# NANOPARTICLE VACCINES AGAINST INFECTIOUS DISEASES

EDITED BY : Rajko Reljic and África González-Fernández 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 978-2-88963-327-2 DOI 10.3389/978-2-88963-327-2

#### 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

# NANOPARTICLE VACCINES AGAINST INFECTIOUS DISEASES

Topic Editors: Rajko Reljic, St George's, University of London, United Kingdom África González-Fernández, Biomedical Research Center (CINBIO), Spain

Citation: Reljic, R., González-Fernández, Á., eds. (2020). Nanoparticle Vaccines Against Infectious Diseases. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-327-2

# Table of Contents


Gustaf Lindgren, Sebastian Ols, Frank Liang, Elizabeth A. Thompson, Ang Lin, Fredrika Hellgren, Kapil Bahl, Shinu John, Olga Yuzhakov, Kimberly J. Hassett, Luis A. Brito, Hugh Salter, Giuseppe Ciaramella and Karin Loré

*20 Corrigendum: Induction of Robust B Cell Responses After Influenza mRNA Vaccination is Accompanied by Circulating Hemagglutinin-Specific ICOS+ PD-1+ CXCR3+ T Follicular Helper Cells*

Gustaf Lindgren, Sebastian Ols, Frank Liang, Elizabeth A. Thompson, Ang Lin, Fredrika Hellgren, Kapil Bahl, Shinu John, Olga Yuzhakov, Kimberly J. Hassett, Luis A. Brito, Hugh Salter, Giuseppe Ciaramella and Karin Loré


Roberto Nisini, Noemi Poerio, Sabrina Mariotti, Federica De Santis and Maurizio Fraziano

*70 Virus-Like Particle, Liposome, and Polymeric Particle-Based Vaccines Against HIV-1*

Yong Gao, Chanuka Wijewardhana and Jamie F. S. Mann

*88 Room Temperature Stable PspA-Based Nanovaccine Induces Protective Immunity*

Danielle A. Wagner-Muñiz, Shannon L. Haughney, Sean M. Kelly, Michael J. Wannemuehler and Balaji Narasimhan

*99 Lipid-Based Particles: Versatile Delivery Systems for Mucosal Vaccination Against Infection*

Blaise Corthésy and Gilles Bioley

*119 Mucosal Delivery of Fusion Proteins With* Bacillus Subtilis *Spores Enhances Protection Against Tuberculosis by Bacillus Calmette-Guérin* Alastair Copland, Gil R. Diogo, Peter Hart, Shane Harris, Andy C. Tran, Mathew J. Paul, Mahavir Singh, Simon M. Cutting and Rajko Reljic

*130 Trimethyl Chitosan Nanoparticles Encapsulated Protective Antigen Protects the Mice Against Anthrax*

Anshu Malik, Manish Gupta, Rajesh Mani, Himanshu Gogoi and Rakesh Bhatnagar


Young-Seok Kim, Ahyun Son, Jihoon Kim, Soon Bin Kwon, Myung Hee Kim, Paul Kim, Jieun Kim, Young Ho Byun, Jemin Sung, Jinhee Lee, Ji Eun Yu, Chan Park, Yeon-Sook Kim, Nam-Hyuk Cho, Jun Chang and Baik L. Seong

*212 Nanoparticle Vaccines Against Infectious Diseases* Rashmirekha Pati, Maxim Shevtsov and Avinash Sonawane

# Editorial: Nanoparticle Vaccines Against Infectious Diseases

#### Rajko Reljic<sup>1</sup> \* and África González-Fernández 2,3

1 Institute for Infection and Immunity, St. George's, University of London, London, United Kingdom, <sup>2</sup> Centro de Investigaciones Biomédicas (CINBIO) (Centro Singular de Investigación de Galicia), Universidad de Vigo, Vigo, Spain, 3 Instituto de Investigación Sanitaria Galicia Sur (IIS-GS), Vigo, Spain

Keywords: nanoparticles, vaccine, liposomes, immunity, infection

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

#### **Nanoparticle Vaccines Against Infectious Diseases**

Vaccines remain the most realistic means of controlling, eliminating, and eventually completely eradicating an infectious disease. Many vaccines have been successfully implemented over the course of recent medical history and some have succeeded in completely (small pox) or near-completely eradicating (polio) their target diseases. However, many human pathogens are still in need of an effective vaccine, including viruses such as human immunodeficiency (HIV), herpes simplex (HSV), respiratory syncytial (RSV), Chikungunya, Dengue, Ebola, and Hepatitis C; bacteria such as Mycobacterium tuberculosis or Salmonella dysenteriae and parasites such as Plasmodium spp., to name a few.

Challenges in developing an effective vaccine against these infectious diseases are multiple, ranging from antigenic variability (HIV), multiple genotypes or serotypes (Norovirus, Dengue, Influenza), complex pathogenesis or life cycle of pathogen (tuberculosis, malaria), lack of correlates of protection (TB) and finally, the paucity of the adjuvants and delivery systems suitable for human application. With regards to the latter, it is true to say that non-live, subunit vaccine development has been historically severely hampered by safety issues associated with the adjuvants and delivery systems.

Nanoparticles (NPs) have emerged as a promising approach for vaccine delivery as both antigen delivery platform and immunomodulators. Their use in vaccine candidates from early preclinical to late stage clinical testing is a testament to their success as a promising, new approach, alongside the more conventional protein plus adjuvant, viral vector, or attenuated whole organism vaccine approaches. There are many different types of NPs, both synthetic and natural, as well as further varieties that are not considered classically as NPs, but are physically and functionally closely related to them, such as liposomes, virus-like particles, bacterial spores, and immunostimulating complexes.

The ability of NPs to interact with immune components and to induce humoral and cellular immune responses make them particularly amenable for vaccine design. Furthermore, they have been successfully applied by different routes (systemic and throughout the mucosa), and have been demonstrated capable of modifying and broadening of the immune profiles. They can increase antigen stability (time, temperature, proteolysis) and confer substantial flexibility to vaccine formulation, allowing the incorporation of diverse antigens and immunostimulants, compared to conventional subunit/adjuvant vaccines.

Many of these advantages are highlighted in this Research Topic but it is also necessary to draw attention to the areas for further improvements, including a better understanding of the mechanisms of NPs vaccine immunogenicity and a more efficient antigen presentation by the molecular histocompatibility complex molecules, especially the so called cross-presentation by class I molecules performed by specialized dendritic cells, following the endocytic/phagocytic uptake of NPs. Also, and very importantly, NPs design and technology may need to be

Edited and reviewed by: Denise Doolan, James Cook University, Australia

> \*Correspondence: Rajko Reljic rreljic@sgul.ac.uk

#### Specialty section:

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

> Received: 20 September 2019 Accepted: 21 October 2019 Published: 07 November 2019

#### Citation:

Reljic R and González-Fernández Á (2019) Editorial: Nanoparticle Vaccines Against Infectious Diseases. Front. Immunol. 10:2615. doi: 10.3389/fimmu.2019.02615

**5**

further improved to allow for an "off-the-shelf " approach to their application and testing rather than the complex and time/resource demanding process of customized production.

In this Research Topic, we include several articles that focus on NP-based vaccine delivery against infectious diseases and also review articles that summarize and critically assess the progress that has been achieved so far in the specific areas of NP vaccine development. It is hoped that the presented evidence will deepen our understanding of their mode of action and the overall potential of NPs for translation of this vaccine approach to human application.

Thus, Kim et al. reported a novel method of generating biodesigned NPs utilizing a bacterial expression system and the capacity of RNA molecules to act as chaperones. Using Middle East respiratory syndrome-coronavirus (MERS-CoV) as the infection target, they demonstrated that NPs can be assembled in the expression host and that this was entirely dependent on chaperoning capacity of RNA since mutations in the RNAbinding domain abolished formation of NPs. The resulting NPs were immunogenic in mice and induced blocking antibodies against MERS virus. This approach of generating protein-only based NPs could be further optimized (i.e., in terms of yields and homogeneity) and developed as a generic NPs platform against other infectious diseases.

In another article, the utility of mucosally delivered chitosanbased vaccine against swine influenza A virus (SwIAV) was demonstrated by Dhakal et al., who reported that strong crossreactive mucosal IgA and cellular immune responses in the respiratory tract were induced in young piglets using this vaccination approach. Intranasal delivery of these NPs loaded with SwIAV antigens resulted in a reduced nasal viral shedding and lung virus titres in pigs, suggesting that this vaccination approach could offer a broader coverage than the current attenuated strain-specific SwIAV.

Chitosan was also tested, alongside arginine-rich protamine (PR) and polyarginine (PARG)-based NPs in a more mechanistic in our own studies (Peleteiro et al.), reporting that PR and PARG showed a superior immunomodulating ability, as measured by enhanced reactive oxygen species production, activation of the complement cascade, cytokine production, and mitogen-activated protein (MAP) kinases/nuclear factor κB activation. When complexed with recombinant Hepatitis B glycoprotein, and compared against each other, protamine-based NPs elicited higher IgG levels than PARG NPs.

The utility of NPs as a delivery system for immunododulators and adjuvants rather than antigens was illustrated in the article by Takahashi et al. They showed that the biodegradable carbonate apatite (CA)-based nanoparticles can serve as the delivery system for the CpG oligodeoxynucleotide (CpG ODN) adjuvant and that this combination was more potent in activating dendritic cells and induced more diverse cytokine profiles than CpG ODN alone. When used with a model antigen, the NPs/CpG ODN induced higher level humoral and cellular immune responses in mice, and in particular enhanced CD8+ T cell responses, suggesting that this vaccination approach is particularly suitable for viral infections which require cytotoxic T cells alongside the neutralizing antibodies to control the viremia.

Another example of the benefit of combining NPs with adjuvants was provided also in the study by Malik et al., who showed that an anthrax antigen when combined with trimethyl-chitosan nanoparticles (TMC-PA) and either CpG ODN or polyinosinic: polycytidylic acid (Poly I:C) adjuvant induced higher immune responses in mice than any other combination in that vaccine formulation. In a different approach, Wagner-Muniz et al. reported that laboratory-generated polyanhydride nanoparticles based on 1,8-bis-(p-carboxyphenoxy)-3,6-dioxaoctane (CPTEG), 1,6 bis-(p-carboxyphenoxy)hexane (CPH) and sebacic acid (SA), could be employed as an effective vaccination strategy against Streptococcus pneumoniae, with strong immune responses and good-level of protection after the challenge of mice with the bacteria.

Finally, in our own studies (Copland et al.), we demonstrated that mucosal delivery of a mycobacterial poly-antigen coated onto heat-inactivated Bacillus subtilis spores ("Spore-FP1") induced systemic and mucosal immune responses in mice, characterized with elevated antigen-specific IgG and IgA titres in the serum and bronchoalveolar lavage, antigen-specific memory T-cell proliferation in both CD4<sup>+</sup> and CD8<sup>+</sup> compartments, and resident memory T cells accumulation in the lungs. When used to boost the current TB vaccine, BCG, this vaccine candidate provided superior protection in mice challenged with aerosolised M. tuberculosis.

In addition, the collection also includes two review articles on lipid-based particles as a highly versatile vaccine delivery system. Thus, Corthesy and Bioley reviewed the potential of liposomes and liposomes derivatives as mucosal vaccine delivery systems, while Nisini et al. performed a critical assessment regarding the application of liposomes in a broader context of infectious diseases. Gao et al. provided an extensive review of the current state of virus-like particles (VLP) as an emerging and highly attractive vaccine delivery system, focusing in particular on the use of VLP in the context of HIV infection. And finally, Pati et al. reviewed the current state of NPs-based vaccines against infectious diseases, highlighting the key challenges and the potential for further progress.

The promising results obtained with several type of NPs highlight the potential of this vaccine approach for the development of new vaccines in the near future. However, further studies are still required to address in greater detail the issues regarding their safety, immunogenicity, stability, cost, scaling-up potential, and the use of appropriate animal models and clinical assays in humans. Even so, the significant body of evidence already generated, as partly illustrated in this Research Topic, underscores the translational potential of NPs in vaccine research and development, not only in the context of infectious diseases but potentially also other conditions such as autoimmune diseases and cancer.

#### AUTHOR CONTRIBUTIONS

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

## FUNDING

This work was financially supported by the EU Horizon2020 Eliciting Mucosal Immunity in Tuberculosis (EMI-TB) project (Grant No. 643558) and the Xunta de Galicia Grupo Referencia Competitiva 2016 (ED431C 2016/041).

**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 © 2019 Reljic and González-Fernández. 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:*

*Rajko Reljic, St George's, University of London, United Kingdom*

#### *Reviewed by:*

*Giuseppe Lofano, Ragon Institute of MGH, MIT and Harvard, United States Ali Ellebedy, Washington University in St. Louis, United States*

#### *\*Correspondence:*

*Karin Loré karin.lore@ki.se*

#### *Specialty section:*

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

*Received: 12 September 2017 Accepted: 27 October 2017 Published: 13 November 2017*

#### *Citation:*

*Lindgren G, Ols S, Liang F, Thompson EA, Lin A, Hellgren F, Bahl K, John S, Yuzhakov O, Hassett KJ, Brito LA, Salter H, Ciaramella G and Loré K (2017) Induction of Robust B Cell Responses after Influenza mRNA Vaccination Is Accompanied by Circulating Hemagglutinin-Specific ICOS+ PD-1+ CXCR3+ T Follicular Helper Cells. Front. Immunol. 8:1539. doi: 10.3389/fimmu.2017.01539*

*Gustaf Lindgren1,2, Sebastian Ols1,2, Frank Liang1,2, Elizabeth A. Thompson1,2, Ang Lin1,2, Fredrika Hellgren1,2, Kapil Bahl <sup>3</sup> , Shinu John3 , Olga Yuzhakov3 , Kimberly J. Hassett3 , Luis A. Brito4 , Hugh Salter4,5, Giuseppe Ciaramella3 and Karin Loré1,2\**

*1Department of Medicine Solna, Immunology and Allergy Unit, Karolinska Institutet, Stockholm, Sweden, 2Center for Molecular Medicine, Karolinska Institutet, Stockholm, Sweden, 3Valera LLC, Cambridge, MA, United States, 4Moderna Therapeutics, Cambridge, MA, United States, 5Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden*

Modified mRNA vaccines have developed into an effective and well-tolerated vaccine platform that offers scalable and precise antigen production. Nevertheless, the immunological events leading to strong antibody responses elicited by mRNA vaccines are largely unknown. In this study, we demonstrate that protective levels of antibodies to hemagglutinin were induced after two immunizations of modified non-replicating mRNA encoding influenza H10 encapsulated in lipid nanoparticles (LNP) in non-human primates. While both intradermal (ID) and intramuscular (IM) administration induced protective titers, ID delivery generated this response more rapidly. Circulating H10-specific memory B cells expanded after each immunization, along with a transient appearance of plasmablasts. The memory B cell pool waned over time but remained detectable throughout the 25-week study. Following prime immunization, H10-specific plasma cells were found in the bone marrow and persisted over time. Germinal centers were formed in vaccine-draining lymph nodes along with an increase in circulating H10-specific ICOS+ PD-1+ CXCR3+ T follicular helper cells, a population shown to correlate with high avidity antibody responses after seasonal influenza vaccination in humans. Collectively, this study demonstrates that mRNA/LNP vaccines potently induce an immunological repertoire associated with the generation of high magnitude and quality antibodies.

#### Keywords: mRNA vaccine, adaptive immune responses, non-human primates, influenza, T follicular helper cells, germinal centers

**Abbreviations:** ASCs, antibody-secreting cells; cTfh, circulating T follicular helper; GCs, germinal centers; GLA, glucopyranosyl lipid adjuvant; HA, hemagglutinin; ID, intradermal; IM, intramuscular; LNP, lipid nanoparticle; LN, lymph node; PC, plasma cell; PD-1, programmed death receptor 1; Th, T helper; Tfh, T follicular helper.

## INTRODUCTION

Emerging infections such as Ebola, Zika, Chikungunya, and pandemic influenza virus need vaccines that can be rapidly produced with antigen precision. Modified mRNA vaccines have received considerable attention as they are attractive in this aspect and were recently shown to induce sterilizing immunity to Zika virus infection and protection against lethal challenge with influenza virus (1–5). Advances in mRNA synthesis technology have led to increased mRNA stability, optimized translation capacity, and less indiscriminate activation of innate immunity by mRNA vaccines (2, 3, 6). mRNA vaccines do not require device-mediated delivery methods as with DNA vaccines, and there is no concern of pre-existing immunity typically associated with viral vector vaccine platforms (3). mRNA vaccines delivered in lipid nanoparticles (LNP) are well tolerated and highly immunogenic in mice, ferrets, non-human primates, and humans (5, 7). However, the fundamental mechanisms by which mRNA vaccines induce strong antibody responses are largely unknown.

Durable vaccine-induced antibody responses with a high degree of affinity maturation confer protection against most pathogens (8). Antibody-secreting plasma cells (PCs) in the bone marrow determine the magnitude and longevity of vaccine responses (8). PCs are derived from germinal centers (GCs) in secondary lymphoid tissues during the process of establishing humoral immunity after vaccination (9, 10). Within the dark zone of GCs, B cells that encounter cognate antigens undergo multiple rounds of proliferation and somatic hypermutation. This is followed by antibody affinity maturation in the GC-light zone in a T follicular helper (Tfh) cell-dependent manner (11). B cells, expressing high-affinity B cell receptors as a result of the GC reaction, differentiate into either memory B cells or PCs (12). Recently, a subset of CXCR5+ICOS+, programmed death receptor 1+ (PD-1) CD4+ T cells in the blood was termed circulating Tfh (cTfh) cells due to their similarities with Tfh cells in the lymph nodes (LNs) (13–15). As for regular T helper cells, cTfh cells can be subdivided based on cytokine profile and effector function (16). cTfh1 cells (CXCR3+) are proposed to excel at conferring protection against intracellular pathogens whereas cTfh2 cells (CXCR3− CCR6−) and cTfh17 (CXCR3− CCR6+) may be particularly important in the defense against extracellular pathogens and fungi (14). cTfh cells can be detected seven days after influenza vaccination in humans (17–19). CXCR3+ cTfh cells in humans as well as CXCR3+ Tfh cells in LNs of rhesus macaques have been reported to correlate with the generation of high-avidity antibodies following vaccination (17, 18, 20).

Details of the dynamics of B cell responses and the profile of T cell help induced by mRNA vaccines are currently not known. We, therefore, analyzed the development and maintenance of vaccine-specific responses to a non-replicating mRNA construct encoding the hemagglutinin (HA) of a pandemic influenza H10N8 strain (H10). We show that H10-specific CXCR3+ cTfh cells appeared in the blood 1 week after both prime and boost immunization. This was accompanied by robust GC formation in vaccine-draining LNs, a continuous increase in antibody avidity and seeding of H10-specific PCs to the bone marrow, altogether resulting in protective HA inhibition antibody titers sustained over the 25-week study period.

## MATERIALS AND METHODS

#### Production of Modified mRNA and LNP

Modified mRNA encoding the HA of H10N8 influenza A virus (A/Jiangxi-Donghu/346/2013) were generated as previously described (21). The lipid mixture was combined with a 50 mM citrate buffer (pH 4.0) containing mRNA at 3:1 ratio (aqueous:ethanol) using a microfluidic mixer (Precision Nanosystems). For formulations containing glucopyranosyl lipid adjuvant (GLA) (Avanti Lipids), lipids were combined in a molar ratio of 50:9.83:38.5:1.5:0.17 (ionizable lipid:DSPC:cholesterol:PEG-lipid:GLA). All formulations were dialyzed against PBS, concentrated using Amicon ultra centrifugal filters (EMD Millipore) and passed through a 0.22 µm filter. Particles were 80–100 nm in size with >95% RNA encapsulation.

#### Immunizations and Sample Collection

This animal study was approved by the Local Ethical Committee on Animal Experiments. Chinese rhesus macaques were housed in the Astrid Fagraeus laboratory at Karolinska Institutet according to guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care, and all procedures were performed abiding to the provisions and general guidelines of the Swedish Animal Welfare Agency. Animals were divided into three groups (*n* = 4/group) receiving H10 mRNA/LNP (50 µg) either by IM or ID delivery or H10 mRNA/LNP coformulated with GLA adjuvant (5 µg) delivered IM. Prime and boost immunizations were delivered at week 0 and 4, respectively. Animals receiving H10 mRNA/LNP with GLA received an additional boost at week 15. The animals were sedated with ketamine 10–15 mg/kg given IM (Ketaminol 100 mg/ml, Intervet, Sweden) during the immunizations, blood and bone marrow sampling. Bone marrow was sampled before immunization and at 2, 6, and 25 weeks from the humerus as previously described (22).

An axillary LN was collected before vaccination, opposite from the planned vaccination site, and a collateral axillary LN after boost. The animals were anesthetized by IM injection of 10–15 mg/kg of ketamine and 0.05 mg/kg of medetomidine. Carprofen (4 mg/kg) was given IM as analgesia. LNs were removed in an aseptic manner using minimal entry holes with the aim of removing a singular LN in each procedure. The anesthesia was reversed with atipamezole, 0.25 mg/kg IM after suturing.

#### Blood and Bone Marrow Processing

Peripheral blood was drawn into EDTA tubes and peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll-Paque™ PLUS (GE Healthcare) and washed with PBS. Red blood cells were removed using red blood cell lysis buffer and cells were frozen in 10% DMSO (Sigma-Aldrich) diluted in heat-inactivated FBS. Bone marrow mononuclear cells were isolated and stored in a similar manner as PBMCs.

#### Lymph Node Processing

Lymph nodes were processed as previously described (23). Briefly, LNs were cleaned of fat and cut into small pieces using surgical scissors before being minced and filtered through 70 µm cell strainers. The cells were frozen as described above.

#### Hemagglutination Inhibition (HAI) Assay

Hemagglutination inhibition assay was performed using 0.5% turkey red blood cells (Rockland Antibodies and Assays) diluted in PBS to investigate protective antibody titers. Serum was incubated overnight at +37°C with receptor destroying enzymes (Denka Seiken) to prevent non-specific HAI. Serial dilutions (1:2) of serum samples were performed in V-bottom 96-well plates in duplicates, starting from 1:10 dilution. Recombinant HA of H10N8 influenza A virus (4 units), A/Jiangxi-Donghu/346/2013 (Medigen Inc.) were added to diluted serum and incubated for 30 min at room temperature. The reciprocal of the last serum dilution that resulted non-agglutinated red blood cells represented the HAI titer. Titers <10 were assigned as 1.

#### IgG Avidity ELISA

Clear flat-bottom immune 96-well plates (Thermofisher) were coated for 3 h in 37°C with 100 ng of H10 protein per well. After washing, wells were blocked for 1 h at 37°C in 2% milk powder (blocking buffer) diluted in PBS. Blocking buffer was discarded and plasma samples diluted in blocking buffer were added and incubated for 1 h at 37°C. Plasma was discarded and wells were incubated for 10 min with PBS or sodium thyocyanate (NaSCN) at 1.5–4.5M. After washing, wells were incubated with anti-monkey IgG HRP antibody (Nordic Labs) in washing buffer (0.05% Tween20 in PBS) for 1 h at 37°C. After washing, 100 μl of TMB solution (BioLegend) was added to each well, the reaction was stopped at 5 min with 100 μl of 1M sulfuric acid. Wells were read at 450 nm using an ELISA reader (PerkinElmer).

#### B Cell ELISpot

The frequency of H10-specific antibody-secreting cells (ASCs) and memory B cells was determined as previously described (24), with some modifications. In brief, MAIPSWU10 96-well plates (Millipore) were coated with 10 µg/ml of anti-human IgG (Fcγ; Jackson ImmunoResearch Laboratories). Cells were transferred in duplicate dilution series and cultured overnight at 37°C. For enumeration of ASCs, cells were plated directly without prior stimulation, whereas for memory B cells, cells were prestimulated for 4 days at 2 × 106 cells/ml with 5 µg/ml CpG-B (ODN 2006; Invivogen), 10 µg/ml Pokeweed mitogen (PWM; Sigma-Aldrich), and 1:10,000 Protein A from Staphylococcus aureus Cowan strain (SAC; Sigma-Aldrich). Plates were washed with PBS containing 0.05% Tween-20 (PBS-T), then incubated with 0.25 µg/ml biotinylated goat anti-human IgG (Fcγ; Jackson ImmunoResearch Laboratories) for total IgG determination, 0.1 µg/ml biotinylated H10 for H10-specific determination, or 0.1 µg/ml biotinylated OVA in PBS-T. The plates were washed and then incubated with streptavidin-conjugated alkaline phosphatase (Mabtech) diluted 1:1,000 in PBS-T. Spots were developed with BCIP/NBT substrate (Mabtech) and counted using an AID ELISpot Reader and version 6 of the accompanying software (Autoimmun Diagnostika). Unspecific spots were subtracted from antigen-specific wells, as defined by spots counted in samples incubated in OVAprobed wells.

Recombinant HA of H10N8 influenza A virus (H10), A/ Jiangxi-Donghu/346/2013 (Medigen Inc.), and ovalbumin (OVA; Invivogen) molecules were biotinylated using the EZ-Link Sulfo-NHS-Biotinylation kit (Thermo Fisher) using a 1:1 M ratio and unreacted biotin was removed using the included Zeba Spin Desalting Columns.

#### Phenotypic Analysis and H10 Recall Assay of Tfh Cells

1.5 × 106 cells from indicated time points were stained with LIVE/ DEAD Fixable Blue Dead Cell kit according to manufacturer's protocol (Invitrogen). Samples were surfaced stained with a panel of fluorescently labeled antibodies (Table S1 in Supplementary Material) to identify specific cell subsets. Additionally, 1.5 × 106 PBMCs were rested for 3 h and then stimulated overnight in complete media (10% FCS, 1% penicillin/streptomycin/glutamine in RPMI, all from Gibco, Stockholm, Sweden) in *U*-bottom 96-well plates with H10 peptides (15mers overlapping by 11 amino acids, 2 µg/ml) and Brefeldin A at 10 µg/ml. Cells were stained with surface-specific antibodies (Table S1 in Supplementary Material), fixed, and permeabilized using fixation and permeabilization solution (BD Biosciences) before being stained for intracellular cytokines (Table S1 in Supplementary Material). Samples were resuspended in 1% paraformaldehyde before acquisition using a Fortessa flow cytometer (BD Biosciences). Results were analyzed using FlowJo version 9.7.6. Background cytokine staining was subtracted, as defined by staining in samples incubated without peptide.

#### Detection of GCs *In Situ*

Extirpated LNs were snap frozen using dry ice in OCT media (Tissue-Tek) and kept in −80°C until use. Tissues were thawed to −20°C and then sectioned (8 µm) and fixed for 15 min in 2% formaldehyde in PBS. Tissues were permeabilized using tris-buffered saline with 0.1% saponin and 1% hepes buffer (permwash with pH 7.4), all future reagents were diluted in permwash. LNs were blocked with 1% FCS and then stained with anti CD3 (Dako), Ki67 (BD), and PD-1 (R&D systems). After this, biotinylated anti-mouse or anti-goat (Dako) or antirabbit (Vector Labs) secondary antibodies were added, which were detected by the addition of streptavidin-conjugated Alexa Fluor 405/555/647 (Invitrogen). Image tiles of entire LNs were acquired using a Nikon Eclipse Ti-E confocal microscope. GCs were defined as dense follicular structures including CD3+PD-1+ cells (light zone) and Ki67+ cells (dark zone) (**Figure 3A**). Image analysis was done using CellProfiler software (Broad Institute Inc.) with in-house algorithms. Briefly, GCs were manually identified in the program to enable automatic enumeration of PD-1+ and Ki67+ cells within the individual GCs and the area of the GCs. PD-1+ cells were almost exclusively CD3+ and Ki67+ cells were mostly CD3− (GC B cells).

## CXCL13 ELISA

Undiluted plasma stored in −20°C was thawed and analyzed using a Quantikine Human CXCL13/BLC/BCA-1 Elisa (R&D systems) according to manufacturer's instructions.

## Detection of Antigen-Specific GC B Cells by Flow Cytometry

2 × 106 cells from LN cell suspensions were stained with LIVE/ DEAD Fixable Blue Dead Cell kit according to manufacturer's protocol (Invitrogen). Samples were then incubated with a titrated amount of H10-tetramer probe for 20 min at 4°C. H10-tetramer probes were prepared beforehand by mixing biotinylated H10 protein with streptavidin-BV421 (Biolegend) at a 4:1 M ratio. Samples were washed and stained with a surface antibody cocktail (Table S2 in Supplementary Material). Before intracellular staining (Table S2 in Supplementary Material), samples were fixed and permeabilized using the Transcription Factor Buffer Set (BD Biosciences). Samples were resuspended in 1% paraformaldehyde before acquisition using a Fortessa flow cytometer (BD Biosciences). Results were analyzed using FlowJo version 9.7.6.

#### Statistical Analysis

Statistical analysis was conducted by Prism Version 6.0 (GraphPad) software. All of the data are presented as the mean ± SEM. Difference between groups was analyzed as described in figure legends. A *p*-Value <0.05 was considered to be statistically significant.

#### RESULTS

#### mRNA Vaccine Encoding H10 Induces Protective Levels of Antibodies

Rhesus macaques were immunized either intramuscularly (IM) or intradermally (ID) with an mRNA vaccine encoding the full-length HA of H10N8 (A/Jiangxi-Donghu/346/2013) (H10) formulated in LNP (**Figure 1A**). A third group received this formulation combined with the TLR4-agonist GLA to evaluate whether an adjuvant could further enhance immune responses. All animals received a homologous prime-boost immunization at 0 and 4 weeks (**Figure 1B**). In addition, the GLA group received a third immunization at 15 weeks.

All animals induced neutralizing antibody titers against HA above the accepted level of protection for seasonal influenza vaccination, as measured by hemagglutination inhibition assay (HAI) (**Figure 1C**) (25). Although some of the animals in the ID group already showed titers at the protective level after the prime immunization, all groups had titers that exceeded this level following boost. The antibody levels persisted above this level for the remainder of the study. The titers were significantly higher in the ID group compared to the IM groups for up to 2 weeks following boost, but were similar thereafter. The GLA group did not show higher HAI titers compared to the other groups, thus indicating that the mRNA/LNP formulation itself was sufficiently immunogenic. The third immunization in the GLA

Figure 1 | mRNA vaccine encoding influenza H10 elicits high antibody titers by intradermal and intramuscular immunization. Rhesus macaques were vaccinated with mRNA encoding for hemagglutinin of H10N8 influenza encapsulated in LNPs. (A) Shows the study groups and (B) shows the immunization and sampling schedule. (C) Antibody levels over time induced in the different groups as assayed by hemagglutination inhibition assay. Arrows indicate immunizations at week 0 and 4, as well as 15 for the glucopyranosyl lipid adjuvant group only. (D) The left *y*-axis shows antigen-specific plasma IgG titers (circles) and the right *y*-axis shows avidity index (triangles). Arrows indicate immunizations at week 0, 4. Titers are displayed as mean ± SEM. Dotted line depicts the accepted level of protection for seasonal influenza. Statistics was calculated by two-way ANOVA with Tukey's multiple comparison test: \**p* < 0.05, \*\**p* < 0.01, \*\*\**p* < 0.001.

group resulted in a transient increase in HAI titers, which returned to similar levels as the other groups 5 weeks later.

Antibody titer, as measured by HAI, is a function of both the quantity and quality of an antibody. However, by combining these two measurements, certain mechanistic insights into antibody development may be lost over time. Therefore, total IgG titers against H10 as well as antibody avidity were measured separately. H10-specific IgG antibody titers were induced in both IM and ID groups after prime immunization. Titers continued to increase at the time of the second immunization and peaked 2 weeks thereafter (**Figure 1D**). The avidity index of H10-specific IgG antibodies as measured by a chaotropic ELISA wash assay, did not increase between 4 and 6 weeks following the second immunization (**Figure 1D**). However, at week 11, there was a clear increase in avidity, which continued to rise until the study end. There were no significant differences in IgG titers or avidity between the ID and IM group, except in the ID group two weeks after boost, which had higher IgG titers (*p*< 0.0001). Collectively, this demonstrates that two immunizations with the H10 mRNA/ LNP formulation were sufficient to induce protective and durable antibody titers.

#### Rapid and Sustained B Cell Responses after mRNA Vaccination

To characterize the kinetics of the B cell responses at the cellular level, the frequency of H10-specific memory B cells was determined by ELISpot (**Figures 2A–C**). Circulating H10-specific memory B cells were readily detectable 2 weeks after the prime immunization (**Figures 2A–C**). Thereafter, the number of H10-specific memory B cells contracted slightly, but expanded following the boost immunization. This was followed by a gradual decline. The GLA group showed an additional increase 2 weeks after the second boost as expected (**Figure 2C**). By study end (25 weeks), the IM and ID groups showed similar levels of circulating H10-specific memory B cells (**Figure 2D**), whereas the GLA group had higher levels due to the second boost.

Similar to the rapid induction of circulating H10-specific memory B cells, we detected H10-specific PCs in the bone marrow 2 weeks after the prime immunization in all groups tested (**Figures 2A–C**). However, in contrast to the memory B cell pool, the number of PCs remained relatively stable throughout the study. This dichotomy in fluctuation between memory B cells and PCs is in line with previous reports on protein vaccine immunizations (24).

The number of H10-specific PCs declined significantly in the IM group compared to the ID group by the end of the study (**Figure 2E**). Next, we analyzed plasmablasts in some of the animals at one week after immunization. The numbers of H10 specific plasmablasts were close to the limit of detection after the prime immunization, but increased to readily detectable levels after the boost immunization (**Figure 2F**).

We had previously shown that the priming of vaccine-specific T cells occurs in the vaccine-draining LNs (23). To investigate whether the priming of B cells also takes place at this site, vaccine-draining LNs were collected 9 days after prime immunization and compared with control LNs that did not drain the vaccine delivery site. H10-specific memory B cells and PCs were detected only in the vaccine-draining LNs (**Figures 2G,H**). This was observed with both ID and IM administration. This demonstrates that the H10 mRNA/LNP vaccine induces robust levels of vaccine-specific memory B cells and PCs following prime immunization and that these levels are sustained.

#### GC Formation in Vaccine-Draining LNs

Durable levels of H10-specific memory B cells and PCs coupled with a steady increase of IgG antibody avidity induced by the H10 mRNA/LNP vaccine are likely the outcome of a strong GC reaction. Since priming of H10-specific B cells appeared to be restricted to the vaccine-draining LNs, we also analyzed the expansion of GCs at this site. Using immunofluorescence and confocal imaging of cryosections, we quantified the area of GCs, the numbers of PD-1+ Tfh cells, and proliferating Ki67+ cells within individual GCs (26) (**Figure 3A**). Since the HAI titers did not differ between the three study groups, we did not stratify the data by group for this analysis. We evaluated LNs where biopsies from the same animal could be obtained both before immunization and 2 weeks post-boost. All vaccine groups were represented among the seven animals where such set of LNs was available (*n* = 1 from group 1, *n* = 4 from group 2, and *n* = 2 from group 3). We normalized the total GC area and total numbers of GC Tfh cells or GC Ki67+ cells by LN area. In five out of the seven animals, there was an increase in the GC area/LN area ratio post-immunization (**Figure 3B**). There was also an increase in GC Ki67+ cells/LN area ratio and GC Tfh cells/LN area ratio (**Figures 3C,D**). One advantage of analyzing tissue sections over cell suspensions is the option to study individual GCs in intact LN architecture rather than all GC cells combined. The area of individual GCs was significantly increased post-vaccination (**Figure 3E**). This was also observed for Tfh cells and GC Ki67+ cells within individual GCs (**Figures 3F,G**). As expected, the increase in the number of GC Ki67+ cells was greater than the increase in the number of GC Tfh cells, since GC B cells have an enhanced capacity for proliferation (9). To this end, it has been shown that the B cell/Tfh cell ratio is higher in the top neutralizers following HIV Env vaccination (27). We observed a significant increase in the Ki67+ cell/Tfh cell ratio post-vaccination in individual GCs (**Figure 3H**). There was a strong positive correlation between the number of Ki67+ cells and the area within each GC (*p* < 0.001, *R*<sup>2</sup> = 0.4613) (**Figure 3I**). This was also found for the number of GC Tfh cells and the GC area of individual GCs (*p* < 0.001, *R*<sup>2</sup> = 0.5869) (**Figure 3J**). Additionally, Tfh cell numbers correlated with Ki67+ cell numbers within each GC (*p* < 0.001, *R*<sup>2</sup> = 0.5617) (**Figure 3K**).

Increased levels of the chemokine CXCL13 in plasma was recently proposed as a biomarker for GC activity in the LNs (28). There was a modest elevation in CXCL13 levels detected transiently 1 week after prime vaccination, with no noticeable increase observed at 1 or 2 weeks following boost vaccination (**Figure 3L**). GC formation after prime vaccination was also investigated using flow cytometry by enumerating the frequency of BCL6+ Ki67+ GC B cells expressing H10-specific B cell receptors in the vaccine-draining LNs (**Figure 3M**). No H10-specific

memory B cells (D) and bone marrow PCs (E) between the groups. (F) Antigen-specific plasmablasts per 1 × 106 animals were immunized and LNs were collected after 9 days. Antigen-specific PCs (G) and memory B cells (H) in vaccine-draining LNs vs non-draining control LNs. Red colored dots corresponding to intradermal immunization, blue to IM immunization, and orange to IM immunization with glucopyranosyl lipid adjuvant adjuvant. Data show mean ± SEM. Statistics was based on Mann–Whitney *U*-test \**p* < 0.05.

GC B cells were found in control non-draining LNs, but both unswitched IgM+ and class-switched IgM− BCL6+ Ki67+ GC H10-specific B cells were detected in vaccine-draining LNs 9 days after prime immunization (**Figure 3M**). Taken together, this suggests that the mRNA/LNP vaccine is a potent inducer of GC activity.

## Circulating H10-Specific ICOS**+** PD-1**+** CXCR3**+** Tfh Cells Are Induced after Vaccination and Correlate with Antibody Avidity

As mentioned above, cTfh cells can be identified in blood as CXCR5+ICOS+PD-1+ CD4+ T cells. The Th1-polarized CXCR3+ cTfh cell subset specifically has been shown to correlate with high-avidity antibodies 7 days after influenza vaccination in humans (17, 18). Therefore, we investigated cTfh cell subset frequencies pre-vaccination and 7 days following prime and boost. We first concluded that there was no general increase in CXCR3+ or CXCR3− total CD4+ T cells (**Figure 4A**). Next, we analyzed CXCR5+ ICOS+ PD-1+ CXCR3+/− cTfh cells within the central memory (CD28+CD95+) CD4+ T cell population (**Figure 4B**). While there was no increase in CXCR3− cTfh cells (**Figure 4C**), there was a significant increase in the number of CXCR3+ cTfh cells both after prime and boost (**Figure 4D**). No notable differences in overall cTfh cells were found between the ID and IM groups.

We investigated whether the CXCR3+ cTfh cells were expanded in the vaccine-draining LNs before entering the circulation by analyzing the set of LNs collected at nine days post-prime. The frequency of Tfh cells was increased in the vaccine-draining LNs compared to control non-draining LNs (**Figure 4E**). In addition, the majority of the Tfh cells in the vaccine-draining LNs expressed CXCR3 (**Figure 4E**). CXCR3+ Tfh cells in vaccine-draining LNs

Figure 3 | Increased germinal center (GC) activation post influenza mRNA vaccine immunization. Axillary lymph nodes (LNs) were obtained before prime and 2 weeks after boost. A variety of readouts for GC activity were evaluated following the boost immunization. (A) LN sections stained for GCs before and after vaccination. Sections were stained with anti-CD3 (blue), PD-1 (green), and Ki67 (red) and follicular structures with CD3+PD-1+ T follicular helper (Tfh) and Ki67+ cells were considered GCs. Images were analyzed with CellProfiler software. (B–D) Each LN was analyzed for average GC area and total number of Ki67+ cells and Tfh cells, divided by the LN area for normalization. (E–H) Individual GCs represented by dots. (E) GC area. (F) Number of Tfh cells. (G) Number of Ki67+ cells. (H) Ki67+ to Tfh cell ratio. (I–K) GC architecture homogeneity was investigated by plotting GC area, Tfh, and Ki67+ cell numbers against each other. (L) CXCL13 in plasma at indicated time points. (M) Antigen-specific GC (CD3− CD20+ BCL6+ Ki67+) B cells in vaccine-draining LNs 9 days after prime vs non-draining control LNs by flow cytometry using an H10 tetramer probe. Data show mean ± SEM. Nonparametric Wilcoxon matched-pairs signed rank test was used for (B–D) and Mann–Whitney *U*-test for (E–H). \**p* < 0.05, \*\**p* < 0.01, \*\*\**p* < 0.001.

also expressed Ki67 at higher levels than CXCR3− Tfh cells in the same and CXCR3+/− Tfh cells from control LNs (**Figure 4F**). Although difficult to prove, these findings indicate that the cTfh1 polarized cells found in peripheral blood after mRNA vaccine administration may have been generated in the vaccine-draining LNs.

The number of CXCR3+ cTfh cells following boost correlated with antibody avidity at week 6 and 25, with a significant correlation at week 11 (*p* = 0.0432, *R*<sup>2</sup> = 0.5916) (**Figure 5A**). To investigate cTfh cell specificity, PBMCs were stimulated with H10 overlapping peptides and tested for cytokine secretion by a standard intracellular cytokine assay (**Figure 5B**). There were few H10-specific cells as evidenced by IFNγ production within the total CD4+ central memory T cell population 1 week post-prime, but there was a clear increase one week following boost (**Figure 5C**). The peptide stimulation and culture procedure resulted in some loss in the resolution of CXCR5+ICOS+PD-1+CXCR3+ cTfh cellular staining. However, a significant increase in the number of IFNγ+CXCR3+ cTfh cells 1 week after prime and boost was still observed (**Figure 5D**). Collectively, this demonstrates that the H10/LNP vaccine induces H10-specific cTfh cells of the CXCR3+ Th1-polarized profile that correlate with the avidity of H10-specific IgG antibodies.

#### DISCUSSION

mRNA vaccines can be produced rapidly with high antigen precision, which offers an advantage over the more time-consuming process of producing live-attenuated or subunit protein vaccines (3, 29). Due to antigenic shifts and drifts, the influenza vaccine is renewed yearly to protect against the circulating strains. The mRNA platform would, therefore, be suitable for seasonal

influenza vaccines as well as pandemic influenza strains (4, 5, 29). Although mRNA vaccines have already advanced to clinical trials, little is known of the adaptive immune profile induced by such vaccines. Therefore, this study focused on key immunological events resulting in antibody responses to an mRNA/ LNP vaccine, using rhesus macaques whose immune system closely resembles that of humans. Influenza HA was chosen as the antigen due to the clinical relevance for the mRNA platform and that the levels of antibodies required for protection against seasonal influenza are well-characterized and could be benchmarked against (25). However, important to note is that all the animals in our study were naïve to influenza before immunization, which is in contrast to a clinical setting where most humans will have some degree of pre-existing immunity due to vaccination or prior infection. Therefore, the presence of cross-reactive HA-specific memory B cells and T cells may result in a stronger priming response than recorded in the rhesus macaques, although pre-existing immunity to HA of the pandemic influenza strain H10N8 used in this study is relatively rare in humans.

Seasonal influenza vaccines have been shown to induce protection in 60–90% of recipients (30, 31). In our study, we found that two animals in the ID group reached the accepted level of protection as measured by HAI titers after prime immunization. Animals across the groups reached this level after boost immunization and remained above this level for the entire 25-week study. It is plausible that all animals would have reached protective levels after the prime immunization if more time had been allowed since both the HAI and IgG titers were gradually increasing prior to the boost.

The memory B cell pool expanded quickly after prime immunization, suggesting that the mRNA platform is efficient at priming naïve B cells. H10-specific IgG titers were readily detectable 2 weeks after prime, but neutralization capacity, as measured by HAI titers, was found in a single animal at this time point. This could be explained by the initial low affinity of the memory B cell population, which would lead to insufficient neutralization.

The seeding of PCs into the bone marrow was also detected at 2 weeks post-prime immunization in all groups. The frequency of vaccine-specific PCs that were generated with the mRNA vaccine is comparable to that reported with recombinant HA protein coadministered with a strong adjuvant in the same animal model (24). Also, additional boosts did not increase the PC compartment in the bone marrow with either of the vaccine platforms (24). Regulation of the PC pool in the bone marrow is not well known. We found that the PC levels remained relatively constant once established after the prime, despite HAI titers and H10-specific IgG titers increasing significantly post-boost. H10-specific IgG titers waned more rapidly than HAI titers. This may be explained by the observed increase in antibody avidity, which suggests increased survival of high affinity PCs and/or ongoing affinity maturation in GCs long after immunization. Supportive of the latter hypothesis is that GC B cells have been found to persist for several weeks following immunization (27). High affinity B cell clones are selected to become long-lived PCs that migrate to the bone marrow (32). Vaccine platforms that are powerful at generating antibody responses have shown superior GC formation capacities (23, 33, 34). With the clear GC activation found at 2 weeks following boost immunization, additional affinity maturation and subsequent differentiation of long-lived PCs likely continued for some time thereafter. Since affinity maturation occurs before the differentiation of PCs, it is possible that the PCs seeded early into the bone marrow were replaced by higher affinity PCs derived from the GC reaction induced by the boost immunization. This would explain the increase in antibody avidity found here.

To improve the efficacy of vaccines, there has been much focus on identifying immunological events, including biomarkers, which lead to the production of high titer protective antibodies. Tfh cells and their circulating counterparts have emerged as key players for dictating the antibody responses (35). As mentioned previously, studies have shown that CXCR3+ cTfh cells correlate with high-avidity antibodies against influenza after vaccination in humans (17, 18). It has also been shown that cTfh cell levels can predict the seroconversion in humans to influenza vaccination (36). In rhesus macaques, CXCR3+ Tfh levels in LNs were shown to correlate with avidity, longevity, and neutralization capacity of HIV vaccine-induced antibodies (20). In line with these observations, we found that the mRNA/LNP vaccine platform specifically induced an increase in CXCR3+ cTfh cells, which correlated with increased IgG avidity. This could indicate that CXCR3+ cTfh cells are important for selecting and expanding B cells of high affinity. Whether inducing cTfh cells toward the CXCR3+ phenotype is caused by the vaccine platform or the HA antigen remains to be elucidated. To this end, a type I IFNmediated Th1-polarization after mRNA vaccine administration has been described and shown to at least partly be caused by TLR7/8 ligation (37, 38).

*In vitro* studies have proposed that CXCR3+ cTfh cells preferentially provide help to memory B cells compared to naïve B cells (14). However, CXCR3+ and CXCR3− Tfh cells sorted from rhesus LNs showed that there was no difference in their B cell help capacity (20). Since the vaccinated animals in our study showed a rapid induction of memory B cells, plasmablasts, and PCs, there was clearly an efficient priming of naïve B cells despite being naïve to influenza.

Furthermore, it has been proposed that the main function of CXCR3+ cTfh cells is to select memory B cells of high affinity, leading to rapid expansion of this population upon new antigen encounter (17). With regards to influenza, where the circulating strain changes every year, the ability to select for high-affinity B cell clones against the circulating strain is critical. Recent studies have shown that the cTfh cells that increase in blood after influenza vaccination represent memory Tfh cells that have been reactivated upon antigen re-exposure (19). cTfh cells can home to the LNs and differentiate into GC Tfh cells to facilitate the GC response (39, 40). Induction of vaccine-specific cTfh cells is, therefore, a central mechanism in vaccine-mediated protection, since these cells facilitate quick re-stimulation of memory B cells in the GC reaction. We found H10-specific cells within the CXCR3+ cTfh cell population. As a considerable proportion of ICOS and PD-1 expression was lost during the overnight stimulation, the number of H10-specific CXCR3+ cTfh cells may be underestimated.

Since CXCR3+ Tfh1 cells have been shown to be inferior to CXCR3− Tfh2/17 cells at providing help to naïve B cells, it was suggested that an influenza vaccine inducing Tfh2/17 cells rather than Tfh1 cells would be preferable (14, 41). However, passive transfer of antibody clones against HA in mice showed that only Th1-polarized IgG2a, and not Th2-polarized IgG1, conferred protection against lethal challenge, despite that the antibodies had the same ability to bind the antigen (35, 42, 43). This was proposed to be due to the different Fc regions of the antibodies and indicates that antibodies generated by help from cTfh cells of the Th1 subtype may be critical for the induction of protection against influenza.

In summary, we show here that an influenza mRNA/LNP vaccine induces robust GC and B cell responses, including PCs seeding into the bone marrow. Antibody avidity increases over time and is accompanied by cTfh cells of the CXCR3+ subtype. Collectively, this gives insights into the adaptive immune profile generated by mRNA/LNP vaccines and may indicate that this platform is particularly powerful for infections such as influenza that require a Th1-profile.

## ETHICS STATEMENT

Chinese rhesus macaques were housed in the Astrid Fagraeus laboratory at Karolinska Institutet according to guidelines of the Association for Assessment and Accreditation of Laboratory Animal Care, and all procedures were performed abiding to the provisions and general guidelines of the Swedish Animal Welfare Agency. This animal study was approved by the Local Ethical Committee on Animal Experiments.

## AUTHOR CONTRIBUTIONS

GL, FL, KB, SJ, KH, LB, HS, GC, and KL designed research. GL, SO, FL, ET, AL, FH, KB, SJ, KH, LB, HS, GC, and KL performed experiments and contributed with vaccines. GL, SO, FL, ET, FH, GC, and KL analyzed data. GL, SO, FL, ET, KB, and KL wrote the paper. All authors reviewed the paper.

## ACKNOWLEDGMENTS

We wish to thank Drs. Mats Spångberg and Bengt Eriksson and their personnel at the Astrid Fagraeus laboratory at Karolinska Institutet for expert assistance and care of the animals. We also thank Dr. Gunilla Karlsson Hedestam for advice and Tyler Sandberg and Joel Holmberg should be thanked for technical assistance. This study was partly supported by a grant to KL supported by Vetenskapsrådet. GL, SO, ET, and AL are recipients of faculty salary grants from Karolinska Institutet. AL is also supported by a grant from the China Scholarship Council.

#### REFERENCES


#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at http://www.frontiersin.org/article/10.3389/fimmu.2017.01539/ full#supplementary-material.

Table S1 | List of antibodies used for flow cytometry.

Table S2 | List of antibodies used for flow cytometry.

memory response in circulating T follicular helper cells. *Sci Immunol* (2017) 2: eaag2152. doi:10.1126/sciimmunol.aag2152


antibody responses. *Immunity* (2013) 39:758–69. doi:10.1016/j.immuni. 2013.08.031


**Conflict of Interest Statement:** The authors declare that no conflict of interest exists. The authors KB, HS, KH, LB, HS, and GC are employees of Moderna Therapeutics.

*Copyright © 2017 Lindgren, Ols, Liang, Thompson, Lin, Hellgren, Bahl, John, Yuzhakov, Hassett, Brito, Salter, Ciaramella and Loré. 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) or licensor 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.*

# Corrigendum: Induction of Robust B Cell Responses After Influenza mRNA Vaccination Is Accompanied by Circulating Hemagglutinin-Specific ICOS+ PD-1+ CXCR3+ T Follicular Helper Cells

Gustaf Lindgren1,2, Sebastian Ols 1,2, Frank Liang1,2, Elizabeth A. Thompson1,2, Ang Lin1,2 , Fredrika Hellgren1,2, Kapil Bahl <sup>3</sup> , Shinu John<sup>3</sup> , Olga Yuzhakov <sup>3</sup> , Kimberly J. Hassett <sup>3</sup> , Luis A. Brito<sup>4</sup> , Hugh Salter 4,5, Giuseppe Ciaramella<sup>3</sup> and Karin Loré1,2 \*

#### Approved by:

Frontiers in Immunology Editorial Office, Frontiers Media SA, Switzerland

#### \*Correspondence:

Karin Loré karin.lore@ki.se

#### Specialty section:

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

> Received: 21 February 2019 Accepted: 07 March 2019 Published: 02 April 2019

#### Citation:

Lindgren G, Ols S, Liang F, Thompson EA, Lin A, Hellgren F, Bahl K, John S, Yuzhakov O, Hassett KJ, Brito LA, Salter H, Ciaramella G and Loré K (2019) Corrigendum: Induction of Robust B Cell Responses After Influenza mRNA Vaccination Is Accompanied by Circulating Hemagglutinin-Specific ICOS+ PD-1+ CXCR3+ T Follicular Helper Cells. Front. Immunol. 10:614. doi: 10.3389/fimmu.2019.00614 <sup>1</sup> Department of Medicine Solna, Immunology and Allergy Unit, Karolinska Institutet, Stockholm, Sweden, <sup>2</sup> Center for Molecular Medicine, Karolinska Institutet, Stockholm, Sweden, <sup>3</sup> Valera LLC, Cambridge, MA, United States, <sup>4</sup> Moderna Therapeutics, Cambridge, MA, United States, <sup>5</sup> Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden

Keywords: mRNA vaccine, adaptive immune responses, non-human primates, influenza, T follicular helper cells, germinal centers

#### **A Corrigendum on**

**Induction of Robust B Cell Responses after Influenza mRNA Vaccination Is Accompanied by Circulating Hemagglutinin-Specific ICOS**+ **PD-1**+ **CXCR3**+ **T Follicular Helper Cells**

by Lindgren, G., Ols, S., Liang, F., Thompson, E. A., Lin, A., Hellgren, F., et al. (2017). Front. Immunol. 8:1539. doi: 10.3389/fimmu.2017.01539

In the original article, Liang et al. (44) was not cited in the article. The citation has now been inserted in the **Results, mRNA vaccine encoding H10 induces protective levels of antibodies**, paragraph two and should read:

"All animals induced neutralizing antibody titers against HA above the accepted level of protection for seasonal influenza vaccination, as measured by hemagglutination inhibition assay (HAI) (Figure 1C) as we have reported earlier (25, 44). Although some of the animals in the ID group already showed titers at the protective level after the prime immunization, all groups had titers that exceeded this level following boost. The antibody levels persisted above this level for the remainder of the study. The titers were significantly higher in the ID group compared to the IM groups for up to 2 weeks following boost, but were similar thereafter. The GLA group did not show higher HAI titers compared to the other groups, thus indicating that the mRNA/LNP formulation itself was sufficiently immunogenic. The third immunization in the GLA group resulted in a transient increase in HAI titers, which returned to similar levels as the other groups 5 weeks later."

### REFERENCES


The authors apologize for this error and state that this does not change the scientific conclusions of the article in any way.

Copyright © 2019 Lindgren, Ols, Liang, Thompson, Lin, Hellgren, Bahl, John, Yuzhakov, Hassett, Brito, Salter, Ciaramella and Loré. 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.

# Single Dose of Consensus Hemagglutinin-Based Virus-Like Particles Vaccine Protects Chickens against Divergent H5 Subtype Influenza Viruses

*Peipei Wu1,2,3†, Jihu Lu1,2,3†, Xuehua Zhang1,2,3†, Mei Mei 1,2,3, Lei Feng1,2,3, Daxin Peng3,4, Jibo Hou 1,2,3, Sang-Moo Kang <sup>5</sup> , Xiufan Liu3,4\* and Yinghua Tang1,2,3\**

*<sup>1</sup> Institute of Veterinary Immunology & Engineering, Jiangsu Academy of Agricultural Sciences, Nanjing, China, 2National* 

#### *Edited by:*

*África González-Fernández, Centro de Investigaciones Biomédicas (CINBIO), Spain*

#### *Reviewed by:*

*Michael Schotsaert, Icahn School of Medicine at Mount Sinai, United States Mathew Paul, St George's, University of London, United Kingdom*

#### *\*Correspondence:*

*Xiufan Liu xfliu@yzu.edu.cn; Yinghua Tang tyinghua@yahoo.com*

*† These authors have contributed equally to this work.*

#### *Specialty section:*

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

*Received: 03 June 2017 Accepted: 10 November 2017 Published: 27 November 2017*

#### *Citation:*

*Wu P, Lu J, Zhang X, Mei M, Feng L, Peng D, Hou J, Kang S-M, Liu X and Tang Y (2017) Single Dose of Consensus Hemagglutinin-Based Virus-Like Particles Vaccine Protects Chickens against Divergent H5 Subtype Influenza Viruses. Front. Immunol. 8:1649. doi: 10.3389/fimmu.2017.01649*

*Research Center of Engineering and Technology for Veterinary Biologicals, Jiangsu Academy of Agricultural Sciences, Nanjing, China, 3 Jiangsu Co-Innovation Center for the Prevention and Control of Important Animal Infectious Disease and Zoonosis, Yangzhou, China, 4Animal Infectious Disease Laboratory, College of Veterinary Medicine, Yangzhou University, Yangzhou, China, 5Center for Inflammation, Immunity & Infection, Institute for Biomedical Sciences, Georgia State University, Atlanta, GA, United States*

The H5 subtype highly pathogenic avian influenza (HPAI) virus is one of the greatest threats to global poultry industry. To develop broadly protective H5 subunit vaccine, a recombinant consensus HA sequence (rHA) was constructed and expressed in viruslike particles (rHA VLPs) in the baculovirus-insect cell system. The efficacy of the rHA VLPs vaccine with or without immunopotentiator (CVCVA5) was assessed in chickens. Compared to the commercial Re6 or Re6-CVCVA5 vaccines, single dose immunization of chickens with rHA VLPs or rHA-CVCVA5 vaccines induced higher levels of serum hemagglutinin inhibition titers and neutralization titers, mucosal antibodies, IFN-γ and IL-4 cytokines in sera, and cytotoxic T lymphocyte responses. The rHA VLPs vaccine was superior to the commercial Re6 vaccine in conferring cross-protection against different clades of H5 subtype viruses. This study reports that H5 subtype consensus HA VLP single dose vaccination provides broad protection against HPAI virus in chickens.

#### Keywords: H5 subtype influenza virus, vaccine, consensus, virus-like particles, broad protection

## INTRODUCTION

The H5 subtype highly pathogenic avian influenza (HPAI) viruses not only affect millions of domestic poultry, including chickens, ducks, and geese, as well as thousands of migratory wild birds, but also risk to the public health. Since the H5 subtype HPAI viruses emerged in Southeast Asia in 1990s and have evolved into 10 phylogenetic clades (0–9) and more than 30 subclades based on their hemagglutinin (HA) genes (1, 2). Recently, H5 HPAI viruses of subclade 2.3.4 revealed a novel propensity to reassort with neuraminidase (NA) subtypes other than N1, including H5N2, H5N5, H5N6, and H5N8 (3–5).

A strategy of culling in combination with vaccination is a major current measure to prevent and control the spread of H5 HPAI viruses in several countries and regions (6–8). In China, a series of recombinant commercial vaccines, Re1, Re3–Re8, Re10, D7, and D8, have been developed and updated since 2004 to ensure an antigenic match between the vaccines and the circulating strains (8). However, H5 subtype HPAI viruses are still the pathogen with leading potential threat for poultry industry in China. It has been an everlasting difficult situation that the existing vaccine does not provide optimal protection against the emergence of new variants. The rapid mutation of the H5 viruses continues to bypass the time-consuming procedures of screening vaccine candidates during development of traditional inactivated whole-virus vaccine. Besides, propagation of influenza virus in the embryonated chicken eggs during preparation of inactivated virus vaccine produces biohazardous waste materials and leads potential threat to animal and human public health.

In this regard, there is an urgent need for developing an effective, broadly cross-protective, and safe H5 vaccine for use in poultry farms. Consensus sequences contain the most common residue at each position after aligning with a population of sequences. Previous studies reported that the consensus HA-based influenza vaccine candidates have elicited cross-reactive immune responses (9–12). There are commercial trivalent HA protein vaccines consisting of three insect-cell-derived full-length HAs similar to the seasonal trivalent inactivated vaccine, which represents an alternative to chicken-egg derived vaccines. In this study, we developed an insect-cell-derived consensus H5 subtype HA gene based virus-like particles (rHA VLPs) vaccine. Single dose immunization of chickens with rHA VLPs vaccines elicited serum and mucosal antibodies and cell-mediated immune responses at high levels, and induced cross-protection against subclade 2.3.4.6 and clade 7 of H5 HPAI viruses.

#### MATERIALS AND METHODS

#### Cell Lines, Viruses

*Spodoptera frugiperda* Sf9 insect cells (CRL-1711; ATCC, Manassas, VA, USA) were maintained in serum-free SF900II medium (Invitrogen, Carlsbad, CA, USA) at 27°C and used for production of recombinant baculoviruses (rBVs) and VLPs. Two HPAI strains (ZJ and DT) used as challenge virus were originated from clade 2.3.4.6 (A/Chicken/Zhejiang/2011, ZJ, H5N2) and clade 7 (A/Chicken/Huadong/4/2008, DT, H5N1). In this study, the recombinant Re6 commercial vaccines, and chicken sera against monovalent Re5, Re6, Re7, and Re8 antigens were purchased from Weike Biotechnology Co., Ltd., Harbin, China. Four commercial recombinant vaccines of Re5, Re6, Re7, and Re8 were inactivated whole-virus-based mineral oil emulsion vaccines (Weike). The six inner genes of four recombinant vaccine strains are derived from A/Puerto Rico/8/1934 virus (PR8, H1N1). The HA and NA genes of Re5–Re8 are derived from different subclades of H5 wild-type viruses, Re5 from strain of A/duck/Anhui/1/2006 (clade 2.3.4) (8), Re6 from virus of A/duck/Guangdong/S1322/2006 (clade 2.3.2) (8), Re7 from strain of A/Chicken/Liaoning/S4092/2011 (clade 7.2) (13), and Re8 from A/chicken/Guizhou/4/13 (clade 2.3.4.4) (14).

#### Preparation of Consensus HA Gene-Based Recombinant Proteins

It is known that nucleotide changes are more often because of non-sense mutations. Thus, the frequency of mutations at the nucleotide level is higher than that of the amino acid changes. Other researchers have also reported the consensus sequence derived from the nucleotide level (15). The complete open reading frame (ORF) of consensus HA was derived from the consensus nucleotide sequences of the H5 subtype HA after mega-aligning available complete H5 subtype HA sequences which isolated from 2004 to 2012 in China and originated from avian species in GenBank with Clustal V program (Lasergene, v.11; DNAStar, Madison, WI, USA). A total number of 304 HA sequences were analyzed in this study and the highest frequency of nucleotides in each position was selected to generate the consensus HA. The additional leucine zipper GCN-pII sequence was fused to the C-terminal end of the selected HA gene to form trimers as described (16). This engineered recombinant HA was named as rHA. In the results of the complete HA ORF consensus sequence after mega-aligning, the most adaption region is the HA1 head domain (61–161 aa), compared to the relatively conserved HA2 domain during evolution. The codon of the rHA ORF was optimized according to the codon usage frequency of *S. frugiperda* cell by Gene Script Co. Ltd (Nanjing, Jiangsu, China).

The optimized rHA ORF genes were cloned into the pFastBac vector plasmid to make recombinant Bacmid baculovirus DNAs using DH10Bac competent cells (rAcNPV, Invitrogen, Carlsbad, CA, USA). A recombinant baculoviruses (rBVs) expressing influenza rHA protein was generated by transfection of *Sf*9 insect cells according to the manufacturer's instruction. *Sf*9 cells were infected (1.0 multiplicity of infection, MOI) with rBVs expressing HA. After 4 days, the cell culture supernatants were harvested for preparation of vaccines.

#### Analysis of the Recombinant HA Proteins with Western Blot

The indirect immunofluorescence assay (IFA) was carried out to test the expression of HA in infected *Sf9* cells after 72 h. The primary antibody was chicken anti-sera against Re6 antigen at a dilution of 1:800, and the secondary antibody was FITC labeled goat anti-chicken IgY (1:500) (Southern Biotech, Birmingham, AL, USA).

The infected insect cells were harvested after inoculation of a single rHA rBVs for 96 h, and disrupted by ultrasonic for 30 min to prepare cell lysates under the condition of maintaining the temperature at 0–4°C. The sonicated cell lysates were cleared by low-speed centrifugation (10,000 × *g* for 20 min at 4°C) to remove cell debris. The rHA VLPs in the supernatants were pelleted by ultracentrifugation (100,000 × *g* for 60 min). The sedimented particles were resuspended in phosphate-buffered saline (PBS) at 4°C overnight and further purified through a 20–30–60% discontinuous sucrose gradient at 100,000 × *g* for 2 h at 4°C (17).

For Western blot analysis to determine the expression of rHA, infected cells were dissolved in sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE) sample buffer (50 mM Tris, 3% β-mercaptoethanol, 2% SDS, 10% glycerol), separated by SDS-PAGE, and then transferred to nitrocellulose membranes, subsequently probed with sera (1: 1,000) derived from Re5 and Re6 vaccine-immunized chickens, respectively. The secondary antibodies were HPR-labeled goat anti-chicken IgY (1:2,000) (SoutherBiotech, Birmingham, AL, USA).

The functionality of rHA protein incorporated into VLPs was quantified by hemagglutination assay (HA assay) using Scientific).

1% (v:v) chicken red blood cells. The concentration of protein was measured by Pierce BCA Protein Assay Kit (Thermo Fisher

## Electron Microscopy

For negative staining of VLPs, sucrose gradient-purified VLPs were applied to a carbon-coated Formvar grid for 30 s. Excess VLPs suspension was removed by blotting with filter paper, and the grid was immediately stained with 1% phosphotungstic acid for 90 s. Excess stain was removed by filter paper, and the samples were examined using a transmission electron microscope.

#### Vaccine and Immunopotentiators

The purified recombinant HA proteins were recovered to same volume before ultracentrifugation treatment. The diluent is phosphate buffer solution (pH 7.2, PBS). The rHA (ca. 12 µg/ dose) were prepared as water-in-oil emulsion with or without immunopotentiator CVCVA5, named as rHA or rHA-CVCVA5, respectively. The rHA antigen stock solutions were normalized so that the vaccine preparations had the same reciprocal of hemagglutination titers equal to or higher than 256 as measured with 1% chicken red blood cells. This range of titers is a prerequisite for the monovalent Re6 vaccine preparations with mineral oilin-water emulsion adjuvant in the manufacture company. The preparation of rHA-CVCVA5 was followed as described in our previous reports (18–20). Briefly, the components of CVCVA5 are composed of an aqueous phase that contains L–D isoform muramyl dipeptide (MDP) (InvivoGen), poly I:C (InvivoGen), levamisole hydrochloride (Sigma) added to rHA antigen solution as well as an oil phase of resiquimod (InvivoGen) and imiquimod (InvivoGen) added to Marcol 52 mineral oil (ESSO). One volume aqueous phase was mixed with three volumes of an oil phase. The rHA vaccine preparation in water-in oil emulsion did not contain the adjuvant ingredients of CVCVA5. Preparation of the emulsion form of CVCVA5 was similar to that of the rHA-CVCVA5, having PBS substituted with the rHA antigen solutions and 10-fold concentration of the ingredients of the CVCVA5 as those in the rHA-CVCVA5. The H5-CVCVA5 vaccines were prepared by mixing the commercial H5 (Re6, Weike) subtype vaccine (v/v, 1:9) with an emulsion form of CVCVA5.

## Immunization and Challenge

Groups (*n*= 25 in each group) of 3-week-old age specific pathogenfree (SPF) chickens (Leghorn breed, *Gallus gallus domesticus*) or 15- to 18-day-old age Hy Line Brown breed commercial chickens (*Gallus gallus domesticus*) were used to determine the immunogenicity and efficacy of the recombinant rHA VLPs. The vaccine test groups included the rHA with or without of CVCVA5. The commercial H5 subtype vaccines (Re6) were set as comparison control with or without CVCVA5, and a group of unvaccinated chickens as a naïve control. Serum samples from all birds were collected at 2 and 3 weeks postvaccination (PV).

Chickens from each test group were randomly selected and intranasal challenged with 100 LD50 (0.1 ml) H5 wild-type influenza viruses (ZJ and DT) at 3 weeks PV. These challenge viruses were derived from the clade of 2.3.4.6 (A/Chicken/Zhejiang/201, ZJ, H5N2, *n* = 10 in each challenge group) and clade 7 (A/Chicken/Huadong/4/2008, DT, H5N1, *n* = 10 in each challenge group) (19). Chickens were monitored and recorded clinical signs and body weights for 14 days post challenge (PC). All survival chickens were killed humanely at the end of monitoring experiments. Oropharyngeal and cloacal swab samples were collected at 3, 5, and 7 days PC, or collected from the dead chickens during the observation period. Virus isolation from the swab samples was performed as previously described (21).

## Antibody and Cytokine Tests

Serum antibody levels were titrated by hemagglutinin inhibition (HI) assay or serum-neutralization (SN) assay in DF-1 cell (ATCC, CRL-12203) (19, 22). The SN assay was performed as following described. Briefly, all serum samples from chickens were heat-inactivated (56°C, 30 min). The twofold serial dilutions of serum samples were mixed with equal volumes of 200 mean tissue culture infective dose (TCID50) of H5 subtype variant viruses (ZJ, DT). After 1 h incubation at 37°C, the mixtures were inoculated into the 96-well plates with a monolayer of DF-1 cells, a chicken embryo fibroblasts cell line, and then incubated to observe cytopathic effects (CPE) daily for up to 5 days. The cell death or CPE were used to determine the end-point titers that were calculated as the highest reciprocal dilution of sera at which virus infection is blocked in DF-1 cells according to the method of Reed and Muench (23).

The chicken cytokine levels of IFN-γ (Invitrogen, CA, USA) and IL-4 (USCN, Wuhan, China) in sera at 2 and 3 weeks PV were measured by commercially available ELISA kits following the manufacturer's instructions as described in our previous reports (19, 20). The mucosal antibodies from bronchoalveolar lavage fluids (BAL) were determined by an HI assay.

## Cytotoxic T Lymphocytes (CTL) Assay

Cytotoxic T lymphocytes activity was measured by the lactate dehydrogenase (LDH) cytotoxicity assay kit (Takara, Kyoto, Japan), which is a non-radioactive method to detect the stable cytosolic enzyme LDH released from lysed cells. The assay was performed with the manufacturer's instruction and our previous study (19, 20). Briefly, the target cells that derived from the inbred SPF chickens (B12/B12) embryo fibroblast cells were infected for 8 h with ZJ or DT virus with MOI of 0.1. The target cells (104 cells in 100 µl/well) infected or uninfected were also seeded to each well. The effector cells derived from peripheral blood mononuclear cells isolated from the inbred SPF chickens at 21 days PV, which previously immunized with H5 rHA vaccine or the commercial H5 subtype vaccine (Re6) with or without CVCVA5 adjuvants, respectively. Various amounts of effector T cells were added to each well. Each cell sample was plated in triplicates. Microtiter plates were centrifuged at 250 × *g* for 5 min and incubated in a humidified chamber at 37°C, 5% CO2. After 4-h incubation, the supernatants were harvested, followed by the addition of the substrate tetrazolium salt. The OD values were read at 490 nm in an enzyme-linked immunosorbent assay reader. The specific LDH activity release was calculated by using the following formula: (experimental release − spontaneous release)/(maximum release − spontaneous release) × 100.

#### Statistics

Experimental data are presented as mean ± SD of the mean. Prism 7 (GraphPad Software, Inc., San Diego, CA, USA) was used for data analysis. The statistical significance was analyzed with Student's *t*-test or a one-way analysis of variance. Comparisons used to generate *P* values are indicated by horizontal lines (\**P* < 0.05; \*\**P* < 0.01; \*\*\**P* < 0.001).

#### Ethics Statement

All animal studies were carried out in accordance with the recommendations in the National Guide for the Care and Use of Laboratory Animals. The protocol (VMRI-AP150306) was approved by the Review Board of National Research Center of Engineering and Technology for Veterinary Biologicals, Jiangsu Academy of Agricultural Sciences and Yangzhou University. The surgery and euthanasia were performed under anesthesia with sodium pentobarbital solution (100 mg/kg body weight) *via* an intravenous route to minimize suffering. All experiments involved in the live H5 subtype viruses were performed in the biosafety level 3 laboratory facilities. At the end of experiments, the discarded live viruses, wastes, and infected animal carcasses were autoclaved and incinerated to eliminate biohazards.

## RESULTS

### Production and Analysis of rHA-Based Influenza VLPs

The recombinant HA proteins were produced in insect cells, which were infected with rBVs expressing the H5 subtype AIVs consensus HA gene with poly-basic residues at the cleavage site. The expression of rHA proteins was observed with indirect IFA in *Sf*9 insect cells 72 h after infection with rHA rBVs (**Figure 1**),

whereas there was no specific fluorescence in control baculovirus infected cells (**Figure 1**). As for further evidence, two major visible bands, HA1 and HA2 fragments, were presented in the Western blot analysis which is consistent with a previous study (24). The functionality of rHA incorporated into VLPs (rHA VLPs) was assessed by the hemagglutination assay with 1% cRBC. No activity of hemagglutination was observed when the VLPs were denatured by heat treatment at 100°C for 10 min (data not shown). The reciprocal hemagglutinin titers of the rHA rBV-infected *Sf*9 cell culture supernatants reached to 64 before treating with ultrasonic, and increased to 2,048 after ultrasonic treatment. The increment in HA titers was speculated due to the release of rHA VLPs from the cellular membranes after the ultrasonic treatment of cell lysis. The total protein concentration of culture supernatants was 120 µg/mL measured by the BCA kit.

The size and morphology of rHA VLPs were examined by electron microscopy (**Figure 1**). The morphology of VLPs resembles that of influenza virus particles, and the spikes were observed on their surfaces which is characteristic of influenza virus HA proteins on virions.

## H5 rHA Displays Cross-Reactivity to Immune Sera Induced by H5 Vaccines

The antigenicity of H5 rHA vaccine was determined *via* HI tests in cross reactions with sera raised by antigenically different H5 vaccines, and compared to those reactions with the viral antigens of Re5, Re6, Re7, and Re8. Except the HI titers from the homologous pairs of serum and antigen, the HI titers against the rHA antigen were higher than those that of the viral antigens, the Re5, Re6, Re7, and Re8, cross reacted with sera induced by Re5, Re6, Re7, and Re8 vaccines, respectively. Compared to the sera derived from Re5, Re6, and Re8 vaccines, the serum derived from Re7 vaccine was poorly reactive to rHA and other viral antigens (**Figure 2**).

Figure 2 | Hemagglutinin inhibition (HI) cross-reactivity of H5 rHA protein to sera raised by different vaccines. The HI tests were carried out with the recombinant protein of consensus H5 HA (rHA), viral antigens of Re5, Re6, Re7, and Re8 as 4 hemagglutinin units (HAU)/25 μl in three replicates. The six inner genes of Re5, Re6, Re7, and Re8 vaccine strains are derived from A/Puerto Rico/8/1934 virus (PR8, H1N1), the HA and neuraminidase genes of Re5 from strain of A/duck/Anhui/1/2006 (clade 2.3.4), of Re6 from virus of A/duck/Guangdong/S1322/2006 (clade 2.3.2), of Re7 from A/Chicken/ Liaoning/S4092/2011 (clade 7.2), and of Re8 from A/chicken/Guizhou/4/13 (clade 2.3.4.4). The statistical significance was analyzed with Student's *t*-test. \*\*\**P* < 0.001.

## H5 rHA VLPs Vaccines Elicit Immune Responses in Chickens

To determine whether influenza H5 rHA VLPs induce immune responses specific to influenza virus HA protein, groups of 3-week-old SPF chickens were subcutaneously vaccinated one time with influenza rHA vaccine (12 µg/dose) with or without immunopotentiator (rHA, rHA-CVCVA5), and Re6 vaccine in the presence or absence of immunopotentiator as controls (Re6, Re6-CVCVA5). The levels of serum antibody against different antigens were measured by HI assay at 2 and 3 weeks after a single dose vaccination. Using 4 HA units (HAU) of Re6 as a testing antigen, the HI titers of chickens shot with the Re6 vaccine were higher than those that induced by the rHA vaccine, similarly, HI titers induced by Re6-CVCVA5 vaccine were higher than those by the rHA-CVCVA5 vaccine at 2- or 3-week PV (**Figure 3A**). However, the HI titers raised by vaccination with rHA-containing vaccines were higher than those induced by Re6-containing vaccines when measured with the 4 HAU of Re7, Re8, or rHA testing antigens, respectively (**Figures 3B–D**).

The efficacy of rHA vaccines was also assessed in 15- to 18-dayold commercial chickens. Consistent with the results of SPF chickens, lower HI levels were observed in the rHA-containing vaccine groups compared to those in the Re6-containing vaccines groups, as measured by using 4 HAU of the Re6 as a testing antigen. In contrast, the HI antibody titers in rHA-containing vaccine groups were higher than those of the Re6-containing vaccine groups when determined with the rHA and Re7 antigens, respectively. Similar levels of HI antibodies against the Re8 antigen were observed in chickens which received Re6- or rHA-containing vaccines (**Figures 3E–H**).

### H5 rHA VLPs Vaccination Induces Enhanced Levels of Serum Virus Neutralization Titers and Mucosal Antibodies in SPF Chickens

Serum-neutralization activity against the variants of ZJ or DT virus strains was carried out. Serum samples from the chickens that were immunized with rHA VLPs vaccine in the presence or absence of CVCVA5 and Re6 plus CVCVA5 vaccines showed significantly higher neutralization titers than those of the chickens that were immunized with Re6 vaccine without immnopotentiators at 2 and 3 weeks PV (**Figures 4A,B**).

The mucosal antibody titers in the BAL fluids at 3 weeks PV were also measured by HI assay with Re6, Re7, Re8, and rHA VLPs as 4 HAU antigens. Higher HI titers in BAL fluids from the rHA-CVCVA5 or Re6-CVCVA5 vaccine groups were observed compared to those groups that received rHA or Re6 vaccine without immunopotentiators (**Figure 4C**).

## H5 rHA VLPs Vaccination Induces Increased Levels of Serum Cytokines and CTL Responses in SPF Chickens

To investigate the induction of cytokines after vaccination, we measured the levels of IFN-γ and IL-4 in serum at 2 and

3 weeks PV. Both levels of IFN-γ and IL-4 were significantly increased in the chickens received rHA-CVCVA5 or Re6- CVCVA5 vaccine, and slightly increased in the rHA vaccine group (**Figures 5A,B**). The increased magnitude of IFN-γ was higher than that of the IL-4.

Cytotoxic T lymphocytes play an important role in the control of virus infections. The adjuvanted rHA and Re6 vaccines elicited a robust CTL response compared to those of the non-adjuvanted rHA and Re6 vaccines. Further analysis demonstrated that vaccination of chickens with the adjuvanted rHA vaccine induced slightly higher levels of the CTL responses than those by vaccination with Re6 vaccines in the presence or absence of adjuvants. The CTL immune responses from the Re6 vaccine groups were low which was similar to that of the control groups (**Figures 5C,D**).

## H5 rHA VLPs Vaccine Confers Cross-Protection against Heterogeneous Strains

To investigate whether rHA vaccine could confer cross-protection against a lethal challenge with heterogeneous virus, the rHAvaccinated chickens were challenged with ZJ (clade 2.3.4.6) and DT (clade 7) viruses, respectively (**Figure 6**). After a lethal ZJ virus challenge, all birds survived in vaccine groups of rHA, rHA-CVCVA5, and Re6-CVCVA5 (**Figure 6A**). In contrast, only 30% (3/10) chickens survived in the Re6 vaccine group, and all chickens died of infection in the control group. In the challenge test with DT virus, all chickens survived the infection in the rHA and the rHA plus CVCVA5 vaccine groups. Two out of ten chickens were dead in the groups of Re6-CVCVA5. Only two out of ten chickens survived in the Re6 (clade 2.3.2) vaccine group and no chicken survived in the control group (**Figure 6B**).

PBMCs derived from the SPF chickens vaccinated with rHA or Re6 vaccine with or without adjuvants, or from control group, were used as effector cells (C,D). \**P* < 0.05, \*\*\**P* < 0.001.

After challenge with ZJ virus, the body weight still slightly increased in chickens received with the rHA and the rHA-CVCVA5 vaccines. The birds in the rHA-CVCVA5 vaccine group gained the most weight among all challenge groups. The body weights of the chickens were nearly unchanged in the Re6-CVCVA5 vaccine group during 14 days of the monitoring period. In contrast, the chickens significantly lost weight in the Re6 vaccine and the control groups after challenge with ZJ virus (**Figure 6C**). However, only the rHA-CVCVA5 vaccine group showed a weight gain after challenge with DT virus. The birds in the rHA VLPs and Re6-CVCVA5 groups slightly lost weight at day 8–11, and then quickly recovered to initial weight during the 14-day of observation period. Similar to the challenge with ZJ virus, the chickens also exhibited rapid weight loss in the groups of the Re6 vaccine and control after challenge with DT virus (**Figure 6D**).

The excreted viruses *via* oropharynx and cloaca were analyzed to determine the virus replication at 3, 5, and 7 days PC with ZJ and DT viruses (**Table 1**). After ZJ virus challenge, virus shedding was not detected in chickens from the rHA-CVCVA5 vaccine group, and one bird was positive of virus isolation in both the rHA and the Re6-CVCVA5 groups. All birds were shedding viruses in the Re6 vaccine and control groups. A similar pattern of virus shedding was observed after DT virus challenge (**Table 1**). Chickens in the rHA-CVCVA5 group were free from virus detection. Three chickens in the rHA group and four chickens in the Re6-CVCVA5 group recovered viruses from swab samples. All birds in the Re6 vaccine and control groups excreted viruses PC.

#### DISCUSSION

In this study, we investigated immune responses induced by consensus HA sequence-based VLPs vaccine (rHA) in chickens, including serological and mucosal HI antibodies, cytokines, and induction of cross-protection against two heterologous subclades of H5 subtype AI virus challenge.

Development of vaccine based on the consensus HA sequences is one of the potential strategies to induce broadly protective immune responses against influenza viruses with high mutation rates and antigenic diversities. Consensus sequences encode the most common residues found at each position for the selected population, which are compatible with the site mutation residues in different clade viruses. We have compared the consensus HA sequence from the amino acid residues with that from the nucleotide level, and there were slightly differences mainly located at 310–340 amino acid residues (data not shown). It is expected

Figure 6 | Survival rates and body weight changes of the specific pathogen-free (SPF) chickens after heterologous virus challenge. At 3 weeks postvaccination, groups of SPF chickens (*n* = 10) were intranasally challenged with 100 LD50 viruses of ZJ (A/Chicken/Zhejiang/2011, H5N2, subclade of 2.3.4.6) (A), or DT (A/ Chicken/Huadong/4/2008, H5N, clade 7) (B), and the clinical symptoms or death, and body weight changes were daily monitored and recorded for 14 days post challenge (C,D).


*ZJ virus is the abbreviation virus strain name of A/Chicken/Zhejiang/2011 (H5N2), attributed to the subclade of 2.3.4.6, and DT is virus of A/Chicken/Huadong/4/2008 (H5N1), clade 7. The oropharyngeal and cloacal swab samples were collected at 3, 5, and 7 days post challenge. Virus positivity or shedding was determined by inoculating each swab solution into 3 eggs of 10-day-old specific pathogen-free chicken embryos. The virus isolation positive swab means one or more of the inoculated egg allantoic fluids with reciprocal of hemagglutination titers higher than 4. rHA, H5 rHA VLPs vaccine. Re6, a commercial vaccine strain with HA and neuraminidase from A/duck/Guangdong/S1322/2006 (H5N1, clade 2.3.2) and six inner genes from PR8 (H1N1).*

that consensus sequences at nucleotide levels would be stable in mRNA translation. Besides, we have developed a wild-type Re6 HA-based vaccine which was poorly immunogenic and conferred partial protection against challenge with ZJ and DT viruses (data not published). In contrast to the single virus strain sequencebased subunit vaccine providing strain-restricted immunity, the consensus-based vaccines have been previously investigated as a strategy for eliciting broadly reactive immune responses for different pathogens, including chikungunya virus (25), hepatitis B virus (26), hepatitis C virus (27), Middle East respiratory virus (28), HIV-1 (29), and influenza (10, 30, 31).

The consensus rHA VLPs based vaccines elicited protective immune responses similar to those of the inactivated wholevirus vaccine, and could be applied to the assessment criteria of commercial inactivated virus vaccine, the index including serum HI titers and protective efficacy PC. We have also tested the efficacy of a single copy of the M2 ectodomain (M2e) as a broad-spectrum influenza vaccine candidate in chickens (22). Nevertheless, its further application was restricted by the lack of an appropriate method to evaluate the immune response elicited by the M2e antigens in chickens.

The rHA vaccine in the absence of CVCVA5 elicited higher levels of serological and mucosal antibody responses than those of the Re6 vaccine without CVCVA5 in terms of the HI antibody titers against the Re7 and Re8 antigens in both of the SPF and commercial chickens, and of the SN titers in DF-1 cells. However, higher levels of HI titers after Re6 vaccination were observed when using Re6 as a test antigen in both the SPF and commercial chickens because of an Re6 antigen exactly matching with the Re6 vaccine. In contrast to the HI titers, the neutralizing titers more closely reflect the capability of antibodies inhibiting the replication and propagation of virus. The results of humoral immune responses indicated that the rHA-based vaccine can elicit broad-spectrum antibodies reactive to different antigens, which is consistent with other influenza vaccines based on consensus HA sequences (10).

The VLPs vaccines were superior to the inactivated wholevirus vaccine in eliciting cell-mediated immune responses in animal models as reported in previous studies (17, 32, 33). The residual recombinant baculoviruses in the rHA vaccine preparations can activate Toll-like receptors, which correlated with cellular immune response. In this study, we have also monitored slightly higher levels of IFN-γ and CTL responses induced by the rHA VLPs vaccine in the absence of CVCVA5 compared to those of the Re6 vaccine without CVCVA5.

The challenge studies were carried out with two heterogeneous H5 subtype virus strains (ZJ and DT). The results from this study provide evidence that the rHA VLPs vaccines could be applicable to the field where there exist different subclades of H5 subtype viruses. In contrast to the Re6 vaccine in the absence of immunopotentiator, the rHA vaccine without immunopotentiator broadly protected the vaccinated chickens free from weight loss and death after infection with heterogeneous viruses. The multiimmune components including humoral antibodies, cytokine responses, and CTL activities elicited by the rHA vaccines with CVCVA5 adjuvants could contribute to broad protection against H5 subtype variants infection.

Besides construction of the broad-spectrum antigen, the adjuvant is another option to improve the efficacy of vaccines. The adjuvant, CVCVA5, was shown to be effective on improving the efficacy of the commercial inactivated monovalent or polyvalent vaccines in our previous studies (18, 19). Compared to the rHA and Re6 vaccines in the absence of adjuvant, CVCVA5 adjuvanted vaccine formulations moderately increased the serum antibody levels, significantly improved antibody levels in BAL, and also significantly enhanced the levels of CTL immune responses and cytokines such as IFN-γ and IL-4 in both adjuvanted rHA and Re6 vaccines. There is no significant difference in the protection rates among the groups of rHA vaccine with or without immunopotentiator, suggesting major roles of humoral antibodies in conferring cross-protection by consensus rHA vaccination. In contrast to the rHA vaccine, the increased cellular immune responses in the commercial Re6-CVCVA5 vaccine appear to play pivotal roles in providing effective protection against heterologous virus challenge when antibodies alone are insufficient for effective protection.

In this study, we have reported HA consensus sequencebased H5 rHA VLPs vaccines, which elicited a broad-spectrum serum and mucosal antibody response, and protected chickens experimentally challenged with different subclades of H5 subtype viruses. The results from this study support that development of subunit VLP vaccines based on consensus HA sequences conferring a broad-spectrum protection can provide a potential application in poultry farm with continuing antigenic drift variants.

## ETHICS STATEMENT

All animal studies were carried out in accordance with the recommendations in the National Guide for the Care and Use of Laboratory Animals. The protocol (VMRI-AP150306) was approved by the Review Board of National Research Center of Engineering and Technology for Veterinary Biologicals, Jiangsu Academy of Agricultural Sciences and Yangzhou University. The surgery and euthanasia were performed under anesthesia with sodium pentobarbital solution (100 mg/kg body weight) *via* intravenous route to minimize suffering. All experiments involved in the live H5 subtype viruses were performed in the biosafety level 3 laboratory facilities. At the end of experiments, the discarded live viruses, wastes, and infected animal carcasses were autoclaved and incinerated to eliminate biohazards.

## AUTHOR CONTRIBUTIONS

Conceived and designed the experiments: YT, PW, JL, XZ, and XL. Performed the experiments: YT, PW, JL, XZ, MM, and LF. Analyzed the data: YT, PW, JL, XZ, XL, and DP. Contributed reagents/materials/analysis tools: YT, PW, JL, XZ, MM, LF, JH, and DP. Wrote and edited the paper: YT, PW, S-MK, DP, and XL. All authors read and approved the final manuscript.

## FUNDING

This work was partially supported by the National Key R&D Program of China (2017YFD0500706) (YT), Central Finance Funds for Agrotechnical Popularization grants (TG(15)067) (YT), Jiangsu Agriculture Science and Technology Innovation Fund (CX(15)1063) (YT), the Special Fund for Agroscientific Research in the Public Interest (201303046) (YT), and NIH AI105170 (S-MK), AI093772 (S-MK), AI119366 (S-MK). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

## 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. YT, PW, JL, XZ, MM, LF, and JH are authors on a pending China patent application describing the culture and produce of rHA in bioreactor system (Title: Preparation method of the H5 subtype HA-based subunit vaccine. Reference number: 201710140542.2).

*Copyright © 2017 Wu, Lu, Zhang, Mei, Feng, Peng, Hou, Kang, Liu and Tang. 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) or licensor 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.*

*Karine Chartrand1 , Marie-Ève Lebel <sup>1</sup> , Esther Tarrab1 , Pierre Savard <sup>2</sup> , Denis Leclerc2 and Alain Lamarre1 \**

*1 Immunovirology Laboratory, Institut national de la recherche scientifique (INRS), INRS-Institut Armand-Frappier, Laval, Quebec, Canada, 2 Infectious Disease Research Center, Department of Microbiology, Infectiology and Immunology, Laval University, Quebec City, Quebec, Canada*

#### *Edited by:*

*Rajko Reljic, St George's, University of London, United Kingdom*

#### *Reviewed by:*

*Pablo Penaloza, Northwestern University, United States Owen Kavanagh, York St John University, United Kingdom*

#### *\*Correspondence:*

*Alain Lamarre alain.lamarre@iaf.inrs.ca*

#### *Specialty section:*

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

*Received: 18 September 2017 Accepted: 11 December 2017 Published: 04 January 2018*

#### *Citation:*

*Chartrand K, Lebel M-È, Tarrab E, Savard P, Leclerc D and Lamarre A (2018) Efficacy of a Virus-Like Nanoparticle As Treatment for a Chronic Viral Infection Is Hindered by IRAK1 Regulation and Antibody Interference. Front. Immunol. 8:1885. doi: 10.3389/fimmu.2017.01885*

Although vaccination has been an effective way of preventing infections ever since the eighteenth century, the generation of therapeutic vaccines and immunotherapies is still a work in progress. A number of challenges impede the development of these therapeutic approaches such as safety issues related to the administration of whole pathogens whether attenuated or inactivated. One safe alternative to classical vaccination methods gaining recognition is the use of nanoparticles, whether synthetic or naturally derived. We have recently demonstrated that the papaya mosaic virus (PapMV)-like nanoparticle can be used as a prophylactic vaccine against various viral and bacterial infections through the induction of protective humoral and cellular immune responses. Moreover, PapMV is also very efficient when used as an immune adjuvant in an immunotherapeutic setting at slowing down the growth of aggressive mouse melanoma tumors in a type I interferon (IFN-I)-dependent manner. In the present study, we were interested in exploiting the capacity of PapMV of inducing robust IFN-I production as treatment for the chronic viral infection model lymphocytic choriomeningitis virus (LCMV) clone 13 (Cl13). Treatment of LCMV Cl13-infected mice with two systemic administrations of PapMV was ineffective, as shown by the lack of changes in viral titers and immune response to LCMV following treatment. Moreover, IFN-α production following PapMV administration was almost completely abolished in LCMV-infected mice. To better isolate the mechanisms at play, we determined the influence of a pretreatment with PapMV on secondary PapMV administration, therefore eliminating potential variables emanating from the infection. Pretreatment with PapMV led to the same outcome as an LCMV infection in that IFN-α production following secondary PapMV immunization was abrogated for up to 50 days while immune activation was also dramatically impaired. We showed that two distinct and overlapping mechanisms were responsible for this outcome. While short-term inhibition was partially the result of interleukin-1 receptor-associated kinase 1 degradation, a crucial component of the toll-like receptor 7 signaling pathway, long-term inhibition was mainly due to interference by PapMV-specific antibodies. Thus, we identified a possible pitfall in the use of virus-like particles for the systemic treatment of chronic viral infections and discuss mitigating alternatives to circumvent these potential problems.

Keywords: plasmacytoid dendritic cells, PapMV, interferon-**α**, IRAK1, Sca-1, antibodies, lymphocytic choriomeningitis virus clone 13

## INTRODUCTION

Type I interferons (IFN-I), mainly IFN-α and IFN-β, are a family of cytokines with potent antiviral and immunomodulatory properties. The effects of these cytokines on their milieu are complex and affect multiple cells of the immune system by: (i) inducing activation of dendritic cells (DCs) (1, 2); (ii) sustaining activation of CD8<sup>+</sup> T cells (3, 4); and (iii) inducing differentiation of B cells into antibody secreting cells (5, 6). Thus early IFN-I production is essential for the control of most viral infections such as mouse hepatitis virus (MHV) (7), lymphocytic choriomeningitis virus (LCMV) (8, 9), or simian immunodeficiency virus (SIV) (10). However, whereas early and transient expression of IFN-I controls the infection, prolonged exposure bears detrimental effects to the host's immune response (10–12). The balancing act between the positive and negative effects of the IFN-I response is observed in the clinical setting as displayed by the well-documented adverse effects of IFN therapy (13, 14). Thus, despite being the standardof-care against various diseases, emerging therapies now focus on IFN-free alternatives.

Upon viral infection, plasmacytoid DCs (pDCs) are characterized as the main producers of IFN-I [reviewed in Ref. (15, 16)]. This major characteristic is mainly due to their Toll-like receptor (TLR) expression profile. While other innate immune cells express a wide array of TLRs, pDCs mainly express the endosomal TLR7 and TLR9 (16), which bind genetic material typically associated with viral pathogens, such as ssRNA and unmethylated DNA, respectively. These receptors are also capable of recognizing analogs of their natural ligands such as imidazoquinoline and CpG for TLR7 and TLR9, respectively (17). Through their production of IFN-I, pDCs also serve as a bridge between innate and adaptive immune responses as illustrated by their ability to activate natural killer cells (18), other DCs (19, 20) as well as T cells (18, 21). Consequently, upon pDC depletion, mice become highly susceptible to viral infection with MHV (7), herpes simplex virus (22, 23), LCMV (24), vesicular stomatitis virus (18), or the murine cytomegalovirus (18) among others.

By harnessing the central role of pDCs, we have demonstrated the potential for the use of a plant virus-like nanoparticle as a vaccine candidate as well as an adjuvant in various infectious models (25–34). Our platform is based on the papaya mosaic virus (PapMV) nanoparticle that contains a non-replicative synthetic ssRNA rendering it safe for future human use. The synthetic ssRNA found inside the capsid is recognized by the TLR7 of pDCs (31), leading to the production of IFN-α (26, 31, 35), IL-6 (30, 31) along with other cytokines and chemokines. PapMV resultantly activates DCs, macrophages, T cells as well as B cells (26, 28, 31, 35), making this platform versatile in activating the immune system. Furthermore, we have shown that PapMV induces protective immune responses against pathogens when used as a vaccine platform or an adjuvant (26–34, 36) and slows melanoma development when used in an immunotherapeutic setting (37).

Considering that PapMV induces strong IFN-I responses, we sought to evaluate its potential as an immune adjuvant for the treatment for chronic viral infections with the objective of replacing exogenous IFN-α administration with endogenous IFN-α secretion following systemic PapMV delivery. This approach would provide a universally applicable immune stimulatory molecule that could be used against all viral infections without requiring expression of defined viral antigens. We observed that treatment of mice chronically infected with the persistent strain of LCMV (LCMV-Cl13) with PapMV was unable to clear the infection. Moreover, multiple administrations of PapMV induced an immune tolerance as shown by the almost complete abrogation of IFN-α production following secondary PapMV administration. We show that this tolerization is the result of a combination of factors including interleukin-1 receptor-associated kinase 1 (IRAKI) degradation and interference by PapMV-specific antibodies. This information will be crucial for further clinical development of the PapMV platform.

#### MATERIALS AND METHODS

#### Ethics Statement

This study was performed in accordance with the Canadian Council on Animal Care guidelines. All *in vivo* experiments were reviewed and approved by the Institut national de la recherche scientifique animal care committee.

#### Mice

Female 6- to 10-week-old C57BL/6 mice were purchased from Charles River Laboratories. JHT mice were kindly provided by Dr. Rolf Zinkernagel (Zurich University, Switzerland).

#### Cells and Virus

Lymphocytic choriomeningitis virus Cl13 was kindly provided by Dr. Rolf Zinkernagel (Zurich University, Switzerland). MC57G fibroblast were cultured in minimal essential medium with Earle's salt (Wisent, St-Bruno, QC, Canada) containing 5% heat inactivated fetal bovine serum (FBS) (PAA Laboratories, Mississauga, ON, Canada).

#### PapMV Nanoparticles

PapMV nanoparticles were provided by Folia Biotech (Quebec, QC, Canada) and were produced as described before (30). Briefly, coat proteins are self assembled *in vitro* around a non-coding ssRNA. Lipopolysaccharide (LPS) contamination was always <50 endotoxin units/mg protein and considered as negligible.

#### LCMV Cl13 Infection and Treatment

Mice were infected with 2 × 106 pfu of LCMV Cl13 i.v. and treated with 400 µg of PapMV i.v. on days 3 and 5. Serum was collected 6 h following each treatment to assess IFN-α production. Blood was

**Abbreviations:** BFA, Brefeldin A; BMpDCs, bone marrow-derived plasmacytoid dendritic cells; dpi, days postinfection; FBS, fetal bovine serum; HBV, hepatitis B virus; HCV, hepatitis C virus; IFN, interferon; IRAK1, interleukin-1 receptorassociated kinase 1; LCMV Cl13, lymphocytic choriomeningitis virus clone 13; LPS, lipopolysaccharide; MFI, mean fluorescence intensity; MHC-I, class I major histocompatibility complex; PapMV, papaya mosaic virus nanoparticle; pDCs, plasmacytoid dendritic cells; Poly I:C, polyinosinic:polycyticylic acid; TLR, tolllike receptor; WT, wild type.

collected 8 days postinfection (dpi) and mononuclear cells were isolated by density gradient over Ficoll-Paque (GE Healthcare Life Sciences, Mississauga, ON, Canada) and centrifuged at 1,200 rpm for 20 min at room temperature. Cells were collected and washed with PBS then stained for 30 min at 37°C with GP33–41 PE tetramers, which were synthesized as previously described (38), to label CD8<sup>+</sup> T cells specific for the MHC-I gp33 epitope of LCMV. Extracellular staining was performed on unwashed cells for 20 min at 4°C. Following a wash, cells were fixed with fixation buffer (Biolegend, San Diego, CA, USA) for 20 min at room temperature then analyzed by flow cytometry on a BDLSR Fortessa (BD Biosciences, Mississauga, ON, Canada). Spleens collected 30 dpi were disrupted between frosted microscope slides and cells were stained to assess CD8<sup>+</sup> T cells GP33–41<sup>+</sup> cells as described above. Cells were also incubated with Brefeldin A (BFA) (Sigma, Oakville, ON, Canada) for 5 h at 37°C to inhibit vesicular transport. Spleen cells were then stained for intracellular cytokine production (see below). Blood, spleen, kidney, liver, and brain were also collected 30 dpi to assess viral burden. Organs were mechanically disrupted and supernatants were tittered on MC57G cells by focus forming assay to assess viral burden as previously described (39).

#### Immunizations

Mice were injected with 100 or 400 µg of PapMV i.v., 50 µg of imiquimod (R837) (InvivoGen, San Diego, CA, USA), 25 µg of LPS (Sigma), 50 µg of Poly I:C (InvivoGen), or control.

#### Organ Processing

Spleen and bones from hind legs were collected at various time points following immunization. Spleens were subjected to digestion with 1 mg/mL of collagenase D (Roche, Mississauga, ON, Canada) for 15 min at 37°C. Femurs, tibias, and iliac crests were flushed and single cell suspensions from both spleen and bone marrow were subjected to red blood cell lysis followed by flow cytometry staining.

### Bone Marrow-Derived Plasmacytoid Dendritic Cells (BMpDCs)

Bone marrow-derived pDCs were prepared by flushing the bone marrow of femurs, tibias and iliac crests, and subjected to red blood cell lysis. Cells were seeded at 2 × 106 cells/mL in RPMI 1640 (Wisent) containing 10% FBS, 100 IU penicillin, 100 µg/ ml streptomycin (Wisent), 55 µM β-mercaptoethanol, 1 mM sodium pyruvate, 1× MEM non-essential amino acids, and 10 mM HEPES (Gibco) supplemented with 200 ng/ml of FLT3-L (BioXcell, Lebanon, PA, USA). On days 7–9, cells were stimulated with 100 µg/ml of PapMV, 25 µg/ml of imiquimod (R837) (InvivoGen), 12.5 µg/ml of polyinosinic:polycyticylic acid (Poly I:C) (InvivoGen) or control. On days 8–10, supernatants were frozen at −20°C for IFN-α detection and cells stained for flow cytometry analysis or cell sorting.

## Serum Transfer

Mice were immunized with 100 µg PapMV i.v. on day 0. On day 5 or 25, mice were euthanized and blood was collected by cardiac puncture. Blood was allowed to clot for 30 min at room temperature. Tubes were then centrifuged at 1,500 *g* for 10 min at room temperature. Sera were pooled and injected i.p. to mice whereby the serum from two mice was used to inject one mouse. After 24 h, mice were immunized with 100 µg of PapMV or control i.v. Serum was collected 6 h later to quantify IFN-α and PapMVspecific antibodies and activation of pDCs was assessed in spleen 24 h postimmunization.

## Flow Cytometry

Analysis of surface antigens were performed with the following antibodies and markers: CD3 (145-2C11), CD4 (RM4-5), CD8 (53-6.7), CD44 (IM7), CD62L (MEL-14), PD-1 (29F1A12), Zombie Aqua, CD11b (M1/70), CD11c (N418), CD45R/B220 (RA3-6B2), CD317 (927), CD86 (GL-1), CD69 (H1.2F3), Sca-1 (E13-161.7) (Biolegend). Fc receptors were blocked using a purified anti CD16/32 antibody (2.4G2) (BioXcell). Identification of pDCs was based on their viability (Zombie Aqua- ) and their surface antigen expression (CD11cint, CD11blo, B220<sup>+</sup>, and CD317<sup>+</sup>). For intracellular staining, IFN-γ (XMG1.2), TNF-α (MP6-XT22), IL-2 (JES6-5H4) (Biolegend), IFN-α (RMMA-1) (PBL Assay Science, Piscataway, NJ, USA), IRAK1 (D51G7) as well as isotype control antibodies were used (Cell Signaling Technologies, Beverly, MA, USA) after permeabilization using the Intracellular Staining Permeabilization Wash Buffer 10× and Fixation Buffer following instructions of manufacturer (Biolegend). Data were acquired using BDLSR Fortessa Flow Cytometer (BD Biosciences) and analyzed using the FlowJo software (FlowJo, LLC).

## IFN-**α** Intracellular Staining

Mice were immunized with 400 µg of PapMV i.v. as described above. Spleens and bone marrows were collected 4 h postimmunization and processed as described above. Cells were then incubated with BFA for 4 h at 37°C followed by IFN-α intracellular staining as described above.

## Cell Sorting

Bone marrow-derived pDCs were stimulated for 24 or 48 h and stained for sorting. Fc receptors were blocked as previously described and pDCs were identified as CD11c<sup>+</sup>B220<sup>+</sup>PDCA1<sup>+</sup>. Cells were collected in FBS then washed twice with cold PBS followed by protein extraction (see below). Sort was performed using a BD FACSJazz (BD Biosciences).

## Immunoblotting

For immunoblotting, cells were harvested and lysed in Triton X-100 lysis buffer [20 mM Tris-HCl pH 8.0, 1% Triton X-100, 10% Glycerol, 150 mM NaCl, protease inhibitor cocktail (Roche)]. Lysates were then loaded on a 10% SDS-PAGE followed by transfer on a polyvinylidene difluoride membranes (BioRad, Mississauga, ON, Canada). Membranes were blocked in 5% dry milk TBS-T (TBS, 0.1% Tween-20) for 2 h at room temperature. Primary antibodies against mouse IRAK1 and β-actin (Cell Signaling Technologies) were diluted in TBS-T and incubated with the membranes o/n at 4°C. Antirabbit IgG HRP (Jackson ImmunoResearch, West Grove, PA, USA) were used as secondary antibodies whereby they were diluted in TBS-T and incubated with the membranes for 1 h at room temperature. Detection was performed with ECL chemiluminescence kit (BioRad).

#### ELISA and Multiplex

Interferon-α levels in mice serum or cell culture supernatants were quantified by ELISA according to the manufacturer's directions (Affymetrix eBiosciences). TNF-α, IL-6, IL-10, IL12p40, IL12p70, IL-9, IL-15, KC, G-CSF, M-CSF, RANTES, MIP-1α, MIP-1β, MIP-2, IP-10, and MCP-1 levels in mice serum were quantified using Milliplex Map Mouse Cytokine/Chemokine Premixed 32 Plex (Millipore, Etobicoke, Canada) according to manufacturer's directions. Measurement of median fluorescence intensity (MFI) was performed using Bio-plex (Biorad). PapMV-specific antibody titers were determined as described previously (25). Detection of PapMV-specific IgM was performed with peroxidase-conjugated goat antimouse IgM (Jackson Immunoresearch Laboratories).

#### Statistical Analysis

Statistical analysis was performed with GraphPad Prism Software (GraphPad Software). Error bars represent SEM. Two-tailed Student's *t*-test was used and Welch's correction was applied when needed.

## RESULTS

#### Treatment of Chronic LCMV Cl13 Infection with PapMV Does Not Improve Viral Clearance

The impact of IFN-α on viral infections in mice was shown to vary according to the kinetics and strength of production during the ongoing infection. Early IFN-α was shown to be essential to the control of the infection (7, 9, 10, 40) while long-term IFN-α was detrimental to the host and favored viral persistence (10–12). In one such study, treatment of LCMV Cl13-infected mice with exogenous IFN-α on days 3 and 5 postinfection resulted in the increase of GP33–41-specific CD8<sup>+</sup> T cells as well as a decrease in viral load 32 dpi (8). This led us to hypothesize that the treatment of a chronic infection such as LCMV Cl13 with PapMV could result in a similar clearance of the virus given the capacity of PapMV to induce potent IFN-α production following immunization (26, 31, 35). We therefore infected mice with LCMV Cl13 and treated them 3 and 5 dpi with 400 µg of PapMV i.v. Blood was collected at various time points to assess viral load, which was not affected by the treatments with PapMV (**Figure 1A**). To assess the efficiency of the PapMV treatment, serum was collected 6 h following the treatment on days 3 and 5 and IFN-α was quantified by ELISA. Although the administration of 400 µg of PapMV induced strong IFN-α production in naive mice, LCMV Cl13-infected mice barely produced IFN-α at all (**Figure 1B**) whether assessed after the first treatment on day 3 or the second treatment on day 5. These results indicate that the infection with LCMV Cl13 hinders IFN-α production following PapMV administration. Of note, PapMV not only induces IFN-α in immunized mice but also other cytokines and chemokines such as IL-6, IP-10, and MCP-1 (31, 35), which rely on different sets of signaling pathways than

Figure 1 | Treatment of an established lymphocytic choriomeningitis virus clone 13 (LCMV Cl13) infection with PapMV does not improve viral clearance. Mice were infected with LCMV Cl13 followed by treatments with PapMV on days 3 and 5 (A) Kinetics of viral burden expressed as LCMV Cl13 PFU/ml of blood (B) ELISA quantification of interferon (IFN)-α in serum of mice 6 h following treatment with control or PapMV. (C) Percentages of GP33-specific CD8+ T cells in the blood 8 dpi. (D) CD44, PD-1, and CD62L expression on CD8+ T cells in blood 8 dpi. (E) Percentages of GP33-specific CD8+ T cells in the spleen 30 dpi. (F) CD44, PD-1, and CD62L expression on CD8+ T cells in the spleen 30 dpi. (G) Percentages of CD8+ T cells producing IFN-γ, TNF-α, or IL-2 in response to a stimulation with the GP33–41 peptide for 5 h. (H) Viral loads of LCMV Cl13 in spleen, kidney, liver, and brain 30 dpi. Results are expressed as PFU/g of each organ. For (D,F), results are expressed as a ratio of the sample's mean fluorescence intensity (MFI) over the average MFI of control samples (\**p* < 0.05; \*\**p* < 0.01; \*\*\**p* < 0.001) (*n* = 2, three mice per group).

IFN-α for their production. It is therefore possible that immune cells were activated through these auxiliary cytokines following treatment with PapMV despite the absence of detectable IFN-α in the serum. To test this, we assessed activation of CD8<sup>+</sup> T cells as well as the proportion of GP33–41-specific CD8+ T cells. The proportion of GP33–41-specific CD8+ T cells in blood on day 8 postinfection was similar in groups treated with the control or PapMV (**Figure 1C**) and a similar trend was noticed for the expression of CD44 on CD8+ T cells (**Figure 1D**). We also assessed exhaustion of CD8<sup>+</sup> T cells by means of PD-1 expression, which was increased in mice treated with PapMV compared to mice treated with control (**Figure 1D**). Expression of CD62L on CD8<sup>+</sup> T cells was reduced in PapMV compared to mice treated with control (**Figure 1D**). Together, these data indicate that the treatment of LCMV Cl13-infected mice with PapMV on days 3 and 5 does not increase the activation of CD8<sup>+</sup> T cells. However, these treatments seem to increase the exhaustion of CD8<sup>+</sup> T cells as indicated by the increased expression of the inhibitory receptor PD-1 (41). PapMV treatment did not lead to homing of CD8+ T cells in lymphoid organs, as shown by their decrease in CD62L expression, an adhesion molecule involved in the homing of lymphocytes to secondary lymphoid organs. The percentage of GP33–41-specific CD8<sup>+</sup> T cells was not different on day 30 postinfection between control and PapMV-treated group (**Figure 1E**). As well, the expression of PD-1, CD44, and CD62L was not significantly different between these two groups (**Figure 1F**). Furthermore, our results revealed that the functionality of GP33–41-specific CD8<sup>+</sup> T cells was not affected by the treatment of LCMV Cl13 infection with PapMV, as shown by the similar percentage of IFN-γ, TNF-α, or IL-2 positive CD8<sup>+</sup> T cells in the spleen (**Figure 1G**) between control and PapMV-treated groups. To further confirm that the PapMV treatments had no effect on the clearance of LCMV Cl13, we collected lymphoid and non-lymphoid organs at 30 dpi and assessed the viral load of LCMV Cl13. In the spleen, kidney, liver and brain, both groups displayed similarly elevated LCMV Cl13 titers (**Figure 1H**), which, along with previous data, indicated that treatment with PapMV was ineffective in clearing an LCMV Cl13 infection. Notably, it has been shown that cells previously infected with chronic viruses such as hepatitis B virus (HBV) (42, 43) or hepatitis C virus (HCV) (44, 45) were unresponsive to further TLR ligand stimulation. Based on our findings, we surmise that a possible TLR tolerance mechanism might be at play after an LCMV Cl13 infection, therefore hindering further activation of immune cells by a subsequent TLR ligand stimulation such as PapMV.

#### Pretreatment with PapMV Inhibits Further Effects of Secondary PapMV Administration

Lymphocytic choriomeningitis virus clone 13 activates immune cells through the TLR7/MyD88 endosomal pathway (46) as well as the RIG-I/Mda5 cytosolic pathway (47). In order to exclude activation and cytokine production caused by engagement of the RIG-I/Mda5 pathway, we decided to move to a PapMV-only based model in which only TLR7 is stimulated. We therefore pretreated mice with PapMV at different time intervals ranging from 1 to 50 days prior to a second immunization with PapMV. Sera were collected 6 h following the second immunization to assess IFN-α production by ELISA. A single administration of PapMV induced strong IFN-α production, as detected in the serum 6 h postimmunization (**Figure 2A**). When mice were pretreated with 100 µg of PapMV, the production of IFN-α was abolished following a secondary immunization for up to 50 days following pretreatment (**Figure 2A**). A similar hindrance was observed in the production of TNF-α, IL-6, IL-12p40, IL-12p70, and IL-15, whereas production of IL-10 and IL-9 was enhanced and unaffected, respectively (Figure S1 in Supplementary Material). Similarly, production of various chemokines and growth factors was also suppressed following a second immunization as observed with the production of M-CSF, RANTES, MIP-1α, MIP-1β, IP-10, and MCP-1 whereas production of KC, G-CSF, and MIP-2 was enhanced or unaffected (Figure S1 in Supplementary Material). While there are other cytokines and chemokines that are differentially regulated upon PapMV administration, we focused on IFN-α production based on its wide-ranging use in therapeutic settings (14).

Since pDCs are major producers of IFN-α upon stimulation with a TLR7 ligand, we sought to determine whether the absence of IFN-α was due to a lack of activation of pDCs. Spleens of mice pretreated with 100 µg of PapMV were collected 24 h following the second immunization and activation of pDCs was assessed by

Figure 2 | Pretreatment with PapMV prevents plasmacytoid dendritic cells (pDCs) from responding to a subsequent PapMV administration. (A) ELISA quantification of interferon (IFN)-α in serum 6 h following immunization with PapMV once (PapMV 1×) or preceded by pretreatment with PapMV at various time intervals. (B) CD69 (left) and CD86 (right) expression on splenic pDCs 24 h following immunization with PapMV once (PapMV 1×) or preceded by pretreatment with PapMV at various time intervals. Results are expressed as a ratio of the sample's mean fluorescence intensity (MFI) over the average MFI of control samples (\**p* < 0.05; \*\**p* < 0.01; \*\*\**p* < 0.001) (*n* = 2–8, two to four mice per group).

flow cytometry. As seen with IFN-α production, a single immunization with PapMV induced upregulation of CD69 and CD86 (**Figure 2B**) on pDCs. Conversely, when mice were pretreated with PapMV 1–5 days prior to a second immunization, pDCs were unable to respond to the second immunization, as observed by the absence of CD69 or CD86 upregulation (**Figure 2B**). Here, we found that the expression of activation markers on pDCs on day 1 after the pretreatment was due to remnants of the initial immunization rather than the activation of pDCs following the second immunization (Figure S2 in Supplementary Material). With a 25-day or more lag between the pretreatment and the second immunization, pDCs were activated by the second immunization although to a lesser intensity than mice treated only once with PapMV (**Figure 2B**). Altogether, these results demonstrate that the administration of PapMV induces a refractory state in pDCs rendering them unable to respond to further PapMV immunizations. In short intervals, this effect is completely inhibitory while for longer intervals the inhibition is only partial, indicating that distinct mechanisms are possibly concomitantly interfering with the response.

## Refractory State Induced by PapMV Pretreatment Affects the Response to Other TLR7 and TLR4 Ligands but Not TLR3

To assess whether the refractory state induced by a pretreatment with PapMV affected only further administrations of PapMV, we pretreated mice with PapMV and then challenged them 5 days later with LPS, R837, or Poly I:C, which are ligands for TLR4, TLR7, and TLR3, respectively. We then assessed the expression of CD69 on pDCs (**Figure 3A**). As expected, mice treated with LPS, R837, or Poly I:C alone showed an upregulation of CD69 on pDCs, indicating activation. When mice were pretreated with PapMV, subsequent immunizations with LPS or R837 were not as efficient in inducing the activation of pDCs as immunizations in naive animals but still showed some degree of CD69 upregulation. These results indicate that the inhibition induced by pretreatment with PapMV not only impacts subsequent immunizations with PapMV but also other TLR7 ligands as well at other TLR pathways, such as TLR4, although the inhibition

Figure 3 | Interleukin-1 receptor-associated kinase 1 (IRAK1) is degraded by PapMV in bone marrow-derived plasmacytoid dendritic cells. (A) CD69 expression on splenic plasmacytoid dendritic cells (pDCs) 6 h following the last immunization with the control and R837 and 24 h following immunization with the control, lipopolysaccharide (LPS), or polyinosinic:polycyticylic acid (Poly I:C) once (−) or preceded with a pretreatment with PapMV (+) 5 days prior (*n* = 2–3, two to three mice per group, one representative experiment is shown for R837 samples). (B) CD69 (left) and CD86 (right) expression on bone marrow-derived BMpDCs 2–48 h poststimulation with control (black bars), Poly I:C (checkered bars), R837 (gray bars), or PapMV (white bars). (C) ELISA quantification of IFN-α in culture supernatant from BMpDCs stimulated 2–48 h with PapMV. (D) Representative overlay histograms of IRAK1 expression in BMpDCs following a stimulation of 24 h with control (filled graph), Poly I:C, R837, or PapMV (full line), or secondary antibody staining control (dashed line). (E) IRAK1 expression in BMpDCs 2–48 h following stimulation with the control (black bars), Poly I:C (checkered bars), R837 (gray bars), or PapMV (white bars). (F) Immunoblot of IRAK1 (top) and actin-β (bottom) of BMpDCs 24 h following stimulation with control, PapMV, R837, or Poly I:C. BMpDCs were sorted to isolate pDCs prior to protein extraction and immunoblotting. (A,B,E) Results are expressed as a ratio of the MFI of the sample on the average MFI of control samples. (A) (*n* = 2, two to three mice per group) (B,D,E) (representative experiment of two to ten experiments, two to three replicates per group) (C) (*n* = 2, 2 replicates per time point) (\**p* < 0.05; \*\**p* < 0.01; \*\*\**p* < 0.001).

is less pronounced. Strikingly, unlike the inhibition observed with LPS and R837, administration of PapMV prior to Poly I:C resulted in the expression of CD69, indicative of pDC activation, which was similar in the control and treated groups (**Figure 3A**). Altogether, these results point toward an inhibitory mechanism induced by PapMV pretreatment that affects TLR7 and TLR4 pathways while the TLR3 pathway remains unaffected.

## Stimulation with PapMV Induces Degradation of IRAK1 in pDCs

Toll-like receptor pathways are not specific to each receptor. Indeed, most of the complexes implicated in the signaling cascades are shared across pathways [reviewed in Ref. (48, 49)]. If PapMV were to affect the TLR7 signaling pathway, other TLR signaling pathways would also be affected, resulting in the cross-inhibition observed when cells are pretreated with PapMV. Indeed, such cross-inhibition was observed in response to stimulations with various TLR ligands such as TLR4 ligands (50–52) TLR7 ligands (52, 53) and TLR9 ligands (51, 53), establishing the presence of homo and hetero tolerance in the TLR signaling pathways. In these studies, one common mechanism reported to be involved in the observed cross tolerance was through the degradation of IRAK1 (52, 54, 55), a kinase implicated in all MyD88-dependant TLR signaling pathways, therefore excluding TLR3, which signals through a MyD88-independent pathway (56). Therefore, we assessed IRAK1 expression in pDCs by flow cytometry and immunoblotting to analyze the regulation of IRAK1 following PapMV stimulation. Given that pDCs account for only 0.2% of all splenocytes, we evaluated the response of pDCs to PapMV *in vitro* using BMpDCs. After differentiation of bone marrow cells with Flt3L for 8 days, cells were stimulated with PapMV. Similar to our observations from *in vivo* splenic pDCs, BMpDCs were readily activated by various TLR ligands including PapMV as shown by the upregulation of CD69, CD86 (**Figure 3B**) and the accumulation of IFN-α in culture supernatants following stimulation (**Figure 3C**). Since PapMV activates BMpDCs, we investigated the regulation of IRAK1 following stimulation with various TLR ligands. We first assessed IRAK1 expression by flow cytometry following various incubation periods of BMpDCs with TLR7 ligands, PapMV and R837, or TLR3 ligand, Poly I:C. As expected, R837 induced a strong downregulation of IRAK1 after 24 h in BMpDCs *in vitro* (**Figures 3D,E**). Surprisingly, stimulation of BMpDCs with Poly I:C induced a small downregulation of IRAK1 starting at 6 h poststimulation. Stimulation of BMpDCs with PapMV induced the degradation of IRAK1 albeit to a lesser extent than R837 and with slower kinetics (**Figures 3D,E**). To confirm the modulation of IRAK1 expression, we sorted pDCs 24 h poststimulation with PapMV, R837 and Poly I:C and extracted total cellular proteins to evaluate the expression of IRAK1 by immunoblotting. We found that 24 h post-R837-stimulation, the expression of IRAK1 was undetectable (**Figure 3F**) while stimulation with Poly I:C did not induce any degradation of IRAK1 (**Figure 3F**). IRAK1 was also degraded following stimulation of BMpDCs with PapMV although the extent of degradation was lower in comparison to that observed with R837 (**Figure 3F**). When comparing IRAK1 expression ratios obtained by western blot and flow cytometry (**Table 1**; **Figures 3E,F**), we noticed similar ratios between the two assays for R837 and PapMV stimulated BMpDCs while the ratios vary for Poly I:C stimulated BMpDCs. Taken together, these results indicate that PapMV induces the degradation of IRAK1 in pDCs, which could in part explain the tolerance observed when mice are pretreated with PapMV.

## PapMV Induces the Upregulation of Sca-1 on pDCs Despite Its Expression Not Being Associated with IFN-**α** Production

Heterogeneity in the pDC population has been described (57–60) although the biological significance of this phenomenon is still largely unknown. Various reports have, however, indicated that two subsets of pDCs expressing different sets of markers were differentially associated with IFN-α production following TLR stimulation (57–59, 61). Among these studies, it has been suggested that expression of Sca-1 could discriminate between IFN-α producing pDCs and those that do not: Sca-1<sup>+</sup> pDCs were weak producers of IFN-α while Sca-1<sup>−</sup> pDCs were strong producers (61). Thus, by evaluating the expression of Sca-1 in pDCs, we sought to determine whether it could explain the inhibition in IFN-α production observed with longer periods between PapMV treatments. As previously described (61), pDCs from the spleen are mostly Sca-1<sup>+</sup> (88.45% **Figure 4A**; Spleen; Naive) while pDCs from the bone marrow, although still in majority Sca-1<sup>+</sup> (69.90% **Figure 4A**; Bone marrow; Naive), display of a larger population of Sca-1<sup>−</sup> than in the spleen. After an immunization with PapMV, the proportion of pDCs expressing Sca-1 increased to close to 100% in both the spleen and the bone marrow (**Figure 4A**) and remained elevated for at least 5 days in both organs. The distribution of Sca-1 expression among pDCs returned to naive and control levels by 25 days post-PapMV immunization. To determine which of the Sca-1 expression profile was associated with IFN-α production, we performed IFN-α intracellular staining on pDCs from spleen and bone marrow of mice immunized with 400 µg of PapMV 4 h prior. In the spleen, most IFN-α+ pDCs were Sca-1<sup>+</sup> (**Figure 4B**). However, in the bone marrow, both Sca-1− and Sca-1<sup>+</sup> pDCs were expressing IFN-α after a PapMV immunization (**Figure 4B**). No significant difference was denoted between the two groups in the bone marrow. Niederquell et al. (61) proposed that Sca-1<sup>−</sup> pDCs could be precursors of Sca-1<sup>+</sup> pDCs. In this regard, it is possible that the proportion of Sca-1<sup>+</sup> IFN-α+ pDCs are in fact Sca-1<sup>−</sup> cells that are upregulating Sca-1 in response to stimulation. When intracellular staining for IFN-α was performed 2 and 6 h postimmunization to assess the progression of

Table 1 | Comparison of IRAK1 regulation ratios by flow cytometry and Western blot.


the IFN-α+ population with respect to Sca-1 expression in both spleen and bone marrow, no difference was observed between the three time points (data not shown). Altogether, these results indicate that although PapMV induces expression of Sca-1, IFN-α production does not seem to be associated with Sca-1 expression in this experimental setting.

## Antibodies Are Responsible for Long-term Attenuation of the Response of pDCs to PapMV Immunization

Our results suggest that inhibition of the response to multiple administrations of PapMV is induced through shared mechanisms between TLR-associated pathways for short periods (**Figure 3A**) and PapMV-specific components that affect IFN-α production (**Figure 2A**) as well as pDC activation (**Figure 2B**) for longer periods. We were therefore interested in assessing the role played by antibodies in the refractory state induced by PapMV pretreatment knowing that they were shown to affect responses to primeboost vaccine regimens in other systems (62, 63). We determined the antibody-mediated impact of PapMV pretreatment on further PapMV immunizations using JHT mice, which lack functional B cells and consequently also lack antibodies (64). As observed in C57Bl/6 mice (**Figure 2A**), a single immunization with PapMV in JHT mice induced the production of IFN-α (**Figure 5A**) although in a slightly more pronounced fashion. When a PapMV pretreatment was administered 5, 9, or 25 days prior to secondary PapMV immunizations, the production of IFN-α following the second immunization was equivalent to the response of naive animals, which differed significantly from results obtained in C57Bl/6 mice (**Figure 5A**). Since IFN-α production was not affected by the pretreatment, we examined the expression of CD69 and CD86 on pDCs after multitreatments with PapMV. Similarly to what was observed with IFN-α production, CD69 and CD86 (**Figure 5B**) expression on pDCs was not significantly different between naive mice and PapMV pretreated mice receiving a PapMV immunization. To confirm that the IFN-α production and pDC activation in JHT mice was due to the absence of antibodies and not of B cells, we performed transfer experiments with immune serum. Since two different profiles are observed at day 5 and day 25 in wild-type (WT) mice with respect to the inhibition generated by a PapMV pretreatment, we assessed the kinetics of IgM and IgG production in the serum of mice after a PapMV immunization (**Figure 5C**). As expected, both IgM and IgG specific for PapMV were found in the serum of WT mice 5 days postimmunization while 25 days postimmunization, only a high IgG titer was

(P). (B,E) Results are expressed as a ratio of the MFI of the sample of the average MFI of controls (*n* = 3–6, one to three mice per group) (\**p* < 0.05; \*\**p* < 0.01;

detected. Transfers were carried out with serum collected either 5 or 25 days after PapMV administration pooled from matched groups. The serum from the equivalent of two mice was injected into one recipient mouse, which was immunized with PapMV 24 h later (**Figure 5D**). IFN-α production was assessed in the serum 6 h postimmunization while pDC activation was assessed 24 h postimmunization. Injection of naive serum did not affect the production of IFN-α (**Figure 5E** compared to **Figure 2A**) nor did it affect the activation of pDCs as shown by the expression of CD69 and CD86 (**Figure 5F** compared to **Figure 2B**). However, when PapMV-immune serum taken at 5 or 25 days was administered to mice, the following immunization with PapMV showed no production of IFN-α (**Figure 5E**) and a significant decrease in the expression of CD69 and CD86 on pDCs (**Figure 5F**) although not as much as observed in WT mice. Altogether, these results indicate that PapMV-specific antibodies play a significant role in

the long-term inhibition and attenuation of the immune response following a pretreatment with PapMV.

## DISCUSSION

Interferon-α has long been a treatment of choice for chronic viral infections, whether used alone or in combination with other treatments such as ribavirin in the treatment of HCV. Due to its toxicity, the medical community is moving away from IFNα-based treatments toward alternatives bearing better adverse event profiles. One alternative toward this end would be to induce endogenous IFN-α production by the host instead of administering high doses of exogenous IFN-α. It is with this aim that we used PapMV nanoparticles, which contain a non-coding ssRNA molecule rendering it non-replicative. PapMV induces the production of IFN-α by pDCs without causing any adverse effects

\*\*\* *p* < 0.001).

when administered systemically and could therefore potentially accelerate the clearance of a persistent LCMV Cl13 infection. This approach would also be potentially applicable to other viral infections, as it would not require expression of virus-specific antigens. We observed that sequential application of PapMV treatment has limitations attributed to regulation of the TLR7 pathway as well as the presence of PapMV-specific antibodies upon the first immunization.

The treatment of an LCMV Cl13 infection with PapMV proved to be inefficient as no changes in viral loads or in LCMV-specific immune responses were observed following treatment (**Figure 1**). Like PapMV, LCMV Cl13 is also a TLR7 ligand [reviewed in Ref. (15, 16)]. It is therefore possible that the stimulation of TLR7 by LCMV Cl13 induces TLR tolerance similar to that observed when other TLR ligands are used as stimulators (50–53). Further stimulation of the TLR pathways would therefore be inefficient in LCMV-infected mice. Similarly, previous research has shown that stimulation of HBV-infected (42, 43) or HCV-infected (44, 45, 65) human cells with TLR ligands was unable to induce the production of cytokines and activate infected DCs.

Whereas we were unable to clear an LCMV Cl13 infection using PapMV administration, treatment of viral infections with exogenous IFN-α early in the course of the infection has been shown to be efficient in the control of LCMV (8), SIV (10), or RSV (40). Of note, in the successful treatment of an LCMV Cl13 infection with IFN-α, Wang et al. administered the IFN-α5 subtype. However, the IFN-α subtype profile elicited by PapMV has yet to be determined. Thus, the discrepancy observed could be due to a difference in the IFN-α subtype given the disparity in the immunomodulatory effects and antiviral capacities borne by different subtypes (66–68). Furthermore, contrary to direct IFN-α injection, treatment with PapMV requires uptake of the nanoparticle, release and degradation of the ssRNA inside the endosome before IFN-α can be produced following activation of the TLR7 signaling cascade (26, 28, 31). Although this sequence of events ensures specificity and safety, it is likely more susceptible to various regulatory mechanisms.

Lymphocytic choriomeningitis virus Cl13 stimulates immune cells not only through the TLR7/MyD88 pathway (15, 16) but also through the RIG-I/Mda5 pathway (47). In order to further study the mechanisms at play in this setting and isolate the TLR7 pathway from other variables of the LCMV infection, we pretreated mice with PapMV followed by a second immunization at various time points. This approach recapitulated the results observed in LCMV-infected mice with almost complete abrogation of IFN-α production and pDC activation following the secondary PapMV immunization for short time intervals between immunizations and significant impairment for longer intervals (**Figure 2**). A similar outcome was observed for the production of various cytokines and chemokines such as TNF-α, IL-6, IL-12p40, IL-12p70, IL-15, M-CSF, RANTES, and IP-10 while others were either not affected or enhanced by the secondary immunization (Figure S1 in Supplementary Material). This suggests that PapMV stimulates other pathways in addition to TLR7 leading to a broad activation of the immune system and that these pathways might be differently affected by multiple PapMV administrations. Nonetheless, the main outcome of multiple systemic administrations of PapMV, at least for the TLR7 pathway, is the suppression of the secondary response. This outcome was also observed in previous studies and is indicative of TLR tolerance (discussed below) (50, 52–54). We posit that the inability of subsequent PapMV immunizations to drive a robust response may also be dependent on the route of administration. This conclusion comes from our previous findings showing that sequential intranasal instillations could potentiate PapMV treatments (30). In the previous study, immunizations were separated by seven days and the last immunization led to a higher production of various cytokines in bronchoalveolar lavages. In an intratumoral injection model, we also observed that multiple administrations of PapMV led to decreased tumor growth when administered alone or in combination with other immunotherapies and sustained IFN-α following multiple administrations (37). Pretreatment with PapMV is therefore able to potentiate further PapMV administrations when delivered locally. Limitations are however observed when PapMV is administered systemically, as shown in this study. The development of immunization regimens alternating between various administration routes could therefore be an interesting alternative to mitigate the pitfall of sequential systemic treatments.

With regard to the tolerization of TLRs, we found that degradation of IRAK1 played a central role. This is in agreement with other studies illustrating that this kinase, which is shared across most of the TLR pathways (56), is degraded following TLR2, TLR4, TLR7 and TLR9 stimulation (52, 55, 69, 70). Although this degradation has been shown to last at least 48 h poststimulation (54, 55) both *in vivo* and *in vitro,* the length of this refractory period has yet to be determined. Here, we show that the stimulation of BMpDCs with PapMV induces a partial degradation of IRAK1, which was observed by flow cytometry and later confirmed by immunoblotting (**Figure 3**). Of note, we observed a stronger IRAK1 degradation with R837 than with PapMV, which might be due to the different nature of both TLR7 agonists. Indeed, R837 is a small synthetic molecule that does not require uptake to reach the endosome of cells. It is therefore easier and faster for this molecule to reach more cells and induce the degradation of IRAK1 in a more robust fashion. On the other hand, PapMV is a particulate molecule that has to be taken up by immune cells to reach the endosome thus elongating the interval between the stimulation and the apparent degradation of IRAK1 (71). It would also be of particular interest to verify the regulation of IRAK1 in pDCs *in vivo.* However, due to the low proportion of pDCs in the spleen, we were limited to conducting our analyses *in vitro* to determine the regulation of IRAK1 in BMpDCs. It is important to note that our results revealed only a partial role played by IRAK1 in the tolerance observed following multiple administrations of PapMV. Indeed, stimulation of BMpDCs with PapMV did not induce complete degradation of IRAK1, suggesting that there could be residual proteins left in the cells capable of proceeding through the signaling cascade when further encountering PapMV. This led us to investigate other potential inhibitory mechanisms.

Niederquell et al. suggested that expression of Sca-1 could discriminate between subsets of pDCs able or not to produce IFN-α in response to TLR stimulation (53). We therefore hypothesized that sequential PapMV administrations could preferentially stimulate or expand pDC subsets unable to produce IFN-α, which might explain the abrogation of IFN-α production upon secondary immunizations. However, in our system, Sca-1 expression on pDCs was not associated with the capacity to produce IFN-α in response to PapMV. While we used a particulate molecule, Niederquell et al. used CpG ODN as a TLR9 ligand. Given that TLR7 and TLR9 are not stimulated by the same ligands (RNA vs. DNA) and similar to R837, CpG ODN is a small synthetic molecule, the kinetics of activation are therefore different in both models. In this regard, other markers such as Ly49Q (57), CD123 (72), and CD9 (59) that have also been associated with IFN-α production might be more informative.

Administration of plant virus-like particles in mice leads to the rapid production of specific antibodies [reviewed in Ref. (73)]. Since antibodies are generated following PapMV injection, we were interested in assessing their impact on multiple administration regimens. We showed by immunizing JHT mice, which are devoid of B cells and antibodies, as well as performing serum transfer, that PapMV-specific antibodies generated after the first administration were largely responsible for the tolerance to a second PapMV injection. In the short-term immunization regimen (5 days), there was retained inhibition of pDC activation in JHT mice as shown by the slightly diminished expression of CD69 following a second immunization relative to the naive group (**Figure 5B**). When transferring PapMV immune serum from day 5 postimmunization into naive C57Bl/6 mice followed by PapMV immunization, we observed lower expression levels of CD69 and CD86 compared to mice receiving a single PapMV administration (**Figure 5F** compared to **Figure 2B**). This is either due to an underlying mechanism independent of antibodies or due to the titer of antibody transferred. Although the titer of PapMV specific antibodies found in mice receiving serum transfers is lower than what is found in PapMV immunized mice (Figure S3 in Supplementary Material), this was enough to interfere with the subsequent PapMV immunization by inhibiting the production of IFN-α on both days 5 and 25. To overcome this limitation, one could administer immunogenically distinct plant virus-like particles to circumvent the effect of antibodies. It would also be interesting to explore the use of different injection routes and whether or not antibodies can also interfere with the response to subsequent injections.

Other directions are currently being assessed to potentiate systemically administered PapMV. We have previously shown that our platform could be modified in order to display various epitopes on the surface of the nanoparticle (25, 27, 28, 33, 74–76). These engineered particles showed immunostimulatory properties similar and sometimes better than the original platform following immunization (25, 27, 28, 33, 74, 76). A new strategy that we are currently investigating is the use of a sortase-mediated antigen coupling technique, which permits the fusion of epitopes directly on the surface of PapMV without the need to genetically modify the sequence of the coat protein (36). Immunizations with such PapMV-fused platforms lead to the development of protective humoral responses (36). Different immunization regimen as well as different chronic viral infection models could here be tested to evaluate the potential of PapMV in treating other diseases. The results obtained during this study open the way to study other potential uses for PapMV such as in the treatment of autoimmune diseases. It was previously shown that in the absence of IFN-α, whether in IFNAR-deficient mice or through the use of IFNAR antibody blockade, autoimmune symptoms of lupus prone mice were improved (77, 78). Multiple systemic administrations of PapMV induced an inhibition of IFN-α production providing a potential therapeutic approach for such an application.

In this study, we showed that treatment of a chronic virus infection with PapMV has limitations and still needs to be improved. Although a single administration of PapMV induces strong immune responses, recall systemic immunizations are much less potent partly due to IRAK1 degradation but mainly to interference by PapMV-specific antibodies. Our results also demonstrate that the PapMV platform is able to induce an immune response following a pretreatment although not yet to a degree that would be able to clear an ongoing viral infection. Although this outcome is not favorable in the context of chronic viral infections, it could be of interest for other diseases such as autoimmunity. Further improvements will therefore be needed for this promising therapeutic approach to be used in the treatment of chronic viral infections or other IFN-dependent chronic diseases.

#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Canadian Council on Animal Care guidelines. The protocol was approved by the Institut national de la recherche scientifique Animal Care Committee.

#### AUTHOR CONTRIBUTIONS

KC, M-ÈL, and AL conceived and designed the experiments; KC, M-ÈL, and ET performed the experiments; PS developed the PapMV manufacturing process; KC, AL, and DL cowrote the article.

## ACKNOWLEDGMENTS

We thank Dr. Armstrong Murira and Dr. Albert Parisien for critically reviewing the manuscript.

## FUNDING

This work was financially supported by Canadian Institutes of Health Research Grant MOP-89833 and the Jeanne and J.-Louis Lévesque Chair in Immunovirology from the J.-Louis Lévesque Foundation. KC and M-ÈL acknowledge studentship support from the Fonds de Recherche du Québec – Santé.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at http://www.frontiersin.org/articles/10.3389/fimmu.2017.01885/ full#supplementary-material.

Figure S1 | Multiplex quantification of cytokines and chemokines in serum 6 h following the last immunization with PapMV. Immunizations were performed 7 or 14 days following a first immunization with PapMV (*n* = 1, five mice per group).

Figure S2 | CD69 expression kinetics on plasmacytoid dendritic cells after PapMV immunization. Results are presented as a ratio of the MFI of the sample

## REFERENCES


over the average mean fluorescence intensity of controls (*n* = 1–7, one to three mice per group).

Figure S3 | ELISA quantification of PapMV-specific IgG in serum of mice transferred with immune sera 6 h following an immunization with control or PapMV. Immune sera were collected 5 and 25 days following PapMV immunization (n = 2, two to three mice per group).


peptide leads to an effective one dose influenza vaccine. *Vaccine* (2015) 33(51): 7245–53. doi:10.1016/j.vaccine.2015.10.123


receptor 7. *Proc Natl Acad Sci U S A* (2009) 106(8):2764–9. doi:10.1073/ pnas.0813037106


78. Baccala R, Gonzalez-Quintial R, Schreiber RD, Lawson BR, Kono DH, Theofilopoulos AN. Anti-IFN-alpha/beta receptor antibody treatment ameliorates disease in lupus-predisposed mice. *J Immunol* (2012) 189(12):5976–84. doi:10.4049/jimmunol.1201477

**Conflict of Interest Statement:** DL is the founder and a shareholder of Folia Biotech. Inc., a Canadian Biotechnology Company with the mandate to commercialize the PapMV Technology. The other authors have no financial conflicts of interest.

*Copyright © 2018 Chartrand, Lebel, Tarrab, Savard, Leclerc and Lamarre. 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) or licensor 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 Multirole of Liposomes in Therapy and Prevention of infectious Diseases

*Roberto Nisini1 , Noemi Poerio2 , Sabrina Mariotti1 , Federica De Santis2 and Maurizio Fraziano2 \**

*1Dipartimento di Malattie Infettive, Istituto Superiore di Sanità, Rome, Italy, 2Dipartimento di Biologia, Università degli Studi di Roma "Tor Vergata", Rome, Italy*

Liposomes are closed bilayer structures spontaneously formed by hydrated phospholipids that are widely used as efficient delivery systems for drugs or antigens, due to their capability to encapsulate bioactive hydrophilic, amphipathic, and lipophilic molecules into inner water phase or within lipid leaflets. The efficacy of liposomes as drug or antigen carriers has been improved in the last years to ameliorate pharmacokinetics and capacity to release their cargo in selected target organs or cells. Moreover, different formulations and variations in liposome composition have been often proposed to include immunostimulatory molecules, ligands for specific receptors, or stimuli responsive compounds. Intriguingly, independent research has unveiled the capacity of several phospholipids to play critical roles as intracellular messengers in modulating both innate and adaptive immune responses through various mechanisms, including (i) activation of different antimicrobial enzymatic pathways, (ii) driving the fusion–fission events between endosomes with direct consequences to phagosome maturation and/or to antigen presentation pathway, and (iii) modulation of the inflammatory response. These features can be exploited by including selected bioactive phospholipids in the bilayer scaffold of liposomes. This would represent an important step forward since drug or antigen carrying liposomes could be engineered to simultaneously activate different signal transduction pathways and target specific cells or tissues to induce antigen-specific T and/or B cell response. This lipid-based host-directed strategy can provide a focused antimicrobial innate and adaptive immune response against specific pathogens and offer a novel prophylactic or therapeutic option against chronic, recurrent, or drug-resistant infections.

Keywords: liposome, infectious disease, therapy, immunotherapy, drug, vaccine, adjuvant, immunomodulation

## INTRODUCTION

Liposomes are small artificial spherical vesicles constituted by one or more phospholipid bilayers with the polar groups of phospholipids oriented to the inner and outer aqueous phase. Such a structure explains the high propensity of liposomes to be encapsulated with hydrophilic, hydrophobic, and amphiphilic drugs within the inner aqueous compartment, the lipid bilayers, and at their interfaces, respectively (1). Liposomes can be classified as unilamellar (ULV) or multilamellar (MLV) vesicles (2), depending on the structure of the bilayer (**Figure 1**). ULV are characterized by the presence of a

#### *Edited by:*

*Rajko Reljic, St George's, University of London, United Kingdom*

#### *Reviewed by:*

*Arun Kumar, Linköping University, Sweden Giampiero Pietrocola, University of Pavia, Italy*

> *\*Correspondence: Maurizio Fraziano fraziano@bio.uniroma2.it*

#### *Specialty section:*

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

*Received: 28 November 2017 Accepted: 17 January 2018 Published: 05 February 2018*

#### *Citation:*

*Nisini R, Poerio N, Mariotti S, De Santis F and Fraziano M (2018) The Multirole of Liposomes in Therapy and Prevention of Infectious Diseases. Front. Immunol. 9:155. doi: 10.3389/fimmu.2018.00155*

**47**

Figure 1 | Schematic representation of liposomes showing the different antigen encapsulation propensity: unilamellar (A) and multilamellar (B) vesicles.

single-lipid bilayer of 20–250 nm diameter, which enclose a large aqueous core and are ideally suited for the encapsulation of hydrophilic drugs/antigens (2). Conversely, MLV are characterized by the presence of two or more concentric lipid bilayers of 1–5 µm diameter that preferentially entrap lipid soluble molecules (2). Liposomes are extensively used as carriers for active molecules in cosmetic and pharmaceutical industries as well as in food and farm industries, where liposomes encapsulation has also been studied to entrap unstable substances such as antioxidants, flavors, and antimicrobials. The phospholipid protective shield forms a barrier, which is normally resistant to the action of enzymes, pH, and free radicals within the body, thus protecting the cargo from degradation until the release occurs at the target cell, organ, or system. Because of the high biocompatibility, biodegradability, low toxicity, and capability to encapsulate hydrophilic and hydrophobic compounds (2), liposomes represent the most successful drug carrier system known to date. Many liposome formulations for the treatment of cancer, fungal, and viral infection as well as pain management are available in clinical use, and many other formulations are being tested in different phases of clinical trials (3). From their first description in 1960s, many liposomes have been produced with distinctive characteristics, which strictly depend on the nature of lipid components, on their possible chemical modifications, and on their surface charge. In particular, early conventional liposomes were mainly constituted by natural phospholipids such as phosphatidylcholine (PC), sphyngomielin, and monosyaloganglioside (4). However, this formulation was subjected to several critical issues, such as the instability in plasma and short blood circulation half-life, due to their interaction with high- and low-density lipoproteins that resulted in the rapid release of the encapsulated drug into the plasma, and the efficient liposome uptake and removal from circulation by the macrophage system. Moreover, in most cases, negatively charged liposomes have a shorter half-life in the blood than neutral liposomes, and positively charged liposomes are toxic and quickly removed from circulation (5). The addition of cholesterol (Chol) in liposome formulation reduced the permeability of lipid bilayer, increased the liposome stability, and reduced the rapid release of encapsulated bioactive compound (6). The size was also shown to be a crucial determinant for circulation half-life of liposomes, as their elimination by phagocytes is directly correlated to the liposome diameter: MLV with diameters ranging from 500 to 5,000 nm are quickly removed by phagocytes, whereas ULV with diameters between 20 and 50 nm show the less propensity to be internalized by macrophages (5), but are characterized by a low volume available for drug entrapment.

Improvement of liposome technology involved the generation of stealth liposome, targeted liposomes, immunoliposomes, and stimuli responsive liposomes (**Figure 2**). The stealth strategy is based on the possibility to coat the external liposome surface with biocompatible hydrophilic polymer conjugates, such as polyethylene glycol (PEG), chitosan, and others, which reduces immunogenicity and macrophage uptake. Stealth liposome technology makes liposome capable to escape phagocytosis, to reduce toxicity due to the presence of charged lipids, and to increase blood circulation half-life. However, an important limitation of stealth liposomes is their large biodistribution in tissues, as encapsulated bioactive molecules cannot be selectively delivered to target cells (7). Targeted liposomes where therefore designed to counterbalance the large body distribution of stealth liposomes. Targeted liposomes are characterized by the presence of membranes functionalized with glycoproteins, polysaccharides, or ligands for specific receptors that determine the preferential accumulation of liposomes in selected tissues, so that the liposome drug/antigen cargo can be preferentially released in predetermined target cells or organs (7). A further strategy to deliver entrapped drug/antigen in the desired tissue/cell is represented by immunoliposomes, which are functionalized by antibodies or antibody fragments (8), and stimuli responsive liposomes. Examples of stimuli responsive liposomes are the pH-sensitive liposomes (9), which undergo conformational and chemical changes in response to acid pH, and temperature sensitive liposomes (10), which keep their cargo encapsulated at body temperature, but discharge it upon local heating.

In addition to the delivery of drugs, liposomes may also be used for many other purposes through the modification of their composition and cargo. In vaccinology, for example, liposomes can be formulated with the inclusion of antigens (lipids, nucleic acids, proteins, and peptides), and/or with pathogen-associated molecular patterns (PAMPs), which confer adjuvant properties aimed at modulating the inflammatory microenvironment where T lymphocyte priming occurs (11, 12). At the moment, a number of liposome formulations are in clinical trials as adjuvant for prophylactic as well as therapeutic vaccines against malaria, influenza, tuberculosis (TB), human immunodeficiency virus

functionalized with hydrophilic polymeric conjugates, such as polyethylene glycol (PEG), to avoid elimination by reticuloendothelial system; (C) targeted liposomes with membranes functionalized with glycoproteins, polysaccharides, or ligands of cell/tissue-specific receptors; (D) immunoliposomes, a specific form of targeted liposomes, with membrane functionalized with antibodies or antibody fragments; (E) stimuli responsive (pH or temperature) liposomes, which release the encapsulate drug following exposure to acid pH or heating.

(HIV), and dengue fever (13), whereas Cervarix®, Inflexal®, and Epaxal® are commercially available liposome vaccines against infection by human papilloma virus (HPV), influenza virus, and hepatitis A virus, respectively. Finally, lipids are important second messengers in the regulation of molecular pathways associated with phagosome maturation and activation of antimicrobial responses (14). Thus, the possibility exists to deliver selected lipid second messengers into infected cells through targeted liposomes to activate antimicrobial molecular pathways against specific bacterial pathogens (15).

This review summarizes the recent developments in liposome technology aimed at the generation of novel prophylactic as well as therapeutic tools to control infectious diseases.

## LIPOSOMES AS CARRIER FOR DRUGS

The fate of drugs administered to a living organism varies according to several variables, including distribution, metabolism, and excretion. After a drug enters the systemic circulation, it is not distributed uniformly to the body's tissues due to differences in organ blood perfusion, tissue affinity, local pH, and permeability of cell membrane. Thus, a significant amount of a given drug does not reach the target cell or organ. The consequence of the uneven distribution is a reduced concentration of drug reaching the target cells and the interaction of the drug with bystander targets. Increasing the dose of the drug helps reaching the therapeutic concentration at the target level and also increases the effects of the drug on other cells, concurring to amplify side or adverse events of drug administration. The ratio between the blood concentration at which the drug is effective and the concentration at which the drug becomes toxic is defined therapeutic index (TI): the larger is the TI, the safer is the drug. It has to be further stressed that the ultimate target of drugs is generally an intracellular reaction site and the interaction of drugs with the cellular/bacterial lipid membrane is expected to occur at some point, affecting, sometimes severely, drug bioavailability, and efficacy (16). In the absence of specific mechanisms of intracellular transportation, the chemical nature of the drug dictates the level of entrance into cells. An interesting example is given by the different intracellular accumulation rates observed for four structurally similar quinolones: ciprofloxacin, levofloxacin, garenoxacin, and moxifloxacin (17). Quinolones were shown to exert a condensing effect on the membrane bilayer and the condensing effect differed among the quinolones: even small differences in their chemical structure were shown to influence their lipophilicity, the consequent capacity to cross cell membranes and to accumulate intracellularly, and the eventual antimicrobial effect (18). Thus, the search of innovative approaches to increase drug transport across the cell membrane or to improve the permeability of small molecule, peptide, and/or protein–drugs to access the cytoplasm is a hot field in pharmacology (19).

Within a few years after the first description of "swollen phospholipid systems" by Bangham and colleagues (20), various enclosed phospholipid structures consisting of single bilayers, initially termed "banghaosomes" and then "liposomes" (21), were described (22). The observation that liposomes can entrap drugs made these nanoparticles an interesting tool exploitable as new drug delivery systems (23, 24). The observation that the anticancer drug cytosine arabinoside was successfully delivered by liposomes to tumor cells with a significant increase in the survival times of mice affected by L1210 leukemia (25) was one of the first demonstrations of the suitability of liposomes as drug delivery system. Murine L1210 leukemia became "model system" for testing the therapeutic outcomes of anticancer drugs delivered by a large variety of liposome with different characteristics (26). Another advantage that became evident using liposome-entrapped drugs was the reduction of drug toxicity: drugs entrapped in liposomes are not bioavailable until they are released in target organs. Amphotericin B (AmB) and doxorubicin entrapped in liposomes (26) were extensively studied and the pioneering *in vitro* and *in vivo* studies ultimately led to the first clinical trials of liposomal drugs (27). However, the first tested "large ULV" for drug delivery showed their limits and many efforts have been made to design liposomes able to target specific cells and favor drug release for an optimized release rate of entrapped drug from the liposomes. For the optimal therapeutic activity, drugs must be released from liposomes to the disease site, where they should become bioavailable at a concentration within their therapeutic window for a sufficient time (26). One of the first obstacles that appeared clear when using large ULV was the rapid clearance of liposomes operated by the phagocyte system in the liver and spleen. Several approaches were tested, ranging from variation in liposome dimension to treatment of liposomes with serum albumin or variation in bilayer composition and Chol content (26). A significant step forward in the field of drug delivered by liposomes was made thanks to the observation that attaching PEG to proteins causes an increase of their circulation half-life (28). When the procedure was applied to the liposome system, it was evident that grafting of PEG to the liposome surface resulted in diminished clearance of liposomes by macrophages (29) and, the PEG-liposomes (stealth liposomes), unlike previously tested ULV, were shown to have dose-independent pharmacokinetics (30). The improvements in the therapeutic outcomes of stealth liposomes in comparison with conventional ULV for a variety of therapeutics was demonstrated in different animal models of disease (26), and then in the humans, where the long circulation half-life of a doxorubicin entrapped in PEG–liposomes was confirmed. Shortly thereafter, doxorubicin entrapped in stealth liposomes was used in the first clinical trial for the treatment of Kaposi's sarcoma in HIV patients (31). The persistence of stealth liposomes in the circulation facilitated their accumulation in highly vascularized sites, such as tumors or inflammatory sites. However, it appeared clear that both the efficacy and the TI would increase if drug-bearing liposomes could be selectively addressed to target cells.

Active targeting could be accomplished by coupling targeting ligands to the surface of liposomes, such as proteins, peptides, carbohydrates, or monoclonal antibodies or their fragments (fragment antigen-binding and single-chain variable fragment) (32). Thus, targeted liposomes can be selectively taken up by cells that overexpress the receptor for the moiety, so improving therapeutic outcomes by increasing efficacy and reducing off-target toxicity (32). Among the different moieties that can be covalently or non-covalently attached to the liposome surface, antibodies and antibody fragments are the most widely employed, producing immunoliposomes (33). Spragg and coworkers demonstrated that E-selectin-targeted immunoliposomes for doxorubicin delivery mediated marked cytotoxicity when incubated with activated human umbilical vein endothelial cells (HUVECs) that express E-selectin, but not when incubated with non-activated HUVECs (34). Following these promising results, methods for antibody coupling to liposomes have been developed (33) and different antibodies or fragments have been attached to liposomes, particularly through reactions with maleimide. Examples include immunoliposomes targeted to soluble *Leishmania* antigens, EFGR for glioma, endoglin (CD105), fibroblast activation protein, and HER2 for breast cancer, among others (35–38).

In addition to the possible systemic use, i.e., intravenous administration (39), liposomes have been designed for aerosol administration (40) or intradermal administration due to their lipophilic properties suitable for skin penetration (41). The functionalization of liposomes makes it possible to design drug delivery systems suitable for the treatment of different diseases according to their pathogenesis and localization. Among other diseases, liposomes have been primarily and deeply studied for cancer treatment. The narrow TI of antitumor drugs, i.e., the narrow window between their effective doses and those at which they produce adverse toxic effects, makes liposomes as ideal carriers for this type of drugs (42). Indeed, the majority of liposome-delivered drugs available for clinical use (**Table 1**) and in advanced phase III studies to date belongs to this category (3).

The property of encapsulation of drugs in liposomes to enhance the TI of various agents was the reason for the first use of anti-infectious drugs in severely ill patients with renal failure. To date, the only drug delivered by liposomes approved for clinical use in the field of infectious disease is AmB, but many studies are ongoing aimed at taking advantage of this versatile platform to treat patients with a variety of different infectious diseases. Three formulations of AmB for parenteral use were made available differing for liposome formulation and indications. Abelcet® was developed in 1995 for the treatment of invasive fungal infections refractory to conventional AmB desoxycholate therapy or when renal impairment or unacceptable toxicity precludes the use of conventional AmB (3). Abelcet® has a 1:1 drug:lipid molar ratio with a drug concentration ranging from 25 to about 50% molar (49). Ambisome® was approved in 1997 for the treatment of severe fungal infections including leishmaniasis, aspergillosis, blastomycosis, coccidioidomycosis in febrile, neutropenic patients, and a severe form of meningitis in individuals infected with HIV (3). Ambisome® is also indicated for the treatment of invasive systemic infections caused by *Aspergillus* sp., *Candida* sp., or *Cryptococcus* sp. in renal impaired patients or in patients who cannot tolerate therapy with free AmB (50). Amphotec® is a freeze-dried lipid-based formulation of AmB. The sodium salt of cholesteryl sulfate forms a thermodynamically stable colloidal complex with AmB at a 1:1 drug-to-lipid molar ratio (3).

In the field of parasitic infections, several studies were performed to test the efficacy of liposome-encapsulated drugs. In the treatment of leishmaniosis, an interesting approach is based on the use of dinitroanilines, because of their specific binding to parasite but not human tubulins. However, their low water solubility and instability have blocked their development as antiparasitic agents (51). Encapsulation of drug by dimyristoyl PC- and dimyristoyl phosphatidylglycerol-based liposomes overcome those limitations and allowed to efficiently deliver the drugs, reaching a therapeutic advantage as demonstrated in animal models (51). Another drug under test for liposome delivery is buparvaquone, an extensively studied drug for the treatment of visceral leishmaniosis. da Costa-Silva et al. reported that buparvaquone, which has an immunomodulatory effect in host cells, when entrapped in phosphatidylserine (PS) exposingliposomes, as a delivery approach to macrophage, is highly effective at low doses at eliminating *Leishmania infantum* parasites in a hamster model (52).

In the fight against malaria, the majority of drugs under development are lipophilic and characterized by poor plasma solubility and large biodistribution volumes with low accumulation in red blood cells (RBC). As a consequence, these drugs have shown limited therapeutic activity against intra-erythrocyte *plasmodia* (53). Rajendran et al. developed lipid formulations of soy-PC Chol containing either stearylamine (SA) or phosphatidic acid (PA) and different densities of distearoyl phosphatidylethanolamine-methoxy-PEG 2000 as a delivery system to test the antimalarial activity of monensin. Monensin entrapped in such liposome formulations was tested both in *in vitro* systems of *Plasmodium falciparum* cultures and in *in vivo* murine models of *Plasmodium berghei* infection and found to be more effective than free monensin given at comparable doses (54). The trans platinum–chloroquine diphosphate dichloride was recently successfully tested after its encapsulation into PEGylated neutral and cationic liposomes to fight parasites resistant to traditional drugs and proposed as a new therapeutic tool against malaria (55). Preliminary assays conducted in 1987 using passively loaded chloroquine into RBC-targeted immunoliposomes resulted in significant *P. berghei* growth inhibition in mice, when compared with administration of the free compound (56). More recently, the immunoliposome strategy against malaria was tested taking advantage of a monoclonal antibody specific for a 52-kDa RBC surface antigen characteristic of the murine erythroid lineage


and specifically expressed from the early proerythroblast to mature RBC stages. These immunoliposomes loaded with two novel antimalarial lipophilic drug candidates in the mouse model of *Plasmodium yoelii* increased the TI and efficacy of the used drug (53). Liposomes exposing PS-specific peptide may represent a further strategy to target *plasmodia* infected RBC in apoptosis (eryptosis:erythrocyte apoptosis) that have been tested in malaria treatment, with promising results (57). Experimental malaria may also represent a model to study the combined effect of anti-infective drugs and anti-inflammatory compounds aimed to limit immunopathogenic reactions in critical anatomic sites, such as the brain. Guo et al. tested liposome-encapsulated β-methasone hemisuccinate in a murine model of experimental cerebral malaria and found that this antiinflammatory liposome-delivered drug prolonged the survival time of infected animals, permitting the administration of antimalarial drug before the development of severe anemia (58). In addition to malaria, liposomes, as part of nanopharmaceuticals, are promising tools under investigation for the treatment of several other parasitic diseases including schistosomiasis (59, 60) and acanthamoebiasis (61), which are grouped under the term of neglected diseases (62).

In 2016, it was estimated that there were nine million new TB cases in the world and around half a million of them were caused by multidrug-resistant (MDR) *Mycobacterium tuberculosis* (Mtb) strains (Global tuberculosis Report 2017. http://www. who.int/tb/publications/global\_report/en/ Health Organization, Geneva, Switzerland). The research is focused in developing new antimicrobials, but the need to reach a sufficiently high drug concentration inside the lung macrophages and the appearance of MDR and extensively drug-resistant (XDR) Mtb strains represent two major obstacles to reach this goal (63, 64). A possible strategy to the development of new anti-TB therapeutics is based on the administration of drugs contained in specific nanodevices (62), which, preferentially targeting macrophages, allows the use of lower drug dosage and the reduction of undesirable side effects as well as the limitation of Mtb resistance mechanisms. In this line, an interesting contribution to the development of innovative anti-TB therapeutic strategies is the inhalation therapy with liposomes carrying anti-TB drugs. The inhaled drug is expected to be rather effective in the overt presence of bacteria as in smear-positive cases of TB in which the bronchial tree may be directly connected with the cavitary lesions, where Mtb rapidly multiplies. Liposomes delivering anti-TB drugs may offer several advantages over dry powder inhalable formulations of anti-TB drug (65, 66) or other aerosol administrable nanoparticles (67), such as the increase of drug half-life and the possibility to target the intracellular pathogen (68). However, difficulties may arise to efficiently entrap some of the drugs required for TB treatment into liposomes. Justo and Moraes showed that passive encapsulation of isoniazid and pyrazinamide in liposomes during dry lipid film hydration at initial drug-to-lipid molar ratios of 13.3 can be achieved with efficiencies of 2.5 and of 2.2%, respectively, equivalent to final drug-to-lipid molar ratios of 0.33 and 0.29 (69). Moreover, they were also able to incorporate ethionamide during the dry lipid film production step with a 42% trapping efficiency, equivalent to a low final drug-to-lipid molar ratio of 0.04 (69). Justo and Moraes were unable to efficiently incorporate rifampicin and streptomycin in the vesicles under the conditions tested. In contrast, Zaru et al. succeeded in entrap rifampicin in liposomes made of phospholipon 90®, soy lecithin, and Chol using the film hydration method followed by procedures of freeze dry (70). A deeper deposition of rifampicin entrapped in these liposomes in comparison with rifampicin suspension was found in lung airways of rats after nebulization of a nose-only exposure chamber, encouraging the use of liposomes in aerosol therapy of TB (71). An additional aspect to be faced is the stability of liposome-entrapped drugs. Stability may be enhanced by the use of pro-liposomes, i.e., free flowing granular products composed of drug and phospholipid precursors, which on hydration lead to liposomes (72, 73) that have a higher stability upon reconstitution. A new single-step and fast spray drying technique for pro-liposome powder preparation was developed (74, 75) using a systematic method known as "Quality by Design" (71): a systematic approach to develop drug products that includes evaluation of formulation parameters to achieve defined final product quality (76). The method was used to prepare rifapentine-loaded pro-liposomes, which resulted in an average size of 578 nm and an efficiency of encapsulation around 70% (71). The proliposome with a drug/hydrogenated soy-PC ratio of 1:2 and SA as a charge-inducing agent (74, 75) are promising formulations and represent possible new tools to approach therapy of infectious diseases based on the optimization of the delivery of old drugs. However, liposomes may be also considered for the delivery of new developed drugs to reduce or delay the appearance of drug resistance.

Liposome-entrapped drugs or immunomodulating agents are being developed for the treatment of many other diseases (see some examples in **Table 2**). However, clinical trials based on innovative liposome strategies should take into account also possible side effects that these drug formulations may have, particularly when given intravenously, and the fulfillment of several regulatory landscape. Both FDA and EMA have developed specific guidelines to be fulfilled in clinical trials with reference to future novel liposomal products (77, 78).


## LIPOSOMES AS "DIRECT" ENHANCERS OF INNATE ANTIMICROBIAL IMMUNE RESPONSE

For long time, the function of lipids was believed to be limited to their role as structural components of membranes or principal form of energy storage in cells. However, during the past 30 years, their important role in cell physiology has been disclosed and lipids have been identified as fundamental players in many cellular functions, including, among others, the regulation of signaling, organization of membranes, intracellular vesicle trafficking, and phagocytosis. Particularly interesting is the role of lipids in the physiology and pathology of phagocytosis, which represents a critical step in the effector function of innate immunity. In fact, the phagosome formation and maturation are functions involving coordinated signaling, targeting, and trafficking events largely regulated by lipid moieties (91, 92). In this process, several lipid species accumulate in membrane microdomains, where they associate in signaling transduction platforms (93). In addition, several lipids can (i) promote positive (convex) or negative (concave) membrane curvature, (ii) recruit signal proteins by binding to specific lipid-binding domains (**Table 3**), and (iii) confer an electrostatic potential on membrane to attract opposite charged key signaling and effector proteins (91–93). Thus, intracellular vesicle trafficking is finely regulated by a topologically and temporally controlled lipid expression whose main role, in the case of phagocytosis, is to promote the fusion/fission events, which starting from phagosomes leads to the generation of a bactericidal phagolysosome (PL) (14). Phagocytosis starts with particle recognition by phagocytic receptors and proceeds with phagocytic cup formation, phagosome sealing (PSL), and internalization (**Figure 3A**). In particular, one of the first events occurring during phagocytic receptor signaling is a local rapid accumulation of phosphatidylinositol (PI) 4,5-bis phosphate [PI(4,5)P2] (94), mediated by PI 4-phosphate 5-kinases (PIP5Ks), whose activity both regulates phospholipase D (PLD) and is regulated by PA, the product of PLD (95). Few minutes after accumulation, PI(4,5)P2 is rapidly degraded by phospholipase Cγ (PLCγ) or phosphorylated, by type I phosphoinositide 3-kinase (PI3K), to PI(3,4,5) P3 (96), whose formation is required for pseudopod extension and phagosome formation (97). The rise of PI(3,4,5)P3 at sites of phagocytic receptor engagement promotes the recruitment of PLCγ through its pleckstrin homology (PH) domain (98), with the consequent hydrolysis of PI(4,5)P2 to diacylglycerol (DAG), which in turn can be converted in PA by DAG kinases (99). PA, PI(4,5)P2, and PI(3,4,5)P3 represent the most important lipid second messengers involved in the early phases of phagocytosis. These lipids contribute to the lateral spreading of phagocyte receptor signaling, which is important to integrate the signals elicited by sparsely phagocytic receptors and to cytoskeletal remodeling. Remodeling culminates with the formation of pseudopods, which surrounding the phagocytic target and sealing at their distal tips, form the phagosome (96). Phosphoinositides are key molecules in modulating cytoskeletal reorganization starting from the early phases of phagocytosis: PI(3,4,5)P3 contributes by recruiting WAVE and myosin X to forming phagosomes, while Table 3 | Lipid-binding domains and protein interaction of the most representative bioactive lipids involved in phagocytosis process.


*PI, phosphatidylinositol; PI(3)P, PI 3-phosphate; PI(4)P, PI 4-phosphate; PI(5)P, PI 5-phosphate; PI(4,5)P2, PI 4,5-bisphosphate; PI(3,5)P2, PI 3,5-bisphosphate; PI(3,4)P2, PI 3,4-bisphosphate; PI(3,4,5)P3, PI 3,4,5-trisphosphate; PA, phosphatidic acid; LBPA, lysobisphosphatidic acid; S1P, sphingosine 1-phosphate; AA, arachidonic acid; PH, pleckstrin homology; PX, phox-homology.*

PI(4,5)P2 can promote actin polimerization by (i) inducing WASP dependent Arp2/3 activation, (ii) removing capping proteins like profilin from the ends of actin filaments, and (iii) stimulating gelsolin-mediated severing of actin filaments and hence allowing fast growth of barbed ends (91). PI 3-kinase (Vps34) is expressed in early phagosomes (EPs) and causes enrichment in PI 3-phosphate (PI(3)P), which is crucial for EP maturation since it recruits multiple effectors through highly specific interactions with FYVE and phox-homology (PX) domains on target proteins, such as early endosome antigen-1, endosomal sorting complex required for transport (ESCRT)-0, sorting nexins, and p40phox of NADPH oxidase (100, 101). In EPs, also PA has an important role because it contributes to the assembly and activation of NADPH oxidase by interacting directly with p47phox and by recruiting and activating a number of protein and lipid kinases, including Fgr, the type I PIP5KI, and type 1 sphingosine kinase (14, 102). During the transition from early to late phagosome (LP) stage, PI(3)P is then lost as a consequence of hydrolysis catalyzed by a phosphoinositide lipid 3-phosphatases and PI 4-phosphate [PI(4)P] is rapidly acquired as a result of the enrichment in PI(4)P kinase 2A and such acquisition is associated with fusion of autophagosomes with lysosomes (LYs) (103). LP stage is also characterized by the presence of PI 5-phosphate [PI(5)P] and PI 3,5-bisphosphate [PI(3,5)P2]. In particular, PI(3,5)P2 can be generated by the enzymatic activity of both PI3K and phosphoinositide 5 kinase (PI5K), starting from PI(3)P and PI(5)P, respectively, and takes part in endolysosome morphology, acidification, and trafficking (104). PI(5)P can be directly produced from PI by the means of the PI5K or by dephosphorylation of PI(3,5)P2 by myotubularin 3-phosphatases and seems to be involved in actin remodeling and in protein sorting in endosomal compartment (105).

Phospholipids may also contribute to the membrane surface charge. Changes in the content of anionic phospholipids are accompanied by marked alterations in the surface charge of the membrane of nascent phagosome, where proteins that associate with the membrane by electrostatic interactions may relocate (128). For example, the dislocation of phosphoinositides [prevalently PI(4,5)P2 and PI(3,4,5)P3] causes the modification of surface charge observed after PSL, and the different charge facilitates the dissociation of target proteins causing the termination of signaling and cytoskeleton assembly (92). Similarly, during phagosome maturation, the cytosolic leaflet of phagosomes remains negatively charged and this is prevalently due to accumulation of PS (129). Such a negative charge serves to target proteins with polycationic clusters or polybasic domains and seems to be an important determinant in the distribution of both synaptotagmins, which control membrane fusion events (130), and small GTPases of the Rab and Rho superfamilies, which are critical for both phagosome formation and maturation (131). Thus, the level of anionic phospholipids, including PS and phosphoinositide species, may provide an electrostatic switch to control protein recruitment and vesicle trafficking.

In addition to activation of different signaling pathways, phospholipids can promote phagosome maturation process by inducing positive or negative membrane curvatures. For example, PA is a cone-shaped (type II) lipid that induces negative (concave) curvature of membranes (132), whereas lysobisphosphatidic acid, an inverted cone-shaped fusogenic lipid, forces positive (convex) membrane curvature (125). The resulting changes in membrane curvature activate specific curvature sensors, which are protein motifs or domains, such as amphipathic lipid-packing sensor motifs and BAR domains, which act as geometry sensing tools (133). These sensors play important roles in membrane

Figure 3 | Contribution of different lipid second messengers to phagocytosis from phagosome formation to PL maturation and examples of interference by pathogens. (A) The first event which occur during particulate engulfment is the PCF and the rapid accumulation of PI(4,5)P2 levels, mediated by PI5K and made possible thanks to the translocation of PI(4)P from Golgi to the membrane. PI(4,5)P2 can also be converted in Ins(1,4,5)P3 and DAG by the means of PLC and, furthermore, DAG can be metabolized to PA through the DAGK activity. PA is also early generated by the activation of PLD2, which hydrolyzes PC to PA and choline. Moreover, in this stage, there is a rapid generation of S1P, produced by the phosphorylation of sphingosine by the means of SK1, which in turn may activate sphingolipid Ca2++-release-mediating protein of endoplasmic reticulum (SCaMPER). Following, at the stage of PSL, PI(4,5)P2 is phosphorylated to PI(3,4,5)P3 by PI3K, which is converted to PI(3,4)P2 by 5-PTase during PIN. EP is characterized by the presence of PI(3)P, resulting by both its translocation from Golgi and the dephosphorylation of PI(3,4)P2, mediated by 4-PTase. Although PA is present in the early stage of phagocytosis process, at this stage its conversion starting to PC is mediated by the action of PLD1. LP is characterized by the presence of PI(3,5)P2, PI(5)P, and LBPA. In particular, PI(3,5)P2 can be generated by the enzymatic activity of both PI5K and PI3K, starting from PI(3)P and PI(5)P, respectively. PI is the precursor of PI(3)P, PI(4)P, and PI(5)P and their phosphorylation is mediated by PIK3, PI4K, and PI5K, respectively, in ER. (B) Examples of pathogen interference with host lipid metabolism associated with phagosome maturation. *Salmonella enterica* secretes the PI phosphatases SopB, which has a direct effect on the PI(4,5)P2 hydrolysis and an indirect effect on PI(3)P production through the recruitment of Rab5 and its associated enzyme PI3K (Vps34). *Mycobacterium tuberculosis* (Mtb) produces a secretory acid phosphatase (SapM) and LAM, which hydrolyzes PI(3)P and inhibites the generation of PI(3)P, respectively. Moreover, Mtb is able to inhibit SK1 and PLD action. *Lysteria monocytogenes* invades cells by binding to the host cell Met receptor, which leads to activation of PI3Ks and PI(3,4,5)P3 production and promotes a partial phagosome maturation resulting in a pH decrease and lysteriolysin activation. *Legionella pneumophila*, sequesters endosomes, enriched in PI(4)P, from TGN through SidC and SdcA bacterial proteins, and hence creating replicating vacuoles for itself. *Abbreviations for lipids*: PI(3)P, PI 3-phosphate; PI(4)P, PI 4-phosphate; PI(5)P, PI 5-phosphate; PI(4,5)P2, PI 4,5-bisphosphate; PI(3,4)P2, PI 3,4-bisphosphate; PI(3,4,5)P3, PI 3,4,5-trisphosphate; LBPA, lysobisphosphatidic acid; PA, phosphatidic acid; AA, arachidonic acid; PI, phosphatidylinositol; PC, phosphatidylcholine; PS, phosphatidylserine; Ins (1,4,5)P3, inositol 3,4,5 trisphosphate; DAG, diacylglycerol; SPH, sphingosine; S1P, sphingosine 1-phosphate; LAM, lipoarabinomannan. *Abbreviations for enzymes*: PLD-1, phospholipase D-1; PLD-2, phospholipase D-2; PLC, phospholipase C; DAGK, diacylglycerol kinase; SK1, sphingosine kinase 1; PI3K, phosphoinositide 3-kinase; PI4K, PI 4-kinase; PI5K, phosphoinositide 5 kinase; 3-PTase, PI 3-phosphatase; 4-PTase, PI 4-phosphatase; 5-PTase, PI 5-phosphatase; *Other abbreviations*: PCF, phagocytic cup formation; PSL, phagosome sealing; PIN, phagosome internalization; EP, early phagosome; LP, late phagosome; PL, phagolysosome; LY, lysosome; EE, early endosome; LE, late endosome; MVBs, multivesicular bodies; RE, recycling endosome; RT, retrograde traffic; TGN, trans Golgi network; ER, endoplasmic reticulum; SCV, *Salmonella* containing vacuole; MCV, *Mycobacterium* containing vacuole; LmCV, *Listeria* containing vacuole; LpCV, *Legionella* containing vacuole.

trafficking and remodeling, by regulating membrane protein concentration, enzyme activation, lipid composition, and vesicle fusion events (134).

The translocation of PS on the outer leaflet of plasma membrane represents one of the early events of apoptosis, occurring in both infected and uninfected apoptotic cells, and represents an "eat-me" signal for macrophages (135–138). The functional consequence of a PS-dependent recognition and phagocytosis of apoptotic bodies (efferocytosis) by macrophages is the release of TGF-β, IL-10, and the inhibition of the production of proinflammatory cytokines, such as TNF-α and IL-1β (139). These features have been previously reported as a possible therapeutic strategy to reduce immunopathologic responses (136, 140). However, if phagocytosis involves apoptotic bodies derived from cells expressing PAMPs, as in the case of apoptosis mediated by an infectious agent, phagocytosis causes activation of antigen-presenting cells (APCs) and release of IL-6, TGF-β, and IL-23, leading to a preferential activation of pro-inflammatory Th17 cells (141). Moreover, apoptotic bodies from macrophages infected by *Mycobacterium bovis* bacille Calmette–Guerin (BCG) have been reported to promote antigen cross-presentation and CD8<sup>+</sup> T cell activation as well as an *in vivo* increased protection against experimental TB in mice (142). Coherently, emulsified PS has been described as a simple and efficient carrier to deliver antigenic peptide epitopes to professional APCs to induce both helper and cytotoxic T cell response *in vivo* (143).

Considering the crucial role of lipids in the activation of antimicrobial response, many intracellular bacterial pathogens evolved strategies aimed at interfering, at different levels, with lipid metabolism to generate specific pathogen-containingvacuoles permitting their intracellular survival (14) (**Figure 3B**). *Salmonella enterica* exploits type III secretion system to secrete the PI phosphatases SopB, a phosphoinositide phosphatases able to dephosphorylate PI(4,5)P2. Decreased levels of PI(4,5)P2 reduce the surface charge of the *Salmonella* containing vacuole (SCV) and promote membrane fission by reorganizing the actin cytoskeleton during bacterial internalization (144). SopB also recruits the RAB GTPase RAB5 protein and the RAB5 effector Vps34, a PI3K that accumulates PI(3)P on the SCV surface and interferes with the maturation to next LP stage (144). *Listeria monocytogenes* invades host cells by recognition of Met receptor by bacterial internalin B (InlB), which leads to the PI3K activation and consequent enhancement of PI(3,4,5)P3 levels. Such increase in PI(3,4,5)P3 promotes the fusion of early endosomes with the bacterial vacuole, allowing partial maturation of phagosome before lysis of the phagosome and escape into cytoplasm due to the pH-dependent activation of listeriolysin (145). *Legionella pneumophila*, through SidC and SdcA proteins, anchors to PI(4)P expressing endosomes from trans Golgi network to form a replicative niche for the pathogen (146). Mtb, as well as BCG, interferes with PL maturation (15, 136, 147) by depleting the phagosome of PI(3)P and hence arresting the vacuole in an immature form (148). The mycobacterial secretory acid phosphatase (SapM) acts in concert with the mycobacterial lipoarabinomannan (LAM), a phosphatydilinositol analog, to avoid the formation of PI(3)P by hydrolysis of the phosphate group (149) and by inhibiting the activity of type III PI3K Vps34 (150), respectively. Finally, an inhibition in host PLD activation was observed during early phases of interaction between Mtb and macrophage (151). This inhibition was not due to differences in protein expression, suggesting a metabolic control of the enzyme. Interestingly, an upstream inhibition of sphingosine-kinase exerted by Mtb was also reported which may be responsible of the reduced calcium mobilization (152) and the consequent inhibition of Ca++-dependent PLD activation (153) observed during *in vitro* Mtb infection.

Given the well-defined role of certain lipids in the physiology of phagocytosis, the possibility to modulate the phagosome maturation process by the use of bioactive lipids has been suggested as a strategy to increase the efficacy of innate bactericidal mechanisms. In this context, Anes et al. showed that selected bioactive lipids, such as arachidonic acid, ceramide, sphingosine, sphingomyelin, and sphingosine 1-phosphate (S1P), activate phagosome actin assembly and maturation resulting in killing of pathogenic mycobacteria in murine macrophages (154). Moreover, S1P and its analog lysophosphatidic acid (LPA) were both described to activate PLD-mediated and PL maturationdependent mycobactericidal response in human macrophages (155, 156) and in type II alveolar epithelial cells (157) in *in vitro* models of Mtb infection, highlighting the role of PA in the activation of antimycobacterial response. Interestingly, and of high translational value, both lysophospholipids promoted antimicrobial response, resulting in a significant intracellular killing of endogenous mycobacteria, in bronchoalveolar lavage (BAL) cells from pulmonary TB patients (155, 158). Finally, the intranasal administration of S1P or LPA in murine models of primary Mtb infection significantly reduced pulmonary mycobacterial burden and histophatology (156, 159, 160). On these grounds, the targeted delivery of several bioactive lipids into macrophages to increase their innate anti-bacterial mechanisms could be useful for the treatment of Mtb infection. However, such a delivery is challenging, due to lipid pharmacokinetic properties and to the difficulties to target specific cells. The possibility to incorporate lipid second messengers in Janusfaced liposomes may represent a strategy to overcome these limitations (15, 136). The main feature of these liposomes is the asymmetric distribution of phospholipids within the liposome membrane, with inner leaflet expressing bioactive lipids involved in phagosome maturation, and the outer surface expressing PS, representing an "eat-me" signal for macrophages. PA delivered by this strategy allowed to enhance PL maturationdependent (myco)bactericidal response of macrophages and BAL cells from TB patients as well as from patients affected by MDR bacterial pneumonia (15, 136). Notably, both PA and PI(5)P delivered by Janus-faced liposomes were able to enhance antimicrobial response against *Pseudomonas aeruginosa* in an *in vitro* cellular model of cystic fibrosis by using macrophages expressing a pharmacologically inhibited or a naturally mutated cystic fibrosis transmembrane conductance regulator (15), which are characterized by impaired phagosome maturation (161). These experimental evidences suggest the possibility to use liposomes to deliver bioactive lipids to enhance phagosome maturation-dependent antimicrobial response and to restore the PL maturation pathway, which can be corrupted by specific pathogens (15).

## LIPOSOMES AS "INDIRECT" ENHANCERS OF ADAPTIVE ANTIMICROBIAL RESPONSE

Vaccines represent a great contribution of medical sciences to the reduction of the deaths caused by infectious diseases. The majority of currently available vaccines induce neutralizing or opsonizing pathogen-specific antibody responses to protect against infectious diseases. However, a number of diseases are caused by pathogens that are not controlled by humoral responses and novel vaccine formulations aimed at generating protective pathogen-specific immune responses are under testing or development. According to the pathogen virulence factors or mechanisms of pathogenesis, novel vaccines are being designed to enhance either cellular immune responses, with prevalence of cytotoxic T cells, or polarization of specific T cell subpopulations, or aimed at increasing the response at mucosal or systemic sites. One of the strategies currently under development is based on the use of recombinant antigens (subunit vaccines) in association with adjuvants able to modulate/polarize the developing immune response. Recombinant antigens show a higher safety profile and lower reactogenicity, but are characterized by a reduced immunogenicity in comparison with inactivated or attenuated whole cell vaccines. As a consequence, adjuvants are generally included to vaccine formulations to provide the necessary inflammatory microenvronment required for the activation of innate immunity cells and the following priming and expansion of naïve T cells for the development of appropriate adaptive immune response. Thus, adjuvants are generally defined as substances that (i) facilitate a depot effect and deliver the antigen in proximity of APC, (ii) generate an inflammatory microenvironment necessary for the activation of APC, and (iii) induce the secretion of specific cytokine patterns by APC for the differentiation of naïve T cells into different CD4<sup>+</sup> and/or CD8<sup>+</sup> T cell subpopulations.

In addition to their suitability as drug carriers, liposomes represent an extraordinary tool for the devise of innovative vaccines, as they can be designed for the antigen delivery (see chapter 5) and/or for their possible adjuvancy function (162). In fact, several liposome adjuvants have been licensed for human use and others are being evaluated in clinical trials (**Table 4**). However, many other formulations are possible and have been or could be tested for efficacy and safety. In this regard, different liposome adjuvant formulations can be produced with different properties according to the (i) lipid composition, (ii) liposome size, (iii) liposome charge, and (iv) inclusion of function modifiers such as PAMPs.

### Lipid Composition

The choice of liposomal lipid composition (in terms of hydrocarbon length, unsaturation, charge, and headgroup species of the lipids) influences the physico-chemical features of the liposomes, such as the lipid bilayer fluidity, which can in turn influence immune response. For instance, the degree of saturation or the length of fatty acids may affect the strength of the van der Waals forces among neighboring chains. Thus, phospholipids with longer saturated hydrocarbon chains show higher tendencies to interact each other and to form rigidly ordered bilayer structures than shorter and/or unsaturated hydrocarbon chains, which form liposomes with fluid and disordered bilayers (163). Membrane liposome fluidity in turn affects antigen presentation as it has been previously shown that fluid disordered phase liposomes promoted MHC class I presentation pathway to a higher degree than the solid ordered liposomes (164). Furthermore, in *in vivo* studies, solid ordered liposomes prepared using dimethyldioleoylammonium (DDA) induced 100-fold higher Th1 response than the fluid liposome counterpart prepared using the unsaturated analog DDA bromide (165). Inclusion of Chol is also known to influence bilayer fluidity and is commonly incorporated within liposomes as it can enhance liposome stability. The impact of Chol on adjuvancy is actually controversial, with some studies showing improvements and others reduction of immune response (166). Cationic adjuvant formulation (CAF)01 is actually a commercially available liposome adjuvant formulation that is composed of DDA stabilized with trehalose 6,6-dibehenate (TDB), a glycolipid synthetic variant of mycobacterial cord factor. TDB activates macrophages and dendritic cells (DCs) *via* Syk– Card9–Bcl10–Malt1 signaling pathway and when administrated


Table 4 | Examples of liposome adjuvants vaccines against infectious disease in market or tested in clinical trials.

*This compilation was generated with data from ClinicalTrials.gov.*

*DOPC, dioleoylPC; DOPE, dioleoly-sn-glycero-phophoethanolamine; DDA, dimethyldioleoylammonium; HA, hemaglutinin; NA, neuroaminidase; TDB, trehalose6,6-dibehenate; MPLA, monophosphoryl lipid A; QS21, triterpene glucoside compound derived from the Quillaja saponaria tree; CCS, N-palmityol-derythro-sphingosyl-Carbamoyl-Spermine; CCS/C, N-palmityol-derythro-sphingosyl-Carbamoyl-Spermine/Cholesterol; EPA, ExoProtein A (a detoxified form of exotoxin A from Pseudomonas aeruginosa); n.d., not determined; CFA, cationic adjuvant formulation; HAV, hepatitis A virus.*

in combination with a Mtb subunit vaccine induced a robust and combined antigen-specific Th1 and Th17 responses and partial protection against Mtb challenge (165, 167).

#### Liposome Size

The dimension of liposomes can impact adjuvant efficacy and several studies have shown that Th1 or Th2 response can be evoked by using particles with different diameter (168). In this context, a significantly higher Th1 response, as measured in terms of IL-12-dependent IFN-γ secretion and serum IgG2a production, has been reported following vaccination with large vesicles (≥225 nm diameter), whereas the same antigen encapsulated in small liposomes (≤155 nm diameter) induced a prevalent Th2 response as determined in terms of IgG1 and increased lymph node (LN) IL-5 production (169). Such a different outcome can reflect differences in particle trafficking to LNs: small particles (20–200 nm) freely traffics to the draining LNs where they are taken up by LN resident DCs, whereas larger vesicles (500–2,000 nm) are internalized by tissue-resident DC (170). Different antigen presentation pathway has also been observed following internalization by phagocytes: antigen encapsulated in large (560 nm) vesicles localized in early immature phagosomes, where class II antigen-processing pathway could be intercepted for recognition by CD4<sup>+</sup> T cells, while antigen encapsulated in small (155 nm) vesicles were rapidly transferred to late endosomes/lysosomes, with a consequent reduced class II-restricted Ag-presenting efficiency (171).

#### Liposome Charge

Liposome charge may have critical effects on the adjuvant properties of liposomes. Cationic liposomes may interact with the negatively charged mucosal surfaces and exibit increased mucoadesion, leading to reduced clearance rate, prolonged exposure time of the antigen at the mucosal surface (depot effect), enhanced endocytosis of liposomes by APC (13, 172–175), and increased cell-mediated immune response (173) when compared with neutral liposomes, which tended to induce antibody immune response (176). However, anionic liposomes may also modulate immune response in the context of novel formulations of liposomal adjuvants. As above described, PS is expressed on the outer membrane leaflet of apoptotic bodies and represents an "eat-me" signal for macrophages and DC, which express specific PS receptors. Phagocytes that internalize apoptotic bodies through PS became polarized toward a tolerogenic response (177). However, when the recognition occurs in the presence of selected PAMPs a Th17-differentiating cytokine profile is produced (170) and a similarly efficient class II- and class I-restricted antigen presentation is induced (143).

#### Function Modifiers

The procedures of liposome preparations make possible to include within the liposome structure different function modifiers molecules, such as PAMPs. The addition of function modifiers molecules may permit to design APC-targeted liposomes able to modulate the effector function of APC, including the profile of secreted cytokine that, in turn, can induce the differentiation of diverse T helper cell subpopulations (178). For example, the immunization of mice with liposomes containing ovalbumin (OVA) and oligodeoxynucleotide (ODN) with CpG motifs (ligand for TLR-9) induced a Th1-type immune response, with enhanced production of IFN-γ and of IgG2a, whereas the same antigen encapsulated with Pam3CSK4 (a TLR-2 ligand consisting of tri-palmitoyl-*S*-glyceryl cysteine lipopeptide with a pentapeptide SKKKK) induced increased levels of IgG1, suggesting a Th2-oriented immune response (179, 180). Another example is given by the lipophilic muramyldipeptide (MDP) analogs MDP phosphatidylethanolamine as well as the MDP glyceroldipalmitate that induced higher antibody titers, Th1 response, and IFN-γ levels if added to liposomal formulations containing hepatitis B surface antigen (HBsAg) (181). Thus, the use of different PAMPs can modulate the immune response against the same antigen, in dependence on the targeted TLR.

Monophosphoryl lipid A (MPLA) is a safe and potent liposome adjuvant used with many candidate antigens for new vaccines in the fight against several types of cancer and anti-infective vaccines such as malaria and HIV-1 (182). For example, RTS,S (a particulate antigen consisting of hepatitis B particles coexpressing epitopes derived from *P. falciparum* circumsporozoite protein) induced an *in vivo* cytotoxic T-lymphocyte response and a dose-dependent enhancement of the specific IgG levels when entrapped in liposomes containing MPLA and not when encapsulated in liposomes missing MPLA (183). Moreover, a recent paper demonstrated that the *P. falciparum* Protein-013 can induce a potent and sterilizing immune response when formulated with small ULV containing a synthetic MPLA derivative (3D-PHAD®) and the QS-21 (184). Similarly, Nagill and Kaur encapsulated the 78-kDa antigen of *Leishmania donovani* with MPLA, resulting in decreased parasite burden after challenge in immunized mice (185).

Korsholm et al. described a novel CD8<sup>+</sup> T-cell-inducing adjuvant, CAF09, consisting of DDA-liposomes stabilized with monomycoloyl glycerol-1 (a synthetic analog of a mycobacterial cell wall lipid and a potent stimulator of human DCs) and combined with the TLR3 ligand Poly(I:C) (186). CAF09 was used to test immunogenicity of the model antigen OVA as well as several antigens of Mtb (TB10.3, H56), HIV (Gag p24), HPV (E7) in mice and resulted in a high frequency of antigen-specific CD8<sup>+</sup> T cells (186). In the mouse model of subcutaneous HPV-16 E7-expressing TC-1 skin tumor, immunization with the E7 antigen and CAF09 as adjuvant significantly reduced the growth of the tumor (186). Another widely studied liposome carrier is the cationic liposome–DNA complex (CLDC) that is prepared by mixing liposomes and ODN with CpG motifs, which activate TLR9. Bernstein and colleagues demonstrated that CLDC increased the immune response against simian immunodeficiency virus induced by vaccines. The immunization of rhesus macaques with CLDC induced a potent virus-specific B- and T-cell response in comparison to control animals (12, 187). Moreover, in mice expressing hepatitis B virus, the CLDC was tested as adjuvant for HBsAg. Whereas HBsAg induces only a B-cell immune response, the viral antigen formulated together with CLDC elicited both T- and B-cell responses (12, 188).

#### LIPOSOMES AS ANTIGEN CARRIER IN INNOVATIVE VACCINE PREPARATIONS

Allison and Gregoriadis were the first to describe the capacity of liposomes to elicit immune responses against associated or incorporated antigens (12, 189). Water-soluble compounds (peptides, proteins, carbohydrates, nucleic acids, and haptens), depending on the chemical properties, can be linked to the surface of liposomes, by stable chemical bond or by adsorption, or can be trapped inside the aqueous inner space, whereas lipophilic compounds (antigens, lipopeptides, linker molecules, and adjuvants) can be inserted into the lipid bilayer (12, 190). Surface-exposed antigens are available for and can stimulate B-cell for antibody production, while both surface-exposed and encapsulated antigens, which require intracellular liposome disruption to be accessible, may induce T-cell responses. Liposome-encapsulated protein antigens have been frequently used and several studies well documented the MHC class-I presentation and induction of cytotoxic T lymphocytes (CTLs) by antigens formulated inside liposomes. H-2 antigens entrapped in egg lecithin plus Chol (30% w/w) liposomes induced a potent anti-H-2 CTL allo-response (191). Human LS174T colon tumor cell membranes encapsulated into liposomes (PC/Chol/PA 7:2:1) elicit *in vitro* specific primary and secondary xenogeneic immune responses in murine splenocytes (192). Induction of antiviral immunity by liposome-encapsulated peptides was tested by Ludewig and colleagues that showed the high immunogenicity of peptides derived from the glycoprotein of the lymphocytic choriomeningitis virus when administered intradermally in mice (193). In addition, these authors used the liposome to prime a strong MHC class I-restricted T-cell response specific for 10 diverse epitopes of hepatitis C virus (HCV) (12, 194). In the fight against Ebola Zaire (EBO-Z) virus, the protective efficacy of liposome-encapsulated [L(EV)] irradiated EBO-Z, containing all of the native EBO-Z proteins was evaluated (195). Mice immunized intravenously with L(EV) and challenged with a uniformly lethal mouse-adapted variant of EBO-Z were totally protected from illness and death compared with mice vaccinated intravenously with the irradiated virus nonencapsulated that presented a 55% of survival (195). Liposomes were also formulated with the pH-sensitive 3-methyl-glutarylated hyperbranched poly(glycidol) (MGlu-HPG) and used to entrap OVA. Such MGlu-HPG formulations elicited a strong T cell activation that was inhibited using blocking antibodies against MHC class-I/MHC class-II molecules, suggesting the involvement of MHC class I- and II-restricted antigen presentation (12, 196). Ding and colleagues developed the so-called RAFTsomes, a liposome incorporating membrane microdomains of APCs with enriched epitope/MHC complexes. OVA epitope loaded RAFTsome immunization gave high anti-OVA IgG1 levels and the immunized mice were protected from OVA-expressing EG.7 tumor cell inoculation challenge (197).

Mucosal surfaces are the main entry site for most environmental antigens and the mucosal immunity plays a critical role in preventing the initial infection of many pathogens. Several studies are focused on figuring out the ability of liposomes to act as efficient mucosal antigen delivery system. A potential mucosal carrier formulation has been developed by Gupta and Vyas that encapsulated HBsAg in liposomes coupled with *Ulex europaeus* agglutinin 1, a lectin isolated from *U. europaeus* seeds. After oral immunization, lectinized liposomes were predominantly targeted to M cells on intestinal Peyer's patches, inducing a strong anti-HBsAg IgG response in serum, anti-HBsAg IgA in various mucosal fluids, and cytokine levels in the spleen homogenates (198). Figueiredo and colleagues described another mucosal formulation composed of *Streptococcus equi* antigens encapsulated inside PC/Chol/SA liposomes or inside chitosan nanoparticles. Mice immunized intranasally with both delivery systems developed mucosal, humoral, and cellular immune responses, but higher secretory IgA levels in the lung were observed following vaccination with chitosan nanoparticles, due to their enhanced mucoadhesive properties (12, 199). Wang and colleagues encapsulated the OVA in a galactosylated liposome carrier in which a galactosyl lipid (galactose conjugated covalently with 1,2-didodecanoyl-sn-glycero-3-phosphoethanolamine) was incorporated into a liposomal bilayer. OVA-encapsulated targeted galactosylated liposome elicited in the nasal and lung wash fluid a significantly higher OVA-specific secretory IgA titers and serum IgG antibody levels than control mice (200). Traditional phosphodiester liposomes are not stable and could be easily degraded in the gastrointestinal tract. Zhang and colleagues described the archaeosomes, a novel oral delivery system based on the polar lipid fraction E isolated from *Sulfolobus acidocaldarius*. The archaeosomes had significant higher stability in simulated gastric and intestinal fluids after oral administration and induced a strong serum IgG as well as mucosal IgA immune response. Moreover, such delivery system elicited antigen-specific MHC class I-restricted T cell proliferation (201).

Liposomes are versatile tool to encapsulate and deliver a wide range of molecules, not only proteins but also glycolipid antigens. Kallert et al. demonstrated that purified mycobacterial LAM can be efficiently delivered into CD1<sup>+</sup> APC *via* liposomes and this delivery system induces robust LAM-specific Th1-biased CD1 restricted T-cell responses in primary human cells (202). The authors used liposomes consisted of egg-PC, Chol, and stearylated octaarginine and proved that octaarginine, a cell-penetrating peptide consisting of eight positively charged arginines, increases the glycolipid antigen accumulation into lipid-APCs. The efficacy of octaarginine-containing liposomes was also tested in *in vivo* models of immunization using both BCG primed guinea pigs (203), where these liposomes induced a delayed type hypersensitivity reaction in the skin and in rhesus macaques, using glucose monomycolate (GMM) as antigen, where a CD1-restricted Th1-skewed immune response was observed (202, 204). In a recent study, GMM-containing liposomes were decorated with a high-affinity glycan ligand of the sialic acid-binding Ig-like lectin-7, a siglec receptor expressed on DC that mediates rapid endocytosis and transport of its cargo to LYs (205). This elegant targeting platform leads to a robust GMM-specific CD1-restricted activation of T cells. One way to ensure rapid and specific liposome uptake by leukocyte subsets is decoration of liposomes with antibodies against a cell subset-specific antigen. Klauber et al. reported a liposomal drug delivery system for robust and specific targeting of monocytes and DCs consisting of sterically stabilized liposomes with surface-conjugated antibodies against C-type lectin "Dendritic Cell Immunoreceptor" (206). Using this method, they were able to activate the targeted cells and improve the ability of the agonist to induce secretion of key anticancer cytokines (IL-12p70, IFN-γ, and IFN-α).

Typically, exogenous antigens residing in proteolytic PLs are directed to MHC class II-expressing compartments (MIIC) where peptides resulting from antigen proteolysis are loaded on the neoformed class II molecules to be expressed on cell membrane. Thus, bioactive lipids like PI(3)P, promoting (auto) PL biogenesis, can in turn promote class II antigen presentation pathway (207). Harding and colleagues (208) used liposomes composed of dioleoyl PC/dioleoyl PS (DOPC/DOPS) or dioleoyl phosphatidylethanolamine/palmitoyl homocysteine (DOPE/ PHC) at 4:1 molar ratios with four different encapsulated protein antigens (OVA, murine hemoglobin, bovine ribonuclease, or hen egg lysozyme) and proved that macrophages efficiently processed the encapsulated Ag. These liposomes were able to sequester their contents from potential endosomal processing events and release them only after delivery to LYs for efficient MHC class II presentation (208).

The antigen presentation pathway leading to the loading of exogenous antigens on MHC class I molecules is called crosspresentation. Cross-presentation can occur *via* a vacuolar and a cytosolic pathway. The vacuolar pathway bypasses the cytosolic steps and TAP- or proteasome-sensitivity, is dependent on lysosomal proteases, and invokes a peptide exchange step for reloading of endocytosed MHC-I complexes, which are then rerouted directly back to the plasma membrane in a Brefeldin A-insensitive manner (209). The cytosolic route, considered to be the most important, is TAP and proteasome dependent and allows the export of the antigen into the cytosol through the acquisition of the traslocon SEC61 on antigen-containing endosomes from the endoplasmic reticulum (ER)-associated degradation pathway (210). Although no specific lipid second messengers promoting cross-presentation pathway have been identified so far, the involvement of specific membrane contact sites between ER and phagosomes (211) suggests that specific signal lipids can be involved. The permanence of the antigen in a non-proteolytic compartment represents a condition promoting cross-presentation and explains why DC, whose phagosomes show reduced acidity and limited proteolysis, have better cross-presentation capability than macrophages. In this context, cationic liposomes, but not anionic liposomes, that increase the lysosomal pH in DCs and reduce antigen degradation, promote antigen cross-presentation and CD8<sup>+</sup> T-cell cross-priming (212, 213). As an alternative strategy to promote antigen delivery to cytoplasm and make it available for class I antigen presentation, Miura et al. developed liposomes carrying OVA modified with KALA peptide, an α-elical cationic peptide derived from the sequence of the N-terminal segment of the hemagglutinin (HA)-2 subunit of the influenza virus HA, which is involved in the fusion of the viral envelope with the cell membrane (214), and show membrane destabilizing properties (215). MHC class I presentation pathway can also be favored using pH-sensitive liposomes. For example, Reddy et al. have shown that OVA entrapped in pH-sensitive liposomes (DOPE/1,2-dioleoyl-snglycero-3-succinate 1:1 and DOPC/PS/Chol 5:2:3) allowed MHC class I presentation of OVA peptides by mouse thymoma cells, which were lysed by OVA-specific CD8<sup>+</sup> T lymphocytes (216).

Liposomes can be designed to carry antigens within the acqueous phase, encapsulated inside the lipid bilayer, or exposed onto the liposome surface and this can result in different outcomes of the immune response. Investigation of HA adsorption versus encapsulation and coencapsulation of ODN with CpG motifs in 3β-[*N*-(*N*′,*N*′-dimethylaminoethane)-carbamoyl] Chol (DC-chol) liposomes demonstrated that encapsulated HA was less immunogenic than adsorbed HA (217). Moreover, Takagi and colleagues immunized human-HLA-transgenic mice with liposomes exposing several epitope-peptides derived from HCV in their external layer and observed that 1 mouse- and 3 human MHC-restricted peptides protected and conferred longterm memory to vaccinated animals (218). Immunization of HLA-A\*0201-transgenic mice with liposomes conjugated with peptides designed according to the sequence of highly conserved antigens of influenza viruses and HLA-A\*0201 binder, induced antigen-specific CD8<sup>+</sup> T-cells, and mediated protection of mice challenged intranasally with the influenza viruses H1N1 or H3N2 (219). Similarly, when the peptide nucleoprotein (NP)366–374, designed according to the sequence of the NP of influenza H3N2 virus, was coupled with liposomes and used to immunize mice induced peptide-specific CD8<sup>+</sup> T-cells and the replication of influenza H3N2 virus was successfully suppressed in the lung of challenged mice (220). Naito and colleagues reported that OVA chemically coupled with the surface of liposome *via* amino groups using glutaraldehyde induced Ag-specific IgG but not IgE Ab production (221), showing that liposomes can be exploited to develop vaccines devoid of allergenic potential. Moreover, liposomes showed different capacities to induce class I or class II presentation of the delivered antigen depending on the mobility of their membranes. Taneichi et al. reported that OVA coupled with liposome constituted by unsaturated fatty acid was presented to both CD4<sup>+</sup> and CD8<sup>+</sup> T lymphocytes, while OVA conjugated to liposomes constituted by saturated fatty acid were not presented to CD8<sup>+</sup> T lymphocytes (222). The *in vivo* induction of CTL and the eradication of E.G7 tumor in mice confirmed the capacity to induce cross-presentation by Ag-conjugated liposomes prepared including unsaturated fatty acid in the bilayer (222). Masek and colleagues used small Ni-chelating liposomes exposing on their surface the *Candida albicans* His-tagged antigen hsp (hsp90-CA) with the MDP derivative C18-*O*-6-norAbuMDP as adjuvant. These liposomes were phagocytosed by human DCs *in vitro* and *in vivo* in mice, where induced antigen-specific Th1 and Th2 responses without side effects (223). At least in theory, the covalent attachment of protein antigens to nanocarriers could modify protein structure and mask epitopes, altering the antibody response. Watson and colleagues tested this hypothesis using metal chelation *via* nitrilotriacetic acid (NTA) to attach antigens to liposomes instead of covalent linkage. OVA and a peptide derived from the membrane-proximal region of HIV-1 gp41 (N-MPR) were attached *via* NTA or covalent linkage. Covalently attached N-MPR or OVA elicited stronger antibody responses than NTA-anchored antigens excluding the possible masking effect of covalent binding (224). Finally, antigen modification may also impact the response against haptens as reported by a study by Matyas et al., where the authors conjugated haptenic compounds to protein carriers and embed them in the outer surface of MPLA-containing liposomes. Four synthetic opiate haptens were conjugated to carrier proteins and induced high levels of specific antibodies to synthetic heroin haptens following *in vivo* immunization that might be useful as a candidate vaccine to heroin and similar opiates (225).

Self-amplifying messenger RNA (mRNA) technology has been developed for *in situ* expression of antigens and represents a new platform for innovative vaccine applications (226) with the advantage that unlike DNA, mRNA-based vaccines do not integrate into chromosomes avoiding the risks of oncogenesis and insertional mutagenesis (227). Liposomes were identified as the more suitable delivery system for non-amplifying or selfamplifying mRNA vaccine, and it was rapidly evident that mRNA encapsulated inside liposomes can activate innate immunity through toll-like and RIG-I-like receptors (228, 229).

Richner and colleagues tested the efficacy of this novel technology in a vaccine against Zika virus using a modified mRNAencoding prM-E gene that produced virus-like particles with a reduced cross-reactivity with the related dengue virus (230). Two immunizations induced a strong protective humoral response against Zika infection in mouse model and reduced the crossreactive response against the related virus dengue. For this experiment, liposomes were obtained taking advantage of a method developed by Chen and colleague, consisting of ionizable lipid, DSPC, CHOL, and PEG-lipid (dissolved in ethanol at the molar ratios 50:10:38.5:1.5) combined with a citrate buffer containing mRNA (3:1 = aqueous:ethanol, pH 4.0) using a microfluidizer. The dialyzed and concentrated liposome preparations resulted in 80–100 nm in size and more than 90% encapsulation (231).

The mRNA-based liposomes were also tested to induce humoral responses against influenza A virus in a mouse model. Lipid nanoparticles were engineered to deliver a synthetic, selfamplifying mRNA encoding seasonal influenza HA and tested their immunogenicity in mouse model. This mRNA-based vaccine elicited an immune response comparable to that obtained with the licensed influenza subunit vaccine and all the immunized animals showed HA inhibition and neutralizing antibody titers against the virus 14 days after the second immunization (232). In another study, the immunogenicity of lipid nanoparticles encapsulating a modified non-replicating mRNA encoding influenza H10 HA was tested in *rhesus macaques*. This study showed that anti-HA protective antibody titers and MHC class II-restricted H10-specific T cell response were elicited after two intradermal or intramuscular immunizations. Moreover, this study provided evidences on cells and mechanisms responsible for the efficacy of mRNA vaccinations: following the administration of mRNA–liposomes, DCs, monocytes, and neutrophils were recruited to the site of immunization and draining LNs, but only DCs and monocytes internalized liposomes, produced virus antigens, translating the mRNA, and upregulated costimulatory molecules. This *in vivo* antigen production and activation of APCs lead to the priming of a potent H10-specific MHC class II-restricted T cell response (233). The superiority of liposome encapsulation versus other delivery systems of mRNA was stressed by the group of Bogers and colleagues, who proved the immunogenicity and the safety of cationic liposomes formulated with the mRNA encoding for a clade C envelope glycoprotein of HIV in *rhesus macaques*. This formulation elicited a cellular immune response and a neutralizing antibody response stronger than those elicited by self-amplifying mRNA encapsulated in a viral replicon particle or by a recombinant HIV envelope protein formulated with MF59 adjuvant (234).

### ENGINEERING MULTIROLE LIPOSOMES ABLE TO DELIVER DRUGS, ANTIGENS, ADJUVANTS, AND/OR LIPID SECOND MESSENGERS AS AN ADDITIONAL TOOL TO CONTRAST INFECTIOUS DISEASES

Infectious diseases are the leading causes of morbidity and mortality worldwide and the increasing incidence of treatmentresistant infections has become a major global concern for their epidemic potential (235). In 2016, WHO reported 6 × 105 new cases of TB with resistance to rifampicin, the most effective firstline anti-TB drug, of which 4.9 × 105 had MDR TB and about 9.5% of MDR-TB cases have additional drug resistance and were defined XDR-TB.1 Moreover, 2.3 × 104 in US2 and 2.5 × 104 deaths in Europe3 are attributable each year to infections caused by Gram-positive (*Streptococcus penumoniae, Staphylococcus aureus*), and Gram-negative (*Pseudomonas* spp., *Acinetobacter* spp., and *Enterobacteriaceae*) bacteria with multidrug- or panantibiotic resistance. As a consequence, the development of novel antimicrobial agents and/or other immunotherapeutic options represents a global health priority (236). To this regard, several therapeutic strategies, targeting the host rather than the pathogen, have been proposed and are under development as adjunctive treatments, in particular for MDR TB (235). Examples of host-directed therapies are the reinfusion of *in vitro* expanded pathogen-specific autologous T cells, the administration of micronutrients, antimicrobial peptides, or immune modulators and therapeutic vaccines (235).

Liposomes may offer several advantages in the design of novel pathogen- or host-directed therapies and vaccines, since these nanoparticles can be engineered as delivery system as well as immune modulators. In clinical applications, liposomal drugs have been proven their ability to "passively" accumulate at sites of increased vasculature permeability and for their ability to reduce the side effects of the encapsulated drugs in comparison to free drugs (237). This has resulted in an overall increase in TI, so that liposomal drug delivery has become an established technology platform in some circumstances and has gained considerable clinical acceptance (26). A new frontier in liposomebased approaches against infectious diseases may be envisaged if liposomes are designed as carrier of antigens or drugs in addition to molecules with immunomodulatory functions (**Figure 4**). For example, antigen-loaded liposome carrying different PAMPs inside or exposed on their outer surface leaflet, would target plasma membrane-associated or intracellular pattern-recognition

<sup>1</sup>http://www.who.int/tb/publications/global\_report/en/.

<sup>2</sup>https://www.cdc.gov/drugresistance/threat-report-2013/index.html.

<sup>3</sup>http://www.ema.europa.eu/ema/index.jsp?curl=pages/news\_and\_events/ news/2009/11/news\_detail\_000044.jsp&mid=WC0b01ac058004d5c1.

receptors, respectively, and consequently activate APC to secrete different cytokines, which in turn may drive diverse polarizations of the antigen-specific adaptive immune response (11). It is also possible to engineer asymmetric liposomes that express different lipids at the outer and inner membrane leaflet (15, 136, 238). In this regard, PS may be included in the constitutive lipids of the outer liposome leaflet to target specific cells, like macrophages, whereas lipid second messengers, such as PA or PI(3)P, can be incorporated in the inner leaflet to be delivered to the cell microvesicular system to modulate intracellular endosome trafficking (15, 136, 238). Pathogens may target host lipid metabolism and PL maturation as an intracellular survival strategy and mechanism of immunoevasion (14, 144–146, 148–152, 238–241). The growing evidences for a crucial role played by lipid second messengers in PL biogenesis (91–94, 129) opens new therapeutic possibilities for the reactivation/enhancement of antimicrobial innate immune response with the specific intracellular delivery of lipid second messengers taking advantage of the liposome technology. Since scaling up, stability and regulatory issues are commonly successfully addressed, the clinical availability of novel liposome formulations can be foreseen in the near future. The possibility to enhance/correct specific molecular pathways by lipid second messengers, used for liposome scaffold, may represent a further added value to the plethora of different possibilities offered by liposomes technology and a possible novel host-directed therapy to face the global emergence of antimicrobial resistance.

## AUTHOR CONTRIBUTIONS

All authors made substantial contributions to the conception or design of the work. NP, SM, and FS contributed to the acquisition, analysis, interpretation of data for the work. NP and FS drew the figures. RN and MF wrote the work. All authors revised critically the work for important intellectual content, gave final approval of the version to be published, and agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

to face pathogen-specific strategies.

#### ACKNOWLEDGMENTS

Authors would like to thank Raffaela Teloni for technical assistance.

#### FUNDING

Research supported by (i) the Horizon 2020 Programme of European Commission, grant "EMI-TB"; Eliciting Mucosal Immunity against Tuberculosis—grant # 643558; and

#### REFERENCES


"TBVAC2020; Advancing novel and promising TB vaccine candidates from discovery to preclinical and early clinical development"—grant# 643381; (ii) Italian Foundation for Cystic Fibrosis, "Preclinical study of a host-directed therapy based on Metformin and bioactive liposomes for the control of multidrug-resistant *P. aeruginosa* infection"—grant #14/2017; (iii) Italian Foundation for multiple sclerosis, grant #2016/R/22; and (iv) Ministero della Salute Italiana, Ricerca Finalizzata grant: "Nanotechnology for the multiplex diagnosis of infectious diseases"—grant# PE-2011-02346849.


phagosome maturation in human macrophages via specific inhibition of sphingosine kinase. *J Immunol* (2003) 170(6):2811–5. doi:10.4049/ jimmunol.170.6.2811


Th2 bias in the immune response and protection against infection. *Vaccine* (2009) 27(27):3643–9. doi:10.1016/j.vaccine.2009.03.040


Siglec-7 induces robust T cell activation. *J Immunol* (2014) 193(4):1560–6. doi:10.4049/jimmunol.1303278


immune response towards rHsp90. *J Control Release* (2011) 151(2):193–201. doi:10.1016/j.jconrel.2011.01.016


delivery of a self-amplifying RNA vaccine expressing HIV type 1 envelope with a cationic nanoemulsion. *J Infect Dis* (2015) 211(6):947–55. doi:10.1093/ infdis/jiu522


**Conflict of Interest Statement:** All the authors declare no business relationship that might lead to a conflict of interest.

*Copyright © 2018 Nisini, Poerio, Mariotti, De Santis and Fraziano. 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.*

*Yong Gao, Chanuka Wijewardhana and Jamie F. S. Mann\**

*Department of Microbiology and Immunology, University of Western Ontario, London, ON, Canada*

#### *Edited by:*

*Rajko Reljic, St George's, University of London, United Kingdom*

#### *Reviewed by:*

*Thorsten Demberg, Immatics Biotechnologies, Germany Evelina Angov, Walter Reed Army Institute of Research, United States*

> *\*Correspondence: Jamie F. S. Mann jmann62@uwo.ca*

#### *Specialty section:*

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

*Received: 29 November 2017 Accepted: 07 February 2018 Published: 28 February 2018*

#### *Citation:*

*Gao Y, Wijewardhana C and Mann JFS (2018) Virus-Like Particle, Liposome, and Polymeric Particle-Based Vaccines against HIV-1. Front. Immunol. 9:345. doi: 10.3389/fimmu.2018.00345*

It is acknowledged that vaccines remain the best hope for eliminating the HIV-1 epidemic. However, the failure to produce effective vaccine immunogens and the inability of conventional delivery strategies to elicit the desired immune responses remains a central theme and has ultimately led to a significant roadblock in HIV vaccine development. Consequently, significant efforts have been applied to generate novel vaccine antigens and delivery agents, which mimic viral structures for optimal immune induction. Here, we review the latest developments that have occurred in the nanoparticle vaccine field, with special emphasis on strategies that are being utilized to attain highly immunogenic, systemic, and mucosal anti-HIV humoral and cellular immune responses. This includes the design of novel immunogens, the central role of antigen-presenting cells, delivery routes, and biodistribution of nanoparticles to lymph nodes. In particular, we will focus on virus-like-particle formulations and their preclinical uses within the HIV prophylactic vaccine setting.

Keywords: HIV-1, vaccine, virus-like particles, nanoparticles, immunogenicity

## INTRODUCTION

Human immunodeficiency virus-1 (HIV-1) and HIV-2 have emerged as the result of zoonotic lentiviral transmissions of simian immunodeficiency viruses (SIV), from chimpanzees/gorillas and sooty mangabeys, respectively, into the human population. First identified as the etiological agent behind acquired immunodeficiency syndrome (AIDS) in humans in the early 1980s, HIV has continued to spread into a global pandemic and major public health concern. At the end of 2015 alone, there were 36.7 million people infected with HIV with an additional 2.1 million new infections that same year. Excluding parenteral infections, HIV acquisition is almost entirely transmitted through sexual intercourse. Worldwide, approximately 80% of the 30 million people infected with HIV acquired the virus through heterosexual transmission, with receptive vaginal and anal intercourse accounting for 60–70 and 5–10%, respectively, of the global total.

Human immunodeficiency virus infection is characterized by the catastrophic depletion of CD4 T cells—the very cells that orchestrate host immune responses against invading pathogens. The combination of viral cytopathic effects, cell suicide by caspase-1 mediated pyroptosis, and caspase-3-mediated apoptosis are the major mechanisms behind the depletion of CD4 T cells (1). Over time, the continued depletion of CD4 T cells causes an AIDS-defining illness where people become susceptible to a plethora of opportunistic infections.

During heterosexual transmission, HIV-1 enters the body through the genital mucosa, while in male to male transmission, HIV-1 enters through the rectal or upper gastrointestinal mucosa. Numerous factors contribute to HIV-1 transmission rates across genital and gut mucosal surfaces, including the viral load in the donor during sexual contact, presence of preexisting mucosal infections, trauma, inflammation, frequency of sexual contacts, male circumcision, receptive anal intercourse, and whether individuals used barrier contraceptives such as condoms. The current paradigm suggests HIV enters anatomical sites with micro-abrasions, likely occurring as a direct result of sexual activity, thus providing more direct access to target cells residing in submucosal tissues. It is important to note that mucosal frontline surfaces, such as the endocervix, transformation zone, and gastrointestinal mucosa, have only a single layer of columnar epithelium to separate the external environment from the sterile inner sanctum of the body. Thus, these surfaces are more susceptible to microtraumas and likely serve as entry points for HIV. After the initial seeding of an infecting founder population, the virus spreads to regional lymph nodes and then disseminates systemically. The narrow timeframe from breaching the mucosal frontline through to primary foci development and immediately prior to lymph node distribution is often referred to as the "window of opportunity" (2). This represents the instance where a potential vaccine candidate might be maximally effective at aborting an infection event. During this window of opportunity, HIV may also be at its most vulnerable state during infection considering only one or few HIV-1 clones were transmitted from a highly diverse HIV population in the recipient.

Due to the advent of antiretroviral therapies (ART), HIV is no longer a death sentence. From 1995–2010, approximately 5.1 million AIDS-related deaths were averted in low- and middleincome countries due to greater accessibility of ART (3). Current combination antiretroviral therapies (cART) inhibits various points of the viral replication cycle and, is most effective as a triple therapy regiment to prevent the induction of escape mutants. Although cART can decrease viral load in the blood to undetectable levels, it remains a non-curative strategy against HIV due to the presence of the latent reservoir (4). This is because cART can only target actively replicating virus and, therefore, cannot impact the latent reservoir.

#### EARLY ATTEMPTS AT HIV VACCINES

Initial HIV vaccine candidates to enter human clinical trials were subunit vaccines based on the surface Env glycoprotein. These were mainly soluble monomeric gp120 or gp160 constructs (5–7). These early attempts at vaccines provided the necessary evidence for protection against homologous SIV-HIV (SHIV) chimeric virus challenge in non-human primates (NHPs) (8). Unfortunately, these vaccines were not protective against a heterologous virus challenge despite inducing strong immune responses and the animals remained susceptible to infections with heterologous SHIV viruses (9). Despite this, numerous gp120 subunit vaccines have been designed and evaluated in human clinical trials to provide proof of principle for vaccine-mediated protection against infection. An example of an early monomeric gp120 vaccine trial conducted in human volunteers was the Phase III efficacy trial (VAX003). This study was conducted within an injection drug using cohort in Thailand. The vaccine (AIDSVAX B/E) consisted of a combination of recombinant clade B and E gp120 antigens adjuvanted with alum. The vaccine ultimately provided no significant levels of protection against infection among the participants, with 8.4% in the vaccine arm and 8.3% in the placebo arm becoming infected (10, 11). Another notable human vaccine trial was the VAX004, which employed the AIDSVAX B/B vaccine and showed near identical results in North America and in the Netherlands. This vaccine was given to men who have sex with men (MSM) and women at high risk for sexually transmitted HIV infection cohorts (12, 13). Despite eliciting high homologous neutralizing antibody titers against the applied vaccine strains, only weak neutralizing antibody responses were detected within recipients. Overall, both VAX003 and VAX004 vaccine trials failed to show protection from HIV infection, despite being immunogenic. Such studies helped initiate the search for new and improved immunogens and the development of oligomeric Env structures to better mimic the HIV-1 functional spike. In theory, trimeric Envs may be advantageous to predecessor monomeric vaccine candidates as it is thought they might be able to induce conformationally dependent antibodies. Most of the trimeric Env immunogens, the so-called gp140 immunogens consisting of gp120 and the ectodomain of gp41 (14–16), elicit significantly higher titers of neutralizing antibodies compared to the monomeric Env forms (17–21), however, the breath of neutralization is still limited. Attempts to enhance the immunogenicity of Env antigens, i.e., increase the neutralizing antibody levels, involved modifying Env molecules by deleting the highly glycosylated V2 domain. Unfortunately, this had minimal impact in eliciting broadly neutralizing antibodies. Trimeric gp140 with new disulfide bonds to stabilize gp120–gp41 interactions were tested for immunogenicity within novel prime-boost immunization protocols (22). When rabbits were primed with DNA encoding a membrane bound form of gp140 by electroporation followed by disulfide-stabilized trimeric gp140 booster immunizations *via* the intramuscular (IM), intradermal (I.D.), or subcutaneous (S.C.) routes, high titers of neutralizing antibodies against homologous viruses could be elicited.

Since these early failed attempts at vaccination, numerous DNA, and viral vector platforms have been created and evaluated as anti-HIV vaccine candidates. While DNA/RNA and viral vectors do not fall under the mandate of this review, it is worth mentioning the only clinical trial to date to have demonstrated any efficacy against HIV acquisition was RV144—the phase III "Thai Trial." In 2009, the results from the RV144 trial were released, demonstrating the effectiveness of an ALVAC vaccine prime followed by an AIDSVAX boost. The ALVAC prophylactic vaccine component was a canarypox vector encoding HIV-1 Gag, Pol, and gp120 Env while the AIDSVAX boost consisted of gp120 B/E adsorbed to alum. A modest benefit of 31.2% vaccine efficacy was demonstrated in the vaccine arm using this regimen (23). Interestingly, in a *post hoc* analysis of the protection data, it was observed that should the study have stopped after the first year of implementation, vaccine efficacy would have been as high as 60%, indicating a non-sustained protection from acquisition was achieved. In terms of the immune correlates of protection a weak neutralizing antibody response was again observed; however, non-neutralizing plasma IgG binding antibodies to the top of the Env V1/V2 variable loop were associated with protection (24).

#### NANOPHARMACEUTICAL-BASED VACCINES

Soluble gp120 monomers and more advanced trimeric forms of Env failed to elicit the protective responses necessary to prevent HIV acquisition. Thus, a critical goal in HIV vaccine design has been to try and understand why this has been the case. Collectively, the knowledge that (1) infused neutralizing antibody protects against viral challenge in non-human primates (NHP) studies, (2) a percentage of individuals will generate neutralizing antibody responses against the autologous infecting virus, and that (3) highly potent neutralizing antibody can be generated in some individuals, solidifies the notion that an anti-HIV B cell response can be protective. As neutralizing antibodies are directed against the surface glycoprotein, presentation of natural Env structures, mimicking the native viral Env is deemed imperative to eliciting a neutralizing immune response. Resolving this scientific challenge has led to re-engineering of the Env itself to achieve a native conformation. As a result, numerous antigenic structures have been developed, including germline revertants targeting the N332 supersite and near native trimers candidates such as BG505 SOSIP.664 (25–27). These new immunogens, designed to harness anti-HIV neutralizing antibody responses can elicit autologous neutralizing antibody responses and depending on how they have been utilized, can elicit heterologous Tier 2 neutralizing antibody. However, these and other such candidate vaccine antigens do require mutations in their amino acid sequences to increase stability and this can come at the expense of losing antigenic conformation. For reviews more focused on HIV-1 Env-based vaccines, several recent articles have been published which may be of interest to readers (28, 29).

#### Virus-Like Particles (VLPs)

To overcome these afore-mentioned issues in antigen design, Env trimers have been expressed in lipid membranes. This enables their conformations to more accurately resemble native and functional Env spikes found on infectious virus. Taking this into consideration, nanoparticle platforms such as VLPs provide the ideal technology to express HIV Env immunogens. VLPs are noninfectious and genomeless multiprotein structures that mimic the conformation of viral proteins in their natural environment. These self-assembling molecules can generate ordered arrays of polypeptides that come together to form the VLP. The resulting repetitive geometry offers maximal VLP–host interactions due to increased avidity (30). As a result, VLPs can present viral antigens in their authentic conformation and maximally stimulate resulting immune responses (31).

The use of VLPs as potential vaccine candidates offers numerous advantages over traditional strategies. VLPs have the unique ability to present viral Env spikes in their natural conformation and, therefore, the potential to elicit neutralizing antibody responses against HIV. This feature is often lost when antigens are purified from pathogens or when pathogens are rendered inert through chemical or heat inactivation. VLP vaccines are also cheaper than subunit vaccines since less of the vaccine can be administered without impairing the resulting immune response. Another attractive property of VLP vaccines are their particulate structure, which allows them to be efficiently taken up by antigenpresenting cells (APCs) and stimulate strong humoral and cellular immune responses (32, 33) since they can be presented on both MHC class I and class II molecules (34, 35). In addition, their polyvalency allows for efficient B cell receptor (BCR) crosslinking and activation.

Although several commercially available VLP vaccine formulations are available to protect against Hepatitis B, Hepatitis E, and human papillomavirus (HPV), none exist for HIV (**Table 1**). One of the prominent prophylactic VLP vaccine formulations against the Hepatitis B virus is RECOMBIVAX®. It mainly consists of Hepatitis B surface antigen that spontaneously assembles into ~20 nm lipid-containing particles (36). The major prophylactic vaccine against the hepatitis E virus is the Hecolin® vaccine, whose major constituent is a small portion of the Hepatitis E capsid protein that self assembles into 20–30 nm particles (36). Gardasil® is a major prophylactic VLP vaccine formulation against HPV. It is synthesized by expressing HPV L1 capsid protein, which spontaneously assembles into immunogenic noninfectious VLPs that can induce the same neutralizing antibody responses as the native virions (37).

There are five main platforms that are currently utilized to produce VLP vaccines: bacteria, yeast, insect cells, mammalian


*IM, intramuscular.*

cells, and plants. Although leveraging bacteria to produce VLPs is the most scalable and cost-effective approach, they are unable to perform the essential post-translational modifications for optimal immunogenicity in human (31). In addition, the use of bacteria may introduce contaminating endotoxins into vaccine formulations. Yeast are also scalable and cost-effective approaches to producing VLPs; however, unlike bacteria, they can perform eukaryotic post-translational modifications such as glycosylation. Insect cells can be induced to produce VLPs through baculovirus expression systems. This system can produce large quantities of VLPs in their natural conformation but must be purified from baculovirus contaminants. In fact, these components may obscure the immune response against the epitopes of the VLP (38). VLP production through mammalian cells cannot be scaled up as efficiently as bacteria, yeast, or insect expression systems. In addition, the use of mammalian cell systems runs the risk of human viral pathogen contamination (39); however, they have the most accurate and complex post-translational modifications (40). Therefore, this system is particularly advantageous for constructing complex VLPs. Plants can be induced to produce VLPs through *Agrobacterium tumefaciens*, a Gram-negative soil-based bacteria used to transform plant cells (41). Although plant-based expression systems are easy to scale up and contain no human-derived viral contamination, these VLPs cannot undergo the same post-translational modifications that are done in mammalian cells, thus reducing immunogenicity.

The HIV-1 Env trimer is synthesized as a 160 kDa precursor and is processed by protease to yield a surface and transmembrane subunit during its passage through the secretory pathway. The resulting surface bound oligomer comprises three non-covalently associated gp120 and gp41 subunits. The surface gp120 is responsible for binding host cell target receptors (CD4) and co-receptors (CCR5) whereas the gp41 anchors Env in the membrane. Taking the well-characterized influenza virus as a model system shows that the virus may require only 2–3 trimers to engage with its cellular receptors to create a pre-fusion structure. However, it is believed that 6–8 trimers may be necessary to build the stable fusion pore (42). This question has been hard to address in the context of HIV-1 for a number of reasons including the fact that (1) a large percentage of HIV virions (>99%) are replication defective, (2) the sparsity of intact Env trimers on the viral surface (7–14 spikes/virion), (3) spontaneous shedding of gp120 from Env complexes, and (4) heterogeneity among the HIV-1 Env complexes (42). This has resulted in a number of conflicting studies suggesting a single trimer is necessary (42) or perhaps between five and eight trimers may be necessary for infection (43, 44). More recently, work done by the Trkola group has suggested divergent HIV strains differ in their stoichiometry for entry, requiring ~1–7 trimers, with the majority of HIV strains requiring 2–3 trimers for infection (45). In the context of eliciting humoral immune responses within the host, the surface Env density is important for efficient BCR crosslinking, leading to B cell clonal expansion and antibody affinity maturation. Unfortunately, HIV's low density of Env spikes (7–14 spikes/virus) is substantially lower than that seen for other viruses, such as influenza (400–500 spikes/ virus), vesicular stomatitis virus (VSV, 1,200 spikes/virion), Rous sarcoma virus (RSV, ~118 spikes/virion), and even compared to the related SIV (~70 spikes/virion) (46). Therefore, attempts to increase surface HIV Env density on VLPs may facilitate more protective immune responses. Using fluorescence activated cell (FACS) sorting to isolate producer cells that are recognized by trimer cross reactive broadly neutralizing antibody, Stano et al.*,* sought to isolate and propagate producer cells displaying higher levels of antigenically correct Env (46). Following multiple rounds of FACS sorting, monodisperse VLP formulations were generated that had a 10-fold higher Env density per particle (~127 spikes/particle) than control viral particles. When the Env was sequenced, a truncation in the C-terminal tail of Env was seen. Although the high-density Env particles displayed a greater infectivity compared to normal pseudoviruses, they exhibited similar levels of neutralization sensitivity (46). However, the high Env density VLPs were superior activators of broadly neutralizing VRC01 B cells as determined by upregulation in CD69, IL-6, and TNF-a production and BCR downregulation. This collectively indicates superior BCR clustering and activation (46).

One of the major roadblocks to an effective HIV-1 vaccine is the sheer diversity of the viral swarm. A preventative vaccine must be protective against HIV strains/subtypes circulating in multiple geographic regions. Polyvalent Env vaccines are one such approach to overcome the viral diversity and involve the use of heterogeneous Env mixtures to expose developing immune responses to a multitude of Env conformations (47). This is thought to help promote immuno-focusing of B cell responses onto more conserved regions of vulnerability such as the Env CD4-binding site and MPER while also promoting antibody breadth. To address this issue, Pankrac et al. developed a chimeric HIV-1 VLP formulation capable of accommodating near full length (NFL) genomes of HIV-1 that captures the viral diversity within infected individuals (47). To do this, the authors took the plasma of five pre-cART HIV positive volunteers and engineered the isolated virus into a mixed VLP formulation. By inserting mutations into integrase and a series of substitutions into the RNA packaging element, the resulting VLPs had dramatically reduced genomic RNA and lost the functional ability of viral integrase. They also deleted the 5′LTR so that the probability of reverse transcription was eliminated. Collectively, these "dead" viral particles were shown to have similar morphology to wild-type virus, contained p24 and p17, as well as expressed functional Env, Tat and Rev (47, 48). Upon exposure of HIV-1 infected CD4 T cells to dendritic cells (DCs) pulsed with the heterologous VLPs, a strong IFN-g response was detected, indicating the ability to generate antigenspecific CD4 T cell recall responses. In addition, when the VLPs were used to pulse PBMC cultures, the authors demonstrated Granzyme B production, a biological marker for cytotoxicity (47). Finally, the VLPs were also shown to be able to prime and boost a CD4 adaptive immune response *in vitro* using healthy donor CD4 T cell-DC co-cultures (47). This polyvalent VLP vaccine formulation is called ACT-VEC and is now being evaluated in non-human primates (NHPs) for its immunogenicity.

Trying to understand the host and viral factors that influence HIV infection and how the virus crosses the mucosal barrier might lead to more efficacious HIV vaccines. This has led to the characterization of the phenotypic properties of transmitted/ founder (T/F) viruses to see if transmission is through stochastic events or whether the genetic bottle neck selects for viruses with certain properties. Features associated with transmitted viruses include CCR5 tropism, short variable loops and less N-linked glycan residues. With such distinguishing features in mind, McClure et al. tested the antigenicity and immunogenicity of DNA and modified vaccinia Ankara expressed VLPs, displaying native forms of T/F clade C Envs (49). The hypothesis being that the VLPs expressed Envs with or without administered gp120 might elicit a broadly neutralizing antibody response. This interesting study proceeded investigations into the clade C infection occurring within an infected individual that eventually developed a broadly neutralizing antibody response against the CH103 CD4-binding site (50). Significantly, the VLP expressing vectored immunogens boosted with gp120 were immunogenic in NHPs and despite a complex regimen elicited a neutralizing antibody response in 50% of the macaques tested, with antibody directed toward the CD4-binding site. Thus, demonstrating that a VLP prime plus gp120 boost using a T/F Env to be capable of inducing autologous Tier 2 neutralizing antibody to the CD4-binding site (49).

#### Liposomes

Antigen delivery by liposomal vesicles was demonstrated over 30 years ago when unilamellar vesicles were generated that were composed of phosphatidylserine. These liposomes contained poliovirus and purified poliovirus RNA. They were found to be infectious and could be potentially utilized as a delivery agent for biological macromolecules (51). Liposomes can either be microparticles or colloidal carriers (52), with a diameter ranging from 20 nm to more than 10 µm (53). There are many types of liposomes (e.g., multilamellar vesicles and large or small unilamellar vesicles), with the conventional ones composed of biodegradable and non-toxic zwitterionic phospholipids, such as phosphatidylcholine, phosphatidylserine, and cholesterol (52–54). The formation of liposomes is a spontaneous occurrence when lipids are subjected to hydration (52), enabling the formation of lipid bilayers surrounding an aqueous core. The first efforts at using liposomal technologies for delivery of candidate HIV vaccine antigens occurred nearly 25 years ago by Bui et al. (55). In this first attempt, the authors utilized a non-glycosylated, denatured gp120 called HIV Env-2-3-SF2 as the vaccine immunogen, and co-formulated the antigen with either alum or liposome. The HIV Env-2-3-SF2 alum and HIV Env-2-3-SF2 liposome immune responses were then potentiated using liposomal IL-7. The result was the demonstration that liposomal HIV Env antigens elicited stronger antibody titers than the alum formulations and that the responses could be enhanced using liposomal IL-7. Notably, cytotoxic T lymphocyte (CTL) responses were greater in the liposomal arm than the alum treatment regimen (55).

Interbilayer-cross-linked multilamellar vesicles (ICMV) with His-tagged gp140 trimers anchored onto the surface of the delivery agent using Ni-NTA functionalized lipids were recently explored (56). The aim being to deliver a high quality, B cell triggering-gp140 trimer immunogen, previously demonstrated to elicit high titers of potent cross clad neutralizing antibody responses (57). The ICMV-gp140 nanoparticles were adjuvanted with TLR4 agonist MPLA and used to assess humoral immune response induction in mice and compared to gp140 adjuvanted with SAS (oil in water adjuvant containing MPLA trehalose-6,6′-dimycolate). The trimer-loaded ICMV particles were superior inducers of humoral immune responses compared to gp140 alone or adjuvanted with SAS, with the breadth of the anti-HIV antibody response increased in response to ICVM surface anchoring—including against the conserved MPER sequence of gp41. Unfortunately, due to the use of the murine model, the authors did not test neutralizing antibody production (56).

In a separate study, Ingale et al.*,* utilized single bilayer liposomes, displaying well-ordered, high-density trimers, such as JRFL-SOSIP, and JRFL-NFL, for better B cell activation (58). The trimers were conjugated to the surface of the liposomes, by electrostatic interaction using the cholesterol substitute, DGS-NTA(Ni), and c-terminal His-tagged HIV-1 trimers. As a result, the liposomal formulation containing the surface trimers more efficiently stimulated B cell germinal center formation compared to soluble trimers demonstrating the necessity for membrane display. Furthermore, there appeared to be a suggestion that the liposomal trimers promoted stronger antibody titers to the native trimers and to generate low level tier 2 homologous neutralizing antibody (58). As antigens coupled to the surface of liposomes are exposed to interstitial lymphatic fluid as they transit into the draining lymph nodes, the Bale et al.'s study suggested that it might be possible for surface displayed antigen that are non-covalently bound to be susceptible to dissociation from the liposome prior to B cell contact (59). Therefore, the authors created a secondgeneration liposomal formulation where HIV-1 Env trimers were coupled to synthetic single layer liposomes by alternative means. Here, the authors found that maleimide-thiol covalent coupling of trimers to liposomes elicited higher anti-HIV Env IgG antibody responses than soluble trimers or trimers coupled to liposomes *via* non-covalent metal chelation. Furthermore, the liposomescoupled antigens were better activators of B cells as determined through Ca2+ flux measurements. Finally, murine immunogenicity studies demonstrate that the covalent coupling of antigens enabled the expansion of B cell germinal center reactions, over and above what was seen with soluble antigens, and that they also blocked the access of B cells and antibody to the c-terminal end of the exposed Env spike, suggesting that the trimers were intact at the point of immune recognition and adaptive immune activation (59). Taken together, this suggest that the particulate display of high-density antigen arrays on the surface of liposomes elicits improved B cell responses over previous soluble versions of the same trimers.

A promising vaccine candidate has been the membrane proximal region (MPER) of Env. This gp41 region is the target for several neutralizing antibodies and is attractive as a vaccine candidate due to its conserved and linear nature. A significant drawback to MPER immunogens is its inherent lack of immunogenicity. Recently Hanson et al. utilized 150 nm liposomes as membrane display vehicles to promote a more native vaccine antigen conformation (60). In this example, the authors evaluated the strength and durability of humoral immune responses elicited by MPER peptides anchored to the surface of liposomes *via* palmitoyl tails. The liposomal anchored peptides were compared to alum and complete Freund's oil-based emulsion adjuvant. Liposomal anchored peptides successfully elicited MPER-specific IgG antibody while alum and Freund's adjuvants did not. Critically, anti-MPER antibody responses could be augmented through the inclusion of adjuvants, such as TLR4 agonist (MPLA), TLR9, or STING Interestingly, intrastructural help through helper epitopes promoted IgG responses to MPER without increasing B cell responses against the help sequence (60).

#### Polymeric Nanoparticles

Polymeric nanoparticles are very promising candidate delivery systems and adjuvants for a range of vaccine antigens. Their favorability stems from their ease of synthesis, biocompatibility, their biodegradable nature, the fact they are non-immunogenic, non-toxic and fairly inexpensive. Numerous different polymeric nanoparticle exist but the most frequently encountered types include chitosan, poly(lactic-co-glycolic acid) (PLGA) and poly(lactic acid) (PLA) (**Table 2**).

#### Chitosan (Poly(d-Glucosamine))

This cationic polysaccharide, polymeric nanoparticle is derived from the deacetylation of chitin, a naturally occurring polymer found in the cuticles of insect species and crustaceans, such as crabs and shrimp. Due to its strong immune-stimulatory properties and low immunogenicity, chitosan has frequently been utilized as an antigen delivery system. A significant advantage of chitosan to conventional adjuvants such as alum is its ability to promote TH1 immune responses and to also act as both delivery system and adjuvant when applied to mucosal surfaces.

In 2016, Cosgrove et al., published the first comparative Phase 1 clinical trial investigating the safety and immunogenicity of three HIV-1 clade C gp140 vaccinations delivered by either the IM, intranasal (IN), and intravaginal routes in women (62). The vaccine antigens were co-formulated with either glucopyranosyl lipid adjuvant (GLA), chitosan nanoparticles or an aqueous

Table 2 | (A) Liposome and polymeric particle vaccines; (B) liposome and polymeric particle vaccines against human immunodeficiency virus (HIV)/simian immunodeficiency virus (SIV).


*ICMV, interbilayer-cross-linked multilamellar vesicles; Ni-NTA, nickel-nitrotriacetic acid; MPLA, 3-O-deacyl-4*′*-monophosphoryl lipid A; SAS, Sigma Adjuvant System; PLGA, poly(d,llactic-co-glycolic acid); CLDC, cationic liposome-DNA complex.*

gel, respectively. Within this study, chitosan was ineffectual at promoting humoral immune responses despite the evidence from previous clinical studies, suggesting that chitosan increased potent antibody responses to vaccine immunogens. Interestingly, the authors noted that gp140 co-formulated with chitosan and intranasally delivered resulted in induction of CD4 T cell responses and upon IM boosting, all individuals generated a detectable serum antibody response. This suggests, that the chitosan-gp140 mucosal vaccination primed for an anamnestic serological response. However, it is unclear as to why the priming was restricted to CD4 T cells or if it had indeed primed B cells, but the responses were too low to measure (62). How nanoparticulate chitosan acts as adjuvant is currently unknown; however, recent evidence suggests that it promotes DC maturation through type 1 interferons and increases antigen-specific T cell responses in a type 1 IFN receptor-dependent manner (66). The latter requiring cytoplasmic sensors cGAS and STING, as well as both mitochondrial reactive oxygen intermediates and cytoplasmic DNA (66).

Chitosan has also served as a formulation adjuvant by promoting penetration enhancement of formulations, enabling their uptake and increasing the bioavailability of vaccine antigens when applied topically to mucosal surfaces. This was clearly demonstrated by Klein et al., who compared various polymeric penetration enhancers to promote trans-mucosal delivery of trimeric CN54gp140 protein (63). In this study, chitosan increased the systemic antibody levels of both IgG and IgA anti-gp140 antibodies following IN and sublingual delivery and increased antigen-specific antibody responses in the vagina of intranasally vaccinated mice. It is clear in this study that various polymeric formulations had differing abilities to augment immune responses and that the route of delivery was also a significant factor (63).

#### PLGA (Poly(d,l-Lactic-co-Glycolic Acid)) Nanoparticles

The use of PLGA as a nanoparticle vaccine delivery system has been studied comprehensively over the last two decades, resulting in numerous publications. However, it is the ability of PLGA nanoparticles to entrap bacterial toxoids/antigen or surface display antigens and induce long-lasting immune responses in small animal models which makes them an attractive delivery system (67–71). Early attempts at using PLGA formulations for delivery of recombinant HIV gp120 were centered on the ability of PLGA to release antigens over an extended period, promoting continuous antigenic exposure and immune education (72). These PLGA vaccines were designed to be single shot vaccines, which provide a pulsatile release of contained antigen and QS-21 adjuvant at intervals of 6 months. The resulting antibody response caused neutralization against the matched strain of HIV-1 and providing evidence for PLGA to be a slow release system (72). Since these early attempts, sophisticated attempts at using PLGA formulations as an HIV vaccine have been made (**Table 2**). For instance, the HIV-1 p24-Nef peptide was chemically conjugated to TLR5 agonist FLiC and co-formulated with PLGA nanoparticles (64). Upon intradermal vaccination in mice, the combination of TLR agonist and nanoparticle was shown to increase vaccine immunogenicity by increasing lymphocyte proliferative responses reducing the immunogenic dose required (64). In another example, Rubsamen et al. utilized PLGA encapsulated TLR9 agonist, CpG, and TLR4 agonist Monophosphoryl lipid A (MPLA) in the injection solution to augment immune responses to the delivered HIV CTL epitopes (65). The formulations resulted in immune responses against multiple splenocyte CTL epitopes as measured by IFN-g release (65).

#### NANO-VACCINE TRAFFICKING AND INTERACTIONS WITH HOST IMMUNE CELLS

The immune response to nanoparticle formulations such as liposomes, poly(d,l-lactide-co-glycolide) (PLGA), and VLPs are known to elicit both arms of the immune response. Understanding the cellular events involved in such vaccines are critical for their development as vaccine strategies and advancing test formulations to human clinical trials. Specifically, the molecular and cellular interactions occurring between nanoparticles and the vaccinated host immune system (i.e., effects on DCs, T and B lymphocytes, and lymphoid tissues) will be essential. Although many alternative vaccination routes have been explored in animal models and in humans, most vaccine studies employ IM vaccination. Intradermal (ID) vaccination demonstrates strong immune reactions to vaccinating antigens in a dose sparing manner compared to IM vaccination; however, ID vaccination is associated with a higher frequency of undesired side effects. Thus, we will concentrate on nanoparticle vaccines administered *via* the IM route.

The skeletal muscle system is one of the largest cellular compartments within the human body. Muscles are composed of large multinucleated syncytial muscle fibers which generate the necessary mechanical forces for locomotion. Following an injury to the muscle fibers, the surrounding satellite cells become activated and proliferate. The resulting myeloblasts fuse to become multinucleated myotubes, which will then go on to differentiate into mature muscle fibers. Under non-inflamed conditions, only a very few resident immune cells have been described to be present in the muscle (73). These include APCs such as macrophages and DCs; however, during pathological conditions, other immune cells such as monocytes, neutrophils, and lymphocytes can infiltrate the tissue (74). The introduction of vaccine antigens such as nanoparticles with adjuvants into the muscle environment induce transient inflammatory reactions at the delivery site, resulting in immune cell infiltration (74). Muscle cells can respond to the inflammatory mediators that are secreted by neighboring myocytes as well as resident and infiltrating immune cells. Myocytes express a range of receptors for cytokines, such as IL-1, IL-6, and IFN-g. Furthermore, they express chemokine receptors, such as CCR2, CCR4, and CCR10, which enable them to participate in surveying the local environment (75). The introduction of nanoparticles into the musculature enables their uptake by APCs such as DC and macrophages that are in the epimysium and perimysium interstitial spaces surrounding the entire muscle and muscle fascicles (76). The APCs produce pro-inflammatory mediators and chemokines that activate the myocytes and attract circulating immune cells and migrating APCs into the tissue. In support of this Gao et al. HIV Nanoparticle Vaccines

pro-inflammatory role of myocytes within the vaccinated muscle was the research published by Mosca et al. They used microarray and immunofluorescence analysis of the murine muscle after oilin-water emulsion MF59 adjuvant, CpG, or alum injection (77). While CpG and alum induced time-dependent changes in 387 and 312 genes, respectively, MF59 induced changes in 891 genes, with recruitment of MHC II and CD11b cells within the injection site. Interestingly, the early response proteins penetraxin 3 and JunB were induced upon MF59 delivery, indicating that MF59 directly activated muscle fibers (77). Furthermore, IM injection was shown to be associated with release of ATP from local muscle fibers, which was important in the recruitment of immune cells to site of vaccination and the magnitude of the resulting immune response (78).

Neutrophils and monocytes have been found to be the first cells to infiltrate vaccinated muscle. This infiltration is normally detected within 3–6 h post-delivery. DCs are essential to adaptive immune response, and it is, therefore, not surprising that their recruitment is also rapid. These APCs take up exogenous nanoparticles, become activated, and in the case of DCs, migrate toward the draining lymph nodes. During this transit, the DCs upregulate co-stimulatory molecules, such as MHC II, CD80/86, and CD40. They also secrete cytokines and lose the ability to phagocytose antigens in preference for increased presenting capabilities. Once in the draining lymph nodes, the DCs initiate the adaptive immune response through activation of T and B cells. Remarkably, it is now becoming apparent that upon vaccination, intact nanoparticles such as VLPs can enter the draining lymph node without the requirement for DC-mediated transport (79). Using a mouse model, Cubas et al., demonstrated that IM vaccination with SHIV VLPs efficiently trafficked to the draining lymph nodes, although some speculation arose regarding VLP nanoparticle uptake by blood capillaries in the muscle and trafficking to the spleen (79). Nevertheless, once the VLPs were in the lymph node, the intact SHIV VLPs were detected in the subcapsular sinus (79). Once antigens enter the lymph node, they have been shown to reside there for extended periods of time. In the case of HIV-1, follicular DCs have been shown to retain non-degraded, viral particles in the form of immune complexes on their dendrites for months (80), thereby promoting germinal center reactions; however, it is unclear how long the SHIV VLPs were detectable in this instance.

Many factors, including, size, shape, charge, and receptor– ligand binding potential of nanoparticles dictate their distribution and cellular uptake upon vaccination, and therefore impact prevailing immune responses. VLPs are typically 20–150 nm in diameter, and excluding charge and receptor tropism, the size of the nanoparticle plays an important role in uptake and immunity. Generally, speaking nanoparticles and VLPs between 25 and 40 nm can effectively penetrate tissues upon delivery (81) and must exceed 10 nm to escape renal filtration. Nanoparticle vaccines exceeding 500 nm are likely taken up by APC at the injection site while nanoparticles and VLPs below 200 nm are mainly internalized by DCs, and in the free, intact, form will more likely accumulate in the liver and secondary lymphoid structures (82).

Over the years, numerous nano-vaccine delivery technologies have been investigated for optimal immune priming by examining the size of the nano-formulation (83–85). Using a combination of electron microscopy and dynamic light scattering, Mann et al., revealed that orally delivered, low diameter lipid-based vesicles (10–100 nm), containing influenza vaccine antigen, elicited a TH2 biased immune response in mice. This contrasted with similar vesicle compositions with an average diameter of 980 nm, which had a significantly greater TH1 bias (86). This work was supported by other previous research using lipid-based nanoparticle formulations delivered parenterally, where small lipid-based nano-vaccines (<155 nm) promoted TH2 responses while larger vesicles (>225 nm) elicited TH1 biased responses (87). It has been suggested that the TH1/TH2 polarizing effects related to nanovaccine size might be due to efficiency of antigen presentation by the phagocytosed nano-vaccines by the APCs. An interesting manuscript by Brewer et al. provided evidence for altered trafficking of internalized antigens dependent on delivery systems size. Using murine macrophages, the authors described the trafficking of antigen loaded, large particles (560 nm), into early endosomelike phagosomes, whereas the smaller particles (155 nm) and soluble antigen were rapidly localized to endo-lysosomes (88). In this case, MHC II was detected in both compartments regardless of the TH1/TH2 bias. The internalization of large vesicles was also found to be mediated by actin-dependent phagocytosis, while the smaller vesicles were taken up in an actin-independent manner (88). These findings taken together suggest that nanoparticle size could significantly influence not only vaccine antigen uptake by APCs, but also influence their intracellular trafficking potential, dissemination around the body, and the ultimate immune outcome.

#### STRATEGIES TO INCREASE IMMUNOGENICITY OF NANO-VACCINES

One of the basic tenants of immunology states that antigens can induce immune responses, if it is perceived as being "foreign" by the host. However, non-self-antigens do not always trigger immune responses. Even if they do, they can be poorly immunogenic. For the immune system to become sufficiently mobilized, it must identify some sort of "danger." This can occur in a number of ways including, (1) the recognition of exogenous signals by innate immune cells through pathogen-associated molecular patterns (PAMPS) and (2) being activated in response to recognition of danger-associated molecular patterns, which are endogenous molecules released by damaged or perturbed tissues/cells (89). Such immune stimulatory signals are often applied to vaccines in the form of adjuvants (90–92). Adjuvants are defined as agents which act to enhance the quantity and quality of immune responses to co-formulated immunogens. Mechanistically, adjuvants work by either simply arranging, aggregating, transporting, shielding antigens, or by stimulating pro-inflammatory signaling pathways conducive to enhancing immunogenicity. In other words, adjuvants increase the strength and effectiveness of vaccines. As an example, live attenuated vaccines do not necessarily require any additional help in generating immune responses as they exhibit low level infections. In contrast, highly purified subunit and inactivated vaccines are often poorly immunogenic and may not sufficiently activate innate immunity and generate the necessary pro-inflammatory environment that is conducive to the production of protective responses. Nanoparticle formulations such as VLPs may also display poor immunogenicity, compared to live attenuated vaccines and are therefore often delivered with an adjuvant to maximize an efficient immune response. Over the years, numerous adjuvant technologies have been applied to nanoparticle formulations. In an attempt to attain more and more protective responses and to also overcome potential side effects associated with highly immune-stimulatory adjuvants, mixed adjuvant regimens have gained in popularity. Recently, a number of excellent reviews have been published concerning adjuvants for HIV-1 vaccines (93, 94), so we will discuss only a few of the more common adjuvants below, some of which are detailed in **Table 3**.

#### Alum

Alum is one of the first adjuvants to be developed for augmenting vaccine elicited immune responses having first been demonstrated to act as an adjuvant by Glenny et al., in 1926. Since then, aluminum adjuvants have become the most widely utilized adjuvant and have been used in diphtheria, tetanus, pertussis and poliomyelitis vaccines in many countries for more than 70 years. Alum salts, form polydisperse crystalline particles (~2–10 μm in diameter) capable of adsorbing subunit vaccines onto their surface, thereby increasing BCR crosslinking, presentation to APCs, and can act to form an antigen depot at the vaccination site (95). Critically, alum adjuvants have been described to be potent inducers of IL-1 secretion *in vitro* within APC population, such as DCs and macrophages via NLRP3 inflammasome activation (96). While Alum has been used as an adjuvant in licensed HPV, HBV, and HEV VLP vaccines, how it functions as an adjuvant in the context of VLP vaccines is not clearly understood. Certainly, VLPs are readily phagocytosed by APCs due to their particulate nature. Furthermore, they are decorated with vaccine antigens on their surface, which enables BCR crosslinking. Thus, the adjuvant effects of alum, in the context of these VLP vaccines, may be due to antigenic depot effects and activation of the inflammasome.

The use of alum to augment antiviral immunity has been evaluated in the context of therapeutic HIV vaccines. Although therapeutic vaccinations are now aimed at a sterilizing or functional cure, initial attempts in the 1990s and early 2000s were made to therapeutically "modulate" immune responses to curb CD4 T cell decline and lower the rate of AIDS progression (48, 104). As decreasing anti-p24 responses had been correlated with clinical progression to AIDS, a p24-based vaccination was developed as a potential method to elevate p24 immune responses in HIV-infected individuals. In a randomized placebo-controlled trial that enrolled 304 HIV-infected individuals (CD4 counts <350 cells/μl), individuals were administered monthly IM injections of either (1) alum, (2) p24 VLP (500 μg) + alum, or (3) p24 VLP (1,000 μg) + alum (105). The volunteers maintained ART during the study which consisted of Zidovudine (ZDU) with or without didanosine or zalcitabine. CD4 T cells counts were monitored for the 6 months of treatment and then for an additional 6 months of follow-up. The study failed to demonstrate any benefit of immunization with p24 VLP despite adjuvanting with alum. Interestingly, no statistically significant antibody responses were detected against p24 or p17 but instead were detected against the carrier protein, Ty. Furthermore, a decline in CD4 T cells was recorded over time with no statistically significant difference between treatment regimens (105). In a similar follow-up


*This list is not exhaustive; it covers the more commonly used adjuvants. AS01 [used in HIV vaccines (102)], AS02, AS03, and CAF01 [used in HIV vaccines (16, 103)] among others are not on the list.*

*LPS, lipopolysaccharide; GLA, glucopyranosyl lipid adjuvant; AF, aqueous nanosuspension; SE, squalene-based oil-in-water nanoemulsion; Poly-IC, polyinosinic–polycytidylic acid; DC1, type 1-polarized dendritic cells.*

study, 61 individuals were recruited and split into three treatment arms and received either (1) ZDU + IM alum, (2) ZDU + IM p24 VLP (500 μg) + alum, or (3) placebo capsules + IM p24 VLP (500 μg) + alum (106). This was conducted to evaluate the therapeutic effect of p24 VLP vaccination with ZDU on p24 antibody production and CD4 T cell counts, as well as any impact on viral loads. Despite the treatment being well tolerated, no significant differences between the treatment arms were recorded in any of the antibody and cellular immune responses measured (106).

Alum has also been used as an adjuvant for VLPs used for prophylactic vaccine strategies for prevention of HIV acquisition. Insect cell expression systems have been used to generate chimeric VLPs that express the V3 domain and a linear section of the discontinuous CD4-binding domain of gp120 within gag (107). Subsequent vaccination of rabbits with the various chimeric VLP constructs without adjuvant generated antibody responses to the Pr55 gag carrier component. Interestingly, most of the antibody response was specific to the gag insertion site. Despite generating weak neutralizing antibody responses, vaccination of mice with the different non-adjuvanted chimeric VLPs could generate strong MHC class I CTL responses. However, when the recombinant antigen was adsorbed with alum, its potential for CTL induction was drastically reduced or inhibited (107). Further studies in NHPs using two differing HIV-1 virus-like Pr55 gag VLP vaccine constructs without adjuvant revealed that IM vaccination with both types of VLP induced elevated titers of anti-gag antibody responses, but it was Pr55 gag/env VLPs expressing the full gp120 at the outer surface of the particle, which generated substantial anti-Env antibody titers with a neutralizing phenotype (108). Furthermore, both Gag and Env cytotoxic T lymphocyte (CTL) responses were induced (108).

Collectively, the literature presented suggests that despite Alum's positive influence in some preclinical experiments and in some clinically approved VLP formulations, its inappropriate use can diminish the effects of test vaccines, resulting in ineffectual immune responses. Thus, inclusion of alum as a VLP adjuvant in any preclinical or clinical trial must be scrutinized.

#### Pattern Recognition Receptor (PRR) Agonists

The most highly characterized PAMP sensors are the innate immune system PRRs for which several families have been classified. The C-type lectin toll-like receptors (TLR) are the best-defined families of PRRs. They are expressed by a plethora of mammalian cells as either endosomal or plasma membrane bound receptors. Cell surface TLRs (4, 1/2, 2/6, 5) detect conserved microbial patterns present on the cell surface such as LPS and flagellin, while endosomal TLRs (3, 7, 8, 9) detect microbial genomic material such as double-stranded RNA and DNA and single-stranded RNA (109). Detection of bacteria or viral particles by the various TLRs expressed by APCs, such as DCs, macrophages, and B cells, is normally followed by endocytosis/phagocytosis and subsequent processing and presentation to T cells *via* MHC. The MHC presentation to T cells occurs in the context of several signals, including, co-stimulatory molecules, cytokines, and by TLR ligation. Collectively, this stimulates T cells to surpass the minimal activation thresholds for naïve T cells to partake in immune responses. Due to their abilities to activate innate signaling pathways, PRR ligands are an expanding field of study for HIV vaccine design. This includes the use of TLR4 (MPL) and 7/8 (R848) agonists encapsulated in PLGA particles admixed with VLPs expressing SIV gp160 and Gag (110), the incorporation of TLR7/8 (Resiquimod) and 9 (CpG) into oil-in-water emulsions for responses against HIV-1 Env gp140 (111) and the use of TLR3 (Poly IC:LC, Hiltonol), TLR4 (MPL derivative, E6020) and TLR 7 (a proprietary benzonaphthyridine from Novartis) agonists to characterize vaccinemediated evolution of Env-specific B cell ontogenies (112). Normally, the microbial antigen and TLR ligand are physically associated (i.e., contained within the same pathogenic organism), and therefore are co-delivered to the same phagosomal/ endosomal compartment within the APC. During vaccinations, PAMPs are generally admixed with antigens and, therefore, may not be associated as having a common derivation, unless the PAMP and antigen enter the same endosome. The latter requiring large excesses of PAMPs over what is required for APC and T cell activation (113). Therefore, formulating TLR ligands onto the surface or within nano-vaccines serves as a promising way to optimally stimulate APCs. The TLR7 agonist Imiquimod has been approved for topical delivery in humans for genital warts as a 5% cream (Aldara™) (114). However, its systemic application has been limited because of toxicity. Due to its hydrophobic nature, entrapment within polymeric nanoparticles offers an alternative strategy for imiquimod delivery, greatly reducing its toxicity. A study by Jimenez-Sanchez et al., reported on polylactide (PLA)-based micelles with entrapped TLR7 and HIV-1 gag p24 decorating the surface of the nanosystem. As would be expected, encapsulated imiquimod induced stronger DC stimulation than free imiquimod with encapsulated ligand triggering NF-κB and MAPK pathways (114). In another study, Francica et al. immunized NHPs with HIV gp140 Env with several adjuvants. When the authors co-formulated gp140 with TLR4/7 agonists and alum, they revealed that anti-Env antibody titers were boosted 3- to 10-fold higher than antibody titers induced with alum and gp140 alone. Interestingly, while TLR4 and alum enhanced a set of inflammatory genes, TLR7 suppressed alum-specific inflammatory genes in preference for interferon-responsive genes (115). This suggests that different TLR ligands can have different transcriptional effects against co-formulated adjuvants.

#### MPL and GLA

The detoxified TLR-4 ligand, 3-*O*-deacyl-4′-monophosphoryl lipid A (MPLA), is a lipopolysaccharide (LPS)-derived, heterogeneous blend of varying length MPL molecules from *Salmonella minnesota R595*. The MPLA extraction and treatment procedure resulted in three distinct modifications compared to the parent molecule: (1) the removal of the core polysaccharide containing the O-antigen, (2) the removal of one phosphate group, and (3) one fatty acid chain (116, 117). MPLA has been shown to promote TH1 immune responses and to have a favorable safety profile when compared to LPS (118–120). Since the identification of MPLA as an adjuvant, newer generations of TLR-4 agonists have been developed, such as a synthetic lipid A-like molecule called GLA. GLA adjuvant is either administered in aqueous nanosuspension (AF) or more often, co-formulated with a squalene-based oil-inwater nanoemulsion (SE). GLA-SE is now a clinical stage vaccine adjuvant that also promotes T-bet-dependent TH1 immune responses and enhances protection against a range of intracellular pathogens (97). In fact, it is unsurprising since most TLRs use MyD88 except TLR3, that GLA-SE has been shown to signal through TRIF and MyD88 and that both are necessary for GLA-SE adjuvant activity (121). CD4<sup>+</sup> T cells are the central orchestrators of adaptive immune responses and provide the critical instructions for directing antibody production by B cells and CD8 T cell memory development (122, 123). Therefore, it is notable that, in terms of T cell responses, GLA-SE has been shown to promote polyfunctional CD4 T cell reactions characterized by CD154, IFN-g, TNF-a, GM-CSF, and IL-2 expression and the potential for GLA-SE induced IFN-a to trap immune cells in draining lymph nodes *via* CD69 expression (97). Immunization with *Mycobacterium tuberculosis* antigen and GLA-SE has resulted in elevated antigen specific B cell responses, elevated antibody concentrations, and increased follicular T helper and TH1 biased CD4 T cell numbers compared with adjuvants alum, GLA without SE, or SE alone (124). Furthermore, there is evidence to suggest that GLA-SE augments early innate IFN-g production by CD8 T and NK cells (97).

More recently, it has become clear that TLR4 agonists and TLR7 agonists may be highly synergistic in amplifying immune responses. This can be seen by increased cytokine secretion, heightened germinal center formation, and both antibody class switching and diversity (125, 126). Hence, the development of a nanoliposome formulation that co-localizes TLR4 and TLR7 agonists to the same APCs, which synergistically enhances immune responses have been developed. These TLR4/7 lipid nanoparticle adjuvants have been co-formulated with VLP containing SIVmac239 Env and Gag with the aim of comparing the immunogenicity of soluble gp140 Env and a protein immunogen to Env expressed on the surface of VLPs (127). Interestingly the nanoparticle adjuvant used in combination with soluble gp140 vaccine antigen performed better than the nanoparticle adjuvant co-formulated with VLPs. Specifically, the anti-Env antibody responses and Env-specific plasmablast numbers were greater in magnitude in serum for the soluble gp140 compared to VLP vaccine groups (127). To test the efficacy of TLR4/7-adjuvanted VLPs compared to soluble gp140, vaccinated macaques were challenged weekly with an SIVsmE660 swarm, and although there was a delay in infection with the adjuvanted VLP treatment group compared to the adjuvanted soluble gp140 treatment group, it was not statistically significant. The inclusion of the adjuvant did, however, enhance protection for both soluble gp140 and VLPs compared to the non-adjuvanted controls (127).

#### Adjuvant System 04 (AS04)

Adjuvant System 04 is a new combination adjuvant developed by GlasxoSmithKline Biologicals and consists of Aluminum salts and MPLA. The AS04 adjuvant is utilized in two licensed vaccines, Cervarix® and Fendrix®. Cervarix® contains VLPs of the L1 capsid protein from oncogenic HPV strains 16 and 18 expressed through an insect cell expression system. Fendrix® works against hepatitis B virus (128–130). It should be pointed out that a second HPV vaccine based on recombinant expression of major capsid antigen L1 in yeast is marketed by Merck &Co. under the brand name Gardasil®(131). In terms of Cervarix®, AS04 was selected as the adjuvant of choice because of its safety performance in mice and NHP studies and because it demonstrated induction of higher quality antibody and neutralizing anti-HPV titers response compared to alum alone (99). In the HPV16/18 clinical study, the VLP formulation adjuvanted with AS04 was proven effective against cervical intraepithelial neoplasia lesions and protective against strains phylogenetically related to HPV16/18 such as HPV31, HPV33, and HPV45 (132, 133). Comparing the AS04-adjuvanted HPV16/18 vaccine to alum adjuvanted HPV6/11/16/18 in women aged 18–45 revealed the former to elicit higher levels of neutralizing antibodies and higher frequencies of memory B cells and CD4 T cells (134).

#### MF59

The adjuvant MF59 was first approved in Italy in 1997 after Novartis conducted clinical trials in 1992 (135). MF59 is a squalene-based oil-in-water emulsion stabilized by Tween 80 and Span 85 surfactants (136). Due to its extensive safety profile and efficacy in preclinical and human clinical trials, it is now FDA approved and a component in licensed influenza vaccines such as FLUAD®, Aflunov®, Focetria®, and Celtura® (137). In addition, it is also being evaluated as an adjuvant in candidate HIV vaccine formulations. Although a complete understanding behind the mechanism of action of MF59's is still being defined, the widely held theory is that MF59 activates injection site tissue-resident APCs such as macrophages along with the transient release of ATP from the injected muscle (78). Upon activation, these tissue-resident cells secrete various chemokines and cytokines, which causes a localized inflammatory reaction, resulting in recruitment of more immune cells into the injection site. Vaccine antigen–APC interaction is, therefore, increased, augmenting transport of antigen into local draining lymph nodes and T cell priming. In terms of humoral immune responses, MF59 has been shown to increase antibody affinity maturation to vaccine antigens as well as the diversity of epitopes recognized (138). How this occurs may be due to findings that MF59 enhances antigen-specific IgG antibody responses to vaccine antigens by promoting T follicular helper cell (TFH) responses (139). While few studies have evaluated ways to augment the development of TFH cell responses through vaccination, these highly specialized cells are critical for humoral immunity through provision of help by means of both contact-dependent and independent mechanisms with TFH cell responses directly controlling the magnitude of germinal center B cell responses (140). Therefore, MF59 represents a very promising adjuvant for future vaccine development. Recently, Vargas-Inchaustegui, et al. demonstrated that an SIV vaccine-prime consisting of an Adenovirus 5 host range (Ad5hr) mutant, encoding SIV Gag, Nef, Rev, and Env followed by IM boosts with SIV monomeric gp120 or oligomeric gp140 protein adjuvanted with MF59 can induce long-lived germinal center, Env-specific IL-21<sup>+</sup>TFH cells in rhesus macaque lymph nodes (140). However, while MF59 has been reported to promote germinal center formation and TFH cell induction in previous studies (139, 141), within this study the corresponding effects of the boost to one of the components of the vaccine (SIVmac239 Env) were not as pronounced as that seen in Ad5hr vaccine-prime (140). Whether this lack of increased TFH cell induction by MF59 adjuvanted boost is due to nature of the vaccine antigen being used or its exposure time in the body compared to the viral vector remains to be determined (140).

Previously, Guillon et al. quantitatively and qualitatively tested the antibody response in rabbits induced by PLA nanoparticles and MF59 adjuvant. Using three different antigens, HIV-1 p24, WT HIV-1 Tat, and a detoxified Tat, both adjuvant platforms induced similar levels and kinetics of serum IgG (142). Strikingly, the authors noted differences in the antigenic domains that elicited the antibody response. Serum from p24 vaccinated rabbits showed MF59 induced antibodies that recognized peptides all along the p24 protein sequence while the PLA nanoparticles directed the antibody response toward an immunodominant domain of p24 (190–224). Potentially indicating that adsorption of antigens onto the surface of particles might unmask, or alter the conformation of antigens, revealing alternative immunodominant epitopes not revealed by soluble antigens (142).

#### ISCOMATRIX

This adjuvant was designed as an improvement over its predecessor, the ISCOM (immunostimulating complex) adjuvant. The original formula was a mix of saponin, cholesterol, phospholipid, and other hydrophobic compounds (143). The major drawback to the original formulation was that the vaccine antigen had to be integrated into the hydrophobic structure of the adjuvant, greatly limiting the types of antigen that could be administered with ISCOM (100). To address these issues, Pearse & Drane developed ISCOMATRIX, which did not require the incorporation of antigen. This newer formulation was similar; however, was produced using purified fractions of *Quillaja Saponaria*, cholesterol, and phospholipids. When formulated together, the lipid and QS21 components assemble into 40–50 nm cage-like structures that entrap the vaccine antigen. Unlike other adjuvant systems, the ISCOMATRIX–antigen preparation infiltrates the draining lymph node and interacts with lymph node-resident DCs and other APCs within the first 2 h of immunization. Interestingly, there are two waves of antigen presentation when ISCOMATRIX is used. The first round of antigen presentation occurs when ISCOMATRIX–antigen complexes infiltrate the draining lymph node, whereas the second round of antigen presentation occurs 24–48 h post-inoculation. This is due to tissue-resident DCs that take up ISCOMATRIX–antigen complexes and traffic to draining lymph nodes.

Morelli et al. describe ISCOMATRIX as an integrated adjuvant system because it combines the innate and adaptive arms of the immune system in a TLR-independent but MyD88-dependent manner. In particular, when DCs take up ISCOMATRIX into endosomes or phagolysosomes, these cellular compartments undergo acidification and releases the ISCOMATRIX–antigen complexes into the cytoplasm of the cell. This enables presentation through MHC class I pathway for cross-presentation to CD8 T cells. As expected, antigen is also presented on MHC class II for presentation to CD4 T cells and induction of B cell responses. In terms of T cell responses toward ISCOMATRIX–antigen complexes, it activates both TH1 and TH2 responses as noted by rapid IL-5 and IFN-g production following immunization (144).

As mentioned previously, Ingale et al. evaluated the *in vivo* delivery of soluble JRFL SOSIP trimers in ISCOMATRIX and JRFL SOSIP trimer-conjugated liposomes in ISCOMATRIX. Isolation of lymph node cells indicated increased percentages of GL7+ germinal center B cells from mice administered the liposome-conjugated vaccine (58). Within this study, the delivery and evaluation of well-ordered, high-density Env trimers was the main goal of the research, as such there was no evaluation of the SOSIP Env trimers in the absence of ISCOMATRIX, making it difficult to understand the precise effect of the adjuvant. In a separate study, evaluating germinal center formation and neutralizing antibody responses to HIV Env trimers, Havenar-Daughton et al., using a modified BG505 SOSIP (BG505 SOSIP. v5.2), compared two different adjuvants (PLGA/R848 + MPL vs ISCOMATRIX) for generation of autologous tier 2 neutralizing antibody production in NHPs. In this example, approximately 75% of the vaccinated NHPs generated a detectable neutralizing antibody response but neither adjuvant significantly outperformed the other in terms of elicited Env antibody titer, V3 peptide antibody binding titer or neutralizing antibody titer (145). Interestingly, three NHPs from the PLGA group and 1 NHP from the ISCOMATRIX group were identified as top tier 2 autologous neutralizing antibody producers (titers >1:200). Although it should be noted that both groups had different numbers of animals, with the PLGA treatment group having the most. While these results were promising, no viral challenge was performed (145). Collectively, this suggests that ISCOMATRIX is a promising vaccine adjuvant with the potential to significantly augment B cell immune responses.

#### Surface Functionalized Nano-Vaccines

The co-delivery of vaccine antigens with adjuvants has long been known to augment both B and T cell immune responses. As previously mentioned, current scientific dogma suggests for optimal adjuvant activity, the vaccine antigen and the adjuvant should be targeted to the same APC. Co-delivery of vaccine antigens and adjuvants can normally be achieved using covalent linkages between the two components, ensuring that the antigen and adjuvant will end up in the same endo-lysosomal compartments for maximal PRR engagement and proteolytic degradation. However, due to the increased costliness and in some cases, difficulty of conjugating certain complex antigens or heavily glycosylated antigens to the adjuvant makes this process unfavorable. Therefore, newer ways to formulate antigens and adjuvants are necessary. Interestingly, the use of nano-vaccines offers an alternative mechanism by which antigens and adjuvants can be co-delivered for optimal immune responses. Through cell surface expression, a range of recombinant vaccine adjuvants can be incorporated into budding VLP nano-vaccines, ensuring that surface expressed immunogens are presented alongside the adjuvant. In a study by Franco et al., immuno-stimulatory murine CD40L was expressed on the surface of HIV VLPs, with the aim of targeting HIV-1 gag proteins to DCs and activating the CD40 receptor (146). Through fusion to the CD40L ectodomain with baculovirus gp64, the authors significantly enhanced CD40L expression by the VLPs and demonstrated that the CD40L/gp160 VLPs enhanced IL-12 production from DCs. Further *in vivo* testing of the CD40L/gp160 VLPs in mice demonstrated the vaccine construct to induce strong CD4 and CD8 cell-mediated immune responses compared to CD40L-VLP constructs expressing either full length murine CD40L or murine CD40L fused to gp41(146).

Numerous studies have shown that cytokines serve to be promising adjuvants for vaccines. For instance, an SIV239-based heterologous prime-boost vaccine, co-expressing GM-CSF, was shown to convey 71% protective efficacy against intrarectal SIVSME660 challenge when compared to 25% in the non-GM-CSF adjuvanted group (147). In this case, the GM-CSF adjuvant was shown to enhance the avidity of the anti-Env IgG antibody response, augment the neutralizing activity of the sera, and enhance ADCC activity (147). More recently, a gain of function fusokine (GM-CSF + IL-4, aka GIFT4) membrane bound construct was utilized as a VLP bearing, B cell adjuvant (148). Anchoring of the fusokine into the Env-enriched VLP membrane was achieved by fusing CD59 glycolipid, glycophosphatidyl-inositol (GPI) sequence to GIFT4. The chimeric VLP constructs were administered as an IM-prime, IN-boost in Guinea pigs resulting in high levels of systemic antibody responses. In addition to this, the elicited antibodies exhibited higher avidity, improved neutralization breadth, and were detected in both serum and mucosal secretions (148).

Flagellin, in the monomeric form, is a TLR5, MyD88 dependent signal agonist derived from Gram-negative bacteria flagella. Flagellin is the major protein of the flagellum filament, which extends from the outer membrane of many locomotive Gram-negative bacteria and is responsible for the organism's locomotive ability. Flagellin is highly pro-inflammatory, activating NF-kB in a variety of cells, such as epithelia and APC populations such as monocytes and DCs (98). As the TLR5 receptor is found to be expressed in APCs, it has been suggested that flagellin-containing VLPs may be phagocytosed after TLR5 binding and enhance antigen presentation due to innate signaling within the same cell. In an interesting study by Wang et al., a membrane-anchored form of flagellin was incorporated into a chimeric influenza VLP (149). The results clearly demonstrated that flagellin-containing VLPs could elicit significantly higher antibody responses than unadjuvanted VLPs and that the adjuvant activity was not restricted to humoral immunity. In the latter case, MHC class I and II peptide stimulated splenocyte cultures were shown to have increased cellular immune responses and the mice vaccinated with adjuvanted VLP to be more resistant to heterologous influenza challenge infection than non-adjuvanted controls (149). The membrane-anchored flagellin was also combined into HIV VLPs as a molecular adjuvant using the guinea pig model (150). Both systemic (IM) and mucosal (IN) routes of vaccination elicited serological and mucosal antibody responses, with IgG sera levels correlating with neutralization. Of note, the neutralizing antibody response was broadened against the five tested strains from clades B and C (150).

Surface functionalizing of nanoparticles can also increase the bioavailability of vaccine antigen cargos through enhanced vaccine penetration in tissues. For instance, through decorating the surface of nano-vaccines with mucosal M cell targeting moieties, the bioavailability of mucosally applied vaccine formulations can be increased and antigen-specific immune responses significantly augmented. For instance, through surface functionalizing of PEGylated PLGA nano-vaccines with integrin binding RGD tripeptide (Arg-Gly-Asp), Garinot et al. were able to significantly increase the transport of particles using an *in vitro* model of follicle associated epithelium (Caco-2 cells and Raji B cells) and also monocultures of Caco-2 cells. Oral application of the RGD-nanoparticles was also shown to be immunogenic in some mice, with some animals having elevated anti-OVA IgG antibody responses and splenic IFN-γ T cell responses (151). Additional M cell targeting moieties, such as fucose binding lectin *Ulex euopaeus* agglutinin1 and reovirus sigma protein-1 have also previously been studied as virus, nanoparticle and microparticle targeting strategies for drugs and vaccines (152–156). Alternatively, cell penetrating peptides such as HIV-1 Tat, or CD71 receptor targeting protein such as transferrin have been bio-coupled to vaccine antigens and nano-vaccines for delivery to various cell types and organs (157). Therefore, the functionalization of the surface of nano-vaccines represents an exciting area of active investigation with the potential to significantly tailor vaccines to specific cells, tissues, and organs for optimal immune protection.

#### SUMMARY AND CONCLUSION

The last three decades has seen tremendous advancements in our understanding of HIV acquisition, treatment, and prevention. During this time, several vaccine efficacy trials have been completed and all have failed to prevent HIV infections apart from the modestly effective RV144. A major impediment to a successful vaccine has been antigen design and delivery. Crucial to overcoming this roadblock has been the development of nanoparticles to stabilize and orientate candidate vaccine immunogens for optimal immune triggering. Within this review we have covered the basis for vaccine development and the promising nanoparticle-based approaches to answer some of the very complicated issues surrounding HIV vaccine progress. Nanoparticle-based vaccines have demonstrated themselves to be versatile, and depending on which antigen, adjuvant and delivery route used, can have significant impacts on prevailing immune responses. While nanoparticle vaccines hold great promise, a significant number of unknowns still exist, and when answered could greatly advance our understanding on how to maximally deliver promising vaccine immunogens. These include a better understanding of the effects of charge, size, structure, and potential toxic side effects with various nanoparticle platforms.

While the major focus of this nanoparticle review has been on the development of neutralizing antibody, nanoparticle formulations targeting the V1/V2 region of Env by non-neutralizing antibody is a significant area of interest and needs to be explored further. Justification for this is drawn from the correlates of protection from the RV144 study, which is being further evaluated and validated by on-going clinical trials in Thailand and Africa. In addition, although most vaccines are delivered systemically, HIV is predominantly transmitted through mucosal surfaces and as such, it is imperative that vaccine candidates be developed that can either augment mucosal immunity or be designed to draw protective systemic responses into the mucosal environment. The latter point would enable the provision of protective frontline responses within tissues at risk of infection. The only caveat being that mucosal tissue activation and vaccine-induced inflammation be carefully managed, to not increase the likelihood of transmission by growing the number of potential target cells.

Finally, the use of nanoparticle technology will also undoubtedly have a significant role to play in therapeutic HIV vaccines. The thought of an HIV-1 cure was once regarded a dream. Now, growing evidence suggests that a cure could be a reality and the generation of vaccine induced antiviral immunity will take center stage. There are numerous cure technologies that currently being evaluated, however it is the "Shock and Kill" tactic that has garnered significant attention. Several latency reversal agents have been devised to reactivate (shock) latent virus into transcriptional

#### REFERENCES


activity; however, vaccines that can prime for protective antibody and cytotoxic T and NK cell responses will be essential for subsequent viral eradication (kill).

We are clearly entering an exciting time for HIV vaccine development and nanoparticle formulations will have a lot to offer. As long as there is continued support for the nanoparticle vaccine effort, prophylactic and therapeutic HIV/AIDS interventions will continue to advance until such time as a vaccine becomes a reality.

## AUTHORS CONTRIBUTIONS

YG, CW, and JM wrote sections of the manuscript and YG, CW, and JM edited the manuscript.

#### ACKNOWLEDGMENTS

The authors would like to thank Dr. Katja Klein for helping design and edit this manuscript.

#### FUNDING

JM is funded through a research award from the Canadian Institute of Health Research (grant number: PJT 149075).


with trimeric HIV-1 envelope glycoproteins. *Virology* (2005) 331:33–46. doi:10.1016/j.virol.2004.09.022


T-cell help promote the immunogenicity of HIV membrane-proximal external region peptides. *Vaccine* (2015) 33:861–8. doi:10.1016/j.vaccine. 2014.12.045


to a reduction in prevalence of HPV 16/18 and closely related HPV types. *Br J Cancer* (2014) 110:2804–11. doi:10.1038/bjc.2014.198


**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 Gao, Wijewardhana and Mann. 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.*

*Danielle A. Wagner-Muñiz1 , Shannon L. Haughney2 , Sean M. Kelly2 , Michael J. Wannemuehler1,3 and Balaji Narasimhan2,3\**

*1Department of Veterinary Microbiology and Preventive Medicine, Iowa State University, Ames, IA, United States, 2Department of Chemical and Biological Engineering, Iowa State University, Ames, IA, United States, 3Nanovaccine Institute, Iowa State University, Ames, IA, United States*

*Streptococcus pneumoniae* is a major causative agent of pneumonia, a debilitating disease particularly in young and elderly populations, and is the leading worldwide cause of death in children under the age of five. While there are existing vaccines against *S. pneumoniae*, none are protective across all serotypes. Pneumococcal surface protein A (PspA), a key virulence factor of *S. pneumoniae*, is an antigen that may be incorporated into future vaccines to address the immunological challenges presented by the diversity of capsular antigens. PspA has been shown to be immunogenic and capable of initiating a humoral immune response that is reactive across approximately 94% of pneumococcal strains. Biodegradable polyanhydrides have been studied as a nanoparticle-based vaccine (i.e., nanovaccine) platform to stabilize labile proteins, to provide adjuvanticity, and enhance patient compliance by providing protective immunity in a single dose. In this study, we designed a room temperature stable PspA-based polyanhydride nanovaccine that eliminated the need for a free protein component (i.e., 100% encapsulated within the nanoparticles). Mice were immunized once with the lead nanovaccine and upon challenge, presented significantly higher survival rates than animals immunized with soluble protein alone, even with a 25-fold reduction in protein dose. This lead nanovaccine formulation performed similarly to protein adjuvanted with Alum, however, with much less tissue reactogenicity at the site of immunization. By eliminating the free PspA from the nanovaccine formulation, the lead nanovaccine was efficacious after being stored dry for 60 days at room temperature, breaking the need for maintaining the cold chain. Altogether, this study demonstrated that a single dose PspA-based nanovaccine against *S. pneumoniae* induced protective immunity and provided thermal stability when stored at room temperature for at least 60 days.

Keywords: pneumococcal infections, pneumococcal surface protein A, polyanhydride, nanovaccine, room temperature stability

## INTRODUCTION

Community-acquired pneumonia is a debilitating disease and is globally responsible for 16% of deaths in children under the age of five annually (1). *Streptococcus pneumoniae*, a causative agent of bacterial pneumonia, is a Gram-positive bacterial pathogen of humans, which colonizes the upper respiratory tract and leads to the development of potentially life-threatening diseases, such as otitis

#### *Edited by:*

*Rajko Reljic, St George's, University of London, United Kingdom*

#### *Reviewed by:*

*Giampiero Pietrocola, University of Pavia, Italy Anita S. Iyer, Harvard Medical School, United States*

> *\*Correspondence: Balaji Narasimhan nbalaji@iastate.edu*

#### *Specialty section:*

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

*Received: 07 December 2017 Accepted: 06 February 2018 Published: 02 March 2018*

#### *Citation:*

*Wagner-Muñiz DA, Haughney SL, Kelly SM, Wannemuehler MJ and Narasimhan B (2018) Room Temperature Stable PspA-Based Nanovaccine Induces Protective Immunity. Front. Immunol. 9:325. doi: 10.3389/fimmu.2018.00325*

**88**

media, sinusitis, and pneumonia, with pneumonia being the most serious of these conditions (2, 3). While otitis media and sinusitis are not as severe, they still account for seven million cases in the United States annually (2). *S. pneumoniae* is asymptomatically carried in approximately 25% of healthy individuals and is spread between individuals through respiratory droplets, with children serving as the primary transmission source to adults (4–7). Pneumococcal infections are most prevalent in infants, young children, and the elderly, accounting for over \$3 billion spent annually in direct healthcare spending; in addition, 60–87% of all pneumococcal bacteremia cases are associated with pneumonia in adults (2). While control of *S. pneumoniae* by antibiotics has been beneficial, up to 50% of all strains are resistant to erythromycin, with 97% of those also being resistant to azithromycin, making intervention strategies such as vaccination a necessary component of health care management (8). To date, the single most effective advance in the field of pneumonia prevention has been through immunization, though existing vaccines are still not completely effective (9, 10).

Current preventive strategies involve the use of pneumococcal polysaccharide vaccines (PPVs) and pneumococcal conjugate vaccines (PCVs). PPVs (T-independent antigens) comprise capsular polysaccharides, which are poorly immunogenic in very young and elderly individuals, while PCVs (T-dependent antigens) are more effective in these at-risk populations because of the adjoined protein component. While the introduction of PPV23, PCV7, and PCV13 have had significant contributions in reducing the overall global burden of pneumococcal pneumonia, there is still room for improvement (11). As it currently stands, infection with strains not covered by vaccine serotype infections account for approximately 50% of all deaths in individuals over 55 years of age (12). In addition, it has been observed that introduction of new vaccine strategies against *S. pneumoniae*, such as the introduction of PCV10 immunization program for children in the Netherlands or use of PCV7 in Spain, may be directly correlated to an increased emergence of disease caused by nonvaccine serotype strains cases (13–15). It was demonstrated that PCV7 immunization, which contains the 19F serotype antigen, actually caused an increase in pneumococcal infections by the closely related 19A serotype (16). Globally, this is of particular concern when considering the development of a universal vaccine, as it is now clear that the prevalence of certain serotypes differs greatly between age, geographic region, and ethnicity (3, 17, 18). PPV23 encompasses the most pneumococcal serotypes of the commercially available vaccines. However, it is not recommended for use in children under the age of 18 months because of their poor antibody response to T-independent polysaccharide antigens, explaining why PCV13 is recommended for this age group. While the use of PCV and PPV vaccine formulations have effectively reduced the incidence of pneumococcal pneumonia, there is a need for newer vaccines to protect against the emergence of non-vaccine strains of *S. pneumoniae* (19). Coupled with the phase variation in expression of capsular and surface protein antigens, broadly protective vaccines against colonization and invasive pneumococcal disease are likely to require the inclusion of more conserved antigens that will also facilitate the induction of antigen-specific CD4<sup>+</sup> T cells (20, 21).

*S. pneumoniae* has several key virulence factors that play a critical role in colonization, transmission, and tissue damage, including pneumolysin, two neuraminidases, and pneumococcal surface protein A (PspA). PspA is a choline binding protein, one of the most abundant proteins located on the pneumococcal cell surface, and has been shown to be of particular importance in facilitating nasopharangyl colonization through inhibiting host complement responses (22). In the design of new vaccines against *S. pneumoniae*, it is important to consider a target immunogen that will be capable of inducing a broad, serotype-independent, protective immune response. PspA has been of recent interest as a potential vaccine candidate due to its location on the cell surface of all 95 strains of *S. pneumoniae* currently known and its critical role in pneumococcal pathogenesis (3, 23, 24). Anti-PspA antibodies to clades 1 and 2 of PspA have been shown to be cross-protective against *S. pneumoniae* strains encompassing all six clades of PspA, and provide protection when passively transferred to naïve mice as a therapeutic intervention following septic *S. pneumoniae* challenge (17, 25–27). The ability to provide protection across many serotypes of *S. pneumoniae* using only two clades of PspA allows for the design of a broadly cross-protective vaccine when compared to vaccines containing numerous capsular polysaccharides.

Polyanhydride nanoparticle-based vaccines (i.e., nanovaccines) represent next-generation vaccine platforms against pathogens such as *S. pneumoniae*. These nanovaccines are formulated using random copolymers based on 1,8-bis-(p-carboxyphenoxy)- 3,6-dioxaoctane (CPTEG), 1,6-bis-(p-carboxyphenoxy)hexane (CPH), and sebacic acid (SA) and have been extensively studied as a nanovaccine platform against infectious pathogens, such as influenza virus, *Yersinia pestis*, and *Bacillus anthracis* (28–31).

Polyanhydrides provide vaccine delivery benefits and adjuvant properties that make them well suited as a vaccine delivery platform. These materials exhibit high biocompatibility with minimal injection site reactivity (32–34) (i.e., tenderness, swelling, and pain) in comparison to traditional adjuvants, such as Alum, which has been associated with immunization site tenderness and pain, and are currently FDA-approved for use to treat malignant gliomas in the brain (35, 36). In addition, polyanhydrides are hydrophobic and exhibit surface erosion characteristics, which helps stabilize labile proteins, protects protein denaturation from enzymatic cleavage and acidic degradation products, and allows for the prolonged release of antigen (28, 37–41). The sustained release of antigen allows for enhanced bioavailability of antigen to drive adaptive immune responses and allows for single dose administration and dose-sparing capabilities (30, 42–44). Previous work has shown that mice immunized with a single dose of nanovaccine encapsulating *Y. pestis* fusion protein F1-V were completely protected against *Y. pestis* lethal challenge after at least 280 days post-immunization (DPI) (31). Varying polyanhydride copolymer composition has shown to modulate internalization and persistence within APCs *in vitro*, as well as induction of both cellular and humoral immune responses *in vivo* indicating the ability to tailor polymer chemistry in order to rationally design nanovaccines that optimally inducing antigen-specific protective immunity (31, 42, 45–54).

Previous work from our laboratories has demonstrated that PspA encapsulated into nanoparticles maintained its stability, conformational structure, as well as biological activity upon release, which was measured using an apolactoferrin binding assay (37). In this same work, mice immunized with a single dose of the nanovaccine generated robust antibody titer and avidity to PspA, equivalent to that of antigen adjuvanted with MPLA. However, the protective efficacy of a PspA-based nanovaccine against lethal challenge has not been demonstrated. In this study, the ability of various polyanhydride nanovaccine chemistries as vaccine candidates was evaluated and it was demonstrated that a lead PspA-based nanovaccine protected animals against lethal challenge with a 25-fold reduction in overall PspA dose. Furthermore, this lead formulation was shown to be capable of storage at room temperature for at least 2 months without any loss of efficacy.

#### MATERIALS AND METHODS

#### Materials

Chemicals used for monomer and polymer synthesis included sodium hydroxide, hydrobenzoic acid, dibromohexane, 1-methyl-2-pyrrolidinone, SA monomer, and triethylene glycol purchased from Sigma Aldrich (St. Louis, MO, USA). Acetic anhydride, ethyl ether, petroleum ether, chloroform, methylene chloride, hexane, acetone, sulfuric acid, potassium carbonate, dimethyl formamide, toluene, acetonitrile, N,N-dimethylacetamide, and acetic acid were purchased from Fisher Scientific (Fairlawn, NJ, USA); 4-p-fluorobenzonitrile was purchased from Apollo Scientific (Cheshire, UK). For 1 H NMR analysis of the copolymers, deuterated chloroform and dimethyl sulfoxide were purchased from Cambridge Isotope Laboratories (Andover, MA, USA). The N-terminal region of a recombinant PspA (UAB055, PspA/Rx1 AA1 to 303, clade 2 PspA of the PspA family 1) was produced by Dr. David McPherson (University of Alabama at Birmingham) as described previously (55). Prior to immunization, endotoxin was removed from the protein using endotoxin removal beads (Miltenyi Biotec, Bergisch Gladbach, Germany) according to the manufacturer's instructions followed by dialysis and lyophilization. The final endotoxin content of the protein was determined to be less than 1.9 EU/mg as determined by a limulus amebocyte lysate chromogenic endotoxin quantification kit (Thermo Fisher Scientific).

#### Copolymer and Nanoparticle Synthesis

Monomers of CPH and CPTEG were synthesized as previously described (28, 49). A 50:50 molar composition copolymer of CPH and CPTEG and a 20:80 molar composition copolymer of CPH and SA was synthesized using melt condensation as described by Torres et al. (28). PspA protein, 1% (w/w), was encapsulated into polyanhydride nanoparticles using solid/oil/ oil nanoprecipitation, as previously described (37). Excess buffer was removed from PspA protein solution using a 5 kDa MWCO dialysis microcentrifuge tube and the resulting protein solution was lyophilized overnight at −40°C under vacuum. The lyophilized protein and the respective copolymer was then dissolved in methylene chloride at a concentration of 20 mg/mL of solvent. The solution was sonicated using the VibraCell ultrasonic probe (Sonics & Materials, Inc., Newtown, CT, USA) to ensure complete dissolution and homogenization of the protein and the copolymer. The resulting solution was then added to pentane at a 1–200 (v/v) ratio of methylene chloride to pentane at 20°C for the 20:80 CPH:SA nanoparticle formulation or at a 1–250 (v/v) ratio of methylene chloride to pentane at −40°C and for the 50:50 CPTEG:CPH nanoparticle formulation. Nanoparticles were collected using vacuum filtration. Surface charge of representative 20:80 CPH:SA and 50:50 CPTEG:CPH nanoparticle formulations were measured using quasi-elastic light scattering (Zetasizer Nano, Malvern Instruments Ltd., Worcester, UK).

#### Animals

CBA/CaHN-Btkxid/J (CBA/N) mice were purchased from Jackson Laboratory (Bar Harbor, ME, USA) and used for the immunization studies. These mice have a mutation in their Bruton's tyrosine kinase gene. Due to this mutation, these mice are unable to initiate a proper antibody response to Type II thymicindependent antigens (e.g., capsular polysaccharides), and are, thus, able to better mimic natural antibody responses in at-risk elderly or young populations. Mice were housed under specific pathogen-free conditions where all bedding, caging, water, and feed were sterilized prior to use. All studies were conducted with the approval of the Iowa State University Institutional Animal Care and Use Committee.

#### Immunization Protocols

Groups of 8–10 CBA/N mice were immunized subcutaneously at the nape of the neck with 0–20 µg PspA, 0–5 µg PspA encapsulated into 50–500 µg polyanhydride nanoparticles, and/or 50 µL of Imject Alum Adjuvant (ThermoFisher Scientific) and delivered in a total volume of 100 µL. Immunization groups for the individual studies (see **Tables 1** and **2**) comprised the following : (i) 25 µg soluble PspA alone, (ii) 25 µg PspA administered with 50 µL Imject Alum, (iii) 5 µg PspA encapsulated into 250 µg 50:50 CPTEG:CPH nanoparticles + 20 µg soluble PspA, (iv) 5 µg PspA encapsulated into 250 µg 20:80 CPH:SA nanoparticles + 20 µg soluble PspA, (v) 1 µg soluble PspA alone, (vi) 1 µg PspA administered with 50 µL Imject Alum, (vii) 1 µg PspA encapsulated into 50 µg 50:50 CPTEG:CPH nanoparticles, (viii) 0.5 µg PspA encapsulated into 25 µg 50:50 CPTEG:CPH nanoparticles + 25 µg blank 50:50 CPTEG:CPH nanoparticles + 0.5 µg soluble PspA, or (ix)

Table 1 | 25 µg Pneumococcal surface protein A (PspA)-containing single dose vaccine treatments groups.


*Separate groups of mice were immunized s.c. and all formulations were delivered in 100 µL total volume. Soluble PspA is represented as sPspA, and nanoparticleencapsulated PspA is represented as ePspA.*

Table 2 | One-microgram pneumococcal surface protein A (PspA)-containing single dose vaccine treatments groups.


*Separate groups of mice were immunized s.c. and all formulations were delivered in 100 µL total volume. Soluble PspA is represented as sPspA and nanoparticleencapsulated PspA is represented as ePspA.*

saline control. Blood was collected from mice *via* saphenous vein and serum was isolated following centrifugation (10,000 rcf for 10 min) at days 17 or 21 DPI. Serum was stored at −20°C until analysis could be performed.

#### *S. pneumoniae* Challenge

*S. pneumoniae* strain A66.1, of PspA family 1, clade 2 was used for challenge and grown as previously described (56, 57). Briefly, the bacteria was grown at 37°C for 18 h in filter-sterilized Todd-Hewitt broth with 0.5% yeast extract, and stored at −80°C at a concentration of 1 × 108 CFU/mL. Inoculum was diluted to a concentration of 2.5 × 104 CFU/mL in sterile PBS and 100 µL was administered intraveneously at 24 DPI. Mice were monitored three times per day for the duration of the challenge (14 days). Body temperatures and body condition scores were also recorded once per day. Mice were sacrificed when determined to be moribund or at 14 days post-challenge.

#### Anti-PspA Antibody Titer

Anti-PspA antibody titers were determined *via* an ELISA as described previously (37). Briefly, high-binding Costar 590 EIA/ RIA microtiter plates (Corning) were coated overnight at 4°C with 0.5 µg/mL PspA. The blocking buffer for this assay comprised 2.5% (w/v) powdered milk dissolved in PBS-T and incubated for 2 h at 56°C to inactivate any endogenous phosphatase activity. PspA-coated plates were incubated with blocking solution for 2 h at room temperature before being washed three times with PBS-T. Serum obtained from immunized mice was added to the first well of a row at a dilution of 1:200 and serially diluted in PBS-T containing 1% (v/v) goat serum. Each serum sample was tested in duplicate. Following incubation overnight at 4°C, plates were washed three times with PBS-T after which secondary antibody was added at a dilution of 1 µg/mL. The secondary antibody used in these studies was alkaline phosphatase-conjugated goat anti-mouse IgG heavy and light chain (Jackson ImmunoResearch) and was incubated on the plates for 2 h at room temperature. Plates were then washed three times with PBS-T and alkaline phospatase substrate (Fisher Scientific, Pittsburgh, PA, USA) was added at a concentration of 1 mg/mL dissolved in 50 mM sodium carbonate, 2 mM magnesium chloride buffer at pH 9.3 for colorimetric development. Plates were analyzed after 30 min using a SpectraMax M3 microplate reader at a wavelength of 405 nm. Titer is reported as the reciprocoal of serum dilution at which optical density value was at least twice that of the saline group average plus two SDs.

## Shelf Stability of Stored PspA Nanovaccine

Shelf stability of the PspA nanovaccine was assessed *via* ELISA and lethal challenge. 50:50 CPTEG:CPH nanoparticles encapsulating PspA were taken out of the freezer (−80°C) and stored as a dry powder formulation at 25°C (room temperature) in a glass desiccator containing drierite for 60 days prior to immunization. Soluble PspA was stored at 4°C and combined with Imject Alum prior to immunization. A separate group of nanoparticles were kept in the freezer over the course of this experiment as a freezer-stored standard control. Following room temperature storage, vaccine efficacy was evaluated by immunizing mice with a nanovaccine dose equivalent to 1 µg PspA per animal from one of the storage conditions above, soluble PspA adsorbed to Alum, or the saline control and assessing antibody titer *via* ELISA 21 DPI (see "Anti-PspA Antibody Titer" for details). Following lethal challenge with 2,500 CFU of *S. pneumoniae* (see *S. pneumoniae* challenge for details), survival was assessed over the next 24 days.

#### Statistical Analyses

Statistical analyses were performed using the Gehan–Breslow– Wilcoxon test and the one-way analysis of variance using Tukey's multiple comparisons test. A *p*-value ≤0.05 was considered statistically significant. All analyses were performed using GraphPad Prism v. 7.0 software.

## RESULTS

#### Synthesis and Characterization of PspA-Encapsulated Polyanhydride Nanoparticles

The nanoparticle chemistries chosen for the current study were based on 50:50 CPTEG:CPH and 20:80 CPH:SA copolymers, consistent with our previous work, in which these chemistries were shown to provide protein structural stability and maintain the functional activity of PspA following release (37). The nanoparticles were synthesized using solid/oil/oil emulsion (54). The PspA release kinetics from these formulations are described in our previous work (37). Scanning electron photomicrographs of the nanoparticles revealed similar sizes for the two formulations, with sizes of 455 ± 175 and 422 ± 163 nm for 50:50 CPTEG:CPH and 20:80 CPH:SA nanoparticles, respectively (**Figure 1**). Zeta potential measurements of empty 20:80 CPH:SA and 50:50 CPTEG:CPH nanoparticles revealed that the particles had a negative surface charge, with zeta potentials of −41.9 ± 0.9 and −33.1 ± 5.1 mV, respectively, consistent with previous publications (45, 48, 58).

#### Immunization with PspA-Based Nanovaccines Induces Protective Antibody Response

Previously, it was demonstrated that mice vaccinated with PspA-based nanovaccines induced a robust humoral response; however, protective efficacy was not evaluated (37). In the current studies, separate groups of CBA/CaHN mice (*n* = 8 per

group) were immunized subcutaneously as follows: (i) 25 µg soluble PspA alone, (ii) 25 µg PspA administered with Imject Alum, (iii) PspA-containing 50:50 CPTEG:CPH nanoparticles, (iv) PspA-containing 20:80 CPH:SA nanoparticles, or (v) saline (**Table 1**). Both the nanovaccine formulations contained 5 µg of encapsulated PspA along with 20 µg of soluble protein, as shown in **Table 1**. Anti-PspA total IgG titers were evaluated at 17 DPI and there were no significant differences in titer between vaccinated groups (**Figure 2A**). Mice were challenged with a lethal dose of *S. pneumoniae* strain A66.1 at 24 DPI and survival was assessed over a period of 2 weeks. All PspA naïve (saline) mice succumbed to infection within 4 days, while all the immunized animals showed prolonged survival following challenge. At 14 days post-challenge, only 25% of the (non-adjuvanted) soluble PspA-immunized animals survived, whereas adjuvanting the protein with alum provided 100% protection. Immunization with the 20:80 CPH:SA nanovaccine resulted in 87% survival, while 100% of the animals immunized with 50:50 CPTEG:CPH nanovaccine were protected, as shown in **Figure 2B**. The survival data, in correlation with the antibody titers indicate that not only is PspA an effective immunogen, but induces protective immunity when delivered with an adjuvant. Though both nanoparticle chemistries protected animals from lethal challenge compared

to saline, the 50:50 CPTEG:CPH PspA nanovaccine performed significantly (*p* ≤ 0.008) better than soluble PspA (i.e., sPspA) alone. Based on these results, we selected the 50:50 CPTEG:CPH nanovaccine as our lead candidate for further evaluation.

### PspA Nanovaccine Demonstrates Dose-Sparing Capabilities

Rational design of vaccines using subunit proteins involves considering total cost of the end product, as well as protein availability that potentially impacts vaccine shortages. Our previous work has shown that encapsulating a lowered dose of antigen into polyanhydride nanoparticles can induce a robust immune response (43). In the current studies, the ability of the nanovaccine to provide dose-sparing (i.e., 25-fold reduction) was evaluated by immunizing separate groups of CBA/N mice (*n* = 8 per group; see **Table 2**) subcutaneously using the following regimens: (i) 1 µg PspA encapsulated into 50:50 CPTEG:CPH nanoparticles (PspA Nanovaccine), (ii) 0.5 µg PspA encapsulated into 50:50 CPTEG:CPH nanoparticles plus 0.5 µg soluble PspA (PspA Nanovaccine + Soluble PspA), (iii) 1 µg soluble PspA administered with Imject Alum, (iv) 1 µg soluble PspA alone, and (v) saline alone (i.e., naïve control). Serum was collected at 21 DPI and analyzed

Figure 2 | Pneumococcal surface protein A (PspA) nanovaccines are capable of initiating protective immune responses. (A) Serum was collected from immunized mice (see Table 1) at 17 days post-immunization (DPI) and analyzed for total anti-PspA IgG (data is represented as log10) using ELISA. Dashed line represents background anti-PspA IgG levels in serum from the saline-immunized mice. (B) *S. pneumoniae* challenge was performed at 24 DPI by i.v. administration of 2,500 CFUs of A66.1 strain and survival was monitored for 14 days post-challenge. \* indicates significance (*p* ≤ 0.0025) compared to sPspA alone and # indicates significance (*p* ≤ 0.001) compared to saline-treated mice. Data are representative of one independent experiment containing *n* = 8 mice per treatment group.

for total anti-PspA IgG. Mice immunized with the nanovaccines or PspA + Alum showed significantly (*p* ≤ 0.008) higher titers to PspA compared to soluble PspA alone (**Figure 3A**). Survival rates following lethal challenge showed significantly (*p* ≤ 0.001) higher survival (87.5%) in mice immunized with both the nanovaccine formulations compared to animals immunized with soluble PspA (12%) (**Figure 3B**). Animals immunized with PspA adjuvanted with Alum performed similarly to the animals receiving the nanovaccine formulations. There were no observable differences in disease severity between the immunized animals, as measured by body condition scores and temperature (data not shown). In addition, soluble PspA administered in conjunction with 100 µg of empty 50:50 CPTEG:CPH nanoparticles, performed similarly to other adjuvanted groups, further indicating the ability for PspA to initiate a protective immune response when adjuvanted properly (data not shown). No significant differences in survival were observed between mice immunized with PspA fully encapsulated within the nanoparticles (i.e., PspA Nanovaccine) and animals that received half of the protein as a soluble bolus (i.e., PspA Nanovaccine + Soluble PspA). Eliminating the use of soluble PspA in the nanovaccine is beneficial when designing a shelf-stable dry powder formulation for the long-term storage of a PspA nanovaccine.

#### PspA Nanovaccine Maintains Efficacy following Room Temperature Storage

To evaluate the potential for a room temperature stored (i.e., non-refrigerated) PspA nanovaccine, 1 µg of PspA was encapsulated within 50 µg of 50:50 CPTEG:CPH nanoparticles (i.e., 2 wt.% loading) and stored at room temperature (25°C) and standard storage conditions (−80°C), as detailed in the Section "Materials and Methods" and as previously described (28). The PspA encapsulated nanovaccine was stored at −80°C (freezer stored) or at 25°C for 60 days. The soluble PspA was stored at 4°C and mixed with Alum prior to immunization. Following storage, separate groups of mice (*n* = 8–10 per group) were immunized subcutaneously with the PspA nanovaccine stored at the different conditions or with alum-adjuvanted PspA and serum was collected and analyzed at 21 DPI. Storage conditions did not have a significant effect on anti-PspA IgG titers, with all of the PspA nanovaccines performing similarly (**Figure 4A**). Mice were then challenged with a lethal dose of *S. pneumoniae* at 24 DPI and survival was monitored for 2 weeks. All of the non-immunized mice succumbed to challenge within 4 days post challenge, while the nanovaccine-immunized mice presented with 60–70% survival with a single dose (**Figure 4B**). Once again, no significant differences were observed between the survival rates of mice immunized with the PspA nanovaccine stored at various conditions and those immunized with soluble PspA adjuvanted with Alum. These data indicate that the PspA nanovaccine is efficacious when stored both in the freezer or at room temperature for time periods up to 60 days and performs similarly to refrigerated PspA adjuvanted with Alum.

### DISCUSSION

*Streptococcus pneumoniae* is a causative agent of debilitating bacterial pneumonia. While many existing vaccines have proven to be valuable in reducing overall global disease burden, none are without limitations. Newly developed vaccine strategies against *S. pneumoniae* have identified several novel antigens, including PspA. Anti-PspA antibodies have been observed, following natural colonization or infection in children (59, 60). Other studies have shown success against lethal intranasal or septic *S. pneumoniae* challenge by vaccinating with PspA, which further indicates its potential as an efficacious vaccine immunogen (22, 61).

Aluminum hydroxide (i.e., Alum)-based adjuvants are widely used because of their ability to promote Th2 immune responses (62). However, Alum-based vaccines are associated with higher levels of immunization site tissue reactogenicity (63, 64). Immunization-site tenderness and pain are the most commonly reported symptoms following immunization with vaccines containing Alum. Prevalence of these adverse reactions has been found to be directly associated with increased protein dose administered and decreased age of the vaccinated individuals (65–67). Previous work with polyanhydride nanovaccines has shown low-level tissue site inflammation and toxicity making them a safe alternative to traditional vaccine adjuvants (32, 33, 41). Therefore, nanovaccines may represent a safe and efficacious alternative to Alum-based vaccines, which have limitations associated with administration site reactogenicity and only promote humoral immune responses against PspA.

Work from other laboratories has shown limited success using nanotechnology-based vaccine platforms with PspA in various animal models though, to date, none of these studies have shown protection against lethal challenge with single-dose immunization (68–70). Previous work from our laboratories has indicated that encapsulation of PspA into polyanhydride nanoparticles is capable of retaining protein structure and function, which enabled initiation of a humoral immune response (37). Building upon these studies, PspA was encapsulated within both 50:50 CPTEG:CPH and 20:80 CPH:SA nanoparticle formulations in the current work. Both formulations produced nanoparticles of similar size and size distribution (**Figure 1**). The size, morphology, and size distribution of these particles were similar to those previously reported for these formulations (28, 30, 37, 42). Following subcutaneous immunization of mice with a single dose of either formulation, anti-PspA IgG antibodies were produced, in agreement with our previous findings (37). Following lethal septic challenge, both nanovaccine formulations (containing 25 µg PspA) were capable of conferring protective immunity in mice, with 87.5% survival in 20:80 CPH:SA nanovaccineimmunized mice and 100% protection in those that received the 50:50 CPTEG:CPH nanovaccine (**Figure 2**). Based on the increased protection and higher antibody titers observed following immunization with a single dose of the 50:50 CPTEG:CPH

(B) *S. pneumoniae* challenge was performed at 24 DPI by i.v. administration of 2,500 CFUs of A66.1 strain and survival was monitored for 14 days. \* indicates significance (*p* ≤ 0.001) compared to saline-treated mice. Data are representative of one independent experiment containing *n* = 8–10 mice per treatment group.

nanovaccine formulation, this chemistry was selected as the lead candidate nanovaccine.

Reducing the amount of protein necessary per dose can greatly reduce the cost associated with using a recombinant protein vaccine, thereby allowing for the production and storage of more doses. Most of the published reports evaluating PspA-based vaccine success immunized mice with a minimum of 5 µg of antigen plus an adjuvant and required multiple boosts (56, 61, 71, 72). Indeed, a recent report demonstrated complete protection in mice receiving 0.5 µg of PspA adjuvanted with poly(I:C) intranasally; however, the immunization schedule required three doses to provide protection (71). The need for a single dose vaccination regimen cannot be overstated from a patient compliance standpoint because according to the WHO, an estimated 29% of children aged 1 month–5 years die each year from vaccine preventable diseases, citing non-compliance with the immunization regimen as a key reason for lack of successful national vaccination programs (73). Therefore, single-dose vaccines can aid in improving patient compliance, reduce cost, and improve national immunization rates worldwide. Recent work from our laboratories has demonstrated that 25 µg of ovalbumin encapsulated into polyanhydride microparticles was able to elicit robust antibody titers equivalent to a 16× larger soluble dose of ovalbumin while 25 µg of soluble ovalbumin alone was unable to elicit detectable antibody responses (43). In order to corroborate these results with PspA, the dose of antigen evaluated was reduced 25-fold to 1 µg. Immunization with the 1 µg PspA-based 50:50 CPTEG:CPH nanovaccine formulation, both with and without soluble PspA, induced production of high anti-PspA IgG titers (**Figure 3A**), with very little reduction in overall antibody titer compared to the previous study with a 25 µg dose of PspA (i.e., data in **Figure 2**). Antibody titers were similar between mice immunized with 1 µg of soluble PspA administered with Alum or with the nanovaccine formulations, with the highest overall titers observed in mice immunized with the fully encapsulated nanovaccine formulation. When compared to soluble protein adjuvanted with Alum, mice immunized with the 50:50 CPTEG:CPH nanovaccine were similarly protected from lethal challenge. Furthermore, 87.5% of the animals challenged with the reduced dose PspA nanovaccine (with or without soluble PspA) were protected (**Figure 3B**). These data demonstrate that reducing the protein dose by 25-fold had little effect on survival, with both the fully encapsulated PspA nanovaccine and the nanovaccine administered with a soluble bolus providing significantly higher survival following lethal challenge, compared to soluble protein alone.

By eliminating the soluble protein component from vaccine formulations, nanovaccines can be stored as a dry powder outside of refrigeration with the potential for long-term shelf storage, thereby circumventing the cold chain. As it currently stands, all prequalified WHO vaccines require refrigeration (73). Nanovaccines may enable next-generation vaccines to be stored without the need for refrigeration (i.e., eliminating the cold chain), allowing for vaccines to be sent to remote locations in need for vaccination, resulting in reduced medical costs and productivity losses associated with disease (74). Previous work from our laboratories demonstrated that nanovaccines encapsulating PspA were able to retain protein structure and activity following release, and a nanovaccine encapsulating the Bacillus anthracis protective antigen (PA) released biologically and immunologically functional protein following storage at both room temperature and above after 4 months (28, 37). By contrast, aluminum salt-based adjuvants have been reported to alter the structure of proteins and decrease thermal stability following adsorption (75), and previous work from our laboratories has shown that PA adsorbed to alum lost its functional activity when stored for 4 months at refrigerated conditions or above (28). Following storage at room temperature for 60 days, the PspA nanovaccine was shown to be efficacious, as measured by antibody titer and survival (**Figure 4**). PspA adsorbed to alum performed similarly, however, because PspA is a stable protein with coiled structure, it is likely that differences may be observed between the efficacy of the nanovaccine and alum-based formulations following storage for an extended period of time, or at elevated temperature, and should be evaluated in future studies (22). As such, this work indicates that the PspA nanovaccine formulation can be stored as a dry powder for at least 60 days while maintaining its immunogenicity.

In summary, this work describes the design of a safe next-generation PspA-based pneumonia nanovaccine with dose-sparing, protective immunity, and enhanced room temperature storage. These studies are promising first steps with respect to establishing the polyanhydride nanovaccine platform as an alternative to traditional vaccines in terms of reduced cost (enabled by the dose sparing and single dose capabilities) and difficulties associated with transportation and storage of vaccines in developing countries (enabled by the superior room temperature storage capabilities).

#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the Iowa State University Institutional Animal Care and Use Committee. The protocol was approved by the Iowa State University Institutional Animal Care and Use Committee.

#### AUTHOR CONTRIBUTIONS

DW-M, SH, and SK performed the experiments and analyzed the data. MW and BN, together with DW-M, SH, and SK, designed the experimental plan. All the authors participated in writing the manuscript.

#### FUNDING

The authors thank Dr. David McPherson of the University of Alabama at Birmingham for producing the recombinant PspA used in this work. The authors acknowledge financial support from NIH-NIAID (R56-AI075026) and the Nanovaccine Institute. BN acknowledges the support of the Vlasta Klima Balloun Faculty Chair.

## REFERENCES


model protein antigens. *J Biol Chem* (2005) 280:13406–14. doi:10.1074/jbc. M500687200

**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 Wagner-Muñiz, Haughney, Kelly, Wannemuehler and Narasimhan. 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.*

# Lipid-Based Particles: versatile Delivery Systems for Mucosal vaccination against infection

#### *Blaise Corthésy and Gilles Bioley\**

*R&D Laboratory, Division of Immunology and Allergy, Centre des Laboratoires d'Epalinges, Centre Hospitalier Universitaire Vaudois (CHUV), Lausanne, Switzerland*

Vaccination is the process of administering immunogenic formulations in order to induce or harness antigen (Ag)-specific antibody and T cell responses in order to protect against infections. Important successes have been obtained in protecting individuals against many deleterious pathological situations after parenteral vaccination. However, one of the major limitations of the current vaccination strategies is the administration route that may not be optimal for the induction of immunity at the site of pathogen entry, i.e., mucosal surfaces. It is now well documented that immune responses along the genital, respiratory, or gastrointestinal tracts have to be elicited locally to ensure efficient trafficking of effector and memory B and T cells to mucosal tissues. Moreover, needle-free mucosal delivery of vaccines is advantageous in terms of safety, compliance, and ease of administration. However, the quest for mucosal vaccines is challenging due to (1) the fact that Ag sampling has to be performed across the epithelium through a relatively limited number of portals of entry; (2) the deleterious acidic and proteolytic environment of the mucosae that affect the stability, integrity, and retention time of the applied Ags; and (3) the tolerogenic environment of mucosae, which requires the addition of adjuvants to elicit efficient effector immune responses. Until now, only few mucosally applicable vaccine formulations have been developed and successfully tested. In animal models and clinical trials, the use of lipidic structures such as liposomes, virosomes, immune stimulating complexes, gas-filled microbubbles and emulsions has proven efficient for the mucosal delivery of associated Ags and the induction of local and systemic immune reponses. Such particles are suitable for mucosal delivery because they protect the associated payload from degradation and deliver concentrated amounts of Ags *via* specialized sampling cells (microfold cells) within the mucosal epithelium to underlying antigen-presenting cells. The review aims at summarizing recent development in the field of mucosal vaccination using lipid-based particles. The modularity ensured by tailoring the lipidic design and content of particles, and their known safety as already established in humans, make the continuing appraisal of these vaccine candidates a promising development in the field of targeted mucosal vaccination.

Keywords: mucosal, vaccination, lipidic particles, delivery system, infections

## INTRODUCTION

Vaccination is considered as one of the most successful medical actions and has greatly contributed to the improvement of world health. Indeed, it has strikingly reduced the prevalence of many infectious diseases, and thus helps nowadays to save millions of lives each year (1, 2). Vaccine administration aims at inducing and harnessing protective effector and memory

#### *Edited by:*

*Rajko Reljic, St George's, University of London, United Kingdom*

#### *Reviewed by:*

*Vijay Panchanathan, Perdana University, Malaysia Beatrice Jahn-Schmid, Medizinische Universität Wien, Austria*

> *\*Correspondence: Gilles Bioley gilles.bioley@chuv.ch*

#### *Specialty section:*

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

*Received: 22 December 2017 Accepted: 19 February 2018 Published: 07 March 2018*

#### *Citation:*

*Corthésy B and Bioley G (2018) Lipid-Based Particles: Versatile Delivery Systems for Mucosal Vaccination against Infection. Front. Immunol. 9:431. doi: 10.3389/fimmu.2018.00431*

**99**

immunity, comprising neutralizing antibodies (Abs) together with cytotoxic and helper T cells (3) able to control subsequent challenge by the target pathogen. Live-attenuated or killed whole-pathogens have originally been administered for vaccination purposes, but due to safety concerns, including important reactogenicity and risks of reversion, the use of subunit vaccines is preferred. The latter are composed of recombinant or purified pathogen-derived antigenic entities, mostly depleted of innate immune stimulus, that require the co-administration of adjuvants and/or the use of delivery vehicles to achieve sufficient immunogenicity. Over the last decades, important pieces of work in the field of vaccine technology have allowed to rationally design and develop formulations that ensure efficient induction of immune responses (4). Synthetic micro-/ nanoparticles, liposomes, immune stimulating complexes (ISCOMs), virosomes, virus-like particles, as well as emulsions, all offer several interesting attributes for vaccine delivery and have already proven efficient in parenteral (intramuscular or subcutaneous) vaccinations by inducing protection against infectious agents (5). These formulations have been designed to mimic biophysical and biochemical features of pathogens, thus ensuring efficient display and delivery of concentrated amounts of antigens (Ags) and adjuvants to innate and adaptive immune cells. Interestingly, this leads to reducing the number of injections required to elicit potent cellular and humoral immune responses with minimal cytotoxicity.

Despite important success in protecting individuals against many deleterious pathological situations, it remains that most of the licensed subunit vaccines are administered parenterally. However, except in previously infected individuals, such a route of administration only induces limited protective effect at the level of mucosal surfaces, the sites where the vast majority of pathogenic agents gain access to the host body (6). In addition to mechanical (epithelium covered with mucus) and chemical (anti-microbial peptides) barriers found at mucosae, adaptive humoral and cellular immunity is of prime importance to efficiently protect against pathogenic insults (7). Thus, to reinforce the efficiency of vaccination, the delivery of vaccine formulations directly to the mucosa represents an asset. It is now well accepted that immune responses have to be elicited locally to ensure efficient imprinting of effector and memory B and T cell homing to mucosal tissues where they will limit entry, colonization, and spreading of pathogens (8–11). Until now, the few licensed mucosal vaccines consist in administration of live-attenuated or killed whole-pathogens that raise similar safety concerns as for parenteral injection, while no subunit vaccines have been approved for human use. This is mainly due to technical difficulties inherent to the administration route and the physiology of the tissues where the vaccine formulations are applied. Identifying the most adequate vaccine formulation deliverable mucosally remains challenging due to (1) the fact that, in contrast to parenteral vaccination where injected Ags and adjuvants are directly in contact with antigen-presenting cells (APCs), Ag sampling has to be performed across the mucus and the epithelium first; (2) the deleterious acidic, proteolytic, and dynamic environment of the mucosal surfaces which impact the stability, integrity, and retention time of the applied Ags; and (3) the tolerogenic nature of the mucosa, which impairs induction of effector immunity to antigenic entities lacking sufficient immunostimulatory signals (12). Such hurdles may, however, be partially overcome thanks to recent progress made in the understanding of mucosal immunity and in the field of vaccine technology (6).

Apart from immunological and physiological aspects, one important point to be considered for vaccination is the compliance of the patients (13). For pediatric vaccination, administration has to be minimally invasive and easy to perform. The ability of vaccinating a large number of people in countries where endemic infections are present, but where access to medical infrastructures is limited, is of great importance as well. In this context, parenteral vaccination is not the most appropriate strategy as injections are invasive, painful and require trained/ skilled medical staff for administration. Moreover, it poses problems related to the risks associated with infection at the site of injection, needle-stick injury, spreading of transmissible diseases, and disposal of used materials. Thus, there is an increasing demand for needle-free vaccination. As an example, mucosal vaccines display several advantages, such as ease of administration and self-delivery allowing mass vaccination, absence of needle-associated risks, and in some cases lower costs and simplified production due to absence of administration devices.

Until now, only few mucosally applicable subunit vaccine formulations have been developed and successfully tested (14), mainly because of the limited number of safe and efficient delivery systems and adjuvants available, coupled to the sometimes important amounts of Ag to be administered. This review will focus on lipid-based micro-/nanoparticles that possess several of the desired characteristics of an interesting Ag-delivery system for vaccination as they are biocompatible, can overcome physiological barriers at mucosae, promote Ag crossing of the epithelium and uptake by APCs, protect the associated payload, are adequate for incorporating adjuvants and may display mucoadhesive properties. In order to achieve the induction of protective anti-pathogen humoral and cellular responses at the relevant mucosal surfaces, the choice of the most potent administration route has to be carefully considered by taking into account the physiological and immunological features of the different target tissues (15). These aspects and the strategies to specifically target vaccines to the portals of entry across the epithelium and increase the efficiency of delivery will first be discussed. However, directing vaccines to the appropriate location is not sufficient to ensure optimal vaccination effect. The architecture, size, and surface chemistry of particles are of prime importance and can be manipulated to influence the intensity and type of immune responses. Physicochemical properties of lipid-based particles, Ag incorporation, mucoadhesion, and association with adjuvants will be discussed next. Examples of mucosal application of such formulations in animal models and their outcome will then be presented. Finally, an overview of the current evaluation of lipid particles and open challenges of mucosal vaccination in humans will be considered.

## VACCINE SAMPLING AT MUCOSAL SURFACES AND THE SELECTION OF THE ROUTE OF ADMINISTRATION

When applying a vaccine formulation *via* any delivery route, the anatomical, functional, and immunological characteristics of the different tissues have to be considered (13). Indeed, the structure and spatial organization of the tissues, the presence of mucus and mechanisms to eliminate deposited materials/particles (e.g., peristaltism in the intestine and physical discharge in the respiratory tract), the pro-tolerogenic environment of mucosae and the presence and localization of particular immune cell subsets, especially dendritic cells (DCs), all impact on the outcome of vaccine administration (16). In addition, safety issues have to be considered. In this section, we will present the characteristics of the mucosal immune system in relationship with vaccination, the different mucosal administration routes, as well as the strategies under evaluation to increase the efficiency of vaccine delivery.

## The Mucosal Immune System and Mucosal Vaccination

Mucosal surfaces are continuously exposed to, and challenged by, numerous environmental Ags present, for example, in food, air, or derived from pathogenic or commensal microorganisms in the lumen. On top of the epithelial barrier covered by mucus and the secretion of anti-microbial peptides, a specialized and complex immune network, called mucosa-associated lymphoid tissues, is involved in immunosurveillance of mucosal tissues (17). Lymphoid cells and effector molecules, such as secretory IgA (SIgA), the chief Ab molecule operating at mucosal surfaces (18), cytokines, and chemokines, tightly orchestrate protection against infections and maintenance of tolerance toward endogenous unharmful microorganisms. Sampling for such agents and their delivery to immune cells underneath the epithelial layer takes place *via* direct uptake by DCs within the epithelium (19–21) or across specialized epithelial cells named microfold (M) cells that are responsible for the selective transport of macromolecules, particulate Ags, and microorganisms (22, 23). Internalization *via* M cells occurs through different mechanisms (clathrin-coated endocytosis, actin-dependent phagocytosis, or macropinocytosis) depending on the nature of the Ags. M cells are present in (a) the follicle-associated epithelium that separates the intestinal lumen (apical side) from underlying immune cells (basolateral side) in Peyer's patches (PPs), (b) in intestinal isolated lymphoid follicles, (c) in nasopharynx-associated lymphoid tissues (NALT), and (d) in bronchial-associated lymphoid tissues (BALT) (24). Such structures are composed of innate and adaptive immune cells, including functionally different DC subsets, T, and B cells. DCs integrate signals derived from the sensing of the luminal environment, and release soluble factors, such as cytokines and chemokines, to orchestrate the generation of tightly controlled mucosal immunity locally or after migration in regional lymph nodes (LNs) (7). In addition, paracellular and transcellular uptake of macromolecules and small particles across the epithelia lead to their uptake by APCs outside inductive sites for induction of local immunity *via* regional LNs. By contrast, in the urogenital tract and in the oral cavity, the stratified epithelium does not contain M cells and sampling occurs by DCs interspersed within the tissue leading to induction of immune responses exclusively in draining LNs (15, 25, 26).

Upon encounter with microorganisms or vaccine formulations, mucosal DCs in combination with neighboring epithelial cells control the expression of specific homing receptors on primed lymphoid cells and modulate the type of ensuing immune response (8, 9, 27). Such imprinting relies on the expression of site-specific integrins and chemokine receptors by B and T cells and allows their transit *via* the lymph and through the blood to migrate to different mucosal sites. Recirculation of lymphocytes to the gut requires expression of α4β7 and CCR9, whereas migration to the airways, the oral cavity, and the reproductive tract relies on L-selectin and CCR10. In the case of pathogenic infections, danger signals generated by the sensing of microorganisms switch immune responses toward an effector type of response relying on both humoral and cellular arms to eliminate the infection (28). Immune exclusion and neutralization by SIgA, as well as production of Th1- or Th17-type cytokines that activate phagocytes and induction of cytotoxic T cells, all contribute to the protection of mucosal surfaces (29, 30). Therefore, the major aim of vaccines would be to elicit specific B and T cell responses at the relevant sites to induce specific SIgA that provide a first line of protection against invading pathogens, together with appropriate cellular immune responses to eliminate both the pathogen and pathogen-infected cells. For example, requirements for proper B cell isotype-switching and the generation of IgA responses include mainly the production of TGF-β, IL-6, retinoic acid, and IL-21 by PPs' cells, together with the CD40–CD40L interaction between T follicular helper cells and B cells (31). Thus, vaccine formulations for mucosal application have to be designed to best induce such immunological environment.

## Administration Routes for Mucosal Vaccination

Each route of mucosal administration has its own characteristics and a balance between the pros and cons for each vaccine has to be considered, taking into account the pathogen to fight against and the formulation to be delivered. However, no standardized studies are available to directly compare safety, profile of induced immune responses, and efficiency of protection. In this section, we will consider the different mucosal administration routes. Oral and nasal/pulmonary are the most studied ones, but sublingual is now recognized as a promising way of vaccination. Vaginal and rectal delivery have also been studied, but more scarcely. Even though the most powerful response is usually elicited in the local inductive and adjacent tissues, the common mucosal immune system predicts that homing to distant mucosal tissues is possible (27). However, a certain degree of compartimentalization does not allow imprinted cells to migrate to every mucosal sites. Such flexibility allows to select for the most appropriate route of vaccination to induce protective immune responses at the desired site. Of note, some recent works demonstrated that transcutaneous immunizations have the potency to promote the induction of immune responses with the ability to traffic to the gut and airways, although up to now with low consistency (32). However, this aspect will not be covered by this review and has been described elsewhere (33, 34).

Nasal administration represents a promising route for mucosal vaccination (35), because nasal tissues display a relatively large surface for Ag absorption covered with only a thin layer of mucus, and are highly vascularized. It does not require the delivery of high Ag doses (e.g., as compared to oral administration), is non-invasive and easily accessible for self-administration. Nasal vaccination allows the generation of a broad range of Ab and T cell responses at different mucosal sites, such as the upper (preferentially) and lower airway mucosae, the local secretions, the salivary glands, and the urogenital tract. It also elicits concomitant robust systemic immunity (15). However, in the nasal environment, the presence of proteases and the local pH, together with a relatively high mucoscillary clearance rate, may impact on the vaccine integrity and retention time, thus affecting the generation of immune responses. The major drawback concerns the safety of nasal administration, as physiological function such as smell perception might be altered by vaccine-induced inflammation and the close relationship with the brain might promote health problems. Thus, every vaccine candidate has to be evaluated carefully for safety. Sometimes achieved by nasal delivery or directly targeted, pulmonary immunization allows vaccine formulation to directly access the respiratory tract which is of interest due to its high permeability, its large surface area and the high density of APCs (alveolar macrophages, DCs, and B cells). This route of administration preferentially induces immune responses in the lower airways and has interestingly been shown to promote cellular and humoral responses in the gut. However, efficient delivery in the lung is not an easy task. Delivery *via* the nasal and pulmonary routes does not necessarily lead to similar outcomes: for example, pulmonary vaccination was shown to be more effective than its nasal counterpart at protecting against *Mycobacterium tuberculosis* infection, because different immune mechanisms were involved after one or the other administration route (36–38). Indeed, elevated levels of SIgA were produced in the lung after pulmonary vaccination, with equivalent responses observed in the nasal passage. In addition, IFN-γ production in the lung following pulmonary vaccination was important to fight against *M. tuberculosis*, whereas there was apparently no role for this cytokine in the nasal environment.

Oral administration represents an interesting strategy in terms of ease of delivery, patient compliance, and safety (39). However, due to the intrinsic high dilution of vaccine formulations and the harsh environment of the digestive tract, substantial amounts of Ags have to be administered. Indeed, the extremely low pH in the stomach, proteolytic enzymes and bile salts in the intestine, the presence of relatively thick one-layered mucus, and the overall low permeability of the intestine greatly affect the integrity and delivery of applied Ags. In this context, oral vaccines are likely to be more efficient if repeated doses are given, provided that adjuvants are incorporated to avoid tolerance induction (13). Oral administration is the most efficient delivery route to achieve induction of gut immunity, which is of high importance to fight against the large burden of enteropathogenic infections worldwide. Induction of immune responses in the colon, stomach, mammary, and salivary glands, as well as systemically, also takes place, but with limited robustness (15).

Sublingual immunization generates immune responses with similar profile and mucosal tropism as nasal delivery, i.e., vigorous and broadly disseminating mucosal and systemic IgA and IgG, as well as helper and cytotoxic T cell, responses (40), without many side effects (41), and formulation concerns associated with nasal or oral immunization (42). It is also easily accessible for self-administration. It has been shown to induce immune responses after administration of soluble Ags, particulate Ags, live/killed bacteria, and viruses (40). Sublingual delivery is interesting because the oral cavity has a milder environment that may not degrade vaccine components, and may, thus, not require large amounts of Ags. As an example, higher Ab responses were obtained in mice after sublingual, as compared to oral, administration with about 10–50 times less Ag applied (43). One limiting factor is the absence of Ag-sampling M cells in the oral cavity lined by a stratified epithelium and the relatively low number of DCs in the upper layer of oral tissues. However, vaccine formulations can be taken up by lingual tonsils for delivery into regional LNs and Langerhans cells in the oral epithelium have been shown to act as potent inducers of immunity (15). Several delivery systems (microneedles, liposomes, inactivated microorganisms) and adjuvants [toll-like receptor (TLR) ligands, cholera toxin (CT), mutants of heat labile toxin (LT) and CT] have been evaluated for sublingual vaccination and protective Th1-type responses in the lung, genital tract, and the gut, together with SIgA in the saliva, intestinal, and vaginal washes, have been obtained with different vaccine formulations (42–45).

Vaginal immunization elicits immune responses in the genital tissues and secretions, but is not efficient to induce systemic immunity. Despite relatively low pH, the vagina is a mild environment that does not impair Ag integrity and, thus, allows to limit the amount of Ag to be delivered (25). However, the presence of a stratified epithelium and the absence of inductive sites imply that induction of vaccine response *via* vaginal delivery requires specific adjuvanted formulations and DC subsets (46). In addition, the changes occurring in term of immunological functions during the estrous cycle complicate such immunization (47–49). Additional studies are required to better understand mucosal immunity in the urogenital tract and define specific requirements for vaccine formulations. Rectal immunization is able to induce potent immune responses in the small intestine and the colon, but not efficiently in the systemic compartment (15). Only limited studies are available to fully appreciate the potential of such an administration route for mucosal vaccination.

## Targeted Delivery of Vaccine Ags

Not only do vaccine formulations have to resist the deleterious environment of some mucosal surfaces, but they also have to face an additional hurdle that is to cross the epithelium to gain access to underlying APCs. In this context, targeting the relatively low number of portals of entry at inducing sites, e.g., M cells that represent 1% of epithelial cells (5–10% of enterocytes within the follicularassociated epithelium), or DCs spread within the epithelium is an asset for efficient vaccination. DC targeting by the mean of C-type lectin receptors (DEC205, DC-SIGN, mannose receptor) or specific Abs directed against DC markers has proven to be an efficient strategy to improve the potency of parenteral vaccination (50). Similar strategies have been developed for mucosal vaccination, such as targeting of Langerin on DCs of the oral cavity, the esophagus or the vaginal mucosa (51). FcRn expressed by airway and gut epithelial cells (52, 53), as well as some DC subsets (54), has also been demonstrated as an efficient strategy to deliver IgGbased complexes across the epithelium and to underlying DCs. Such an approach efficiently led to the induction of both CD4 and CD8 T cell effector responses (55–57). Galactosyl ceramide may function as a targeting moiety in the intestine, the rectum, and the endocervical mucosa (58), and the ganglioside GM1 molecule can be targeted by a specific peptide developed by phage display (59). As far as M cells are concerned, specific delivery can be achieved *via* different strategies: the tight junction molecule claudin-4 (60), the bacterial FimH receptor GlycoProtein-2 (61, 62), the complement C5a receptor and its ligand Co1 (63, 64), a M-cell-specific peptide referred to as CKS9 (65), or the unique glycosylation pattern involving α-1-fucose; for the later, the use of Ulex Europaeus Lectin-1, or a specific monoclonal Ab have been successfully demonstrated (66–68). However, in humans, the lack of expression of this particular sugar moiety on M cells (69), together with the extra-M cell expression of GP-2 (70), precludes the use of such strategies for specific targeting purposes. By contrast, a promising approach consists in coupling vaccine Ags with SIgA in order to induce M-cell-specific retrotransport across the epithelium and DC targeting *via* Dectin-1 in both mice and humans (71). This Ab molecule can potentially serve as a cargo for the controlled delivery of the associated payload as this occurs naturally with microorganisms sampled from the mucosal lumen (72). An additional advantage is the resistance of SIgA to protease degradation and its ability to anchor in mucus (18), two features that may improve both stability and retention time of the associated Ags.

## FORMULATION CONSIDERATIONS

Particulated Ags have been designed to mimic the shape, size, and antigenic display of pathogens with the aim of improving vaccine efficiency (73). Ags associated with micro-/nanoparticles display increased depot effect upon administration, are better protected from degradation, and are more efficiently taken up by APCs and presented to B and T cells than soluble Ags (74). An important number of studies have evaluated the effect of particle properties (type of material, size and charge, Ag incorporation, or association of adjuvants) on the profile and strength of induced Ab and T cell responses. When composed of natural lipids, lipid-based particles have the advantage of being biocompatible. In addition, they are very flexible in terms of formulation, implying that lipid exchange within the particle shell is achievable, leading to modulation of their physico-chemical properties. This is of importance in the biological environment because all these parameters will influence the stability and immunological consequence of delivered particles. However, minor changes in the composition of the particle may impact on its efficiency and protective ability, meaning that any formulation needs to be individually evaluated *in vivo*.

## Size and Charge of Particles

The size of particulate Ags has an impact on the type of immune responses that are generated because it influences the mechanism of uptake by APCs. Indeed, receptor-mediated endocytosis, pinocytosis, macropinocytosis, or phagocytosis, all lead to different ways of trafficking within the cells and, therefore, induce preferentially presentation *via* the MHC I or MHC II pathway for CD8 or CD4 T cell priming, respectively (75). Small particles (up to 200 nm) are sensed as viruses and are taken up by receptor-mediated endocytosis leading to predominant T cell responses, whereas larger particles (more than 500 nm) are taken up *via* micropinocytosis or phagocytosis to preferentially induce Ab responses (76). Similar size-dependent uptake by M cells or enterocytes takes place, leading to differential sampling of the particulated Ags. Other studies demonstrated that vesicles smaller than 250 nm induced a balance toward Th2-type of responses, whereas the opposite was observed for larger vesicles (77–79). Moreover, in the context of mucosal delivery, the size of the particles influences the tissue localization and the diffusion across the mucus. Following nasal administration, small particles are better transported across the nasal mucosa, whereas larger ones are better deposited in the respiratory tract to be taken up by alveolar macrophages (80, 81). In order to get access to the epithelium, both viscosity and pore size of mucus can impact on the penetration of vaccine components. Apparently, the average size of pores in the mucus is in between 200 and 500 nm, e.g., in the cervicovaginal and small intestinal mucus. This suggests that particles smaller than this cutoff freely diffuse across the mucus, whereas larger ones take more time to reach the epithelium or possibly never reach it (82, 83). One major point to be considered is that correlation between size and immunogenicity is difficult to strictly assess for lipid-based particles, because homogeneous and monodispersed preparations have been challenging to obtain, and when feasible, such preparations require technical issues that may dramatically increase the cost of vaccine formulations.

Diffusion across the mucus is not only governed by size of particulated Ags or mucus pores but also by chemical characteristics such as the surface charge of particles. Hydrophobic and electrostatic interactions mediated by particles aggregate mucus microstructure and impede diffusion of vaccines, while hydrophilic and neutral vaccine formulations promote mucus penetration (23). Mucoadhesion is promoted by positively charged particles that interact with negatively charged mucus. For example, electrostatic interactions between cationic lipids and the nasal mucosa promote enhanced contact time with the tissue, higher local concentration, and thus improved penetration of liposomes (84). Similarly, cationic particles better interact with negatively charged cell membranes, such as those of M cells and enterocytes, therefore limiting vaccine clearance and improving sampling *via* endocytosis or membrane fusion (85, 86); it also improves the uptake by DCs (87). However, cationic particles may have charge-dependent cytotoxicity against target cells; therefore, the density of cationic lipids within the particle shell has to be carefully defined and a tight balance between strong adhesion and safety has to be achieved (23). Interestingly, it seems that the presence of the mucus limits cytotoxicity (85), meaning that cationic particles keep their validity for mucosal vaccination.

#### Corthésy and Bioley Mucosal Vaccination with Lipid-Based Particles

#### Incorporation of Ag

There are different ways of associating Ags to lipid-based delivery systems and the choice depends mainly on the administration route and the nature of the Ags (88, 89). For oral administration, encapsulation seems favorable in terms of ensuing immune responses, because it prevents rapid degradation of the Ag within the gastrointestinal environment and, hence, increases its half-life (90). Encapsulation of Ags is relatively easy to perform during the manufacture process, but this may alter antigenic structures. By contrast, maintenance of the integrity of the Ags is less affected by the nasal route, suggesting that surface association *via* charge interaction is sufficient. Such an approach is technically not demanding, owing that opposite charges of either the particles or the Ags favors it (90). Alternatively, covalent binding at the surface of particles is achievable, although more complicated to perform; this precludes the undesired release of the payload that may occur within the tissue environment. In terms of immune response induction, encapsulation within liposomes preferentially induces IgG production, whereas surface display of the Ag induces both IgM and IgG responses (91), with elevated levels (92). In addition, Ag density at the surface of particles, as well as the Ag-to-lipid ratio, has been documented to influence the elicited immune responses following immunization (93, 94). This may suggest that both encapsulation and surface location of the Ag within the same formulation would promote optimal induction of T cell and B cell responses.

Apart from protein Ags, plasmid DNA coding for pathogenderived Ags have been evaluated for vaccination (95, 96). Such strategy has an established record of efficacy in preclinical studies and can be safely used in humans, even in immunocompromised individuals. However, based on results obtained in the field of veterinary vaccination, naked DNA induces only weak immune responses. In order to improve immunogenicity, association of DNA with cationic liposomes leads to increased uptake by target cells and delivery into the nucleus. DNA immunization through the nasal or the oral route can effectively induce protective humoral and cellular immunity at related mucosal surfaces, but necessitates association with cationic delivery systems, presumably to increase mucus penetration, to reduce mucociliary clearance, and to improve permeation across the epithelium (97, 98). DNA sequences such as the canonical CpG motifs have been shown to display immunostimulatory properties. In a similar way, messenger RNA-based vaccines, when approprietely protected from ribonucleases are translated in the cytoplasm and do not require nuclear transport (99). This has been mainly evaluated with cationic lipid-based vesicles in the context of cancer immunotherapy, and the efficient nasal application of particle-associated mRNA has been demonstrated (100).

#### Mucoadhesive Properties

Upon mucosal administration, vaccine formulations are diluted in mucosal fluids and have to face bulk flow, leading to limited retention time and suboptimal access to the epithelium for sampling. Such deleterious effects can be compensated by incorporation of mucoadhesive and mucus-penetrating components. In this case, the surface structure of lipid-based particles has to be carefully designed to obtain an adequate balance between strong adhesion and mucus penetration. Some possible strategies are described below. The first one consists in incorporating polyethylene glycol (PEG) at the surface of particles. PEG has originally been used for systemic administration in order to avoid adsorption of plasma proteins and the formation of a corona that may mask targeting ligands, adjuvants, or Ags at the surface of particles (101). In addition, the presence of PEG increases the stability upon administration. A similar stabilization effect has been reported in the case of oral or sublingual delivery of liposomes (93, 102, 103). PEG is a hydrophilic component that has been reported to help particles to penetrate the nasal mucosa by preventing aggregation and thus facilitating diffusion across the mucosal barrier. Moreover, it can form hydrogen bonds with mucus leading to mucoadhesion, but also helps diffusion across the mucus; indeed, such intriguing bifunctionality has been correlated with the molecular weight of PEG. High molecular weight polymers (>PEG5000) are preferentially mucoadhesive whereas lower ones (PEG2000) better diffuse within the mucus (103–105). An additional non-negligible advantage of PEG is that it provides cryopreservative functions during particle manufacture.

The second strategy is to associate with micro-/nanoparticles some mucoadhesive components, such as chitosan (deacetylated chitin), alginate, polyvinyl alcohol, hyaluronan, or cellulose derivatives that all boost particle-based vaccination (106–108). Addition of bioadhesive components (xanthan gum or tramella) within formulations helps to increase the viscosity of the vaccine and, thus, the retention time at mucosal surfaces. The most studied mucoadhesive molecule is chitosan, whose relevant properties for mucosal vaccination are as follows: (1) it is a positively charged molecule that can interact with negatively charged mucus to improve adhesion; (2) it is a permeation enhancer due to its ability to transiently open epithelial tight junctions and, thus, improve Ag sampling (109); and (3) it has adjuvant properties, promoting induction of IFN-γ, IgG, and SIgA (110). Chitosan has been explored for delivery *via* oral, nasal, and pulmonary routes in association with liposomes leading to increased stability and mucoadhesion for absorption by mucosal surfaces (111). Interestingly, it did not demonstrate detrimental effects toward mucosal tissues (112).

#### Incorporation of Adjuvants

Several adjuvants have been evaluated during the last decades for mucosal vaccination (113). Essential properties of the ideal adjuvant include the following: to be effective with low-dose Ag; to be suitable with many different Ags; to be effective enough to reduce the number of vaccine administrations; to be able to induce long-term immune responses; and to display limited or absent toxicity. Innate immune triggers have been used as adjuvants as they have the capacity to elicit pro-inflammatory responses to recruit phagocytes, to enhance Ag presentation by APCs, and to activate APCs in order to generate the adequate environment for efficient priming of adaptive immunity. Studies in animals have demonstrated an important adjuvant effect of the *Vibrio cholerae* endotoxin CT and *Escherichia coli* LT, ensuring enhanced Ag permeation through the epithelium, enhanced targeting of M cells, increased Ag presentation by DCs and improved activation of DCs (114); a direct effect on B and T cells has additionally been reported (15). However, such adjuvants are inadequate for human use because of their toxicity and unacceptable side effects, as for example: induction of deleterious inflammatory response leading to altered function of olfactory nerves or to Bell's palsy after nasal administration, and diarrhea symptoms after oral administration (115, 116). This has oriented research toward the generation of less toxic derivatives engineered by introduction of mutations in the A subunit of CT and LT (117–120). The most promising derivative is the double mutant LT (R192G/L211A, dmLT) that has no demonstrated side effects in animal application while retaining important adjuvant activity after oral or sublingual administration (121–123). Similar mutations R192G/L211A applied to CT similarly reduced its toxicity, although to a level still not acceptable for human use. Introduction of additional mutations within the amino acid 189–197 stretch recently demonstrated more safety with an ability to induce both Ab and T cell responses close to that of the non-mutated CT following nasal, oral, and sublingual vaccinations (124). An alternative approach is the use of the B subunit of LT or CT only. LTB and CTB are not very efficient *via* the oral route, however, nasal administration demonstrated some efficiency when the Ag was physically linked to the adjuvant resulting in increased uptake across the epithelium and by DCs (15). Fusion protein obtained by association of the A1 subunit of CT and *S. aureus* protein A derivative (CTA1-DD) proved efficient at boosting B cell responses after nasal administration (125). In another report, edema toxin from *Bacillus anthracis* and diphtheria toxoid within lipidic particles have been evaluated for nasal administration and showed efficient induction of immune responses leading to reduced bacterial load after pathogen challenge (126).

In parallel, evaluation of immunostimulatory molecules active for parenteral administration, such as TLR ligands, have been performed (113). CpG oligodeoxynucleotides (CpG), monophosphoryl lipid A (MPLA), and flagellin that are ligands for TLR-9, TLR-4, and TLR-5, respectively, have been administered orally or intranasally and demonstrated immunostimulatory properties for mucosal immune responses, including induction of SIgA (127–131). Pulmonary delivery of a *M. tuberculosis*-derived Ag together with CpG or MPLA promoted the generation of IFN-γ production in the lung; MPLA was more potent to induce IL-17 production and to decrease the bacterial load following challenge (132). Flagellin, expressed by different pathogenic bacteria, can indirectly stimulate local DCs following nasal delivery, and induce mucosal IgA responses and protection upon Influenza vaccine administration (133, 134). Alternatively, trehalose dibehenate (TDB), a synthetic analog of a *M. tuberculosis* cord factor known to interact with Mincle and promote Th1/Th17 type of responses (135), has also been shown to be effective by the nasal route (136). Saponin QS21 also demonstrated potent adjuvant effect when nasally administered (128, 129). In addition, STING ligands 3′3′-cGAMP, c-di-AMP, and c-di-GMP have been efficiently delivered *via* the nasal or the sublingual route to elicit Th1/Th17 responses and high-affinity SIgA (137). Activation of NKT cells by administration of α-galactosylceramide is also of interest for nasal, oral, and sublingual vaccination due to its ability to enhance immunogenicity of different mucosal vaccine formulations (138–140). All these adjuvants can be incorporated within lipid-based particles or associated at their surface depending on the localization of their cognate receptor in target cells. This has been successfully achieved and resulted in improved uptake by and activation of DCs (141). Moreover, cationic lipids *per se* have been shown to directly activate APCs (142).

## LIPID-BASED PARTICLES FOR MUCOSAL VACCINATION

In order to induce efficient and protective immune responses by vaccination, not only the Ags and the adjuvants have to be carefully defined, but also an appropriate delivery system is of prime importance. When aiming at using mucosal routes of administration, they must be designed to resist chemical degradation by low pH, proteolytic enzymes, and the harsh environment of mucosal surfaces. Lipid-based particles represent interesting delivery systems to incorporate Ags and adjuvants, allowing targeted and concentrated delivery of relatively low amounts in tissues, together with limiting toxicity associated with potential spreading of the payload (**Figure 1**). Many lipid-based particles have been tested in animal models of immunization and/or infection and are reviewed below (**Table 1**). Advantages, limitations, and necessary refinements for use as effective mucosal vaccine are discussed sequentially.

#### Liposomes

The enormous potential of liposomes for drug delivery has been acknowledged for decades. Indeed, they display features including controlled release, protection from degradation, improved pharmacokinetics, increased circulation time, and targeting to specific tissues (143, 144). They have been progressively adapted for administration of diverse antigenic entities, such as proteins, peptides, and DNA, in order to produce vaccine formulations to fight against several viral and bacterial infections (89). Liposomes are spherical vesicles consisting in unilamellar or multilamellar shell of phospholipid bilayer(s) entrapping an aqueous core and range in size from tens of nanometers to several micrometers in diameter. The amphiphilic nature of phospholipids mediates self-assembling of liposomes in an aqueous environment leading to a bilayer configuration. They can incorporate both hydrophilic molecules encapsulated within the aqueous core and hydrophobic molecules hooked at their surface or inserted within the inner hydrophobic space of the lipid bilayer. Biocompatible neutral and anionic phospholipids, such as phosphatidyl cholines [e.g., distearoylphosphatidylcholine (DSPC)], and cholesterol, are the most commonly used constituents of the shell that ensures proper stability of the structure and improved immunogenicity of the formulation (145, 146). The length and degree of saturation of acyl chains influence both the permeability and the fluidity of the shell, leading to increased or decreased stability. Liposomes are versatile delivery systems that are interesting for vaccination formulations because their physicochemical properties can be modulated by altering their composition in lipids. Among possible modifications, pH titrable lipids to induce controlled release of payloads (147, 148) and synthetic cationic lipids to

linked at the surface of particles. Not drawn on scale.

improve immunogenicity (90) have been generated. In addition, functionalization of liposomes with specific targeting moieties has emerged as a promising strategy to improve delivery. Vesicles bearing the DC-SIGN-specific ligand Lewis x glycan showed increased DC-targeting properties and subsequent activation of T cells, especially when adjuvanted (141) and IgG-coupled liposomes have demonstrated enhanced transmucosal transport in nasal tissues (55).

Cationic liposomes prepared with dioleoyltrimethylammoniumpropane (DOTAP), dimethyldioctadecylammonium bromide (DDA), dimethylaminoethane-carbamoyl (DC)-cholesterol have been successfully evaluated. Nasal administration of DDA-based liposomes induced greater local and vaginal IgA production as compared to vesicles without cationic lipid. Moreover, the incorporation of PEG further increased the observed immune responses (149). Similarly, delivery of cationic liposomes composed of DOTAP and DC-cholesterol *via* the nasal route allowed efficient uptake by DCs in NALT and subsequent induction of specific IgA and T cells in nasal tissues (150). The adjuvant CAF01 is a prime example of efficient cationic liposomes to be used for mucosal vaccination. Incorporation of both DDA and the immunostimulatory molecule TDB, has been evaluated in several animal models of infections with Influenza, *Chlamydia* and *M. tuberculosis* (151). In such context, the presence of the adjuvant had a substantial beneficial effect on immunogenicity (152). Nasal vaccination against Influenza or *Streptococcus pyogenes* with CAF01-based formulations allowed to generate mucosal effector T cell and IgA responses and to protect vaccinated animals (136, 153). Furthermore, preparation of liposomes with the cationic lipid ceramide carbamoylspermine efficiently stimulated systemic and mucosal immunity following intranasal administration (154). The use of cationic preparations is also an interesting approach for alternative forms of antigenic entities, as liposomes incorporating DOTAP and a plasmid DNA coding for a mycobacterial heat-shock protein given nasally induced local mucosal immune responses able to reduce *M. tuberculosis* load in the lung (155). Overall, liposome-based vaccination *via* the nasal route leads to the induction of robust immune responses whatever the nature of the Ag and its mode of incorporation. Thus, fine-tuning modulation of the profile of vaccine-elicited responses appears to depend on the composition of the formulation, including the type of lipids and/or the presence of adjuvants. Interestingly, most


liposomal preparations seem to be well tolerated, inducing only limited inflammatory responses, irritation, sneezing, or burning syndroms. Oral administration of liposomes has been documented, but its stability in the gastrointestinal tract remains the main concern. As already discussed in Section "Mucoadhesive Properties," promising approaches can be envisaged to improve the stability of liposomal preparations. Incorporation of mannose, chitosan, and PEG are all possible scenarios resulting in reinforced stability, better targeted delivery across the epithelium and to APCs, and improved immunogenicity. Stabilization of liposomes with layer-by-layer deposition of polyelectrolytes also increased the generation of Ab and T cell responses in mucosal tissues (156). The administration of multilamellar preparations is an alternative strategy. Finally, as discussed in Section "Administration Routes for Mucosal Vaccination," vaccination *via* the sublingual route is a promising development that requires to be further evaluated for liposomal preparations in the context of infectious diseases. For example, PEG-modified liposomes incorporating Influenza-derived Ags, together with the TLR-4 agonist CRX-601 as adjuvant, were effective at eliciting elevated levels of serum neutralizing Abs and mucosal IgA (103).

#### Liposome-Derivatives

Derivatives of liposomes have been explored to circumvent some of the drawbacks associated with liposomes and to improve their efficiency. For example, association of non-ionic surfactants with cholesterol or its derivatives to generate a structure called niosomes has allowed to increase the stability of the bilayer vesicles by preventing oxidation of the lipids (157). Addition of mannan at the surface of niosomes further increased the stability of the vesicles and helped to target specific receptors on APCs following oral administration. Vaccination with niosomes incorporating plasmid DNA coding for an Hepatitis B Ag induced SIgA production in the salivary and intestinal fluids, together with systemic Th1-type T cell responses (158). Moreover, incorporation of bile salts within niosome structures (bilosomes) has been shown to increase the stability of the vesicles and thus to improve oral delivery of peptides and proteins to the gut immune system (159, 160). Bile salts, such as deoxycholic acid or taurocholic acid, are amphiphatic molecules that can be easily incorporated within lipid bilayers and can promote the passage of lipophilic components across cell membranes (161). Thus, bilosomes have the ability to reinforce the biovailability of associated Ags mainly for oral vaccination (162). Different examples of bilosome application have been reported in association with Hepatitis B-derived Ags and Tetanus toxoid. In this context, induction of SIgA in mucosal secretions and IgA-positive plasma cells were observed (162–164) and showed elevated responses as compared to parenteral injection or use of niosomes without bile salts. An alternative approach is the inclusion of polar lipids with fully saturated isoprenoid chains extracted from Archaea to generate vesicles called archaeosomes or archaeal lipid mucosal vaccine adjuvant and delivery (AMVAD) (165). They have been shown to induce robust long-lasting protective Ab and T cell responses, including cytotoxic T lymphocytes responses after systemic injection (166). Advantages of such structures for mucosal vaccination comprise increased pH-dependent and thermal stability due to prevention of lipid oxidation and resistance to phospholipases and bile salts. In this context, mice immunized by the nasal route demonstrated sustained robust local and distant IgA responses in mucosal fluids, strong systemic IgG responses, and T cell responses (167). In addition, nasal vaccination with archaeosomes and cell-free extracts of *Francisella tularensis* led to reduced bacterial burden in the lung and spleen in a mouse model of tularemia (168). Oral immunization with archaeosomes is possible as well, although with higher amounts of Ags. Improved stability and retention time of such vesicles has been observed in the intestine, leading to potent IgG and IgA production (169). One non-negligible drawback of this approach is the access to achaeal polar lipids, as the purification from Archaea is a relatively demanding process. Nevertheless, production of synthetic polar lipid structures is under development.

### Virosomes

Virosomes are a special category of liposomes, where part of the lipid content is derived from viral components that self-assemble into an organized three-dimensional structure that mimics the antigenic structure of the original virus (170). Interestingly, they have been demonstrated to be immunogenic without further addition of adjuvants (171), although addition of immunopotentiating agents further improves their vaccine efficiency (172). Originally called immunopotentiating reconstituted Influenza virosomes, they harbor hemagglutinin and neuraminidase proteins from Influenza virus. These proteins target sialic acid on cell membranes, leading to fusion between the target cell and virosomes, followed by intracellular delivery of their payload. They exhibit similar flexibility and advantages as standard liposomes; however, the process to extract all the necessary components from Influenza virus is relatively complex. Virosomes have been mainly investigated for parenteral vaccination, but reports on their use for mucosal administration exist. They have been used as prime-boost vaccination strategy in a simian model of HIV infection, where intramuscular injections have been followed by nasal administration. It induced full protection against vaginal simian-HIV challenge that was correlated with the presence of mucosal IgA and IgG with blocking activity against virus transcytosis and neutralizing/Ab-dependent cellular cytotoxicity properties, respectively (173). In mice, nasal or sublingual administrations of adjuvanted virosomes were able to protect against Influenza and respiratory syncytial virus infections by promoting mucosal and systemic Ab responses, together with Th1-type cellular responses (174–176).

#### Gas-Filled Microbubbles

Gas-filled microbubbles are microsized spherical structures composed of a lipidic, denatured protein-based, or crosslinked polymer shell generally entrapping inert high molecular weight gases to ensure resistance to pressure once administered (177). Due to their strong echogenicity in presence of low ultrasound intensities, they are currently used for human application as intravenously delivered echo-contrast agents to more precisely visualize for example angiogenesis in malignant tumors, left ventricular opacification, and myocardial perfusion. In addition, cavitation induced by higher ultrasound application leads to the transient nonlethal permeability of the surrounding tissue (e.g., vascular barriers or cell membrane) allowing enhanced local on-demand extravasation and bioavailabilty of microbubbleassociated payload (178). In the last decades, such a process, known as sonoporation, has received important attention in order to improve delivery of a wide range of therapeutic molecules, including chemotherapeutic agents, siRNA, miRNA, oligonucleotides, or plasmid DNA, to tumor or immune cells (177, 179). Typically, sonoporation has been used for improved delivery of Ags into DCs with the aim of boosting immune responses (180). Interestingly, lipid-based microbubbles can be taken up by APCs and deliver intracellularly their antigenic payload without ultrasound application, leading to processing and presentation of the Ag to responsive T cells (181, 182). Furthermore, microbubbleassociated Ags can be injected parenterally as a vaccine formulation to elicit potent and long-lasting immune responses against systemic bacterial infection (183, 184). Lipid-based microbubbles are usually composed of phospholipids (e.g., DSPC) and palmitic acid, but tailored formulations can be prepared by incorporation of cationic lipids in their shell in order to better associate DNA (179). In addition, to improve the specificity of imaging and drug delivery, microbubbles can be targeted to particular tissues by linking cell-specific ligands or Abs has been developed at their surface (185, 186). Such aspects are of interest for mucosal vaccination using targeting strategies as discussed in Section "Targeted Delivery of Vaccine Ags." Moreover, adjuvants can be associated with microbubbles, which results in enhanced immunogenicity of the vaccine preparations. As an example, nasal delivery of α-galactosylceramide-adjuvanted microbubbles displaying the *Salmonella*-derived SseB Ag at their surface were able to induce potent IgA, IgM, and IgG humoral responses in the gut, which were associated with a Th1-/Th17-type cellular response. This resulted in a significant decrease in local and systemic bacterial load following oral infection with *Salmonella enterica* Typhimurium in prophylactically vaccinated mice; such effect was more potent than parenteral injection of the same microbubble formulation (140). Despite so far limited induction of local immune responses after oral administration, improvement of microbubble formulations may lead to enhanced immunogenicity. Moreover, sublingual administration remains to be tested owing to its valuable advantages in the context of mucosal vaccination. In recent years, nanosized bubbles have been developed that showed increased stability and extravasation following systemic administration, suggesting that such derivatives might be even more suitable for vaccination purposes (187).

#### Immune Stimulating Complexes

Immune stimulating complexes (ISCOMs) are negatively charged self-assembling pentagonal dodecahedrons cage-like rigid structures with a size of 30–40 nm. They can form spontaneously after mixing Ags with cholesterol, phospholipids (usually phosphatidylethanolamine and phosphatidylcholine), and the saponin Quil A extracted from the bark of *Quillaja saponaria* Molina tree. Interestingly, such formulation allows to reduce the toxicity associated with saponin administration, while retaining its adjuvant activity (188, 189). Proteins or glycoproteins that are normally anchored by a hydrophobic transmembrane sequence into the cell membrane can be incorporated as such. Non-amphipatic proteins or peptides have to be modified by attachment of a lipid tail (e.g., palmitic acid). Immunization with ISCOMs induced both Th1-type humoral and cellular responses, including cytotoxic T lymphocytes that are important to fight against intracellular pathogens (190). Several studies have reported potent induction of mucosal immune responses, including robust IgA production in nasal washes and the lung, after nasal/pulmonary vaccination with ISCOMs harboring antigenic entities from Influenza virus (191–194), respiratory syncitial virus (195), Hepatitis B virus (196) and measles (197). Protective efficacy was observed as well after vaccination with an Influenza subunit vaccine composed of ISCOMs (198), adjuvanted ISCOM-based anti-*M. tuberculosis* and anti-Influenza vaccines (199, 200) and *Helicobacter pylori*-Ags delivered *via* ISCOMs (201). In some cases, such immunization proved more efficient than parenteral injection (196, 202). Production of ISCOMs with an alternative saponin, derived from *Quilaja brasiliensis*, also allowed to induce mucosal local and distant IgA production after nasal delivery of an OVA-based vaccine (203). In addition, incorporation of DNA plasmid within the ISCOM matrix elicited potent anti-*Haemophilus influenzae* cellular and Ab responses in the nasopharynx of nasally immunized animals (204). Oral administration of ISCOM-based vaccines has been evaluated (205, 206); however, it seems that the generation of intestinal IgA responses was limited (207). Although ISCOMs are self-adjuvanted delivery systems, incorporation of the adjuvant CTA1-DD within the structure allowed to induce robust mucosal IgA production and T cell proliferation, together with systemic responses, after nasal administration (208). Such an approach has also been evaluated *via* oral delivery. Potent systemic Th1-type immune responses were induced, but unfortunately the mucosal compartments were not analyzed (209). CTA1-DD/ISCOMs incorporating major outer membrane protein from *Chlamydia muridarum* have also been administered *via* the vaginal route. Vaccination induced limited Ab responses, but clearly detectable CD4 T cell responses in vaginal tissues. This led to protection against a bacterial challenge, as demonstrated by reduction in bacterial shedding from the genital tract (49). Overall, the use of ISCOMs as a delivery vehicle for mucosal vaccination finds its best applicable for nasal administration, even though the sublingual route remains to be explored. Nevertheless, the difficulties related to the use of hydrophilic Ags that have to be modified before incorporation within ISCOMs, together with the reported toxicity of saponin, somehow limits the wide use of such vaccination approach.

#### Others

Oil-in-water nanoemulsions disperse into nanosized droplets and exhibit long-term colloidal stability. They can encapsulate hydrophilic or hydrophobic payload, respectively, and have been tested for nasal vaccination. A nanoemulsion based on soybean oil and cetylpiridinium chloride (W805EC) has been shown to deliver its antigenic payload across ciliated nasal epithelial cells and to the regional LNs in the NALT through migrating activated DCs (210). *Via* TLR-2 and TLR-4, such vaccine formulation promoted the induction of robust Ab and Th1-/Th17-type cellular responses and when associated with inactivated Influenza vaccine, generated a protective immunity against Influenza challenge (211). Such an approach similarly proved efficient in animal models to fight against *M. tuberculosis*, Hepatitis B, and *Bacillus anthracis* infections, or to generate HIV-1-specific mucosal immune responses (212, 213). Improved stability of nanoemulsions for mucosal delivery can now be achieved by the double emulsion water-in-oil-in-water technology, which has been applied both nasally and orally and resulted in robust production of systemic IgG and mucosal IgA (214). Alternatively, coupling of lipopeptides with a polylysine core induces the formation of 5–15 nm particles that can promote the generation of systemic and mucosal IgG/IgA and T cell responses after nasal administration. In these conditions, protective responses have been obtained against *S. pyogenes* infection (215, 216).

#### HUMAN APPLICATION OF LIPID PARTICLES FOR MUCOSAL VACCINATION

Currently approved mucosal vaccines are composed of liveattenuated or killed whole-pathogen cells that offer relatively good efficacy, but cannot be administered to young infants, immunocompromised people and the elderly due to potential safety issues. The majority of mucosally administered vaccines in humans are delivered *via* the oral route and directed against enteric infections such as polio, cholera, typhoid fever, and rotavirus infection (14). Oral polio vaccine has been used for more than 50 years with great success and is a prototypical vaccine for polio eradication in many countries. Interestingly, it demonstrated improved efficacy as compared to an inactivated pathogen vaccine injected parenterally (217–219). Vivotif®, as well as Dukoral®, Shanchol™, Orochol®/Vaxchora™, and mORC-Vax™, are live-attenuated or whole-killed vaccines against *Salmonella*induced typhoid fever or *Vibrio cholerae* infections, respectively (220–222). As expected based on preclinical studies in animal models, their protective efficacy has been correlated with effector immune responses present at mucosal surfaces (in most cases detection of SIgA in mucosal fluids) and induction of plasma cells expressing gut-homing molecules specific for the small intestine and the colon (223–225). In addition, and similar to observations in animal models, the choice of the administration route impacts on the tropism of the induced mucosal immune responses (226). Only FluMist®, a live-attenuated Influenza virus vaccine, is licensed for nasal administration (227). It demonstrates high level of protection against matched and mismatched viral strains in children and adults and proved more efficient than parenteral vaccination (228). Virus-specific mucosal IgA and systemic IgG responses with a possible role for cell-mediated immunity has been documented in vaccinated individuals (229). So far, the only adjuvant used for vaccinal application to mucosae is the B subunit of CT, which has been included in Dukoral® to improve the immunogenicity of the killed whole-pathogen *Vibrio cholerae* vaccine.

Despite many encouraging results obtained in proof-of-concept and preclinical animal models, a limited number of subunit vaccines based on lipidic delivery systems has been evaluated and/or approved for human use, especially for mucosal administration (**Table 2**) (230). Interestingly, in the context of mucosal vaccination, at least three formulations have been evaluated in early phase clinical trials. Nasal administration of the oil-in-water nanoemulsion W805EC combined with the approved inactivated systemic Influenza vaccine Fluzone® has demonstrated induction of IgA responses in nasal washes in a clinical evaluation (231). In addition, virosome-based and ISCOM-based Influenza vaccines are currently under development for administration *via* the nasal route (232, 233). Systemic injection of virosomes has been


TABLE 2 | Examples of licensed and in development lipid-based vaccines for human use.

*I.m., intramuscular; I.n., intranasal; I.v., intravenous.*

licensed for human vaccination against hepatitis A (Epaxal®) and Influenza (Inflexal V®) (234). Additional vaccines based on such technology have been tested in Phase 1 clinical trial for malaria and Influenza *via* the systemic route (232, 235). The advantage of such approach is that virosomes are self-adjuvanted, which is not the case for all other lipid-based delivery systems. Two Influenza vaccine formulations composed of squalene-based nanoemulsions (MF59® and AS03) are also approved for intramuscular injection in humans, with a particular focus on use in young children and elderly (236, 237). At least three additional strategies have been evaluated in humans. The most promising one is the use of the AS01 adjuvant, composed of liposomes made of highly unsaturated neutral phospholipids including MPLA and saponin QS21 (238). Successful phase III vaccination trials of such delivery system have been performed with the RTS,S/Mosquirix™ vaccine formulation against malaria (239) and the herpes-zooster vaccine (HZ/su) (240). CAF01 is an alternative formulation that is currently tested for parenteral vaccination against tuberculosis. Such liposomal bilayer preparation contains the cationic lipid DDA and the glycolipid TDB as immunostimulator (152). ISCOMs incorporating Ags E6 and E7 from HPV16 have been tested in women with cervical intraepithelial neoplasia and HIVpositive individuals with oncogenic HPV infections (241, 242). They demonstrated a safety profile and induced specific humoral and effector T cell responses. Similar results were obtained following vaccination with HCV-derived antigenic entities (243). Additional vaccine preparations based on lipidic constructs are currently evaluated to fight against pathogen infections, such as malaria, dengue fever, HIV, or Influenza (90, 230).

Nevertheless, the above-mentioned studies are mainly performed with systemically injected formulations and only rare mucosal applications have been assessed. The use of mucosal route of administration requires that anatomical, functional, and immunological characteristics are taken into consideration and differences between humans and animals may results in poor inter-species translation of promising results (13). An organized NALT similar to that present in mice is not found in humans. By contrast, alternative inductive sites in the form of immune nodules are present in humans in the upper nasal cavity, in the concha, and in Waldeyer's rings (adenoids, tonsils) (244). Pulmonary delivery of vaccines does not seem to be optimal as well, because BALT have only been reported in fetuses and young children, but not in healthy adults. In addition, the localization and phenotype of DCs in the nasal cavity and in the lung all differ between mice and humans. Taking into account these considerations, sublingual administration sounds like a promising strategy (42). Indeed, immunological and physiological organization of the oral cavity is similar in both mice and humans, with documented presence of the same DC subsets, such as Langerhans cells, capable of eliciting immunogenic or tolerogenic responses depending on the applied formulations. Vaccination *via* the sublingual route has been mostly evaluated in humans for allergen immunotherapy, and has been shown to induce systemic IgG (focus of allergy) (245). Therefore, evaluation of the mucosal vaccination approach to protect against infectious diseases is highly relevant and needs evaluation. In addition, although the gut immune system is relatively similar between both species, oral administration of subunit vaccines is further complicated due to the constraints related to the stability of formulations in the aggressive environment of the digestive tract. On the top of anatomical considerations, the age-related decline in the immune function, possibly related to the documented decrease in M cell differentiation with age, represents a drawback for immunization in eldery people.

#### CONCLUSION AND PERSPECTIVES

Although vaccination has led to the control of several diseases and has demonstrated substantial technological progresses, humans still suffer from infections leading to death and increasing health costs. Many infectious diseases for which the development of effective vaccines is urgently needed include those transmitted through various mucosal routes that affect the gastrointestinal tract (*E. coli, Salmonella, Shigella, Vibrio cholerae, H. pylori*), the respiratory tract (Influenza, *M. tuberculosis*, respiratory syncitial virus) or are sexually transmitted (HIV, *Chlamydia*) (15). To date, parenteral vaccination represents an important part of the administered vaccines, despite the fact that they poorly induce mucosal immune responses. Furthermore, the requirement for sterile needles, their subsequent elimination, the associated cost, and the cold chain's requirement in many instances prompt a shift toward reduced frequencies of intramuscular vaccination. In addition, most of the currently licensed vaccines can only be administered over 2 years of age for safety reasons; similar considerations apply for immunodeficient individuals and elderly (13). Therefore, the mucosal application of subunit vaccines represents a sound alternative to broaden the target population that could benefit from vaccination. Cues into this direction include recent advances in the understanding of mucosal immunity as well as assessment of correlates of protection may help to develop promising mucosal vaccines; in this respect, design of novel effective delivery strategies will permit to achieve mucosal vaccines that induce protective neutralizing SIgA, together with CTLs and effector CD4 T cells mainly secreting IFN-γ and IL-17. Moreover, because they are considered as safe, subunit vaccines can certainly be administered to neonates and young infants who already possess a functional mucosal immune system (13).

Lipid-based particles fulfill the requirements for better efficiency, safety, low-dose Ag, and ease to handle logistically. They can deliver a wide range of antigenic entities upon mucosal delivery and can be tailored to obtain vaccine formulations with appropriate properties to address questions related to the mechanisms invlolved in the control of the pathogen and the route of administration. Robust and sustained induction of immune responses, comprising production of SIgA at mucosal surfaces, together with helper and cytotoxic T cells, often correlate with protection in defined animal models under study; this will undoubtly help to drive vaccine development toward the

#### REFERENCES


right direction. It remains that a strict comparative analysis of the formulations and administration routes to be used against a particular infectious disease is rarely performed within the same study. To contribute to the identification of such a missing piece in this complex puzzle would be an asset in order to optimize mucosal vaccination. The same lack of information must be filled up when one deals with the definition of the optimal dosing and schedule of administration to ensure efficient priming and boosting of immune responses aiming at reaching optimal magnitude and maintenance.

Some lipid-based preparations have the advantage of being lyophilized, thus allowing to simplify the logistics usually necessary for cold chain. Indeed, some of these formulations can be stored at room temperature for several months and can be administered in such form *via* the nasal, oral, or sublingual routes. Dry powder nasal vaccines have already been tested (246), oral delivery of capsules is not a problem (222), and many allergy-related immunotherapy tablets have been considered for sublingual administration (247). Moreover, lipid-based preparations can be aerosolized, which might represent an alternative procedure to keeping stable vaccine preparations. Overall, great expectations are coming from the lipid-based vaccine formulations currently evaluated in clinical trials in humans, which together with the different mechanisms of mucosal immunity recently unraveled, may likely favor the development of future mucosal vaccines suitable for a majority of individuals, thanks to the combinatorial flexibility offered by the nature of the constituents available.

#### AUTHOR CONTRIBUTIONS

BC and GB planned the manuscript and wrote sections of the manuscript. Both authors read and approved the submitted version of the manuscript.

#### FUNDING

This work was supported by the Swiss Science Research Foundation (grant 3100-156806 and grant 310030\_175686 to BC).


secretions for delivery of antigens to mucosal dendritic cells. *Immunity* (2004) 20(6):769–83. doi:10.1016/j.immuni.2004.05.007


Th2 bias in the immune response and protection against infection. *Vaccine* (2009) 27(27):3643–9. doi:10.1016/j.vaccine.2009.03.040


lipid – ceramide carbamoyl-spermine (CCS) I. Immunogenicity and efficacy studies in mice. *Vaccine* (2006) 24(18):3990–4006. doi:10.1016/j.vaccine. 2005.12.017


antibodies protecting nonhuman primates against vaginal SHIV challenges. *Immunity* (2011) 34(2):269–80. doi:10.1016/j.immuni.2011.01.015


glycoproteins in micelles and immunostimulatory complexes (iscoms). *Scand J Immunol* (1988) 27(6):645–52. doi:10.1111/j.1365-3083.1988.tb02397.x


the common mucosal immune system in humans. *Infect Immun* (1998) 66(12):5630–5.


247. Passalacqua G, Canonica GW. Sublingual immunotherapy: focus on tablets. *Ann Allergy Asthma Immunol* (2015) 115(1):4–9. doi:10.1016/j.anai.2015. 03.022

**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 Corthésy and Bioley. 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.*

# Mucosal Delivery of Fusion Proteins with *Bacillus subtilis* spores enhances Protection against Tuberculosis by Bacillus calmette-guérin

*Alastair Copland1†, Gil R. Diogo1†, Peter Hart1†, Shane Harris1 , Andy C. Tran1 , Mathew J. Paul1 , Mahavir Singh2 , Simon M. Cutting3 and Rajko Reljic1 \**

*1St George's Medical School, London, United Kingdom, 2 Lionex GmbH, Braunschweig, Germany, 3School of Biological Sciences, Royal Holloway University of London, Egham, United Kingdom*

#### *Edited by:*

*Jeffrey K. Actor, University of Texas Health Science Center at Houston, United States*

#### *Reviewed by:*

*Camille Locht, INSERM, France Owen Kavanagh, York St John University, United Kingdom*

> *\*Correspondence: Rajko Reljic*

*rreljic@sgul.ac.uk*

*† These authors have contributed equally to this work.*

#### *Specialty section:*

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

*Received: 13 November 2017 Accepted: 07 February 2018 Published: 12 March 2018*

#### *Citation:*

*Copland A, Diogo GR, Hart P, Harris S, Tran AC, Paul MJ, Singh M, Cutting SM and Reljic R (2018) Mucosal Delivery of Fusion Proteins with Bacillus subtilis Spores Enhances Protection against Tuberculosis by Bacillus Calmette-Guérin. Front. Immunol. 9:346. doi: 10.3389/fimmu.2018.00346*

Tuberculosis (TB) is the most deadly infectious disease in existence, and the only available vaccine, *Bacillus Calmette-Guérin* (BCG), is almost a century old and poorly protective. The immunological complexity of TB, coupled with rising resistance to antimicrobial therapies, necessitates a pipeline of diverse novel vaccines. Here, we show that *Bacillus subtilis* spores can be coated with a fusion protein 1 ("FP1") consisting of *Mycobacterium tuberculosis* (Mtb) antigens Ag85B, ACR, and HBHA. The resultant vaccine, Spore-FP1, was tested in a murine low-dose Mtb aerosol challenge model. Mice were primed with subcutaneous BCG, followed by mucosal booster immunizations with Spore-FP1. We show that Spore-FP1 enhanced pulmonary control of Mtb, as evidenced by reduced bacterial burdens in the lungs. This was associated with elevated antigen-specific IgG and IgA titers in the serum and lung mucosal surface, respectively. Spore-FP1 immunization generated superior antigen-specific memory T-cell proliferation in both CD4+ and CD8+ compartments, alongside bolstered Th1-, Th17-, and Tregtype cytokine production, compared to BCG immunization alone. CD69+CD103+ tissue resident memory T-cells (Trm) were found within the lung parenchyma after mucosal immunization with Spore-FP1, confirming the advantages of mucosal delivery. Our data show that Spore-FP1 is a promising new TB vaccine that can successfully augment protection and immunogenicity in BCG-primed animals.

Keywords: tuberculosis, spores, vaccine, immunity, adjuvants

#### INTRODUCTION

In 2015, tuberculosis (TB) overtook HIV/AIDS as the leading cause of death due to infection (1). This statistic contrasts starkly with the fact that the only available TB vaccine, *Mycobacterium bovis* Bacillus Calmette-Guérin (BCG), is the most widely administered vaccine in history (2), having been developed almost a century ago. The proposed reasons for the failure of BCG to adequately protect against TB are many and varied. They include (i) BCG sub-strain heterogeneity (3), (ii) pre-exposure of the host to environmental non-tubercle mycobacteria (4), (iii) a failure to prevent pulmonary infection (5), and (iv) limited protection in adults compared to children (6). Despite these limitations, BCG is unlikely to be discontinued in clinical use. While its efficacy in many demographics is modest, there are accumulating data indicating that BCG may protect against non-TB diseases by training the innate immune system to respond non-specifically to diverse microbial threats (7, 8). A novel TB vaccine is therefore likely to supplement, rather than replace, BCG.

In 2011, the novel viral vector TB vaccine MVA85A, comprising Ag85A, was tested for safety and efficacy in a phase 2 clinical trial in South Africa, and it was found that parenteral administration of MVA85A in BCG-immunized infants offered no significant protection above that of BCG alone (9). The reasons for its failure are still unclear, since MVA85A protected against *Mycobacterium tuberculosis* (Mtb) in multiple animal models (10). But it is becoming increasingly apparent that the development pipeline for new TB vaccines will require technological diversity in order to maximize chances of success. In recent years, vaccines that are based upon particulate nano- or microscale delivery systems have made remarkable strides in both oncology and infectious diseases (11–13).

*Bacillus subtilis* is an environmental Gram-positive bacterium that is also found as a gut commensal in humans (14). Its spores have the desirable properties of being both safe and adjuvantic (15). But more importantly, they possess hydrophobic and electrostatic properties that allow them to readily bind protein antigens, making these spores pertinent to vaccine development as potential antigen delivery systems (16). The combination of intrinsic adjuvanticity and antigen-binding biophysical properties allows *B. subtilis* spores to act simultaneously as adjuvants and antigen carriers. Studies have shown that immunization of mice with *B. subtilis* spores coated with the influenza antigen M2e can induce strong antibody responses and protect against lethal challenge (17, 18). Similar findings have been observed in other immunization models, including immunogenicity against HIV and streptococci (19, 20). *B. subtilis* spores are thus an attractive platform for subunit vaccine enhancement.

We have previously shown that *B. subtilis* spores coated with TB antigens (21) or genetically engineered to express a TB antigen (22) can enhance protection against TB by BCG (prime-boost) in a mouse intranasal infection model. Although this provided a proof-of-principle framework for vaccine efficacy, the use of genetically modified components in a vaccine presents numerous regulatory barriers for clinical application (23). Here, we developed a novel TB vaccine—"Spore-FP1"—composed of *B. subtilis* spores non-covalently coated with a fusion protein (FP1) consisting of the antigens Ag85B, ACR and the epitheliumbinding domain of HBHA ("FP1"). Ag85B and ACR were chosen to represent early and late stages of *Mtb* infection, respectively, while HBHA (heparin-binding domain only) was used for epithelial targeting in the lungs. Mucosal booster immunization with Spore-FP1 in BCG-primed mice enhanced protection in a low-dose aerosol Mtb challenge model, compared to BCG alone. The enhanced protection was concomitant with a wide array of boosted immunological parameters, including enhanced antigen-dependent T-cell proliferation and antibody production. Spore-FP1 is therefore a novel TB vaccine that has the potential to supplement pre-existing immunity conferred by BCG in human populations.

#### MATERIALS AND METHODS

#### Ethics Statement

All animals were used with approval from St. George's University of London Ethics Committee under an approved Home Office animal project license (70/7490) and used in accordance with the Animals (Scientific Procedures) Act 1986.

#### Mice and Immunizations

Female C57BL/6 mice were obtained from Charles River, UK, and were between 8 and 12 weeks of age before experimental use. For all bacterial challenge or immunogenicity experiments (except lung T-cell analysis), mice were immunized with 5 × 105 CFU BCG Pasteur (100 µL) subcutaneously or vehicle control. Intranasal booster immunizations consisted of 1 × 109 *B. subtilis* spores coated with 10 µg FP1 in 40 µL volumes per animal per dose, or vehicle control. These immunizations were performed under light anesthesia. For experiments involving dead spores, spores were autoclaved before protein adsorption. For some experiments, the adjuvant poly(I:C) (Sigma-Aldrich) was used intranasally at a dose of 20 µg.

Mice were infected with approximately 200 *M. tuberculosis* bacilli per animal delivered *via* low-dose aerosol, using a Biaera aerosol generator (Biaera Technologies). Infectious dose was routinely verified by standard plating techniques.

### Bacteria and Colony Forming Unit Quantification

Bacillus Calmette-Guérin Pasteur and Mtb (strain H37Rv) were used for the *in vivo* experiments; these were kind gifts of Professor Juraj Ivanyi (King's College, London). Both strains were grown to log phase at 37°C in 7H10 broth (Becton Dickinson) supplemented with ADC (Becton Dickinson), 0.05% Tween-80 and Selectab (Mast Diagnostics). Bacteria were then enumerated by the standard CFU method on 7H11 agar plates [supplemented with OADC (Becton Dickinson), glycerol and Selectab (Mast Diagnostics)] and cryopreserved in liquid nitrogen until use.

Bacterial burden from mouse organs was assessed by CFU enumeration. Lung and spleen homogenates were prepared in a stomacher containing 0.1% Triton X-100. Homogenates were plated in technical duplicates (lungs) or singlets (spleens) on 7H11 agar supplemented with OADC, glycerol and Selectatab. CFUs were counted after a 3–4-week incubation at 37°C.

#### Vaccines and Immunization

Amino acid sequences of the Mtb proteins ACR, Ag85B (pos. 23–25), and HBHA (pos. 160–199) were connected via linker peptides (GGGSGGGS), and six histidine residues were added to the C-terminus resulting in FP1. The amino acid sequence of FP1 was retranslated to DNA considering the codon usage of *Escherichia coli* DH5α, the host strain for protein production. In order to enable site-directed cloning, restriction sites for *Nco*I and *Hind*III were added to the 5′ and 3′ end, respectively. The synthetic gene was provided by GenScript (USA) inserted in pUC57. The gene of FP1 was excised from this plasmid using the abovementioned restriction endonucleases and ligated to expression vector pLEXWO481, an IPTG-inducible derivative of pMV261 (24), digested with the same enzymes as before. For production of FP1, the gene was expressed under control of lac-promoter while growing the host strain in APS medium at 30°C. Recombinant protein was isolated from inclusion bodies after denaturation in 8 M urea using metal chelate chromatography (Ni-NTA Superflow, Qiagen). Highly enriched FP1 was refolded by gel-filtration using sephadex G-200 material (GE Healthcare). Purity was assessed by fully automated SDS-PAGE with fluorimetric detection and densitometric purity (>97% purity). Western blots specific for component antigens were used to confirm the identity of the protein band. Endotoxin content was measured by LAL assay and determined to be <7 IU/mg. For formulation, FP1 was incubated with spores for 1 h at room temperature prior to the addition of polyI:C (if used). Vaccines were delivered immediately after formulation.

#### Antibody and Antigen Quantification

Antibody levels (IgA and IgG) in BAL and serum were quantified by ELISA. Antigens [Ag855, ACR; Lionex GmbH (Braunschweig, Germany)] were coated onto a plate at 2 µg/mL overnight, followed by blocking for 2 h with PBS containing 1% bovine serum albumin (BSA). BAL and serum were diluted 1:250 and 1:1,000, respectively, in PBS with 1% BSA and incubated on the plate in triplicate for 1 h at 37°C. Levels of IgA or IgG were detected using peroxidase-conjugated anti-mouse IgA or anti-mouse IgG (Sigma) and OPD substrate (Sigma). Plates were read on a Tecan200 plate-reader at 450 nm absorbance.

Assessment of protein loading onto spores and stability of final product was done by ELISA (quantifying unbound protein after adsorption) and by measuring charge and size using a ZetaSizer NanoZS (Malvern) according to manufacturer's instructions and proprietary software. Significance was tested with a paired *t*-test. For ELISA measurements, FP1 was coated onto plates at varying concentrations as described for the antibody measurements, followed by detection by a peroxidase-conjugated anti-His antibody (Sigma) in conjunction with OPD substrate. Plates were read as described above.

#### General Flow Cytometry

For most experiments, cells were first stained with Fixable Viability Dye eFluor® 780 (1:1,000 dilution; eBioscience) in the presence of Fc receptor blockade (TruStain, 1:500 dilution; Biolegend). For surface staining, cells were then stained in flow cytometry buffer (PBS containing 0.5% BSA and 0.1% sodium azide—all from Sigma-Aldrich) for 30–45 min at 4°C. For some experiments, cells were subsequently fixed in the appropriate fixative for 30 min at 4°C, and then stained in a permeabilization buffer for 45 min, followed by acquisition on a BD FACSCanto II, unless otherwise specified. For compensation matrices, UltraComp beads were used according to the manufacturer's instructions (eBioscience). Staining boundaries were determined by a combination of antibody titration, biological controls and fluorescence-minus-one samples.

#### Antigen-Presenting Cell (APC) Activation

Dendritic cells (DCs) were obtained according to a well-established protocol (25). Briefly, mouse femurs were aseptically flushed with complete RPMI (RPMI-1640 containing 100 U/mL penicillin/ streptomycin, 2 mM l-glutamine, 10% fetal calf serum, and 50 µM 2-mercaptoethanol—all from Sigma-Aldrich) and the bone marrow cells were cultured in complete RPMI with 50 ng/mL GM-CSF (Peprotech) for 2 days, followed by complete removal of the liquid media containing non-adherent granulocytes, and replacement with fresh GM-CSF-supplemented media. Cells were then cultured for a further 3–4 days, and non-adherent and loosely adherent cells were gently detached. DCs were phenotyped by flow cytometry and were found to be>85% CD11c<sup>+</sup> and expressing high levels of MHC Class II. DCs were cryopreserved in 10% DMSO until use. For experiments involving macrophages, the J774 cell line was used. Macrophages were cultured in complete DMEM (from Sigma, see RPMI), and sub-cultured every 3 days at ~80% confluency. Cells were>99% CD11b<sup>+</sup> as assessed by flow cytometry.

To measure activation, APCs were stimulated for 48 h with *B. subtilis* spores at an MOI of 1, 10, or 100, or *E. coli* LPS (100 ng/mL; Sigma-Aldrich), and stained with a panel of antibodies: CCR7- PerCP/Cy5.5, CD80-APC, CD86-PE/Cy7, MHC Class I-FITC, MHC Class II-Brilliant Violet 510, PD-L1-Brilliant Violet 421, and PD-L2-PE—all from Biolegend. Supernatants were tested for IL-1β, IL-6, and TNF-α. IL-12p40 was detected by intracellular cytokine staining after 20 h stimulation of macrophages in the presence of 10 µg/mL brefeldin A (Sigma-Aldrich) using IL-12p40-PE (Biolegend) in flow cytometry buffer containing 0.5% saponin (Sigma-Aldrich). To detect transcription factor phosphorylation, macrophages were stimulated for 4 h and then fixed in 90% methanol as previously described (26), followed by staining with antibodies against phosphorylated forms of c-Jun (AP-1), NF-κB (p60), and IRF-3 (Cell Signaling).

#### T-Cell Proliferation

Splenocytes were obtained from mouse spleens that had been mechanically homogenized and treated with ACK lysis buffer (Sigma-Aldrich) to remove erythrocyte contamination. Splenocytes were cultured at 1 × 106 /well in a 96-well plate. Cells were stimulated with 5 µg/mL recall antigen or 1 µg/mL α-CD3 (Biolegend) for 5–6 days, followed by surface staining with CD4-PerCP/Cy5.5, CD8-Brilliant Violet 510, CD44-FITC, CD62L-PE, and CD90.2-Brilliant Violet 421—all from Biolegend. Cells were then fixed and permeabilized using the eBioscience Foxp3/Transcription Factor Staining Buffer Set and stained with Ki67-APC.

#### Cytokine Quantification

Cytokines were measured by ELISA or Multiplex immunoassay. IL-1β, IL-6, and TNF-α in culture supernatant were measured by ELISA using eBioscience Ready-Set-Go kits according to the manufacturer's instructions. For IL-4, IL-10, IL-17A, and IFN-γ, a 4-plex multiplex immunoassay (Biolegend) was used according to manufacturer's instructions. Data was acquired on a BD FACSCanto, and analysis performed using Legendplex software (Biolegend).

#### Lung Cell Isolation and Analysis

For these experiments, mice were immunized with PBS or BCG (subcutaneously) or mucosally with the indicated vaccine component. Lungs were perfused of blood by flushing PBS through the right ventricle. Tissue was then dissected into 1 mm pieces using a scalpel, followed by digestion in 1 mg/mL collagenase and 0.5 mg/mL DNase I (Roche). Cells were then passed through a 70 µm strainer (Becton Dickinson), contaminating erythrocytes were lysed, and mononuclear cells were stained for CD3-APC, CD4-PerCP/Cy5.5, CD8-Brilliant Violet 510, CD44- FITC, CD62L-PE, CD69-PE/Cy7, and CD103-Brilliant Violet 421—all from Biolegend.

#### Statistical Analysis

Statistical tests are described in the relevant figure legends. All analysis was performed using FlowJo v10, Microsoft Excel 2010 and GraphPad Prism 7.

#### RESULTS

#### Spores Can Effectively Bind to Mtb Fusion Proteins

We tested whether a biological carrier system, such as spores, could bind and facilitate carriage of the FP-1 fusion-protein. Spores were incubated with FP-1 prior to centrifugation to quantity the amount of free FP1 present in the supernatant after adsorption. Of 100 µg FP1, less than 2 µg free FP1 was detected in supernatants after adsorption indicating a high binding efficiency of over 98% (*n* = 3). To characterize the formulated Spore-FP1 vaccine, we assessed the size and charge of Spores either pre- or postadsorption using a ZetaSizer NanoZS. Size of the spores increased following FP-1 adsorption from 1,337 ± 13.8 to 1,389 ± 13.53 nm for naked and loaded spores, respectively (*n* = 3). Spores became marginally more negative when FP-1 was loaded, with charge decreasing from −47.13 ± 0.95 to −49.1 ± 0.44 mv (*n* = 3). While neither of the changes were significant, both trended toward significance (*p* = 0.064 and 0.068 for charge and size, respectively) and, in conjunction with the ELISA data, are suggestive of protein loading onto the spore surface. Importantly, FP1 loaded spores were moderately negatively charged and had a low polydispersity index (PDI = 0.237), indicating strong colloidal stability.

#### Spore-FP1 Can Enhance Bacterial Control Afforded by BCG

Mucosal vaccination against respiratory diseases offers distinct advantages over parenteral delivery routes, such as enhanced control of the pathogen, presumably due to localized immune effector cells (27). We therefore tested the ability of mucosallydelivered Spore-FP1 to enhance control of Mtb, compared to BCG immunization alone, in a "prime-boost" strategy. We hypothesized that a booster immunization with Spore-FP1 would lead to better protection than immunization with BCG alone. Mice were first primed with BCG or vehicle control, followed by two intranasal boosts with Spore-FP1. Mice were then challenged with ~100 CFU aerosolized Mtb, and bacterial burdens were quantified in the lungs and spleen. As shown in **Figure 1A**, mice immunized with BCG alone were better able to control Mtb compared to mock-immunized mice, as evidenced by reduced CFUs in the lungs (PBS CFUs: 5.54 ± 0.06; BCG CFUs: 4.84 ± 0.18, *p* < 0.05). When mice received the Spore-FP1 booster immunization, however, there was a significant improvement compared to BCG alone, with a near −1 log reduction in bacterial burden (Spore-FP1 CFUs: 4.10 ± 0.20, *p* < 0.0001 vs. PBS, *p* < 0.05 vs. BCG). In the spleens, BCG and Spore-FP1 offered comparable levels of protection, with both groups exhibiting significant protection compared to mock immunization (*p* < 0.0001). Next, Spore-FP1 was tested with codelivery of the adjuvant poly(I:C), a known inducer of the Th1 subset when used in the respiratory tract (28). As can be seen in the lungs (**Figure 1B**), Spore-FP1 was again able to induce significantly better protection than BCG (*p* < 0.01), with a trend—though not a statistically significant difference—for increased protection in the spleen (*p* = 0.06, BCG vs. Spore-FP1). These data collectively demonstrate that Spore-FP1 immunization could improve protection offered by BCG alone in multiple contexts, and therefore

we sought to uncover any immunological phenomena that could be associated with efficacy.

## Spore-FP1 Enhances Antigen-Specific Antibody Production

Evidence suggests that antibodies may play a role in protecting against TB, either directly (Fab-mediated) or indirectly (Fc-mediated): monoclonal IgA therapy can reduce pulmonary Mtb burden (29), and adoptive transfer of antibodies from hosts with latent TB can improve macrophage functionality (30). We therefore probed whether Spore-FP was generating antibodies against Ag85B and ACR. Spore-FP1 immunization significantly enhanced titers of Ag85B-specific IgA in the BAL compared to PBS (*p* < 0.05), whereas there was no difference with BCG (**Figure 2A**). There was also a trend for increased levels of Ag85Bspecific IgG induced by Spore-FP1. With regards to ACR-specific antibodies (**Figure 2B**), Spore-FP1 was able to significantly enhance levels of α-ACR IgG in the serum, compared to PBS (*p* < 0.01). However, there were no changes in the α-ACR IgA within the BAL.

### Spore-FP1 Generates Abundant Tcm and Tem Cells with High Proliferative Capacity

The observation that Spore-FP1 immunization led to higher Mtb-specific IgG and IgA titers suggested that T-cell immunity

Figure 2 | Enhanced humoral immunity caused by Spore-FP1. Immunized mice were tested for the presence of antigen-specific IgG in the serum (1:1,000 dilution) and IgA in the BAL (1 mL PBS flush; 1:10 dilution) by ELISA, with optical density read at 450 nm in duplicate. (A) Levels of IgG and IgA specific to Ag85B. (B) Levels of IgG and IgA specific to ACR. Results are expressed as mean ± SEM. Data shown are derived from *n* = 3 individual mice and are representative of two independent experiments. Significance was tested against the unstimulated control by one-way ANOVA with Tukey's posttest, \**p* < 0.05 and \*\**p* < 0.01.

was also being modulated. We hypothesized that Spore-FP1 was inducing stronger T-cell immunity than BCG alone, leading to enhanced antibody levels. To test this, splenocytes from immunized mice were assessed for the expression of the cell cycle and proliferation marker Ki67 after exposure to the recall antigens Ag85B, ACR and FP1. The Ki67<sup>+</sup> cells were then divided into naive (CD44loCD62Lhi), T central memory (Tcm; CD44hiCD62Lhi) or T effector memory (Tem; CD44hiCD62Llo) phenotypes. As shown in **Figure 3**, as expected, there was minimal proliferation in the PBS group in response to all antigens, with a background level of ~3% Ki67<sup>+</sup> in memory cell subsets. There were modestly more proliferating cells in the BCG group, which is consistent with other studies showing that BCG induces a very small percentage of antigen-specific splenic T-cells (31, 32). For instance, there were 6.48% Ki67<sup>+</sup> CD4<sup>+</sup> Tem cells after ACR stimulation in this group, and a similar level in the CD8<sup>+</sup> Tem cells. However, in the Spore-FP1 group, there was a sharp overall increase in the percentage of Ki67<sup>+</sup> cells, with notable spikes (>20%) in proliferating CD8<sup>+</sup> Tcm and Tem cells in response to Ag85B. Similarly, Spore-FP1 had the highest percentage of CD4<sup>+</sup> Tem cells responding to Ag85B (>10%). Results for ACR in this group were more modest, but the trend remained consistent. These data support the ability of a mucosal vaccine to induce substantial T-cell responses at primary lymphoid sites.

## Spore-FP1 Immunization Results in a Mixed T-Cell Cytokine Profile

The increase in T-cell proliferation in response to mucosal immunization with Spore-FP1 led us to question which subsets of T helper cells and cytotoxic T-cells were responding to antigen. Therefore, splenocytes from immunized animals were cultured with recall antigens (Ag85B, ACR and FP1) and assessed for the production of IFN-γ, IL-4, IL-10, and IL-17A, which are secreted from Th1/Tc1, Th2/Tc2, Treg, and Th17/Tc17 subsets, respectively. We found (**Figure 4**) that there was muted cytokine production across all analytes in the BCG group when cells were stimulated with recall antigens, with the exception of minor IFN-γ secretion (570.36 pg/mL) after ACR pulsing. In the Spore-FP1 group, however, there was profound cytokine release in response to all three antigens. After Ag85B pulsing, Spore-FP1 splenocytes produced copious amounts of IFN-γ (3519.6 pg/mL), IL-10 (86.26 pg/mL), and IL-17A (1837.5 pg/mL), suggesting that Spore-FP1 immunization generated mixed T-cell subsets that were specific for Mtb antigens. Similar results were observed for FP1 antigen recall, and there were modest levels of cytokines for ACR. No IL-4 was detected in any of the groups.

## Evidence of Tissue-Resident Memory T-Cells after Mucosal Immunization

Since Spore-FP1 was causing the development of antigen-specific T-cells in central lymphoid organs, we next interrogated the lungs for the presence of tissue-resident memory T-cells. Lungs were perfused and harvested from immunized animals, and then CD44hi (i.e., memory) T-cells were assessed for the expression of tissue retention markers CD69 and CD103. As shown in **Figure 5**, PBS and BCG immunization induced minimal levels

of these cells (<4% in both CD4<sup>+</sup> and CD8<sup>+</sup> T-cells), with only a minor increase induced by FP1 alone. Notably, the mucosal delivery of *B. subtilis* spores alone did not lead to the generation of Trm, while the full vaccine construct, Spore-FP1, was able to induce 14.9% CD4<sup>+</sup> and 12.5% CD8<sup>+</sup> Trm, respectively.

#### Bacillus Spores Activate Macrophages and DCs

Antigen-presenting cells are essential for the generation of T-cell immunity after immunization (33, 34). Empirical and systems biology approaches have revealed a correlation between APC activation by some antibody-inducing vaccines and protective immunity (35–37). Consequently, we next tested whether *B. subtilis* spores could activate DCs and macrophages (**Figure 6**). DCs and macrophages were pulsed with *B. subtilis* spores for 2 days at a range of MOIs and assessed for the upregulation of maturation markers. In DCs, spores significantly upregulated CD80, MHC Class I and CCR7 (CD80: *p* < 0.05, MHC Class I: *p* < 0.001, CCR7: *p* < 0.01), with strong trends for upregulation of CD86 and PD-L1 (**Figure 6A**). Interestingly, spores induced

the downregulation of MHC Class II and PD-L2. This may reflect the time-point at which the markers were measured, since Class II is known to be upregulated initially, followed by late-phase downregulation in activated DCs (38). In macrophages, spores were able to induce the upregulation of PD-L1 (*p* < 0.05), PD-L2 (*p* < 0.001), and CCR7 (*p* < 0.001), with similar trends for CD80, CD86, and MHC Class II. With regards to the production of proinflammatory cytokines, there was evidence of secretion of IL-6 and TNF-α at an MOI of 100 in DCs and macrophages (**Figure 6B**), albeit at lower levels compared to the positive control LPS. There was modest IL-1β production at the highest dose of spores. Next, we used intracellular cytokine staining to detect IL-12p40 production after spore exposure (**Figure 6C**). Spores were found to induce comparable levels of IL-12p40<sup>+</sup> APCs (23.3%) compared to LPS (18.3%). As expected, there was minimal production of IL-12p40 in unstimulated cells. Finally, we investigated three transcription factors downstream of Tolllike receptors (TLRs) to understand the cause of the phenotype. Intracellular flow cytometry at 4 h poststimulation (**Figure 6D**) revealed that *B. subtilis* spores were activating AP-1 (c-JUN), NF-κB and IRF-3 to a greater extent than LPS, suggesting engagement of both MyD88 and TRIF adaptors, in tandem with mitogen-activated protein kinase activity.

#### DISCUSSION

Tuberculosis is a disease defined by an immunological complexity that has hindered efforts toward vaccine development and allowed Mtb to persist perniciously across the globe. The cause of this complexity is a raft of exquisite immune evasion mechanisms that has evolved over thousands of years to manipulate host immunity. Therefore, at present, animal infection models are the most useful tool for predicting TB vaccine success (10).

We have previously reported that spores coated with Mtb antigens can provide good protection against Mtb in an intranasal infection model (21). A major limitation of this model, however, is that intranasal infection is not physiologically representative of natural Mtb infection. Furthermore, this study did not address whether such vaccine constructs could enhance pre-existing BCG-mediated immunity in the mouse model in a prime-boost immunization schedule. In the present study, we used a technologically superior low-dose aerosol Mtb challenge model to test the efficacy of a novel TB vaccine, Spore-FP1, on BCG-immunized mice, with the hypothesis that Spore-FP1 could boost the protection afforded by BCG. Spore-FP1 was designed to include antigens expressed during the early (Ag85B) and late (ACR) phases of Mtb infection, which could help to account for some of the observed dynamic changes in antigen expression during disease (39). Moreover, unlike previous studies, Spore-FP1 also contained a portion of the epithelium-binding domain of HBHA, which would encourage attachment to lung epithelium and dissemination of the vaccine (40).

Our most important finding was that intranasal immunization with Spore-FP1 in a BCG-primed mouse (i.e., the "prime-boost" strategy) could significantly enhance protection above that given by BCG alone. This effect was independent of any coadjuvants or whether the spores were viable or not. Bacterial burdens in the lungs were significantly reduced by the Spore-FP1 booster immunization compared to BCG alone. Interestingly, there was no statistically significant reduction in extra-pulmonary Mtb. This may reflect a limitation of detection in our assays, or at least an area of potential development for future utilization of spore vehicle systems. Distinct T-cell-derived cytokine patterns are known to exert dichotomous effects on bacterial burdens in the lungs and spleen (41), which may be the case here. Furthermore there was also a slight reduction in protective effect when Spore-FP1 was used with adjuvant. This may reflect ordinary biological variation between experiments, or perhaps non-optimal adjuvant choice, and will be the subject of future investigations.

To investigate immunological events associated with protection, we first evaluated antigen-dependent antibody production in serum and BAL. The role of antibodies in TB is contentious, although there have been recent reappraisals of the field generally in favor of their protective role (42). Mice immunized with Spore-FP1 were found to produce more Ag85B-specific IgG in the serum and IgA in the BAL than the BCG group; a similar

trend was observed for ACR. These data strongly suggest that a Spore-FP1 boost immunization was better at inducing humoral immunity than a single BCG immunization.

T-cells are essential for protection against Mtb: Th1 cells prime macrophages for activation via IFN-γ (43), and Th17 cells can upregulate the production of antimicrobial peptides and lymphocyte chemoattractants (44, 45). Deficiency in either of these cytokines is extremely detrimental to the host during disease. It has been shown that during natural infection, Mtb can subvert the host immune system in order to restrict antigen presentation (46, 47). Hence a vaccine that enhances antigen presentation, and thus leads to a higher frequency of antigen-specific T-cells, is highly desirable. In our experiments, we observed a higher frequency of proliferating splenic T-cells in response to recall antigens in the Spore-FP1 group compared to mice that had only received BCG immunization. BCG is also able to restrict antigen presentation *in vivo* to a certain extent (47–49), and consistent with this fact, we observed minimal proliferative responses to Ag85B/ACR in BCG-immunized animals. Such small-magnitude responses in BCG-immunized mice are highly typical and described elsewhere in the literature, wherein cells specific for Ag85 typically represent ~0.1% of the total splenic polyclonal T-cell pool in cytokine capture assays (31, 32). In contrast to these constrained responses, Spore-FP1 was able to induce a dramatically larger percentage of proliferating T-cells, indicating either a higher frequency of memory cells, or at the very least cells with a higher proliferative capacity. Many of the proliferating CD8<sup>+</sup> Ki67<sup>+</sup> cells were of the Tcm phenotype, which act as a "reservoir" of cells in primary and secondary lymphoid organs with high potential for differentiation into effector cells in distal sites. For chronic diseases such as TB that include T-cell exhaustion as a definitive mechanism of immune evasion (i.e., terminal differentiation), the generation of proliferative Tcm by a prophylactic vaccine offers a distinct advantage.

In line with proliferative responses, Spore-FP1 was also a potent inducer of IFN-γ, IL-10, and IL-17A release after splenocyte exposure to recall antigens. Thus, the antigen-specific cells were fully functional by producing effector cytokines during proliferation. It could be surmised that Spore-FP1 therefore induced a mixed Th1-Th17-Treg response. The absence of IL-4 release is interesting, and suggests that Spore-FP1 induced a T-cell skewing away from the Th2 to a Th1/Th17 phenotype. IL-4 is largely believed to be detrimental during Mtb infection, since it antagonizes the biological effects of IFN-γ to promote alternatively activated macrophages (50). The role of IL-10 in TB is more contentious. While IL-10 can hamper antimycobacterial immunity during BCG immunization (51), recent evidence from *Rhesus macaque* infection models has suggested that CD4<sup>+</sup> T-cells coexpressing a balance of pro- and anti-inflammatory cytokines are significantly associated with granuloma sterilization, possibly due to a reduction in "collateral damage" to the lung tissue (52). Furthermore, IL-10 is important for shielding CD8<sup>+</sup> memory T-cells from apoptosis in inflammatory contexts (53), and IL-10 deficient mice are highly susceptible to reinfection by intracellular pathogens (54). We believe that the T-cell profile induced by Spore-FP1 is therefore beneficial in the context of immunization.

It is worth noting that for both humoral and cellular immunogenicity, there was generally a greater response to Ag85B than to ACR. This is perhaps due to the fact that Ag85B is a strong immunodominant antigen (55) that has formed the basis of many new TB vaccines. Notably, however, ACR was still able to elicit potent IFN-γ production in splenocytes from Spore-FP1 immunized mice.

Alongside conventional T-cell activation signatures, we also observed a striking accumulation of gross CD69<sup>+</sup>CD103<sup>+</sup> Trm in lung tissue after immunization with Spore-FP1. These cells are likely to be directed toward epitopes found within FP1, since the vehicle control (spores alone) failed to induce any appreciable quantities of these cells. As to why no Trm were directed against *B. subtilis* spores themselves, it may be that *B. subtilis,* as a mammalian commensal (56) (in the absence of a "foreign" antigen such as those included in FP1), can suppress the mobilization of effector T-cells that would lead to its own clearance. In support of this hypothesis, *B. subtilis* secretory products can induce a Foxp3-dependent tolerogenic environment in the gut (57), and consistent with this fact, we observed modest IL-10 responses from splenocytes exposed to recall antigen (although much lower than IFN-γ and IL-17A). A recent study has elegantly demonstrated that mucosal immunization with BCG—as opposed to parenteral immunization—leads to the accumulation of Trm in the pulmonary tissue (27). These cells are sufficient for protection, since adoptive transfer of Trm into BCG-naive mice protects against Mtb challenge. We speculate that the enrichment of this cell type in the lungs, induced by Spore-FP1 in our experiments, is playing a major role in the protection afforded by our novel vaccine.

Turning our attention to the innate immune system, we detected potent activation signatures in macrophages and DCs pulsed with *B. subtilis* spores. While it is known that *B. subtilis* spores can activate TLR-2→MyD88 downstream pathways, these studies have largely restricted maturation marker analysis to CD40 and MHC Class I and II expression on DCs (19, 58). Here, we showed for the first time that spores can also simultaneously induce CCR7, PD-L1 and PD-L2 upregulation. Since minimal T-cell priming occurs in the lung (59, 60), CCR7 expression will be critical for DCs that have taken up Spore-FP1 to migrate to the lung-draining lymph nodes and present antigen to naive T-cells. The upregulation of PD-L1 and PD-L2, on the other hand, may mitigate the overall inflammatory response, which is an important boon for mucosal delivery. In justification of this notion, PD-L1 blockade during antigen delivery into the lungs leads to exacerbated irritation and inflammation via Treg depletion, which is ameliorated upon immune reconstitution (61). Underscoring all of these phenotypic characteristics was the observation that IRF-3 was phosphorylated alongside NF-κB upon APC stimulation with spores. These data allude to a novel activation pathway besides the TLR-2→MyD88 axis, which is driving APC activation by *B. subtilis* spores, and has hitherto remained unexplored. This proposition warrants further biochemical investigation.

To conclude, we have shown that Spore-FP1 can enhance protection offered by BCG and also activate multiple arms of the innate and adaptive immune systems. These data demonstrate the potential applicability of Spore-FP1 as a TB vaccine, but also offer fresh insights into the mechanisms of *B. subtilis* spores as a vaccine development platform.

#### ETHICS STATEMENT

The animal work was reviewed and approved by St George's University of London Ethics Committee for animal experimentation and studies performed under a valid UK Home Office Project Licence.

## AUTHOR CONTRIBUTIONS

AC, PH, and GD performed most of the immunization and MTB challenge experiments. SH and ACT performed *in vitro* immunogenicity experiments. MS provided recombinant proteins. SC provided spores. MP performed immunological evaluations. RR conceived the study and wrote up the manuscript with AC.

## FUNDING

This study was funded by the European Commission H2020 grant no. 643558 awarded to the EMI-TB Consortium.

#### REFERENCES


mechanisms of action. *PLoS One* (2014) 9:e87454. doi:10.1371/journal. pone.0087454


of invariant chain degradation. *Immunity* (2001) 14:739–49. doi:10.1016/ S1074-7613(01)00148-0


responses and increases protection to *Mycobacterium tuberculosis* infection. *J Immunol* (2012) 189:4079–87. doi:10.4049/jimmunol.1201061


**Conflict of Interest Statement:** Author MS was employed by company Lionex. All other authors declare no competing interests.

*Copyright © 2018 Copland, Diogo, Hart, Harris, Tran, Paul, Singh, Cutting and Reljic. 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.*

# Trimethyl chitosan nanoparticles encapsulated Protective antigen Protects the Mice against anthrax

*Anshu Malik, Manish Gupta, Rajesh Mani, Himanshu Gogoi and Rakesh Bhatnagar\**

*Molecular Biology and Genetic Engineering Laboratory, School of Biotechnology, Jawaharlal Nehru University, New Delhi, India*

Anthrax is an era old deadly disease against which there are only two currently available licensed vaccines named anthrax vaccine adsorbed and precipitated (AVP). Though they can provide a protective immunity, their multiple side-effects owing to their illdefined composition and presence of toxic proteins (LF and EF) of *Bacillus anthracis,* the causative organism of anthrax, in the vaccine formulation makes their widespread use objectionable. Hence, an anthrax vaccine that contains well-defined and controlled components would be highly desirable. In this context, we have evaluated the potential of various vaccine formulations comprising of protective antigen (PA) encapsulated trimethyl-chitosan nanoparticles (TMC-PA) in conjunction with either CpG-C ODN 2395 (CpG) or Poly I:C. Each formulation was administered *via* three different routes, viz., subcutaneous (SC), intramuscular (IM), and intraperitoneal in female BALB/c mice. Irrespective of the route of immunization, CpG or Poly I:C adjuvanted TMC-PA nanoparticles induced a significantly higher humoral response (total serum IgG and its isotypes viz., IgG1, IgG2a, and IgG2b), compared to their CpG or Poly I:C PA counterparts. This clearly demonstrates the synergistic behavior of CpG and Poly I:C with TMC nanoparticles. The adjuvant potential of TMC nanoparticles could be observed in all the three routes as the TMC-PA nanoparticles by themselves induced IgG titers (1–1.5 × 105 ) significantly higher than both CpG PA and Poly I:C PA groups (2–8 × 104 ). The effect of formulations on T-helper (Th) cell development was assessed by quantifying the Th1-dependant (TNF-α, IFN-γ, and IL-2), Th2-dependant (IL-4, IL-6, and IL-10), and Th17-type (IL-17A) cytokines. Adjuvanation with CpG and Poly I:C, the TMC-PA nanoparticles triggered a Th1 skewed immune response, as suggested by an increase in the levels of total IgG2a along with IFN-γ cytokine production. Interestingly, the TMC-PA group showed a Th2-biased immune response. Upon challenge with the *B. anthracis* Ames strain, CpG and Poly I:C adjuvanted TMC-PA nanoparticles immunized *via* the SC and IM routes showed the highest protective efficacy of ~83%. Altogether, the results suggest that CpG or Poly I:C adjuvanted, PA-loaded TMC nanoparticles could be used as an effective, non-toxic, second generation subunit-vaccine candidate against anthrax.

Keywords: anthrax, vaccine, protective antigen, trimethylchitosan nanoparticles, CpG, PolyI:C

## INTRODUCTION

*Bacillus anthracis*, a Gram-positive microorganism, is the cause of an acute disease Anthrax. It can form spores, which could remain dormant for years and accidental exposure by inhalation may result in death (1). Although inhalational anthrax has been reported to have a high fatality rate near to 100%, few deaths have also been documented due to cutaneous and gastrointestinal anthrax (2).

#### *Edited by:*

*África González-Fernández, Centro de Investigaciones Biomédicas (CINBIO), Spain*

#### *Reviewed by:*

*Noemi Csaba, Universidade de Santiago de Compostela, Spain Arun Kumar, Linköping University, Sweden*

*\*Correspondence: Rakesh Bhatnagar rakeshbhatnagar@jnu.ac.in*

#### *Specialty section:*

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

*Received: 18 December 2017 Accepted: 06 March 2018 Published: 20 March 2018*

#### *Citation:*

*Malik A, Gupta M, Mani R, Gogoi H and Bhatnagar R (2018) Trimethyl Chitosan Nanoparticles Encapsulated Protective Antigen Protects the Mice Against Anthrax. Front. Immunol. 9:562. doi: 10.3389/fimmu.2018.00562*

**130**

It is also recognized for many bioterror attacks in history, including the famous 2001 bioterror attack of anthrax spores in the US (3). A prophylactic approach would be the best way to cease the disease and protect people against any bioterror attack. The currently existing vaccines BioThrax (also known as anthrax vaccine adsorbed) and anthrax vaccine precipitated (AVP) are the only licensed vaccines present globally (4). The major caveats of these vaccines are the presence of toxic proteins [lethal factor (LF) and edema factor (EF)] and the undefined composition of anthrax proteins (5). At present, only the high-risk population such as defense personnel are vaccinated with BioThrax. However, there have been several reports of reactogenicity in such individuals (6). Another drawback associated with this vaccine is the cumbersome dosing schedule, which comprises of multiple initial doses and subsequent annual boosters for sustained immunity (7).

In recent times, the focus has shifted to second generation subunit vaccines where the composition is pharmaceutically defined, and highly immunogenic proteins are taken into account making it safer for the mass use. In case of anthrax, *B. anthracis,* the causative organism of anthrax, secrete two toxins comprised of three proteins: the protective antigen (PA), the LF, and the EF. The edema toxin (PA + EF) causes edema and lethal toxin (PA + LF) causes death in animals (8). PA has always been the choice of antigen as it binds to the host receptor and binds to LF and EF to make lethal toxin and edema toxin, which eventually cause the toxicity in the host (9). Hence, the antibodies against PA have been shown to protect against aerosolized *B. anthracis* Ames spore challenge (10).

A soluble antigen when injected alone has less residence time and often lack essential danger signals necessary to provoke dendritic cells and consequently, T-cells. To surpass these limitations, antigen encapsulation into particulate system helps by increasing the retention time of the antigen in the host system and creating a depot, which releases the antigen into the system for a longer duration in a controlled manner (11, 12). The susceptibility of particles to be phagocytized by the macrophages is 1,000–10,000 times more efficient, and it also allows multimeric antigen presentation and delivery to antigen-presenting cells (APCs), in comparison to the soluble antigens (13).

The intervention of natural, biodegradable polymers-based encapsulation carriers makes them reasonably safe to employ for vaccine delivery (14, 15). Poly (lactic-co-glycolic acid) and chitosan are the most studied among them. Chitosan is derived from chitin, which is a ubiquitous natural polymer and is soluble in water under acidic conditions. Chitosan particles are usually positively charged and irregular in shape. However, they are susceptible to dissociation with ionic fluctuations and unstable at physiological pH. Several chemically modified (at functional groups -OH and -NH2) derivatives of chitosan have been studied. For instance, trimethyl-chitosan (TMC) is preferred over chitosan due to its stability at several pH ranges and its positive surface charge and has been utilized for vaccine delivery against influenza and Hepatitis B (14, 16–21).

Trimethyl-chitosan as well as chitosan have also been claimed to possess mucoadhesive and intrinsic adjuvant properties, and evidently, its nanoparticles can stimulate *in vitro*

T-cell maturation and proliferation (22). Furthermore, it also possesses the ability to open tight junctions and cross epithelial barriers by redistributing the cytoskeletal F-actin and ZO-1 tight junction protein (23–26). The precursor of TMC and chitosan, chitin, has also been demonstrated to be a size-dependent pathogen-associated molecular pattern (PAMP) and shown to interact and activate the macrophages through different TLRs and dectin-1 (27, 28).

In addition to TMC nanoparticles, we sought to investigate further whether the immune response of nanoparticles could be synergistically improved upon addition of an adjuvant. Alum or aluminum salts-based adjuvants are currently licensed by FDA for human use and are most commonly employed in combination with several antigens. Despite this, they too suffer from various shortcomings such as variable antigen adsorption, induction of weak T-cell-mediated immune response, poor maturation of APCs, and intermittent occurrence of hypersensitivity reactions (29–31). Hence, PAMP-based adjuvants CpG-C ODN 2395 and Poly I:C were used in the study, which has been documented to provide protective immunity against anthrax infection (32, 33). CpG interacts with the TLR9 receptor and initiates a cascade of immuno-stimulatory signaling events culminating in secreting various cytokines and chemokines. This, in turn, causes maturation, differentiation, and proliferation of several immune cell types and eventually perpetuates a pro-inflammatory and TH1-biased immune response (34, 35). Similarly, the interaction of Poly I:C with TLR-3 induces dendritic cells maturation and is also reported to promote the *in vitro* survival of activated CD4+ cells and *in vivo* survival of antigen-activated CD8+ cells (36–39). Cytokines also perform a crucial part in the generation of humoral and cell-mediated immunity, and hence were examined in this study post immunization. The inflammatory Th1 cytokines, comprising of IFN-γ, TNF, and IL-2, stimulate predominantly cell-mediated immunity and T-cell proliferation, along with T-helper cell differentiation (40, 41). Immune modulating Th2 cytokines, such as IL-4, IL-6, and IL-10, support humoral immunity and adaptive immunity, mediate allergic diseases, and regulate chronic infections (42, 43).

The following parameters are known to affect the immune response elicited by the particulate-antigen delivery system: (i) administration route, (ii) booster administration, and the interval between boosters (iii) delivery system and amount of antigen, (iv) inclusion of immune modulating molecules like tolllike receptors (TLRs) ligands. Hence, in this study, we evaluated the vaccine potential of various TMC-PA nano-formulations with adjuvants through subcutaneous (SC), intramuscular (IM), and intraperitoneal (IP) delivery platforms. The protective efficacy of the various TMC-PA vaccine formulations was also assessed.

#### MATERIALS AND METHODS

#### Materials

Chitosan (75–85% deacetylated, mol wt 50–190 kDa), Sodium tripolyphosphate (TPP), Poly I:C and Tween 80 were purchased from Sigma (Sigma, India). CpG-C ODN 2395 was purchased from Hycult Biotech Pvt. Ltd. All other materials used were of the analytical or pharmaceutical grade.

#### Preparation of TMC Form Chitosan

Trimethyl-chitosan was derived from chitosan as described previously by Muzarelli and Tanfani and Verheul et al. (17, 44), with some modifications. To begin with, for formic acid–formaldehyde methylation (Eschweiler-Clarke), 3 g of chitosan was dissolved in 10 ml of formic acid, 10 ml of formaldehyde, and 60 ml distilled water added one by one in a round bottom flask. Under reflux condensing conditions, it was heated in an oil bath at 70°C for 118 h, under constant magnetic stirring. The resultant viscous liquid was evaporated and treated with 1N NaOH to increase its pH to 12. The so-formed dimethyl-chitosan (DMC) was washed with distilled water and then dissolved in 25 ml of 1-methyl-2-pyrrolidinone in the presence of 0.5 ml of iodomethane. It was heated up to 40°C on an oil bath under constant magnetic stirring for overnight. The solution was then dissolved in ethanol/diethyl ether mixture (50:50) resulting in the precipitation of TMC. The precipitated TMC was separated by centrifugation and washed thrice with diethyl ether. After complete evaporation of diethyl-ether, TMC was dissolved in 50 ml of 10% NaCl and dialyzed against deionized water for 3 days. After dialysis, TMC was lyophilized to obtain a dried powder form. 1 H-nuclear magnetic resonance (1 H-NMR) spectroscopy was done to assess the presence of trimethyl groups in chitosan. The degree of quaternization (DQ) of the end product was estimated by:

$$\mathrm{DQ}\left(\%\right) = \frac{\left[\left(\mathrm{CH}\_{3}\right)\_{\flat}\right]}{\left[\mathrm{H}\right]} \times 1/\,\,\Psi \times 100\tag{1}$$

#### Preparation of PA-loaded TMC Nanoparticles (TMC-PA)

The nanoparticles were prepared by using the ionic gelation method as described previously (22, 45). A schematic representation of the preparation process is illustrated by Figure S1 in Supplementary Material. Briefly, 10 mg of TMC was dissolved in 5 ml 10 mM HEPES (pH 7.4), and PA was also prepared in the same buffer (8 mg/ml). The protein was dissolved in the TMC solution achieving a final concentration of 0.1 mg/ml under constant stirring. While continuous stirring (350 rpm), 1 ml of the crosslinking agent, TPP solution (1.7 mg/ml) was added to the TMC-PA solution drop by drop to induce ionic complexation. Weight ratio TMC:PA:TPP of 10:0.5:1.7 was taken to get the desired size of particles. An opalescent dispersion forms after TPP addition indicating the formation of nanoparticles. The nanoparticles were collected by centrifugation at 12,000 rpm for 15 min on a 10 µl glycerol bed. The particles were stored at −20°C until further use. The supernatant obtained was used for Micro BCA protein estimation. The protein encapsulation efficiency of the particles was further calculated by using Eq. 2:

$$\text{Encapsulation efficiency} \left(\% \right) = \frac{\text{protein}}{\text{weight of total protein used for}} \times 100 \qquad \text{(2)}$$

$$\text{encapsulation}$$

For size analysis, a separate batch of nanoparticles was prepared with a minor modification. 0.05% of Tween 80 was added after dissolving TMC in the HEPES buffer.

#### Physical Characterization and *In Vitro* Release Profile of TMC-PA Nanoparticles

The particles were dissolved in HEPES buffer pH 7.4 to measure the size, polydispersity index and zeta potential of the particles with the help of a Nanosizer ZS apparatus (Malvern Instruments, Malvern, UK). For scanning electron microscopic (SEM) imaging, particles were coated on a carbon tape on an aluminum stub and then coated with gold particles at 2 kV for 200 s under inert Argon condition. SEM images were captured using an electron microscope [Zeiss EVO40 (Carl Zeiss, Thornwood, NY, USA)]. For transmission electron microscopy [JEM 2100 F (Jeol Ltd., Tokyo, Japan), Electron Microscopy Sciences, Hatfield, PA] the nanoparticles dissolved in HEPES buffer were dropped (20 µl) on a copper mesh grid (46). After evaporating the buffer, the grid was inserted in the instrument and imaging was done in a high vacuum, at 200 kV and direct magnification of 3,000X.

For the *in vitro* protein release profile of the nanoparticles, a definite amount of particles were dissolved in 1× PBS and were stirred at 100 rpm at 37°C in several aliquots. Each aliquot was taken out from stirring at a specific time point and was centrifuged to spin down the nanoparticles, and the supernatant was taken to estimate the protein released by the nanoparticles in that specific time interval. Consequently, at different time points (2, 4, 6, 8, 10, 12, 24, 48, and 96 h), the release of protein was measured to assess the release of nanoparticles. The protein release was monitored for 96 h.

#### *In Vivo* Immunological Study

Mice were obtained from National Centre for Laboratory Animal Sciences, NIN, Hyderabad, India, and were maintained in animal holding room of BSL3 laboratory. All the animal experiments including challenge studies were performed in compliance with Institutional Animal Ethics Committee (Jawaharlal Nehru University) and Council for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA, Ministry of Social Justice and Empowerment, Government of India). Female Balb/c mice (6–8 weeks of age) were immunized with the placebo nanoparticles (Blank NP), TMC-PA nanoparticles (TMC-PA), TMC-PA nanoparticles in combination with CpG (CpG TMC-PA), and Poly I:C (Poly I:C TMC-PA) (*n* = 21 per group). All these formulations were administered in mice *via* SC, IM and IP routes. The amount of PA administered with every formulation was 20 μg/mice. CpG (20 μg/mice) as well as Poly I:C (10 μg/mice) were mixed with the nanoparticles right before the administration into mice. After an initial prime dose, two successive booster doses were given at an interval of 7 and 15 days, respectively (as depicted in Figure S2 in Supplementary Material). Sera were collected on 28th and 42nd day and stored at −80°C until further use. The sera samples were then used for the estimation of total IgG, IgG1, and IgG2a endpoint antibody titer against PA (anti-PA) by ELISA. Administration routes used in the study and the corresponding volume of dose injected into each animal are detailed in Table S1 in Supplementary Material.

## Cytokine Analysis

After 42 days of primary immunization, three mice from each group were sacrificed, and their spleens were crushed to make a single cell suspension. 106 cells were seeded in each well of a 96-well round bottom (Nunc) plate. The cells were stimulated with PA (10 µg/ml) to induce the production of cytokines in the supernatant. Unstimulated and Con A (5 μg/well) treated cells were taken as negative and positive control, respectively. After 48 h of stimulation, the supernatant was collected from each well and was used for cytokine measurement using CBA (Cytometric Bead Array) Mouse Th1/Th2/Th17 Cytokine Kit (BD Biosciences), according to manufacturer instructions. The data were acquired on a BD LSR™ II flow cytometer (Becton Dickinson) and analyzed using the FCAP Array software V3.0 (Becton 519 Dickinson).

## Survival Assay/Protective Efficacy Studies

At 43rd day, remaining mice in each group were challenged with 0.5 × 103 spores of *B. anthracis* Ames strain. As a control, 20 µg of PA adsorbed on alum was subcutaneously injected into a separate mice group (*n* = 12) and then challenged simultaneously. Mice were monitored for 15 days for death events in each group. Survival curve was plotted to compare the efficiency of protection in vaccinated mice groups over control blank nanoparticle immunized mice group.

#### Statistical Analysis

The results are reported as mean ± SE post data preparation and statistical analysis using GraphPad Prism v6.05 software. The statistical significance of antibody titer and cytokine level data was calculated by using one-way ANOVA, followed by Tukey's multiple comparisons test. The survival curve for anthrax spore challenge experiment was evaluated using Kaplan–Meier survival estimates (GraphPad Prism, La Jolla, CA, USA). Statistically significant differences between the groups are highlighted by following denotations: \* for *P*-value <0.05, \*\* for *P*-values <0.01, \*\*\* for *P*-values <0.001, and \*\*\*\* for *P*-values <0.0001.

## RESULTS

#### NMR Characterization of the Trimethyl Chitosan

First, a dimethyl derivative of chitosan (DMC) was obtained with formic acid and formaldehyde treatment. Followed by derivation of TMC from DMC, 1 H-NMR spectrum was analyzed for the presence of trimethyl group in chitosan. The peak present at 3.33 ppm represents the trimethyl group incorporated into the polymer at the -NH2 group of chitosan, imparting a positive charge to the polymer (Figure S3 in Supplementary Material). The final product was completely soluble in water, and the NMR was carried out by dissolving the final product in D2O. The DQ calculated by the Eq. 1 above was 80.4 ± 5.2%. Negligible O-methylation was observed in the product as seen in the NMR profile.

## PA Encapsulated Nanoparticles Preparation

TMC-PA nanoparticles were prepared by ionic gelation method with the help of TPP anion. The particles were in the nanometer range with an average particle size of 254 ± 24 nm as depicted in **Figure 1D**. The particles shape and morphology was observed by SEM and TEM as shown in **Figures 1A,B**, and the nanoparticles were found to be smooth in texture but irregular in shape. The encapsulation efficiency of the particles was found to be 78.33 ± 4.75%. The polydispersity index of the particles was found to be 0.2 indicating fine homogeneity of the particles. **Figure 1E** reveals the zeta potential of the protein-loaded nanoparticles was found to be 8.84 ± 0.25 mV, which confirmed that the particles carried a positive charge.

## *In Vitro* Antigen Release

The antigen release profile of the TMC-PA nanoparticles was taken into account to estimate the amount of protein released from the particles over a specified period. The release of protein was studied until 96 h, illustrated in **Figure 1C**. Around 30% of the protein was released in the first 6 h only, and approximately 55% of the protein was released within first 24 h. Afterward, protein release was quite slow, and in next 24 h, only 7% protein was released. For the next 24 h also, only 8% release was observed. Therefore, in the later period, the release was slow and consistent inferring that remaining 30% of the protein would also be released at the same pace at these conditions. This suggests that at physiological pH, PA would be completely released within a week from the particles. Since 50% of the protein was burst released in the first 24 h; therefore, for immunization studies, 20 µg of protein encapsulating nanoparticles were used per mice. The encapsulated protein was analyzed on SDS polyacrylamide upon dissolution of the nanoparticles with 1% NaCl. The protein was observed at its molecular weight of 83KDa, and no protein degradation was observed, as seen in **Figure 1F**.

## Immune Response Generated in Mice by TMC-PA Alone and Upon Adjuvanation with CpG and Poly I:C

The immune response was elucidated in terms of the total IgG titer raised against the antigen. Since the mice were immunized *via* three routes: subcutaneous, IM, and IP; therefore, the comparison between different groups was also carried out within one frame of the route at a time.

## Total IgG Titer

**Figure 2** illustrates the highly elevated total IgG titer in almost all the vaccinated mice groups when compared to only PA vaccinated control mice. In all three routes, the descending order for the titers was CpG TMC-PA > Poly I:C TMC-PA > TMC-PA > Poly I:C PA > CpG PA. It could be seen that CpG and Poly I:C adjuvanation increased the total IgG titer from TMC-PA group significantly in almost all the three routes. In the subcutaneous route, the total IgG titer of the CpG TMC-PA group was significantly (*P* < 0.0001) higher than Poly I:C TMC-PA group on the 28th day (1.45-fold higher), whereas on 42nd day, the two titers were

not significantly different. However in IM route, on both 28th and the 42nd day, the total IgG titer of CpG TMC-PA group is significantly (*P* < 0.01 and *P* < 0.0001, respectively) higher than Poly I:C TMC-PA group (1.44-fold on the 28th day and 1.5-fold on the 42nd day). Also, the TMC-PA, CpG TMC-PA, and Poly I:C TMC-PA groups showed significantly higher titer than the only PA group. These results elucidate the potential of TMC in mediating a very good humoral response in comparison to the only protein group and upon adjuvanation with CpG or Poly I:C, it increased in a synergistic manner. In all the three routes, the TMC-PA group showed significantly higher titer than CpG PA group and Poly I:C PA group (*P* < 0.0001). This shows that the TMC acts as a strong and better adjuvant in the form of nanoparticles with the PA protein and CpG as well Poly I:C adjuvanation improves the immune response mounted by TMC-PA alone.

#### IgG Isotypes

The levels of IgG1 (**Figure 3**) were higher in all TMC-PA containing groups in comparison to other groups, and this pattern was similar for all the three routes. On 28th and 42nd day, the TMC-PA group showed the highest titer in SC and IM route. However, for IP route, Poly I:C TMC-PA group showed the highest titer on both 28th and 42nd day followed by the TMC-PA group. The Poly I:C PA and CpG PA groups showed substantially lower levels of IgG1 in comparison to the other nanoformulation vaccinated groups at both the time points and all the routes (*P* < 0.0001).

In case of IgG2a subtype (**Figure 4**), among all routes, on the 28th and 42nd day, the Poly I:C TMC-PA and CpG TMC-PA groups showed the highest titers. The TMC-PA, CpG PA, and Poly I:C PA groups showed significantly lower titers when compared to the CpG and Poly I:C adjuvated TMC-PA groups, through all the routes, and at 42nd day, there was no significant difference in their titers as well, except in SC route. IgG isotypes have been of importance as to determine the Th biasness of the immune response. TMC has been reported to have a Th2-directing effect on the immune response, whereas CpG and Poly I:C have been reported to have a Th1 biasing effect on the immune response.

#### Cytokine Release and Th1/Th2 Immune Response in Mice Groups CpG TMC-PA Group

Based on the cytokine levels and the levels of IgG isotypes (IgG1 and IgG2a), it was seen that CpG TMC-PA SC/IM/IP groups of mice developed a dominant Th1-biased immune response. By 42nd day, the ratio of IgG2a vs IgG1 was found to be 2.4 for SC route, 2.3 for IM route, and 2.3 for IP route (**Figures 3** and **4**). The IFN-γ levels for these groups for SC, IM, and IP routes were highest among all the groups with no significant difference between them (**Figure 5A**). IL-4 levels for this group were <10 pg/ml for all three routes with no significant difference among them (**Figure 5D**). Similarly, IL-10 levels were <150 pg/ml and IL-6 levels were ~200 pg/ml, with no significant difference among all

three routes (**Figures 5E,F**). TNF-α level for the SC, IM, and IP routes were ~500 pg/ml, with no significant difference between different routes (**Figure 5B**). However, IL-2 levels for IM route were significantly higher (*P* < 0.0001) than SC and IP routes (**Figure 5C**).

#### CpG PA Group

This group also showed a Th1-skewed response as the ratio of IgG2a:IgG1 for SC, IM, and IP routes were 2.1, 1.8, and 1.7, respectively at 42nd day (**Figures 3** and **4**). The IFN-γ levels were significantly different between SC vs IM route (*P* < 0.0001) and SC vs IP route (*P*< 0.0001); however, no significant difference was observed between IM vs IP routes (**Figure 5A**). The Th2 cytokine IL-4 was observed to have very low levels (<10 pg/ml) supporting the Th1 biasness (**Figure 5D**). TNF-α levels for SC, IM, and IP routes were ~400 pg/ml (**Figure 5B**). The IL-2 levels of the SC route were significantly higher in comparison to IM (*P* < 0.0001) and IP (*P* < 0.0001) routes (**Figure 5C**). IL-10 and IL-6 levels for all three routes were <160 pg/ml, with no significant difference amongst the three routes (**Figures 5E,F**).

#### Poly I:C TMC-PA Group

At 42nd day, the ratios of IgG2a:IgG1 for SC, IM, and IP routes were 3.2, 3.0, and 2.5, respectively, suggesting a dominant Th1 response (**Figures 3** and **4**). In contrast, the IFN-γ and IL-4 levels do not support a strong Th-1 response (**Figures 5A,D**). Similarly, the levels of TNF-α and IL-2 were relatively low and exhibited no significant difference between the routes (**Figure 5B**). Both IL-10 and IL-6 levels were relatively higher with the IL-10 levels of IP route significantly higher than SC (*P* < 0.0001) and IM route (*P* < 0.001) (**Figures 5E,F**). The IL-6 levels for SC, IM and IP routes also revealed a highly significant inter-route difference (*P* < 0.0001) (**Figure 5F**).

#### Poly I:C PA Group

At 42nd day, the ratios of IgG2a:IgG1 for SC, IM, and IP routes were 1.9, 1.6, and 1.4, respectively, suggesting a moderate Th1 response (**Figures 3** and **4**). Similarly, the IFN-γ levels were relatively high showing a significant difference between SC vs IM route (*P* < 0.001) and SC vs IP (*P* < 0.01) routes whereas the IL-4 levels for all routes were >20 pg/ml, also suggesting a moderate Th-1 response (**Figures 5A,D**). The levels of TNF-α for the IM route were significantly higher than SC (*P* < 0.001) and IP (*P* < 0.0001) routes (**Figure 5B**). The IL-2 levels of IM and IP routes were found to be significantly higher (*P* < 0.0001) than SC route (**Figure 5C**). In contrast to the moderate Th-1 response demonstrated by above parameters, IL-10 values of SC and IP

routes were higher, whereas IM route response was moderate (**Figure 5E**). The IL-6 levels for all three routes were observed to be <350 pg/ml (**Figure 5F**).

#### TMC-PA Group

At 42nd day, the ratios of IgG1:IgG2a for SC, IM, and IP routes were 2.9, 3.5, and 1.6, respectively, suggesting a dominant Th2 response (**Figures 3** and **4**). However, IFN-γ levels and IL-4 levels suggest a Th1 biased immune response (**Figures 5A,D**). As the values of IL-10 and IL-6 fall in the range of 500–1,200 pg/ml, it suggests a Th2 response, which is also supported by low levels (<500 pg/ml) of TNF-α and IL-2 (**Figures 5B–F**).

#### IL-17 A Cytokine

**Figure 5G** illustrates that the highest levels of IL-17A were observed in the TMC-PA IP group, followed by Poly:IC TMC PA IP group and Poly:IC TMC PA IM being ~4,000 pg/ml. The Poly:IC PA SC and TMC-PA SC groups produced similar levels of this cytokine. Poly:IC PA IP and Poly:IC TMC PA SC groups secreted ~1,000 pg/ml of IL-17. All the remaining groups secreted relatively lower amount of cytokine.

#### Survival Curve

After 6 weeks of immunization, all the mice groups were challenged with 0.5 × 103 spores of *B. anthracis* Ames strain. The mice were monitored for 2 weeks to observe the protective efficacy of different formulations as shown in **Figure 6**. The mice vaccinated with Alum and PA were taken as a positive control and exhibited a survival of 83.3%, which was the maximum of all the groups examined. The CpG TMC-PA SC, Poly I:C TMC-PA SC, and Poly I:C TMC-PA IM groups also showed a survival rate of 83.3%, equivalent to the positive control used. Except for the blank nanoparticles and only PA-treated groups, all other mice groups showed a survival rate of more than 50%. The TMC-PA SC group showed a survival of 75% and the TMC-PA IM and TMC-PA IP groups showed a survival of 66.6%. Our study demonstrates that the SC route of immunization showed a better survival rate in comparison to the other two routes.

## DISCUSSION

The need to expand the arsenal of adjuvants and development of new subunit vaccine against anthrax with fewer side effects has set out an exploration of the natural, biodegradable polymers with enhanced bioavailability. This study investigated a novel combination of TMC nanoparticles with PA as a potential vaccine candidate against anthrax. To the best of author's knowledge, this novel approach has not been reported to-date.

The preparation of TMC and its nanoparticles, its subsequent characterization of physicochemical properties, and *in vitro* release kinetics was studied before performing the murine studies. The synthesis of TMC polymer followed an intermediate step of formation of dimethyl chitosan. This method was adapted to avoid the O-methylation of the chain and chain scission which occurs commonly due to prolonged harsh conditions during the preparation of TMC. A DQ of >30% and lower extent of O-methylation enhances the water solubility of TMC. This property of TMC was leveraged for the preparation of PA-loaded TMC nanoparticles at pH 7 in HEPES buffer (17, 44). The TPPbased ionic gelation method follows the principle of complexation of oppositely charged molecules (protonated amine of TMC and TPP anion), resulting in the formation of nanoparticles (19, 47, 48). Unlike the widely used methods utilizing organic solvents and high temperature, the ionic gelation method is very mild, and thus we did not observe any degradation products in the SDS PAGE analysis (49–51).

The average size of PA-loaded TMC nanoparticles (254 nm) was found to be suitable for an efficient uptake by APCs. As the TMC nanoparticles were positively charged at physiological pH, an electrostatic interaction with the negatively charged cell membrane is highly likely. This is in alignment with a previously reported study where insulin, ovalbumin (OVA), and tetanus toxoid-based chitosan nanoparticles were shown to have a similar particle size, charge distribution (47, 52–54). In terms of the release kinetics of protein, the observed pattern of an initial burst release followed by a controlled release is also supported by other studies (52, 53). Thus, the *in vitro* antigen release study confirms the long-term antigen-releasing capacity of PA-TMCs without any protein degradation post-encapsulation.

After successful preparation and characterization of TMC nanoparticles, mice studies were conducted. Although the nasal route has been extensively explored for TMC nanoparticle-based immunization (19, 23, 47, 52, 55), we did not achieve a satisfactory immune response modulation. This could potentially be attributed to the large size of TMC-PA nanoparticles and subsequently its poor diffusion across the epithelial layer covering nasal-associated lymphoid tissue. In a similar instance, Boonyo et al. and Hagenaars et al. reported a subdued intranasal immune response for an OVA and HI-loaded TMC nanoparticle system (52, 55). Thus, the SC, IM, and IP routes were subsequently pursued for immunization.

The results of our study, as evident from **Figures 2**–**6**, demonstrated an effective immune response for TMC nanoparticles by all three routes (SC, IM, and IP). The striking results were that the TMC-PA groups *via* all routes were able to produce a sufficient amount of anti-PA titer as well as provided nearly 60–70%

protection in anthrax challenged mice. More interestingly, the adjuvants CpG and Poly I:C behaved synergistically with TMC-PA in terms of generating an elevated anti-PA antibody titer and survival efficacy of nearly 70–80%. However, SC route was found be more effective in generating the highest IgG titer. Such response could be attributed to longer persistence, slower clearance, and prolonged antigen presentation after SC administration (56). Consequently, we could relate to the large number of vaccines, which are currently being administered *via* SC route like mumps, rubella, yellow fever, Japanese encephalitis, measles, Haemophilus influenza type b, inactivated polio vaccine, and *S. pneumonia* (25, 56).

In line with the previously reported studies, the TMC-PA group generated a Th2-biased immune response (25, 34). On the other hand, Poly I:C and CpG are reportedly Th1 biasing adjuvants (57–60). CpG alone has been reported to establish a Th1-biased immune response with several antigens (61–63). In this study, Poly I:C administration modulated the immune response toward Th1 with PA and TMC-PA nanoparticles. Supporting the earlier evidence, the CpG PA groups generated a Th1 response *via* all the three routes. Despite the fact that the PA is a Th2 biased antigen and TMC is also a Th2 biased adjuvant, the CpG adjuvated TMC-PA-treated groups also generated a dominant Th1 response *via* all the three routes. This observation is supported by Slütter and Jiskoot (34), which reported that the TMC-CpG nanoparticles encapsulating OVA as an antigen generated a 10-fold higher Ig2a titer than the TMC-TPP nanoparticles encapsulating OVA, thus provoked a strong Th1 response. The same report also revealed that the antigen-stimulated splenocytes secreted elevated levels of IFN-γ. This was also observed in our study that the CpG PA and CpG adjuvated TMC-PA groups showed highest levels of IFN-γ. In another study by Wang et al., virus-like particles encapsulating CpG-gold nanoparticle conjugates elicited a strong IFN-γ secretion (64). Therefore, we conclude that CpG enhances the production of IFN-γ, upon adjuvanation with an antigen. However, the levels of IL-4 produced by the splenocytes were low for all the groups, but the TMC-PA group relatively secreted the highest IL-4 irrespective of the route of immunization. Although, the CD4+ and CD8+ T-cell count gives a clear indication about the Th1 or Th2 response, but in this study, we did not analyze these parameters. However, the cytokines levels do indicate the direction of biasness of the immune response.

IL-17 T-helper cells are well known to serve as a bridge between innate and acquired immunity. Upon sensing the cytokines in the mucosal sites, the memory effector subset of Th17 cells can transform into Th1 or Th2 phenotype. IL-17A is the hallmark cytokine of the proinflammatory Th cell subset of Th17. As evident from **Figure 5**, Poly I:C PA and Poly I:C TMC-PA groups generated

high levels of IL-17A. The TMC-PA group also secreted high levels of IL-17A, except the IM route. In a study by Holm et al., Poly I:C was reported to stimulate the production of IL-17A and IL-21 directly and consequently, drive the human naïve CD4+ T-cells differentiation, whereas CpG did not affect IL-17A on either mRNA or protein level (65). Similarly, Vultaggio et al. reported a positive correlation between Poly I:C and 1L-17A production (66). The chitin-based adjuvants have also been reported to be an immune response modulator (67) with a size-dependent response on macrophage IL-17A production (28). Thus, the growing scientific evidence indicates that TMC plays a role in stimulating IL-17A production. However, the mechanism of action remains unclear and need further exploration.

CpG and Poly I:C are PAMP-based adjuvants and act by initiating a cascade of innate immune system stimulation by interacting with TLR-3 and TLR-9, respectively. However, the immune modulation mechanism behind TMC nanoparticles remains elusive. We hypothesize that this immune-modulated response is due to the positively charged TMC nanoparticles, which might act as "danger signal" for dendritic cells.

In conclusion, the PA-loaded TMC nanoparticles as well as CpG and Poly I:C adjuvanted TMC-PA formulations elicited strong IgG antibody response *via* SC, IM, and IP routes in mice. However, the SC route showed the strongest response for IgG titer and survival efficacy. The TMC nanoparticle formulations were able to protect the mice against anthrax challenge and were comparable in protective efficacy to alum. In our opinion, TMC-based formulations possess a great potential as a vaccine candidate against anthrax, and further studies are needed to explore its adjuvant mechanisms.

#### ETHICS STATEMENT

All mice experiments were performed in compliance with Institutional Animal Ethics Committee (Jawaharlal Nehru University) and Council for the Purpose of Control and Supervision of Experiments on Animals (CPCSEA, Ministry of Social Justice and Empowerment, Government of India). Mice were housed in the individually ventilated animal caging system.

#### AUTHOR CONTRIBUTIONS

Experiments conceptualized and designed by AM. Performed by AM, MG, and RM. Data analysis by AM. Wrote the article: AM, HG, and RB. Supervised by RB.

#### ACKNOWLEDGMENTS

We would like to thank all the members of the laboratory for thoughtful and lively discussions during this work. The author AM would like to thank Council of Scientific and Industrial Research (CSIR), India, Department of Biotechnology (DBT), India, and JC Bose fellowship for providing financial support during this research term.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.00562/ 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 Malik, Gupta, Mani, Gogoi and Bhatnagar. 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:*

*Rajko Reljic, St George's, University of London, United Kingdom*

#### *Reviewed by:*

*Sylvie Fournel, Université de Strasbourg, France Michael Schotsaert, Icahn School of Medicine at Mount Sinai, United States*

#### *\*Correspondence:*

*Yasuo Yoshioka y-yoshioka@biken.osaka-u.ac.jp*

#### *Specialty section:*

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

*Received: 28 December 2017 Accepted: 28 March 2018 Published: 18 April 2018*

#### *Citation:*

*Takahashi H, Misato K, Aoshi T, Yamamoto Y, Kubota Y, Wu X, Kuroda E, Ishii KJ, Yamamoto H and Yoshioka Y (2018) Carbonate Apatite Nanoparticles Act as Potent Vaccine Adjuvant Delivery Vehicles by Enhancing Cytokine Production Induced by Encapsulated Cytosine-Phosphate-Guanine Oligodeoxynucleotides. Front. Immunol. 9:783. doi: 10.3389/fimmu.2018.00783*

*Hideki Takahashi <sup>1</sup> , Kazuki Misato1 , Taiki Aoshi 2,3, Yasuyuki Yamamoto4 , Yui Kubota5 , Xin Wu5 , Etsushi Kuroda6,7, Ken J. Ishii 6,7, Hirofumi Yamamoto5 and Yasuo Yoshioka1,4,8,9\**

*1Vaccine Creation Project, BIKEN Innovative Vaccine Research Alliance Laboratories, Research Institute for Microbial Diseases, Osaka University, Suita, Japan, 2Vaccine Dynamics Project, BIKEN Innovative Vaccine Research Alliance Laboratories, Research Institute for Microbial Diseases, Osaka University, Suita, Japan, 3Vaccine Dynamics Project, BIKEN Center for Innovative Vaccine Research and Development, The Research Foundation for Microbial Diseases of Osaka University, Suita, Japan, 4Vaccine Creation Project, BIKEN Center for Innovative Vaccine Research and Development, The Research Foundation for Microbial Diseases of Osaka University, Suita, Japan, 5Division of Health Sciences, Department of Molecular Pathology, Graduate School of Medicine, Osaka University, Suita, Japan, 6 Laboratory of Adjuvant Innovation, Center for Vaccine and Adjuvant Research, National Institutes of Biomedical Innovation, Health and Nutrition, Ibaraki, Japan, <sup>7</sup> Laboratory of Vaccine Science, WPI Immunology Frontier Research Center, Osaka University, Suita, Japan, 8 Laboratory of Nano-Design for Innovative Drug Development, Graduate School of Pharmaceutical Sciences, Osaka University, Suita, Japan, 9Global Center for Medical Engineering and Informatics, Osaka University, Suita, Japan*

Vaccine adjuvants that can induce not only antigen-specific antibody responses but also Th1-type immune responses and CD8+ cytotoxic T lymphocyte responses are needed for the development of vaccines against infectious diseases and cancer. Of many available adjuvants, oligodeoxynucleotides (ODNs) with unmethylated cytosinephosphate-guanine (CpG) motifs are the most promising for inducing the necessary immune responses, and these adjuvants are currently under clinical trials in humans. However, the development of novel delivery vehicles that enhance the adjuvant effects of CpG ODNs, subsequently increasing the production of cytokines such as type-I interferons (IFNs), is highly desirable. In this study, we demonstrate the potential of pH-responsive biodegradable carbonate apatite (CA) nanoparticles as CpG ODN delivery vehicles that can enhance the production of type-I IFNs (such as IFN-α) relative to that induced by CpG ODNs and can augment the adjuvant effects of CpG ODNs *in vivo*. In contrast to CpG ODNs, CA nanoparticles containing CpG ODNs (designated CA-CpG) induced significant IFN-α production by mouse dendritic cells and human peripheral blood mononuclear cells *in vitro*; and production of interleukin-12, and IFN-γ was higher in CA-CpG-treated groups than in CpG ODNs groups. In addition, treatment with CA-CpG resulted in higher cytokine production in draining lymph nodes than did treatment with CpG ODNs *in vivo*. Furthermore, vaccination with CA-CpG plus an antigen, such as ovalbumin or influenza virus hemagglutinin, resulted in higher antigen-specific antibody responses and CD8+ cytotoxic T lymphocyte responses *in vivo*, in an interleukin-12- and type-I IFN-dependent manner, than did vaccination with the antigen plus CpG ODNs; in addition, the efficacy of the vaccine against influenza virus was higher with CA-CpG as the adjuvant than with CpG ODNs as the adjuvant. These data show the potential of CA nanoparticles to serve as CpG ODN delivery vehicles that increase the production of cytokines, especially IFN-α, induced by CpG ODNs and thus augment the efficacy of CpG ODNs as adjuvants. We expect that the strategy reported herein will facilitate the design and development of novel adjuvant delivery vehicles for vaccines.

Keywords: adjuvant, cytosine-phosphate-guanine oligodeoxynucleotide, influenza virus, interferon-**α**, nanoparticle, vaccine

#### INTRODUCTION

As the recent Ebola virus outbreak and past worldwide influenza pandemics have demonstrated, infectious diseases are a serious global health problem (1). Vaccines are the most efficient way to prevent infectious diseases, and as a result of immunological innovations such as the development of checkpoint inhibitors, vaccines are also expected to be effective against cancer in the coming years (2).

There are various types of vaccines, including attenuated live vaccines, inactivated whole vaccines, and protein- or peptidebased subunit vaccines (3–5), and they all have various advantages and disadvantages. For example, although attenuated live vaccines can generate strong adaptive immunity, the pathogens in these vaccines have the potential to cause infectious disease in rare cases, especially in individuals with compromised immune systems (3–5). In contrast, protein- or peptide-based subunit vaccines are much safer than attenuated live vaccines and inactivated whole vaccines; however, when used alone, subunit vaccines evoke only weak adaptive immunity. The efficacy of subunit vaccines can be improved by using adjuvants (6). Aluminum salts (alum) are the most widely used adjuvants and are present in many important vaccines, including the diphtheria–tetanus–pertussisinactivated poliomyelitis vaccine, the pneumococcal conjugate vaccine, and hepatitis B vaccine in humans (7). Unfortunately, although alum induces strong Th2-type immune responses, it cannot induce Th1-type immune responses or CD8<sup>+</sup> cytotoxic T lymphocyte (CTL) responses, which are necessary for vaccines against infectious diseases such as influenza virus, hepatitis C, and malaria as well as for cancer vaccines designed to eliminate infected or malignant cells (8). Therefore, the development of adjuvants that can induce Th1-type immune responses and CD8<sup>+</sup> CTL responses is necessary.

Oligodeoxynucleotides (ODNs) with unmethylated cytosinephosphate-guanine (CpG) motifs, which act as Toll-like receptor 9 (TLR9) agonists, are among the most effective adjuvants for inducing Th1-type immune responses and CD8<sup>+</sup> CTL responses (9). CpG ODNs, which are short single-stranded synthetic DNA fragments containing immunostimulatory CpG motifs, bind to TLR9 at endosomes after being taken up by B cells and dendritic cells (DCs). This binding initiates an innate immune response characterized by the production of cytokines that can promote Th1-type immune responses. Therefore, CpG ODNs are expected to be useful as vaccine adjuvants and immunotherapeutic agents against infectious diseases and cancer (9). In fact, in Phase II clinical trials, a hepatitis B vaccine with CpG ODNs as an adjuvant has demonstrated superiority to a currently licensed hepatitis B virus vaccine with alum as an adjuvant (10, 11). There are at least four types of CpG ODNs, each of which has a different structure and different physical and immunostimulatory properties (9). Of the various types, D-type CpG ODNs (also known as A-type CpG ODNs) and K-type CpG ODNs (also known as B-type CpG ODNs) are the major types for experimental and clinical uses. D-type CpG ODNs can induce plasmacytoid DCs to produce massive amounts of type-I interferons (IFNs) such as IFN-α. IFN-α is important for the activation and cytotoxicity of natural killer T cells (12), for CD8<sup>+</sup> T-cell activation (13), and for the maturation of DCs, thus indicating the importance of adjuvants that can induce IFN-α production. D-type CpG ODNs selfassemble into multimers because of their palindromic and poly-G sequences, and these multimers are important for IFN-α induction (14). However, uncontrollable aggregation hinders the clinical utility of D-type CpG ODNs. In contrast, K-type CpG ODNs can activate the production of interleukin (IL)-6 by B cells and DCs but only weakly induce IFN-α production by plasmacytoid DCs. These differences in cytokine production between D-type and K-type CpG ODNs are thought to be the reason for their different adjuvant effects in the treatment of some infectious diseases. For example, K-type CpG ODNs induce better antibody production and cytokine responses than D-type CpG ODNs when used in a malaria vaccine in cynomolgus monkeys (15), whereas D-type CpG ODNs elicit better immune responses than K-type CpG ODNs in a leishmania vaccine in rhesus monkeys (16). In addition, when used as a monotherapeutic for leishmaniasis, D-type CpG ODNs have better therapeutic effects than K-type CpG ODNs (17). Therefore, the design of CpG ODN-based adjuvants with the functions of both D-type and K-type CpG ODNs would be highly desirable.

**Abbreviations:** alum, aluminum salts; CA, carbonate apatite; CpG, cytosinephosphate-guanine; CTL, cytotoxic T lymphocyte; DCs, dendritic cells; ELISA, enzyme-linked immunosorbent assays; FCS, fetal calf serum; HRP, Horseradish peroxidase; IFNs, interferons; Ifnar2, IFN-α/β receptor 2; IL, interleukin; ODNs, oligodeoxynucleotides; OVA, ovalbumin; PBMCs, peripheral blood mononuclear cells; SV, ether-treated hemagglutinin antigen-enriched virion-free split vaccine; TLR9, toll-like receptor 9.

To design such multifunctional CpG ODNs, researchers have tried to use multimerization to give K-type CpG ODNs the ability to induce IFN-α production. There are two main methods for multimerizing K-type CpG ODNs: the use of biological molecules and the use of synthetic particles. With regard to the former, several types of peptides (e.g., antimicrobial peptide LL37 and Tat peptide) and β-glucan have been used to enhance the ability of K-type CpG ODNs to induce IFN-α production and to increase their adjuvant activity both *in vitro* and *in vivo* (18–20). In contrast, only limited data are available on the use of synthetic particles to enhance the adjuvant activity of K-type CpG ODNs *in vivo* by giving them the ability to induce IFN-α production, although the encapsulation of K-type CpG ODNs in, or their conjugation with, various types of particles has been shown to enhance IFN-α production *in vitro* (14, 21, 22). Therefore, the development of novel particles that contain encapsulated K-type CpG ODNs and can induce IFN-α production would be useful and would facilitate investigations of the relationship between IFN-α production *in vitro* and vaccine effects *in vivo*.

Among the synthetic particles that have been studied to date, pH-sensitive carbonate apatite (CA) particles have been shown to serve as efficient intracellular delivery vehicles for small molecules, as well as for DNA, RNA, and proteins (23–25). For example, the DNA transfection efficiency of CA nanoparticles is 10–100 times that of a general transfection reagent, Lipofectamine, and calcium phosphate precipitation in mammalian cells *in vitro* (23). Furthermore, Yamamoto and colleagues recently developed CA nanoparticles of 10–20 nm diameters by modifying the process by which the nanoparticles were generated (26). Wu et al. reported a simple method for generating CA nanoparticles in which siRNA is encapsulated with high efficiency (26). The resulting CA nanoparticles are highly stable at pH 7.4 but quickly degrade in acidic environments such as endosomal compartments in cells. These siRNA-containing CA nanoparticles show dramatic antitumor effects after intravenous administration and are nontoxic in mice and monkeys (26). These facts suggest that CA nanoparticles have the potential to serve as CpG ODN delivery vehicles owing to their high encapsulation efficiency, low cost, biodegradability, safety, and biocompatibility.

In this study, we show the usefulness of CA nanoparticles as CpG ODN delivery vehicles that give K-type CpG ODNs the ability to induce IFN-α production. We found that CA nanoparticles containing K-type CpG ODNs (designated CA-CpG) showed stronger antigen-specific antibody responses and CD8<sup>+</sup> CTL responses than those of K-type CpG ODNs, and when mice were vaccinated with the nanoparticles in combination with an antigen, the preventive vaccine effects against influenza virus were better than those of the antigen plus K-type CpG ODNs. This is the first report of the use of pH-responsive biodegradable CA nanoparticles to improve the cytokine production pattern of K-type CpG ODNs and enhance their adjuvant activity.

#### MATERIALS AND METHODS

#### Reagents

K-type CpG ODNs (CpG K3: 5′-atcgactctcgagcgttctc-3′) and Alexa 488-labeled K-type CpG ODNs (Alexa 488-labeled CpG K3) were purchased from GeneDesign (Osaka, Japan). EndoGrade endotoxin-free ovalbumin (OVA) (<0.1 EU/mg protein) purchased from Hyglos GmbH (Bayern, Germany) was used for immunization. Grade V OVA purchased from Sigma-Aldrich (St. Louis, MO, USA) was used for enzyme-linked immunosorbent assays (ELISA). Horseradish peroxidase (HRP)-conjugated goat anti-mouse IgG and IgG1 antibodies were purchased from Merck Millipore (Darmstadt, Germany). HRP-conjugated goat anti-mouse IgG2c was purchased from GeneTex, Inc. (Irvine, CA, USA). Alhydrogel adjuvant 2% (alum) was purchased from InvivoGen (San Diego, CA, USA).

#### Mice

C57BL/6J mice were purchased from SLC (Kyoto, Japan). *IFN-α/β receptor 2 (Ifnar2)-Il-12 p40* double-deficient mice have been described previously (27) and were kindly provided by Dr. Ishii (National Institutes of Biomedical Innovation, Health and Nutrition, Ibaraki, Osaka, Japan). Mice were used at 7 to 8 weeks of age and were housed in a room with a 12-h-light/12 h-dark cycle (lights on at 8:00 am, lights off at 8:00 pm) with access to food and water. All animal experiments were performed in accordance with institutional guidelines of Osaka University for the ethical treatment of animals (protocol number H26-11-0).

#### Synthesis of CA Nanoparticles and CA-CpG

Carbonate apatite nanoparticles and CA-CpG were prepared as described previously (26). We incubated a 25 mL solution of NaHCO3 (44 mM), NaH2PO4 (0.9 mM), and CaCl2 (1.8 mM) (pH 7.5) without or with 1.9, 5.6, 16.7, 50, or 150 µg of K-type CpG ODNs at 37°C for 30 min. The solution was then centrifuged at 12,000 rpm for 3 min, and the pellet was dissolved in saline containing 0.5% mouse serum. The resulting solution of CA nanoparticles or CA-CpG was sonicated in a water bath for 10 min. For determination of the amount of endotoxin, the collected CA-CpG was dissolved in 1 mL of 0.02 M ethylenediaminetetraacetic acid and the amount of endotoxin was determined by means of a Toxicolor LS-50M kit (Seikagaku Corporation, Tokyo, Japan). CA-CpG was found to include <0.2 EU/mL of endotoxin.

#### Efficiency of K-Type CpG ODN Encapsulation Into CA Nanoparticles

For determination of the K-type CpG ODN encapsulation efficiency, the collected CA-CpG was dissolved in 1 mL of 0.02 M ethylenediaminetetraacetic acid and the amount of K-type CpG ODNs was determined with a Qubit ssDNA Assay Kit (Thermo Fisher Scientific, Waltham, MA, USA).

### Properties of CA-CpG and CA Nanoparticles

The size distributions of CA nanoparticles and CA-CpG were analyzed with a Zetasizer Nano ZS (Malvern Instruments, Malvern, Worcestershire, UK). Specifically, the mean diameters and particle size distributions of the nanoparticles in mouse serum albumin (Sigma-Aldrich) were measured by means of a dynamic light-scattering method using capillary cells. The size distributions of both types of nanoparticles were also analyzed by atomic force microscopy conducted with a scanning probe microscope (SPM-9500, Shimadzu, Kyoto, Japan) operated in dynamic mode and equipped with a microcantilever (OMCL-AC240TS-R3, Olympus, Tokyo, Japan).

## Preparation and Stimulation of Mouse Bone Marrow-Derived DCs and Human Peripheral Blood Mononuclear Cells

To generate bone marrow-derived DCs, we isolated bone marrow cells from the femurs of C57BL/6J mice and cultured the cells for 7 days with 100 ng/mL human Flt-3L (PeproTech, Rocky Hill, NJ, USA). Cells were seeded at a density of 5 × 105 cells/well in a 96-well flat-bottom culture plate (Nunc, Roskilde, Denmark) and were cultured in complete RPMI medium (RPMI 1640 supplemented with 10 vol% fetal calf serum (FCS), penicillin, and streptomycin). These cells were stimulated with K-type CpG ODNs or CA-CpG for 24 h. Supernatants were subjected to ELISA to determine the levels of IFN-α (PBL Assay Science, Piscataway, NJ, USA), IFN-γ (BioLegend, San Diego, CA, USA), and IL-12 p40 (BD Biosciences, San Jose, CA, USA) according to the manufacturer's instructions. Human peripheral blood mononuclear cells (PBMCs) were obtained from two healthy adult male Japanese volunteers (26 years of age) with informed consent. After the PBMCs were prepared with Ficoll, they were plated at a concentration of 1 × 107 cells/mL in a 96-well flat-bottom culture plate and maintained in complete RPMI medium. They were stimulated with K-type CpG ODNs or CA-CpG for 24 h. Supernatants were subjected to ELISA to determine the levels of pan-IFN-α and IFN-γ according to the manufacturers' instructions. All experiments using human PBMCs were approved by the Institutional Review Board of the Research Institute for Microbial Diseases, Osaka University.

## Cellular Uptake of CA-CpG and K-Type CpG ODNs

Mouse bone marrow-derived DCs (1 × 107 cells/mL in a 96-well flat-bottom culture plate) were treated with Alexa 488-labeled K-type CpG ODNs (1.25 µg CpG ODNs/mL) or CA nanoparticles containing Alexa 488-labeled K-type CpG ODNs (CA-Alexa 488-CpG; 1.25 µg CpG ODNs/mL) for 10, 30, 90, 120, or 180 min. Cells were stained with 0.4 w/v% trypan blue solution (Wako, Osaka, Japan) to quench any Alexa 488-labeled K-type CpG ODNs bound to the cell surface, and then the cells were analyzed by means of flow cytometry (NovoCyte Flow Cytometer, ACEA Bioscience, San Diego, CA, USA). To separate various subsets of DCs from the collected cells, we incubated the cells with antimouse CD16/CD32 antibody (TONBO biosciences, San Diego, CA, USA), anti-mouse B220 antibody (BioLegend), anti-CD11c antibody (BioLegend), and anti-CD11b antibody (BioLegend) in the absence of trypan blue. In this way, the DCs were separated into the following subsets: B220<sup>+</sup> CD11c<sup>+</sup> plasmacytoid DCs, CD11b+ CD11c+ conventional DCs, and CD11b− CD11c+ conventional DCs.

## Biodistribution of CpG ODNs *In Vivo*

C57BL/6J mice were treated with saline, Alexa 488-labeled K-type CpG ODNs (8 or 40 µg CpG ODNs/mouse), or CA-Alexa 488- CpG (8 µg CpG ODNs/mouse) into the ear pinna. Twenty-four hours after administration, draining lymph nodes were collected. To prepare single-cell suspensions, we incubated the draining lymph nodes with 5 µg/mL collagenase D (Wako, Tokyo, Japan) for 60 min at 37°C. We used flow cytometry to analyze the uptake of CpG ODNs by cells in draining lymph nodes.

### *In Vivo* Activation of Lymph Node Cells

C57BL/6J mice were treated with K-type CpG ODNs (8 or 40 µg CpG ODNs/mouse) or CA-CpG (8 µg CpG ODNs/mouse) into the ear pinna. Twenty-four hours after administration, draining lymph nodes were collected. To prepare single-cell suspensions, we incubated the draining lymph nodes with collagenase D (5 µg/mL) for 60 min at 37°C. Prepared cells were incubated for 8 h, culture supernatants were collected, and cytokines were measured by means of ELISA.

#### Immunization

C57BL/6J mice were treated with OVA (100 µg/mouse) without or with K-type CpG ODNs (10 or 50 µg CpG ODNs/mouse) or OVA with CA-CpG (10 µg CpG/mouse) subcutaneously at the base of the tail at days 0 and 14. Using hematocrit capillary tubes (Terumo, Tokyo, Japan), we obtained blood samples from the retro-orbital venous plexus at day 21 and centrifuged the samples at 3,000 × *g* at 4°C. The resulting plasma was stored at −80°C until analysis. To investigate the adjuvant effects of CA-CpG administered by another route, we treated C57BL/6J mice with OVA (80 µg/mouse) without or with K-type CpG ODNs (8 or 40 µg CpG ODNs/mouse) or OVA with CA-CpG (8 µg CpG/ mouse) into the ear pinna at days 0 and 7, and we obtained blood samples at day 14.

#### Detection of OVA-Specific Antibodies

The levels of OVA-specific antibodies in plasma were determined by means of ELISA. To detect OVA-specific IgG, IgG1, and IgG2c, we coated ELISA plates (Corning, Corning, NY, USA) with OVA in carbonate buffer (10 µg/mL) overnight at 4°C. The coated plates were incubated with PBS containing 10% FCS. Plasma samples were diluted with PBS containing 10% FCS, and the dilutions were added to the OVA-coated plates. After incubation with plasma for 2 h, the coated plates were incubated with an HRP-conjugated goat anti-mouse IgG, IgG1, or IgG2c solution for 1 h at room temperature. After incubation, the color was developed with tetramethyl benzidine (Moss, Pasadena, MD, USA), the reaction was stopped with 2 N H2SO4, and the OD450–570 was measured on a microplate reader (Power Wave HT, BioTek, Winooski, VT, USA).

#### Detection of OVA-Specific CD8**<sup>+</sup>** CTL Induction

C57BL/6J mice were treated with OVA (80 µg/mouse) without or with K-type CpG ODNs (8 or 40 µg CpG ODNs/mouse) or CA-CpG (8 µg CpG ODNs/mouse) into the ear pinna. Draining lymph nodes were collected at day 7, and lymph node cells were prepared for determination of OVA257–264-specific CD8<sup>+</sup> CTL responses. Lymph node cells (1 × 106 cells) were added to the wells of a 96 half-well plate and then stimulated with OVA257–264 peptide (SIINFEKL, final concentration 5 µg/mL) for 24 h at 37°C. After the incubation, the concentration of IFN-γ in the supernatants was analyzed by means of ELISA. For the tetramer assay, lymph node cells (1 × 106 cells) were stained with purified anti-mouse CD16/CD32 antibody and then with the phycoerythrin-labeled H-2Kb /OVA257–264 tetramer (MBL, Aichi, Japan). Fluorescein isothiocyanate-conjugated CD8α (Abcam, Cambridge, UK), allophycocyanin-conjugated CD44 (BioLegend), and 7-aminoactinomycin D (BioLegend) were then added. The number of CD8α+ CD44<sup>+</sup> tetramer<sup>+</sup> cells was determined by flow cytometry.

#### Vaccine Against Influenza Virus

C57BL/6J mice were treated with ether-treated hemagglutinin antigen-enriched virion-free split vaccine (SV; strain A/California/7/2009, H1N1, 0.5 µg/mouse) without or with K-type CpG ODNs (10 or 50 µg CpG ODNs/mouse), CA-CpG (10 µg CpG ODNs/mouse), CA nanoparticles, or alum at days 0 and 14. At day 21, we obtained blood samples from the retro-orbital venous plexus and centrifuged the samples at 3,000 × *g* at 4°C; the resulting plasma was stored at −80°C until analysis. To detect SV-specific IgG, IgG1, and IgG2c, we coated ELISA plates with SV in carbonate buffer (10 µg/mL) overnight at 4°C. The coated plates were incubated with PBS containing 10% FCS. Plasma samples were diluted with PBS containing 10% FCS, and the dilutions were added to the SV-coated plates. After incubation with plasma for 2 h, the coated plates were incubated with a solution of HRP-conjugated goat anti-mouse IgG, IgG1, or IgG2c for 1 h at room temperature. After incubation, the color was developed with tetramethyl benzidine, the reaction was stopped with 2 N H2SO4, and OD450–570 was measured on a microplate reader. For analysis of hemagglutination inhibition (HI) titers, plasma samples were incubated with RDE (II) (DENKA SEIKEN, Tokyo, Japan) for 18 h at 37°C and then heated at 56°C for 30 min to deactivate the enzyme. To avoid nonspecific hemagglutination, plasma samples were treated with guinea pig red blood cells for 1 h at room temperature, and the supernatants were serially diluted 2-fold in microtiter plates. An equal volume of influenza virus (strain A/California/7/2009, H1N1) was added to each well, and the plates were incubated for 1 h at room temperature. Guinea pig red blood cells were added and allowed to settle for 1 h at 4°C. The HI titer was determined from the dilution of the last well that contained non-agglutinated guinea pig red blood cells. Moreover, immunized mice were challenged intranasally with 7.2 × 103 pfu (10-LD50) of influenza virus (strain A/PR/8/34, H1N1) (28). Body weights and survival rates of challenged mice were monitored for 14 days. Ether-treated hemagglutinin antigen-enriched virion-free split vaccine and influenza virus were kindly provided by the Research Foundation for Microbial Diseases of Osaka University, Suita, Japan.

#### Weight of Tissues

K-type CpG ODNs (10 or 50 µg CpG ODNs/mouse) or CA-CpG (10 µg CpG ODNs/mouse) was administered to C57BL/6J mice subcutaneously at the base of the tail at days 0, 2, and 4. Twentyfour hours after the final administration, the entire spleen, the entire liver, and draining lymph nodes were isolated and weighed.

#### Statistical Analyses

Statistical analyses were performed with GraphPad Prism (GraphPad Software, San Diego, CA, USA). All data are presented as mean with SD or SEM. Significant differences between control groups and experimental groups were determined by means of Dunnett's test, Tukey's test, or Student's *t*-test. A *p* value of <0.05 was considered to indicate statistical significance.

## RESULTS

## CA-CpG Had High CpG ODN Encapsulation Efficiency

We evaluated the usefulness of CA nanoparticles as CpG ODN delivery vehicles. We began by determining the amount of K-type CpG ODNs encapsulated in CA-CpG. We found that the amount of K-type CpG ODNs used to generate CA-CpG affected the efficiency with which the K-type CpG ODNs was encapsulated in CA-CpG (**Figure 1A**). When we used 50 µg K-type CpG ODNs, the encapsulation efficiency was high (58%). Therefore, we used CA-CpG prepared with 50 µg K-type CpG ODNs for subsequent experiments. Dynamic light scattering showed that both the empty CA nanoparticles and CA-CpG were uniform and monodispersed, with hydrodynamic diameters ranging from 30 to 40 nm, respectively (**Figure 1B**). In addition, we confirmed that CA-Alexa 488-CpG was also uniform and monodispersed, with hydrodynamic diameters ranging from 30 to 40 nm, as was the case for both the empty CA nanoparticles and CA-CpG (data not shown). Atomic force microscopy showed that the sizes of the CA nanoparticles and CA-CpG ranged from 13 to 25 nm (**Figure 1C**). These results suggest that we successfully generated nanosize CA-CpG with high encapsulation efficiency for K-type CpG ODNs.

#### CA-CpG Induced IFN-**α** Production and Enhanced the Production of Other Cytokines in Mouse DCs and Human PBMCs

To determine the immunostimulatory effects of CA-CpG on DCs, we treated mouse-derived DCs with CA-CpG or K-type CpG ODNs *in vitro* and measured cytokine levels in the supernatants by means of ELISA (**Figure 2A**). DCs stimulated with CA-CpG produced substantial amounts of not only IFN-γ and IL-12 p40 but also IFN-α, whereas K-type CpG ODNs did not induce cytokine production at the tested doses. In human PBMCs, higher levels of IFN-α and IFN-γ were produced in response to CA-CpG than in response to K-type CpG ODNs (**Figure 2B**). These data suggest that CA-CpG might be superior to K-type CpG ODNs as an adjuvant. In addition, they suggest that CA-CpG has some characteristics of D-type CpG ODNs, because IFN-α is known to be a D-type CpG ODN-specific cytokine.

#### contain K-type CpG ODNs.

## Pattern of CA-CpG Uptake by DCs Differed From That of K-Type CpG ODNs *In Vitro*

To elucidate the mechanism of the immunostimulatory effects of CA-CpG, we compared the cellular internalization of CA-CpG and K-type CpG ODNs in mouse DCs by using Alexa 488-labeled K-type CpG ODNs. DCs were treated with CA-Alexa 488-CpG or with Alexa 488-labeled K-type CpG ODNs, and cells were stained with trypan blue solution to quench any Alexa 488-labeled K-type CpG ODNs bound to the cell surface, so that only internalized K-type CpG ODNs were detected during flow cytometry (**Figures 3A–C**). The amount of Alexa 488-labeled K-type CpG ODNs in DCs treated with CA-Alexa 488-CpG or Alexa 488-labeled K-type CpG ODNs increased over time, and at 90, 120, and 180 min after treatment the groups treated with Alexa 488-labeled K-type CpG ODNs had more cells that were positive for K-type CpG ODNs than did the groups treated with CA-Alexa 488-CpG (**Figures 3A,B**). For more precise analysis, the fractions that were positive for CpG ODNs were divided into two fractions, one with low fluorescence and one with high fluorescence; then we again compared the groups treated with CA-Alexa 488-CpG and the groups treated with Alexa 488-labeled K-type CpG ODNs (**Figures 3A,C**). In the low-fluorescence fraction, at 90, 120, and 180 min after treatment, the groups treated with Alexa 488-labeled K-type CpG ODNs had more cells that were positive for K-type CpG ODNs than did the CA-Alexa 488-CpG-treated groups. In contrast, in the high-fluorescence fraction at all time points, the CA-Alexa 488-CpG-treated groups had significantly more cells that were positive for K-type CpG ODNs than did the groups treated with Alexa 488-labeled K-type CpG ODNs. On the basis of these results, we speculated that certain subsets of DCs might have taken up CA-CpG more efficiently than others. To evaluate this possibility, we categorized DCs into the following subsets: B220<sup>+</sup> CD11c<sup>+</sup> plasmacytoid DCs, CD11b<sup>+</sup> CD11c<sup>+</sup> conventional DCs, and CD11b<sup>−</sup> CD11c<sup>+</sup> conventional DCs. Then we analyzed the CpG ODN-positive cells in each subset at both 37°C and 4°C, so that we could evaluate nonspecific binding of CA-Alexa 488- CpG or Alexa 488-labeled K-type CpG ODNs to the cell surface (**Figure 3D**). There were fewer CpG ODN-positive cells at 4°C

were then measured by ELISA. (A) In DCs, CA-CpG induced higher IFN-α, IFN-γ, and IL-12 p40 production than did K-type CpG ODNs, *n* = 4 per group. (B) In PBMCs, CA-CpG induced higher levels of pan-IFN-α and IFN-γ production than did K-type or D-type CpG ODNs, *n* = 5 per group. These data suggest that CA-CpG has some characteristics of D-type CpG ODNs. Data are given as mean ± SD. \**p* < 0.05, \*\**p* < 0.01 vs. K-type CpG ODN-treated group as indicated by Student's *t*-test (A) or Dunnett's test (B).

than at 37°C, indicating that the binding of CA-CpG or K-type CpG ODNs on the cell surface could be ignored in this assay. At 37°C, in all the subsets of DCs, the number of CpG ODNpositive cells in the CA-CpG-treated groups was lower than that in the K-type CpG ODN-treated groups (**Figure 3D**), indicating that CA-CpG was taken up into all the subsets of DCs equally, as was the case for the K-type CpG ODNs.

## CA-CpG Enhanced Cytokine Production *In Vivo*

Generally, the delivery of an adjuvant to lymph nodes, and specifically to DCs in lymph nodes, is important for the efficient induction of adaptive immunity. However, the efficiency with which CA-CpG delivers CpG ODNs to DCs in lymph nodes is unknown. Therefore, to assess the distribution and uptake of

Figure 3 | Cellular uptake of carbonate apatite (CA)-cytosine-phosphate-guanine (CpG). (A–C) Comparison of cellular uptake patterns of CA-CpG and K-type CpG oligodeoxynucleotides (ODNs). Mouse-derived dendritic cells (DCs) were treated with CA-Alexa 488-CpG or Alexa 488-labeled K-type CpG ODNs. At various time points after treatment, K-type CpG ODNs internalized in DCs were measured by flow cytometry after trypan blue quenching. (A) Representative dot plots show Alexa 488-labeled CpG ODNs in DCs vs. forward scatter (FSC-A). (B) Quantification of fluorescence revealed that uptake of CA-CpG into DCs was lower than uptake of K-type CpG ODNs, *n* = 5 per group. (C) Separation of the dot plots for the CpG ODN-positive fraction in panel (A) into graphs for high-fluorescence and low-fluorescence fractions revealed that the number of CpG ODNs taken up by each DC in the CA-CpG-treated groups was larger than that in the K-type CpG ODN-treated groups, *n* = 5 per group. (D) Subsets of DCs that took up CA-CpG. Mouse-derived DCs were treated with CA-Alexa 488-CpG or Alexa 488-labeled K-type CpG ODNs for 3 h at 4°C or 37°C. Uptake of CA-Alexa 488-CpG or Alexa 488-labeled K-type CpG ODNs into each subset of DCs was measured by flow cytometry without trypan blue quenching, *n* = 5 per group. Data are given as mean ± SD. \**p* < 0.05, \*\**p* < 0.01 vs. K-type CpG ODN-treated group as indicated by Student's *t*-test (B,C) or Dunnett's test (D). ND, not detected.

CA-CpG into lymph nodes, we measured the quantity of CpG ODN-positive cells in draining lymph nodes by flow cytometry after administration of CA-Alexa 488-CpG or Alexa 488-labeled K-type CpG ODNs to mice (**Figure 4A**). Whereas in the group treated with Alexa 488-labeled K-type CpG ODNs at 40 µg CpG ODNs/mouse (the positive control), large proportions of the lymph node cells were CpG ODN-positive, there were no significant differences between the proportions of CpG ODN-positive cells in the groups treated with CA-Alexa 488-CpG at 8 µg CpG ODNs/mouse and the groups treated with Alexa 488-labeled K-type CpG ODNs at 8 µg CpG ODNs/mouse at any time point. These results suggest that the CA nanoparticles did not increase the number of CpG ODNs taken up by DCs in lymph nodes.

Next, to determine the immunostimulatory effects of CA-CpG *in vivo*, we measured cytokine production in draining lymph nodes after CA-CpG administration (**Figure 4B**).

\*\**p* < 0.01 as indicated by Tukey's test (A) or vs. K-type CpG ODNs (8 µg CpG ODNs/mouse)-treated group as indicated by Dunnett's test (B). ND, not detected.

We found that production of the cytokines IL-6, IL-12 p40, and IFN-γ in draining lymph nodes was substantially higher after treatment with CA-CpG than after treatment with K-type CpG ODNs (**Figure 4B**). These data suggest that CA-CpG markedly enhanced innate immune activation responses, such as cytokine production, relative to K-type CpG ODNs both *in vitro* and *in vivo*.

## CA-CpG Enhanced Antigen-Specific Antibody Responses and CD8**<sup>+</sup>** CTL Responses *In Vivo*

To evaluate the *in vivo* effects of the use of CA-CpG as a vaccine adjuvant in mice, we assessed antibody responses and CD8<sup>+</sup> CTL responses following co-administration of CA-CpG or K-type CpG ODNs with OVA, a model protein antigen. Mice were immunized with OVA plus CA-CpG (10 µg CpG ODNs/mouse) or OVA plus K-type CpG ODNs (10 or 50 µg CpG ODNs/mouse) subcutaneously at the base of the tail. As a positive control, we used alum, which is widely used throughout the world as an adjuvant for human vaccines. Seven days after the final immunization, the levels of OVA-specific total IgG, IgG1, and IgG2c antibodies in plasma were analyzed by ELISA (**Figure 5A**). Levels of antigenspecific IgG subclasses reflect the subset of CD4<sup>+</sup> T cells that are induced by vaccination, with production of IgG1 and IgG2c corresponding to Th2- and Th1-type responses, respectively. Although mice immunized with OVA plus alum showed high levels of total IgG and IgG1 production, mice immunized with OVA plus CA-CpG or OVA plus K-type CpG ODNs produced high levels not only of total IgG and IgG1 but also of IgG2c. Mice immunized with OVA plus CA-CpG (10 µg CpG ODNs/mouse) produced significantly higher levels of OVA-specific total IgG, IgG1, and IgG2c than did mice immunized with OVA plus K-type CpG ODNs (10 µg CpG ODNs/mouse). Antibody production in groups treated with OVA plus CA-CpG (10 µg CpG ODNs/ mouse) was comparable to that in groups treated with OVA plus K-type CpG ODNs (50 µg CpG ODNs/mouse).

Next, to evaluate the adjuvant effects of CA-CpG when administered by a different route, we immunized mice with OVA plus CA-CpG or OVA plus K-type CpG ODNs into the ear pinna and then measured OVA-specific antibody responses (**Figure 5B**). As was the case for immunization subcutaneously at the base of the

7 days after the final immunization, *n* = 5 per group. (C,D) CD8+ cytotoxic T lymphocyte (CTL) responses. OVA257–264-specific CD8+ CTL responses were monitored by means of a tetramer assay in untreated mice and in mice immunized with OVA alone, with OVA plus K-type CpG ODNs, or with CA-CpG (C), *n* = 5 per group. Lymph node cells were stimulated with OVA257–264 peptide, and the concentration of IFN-γ in supernatants was analyzed by ELISA (D), *n* = 5 per group. The data in this figure suggest that a simple mixture containing CA-CpG and an antigen-induced antigen-specific antibody responses and CD8+ CTL responses that were superior to those induced by K-type CpG ODNs and alum, thus demonstrating the potential of CA-CpG as a vaccine adjuvant. Data are given as mean ± SEM. \**p* < 0.05, \*\**p* < 0.01 vs. K-type CpG ODNs (10 or 8 µg CpG ODNs/mouse)-treated group as indicated by Dunnett's test.

tail, the level of OVA-specific total IgG induced by immunization with OVA plus CA-CpG (8 µg CpG ODNs/mouse) was significantly higher than the level induced by immunization with OVA plus K-type CpG ODNs (8 µg CpG ODNs/mouse) and was comparable to the level induced by administration of OVA plus K-type CpG ODNs (40 µg CpG ODNs/mouse).

Clearance of viruses is known to require not only antibody responses but also strong CD8<sup>+</sup> CTL responses, which are characterized by IFN-γ production by CD8+ T cells. To investigate the ability of CA-CpG to induce OVA-specific CD8<sup>+</sup> CTL responses, we vaccinated mice with OVA plus CA-CpG and then examined the frequency of H-2Kb /OVA257–264 tetramer<sup>+</sup> CD8<sup>+</sup> T cells (**Figure 5C**) and measured the production of H-2Kb / OVA257–264-specific IFN-γ (**Figure 5D**). The frequency of H-2Kb / OVA257–264 tetramer<sup>+</sup> CD8<sup>+</sup> T cells induced by OVA plus CA-CpG was significantly higher than the frequency induced by OVA plus K-type CpG ODNs (**Figure 5C**). Furthermore, the production of OVA257–264-specific IFN-γ induced by OVA plus CA-CpG was significantly higher than that induced by immunization with OVA plus K-type CpG ODNs (**Figure 5D**). Taken together, these results suggest that, *in vivo*, CA-CpG stimulated antigen-specific antibody responses and CD8<sup>+</sup> CTL responses more effectively than did K-type CpG ODNs.

### CA-CpG Showed Strong Adjuvant Effects and Preventive Effects in Murine Influenza Models

We examined the vaccine adjuvant effects of CA-CpG by using clinically relevant influenza vaccination models in mice. Mice were immunized with A/California/7/2009 (H1N1) SV plus CA-CpG (10 µg CpG ODNs/mouse), K-type CpG ODNs (10 or 50 µg/mouse), CA nanoparticles, or Alum. The levels of SV-specific total IgG, IgG1, and IgG2c antibodies in plasma were analyzed by ELISA after the final immunization (**Figures 6A,B**). Mice immunized with SV plus CA-CpG (10 µg CpG ODNs/ mouse) showed significantly higher levels of SV-specific total IgG, IgG1, and IgG2c than did mice immunized with OVA plus K-type CpG ODNs (10 and 50 µg CpG ODNs/mouse). The levels of SV-specific total IgG and IgG2c in mice treated with SV plus CA-CpG were significantly higher than those in mice treated with SV plus alum, whereas there was no difference in SV-specific IgG1 levels between these two groups of mice. The levels of total IgG and IgG1 in mice treated with SV plus CA nanoparticles were higher than the levels in mice treated with SV alone but were significantly lower than the levels in CA-CpGtreated mice. Next, we examined the neutralizing potential of these antibodies by using the HI assay against influenza virus A/California/7/2009 (**Figure 6C**), because this assay is widely used to evaluate anti-influenza virus neutralizing and protective antibodies. The HI titers showed the same trends as the levels of SV-specific total IgG, and mice immunized with SV plus CA-CpG had significantly higher HI titers than did mice immunized with SV plus K-type CpG ODNs (10 µg CpG ODNs/mouse). After the final immunization, we challenged the immunized mice with mouse-adapted influenza virus A/PR/8/34 (H1N1), which is a heterologous influenza virus from the one used as a vaccine antigen. Weights and survival rates of the infected mice were observed every day for 2 weeks (**Figure 6D**). Much greater pathogenesis was observed in the mice immunized without an adjuvant or with alum than in the mice co-immunized with CA-CpG or K-type CpG ODNs (10 or 50 µg CpG ODNs/mouse), as indicated by survival rate and weight loss. All mice died within 10 days after viral infection in the groups that received SV alone or SV plus alum, and 80% of the mice immunized with SV plus CA nanoparticles died within 7 days. Although only 40% of the mice immunized with SV plus K-type CpG ODNs (10 µg CpG ODNs/mouse) survived 14 days after challenge, 80% of mice immunized with SV plus CA-CpG (10 µg CpG ODNs/mouse) or K-type CpG ODNs (50 µg/mouse) survived, and they regained body weight more rapidly. These results indicate that CA-CpG is a potent vaccine adjuvant that protects against viral infection better than K-type CpG ODNs or alum.

## Adjuvant Effects of CA-CpG Depended on Both IFN-**α** and IL-12

Activation of CD8<sup>+</sup> CTL responses requires T-cell activation by DCs via costimulatory signals and cytokine-mediated signals. For example, IFN-α and IL-12 work as critical survival signals that expand and differentiate effector CD8<sup>+</sup> T cells (29–31). Therefore, to elucidate the involvement of IFN-α and IL-12 in antibody responses and CD8<sup>+</sup> CTL responses induced by CA-CpG, we treated mice deficient in both *Ifnar2* and *Il-12 p40* with OVA plus CA-CpG (**Figure 7**). The levels of OVA-specific total IgG, IgG1, and IgG2c in these double-deficient mice immunized with OVA plus CA-CpG tended to be lower than the levels in wild-type mice (**Figure 7A**). In addition, the frequency of H-2Kb /OVA257–264 tetramer<sup>+</sup> CD8<sup>+</sup> T cells in the double-deficient mice tended to be lower than that in the wild-type mice (**Figure 7B**). These results suggest that CA-CpG enhanced both Th1-type humoral and cellular immune responses *via* the IFN-α pathway, the IL-12 pathway, or both.

## Adverse Effects of CpG ODNs Were Not Enhanced by CA-CpG

Cytosine-phosphate-guanine ODNs are known to cause splenomegaly and to disrupt splenic microarchitecture in mice after repeated administration (32). To assess the adverse effects of CA-CpG, we treated mice three times every other day with CA-CpG or K-type CpG ODNs and then measured tissue weights (**Figure 8**). No significant differences in liver weight were observed between the groups. In contrast, spleen weights and draining lymph node weights in mice treated with K-type CpG ODNs at 50 µg CpG ODNs/mouse were higher than those in mice treated with CA-CpG (10 µg CpG ODNs/mouse), although the difference was statistically significant only for spleen weights. The effects of CA-CpG (10 µg CpG ODNs/mouse) were almost the same as those of K-type CpG ODNs at 10 µg CpG ODNs/ mouse. These results suggest that CA-CpG improved the efficacy of CpG ODNs without enhancing their adverse effects.

## DISCUSSION

Because systemic dissemination of CpG ODNs away from the site of injection occurs rapidly after administration, high doses of CpG ODNs are required for induction of vaccine effects, thus increasing the risk of adverse events (33). Therefore, the development of easy to fabricate, high encapsulation capacity CpG ODN delivery vehicles that enhance the adjuvant activity

Figure 6 | Vaccine adjuvant effect of carbonate apatite (CA)-cytosine-phosphate-guanine (CpG) against influenza virus. Mice were immunized subcutaneously at the base of the tail with SV alone or with SV plus CA nanoparticles that did not contain CpG oligodeoxynucleotides (ODNs), with SV plus K-type CpG ODNs (10 or 50 µg CpG/mouse), with SV plus CA-CpG (10 µg CpG/mouse), or with SV plus alum. (A,B) SV-specific antibody responses. Levels of SV-specific total IgG (A), IgG1 (B), and IgG2c (B) in plasma were evaluated by ELISA 7 days after the final immunization. (C) Hemagglutination inhibition (HI) assay. HI titers in plasma samples were evaluated 7 days after the final immunization. (D) Preventive effects against influenza virus. Fourteen days after the final immunization, mice were challenged with a 10-LD50 dose of influenza virus A/PR/8 (H1N1). Body weights and survival rates were monitored for the next 14 days. The data in this figure suggest that CA-CpG acted as an influenza vaccine adjuvant in mice. *n* = 5 per group. Data are given as mean ± SEM. \*\**p* < 0.01 vs. K-type CpG ODNs (10 µg CpG ODNs/mouse)-treated group as indicated by Dunnett's test.

Figure 7 | Adjuvant activities of carbonate apatite (CA)-cytosine-phosphate-guanine (CpG) in *IFN-α/β receptor 2 (Ifnar2)* and *Il-12 p40* double-deficient mice (DKO mice). Wild-type (WT) mice or DKO mice were immunized subcutaneously at the base of the tail with ovalbumin (OVA) plus CA-CpG (10 µg CpG/mouse) at days 0 and 14. (A) OVA-specific antibody responses. Levels of OVA-specific total IgG, IgG1, and IgG2c in plasma were evaluated by ELISA 7 days after the final immunization, *n* = 5 per WT group and *n* = 4 per DKO group. (B) CD8+ cytotoxic T lymphocyte (CTL) responses. OVA257–264-specific CD8+ CTL responses were monitored in mice by tetramer assay 7 days after the final immunization, *n* = 5 per WT group and *n* = 4 per DKO group.

indicated by Tukey's test.

of CpG ODNs and reduce their adverse effects is necessary. In addition, after cellular uptake, the delivery vehicle must degrade rapidly *via* an endosomal or lysosomal processing pathway so that the CpG ODNs can bind to endosomal TLR9. Several types of particle-based delivery vehicles have been reported, including cationized gelatin nanoparticles (34), calcium phosphate nanoparticles (35), poly(lactic-*co*-glycolic acid) nanoparticles (36, 37), Ficoll nanoparticles (38), poly(propylene sulfide) nanoparticles (39), and liposomes (40, 41). However, there have been no reports

suggesting that these particles can enhance the adjuvant activity of K-type CpG ODNs by giving them the ability to induce IFN-α production *in vivo*. We showed here that CA nanoparticles can serve as K-type CpG ODN delivery vehicles and that they enhance the adjuvant activity of K-type CpG ODNs by inducing IFN-α production both *in vitro* and *in vivo*.

Carbonate apatite nanoparticles have been shown to efficiently encapsulate siRNA (26). In this study, we showed that CA nanoparticles can also encapsulate K-type CpG ODNs, with an encapsulation efficiency of about 60% (**Figure 1A**). Researchers have tried to enhance the adjuvant activity of CpG ODNs by encapsulating them in microparticles and nanoparticles, but the encapsulation efficiency has been very low (42–44). Because CpG ODNs are negatively charged, cationic materials such as protamine and chitosan have been used to bind CpG ODNs and facilitate their incorporation into particles. For example, Fisher et al. showed that CpG ODNs could be encapsulated with approximately 90% efficiency into protamine-coated poly(lactic*co*-glycolic acid) particles, whereas the encapsulation efficiency was only 8% for bare particles (43). Peine et al. (44) and Zhang et al. (42) showed that CpG ODNs could be encapsulated with 30% efficiency into acetalated dextran particles and poly(lactic*co*-glycolic acid) particles, respectively, without the use of cationic materials, by optimization of the conditions used to generate the particles. These results suggest that encapsulation of CpG ODNs into CA nanoparticles with relatively high efficiency can be achieved without the use of cationic materials.

Here, we showed that, compared with K-type CpG ODNs, CA-CpG induced higher levels of cytokine production in mouse DCs and human PBMCs (**Figure 2**). Notably, IFN-α production induced by CA-CpG was not only higher than that induced by K-type CpG ODNs but also higher than that induced by D-type CpG ODNs. Because IFN-α is a D-type CpG ODN-specific cytokine, these results suggest that CA-CpG-encapsulated K-type CpG ODNs gained characteristics of D-type CpG ODNs without losing their K-type CpG ODN characteristics. Many studies have shown that the difference between the cytokine production patterns induced by K-type and D-type CpG ODNs might be due to differences in their subcellular distribution in DCs (45, 46). It has been suggested that K-type CpG ODNs localize to late endosomes after cellular internalization and activate the IFN regulatory factor 5 pathway *via* TLR9, whereas D-type CpG ODNs localize to early endosomes and activate the IFN regulatory factor 7 pathway *via* TLR9 (47). In fact, a change in the localization of K-type CpG ODNs to early endosomes by the use of cationic liposomes has been shown to induce IFN-α production (47). In contrast, a new model was recently proposed in which adaptor protein 3-mediated TLR9 trafficking to lysosome-related organelles is essential for the induction of IFN-α production (48); this model suggests that the mechanisms of IFN-α production are more complicated than previously thought. Therefore, we cannot determine the precise relationship between the cellular localization of CA-CpG and its potential to induce IFN-α production. However, several studies have demonstrated that the particle size of CA-CpG affects its intracellular trafficking and localization. For example, small particles (<200 nm) are rapidly transported to late lysosomes, which have an acidic environment; whereas large particles (>500 nm) are localized in nearly neutral environments such as early endosomes and phagosomes (49, 50). These size-related differences may be the result of particle size-dependent differences in cellular internalization pathways (51). Therefore, we speculate that CA-CpG may localize in lysosome-related organelles, not in early endosomes, and may release CpG ODNs, which then induce IFN-α production. However, further study is required to clarify the exact cellular localization of CA-CpG.

In this study, we showed that, *in vitro*, CA-CpG was taken up by fewer DCs than were K-type CpG ODNs (**Figures 3A,B**). However, the number of CpG ODNs taken up by each DC was larger than the number of K-type CpG ODNs taken up by each DC, probably because CpG ODNs were condensed into a small space in the CA-CpG (**Figures 3A,C**). The greater uptake of CpG ODNs by cells treated with CA-CpG than by cells treated with K-type CpG ODNs might result in higher cytokine production in the former than in the latter.

With regard to the biodistribution of particles *in vivo*, several studies have suggested that small nanoparticles (<100 nm or <200 nm, depending on the study) can be trafficked to draining lymph nodes preferentially and can be taken up by DCs in lymph nodes, whereas larger particles (>100 nm or >200 nm, depending on the study) tend to be deposited at the injection site. These results suggest that nanoparticles <100 nm might be optimal for enhancing the efficacy of antigen and adjuvant delivery vehicles (52–54). We showed that the proportion of DCs that took up CpG ODNs in draining lymph nodes in CA-CpG-treated mice (8 µg CpG ODNs/mouse) was almost the same as the proportion in K-type CpG ODN-treated mice (8 µg CpG ODNs/mouse) (**Figure 4A**). However, cytokine production by lymph node cells was higher after administration of CA-CpG (8 µg CpG ODNs/mouse) than after administration of K-type CpG ODNs (8 µg CpG ODNs/mouse) (**Figure 4B**). These data suggest that, *in vivo* as well as *in vitro*, CA-CpG strongly induced cytokine production even though the number of CpG ODN-positive DCs was lower. It is not clear whether CpG ODN-positive individual DCs in the draining lymph nodes of CA-CpG-treated mice took up larger amounts of CpG ODNs than did individual DCs in the draining lymph nodes of K-type CpG ODN-treated mice. Further investigation of the biodistribution of CA-CpG *in vivo* is necessary.

The antigen (OVA and SV)-specific antibody responses (**Figures 5A,B** and **6**) and CD8<sup>+</sup> CTL responses (**Figures 5C,D**) induced by CA-CpG were superior to those induced by K-type CpG ODNs. The strength of the antigen-specific immune responses induced by CA-CpG was significantly higher than the strength of the responses induced by the same dose of K-type CpG ODNs and was almost the same as the strength of the responses induced by a dose of K-type CpG ODNs that was five times high. However, unlike K-type CpG ODNs (50 µg CpG ODNs/ mouse), CA-CpG (10 µg CpG ODNs/mouse) did not induce splenomegaly, which suggests that CA-CpG may be a superior CpG ODN delivery vehicle in that it has better vaccine adjuvant effects than K-type CpG ODNs, without enhanced adverse effects (**Figure 8**). Many nanomaterials have the potential to enhance adaptive immunity, that is, they have adjuvant effects (55, 56). In this study, we found that the antibody responses in mice treated with SV plus CA nanoparticles without K-type CpG ODNs were slightly higher than the responses in mice treated with SV alone (**Figures 6A,B**). In contrast, the responses in mice treated with SV plus CA nanoparticles without K-type CpG ODNs were significantly lower than the responses in CA-CpG-treated mice (**Figures 6A,B**). These data suggest that the adjuvant activity of CA nanoparticles alone was low. In addition, we showed that CA-CpG performed better against influenza virus infection than did the same dose of K-type CpG ODNs or alum (**Figure 6D**). Following influenza infection, not only neutralizing antibodies but also CD8<sup>+</sup> CTL responses play crucial roles in eliminating the virus and preventing viral persistence (57, 58). Furthermore, vaccines that induce CD8<sup>+</sup> CTL responses are expected to show efficacy against influenza virus, because CD8<sup>+</sup> CTL responses can provide cross-subtype or heterosubtypic protection (57, 58). Therefore, enhancement of CD8<sup>+</sup> CTL responses by CA-CpG might lead to the development of a vaccine with strong protective effects against not only homologous but also heterologous influenza viruses. Taken together, these data suggest that CA-CpG can be used as a potent adjuvant for protein-based vaccines such as influenza SVs.

We showed that the enhancement of antibody responses and CD8<sup>+</sup> CTL responses induced by CA-CpG tended to be negated in *Ifnar2-* and *Il-12 p40* double-deficient mice. We showed that CA-CpG induced the production of IFN-α and IL-12 in DCs more strongly than did K-type CpG ODNs *in vitro* (**Figure 2**). Previously, Kobiyama et al. showed that IFN-α and IL-12 are produced from plasmacytoid DCs and CD8α+ DCs, respectively, *in vitro* after stimulation with a complex consisting of CpG ODNs and β-glucan (19, 27). Furthermore, these investigators also showed that both IFN-α and IL-12 are indispensable for the induction of Th1-type antibody responses and CD8<sup>+</sup> CTL responses induced by the CpG ODN complex (19, 27). These findings suggest that IFN-α produced by plasmacytoid DCs or IL-12 produced by CD8α+ DCs, or both, play important roles in the strong immune responses induced by CA-CpG.

In summary, we have shown that CpG ODNs encapsulated in CA nanoparticles as delivery vehicles induce enhanced cytokine production—especially IFN-α production—relative to that induced by CpG ODNs alone. Furthermore, we have shown CA-CpG is a potent adjuvant for protein-based subunit vaccines. We believe that the use of CA-CpG can improve the efficacy of influenza vaccines, such as vaccines for pandemic influenza and influenza vaccines for use among the elderly. In addition, CA-CpG may be superior to CpG ODNs as an adjuvant for vaccines against cancer, because CA-CpG can induce not

#### REFERENCES


only antigen-specific antibody responses but also CD8<sup>+</sup> CTL responses. Thus, we believe that the novel pH-responsive CA nanoparticles reported herein have the potential to open up new avenues of research on vaccine adjuvant delivery vehicles. In addition, our results can be expected to lead to an improved understanding of the mechanisms of adjuvant effects and to the development novel particles or adjuvant delivery vehicles designed to improve vaccine efficacy.

#### ETHICS STATEMENT

All animal experiments were performed in accordance with the institutional guidelines of Osaka University for the ethical treatment of animals. All experiments using human peripheral blood mononuclear cells were approved by the Institutional Review Board of the Research Institute for Microbial Diseases, Osaka University.

### AUTHOR CONTRIBUTIONS

HT, TA, HY, and YY designed the experiments and interpreted the results. HT, KM, YYamamoto, and YK performed the experiments. HT, KM, YYamamoto, YK, and YY collected and analyzed the data. XW, EK, and KI provided technical support and conceptual advice. HT and YY wrote the manuscript. YY supervised the study.

#### ACKNOWLEDGMENTS

We thank Dr. Yasuyuki Gomi (The Research Foundation for Microbial Diseases of Osaka University) and Dr. Toshihiro Akaike (Foundation for Advancement of International Science).

## FUNDING

This study was supported by a grant from the Japan Society for the Promotion of Science (no. JP17H04009 to YYoshioka).


immunity and protects against influenza virus infection. *J Immunol* (2013) 190(12):6221–9. doi:10.4049/jimmunol.1202654


**Conflict of Interest Statement:** The following authors have conflict of interests to declare. TA, YYamamoto, and YYoshioka are employed by The Research Foundation for Microbial Diseases of Osaka University. HY and YYoshioka have filed a patent application related to the content of this manuscript. The other authors have no conflicts of interest to declare.

*Copyright © 2018 Takahashi, Misato, Aoshi, Yamamoto, Kubota, Wu, Kuroda, Ishii, Yamamoto and Yoshioka. 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.*

# Polymeric nanocapsules for Vaccine Delivery: influence of the Polymeric shell on the interaction With the immune system

*Mercedes Peleteiro1†, Elena Presas2,3†, Jose Vicente González-Aramundiz3,4, Beatriz Sánchez-Correa1,5, Rosana Simón-Vázquez1 , Noemi Csaba2,3, María J. Alonso2,3\* and África González-Fernández1 \**

*<sup>1</sup> Inmunología, Centro de Investigaciones Biomédicas (CINBIO) (Centro Singular de Investigación de Galicia), Instituto de Investigación Sanitaria Galicia Sur, Universidade de Vigo, Vigo, Spain, 2Department of Pharmacy and Pharmaceutical Technology, School of Pharmacy, University of Santiago de Compostela, Santiago de Compostela, Spain, 3Center for Research in Molecular Medicine and Chronic Diseases (CIMUS), University of Santiago de Compostela, Santiago de Compostela, Spain, 4 Departamento de Farmacia, Facultad de Química, Pontificia Universidad Católica de Chile, Santiago, Chile, 5 Immunology Unit, University of Extremadura, Cáceres, Spain*

#### *Edited by:*

*Urszula Krzych, Walter Reed Army Institute of Research, United States*

#### *Reviewed by:*

*Thorsten Demberg, Immatics Biotechnologies, Germany Serge Muyldermans, Vrije Universiteit Brussel, Belgium*

#### *\*Correspondence:*

*María J. Alonso mariaj.alonso@usc.es; África González-Fernández africa@uvigo.es*

*† These authors have contributed equally to this work.*

#### *Specialty section:*

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

*Received: 18 January 2018 Accepted: 29 March 2018 Published: 19 April 2018*

#### *Citation:*

*Peleteiro M, Presas E, González-Aramundiz JV, Sánchez-Correa B, Simón-Vázquez R, Csaba N, Alonso MJ and González-Fernández Á (2018) Polymeric Nanocapsules for Vaccine Delivery: Influence of the Polymeric Shell on the Interaction With the Immune System. Front. Immunol. 9:791. doi: 10.3389/fimmu.2018.00791*

The use of biomaterials and nanosystems in antigen delivery has played a major role in the development of novel vaccine formulations in the last few decades. In an effort to gain a deeper understanding of the interactions between these systems and immunocompetent cells, we describe here a systematic *in vitro* and *in vivo* study on three types of polymeric nanocapsules (NCs). These carriers, which contained protamine (PR), polyarginine (PARG), or chitosan (CS) in the external shell, and their corresponding nanoemulsion were prepared, and their main physicochemical properties were characterized. The particles had a mean particle size in the range 250–450 nm and a positive zeta potential (~30–40 mV). The interaction of the nanosystems with different components of the immune system were investigated by measuring cellular uptake, reactive oxygen species production, activation of the complement cascade, cytokine secretion profile, and MAP kinases/nuclear factor κB activation. The results of these *in vitro* cell experiments showed that the NC formulations that included the arginine-rich polymers (PR and PARG) showed a superior ability to trigger different immune processes. Considering this finding, protamine and polyarginine nanocapsules (PR and PARG NCs) were selected to assess the association of the recombinant hepatitis B surface antigen (rHBsAg) as a model antigen to evaluate their ability to produce a protective immune response in mice. In this case, the results showed that PR NCs elicited higher IgG levels than PARG NCs and that this IgG response was a combination of anti-rHBsAg IgG1/IgG2a. This work highlights the potential of PR NCs for antigen delivery as an alternative to other positively charged nanocarriers.

Keywords: nanocapsules, vaccination, antigen, adjuvant, hepatitis B, rHBsAg, nanovaccines

## INTRODUCTION

Vaccination is one of the most cost-effective health interventions for the prophylaxis of infectious diseases. Conventional vaccines use inactivated or attenuated pathogens—or even subunits of these organisms, such as toxoids and carbohydrates—as antigens (1). Unfortunately, the immune protection provided by these subunit vaccines, even with classical adjuvants, i.e., alum, is often insufficient.

**159**

Moreover, these vaccines suffer from additional problems, such as limited storage stability and the need for multiple-dose injection schemes to achieve effective protection. Thus, vaccine development has been required to move toward a new generation of vaccines that could provide better and more sustainable protective immune responses (2).

Interest in the use of innovative antigens such as recombinant proteins, peptides, or nucleic acids has increased in the last few decades (3–5). In spite of the many advantages of these new vaccine prototypes, including an enhanced immunogenicity/ risk ratio profile and easier manufacture, the development of improved adjuvants that may enhance the potency of these vaccines has become essential. Several inmunostimulant molecules (i.e., cytokines, toll-like receptor ligands, toxins, or saponins) and antigen delivery systems (i.e., virosomes, emulsions, or nanoparticles) are among the potential adjuvants studied in this scope. In particular, the use of nanosystems has attracted significant attention, especially for vaccines requiring an improvement in cell-mediated immune responses (6).

Polymer- and lipid-based nanoparticles have been the most widely explored nanosystems in vaccination. It has already been reported that nanostructures are able to enhance both B and T cell responses due to their similarity with viruses in terms of size and surface properties (7). In addition, due to their prolonged presentation to the immune system, these nanostructures can induce long-lasting immune responses, thereby offering the possibility of decreasing the number of doses required to achieve protective antibody levels. Moreover, nanoparticle formulations have shown the ability to avoid overactivation of the immune response and, simultaneously, to trigger the production of pro-inflammatory cytokines (8).

Our group has developed a number of antigen delivery technologies and validated them using rHBsAg as a model antigen. Namely, we have produced various chitosan-based nanocarriers, including chitosan-coated PLGA nanoparticles (9), CS nanoparticles (10), and CS nanocapsules (NCs) (11). More recently, we also developed arginine-rich based (PARG and PR) NCs (12) and nanoparticles including different anionic polymers (dextran sulfate and alginate) (13). Overall, the *in vivo* administration of these nanocarriers through nasal or parenteral routes has led to significant IgG responses in mice (11, 14). Interestingly, the differences obtained in the elicited immune responses highlighted the importance of the composition of the nanosystem, especially concerning the type of biopolymer exposed on the nanoparticle surface.

The aim of the work described here was to gain an understanding of the role of the polymeric shell of NCs in their interaction with immune cells. For this purpose, three types of NCs with different coating polymers (PARG, PR, and CS), and the corresponding nanoemulsion (NE), were compared in a systematic study. This included the *in vitro* evaluation of cellular uptake, the reactive oxygen species (ROS) production, the activation of the complement cascade, the cytokine secretion profile, the MAP kinases/nuclear factor κB (NFκB) activation, and gene expression. The influence of size, composition of the outer shell layer, and superficial charge in these processes were analyzed, and the two most promising prototypes were selected to perform *in vivo* studies with the rHBsAg antigen. In addition to the systematic study, the four different prototypes were converted into freeze-dried products, to enhance the stability of the formulations under storage and avoid the need of the maintenance of the cold chain. The physicochemical characterization of the resulting products is disclosed in Supplementary Material.

### MATERIALS AND METHODS

#### Materials

Three polymers were used: protamine from Yuki Gosei Kogyo, Ltd. Company (Japan), chitosan hydrochloride salt (Protasan UP CL113 deacetylation degree of 75–90%, *M*w: 125 kDa) was purchased from NovaMatrix (Sandvika, Norway) and polyarginine (PARG) (*M*w 5–15 kDa) was obtained from Sigma-Aldrich (Barcelona, Spain). Miglyol® 812, a mixture of three different triglycerides of medium chain fatty acids, was obtained from Sasol GmbH (Witten, Germany). Recombinant hepatitis B surface antigen (rHBsAg) (*M*w 24 kDa) in an aqueous suspension [0.16 mg/mL in phosphate-buffered saline (PBS)] was kindly donated by Shantha Biotechnics Limited (Hyderabad, India). PEG-stearate (Simulsol® M52) was purchased from Invitrogen (UK). Cobra venom factor (CVF) was supplied by Quidel Corporation (USA), and the PVDF membranes were obtained from Bio-Rad (USA).

Aluminum hydroxide, sodium hydroxide, rhodamine 6G, Triton 100x, PBS, 5-bromo-4-chloro-3-indolyl phosphate (BCIP), glycerol, sodium cholate, 2ʹ,7-dichlorofluorescein diacetate (DCFH-DA), lipopolysaccharide (LPS), phorbol myristate acetate (PMA), and phytohemagglutinin (PHA) were all obtained from Sigma-Aldrich (USA).

Cell culture media (DMEM and RPMI 1640) and the antibiotics penicillin and streptomycin were supplied by Gibco (Life Technologies, Scotland). Heat inactivated fetal bovine serum (FBS) and Accutase® were purchased from PAA Laboratories (Austria), and Ficoll-Hypaque was obtained from GE Healthcare (Sweden).

#### Antibodies

#### *Enzyme-Linked Immunosorbent Assay (ELISA) for Recombinant Hepatitis B Antigen (rHBsAg)*

Enzyme-linked immunosorbent assay kit (Murex HBsAg Version 3) was obtained from Murex Biotech Ltd. (Dartford, UK), and the polyclonal chicken antibodies against HBsAg were purchased from Abcam (UK). Mouse and rabbit antibodies against HBsAg used in the ELISA test were purchased from Acris Antibodies GmbH (Hiddenhausen, Germany) and Biokit (Barcelona, Spain), respectively. Secondary mouse antibodies against IgG, IgG1, IgG2a conjugated with horseradish peroxidase (HRP) were acquired from Southern Biotech (USA).

*Western Blot and ELISA for Complement Factors (C3 and C5)* Polyclonal secondary antibodies used in the Western blot assay were purchased from Dako (Glostrup, Denmark), and the monoclonal antibody (mAb) against complement factor C3 was obtained from Abcam (UK). Human C5a ELISA Kit II BD OptEIA™ used for the detection of complement factor C5 was purchased from BD Biosciences Pharmingen (CA, USA).

#### *Flow Cytometry: Cell Markers and Cytokines*

Monoclonal antibodies for flow cytometry assays were anti-CD3-PECy5.5, anti-CD19-PECy5.5 (Southern Biotech, AL, USA), anti-HLA-DR-FITC (Beckman Coulter, USA), and anti-CD71-PE (NIT Zipper, Nanoimmunotech, Spain).

Cytokine detection was performed with the FlowCytomix kit Th1/Th2 from eBioscience (San Diego, CA, USA).

#### *Analysis of Kinase Routes by Western Blot*

Rabbit monoclonal antihuman p-extracellular signal-regulated kinase (ERK), p-p38, p-SAP/JNK, IKBα, and goat anti-rabbit IgG-HRP antibodies used for the Western blot analysis were supplied by Cell Signaling Tech. (Danvers, MA, USA), and the anti-GAPDH antibody was obtained from Sigma-Aldrich (Barcelona, Spain).

#### Nanoparticle Preparation

Nanocapsules were prepared using the solvent displacement technique previously described by our research group (15). Briefly, a mixture of 12 mg of PEG-st and 5 mg of sodium cholate was dissolved in 0.750 mL of ethanol, followed by the addition of 62.5 µL of Miglyol® 812 to the previous mixture. The volume of the organic phase was increased by adding 4.25 mL of acetone. This organic phase was added to an aqueous phase composed of 10 mL of a solution of 0.05% w/v of PR, PARG, or CS for the preparation of the PR and CS NCs or 10 mL of milliQ water in the case of the control, uncoated NE. In all cases, the nanostructures were spontaneously formed upon diffusion of the solvents. The nanostructures were isolated by ultracentrifugation of the samples at 61,690 *g* for 1 h at 15°C (Optoma TM L-90K Ultracentrifuge, Beckman Coulter, USA).

#### Physicochemical Characterization of the Nanostructures

The hydrodynamic diameter and polydispersity index (PDI) of the NCs and the corresponding NE were measured by photon correlation spectroscopy. Zeta potential was determined by laser-Doppler anemometry (Zetasizer®, NanoZS, Malvern Instruments, Malvern, UK). Samples were measured after diluting in milliQ water (970 µL of water: 30 μL of NCs) or in 1 mM KCl for size and zeta potential, respectively.

#### Fluorescent Labeling of the Nanocarriers

Nanocapsules were prepared as described in Section "Nanoparticle Preparation" by adding an aliquot of the chromophores to the oily phase. Rhodamine B, rhodamine 6G, and DiD (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine, 4-chlorobenzenesulfonate salt) were selected as fluorescent markers at two different concentrations (10 and 50 µg/mL dissolved in ethanol), evaluating both the influence of the nature of the chromophore and its concentration. The evaluation of the encapsulation efficiency of the chromophores was carried out indirectly by quantifying the non-encapsulated fluorophore remaining in the undernatants after the centrifugation step. Rhodamines were measured at an emission wavelength of 590 nm (LB 940 Multimode Reader Mithras, Berthold Technologies GmbH & Co KG, Germany); UV–VIS spectroscopy at λ= 646 nm (Du-BoLife Science UV/VIS Beckman Coulter) was used to quantify DiD. The three different fluorophores were incorporated into the NE, PR and CS NCs, assuming no differences in the release and the encapsulation efficiency between the PR and the PARG NCs. The evaluation of the loading efficiency and release profile of the different fluorescent dyes are disclosed in Supplementary Material (Supplementary Figure 1 and Supplementary Table 1).

### Association of Hepatitis B Surface Antigen (rHBsAg) With the NCs

PR and PARG NCs, prepared as described in Section "Nanoparticle preparation," were incubated with rHBsAg in equal volumes at a mass ratio of 4:1 (theoretical concentration of the cationic polymer adsorbed on the NC surface:rHBsAg) to achieve adsorption onto the polymeric corona. The process was performed under mild conditions (RT, 1 h), and the loaded NCs were subsequently isolated (30 min, 12,872 *g*, 15°C). The amount of rHBsAg associated to the NCs was indirectly quantified by measuring the concentration of antigen remaining in the undernatant after the ultracentrifugation step. An ELISA commercial kit was used to quantify the rHBsAg concentration in the samples, according to the manufacturer's instructions. The association efficiency for rHBsAg (A.E.%) was calculated indirectly as the difference between the concentration of free antigen detected in the supernatant and the total concentration in the initial suspension.

## Freeze Drying of the NCs

To obtain a lyophilized product, preliminary studies were carried out to determine the best conditions to preserve the original physicochemical properties of the prototypes upon reconstitution (Supplementary Figure 2). The conversion into a dry powder was performed using a freeze-drying process (Genesis SQ Freeze dryer, Virtis, USA) (11). The studies were conducted with aliquots of the four nanosystems in a range of concentrations (0.25–1% w/v) both with and without cryoprotectant (glucose, trehalose, and sucrose at 5 and 10% w/v). Based on the results obtained, and to standardize the process conditions for the four prototypes, a solution of each of the different nanosystems (0.75% w/v) was frozen overnight at −20°C in the presence of sucrose as cryoprotectant (10% w/v). Subsequently, the samples were transferred to the lyophilizer (Genesis SQ Freeze Drier, Virtis, USA) and submitted to an initial drying step for 24 h at −35°C and 2–10 mTorr followed by a second and a third drying step (24 h, 0°C and 16 h, 20°C, respectively). After the process, the freeze-dried cake had an overall good aspect and could be resuspended swiftly in ultrapure water.

## Activation of the Complement Cascade

The activation of the complement cascade upon contact with the different nanostructures was analyzed by two methods: degradation of the complement C3 factor by Western blot and quantification of the C5a levels by ELISA. A pool of human plasma from healthy donors was incubated with different concentrations of nanosystems (10, 100, and 1,000 µg/mL of the constituent polymer and the corresponding amount of NE) and veronal buffer (pH 7.4). An aliquot of 50 µL of each formulation and controls were incubated at 37°C for 1 h. CVF and PBS were used as positive and negative controls, respectively. After the incubation step, samples were centrifuged at 16,000 *g* for 30 min to separate the nanosystems from the other components.

#### Qualitative Determination by Western Blot

After the centrifugation step, supernatants containing the complement proteins were loaded onto a 10% SDS-PAGE gel and transferred to a PVDF membrane with the Transblot-Turbo Transfer System (Bio-Rad, Hercules, CA, USA). Membranes were blocked overnight at 4°C with TBST with 5% of non-fat dried milk. Membranes were then incubated with a mouse mAb against human C3 (1:1,000) (90 min, RT). A secondary incubation step was performed under the same conditions (90 min, RT) with secondary polyclonal goat anti-mouse IgG antibodies conjugated with alkaline phosphatase (1:2,000). Intensive washes were carried out between all steps. Membranes were revealed with BCIP. To quantify the degradation, the intensity of the lower band was normalized to the negative control, where some basal C3 degradation was observed.

#### Quantitative Determination by ELISA

An ELISA assay was performed using the Human C5a ELISA Kit II to confirm the activation of the complement cascade quantifying C5a factor levels. Briefly, standards and samples were added for 2 h at RT to the wells previously coated with an mAb against the C5a factor. After several washes, wells were incubated for 1 h with a mixture of biotinylated antihuman C5a antibody and streptavidin-HRP. Intensive washes were performed between all these steps. To detect the presence of the antibody-antigen complexes, TMB Substrate Reagent was added. After 30 min of incubation, the reaction was stopped, and the absorbance was measured at 450 nm.

## *In Vitro* Cell Studies

#### Cells

Raw 264.7 (mouse macrophage cell line), Jurkat (human leukemic T-cell line), Hmy (human B lymphoblast cell line), and HL60 (human promyelocytic leukemia cell line) were purchased from ATCC (American Type Culture Collection, Middlesex, UK). All lines were cultured in RPMI supplemented with 10% (v/v) of heated-inactivated FBS, 2 mM glutamine, and 100 U/mL of penicillin/streptomycin, at 37°C in a 5% CO2 atmosphere. Cells were split every other day to maintain 70–80% confluent cultures.

Human peripheral blood mononuclear cells (hPBMCs) were obtained from three healthy voluntary donors. To separate mononuclear cells, 15 mL of EDTA-anticoagulated blood was diluted with 15 mL of PBS and centrifuged (180 *g*, 30 min, 20°C) through a Ficoll-Hypaque gradient using a 7:3 ratio (diluted blood:ficoll). Mononuclear cells were collected at the interface between the ficoll and the plasma and washed twice by centrifugation (100 *g*, 5 min, 20°C) with complete medium.

#### Cell Viability Assay: xCELLigence**®** System

An xCELLigence® RTCA DP Instrument (Roche Diagnostics, Penzberg, Germany) was used according to the manufacturer's instructions to analyze cell viability. Raw 264.7 cells were cultured at a density of 1.5 × 104 cells/well with 200 µL of RPMI supplemented with 10% FBS until they reached the exponential phase (37°C and 5% CO2) (around 18 h). The different prototypes were added in a range of six different concentrations from 250 to 7.8 µg/mL. As negative controls, cell culture media and the blank nanosystems were added to the wells. The impedance was monitored at 15 min intervals for 72 h.

#### Cellular Uptake by Macrophages

To evaluate the internalization of the nanosystems by fluorescence microscopy, 1 × 105 of Raw 264.7 cells were seeded in an NUNC 96-Well Optical-Bottom Plates with Coverglass Base (Thermo Fisher Scientific, Langenselbold, Germany) with RPMI 10% FBS in the presence and absence of the fluorescent prototypes at 10 and 50 µg/mL of the constituent polymer and the corresponding amount of NE for 30 min. Three washes were performed to remove the excess nanosystems, and cells were observed with an inverted fluorescence microscope (IX50, Olympus Optical Co. GmbH, Hamburg, Germany).

A similar protocol was followed to perform the flow cytometry analysis. In this case, after incubation with the fluorescently labeled prototypes followed by three washes with PBS, cells were detached using Accutase® (10 min, 37°C and 5% CO2). Finally, cells were washed once with complete medium to inactivate Accutase®, and the suspension was analyzed by flow cytometry (Accuri Cytometers, Ann Arbor, MI, USA).

A kinetic and more detailed study of the internalization was carried out by Confocal Laser Scanning Microscopy (Leica SP5) using the High Content Screening Automation HCS A module and the LAS AF MATRIX software. Images were acquired every 5 min during 3 h.

#### Cytokine Profile Evaluation

Cytokine production was assessed by incubating 2 × 105 hPBMCs during 24 h in 96-well plates (37°C, 5% CO2) in the presence of the prototypes at two different concentrations: 10 and 100 µg/ mL. As negative and positive controls, cells were incubated with complete medium or with a combination of a solution of 1 µg/ mL of LPS and 10 µg/mL of PHA, respectively. After 24 h, the plate was centrifuged (100 *g*, 5 min, 4°C), and supernatants were collected and stored at −20°C before analysis. The cytokine levels (IL-12p70, INFγ, IL-2, IL-10, Il-8, IL-6, IL-4, IL-5, IL-1β, TNFα, and TNF β) were quantified using the FlowCytomix kit Th1/ Th2 according to the manufacturer's instructions. Briefly, 25 µL of antibody-coated microspheres was incubated with 25 µL of culture supernatants and 50 µL of biotin-conjugated secondary antibodies (2 h, RT) using a microplate shaker. After several washes, 50 µL of streptavidin conjugated to phycoerythrin and 100 µL of PBS-T were added to the preparation and incubated on a microplate shaker (1 h, RT). Finally, phycoerythrin-bound beads were studied by flow cytometry (FC500, Beckman Coulter, Miami, FL, USA), and data were analyzed using FlowCytomix Pro 3.0 Software (eBioscience, San Diego, CA, USA).

#### ROS Production

The production of intracellular ROS was detected by measuring the oxidation of 2′,7-dichlorofluorescein diacetate (DCFH-DA). This marker can be oxidized by ROS and converted to a fluorescent compound.

The human promyeloblast cell line HL60 was used to evaluate the production of ROS. 2.5 × 105 cells were cultured in 24-well plates in contact with 10 or 100 µg/mL of the different prototypes after 1 and 12 h. As negative and positive controls, complete medium and a solution of 10 µM of PMA were used, respectively. After this incubation step, cells were collected and centrifuged (100 *g*, 5 min) and resuspended in PBS containing 5 µM of DCFH-DA. Afterward, the samples were incubated at 37°C during 30 min. Finally, cells were washed twice with PBS and analyzed by flow cytometry (Accuri Cytometers, Ann Arbor, MI, USA). The median of the fluorescence intensity was normalized to the negative control.

#### Activation Markers

Changes in the pattern of expression of membrane markers were studied in hPBMCs (CD3, CD19, CD71, and HLA-DR). 2 × 105 cells were seeded in a 96-well plate and incubated with and without the nanosystems at two different concentrations (10 and 100 µg/mL of the constituent polymer and the corresponding amount of the NE) during 24 h. As a positive control, a solution of 10 µg/mL of PHA was used. Cells were labeled with the antibodies during 30 min at 4°C. In a final, the cells were washed twice with PBS and analyzed by flow cytometry (FC500, Beckman Coulter, Miami, FL, USA).

#### Routes of Activation. MAP Kinases and NF**κ**B

The study of the signaling pathway activation (ERK1/2, p38, SAPK/JNK and NFκB) was performed with the human tumoral cell lines Jurkat and Hmy. These two cell lines were incubated with 10 µg/mL of the constituent polymer of PR NCs and PARG NCs for 1 and 3 h.

To perform the Western blot analysis, cells were washed with PBS and then suspended in a lysis buffer (Tris–HCl 10 mM, pH 8, NaCl 150 mM, EDTA 2.5 mM and 1% NP-40) containing a protease and phosphatase inhibitor (Complete Mini and PhosphoStop from Roche Ltd., Basel, Switzerland). Cell lysates were centrifuged (16,000 *g*, 4°C, 5 min) using an Eppendorf 5415R centrifuge (Eppendorf AG, Hamburg, Germany) to remove cell residues.

Cell extracts were loaded onto a 10% SDS-PAGE gel and transferred to a PVDF membrane using the Transblot-Turbo Transfer System (Bio-Rad, Hercules, CA, USA). PVDF membranes were blocked with 5% of non-fat dried milk in TBST (1 h, RT). Membranes were washed and incubated overnight at 4°C with specific rabbit monoclonal antihuman p-ERK, p-, and p-SAP/JNK antibodies to determine the phosphorylation of ERK, p38, and SAP/JNK kinases (dilutions from 1:10,000 to 1:20,000 in TBST) and with anti IKBα (1:10,000), which is the NFκB inhibitor, to evaluate its activation indirectly. After intensive washes, membranes were incubated with the anti-GAPDH antibody as internal control (1:40,000) for 1 h at RT. Goat anti-rabbit IgG antibodies conjugated to HRP diluted 1:50,000 in TBST with 2.5% of skimmed milk were used as secondary antibodies.

Membranes were revealed with the Clarity Western ECL Substrate (Bio-Rad Laboratories Inc.), and the protein bands were analyzed and quantified using the ChemiDoc XRS imaging system (Bio-Rad Laboratories Inc.).

## *In Vivo* Studies

#### Animals

Female BALB/c mice (4–5 weeks old) were housed in filter-top cages with a 12 h light/12 h dark cycle with a constant temperature environment of 22°C. Food and water were provided *ad libitum*.

Groups of five animals were randomly distributed and immunized twice (0 and 28 days) with 60 µL of PR NCs or PARG NCs incorporating 10 µg of rHBsAg. The injection was performed intramuscularly (i.m.) in the hind leg of the mouse and rHBsAg adsorbed to aluminum hydroxide was used as a positive control. Aluminum hydroxide and rHBsAg solutions were incubated in a volumetric ratio of 3:1 (rHBsAg:alum) for 30 min at 4°C under moderate agitation. The suspension was centrifuged (10,000 *g*, 10 min, 4°C), and the pellet was resuspended in a suitable volume of isotonic saline solution. Blood samples were collected from the mouse maxillary vein at 6 and 10 weeks for the quantification of HBsAg-specific antibodies.

#### Quantification of rHBsAg-Specific IgG and IgG Subtypes by ELISA

96-Well plates were coated with rHBsAg diluted in carbonate buffer (pH 9.6) at a concentration of 5 µg/mL and incubated overnight at 4°C. The plates were then blockaded with PBS-BSA 1% for 1 h at 37°C to reduce the non-specific interactions. A mouse monoclonal IgG anti-HBsAg was used to quantify the levels of specific rHBsAg IgG: the μg/mL of the specific IgG were converted into international units using a solution of anti-HBsAg rabbit IgG with a known concentration into mIU/mL. The controls and a pool of serum samples from each group were serially diluted and incubated for 2 h at 37°C. Two secondary antibodies (goat anti-mouse and anti-rabbit IgG conjugated with HRP) were added and incubated with the samples for 1 h at 37°C. Bound antibodies were revealed with 2,2′-azino-bis(3 ethylbenzothiazoline-6-sulfonic acid) (ABTS), and the optical density was read at 405 nm.

To calculate the ratio IgG1/IgG2a, a pool of sera collected from the immunized mice were analyzed using the same ELISA protocol, but using the goat anti-mouse IgG1 and IgG2a (conjugated to HRP) as secondary antibodies.

#### Restimulation of Splenocytes *Ex Vivo* for Quantitative PCR (qPCR) Gene Expression Assays

Changes in gene expression were evaluated by qPCR using TaqMan® Gene Expression Assays for several genes and the TaqMan® Fast Advanced Master Mix from Life Technologies™. The qPCR was performed in a 7900HT Fast Real-Time PCR System (AB, Life Technology Co.).

Mice previously immunized with two intramuscular doses (0, 4 weeks) were sacrificed at the end of the study (11 weeks after the first immunization). To obtain cell suspensions, spleens were removed under sterile conditions, and cells were dissociated by gentle teasing in complete medium (DMEM with 10% FBS) and filtered with a cell strainer (Falcon, NJ, USA). The suspension was centrifuged (100 *g*, 5 min), and the pellet was resuspended in 7 mL of DMEM. The same protocol followed to obtain hPBMCs was carried out to eliminate spleen erythrocytes and granulocytes.

Mononuclear cells separated by gradient centrifugation at a concentration of 5 × 106 cells/mL were restimulated with 10 µg/ mL of rHBsAg for 12 or 24 h. RNA was extracted and purified using the ReliaPrep™ RNA Miniprep Systems kit (Promega), and genomic DNA was eliminated with the same kit. The cDNA was synthetized using the Maxima First Strand cDNA Synthesis kit (Thermo Fisher Scientific Inc.). To check the optimal concentration of cDNA per sample, a qPCR for GAPDH was previously performed at different dilutions.

Data from three mice per group were normalized to the internal control GAPDH, and ΔΔCt was calculated using as a baseline the control data from mice immunized with PBS 1× and restimulated *in vitro* with rHBsAg. These data were averaged, and relative quantification (RQ) was calculated.

#### Statistical Analysis

Results are presented as Mean ± SD. Statistical comparisons were made by the Mann–Whitney *U* test for *in vivo* experiments and *T* test for *in vitro* experiments. The accepted level of significance was a *p* value < 0.05.

#### Ethical Issues

Institutional ethics approval to work with human samples from healthy donors was obtained from the Ethics Committee for Clinical Research (Xunta de Galicia, Spain, 2013/272). All participants included in the study gave their written informed consent.

All protocols developed with mice were adapted from the guidelines of the Spanish regulations (Royal Decree 53/2013) regarding the use of animals in scientific research and under the approval of the Ethical Committee of the University of Vigo.

#### RESULTS AND DISCUSSION

#### Development and Characterization of Polymeric NCs

The method selected for the preparation of the nanostructures was the solvent displacement technique previously described by our research group (16). The procedure is based on the nanoprecipitation of the lipid components upon the diffusion of a polar solvent into an aqueous phase, where the cationic polymer interacts with the oily nanodroplet cores, thus forming a polymeric coating that surrounds the oily core. Polycationic NCs made of CS (11, 14), PARG (17) and PR (12) with different oil/surfactant ratios were previously developed by our research group. In this work, the type and the ratio of the lipidic constituents were adjusted to develop nanocarriers with the identical oily core composition but varying the polymeric surface coatings. As a result, three different NC formulations and the corresponding negatively charged NE were developed (**Table 1**). As expected, the three NC formulations had a positive zeta potential and their size varied depending on the nature of the polymer shell. PR and PARG NCs presented zeta potential values close to +30 mV, whereas the value for CS NCs was higher (+47 mV), probably due to the presence of a larger number of positive groups on their surface. With respect to the particle size, PR or PARG NCs had similar sizes, within the range 250–300 nm, and a low PDI (0.2 and 0.1, respectively). By contrast, CS NCs had a larger size and higher PDI (456 ± 2; 0.3). This result can be explained by the different entanglement of the tensioactive agents (PEG-stearate and sodium cholate) with the cationic polymers at the interface of the oily cores. In fact, CS NCs with reduced sizes could also be produced by adjusting the surfactant composition (11).

#### Loading of rHBsAg Antigen Onto the NCs

Based on the results obtained in the different cellular studies, PR and PARG NCs were selected for loading with rHBsAg. The incorporation of the antigen into the nanostructure surface was attributed to electrostatic forces occurring between the cationic polypeptide of the polymer coating and the negatively charged antigen (−20 mV), as well as to hydrophobic interactions between the antigen and the tensioactive molecules. To associate rHBsAg to the NC surface, the antigen solution was incubated with the previously isolated NCs (11). The mass ratio selected was 4:1 (theoretical concentration of the cationic polymer adsorbed into


*PDI, polydispersity index; HBsAg, hepatitis B surface antigen.*

the NC surface:rHBsAg). Quantification of the non-associated antigen was performed by ELISA. Both prototypes showed an association efficiency greater than 70% (**Table 1**). Regarding the physicochemical properties of the nanostructures, the association of the antigen to the NC surface led to a size increase and a decrease in the zeta potential. These changes support the presence of the antigen on the NC surface, a location that would theoretically facilitate recognition by antigen-presenting cells (18, 19).

#### Activation of the Complement Cascade

The complement system is one of the most important constituents of the innate humoral system. It includes different serum proteins that are activated in cascade upon contact with different stimuli. Three different routes (classic, alternative, and lectins) can activate this system, and the degradation of the C3 is a common step in all of them. The stimulation and the consequent activation of the cascade can promote phagocytosis processes, it can mediate inflammation, and it is involved in the recognition and clearance of pathogens (20). The complement serves as a first line of defense, and it plays an important role in promoting antigen-specific responses by enhancing both B- and T-cell immunity (21). In fact, the bioactivity of alum (the most used adjuvant in vaccines worldwide) is strongly related to its ability to activate the complement cascade (22).

Analysis of the C3 degradation by Western blot showed that PR NCs slightly induced the activation of the complement cascade. The highest rate of activation was achieved at the highest concentration tested (1 mg/mL of theoretical concentration of protamine), and the degradation rate was double that found in the negative control (**Figure 1**). This result was confirmed by ELISA using the quantification of the C5a factor levels, which were two times higher in the plasma exposed to the NCs than in the untreated negative control (data not shown).

It has been previously reported that the nanoparticle surface composition plays a crucial role in the complement cascade activation: the presence of a high surface density of amino groups in the polymer chain can enhance the interaction with the C3b α-chain (23). PR and PARG are guanidine-rich polymers, and they would be expected to activate the complement cascade. By contrast, the results on the complement activation properties of chitosan reported to date are quite variable (24, 25), probably due to the different types and sources of chitosan on the market. In our case, activation was not observed after contact between this polymer and the complement cascade proteins.

Western blot analysis showed that only PR NCs, and not PARG NCs, were able to achieve significant levels of the C3 split products at the highest concentration tested. This finding suggests that another mechanism, besides the interaction with the C3b α-chain, may be contributing to the activation of the complement cascade. This difference could also be attributed to a higher percentage of adsorption of the PR when compared with the PARG. Further investigation need to be perform to validate this hypothesis. Although little is known about the mechanism of complement activation induced by arginine-rich polymers, it has previously been reported that PR, which is commonly used to neutralize the effect of heparin after cardiopulmonar bypass, can induce complement cascade activation through the classic route when heparin–PR complexes are formed (26).

#### *In Vitro* Cell Studies Cell Viability

Six different concentrations of the nanosystems (from 250 to 7.8 µg/mL) were used to evaluate the cell viability. As shown in **Figure 2**, a dose-dependent trend was observed for all formulations. PARG NCs reduced cell viability to a greater extent than the other prototypes, with an IC50 close to 30 µg/mL, followed by PR NCs > NE > CS NCs. A major reduction in the cell viability has been previously reported with NCs presenting PARG in the surface composition, being this toxicity reported as dose dependent and influenced by the composition of the nucleus (27). In addition, it has previously been reported that the toxic effect induced by nanoparticles decreases with increasing particle size (28). This may explain why CS NCs are the prototype that reduces less the cell viability. In the case of the NE, this prototype was only able to reduce the cell viability at the higher concentrations tested: the negative charge on the surface may play an important role in the reduction of cell viability (29, 30).

#### Cellular Uptake and ROS Production by Macrophages

Fluorescence microscopy and flow cytometry results (**Figure 3**) indicate that all prototypes were efficiently internalized by the cells in a dose-dependent manner. A real-time uptake study using Confocal Laser Scanning Microscopy and the High Content Screening Automation HCS A module was performed by incubating macrophages with fluorescent NCs during 90 min. All of the prototypes were quickly internalized, and the fluorescence was detected after 15 min from the starting time point of the experiment (**Figure 4**). The fluorescence intensity increased up to the last time point of the experiment.

It has been reported in the literature that particulate delivery systems are commonly taken up by phagocytic cells and that this internalization is dependent on particle size, composition, and surface charge (31–33). In general, charged large particles are internalized more rapidly than small neutral particles, with the cationic prototypes being particularly prone to uptake within

charged systems (34). In this study, all the formulations are charged, positively or negatively, and have proven to be efficiently internalized. The charge, regardless of its sign, may have facilitated capture by macrophages: while cationic nanoparticles would have easily interacted with anionic cell surfaces, the internalization of anionic NEs may have occurred through the nonspecific binding of the particles on cationic domains followed by endocytosis (35). It has been described in other polymeric nanomaterials, that once they are internalized in endosomes, they go to lysosome where they are processed. They can induce ROS production, and eventually, trigger the immune response (36–38). In addition, it has been reported that ROS production by the dendritic cells plays an important role in the antigen presentation in that it stimulates the cross-presentation in these cells and therefore it enhances the CD8<sup>+</sup> T lymphocyte responses (39). As a consequence, ROS production in the promyelocytic cell line HL60 was studied as a measure of cell activation after the internalization of these prototypes, where the cell activation was at the same time an indicator of the initiation of the adaptive immune response.

The ROS production was studied after cell exposure to nanostructures during 1 or 12 h (Figure S3 in Supplementary Material). All the prototypes induced the production of ROS in a dose- and time-dependent manner. It has been previously reported that the induction of ROS by inorganic nanoparticles could be a mechanism for the promotion of cell death (40). However, little is known about ROS production by polymeric nanoparticles as a mechanism of cell defense against infections.

#### Cytokine Profile Evaluation

Cytokines are peptides, small proteins, or glycoproteins that play an important regulatory role in several biological processes. They

are produced mainly by immune cells, but also by several other cell types. Cytokines are the main messengers between cells of the immune system and they regulate the innate and adaptive responses and also modulate the inflammatory response (41). In this study, our main goal was to investigate if the NCs, to be used as antigen carriers, are inherently able to stimulate cytokine secretion by hPBMCs.

It can be seen from the results in **Table 2** that the PR and PARG NCs have a slightly greater tendency to induce the production of pro-inflammatory cytokines than CS NCs or the control NE. The main difference was observed in the secretion of TNFα, where two out of the three donors responded in the case of PR and PARG NCs, only one out of three donors was positive for CS NCs, and no positive response was observed for the control NEs. Although these responses are not remarkable, they are in agreement with previous studies in which it was found that protamine and PARG are able to induce the production of TNFα; PR was included in a nanoparticulate form and incorporated CpG (42), while in the case of PARG it was the bulk material (43). On the other hand, it has also been reported that cationic particles are more efficient at inducing inflammatory responses than anionic or neutral particles (34). This observation is consistent with the overall cytokine profiles observed for the three types of NCs. By contrast, the anionic NE was only able to induce the production of IL-8, a cytokine that is basally produced at high levels (**Table 2**). The differences in the cytokine secretion profiles for the three cationic prototypes could be explained by the smaller size of PR and PARG NCs when compared with CS NCs: the higher specific surface area may contribute to a more efficient interaction with the mononuclear cells (44).

The cytokine profile induced by PR and PARG prototypes is dominated by IL-6, Il-1β, IL-8, and TNFα (at the highest dose tested). All of these cytokines are mainly produced by macrophages when they are exposed to inflammatory stimuli (45). While IL-8 is a potent chemotactic factor for lymphocytes and neutrophils that is often released during phagocytosis, IL-6 induces B cell differentiation and increases antibody titers (46, 47). In addition, IL-1β is released upon activation of macrophages *via* Caspase 1 pathway, and similarly to this one, TNFα is also an endogenous pyrogen that is produced and released at the early stages of the immune response to infections (45). Both prototypes have proved to be efficiently internalized by phagocytic cells (**Figures 3** and **4**), hence a high internalization rate during the first steps of the immune response may have contributed to the activation of human monocytes, leading to the release of all these pro-inflammatory cytokines generating an adequate scenario to initiate a specific immune response.

#### Activation Markers

Human CD71 (transferrin receptor) and class II molecules of the major histocompatibility complex (HLA-DR) are



*NE, nanoemulsion; CS NC, chitosan nanocapsule; PR NC, protamine nanocapsule; PARG NC, polyarginine nanocapsule.*

*The positive control was lipopolysaccharide (LPS) (1 µg/mL) with phytohemagglutinin (PHA) (10 µg/mL).*

+*: 1–10;* ++*: 10–100;* +++*: 100–1,000;* ++++*:* >*1,000-fold higher than negative control (cells incubated in culture media).*

+*N/tested: number of positive responsive persons/three donors tested.*

membrane markers that can be modulated during cell activation. Transferrin receptor is expressed at different levels in many cell types and increases its expression in activated cells such as T and B lymphocytes, because this mechanism is used to import extracellular iron for metabolic needs (48, 49). HLA-DR is constitutively expressed in B cells and it has been shown to be down regulated under certain stimuli. The modulation of this marker changes depending on the stage of differentiation of the B cells upon its activation. While an increase during the first contact with stimuli is commonly observed, a decrease occurs during the last steps of differentiation (50–52). On the other hand, T cells do not express HLA-DR unless they are activated (53).

The expression of the activation markers CD71 (transferrin receptor) and HLA-DR in T (as CD3<sup>+</sup>) and B (as CD19<sup>+</sup>) cells incubated with the different nanostructures at two different doses (10 and 100 µg/mL) was evaluated by flow cytometry (**Figure 5**). The results indicate that only PARG NCs were able to increase the percentage of CD71<sup>+</sup>HLA-DR<sup>+</sup> T cells and also the expression of CD71 in B cells. Furthermore, PARG NCs induced the modulation of HLA-DR in B cells. While this effect seems to correlate to cell activation, it cannot be excluded that a cytotoxic effect could have been the responsible for this modulation in the expression of these markers as consequence of exposure to damage-associated molecular patterns (DAMPs, danger signals, or alarmins). In fact, these molecules have been described as the possible responsibles for the adjuvant effect of other molecules such as alum (54).

These results are in agreement with the cytokine data, where PARG NCs induced the production of different cytokines in hPBMCs. A synergistic effect of all of these cytokines may play a crucial role in the lymphocyte activation and subsequently in the modulation of markers such as CD71 or HLA-DR.

CD71+HLADR+ cells are represented. The results are the average of three independent experiments with hPBMCs from three different healthy donors (\**p* < 0.05).

#### Routes of Activation. MAP Kinases

Mitogen-activated protein kinases (MAPKs) are a protein family that is involved in several processes such as gene expression, mitosis, cell migration, metabolism, and programmed cell death. MAPKs are activated through the phosphorylation of certain residues and, in turn, they activate their target proteins, including other protein kinases, transcription factors, or cytoskeleton proteins.

The MAPK families that are most widely reported in the literature are the ERK, the c-Jun NH2-terminal kinases (JNK), and the p38 family. While ERKs are involved in the regulation of processes such as mitosis or meiosis, JNK are mainly involved in the control of programmed cell death or apoptosis. On the other hand, p38 participates in the immune response activation (55). In addition, the NFκB is a transcription factor that plays an important role in the regulation of certain immune responses by modulating the expression of receptors or surface adhesion molecules, inflammatory cytokines, and chemokines. Furthermore, it is involved in the response to stress signals and in cell survival (56, 57).

In this study, we investigated the capacity of PR and PARG NCs to induce the secretion of MAPK and NFκB in two different cell lines, namely, the Jurkat (human leukemic T-cell line) and the Hmy (human B Lymphoblast cell line) cell lines. It has previously been reported that activation of the NFκB pathway leads to the degradation of the IKBα inhibitor and therefore a decrease in IKBα was expected in these cells.

The Western blot analysis only showed a slight increase in the phosphorylation levels of the p38 family after 3 h in contact with PARG NCs (**Figure 6**). Considering Hmy and Jurkat cell lines as representative B and T cells, respectively, of the main PBMCs, this result could correlate with the observed capacity of these NCs to induce the production of Il-1β, IL-6 and IL-8 in human PBMC, as it is known that the p38 family is activated in response to different inflammatory cytokines and the triggering of different immune responses (55, 58, 59). In addition, p-p38 has also been shown to be involved in the proliferation of Th1 cells (60). This explains why p38 is activated mainly by PARG NCs, as this NCs formulation showed a Th1 profile characterized by the production of TNFα.

In the case of ERK, both prototypes, PR and PARG NCs, induced its activation in Hmy and Jurkat cells, with this effect being time dependent only in Jurkat cells (**Figure 6**). ERK is activated in response to several cytokines, and it is involved not only in cell survival but also in cell differentiation (55). In addition, ERK phosphorylation is an important step during both T and B cell activation (61, 62) and in the Th2 differentiation process (63).

Regarding p-SAP/JNK, activation was not detected upon contact with any of these prototypes (**Figure 6**). This fact can be explained by the pro-apoptotic activity of the p-SAP/JNK family:

the lack of activation was expected (58, 64) since none of the prototypes showed a cytotoxic effect at low doses and in short periods of time.

The activation of NFκB was detected by the decrease in its IkBα inhibitor levels (**Figure 6**). This fact was observed in both Jurkat and Hmy cells and with both protamine NCs and PARG NCs. NFκB is activated by a wide range of signals such as antigen receptors, pattern-recognition receptors, and different TNF and IL-1 receptors. This activation regulates not only the innate but also the adaptive immunity and different inflammatory responses (65). The decrease in the IkBα levels in both Jurkat and Hmy cell lines upon contact with this immune stimulator was expected considering that this transcription factor is essential to trigger some processes related to immune cell activation.

In light of the results obtained, it can be concluded that PR and PARG NCs are able to induce the phosphorylation of p38 and ERK, as well as the degradation of IκBα in both T and B cell lines, with PARG NCs having a slightly superior activation capacity.

Overall, data obtained by *in vitro* studies, postulate PR and PARG NCs as the most promising prototypes as antigen delivery systems. On the other hand, NE and CS NC have not demonstrated to be strong inducers of immune responses and do not activate the complement cascade, thus they could be useful as carriers intended to avoid the immune system recognition, for example to target cells for gene therapy (66).

#### *In Vivo* Studies

Taking into account the results obtained in the *in vitro* studies, PARG and PR NCs were selected for the *in vivo* evaluation of their capacity to induce the production of an immune response against a model antigen, rHBsAg. To this end, groups of five Balb/C female mice were intramuscularly immunized with two doses of rHBsAg adsorbed onto PR NCs and PARG NCs or alum. The humoral-specific immune response (anti-rHBsAg serum IgG) and the type of immune response elicited (measured by the IgG1/ IgG2a ratio) were both analyzed. In addition, a panel of several genes involved in immune responses was studied *ex vivo* in cells from immunized mice.

#### Immune Response Elicited by Immunostimulating rHBsAg-Loaded NCs

It can be seen from the results in **Figure 7** that PR NCs and PARG NCs were able to elicit antigen-specific antibody levels at 6 weeks after immunization, with levels close to 400 and 130 mIU/mL, respectively. These levels increased slightly to 500 and 200 mIU/ mL, respectively, after 10 weeks. It is worthy to note that in both cases, seroprotection levels (>100 mUI/mL) were achieved, although this increase was not as high as the one observed in control mice immunized with alum-rHBsAg (from 1,100 to 1,600 mUI/mL) (**Figure 7**).

Despite the response induced by the marketed alum-rHBsAg vaccine was much higher, the average levels achieved by PR NCs and PARG NCs were over 100 mIU/mL, which has been established as the threshold of antibody protective levels in humans (67). It can also be stated that PR NCs showed better immunostimulant properties than PARG NCs as they induced higher specific antibody levels in mice (5/5). This difference can be explained by the higher ability of PR NCs to activate the complement cascade compared with PARG NCs, which may play a crucial role in the adjuvant effect of the NC. Thus, independently on the NCs under study, the immune response induced by both prototypes could be due to a combination of several routes such as p38 and ERK. Moreover, the lack of a strong activation of the

Figure 7 | Systemic immune response upon intramuscular immunization with rHBsAg PR NCs and PARG NCs. The top graphs represent individual IgG concentrations of anti-HBsAg (expressed in mUI/mL) in each mouse (triangles) and the average of five mice (dash) at 6 weeks (left graph) and 10 weeks (right graph) after the first administration. The number of mice with antibody levels over 100 mIU/mL and the IgG1/IgG2a ratios at different time points are also represented in the lower table.

NFκB pathway could be related to the moderate *in vivo* immune response obtained in the presence of NCs.

#### Modulation of the Immune Response

To determine the type of response induced by these prototypes, the IgG1/IgG2a ratio was calculated for both prototypes. These two IgGs were selected because while the IgG2a subtype is related to cellular immune responses mediated by Th1 lymphocytes, IgG1 is mostly produced after Th2 activation, which mediates the humoral immune response.

Both humoral and cellular responses were elicited by both prototypes. The highest levels detected for Th2, with ratios ranging from 2.6 to 1.7. While this ratio did not change over time for PR NCs, it decreased slightly for PARG NCs, which indicates that an increase in Th1 response had occurred (**Figure 7**). This finding is in agreement with the combination of cytokines produced by hPBMCs upon contact with PARG NCs and also due to the activation of both p-ERK and p-p38, which are commonly involved in Th2 and Th1 responses, respectively (60, 63). The same trend was observed in the control group, although in this case the initial IgG1/IgG2a ratio was higher (close to 5). This is in agreement with the known fact that alum is a very effective Th2 activator (68).

#### Gene Expression Assays in Restimulated Splenocytes From Immunized Mice

To understand the mechanism involved in the immunostimulant properties of PARG NCs and PR NCs, the expression of different genes was studied by qPCR. For this purpose, splenocytes from immunized mice were restimulated with the HBsAg during 12 or 24 h. RNA from these cells was extracted to analyze the expression of genes involved in apoptosis and inflammation (Caspase 1, 3, 8, and 9; Fas and Bcl2) of different cytokines (IL-10, 1L-4, IL-2, INF-γ, IL-6, TNF-α, IL-1β, and IL-17) and of certain activation markers (CD74, B2M, CD71, CD69, CD154, CD80, and CD86). The results of these studies are summarized in **Table 3**, and the values highlighted in red show an increase in the RQ value greater than 2. Significant differences were only registered for IL17A in mice immunized with PR NCs-HBsAg both 12 and 24 h after restimulation. A minor increase was also observed in IL-1β levels in the same group after 24 h of restimulation.

IL-17A is a cytokine produced by Th17 cells and by other immune cells, including γδ T cells, NKT cells, NK cells, neutrophils, and eosinophils. In fact, one of its functions is to connect innate and adaptive immune responses. IL-17A is a pro-inflammatory cytokine that induces the production of other cytokines and chemokines, which are crucial in the recruitment, activation, and migration of neutrophils to the site of inflammation. It has been reported that this cytokine is also involved in the development of germinal centers (GC) (69). IL-1β, which is produced through the NALP3 inflammasome route and whose release can be initiated by the complement cascade activation (70), is one of the major mediators of inflammation and it is the main known endogenous pyrogen. IL-1β is also very important in the initiation of innate immune responses and, as a consequence, it has a critical role in the control of pathogenic infections (71). Both cytokines IL-17

Table 3 | Changes in the expression of selected genes in splenocytes from mice immunized with alum-rHBsAg, PRNC-rHBsAg, and PARGNC-rHBsAg restimulated for 12 or 24 h with rHBsAg.


*Values in this table represent the RQ with respect to the control (splenocytes from mice immunized with phosphate-buffered saline 1*× *and restimulated in vitro with rHBsAg) calculated from independent data for 3 mice per group.*

*RQ values greater than 2 are represented in red (% CV of* Δ*Ct* < *25).*

*RQ, relative quantification; PR NC, protamine nanocapsule; PARG NC, polyarginine nanocapsule; rHBsAg, recombinant hepatitis B surface antigen.*

and IL-1 have been proposed as the main mediators of the adjuvant effect of several vaccines (72–74). The induction of IL-17 and IL-1β production by PR NCs in cells from immunized mice may have an important role in initiating the immune response after vaccination. These could induce an inflammatory response that would create the appropriate environment for the development of a specific immune response.

PR NCs have proved to be efficient antigen carriers *in vivo* using the HBsAg as a model antigen. We postulate that the incorporation of immunostimulant molecules into this formulation, such as imiquimod, CpG, polyI:C (13, 14, 75), could lead to the design of a prototype able to induce an enhanced immune response, even stronger than that induced by alum, and also to an effective system that could modulate this immune response toward a Th1 profile.

#### CONCLUSION

The results underline the potential of arginine-rich NCs as antigen delivery carriers. The systematic analysis performed in this study provides valuable information regarding the feasibility of modulating the immunostimulatory properties of NCs by selecting the nature of the polymer shell. This information could help in the design and development of novel nanocarriers for vaccine delivery, taking as a starting point the physicochemical properties of PR NCs, which was the prototype that showed the best immunological profile in the *in vivo* evaluation.

#### ETHICS STATEMENT

Institutional ethics approval to work with human samples from healthy donors was obtained from the Ethics Committee for Clinical Research (Xunta de Galicia, Spain, 2013/272). All

#### REFERENCES


participants included in the study gave their written informed consent. All protocols developed with mice were adapted from the guidelines of the Spanish regulations (Royal Decree 53/2013) regarding the use of animals in scientific research and under the approval of the Ethical Committee of the University of Vigo.

### AUTHOR CONTRIBUTIONS

MP and EP have contributed to the design, acquisition, analysis, and interpretation of data; the drafting and the revision of the work. JG-A, BS-C, and RS-V have contributed to the design, acquisition, analysis, and interpretation of data for the work. NC, MA, and ÁG-F have contributed to the design, the drafting, and the revision of the work.

#### ACKNOWLEDGMENTS

This work was supported by the Spanish Ministry of Economy and Competitiveness (SAF2011-30337-C02-02 and BIO2014-53091- C3-1-R). Financial support from the Xunta de Galicia (Centro singular de investigación de Galicia 2016–2019 and Grupo de referencia competitiva, ED431C 2016041) and the European Union (European Regional Development Fund—ERDF) is gratefully acknowledged. The authors would like to thank Jesús Méndez from CACTI facilities of the University of Vigo. MP acknowledges fellowships from the Spanish Ministry of Education (FPU predoctoral grants program).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.00791/ full#supplementary-material.


inflammation and IL-1β release. *J Immunol* (2013) 191:1006–10. doi:10.4049/ jimmunol.1300489


**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 Peleteiro, Presas, González-Aramundiz, Sánchez-Correa, Simón-Vázquez, Csaba, Alonso and González-Fernández. 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.*

## Mucosal immunity and Protective efficacy of intranasal inactivated influenza Vaccine is improved by chitosan nanoparticle Delivery in Pigs

*Santosh Dhakal <sup>1</sup> , Sankar Renu1 , Shristi Ghimire1 , Yashavanth Shaan Lakshmanappa1 , Bradley T. Hogshead1 , Ninoshkaly Feliciano-Ruiz1 , Fangjia Lu2 , Harm HogenEsch2 , Steven Krakowka3 , Chang Won Lee1 and Gourapura J. Renukaradhya1 \**

#### *Edited by:*

*África González-Fernández, Centro de Investigaciones Biomédicas (CINBIO), Spain*

#### *Reviewed by:*

*Jose Crecente-Campo, Centro de Investigación en Medicina Molecular y Enfermedades Crónicas (CiMUS), Spain Rong Hai, University of California, Riverside, United States*

#### *\*Correspondence:*

*Gourapura J. Renukaradhya gourapura.1@osu.edu*

#### *Specialty section:*

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

*Received: 21 January 2018 Accepted: 16 April 2018 Published: 02 May 2018*

#### *Citation:*

*Dhakal S, Renu S, Ghimire S, Shaan Lakshmanappa Y, Hogshead BT, Feliciano-Ruiz N, Lu F, HogenEsch H, Krakowka S, Lee CW and Renukaradhya GJ (2018) Mucosal Immunity and Protective Efficacy of Intranasal Inactivated Influenza Vaccine Is Improved by Chitosan Nanoparticle Delivery in Pigs. Front. Immunol. 9:934. doi: 10.3389/fimmu.2018.00934*

*<sup>1</sup> Food Animal Health Research Program, Department of Veterinary Preventive Medicine, The Ohio State University, Wooster, OH, United States, 2Department of Comparative Pathobiology, College of Veterinary Medicine, Purdue University, West Lafayette, IN, United States, 3Department of Veterinary Biosciences, The Ohio State University, Columbus, OH, United States*

Annually, swine influenza A virus (SwIAV) causes severe economic loss to swine industry. Currently used inactivated SwIAV vaccines administered by intramuscular injection provide homologous protection, but limited heterologous protection against constantly evolving field viruses, attributable to the induction of inadequate levels of mucosal IgA and cellular immune responses in the respiratory tract. A novel vaccine delivery platform using mucoadhesive chitosan nanoparticles (CNPs) administered through intranasal (IN) route has the potential to elicit strong mucosal and systemic immune responses in pigs. In this study, we evaluated the immune responses and cross-protective efficacy of IN chitosan encapsulated inactivated SwIAV vaccine in pigs. Killed SwIAV H1N2 (δ-lineage) antigens (KAg) were encapsulated in chitosan polymer-based nanoparticles (CNPs-KAg). The candidate vaccine was administered twice IN as mist to nursery pigs. Vaccinates and controls were then challenged with a zoonotic and virulent heterologous SwIAV H1N1 (γ-lineage). Pigs vaccinated with CNPs-KAg exhibited an enhanced IgG serum antibody and mucosal secretory IgA antibody responses in nasal swabs, bronchoalveolar lavage (BAL) fluids, and lung lysates that were reactive against homologous (H1N2), heterologous (H1N1), and heterosubtypic (H3N2) influenza A virus strains. Prior to challenge, an increased frequency of cytotoxic T lymphocytes, antigen-specific lymphocyte proliferation, and recall IFN-γ secretion by restimulated peripheral blood mononuclear cells in CNPs-KAg compared to control KAg vaccinates were observed. In CNPs-KAg vaccinated pigs challenged with heterologous virus reduced severity of macroscopic and microscopic influenza-associated pulmonary lesions were observed. Importantly, the infectious SwIAV titers in nasal swabs [days post-challenge (DPC) 4] and BAL fluid (DPC 6) were significantly (*p* < 0.05) reduced in CNPs-KAg vaccinates but not in KAg vaccinates when compared to the unvaccinated challenge controls. As well, an increased frequency of T helper memory cells and increased levels of recall IFNγ secretion by tracheobronchial lymph nodes cells were observed. In summary, chitosan SwIAV nanovaccine delivered by IN route elicited strong cross-reactive mucosal IgA and cellular immune responses in the respiratory tract that resulted in a reduced nasal viral shedding and lung virus titers in pigs. Thus, chitosan-based influenza nanovaccine may be an ideal candidate vaccine for use in pigs, and pig is a useful animal model for preclinical testing of particulate IN human influenza vaccines.

Keywords: swine influenza virus, chitosan nanoparticles, mucosal immune response, intranasal vaccination, pigs

## INTRODUCTION

Influenza is caused by influenza A virus (IAV) of *Orthomyxoviridae* family. It is an economically important disease in the global pig industry (1, 2). Virulent swine IAV (SwIAV) infection leads to acute febrile respiratory disease which is often complicated with secondary bacterial infections (3). SwIAV increases its genetic diversity through frequent antigenic drift and antigenic shift. So far, H1N1, H1N2, and H3N2 subtypes are the major SwIAV circulating in pig populations (4). Since epithelial cells lining the porcine respiratory tract bear receptors for both avian and human IAVs, pigs can be infected with IAV from different hosts, and this event favors genetic assortment and adaptation of novel influenza strains of zoonotic and even pandemic potential (5). The pandemic H1N1 virus of 2009 and the more recent "H3N2 variant" virus in the USA are recent examples of swine-origin IAVs which cause infection and resultant pulmonary disease in humans (6, 7). Controlling influenza in pigs through vaccination serves dual benefits by protecting economic loss in swine industry and preventing possible public health risk that these reassorted SwIAVs pose for humans.

Swine influenza vaccines are commercially available. These are multivalent whole-inactivated virus (WIV) vaccines that are administered intramuscularly (IM) (8). The WIV vaccines provide protection against homologous virus infections but do not induce adequate heterologous immunity against constantly evolving IAVs that develop by point mutation(s) (8, 9). Moreover, the IM route used for WIV vaccines does not elicit adequate mucosal immune responses which are essential for providing cross-protective immunity against multitude of variant IAVs (10, 11). Intranasal (IN) vaccine that targets mucosal immune system of the respiratory tract can be a useful alternative to the current IM influenza vaccines used in pigs. Nasal mucosal vaccination not only induces strong protective immune responses at mucosal sites in the respiratory tract but also enhances immunity at distal mucosal and systemic sites (12, 13).

Biodegradable and biocompatible polymer-based nanoparticle (NP) formulation(s) provide an innovative strategy of vaccine antigen delivery to mucosal sites (14). Particulate vaccines facilitate antigen uptake by professional antigen-presenting cells (APCs), maintain slow and sustained antigen release, prevent the antigen(s) from undesirable enzymatic degradation, and potentiate the levels of protective immunity (14, 15). Different types of NPs are investigated for IN delivery of influenza vaccine antigens. For example, IN immunization in mice using liposome-based DNA and subunit influenza nanovaccines are shown to elicit mucosal, cellular, and humoral immune responses (16, 17). Poly(lactic-co-glycolic) acid (PLGA) NP-entrapped highly conserved H1N1 influenza virus peptides administered IN enhances the epitope-specific T cell response and protective efficacy in pigs (18). Ferritin-based IN influenza nanovaccine is shown to enhance mucosal secretary IgA and T cell response and confers homo- and heterosubtypic protection in mice (19). In our previous study, killed SwIAV antigen (KAg) encapsulated in PLGA polymer-based NP and delivered IN induced a robust cross-reactive cell-mediated immune response associated with a significant clearance of challenge heterologous virus from the lungs of pigs (20). In another study, the encapsulation of KAg in polyanhydride polymer-based NP also enhanced the crossreactive cell-mediated immune response against SwIAV (21). However, both PLGA and polyanhydride polymer-based NP SwIAV vaccines used IN in these studies failed to elicit mucosal IgA and systemic IgG antibody responses, most likely due to their biased ability to induce strong T helper 1 (Th1) but not T helper 2 (Th2) responses. This Th1-biased response failed to reduce the nasal virus shedding in pigs (20, 21).

In the present study, we used chitosan, a natural mucoadhesive polymer derived NPs (CNPs) for the encapsulation of SwIAV KAg (CNPs-KAg) and performed a heterologous vaccine challenge trial in nursery pigs. Due to its cationic nature, chitosan binds readily to mucosal surfaces. Chitosan also possesses adjuvant properties, a feature which promotes immune activation (22). Previous studies have shown that CNPs form an attractive platform for mucosal vaccine delivery. For example, live Newcastle disease virus (NDV) encapsulated in CNPs and delivered through oral and IN route in chickens induced a higher secretary IgA antibody response in intestinal mucosa and enhanced the protective efficacy against highly virulent NDV strain challenge infection (23). Similarly, influenza subunit/split virus vaccine delivered in CNPs by IN route improves systemic and mucosal antibody and cell-mediated immune responses in mice (24–26). Hence, we hypothesized that IN delivery of chitosan-based nanovaccine would enhance both mucosal antibody and cellular immune responses and provide better protective immunity against SwIAV in pigs compared to soluble IN-inactivated vaccine. Our results demonstrated that CNPs-KAg IN vaccination improved mucosal IgA response in the entire respiratory tract and also elicited cellmediated immune response against different subtypes of SwIAV, resulting in a reduced nasal viral shedding and an infectious virus burden in the pulmonary parenchyma.

#### MATERIALS AND METHODS

#### SwIAV Propagation and Inactivation

Field isolates of IAVs A/Swine/OH/FAH10-1/10 (H1N2) (27), A/Swine/OH/24366/2007 (H1N1) (28), and A/Turkey/OH/ 313053/2004 (H3N2) (29) were propagated in Madin–Darby canine kidney (MDCK) cells. The H1N2 A/Swine/OH/FAH10- 1/10 (H1N2-OH10) was used for CNPs-KAg vaccine preparation and H1N1 A/Swine/OH/24366/2007 (H1N1-OH7) was used for the virulent virus challenge infection. The H3N2 A/Turkey/OH/313053/2004 (H3N2-OH4) was used together with H1N2-OH10 and H1N1-OH7 for *ex vivo* cross-reactive immune analysis. The H1N2-OH10 vaccine virus and H1N1-OH7 challenge virus are heterologous to each other with 77% HA gene identity, whereas H3N2-OH4 virus, originally isolated from turkeys, is heterosubtypic to other two SwIAVs with HA gene identity of 63% (27–29). For vaccine preparation, cell culture fluid of H1N2-OH10 virus grown in MDCK cells was harvested and subjected to sucrose gradient ultracentrifugation. The virus pellet was suspended in phosphate-buffered saline (PBS), titrated for infectious virus titer, and inactivated by using binary ethyleneimine (Sigma, MO, USA) as described previously (20).

## Preparation of Chitosan-Based Nanovaccine and *In Vitro* Characterization

Chitosan NPs-loaded-killed SwIAV antigen (KAg) (CNPs-KAg) formulation was prepared by the ionic gelation method as described previously (23, 30–32) with some modifications. Briefly, 1.0% (w/v) low-molecular weight chitosan polymeric (Sigma, MO, USA) solution was prepared in an aqueous solution of 4.0% acetic acid under magnetic stirring until the solution became clear. The chitosan solution was sonicated; pH was adjusted to 4.3 and filtered *via* a 0.44-µm syringe filter. Five milliliters of 1.0% chitosan solution was added to 5.0-mL deionized water and incubated with 3.0-mg SwIAV KAg dissolved in 1.0 mL 3-(N-morpholino) propanesulfonic acid (MOPS) buffer at pH 7.4. Consequently, 2.5 mL of 1.0% (w/v) tripolyphosphate (TPP) (Sigma, MO, USA) dissolved in 2.5-mL deionized water was added into the chitosan polymer solution with continuous magnetic stirring at room temperature (RT) (22° C). The formulated SwIAV nanovaccine was centrifuged at 10,000 rpm for 10 min, dispersed in MOPS buffer at pH 7.4, lyophilized with a cryoprotectant, and stored at −80° C.

Particle size and zeta potential of empty and vaccine antigenloaded NPs were measured after dispersion in PBS (pH 7.4) and stored at 4° C for at least 30 h by dynamic light-scattering (DLS) method using a zeta-sizer coupled with an MPT-2 titrator (Malvern) as described previously (33). During each vaccination, CNPs-KAg were freshly prepared and used. The morphology of NPs was obtained by using the cold field emission Hitachi S-4700 scanning electron microscope (SEM) (20). Briefly, the powder form of NPs was loaded on to aluminum stubs and coated with platinum prior to examination under the microscope. Protein loading efficiency in CNPs-KAg was estimated indirectly by determining the difference between the initial amount of protein used for loading CNPs and the protein left in the supernatant (23). *In vitro* protein release profile in CNPs-KAg suspended in PBS for up to 15 days was estimated and expressed as the cumulative percentage release of SwIAV antigen at each time point as described previously (20). In brief, CNPs-KAg suspended in 500 µL PBS (pH 7.4) in triplicate in eppendorf tubes was incubated at 37° C in a revolving roller apparatus. At indicated time point, tubes were centrifuged, supernatant collected, and pellet was resuspended in fresh 500 µL PBS. Protein released on to the supernatant was estimated by micro-BCA protein assay kit (Thermo Scientific, MA, USA) and expressed as the percentage of cumulative protein released over the initial amount at time zero in particles.

## *In Vitro* Uptake of CNPs-KAg by APCs

Peripheral blood mononuclear cells (PBMCs) isolated from 9- to 10-week-old pigs were used for the *in vitro* antigen uptake study. Cells were suspended in enriched-Roswell Park Memorial Institute medium and seeded on to 1 million cells/well in 96-well cell culture plates. After overnight incubation at 37° C in 5% CO2, unattached cells were removed. The attached monocyte/ macrophage cells were treated with SwIAV KAg or CNPs-KAg containing the antigen at 10 µg/mL concentration for 10, 30, and 150 min. After the indicated period of incubation, the cells were fixed with 80% acetone, stained with IAV nucleoprotein-specific antibody (CalBioreagents, CA, USA) followed by Alexa Fluor 488 conjugated goat anti-mouse IgG antibody (Life technologies, OR, USA). Cells were evaluated under fluorescent microscope (Olympus IX70) and photomicrographs were taken (20×). For evaluation of SwIAV antigen uptake from CNPs-KAg-treated porcine monocyte/macrophages prepared from three pigs, PBMCs separately were incubated in 48-well plates seeded with 2 × 106 cells per well overnight as described above. Cells were treated with KAg or CNPs-KAg at SwIAV antigen concentration 10 µg/mL for 10, 30, and 150 min. A positive control was MDCK cells infected with SwIAV H1N2-OH10 at multiplicity of infection (MOI) 1 for 12 h. After the indicated period of treatment, cells were fixed using 1% paraformaldehyde (PFA), permeabilized, and stained with IAV nucleoprotein-specific antibody (CalBioreagents, CA, USA) followed by treatment with goat anti-mouse IgG Alexa Flour 488 conjugated secondary antibody. We acquired 50,000 events in BD Aria II flow cytometer (BD Biosciences, CA, USA) and the data analyzed by using the FlowJo software (Tree Star, OR, USA).

### *In Vitro* Generation and Stimulation of Porcine Dendritic Cells (DCs)

Porcine monocyte-derived dendritic cells (MoDCs) were prepared from PBMCs isolated from seven pigs as described previously (34) with few modifications. Briefly, 25 million PBMCs per mL were seeded in each well of six-well culture plates. After overnight incubation at 37° C in a 5% CO2 incubator, non-adherent cells were discarded and adhered cells were treated with GM-CSF (25 ng/mL) and interleukin (IL-4) (10 ng/mL) cytokines. Half of the culture media was replaced on every third day. On day 7, the plate was centrifuged at 2,000 RPM at 4° C for 5 min and the supernant was harvested gently, and the generated immature MoDCs were stimulated in the same plate without seeding into fresh plates with medium only, LPS control (10 µg/mL), KAg (10 µg/mL), and CNPs-KAg containing 10 µg/mL of KAg for 48 h. The culture supernatant was harvested, and the levels of innate, pro-inflammatory and Th1 cytokines, IFN-α, TNF-α, IL-1β, IL-12, IL-6, and IL-10 were estimated by ELISA as described previously (35).

## Experimental Design

Cesarean-delivered colostrum-deprived and bovine colostrum-fed influenza antibody-free Large White-Duroc crossbred piglets were raised in our BSL2 animal facility at OARDC. Piglets at 4 weeks of age (male and female) were randomly assigned into one of the three experimental groups and kept in separate isolation rooms (**Table 1**). The first IN vaccination was performed at 5 weeks of age and the second IN booster vaccination at 8 weeks of age. All piglets receiving virulent SwIAV were challenged at 10 weeks of age. Separate groups of pigs were vaccinated IN with DMEM (Gibco) or with 1 × 107 TCID50 equivalent of KAg or CNPs-KAg suspended in 2 mL DMEM by IN mist as described previously (20). The challenge infection was done using heterologous H1N1-OH7 SwIAV (6 × 106 TCID50) in 2 mL, divided into 1 mL administered IN and 1 mL intratracheally as described previously (18, 20, 21).

Serum samples were collected at days post-vaccination (DPV) 21 and 35. The rectal temperatures were recorded daily from day post-challenge (DPC) 0 onward, and nasal swab samples were collected at DPC 0, DPC 4, and DPC 6. Pigs were euthanized at DPC 6, and serum and bronchoalveolar lavage (BAL) fluid were collected. During necropsy, lungs were examined for macroscopic pneumonic lesions and scored as described previously (20). Lung lysates were prepared by homogenization of 1.0 g of lung tissue collected from the right apical lobe (20). Nasal swabs, sera, BAL fluid, and lung lysate samples were stored at −80° C until processed for antibody and virus titration. The PBMCs were isolated from blood at DPV 35/DPC 0 and DPC 6 (20). Mononuclear cells were harvested from tracheobronchial lymph nodes (TBLN-MNCs) at DPC 6 as described previously (35).

#### Antibody Titration

Hemagglutination inhibition (HI) antibody titers against IAVs H1N1-OH7, H1N2-OH10, and H3N2-OH4 in sera and BAL fluid samples were determined as described previously (20). The SwIAVspecific IgG and IgA antibodies in nasal swabs, sera, BAL fluids, and lung lysates were determined by ELISA (20). Briefly, 96-well plates (Greiner bio-one, NC, USA) were coated overnight with respective pre-titrated IAV antigen (5 µg/mL) and blocked with 5% skim milk powder containing 0.05% Tween-20 for 2 h at RT. After washing, fivefold dilutions of nasal swab, serum, BAL fluid, and lung lysate samples in PBS containing 2.5% skim milk powder and 0.05% Tween-20 were added to marked duplicate wells, incubated for 2 h at RT, washed, and horse radish peroxidase-conjugated goat anti-pig IgA (Bethyl Laboratories Inc., TX, USA) or goat anti-pig IgG (KPL, MD, USA) was added. Finally, the antigen and antibody interaction were detected by using 1:1 mixture of peroxidase substrate solution B and TMB peroxidase substrate (KPL, MD, USA). The reaction was stopped using 1.0 M phosphoric acid, and optical



density was measured at 450 nm using Spectramax microplate reader (Molecular devices, CA, USA).

#### Antigen-Specific Cell Proliferation Assay

The PBMCs isolated at DPV 35/DPC 0 were cultured together with H1N1-OH7, H1N2-OH10, or H3N2-OH4 SwIAV at 0.1 MOI and incubated at 37° C in 5% CO2 incubator for 72 h. Antigen-specific lymphocyte proliferation was determined by using the cell titer 96 aqueous non-radioactive proliferation assay kit (Promega, WI, USA). The cell proliferative response was compared among groups using lymphocyte stimulation index values as described previously (20).

#### Cytokine ELISA

The PBMCs isolated at DPC 0 and TBLN-MNCs at DPC 6 were cultured with H1N2-OH10, H1N1-OH7, or H3N2-OH4 SwIAV at 0.1 MOI. After 72 h of stimulation, the supernatant was collected and interferon gamma (IFNγ) secretion was determined by ELISA as described previously (20). Similarly, the production of interleukin-6 (IL-6) in BAL fluid collected at DPC 6 was determined by ELISA (35).

### Virus Titration

Viral titers contained in nasal swabs and BAL fluids were determined in a 10-fold dilution of the samples in DMEM containing TPCK-trypsin (1 µg/mL). The samples were transferred to quadruplicate 96-well cell culture plate wells containing overnight cultured monolayers of MDCK cells and incubated for 72 h, 37° C, 5.0% CO2. Cells were fixed with acetone and immunostained with IAV nucleoprotein-specific antibody (#M058, CalBioreagents, CA, USA) followed by Alexa Fluor 488 conjugated goat antimouse IgG (H + L) antibody (Life technologies, OR, USA). Virus replication in cells was determined by using immunofluorescence technique as described previously (20).

#### Histopathology of Lungs

For histopathological analysis of pulmonary tissues, 10% formalininflated apical, cardiac, and diaphragmatic lobes were collected and further emulsion-fixed in 10% neutral buffered formalin. Five-micrometer sections of formalin-fixed, paraffin-embedded apical, cardiac, and diaphragmatic lung lobes were stained with hematoxylin and eosin (H&E) as previously described (20). The H&E-stained tissue sections were examined for microscopic changes of interstitial pneumonia, peribronchial and perivascular accumulation of mononuclear cells, bronchial exudates, and epithelial changes related to influenza infection. All these parameters were scored by a board-certified veterinary pathologist (SK) who was not provided with any vaccination history of pig groups in a scale of 0 (no change compared from normal) to 3 (marked changes from normal) as described previously (20).

#### Flow Cytometry

The PBMCs isolated at DPC 0 and TBLN-MNCs at DPC 6 were immunostained for T lymphocyte subset phenotyping as described previously (20). Antibodies used in the flow cytometry were anti-porcine CD3 (Southernbiotech, AL, USA), CD4α (Southernbiotech, AL, USA), CD8α (Southernbiotech, AL, USA), and CD8β (BD Biosciences, CA, USA). Briefly, the cells were blocked with 2% pig serum in fluorescence-activated cell sorting (FACS) buffer and surface labeled with pig lymphocytespecific purified, biotin or fluorochrome-conjugated antibodies or their respective antibody isotypes. Cells were fixed using 1% PFA, washed, suspended in FACS buffer, and acquired using BD Aria II flow cytometer (BD Biosciences, CA, USA). Data analysis was done using FlowJo software (Tree Star, OR, USA).

## Quantitative Reverse Transcription PCR (RT-qPCR)

Total RNA was extracted from PBMCs at DPC 0 and TBLN at DPC 6 using TRIzol reagent (Invitrogen, CA, USA) as per the manufacturer's instructions. NanoDrop™ 2000c Spectrophotometer (Thermo Fisher Scientific, MA, USA) was used to determine the concentration and purity of RNA. cDNA was prepared from 1 µg of total RNA using the QuantiTect Reverse Transcription Kit (AIAGEN). Primers of housekeeping gene (β actin) and target genes (T-bet and GATA-3) used in this experiment were described previously (36). The mRNA expression was analyzed by 7500 Real-Time qPCR system (Applied Biosystems, CA, USA) using the qScript™ One-Step SYBR Green qRT-PCR kit, Low ROX™ (Quantabio, MA, USA). The target gene expression level was normalized with housekeeping gene levels, and the fold change was determined by comparative 2−ΔΔ*<sup>C</sup>*T method (37).

#### Statistical Analysis

Statistical analysis was performed by using non-parametric Kruskal–Wallis test followed by Dunn's *post hoc* test using the software GraphPad Prism 5 (GraphPad Software, Inc., CA, USA). Pig rectal temperature data were analyzed by repeated measure

Figure 1 | *In vitro* characteristics of chitosan nanoparticles (CNPs)-KAg. Diameter of (A) empty CNPs and (B) swine influenza A virus (SwIAV)-killed antigen (KAg)-loaded CNPs (CNPs-KAg) determined by dynamic light scattering. Scanning electron microscope (SEM) images of (C) empty CNPs and (D) CNPs-KAg. (E) Release of KAg from CNPs-KAg suspended in phosphate-buffered saline over a period of 15 days. (F) Uptake of soluble SwIAV KAg or CNPs-KAg formulation by monocytes/macrophages at indicated time points determined by fluorescent microscopy (Olympus, IX70, 20× magnifications). The frequency of monocytes/ macrophages uptaken SwIAV KAg treated with soluble antigen or CNPs-KAg determined by flow cytometry: (G) SwIAV-infected Madin-Darby canine kidney (MDCK) cells as positive control; (H) a representative picture of SwIAV KAg or CNPs-KAg uptake by porcine monocytes/macrophages after 150 min treatment; and (I) percentage of cells with internalized SwIAV antigen at 10, 30, and 150 min treatment.

ANOVA using Friedman test followed by Dunn's pairwise comparison. Cytokine data (**Figure 2**) between two groups were analyzed by Mann–Whitney test. A *p*-value of less than 0.05 was considered to be statistically significant. The infectious virus titer was determined using Reed and Muench method. Data were presented as the mean ± SEM of three to five pigs except for the HI titers which were expressed as the geometric mean with 95% confidence interval.

#### RESULTS

#### Characterization of CNPs-KAg Vaccine Candidate

The encapsulation efficiency of SwIAV KAg in chitosan nanovaccine formulation was 67%. This result was comparable to the encapsulation efficiency of chitosan NPs entrapped with *Salmonella* outer membrane protein antigens (70%) (Renu et al., 2018, manuscript under review). As determined by DLS, the average size of the empty (**Figure 1A**) and antigen-loaded (**Figure 1B**) NPs was 414.2 and 571.7 nm, respectively. Empty NPs showed two peaks at 36 (~10%) and 323 nm (~90%) with polydispersity index (PDI) of 0.39. Likewise, antigen-loaded NPs also had two peaks at 70 (~15%) and 468 nm (~85%) with PDI of 0.60. Data show that the CNPs-KAg were polydispersed in nature. SEM analysis showed that the morphology of the empty NPs was spherical with a smooth surface (**Figure 1C**), while antigen-loaded NPs had a relatively rough and irregular surface (**Figure 1D**). The surface charge of empty and antigen-loaded chitosan NPs was +1.88 and +1.69 mV, respectively. We observed 6% burst release, i.e., surface-associated antigen release during the first 1 h, and on an average, 9% of antigen was released after 24 h of incubation. Further, a slow and sustained release of antigen was observed with a cumulative release of approximately 46% after 15 days (**Figure 1E**).

To determine whether chitosan encapsulation of KAg enhances the uptake of antigen by APCs, we prepared monocyte/ macrophages from PBMCs and allowed for interaction with KAg or CNPs-KAg and stopped the reaction at three different time points. Internalization of CNPs-KAg vaccine by monocytes/macrophages was observed within 10 min of treatment indicated by a higher number of influenza-specific fluorescent signals compared to KAg treatment (**Figure 1F**). Further, the uptake of CNPs-KAg was substantially increased after 30 and 150 min post treatment compared to control KAg-treated cells. We also performed flow cytometry analysis of monocyte/macrophages treated with KAg

or CNPs-KAg to determine the frequency of specific uptake of influenza antigens in APCs. MDCK cells infected with SwIAV H1N2-OH10 were used as positive control (**Figure 1G**) and a representative picture of flow cytometry analysis of KAg or CNPs-KAg uptake by monocyte/macrophages is also shown (**Figure 1H**). In soluble KAg-treated cells, an average of 2.7, 7.1, and 10.1% cells positive for influenza antigen, and in CNPs-KAg-treated cells, 7.2, 11.7, and 16% cells with uptaken influenza antigen after 10, 30, and 150 min of incubation were noticed, respectively (**Figure 1I**). These data clearly demonstrated that CNPs-KAg was efficiently internalized by pig APCs better than soluble antigens.

### CNPs-KAg Formulation Induced the Secretion of Cytokines by Porcine MoDCs *In Vitro*

In order to elucidate the adjuvant property of chitosan NPs in porcine APCs, we treated porcine MoDCs with medium and LPS as control to compare the effect of soluble KAg and CNPs-KAg treatment in inducing the secretion of different cytokines. As expected, the medium control cells had very little secretion of all the detected cytokines, while LPS treatment induced the production of all the analyzed cytokines except IFN-α (**Figures 2A–F**). Cells treated with KAg secreted significantly higher levels of pro-inflammatory cytokines TNF-α (**Figure 2B**) and IL-6 (**Figure 2E**) and Th1 cytokine IL-12 (**Figure 2D**) compared to medium control. In DCs treated with CNPs-KAg, the production of innate IFN-α (**Figure 2A**), TNF-α (**Figure 2B**), IL-1β (**Figure 2C**), and IL-12 (**Figure 2D**) was significantly higher in CNPs-KAg-treated compared to that in soluble KAgtreated cells. In CNPs-KAg-treated cells, the production of IL-6 (**Figure 2E**) and Th2 cytokine IL-10 (**Figure 2F**) was higher than medium control cells, but not significantly higher compared to KAg-treated cells. Our *in vitro* DCs treatment data suggest that chitosan nanovaccine formulation has a potent adjuvant effect on porcine DCs.

## CNPs-KAg Vaccine Augmented the IAV-Specific Mucosal Antibody Response in the Respiratory Tract of Pigs

Secretary IgA antibody levels in nasal swab samples collected after IN prime-boost vaccination at DPV 35/DPC 0 were significantly higher (*p* < 0.05) in CNPs-KAg-vaccinated pigs compared to those in pigs receiving soluble KAg when tested against the homologous H1N2-OH10 (**Figure 3A**), heterologous H1N1-OH7 (**Figure 3B**), and heterosubtypic H3N2-OH4 IAVs (**Figure 3C**). A significant difference in antibody response was observed between CNPs-KAg and KAg vaccinates in serial twofold diluted nasal swab samples (**Figures 3A–C**). These data suggest that the CNPs-KAg IN delivery induced an enhanced cross-reactive mucosal secretary IgA antibody response in pigs. Specific IgG antibody response in sera after prime-boost vaccination in KAgvaccinated pigs against the vaccine virus was comparable to the CNPs-KAg vaccine group (**Figure 3D**). A significantly higher (*p* < 0.05) cross-reactive IgG response was observed in CNPs-KAg vaccinates against heterologous H1N1-OH7 (**Figure 3E**) and heterosubtypic H3N2-OH10 (**Figure 3F**) IAVs compared to KAg-vaccinated animals. In CNPs-KAg-vaccinated pig sera, IAVspecific HI antibody titers against H1N2-OH10 (**Figure 3G**), H1N1-OH7 (**Figure 3H**), and H3N2-OH4 (**Figure 3I**) were significantly higher (*p* < 0.05) compared to mock pig group. The HI titers in CNPs-KAg vaccinates were around twofold higher compared to KAg vaccinates against heterologous (**Figure 3H**) and heterosubtypic (**Figure 3I**) IAVs, but the data were not statistically significant (*p* > 0.05). The expression of Th2-specific transcription factor GATA-3 mRNA in PBMCs of pigs at DPV 35/DPC 0 was 4- and 1.5-fold higher in CNPs-KAg-vaccinated pigs compared to that in unvaccinated control (*p* < 0.05) and KAg-vaccinated pigs (*p* > 0.05) (**Figure 4A**). The expression of Th1-specific transcription factor T-bet in PBMCs was not significantly different among the pig groups (**Figure 4B**).

The mucosal IgA response in pigs post-challenge at DPC 6 was determined, and the data indicate that specific IgA in CNPs-KAg-vaccinated pig group was significantly higher (*p* < 0.05) compared to unvaccinated-challenged animals and remarkably higher compared to KAg-vaccinated and -challenged animals against all three IAV subtypes in nasal swabs (**Figures 5A–C**), BAL fluids (**Figures 5D–F**), and lung lysates (**Figures 5G–I**). These data indicated the secretion of robust mucosal IgA antibody in the upper respiratory tract (nasal swabs), lower respiratory tract (BAL fluids), and lung parenchyma (lung lysates) of pigs. Similarly, systemic IgG antibody response in serum at DPC 6 was also enhanced in the CNPs-KAg vaccinates compared to that in unvaccinated (*p* < 0.05) and KAg-vaccinated and viruschallenged animals (**Figures 6A–C**). However, HI antibody titers in BAL fluid at DPC 6 were comparable between KAg and CNPs-KAg vaccinates (**Figures 6D–F**).

## CNPs-KAg Vaccine Enhanced Systemic-Specific Cell-Mediated Immune Response Against IAVs

To understand the role of chitosan delivered IAV nanovaccine in the induction of specific cell-mediated immune response after IN vaccinations, isolated PBMCs at DPV 35 were restimulated with H1N2-OH10, H1N1-OH7, and H3N2-OH4 viruses. The harvested cell culture supernatants were analyzed for IFNγ secretion, and significantly higher (*p* < 0.05) levels of IFNγ in homologous H1N2-OH10 virus restimulated CNPs-KAg compared to soluble KAg-vaccinated pigs were observed (**Figure 7A**). Though not statistically significant, the IFNγ recall response in heterologous and heterosubtypic viruses restimulated cells was noticeably of higher levels in CNPs-KAg vaccinates than in KAg-vaccinated pigs (**Figures 7B,C**). The average IFNγ amounts in CNPs-KAg vaccine group against H1N1-OH7 and H3N2-OH4 viruses were 463 and 332 pg/mL compared to 91 and 16 pg/mL in KAg vaccinates, respectively (**Figures 7B,C**). We also performed phenotyping of PBMCs isolated at DPC 0 by flow cytometry. The frequency of cytotoxic T cells (CTLs) in CNPs-KAg-vaccinated pigs (average: 18.1%) was higher compared to that in KAg-vaccinated (average: 15.7%, *p* > 0.05) and unvaccinated pigs (average: 13.4%, *p* < 0.05) (**Figure 7D**). This finding is consistent with enhanced IFNγ response in CNPs-KAg-vaccinated pigs, as activated CTLs are one of the major T cell subsets which secrete high levels of antiviral cytokine IFNγ. In addition, in PBMCs at DPC 0, virusspecific cell proliferation was detected in an increased trend upon restimulation with homologous (**Figure 7E**) and heterologous (**Figure 7F**), but not with heterosubtypic (**Figure 7G**) viruses in CNPs-KAg-vaccinated pigs. Overall, these data suggested the

Figure 4 | Expression of T helper 1 (Th1) and T helper 2 (Th2) response inducing specific transcription factors after prime-boost vaccination in pigs. The expression of (A) Th2 transcription factor GATA-3 and (B) Th1 transcription factor T-bet in peripheral blood mononuclear cells of pigs at day post-vaccination (DPV) 35/day post-challenge (DPC) 0 were determined by quantitative reverse transcription PCR (qRT-PCR). Data represent the mean value of three to five pigs ± SEM. Statistical analysis was carried out using Kruskal–Wallis test followed by Dunn's *post hoc* test. Asterisk refers to the statistical significant difference between the indicated two pig groups. \**p* < 0.05.

(C,F,I) influenza A virus (IAVs). Data represent the mean value of three to five pigs ± SEM at all indicated dilutions. Statistical analysis was carried out using Kruskal– Wallis test followed by Dunn's *post hoc* test where A, B, and C refer to significant difference (*p* < 0.05) between unvaccinated vs KAg vaccinates, unvaccinated vs CNPs-KAg vaccinates, and KAg vs CNPs-KAg vaccinates, respectively, at the indicated dilution.

presence of superior cross-reactive effector memory lymphocyte response in pigs induced by chitosan encapsulation of inactivated SwIAV antigen.

### CNPs-KAg Vaccine Reduced the Inflammatory Changes in the Lungs of Virulent and Heterologous Virus-Challenged Pigs

Rectal temperature of pigs was recorded daily post-challenge until euthanized. Pigs in all groups had fever (≥104° F) for the first 2 days after challenge. However, there was no statistical difference in temperature profile among the pig groups (**Figure 8A**). Macroscopic pulmonary lesions were scored for percent consolidation induced by influenza infection and observed lower pulmonary consolidation in CNPs-KAg vaccinates (mean score 15) compared to KAg (mean score 17) and unvaccinated animals (mean score 19) (**Figure 8B**), but the data were not statistically significant (*p* > 0.05). Microscopic pulmonary lesions were subjectively scored on H&E-stained lung sections where a score of 0 = no change from normal, 1 = minimal change from normal, 2 = moderate change from normal, and 3 = severe change from normal (**Figure 8C**). The mean scores of interstitial pneumonia (2, 1.6, and 0.8), peribronchial inflammation (2, 1.8, and 1.8), perivascular inflammation (1.6, 0.3, and 0.5), bronchial exudates (0.7, 0.1, and 0.2), and epithelial changes (0.3, 0.3, and 0.1) were observed in virus-challenged unvaccinated, KAg, and CNPs-KAg vaccinates, respectively. All the microscopic evaluation of pulmonary tissues was conducted by a board-certified veterinary pathologist. A moderate reduction in inflammatory changes was observed in both the

vaccinated pig groups when compared to the lesion scores in the unvaccinated and challenged group. In particular, the interstitial pneumonia and epithelial changes were much reduced in CNPs-KAg group compared to those in soluble KAg-vaccinated pigs.

We also evaluated the levels of pro-inflammatory cytokine IL-6 secretion in the BAL fluid and observed relatively lower levels in CNPs-KAg-vaccinated pigs, consistent with the lower macroscopic and microscopic lung lesions (**Figure 8D**).

### CNPs-KAg Vaccine Enhanced the Mucosal Cellular Immune Response in the Tracheobronchial Lymph Nodes of Virulent and Heterologous Virus-Challenged Pigs

In the CNPs-KAg-vaccinated pigs, the frequency of CTLs, IFNγ, and specific lymphocyte proliferation index values were augmented in PBMCs (**Figure 7**). The cell-mediated immune response in the lung draining TBLN was also examined in these vaccinated pigs. Our data demonstrated a significantly higher (*p* < 0.05) secretion of IFNγ by TBLN-MNCs restimulated with vaccine (H1N2-OH10) and challenge (H1N1-OH7) viruses in CNPs-KAg, but not in KAg-vaccinated compared to mock group (**Figures 9A,B**). Cells similarly stimulated with heterosubtypic (H3N2-OH4) IAV showed an increase in IFNγ secretion in CNPs-KAg-vaccinated pig group but this increase was not statistically significant (**Figure 9C**). We performed flow cytometry analysis of TBLN-MNCs isolated at DPC 6 and observed a significantly higher (*p* < 0.05) frequency of T helper/memory cells (CD3<sup>+</sup>CD4<sup>+</sup>CD8α+), one of the principle contributors of IFNγ production in pigs (18), in CNPs-KAg-vaccinated pig group compared to that in unvaccinated and challenged animals (**Figure 9D**). The expression of Th1 and Th2 transcription factors mRNA level in TBLN collected at DPC 6 was analyzed. Consistent with augmented cellular response in TBLN-MNCs of CNPs-KAgvaccinated pigs, in frozen TBLN tissues mRNA, the expression of the Th1-specific transcription factor T-bet was significantly higher (*p* < 0.05) in CNPs-KAg compared to KAg vaccinates (**Figure 9E**). The expression of Th2 transcription factor GATA-3 mRNA was not increased in TBLN (**Figure 9F**).

## CNPs-KAg Vaccine Reduced Virus Shedding in the Nasal Cavity and Also Pulmonary Viral Titers in SwIAV-Challenged Pigs

We observed a significantly reduced (*p* < 0.05) challenge virus shedding at DPC 4 from the nasal passage of CNPs-KAg vaccinates compared to that of unvaccinated and challenged animals (**Figure 10A**). By DPC 6, infectious virus was detected in the nasal passage of only one of five pigs (20%) vaccinated with CNPs-KAg vaccine, while all pigs in KAg-vaccinated and unvaccinated groups were shedding virus ranging from 102.5 to 103.3 TCID50/mL (**Figure 10B**). The average virus titers in nasal swab at DPC 6 in unvaccinated, KAg-, and CNPs-KAg-vaccinated and -challenged pigs were 102.8, 102.5, and 100.5 TCID50/mL, respectively (**Figure 10B**). Similarly, the virus titer in BAL fluid

on DPC 6 was significantly reduced (*p* < 0.05) in CNPs-KAg but not in KAg group compared to unvaccinated virus challenge pigs (**Figure 10C**). The average virus titers in BAL fluid at DPC 6 in unvaccinated, KAg-, and CNPs-KAg-vaccinated and IAV-challenged pig groups were 106.3, 105 , and 103 TCID50/mL, respectively.

#### DISCUSSION

Chitosan is a natural polymer synthesized by deacetylation of chitin, one of the most abundant polysaccharides in nature (38). Chitosan forms an attractive excipient for drug and vaccine delivery as it bears biocompatible, biodegradable, mucoadhesive, polycationic, and immunomodulatory properties (38, 39). Chitosan is often coupled with TPP, a polyanion that helps in the encapsulation of the biochemical agents through ionotropic gelation. The chitosan and TPP (CS/TPP) NPs formulation in mice were shown to induce both cell-mediated (Th1) and humoral (Th2) immune responses when immunized through IN route against *Streptococcus equi* (40). Similarly, tetanus toxoid loaded in CS/TPP NPs IN delivered in rat was efficiently transported through the nasal epithelium, and in mice, it induced a longlasting systemic and mucosal antibody response compared to soluble antigen (41). Mice immunized through IN route using CS/TPP-based influenza split virus vaccine were shown to induce a higher systemic and mucosal antibody response than soluble antigens and also enhanced the cell-mediated immune response indicated by an increased IFNγ-secreting cell frequency in spleen (25). Unlike the preparation of PLGA and polyanhydride NPs, the process of preparing CNPs does not need any organic solvents and thus involves a simple and mild procedure protecting sensitive biochemical agents including proteins and provides scope for the easy modification of particles (42–45).

In this study, we prepared chitosan-based influenza nanovaccine using TPP by ionotropic gelation technique. The resulting NPs were around 500 nm in diameter which is adequate for efficient uptake by APCs (18, 20, 21, 46). The size of NPs was slightly increased after antigen loading like reported earlier (47). But the

(H&E)-stained lung pictures showing bronchial exudates (dotted black circle), perivascular inflammation (black arrow), peribronchial inflammation (dashed black arrow), and interstitial pneumonia (small black triangle). (D) Secretion of cytokine interleukin (IL-6) in bronchoalveolar lavage (BAL) fluid. Data represent the mean value of three to five pigs ± SEM.

surface charge of our NPs did not change much with or without antigen loading, and the charge (+2.84 mV) was comparable to NPs entrapped with NDV, which was also loaded in CS:TPP at 2:1 ratio formulation like our CNPs-KAg (23). We evaluated the stability of CNPs-KAg NPs suspended in physiological buffer until 30 h maintained at 4° C. For vaccination of pigs, CNPs-KAg was freshly prepared and maintained on ice until delivered IN (1–2 h) which ensured the stability of NPs vaccine.

For better stability and long-term storage of NP vaccines, the surface charge should be highly negative or positive (38, 48). But our CNPs-KAg NPs were polydispersed in nature and had a weak positive surface charge, suggesting that further improvements to our CNPs-KAg formulation are required through optimization of the ratio of CS:TPP:KAg to ensure better physicochemical properties. The optimal CS:TPP:KAg combination should yield a higher positive surface charge, monodispersed nature, relatively smaller size NPs (100–300 nm), and stable for a long time at different storage conditions. The encapsulation efficiency of KAg in chitosan NPs formulation was 67%, higher than the encapsulation efficiency of H1N2-OH10 KAg (~50–55%) in PLGA and polyanhydride NPs (20, 21). The higher encapsulation efficiency of vaccine antigens is desirable to reduce the cost of vaccine production. The protein release from CNPs-KAg was slower than previously reported similar CS/TPP NPs formulation, wherein close to 50% of NDV antigens were released from CNPs within the first 3 days (23). CNP encapsulation enhances the antigen uptake by APCs, increases the expression of activation markers, and secretion of pro-inflammatory cytokines by APCs (49). As expected, SwIAV antigens delivered in chitosan NPs were efficiently internalized by porcine APCs compared to soluble KAg, and importantly, induced the higher production of innate, pro-inflammatory, and Th1 cytokines compared to soluble KAg.

Induction of strong mucosal immunity is associated with an increased breadth of protective efficacy against influenza, and inactivated IM vaccines do not elicit high levels of antigenspecific mucosal IgA antibody response in the respiratory tract (10, 11). Moreover, IM influenza vaccines in pigs have a limitation of not being effective in the presence of maternal-derived antibody (MDA) (50). However, successful IN vaccination has a potential to overcome MDA interference because of the induction of robust local mucosal immunity in the respiratory tract with minimal MDA interference (51). Chitosan is an attractive

(CD3+CD4−CD8α+) in TBLN-MNCs of CNPs-KAg-vaccinated pigs was analyzed by flow cytometry. The expression of Th1 (E) and Th2 (F) transcription factors were also determined in TBLN at DPC 6. Data represent the mean value of three to five pigs ± SEM. Statistical analysis was carried out using Kruskal–Wallis test followed

by Dunn's *post hoc* test. Asterisk refers to the statistical significant difference between the indicated two pig groups (\**p* < 0.05).

polymer for IN immunization (22). It enhances the absorption of vaccine particles across the nasal epithelium (52). Further, when compared to aqueous chitosan solution, insulin-loaded CNPs (300–400 nm diameter) increased the nasal absorption of insulin (53). Due to the positive charge of chitosan, it can interact with anionic components such as sialic acid of glycoproteins on epithelial cell surfaces, thereby prolonging local retention time and decreasing antigen clearance on mucosal surfaces. In addition to its bioadhesive properties, chitosan enhances paracellular and intracellular transport of particulate antigens into the subepidermal space for optimal contact with APCs and other cells associated with immune responses (54, 55). In mice, the IN delivery of CNP-based hepatitis B vaccine enhances the mucosal IgA antibody response (56, 57). Other murine studies have shown that IN immunization with chitosan-based nanovaccine formulations induces robust mucosal and systemic antibody responses against *Pneumococcus* spp., *Diphtheria* spp., and *Bordetella* spp. (58–60).

An influenza subunit vaccine coadministered IN with chitosan delivery system enhanced both mucosal and systemic antibody response in mice (61). The IN delivery of chitosan-delivered DNA vaccine against *Coxsackievirus* in mice enhanced the secretion of both serum IgG and mucosal IgA as well as CTLs activity in spleen (62). Consistent with the previous studies in mice (58–60, 63), the prime-boost vaccination of CNPs-KAg in pigs improved the IgA antibody secretion in the nasal passage and lungs. Importantly, robust secreted antibodies were cross-reactive against heterologous and heterosubtypic IAV and helped in significant reduction in nasal virus shedding and lung load of a heterologous challenge virus. In a previous experiment, PLGA-SwIAV KAg nanovaccine failed to reduce nasal virus shedding in spite of inducing a robust-specific cellmediated immune response and reducing virus load in the lungs of most of the pigs. This anomaly was likely due to the inability of PLGA-encapsulated vaccine to induce mucosal IgA response (20). Similarly, polyanhydride-SwIAV KAg nanovaccine also

enhanced specific cell-mediated immunity but did not enhance mucosal antibody responses and hence did not significantly reduce the nasal virus shedding (21). Like earlier murine studies (64, 65), IN vaccination with CNPs-KAg also induced influenzaspecific systemic IgG antibody and HI titers.

Dunn's *post hoc* test. Asterisk refers to the statistical significant difference

between the indicated two pig groups (\**p* < 0.05).

Cell-mediated immunity is of prime importance for providing complete protection against intracellular pathogens. The Th1 cytokine IFN-γ is a critical cytokine involved in antiviral responses (66, 67). Chitosan is superior to alum adjuvant in enhancing the cell-mediated immune responses (68). It also induces type I IFN secretion from immature DCs which helps in DC maturation and generation of Th1-mediated cellular immune responses (69). In this study, enhanced IFNγ secretion by activated lymphocytes in a recall response with genetically variant IAVs was observed in both PBMCs and TBLN-MNCs of CNPs-KAg-vaccinated pigs. The observed spike in IFNγ recall response was associated with an enhanced virus-specific cellular response both at mucosal sites and systemically. Activated T cell subsets such as T helper and CTLs and innate NK cells are the sources of IFNγ (70). The prime-boost vaccination schedule employed in this study with CNPs-KAg increased the CTLs in PBMC cultures, the major source of IFNγ, the cytokine that helps in clearing virus from infected cells (67).

Another important T cell subset in pigs is T helper/memory cells (CD3+CD4+CD8α+) (71) which possesses cytolytic function and also secretes IFNγ. The protective response against pseudorabies virus infection has been attributed to the increased frequency of T helper/memory cells (71, 72). The frequency of T helper/memory cells in TBLN-MNCs was significantly enhanced in CNPs-KAg-vaccinated pigs. Thus, both T helper/memory and CTLs appear to contribute substantially in improving the crossprotective cellular immune response in pigs vaccinated with chitosan-based influenza nanovaccine.

In conclusion, the mucoadhesive chitosan-based IAV nanovaccine formulation delivered as IN mist augmented cross-reactive T and B lymphocytes response in pigs at both mucosal (upper and lower respiratory tracts and regional lymph nodes—TBLN) and systemic (blood) sites by augmenting secretary IgA, systemic IgG, and T cell responses against highly variant IAVs. This augmented virus-specific cross-reactive immune response resulted in a reduced nasal virus shedding, reduced viral titers in the pulmonary parenchyma, and relatively reduced inflammatory changes in the lungs. Thus, our study indicates that chitosan IAV nanovaccine might be an ideal vaccine candidate against constantly evolving influenza infections in swine herds. Future studies will focus on the optimization of CS:TPP:KAg combination to ensure a monodispersed nature, a higher positive charge, and better stability of CNPs-KAg vaccine. In our future vaccine challenge studies, the efficacy of IN CNPs-KAg vaccine will also be compared with commercial IM-killed and IN-modified live IAV vaccines, and in MDA positive piglets against variant field IAV isolates.

#### ETHICS STATEMENT

This animal study was carried out in strict accordance with the recommendations by Public Health Service Policy, United States Department of Agriculture Regulations, the National Research Council's Guide for the Care and Use of Laboratory Animals, and the Federation of Animal Science Societies' Guide for the Care and Use of Agricultural Animals in Agricultural Research and Teaching. We followed all relevant institutional, state, and federal regulations and policies regarding animal care and use at The Ohio State University. All the pigs were maintained, samples collected, and euthanized, and all efforts were made to minimize the suffering of pigs. This study was carried out in accordance with the approved protocol of the Institutional Animal Care and Use Committee at The Ohio State University (Protocol number 2014A00000099).

#### AUTHOR CONTRIBUTIONS

SD and GR conceived and designed the project. SD, SR, CL, and GR performed the experiments, analyzed the data, and composed the manuscript. Vaccination trial in pigs, sample collection, and laboratory experiments were supported by SG, YL, BH, and NF-R. FL and HH worked on *in vitro* characterization of NPs. SK provided the pathological analysis of the pulmonary tissues of pigs. All the authors provided critical feedback on the manuscript prior to publication and have agreed to the final content.

#### ACKNOWLEDGMENTS

We are thankful to Dr. Juliette Hanson, Megan Strother, Sara Tallmadge, Hyesun Jang, John M. Ngunjiri, and Mohamed Elaish for their help in animal studies. We also thank Dr. Tea Meulia,

#### REFERENCES


Horst Leona, and the team at MCIC, OARDC, for their help in SEM analysis.

#### FUNDING

This work was supported by OARDC OSU industry match grant and Agriculture and Food Research Initiative Competitive Grant No. 2013-67015-20476 from the USDA National Institute of Food and Agriculture.


codelivery of a CpG ODN in alginate coated chitosan nanoparticles. *Eur J Pharm Biopharm* (2008) 69:405–16. doi:10.1016/j.ejpb.2008.01.019


**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 Dhakal, Renu, Ghimire, Shaan Lakshmanappa, Hogshead, Feliciano-Ruiz, Lu, HogenEsch, Krakowka, Lee and Renukaradhya. 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.*

*Young-Seok Kim1,2, Ahyun Son1 , Jihoon Kim1,2, Soon Bin Kwon1,2, Myung Hee Kim3 , Paul Kim1,2, Jieun Kim4 , Young Ho Byun1 , Jemin Sung1,2, Jinhee Lee1,2, Ji Eun Yu1,2, Chan Park1,2, Yeon-Sook Kim5 , Nam-Hyuk Cho6,7, Jun Chang3 and Baik L. Seong1,2\**

*1Department of Biotechnology, College of Life Sciences and Biotechnology, Yonsei University, Seoul, South Korea, 2Vaccine Translational Research Center, Yonsei University, Seoul, South Korea, 3Graduate School of Pharmaceutical Sciences, Ewha Womans University, Seoul, South Korea, 4 Life Science and Biotechnology, Underwood International College, Yonsei University, Seoul, South Korea, 5Division of Infectious Diseases, Department of Internal Medicine, Chungnam National University School of Medicine, Daejeon, South Korea, 6Department of Microbiology and Immunology, Seoul National University College of Medicine, Seoul, South Korea, 7Department of Biomedical Sciences, Seoul National University College of Medicine, Seoul, South Korea*

#### *Edited by:*

*Rajko Reljic, University of London, United Kingdom*

#### *Reviewed by:*

*Pietro Speziale, Università degli studi di Pavia, Italy Rong Hai, University of California, Riverside, United States Jamie Mann, University of Western Ontario, Canada*

> *\*Correspondence: Baik L. Seong blseong@yonsei.ac.kr*

#### *Specialty section:*

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

*Received: 29 December 2017 Accepted: 01 May 2018 Published: 17 May 2018*

#### *Citation:*

*Kim Y-S, Son A, Kim JH, Kwon SB, Kim MH, Kim P, Kim JE, Byun YH, Sung JM, Lee JH, Yu JE, Park C, Kim Y-S, Cho N-H, Chang J and Seong BL (2018) Chaperna-Mediated Assembly of Ferritin-Based Middle East Respiratory Syndrome-Coronavirus Nanoparticles. Front. Immunol. 9:1093. doi: 10.3389/fimmu.2018.01093*

The folding of monomeric antigens and their subsequent assembly into higher ordered structures are crucial for robust and effective production of nanoparticle (NP) vaccines in a timely and reproducible manner. Despite significant advances in *in silico* design and structure-based assembly, most engineered NPs are refractory to soluble expression and fail to assemble as designed, presenting major challenges in the manufacturing process. The failure is due to a lack of understanding of the kinetic pathways and enabling technical platforms to ensure successful folding of the monomer antigens into regular assemblages. Capitalizing on a novel function of RNA as a molecular chaperone (chaperna: chaperone + RNA), we provide a robust protein-folding vehicle that may be implemented to NP assembly in bacterial hosts. The receptor-binding domain (RBD) of Middle East respiratory syndrome-coronavirus (MERS-CoV) was fused with the RNA-interaction domain (RID) and bacterioferritin, and expressed in *Escherichia coli* in a soluble form. Site-specific proteolytic removal of the RID prompted the assemblage of monomers into NPs, which was confirmed by electron microscopy and dynamic light scattering. The mutations that affected the RNA binding to RBD significantly increased the soluble aggregation into amorphous structures, reducing the overall yield of NPs of a defined size. This underscored the RNA-antigen interactions during NP assembly. The sera after mouse immunization effectively interfered with the binding of MERS-CoV RBD to the cellular receptor hDPP4. The results suggest that RNA-binding controls the overall kinetic network of the antigen folding pathway in favor of enhanced assemblage of NPs into highly regular and immunologically relevant conformations. The concentration of the ion Fe2+, salt, and fusion linker also contributed to the assembly *in vitro*, and the stability of the NPs. The kinetic "pace-keeping" role of chaperna in the super molecular assembly of antigen monomers holds promise for the development and delivery of NPs and viruslike particles as recombinant vaccines and for serological detection of viral infections.

Keywords: nanoparticle, virus-like particle, chaperone, ferritin, Middle East respiratory syndrome coronavirus, receptor-binding domain, Lysyl-tRNA synthetase, RNA-interacting domain of human LysRS

## INTRODUCTION

Various types of viral vaccines have been developed over the last century with a wide spectrum of efficacy and safety (1, 2). The manufacturing of most conventional vaccines—live attenuated, inactivated, or subunit vaccines—invariably require the culturing of infectious viruses in cell substrates (3). Despite dedicated efforts, conventional cell culture often fails to produce sufficient amounts of virus for evaluating the immunogenicity, protective efficacy, and safety of viral vaccines. Moreover, some emerging viruses cause high-mortality rates, without options for treatment or prophylaxis, necessitating their manipulation, and manufacture under stringent bio-safety environment (4). Not surprisingly, alternative technologies that circumvent these limitations are a high priority in the areas of vaccine development and production. Nanoparticles (NPs), virus-like particles (VLPs), and assembly of multimeric peptides each provide attractive platforms for vaccine design (5).

Virus-like particles and NPs structurally resemble infectious virions, but are non-infectious due to the lack of viral genomes. Recombinant surface antigens from natural virions are assembled into highly ordered conformations as empty particles devoid of genetic material. Antigenic epitopes are presented on the multivalent and highly repetitive outer structure of the NPs, which leads to the crosslinking of B-cell receptors and the induction of long-lasting immune responses (6–8). By mimicking the morphology of the natural infectious virions, the regularly assembled particles are highly immunogenic, and are amenable to diagnostic and prophylactic exploitation. Among the simplest targets are the VLPs of non-enveloped viruses, such as hepatitis E virus or human papilloma virus, and are composed purely of viral capsid proteins (9–11).

In contrast to non-enveloped viruses, where virion assembly is exclusive to capsid proteins, enveloped viruses (e.g., coronavirus or flavivirus), require an additional membrane component for assembly into mature virions. Consequently, in enveloped VLPs, the assembly of matrix proteins provides a molecular scaffold, and viral antigens are embedded into lipid membranes. Different types of glycoproteins may be embedded in the lipid membrane as target antigens for generating immunological responses (12). However, this process requires multiple proteins (surface antigens and matrix proteins), and the enveloped VLPs are not structurally uniform and are difficult to characterize. A promising alternative is to present the target antigens on the surfaces of self-assembled NPs, which, in lieu of lipid membranes, serve as the macromolecular scaffold for the presentation of the antigens of interest.

In developing NP vaccines, consideration should be given regarding the selection of a robust and faithful system for NP assembly that enables the cost-effective development and delivery of vaccines in a timely manner. Structure-based approaches *in silico* and their underlying principles are relatively advanced for NP assembly (13–15). Most of the approaches consider the thermodynamic stability of the final assembled NPs, without due recognition for the kinetic complexities controlling regular assemblage over random interactions that lead to misfolded aggregations. Therefore, it is not surprising that most engineered NPs are refractory to soluble expression, which presents practical challenges in production, both at a laboratory-scale and in commercial manufacturing processes. This problem becomes augmented when expressed in bacterial hosts because of a lack of folding assistance in the bacterial cytoplasm for viral antigens. Therefore, due to advantages in assisted folding, posttranslational modifications, and the possibility of generating multiple-component NPs and VLPs, eukaryotic hosts such as yeast, insects, and mammalian cells have been favored over bacterial hosts (16–18). However, these systems are significantly more expensive than bacterial systems, are more time-consuming, and the down-stream processes are usually more complex. Moreover, the purification of VLPs from insect cell systems poses a challenge due to similar physicochemical properties between the VLPs and the baculoviruses (1, 16). Bacterial systems, if available, would provide a cost-effective means to develop and deliver vaccines, as well as sero-diagnostic antigen kits used to diagnose-specific infection diseases.

Middle East respiratory syndrome (MERS) was first reported in Saudi Arabia in 2012 and has caused multiple cases of infection with high mortality in Europe and Asia (19, 20). MERS is caused by MERS-coronavirus (MERS-CoV), which can be transmitted from camels to humans, and from humans to other humans (21, 22). Worldwide transmission is increasing in direct household and community-wide transmission, as well as in nosocomial settings, as exemplified in a 2015 outbreak in Korea (23, 24). Neither effective vaccines nor therapeutic interventions are currently available. Because of this, assembly of MERS-CoV antigens into immunologically relevant conformation as NPs would be of interest and may be helpful in developing vaccines, sero-diagnostic tools, and therapeutic monoclonal antibodies.

In the current study, we present a novel bacterial NP of MERS-CoV antigen using ferritin as a molecular scaffold for self-assembly. Ferritin, which is present in most living organisms, has 24 identical subunits that spontaneously self-assemble and form NP complexes with internal and external diameters of 8 and 12 nm, respectively (25, 26). Previous studies show that ferritins of *Helicobacter pylori* from a human isolate can be used as scaffold for HIV and influenza NP vaccines, using eukaryotic host cells such as human embryonic kidney cells (HEK293F or HEK293S) (27, 28). Likewise, bacterioferritin (FR), which self-assembles into nanocages with octahedral symmetry, has also been evaluated as a potential drug delivery system (29). However, viral antigens of human pathogens are prone to misfolding into aggregates, which necessitates chemical refolding of the insoluble aggregates in order to regain solubility and to allow regular assembly of the antigen (30, 31). In addition, displaying antigens on the surface of multi-molecularly assembled scaffolds in bacterial hosts remains a daunting challenge.

We hypothesized that NPs displaying the receptor-binding domain (RBD) of the spike protein from MERS-CoV could be produced in a bacterial system by harnessing the function of a molecular chaperone. Conventionally, protein folding and the prevention of non-functional aggregation have been ascribed to molecular chaperones (32–34). Recently, it has been shown that RNA molecules are able to provide novel functions as molecular chaperones (35–37). Based on novel findings, the concept of chaperna (chaperone + RNA) function was established (38). In this report, chaperna function was harnessed for the folding and assembly of hybrid ferritin monomers into NPs using a bacterial expression system. We also demonstrated that the biophysical properties, including solubility, yield, and stability of MERS-CoV NPs, could be improved by properly controlling the RNA-binding affinity, and the concentrations of Fe2+ and salts. The chapernabased NP assembly may prove to be a versatile tool for developing and delivering recombinant vaccines and for serological detection of emerging/re-emerging viruses.

## MATERIALS AND METHODS

#### Ethics Statement

All animal research was performed according to the guidelines of Ministry of Food and Drug Safety of Republic of Korea. All experiments were approved by the YLARC Institutional Animal Care and Use Committee (IACUC; permit number: IACUC-A-201710-377-01). Six-week-old female BALB/c mice were purchased from ORIENT BIO Inc. (Seoul, Korea). Sera from the recovered MERS patients were used after ethical approval granted by the institutional review boards of Chungnam National University Hospital (IRB no. 2015-08-029) and Seoul National University Hospital (IRB no. 1509-103-705). This study was performed in accordance with the ethical standards laid down in the 1964 declaration of Helsinki and all subsequent revisions. Informed consent was obtained from all patients participated in this study.

#### Construction of Expression Vectors

The expression vector pGE-hRID(3) was constructed from the parental vector pGE-LysRS (3) (39). The pGE-LysRS(3) vector was enzymatically cut with NdeI and KpnI. The PCR product of hRID, which carries the TEV protease cleavage site and a 6-histidine tag at the C-terminus, was cut using the same restriction enzymes and the digested fragment inserted into the vector to generate pGE-hRID(3). FR (Genebank accession No. NC\_000913.3) DNA was synthesized by, and purchased from, COSMO GENETECH (Korea). The DNA was cleaved with SalI and HindIII, and inserted into pGE-hRID(3) to generate hRID(3)-FR. The receptor binding domain (RBD), N-terminal residues 367–606, of the MERS-CoV S protein (GenBank accession No. AFS88936.1), was generated by gene synthesis, cut with KpnI and SalI, and inserted into hRID-FR to generate pGE-hRID(3)-RBD-FR. Linker SSG or ASG was inserted into the C-terminus of the RBD using overlapping PCR, cleaved with KpnI and SaI, and ligated into hRID-FR, generating pGE-hRID(3)-RBD-[SSG]-FR or pGE-hRID(3)-RBD-[ASG]- FR, respectively. The schematic diagrams of each expression vector are illustrated in **Figure 1B**. The genes of mutant hRID(2 m) (K19A and K23A) and hRID(9 m) (K19A, K23A, R24A, K27A, K30A, K31A, K35A, K38A, and K40A) were generated by gene synthesis, cleaved with NdeI and KpnI, and inserted into pGEhRID(3)-RBD-FR, generating pGE-hRID(2 m)-RBD-FR and pGE-hRID(9 m)-RBD-FR, respectively. The mutation sites and amino acid sequences of the mutants are shown in Table S1 in Supplementary Material.

## Protein Expression and Purification

The resulting expression vectors were transformed into the *Escherichia coli* strain SHuffle® T7. The cells were grown in 50 ml of LB medium with ampicillin (50 µg/ml) at 30°C overnight. Each type of transformant was inoculated into 500 ml of LB medium with ampicillin, grown at 30°C until an optical density (OD600) of 0.6–0.8 was reached. Protein expression was induced with 1 mM IPTG for 12 h. Each sample was harvested by centrifugation, lysed by sonication in lysis buffer (50 mM Tris–HCl, pH 7.5; 10% glycerol; 2 mM 2-mercaptoethanol; and 0.1% Tween-20). The soluble fraction of each lysate was purified on a Ni-affinity Histrap™ HP column by ATKA prime (GE Healthcare) and concentrated with Centriprep™ (Merck Millipore Ltd.). The purified proteins were treated with TEV protease to remove the fusion partner hRID. The assembled NPs were purified by gel filtration on 10/300 Superose™ 6 Increase columns (GE Healthcare).

#### Homology Modeling and Trimer Simulation of RBD-FR

A homology model for the fusion complex of MERS-CoV RBD and FR was generated by MODELER, version 9.16 (Sali Lab of California) (13) using data from protein data base (PDB) for MERS-CoV RBD (PDB ID code 4kqz) and bacterial ferritin (PDB ID code 1bcf) as templates. The linker domains were improved using refinements in the loop domain (40). Energystable models of RBD-FR, RBD-[SSG]-FR, RBD-[ASG]-FR, and RBD-[D6]-FR for trimer structure formation were predicted using Multimer Docking software, ClusPro (ABC Group and Structural Bioinformatics Lab Boston University and Stony Brook University). The thermodynamic stabilities were calculated using the Cluspro formula for Coefficient Weights (*E* = 0.40Erep + −0.40Eatt + 600Eelec + 1.00EDARS) (41, 42).

#### Characterization of NPs Using Transmission Electron Microscopy (TEM) and Cryo-Electron Microscopy (cryo-EM)

To examine the size and structure of the purified NPs, microscopic evaluations using TEM and cryo-EM were performed. For TEM analysis, a drop of the NPs was placed onto a formvar/carboncoated TEM grid (SPL). The grid was negatively stained with 2% uranyl acetate, dried, and examined using a JEM-1011 electron microscope (JEOL) at an accelerating voltage of 80 kV. The particle sizes were calculated using Camera-Megaview III (Soft imaging system-Germany) for measuring the NPs in random image fields. For cryo-EM, the NPs were placed onto plasma-treated formvar/ carbon 200 copper grid (EMS) and negatively stained with 2% uranyl acetate. The grid was accelerated at 200 kV with an FEI CryoTecnai F20 cryo-EM microscope made available through the Korean Institute of Science and Technology. The NPs were examined and photographed in high resolution.

## Dynamic Light Scattering (DLS)

Nanoparticle samples (3 ml) were placed into a Dispo-H cell, and analyzed using a Zeta-potential & Particle size Analyzer (ELS-2000ZS; Otsuka Electronics). The intensity distribution diameter

Figure 1 | Design and soluble expression of Middle East respiratory syndrome-coronavirus (MERS-CoV) receptor-binding domain (RBD) nanoparticles (NPs) using the chaperna-based expression platform. (A) Computational modeling of FR, RBD-FR, RBD-[SSG]-FR, and RBD-[ASG]-FR. The hRID domain, RBD, FR, and linkers are colored as blue, green, gray, and red, respectively. (B) Schematic diagram of the expression vector system. The two linkers are shown in red, and tev indicates the TEV protease recognition site. (C) Expression of hRID-RBD-FR in the presence or absence of the hRID fusion partner and linkers. The proteins were expressed at various temperatures (37, 30, and 18°C) and the cell lysates were separated into total (T), soluble (S), and insoluble (P) fractions by centrifugation (left panel). The solubility of each protein expressed at 18°C was measured by a gel densitometer and the data were summarized and shown in the right panel (*n* = 3). Statistical significance (\*\**p* < 0.01, \*\*\**p* < 0.001) was indicated for the samples compared with the control using a two-tailed Student's *t*-test. (D) Illustration of MERS-CoV RBD-FR NPs using the chaperna-based hRID fusion partner. The hRID facilitated folding of the aggregation-prone RBD-FR through interaction with RNA. The monomer of RBD-FR formed a properly folded trimeric structure by cleaving hRID with TEV protease. Eight trimers assembled and formed into MERS-CoV-like NPs. Red triangles indicate the RBD trimer on the FR NPs.

of the NPs was measured twice at 25°C in water as a solvent with the sample accumulation time at 200 s.

## Effect of Salt and Fe2**<sup>+</sup>** Concentrations on NP Assembly and Stability

Cultured cells (3 ml) were lysed with lysis buffer in the presence of various concentrations of NaCl (0, 50, 100, 150, 200, 225, 250, 275, and 300 mM) to evaluate the intracellular proteins. All samples were performed in triplicate. The cell lysates were separated into soluble and insoluble fractions by centrifugation, and the protein stabilities analyzed by sodium dodecyl sulfatepolyacrylamide gel electrophoresis (SDS-PAGE). Thus, the proteins from cell lysates (500 ml culture) were purified using HisPur™ Ni-NTA Resin (Thermo Fisher Scientific) in buffer A depending on NaCl concentration (0–300 mM). To evaluate the effects on Fe2<sup>+</sup> on NP formation, cells were cultured in LB media with various concentrations of Fe2<sup>+</sup> (0, 200, 500, and 1,000 µM). NP formation was examined by size exclusion chromatography (SEC), SDS-PAGE, TEM, and DLS at the various concentrations of NaCl or Fe2<sup>+</sup>.

## Analysis of Protein Stability and RNA Binding

The cells were harvested, sonicated with lysis buffer, and separated into soluble and pellet fractions by centrifugation. Target proteins in the soluble fraction were purified using HisPur™ Ni-NTA Resin (Thermo Fisher Scientific), following the manufacturer's instruction. T (total lysate), S (soluble fraction), P (pellet fraction), W (wash fraction), and E (the elution fraction were analyzed by SDS-PAGE. Co-purification of the nucleic acids and proteins in the wash and elute were analyzed on a native agarose gel. The nucleic acids were visualized with ethidium bromide (EtBr), and the proteins with Coomassie staining.

#### RNase A Treatment

Cultured cells (10 ml) were harvested using the same method described above. The cells were lysed with 500 µl of protein extraction reagent B-PER™ II (Thermo scientific) and separated into soluble and pellet fractions by centrifuged 12,000 rpm for 10 m. A 200 µl aliquot of each soluble fraction was further treated with 250 µg/ml of RNase A (iNtRON Biotechnology) and incubated at 37°C for 15 min. The nuclease treated samples were clarified by centrifugation at 12,000 rpm for 15 min and the soluble supernatants and the pelleted precipitates were analyzed on an SDS-PAGE gel followed by western blot analysis.

## hDPP4-Binding Enzyme-Linked Immunosorbent Assay (ELISA)

To confirm the proper folding of RBD-FR and its variant (RBD- [SSG]-FR), the binding of the purified proteins with the MERS-Cov receptor hddp4 was performed by ELISA. FR only and phosphate-buffered saline (PBS) were used as negative controls. Nunc 96-well microtiter immunoplates (Thermo Fisher Scientific) were coated with 100 ng/well of hDPP4 proteins (Abcam) and incubated at 4°C overnight. The plates were washed and blocked with 150 µl/well of blocking buffer (1% BSA) for 1 h at room temperature. RBD (SSG linker, WT, 2 m, or 9 m)-FR (100 ng/ well) were added for 2 h at 37°C. An anti-penta His antibody (100 µl/well; Qiagen) was serially diluted (1/100 to 1/12,800) in TBST [50 mM Tris–Cl (pH 7.4), 0.05% Tween-20], added to the wells, and incubated for 1 h at 37°C. A secondary goat anti-mouse IgG antibody conjugated with HRP in a 100-µl volume (1:5,000, Sigma-Aldrich) was added and incubated for 1 h at 37°C. The plates were washed three times with TBST at the end of each step. After washing, 100 µl/well of substrate TMB solution (BD Biosciences) were added to the well and the plates were incubated at 37°C for 30 min in the dark. 50 µl of stop solution (2 N H2SO4) was added to the well to stop the colorimetric reaction, and the absorbance at 450 nm was measured using an ELISA reader, FLUOstar OPTIMA (BMG LABTECH).

### Serological Detection of MERS-CoV Infection

Each well of a 96-well microplates (NUNC, Roskilde, Denmark) was coated with 250 ng of four proteins [RBD-[SSG]-FR, RBD-FR(WT), RBD-FR(2 m), and RBD-FR(9 m)] and incubated overnight at 4°C. MERS-CoV RBD protein (MERS-RBD-005P; eEnzyme) was used as a positive control. Following each subsequent step, the wells were washed three times with buffer (0.05% TWEEN-20-PBS; Sigma-Aldrich, St. Louis, MO, USA). The coating antigens were removed, and the wells were blocked with PBST (5% skim milk in PBS and Tween-20) for 1 h at 37°C. After 2 h, the blocking solution was removed. Twofold serially diluted sera from four patients (CNNH-0709, 0809, 1009, and 1309) were added to each well and incubated at 37°C for 2 h. The antigencoated wells were incubated with peroxidase-conjugated goat anti-human IgG antibody (KPL, SeraCare Life Sciences, Milford, MA, USA) at 37°C for 1 h. The primary antibody was removed and 3,3′,5,5′-tetramethylbenzidine (TMB; Sigma-Aldrich) was added to each well as colorimetric substrate. Immediately after treatment of the reactions with stopping solution (Sigma-Aldrich), the OD was read at 450 nm.

#### Mouse Immunization and Sera Collection

Six-week-old female BALB/c mice were immunized with 20 µg/ mouse of the RBD-FR, RBD-[SSG]-FR, or RBD protein generated as described above, or with commercially available MERS-CoV RBD protein (MERS-RBD-005P; eEnzyme) as antigen in BSL-2 facility in YLARC. Antigens were diluted in PBS. For the first group, equal volume of MF59 adjuvant (AddaVax, Cat. No vacadx-10) (43) was mixed by pipetting. For the other group, equal volume of antigens and alum adjuvant (Thermo Fisher Scientific) were mixed by pipetting following the manufacturers' protocol. PBS plus adjuvant and FR were used as negative controls. The immunized mice were boosted twice with intramuscular injections on days 14 and 28. Mice were anesthetized on days 27 and 41 for ocular bleeding from the orbital sinus (Figure S10 in Supplementary Material). Immune sera were processed by centrifugation of the collected blood at 12,000 × *g* for 30 min. The spleen and the BALF (bronchoalveolar lavage) were obtained at 7 days after the last immunization from sacrificed mice. BALF was taken by washing the airways with 1 ml of PBS.

#### Flow Cytometric Analysis

T-cell population from immunized mice were analyzed by Flow cytometric analysis (43, 44). The spleens were taken at 7 days after the last immunization from the sacrificed mice. To obtain single-cell suspensions, the tissues were homogenized and passed through 70 µm cell strainers (SPL). After centrifugation, erythrocytes were removed by Red Blood Cell Lysing Buffer (Sigma). The cells were washed and resuspended in Iscove's Modified Dulbecco's Media containing 10% FBS. For intracellular cytokine staining, the splenocytes were stimulated with 10 µg/ml RBD protein or phorbol myristate acetate/ionomycin in the presence 10 ng/ml recombinant human IL-2 (BioLegend) and Brefeldin A (1:1,000; eBioscience) at 37°C for 5 h. After stimulation, the cells were blocked with rat anti-mouse CD16/CD32 (BD Biosciences) and surface stained with anti-CD8 (FITC, clone 53-6.7; BioLegend) and anti-CD4 (PE/Cy7, clone GK1.5; BioLegend) at 4°C for 30 min. The stained cells were fixed in FACS lysing solution (BD Biosciences) at room temperature for 20 min, and permeabilized with FACS buffer (0.5% FBS, 0.1% NaN3 in PBS) containing 0.5% saponin (Sigma) at room temperature for 15 min. Then, the cells were stained with anti-IFN-γ (PE, clone XMG1.2; BioLegend) and anti-TNF-α (APC, clone MP6-XT22; BioLegend) at room temperature for 40 min. All data were collected by BD LSR Fortessa (BD Biosciences) and analyzed with FlowJo software (Tree Star Inc., Ashland, OR, USA).

#### Competition ELISA Between RBD Protein and hDPP4 Receptor

Competition ELISA was performed to determine whether MERS-CoV antigen [RBD-[SSG]-FR, RBD-FR, RBD, and FR (negative control)]-immunized mouse serum inhibited binding of RBD protein to hDPP4 receptor (45, 46). 500 ng/well hDPP4 protein (Abcam) was coated on Nunc 96-well microtiter immunoplates (Thermo Fisher Scientific) and incubated overnight at 4°C. Plates were washed and blocked with 150 µl/well of blocking buffer [5% skim milk in PBS and Tween-20 (PBST)] for 1 h at 37°C. At the same time, mouse sera immunized with RBD, RBD-[SSG]-FR, RBD-FR, and FR were serially diluted (1/10 to 1/160) with 500 ng/well RBD protein (MERS-RBD-005P; eEnzyme) in TBST [50 mM Tris-Cl (pH 7.4), 0.05% Tween-20], added to new wells, and incubated for 1 h at 37°C. 100 µl solution was added to each well at 37°C and incubated for 2 h. After that, 100 µl of anti-6xHis tag antibody conjugated with horseradish peroxidase (1:1,000, Thermo Fisher Scientific) was added to each well and incubated for 1 h at 37°C. Plates were washed three times with TBST, and 100 µl/well of substrate TMB solution (BD Biosciences) was incubated at 37°C for 30 min in the dark. 50 µl of stop solution (2 N H2SO4) was added to the well to stop the color reaction and measure the absorbance at 450 nm using an ELISA reader FLUOstar OPTIMA (BMG LABTECH).

## RESULTS

#### The hRID Facilitated the Solubility of MERS-CoV RBD-FR

The spike glycoprotein (S) of MERS-CoV was used for the generation of MERS-CoV-like NPs. S protein forms trimers, resulting in large spikes on the virus envelope (47). It is challenging to express the full-sized S protein (~200 kDa) in *E. coli*. Thus, the S1 domain of S protein (~80 kDa), which includes the receptor-binding ability, was used. Our initial attempt to express the S1 domain, either as S1 or as an S1-FR fusion protein, failed; the expression level and solubility of the protein was below the lower limit of detection by SDS-PAGE and western blotting (Figure S1 in Supplementary Material). We therefore used the RBD (367–606 a.a.) of the S1 protein, which has a pivotal function as illustrated in **Figure 1B** (48, 49). When expressed alone in *E. coli*, the RBD is not able to form the trimeric assembly (unpublished observation), due to the lack of the HR2 domain within the S2 domain (50). To overcome this problem, FR was used as scaffold for the assembly. FR is a spherical NP whose subunits form trimers that subsequently result in octahedral structures composed of 24 identical subunits (51). We therefore performed computational modeling to evaluate the potential of FR as scaffold for trimer formation of the RBD.

Possible trimer formation was analyzed by computational modeling using MODELER (13, 52) and ClusPro (41, 42). Various linkers, including SSG, ASG, and D6, were introduced between the RBD and FR with a goal to minimize steric hindrance between the two domains so as to enhance trimer and NP formation. *In silico* analysis showed energy-stable trimeric models of RBD-FR, RBD-[SSG]-FR, and RBD-[ASG]-FR, whereas RBD-D6-FR failed to form a trimeric structure (**Figure 1A**). The RBD- [SSG]-FR was predicted to be the most stable and well-structured compared with RBD-FR and RBD-[ASG]-FR. Initial testing of the RBD-FR constructs without hRID fusion showed that none of the constructs were solubly expressed, even under low-temperature culture conditions (**Figure 1C**) (10.5 and 8.8%, for RBD-FR and RBD-[SSG]-FR, respectively). In addition, the yield of purified RBD-FR from a 2 l culture was only 30 µg of protein. Because of the low-soluble expression of MERS-CoV RBD, we fused hRID to the N-terminus of RBD-FR as a chaperna-based fusion partner (**Figure 1B**). We previously confirmed that by using chaperna, the globular domain of influenza hemagglutinin (HA) is efficiently assembled into a trimeric complex with an immunologically relevant conformation (Yang et al., *in press*). As shown in **Figure 1C**, the hRID fusion significantly increased the solubility of both RBD-FR (59.1%) and RBD-[SSG]-FR (62.83%), indicating that the chaperna platform effectively increased both the solubility and the folding of its fused target proteins. Because of the poor expression level and low solubility of the RBD-[ASG]-FR construct (**Figure 1C**), further experiments were performed using only the RBD-FR and RBD-[SSG]-FR constructs.

### The SSG Linker Increased the Proper Assembly of MERS-CoV RBD NPs

After purification of the soluble proteins (Figure S2 in Supplementary Material), we determined the potential effects of using hRID as a fusion partner for the self-assembly of the NPs. As shown in Figure S3 in Supplementary Material, hRID-RBD-FR failed to form NPs. Because of this, we performed TEV protease cleavage of the hRID. Removal of the hRID domain facilitated the self-assembly of the RBD-FR monomers, and also eliminated the immune response against the non-self hRID domain in BALB/c mice (Figure S4 in Supplementary Material). After hRID cleavage, RBD-FR and RBD-[SSG]-FR were purified using SEC (**Figure 2A**). As expected, RBD-[SSG]-FR assembled into properly formed NPs (1,080 kDa) more efficiently than did RBD-FR NPs, which were mainly detected in the void-volume fractions, suggesting they were irregularly assembled soluble aggregates. The size of the RBD-[SSG]-FR NPs was further confirmed by TEM. TEM images of the RBD-[SSG]-FR NP structures showed hollow, spherical particles that were more compact than the RBD-FR NPs. The average diameter of the RBD-[SSG]-FR NPs was 28–30 nm (**Figure 2B**). In contrast, DLS analysis of the RBD-FR NP structure without the SSG linker appeared to be smaller with an average intensity diameter of 26.3 nm, and this compared with RBD-[SSG]-FR that had an average intensity distribution diameter of 30.5 nm (**Figure 2C**). Consistent with the SEC analysis, RBD-FR without a fusion partner was mostly produced in a soluble aggregated form. Therefore, we identified that the protein folding did not occur properly without hRID, and the formation of NPs was confirmed by both SEC and SDS-PAGE analyses. As shown in Figure S5 in Supplementary Material, the purified NPs retained their stability over an extended period of time at various temperatures (25, 4, and −20°C). Thus, these results indicate that the SSG linker allowed the RBD-FR to generate properly assembled NPs. It should also be noted that the efficiency of protein folding and NPs formation may be further enhance through appropriate linker selection.

## NaCl Concentration Had a Pivotal Role in Assembly of RBD-[SSG]-FR NPs

It has been reported that ionic strength plays an important role in the stability and self-assembly of ferritins (53, 54). We examined the effect of salt concentration on the formation and stability of the RBD-[SSG]-FR NPs at various concentrations (0–300 mM). Consistent with the previous studies, the stability of the protein was highly affected by the concentration of NaCl in the lysis buffer by SDS-PAGE (**Figure 3A**) (*n* = 3). The solubility of the protein significantly decreased as the concentration of NaCl increased from 0 to 100 mM, with the solubility being about 8.79-fold lower at 100 mM compared with 0 mM. Unlike previous studies, the solubility of the protein was gradually recovered at higher NaCl concentrations (>100 mM); the solubility at 300 mM was 1.45-fold higher than at 0 mM. Furthermore, the yield of soluble of protein per liter of culture increased in a salt concentrationdependent manner (**Figure 3B**).

To further investigate the effect of salt concentration, the physicochemical and morphological properties of the RBD-[SSG]-FR protein were examined by SEC, TEM, and DLS. In 50 mM NaCl, most of the protein was aggregated during the purification process, and the purified protein failed to form spherical structures, but instead, existed predominantly as 45 kDa monomers (**Figures 3C,D**; Figure S6 in Supplementary Material). In contrast, the protein that was lysed in 0 mM NaCl and purified in 200 mM NaCl, developed well-structured NPs according to TEM and DLS analyses (**Figures 3C,D**; Figure S6 in Supplementary Material). However, based on SEC analysis, at high-salt concentrations (>250 mM), the protein failed to form stable structures with the proteins being eluted predominantly in the void volume, suggesting they were soluble aggregates under the high-salt concentrations (**Figure 3C**).

Transmission electron microscopy images under the various salt concentrations clearly supported the conclusion, showing that the tendency for aggregation was dependent on the salt concentration (**Figure 3D**). Taken together, the results underscored the importance of salt concentration on the solubility of monomers and the quality of multimeric assembly of hybrid NPs.

## Fe2**<sup>+</sup>** Had an Effect on the NPs Formation and Stability

Ferritin has an intrinsic ability to interact with Fe2<sup>+</sup> to form ferritin-iron cores (55). Thus, it was worth investigating the effect of Fe2<sup>+</sup> on the assembly and stability of RBD-[SSG]-FR NPs. Cells were grown in LB medium with various concentrations of Fe2<sup>+</sup>. As shown in **Figure 4A**, the yield of purified protein was significantly increased from cultures with 500 µM Fe2<sup>+</sup>, reflecting a 2.7-fold increase compared with similar cultures 0 µM Fe2<sup>+</sup>. The cell growth and purification yield at 1,000 µM Fe2<sup>+</sup> were slightly decreased, presumably due to the toxicity of ferric acid. NP formation under the various concentrations of Fe2<sup>+</sup> was analyzed by SEC (**Figure 4B**). Consistent with the previous results, the proteins were eluted mainly in the fractions expected for the size of assembled NPs (1,080 kDa).

Of note, the ratio between NPs and soluble aggregates in the SEC analysis showed that NP formation was facilitated at high concentrations of Fe2<sup>+</sup> (**Figure 4B**). The formation of RBD-[SSG]-FR NPs at an Fe2<sup>+</sup> concentration of 1,000 µM was confirmed by TEM (**Figure 4C**) and DLS (**Figure 4D**). The TEM analysis clearly showed that the morphology of the proteins was more compact, and probably highly stable, when assembled at high Fe2+ concentrations (500 µM) than at lower concentrations (0 µM) (**Figure 4C**). As shown in **Figure 4D**, the average diameter of NPs examined by DLS was 25.1 nm at high Fe2<sup>+</sup> concentration (500–1,000 µM) and 27.7–32.2 nm at lower concentration (0–200 µM). These results suggest that both Fe2<sup>+</sup> and salts concentrations influenced the efficiency and quality of the regular assembly of hybrid ferritin monomers into NPs.

## RNA Binding Was a Key Factor for the Solubility of Hybrid Ferritin

Our previous studies show that an RNA–protein interaction is crucial for transducing the chaperone function of RNA into the folding of client proteins (38). Consistent with that, our present study showed that RNA facilitated the folding of its interacting proteins. The solubility of hRID(WT)-RBD-FR was 5.69-fold higher than RBD-FR without hRID fusion (**Figure 1**), strongly supporting the previous studies. In addition, the solubility of RBD alone was completely insoluble (**Figure 5B**; Figure S7 in Supplementary Material). It has been shown that the positively charged residues of lysine moieties in hRID contribute to tRNA binding (56). In the current study, the tRNA binding induced the intrinsically disordered protein (IDP) status of hRID to form alpha-helical structures (**Figure 5A**). Thus, two RNA-binding

size of the proteins were shown with the arrows. The eluted fractions were analyzed on sodium dodecyl sulfate-polyacrylamide gel electrophoresis (right panel). Red arrows indicate the purified target protein and the separated hRID. (B) Negative-stained transmission electron microscopy images of RBD-[SSG]-FR and RBD-FR. (C) The intensity distribution diameter of RBD-[SSG]-FR and RBD-FR as determined by dynamic light scattering.

mutants, double-mutant hRID (K19A and K23A) and nonamutant hRID (K19A, K23A, R24A, K27A, K30A, K31A, K35A, K38A, and K40A) were constructed (Table S1 in Supplementary Material). The total *E. coli* lysate (T) was fractionated into the soluble fraction (S) and the pellet fraction (P) by centrifugation. As expected, both RBD and RBD-FR without fusion to hRID domain, were refractory to being produced as soluble proteins (**Figure 5B**). Interestingly, the solubility of the RNA-binding mutants did not decrease, but actually increased to 75.3% for the 2 m mutant and 93.4% for the 9 m mutant compared with wild-type protein at 60.1% (**Figure 5B**). Considering that hRID is relatively unstructured in the absence of tRNA binding, the results are consistent with previous reports that the fusion with IDPs promotes the solubility of target proteins (57–59). Following purification of wild-type hRID-RBD-FR (hRID(WT)-RBD-FR), electrophoretic mobility shift assays showed that greater amounts of nucleic acids were co-purified

with hRID(WT)-RBD-FR protein than with the mutant hRID-RBD-FRs (2 and 9 m) under non-denaturing conditions

Each soluble fraction from cell lysates was purified using Hispur™ Ni-NTA Resin.

(**Figure 5C**). The relative ratio of nucleic acid based on EtBr staining and proteins based on Coomassie staining in the eluted fraction confirmed the reduced affinity of mutants to nucleic acids.

To test if RNA had a role in maintaining the stability of the target proteins, the lysates were treated with RNase A to eliminate RNA, and the solubility of each protein was analyzed by SDS-PAGE and western blotting. The soluble fractions of the lysates (S) were incubated at 37°C in the presence and absence of RNase A and the samples were further separated into soluble fraction (SS′) and insoluble fraction (SP′) by centrifugation. As shown in the left panel of **Figure 6**, RNase A treatment completely abolished the effect of RNA on protein solubility as compared with the control (RNase A−) or with samples prior to RNase treatment. Parallel experiments with the 2 and 9 m mutants showed much less RNA co-purified with the proteins, confirming the reduced affinity to nucleic acids and the complete depletion of RNA by RNase A treatment (**Figure 6**, left panel). Remarkably, the solubility of hRID(WT)-RBD-FR was greatly reduced by depletion of RNA as reflected in the ratio of [SS′]/ [SP′] [0.1 and 0.4 for RNase (+) and RNase (−), respectively] by both Coomassie staining and western blot analyses (**Figure 6**, right panel). However, the solubility of the mutants (2 and 9 m), was not significantly affected by RNase A treatment, probably due to their lower affinity to RNA (**Figure 6**, right panel). Taken together, the results demonstrate that hRID(WT)-RBD-FR maintained a strong affinity for RNA, and that affinity was pivotal for maintaining the solubility of the protein.

#### RNA-Binding Facilitated NP Assembly and Prevented Formation of Soluble Aggregates

To further define the RNA dependence of solubility of the ferritin hybrids (**Figure 6**), we investigated if the RNA binding had a role in the formation of NPs. RBD-FR and the various hRID-RBD-FR (WT, 2, and 9 m) proteins were purified by nickel-affinity chromatography (Figure S2 in Supplementary Material) and their physicochemical properties analyzed by SEC (**Figure 7A**), TEM (**Figure 7B**), and DLS (**Figure 7C**). The soluble yields of RBD-FR (hRID fusion) was approximately 1.6 mg/l of culture, representing greater than 1,000-fold higher levels than its hRID (−) counterpart (~15 μg/l culture), again confirming the role of hRID as a robust enhancer for solubility and assembly. It was striking to note that the two mutant proteins, despite high solubility (**Figure 5B**), were detected at disproportionately higher amounts in the void fractions of SEC, indicating that they failed to form NPs of a defined size, and existed predominantly as soluble aggregates (**Figure 7A**). However, hRID(WT)-RBD-FR predominantly formed NPs of a defined size (~1,080 kDa). It is also interesting to note that there was a slight shift of the RNA-binding mutants (2 and 9 m) in the elution pattern, suggesting a larger size of NPs compared with wild-type NPs. Overall, the ratio between soluble aggregates in the void volume and the NPs of defined size clearly showed that RNA binding was crucial for assembly of the monomers into NPs. As a control, RBD-FR (without hRID fusion) existed predominantly as soluble aggregates (Figure S8 in Supplementary Material). Consistent with these results, EM analysis confirmed well-structured NPs by hRID(WT)-RBD-FR, compared to largely aggregated structures by the mutant proteins (**Figure 7B**). Even if multi-molecular structure was formed, the structure becomes unstable, mostly as soluble aggregates. Consistently, the intensity distribution diameter of the wild-type protein, as estimated by DLS analysis, was 25 nm compared with larger sizes of hRID(2 m) at 34.2, 519.2 nm and hRID(9 m) at 52, 717.7 nm (**Figure 7C**; Figure S9 in Supplementary Material). It is conceivable that soluble aggregates may shield the exposed 6-histidine tag, resulting in a decreased binding affinity to Nickel resins and elution in earlier fractions compared with WT protein (Figure S2 in Supplementary Material). Taken together, the data demonstrate that RNA binding prevented aggregation into irregular conformations and guided the self-assembly of the hybrid ferritin monomers into NPs of a stable structure.

### Wild-Type Proteins Showed Strong-Binding Ability to Both the hDPP4 Receptor Protein and MERS-CoV Infected Patient Sera

The immunological properties of ferritin NPs were analyzed by ELISA. The hDDP4 (human DPP4) receptor has been previously identified as the receptor for MERS-CoV human infection (46). Therefore, using hDPP4 as a coating antigen, ELISA-binding assays between RBD NPs and the receptor were performed (**Figure 8**). FR without RBD fusion failed to bind, and was similar to the PBS negative control. Strikingly, the binding ability increased in the same order as the RNA-binding ability (hRID(WT) > 2 m > 9 m), with highest absorbance observed in the WT with the SSG linker (hRID(WT)-RBD-[SSG]-FR). The results show that the conformation of RBD in the WT NPs better resembled the protective antigen of MERS-CoV RBD from 293 cells, compared with the RNA-binding mutants 2 and 9 m. Again, judicious choice of linker between the ferritin carrier and the antigen was important for receptor binding and was reflected in its importance for NP assembly into a stable conformation (**Figure 2**). Finally, the ELISA results for NP against human patients was investigated using the sera from four MERS-CoV-infected patients (**Figure 9**). Six different proteins, including five recombinant NPs (hRID(WT)-RBD-FR, hRID(2 m)-RBD-FR, hRID(9 m)-RBD-FR, hRID(WT)- RBD-[SSG]-FR, and FR), and MERS-CoV RBD protein were compared by ELISA using them as capture antigens. Strong ELISA signals were detected for the four recombinant NPs and MERS-CoV RBD from 293 cells (positive control). The WT form consistently showed a higher response than the RNAbinding mutants (hRID(WT) > 2 m > 9 m), with hRID(WT)- RBD-[SSG]-FR being the best binder among constructs tested. These results address to the utility of the *E. coli* assembled MERS-CoV RBD-FR NPs as useful tools for sero-diagnosis of MERS-CoV infection. Taken together, the results confirmed the immunologically relevant conformation of the MERS-CoV RBD displayed on the hybrid ferritin particles, and the crucial role of RNA in controlling the kinetic pathway for the assembly of viral antigen monomers into stable NPs.

### Immunization of MERS-CoV NPs Induces RBD-Specific Antibodies

To evaluate the immunogenicity of ferritin-based NPs, BALB/c mice (*n* = 5) were immunized with RBD, RBD-FR, and RBD- [SSG]-FR NPs antigens. The tRNAs were found to be removed from the hRID protein during the purification process. Before immunization, potential RNA contamination in the purified proteins was determined by gel electrophoresis. As shown in Figure S11 in Supplementary Material, RNA was below detection level, if any, after several purification steps, compared with the proteins purified in the first step. Previously, MF59-adjuvated and alum-adjuvated MERS-CoV antigen have been reported to increase the antibody and T-cell responses in mice (44, 60). Thus, the first group and second group were immunized twice with 20.0 µg of antigen containing the equal volume of alum

western blot (right, lower panel).

and MF59, respectively (Figure S10 in Supplementary Material). After collecting serum, the RBD-specific IgG titer was analyzed by ELISA. Alum adjuvanted RBD-[SSG]-FR NPs and RBD-FR NPs induced 14-fold and 16-fold higher (*p* < 0.0001) than RBD (**Figure 10A**). The antibody titers of MF59 adjuvanted RBD-[SSG]-FR NPs and RBD-FR NPs were 4.5- and 3-fold

Microscopy (B) and dynamic light scattering (C).

higher than RBD, respectively. The antibody responses by RBD-FR and RBD-[SSG]-FR NPs were much stronger than the RBD in all antibody subtypes tested (IgG1, IgG2a, and IgG2b)

binding domain (RBD)-FR bound with hDPP4 receptor protein. Each purified protein shown in Figure S2 in Supplementary Material and size exclusion chromatography was used to explore hDPP4 receptor-binding affinity to the protein. All data are shown as mean ± SD from triplicate samples. FR alone and phosphate-buffered saline were used as negative controls.

(**Figures 10B–D**). As a test of mucosal immune responses, the RBD-specific IgA antibody levels from BALF were also analyzed by ELISA (**Figure 10E**). MF59 adjuvanted RBD-[SSG]-FR NPs presented significantly higher OD values than RBD and FR (negative control). These results suggested that RBD-[SSG]-FR NPs induces local mucosal immune response stronger than RBD. In addition, it was confirmed that antibody responses of IgG, IgG1 (Th1), IgG2a, and IgG2b (Th2) against MF59-adjuvated antigens were higher than those from alum-adjuvated antigens. In contrast, PBS and FR control groups failed to, or only weakly induce an antibody response against RBD protein. These results suggest that FR-based NPs significantly enhance various antibody responses than monomeric antigens.

#### Specific Cellular Immune Responses Were Induced by MERS-CoV NPs

The cellular immune responses were investigated in mice immunized with protein (RBD, RBD-FR, RBD-[SSG]-FR) and FR (negative control). Splenocytes of mice (*n* = 4) were harvested 1 week after the last immunization, stimulated with

Figure 9 | Sero-diagnosis of Middle East respiratory syndrome-coronavirus-infected patients using receptor-binding domain (RBD)-FR nanoparticles. The purified proteins shown in Figure S2 in Supplementary Material were used as coating antigens. FR alone and infected cell lysates were used as negative and positive controls, respectively. Virus-infected sera from four patients were serially diluted from 1:100 (twofold dilution). All data are presented as mean ± SD of duplicate samples.

Figure 10 | Immune responses in receptor-binding domain (RBD) nanoparticles (NPs) immunized mice (*n* = 5). Endpoint titer of IgG (A), IgG1 (B), IgG2a (C), and IgG2b (D) antibody binding to Middle East respiratory syndrome-coronavirus RBD were detected using mice serum after two immunizations. RBD-specific antibodies were detected after immunizations of RBD NPs, RBD, FR with adjuvant (alum and MF59) using enzyme-linked immunosorbent assay. (E). RBD-specific IgA antibodies were detected using BALF (diluted 1:8) after immunization of protein with MF59. OD, optical density. Each endpoint titer was shown by individual. All error bars were shown as mean ± SD (*n* = 5) and all *p*-values were obtained using Student's two-tailed tests (\*\**p* < 0.01, \*\*\**p* < 0.001).

RBD protein, and analyzed for cytokines by flow cytometry. In the RBD-immunized group, IFN-γ and TNF-α-producing CD4<sup>+</sup> T-cell responses were detected at low levels. However, IFN-γ and TNF-α-producing CD4<sup>+</sup> T cells were significantly increased in RBD-FR and RBD-[SSG]-FR-immunized groups compared with RBD and FR-immunized group (Figure S12 in Supplementary Material). These results demonstrated that the RBD NPs vaccination induced antigen-specific CD4<sup>+</sup> T cells that produced IFN-γ and TNF-α upon antigen stimulation.

### Anti-NPs Serum Effectively Blocked RBD Protein Binding to the hDPP4 Receptor

Middle East respiratory syndrome-coronavirus infection is mediated by the interaction of RBD and the host receptor hDPP4 (45, 46). As a correlate of protection, a competition ELISA was performed to investigate whether antibodies generated from NPs immunization were able to interfere with the binding to hDPP4. Thus, after incubation of RBD protein with mouse serum (1:10), the binding of serum-mixed samples to hDPP4 protein was measured. As shown in **Figure 11**, RBD-[SSG]-FR, RBD-FR, and RBD-immunized sera strongly abolished the binding of RBD to hDPP4 receptor (93.3, 82.2, and 75.67%, respectively). Interestingly, the relative efficiency of interference correlates with that of NP assemblage (**Figure 11**). In contrast, the FR-immunized mouse serum (negative control) failed to inhibit the interaction. Taking together, these results demonstrate that immunization of NPs greatly stimulates MERS-CoVspecific antibody response that effectively interferes with the cellular receptor binding, suggesting its possibility as a vaccine. However, protection efficacy should ultimately be tested in a live virus challenge model.

#### DISCUSSION

Having key immunologic features, like a highly repetitive nanostructure, provides a designing principle for NPs in inducing potent and long-lasting antibody responses. For VLPs of non-enveloped viruses, assembly is made purely by capsid proteins. For enveloped viruses, however, additional membrane components and matrix proteins are required to display the target antigens on the surface of assembled VLPs. A promising alternative is to present target antigen on the surfaces of selfassembled NPs, which, in lieu of lipid membranes and matrix proteins, serve as a macromolecular scaffold for the presentation of antigens of interest (61). Ferritins, as a substitute for matrix proteins and membranes, have been used as scaffold for the regular assembly of target antigens. However, ferritin-based NPs have been produced only in host cells of mammalian or insect origin (28, 62). Previously, we showed that influenza HA

could be assembled in a soluble, trimeric, and immunologically relevant conformation by exploiting chaperna activity (63). The present study is the first report of using RNAs as molecular chaperone for supra-molecular structures. Here, we present a novel bacterial system for NP assembly of hybrid ferritin displayed surface antigens from MERS-CoV. The NPs reacted strongly with sera derived from MERS-CoV-infected patients (**Figure 9**) confirming their utility in sero-diagnosis of infection. Moreover, the antisera, generated from immunization of mice, were able to interfere with the binding to the cellular receptor hDPP4 (**Figure 8**), in part of essential protective immune responses. The efficiency of receptor-binding inhibition (**Figure 11**), as well as the ability for inducing the mucosal responses (**Figure 10E**), correlated with the regular assembly of NPs as examined by DLS or EM (**Figure 2**), confirming that presentation of antigenic epitopes on a multivalent and highly repetitive structure is indeed important for the quality of immune responses. Overall, the quality of NPs and consequent immune responses were governed by the RNA-mediated assembly of antigens.

We hypothesized that chaperna function could be harnessed for presenting target antigens as highly repetitive nanostructures (**Figure 1D**). The hRID is the N-terminal domain of hLysRS and was previously identified as a nucleic acid-binding domain (**Figure 5**) (63). In this report, the hRID was exploited as a transducer for chaperna function (TCF) by serving as a docking-tag for cellular RNA for the folding/assembly of the hybrid FR containing client antigen proteins [RBD of MERS-CoV (**Figure 1D**)]. The advantage of using hRID as a TCF could be many fold. First, hRID is small (8.3 kDa), monomeric, and was flexible enough to allow the access of site-specific protease for the removal of hRID (Figure S3 in Supplementary Material). Of note, hRID belongs to IDPs, which switches into stabile α-helixes upon binding with tRNAs. Second, the bound RNA, due to its highly negative charge, may resist uncontrolled intermolecular interactions among monomers into amorphous aggregation. Finally, even the naked hRID (in the absence of RNA binding), due to its intrinsically flexible nature, may not pose physical hindrance to multiple interactions among monomers, enabling assembly into stable super-structures, upon removal of the hRID. Thus, the potential "pace-making" function harnessed with the RNA molecule, allows a regular assembly of monomers as highly repetitive nanostructures. Consequently, in the current study, hybrid FR was produced in soluble forms, could be purified by one-step affinity chromatography, and most remarkably, assembled into NPs of defined sizes upon removal of the hRID (Figure S2 in Supplementary Material). Consistent with the principles of design, the loss of RNA binding by hRID significantly hampered the regular assembly of the ferritin monomers and increased the amount of non-functional misfolded proteins as soluble aggregates (**Figure 7**). Thus, the overall yield, as well as the quality of NPs, were dependent on the chaperna function transduced by the hRID, which in turn was mediated by interaction with cellular RNAs (likely to be tRNAs).

The driving and controlling factors for *de novo* assembly of biomolecules are poorly understood. Historically, host factors like GroEL/S were initially discovered as molecular chaperones for supporting viral growth in *E. coli* and supporting the assembly of viral capsid proteins (64, 65). Moreover, GroEL/S also cooperates with RbcX in plant cells for the assembly of multicomponent RUBISCO, which is the most abundant protein in the biosphere responsible for photosynthesis (66). Therefore, it is intriguing that RNA could provide such a robust folding/ assembly of a supra-molecular structure. We recently confirmed that the present strategy could be successfully applied to the assembly of bacterially synthesized monomers of norovirus into VLPs composed of 180 monomers (unpublished observation, Seong, B.L.). Whether RNA can substitute for, or collaborate with pre-existing protein-based molecular chaperones remains an exciting avenue for future investigations. It should be noted that the defined versatile functions are being expanded for RNA molecules. As an engineered system for harnessing chaperna function, the present report may prove to be the tip of an iceberg for pivotal function of RNA molecules as chaperones for the folding and supra-molecular assembly of proteins in living organisms (36, 38).

Various factors were identified as important for efficient assembly of MERS-CoV NPs. As an extrinsic factor, the binding affinity of hRID to cellular RNAs was crucial for the assembly and the quality of the assembled NPs (**Figure 7**). As intrinsic factors, the concentration of salts and Fe2<sup>+</sup> also influenced the assembly and stability of NPs (**Figures 3** and **4**). The ionic strength played an important role in the stability and self-assembly of ferritins, and aggregation increased with increasing concentrations of NaCl (54). The assembly of the hybrid MERS-CoV NPs revealed an interesting change in salt dependence, with 200–225 mM NaCl buffer as optimal condition as confirmed by EM and DLS analyses (**Figure 3**). The change in salt dependence was probably due to the presence of electrostatic interactions among RBD domains (54, 67). The dependence on Fe2<sup>+</sup> was not surprising considering that ferritin has an intrinsic ability to interact with Fe2<sup>+</sup> to form ferritin-iron cores (55). Based on our experience, to enhance the quality of NPs, it is advisable to control Fe2<sup>+</sup> concentrations, both during the culturing of the bacterial cells and during the purification of the soluble monomer proteins (**Figure 4**). First, the yield of the purified protein was increased in the presence of 500 µM Fe2<sup>+</sup> (**Figure 4B**), up to 2.7-fold greater compared with the control conditions lacking Fe2<sup>+</sup>. Second, the ratio between NPs and soluble aggregates in SEC showed that NPs formation was facilitated at high concentrations of Fe2<sup>+</sup>, and resulted in a more compact morphology under EM (**Figures 4B,C**). Thus, both the overall yield and the quality of NPs were governed by their intrinsic ability to interact with Fe2<sup>+</sup>. Finally, our data show that the presence and the nature of the linker between the ferritin and the RBD antigen was also important to the assembly of NPs. It is possible that a linker with flexibility and sufficient length would accommodate the steric requirements for assembly of multimeric NPs. However, it is difficult to precisely predict the effect of the linker, and therefore it is advisable to screen multiple constructs during the early stages of testing the assembly of NPs displaying antigens of interest.

In conclusion, the chaperna-based antigen assembly platform holds promise for the development and delivery of NP-based vaccines to enhance RBD-specific antibody responses, and the serological detection of emerging viruses. Various types of designing principles have advanced the structure-based approaches to NP assembly (61, 68). However, most of the *in silico* methods consider the thermodynamic stability of the final assembled NPs, but not necessarily the kinetic pathways leading to their successful folding into regular assemblages. Consequently, most NPs are refractory to soluble expression and fail to assemble as designed, resulting in significant, and practical challenges in the manufacturing process. The chaperna-mediated folding and the "pace-keeping" assembly of monomers into higher ordered structures will enable faithful production of NP and VLP-based vaccines against emerging and re-emerging viral infections.

#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of "Ministry of Food and Drug Safety (MFDS) of Republic of Korea." The protocol was approved by the "YLARC Institutional Animal Care and Use Committee (IACUC; permit number: IACUC-A-201710-377-01)."

#### AUTHOR CONTRIBUTIONS

Y-SK and BS designed the research study. Y-SK, AS, and BS wrote the manuscript. Y-SK performed overall experiments. JHK, JEK, SK, YB, and JS conducted the research study. JL and JY provided technical assistance and vector. PK developed the necessary software for performing analyzing computer data. CP provide technical assistance to conduct RNase A treatment experiments. YK and NC provide patients sera infected with MERS-CoV. MK and JC analyzed cellular immune response by flow cytometry.

#### ACKNOWLEDGMENTS

We thank J. I. Kim and M.-S. Park at Korea University Medical School for providing information on sero-diagnosis of MERS-CoV-infected patient. We thank Y. H. Jang for information on cellular immune analyses.

#### REFERENCES


#### FUNDING

This study was supported by grants from the Ministry of Health & Welfare, Republic of Korea (grant numbers HI13C0826 and HI15C2934) and the National Research Foundation of Korea (2014M3A9E4064743).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at https://www.frontiersin.org/articles/10.3389/fimmu.2018.01093/ full#supplementary-material.

a baculovirus system compared with a vaccinia virus system. *Biotechnol Appl Biochem* (2000) 32(1):27–33. doi:10.1042/BA20000001


**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 Kim, Son, Kim, Kwon, Kim, Kim, Kim, Byun, Sung, Lee, Yu, Park, Kim, Cho, Chang and Seong. 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.*

# Nanoparticle Vaccines Against Infectious Diseases

#### Rashmirekha Pati <sup>1</sup> , Maxim Shevtsov 2,3,4 and Avinash Sonawane1,5 \*

*<sup>1</sup> School of Biotechnology, KIIT University, Bhubaneswar, India, <sup>2</sup> Institute of Cytology of the Russian Academy of Sciences (RAS), St. Petersburg, Russia, <sup>3</sup> Klinikum Rechts der Isar, Technical University of Munich, Munich, Germany, <sup>4</sup> First Pavlov State Medical University of St.Petersburg, St. Petersburg, Russia, <sup>5</sup> Discipline of Biosciences and Biomedical Engineering, Indian Institute of Technology Indore, Indore, India*

Due to emergence of new variants of pathogenic micro-organisms the treatment and immunization of infectious diseases have become a great challenge in the past few years. In the context of vaccine development remarkable efforts have been made to develop new vaccines and also to improve the efficacy of existing vaccines against specific diseases. To date, some vaccines are developed from protein subunits or killed pathogens, whilst several vaccines are based on live-attenuated organisms, which carry the risk of regaining their pathogenicity under certain immunocompromised conditions. To avoid this, the development of risk-free effective vaccines in conjunction with adequate delivery systems are considered as an imperative need to obtain desired humoral and cell-mediated immunity against infectious diseases. In the last several years, the use of nanoparticle-based vaccines has received a great attention to improve vaccine efficacy, immunization strategies, and targeted delivery to achieve desired immune responses at the cellular level. To improve vaccine efficacy, these nanocarriers should protect the antigens from premature proteolytic degradation, facilitate antigen uptake and processing by antigen presenting cells, control release, and should be safe for human use. Nanocarriers composed of lipids, proteins, metals or polymers have already been used to attain some of these attributes. In this context, several physico-chemical properties of nanoparticles play an important role in the determination of vaccine efficacy. This review article focuses on the applications of nanocarrier-based vaccine formulations and the strategies used for the functionalization of nanoparticles to accomplish efficient delivery of vaccines in order to induce desired host immunity against infectious diseases.

#### Edited by:

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

#### Reviewed by:

*Randy A. Albrecht, Icahn School of Medicine at Mount Sinai, United States Michael Schotsaert, Icahn School of Medicine at Mount Sinai, United States Katie Louise Flanagan, RMIT University, Australia*

> \*Correspondence: *Avinash Sonawane asonawane@iiti.ac.in*

#### Specialty section:

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

> Received: *13 April 2018* Accepted: *07 September 2018* Published: *04 October 2018*

#### Citation:

*Pati R, Shevtsov M and Sonawane A (2018) Nanoparticle Vaccines Against Infectious Diseases. Front. Immunol. 9:2224. doi: 10.3389/fimmu.2018.02224* Keywords: nanoparticles, vaccine development, human diseases, targeted vaccine delivery, antigens

## INTRODUCTION

In twenty-first Century, infectious diseases have emerged as a serious threat to the health of millions of people across the globe (1)**.** According to the World Health Organization (WHO) report for 2016, ∼3.2 million deaths have occurred due to lower respiratory infections and 1.4 million from tuberculosis alone worldwide (2). Over the past few decades, many new infectious diseases have emerged and few old diseases re-emerged, which were once considered to be no longer a threat to the human being (3–5). Collectively, these diseases account for millions of deaths that cause enormous impact on the global socio-economical and health-care sectors. The major challenges to combat such diseases are that for many of them, there are no effective drugs available. One of the plausible approaches could be based on the application of nanocarrier based vaccination (6)**.** However, there are still no effective vaccines available against some of the most prevalent diseases including immune deficiency syndrome (AIDS) and tuberculosis. This underlines an urgent need for the development of desired vaccines against these diseases. Some of the important aspects of any optimal vaccine includes (i) safety, (ii) stability, and (iii) the ability to elicit durable and adequate immune response with a minimum number of doses (7–9). Presently, different generation vaccines such as attenuated or killed whole organisms (first generation), subunit (second generation) and RNA or DNA vaccines (third generation) are used to elicit protective immunity against diseases (10–12). Despite several advantages of RNA or DNA vaccines such as minimal risk of infection, ability to elicit immune response against specific pathogen and cost effective (13); there are a number of challenges associated with the efficient delivery of these vaccine molecules to the target sites and the requirement of the prime-boost vaccination regimens with other immunogenic agents. These includes premature degradation of molecules and the inability to translate into a functional immunogen (14). Similarly, protein based vaccines are used successfully against several infectious diseases such as Haemophilus influenza type b, diphtheria, tetanus, acellular pertusis, meningococcus and pneumococcus (15), however they require an adjuvant to potentiate their immunogenicity, and also encounter early degradation after exposure to hostile milieu. Introduced recombinant proteinbased vaccines (e.g., recombinant hemagglutinin vaccine for influenza) further enhance the immunity toward infection indicating the applicability of the recombinant technology for the vaccine production (16). To overcome these hurdles, an efficient vaccine delivery system is required which not only delivers the vaccine molecules to the target site to evoke enduring immune responses but also has minimal side effects and requires less doses. Moreover, there is an increasing need to develop new generation composite vaccine molecules that will act as immunogen as well as an adjuvant. Nanotechnology based formulations offer numerous advantages for the development of new generation vaccines. Nanocarrier based delivery system can protect the vaccines from premature degradation, improve stability, has good adjuvant properties, and also assists in targeted delivery of an immunogen to the antigen presenting cells (APCs). There are several mechanisms by which vaccines can be delivered to the specific sites using nanocarriers. Vaccine antigens can be encapsulated within the nanocarriers or decorated on their surface (**Figure 1**). Encapsulation within the nanoparticles (NPs) can protect the antigen from premature protease degradation and elicit sustainable release, whereas the surface adsorption facilitates their interaction with cognate surface receptors such as toll like receptors (TLRs) of APCs (17). Nanocarrier based delivery systems provide a suitable route of administration of vaccine molecules and enhance cellular uptake thereby resulting in robust innate, humoral, cellular as well as mucosal immune responses when compared with unconjugated antigens. This review mainly focuses on the potential use of nano delivery systems as novel vaccine strategies for the induction of innate as well as adaptive immune responses against infectious diseases.

## KEY CELLULAR COMPONENTS OF THE IMMUNE SYSTEM

The immune system is composed of a collection of mobile cells that traffic throughout the body as well as reside at the site of entry (i.e., skin, respiratory, gastrointestinal, and genital tracts) in search of invading pathogens. These cells belong to two major types of innate and adaptive immune system. The innate immune cells like macrophages and neutrophils rapidly respond to the pathogens by recognizing pathogen surface moieties, phagocytosis, and the elimination of pathogens through activation of different antibacterial effector functions. Similarly, two major components of the adaptive immunity i.e., T and B-cells are important for the generation of cell mediated and humoral immune responses, respectively. T cells including CD4<sup>+</sup> helper T cells secrete different cytokines to modulate the functions of B cells, whereas CD8<sup>+</sup> T cells recognize and destroy virally infected cells. Antibodies produced by the B cells can further neutralize the invading microbes or clear infected cell or opsonized pathogens through cell-mediated systems. APCs, in particular dendritic cells (DCs) and macrophages, migrate through the body to sample, process and present the antigens to T-cells to activate cellular immune responses. These cells express various surface receptors to recognize cognate ligands and danger signals to trigger activation of different signaling pathways that eventually lead to the activation of T-cells (18). After sampling the antigens, DCs migrate from the peripheral tissues into the draining lymph nodes to activate naive T-cells (19), whereas macrophages after ingestion of antigens increase their lysosomal degradative machinery to enhance the antigen presentation to activate helper T cells.

## TYPES OF NANO-IMMUNO ACTIVATORS

Some NPs are themselves able to stimulate different immune cells to boost the host immunity. The size, shape and surface chemistry of NPs (described below in more detail) are important factors that determine their potential to activate immune responses. In general, NPs are able to stimulate immune reactions by increasing the synthesis of defense genes and inflammatory reactions (20). Various types of NPs like gold, carbon, dendrimers, polymers and liposomes have the capability to induce cytokine and antibody responses (21–26). This was observed in the case of administration of empty PEGylated liposomes, which were able to elicit IgM response in an in-vivo model. (27, 28). Besides their potential to deliver various immune stimulators to the specific sites as well as into the deep tissues where vaccine molecules alone may not able to reach, these NPs have also been exploited as adjuvants to augment immunogenicity of vaccine candidates. Nano-immuno stimulators are the nano scale (20– 100 nm) vaccine particles that can improve the vaccine efficacy in vivo better than bulk molecules (20, 29). Some of the known nano-immuno stimulators that have been used for this specific purpose are inorganic NPs (iron and silica) (30, 31), polymeric NPs (chitosan, PLGA, PVPONAlk, γ-PGA) (32–37), liposomes (cholesterol and lipids) (33, 38) and virus like particles (VLPs) (39, 40). Different types of NPs used to deliver antigens to give protection against different diseases have been listed in **Table 1**.

## Inorganic NPs

Some biocompatible inorganic NPs such as gold, carbon and silica have been exploited in the vaccine delivery studies (50, 79–81). These NPs can be synthesized in various shapes, size and surface modified forms. Some of the viral antigens were successfully delivered using inorganic NPs as carriers. This caused increase in antigen stability by protecting them from premature degradation by proteolytic enzymes. Delivery of viral and bacterial antigens using gold NPs was also found to induce quite robust host immune responses against influenza, immunodeficiency virus, foot and mouth, and tuberculosis diseases in mice (51, 52, 82, 83). Encapsulation of plasmid DNA that encode mycobacterial hsp65 antigen in gold NPs exhibited significant reduction in the Mycobacterium tuberculosis, causative agent of human tuberculosis, burden in infected mice (52, 82). Few studies have used hollow mesosporous silica, nanotube and spherical forms of carbon NPs as adjuvants to improve the immunogenicity and delivery of protein and peptide antigens against viral infections (79, 83, 84). Silica based NPs contain abundant silanol groups that can be utilized to introduce specific functional groups on their surface to gain access for vaccine molecules into target cells (84–86). The major advantages of inorganic NPs include low production cost, reproducibility and safety in application.

#### Polymeric NPs

In recent years, polymeric NPs have received great attention for their applications in the delivery of a number of vaccines. This is primarily due to their ease in preparation, biodegradability, biocompatibility, reduced cytotoxicity, and the possibility to fine-tune surface properties as needed (87). Moreover, it is relatively easy to control the rate of vaccine release by altering the composition or ratio of co-polymers during the NP synthesis process (87). The most commonly used polymeric NPs for vaccine delivery are poly (lactic-co-glycolic acid; PLGA) or poly (lactic acid; PLA). PLGA NPs have already been tried in the delivery of a broad range of antigens, including hydrophobic antigens (34, 35), hepatitis-B virus antigens (54), Bacillus anthracis (41), tetanus toxoid (35), and ovalbumin (88). The use of PLGA conjugated antigens exhibited strong immunostimulatory property by inducing cytokine and nitric oxide production against mycobacteria infection (89). In addition to synthetic polymers, some natural biopolymers such as alginate, pullans, inulins, and chitosan have been used as adjuvants (90–93). Inulin, a known activator of the complement cascade (94), conferred better protection against hepatitis B and influenza viruses (92, 93). Similarly, chitosan NPs were demonstrated as nanocarrier molecules for HBV antigens (55), DNA vaccine (56), and Newcastle disease vaccine (42). The delivery of PLGA and chitosan NP conjugated vaccine molecules enhanced the immune responses at the mucosal site (95, 96). Our recent study also showed that delivery of M. tuberculosis lipids using biocompatible chitosan NPs was able to induce significant humoral as well as cellular immune responses when compared to lipids alone in mice (43). We also found that intraperitoneal administration of these conjugates showed better activation of splenic T-cells. Another study by de Titta et al. has shown that intradermal administration of CpG conjugated polymeric NPs increased dendritic cell activation by several fold, exhibited comparable vaccine efficacy at ∼400 times lower dose, and also caused enduring cellular immunity in comparison to free CpG (97). These desired properties along with already known reduced toxicity and biocompatibility under both in vitro and in vivo conditions make polymeric NPs plausible candidates for further preclinical pharmacokinetics and therapeutic applications (98).

#### Liposomes

In addition to polymeric NPs, liposomes are the second most widely explored vaccine and drug delivery vehicle in the nanomedicine field. The synthesis of liposomes is a spontaneous process, where hydration of lipids enables the lipid bilayer formation around an aqueous core (99). So far, different types of liposomes, including unilamellar or multilamellar vesicles composed of biodegradable phospholipids (e.g., phosphatidylserine, phosphatidylcholin and cholesterol) were included in the vaccine studies (100). Liposomes deliver vaccines by fusion with the target cell membrane (101).The structurally flexible and versatile liposomes are able to encapsulate both hydrophilic and hydrophobic substances. The hydrophilic molecules can be incorporated into the aqueous core, while hydrophobic molecules are encased within the phospholipid bilayer. Earlier reports have shown that delivery of antigenic proteins entrapped in multilamellar lipid vesicles elicit strong T and B-cell responses (102). Similarly antigenic peptides conjugated to phosphatidylserine (PS)-liposomes were readily TABLE 1 | List of antigens delivered by using different nanocarriers for the treatment of different diseases.


internalized by APCs to potentiate T-helper cell mediated immune responses (103) and delivery of heat shock protein encoding vaccine DNA using liposomes elicited strong protective immunity against fungal infection (104). Because of their foreseen applications, several liposome based vaccine nano-formulations have been approved for clinical trials against intracellular pathogens, including viruses and M. tuberculosis (105). One such study already demonstrated the potency of liposomal aerosol carriers in the generation of protective immunity against M. tuberculosis infection (106, 107). Other studies have tried a combination of dimethyl dioctadecyl ammonium (DDA) lipid based liposomes and various immunomodulators to enhance immunity against influenza, chlamydia, erythrocytic-stage malaria, and tuberculosis infections (108–112). In the context of DNA vaccines, lipid-DNA complexes have been successfully delivered to the lungs of monkeys (101).

#### VLPs (Virus Like Particles)

There are several reports that adequately proved applications of VLPs as a vaccine carrier, and also their ability to stimulate the host immune responses (113–115). VLPs are composed of selfassembled viral membrane that forms a monomeric complex displaying a high density of epitopes (115, 116). Interestingly, VLPs can also be engineered to express additional proteins either by fusion of proteins with the particles or by endogenous expression of multiple antigens (113, 117). It is also possible to chemically couple non-protein antigens and small organic molecules onto the viral surface to produce bioconjugates with VLPs (118, 119). Due to these distinct features, VLPs can provide protection not only against virus, but also against heterologous antigens (116). A specific immune response was successfully generated after the delivery of an antigen using virus capsid protein SV40 in mammalian cells (120). VLPs were also found to increase the immunogenicity of weak antigens. For example Salmonella typhi membrane antigen, influenza A M2 protein and H1V1 Nef gonadotropin releasing hormone (GnRH) assembled VLPs produced strong antigen specific humoral as well as cellular immune responses (121, 122). It is presumed that the use of VLP based nanoformulations could enable the antigens to achieve conformations resembling to native antigen structure, thus it may result in better stimulation of the host immune response (122).

#### Dendrimers

Dendrimers are three dimensional, mono-dispersed and hyperbranched nano structures that are made up of a mixture of amines and amides. Few studies have explored the application of dendrimers in the delivery of different antigenic molecules. The most commonly used dendrimers for vaccine delivery are polypropyleneimine (PPI) and polyamido amine (PAMAM) dendrimers. A single dose of dendrimer encapsulated multiple antigens was found to produce strong antibody and T-cell responses against Ebola virus, H1N1 influenza, and Toxoplasma gondii (123). This generation of robust immune response was found to be due to efficient uptake of dendrimers by the host cells. Similarly a significant increase in the vaccine efficacy of HIV transactivator of transcription (TAT) based DNA vaccine was observed due to enhanced cellular uptake of PMAM dendrimer (124). Hence, the possibility to tailor the dendrimers to attain certain biological and physico-chemical properties, and also the feasibility to conjugate several ligands to the single molecule have made dendrimers promising candidates for the development of new generation vaccines with enhanced immunogenic properties.

### DELIVERY OF IMMUNE STIMULATORS USING NANOCARRIERS

#### Cytokines

Cytokines are known as important signaling molecules secreted by different cells in response to external stimuli. Some of the cytokines are able to activate immune cells to generate protective immunity against several diseases. However, cytokines are mostly susceptible to early degradation that subdue their participation in the generation of host immunity. Moreover, uncontrolled release of cytokines as immune responders may sometimes lead to harmful side effects (125). To overcome these limitations, several studies have attempted to synthesize engineered nanocarriers to achieve effective and controlled delivery of cytokines to the target sites. This approach was found to reduce their toxicity, improve circulation time and antigen specific T-cell responses in comparison to free cytokines (126, 127). Incorporation of granulocyte macrophage colony stimulating factor (GM-CSF) and interferon alpha (IFN-α) into nano-carriers exhibited great application in cancer therapy (128, 129). Nano-carrier conjugated cytokines also showed great potential in the treatment of infectious diseases. For example, IL-12 encapsulated microspheres induced strong protective immunity against tuberculosis (130). This effect was due to production of high antibody titers as a result of sustained and controlled release of IL-12 from the microspheres in immunized mice (130).

### Toll Like Receptor Agonists

Like cytokines, several toll-like receptor (TLR) agonists were also explored as immune activators to augment immune surveillance mechanisms. Different immune effector cells such as macrophages, B-cells and DCs express different types of TLRs, which are known to interact with specific pathogen associated molecular patterns (PAMPs). These specific interactions eventually initiate downstream signaling cascades to ensure the elimination or generation of immunity against pathogens (131, 132). Conjugation of TLR specific agonists on nanocarriers helps to target the molecules to specific immune cells and therefore reduce the possibility of systemic biodistribution. One such study has shown that conjugation of TLR-7/8 agonist on nano polymers caused efficient internalization by APCs and also prolonged the T cell responses (133). Administration of NPs loaded with vaccine peptide antigen and TLR-7 and 9 ligands were also found to induce strong memory and effector CD8<sup>+</sup> T-cell response (134). Another study has shown that conjugation of TLR-8 agonist to a polymer nanocarrier increased activation and maturation of naive DCs due to selective endocytosis and prolonged release of an immunogen by the nanocarrier inside DCs (135). Moreover, intradermal injection of CpG and antigen encapsulated polymeric NPs were rapidly drained into the lymph nodes to activate DCs (97). These studies indicate that NPs can be used as a tool to appropriately target presentation of antigens to T and B-cell rich lymphoid organs.

#### Nucleic Acids

The genetic molecules such as DNA, plasmids and RNA can also act as immuno-stimulants. Due to these characteristics, in addition to less risk to cause disease particularly in immunocompromised individuals, these genetic materials are considered as promising candidates for the development of next generation vaccines. After administration, the plasmid vector translocates to the nucleus to initiate transcription of recombinant genes using the host cellular machinery. A recombinant DNA segment encoding HspX-PPE44-esxV fusion antigen of M. tuberculosis showed great potential as a new tuberculosis DNA vaccine candidate (136). A similar type of study has been conducted in the past where the vaccination of DNA or RNA constructs expressing mycobacterium antigens were capable of inducing humoral as well as cellular immune responses (137). Likewise, plasmids harboring genes encoding for viral antigen have been encapsulated into alginate nanocarriers and targeted against viral infections (138).

## IMPORTANCE OF PHYSICOCHEMICAL PROPERTIES IN DESIGNING NANO-IMMUNO FORMULATIONS

In order to improve their delivery and vaccine characteristics, different approaches have been practiced to conjugate vaccine molecules to different nanocarriers. Vaccine molecules can be surface conjugated, encapsulated or surface adsorbed with the nanocarriers. Antigen adsorption on the nanocarrier is simply based on the presence of a charge or hydrophobic interactions between NPs and the candidate molecule (139, 140). This type of interaction is usually non-covalent, which may lead to rapid dissociation of antigens from nanocarriers depending upon the external milieu such as pH, ionic strength, temperature, and the antigen hydrophobicity. On the other hand, encapsulation and chemical conjugation of antigen to nanocarriers is more stable due to strong interactions and chemical bond formation between the target molecule and the nanocarrier. Further, antigens can also be encapsulated into nanocarriers by simple mixing reaction during the synthesis. In this case, the antigens are released only after partial or complete dissociation of the nanocarrier (141). These processes have already been used with silica and gold NPs (142). Similarly, chitosan and dextran sulfate NPs were used for the preparation of cationic and anionic antigenic formulations. Some viral antigens are known to bind to both positive as well as negative charged NPs through immobilization process and hydrogen bonds (143). The immobilization process depends on the charge, pH, ratio of NPs and antigens, and the protein partition coefficient between the solution and the colloid (143). Several antigens were successfully delivered to the target sites by chemical conjugation, adsorption and encapsulation to soft nanocarriers like VLPs, liposomes and immune stimulating complexes (ISCOM) (144– 147). ISCOMs are a class of adjuvant formulations that consist of saponins, cholesterol and phospholipids in specific ratios. Antigens can be formulated into ISCOMs directly (148) or after the surface modification (149, 150). Since ISCOM particles are negatively charged, direct conjugation of most of the soluble proteins is a limiting factor. Nanocarriers can augment immunogenicity of a molecule. For example, influenza antigen H1N1 conjugated chitosan NPs and Yersinia pestis F1-antigen coated gold NPs (AuNPs) produced higher levels of antibody and cytokine responses in comparison to mice administered with unconjugated antigens (151).This was found to be due to stabilization and increased immunogenicity of vaccine antigens due to conjugation with NPs.

Another important aspect in the development of nanoimmuno formulations is that they improve antigen delivery and presentation (152). In this context, NP shape, size and surface charge are key factors that affect NP circulation, biodistribution, bioavailability and specificity by crossing biological barriers. Besides these factors, particle geometry such as surface to volume ratio plays an important role in the determination of immunogen release and degradation kinetics (153, 154). Here, the importance of different physicochemical parameters such as size, shape, surface area, porosity, hydrophobicity, hydrophilicity and crystallinity in the interaction between NPs and the target cell is discussed.

#### Size

The size of NPs determines the mode of cellular uptake and specificity (155, 156). PLGA NPs of large size (1, 7 and 17µm) showed reduced internalization rate in comparison to smaller NPs (300 nm) (157). The size of NPs also determines the cellular specificity and migration. Smaller NPs (20–200 nm) were readily endocytosed by the resident DCs, whereas larger size (500– 2,000 nm) NPs were effectively taken up by the migratory DCs (158). NPs of less than 200 nm size were drained into the lymph nodes (159), while particles up to 20 nm range were suitably transported to the APCs (152, 160). Notably, NP curvature also affects the cellular interaction and phagocytosis rate (161). NPs of 150 nm diameter and 450 nm height showed more cellular uptake as compared to the particles having 1,200 × 200 nm size. Of note the size of NPs was also found to influence the activation of signaling pathways. A study has demonstrated that smaller NPs are able to alter the cell signaling processes more efficaciously than the large NPs (31).

#### Surface Charge

Vaccine loaded NPs can also be targeted to specific sites by modifying the NP surface charge. Delivery of such NPs at appropriate sites elicit strong immune responses against antigens. NP surface charge is responsible for the interaction with congnate surface molecules present on the target cells. This was exemplified from the observation that cationic polysterene NPs were efficiently internalized by the APCs in comparison to neutral surface charged NPs. This may be due to electrostatic interactions between the cationic NPs with anionic cell membranes (162, 163). Interestingly, pulmonary instillation of cationic and anionic NPs showed similar endocytosis rate in macrophages and draining lymph nodes, however cationic formulations showed more expression of Ccl2 and Cxc10 chemokines that caused more recruitment and maturation of CD11b DCs in comparison to anionic NPs in the lung (125, 156). Similarly, neutral silica-silane shell polymer NPs were less effective in the activation of innate immune cells (128). These studies clearly indicate appropriate surface modifications of NPs may help to generate stronger immunological responses against specific infection.

#### Shape

Beside size and surface charge, NP shape is also a critical determinant in the cellular interaction, intracellular trafficking and the rate of antigen release inside the host cells (79, 141). Spherical gold NPs were actively internalized by bone marrow derived dendritic cells in comparison to rod shaped particles of similar dimensions (33, 34), and that spherical NPs were able to induce strong immune response than cube or rod shaped NPs (164). Another study reported that worm-like particles were impaired in phagocytosis as compared to spherical NPs (151). These distinctions were ascribed to the differences in contact area between NPs and the target cell membrane. The shape of NPs also determines the localization of NPs inside the host cells. This was demonstrated by the fact that although nano rods and nano sheets were internalized via clathrin mediated endocytosis, nano rods were particularly delivered to the nucleus while nanosheet were retained in the cytoplasm (146, 147, 155). This is an important aspect in the context of improving antigen processing and presentation to T-cells. It is well established that enhanced antigen processing and presentation can be achieved if the candidate molecules are delivered to the lysosomal compartment of the cells.

#### Hydrophobicity

Hydrophobicity of NPs plays a significant role in the interaction with soluble proteins and immune cells through recognition of hydrophobic moieties (165). Previous studies have shown that hydrophobic polymeric NPs are strong inducers of cytokines and co-stimulatory molecules than hydrophilic polymeric NPs (53, 105, 166). Exposure to hydrophobic NPs showed enhanced activation of DCs by inducing the expression of CD86 costimulatory molecules when compared with hydrophilic ones. Similar observations were reported in other innate immune cells, in which hydrophobic NPs were able to activate these cells by up-regulating the expression of proinflammatory cytokine encoding genes (102), and also facilitated opsonization process by increasing the adsorption of immunoglobulins on the cell surfaces (103). However, other studies have reported that polyethylene glycol coating (PEGylation) reduced the interaction of NPs with immune receptors (50, 80). This property is considered useful in the prevention of non-specific adsorption of proteins on NPs and thereby prevent their up-take by APCs (50). Such non-specific adsorption of proteins and their uptake by phagocytic cells can also be preventing by the incorporation of an alkyl linker between the PEG and thiol moieties on NPs (80).

#### Surface Modification

Surface modification of NPs alters ligand specificity and interaction with APCs (160). Conjugation of CD47 molecules on the surface of NPs modulated the down-stream signaling cascades and also reduced NP internalization by phagocytic cells (131). Functionalization of NPs with TLR-7, TLR-8, and TLR-9 agonists increased cytokine production and the expression of immunoregulatory genes (132–134). Similarly, conjugation of poly (methyl vinyl ether-co-maleic anhydride; PVMA), TLR2, and TLR4 agonists, and galactose polymer to NPs were shown to activate the complement pathway as a result of stable binding to C3b complement factor (139, 142). Further, lipoproteinlike NPs showed LPS scavenging activity, thereby resulting in the inhibition of TLR-4 dependent inflammatory responses (140). Overall, these studies strongly demonstrated that tuning of physico-chemical properties of NPs could be used as a fundamental tool to target vaccine molecules to specific sites to induce desired immune responses.

## IMPLICATIONS OF THE NANOCARRIERS IN THE VACCINE DEVELOPMENT

Emerging studies have proved that nanocarriers can be useful mediators in the development of vaccines against various diseases. In this context, it is important to develop NP formulations that can deliver immunogens to APCs especially DCs to induce effective antigen-specific T-cell responses (**Figure 2**). Several nanocarriers have been shown to specifically activate DCs to effectuate anti-tumor or anti-viral immune responses (167–170). Zhu et al. proposed that nano-TiO<sup>2</sup> and Fe3O4-TiO<sup>2</sup> particles could function as a useful vector to promote vaccine delivery in immune cells (168). Coincubation of nano-TiO<sup>2</sup> and Fe3O4-TiO<sup>2</sup> with DCs resulted in an increased production of TNF-α, and also upregulated the expression of CD80, CD86 and MHC class II molecules through the NF-κB signaling pathway (163). In this way, immunization efficacy of various NP formulations such as erythrocyte membrane-enveloped poly(D,L-lactide-co-glycolide) (PLGA) NPs for antigenic peptide (hgp10025-33) and TLR-4 agonist, VLPs expressing RSV glycoproteins, chitosancoated EphrinA1-PE38/GM-CSF, and several others have been improved (171–177). NPs can also control cell polarization and differentiation. Branched polyethylenimine-superparamagnetic iron oxide NPs (bPEI-SPIONs) promoted Th1 polarization of DCs (178). Another study by Sehgal et al. showed that NPs can also be used to target subsets of particular immune cells. They have shown that simultaneous targeting of DC subsets (i.e., DC-SIGN+ and BDCA3+DC) by NPs synergistically stimulated the activation of T cell-mediated immunity when compared with targeting of each DC subset separately (170).

Preclinical studies by different research groups have successfully demonstrated the efficacy of NP based vaccines in the induction of specific immune responses against tuberculosis (42, 179–182). Feng et al. developed a NP-based recombinant DNA vaccine that consists of Esat-6 and fms-like tyrosine kinase 3 ligand enveloped with chitosan NPs (42). Intramuscular prime vaccination followed by nasal boost of this recombinant DNA vaccine remarkably enhanced T cell responses in Mycobacterium tuberculosis challenged mice (42). Another study has shown that pulmonary administration of M. tuberculosis Ag85B antigen and CpG adjuvant conjugated polypropylene sulfide NPs (NP-Ag85B) induced M. tuberculosis specific polyfunctional Th1 responses and also reduced the lung bacterial burden (183).

## TARGETED DELIVERY OF NANOPARTICLES CAN ACTIVATE INNATE AND ADAPTIVE IMMUNE RESPONSES

#### Innate Immunity

Macrophages and monocytes are highly heterologous cells that are distributed throughout the body. Macrophages process and present the antigens to elicit adaptive immune response. Due to their intrinsic phagocytic nature, macrophages can be easily targeted by surface engineered NPs, in which cognate ligands agonist to macrophage receptors can be conjugated on the NP surface (**Figure 1**). As discussed above several physicochemical parameters of NPs such as size, surface charge, hydrophobicity, surface topography, and material composition can be optimized to facilitate the interactions between NPs and macrophage receptors (184–186). The rate of NP endocytosis also depends upon the type of cell surface receptors and the ligand conjugated to the NP surface. For example, NPs targeted via mannose and Fc receptors were rapidly internalized as compared to scavenger receptors (187). Endocytosis of IgG and anti-F4/80 antibody coated NPs showed more uptake rate and retention time inside the macrophages without affecting the cell viability (188, 189). Also, positively charged NPs interact more strongly with negatively charged phospholipid components of the cell membrane (190). Hyperactivation of some inflammatory cells can also be restricted through controlled release of stimulants using NPs. Upon activation, neutrophils can secrete variety of cytokines and hydrolytic enzymes in response to infection (191). Prolonged neutrophil activation often leads to acute inflammation and tissue damage at the localized site. Therefore, controlled release of molecules is

necessary to prevent the hyperactivation and massive recruitment of neutrophils. It has been reported that bovine serum albumin (BSA) NPs were able to modulate the functions of neutrophils following their internalization. Intravenous injection of antiinflammatory peptide encapsulated polymeric NPs reduced neutrophil recruitment and subsequently hyperinflammation to prevent further tissue damage (192). The use of NPs to deliver vaccine/drugs in a controlled fashion is now considered as an attractive approach to develop therapeutic strategies against a range of acute and chronic inflammatory diseases (193).

## Adpative Immunity

T and B-cells of the adaptive immune system express a repertoire of receptors to recognize a range of antigens. Activation or suppression of T-cell immunity can determine the fate of a disease. A number of NP based therapeutic strategies have been developed to regulate T-cell activity against viral, bacterial, or fungal infections. For example, antiviral siRNA or retroviral drug encapsulated lipid NPs or dendrimers were effectively delivered to CD4<sup>+</sup> T-cells to block HIV replication. This caused a significant reduction in HIV titer when compared with the use of non-encapsulated retroviral drugs (191, 194). T-cell activation also depends up on the type and size of NP used for the delivery of antigen. Liposome encapsulated antigens were better presented to CD4<sup>+</sup> T cells by APCs (195, 196) and delivery of 200 nm ova conjugated NPs increased MHC class I and II expression and also produced a higher percentage of antigen specific CD4+T cells as compared to 30 nm ova conjugated particles (197).

B cells are able to recognize and respond to the microbial surface antigens through B-cell receptors (198). Activation and clonal expansion of antigen specific B-cells using engineered NPs have been exploited for the development of vaccines against different diseases (**Figure 2**). Encapsulation of antigen in virus like particles (VLPs) was able to induce strong and durable humoral responses when compared with the administration of exposed vaccine molecules (199). The potency of immune responses also depends upon the mode of antigen presentation to the target cells. Surface conjugated immunogenic proteins and peptides were able to activate B cells much stronger than encapsulated antigens (200). A single dose of PLGA NPs with surface displayed ovalbumin (OVA) elicited strong antibody responses in vivo as compared to free OVA (201, 202). NPs can also be used to activate specific immune responses. A study has shown that peptide conjugated carbon nanotubes showed significant antigen specific IgG response in comparison to peptide or adjuvant alone (83).

### NANOPARTICLES CAN BE USED TO INCREASE CROSS ANTIGEN PRESENTATION

In general, antigens captured by APCs from the extracellular environment are targeted to the endo-lysosomal compartments, where they are first processed into peptides and then loaded onto class II MHC molecules before presentation to CD4<sup>+</sup> helper T cells. However, cytosolic antigens are loaded on MHC class I molecules and presented to CD8<sup>+</sup> T-cells, which are crucial for the clearance of viral and intracellular infections (203). It is reported that some fraction of antigens delivered through NPs are trafficked to cytosolic vacuoles of APCs and presented by MHC class I molecules (203–205). The NP mediated cross antigen presentation was first demonstrated in antigens conjugated to iron oxide polymer NPs (206–209). In addition, inorganic and polymeric NPs have also been used for antigen delivery to cytosol (210–212). In this context, lipid NPs were shown to induce CD8<sup>+</sup> T cell expansion by efficient antigen cross presentation against viral infection in in-vivo models (102, 213). Similarly, invariant natural killer T cells (iNKT), which are a special subset of T-cells, recognize lipid antigens presented by CD1d cells. PLGA NPs conjugated with α-galactosylceramide glycolipid, an iNKT cell stimulant, increased cytokine release as well as expansion of antigen specific CD8+ T cells (214). The cross antigen presentation also depends upon the particle-antigen linkages. It has been shown that disulfide bonding between NP and antigens caused release of antigens into the endosomal compartment and also enhanced CD8<sup>+</sup> T cell formation as compared to non-degradable linkers (215, 216). Similarly, pulmonary administration of NPs efficiently enhanced cross antigen presentation, which resulted in at least 10-fold more effector CD8+T cells in lungs (217).

#### NANOPARTICLES AS ADJUVANTS TO GENERATE IMMUNE RESPONSES IN LYMPHOID ORGANS

Adjuvants are known to enhance and prolong the immune responses against antigens. Delivery of adjuvants and antigens using NPs have been found useful to prolong their exposure in the lymphoid organs such as lymph nodes to generate robust immune responses. This is especially important for small adjuvant molecules, which are rapidly cleared from the bloodstream. NPs with a size ranging from 10–100 nm can penetrate the extracellular matrix and travel to the lymph nodes where they can be internalized by the resident macrophages to activate T-cell responses (218–220). The bio-distribution of NPs also depends upon the route of administration and size. It was observed that larger particles accumulated near the site of NPs and were subsequently endocytosed by the local APCs (160), whereas the smaller NPs drained to the blood capillaries (158, 218). PEG coated liposomes of 80–90 nm diameter showed higher accumulation in lymph nodes after subcutaneous administration as compared to intravenous and intraperitoneal administration (221).

## CONCLUSIONS

The nano-immuno formulations can improve the antigen stability, targeted delivery and also enhance their immunogenicity properties. Most soluble antigens cannot be efficiently endocytosed by the APCs and hence are poorly effective in inducing protective immunity. The immunogenicity of such soluble vaccine antigens can be improved by conjugating them with nanocarriers that can facilitate the recognition and uptake by APCs. This strategy has already been proved effective for inducing/increasing the immunogenicity of poorly immunogenic antigens, such as polysaccharides of pneumococcal vaccines (222). In the last few years, the application of nanotechnology in the field of immune engineering is growing rapidly with a number of new carrier synthesis strategies. Furthermore, novel nano formulations also contain immunostimulatory molecules to enhance the adjuvant properties of the nanoparticles. Co-encapsulation of the TLR agonists [e.g., CpG, poly(I:C)] (77) or imiquimoid (78) into dextran or chitosan NPs, respectively enhanced receptor-based recognition of the nanovaccines with subsequent cell activation. The recent study by Margaroni et al. showed that vaccination with poly(D,L-lactide-co-glycolide; PLGA) nanoparticles with Leishmania infantum antigens (sLiAg) and surface-modified with a TNFα-mimicking eight-amino-acid peptide (p8) induced significant protection against parasite infection in BALB/c mice accompanied by activation of CD8+ T cells and increase in IFNγ production (223).

Additionally, NPs can be tailored for non-invasive administration and prolonged delivery of the vaccine antigens to a specific location, thus providing the possibility for formulation of the single dose vaccine. Several studies clearly demonstrated the efficacy of the non-invasively administered vaccines such as intranasal application of influenza nano vaccine (224), chitosan NPs with hemagglutinin protein of H1N1 influenza virus (225), Streptococcus equi proteins (226), hepatitis B surface antigen (pRc/CMV-HBs) (227) and plasmid encoding a multi-epitope protein against M. tuberculosis (pHSP65pep) (228) or antigen 85B (229) were used to provide protective immunity against infections. These considerations can improve the progress of ongoing strategies in the development of nanoparticle-based vaccines. In future, development of nanovaccines will address not only the possibility to induce the immune response but also the anti-infective therapeutic activity of NPs thus representing the feasibility to apply multifunctional particles for the treatment of diseases.

## AUTHOR CONTRIBUTIONS

AS and RP wrote the manuscript. AS supervised the process. MS wrote the part on Use of nanocarriers in vaccine delivery to dendritic cells.

#### FUNDING

For financial support, we thank Department of Science and Technology (SR/NM/NS-1159/2016), Govt of India and Alexander von Humboldt Fellowship to AS. MS was supported by the Alexander von Humboldt Fellowship and by a grant of

#### REFERENCES


the Russian Science Foundation 14-50-00068 and by the Federal Agency of Scientific Organizations, Russia.

#### ACKNOWLEDGMENTS

The authors thank Nan-Jong Lee for figures preparation.


in biodegradable nanospheres. J Control Release (2002) 85:247–62. doi: 10.1016/S0168-3659(02)00275-4


vaccine against fungal infection. J Biomed Nanotechnol. (2013) 9:221–30. doi: 10.1166/jbn.2013.1491


activation of innate immune receptors. Mol Pharm. (2017) 14:4098–112. doi: 10.1021/acs.molpharmaceut.7b00795


**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 Pati, Shevtsov and Sonawane. 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.