a greater extent in sphingosine, also triggers membrane permeability to solutes (82). While some studies proposed that sphingosine-based lipids could form large channels in membranes through an "all or none" mechanism (83), others suggested that sphingolipids actually promote membrane permeabilization by a graded process involving rigidification of membrane domains and subsequent creation of local structural defects (82, 84) (**Figure 1**, lower right panel). Sphingosine synthesis results from ceramide deacetylation by two ceramidases, encoded by the Asah1 and Asah2 genes, and respectively functioning at acid or neutral pH. Notably, the expression of both enzymes is higher in DC1s than in DC2s (immgen.org), suggesting that ceramide conversion into membrane-disrupting sphingosine could be increased in DC1 endocytic compartments. This DC1-specific enrichment in sphingosine could possibly be mediated by lipid bodies, which have also been proposed to destabilize some ER or phagosomal membrane domains during their formation, thereby causing leakage of the content of these compartments (85). Along the same line, BMDCs deficient for Igtp, a GTPase controlling accumulation of lipid bodies, show a selective defect for crosspresentation (86). Moreover, intracellular accumulation of lipid droplets correlates with cross-presentation efficiency, as DC1s display significantly higher amounts of these organelles than DC2s (86). However, pharmacological interference with lipid body formation fails to influence antigen export to the cytosol in the context of saponin adjuvant-based cross-presentation (87). Thus, the precise role of lipid bodies in cross-presentation and cytosolic antigen leakage remains to be specified.

#### Compensatory Mechanisms for Endocytic Membrane Rupture

Although several lines of evidence point to a contribution of endocytic membrane disruption and subsequent antigen leakage into the cytosol, this model has been repeatedly dismissed owing to its presumable lack of regulation and ensuing cell toxicity. Indeed, links between endocytic leakage and cell death were reported in different systems. Silica crystal-dependent phagosomal rupture, for example, leads to cytosolic release of intraluminal cathepsin B, which in turn activates the NLRP3 inflammasome, resulting in pyroptosis (88). Hydroxychloroquine-mediated cathepsin release from lysosomes can also trigger caspase activation and apoptosis (89), suggesting a requirement for control mechanisms to contain damaging consequences of leakage.

In this regard, the ESCRT machinery, formerly known for its key role in viral budding or cytokinetic abscission (90, 91), was recently identified as a core component of biological membrane repair following damage (92–96). A role for the ESCRT-I protein TSG101 in antigen export to the cytosol and cross-presentation has been previously suggested (73). However, increased cytosolic export observed following TSG101 silencing had been attributed to TSG101-dependent inhibition of MR poly-ubiquitination, required for cytosolic antigen translocation. Yet, considering the dispensable role of ubiquitination in antigen export to the cytosol (43) and the fact that TSG101 is also required for ESCRT-III-mediated repair of endolysosomal membranes (95, 96), defects in endocytic membrane repair, concomitant with TSG101 depletion, may have also contributed to the observed increased export phenotype (73). Yet, the possible involvement of ESCRT-III in controlling antigen export to the cytosol has not been investigated so far.

#### REFERENCES


#### CONCLUSION

Identification of several critical players in antigen cross-presentation, such as Sec22b (4), or Rab43 (40), and their subsequent validation in conditional knock-out mouse pre-clinical models established a major role for this pathway in different types of immunes responses, including anti-tumor immune responses. Yet, the way antigens gain access to the cytosol during cross-presentation is far from being entirely resolved. Export to the cytosol is not only the last event in the pathway that remains largely obscure, but it is also a rate-limiting step in the process (52, 97). Identifying the molecular mechanism involved will certainly provide relevant targets to manipulate antigen cross-presentation for vaccination and immunotherapy purposes.

#### AUTHOR CONTRIBUTIONS

MG and SA designed, prepared and wrote the manuscript.

#### FUNDING

MG is supported by a Fondation pour la Recherche Médicale grant (grant No. FDT201805005336). MG and SA are supported by ANR-10-IDEX-0001-02 PSL<sup>∗</sup> , ANR-11-LABX-0043 grants, and ERC grant 2013-AdG No. 340046 DCBIOX.

#### ACKNOWLEDGMENTS

The authors thank Dr. Philippe Benaroch and Dr. Marianne Burbage for critical reading and feedback on the manuscript. This work benefitted from data assembled by the Immgen consortium.


processing due to decreased MHC-I stability at phagolysosomal pH. J Immunol. (2003) 170:5825–33. doi: 10.4049/jimmunol.170.12.5825


cross-presentation compartment in dendritic cells. Nature (2003) 425:397–402. doi: 10.1038/nature01911


of p97 and cytosolic antigen translocation for cross-presentation. Proc Natl Acad Sci USA. (2011) 108:9933–8. doi: 10.1073/pnas.1102397108


97. Moore MW, Carbone FR, Bevan MJ. Introduction of soluble protein into the class I pathway of antigen processing and presentation. Cell (1988) 54:777–85.

**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Gros and Amigorena. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Dendritic Cell Targeting Using a DNA Vaccine Induces Specific Antibodies and CD4<sup>+</sup> T Cells to the Dengue Virus Envelope Protein Domain III

Arthur Baruel Zaneti <sup>1</sup> , Marcio Massao Yamamoto<sup>1</sup> , Fernando Bandeira Sulczewski <sup>1</sup> , Bianca da Silva Almeida<sup>1</sup> , Higo Fernando Santos Souza<sup>1</sup> , Natália Soares Ferreira<sup>1</sup> , Denicar Lina Nascimento Fabris Maeda<sup>2</sup> , Natiely Silva Sales <sup>2</sup> , Daniela Santoro Rosa3,4 , Luís Carlos de Souza Ferreira<sup>2</sup> and Silvia Beatriz Boscardin1,4 \*

#### Edited by:

Urszula Krzych, Walter Reed Army Institute of Research, United States

#### Reviewed by:

Tejram Sahu, Johns Hopkins University, United States Sri H. Ramarathinam, Monash University, Australia

> \*Correspondence: Silvia Beatriz Boscardin

#### Specialty section:

sbboscardin@usp.br

This article was submitted to Vaccines and Molecular Therapeutics, a section of the journal Frontiers in Immunology

> Received: 05 October 2018 Accepted: 10 January 2019 Published: 29 January 2019

#### Citation:

Zaneti AB, Yamamoto MM, Sulczewski FB, Almeida BdS, Souza HFS, Ferreira NS, Maeda DLNF, Sales NS, Rosa DS, Ferreira LCdS and Boscardin SB (2019) Dendritic Cell Targeting Using a DNA Vaccine Induces Specific Antibodies and CD4<sup>+</sup> T Cells to the Dengue Virus Envelope Protein Domain III. Front. Immunol. 10:59. doi: 10.3389/fimmu.2019.00059 <sup>1</sup> Department of Parasitology, Institute of Biomedical Sciences, University of São Paulo, São Paulo, Brazil, <sup>2</sup> Department of Microbiology, Institute of Biomedical Sciences, University of São Paulo, São Paulo, Brazil, <sup>3</sup> Department of Microbiology, Immunology and Parasitology, Federal University of São Paulo (UNIFESP/EPM), São Paulo, Brazil, <sup>4</sup> Institute for Investigation in Immunology (iii)-INCTiii, São Paulo, Brazil

Dengue fever has become a global threat, causing millions of infections every year. An effective vaccine against all four serotypes of dengue virus (DENV) has not been developed yet. Among the different vaccination strategies available today, DNA vaccines are safe and practical, but currently induce relatively weak immune responses in humans. In order to improve immunogenicity, antigens may be targeted to dendritic cells (DCs), the main antigen presenting cells and orchestrators of the adaptive immune response, inducing T and B cell activation. It was previously shown that a DNA vaccine encoding a fusion protein comprised of an antigen and a single-chain Fv antibody (scFv) specific for the DC endocytic receptor DEC205 induced strong immune responses to the targeted antigen. In this work, we evaluate this strategy to improve the immunogenicity of dengue virus (DENV) proteins. Plasmids encoding the scFv αDEC205, or an isotype control (scFv ISO), fused to the DENV2 envelope protein domain III (EDIII) were generated, and EDIII specific immune responses were evaluated in immunized mice. BALB/c mice were intramuscularly (i.m.) immunized three times with plasmid DNAs encoding either scDEC-EDIII or scISO-EDIII followed by electroporation. Analyses of the antibody responses indicated that EDIII fusion with scFv targeting the DEC205 receptor significantly enhanced serum anti-EDIII IgG titers that inhibited DENV2 infection. Similarly, mice immunized with the scDEC-EDIII plasmid developed a robust CD4<sup>+</sup> T cell response to the targeted antigen, allowing the identification of two linear epitopes recognized by the BALB/c haplotype. Taken together, these results indicate that targeting DENV2 EDIII protein to DCs using a DNA vaccine encoding the scFv αDEC205 improves both antibody and CD4<sup>+</sup> T cell responses. This strategy opens perspectives for the use of DNA vaccines that encode antigens targeted to DCs as a strategy to increase immunogenicity.

Keywords: dengue fever, dendritic cells, envelope protein domain III, single-chain Fv antibody, DNA vaccine

# INTRODUCTION

Dengue virus (DENV) is the causative agent of dengue fever, an infection that has become a serious public health issue. In the last decades, the alarming increase in the number of cases [50–100 million per year, (1)] and also the increase in the incidence of more severe clinical forms of the disease (like dengue hemorrhagic fever, DHF or dengue shock syndrome, DSS) led the World Health Organization to prioritize the development of a vaccine against dengue (2). DENV is transmitted to humans by the bite of mosquitoes of the genus Aedes (such as Aedes aegypti and Aedes albopictus) infected with one of the four virus serotypes (DENV 1–4) (3).

The virus genome is translated into a polyprotein which is processed by virus and host proteases to produce three proteins that make up the viral particle: capsid (C), premembrane/membrane (prM/M) and envelope (E), and seven other non-structural proteins, NS1, NS2a, NS2b, NS3a, NS4a, NS4b, and NS5 (3). The E protein plays an important role in the protective immunity against DENV as it contains the majority of epitopes that elicit neutralizing antibodies (4–6). This protein can be divided into three domains: the central domain (EDI), the domain responsible for dimerization containing the fusion peptide (EDII), and the domain that binds to the surface cell receptor (EDIII) (2). EDIII has been extensively used in vaccine development for its ability to induce antibodies able to block DENV infection (7–9). In addition to neutralizing antibodies, T cell responses also play a relevant role in the development of protection. T cells limit the spread of viral infection because they kill infected cells and secrete pro-inflammatory cytokines (10, 11). IFNγ-secreting Th1 and CX3CR1<sup>+</sup> cytotoxic CD4<sup>+</sup> T cells are also associated with protection (12, 13).

Dendritic cells (DCs) are central for immunity induction, activating both T and B cells. These cells are excellent antigen presenting cells (APCs) because of their ability to acquire different antigens (either by pinocytosis, endocytosis, or phagocytosis), when compared to other cell types such as macrophages and B cells (14). To accomplish their role as APCs, DCs express a large number of extra and intercellular receptors that are responsible for their ability to "sense" the environment. When they encounter an antigen in the context of infection/inflammation, DCs undergo a maturation process that results in the up-regulation of co-stimulatory and MHCII molecules, and increases their ability to present antigens in the context of MHC I and II (15).

The last decades have proven to be extremely prolific for the study of DC biology and function, as different DC subsets were identified both in humans and in mice (16). Each subset is normally characterized by the expression of different surface markers. The DEC205 is an endocytic C-type lectin receptor expressed by murine and human DCs in different organs (17– 20) that uptakes antigens and directs them to MHCII rich late endosomes, increasing antigen presentation to CD4<sup>+</sup> T cells (21). In mice, DEC205<sup>+</sup> DCs are resident in the T cell zone of lymphoid organs, and also express the CD8α marker (22). The DEC205+CD8α <sup>+</sup> DCs were involved in the uptake of dying cells, and in the resistance to some viral infections (23–25). Antigens derived from different pathogens have been targeted to DEC205+CD8α <sup>+</sup> DCs using chimeric anti-DEC205 monoclonal antibodies (mAbs) genetically fused to them, administered in the presence of a DC maturation stimulus (26–34).

The use of chimeric mAbs to target antigens to different endocytic receptors has become more spread, and this concept was employed in the development of more effective DNA vaccines. These vaccines are usually safe, cheap, easy to produce but may fail to induce strong immune responses, especially in humans (35). An improvement in the immune response was obtained after antigen targeting to the DEC205<sup>+</sup> DCs using DNA vaccines consisting of plasmids encoding a single chain variable fragment (scFv) fused to the antigen of interest (36–41). However, other groups have shown that antigen targeting to DEC205<sup>+</sup> DCs using DNA vaccines was also able to induce immune tolerance to the antigen of interest and consequently protect against experimental allergic encephalomyelitis (42).

In this work, we generated a DNA vaccine encoding the anti-DEC205 scFv fused to DENV2 EDIII (pscDEC-EDIII). Vaccine administration by intramuscular immunization followed by electroporation elicited high titers of anti-EDIII antibodies in mice immunized with pscDEC-EDIII capable of blocking DENV2 infection in VERO cells. In addition, EDIII targeting to DEC205<sup>+</sup> DCs within the context of a DNA vaccine elicited specific CD4<sup>+</sup> T cell proliferation and pro-inflammatory cytokine production.

#### MATERIALS AND METHODS

#### Plasmid Generation

Constructions of the pcDNA3 scDEC-OVA and scISO-OVA plasmids were described previously (37). We digested the DNA fragment encoding ovalbumin (OVA) with the restriction enzymes NotI and XbaI (New England Biolabs) and purified the open vectors with the "PureLink Quick Plasmid DNA" kit (Invitrogen). The sequence corresponding to the ectodomain of the DENV2 envelope protein (HQ026763, lineage DENV-2/BR0690/RJ/2008) was synthetized and cloned into the pUC57 plasmid (Genscript USA Inc.). We amplified the EDIII sequence (aa 297–394) with the primers sense 5′ GGCGGCCGCATGTCCTACTCTATGTGCAC 3′ and anti-sense 5′ TCTAGATCAGTGATGGTGATGGTGAT-GTTTCTTAAACCAATTCAGCTTC 3′ with the Phusion High Fidelity DNA Polymerase (New England Biolabs) according to the manufacturer's instructions. The anti-sense primer was also designed to insert a 6× His-tag at the end of the sequence. The PCR product was cloned into the pJET1.2/blunt vector (Thermo Scientific) and digested with the restriction enzymes NotI and XbaI (New England Biolabs). The digestion product was purified with the "PureLink Quick Plasmid DNA" kit (Invitrogen) and cloned in frame with the open reading frames of scDEC and scISO in the pcDNA3 vectors using the T4 DNA ligase enzyme (New England Biolabs). The final vectors, named pscDEC-EDIII and pscISO-EDIII were sequenced to confirm the presence of the EDIII sequence in frame with the antibody sequences. We transformed the plasmids into DH5α bacteria and purified the DNA in large scale using the "EndoFree Plasmid Maxi Kit" (QIAGEN) for subsequently transfection of human embryonic kidney (HEK) 293T cells and mice immunization.

#### HEK 293T Transient Transfection

The transfection of HEK293T cells was performed as described previously (30). After 5 days in culture, the supernatants of cultures transfected with pscDEC-EDIII or pscISO-EDIII were collected, centrifuged at 1,000 × g for 30 min and filtered in 0.22µM filters (Corning). The samples were concentrated in a dialysis membrane surrounded by sucrose and dialyzed twice in PBS for 4 h at 4◦C.

#### Western Blotting

The scDEC-EDIII or scISO-EDIII containing samples were sorted in a 12% SDS-PAGE gel under reducing conditions. The proteins were transferred to a nitrocellulose membrane (GE Healthcare) and the membrane was blocked overnight at 4◦C with PBS containing 0.05% Tween 20 (PBS-T 0.05%), 2.5% BSA and 5% non-fat milk. After three 5-min washes with PBS-T 0.05%, the membrane was incubated with 6x-HIS tag monoclonal antibody (1:5,000; Thermo Fisher Scientific) at room temperature (rt) for 2 h. Next, the membrane was incubated with goat anti-mouse total IgG-HRP antibody (1:2,000; Jackson ImmunoResearch Laboratories) for 2 h at rt and developed using quimioluminescence (ECL kit, GE Healthcare).

#### CHO-DEC Binding Assay

Transgenic CHO cells stably expressing the mouse DEC205 receptor (kindly provided by Dr. Michel Nussenzweig, The Rockefeller University, New York) were used for the binding assays. One hundred thousand cells were incubated with 4, 2, or 1µg/mL of the scDEC-EDIII or scISO-EDIII scFvs for 40 min on ice. After two washes with FACS buffer (2% fetal bovine serum in PBS), cells were incubated with the 6x-HIS tag monoclonal antibody (1:5,000; Thermo Fisher Scientific) for 20 min on ice. The cells were washed two times again with FACS buffer and incubated with the anti-mouse IgG-Alexa488 antibody (1:2,000; Life technologies). After another round of washes, the cells were analyzed by flow cytometry and 20,000 events were acquired in the FACSCaliburTM flow cytometer (BD Biosciences).

#### Mice and Immunization

Six- to eight-weeks-old male BALB/c mice were bred at the Isogenic Mouse Facility of the Parasitology Department, University of São Paulo, Brazil. This study was carried out in accordance with the recommendations of the Federal Law 11.794 (2008), the Guide for the Care and Use of Laboratory Animals of the Brazilian National Council of Animal Experimentation (CONCEA) and the ARRIVE guidelines. The Institutional Animal Care and Use Committee (IACUC) of the University of São Paulo approved the protocol under the following number: 36/2016. Groups of eight animals were immunized with 100 µg of pscDEC-EDIII or pscISO-EDIII diluted in saline (0.9% NaCl). A control group consisting of 4 animals was injected with saline alone. Briefly, the animals were anesthetized by intraperitoneal (i.p) injection of a mixture of Ketamine and Xylazine (100 and 10 mg/kg, respectively). Next, the skin over the hind leg was sterilized with ethanol and the injections were carried out intramuscularly (i.m.) in the anterior tibial muscle of the mice followed immediately by electroporation. For the electroporation, two 130 V pulses with 1 ms duration and four 70 V pulses with 50 ms duration were applied with the CUY560- 5-0.5 electrode using the NEPA21 Super Electroporator (Nepa Gene Co., Ltd.). The interval between each pulse was 450 ms. Three doses were administered in 2-week intervals. Animals were euthanized 2 weeks after the last dose, their sera were collected via cardiac puncture and the spleens were removed for subsequent analysis.

## ELISA

We used sera collected from the different immunization groups for the detection of EDIII-specific antibodies by ELISA. High binding ELISA plates (Costar) were coated overnight at rt with 100 ng/well of the recombinant EDIII protein (43) diluted in PBS. In the following day, the plates were washed three times with PBS containing 0.02% Tween 20 (PBS-T 0.02%) and then blocked with PBS-T 0.02%, 1% BSA, and 5% non-fat milk for 1 h at rt. After three washes, serum samples were serially diluted in PBS-T 0.02%, 0.25% BSA, and 5% non-fat milk and incubated for 2 h at rt. Goat antimouse IgG antibody conjugated with horseradish peroxidase (HRP) (1:2,000; Jackson ImmunoResearch Laboratories) or HRP conjugated subclass-specific anti-mouse IgG (1:2,000; SouthernBiotech) were used as secondary antibodies, and plates were incubated for 2 h at rt. ELISA was developed with orthophenylenediaminedihydrochloride (Sigma) and H2O<sup>2</sup> diluted in phosphate–citrate buffer, pH 4.7. The reaction was stopped with 4N H2SO<sup>4</sup> and the OD<sup>490</sup> was measured in a microplate reader (Biotek). Titers represent the highest serum dilution showing an OD<sup>490</sup> > 0.1 normalized in a log10 scale. The IgG1/IgG2a ratio was calculated by dividing the mean values of the highest serum dilution obtained for IgG1 by the mean value of the highest serum dilution obtained for IgG2a without normalization. To determine the avidity of the antibodies, we performed an extra step before adding the secondary antibody. Fixed dilutions (OD<sup>490</sup> = 0.7) of the samples were incubated with 7 M urea or PBS for 5 min. After three washes, the procedure continued exactly as described for the standard ELISA protocol. The avidity index was calculated by the sample's OD<sup>490</sup> × 100 in 7 M urea divided by the OD<sup>490</sup> in PBS.

#### Virus Neutralization Assay and Competition Assays

Viral neutralization was assessed via a flow cytometry based assay adapted from (44). VERO cells (1 × 10 5 cells/well) were cultured in flat-bottomed 96-well plates (Costar) overnight at 37◦C and 5% CO2. Sera from immunized mice were heat inactivated at 56◦C for 30 min. Two-fold serially diluted sera were incubated with virus particles of the DENV2 NGC strain (MOI of 0.1) for 30 min at 37◦C and 5% CO2. The sera/virus mixtures were then incubated with the cells for 1 h at 37◦C and 5% CO2. Next, the sera/virus mixtures were removed and DMEM containing 5% fetal bovine serum (FBS) was added to the cells that were then incubated for 24 h at 37◦C and 5% CO2. Trypsin was used to detach the cells that were resuspended in DMEM 5% FBS and transferred to V-bottomed 96-well plates. Cells were fixed and stained as described previously (45) using 4G2 (10µg/mL; mouse anti-flavivirus envelope antibody) as a primary antibody and anti-mouse IgG-Alexa488 (1:2,000; Life technologies) as a secondary antibody. The cells were resuspended in FACS buffer and 20,000 events were acquired in the BD FACSCaliburTM flow cytometer (BD biosciences). The sera effect on virus neutralization was determined in comparison to a control infection with sera derived from mice injected with saline only.

For the competition assay, adapted from (43), a fixed dilution of the sera from the groups (1:20) was incubated with different molar concentrations of the recombinant EDIII protein for 30 min. The sera/protein mixture was then incubated with the DENV2 NGC strain for 30 min and the experiment continued as described above. Neutralization of infection was determined in comparison to a control infection with DENV2 incubated with recombinant EDIII plus sera from saline injected mice.

#### Imunofluorescence

VERO cells (25 × 10<sup>3</sup> cells/well) were cultured on glass coverslips inside 24-well plates (Costar) overnight at 37◦C and 5% CO2. Infections were performed with MOI of 0.1 with the DENV2 NGC strain for 1 h. After this period, supernatants containing the virus were discarded and the cells were incubated for 24 h in DMEM 5% FBS. Cells were fixed with ice-cold methanol for 5 min and the coverslips were blocked in PBS containing 1% BSA for 1 h at rt. Next, cells were incubated with pooled sera from the different mouse groups, or with the 4G2 mAb (as a positive control), during 1 h at rt, followed by another incubation with anti-mouse IgG-Alexa488 (1:2,000; Life technologies) in the same conditions. Nuclei were labeled with DAPI (1µg/mL) and the images were acquired in a fluorescence microscope (Leica DMI6000B/AF6000, Buffalo Grove, IL, USA) coupled to a digital camera system (DFC 365 FX, Leica) and processed by the Leica Application Suite X (LAS X).

#### Splenocyte Isolation

Two weeks after the last vaccine dose, spleens were removed aseptically and processed exactly as previously described (30). Pools (n = 4; two pools per group) of bulk splenocytes were resuspended in R10 [RPMI supplemented with 10% of FBS (GIBCO), 2 mM L-glutamine (GIBCO), 10 mM Hepes (GIBCO), 1 mM sodium pyruvate (GIBCO), 1% vol/vol nonessential aminoacid solution (GIBCO), 1% vol/vol vitamin solution (GIBCO), 20µg/mL of ciprobacter (Isofarma, Brazil) and 5 × 10−<sup>5</sup> M 2-mercaptoetanol (GIBCO)]. Cell viability and concentration were estimated using the CountessTM Automated Cell Counter (Invitrogen).

#### Peptide Library

A peptide library comprising the DENV 2 E protein (HQ026763, lineage DENV-2/BR0690/RJ/2008) amino acids 161–404 was synthesized by GenScript USA Inc. This library contained 29 overlapping 20-mer peptides that were synthesized with more than 75% purity. Peptides were resuspended in water (10 mg/mL) and stored at −20◦C. For in vitro stimulation experiments, peptides were divided into 3 pools as depicted in **Table 1**.

#### ELISpot

We used the mouse IFN gamma ELISPOT Ready-SET-Go! <sup>R</sup> (eBioscience) to detect IFN-γ producing splenocytes. The procedure was performed according to the manufacturer's instructions. Briefly, ELISpot plates (MAIPS4510; Millipore) were coated with the capture antibody and incubated overnight at 4◦C. The plates were washed twice with PBS and blocked for 1 h with R10 at rt. Splenocytes were incubated in the presence of 2µg/mL pooled EDIII or EDI/II (negative control) peptides for 20 h at 37◦C with 5% CO2. Unpulsed cells were used as controls for each group. After incubation, the plates were washed three times with PBS-T 0.05% and incubated for 2 h at rt with the biotinylated anti-IFN-γ antibody. Following

TABLE 1 | List of peptides derived from the E protein.


\*Bold shows the EDIII amino acid sequence.

another round of washes, the plates were incubated with avidin-HRP for 45 min at rt. Plates were washed three times with PBS-T 0.05% and the spots were developed with the "AEC substrate set" kit (BD biosciences). We used an automated stereomicroscope (KS ELISPOT, Zeiss, Oberkochem, Germany) to count the number of spots. The formula (# of spots in the pulsed well – # of spots in the unpulsed well) was used to calculate the number of IFN-γ producing cells/10<sup>6</sup> cells.

#### Proliferation and Intracellular Staining (ICS)

To analyze T cell proliferation, splenocytes from mice were labeled with carboxyfluoresceinsuccinimidyl ester (CFSE). Briefly, 50 × 10<sup>6</sup> splenocytes were resuspended in pre-heated PBS and labeled with 1.25µM of CFSE for 10 min at 37◦C. Cells were then washed, resuspended in R10 and 3 × 10<sup>5</sup> cells/well were incubated at 37◦C and 5% CO<sup>2</sup> in 96-well round-bottomed plates with 2µg/mL of pooled EDIII or EDI/II (negative control) peptides. Unpulsed cells were used as controls for each group. After 3 days in culture, cells were restimulated with the same peptide pools plus 2µg/mL of αCD28. After 1 h incubation at 37◦C and 5% CO2, 0.5 µg of Golgi Plug (Brefeldin A, BD Pharmingen) was added per well, and the plate was incubated for 12 h at 37◦C and 5% CO2. After the incubation period, the cells were washed with FACS buffer and transferred to V-bottomed 96-well plates. Cells were stained with LIVE/DEAD <sup>R</sup> dye (Life Technologies) and <sup>α</sup>CD4-PerCP (clone RM4-5) for 40 min on ice and in the dark. After 4 washes with FACS Buffer, the cells were fixed and permeabilized using the Cytofix/Cytoperm kit (BD Pharmingen) according to the manufacturer's instructions. The cells were then washed three times with PermWash buffer (BD Pharmingen). The intracellular staining was performed using αCD3-APC/Cy7 (clone 145-2C11), αIFNγ-APC (clone XMG1.2), αIL2-PE (clone JES6-5H4), and αTNFα-PE/Cy7 (clone MP6-XT22) for 40 min on ice and in the dark. The cells were washed twice and resuspended in FACS Buffer. All antibodies were purchased from BD Pharmingen. Flow cytometer readings were carried out with 200,000 events acquired in the BD LSRFortessa flow cytometer (BD Biosciences) and analyzed using FlowJo software (version 9.3, Tree Star, San Carlo, CA). The analysis of proliferating (CFSElow) cells producing different combinations of cytokines (IFNγ, IL-2, and TNFα) was performed with the Boolean gating platform (FlowJo Software). The percentages of proliferating or/and cytokine-producing cells were calculated by subtracting the values obtained with unpulsed cells.

#### Data Analysis

We used the Prism 7 software (GraphPad Software Inc, LA Jolla, CA) for all tests. Statistical differences were considered significant when p ≤ 0.05. One-way ANOVA followed by Tukey's honestly significantly different (HSD) were used for the ELISA data and Two-way ANOVA followed by Bonferroni correction was used for the ELISpot, CFSE and ICS data. The NT<sup>50</sup> values for the neutralization assays were determined with the non-linear regression (curve fit) analysis.

# RESULTS

#### Production of the Recombinant scFvs

The DENV2 EDIII nucleotide sequence (encoding amino acids 297–394) was cloned in frame into plasmids encoding the variable regions of the heavy and light chains of the anti-DEC205 (clone NLDC145) and the isotype control (clone III/10) as previously described (37). **Figure 1A** shows a schematic representation of pscDEC-EDIII and pscISO-EDIII that were then used to transfect HEK293T cells. Western blot analyses of concentrated cell culture supernatants confirmed secretion of scDEC-EDIII and scISO-EDIII by transfected cells (∼46 kDa, **Figure 1B**). To demonstrate that scDEC-EDIII retained the capacity to bind to the DEC205 receptor, CHO cells stably expressing the murine DEC205 receptor were incubated with different concentrations of either scDEC-EDIII or scISO-EDIII. **Figure 1C** shows that only the scDEC-EDIII bound to DEC205 receptor in a concentration dependent manner. Taken together, these results indicate that both scFvs were successfully secreted from transiently transfected cells, and that the scDEC-EDIII preserved its binding capacity to the DEC205 receptor.

### In vivo EDIII Targeting to DCs Improves Antibody Responses

Next, we assessed if immunization with pscDEC-EDIII could improve the anti-EDIII antibody response. For that purpose, mice received three doses of each plasmid administered i.m. followed by electroporation (**Figure 2A**). Twelve days after the administration of the first and second doses, and 14 days after the third dose, mice were bled and the sera were tested individually for reactivity against the recombinant EDIII produced in bacteria. **Supplementary Figure 1** shows an increase in anti-EDIII antibody titers in mice immunized with both plasmids, especially after the administration of the third dose. Moreover, the anti-EDIII antibody titers observed in the animals immunized the pscDEC-EDIII were higher than those observed in mice immunized with pscISO-EDIII (**Figure 2B**). When IgG subclasses present in the sera of immunized mice were analyzed, IgG1, IgG2a, and IgG2b but not IgG3 were detected (**Figure 2C**). Notably, the IgG1/IgG2a ratio detected in mice immunized with pscDEC-EDIII was approximately 10× higher (6.62) than the one obtained in mice immunized with pscISO-EDIII (0.60). When antibody avidity was measured, we noticed that it was higher in sera from mice immunized with pscDEC-EDIII than in mice immunized with pscISO-EDIII (**Figure 2D**). Anti-EDIII antibodies were also tested for binding to the viral E protein by immunofluorescence and, as expected, sera collected from mice immunized with pscDEC-EDIII or pscISO-EDIII reacted with VERO cells previously infected with DENV2 (**Figure 2E**). The labeling patterns were similar to those observed in infected cells stained with 4G2 mAb that recognizes the E protein of Flaviviruses. These results indicate that there are differences in the magnitude, and in the quality of anti-EDIII antibodies raised after immunization with pscDEC-EDIII or pscISO-EDIII.

mAb followed by incubation with goat anti-mouse IgG-Alexa488. Analysis was performed using the FlowJo software (version 9.3, Tree Star).

# Anti-EDIII Antibodies Raised in Immunized Mice Inhibit DENV2 Infection

Anti-EDIII antibodies were shown to be effective to block virus entry into eukaryotic cells (7–9, 46, 47). We then analyzed if anti-EDIII antibodies present in the sera of immunized mice were able to block DENV2 infection. **Figure 3A** shows that antibodies raised in mice immunized with both scFvs plasmids inhibited virus entry with the same efficiency and in a dilution dependent manner. The 50% neutralization titers (NT50) were also similar for sera derived from pscDEC-EDIII (1.66) or from pscISO-EDIII (1.69) immunized mice. To verify if the anti-EDIII antibodies were able to block DENV2 infectivity by binding to EDIII, we performed a competition assay using different concentrations of recombinant EDIII and a fixed serum dilution. Sera derived from mice immunized with pscDEC-EDIII, or pscISO-EDIII were then incubated with increasing amounts of EDIII. We observed a reduction in the sera capacity to inhibit DENV infection as the amount of EDIII was increased (**Figure 3B**). Of note, although we noticed a difference in the slope of the curves between the two groups, no statistical significance was observed, indicating that the anti-EDIII antibodies induced in both groups bound equally well to recombinant EDIII.

These results indicate that immunization with a DC targeted DNA vaccine was able to elicit higher anti-EDIII antibody titers with higher avidity. However, their blocking capacity did not differ from the blocking capacity of anti-EDIII antibodies induced by the vaccine that did not target DCs.

#### DC-Targeted DNA Vaccine Elicits Specific IFNγ Production and CD4<sup>+</sup> T Cell Proliferation

We also investigated if EDIII targeting to DCs would impact cellular immune responses, particularly CD4<sup>+</sup> T cells. Splenocytes harvested 14 days after the administration of the last immunization dose were incubated with three peptide pools containing peptides derived from a library comprising EDI/II and EDIII 20-mer overlapping peptides (**Table 1**). Pool 1 contained 10 peptides restricted to EDI/II domains (not present

in the EDIII sequence encoded by the DNA vaccine plasmids), pool 2 contained five peptides derived from EDI/II and five peptides derived from EDIII sequence, and pool 3 contained nine peptides derived from EDIII amino acid sequence. All three pools were used to stimulate splenocytes from mice immunized with pscDEC-EDIII, pscISO-EDIII or saline. ELISpot assays showed that the number of IFNγ-producing splenocytes derived from mice immunized with pscDEC-EDIII was higher than that observed in mice immunized with the isotype control DNA vaccine or saline (**Figure 4A**). In addition, the response was mainly directed to a peptide(s) contained in pool 3 (**Figure 4A**). There was also a statistically significant difference in the number of IFNγ-expressing cells derived from mice immunized with pscDEC-EDIII after stimulation with peptide pool 2. This result suggests that pools 2 and 3 contain peptides able to bind to the BALB/c H-2K<sup>d</sup> haplotype. When CD4<sup>+</sup> T cell proliferation (representative gating strategy shown in **Supplementary Figure 2**, CFSElow

FIGURE 3 | Antibodies from mice immunized with pscDEC-EDIII and pscISO-EDIII partially inhibit DENV infection by binding to EDIII. (A) Neutralization in VERO cells. Pooled sera from mice immunized as described in Figure 2 were heat inactivated at 56◦C for 30 min. Two-fold serially diluted sera were incubated with the DENV2 particles for 30 min at 37◦C and 5% CO2. The sera/virus mixture was then incubated with the cells for 1 h at 37◦<sup>C</sup> and 5% CO2. The cultures supernatant was replaced by DMEM 5% FBS followed by another incubation at the same conditions for 24 h. Cells were stained with the 4G2 (mouse anti-flavivirus envelope antibody) and anti-mouse IgG-Alexa488. The neutralization of infection was determined in comparison to a control infection. Results are represented by means and SEM from pooled data of four independent experiments. (B) Competition assay with the recombinant EDIII protein. A fixed dilution (1:20) of sera was incubated with increasing molar concentrations of EDIII protein prior to incubation with the virus. The cells were resuspended in FACS buffer and 20,000 events were (Continued) FIGURE 3 | acquired in the BD FACSCaliburTM (BD biosciences) flow cytometer. The neutralization of infection was determined in comparison to a control infection with sera derived from mice injected with saline. Data were analyzed by a two-way ANOVA for repetitive measures followed by Sidak's multiple comparisons test. ns = not-significant. Representative of three independent experiments.

panel) was analyzed, we also detected a higher frequency of CD3+CD4+CFSElow cells in the DC-targeted DNA vaccine group when compared to the groups immunized with the isotype control plasmid or saline (**Figure 4B**). Similarly to the previous experiment, the highest frequency of proliferation was directed against pool 3. As expected, pool 1 (comprising unrelated peptides derived from EDI/II domain) elicited a lower response in mice immunized with scFv plasmids that was not different from the response induced in animals that received saline.

### DC-Targeted DNA Vaccine Induces CD4<sup>+</sup> T Cells That Proliferate and Produce Pro-Inflammatory Cytokines at the Same Time

We next assessed if CD4<sup>+</sup> T cells that proliferated in response to pools 2 and 3 were also able to produce the pro-inflammatory cytokines IFNγ, IL-2, and TNFα (representative gating strategy shown in **Supplementary Figure 3**). As shown in **Figure 5A**, the frequency of CD4<sup>+</sup> T cells that proliferated and produced IFNγ was higher in animals immunized with pscDEC-EDIII pulsed with pools 2 and 3 when compared to animals that received saline. For pool 3, as observed in the ELISpot assay, the frequency of CD4+CFSElow that produced IFNγ was also higher in mice immunized with pscDEC-EDIII than in the group immunized with pscISO-EDIII. We did not observe significant differences among the groups when we compared the frequency of CD4+CFSElow cells that produced IL-2 (**Figure 5B**). TNFα production by proliferating CD4<sup>+</sup> T cells was also higher in cells derived from pscDEC-EDIII immunized mice, especially when they were incubated with pool 3. A higher response against pool 2 was also observed, but it did not reach statistical significance when compared to the other two groups (**Figure 5C**).

The results in **Figure 5** showed that immunization with pscDEC-EDIII induced CD4<sup>+</sup> T cells that proliferated and produced three pro-inflammatory cytokines mainly to peptides contained in pools 2 and 3. To explore this response in more detail, we performed a Boolean analysis of the data (representative gating strategy shown in **Supplementary Figure 2**, CFSElow and cytokine<sup>+</sup> panels), and showed that the CD4+CFSElow cells were polyfunctional and able to produce combinations of the tested cytokines. For example, proliferating CD4<sup>+</sup> T cells from mice immunized with pscDEC-EDIII produced all three cytokines simultaneously when pulsed with pool 2 (**Figure 6A**). Despite not statistically significant, we also observed an increase in the frequencies of CD4<sup>+</sup> T cells that proliferated and produced IL-2/TNFα,

or only IL-2 or TNFα in pscDEC-EDIII immunized animals. The only exception was due to the IFNγ/TNFα double positive cells whose frequency was higher in pscISO-EDIII immunized animals. Similar results were observed in assays using pool 3 (**Figure 6B**). Mice immunized with pscDEC-EDIII showed a statistically significant increase in the frequency of triple positive CD4<sup>+</sup> T cells when compared to the group that received saline. Moreover, the percentage of CD3+CD4<sup>+</sup> cells that proliferated and produced only TNFα was statistically higher in mice immunized with pscDEC-EDIII than in those that received pscISO-EDIII or saline. Despite not significant, we observed that animals immunized with pscDEC-EDIII presented a higher frequency of cells positive for IFNγ/TNFα, only IFNγ, or only IL-2 when compared with the other two groups. Taken together, these results indicate that EDIII targeting to DCs using a DNA vaccine was able to elicit a polyfunctional CD4<sup>+</sup> T cell response.

#### Identification of EDIII-Specific Epitopes Recognized by CD4<sup>+</sup> T Cells Elicited in Mice Immunized With pscDEC-EDIII

In order to identify the peptide(s) present in the pools able to specifically activate CD4<sup>+</sup> T cell responses in mice immunized with pscDEC-EDIII, we performed an ELISpot assay using individual peptides comprising the complete EDIII sequence plus two control peptides derived from EDI/EDII (EIKITPQSSTTEAELTGYGT and STTEAELTGYGTVTMECSPR). Splenocytes from mice

immunized with pscDEC-EDIII were pulsed with each individual peptide and the number of IFNγ producing splenocytes/10<sup>6</sup> total cells was recorded. **Figure 7** indicates that the response was mainly directed to two peptides: MDKLQLKGMSYSMCTGKFKI present in pool 2, and RHVLGRLITVNPIVTEKDSP in pool 3. In addition, we

observed that peptide RHVLGRLITVNPIVTEKDSP represented the immunodominant epitope since the response directed to it was almost three times higher than that detected after stimulation with peptide MDKLQLKGMSYSMCTGKFKI.

#### DISCUSSION

Dengue infection has become a major public health concern as the disease outbreaks and complications have increased substantially in the last five decades (48). Since then, the development of a vaccine has become a global health priority.

The challenge is enormous as dengue is caused by four different serotypes and a previous immune response against one particular serotype can exacerbate the disease caused by another (8). Different approaches are being evaluated and two vaccines based on live attenuated viruses have reached phase III trials: the CYD-TDV by Sanofi Pasteur and the TV003/TV005 by US National Institutes of Health (49). However, results of the CYD-TDV vaccine indicating that the risk of severe disease could increase in seronegative individuals led WHO to recommend that the vaccine would only be administered in populations with dengue serological prevalence rates above 80% (50). In this way, other approaches are currently being developed.

Antigen targeting to DCs through the use of chimeric mAbs has been a promising strategy to induce either humoral or cellular immune responses against different antigens such as: ovalbumin (26–28), Plasmodium yoelii circumsporozoite protein (27), Yersinia pestis LcrV (29), DENV2 non-structural protein 1 (30), Trypanosoma cruzi amastigote surface protein 2 (31), Plasmodium vivax merozoite surface protein 1 (32, 33), and HIV gag (51, 52), among others.

However, the production of such mAbs is time consuming and expensive. DNA vaccines, on the other hand, are cheap, safe and easier to produce and purify, but in general are less immunogenic (35). Different approaches have been developed to increase the immunogenicity of DNA vaccines. Among them are the use of electroporation (53, 54) and the use of plasmids encoding scFv specific for the DEC205 receptor coupled to the antigen of interest.

Although EDIII has been recently used in a DNA immunization strategy (9), in this work we sought to produce a DNA vaccine able to target EDIII from DENV2 to the DEC205<sup>+</sup> DCs in vivo. This was accomplished when we fused the sequence of anti-DEC205 scFv to the EDIII, generating the DNA plasmid named pscDEC-EDIII. As a negative control, we also fused the scFv of a control mAb that was not able to bind to DCs (pscISO-EDIII). Interestingly, a similar construct was engineered by Coconi-Linares et al. that expressed a scFvDEC-EDIII in the plant Nicotiana benthamiana (55). In this case, the authors purified the recombinant scFvDEC205-EDIII protein and immunized BALB/c mice in the presence of anti-CD40 and poly (I:C). Their results showed that the scFvDEC205-EDIII was immunogenic, inducing antibodies with neutralization capacity and proliferating T cells. Nonetheless, a more detailed evaluation of the antibody and T cell responses was not performed.

We decided to use the pscDEC-EDIII as a DNA vaccine for its simplicity and potential to induce strong immune responses when administered together with electroporation. Initially we showed that both plasmids (pscDEC-EDIII or its isotype control) were able to drive the production of chimeric scFvs when transfected in HEK293T cells. More importantly, the scFvs were successfully secreted from the cells, indicating availability in the extracellular medium and possible targeting to the DEC205<sup>+</sup> DCs. Indeed, scDEC-EDIII showed a concentration dependent binding to the DEC205 receptor. Similar results were obtained with other antigens coupled to the same scFvs (37, 56).

Once production and DEC205 receptor specific binding were confirmed for scDEC-EDIII, we used scFv plasmids to immunize mice. We showed that the anti-E antibody titers after the administration of three doses were higher in the group immunized with pscDEC-EDIII when compared to the nontargeted control. Others obtained similar results using different antigens like ovalbumin, HIV gag p41 (37), HER2/neu (38), hepatitis B virus (57), human respiratory syncytial virus (40), and botulinum neurotoxin (39). Although the number of doses varied as well as the amount of plasmid DNA, all these studies used electroporation following intramuscular injection. Interestingly, when we analyzed the IgG isotypes elicited by immunization with the scFvs, we noticed that there was a difference in the IgG1/IgG2a ratio in the sera of mice immunized with pscDEC-EDIII or pscISO-EDIII. Although some groups, using different antigens, also obtained similar results (37, 58), others showed differences when both groups were compared (38). Despite the differences from one study to another, it has become clear that antigen targeting to the DEC205<sup>+</sup> DCs modulates the humoral immune response differently than the non-targeted antigen. We also noticed a significant increase in the avidity of the anti-E antibodies raised in mice immunized with pscDEC-EDIII, while antibodies derived from both immunized groups recognized infected cells. A similar recognition pattern was also obtained after intradermal immunization with a DNA plasmid encoding EDIII (9). This result indicated that the EDIII recognized by the mice sera presented a similar conformation when compared to the EDIII present in the viral particle.

EDIII has been previously described as a target for neutralizing antibodies (46, 47, 59, 60). We then decided to analyze if sera from scFv immunized animals were able to block virus invasion in vitro. The results showed that sera from mice immunized with pscDEC-EDIII or with pscISO-EDIII blocked DENV2 infection in a dilution dependent manner. This result contrasted with our previous results showing higher titers and avidity in the sera of mice immunized with pscDEC-EDIII. A positive correlation between neutralization capacity and higher avidity to the viral particle was observed previously on dengue-infected patients (61). In contrast, other results using HIV envelope proteins showed that higher avidity not always correlates with neutralization capacity (62). More importantly, when sera from these mice were previously incubated with different amounts of recombinant EDIII, we observed a reversion in the sera capacity to block infection. This result more clearly demonstrated that the antibodies directed to EDIII mediate DENV infection inhibition in this model. Similar results were also obtained when an EDIII recombinant protein was administered in the presence of the heat-labile toxin (LT) or its non-toxic B subunit (43), or when a chimeric protein containing EDIII was administered to monkeys (63).

Our group and others described that antigen targeting to the DEC205<sup>+</sup> DCs is a very efficient way of elicit CD4<sup>+</sup> T cell responses (27, 30–34, 51, 64, 65). We then sought to analyze the CD4<sup>+</sup> T cell response induced after immunization with plasmids encoding scFvs genetically fused to EDIII. We took advantage of a peptide library comprising overlapping peptides derived from the E protein amino acids 161 to 404. Our data showed that the pscDEC-EDIII immunization induced CD4<sup>+</sup> T cells that proliferated when pulsed with peptide pools comprising the EDIII portion of the molecule. An analysis of pro-inflammatory cytokines produced by the animals immunized with pscDEC-EDIII showed a higher frequency of IFNγ and TNFα producing CD4<sup>+</sup> T cells when compared to the animals immunized with the isotype control, although IL-2 levels were comparable. Interestingly, there was a difference in the magnitude of the response when splenocytes were pulsed with pools 2 or 3. The frequency of CD4<sup>+</sup> T cells that proliferated and produced IFNγ or TNFα was higher after pulse with pool 3, indicating the presence of an immunodominant epitope(s) capable of binding the BALB/c haplotype (H-2K<sup>d</sup> ). A more detailed analysis showed that pscDEC-EDIII immunization elicited polyfunctional CD4<sup>+</sup> T cells producing one, two, or three pro-inflammatory cytokines, even though statistical significance in comparison to the isotype control group was only reached when single TNFα producers were compared. CD4<sup>+</sup> T cells producing the same combination of pro-inflammatory cytokines were also observed in animals immunized with scDEC-HIV p41 (37) or with scDEC-HER2 (38). The presence of polyfunctional CD4<sup>+</sup> T cells with protective capacity was first described in a mouse model testing a vaccine against Leishmania major (66). After that, many groups set out to investigate if polyfunctional CD4<sup>+</sup> T cells could be related with protection in other models. Studies with HIV-1 infected individuals showed that those displaying polyfunctional CD4<sup>+</sup> T cells were able to better control disease (67, 68). In dengue infection, one study showed that CD4<sup>+</sup> T cells that produced either IFNγ or IL-2 correlated with protection from secondary virus infection in children (69). Polyfunctional CD4<sup>+</sup> T cells were also identified in individuals submitted to a DENV-1 vaccine candidate, although protection against infection was not investigated in this particular study (70).

Worth mentioning is the fact that, in some cases, DNA vaccination with a scFv encoding the scDEC205 fused with an antigen elicited weaker immune responses when compared to the non-targeted controls (56, 58, 71). The reason for these results is still unclear. In fact, DEC205 targeting using chimeric anti-DEC205 mAbs has been known to induce tolerance if the antigen is delivered in the absence of a DC maturation stimulus (72). However, electroporation facilitates DNA uptake and makes more DNA available for detection by intracellular DNA sensors, thereby activating the production of cytokines as an innate reaction (73). In addition, the plasmid backbone should also be considered. As a mechanistic explanation is still elusive, additional studies are necessary.

Finally, we attempted to map the epitopes responsible for the CD4<sup>+</sup> T cell proliferation and cytokine production. When splenocytes from mice immunized with pscDEC-EDIII were pulsed with each individual peptide, we identified two peptides that were probably responsible for the response observed against pools 2 and 3: MDKLQLKGMSYSMCTGKFKI and RHVLGRLITVNPIVTEKDSP, respectively. Peptide RHVLGRLITVNPIVTEKDSP comprises the sequence of peptide RHVLGRLITVNPIVT that was shown to induce IFNγ production when this peptide was used to immunize C57BL/6J mice (74). In addition, the same peptide was also shown to bind to HLA-DRB1<sup>∗</sup> 08:02 in patients from Nicaragua (75) and Sri Lanka (76). This result indicates that our immunization strategy may have the potential to induce CD4<sup>+</sup> T cells in humans. Peptide MDKLQLKGMSYSMCTGKFKI is contained in the sequence of peptide SGNLLFTGHLKCRLRMDKLQLKGMSYSMCTG, which was previously used to immunize BALB/c mice. Proliferation and release of IL-2 were detected in this case (77). As DEC205 targeting greatly improves CD4<sup>+</sup> T cell responses, it is relatively easy to map antigenic peptides. CD4<sup>+</sup> T cell epitopes derived from different proteins have been more easily mapped in samples derived from animals submitted to antigen targeting to DCs. CD4<sup>+</sup> T cells epitopes were detected in the Plasmodium yoelii circunsporozoite protein (27), in the HIV p24 gag (51), in the Leishmania major LACK antigen (64), in the Yersinia pestis LcrV antigen (65) and in the Trypanosoma cruzi ASP-2 protein (31).

The role of CD4<sup>+</sup> T cells in dengue infection is still not very well defined. CD4<sup>+</sup> T cells from infected patients were found to have ex vivo specific cytolytic activity against DENV (13). Individuals vaccinated with an experimental live attenuated DENV1 vaccine exhibited CD4<sup>+</sup> T cells with cytotoxic and proliferative capacities in vitro (78, 79). In mouse models, a DNA vaccine encoding the NS1 protein induced protection via CD4<sup>+</sup> T cells and antibodies (80). Immunization with a chimeric αDEC205 mAb fused to NS1 was also able to induce CD4<sup>+</sup> T cells that contributed for protection (30). Yauch et al. showed that vaccination with CD4<sup>+</sup> peptides in an IFN-α/βR <sup>−</sup>/<sup>−</sup> mouse model reduced viral loads. Interestingly, CD4<sup>+</sup> T cells did not seem to have an impact on antibody neutralization of the virus measured by infection of C6/36 cells (81). Further studies will be needed to address the role of CD4<sup>+</sup> T cells in our model.

Although in the literature CD8α <sup>+</sup> DCs are described as playing an important role in the uptake of apoptotic cells and in antigen cross-presentation in the context of MHC Class I (82, 83), we did not detect a robust CD8<sup>+</sup> T cell response induced by our scFv DNA vaccines (data not shown). One reason for that might be related to the choice of the EDIII as an antigen, as most CD8<sup>+</sup> T cell epitopes are localized on the non-structural proteins, especially NS3 and NS5 (84). Our group has previously demonstrated that NS1 targeting to DCs via DEC205 induced protection partially mediated by CD8<sup>+</sup> T cells (30). In addition, studies mapping CD8<sup>+</sup> T cell epitopes usually use peptides of no more than 12-15 amino acids. Our peptide library consisted of 20-mers, which might have restricted the identification of CD8<sup>+</sup> T cell responses.

Taken together, our results show that antigen targeting to CD8α <sup>+</sup> DCs using DNA vaccines is a promising strategy to induce cellular and humoral responses and may be used in the development of more efficient dengue vaccines.

#### AUTHOR CONTRIBUTIONS

AZ and SB designed the experiments. AZ, FS, BA, HS, NF, NS, and MY conducted most of the experiments. AZ and SB analyzed the data. AZ and SB prepared the figures and wrote the manuscript. LF, DM, and DR contributed reagents. DR, LF, FS, NS, and MY revised the manuscript. All authors read and approved the final version of the manuscript.

#### REFERENCES


# FUNDING

The São Paulo Research Foundation (FAPESP, grant numbers 2013/11442-4, 2014/50631-0, 2014/17595-0, and 2014/15061-8), the Brazilian National Research Council (CNPq, grant number 472509/2011-0) and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior—Brasil (CAPES, Finance Code 001) funded this research. AZ, FS, and BS received fellowships from FAPESP (2016/04477-4, 2015/18874-2, and 2015/16565- 2, respectively). SB, DR, and LF are recipients of CNPq fellowships.

#### ACKNOWLEDGMENTS

The authors would like to thank Danielle Chagas, Anderson Domingos Silva and Doroty Nunes da Silva for assistance in the animal facility, and Dr. Mauro Javier Cortéz Veliz for the use of the immunofluorescence microscope.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.00059/full#supplementary-material


protein of DENV type 2 to DEC-205 receptor elicits neutralizing antibodies in mice. Vaccine (2013) 31:2366–71. doi: 10.1016/j.vaccine.2013.03.009


receptor modulates the cellular and humoral immune response. Int Immunol. (2013) 25:247–58. doi: 10.1093/intimm/dxs112dxs112


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Zaneti, Yamamoto, Sulczewski, Almeida, Souza, Ferreira, Maeda, Sales, Rosa, Ferreira and Boscardin. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Are Conventional Type 1 Dendritic Cells Critical for Protective Antitumor Immunity and How?

Jean-Charles Cancel † , Karine Crozat\* † , Marc Dalod\* † and Raphaël Mattiuz †

CNRS, INSERM, Centre d'Immunologie de Marseille-Luminy, Turing Center for Living Systems, Aix Marseille University, Marseille, France

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Kai Hildner, Universitätsklinikum Erlangen, Germany Patrizia Stoitzner, Innsbruck Medical University, Austria

#### \*Correspondence:

Karine Crozat crozat@ciml.univ-mrs.fr Marc Dalod dalod@ciml.univ-mrs.fr

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 30 September 2018 Accepted: 04 January 2019 Published: 12 February 2019

#### Citation:

Cancel J-C, Crozat K, Dalod M and Mattiuz R (2019) Are Conventional Type 1 Dendritic Cells Critical for Protective Antitumor Immunity and How? Front. Immunol. 10:9. doi: 10.3389/fimmu.2019.00009 Dendritic cells (DCs) are endowed with a unique potency to prime T cells, as well as to orchestrate their expansion, functional polarization and effector activity in non-lymphoid tissues or in their draining lymph nodes. The concept of harnessing DC immunogenicity to induce protective responses in cancer patients was put forward about 25 years ago and has led to a multitude of DC-based vaccine trials. However, until very recently, objective clinical responses were below expectations. Conventional type 1 DCs (cDC1) excel in the activation of cytotoxic lymphocytes including CD8<sup>+</sup> T cells (CTLs), natural killer (NK) cells, and NKT cells, which are all critical effector cell types in antitumor immunity. Efforts to investigate whether cDC1 might orchestrate immune defenses against cancer are ongoing, thanks to the recent blossoming of tools allowing their manipulation in vivo. Here we are reporting on these studies. We discuss the mouse models used to genetically deplete or manipulate cDC1, and their main caveats. We present current knowledge on the role of cDC1 in the spontaneous immune rejection of tumors engrafted in syngeneic mouse recipients, as a surrogate model to cancer immunosurveillance, and how this process is promoted by type I interferon (IFN-I) effects on cDC1. We also discuss cDC1 implication in promoting the protective effects of immunotherapies in mouse preclinical models, especially for adoptive cell transfer (ACT) and immune checkpoint blockers (ICB). We elaborate on how to improve this process by in vivo reprogramming of certain cDC1 functions with off-the-shelf compounds. We also summarize and discuss basic research and clinical data supporting the hypothesis that the protective antitumor functions of cDC1 inferred from mouse preclinical models are conserved in humans. This analysis supports potential applicability to cancer patients of the cDC1-targeting adjuvant immunotherapies showing promising results in mouse models. Nonetheless, further investigations on cDC1 and their implications in anti-cancer mechanisms are needed to determine whether they are the missing key that will ultimately help switching cold tumors into therapeutically responsive hot tumors, and how precisely they mediate their protective effects.

Keywords: conventional type 1 dendritic cells, tumor, type I IFN, CD8<sup>+</sup> T cells, NK cells, immunotherapy, cancer immunosurveillance, clinical trials

# INTRODUCTION

Immune responses against cancer are sculpted by the tumor microenvironment, including its composition in terms of cell types and their physiological states. Indeed, tumors escape host immune defenses not only through decreasing their intrinsic immunogenicity but also by shaping a specific immunosuppressive microenvironment (1, 2). Exogenous factors such as the microbiota and its metabolites also modulate the tumor microenvironment and hence antitumor immune responses (3). According to their degree of infiltration by immune cells and to their capacity to activate antitumor immune responses, tumors have been classified as immunologically "Hot" or "Cold." "Hot" tumors are immunogenic, T cellinflamed, and efficiently rejected by the immune system. They are characterized by the presence of activated CD8<sup>+</sup> cytotoxic T lymphocytes (CTLs), by the expression of T cell-attracting chemokines, and by a type I interferon (IFN-I) transcriptional signature (4). "Cold" tumors lack T cell infiltration, which is correlated with an absence of IFN-I signature and with a poor chemokine production (4). They are ignored by the immune system due to their poor immunogenicity and are very poorly responsive to immunotherapies. We propose to refine this bimodal classification through the addition of two other tumor states, which we called "Icy" and "Warm." We define as "Icy" the tumors that develop potent, active, mechanisms to prevent immune recognition and T cell activation, by inducing a highly immunosuppressive microenvironment very early on during their development. Hence, "Icy" tumors are even more refractory to immune control than "Cold" tumors. "Warm" tumors present an intermediate level of infiltration by "exhausted" CTLs, which have been functionally paralyzed by the local immunosuppressive environment that has been progressively shaped during tumor development. The exhaustion of CTLs is at least in part due to engagement of their immune checkpoint receptors by ligands expressed by the tumor cells themselves or by infiltrating antigen (Ag)-presenting cells. "Warm" tumors are more prone to be controlled by immune checkpoint blockade (ICB) treatments. These monoclonal antibody (mAb) based immunotherapies have revolutionized cancer patient care, by significantly increasing not only overall survival rates but also very long-term remissions for tumor types previously difficult to treat. Despite this major advance, the majority of patients with difficult-to-treat cancers do not respond to ICB. To overcome this issue, it is critical to find additional means of manipulating the microenvironment of the "Cold" or "Warm" tumors unresponsive to ICB, in order to convert them into "Hot" tumors. It should be possible to achieve this by combining ICB with adjuvant immunotherapies able to counteract the other immune escape mechanisms established by these tumors, in order to (i) trigger de novo or enhance T cell infiltration, (ii) enhance cross-presentation of tumor-associated Ag, and (iii) promote a better induction or reactivation of CTL effector functions.

Dendritic cells (DCs) are the most potent Ag-presenting cells, with a unique efficacy for priming naïve T cells and inducing their functional polarization. They are more generally in charge of orchestrating the expansion and functions of T and natural killer (NK) cells in lymphoid and non-lymphoid tissues. Many clinical trials have been performed over the last 25 years to attempt harnessing DC functions for boosting protective antitumor CTL responses in cancer patients (5). Up to now, the results have been disappointingly far below expectations. These failures occurred at least in part because of the almost exclusive use of monocyte-derived DCs (MoDCs) for ACT in cancer patients. Indeed, later advancement of our basic understanding of the heterogeneity and functional plasticity of DCs suggested that other types than MoDCs should be better suited for this purpose (6–8). A relatively recent consensus has emerged on a universal and simplified classification of DC types both in mice and in humans, based on their ontogeny, gene expression programs, phenotype, functions and localization (9, 10). Five major types of DCs can be distinguished: plasmacytoid DC (pDCs), type 1 conventional DCs (cDC1), type 2 cDCs (cDC2), Langerhans cells and MoDCs. In mice, cDC1 encompass both the lymphoid tissue-resident CD8α <sup>+</sup> cDCs as well as the CD103+CD11b<sup>−</sup> cDCs that reside in the parenchyma of non-lymphoid tissues and, once matured upon activation, can migrate to the draining lymph nodes. In humans, cDC1 correspond to the CD141 (BDCA3)high CD11b−/low cDCs. Both mouse and human cDC1 express specifically the chemokine receptor XCR1 and selectively the Ctype lectin endocytic receptor CLEC9A (11). cDC1 can directly enter tissues from the blood, or differentiate locally from a dedicated progenitor, the pre-cDC1 that has been characterized both in the mouse and the human (12, 13). Mouse cDC2 correspond to the CD11b<sup>+</sup> cDCs, and human cDC2 to the CD1c (BDCA1)high CD11b+/high cDCs. For a very long time, MoDCs were the only DC type that could be produced in vitro, in high numbers and under clinical-grade conditions (5, 6, 8). They were therefore used for most immunotherapeutic clinical trials based on adoptive cell transfer (ACT) of in vitro derived autologous DCs. However, MoDCs strikingly differ from cDC1 and cDC2 that are the major types of DCs residing in secondary lymphoid organs and orchestrating immune responses in vivo (14–16). For example, MoDCs do not migrate efficiently to lymph nodes and are particularly prone to develop immunosuppressive functions, whereas cDC1 excel in the activation of CTLs, which are critical effector cell types for antitumor immunity (17). Thus, major efforts have been conducted in the last 10 years to investigate whether cDC1 might be critical for defense against cancer, and how. Here, we are reporting on studies addressing this issue in mice, under experimental conditions of spontaneous immune rejection of tumor grafts in syngeneic recipients, or in preclinical models of immunotherapies. We also summarize human studies that mined large datasets of tumor gene expression profiles to investigate correlations between clinical outcome and digital deconvolution of the tumor immune infiltrate. We discuss how the knowledge generated by these studies can instruct innovative immunotherapeutic strategies to harness cDC1 functions for the benefits of cancer patients.

## NO CURRENTLY AVAILABLE MUTANT MOUSE MODEL IS SPECIFICALLY TARGETING ONLY cDC1 IN VIVO

To determine whether and how a given type of immune cells plays a non-redundant role in antitumor immunity in vivo, it should be specifically and efficiently manipulated in mice. Different mutant mouse models have been generated to either deplete DCs, or inactivate candidate genes in DCs, as recently reviewed (18, 19). Here, we will specifically discuss the use of mutant mouse models to investigate the functions of cDC1 or their molecular regulation (**Table 1**) (14, 20–47). Mouse models expressing the Cre DNA recombinase under the control of the promoter of a gene selectively expressed in DCs have been generated to enable conditional deletion of candidate floxed genes in the targeted cells (e.g., Itgax-Cre targeting CD11c<sup>+</sup> cells and Xcr1-Cre targeting cDC1). Constitutive depletion models have been generated using two types of strategies. The first corresponds to the knock-out of a transcription factor shown to be crucial selectively for the development/homeostasis of cDCs (Zbtb46) or cDC1 (Batf3) (**Table 1**). The second consists in ectopic expression of the active subunit of the diphtheria toxin (DTA) selectively in DCs (e.g., Xcr1-Cre;Rosa26-LSL-DTA mice for cDC1, **Table 1**). Conditional depletion can be achieved upon diphtheria toxin administration in mutant animals engineered for ectopic expression of the gene encoding the human diphtheria toxin receptor (hDTR) selectively in DCs (e.g., Karma-hDTR or Xcr1-hDTR mice for cDC1).

One major caveat of using CD11c for targeting DCs is that the gene encoding this molecule, Itgax, is expressed by other immune cell types, including some that play critical roles in anti-tumor immunity, such as NK cells, effector memory CTLs, intraepithelial lymphocytes (IELs), plasmablasts, and subsets of monocytes or macrophages (32). Knock-in within the Zbtb46 gene has been used to target all cDCs. However, this gene is also expressed by endothelial cells and committed erythroid progenitors (14, 32–34). Since angiogenesis critically affects solid tumor development, experiments should be performed using bone marrow chimera mice generated by engrafting mutant bone marrow cells into a wild type (WT) recipient animal. Batf3−/<sup>−</sup> mice have been the most frequently used model to investigate whether cDC1 play a critical role in physiological processes. However, even in this model, complementary strategies are needed before drawing final conclusions, because Batf3 is also expressed in cDC2 and effector CD4<sup>+</sup> T cells, and because it represses Foxp3 expression in CD4<sup>+</sup> T cells leading to increased numbers of regulatory T cells (Treg) in knock-out mice (35, 36). In addition, the impact of Batf3 inactivation on cDC1 homeostasis is less efficient in the C57BL6/J genetic background than in the 129svEv one. Under inflammatory settings, the knock-out of Batf3 can be compensated for cDC1 development, by the induction in DC precursors of the paralog genes Batf and Batf2 (37–39). We have engineered mutant mouse models for cDC1 targeting based on the knock-in of Cre (42) or hDTR (43) into the Gpr141b (alias A530099j19rik or Karma) gene, but these models also target mast cells (42). Finally, the Xcr1 gene was targeted to generate mutant mouse models for specific, conditional or constitutive, cDC1 depletion, as well as for their genetic manipulation (42, 44, 46). The Xcr1 gene is preserved in our models (42). In contrast, it is knockedout in the other ones (46); hence, only heterozygous mice should be used for these models in order to avoid possible phenotypic effects due to a complete XCR1 deficiency. Besides cDC1, only a minute proportion of CD4<sup>+</sup> T cells are targeted in Xcr1-Cre mice (42). Although still imperfect, the mutant mouse models based on the manipulation of the Xcr1 gene are the best to target cDC1 in vivo. In conclusion, none of the mutant mouse models used to date for cDC1 targeting are entirely specific and efficient, but some are better suited than others for this purpose. In any case, it is always important to use complementary methods to ensure that the phenotypes observed are only or mostly due to the manipulation of cDC1. For example, depleted mice should be replenished with wildtype cDC1 if possible. Alternately, results should be confirmed in other mutant models also targeting cDC1 but no other cell types in common.

#### THE ROLE OF cDC1 IN CANCER IMMUNOSURVEILLANCE HAS NOT YET BEEN INVESTIGATED

Cancer development is a multistep process consisting in the accumulation of genetic mutations within a cell leading to increased or deregulated proliferation and survival, with clonal selection of neoplastic progeny (48). There is a strong contribution of the host immune responses in this dynamical process of tumor selection, which has been described as the three E of cancer immunoediting: Elimination, Equilibrium and Escape of cancer cells (49). A failure of the immune system to eliminate all transformed cells early during their development is followed by an equilibrium state during which the immune system exerts a relentless pressure on surviving tumor cells, ultimately leading to tumor escape from the exhausted immune system. The initial elimination phase is therefore critical to restrict tumor growth very rapidly to prevent relapse or metastasis. Efficient recognition and elimination of transformed cells implies constant monitoring of the body by both the innate and adaptive immune systems, a process called cancer immunosurveillance. Upon monitoring spontaneous, carcinogen- or genetically-induced tumor development in mice bearing various immune deficiencies, critical roles in cancer immunosurveillance have been uncovered for αβ and γδ T cells, NKT cells and NK cells, as well as for the cytokines IFN-γ, IFN-I, IL-12 and for the cytotoxic effector molecules Perforin and TRAIL (50). The role of cDC1 in cancer immunosurveillance remains to be assessed. However, a wealth of data has accumulated on their role in the spontaneous immune rejection of tumor grafts in mice, a popular surrogate model for immunosurveillance (**Table 2**) (35, 51, 54–58).

#### TABLE 1 | Mouse models to deplete DCs, cDCs or cDC1 in vivo.


Bold: First publication. \*Mouse models expressing the Cre DNA recombinase under the same gene promoter have been generated. \*\*Xcr1-Cre;Rosa26-LSL-hDTR mice; \*\*\*Xcr1- Cre;Rosa26-LSL-DTA mice.

#### BATF3−/<sup>−</sup> MICE FAIL TO REJECT SYNGENEIC TUMOR GRAFTS, SUGGESTING A CRITICAL ROLE FOR cDC1 IN SPONTANEOUS ANTITUMOR IMMUNE DEFENSES

Most tumor cells are not able to directly prime naïve T cells, due to their low expression of MHC class I and co-stimulation molecules or to their acquisition of immunosuppressive functions such as high expression of ligands for immune checkpoint receptors. Thus, induction of CTL responses against most tumors requires accessory cells able to take-up, process and present exogenous tumor Ag in association with MHC-I molecules, a process known as cross-presentation. cDCs are highly efficient in initiating and globally orchestrating adaptive immunity, due to their professional capacities to simultaneously deliver all necessary signals to T cells, namely

Ag presentation as signal 1, co-stimulation as signal 2, and cytokines as signal 3 (59). Mouse and human cDC1 excel at activating CTLs, due to their higher capacity to cross-present cellular Ag as compared to other types of Ag-presenting cells (11). It seemed therefore logical that cDC1 should play a critical role in anti-tumor immunity (**Table 2**). Kenneth Murphy's group was the first to confirm this hypothesis, by showing loss of spontaneous rejection of transplantable tumors in Batf3−/<sup>−</sup> mice (35). Contrary to their WT counterparts, Batf3−/<sup>−</sup> cDCs failed to induce proliferation of OT-I cells when co-cultured with cells loaded with the OVA protein, suggesting that crosspresentation of cellular Ag indeed constitutes one of the critical, non-redundant functions of cDC1. Several other studies have since reported similar results, altogether using a variety of transplantable tumors (**Table 2**). These studies strongly support a critical role for cDC1 in spontaneous antitumor immune defenses. However, a possible role of the loss of Batf3 expression in cDC2 or in effector T cells has not been ruled out. In addition,



Bold: NK cell dependent rejection. Underlined: NK cell independent rejection. Fibrosarcoma: 1969, 1773RS100, 7835, d38m2, d42m1, F515, GAR4.GR1, H31m1, MC-57.SIY. Melanoma: B16.SIY, Ptgs1/Ptgs2−/<sup>−</sup> BRAFV600<sup>E</sup> . Mastocytoma: P198. 129/SvEv background: 1773RS100, d38m2, d42m1, F515, GAR4.GR1, H31m1. C57BL6/J background: 1969, 7835, B16.SIY, MC-57.SIY, Ptgs1/Ptgs2 <sup>−</sup>/<sup>−</sup> BRAFV600<sup>E</sup> . DBA/2 background: P198.

Batf3−/<sup>−</sup> mice can still achieve partial tumor control and mount tumor-specific CTL response under low-dose tumor challenge (35, 51, 54), which might be explained either by the incomplete cDC1 loss or by partial redundancy between cDC1 and other cell types for the cross-presentation of cellular Ag and the induction of antitumor adaptive immunity. Further studies are warranted to address these issues.

### INSIGHTS INTO HOW cDC1 COULD PROMOTE PROTECTIVE SPONTANEOUS ANTITUMOR IMMUNITY

#### Cross-Presentation by cDC1 Is Necessary but Not Sufficient for Immune Control of a Regressor Fibrosarcoma

The importance of cross-presentation in cancer immunology has been extensively reviewed (60). Very recently, the WDFY4 molecule, a member of the BEACH (Beige and Chediak-Higashi) domain–containing family of proteins, was reported to be specifically required for cross-presentation of cell-associated Ag by cDC1, and for cDC1-dependant immune control of the highly immunogenic 1969 regressor fibrosarcoma (52). The demonstration of a cell-intrinsic requirement of WDFY4 in cDC1 for immunity against cancer was achieved by comparing tumor growth between Wdfy4−/−:WT vs. Wdfy4−/−:Batf3−/<sup>−</sup> mixed bone marrow chimera mice. Importantly, Wdfy4−/<sup>−</sup> cDC1 were not compromised in their abilities to produce IL-12 and to present Ag in association with MHC class II molecules for CD4<sup>+</sup> T cell activation. Wdfy4-deficient cDC1 appeared to be selectively impaired in their ability to crosspresent Ag but not in other functions also required for efficient CTL priming and expansion. To our knowledge, this study is the first to demonstrate that a specific defect in cDC1 cross-presentation in vivo leads to a failure of mice to control spontaneously tumor growth. Further studies are warranted to confirm these data and extend it to other preclinical tumor models. A major role of Batf3 in cDC1 is to sustain their expression of Irf8. Consistent with this, the development of cDC1 and their ability to cross-present cellassociated Ag are rescued in Batf3−/<sup>−</sup> animals transgenic for Irf8. Nevertheless, these mice still fail to control the growth of a regressor fibrosarcoma, likewise to Batf3−/<sup>−</sup> animals. Thus, in addition to cross-presentation, other functions of cDC1 are also necessary for the promotion of protective antitumor immunity but remain to be identified (53). Moreover, both in mice and humans, cDC2 and pDCs can also perform crosspresentation of cell-associated Ag, under specific conditions of stimulation, less efficiently than cDC1 (11). Hence, we propose that cDC1 play a critical role in antitumor immunity not only due to their strong cross-presentation activity but rather because they uniquely combine several key features that are not simultaneously expressed together in other cell types, as detailed below (**Figure 1**).

#### Proposed Key Features Underlying cDC1 Non-redundant Role in Anti-tumor Immunity

First, the expression of XCR1 and CCR5 by cDC1 may enable their local recruitment by cytotoxic lymphocytes producing the ligands for these chemokine receptors, XCL1 and CCL4/5 (45, 55, 61–63). Second, reciprocally, cDC1 ability to produce high levels of CXCL9/10 may promote local recruitment of effector and memory CTLs expressing CXCR3 (43, 57, 64). Third, cDC1 can deliver positive co-stimulation signals. Fourth, cDC1 are a major source of IL-12, IFN-β, and IL-15, thereby promoting the survival and proper activation of NK, NKT cells and CTLs (43, 65–69). In a model of lung metastasis, cDC1 were the major source of IL-12, which was critical to control metastasis in a NK cell- and IFNγ-dependent manner (66). Fifth, cDC1 can promote Th1 induction (70–72) and favor CD4<sup>+</sup> T cell help delivery to CTL through simultaneous presentation of Ag in association to MHC-I and MHC-II (73, 74). Depending on the cues that they receive during their activation at the time of Ag processing and presentation, DCs will polarize into different functions during their maturation (75, 76). At steady state, during their homeostatic activation, DCs acquire the ability to induce immune tolerance by causing the death, anergy or polarization into regulatory functions of self-reactive T cells, a process referred to as DC tolerogenic maturation. On the contrary, in proper activating contexts, DCs undergo an immunogenic maturation by acquiring the combined expression of activating co-stimulation molecules and cytokines leading to the induction of strong Ag-specific effector lymphocyte responses. The immunogenic maturation of cDC1 is promoted by IFN-I (51, 54, 68, 75), including through cross-talk with pDCs as a major source of these cytokines (77). Cell-intrinsic responses of cDC1 to IFN-I appear to be critical for spontaneous tumor rejection by enhancing their cross-presentation capacity (51, 54), and perhaps also their trans-presentation of IL-15 which promotes the proliferation and effector differentiation of CTLs

(68). However, cross-presentation was not totally abolished in Ifnar1−/<sup>−</sup> DCs (51, 54). Upon exposure to high doses of Ag in vitro, cross-presentation was even as efficient in Ifnar1−/<sup>−</sup> DCs as in WT DCs. Although, in spontaneously rejected tumor grafts, the cellular source of IFN-I was identified as expressing CD11c, IFN-β production was not altered in Batf3−/<sup>−</sup> mice. Further investigations are required to assess the roles of different types of DCs in CTL activation and in the production of, or responses to, IFN-I, during spontaneous tumor control. In summary, cDC1 constitute a versatile and efficient platform for CTL activation by uniquely bridging several components of innate and adaptive immune responses in a manner promoting mutually beneficial cross-talk (**Figure 1**). However, further studies are warranted to determine whether the different mechanisms detailed above are each critical for the protective antitumor functions of cDC1, as well as their respective importance.

#### When and Where Are cDC1 Functions Exerted During Cancer Immunosurveillance?

Intra-tumoral cDC1 have been suggested to be crucial for in situ maintenance of the effector functions of pre-activated CTLs (65). cDC1 promote memory CTL recall upon secondary infections (43). In an experimental model of established immune memory, only tumors that could be infiltrated by both cDC1 and CTLs were spontaneously controlled (57). While conditional cDC1 depletion was not performed in these

antitumor immune responses. DAMPs, danger associated molecular patterns; F-Actin, filamentous actin; ICD, immunogenic cell death; MDSCs, Myeloid-derived

suppressor cells; MoDCs, Monocyte-derived dendritic cells; TAMs, Tumor associated macrophages; Tregs, Regulatory T cells.

settings to functionally confirm the importance of cDC1 for the reactivation of antitumor CTLs, this point was addressed in another study examining the reactivation of adoptively transferred antitumor central memory CTLs into WT vs. Batf3−/<sup>−</sup> recipient mice (78). In conclusion, cDC1 might not only be required for the initiation of adaptive immunity against intracellular pathogens or tumors but all along the life cycle of CTLs, including for their maintenance in the tumor as well as for the generation and recall of memory to prevent relapse or metastases.

Several studies have suggested that T cell priming in the tumor draining lymph node is required to mount anti-tumor immunity (79, 80) A study showed that tumor-associated cDC1 bearing intact tumor Ag traffic to the draining lymph node to prime naïve CTLs in a CCR7-dependent manner (80). However, Ccr7 knock-out had little impact on tumor growth (80). Moreover, CTL priming, activation, proliferation and effector function acquisition in tumor was observed when T cell egress from lymph nodes was blocked (81) or in mice lacking lymph nodes (82). Although these experimental settings could alter cDC1 and lymphoid cell trafficking (83, 84), they nevertheless show that the activation of antitumor adaptive immunity can occur directly at the tumor site (82), possibly in tertiary lymphoid structures developing locally (85). In any case, for efficient tumor rejection without relapse or metastases, systemic immunity is likely important in addition to in situ responses, as recently appreciated in the context of immunotherapy (86).

#### Proposed Model of cDC1 Role in Antitumor Immunity

Based on the knowledge discussed in the previous sections, we propose a putative model of the mechanisms through which cDC1 promote the rejection of syngeneic tumor grafts in preclinical mouse models (**Table 2**) and may physiologically contribute to cancer immunosurveillance (**Figure 2**). cDC1 take up cell-associated Ag in the tumor after immunogenic cell death, undergo immunogenic maturation, and traffic to the tumor-draining lymph node. There, cDC1 prime naïve CTLs and polarize them toward protective effector functions. CTLs expand and migrate to tumor, where they can be attracted by chemokines secreted locally by cDC1. The tumor-associated cDC1 also sustain infiltrating CTL protective functions (expansion, maintenance and memory recall), and might also prime naïve CTLs in situ.

#### FAILURE OF IMMUNOSURVEILLANCE: ARE cDC1 DIRECT TARGETS OF TUMOR ESCAPE MECHANISMS?

We propose a classification of tumors (Hot/Warm/Cold/Icy) according to their immunogenicity, their cDC1 infiltration, maturation and phenotype, and the characteristics of the antitumor CTL response.

"Hot" tumors are characterized as strongly infiltrated by effector CTLs. They are spontaneously controlled by the immune system (**Figure 2**). They include the syngeneic cancer cell lines used to study spontaneous rejection of tumor grafts (**Table 2**).

"Warm" tumors express tumor neoAg and are infiltrated by cDC1 and CTLs (**Figure 3** Right). Experimental studies in mice suggest that the correlation between high CTL numbers and increased cDC infiltration in tumors is due to a positive feedback loop between these two cell types mutually promoting their local recruitment and survival. It is not clear how this process is initiated, i.e., which cell type is recruited first to the tumor site. This might depend on the combination of tumor type and host characteristics. "Warm" tumors are ultimately not controlled by the immune system, due to their late selection for harboring immune escape mechanisms, such as intrinsic impairment of Ag processing and presentation (87, 88) or induction of CTL exhaustion (89). In these tumors, cDC1 could have undergone immunogenic maturation but may present Ag to CTL in a manner contributing to their chronic activation and exhaustion, e.g., through engagement of checkpoint receptors such as PD-1 or CTLA4.

"Cold" tumors are weakly immunogenic and poorly infiltrated but induce some level of adaptive immunity (**Figure 3** Top). In such tumors, cDC1 could also be direct targets of immune escape mechanisms, such as local production of factors inhibiting DC differentiation or promoting tolerogenic over immunogenic maturation. Those factors include TGF-β, IL-10, IL-6, CSF-1, and VEGF (90). Although cDC1 are proposed to contribute to central and peripheral tolerance (75, 91, 92), whether they can be hijacked by tumors to promote local immunosuppression has not been rigorously investigated.

"Icy" tumors are not immunogenic per se, are not infiltrated by T cells and fail to induce immune responses (**Figure 3** Left). Those tumors have evaded or hijacked innate immunity in a manner preventing immune cell infiltration at very early stages of the cancer immunoediting process (87). cDC1 can be direct targets of these early immune escape mechanisms. In melanoma, the WNT/β-Catenin signaling pathway prevents the recruitment of cDC1 and CTLs into the tumor, at least in part by inhibiting the local production of CCL4 and CXCL9 (57, 93). CCL4 contributes to the recruitment of cDC1 through their CCR5 chemokine receptor (93). CXCL9 helps promoting the recruitment of both pre-cDC1 (94) and memory/effector T cells (57), through CXCR3. Another mechanism of evasion of innate immunity by melanoma is tumor-intrinsic elevated COX activity leading to PGE<sup>2</sup> production and downstream inhibition of NK cell, cDC1 and CTL infiltration (55, 56), by disrupting the XCL1/XCR1 and CCL5/CCR5 chemotactic axes. Impairment of CTL infiltration into the tumor is proposed to occur downstream of the failure of cDC1 recruitment (55, 56).

In brief, cDC1 are direct targets of tumor escape mechanisms since the tumor microenvironment can modulate all of the processes necessary to promote their protective antitumor functions. It can determine the tolerogenic vs. immunogenic nature of tumor cell death (95–97), control the expression of the growth factors and chemokines promoting local recruitment, differentiation, expansion and survival of cDC1 or their progenitors (55, 93, 98), dampen cDC1 production of activating

cytokines (56, 67), and inhibit their maturation or even polarize it toward tolerance (99, 100).

#### STUDIES OF THE NATURAL ROLE OF cDC1 IN IMMUNOTHERAPIES

In the last two decades, cancer treatments have successfully shifted from only targeting the cancer itself to also manipulating the immune system, with the aim to boost or induce de novo protective antitumor cellular immune responses, mainly CTLs but also NK and NKT cells. These novel treatments called immunotherapies encompass different strategies. Here we will specifically discuss studies performed in experimental settings mirroring the two types of immunotherapies that have shown the best clinical benefits in cancer patients. First, we will focus on treatments providing exogenous effector cells through ACT of autologous antitumor CTLs, after their expansion and activation in vitro (eventually combined with genetic engineering for CAR T cells), i.e., CTL ACT. Second, we will discuss mAb

functions (expansion, maintenance, and memory recall). They might also prime naïve CTLs in situ. TdLN, Tumor draining lymph node.

immunomodulation (mAIM) to block checkpoint receptors on CTLs or NK cells (101–106), i.e., ICB, which has proven more efficient than conventional chemotherapies or radiotherapies in several cancer types, with better overall responses, and, most strikingly, significantly increased long-term survival (107). ACT or ICB monotherapy promotes durable disease control only in 30% of the patients. While they dramatically improve the response rate in patients with metastatic melanoma, ICB bitherapies cause significant adverse effects and toxicities, with high incidences of autoimmune manifestations (108, 109).

COX1/2, Cyclo-oxygenase 1/2; iTreg, induced regulatory T cell; mAb, monoclonal antibodies; PGE2, Prostaglandin E2.

Understanding the mechanisms controlling responsiveness to ACT or ICB is thus a prerequisite before complementing these immunotherapies by adjuvant treatments able to further improve the rate and duration of remission for cancer patients. One hypothesis to explain patient non-response to immunotherapies is an impairment of the accessory cells needed to promote CTL reactivation and to sustain their effector functions, rather than cell-intrinsic defects in the CTLs themselves. In this scenario, cDC1 are likely candidates, based on their critical role in promoting the spontaneous rejection of tumors in preclinical mouse models, and on their unique functional features endowing them with a high efficiency for nurturing cytotoxic cells all along their life cycle.

# Role of cDC1 in Promoting CTL ACT Efficacy

The ACT procedure the most commonly used so far consists in isolating endogenous CTLs from a cancer patient, expanding them in vitro through tumor Ag-specific re-stimulation under conditions allowing reversal of exhaustion, and then re-infusing them into the host. By using autologous cells for the treatment, this strategy alleviates any side effects that could arise in allogenic settings. However, one major issue is that only few cancer types respond to this treatment. This might be due to immune escape mechanisms in the tumor limiting locally CTL access to the activating signals necessary to prevent their exhaustion and promote their proliferation, sustained activation and survival. Preclinical mouse models have been used to address this issue, aiming at determining whether professional Ag cross-presentation in the context of positive co-stimulation and delivery of specific cytokines is necessary for ACT efficacy. Since cDC1 excel at this combination of functions (**Figure 1**), they could promote ACT efficacy. Indeed, injection of diphtheria toxin in ACT recipient Zbtb46-DTR mice significantly decreased their response to immunotherapy. cDC1 but not cDC2 from tumor-engrafted control mice were shown to cross-present tumor Ag and produce IL-12 ex vivo. Thus, it was concluded that cDC1 are necessary for ACT efficacy in these experimental settings (65). However, opposite results were recently reported under similar experimental conditions, showing a lack of cDC requirement for ACT success (110). Differences between the experimental set-up of these two studies might explain their different conclusions, since only the second study used bone marrow chimera mice rather than directly Zbtb46-DTR animals, which is necessary to rule out any impact of loss of Zbtb46 expression in other cells than cDCs (33). Therefore, additional studies are necessary to determine whether cDCs are required for maximal ACT efficacy, and how. If those studies unravel specific pathways that can be potentiated, this could allow designing of a "DC adjuvant" therapy for ACT, which might broaden its success rate to more patients and for additional cancer types.

#### Role of cDC1 in Promoting Responses to mAIM

In the course of a normal immune response, Ag-presenting cells regulate their expression of ligands for T cell activating vs. inhibitory co-receptors. This contributes to fine tune the intensity and kinetics of the adaptive immune response, in order to balance efficient immune control of pathogens with the risk of developing an immunopathology due to an excessive T cell activation. Tumors can hijack this process by expressing ligands for T cell inhibitory receptors leading to premature termination/exhaustion of CTL responses (48, 107). This tumoral immune evasion strategy can be overcome by mAIM through infusion of mAbs capable of either inhibiting the engagement of T cell inhibitory co-receptors (i.e., ICB mAbs) or mimicking the engagement of T cell activating co-receptors (co-stimulation activating mAbs) [listed in (106)]. These mAbs can be used as monotherapy or bi-therapy. The ICB mAbs the most commonly used in clinics are directed against programmed cell death protein 1 (PD-1) and cytotoxic T-lymphocyte associated protein 4 (CTLA-4). Although their use has dramatically improved patient survival for different types of cancer, their precise mode of action is still a matter of debate. The mechanisms underlying lack of response in the majority of patients remain elusive. Preclinical mouse models have been used to address this issue and showed that treatment efficacy is abrogated in cDC1-deficient Batf3−/<sup>−</sup> animals (**Table 3**) (56, 57, 65, 69, 117, 118). This is the case for anti-CTLA4 (117) or anti-PD-L1 (79) monotherapies, for a bi-therapy combining anti-PD-1 and CTLA4 mAbs (69), and for a bi-therapy combining the anti-PD-1 mAb with the co-stimulation activating anti-CD137 mAb (118). However, these studies did not determine whether the lack of mAIMdependent tumor control in Batf3−/<sup>−</sup> mice was due to a lack of antitumor CTL priming at the time of tumor engraftment, before immunotherapy, or to a failure of mAIM at inducing the reactivation of previously primed but exhausted antitumor CTLs, at the time when the immunotherapy was administered. Efficient activation of anti-tumor CTLs, for proliferation and acquisition of effector functions, requires cross-presentation of tumorassociated Ag, activating co-stimulation and delivery of specific cytokines from accessory cells. As described previously, cDC1 excel at simultaneously delivering all these signals (**Figure 1**). In particular, one of the critical functions of cDC1 during mAIM immunotherapies may be to deliver IL-12 (69). In addition, cDC1 may also be a major source of CXCL9/10 (43) for recruiting activated or memory CTLs into the tumors (57, 93) (**Figures 1**, **2**).

#### Current Limitations, Controversies or Unknowns

Many of the conclusions drawn above are based on the use of Batf3−/<sup>−</sup> mice, or on the assumption that cDC1 are the main source of the cytokines or chemokines promoting response to mAIM therapies, without formal demonstration of this point by functional inactivation of candidate functions selectively in cDC1. Moreover, the respective importance of Ag cross-presentation vs. delivery of specific activating co-stimulation or cytokine signals by cDC1 has not been delineated yet under immunotherapies condition. Thus, further studies are required to confirm and extend these analyses, by using other mutant mouse models allowing specific cDC1 depletion or selective manipulation of each of their candidate functions.

#### HARNESSING cDC1 FUNCTIONS TO IMPROVE IMMUNOTHERAPIES AGAINST CANCER

In parallel of developing immunotherapies to directly boost lymphocyte effector responses against tumor cells, the



MOA : Mode of action.

**344**

community has put much effort in trying to elaborate vaccines to ignite or reactivate endogenous antitumor immune responses in patients. Among all Ag presenting cells identified so far, cDC1 are the only ones to express selectively unique cell surface markers, such as CLEC9A or XCR1, which enables their specific targeting with mAb in vivo. Intratumor injection of bone marrow-derived DCs highly enriched in cDC1 increased local CTL infiltration and improved response to ICB (57). Therefore, in combination with other immunotherapies, cDC1 represent a very good candidate immune cell type to mobilize with off-the-shelf compounds for boosting patient antitumor immunity.

## Specific Targeting of cDC1 for Vaccination Purposes

Many preclinical studies in various mouse models have demonstrated the efficacy of in vivo targeting of Ag specifically to cDC1 in combination with the administration of a proper adjuvant for priming or reactivating adaptive immunity, leading to a rapid yet long term immune protection against infections by intracellular pathogens or against tumors (119) (**Table S1**). Adjuvants used to induce a beneficial inflammation promoting an immunogenic environment to prevent or counterbalance tumor immunosuppressive functions include the Toll-Like-Receptor ligands LPS, Imiquimod, CpG or Poly(I:C). Other adjuvants include drugs which directly stimulate accessory lymphocytes, such as αGalCer for NKT cell activation (120, 121), or agonistic anti-CD40 antibodies which mimic the helper signal delivered by CD4<sup>+</sup> T cells to DCs for promoting their production of the lymphocyte activating cytokines IL-12 and IL-15/IL-15Rα (68, 122, 123). Vaccine formulation including naked DNA (124, 125), porous polymer matrices (126), or oil in water nano-emulsion (127) are intrinsically immunogenic. Vaccination based on macroporous polymer matrices encapsulating tumor lysates, GM-CSF and CpG, were quite effective in attenuating tumor growth (126), although not targeting specifically cDC1. DEC-205 has been by far the cell surface marker the most used to target cDC1 in vivo (**Table S1**). However, it is not specific of cDC1 since it is expressed on Langerhans cells in the epidermis, on all migratory DC in lymph nodes (128) and it is highly upregulated on various DC subsets in tumors (30). The same issue applies to CD40. This raises the question of the respective roles of cDC1 vs. other types of DC in the protection conferred by vaccines based on in vivo Ag delivery through DEC-205 or CD40. Indeed, tumor Ag delivery to pDCs or cDC2 by using anti-BST2 (129) or anti-DCIR2 mAb (130) respectively, or administration of tumor Ag-pulsed pDC (131), are highly efficient in conferring protection against cancer. This shows that not only cDC1 but also cDC2 or pDCs can induce protective antitumor immunity, providing that Ag is delivered to these cells through adequate endocytic receptors in the presence of proper maturation signals. Interestingly, immunization with tumor-associated exogenous cDC1 or cDC2 prior to tumor engraftment revealed complementary functions of these two DC types (30). In a model of challenge with Lewis Lung carcinoma, only cDC2 vaccination led to reduced tumor growth rate and weight, correlating with reduced tumor infiltration by myeloid-derived suppressor cells, functional polarization of tumor-associated macrophages toward a M1 like antitumor phenotype, and promotion of Th17 rather than Treg CD4<sup>+</sup> T cell responses (30). cDC1 were confirmed to be more efficient than cDC2 for the induction of antitumor CTL responses, which protected against a challenge with B16 melanoma (30).

In summary, even though other DC types can be successfully harnessed for cancer vaccines in mouse preclinical models, many studies showed that in vivo targeting of cDC1 is highly efficient for the activation of antitumor CTL responses able to induce complete tumor rejection in prophylactic settings and to delay significantly tumor progression or metastasis in therapeutic settings (**Table S1**). The efficacy of DC-targeted vaccines depends on three critical parameters: (i) the mode of delivery of the Ag, (ii) the nature of the Ag, and (iii) the nature of the adjuvant. Targeting Ag to cell surface receptors trafficking into late endosomes or lysosomes promotes more efficient cross-presentation by human cDC1 as compared to cDC2, whereas both cell types can mediate this function upon Ag delivery to early endosomes (132). The route of administration of the vaccine should be carefully determined depending on the necessity to induce mucosal and/or systemic immunity according to the type of cancer involved (86, 133). Once activated, tumorassociated DCs have the capacity to migrate to tumor-draining lymph nodes to prime T cells (80) and may rather favor a local antitumor immunity. There is also evidence that CTL priming can occur directly in the tumor (82). Hence, intra- or peritumoral administration of cDC1-targeted Ag for solid tumors may be the best way to enhance priming of CTLs both inside the tumor, and through migration of tumor-associated DCs to the draining lymph node. The tumor Ag should be well selected as the immune system can be almost irreversibly tolerized against certain self Ag (134). Some adjuvants are more efficient in promoting a beneficial inflammatory microenvironment in the tumor, linked to their ability to induce IFN-I. It might be desirable to include adjuvants that directly engage cDC1 since exposure to inflammatory mediators in the absence of direct signaling by pattern recognition receptors might not be sufficient to promote immunogenic DC maturation (135).

#### Mobilizing cDC1 Functions in Combination Immunotherapies

In most of the preclinical models discussed above, cDC1 targeting therapeutic vaccines delay tumor progression or metastasis, or even promote a better tumor control over a long time, but fail by themselves in inducing complete tumor rejection. However, combining strategies mobilizing cDC1 with current immunotherapies, in particular with ICB, should promote the induction of long lasting protective antitumor immunity in more patients, and should more generally improve the objective response rate, the response duration and the overall survival of patients. In preclinical mouse models of immunotherapies, the antitumor effects of various off-the-shelf treatments were shown to require cDC1 functions (**Table S1**), and a variety of strategies were specifically designed to harness cDC1 against cancer (**Table 4**). Hereafter, we discuss how these studies advanced our understanding of when, where and how to mobilize cDC1 functions in combination immunotherapies.

Upon immunotherapy, cDC1 are increased in the tumor very early after the beginning of the treatment, and have left the tumor in favor of cDC2 during the phase of immunotherapy-induced rejection of the tumor (86). Therefore, the location and timing of cDC1 booster administration in combination with immunotherapies are likely to be determinant for treatment success.

One way to attempt improving the response of cancer patients to immunotherapy is to boost the ability of their cDC1 to crosspresent tumor Ag (60). Combined administration in mice of mAbs directed against tumor Ag with a stabilized form of IL-2 enhances antitumor immunity in a cDC1-dependent manner (**Table 4**) (114, 139). This is because antibody-dependent cellmediated cytotoxicity provokes an immunogenic tumor cell death favoring the up-take and cross-presentation by cDC1 of tumor cell fragments. Indeed, tumor cell lysates or tumor plasma membrane vesicles may represent the best sources of Ag for crosspresentation, because they include a constellation of neoAg. Tumor Ag cross-presentation by cDC1 can also be triggered upon administration of tumor Ag coupled to mAb directed against cDC1 surface markers (**Table S1**).

Cross-presentation of tumor Ag by cDC1 must occur simultaneously to their immunogenic maturation such that they can deliver all of the signals required for the efficient priming of naïve CTLs or the reactivation of exhausted CTLs, including proper co-stimulation, activating cytokines, chemokines and CD4<sup>+</sup> T cell help, in the tumor bed or upon migration to the draining lymph node. This implies administrating the good adjuvant at the right time and in the proper place. TLR3, CpG, or STING agonist adjuvants promoting a strong production of IFN-I are especially efficient at promoting antitumor immunity, even more upon peritumoral rather than systemic delivery (79, 140–142). To further promote the beneficial anti-tumor activity of IFN-I and limit their deleterious side effects, a synthetic mutated IFNα2 has been engineered and coupled to anti-Clec9a mAb, allowing delivery of IFN-I activity specifically on cDC1. The administration of this cDC1-targeted adjuvant synergizes with mAIM, chemotherapy, or with low dose of TNF, resulting in a regression or a long-lasting protection against melanoma and breast carcinoma in the absence of toxic effects (115). Targeting IFN-I on tumor cells also improves the antitumor effects of mAIM (112, 114, 143), in part through direct effects on cDC1 and/or cDC2 (143) but also more generally by modulating the responses of many other immune cells in the tumor microenvironment. Importantly, to promote protective antitumor immunity, IFN-I must be delivered simultaneously to, or shortly after, the tumor Ag. Indeed, IFN-I-induced cDC1 maturation strongly decreases their phagocytic capacity and thus prevents their ability to cross-present if occurring before tumor Ag uptake (114).

IL-12 production by cDC1 is proposed to significantly contribute to their protective antitumor activity, at least in part by promoting Th1 response and activating IFN-γ production by NK cells and CTLs. Administration of recombinant IL-12 in combination or not with mAIM therapy displayed antimetastatic (66) or immunotherapy-induced antitumor effect (118) in WT animals (**Table 4**). However, interestingly, these potentiating effect of the mAIM therapy was lost in Batf3−/<sup>−</sup> mice (118), showing that IL-12 administration is not sufficient to replace the antitumor functions of cDC1.

Another function of cDC1 that could be exploited for boosting current immunotherapies is their ability to respond to the chemoattractant XCL1, due to their specific expression of the chemokine receptor XCR1. At steady state, high levels of the Xcl1 transcript are detected in NK cells, NKT cells and memory CTLs. Upon activation, Xcl1 expression is further upregulated in these cells and induced in effector CTLs, which promotes the recruitment of cDC1 into inflamed tissues in close contact to XCL1-producing cells, leading to a cross-talk amplifying the responses of both cell types (77). Therefore, intra-tumoral delivery of XCL1 seemed a promising strategy to enhance local recruitment of cDC1 in order to harness their protective functions in combination immunotherapies. Certain types of melanoma or colon carcinoma tumors engineered to express high amount of XCL1 harbored a higher cDC1 infiltration and were rejected faster or grew more slowly in WT but not in Batf3−/<sup>−</sup> mice, as compared to control tumors. However, this process was inhibited in tumors producing PGE2, due in part to the ability of this molecule to decrease XCR1 expression in cDC1 (55). This study illustrates well the necessity not only to mobilize cDC1 in combination immunotherapies, but at the same time to dampen the immunosuppressive pathways targeting cDC1 functions in the tumor microenvironment. Hence, in addition to directly targeting CTL and cDC1 functions, combined immunotherapies should probably include means to counteract the tumor immunosuppressive pathways acting indirectly on these cells, such as inhibiting β-catenin, PGE<sup>2</sup> or adenosine receptor signaling (55, 57, 93, 144), or depleting/reprogramming the tumor-associated mononuclear phagocytes endowed with immunosuppressive functions including macrophages, MDSCs and pDCs (67, 145, 146).

Because cDC1 are the rarest subset of Ag presenting cells in tumors (30) and their numbers have been shown to decrease in the course of certain immunotherapies (86), strategies aiming at harnessing their functions for cancer treatment should include methods to promote their expansion in vivo. Tumor-infiltrating NK and T cells upregulate FLT3-L, which seems to contribute to the local expansion of tumor cDC1 (62), and most likely cDC2. Administration of recombinant FLT3-L to tumor-bearing mice as a supportive treatment to mAIM immunotherapy reinforces CTL infiltration and activation in the tumor (137), and the combined administration of FLT3-L and poly(I:C) which respectively support cDC1 expansion and activation significantly improved antitumor mAIM immunotherapy in mice (79, 118) (**Table 4**). Alternatively, large quantities of cDC1 could be injected peritumorally simultaneously to ICB administration, in order to further promote the priming of naïve CTLs toward neoAg or the reactivation of endogenous antitumor CTL responses. This should be achievable since recent studies showed that large numbers of fully functional cDC1 can be


\*Of note: anti-TIM3 alone exerts its effect independently

 of CD11c+ cells (138).

TABLE 4 |

Anti-tumor

 off-the shelf therapies relying on cDC1 functions.

generated in vitro from hematopoietic progenitors cultured with FLT3-L on feeder cells expressing the Notch ligand Delta-like 1 (147, 148).

In summary, several studies have attempted to improve the response to cancer chemotherapies, radiotherapies or mAIM immunotherapies by combining these treatments with putative or known cDC1 boosters (**Table S1** and **Table 4**). In all cases, tumor progression was greatly dampened in parallel with enhanced CTL activation and sometimes with a documented increased maturation of cDC1. In many studies, this beneficial effect was shown to be abrogated in Batf3−/<sup>−</sup> mice. These studies in mouse preclinical models of combined immunotherapies strongly enforce the hypothesis that harnessing cDC1 functions in cancer patients should improve their response rate and longterm survival to already existing immunotherapies including ICB, and show how this could be achieved.

### WHAT FUNCTIONAL SPECIFICITIES MAKE HUMAN cDC1 GOOD CANDIDATE AG-PRESENTING CELLS FOR THE PROMOTION OF PROTECTIVE ANTI-TUMOR IMMUNITY?

#### Comparative Genomics Established Overall Homology Between Mouse and Human cDC1

A striking overall homology between human and mouse cDC1 was established through cross-species comparative genomics of several immune cell types (14, 149–153). This provided a very strong incentive to investigate the role of human cDC1 in antitumor immunity, considering the body of evidence discussed above supporting a critical role of mouse cDC1 in promoting NKand CTL-mediated tumor control in preclinical cancer models.

### Conservation of Key Characteristics Proposed to Underlie Mouse cDC1 Protective Role Against Cancer

A number of shared and distinctive features of mouse and human cDC1 are summarized in **Table 5** (11, 14, 15, 61, 63, 79, 147–150, 154–162, 165–174), with their possible relevance for immune defense against cancer. Globally, the combination of features proposed to endow mouse cDC1 with their unique efficacy to promote protective anti-tumor immunity is well conserved in human cDC1. Differences in cross-presentation efficacy appear to be more subtle between human than mouse DC subsets (155, 175). Of note, however, a consensus has emerged from various studies that human cDC1 are more efficient than other DC types for the cross-presentation of cell-associated Ag (15, 45, 63, 155–157), likewise to the situation in the mouse. Human cDC1 were reported by several teams not to produce IL-12 (150, 167). However, other studies have shown that under optimal conditions of stimulation human cDC1 can produce this cytokine to levels equivalent or higher than those made by cDC2 or MoDCs (147, 156, 168, 169, 173, 174).

# Current Limitations, Controversies or Unknowns

One study has recently reported that human cDC1 do not migrate efficiently from the parenchyma of non-lymphoid tissues to their draining lymph nodes (176). This bears important implications for vaccination or immunotherapies if it is confirmed.

The mechanisms that make human cDC1 especially efficient for cross-presentation of cell-associated Ag are still not understood. One of the main limitations to address this issue, and more generally to study the functions of human cDC1 and their molecular regulation, is their rarity and fragility.

### WHAT EVIDENCES EXIST THAT HUMAN cDC1 CORRELATE WITH A BETTER OUTCOME IN CANCER PATIENTS AND WHAT CAN BE INFERRED FROM THESE STUDIES REGARDING THEIR PROTECTIVE MODE OF ACTION?

#### A Higher Expression of cDC1 Gene Signatures in Tumors Correlates With a Better Clinical Outcome

#### State-of-the-Art in Assessing cDC1 Infiltration From Whole Tumor Tissue Gene Expression Profiles

Several public datasets are available with gene expression profiles of whole tumor tissue from large cohorts of patients with well documented clinical characteristics. Increasing numbers of teams are querying this gold mine to test whether higher expression in tumors of gene signatures specific for various cell types or biological pathways are associated with a better or worse clinical outcome. Such analyses could allow high throughput testing of the possible relationship between overall survival and tumor infiltration by specific cell types in a given activation state. Such analyses would then allow focusing further studies on the most promising observations, to test whether they are confirmed by using immunohistofluorescence or flow cytometry to directly measure the frequency of specific combinations of immune cell types and activation states in the tumors. However, there is currently no consensus on which gene signatures are the most specific and robust for each immune cell type of interest. In particular, until very recently, to assess the prognostic value of DC infiltration into the tumors, the gene signatures used were those from in vitro derived MoDCs. The extent of DC infiltration into tumors as computationally inferred in these studies had no significant prognostic value for overall patient survival, or was even associated to an increased hazard risk (177– 181). However, based on the known major differences between MoDCs and cDCs (14–16) and on the beneficial role of mouse cDC1 in antitumor immunity, further studies were needed to assess whether higher infiltration of human tumor by other DC types, in particular cDC1, could be associated with a better clinical outcome.

In the last four years, from the few studies performed to address this issue, a consensus has been emerging that TABLE 5 | Shared and distinctive features of mouse and human cDC1.


higher expression of cDC1 transcriptomic fingerprints in various tumors correlates with a better clinical outcome (**Figure 4**, green cells, in the bold rectangle).

In the case for breast cancer (BRCA), a good prognosis of a higher tumor infiltration by cDC1 has been documented independently by 4 studies (55, 65, 66, 182), altogether interrogating three patient cohorts [TCGA, METABRIC, and the meta-cohort generated by Györffy et al. (183)]. A higher expression of the cDC1 transcriptomic signature in tumor was at least as powerful a predictor of prolonged patient survival to cancer as that of the CTL signature (55, 182). Transcriptomic fingerprints or genes associated to certain other immune cell types including cDC2, pDCs or monocytes/macrophages did not have a positive prognostic value (**Figure 4**, gray or red cells). This supports the hypothesis of a specific protective role of high infiltration of breast tumors by cDC1, rather than the alternate hypothesis that differential levels of cDC1 gene expression in tumors reflect differences in their overall leukocyte infiltration and lead recapitulates the known different clinical outcome of "Hot" or "Warm" vs. "Cold" or "Icy" tumors. However, more studies are warranted to address this issue. For triple negative breast cancer (TNBC), the positive prognostic value of higher cDC1 infiltration in tumors was even better than for all types of BRCA or for luminal BRCA. This was observed in three studies, encompassing altogether the analyses of two patient cohorts (55, 65, 66).

Similar analyses were performed for other types of cancer. For head and neck squamous cell carcinoma (HNSC) and lung adenocarcinoma (LUAD), a higher expression of the cDC1 transcriptomic signature in tumor was also associated to a better clinical outcome by at least two independent studies (55, 62, 80). This was also the case for skin cutaneous melanoma (SKCM) (**Figure 4**) (47, 72), on two distinct patient cohorts, TCGA and the cohort described by Boguvonic et al. (184). In addition, for metastatic melanoma, the specific positive prognostic value of high cDC1 infiltration in the

tumor bed was confirmed by flow cytometry analyses, whereas no significant prognostic value was observed for many other cell types including cDC2, pDCs, Mono/Mac and most surprisingly CTLs (62) (**Figure 4**). Finally, for LUAD, single cell RNA sequencing and paired CyTOF analyses of tumors and their neighboring normal lung tissue showed that cDC1 were significantly reduced in tumors, contrasting to increased numbers of macrophages in an immunosuppressive state and of cDC2/MoDCs (185). This study further supports the previously proposed hypothesis that the balance between cDC1 vs. cells of the monocyte/macrophage/neutrophil lineages in the tumor leukocyte infiltrate strongly determines the degree of local immunosuppression (65, 93).

#### Limitations, Controversies, or Unknowns

All of the above studies suggest that, in a variety of human cancers, intra-tumoral cDC1 abundance correlates with a better clinical outcome. However, further studies are required to confirm these results, to extend these types of analyses to other types of cancer, and to deepen our understanding of the underlying mechanisms.

A first issue that clearly stands out in **Figure 4** is the lack of a consensus definition of the transcriptomic fingerprints used for each immune cell type across studies, not only for cDC1 but even for CTLs or NK cells. Indeed, there is relatively little overlap between the gene signatures used for the same cell types across studies (blue names in **Figure 4**). Several genes used in some of the cDC1 transcriptomic fingerprints are known to have a promiscuous expression across many cell types (55). CCR7 and ITGAE (CD103) are expressed on all mature DC types and on T cell subsets. BATF3 and ZBTB46 are shared with cDC2, and FLT3 with cDC2, pDCs and hematopoietic progenitors. IRF8 is highly expressed in pDCs and certain types of monocytes or macrophages. THBD (CD141/BDCA3) can be expressed on cDC2, pDCs, MoDCs and non-immune cell types. One study undertook the "tour de force" of profiling by microbulk RNAseq all the distinct mononuclear phagocyte types that they could identify in, and isolate from, BRCA or TNBC, in order to generate the cell-type specific transcriptomic signatures the most relevant to the cancer types studied (182). However, in this study, the TNBC gene signature of the cell population enriched in cDC1 (cDC1e) (182) encompasses only 8% of genes known to be selectively expressed in cDC1 but 42% of genes known to be expressed in NK cells. This raises the question of the interpretation of the positive prognostic value of that signature. It might not only reflect the infiltration of cDC1 but also that of NK cells, in consistency with other analyses in an independent study (55). Indeed, depending on individual samples, the cDC1e population encompassed 50– 95% of other cells than cDC1. Using CD16 and CD56 for excluding NK cells from cDC1e cells might have been insufficient, since the strongly activated human NK cells expressing the highest levels of XCL1 and XCL2 express neither of these cell surface markers (55). More generally, it is likely that the much higher infiltration of TNBC by lymphocytes, as compared to luminal BRCA, led to major differences in the cell types other than cDC1 that were included in the cDC1e cell population between these two types of cancers. This could confound interpretation of the results of the enrichment analysis of these signatures.

There is a need to define better transcriptomic signatures for human immune cell types, allowing to more rigorously computationally deconvolute the extent of their infiltration in tumors and its eventual correlation with the clinical outcome. One strategy to achieve this aim is to select genes which show high selective expression in the targeted immune cell type across tissues and activation conditions, as well as between human and mouse (14, 45, 153). An alternative strategy could be to perform single cell RNA sequencing from tumor samples, in order to define the transcriptomic signatures specific to various combinations of relevant immune cell types and activation states in the most unbiased way. This strategy would also alleviate the potential confounding effect of cross-contamination between populations as can occur with microbulk gene expression profiling studies (11, 182). Moreover, it will generate transcriptomic signatures specific to the combination of the cell types and of the cancer studied. Indeed, it has been reported that using gene signatures derived from another tissue does not always work adequately to computationally deconvolute the immune cell type composition of tumors, due to differential imprinting of cells in distinct local microenvironments (181, 182).

A second issue is the necessity to include signatures of various types of immune cells, to ensure that the better prognostic value observed in the patients whose tumors harbor higher levels of the genes specific to the candidate immune cell type is not merely a reflect of a higher overall leukocyte infiltration. Indeed, the goal is not just to compare globally "Hot" or "Warm" vs. "Cold" tumors. Rather, it is to pinpoint which immune cell types specifically promote tumor control, or on the contrary contribute to local immunosuppression, in order to identify how to best manipulate the tumor infiltrate, for the benefits of cancer patients, through combined immunotherapies. Thus, the cell types considered to be functionally and/or developmentally the most closely related to the candidate one should be included, for example cDC2 or MoDCs for cDC1, NK cells and γδ T cells for αβ T cells. In addition, one should also include cell types expected to have no, or opposite, impact on tumor growth, for example neutrophils, macrophages and regulatory T cells which are considered as promoting immunosuppression.

A third issue is to deepen our understanding of when, where and how cDC1 promote tumor control. In several studies, the inferred higher cDC1 infiltration in tumors was correlated with higher inferred infiltrations of CTLs or NK cells, and with higher expression of FLT3L, XCL1, XCL2, CCL4, CCL5, LAMP3, CCR7, CXCL9, CXCL10, and CXCL11 (**Figure 4**) (55, 57, 62, 64, 80, 93). These observations need further independent confirmation through the analysis of other cohorts of patients, and by using complementary methodologies including immunohistochemistry, CyTOF or single cell RNA sequencing to measure the correlation between cDC1 infiltration into the tumors and the status of antitumor

#### TABLE 6 | Completed clinical trials targeting cDC1.


SLN, Sentinel lymph nodes.

NK and CTL responses. In any case, these studies support our proposed model of a critical positive cross talk between cDC1, cytotoxic lymphocytes and CD4<sup>+</sup> T cells for promoting effective antitumor immunity (**Figure 2**). Finally, it would be of utmost interest to extend to cohorts of patients benefiting from various types of immunotherapies these analyses aiming at deconvoluting the gene expression profiles of whole tumor tissue into immune cell type composition. This should help determining whether the clinical response can be predicted from cDC1 infiltration in the lesions, and to adapt the treatments accordingly for example by combining to ICB the use of drugs promoting cDC1 recruitment and activation into the tumors of patients when this process is defective (**Figure 3**).

#### Efficacy of Immunotherapeutic Protocols That May Preferentially Target/Harness Human cDC1

A few clinical trials have already been conducted using treatment protocols that have been proposed to preferentially target/harness human cDC1 (**Table 6**) (186, 187, 190). They gave encouraging results, which further supports the rationale of specifically targeting human cDC1 for the design of novel combined immunotherapies against cancer (189).

### WHAT TOOLS ARE AVAILABLE TO STUDY AND MANIPULATE HUMAN cDC1 FOR THE BENEFITS OF CANCER PATIENTS?

Building on the conservation of cDC1 molecular makeup and functions between mouse and human, similar tools have been generated in both species to specifically target these cells for immunotherapy against cancer. This should accelerate translation from mouse preclinical studies to human clinical trials. Hence, most of the tools and approaches that have been detailed in the section on mouse experimental models (**Figure 3** and **Tables 3**, **4**) could be implemented in humans, as briefly summarized below.

#### Generation and Study of Novel in vitro Models of Human DC Types

To overcome the roadblock of the rarity of human cDC1 and of their fragility upon ex vivo isolation, we and others recently developed optimized in vitro culture systems to generate high numbers of cDC1, cDC2 and pDCs from CD34<sup>+</sup> hematopoietic progenitors (15, 147, 148). These novel in vitro models will allow rigorous comparison of the functions of the different human DC types, dissection of their molecular regulation, and better understanding of their cross talk. Further adaptations of these protocols are warranted to derive in vitro autologous cDC1 from the circulating CD34<sup>+</sup> cells of patients, load them with Ag and mature them, under conditions compatible for clinical use in vaccination or immunotherapy. It should be noted that encouraging results have been obtained with clinical trials of autologous ACT of ex vivo loaded and matured pDCs and cDC2 in melanoma patients, which seem superior to MoDCs to prime or boost endogenous CTL responses against the tumor. This emphasizes that, as in mice, cDC1 are not the only DC type that could be successfully harnessed for combined immunotherapy in cancer patients (6, 7, 191, 192).

#### Means to Specifically Deliver Ag and Maturation Signals to Human cDC1

Considering their conserved specific expression pattern on mouse and human cDC1, and the very encouraging results obtained in mouse preclinical models, the CLEC9A and XCR1 receptors are the best candidates for Ag, or Ag+adjuvant cargo, delivery to human cDC1, using recombinant ligands (193, 194) or monoclonal antibodies. A combination of TLR3- and TLR8-specific agonists is desirable to promote an immunogenic maturation associated with the production of both IL-12 and IFN-β/λ (**Table 6**) (195). Targeting delivery of IFN-I activity to cDC1 is another very promising adjuvant based on the proof-of-principle published in mice (**Tables 3**, **5**). Additional means could be envisioned to favor the crosstalk between cDC1 and NK or NK T cells (196), e.g., use of NK cell immune checkpoint blockers (101–105) or targeted delivery to cDC1 of activating antigenic ligands for NK T cells (120).

#### Means to Promote cDC1 Differentiation, Survival and Local Recruitment in the Tumor Bed

Systemic injection of FLT3-L could promote cDC1 differentiation and survival (79). Local delivery of XCL1 could promote their recruitment in the tumor bed. In patients responding to checkpoint blockade inhibitors, these functions might be achieved upon local NK and CTL activation for FLT3-L, XCL1, and CCL4/5 production (55).

#### Blockade of cDC1 Checkpoint Inhibitors

A systematic analysis of immune factor checkpoint expression on human DC types is ongoing in order to investigate which ones could be reasonable candidates as components of combined immunotherapies targeting both CTLs and DCs (197).

# CONCLUDING REMARKS

Lately, cDC1 have been in the spotlight of many studies investigating in mice the immune mechanisms driving tumor rejection, spontaneously or upon immunotherapy. All these studies converge toward a hub role of cDC1 in providing the initial priming, or in sustaining the activation, of antitumor T and NK cell responses. These advancements in our understanding of the role of cDC1 in antitumor immunity have been made possible by the recent blossoming of genetic tools allowing cDC1 manipulation. However, so far, most conclusions have been drawn from results obtained under experimental conditions that were not solely targeting cDC1, whether it was the use of genetically engineered mouse models or of mAb directing against cell surface markers. In fact, to be protective against immunosuppressive tumors such as those treated in the clinic, the immune response is necessarily complex and multi-parametric. More and more observations pinpoint that, in addition to cDC1, other DCs, type 1 CD4<sup>+</sup> T cells, and sometimes neutrophils are also central in promoting protective antitumor immunity, whereas Treg or type 17 CD4<sup>+</sup> T cells, monocytes and macrophages may rather play immunosuppressive roles. Further studies using models allowing conditional depletion of cDC1 will be critical in rigorously investigating whether cDC1 functions are instrumental at the time when immunotherapies are delivered. These studies will definitely settle the currently prevailing hypothesis that cDC1 functions, when specifically boosted, could provide great support to boost patient responses to currently used anticancer immunotherapies.

In human tumors, enrichment of genetic signatures described as cDC1-specific is associated with a good prognosis and a better clinical outcome in a several cancers, including luminal and TN breast cancer. These correlative analyses should be extended to additional types of cancer and to different patient treatment regimen. It is possible that the extent of cDC1 infiltration in the tumor fluctuates over time following the development or the suppression of an efficient antitumor immune response, as observed in mice during immunotherapy (86), and that cDC1 infiltration may not be protective against all types of cancer. Still, the perspective of exploiting cDC1 to improve current immunotherapies is extremely encouraging, and completion of cDC1-targeting vaccine clinical trials in human will surely help in gaining insight into their importance in cancer.

# AUTHOR CONTRIBUTIONS

All authors wrote the manuscript, contributed to manuscript revision, read and approved the submitted version.

# FUNDING

This work was supported by grants from the European Research Council under the European Community's Seventh Framework Program (FP7/2007–2013 grant agreement number 281225 to MD), from the Agence Nationale de la Recherche (ANR; XCR1-DirectingCells to KC), from the Fondation pour la Recherche Médicale (label Equipe FRM 2011, project number DEQ20110421284 to MD), from the Fondation ARC pour la recherche sur le cancer (to KC) and from the Institut National du Cancer (INCa PLBIO 2018-152). This work also benefited from institutional funding from CNRS and INSERM. J-CC was supported by doctoral fellowship from La Ligue Nationale Contre le Cancer. RM was supported by doctoral fellowships from the Biotrail Ph.D. program (Fondation A∗MIDEX) and from Fondation ARC pour la recherche sur le cancer.

#### ACKNOWLEDGMENTS

**Figures 1**–**3** have been created with BioRender (https:// biorender.io/) under academic license.

#### REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.00009/full#supplementary-material


independently of cross-presentation. Cancer Immunol Res. (2018). 7:29–39. doi: 10.1158/2326-6066.CIR-18-0138


from transcriptional loss of class I HLA. Nat Commun. (2018) 9:3868. doi: 10.1038/s41467-018-06300-3


to potent and systemic tumor regression and immunity. Cell Rep. (2015) 11:1018–30. doi: 10.1016/j.celrep.2015.04.031


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Cancel, Crozat, Dalod and Mattiuz. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Molecular Aspects of Dendritic Cell Activation in Leishmaniasis: An Immunobiological View

Rafael Tibúrcio1,2, Sara Nunes 1,2, Ivanéia Nunes 1,2, Mariana Rosa Ampuero1,2 , Icaro Bonyek Silva1,2, Reinan Lima1,2, Natalia Machado Tavares 1,2,3 \* and Cláudia Brodskyn1,2,3 \*

<sup>1</sup> Gonçalo Moniz Institute, Oswaldo Cruz Foundation, Salvador, Brazil, <sup>2</sup> Federal University of Bahia, Salvador, Brazil, <sup>3</sup> Instituto Nacional de Ciência e Tecnologia (INCT) iii Instituto de Investigação em Imunologia, São Paulo, Brazil

#### Edited by:

Daniela Santoro Rosa, Federal University of São Paulo, Brazil

#### Reviewed by:

Laila Gutierrez Kobeh, National Autonomous University of Mexico, Mexico Mayda Gursel, Middle East Technical University, Turkey Nahid Ali, Indian Institute of Chemical Biology (CSIR), India Jude Ezeh Uzonna, University of Manitoba, Canada

#### \*Correspondence:

Natalia Machado Tavares natalia.tavares@bahia.fiocruz.br Cláudia Brodskyn brodskyn@bahia.fiocruz.br

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 28 September 2018 Accepted: 28 January 2019 Published: 22 February 2019

#### Citation:

Tibúrcio R, Nunes S, Nunes I, Rosa Ampuero M, Silva IB, Lima R, Machado Tavares N and Brodskyn C (2019) Molecular Aspects of Dendritic Cell Activation in Leishmaniasis: An Immunobiological View. Front. Immunol. 10:227. doi: 10.3389/fimmu.2019.00227 Dendritic cells (DC) are a diverse group of leukocytes responsible for bridging innate and adaptive immunity. Despite their functional versatility, DCs exist primarily in two basic functional states: immature and mature. A large body of evidence suggests that upon interactions with pathogens, DCs undergo intricate cellular processes that culminate in their activation, which is paramount to the orchestration of effective immune responses against Leishmania parasites. Herein we offer a concise review of the emerging hallmarks of DCs activation in leishmaniasis as well as a comprehensive discussion of the following underlying molecular events: DC-Leishmania interaction, antigen uptake, costimulatory molecule expression, parasite ability to affect DC migration, antigen presentation, metabolic reprogramming, and epigenetic alterations.

Keywords: dendritic cell activation, leishmania- dendritic cell interaction, parasite uptake, dendritic cells migration, metabolism of infection, epigenetic modifications

#### INTRODUCTION

#### Important Considerations in Leishmaniasis

Leishmaniasis comprises a collection of neglected protozoan infections caused by unicellular organisms belonging to the genus Leishmania spp. According to the current World Health Organization estimation, 12 million people are affected by leishmaniasis and 350 million are at risk of infection worldwide (1–3).

The pathology of this disease results in a wide spectrum of clinical manifestations not only associated with the biological aspects of Leishmania species and strains, but also with host immune responses. Interestingly, it has been recently suggested that the clinical progression of the disease is influenced by several other factors, ranging from the host's nutritional status to the presence of RNA viruses in the Leishmania species (4–7).

These manifestations are dichotomically divided into Visceral (VL) and Tegumentary Leishmaniasis (TL). The former is characterized by the dissemination of parasites to visceral organs, while the latter branch includes Localized Cutaneous Leishmaniasis (LCL), a frequent form of TL in which ulcerated skin lesions are common. It has been abundantly reported that a modest fraction of LCL cases can evolve into mucosal lesions, which is termed as Mucocutaneous Leishmaniasis (MCL). Additionally, TL can also present as a variety of clinical manifestations, such as Disseminated Cutaneous Leishmaniasis (DCL), which comprises multiple nodular ulcerated lesions, whereas Diffuse Leishmaniasis (DL) is characterized by scattered nonulcerated lesions (5, 8, 9).

Leishmania transmission occurs when infected sandflies inoculate the promastigote forms of the parasite into the host skin. Additionally, the arthropod vector also introjects various parasite-associated compounds, along with other molecules found in salivary secretions, which collectively exert immunomodulatory effects on the host defense (10). The early events of infection are characterized by the engagement of different phagocytic cells (e.g., tissue-resident macrophages, dermal DCs, and neutrophils) in the recognition and uptake of parasites (8). Emerging pieces of evidence indicate that neutrophils are one of the first cell types to interact with Leishmania parasites (11). Subsequently, depending chiefly on the Leishmania species, infected neutrophils become apoptotic and can be phagocytized by macrophages (12). Accordingly, parasite transmission to these cells becomes facilitated, leading to the subsequent differentiation of promastigotes into intracellular replicative amastigotes that occurs in the interior of macrophages phagolysosomes. Additionally, the literature upholds that dendritic cells (DCs) are also key elements in the early interaction with Leishmania parasites, thusly these are thought to be a decisive in the outcome of infection (13). Indeed, the complex interactions occurring between DCs and parasites may lead to long-term Leishmania replication, or to the establishment of an effective immune response against this pathogen.

#### The Immunobiology of Dendritic Cells

DCs are competent antigen presenting cells (APC) that take center stage in both the induction of immunological responses and the generation of tolerance (14). In the context of inflammation and infection, DCs are responsible for orchestrating the connection between the innate and adaptive axis of immunity. Interestingly, despite the significant importance of DCs in several immunological processes, these cells do not comprise a homogeneous population, and are further classified into distinct subtypes according to origin, differential expression of surface proteins, cell localization, and immunological function (15).

#### Dendritic Cell Origin

It has been long hypothesized that DCs stem from a bonemarrow resident population of hematopoietic stem cells (HSC), which eventually give rise to both granulocyte-macrophage progenitors (GMP), and multi-lymphoid progenitors (MLP), the precursors of all DC subsets (16). Subsequent stages in DC ontogeny involve precursors, such as CD14+ monocytes, circulating blood myeloid DCs (mDCs), or plasmacytoid DCs, from which all myeloid and lymphoid DCs are derived. It is noteworthy that mDC precursors comprise a heterogeneous lineage of cells predetermined to develop into CD1+ or CD141+ DCs. Additionally, human mDCs express conventional myeloid markers, including CD11c, CD11b, CD13, and CD33 (17). In mice, these cell populations are often referred to as conventional DCs. Interestingly, it has been well-elucidated that in humans, both CD14+DCs and inflammatory DCs are derived from classical monocytes, which justifies the fact that these cells present greater similarity to monocytes and macrophages than other DC subsets (18).

#### Dendritic Cell Subtypes

#### **Myeloid/Conventional Dendritic Cells**

Typically, myeloid DCs are classified into two subtypes: cDC1 and cDC2. The human cDC1 subset is identified by the expression of CD141 (BDCA-3), while the murine equivalent is subdivided into a splenic CD8α-bearing population and another CD141+ DC subset residing in non-lymphoid tissues (19–21). Human and mouse cDC1 express Clec9A (C-type lectin domain family 9-member A) and XCR1 (a chemokine receptor), which provide specificity for their biological activities in combating invasive microorganisms and tumors (22). In regard to the expression of transcription factors, cDC1s are characterized as producing both BAFT3 (Basic leucine zipper transcription factor) and IRF8 (Interferon regulatory factor 8). It has been long suggested that cDC1s have the capacity to effectively induce the activation of CD8+ T cells via the process of antigen cross-presentation, as well as produce copious amounts of IL-12p70 (23, 24).

cDC2s express both common myeloid markers, such as CD11b, CD11c, CD13, and CD33, in addition to other antigens more recently identified in these cells: CD1c, CD2, FceR1, and SIRPA (15). cDC2s comprise a large portion of the human conventional DCs found in blood and tissues. The immunological function of cDC2s is granted by a myriad of immune receptors, including Toll-like receptors (TLRs) 2,4,5,7 and 8, C-type lectins, including Dectin-1 and−2 as well as Nodand RIG-like receptors (**Figure 1** and **Table 1**) (15).

#### **Plasmacytoid Dendritic Cells**

Plasmacytoid DCs (pDCs) comprise a group of Type I interferon (IFN)-producing cells whose distinguishing feature is their participation in the response against viral infection. Interestingly, human pDCs were first identified as a population of cells found in the peripheral blood and tonsils. With respect to morphology, blood pDCs are mainly recognized by their resemblance to lymphocytes, whereas IL-3/CD40L-cultured pDCs possess a microscopic appearance similar to mDCs (22, 25). Typically, human pDCs are characterized by the unique expression of both cell-surface receptors and transcription factors. Of note, the distinctly expressed receptors include both conventional and other recently identified markers. The former includes CD123 (IL-3R), CD303 (BDCA-2), and CD304 (BDCA-4), while the later includes FceR1, DR6 (CD358), and CD300A. As previously mentioned, these cells also exhibit distinctive TF expression, including E2-2, IRF8, and IRF4 (26, 27). It is particularly interesting to note that, in contrast to cDCs, human pDCs do not express any conventional myeloid markers, e.g., CD11b, CD11c, CD13, and CD33 (15). Murine pDCs express siglecH (Sialic acidbinding immunoglobulin-type lectin), bst2 and Ly6C. SiglecH is a surface receptor responsible for binding glycans presenting sialic acids residues. Interestingly, Blasius et al demonstrated that siglecH regulates type I IFN production in a DAP12 dependent manner (28). Bst2, an integral membrane protein associated with lipid rafts, has also been associated with IFNmediated responses against viral infection (29). Additionally, several reports have demonstrated that both murine and human pDCs rely on TLR7 and TLR9 expression for IFN-α production

and immunity against viruses (30). Interestingly, mounting evidence have demonstrated that pDCs also play a role in the establishment of peripheral tolerance by delivering antigens to the lymph nodes (**Figure 1** and **Table 1**) (31–33).

#### **Langerhans Cells**

A distinct lineage of epidermis-resident DCs, known as Langerhans cells (LC), mainly characterized by the expression of C-type lectin Langerin and CD1a, grant organisms immunity against several skin pathogens, such as fungi and bacteria (34). Uniquely, LCs possess Langerin-replete organelles, known as Birbeck granules. Although their main function has not been well-elucidated, the depletion of these granules has not been determined to mitigate the process of antigen presentation (**Figure 1** and **Table 1**) (35).

#### **Monocyte-Derived Dendritic Cells**

Monocytes constitute a very plastic group of mononuclear phagocytic cells long thought to be the source of macrophages and DCs. Several reports suggest that monocytes possess both pro- and anti-inflammatory functional specializations which are, in turn, chiefly regulated by tissue environments (36). Human monocytes comprise two subsets of peripheral blood circulating cells mainly characterized by the expression of CD14 and CD16, whereas murine monocytes can be identified by the presence of Ly6C, CCR2, and CX3CR1 (22). Of note, in the presence of inflammation, blood-circulating monocytes invade tissues, subsequently differentiating into monocyte-derived Dendritic cells (moDCs). It has been long established that in vitro moDCs are obtained by stimulating monocytes with GM-CSF and IL-4 (37). These cells possess a broad functional repertoire, including lymphocyte activation and the production of cytokines, such as IL-6, TNF-α, IL-12, IL-23, and IL-1 (15). Inflammatory DCs (iDCs) express both CD14 and CD16, in addition to CD206, CD209 (DC-SIGN), and CD163. Importantly, a novel tumor necrosis factor (TNF)-and inducible nitric oxide synthase (iNOS)-producing DCs (TIP-DCs) subset has been reported to exert a pivotal role in the course of several infectious diseases, including experimental leishmaniasis (**Figure 1** and **Table 1**) (38).

#### **Adaptative Immunity Gatekeepers: The Role of DCs and T Cells Activation**

In an immature state, DCs are typically located in peripheral tissues and express low levels of major histocompatibility complex II (MHC II) and costimulatory molecules. These cells possess highly efficient cellular machinery for antigen recognition and capture (39). In response to signals associated with infection and inflammation, such as the presence of pathogens and other damaging elements, DCs undergo intricate molecular processes that culminate in the acquisition of a mature functional state, whose main characteristic is the ability to induce both naïve CD4<sup>+</sup> T cell activation and proliferation via antigen presentation (40). Most importantly, the signaling process that induces DC maturation involves the recognition of pathogen-associated molecular patterns (PAMPs) by way of a sophisticated surface and intracellular molecular detection system consisting of pattern recognition receptors (PRRs) and downstream signaling (41).

After interaction with antigen-bearing DCs, naïve CD4<sup>+</sup> T cells are capable of differentiating into two main functional phenotypes: T helper 1 (Th1) and T helper 2 (Th2) profiles. It should be noted that this dichotomy is a rather simplistic representation of the Th cell repertoire. In recent decades, several studies have identified other Th subtypes, including Th17 (whose hallmark is the production of IL-17 in response to viruses, bacteria and fungi), Th9 (a producer of IL-9 and IL-10, and also a key element in humoral interplay with B cells), Th follicular (characterized by the production of IL-4 and IL-2 (also related to supporting B cell-mediated immunity), and T regulatory (Tregs), involved in the promotion of selftolerance (42). Since T regs exert a prevalent immunological role in the regulation of other immune cells, their populations


BMDC, bone marrow -derived dendritic cell; IFN-γ, interferon gamma; IL-12, interleukine-12; NK, natural killer cells; TLR, toll-like receptors; HIF1α, hypoxia-inducible factor 1 α.

heterogeneity and functional specializations are of particular interest. Commonly, T regs are dichotomically classified as "natural" (CD4+CD25+Foxp3<sup>+</sup> T cells) or inducible T regs (a group that includes the IL-10-secreting Treg1 cells, the Th3 population that produces both TGF-b and IL-10, and foxp3<sup>+</sup> Tregs). A more detailed description of the diversity and functions of T regulatory cells can be found elsewhere (43).

#### **DC as Modulators of the Adaptive Immune Response in Leishmaniasis**

The biological features of pathogens and activation PRRs as well as the underlying signaling processes, are determinant in the specific cytokines secreted by activated DCs, which in turn are one of the key element in the polarization of Th cell subtypes (44).

In general, Th1 cells produce pro-inflammatory cytokines, such as Interferon Gamma (IFN-γ), which lead to Tumor Necrosis Factor Alpha (TNF-α) production by innate immune cells, promoting a resistance profile against Leishmania. The hallmark of IFN-γ leishmanicidal activity relies on the classical activation of infected macrophages, leading to increased production of nitric oxide (NO) and reactive oxygen species (ROS), which subsequently culminate in intracellular Leishmania elimination (13). By contrast, the Th2 profile is characterized by the production of IL-4, IL-5, and IL-13, which are mostly associated with enhanced arginase activity accompanied by the alternative activation of macrophages, parasite survival and proliferation, and pronounced susceptibility (13). Additionally, in recent decades, the contributions of the Th17 subtype on the progression of leishmaniasis has become a growing concern. The hallmark of the Th17 profile is the production of IL-17, and the subsequent recruitment of neutrophils to the site of inflammation. As reviewed elsewhere, the joint actions of this subtype paradoxically play a dual role in leishmaniasis, since these cells are not only responsible for the elimination of parasites, but also for the exacerbation of the inflammatory process and tissue damage (45). Crosstalk between Leishmania and DCs via the stimulation of various cellular apparatuses and the engagement of multiple signaling processes culminates in phenotypic and functional alterations in DCs. Such modifications are imperative for proper cytokine production and the activation of Th cells, which induce immune events that can result in parasite control (46).

# DENDRITIC CELL-LEISHMANIA INTERACTION

# Leishmanial Signals Prompt DCs Activation

One of the biological functions of DCs is to recognize molecular patterns associated with pathogens (PAMPs). To this end, DCs employ PRRs that interact with a variety of PAMPs expressed by distinct species of Leishmania. The activation of DCs can be substantially modulated by these interactions, which greatly influence the immune response against Leishmania (46).

#### The DC Recognition Apparatus: the Role of PRRs in Leishmania-DC Interplay

Toll-like receptors (TLRs) are germline-encoded immune receptors that play a pivotal role in the immunosurveillance function of DCs. These receptors are subdivided into 10 families in humans (TLR1 to TLR10) and 12 families in mice (TLR1 to TLR9, and TLR11 to TLR13) (47), and are expressed on either the cell membrane surface or in intracellular compartments. TLRs possess leucine-rich repeats that serve as molecular docking sites for ligand-receptor interactions. Upon ligand-mediated activation, TLRs undergo an intricate dimerization process that activates a variety of biochemical pathways, culminating in the transcription of several inflammatory genes. Of note, it has been proposed that TLRs may be central elements in the establishment of immune homeostasis, as these cells participate in the delicate balance between pro-inflammatory and antiinflammatory responses (48). However, despite their relevance in the recognition of several pathogens and the induction of immune responses, only TLR2, TLR4, and TLR9 have been described in the mediation of DC-Leishmania interaction.

It has been demonstrated that the neutralization of TLR2 and TLR4 decreases the expression of molecules involved in the process of antigen presentation during L. major infection, which suggests that both receptors may be key players in the establishment of effective responses against Leishmania (49). Interestingly, a study documented that TLR2 deficiency increases DC activation, leading to IL-12 production during L. braziliensis infection. On the other hand, a deficiency of MyD88 results in lower levels of DC activation and IL-12 production, both essential elements in mounting protective immunity against L. braziliensis (50).

TLR2 also recognizes Lipophosphoglycan (LPG), a surface molecule conserved in all Leishmania species that is considered an important virulence factor, especially due to its role in the modulation of immune cell activation (51). When LPG of L. mexicana is recognized by the TLR2 of moDCs, the expression of MHC-II and CD86 as well as the secretion of IL-12p70, are enhanced. Subsequently, the interaction of DCs with NKT cells culminates in higher IFN-γ production. This cellular interaction could possibly contribute to the protective state observed during the acute phase of L. mexicana infection (52).

TLR9 has been described as important in DC activation as well as in the production of neutrophil chemoattractant during L. infantum infection in C57BL/6 mice (53). Additionally, TLR9 is also required for the induction of IL-12 production in mouse bone marrow-derived DCs (BMDCs) infected with L. major, leading to both IFN-γ expression and cytotoxicity enhancement in NK cells (54). Collectively, these findings contribute to the understanding of how the intracellular TLRs in DCs mediate the stimulation of other immune cells that promote parasite eradication.

Different Leishmania species can mitigate the signaling pathways of CTLRs to promote parasite proliferation and survival. Iborra et al. showed that L. major releases a soluble protein ligand of Mincle (Macrophage inducible Ca2+ dependent lectin receptor) that targets an inhibitory ITAM signaling pathway, resulting in the impairment of DC activation and migration (55). In a similar vein, Zimara et al. experimentally demonstrated the importance of CTLRs in mounting an adaptive immune response against L. major. These researchers observed an increased expansion of Dectin-1<sup>+</sup> DCs following L. major inoculation in C57BL/6 and BALB/c mice as well as in the peripheral blood of CL patients. Additionally, experiments with BMDM stimulated with a Dectin-1 agonist revealed both high levels of DC maturation and the expansion of CD 4 T cells (56). Collectively, these findings serve to indicate the significance of both CTRL signaling and the physical interactions between these cells and pathogens in the promotion of an effective immune response.

#### Mechanisms of Leishmania Uptake

In addition to their importance as major mediators of the innate and adaptive branches of immunity, DCs are also recognized for their highly efficient phagocytic activity (57). These cells actively collect antigens in their surroundings, and couple subsequent antigen processing with epitope exhibition via antigen presentation platforms–the major histocompatibility complex molecules (58).

Typically, the mechanisms of capturing pathogens involve specific receptor-ligand interactions as well as the mobilization of cytoskeleton elements that promotes the internalization of parasites (59). Several studies have proposed that the uptake of Leishmania by DCs occurs in a parasite life form-dependent manner, since DCs preferentially phagocytose IgG -coated amastigotes. In fact, amastigotes internalization involves the participation of FcγRI and FcγRIII (60). It has been suggested that GP63, a protease found on leishmanial membranes, mediates the conversion of C3b into its inactive form, iC3b. Subsequently, iC3b binds to CR3, resulting in the adherence of leishmania to the surface membrane of DCs (60).

Argueta-Donohué et al. demonstrated that DC-SIGN (Dendritic Cell-Specific Intercellular adhesion molecule-3- Grabbing Non-integrin), a surface receptor mainly found in DCs, mediates a more efficient internalization rates of L. mexicana promastigotes after 3 h of in vitro infection (52). Moreover, these authors also confirmed that the experimental neutralization of DC-SIGN significantly reduces the rates of infection in moDCs. These intriguing findings illustrate the fundamental role of DC-SIGN in several instances of Leishmania-DC interplay, ranging from the initiation of parasite phagocytosis to discrimination between Leishmania life cycle stages (52). Additionally, mounting evidence indicates that DC-SIGN also recognizes L. pifanoi surface molecules, contributing to the subsequent uptake of parasite amastigotes (61). Furthermore, L. major and L. donovani- infected moDCs exhibit reduced surface expression of DC-SIGN in contrast to uninfected cells, with this immunomodulation being accentuated in cells stimulated with excreted-secreted antigens (ESA) of both Leishmania species (62). Together, these studies suggest that the consequences of the DC-SIGN-mediated crosstalk between Leishmania and host DCs may have profound biological consequences in Leishmania infection.

#### The Effects of Purinergic Receptors on DC Activation During Leishmania Infection

Purinergic receptors play a significant role in the recognition of damage-associated molecular pattern (DAMPs), including the detection of extracellular Adenosine Triphosphate (ATP), a potent pro-inflammatory trigger of immune responses (63). In pathophysiological contexts, ATP is converted into Adenosine (ADO) via the action of the ectonucleotidases CD39 and CD73, and the accumulation of ADO in the extracellular milieu results in the activation of its A2 receptor. This phenomenon has been observed during L. amazonensis infection, which was accompanied by the suppression of DC functions via decreased rates of CD40 expression and IL-12 production. Additionally, the activation of the A2b receptor of DCs decreases the capacity of these cells to stimulate T cell proliferation and the production of IFN-γ, leading to an insufficient protective immune response, a peculiarity of L. amazonensis infection (64). While increases in CD39 and CD73 expression are also observed in L. braziliensis and L. major infection, A2b receptor activation has not been detected. Interestingly, it has been proposed that the main evasion mechanism employed by these two species is reduced expression of the molecules involved in antigen presentation, which includes the exploitation of the IL-10 receptor (IL-10R). Notwithstanding, this evasion mechanism employed by L. braziliensis and L. major is followed by the upregulation of CD40, which may suggest that it does not prevent T cell activation (65).

Together, these findings provide evidence that early interactions between DCs and Leishmania can have profound effects on disease outcome. Several of the mechanisms of immune evasion employed by Leishmania include the mitigation of DC immunobiological functioning via the exploitation of different receptors and the disruption of downstream signaling pathways. In addition, recent data indicate that the impairment of DC activation is directly associated with the enhancement of parasite survival and persistence in hosts.

#### CO-STIMULATORY MOLECULES AND ANTIGEN PRESENTATION

Following the recognition and internalization of pathogens, DCs migrate to secondary lymphoid organs to present processed antigens to naïve T cells (66). Subsequently, the adaptive immune response becomes initiated via the presentation of small peptides through either MHC class I or class II molecules. Basically, the former class mediates the recognition of endogenous peptides by cytotoxic CD8+ T cells, while the latter is involved in the presentation of exogenous peptides to CD4+ T helper cells. Notably, the process of antigen fragmentation is of paramount importance to allow for proper antigen presentation, since MHCII molecules only present peptides with a specific number of amino acids (67). Alternatively, DCs are also capable of mobilizing MHC I molecules in order to display exogenously derived-antigens, a process known as cross-presentation (68). Additionally, co-stimulatory molecules (such as CD40, CD80, and CD86) are essential to effective antigen presentation, by providing secondary signals for T cell expansion and differentiation (69).

Several species of Leishmania employ distinct strategies to regulate the expression of co-stimulatory molecules, which dampens the process of antigen presentation (70). Accordingly, the modulation of co-stimulatory molecules can be associated not only with enhanced parasites survival and growth, but also with subsequent disease outcome.

Figueiredo et al. experimentally demonstrated that L. amazonensis induced lower levels of MHCII, CD86 and CD40 expression in BMDCs (bone marrow -derived DCs) from C57BL/6 mice, resulting in a decline in T-cell proliferation (65). Furthermore, the adoptive transfer of BMDCs expressing low levels of CD40 was associated not only with T regulatory cell expansion, but also with an increase in L. donovani burden in BALB/c mice (71). It has been also shown that CD40 and its ligand are important for the development of resistance against L. major infection (72, 73). Hai Qi et al. reported that L. amazonensis amastigotes mitigated IL-12 production in a CD40-dependent manner in a BALB/c infection model, which was followed by an increase in levels of IL-4 (74). Subsequently, amastigote-infected DCs were observed to be able to activate pathogenic CD4+ T cells, which could potentially lead to exacerbated Leishmania proliferation and the progression of pathogenesis (72). Thus, emerging evidence suggests that reduced CD40 expression could possibly facilitate Leishmania infection.

The importance of CD80 and CD86 expression has been highlighted in the establishment of early immune responses. For instance, the infection of human moDCs with L. amazonensis downregulates the expression of CD80 and upregulates the expression of CD86, which is followed by a decrease in IL-6 production during DC differentiation (75). Although CD86 possibly takes center stage in this context, the equivalent expression of other costimulatory molecules can lead to the early production of IFN-γ or IL-4 during infection by L. major depending on the experimental model (76). Together, these results reinforce the contribution of these molecules in the production of different cytokines by properly stimulated T cells.

Several species of Leishmania can modulate antigen processing and the expression of MHC II molecules (77). DCs infected with L. major amastigotes not only upregulate the expression of several molecules involved in antigen presentation,

such as MHC class II, CD40, CD54, CD80, and CD86, but also exhibit elevated rates of IL-12 production. It should be noted that only the amastigote forms of these parasites were capable of inducing this increase in L. major-infected DCs (78). Conversely, L. mexicana amastigotes do not promote increased expression of CD80, CD54, and MHC II molecules in BMDCs, suggesting that these discrepancies in the immune response by DCs occur in a species-specific manner (79).

Interestingly, a recent study by Resende et al. reported a dichotomic response between L. infantum-infected and noninfected DCs. In this study, the authors observed that uninfected DCs expressed higher levels of IL-12p40 and other co-stimulatory molecules, which enabled DCs to elicit appropriate CD4<sup>+</sup> T cell immunoprotective responses, whereas infected DCs expressed lower levels of co-stimulatory molecules and high IL-10 production (80). This finding suggests that L.infantum-infected DCs and their uninfected counterparts exert antagonistic roles in the activation and polarization of T cells, mechanistically revealing a novel evasion strategy employed by this species. Along the same lines, a study carried out by Carvalho et al demonstrated that, in contrast to L-braziliensis-infected DCs, only uninfected DCs upregulate the expression of MHC II, CD80, and CD86. Interestingly, it was also observed that despite enhancing the expression of such molecules, L-braziliensisinfected DCs produced higher levels of TNF-α in response to stimulation with LPS. These findings corroborate the hypothesis that uninfected and Leishmania-infected DCs can act conjointly, yet distinctly, to promote immune responses against the parasite, since uninfected DCs can lead to enhanced T cell activation, while the production of TNF-α by infected DCs may

contribute to the control of parasite proliferation at the site of infection (81).

Paradoxically, DCs are also able to present exogenous antigens through MHC-I, with significant consequences on the activation of CD8 T cells (82). It has been reported that this phenomenon, also known as cross-presentation, is of great importance to the expansion of antigen-specific cytotoxic CD8 T cells, which are responsible for eliciting an effective immune responses against Leishmania. Accordingly, in this context, DC figure as the most potent inducers of IFN-γ production by CD8+ T lymphocytes. Brewig et al. demonstrated that, in experimental leishmaniasis, the priming of CD4<sup>+</sup> and CD8<sup>+</sup> T cell relies essentially on the activity of distinct DC subtypes (83). Indeed, it was reported that the depletion of Langerin<sup>+</sup> DCs was associated with the reduced proliferation of L. major-specific CD8+ T cells. As a consequence, the amount of primed CD8 T cells found at the site of infection and in lymph nodes was significantly reduced (83). In a similar vein, a study conducted by Ashok et al revealed the importance of cross-priming DCs in the effective constraint of L. major infection. It was shown that Batf3−/<sup>−</sup> mice (which lack CD8+/CD103<sup>+</sup> DCs) exhibited increased susceptibility to L. major (84). Furthermore, a study by Lemos et al explored the function of CD8<sup>+</sup> DCs in antigen presentation during L. major infection in a murine model that restricted the expression of MHC-II to CD8a+/CD11b<sup>+</sup> DCs. Notably, it was observed that CD8a+/CD11b<sup>+</sup> DCs could efficiently restrain L. major infection by eliciting the effective constraint of parasites by CD4 T cells (85).

Numerous studies have revealed that the underlying mechanisms of antigen processing depend not only on the constitutive proteasome or the immunoproteosome, but also on the involvement of alternative molecular machineries of cytosolic degradation, such as tripeptidyl peptidase II (TPPII) and nardilysin (86–88). TPPII is a known eukaryotic peptidase related to several cellular processes, such as antigen processing, apoptosis and cell division. However, it should be noted that TPPII activation occurs mainly when proteasome function becomes compromised (89).

Although little is known about the detailed mechanisms of cytosolic endopeptidases, such as Nardilysin, their role seems to be indispensable in the generation of some specific epitopes (90). The importance of alternative antigen processing machinery should be further investigated in the context of leishmaniasis.

Although the participation of CD8<sup>+</sup> T cells in Leishmania infection is still controversial, growing evidence indicates that protective responses rely substantially on the effective dendritic cell-mediated activation of cytotoxic lymphocytes (86, 87, 91, 92).

### LEISHMANIA AFFECTS THE MIGRATION OF DCS

Immature DCs strategically reside in peripheral tissues, where they exercise their main function as immune guards. As discussed previously, these sentinels specialize in antigen uptake via their apparatus to internalize foreign particles. In peripheral tissues, PAMP-mediated activation confers an immunostimulatory phenotype to DCs, characterized by the upregulation of molecules also associated with an enhanced migratory ability. Subsequently, DCs migrate toward lymph nodes, where they exchange information with naïve T cells via the antigen presentation process (93).

In order to ensure precise mobilization, the migration of DCs needs to be highly coordinated and regulated by particular recruitment signals (94). The primary mechanisms of DC migration involve the cooperative action of chemokines and their receptors. Chemokine receptors are typically transmembrane proteins associated with G-proteins whose activation triggers signaling pathways responsible for the promotion of cell mobilization (95). Some evidence seems to suggest that DC subtypes exhibit diverse chemokine receptors, conferring subtype-specific migration dynamics. Commonly, immature DCs express CCR2, CCR5, CCR6, CXCR2, and CXCR4 in a predominant fashion. Upon pathogen-mediated activation, DCs undergo a maturation processes that culminates in the crucial upregulation of CCR7 (**Figure 2A**) (96).

By way of evolution, protozoan parasites developed strategies to mitigate DC functioning by inhibiting access to T cells, thusly restricting the establishment of efficient adaptive immune responses. In the context of flagellate protozoan infection, successful DC migration to draining lymph nodes (dLNs) is substantially dependent on both CCR2 and CCR7 expression (97). Several studies have reported that Leishmania can induce a reduction in rates of DC migration (98, 99). In vitro studies have elucidated the roles of both soluble products and membrane constituents, such as L. major LPG, in the inhibition of DC motility (98, 100). The underlying molecular mechanisms of Leishmania-induced mitigation of DC motility remain elusive. A recent study suggested that L. major exploits the junction adhesion molecule C (JAM-C) to reduce DC migration rates, and demonstrated that the experimental blockage of this molecule enhanced the immunological response against this parasite (101). An investigation in an animal model exhibited an L. majorsusceptible phenotype, suggesting that the depletion of CCR2 culminated in poor DC migration and a skewed Th2 immune response (102). Furthermore, infection with L. donovani, a viscerotropic species, promotes an inhibition in the expression of CCR7 mediated by IL-10 production, which ensures that DCs will not be able to reach splenic regions, thereby contributing to the progression of visceral leishmaniasis (**Figure 2B**) (103). Another in vivo study associated CCL19/CCL21 deficiency with a reduction in both DC mobility and resistance to L. donovani infection (103).

Collectively, the current data suggest that efficacious DC migration is essential to the establishment of effective responses against Leishmania parasites. The literature clearly indicates that these parasites employ a plenitude of strategies to prevent DCs from activating T cells during the course of several clinical forms of leishmaniasis. Deciphering the complex dynamics surrounding the Leishmania-mediated impairment of DC mobilization will provide new insights into the evasion mechanisms employed by these parasites and elucidate their effects on the immunopathogenesis of leishmaniasis.

# METABOLIC REPROGRAMMING DURING DC ACTIVATION

Faced with infection and inflammation, DCs must cope with increasing catabolic and anabolic demands via the redirection of a plethora of metabolic pathways to support their major immune functions (104). Typically, the metabolism of inactive DCs is characterized by the central roles of oxidative phosphorylation (OXPHOS) and fatty acid oxidation (FAO), for energy supply and biomolecule synthesis, respectively (105). New evidence suggests that, after the initiation of PAMP-mediated activation, DCs undergo metabolic reprogramming, relying substantially on anaerobic glycolysis for ATP production, a process characterized by the conversion of pyruvate into lactate. Despite being ineffective in the generation of ATP, glycolysis can be coupled with several anabolic pathways, such as fatty acid synthesis and the pentose phosphate pathway, allowing for the biosynthesis of other macromolecules, namely lipids and nucleotides, respectively (106, 107). In this scenario, DCs exhibit low rates of oxidative phosphorylation. These deviations in the metabolic repertoire of DCs are prominent regulators of immune responses, as metabolic enzymes and their products can influence the establishment of inflammation (108).

In this inflammatory milieu, immune cells are poorly supplied with oxygen and nutrients for their metabolic processes, leading to the activation of hypoxia-inducible transcription factor 1α (HIF-1α) (109). Recently, HIF-1α was recognized as a major player in the induction of glycolysis, since it promotes the transcription of several enzymes involved in glucose metabolism (110). Nevertheless, its expression was shown to favor L. donovani infection in a model of chronic visceral leishmaniasis, as increased HIF-1a expression in murine splenic DCs was correlated with decreased IL-12 production, allowing parasite survival through limited Th1 cell expansion (111). In consonance with these observations, Hammani et al. demonstrated the importance of the IRF-5/HIF1α transcription factor axis in the impairment of DCs to promote the expansion of CD8<sup>+</sup> T cells (112). Conversely, in vitro experiments showed that HIF1α enhanced both Leishmania major elimination and levels of NO production in macrophages (**Figure 3**) (113). Together, while these observations suggest that HIF1α downregulates some DC functions against Leishmania, this effect may be cell-specific. Moreover, a recent study highlighted the contribution of two energetic sensors, Sirtuin 1 (SIRT1) and AMP-activated protein kinase (AMPK), to L. infantum survival and replication in macrophages (114).

Given the relevance of the metabolic processes of DCs in supporting the immunobiological functioning of these cells, it is unsurprising that an increasing number of studies have contemplated this interesting topic in recent years. Nevertheless, few studies have attempted to investigate the role of DC immunometabolism in Leishmania infection. Currently, the molecular players involved in metabolic reprogramming and the mechanistic basis of immunometabolism continue to remain elusive in the context of leishmaniasis.

#### EFFECTS OF EPIGENETIC MODIFICATIONS ON DC DEVELOPMENT AND COURSE OF INFECTION

Interactions between host cells and parasites prompt several alterations in a range of biological processes occurring in DCs, including epigenetic alteration via modified gene expression. This phenomenon is not dependent on DNA sequence modifications and includes DNA methylation, histone modifications, chromatin remodeling and regulation by non-coding RNAs (115–117).

The activation of transcription factors is one of the major regulatory elements occurring in epigenetic alterations (118, 119). PU.1 transcription factor has been described as an essential TF for the development and functioning of DCs, as evidenced by the expression of FLT3, granulocyte-macrophage colonystimulating factor receptor (GM-CSFR) and macrophage colony stimulating factor receptor (M-GSFR) (119, 120). It has been demonstrated that PU.1 is also involved in the regulation of basal expression of DC-SIGN, which in turn influences the repertoire of antigen uptake in DCs (121). PU.1 can also regulate the promoter region of genes CD80 and CD86 in murine bone marrow–derived DCs, leading to the overexpression of these costimulatory molecules, thereby enhancing DC migration and the activation of T cells (122).

In face of tissue damage or infection, several modifications in the histones alter chromatin conformation, leading to changes in the expression profile of critical genes in specific DC subsets (123, 124). Tserel et al. showed by GWAS (Genome-wide Association Study) that histone modifications can influence the processes of differentiation, phagocytosis and antigen presentation in moDCs through the upregulation of surface marker expression and chemokine production. Similar findings have been reported in macrophages, reinforcing the similarity of epigenetic mechanisms in the development of both cell types (125).

The importance of epigenetic changes in DCs infected by Leishmania remains unclear. However, L. donovani infection in macrophages was shown to lead to changes in the methylation

#### REFERENCES

1. Alvar J, Vélez ID, Bern C, Herrero M, Desjeux P, Cano J, et al. Leishmaniasis worldwide and global estimates of its incidence. PLoS ONE (2012) 7:e35671. doi: 10.1371/journal.pone.0035671

of CpGs sites via parasite exosome secretion, which may enhance parasite replication and survival (126). Furthermore, L. amazonensis infection promoted epigenetic modifications at the IL-10 locus in murine macrophages, which activated ERK1/2 pathways and promoted parasite survival (127). In addition, this parasite species can upregulate histone deacetylases (HDACs), which enhances iNOS promoters in macrophages, thusly favoring infection (128).

Taken together, in addition to playing a crucial role in the development of DCs, these findings seem to suggest that interactions between Leishmania and immune cells can trigger epigenetic modifications that may alter the course of the infection. However, much remains to be elucidated with regard to this topic.

#### CONCLUSION REMARKS

DCs are relevant immunological agents in the concatenation of innate and adaptative branches of immunity. Here, we have attempted to integrate recent advances in molecular aspects of the immunobiological functioning of DCs with the current state of understanding regarding the pathogenic mechanisms of leishmaniasis. Although a large body of evidence supports the central role of DC activation in the establishment of responses against Leishmania parasites, the overwhelming complexity of Leishmania-DC interactions impedes the attainment of a comprehensive understanding of the molecular processes involved in DC activation. Further clarification is required to unravel the interplay between different DC subtypes and different species and life cycle stages of Leishmania as well as how parasites subvert particular aspects of DC activation in the effort to successfully establish infection. Finally, an enhanced understanding of the fundamental molecular events underlying DC activation will lead to the expansion of our current base of knowledge surrounding leishmaniasis as well as offer new therapeutic targets.

#### AUTHOR CONTRIBUTIONS

RT and SN designed the review and wrote the manuscript. IN, MR, IS, and RL assembled the review and wrote the manuscript. NM and CB supervised the work, designed the review, and wrote the manuscript.

#### FUNDING

This review was supported by CNPq, CAPES, and Fapesb. These financial agencies are responsible for the payment of fellowships of CB, RT, SN, IN, IS, RL, and NM.


of plasmacytoid dendritic cell development. Cell (2008) 135:37–48. doi: 10.1016/j.cell.2008.09.016


absence of CCR2 shifts the Leishmania major-resistant phenotype to a susceptible state dominated by Th2 cytokin. J Exp Med. (2000) 192:205–18. doi: 10.1084/jem.192.2.205


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Tibúrcio, Nunes, Nunes, Rosa Ampuero, Silva, Lima, Machado Tavares and Brodskyn. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Neoadjuvant Radiochemotherapy Significantly Alters the Phenotype of Plasmacytoid Dendritic Cells and 6-Sulfo LacNAc<sup>+</sup> Monocytes in Rectal Cancer

#### Edited by:

Silvia Beatriz Boscardin, University of São Paulo, Brazil

#### Reviewed by:

Shohei Koyama, Osaka University, Japan Veronika Lukacs-Kornek, Saarland University, Germany

\*Correspondence:

Marc Schmitz marc.schmitz@tu-dresden.de

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

> Received: 28 August 2018 Accepted: 06 March 2019 Published: 29 March 2019

#### Citation:

Wagner F, Hölig U, Wilczkowski F, Plesca I, Sommer U, Wehner R, Kießler M, Jarosch A, Flecke K, Arsova M, Tunger A, Bogner A, Reißfelder C, Weitz J, Schäkel K, Troost EGC, Krause M, Folprecht G, Bornhäuser M, Bachmann MP, Aust D, Baretton G and Schmitz M (2019) Neoadjuvant Radiochemotherapy Significantly Alters the Phenotype of Plasmacytoid Dendritic Cells and 6-Sulfo LacNAc<sup>+</sup> Monocytes in Rectal Cancer. Front. Immunol. 10:602. doi: 10.3389/fimmu.2019.00602 Felix Wagner 1†, Ulrike Hölig1†, Friederike Wilczkowski 1†, Ioana Plesca1,2, Ulrich Sommer <sup>3</sup> , Rebekka Wehner 1,4,5, Maximilian Kießler <sup>1</sup> , Armin Jarosch<sup>3</sup> , Katharina Flecke1,6 , Maia Arsova1,6, Antje Tunger 1,4, Andreas Bogner <sup>7</sup> , Christoph Reißfelder <sup>8</sup> , Jürgen Weitz 4,5,7, Knut Schäkel <sup>9</sup> , Esther G. C. Troost 2,4,5,10,11, Mechthild Krause2,4,5,10,11 , Gunnar Folprecht 4,5,6, Martin Bornhäuser 4,5,6, Michael P. Bachmann4,5,12, Daniela Aust 3,4,5 , Gustavo Baretton3,4,5 and Marc Schmitz 1,4,5 \*

1 Institute of Immunology, Faculty of Medicine Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany, <sup>2</sup> Department of Radiotherapy and Radiation Oncology, University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany, <sup>3</sup> Institute of Pathology, University Hospital of Dresden, Dresden, Germany, <sup>4</sup> Partner Site Dresden, National Center for Tumor Diseases (NCT), Dresden, Germany, <sup>5</sup> Partner Site Dresden, German Cancer Consortium (DKTK), and German Cancer Research Center (DKFZ), Heidelberg, Germany, <sup>6</sup> Department of Medicine I, University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany, <sup>7</sup> Department of Gastrointestinal, Thoracic, and Vascular Surgery, University Hospital Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany, <sup>8</sup> Department of Surgery, Mannheim University Medical Centre, University of Heidelberg, Mannheim, Germany, <sup>9</sup> Department of Dermatology, University Hospital of Heidelberg, Heidelberg, Germany, <sup>10</sup> OncoRay – National Center for Radiation Research in Oncology, Dresden, Germany, <sup>11</sup> Institute of Radiooncology – OncoRay, Helmholtz-Zentrum Dresden-Rossendorf, Dresden, Germany, <sup>12</sup> Department of Radioimmunology, Institute of Radiopharmaceutical Cancer Research, Helmholtz Center Dresden-Rossendorf, Dresden, Germany

Neoadjuvant radiochemotherapy (nRCT) can significantly influence the tumor immune architecture that plays a pivotal role in regulating tumor growth. Whereas, various studies have investigated the effect of nRCT on tumor-infiltrating T cells, little is known about its impact on the frequency and activation status of human dendritic cells (DCs). Plasmacytoid DCs (pDCs) essentially contribute to the regulation of innate and adaptive immunity and may profoundly influence tumor progression. Recent studies have revealed that higher pDC numbers are associated with poor prognosis in cancer patients. 6-sulfo LacNAc-expressing monocytes (slanMo) represent a particular proinflammatory subset of human non-classical blood monocytes that can differentiate into DCs. Recently, we have reported that activated slanMo produce various proinflammatory cytokines and efficiently stimulate natural killer cells and T lymphocytes. slanMo were also shown to accumulate in clear cell renal cell carcinoma (ccRCC) and in metastatic lymph nodes from cancer patients. Here, we investigated the influence of nRCT on the frequency of rectal cancer-infiltrating pDCs and slanMo. When evaluating rectal cancer tissues obtained from patients after nRCT, a significantly higher density of pDCs in comparison to pre-nRCT tissue samples was found. In contrast, the density of slanMo was not significantly altered by nRCT. Further studies revealed that nRCT significantly enhances the proportion of rectal cancer-infiltrating CD8<sup>+</sup> T cells expressing the cytotoxic effector molecule granzyme B. When exploring the impact of nRCT on the phenotype of rectal cancer-infiltrating pDCs and slanMo, we observed that nRCT markedly enhances the percentage of inducible nitric oxide synthase (iNOS)- or tumor necrosis factor (TNF) alpha-producing slanMo. Furthermore, nRCT significantly increased the percentage of mature CD83<sup>+</sup> pDCs in rectal cancer tissues. Moreover, the proportion of pDCs locally expressing interferon-alpha, which plays a major role in antitumor immunity, was significantly higher in post-nRCT tissues compared to pre-nRCT tumor specimens. These novel findings indicate that nRCT significantly influences the frequency and/or phenotype of pDCs, slanMo, and CD8<sup>+</sup> T cells, which may influence the clinical response of rectal cancer patients to nRCT.

Keywords: plasmacytoid dendritic cells, 6-sulfo LacNAc<sup>+</sup> monocytes, CD8<sup>+</sup> T cells, tumor immune architecture, radiochemotherapy, rectal cancer

#### INTRODUCTION

Colorectal cancer is one of the most common malignancies in the United States with an estimated incidence of 140,250 cases and an estimated number of 50,630 deaths in 2018 (1). Previous reports have provided evidence that the immune contexture plays a major role for the clinical outcome of colorectal cancer patients (2). Thus, it has been shown that high densities of CD45RO<sup>+</sup> T helper (Th) 1 cells and CD8<sup>+</sup> T cells are associated with improved survival of colorectal cancer patients (3, 4). However, patients with high expression of Th17 genes had a poor prognosis. These results led to the development of a so-called "immunoscore" for an optimized tumor classification (5).

Neoadjuvant radiochemotherapy (nRCT) followed by total mesorectal excision constitutes the current standard of care for locally advanced rectal cancers (6, 7). nRCT can efficiently reduce tumor size, resulting in a higher rate of sphincter-preserving surgical interventions, and an increased rate of R0-resections. In addition, this treatment modality decreases the local recurrence rate. Recent findings have revealed that nRCT can significantly influence the tumor immune contexture, affecting the tumor responsiveness to this treatment modality (2, 8–11). Thus, it has been reported that several chemotherapeutic agents as well as radiotherapy can efficiently stimulate antitumor immune responses by triggering immunogenic cell death in tumor cells (12, 13). This process is characterized by the translocation of intracellular calreticulin to the surface of tumor cells and the release of high-mobility-group box 1. Surface-exposed calreticulin markedly enhances the phagocytosis of tumor cells by dendritic cells (DCs) that play a crucial role in the induction and regulation of antitumor immunity (14). Released high-mobilitygroup box 1 from chemo- or radiotherapy-treated tumor cells promote the maturation and activation of DCs, resulting in an efficient processing and presentation of tumor-associated antigens and the stimulation of potent tumor-directed T-cell responses. In addition to immunogenic cell death induction, radiotherapy has been shown to reduce the surface expression of CD47, which acts as an antiphagocytosis signal to promote immune evasion (15). Inhibition of CD47 function significantly augments the engulfment of tumor cells by DCs, resulting in effective antitumor responses (16). In contrast to these immunostimulatory effects, radiotherapy, and chemotherapeutic agents can also induce immunosuppressive effects. They include the increase of tumor-infiltrating regulatory T cells and myeloidderived suppressor cells, which produce immunosuppressive molecules (8, 9, 17). These immune cell subsets can profoundly impair the functional properties of effector T cells and can promote tumor growth and resistance.

Plasmacytoid DCs (pDCs) represent an important subset of human blood DCs that are capable of producing large amounts of interferon (IFN)-α upon stimulation (18, 19). In addition, pDCs can efficiently enhance the antitumoral capabilities of natural killer (NK) cells and T lymphocytes (20, 21). Based on these functional properties, stimulated pDCs can promote antitumor responses in vivo. Thus, the intratumoral administration of activated pDCs led to tumor regression in a B16 melanoma mouse model (22). Furthermore, it has been demonstrated in an orthotopic murine mammary tumor model that the intratumoral application of a toll-like receptor 7 ligand results in the activation of tumor-associated pDCs and tumor regression (23). In a clinical trial, intranodal injections of activated pDCs loaded with tumor-associated antigen-derived peptides in patients with metastatic melanoma induced specific CD8<sup>+</sup> and CD4<sup>+</sup> T cell responses (24). However, pDCs can also act as tolerogenic cells by suppressing T cell responses (19, 25). Previous studies have shown that pDCs infiltrate a variety of human cancers including head and neck, breast and ovarian cancer, and that a higher density is associated with poor clinical outcome (26–28). In addition, it has been demonstrated that tumor-infiltrating pDCs produce reduced amounts of IFN-α upon activation and can efficiently promote the expansion of regulatory T cells (29). 6-sulfo LacNAc (slan) monocytes (slanMo, formerly termed

M-DC8<sup>+</sup> DCs or slanDCs) are a subset of human non-classical blood monocytes, which can differentiate into DCs (30–40). Previously, we have reported that slanMo produce high levels of various proinflammatory cytokines and display a marked capability to handle IgG-complexed antigens (31, 32, 37). We have also demonstrated that slanMo mediate direct cytotoxicity against tumor cells (38, 39). Further studies have revealed that they efficiently induce neoantigen-specific CD4<sup>+</sup> T cells, activate tumor-reactive CD8<sup>+</sup> T cells, and promote the polarization of naïve CD4<sup>+</sup> T lymphocytes into Th1 or Th17/Th1 cells (30– 32, 36). Moreover, slanMo have been shown to stimulate IFN-γ production and cytotoxic activity of NK cells (39, 40).

In the present study, we determined the impact of nRCT on the frequency of rectal cancer-infiltrating pDCs, slanMo, CD3<sup>+</sup> T cells, total CD8<sup>+</sup> T lymphocytes, and GrzB-expressing CD8<sup>+</sup> T cells. Furthermore, we analyzed the impact of nRCT on the percentage of rectal cancer-associated slanMo locally producing inducible nitric oxide synthase (iNOS) or tumor necrosis factor (TNF)-α, which play an important role in regulating tumor growth. Following recent findings, indicating that IFNα essentially contributes to the antitumor effects mediated by RCT, the influence of nRCT on the proportion of rectal cancerinfiltrating pDCs locally expressing this cytokine was evaluated.

# MATERIALS AND METHODS

#### Patients and Study Design

This is a retrospective study including 60 rectal cancer patients treated with nRCT followed by surgery at the University Hospital Carl Gustav Carus of Dresden between 2001 and 2013. From 20 of these patients, tumor biopsies prior to nRCT were available. Additionally, a cohort of 28 primarily resected rectal cancer patients without nRCT was matched according to gender, age, and TNM-stage. **Table 1** summarizes the clinicopathological characteristics of the study population.

#### Immunohistochemistry

Formalin-fixed and paraffin-embedded tissue sections were cut into 3–5µm sections. Subsequently, these sections were deparaffinized in xylene (2 × 15 min, VWR International, Fontenay-sous-Bois, France) and hydrated by washes of graded ethanol (Berkel AHK, Ludwigshafen, Germany) to water (B. Braun, Melsungen, Germany). Tissue sections were boiled in citrate buffer (Zytomed Systems GmbH, Berlin, Germany) at pH 6.0 for 20 min for antigen retrieval. Subsequently, tissues were stained overnight at 4◦C with either the polyclonal goat anti-BDCA-2 antibody (1:200, R&D Systems, Minneapolis, MN, USA) to evaluate pDCs (41) or the monoclonal mouse anti-slan antibody DD2 (1:10, Institute of Immunology, Faculty of Medicine Carl Gustav Carus, Technische Universität Dresden, Dresden, Germany) to analyze slanMo (32, 34–36). Then, tissues used for pDC staining were incubated with a mouse anti-goat antibody solution (Thermo Fisher Scientific, Rockford, IL, USA) for 60 min. Afterwards, all tissues were incubated with dextran-labeled antibodies against mouse immunoglobulins (Dako, Glostrup, Denmark) for 30 min. pDCs and slanMo were visualized by the alkaline phosphatase-based EnVisionTM detection system according to the manufacturer's instructions (Dako). All tissue sections were counterstained with Mayer's hematoxylin (Merck, Darmstadt, Germany). For pDC quantification, positively stained cells were counted in three different high power fields (HPF) of a section by using AxioVision 4.8.1.0 (Zeiss, Jena, Germany) and the mean value was determined. The mean number of pDCs per HPF (area: 0.237 mm²) was converted to square millimeter. For slanMo, slides were digitized by an iScan Coreo slide scanner (Ventana Medical Systems, Oro Valley, AZ, USA) and evaluated using the same HPF method.

To determine the frequency of rectal cancer-infiltrating CD3<sup>+</sup> T cells, CD8<sup>+</sup> T cells, and granzyme B (GrzB)-expressing CD8<sup>+</sup> T cells, formalin-fixed, and paraffin-embedded tissue sections were deparaffinized in xylene BenchMark XT (Ventana Medical Systems) and then exposed to the Cell Conditioning 1 solution for antigen retrieval (Ventana Medical Systems). Two double immunohistochemical stainings were performed: CD3 / Ki67 and CD8 / GrzB. For the first double reaction, the monoclonal mouse anti-CD3 antibody (clone 2GV6, readyto-use, Ventana Medical Systems) and the monoclonal mouse anti-Ki67 antibody (clone Mib-1, 1:50, Dako) were used. For the second double staining, the monoclonal mouse anti-CD8 antibody (clone C8/144B, 1:10, Dako) and the monoclonal mouse anti-GrzB antibody (clone GrzB-7, 1:10, Dako) were applied. All tissue sections were counterstained with Mayer's hematoxylin. Subsequently, the tissue sections were digitized by an iScan Coreo slide scanner, followed by T-cell quantification by using the Image viewer v. 3.1 (Ventana Medical Systems). Positively stained T lymphocytes were counted in three different HPF of a section and the mean value was determined. The mean number of T cells per HPF (area: 0.237 mm²) was converted to square millimeter. To determine the percentage of GrzB-expressing CD8<sup>+</sup> T lymphocytes, between 65 and 576 CD8<sup>+</sup> T cells per tissue section were evaluated dependent on their frequency in the three HPF.

# Immunofluorescence Staining

Tissues were deparaffinized, hydrated, and heat-treated as described above. After antigen retrieval, tissue sections were incubated overnight with primary antibody solutions containing goat anti-human BDCA-2 (1:50, R&D Systems) and either mouse anti-human CD83 (clone 1H4b, 1:100, Abcam, Cambridge, UK) or mouse anti-human IFN-α (clone F-7, 1:500, Santa Cruz Biotechnology, Heidelberg, Germany). Subsequently, tissue sections were incubated with a rabbit anti-goat antibody solution (1:100, Abcam) for 10 min. Afterwards, secondary antibody solution containing donkey anti-rabbit AF488 (1:100, Abcam) and donkey anti-mouse AF546 (1:100, Thermo Fisher Scientific) was applied for 30 min.

For staining of slanMo, tissue sections were incubated overnight with primary antibody solutions containing rabbit anti-human TNF-α (1:100, Abcam) and mouse anti-human DD2 (1:20), followed by 30 min of incubation with the secondary antibodies goat anti-mouse IgM biotin (1:100, SouthernBiotech, Birmingham, AL, USA) and fluorescence-labeled goat antirabbit IgG AF488 (1:100, Thermo Fisher Scientific). Finally,



Streptavidin AF546 (1:500, Thermo Fisher Scientific) was applied for 15 min. To determine iNOS expression in rectal cancerinfiltrating slanMo, tissue sections were incubated with a mouse anti-human iNOS antibody (1:50, BD Biosciences, San Jose, CA, USA) for 60 min, followed by the application of a goat antimouse IgG (1:400, Abcam) and fluorescence-labeled donkey antigoat AF488 antibody (1:100, Thermo Fisher Scientific), each for 20 min. Mouse serum (1:100, Dako) was applied for 10 min to prevent unspecific binding of the following antibody. For the detection of slanMo, mouse anti-human DD2 antibody (1:2) was applied for 60 min. Afterwards the tissues were incubated with a secondary antibody solution containing goat anti-mouse IgM Biotin (1:100, SouthernBiotech), followed by the application of fluorophore AF546 labeled Streptavidin (1:500, Thermo Fisher Scientific), each for 20 min. Then, tissues were mounted with 4,6 diamidino-2-phenylindole-containing AKLIDES <sup>R</sup> ANA plus medium (Medipan, Dahlewitz/Berlin, Germany), coverslipped, and analyzed with a Keyence fluorescence microscope BZ-9000 (Keyence, Osaka, Japan). To determine the percentage of CD83<sup>+</sup> and IFN-α <sup>+</sup> pDCs or iNOS<sup>+</sup> and TNF-α <sup>+</sup> slanMo, between 20 and 50 cells per tissue section were evaluated dependent on their frequency.

For additional experiments, immunofluorescence multiplex staining was accomplished by using the Opal kit and the Vectra imaging platform (Perkin Elmer, Hopkinton, MA, USA). Tissues were deparaffinized and hydrated as described above. Antigen retrieval was performed in AR6 or AR9 buffer (both from PerkinElmer) using microwave treatment. Afterwards, the Opal kit was used according to the manufacturer's instructions (PerkinElmer). Therefore, tissue sections were blocked for 10 min with the Antibody Dilutent/Block (PerkinElmer), then incubated with the primary antibody for one hour, followed by 10 min incubation with a horseradish peroxidase-conjugated secondary antibody (PerkinElmer). In case of goat antihuman primary antibodies, another 10 min incubation with a bridge mouse anti-goat antibody (1:100, Thermo Fisher Scientific) was required prior to the application of the secondary antibody. Finally, a TSA fluorophore was added to the tissue sections for 10 minutes. Subsequent stripping of the primary together with secondary antibodies was performed by microwave treatment. In between all the steps mentioned above, except prior to the primary antibody application, tissue sections were washed for 2 x 3 min in TBST buffer. Every incubation step took place in a humidified chamber on a rocking platform at room temperature. Blocking, incubation with primary and secondary antibody, visualization by a TSA fluorophore and microwave treatment were repeated for each primary antibody. Finally, all tissue slides were counterstained with spectral DAPI (PerkinElmer) for 5 min, washed with TBST buffer and with autoclaved water for 2 min, and then coverslipped with fluoromount medium (SouthernBiotech).

For the four-color Opal multiplex staining, primary antibodies directed against CD8 (clone C8/144B, 1:100, Dako, high pH retrieval), BDCA-2 (goat polyclonal, 1:100, R&D Systems, low pH retrieval), and panCK (clone AE1/AE3, 1:100, Thermo Fisher Scientific, low pH retrieval) were used and visualized by the TSA fluorophores 570 (1:100), 650 (1:100), and 690 (1:100, all from PerkinElmer), respectively. For the two-color Opal staining, the BDCA-2 primary antibody was used together with the polyclonal goat anti-human CXCL10, CCL4, or CCL5 primary antibody (1:100, all from R&D Systems, low pH retrieval) and combined with the TSA fluorophores 650 (1:100) and 570 (1:100), respectively. Acquisition of the multispectral images was performed with the Vectra 3.5 Automated Imaging System (Perkin Elmer). Spectral unmixing was done in inForm <sup>R</sup> using a library built from single stained tissue slides for each primary antibody-TSA fluorophore combination. ImageJ software was then used for final image processing.

# Statistical Analysis

Statistical analysis was performed using unpaired student's t-test for the evaluation of pDCs and slanMo in non-matched tissues of untreated and nRCT-treated rectal cancer patients. Paired student's t-test was used for the analysis of pDCs, slanMo, and T cells in matched pre-nRCT and post-nRCT tumor tissues. Values of <sup>∗</sup>p ≤ 0.05; ∗∗p ≤ 0.01; ∗∗∗ p ≤ 0.001 were considered as significant.

# RESULTS

### nRCT Significantly Increases the Frequency of pDCs in Rectal Cancer

pDCs essentially contribute to the regulation of innate and adaptive immunity and may play an important role in the immune defense against tumors. Recently, it has been demonstrated that pDCs are present in a variety of primary

human tumor entities (26–28). However, little is known about the impact of nRCT on the density of tumor-infiltrating pDCs. Here, we address this issue by analyzing the frequency of pDCs in paraffin-embedded tissue specimens from non-treated (n = 28) and nRCT-treated (n = 40) rectal cancer patients (non-matched) with different clinicopathological characteristics (**Table 1**). BDCA-2<sup>+</sup> pDCs were present in all non-treated and nRCT-treated rectal cancer tissue specimens (**Figures 1A–D** and data not shown) at varying frequencies. pDCs were preferentially located in the tumor stroma, where they were not equally distributed. Whereas some regions are characterized by an accumulation of pDCs (**Figures 1A–D**), other regions contained only low pDC numbers. Interestingly, the number of pDCs was significantly higher in the nRCT-treated cohort (219.3 ± 106.6 pDCs/mm²) than in the untreated cohort (160.3 ± 48.7 pDCs/mm²) as depicted in **Figure 2A**. In further experiments, matched pre-nRCT and post-nRCT tumor specimens from 18 rectal cancer patients were analyzed. pDC infiltration increased >10% after nRCT in 15 out of 18 patients (**Figure 2B**). As shown in **Figure 2C**, the frequency of pDCs was significantly higher in the nRCT-treated cohort (172.0 ± 89.6 pDCs/mm²) in comparison to the untreated cohort (99.2 ± 44.1 pDCs/mm²), confirming the results obtained with the non-matched cohorts.

#### nRCT Does Not Significantly Modulate the Frequency of Rectal Cancer-Infiltrating slanMo

Previous studies have demonstrated that slanMo accumulate in primary tumor tissues of clear cell renal cell carcinoma (ccRCC) patients and in metastatic lymph nodes from cancer patients (42, 43). In the present study, we investigated whether infiltrating slanMo are detectable in rectal cancer tissues and whether nRCT can modulate their frequency. When analyzing the tissue specimens from non-treated and nRCT-treated rectal cancer patients (non-matched), we found that slanMo are present in 27 out of 28 non-treated and in all 40 nRCT-treated tissues (**Figures 3A–D** and data not shown) at varying frequencies. slanMo were preferentially located in the tumor stroma. In contrast to pDCs, nRCT only slightly increased the density of slanMo in the treated cohort (18 ± 13.8 slanMo/mm²) compared to the untreated (13.4 ± 7.8 slanMo/mm²) cohort (**Figure 3E**). These findings were confirmed when evaluating matched pre-nRCT and post-nRCT tumor specimens from 20 rectal cancer patients. Again, nRCT did not significantly modulate the frequency of rectal cancer-infiltrating slanMo (pre-nRCT: 12.4 ± 7.4 slanMo/mm², post-nRCT: 14.6 ± 6.6 slanMo/mm²) as depicted in **Figure 3F**.

# nRCT Significantly Enhances the Proportion of GrzB-Expressing CD8<sup>+</sup> T Cells in Rectal Cancer

Recent studies have demonstrated that tumor-infiltrating T cells play an important role for the clinical outcome of colorectal cancer patients (2–5). Based on these findings, we explored the impact of nRCT on the frequency of rectal cancer-infiltrating

frequency of infiltrating pDCs in tissue specimens from non-matched, untreated and nRCT-treated rectal cancer patients. (B,C) In addition, pDC density in matched pre-nRCT or post-nRCT tumor specimens from 18 rectal cancer patients was analyzed. (A) The number of pDCs in non-matched untreated (n = 28) compared to nRCT-treated (n = 40) rectal cancer tissues is depicted. The results are presented as mean value ± SD of rectal cancer-infiltrating pDCs. Asterisks indicate a statistically significant difference (\*\*p < 0.01). (B) pDC number in 18 matched pre-nRCT or post-nRCT tumor specimens is depicted for each patient. (C) pDC frequency in 18 matched pre-nRCT or post-nRCT tumor specimens is presented as mean value ± SD. Asterisks indicate a statistically significant difference (\*\*p < 0.01).

CD3<sup>+</sup> T lymphocytes, total CD8<sup>+</sup> T lymphocytes, and GrzBexpressing CD8<sup>+</sup> T cells in matched pre-nRCT and post-nRCT tumor samples of 18 patients. As demonstrated in **Figure 4A**, the number of rectal cancer-infiltrating CD3<sup>+</sup> T cells was not significantly altered by nRCT (pre-nRCT: 1598.1 ± 842.4 CD3<sup>+</sup> T cells/mm², post-nRCT: 1228.8 ± 671.6 CD3<sup>+</sup> T cells/mm²). In contrast, nRCT significantly increased the frequency of total CD8<sup>+</sup> T cells in the nRCT-treated cohort (429.9 ± 284.2 CD8<sup>+</sup> T cells/mm²) compared to the untreated cohort (286.8 ± 162.6 CD8<sup>+</sup> T cells/mm²) as depicted in **Figure 4B**. The number of the CD8<sup>+</sup> T cells was increased in 14 out of 18 post-nRCT tumor samples (data not shown). In 12 out of these 14 tissues, a simultaneous accumulation of CD8<sup>+</sup> T cells and pDCs was detected. Within the total rectal cancer-infiltrating CD8<sup>+</sup> T cell compartment, CD8<sup>+</sup> T lymphocytes expressing the cytotoxic

examples, the presence of infiltrating slanMo in an (A,B) untreated and (C,D) treated rectal cancer tissue is shown. Scale bars are (A,C) 100µm or (B,D) 50µm, respectively. (E) The number of slanMo in non-matched untreated (n = 28) compared to nRCT-treated (n = 40) rectal cancer tissues is depicted. The results are presented as mean value ± SD of rectal cancer-infiltrating slanMo. (F) The frequency of slanMo in 20 matched pre-nRCT or post-nRCT tumor specimens is demonstrated. The results are presented as mean value ± SD of rectal cancer-infiltrating slanMo.

effector molecule GrzB were also detectable (**Figures 4C–F**). Notably, the percentage of GrzB<sup>+</sup> CD8<sup>+</sup> T cells was significantly higher in the nRCT-treated cohort (57.5 ± 14.4% GrzB<sup>+</sup> CD8<sup>+</sup> T cells) in comparison to the untreated cohort (44.2 ± 9.4% GrzB<sup>+</sup> CD8<sup>+</sup> T cells) as shown in **Figure 4G**. Following these findings, we explored whether CD8<sup>+</sup> T cells co-localize with pDCs in four nRCT-treated rectal cancer tissues. As demonstrated in **Figure 5**, regions containing a high density of pDC and CD8<sup>+</sup> T cells were detectable in all analyzed rectal cancer tissues.

# nRCT Increases the Proportion of iNOS- or TNF-α-Expressing slanMo in Rectal Cancer

iNOS and TNF-α were shown to mediate either tumorpromoting or antitumor effects and may therefore influence the efficacy of nRCT in cancer patients (44, 45). Following our previous findings that iNOS- and/or TNF-α-expressing slanMo are detectable in tissue specimens of patients with various inflammatory disorders (35, 36, 46), we investigated whether nRCT modulates the percentage of iNOS- or TNF-α-producing

explore the frequency of rectal cancer-infiltrating CD3<sup>+</sup> T lymphocytes, total CD8<sup>+</sup> T lymphocytes, and GrzB<sup>+</sup> CD8<sup>+</sup> T cells in matched pre-nRCT and post-nRCT tumor samples. The frequency of (A) CD3<sup>+</sup> cells and (B) CD8<sup>+</sup> T cells in 18 matched pre-nRCT or post-nRCT tumor specimens is presented as mean value ± SD. Asterisks indicate a statistically significant difference (\*p < 0.05). (C–F) As representative examples, the presence of infiltrating GrzB<sup>+</sup> CD8<sup>+</sup> T cells in an (C,D) untreated and (E,F) nRCT-treated rectal cancer tissue is demonstrated. Scale bars are (C,E) 100µm or (D,F) 50µm, respectively. (G) The percentage of GrzB-expressing CD8<sup>+</sup> T cells in 18 matched pre-nRCT or post-nRCT tumor specimens is presented as mean value ± SD. Asterisks indicate a statistically significant difference (\*\*p < 0.01).

slanMo in matched pre-nRCT and post-nRCT tumor specimens of 10 patients. While iNOS<sup>+</sup> slanMo were present in 9 out of 10 post-nRCT rectal cancer tissues, TNF-α <sup>+</sup> slanMo were detectable in 7 out of 10 tumor tissues obtained after treatment at varying percentages (**Figures 6A–D**). In contrast, iNOS<sup>+</sup> slanMo were only found in 2 out of 10 pre-nRCT rectal cancer tissues, while

FIGURE 5 | Co-localization of pDC and CD8<sup>+</sup> T cells in nRCT-treated rectal cancer tissues. Immunofluorescence multiplex staining was performed to detect a co-localization of pDC and CD8<sup>+</sup> T cells in four post-nRCT tumor samples. As a representative example, an image of a tissue region containing high numbers of rectal cancer-infiltrating pDC and CD8<sup>+</sup> T cells is shown. Scale bar is 100µm.

TNF-α <sup>+</sup> slanMo were absent in these tissues (**Figures 6C,D**). As depicted in **Figure 6E**, the proportion of iNOS<sup>+</sup> slanMo was significantly higher in post-nRCT tumor tissues (26 ± 24% iNOS<sup>+</sup> slanMo) compared to pre-nRCT tumor specimen (1 ± 2% iNOS<sup>+</sup> slanMo). In addition, nRCT significantly increased the percentage of slanMo locally expressing TNF-α in rectal cancer tissues (**Figure 6F**).

### nRCT Significantly Increases the Proportion of CD83<sup>+</sup> pDCs in Rectal Cancer

To investigate whether nRCT influences the maturation status of rectal cancer-infiltrating pDCs, we analyzed the proportion of pDCs expressing the maturation marker CD83 in matched pre-nRCT and post-nRCT tumor samples of 18 patients. CD83<sup>+</sup> pDCs were detectable in all post-nRCT rectal cancer tissues, but only in 11 out of 18 tumor tissues before nRCT at varying percentages (**Figures 7A,B**). In 13 post-nRCT rectal cancer tissues, ≥30% of pDCs expressing CD83 were present, providing evidence that these tissues can contain a marked proportion of mature pDCs (**Figure 7B**). As shown in **Figure 7C**, the percentage of CD83<sup>+</sup> pDCs was significantly higher in postnRCT tumor tissues (36.7 ± 18.9% CD83<sup>+</sup> pDCs) in comparison to pre-nRCT tumor specimens (3.2 ± 4.2% CD83<sup>+</sup> pDCs), indicating that nRCT can profoundly enhance the proportion of mature pDCs in rectal cancer tissues.

# nRCT Significantly Enhances the Percentage of IFN-α-Expressing pDCs in Rectal Cancer

Activated pDCs are major producers of IFN-α, which may essentially contribute to the antitumor effects mediated by radio- and chemotherapy (18, 19). Following these findings, we determined the impact of nRCT on the proportion of IFN-αexpressing pDCs in matched pre-nRCT and post-nRCT tumor samples of 18 patients. Whereas, IFN-α <sup>+</sup> pDCs were only present in 11 out of 18 pre-nRCT rectal cancer tissues, infiltrating pDCs locally expressing IFN-α were found in all post-nRCT tumor samples at varying percentages (**Figures 8A,B**). In 16 post-nRCT rectal cancer tissues, ≥30% of IFN-α <sup>+</sup> pDCs were detectable (**Figure 8B**). The percentage of IFN-α <sup>+</sup> pDCs was significantly higher in post-nRCT tumor tissues (52.0 ± 20.5% IFN-α <sup>+</sup> pDCs) compared to pre-nRCT tumor specimen (5.5 ± 8.1% IFN-α + pDCs) as depicted in **Figure 8C**. These results provide evidence that nRCT can profoundly increase the proportion of pDCs locally expressing IFN-α in rectal cancer tissues.

# Rectal Cancer-Infiltrating pDCs Express the Chemokines CXCL10 and CCL4

Recently, it has been shown that pDCs are able to produce various chemokines such as CCL4, CCL5, and CXCL10 (19, 47) that can promote the migration of T cells to tumor sites. Following these observations, we investigated the presence of pDCs expressing these chemokines in 9 pre-nRCT and 18 post-nRCT tumor specimens of the matched cohort. Whereas, CXCL10- or CCL4 expressing pDCs were not found in pre-nRCT tumor samples, CXCL10<sup>+</sup> pDCs were detectable in 14 out of 18 and CCL4<sup>+</sup> pDCs in 12 out of 18 post-nRCT tumor samples (**Figures 9A,B**). CCL5-expressing pDCs were not present in these tumor tissues (data not shown). These results imply that pDCs can contribute to the secretion of important chemokines for T-cell migration to rectal cancer tissues after nRCT.

# DISCUSSION

Recent studies have revealed that DCs, as key regulators of innate and adaptive immunity, are a component of the immune architecture in colorectal cancer and may influence the clinical outcome of patients. When exploring the presence of human DCs in colorectal cancer, it has been reported that S-100<sup>+</sup> DCs are detectable in almost all colorectal specimens and are mainly located in the tumor stroma (48–50). Increasing numbers of S-100<sup>+</sup> DCs infiltrating tumor epithelium correlated with higher numbers of intraepithelial CD4<sup>+</sup> and CD8<sup>+</sup> T cells (49). Whereas, several studies provided evidence that a higher density of S-100<sup>+</sup> DCs was associated with improved survival of patients (48, 51–53), other studies did not find a statistically significant correlation between S-100<sup>+</sup> DC infiltration and a favorable clinical outcome (49, 50). In addition, it has been shown that DCs expressing the maturation marker CD83 are present in

cancer patients. (A,B) As representative examples, images of (A) single iNOS or slan and (B) single TNF-α or slan stainings as well as merged images are depicted. Original magnification was x400. (C,D) Percentage of (C) iNOS- and (D) TNF-α-expressing slanMo in matched pre-nRCT or post-nRCT tumor specimens is shown for each patient. The results are presented as mean value ± SD of the proportion of (E) iNOS- or (F) TNF-α-expressing slanMo in matched pre-nRCT or post-nRCT tumor specimens. Asterisks indicate a statistically significant difference (\*p < 0.05, \*\*p < 0.01).

colorectal cancer (53–55). These DCs were found predominantly in the invasive margin, often in clusters with lymphocytes (55). Gulubova et al. demonstrated that patients with locally advanced tumors had significantly lower infiltration of CD83<sup>+</sup> DCs in tumor stroma and in the invasive margin (53). Whereas all these findings are based on the detection of general marker molecules for DCs, studies investigating the presence of distinct human DC subsets in colon cancer tissues are rather limited. When exploring the density and distribution of pDCs in rectal cancer, we found that pDCs are present in all non-treated rectal cancer specimens at varying frequencies and are preferentially located in the tumor stroma. These results further substantiate recent studies, indicating that pDCs are detectable in colorectal cancer (51, 56). Previously, we and others have determined the presence of slanMo in primary tumor samples and derived metastases. Vermi et al. have found an accumulation of slanMo in metastatic lymph

nodes from cancer patients (42). In addition, we have detected slanMo in the majority of primary tumor specimens, metastatic lymph nodes, and distant metastases from ccRCC patients (43). Further findings have revealed that ccRCC-infiltrating slanDCs display a tolerogenic phenotype and that higher slanDC numbers are associated with a reduced survival of ccRCC patients (43). More recently, it has been shown that slanMo are also present in bone marrow specimens of multiple myeloma patients and in various types of non-Hodgkin lymphomas (57, 58). In the present study, we observed that slanMo are detectable in almost all non-treated rectal cancer specimens at varying frequencies and are preferentially located in the tumor stroma. The slanMo frequency is markedly lower in comparison to rectal cancerinfiltrating pDCs, but higher compared to RCC-infiltrating slanMo (43). So far, only little is known about the impact of nRCT on the frequency of tumor-infiltrating DCs. When exploring

the influence of nRCT on the number of infiltrating pDCs and slanMo in rectal cancer, we found a significantly higher number of pDCs in nRCT-treated tissue specimens, whereas the frequency of slanMo remained stable after nRCT.

CD4<sup>+</sup> and CD8<sup>+</sup> T lymphocytes are important effector cells of adaptive antitumor immunity. CD8<sup>+</sup> effector T cells efficiently recognize and destroy tumor cells. CD4<sup>+</sup> effector T cells enhance the ability of DCs to induce CD8<sup>+</sup> T cell responses. They also provide help for the maintenance and expansion of CD8<sup>+</sup> T cells and can eliminate tumor cells directly. In addition, CD4<sup>+</sup> T cells are able to promote the differentiation of B cells into antibodyproducing plasma cells. Based on these properties, tumorinfiltrating effector T cells efficiently influence tumor growth and play an essential role for the clinical outcome of colorectal cancer patients (2–5). Here, we explored the impact of nRCT on the frequency of rectal cancer-infiltrating T lymphocytes in matched pre-nRCT and post-nRCT tumor samples. We found that nRCT significantly increases the number of total CD8<sup>+</sup> T

proportion of IFN-α

cells as well as the percentage of CD8<sup>+</sup> T cells expressing GrzB in the nRCT-treated cohort. These results are in agreement with previous studies, indicating that nRCT can significantly enhance the density of infiltrating CD8<sup>+</sup> T cells in rectal cancer. Thus, it has been reported that the frequency of rectal cancer-infiltrating CD8<sup>+</sup> T cells and the proportion of GrzB<sup>+</sup> CD8<sup>+</sup> T cells is markedly increased by nRCT (59–62). In addition, a high density of CD8<sup>+</sup> T cells prior to treatment was associated with a good response to nRCT and was linked to a better clinical outcome (59–61). However, McCoy et al. did not observe an association between the frequency of rectal-cancer infiltrating CD8<sup>+</sup> T cells prior to treatment and response to nRCT (63).

iNOS and TNF-α are important molecules that have an impact on carcinogenesis and cancer progression and may influence the clinical response of patients to various treatment modalities. Both molecules play a dual role in cancer by mediating tumorpromoting or antitumor effects (44, 45). Thus, iNOS and TNFα were shown to promote tumor proliferation, angiogenesis, invasiveness, and metastasis. Further studies have revealed that these molecules can also efficiently impair tumor growth by various mechanisms such as the inhibition of proliferation, the induction of apoptosis, and the recruitment of tumor-reactive T cells (44, 45). Previously, we have identified slanMo as inflammatory, iNOS- and/or TNF-α-expressing cells in tissues specimens of patients with psoriasis, lupus erythematosus, or graft-vs.-host disease (35, 36, 46). Here, we determined the impact of nRCT on the percentage of iNOS- or TNF-α-expressing slanMo in matched pre-nRCT and post-nRCT rectal cancer specimens. iNOS<sup>+</sup> or TNF-α <sup>+</sup> slanMo were rare or absent in pre-nRCT tissues. However, nRCT significantly augmented the proportion of infiltrating slanMo locally expressing iNOS- or TNF-α in rectal cancer. The nRCT-mediated increase of iNOSproducing slanMo is in line with a recent study, demonstrating that low dose irradiation induces iNOS expression in melanomainfiltrating mouse macrophages, resulting in an enhanced recruitment of T cells (64). In addition, Klug et al. observed that low dose irradiation leads to an accumulation of iNOS<sup>+</sup> macrophages and intraepithelial T cells in tissue specimens of pancreatic cancer patients (64).

Type I IFN play a key role in antitumor immunity (65). They promote the maturation and antigen-presenting capacity of DCs as well as their migration to lymph nodes. Furthermore, type I IFN stimulate the release of proinflammatory cytokines by macrophages and improve the tumor-directed cytotoxic activity mediated by CD8<sup>+</sup> T cells and NK cells. Accumulating evidence indicates that type I IFN can essentially contribute to the beneficial effects mediated by chemotherapy and radiotherapy (9, 66). Thus, it has been demonstrated that the efficacy of anthracycline-based chemotherapy against established tumor in mice is critically dependent on type I IFN (66). Furthermore, it has been reported that radiotherapy increases the intratumoral type I IFN expression in mice (67). Type I IFN were shown to enhance the cross-priming and T-cell stimulatory capacity of tumor-infiltrating DCs leading to tumor regression (67). pDCs are major producers of type I IFN following stimulation with toll-like receptor 7 and 9 agonists (18, 19). However, recent studies have revealed that tumor-infiltrating pDCs are defective at producing type I IFN in response to toll-like receptor agonists (26, 28, 29). When analyzing matched pre-nRCT and post-nRCT rectal cancer specimens, we found that only a small proportion of IFN-α-expressing pDCs is detectable prior nRCT. Notably, nRCT profoundly increased the proportion of pDCs locally expressing IFN-α. Together with our finding that nRCT also enhances the percentage of CD83<sup>+</sup> pDCs, these results reveal that nRCT promotes the maturation and IFN-α production of rectal cancer-infiltrating pDCs.

pDCs are capable of producing various chemokines such as CCL4, CCL5, and CXCL10, which attract T cells to sites of infection and inflammation (19, 47). In addition, it has been reported that these chemokines play an essential role for the trafficking of T cells to tumor tissues. Thus, it has been shown that the expression of CCL4, CCL5, and CXCL10 in melanoma metastases is associated with the recruitment of CD8<sup>+</sup> T cells (68). Moreover, a recent study has revealed that CCL5 and CXCL10 expressed by colorectal cancer tissues promote the migration of GrzB<sup>+</sup> CD8<sup>+</sup> T cells (69). When investigating the expression of these chemokines by rectal cancer-infiltrating pDCs, we observed that CCL4- or CXCL10-expressing pDCs are present in the majority of post-nRCT tumor specimens, whereas they are absent in pre-nRCT tissue samples. These findings indicate that nRCT can induce CXCL10 and CCL4 expression in rectal cancer-infiltrating pDCs. In addition, they imply that pDCs can contribute to the production of important chemokines for T-cell migration to rectal cancer tissues after nRCT.

Taken together, we found that nRCT significantly increases the percentage of rectal cancer-infiltrating slanMo locally expressing iNOS and TNF-α, which can either mediate tumor-promoting or antitumor effects and may affect the efficacy of nRCT in rectal cancer patients. Moreover, we demonstrated for the first time that nRCT results in an accumulation of pDCs as well as an increased proportion of CD83- and IFN-α-expressing pDCs in rectal cancer. In addition, the density of infiltrating GrzB<sup>+</sup> CD8<sup>+</sup> T cells was significantly enhanced by nRCT. These findings indicate that nRCT can harness antitumor responses by converting immature, non-activated pDCs into mature, inflammatory cells and by increasing the frequency of CD8<sup>+</sup> T cells expressing the

#### REFERENCES


cytotoxic effector molecule GrzB. Activated pDCs and GrzB<sup>+</sup> CD8<sup>+</sup> T cells may contribute to the beneficial effect of nRCT in rectal cancer patients.

#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the ethical committee at the Faculty of Medicine Carl Gustav Carus of the Technische Universität Dresden, Germany. All subjects gave written informed consent in accordance with the Declaration of Helsinki. The protocol was approved by the ethical committee at the Faculty of Medicine Carl Gustav Carus of the Technische Universität Dresden, Germany.

#### AUTHOR CONTRIBUTIONS

FeW, UH, FrW, and IP performed experiments, analyzed data, and wrote the manuscript. RW, MaK, KF, and MA performed experiments and analyzed data. US, AJ, and AT analyzed data and wrote the manuscript. AB, CR, JW, KS, ET, MeK, GF, MB, MPB, DA, and GB analyzed data and revised the article. MS designed research, interpreted the data, and wrote the manuscript.

#### FUNDING

This study was supported in part by grants from the Faculty of Medicine Carl Gustav Carus of the Technische Universität Dresden, Germany to RW and from the German Research Foundation (DFG, SCHA 1693/1-1, and SFB TR 156) to KS.

#### ACKNOWLEDGMENTS

We thank Bärbel Löbel for excellent technical assistance.


clearance of human papillomavirus-positive cancer. Int J Cancer. (2013) 133:120–9. doi: 10.1002/ijc.28015


between human slan<sup>+</sup> non-classical monocytes and myeloid dendritic cells. J Leukoc Biol. (2017) 102:1055–68. doi: 10.1189/jlb.3MA0117-037R


against tumor-associated antigens and survival. J Transl Med. (2007) 5:62. doi: 10.1186/1479-5876-5-62


B+ (GrzB) cytotoxic T-cells in rectal cancer. Oncoimmunology. (2017) 7:e1393133. doi: 10.1080/2162402X.2017.1393133


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Wagner, Hölig, Wilczkowski, Plesca, Sommer, Wehner, Kießler, Jarosch, Flecke, Arsova, Tunger, Bogner, Reißfelder, Weitz, Schäkel, Troost, Krause, Folprecht, Bornhäuser, Bachmann, Aust, Baretton and Schmitz. This is an openaccess article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Harnessing Dendritic Cells for Poly (D,L-lactide-co-glycolide) Microspheres (PLGA MS)—Mediated Anti-tumor Therapy

#### Julia Koerner <sup>1</sup> , Dennis Horvath<sup>1</sup> and Marcus Groettrup1,2 \*

<sup>1</sup> Division of Immunology, Department of Biology, University of Konstanz, Konstanz, Germany, <sup>2</sup> Biotechnology Institute Thurgau at the University of Konstanz, Kreuzlingen, Switzerland

With emerging success in fighting off cancer, chronic infections, and autoimmune diseases, immunotherapy has become a promising therapeutic approach compared to conventional therapies such as surgery, chemotherapy, radiation therapy, or immunosuppressive medication. Despite the advancement of monoclonal antibody therapy against immune checkpoints, the development of safe and efficient cancer vaccine formulations still remains a pressing medical need. Anti-tumor immunotherapy requires the induction of antigen-specific CD8+ cytotoxic T lymphocyte (CTL) responses which recognize and specifically destroy tumor cells. Due to the crucial role of dendritic cells (DCs) in initiating anti-tumor immunity, targeting tumor antigens to DCs has become auspicious in modern vaccine research. Over the last two decades, micron- or nanometer-sized particulate delivery systems encapsulating tumor antigens and immunostimulatory molecules into biodegradable polymers have shown great promise for the induction of potent, specific and long-lasting anti-tumor responses in vivo. Enhanced vaccine efficiency of the polymeric micro/nanoparticles has been attributed to controlled and continuous release of encapsulated antigens, efficient targeting of antigen presenting cells (APCs) such as DCs and subsequent induction of CTL immunity. Poly (D, L-lactide-co-glycolide) (PLGA), as one of these polymers, has been extensively studied for the design and development of particulate antigen delivery systems in cancer therapy. This review provides an overview of the current state of research on the application of PLGA microspheres (PLGA MS) as anti-tumor cancer vaccines in activating and potentiating immune responses attempting to highlight their potential in the development of cancer therapeutics.

Keywords: PLGA, microspheres, cancer vaccine, dendritic cell, anti-tumor response, spray drying, immunotherapy, CTL

# INTRODUCTION

With an annual incidence of several million new cases worldwide, cancer represents one of the most prevalent malignancies and leading causes of pain and mortality. Conventional treatment options usually include a combination of primary resection, radiotherapy and/or chemotherapy. However, cancer patients suffer from devastating adverse side-effects and poor quality of life after chemo- or

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Elizabeth Ann Repasky, University at Buffalo, United States Yvette Van Kooyk, VU University Medical Center, Netherlands

\*Correspondence: Marcus Groettrup marcus.groettrup@uni-konstanz.de

#### Specialty section:

This article was submitted to Vaccines and Molecular Therapeutics, a section of the journal Frontiers in Immunology

> Received: 06 September 2018 Accepted: 14 March 2019 Published: 05 April 2019

#### Citation:

Koerner J, Horvath D and Groettrup M (2019) Harnessing Dendritic Cells for Poly (D,L-lactide-co-glycolide) Microspheres (PLGA MS)—Mediated Anti-tumor Therapy. Front. Immunol. 10:707. doi: 10.3389/fimmu.2019.00707

**390**

radiation therapy. Moreover, therapeutic failure of standard therapeutics results in increased risk of tumor relapse and metastasis formation (1). Hence, there is an urgent need for safe and effective vaccine development against this life-threating group of diseases. With the identification of multiple unique cancer antigens (tumor-associated antigens, TAA) and the investigation of manifold immune evasion pathways of tumors, immunotherapy has become a growing focus in clinical research.

Cancer immunotherapy encompasses therapeutic modulation of the host's immune system to defend against foreign or selfantigens that have gone awry in tumor development. Cancer vaccines aim at triggering immune activation to specifically target and eliminate tumor cells. Ideally, a memory response is generated to impede metastasis formation and further spread of the disease. In contrast to passive immunotherapy which aims at delivering neutralizing antibodies, active forms of immunotherapy are supposed to induce multi-faceted cell mediated immunity by simultaneous activation of APCs, CD4+ and CD8+ T cells, as well as B cells and innate immune cells, as for instance NK cells, granulocytes and macrophages. Compared to standard tumor therapies, immunotherapeutic anti-tumor vaccines offer distinct advantages, namely: increased specificity and reduced toxicity by activation of antigen-specific CTL responses. Effector CTLs are able to decrease large tumor masses and induce long-term protection against tumor recurrence through induction of immunological memory (2). Recent advances in cancer immunotherapy have paved the way for the discovery of versatile methods for prevention or treatment of various types of cancer. As a result, several cancer vaccines are currently investigated in clinical trials. However, most of them have not progressed beyond phase III studies. Although antigenspecific responses were generated and increases in overall survival rates were obtained, there is no consistency in clinical benefit. Most of the approaches were presented with major drawbacks in vaccine delivery and efficacy. Administration of soluble antigenic formulations, e.g. synthetic peptides or purified tumor-associated antigens was not promising due to poor immunogenicity, limited bioavailability, short half-life and rapid degradation or elimination of the antigensin vivo, demanding the need for repeated injections (3).

Due to the unique ability of DCs to prime and activate naïve T cells (4, 5), DC-based vaccination strategies have shown to be a promising approach in the development of polyvalent cancer vaccines. The first promising results have been achieved using ex vivo derived autologous tumor cells or DCs that have been pulsed with various tumor-associated proteins or peptides (6). However, major drawbacks were seen in suboptimal antigen presenting capacity of isolated DCs or simple lack of autologous tumor samples (7, 8). Several promising immunotherapeutic advances came across with the use of allogeneic tumor-lysate pulsed DCs, loading of DCs with MHC class I restricted tumor antigens (9–11), or via transfection of cDNA encoding TAAs (8, 12). Whole tumor lysate contains a large repertoire of tumor antigens capable of inducing immune responses against a broad spectrum of multiple epitopes including those that are unique to the patient's tumor. The development of DC-based vaccination has led to the first therapeutic cancer vaccine. In April 2010, Provenge <sup>R</sup> (Sipuleucel-T) was approved by the FDA for treatment of castration-resistant, metastatic prostate cancer (13). This immunotherapy involves ex vivo stimulation of autologous, blood-derived antigen presenting cells from prostate cancer patients that are pulsed with a prostate cancer-associated antigen [PAP (prostate acid phosphatase)–GM-CSF fusion protein]. DCs were subsequently re-introduced into patients to stimulate an immune response against PAP-expressing prostate cancer cells. These well-tolerated approaches using ex vivo loaded DCs were tested in a variety of experimental models and clinical trials [reviewed in Tacken et al. (14)], and seemed to be encouraging due to good safety records, the generation of enhanced T cell responses and partial reduction of tumor load. However, clinical application is still limited as these ex vivo procedures are laborious and time-consuming, extremely expensive and lack universal applicability (15). More importantly, the overall clinical response rates in cancer patients were only 7% (16).

To circumvent the limitations associated with in vitro manipulation of cells, direct in vivo targeting of DCs along with appropriate adjuvants for simultaneous activation of dendritic cells has gained major focus. Particulate delivery systems have shown to overcome the main obstacles related to traditional cancer therapeutics. Instead of causing the risk to induce systemic, adverse immunity, vaccine antigens are delivered to DCs in a targeted manner. We and others have established the use of PLGA MS as an efficient vaccine delivery system for dendritic cell targeting. Subsequent induction of potent immune responses has led to remarkable protective and therapeutic anti-tumor activity in vivo. In this article we review how DCs can be antigen charged and matured with PLGA MS in vitro and in vivo and how microspheres can be produced and formulated to optimally be taken up by DCs. Moreover, we discuss the parameters how antigen presentation and T cell stimulation by PLGA MS-loaded DCs can be improved to elicit a vigorous and effective anti-tumor immune response.

#### COMPARISON OF PARTICULATE ANTIGEN DELIVERY SYSTEMS

At present, several particulate drug delivery systems for cancer immunotherapy–other than PLGA based particles–have passed pre-clinical investigations and are currently tested for human application, such as liposomes, virosomes, immune-stimulatory complexes (ISCOMs) or gold particles. These systems are reviewed elsewhere (17) and are beyond the scope of this article. Furthermore, detailed analysis of nano-sized particulate vaccine delivery systems has been already extensively reviewed (18–20) and is only of specialized focus in this review.

Multiple different natural or synthetic polyesters have been reported for the development of (sub)micron sized colloidal drug delivery systems, such as poly(lactic acid) (PLA), poly(glycolic acid) (PGA), poly(ε-caprolactone) (PCL), poly(methyl methacrylate) (PMMA), poly(β-amino esters) as well as other ester derivates [poly(anhydrides), poly(orthoesters), poly(phosphoesters), poly(phosphazenes) or poly(cyanoacrylate)]. Due to their excellent bioavailability,

biodegradable and biocompatible properties, controlled release and low toxicity, these polymers have been extensively studied as delivery systems of various therapeutic vaccines as well as for cancer immunotherapy in preclinical settings (21–23). Based on the method of preparation, different types of polymeric particles can be designed: spheres, capsules, cubes and other shapes. While the active compound of micro/nanocapsules is contained inside a cavity underneath the polymeric layer, micro/nanospheres homogenously entrap the encapsulated materials into the inner polymer matrix core (24).

The aliphatic co-polymer PLGA is one of the most frequently used and explored polymers for controlled delivery of bioactive molecules in microspheres and nanoparticles (NP) (25). The amorphous PLGA is composed of varying proportions of lactic and glycolic acids (**Figure 1**). Due to its ideal in vivo properties of biodegradability, biocompatibility and its clear safety records, PLGA has been licensed by the U.S. Food and Drug Administration (FDA) and European Medicines Agency (EMA) for use in pharmaceutical application via parenteral (subcutaneous, intradermal, intramuscular) and mucosal routes as well as for suspension formulations of biomedical devices including surgical sutures and bone implants (26). At present, there are 12 PLGA-based microparticle cancer vaccine formulations approved by the FDA for clinical use. Most of these PLGA MS systems are targeting prostate cancer, for example Pamorelin LA <sup>R</sup> which encapsulates the gonadotropin releasing hormone (GnRH) agonist triptorelin pamoate for palliative treatment of advanced prostate cancer (27). Of note, not a single nanoparticulate vehicle has reached clinical approval due to associated toxicity issues (28) as discussed later in this review.

#### PROPERTIES OF PLGA PARTICLES

A wide range of biologically active compounds including hormones, antibiotics, and drugs can be encapsulated into PLGA particles (29, 30). Thus, PLGA micro- and nanoparticles have been well-established as delivery systems of innumerable antigens such as proteins, peptides, lipopeptides, viral or bacterial DNA as well as immunomodulatory molecules (31–37). PLGA particles exhibit a vast array of advantages over soluble vaccine formulations. At first, GMP (good manufacturing practice) grade polymer is commercially available (for example, PLGA Resomer <sup>R</sup> from Evonik Industries) meeting the GMP requirements of regulatory authorities. Encapsulation within PLGA particles protects the encapsulated bioactive molecules from premature degradation by proteolytic enzymes or metabolic turnover and minimizes loss of therapeutic activity prior to delivery. The enhanced bioavailability is due to sustained and controlled release of encapsulated substances over extended time periods of several weeks to months thus creating a depot effect at the site of injection. Prolonged antigen presentation and continuous T cell stimulation would circumvent the need for conventional multiple dose immunization schedules, e.g. primeboost vaccination (38, 39). Hence, PLGA MS would provide a valuable approach for single administration vaccine design so that clinical intervention is only limited to one therapeutic injection. Encapsulation of peptides into PLGA MS was shown to enhance and extend antigen presentation on MHC class I and II by DCs and macrophages (29, 38) which is possibly due to higher total load of antigens and prolonged degradation time of larger microparticles compared to nanoparticles (40). Furthermore, entrapment of proteins or peptides into biodegradable PLGA microspheres increases the immunogenicity of poorly immunogenic antigens, e.g. weak self-antigens in tumor tissue. While soluble peptide immunizations elicited only very poor CD8+ T cell responses, microencapsulation of the HLA-A ∗ 0201 restricted immunodominant epitope STEAP 1 (six transmembrane epithelial antigen of the prostate) was shown to induce potent prostate cancer peptide-specific CTL activation and cytotoxic effector function (36, 41).

#### Release and Encapsulation Qualities of PLGA Microspheres

Upon encountering aqueous medium, PLGA is slowly hydrolyzed into its original monomeric components. The resulting products lactic and glycolic acid are physiological metabolites of the citric acid cycle and thus completely eliminated from the human body as carbon dioxide and water (42). The degradation rate and subsequent drug release is primarily dependent on the PLGA polymer composition and the molecular weight of the polymer. These two factors also impact hydrophilicity, the hydration rate as well as the glass transition temperature (Tg) of the respective polymer type, which in turn also affect the release profile (43). A high content of glycolic acid in the co-polymer leads to higher hydrolysis rates and a more rapid release, as glycolic acid is slightly more hydrophilic than the crystalline lactic acid, which fosters water permeability into the polymer matrix. Several other factors contribute to release rates of PLGA MS including concentration of the polymer in the organic solvent during PLGA MS fabrication, PLGA particle size and morphology, as well as storage conditions such as temperature and humidity and of course, the encapsulated material itself. The PLGA 50:50 polymer is preferred over other PLGA polymers with different lactic:glycolic ratios (65:35; 75:25; 80:20) in controlled release vaccine formulation since encapsulated molecules are homogenously dispersed inside the polymer matrix. Additionally, it is slightly more hydrophilic and thus possesses the fastest degradation rate resulting in complete degradation within 30 to 60 days in aqueous medium (44). It also occupies the least crystallinity hence being more Koerner et al. PLGA MS for DC-Centered Cancer Therapy

prone to (enzyme-independent) hydrolysis and bulk erosion. Only when the PLGA polymer becomes porous and hydrated, encapsulated material of high molecular weight can be released. This will prevent early release of antigens or adjuvants before internalization of the particles by DCs and thus reduces systemic distribution of the encapsulated molecules. The release profile of PLGA degradation encompasses two phases with an initial burst that is followed by progressive release of the encapsulated material. The burst release is likely attributed to weakly bound or adsorbed proteins on the PLGA particle surface that are rapidly dispersed upon submersion into aqueous media (45). Noticeably, about 30% of the entrapped material can be released within a few days, though the percentage markedly depends on the physical properties of the microparticles (46).

### Physico-Chemical Characteristics of PLGA Particles

A major advantage of using PLGA polymers is attributed to its great flexibility and ease to manipulate and modify its physicochemical properties such as: molecular mass of the polymer, hydrophilicity and crystallinity (monomer ratio), endgroup chemistry, particle size and surface charge. All these factors can be modified to obtain desired and suitable degradation rates and subsequent release patterns for individual treatment regimen. Furthermore, these properties also dictate intracellular trafficking and can thus be individually adjusted to the needs of the encapsulated material (47). The main improvement of using PLGA particles as vaccine delivery system relies on the ability to simultaneously stimulate innate and adaptive immunity through directing intracellular antigen processing toward the crosspresentation pathway. Furthermore, maintenance of integrity and thus activity of the encapsulated material ensures their bioavailability and their ability to mount effective immune responses (48).

# PRODUCTION METHODS FOR PLGA MS

There are several methods employed to produce micro- and nanoparticles such as emulsification-solvent-evaporation, organic phase separation (coacervation), nano-precipitation (diafiltration), and newer strategies such as supercritical microfluidics, coaxial electrospray or the PRINT (particle replication in non-wetting templates) technology (49). However, major drawbacks of the most widely used single or double emulsification solvent evaporation techniques is poor encapsulation efficiency, which either requires increased drug loading or the use of surfactants (e.g. PVA, poly-vinyl alcohol) to stabilize the oil-in-water emulsion until particles have been formed. Moreover, high shear or cavitation forces, excessive use of energy or freezing and drying cycles cause significant risk of aggregation or degradation of encapsulated material of these particles, thereby rendering emulsion techniques difficult for mass production (50). Furthermore, the initial burst is very high due to poor drug loading into the particles while adsorption onto the particle surface is very common (22). Nevertheless,

we have used and optimized the spray-drying technology in our laboratory.

#### Microencapsulation by Spray Drying

Spray-drying is a very suitable and rapid one-step process for encapsulation of both hydrophobic as well as hydrophilic proteins and peptides into PLGA particles. The principle is based on nebulization of a solid-in-oil dispersion or water-inoil-emulsion composed of antigen and adjuvants in an aqueous phase that is mixed with the volatile, water-immiscible organic solvent [e.g. dichloromethane (DCM)] used to dissolve the PLGA polymer. The fluid is spray-atomized into a gas stream of compressed air or compressed nitrogen into a desiccating chamber, where liquid droplets pass a current of warm airstream subsequently creating microparticles at the spray nozzle by evaporation of the organic solvent (51, 52). Evaporation keeps the product temperature at low levels, thus only little temperature deterioration occurs (53). As the fluid is converted into a dry powder in the drying chamber, the particle-loaded air stream enters tangentially into the cyclone, which results in a centrifugal force that creates a downward spiral movement in the cyclone causing particle deposition at the bottom of the cyclone separator and the collecting vessel (54) (see **Figure 2**).

This microencapsulation process warrants stability and the biological activity of the encapsulated epitopes and guarantees high yield and encapsulation efficiencies of more than 85% (55). The low preparation temperature of the spray drying method avoids thermal denaturation of encapsulated compounds. The produced microspheres do not exhibit aggregation and show good suspensibility in injection solution. Spray drying usually produces particles with a particle size distribution of about 500 nm−5µm. Besides process parameters such as the liquid feed rate, the drying air flow rate or the inlet air temperature, molecular weight and concentration of the polymer in the organic solvent critically determines particle size and affects microsphere morphology and subsequently degradation and drug release of PLGA MS (56). The spray-drying method has been successfully used with several biodegradable polymers such as PLA, PCL, gelatin, and polysaccharides or related biopolymers (57–59). It has several advantages over other particle production methods such as lower residual organic solvent, higher yield, and higher encapsulation efficiency or prolonged sustained release. Mentionable, particle sizes can now be easily controlled by using the nano-spray drying method based on the vibrating mesh technology (60, 61). Sticky adhesion of PLGA microparticles to the interior surface of the spray-drier's glass ware, as always referred to a salient drawback of spray drying (62), has been overcome by the use of the non-ionic surfactant Poloxamer <sup>R</sup> 188 to wash out spray-dried particles.

The optimized drying procedure after spray drying by vacuum drying over several days reduces the residual amount of organic solvent to a minimum (63). The authorized pharmaceutical limit for residual organic volatile impurities of DCM by the U.S. (USP) and European Pharmacopeia (PhEur) is 0.06%. This is pertinent, as incomplete solvent removal or solvent impurities may cause chemical degradation of the encapsulated compounds within the polymer matrix. By efficient removal of the solvent, spraydried PLGA MS are highly stable as dry powder for long-term storage without degradation of the encapsulated compounds thus preserving therapeutic activity. Furthermore, spray drying can be easily scaled up to produce large batches. Polymer-drug solutions of high volumes are rapidly spray-dried within minutes, which would facilitate industrial production processes for potential clinical application.

#### In vivo Uptake of PLGA Particles by APCs

Without specific recognition, PLGA MS provide non-specific and untargeted antigen delivery toward APCs (mainly DCs, but also macrophages) because particle sizes of 0.5–5µm exhibit similar dimensions to pathogens (44, 64). Conceptionally, the particulate matter facilitates cellular uptake and internalization by APCs and allows for faster degradation and rapid intracellular release of the antigenic cargo (25). Thus, encapsulated antigens are better processed and presented by APCs compared to antigens in soluble form. Consequently, PLGA MS-mediated antigen delivery induces a more efficient recognition of presented epitopes by the immune systems (65, 66). DCs, but also macrophages, are highly phagocytic cells being equally able to internalize large, micron-sized particles and small nanoparticles. Several studies indicate that the majority of DCs are able to take up PLGA MS (as well as PLGA nanoparticles) within 24 h. Although, the ideal particle size for uptake by APCs still remains a matter of debate, the particle size critically influences cellular uptake mechanisms but also dictates fate of intracellular endocytic pathways and DC activation and thus affects the generated immune response (20).

### Particle Size Influences the Immunogenicity of PLGA Particles

It has been demonstrated that DCs preferably engulf smaller, submicron- or virus–sized particles of 20–200 nm, whereas large particulate vaccines of bacterial size (>500 nm; e.g. microspheres) are mainly taken up by macrophages (67, 68). PLGA particle uptake by human DCs in vitro was less efficient at sizes exceeding 500 nm (69). A comparative study by Joshi et al. analyzed OVA (ovalbumin) -specific CTLs in blood after in vivo administration of PLGA particles containing OVA/CpG of 300 nm, 1, 7, and 17µm size. The smallest particles induced the highest antigen specific T cell response suggesting that the smaller the particle the stronger the response (70). Noteworthy, PLGA particles were injected intraperitoneally and tetramerpositive signals were analyzed after two booster vaccinations– incomparable to our vaccination regimen and analysis of peak T cell response on day 6 after PLGA MS vaccination in vivo (71). In fact, it was reported, that immature DCs (iDCs) are also able to internalize larger particles by either phagocytosis or micropinocytosis (72, 73). As well, Gutierro et al. have demonstrated increased access of large sized PLGA particles (1µm) to APCs which in turn elicited a higher total IgG response and increased IFN-γ production of T cells (74). We and others have demonstrated efficient uptake of PLGA microparticles by human peripheral blood monocyte-derived DCs (moDCs), murine immature bone-marrow derived DCs (BMDCs), as well as macrophages in vitro and by CD11c+ dendritic cells after subcutaneous immunization in vivo (75–77). The entrapped content in DCs is efficiently transported from peripheral tissue to the site of antigen-presentation in secondary lymphoid organs (SLOs) like spleen and liver, providing direct evidence for migration of immature, skin-resident DCs to draining lymphnodes after PLGA MS uptake (78). This was experimentally confirmed by the presence of Quantum-Dot (QD) positive PLGA microspheres in CD169+ subcapsular sinus macrophages (SSM) in draining lymph nodes (dLN) after immunization with these fluorescent microspheres (79). In contrast to subcutaneous PLGA administration into dermis or epidermis, macrophages are the predominant cell type entrapping PLGA particles after i.p. administration (80, 81). PLGA MS uptake by human moDCs in vitro does not negatively influence biological properties, such as survival, cytokine secretion, antigen presentation or subsequent T cell stimulation (75, 82). Also, uptake of PLGA nanoparticles has been investigated using in vitro generated human and mouse DC population (83–85). Human moDCs, CD34+ stem cell-derived DCs and mouse BMDCs were able to engulf PLGA NP. Uptake of PLGA MS and NPs was prevented using cytochalasin B, which points to involvement of actin-polymerization during phagocytosis of PLGA particles (86, 87). In fact, it was shown that PLGA nanoparticles are partly internalized via fluid phase pinocytosis but also through clathrin-dependent receptor mediated endocytosis, while uptake of PLGA microparticles by DCs was achieved by non-specific phagocytosis (88).

#### Present Challenges of PLGA Nanoparticle Mediated Cancer Vaccines

With respect to vaccine design, one must consider that nanoparticles with a size range of < 200 nm are able to directly enter the lymphatic vessel system from the interstitial space by diffusing through endothelial cell junctions. Additionally, NPs even can easily cross physiological barriers, such as the pulmonary tract, epithelial tight junctions or the blood-brainbarrier (BBB) without specific targeting. On the one hand, PLGA nanoparticles might facilitate stimulation of immune responses via direct delivery of antigens to lymph node (LN) resident DCs and macrophages within hours after administration (82, 89). On the other hand, it has been established that premature antigen presentation may lead to induction of antigen tolerance. Furthermore, toxicity issues of unspecific uptake by other endocytic cells or non-specific distribution is still a problem of PLGA based nanoparticle-mediated vaccine delivery (90). In contrast, PLGA microspheres remain at the subcutaneous injection site in peripheral tissues and require active uptake by immature DCs resulting in proper activation of DCs and migration to skin-draining LNs where they efficiently present the processed antigens to naïve T cells. Additional toxicity concerns of nano-polymers have emerged, namely electrostatic interaction of positively charged nanoparticles with cell membranes, the recognition of hydrophobic NPs with cells of the reticuloendothelial system (RES) or aggregation of small cationic nanoparticles with serum proteins, potentially causing severe immunotoxicity by hemolysis or platelet aggregation ("nanotoxicology") (90). To improve directed targeting and to minimize safety issues of undesired biodistribution in vivo—a problem we are not facing with the use of PLGA microspheres– nanoparticles need to be either surface-modified by hydrophilic moieties like the non-ionic polymer poly ethylene glycol (PEG) or need to be decorated with anchoring endocytosis molecules such as mannose, fucose, N-acteylglucosamine directed against DC-specific surface receptors (e.g. DC-SIGN, mannose receptor, DEC-205) or with DC-specific antibodies such as anti-CD11c (91, 92). The attachment of DC targeting moieties on PLGA NP surfaces has resulted in enhanced vaccine efficacy due to selective cellular binding, facilitated receptor-mediated endocytosis, and subsequent increased antigen cross-presentation to CD8+ T cells (93, 94). Despite enhancing homing mechanisms, preclinical and clinical data over the last decade have unveiled that targeting optimizations did not increase intratumor delivery of NP, which is below 1% of the injected nanoparticle dose (95, 96).

Though present particle-based cancer vaccine strategies have been built upon the hypothesis of preferential uptake of nanoparticles (smaller than 200 nm) and subsequent superiority at priming of cytotoxic responses over microparticles (>1µm) (97), the optimum particle size for eliciting maximum immune responses has been a challenging topic ever since. Particle size is an important but not the only factor for dictating cellular uptake and intracellular trafficking. In contrast, the induction of specific and potent immune responses depends on a vast array of parameters including physico-chemical properties of PLGA, polymer composition, molecular weight and preparation methods, as well as routes of administration and nature and content of the encapsulated material. We suggest that PLGA microspheres exhibit an ideal adjuvant particle size inducing consistent and very effective immune responses in vivo that encourages ongoing use and future optimization of PLGA microsphere-based anti-cancer vaccines (see **Figure 3**).

## DC-MEDIATED ANTIGEN PRESENTATION FROM PLGA PARTICLES

Upon endocytic uptake of PLGA microspheres by iDCs, the particles are internalized into early endosomes. A combination of homogenous bulk polymer erosion and slow hydrolysis of microspheres leads to release of the micro-encapsulated antigens and molecules over a period of about 30–60 days, which elicits a low micro-environmental pH that further enhances PLGA hydrolysis (44). Inside the acidic endosomal compartment, lysosomal proteases, and peptidases cleave released antigens into peptides of 12–25 amino acids in length which normally enter classical endocytic pathway via MHC class II presentation for interaction with CD4+ T cells (98). Furthermore, release of the antigenic cargo, including TLR ligands with receptors located at the inner endosomal membrane, leads to endosomal acidification and maturation of the phagosome associated with TLR triggering (99). Reversion of the anionic particle surface charge (from negative to positive) in the acidic lysosomal compartment enables local interaction with endo-lysosomal membranes and facilitates escape from phagosomes into the cytoplasmic compartment. In fact, PLGA micro/nanoparticles rapidly escape the endo-lysosomal compartment within minutes (65, 76). Another possibility of endosomal escape has built upon the "proton-sponge mechanism." The influx of chloride and hydronium ions during endosomal acidification causes osmotic pressure and leads to rupture of the endosomal membrane and subsequent release of its content (25, 100). Cytosolic release of the encapsulated proteins leads to antigen degradation into 8–11 aa long peptides by the proteasome before loading of these peptide fragments onto MHC class I molecules in the ER, a process known as "cross-presentation" (101). MHC class I—peptide complexes are subsequently transported to the cell surface to be presented to CD8+ T cells, thereby inducing the differentiation of CTLs. PLGA encapsulated antigens can be cross-presented by either endosomal escape (phagosome-to-cytosol pathway) or even simultaneously via the vacuolar pathway in the endocytic compartment. Compared to that, other particulate antigen formulations are exclusively relying on the TAP/proteasomedependent pathway (102). Via exploiting distinct pathways of antigen presentation, PLGA-based particles increase the peptide

FIGURE 3 | Schematic description of PLGA microsphere mediated anti-tumor response. After subcutaneous immunization, PLGA microspheres are efficiently taken up by immature, skin-resident APCs, mainly DCs. Co-delivery of antigens and TLR ligands leads to enhanced DC activation and maturation by upregulation of co-stimulatory surface maturation marker and MHC class molecules I and II during migration to lymph nodes. In the draining lymph node, encapsulated cancer antigens are processed and presented on either MHC class II to naive CD4+ T helper cells or via cross presentation to CD8+ T cells. Priming and activation of CD8+ T cells leads to differentiation and proliferation of tumor antigen-specific effector CTLs. Clonal expansion and CTL infiltration into the tumor environment results in recognition and eradication of target tumor cells mediated via IFN-γ release and enhanced Th1 polarized immune functions.

pool that is presented on MHC class I and subsequently, the magnitude of the resulting CTL response. Furthermore, downregulation or loss of TAP activity is a major mechanism of tumor immune evasion (103). Thus, TAP deficiency in tumors won't necessarily hamper PLGA MS-mediated antigen crosspresentation by usage of the vacuolar pathway. Involvement of the cross-presentation pathway in processing of encapsulated protein and peptide antigens is further underlined by blockage of their presentation using proteasome inhibitors or brefeldin A (104). Cross-presentation is highly relevant for anti-tumor vaccines that rely on proper induction of tumor killing CTLs (25, 29, 38, 65, 79). Simultaneously, release of antigens into the cytosol may protect the antigenic content from further lysosomal degradation resulting in prolonged antigen presentation. Efficient presentation of PLGA MS delivered proteins and peptides onto MHC class I and II leads to development of a full-blown immune response, since activation of CD4+ T cells, particularly T helper 1 (Th1) cells, are central for activation and stimulation of antigen-specific CTLs through secretion of IFN-γ, IL-2, and IL-12. In addition to direct tumor cytolytic functions, IFN-γ secretion further recruits crucial mediators of the innate immune response, such as NK cells and macrophages thereby potentiating tumor cell killing or apoptotic tumor body clearance (105, 106). The only limitation of PLGA microparticles for use as anti-cancer vaccine is attributed to high initial burst due to dissolution of molecules that are adsorbed at the particle surface which may cause unintentional toxic sideeffects (45). However, it has been demonstrated that the initial burst is of lower magnitude in larger (micro-)particles compared to smaller particles (46).

#### CO-ENCAPSULATION OF ANTIGEN AND IMMUNOSTIMULATORY PATTERN MOLECULES

Encapsulation of antigen together with immunomodulatory molecules overcomes obstacles associated with present adjuvant containing vaccines. For instance, an ameliorated safety profile of adjuvants is accomplished by dose reduction, thus limiting undesired toxicities due to systemic administration of the immune potentiators at non-targeted tissues. Immunogenicity of the encapsulated antigen can further be improved or increased using immunostimulatory adjuvants through providing cellular, humoral, and/or mucosal immunity. Besides ensuring efficient antigen presentation due to proper DC activation and maturation, co-delivery of antigen and adjuvants in PLGA MS/NPs may further potentiate the induced immune response through secretion of NK cell recruiting and activating cytokines by the stimulated DC. Hence, activation of both CTL and NK cell mediated anti-tumor responses are able to eliminate MHC class I positive as well as negative tumors.

The choice of the adjuvant critically determines the outcome and spectrum of the elicited immune response. Thus, addition of adjuvants improves the induction of immune responses of poorly immunogenic tumor self-antigens and potentially supports reduction of the required antigen amount.

# Currently Used Adjuvant Agents in Vaccine Formulations

Delivery of both, the antigen and an appropriate DC maturation stimulus in physiological and temporal vicinity improves migratory capacity toward LNs and efficiently stimulates proper T cell responses. Indeed, T cell activation by single encapsulated antigens in the absence of costimulatory molecules or proinflammatory cytokines may induce Th2-associated unfavorable immune responses or may even result in tolerance induction against the antigen. The most common adjuvant which has been introduced for vaccination trials over 60 years ago is the water-in-oil emulsion incomplete Freund's adjuvants IFA (107), commonly used as MontanideTM ISA-51 in clinical trials of DC-based immunotherapy (108, 109). The adjuvant effect relies on formation of a local depot providing slow release and prolonged presentation of the antigen (110). Although, IFA is primarily known to induce Th2-biased responses and stimulating humoral responses of long-term IgG production, it can also stimulate CTL or Th1 immunity directed against the antigen that is emulsified in IFA (111). Due to emerging adverse effects such as local skin reactions, abscesses, inflammation or granulomas at the injection site, IFA is not allowed for routine immunotherapy (112). Aluminum salts (alum, and its derivate MF-59) were the first adjuvants approved by the FDA and EMA for clinical use in humans (113, 114) and are currently present in the composition of the majority of vaccines (115). Although generally well-tolerated, alum adjuvants skew immune responses toward humoral mediated and Th2-polarizing conditions and only poorly stimulate CTL responses (116), additional to critical safety concerns and poor therapeutic benefit (117). Apart from alum, there are only two other adjuvants clinically approved for human use, which are AS03 [used in the H5N1 vaccine Prepandrix <sup>R</sup> (118)] and AS04 (a combination of alum and TLR 4 ligand monophosphorly lipid A (MPL <sup>R</sup> ) applied in hepatitis B virus (HBV, Fendrix <sup>R</sup> ) and human papilloma virus (HPV, Cervarix <sup>R</sup> ) vaccines) (119).

Toll-like receptor (TLR) ligands have been demonstrating a huge impact on cancer immunotherapy due to their capacities of DC activation and promotion of desired Th1 polarized immune responses. Several TLR ligands including oligonucleotides, single- or double-stranded RNA (ssRNA, dsRNA), flagellin or lipopeptides have already been investigated in clinical trials of a plethora of cancer types as reviewed in Temizoz et al. (120).

# Encapsulated TLR Ligands as DC Priming Adjuvants

TLR stimulation greatly enhances PLGA vaccine efficacy through powerful activation of DCs including the three signals required for proper T cell activation: increased expression of peptide-MHC complexes, upregulation of co-stimulatory molecules and cytokine secretion (121, 122). Furthermore, TLR triggering enhances cross-presentation to CD8+ T cells and stimulates a Th1-polarized immune response (123). Co-encapsulation of the antigen with either TLR7 or TLR9 ligands into PLGA MS stimulates DC maturation as well as cytokine secretion, and facilitates cross-presentation in vitro as shown by Heit et al. (124). Encapsulation of other so-called pathogen recognition receptor (PRR) agonists such as NOD (nucleotide-binding oligomerization domain-like receptor) ligands into either PLGA NP or MS have resulted in similar improvement of vaccine efficiency through enhanced maturation and pro-inflammatory cytokine secretion of human moDCs (125, 126). A detailed list of studies demonstrating improved cellular responses elicited by PLGA particulate systems via association of TLR ligands compared to the antigen alone or over soluble counterparts was extensively reviewed by Silva et al. (20).

We have incorporated at least two TLR ligands into our PLGA MS regimen, which were chosen due to their described Th1 inducing immunomodulation and stimulation of both humoral and cellular immunity (127, 128), namely CG-rich unmethylated Oligodeoxynucleotides (CpG ODN) and the RNA virus associated danger signal polyI:C (polyinosinic:polycitidylic acid) (29, 71, 129, 130). Their receptors, TLR9 and TLR3 respectively, are localized in the membrane of the endosomal compartments of most APCs where PLGA MS are internalized after endocytic uptake (131). Importantly, actual expression pattern of the respective TLRs has to be considered for particle vaccine design and the preferred targeting cell type. While TLR9 expression is limited to plasmacytoid DCs (pDCs), B cells and keratinocytes, TLR3 is expressed more broadly (132). Co-encapsulation of the model antigen ovalbumin together with CpG ODNs or polyI:C into PLGA microspheres efficiently elicited potent antigen-specific CTL responses and Th1 differentiation in comparison with soluble antigen after a single subcutaneous PLGA MS immunization in vivo (71, 130). Mice immunized with PLGA MS OVA/CpG generated a 2-fold increase in antigen-specific CD4+ and CD8+ T cell proliferation and IFN-γ production compared to a mixture of MS loaded separately with either the antigen or the adjuvant (PLGA MS OVA + PLGA MS CpG) (71). Pulsing of DCs with empty PLGA microspheres did not induce DC maturation in vitro (130), nor did vaccination of mice with empty PLGA MS elicit undesirable T cell responses (36, 71, 79, 133) thus confirming antigen-specificity of immune responses induced with PLGA microspheres. In stark contrast to that, proinflammatory adjuvant properties of PLGA microparticles (in comparison to PLGA nanoparticles) have been observed in macrophages (134). Several other TLR agonists have proven strong potential of enhancing the immunogenicity and efficacy of PLGA particle mediated cancer therapy in preclinical settings such as the TLR4 ligand monophospholipid A (MPLA), a chemically modified derivative of the S. minnesota derived endotoxin lipid A (135). Indeed, co-administration of TLR agonists in protein and peptide based cancer vaccines have entered clinical phase such as the TLR3 ligand poly ICLC (Hiltonol <sup>R</sup> ) demonstrating tumor regression of advanced facial rhabdomyosarcoma (136), or the TLR7 agonist Imiquimod which has been approved for treatment of basal cell carcinoma due to its ability of CTL-mediated tumor regression by DC and NK cell recruitment (137).

#### Enhancing PLGA Mediated Cancer Vaccines by Co-delivery of a Second TLR Ligand

Improvement of the PLGA MS system by adding a second TLR ligand, separately encapsulated has been shown to positively influence Th cell polarization to Th1—mediated immune response by targeted DCs (71, 129), suggesting that optimal DC activation depends on synergistic triggering of several TLR signaling pathways (138). Immunization of mice with PLGA MS OVA/CpG together with PLGA MS polyI:C resulted in greater number of KLRG1+ effector T cells (139) and increased cytotoxic effector functions of OVAspecific CD8+ CTLs, as demonstrated by IFN-γ production, oncolytic granzyme B and perforin secretion and increased CD107α expression (71). Several other groups have similarly demonstrated that concomitant delivery of antigen and adjuvant in the same endo-lysosomal compartment is required for proper activation of DCs and superior CTL induction in vivo (124). In any case, cross-presentation of the internalized antigen was enhanced with simultaneous co-encapsulation of either TLR3 or TLR9 ligands and the antigen (140, 141). The enhanced vaccine efficiency manifests in prolonged presentation of antigen derived epitopes and superior antitumor responses in mice (71, 124, 142). For example, codelivery of PLGA NPs-OVA together with the TLR4 ligand MPLA (143) or the melanoma antigen TRP2 with another TLR4 ligand (7-acyl lipid A) (144) generated improved antigenspecific responses.

Surprisingly, other studies have come to the opposite conclusion, namely that co-administration of antigen and TLR ligand in different PLGA particles [PLGA NP OVA + PLGA NP (MPLA + R837)] yields better results compared to co-delivery of antigen and adjuvant in the same nanoparticle (PLGA NP OVA/MPLA/R837). Mentionable, these studies only focused on the humoral response and IgG1 and IgG2a production and have not analyzed cellular immunity. Of note, TLR7 (the receptor for R837) is not expressed in the cross-presenting CD8α+ splenic DC subset (145) which may cause inferior responsiveness toward imidazoquinolines. Moreover, the discrepancy between coencapsulation and co-administration strategies probably depends on the particulate nature, the encapsulated antigen, the route of administration and the choice of adjuvant. Another possibility to enhance immunogenicity of PLGA MS mediated vaccine delivery system is co-encapsulation of multiple specific CTL epitopes. It was already demonstrated that administration of two OVAderived epitopes into one PLGA microsphere elicited substantial IFN-γ secretion in vivo (146).

### ANTI-TUMOR RESPONSES TO IMMUNOTHERAPY WITH PLGA PARTICLES

Co-delivery of antigen and adjuvant to DCs is required for PLGA MS-mediated anti-tumor immunotherapy. Reduction of tumor growth in various syngeneic tumor models in mice was better compared to the same antigen emulsion in IFA. Both, protective as well as therapeutic treatment with PLGA MS OVA/CpG + PLGA MS polyI:C elicited potent antitumor activity in subcutaneous tumor models as well as in lung metastasis models using EG-7 thymoma or the aggressive MO-5 melanoma tumor cells in mice (129). Remarkably, even a single administration of co-encapsulated OVA/CpG microspheres completely protected mice from tumor growth (129). Increased anti-tumor activity using PLGA associated nanoparticulate vaccines was shown by others as well. PLGA NP OVA/polyI:C or PLGA NP OVA/CpG exerted potent anti-tumor activity against subcutaneously implanted EG-7 tumor cells (147). Noticeably, Silva et al. demonstrated decreased growth of B16F10 melanoma in both therapeutic and prophylactic settings using MHC class I or II restricted melanoma peptides Melan-A and gp100 encapsulated into PLGA NP together with either one or both of the TLR3 and TLR9 ligands polyI:C and CpG (148). This study offered several distinct conclusions besides confirmation of the fact that co-encapsulation of antigens and adjuvants in PLGA particles improves antigen-specific anti-oncogenic immunity. First, the study shows synergistic effects of enhanced anti-tumor activity by co-encapsulation of the two immunopotentiators CpG and poly(I:C) into one particle. Second, mice were slightly more protected from tumor growth after immunization with nanoparticles containing two MHC class I-restricted melanoma epitopes simultaneously. Furthermore, the authors propose that co-administration of PLGA NPs with either an MHC class I or an additional MHC class II restricted epitope along with

both TLR ligands induced almost complete blockage of tumor growth, suggesting the important activation of both CD8+ and CD4+ T cell responses for efficacy of anti-tumor immunity. IFN-γ secreting CD4+ T cells facilitate the differentiation of tumor antigen-specific CTLs and promote the recruitment of cells from the immune system participating in tumor cell containment (149). Further, tumor specific CD4+ T cells regulate the survival of CD8+ memory T cells (150). Combined TLR ligation on DCs triggering both MyD88 dependent and independent TLR mediated signaling pathways in parallel has already been demonstrated to promote broader activation of DCs. The marked increase in pro-inflammatory cytokine production and expression of co-stimulatory molecules resulted in enhanced T cell responses in vivo or even insensitivity to the immunosuppressive activity of Tregs at tumor sites (151– 153). Additionally, tumor-induced immunosuppression of DCs is one of the main causes for ineffective anti-tumor responses (154). Thus, co-delivery of tumor antigens together with TLR ligands in PLGA MS not only targets the antigen to DCs, but might also rescue impaired DC function from tumor induced immunosuppression (155, 156).

## Alternatives to TLR Ligands as Immunomodulatory Compounds

In addition to TLR ligands, it is possible to include lipid antigens (e.g. the extremely potent glycosphingplipid α-galactosylceramide, α-GalCer), which activate natural killer T (NKT) cells by binding to the non-classical MHC CD1 molecules. This unique subset of the T cell lineage acts as a potent adjuvant in immune responses against cancer by downstream activation of both innate and adaptive immune responses (157). Although not directly killing tumor cells, NKT cells simulate the cross-priming of tumor antigens by DCs through rapid secretion of large amounts of IFN-γ, IL-12, and IP10 (IFN-γ inducible protein 10) and are able to induce further recruitment of NK cells, macrophages, DCs, CD4+ and CD8+ T cells to tumor sites (157). A combination of α-GalCer and the TLR4 agonist MPLA into PLGA microspheres markedly increased cellular immune responses (158). Moreover, coencapsulation of the invariant NKT cell agonist, together with the TLR 7/8 agonist R848 (Resiquimod) and polyI:C into PLGA nanoparticles enhanced CD4+ and CD8+ mediated anti-tumor responses mainly dependent on DC condition via NKT cells (159). Interestingly, a non-glycosidic derivate of α-GalCer, threitolceramide (ThrCer) has already proven clinically effectiveness in human and mice (160).

# Tumor Lysate as Antigen Source for Particulate PLGA Mediated Cancer Immunotherapy

As outlined above, endogenous and exogenous antigen supply in DC-mediated cancer immunotherapy has faced major limitations such as peptide degradation, rapid turnover of peptide/MHC complexes or dissociation of peptide from MHC during DC preparation/injection (161). This was likely attributed to the fact that only a limited number of peptides with few if any T helper peptides were used (162). Additionally, immunotherapy of solid malignancies is often hampered by low numbers of tumor-specific T cells due to inefficient antigen delivery of DC-based immunotherapy. Moreover, re-administered DCs displayed poor migratory capacity, thus limiting the amount of antigen presented to T lymphocytes in local dLN (163). The use of whole tumor lysates (TL) bypasses the limited potency of single antigen delivery thus broadening the repertoire of defined TAAs and neoantigens and thereby enhancing the probability of generating polyvalent, tumor-associated and antigen-specific CTL responses. Simultaneous stimulation of both CD8+ restricted CTL responses and CD4+ T helper cells is potentially complex enough to overcome the ability of tumors to down-regulate targeted antigens. PLGA MS coencapsulating TRAMP-prostate derived tumor lysate and TLR ligands showed promising ex vivo cytotoxic T lymphocyte responses and achieved elimination of large tumor masses in vivo in TRAMP mice, a transgenic mouse model for prostate cancer (133). The anti-tumor efficacy of tumor lysate co-encapsulated with CpG ODNs in PLGA MS was also shown by Goforth et al. in a mouse model for melanoma (164). As well, a prime boost regimen of microspheres containing lysates of mammary gland tumor cells followed by a booster vaccination of bulk tumor lysate together with TLR ligands in liposome formulation was able to ameliorate tumor growth in a murine breast cancer model (165). Noticeably, patient-derived DCs loaded with PLGA NPs encapsulating lysed tumor tissue from patients with advanced head and neck squamous cell carcinoma (HNSSC) could efficiently induce IFN-γ production and could significantly reduce IL-10 secretion in autologous CD8+ T cells (166). Similar findings were made by Hanlon et al. demonstrating increased production of pro-inflammatory cytokines in healthy donor DCs that were pulsed with PLGA nanoparticles encapsulating tumor lysate of an ovarian cancer cell line (167). Malignant cells have developed prodigiously smart mechanisms to coopt immune cells for tumor progression thereby creating an immunosuppressive microenvironment. Cancer cells are able to attract immunosuppressive cell types such as regulatory T cells (Tregs) and myeloid derived suppressor cells (MDSCs) and are known to drive TAM (tumor-associated macrophage) differentiation to the pro-tumorigenic M2 phenotype. PLGA MS mediated cancer therapy might be an ideal strategy to revert these immunosuppressive mechanisms by inducing factors that are essential for cytotoxicity against cancer cells such as intratumoral activated CD8+ T cell response and IFN-γ production as well as recruitment of NK cells. Moreover, upregulation and overexpression of immune checkpoints CTLA-4 (cytotoxic T lymphocyte associated antigen 4) or PD-L1 (programmed cell death protein ligand 1) on cancer cells induces T cell anergy and maintains Treg induced immunosuppression. Thus, it might also be interesting to combine PLGA microsphere-based immunotherapy with immune checkpoint inhibitors to restore T cell anti-tumor effector function.

Apart from generation of anti-tumor responses, we additionally could demonstrate the preeminence of PLGA MS in infectious diseases. PLGA MS encapsulated Influenza virus matrix M1 peptide together with CpG induced potent TABLE 1 | Main advantages of PLGA microspheres as a DC-mediated particulate vaccine delivery system for cancer immunotherapy.


anti-viral CTL responses and protected against Influenza A infection (168).

In relation to a potential use of PLGA MS in clinical application, sterilization of PLGA MS by γ-irradiation did not negatively affect T cell responses (133). The biggest advantage of spray-dried PLGA microspheres is the high reproducibility of the low-cost MS production meeting GMP requirements of efficacy, safety and stability of pharmaceuticals.

#### FUTURE PERSPECTIVES

PLGA microspheres have demonstrated great proficiency for potential use in cancer immunotherapy (see **Table 1**). They have overcome the major challenges of drug delivery systems, such as protection of encapsulated material from rapid degradation and clearance. PLGA MS exhibit ideal properties for facilitated and untargeted uptake of mainly DCs after subcutaneous injection. Concomitant delivery of antigens and adjuvants to the same APC leads to efficient DC activation and increased stimulation of CD4+ T cells as wells as of CD8+ T cells via cross-presentation by coordinate and synergistic pathways. PLGA MS mediated drug delivery allows particularly low doses of antigens and adjuvants–still inducing strong CTL responses but minimizing potential sideeffects of unspecific activation of systemic immune responses. Reducing the doses of antigen or immunostimulants is generally desired regarding potential clinical application or approval by international regulatory agencies. Sustained and prolonged antigen release induces superior immune responses and CD8+ T cell memory while simultaneously avoiding the risk of tolerance induction. The depot effect created at the injection site substitutes the need for conventional booster injections to maintain immune responses. Co-encapsulation of antigens together with toll-like receptor ligands yields potent and longlasting CTL and T helper cell responses in vivo leading to protective and therapeutic anti-tumor activity in several tumor mouse models.

Despite the mentioned advantages of PLGA particles, particulate cancer vaccines are not available for clinical application at present. By far, most in vitro and preclinical mouse studies have been performed with model antigens and model tumors. It is important to switch to clinically relevant antigens and autochthonous, transgenic or carcinogeninduced tumor models for more realistic efficacy assessments in the future. Moreover, the production of GMP-grade PLGA MS needs to be established and refined to get approval for clinical studies. Translation form bench-side into the clinic has always been challenging due to various aspects including characterization of all materials used, availability of cGMP products, the presence of residual organic solvent impurities, difficulties in controlling encapsulated drug release including high initial burst and incomplete release, variability in particle size or morphology between different batches and safety issues including effectiveness and ease of administration in human cancer patients. Increasing the implementation of process analytical technologies (PAT) will control manufacturing and development of PLGA particles to ensure reproducible, effective and safe vaccines and clinical transition. The spray drying process would overcome limits of applicability in larger clinical settings, since the production of PLGA MS is easy to scale-up, cost-effective and amenable to sterile manufacturing. Unlike vaccines for infectious diseases, cancer vaccines might need to be tailored for individual patients due to diverse gene mutations in cancer cells creating neo-antigens. Hence, the development of custom-designed whole tumor lysate encapsulated into personalized PLGA MS might introduce a very promising, rapid and potent cancer treatment approach. Tumor lysates provide a pool of tumor-associated antigens to trigger suitable CD8+ and CD4+ T cell mediated anti-tumor responses that overcome the infirmities of single peptide vaccinations. Currently we are investigating PLGA MS mediated immune responses of used immunostimulatory molecules in VaccigradeTM , GMP certified and endotoxin-free formulations as well as other adjuvant candidates.

In summary, concomitant delivery of antigens and immunomodulators in PLGA microparticles reveals a potent DC—centered therapeutic approach for inducing strong antitumor immunity in various cancer settings which might pave the way for PLGA microspheres to become a key member of current cancer vaccines.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

JK wrote the manuscript. DH prepared the figures. MG supervised associated projects and corrected and refined the manuscript.

#### FUNDING

This project has received funding from the European Union's Horizon 2020 research and innovation program under grant agreement No 686089. The underlying studies have previously been supported by grants nrs. 70112413, 107943, and 102290 of Deutsche Krebshilfe.

#### ACKNOWLEDGMENTS

The authors would like to thank Bruno Gander for advice and Ying Waeckerle-Men, Edith Uetz-von Allmen, Eva Schlosser, Marc Mueller, Valerie Herrmann, and Annette Sommershof for previous experimental work that contributed to this review.


carrier platforms for vaccine delivery. Nanomedicine. (2014) 9:2703–18. doi: 10.2217/nnm.14.156


induce dendritic cell cross-presentation, CTL responses, and antiviral protection. J Immunol. (2011) 186:2422–9. doi: 10.4049/jimmunol.1002845


antigen using fusion molecules or biodegradable microparticles. J Pharm Sci. (2007) 96:3283–92. doi: 10.1002/jps.20978


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Koerner, Horvath and Groettrup. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Sources of Type I Interferons in Infectious Immunity: Plasmacytoid Dendritic Cells Not Always in the Driver's Seat

Shafaqat Ali 1,2, Ritu Mann-Nüttel <sup>1</sup> , Anja Schulze<sup>1</sup> , Lisa Richter <sup>1</sup> , Judith Alferink 2,3 and Stefanie Scheu<sup>1</sup> \*

1 Institute of Medical Microbiology and Hospital Hygiene, University of Düsseldorf, Düsseldorf, Germany, <sup>2</sup> Cluster of Excellence EXC 1003, Cells in Motion, Münster, Germany, <sup>3</sup> Department of Psychiatry, University of Münster, Münster, Germany

Type I Interferons (IFNs) are hallmark cytokines produced in immune responses to all classes of pathogens. Type I IFNs can influence dendritic cell (DC) activation, maturation, migration, and survival, but also directly enhance natural killer (NK) and T/B cell activity, thus orchestrating various innate and adaptive immune effector functions. Therefore, type I IFNs have long been considered essential in the host defense against virus infections. More recently, it has become clear that depending on the type of virus and the course of infection, production of type I IFN can also lead to immunopathology or immunosuppression. Similarly, in bacterial infections type I IFN production is often associated with detrimental effects for the host. Although most cells in the body are thought to be able to produce type I IFN, plasmacytoid DCs (pDCs) have been termed the natural "IFN producing cells" due to their unique molecular adaptations to nucleic acid sensing and ability to produce high amounts of type I IFN. Findings from mouse reporter strains and depletion experiments in in vivo infection models have brought new insights and established that the role of pDCs in type I IFN production in vivo is less important than assumed. Production of type I IFN, especially the early synthesized IFNβ, is rather realized by a variety of cell types and cannot be mainly attributed to pDCs. Indeed, the cell populations responsible for type I IFN production vary with the type of pathogen, its tissue tropism, and the route of infection. In this review, we summarize recent findings from in vivo models on the cellular source of type I IFN in different infectious settings, ranging from virus, bacteria, and fungi to eukaryotic parasites. The implications from these findings for the development of new vaccination and therapeutic designs targeting the respectively defined cell types are discussed.

Keywords: type I interferon, plasmacytoid dendritic cells, interferon producing cells, infection, pathogen, virus, immunopathology, immune activation

# INTRODUCTION

The cytokine family of type I IFNs fulfills key functions in anti-viral immunity but is also produced in the immune responses to other classes of pathogens covering viruses, bacteria, parasites, and fungi (1). Additionally, these cytokines are functionally involved in the pathogenesis of inflammatory autoimmune diseases (2).

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Susan Kovats, Oklahoma Medical Research Foundation, United States Jennifer Johanna Lühr, University Hospital Erlangen, Germany

\*Correspondence: Stefanie Scheu stefanie.scheu@uni-duesseldorf.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

> Received: 31 August 2018 Accepted: 25 March 2019 Published: 12 April 2019

#### Citation:

Ali S, Mann-Nüttel R, Schulze A, Richter L, Alferink J and Scheu S (2019) Sources of Type I Interferons in Infectious Immunity: Plasmacytoid Dendritic Cells Not Always in the Driver's Seat. Front. Immunol. 10:778. doi: 10.3389/fimmu.2019.00778

Together with IFNβ, type I IFNs comprise multiple IFNα subtypes (11 in mice and 13 in humans), IFNε, IFNκ, and IFNω in most mammals. In addition, IFNδ, IFNζ (limitin), and IFNτ have been detected exclusively in pigs, mice, and ruminants, respectively (3–6). Type I IFNs are encoded by intronless genes clustered in mice on chromosome 4 and in humans on chromosome 9 (3–6). Induction of type I IFN expression is facilitated after activation of a diverse set of pathogen sensing pattern recognition receptor (PRR) pathways by binding of IFN regulatory factors (IRFs) and NF-κB to acute response elements in the promoters of type I IFN gene loci (7). All type I IFNs bind to a common heterodimeric IFNα receptor (IFNAR), which is composed of the IFNAR1 and IFNAR2 subunits and is expressed by virtually all nucleated cells of the body. Following IFNAR engagement by its ligands, canonical type I IFN signaling activates the Janus kinase (JAK) signal transducer and activator of transcription (STAT) pathway, leading to transcription of IFN-stimulated genes (ISGs) (5, 8). ISG-encoded proteins mediate induction of cell-intrinsic antimicrobial states in infected and neighboring cells that limit the spread of infectious agents, particularly viral pathogens. Additionally, ISGs influence innate and adaptive immune responses by promoting antigen presentation and NK cell functions, modulating inflammatory cytokine production, and activating high-affinity antigen-specific T and B cell responses and immunological memory (9). Type I IFN production, however, can also have deleterious roles in chronic viral and bacterial infections, and can lead to immunopathologies such as inflammatory disorders and autoimmunity (1, 2, 10, 11).

IFNβ was originally defined as the antiviral factor produced by fibroblasts after viral infections (12) and has been thought to be produced by virtually all cells of the body. Later pDCs specialized in the rapid secretion of high amounts of type I IFN have been termed the natural "IFN producing cells" (IPCs). Recent findings, however, indicate that production of type I IFN, especially the early synthesized IFNβ, in anti-infectious immune responses can occur independently of pDCs and that the cell type responsible for type I IFN production rather depends on the specific infectious setting. In this review we summarize the recent findings on the identity and function of type I IFN producing cells in infection by focusing on insights gained from in vivo mouse models covering type I IFN reporter mice and models of cell type specific ablation.

#### Pathways of Type I IFN Activation in Different Cell Types

To devise novel anti-infectious treatment regimens targeting a specific cellular subtype, it is crucial to know the identity of the cells responsible for the production of type I IFN in the course of an infection. Early on, pDCs were considered primary producers of IFNα during virus infections (13, 14). For human pDCs it has been reported that IFNα/β transcripts account for an astounding 50% of all mRNAs in the cell after viral activation (15). More than 40 years ago, pDCs were first described in humans as natural IPCs that activate NK cells upon exposure to viruses (16, 17). The murine equivalent was described in 2001 as type I IFN producing cells with plasmacytoid morphology (18–20). These cells detect RNA and DNA viruses through two endosomal sensors, TLR7 and TLR9, respectively, which induce secretion of type I IFN through the MyD88-IRF7 signaling pathway (21– 24). Specifically, TLR7/9-ligand interactions in early endosomes result in type I IFN production while ligand recognition in late endosomes or lysosomes rather leads to inflammatory cytokine production and pDC maturation (25, 26). At least in the mouse, TLR7 and 9 are also expressed by monocytes, conventional DCs (cDCs), and B cells (27, 28). Therefore, the contribution of those cell types to type I IFN production triggered via the TLR7/9- MyD88-IRF7 pathway has to be considered. B cells, for instance, have recently been shown to produce type I IFN in vivo after optimized stimulation conditions using the TLR9 ligand CpG-A (29). A specific feature of pDCs is that they can produce type I IFN independently of IFNAR mediated feedback signaling (30). However, they do respond to type I IFN by generating an autocrine circuit through IFNAR, which augments type I IFN secretion and induces their activation and migration (31, 32).

In humans, pDCs, monocytes, and other myeloid cells also produce type I IFN after stimulation of the TLR8-MyD88-IRF7 pathway by viral single-stranded RNA (ssRNA) (33, 34). The mouse TLR8 was initially considered non-functional (33, 34). More recently it has been shown that mouse TLR8 can be stimulated by a combination of oligodeoxynucleotides (ODNs) and human TLR8 ligands. Further, mouse pDCs produce type I IFN after stimulation with vaccinia virus (VV) in a TLR8 dependent way (35, 36). Two additional TLRs, TLR3 and 4, are able to induce type I IFN expression independently of the MyD88 pathway via recruiting the TIR domain-containing adaptor protein inducing interferon beta (TRIF; also known as TIR domain-containing adapter molecule 1, TICAM-1). This activates the transcription factor IRF3 thus initiating type I IFN, in particular IFNβ expression (37, 38). TLR3 is absent in mouse pDCs but highly expressed in endosomes of murine CD8α + and CD103<sup>+</sup> and human CD141<sup>+</sup> cDCs of the DC1 subtype that are efficient in cross-presenting (39, 40). It recognizes double-stranded RNA (dsRNA) as viral replication intermediates as well as ssRNA containing stem loops (41). In addition to DCs, TLR3 activation can lead to type I IFN expression in macrophages, fibroblasts, and epithelial cells (42). While TLR3 exclusively signals via the TRIF pathway, TLR4 utilizes MyD88 as well as TRIF signaling routes after recognizing its cognate ligand bacterial lipopolysaccharide (LPS). Analogous to TLR3 activation, LPS binding to TLR4 induces type I IFN expression via TRIF-IRF3 (43). The majority of hematopoietic cells of the myeloid and lymphoid lineage, with the exception of human pDCs, and few other cell types such as pancreatic β-cells express TLR4 (44).

In contrast to pDCs, cDCs, and macrophages mainly produce type I IFN in response to virus challenge by utilizing retinoic acidinducible gene I (RIG-I)-like helicases (RLHs) (43, 45–47). RLHs, including RIG-I and melanoma differentiation-associated gene 5 (MDA5), are cytoplasmic dsRNA receptors that transmit their signal through the mitochondrial antiviral-signaling protein, virus-induced signaling adapter (MAVS, aka IFNb promoter stimulator (IPS)-1 or Cardif). This activates IRF3 and IRF7 to induce the transcription of type I IFN and other antiviral genes (48–50).

Finally, soluble sensors in the cytoplasm detect dsDNA in a sequence-independent manner, exhibit a broad expression spectrum including pDCs, cDCs, macrophages, and mouse embryonic fibroblasts (MEFs), and activate signaling pathways leading to type I IFN expression (47, 51). These sensors include the cyclic GMP-AMP synthase (cGAS)/STING pathway, the RNA polymerase III/RIG-I/MAVS pathway, DNA-dependent activator of IRFs (DAI), IFNγ-inducible protein 16 (IFI16), and the DDX family (47, 51–58).

#### Mouse Models and in vivo Experimental Strategies for the Definition of the Cellular Source of Type I IFNs in Infection

Several models of cytokine reporter mice have been developed for the detection of type I IFN production in vivo, as intracellular staining is not sensitive in most cases (**Table 1** and **Figure 1**). Earlier published IFNβ knock-out mouse lines already contained reporter elements to detect Ifnb promoter-driven gene transcription. For example, coding sequences for the mouse immunoglobulin λ2 chain, a green fluorescent protein (GFP), or the human CD2 had been inserted immediately downstream of the Ifnb promoter to visualize IFNβ expression on a cellular level (73–75). However, the reporter features in these mouse strains have not been used in vivo so far.

More recently, a mouse line expressing GFP under the control of the Ifna6 promoter (Ifna6gfp/+) recapitulates the expression of various IFNα genes and has been employed to define the cellular source of IFNα in virus infection models (32, 46, 76). Also, for IFNβ a fluorescence reporter-knock-in mouse model (IFNβ mob/mob) has been generated. Here, yellow fluorescent protein (YFP) is expressed from a bicistronic mRNA linked by an internal ribosomal entry site (IRES) to the endogenous IFNβ mRNA (59). Ifna6gfp/<sup>+</sup> as well as IFNβ mob/mob reporter mice have each been shown to report for the majority of type I IFNs. However, in vitro analyses on BM-derived DCs from the double reporter mouse line generated by intercrossing the Ifna6gfp/<sup>+</sup> and IFNβ mob/mob reporter strains revealed that specific type I IFN subtypes can be produced by distinct cell subpopulations (77).

In an alternative reporter mouse system, a firefly luciferase reporter gene has been placed under the control of the Ifnb promoter (IFN-β <sup>+</sup>/1β−luc). Rather than IFNβ expression on a single cell level, this model detects in vivo kinetics of IFNβ expression in the mouse paralleling the spread of pathogens through the organism under infectious conditions (60). Additionally, in this mouse line the IFNβ coding sequence is flanked by loxP sites (IFN-β floxβ−luc/floxβ−luc) providing the possibility to characterize the impact of IFNβ production by a given cell type on the pathophysiology of various infections via tissue- or cell-specific Cre-mediated deletion of IFNβ (60).

#### Methods and Models for Assessing the Impact of Type I IFN Producing Cell Populations in vivo

Several experimental strategies have been developed to determine the in vivo contribution of a specific cell type to the type I IFN response during infections (78). Initially, antibody mediated depletion has been utilized frequently to ablate pDCs and monocytes (79). Antibodies against Ly6G/C (also known as Gr1) and CD317 (also known as BST-2) have been used to deplete pDCs in vivo and in vitro (18, 79–87). However, these antibodies generally target multiple cell types in addition to pDCs: The antibody RB6-8C5 directed against Ly6G/C reacts strongly with neutrophil-specific Ly6G antigen, but cross-reacts also with the Ly6C Ag (88) expressed on pDCs as well as on monocytes/macrophages, activated T cells, NK cells, plasma cells, and endothelial cells (89–92). Likewise, CD317 is recognized by the three different antibody clones 120G8.04, JF05-1C2.4.1 (also known as PDCA-1), and eBio927, and is expressed in naïve mice by pDCs, but also plasma cells. Following stimulation with type I IFNs and IFNγ CD317 is upregulated, additionally, on several other myeloid and lymphoid cells (79, 93). Finally, in vivo treatment with clodronate-containing liposomes depletes phagocytes in mice, but also disturbs the microarchitecture of secondary lymphoid organs (94, 95).

In the past years, several genetically modified mouse lines with a constitutive or inducible lack of specific cell types attributed to produce type I IFN have become available (**Table 1** and **Figure 1**) (78). For pDCs, already several mouse models exist for constitutive or inducible ablation. Mice carrying a hypomorphic mutation at the Ikaros locus express low levels of the transcription factor Ikaros (IkL/<sup>L</sup> ) and therefore lack peripheral pDCs, but no other DC subsets (61). When using this line as a "pDC-less" model, one has to take into account that other hematopoietic lineages including T and B cells and neutrophils are also affected by the IkL/<sup>L</sup> mutation, and that IkL/<sup>L</sup> mice start to develop thymic lymphomas by 10 weeks of age (62, 96, 97). Constitutive deletion of E2-2, the basic helix-loop-helix transcription factor, also known as TCF4, that controls development and maintenance of pDCs, results in perinatal lethality in mice (98). To overcome this lethality and to specifically ablate the pDC lineage, mice harboring a constitutively deleted and a floxed Tfc4 allele (Tcf4flox/−) have been crossed to Itgax-Cre (CD11c-Cre) or Rosa26-CreER mice in which Cre is expressed in DCs or can be induced ubiquitously after tamoxifen administration, respectively (63, 64, 99, 100). Another strategy uses Diphtheria toxin receptor (DTR) mediated conditional and targeted cell depletion. CLEC4A-DTR-tg mice express DTR under the human pDC specific promotor of the C-type lectin domain family 4 member A (CLEC4A; also known as blood dendritic cell antigen 2, BDCA2). Administration of diphtheria toxin (DT) in these CLEC4A-DTR-tg mice results in transient but specific depletion of pDCs (65). In an alternative approach, a cDNA encoding the human DTR fused to the enhanced green fluorescent protein (EGFP) and preceded by an IRES was inserted into the 3′ untranslated region of the Siglech gene. This Siglechdtr/dtr mouse model allows specific elimination of pDCs in vivo via injection of DT (66). An analogous mouse line termed SiglecH-DTR-tg was generated using bacterial artificial chromosome (BAC) transgenic technology (67). SiglecH represents a sialic acid–binding Ig-like lectin that exerts immunomodulatory roles in antiviral immune responses. In SiglecHeGFP/+mice, heterozygous for the reporter gene, it was shown that in addition to pDCs, SiglecH was

TABLE 1 | Genetically modified mouse models to visualize or define the function of type I IFN producing cells.


expressed in specialized macrophage subsets, such as marginal zone macrophages (MZM), lymph node medullary macrophages, and microglia. SiglecH was also found in immediate precursors of pDCs (pre-pDCs) in the BM, which have the plasticity to differentiate into pDCs and cDCs (67, 101). Despite of SiglecH expression on above described other cell types Loschko et al. showed pDC specific antigen delivery in mice by using SiglecH as a target structure (102) suggesting the usability of SiglecH as a lead molecule for the generation of pDC specific transgenic animals. A side by side comparison showed a higher susceptibility to Listeria monocytogenes infection in DT-treated SiglecH-DTRtg vs. CLEC4A-DTR-tg mice. This finding was attributed to the additional lack of MZM in SiglecH-DTR-tg mice after DT treatment which was not observed in CLEC4A-DTR-tg mice (67).

With the aim to specifically express the Cre recombinase in pDCs a BAC-tg "pDCre" mouse line was generated which expresses Cre under the control of the Siglech promoter (68). By crossing these mice with a reporter mouse line that indicates Cre activity via red fluorescent protein (RFP) expression the authors found ∼30% of SiglecH<sup>+</sup> pDCs terminally labeled with RFP. Additionally, RFP expression was observed in a minor fraction of SiglecH<sup>−</sup> B-, T-, NK-, and NK-T cells and splenic cDCs and CD11cint BM cells suggesting that a small fraction of early lymphoid progenitors actively transcribes the SiglecH locus. Thus, the broader expression pattern of SiglecH should be considered when using SiglecH-DTR-tg mice to evaluate pDC functions in vivo.

Recently, a novel mouse model has been described in which type I IFN production is restricted to pDCs. In this knockin model Irf7 expression is driven by the Siglech promoter (SiglechIrf7/+). The SiglechIrf7/<sup>+</sup> mice were then backcrossed onto Irf3−/−/Irf7−/<sup>−</sup> double knock-out mice which are deficient

FIGURE 1 | Each model system harbors specific advantages and caveats as further described in Table 1. B, B cell; BM, bone marrow cell; cDC, conventional dendritic cell; MMM, marginal metallophilic macrophage; MZM, marginal zone macrophage; MO, monocyte; NK, natural killer cell; Nϕ, neutrophil; SSM, subcapsular sinus macrophage; T, T cell; pDC, plasmacytoid dendritic cell. The figure was created using Servier Medical Art according to Creative Commons Attribution 3.0 Unported License (https://creativecommons.org/licenses/by/3.0/). Changes were made to the original cartoons.

in type I IFN production. This yielded animals (referred to as "pDC:Irf7+" mice) in which IRF7 signaling required for type I IFN expression is functional exclusively in pDCs (69). Additionally, in these mice an IRES site followed by coding sequences for Cre fused to the mutated ligand binding domain of the human estrogen receptor (ERT2) (103) was inserted behind the Irf7 gene into the Siglech gene locus (69). Therefore, this mouse line can potentially be used in the future for tamoxifen inducible pDC specific Cre expression and thus pDC specific gene deletion when crossed to the respective floxed mouse lines.

Also, for other cell types than pDCs, the DTR-mediated depletion approach has been employed. In recent years, mice expressing the DTR under the control of the CD11b-, CD11c-, and CD169-promoters have been generated and successfully used for depletion of monocytes, cDCs, and CD169<sup>+</sup> macrophage subpopulations such as MZMs and subcapsular sinus macrophages in the spleen and lymph nodes (70–72, 104–109).

In the following sections we will discuss approaches designed to define the type I IFN producing cell types in infection using the in vivo mouse models described above.

# VIRAL INFECTIONS

In this chapter we will focus on more recent findings from in vivo models aimed at visualizing IFNα/β producing cell types and defining their contribution to the overall type I IFN production and their impact on the course of viral infections (**Table 2**). For a more generalized overview of the cellular sources of type I IFN in viral infections we kindly refer to an expert review by Swiecki et al. (127).

#### DNA Viruses

Findings on the cellular sources of type I IFN during relevant infection models for DNA viruses and the respective in vivo experimental strategies are highlighted in the following sections.

#### Human and Mouse Cytomegalovirus

Infection with the human cytomegalovirus (HCMV) causes mostly asymptomatic, latent infections in the immunocompetent host. In immunosuppressed individuals or newborns infected in utero, an infection with this virus can lead to severe illness and permanent organ damage. The murine cytomegalovirus (MCMV) exhibits high structural and biological similarity to HCMV and is thus widely used as a model system for antiviral immune responses (128). MCMV induces a biphasic type I IFN response, with peak expressions occurring at 8 h and 36–72 h p.i. which are triggered by the initial virus contact and viral particles entering the system after completion of the first viral replication cycle, respectively (110). Early type I IFN expression is independent of TLR signaling and predominantly generated by stromal cells infected by the virus (110). Using IFNβ mob/mob reporter mice, IFNβ production was detected in splenic pDCs as early as 6–12 h p.i. (59, 111). After in vivo depletion of pDCs by anti-CD317 or anti-Ly6G/C treatment IFNα serum levels were severely reduced 36 h after MCMV infection (18, 80, 81, 113). Under these conditions, however, other cell types secrete IL-12 and ensure sufficient IFNγ and NK cell responses leading to control of MCMV infection (18, 80). Of note, 44 h after MCMV infection IFNα serum levels in pDC depleted mice were no longer reduced as compared to untreated mice (113). Similar observations were made in IkL/<sup>L</sup> mice that lack pDCs (61) or CLEC4A-DTR-tg mice that have been transiently depleted of pDCs (65, 113). Thus, transient type I IFN production at the first day of MCMV infection was pDC-dependent, while cells other than pDCs are responsible for the type I IFN levels measured at later timepoints, at least when relatively high inocula of MCMV are used. In contrast, at lower doses of MCMV which are presumably closer to a natural infection setting, pDCs can limit viral burden in the spleen and liver. Here, pDCs have been shown to promote NK cell activation and cytotoxicity in the early phase of MCMV infection (65). While it is well-established that pDCs sense the MCMV via the TLR9 and TLR7 mediated pathways (18, 80, 113, 129, 130), also the TLR3 and TLR2 pathways which are functionally used by other cells than pDCs have been shown to be involved in the induction of type I IFN production (104, 131, 132). These findings are in accordance with multiple observations that defects in MyD88 signaling have a more severe impact on anti-MCMV immune responses than TLR9 deficiency or pDC depletion (113, 129). So far, the identity of the non-pDC cell types involved in anti-MCMV type I IFN response remain incompletely defined.

#### Vaccinia Virus

One report indicated that vaccinia virus (VV) and to a lesser extend MCMV induce type I IFN in CD11c<sup>−</sup> CD11b<sup>+</sup> Ly6C<sup>+</sup> inflammatory monocytes, but not macrophages or other types of DCs, in a TLR2 dependent way using IFNβ mob/mob reporter mice. Further, CD11b-DTR-tg mice depleted of monocytes exhibited increased viral titers in the liver and decreased serum levels of type I IFN after VV infection (104). This is similar to other studies using footpad infection of modified vaccinia virus Ankara (MVA) and pDC depletion in the CLEC4A-DTR-tg mouse model, where type I IFN levels in the draining lymph nodes were comparable to control mice indicating that pDCs are not required for mounting an intact type I IFN response after local infection with this dsDNA virus (118).

#### Adenovirus

The dsDNA adenovirus is used as a vector for the development of gene therapy applications but can also cause severe disease in immunocompromised individuals. By using CD11c-DTR-tg TABLE 2 | Cellular sources of type I IFN production in viral infections in vivo.


(Continued)

#### TABLE 2 | Continued


mice and anti-CD317 treatment to ablate cDCs vs. pDCs in vivo it has been shown that wildtype (WT) adenovirus as well as adenoviral vectors induce rapid IFNα/β production almost exclusively in splenic cDCs rather than in pDCs (105).

#### Herpes Simplex Virus

For Herpes Simplex Virus (HSV) local (subcutaneous or genital) as well as systemic (i.v.) infection models have been analyzed. After subcutaneous HSV-1 infection, pDCs were shown to provide type I IFN necessary for licensing of cDCs which in turn induce effective cytotoxic T cell responses. Here, mice depleted for pDCs by anti-Ly6G/C treatment displayed increased viral titers in the draining lymph nodes at day 7 p.i. as compared to controls (82). Similarly, in a genital HSV-2 herpes model, mice depleted for pDCs using anti-CD317 antibodies succumbed earlier to the infection and exhibited reduced local IFNα levels, while the Th1 response in draining lymph nodes developed normally (117). In contrast to findings from antibodymediated depletion, in pDC depleted CLEC4A-DTR-tg mice neither differences in viral burden nor survival after vaginal HSV-2 infection was found nor were pDCs found to contribute significantly to antiviral CD8 T cell responses after subcutaneous HSV-1 inoculation (114). These contradicting findings have been explained by the antibody-mediated depletion of additional cell types other than pDCs in contrast to the more restricted depletion in the CLEC4A-DTR-tg genetic mouse model. On the other hand, it cannot be excluded that DTR mediated depletion is less effective and therefore a residual pDC activity retained after DT administration. Slight differences in the respective experimental settings might contribute as well as e.g. after antibody-mediated pDC depletion IFNα levels were measured in vaginal washes while in the genetic depletion model total protein amount was assessed in the vaginal and cervical tissue itself. As for MCMV, TLR3-expressing cells, such as CD8<sup>+</sup> DCs or other hematopoietic and non-hematopoietic cells, are essential for type I IFN production in local HSV infection at later timepoints rather than pDCs (114).

After systemic challenge with UV-irradiated HSV in an early study immunohistological stainings for IFNα/β indicated that the majority of type I IFN producing cells in the spleen represent marginal metallophilic macrophages and to a lesser extend MZMs (133). However, IFNα levels were markedly reduced in pDC depleted Siglechdtr/dtr mice 6 h after i.v. infection with HSV-1 and viral titers were found increased in the spleen as compared to control animals pointing toward pDCs as the major type I IFN producers in this situation (66). Similar results were obtained in pDC depleted CLEC4A-DTR-tg mice, with the exception that no viral replication was detectable in the spleens of either DT-treated CLEC4A-DTR-tg or control mice (114). This discrepancy may reflect differences in the strains of HSV-1 used or differences in the promoters used to drive DTR expression (CLEC4A vs. SiglecH) with a slightly divergent expression pattern as discussed above. For systemic HSV-2 infections, results from antibody depletion and pDC ablation in CLEC4A-DTRtg mice corresponded well since in both cases a reduction of IFNα serum levels were observed together with increased viral titers and reduced survival (114–116). Thus, similar to vaccinia virus the cell type responsible for the production of type I IFN in HSV infection may depend on the route of pathogen entry with pDCs controlling the infection once the virus has spread systemically.

#### Ectromelia Virus

Ectromelia virus (ECTV), a large DNA orthopoxvirus, is the causative agent of mousepox, the mouse homolog of human smallpox. ECTV causes systemic disease after s.c. infection of the footpad. In vivo it was shown by clodronate and anti-CD317 mediated depletion of monocytes vs. pDCs and ex vivo sorting and RT-PCR analyses that infected inflammatory monocytes are the major producers of type I IFN in the draining lymph nodes (119).

In summary, the cellular source for type I IFN production during DNA virus infection depends on the virus type itself, the dosage, timepoint as well as route of infection. Early after infection with MCMV pDCs are the primary source of type I IFN production capable of reducing virus titers at low concentrations of the virus. However, at later timepoints of infection CD8<sup>+</sup> DCs rather than pDCs become the key source of type I IFN production. In addition to pDCs, other cell types such as cDCs in adenovirus infection, metallophilic macrophages and MZMs during HSV exposure, stromal cells in MCMV infection and inflammatory monocytes in response to ECTV are an essential source of type I IFN.

#### RNA Viruses

A recent meta-transcriptomics survey defined 196 vertebratespecific RNA virus species the majority of which is able to infect humans and cause diseases of varying severity (134, 135). At the moment only few mouse models are available to elucidate the host immune response to these viruses. In this chapter we summarize the in vivo model studies aimed at visualizing type I IFN producing cell types and defining their contribution to the type I IFN production and RNA virus control.

#### Newcastle Disease Virus

For systemic infections with RNA viruses, such as after i.v. inoculation with the paramyxovirus Newcastle disease virus (NDV), it has been shown that pDCs and also cDCs, macrophages, and monocytes, produced IFNα (46). Here, pDCs mount an antiviral type I IFN response in a viral replicationindependent manner through virus recognition by TLR7 and the activation of the type I IFN positive feedback loop. Only in the absence of this type I IFN positive feedback, the virus infects and also replicates in pDCs. In this case, type I IFN induction occurs in pDCs via cytoplasmic RLHs (32). However, other ssRNA viruses have been reported to induce type I IFN expression in pDCs in a replication dependent manner (136, 137). Especially for vesicular stomatitis virus (VSV), the capture of the replicating virus in the autophagosome is required for its transfer to the TLR7 containing endosomal compartment (137). After local infection with NDV, here after intranasal infection, the IFNα-producing cells shifted from pDCs to alveolar macrophages and cDCs that utilize the RLH system for type I IFN induction (46).

#### Vesicular Stomatitis Virus

Upon s.c. VSV infection, draining LNs contained ∼90% less IFNα when depleted of macrophages by clodronate liposomes. However, when pDCs were depleted by anti-CD317 treatment, IFNα levels induced by VSV were reduced only by half as compared to controls. It was concluded that infected CD169<sup>+</sup> subcapsular sinus macrophages produce IFNα, yet half of the type I IFN is produced by pDCs stimulated directly or indirectly by the infected macrophages (108). Later it was shown that CD169<sup>+</sup> macrophages in the spleen represent a compartment of enhanced viral replication (138). Thus, it is conceivable that CD169<sup>+</sup> macrophages potentiate the type I IFN response indeed indirectly via activating pDCs. When VSV was inoculated i.v. in CLEC4A-DTR-tg mice transiently depleted for pDCs, IFNα was found reduced and viral titers increased only at very early timepoints, again pointing to a rather transient role of pDCs in anti-viral immunity (65).

#### Dengue and Chikungunya Virus

For the distantly related arboviruses Dengue (DENV) and Chikungunya (CHIKV) virus it was recently shown that pDCs are sufficient to control these viruses via IRF7-regulated type I IFN responses in both systemic as well as local infection settings. In this report novel pDC:Irf7<sup>+</sup> mice were introduced in which IRF7-driven type I IFN production is restricted to pDCs and were compared to IRF3/7 double deficient mice that are completely devoid of type I IFN expression (69). After i.v. infection with DENV pDC:Irf7<sup>+</sup> mice exhibited a lower viral load than Irf3/7 double deficient mice. However, as compared to WT mice higher viral tiers were detected in pDC:Irf7<sup>+</sup> mice (69). After s.c. infection with CHIKV Irf3/7 double deficient mice succumb to the virus while 100% of pDC:Irf7<sup>+</sup> mice survive the infection exhibiting no overt clinical symptoms similar to WT mice. Early control of viremia in pDC:Irf7<sup>+</sup> mice was reduced as compared to WT but still improved as compared to Irf3/7 double deficient mice. Thus, analogous to findings from other virus infection models also for these RNA viruses, antiviral response mounted by pDCs controls infection once the virus spreads systemically.

#### La Crosse Virus, Rabies Virus, and Theiler's Murine Encephalomyelitis Virus

In infection models with RNA viruses exhibiting a specific tropism, pDCs play only a minor role. In the brain of mice infected with the ssRNA La Crosse virus, IFNβ production was assessed by the IFN-β <sup>+</sup>/1β−luc luciferase reporter mouse model (60) and detected in astrocytes, microglia, and to a lesser extend also in infected neurons (122). This confirmed earlier findings where IFNα/β expression in these cell types after La Crosse virus infection was visualized by immunostaining and RNA in situ hybridization (123). Utilizing the conditional reporter activity of the IFN-β floxβ−luc/floxβ−luc mice it was shown for several neurotropic viruses such as rabies virus (RABV), Theiler's murine encephalomyelitis virus (TMEV), and VSV that astrocytes are the main producers of IFNβ after infection of the brain (139).

#### Encephalomyocarditis Virus

Another example for type I IFN expression by non-myeloid cells represent β-islet cells. The encephalomyocarditis virus (EMCV) strain D, an ssRNA picornavirus with tropism for the insulinproducing β cells of the pancreas, can induce diabetes and myocarditis in certain mouse strains. CD11c<sup>+</sup> cells in this model have been shown to be protective as DT treated CD11c-DTR-tg mice developed diabetes and exhibited increased viral titers in the pancreas, spleen, and heart associated with reduced type I IFN levels as compared to non-depleted controls (106).

#### Pneumonia Virus of Mice

Pneumonia virus of mice (PVM) infection led to a marked infiltration of pDCs and increased expression of type I IFN in WT but not TLR7- or MyD88-deficient mice. Transfer of TLR7 competent, but not TLR7-deficient pDCs led to a significantly diminished virus recovery in TLR7−/<sup>−</sup> animals on day 7 after infection with PVM indicating that TLR7-mediated signaling by pDC is required for appropriate innate responses to acute PVM infection (140).

#### Respiratory Syncytial Virus

For intratracheal infection with respiratory syncytial virus (RSV) it has been shown that anti-CD317 mediated depletion of pDCs completely abolished IFNα expression and protein levels in the lungs. This correlated with increased viral titers and exacerbated immunopathology of the lungs of pDC depleted mice (126). Thus, pDCs fulfill a substantial protective role during local RSV infection.

#### Influenza Virus and Influenza Virus-Like Orthomyxovirus Thogoto Virus

Initial in vitro studies showed that spleen cells from mice that were depleted for pDCs by anti-Ly6G/C injection did not produce IFNα in response to stimulation with inactivated influenza virus in contrast to splenocytes from untreated animals (18). IFNα production in vitro could be attributed to the CD317<sup>+</sup> CD11c<sup>+</sup> pDC population of sorted mouse spleen cells (86). However, in vivo intranasal infection with sublethal doses of influenza virus in pDC-deficient IkarosL/<sup>L</sup> and WT mice revealed a similar course of disease, as determined e.g. by weight loss and viral titers (120). Thus, pDCs are able to produce type I IFN after stimulation by influenza but are dispensable for a successful antiviral immune response. Albeit, type I IFN levels in vivo were not assessed for this infection model.

For the influenza virus-like orthomyxovirus Thogoto virus (THOV) type I IFN production in the peritoneal cavity was mainly attributed to CD11b<sup>+</sup> F4/80<sup>+</sup> myeloid cells that was independent of the type I IFN receptor mediated feedback loop and coincided with the tropism of this virus (121).

#### Mouse Hepatitis Virus

After i.p. infection with Mouse hepatitis virus (MHV), pDC depletion by anti-CD317 was accompanied by severely diminished IFNα serum levels (124). The transient pDC depletion did not lead to lethality following the low-dose MHV infection used in this study. Nevertheless, initial viral titers in spleens were found increased more than 1,000-fold in pDC-depleted compared to control mice (124). Very similar observations were made in Itgax-Cre<sup>+</sup> Tcf4flox/<sup>−</sup> mice lacking pDCs. These mice show reduced serum IFNα levels and elevated viral loads in the liver and spleen (100). Thus, pDCs appear to be essential for type IFN I mediated protection against systemic infection with the prototypical acute cytopathic coronavirus MHV.

#### Lymphocytic Choriomeningitis Virus

Lymphocytic choriomeningitis virus (LCMV) infection is widely used to study acute as well as chronic infections. In an acute infection setting in Ifna6gfp/<sup>+</sup> reporter mice pDCs were found to be the major type I IFN producers after infection with the WE strain of LCMV. Additionally, few cDCs and macrophages specifically in the spleen exhibited GFP-reporter activity (76). Also, in IFNβ mob/mob reporter mice macrophages could be excluded as major type I IFN producers and depletion of phagocytic cells by clodronate liposomes did not affect type I IFN serum levels (94). In contrast, another study using the Armstrong strain of LCMV reported severely reduced IFNα/β serum levels after clodronate treatment (95). Specifically, a small population of CD169<sup>+</sup> macrophages in the spleen and lymph nodes has recently been shown to release high amounts of type I IFN after LCMV infection. Selective depletion of these cells in CD169-DTR-tg mice resulted in reduced type I IFN levels from day 4 p.i. onward and persistent viral titers. As a consequence, CD169 depleted mice exhibited severe immunopathology and died quickly after infection (107). In line with this, production of serum type I IFN was not reduced in LCMV infected mice treated with the pDC depleting anti-Ly6G/C antibody as compared to those injected with control antibody (81). Also, in congenitally pDC-deficient Itgax-Cre<sup>+</sup> Tcf4flox/<sup>−</sup> mice, virus titers early after infection were comparable to WT controls confirming that pDCs are dispensable for the control of acute LCMV infection (100). Still, pDCs have been shown to be a transient source of type I IFN as pDC depletion in CLEC4A-DTR-tg mice led to reduced serum IFNα levels at 16 h p.i. with LCMV Armstrong or clone 13, but not at later timepoints (125). Contrasting the observations in acute LCMV infection, in a chronic infection setup using LCMV

Docile the virus persisted until day 53 in the blood of Itgax-Cre<sup>+</sup> Tcf4flox/<sup>−</sup> mice while the virus was cleared between day 21 and 28 in WT mice. This was attributed to a failure of sufficient CD4<sup>+</sup> and CD8<sup>+</sup> T cell activation in the absence of pDCs and indicated that pDCs are essential for generating a functional adaptive immunity to chronic viral infections (100).

Taken together, pDCs are a major source of type I IFN and are required for type I IFN mediated protection against systemic infection in most of the RNA virus infections such as NDV, VSV, DENV, CHIKV, PVM, RSV, MHV, and LCMV. However, contribution of pDCs in type I IFN release and type I IFN mediated protection depends on the titer of the virus, time after infection, and the route of the infection. In addition to pDCs, other cell types such as cDCs, macrophages, and monocytes in NDV, macrophages in VSV, astrocytes, microglia, and neurons in La Crosse virus, astrocytes in RABV, TMEV, and VSV, β-islet cells and cDCs in EMCV, and cDCs and macrophages in LCMV infection significantly contribute to type I IFN production. Thus, similar to infection with DNA viruses, also after infection with RNA viruses pDCs are functionally involved in type I IFN production mostly early during infection but are dispensable for virus control during later stages of infection. In chronic infection, however, pDCs provide type I IFN to support and preserve T cell functions.

#### Retroviruses

HIV activates pDCs to produce high levels of IFNα most likely via activation of TLR7 (141). Also, it is assumed that type I IFNs are produced during HIV infection predominantly by pDCs as decreased IFNα production in HIV-infected patients correlates with numerical and functional deficiencies in circulating pDCs (142). A direct assessment of the contribution of pDCs to type I IFN levels in HIV-infection, however, has not been performed. Although type I IFNs are known to mediate antiviral immunity, there has always been caution toward a detrimental role of type I IFNs during HIV/AIDS because of their proinflammatory nature (11, 143, 144). Thus, many studies have shown that pDCs are a source of type I IFN in retroviral and other virus infections in vivo. However, additional cellular sources of type I IFN are required to fully control viral infections. In summary, pDCs are a known source of type I IFN in retroviral infection. However, the relative contribution of pDCs vs. other type I IFN producers to the overall type I IFN response and immune control or pathology after retrovirus infection, is not fully understood.

# BACTERIAL INFECTIONS

While a considerable number of studies have been undertaken to define the cellular source of type I IFN and the functions of these cell types in viral infections, fewer data exist for non-viral infections. In bacterial infections, type I IFNs can act as activators of protective immune responses or mediate immunosuppressive functions leading to exacerbation of the infection. This ambivalent role of type I IFN has been reviewed recently (1, 11, 145). In this chapter we will focus on the efforts to clarify the identity and impact of type I IFN producing cells as knowledge on these has increased significantly paralleling the availability of newly developed mouse models.

#### Mycobacteria

It has been well established that CD4 T cells as well as secreted effector cytokines TNF, IL-12, and IFNγ exert protective functions in host resistance to the intracellular bacterial pathogen Mycobacterium tuberculosis (Mtb) (146). In contrast, the role of type I IFN during Mtb infection appears to promote infection instead of controlling infection. Type I IFNs downregulate IFNGR1 expression and thereby suppress IFNγ signaling (147, 148) and IFNAR-deficient mice displayed increased bacterial clearance to infection with Mtb, although bacterial growth in the lung was unaffected (149). In in vitro studies, BM-derived macrophages and DCs have been identified as a possible source of type I IFN in response to Mtb (149, 150). Also, human peripheral blood mononuclear cell (PBMC)-derived macrophages and especially DCs were shown to produce type I IFN after in vitro infection with Mtb (151, 152).

Similar to Mtb, IFNγ promotes antimicrobial activity against Mycobacterium leprae whereas type I IFNs contribute to pathogenesis (153). Here, PBMC-derived monocytes expressed IFNβ and IFN-stimulated genes including the immunosuppressive cytokine IL-10 during M. leprae infection in vitro (153). So far the cell type expressing type I IFN in the context of mycobacterial infections in vivo as well as definition of the functional impact of these cells await clarification.

#### Listeria monocytogenes

Type I IFN not only inhibits antibacterial signaling pathways and promotes infection in the case of mycobacteria. Also, L. monocytogenes has evolved mechanisms to activate the type I IFN pathway for the benefit of this intracellular pathogen. Mice deficient in IFNAR signaling are more resistant to systemic L. monocytogenes infection as compared to WT controls. Mechanistically, type I IFNs enhance susceptibility to systemic Listeria infection by reducing responsiveness to IFNγ, decreasing the number of pro-inflammatory myeloid cells, promoting the expression of proapoptotic genes, and enhancing T cell sensitivity to apoptosis (148, 154–156). Of note, in intragastric or foodborne infection with L. monocytogenes type I IFN receptor mediated signaling contributed positively to survival of infected mice or did not have an impact at all, respectively (157, 158). This emphasizes again, that the route of infection contributes significantly to differences in the impact of type I IFN in infection.

Four distinct cell types have been reported as sources for type I IFN production during systemic L. monocytogenes infection in vivo (**Table 3**). For one, a FACS-purified splenic cell population from infected mice that displays surface antigens typical of macrophages and not pDCs was identified as the main producer of type I IFN (159). Also, the apathogenic Listeria mutant lacking listeriolysin O which is unable to escape from the phagolysosome into the cytoplasm of the infected cell, does not stimulate IFNβ synthesis (164, 165). Later, Tip-DCs, an effector subtype of Mac-3 hi inflammatory monocytes, which produce TNF and iNOS were identified as the major IFNβ-producing cells in vivo in systemic L. monocytogenes infections using IFNβ mob/mob and IFN-β <sup>+</sup>/1β−luc reporter mice (160, 161). IFNβ-producing TiP-DCs harbored high bacterial loads and were located within the foci of infection in the splenic white pulp ideally positioned to activate T cells as well as NK cells via type I IFN (160). Bacterial loads in the spleen were severely increased in mice deficient in CCR2 and thus lacking TiP-DCs (162). Thus, this subtype of inflammatory monocytes has been attributed an important role in early containment of L. monocytogenes infection (162, 166). The overall role of TiP-DCs in this infection may therefore be ambiguous, having a regulatory function in controlling the balance between containment of infection and at the same time mediating detrimental effects of type I IFN on the host. Interestingly, in the spleens of Listeria-infected CCR2−/<sup>−</sup> mice increased levels of type I IFN were observed indicating that alternative cell types produce type I IFN in the absence of TiP-DCs which are triggered additionally by increased bacterial load. Along this line, a detrimental role for pDCs in controlling L. monocytogenes infection was demonstrated in Siglechdtr/dtr mice where ablation of pDCs caused significantly increased survival and decreased bacterial burden at day 3 p.i., while type I IFN levels themselves were not analyzed under these conditions (66). At earlier timepoints, however, anti-PDCA-1 mediated depletion of pDCs did not lead to a difference in bacterial load or levels of type I IFN in the spleen as compared to control animals (161). In one report, CD317<sup>+</sup> SiglecH<sup>−</sup> CD19<sup>+</sup> B cells have been found to be able to induce IFNα after stimulation with heatkilled L. monocytogenes (163). Ex vivo isolated CD317<sup>+</sup> SiglecH<sup>−</sup> CD19<sup>+</sup> B cells activated cytotoxic function of NK cells in an IFNα-dependent manner. In vivo, this B cell subset contributed positively to resistance to L. monocytogenes infection as Btk−/<sup>−</sup> mice deficient for B-cells and unable to generate CD317<sup>+</sup> CD19<sup>+</sup> B cells displayed increased susceptibility to L. monocytogenes infection, while adoptive transfer of CD317<sup>+</sup> CD19<sup>+</sup> B cells to Btk−/<sup>−</sup> mice normalized their resistance to L. monocytogenes infection (163).

#### Extracellular Bacteria

As for intracellular bacteria also for extracellularly replicating bacterial pathogens type I IFN can either be detrimental or essential for host defense (145). Group B streptococci (GBS) are important neonatal pathogens and type I IFN receptor signaling is reported to contribute to host resistance against this pathogen (167). Mice i.p. infected with GBS express elevated levels of IFNβ and IFNα4 mRNA in the spleen. In vitro, GBS activated type I IFN expression in peritoneal macrophages, BMderived cDCs and to a lesser extent also in macrophages, while pDCs were completely unable to produce type I IFN after GBS stimulation (167, 168).

In contrast to GBS, type I IFN induction in the mixed bacterial sepsis model of colon ascendens stent peritonitis (CASP) has been shown to have a detrimental effect on the host. Septic peritonitis induced in IFNAR−/<sup>−</sup> mice showed improved survival and bacterial clearance as compared to WT controls. Splenic CD11b<sup>+</sup> CD11c<sup>−</sup> macrophage-like cells could be identified as major producers of IFNβ ex vivo by TABLE 3 | Cellular sources of type I IFN production in Listeria monocytogenes infection in vivo.


RT-PCR analyses from sorted cells, while no IFNα subtypes were detected (169).

In summary, type I IFN production has a detrimental effect for the host after infection with intracellular bacteria such as mycobacteria and L. monocytogenes. BMDCs and PBMC-derived DCs and macrophages are the responsible cell types for type I IFN production during mycobacteria infection. For L. monocytogenes, four cell types have been identified as type I IFN producers, namely macrophages, Tip-DCs, inflammatory monocytes, and B cells. In the case of extracellular bacteria, the cell types identified as type I IFN producers include macrophages and BMDCs. However, with the exception of the intracellular model organism L. monocytogenes, the knowledge on the cellular source of type I IFN in bacterial infection is rather scarce.

# FUNGAL INFECTIONS

As for bacterial infections, the effect of type I IFN in mouse models for infections with pathogenic fungi has been reported as beneficial or detrimental for the host depending on the fungal species and the route of infection. Additionally, controversial results obtained from very similar infection settings have been explained by the possible impact of differences in the microbiome in the respective mouse colonies (1, 11). The cell type responsible for type I IFN production in fungal infections in vivo, however, awaits clarification. To our knowledge only for the important opportunistic fungal pathogen Aspergillus fumigatus in vivo studies in this direction have been undertaken. The type I IFN response triggered by A. fumigatus was analyzed initially in human pDCs isolated from PBMCs. When these cells were stimulated in vitro with A. fumigatus hyphae IFNα was detected in the supernatant (170). IFNAR−/<sup>−</sup> mice or mice depleted of pDCs by anti-CD317 treatment exhibited an increased susceptibility to pulmonary or i.v. infection with A. fumigatus

conidia. A direct impact of pDC depletion on type I IFN levels in vivo after infection, however, has not been analyzed in this study (170). Therefore, the hypothesis that pDCs mediate their protective function in this fungal infection directly via type I IFN remains to be tested.

#### INFECTIONS WITH PROTOZOAN PARASITES

Infection with a wide variety of protozoan parasites can trigger type I IFN expression in mammalian hosts as reviewed recently (171, 172). For Plasmodium, Leishmania, and Trypanosoma in vivo infection models several studies have been carried out in the last few years which allowed the identification of cellular sources of type I IFN in response to intracellular parasite infections. This will be the focus of the following chapter and summarized in **Table 4**.

#### Plasmodium

Malaria is an important parasitic disease predominantly in tropical and subtropical African regions. It is caused by the protozoan parasite Plasmodium with P. falciparum being responsible for its most severe forms. In humans, malarial parasites are transmitted at sporozoite-stage by infected mosquitoes (182). The transmitted sporozoites rapidly travel to the liver, where they infect hepatocytes and initiate clinically silent but immunologically active liver-stage infection (171). Well-established in vivo mouse models include the lethal Plasmodium yoelii YM and P. berghei ANKA leading to high parasitemia and cerebral malaria (CM), respectively, after inoculation with Plasmodium-infected erythrocytes. Further, P. chabaudi is used as a chronic infection model. Various cellular sources for type I IFNs have been proposed after Plasmodium infection in vitro (182–184).

After inoculation with P. berghei ANKA infected erythrocytes in vivo, isolated splenic pDCs as well as CD8<sup>−</sup> cDCs expressed type I IFN (173, 174). Using anti-CD317 mediated pDC depletion and cDC depletion in CD11c-DTR-tg mice it was shown that cDCs but not pDCs are required for the induction of CM (173). Additionally, cDCs require IFNAR dependent signaling for systemic IFNα production in this model as indicated by substantially lower levels of serum IFNα in CD11c-Cre Ifnar1fl/fl mice, compared to those in infected Ifnar1fl/fl littermate controls (174).

In contrast to the P. berghei ANKA model, P. chabaudi infection did not induce IFNα in splenic cDCs but rather in pDCs via the TLR9 sensing pathway (175). However, pDCs were not essential for parasite clearance in P. chabaudi infection (175). Direct in vivo analysis performed in IFNβ mob/mob reporter mice (59) revealed that in P. chabaudi infection about 75% of IFN producing cells are pDCs (176). In addition to pDCs, splenic red pulp macrophages (RPMs) can generate significant quantities of IFNβ in response to P. chabaudi infection. Contribution of both cell types to the type I IFN response in this system was defined by pDC depletion via anti-CD317 treatment and in RPM deficient SpiC−/<sup>−</sup> mice (176).

In the lethal malaria mouse model of P. yoelii YM infection, type I IFN enhances inflammatory blood leukocyte activation and lethal outcome (177). IFNβ mob/mob reporter mice indicated here that type I IFN is produced in high amounts by BM and blood pDCs and to lesser extent by tissue resident pDCs (177). Depletion of pDCs by anti-CD317 or using pDC specific CLEC4A-DTR-tg mice confirmed pDCs as the major cellular source of type I IFN in this severe malaria model (177, 178). However, depletion of pDCs also resulted in a slight but significant increase of parasitemia (178). Further, priming of pDCs by plasmodium activated CD169<sup>+</sup> macrophages was essential (177). It was proposed that in in vivo settings the low levels of secreted type I IFN produced by monocytes and macrophages prime pDCs for systemic production of type I IFN in malaria.

From data available so far, pDCs as well as cDCs and macrophage subtypes are the cell types responsible for the generation of the type I IFN response, depending on the Plasmodium species. Similar to LCMV, Mycobacteria, or Listeria infections (153, 156, 185), it is thought that early robust production of type I IFN in the first 24 h is essential to induce protective innate and adaptive immunity against Plasmodium, while late production of type I IFN impairs host anti-malaria immune responses by induction of negative immune regulators such as PD-L1 and IL-10 (178).

#### Leishmania

Leishmania spp. are transmitted to mammalian organisms by the bite of infected sand flies (186). The parasites preferentially infect macrophages, but can also be found in other cell types, such as fibroblasts, neutrophils, and DCs (172). Depending on the parasite species and strain Leishmania causes a mild to severe cutaneous, mucocutaneous or visceral leishmaniasis (171, 187). Increased production of type I IFN has been observed in local tissues and in the draining lymph nodes of L. major infected mice (187, 188). There are diverging reports about the role of type I IFN production in the control of parasite burden and development of disease pathology. Depending on the time course of infection and type I IFN induction it can exert detrimental or protective effects for the host in Leishmaniasis (189). Most of the studies addressing the cellular source of type I IFN in Leishmania infection were performed in vitro. For example, infection of murine macrophages with L. major or L. amazonensis lead to type I IFN production (188, 190). In vitro exposure of BM-derived as well as splenic pDCs to L. major, L. infantum, or L. braziliensis promastigotes induces release of IFNα and IFNβ in a TLR9 dependent manner (191). Intriguingly, the amounts of type I IFN produced in response to Leishmania spp. are comparable to the type I IFN levels produced in response to stimulation with CpG ODNs in these experiments (191). Recently in vitro exposure to the parasite L. donovani was reported to trigger IFNβ production in splenic B cells. Here, also high levels of type I IFN mRNA were detected in splenic B cells purified from in vivo L. donovani infected mice (179). Taken together, depending on the Leishmania subtypes, pDCs and B cells are the source of type I IFN when the cells are directly exposed to the pathogen in vitro. Information on the type I IFN producers in vivo remain


TABLE 4 | Cellular sources of type I IFN production in intracellular parasite infections in vivo.

scarce so far for this important protozoan parasite model but could be increased significantly making use of the now available mouse models.

#### Toxoplasma

Toxoplasma gondii is an intracellular protozoan parasite that has infected at least 50% of the human population. It causes severe toxoplasmosis in immune-suppressed patients. T. gondii can infect a wide range of warm-blooded animals, is able to invade any nucleated cell but survives outside of the mammalian host as well (171, 172). The gut epithelium is a strategic barrier to prevent or limit parasite dissemination upon oral infection with T. gondii. In the initial phase of oral T. gondii infection elevated IFNβ mRNA levels were observed in the small intestine. Intestinal epithelial cells (IECs) and cells from the lamina propria are the source of local IFNβ production in early infection as assessed by real-time PCR performed on cells isolated from infected mice (180). In in vitro infection, T. gondii has been reported to induce or suppress type I IFN induction depending on the host species, the cell type, and the parasite strain analyzed. One publication showed that BMderived murine pDCs produce IFNα after infection with T. gondii (192). Murine pDCs recognized T. gondii profilin via TLR11 and TLR12 and produce type I IFN in a MyD88 dependent fashion (192, 193). In contrast to murine pDCs, human pDCs lack TLR11 and TLR12 and are unable to produce type I IFN despite of direct infection with T. gondii. Active infection with T. gondii in vitro rather functionally inactivates human pDCs (194). In particular macrophages and DCs serve as reservoirs of T. gondii infection and facilitate early dissemination (195). Most of the Toxoplasma strains tested are unable to induce type I IFN production in murine BM-derived macrophages after in vitro infection (196, 197). T. gondii mediated suppression of type I IFN expression has been reported also for monocytes, macrophages, and several DC subsets in vitro (181, 195, 196). On the other hand, few atypical Toxoplasma strains such as COUGAR and RUB can induce IFNβ production in murine BMderived macrophages as well as in human skin fibroblasts in in vitro infection systems (196). In a physiological oral infection mouse model ex vivo isolated inflammatory monocytes in the gut-draining mesenteric lymph nodes were identified the major producers of IFNβ. The expression of IFNβ by inflammatory monocytes required phagocytic uptake of T. gondii, while active invasion did not trigger IFNβ induction (181). Thus, depending on the host species, the cell type, and the parasite strain, T. gondii may induce or suppress type I IFN production. Epithelial, skin fibroblasts, pDCs, macrophages and inflammatory monocytes here are known cellular sources of type I IFN. In T. gondii infection most of the knowledge about the cellular sources of type I IFN is deduced from in vitro experiments. Analysis of type I IFN reporter mouse models with and without ablation of different cell types is missing, so far.

Taken together, the major type I IFN producing cell types and their contribution to immunity against many protozoan parasites remain to be defined. To our knowledge, no direct study to elucidate the cellular sources of type I IFN in multicellular parasite such as helminth infections in vivo has been published. In order to understand the cellular sources of type I IFN and their relevance with regard to disease elimination in multicellular parasites such as helminths, type I IFN reporter mouse models and cell specific depletion models remain to be analyzed. As for the other pathogen types reviewed above, parasite numbers and the site of infection might influence the sensing pathway and cell type activated to produce type I IFN.

#### CONCLUDING REMARKS AND FUTURE PERSPECTIVES

In recent years the generation of novel animal models has remarkably advanced our understanding of the mode of action of IFNs and the cell type responsible for its production in the context of an infection. The existing knowledge does not allow to depict any cell type as a single cell type responsible for the entire type I IFN production in the course of any infection. Rather, depending on the type of infection a wide variety of cells have exhibited the capacity to produce type I IFN. Decisive factors for the type of cell initiating type I IFN production are the type and amount of pathogen and the site and stage of the infection. Additionally, the genetic background of the mouse model and its microbiome status contribute as well and need to be further analyzed.

Even though pDCs are more specialized than other cell types in type I IFN production, it is getting increasingly clear that in vivo their contribution to antiviral immunity and also to immune responses to bacterial, fungal, and parasitic infection exhibits restricted patterns in time of induction and duration. The importance of pDCs as the source of type I IFN early in virus infections does not hold true at later timepoints when other host cells take over as dominant producers of type I IFNs. The impact of pDCs also depends on the route of infection. While pDCs provide an important source of type I IFN in systemic infections, their requirement for I IFNmediated antiviral immune responses in local tissues seems to be necessary only if other lines of defense are broken. However, there are exceptions to the rule as shown for local infections with MHV and HSV-2 where pDC-derived type I IFN in mice is critical for viral control and survival. Indeed, the limitation of pDC responses is caused by an upregulation of pro-apoptotic molecules and apoptosis induction in pDCs in a type I IFN-dependent manner during systemic viral infections (198). This has been suggested as a mechanism to prevent immunopathology due to sustained pDC-mediated type I IFN production. Besides pDCs, mainly macrophages, inflammatory monocytes and cDCs are able to mount significant anti-infectious type I IFN responses in vivo. Instead of a single specialized cell type, it is rather the orchestrated type I IFN expression by multiple cellular sources that ensures protective anti-infectious immune responses mediated by type I IFN. To elucidate synergisms and redundancies between the different type I IFN producing cells will be a topic of future studies.

#### REFERENCES


Advanced single cell functional profiling and systems biology approaches will contribute significantly in the near future to identify the exact functions of specific cell types, even cell subtypes, in the different stages of an infection. The spatiotemporal interaction of the type I IFN producing cell with the pathogen and the immune cells that are activated by type I IFN could help to better dissect the diverse functions of type I IFN in the immune response at different stages of infection. Importantly, due to the severe side effects of type I IFN treatment, there is a dire need to better control its activity and thereby increase its beneficial net effect. Strategies such as modifying the affinity of type I IFNs or modulating its time of availability have been reviewed recently (199). These new approaches to develop and improve vaccination strategies and to define novel therapeutic leads for infectious diseases are urgently called for in a time where antibiotic resistances are projected to increase rapidly.

#### AUTHOR CONTRIBUTIONS

SA, RM-N, AS, LR, JA, and SS wrote the manuscript. SA, RM-N, and SS designed and generated tables. RM-N designed and generated the figure. All authors read and approved the final manuscript.

#### FUNDING

This work was supported by the Deutsche Forschungsgemeinschaft (RTG 2158 Natural products and natural product analogs against therapy-resistant tumors and microorganisms: new lead structures and modes of action), the Manchot Graduate School Molecules of Infection II and III, the Research Commission of the Medical Faculty of the University of Düsseldorf, Germany (30/2016) (to SS) and by the Deutsche Forschungsgemeinschaft (FOR 2107, AL 1145/5–2), the IZKF (Alf3/018/16), the DFG EXC 1003, Grant FF-2014-01 Cells in Motion–Cluster of Excellence, Münster, Germany, and the Alzheimer Forschung Initiative e.V. (14835) (to JA).

factors. Immunity. (2006) 25:349–60. doi: 10.1016/j.immuni.2006. 08.009


produce large amounts of type I interferon. Nat Med. (1999) 5:919–23. doi: 10.1038/11360


promotes the initial type I interferon response to cytomegalovirus. Cell Host Microbe. (2008) 3:67–76. doi: 10.1016/j.chom.2007.12.008


persistent LCMV infection. Science. (2013) 340:202–7. doi: 10.1126/science. 1235208


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Ali, Mann-Nüttel, Schulze, Richter, Alferink and Scheu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Langerin+CD8<sup>+</sup> Dendritic Cells in the Splenic Marginal Zone: Not So Marginal After All

#### Ronald A. Backer\*, Nathalie Diener and Björn E. Clausen\*

Paul Klein Center for Immune Intervention, Institute for Molecular Medicine, University Medical Center of the Johannes Gutenberg-University, Mainz, Germany

Dendritic cells (DC) fulfill an essential sentinel function within the immune system, acting at the interface of innate and adaptive immunity. The DC family, both in mouse and man, shows high functional heterogeneity in order to orchestrate immune responses toward the immense variety of pathogens and other immunological threats. In this review, we focus on the Langerin+CD8<sup>+</sup> DC subpopulation in the spleen. Langerin+CD8<sup>+</sup> DC exhibit a high ability to take up apoptotic/dying cells, and therefore they are essential to prime and shape CD8<sup>+</sup> T cell responses. Next to the induction of immunity toward blood-borne pathogens, i.e., viruses, these DC are important for the regulation of tolerance toward cell-associated self-antigens. The ontogeny and differentiation pathways of CD8+CD103<sup>+</sup> DC should be further explored to better understand the immunological role of these cells as a prerequisite of their therapeutic application.

#### Edited by:

Silvia Beatriz Boscardin, University of São Paulo, Brazil

#### Reviewed by:

Kristian Michael Hargadon, Hampden–Sydney College, United States Joke M. M. Den Haan, VU University Medical Center, Netherlands

#### \*Correspondence:

Ronald A. Backer r.backer@uni-mainz.de Björn E. Clausen bclausen@uni-mainz.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

> Received: 12 October 2018 Accepted: 19 March 2019 Published: 12 April 2019

#### Citation:

Backer RA, Diener N and Clausen BE (2019) Langerin+CD8<sup>+</sup> Dendritic Cells in the Splenic Marginal Zone: Not So Marginal After All. Front. Immunol. 10:741. doi: 10.3389/fimmu.2019.00741 Keywords: Conventional dendritic cells, cross-presentation, dendritic cell subsets, immunotherapy, macrophages, marginal zone, plasmacytoid dendritic cells, spleen

#### INTRODUCTION

Dendritic cells (DC) link pathogen sensing and activation of innate immunity to the initiation of (primary) adaptive immune responses. For the latter, DC function as professional antigen (Ag)-presenting phagocytes that orchestrate the priming and polarization of naïve T cells. Importantly, next to stimulating protective immunity following infection, cancer or vaccination, DC are also crucial for the maintenance of immunological (self-) tolerance.

In the steady state, the murine DC family encompasses several cell populations that are very heterogeneous in development, phenotype and differ in their immune-regulatory functions. This variety among DC that have evolved at distinct immunological sites, allows immune responses to be specifically tailored to a given pathogenic threat (1, 2). In general, DC can be categorized into two classes (**Figure 1A**). The first class consists of the natural type I interferonproducing plasmacytoid DC (pDC: CD11cintCD45RA+Ly6C<sup>+</sup> cells). These pDC are poor in Agpresentation but play a crucial role as first-line defense against viral infections and are involved in anti-tumor responses as well (3). The second class of DC comprises conventional (classical) DC (cDC), which are characterized by the expression of high levels of CD11c and MHC class II (MHCII). These cDC can be further separated into functionally specialized cDC1 and cDC2 populations, initially according to their phenotype, and later through their molecular signatures, ontogeny and unique transcription factor dependency (4) (**Table 1**). These cDC1 and cDC2 populations are defined across different organs, and display distinct responses to pathogen- and danger-associated signals and, subsequently, specialized capacities to interact with T cells (5, 6).

**426**

Understanding DC biology becomes even more complex as both the cDC1 and cDC2 populations can be further divided based on their localization and migratory abilities into (i) peripheral tissue (migratory) DC and (ii) lymphoid organresident DC subpopulations. Whereas, resident DC do not leave the lymph nodes (LN), spleen or thymus, migratory DC are the prototypic DC described by the Langerhans paradigm (7, 8). These migratory DC strategically line the barrier organs toward the external environment (e.g., skin and mucosa), and sample the tissues for invading pathogens (including commensals) and incoming immunogenic particles. Upon Ag encounter, together with pro-inflammatory stimuli, these DC move from the tissues into the T cell areas of local LN where they initiate protective T cell responses (9).

In this review, we will first recapitulate cDC heterogeneity in the spleen, and then zoom in on one particular splenic DC subset, namely, Langerin+CD8<sup>+</sup> cDC1. In particular, we will summarize recent highlights in the biology of this DC subset, discuss its functional specialization in mice, touch upon the human equivalents and finally conclude by discussing potential concepts to harness these Langerin+CD8<sup>+</sup> cDC1 to develop improved therapeutic and / or vaccination strategies.

#### HETEROGENEITY OF SPLENIC cDC SUBSETS

The spleen is the largest secondary lymphoid organ of the body and is functionally linked to the systemic blood circulation (**Figure 2A**). Histologically, the spleen consists of red pulp (RP) and white pulp (WP). The RP is a loose venous sinusoidal meshwork involved in blood filtration, while the WP contains T cell-rich periarteriolar lymphoid sheaths (PALS) and discrete B cell follicles. Thereby, the WP resembles the lymphoid structures found in LN and is thus essential for the induction of adaptive immune responses. A specialized environment called the marginal zone (MZ) is uniquely situated at the transition site between the scavenging RP and the lymphoid WP. As the arterial bloodstream opens into the marginal sinuses, most of the blood entering the spleen passes the MZ (**Figure 2A**). The splenic MZ, therefore, is together with the RP involved in the filtration of the blood and constitutes the prime site for the detection of blood-borne Ag (10).

Ag larger than 75 kDa are trapped and cleared by a large number of specialized MZ-resident phagocytic cells, including marginal zone macrophages (MZM), marginal metallophilic macrophages (MMM) and marginal zone B cells (MZB), thereby initiating immune responses against systemic pathogens (10– 13) (**Figure 2B**). Moreover, the MZ is of vital importance for the clearance of apoptotic cells and the subsequent induction of self-tolerance, which can be abrogated by the depletion of macrophages (Mφ) in the MZ (14, 15).

The splenic DC compartment only consists of resident DC as the spleen is not connected to the afferent lymphatic system by which migratory DC traffic from the peripheral tissues to LN. Historically, splenic cDC were defined based on the reciprocal expression of CD4 and CD11b or the CD8αα homodimer into at least three distinct DC subsets: (i) a CD8αα-expressing CD8+CD11b<sup>−</sup> cDC1 subset, and a CD11b<sup>+</sup> cDC2 subpopulation that can be further divided into (ii) CD4+CD8<sup>−</sup> DC and (iii) CD4−CD8<sup>−</sup> double-negative DC subsets. To date, unsupervised phenotypic analysis, for example using (single cell) RNA sequencing and high-dimensional flow cytometry or mass cytometry, has added a large number of additional subpopulation-specific markers, confirming the existence of heterogeneity (DC subsets) within both cDC1 and cDC2 subpopulations (16). All of these phenotypically distinct cDC subsets may exert specialized roles in, respectively, promoting and suppressing different facets of immunity (**Table 1**).

#### Splenic cDC1

Analysis by flow cytometry indicated that the majority of splenic CD8<sup>+</sup> cDC1 co-express the C-type lectin receptors DEC205 (CD205) and Langerin (CD207) (**Figure 1B**). Initially, staining spleen sections for DEC205 localized CD8<sup>+</sup> cDC1 in the PALS only (11, 15, 17–20), resulting in the dogma that CD8<sup>+</sup> cDC1 were restricted to the WP (17, 19, 21–23). In contrast, Langerin was predominantly detected in the MZ and only in limited amounts in the RP and the PALS by histology (24–28). This discrepancy in (co-) localization of Langerin and DEC205 between methods may be due to DEC205 levels too low to be detected by histology, resulting in variable DEC205 expression on slides. Therefore, it is now generally accepted that in the steady state CD8<sup>+</sup> cDC1 are mainly located in the MZ and RP, and that they are not limited to the WP (28–30) (**Figure 2B**).

CD8<sup>+</sup> cDC1 are characterized by a high ability to cross-present cell-associated and soluble Ag (31–36), and predominantly induce TH1-type helper T cell responses (36–38), as well as regulatory T cells (TREG) via TGFβ (**Figure 2C**). Moreover, CD8<sup>+</sup> cDC1 can activate and polarize invariant natural killer T (iNKT) cells via CD1d presentation of glycolipid Ag (39).

Although multiple reports revealed considerable heterogeneity within this subpopulation, functional features (e.g., cross-presentation) are, nevertheless, mainly attributed to the cDC1 subpopulation as a whole. However, differential expression of DEC205 and CX3CR1, for example, is believed to divide the CD8<sup>+</sup> DC subpopulation into subsets that have distinct functions in pathogen-recognition and immune-modulation (40, 41) (**Figure 1B**). Although the origin of CX3CR1+CD8<sup>+</sup> DC is not clear yet, these cells seem to lack many functional hallmarks of classical CD8<sup>+</sup> cDC1, including cross-presentation and IL-12 secretion in response to microbial challenge. In addition, CX3CR1-expressing DC rearranged immunoglobulin genes and are thought to rather resemble pDC and to be closely related to CD8<sup>−</sup> DC (41), and therefore might not be considered as cDC1. Another chemokine receptor highly expressed on splenic CD8<sup>+</sup> cDC1 is XCR1 (42), which potentially allows close interaction with activated T cells and NK cells. Surprisingly however, Diphtheria-toxin (DT) treatment of XCR1-DTR knock-in mice did not result in complete depletion, indicating that splenic CD8<sup>+</sup> cDC1 include a distinct population that is not eliminated due to heterogeneous XCR1 expression (43). Also in the absence of functional Notch2 signaling the number of CD8<sup>+</sup> DC is diminished, suggesting that at least a subset of splenic CD8<sup>+</sup> cDC1 also depend on Notch2 (44).

Taken together, these observations indicate that several distinct resident CD8<sup>+</sup> cDC1 subsets are present in the spleen, but that the potential functional heterogeneity within this cDC1 subpopulation is currently underappreciated, and that several cDC1-specific functions might turn out to be rather CD8<sup>+</sup> subset-restricted characteristics.

#### Splenic cDC2

CD11b+CD8<sup>−</sup> cDC2 are the most abundant cDC in the lymphoid organs. In contrast to CD8<sup>+</sup> cDC1, this cDC2 subpopulation is known to be heterogeneous, but less well defined in function. In general, CD8<sup>−</sup> cDC2 (also characterized by the specific expression of the C-type lectin receptor DCIR2) are preferentially involved in MHCII-restricted Ag presentation and TH2 priming (31, 45), although they also have the ability to cross-present exogenous Ag under certain circumstances (34, 46, 47).

The heterogeneous cDC2 population can be further subdivided according to the differential expression of CD4 and the endothelial cell-selective adhesion molecule (ESAM), although this does not result in clearly defined homogenous populations (48), which makes it difficult to determine individual immune-modulatory capacities. Due to their similarities in phenotype and gene expression profiles, both CD4+CD8<sup>−</sup> cDC2 (which largely co-express ESAM) and CD4−CD8<sup>−</sup> doublenegative cDC2 are often collectively referred to as CD8<sup>−</sup> cDC2 (19, 49–52), however, according to recent studies these two subsets appear different (44, 53). For example, ESAMlow cDC2 TABLE 1 | Steady state cDC subset characteristics in mouse and human.


++, +, +/−, and −, represent very high to low to absent expression; #, inducible expression; BC, bridging channel; MZ, marginal zone; n.d., not determined in detail; PALS, Periarteriolar Lymphoid Sheaths.

exhibit a more myeloid signature with Csf-1R, Csf3R, CCR2 and Lysozyme expression, suggesting that they are related to monocytes rather than to cDC. As migratory ESAM−CD11b<sup>+</sup> tissue cDC2 can arise from both bone marrow (BM)-DC progenitors and monocytes, it is still under debate whether these splenic ESAMlow cDC originate from circulating monocytes or not (44, 54). Most likely they arise from early progenitors such as Macrophage and Dendritic Cell Precursors (MDP) without the contribution of the Common Dendritic Cell Precursor (CDP) (44, 55).

CD8<sup>−</sup> cDC2 reside in the MZ and bridging channels of the spleen (18, 31, 56), which are interruptions in the MZ where the PALS is in contact with the RP allowing T cell entry into the WP (57, 58) (**Figure 2B**). The development of CD8<sup>−</sup> cDC2 (and more specifically, of the ESAM<sup>+</sup> CD8<sup>−</sup> cDC2 subset) depends on Notch2 (59). Furthermore, the G protein-coupled receptor EBI2 determines the specific MZ positioning, thereby allowing signaling and crosstalk with MZ B cells and other cells via LTßR and SIRPα, which is essential for the homeostasis of CD8<sup>−</sup> cDC2 (18, 56, 60). In addition, Runx3 is required for the specification and homeostasis of CD8<sup>−</sup> cDC2, as ablation of Runx3 expression resulted in a substantial decrease of CD8<sup>−</sup> cDC2 numbers in the spleen (55).

FIGURE 2 | Structure and cellular composition of the murine spleen. (A) The spleen consists of red pulp (RP) and white pulp (WP). Blood enters the spleen via the splenic artery, which is subsequently branching into the trabecular arteries and central arteries. Finally, small arterioles and capillaries end up in the RP. The RP is a venous sinusoidal system containing connective tissue, sinuses and venules. Here, blood can leave the open ends of splenic RP capillaries, allowing free percolation into the RP and subsequent re-collected into the sinuses for venous drainage. In mice, the WP is composed of B cell follicles and T cell areas (the periarterial lymphatic sheaths, PALS) surrounding a central arteriole. The marginal zone (MZ) separates the WP from the RP. As marginal sinuses are opening in the MZ, most of the arterial blood that enters the spleen is running through the MZ. Furthermore, re-circulating lymphocytes can leave the blood in the MZ. (B) At least 2 types of macrophages are present in the MZ. Marginal metallophilic macrophages (MMM) are located as a tight network in the inner part of the MZ near the WP. Marginal zone macrophages (MZM) can be found in the outer MZ facing the RP. Scattered between these MZM are marginal zone B cells (MZ B cells) and Langerin+CD8<sup>+</sup> cDC1, whereas cDC2 are mainly located in so called bridging channels, which are interruptions in the MZ sinus and macrophage rims. Some Langerin+CD8<sup>+</sup> cDC1 are also present in the RP and WP T cell areas. In the RP, red pulp macrophages (RP Mφ) can be identified. (C) Langerin+CD8<sup>+</sup> cDC1 are involved in the direct uptake, processing and cross-presentation of blood-borne antigens (Ag). Upon Ag encounter ①, Langerin+CD8<sup>+</sup> cDC1 migrate out of the MZ into the WP T cell areas ② to prime Ag-specific T cell responses ③. Depending on the type of Ag, this results in CD8<sup>+</sup> T cell activation, or in CD8<sup>+</sup> T cell tolerance ④. Moreover, Langerin+CD8<sup>+</sup> cDC1 are able to acquire Ag from other cells (e.g., potentially from MMM), via a process called Ag-transfer ⑤.

# SPLENIC LANGERIN+CD8<sup>+</sup> cDC1

Expression of the endocytic receptor Langerin is a classical hallmark of Langerhans cells (LC) in the epidermis and skindraining lymph nodes (61, 62). However, Langerin expression is not restricted to LC as also other skin DC subsets (i.e. dermal CD103<sup>+</sup> DC) express Langerin and are functionally distinct (63– 66). In addition, Langerin<sup>+</sup> DC can be found as interdigitating cells in the T cell zones of LN, as well as in the gut and the lung (25, 62, 67–70). Among splenic cDC, Langerin expression is mainly found on CD8<sup>+</sup> cDC1, though its expression is lower than on LC and primarily intracellular in location (71). Although percentages vary depending on the genetic background of the experimental mice, the Langerin+CD8<sup>+</sup> DC subset constitutes the majority of CD8<sup>+</sup> cDC1 in the spleen and DT-mediated ablation in Lang-DTREGFP mice can reach about 70% of the splenic CD8<sup>+</sup> cDC1 subpopulation (24, 72–75).

Langerin+CD8<sup>+</sup> cDC1 are primarily localized in the MZ, just internal to the F4/80<sup>+</sup> RP Mφ, interspersed with MZM and forming a ring around the CD169<sup>+</sup> MMM (**Figure 2B**). In addition, a minor fraction of Langerin+CD8<sup>+</sup> cDC1 can be found in the RP and PALS (74). Compared to their Langerin<sup>−</sup> counterparts, Langerin+CD8<sup>+</sup> cDC1 share a common morphology with a similar expression profile of the classical splenic cDC1 markers like CD8αα, CD24, CD36, DEC205, Clec9a, ICAM, and XCR1 (**Figure 1B** and **Table 1**). In addition, Langerin+CD8<sup>+</sup> cDC1 co-express high levels of the integrin CD103. In steady state, splenic cDC display a rather immature phenotype with low levels of MHCII and co-stimulatory molecules (76). In contrast to Langerin<sup>−</sup> cDC1, the baseline expression of the activation markers CD80 and CD86 are slightly higher on the Langerin+CD8<sup>+</sup> cDC1 subset (24, 72, 75), but whether this reflects a functionally more mature state remains unknown. However, steady state levels of serum IL-12 were significantly decreased in mice depleted of Langerin+CD8<sup>+</sup> cDC1, indicating that these DC are responsible for basal IL-12 production (77). At least, Langerin−CD8<sup>+</sup> cDC1 are not unresponsive to inflammation as the characteristics of upregulated activation markers during Tolllike receptor (TLR) or glycolipid antigen α-galactosylceramide (α-Gal-Cer) stimulation are similar to Langerin+CD8<sup>+</sup> cDC1 (75). Conclusively, Langerin marks a proportion of CD8<sup>+</sup> cDC1 in the splenic MZ. Based on their specific localization and phenotypic characteristics, it is suggestive that Langerin+CD8<sup>+</sup> cDC1 are important regulators of immune responses toward blood-borne Ag in the steady state and during inflammation.

#### ONTOGENY AND MOLECULAR REGULATION OF LANGERIN+CD8<sup>+</sup> cDC1

To date, conclusive data are lacking to define Langerin+CD8<sup>+</sup> and Langerin−CD8<sup>+</sup> DC as distinct steady state cDC1 subsets, because alternatively, Langerin-expression could merely reflect different developmental stages within the CD8<sup>+</sup> cDC1 subset (**Figure 1B**). CD8<sup>+</sup> cDC1 express much lower levels of Langerin as compared to LC and they lack the LC-specific intracellular organelles known as Birbeck granules. In addition, the spleen does not drain the skin. Therefore, Langerin+CD8<sup>+</sup> cDC1 in the spleen are unrelated to LC and are continuously replaced by blood-borne precursors of a non-LC origin (65, 67).

In the tissue, cDC have, in general, a relatively finite halflife of about 4–6 days, with CD8<sup>+</sup> cDC1 in the spleen having even higher turnover rates (36). Indeed, DT-mediated depletion of a significant proportion of splenic CD8<sup>+</sup> cDC1 in Lang-DTREGFP knock-in mice was evident for a period of 2–3 days after which Langerin+CD8<sup>+</sup> cDC1 repopulated the spleen and reached homeostatic levels again by day 7 (72, 73, 78). This differs somewhat from Langerin+CD8<sup>−</sup> dermal cDC1, which follow a similar kinetics after DT-mediated depletion, but fail to reach full reconstitution (79). Langerin+CD8<sup>+</sup> cDC1 exhibit decreased survival as compared to Langerin−CD8<sup>+</sup> cDC1 upon in vitro activation or upon in vivo cell transfer. For example, treatment of mice with TLR-ligands or the innate invariant NKT (iNKT) cell ligand α-Gal-Cer also resulted in a fast decline of splenic Langerin+CD8<sup>+</sup> cDC1 numbers, which peaked after 15–24 h. However, Langerin−CD8<sup>+</sup> cDC1 numbers remained unchanged, suggesting that activation does not convert Langerin<sup>+</sup> into Langerin<sup>−</sup> cDC1 but that Langerin+CD8<sup>+</sup> cDC1 are rather sensitive to activation-induced cell death (24, 75, 80). Although the exact mechanisms remain elusive, in this setup TNFα may be one factor inducing cell death in Langerin+CD8<sup>+</sup> cDC1 (80).

cDC are of myeloid origin and develop from hematopoietic stem cells (HSC) in the BM, but the exact developmental pathways of different cDC lineages remain controversial and difficult to elucidate (1, 5, 50, 54, 81–86) (**Figure 1**). It is generally accepted that all cDC precursors share a common differentiation pathway depending on the transcription factors PU.1 and Zbtb46 until they become committed common DC precursors (CDP) (87, 88). Subsequently, the developmental pathway of pDC and cDC diverges as CPD have the ability to differentiate into either cells of the pDC lineage or cDC precursors (pre-DC). Pre-DC, that are dependent on FLT3-L (89, 90), migrate into peripheral tissues to further mature into either cDC1 or cDC2 driven by specific transcription factors and cytokine combinations (1, 91, 92). Further differentiation of cDC1 is strictly guided by the hierarchical expression of Irf8, Id2, and Batf3, as targeted deletion of these transcription factors in mice leads to severe developmental defects in cDC1 causing a marked decrease of splenic CD8<sup>+</sup> DC numbers (34, 93–95).

So far, no Langerin+CD8<sup>+</sup> cDC1-specific transcriptional program has been identified. Moreover, both Langerin+CD8<sup>+</sup> and Langerin−CD8<sup>+</sup> cDC1 express similar levels of the cDC1 associated transcription factors Irf8, Id2, Nfil3, and Batf3 (75), indicating that the two subsets share a similar ontogeny and thus do not arise from distinct developmental pathways. Indeed, the complete lack of CD8<sup>+</sup> cDC1 in Irf8-deficient mice suggests that Irf8 is critically involved in Langerin+CD8<sup>+</sup> cDC1 development (96–99). Interestingly, infection with intracellular bacteria (e.g., Mycobaterium tuberculosis, Listeria and Toxoplasma) could functionally restore the CX3CR1−Langerin−CD8<sup>+</sup> cDC1 compartment in Batf3 KO mice, while these mice still lack the majority of splenic Langerin+CD8<sup>+</sup> cDC1 (100, 101). Purified Langerin−CD8<sup>+</sup> cDC1 started to express Langerin upon transfer into naïve mice, with up to 60–70% of cells stably expressing the Langerin receptor 40 h post transfer (75). Moreover, gain of Langerin expression by Langerin−CD8<sup>+</sup> cDC1 has been associated with their differentiation into a more mature cDC population with increased capacity to phagocytose dead cells, secrete IL-12 and cross-prime CD8<sup>+</sup> T cells (75). The cytokine granulocyte-macrophage CSF (GM-CSF) enhances the differentiation of cross-presenting splenic CD103<sup>+</sup> cDC1 during bacterial infection (102), but whether GM-SCF signaling is also able to induce Langerin expression on these cDC1 is not clear yet. Conclusively, these data suggest that Langerin−CX3CR1−CD8<sup>+</sup> DC might be precursors of the (functionally mature) Langerin+CD8<sup>+</sup> subset, and that the differentiation into mature Langerin+CD8<sup>+</sup> DC depends on Batf3 and yet undefined conditions (75, 94).

#### CROSS-PRESENTATION OF CELL-ASSOCIATED AG BY LANGERIN+CD8<sup>+</sup> DC

CD8<sup>+</sup> cDC1 not only cross-present Ag under inflammatory settings, as cross-presentation by these cDC1 under steady state conditions is important for efficient induction of tolerance to self-Ag. Moreover, antigen-presenting cells (APC) in the splenic MZ exhibit a high phagocytic capacity for dying / apoptotic cells, suggesting that the MZ is essential for the initiation of immune self-tolerance (14). Indeed, experiments using intravenously injected apoptotic cells revealed that these cells initially accumulated in the MZ where they were preferentially phagocytosed by CD8<sup>+</sup> cDC1 rather than by CD8<sup>−</sup> cDC2. Notably, not all CD8<sup>+</sup> cDC1 had the ability to cross-present, as only about half of the CD8<sup>+</sup> cDC1 phagocytosed apoptotic cells, independent of the number of injected apoptotic cells (24). Phagocytic active CD8<sup>+</sup> cDC1 specifically expressed Langerin and CD103, whereas Langerin<sup>−</sup> cDC1 did not acquire apoptotic cells (24, 75). Phagocytosis of apoptotic cells by Langerin+CD8<sup>+</sup> cDC1 induced their migration from the MZ to the WP (24). This is in line with previous studies demonstrating redistribution of CD8+XCR1<sup>+</sup> cDC1 from the MZ to the center of the T cell zones, where CD8<sup>+</sup> T cells concentrate upon challenge with LPS (21, 28). Here, phagocytosis of apoptotic cells by Langerin+CD8<sup>+</sup> DC resulted in efficient cross-presentation of their cell-associated Ag, while no significant CD4<sup>+</sup> T cell priming was detected (24, 72).

The observed differences in uptake and cross-presentation between Langerin+CD8<sup>+</sup> and Langerin−CD8<sup>+</sup> cDC1 could result from a general inability of Langerin−CD8<sup>+</sup> cDC1 to phagocytose apoptotic cells due to a lack of the respective uptake receptors for apoptotic cells. For example, antibody (Ab) mediated Ag targeting to the recognition receptors DEC205 and Clec9a resulted in efficient cross-presentation and CTL priming (31, 103). Because Langerin+CD8<sup>+</sup> cDC1 highly express DEC205 and Clec9a, as well as increased levels of the supposed dead-cell receptor CD36 (24), this cDC1 subset may represent a functionally distinct population with specific phagocytic capacities for apoptotic cells as compared to Langerin−CD8<sup>+</sup> cDC1. Indeed, Langerin+CD8<sup>+</sup> cDC1 displayed an intrinsic activity for the uptake of cell-associated Ag, while Langerin+CD8<sup>+</sup> and Langerin−CD8<sup>+</sup> cDC1 did not differ in their capacity to phagocytose bacteria, beads or soluble Ag (24). Nevertheless, CD8<sup>+</sup> T cell activation in response to soluble Ag was much stronger in Langerin+CD8<sup>+</sup> cDC1 compared to Langerin−CD8<sup>+</sup> cDC1. This suggests that, although both cDC1 subsets acquired comparable amounts of Ag, Langerin+CD8<sup>+</sup> and Langerin−CD8<sup>+</sup> cDC1 subsets exhibit inherent differences in their Ag-processing machinery. To examine this in more detail, mice were challenged with exogenous cytochrome c (cyt c). As this pro-apoptotic molecule induces apoptosis when diverted into the cytoplasm, cyt c specifically depleted cells that possess cytosolic export mechanisms required for cross-presentation (104). Indeed, cyt c treatment selectively and dose-dependently ablated Langerin+CD8<sup>+</sup> cDC1 but not Langerin−CD8<sup>+</sup> DC. Moreover, depletion of splenic Langerin+CD8<sup>+</sup> cDC1 in Lang-DTR mice also abrogated CD8<sup>+</sup> T cell responses (72, 77). These data would identify Langerin+CD8<sup>+</sup> cDC1 as the main professional cross-presenting subset within the CD8<sup>+</sup> cDC1 subpopulation (24, 41, 72, 104).

Presentation of cell-associated Ag by Langerin+CD8<sup>+</sup> cDC1 is critical for the maintenance of self-tolerance. Depletion of Langerin+CD8<sup>+</sup> cDC1 prior to injection of MOG-expressing apoptotic cells and subsequent MOG/CFA immunization resulted in Ag-specific CD8<sup>+</sup> T cell hypo-responsiveness and impaired EAE-progression (24). Whether this tolerance depends on the induction of regulatory T cells (TREG) by Langerin+CD8<sup>+</sup> cDC1 remains unknown. However, DEC205 and Langerin Ag-targeting experiments revealed that, in contrast to Langerin<sup>+</sup> migratory skin cDC, Langerin+CD8<sup>+</sup> cDC1 in the spleen were inefficient in generating TREG in vivo (105). Splenic Langerin+CD8<sup>+</sup> cDC1 can also acquire, process and cross-present lymphoma-derived Ag, both in vitro and in vivo (106). In this setting, Langerin+CD8<sup>+</sup> cDC1 exhibit a tolerogenic function, indicated by decreased antitumor immunity, resulting from impaired naïve CD8<sup>+</sup> T cells priming, possibly due to the lack of DC maturation and enhanced expression of the T cell suppressive ligand PD-L2 (106). In response to phagocytosis of dead cells, Langerin+CD8<sup>+</sup> cDC1 strongly upregulate the expression of CD80 and the TNF superfamily ligand 4-1BBL (24), both T cell co-stimulatory molecules. Therefore, in combination with the appropriate stimulation such as TLR or licensing by bystander iNK T cells, Langerin-targeted Ag could stimulate immunity and thus Agspecific CTL responses (72, 103). In line, Langerin+CD8<sup>+</sup> cDC1 specifically enhanced protective immune responses when preactivated CD8<sup>+</sup> T cells were transferred as antitumor treatment strategy (106).

Notably, the spleen is, next to the liver and BM, important for the clearance of aged red blood cells (RBC), where mainly Mφ in the peripheral RP actively remove these senescent RBC (erythrophagocytosis) (107). However, damaged RBC that undergo 'programmed cell death'-like apoptosis (eryptosis) are primarily taken up by splenic MZM and Langerin<sup>+</sup> cDC (108).

In summary, the Langerin+CD8<sup>+</sup> cDC1 subset is predominantly involved in clearance and cross-presentation of circulating apoptotic cells, but not, or only minimally, in MHCII presentation and subsequent CD4<sup>+</sup> T cell priming in responses to the same Ag. Again, this does not answer whether the ability to cross-present depends on a certain maturation stage characterized by the expression of Langerin, or whether Langerin+CD8<sup>+</sup> and Langerin−CD8<sup>+</sup> cDC1 are two functionally discrete cDC1 subsets, and this should be examined.

# FUNCTION OF LANGERIN+CD8<sup>+</sup> DC DURING SYSTEMIC INFECTION

Langerin-expression identifies the cross-presenting cDC1 subset in the spleen. So far, it is not clear whether Langerin is merely a phenotypic marker for this specific CD8<sup>+</sup> cDC1 subset, or whether the receptor actually exhibits functional properties. In LC, Langerin is associated with the formation of Birbeck granules. However, these structures are absent in Langerin+CD8<sup>+</sup> cDC1, and their formation cannot be induced by stimulation with anti-Langerin monoclonal Ab (67). In general, C-type lectin receptors like Langerin function as innate pattern recognition receptors. Langerin itself recognizes cellsurface carbohydrate structures on pathogens (e.g., mannose, fucose and n-acetylglucosamine). This recognition normally results in the internalization and subsequent presentation of pathogen-associated Ag on MHC molecules (109, 110). DEC205 is found in MHCII-rich late endosomes and lysosomes (111), and Ab-mediated Ag targeting to DEC205 in the absence of adjuvant resulted in CD4<sup>+</sup> T cell priming and TREG expansion (112). However, direct comparison of Ab-mediated Ag targeting to Langerin and DEC205 indicate that, although targeting Langerin and DEC205 both resulted in comparable TH1 responses as determined by CD4<sup>+</sup> T cell IFNγ production, DEC205 targeting resulted predominantly in CD8<sup>+</sup> T cell proliferation (31, 103). In contrast, langerin-targeting resulted in efficient priming of both CD4<sup>+</sup> and CD8<sup>+</sup> T cells which was persistent for at least 14 days (113), indicating that Langerin on Langerin+CD8<sup>+</sup> cDC1 may be involved in the delivery of Ag for presentation on both MHCI and MHCII molecules (103). As DEC205 and Langerin are co-expressed by CD8<sup>+</sup> cDC1 subset, these data might suggest that both lectin receptors feed into different Agprocessing and presentation pathways. Since Langerin deficiency did not impair Ag-presentation of soluble Ag by LC (114), the expression of Langerin on Langerin+CD8<sup>+</sup> cDC1 per se may not be a prerequisite for cross-presentation.

CD8<sup>+</sup> cDC1 are specialized cross-presenting cells and the most potent producers of IL-12 under several inflammatory settings, such as CD40 stimulation (29, 36, 37, 51, 115, 116). IL-12 is a pro-inflammatory cytokine involved in NK cell responses and the differentiation of TH1 T cells (117). Langerin+CD8<sup>+</sup> cDC1 produce high levels of IL-12 upon systemic stimulation, whereas Langerin−CD8<sup>+</sup> cDC1 are poor IL-12 producers (77, 118). However, the requirement of IL-12 production by CD8<sup>+</sup> cDC1 seems to depend on the type and timing of infection. While during the first hours of infection, Langerin−CD8<sup>+</sup> cDC1 were essential for early and transient IL-12 production, later on and at least until 3 weeks post-infection, Langerin+CD8<sup>+</sup> cDC1 were the dominant source of IL-12 (77). This study also identified that the depletion of Langerin+CD8<sup>+</sup> cDC1 resulted in diminished protective immune responses against intravenous Mycobacterium bovis infection. Langerin+CD8<sup>+</sup> cDC1-depleted mice displayed increased bacterial loads, due to decreased IL-12 production in combination with delayed and diminished CD8<sup>+</sup> T cell responses (77). Interestingly, although CD8<sup>+</sup> T cell responses recovered over time, the bacterial load continued to increase and could not be controlled. This indicates that early immune priming effects by Langerin+CD8<sup>+</sup> cDC1 are essential for the fate of the immune response (77), which was also found for the negative regulatory role of LC during cutaneous Leishmania major infection (119).

Upon activation iNKT cells rapidly produce proinflammatory cytokines. Due to their immunoregulatory function, iNKT cells are implicated to play a role in infectious diseases, autoimmune diseases and cancer. iNKT cells can be activated by cDC and in turn activate cDC to produce IL-12. Although Langerin+CD8<sup>+</sup> cDC1 are not required for the initial activation of iNKT cells (80), conditioning of Langerin+CD8<sup>+</sup> cDC1 by these iNKT cells in combination with TLR stimulation synergistically enhanced cytokine secretion and sustained T cell priming capacities of Langerin+CD8<sup>+</sup> cDC1 (120).

Although the crucial role of the spleen and its CD8<sup>+</sup> cDC1 compartment for bacterial and viral clearance and for providing protective immunity is known, evidence about the specific contribution of Langerin+CD8<sup>+</sup> cDC1 in these models is so far limited. Therefore, further studies will be needed to pursue the implication of Langerin+CD8<sup>+</sup> cDC1-specific functions during systemic infections.

# LANGERIN+CD8<sup>+</sup> DC AND MACROPHAGE INTERACTIONS

In general, CD8<sup>+</sup> cDC1 obtain Ag directly from their surrounding environment. The specific localization of Langerin+CD8<sup>+</sup> cDC1 within the MZ strongly suggests that these cells are involved in efficient sampling of the blood (**Figure 2C**). However, these DC poorly phagocytose bloodborne Ag as compared to the various Mφ subsets in the MZ (74). Using polystyrene particles or bacteria, <10% of the Langerin+CD8<sup>+</sup> cDC1 were able to phagocytose these Ag as compared to the majority of MZ Mφ (24, 113). Thus, the bulk of particles from the blood is cleared by Mφ and not DC, even though the cells are in close proximity. Notably, phagocytosis assays primarily assess the level of particle scavenging or clearance, a feature of Mφ, but they do not provide information regarding the efficiency of downstream steps, such as Ag-processing and Ag-presentation by DC to T cells.

Another, although a less well appreciated mechanism of Ag acquisition is the transfer of Ag between APC (121, 122). This functional interaction would allow cDC to initiate protective T cell responses even if Ag availability and accessibility is limited. For example, Langerin+CD8<sup>+</sup> cDC1 are able to cross-present Ag from injected Ag-loaded allogeneic BM-DC and mount CD8<sup>+</sup> T cell responses without affecting CD4<sup>+</sup> T cell responses (118). Depletion of the Langerin+CD8<sup>+</sup> DC population in the Lang-DTREGFP mice abrogated this indirect Ag-presentation and thus subsequent CD8<sup>+</sup> T cell priming. These data indicate that Langerin+CD8<sup>+</sup> cDC1 were able to acquire Ag indirectly from other APC populations via Ag-transfer in the presence of a potent adjuvant. Notably, this Ag transfer is not limited to protein Ag, as the glycolipid α-Gal-Cer could also be acquired and presented by endogenous Langerin+CD8<sup>+</sup> cDC1 (118).

Moreover, Ag initially acquired by CD169<sup>+</sup> MMM, either after monoclonal Ab-mediated Ag targeting or during adenoviral infection, could specifically be presented by CD8<sup>+</sup> DC, suggesting transfer from MMM to CD8<sup>+</sup> cDC1 (123–125). Unfortunately, Langerin-expression on this cross-presenting CD8<sup>+</sup> cDC1 has not been studied, but as this Ag-transfer absolutely requires a Batf3-dependent, Clec9a<sup>+</sup> cDC1 (125), it is very suggestive that the Langerin+CD8<sup>+</sup> cDC1 subset is the prime candidate to govern this process due to its Clec9a expression, Batf3-dependency and localization in the MZ (**Figure 2C**).

# LANGERIN+CD8<sup>+</sup> cDC1: IMPLICATIONS FOR IMMUNOTHERAPY

The ultimate goal of every vaccination strategy to treat chronic infections and cancer is the induction of durable and protective T cell responses. For this purpose, proteins are very useful, were it not that they are poorly immunogenic. Notably, the immunogenicity of proteins can be immensely enhanced via targeting to cDC. This Ag-targeting, in combination with appropriate DC maturation signals like αCD40, PolyIC or conditioning by activated NKT cell strongly boost Ag-specific T cell responses. Therefore, identification and functional characterization of cDC (subsets) to reinforce vaccination efficiency is of great interest. Currently, many protein-targeting strategies utilize DEC205 and Clec9a receptors, but this will result in targeting of additional cell types as their expression is not cDC restricted (103). In contrast, murine Langerin expression is confined to the CD8<sup>+</sup> cDC1 subpopulation. Furthermore, the Langerin+CD8<sup>+</sup> cDC1 subset appears to be specialized in prolonged Ag cross-presentation with sustained T cell priming capacities and IL-12 production, making it a prime candidate for improved DC targeting and vaccination strategies (120).

However, do human equivalents of murine Langerin+CD8<sup>+</sup> cDC1 actually exist? And if so, what will be their relevance and how much of the murine knowledge is translatable to the human system? At first glance, similar to mice, the human DC network can be divided into multiple phenotypically and functionally distinct cDC1 and cDC2 resident and migratory DC subpopulations (16, 126, 127). However, direct comparison of the murine and human cDC subsets remains challenging (85). First of all, several differences in DC ontogeny between mice and men exist. For example, Irf8-deficiency in human resulted in a lack of both cDC1 and cDC2 subsets (128). Secondly, many of the markers used to phenotypically discriminate between the different murine DC subsets cannot be used in humans. Yet, it is now widely accepted that the expression of CD141 (BDCA3), Clec9a and XCR1 marks human cDC1, while human cDC2 are identified by CD1c (BDCA1) expression (4, 126, 129–133) (**Table 1**). Indeed, the CD141<sup>+</sup> cDC exhibit several phenotypical (16, 134–137), transcriptional (138, 139) and functional characteristics (140–142) corresponding to murine CD8<sup>+</sup> cDC1. On the other hand, no expression of Langerin could be detected on these human CD141<sup>+</sup> cDC1. Moreover, the division of labor between human cDC1 and cDC2 might be less strict as compared to mice. Generally, human CD141<sup>+</sup> cDC1 and CD1c<sup>+</sup> cDC2 are specialized in MHCI and MHCII presentation, respectively. However, depending on the type of Ag they encounter the both human resident cDC subpopulations can do both (143–146). Furthermore, human cDC1 produce IL-12, but in contrast to the mouse, also human cDC2 produce IL-12 at similar or even higher levels. These observations essentially suggest that human cDC2 are involved in orchestrating TH1 immune responses. In line, a population of human CD1a<sup>+</sup> cDC, closely related to CD1c<sup>+</sup> cDC2, expresses low levels of Langerin (142, 147–149). Accordingly, a fraction of Langerin<sup>+</sup> cDC in the mouse lacks the expression of various markers that are associated with cross-presenting cDC1 (e.g., CD103), suggesting that some cDC2 are included in the Langerin<sup>+</sup> cDC fraction as well (150). Therefore, the question whether human CD141<sup>+</sup> cDC1 are functional equivalents of the murine Langerin+CD8<sup>+</sup> cDC1 remains open, leaving the possibility that these counterparts may be found within the human CD1c<sup>+</sup> cDC2 subpopulation.

Another factor potentially determining the functional specialization of the murine Langerin+CD8<sup>+</sup> cDC1 subset might be their unique micro-anatomical niche within the spleen (**Figure 2**). Like the murine spleen, human spleen consists of RP and WP with similar functions, except that its micro-architecture differs in several ways. Importantly, humans lack marginal sinuses, and therefore the well-defined MZ found in rodents is as such absent in human spleen. Instead, humans possess a distinct histological compartment consisting of an inner and outer MZ surrounded by the perifollicular zone (10, 151, 152). Although this region might functionally represent the murine MZ, it is characterized by a different blood flow and different cellular composition (153, 154). To prevent confusion, this region may therefore better be described as the superficial zone (153). DEC205<sup>+</sup> cDC are abundantly localized in this superficial zone (155), indicating that these cells, equivalent to mice, are involved in initiating (adaptive) immune responses toward blood-borne Ag. This incomplete picture also illustrates that still many open questions remain, which should be the subject of further research into human cDC subpopulations in order to harness these cells for immunotherapy.

#### CONCLUSIONS AND FUTURE PERSPECTIVES

Many (if not all) immune functions attributed to the splenic cDC1 subpopulation appear to be exerted by the Langerin<sup>+</sup> subset. However, several developmental and functional insights regarding the Langerin+CD8<sup>+</sup> cDC1 subset and, in particular, the identification of its human counterpart remain to be clarified. On one hand, Langerin expression could reflect a more mature state of CD8<sup>+</sup> cDC1, enabling them to perform their specific immune regulatory functions. On the other hand, certain factors expressed by cell types unique to the MZ (including MZM, MMM, MZ B cells, and sinus lining cells) may facilitate the specific properties of, exclusively, the Langerin+CD8<sup>+</sup> cDC1 subset. Therefore, the elucidation of the relationship between Langerin+CD8<sup>+</sup> cDC1 and the MZ, including the determination of factors supporting the unique properties of Langerin+CD8<sup>+</sup> cDC1, may be of particular interest. The combination of high-dimensional techniques and unbiased analysis has already revealed distinct differentiation stages and/or subpopulations of human cDC1 and cDC2 (16, 126), and might allow the identification of human equivalents of the murine Langerin+CD8<sup>+</sup> cDC1. These human cDC could then potentially be exploited for future therapies of e.g., chronic inflammatory diseases or cancer.

#### REFERENCES


#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to writing this review article, and approved it for publication.

#### ACKNOWLEDGMENTS

The authors would like to thank the former and present members of their laboratories for discussion and their contributions to the work discussed in this review. We also thank Marlene Backer for comments on the manuscript. Currently, the work in our laboratory is supported by grants from the German Research Foundation (DFG) to RAB (BA 5939/2-1) and BEC (CL 419/2-1, CL 419/4-1 and CRC 1292). BEC and RAB are members of the Research Center for Immunotherapy (FZI) Mainz.

immune responses to apoptotic cell-associated antigens. J Clin Invest. (2007) 117:2268–78. doi: 10.1172/JCI31990


infections with Listeria monocytogenes or certain viruses. J Exp Med. (2016) 213:75–92. doi: 10.1084/jem.20142350


cells homologous to mouse CD8alpha+ dendritic cells. J Exp Med. (2010) 207:1283–92. doi: 10.1084/jem.20100223


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Backer, Diener and Clausen. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Poly(I:C) Potentiates T Cell Immunity to a Dendritic Cell Targeted HIV-Multiepitope Vaccine

Juliana de Souza Apostólico1,2, Victória Alves Santos Lunardelli 1,2 , Marcio Massao Yamamoto<sup>3</sup> , Edecio Cunha-Neto2,4, Silvia Beatriz Boscardin2,3 and Daniela Santoro Rosa1,2 \*

<sup>1</sup> Laboratory of Experimental Vaccines, Department of Microbiology, Immunology and Parasitology, Federal University of São Paulo, São Paulo, Brazil, <sup>2</sup> Institute for Investigation in Immunology (iii)—INCT, São Paulo, Brazil, <sup>3</sup> Department of Parasitology, Institute of Biomedical Sciences, University of São Paulo, São Paulo, Brazil, <sup>4</sup> Laboratory of Clinical Immunology and Allergy (LIM60), School of Medicine–University of São Paulo, São Paulo, Brazil

#### Edited by:

Irina Caminschi, Monash University, Australia

#### Reviewed by:

Pablo Alejandro Silveira, Anzac Research Institute, Australia Kirsteen Tullett, Monash University, Australia Jessica Li, Peter MacCallum Cancer Centre, Australia

> \*Correspondence: Daniela Santoro Rosa dsantororosa@gmail.com

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 28 September 2018 Accepted: 01 April 2019 Published: 24 April 2019

#### Citation:

Apostólico JS, Lunardelli VAS, Yamamoto MM, Cunha-Neto E, Boscardin SB and Rosa DS (2019) Poly(I:C) Potentiates T Cell Immunity to a Dendritic Cell Targeted HIV-Multiepitope Vaccine. Front. Immunol. 10:843. doi: 10.3389/fimmu.2019.00843 Cellular immune responses are implicated in resistance to HIV and have been considered for the development of an effective vaccine. Despite their safety profile, subunit vaccines need to be delivered combined with an adjuvant. In the last years, in vivo antigen targeting to dendritic cells (DCs) using chimeric monoclonal antibodies (mAb) against the DC endocytic receptor DEC205/CD205 was shown to support long-term T cell immunity. Here, we evaluated the ability of different adjuvants to modulate specific cellular immune response when eight CD4<sup>+</sup> HIV-derived epitopes (HIVBr8) were targeted to DEC205<sup>+</sup> DCs in vivo. Immunization with two doses of αDECHIVBr8 mAb along with poly(I:C) induced Th1 cytokine production and higher frequency of HIV-specific polyfunctional and long-lived T cells than MPL or CpG ODN-assisted immunization. Although each adjuvant elicited responses against the 8 epitopes present in the vaccine, the magnitude of the T cell response was higher in the presence of poly(I:C). Moreover, poly(I:C) up regulated the expression of costimulatory molecules in both cDC1 and cDC2 DCs subsets. In summary, the use of poly(I:C) in a vaccine formulation that targets multiple epitopes to the DEC205 receptor improved the potency and the quality of HIV-specific responses when compared to other vaccine-adjuvant formulations. This study highlights the importance of the rational selection of antigen/adjuvant combination to potentiate the desired immune responses.

Keywords: HIV, multiepitope vaccine, dendritic cell targeting, DEC205, adjuvants

# INTRODUCTION

Vaccine induced T cell immunity is required for effective protection against intracellular pathogens responsible for diseases classified as global threats like AIDS, tuberculosis, malaria, and also against cancer. The ability of dendritic cells (DCs) to uptake, process and present antigens is crucial to induce and regulate T cell immunity (1). Thus, activation of DCs has been considered key in vaccines designed to induce cellular immunity (2). DCs express a wide range of receptors including pattern recognition receptors (PRRs), like toll-like receptors (TLRs), cytosolic receptors, and Ctype lectin receptors, that are able to recognize pathogen- or damage- associated molecular patterns (PAMPs or DAMPs, respectively) (3). The nature of the signal delivered to the DC does not only affect the magnitude of T cell responses, but also influences the generation of memory precursors and the overall quality of immune response (4, 5).

Human and mouse DCs can be divided in two major subsets: plasmacytoid DCs and conventional/myeloid DCs with specific functions in the steady state (6–8). Recently, DCs were classified based on their ontogeny in conventional type 1 DCs (cDC1) and conventional type 2 DCs (cDC2) (9, 10). Conventional type 1 DCs encompass lymphoid CD8α <sup>+</sup> and non-lymphoid CD103+, both of which express DEC205. DEC205 also known as CD205 is a C-type lectin endocytic receptor and was the first identified DC-specific receptor (11). DEC205 is highly expressed on cDC1, but can also be found on thymic epithelial cells, Langerhans cells and, at relatively low levels, on B cells (12, 13). Recently, synthetic CpG oligonucleotides (ODNs), a potent immunostimulator, were identified as ligands that bind to the surface DEC205 (14, 15).

A promising strategy to improve vaccine efficacy is to selectively target the desired antigen to a DC subset by linking it to a monoclonal antibody (mAb) against the specific DC receptor. During the last decade, several reports revealed the feasibility of in vivo antigen targeting to cDC1 using a mAb against DEC205 (αDEC205) to improve both humoral and cellular responses (2, 16–20). Vaccination with DEC205 targeted antigens also induced protection in different infection and tumor models (21–23). However, for this particular receptor, inflammatory signals such as adjuvants must be co-administered with the targeted antigen to induce DC maturation, cellular immunity and avoid tolerance (24–26).

Different microbial products such as TLR ligands have been characterized and used as adjuvants to trigger intracellular signaling cascades that result in cytokine production, up regulation of costimulatory molecules and DCs maturation (27– 30). Mouse conventional DC subsets differentially express a broad repertoire of TLRs that result in different activating phenotypes and adaptive immunity (31). The co-delivery of TLR ligands and DEC205 targeted antigens has been shown to significantly improve vaccine immunogenicity in mice and in non-human primates (16).

Polyinosinic:polycytidylic acid [poly(I:C)] is a synthetic analog of viral double-stranded RNAs (dsRNAs) that activates TLR3 and RIG-I-like receptors (retinoic acid-inducible gene -I- like receptors, or RLRs) (32). Poly(I:C) is the most commonly administered adjuvant in mice in the context of DCtargeted vaccines using αDEC205 mAbs fused with proteins of interest (18). This strategy has already been tested with chimeric mAbs containing proteins derived from dengue virus (33), Trypanosoma cruzi (34), Plasmodium sp (26, 35, 36), Mycobacterium tuberculosis (37), Yersinia pestis (22), Toxoplasma gondii (23), HIV (21, 38, 39) and also from tumors (40). The excellent results obtained with this adjuvant, justified its use in clinical trials. To improve poly I:C stability (32) in humans, a modified version (poly-ICLC) was developed and used in different trials (41, 42).

Monophosphoryl lipid A (MPL), a chemically derivative of bacterial lipopolysaccharide (LPS), is a TLR4 agonist that preferentially activates the TIR-domain-containing adapterinducing interferon-β (TRIF) signaling pathway to drive the production of Th1 cytokines and activate CD4<sup>+</sup> T cells (43) (44, 45). MPL is the first and only TLR ligand licensed in a human vaccine (MelacineTM, approved as a melanoma vaccine). More recently, other MPL-containing vaccines became available (FendrixTM and CervarixTM, both from GSK) (46). CpG oligodeoxynucleotides (ODN) are unmethylated CpG motifs that interact with endosomal TLR9 and lead to proinflammatory cytokine production by DCs (47). B type ODN has a protective phosphorothioate backbone that protects it from nuclease digestion and enhances its half-life in vivo (48). Several clinical trials were conducted and CpG ODN emerged as a potent adjuvant to induce high antibody titers more quickly and after fewer doses (49, 50). Moreover, CpG ODN has been used along with αDEC205 mAb to target HIV and Plasmodium proteins (51, 52).

Here, we used eight promiscuous HIV-derived CD4<sup>+</sup> T cell epitopes (HIVBr8) fused with αDEC205 to target CD11c<sup>+</sup> CD8α <sup>+</sup> DCs in the presence of different TLR ligands. The hierarchy of adjuvant potency shows that poly(I:C) is a superior adjuvant for the multiepitope DC-targeted vaccine in magnitude, breadth, and longevity.

#### MATERIALS AND METHODS

## Generation of the Fusion Monoclonal Antibody (mAb)

Plasmids encoding the light and heavy chain of the mouse αDEC205 antibody were kindly provided by Dr. Michel C. Nussenzweig (The Rockefeller University, New York, USA). The plasmid encoding the heavy chain of the mouse DEC205 fused to eight HIV-1 epitopes was previously described and contains the following epitopes: p6 (32-46), p17 (73-89), pol (785-799), gp160 (188-201), rev (11-27), vpr (65-82), vif (144-158), and nef (180-194) (39).

# Expression and Purification of αDECHIVBr8 mAb

The production of αDECHIVBr8 mAb [original clone NLDC145 (24)] was performed after transient transfection of human embryonic kidney (HEK) 293T cells (ATCC, CRL-11268) exactly as described elsewhere (33). Briefly, 293T cells were cultured in 150 mm plates (Sarstedt) under standard conditions in Dulbecco's modified Eagle's medium (Invitrogen) supplemented with 1% (v/v) antibiotic-antimycotic (Invitrogen), 1% (v/v) lglutamine (Invitrogen), and 5% (v/v) Ultra low IgG Fetal Bovine Serum (Invitrogen). When cell confluence reached 70%, 293T cells were transfected using 10 µg of the plasmids encoding the light and the heavy chains in the presence of 150 mM NaCl and 0.45 mg/mL polyethyleneimine (PEI) (Sigma Aldrich). After 7 days in culture at 37◦C with 5% CO2, the culture supernatants containing secreted antibodies were collected by centrifugation at 1,000 x g for 30 min at 4◦C and filtered through 0.22µM membrane. The chimeric αDECHIVBr8 mAb was precipitated by addition of ammonium sulfate (Sigma Aldrich) to 60% of the total culture volume, and resuspended/dialyzed overnight against PBS at 4◦C. After purification by affinity chromatography with protein G beads column (GE Healthcare), fusion mAb was dialyzed against PBS, resolved on a SDS-12% polyacrylamide gel, quantified, and stored at −20◦C until use.

#### Mice

Female BALB/c (H-2<sup>d</sup> ) mice with 6-to 8-week old were purchased from Centro de Desenvolvimento de Modelos Experimentais para Medicina e Biologia (CEDEME)- Brazil. Mice were housed and manipulated under specific-pathogen-free (SPF) conditions at the animal care facility of the Division of Immunology, Federal University of São Paulo (UNIFESP).

#### Immunization

Groups of six mice were immunized twice, 2 weeks apart, with 4 µg of αDECHIVBr8 mAb by intraperitoneal (I.P) route in the presence of the following adjuvants: 50 µg of poly(I:C) (Invivogen), 20 µg of Monophosphoryl Lipid A (MPL) (Invivogen), or 10 µg of CpG ODN 1826 (Invivogen). The amount of adjuvants used was previously determined (53). Control groups were immunized with 4 µg of αDECHIVBr8 in the absence of adjuvant or with PBS only.

#### Spleen and Mesenteric Lymph Node Cell Isolation

Fifteen and sixty days after the administration of the second dose (boost), mice were deeply anesthetized by ketamine/xylazine solution (300 and 30 mg/kg, respectively) and mesenteric lymph nodes and the spleen were aseptically removed. After obtaining single cell suspensions, cells were washed in 10 mL of RPMI 1640 (Gibco). Splenic red blood cells were lysed with 1 mL of ACK solution (150 mM NH3Cl, 10 mM KHCO3, 0.1 mM EDTA) for 2 min at room temperature. After two additional washes with RPMI 1640, splenocytes and lymph node cells were then resuspended in R10 (RPMI supplemented with 10% of fetal bovine serum, 2 mM L-glutamine, 1% v/v vitamin solution, 1mM sodium pyruvate, 1% v/v non-essential amino acids solution, 40µg/mL of Gentamicin, 5 x 10−<sup>5</sup> M 2-mercaptoetanol (all from Gibco) and 20µg/mL of Cyprofloxacin (Ciprobacter, Isofarma). The viability of cells was evaluated using 0.2% Trypan Blue exclusion dye to discriminate between live and dead cells. Cell concentration was estimated with the aid of a cell counter (Countess, Invitrogen) and adjusted in cell culture medium.

# Cytokine Determination

One million splenocytes were incubated for 48 h in the presence of pooled HIV-1 peptides (5µM) or medium alone as negative control. Culture supernatants were harvested and stored at −20◦C until analysis. IL-2, IL-4, IL-6, IL-10, IL-17, IFNγ, and TNFα were detected simultaneously using mouse Th1/Th2/Th17 cytokine bead array (CBA) kit (BD Pharmingen), according to the manufacturer's instructions. The range of detection was 20– 5,000 pg/mL for each cytokine.

# T Cell ELISpot Assay

The ELISpot assay was performed using mouse IFNγ ELISpot Ready-SET-Go! (eBiosciences) according to manufacturer's instructions. Splenocytes from immunized mice were obtained as described and assayed for their ability to secrete IFNγ after in vitro stimulation with individual or pooled HIV-1 peptides (5µM) or medium alone as negative control. Spots were counted using an AID ELISPOT Reader System (Autoimmun Diagnostika GmbH, Germany). The number of IFN-γ producing cells/10<sup>6</sup> splenocytes was calculated after subtracting the negative control values and the cutoff was 15 SFU per million splenocytes.

#### Analysis of HIV-Specific Proliferation and Intracellular Cytokine Production by Flow Cytometry

To analyze HIV-specific T cell expansion, proliferation, and cytokine production, splenocytes from immunized mice were labeled with carboxyfluorescein succinimidyl ester (CFSE) (54). In summary, freshly isolated splenocytes were resuspended (50 × 10<sup>6</sup> /mL) in PBS and labeled with 1.25µM of CFSE (Molecular Probes) at 37◦C for 10 min. The reaction was quenched with RPMI 1640 supplemented with 10% FBS (R10) and cells were washed/resuspended with R10. Cells were cultured in 96-well round-bottomed plates (5 × 10<sup>5</sup> /well in triplicate) for 5 days at 37◦C and 5% CO<sup>2</sup> with medium alone or with pooled HIV-1 peptides (5µM). After 4 days, cells were restimulated with pooled HIV-1 peptides (5µM) in the presence of 2µg/mL anti-CD28 (BD Pharmingen) and Brefeldin A- GolgiPlugTM (BD Pharmingen) for further 12 h. After the incubation period, cells were washed with FACS buffer (PBS with 0.5% BSA and 2 mM EDTA) and surface stained with anti-mouse CD3 APCCy7 (clone 145-2C11), CD4 PerCP (clone RM4-5), and CD8 Pacific Blue (clone 53-6.7) monoclonal antibodies for 30 min at 4◦C. Cells were fixed and permeabilized using Cytofix/CytopermTM kit (BD Pharmingen), according to manufacturer's instructions. After permeabilization, cells were washed with Perm/Wash buffer (BD Biosciences) and stained with anti-mouse IL2 PE (clone JES6- 5H4), TNFα PECy7 (clone MP6-XT22), and IFNγ APC (clone XMG1.2) monoclonal antibodies for 30 min at 4◦C. Following staining, cells were washed twice and resuspended in FACS buffer. All antibodies were from BD Pharmingen. Samples were acquired on a FACSCanto II flow cytometer (BD Biosciences) and then analyzed using FlowJo software (version 9.9, Tree Star, San Carlo, CA). To analyze cellular polyfunctionality, we used the Boolean gate platform (FlowJo software) to create combinations of the three cytokines (IL-2, TNFα, and IFNγ) within the CFSElow population (cells that have undergone at least one cycle of division) resulting in seven distinct patterns. Polyfunctionality was defined as the ability of cells to exert at least two functions. The gating strategy, illustrated using data from one representative experiment, is shown in **Figure S1**. The frequencies of cytokine producing cells were calculated by subtracting the frequency of cells that were stimulated in vitro with HIV peptides by the frequency of the cells that were cultured in the presence of medium alone (background). For each experiment performed, unstained and all single-color controls were processed to allow proper compensation.

#### Expression of Costimulatory Molecules on DC Surface

Mice were immunized once with 4 µg of αDECHIVBr8 mAb combined with the different adjuvants (poly(I:C), MPL or Apostólico et al. Poly(I:C) Potentiates Multiepitope Targeted Vaccine

CpG ODN 1826). After 12 h, splenocytes were stained with biotinylated anti-mouse CD3 (clone 145-2C11), CD19 (clone 1D3), and CD49b (clone DX5). After 30 min, cells were washed with FACS buffer and stained with streptavidin APCCy7, antimouse CD11c APC (clone HL3), IAIE PE (clone 2G9), CD8 Pacific Blue (clone 53-6.7), CD40 FITC (clone 3.23), CD80 PerCP (clone 16-10A1), and CD86 PECy7 (clone GL1). Samples were acquired on a FACSCanto II flow cytometer (BD Biosciences) and then analyzed using FlowJo software (version 9.9, Tree Star, San Carlo, CA). For each experiment performed, unstained and all single-color controls were processed to allow proper compensation. Three million events were acquired in a live lymphocyte gate.

#### Data Analysis

Statistical significance (p-value) was calculated by Two-way ANOVA followed by Bonferroni post hoc test or unpaired ttest (different time points comparison). Statistical analysis and graphical representation of data was performed using GraphPad Prism version 7.0 software.

# RESULTS

# Multiepitope Targeting to DEC205<sup>+</sup> DCs With Different Adjuvants Induces Type 1 Cytokine Production

To examine the effect of different adjuvants on HIV-specific cellular immune response, mice were immunized with two doses of αDECHIVBr8 mAb in the presence of the TLR agonists poly(I:C), MPL or CpG ODN 1826. Fifteen or Sixty days after the boost, splenocytes from immunized mice were incubated with pooled HIV-1 peptides to analyze specific cytokine production. First, we evaluated IFNγ production by ELISpot assay (**Figure 1A**). We observed that 15 days after the boost splenocytes from mice immunized with αDECHIVBr8 mAb combined with poly(I:C) presented higher number of specific IFNγ producing cells (716 SFU/10<sup>6</sup> cells) when compared to the groups immunized in the presence of MPL or CpG ODN 1826 (404 and 286 SFU/10<sup>6</sup> cells, respectively). Moreover, a significant difference was observed between MPL and CpG ODN 1826 groups (**Figure 1A**, left). Sixty days after the boost, we detected the same profile albeit with lower magnitude. Mice immunized with αDECHIVBr8 combined with poly(I:C) displayed 514 SFU/10<sup>6</sup> cells while MPL and CpG ODN 1826 presented 284 and 142 SFU/10<sup>6</sup> cells, respectively (**Figure 1A**, right). A comparison between 15 and 60 days revealed a significant decrease in the magnitude for poly(I:C) (p < 0.001), CpG ODN 1826 (p < 0.001), and MPL (p < 0.01) immunized groups. Splenocytes from mice immunized with αDECHIVBr8 in the absence of adjuvant or PBS (control groups) presented negligible numbers of IFNγ producing cells.

We also analyzed the cytokine profile by CBA assay using supernatant culture of splenocytes stimulated with pooled HIV peptides. Splenocytes from mice that received αDECHIVBr8 combined with poly(I:C) produced higher levels of IFNγ when compared to MPL or CpG ODN 1826, corroborating the ELISpot findings (**Figure 1B**). Interestingly, in the poly(I:C) adjuvanted group IFNγ production was even higher 60 days after the boost when compared to the 15 days time point (p < 0.001). Poly(I:C) also induced superior IL-2 production 15 days after the boost (**Figure 1C**, left). However, 60 days after the boost, IL-2 production significantly decreased in the group immunized with αDECHIVBr8 plus poly(I:C) and increased in the group that received the mAb in the presence of MPL (p < 0.001) (**Figure 1C**, right). IL-2 production by the group that received the mAb with CpG ODN 1826 slightly increased 60 days after the boost when compared to 15 days time point (p < 0.001). Regarding TNFα production 15 days after the boost, we observed that αDECHIVBr8 mixed with MPL produced the highest levels (**Figure 1D**). TNFα levels increased 60 days after the boost for the poly(I:C) group (p < 0.001) and decreased for the CpG ODN 1826 group (p < 0.05). No difference was observed for the MPL immunized group. Inflammatory IL-6 (**Figure 1E**) was higher in the group immunized with αDECHIVBr8 plus MPL 15 days after boost, but at the later time point the levels of this cytokine significantly decreased (p < 0.01). In contrast, 60 days after the boost with mAb and poly(I:C), IL-6 (p < 0.01) production increased considerably. IL-10 (**Figure 1F**) was superior in the poly(I:C) immunized group in both time points followed by MPL immunized group. However, after 60 days, IL-10 production decreased in the MPL (p < 0.001) and in the CpG ODN 1826 (p < 0.05) groups. Of note, IL-4 and IL-17 production was below the assay detection limit (data not shown). Taken together, these results indicate that different adjuvants induce a type 1 immune response when multiple HIV-antigens are delivered to CD8α + DCs by the endocytic receptor DEC205.

#### Poly(I:C) Promotes Robust and Long-Lived Polyfunctional T Cell Responses

In an attempt to evaluate HIV-specific CD4<sup>+</sup> and CD8<sup>+</sup> T cell responses, splenocytes from immunized mice were labeled with CFSE and pulsed in vitro with HIV-1 peptides. After culture, the frequency of CD3+CD4+CFSElow (**Figure 2A**) and CD3+CD8+CFSElow (**Figure 2B**) were evaluated by flow cytometry. Fifteen days after boost, splenocytes from mice that received αDECHIVBr8 along with poly(I:C) presented higher frequency of proliferating CD4<sup>+</sup> (9.96%) and CD8<sup>+</sup> (5.90%) T cells when compared to MPL immunized groups (6.83 and 4.86%, respectively). In contrast, CpG ODN 1826 displayed the lowest frequency of proliferating T cells. The same profile was observed 60 days after the boost, with the group that received αDECHIVBr8 plus poly(I:C) displaying higher CD4<sup>+</sup> (11.30%) and CD8<sup>+</sup> (6.17%) specific proliferation when compared to MPL (CD4+CFSElow 4.86% and CD8+CFSElow 2.47%) or CpG ODN 1826 (CD4+CFSElow 3.60% and CD8+CFSElow 1.31%) (**Figures 2A,B** right, respectively). Comparative analyses showed significant difference on the frequency of CD4+CFSElow cells between 15 and 60 days only for the group that received αDECHIVBr8 plus MPL (p < 0.05). Regarding the CD8<sup>+</sup> T cell compartment (CD8+CFSElow cells), a significant difference was observed for MPL (p < 0.05) or CpG ODN 1826 (p < 0.01) groups. In contrast, mice immunized with αDECHIVBr8

followed by Bonferroni post hoc test or unpaired t-test (different time points comparison) \*p < 0.05, \*\*p < 0.01; \*\*\*p < 0.001; a p < 0.05; b p < 0.01; c p < 0.001 when 15 days was compared to 60 days time point. Data represent mean ± SD and are representative of 3 independent experiments.

in the presence of poly(I:C) displayed similar frequency of proliferating CD4<sup>+</sup> and CD8<sup>+</sup> T cells in all time points. To further characterize the functional profile of antigen-specific T cells, we assessed the ability of single cells to proliferate and produce the cytokines IFNγ, TNFα, and IL2 individually or simultaneously. The flow cytometry profile demonstrated that immunization with αDECHIVBr8 mAb along with poly(I:C) induced higher frequency of CD4<sup>+</sup> T cells that proliferated and produced IFNγ <sup>+</sup>IL2+TNFα <sup>+</sup> or IFNγ <sup>+</sup>TNFα <sup>+</sup> simultaneously or only one cytokine (IFNγ or TNFα) 15 or 60 days after the boost (**Figures 3A,B**, respectively). Interestingly, for the poly(I:C) and MPL groups 60 days after the boost, the frequency of polyfunctional CD4<sup>+</sup> T cells that proliferated and produced IFNγ, TNFα, and IL-2 simultaneously decreased, leading to an increase in the double or single cytokine producers (**Figure 3**– pie charts). Moreover, αDECHIVBr8 mixed with poly(I:C) also displayed higher frequency of proliferating CD8<sup>+</sup> T cells that produce IFNγ or TNFα 15 or 60 days after the boost when compared with other groups (**Figures 3C,D**, respectively). Similarly to what was observed with the CD4 compartment at the later time point (60 days), there was also a shift in the CD8<sup>+</sup> T cell polyfunctional profile in all groups when compared to 15 days after the boost; the frequency of three cytokine producing cells diminished while the single cytokine producers augmented (**Figure 3** pie charts). Altogether, these results demonstrated that immunization with two doses of αDECHIVBr8 along with poly(I:C) induced higher and long-lasting specific polyfunctional CD4<sup>+</sup> and CD8<sup>+</sup> T cells responses.

#### Poly(I:C) Increases Epitope Coverage

To assess the breadth of T cell responses, splenocytes from immunized mice were incubated with single HIV-1 peptides present in the fusion vaccine and the number of IFNγ producing cells was determined by ELISpot. Fifteen days after last dose (**Figure 4A**), all adjuvants tested were able to induce positive responses against all peptides, albeit at different magnitudes (poly(I:C) > MPL > CpG ODN). At a later time point (**Figure 4B**), poly(I:C), and CpG ODN adjuvanted groups sustained IFNγ production against all peptides (head-to-head comparison in **Figures S2A,C**). On the contrary, in the MPL group, the magnitude of the response was more significantly reduced when we compared the 15 and 60 days time points (**Figure S2B**). Thus, multiepitope in vivo targeting to DEC205<sup>+</sup> DCs when combined with poly(I:C) induced broad, potent and long-lasting T cell responses.

#### Differential Expression of Costimulatory Molecules in Splenic DCs Subsets

To further characterize phenotypic differences among the adjuvants, we compared the maturation status of splenic DCs after in vivo administration of the mAb combined with poly(I:C), MPL or CpG ODN 1826. The gating strategy, illustrated using data from one representative experiment, is shown in **Figure S3**. Twelve hours after injection, CD11c+CD8α <sup>+</sup> DCs from poly(I:C) group considerably up-regulated the expression of CD80 compared to other groups (**Figures 5A,B**). CpG ODN 1826 slightly increased CD80 expression only when compared to MPL. However, none of the adjuvants up regulated CD80 expression on CD11c+CD8α <sup>−</sup> DCs. Furthermore, poly(I:C) was the only adjuvant to significantly up regulate CD86 expression in both DCs subsets (**Figures 5C,D**). Similarly, we observed a significant increase in the MFI of CD40 molecule by poly(I:C) in both DCs subsets when compared to other adjuvants (**Figures 5E,F**). In addition, to assess whether DC activation could occur

earlier than 12 h, we analyzed the expression of costimulatory molecules 6 h after injection, and observed the same pattern of CD80, CD86, and CD40 expression in both DCs subsets (**Figures S4A–C**, respectively). We also analyzed the activation profile on mesenteric lymph nodes and the same pattern of expression was observed (data not shown). Taken together, these results strength the idea that poly(I:C) is a superior adjuvant than MPL or CpG ODN 1826 since it up regulates costimulatory molecules in both splenic DCs subsets (CD8α <sup>+</sup> and CD8α −).

#### DISCUSSION

Antigen targeting to DCs through DEC205 endocytic receptor is an effective way to enhance antigen uptake. However, the induction of cell immunity is only accomplished when αDEC205 chimeric mAbs are delivered together with an adjuvant (55–57). Adjuvants enhance immunity to vaccine antigens by influencing the magnitude, breadth/immunodominance, and persistence of immune responses (27). Hence, the choice of the adjuvant formulation is of utmost importance to induce the desired immune response (58). Although a limited number of vaccine adjuvants are currently licensed for human use (aluminum salts, MF59, AS03, and AS04), several compounds have entered clinical trials with demonstrated efficacy (27).

Antigen targeting to DCs through DEC205 receptor is used as a vaccination strategy to induce strong antigen-specific immune responses against several pathogens (26, 34, 55) and tumors In the HIV vaccine scenario, antigen targeting to cDC1 through DEC205 was performed using the full-length gag (p24) protein (21, 51, 59–62). The success in different pre-clinical studies using mice and non-human primates (16) quickly pushed forward the translation of this strategy to humans. Recently, two phase I clinical trials (NCT01889719 and NCT01127464) delivered HIV p24 using a human αDEC205 mAb plus poly-ICLC as adjuvant. Promising results were obtained when a human αDEC205

mAb fused to the full-length tumor antigen NY-ESO-1 was administered together with poly-ICLC (41, 42). In fact, three phase I/IIb clinical trials are currently under way (NCT02166905, NCT03206047, NCT03358719) and two others are already completed (NCT01522820, NCT00948961, NCT01834248).

Previously, we generated an αDEC205 multiepitope fusion mAb (αDECHIVBr8) to target eight promiscuous CD4<sup>+</sup> T cell epitopes from several HIV proteins to cDC1s. The αDECHIVBr8 mAb was administered to mice in the presence of poly(I:C) as adjuvant and compared to DNA plasmid immunization in homologous and heterologous prime-boost regimens. We found that αDECHIVBr8 homologous primeboost regimen induced stronger T cell immune responses against all epitopes when compared to homologous DNA vaccination (39). Here, we compared the adjuvant properties of poly(I:C), MPL, and CpG ODN 1826 to induce HIVspecific cellular immune response when formulated with the fusion αDECHIVBr8 mAb. To our knowledge, this is the first time that multiple epitopes derived from different proteins of the same pathogen are targeted in vivo to DCs and tested in the context of different adjuvants. This is an important issue since adjuvants can influence immunodominance by altering the immune repertoire of CD4 T cell responses (63). Overall, our data reveal the potential of poly(I:C) as a superior adjuvant for the development of a multiepitopebased vaccine that targets CD8α <sup>+</sup> DCs through the DEC205 endocytic receptor.

Initially, we found that poly(I:C) induced higher magnitude of specific IFNγ producing cells and also Th1 cytokine production when compared to MPL or CpG ODN 1826. Likewise, Longhi et al. showed that poly(I:C) is a more potent adjuvant to induce specific immune responses against a DC-targeted HIV gag protein (51). Indeed, poly(I:C) has been the most commonly administered adjuvant with DC-targeted vaccines using αDEC205 mAbs fused with full-length proteins from

different pathogens in both mice and non-human primates (21– 23, 33–39, 60).

Poly(I:C) is sensed by TLR3 and RLR receptors, and triggers up regulation of costimulatory molecules, strong type I IFN production by DCs and Th1 responses (32). Type I IFNs mediate the adjuvant effect of poly(I:C) acting as a third signal by promoting and sustaining clonal expansion of T cells (64–68). Indeed, our results demonstrate that immunization with αDECHIVBr8 along with poly(I:C) also induced higher frequency of proliferating CD4<sup>+</sup> and CD8<sup>+</sup> T cells. Moreover, we found that administration of the αDECHIVBr8 mAb concomitant with poly(I:C) induced higher frequency of specific polyfunctional T cells, i.e., cells that proliferated and simultaneously produced Th1 cytokines (IFNγ, IL2, and TNFα). Ours results corroborate with previous reports showing the development of polyfunctional T cells after HIV gag protein targeting to DCs along with poly(I:C) (21, 38, 62).

Additionally, the presence of polyfunctional T cells is also a hallmark after vaccinia and yellow fever virus vaccinations (69, 70), and correlates with non-progressive HIV infection (71, 72). Recent HIV vaccine trials suggest that a broad (multiple specificities) and potent (high magnitude) response against conserved epitopes would be a desirable attribute of a T-cell based vaccine (73, 74). Indeed, vaccine induced broad T cell responses conferred protection after simian immunodeficiency virus challenge (75). We showed that poly(I:C) and MPL induced T cell responses against all epitopes (broad responses) present in the αDECHIVBr8 fusion mAb, although poly(I:C) was more potent. Likewise, Teixeira et al. demonstrated the ability of a bacterial adjuvant (Propionibacterium acnes) to expand the breath of a multiepitope DNA-based HIV vaccine (76).

A central feature of successful vaccines is their ability to induce immunological memory. Cross-sectional studies of smallpox and yellow fever vaccines showed that specific humoral and

T cell responses can be detected for many years (77, 78). When we analyzed the longevity of the immune response, only poly(I:C) vaccine group had sustained T cell proliferation and IFNγ responses against all peptides ∼2 months after the second immunization. It is important to note that MPL was the second most potent adjuvant tested and better to induce pro-inflammatory cytokines such as TNFα and IL-6. Previous reports provided evidence that MPL, a TLR2, and TLR4 agonist, is effective to induce TNFα, IL-10, and IL-12 production (44, 79). MPL induced a broad T cell response after the boost but narrowed after 2 months. Previous reports using αDEC205 mAb fused with HIV gag protein showed that MPL or LPS were as effective as poly(I:C) to induce specific humoral responses but less potent to induce Th1 CD4<sup>+</sup> T cell immunity (38, 51).

Interestingly, immunization with αDECHIVBr8 in the presence of CpG ODN induced weak T cell responses and narrowed epitope positivity. B class CpG ODN is a fully phosphorothioate TLR9 agonist that binds to surface DEC205 receptor (14, 15) and could therefore compete with the fusion αDEC205 mAb for cellular uptake. Our data are in line with a previous study demonstrating that immunization with αDEC-Gag plus CpG ODN 1826 induces lower frequency of responding CD4<sup>+</sup> T cells compared with poly(I:C) (51).

Anti-DECHIVBr8 combined with poly(I:C) was the most effective strategy to modulate DC activation by up regulating costimulatory molecules in a more pronounced way in the CD11c<sup>+</sup> CD8α <sup>+</sup> subset but also in CD11c<sup>+</sup> CD8α <sup>−</sup> DCs. This may be due to the fact that CD8α <sup>+</sup> DEC205<sup>+</sup> DCs express higher levels of TLR3 when compared to CD8α <sup>−</sup> DCs (2, 18, 80). As a consequence of DC maturation, poly(I:C) enhanced T cell immunity. As stated before, it was shown that poly(I:C) was most effective to induce Th1 CD4<sup>+</sup> T cell immunity compared to LPS or CpG ODN 1826 using the HIV gag targeted protein (51).

< 0.001. Data represent mean ± SD and are representative of 3 independent experiments.

The use of mouse model to select an adjuvant may be a caveat since the pattern of expression of TLR in the target DEC205<sup>+</sup> DC subset can differ between human and mouse (18). However, the adjuvant effect after antigen targeting does not necessarily rely on the direct activation of its respective TLR. For example, the effect of poly(I:C) on cDC1 is mediated by type I IFN receptor (51) suggesting that it is possible to have immune activation even if the targeted DC does not express a certain TLR.

Collectively, the observations demonstrate that combination of poly(I:C) with multiepitope targeting to DEC205<sup>+</sup> DCs modulates DC activation and elicits strong, broad, polyfunctional, and long-lived Th1 responses superior to other adjuvants both in quantity and quality. Therefore, the pursuit of a safe and effective T cell-based vaccine may benefit from the proper association of multiple epitope targeting to DC populations using a potent adjuvant formulation.

#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Federal Law 11.794 (2008) and the Guide for the Care and Use of Laboratory Animals of the Brazilian National Council of Animal Experimentation (CONCEA). The protocol was approved by the UNIFESP Animal Care and Use Committee (IACUC).

#### AUTHOR CONTRIBUTIONS

JA, SB, and DR conceived and designed the experiments. JA, VL, MY, and DR performed the experiments. JA, VL, and DR analyzed the data and prepared the figures. DR, SB, and EC-N contributed with reagents and materials. JA, VL, SB, and

#### REFERENCES


DR wrote the manuscript. SB, EC-N, and DR performed the final review of the article. All authors read and approved the final article.

#### FUNDING

This research was supported by the São Paulo Research Foundation (FAPESP, grant numbers 2014/50631-0 and 2017/17471-7), the Brazilian National Research Council (CNPq)/Institute for Investigation in Immunology and the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-Brasil (CAPES, Finance Code 001). JA, VL, EC-N, SB, and DR received fellowships from CNPq/ FAPESP.

#### ACKNOWLEDGMENTS

We thank Mr. Geová Santos for assistance in the animal facility.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.00843/full#supplementary-material

antibody NLDC-145. I Expression on dendritic cells and other subsets of mouse leukocytes. Cell Immunol. (1995) 163:148–56.


promote immunogenicity and protective efficiency against toxoplasmosis. Front Immunol. (2018) 9:317. doi: 10.3389/fimmu.2018.00317


to the dendritic cell receptor DEC-205. Sci Transl Med. (2014) 6:232ra251. doi: 10.1126/scitranslmed.3008068


in nonhuman primates. Proc Natl Acad Sci USA. (2011) 108:7131–6. doi: 10.1073/pnas.1103869108


not noncontrollers maintain CD4 T cells coexpressing three cytokines. J Virol. (2007) 81:12071–6. doi: 10.1128/JVI.01261-07


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Apostólico, Lunardelli, Yamamoto, Cunha-Neto, Boscardin and Rosa. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Beyond cDC1: Emerging Roles of DC Crosstalk in Cancer Immunity

#### Rajkumar Noubade\*, Sonia Majri-Morrison and Kristin V. Tarbell

Department of Inflammation and Oncology, Amgen Research, Amgen Inc., South San Francisco, CA, United States

Dendritic cells (DCs) efficiently process and present antigens to T cells, and by integrating environmental signals, link innate and adaptive immunity. DCs also control the balance between tolerance and immunity, and are required for T-cell mediated anti-tumor immunity. One subset of classical DCs, cDC1, are particularly important for eliciting CD8 T cells that can kill tumor cells. cDC1s are superior in antigen cross-presentation, a process of presenting exogenous antigens on MHC class I to activate CD8<sup>+</sup> T cells. Tumor-associated cDC1s can transport tumor antigen to the draining lymph node and cross-present tumor antigens, resulting in priming and activation of cytotoxic T cells. Although cross-presenting cDC1s are critical for eliciting anti-tumor T cell responses, the role and importance of other DC subsets in anti-tumor immunity is not as well-characterized. Recent literature in other contexts suggests that critical crosstalk between DC subsets can significantly alter biological outcomes, and these DC interactions likely also contribute significantly to tumor-specific immune responses. Therefore, antigen presentation by cDC1s may be necessary but not sufficient for maximal immune responses against cancer. Here, we discuss recent advances in the understanding of DC subset interactions to maximize anti-tumor immunity, and propose that such interactions should be considered for the development of better DC-targeted immunotherapies.

#### Edited by:

Diana Dudziak, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Joke M. M. Den Haan, VU University Medical Center, Netherlands Roxane Tussiwand, Universität Basel, Switzerland

> \*Correspondence: Rajkumar Noubade rnoubade@amgen.com

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

> Received: 30 August 2018 Accepted: 23 April 2019 Published: 09 May 2019

#### Citation:

Noubade R, Majri-Morrison S and Tarbell KV (2019) Beyond cDC1: Emerging Roles of DC Crosstalk in Cancer Immunity. Front. Immunol. 10:1014. doi: 10.3389/fimmu.2019.01014 Keywords: dendritic cells, cDC1, cDC2, crosstalk, cancer immunity

# INTRODUCTION

The interaction between various myeloid and lymphoid cell populations is crucial to initiate and orchestrate a robust anti-tumor response. By processing tumor associated antigens (TAAs) and migrating to draining lymph nodes (dLN), where T cell priming occurs, dendritic cells (DCs) are considered the most potent professional antigen presenting cells (APCs) to elicit adaptive anti-tumor immunity (1). In addition to presenting antigens, DCs use soluble molecules such as cytokines and chemokines as well as direct cell-cell contacts to prime and activate TAA-specific T cells. DCs were discovered by Ralph Steinman and Zanvil Cohn in 1973 as an APC population, distinct from macrophages, that initiate adaptive immune responses (2). As a result of more recent deep-phenotyping, DCs are now recognized to be a heterogenous population comprising several subsets distinguished by their development, phenotypic differences, localization, and functional specialization (2–6). This functional specialization of each subset allows DCs to initiate distinct immune responses in different immunological contexts (7). Here, we review literature supporting the hypothesis that, although one DC subset, conventional DC1(cDC1), has been shown to be crucial for anti-tumor immunity, multiple DC subsets, and interactions with other cells are needed for maximal responses.

#### DC Subsets Are Functionally Specialized

DCs are broadly classified as classical (or conventional) DCs (cDCs) and plasmacytoid DCs (pDCs), each with specialized functions. cDCs, specialized in antigen presentation to naïve T cells can be further segregated into cDC1s and cDC2s, excelling in MHC class I- and class II-mediated antigen-presentation, respectively (3, 6, 8–10). cDCs are found both as lymphoid and non-lymphoid tissue cells, the latter of which can migrate via the lymph to dLN to present tissue-derived antigens (3, 11). cDC1s, present at lower frequency compared to cDC2s, are identified by the expression of XCR1 (12), and in humans, also by the expression of CD141 (BDCA3) (5, 13, 14). cDC1s possess specialized mechanisms to mediate efficient antigen recognition, antigen transport to appropriate endosomal compartments and subsequent processing for the presentation to CD8 T cells in a process known as cross-presentation (15–18). cDC1s can also activate CD4 T cells through MHC class II antigen presentation and can polarize activated CD4 T cells toward a Th1 phenotype through the secretion of IL-12 (19).

cDC2s are specialized in MHC class II-mediated antigen presentation and are the most efficient APCs for activation and expansion of CD4 T cells (5, 13, 20). They are the most frequent DC population present in blood, lymphoid organs and tissues and promote a wide range of immune responses including Th1, Th2, and Th17 in specific contexts (13, 19, 21–25). Human cDC2s can be identified by their preferential expression of CD1c (BDCA1) and CD172a (SIRPα) (26). cDC2s are more heterogenous than cDC1s, and express various receptors that enable them to respond to broad spectrum of microbial products (22, 26–28). A subset of Notch2-dependent cDC2s specializes in IL-23 production and contributes to innate defense and adaptive immune responses (27, 29).

pDCs, distinguished by their ability to produce large amounts of type I IFN upon viral infection (30–33) are identified, in humans, by the expression of surface markers CD303 (BDCA-2), CD304 (BDCA-4/Neuropilin) and CD123 (5, 13). They are present mainly in lymphoid organs and can migrate to the LN through blood circulation (5, 34). Mature pDCs can also act as APCs and have distinct regulation of MHC class II surface expression that results in sustained membrane peptide-MHC complex and antigen presentation (30). A heterogeneity of pDCs is also described in terms of their ability to produce type I IFN and/or antigen presentation (35, 36).

Another related but developmentally distinct population, derived from monocytes, termed monocyte DCs (moDC) upregulates certain functional properties of DCs in some contexts and express tumor necrosis factor (TNF)- α and intracellular nitric oxide synthase (iNOS) (37). More commonly, the term moDCs refers to monocyte isolated from human peripheral blood mononuclear cells (PBMC) that are in vitro differentiated in the presence of granulocyte-macrophage colony-stimulating factor (GM-CSF) and interleukin (IL)-4 into cells sharing several phenotypic and functional features of DCs (26, 38, 39). moDCs are the most common in vitro model of DCs, yet are quite heterogeneous in both mouse and human, with unclear relationship to in vivo cell populations (40–42).

All DC subsets, including cDCs and pDCs, are found in the tumor microenvironment (TME) (30, 43–47) and among the cDCs, the cDC2s outnumber cDC1s, with cDC1s being the rarest APCs within the TME (43, 48). The role of pDCs in tumor immunity remains elusive and contradictory. Similarly, the precise role of cDC2s in anti-tumor immunity has been difficult to delineate due to lack of proper genetic tools. On the contrary, mounting evidence suggests cDC1s to be the critical antigen presenting DC subset for the generation of anti-tumor immunity. Here we summarize data supporting the importance of cDC1s in anti-tumor immunity, and then review the recent literature that documents DC crosstalk being necessary for effective immune responses, in other contexts such as anti-viral immune responses, and apply these principles to tumor immunity.

#### cDC1s Are Necessary for Anti-tumor Immunity

Since MHC class I molecules are expressed by every cell in the body (not just infected cells and cancer cells), to avoid bystander killing of healthy cells by CTLs, extracellular antigens do not enter the MHC class I-loading machinery (15, 18). Therefore, to generate an immune response, cancer cell antigens need special processing in APCs to be presented to naïve CD8 T cells. Moreover, naïve CD8 T cells primarily circulate through secondary lymphoid organs (15). Hence, cancer antigens must be brought to secondary lymphoid organs to be presented to naïve CD8 T cells. cDC1s fulfill both functions by patrolling tumor tissues, and by capturing, processing and presenting tumorantigens on their surface through MHC class I molecules via antigen cross-presentation. cDC1s then migrate to dLN and deliver peptide/MHC class I signal to CD8 T cells which leads to their activation and the initiation of an immune response against tumor cells (15, 18).

Although other cell types have been reported to cross-present antigens (11, 49), this specialized function is mostly attributed to the cDC1 subset, owing to their unique adaptations of subcellular molecular machinery and vesicular trafficking (15, 18). Such adaptations include efficient antigen uptake of dying cells, delivery of cell-associated antigen to early endosomes, (15, 50– 52), efficient phagosome-to-cytosol export of an ingested antigen possibly aided by ER-derived translocons and ER-associated degradation (ERAD) components such as Sec61, Derlin, p97 ATPase, Sec22 (15, 53–55), lower expression of lysosomal proteases (50) and antagonizing their degradative functions via NOX2-mediated ROS generation (56–60). The end result of such lower proteolysis, and therefore, increased antigen retention in cDC1s, is eventually an enhanced ability to carry the antigen all the way from peripheral tissues where the antigen is captured, to the dLN, where priming and activation of CD8 T cells occurs (56). The importance of cDC1s' ability to cross-present antigen in its immune functions is recently demonstrated using Wdfy4 deficient mice, which selectively lack cross-presentation (61).

Beyond their role in antigen cross-presentation, cDC1s are the major source of IL-12 production and thus influence antitumor immunity by activating NK cells and driving CD4 T cell responses toward Th1 responses (19, 62–64).

The critical role of cDC1s in anti-tumor immunity has been shown using mice deficient in basic leucine zipper transcription factor ATF-like 3 (Batf3), a transcription factor required for cDC1 differentiation (65). Batf3 knockout mice lack cDC1 cells but not other APCs and display impaired anti-tumor immunity in several models (43, 65–68). Expansion and activation of cDC1s using fms-related tyrosine kinase 3 ligand (Flt3L) and poly I:C leads to significant enhancement of antitumor responses (45). Immunotherapies such as PD1/PD-L1 blockade or CD137 agonists are ineffective in Batf3-deficient mice, highlighting the crucial role cDC1s in tumor immunity (68, 69). Furthermore, tumor-resident cDC1s are required for trafficking of adoptively transferred CD8 T cells into tumors through their ability to produce CXCL9 and CXCL10 (67, 70). DC-specific deletion of Sec22b leads not only to impaired cross-presentation of TAAs and reduced anti-tumor immune responses but also abolishes the efficacy of anti-PD1 therapy (53). In humans, the presence of cDC1s within the TME is associated with better prognosis and response to immunotherapy. Analysis of the cancer genome atlas (TCGA) dataset shows that a higher ratio of a cDC1 gene signature to a signature of all other myeloid cells (including monocyte/macrophage, and not just other DC subsets) is associated with better prognosis across human tumors (44, 71). Abundance of CD8 T cells positively correlates with cDC1 markers in pancreatic tumors (70). Taken together, these data show that cross-presenting cDC1s are crucial and necessary for the generation of an effective anti-tumor immunity.

#### cDC1 Are Not Sufficient for Maximal Anti-tumor Immunity: Potential Roles of Other DC Subsets

Tumor immunology is built upon the tenet that cytotoxic CD8 T cells (CTLs) eliminate tumor cells (72) and the prevailing dogma is that cDC1s are the most potent APCs for the CTL response against tumor. Because of the strong evidence for the importance of cDC1 in tumor immunity, as presented in the previous section, in one scenario it is possible that cDC1s are the sole DC subset sufficient for optimal anti-tumor CTL generation through antigen presentation via MHC class I as well as MHC class II (**Figure 1A**). A major driver of the current dogma is the studies conducted in mice genetically manipulated to lack cDC1 such as Batf3-deficient or Zbtb46-Cre mice. However, these tools are imperfect. For example, Batf3 is expressed in cDC2 and effector CD4 T cells (65, 73) and Zbtb46 is also expressed in DC2s as well as in endothelial cells (74–76), raising the possibility of contributions from additional DC subsets and other cell types. Hence, it is not clear whether the cDC1 subset alone is sufficient to provide the maximal immunity against tumor. Recent evidence in non-tumor settings has demonstrated that cDC1s require significant contributions from other DC subsets and are not sufficient for an optimal CTL response (77–79), pointing toward a role for the other cells in shaping a robust and durable anti-tumor immunity.

Therefore, we describe a second scenario that includes possible roles of other DC subsets for a more robust anti-tumor immunity, directly and indirectly (**Figure 1B**). This scenario

FIGURE 1 | Potential scenarios of DC crosstalk in anti-tumor immunity. (A) Describes a scenario where an effective anti-tumor immune response would rely solely on cDC1 functions. cDC1s can activate both CD8 T cells and CD4 T cells through MHC class I- and MHC class II-mediated antigen presentation, respectively. Activated CD4 T cells provide licensing signal to cDC1s, which relay that help to CD8 T cells. Helped CD8 T cells have enhanced cytotoxic properties to efficiently kill tumor cells. (B) Describes multi-cellular interactions to achieve full-strength CTL responses against tumor. In this scenario, cDC1s predominantly activate CD8 T cells and cDC2s predominantly activate CD4 T cells. Activated CD4 T cells, in addition to providing help to maximize CTL responses can directly exhibit anti-tumor responses. Activated pDCs can modulate the TME mainly via type I IFN production, but can also activate CD4 T cells via MHC class II-mediated antigen presentation. Solid line indicates strong experimental evidence in tumor setting and dashed line indicates data in non-tumor setting. Thick line indicates predominant function.

incorporates recent findings of spatiotemporal segregation of cDC1 and cDC2 activation within dLN to activate CD8 and CD4 T cells, respectively, during antiviral response. This robust CTL response requires interactions between multiple DC subsets, including cDCs and pDCs in a two-step priming process (77– 79). Even though these responses are context dependent and are observed in anti-viral response, the general principals remain the same in anti-tumor immune response. Accordingly, in this scenario, the tumor-derived cDC1 primes CD8 T cells while tumor-derived cDC2 activates CD4 T cells in the first step of the CTL priming process and then in the second step, the activated CD4 T cells licenses a LN-resident cDC1 to relay the help for CTLs. Contributions of activated CD4 T cells to antitumor immunity can be more than just providing the help to CTLs, but also include activation of NK cells and macrophages through IFN-γ, modulation of tumor stroma and angiogenesis or direct cytolytic effects (80–83).

Additionally, during the two-step priming process, pDCs are recruited to cDC1-CD8 T cell priming sites, providing critical licensing signal to cDC1s through type 1 IFN. In this regard, lack of type 1 IFN receptor in cDC1s impairs their ability to reject tumors (84, 85). Furthermore, pDCs are usually weak APCs in the absence of activating signals but direct antigen presentation and T cell stimulation by pDCs has been described (30, 86). In fact, adoptive transfer of tumor-antigen-loaded pDCs induced potent anti-tumor T cell responses in melanoma patients (87), suggesting the possibility of anti-tumor immunity directly through APC functions by pDCs.

In the following sections, we mainly focus on this latter scenario of non-synchronous activation events by cDC1s and cDC2s and the reorganization of pDCs to the sites of CTL priming to describe the crosstalk between DC subsets and propose an integrated model of multi-DC subsets, multi-cell type interactions in achieving full-strength CTL responses in anti-tumor immunity.

#### Crosstalk Between DC Subsets

One of the goals of cancer immunotherapy is to promote tumorantigen specific T cell responses. The current data supports the notion that cDC1s are well-suited for this purpose and that they are usually necessary for the generation of an antitumor response. However, as discussed below, they may not be sufficient for full-strength anti-tumor cytotoxic T cell responses and interactions with other DC subsets contribute to this process. In the following sections, we will review the interactions between each DC subsets separately.

#### cDC-pDC Crosstalk

cDCs and pDCs are co-localized in many immune contexts, e.g., non-inflamed LNs, skin biopsies from lupus erythematosus patients, thyroid glands from autoimmune thyroiditis patients and spleens of cancer patients (88–90). Such close-proximity of pDCs and cDCs suggests possible functional coordination. Indeed, local production of type I IFN by pDCs induces stimulatory molecules on cDCs driving their maturation during an effective immune response (79). Intravital two-photon microscopic analysis of DC subsets within dLN during vaccinia virus infection showed active, CCR5-mediated recruitment of pDCs to the site of CD8 T cell priming by virus-infected cDC1. The activated CD8 T cells also orchestrate, via XCL1, recruitment of resident, non-infected XCR1+cDC1s. pDCs produce type I IFN to induce upregulation of costimulatory molecules including CD40, CD80, and CD86 on non-infected resident-cDC1s (79), driving their maturation and antigen-presentation functions leading to robust CTL response. pDC help for CTL response, either through type I IFN or other costimulatory molecules such as CD40L has been described in other viral infection models (91– 93). Depletion of pDCs results in impaired CTL responses in many viral infections, e.g., VSV infection (94), LCMV infection (95), and cutaneous herpes simplex virus (HSV) (92). In the LCMV infection model, pDC-mediated CD4 T cell activation was essential in providing help and generation of anti-viral CTL response (95). These observations underscore the pivotal role of the crosstalk between DC subsets in maximizing immune response against cell-associated antigens.

Similarly, in the context of anti-tumor immune responses, cooperation between pDCs and cDC1s and the resulting synergistic effects dependent on soluble factors such as type I IFN and/or cell-cell contact between the two DC subsets are described (11, 47). The potent anti-tumor T cell responses induced in melanoma patients by adoptive transfer of tumor-antigen-loaded pDCs (87) could be either a result of direct priming by pDCs or via interactions with other cells, including cDCs. However, tumor infiltrating pDCs exhibit an abnormal or hypofunctional state, most likely due to immuno-suppressive effects of the TME such as TGFβ (96). The presence of pDCs in tumors is associated with poor prognosis in cancers such as breast and ovarian cancers (97, 98). pDCs are generally thought to contribute to tolerance induction and tumor promotion in this setting, most likely due to Treg induction and expression of immunosuppressive factors such as indoleamine 2,3-dioxygenase (IDO) (98, 99). Thus, the role of pDCs in shaping adaptive tumor immunity remains elusive. It likely depends on their activation status and involves cooperativity with other cells but how pDCs are activated needs further investigation.

#### cDC1-cDC2 Crosstalk

The two cDC subsets communicate not only through soluble mediators such as IL-12 but also through a third cell viz., activated CD4 T cell. Even though both cDC subsets are adept in priming naïve T cells, cDC2s are more proficient in activating CD4 T cells than CD8 T cells while cDC1s are potent activators of CD8 T cells but present antigen to CD4 T cells less efficiently, both in vitro and in vivo (8, 20, 43). However, recent literature demonstrates that robust and maximal induction of cytotoxic CD8 T cell responses against cell-associated antigens not only requires interactions with cDC1s, but also interactions involving cDC2s (77, 100). Intravital microscopy demonstrated that, in the dLN, the two cDC subsets exhibit differential localization wherein cDC1s are largely segregated to the T cell zone in deep paracortical regions and cDC2s are more peripherally distributed (78, 100–103) and that CD8 T cells cluster with cDC1s and CD4 T cells cluster with cDC2s during step one of two-step T cell priming event in anti-viral immunity (78, 100, 104), suggesting parallel activation of the two T cell subsets by two different cDCs in an asynchronous manner. Such differential localization of the cDC subsets into non-overlapping T cell regions is also reported in the spleen (105).

The peripheral DC subsets also exhibit different kinetics during their migration to dLN (106), with an implication that cDC2s might access CD4 T cells earlier. The CD4 T is cell activated in the first step of the priming process, then gets recruited to LN-resident, XCR1+ cDC1 during the second step of the priming process and delivers help signals to that cDC1. The receiver-cDC1 then transmits the help signal to CD8 T cell activated in the first step, resulting in a robust expansion of highly effective CTLs. In this regard, it is well-established that, in the absence of CD4 T cell help, CD8 T cell responses are weaker and insufficient to generate long-lasting memory (107–109). The CD4 T cell help includes molecules such as CD40L expressed on CD4 T cells, that induces expression of costimulatory molecules including CD70, CD80, CD86, and cytokines such as IL-12, IL-15 by cDC1 (66, 110–112). The molecular nature of CD4 T cell help in shaping the CTL response is recently reviewed (104) and will not be discussed here in detail. Signaling though type I IFN is critical for proper functioning of cDC1s (85) and cDC2s are one of the important sources of this cytokine, as shown by depletion of pDCs using anti-pDCA antibodies in Batf3-deficient mice (84).

cDC1s and cDC2s may also collaborate for optimal Th1 induction. In the context of leishmania infection, targeting antigen to either cDC1 or cDC2 can elicit IFNγ-producing T cells, but interestingly, the cDC2s require IL-12 produced by the cDC1s to induce Th1 responses, whereas the cDC1s induce Th1 responses via CD70, independent of IL-12 (19). Therefore, each DC subset provides different signals that can contribute to effector T cell responses. Among the activated CD4 T cells, Th1 cells excel in providing the help to cDC1s to prime and expand CTLs through of production large amounts of IFNγ (113), thus fostering an important crosstalk between the two cDCs.

The majority of the experimental data described above originates from studies in anti-viral immunity. However, where and how naïve cancer cell-specific CD4 T cells get activated in a tumor setting is less clear. Lessons learnt on the importance of MHC class II-restricted CD4 T cell responses in autoimmune pathogenesis may shed light on this question in anti-tumor responses as well, since the anti-tumor response is essentially a self-specific response (114). The highest genetic risk for autoimmunity is conferred by HLA class II genes, with odds ratios >6, suggesting that CD4 T cell responses are necessary for immunity against self. In the context of autoimmunity, although some priming in the target tissue may occur (115–117), most studies suggest that self-specific CD4 T cells are first primed in the dLN, suggesting that a similar phenomenon might be happening in the generation of an anti-tumor immune response.

#### Evidence for the Importance of Tumor-Derived cDC2s and Activation of CD4 T Cells in the Draining Lymph Node

A large body of literature shows that naïve CD8 T cell activation for the generation of anti-tumor immunity occurs in dLN and is mediated by DCs (118–121). Interestingly, requirement of CD4 T cell help for optimal CD8 T cell effector functions in the context of tumor immunity is also well-documented, including the ability of CTLs to infiltrate the tumors (8, 119, 122–127). Non-helped CD8 T cells exhibit dysfunctional state with high expression of exhaustion markers in metastatic lung tumor model (127). In this regard, it is also well-established that the TME contains both cDC1 and cDC2 subsets (43–46). But importantly, both cDC1s and cDC2s scavenge tumor antigens (44) and migrate to dLN in a CCR7-dependent manner (46). Under right conditions, cDC2s can induce CD4 T cell activation in response to cellassociated antigen (51). Consistent with this, tumor-derived and dLN-derived cDC2s stimulate CD4 T cells more efficiently, ex vivo, in Lewis lung carcinoma model expressing ova as a model antigen (43). Furthermore, in this experimental setting, while cDC1 efficiently primed CD8 T cells, cDC2s are the most efficient activators of CD4 T cells. In addition, vaccination with the activated cDC2s reduced tumor growth, similar to that observed with cDC1s (43). Delivery of tumor antigen to cDC2 using dendritic cell immunoreceptor 2 (Dcir2) leads to significant anti-tumor effects in a mouse melanoma model (128). In a lung adenocarcinoma mouse model engineered to express MHC class II-restricted cytosolic antigen, activated cDC2 are observed both in the tumor and dLN and antigen-specific naïve CD4 T are activated in the dLN (129). In breast cancer patients gene signature of cDC2s positively correlates with better survival, similar to that observed with cDC1s, (130) and MHC class II expression predicts response to anti-PD1/PD-L1 therapy in melanoma patients (131). Collectively, tumor-derived cDC2s are likely to contribute to CD4 T cell activation in the dLN.

#### Integrated Model of DC Crosstalk in Tumor Draining Lymph Node

The spatiotemporal nature of DC crosstalk suggests two distinct DC-mediated events for maximal CD8 T cell responses: one after the initial antigen capture and another after the antigen is transferred to dLN-resident cDC1 cells (8). This sequential CTL activation is demonstrated by the exclusive clustering of migratory cDC1s with CD8 T cells early on during the initiation of an antiviral immune response. Subsequent clustering of activated CD8 T cells with the LN-resident cDC1s acts as a platform for signal relay from pDCs and activated CD4 T cells (79). According to this "consecutive interaction" model (79, 112), the generation of maximal CTL response and therapeutic anti-tumor immunity requires a multicellular orchestration of events in the tumor dLN (**Figure 2**) wherein migratory cDC1s capture the antigen in tumors, migrate to the dLN and form the initial priming site to activate CD8 T cells. The activated CD8 T cells produce CCL3/CCL4 and XCL1 to mediate recruitment of CCR5+pDCs and XCR1+LN-resident cDC1s, respectively. The migratory cDC1s handoff antigen to resident cDC1s in a yet-tobe-described mechanism (44, 106). In parallel, migratory cDC2s that have captured the antigen also move from the tumor to dLN and activate CD4 T cells. The pDCs induce the maturation of newly recruited, LN-resident cDC1s and the activated CD4 T cells licenses them for superior CTL responses. The overall effect of such orchestration and functional-cooperativity of pathways between different DC subsets is the amplification of CTL responses against a given antigen, without potentially missing out on the critical help necessary for CTLs to function at their peak. In fact, vaccine-mediated induction of such coordinated efforts of multiple DC subsets is known to trigger sustained and potent CTL responses while inhibiting immunosuppressive pathways in preclinical models (132). Ex vivo analyses of individual DC subsets might fail to identify such cellular orchestration to appreciate the relative contribution of each interaction between the different DC subsets in the generation of potent immune response.

#### DC Crosstalk in Tumors in situ

Accumulating evidence suggest that cross-priming by tumorresident cDC1 in situ is also an important phenomenon in the generation of an anti-tumor immune response. Local T cell priming and activation within tumors were observed in mice that lacked LN, or when T cell recirculation was blocked (133–135). Furthermore, intratumoral cDCs are required for the tumor regression achieved with adoptively transferred T

cells in an experimental setting where migration of T cells to dLN was prevented (44). Moreover, tumor-resident cDC1s are the predominant sources of CXCL9 and CXCL10 and mediate recruitment effector T cells into the tumor (67). Similar to the events described for the dLN in the previous section, activated CD8 T cells could potentially orchestrate events in situ in the tumor where LN-like structures known as tertiary lymphoid structures (TLS) are present. A hallmark of TLS is the presence of high endothelial venules (HEVs) and expression of CCL19 and CCL21, the ligands for CCR7 (136, 137). DCs migrate in a CCR7-dependent manner (43, 45, 46, 138, 139). Moreover, well-organized TLSs contain B cell and T cells areas with mature DC subsets including cDCs and pDCs. Such organization makes TLS an ideal place to sustain proximity and the crosstalk between various subsets, and orchestrating local events required for maximal tumor immunity (135, 136). In fact, tumor-associated TLSs are functional structures capable of recruiting antigen-specific T cells and facilitating their activation through interactions with DCs (140). Interestingly, TLSs have been observed in several human tumors and their presence, particularly the ones containing high amounts of DCs and Th1 cells within the TLS, is associated with better prognosis (137, 141, 142) and increased TLS density is associated with strong infiltration of effector and memory CD8 T cells within the tumors (141), reflecting the importance of crosstalk between DC subsets and, CD4 help in increased CTL trafficking. Lung cancer patients with intratumoral CD8 T cells but no TLS had poor survival, indicating the necessity of their in situ education within the TLS for better effector functions (141, 143). In a metastatic lung tumor model, administration of TLR9 activator

leads to CD8 T cell infiltration concurrent with TLS formation. The presence of TLS in this model was completely dependent on CD4 help (127). Taken together, these data suggest that TLSs promote DC crosstalk and anti-tumor immunity. Thus, induction of TLS provides another opportunity to promote communication between DC subsets to augment the magnitude of protective immunity, particularly against neoantigens that arise during the later phases of tumor progression (121). Moreover, induction of simultaneous trafficking and activation of cDCs and pDCs, using a vaccination strategy that combined DC subset-specific adjuvants (e.g., CpG-ODN and GM-CSF) leads to local accumulation of CD8 T cells and superior anti-tumor responses (132) suggesting that, even in the absence of TLS, evoking appropriate DC-crosstalk within the tumor tissue has the potential to boost superior CTL responses than targeting a single DC subset.

#### Influence of DC Crosstalk With Other Cells in the TME on Anti-tumor Immunity

DCs can also engage with other immune cell types in the TME and lymphoid organs. Such interactions can enhance or dampen DC functions and anti-tumor immunity, depending on the cell types involved. For example, DCs interact with Treg cells, resulting in the suppression of CD8 T cell-mediated antitumor immunity (144). Two-photon laser-scanning microscopy analysis showed that Treg cells engage in prolonged physical interactions with DCs, six times longer than that of DC-CD8 T cell interaction in tumor. This extended physical contact between Treg cells and DCs results in upregulation of the immunosuppressive molecules such as IDO and lower maturation molecules on DC surface (144).

Interactions with other immune cell types such as natural killer (NK) cells with DCs can boost the immune response against tumors. It has long been established that, through the secretion of IL-12, cDC1s can license NK cells to kill tumor cells (145–147). However, recent studies have shown that NK cells can also influence DC functions in the context of tumors. In fact, NK cells produce XCL1 to recruit XCR+cDC1s to the TME (148). In addition, NK cells are one source of Flt3L within the tumor and dictate intratumoral accumulation of cDC1 cells by supporting DC survival, proliferation or development (71). Stimulation of NK cells with DC-derived factors such as IL-12, IL-15/IL-15Rα complex or contact–dependent interactions of OX40-OX40L augment NK cell functions to eliminate tumor cells (149– 151). TCGA analysis suggests that NK cell/XCL1/cDC1 axis is associated with better survival in many cancer indications (148).

DCs also interact with NKT cells, the unconventional T lymphocytes expressing a semi-invariant T cell receptor (TCR) that recognize glycolipids presented by CD1d. (152). Although CD1d can be expressed by many hematopoietic cell types, DCs constitutively express CD1d and are the most potent APCs for exogenous glycolipids (153–155). The NKT cell ligand αgalactosylceramide (α-GalCer) acts as a potent in vivo adjuvant for DCs, resulting in increased expression of MHC class II and other costimulatory molecules (155). In addition, α-GalCer presented by DCs strongly activates NKT cells through CD40/CD40L interaction to induce IFN-γ production (156). Administration of α-GalCer was efficacious in preclinical tumor models (157) but not in patients (158), most likely due to soluble α-GalCer-induced anergy of NKT cell (159). Administration of α-GalCer, either soluble or loaded in DCs, is currently being explored to enhance anti-tumor immunity (160). Endogenous glycolipids are known to activate NKT cells (161) and CD1d expression is observed on tumor cells (162). In fact, the level of CD1d expression on tumor cells dictates NKT-mediated cytotoxicity (163).

Tumor-associated macrophages, in most carcinomas, are linked to poor prognosis primarily due to their immunosuppressive phenotype (164, 165). Macrophages produce IL-10 and in turn prevent IL-12 secretion of by DCs, resulting in dampened tumor-specific CD8 T cell activation (166). Among mononuclear phagocytes, monocytederived cells (including macrophages) are found at higher frequencies in tumors compared to DCs, and a higher monocytemacrophage signature is associated with worse clinical prognosis (130, 167). These cells maintain a phenotype similar to in vitro M2 macrophages and contribute to the suppressive tumor microenvironment primarily via expression of antiinflammatory mediators such as IL-10, TGF-β and IDO. Many of these signals dampen the ability of cDCs to present antigen in an immunogenic manner (164). However, in other contexts, macrophages can be inflammatory and effective APCs for eliciting T cell responses (168, 169). Thus, with the addition of the right signals, tumor macrophages have the potential to contribute to anti-tumor immunity.

Additionally, even though B cells have been described to play varied and often contrasting roles in the contexts of tumor immunity, emerging evidence suggests that B cells may also contribute to tumor immunity, both via antibody-mediated effects and by acting as APCs (170–172). Specifically, in terms of the crosstalk, DCs engage with B cells to promote their growth and differentiation, resulting in the production of antibodies. pDCs, through type I IFN production, can increase TLR7 expression and other activation markers on B cells (173). pDCs are specifically capable of inducing differentiation of activated B cells into Ig-secreting plasma cells through the secretion of type I IFN and IL-6 (174). Additionally, DCs dramatically enhance the secretion of IgG and IgA through the ligation of CD40 (175). B cells isolated from TLS-containing lung cancers showed significant antibody response against many TAAs (143, 176).

Finally, DC crosstalk with cancer cells has tremendous impact on the immune surveillance of the tumors. Cancer cells express several immunosuppressive factors such as PGE2, β-catenin and cytokines such as IL-10. PGE2 renders cDC1s unresponsive to XCL1 and CCL5 by downregulating XCR1 and CCR5 expression (148). β-catenin expression in cancer cells causes ATF3-mediated suppression of CCL4, the ligand for CCR5, leading to defective recruitment of cDC1 to the TME, and adversely affecting CD8 T cell priming against TAAs (177). Interestingly, PGE2 also induces the expression of β-catenin not only in tumor cells but also in stromal cells such as cancer associated fibroblasts (CAFs). CAFs respond to tumor-derived TNFα and IL-1β to secrete thymic stromal lymphopoietin (TSLP). TSLP is a strong driver of cDCs to activate Th2 CD4 T cells that are considered pro-tumorigenic (178). CAFs also produce stromal cell-derived factor 1 (SDF1) which drives cDCs toward tolerogenic DCs secreting IDO in a STAT3-dependent manner and promoting the recruitment and differentiation of Treg cells in tumors (179). However, co-targeting fibroblasts in combination with DC-based vaccine enhances the anti-tumor immune responses (180), suggesting that DC/stromal cell interactions can be manipulated to improve immunotherapies. Overall, with the property of bridging the innate and adaptive immune cells, DCs have a pivotal role in orchestrating an anti-tumor immune response by engaging interactions with many cell types within the TME.

# Potential Therapeutic Applications of Tumor DC-Crosstalk

The field of cancer immunotherapy, energized by the effect of T cell checkpoint inhibitors (CPI) in some patients, is beginning to focus on ways to treat "cold" tumors that lack T cells which can be activated with an anti-PD1 or other CPI. There is a large unmet medical need to increase the proportion of patients who respond to immunotherapy. Enhancing innate immunity, and DC function in particular, is one way to make tumors "warmer" that has tremendous potential. To date, most cell-based DC cancer therapies have utilized moDCs and have shown limited efficacy (121, 181, 182). With our current knowledge of both the importance of cross-presenting cDC1s for tumor immunity and the plasticity of monocytederived cells, moDCs are likely not the best cell type to use for inducing optimal clinical outcomes against cancer. Most studies show that moDCs have limited capacity for both cross-presentation and migration to draining LN compared to Batf3-dependent cells (43, 183). In addition, most monocytederived cells in the TME are immunosuppressive, and even if ex vivo moDCs can be activated to sustain cDC1-like properties, these are not likely maintained in the TME (121, 181). Therefore, moDC-based vaccines may not be the answer, and a new generation of DC-focused cancer immunotherapies are needed.

Increasing cDC1 function is one important goal, but as described here, some of this can occur indirectly via the cooperative interactions with other cells. In addition, both cDCs and pDCs have the potential to directly activate T cells that can kill cancer cells if exposed to the right activating signals (**Figure 1B**). Therefore, targeting maturation signals specifically to just cDC1s may not be the optimal therapy, and delivering signals that can enhance the function of all DC subsets may enhance efficacy and durability. For example, although tumor pDCs often correlate with poor prognosis, they are the most efficient producers of type 1 IFN and have the capacity for sustained MHC class II expression; these functions together may inflame the tumor and elicit strong T cell help that in turn could

#### REFERENCES


be sustained by newly matured cDC1s. Therefore, identifying signals that target and activate all DC subsets, and the cells that crosstalk with them will help provide novel insights into the cellular and molecular nature of tumor-specific CTL priming. The goal is to design therapies that build a site of sustained, immunostimulatory tumor-antigen presentation and increase the magnitude of anti-tumor immunity, so we can successfully treat a broader set of patients.

#### AUTHOR CONTRIBUTIONS

RN, SM-M, and KT designed and wrote the manuscript.

#### FUNDING

The authors, employees of Amgen Inc., declare that this study received funding from Amgen Inc. in its entirety. Amgen Inc. did not have a role in the study design.

#### ACKNOWLEDGMENTS

We thank Carlos Briseno, Michael Gonzalez, and Jackson G. Egen for critical reading of the manuscript and helpful discussions.

plasmacytoid DCs by repressing Id2. J Exp Med. (2016) 213:897–911. doi: 10.1084/jem.20151715


crosspresentation selectively in CD8(+) dendritic cells. Immunity. (2009) 30:544–55. doi: 10.1016/j.immuni.2009.01.013


antiviral NK and CD8<sup>+</sup> T cell accrual. Immunity. (2010) 33:955–66. doi: 10.1016/j.immuni.2010.11.020


cells in tumor eradication in vivo. J Exp Med. (1999) 190:617–27. doi: 10.1084/jem.190.5.617


Proc Natl Acad Sci USA. (2004) 101:16606–11. doi: 10.1073/pnas.040 7522101


**Conflict of Interest Statement:** RN, SM-M, and KT are full-time employees of Amgen Inc.

Copyright © 2019 Noubade, Majri-Morrison and Tarbell. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Current Concepts on 6-sulfo LacNAc Expressing Monocytes (slanMo)

Fareed Ahmad<sup>1</sup> , Thomas Döbel 1,2, Marc Schmitz 3,4 and Knut Schäkel <sup>1</sup> \*

<sup>1</sup> Department of Dermatology, Heidelberg University Hospital, Heidelberg, Germany, <sup>2</sup> Dermatology Branch, National Institute of Arthritis and Musculoskeletal and Skin Diseases, Bethesda, MD, United States, <sup>3</sup> Institute of Immunology, Faculty of Medicine Carl Gustav Carus, Technische Universtät Dresden, Dresden, Germany, <sup>4</sup> Partner Site Dresden, National Center for Tumor Diseases (NCT), Dresden, Germany

The human mononuclear phagocytes system consists of dendritic cells (DCs), monocytes, and macrophages having different functions in bridging innate and adaptive immunity. Among the heterogeneous population of monocytes the cell surface marker slan (6-sulfo LacNAc) identifies a specific subset of human CD14<sup>−</sup> CD16<sup>+</sup> non-classical monocytes, called slan<sup>+</sup> monocytes (slanMo). In this review we discuss the identity and functions of slanMo, their contributions to immune surveillance by pro-inflammatory cytokine production, and cross talk with T cells and NK cells. We also consider the role of slanMo in the regulation of chronic inflammatory diseases and cancer. Finally, we highlight unresolved questions that should be the focus of future research.

psoriasis

#### Edited by:

Silvia Beatriz Boscardin, University of São Paulo, Brazil

#### Reviewed by:

Elizabeth Mellins, Stanford University, United States Pieter J. M. Leenen, Erasmus University Rotterdam, Netherlands

\*Correspondence: Knut Schäkel knut.schaekel@med.uni-heidelberg.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

Received: 15 October 2018 Accepted: 12 April 2019 Published: 22 May 2019

#### Citation:

Ahmad F, Döbel T, Schmitz M and Schäkel K (2019) Current Concepts on 6-sulfo LacNAc Expressing Monocytes (slanMo). Front. Immunol. 10:948. doi: 10.3389/fimmu.2019.00948 Keywords: slan<sup>+</sup> monocytes, slanMo, non-classical monocytes, inflammation, autoimmunity, cancer, infection,

# CD16<sup>+</sup> MONOCYTES

Monocytes are important regulatory cells in innate and adaptive immunity (1, 2). Studies on blood leukocytes showed that monocytes are a heterogeneous cell population that can be roughly separated into three populations: classical monocytes CD14+CD16−, intermediate monocytes CD14+CD16+, and non-classical monocytes CD14−CD16<sup>+</sup> (1, 3–5). The murine counter part of non-classical monocytes was identified as Ly6ClowCCR2−CX3CR1hi cells (4, 6, 7). So far the most distinctive and best-studied function of mouse non-classical monocytes is their migration independent of the direction of blood flow along the luminal side of the vascular endothelium (8–10). There, they function in immune surveillance of the vasculature and exert both anti-inflammatory and pro-inflammatory functions. Therefore, they are also called patrolling monocytes (8). Patrolling behavior is a common feature of both murine and human non-classical monocytes (8, 9, 11, 12). However, murine non-classical monocytes are currently considered to be cells of vascular homeostasis, while the majority of studies describe an overall pro-inflammatory function of human non-classical monocytes (9, 11, 13, 14). The pro-inflammatory function of human non-classical monocytes is mainly attributed to the production of TNF-α and IL-12 (6, 11, 15–17). Concerning the origin of non-classical monocytes, there is now evidence from studies in mice and humans that classical monocytes give rise to non-classical monocytes (18, 19). The transcription factor Nur77 (NR4A1) is upregulated in human and murine non-classical monocytes (17, 19, 20). Mice, having a deletion in the NR4A1 super enhancer, lack non-classical monocytes and serve as a model to study their function in vivo (10).

In the absence of specific markers, studies on human CD16<sup>+</sup> monocytes are largely descriptive, and rely on CD14/CD16-gating strategies with no clear-cut definition. Numbers of intermediate monocytes and non-classical monocytes in blood were found altered under various conditions (15, 21–25). These studies are confined to blood leukocytes, as there is no stable expression of CD14 and CD16 on non-classical monocytes entering into tissues or differentiating into macrophages and DCs.

Within human CD14−CD16<sup>+</sup> non-classical monocytes, our group defined a 6-sulfo LacNAc (slan) expressing cell population (slanMo) in peripheral blood (16, 26, 27). Subsequently, slan expressing cells have been identified in tissues (16, 28–31). Therefore, the stably expressed slan antigen provides a unique opportunity to study these cells in different organs.

# IDENTITY OF slanMo EXPRESSING CELLS

slanMo research began in 1998 when a CD16<sup>+</sup> cell population accounting for 50% of non-classical monocytes was defined by the mAb M-DC8 (32, 33). The mAb M-DC8 (IgM) was generated by immunizing mice with peripheral blood mononuclear cells (PBMCs), depleted of CD14<sup>+</sup> monocytes, T cells and B cells (33). DD1 and DD2 (IgM, generated by immunization with slanMo) are additional slan-specific mAbs that allowed for the detection of slan<sup>+</sup> cells in paraffin-embedded tissue sections (30, 31, 34). slanMo specifically express the eponymous "slan" antigen (6-sulfo LacNAc), an O-linked glycosylated variant of P-selectin glycoprotein ligand-1 (PSGL-1) (25, 30). At the molecular level, the slan-antigen is a non-sialylated and nonfucosylated 6-sulfated N-acetyllactosamine (LacNAc) (26). This is in contrast to the cutaneous lymphocyte-associated antigen (CLA, also known as sialyl 6-sulfo Lewis X), which is a sialylated and fucosylated variant of 6-sulfo LacNAc. While CLA binds to E-selectin and thereby facilitates skin homing of T cells, slan was shown to be devoid of binding to E- and -L-selectin (35). The exact function and the binding partners of slan are unknown. However, sulfated terminal glycotopes as found in the slanantigen were shown to serve as ligands for lectins other than E- and–L-selectin, including members of the galectins and siglec families (36–41).

Transcriptomic studies on blood leukocytes clearly identified slan<sup>+</sup> cells as a subset of monocytes and accordingly they were called slanMo (4, 11, 42, 43). While being of monocyte origin, slanMo may either rapidly acquire dendritic cell functions (4, 42, 44) or differentiate into macrophages (29, 45). Their initial recognition as dendritic cells (DCs) (33) was based on their DClike phenotype with very low or undetectable levels of the classical monocytes markers CCR2, CD14, CD62L, CD11b, and CD36 as well as their function as professional antigen presenting cells as revealed by T cell stimulatory experiments (16, 30). Similarly, in skin tissue of psoriasis patients, slan<sup>+</sup> cells showed a DC-like phenotype (CD14−, CD163−) and function (IL-23p19+) (30).

slanMo purified from human tonsil tissue resembled DCs by morphology and function (28). They co-localized with T cells in tonsils and induced their proliferation several times more efficient than macrophages and similar to bona fide DCs (DC1, DC2, and pDC). In addition, peripheral blood slanMo cultured in tonsil-derived condition medium acquired the phenotype of slanMo in tonsils (28). slan<sup>+</sup> cells in lymph nodes of patients with diffuse large B-cell lymphoma, exhibited a phenotype of either immature DCs (CD163low/CD14low/CD64low/CD16low) or macrophages (CD163hi/CD14hi/CD64hi/CD16hi) (29). Furthermore, in vitro studies revealed that GM-CSF and IL-4-treated slanMo can differentiate into cells with a DC-like phenotype, while IL-34-treated slanMo revealed a macrophagelike phenotype (28). Thus, slanMo may be considered as a type of circulating and tissue myeloid cell population with remarkable plasticity (28, 29, 46).

Recently, Hamers et al. defined heterogeneity within human monocytes (**Table 1**) using mass cytometry combined with single cell sequencing data (47). slanMo, but not slan-negative nonclassical monocytes, were shown to express CXCR6, which facilitated chemotactic migration toward CXCL16 (47, 48). Interestingly, CXCL16 was previously shown to be upregulated in psoriasis, lupus nephritis as well as in cardiovascular disease (47, 49–52). In line with this study describing slanMo as having phenotype and functions distinct from other nonclassical monocytes, Hofer et al. reported on a selective depletion of slan-negative CD16<sup>+</sup> cells in patients with sarcoidosis (53). Furthermore, they demonstrated a 5-fold depletion of slan-positive monocytes in patients with hereditary diffuse leukoencephalopathy with axonal spheroids (HDLS), a disease caused by macrophage colony-stimulating factor (M-CSF) receptor mutations.

# FUNCTION OF slanMo

The selective slan-marker opened the possibility for functional studies (**Figures 1**, **2**) after mAb-directed purification of slanMo. In blood, slanMo circulate as cells with low-level expression of HLA-DR and co-stimulatory molecules (26, 30, 47). They express a broad range of toll-like receptors (TLRs) but lack TLR3 and TLR9 (46). Stimulation of freshly isolated or immature slanMo with lipopolysaccharide (LPS) or CD40 ligand resulted in high-level TNF-α production (16, 26, 54). TNFα-producing slanMo were identified in psoriasis, lupus skin lesions, glomerular capillaries of lupus nephritis, and tumor draining lymph nodes (30, 31, 43, 46). Stimulating freshly isolated slanMo did not induce IL-12 or IL-23 production (11, 26, 42) however, slanMo revealed an outstanding capacity to produce IL-12 and IL-23 compared to blood monocytes and DCs, when stimulated after a brief culture period of 6 h (16, 31, 44). This functional maturation occurred when slanMo were left unseparated as in whole PBMC cultures and also after their purification by slan-directed magnetic cell sorting. The phenotypic maturation was reflected by upregulation of CD83, CD80, and HLA-DR, while CD16 was shed from the surface by activation of a disintegrin and metalloproteinase domain-containing protein 17 (ADAM17) (16). During this maturation process, expression of the slan antigen remained stable. Interestingly, maturation of slanMo could be completely prevented when erythrocytes were added to in vitro cultures of already purified slanMo. Therefore, in peripheral blood, maturation of slanMo may be tightly controlled by circulating erythrocytes (16). mAb-directed blocking experiment revealed

**Abbreviations:** PSGL1, P selectin glycoprotein ligand 1; slanMo, 6-sulfo LacNAc expressing monocytes; DCs, Dendritic cells; Ly, Lymphocyte antigen; ROS, Reactive oxygen species; HO-1, Heme oxygenase 1; PD-L1, Programmed deathligand 1; TLR, Toll like receptor; NR4A1, Nuclear receptor transcription factor 4A1; G-CSF, Granulocyte-colony stimulating factor; LPS, Lipopolysaccharides; PBMCs, Peripheral blood mononuclear cells; COPD, Chronic Obstructive Pulmonary Diseases; HGF, Hepatocyte Growth Factor.


TABLE 1 | Human monocyte heterogeneity.

\*Defined by a rather complex set of differentially expressed molecules (47). The table summarizes phenotypic and functional aspects of human monocyte heterogeneity. According to previous work and the recent study, 4 classical, 1 intermediate, and 3 non-classical monocyte populations can be defined (represented by different color code) (47). The 3 non-classical monocyte populations are identified by the differential expression of slan and CD9. General differences in phenotype and cytokine production are depicted as well. Intensity of color represent the expression level of surface marker and cytokines production from non-classical monocytes to classical monocytes and vice-versa.

that the inhibitory effect of erythrocytes depended on the expression of CD47 on erythrocytes and its binding to signalregulatory protein α (SIRPα) on slanMo (16). The in vitro findings of slanMo producing TNF-α and IL-23 are mirrored by studies on psoriasis skin lesions where 85% of dermal slanMo were found to express IL-23p19 and 50% of the cells expressed TNF-α (31). Conditioning slanMo with IFN-γ for 6 h before stimulation with LPS or R848 increased (10-fold) their IL-12 secretion acknowledging the relevance of a positive feedback loop with IFN-γ producers such as Th1 cells and NK cells (55). slanMo revealed a strong response to TLR7 and TLR8 ligands with high IL-12 and IL-23 production (31, 54). Interestingly, IL-23 production required autocrine signaling by TNF-α and IL-1β (56, 57). The responsiveness to TLR7 and TLR8 stimulation is relevant for the activation of slanMo in autoimmune diseases and psoriasis where single stranded RNA motives are either contained within autoimmune complexes or being complexed by the antimicrobial peptide LL37 as in psoriasis (31, 46).

Leeuwen-Kerkhoff et al. and Cros et al. reported a low IL-12 production and Th1 programming capacity of slanMo (11, 42). In these studies, the short maturation step through which slanMo gain their outstanding IL-12 and IL-23 producing capacity was not taken into account. In addition, some groups rely on staining of CD16 in addition to slan for the isolation of slanMo. However, CD16 cross-linking can induce an inhibitory signal (inhibitory ITAM signaling, ITAMi), reducing pro-inflammatory cytokine production (58, 59). In addition, we realized that slanMo are sensitive to flow cytometric cell sorting as they rapidly undergo apoptosis thereafter. In summary, studies revealed that slanMo circulate in blood as immature cells that readily produce TNF-α and acquire the capacity to produce IL-12 and IL-23.

#### IMMUNE CROSS-TALK OF slanMo WITH OTHER CELLS

#### T Lymphocytes

Mononuclear phagocytes largely differ in their function to regulate adaptive immune responses by directing the quality and magnitude of T cell responses (**Figure 1**). Different studies assessed the function of slanMo to stimulate T cell proliferation and direct the production of T cell derived cytokines (16, 31, 60–62). slanMo revealed a better capacity to stimulate the proliferation of allogeneic CD4<sup>+</sup> T cells than CD14<sup>+</sup> monocytes (16, 26, 31). In contrast to CD14<sup>+</sup> monocytes, slanMo efficiently primed T cells for the neoantigen keyhole limpet hemocyanin (KLH), and induced allo-antigen specific CD8<sup>+</sup> cytotoxic T cells (33). Similar to CD1c<sup>+</sup> DCs (DC2), slanMo primed naïve allogenic cord blood T cells (26). Further, slanMo demonstrated a stronger programming of Th1 cells as compared to DC2 when cultured for 6 h before stimulation with LPS and then co-culture with cord blood T cells (26, 45). This is in line with the superior IL-12 production of slanMo when compared with DC2 after 6 h of spontaneous maturation (54). Another study assessed the Th17 programming capacity using allogenic naïve T cells. Here again a higher capacity of slanMo to induce Th17/Th1 T cells was observed in cultures stimulated with slanMo instead of DC2 (31). The strong T cell stimulatory capacity of slanMo may be relevant for recall responses in peripheral tissues as well as for the priming of naïve T cells in lymphoid tissue (46).

#### Natural Killer Cells

The interaction of mononuclear phagocytes and natural killer (NK) cells is well known. The main mechanisms by which mononuclear phagocytes can activate NK cells are soluble mediators as well as through direct cell-to-cell contacts.

Co-culture of slanMo with NK cells promotes mutual activation (63, 64). Stimulation of slanMo with LPS induced an IL-12 production that stimulated NK cells to produce IFN-γ, which in a positive forward feedback loop potentiated the IL-12 production of slanMo and the IFN-γ production of NK cells. This resulted in an increased NK cell activation (CD69, NKp30, NKp44, NKG2D) as well as an increased tumordirected cytotoxicity against chronic myeloid leukemia (CML) blasts and the leukemia cell line K562 compared to those NK cells stimulated without slanMo (63, 65). This cross talk of slanMo and NK cells also improved the slanMo-mediated differentiation of näive CD4<sup>+</sup> T cells into IFN-γ producing Th1 cells (66).

Optimal reciprocal activation of slanMo and NK cells required direct cellular contact. Tufa et al. identified a cellular communication circuit through transmembrane TNF-α expressed by slanMo and its interaction with upregulated TNFR2 on NK cells leading to higher secretion of GM-CSF by NK cells (65). Similarly, ICAM-1 expressed by slanMo bound to LFA-1 on NK cells thereby promoting an enhanced IL-1β secretion by slanMo (63). Stimulation of slanMo with TLR7/8 ligands resulted in a pronounced production of TNF-α, IL-1β, IL-12, and IL-6 allowing for an improved tumor directed cytotoxicity of slanMo and NK cells (67).

#### Neutrophils

Human neutrophils were shown to directly interact with both NK cells and slanMo in vitro, which eventually enhanced the activity of both cell types—NK cells and slanMo—after LPS, IL-12 or IL-12/IL-18 stimulation (64). Neutrophils engaged with slanMo via CD18 (integrin ß2) and intracellular adhesion molecule 1 (ICAM-1) that boosted the release of IL-12 by slanMo, which further stimulated activated NK cells to produce IFN-γ. Neutrophils were also shown to interact with NK cells via CD18 and ICAM-3 thereby augmenting IFN-γ production by NK cells (68). Co-localization of slanMo, NK cell, and neutrophils in inflamed tissue of psoriasis and Crohn's disease provided evidence for cooperation between these cells in which neutrophils may function as amplifiers of immune responses mediated by slanMo and NK cells.

# Mesenchymal Stem Cells

Mesenchymal stem cells (MSCs) are well known for their immunomodulatory properties. Results from therapeutic studies are encouraging and there is hope that the application of MSCs open new options for the therapy of immune-related diseases (69). slanMo were found in increased numbers in tissues affected by chronic inflammatory diseases such as Crohn's disease, multiple sclerosis, rheumatoid arthritis, and lupus erythematosus. Treatment with MSCs was regarded successful in these diseases. Co-culture of slanMo with MSCs resulted in a reduced production of TNF-α, IL-6, and IL-12, while production of the immunosuppressive cytokine IL-10 was enhanced in

response to LPS stimulation (70). MSCs also inhibited slanMoinduced proliferation of allogeneic CD4<sup>+</sup> and CD8<sup>+</sup> T cells and dampened the polarization of naïve CD4<sup>+</sup> T lymphocytes into Th1 cells (70). In these experiments prostaglandin E2 (PGE2) was identified as a main MSC-derived immune regulatory molecule. These findings fit well with the overall function of MSCs. Other MSC-derived immunoregulatory molecules are IL-10, IL-4, TGFβ, HGF, and PDL-1, all of which act by inhibiting differentiation of autoreactive CD4<sup>+</sup> T cells into pathogenic Th1 cells by stimulating their differentiation into Th2 and Treg lymphocytes (69). These data suggest that MSCs considerably impair the immunostimulatory properties of inflammatory slanMo.

#### slanMo IN VIVO

In healthy individuals, roughly 1% of PBMCs stain positive for the slan marker (71). In healthy stem cell donors treated with granulocyte-colony stimulating factor (G-CSF), the frequency of slanMo increased from 14.9 × 10<sup>6</sup> /L to 64.0 × 10<sup>6</sup> /L. G-CSF was described to increase the numbers of tolerogenic DCs and T cells among mobilized blood leukocytes in the graft (72). In contrast, slanMo mobilized by G-CSF retained their capacity to produce IL-12 and TNF-α (73). Furthermore, G-CSF–mobilized slanMo programmed the differentiation of Th1 cells and displayed a strong capacity to stimulate the proliferation of naïve allogeneic cord blood T cells (73). Thus, slanMo transfused into recipients of allogeneic peripheral blood stem cell (PBSC) transplants are functionally fully capable and may support graft-vs. -host disease as well as graft- vs. leukemia effects.

During the first month after allogeneic stem cell transplantation slanMo showed slow reconstitution in blood compared to cDCs and pDCs (74), however, a steady increase in the frequency of slanMo has been observed in the 2nd and 3rd month after post-transplantation (75, 76). The slow reconstitution of slanMo after bone marrow transplantation as observed in this study is reflected by reports on non-classical monocytes demonstrating the same slow reconstitution in blood (18). Whether these findings argue for slanMo to develop from classical monocytes, as described for non-classical monocytes has not been addressed and requires further studies.

#### slanMo IN DISEASES

The contribution of slanMo to the immune pathogenesis of different diseases has been studied (**Table 2**) and will be discussed in the following chapter.

#### Psoriasis

Psoriasis is a chronic inflammatory skin disease with an immune response steered by IL-23 and TNF-α producing antigen presenting cells (16, 31, 80–82), thereby stimulating T cells to produce IL-17, a cytokine that is now identified to be of chief importance for inducing skin inflammation in psoriasis (83). Therapeutic responses to antibody mediated neutralization of IL-17, IL-23, and TNF-α (84, 85) underscore the role of these cytokines as predominant drivers of the disease.

slanMo have been found at increased frequencies in psoriasis skin lesions and these numbers rapidly normalized with clinically effective anti-TNF therapy (31, 84, 85). In parallel to the reduced numbers of slanMo in skin lesions their frequency in blood increased. Interestingly, these cells showed a decrease in their expression of HLA-DR (76, 85). Lesional slanMo expressed IL-23, TNF-α as well as inducible nitric oxide synthase (iNOS). The phenotype (CD1c<sup>−</sup> and CD11c+) and function (IL23+, TNF-α +, iNOS+) of slanMo in active psoriasis skin lesions corresponded to TNF-α-producing iNOS expressing (TIP)-DCs, that were earlier defined by Lowes et al. (82). In vitro slanMo demonstrated the capacity to program T cells producing IFN-γ, IL-17, IL-22 but not IL-10 (16, 31, 81). These data lend additional support to the role of slanMo as relevant stimulatory cells in psoriasis.

Autocrine TNF-α stimulation of slanMo allows for high level production of IL-12, IL-23, IL-1ß, and IL-6 (56). In accord with the general role of TNF-α as a stimulatory cytokine, treatment with the potent TNF-α-inhibitor infliximab rapidly reduced IL-12, IL-1β, and CCL20 mRNA expression in psoriasis patients (84). The migration of slanMo from the peripheral blood into psoriasis skin lesions may be facilitated by the local expression of the anaphylatoxin C5a, fractalkine (CX3CL1), and CXCL12 for which the respective receptors are expressed by slanMo (C5aR, CX3CR1, and CXCR4) (31). Self-nucleic acid complexed to the antimicrobial peptide cathelicidin (LL37) is the beststudied autologous immune stimulus in psoriasis. Stimulating slanMo with LL37-RNA-complexes induced TNF-α production at higher levels compared to DC2 (31). The cytokine production clearly places these cells on center stage for orchestrating Th17 mediated immune responses in psoriasis. As there are other slannegative antigen presenting cells producing IL-23 and TNF-α in psoriasis skin lesions, it remains to be elucidated whether slanMo have a unique and non-redundant stimulatory role in psoriasis skin inflammation. Given the selective expression of the slan on a pro-inflammatory cell type in psoriasis and other diseases, an antibody-directed targeting approach of slanMo has been developed (29, 31, 81, 86) that may have potential of serving as a new treatment option in psoriasis and other inflammatory diseases.

#### Atopic Dermatitis

Atopic dermatitis (AD) is a chronic relapsing inflammatory skin disease affecting 15–25% of children and 1–3% of adults (87, 88). The changes within the mononuclear phagocyte system in AD are complex. Inflammatory epidermal dendritic cells (IDECs) (CD1a+, Langerin−, FcεRI+) are believed to enhance local inflammation and eczema severity in AD patients (89). Higher frequencies of dermal mononuclear phagocytes expressing CD11c, CD1a, CD206, and DC-SIGN have been identified in AD patients (90). Similar to psoriasis, a higher frequency of slanMo is also reported in the dermis of active skin lesions of AD patients. These slanMo lacked expression of FcεRI, CD1a, CD14, and CD163, thereby displaying a phenotype different from already described mononuclear phagocytes in AD patients (54). Peripheral blood slanMo of these patients retained their capacity to produce inflammatory cytokines and produced more TNF-α and IL-12 than myeloid DCs or classical monocytes


TABLE 2 | The observed location and potential role of slanMo in different diseases.

iNOS- Inducible nitric oxide synthase, TLR- Toll-like receptors. ccRCC- clear cell renal cell carcinoma Ref.- References.

after LPS or R848 stimulation (54). Mental stress is a wellknown factor to trigger flares of AD (91). A standardized mental stress test in patients with AD induced an instant mobilization of slanMo into the blood circulation. Testing for their TNFα-production showed their unchanged capacity to do so (54). The mobilization of CD16<sup>+</sup> monocytes was previously shown for psoriasis patients (92). Whether this mobilization includes all CD16<sup>+</sup> monocytes or applies preferentially to slanMo has not been addressed. Non-classical monocytes are known to function as patrolling monocytes along endothelial cells. Therefore, the observed stress induced mobilization may reflect detachment of slanMo from the vasculature into blood circulation. This process was shown to be induced by a transient rise of catecholamines induced by mental stress (54).

Cytokine production of slanMo is not a fixed condition and can be modulated by micro environmental factors relevant to AD and allergic diseases. Histamine is an important regulator of allergic inflammation that modulates pro-inflammatory functions of slanMo. Different histamine receptors are expressed by slanMo, particularly the recently identified histamine H4 receptor (H4R). Histamine effectively blunted TNF-α and IL-12 production of slanMo, a reduction mediated via the H4R and the combined action of H2R and H4R (93). Hence, H4R agonists might have therapeutic potential to down-regulate immune reactions, e.g., in allergic inflammatory skin diseases (93, 94). Birch pollen contains antigens potentially inducing allergic IgEmediated sensitization. Pollen also contain immunomodulatory substances. In this context, pollen-associated E1-phytoprostanes (PPE1) were shown to license human monocyte-derived dendritic cell for T-helper type 2 (Th2) polarization of naïve T cells (95). Aqueous birch pollen extracts inhibited IL-12 production by slanMo in a dose-dependent manner, while the levels of IL-6 remained unaffected. PPE1 inhibited secretion of both IL-12p70 and IL-6. slanMo exposed to aqueous pollen extracts were impaired in eliciting an IFN-γ response in naïve CD4<sup>+</sup> T cells (95). These data demonstrated that slanMo having a constitutively high potency to induce Th1 responses, are susceptible to the Th2 polarizing effect of low molecular weight, non-protein factors derived from pollen.

#### Lupus Erythematosus

Lupus erythematosus is an autoimmune disease in which genetic and environmental factors lead to autoantibody production and induction of inflammation manifesting to multiple organs (96). The autoantibodies in lupus erythematosus patients are directed against nuclear antigens and form immune complexes containing double-stranded DNA (dsDNA) and single-stranded RNA (ssRNA) that activate DCs and drive pathogenic T cell responses (97–99). In response to ssRNA and dsDNA, plasmacytoid dendritic cells (pDCs) produce IFN-α, a critical immunoregulatory cytokine in lupus erythematosus (100, 101). slanMo were shown to lack IFN-α production but may contribute to the disease progression through high TNF-α production (46). Immunohistochemistry showed an increased frequency of slanMo in skin lesions of patients with cutaneous and systemic lupus erythematosus (46). slanMo were found scattered in the dermis where they locally expressed TNF-α. They appeared to cluster in lymph follicle-like structures where they colocalized with T cells. Incubating slanMo with serum from lupus erythematosus patients induced production of TNF-α (46). The stimulatory components of the lupus erythematosus sera are autoimmune complexes containing single-stranded RNA (ssRNA) binding to TLR7 and TLR8 or double stranded DNA binding to TLR9 (11). slanMo lack the DNA-sensor TLR9, but instead express TLR7 and TLR8 (46). In fact, ssRNA or selective TLR7 and TLR8 ligands induce TNF-α and IL-12 production in slanMo at higher levels compared to conventional dendritic cells (cDCs) or plasmacytoid dendritic cells (pDCs) (26, 31, 54).

Immune complexes binding to the vasculature frequently causes vasculitis in lupus patients (102, 103). In lupus nephritis, intracapillary accumulation of immune complexes can prime the activation of Fc receptor-bearing myeloid cells (99, 104, 105). The observation that slanMo have a CD16-mediated capacity to bind IgG-ICs (34) and to be present in lupus skin lesions (46) made us to investigate the role of IgG-ICs for the direct recruitment and activation of slanMo from the blood flow in lupus nephritis (34). Among the different types of lupus nephritis, intracapillary IC deposition and accumulation of monocytes are hallmarks of diffuse proliferative lupus nephritis class III and IV frequently leading to end stage renal disease (22, 106). The relevance of intracapillary IgG-ICs in terms of monocytes recruitment and activation, as well as the nature and function of these monocytes were not well understood. For the early focal form of lupus nephritis (class III) we demonstrated a selective accumulation of slanMo, which locally expressed TNF-α (43). In vitro and in vivo mouse studies showed that immobilized IgG-ICs induced a direct recruitment of slanMo from the microcirculation via interaction with FcγRIIIA (CD16) (43). Intravenous immunoglobulins block CD16 and completely prevented slanMo recruitment (34). Engagement of immobilized IgG-ICs by slanMo induced the production of neutrophilattracting chemokine CXCL2 as well as TNF-α, which in a forward feedback loop stimulated endothelial cells to produce the slanMo-recruiting chemokine CX3CL1 (fractalkine) (43). These studies demonstrated that expression of CD16 equips slanMo with a capacity to orchestrate early IC-induced inflammatory responses in glomeruli and identified slanMo as a pathogenic cell type in lupus nephritis.

#### Multiple Sclerosis (MS)

Multiple sclerosis is a chronic inflammatory disease of the central nervous system characterized by injury to the myelin sheath and axonal loss (107). Discussions of MS pathophysiology frequently put cells of the adaptive immune response in the spotlight. However, dendritic cells, monocytes, macrophages, and microglia, collectively referred to as mononuclear phagocytes, appear to have prominent roles in MS pathogenesis. These populations of mononuclear phagocytes function as antigen presenting effector cells in neuroinflammation (108–110). In a study on MS, slanMo were found in the patient's cerebrospinal fluid and accumulated in inflammatory brain lesions. The degree of local inflammation positively correlated with the number of slanMo (77). Recruitment of CXCR4 expressing slanMo to brain lesions may be induced by CXCL12, which was found elevated in MS patients (111, 112).

#### Crohn's Disease

Crohn's disease is characterized by patchy inflammatory lesions and affects the entire gastrointestinal tract (113, 114). In humans, intestinal lamina propria, a subset of myeloid cells HLA-DRhigh Lin<sup>−</sup> CD14<sup>+</sup> CD163low, have been identified that can enhance immunity and differentiation of Th17 cells (62). A study on slanMo revealed an increased frequency of IL-1β and TNF-α-producing slanMo in the mesenteric lymph nodes of Crohn's disease patients. slanMo accumulated in inflamed colons of Crohn's disease but not in ulcerative colitis patients (78). In parallel to the presence of slanMo in peripheral tissues, their frequency in blood circulation was reduced. Thus, slanMo may contribute to the immunopathogenesis of Crohn's disease.

#### HIV Infection

Chronic immune activation and a breakdown of the gastrointestinal mucosal barrier allow translocation of microbial products (e.g., LPS) from gut associated lymphoid tissue into the circulation (115). LPS activates monocytes and DCs that produced pro-inflammatory cytokines such as TNF-α and IL-1β (71). Increased serum TNF-α has been reported for HIV-infected individuals and is known to promote viral replication in infected CD4<sup>+</sup> T lymphocytes (116, 117). Therefore, the potential role of slanMo in fueling chronic immune activation during HIV-1 infection has been evaluated (71, 79). Dutertre et al. investigated the role of slanMo (referred to as mAb M-DC8<sup>+</sup> monocytes) in peripheral blood of HIV infected individuals (79). Specifically, they addressed chronic immune hyperactivation caused by production of TNF-α. Viremic HIV patients showed an increase in CD16<sup>+</sup> monocytes and a marked increase in slanMo (M-DC8<sup>+</sup> cells). PBMCs of viremic patients displayed an overproduction of TNF-α in response to LPS that was mostly attributed to slanMo (79). Tufa et al. reported higher relative and absolute numbers of slanMo in peripheral blood of untreated HIV infected individuals, which were activated and secrete increased amounts of IL-1β, TNF-α and IL-12 compared to healthy controls. Furthermore, the frequency of IL-1β <sup>+</sup> slanMo directly correlated with TNF-α <sup>+</sup> slanMo and viral load, suggesting virus-driven immune activation of slanMo in HIV-infected individuals (71). These data are in support of a role of slanMo in the maintenance of chronic immune activation and HIV disease progression.

#### slanMo IN CANCER

Recently, slanMo have been implicated in a novel type of immune surveillance in cancer (45). Vermi et al. demonstrated that slanMo are recruited to metastatic tumor-draining lymph nodes (M-TDLN) where they are aligned along the tumor tissue (30). The recruitment of slanMo depended on the arrival of cancer cells to M-TDLN, as slanMo were absent in unaffected lymphnodes and at primary carcinoma sites. Within M-TDLN, slanMo were found adjacent to dead cells where they phagocytosed tumor cells (30). These slanMo expressed HLA-DR, CD40 and TNF-α. More importantly, unlike pDCs from the same patient cohort, circulating slanMo from patients with advanced colorectal cancer remained substantially intact in terms of numbers, cytokine production (TNF-α and IL-12p70) and induction of T-cell proliferation (62). Thus, in contrast to other mononuclear phagocytes these data suggested that circulating slanMo are not developmentally or functionally hijacked or converted into immunosuppressive cells by growing tumors.

A study on diffuse large B cell lymphoma highlighted slanMo as prominent effectors of antibody-mediated tumor cell targeting (29). slanMo from these patients showed an effective rituximab-mediated antibody dependent cell-mediated cellular cytotoxicity (ADCC) slightly lower when compared with the one displayed by NK cells. Moreover, slanMo acquired a macrophage-like phenotype and became very efficient in rituximab-mediated antibody dependent cellular phagocytosis (29). Previous studies identified the critical role of CD16 in slanMo mediated ADCC (118).

In multiple myeloma, numbers of circulating slanMo significantly reduced compared to healthy controls (119). Stimulation of bone marrow or peripheral blood from multiple myeloma patients with TLR7/8 ligand (R848) showed a reduced IL-12 production by slanMo. Further co-culture of slanMo with a multiple myeloma cell line or cells isolated from patients revealed a phenotypic shift of slanMo toward intermediate monocytes and these cells demonstrated a reduced capacity to induce T cell immune responses (119). In the tumor tissue of renal cell carcinoma, an increased number of slanMo have been reported, where they produced IL-10 and revealed a macrophage like phenotype (45).

Taken together, these studies identify different roles for slanMo in cancer. slanMo may be helpful by stimulating tumor specific T cells responses (32) and by conducting a tumordirected cytotoxicity (ADCC) (29). On the other hand, slanMo can differentiate into cells that are part of a tolerogenic immune response (29, 30, 46). Thus, in cancer slanMo seem to display a remarkable functional plasticity.

#### CONTROLLING THE PRO-INFLAMMATORY FUNCTION OF slanMo

In this chapter, we discuss studies investigating how the immune related function of slanMo is modulated by several common therapeutics that are applied for the treatment of chronic inflammatory diseases or cancer.

#### PDE4-Inhibitor

A new option for the treatment of psoriasis and psoriasis arthritis is the phosphodiesterase 4 (PDE4)-inhibitor apremilast. PDE4-inhibitors increase intracellular cAMP levels and were shown to attenuate pro-inflammatory functions in different cell types and diseases (120, 121). Apremilast is currently tested in a phase III trial in Behçet's disease and is under study in a number of other inflammatory diseases (122, 123). Previous studies demonstrated that apremilast could reduce the production of GM-CSF, IL-12p70, TNF-α, and IFN-γ while increasing the production of IL-10 and IL-6 in LPS-stimulated PBMCs (124). Studies on ultraviolet B-irradiated keratinocytes showed a reduced production of TNF-α when cultured in the presence of apremilast while skin fibroblasts exhibited a reduced migratory capacity (125). Inhibition of PDE4 in slanMo reduced IL-12 and TNF-α production while this treatment enhanced their IL-23 production. As a consequence, apremilast-treated slanMo showed a reduced induction of Th1 cells while at the same time sustaining Th17 responses. A strong Th17-promoting effect of a drug that is effective in the treatment of an IL-17-mediated disease is unexpected. The enhanced IL-23p19 production in response to PDE4-inhibition can be explained by cAMP-dependent activation of protein kinase A and subsequent phosphorylation of the cAMP-response element binding protein (CREB) (126). Recently, the PDE4 inhibitor roflumilast licensed for the treatment of COPD was studied in mouse DCs generated in vitro from bone marrow precursor cells (127). In line with our study, these authors also demonstrated a PDE-4-inhibitor induced production of IL-23 in DCs and of IL-17 in T cells. Therefore, PDE4-inhibitors possibly exert their good therapeutic effects through modulation of functions on other immune and non-immune cells.

#### Dimethylfumarate

Dimethylfumarate (DMF) is a small molecule licensed for the treatment of psoriasis and multiple sclerosis (128). Skin lesions of psoriasis patients treated with DMF (in combination with monomethylfumarate—fumaderm <sup>R</sup> ) showed a reduced frequency of slanMo. Studying the function of slanMo in the presence of DMF demonstrated an inhibition of CX3CL1- and C5a-induced migration of slanMo. Both, CX3CL1- as well as C5a are expressed in psoriasis plaques. DMF also attenuated the rapid spontaneous phenotypic maturation of slanMo, as judged by reduced expression of CD80, CD86, CD83, and HLA-DR (124). In addition, slanMo showed a DMF-dependent decrease in the production of IL-23, IL-12, TNF-α, and IL-10, and a reduced capacity to stimulate Th17/Th1 responses. At the level of intracellular signaling, DMF-treated slanMo showed an increased expression of heme oxygenase 1 (HO-1) (124). HO-1 is an enzyme with import antioxidant, anti-inflammatory, and cytoprotective functions. Treatment of slanMo with DMF also inhibited phosphorylation of NFκB p65. This may directly affect IL-12p70 transcription as NFκB p65 binding sites were found within the IL-23p19 promoter (124). Moreover, the observed DMF-dependent reduction in STAT1 phosphorylation would explain the reduced IL-12/IL-23 production of DMF-treated slanMo, as STAT1 signaling is essential in this respect (124, 129). Collectively, these findings demonstrated that slanMo found in psoriasis as well as in MS are a relevant target for the therapeutic immunomodulatory effects of DMF.

#### Chemotherapeutic Agent

Treatment of cancer with chemotherapeutic agents remains a challenge for immunological researcher. An ideal therapy should target the proliferation of cancer cells and leave the function of tumor-directed immune responses intact. In a study, comparing different chemotherapeutic agents for their in vitro capacity to modulate the function of slanMo, mitomycin-c, methotrexate, and paclitaxel have no influence on the ability of slanMo to secrete pro-inflammatory cytokines. The ability of treated slanMo to activate T lymphocytes and NK cells also remained intact (130). These observations provided arguments of slanMo contributing to tumor cell elimination in patients treated with respective drugs. However, in this context, doxorubicin and vinblastine significantly impaired production of TNF-α, IL-12, and IL-6 by slanMo (130). Both drugs also inhibited slanMomediated T cell proliferation and suppressed their ability to stimulate NK cells.

Bortezomib is an efficient targeted form of chemotherapy for treatment of multiple myeloma (131). The anti-tumor activity of bortezomib is mediated by proteasome inhibition, leading to NFκB inhibition, decreased cell proliferation and induction of apoptosis (132). Bortezomib mediated proteasome inhibition efficiently impaired in vitro maturation of slanMo as well as release of TNF-α and IL-12 upon LPS stimulation. In addition, it also inhibited slanMo-mediated proliferation and differentiation of CD4<sup>+</sup> T cells. Furthermore, bortezomib impaired the ability of slanMo to stimulate IFN-γ secretion and tumor-directed cytotoxicity of NK cells (133).

#### CONCLUSIONS AND PERSPECTIVES

Many studies have contributed to the current understanding of slanMo as cells with a pronounced potential to stimulate innate and adaptive immune responses. slanMo appear similar but not identical to non-classical monocytes with some genes such as CXCR6 being differentially expressed. In the case of the most obvious difference, namely the selective expression of the slanantigen, the function remains to be determined. Other open questions regard the precursor cells of slanMo and the signals guiding their development.

In regard to their likely function as patrolling monocytes, slanMo may play an unexplored role in immune surveillance of the vasculature. Our recent study in immune complex induced lupus nephritis would be in line with this task. Here slanMo were shown to play an important role for the initiation of the immune complex induced immune response in the glomerular capillaries. Whether this early stimulatory function during the beginning of an immune response holds true also for other inflammatory diseases such as psoriasis is an open question.

#### REFERENCES


Additional investigations to the function of slanMo in cancer are to be awaited and appear relevant asin vitro studies demonstrated their effective tumor-directed cytotoxicity and stimulation of tumor-directed immune responses.

Overall, there is a high interest in gaining a better understanding of the function of slanMo and slan-negative non-classical monocytes. Further in-depth studies of slanMo can be highly informative for understanding immunopathology and provide an attractive target for therapeutic intervention.

#### AUTHOR CONTRIBUTIONS

FA, KS, MS, and TD have written and edited the manuscript.

#### ACKNOWLEDGMENTS

We thank to Steffi Oehrl, Ramona Geske, Lukas Freund, Felix Funck, and Hao Zhang for their suggestions and critical discussion. This work was funded by the Deutsche Forschungsgemeinschaft (DFG, German Research Foundation) to KS—SFB TRR 156, SCHA 1693/1-1, and project number 259332240/RTG 2099.

viruses via TLR7 and TLR8 receptors. Immunity. (2010) 33:375–86. doi: 10.1016/j.immuni.2010.08.012


of human 6-sulfo LacNAc+ (slan) dendritic cells. Int J Cancer. (2013) 132:1351–9. doi: 10.1002/ijc.27786


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Ahmad, Döbel, Schmitz and Schäkel. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Co- but not Sequential Infection of DCs Boosts Their HIV-Specific CTL-Stimulatory Capacity

Manuela Schönfeld1†, Ulla Knackmuss 1†, Parul Chandorkar <sup>1</sup> , Paul Hörtnagl <sup>2</sup> , Thomas John Hope<sup>3</sup> , Arnaud Moris 4,5, Rosa Bellmann-Weiler <sup>6</sup> , Cornelia Lass-Flörl <sup>1</sup> , Wilfried Posch<sup>1</sup> \* and Doris Wilflingseder <sup>1</sup> \*

#### Edited by:

Diana Dudziak, Hautklinik, Universitätsklinikum Erlangen, Germany

#### Reviewed by:

Quentin James Sattentau, University of Oxford, United Kingdom Morgane Bomsel, Institut National de la Santé et de la Recherche Médicale (INSERM), France

#### \*Correspondence:

Doris Wilflingseder doris.wilflingseder@i-med.ac.at Wilfried Posch wilfried.posch@i-med.ac.at

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

> Received: 23 July 2018 Accepted: 02 May 2019 Published: 24 May 2019

#### Citation:

Schönfeld M, Knackmuss U, Chandorkar P, Hörtnagl P, Hope TJ, Moris A, Bellmann-Weiler R, Lass-Flörl C, Posch W and Wilflingseder D (2019) Co- but not Sequential Infection of DCs Boosts Their HIV-Specific CTL-Stimulatory Capacity. Front. Immunol. 10:1123. doi: 10.3389/fimmu.2019.01123 <sup>1</sup> Division of Hygiene and Medical Microbiology, Medical University of Innsbruck, Innsbruck, Austria, <sup>2</sup> Central Institute for Blood Transfusion and Immunological Department, Medical University of Innsbruck, Innsbruck, Austria, <sup>3</sup> Department of Cell and Molecular Biology, Feinberg School of Medicine, Northwestern University, Chicago, IL, United States, <sup>4</sup> Sorbonne Université, INSERM, CNRS, Center for Immunology and Microbial Infections - CIMI-Paris, Paris, France, <sup>5</sup> Institute for Integrative Biology of the Cell (I2BC), CEA, CNRS, Université Paris-Sud, Université Paris-Saclay, Gif-sur-Yvette, France, <sup>6</sup> Department of Internal Medicine II, Medical University of Innsbruck, Innsbruck, Austria

Pathogenic bacteria and their microbial products activate dendritic cells (DCs) at mucosal surfaces during sexually transmitted infections (STIs) and therefore might also differently shape DC functions during co-infection with HIV-1. We recently illustrated that complement (C) coating of HIV-1 (HIV-C), as primarily found during the acute phase of infection before appearance of HIV-specific antibodies, by-passed SAMHD1-mediated restriction in DCs and therefore mediated an increased DC activation and antiviral capacity. To determine whether the superior antiviral effects of HIV-C-exposed DCs also apply during STIs, we developed a co-infection model in which DCs were infected with Chlamydia spp. simultaneously (HIV-C/Chlam-DCs or HIV/Chlam-DCs) or a sequential infection model, where DCs were exposed to Chlamydia for 3 or 24 h (Chlam-DCs) followed by HIV-1 infection. Co-infection of DCs with HIV-1 and Chlamydia significantly boosted the CTL-stimulatory capacity compared to HIV-1-loaded iDCs and this boost was independent on the opsonization pattern. This effect was lost in the sequential infection model, when opsonized HIV-1 was added delayed to Chlamydia-loaded DCs. The reduction in the CTL-stimulatory capacity of Chlam-DCs was not due to lower HIV-1 binding or infection compared to iDCs or HIV-C/Chlam-DCs, but due to altered fusion and internalization mechanisms within DCs. The CTL-stimulatory capacity of HIV-C in Chlam-DCs correlated with significantly reduced viral fusion compared to iDCs and HIV-C/Chlam-DCs and illustrated considerably increased numbers of HIV-C-containing vacuoles than iDCs. The data indicate that Chlamydia co-infection of DCs mediates a transient boost of their HIV-specific CTL-stimulatory and antiviral capacity, while in the sequential infection model this is reversed and associated with hazard to the host.

Keywords: HIV-1, STIs, dendritic cell, complement, CTL

# INTRODUCTION

Dendritic cells (DCs) play a pivotal role in the defense against invading pathogens. DCs reside in the peripheral tissue, where they capture antigens and transport them to lymph nodes to present them to naive T cells. Hence, DCs play a key role in shaping the adaptive immune response. Of all new HIV-1 infections, 60–90 % are caused by sexual transmission (1, 2). Since HIV-1 transmission occurs at mucosal surfaces, DCs are amongst the first cells to encounter the virus (3). At the same time, HIV-1 spontaneously activates the classical complement (C) pathway (4), even in seminal fluid (5), through direct binding of C1q to the viral surface. Therefore, C-opsonized HIV (HIV-C) is accumulating at mucosal sites during early HIV-1 infection (6, 7).

We have previously shown that HIV-C interacts with complement receptors 3 (CR3) and 4 (CR4) on iDCs, whereas non-opsonized HIV binds DCs via gp120 to DC-SIGN (8) and via CD169 (Siglec-1) binding to virions. Furthermore, iDCs were efficiently infected with HIV-C compared to non- or antibody-opsonized HIV (7, 9). HIV-C was able to bypass SAMHD1 restriction in DCs, an intrinsic cellular defense mechanism, which usually inhibits HIV-1 replication in myeloid cells. Thus, complement opsonization of the virus counteracted viral defenses in DCs. DCs exposed to HIV-C had a significantly higher maturation and costimulatory capacity compared to DCs exposed to non-opsonized HIV (9).

In general, efficiency of HIV-1 transmission is low (10). However, it is known that viral and bacterial genital infections that cause inflammation or ulcers increase risk of infection and/or susceptibility to HIV transmission (10). Epidemiological studies also revealed a link between an increased incidence of STIs with increased efficiency to transmit the virus to an uninfected partner (11). Among the STIs most commonly associated with high genital HIV loads are Gardnerella vaginalis (12, 13) associated with bacterial vaginosis (BV), herpes simplex virus type 2 (HSV-2), Chlamydia trachomatis, Neisseria gonorrheae, and Trichomonas vaginalis (10). Dendritic cells incubated with mucosal fluid from women with BV were found to up-regulate maturation and activation markers like HLA-DR, CD40, and CD83, and to have an increased T cellstimulatory capacity indicating an impact on mucosal immunity (14). To determine if model pathogenic bacteria could similarly pereturb the complement-mediated avoidance of antiviral effects when DCs are exposed to bacteria, we added Chlamydia and opsonized HIV-1 either simultaneously mimicking a coinfection (HIV-C/Chlam-DCs) or by delayed addition of HIV-C (Chlam-DCs). Chlamydia (C.) trachomatis are gram-negative obligate intracellular bacteria and a primary agent causing nongonococcal urethritis (15). During infection of cells within the vaginal mucosa, C. trachomatis initiates disruption of the mucosal-epithelial layer allowing better tissue entry of HIV-1 (10). Immunological alterations due to the presence of C. trachomatis may further support the transmission of HIV to susceptible cells or impact the antigen-presenting capacity of DCs (10).

Given that infection of iDCs is modulated by the opsonization pattern of HIV-1, which also had an impact on outcomes of both humoral and cellular antiviral immune responses (9, 16, 17) and given that HIV-1 particles are opsonized in vivo (18) and in vitro (4, 5), we analyzed whether the presence of Chlamydia modulates DC properties and function during co-infection with HIV-C.

# MATERIALS AND METHODS

#### Ethics Statement

This study was carried out in accordance with the recommendations of the Ethics Committee of the Medical University of Innsbruck. The protocol was approved by the Ethics Committee of the Medical University of Innsbruck. All subjects gave written informed consent in accordance with the Declaration of Helsinki.

## Generation of Human Monocyte-Derived iDCs and mDCs

Monocytes were isolated from whole blood of healthy donors by using CD14 BD IMAG Beads (Becton-Dickinson), according to the manufacturer's instructions. Differentiation into iDCs was done using IL-4 (200 U/ml) and GM-CSF (1,000 U/ml) and the iDC phenotype was routinely confirmed on day 5 by flow cytometry using CD11b, CD11c, DC-SIGN, HLA-DR, and CD83 (9, 16, 19). Representative histogram plots of these markers on iDCs are illustrated in **Figure S1** (upper panel, red; isotype, blue). To generate LPS-DCs, day 5 iDCs were stimulated for 24 h with 100 ng/ml pure LPS-EB (Sigma) prior to HIV-1 infection.

Acute and chronic Chlamydia exposure was mimicked by stimulation of day 5 iDCs with infectious or heat-inactivated Chlamydia for either 24 h prior to (Chlam-DCs) or at the same time (HIV-C/Chlam-DCs or HIV/Chlam-DCs) as HIV-1 infection. For first experiments (DC maturation, binding, internalization) infectious as well as heat-inactivated bacteria were used. Since no differences were observed and since we intended to study PAMP-associated changes in DCs induced by Chlamydia, for all other analyses we used heat-inactivated bacteria. A representative histogram plot of CD83 expression on iDCs (red), DCs treated with heat-inactivated Chlamydia (dark green) or live Chlamydia (light green) is depicted in **Figure S1**, lower panel. Since isotype controls between the conditions did not differ, the iDC isotype control is shown (**Figure S1**, lower panel).

#### Bacteria

Chlamydia spp. propagated in human epithelial HL cells and aliquots of purified bacteria were stored in sucrose phosphate glutamic acid at −80◦C until use (20). For quantification of infection, coverslips overlaid with HL-cell-monolayers were fixed in methanol and stained with FITC-conjugated anti-Chlamydia LPS monoclonal antibody (OXOID (Ely) Ltd., Ely, UK). Chlamydial inclusion bodies within cells were counted by fluorescence microscopy at a magnification of x100 with a Scope A1 microscope (Zeiss). For experiments using infectious, purified Chlamydia cells were infected at a multiplicity of infection/MOI of 10 as described earlier (20). An aliquot of tested, purified Chlamydia spp. was used for heat-inactivation at 70◦C for 20 min.

#### Plasmids

The infectious R5-tropic HIV-1 proviral clone R9Bal was used for maturation, binding/internalization and DC infection studies. For HIV-1 fusion assays the R9Bal and vpr/β-lam expression constructs were used to generate chimeric R9Bal/β-lam proviral clones. Confocal microscopic analyses and HC/HT imaging analyses were performed by using chimeric R9Bal/mCherry virus preparations originating from R9Bal and vpr/m-Cherry expression plasmids (21).

### Virus Production

HIV-1 proviral clones were produced by transfecting HEK293T cells. R9Bal/β-lam and R9Bal/mCherry virus stocks were prepared by co-transfection of HEK293T cells with the proviral R9Bal DNA and the vpr-β-lam or vpr-mCherry expression constructs (9). Freshly produced virus was obtained via ultracentrifugation (70,000 × g/90 min/4◦C). Concentration of the ultracentrifuged virus was measured by p24 ELISA (22) and viral infectivity was confirmed by the determination of the TCID<sup>50</sup> using PHA/IL-2-stimulated PBLs. To monitor productive infection of DCs or DC/T cell co-cultures, p24 ELISA was used.

# Opsonization of Viral Stocks

Viruses were opsonized by incubation with normal human serum (NHS) as a complement (C) source in a 1:10 dilution for 1 h at 37◦C (HIV-C). As negative control, the viruses were incubated under the same conditions in medium, which reflects non-opsonized HIV-1 (HIV). After opsonization, viruses were thoroughly washed, pelleted by ultracentrifugation (25,000 × g/90 min/4◦C) and re-suspended in RPMI medium. The opsonization pattern was determined by virus capture assay (VCA) as described (7). 96-well high-binding plates were coated using anti-human C3c, C3d, or IgG antibodies. Mouse IgG antibody was used as a control for background binding. Plates were then incubated overnight at 4◦C with the differentially opsonized virus preparations (10 ng p24/well). After extensive washing, virus was lysed and p24 ELISA was performed.

### Capture of HIV-1

Differentially matured DCs (1 × 10<sup>5</sup> cells/well/100 µl) were exposed to 25 ng p24/ml of R5 tropic non-opsonized (R9Bal) or complement-opsonized (R9Bal-C) HIV-1. After 6 h incubation at either 4◦C for binding or 37◦C for internalization, cells were washed 4 times to remove unbound virus. Cell pellets were lysed with 2% Igepal and viral amount was assessed by p24 ELISA.

# Viral Fusion Assay

DCs were plated into 96-well plates (1 × 10<sup>5</sup> cells/well/100 µl) and infected with 250 ng p24/ml non-opsonized or opsonized R9Bal/β-lam. After 5 h incubation cells were washed and loaded for 1 h with CCF2-AM substrate solution according to the manufacturer's instructions (LiveBLAzerTM FRET-B/G Loading Kit with CCF2-AM, LifeTechnologies). Cells were washed again and developed for 16 h in CO2-independent medium (Gibco) containing 10% FCS and 2.5 mM probenicid. Cleavage of CCF2 was analyzed by flow cytometry after fixation of DCs in 4% paraformaldehyde.

#### Microscopy

To visualize intracellular HIV-1 localization by confocal microscopy, iDCs, HIV-C or HIV/Chlam, Chlam- and LPS-DCs were plated onto Poly-L-lysine (Sigma)-coated coverslips and exposed to R9Bal/mCherry or –GFP (350 ng p24/ml) for 24 h. For HC/HT screening analyses, various matured DCs (50,000/well) were seeded in CellCarrier Ultra plates (Perkin Elmer) and infected over night with fluorescently labeled HIV-C (350 ng p24/ml). DCs were fixed with 4% paraformaldehyde, labeled using Hoechst 33342 (Cell Signaling Technologies), permeabilized (Permeabilization Wash Buffer, BioLegend), and stained with HLA-DR (BioLegend). Following staining, cells were washed and mounted (confocal microscopy) or re-suspended in D-PBS (HC/HT Screening). Confocal microscopy was performed on a Leica SP5 (Leica Microsystems) using a glycerol objective. Images were analyzed using LAS AF Lite (Leica Microsystems) and Fiji (ImageJ). For 3-D-rendered stacks, Imaris (Bitplane) was used. HC/HT analyses were performed using an Operetta CLSTM (Perkin Elmer) and co-localization of mCherry/HLA-DR or GFP/Siglec-1 automatically quantified using the HarmonyTM Software and RMS Spot Analyses (Perkin Elmer). For these automated analyses, first fluorescence intensities were measured, since if HIV particles are in the cytoplasm, the fluorescence intensities are significantly lower compared to packed virus in endosomes. Lower intensities can then be excluded from the automatic screening process. Then co-localization of virus particles with HLA-DR, which is in endosomal compartments only, was measured.

#### DC Infection

Day 6 iDCs, HIV-C or HIV/Chlam-DCs, Chlam-DCs and LPS-DCs (1 × 10<sup>5</sup> cells/well/100 µl) were infected in triplicate with 25 ng p24/ml of R9Bal or R9Bal-C. After 24 h incubation, DCs were thoroughly washed and cultured at 37◦C and 5 % CO<sup>2</sup> for 15 days. For co-infection experiments, autologous CD4<sup>+</sup> T cells were added to washed DCs the day after HIVinfection. After several days post-infection, supernatants were taken and diluted 1:10 with 2% Igepal to lyse the virus. Productive infection was determined by measuring p24 concentrations in the supernatant.

# Interferon-γ Elispot

SL9 clone 2, a HIV-specific CD8<sup>+</sup> CTL clone, was derived from an HIV-infected patient and recognizes the well-characterized immune-dominant epitope of Gag p17 SLYNTVATL (SL9) presented by HLA-A<sup>∗</sup> 02:01 (23, 24). The human immune response to the HLA-A<sup>∗</sup> 02:01-restricted Gag77−<sup>85</sup> SLYNTVATL epitope is the most studied—SL9 is a highly immunogenic, helpindependent HIV-1 epitope and a strong negative association was demonstrated between SL9-CTL levels and viral load (25). DCs were co-cultured overnight with SL9-CTLs (2,500– 10,000 clones/well). As positive controls, DCs were incubated with 1µg/ml of cognate peptide before washing and addition of the HIV-specific CTL clones overnight. IFN-γ production was monitored in an Elispot assay as described (24). All antibodies (Abs) used for the IFN-γ Elispot were purchased from Mabtech.

#### Multicolor FACS Analyses

Differentiation and maturation of DCs exposed to Chlamydia or LPS and HIV-1 were analyzed by using anti-human CD11c-AlexaFluor488, HLA-DR-PerCP/Cy5.5, DC-SIGN-PE (Biolegend), CD86-BV421, CD83-APC, CD169-PE (BD Biosciences) on a FACS Verse flow cytometer (BD Biosciences). Cell surface expression of receptors for HIV and HIV-C binding was determined by flow cytometry as described using antihuman CD11b-APC, CD4 PerCP/Cy5.5 and DC-SIGN-PE (BioLegend). Data was analyzed using FACS DIVA software (BD Biosciences) and R.

# Statistical Analysis

Data were analyzed using GraphPad Prism software (GraphPad Software Inc.). Statistical analyses were performed using two-way ANOVA with Dunnett's posttest for multiple comparisons.

### RESULTS

#### Reduced Maturation of DCs During Chlamydial Co-infection Compared to Sequential Infection

We initially evaluated whether exposure to Chlamydia induced maturation of DCs similarily to the positive control LPS. Therefore, we analyzed cell surface expression of the specific markers CD83, CD86 and HLA-DR after the different treatments. We found that long-term exposure (24 h) of iDCs to Chlamydia induced significant up-regulation of CD83, CD86, and HLA-DR compared to untreated iDCs (**Figure 1A**). However, expression levels of CD83, CD86, and HLA-DR on Chlam-DCs were lower compared to LPS-stimulated DCs (LPS-DCs) in all donors tested (**Figure 1A**, n = 6). Independent of DC stimulation, the expression levels of DC-SIGN and the complement receptors 3 and 4 (CR3, CD11b/CD18; CR4, CD11c/CD18) were only moderately changed and CD4 expression was slightly reduced under all maturation conditions as also shown by Chen et al. (26) (not shown). Exposure of such various matured DCs (Chlam-DCs, LPS-DCs) to HIV-C did not change the expression of up-regulated markers CD83 (**Figure 1B**, left panel), CD86 (**Figure 1B**, middle panel), and HLA-DR (**Figure 1B**, right panel). In contrast, a reduced maturation of DCs was observed upon co-infection with HIV-C and Chlamydia (HIV-C/Chlam-DCs) and this maturation was comparable to that when iDCs were exposed to HIV-C only (**Figure 1B**, CD83—left panel, CD86—middle panel, HLA-DR—right panel). Expression of all maturation and activation markers was significantly higher on HIV-C- and HIV-C/Chlam-DCs compared to iDCs. We demonstrated that stimulation of DCs with Chlamydia caused a lower DC maturation compared to LPS and this maturation was not increased due to additional HIV-C exposure.

# Binding of HIV-C Depends on the DC Maturation Status

Since expression of activation markers was shown to be different on iDCs, Chlam- and LPS-DCs, we assessed whether this might lead to differential binding of HIV-C to DCs. To characterize binding of HIV-C co-cultures of various matured DCs and HIV-C were incubated for 6 h at 4◦C (8). At 4◦C, DCs just bind but do not internalize viral pathogens (8). HIV-C (25 ng p24/ml) was added to iDCs, Chlam- and LPS-DCs for 6 h at 4◦C. Using the co-infection model, DCs were incubated with simultaneously added HIV-C and Chlamydia under above mentioned conditions. Cell-bound virus was determined after thorough washing and lysing of DCs by quantification of p24 protein. Similar amounts of HIV-C were attached to iDCs and HIV-C/Chlam-DCs, while Chlam- and LPS-DCs depicted a significantly increased binding of HIV-C (**Figure 2A**). A similar binding pattern was analyzed for non-opsonized HIV-1 (HIV, **Figure S2A**). Therefore, binding to DCs was independent of the opsonization pattern, but was modulated by DC maturation status.

### DC Maturation Affects HIV-C Internalization

To also see if internalization of HIV-C in iDCs, HIV-C/Chlam-, Chlam-, and LPS-DCs differs, we incubated differentially stimulated cells for 6 h at 37◦C. Virus was added as described above and bound/internalized HIV-C was determined by p24 ELISA after washing and lysing the cells. These analyses revealed that LPS-DCs show a ∼5-fold higher internalization compared to iDCs and HIV-C/Chlam-DCs (**Figure 2B**). Internalization of HIV-C into LPS-DCs was significantly higher compared to its non-opsonized counterpart (**Figure S2B**, p = 0.005). Though the internalization of HIV-C in Chlam-DCs was lower compared to LPS-DCs, a significantly higher internalization of HIV-C compared to both iDCs (p = 0.0030) and HIV-C/Chlam-DCs (p = 0.0071) was identified (**Figure 2B**). The increase in HIV-1 internalization upon DC maturation was observed independent on whether the virus was opsonized (**Figure 2B**) or not (**Figure S2B**).

#### DC Maturation Affects HIV-C Fusion

To further evaluate the impact of iDC maturation by the different treatments on the interaction of cell and virions, we analyzed virion fusion using Vpr-β-lactamase (Vpr-blam) containing HIV-C (**Figure 2C**) or HIV (**Figure S2C**). We found that fusion was not inhibited in HIV-C/Chlam-DCs relative to HIV-C-exposed iDC controls (**Figure 2C**). In contrast fusion was significantly decreased in Chlam-DCs and LPS-DCs (**Figure 2C**). It is notable that fusion was completely inhibited in the LPS-DCs independent of the opsonization pattern of the virus (**Figure 2C** and **Figure S2C**). iDCs and co-infection of DCs with Chlamydia were associated with the highest fusion with HIV-1, while sequential infection with Chlamydia displayed significantly lower fusion levels with a complete inhibition of fusion in LPS-DCs.

#### Siglec-1 Does Not Play a Role With Respect to HIV-C Capture

Since Siglec-1 (CD169) was described—at least in vitro - to exert a prominent role with respect to capture and transfer of HIV-1 in LPS-stimulated mDCs (27–29), we analyzed colocalization of this molecule with GFP-tagged HIV or HIV-C in differently stimulated DCs. For this, we performed high content screening of differentially stimulated and infected DCs and analyzed the co-localization of GFP-tagged virus with PE-labeled Siglec-1. We automatically analyzed two fields á 100 cells for their co-localization of HIV-1 and Siglec-1 using the HarmonyTM software (Perkin-Elmer) and mean values of spots co-localizing within 100 cells are depicted in **Figure S3**. These analyses revealed no significant differences but only slightly higher Siglec-1/HIV-C co-localization in Chlam-DCs compared to iDCs or HIV/Chlam-DCs and compared to background fluorescence of non-infected cells, which served as negative controls (**Figure S3**). As positive controls, mature DCs (Chlam-DCs or LPS-DCs) infected with non-opsonized HIV-1 (HIV) were used, which displayed significantly higher co-localization compared to HIV-C-infected DCs. the results suggest that a modulation of the interaction of HIV-1 and Siglec-1 is not playing a major role in viral capture.

### HIV-C Localizes to HLA-DR-Containing Compartments in Chlam- and LPS-DCs

To gain additional insights into potential differences in the interaction of HIV-1 with iDCs matured by the different treatments, we evaluated the intracellular localization of HIV-C in iDCs, HIV/Chlam-, Chlam-, and LPS-DCs. To this end, we infected the respective different DC populations using fluorescently labeled HIV-C and analyzed viral particle distribution of internalized HIV-C by high-content/highthroughput (HC/HT) image analyses and confocal microscopy (**Figure 3**, **Figure S4**). For these analyses, cells were additionally labeled using a nuclear stain (**Figure 3**, **Figure S4**, Hoechst, blue) and HLA-DR as marker for endosomal compartments including virus containing compartments (VCCs) (**Figure 3**, **Figure S4**, green). The analyses revealed significantly lower levels of HIV-C-containing HLA-DR-containing compartments in iDCs compared to HIV-C/Chlam-, Chlam-, and LPS-DCs (**Figure 3**, left, histogram plot). Significantly higher HLA-DR-containing compartment levels were detected in LPS-DCs compared to both, HIV-C/Chlam- and Chlam-DCs (**Figure 3**, left, histogram plot). The accumulation of the virus in HLA-DR-containing compartments in LPS-DCs was confirmed using confocal microscopic analyses (**Figure 3**, right)—these revealed vacuolar and cytoplasmic distribution of HIV-C in iDCs (**Figure 3**, right,

upper panel, **Figure S4**, left), while only HLA-DR-containing compartments, but no cytoplasmic HIV-1, were detected in LPS-DCs (**Figure 3**, right, lower panel, **Figure S4**, right). **Figure 3** shows volume projections of maximal pixel intensity of all layers analyzed and **Figure S4** the respective 3D-rendered zstacks. Fusion and image analyses by HC/HT screening and confocal microscopy revealed a high cytoplasmic distribution of HIV-C in iDCs and under conditions of co-infection. The cytoplasmic distribution was reduced in sequentially infected DCs and completely abrogated in LPS-DCs. In contrast HIV-1 containing vacuoles were mainly detected in LPS-exposed DCs, to lower levels in Chlamydia-exposed DCs and only marginally in iDCs.

# DC Infection Is Enhanced by Chlamydia Co- and Sequential Infection

We recently demonstrated that HIV-C overcomes restriction in iDCs resulting in significantly higher productive DC infection, improved antigen-presentation as well as humoral antiviral immune responses (30). We analyzed productive DC infection using HIV-C in co- (HIV+Chlam-) and sequential infection (Chlam-) DC models as well as LPS-DCs. iDCs were used as controls—again we found that non-opsonized HIV caused a significantly lower productive infection of iDCs (**Figure 4**, upper panel, dotted green line, HIV, vs. green line, HIV-C and **Figure S5**) compared to HIV-C despite similar binding and internalization (**Figure 4**, upper panel, and **Figure S1**). However, neither HIV (not shown) nor HIV-C (**Figure 4**, upper and lower panels) caused any productive infection in LPS-DCs. Within the co-infection model, both HIV-C (**Figure 4**, upper and lower panels) and HIV (**Figure S5**) exerted an enhanced productive DC infection compared to iDCs (**Figure S5**) nevertheless, complement opsonization of HIV-1 still promoted a significantly increased DC infection compared to its nonopsonized counterpart (**Figure S5**). Although Chlam-DCs representing the sequential infection—showed a high maturation and low viral fusion, they were infected to high levels with both HIV-C (**Figure 4**) and HIV (**Figure S5**). These data suggest that HIV-C facilitated productive infection in DCs during chlamydial co- and sequential infection as well as a different maturation status between LPS- and Chlamydia-matured DCs.

#### HIV-C/Chlamydia Co- but not Sequential Infection of DCs Is Associated With Reduced HIV Transfer

Lastly, we evaluated HIV-1 trans-infection from differently matured DCs to autologous, stimulated CD4<sup>+</sup> T cells as revealed by a co-culture with T cells. In these studies, we found that simultaneous stimulation of DCs with HIV-C and Chlamydia

resulted in similar infection rates to CD4<sup>+</sup> T cells as HIV-C-exposed iDCs. Compared to Chlam- and LPS-DCs, these conditions illustrated significantly reduced trans-infection in coculture experiments (**Figure 5**). As demonstrated previously for non-opsonized HIV (27, 31), LPS-matured DCs, too, transmitted significantly more virus when complement-opsonized compared to iDCs, HIV/Chlam- and Chlam-DCs (**Figure 5**). Therefore, levels of transmitted HIV-C in Chlamydia-matured DCs differ in co- and sequential infection models and transfer does not correlate with Siglec-1 co-localization in the HIV-C model (**Figure S3**).

# Chlamydia Co-infection Promotes Significant Activation of HIV-Specific CTLs, While Reversing the Situation During Sequential Infection

Bypassing of restriction mechanisms in iDCs and enhanced productive infection using HIV-C rendered the cells capable to activate highly specific anti-HIV-cellular and humoral immune responses (9, 17). To determine the potential impact of Chlamydia on cellular HIV responses, we evaluated the ability of differently matured DCs (iDCs, HIV-C or HIV/Chlam-, Chlam-DCs, and LPS-DCs) exposed to HIV-C (**Figure 6**) or HIV (**Figure S6**) to stimulate HLA-matched HIV-specific CTLs. While in the co-infection model, when Chlamydia and HIV-C were added simultaneously, we detected a significantly higher CTL stimulatory capacity compared to HIV-C-exposed iDCs (**Figure 6**). Within the sequential infection model (Chlam-DCs, LPS-DCs) a significantly abrogated potential to stimulate HIVspecific CD8<sup>+</sup> T cells was observed (**Figure 6**). SLYNTVATLexposed DCs were used as positive controls (**Figure 6**). The CTL-stimulatory power of DCs was also drastically augmented using co-infection of the cells with bacteria and non-opsonized HIV-1 (**Figure S6**). As already observed during our earlier work, HIV-loaded iDCs exerted a very weak CTL-stimulatory capacity (**Figure S6**) (9, 18). These observations illustrate that coinfection of DCs with Chlamydia and HIV-C or HIV is associated with induction of HIV-specific CTL responses, while sequential infection results in increased hazard with respect to the weak CTL-stimulatory capacity of DCs.

# DISCUSSION

The studies presented here reveal that infection of the host with Chlamydia and HIV-1 have both potential positive and negative impact on HIV infection. Simultaneous infection of DCs with Chlamydia and HIV might be beneficial for the host as this triggers a higher HIV-specific CTL activation and lower transfer of HIV to autologous CD4<sup>+</sup> T cells. In

contrast, sequential infection of DCs with Chlamydia and HIV, which might be a common situation in the host, results in detrimental outcomes as it is associated with higher productive DC infection and viral transmission to susceptible CD4<sup>+</sup> T cells as well as poorer stimulation of HIV-1-specific CD8<sup>+</sup> T cell clones.

Upon simultaneous stimulation of DCs with Chlamydia and either complement-opsonized HIV-1 or untreated control HIV, a significantly improved CTL response was observed. This is in contrast to the requirement for complement-opsonization we previously reported in the absence of Chlamydia exposure to act as an endogenous adjuvant for DC-mediated CTL activation of iDC (16). We also find that HIV-exposed DCs co-infected simultaneously with Chlamydia exerted a superior

FIGURE 5 | HIV-C is efficiently transferred from Chlam- and LPS-DCs. In co-culture experiments differentially stimulated DCs (iDCs, green; HIV/Chlam-DCs, blue; Chlam-DCs, red; LPS-DCs, turquoise) were infected using HIV-C (25 ng p24/ml), thoroughly washed and autologous CD4<sup>+</sup> T cells were added. Significantly higher infectivity was measured in Chlam-DC- and LPS-DC co-cultures compared to iDC- and HIV-C/Chlam-DC-CD4<sup>+</sup> T cell co-cultures. p24 ELISAs of differently stimulated DC/T cell co-cultures performed in triplicates from two donors exposed to HIV-C are summarized and statistical analyses were performed using two-way ANOVA with Dunnett's posttest for multiple comparisons.

FIGURE 6 | Enhanced stimulation of HIV-specific T cell clones at simultaneous addition. IFNγ induction in CD8<sup>+</sup> T cell clones by HIV-C-exposed iDCs and HIV-C/Chlam-DCs was significantly stronger than that of non-opsonized HIV-loaded DCs (HIV-DCs; p < 0.0001 for CD8<sup>+</sup> T cell clones), or Chlam- and LPS-DCs exposed to HIV-C (p < 0.0001 for all). As positive controls specific peptide-loaded DCs for CD8<sup>+</sup> T cell clones were used (iDCs/SLYNTVATL). IFNγ Elispots of CD8<sup>+</sup> T cell clones were repeated using HLA-matched and differently stimulated DCs from three donors exposed to HIV-C, or HIV. Statistical analyses were performed using two-way ANOVA with Dunnett's posttest for multiple comparisons.

CTL-stimulatory capacity of HIV-specific CD8<sup>+</sup> T cell clones compared to their HIV-iDC counterparts (9, 16, 18). As shown recently in a murine vaginal co-infection model (32), chlamydial

and day 15 post infection is depicted.

pre-infection protected the mice from subsequent Herpes Simplex Virus (HSV)-2 challenges. This Chlamydia-mediated protection was transient and only detectable in mice prechallenged with Chlamydia before, simultaneously with, or shortly after infection with HSV-2 (32). These findings are in accordance with our data, where co-infection of DCs with Chlamydia and HIV or HIV-C resulted in significantly higher CTL activation via DCs. In contrast, DC sequential infection for 3 h or 24 h with Chlamydia followed by HIV-C infection had detrimental outcomes (**Figures 4**, **5**). Under these conditions, sequentially infected DCs only had a poor capacity to stimulate HIV-specific CTLs and allowed significantly higher productive HIV infection (cis infection). In contrast, no cis-infection was analyzed at all in DCs challenged for 3 or 24 h with LPS prior infection with HIV-C. The impact of pre-existing STIs on HIV immune responses was studied by Sheung et al. (33) in high risk Kenyan female sex workers. They found that mucosal Neisseria gonorrhoeae co-infection during HIV-1 acquisition was associated with substantially enhanced HIV-specific CD8<sup>+</sup> T cell responses (33). The enhanced CTL response was not seen in women with Chlamydia co-infection, which correlates well with our findings within the sequential infection model of DCs with Chlamydia and HIV-1, which exerted a weak HIV-specific CTL activation. However, to study the impact of simultaneous STI on HIV immune responses is logistically impossible in the human host.

LPS-DCs had the highest binding and internalization of HIV-C followed by Chlam-DCs, while iDCs and HIV-C/Chlam-DCs showed similar HIV-C up-take levels. As described earlier, maturation of DCs—as seen in LPS- or Chlam-DCs—enhances their virus capture and trans infection capacity while reducing viral fusion events (34). HIV-C/Chlam-DCs are not as mature as Chlam-DCs, when binding and internalization were measured. Therefore, HIV-C/Chlam-DCs more act like iDCs, which show less binding and internalization, but enhanced fusion. Consistent with this interpretation, the highest levels of fusion were measured in iDCs and HIV-C/Chlam-DCs, while Chlam- and LPS-DCs demonstrated considerably reduced fusion levels (34). Consistent with our fusion data, the accumulation of HIV-C in HLA-DR-containing compartments was highest in LPS-DCs and also Chlam-DCs showed significantly higher HIV-C-containing compartments compared to iDCs. In macrophages, virus containing compartments (VCCs) were described to resemble late endosomes or multi-vesicular body (MVB) compartments and to show enrichment of CD9, CD53, CD81, CD82, and MHC class II (35, 36). We previously illustrated co-localization of HIV-C with these markers (7). VCCs are non-acidic and often express surface-connected tubular conduits to the plasma membrane (35, 37, 38). VCC formation was demonstrated to greatly facilitate trans-infection of HIV-1 from macrophages to autologous CD4<sup>+</sup> T cells (39). Accumulation of viral particles within intracellular DC compartments was illustrated to share multiple features with macrophage VCCs (30, 40, 41). Concentration of non-opsonized HIV-1 particles in large saclike and tetraspanin-rich/MHC II compartments within LPSmDCs was shown by various imaging studies (27, 42, 43). We also show a similar distribution of HIV-C in MHC II (HLA-DR-) compartments particularly in Chlam- and LPS-matured DCs. Transfer of such trapped viral particles, which were nonopsonized, from mDCs to CD4<sup>+</sup> T cells was highly effective (44–46). Localization of internalized virus differs greatly in endocytically active iDCs compared to mDCs—mDCs storing intact HIV particles within large vesicles correlate with increased trans-infection abilities (34). We here demonstrate (47), that similar to non-opsonized HIV-1, mature DCs (i.e., LPS-DCs and Chlam-DCs) retained HIV-C particles in an infectious form and efficiently transmitted the virus particles to target CD4<sup>+</sup> T cells through trans infection. Despite co-infection with Chlamydia, DCs displayed significantly higher amounts of trapped virus particles compared to iDCs loaded with HIV-C. Such co-infected DCs exerted superior antiviral functions as increased HIV-specific CTL-stimulation and reduced transfer to CD4<sup>+</sup> T cells. These effects were likely a consequence of higher viral fusion of HIV-C during co-infection compared to LPS-DCs and the sequential infection model, where DCs were incubated with Chlamydia for a prolonged period prior to addition of HIV-C.

Siglec-1 was recently described to play a major role during HIV-1 capture and transfer in LPS-mDCs. Here, we also analyzed co-localization of GFP-tagged complement-opsonized HIV-1 and Siglec-1 in iDCs, HIV+Chlam-, Chlam-, and LPS-DCs. We did not find any correlation between co-localization of Siglec-1/HIV-C, the maturation status of DCs and transfer to susceptible T cells. These findings are consistent with recent in vivo studies by Martinez-Picado et al. where they demonstrated that Siglec-1 protein truncation did not have a measurable impact on HIV-1 acquisition or AIDS outcomes in vivo (48). The missing correlation of Siglec-1/HIV-C and transfer from differently matured DCs to target cells which was described in vitro for non-opsonized HIV-1 by recent studies (41, 49–51) might rely on the fact that C3 fragments covalently bind to the surface of HIV-1 (52) potentially hampering interactions of Siglec-1 with virus-incorporated host-cell-derived glycosphingolipid GM3. GM3 was shown to allow capture by DCs, monocytes and macrophages in vitro (51). In our analyses, we, too, found higher co-localization of non-opsonized HIV with Siglec-1, in particular in the sequential infection model, but also in LPS-mDCs. In vivo, HIV-1 was found to be opsonized with complement fragments or specific antibodies in all compartments tested so far (53–57). Therefore, the findings by Martinez-Picado et al. that Siglec-1 protein truncation did not correlate with HIV-1 acquisition or AIDS outcomes in vivo could be explained by covalent coating of the virus with C3. C3 bound to the viral particles would mediate interactions with complement receptors 3 and 4 (CR3, CD11b/CD18; CR4, CD11c/CD18) abundantly expressed by immature and mature DCs rather than allowing interactions of GM3 with Siglec-1. We earlier found that the covalently linked cloud of C3 fragments on the viral surface impaired interactions of the HIV envelope glycoproteins with C-type lectins expressed on iDCs (8).

The presented data shows that co- or sequential infection of DCs with Chlamydia alters the progression of subsequent HIV-1 infection with implications for HIV-1 processing into peptides for MHC presentation, transfer to target cells via trans-infection and CTL responses (58, 59). STIs are an important public health issue and in HIV-positive women, STIs are associated not only with gynecological complications but with increased risk of HIV transmission to HIV-negative partners and newborns (60). We find that infection of DCs with HIV-C (or HIV) and Chlamydia are associated with mechanisms but only if added simultaneously. The mechanisms are likely due to simultaneous stimulation of innate immune mechanisms on DCs. One such trigger might be activation of Toll-like receptors (TLRs), since Chlamydia was illustrated to activate TLR2/6 (61). Therefore, within the chlamydial/HIV-C co-infection model TLRs in concerted action with CR3/CR4 (HIV-C) or C-type lectins (HIV) could stimulate a more robust DC activation compared to HIV-C- or HIV-DCs alone. This would result in even higher stimulation of HIVspecific CTLs and reduction of viral infectivity in the coinfection model. Other host innate immune responses, which might contribute to the higher anti-HIV-1 activity of coinfected DCs comprise superior induction of pro-inflammatory cytokines and/or antimicrobial peptides (62–64). However, the sequential infection model, which probably occurs more often in vivo compared to simultaneous DC stimulation with both pathogens, was associated with harm to the host due to significantly enhanced cis and trans infection with HIV-1 and significantly reduced HIV-specific CTL-stimulation. In future studies, we want to elucidate the mechanisms in DCs involved in the observed differences in Chlamydia-mediated effects to characterize factors associated with protection, which might be applied as therapeutic interventions during STIs to lower the risk of HIV-1 transmission and infection.

#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Ethics Committee of the Medical University of Innsbruck. The protocol was approved by the Ethics Committee of the Medical University of Innsbruck [ECS 1166/2018].

#### AUTHOR CONTRIBUTIONS

MS, UK, and PC performed experiments, analyzed data, and read the manuscript, PH, TJH, AM, and RB-W contributed essential components, read and discussed the manuscript, CL-F helped in designing the study, provided financial support and read and discussed the manuscript, WP and DW designed the study, conducted experiments, analyzed data, and wrote the manuscript with input from all authors. All authors provided critical feedback and helped shape the research, analysis and manuscript.

# ACKNOWLEDGMENTS

We would like to thank our technician Karolin Thurnes, Divison of Hygiene and Medical Microbiology, and Prof. Oliver Keppler, Max-von-Pettenkofer Institute, Munich, Germany, for supplying the HIV plasmids. We would like to thank the Austrian Science Fund (MCBO graduate program/W011010-21 and P24598 to DW, P25389 to WP) and the Oesterreichische Nationalbank Anniversary Fund (Project number: 17614 to WP) for supporting this work. Further, this publication was made possible with help from the HIV Vaccine Trials Network Mucosal Immunology Group Program, an NIH funded program (HVTN LC Grant UM1AI068618 to TJH and DW) and the Tyrolean Science Fund (to DW). AM is supported by ANRS, Sidaction and ANR fundings.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.01123/full#supplementary-material

Figure S1 | Flow cytometric analyses of DC profiles of iDCs and differentially treated DCs. (Upper panel) Monocyte-derived iDCs are routinely checked for characteristic markers CD11b, CD11c, and DC-SIGN, which are homogenously expressed on day 5 iDCs. Characteristic maturation markers CD83 and HLA-DR are not expressed or do show a low expression on day 5 iDCs dependent on the donor. Representative histogram plots for the various markers are illustrated. (Lower panel) Day 5 iDCs were treated for further 2 days with live (light green) or heat-inactivated (dark green) Chlamydia or not (iDCs, red) and analyzed for expression of characteristic maturation markers. A representative histogram plot for CD83 is depicted.

Figure S2 | Chlam- and LPS-DCs efficiently capture HIV. Binding at 4◦C (A) and internalization at 37◦C (B) were performed in triplicates using 25 ng/ml of R5-tropic non-opsonized HIV-1. Bar graphs show means ± SD from three independent experiments. p24 levels within the cell lysates were determined by ELISA. Prior to cell lysate preparation, cells were thoroughly washed to remove unbound virus. Statistical analysis shows 2-way ANOVA with Tukey's multiple comparisons test. Six donors are summarized. (C) Fusion assays were performed by exposure of HIV/Chlam-DCs and LPS-DCs to HIV bearing the chimeric protein β-lactamase-vpr. The amount of fused virus was determined by flow cytometric analyses of cleaved CCF2 in the cytoplasm. Percentages of cleaved CCF2-positive cells from three independent donors are depicted.

Figure S3 | Siglec-1-independent transfer of HIV-C. Enhanced transfer of HIV-C from Chlam- and LPS-DCs was independent on Siglec-1 as analyzed by high content screening as depicted (upper panel). Only low spots of HIV-C/Siglec-1-co-localization were quantified in 2 fields of 100 cells each (lower panel, right). The co-localization was compared to non-infected differentially stimulated DCs, which represent background values (lower panel, left), and HIV-infected differentially stimulated DCs (lower panel, middle). 200 cells were analyzed in total.

Figure S4 | Localization of HIV-C in iDCs and LPS-DCs. For three-dimensional reconstructions, confocal z stacks of iDCs and LPS-DCs exposed to HIV-C were processed with Imaris software using surface reconstruction (Surpass, IMARIS 8.2). About 30 cells per condition were analyzed.

Figure S5 | Enhanced DC infection by HIV-C independent of stimulation. iDCs, HIV/Chlam- and Chlam-DCs exerted a significantly enhanced infection using HIV-C (gray) compared to HIV (white). Nevertheless, also productive DC infection of HIV/Chlam-DCs was significantly increased compared to the low-level infection of iDCs using non-opsonized HIV. Three independent donors were summarized in the graph and means ± SD are shown.

Figure S6 | Enhanced CTL stimulation by HIV+Chlam DCs. IFNγ induction in CD8+ T cell clones by DCs simultaneously exposed to HIV and Chlamydia was significantly higher than that iDCs, Chlam-, and LPS-DCs exposed to HIV (p < 0.0001 for all). Means ± SD of three independent experiments are illustrated.

#### REFERENCES


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Schönfeld, Knackmuss, Chandorkar, Hörtnagl, Hope, Moris, Bellmann-Weiler, Lass-Flörl, Posch and Wilflingseder. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# What Makes a pDC: Recent Advances in Understanding Plasmacytoid DC Development and Heterogeneity

Dendritic cells (DCs) are professional antigen presenting cells (APCs) that originate in the bone marrow and are continuously replenished from hematopoietic progenitor

Andrea Musumeci, Konstantin Lutz, Elena Winheim and Anne Barbara Krug\*

Institute for Immunology, Biomedical Center, Ludwig-Maximilian-University, Munich, Germany

Edited by: Christian Muenz, University of Zurich, Switzerland Reviewed by: Susan Kovats, Oklahoma Medical Research Foundation, United States Diana Dudziak, Universitätsklinikum Erlangen, Germany cells. Conventional DCs (cDCs) and plasmacytoid DCs (pDCs) are distinguished by morphology and function, and can be easily discriminated by surface marker expression, both in mouse and man. Classification of DCs based on their ontology takes into account their origin as well as their requirements for transcription factor (TF) expression. cDCs and pDCs of myeloid origin differentiate from a common DC progenitor (CDP) through committed pre-DC stages. pDCs have also been shown to originate from a lymphoid progenitor derived IL-7R<sup>+</sup> FLT3<sup>+</sup> precursor population containing cells with pDC or B cell potential. Technological advancements in recent years have allowed unprecedented resolution in the analysis of cell states, down to the single cell level, providing valuable information on the commitment, and dynamics of differentiation of all DC subsets. However, the heterogeneity and functional diversification of pDCs still raises the question whether different ontogenies generate restricted pDC subsets, or fully differentiated pDCs retain plasticity in response to challenges. The emergence of novel techniques for the

regarding DC development and plasticity in the near future.

Natalio Garbi, University of Bonn, Germany

\*Correspondence: Anne Barbara Krug anne.krug@med.uni-muenchen.de

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

> Received: 04 October 2018 Accepted: 13 May 2019 Published: 29 May 2019

#### Citation:

Musumeci A, Lutz K, Winheim E and Krug AB (2019) What Makes a pDC: Recent Advances in Understanding Plasmacytoid DC Development and Heterogeneity. Front. Immunol. 10:1222. doi: 10.3389/fimmu.2019.01222 Keywords: plasmacytoid dendritic cells, hematopoiesis, dendritic cell development, DC progenitor, plasticity, heterogeneity

integration of high-resolution data in individual cells promises interesting discoveries

# INTRODUCTION

Plasmacytoid dendritic cells (pDCs) and two major subsets of conventional dendritic cells (cDC1 and cDC2) have been identified in mice and humans as well as other mammalian species including non-human primates and pigs, with high similarities between species (1–3). cDC subsets recognize both extracellular and intracellular pathogens, efficiently process and present exogenous antigens to naive CD4<sup>+</sup> and CD8<sup>+</sup> T cells and elicit effective adaptive immunity. pDCs are highly effective in sensing intracellular viral or self DNA and RNA mainly via Toll-like receptors (TLRs) and rapidly producing large amounts of type I and III interferons (IFNs) (4). Thus, they play an important role in antiviral immunity and systemic autoimmunity (5–8). pDCs are distinguished from cDC subsets by expression of surface markers CD45R (B220), CD45RA, Ly-6C, Siglec-H, and BST2 (CD317) in the mouse and CD303 (BDCA2), CD304 (BDCA4), CD123 (IL-3R), and CD45RA in humans.

DC subpopulations originate from proliferating progenitor cells in the bone marrow (BM) and require fms-like tyrosine kinase 3 ligand (FLT3L)–FLT3 interaction for their development. Lin<sup>−</sup> FLT3<sup>+</sup> c-Kitlow/int M-CSFR<sup>+</sup> murine BM cells, so called common DC progenitors (CDP), which are derived from the myeloid macrophage DC progenitors (MDP) or lymphoid primed multipotent progenitors (LMPP), were shown to be DCcommitted and to generate pDCs, cDC1 and cDC2 [**Figure 1**, (9, 10)]. Clonal assays and subsequent single cell transcriptome and imaging analyses demonstrated that the majority of CDPs are already pre-committed to pDC or cDC subsets (9–13). This is also the case for the pre-cDCs, which already contain pre-cDC1, and pre-cDC2 (13, 14). In contrast, pDCs are also produced from a lymphoid progenitor (LP) (15) in the steady state whereas this happens for cDCs only in situation of cDC ablation (16).

DC subpopulations can be defined by their ontogeny and by the requirement of specific transcription factors (TF) for their development. pDCs require high-level expression of IRF-8, TCF-4 (also known as E2-2) and BCL-11A for their development, functional specification and maintenance (17–21). Expression of DNA-binding protein inhibitor ID-2, which prevents the activity of the major pDC TF E2-2, needs to be suppressed to allow the generation of pDCs from CDPs (22, 23). On the other hand, the major cDC branches can be distinguished by distinct requirements for IRF-8 (for cDC1) and IRF-4 (for cDC2) (14, 24–27).

DC subpopulations are also distinguished by a high degree of functional specialization (28). While cDC1 efficiently crosspresent antigens to CD8<sup>+</sup> T cells (27, 29, 30) and produce high levels of IL-12p70, thus promoting cytotoxic T cells and Th1 cells (31, 32), cDC2 are superior in presenting antigens on MHC class II, supporting Th1, Th2, and Th17 polarization (26, 27, 33). pDCs participate in the first line of defense against viral infections by acting as innate effector cells, which initiate IFN-induced antiviral responses in adjacent cells and recruit cytotoxic NK cells (5). Resting pDCs are weak antigen presenting cells and in contrast to cDCs do not prime naïve T cells. After activation, pDCs can acquire the capacity to present antigens and activate T cells directly. Their ability to prime T cells, thus performing truly like DCs, is debated and complicated by the finding that pDC-like cells, which were shown to be related to cDCs (13, 15, 34, 35) have been included in the pDC population in many functional studies, as discussed below. By producing cytokines and chemokines activated pDCs modulate T cell responses elicited by cDCs (5). During viral infection pDCs were shown to cooperate with cDC1 in lymph nodes, promoting their maturation and cross-presentation activity to induce antiviral CD8<sup>+</sup> T cells (36). But there is also evidence for a role of pDCs in the induction of immune tolerance by generation of hyporesponsive and regulatory T cells (37–39).

Recent technological developments have allowed unprecedented resolution, down to the single cell level, in the analysis of cell transcriptomes as well as in in vivo lineage tracing, overcoming the limitations of discrimination based solely on surface markers (40–44). The characterization of transcriptional profiles of individual cells (13, 42, 45) and more recently the integrated analysis of single cell transcriptome and chromatin accessibility (46) has revealed unexpected heterogeneity and signs of very early lineage priming of individual hematopoietic BM progenitor cells, which were previously considered multior oligopotent. For example, single cell barcoding and tracing showed that DC and even pDC commitment can already be imprinted in early LMPP and at the HSPC stage (12, 41, 47). cDC subtype specification was detected already at the CDP and pre-cDC stage of development (12–14). In some instances, these analyses led to the definition of more stringent surface marker combinations that allow the discrimination of largely committed progenitor cells within the "oligopotent" population (13, 15).

Combining CRISPR/Cas9-based genomic perturbation with transcriptome profiling in the same cells revealed differentiation trajectories and regulatory networks during hematopoiesis (40, 48). Integration of clonal labeling and lineage tracing experiments and single cell time-lapse imaging experiments may lead to a better understanding of immune cell differentiation dynamics and regulation in the future (11, 40, 43, 49).

#### PLASMACYTOID DENDRITIC CELL DEVELOPMENT FROM MYELOID AND LYMPHOID PROGENITORS

Early works indicated that DCs can be derived from both FLT3<sup>+</sup> CMP and CLP (50, 51). Competitive in vivo transfer experiments with CMPs and CLPs showed that pDCs can also be generated from both, but are mainly of "myeloid" origin (52). Subsequent studies indicated that CMP and CLP-derived pDCs differ in their ability to produce type I IFN and to stimulate T cells (53, 54). Interestingly, a significant proportion of pDCs expresses recombination activation genes (Rag1/Rag2) and undergoes immunoglobulin DH-J<sup>H</sup> rearrangement indicating a "lymphoid" past. But the expression of Rag genes and detection of Ig rearrangements in pDCs derived from both CMP and CLP suggested that these are by-products of a "lymphoid" transcriptional program expressed only transiently in the pDC lineage (55, 56). However, the issue was revisited by Sathe et al. who found that RAG1 expression and Ig rearrangement are mainly found in CLP-derived pDCs (54). pDC generation from CLPs but not CDPs required constitutive type I IFN signals for upregulation of FLT3, suggesting differential requirements for instructive cytokines for the two developmental pathways (57). After the discovery that myeloid progenitor derived CDPs generate both cDCs and pDCs, research mostly focused on the branching of pDC and cDC development.

We found that CCR9low pDC-like precursor cells (CD11c+ Siglec H+ BST2+ B220lo/hi), which express lower levels of E2-2 and higher levels of Id2 than pDCs, can be generated from murine CDPs and these can give rise to CCR9high pDCs as well as cDCs [**Figure 1**, (11, 58, 59)]. The CCR9low pDClike precursor population in the BM contains only a small fraction of proliferating cells indicating heterogeneity within this population regarding differentiation stage (58). It remains to be determined if this population, which can also be detected in lymphoid organs at low frequency contains differentiated cells

with plasticity to develop into pDCs and cDCs or precursors with dual potential or both. Interestingly, pDC-like cells with a similar phenotype accumulated in the BM of Mtg16-deficient mice, which failed to downregulate Id2 expression, thereby blocking the activity of E2-2 and further pDC differentiation (60). In addition, Zeb2 has been identified as an important regulator of Id2 expression, which allows pDC development from CDPs by suppressing the alternative cDC1 fate at a common precursor stage (22, 23). More recently Etv6 was shown to cooperate with IRF8 to refine cDC1-specific gene expression and repress the pDC gene expression signature indicating the close relationship between cDC1 and pDCs (61). Siglec-H, a canonical marker distinguishing mature pDCs from cDCs, is expressed at very early stages of differentiation, but does not denote a plasmacytoid commitment. Within the CDP and the pre-DC fraction in the BM, Siglec-H<sup>+</sup> cells expressing TF Zbtb46 are exclusively committed to cDCs (62) and were shown to contain precursors committed to cDC1 and cDC2 (13, 14). Similarly, Siglec-H<sup>+</sup> Ly-6C<sup>+</sup> cells in the pre-DC compartment (defined as Lin<sup>−</sup> CD135<sup>+</sup> CD11c<sup>+</sup> MHCII<sup>−</sup> CD172α <sup>−</sup>) were shown to give rise to both subsets of cDCs, whereas Siglec-H<sup>+</sup> Ly-6C<sup>−</sup> pre-DCs gave rise to cDC subsets and pDCs (13). Using the single cell imaging and tracking method we could show that CDP progeny transit through a CD11c<sup>+</sup> CCR9low Siglec-H<sup>+</sup> pDC-like stage during their development into CCR9high pDCs (11). The CDP-derived pDC-committed precursor, which must be present within this population, is still a missing link. M-CSFR<sup>+</sup> CDPs give rise to pDCs, however their output is rather low. Interestingly, Onai et al. found that the pDC potential was higher in the M-CSFR<sup>−</sup> E2-2<sup>+</sup> fraction of CDPs in murine BM (12, 63). They also demonstrated that E2-2high cells within M-CSFR<sup>−</sup> IL-7R<sup>−</sup> CDPs gave rise exclusively to pDCs in spleen and lymph nodes, but also to cDCs in the small intestine, showing the plasticity of this pDC-primed CDP subset or its progeny in the local tissue environment (63).

More recently Rodrigues et al. found that FLT3<sup>+</sup> IL-7R (CD127)<sup>+</sup> CD117lo/int lymphoid progenitor (LP) cells in murine BM, which differ from CDPs only by expression of IL-7R and lack of M-CSFR expression, have a 5-fold higher output of pDCs compared to CDPs (15). Within this LP pool, three subpopulations were distinguished by diverse expression of Siglec-H and Ly-6D. Of these, only the Siglec-H Ly-6D double positive (DP) population had exclusive pDC potential, while the Ly-6D single positive (SP) population generated both B cells and pDCs, congruent with the results of a recently published computational fate mapping analysis of single cell RNAseq data (64). Further analysis showed the SP population to contain cells committed either to B cell or to pDC differentiation. The model proposed by Rodrigues et al. suggests that IL-7R<sup>+</sup> Siglec-H and Ly-6D DN LPs proceed to upregulate Ly-6D (SP) and, under the influence of lineage defining TFs IRF8 and EBF1 induced by FLT3L and IL-7 respectively, proceed either to the pDC lineage or towards B cells (**Figure 1**). Interestingly, mice lacking Zeb2 in CD11c<sup>+</sup> cells were shown to have a severe defect in pDC numbers, which was attributed to failed repression of Id2 leading to diversion of precursors to cDC1 (22, 23). Since a substantial proportion of pDCs was shown to be derived from the LP which lacks cDC potential in the steady state (65), it remains to be investigated if the transcriptional repressor Zeb2 is also involved in suppressing alternative cell fates in the LP.

Functionally, the IL-7R<sup>+</sup> DP cells described by Rodriguez et al. as pDC precursors can be considered immature progenitors, as they do not yet express genes important for pDC function (such as Irf7 and Spib) and require further cell divisions to generate mature pDCs (15). In contrast to the CDP-derived CD11c<sup>+</sup> Siglec-H<sup>+</sup> CCR9low pDC-like precursors, the IL-7R<sup>+</sup> DP cells lack CD11c and B220 expression and fail to produce type I IFNs in response to TLR9 stimulation by CpG-A, a hallmark of the pDC-lineage, but acquire this capacity after culture with FLT3L (15).

IL-7R<sup>+</sup> Siglec-H<sup>+</sup> Ly-6D<sup>+</sup> pDC-committed precursors make a substantial contribution to the pool of differentiated pDCs. Thus, pDC generation seems to be regulated by the cell fate decision between pDC and cDC1, but also by the pDC versus B cell dichotomy. The contribution of the two pathways to pDC generation under conditions of inflammation or infection and the functional consequences of the distinct ontogeny of pDCs remain to be investigated.

# HETEROGENEITY OF pDCs AND pDC-LIKE CELLS IN MURINE LYMPHOID ORGANS

Different subsets of pDCs have been identified in the BM, mostly differing in their degree of differentiation and their capacity to produce type I IFNs or pro-inflammatory cytokines (4, 66). Markers such as CCR9, SCA-1, CD9, and Ly-49Q, which are expressed by the majority of peripheral mouse pDCs, can be used to discriminate these subsets (59, 67, 68). More recently, single cell RNAseq analysis confirmed the presence of two subsets within Lin<sup>−</sup> CD11c<sup>+</sup> BST2<sup>+</sup> Siglec-H<sup>+</sup> cells in spleen and BM (15). The "pDC-like cells" described in this paper express several genes characteristic of cDCs and other myeloid cells (including Zbtb46) but lack or express low levels of Ccr9, Ly6d, and Dntt. By gene expression profile and surface phenotype (lower levels of Siglec-H, BST2, MHCII, higher levels of CD11c, Ly-6C, and CX3CR1 compared to pDCs) they greatly resemble the CCR9low MHCIIlow CX3CR1<sup>+</sup> pDC-like precursors described previously in BM (58, 59) and are a subset of those. Interestingly, Rodrigues et al. also found that the minor subset of pDC-like cells (defined as Zbtb46-eGFP<sup>+</sup> Siglec-Hint BST2+), responded with IFN-α production to CpG-A and showed better antigen processing and presenting ability than "regular" pDCs. It was also previously shown that IFN-β production in the spleen is limited to a small subset of CD9<sup>−</sup> cells within the CCR9<sup>+</sup> mature pDC population in murine spleen (69).

These works suggest the existence of minor subsets of pDCs in peripheral organs, differing in the extent of IFN-I production and the capacity of antigen processing and presentation. Considering that these subsets identified by differential expression of surface markers are largely overlapping and often very rare, it remains unclear whether the functional differences observed are due to functional specialization or are the result of lineage imprinting, or whether they are simply sequential stages of pDC differentiation leading to the mature pDC.

## REVISITING THE DEFINITION OF HUMAN pDCs

The pDC-like cells described in the mouse which express pDC markers and TFs, but rapidly give rise to cDCs and behave like cDCs in antigen presentation assays greatly resemble the subset of CD123<sup>+</sup> CD45RA<sup>+</sup> CD33<sup>+</sup> CX3CR1<sup>+</sup> pre-DCs recently identified in human blood (35) and the AXL<sup>+</sup> SIGLEC6<sup>+</sup> human blood DC subset (AS-DC) described by Villani et al. (34). These "pDC-like cells," which are hidden in the pDC population as defined by surface marker expression (Lin<sup>−</sup> HLA-DR<sup>+</sup> CD123<sup>+</sup> CD45RA<sup>+</sup> CD303+), are functionally distinct from pDC in that they do not produce type I IFN in response to TLR7 and 9 stimulation. In that respect they are different from the Zbtb46<sup>+</sup> Siglec-H<sup>+</sup> pDC-like cells found in murine spleen. As to their classification as precursors of cDCs, it is based mainly on the observation that the pre-DCs acquire cDC phenotype and function in culture (35). The human pre-DC population contains pre-cDC1 and pre-cDC2 (35, 70). However, these cells are not proliferating in the steady state and appear to be functionally mature and could therefore actively participate in immune responses (34, 35). Cells in human blood, BM and tonsil defined as a CD2<sup>+</sup> CD5<sup>+</sup> (and CD81+) subpopulation of human pDCs were studied previously and were found to produce IL-12 but not IFN-α and to stimulate naïve CD4<sup>+</sup> T cells (71–74). This population is largely overlapping with the recently described pre-DC and AS-DC (34, 35). It is currently not resolved to which extent cytokine responses and T cell activation capacity attributed to human pDCs in earlier studies were influenced by contamination by cDC precursors, especially because most studies were performed with pDCs that had been stimulated e.g., with IL-3, CD40L or viruses (75–77). It was shown recently that human blood pDCs diversify into functionally distinct and stable subsets after activation by influenza virus or CpG even after prior exclusion of contaminating pre-DCs demonstrating great functional plasticity of this cell type (78). In the light of these recent findings the functional properties of bona fide pDCs in innate and adaptive immune responses need to be reexamined.

#### FUTURE PERSPECTIVES

Technological advances including single cell transcriptome, epigenome, and mass cytometry analyses as well as single cell tracking methods have revealed that development and functional specification of DC subpopulations is much more complex than anticipated. Several questions regarding pDC development and functional plasticity remain unanswered. It would be important to address the contribution of the CDP and LP to pDCs during infections or inflammation and to clarify if the developmental history of pDCs is really relevant for their function. Furthermore, it is unclear at this point, which functions ascribed to human pDCs are mediated by bona fide pDCs and which are mediated by the contaminating pre-DCs. This is especially important for developing pDC-targeted or adoptive transfer therapies for induction of immunity or tolerance. Similarly, the functional diversification of pDCs after activation and also the phenomenon of pDC exhaustion during chronic infection (79) are important topics for further study. An exciting area of research is the correlation of gene expression with chromatin accessibility and epigenetic modifications on the single cell level and the integration of all this data (80), which will allow to unravel the transcriptional regulation of cell fate decisions leading to pDC development and functional diversification. Combined with CRISPR/Cas9-based genetic screening and

#### REFERENCES


functional assays these new single cell analysis methods will lead to a thorough understanding of development, plasticity and function of DC subpopulations with implications for DC targeted therapy.

#### AUTHOR CONTRIBUTIONS

All authors listed have made a substantial, direct and intellectual contribution to the work, and approved it for publication.

#### FUNDING

AK received funding from the German Research Foundation (SFB 1054 TP A06 and TRR 237 TP B14) and Friedrich-Baur-Foundation.


tuning in tissue microenvironments. Int Immunol. (2017) 29:443–56. doi: 10.1093/intimm/dxx058


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Musumeci, Lutz, Winheim and Krug. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

# Flagellin/NLRC4 Pathway Rescues NLRP3-Inflammasome Defect in Dendritic Cells From HIV-Infected Patients: Perspective for New Adjuvant in Immunocompromised Individuals

Edione Cristina dos Reis <sup>1</sup> \*, Vinícius Nunes Cordeiro Leal <sup>1</sup> , Jaíne Lima da Silva Soares <sup>1</sup> , Fernanda Pereira Fernandes <sup>1</sup> , Dhêmerson Souza de Lima<sup>1</sup> , Bruna Cunha de Alencar <sup>2</sup> and Alessandra Pontillo<sup>1</sup>

<sup>1</sup> Laboratory of Immunogenetics, Department of Immunology, Institute of Biomedical Sciences/ICB, University of São Paulo/USP, São Paulo, Brazil, <sup>2</sup> Laboratory of Immune System Cell Biology, Department of Immunology, Institute of Biomedical Sciences/ICB, University of São Paulo/USP, São Paulo, Brazil

Edited by:

Daniela Santoro Rosa, Federal University of São Paulo, Brazil

#### Reviewed by:

Osamu Takeuchi, Kyoto University, Japan Ezequiel Ruiz-Mateos, Institute of Biomedicine of Seville (IBIS), Spain

#### \*Correspondence:

Edione Cristina dos Reis edionereis@usp.br

#### Specialty section:

This article was submitted to Antigen Presenting Cell Biology, a section of the journal Frontiers in Immunology

> Received: 30 July 2018 Accepted: 21 May 2019 Published: 11 June 2019

#### Citation:

Reis EC, Leal VNC, Soares JLS, Fernandes FP, Souza de Lima D, de Alencar BC and Pontillo A (2019) Flagellin/NLRC4 Pathway Rescues NLRP3-Inflammasome Defect in Dendritic Cells From HIV-Infected Patients: Perspective for New Adjuvant in Immunocompromised Individuals. Front. Immunol. 10:1291. doi: 10.3389/fimmu.2019.01291 Introduction: NLRP3 inflammasome plays a key role in dendritic cells (DC) activation in response to vaccine adjuvants, however we previously showed that it is not properly activated in DC from HIV-infected patients (HIV-DC), explaining, at least in part, the poor response to immunization of these patients. Taking in account that several cytoplasmic receptors are able to activate inflammasome, and that bacterial components are considered as a novel and efficient adjuvant, we postulated that bacterial flagellin (FLG), a natural ligand of NAIP/NLRC4 inflammasome, could rescue the activation of the complex in HIV-DC.

Objective: Demonstrate that FLG is able to activate monocyte-derived dendritic cells from HIV-infected individuals better than LPS, and to what extent the entity of inflammasome activation differs between DC from HIV-infected patients and healthy donors.

Methods: Monocyte-derived dendritic cells from HIV-infected patients (HIV-DC) and healthy donors (HD-DC) were stimulated with FLG, and inflammasome as well as DC activation (phenotypic profile, cytokine production, autologous lymphocytes activation) were compared. Chemical and genetic inhibitors were used to depict the relative contribution of NLRC4 and NLRP3 in HIV/HD-DC response to FLG.

Results: FLG properly activates HD-DC and HIV-DC. FLG induces higher inflammasome activation than LPS in HIV-DC. FLG acts through NLRC4 and NLRP3 in HD-DC, but at a lesser extent in HIV-DC due to intrinsic NLRP3 defect.

Conclusions: FLG by-passes NLRP3 defect in HIV-DC, through the activation of NAIP/NLRC4 inflammasome, indicating possible future use of the bacterial component as an efficient adjuvant in immunocompromised individuals.

Keywords: dendritic cell, inflammasome, HIV, flagellin, NLRC4, adjuvant

# INTRODUCTION

Dendritic cells (DC) are a specialized professional antigen presenting cells (APC) with unique capability to initiate and maintain primary immune responses when pulsed with antigens (1–3). Following recognition of pathogen- or damage-associated molecular patterns (PAMPs or DAMPs, respectively) by innate pattern recognition receptors (PRRs), DC activate and turn into a potent APC. The final differentiation of activated DC is characterized by plasma membrane up-regulation of MHC-II and co-stimulatory molecules (i.e., CD86, CD80), and the production of cytokines important for T CD4<sup>+</sup> lymphocytes activation at the immunologic synapsis (i.e., IL-12 and IL-18; and/or IL-1ß; or IL-4). The result of DC activation drives the polarization of T CD4<sup>+</sup> lymphocytes, and therefore of the immune response (4).

Successful vaccine preparations have to properly activate DC to induce a long-term memory protective immunity. Together with pathogen' antigens, adjuvants strongly contribute to the effectiveness of a vaccine. Their action is mediated by DC PRRs, such as Toll-like receptors (TLRs) and NACHT and LRR containing receptors (NLRs), through the activation of intracellular pathways leading to the production of cytokines important for T cell activation (5). Alum (aluminum hydroxide), a commonly used adjuvant, activates murine DC through the induction of the NLRcontaining a PYD domain 3 (NLRP3) and the consequent mounting of the cytoplasmic complex, known as inflammasome, which results in caspase-1 activation and IL-1ß and IL-18 production. The absence of NLRP3 results in the loss of adjuvant responsivity (6), emphasizing the central role of inflammasome in the activation of DC and in the induction of an efficient immune response.

Accordingly, recent findings have reported that individuals with a low response to vaccines, such infants (7) or cancer patients (8) present a substantial alteration in inflammasome expression and/or activity.

Immune response to many current vaccines is known to be impaired and/or less effective in chronically HIV-infected individuals (9). This impairment has been associated to both a reduced frequency of DC (10, 11), together with phenotypic and functional alterations of these cells (12). As HIV-infected patients present well-documented DC impairment, it has been proposed that a poor response to vaccination could be caused by a diminished and/or defective response to common adjuvants (13).

We have previously demonstrated that NLRP3 inflammasome is not correctly activated by bacterial lipopolysaccharide (LPS) in monocyte-derived dendritic cells (MDDC) from HIV-infected patients (HIV-DC) (14), possibly as a result of the HIVassociated chronic inflammation and the consequent immune system exhaustion (15). As NLRP3 inflammasome is involved in the activation of DC by vaccine adjuvants (6, 16), the defect observed in NLRP3 inflammasome possibly contributes to the less extend immunization response in HIV-infected individuals.

To counteract the low immune response, new vaccination strategies have been proposed, such as the use of PRRs agonists, such as LPS or flagellin (FLG), as largely reviewed in (17).

FLG is the main component of a bacterial flagellum, and it is recognized extracellularly by TLR5 inducing a Myd88 signaling and promoting the transcription of NF-κB-related genes (18, 19); and by the intracellular receptors, NLR-containing a BIR domain (NAIP) and NLR-containing a CARD domain (NLRC)-4. NAIP directly binds FLG, while NLRC4 recognizes the NAIP:FLG complex and mounts the NAIP/NLRC4 inflammasome, resulting in IL-1ß and IL-18 production (20–22).

FLG has already been used as an adjuvant in a number of clinical trials of healthy individuals (23, 24), however to our knowledge none or poor data are available about its ability to activate MDDC in both healthy or immuno-compromised individuals via NAIP/NLRC4 inflammasome.

Taking in account the key role of the inflammasome in proper DC activation, and the impairment of NLRP3 activation observed in HIV-DC, we hypothesize that FLG could represent an alternative adjuvant for HIV-infected patients, by activating inflammasome in DC through NAIP and NLRC4 receptors, and in this way by-passing the NLRP3 defect. Therefore, the aim of this study was to demonstrate that FLG is able to activate MDDC from HIV-infected individuals better than LPS, and to what extent the entity of inflammasome activation differs between HIV-DC and HD-DC.

### MATERIALS AND METHODS

#### HIV-Infected Patients

Twenty-seven HIV-infected adults patients (16 males/11 females; 51.9 ± 11.7 years), proceeding from the metropolitan area of São Paulo (SP, Brazil), seropositive for at least 5 years (26.9 ± 16.9 years), in antiretroviral therapy (ART), with blood CD4<sup>+</sup> T cells count >500 cells/µl, without clinical AIDS or other chronic diseases (i.e., neoplasias, cardiovascular disease, autoimmune disease, kidney disease, obesity) or infectious diseases (i.e., human T-lymphotropic virus/HTLV, hepatitis B or hepatitis C virus), were recruited from January 2016 to May 2018 at the "Serviço de Extensão ao Atendimento de Pacientes HIV/AIDS" (SEAP) of the Faculty of Medicine, University of São Paulo (São Paulo, SP, Brazil). Fifty milliliter of the peripheral blood was collected in heparin tubes. All volunteers assigned the informed consent approved by the Institutional Ethical Committee. Patients' main characteristics are summarized in **Table 1**.

### Healthy Donors (HD)

Twenty-seven adults (15 males/12 females; 45.5 ± 13.4 years), proceeding from the metropolitan area of São Paulo (SP, Brazil), without clinical HIV or other chronic or infectious diseases, were recruited at the Blood Bank Service of the Hospital "Oswaldo Cruz" (São Paulo, SP, Brazil). Fifty milliliter of the peripheral blood was collected in heparin tubes. All volunteers assigned the informed consent approved by the Hospital Ethical Committee. HD demographic data were included in **Table 1**. Of note, any significant difference exists in gender ratio (Fisher test p > 0.05) or age mean value (t- test p > 0.05) between HD and HIV-infected patients.



Gender and age are reported for healthy donors (HD) and HIV-infected patients (HIV). Time from HIV-1 diagnosis as well as plasma viral load (PVL) and CD4<sup>+</sup> T cells counts at the time of blood collection (\*), before (<sup>0</sup> ) and after (<sup>1</sup> ) the start of anti-retroviral therapy (ART) were included for HIV-infected patients. The detection limit of PVL was 1.70 log HIV-1 RNA copies/ml. M, males; F, females; n, number of individuals; SD, standard deviation.

#### Human Monocyte-Derived Dendritic Cells (MDDC)

Mononuclear cells were isolated from 50 mL of peripheral blood by Ficoll-Hypaque (GE Healthcare) density gradient, and monocytes were separated from lymphocytes by plastic adherence. Briefly, 4 × 10<sup>6</sup> mononuclear cells/well were incubated in 24-wells culture plates (Corning-Costar). After 2 h, non-adherent cells (mainly lymphocytes) were removed and cryopreserved at −80◦C for co-culture assays, while adherent cells (mainly monocytes) were cultured with 50 ng/mL GM-CSF (Peprotech) and 50 ng/mL IL-4 (Peprotech) in RPMI-1640 (Gibco, Thermo Fisher Scientific) supplemented with 10% of fetal bovine serum (FBS; Gibco) at 37◦C in 5% CO<sup>2</sup> for 5 days (25). Monocytes-to-DC differentiation was confirmed by flow cytometry analysis of CD14 and CD11c surface markers (**Supplementary File 1**).

MDDC (HIV-DC or HD-DC) were stimulated with purified 5µg/mL FLG from S. typhimurium (FLA-ST, Invivogen) or 1µg/mL LPS from E. coli (Sigma-Aldrich, Merck) for 3, 8, 18, and 24 h and 1 mM ATP was added for more 15 min (26). In some experiments, MDDC were pre-incubated with 10µg/mL MCC-950 (Invivogen), a specific NLRP3 inhibitor (27), or 10µM parthenolide (PTD; Sigma-Aldrich, Merck), a NF-κB and caspases inhibitor (28); or with 1,000 UI/mL IFN-α-2b (Schering-Plough) for 18 h (29, 30). Cell supernatants were collected for cytokines measurement. MDDC were used for cytometric assays or RNA isolation and genes expression analysis.

To assess MDDC ability to activate CD4<sup>+</sup> T lymphocytes, 0.4 × 10<sup>5</sup> MDDC were distributed in 96-well U-bottom suspension culture plates with 4.0 × 10<sup>5</sup> autologous lymphocytes (cryopreserved non-adherent peripheral blood mononuclear cells) (co-culture MDDC/lymphocytes ratio: 1/10) in duplicates and cultured in the presence of unspecific (not antigen-specific) stimuli for 96 or 120 h to measure IFN-γ production and lymphocytes proliferation, respectively. Lymphocytes alone were used as a negative control (Neg).

#### MDDC Phenotype Analysis

2 × 10<sup>5</sup> DC/mL were incubated in Phosphate Buffer Saline (PBS; Sigma-Aldrich, Merck) with the optimal dilution of anti-CD14 PE (MEM15; Exbio), anti-CD11c (3.9; Biolegend), anti-HLA-DR V500 (G46-6; BD Biosciences), anti-CD86 PE-cy7 (2331/FUN-1, B), and anti-CD40 Horizon 450 (5C3; BD Biosciences) antibodies for 20 min at 4◦C. Cells were then washed twice with PBS and resuspended in 200 µL of 4% Formaldehyde-PBS. The Live/Dead Fixable Cell Stain Kit (Life Technology, Thermo Fisher Scientific) was added to the assay according to the manufacturer's instructions. A minimum of 50,000 events was acquired on a LSRFortessaTM X-20 flow cytometer (BD Biosciences) using the FACS Diva software (BD Biosciences). Data were analyzed using the FlowJo software (Tree Star). The gates strategy for one representative experiment was reported in **Supplementary File 2**.

# CD4<sup>+</sup> T Lymphocytes Activation Assay

CD4<sup>+</sup> T lymphocytes activation was measured by the meaning of intracellular staining of IFN-γ. Briefly, autologous co-cultures of MDDC and lymphocytes were treated with lymphocyte mitogens Ionomycin (1µg/mL; Sigma-Aldrich, Merck) and Phorbol Myristate Acetate (10 ng/mL; Sigma-Aldrich, Merck) for 96 h (31). Twenty microgram per milliliter Brefeldin A (Sigma-Aldrich, Merck) was added 6 h before the end of co-culture to block Golgi secretory pathway. Cells were then labeled for surface marker anti-CD3 APC (MEM-57) and anti-CD4 PE (RPAT4) (BD Biosciences), permeabilized with Cytofix/Cytoperm solution (BD Biosciences), and finally stained for anti-IFN-γ V450 (B27; BD Biosciences). The Live/Dead Fixable Cell Stain Kit was added to the assay according to the manufacturer's instructions. Cells were then washed twice with PBS and resuspended in 200 µL of 4% Formaldehyde-PBS to proceed to flow cytometry analysis as above-mentioned.

# CD4<sup>+</sup> T Lymphocytes Proliferation Assay

The CFSE Cell Division Tracker Kit (Biolegend) was used for flow cytometry analysis of in vitro CD4<sup>+</sup> T cells proliferation assay according to manufacturers' protocol. Briefly, autologous lymphocytes were pre-treated with 0.1µM CFSE before being added to MDDC cultures in the presence of the lymphocyte mitogen Concanavalin A (5µg/mL; Sigma Aldrich, Merck) for 120 h (32). At the end of assay, cells were then labeled for anti-CD3 APC (MEM-57) and anti-CD4 PE (RPAT4) (BD Biosciences). The Live/Dead Fixable Cell Stain Kit was added to the assay according to the manufacturer's instructions. Cells were then washed twice with PBS and resuspended in 200 µL of 4% Formaldehyde-PBS to proceed to flow cytometry analysis as above-mentioned.

#### Cytokines Measurement in Culture Supernatants

IL-1β, IL-18, TNF-α, and IL-12p70 were measured in MDDC culture supernatants by ELISA according to the manufacturers' protocols (Biolegend for IL-1β; IL-18, TNF; eBioscience for IL12p70). Data were reported as pg/mL.

#### Caspase-1 Activity Assay

The detection of caspase-1 activity in MDDC was measured with the FAM FLICA Caspase-1 Assay Kit (Immunochemistry Technologies) and flow cytometry according to the manufacturer's protocol. Briefly, 10 µL 30x FLICA was added to 2 × 10<sup>5</sup> MDDC in 300 mL and cells incubated for 1 h at 37◦C 5% CO2. The Live/Dead Fixable Cell Stain Kit was used. Cells were then washed twice with PBS and resuspended in 200 µL of 4% Formaldehyde-PBS to proceed to flow cytometer analysis as above-mentioned. Live MDDC were gated based on their forward (FSC) and side light scatter (SSC). Histograms for one representative experiment was reported in **Supplementary File 3**.

#### Inflammasome Genes Expression Analysis

Total RNA was isolated from 2 × 10<sup>5</sup> MDDC using the RNAqueous-Micro kit (Ambion, Thermo Fisher Scientific) according to manufacturer's protocol and quantified using Nanodrop N-1000 (Agilent). 0.5 µg of total RNA was converted into cDNA using Superscript III RT kit and random primers (Invitrogen, Thermo Fisher Scientific). NLRP1 (hs00248187), NLRP3 (hs00366465), NAIP (hs03037952), NLRC4 (hs00368367), CASP1 (hs00354836), IL1B (hs01555410), IL18 (hs01038788), CARD8 (hs01088221), BRCC3 (hs02386484), and NEDD8 (hs01921826) genes were amplified using TaqMan <sup>R</sup> gene-specific assays (Applied Biosystems, Thermo Fisher Scientific) and qPCR on the QuantStudio 3.0 Real-Time PCR equipment (Applied Biosystems, Thermo Fisher Scientific). The QuantStudio 3.0 software was used to obtain cycle threshold values (Ct) for relative gene expression analysis according to Fold Change (FC) method (33). Raw expression data (Ct) were normalized with the expression of the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase/GAPDH (hs02758991; TaqMan <sup>R</sup> assay) (1Ct), and the FC was calculated comparing stimulated and unstimulated (UN) conditions (FC = 2 <sup>−</sup>11Ct; 11Ct = 1Ctstimulated-1CtUN). Alternatively, the basal (constitutive) gene expression was calculated as 2−1Ct .

#### miR-223 Expression Analysis

Total RNA was isolated from 2 × 10<sup>5</sup> MDDC by mirVanaTM miRNA Isolation Kit (Ambion, Thermo Fisher Scientific) according to the manufacturer's instructions. 0.5 µg of total RNA were converted into cDNA using kit TaqManTM MicroRNA Reverse Transcription and miRNA-specific primers (Applied Biosystems, Thermo Fisher Scientific). miR-223 was amplified using TaqMan <sup>R</sup> miR-specific assays (TM:002098; Applied Biosystems, Thermo Fisher Scientific) and qPCR on the QuantStudio 3.0 Real-Time PCR equipment. The QuantStudio 3.0 software was used to obtain cycle threshold values (Ct) for relative gene expression analysis according to Fold Change (FC) method (33). Raw expression data (Ct) were normalized with the expression of and non-coding small RNA control U6 (TM:001973; TaqMan <sup>R</sup> assay) (1Ct), and the FC was calculated comparing stimulated and unstimulated (UN) conditions (FC=2 <sup>−</sup>11Ct; 11Ct= 1Ctstimulated-1CtUN). Alternatively, the basal (constitutive) gene expression was calculated as 2−1Ct .

#### Detection of "Specks" Formation

Detection of inflammasome "specks" formation was performed by confocal microscopy and immunofluorescence, as previously described (34). Briefly, 2 × 10 <sup>5</sup> MDDC were cultured in 16 wells chamber slides (Thermo Fisher Scientific) and stimulated with 5µg/mL FLG or 1µg/mL LPS for 24 h at 37◦C 5% CO<sup>2</sup> with or without 1 mM ATP. Cells were then fixed and permeabilized with Cytofix/Cytoperm reagent (BD Biosciences) for 30 min at 37◦C 5% CO2, and incubated with primary antibody for NLRP3 (1:100 mouse anti-human NLRP3, Abcam) and/or NLRC4 (1:200 rabbit anti-human NLRC4; Biolegend) overnight at room temperature. Fluorescent secondary antibodies (Alexa 488-conjugated goat-anti-mouse IgG1, or Alexa 647-conjugated goat-anti-rabbit IgG1; Biolegend) were then added for 1 h. Finally, cells were washed and fixed to image acquisition at DMi8 confocal laser scanning microscope (Leica). 4′ ,6-Diamidine-2′ phenylindole dihydrochloride (DAPI; Sigma-Aldrich, Merck) was used for nuclear counterstaining. ImageJ software and related plugins (National Institutes of Health) were used for image processing. The counting of NLRP3+ and NLRC4+ specks in MDDC was performed manually by observing specks formation within the cells (34), and through the ImageJ software by calculating the corrected total cellular fluorescence (CTCF) for each marker as integrated density–(area of selected cell × mean fluorescence of background readings) (35).

#### NLRC4 and NLRP3 Silencing

Pre-validated shRNA for human NLRC4 and NLRP3 was obtained from MISSION <sup>R</sup> shRNA Plasmid DNA (Sigma-Aldrich, Merck). The shRNAs for NLRP3 and NLRC4 used in the study are listed in **Supplementary File 4**. 2 × 10<sup>5</sup> MDDC were transduced using with the same amounts of lentiviral particles encoding non-targeting control or gene-specific shRNA in the presence of SIV3+ VLP for 48 h. Thereafter, cells were treated with 5µg/mL FLG or 1µg/mL LPS for 24 h. The shRNA knockdown efficiency of the target protein in lentivirus-transduced cells was assessed by gene expression (**Supplementary File 5**). The concentrations of IL-1β in cell culture supernatants was measured by ELISA.

#### Data Analysis

All data were collected and analyzed from at least three independent experiments. Normality test was applied to the data, and parametric or non-parametric analysis was used accordingly to compare two or more data sets as specified for each graph. The level of significance was p < 0.05. Calculations were performed using the statistical software package GraphPad Prism 7.0.

#### Biosecurity and Institutional Safety Procedures

All research was performed following the guidelines of biosecurity and safety of Institute of Biomedical Science (ICB/USP).

### RESULTS

#### Flagellin Similarly Activates MDDC From HIV-Infected Patients and Healthy Donors

MDDC were treated with 5µg/mL FLG for 24 h and phenotypic profile, TNF and IL-12 secretion as well as CD4<sup>+</sup> T lymphocytes activation in co-culture experiments were assessed based on previously published protocols (31, 32).

FLG activates MDDC, both HIV-DC and HD-DC, as indicated by the increase of co-stimulatory molecules, cytokines release and CD4<sup>+</sup> T lymphocytes activation (**Figure 1**).

As expected, FLG induced the up-regulation of HLA-DR (95.83 ± 0.68 % positive cells) compared to untreated cells (UN: 77.5 ± 4.4 % positive cells) and the significant increase of CD40 (FLG: 39.8 ± 3.7 %, vs. UN: 7.8 ± 2.0 % positive cells; p = 3 × 10−<sup>5</sup> ) in HD-DC, but not of CD86, which appeared to be decreased in HD-DC+FLG compared to untreated cells (FLG: 62.8 ± 4.3 %, vs. UN: 91.1 ± 2.9 % positive cells; p = 0.001) (**Figure 1A**). However, this result could be due to a previously reported positive feedback mechanism and not to a negative effect of FLG (36). A significant augment of HLA-DR (FLG: 83.5 ± 6.1 %, vs. UN: 49.2 ± 9.5 % positive cells; p = 0.006) and CD40 (FLG: 48.7 ± 5.4 %, vs. UN: 17.4 ± 1.7 % positive cells; p = 0.020) was observed in HIV-DC challenged with FLG. The expression of CD86 did not change in treated or untreated cells (**Figure 1A**).

It is interesting to emphasize that the entity of surface markers expression did not significantly differ between HIV-DC and HD-DC in both untreated or treated conditions (**Figure 1A**). Even if we observed lower viability in HIV-DC compared to HD-DC, this difference did not result statistically significant (p > 0.999).

FLG induced the secretion of a good and similar amount of TNF in HD-DC and HIV-DC (p < 0.05) (**Figure 1B**). On the other hand, the production of IL-12 differs between HD-DC and HIV-DC (p = 0.029): while HD-DC produced significant level of IL-12 in response to the molecular pattern (FLG: 212.7 ± 53.1 pg/mL, vs. UN: 50.5 ± 2.9 pg/mL; p = 7 × 10−<sup>4</sup> ), the induction of cytokine appeared to be less pronounced in HIV-DC (FLG: 63.9 ± 6.9 pg/mL, vs. UN: 38.2 ± 12.1 pg/mL; p > 0.05) (**Figure 1C**).

Altogether these data indicate that FLG is able to activate HIV-DC in a similar way to that seen for HD-DC.

To test whether FLG-treated MDDC are able to induce a properly adaptive immunity response, a MDDC/lymphocytes coculture assay was performed using autologous cells as previously described (31, 32). Lymphocytes alone were used as a negative control (Neg) (**Figures 1D,E**).

A significant increment of IFN-γ <sup>+</sup> CD4<sup>+</sup> T cells was observed in healthy donors (FLG: 12.9 ± 5.5 % vs. UN: 1.3 ± 0.4 % positive cells; p = 0.022) as well as in HIV-infected individuals (FLG: 9.6 ± 1.8 % vs. UN: 0.1 ± 0.1 % positive cells; p = 0.014). Of note, the percentage of IFN-γ <sup>+</sup> CD4<sup>+</sup> T cells after FLG treatment was similar in healthy donor and patients (p > 0.999). Negative control resulted similar to untreated co-cultures (p > 0.05) (**Figure 1D**), emphasizing that the increasing percentage of positive cells is not an artifact.

Moreover, FLG-treated MDDC were able to induce a significant proliferation of CD4<sup>+</sup> T lymphocytes in healthy donors (FLG: 51.0 ± 12.1 %, vs. UN: 3.7 ± 0.9 % positive cells; p = 0.026) and patients (FLG: 24.3 ± 2.8 %, vs. UN: 3.5 ± 0.8 % positive cells; p = 0.038), even if in a lesser extent in patients compared to HD (p = 0.004). Negative control presented a proliferation rate similar to untreated co-cultures (p > 0.05) (**Figure 1E**).

It is interesting to underline that the limited percentage of positive cells in this type of assay is in accord with previously published data for autologous MDDC and T cells co-culture both in healthy donors (32) and even in HIV-infected patients (31) treated with unspecific stimuli (mitogens). Moreover, despite its limited entity, the activation of lymphocytes is consequence of MDDC stimulation as in the absence of MDDC (negative control) mitogens cannot activate T cells (**Figures 1D,E**).

Taking in account the activation status of FLG-treated HIV-DC (**Figures 1A–C**) together with their ability to induce lymphocytes activation (**Figures 1D,E**), we clearly demonstrated that FLG is able to activate HIV-DC similarly to what observed for HD-DC. It is interesting to emphasize that, on the contrary, we have previously shown that HIV-DC did not properly respond to bacterial LPS (14).

#### Flagellin, but Not LPS, Induces Comparable Inflammasome Activation in MDDC From HIV-Infected Patients and Healthy Donors

We then investigated the ability of flagellin to stimulate inflammasome in MDDC by the meaning of inflammasome cytokines production and caspase-1 activity. LPS alone or in combination with ATP was added to the assay as a positive control for inflammasome or NLRP3 inflammasome activation, respectively (26) (**Figure 2**). A time-course assay treating MDDC with 5µg/mL FLG or 1µg/mL LPS was performed to determine the best experimental time for IL-1ß detection in this model, (**Supplementary File 6**), and selected 24 h for all the experiments.

FLG induced a significant IL-1ß release in HD-DC (155.3 ± 20.7 pg/mL), compared to unstimulated cells (4.0 ± 1.9 pg/mL; p < 0.0001) and similarly to LPS (132.4 ± 9.9 pg/mL; p < 0.0001). The treatment with LPS+ATP resulted in a significant increase of IL-1ß release compared to LPS (p = 0.002), indicating a proper response of NLRP3 inflammasome in HD-DC (**Figure 2A**).

In HIV-DC, the entity of IL-1ß production resulted lower than in HD-DC but significantly augmented compared to resting cells (FLG: 121.3 ± 34.3 pg/mL; vs. UN: 18.3 ± 9.3 pg/mL; p = 2 × 10−<sup>4</sup> ) and also to LPS-treated cells (84.4 ± 17.5 pg/mL; p = 0.002). However, ATP did not alter LPS-induced IL-1ß production (p > 0.05) (**Figure 2A**), confirming the previously observed dysregulation of NLRP3 inflammasome in MDDC from HIV-infected patients (14).

In a similar way, FLG also induced significantly IL-18 release in HD-DC (FLG: 91.1 ± 19.3 pg/mL, vs. UN: 12.8 ± 6.1 pg/mL; p = 0.018) and at lower extent in HIV-DC (FLG: 50.5 ± 5.6 pg/mL, vs. UN: 2.9 ± 1.5 pg/mL; p = 0.009). HD-DC better respond to LPS and LPS+ATP than HIV-DC also in term of IL-18. There

FIGURE 1 | Flagellin similarly activates HIV-DC and HD-DC. 2 × 10<sup>5</sup> MDDC from healthy donors (HD-DC; n = 5) and HIV-infected patients (HIV-DC; n = 5) were stimulated with 5µg/ml flagellin (FLG) for 24 hours. Viability, expression of characteristic DC surface markers (A) as well as TNF (B) and IL-12p70 (C) secretion were analyzed and compared between untreated (UN) and stimulated (FLG) conditions as well between HD-DC and HIV-DC groups 0.4 × 10<sup>5</sup> FLG-treated MDDC were cultured with 4 × 10<sup>5</sup> autologous lymphocytes (MDDC/lymphocytes ratio: 1/0) for 96 hours to detect IFN-γ production in CD4<sup>+</sup> T lymphocytes (percentage of CD3<sup>+</sup> CD4<sup>+</sup> T IFN-γ <sup>+</sup> cells) (D), or 120 hours to measure CD4<sup>+</sup> T lymphocytes proliferation (percentage of CD3<sup>+</sup> CD4<sup>+</sup> T CSFElow cells) (E). Lymphocytes alone were used as a negative control (Neg). Data are reported as mean ± standard error. Multiple t-test (A) and Two-Way ANOVA test (B–E) were applied to compare conditions within a group (HIV-DC or HD-DC; \*p < 0.05) and between groups (HIV-DC vs. HD-DC; §p < 0.05).

were no statistical differences in the production of IL-18 between HD- and HIV-DC (**Figure 2B**).

Inflammasome cytokines release revealed that FLG is able to induce complex activation in MDDC from healthy as well as HIV-infected individuals. Accordingly, FLG increased caspase-1 activity in HD-DC (FLG: 6.8 ± 2.0 %, vs. UN: 0.3 ± 0.0 % positive cells) and in HIV-DC (FLG: 6.9 ± 0.7 %, vs. UN: 3.5 ± 0.5 % positive cells). HIV-DC presented a constitutively activated caspase-1, however a lower activation in response to LPS+ATP compared to HD-DC, once more emphasizing the specific dysregulation of NLRP3 pathway (**Figure 2C**). Even if these differences did not reach statistical significance (p = 0.07), we underline that caspase-1 activity accompanies above-mentioned cytokines data, and that this is the first study showing a tendency in caspase-1 activation defect in MDDC from HIV-infected individuals. Although other studies have reported statistically significant differences in caspase-1 activity in healthy donors and HIV-infected patients, those results referred to lymphoid compartment or peripheral blood mononuclear cells, while little is known in myeloid cells (37–40).

#### Flagellin Induces Inflammasome Activation by Stimulated NAIP/NLRC4 and NLRP3 Receptors in HD-DC, but Not in HIV-DC

Once assessed that FLG is able to induce inflammasome activation in MDDC, we therefore tried to depict the pathways involved in complex formation and to detect any differences between HD-DC and HIV-DC.

inflammasome and NLRP3 inflammasome, respectively. Culture supernatants were used to measure IL-1β (A) and IL-18 (B) concentration (pg/mL). Cells were harvested for analysis of caspase-1 activity by FAM-FLICA assay and flow cytometry. Percentage of FAM-FLICA+CD11c<sup>+</sup> cells were reported for FLG-treated (FLG) and untreated (UN) MDDC (C). Data are represented as mean ± standard error. Two-Way ANOVA test was applied to compare conditions within a group (HIV-DC or HD-DC; \*p < 0.05; LPS+ATP vs. LPS: #p < 0.05) and between groups (HIV-DC vs. HD-DC; §p < 0.05).

Gene expression analysis revealed that HIV-DC presents a significant higher constitutive expression of IL1B and IL18 compared to HD-DC. NLRP3 also resulted increased even if not in a statistically significant way, while the level of NAIP and NLRC4 was similar between MDDC (**Figure 3A**).

When the effect of flagellin was evaluated in genes modulation, we observed a revealed a different expression profile in HD-DC and HIV-DC at all the analyzed time-points (3, 8, 18, and 24 h) (**Figure 3B**).

In particular, we focused our attention on the two receptors NLRP3 and NLRC4. While, as expected, FLG induces NLRC4 gene modulation in HD-DC and HIV-DC (**Figure 3C**), NLRP3 appeared to be defective in HIV-DC, as FLG was able to induce NLRP3 expression in HD, but not in HIV-DC (**Figure 3D**), according to our previously published data (14).

The dysregulation observed for NLRP3 could be due to an imbalance of inhibitor and activator signals. NLRP3 is tightly regulated by endogenous proteins CARD8 (41), BRCC3 (42) and NEDD8 (43), and by miR-223 (44). Interestingly, the basal expression of CARD8 and BRCC3, as well as of miR-223, resulted significantly augmented in HIV-DC compared to HD-DC (**Figure 3E**), suggesting a possible cause of low responsiveness of NLRP3. Moreover, taking in account that IFN-I also contributes to the negative regulation of NLRP3 (29, 30) and that HIV-infected patients are known to present high level of circulating IFN-I (45–47), we shown that the treatment of HD-DC with IFN-a significantly reduced IL-1ß release specifically in LPS+ATP treated cells up to cytokine level observed in HIV-DC (**Figure 3F**), emphasizing the inhibitory role of IFN-I on NLRP3 and suggesting that this could be another cause of a specific NLRP3 defect in HIV-DC.

According to our initial hypothesis, flagellin appears to be able to by-pass this defect of NLRP3 in HIV-DC, and at the same time our results have shown the involvement of both NLRC4 and NLRP3 receptors in HD-DC response to flagellin.

This hypothesis of a "two-receptors" mechanism is supported also by inflammasome "specks" detection through immunofluorescence staining of NLRP3 and NLRC4 (**Figure 4**). NLRC4+ specks were evidenced in confocal imagines of FLGtreated HD-DC (**Figure 4A**) and HIV-DC (**Figure 4B**). NLRP3+ specks resulted more in FLG-treated HD-DC (**Figure 4A**) than in HIV-DC (**Figure 4B**). In general NLRC4 and NLRP3 staining localized in the same cells. By the use of CTCF index, we showed that FLG significantly induced both NLRC4+ and NLRP3+ specks in HD-DC (FLG vs. UN: p < 0.0001) (**Figure 4C**). On the other hands, in HIV-DC FLG induced preferentially NLRC4+ specks (CTCF FLG vs. UN: p = 0.0001) and at lesser extent NLRP3+ specks (p > 0.05) (**Figure 4D**). These findings demonstrated that in healthy donor cells, FLG not only activates inflammasome through the expected NAIP/NLRC4 pathway but also through the NLRP3 one.

To better investigate the involvement of NAIP/NLRC4 and NLRP3 in response to FLG in our model, we evaluated the IL-1ß production in FLG-treated MDDC previously incubated with 10µM MCC-950, a specific NLRP3 inhibitor (27), or with 10µM parthenolide/PTD, a large spectrum inflammasome inhibitor (28) (**Figures 5A,B**).

MCC-950 and PTD significantly inhibited IL-1ß production in FLG-treated HD-DC (78 and 90% of inhibition, respectively; p < 0.05), as well as LPS+ATP-treated cells (78 and 90% of inhibition, respectively; p < 0.05), and partially LPS-induced inflammasome activation (78 and 90% of inhibition, respectively; p < 0.05) (**Figure 5A**), due to the contribution of other pathway in inflammasome activation by LPS (48).

On the other side, MCC-950 was not able to significantly reduce cytokine release in HIV-DC stimulated with FLG (48% of inhibition; p > 0.05), LPS (48% of inhibition; p > 0.05) or LPS+ATP (48% of inhibition; p > 0.05) (**Figure 5B**). Accordingly, in the presence of MCC-950 a reduction not significantly of NLRP3+ specks was observed in HIV-DC (**Supplementary File 7**). PTD similarly inhibited IL-1β production in both FLG-treated HIV-DC (84% of inhibition p = 0.002), LPS or LPS+ATP-treated HIV-DC (84% of inhibition; p = 0.002), as expected due to

IFN-α-2b pre-treatment and compared with HIV-DC (n = 3) stimulated with FLG. Data are represented as mean ± standard error. Multiple t-test was applied to compare HD- and HIV-DC in (A,C–E). Kruskall-Wallis test was applied to compare HD-DC, HD-DC + IFN-α-2b, and HIV-DC sets in (F) (\*p < 0.05).

the caspase-1 and NF-kB dependency for FLG-induced response (49), as well as for LPS-mediated one (48). These data confirm the greater involvement of NLRP3 in response to FLG in healthy donors compared to HIV-infected individuals.

Then we compared the differences in IL-1ß release in FLGtreated MDDC following NLRC4 or NLRP3 shRNA knockdown (**Figures 5C,D**). In NLRC4 shRNA-transduced MDDC, FLG induced a lower level of IL-1ß production compared to untreated cells both in HD-DC (**Figure 5C**) and in HIV-DC (**Figure 5D**). In NLRP3 siRNA-transducted MDDC, FLG induced a lower level of IL-1ß compared to untreated cells in HD-DC (**Figure 5C**), however this effect was not observed in HIV-DC (**Figure 5D**). These data are in accord with above-reported effect of chemical NLRP3 inhibitor MCC-950 (**Figures 5A,B**).

Altogether these findings allow us to suggest that FLG activates the inflammasome in human MDDC through the induction of NAIP/NLRC4 and NLRP3 receptors; the contribution of NLRP3 in FLG-signaling is less pronounced in HIV-DC due to the well-known "exhausted" profile of these cells (50).

# DISCUSSION

A number of microbial components have been proposed as alternative adjuvants to augment the immune responses of poorly immunogenic vaccines and/or of not fully immunocompetent individuals. Emerging evidence pointed out the possible use of bacterial flagellin in this context, firstly in mice [as extensively revised in Hajam et al. (17)], but also in a human clinical trial of prophylactic vaccine (23). Studies in mice show the stimulatory capacity of flagellin to induce both humoral and cellular immune responses when implied together with pathogen' antigens as adjuvant (51, 52) or in cancer immunotherapy (53).

Taking in account that, even in ART treatment, HIV-infected individuals continue to experience immune dysfunction, leading, among others side effects, to a deficient vaccine response (9), the necessity of new vaccine strategy in this population appear to be urgent.

Using the in vitro model of peripheral blood monocytesderived-dendritic cells developed for HIV-infected patient's immunotherapy by Lu and collaborators (54), we demonstrated

that flagellin is able to activate MDDC from HIV-infected patients as well as from healthy donors. Previous studies have evidenced the ability of flagellin to activate primary human monocytes, or human pro-monocytic cell line U38 (55), however here we demonstrated that flagellin is able to activate human MDDC considering both the MDDC profile as well as the MDDC-mediated lymphocytes activation (**Figure 1**), with the exception of IL-12 production which results lower in HIV-DC compared to HD-DC, but in accord with previously published data (56).

Flagellin is sensed by two main innate immune receptors, TLR5 (18) and NAIP/NLRC4 (57–59). While it has been shown that flagellin induces NAIP/NLRC4 inflammasome activation in primary human macrophages and monocytes (49, 60, 61), little is known about its role in human DC. Despite it has been previously reported that flagellin stimulates IL-1β production in human MDDC (62), any evidence about NLRC4 or NLRP3 pathway in IL-1ß induction have been shown nor hypothesized. Besides the description of these two receptors within inflammasome activation by flagellin, our study also demonstrated for the first time the different contribution of the two receptors in HIV-DC response to flagellin.

As expected, flagellin induces the inflammasome activation through NAIP/NLRC4 both in HD-DC as well as HIV-DC (**Figure 2**), suggesting that this pathway is still effective in HIV-infected patients, contrary to what seen for NLRP3 (14). Actually, HIV-DC presents a basal increased expression of NLRP3 inhibitory molecules, namely CARD8 (41), BRCC3 (42) and miR-223 (44), compatible with specific inhibition of this receptor (**Figure 3E**).

While the low rate of NLRP3 response in HIV-DC has been attributed to the chronic inflammation of HIVinfected patients, consequent of HIV-1 persistence, endogenous viruses reactivation (i.e., herpes simplex virus), immune system exhaustion, increased intestinal permeability and microbial translocation (mainly LPS) and antiretroviral drugs cytotoxicity (15, 63) and taking in account that flagellate bacteria also could be present in gut microbiota, it remains obscure why the NAIP/NLRC4 pathway continues "ready-to-go." The administration of continues doses of LPS was found to be tolerogenic in mice (64), however any data are available about flagellin. We speculated that the increased IFN-a production in HIV-infected individuals (even if at lower levels in ARTtreated patients) could be a possible cause of specific NLRP3 inhibition due to the known effect of the anti-viral mediator as NLRP3 negative regulators (30). It is important to underline that our cohort of ART-treated HIV-infected individuals presented a mean value of plasma IFN-a of about 50 pg/mL, 5-fold more than healthy individuals (about 10 pg/mL) (data not shown). Moreover, when HD-DC were treated with IFN-a their NLRP3 response is diminished similarly to what observed in HIV-DC (**Figure 3F**), supporting our hypothesis. Another possible explication concerns the higher constitutive recruitment of adaptor molecule ASC (Apoptosis speck like with a CARD) in inflammasome complex in HIV-infected patients compared to healthy controls as recently observed by Ahmad and colleagues (65) in PBMC. In this case, inflammasome receptors, which need ASC to mount the complex (i.e., NLRP3), could be disadvantaged in respect to sensors that directly recruit caspase-1, such as NLRC4.

Beyond the main purpose of this article, and for the first time to our knowledge, we demonstrated that NLRP3 also contribute to flagellin response in human MDDC, as revealed by NLRP3+ specks formation in FLG-treated cells (**Figure 4**) and by specific (chemical and genetic) inhibition of this receptor (**Figure 5**). Our findings are in lines with previous reports about NLRC4 and NLRP3 co-localization into a unique complex in HEK293 cells (66) and in mice bone marrow-derived macrophages during Salmonella infection (67).

Therefore, flagellin induces the two pathways in HD-DC, whereas it preferentially activates NLRC4 pathway in HIV-DC due to the defect in NLRP3 in these cells. Despite the immunofluorescence visualization of both NLRP3+ and NLRC4+ specks, our data are not sufficient to determine whether the two receptors are truly co-localized in the same inflammasome complex, and further investigations will be needed to finally prove it.

In conclusion, our data support the use of flagellin in the design of future vaccines effective also in immunocompromised individuals, such as HIV-infected patients. According to our findings, flagellin activates human MDDC from healthy and HIV-infected individuals through the NAIP/NLRC4 inflammasome, with the participation of NLRP3 at least in healthy donors cells. Flagellin was able to by-pass the NLRP3 defect in HIV-DC, contributing to inflammasome activation and consequently full MDDC maturation in HIV-infected patients.

#### ETHICS STATEMENT

All research involving human participants was approved by the Institutional Ethics Committees of Oswaldo Cruz Hospital, Hospital das Clinicas/Faculty of Medicine of the University of São Paulo (FMUSP) and of the Institute of Biomedical Science (ICB/USP). Written informed consent was obtained from all participants, and clinical investigations were conducted according to the principles expressed in the Declaration of Helsinki.

# AUTHOR CONTRIBUTIONS

EdR, BdA, and AP: conceived and designed experiments. EdR: performed MDDC experiments. VL: collection of samples. DdL:

# REFERENCES


performed the immunofluorescence experiments. FF and JdS: ELISA. EdR, DdL, and AP: statistical analysis. EdR, DdL, VL, FF, and AP: discussion of results. EdR and AP: wrote the article. BdA: designed and discussed silencing experiments.

# FUNDING

This project was supported by the Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP) (research grant 2015/23395-6 and 2015/50660-7 to AP; 2014/23225-0 to BdA). ECR (2015/17373-0), VL (2017/10824-1) and FF (2018/04361- 1) receive FAPESP post-graduation fellowships; JdS had received a CAPES master' degree fellowships; DdL receives a Ph.D. fellowship from the Fundação de Amparo à Pesquisa do Estado Amazonas (FAPEAM). AP receives a CNPq Researcher Fellowship. The funders had no role in this study design, data collection and analysis, decision to publish, or preparation of the manuscript.

# ACKNOWLEDGMENTS

We are very grateful to the Oswaldo Cruz Hospital Blood Bank Service and the Serviço de Extensão ao Atendimento de Pacientes HIV/AIDS (SEAP) of the FMUSP (SP, Brazil) for the recruitment of healthy donors and HIV-infected patients, respectively. We acknowledge Prof. Niels Olsen Câmara for slide chambers and qPCR reagents; Dr. Telma Miyuki Oshiro for anti-human NLRP3 antibody, and IL-12p70 ELISA reagents; Prof. Karina Ramalho Bortoluci, Dr. Paula Ordonhez Rigato, and Prof. Silvia Boscardin for helpful suggestions and discussion. We acknowledge Prof. Eliana Faquim and the technician Alexsander Souza for the microscope facility of the Laboratory of Cellular Biology at the Butantan Institute (São Paulo, SP, Brazil) by the use of a DMi8 confocal laser scanning microscope; the Laboratory of Immune Cell Biology (Department of Immunology, ICB, USP) for technical support in silencing experiments.

# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fimmu. 2019.01291/full#supplementary-material


cells in patients infected with HIV-1. Blood. (2003) 101:4505–11. doi: 10.1182/blood-2002-10-3189


continued class II MHC synthesis and antigen processing. J Immunol. (2012) 188:3116–26. doi: 10.4049/jimmunol.1101313


recurrent macrophage activation syndrome. Nat Genet. (2014) 46:1140–6. doi: 10.1038/ng.3089


**Conflict of Interest Statement:** The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Copyright © 2019 Reis, Leal, Soares, Fernandes, Souza de Lima, de Alencar and Pontillo. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.