# TOXOPLASMA GONDII HOST INTERACTIONS: A STORY OF IMMUNE ATTACK AND PARASITE COUNTERATTACK

EDITED BY : Jeroen P. J. Saeij, Eva Frickel, Kirk Jensen and Nicolas Blanchard PUBLISHED IN : Frontiers in Cellular and Infection Microbiology

#### 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 978-2-88963-402-6 DOI 10.3389/978-2-88963-402-6

#### 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

# TOXOPLASMA GONDII HOST INTERACTIONS: A STORY OF IMMUNE ATTACK AND PARASITE COUNTERATTACK

Topic Editors:

Jeroen P. J. Saeij, University of California, Davis, United States Eva Frickel, Francis Crick Institute, United Kingdom Kirk Jensen, University of California, Merced, United States Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

Toxoplasma gondii is an obligate intracellular parasite that can infect all warm-blooded animals, including an estimated ~30% of humans. It can cause severe disease in immune-suppressed individuals and in fetuses as well as blinding chorioretinitis in adults and children. Toxoplasma-innate immune system interactions determine early parasite control and activation of the adaptive immune system by the host and are therefore critical in determining host survival during the acute phase of infection. However, induction of an exaggerated inflammatory response can also lead to pathology. Only the chronic tissue cyst form of Toxoplasma is orally infectious. It is therefore critical for the parasite's survival during the chronic phase to escape immune responses at this stage as well.

Toxoplasma exists as genetically divergent strains mostly depending on geography, with the most strain diversity being found in South America. The key to Toxoplasma's successful co-option of the host are proteins secreted from its rhoptry and dense granule secretory organelles. Rhoptry proteins (ROPs) are secreted into the host cell cytoplasm upon invasion while dense granule proteins (GRAs) are secreted once the parasite establishes itself in its parasitophorous vacuole (PV). GRAs can localize to the PV, the PV membrane, or are secreted beyond the PVM into the host cytoplasm. Many ROPs and GRAs are involved in modulating host cell signaling pathways and evasion of host immune responses and play important roles in Toxoplasma virulence. Polymorphisms in Toxoplasma's ROPs and GRAs, likely determine how well these effectors bind to the divergent substrates in different host species, which can explain Toxoplasma strain differences in virulence in a particular host species.

By studying Toxoplasma we have not only started to unravel how the parasite modulates immune responses to enhance its survival, replication, and transmission but we have also learned a lot about the immune system. Many unique mechanisms of immunity have indeed been defined using Toxoplasma and this parasite has aided our understanding of tissue-specific immune responses in the brain and intestine.

This Research Topic will give a comprehensive overview of Toxoplasma-host immune response interactions. Most Toxoplasma virulence determinants to date have been established in murine systems and it is unclear how the parasite interacts with other intermediate hosts and humans. In addition, the interactions of Toxoplasma with some of the most relevant cell types during infection, including dendritic cells, neurons, intestinal epithelial cells or vascular endothelial cells, remain poorly understood.

Citation: Saeij, J. P. J., Frickel, E., Jensen, K., Blanchard, N., eds. (2020). Toxoplasma gondii Host Interactions: A Story of Immune Attack and Parasite Counterattack. Lausanne: Frontiers Media SA. doi: 10.3389/978-2-88963-402-6

# Table of Contents


Agnieszka Lis, Mandi Wiley, Joan Vaughan, Peter C. Gray and Ira J. Blader

*39 From Entry to Early Dissemination—Toxoplasma gondii's Initial Encounter With its Host*

Estefania Delgado Betancourt, Benjamin Hamid, Benedikt T. Fabian, Christian Klotz, Susanne Hartmann and Frank Seeber

*48 Calling in the CaValry—Toxoplasma gondii Hijacks GABAergic Signaling and Voltage-Dependent Calcium Channel Signaling for* Trojan horse*-Mediated Dissemination*

Amol K. Bhandage and Antonio Barragan

*61 Toxoplasma-Induced Hypermigration of Primary Cortical Microglia Implicates GABAergic Signaling*

Amol K. Bhandage, Sachie Kanatani and Antonio Barragan

*73 Comprehensive Kinetic Survey of Intestinal, Extra-Intestinal and Systemic Sequelae of Murine Ileitis Following Peroral Low-Dose Toxoplasma gondii Infection*

Markus M. Heimesaat, Ildiko R. Dunay and Stefan Bereswill


Corinne Loeuillet, Anais Mondon, Salima Kamche, Véronique Curri, Jean Boutonnat, Pierre Cavaillès and Marie-France Cesbron-Delauw *126 Interplay Between Toxoplasma gondii, Autophagy, and Autophagy Proteins*

Carlos S. Subauste

*137 Impact of Toxoplasma gondii Infection on Host Non-coding RNA Responses*

Kayla L. Menard, Breanne E. Haskins and Eric Y. Denkers


Kerrie E. Hargrave, Stuart Woods, Owain Millington, Susan Chalmers, Gareth D. Westrop and Craig W. Roberts

# Toxoplasma Does Not Secrete the GRA16 and GRA24 Effectors Beyond the Parasitophorous Vacuole Membrane of Tissue Cysts

Shruthi Krishnamurthy and Jeroen P. J. Saeij\*

Department of Pathology, Microbiology and Immunology, School of Veterinary Medicine, University of California, Davis, Davis, CA, United States

#### Edited by:

Mohamed Ali Hakimi, Institut National de la Santé et de la Recherche Médicale (INSERM), France

#### Reviewed by:

Markus Meissner, Ludwig-Maximilians-Universität München, Germany Renato Augusto DaMatta, State University of Norte Fluminense, Brazil

\*Correspondence:

Jeroen P. J. Saeij jsaeij@ucdavis.edu

#### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 06 July 2018 Accepted: 01 October 2018 Published: 18 October 2018

#### Citation:

Krishnamurthy S and Saeij JPJ (2018) Toxoplasma Does Not Secrete the GRA16 and GRA24 Effectors Beyond the Parasitophorous Vacuole Membrane of Tissue Cysts. Front. Cell. Infect. Microbiol. 8:366. doi: 10.3389/fcimb.2018.00366 After invasion, Toxoplasma resides in a parasitophorous vacuole (PV) that is surrounded by the PV membrane (PVM). Once inside the PV, tachyzoites secrete dense granule proteins (GRAs) of which some, such as GRA16 and GRA24, are transported beyond the PVM likely via a putative translocon. However, once tachyzoites convert into bradyzoites within cysts, it is not known if secreted GRAs can traffic beyond the cyst wall membrane. We used the tetracycline inducible system to drive expression of HA epitope tagged GRA16 and GRA24 after inducing stage conversion and show that these proteins are not secreted beyond the cyst wall membrane. This suggests that secretion of GRA beyond the PVM is not important for the tissue cyst stage of Toxoplasma.

Keywords: Toxoplasma gondii, secreted effectors, tissue cyst wall, tetracycline inducible expression, GRA16, GRA24

# INTRODUCTION

Toxoplasma gondii, which belongs to the phylum Apicomplexa, is an obligate intracellular parasite that can cause disease in immuno-compromised patients and fetuses (Montoya and Liesenfeld, 2004; Weiss and Dubey, 2009). It is the causative agent of toxoplasmosis, the 2nd most common cause of food-borne illness in the USA (Jones and Dubey, 2012). Infectious tissue cysts are present in brain and muscles of many warm-blooded chronically infected hosts (Kim and Weiss, 2004). Infected cats, which are the definitive hosts, shed infectious oocysts in their feces contaminating food and water sources. Infection is initiated by ingestion of either tissue cysts containing the bradyzoite life-cycle stage or oocysts (Dubey, 1998). Upon stage conversion into tachyzoites and invasion of a host cell, Toxoplasma forms a parasitophorous vacuole (PV) that is surrounded by the PV membrane (PVM) (Black and Boothroyd, 2000). During invasion, Toxoplasma secretes effector proteins from rhoptries (ROPs), which mediate invasion, inhibition of host restriction factors, and modulation of host signaling pathways (Dubremetz, 2007; Boothroyd and Dubremetz, 2008; Hakimi et al., 2017). Once inside the PV, proteins from the dense granule secretory organelles (GRAs) are secreted onto, and beyond the PVM into the host cell cytoplasm (Hakimi and Bougdour, 2015; Hakimi et al., 2017). Three putative translocon proteins: Myc-regulation 1 (MYR1) along with MYR2 and MYR3 determine transport of GRA16 and GRA24 across the PVM into the host cell cytoplasm after which they traffic to the host cell nucleus (Franco et al., 2016; Marino et al., 2018). In addition to these putative translocon proteins, an aspartyl protease, ASP5 cleaves many secreted GRA proteins at a characteristic RRLxx motif also known as the Toxoplasma export element (TEXEL) motif which is important for their localization and function (Coffey et al., 2015; Hammoudi et al., 2015). Most of these effectors that have been characterized are from the non-orally infectious tachyzoite stage. It is unclear if bradyzoites within tissue cysts, akin to tachyzoites within the PV, can secrete GRAs beyond the PVM as the cyst wall is built on the inside of the PVM (Jeffers et al., 2018) and presents a potential barrier for GRA secretion into the host cell.

GRA16, GRA24, GRA28 and IST (T. gondii inhibitor of STAT1 transcriptional activity) are secreted by tachyzoites beyond the PVM and traffic to the host nucleus (Bougdour et al., 2013; Braun et al., 2013; Gay et al., 2016; Nadipuram et al., 2016; Olias et al., 2016) where they modulate host signaling pathways important for parasite fitness. In this brief report we use the tetracycline inducible system to induce the expression of epitope tagged GRA16 and GRA24 after in vitro stage conversion. We observed that anhydrotetracycline (ATc) induced GRA16-HA and GRA24- HA are not secreted beyond the PVM and are not localized to the host cell nucleus. Instead, they accumulate within the in vitro tissue cysts.

# MATERIALS AND METHODS

#### Host Cells and Parasite Strain

Human foreskin fibroblasts (HFFs) were used as host cells and were cultured under standard conditions using Dulbecco Modified Eagle Medium (DMEM) with 10% fetal bovine serum (FBS) (Rosowski et al., 2011). We chose GT1 parasites expressing tetracycline repressor (Tet-R) (Etheridge et al., 2014), a type I strain that is capable of forming cysts during in vitro stage conversion (Lindsay et al., 1991; Fux et al., 2007) induced by pH 8-8.2+low CO<sup>2</sup> (Skariah et al., 2010).

#### Plasmid Construction

Using Gibson assembly (Gibson et al., 2009), we constructed Tet-On plasmids with a phleomycin resistance cassette to express GRA16-HA and GRA24-HA. The pTeton vector backbone was amplified with the following primers which were used to construct both the pTetONGRA16-HA SRS22A3′UTR as well as the pTetONGRA24-HA SRS22A3UTR constructs:

vector TetON for-GCATCCACTAGTGCTCTTCAAGGTT TTACATCCGTTGCCT

Vector TetON rev-AATTGCGCCATTTTGACGGTGACGA AGCCACCTGAGGAAGAC

The following primers were used to amplify the pieces for pTetONGRA16-HA SRS22A3′ UTR Gibson assembly:

Vector GRA16 for-ACCGTCAAAATGGCGCAATTATGTAT CGAAACCACTCAGGGATAC

SRS22UTRGRA16 rev-AATGACAGGTTCAAGCATAATCG GGAACGTCGTATG

GRA16HA SRS22A for-TTATGCTTGAACCTGTCATTTAC CTCCAGTAAACATG

SRS22Avector rev-TGAAGAGCACTAGTGGATGCGTTCT AGTGCTGTACGGAAAAGCAAC

The following primers were used to amplify the pieces for pTetONGRA24-HA SRS22A3′ UTR Gibson assembly:

vectorGRA24 for –ACCGTCAAAATGGCGCAATTATGCTC CAGATGGCACGATATACCG

SRS22AUTRGRA24HA rev-AATGACAGGTTTAAGCATAA TCGGGAACGTCGTATG

GRA24HASRS22A For-TTATGCTTAAACCTGTCATTTA CCTCCAGTAAACATG.

# Parasite Transfection and Selection

The pTetOn vectors containing GRA16-HA and GRA24-HA were linearized using the AseI restriction enzyme. The linearized plasmids (50 µg) were electroporated into 5 × 10<sup>7</sup> GT1tetR parasites using the protocol described in (Gold et al., 2015). After lysis of parasites from host cells, they were selected twice with 50µg/ml phleomycin (Ble) and maintained in 5µg/ml Ble (Krishnamurthy et al., 2016) after which they were cloned by limiting dilution.

# Indirect Immunofluorescence With Tachyzoites

Coverslips with a monolayer of HFFs were infected in the presence or absence of 2µM ATc with GT1-TetR parasites or single clones of parasites expressing GRA16-HA and GRA24- HA under the RPS13 promoter downstream of a Tet-O7 operator. The coverslips were fixed 16 h post infection with 3% formaldehyde and processed for IFA using rabbit anti-HA primary (Roche) antibody followed by goat anti-rabbit Alexa-555 secondary antibody and Hoechst (Invitrogen).

# In Vitro Stage Differentiation and IFA With Tissue Cysts

HFFs were infected with a single clone of GT1-tetR parasites that expressed GRA16-HA only in the presence of ATc. The media was switched from DMEM with 10% FBS to tricine-buffered RPMI media with pH 8.0 and put in an incubator with low CO<sup>2</sup> after 24 h of infection (MOI = 0.1) to induce tachyzoite to bradyzoite stage conversion. After 5 days, 2µM ATc was added and 2 days later the coverslips were fixed with 100% cold methanol (also eliminates YFP signal from TetR) and processed for IFA. The cyst wall was stained with DBA-FITC (Boothroyd et al., 1997) along with HA and Hoechst as described above.

#### RESULTS

To test if GRAs are secreted beyond the tissue cyst wall membrane, we utilized the Tet-inducible system (Etheridge et al., 2014) to express an HA-tagged copy of GRA16 or GRA24 under the Tet operator (Tet-O) in parasites expressing Tet-R. We could not just use GRA16-HA or GRA24-HA expressed from the endogenous promoter because if we saw these proteins in the

**Abbreviations:** Tet-on, tetracycline inducible system; TetR, tetracycline repressor protein; ATc, anhydrotetracycline; HA, hemagglutinin epitope tag; YFP, yellow fluorescent protein; Ble, phleomycin; HFF, human foreskin fibroblast; DBA lectin, dolichos biflorus agglutinin; IFA, indirect immunofluorescence; GRA, dense granule proteins; ROP, rhoptry proteins; IST, T. gondii inhibitor of STAT1 transcriptional activity; PV, parasitophorous vacuole; PVM, PV membrane.

host nucleus we would not know if they were secreted beyond the PVM before the cyst wall was made (as tachyzoites) or after the cyst wall was made (as bradyzoites). In the absence of anhydrotetracycline (ATc), the Tet-R binds to Tet-O and represses the transcription of either GRA16-HA or GRA24-HA under the RPS13 promoter (Etheridge et al., 2014; Wang et al., 2016). ATc binds TetR and relieves repression of transcription which allows for the expression of HA-tagged GRA16 or GRA24. Since ATc is smaller than the size-exclusion limit of the cyst wall (Lemgruber et al., 2011), we decided to use this system to answer our question.

To check if our constructs were able to stably express functional GRA16 and GRA24, we first transfected them into the RH parasite strain and observed nuclear localization of these proteins (data not shown). After transfection of the GT1 Tet-R expressing strain with the Tet-inducible GRA16-HA or GRA24- HA construct and subsequent selection with phleomycin for stable integration, we show by IFA that in tachyzoites GRA16- HA (**Figures 1A,B**) and GRA24-HA (**Figures 1C,D**) were only expressed in the presence of ATc.

A single parasite clone that expressed GRA16-HA or GRA24- HA only in the presence of ATc was chosen for induction of stage differentiation in vitro in human foreskin fibroblast (HFFs). Five days post-switching, 2µM of ATc was added to the cultures to induce the expression of GRA16-HA and GRA24-HA since we observed that at least 50% of the parasites had converted to cysts by staining the cyst wall with DBA-lectin (Boothroyd et al., 1997) (data not shown). The parasites were fixed 48 h following addition of ATc to allow for sufficient expression of GRA16-HA and GRA24-HA. We performed an indirect immunofluorescence assay (IFA) to determine the localization of GRA16 and GRA24 using anti-HA antibody as well as DBA-lectin to detect the cyst wall. We show that in host cells containing tissue cysts, GRA16-HA and GRA24-HA were not detected in the host cell nucleus or beyond the tissue cyst wall membrane and that instead they accumulated underneath the cyst wall. Almost 100% of vacuoles we observed were DBA positive. We decided to observe HFFs only infected with one parasite and therefore containing only one cyst as differences in the timing of conversion could affect the localization of GRA16 and GRA24. Out of 189 (80 for GRA16-HA and 109 for GRA24-HA from three biological replicates) images of singly infected host cells containing DBA positive cysts, both GRA16 and GRA24 were expressed exclusively beneath the cyst wall only in the presence of ATc (**Figures 2A,B**). We observed GRA16 and GRA24 localized to the host cell nucleus only in multiple infected cells containing tachyzoites, along with in vitro cysts (**Supplemental Figure 1**). Thus, our results show that GRA16 and GRA24 are

not secreted beyond the tissue cyst membrane into the host cell.

#### DISCUSSION

We show here that GT1 parasites are able to form DBA lectin positive in vitro cysts. We also show for the first time that ATc is able to cross the cyst wall in vitro. Even though bradyzoites within tissue cysts are not as metabolically active compared to tachyzoites, it is becoming clear that they are also not in a dormant state (Sinai et al., 2016). However, we observed that bradyzoites do not secrete GRA16 and GRA24 beyond the in vitro cyst wall membrane. These proteins accumulated within the cyst wall suggesting that their role in the host cell nucleus is not required at this stage. In preparing for the chronic phase of their life cycle, bradyzoites lose connectivity between themselves through the intravacuolar network (IVN) and undergo asynchronous division (Frénal et al., 2017). Our data indicates that the bradyzoites within cysts also lose connectivity to host cells by not secreting GRAs, which usually modulate host signaling pathways in tachyzoites, beyond the PVM. Possibly bradyzoites require these proteins during natural oral infections after excystation from tissue cysts to establish infection in gut epithelial cells of the host. Not secreting parasite proteins beyond the cyst wall might help Toxoplasma to remain invisible and undetected by the host immune response during the chronic phase of infection. This hypothesis is in conjunction with published literature wherein proteins from dense granules (Ferguson, 2004; Lemgruber et al., 2011) and rhoptries (Schwarz et al., 2005) were shown to be secreted in bradyzoites but never beyond the tissue cyst wall. Another possibility may be that the translocon proteins MYR1/2/3 or ASP5 are not sufficiently expressed at this stage to effectively mediate transport of secreted GRAs beyond the cyst wall membrane. Even though all the MYRs and ASP5 are expressed in tachyzoites, sporozoites and bradyzoites, their expression is significantly lower in bradyzoites (Marino et al., 2018). However, even if MYR1-3 and ASP5 are expressed, our data indicate that the cyst wall seems to act as a barrier as ATc- induced GRA16 and GRA24 accumulated beneath the wall (**Figure 2**).

#### AUTHOR CONTRIBUTIONS

SK generated all the data. SK and JS wrote and edited the manuscript.

# FUNDING

This research was funded by NIH-5R01AI080621-10.

#### ACKNOWLEDGMENTS

We thank all the members of the Saeij lab for their feedback on this manuscript. We also thank Dr. Ronald D. Etheridge for proving us with GT1-TetR parasites.

# REFERENCES


# SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00366/full#supplementary-material

Supplemental Figure 1 | Localization of GRA16-HA and GRA24-HA to the host cell nucleus was observed only in multiple infected host cells containing parasites in different stages of in vitro stage conversion. Parasites expressing GRA16-HA (A) or GRA24-HA (B) localized the respective epitope tagged GRAs in the parasites as well as in the host cell nucleus. Non-uniform DBA staining suggests that the parasites were in different stages of conversion into tissue cysts. Images are scaled to 10µm.


**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 Krishnamurthy and Saeij. 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.

# Virulence in Mice of a *Toxoplasma gondii* Type II Isolate Does Not Correlate With the Outcome of Experimental Infection in Pregnant Sheep

#### *Edited by:*

Jeroen P. J. Saeij, University of California, Davis, United States

#### *Reviewed by:*

José Roberto Mineo, Federal University of Uberlandia, Brazil Bellisa Freitas Barbosa, Federal University of Uberlandia, Brazil

#### *\*Correspondence:*

Luis Miguel Ortega-Mora luis.ortega@ucm.es Julio Benavides julio.benavides@csic.es

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 18 September 2018 *Accepted:* 10 December 2018 *Published:* 04 January 2019

#### *Citation:*

Sánchez-Sánchez R, Ferre I, Regidor-Cerrillo J, Gutiérrez-Expósito D, Ferrer LM, Arteche-Villasol N, Moreno-Gonzalo J, Müller J, Aguado-Martínez A, Pérez V, Hemphill A, Ortega-Mora LM and Benavides J (2019) Virulence in Mice of a Toxoplasma gondii Type II Isolate Does Not Correlate With the Outcome of Experimental Infection in Pregnant Sheep.

Front. Cell. Infect. Microbiol. 8:436. doi: 10.3389/fcimb.2018.00436 Roberto Sánchez-Sánchez <sup>1</sup> , Ignacio Ferre<sup>1</sup> , Javier Regidor-Cerrillo<sup>1</sup> , Daniel Gutiérrez-Expósito<sup>2</sup> , Luis Miguel Ferrer <sup>3</sup> , Noive Arteche-Villasol <sup>2</sup> , Javier Moreno-Gonzalo<sup>1</sup> , Joachim Müller <sup>4</sup> , Adriana Aguado-Martínez <sup>4</sup> , Valentín Pérez <sup>2</sup> , Andrew Hemphill <sup>4</sup> , Luis Miguel Ortega-Mora<sup>1</sup> \* and Julio Benavides <sup>2</sup> \*

<sup>1</sup> SALUVET, Animal Health Department, Faculty of Veterinary Sciences, Complutense University of Madrid, Madrid, Spain, 2 Instituto de Ganadería de Montaña (CSIC-Universidad de León), León, Spain, <sup>3</sup> Departamento de Patología Animal, Facultad de Veterinaria, University of Zaragoza, Zaragoza, Spain, <sup>4</sup> Vetsuisse Faculty, Institute of Parasitology, University of Bern, Bern, Switzerland

Toxoplasma gondii is an apicomplexan parasite that infects almost all warm-blooded animals. Little is known about how the parasite virulence in mice extrapolates to other relevant hosts. In the current study, in vitro phenotype and in vivo behavior in mice and sheep of a type II T. gondii isolate (TgShSp1) were compared with the reference type II T. gondii isolate (TgME49). The results of in vitro assays and the intraperitoneal inoculation of tachyzoites in mice indicated an enhanced virulence for the laboratory isolate, TgME49, compared to the recently obtained TgShSp1 isolate. TgShSp1 proliferated at a slower rate and had delayed lysis plaque formation compared to TgME49, but it formed more cyst-like structures in vitro. No mortality was observed in adult mice after infection with 1–10<sup>5</sup> tachyzoites intraperitoneally or with 25–2,000 oocysts orally of TgShSp1. In sheep orally challenged with oocysts, TgME49 infection resulted in sporadically higher rectal temperatures and higher parasite load in cotyledons from ewes that gave birth and brain tissues of the respective lambs, but no differences between these two isolates were found on fetal/lamb mortality or lesions and number of T. gondii-positive lambs. The congenital infection after challenge at mid-pregnancy with TgShSp1, measured as offspring mortality and vertical transmission, was different depending on the challenged host. In mice, mortality in 50% of the pups was observed when a dam was challenged with a high oocyst dose (500 TgShSp1 oocysts), whereas in sheep infected with the same dose of oocysts, mortality occurred in all fetuses. Likewise, mortality of 9 and 27% of the pups was observed in mice after infection with 100 and 25 TgShSp1 oocysts, respectively, while in sheep, infection with 50 and 10 TgShSp1 oocysts triggered mortality in 68 and 66% of the fetuses/lambs. Differences in vertical transmission in the surviving offspring were only found with the lower oocyst doses (100% after infection with 10 TgShSp1 oocysts in sheep and only 37% in mice after infection with 25 TgShSp1 oocysts). In conclusion, virulence in mice of T. gondii type II isolates may not be a good indicator to predict the outcome of infection in pregnant sheep.

Keywords: *Toxoplasma gondii*, type II, virulence, phenotypic traits, mice, sheep, congenital toxoplasmosis

#### INTRODUCTION

Toxoplasma gondii is an apicomplexan parasite capable of infecting almost all warm-blooded animals and causing potentially fatal disease in humans and some relevant domestic species, such as small ruminants (Dubey, 2010). Toxoplasma gondii may be transmitted horizontally by oral ingestion of infectious oocysts from the environment and tissue cysts contained in raw and undercooked meat or vertically by transplacental transmission of tachyzoites (Tenter et al., 2000). Toxoplasma gondii diverges in three main clonal lineages, I, II, and III, with marked differences in mouse virulence (Howe and Sibley, 1995) although atypical strains have been also described (Su et al., 2006). Type I isolates are highly virulent in mice, whereas types II and III show a dose-dependent mortality (Saeij et al., 2006). Laboratory isolates from types I, II, and III are used in T. gondii research and have been maintained in successive passages in cell culture and mice. Enhanced virulence throughout successive passages in cell culture and mice for T. gondii type I isolates (e.g., RH isolate) has been widely reported (Villard et al., 1997; Mavin et al., 2004; Khan et al., 2009), but little is known about the influence of continuous passages in type II isolates. Likewise, there is little information on how T. gondii virulence in mice compares to virulence in other species, particularly in those experiencing clinical toxoplasmosis, such as sheep and humans.

Transmission of T. gondii from dams to offspring during pregnancy (congenital toxoplasmosis) is one of the consequences of infection (Innes et al., 2009; McAuley, 2014). Many animal experimental models have been developed; among them, mouse models are the most frequently used due to similarities in placental histology between rodents and humans (Vargas-Villavicencio et al., 2016). However, the structure of the placenta, reproductive physiology and immune responses greatly differ between rodents and ruminants, which clearly could influence the passing of T. gondii through the maternofetal interface and the modulation of the host immune responses during pregnancy (Entrican, 2002). In sheep, T. gondii is one of the main abortifacient agents (Dubey, 2009). Type II T. gondii isolates are the most prevalent in all of the hosts in Europe, including sheep (Chessa et al., 2014). In most of the experimental studies in pregnant sheep, type II isolates M1, M3, and M4 have been used (Dubey, 2009; Castaño et al., 2014), but the virulence in mice of these isolates was never assessed.

The aim of this study was to compare the phenotype in vitro and the virulence in mice of a newly obtained T. gondii type II isolate (TgShSp1) with that of the laboratory type II reference isolate (TgME49) and to compare congenital infection in mice and in sheep.

#### MATERIALS AND METHODS

#### Ethics Statement

Animal procedures for the T. gondii isolation by a mouse bioassay of field samples from sheep abortions (PROEX 274/16), for evaluation of virulence by intraperitoneal inoculation of tachyzoites in non-pregnant mice (PROEX 274/16) and for T. gondii infection of mice and cats (PROEX 166/14) for oocyst production were approved by the Animal Welfare Committee of the Community of Madrid, Spain, following proceedings described in Spanish and EU legislation (Law 32/2007, R.D. 53/2013, and Council Directive 2010/63/EU). Animal procedures to characterize TgShSp1 oocysts in pregnant mice were approved by the Animal Welfare Committee of the Canton of Bern (approval No. BE 101/17). All sheep handling practices were approved by the local government and followed the recommendations of the Directive 2010/63/EU of the European Parliament, the Council on the protection of animals used for scientific purposes, and the IGM-CSIC Animal Experimentation Committee (protocol number 416-2016). All animals used in this study were handled in strict accordance with good clinical practices, and all efforts were made to minimize suffering.

#### *T. gondii* Isolates, Isolation of the TgShSp1, and Genotyping

The T. gondii type II ovine isolate TgME49 (genotype #1) isolated in 1958 from sheep muscle (Lunde and Jacobs, 1983) was kindly donated by Dr. J. C. Boothroyd and had an unknown passage number, but it had been routinely maintained in cell culture and mice. The T. gondii type II (genotype #3) isolate (TgShSp1) was obtained from a T. gondii ovine abortion outbreak in September and October 2015 in a Spanish sheep flock (Assaf breed) in the province of Palencia (northwest Spain), which suffered abortion in 30 out of 239 pregnant sheep (12.5%).

TgShSp1 was isolated by passage in mice from a brain of an ovine aborted fetus, in which T. gondii infection was confirmed by PCR within 24 h after collection as described (Regidor-Cerrillo et al., 2008). Briefly, fetal brain (6 g) was homogenized in 6 mL of PBS containing 2% antibiotic-antimycotic solution (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), filtered in sterile gauze, and centrifuged at 1,350 g for 15 min. The supernatant was discarded, and the sediment was suspended in 1,400 µL of PBS with antibiotic-antimycotic solution, and 400 µL was inoculated subcutaneously into one 8-weeks-old female CD1 mouse that was followed for clinical signs. At day 40 pi, asymptomatic mouse were euthanized, and the brain collected and tested by PCR (see below). The PCR-positive mouse brain was homogenized in 1,000 µL of PBS with antibiotics by passing through a descending series of needles (20–25 G) and immediately subcutaneously inoculated into two other CD1 mice (500 µL to each mice). At 11 dpi, mice were euthanized, and peritoneal flushes were used for isolation in MARC-145 cell culture (Regidor-Cerrillo et al., 2008) and checked for the presence of parasite by PCR. Toxoplasma gondii tachyzoites were observed in cell cultures at 4 days after inoculation of peritoneal flushes on MARC-145 cells and maintained by successive passages until cryopreservation.

The T. gondii isolate obtained was genotyped by polymerase chain reaction–restriction fragment length polymorphism (PCR-RFLP) using 12 molecular markers (SAG1, 3′ -SAG2, 5′ -SAG2, Alt.SAG2, SAG3, BTUB, GRA6, c22-8, c29-2, L358, PK1, and Apico) as previously described (Su et al., 2006). The digested PCR products were visualized by 2.5% agarose gel electrophoresis, stained with Gel Red <sup>R</sup> Nucleic Acid Gel Stain (Biotium <sup>R</sup> , Fremont, California, USA), observed under UV light, assigned to a T. gondii type and classified according to genotypes present in ToxoDB (http://toxodb.org/toxo/).

#### *In vitro* Assays

MARC-145, Vero and Human Foreskin Fibroblasts (HFF) cell lines were used for studying the tachyzoite-to-bradyzoite conversion and proliferation of T. gondii isolates in vitro. Cells were routinely maintained in DMEM (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) with phenol red supplemented with 10% heat-inactivated, sterile, filtered fetal calf serum (FCS) (Gibco, Thermo Fisher Scientific, Waltham, MA, USA), 2 mM glutamine (Lonza Group, Basel, Switzerland) and a mixture of penicillin (100 U/ml), streptomycin (100µg/ml), and amphotericin B (Lonza Group, Basel, Switzerland) at 37◦C in a humidified atmosphere of 5% CO2. TgME49 and TgShSp1 were maintained by serial passages in MARC-145 and Vero cells in the same culture medium with 2% FCS. Tachyzoites used for in vitro assays were harvested 3 days pi (TgME49) or 5 days pi (TgShSp1, passage 10), when the majority of parasites were still intracellular and purified by a PD-10 column (GE Healthcare, Little Chalfont, United Kingdom), as described (Regidor-Cerrillo et al., 2011). Tachyzoite viability was confirmed by trypan blue exclusion, and numbers were determined by counting in a Neubauer chamber. All assays were carried out in triplicate, including at least three replicates in each assay.

#### Evaluation of the Tachyzoite-to-Bradyzoite Differentiation in vitro Through Cyst Wall-Specific DBL Staining

Purified T. gondii parasites of the TgME49 or TgShSp1 isolate (2 × 10<sup>3</sup> tachyzoites) were added to MARC-145 monolayers grown to confluence in 24-well plates. At 24 h, culture medium was replaced by DMEM with alkaline pH (8–8.2), and plates were incubated at 37◦C without CO<sup>2</sup> supplementation for 3–4 days (Skariah et al., 2010) to evaluate induced differentiation. Duplicate plates were maintained for 3 days under regular conditions to evaluate spontaneous differentiation (pH 7.2– 7.4 and 5% CO2). Tachyzoite-to-bradyzoite conversion was evaluated at 72–96 h by double immunofluorescence staining in cell monolayers fixed with paraformaldehyde 3% and glutaraldehyde 0.05% and permeabilized with Triton X-100 0.25%, using a polyclonal mouse-anti T. gondii antiserum at a dilution of 1:100 as primary antibody, Alexa Fluor <sup>R</sup> 594 Goat Anti-Mouse IgG (H + L) (Life technologies, Carlsbad, CA, USA) at a dilution of 1:1,000 as secondary antibody for parasite staining, and the Dolichos biflorus lectin (DBL) (Vector Labs, Burlingame, United States) at a dilution of 1:50 for cyst wall staining. Cell nuclei were stained with DAPI. Finally, the total numbers of DBL-positive cysts and DBL-negative structures compatible with parasite structures, including lysis plaques, were counted using an inverted fluorescence microscope (Nikon Eclipse TE200) at 200x magnification. The percentage of conversion for each well was determined.

#### In vitro Intracellular Proliferation Assays

Toxoplasma gondii proliferation was evaluated in Vero cells by a plaque assay and in HFF determining the tachyzoite yield at 48 h pi. For this assay, Vero cultures grown to confluence in 24-well plates were infected with 5 × 10<sup>4</sup> purified tachyzoites of either TgME49 or TgShSp1 and further maintained at 37◦C and 5% CO<sup>2</sup> for 4 days and stained with 0.2% crystal violet (Alfa Aesar, Haverhill, Massachusetts, United States) solution in 2% ethanol (Ufermann et al., 2017). Images were captured using a SMZ1000 binocular loupe (Nikon <sup>R</sup> , Tokyo, Japan).

Tachyzoite yield in HFF was determined by quantifying the number of tachyzoites at 48 h pi (TY48h) by real-time PCR (qPCR). HFF cultures grown to confluence in 24-well plates were infected with 10<sup>5</sup> purified T. gondii tachyzoites and maintained for 48 h at 37◦C in 5% CO<sup>2</sup> as previously described (Regidor-Cerrillo et al., 2011). Then, the medium was removed, and cells were recovered in 150 µL lysis buffer and 10 µL proteinase K (Macherey-Nagel, Düren, Germany) for DNA extraction and quantification of parasite genomic DNA by qPCR.

#### *In vivo* Experimental Infections

#### Virulence Assessment in Non-pregnant Mice Intraperitoneally Inoculated With TgShSp1 and TgME49 Tachyzoites

TgShSp1 and TgME49 virulence was determined in mice following recommendations for standardization described by Saraf et al. (2017). For the in vivo challenge, tachyzoites of TgShSp1 (passaged 10 times in cell culture) and TgME49 (unknown passage number) were recovered from Vero cultures when they were still largely intracellular (>80% of undisrupted parasitophorous vacuoles) (Regidor-Cerrillo et al., 2010), repeatedly passed through a 27-gauge needle at 4◦C and filtered through a 5-µm polycarbonate filter (IpPORE <sup>R</sup> , IT4IP, Louvainla-Neuve, Belgium) (Saraf et al., 2017). The tachyzoite viability was determined by Trypan blue exclusion and counted in a Neubauer chamber. Tachyzoite 10-fold serial dilutions were performed starting from 10<sup>5</sup> to 1 tachyzoite(s) suspended in 200 µL of PBS, and each dilution was inoculated intraperitoneally into five 8-weeks-old female CD1 mice (Janvier-Labs, Laval, France) within 30 min of harvesting the parasites from cell culture. Five control female mice were inoculated with PBS. Mice were observed daily and clinical signs (morbidity) were scored according to the description made by Arranz-Solis et al.

(2015a). Briefly, scores were classified as 0 (no alterations), 1 (ruffled coat), 2 (rounded back), 3 (noticeable loss of body condition/severe weight loss), or 4 (nervous signs such as activity decrease, hind limb paralysis, walking in circles, or head tilt). As a humane endpoint, mice exhibiting evident loss of body condition (score of 3) or nervous signs (score of 4) were culled to limit unnecessary suffering. Mice with clinical scores of 0, 1, and 2 were euthanized at 6 weeks p.i. Samples of blood were collected from mice for serology by IFAT and brain and lung for parasite detection by PCR.

#### Generation, Purification, and Sporulation of T. gondii Oocysts

Oocysts of the TgShSp1 isolate were obtained through oral infection of cats as previously described (Müller et al., 2017). Briefly, ten 8-weeks-old female CD1 mice (Janvier-Labs, Laval, France) were inoculated intraperitoneally with 10<sup>5</sup> tachyzoites of TgShSp1 (passage 10). At 2 months post-inoculation, mice were euthanized, and the brains were collected. Two 12-weeksold kittens free of T. gondii and other relevant feline pathogens (Isoquimen S.L., Barcelona, Spain) were fed a pool of 5 brains each. Feces were collected from kittens daily and examined to detect shedding of T. gondii oocysts. Unsporulated oocysts were harvested from feces and sporulated by resuspending in 2% H2SO<sup>4</sup> for 4 days at room temperature. Sporulated oocysts were kept at 4◦C until used. The same batch of sporulated oocysts of the TgME49 and TgShSp1 was used in mouse and sheep infections. Sporulated oocysts of TgME49 originated from the same batch as described earlier (Müller et al., 2017).

#### Assessment of TgShSp1 Oocyst Infection in Pregnant Mice

TgShSp1 was evaluated in pregnant mice out similarly as previously described for TgME49 (Müller et al., 2017). CD1 females (50 mice) and males (25 mice) were purchased from Charles River Laboratories (Sulzberg, Germany) at the age of 8 weeks and were maintained in a common room under conventional day/night cycle housing conditions. Females at 9 weeks of age were synchronized with respect to estrus and were distributed into cages, where two females and one male were housed together for 3 days (during which 3 females died). Subsequently, the female mice were orally infected by gavage with high doses of oocysts: 2,000 oocysts (group A, n = 9) and 500 oocysts (group B, n = 9), an intermediate dose of oocysts: 100 oocysts (group C, n = 10) and a low dose of oocysts: 25 oocysts (group D, n = 10) suspended in 100 µL of carboxymethyl cellulose solution (0.5% in water) at day 7 post-mating. The control group (group E, n = 9) received carboxymethyl cellulose solution alone. Pregnancy was confirmed 2 weeks post-mating by weighing, and pregnant mice were then allocated into single cages to give birth on days 19–22 and to rear their pups for an additional 4 weeks. During this time, those females that had remained non-pregnant were maintained in cages of three to five mice. Dams and their offspring were evaluated daily from birth to day 28 post-partum (pp). Despite the numerous parameters evaluated, pup mortality (number of pups born dead or euthanized due to severe clinical signs as described above for intraperitoneal inoculation) and vertical transmission (surviving pups being PCR-positive in the brain) were the most relevant assessments. Data on pregnancy rate (percentage of female mice that became pregnant), litter size (number of delivered pups per dam), and clinical signs (morbidity) of dams and non-pregnant mice were recorded during this time as described above for intraperitoneal inoculation. Neonates were weighed every second day from day 14 pp until the end of the experiment (day 28 pp) to evaluate morbidity in the offspring. Day 14 pp was chosen as a starting point for weight monitoring to avoid excessive handling of the pups during the first 2 weeks after birth, which can result in rejection by the dams. Dams, non-pregnant mice and pups were euthanized in a CO<sup>2</sup> chamber at 28 days pp. Blood from dams and non-pregnant mice was recovered by cardiac puncture, and sera were obtained to test humoral immune responses. Brains and lungs were removed from dams and non-pregnant mice and stored at −20◦C until determination of parasite load. The heads of pups that survived were collected and stored at −20◦C. Subsequently, the frozen heads were cleaved and brains were removed. The frozen brains from pups were immediately processed for DNA purification and then parasite quantification. Whenever possible, dead pups succumbing to the infection early after birth were removed, their heads sampled and their brains analyzed.

#### Assessment of TgShSp1 and TgME49 Oocyst Infections in Pregnant Sheep

Fifty-four pure Rasa Aragonesa breed female ewes aged 12 months were selected from a commercial flock. All animals were seronegative for T. gondii, N. caninum, border disease virus (BDV), Schmallenberg virus (SBV), Coxiella burnetii, and Chlamydia abortus as determined by enzyme-linked immunosorbent assay (ELISA). They were estrus-synchronized and mated with pure-bred Rasa Aragonesa tups for 2 days, after which the rams were separated from the ewes. Pregnancy and fetal viability were confirmed by ultrasound scanning (US) on day 40 post-mating. Pregnant ewes (n = 37) were randomly distributed into seven experimental groups and housed at the Instituto de Ganadería de Montaña (CSIC-Universidad de León), León, Spain.

Thirty-three pregnant ewes were orally dosed on day 90 of pregnancy with a high dose of oocysts (500 sporulated oocysts) of TgShSp1 (group 500A, G500A, n = 6) or TgME49 (group 500B, G500B, n = 5), an intermediate dose of oocysts (50 oocysts) of TgShSp1 (group 50A, G50A, n = 6) or TgME49 (group 50B, G50B, n = 5) or a low dose of oocysts (10 oocysts) of TgShSp1 (group 10A, G10A, n = 6) or TgME49 (group 10B, G10B, n = 5). The four remaining sheep were used as negative controls of infection (uninfected) and received 50 mL of PBS on day 90 of pregnancy.

Pregnant ewes were observed daily throughout the experimental period. Rectal temperatures were recorded daily from day 0 until 14 days pi and then weekly to evaluate morbidity. The physiological range for rectal temperatures in sheep was obtained from Diffay et al. (2002), and rectal temperatures above 40◦C were considered hyperthermic. Fetal viability was assessed by US monitoring of fetal heartbeat and movements twice a week after infection. When fetal death occurred, or immediately after parturition, dams and lambs were first sedated with xylazine (Rompun, Bayer, Mannheim, Germany) and then euthanized by an intravenous overdose of embutramide and mebezonium iodide (T61, Intervet, Salamanca, Spain). Animals from the uninfected group were examined by US every 2 weeks.

According to the survival in fetuses/lambs, sheep were classified into three categories: (a) suffering early abortions (i.e., between 8 and 11 dpi); (b) suffering late abortions, which occurred from 12 to 50 dpi; and (c) sheep delivering stillbirths, mummified fetuses or live lambs from 51 dpi. After birth, lambs were clinically inspected and then sedated and euthanized. Lambs showing weakness in relation to all live lambs were used to calculate morbidity in the offspring. In spite of the numerous parameters evaluated, fetal/lamb mortality and vertical transmission in live lambs (seropositivity and parasite detection in brain or lung) were the most relevant assessments.

Blood samples to evaluate humoral immune responses were collected prior to infection, at 3, 5, 7, and 10 days pi and then weekly by jugular blood draw. Precolostral serum was collected from lambs immediately after delivery from dams. To prevent any transmission of colostral antibodies from dams, udders were covered with a piece of cloth 1 week before the expected date of delivery as a preventive measure, and lambs were separated from their mothers immediately after birth. Serum samples were stored at −80◦C until analysis.

During necropsy, six randomly selected placentomes or cotyledons from aborted dams and dams that gave birth, respectively, were recovered from each placenta, transversally cut into 2–3 mm-thick slices, and fixed in 10% formalin for histopathological examination, whereas the remaining tissues from these placentomes/cotyledons were stored at −80◦C for further DNA extraction and PCR analyses. Samples from fetal tissues, including brain and lungs, were stored at −80◦C for DNA extraction or were fixed in 10% formalin for histopathology. Thoracic and abdominal fluids were also collected from fetuses and stillborn lambs from which precolostral sera could not be obtained, and maintained at −80◦C for serology.

#### Serological Analyses: IFAT and ELISA

The serum samples from mice used for isolation and determination of virulence were analyzed by the immunofluorescence antibody test (IFAT) for the detection of anti-T. gondii IgG as previously described (Alvarez-Garcia et al., 2003), using an anti-mouse IgG conjugated to FITC (Sigma-Aldrich, Madrid, Spain) diluted 1:64 in Evans Blue (Sigma-Aldrich). We used the cut-off of 1:25. Serum titers for T. gondii in oocyst-infected mice were assessed by ELISA as previously described for Neospora caninum-infected mice (Debache et al., 2008, 2009), except that soluble antigen extract from T. gondii tachyzoites was used (Alaeddine et al., 2005).

Toxoplasma gondii-specific IgG antibody levels in sheep were measured using an in-house indirect ELISA similarly as previously described (Castaño et al., 2014). The indirect fluorescent antibody test (IFAT) was used to detect specific IgG anti-Toxoplasma antibodies in fetal fluids and precolostral sera, adapting the technique previously described for IFAT analysis in N. caninum-infected animals (Alvarez-Garcia et al., 2003), using an anti-sheep IgG (Sigma-Aldrich) diluted 1:200 in Evans blue (Sigma-Aldrich). Fetal fluids and precolostral sera were diluted at 2-fold serial dilutions in PBS starting at 1:8 (for fetal fluids) and 1:50 (for precolostral sera) up to the endpoint titer. Continuous tachyzoite membrane fluorescence at a titer ≥8 for fetal fluids or ≥50 for precolostral sera was considered a positive reaction.

#### DNA Extraction and PCR for Parasite Detection and Quantification in Tissues

Genomic DNA from in vitro samples was extracted from these samples using the NucleoSpin <sup>R</sup> DNA RapidLyse Kit (Macherey-Nagel, Düren, Germany) according to the manufacturer's instructions. DNA concentrations were adjusted to 20 ng/µL and quantified using qPCR with primer pairs for the 529-bp repeat element for T. gondii for parasite quantification and primer pairs for the 28S rRNA gene for quantify cell DNA under conditions previously described (Collantes-Fernández et al., 2002; Castaño et al., 2016, respectively).

Genomic DNA was extracted from mice that were used for isolation and determination of virulence out using the commercial Maxwell <sup>R</sup> 16 Mouse Tail DNA Purification Kit. The T. gondii DNA detection was carried out by an ITS-1 PCR adapted to a single tube as previously described (Castaño et al., 2014). DNA extraction and qPCR analysis from oocyst-infected mice were performed as previously described (Müller et al., 2017).

In sheep, genomic DNA was extracted from three 50–100 mg samples taken from each location: six placentomes in aborted dams or six cotyledons in dams that gave birth, as well as fetal brain and lung, using the commercial Maxwell <sup>R</sup> 16 Mouse Tail DNA Purification Kit. T. gondii DNA detection was carried out by an ITS-1 PCR as described above (Castaño et al., 2014). DNA that tested positive by nested-PCR was adjusted to 20 ng/µL and quantified using qPCR as previously described (Castaño et al., 2014). Parasite number in tissue samples (parasite burden) was expressed as parasite number/mg ovine tissue. Standard curves for T. gondii and sheep DNA showed an average slope of −3.44 and −3.30, respectively, and an R <sup>2</sup> > 0.99. Parasite-negative DNA samples were included in each round of DNA extraction and PCR as negative controls.

#### Histological Processing

After fixation for 5 days, placental and fetal sheep tissues were cut coronally, embedded in paraffin wax and processed by standard procedures for hematoxylin and eosin (HE) staining. Conventional histological evaluation was carried out on all sections. To quantify the lesions in the brain of stillborn lambs and live lambs, the number and size of glial foci, as well as the total area of lesion in the examined tissue, were calculated through a computer-assisted morphometric analysis on HEstained sections following the procedure described previously (Arranz-Solis et al., 2015b).

# Statistical Analysis

The growth rate and percentage of DBL-positive cysts in vitro of TgME49 and TgShSp1 were compared using the Mann–Whitney test. In pregnant mice, differences in seroconversion, pregnancy rates, litter size, pup mortality, and parasite presence in tissues were analyzed by the χ 2 -test or Fisher's exact F-test. One-way ANOVA followed by Tukey's multiple comparisons test were employed to compare body weights. Parasite burdens and anti-T. gondii antibody levels were analyzed using the non-parametric Kruskal–Wallis test followed by Dunn's test for comparisons between groups, as well as the Mann–Whitney test for pairwise comparisons.

In pregnant sheep, the number of fetuses/lambs suffering mortality and the number of weak lambs (morbidity) were compared using the χ 2 -test or Fisher's exact F-test. Rectal temperatures and humoral immune responses were analyzed using one-way ANOVA followed by Tukey's multiple comparisons test until 14 days pi or until the end of the experiment. Differences in frequency of PCR detection of parasite DNA and in the percentage of cases showing lesions were evaluated using the χ 2 -test or Fisher's exact F-test. Differences in parasite burdens and histological measurements of lesions were analyzed using the non-parametric Kruskal–Wallis test followed by Dunn's test for comparisons between groups, as well as the Mann–Whitney test for pairwise comparisons.

Differences between mice and sheep in the number of fetuses/pups/lambs that died in relation to the total number of fetuses/pups/lambs or in the number of surviving offspring infected with T. gondii in relation to all live offspring were assessed using the χ 2 -test or Fisher's exact F-test. Likewise, a categorization of the parameters to evaluate congenital infection was done, into high (>67%), medium (66–34%), low (<33%), or none (0%) of the fetuses/pups/lambs with clinical signs, offspring mortality or vertical transmission. Statistical significance for all analyses was established at P < 0.05. All statistical analyses were performed using GraphPad Prism 6.01 software (San Diego, CA, USA).

# RESULTS

### Isolation of the *T. gondii* Isolate TgShSp1 From a Sheep Flock in Spain

On day 40 post-inoculation, one mouse inoculated with the brain homogenate of the sheep abortion was T. gondii PCR-positive in the brain. The peritoneal fluid of one of the other two mice inoculated with the positive mouse brain was PCR-positive on day 11 pi. Four days after inoculating the PCR-positive peritoneal flush into cell culture, the isolation of TgShSp1 was confirmed. Genotyping classified TgShSp1 as genotype #3 (a type II variant, type II for nine alleles/type I for Apico) (ToxoDB).

# TgShSp1 and TgME49 Differ in Behavior *in vitro*

#### TgShSp1 Exhibits a High Capacity to Form Cysts in vitro

At difference in TgME49 free-floating cyst-like structures was often identified by light microscopy in the TgShSp1 infected cultures at 3 days p.i. in successive passages. TgShSp1 cultured under regular conditions (at a neutral pH) demonstrated spontaneous conversion to bradyzoite with a statistically higher number of DBL-positive cysts (14%) compared to TgME49 (2%) (P < 0.0001). Additionally, after induction of bradyzoite development (at a basic pH), TgShSp1 formed a higher number of DBL-positive cysts (55%) compared to TgME49 (33%) (P < 0.0001) (**Figure 1A**).

#### TgME49 Shows a Higher Growth Rate in vitro Than TgShSp1

Evaluation of parasite growth in Vero cells by a lysis plaque assay showed that TgME49 produced large clear zones due to host cell lysis, while during the same period, TgShSp1 showed essentially an intact monolayer cell (**Figure 1B**). Determination of the TY48h in HFF was also assessed to confirm differences in parasite growth. The TY48h values for TgME49 were significantly higher compared to those from TgShSp1 (P < 0.0001) (**Figure 1C**).

# TgShSp1 and TgME49 Differ Greatly in Virulence in Mice

A summary of clinical signs, serology and parasite detection in mice is shown in **Table S1**. Most of the mice inoculated with doses from 10<sup>5</sup> to 10<sup>2</sup> tachyzoites of TgShSp1 only exhibited a ruffled coat between days 4 and 13 pi, but they did not have to be euthanized due to severe clinical signs (**Figure 2A**; **Table S1**). In contrast, upon infection with TgME49, several mice had to be euthanized due to clinical scores (**Figure 2B**). The surviving mice infected with doses from 10<sup>5</sup> to 10 tachyzoites of TgME49 exhibited clinical signs (rounded back) between days 8 and 14 pi (**Table S1**). The LD<sup>50</sup> for TgME49 was approximately 10<sup>3</sup> tachyzoites vs. >10<sup>5</sup> tachyzoites for TgShSp1.

All mice inoculated with doses from 10<sup>5</sup> to 10 tachyzoites of TgShSp1 were seropositive at 6 weeks pi, with IFAT titers ranging from 1:400 to 1:3,200. Concerning mice infected with TgME49, mice euthanized prior to day 24 pi were seronegative, while one mouse infected with 10<sup>3</sup> tachyzoites euthanized on day 33 pi had an IFAT titer of 1:25. Survivors infected with doses from 10<sup>5</sup> to 10 tachyzoites of TgME49 were seropositive at 6 weeks pi, with IFAT titers ranging from 1:50 to 1:800 (**Table S1**).

All mice infected with doses from 10<sup>5</sup> to 10 tachyzoites of TgME49 and TgShSp1 were PCR-positive in the brain, confirming T. gondii infection. Likewise, almost all lung samples from mice infected with doses from 10<sup>5</sup> to 10 tachyzoites of both isolates were PCR-positive, except lung samples from surviving mice infected with 10 tachyzoites of TgME49. Mice infected with 1 tachyzoite did not show clinical signs, they were seronegative, and all tissue samples were PCR-negative, identical to uninfected mice (**Table S1**).

# TgShSp1 Shows Low Virulence in Mice Infected With Oocysts but Is Efficiently Transmitted to Offspring

#### Evaluation of TgShSp1 Infection in Dams

Clinical signs in infected dams were generally mild. Therefore, none of the dams had to be euthanized due to severe clinical signs. At 12 days pi, one out of five dams infected with 2000

violet-stained Vero monolayer background. (C) A column-plot graph representing the tachyzoite yield (TY48h) of TgME49 and TgShSp1. Values of replicates from experiments performed in triplicate for each isolate. Error bars indicate the SD. \*\*\*\*Marks the significantly higher TY48h values for TgME49 compared to TgShSp1.

oocysts (group A) exhibited ruffled coat (1), and one out of six dams infected with 500 oocysts (group B) displayed rounded back (2). No clinical signs were observed in dams infected with 100, 25, or 0 oocysts (groups C, D, and E). Pregnancy rates ranged from 55 to 66%, with no significant differences between them. Similarly, no differences between the groups were found in litter size (11.8–14.4 delivered pups), suggesting that pregnancy

FIGURE 2 | Survival curve of CD1 mice after infection with T. gondii tachyzoites of TgShSp1 isolate (A) or TgME49 isolate (B). Five mice per group were infected i.p. with 10<sup>5</sup> , 10<sup>4</sup> , 10<sup>3</sup> , 10<sup>2</sup> , 10, or 1 tachyzoite of the TgShSp1 isolate or the TgME49 isolate. Survival was monitored for 42 days. Each point represents the percentage of surviving animals at that day, and downward steps correspond to euthanasia due to severe clinical signs.

was not noticeably altered by infection with TgShSp1 oocysts (**Table 1**).

All infected dams with 2,000, 500, and 100 oocysts (groups A, B, and C) developed Toxoplasma-specific humoral immune responses at day 28 pp. However, although only seroconversion in half of the dams was observed in the group infected with 25 oocysts (group D), there were no statistically significant differences in the number of dams showing seroconversion between infected groups (**Table 1**). Anti-T. gondii IgG levels were significantly increased in groups infected with 2,000 (P < 0.05), 500 (P < 0.01), and 100 oocysts (P < 0.01) in comparison to the unchallenged group in which all dams were seronegative (**Figure S1A**). Toxoplasma gondii DNA was detected in the brain of all dams from infected groups, with the exception of three dams in the group infected with 25 oocysts (group D) (**Table 1**). Quantitative evaluation of parasite burdens in brain showed no significant differences between infected groups (**Figure S2A**). In the lungs, parasite DNA was detected in the 50–83% of the dams from oocyst-infected groups, without significant differences in parasite detection or parasite load between them (**Figure S3A**; **Table 1**).

#### Evaluation of TgShSp1 Infection in Offspring Mice

Most pups were born dead (82/93; 88%), although a few pups had to be euthanized due to severe clinical signs between day 2 and day 21 pp (11/93; 12%). Half of the pups in the group infected


with 2,000 oocysts (group A) and with 500 oocysts (group B) were born dead or had to be euthanized due to severe clinical signs and had significantly higher pup mortality compared to the uninfected group (P < 0.05) (**Table 1**). In groups infected with 100 oocysts (group C) and 25 oocysts (group D), respectively, 9 and 27% of pups were born dead or had to be euthanized due to severe clinical signs. Only one of seventy-two pups was born dead in the uninfected group (group E) (**Table 1**). Starting from day 14 pp, offspring of the uninfected group (group E) showed significantly higher body weight than groups infected with 500 (P < 0.0001) and 100 oocysts (P < 0.001), and the same was true of those infected with 2,000 oocysts (group A) from day 22 pp (P < 0.001). However, no decreased body weight was noted in pups infected with 25 oocysts (group D) (**Figure 3A**).

Vertical transmission was detected upon PCR analyses in all brains (100%) of surviving pups from groups infected with 2,000, 500, and 100 oocysts (groups A, B, and C). However, parasite was only detected in 37% of the brains from pups infected with 25 oocysts (group D), with significantly lower parasite detection compared to groups infected with 2,000, 500, and 100 oocysts (P < 0.0001) (**Table 1**). Pups infected with 25 oocysts (group D) and with 100 oocysts showed lower parasite burden compared to those infected with 2,000 and 500 oocysts (groups A and B) (P < 0.0001) (**Figure 3B**). No T. gondii DNA could be detected in the brain of pups that had died on day 0 or 1 pp.

#### Evaluation of TgShSp1 Oocyst Infection in Non-pregnant Mice

Similarly, to pregnant mice, clinical signs in non-pregnant mice infected with oocysts were generally mild. Therefore, none of the non-pregnant mice had to be euthanized due to severe clinical signs. Only ruffled coat was observed in all mice infected with 2,000 oocysts (group A), in one out of three mice in the group infected with 500 oocysts (group B) and in one out of four non-pregnant mice in the group infected with 100 oocysts (group C). All infected non-pregnant mice infected with 2,000, 500, and 100 oocysts (groups A, B, and C) developed Toxoplasma-specific humoral immune responses at day 28 p.p. that were significantly increased in comparison to the unchallenged group, with basal IgG levels (P < 0.05) (**Figure S1B**). In the group infected with 25 oocysts (group D), only three of the four non-pregnant mice seroconverted (**Table 1**). Further analyses of antibody responses of pregnant and non-pregnant mice in each group did not reveal any significant differences. Toxoplasma gondii DNA was detected in the brains of all non-pregnant mice from infected groups, with the exception of one mouse infected with 25 oocysts (group D) (**Table 1**). In the lungs, parasite DNA was only detected in three out of four samples from mice infected with 100 oocysts (group C). Quantitative evaluation showed no significant differences in brain and lung parasite loads in non-pregnant mice (**Figures S2B**, **S3B**). The comparison of parasite load in brain and lungs in pregnant mice vs. non-pregnant mice revealed no significant differences.

 of pups surviving at day 28 pp (percentage).

\*P < 0.05 and \*\*\*\*P < 0.0001 significant differences.

#### TgShSp1 and TgME49 Oocyst Infection Cause Similar Fetal/Lamb Mortality and Vertical Transmission in Pregnant Sheep Clinical Observations

individual values of parasite burden (number of parasites per µg of DNA), and

medians are represented as horizontal lines. \*\*\*\*P < 0.0001.

No mortality was found in any sheep during the experiment. All ewes showed fever after infection. Significant increases in body temperature were found for 3–4 days after day 4 p.i. in all infected groups. From day 14 pi until the end of the experiment, no changes were detected in the infected groups. Likewise, compared to groups infected with 500 oocysts, 1 day of delay in the increase of rectal temperature (from day 5 to 6 pi) was found in groups infected with 50 and 10 oocysts (**Figure 4**). Differences in temperature increase between groups receiving the same dose of oocysts of the different T. gondii isolates were generally not found. As an exception, lower rectal temperatures were only found on day 8 pi in groups infected with 500 and 10 oocysts of TgShSp1 compared to groups receiving the same

euthanized, and their data are not available. Each point represents the mean + S.D. at the different sampling times for each group.

doses of TgME49 oocysts (G500A vs. G10A and G500B vs. G10B) (P < 0.05). The mean rectal temperature in the uninfected group remained below 39.5◦C throughout the monitoring period.

Fetal/lamb mortality in groups receiving 500 oocysts was 100%. In groups receiving 50 and 10 oocysts, 68% of fetuses/lambs in G50A (in 6/6 ewes), and 66% in G10A (in 4/6 ewes) died after infection with TgShSp1, and 42% (in 2/5 ewes) of them died in both groups after infection with TgME49. Fetal/lamb mortality was increased with the oocyst doses. A lower fetal/lamb survival rate was found in the group infected with 500 TgME49 oocysts compared with that infected 50 oocysts (G500B vs. G50B) (P < 0.05). In addition, a significantly lower fetal/lamb mortality was found in those infected with 500 oocysts compared with those infected with 10 oocysts for both isolates (G500A and G500B compared to G10A and G10B, respectively) (P < 0.05). No differences in fetal/lamb mortality were found between groups infected with 50 and 10 oocysts (G50A and G50B compared to G10A and G10B, respectively). Comparing groups that received the same dose of sporulated oocysts but different isolates, no significant differences were found in the fetal/lamb survival rate. Concerning fetal death during pregnancy, all ewes challenged with 500 oocysts (G500A and G500B) aborted, and in


TABLE 2 | Fetal/lamb mortality in sheep and percentages of placentomes/cotyledons or fetuses/lambs showing histological lesions and parasite detection.

NA, not available.

the groups infected with lower doses, abortions were found in 3/6 and 2/5 pregnant ewes infected with 50 oocysts of TgShSp1 and TgME49, respectively, and in 1/6 and 1/5 pregnant ewes infected with 10 oocysts of TgShSp1 and TgME49, respectively (**Table 2**). Non-aborted dams gave birth between days 143 and 149 of pregnancy, except one ewe infected with 50 TgME49 oocysts, which gave birth prematurely on day 134. In G50A, 1 mummified fetus and 3 stillborn lambs were found, and in G50B, G10A, and G10B, one, eight and one stillborn lambs were delivered, respectively. Concerning morbidity in lambs born alive, weakness was found in 5 out of 5 and 1 out of 4 in groups infected with 50 oocysts, G50A and G50B, respectively, and in 2 out of 5 and 0 out of 4 in those infected with 10 oocysts, G10A and G10B, respectively. Therefore, morbidity in lambs born alive from the group infected with 50 TgShSp1 oocysts, G50A, was significantly higher than the corresponding TgME49 group, G50B (P < 0.05), whereas no significant differences in the number of weak lambs were found between lambs from groups infected with 10 oocysts or between lambs from groups receiving different doses of both isolates. Dams from the pregnancy control group gave birth two stillborn lambs and six healthy lambs between days 147 and 152 of pregnancy.

#### Parasite Detection and Burden in Placental and Fetal Tissues

#### **Placental tissues**

In ewes suffering early abortions, no T. gondii DNA was detected in placentomes in TgME49-infected animals (i.e., G500B, G50B, and G10B), and in ewes infected with TgShSp1, parasite DNA was only detected in one ewe infected with 500 oocysts (G500A), which aborted on day 9 pi (one positive placentome sample out of 30), and in one ewe infected with 50 oocysts (G50A), which aborted on day 10 pi (one positive placentome samples out of 18). In contrast, all placentomes from ewes showing late abortion were PCR-positive. In ewes that delivered stillbirths or live lambs, all cotyledons from TgShSp1-infected animals were PCR-positive. Among those challenged with TgME49 oocysts, 100% of cotyledon samples were positive in the group infected with 500 oocysts (G50B), while 75% of cotyledon samples were positive in the group infected with 10 oocysts G10B (**Tables 2**, **S2**). Concerning parasite burden (measured as the number of tachyzoites per milligram of tissue) in cotyledons from ewes that delivered stillbirths or live lambs, no differences were found between groups infected with 50 and 10 oocysts in any of the isolates. However, comparing both T. gondii isolates, parasite loads in cotyledons from ewes that gave birth in groups infected with 50 and 10 oocysts of TgShSp1 (groups G50A and G10A), both were lower compared to those infected with TgME49 (G50B and G10B) (P < 0.05) (**Figure 5A**).

#### **Fetal tissues**

In tissues from fetuses undergoing early abortion upon TgME49 infection, no T. gondii DNA was detected. The same was true for fetuses from early abortions after challenge with TgShSp1 oocysts, except for one positive fetal lung sample (1 positive sample out of 20) from one ewe that aborted on day 10 pi in one group infected with 50 oocysts (G50A) (**Tables 2**, **S2**).

In late abortions, T. gondii DNA was detected in every fetus and all organs analyzed (**Table 2**). The percentage of positive samples in every individual organ ranged from 33% in the brain from one fetus in G10A to 100% in the rest of the brains and lungs from late abortions (**Table S2**).

FIGURE 5 | Dot-plot graphs of T. gondii burdens in cotyledons from ewes that gave birth (A) and brain (B) and lung (C) from stillborn lambs and live lambs from T. gondii-infected ewes. Each dot represents individual values of parasite burden (number of parasites per milligram of host tissue), and medians are represented as horizontal lines. Considering that the T. gondii detection limit by real-time PCR is 0.1 parasites, negative samples (0 parasites) were represented on the log scale as <0.1 (i.e., 10−<sup>2</sup> ). The unbroken line is used to indicate differences between isolates, and the dashed line (———) is used to indicate differences between doses. For significant differences between infected groups in each tissue, \*P < 0.05 and \*\*P < 0.01.

In all stillbirths and live lambs, T. gondii DNA was found in at least one of the studied organs, except in one stillborn lamb and one live lamb born from one ewe infected with 10 TgME49 oocysts (G10B) in which T. gondii DNA was not detected in any analyzed tissue (**Table S2**). Concerning parasite detection in the brain of stillbirths/live lambs born from ewes infected with TgME49 oocysts, more samples were PCR-positive in the group infected with 50 oocysts, G50B (91.6%; 11/12; 4 out of 4 fetuses) compared to the group infected with 10 oocysts, G10B (53.3%; 8/15; 3 out of 5 fetuses) (P < 0.05). In lambs from ewes infected with TgShSp1 oocysts, a lower number of brain samples were found to be PCR-positive in the group infected with 50 oocysts, G50A (33.3%; 5/15; 7 out of 9 fetuses), compared to the group infected with 10 oocysts, G10A (66.6%; 26/39; 13 out of 13 fetuses) (P < 0.05). Similarly, lower parasite burden in brain from lambs was found in the group infected with 50 TgShSp1 oocysts (G50A vs. G10A (P < 0.01). Comparing both isolates, parasite detection and parasite burden were higher in brain samples from lambs in the group infected with 50 oocysts of TgME49 compared to the corresponding TgShSp1 group G50B vs. G50A) (P < 0.01), while no significant differences were found between groups infected with 10 oocysts (G10A vs. G10B) (**Figure 5B**; **Table 2**; **Table S2**). In lung tissues from stillbirths/live lambs, all samples were PCR-positive in those groups infected with 50 oocysts (G50A and G50B). Additionally, 100% parasite detection was observed in the group infected with 10 oocysts of TgShSp1 (G10A), whereas a significantly lower parasite detection rate was found in the group infected with 10 oocysts of TgME49, G10B (60%; 9/15; 3 out of 6 animals) (P < 0.05). Comparison of the same oocyst dose from both isolates revealed no differences in parasite detection in lung samples from groups infected with 50 oocysts (G50B and G50A), but a higher parasite detection rate was found in the group infected with 10 TgShSp1 oocysts, G10A, compared to the group infected with 10 TgME49 oocysts, G10B (P < 0.001) (**Table 2**; **Table S2**). No differences in parasite burden in lungs from lambs were found between different doses or isolates (**Figure 5C**). Likewise, no differences in parasite detection or parasite burden were found in any fetal tissue between stillborn lambs and live lambs (data not shown). Samples from fetal tissues exhibiting DNA degradation and mummification were excluded from PCR analysis.

#### Histological Lesions and Lesion Quantification

The only evident histological lesions in the studied organs were found in the brain from fetuses/lambs and placenta. Only the placenta from late abortions detected through US was available for histological study. As in those cases of early abortions, lambing or delivery of stillbirths, it was too autolytic to allow proper histological evaluation.

In early abortions, multifocal areas of coagulative necrosis at the white matter (leukomalacia) were found in the brain from all the fetuses aborted in this period. In addition, no evident differences in the severity or number of lesions were noted between groups. In late abortions, lesions (multifocal necrotic placentitis) were found in all placentas studied. Likewise, there were brain lesions (multifocal non-purulent encephalitis) in all the fetuses from late abortions (**Table 2**).

In stillbirths/live lambs, glial foci with or without a central area of necrosis were observed in the brain. These lesions were found in lambs from all groups, with a prevalence between 60 and 100%, depending on the group (**Table 2**). When the percentages of brain lesions in lambs from ewes infected with TgME49 and TgShSp1 were compared, there were no differences between isolates or oocyst doses tested. Furthermore, there was no difference in the percentage of cases with brain lesions between stillbirths and live lambs (data not shown). Lesion quantification was carried out in brain samples from the stillbirths and live lambs, and no significant difference was found in the number of lesions, individual focus area or percentage of damaged area between groups (**Figure S4**).

(RIPC), according to the formula: RIPC = (OD405 sample – OD405 negative control)/(OD405 positive control – OD405 negative control) × 100.

#### Humoral Immune Responses

The Toxoplasma-specific IgG antibody responses in dams are shown in **Figure 6**. No increase in IgG level compared to the uninfected group and no seroconversion was found in any of the ewes showing abortion during the acute phase of the infection. However, ewes with late abortion or those giving birth seroconverted on day 21 pi. In the group infected with 500 TgShSp1 oocysts, 500A, all ewes except one suffered early abortions, so this group was excluded from statistical analysis. From day 21 pi onwards, ewes infected with 50 and 10 oocysts seroconverted and exhibited higher IgG compared to the control group (P < 0.05). When analyzing the IgG levels of animals infected with TgShSp1 sporulated oocysts, no significant differences were found between ewes infected with 50 and 10 oocysts, G50A and G10A (**Figure 6A**). However, in animals infected with TgME49 sporulated oocysts, it is noteworthy that the group infected with 10 oocysts, G10B, had higher IgG on day 21 pi than the group infected with 50 oocysts, G50B (P < 0.01) (**Figure 6B**). Comparing groups receiving the same dose of sporulated oocysts, no significant differences in IgG level were found between groups infected with 50 oocysts, G50A and G50B. However, the group infected with 10 oocysts of TgShSp1, G10A, displayed lower IgG than that infected with TgME49, G10B, from day days 21 to 35 pi (P < 0.01). All uninfected control animals exhibited basal IgG levels within the reference range throughout the experimental study.

None of the fetuses that were aborted before day 11 pi had detectable IgG against T. gondii antigen. In contrast, fetuses undergoing late abortions were IgG-positive. Of the lambs born from TgShSp1-infected ewes, seven out of eight and nine out of twelve lambs were positive in groups infected with 50 and 10 TgShSp1 oocysts, G50A and G10A, respectively. Similarly, of the lambs born from TgME49-infected ewes, two out of three and 50% of lambs born from those groups infected with 50 and 10 TgME49 oocysts, G50B and G10B, respectively, were positive (**Table S3**). Specific IgG responses against parasite antigen were not detected in lambs from the uninfected group.

#### Comparative Assessment of Congenital Infection in Mice and Sheep After Infection With TgShSp1 Oocysts

Offspring mortality after infection with 500 TgShSp1 oocysts occurred in sheep at a statistically higher rate compared to mice (P < 0.001), since in sheep all fetuses died and in mice only 50% of the pups died. Similarly, higher offspring mortality was found in sheep compared to mice after infections with intermediate (P < 0.0001) and low doses of oocysts (P < 0.01), since infection with 50 and 10 TgShSp1 oocysts triggered mortality in 68 and 66% of fetuses/lambs, whereas in mice mortality of 9 and 27% of the pups was observed after infection with 100 and 25 TgShSp1 oocysts, respectively.

Since no lambs were born in the group infected with 500 TgShSp1 oocysts, offspring morbidity in this group could not be assessed. However, in mice infected with 500 TgShSp1 oocysts, a decrease in pup body weight was found from day 14 pi onwards. Offspring from mice infected with 100 TgShSp1 oocysts showed a decrease in bodyweight from day 14 pi onwards, and all lambs born alive from ewes infected with 50 TgShSp1 exhibited weakness at birth. At low doses of oocysts, no body weight decrease was noted in pups born from mice infected with 25 TgShSp1 oocysts; however, 2 out 5 live lambs from ewes infected

with 10 TgShSp1 oocysts were born with obvious weakness and impaired health.

All (100%) of the surviving mouse pups were PCR-positive in the brain, and 100% live lambs were seropositive or with a PCR-positive result in at least one tissue after infection with intermediate doses of oocysts (100 TgShSp1 oocysts in mice and 50 TgShSp1 oocysts in sheep). Additionally, 100% of the surviving pups were PCR-positive in the brain after infection with 500 TgShSp1 oocysts (high dose of oocysts). However, a statistically higher number of T. gondii-positive offspring were found in sheep compared to mice after infection with low doses of oocysts (P < 0.05), since 100% lambs were seropositive or with a PCR-positive result in at least one tissue after infection with 10 TgShSp1 oocysts and only 37% of the surviving pups were PCR-positive in the brain after infection with 25 TgShSp1 oocysts (**Table 3**).

### DISCUSSION

Toxoplasma gondii is an apicomplexan parasite that is distributed worldwide (Dubey, 2010). In Europe and North America, T. gondii isolates display a clonal population structure, with the vast majority of T. gondii isolates being grouped into three lineages, namely, types I, II and III (Howe and Sibley, 1995). Type II T. gondii is the most prevalent in all hosts in Europe, including sheep (Dumètre et al., 2006; Halos et al., 2010; Su et al., 2010). Previous studies in Europe have shown that T. gondii type II is associated with ovine abortion (Owen and Trees, 1999; Jungersen et al., 2002; Chessa et al., 2014). In Spain, type II is the most prevalent genotype in wild animals and cats (Montoya et al., 2008; Calero-Bernal et al., 2015), as well as in previously obtained ovine isolates (Fuentes, 1999). The TgShSp1 isolate belongs to genotype #3 (a type II variant, type II for nine alleles/type I for Apico), sharing genotype with the Prugniaud (PRU) isolate.

Toxoplasma gondii PRU isolates exhibit a similar genetic pattern to the T. gondii type II reference isolate, TgME49 (genotype #1, type II for the studied alleles) (Su et al., 2012). In addition, both type II isolates, TgME49, and PRU, activate the host cell transcription factor NF-κB, an integral component of the immune response to T. gondii, and they display identical GRA15 gene sequences, which is involved in NF-κB activation (Rosowski et al., 2011). TgME49 was isolated from sheep muscle in 1958 (Lunde and Jacobs, 1983) and has since then undergone long-term passaging in cell culture and mice (Sibley et al., 2002). Previous studies have demonstrated changes in biological characteristics of T. gondii isolates after passages in mice and cell culture (Frenkel et al., 1976; Lindsay et al., 1991; Harmer et al., 1996; Saraf et al., 2017). This fact has been widely studied in T. gondii type I isolates (Cesbron and Sabin, 1994; Villard et al., 1997; Dubey et al., 1999; Mavin et al., 2004; Khan et al., 2009). However, whether these changes also occur in type II isolates, and how they compare to recently obtained isolates, remains unknown. Increased growth in vitro can be found after repeated passages (Yano et al., 1987). The dramatic differences observed during in vitro growth of TgME49 and TgShSp1 might reflect the highly different passage history of the two isolates. Plaque formation is commonly used to measure growth of T. gondii, and this process is the result of several events, including invasion, growth, egress, and migration (Roos et al., 1994). Notably, TgME49 tachyzoites formed plaques at 4 days pi, but TgShSp1 did not. The finding that TgShSp1 did not form plaques in vitro could have resulted from the limited growth rate, associated with a higher capacity of bradyzoite conversion, as previously described (Khan et al., 2009). This was confirmed by monitoring spontaneous cyst formation through labeling with the fluorescent lectin DBL, demonstrating tissue cyst formation under standard cell culture procedures in TgShSp1 that was greater than TgME49. Due to its high passage number in cell culture or mice, our TgME49 isolate may not accurately represent natural virulence traits of the type II lineage, which suggests that comparisons of phenotypes between T. gondii isolates should be conducted using low-passage stocks.

Traditionally, mouse models are utilized to evaluate virulence by monitoring survival after experimental infection. Type I isolates are highly virulent in mice (LD<sup>100</sup> of 1 tachyzoite), whereas types II and III exhibit median lethal doses (LD50) that range from 10<sup>2</sup> to 10<sup>5</sup> (Saeij et al., 2006). Conventionally, TgME49 is a cystogenic type II isolate with low virulence in mice (Ferreira et al., 2001; Gavrilescu and Denkers, 2001; Oliveira et al., 2016). Intraperitoneal inoculation of 10<sup>3</sup> and 5 × 10<sup>4</sup> TgME49 tachyzoites intraperitoneally has not caused mortality (Ferreira et al., 2001; Oliveira et al., 2016). However, in this study, TgME49 displayed a LD<sup>50</sup> of 10<sup>3</sup> , so the virulence of our TgME49 could be considered stronger than previous descriptions (Ferreira et al., 2001; Oliveira et al., 2016). Enhanced virulence in mice for T. gondii strains maintained for several passages has also been reported previously (Shimizu et al., 1967; Sibley and Boothroyd, 1992; Frenkel and Ambroise-Thomas, 1996). Hence, it seems logical to speculate that results from studies using laboratory isolates should be validated with more recent isolates before they can be extrapolated as general features of the respective lineage. In contrast to TgME49, mice inoculated with tachyzoites of the recently obtained type II isolate TgShSp1 exhibited only moderate, low-level clinical signs, but no mortality (LD<sup>50</sup> > 10<sup>5</sup> ), similar to what was described earlier after intraperitoneal inoculation of Swiss Webster mice with 10<sup>3</sup> tachyzoites of a PRU isolate (Wang et al., 2013). In addition, although oocysts are considered more virulent than tachyzoites in mice (Dubey and Frenkel, 1973), no mortality in adult mice was found after infection with oocysts of TgShSp1 in pregnant and non-pregnant mice, suggesting very low virulence in mice.

In the present work, we also investigated congenital toxoplasmosis in pregnant mice by inoculating them orally with different doses of TgShSp1 oocysts. The risk of congenital toxoplasmosis depends on the virulence of the parasite (Tenter et al., 2000). Based on the previously established toxoplasmosis model using TgME49 oocysts (Müller et al., 2017), we infected mice at day 7 post-mating, which represents the beginning of the second term of gestation. Few mice infected with TgShSp1 oocysts showed mild clinical signs, contrary to the large number of mice succumbing to infection after the same oocyst doses of TABLE 3 | Summary of the outcome of the congenital infection in mice and sheep after infection with different doses of TgShSp1 oocysts.


<sup>a</sup>Mortality of fetuses during pregnancy in sheep and of pups/lambs after birth in mice and sheep.

<sup>b</sup>Morbidity in the offspring was evaluated by decrease of the body weight of pups from day 14 pp in mice and by clinical signs in live lambs (weakness) at birth.

c In mice, PCR-positive brains in surviving pups at day 28 pp. In sheep, seropositive live lambs with a PCR-positive result in at least one tissue.

NA, not available. High, medium, low or minus (–) mean presence of clinical signs, offspring mortality, or vertical transmission of the parasite in >67, 66–34, <33, and 0% of the fetuses/pups/lambs, respectively. Body weights lower than the uninfected group are considered high morbidity, while no difference compared to uninfected group is represented as minus (–). \*P < 0.05, \*\*P < 0.01, \*\*\*P < 0.001, and \*\*\*\*P < 0.0001 significant differences between mice and sheep in the number of fetuses/pups/lambs died or in the number of surviving offspring infected with T. gondii.

TgME49 (Müller et al., 2017). There is a possibility that there has been a selection toward increased virulence within TgME49 due to the successive passages, as explained above but also due to the sulfadimidine treatment that was applied in mice used for infection of cats to generate TgME49 oocysts used in this study (Müller et al., 2017). This sulfonamide treatment could act as a bottleneck in selecting tachyzoites with a faster replication and therefore increasing the virulence in mice of the final TgME49 parasites. Unlike what was observed for TgME49 (Müller et al., 2017), with which a clear effect on pregnancy rate was found after infection with 2,000 oocysts and on litter size after infection with 500 oocysts, infection with TgShSp1 oocysts did not result in alteration of pregnancy rate or litter size. Likewise, while 92% of the pups died after infection of dams with 25 TgME49 oocysts (Müller et al., 2017), a significant decrease in pup survival was found only after infection with 2,000 and 500 TgShSp1 oocysts, where there was mortality in 50% of the pups. In conclusion, infection of mice with TgShSp1 oocysts at mid-pregnancy did not generate severe clinical signs in adult mice, but infection doses of 2,000 and 500 oocysts in the dams resulted in mortality of 50% of the pups and decreased body weight in surviving pups, and 100% vertical transmission occurred with doses of up to 100 oocysts.

Sheep are a relevant host of the parasite and could suffer abortions when primo infected during gestation (Vargas-Villavicencio et al., 2016). Fetal/lamb mortality was similar for both isolates. However, those ewes infected with 10 and 50 TgME49 oocysts and that delivered stillbirths/live lambs exhibited higher parasite load in cotyledons than those infected with the same doses of TgShSp1 oocysts. Similarly, a higher parasite load was found in the brain from lambs born in the group infected with 50 TgME49 oocysts compared to the corresponding TgShSp1 group. Therefore, it is tempting to hypothesize that the enhanced virulence of our TgME49 contributed to the abovementioned effects. Comparing different doses of infection, there is a correlation between the dose of infection and the rate of early abortions, as previously suggested (Mévélec et al., 2010; Benavides et al., 2017). Infection with 500 oocysts triggered abortion in all fetuses, similar to previous experimental infections in pregnant sheep at mid-pregnancy using 2,000 oocysts of M1 and M4 isolates (Owen et al., 1998; Castaño et al., 2014). After infection with 50 TgShSp1 oocysts or 50 TgME49 oocysts, 68 and 42% of fetuses/lambs died, respectively, similar to what was previously reported after infection with 50 M4 oocysts (Castaño et al., 2014). The occurrence of abortions after infection with 10 oocysts was low, but large numbers of stillbirths and weak lambs were found, mainly in the 10 TgShSp1 oocysts group. There seems to be a correlation between the presence of the parasite and the occurrence of stillbirths, since stillbirths from the group infected with 10 TgShSp1 oocysts exhibited higher parasite detection and load in the brain than those in the group infected with 50 TgShSp1 oocysts. Regardless of the isolate or dose, no differences were found in the congenital infection of lambs born, since vertical transmission was found in all them except in one stillborn lamb and one live lamb born from one ewe infected with 10 TgME49 oocysts. Likewise, no differences were found between doses or isolates with respect to brain lesion presence and lesion severity in lambs born.

Most of the experimental studies in pregnant sheep carried out so far used M1, M3, and M4 T. gondii type II isolates for infection (Dubey, 2009; Castaño et al., 2014). However, although numerous experimental infections were also carried out in mice with these isolates (Nicoll et al., 1997; Owen et al., 1998; Hamilton et al., 2018), their virulence in mice models has not been studied in depth. Therefore, the correlation between virulence in mice and outcome of experimental infections in pregnant sheep has not been elucidated. Despite the clear differences in body weight between mice and sheep, in the current study, similar doses of oocysts were used to compare both hosts. None of the adult mice challenged with 25 TgShSp1 oocysts exhibited clinical signs, whereas all ewes challenged with 10 TgShSp1 oocysts had fever. Therefore, morbidity in sheep seems to be higher than in mice. In addition, no mortality was observed in adult mice or sheep infected with TgShSp1 oocysts. When comparing the congenital infection after challenge at midpregnancy between both hosts, 50% mortality was caused in mice by infection with 500 TgShSp1 oocysts, whereas in sheep infection with the same oocyst dose caused mortality in all fetuses. In brief, mice seem to be less susceptible to offspring mortality than sheep, despite the fact that vertical transmission was similar in both species. High vertical transmission and low offspring mortality could be an evolutionary strategy of the parasite to generate a large infected offspring group in mice, one of the most relevant hosts of T. gondii (Müller and Howard, 2016).

There are several differences between mice and sheep that could underlie the differences found in this study. The histological structure of the placenta is very different between mice and sheep (Entrican, 2002), and although maternal blood and fetal tissue are closer in mice, allowing an easy crossing of tachyzoites but also of antibodies, the longer period of gestation, the lack of maternal antibodies crossing the placental barrier and fewer fetuses may facilitate vertical transmission in sheep. In addition, host genetics are likely important in determining susceptibility and severity of infection (Howe et al., 1996; Müller and Howard, 2016). Small rodents, natural intermediate hosts, are often exposed to a higher dose and more virulent parasites than other groups of mammals. It may therefore be that Toll-Like-Receptors (TLR)11 and TLR12 and the polymorphism of immunity-related GTPases (IRG proteins) have been positively selected in rodents, because of their critical importance in host resistance against high infection loads or more virulent clones of T. gondii. In mice, TLR11 and TLR12 on dendritic cells detect the apicomplexan actin-binding protein profilin leading to the secretion of interleukin 12 (IL12), which can subsequently induce production of IFNγ by T cells. IFNγ induces a variety of parasiticidal mechanisms, which in mice are dominated by upregulation of the IRGs (Gazzinelli et al., 2014). IRGs can destroy the vacuole in these parasites live and subsequently the parasite itself. Considering the ubiquity of T. gondii in nature, it is intriguing that genes encoding TLR11, TLR12, and IRG proteins are not found in many mammalian species (Gazzinelli et al., 2014). Although further studies are needed, and despite the influence of genetic polymorphisms in ovine abortions (Darlay et al., 2011), this fact could render sheep less resistant. Similarly, differences in immune cell populations may influence the pathogenesis of toxoplasmosis in these hosts. γδ T cells, which rapidly recognize and respond to non-processed antigens and seem to have an important role in T. gondii infection (Egan et al., 2005), represent a relevant subset of circulating T cells in sheep compared to mice (Holderness et al., 2013). Further studies are needed to characterize the cellular and molecular bases contributing to transmission dynamics and disease in different hosts of T. gondii (Dubremetz and Lebrun, 2012; Hunter and Sibley, 2012).

In conclusion, we have demonstrated that infection with tachyzoites and oocysts of the type II T. gondii isolate TgShSp1 in mice does not cause mortality, but this isolate is efficiently vertically transmitted in pregnant mice, and compared to sheep, it triggers lower offspring mortality and morbidity. Thus, at least for this isolate, the disease caused in pregnant mice and offspring is not a reliable predictor/indicator for disease caused in pregnant sheep at mid-gestation. Whether this conclusion is also valid for other type II T. gondii strains needs to be addressed in future studies. In addition, our results suggest that the laboratory isolate TgME49 exhibits an enhanced virulence due to successive passages in cell culture and mice. Thus, virulence traits may have been modified, and it might be advisable to use low-passage isolates in experimental studies, as these probably provide a more realistic picture of the true nature of the parasite biology in the field.

#### AUTHOR CONTRIBUTIONS

IF, JR-C, VP, AH, LO-M, and JB conceived the study and participated in its design. LO-M coordinated the isolation, in vitro and mouse studies, and JB and LO-M coordinated the studies in sheep. RS-S, LO-M, and JB wrote the manuscript, with result interpretation and discussion inputs from IF, JR-C, and AH. JR-C, JM-G, and JB carried out the isolation of TgShSp1. RS-S and JR-C carried out in vitro experiments. LF selected sheep and executed the reproductive program. RS-S, IF, JR-C, and JM prepared tachyzoites or oocysts and performed the infections. RS-S, DG-E, NA-V, JM-G, AA-M, VP, and JB participated in inoculation and clinical examination of animals, performed necropsies and sampling of the animals and performed histopathological analyses. RS-S performed PCR and qPCR analyses, serological assays, and statistical analysis and interpreted the results. All authors read and approved the final manuscript.

# FUNDING

RS-S is supported by a fellowship from the Spanish Ministry of Education, Culture, and Sports (MECD), as a part of the Program of Training of University Teaching Staff (FPU, grant number FPU13/03438) and a mobility grant for predoctoral short stays in R+D centers (EST16/0719). DG-E is the recipient of a postdoctoral contract from the Junta de Castilla y León, partially funded by the European Social Fund (European Union). NA-V is the recipient of a predoctoral contract from the Ministerio de Economía, Industria y Competitividad (Ref. BES-2016-076513). This work was supported by the Ministry of Economy and Competitiveness (AGL2016-75935-C2-1-R and C2-2-R), the Community of Madrid, Spain (PLATESA, S2013/ABI2906), Junta de Castilla y León (LE080U16), and a grant of the Swiss National Science Foundation to AH (project No. 310030\_165782).

#### ACKNOWLEDGMENTS

We thank Dr. J. C. Boothroyd for the donation of the TgME49 isolate. We gratefully acknowledge Norbert Müller, Pablo Winzer and Nicoleta Anghel and Vreni Balmer from the University of Bern (Switzerland), Melchor Molero Crespo, Iván Panero Frade, Carmen Espiniella García, and María Jesus González Fraile from IGM (León, Spain), Alejandro Jiménez, Cristina Guerrero, Alicia Colos, and Irene Écija from SALUVET group, Müller Ribeiro from University of Pernambuco (Recife, Brazil) and Jose Calasanz Jiménez, Francisco Saura, and Jose Maria González from University of Zaragoza (Zaragoza, Spain) for their excellent technical assistance. The Animal Experimentation Service (SEA) at the University of Zaragoza is acknowledged for providing their facilities to carry out the reproductive program in sheep.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2018.00436/full#supplementary-material

Figure S1 | Box-plots showing IgG serum titers generated in dams (A) and non-pregnant mice (B) infected with TgShSp1 oocysts and the uninfected group. Graphs represent the median percentage, the lower and upper quartiles (boxes) and minimum and maximum values (whiskers). For significant differences in serum titres, <sup>∗</sup>P < 0.05 and ∗∗P < 0.01.

Figure S2 | Dot-plot graphs of T. gondii burdens in brain from dams (A) and non-pregnant mice (B). Each dot represents individual values of parasite burden (number of parasites per microgram of DNA), and medians are represented as horizontal lines.

#### REFERENCES


Figure S3 | Dot-plot graphs of T. gondii burdens in lung from dams (A) and non-pregnant mice (B). Each dot represents individual values of parasite burden (number of parasites per microgram of DNA), and medians are represented as horizontal lines.

Figure S4 | Dot-plots showing number of lesions (A), individual focus area (B), and percentage of damaged area (C) in the brains from stillborn lambs and live lambs. Each dot represents individual, and medians are represented as horizontal lines.

Table S1 | Clinical signs, serological titers, and parasite detection in mice intraperitoneally infected with TgME49 and TgShSp1 tachyzoites.

Table S2 | Individual frequency of parasite DNA detection in infected animals.

Table S3 | Individual serological titers in fetuses/lambs from infected ewes.


gondii isolates obtained from livestock in northeastern Brazil. Memórias do Instituto Oswaldo Cruz 111, 391–398. doi: 10.1590/0074-02760150459.


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

Copyright © 2019 Sánchez-Sánchez, Ferre, Regidor-Cerrillo, Gutiérrez-Expósito, Ferrer, Arteche-Villasol, Moreno-Gonzalo, Müller, Aguado-Martínez, Pérez, Hemphill, Ortega-Mora and Benavides. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.

\*

# The Activin Receptor, Activin-Like Kinase 4, Mediates Toxoplasma Gondii Activation of Hypoxia Inducible Factor-1

#### Agnieszka Lis <sup>1</sup> , Mandi Wiley 2†, Joan Vaughan<sup>3</sup> , Peter C. Gray <sup>3</sup> and Ira J. Blader <sup>1</sup>

<sup>1</sup> Department of Microbiology and Immunology, Jacobs School of Medicine and Biomedical Sciences, University at Buffalo, Buffalo, NY, United States, <sup>2</sup> Department of Microbiology and Immunology, University of Oklahoma Health Sciences Center, Oklahoma City, OK, United States, <sup>3</sup> Salk Institute, La Jolla, CA, United States

Edited by:

Jeroen P. J. Saeij, University of California, Davis, United States

#### Reviewed by:

Carsten Lüder, Universitätsmedizin Göttingen, Germany Eric Denkers, University of New Mexico, United States Melissa Lodoen, University of California, Irvine, United States

#### \*Correspondence:

Ira J. Blader iblader@buffalo.edu

#### †Present Address:

Mandi Wiley, Arthritis and Clinical Immunology Research Program, Oklahoma Medical Research Foundation, Oklahoma City, OK, United States

#### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 14 September 2018 Accepted: 04 February 2019 Published: 05 March 2019

#### Citation:

Lis A, Wiley M, Vaughan J, Gray PC and Blader IJ (2019) The Activin Receptor, Activin-Like Kinase 4, Mediates Toxoplasma Gondii Activation of Hypoxia Inducible Factor-1.

Front. Cell. Infect. Microbiol. 9:36. doi: 10.3389/fcimb.2019.00036 To grow and cause disease, intracellular pathogens modulate host cell processes. Identifying these processes as well as the mechanisms used by the pathogens to manipulate them is important for the development of more effective therapeutics. As an example, the intracellular parasite Toxoplasma gondii induces a wide variety of changes to its host cell, including altered membrane trafficking, cytoskeletal reorganization, and differential gene expression. Although several parasite molecules and their host targets have been identified that mediate- these changes, few are known to be required for parasite replication. One exception is the host cell transcription factor, hypoxia-inducible factor-1 (HIF-1), which is required for parasite replication in an oxygen-dependent manner. Toxoplasma activates HIF-1 by stabilizing the HIF-1α subunit, and this is dependent on the signaling from the Activin-Like Kinase (ALK) receptor superfamily. Here, we demonstrate that specific overexpression of the ALK family member, ALK4, increased HIF-1 activity in Toxoplasma-infected cells, and this increase required ALK4 kinase activity. Moreover, Toxoplasma stimulated ALK4 to dimerize with its co-receptor, ActRII, and also increased ALK4 kinase activity, thereby demonstrating that Toxoplasma activates the ALK4 receptor. ALK4 activation of HIF-1 was independent of canonical SMAD signaling but rather was dependent on the non-canonical Rho GTPase and JNK MAP kinase signaling pathways. Finally, Toxoplasma increased rates of ALK4 ubiquitination and turnover. These data provide the first evidence indicating that ALK4 signaling is a target for a microbial pathogen to manipulate its host cell.

Keywords: toxoplasma and toxoplasmosis, hypoxia, transcripional regulation, parasite - host interactions, tgf-beta signaling

# INTRODUCTION

When a host cell encounters an intracellular pathogen, DNA-binding transcription factors are activated either directly by pathogen-derived factors or indirectly by host-derived autocrine/paracrine-acting chemokines, cytokines, and growth factors. In general terms, the function of these host transcription factors can be categorized into three groups: (i) pro-host–protects the host from infection; (ii) pro-parasite–promotes pathogen growth; and

(iii) bystander–genes that have no apparent impact on the host–pathogen interaction. STAT1 is a prototypical pro-host transcription factor since it regulates the expression of IFNγeffector genes, which is required for resistance to Toxoplasma (Collazo et al., 2002; Lieberman et al., 2004). Although few proparasite transcription factors are known, the pathogen-derived factors that activate them, and the pathogen processes that rely on them are important to identify as they represent novel drug targets.

Hypoxia-inducible factor-1 (HIF-1) is a host cell transcription factor that is activated by a wide array of microbial pathogens (Nizet and Johnson, 2009). In mice infected with extracellular pathogens such as Streptococcus pyogenes and Pseudomonas aeruginosa, HIF-1 is important for host resistance by regulating neutrophil and macrophage phagocytic functions (Cramer et al., 2003; Peyssonnaux et al., 2005). HIF-1 is also activated by diverse classes of intracellular pathogens (Sodhi et al., 2000; Wakisaka et al., 2004; Arrais-Silva et al., 2005; Kempf et al., 2005; Peyssonnaux et al., 2005; Spear et al., 2006; Hartmann et al., 2008; Nakamura et al., 2009; Metheni et al., 2015; Fecher et al., 2016). In some of these infections, HIF-1 plays a key role in host defenses by regulating the expression of immunomodulatory molecules and metabolic enzymes. Less is known about whether HIF-1 can support the growth of intracellular pathogens. One notable exception is work from our laboratory demonstrating that HIF-1 is a key pro-parasite transcription factor in cells infected with the protozoan parasite Toxoplasma gondii (Spear et al., 2006).

HIF-1 is a heterodimer composed of α and β subunits that is activated when O2-dependent degradation of the HIF-1α subunit is prevented due to hypoxic stress. However, Toxoplasma does not activate HIF-1 merely by consuming O<sup>2</sup> and triggering localized hypoxic responses. Rather, the parasite activates HIF-1 by down regulating the prolyl hydroxylase 2 (PHD2) enzyme (Wiley et al., 2010) whose hydroxylation of HIF-1α targets it for proteasomal degradation. Using pharmacological, cellular, and genetic inhibitors, we demonstrated that signaling from the Activin-Like Kinase receptor superfamily (ALK4,5,7) is required for HIF-1 activation in Toxoplasma-infected cells (Wiley et al., 2010; Brown et al., 2014).

The ALK4,5,7 receptors are a family of conserved serine/threonine kinase receptors that bind about 30 different ligands. These ligands include Activin A (ALK4), TGFβ (ALK5), and Nodal (ALK7) (Miyazono et al., 2001). ALK4,5,7 receptors do not bind ligands on their own but do so only after their ligand binds a second serine/threonine kinase type II receptor [TGFBR2 (ALK5) and ActRII (ALK4 and ALK7)], which induces the homodimers of each receptor type to multimerize (Attisano and Wrana, 2002). Receptor multimerization allows the cytoplasmic domain of the type II receptor to phosphorylate ALK4,5,7, which leads to activation of the type I receptor's kinase activity. Substrates of ALK4,5,7 kinases mediate most of the physiological changes induced by the ligand, and the SMAD2/SMAD3 transcriptional regulators are the best characterized effectors and are therefore considered the canonical signaling pathway (Attisano and Wrana, 2002). Other non-canonical pathways are also activated by ALK4,5,7; and these include MAP and PI-3 kinases (Zhang, 2009) although the mechanisms by which they are activated is less clear. Here, we demonstrate that Toxoplasma activates ALK4 to trigger HIF-1 activity. We also demonstrate that HIF-1 activation by ALK4 is independent of SMAD2/3 but rather requires host Rho GTPase and JNK MAP kinase signaling.

# MATERIALS AND METHODS

#### Cells and Parasites

The RH Toxoplasma strain (from ATCC; Manassas, VA) and the GRA24 knockout (from Dr. Mohamed Ali Hakimi (CNRS; Grenoble, France) was passaged in human foreskin fibroblasts (HFFs) and murine embryonic fibroblasts (MEFs) in Dulbecco's Minimal Essential Medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum, glutamine and penicillin (100 U/mL)/streptomycin (10 mg/mL) as described (Wiley et al., 2010). All cells and parasites were routinely tested for Mycoplasma contamination (MycoAlert from Lonza; Basel, Switzerland) and found to be negative. Unless otherwise noted, the highest grade possible of chemicals were purchased from Sigma (St. Louis, MO). In addition, experiments were performed under normoxic conditions unless otherwise noted. Cells were grown under low O<sup>2</sup> conditions using an INVIVO<sup>2</sup> Hypoxia Chamber (Baker Instruments; Sanford, ME).

#### Luciferase Assay

Luciferase assays were performed as previously described (Wiley et al., 2010). Briefly, cells were transfected using Lipofectamine 2,000 (Invitrogen; Carlsbad, CA) in 24-well plate with the indicated plasmids (400 ng total) and grown for 24 h at 37◦C. The cells were then mock or parasite infected at a MOI of 4 and incubated for 18 h. The cells were harvested and luciferase activity measured using the Dual Glo Luciferase Reporter Assay (Promega; Madison, WI). U0126 was dissolved in DMSO. Recombinant Lethal Factor/Protective Antigen (LF; kindly provided by Dr. Jimmy Ballard from the University of Oklahoma Health Sciences Center) was used as previously described (Phelps et al., 2008). The plasmids used for this study are described in **Supplemental Table 1.**

#### Western Blotting

Host cells were mock or parasite infected (MOI of 4) for the indicated times, washed 3 times with ice-cold PBS, and lysed on ice with lysis buffer (50 mM TRIS-HCl pH 7.4, 1% NP-40, 150 mM NaCl, 0.1% SDS, 0.25% Sodium Deoxycholate, 10 mM NaF, 20 mM Na3VO4, 10 mM EDTA, 100 mM beta-glycerophosphate, and 1 X protease inhibitor cocktail (Roche; Indianapolis, IN). Lysates were collected, centrifuged at 16,000 xg to remove cell debris, and protein concentrations determined. Equal amounts of proteins were separated by SDS-PAGE, transferred to nitrocellulose membrane, blocked for 1 h with LI-COR blocking solution (LICOR; Lincoln, NE), incubated overnight at 4 ◦C with primary antibodies (see **Supplemental Table 2**) in 5% bovine serum albumin in TTBS (50 mM Tris pH 7.4, 150 mM NaCl, 0.1% Tween-20), followed by a 2 h incubation with Alexa Fluor 680- or 800-conjugated secondary antibodies (Li-COR). Blots were visualized using the LI-COR Odyssey scanner and quantified using the Imager's software.

#### Immunoprecipitation

Mock- or parasite-infected host cells were lysed on ice with immunoprecipitation buffer (50 mM Tris-HCl buffer pH 7.5, 150 mM NaCl, 0.5% NP-40, 5 mM EDTA, 10% glycerol, 10 mM NaF, 20 mM Na3VO<sup>4</sup> plus 1 X protease inhibitor cocktail). Lysates were clarified and equal amounts of protein were incubated overnight with primary antibodies (see **Supplemental Table 2**) at 4◦C, and then incubated for an additional 2–3 h with protein G agarose (ThermoFisher; Waltham, MA) at 4◦C. Beads were washed 5 times with lysis buffer and eluted with SDS-PAGE sample buffer for 5 min at 95◦C and analyzed by Western blotting.

#### In vitro Kinase Assay

Host cells were incubated overnight with DMEM containing 0.2% heat-inactivated serum and then mock or parasite-infected (MOI = 4) for the indicated times. Cells were then lysed in modified RIPA buffer (25 mM Tris-HCl pH 7.4, 0.1% SDS, 1% NP-40, 150 mM NaCl, 1 mM NaF, 10 mM Na3VO4, 25 mM beta-glycerophosphate, 0.25% sodium deoxycholate, and 1 X protease inhibitor cocktail), and 500 µg of the lysates were incubated overnight with anti-ALK4-purified rabbit antibody (Lebrun and Vale, 1997) at 4◦C. Protein A agarose beads were added for 3 h, collected by centrifugation, washed 3 times with modified RIPA buffer and 2 times with kinase assay buffer (25 mM HEPES pH 7.6, 1 mM DTT, 10 mM MgCl2, 25 mM betaglycerophosphate, 200µM ATP). Kinase reactions were started by adding 0.8 µg of SMAD2–GST (Sigma) and 5 µCi32P-ATP (6,000 Ci/mmole; Perkin Elmer; Waltham, MA) to each tube. After 2 h, the reactions were extensively washed with kinase assay buffer, separated by SDS-PAGE, and the gels were dried and subjected to autoradiography.

#### Immunofluorescence

HFFs plated on glass coverslips were treated with either 50 ng/mL of Activin A or parasite-infected (MOI of 4) for 2 h and then fixed with 2% paraformaldehyde (in phosphate buffered saline (PBS) for 20′ at room temperature and then washed 3 times with ice-cold PBS. Cells were permeabilized with 0.1% Triton X-100 for 30′ , blocked overnight with 3% bovine serum albumin in PBS at 4◦C. Then, they were incubated with primary antibodies (Tables) for 2 h, washed 3 times in PBS and incubated with Alexa 488- or Alexa 594-conjugated secondary antibodies for 1 h. Coverslips were washed and then mounted on slides in Vectashield-containing DAPI (Vector Labs, Burlingame, CA) mounting medium. Coverslips were imaged by fluorescence microscopy and images acquired using a 100 X Plan Apo oil immersion 1.46 527 numerical aperture lens on a motorized Zeiss Axioimager M2 microscope equipped with 528 an Orca ER charge-coupled-device (CCD) camera (Hamamatsu, Bridgewater, NJ).

#### RESULTS

#### ALK4 Potentiates HIF-1 Activation in Toxoplasma-Infected Cells

Previously, we demonstrated that Toxoplasma signals through the ALK4,5,7 receptor superfamily to activate HIF-1 (Wiley et al., 2010). To determine which receptor(s) is involved in HIF-1 activation, we hypothesized that overexpression of a relevant receptor would potentiate HIF-1 activity in Toxoplasmainfected cells. Thus, cells were transfected with plasmids expressing a HIF-1–regulated firefly luciferase reporter (HREluc), a constitutively expressed Renilla luciferase reporter (to normalize firefly luciferase activity), and either ALK4, ALK5, or ALK7 (or an empty vector as a control). The transfected cells were infected with Toxoplasma, and 18 h later the cells were lysed and their luciferase activity was measured. We found that only ALK4 overexpression significantly increased HRE-luc activity in Toxoplasma-infected cells (**Figure 1**). Increases in luciferase activity required ALK4 kinase activity since an ALK4 kinase dead mutant was unable to increase luciferase activity following Toxoplasma infection. We did not, however, note a dominant negative effect of the mutant on HIF-1 activation indicating that either expression levels were not high enough or that Toxoplasma and ALK4 interact trough novel mechanism.

### Toxoplasma Activates ALK4

ALK4 activation is a multistep process in which ALK4 homodimers interact with homodimers of a type II ActRII receptor only after the type II receptors become ligand bound. Type II receptors are constitutively active serine/threonine kinases and ligand-binding allows to bind and phosphorylate ALK4, leading to an increase in the kinase activity of ALK4. To test whether Toxoplasma induces ALK4/ActRII multimerization, cells were mock- or parasite-infected for increasing amounts of time and then immunoprecipitated with anti-ActRII antisera or IgG as a control. The immunoprecipitates were collected and Western blotted to detect ALK4. ALK4/ActRII interactions increased within 30′ post infection and then decreased by 2 hpi (**Figure 2A** and **Supplemental Figure 1**). To confirm Toxoplasma ALK4/ActRII interactions, cells were transfected with either an empty vector or a HA-tagged ALK4 expression construct and were mock- or parasiteinfected 30′ . HA-tagged ALK4 was immunoprecipitated with anti HA antibody and immune complexes blotted to detect ActRII. Consistent with the ActRII immunoprecipitation data, Toxoplasma increased the abundance of ActRII in the ALK4 immunoprecipitates (**Figure 2B**).

ALK4 kinase activity is stimulated after it multimerizes with ActRII. We therefore immunoprecipitated ALK4 from the mock- and parasite-infected cells, and the immunoprecipitates were incubated in kinase assay buffer with γ-<sup>32</sup>P-ATP and recombinant SMAD2, which is an ALK4 substrate (Heldin et al., 1997). The reactions were separated by SDS-PAGE, and radiolabeled SMAD2 was detected by Phosphorimager analysis. We found that infection significantly increased ALK4 kinase activity 2 hpi (**Figure 2C**), suggesting that ALK4 remains active even after ActRII and ALK4 dissociate from one another. These

kinase-deficient ALK4 (A), ALK5 (B), or ALK7 (C) (or empty vector as a control). The cells were mock or parasite infected for 18 h and luciferase activity was measured. Shown are the averages ± standard errors of at least 3 independent experiments. \*<p< 0.05 (unpaired Student's t-test).

Immune complexes were collected and then incubated with recombinant SMAD2 and 32Pγ-ATP. Reactions were separated by SDS-PAGE and then the gels were dried, exposed to a Phosphorimager screen, and analyzed by densitometry. Shown is a representative autoradiograph and the graph represents the average and standard deviation of three independent experiments.\*<p< 0.05 (One Way ANOVA).

data indicate that Toxoplasma activates ALK4 and uses this receptor to activate HIF-1.

### Toxoplasma Activates HIF-1 Independently of Canonical SMAD Signaling

ALK downstream signaling comprises the canonical SMAD2/3 pathway and non-canonical pathways that includes MAP kinases. SMAD2 and SMAD3 are resident cytoplasmic DNAbinding proteins that are phosphorylated by activated ALK4 receptors. Once phosphorylated, SMAD2/3 binds SMAD4 and the complex traffics to the nucleus, where it binds DNA and regulates transcription (Macías-Silva et al., 1996). To assess SMAD2/3 activity in Toxoplasma-infected cells, SMAD2 subcellular localization was compared among mock- and Toxoplasma-infected cells (2 hpi) and activin-treated cells as a control. SMAD2 accumulated in nuclei of activin-treated cells but not in Toxoplasma-infected cells (**Figure 3A**). We did note that SMAD2 levels were apparently slightly lower in nuclei of Toxoplasma-infected cells but were precluded from drawing any conclusions since the SMAD2 antisera cross reacts with tachyzoites.

We next examined SMAD2/3 activity in Toxoplasmainfected cells using a SMAD2/3-dependent luciferase (SMRE-luc) reporter plasmid. Thus, pSMRE-luc-transfected host cells were either mock- or parasite-infected or treated with activin as a positive control. The cells were lysed 18 h later and luciferase activity was measured. We found that while activin increased luciferase activity Toxoplasma infection did not (**Figure 3B**). We next tested whether Toxoplasma infection stimulates SMAD2/3 phosphorylation by Western blotting lysates from mock- and

parasite-infected cells using antibodies that recognize either total or phosphorylated SMAD2. In contrast to activin, which robustly increased SMAD2 phosphorylation, infection with Toxoplasma was unable to stimulate phosphorylation (**Figure 3C**). Together, these data indicate that while Toxoplasma engages and activates ALK4 it does so in a manner that prevents it from activating canonical SMAD2/3 signaling.

# Host JNK MAP Kinase Signaling Is Necessary for Toxoplasma Activation of HIF-1

We next examined MAP kinases (MAPKs) because they; (i) are ALK4,5,7-regulated non-canonical signaling pathways (Zhang, 2009),(ii) regulate HIF-1(Comerford et al., 2002; Semenza, 2002), and (iii) are activated by Toxoplasma (Derynck and Zhang, 2003; Kietzmann et al., 2016). To test whether MAPK signaling was involved in HIF-1 activation in Toxoplasma-infected cells, HREluc–expressing cells were mock-treated or treated with Bacillus anthracis Lethal Factor (LF), which inactivates MAPK signaling by cleaving MAP kinase kinases (Bardwell et al., 2004). We found that luciferase activity was significantly reduced by LF, indicating the involvement of a MAPK signaling module in Toxoplasma activation of HIF-1 (**Figure 4A**).

Mammalian cells express three major MAPK signaling cascades (ERK, p38 and JNK). Because pharmacological inhibitors against p38 and JNK MAPKs block parasite growth (Wei et al., 2002; Dittmar et al., 2016), we first tested the effect of the ERK inhibitor, U0126 (10µM), which does not affect parasite growth (Dittmar et al., 2016), We found that HIF-1 activation was unaffected by treatment with this compound (**Figure 4B**) which is consistent with the finding that this compound has no

FIGURE 4 | JNK Signaling Is Important for HIF-1 Activation in

Toxoplasma-infected Cells. (A) pHRE-luc-transfected cells were mock treated or treated with B. anthracis Lethal Factor (LF) for 6 h and then infected with Toxoplasma. Lysates were collected 18 h later and luciferase activities were measured. Shown are the average infection-induced fold inductions and standard deviations of 3 independent experiments. \* < p < 0.05 (One Way ANOVA). (B) Mock- or Toxoplasma-infected pHRE-luc-transfected cells were treated with either DMSO or the ERK inhibitor U0126 (10µM). Lysates were collected 18 h later and luciferase activities measured. Shown are average infection-induced fold inductions ± standard deviations of 3 independent experiments. (C) pHRE-luc or pEGR-luc transfected cells were mock-infected or infected with wild-type RH or GRA24 knockout parasites. Lysates were collected 18 h later and luciferase activities measured. Shown are the average infection-induced fold inductions ± standard deviations of 3 independent experiments. NS p > 0.05 (Student's t-test). \*p < 0.05 (Student's t-test). (D) pHRE-luc-transfected cells were co-transfected with either empty vector or JNK1 dominant negative (DN) and then mock or parasite infected. Lysates were collected 18 h later, and luciferase activities was measured. Shown are average infection-induced fold inductions ± standard deviations of 3 independent experiments.\*p< 0.05 (Student's t-test). (E) Host cells treated with either DMSO or SB505124 (5µM) for 2 h and then mock or parasite infected. Lysates were collected 2 h later and Western blotted to detect total or phosphorylated JNK1/2. Graph shows averages of phosphorylated:total JNK1/2 ratios ± standard deviations from 3 independent experiments. NS p> 0.05 (Student's t-test).\*p< 0.05 (Student's t-test). (F) Host cells were pretreated for 2 h with SB505124 (5µM) and then mock or parasite infected for the indicated times. Lysates were prepared and Western blotted to detect total or phosphorylated p38 MAPK.

of 3 independent experiments. (D) pEGR-luc-transfected cells were co-transfected with empty vector or RhoA dominant negative (DN) mutant and then mock or parasite infected. Lysates were collected 18 h later and luciferase activities measured. Shown are the average fold inductions +/- standard deviations of 3 independent experiments. \*p < 0.05 (Student's t-test).

effect on parasite growth at either 21 or 3% O<sup>2</sup> [(Dittmar et al., 2016) and data not shown]. p38 MAPK activation in Toxoplasmainfected cells is dependent on a dense granule protein named GRA24 (Braun et al., 2013). We therefore compared the ability of wild-type or GRA24 knockout parasites to activate HIF-1 or EGR2, which is a host cell transcription factor whose activation by Toxoplasma is dependent on GRA24/p38 MAPK signaling (Braun et al., 2013; Dittmar et al., 2016). While EGR2 activation was reduced by the loss of GRA24, HIF-1 activation was comparable between the wild-type and GRA24 knockout parasites (**Figure 4C**).

To examine the role of JNK signaling in HIF-1 activation, cells were transfected with the HIF-1 luciferase reporter along with a plasmid encoding a dominant negative JNK mutant. Cells were infected and after 18 h luciferase activity was measured and found to be significantly reduced in the Toxoplasmainfected cells expressing the JNK mutant (**Figure 4D**). JNK is activated in Toxoplasma-infected cells (Valère et al., 2003; Morgado et al., 2011), and we therefore tested whether JNK activation was ALK4 dependent. We previously showed that SB505124, an ALK4,5,7 inhibitor, blocks HIF-1 activation. It also prevents Toxoplasma replication by inhibiting a parasite MAPK homolog, but this effect is only apparent at later times post infection (DaCosta Byfield et al., 2004; Wiley et al., 2010;

Brown et al., 2014). Thus, we pretreated cells with SB505124 or DMSO vehicle (as a control) for 30′ and then mock- or parasite-infected them for 2 h. Cell lysates were Western blotted to detect phospho-specific or total JNK. Toxoplasma-induced JNK1 phosphorylation was reduced by SB505124 (**Figure 4E**). In contrast, SB505124 did not inhibit Toxoplasma induction of p38 MAPK phosphorylation and appeared to potentiate its phosphorylation. These data indicate that that the drug does not unexpectedly inhibit other host cell signaling pathways impacted by Toxoplasma infection (**Figure 4F**).

#### Host Rho GTPase Signaling Is Required for HIF-1 Activation

Two distinct pathways mediate TGFβ activation of JNK. First, the JNK-activating MAPK kinase, TAK1, is activated via recruitment of the TRAF6 ubiquitin ligase to the ALK5/TGFβRII receptor complex (Yamashita et al., 2008). To test whether Toxoplasma increases TRAF6 association with ALK4/ActRII, cells were mock- or parasite-infected for 2 h. Then TRAF6 was immunoprecipitated and the immune complexes were Western blotted to detect ActRII. In contrast to TGFβ (Yamashita et al., 2008), TRAF6 recruitment to the ALK4/ActRII complex was significantly decreased early after infection although it was restored at 18 hpi (**Figure 5A**). Since receptor binding is required for TRAF6 activity, these data therefore indicate that TRAF6 is most likely not involved in ActRII/ALK4-mediated activation of HIF-1 in Toxoplasma-infected cells.

Rho GTPase signaling is a second pathway that mediates TGFβ activation of JNK (Atfi et al., 1997). To test whether this GTPase is important for HIF-1 activation in Toxoplasmainfected cells, HRE-luciferase reporter transfected host cells were co-transfected with either a plasmid that expresses a dominant negative Rho GTPase mutant or an empty vector as a control. The cells were then mock- or parasite-infected for 18 h at which time luciferase activity was measured. The data indicated that HIF-1–dependent luciferase activity was significantly reduced in Toxoplasma-infected cells expressing the dominant negative Rho GTPase mutant (**Figure 5B**). As a positive control, we demonstrated that the dominant negative Rho mutant reduced hypoxic activation of the HRE-luc reporter, a finding consistent with previous reports that Rho signaling is important for HIF-1 activation (Hayashi et al., 2005) (**Figure 5C**). In contrast, expression of the Rho GTPase mutant had no apparent effect on parasite activation of the EGR-luciferase reporter (**Figure 5D**) indicating that expression of the mutant did not have unexpected effects on host cell signaling.

#### Toxoplasma Induces ALK4 Turnover

Once activated, ALK4,5,7 receptors are endocytosed and degraded by both proteasome-dependent and -independent mechanisms (Kavsak et al., 2000; Di Guglielmo et al., 2003). To test whether Toxoplasma induces a similar fate for ALK4, we first assessed its steady state levels after 0.5, 2, and 18 h post infection and found that ALK4 levels were significantly reduced at 18 hpi (**Figure 6A**). Since infection does not reduce ALK4 mRNA levels [(Blader et al., 2001; Kim et al., 2007; Saeij et al., 2007) and data not shown], we tested whether Toxoplasma decreased ALK4 halflife. Thus, cells were pretreated for 30′ with the protein synthesis inhibitor, cycloheximide (CHX), and then mock- or parasiteinfected for 10′ or 2 h. Lysates were prepared and Western blotted to detect ALK4, ActRII, or histone H3 as a loading control. In mock-infected cells, CHX reduced ALK4 levels as early as 10 min and ALK4 levels were significantly reduced by 2 h. While infection did not have an apparent affect on ALK4 turnover after 10′ , it did so at 2 hpi (**Figure 6B**). In contrast, turnover rates of either ActRII isoforms were unaffected by infection.

Next, we tested whether ALK4 turnover was proteasomedependent by examining ALK4 levels in CHX-pretreated cells that were mock- or parasite-infected in the absence or presence the proteasome inhibitor, MG132. We found that MG132 led to increased ALK4 levels in the Toxoplasma- but not mockinfected cells (**Figure 6C**). Finally, ALK4 was immunoprecipated from mock- or parasite-infected cells (2 hpi) that were treated with MG132 and then ubiquitin was detected in the immune complexes by Western blotting. The data indicated that infection led to increased ubiquitin staining in the MG132 treated Toxoplasma-infected cells (**Figure 6D**). Taken together, these data indicate that infection increases ubiquitination and proteasomal degradation of ALK4.

#### DISCUSSION

Intracellular pathogens create their replicative niches by secreting factors that interact with host cell proteins. Many of these host cell proteins are intracellular and are modulated by pathogenderived factors that are exposed to the host cytoplasm. Others are cell surface receptors that are engaged either by the pathogen before they infect a host cell or by a host-derived factor whose release is stimulated by exposure to the pathogen. Here, we extend on our previous work (Wiley et al., 2010) and report that ALK4 is a host cell surface receptor kinase that is activated by Toxoplasma and is important for activating HIF-1. Our work, however, does not resolve whether ALK4 is necessary and sufficient for this process since ALK4 deficient cells are unavailable. In addition, ALK4 siRNAs, which had no dramatic effect on Toxoplasma activation of HIF-1 (not shown) only reduced target protein abundance by only ∼70%. Thus, we cannot exclude possibilities that other receptors can compensate for decreased ALK4 activation or that limited ALK4 expression remains sufficient to mediate HIF-1 activation.

In Toxoplasma-infected cells, ALK4 only appears to regulate JNK without triggering other known downstream signaling pathways such as the SMADs. The mechanism(s) underlying this unique signaling specificity remains unknown. One possibility is that besides regulating receptor abundance, ubiquitin dependent degradation of ALK4,5,7 creates a signaling platform that promotes specific activation of JNK while blocking canonical SMAD2/3 signaling (Zuo and Chen, 2009). In addition, other host cell signaling pathways may be more directly and robustly regulated by Toxoplasma effector proteins that are introduced into the host cell during infection (e.g., GRA24/p38 MAPK, Braun et al., 2013) and therefore sequestered from interacting with ALK4.

Consistent with earlier work that Toxoplasma activates host Rho GTPases (Na et al., 2013), we found that Rho GTPase signaling was important for activating HIF-1 although it remains unclear how it integrates with JNK signaling. One possibility is that a Rho-GTP effector protein may directly signal to activate JNK or an upstream MAPK kinase. Alternatively, Rho may regulate JNK indirectly by affecting ALK4 endocytosis/intracellular trafficking. This may occur by virtue of Rho regulating actin stress fiber formation, which is necessary to coordinate caveolin-associated membrane domains (Stahlhut and van Deurs, 2000; Prieto-Sánchez et al., 2006). However, Toxoplasma does not appear to induce host cell stress fiber assembly (Morisaki et al., 1995; Coppens et al., 2006), and the rates of endocytosis/pinocytosis are not increased by infection (Jones and Len, 1976; Sweeney et al., 2010). Thus, we hypothesize that Toxoplasma utilizes Rho signaling to activate JNK through a more direct approach.

Toxoplasma activates HIF-1 via a parasite-derived secreted factor (Spear et al., 2006) that either directly engages ALK4 or stimulates the expression and release of a host factor that does so. In this manner, Toxoplasma can activate ALK4 in either the cells that it is infecting or neighboring uninfected cells. This suggests that besides the infected cell Toxoplasma may be modulating its microenvironment to establish favorable growth conditions. Besides HIF-1, the parasite also impacts its microenvironment by modulating cell cycle progression of neighboring uninfected cells (Lavine and Arrizabalaga, 2009). In addition, Toxoplasmaderived micronemal proteins that are shed from the parasite during invasion and therefore could interact with neighboring uninfected cells (Muniz-Feliciano et al., 2013) activate host Epithelial Growth Factor Receptor. Thus, we hypothesize that activation of HIF-1 and these other pathways may not only impact the infected cell but neighboring cells that are awaiting infection as well as other cells (e.g., monocytes and T-cells) recruited to the site of the infection. This may be potentially important as ALK4 signaling has been demonstrated to regulate regulatory T-cell development (Huber et al., 2009; Semitekolou et al., 2009; Tousa et al., 2017) as well as macrophage activation, polarization, and function (Ogawa et al., 2006; Zhou et al., 2009; Sierra-Filardi et al., 2011). Our future work will explore the global impact that the ALK4/HIF-1 signaling may have in the fate of Toxoplasma infections.

# AUTHOR CONTRIBUTIONS

AL designed and performed experiments, interpreted results, and wrote the manuscript. MW designed and performed experiments, interpreted results, and edited the manuscript. JV and PG generated reagents, interpreted results, and edited the manuscript. IB designed experiments, interpreted results, and wrote the manuscript.

#### ACKNOWLEDGMENTS

We thank members of the Blader laboratory for helpful discussions as well as Drs. Jimmy Ballard for providing LF and Eric Howard for the JNK expression constructs. This work was supported by NIH Grant R01-AI069986.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00036/full#supplementary-material

Supplemental Figure 1 | Toxoplasma Induces ALK4/ActRII Dimerization and Activation. ActRII was immunoprecipitated from lysates prepared from mock- or parasite-infected cells (30′ post-infection) using anti ActRII or IgG as a control). Immune complexes were Western blotted to detect ALK4. ALK4 levels were assessed in whole cell lysates (WCL).

Supplemental Table 1 | List of plasmids used in the study.

Supplemental Table 2 | List of antibodies used in the study.

# REFERENCES


regulatory T cells that restrain asthmatic responses. Proc. Natl. Acad. Sci. U.S.A. 114, E2891–E2900. doi: 10.1073/pnas.1616942114


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

Copyright © 2019 Lis, Wiley, Vaughan, Gray and Blader. 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.

# From Entry to Early Dissemination—Toxoplasma gondii's Initial Encounter With Its Host

Estefania Delgado Betancourt 1†, Benjamin Hamid2†, Benedikt T. Fabian<sup>1</sup> , Christian Klotz <sup>1</sup> , Susanne Hartmann<sup>2</sup> and Frank Seeber <sup>1</sup> \*

<sup>1</sup> FG 16: Mycotic and Parasitic Agents and Mycobacteria, Robert Koch-Institute, Berlin, Germany, <sup>2</sup> Department of Veterinary Medicine, Institute of Immunology, Freie Universität Berlin, Berlin, Germany

Toxoplasma gondii is a zoonotic intracellular parasite, able to infect any warm-blooded animal via ingestion of infective stages, either contained in tissue cysts or oocysts released into the environment. While immune responses during infection are well-studied, there is still limited knowledge about the very early infection events in the gut tissue after infection via the oral route. Here we briefly discuss differences in host-specific responses following infection with oocyst-derived sporozoites vs. tissue cyst-derived bradyzoites. A focus is given to innate intestinal defense mechanisms and early immune cell events that precede T. gondii's dissemination in the host. We propose stem cell-derived intestinal organoids as a model to study early events of natural host-pathogen interaction. These offer several advantages such as live cell imaging and transcriptomic profiling of the earliest invasion processes. We additionally highlight the necessity of an appropriate large animal model reflecting human infection more closely than conventional infection models, to study the roles of dendritic cells and macrophages during early infection.

Keywords: intestinal organoids, Apicomplexa, intestinal epithelial barrier, innate response, Toxoplasma gondii, Paneth cells

#### INTRODUCTION

Infection by the intracellular apicomplexan parasite Toxoplasma gondii affects an estimated 25–30% of humans worldwide (Montoya and Liesenfeld, 2004), making this zoonotic parasite one of the most widespread human pathogens in the world. Infected felids excrete up to several hundred million environmentally resistant oocysts with their feces, which can infect any warm-blooded animal upon ingestion. There, T. gondii reproduces asexually via two distinct life cycle stages, the fast growing tachyzoite and the slower reproducing bradyzoite stage. The latter forms cysts in various host tissues, which may be consumed by carnivores or omnivores. Following ingestion, bradyzoites are released from cysts, reverting to the tachyzoite stage, replicating, and invading surrounding tissues before eventually disseminating throughout the body to other organs (Blader et al., 2015).

#### Different Outcomes Are Observed Following Experimental Infection With Different Parasite Stages and in Different Host Species

While T. gondii can be transmitted via any of the above-mentioned paths, it is known that infections with different forms of the parasite have different effects in different hosts. Sporozoites differ

#### Edited by:

Jeroen P. J. Saeij, University of California, Davis, United States

#### Reviewed by:

Antonio Barragan, Stockholm University, Sweden Melissa Lodoen, University of California, Irvine, United States

\*Correspondence:

Frank Seeber seeberf@rki.de

†These authors have contributed equally to this work

#### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 15 December 2018 Accepted: 13 February 2019 Published: 05 March 2019

#### Citation:

Delgado Betancourt E, Hamid B, Fabian BT, Klotz C, Hartmann S and Seeber F (2019) From Entry to Early Dissemination—Toxoplasma gondii's Initial Encounter With Its Host. Front. Cell. Infect. Microbiol. 9:46. doi: 10.3389/fcimb.2019.00046 biochemically and cell biologically from tachyzoites and bradyzoites (Speer et al., 1995; Dubey et al., 1998; Jerome et al., 1998; Fritz et al., 2012). Understanding innate immune mechanisms will therefore require comparisons of infections with oocyst-derived sporozoites and bradyzoites, as well as consideration of naturally occurring species-specific transmission pathways.

The natural predator-prey interaction of cats and rodents serves as a convincing argument for studying rats and mice as natural hosts for T. gondii. Experimental data (Dubey, 2001, 2006) support the hypothesis that T. gondii has primarily evolved for transmission by carnivory in cats and via the fecal-oral route in herbivores (Dubey, 2006). Mus musculus is almost entirely an herbivorous organism, with occasional insectivorism (Butet and Delettre, 2011). Maternal cannibalism, as seen under lab conditions, is rather a stress-related behavior (Weber and Olsson, 2008; Weber et al., 2013) and is presumably much less often observed in nature. Consequently, the common use of oral T. gondii infection with bradyzoites in mice as a research model is problematic, particularly if we wish to gain insights relevant to human infection (Ehret et al., 2017; Sher et al., 2017). Notably, islands free of felids exhibited a low seroprevalence of T. gondii in wild pigs and humans, likely resulting from a lack of oocysts in the environment (Dubey et al., 1997a; de Wit et al., 2019). This underlines the important role that oocysts play in parasite dissemination, even for omnivorous species such as pigs or humans.

## The Early Events of Intestinal Entry of T. gondii

Very little is known in any organism about the very early phase of infection with either bradyzoites or sporozoites regarding mechanisms employed by the parasite to pass through the intestinal epithelial barrier (IEB). Early in vivo studies reported that excysted sporozoites were observed in enterocytes 30 min post-infection, with few cytopathological lesions such as villi enlargement detected at the ultrastructural level (Dubey et al., 1997b; Speer and Dubey, 1998). Sporozoites could pass through enterocytes and goblet cells of the ileal epithelium 2 h post-infection and enter the lamina propria where parasites differentiated. However, recent reports have concluded that parasites are only reliably detectable by in vivo imaging 3–5 days post-infection (Coombes et al., 2013; Gregg et al., 2013).

There is therefore a clear need for cellular systems which mimic the in vivo situation and allow live cell imaging and transcriptomic profiling of the earliest invasion processes. Great advances in generation, cultivation and cell-type characterization of intestinal organoids (IOs) offer unique opportunities to observe these early events in different hosts with different T. gondii stages (Klotz et al., 2012; Derricott et al., 2019).

# Intestinal Organoid Models to Study T. gondii Infections

IOs can serve as an unlimited source for primary intestinal epithelial tissues. They reflect the cellular content and functionality of the in vivo organ (**Figure 1A**) including the unique properties of the IEB, like composition of tight junctions (Chiba et al., 2008; Kozuka et al., 2017). By manipulating culture conditions IOs can display the different cell populations that vary throughout both small and large intestine. A major advantage of IOs is their long-term survival in vitro (in contrast to ex vivo organ cultures) and the savings on animal experiments once IOs have been established. They are also genetically tractable (Schwank et al., 2013). However, IOs also have drawbacks, like issues with reproducibility and comparability of results between labs due to non-standardized culture conditions and/or different donors for organoid preparations (Bartfeld, 2016; Yu et al., 2017). Moreover, although IOs provide a good representation of the complexity of the intestinal tissue, terminally differentiated IOs with all cell populations, in particular the less frequent Tuft- and M-cells, are difficult to obtain. Other aspects that are missing in IOs compared to whole animals or ex vivo short-term organ cultures are immune cells and the microbiota. However, recent technological advances allow these "missing" components to be gradually incorporated into the system to consecutively reproduce the in vivo situation (Bartfeld, 2016; Hill et al., 2017; Noel et al., 2017; Williamson et al., 2018). Another hurdle, in particular for infection studies, is the inverted topology of the IOs, i.e., the apical side of the epithelial cells faces the lumen of the organoid (**Figures 1C,D**). This requires the pathogens to be introduced via microinjection (Bartfeld and Clevers, 2015; Hill et al., 2017; Heo et al., 2018; Williamson et al., 2018). However, in the case of T. gondii infection of IOs is efficiently occurring (**Figures 1B–D**) when its lumen becomes accessible by physically breaking it open via simple pipetting. From this short discussion it is evident that for studying T. gondii biology, depending on the research question, three-dimensional IOs are superior to 2D- or even 3D-cultures of cell lines (Barrila et al., 2018; Danielson et al., 2018) but that they cannot yet fully replace mouse experiments.

Intestinal epithelial cells (IECs) are constantly renewed every 3–4 days (Clevers and Bevins, 2013). The main absorptive cell type, enterocytes, are characterized by a columnar architecture and microvilli at the apical surface. Goblet cells secrete mucins, which form a thick mucus layer along the epithelial surface, limiting access and promoting removal of potential invading pathogens (Johansson and Hansson, 2016). Enteroendocrine cells release hormones at the basolateral site, mediating paracrine effects to neighboring cells (Allaire et al., 2018). Tuft cells are thought to serve as luminal chemosensors, but their main function is largely unknown (Gerbe and Jay, 2016). Paneth cells are located in the crypt base and play a pivotal role in innate immune defense in the intestine by secreting antimicrobial molecules, such as defensins/cryptidins, lysozyme and phospholipases (Cheng and Leblond, 1974; Clevers and Bevins, 2013). The release of these granules is highly dependent on several stimuli such as cholinergic agonists (Satoh et al., 1995; Clevers and Bevins, 2013). Paneth cell-specific autophagy has been shown recently as essential for protection against interferon-γ (IFN-γ) dependent intestinal inflammation and crypt integrity, also in the context of a T. gondii infection (Raetz et al., 2013; Burger et al., 2018). Another important role of Paneth cells is the maintenance of the stem cell niche in the small intestine (Sato et al., 2011; Clevers and Bevins, 2013).

reach the lamina propria. The principle strategies used by the parasite to cross this barrier are either via invasion of the intestinal epithelial cells (a) or by transepithelial migration (b). Left: Infected murine DC. T. gondii ROP5 and ROP18 phosphorylate host IRGs, thus protecting the PV from degradation. Extracellular profilin is recognized by the cell via TLRs 11 and 12, inducing the release of IL-12 and TNF-α. Right: Extracellular matrix-embedded crypt (generated via physical disruption of the intestine) leads under appropriate culture conditions to the initial formation of a crypt-derived organoid which then can differentiate into a more complex "mature" intestinal organoid. (B–D) Microscopic images of a mouse small intestine-derived organoid infected with T. gondii RH strain tachyzoites. (B) Representative bright-field confocal image of a mouse IO after 7 days in culture. Scale bar 50µm. (C) Projection of a confocal z-stack of the IO shown in C, stained with FITC-phalloidin for apical F-actin (green), TRITC-labeled UEA-1 lectin for Paneth cell granules (red) and DAPI for nuclei (blue). Note that due to the projection of the stack fluorescent signals might appear mis-localized within the organoid compared to the single plane shown in (D). Scale bar 50µm. (D) Enlarged view of an organoid villus-like structure (white square in C) of a single plane. Parasites (identifiable by their GFP-tagged green tubular mitochondrion (Thomsen-Zieger et al., 2003; black arrow) had replicated in IECs for 48 h. Paneth cells, identifiable by their multiple granules (white arrow) can be detected in the villus-like structure. Due to its granular appearance the red arrow indicates a possible Paneth cell containing replicating parasites. The IO's lumen is filled with cell debris from apoptotic cells, constantly shed as part of the high turnover rate of IECs. The red structures in the lumen marked with white arrow heads might indicate Paneth cell degranulation, as previously described by Farin et al. (2014). Scale bar 20µm.

IOs allow the real-time study of early infection event dynamics in specific gut epithelial cell types upon T. gondii infection. This includes the ability to study IOs derived from a range of host species, including rodents, pigs and humans (Klotz et al., 2012; Derricott et al., 2019). They can therefore serve to highlight differences in these processes in different species under comparable experimental conditions, due to the absence of the immune system and microbiota. In the past, infection studies were performed using either the murine in vivo model, or small intestinal cell lines derived from immortalized or cancer cells. It will therefore be crucial to compare results to T. gondii-infected IOs.

#### Innate Defense Mechanisms of Intestinal Epithelial Cells to T. gondii Infection

Studies in mice have provided a comprehensive picture of innate immune responses in the lamina propria and beyond (Yarovinsky, 2014; Cohen and Denkers, 2015; Dunay and Diefenbach, 2018) upon T. gondii infection, as well as the role of the microbiota (Cohen and Denkers, 2014; Leung et al., 2018). However, much less is known about how parasites are able to overcome the IEB or their fate in the individual cells of it (Jones et al., 2017).

The intestinal mucosa is protected by physical barriers that include the mucus layer, the glycocalyx of enterocytes and the tight junctions between intestinal epithelial cells (Pelaseyed et al., 2014; Okumura and Takeda, 2017). It shields the mucosa from invasion by the microbiota. Although T. gondii is apparently able to penetrate this barrier the efficiency of this process and how the hurdles are overcome is unknown.

Few studies have addressed the role of glycocalyx and mucus during T. gondii infection. However, one described an increase in mucus-producing goblet cells in rats upon infection with oocysts (Trevizan et al., 2016). Conflicting results were reported for the role that trefoil factor family (TFF) peptides, major constituents of intestinal mucus, play in a T. gondii infection. While one study in mice showed a protective effect of TFF2 against immunopathology (McBerry et al., 2012), another reported the opposite effect for TFF3 (Fu et al., 2015).

Different pathways have been proposed to be used by T. gondii for transmigration into the intestinal epithelial tissue (Jones et al., 2017) (**Figure 1A**). The first one is paracellular transmigration, in which the parasites, aided by their gliding motility, move through the intercellular junctions without altering the barrier integrity. In vitro studies using intestinal polarized monolayers showed that tachyzoites of type I strains exhibited a higher migratory capacity compared to type II and type III strains. This might contribute to the higher virulence of type I strains seen in lab mice (Barragan and Sibley, 2002). It was later shown that parasites rapidly cluster between the cellular junctions upon entry (Weight and Carding, 2012; Briceño et al., 2016; Jones et al., 2017), and the tight junction protein occludin was identified as a specific target of tachyzoites during passage through the paracellular pathway (Weight et al., 2015). T. gondii might use it to efficiently cross the monolayers by the interaction of intracellular adhesion molecule-1 with the parasite adhesion molecule MIC2, without affecting barrier permeability (Barragan et al., 2005). However, contradictory results were reported in an experimental set-up with Caco-2 cells (Briceño et al., 2016) where it was shown that intestinal barrier function was disturbed. These examples highlight the need to evaluate these crucial events in a cellular system like IOs that closely resemble the in vivo IEB.

The second reported entry pathway is by penetration of the apical cell membrane and passing through the basolateral side in order to reach the underlying lamina propria where leukocytes reside (Barragan et al., 2005; Lambert and Barragan, 2010). Several authors have proposed a third pathway, which involves a Trojan-horse-like model (Gregg et al., 2013; Jones et al., 2017). Upon infection of IECs neutrophils are rapidly recruited to the site of infection and subsequently infected by the parasite. These are then capable of migrating through the epithelial cell layer and crossing the lumen, thereby facilitating parasite spread not only in the intestine but also to other tissues (Coombes et al., 2013). However, most of these data were generated in the above-mentioned traditional model systems. Therefore, it will be interesting to see how IOs compare to these models and which reflect best the in vivo situation.

Besides the physical barrier discussed above an independent biochemical barrier exists, composed mainly of antimicrobial peptides and proteins (e.g., cryptidins, defensins, lysozyme). The vulnerability of T. gondii bradyzoites or sporozoites toward these molecules is largely unknown. A differential effect of oocysts of different genetic backgrounds on Paneth cell-derived lysozyme expression following infection of BALB/c mice was reported (Lu et al., 2018). Likewise, only a single study provided some indirect evidence for an effect of the antimicrobial activity of cryptidins on bradyzoites in the lumen of mice prior to invasion of the epithelial layer (Foureau et al., 2010). Evidently, there is still more to learn in this area.

Upon parasite exposure IECs (and immune cells, see below) recognize the parasite through pattern recognition receptors on the cell surface, such as Toll-like receptors (TLRs), which activate secretion of pro-inflammatory cytokines that induce a subsequent Th1 response (Gopal et al., 2008). Among these, Paneth cell-resident TLR9 has been shown to modulate recognition of external pathogens and to induce the immune response through mechanisms such as defensin release (Rumio et al., 2004; Buzoni-Gatel et al., 2006; Foureau et al., 2010). Surprisingly, a recent transcriptomic study with rat intestinal IEC-18 cells did not find evidence of pathogen-associated molecular patterns being induced upon infection with oocystderived T. gondii sporozoites (Guiton et al., 2017).

#### Early Interaction of T. gondii With Host Immune Cells

The lamina propria and Peyer's patches are rich in dendritic cells (DCs) and macrophages (M8s). Once the intestinal barrier is overcome by T. gondii, these are the first immune cells to recognize parasite infection and to initiate the mounting of the host immune response. This recognition can occur via at least three distinct routes. (1) DCs and M8s directly phagocytose free, opsonized parasites upon crossing the epithelial barrier. (2) Both cell types also phagocytose infected apoptotic IECs (Buzoni-Gatel and Werts, 2006). (3) DCs are able to elongate through the tight junctions of the epithelium, and in mice recognize the soluble T. gondii antigen profilin in the lumen via TLR 11 and 12 (Yarovinsky et al., 2005; Koblansky et al., 2013). DCs and M8s exhibit directly toxoplasmacidal effects to phagocytosed parasites. However, once stimulated they also begin to secrete interleukin (IL) 12 and tumor necrosis factor α (TNFα) (Buzoni-Gatel and Werts, 2006). IL-12 and TNF-α induce the differentiation of CD4<sup>+</sup> T cells into Th1 cells, which secrete IFNγ. In parallel, IL-12, along with IL-15 secreted by infected IECs, stimulate natural killer (NK) cells and CD8<sup>+</sup> T cells to begin secreting IFN-γ, the primary mediator of resistance to T. gondii (Suzuki et al., 1988). This leads to containment of the parasite and its conversion into the bradyzoite form, thereby hiding from the immune system (Hunter and Sibley, 2012; Ahmed et al., 2017).

### Role of T. gondii Rhoptry Proteins in Early Immune Cell Modification

One mechanism by which IFN-γ mediates parasite destruction in mice is through upregulation of immunity-related GTPases (IRGs) (Gazzinelli et al., 2014; Müller and Howard, 2016). IRGs are intracellular host proteins, some of which localize to the PV membrane in an infected cell causing membrane rupture, parasite release into the host cell cytosol and its subsequent degradation. In order to avoid destruction, T. gondii has evolved a means to subvert this host defense mechanism (Buzoni-Gatel and Werts, 2006; Gazzinelli et al., 2014). It is dependent on several parasite proteins which are derived from unique secretory organelles (rhoptries and dense granules) and transported into the infected cell. ROP18 is able to phosphorylate host IRGs such as Irga6 while ROP5 modulates this activity, all eventually resulting in PV membrane destruction (Reese et al., 2011; Behnke et al., 2012; Fleckenstein et al., 2012; Niedelman et al., 2012; Etheridge et al., 2014). However, depending on the genetic background of the mouse, this virulence mechanism of the parasite can be overcome by a highly polymorphic IRG protein (Irgb2-b1). Some of its variants can act as decoys for ROP5/18 binding, enabling other IRGs to degrade the PV (Lilue et al., 2013). Surprisingly, it is unknown if or to what extent this IRG response plays a role in the intestine. Mouse IOs could be very useful to shed light on this immediate obstacle the parasite has to overcome in order to proliferate and disseminate.

#### Phenotypic Changes of DCs Utilized as "Trojan Horse" Vehicles for Dissemination of T. gondii

After infection T. gondii is able to rapidly disseminate throughout the body. Within hours it is found in the spleen and it is also able to cross the blood-brain barrier, the placental barrier in pregnant hosts, and enter immune privileged sites such as the eyes (Lambert and Barragan, 2010; Harker et al., 2015). This is achieved through invasion and utilization of migratory leukocytes as "Trojan horses." There is evidence suggesting that several cell types may be used for this purpose, including DCs, M8s, neutrophils, NK cells, and T cells (Courret et al., 2006). However, extracellular tachyzoites released from infected endothelial cells in the brain vasculature have also recently been implicated in overcoming the blood-brain barrier (Konradt et al., 2016).

Mostly DCs, inherently able to become migratory, have been implicated in early dissemination (Weidner and Barragan, 2014; Kanatani et al., 2015; Brasil et al., 2017; Ólafsson et al., 2018). Following activation by an antigen they undergo a series of phenotypic changes required for its efficient presentation. This includes comprehensive remodeling of the actin cytoskeleton and the loss of actin-rich structures called podosomes. These changes are essential for switching from a strongly adhesive to a migratory phenotype, allowing cells to reach the lymph nodes. Using lipopolysaccharide to activate DCs indicated that remodeling was dependent on TLR4 signaling and prostaglandin E2 (PGE2) secretion (van Helden et al., 2010; Weidner et al., 2013). In contrast, following infection with T. gondii tachyzoites, phenotypic changes occurred <10 min post-invasion and were not reliant on TLR4 or PGE2. This was shown experimentally to require active manipulation by live T. gondii (Weidner et al., 2013). Recent studies indicated that T. gondii infection results in a marked reduction in pericellular proteolytic activity by DCs, mediated via the release of tissue inhibitor of metalloproteinase 1. This suggests a compensatory mechanism for an upregulation of matrix metalloproteinases, which have been demonstrated to perform diverse catalytic and non-catalytic functions in amoeboid migration (Orgaz et al., 2014; Ólafsson et al., 2018).

### Differences in the Macrophage/DC Responses of Mice and Humans to T. gondii: the Pig as Human-Relevant Model

As discussed, murine DCs are able to undergo maturation in response to the detection of the soluble T. gondii antigen profilin via TLR11 and TLR12. However, in humans TLR12 is entirely absent, and TLR11 is apparently a non-functional pseudogene (Zhang et al., 2004; Roach et al., 2005; Ishii et al., 2008). Consequently, profilin does not elicit an immune response in humans; instead it relies on phagocytosis of tachyzoites (Tosh

TABLE 1 | Markers of monocyte and DC subsets in mice, humans, and pigs (Fairbairn et al., 2013; Summerfield et al., 2015; Sher et al., 2017).


Murine and human subsets highlighted in yellow produce IL-12 in response to stimulation with T. gondii tachyzoites in vitro (Sher et al., 2017). The porcine subsets which respond to T. gondii exposure are currently unknown.

et al., 2016; Sher et al., 2017). Although the pattern recognition receptors responsible for the recognition of T. gondii in humans have not been definitively identified, human PBMCs produce pro-inflammatory cytokines following stimulation with T. gondii RNA or DNA. This implicates the involvement of TLRs 7, 8, and 9 which are responsible for the recognition of nucleic acids from pathogens (Forsbach et al., 2008; Andrade et al., 2013; Jennes et al., 2017). The specific subsets of monocytes and DCs secreting IL-12 in response to T. gondii also differ between mice and humans. In mice, inflammatory monocytes and CD8α <sup>+</sup> DCs respond, whereas the human analogs—classical monocytes and the cDC2 subset—do not. In contrast, human non-classical and intermediate monocytes and the cDC1 subset produce IL-12, which are analogous to murine patrolling monocytes and CD8α − DCs (Tosh et al., 2016; Sher et al., 2017).

There is a clear need for an immunologically more human-like large animal model to understand the mechanisms underlying T. gondii infection and immunity in humans (Ahmed et al., 2017; Sher et al., 2017). The pig is one such candidate that could be utilized for this purpose. Genomic studies have indicated that 80% of porcine immune response genes resemble human equivalents, whereas for mice <10% are similar (Meurens et al., 2012; Mair et al., 2014) (**Table 1**). Of particular note is that like humans, pigs lack TLRs 11 and 12 (Uenishi and Shinkai, 2009; Mair et al., 2014) and so are presumably also unable to respond to profilin. They do however exhibit TLRs 7, 8, and 9, and so are likely able to recognize T. gondii via the same mechanism as humans (Uenishi et al., 2012; Jennes et al., 2017). Thus, the initial porcine DC and M8 responses to T. gondii deserve further examination regarding their similarity to the human response.

There are also clinical similarities between the human and porcine responses to T. gondii, which further suggest the pig may be an appropriate model for human infection. For example, postnatal infection with T. gondii is usually asymptomatic or mild in humans and pigs, whereas infections with some parasite strains can be fatal in mice (Dubey, 1986; Nau et al., 2017). During pregnancy in humans and pigs parasites can often cross the placental barrier and result in abortion or congenital toxoplasmosis (Jungersen et al., 2001), whereas fetal infections are rare in immunocompetent mice (Shiono et al., 2007; Nau et al., 2017). Furthermore, as omnivorous mammals pigs, like humans, are naturally at risk of exposure to both T. gondii tissue cysts and oocysts in their diet (Meurens et al., 2012). This makes them a more natural host for research into the early stages of infection with both bradyzoites and sporozoites.

Although fewer immunological reagents are currently available for swine in comparison to mice and humans, this is an area undergoing rapid progress, not the least because of the increased interest in pig organs for xenotransplantation

#### REFERENCES

Ahmed, N., French, T., Rausch, S., Kühl, A., Hemminger, K., Dunay, I. R., et al. (2017). Toxoplasma co-infection prevents Th2 differentiation and leads to a Helminth-specific Th1 response. Front. Cell. Infect. Microbiol. 7:341. doi: 10.3389/fcimb.2017.00341

(Meier et al., 2018). After mice and primates, the porcine immune system is perhaps the next most thoroughly characterized, with pigs being firmly established as a model organism for infection research. This includes their use as a model for infection with other human-relevant, orally-acquired pathogens such as Helicobacter pylori and human rotavirus, as well as the protozoan parasite Cryptosporidium parvum (Meurens et al., 2012). In recent years the body of literature on the porcine cellular immune response specifically to T. gondii also increased (e.g., Miranda et al., 2015; Jennes et al., 2017; Nau et al., 2017). Notably, porcine IOs have also been described recently (Derricott et al., 2019).

# CONCLUDING REMARKS

IOs closely resemble the in vivo intestinal barrier and represent a source of species-specific IECs. To mechanistically study interactions of pathogens with such a complex organ it is advantageous to examine the contribution of individual epithelial cells in the absence of immune cells and microbiota. However, several reports have illustrated that IOs can be co-cultured with DCs, M8s, IELS (Nozaki et al., 2016; Noel et al., 2017; Ihara et al., 2018; Nakamura, 2018) and also with bacteria (Hill et al., 2017; Williamson et al., 2018), thereby complementing this system as required.

The pig allows for tissue-specific translational research since the immune parameters depicted so far closely resemble humans. Future studies will show whether porcine intestinal innate and adaptive parameters better reflect human early infection events in comparison to mice.

# ETHICS STATEMENT

The use of animal material was approved by the responsible local authorities of the German Federal State Berlin (permit T0173/14).

# AUTHOR CONTRIBUTIONS

ED provided parts of **Figure 1A** and **Figures 1B–D**. BH provided parts of **Figure 1A**. All authors contributed to the text and approved its final version.

# FUNDING

The work was supported by the German Research Council: GRK 2046 to ED, BH, CK, SH, and FS. Work by CK and FS cited is supported by the Robert Koch-Institute. BF is supported by the Robert Koch-Institute as part of the German One Health Initiative (GOHI).

Allaire, J. M., Crowley, S. M., Law, H. T., Chang, S. Y., Ko, H. J., and Vallance, B. A. (2018). The intestinal epithelium: central coordinator of mucosal immunity. Trends Immunol. 39, 677–696. doi: 10.1016/j.it.2018.04.002

Andrade, W. A., Souza Mdo, C., Ramos-Martinez, E., Nagpal, K., Dutra, M. S., Melo, M. B., et al. (2013). Combined action of nucleic acid-sensing Toll-like receptors and TLR11/TLR12 heterodimers imparts resistance to Toxoplasma gondii in mice. Cell Host Microbe 13, 42–53. doi: 10.1016/j.chom.2012. 12.003


mode encompassing podosome dissolution, secretion of TIMP-1, and reduced proteolysis of extracellular matrix. Cell. Microbiol. 20:e12808. doi: 10.1111/cmi.12808


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

Copyright © 2019 Delgado Betancourt, Hamid, Fabian, Klotz, Hartmann and Seeber. 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.

# Calling in the CaValry—*Toxoplasma gondii* Hijacks GABAergic Signaling and Voltage-Dependent Calcium Channel Signaling for *Trojan horse*-Mediated Dissemination

#### Amol K. Bhandage and Antonio Barragan\*

Department of Molecular Biosciences, The Wenner-Gren Institute (MBW), Stockholm University, Stockholm, Sweden

#### *Edited by:*

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### *Reviewed by:*

Tajie Harris, University of Virginia, United States Friedrich Frischknecht, Universität Heidelberg, Germany

> *\*Correspondence:* Antonio Barragan antonio.barragan@su.se

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 30 December 2018 *Accepted:* 27 February 2019 *Published:* 20 March 2019

#### *Citation:*

Bhandage AK and Barragan A (2019) Calling in the CaValry—Toxoplasma gondii Hijacks GABAergic Signaling and Voltage-Dependent Calcium Channel Signaling for Trojan horse-Mediated Dissemination. Front. Cell. Infect. Microbiol. 9:61. doi: 10.3389/fcimb.2019.00061 Dendritic cells (DCs) are regarded as the gatekeepers of the immune system but can also mediate systemic dissemination of the obligate intracellular parasite Toxoplasma gondii. Here, we review the current knowledge on how T. gondii hijacks the migratory machinery of DCs and microglia. Shortly after active invasion by the parasite, infected cells synthesize and secrete the neurotransmitter γ-aminobutyric acid (GABA) and activate GABA-A receptors, which sets on a hypermigratory phenotype in parasitized DCs in vitro and in vivo. The signaling molecule calcium plays a central role for this migratory activation as signal transduction following GABAergic activation is mediated via the L-type voltage-dependent calcium channel (L-VDCC) subtype Cav1.3. These studies have revealed that DCs possess a GABA/L-VDCC/Cav1.3 motogenic signaling axis that triggers migratory activation upon T. gondii infection. Moreover, GABAergic migration can cooperate with chemotactic responses. Additionally, the parasite-derived protein Tg14-3-3 has been associated with hypermigration of DCs and microglia. We discuss the interference of T. gondii infection with host cell signaling pathways that regulate migration. Altogether, T. gondii hijacks non-canonical signaling pathways in infected immune cells to modulate their migratory properties, and thereby promote its own dissemination.

#### Keywords: apicomplexa, CNS infection, dendritic cell, microglia, motility, GABA receptor

# INTRODUCTION

The apicomplexan parasite Toxoplasma gondii infects a diverse repertoire of hosts, including humans and rodents (Tenter et al., 2000). As an obligate intracellular pathogen, Toxoplasma is able to infect and replicate within virtually any type of nucleated cell from warm-blooded vertebrates. Despite that up to one-third of humans become chronically infected during their lifetime, most infections are considered asymptomatic. Yet, Toxoplasma can cause life-threatening disease in immunocompromised individuals or to the developing fetus (Montoya and Liesenfeld, 2004). Congenital toxoplasmosis is also a significant problem in veterinary medicine.

After oral ingestion, systemic parasite dissemination during primary infection precedes the establishment of chronic infection. Early studies in rodents showed that, shortly after invasion of the intestinal tissue (Dubey, 1997), parasites (tachyzoite stage) are retrieved in the blood circulation (Derouin and Garin, 1991; Zenner et al., 1998). At this early stage–before the infection is controlled by the immune response– the parasite can be isolated from virtually any organ in rodents, including immunoprivileged organs such as brain, eyes, and testis (Hitziger et al., 2005). Consequently, breaching restrictive biological barriers, such as the intestine, the blood brain-barrier, blood-retina barrier, or the placenta, is a requisite for the establishment of primary Toxoplasma infection and subsequently chronic infection or congenital infection.

The rapidly replicating stage of T. gondii (the tachyzoite) mediates dissemination. Tachyzoites have an obligate intracellular existence and their active locomotion, termed gliding motility, is instrumental for invasion of host cells (Dobrowolski and Sibley, 1996). Gliding motility also powers parasite migration within the tissue microenvironment (Barragan and Sibley, 2002). Additionally, transportation by the blood and the lymphatic circulation assures rapid systemic dissemination to distant organs.

The onset of immune responses against T. gondii is accompanied by the transformation of the parasite into tissue cysts (bradyzoite stage) and is believed to result in longlasting or lifelong chronic infection in humans (Joynson and Wreghitt, 2001). In rodents, cellular immune responses mediated by dendritic cells (DCs), T cells, NK cells, macrophages, and cytokine responses (IL-12 and IFN-γ) are essential to overcome primary infection and for establishing latent chronic infection (Yap and Sher, 1999; Sacks and Sher, 2002). Leukocytes traffic the tissues, populate the blood, drain to the lymphatic system and back into the circulation (Friedl and Weigelin, 2008). Thus, while constitutive epithelial and connective tissues may provide refuge and replicative niches for T. gondii, leukocytes additionally mediate immune surveillance and are essential for pathogen clearance. However, the migratory functions of leukocytes make them also suitable vehicles for Toxoplasma to mediate its dissemination in the organism by a Trojan horse mechanism (Weidner and Barragan, 2014).

In this review, we discuss the current knowledge on how Toxoplasma modulates the migratory properties of immune cells, primarily DCs, and microglia, by inducing a hypermigratory phenotype that promotes parasite dissemination. The mechanisms involved in Toxoplasma-induced hypermigration include a hijacking of the GABAergic signaling system and voltage-dependent calcium channel (VDCC) signaling in parasitized cells.

# GABA, A SIGNALING MOLECULE IN THE CNS AND PERIPHERAL TISSUES

The γ-aminobutyric acid (GABA) is most commonly known as an inhibitory neurotransmitter in the CNS of vertebrates, where it contributes in maintaining the balance between excitatory and inhibitory neurotransmission. GABA executes its action in the CNS through GABA-A and GABA-B receptors. GABA-A receptors are chloride (Cl−) -permeable ion channels formed by pentameric combinations of subunits (2 α + 2 β + 1 additional subunit) out of 19 known subunits to date (α1-6, β1-3 γ1-3, δ, ε, θ, π, ρ1-3) (Olsen and Sieghart, 2008; Sieghart et al., 2012). In contrast, GABA-B receptors are metabotropic G-protein coupled receptors formed as a heterodimer of 2 subunits (Bowery et al., 2002; Bowery, 2010).

The depolarizing (excitatory) or hyperpolarizing (inhibitory) effects of GABA depend on the intracellular Cl<sup>−</sup> concentration [Cl−]<sup>i</sup> which is set by cation chloride co-transporters (CCCs) -NKCCs and KCCs (Blaesse et al., 2009; Kaila et al., 2014). For instance, higher relative expression of NKCC1 than KCC2 in the immature neurons results in higher [Cl−]<sup>i</sup> leading to depolarizing effects of GABA (Ben-Ari et al., 2007; Kilb, 2012). Thus, GABA can mediate depolarizing or hyperpolarizing responses in immature and mature neurons, respectively, indicating a developmental switch between the actions of GABA. Additionally, certain neuronal populations in the adult rat brain exhibit depolarizing responses to GABA (Chiang et al., 2012; Haam et al., 2012; Sauer et al., 2012). Further, experimental down-regulation of KCC2 results in excitatory effects upon GABA-A receptor activation in the adult neurons (Sarkar et al., 2011). Overall, these studies highlight that the actions of GABA are highly dependent on activity of Cl<sup>−</sup> transporters, developmental stage, cell type and sub-cellular localization (Bortone and Polleux, 2009).

Beyond neurotransmission, GABA has also been attributed roles in cellular processes such as migration, proliferation, differentiation, synapse formation, axonal growth, and neuronal death (Birnir and Korpi, 2007; Kilb, 2012). In addition to its predominant presence in the CNS, GABA is also synthesized by non-neuronal tissues like pancreatic islets, glia cells, adrenal medulla, germ cells, testes, and immune cells and these tissues also often express GABA receptors (Gladkevich et al., 2006; Jin et al., 2013). Of note, GABA precedes the existence of the vertebrate and invertebrate nervous systems and is synthesized by prokaryotes for metabolic purposes (Feehily et al., 2013; Xiong et al., 2014), and by plants for signaling (Michaeli and Fromm, 2015). T. gondii possesses a GABA shunt pathway and can utilize GABA as an energy source, for instance, to sustain gliding motility under nutrient-limited conditions (MacRae et al., 2012). Thus, GABA's function as a neurotransmitter is likely an evolutionary adaption posterior to metabolic and other signaling functions. In other words, GABA precedes the development of a CNS in metazoa and, consequently, alternative functions exist.

#### A GABAergic SIGNALING SYSTEM IN IMMUNE CELLS AND PERIPHERAL TISSUES

Research over the past decade has identified the functional implications of GABA and its receptors outside the CNS, i.e., in pancreatic islets, immune system, digestive system, and reproductive system (Gladkevich et al., 2006; Bhandage et al., 2018a; Korol et al., 2018). Mounting evidence indicates that brain-resident immune cells, such as microglia, and circulating immune cells, such as neutrophils, T lymphocytes, monocytes, macrophages and DCs, secrete GABA, and/or express GABA receptors and associated proteins (Rane et al., 2005; Alam et al., 2006; Bhat et al., 2010; Lee et al., 2011; Fuks et al., 2012; Bhandage et al., 2015, 2018a).

Further, GABA impacts on the effector functions of immune cells, i.e., migration, proliferation, and cytokine secretion via GABA receptors. For instance, GABA mediates protective roles by inhibiting auto-reactive T cells in human autoimmune inflammatory diseases such as type 1 diabetes, rheumatoid arthritis and multiple sclerosis (Tian et al., 2004, 2011; Bhat et al., 2010; Bhandage et al., 2018a). Gephyrin-dependent enhanced GABA signaling has been shown to participate in the conversion of alpha cells to insulin producing beta cells in diabetic pancreatic islets resulting in cure of type 1 diabetes (Li et al., 2017). GABA and GABA-A receptor signaling have also been implicated in maturation and proliferation of adult stem cells and GABA has been suggested as a tumor-signaling molecule in cancer cells (Andäng et al., 2008; Young and Bordey, 2009). Moreover, a recent study demonstrates antimicrobial and autophagy-related roles for GABA in macrophages upon mycobacterium infection (Kim et al., 2018). Jointly, this highlights the significance of GABAergic signaling in human diseases.

Thus, mounting evidence shows that the immune system harbors components or the complete machinery for GABAergic signaling and that GABA can serve as a modulator of the effector functions of immune cells (**Table 1**). It evidences a possible crosstalk between the nervous system and the immune system and the presence of novel "neuro-immune-signaling" axes (Sospedra and Martin, 2005; Levite, 2012). However, the understanding of GABAergic signaling and its patho-/ physiological roles in immune cells is still rudimentary or non-existent compared with neuronal systems and thus, crucial to be further studied in depth.

#### VDCC SIGNALING IN IMMUNE CELLS

VDCCs are formed from α1, α2, β, δ, and γ subunits where the transmembrane α1 subunit forms the tetrameric ion channel with a central pore permeable specifically for calcium ions. The 10 members of the VDCC (also called CaV) family can be divided in 3 subfamilies high voltage activated L-type, moderate voltage activated P/Q, N, R-type and low voltage activated T-type channels (Catterall, 2011).

The expression of VDCCs by immune cells has long remained unresolved. However, their expression and implication in immune functions has more recently received attention (**Table 1**). In excitable cells like neurons and pancreatic islets beta cells, the channels are opened by depolarizing voltage changes in the membrane potential but how they are operated in immune cells is still ambiguous (Badou et al., 2013). It is possible that immune cells can sense these depolarizing changes by activation of ligand-gated ion channels such as GABA-A receptors and N-methyl-D-aspartate (NMDA) receptors.

Some of the implications of VDCCs in the immune cell physiology are described here. The CaV1.4 channels, L-type VDCC, have been shown to contribute in controlling naive T cell homeostasis, T cell receptor (TCR) signaling and antigen-driven T cell immune responses (Omilusik et al., 2011). Additionally, CaV1.2 and CaV1.3, other L-type VDCC channels, participate in TCR-induced calcium flux in T cells (Stokes et al., 2004). Not only α subunits but also β subunits of Ca<sup>V</sup> channels are shown to be important for normal T cell functions such TCR-mediated calcium entry, nuclear factor of activated T cells (NFAT) activation, and cytokine production (Badou et al., 2006). The CaV3.1, T-type channel, shapes the cytokine profile of T helper cells and can ameliorate autoimmune responses (Wang et al., 2016). The CaV1.2 channels can activate intracellular calcium receptors and contribute in the surface expression of MHC class II molecules in DCs during antigen presentation to T cells (Vukcevic et al., 2008). While characterization has just begun, these evidences indicate functional roles for VDCCs in immune cell physiology.

#### INFECTION OF LEUKOCYTES BY *T. gondii* AND THEIR ROLE IN PARASITE DISSEMINATION

Following oral infection, Toxoplasma rapidly disseminates in its host. Early studies in rodents detected tachyzoites in the blood, lymph nodes and peripheral organs rapidly after infection (Derouin and Garin, 1991; Dubey, 1997; Zenner et al., 1998). It was demonstrated that not only the intestinal tissue becomes parasitized but also the intra-epithelial leukocytes (Dubey et al., 1997). Following studies showed that both resident and nonresident leukocytes become infected in the intestine (Courret et al., 2006; Gregg et al., 2013).

The rapid dissemination of T. gondii tachyzoites (Hitziger et al., 2005) combined with the parasite's ability to infect and replicate within leukocytes (Channon et al., 2000) raised the hypothesis that the systemic spread of parasites was mediated by a Trojan horse type of mechanism. DCs were identified as important mediators of dissemination and attributed shuttling functions for T. gondii (Courret et al., 2006; Lambert et al., 2006; Bierly et al., 2008). Additionally, all strains tested to date from the three predominant T. gondii lineages (types I, II, III) induce a hypermigratory phenotype in DCs upon challenge with tachyzoites (**Figures 1A–C**) (Lambert et al., 2009). The characteristics and criteria of the hypermigratory phenotype have been previously reviewed (Weidner and Barragan, 2014) and its impact on the dissemination of T. gondii is discussed below.

Similar to DCs, monocytic cells have been attributed a hypermigratory phenotype upon Toxoplasma challenge (Harker et al., 2013; Cook et al., 2018). Additionally, parasitetransportation functions to the brain have been attributed to monocytes (Courret et al., 2006; Lachenmaier et al., 2011). Other leukocytes also become infected in vivo and thus may also contribute to the systemic spread of the infection. These include T cells (Persson et al., 2007; Chtanova et al., 2009), NK cells (Persson et al., 2009; Sultana et al., 2017), neutrophils (Norose et al., 2008; Coombes et al., 2013), and macrophages (Da Gama et al., 2004; Lambert et al., 2011). While NK cells do not seem to facilitate passage of T. gondii across the blood-brain barrier (Petit-Jentreau et al., 2018), the relative contribution of the


<sup>a</sup>peripheral blood mononuclear cells.

different leukocyte types to dissemination at the different phases of the infection remains undetermined.

#### *T. gondii* INFECTION AND GABAergic SIGNALING IN DCs

#### Primary Human and Murine DCs Express a Functional GABAergic System

Murine bone marrow-derived DCs (mBMDCs) express mRNAs for five GABA-A receptor subunits (α3, α5, β1, β3, ρ1), the enzyme responsible for GABA synthesis (glutamate decarboxylase GAD65) as well as a GABA transporter (GAT4) (Fuks et al., 2012). Since the classical GABA-A channel pentamer requires 2 α, 2 β, and 1 additional subunit, the expression pattern in mBMDCs indicated the possibility of subunit assembly into a pentamer that can be trafficked to the cell membrane **(Figure 2A)**. GABA-induced whole-cell inwards currents recorded in mBMDCs and human monocyte-derived DCs (hMDDCs) using patch clamp electrophysiology (Fuks et al., 2012) demonstrated presence of functionally active GABA-A receptors. It may be assumed that the activation of GABA-A receptors results in efflux of Cl<sup>−</sup> ions out of cells, leading to membrane depolarization. Thus, DCs harbor a functional GABAergic system.

Knowledge remains limited on the expression and function of the GABAergic system in other immune cells (Barragan et al., 2015). It will be instrumental to determine the subunit composition of the active subtype of GABA-A receptors and of additional GABAergic signaling components such as CCCs, GADs, GATs, GABA transaminase (GABA-T), and GABA-A receptor anchoring proteins. Further extending this knowledge to human DCs (native myeloid/plasmacytoid DCs) will be crucial to understand the implication of GABAergic signaling in human diseases.

#### *Toxoplasma*-Infected DCs Secrete GABA

Independent of the parasite strain, Toxoplasma infection induced GABA secretion in mBMDCs in a time- and dose-dependent manner (Fuks et al., 2012). Further, hMDDCs also consistently secreted GABA upon parasite infection with a donor-to-donor quantitative variability (Fuks et al., 2012). Parasitized DCs, but not by-stander DCs, secreted GABA, indicating that intracellular localization of live parasites was necessary for GABAergic activation of DCs. In line, GABA secretion was associated with active tachyzoite invasion but not adhesion of parasites to the cell membrane, challenge with heat-inactivated tachyzoites or lysate, infected cell supernatants and LPS. Pharmacological inhibition of GABA synthesis or transport did not impact on parasite replication inside DCs or on the DC viability, suggesting that GABAergic signaling was independent of the host cell and parasite developments.

GABAergic cells are defined by their ability to produce GABA through the expression of GABA synthesizing enzymes. GAD has two isoforms: GAD65 and GAD67 (Kilb, 2012). The abundant secretion of GABA and expression of GAD65 in Toxoplasma infected-DCs show that DCs become GABAergic cells upon parasite infection.

It remains unknown how the parasite infection regulates GABA synthesis and secretion by DCs. It is plausible that parasite effector molecules interact with components of the GABAergic system such as enzymes synthesizing or metabolizing GABA (GADs) and GABA-T or transporters of GABA (GATs) to induce these secretions **(Figure 2B)**. Upon parasitic infection in

dissolution of podosomes. Scale bar = 10µM. (B,C) Representative motility plots of unchallenged (black) and Toxoplasma-infected (red) murine bone marrow-derived DCs (mBMDCs), respectively. Infected cells exhibit prolonged migratory paths and elevated velocities. Cells were imaged and tracked as described in Weidner et al. (2013). X- and Y-axes indicate µm.

mBMDCs, transcriptional expression of GAD65 was unaltered but expression of GAT4 was seven-fold upregulated (Fuks et al., 2012). Also, the intervention in GABA synthesis by pharmacological inhibition of GAD (using semicarbazide) abolished GABA secretion in the supernatants whereas inhibition of the transporter GAT4 (using SNAP 5114) diminished GABA secretion to around half (Fuks et al., 2012; Kanatani et al., 2017). This suggests that GABA is not only synthesized de novo, but also transported more efficiently outside the cells implying roles for both GAD65 and GAT4 in the hypermigratory phenotype.

Of note, Toxoplasma utilizes a metabolic GABA-shunt as an energy source under nutrient-limited conditions (MacRae et al., 2012). In respect to hypermigration of parasitized DCs, inhibition of GABA synthesis or transport decreased the concentration of GABA only in the supernatants of infected DCs but not in the supernatants from free (extracellular) tachyzoites, indicating that the host cell GABAergic machinery is responsible for the augmented secretions of GABA.

Recently (Brooks et al., 2015) identified delocalization of GAD67 in Toxoplasma-infected rodent brain as an indication of disturbed GABA signaling in CA3 pyramidal neurons and increased susceptibility of the animals to experimentally induced epileptic seizures. If GAD67 plays a role in the hypermigration of primary DCs remains to be investigated.

### Modulation of GABA-A Receptors in Parasitized DCs

The abundant GABA secretions in the supernatants of infected DCs would presumably act in an autocrine fashion on GABA-A receptors in the host cell membrane. Since GABA-A receptors are fast Cl<sup>−</sup> ion channels with opening time in the millisecond range and [Cl−]<sup>i</sup> in monocytic cells is higher than in mature neurons (ranging from 24 to 75 mM), GABA-mediated activation of GABA-A channels in DCs results in inwards currents, as shown by patch clamp electrophysiology (Fuks et al., 2012) i.e.,

(Continued)

FIGURE 2 | The CCCs i.e., NKCCs/KCCs are involved in the maintenance of Cl<sup>−</sup> gradient in DCs. In murine DCs, expression of the L-type VDCC CaV1.3 predominates over other expressed VDCCs while microglia express a broader set of VDCCs. (B) T. gondii actively invades host cells and resides intracellularly in a parasitophorous vacuole (PV). Shortly after parasite invasion, DCs exhibit (i) enhanced GABA synthesis through GAD, (ii) upregulation of the transporter GAT4, (iii) elevated expression of GABA-A R subunit mRNAs, indicative of increased receptor trafficking to the membrane, and (iv) elevated NKCC1 activity leading to an increase in [Cl−] i . A similar GABAergic activation occurs in microglia upon Toxoplasma infection. The secreted GABA acts in an autocrine fashion and activates GABA-A R. Opening of GABA-A channels results in Cl<sup>−</sup> efflux from the cell producing depolarization of the plasma membrane. Depolarization activates VDCCs, preferentially the subtype CaV1.3 in DCs, and leads to calcium (Ca2+) entry into the cell. Hypothetically, Ca2<sup>+</sup> acts as second messenger to promote cellular signaling implicated in motility, transmigration, chemotaxis and transcriptional modulation. Hitherto unidentified intracellular targets may include 14-3-3-regulated MAP kinase activity. Higher intracellular Ca2<sup>+</sup> concentrations or fluxes are likely required for the observed rapid cytoskeletal rearrangements, such as the dissolution of podosomes and integrin redistributions implicated in the hypermigratory phenotype.

efflux Cl<sup>−</sup> ion out of the cell. Thus, GABAergic activation will increase the positive charge inside the cell and depolarize the plasma membrane (Tian et al., 2004; Cahalan and Chandy, 2009; Bhandage et al., 2018a).

As indicated above, mBMDCs express mRNAs for five GABA-A receptor subunits (α3, α5, β1, β3, ρ1). Interestingly, the expression of α3 and ρ1 subunits was upregulated by 2 h postinfection (Fuks et al., 2012). This probably contributes to de novo assembly of GABA-A pentamers that can traffic to the membrane. Increase in the number of functional GABA-A pentamers on the membrane would enhance GABAergic signaling, and thereby depolarization in DCs **(Figure 2B)**. GABA has been shown to induce migration and chemotaxis in immature neurons in vitro through the activity of ρ subunit-containing GABA-A receptors (Denter et al., 2010). Similarly, the upregulation of the ρ1 subunit in Toxoplasma-infected DCs raises the possibility that ρ1-containing GABA-A receptors may be involved in the hypermigration of DCs. Further, parasitic infection may possibly impact on intracellular molecules such as protein kinases that regulate GABA-A receptor activity. Given the expression level of GABA-A receptor subunits and their functional status upon parasite infection in mBMDCs, we speculate that opening of few GABA-A channels per cell per unit time may be sufficient to change the membrane potential significantly, further activating downstream signaling (Chandy et al., 2004; Tian et al., 2004; Cahalan and Chandy, 2009; Feske et al., 2012).

Altogether, Toxoplasma-induced activation of GABAergic signaling will depolarize the DCs, reaching the threshold membrane potential for opening of VDCCs and resulting in calcium entry into the cell. Previous reports have shown that depolarizing GABA-A receptor activation can cause calcium influx through VDCCs without triggering action potentials in immature neurons and stimulate neuronal migration, chemotaxis and maturation (Lin et al., 1994; Behar et al., 1996; Ganguly et al., 2001; Owens and Kriegstein, 2002; Ben-Ari, 2012). It was therefore reasonable to explore whether T. gondii infection induced DC migration through a similar mechanism, as delineated below.

#### GABAergic Signaling Mediates Enhanced Transmigration and Hypermotility of *Toxoplasma*-Infected DCs

GABA-A receptor blockade inhibits hypermigration of Toxoplasma-infected DCs (Fuks et al., 2012). Similarly, interference of GABA synthesis or transportation by pharmacological inhibition of GADs or GATs, respectively, abrogates transmigration and hypermotility in parasitized DCs. Addition of supernatants from infected DCs or exogenous GABA in presence of the inhibitors of GAD and GAT reconstituted the transmigration frequencies and hypermotility of DCs (Kanatani et al., 2017). Further, adoptively transferred parasitized DCs in mice exhibited impaired migration and reduced dissemination upon treatment with inhibitors of GAD and GAT as compared to non-treated cells. Additionally, the parasite load in brain, spleen and mesenteric lymph nodes of mice was substantially lower upon GABAergic inhibition. This suggests that the hypermigratory phenotype cannot be induced without active GABAergic signaling in DCs.

#### Regulation of the GABA-A Receptor Activity by CCCs

The activity of GABA-A receptors i.e., depolarization (excitation) or hyperpolarization (inhibition) depends on the [Cl−]<sup>i</sup> which, in turn, is set by CCCs. The most commonly studied CCCs in the CNS are NKCC1 and KCC2 out of 2 and 4 from NKCC and KCC family of solute carriers, respectively, and their expression varies depending on developmental stages, cell types and sub-cellular localization (Blaesse et al., 2009; Kaila et al., 2014). Thus, CCCs are essential for maintaining Cl<sup>−</sup> homeostasis to obtain depolarizing or hyperpolarizing GABAergic responses (Glykys et al., 2014).

Since Toxoplasma infection modulates the GABAergic system in DCs, it would be interesting to know which CCCs are instrumental in DCs for maintenance of [Cl−]<sup>i</sup> . Any changes in CCC expression could alter [Cl−]<sup>i</sup> which eventually will alter the activity of functional GABA-A receptors in DCs. Some evidences indicate that monocytic cells have significantly higher [Cl−]<sup>i</sup> than that of mature neurons (Ince et al., 1987; DeFazio et al., 2000). With this high [Cl−]<sup>i</sup> , DCs ought to depolarize upon GABA-A receptor activation, similar to the depolarizing effects of GABA on immature neurons (Kilb, 2012). However, the CCC expression repertoire of CCC in DCs and the putative impact by Toxoplasma infection remain uncharacterized. Yet, recent data indicates that the CCC expression is modulated in primary microglia upon Toxoplasma infection (Bhandage et al., 2019).

# CALCIUM SIGNALING DOWNSTREAM TO GABAergic SIGNALING IS ESSENTIAL FOR THE INDUCTION OF A HYPERMIGRATORY PHENOTYPE IN DCs

## CaV1.3 Channels Are the Active L-Type VDCCs Mediating Hypermigration in DCs

In sub-physiological concentrations of calcium, the Toxoplasmainduced hypermigratory phenotype of DCs was abrogated (Kanatani et al., 2017). In addition, nickel ions dosedependently inhibited induction of hypermotility. Since nickel ions compete with physiological calcium by blocking the permeation path of VDCCs, this suggested the involvement of VDCCs.

A transcriptional analysis showed that mBMDCs constitutively expressed 9 different Ca<sup>V</sup> channel poreforming α1 subunits, indicating a possibility for formation of functional Ca<sup>V</sup> channels (Kanatani et al., 2017). When the most prominently expressed CaV1.3 channels were pharmacologically inhibited or genetically silenced, the abolished transmigration and hypermotility in mBMDCs were not recovered by addition of exogenous GABA **(Figure 2B)**. Inhibition of other calcium channels, e.g., silencing of CaV1.2 channels or antagonism of purinergic receptors, non-significantly impacted on hypermotility, indicating that CaV1.3 was the key VDCC subtype in DCs responsible for mediating Toxoplasma-induced hypermigration.

Additionally, calcium influx through L-type VDCCs can regulate the activity of GABA-A receptors through phosphorylation of the β3 subunit by calmodulin-dependent protein kinase II (CaMKII) (Saliba et al., 2012). This suggests that CaV1.3 channels and GABA-A receptors may cooperate and mutually regulate each other in DCs.

# GABAergic Signaling Induces VDCC-Mediated Calcium Signaling in DCs

Exposure of DCs to GABA evoked a transient calcium influx, suggesting that GABA-A receptor-induced depolarization can open VDCCs for calcium influx (Kanatani et al., 2017). Additionally, inhibition of GABAergic signaling abrogated hypermotility, which was reconstituted by VDCC activation (Bay-K8644). In sharp contrast, GABAergic activation was unable to recover the hypermotility of DCs upon blockade of Ca<sup>V</sup> channels by broad inhibitors of VDCCs (nifedipine, benidipine), by a selective CaV1.3 channel antagonist or by CaV1.3 gene silencing. Jointly, this indicated that VDCC-dependent calcium signaling mediated hypermotility downstream of GABAergic signaling **(Figure 2B)**.

# Calcium Signaling Through VDCCs Results in Hypermigration

Calcium, being an important second messenger, participates in multiple cellular processes. Local openings of VDCCs on the plasma membrane are crucial for neurotransmission in neurons or the fusion of insulin-containing vesicles in pancreatic islet beta cells (Lin et al., 1994; Behar et al., 1996; Gandasi et al., 2017). Similarly, in Toxoplasma-infected hypermotile DCs, the subcellular microdomains in the plasma membrane may be important for signaling, and therefore for conferring the migratory activation. The observed transient calcium fluxes in response to GABA may participate in the dissolution of adhesion-mediating podosomes (Weidner et al., 2013), the redistribution of integrins (Kanatani et al., 2015), the balance between activity of matrix metalloproteinases (MMPs) and tissue inhibitor of metalloproteinase 1 (TIMP1) (Olafsson et al., 2018) and thus, ultimately in the cytoskeletal remodeling driving the DCs toward a hypermigratory phenotype. Of note, TIMP1 is released by calcium-dependent vesicular exocytosis (Dranoff et al., 2013) and is instrumental in the reduced proteolytic activity and migratory activation of parasitized DCs (Olafsson et al., 2018). Possibly, high-resolution live cell calcium imaging might elucidate the regulation of these processes by calcium.

Why is Toxoplasma manipulating inotropic receptors in the host cell? The effector functions achieved through the modulation of inotropic receptors are rapid, in range of seconds, and therefore offer the advantage of bypassing transcriptional regulation in the host cell. Thus, it might be in favor of Toxoplasma to hijack GABAergic and VDCC signaling in DCs to achieve a rapid onset of hypermigration shortly after hostcell invasion and thereby speed-up the process of systemic dissemination (Lambert et al., 2006; Fuks et al., 2012).

# MODULATION OF IONOTROPIC SIGNALING IN MICROGLIA AND OTHER BRAIN-RESIDENT CELLS

Recent work has shown that Toxoplasma infection can alter GABAergic synapses in the rodent brain (Brooks et al., 2015). Altering of the distribution of the GABA synthesis enzyme GAD67 led to disturbed GABA signaling and increased the susceptibility of animals to experimentally induced epileptic seizures. Also, Toxoplasma infection decreased the expression of the astrocytic glutamate transporter, GLT1, and increased extracellular glutamate levels (David et al., 2016). Thus, the observed dysregulation of GABAergic and glutamatergic signaling in toxoplasmosis is intriguing, also in perspective of the infiltration of DCs to the brain parenchyma during Toxoplasma infection (John et al., 2011). Additionally, astrocytes and microglia are both permissive to parasite invasion and replication in vitro but only microglia exhibited enhanced transmigration and hypermotility upon challenge with T. gondii (Dellacasa-Lindberg et al., 2011; Contreras-Ochoa et al., 2012; Bhandage et al., 2019). This raises the question whether microglia serve as Trojan horses for Toxoplasma dissemination within the brain parenchyma.

Human microglia have been reported to express GABA-T and 3 GABA-A receptor subunits (α1, α3, and β1) (Lee et al., 2011). In addition, microglia respond to GABA by suppressing IFN-γ production through inhibition of inflammatory pathways mediated by NF-kB and P38 mitogen-activated protein (MAP) kinases. These inhibitory effects of GABA are partially mimicked by the GABA-A receptor agonist–muscimol and the GABA-B receptor agonist -baclofen, implying functionality for both types of GABA receptors in human microglia (Lee et al., 2011). In mouse retinal microglia, endogenous GABAergic signaling negatively regulated dendritic morphology in vivo in the brain as the inhibition of GABA-A or GABA-B receptors resulted in more profound increase in dendritic structures (Fontainhas et al., 2011). A recent characterization in murine primary microglia revealed that (i) microglia exhibit hypermotility upon challenge with T. gondii, (ii) secretion of GABA upon T. gondii infection, (iii) transcriptional expression of a complete GABAergic machinery including GABA-A receptors, (iv) T. gondii infection modulated the expression of the GABAergic machinery, and (v) pharmacological inhibition of GABAergic signaling at different levels (synthesis, receptor, receptor regulator, and VDCC inhibition) abrogated hypermotility of microglia (Bhandage et al., 2019). Jointly, this indicates that T. gondii infection activates migration of microglia via GABAergic signaling, similar to DCs.

Additional immune cells that mediate important immune responses to Toxoplasma infection, such as monocytes, T cells, NK cells, macrophages, and neutrophils, have been associated with the parasite dissemination, as delineated above. Whether Toxoplasma infection modulates GABAergic signaling in other immune cells—if expressed—needs to be addressed in future investigations.

# A ROLE FOR Tg14-3-3 IN *T. gondii*-INDUCED HYPERMIGRATION OF DCs AND MICROGLIA

Much of the current work in understanding the Trojan horse mechanism has focused on the infected host cell. Cytoskeletal changes with dissolution of podosomes, elevation in motility and transmigration, modulated interactions with extracellular matrix and regulation of integrin receptors, among others, have been reported in DCs or monocytic cells following infection by T. gondii (Lambert et al., 2006; Harker et al., 2013; Weidner et al., 2013; Kanatani et al., 2015; Cook et al., 2018; Olafsson et al., 2018).

By analyzing the impact of fractionated total tachyzoite lysates on DC motility, T. gondii 14-3-3 (Tg14-3-3), a parasite-derived orthologous protein of the ubiquitously expressed 14-3-3 protein family of eukaryotic cells, was linked to hypermigration of DCs and microglia (Weidner et al., 2016). In the absence of other T. gondii proteins, recombinant Tg14-3-3 was sufficient to induce a hypermigratory state in DCs and microglia, with velocities comparable to that of a live Toxoplasma infection. Tg14-3-3 was detected in secreted parasite fractions and localized to the parasitophorous vacuolar space in infected DCs. Interestingly, a rapid recruitment of host cell 14-3-3 to the parasitophorous vacuole membrane (PVM) was observed (Weidner et al., 2016). Thus, the PVM may serve as the potential interface for Tg14-3-3 interaction with host cellular proteins.

Little is known about the functions of the 14-3-3 proteins in apicomplexan parasites, and specifically Tg14-3-3 (Assossou et al., 2003, 2004; Lorestani et al., 2012). In mammalian eukaryotic cells 14-3-3 proteins are involved in the MAP kinase-mediated regulation of cellular processes including the organization of the cytoskeleton and the cell motility (Sluchanko and Gusev, 2010). Further, specific human isoforms of 14-3- 3 can promote cell migration and metastasis of cancer cells through cytoskeletal remodeling (Somanath and Byzova, 2009; Freeman and Morrison, 2011). It is therefore plausible that Tg14- 3-3, directly or indirectly through sequestration of host 14-3-3, impacts on MAP kinase signaling and thereby motility.

### GABAergic MIGRATORY ACTIVATION OF DCs CAN COOPERATE WITH CHEMOTAXIS OF *Toxoplasma*-INFECTED LEUKOCYTES

Upon contact with microbes in peripheral tissues, DC maturation implies changes in the expressed chemokine receptors that license migration to draining lymph nodes. As delineated above, Toxoplasma-induced hypermigration of DCs depends on GABAergic signaling (Fuks et al., 2012), but not on classical chemokine receptors, e.g., chemokine receptor 7 (CCR7) or Toll-like receptor (TLR)/MyD88 signaling (Lambert et al., 2006; Olafsson et al., 2018). Yet, hypermigratory DCs down-regulate CCR5 and up-regulate CCR7 upon Toxoplasma challenge (Fuks et al., 2012; Weidner et al., 2013), in line with reported chemotactic responses by CD34<sup>+</sup> DCs (Diana et al., 2004). Secretion of the dense granule protein GRA5 has been associated with the upregulation of CCR7, and CCR7/CCL19-driven chemotaxis (Persat et al., 2012). Indeed, DCs challenged with soluble GRA5, or live tachyzoites, chemotaxed in a CCL19 gradient (Fuks et al., 2012; Persat et al., 2012; Weidner et al., 2013).

The impact of chemokine receptor modulation in Toxoplasma-infected DCs remains however unexplored in vivo. In vitro, the onset of GABAergic hypermigration following parasite invasion is rapid (minutes) while the onset of measurable chemotactic responses is significantly slower in vitro (12–24 h) (Fuks et al., 2012; Weidner et al., 2013; Kanatani et al., 2015). Yet, hypermotility and chemotaxis cooperatively potentiated the speed and directional motility of parasitized DCs in vitro. Thus, GABAergic hypermotility and CCR7-mediated chemotaxis may, in theory, jointly potentiate migration of infected DCs in vivo, and facilitate parasite dissemination.

#### THE *Trojan horse* MECHANISM AND FREE EXTRACELLULAR TACHYZOITES: TWO CO-EXISTING MODES OF PARASITE DISSEMINATION?

Intracellular localization in migratory leukocytes represents a secluded niche for dissemination, partly protecting from immune attack in the hostile extracellular environment. However, despite T. gondii's obligate intracellular existence for replication (Dobrowolski and Sibley, 1996), its extracellular gliding motility mechanism provides a means for migration in the microenvironment in tissues (Barragan and Sibley, 2002). Freshly egressed tachyzoites can actively traverse polarized epithelial and endothelial cell monolayers. The identified process of paracellular transmigration implicates interactions between host cell ICAM-1 and the parasite adhesin MIC2 (Barragan and Sibley, 2002, 2003; Barragan et al., 2005; Furtado et al., 2012b).

Shortly after inoculation of parasites in murine experimental models, leukocyte-associated tachyzoites can be detected in the circulation but also free extracellular tachyzoites that increase in numbers as infection develops (Lambert et al., 2009; Konradt et al., 2016). The relative contribution of these two possible modes for dissemination in natural infections in rodents or humans remains unexplored. However, it is likely that, following systemic dissemination in the blood, passage to the brain occurs with low parasitemia during natural primary infection. Moreover, extracellular tachyzoites are exposed to neutralization by the complement system and IgM (Couper et al., 2005). Also, parasite genotype-related differences have been described. For type II and III strains, the relative leukocyte-associated fraction of tachyzoites predominated early during infection (Lambert et al., 2009), in line with observations of leukocyte-associated type II parasites early after infection (Courret et al., 2006; Unno et al., 2008). In contrast, for type I parasites, the relative extracellular tachyzoite fraction was predominant in the spleen (Lambert et al., 2009). Additionally, transfer of tachyzoites between different leukocyte types has been shown to occur in the blood and tissues (Persson et al., 2007, 2009; Kanatani et al., 2017).

One additional possibility is that infected leukocytes in circulation "deliver" the parasites to the endothelium (Lambert and Barragan, 2010), as recently reported for a pulmonary infection model in mice (Baba et al., 2017). In infection models of toxoplasmosis, both parasitized leukocytes and free extracellular tachyzoites have been suggested to mediate dissemination in ocular infection (Furtado et al., 2012a,b, 2013) and for intraluminal intestinal spreading in mice (Coombes et al., 2013; Gregg et al., 2013). Jointly, this indicates that both intracellular and extracellular dissemination strategies co-exist and may even act in a complementary fashion at different phases of the infection.

### REFERENCES


# CONCLUSIONS AND PERSPECTIVES

Toxoplasma gondii has developed strategies for hijacking the migratory functions of infected leukocytes, which simultaneously serve as a replicative niche. Thus, T. gondii reconciles the obligate need for intracellular replication with the establishment of infection in peripheral organs. To this end, T. gondii modulates the motogenic GABAergic/VDCC signaling axis of DCs and microglia to hijack migratory functions and promote its dissemination. Similarly, upcoming evidences indicate that parasites, bacteria and viruses modulate GABAergic signaling in immune cells for survival (Fuks et al., 2012; Zhu et al., 2017; Kim et al., 2018).

The impact of GABA and other neuroactive molecules in immune cells is an emerging field. Infection models, such as toxoplasmosis, can increase the current understanding of the GABAergic signaling system in immune cells and on how GABAergic signaling impacts on physiological and pathophysiological conditions. Additionally, novel insights into the pathogenesis of infections are provided. Research from recent years has revealed that immune cells are GABAergic and future studies will likely uncover novel functions for GABA signaling in immune cells.

#### AUTHOR CONTRIBUTIONS

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

#### FUNDING

The work was supported by the Swedish Research Council (2018–02411), the Fredrik and Ingrid Thurings Foundation (2017-00349, 2018-00404) and the Olle Engkvist Foundation (193–609).


neurons via calcium-dependent mechanisms. J. Neurosci. 16, 1808–1818. doi: 10.1523/JNEUROSCI.16-05-01808.1996


interactions during recall responses in the lymph node. Immunity 31, 342–355. doi: 10.1016/j.immuni.2009.06.023


tolerance and succinate biosynthesis. Appl. Environ. Microbiol. 79, 74–80. doi: 10.1128/AEM.02184-12


entwined? Front. Plant Sci. 6:419. doi: 10.3389/fpls.2015. 00419


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

Copyright © 2019 Bhandage and Barragan. 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.

# *Toxoplasma-*Induced Hypermigration of Primary Cortical Microglia Implicates GABAergic Signaling

Amol K. Bhandage† , Sachie Kanatani † and Antonio Barragan\*

Department of Molecular Biosciences, The Wenner-Gren Institute, Stockholm University, Stockholm, Sweden

#### *Edited by:*

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### *Reviewed by:*

Ildiko Rita Dunay, Universitätsklinikum Magdeburg, Germany Melissa Lodoen, University of California, Irvine, United States

> *\*Correspondence:* Antonio Barragan antonio.barragan@su.se

†These authors have contributed equally to this work

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 30 December 2018 *Accepted:* 05 March 2019 *Published:* 20 March 2019

#### *Citation:*

Bhandage AK, Kanatani S and Barragan A (2019) Toxoplasma-Induced Hypermigration of Primary Cortical Microglia Implicates GABAergic Signaling. Front. Cell. Infect. Microbiol. 9:73. doi: 10.3389/fcimb.2019.00073 Toxoplasma gondii is a widespread obligate intracellular parasite that causes chronic infection and life-threatening acute infection in the central nervous system. Previous work identified Toxoplasma-infected microglia and astrocytes during reactivated infections in mice, indicating an implication of glial cells in acute toxoplasmic encephalitis. However, the mechanisms leading to the spread of Toxoplasma in the brain parenchyma remain unknown. Here, we report that, shortly after invasion by T. gondii tachyzoites, parasitized microglia, but not parasitized astrocytes, undergo rapid morphological changes and exhibit dramatically enhanced migration in 2-dimensional and 3-dimensional matrix confinements. Interestingly, primary microglia secreted the neurotransmitter γ-aminobutyric acid (GABA) in the supernatant as a consequence of T. gondii infection but not upon stimulation with LPS or heat-inactivated T. gondii. Further, microglia transcriptionally expressed components of the GABAergic machinery, including GABA-A receptor subunits, regulatory molecules and voltage-dependent calcium channels (VDCCs). Further, their transcriptional expression was modulated by challenge with T. gondii. Transcriptional analysis indicated that GABA was synthesized via both, the conventional pathway (glutamate decarboxylases GAD65 and GAD67) and a more recently characterized alternative pathway (aldehyde dehydrogenases ALDH2 and ALDH1a1). Pharmacological inhibitors targeting GABA synthesis, GABA-A receptors, GABA-A regulators and VDCC signaling inhibited Toxoplasma-induced hypermotility of microglia. Altogether, we show that primary microglia express a GABAergic machinery and that T. gondii induces hypermigration of microglia in a GABA-dependent fashion. We hypothesize that migratory activation of parasitized microglia by Toxoplasma may promote parasite dissemination in the brain parenchyma.

Keywords: apicomplexa, central nervous system, glia, leukocyte migration, neurotransmission, GABA receptor

# INTRODUCTION

Toxoplasma gondii is a globally widespread parasite that infects virtually all warm-blooded organisms, including humans and rodents (Joynson and Wreghitt, 2001). Chronic carriage of T. gondii without major symptomatology is common. However, systemic dissemination of T. gondii can cause life-threatening infection that manifests as Toxoplasma encephalitis in immune-compromised patients (Joynson and Wreghitt, 2001). T. gondii is obligate intracellular and the tachyzoite stage actively invades and replicates within nucleated cells in the host (Frénal et al., 2017). Tachyzoites subvert the migratory properties of leukocytes, e.g., dendritic cells (DCs) (Lambert et al., 2006), and use shuttle leukocytes (Trojan horse) to rapidly reach the blood circulation and the central nervous system (CNS) (Courret et al., 2006; Lambert et al., 2006, 2009). Fast-replicating tachyzoite stages can infect microglia, astrocytes, and neurons in vitro (Lüder et al., 1999; Scheidegger et al., 2005).

Microglia are the principal resident immune cell in the CNS and originate from primitive hematopoietic precursors outside the CNS, in the embryonic yolk sac (Ginhoux et al., 2010). Microglia rapidly respond to tissue injury and inflammation (Kreutzberg, 1996; Nimmerjahn et al., 2005). They are likely an important source of IFN-γ and interact with CD4<sup>+</sup> and CD8<sup>+</sup> lymphocytes, thus contributing to acquired immunity during toxoplasmic encephalitis (Suzuki et al., 2005). Astrocytes also play important roles in resistance to T. gondii during chronic infection (Wilson and Hunter, 2004). Thus, multiple functions have been attributed to glia cells during Toxoplasma infection in murine models, including cytokine production and phagocytosis (Strack et al., 2002; Wilson and Hunter, 2004; Suzuki et al., 2005; Carruthers and Suzuki, 2007). Microglial nodules in the CNS have also been described during Toxoplasma infection in humans (Nebuloni et al., 2000). Yet, the mechanisms for parasite dissemination locally in the CNS parenchyma, that is associated with life-threatening encephalitis, remain unknown.

Ŵ-aminobutyric acid (GABA), the main inhibitory neurotransmitter in the vertebrate brain, has also been attributed motogenic functions outside the CNS including cell migration and metastasis (Azuma et al., 2003; Wheeler et al., 2011). Along these lines, Toxoplasma induces hypermigration of infected DCs through GABAergic signaling (Fuks et al., 2012; Kanatani et al., 2017). GABA-A receptors (GABA-A R) are ionotropic chloride channels composed from pentameric combinations of 19 different subunits (Olsen and Sieghart, 2008) and whose functions are regulated by cation-chloride co-transporters (CCCs) (Kahle et al., 2008). GABA is shuttled in and out of cells via GABA transporters (GAT) (Höglund et al., 2005) and GABAergic cells synthesize GABA via glutamate decarboxylases (GAD65/67) and can metabolize it by GABA-transaminase (GABA-T) (Soghomonian and Martin, 1998). GABA can also be synthesized from putrescine (Seiler et al., 1973) and more recent work has characterized an alternative GABA synthesis pathway via monoamine oxidase B (MAOB) and aldehyde dehydrogenases (ALDH2 and ALDH1a1) in neurons and astrocytes (Yoon et al., 2014; Kim et al., 2015). Thus, GABAergic signaling has been extensively studied in neurons and astrocytes (Lee et al., 2011; Kilb, 2012) but remains chiefly unexplored in microglia (Barragan et al., 2015; Bhandage and Barragan, 2019).

Here, we report that migratory activation of Toxoplasmainfected microglia occurs via GABAergic signaling. We discuss the possible implications of GABA secretion and the migratory activation of microglia in the pathogenesis of Toxoplasma encephalitis.

# MATERIALS AND METHODS

#### Parasites and Cell Line

Toxoplasma gondii lines used include GFP-expressing PTGluc (type II, cloned from ME49/PTG-GFPS65T) and RFP-expressing PRU-RFP (type II). Tachyzoites were maintained by serial 2-day passaging in human foreskin fibroblast (HFF-1 SCRC-1041, American Type Culture Collection) monolayers cultured in Dulbecco's modified Eagle's medium (DMEM; Thermofisher scientific) with 10% fetal bovine serum (FBS; Sigma), gentamicin (20µg/ml; Gibco), glutamine (2 mM; Gibco), and HEPES (0.01 M; Gibco), referred to as complete medium (CM). The murine microglia cell line BV2 (American Type Culture Collection) was cultured in CM.

# Primary Glia Cells and Microglia

Primary glia cell cultures were generated as follows. One- to three-day-old pups from C57BL/6 mice were euthanized. The brains were dissected and cortices collected. Cortices were further washed in ice-cold Ca2+- and Mg2<sup>+</sup> free Hanks' buffered salt solution (HBSS; Gibco), minced, and resuspended in icecold HBSS. After being washed, tissues were incubated for 15 min in HBSS containing 0.1% trypsin and resuspended in astrocyte medium containing DMEM F-12 (Gibco), 10% FBS, 1% G5 supplement (Gibco), and gentamicin (20µg/ml; Gibco). Medium was changed every 2–3 days. Microglia were harvested as described previously (Dellacasa-Lindberg et al., 2011). Briefly, confluent astrocyte monolayers derived from primary glia cultures were sub-cultivated in microglia medium containing DMEM F-12, 10% FBS, glutamine (2 mM; Gibco), and gentamicin (10µg/ml). Microglial cells were harvested from confluent astrocyte monolayers after 7 days by tapping the side of the culture flasks, removing loosely adherent microglia from astrocyte monolayers. Microglia purity was assessed by transcriptional analysis of a marker panel (**Figure S1** and **Table S1**).

#### Reagents

Lipopolysaccharide (LPS), 4-Diethylaminobenzaldehyde (DEAB; aldehyde dehydrogenase inhibitor), selegiline (MAO-B inhibitor), nifedipine (L-type VDCC inhibitor, all from Sigma-Aldrich), L-allylglycine (L-AG; GAD inhibitor), picrotoxin (GABA-A receptor inhibitor), bumetanide (NKCC1 inhibitor), benidipine (broad VDCC inhibitor, all from Tocris Bioscience), and 1-(3-Chlorophenethyl)-3-cyclopentylpyrimidine-2,4,6- (1H,3H,5H)-trione (CPCPT, CaV 1.3 inhibitor, Merck Millipore) were used at the indicated concentrations. Heat inactivation of T. gondii tachyzoites was performed at 56◦C for 30 min. Supernatants were collected from microglia incubated with freshly egressed T. gondii tachyzoites (MOI 1, 24 h) and added at a final concentration of 1:1.

#### Motility Assays

Motility and velocity analyses were performed as previously described (Weidner et al., 2013). Briefly, microglia were challenged with freshly egressed tachyzoites or treated with

localization of individual microglia (DAPI) at indicated conditions. (F) Mean migrated distances by microglia under same condition as in E. Data represent compiled analysis of 500 randomly chosen cells from 2 independent experiments in duplicate. For (B,D,F), bar graphs represent mean + SEM. Statistical significance was tested by One-Way ANOVA with Dunnett's post-hoc test. ns p ≥ 0.05, \*p < 0.05, \*\*p < 0.01, \*\*\*p < 0.001.

antagonists. Cells were seeded on 96-well plates or matrigelcoated (100 µg/cm<sup>2</sup> , Corning) labtech chambers (Thermofisher). The cells were imaged every min for 60 min (Zeiss Observer Z.1). Motility patterns for 50–60 cells per experimental group were compiled using ImageJ (image stabilizer software and manual tracking plugins). Transmigration assays with astrocytes were performed in transwell filters (8µm pore size, BD Falcon), as previously described (Fuks et al., 2012). Pharmacological inhibitors were added at concentrations that non-significantly impacted on cell morphology, baseline motility of unchallenged cells, and on parasite infection rates (Fuks et al., 2012).

#### 3D Migration Assay

The assay was performed as previously described (Kanatani et al., 2015). Briefly, a collagen layer was prepared using bovine collagen I (0.75 mg/ml, GIBCO) in a 96-well plate (80 µl/well). Microglia were challenged with freshly egressed T. gondii tachyzoites (MOI 3) in CM for 4 h. The cell suspension (5 × 10<sup>4</sup> cells) was applied to the collagen layer and incubated for 18 h at 37◦C and 5% CO2. Gels were fixed and DAPI-stained. Image stacks were generated (200 optical sections) by confocal microscopy (LSM 780, Zeiss) and migrated distances by cells were analyzed using Imaris x64 v.8.1.1 software (Bitplane AG, Zurich, Switzerland).

#### Immunocytochemistry

Host cells were cultured on glass coverslips. Fixation was performed with 4% PFA in PBS for 15–20 min at RT. The cells were permeabilized using 0.5% Triton X-100 in PBS. To visualize host cell F-actin and podosomes, cells were stained with Alexa Fluor 488- or 594- or 647-conjugated phalloidin (Invitrogen). To probe GABA-A subunits, cells were incubated with rabbit anti-GABA-A R α3 polyclonal antibody (Alomone labs), rabbit anti-GABA-A R α5 polyclonal antibody (Synaptic system), and mouse anti-GABA-A R β3 monoclonal antibody (NeuroMab) ON at 4◦C. Following staining with Alexa Fluor 488-conjugated secondary antibodies (Invitrogen) and DAPI, coverslips were mounted and imaged by confocal microscopy (LSM 780 or LSM 800, Zeiss).

# Scoring of Cell Morphology

Phalloidin-stained microglia were monitored by epifluorescence microscopy (Leica DMRB). For each preparation, 20–30 randomly chosen fields of view were assessed and an average of 100 cells were counted per condition. The microglia were scored based on morphological criteria, as previously described for DCs (Weidner et al., 2013).


#### Real-Time Quantitative PCR

Total RNAs were extracted using Direct-zol miniprep RNA kits with TRIzol reagent (Zymo Research). First-strand cDNA was synthesized using Superscript III or IV Reverse Transcriptase (Invitrogen) using a standard protocol provided by manufacturer. Real-time quantitative PCR (qPCR) was performed in Rotor-Gene 6000 (Corbett Research) or QuantStudio 5 384 Optical well plate system (Applied Biosystem) in a standard 10 µl with the 2X SYBR FAST qPCR Master Mix (KAPA Biosystems) and gene specific primers using a standard amplification protocol followed by melt curve analysis. The gene specific DNA primer pairs were designed to cover all the transcripts available currently using GETprime or NCBI primer blast tool, ordered from Sigma-Aldrich and validated on whole brain homogenates (**Table S2**). The criteria for positive detection of a signal were presence of single peak at specific temperature in the melt curve and presence of a single band in agarose gel electrophoresis. The relative expression levels (2−1Ct) were calculated for each target relative to a normalization factor -geometric mean of reference genes, glyceraldehyde 3-phosphate dehydrogenase (GAPDH), and Actin-β or TATA-binding protein (TBP) and importin 8 (IPO8).

# GABA Enzyme-Linked Immunosorbent Assays (ELISA)

ELISA (Labor Diagnostica Nord, Nordhorn, Germany) was performed as previously described (Fuks et al., 2012). Briefly, microglia were plated at a density of 5 × 10<sup>5</sup> cells per well and incubated for 24 h in presence of T. gondii tachyzoites or reagents, as indicated. GABA concentrations in supernatants quantified at a wavelength of 450 nm (VMax <sup>R</sup> Kinetic ELISA Microplate Reader, Molecular Devices).

#### Statistical Analysis

Statistical analyses were performed using GraphPad Prism 7.0 (La Jolla, CA, USA) and R Stats Package version 3.0.2 (R Foundation for Statistical Computing, Vienna, Austria). The significance level was set to p < 0.05.

# RESULTS

#### Primary Cortical Microglia Exhibit Morphological Changes and Enhanced Migration Upon Infection With *T. gondii*

We previously described that Toxoplasma-infected DCs undergo rapid morphological changes and exhibit hypermotility in absence of chemotactic cues (Weidner et al., 2013). Similarly, infected microglia exhibited enhanced transmigration in vitro (Dellacasa-Lindberg et al., 2011). To further determine the cellular processes underlying this migratory activation, we characterized morphological changes in microglia upon challenge with T. gondii tachyzoites by staining F-actin filaments (**Figure 1A**). Toxoplasma-infected microglia were consistently characterized by loss of actin-rich cytoskeletal podosomes and a rounded morphology (**Figures 1A,B**). These observations were confirmed in the microglia cell line BV2 (**Figure S2A**). Next, a motility analysis was performed to address a possible link between the infection-associated morphological changes and the previously described enhanced transmigration of Toxoplasma-infected microglia (Dellacasa-Lindberg et al., 2011). Infected primary microglia and BV2 cells migrated significantly longer distances at higher velocities compared with unchallenged microglia or microglia challenged with LPS, heat-inactivated T. gondii or supernatant from infected microglia (**Figures 1C,D** and **Figures S2B,C**). This indicated that the migratory activation was linked to the presence of live intracellular parasites. In sharp contrast, infected primary astrocytes exhibited undistinguishable morphological changes compared to uninfected astrocytes (**Figure S3A**) and nonsignificant changes in motility (**Figures S3B,C**). Further, we analyzed the migratory activation of Toxoplasma-challenged microglia in a collagen matrix (Kanatani et al., 2015). In this 3D setting, infected microglia and BV2 cells migrated significantly longer distances compared with untreated or LPS-treated microglia (**Figures 1E,F** and **Figure S2D**). Altogether, we conclude that, upon infection with T. gondii, microglia undergo morphological changes and exhibit enhanced migration in 2D and 3D confinements.

# Expression of GABAergic and VDCC Signaling Components by Primary Cortical Microglia

To date, GABAergic signaling in microglia has remained chiefly unexplored (Barragan et al., 2015). The implication of microglia during toxoplasmic encephalitis (Dellacasa-Lindberg et al., 2011) and the participation of GABAergic/VDCC signaling in hypermigration of Toxoplasma-infected DCs (Fuks et al., 2012; Kanatani et al., 2017) motivated an assessment of GABAergic and VDCC signaling components in microglia. First, microglia and astrocyte preparations were characterized by transcriptional expression of a panel of markers. A prominent differential expression of Iba1, CD11b (microglia markers) and GFAP, GLT1, Aquaporin 4 (astrocyte markers) was detected (**Figure S1** and **Table S1**). Second, a transcriptional analysis of primary cortical microglia revealed presence of mRNAs for (i) GABA enzymes GAD65, GAD67, and GABA-T (**Figure 2A**), (ii) GABA transporters GAT2, GAT4, and bestrophin 1 (BEST1) (**Figure 2B**), (iii) 15 GABA-A R subunits (α1-5, β1-3, γ1-3, δ, ε, ρ1-2) (**Figure 2C**), (iv) CCCs including NKCC1-2, KCC1-4, and NCC (**Figure 2D**), and (v) 10 VDCC Ca<sup>V</sup> subunits (**Figure 2E**). Additionally, the expression of these components was assessed in astrocytes and whole brain (**Figures 2A–E**). All the components analyzed were detected in the whole brain samples. Further, an analysis of the relative expression in astrocytes and microglia revealed differential expression of GABA-T, GAT1, GAT3, GAT4, BEST1, GABA-A R subunits α4, β1, and δ, and a VDCC channel Ca<sup>V</sup> 2.3 (**Table S3**). Two of the GABA transporters, GAT1 and GAT3, were undetectable in microglia whereas BEST1 was undetectable in astrocytes. Additionally, the relative quantitative expression of GABA-T, GAT4, Ca<sup>V</sup> 2.3, α4, and β1 GABA-A R subunits was ∼40- to 60-fold higher in astrocytes whereas the expression of the δ GABA-A R subunit was ∼14-fold higher in microglia compared with astrocytes (**Table S3**). In general, microglia exhibited a broad expression of GABA-A R subunits with predominance of the β3 subunit (**Figure 2C**), similar to the expression profile of astrocytes. In addition, immunocytochemical stainings were consistent with protein expression of the α3, α5, and β3 GABA-A R subunits in microglia (**Figure 3**). A similar setup of CCCs and VDCCs was transcribed in microglia, astrocytes, and brain (**Figures 2D,E**), advocating for constitutive roles in these cells and brain tissue. Overall, we conclude that primary microglia transcriptionally express the necessary components to form functional GABAergic and VDCC signaling systems.

#### Challenge With *T. gondii* Modulates the Expression GABAergic and VDCC Signaling Components in Microglia

To address the impact of T. gondii infection on the microglial GABAergic and VDCC signaling components, the mRNA expression was assessed at 2, 4, 12, and 24 h post-infection and related to unchallenged microglia at the corresponding time point. The data revealed modulated expression of multiple components. Shortly after Toxoplasma challenge, microglia upregulated the expression of the GABA synthesis enzymes GAD65/67 while expression of the GABA degradation enzyme GABA-T was downregulated (**Figure 4A** and **Table S4**). Additionally, expression of the GABA transporter GAT4 was upregulated (**Figure 4B** and **Table S4**). Jointly, these changes are all in theory consistent with elevated GABA concentrations. Further, additional components modulated (>50% up- or down) by Toxoplasma challenge included: (i) GABA-A R subunits α2- 5, β1, β3, γ1-3, δ, ρ1, and ρ2 (**Figure 4C** and **Table S4**), (ii) CCCs NKCC1-2, KCC1-3, and NCC (**Figure 4D** and **Table S4**), and (iii) VDCCs CaV1.3, 1.4, 2.1, and 3.1 (**Figure 4E** and **Table S4**). Thus, Toxoplasma infection modulates the expression of components of the GABAergic and VDCC signaling systems in microglia.

#### Implication of the GABAergic and VDCC Signaling Systems in *Toxoplasma*-Induced Hypermotility of Microglia

Upon Toxoplasma challenge of microglia, we observed transcriptional upregulation of GABA synthesis enzymes GAD65/67 and GABA transporters, with down-modulation of GABA-degrading enzyme GABA-T. To determine if this, in fact, translated into elevation of GABA, we assessed secretion of GABA in cell supernatants. Importantly, both primary microglia and BV2 microglia cells challenged with T. gondii tachyzoites exhibited significantly elevated GABA concentrations in the supernatant, contrasting non-significant effects upon challenge with heat-inactivated tachyzoites or LPS (**Figure 5A** and **Figure S2E**). In addition to the enzymes GAD65/67 that constitute the conventional GABA synthesis pathway, the mRNA for other enzymes, MAO-B, ALDH2, and ALDH1a1, known to synthesize GABA from putrescine by an alternative pathway in astrocytes and neurons, were also detected in primary microglia (**Figure 5B**). Additionally, T. gondii infection modulated their expression (**Figure 5C**).

Further, direct pharmacological inhibition of the GABA synthesis enzymes, GAD65/67 by L-allylglycine (**Figures 5D,E**) and ALDH1a1 and ALDH2 by DEAB (**Figures 5D,F**) abolished hypermotility, whereas an inhibitor of MAO-B (selegeline, acting upstream of ALDH2/ALDH1a1) non-significantly reduced hypermotility (**Figures 5D,F**). This indicated a putative activation of both the conventional and the alternative pathway for GABA synthesis (Kim et al., 2015) and the implication of GABA in the induction of hypermotility in Toxoplasma-infected microglia.

Next, we assessed the implication of GABA-A Rs and VDCCs by applying pharmacological antagonism. Indeed, antagonism of (i) GABA-A R by the direct channel pore blocker picrotoxin, (ii) NKCC1, a regulator of GABA-A R, by bumetanide, and (iii) VDCCs by the broad inhibitor benidipine and L-type inhibitor, nifedipine abolished hypermotility, with non-significant effects on the base-line motility of unchallenged microglia (**Figures 5G,H**). In contrast to DCs (Kanatani et al., 2017), the CaV1.3 inhibitor CPCPT non-significantly impacted the hypermotility of Toxoplasma-infected microglia (**Figures 5G,H**). We conclude that, upon Toxoplasma challenge, microglia secrete GABA and that inhibition of GABAergic signaling by

targeting GABA synthesis, GABA-A R antagonism or regulator antagonism, and VDCC antagonism abolishes hypermotility.

#### DISCUSSION

Previous work has established that T. gondii tachyzoites induce a hypermigratory phenotype in parasitized DCs and monocytic cells (Lambert et al., 2006; Kanatani et al., 2017; Cook et al., 2018). Yet, the impact of Toxoplasma infection on the migratory functions of glial cells, that mediate immune surveillance in the CNS, has remained unknown.

Our data demonstrate that a migratory activation of primary cortical microglia sets in upon T. gondii infection, in both 2D and 3D matrix confinements. The onset of highvelocity locomotion was accompanied by the dissolution of adhesion-mediating podosome structures and the acquisition of rounded cell morphology. Thus, parasitized microglia acquired morphological characteristics and locomotion that are reminiscent of the high-speed amoeboid migration mode of activated DCs (Lämmermann et al., 2008) and of hypermotile Toxoplasma-infected DCs (Weidner et al., 2013; Kanatani et al., 2015). Interestingly, similar morphological changes and migratory activation were confirmed in the microglia cell line BV2 but were absent in primary astrocytes. In spite that primary microglia and astrocytes are similarly permissive to Toxoplasma infection in vitro (Dellacasa-Lindberg et al., 2011), this underlines that the hypermigratory response differs between the two cell types. The data are in line with the previously observed enhanced transmigration by Toxoplasmainfected microglia, but not by astrocytes (Dellacasa-Lindberg et al., 2011) and also differences among leukocyte types (Lambert et al., 2011). Also, microglia are highly migratory cells in response to inflammatory cues while astrocytes have additional important structural functions in the CNS (Sofroniew and Vinters, 2010). Further, the finding that LPS, supernatants

from infected microglia or heat-inactivated tachyzoites were unable to induce hypermotility in microglia indicates that the migratory activation was linked to the presence of intracellular T. gondii. In inflammatory conditions, highly migratory amoeboid microglial cells (AMC) have been reported (Deng et al., 2009). However, contrasting with the classical activation observed for AMC, a down-modulation of activation markers, e.g., MHC II and CD86, was observed in Toxoplasmainfected microglia (Dellacasa-Lindberg et al., 2011). This indicated that alternative activation mechanism(s) lied behind Toxoplasma-induced hypermotility.

We report that GABAergic signaling is implicated in the Toxoplasma-induced hypermotility of primary cortical microglia. In relation to neurotransmission, the GABAergic systems of neurons and astrocytes have been extensively studied (Lee et al., 2011; Kilb, 2012). However, because chiefly immune surveillance functions have been attributed to microglia, their GABAergic potential has remained unexplored (Barragan et al., 2015). Human microglia have been previously reported to express GABA-Transaminase and 3 GABA-A R subunits (α1, α3, and β1), indicating GABAceptive functions (Lee et al., 2011). Notably, we report that murine primary microglia secrete GABA upon Toxoplasma challenge, indicating they are GABAergic cells. Additionally, microglia transcriptionally expressed GABA synthesis and degradation enzymes, GABA transporters and GABA-A R subunits, all consistent with the existence of a GABAergic system.

We show that both primary cortical microglia and a microglia cell line (BV2) respond to Toxoplasma infection with GABA secretion and hypermotility. Primary microglia transcriptionally expressed the enzymes that constitute the conventional pathway of GABA synthesis (GAD 65/67) and upon challenge with Toxoplasma, both GAD65 and GAD67 mRNA exhibited a significant up-regulation (∼2- to 4-fold). Somewhat in contrast, only GAD65 mRNA was detected in infected DCs (Fuks et al., 2012) and the mRNA of the GABA catabolic enzyme ABAT was more strongly expressed in astrocytes compared with microglia (∼44-fold difference). Jointly, this is indicative of differences in GABA metabolism between the two cell types. Further, pharmacological inhibition of GAD65/67 inhibited Toxoplasma-induced hypermotility in microglia but required high concentrations of the inhibitor. However, pharmacological inhibition of the alternative pathway enzymes ALDH2/ALDH1a1 also inhibited hypermotility. Jointly, this advocates that the conventional and alternative GABA synthesis pathways cooperate in GABA production in microglia and that GABA synthesis is necessary to maintain hypermotility, as previously demonstrated in DCs (Fuks et al., 2012; Kanatani et al., 2017). Also, similar to observations in DCs, supernatants from infected microglia (containing secreted GABA) were insufficient to induce hypermigration of naïve microglia. This is

FIGURE 5 | (MAO-A, MAO-B) was analyzed in unchallenged microglia as indicated under Materials and Methods. (C) Modulation of enzymes indicated in (B) in Toxoplasma-challenged microglia in comparison to unchallenged microglia is presented as percentage increase (red color intensity scale) or decrease (blue scale) in a heat map at indicated time points. (D) Representative motility plots of microglia incubated with PRU tachyzoites and treated for 4 h with L-allylglycine (L-AG), DEAB, and selegiline as indicated under Materials and Methods. (E) Mean velocities of unchallenged and Toxoplasma-challenged microglia, as in (D), treated with GAD inhibitor L-allylglycine (L-AG) at concentrations ranging from 62.5 to 1,000µM. (F) Mean velocities of unchallenged and Toxoplasma-challenged microglia, as in (D), incubated with complete medium (CM), DEAB (10µM) or selegeline (10µM). (G) Representative motility plots of microglia incubated with PRU tachyzoites and treated for 4 h with picrotoxin, benidipine or CPCPT. (H) Mean velocities of unchallenged and Toxoplasma-challenged microglia incubated with complete medium (CM), picrotoxin (50µM), bumetanide (10µM), benidipine (10µM), nifedipine (10µM), or CPCPT (1µM). For (A,B,E,F,H), bar graphs represent mean + SEM from 3 independent experiments. For (A), statistical significance was tested by Student's t-test and for (E,F,H), by One-Way ANOVA with Dunnett's post-hoc test. ns p ≥ 0.05, \*p < 0.05, \*\*p < 0. 01, \*\*\*p < 0. 001.

also in line with the concept that a live intracellular tachyzoite is necessary to trigger the hypermigratory phenotype (Fuks et al., 2012; Kanatani et al., 2017) and may indicate that prior GABA-A R activation is necessary.

Another major difference between microglia, astrocytes, and DCs was the transcriptional expression of GABA transporters. While microglia preferentially expressed GAT2 (and GAT1 was undetectable), astrocytes preferentially expressed GAT1 (while BEST1 was undetectable). Additionally, microglia expressed the GABA regulators CCCs and VDCCs, with modulated expression upon Toxoplasma challenge.

Further, the subunit repertoire of transcriptionally expressed GABA-A R subunits by microglia was broader than that of myeloid DCs (Fuks et al., 2012) and remarkably similar to the repertoire expressed by astrocytes, despite the different ontogenic origins of these two cell types (Hochstim et al., 2008). However, quantitative differences were present. For example, the δ subunit exhibited ∼14-fold higher relative expression in microglia while the α4 and β1 subunits exhibited ∼40- to 60-fold higher relative expression in astrocytes. A caveat of the purification protocol is always the possibility of contaminating astrocytederived mRNA in microglia purifications and vice versa. However, quantitative analyses based on transcription of classical microglia and astrocyte markers showed that this contamination should be a minor contributor and cannot explain the overall qualitative and quantitative expression differences between microglia and astrocytes. Jointly, the data show that microglia are GABAergic cells. Future research needs to determine how GABA receptors and other GABAergic components traffic in the infected cell and become activated and/or modulated by the infection.

Importantly, pharmacological antagonism targeting various levels of the GABAergic motogenic axis, i.e., GABA synthesis, GABA-A R antagonism, GABA-A R regulator or VDCC antagonism, abolished Toxoplasma-induced hypermotility, concordant with results in DCs (Fuks et al., 2012; Kanatani et al., 2017). However, differences also exist. For example, a specific inhibitor of the VDCC subtype CaV1.3 abolished the Toxoplasma-induced hypermotility of DCs (Kanatani et al., 2017) but non-significantly impacted on the hypermotility of microglia. This corresponds well with the predominant expression of CaV1.3 over other VDCC subtypes in DCs and the broader expression, of CaV1.3 but also additional VDCC subtypes, in cortical microglia. Jointly, the data underline the functional implication of GABAergic signaling in Toxoplasma-induced hypermigration of primary microglia.

Additionally, the parasite-derived 14-3-3 molecule has been linked to the hypermigratory phenotype in both DCs and microglia (BV2), with sequestration of host-cell 14-3-3 (Weidner et al., 2016). 14-3-3 is an abundant molecule in the CNS (Sluchanko and Gusev, 2010) that can regulate GABA receptor function (Laffray et al., 2012). Thus, future research needs to address how 14-3-3 is implicated in Toxoplasma-induced GABAmediated hypermotility of microglia.

The finding that Toxoplasma infection induces GABA secretion by microglia raises additional questions in relation to infection in the CNS. As the neurotransmitter systems are tightly regulated, any changes in the expression of components or the regulation of these systems might lead to altered balance between excitation and inhibition in the CNS (Semyanov et al., 2004). Our data are well in line with recent descriptions of GABAergic dysregulation (Brooks et al., 2015) and glutamatergic dysregulation (David et al., 2016) in murine neurotoxoplasmosis. Hypothetically, local elevations of GABA concentrations by Toxoplasma infection may alter the neuronal functions, with impacts on normal brain physiology and altered host behavior (Brooks et al., 2015). Additionally, GABA has been attributed chemokinetic and chemotactic effects on migrating embryonic neurons (Behar et al., 1996). In DCs, a GABA-mediated motogenic effect is observed but no chemotactic effect was detected (Fuks et al., 2012). Of note is that astrocytes, despite expressing GABA receptors and secreting GABA, did not respond with hypermotility upon Toxoplasma infection. This may be due to differential composition of subunits or differential activation profile (and therefore functionality) of GABA receptors or due to differences in signaling downstream of GABA receptor activation, e.g., signal transduction via VDCCs and other downstream signaling pathways (Kanatani et al., 2017). Finally, mounting evidence implicates GABA in the immune functions of T cells and DCs (Barragan et al., 2015). This raises the question whether GABA should be considered a "neuro-immuno-transmitter," as recently suggested for dopamine (Levite, 2016).

Based on the data at hand, we hypothesize that the induced hypermigration of infected microglia may facilitate parasite dispersion in the brain parenchyma. The findings also open up for the speculation that GABA secretion by Toxoplasmainfected glial cells may have migratory effects of surrounding cells in the parenchymal microenvironment and modulate GABA levels locally leading to altered neuronal functions. Finally, leukocytes have been implicated in the delivery of Toxoplasma to the brain parenchyma (Courret et al., 2006) with rapid parasite transfer between microglia and T cells (Dellacasa-Lindberg et al., 2011). Thus, the production of GABA by infected cells could hypothetically serve as a signal attracting new host cells and thereby facilitating dissemination within the parenchyma. These alternatives need to be explored and constitute novel perspectives on the pathogenesis of toxoplasmic encephalitis.

#### DATA AVAILABILITY

All datasets generated for this study are included in the manuscript and/or the supplementary files.

#### ETHICS STATEMENT

The Regional Animal Research Ethical Board, Stockholm, Sweden, approved protocols involving extraction of cells from mice, following proceedings described in EU legislation (Council Directive 2010/63/EU).

# REFERENCES


# AUTHOR CONTRIBUTIONS

AKB and SK performed experiments and analyzed data. AKB, SK, and AB conceived experimental design and wrote the manuscript.

# FUNDING

The work was supported by the Swedish Research Council (2018- 02411), the Fredrik and Ingrid Thurings Foundation (2017- 00349, 2018-00404), and the Olle Engkvist Foundation (193-609).

# ACKNOWLEDGMENTS

We are grateful to the members of the Barragan lab and Prof Bryndis Birnir, Uppsala University, for critical discussion. Imaging was performed at the Stockholm University Imaging Facility (IFSU).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00073/full#supplementary-material


dendritic cells infected by Toxoplasma gondii. PLoS Pathog. 13:e1006739. doi: 10.1371/journal.ppat.1006739


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

Copyright © 2019 Bhandage, Kanatani and Barragan. 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.

# Comprehensive Kinetic Survey of Intestinal, Extra-Intestinal and Systemic Sequelae of Murine Ileitis Following Peroral Low-Dose *Toxoplasma gondii* Infection

#### Markus M. Heimesaat <sup>1</sup> \*, Ildiko R. Dunay <sup>2</sup> and Stefan Bereswill <sup>1</sup>

1 Institute of Microbiology, Infectious Diseases and Immunology, Charité - University Medicine Berlin, Corporate Member of Freie Universität Berlin, Humboldt-Universität zu Berlin, and Berlin Institute of Health, Berlin, Germany, <sup>2</sup> Medical Faculty, Institute of Inflammation and Neurodegeneration, University Hospital Magdeburg, Magdeburg, Germany

#### *Edited by:*

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### *Reviewed by:*

Chrystelle Bonnart, INSERM U1220 Institut de Recherche en Santé Digestive, France Françoise Debierre-Grockiego, Université de Tours, France

#### *\*Correspondence:*

Markus M. Heimesaat markus.heimesaat@charite.de

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

> *Received:* 22 January 2019 *Accepted:* 25 March 2019 *Published:* 12 April 2019

#### *Citation:*

Heimesaat MM, Dunay IR and Bereswill S (2019) Comprehensive Kinetic Survey of Intestinal, Extra-Intestinal and Systemic Sequelae of Murine Ileitis Following Peroral Low-Dose Toxoplasma gondii Infection.

Front. Cell. Infect. Microbiol. 9:98. doi: 10.3389/fcimb.2019.00098 We have recently shown that following peroral low-dose Toxoplasma gondii infection susceptible mice develop subacute ileitis within 10 days. Data regarding long-term intestinal and extra-intestinal sequelae of infection are scarce, however. We therefore challenged conventional C57BL/6 mice with one cyst of T. gondii ME49 strain by gavage and performed a comprehensive immunopathological survey 10, 36, and 57 days later. As early as 10 days post-infection, mice were suffering from subacute ileitis as indicated by mild-to-moderate histopathological changes of the ileal mucosa. Furthermore, numbers of apoptotic and proliferating/regenerating epithelial cells as well as of T and B lymphocytes in the mucosa and lamina propria of the ileum were highest at day 10 post-infection, but declined thereafter, and were accompanied by enhanced pro-inflammatory mediator secretion in ileum, colon and mesenteric lymph nodes that was most pronounced during the early phase of infection. In addition, subacute ileitis was accompanied by distinct shifts in the commensal gut microbiota composition in the small intestines. Remarkably, immunopathological sequelae of T. gondii infection were not restricted to the intestines, but could also be observed in extra-intestinal tissues including the liver, kidneys, lungs, heart and strikingly, in systemic compartments that were most prominent at day 10 post-infection. We conclude that the here provided long-term kinetic survey of immunopathological sequalae following peroral low-dose T. gondii infection provides valuable corner stones for a better understanding of the complex interactions within the triangle relationship of (parasitic) pathogens, the host immunity and the commensal gut microbiota during intestinal inflammation. The low-dose T. gondii infection model may be applied as valuable gut inflammation model in future pre-clinical studies in order to test potential treatment options for intestinal inflammatory conditions in humans.

Keywords: subacute and chronic ileitis, low-dose *Toxoplasma gondii* infection, Th1-type immunopathology, gut microbiota changes, host-pathogen-interactions

# INTRODUCTION

Within 1 week following peroral high-dose infection with more than 50 cysts of the intracellular protozoan parasite Toxoplasma gondii strain ME49 susceptible mice irrespective whether harboring a murine or human gut microbiota develop acute necrotizing inflammation of the terminal ileum with lethal outcome (Heimesaat et al., 2006; Munoz et al., 2011; von Klitzing et al., 2017a,b). This CD4+ T lymphocyte dependent inflammatory condition is characterized by massive secretion of pro-inflammatory mediators such as TNF, IFN-γ, and nitric oxide, whereas IL-10 expression is upregulated as counterregulatory measure (Liesenfeld et al., 1996, 1999; Jankovic et al., 2010; Munoz et al., 2011). Given the terminal ileum as predilection site, the underlying Th1-type immunopathology, and the pronounced changes in the commensal gut microbiota composition (i.e., dysbiosis) characterized by an overgrowth of the inflamed ileal lumen with commensal Escherichia coli and Bacteroides / Prevotella species, the high-dose T. gondii infection model mimics key features of human inflammatory bowel diseases such as Crohn's disease (Liesenfeld, 2002; Heimesaat et al., 2006; Munoz et al., 2009, 2011, 2015). However, data regarding long-term intestinal and extra-intestinal sequelae of peroral murine T. gondii post-infection resulting in intestinal inflammation are lacking due to the fatal immunopathological course within 1 week post infection (p.i).

We have previously established a subacute and non-lethal ileitis model following peroral low-dose (i.e., one cyst) T. gondii ME49 strain infection (Dunay et al., 2008; Escher et al., 2018; Heimesaat et al., 2018a). In order to further dissect the interplay between parasite, host immunity and gut microbiota (the latter mimicking human conditions) during infectioninduced inflammation we had generated mice harboring a complex human gut microbiota prior ileitis induction (Bereswill et al., 2011; von Klitzing et al., 2017a,b,c,d; Escher et al., 2018; Heimesaat et al., 2018a,b). Within 9 days p.i. mice were suffering from T cell dependent subacute ileitis that was characterized by rather mild symptoms and mild-to-moderate ileal histopathological changes without necrosis (Escher et al., 2018; Heimesaat et al., 2018a). We did, however, not follow-up mice suffering from subacute ileitis beyond day 9 p.i.

This prompted us in the present study to perform a comprehensive long-term immunopathological survey of conventional C57BL/6j mice until day 57 following ileitis induction. We here demonstrate that immunopathological sequelae of ileitis induction (i) were most severe during the early phase of peroral low-dose T. gondii infection (i.e., around day 10 p.i.), (ii) affected not only the terminal ileum but also the colon, (iii) were accompanied by distinct shifts in the commensal microbiota composition within the ileal lumen, and (iv) were not restricted to the intestinal tract but could also be observed in extra-intestinal and (v) strikingly, even in systemic compartments.

# MATERIALS AND METHODS

# Ethics Statement

Mouse experiments were performed in accordance with the European Guidelines for animal welfare (2010/63/EU). The protocols for these experiments were approved by "Landesamt für Gesundheit und Soziales," LaGeSo, Berlin (registration numbers G244/98). Clinical conditions of mice were assessed at least once daily.

# Mice and *T. gondii* Infection

C57BL/6j mice were reared and housed under specific pathogen free (SPF) conditions in the Forschungseinrichtungen für Experimentelle Medicine (Charité—University Medicine Berlin). Conventionally colonized female mice 12 weeks of age were included into the experiments. On day (d) 0, mice were perorally subjected to low-dose T. gondii infection (i.e., one cyst of strain ME49 in a volume of 0.3 mL by gavage) as described recently (Heimesaat et al., 2006, 2018a). T. gondii DNA was quantitated in homogenized ileal tissue (∼1 cm<sup>2</sup> ) as stated elsewhere (Munoz et al., 2009).

#### Sampling Procedures

At defined time points, namely d10, d36, or d57 p.i., mice were sacrificed by isofluran inhalation (Abbott, Germany). Naive mice served as uninfected controls. Upon necropsy, cardiac blood, luminal ileal samples as well as ex vivo biopsies from mesenteric lymph nodes (MLN), spleen, liver, kidneys, lungs, heart, colon and ileum were obtained under sterile conditions.

#### Histopathology

After removal ileal ex vivo biopsies were immediately fixed in 5% formalin and embedded in paraffin. After hematoxylin and eosin (H&E) staining, histopathological changes were quantitatively assessed in 5 µm thin sections applying a standardized scoring system ranging from 0 to 6 as stated elsewhere (Heimesaat et al., 2006).

#### IMMUNOHISTOCHEMISTRY

For in situ immunohistochemical analyses ex vivo biopsies were obtained from the ileum, colon, liver, kidneys, lungs and the heart, immediately fixed in 5% formalin and embedded in paraffin. In order to quantitatively determine apoptotic and proliferating epithelial cells, macrophages and monocytes, T lymphocytes and B lymphocytes cells, paraffin sections (5µm) were stained with primary antibodies directed against cleaved caspase 3 (Asp175, Cell Signaling, Beverly, MA, USA, 1:200), Ki67 (TEC3, Dako, Denmark, 1:100), F4/80 (# 14-4801, clone BM8, eBioscience, San Diego, CA, USA, 1:50), CD3 (#N1580, Dako, 1:10), and B220 (No. 14-0452- 81, eBioscience; 1:200), respectively, as described earlier (Heimesaat et al., 2018b). Positively stained cells were then determined applying light microscopy (magnification 100

**Abbreviations:** CBA, cytometric bead array; H&E, hematoxylin and eosin; IBD, inflammatory bowel disease; IFN, interferon; IL, interleukin; MLN, mesenteric lymph nodes; NOD, nucleotide-oligomerization-domain; PBS, phosphate-buffered saline; p.i., post-infection; SPF, specific pathogen free; TLR, toll-like receptor; TNF, tumor necrosis factor; WT, wildtype.

x and 400 x), and for each mouse the average number of respective positively stained cells was assessed within at least six high power fields (HPF, 0.287 mm<sup>2</sup> , 400 x magnification) by a blinded independent investigator. In order to provide a broader overview of the T. gondii induced immunopathological sequelae, representative photomicrographs of the immunohistochemically stained paraffin sections are provided with lower magnification (100 x) in the Online **Supplemental Materials**.

#### Pro- and Anti-inflammatory Mediator Detection

Intestinal ex vivo biopsies were cut longitudinally, washed in phosphate buffered saline (PBS; Gibco, Life Technologies, UK), and strips of ∼1 cm<sup>2</sup> tissue as well as ex vivo biopsies derived from MLN (3 lymph nodes), liver (∼1 cm<sup>3</sup> ), one kidney (cut longitudinally), one lung, and spleen (one third) were transferred to 24-flat-bottom well-culture plates (Nunc, Germany) containing 500 µL serumfree RPMI 1640 medium (Gibco, life technologies, UK) supplemented with penicillin (100 U/mL) and streptomycin (100µg/mL; PAA Laboratories, Germany). After 18 h at 37◦C, respective culture supernatants as well as serum samples were tested for TNF, IFN-γ, MCP-1, IL-6, IL-12p70, and IL-10 by the Mouse Inflammation Cytometric Bead Assay (CBA; BD Biosciences, Germany) on a BD FACSCanto II flow cytometer (BD Biosciences). Nitric oxide was measured by the Griess reaction as stated elsewhere (Heimesaat et al., 2006).

#### Gut Microbiota Analyses

DNA was extracted from ileal luminal samples as reported recently (Heimesaat et al., 2006; Bereswill et al., 2014). In brief, DNA was quantitated applying the Quant-iT PicoGreen reagent (Invitrogen, UK) and adjusted to 1 ng per µL. The total eubacterial load as well as the main bacterial groups within the murine gut microbiota such as enterobacteria, enterococci, lactobacilli, bifidobacteria, Bacteroides / Prevotella species, Clostridium coccoides group, and Clostridium leptum group were determined by quantitative real-time polymerase chain reaction (qRT-PCR) with species-, genera- or groupspecific 16S rRNA gene primers (Tib MolBiol, Germany) as stated elsewhere (Heimesaat et al., 2010; Bereswill et al., 2011; Rausch et al., 2013) (expressed as numbers of 16S rRNA gene copies per ng DNA).

#### Statistical Analysis

Medians and levels of significance were determined by oneway ANOVA or Kruskal-Wallis test followed by Tukey postcorrection for multiple comparisons (GraphPad Prism v7, USA) as indicated. Two-sided probability (p) values ≤ 0.05 were considered significant. Experiments were reproduced three times.

#### RESULTS

#### Intestinal Inflammatory Changes Upon Peroral Low-Dose *T. gondii* Infection

In order to induce subacute ileitis, conventionally colonized mice were perorally subjected to low-dose (i.e., one cyst of) T. gondii challenge by gavage and to a comprehensive immunopathological survey for almost 2 months p.i. Within 10 days upon infection mild-to-moderate histopathological changes of the ileal mucosa could be observed ranging from edematous mucosal blubbing, leukocytic infiltrates of the ileal mucosa and lamina propria, cell-free exudates into the small intestinal lumen, and cellular shedding (p < 0.001 vs. naive; **Figure 1A** and **Figure S1A**). As compared to day 10 p.i., median histopathological scores quantitatively assessing ileal mucosal damage tended to be lower at later time points, but did not reach statistical significance due to high standard deviations (n.s.; **Figure 1A** and **Figure S1A**). We additionally quantitatively surveyed intestinal inflammatory responses applying in situ immunohistochemistry of intestinal paraffin sections. In T. gondii infected mice, multifold increased apoptotic ileal epithelial cell numbers could be determined at days 10 and 57 (p < 0.001 and p < 0.01, respectively vs. naive) with highest counts in ex vivo biopsies taken at day 10 p.i. (p < 0.001 vs. day 36 and day 57 p.i.; **Figure 1B** and **Figure S1B**). In parallel, Ki67+ ileal epithelial cells indicative for cell proliferation and regeneration counteracting T. gondii induced cell damage increased to a maximum at day 10 p.i. (p < 0.001), declined thereafter, but were still higher as compared to those in naive control mice (p < 0.001; **Figure 1C** and **Figure S1C**).

Given that T. gondii induced ileitis is highly T cell dependent (Munoz et al., 2011), we quantitated CD3+ T lymphocytes in the ileal mucosa and lamina propria. In fact, T cell numbers almost doubled with 10 days p.i., but declined back to naive counts thereafter (p < 0.001; **Figure 1D** and **Figure S1D**). In addition, T. gondii infected mice also displayed elevated B cell numbers in their ileal mucosa and lamina propria at day 10 p.i. (p < 0.01; **Figure 1E** and **Figure S1E**), but not at later time points.

We have recently observed that T. gondii induced immunopathological responses within the intestinal tract do not exclusively affect the terminal ileum, but also the large intestines (Escher et al., 2018; Heimesaat et al., 2018a) and therefore expanded our inflammatory survey to the colon. Histopathological mucosal changes within in colonic mucosa were rather minor, however (**Figure S2A**), and did not reach statistical significance due to high standard deviations within respective groups (n.s.; not shown). When quantitating apoptotic cells by in situ immunohistochemistry, multifold increased numbers of apoptotic colonic epithelial cells could be assessed only at day 10 p.i. (p < 0.001; **Figure 2A** and **Figure S2B**), which also held true for T lymphocytes (p < 0.001; **Figure 2C** and **Figure S2D**) and B lymphocytes in the large intestinal mucosa and lamina propria (p < 0.001; **Figure 2D** and **Figure S2E**). Like in the ileum, elevated Ki67+ cells numbers could be observed in the colonic epithelia upon T. gondii infection (p < 0.001; **Figure 2B** and **Figure S2C**),

Hence, apoptotic, proliferating and T cell as well as B cell responses within both, the small as well as the large intestines

post-correction for multiple comparisons and numbers of analyzed mice (in parentheses) are indicated. Data were pooled from four independent experiments.

were most pronounced 10 days following peroral low-dose T. gondii infection.

#### Intestinal Inflammatory Mediator Secretion Upon Peroral Low-Dose *T. gondii* Infection

We next assessed pro-inflammatory mediator secretion in ex vivo biopsies derived from distinct intestinal compartments (**Figure 3**). At day 10 p.i., increased ileal nitric oxide and IFN-γ concentrations could be measured (p < 0.001; **Figures 3A,C**), that declined to naive levels thereafter. In the colon, however, increased TNF concentrations could be observed at either time point p.i. (p < 0.05-0.001; **Figure 3E**) with maximum levels at day 10 p.i. (p < 0.001; **Figure 3E**). Alike the ileum, increased IFN-γ concentrations were measured in large intestinal ex vivo biopsies exclusively at day 10 p.i. (p < 0.001; **Figure 3F**), whereas elevated colonic TNF levels could be measured at days 10 and 57 p.i. (p < 0.001 and p < 0.05, respectively; **Figure 3E)**.

We further surveyed pro-inflammatory mediator secretion in MLN draining the inflamed intestines. As early as day 10 p.i., TNF and IFN-γ concentrations had increased in the MLN (p < 0.001; **Figures 3H,I**), but declined to naive levels later-on, which also held true for nitric oxide (p < 0.05–0.005; **Figure 3G**).

FIGURE 3 | Intestinal pro-inflammatory mediator secretion over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.; closed circles). Naive (N) mice served as uninfected controls (open circles). Pro-inflammatory mediators such as (A,D,G) nitric oxide, (B,E,H) TNF and (C,F,I) IFN-γ were measured in ex vivo biopsies taken from distinct intestinal compartments including the (A–C) ileum, (D–F) colon and (G–I) mesenteric lymph nodes (MLN) at respective time points. Medians (black bars), levels of significance (p-values) as determined by one-way ANOVA test followed by Tukey post-correction for multiple comparisons and numbers of analyzed mice (in parentheses) are indicated. Data were pooled from four independent experiments.

Hence, peroral low dose T. gondii infection was accompanied by pronounced pro-inflammatory mediator secretion in the entire intestinal tract, particularly during the early phase (i.e., the first 10 days p.i.).

#### Intestinal Microbiota Changes Upon Peroral Low-Dose *T. gondii* Infection

Given that inflammatory processes in the intestinal tract are associated with pronounced shifts in the gut microbiota composition (Heimesaat et al., 2006, 2007a,b; Erridge et al., 2010; Bereswill et al., 2011; Fiebiger et al., 2016), we quantitatively surveyed changes in distinct intestinal gut bacterial groups and species applying culture-independent 16S rRNA based methods (**Figures 4A–I**). Within 10 days p.i., enterobacteria and enterococci had increased in luminal samples taken from the inflamed ileum (p < 0.001; **Figures 4B,C**), but declined thereafter (p < 0.05–0.001; **Figures 4B,C**), whereas obligate anaerobic Gram-negative species such as Bacteroides / Prevotella species were slightly higher at either time point p.i. as compared to naive mice (p < 0.05; **Figure 4F**). Conversely, lactobacilli were lower in ileal samples derived from T. gondii infected mice (p < 0.01–0.001; **Figure 4D**), which also held true for bifidobacteria at days 36 and 57 p.i. (p < 0.05–0.01; **Figure 4E**). Converse to enterobacteria and enterococci, the obligate anaerobic Clostridium coccoides and Mouse Intestinal Bacteroides decreased to lowest levels until day 10 p.i. (p < 0.05–0.005; **Figures 4G–I**), but increased later-on (p < 0.05– 0.001; **Figures 4G,I**). For Clostridium leptum, a trend toward lower gene numbers at day 10 p.i. as compared to naïve conditions could be observed (n.s. due to high standard deviations; **Figure 4H**), whereas bacterial loads were higher at days 36 and 37 vs. day 10 p.i. (p < 0.001 and p < 0.01, respectively; **Figure 4H**).

Of note, increased T. gondii DNA could be detected in small intestinal ex vivo biopsies at day 10 p.i., but not later-on (p < 0.001; **Figure 4J**).

Hence, T. gondii induced immunopathological sequelae observed in the ileum were accompanied by pronounced shifts in the luminal gut microbiota composition in the distal small intestines.

#### Extra-Intestinal Inflammatory Changes Upon Peroral Low-Dose *T. gondii* Infection

We next surveyed potential T. gondii induced inflammatory changes in extra-intestinal compartments over time. T. gondii infected mice displayed elevated numbers of caspase3+ cells in their livers exclusively at day 10 p.i. (p < 0.001; **Figure 5A** and **Figure S4A**). In addition, hepatic F4/80+ cells had increased as early as day 10 p.i. (p < 0.001), but declined to naive numbers until day 57 p.i. (p < 0.001; **Figure 5B** and **Figure S3B**). At either time point p.i., mice exhibited increased numbers of both, T and B lymphocytes in their livers (p < 0.001; **Figures 5C,D** and **Figures S3C,D**). Furthermore, T. gondii

representative high power fields (HPF, 400x magnification) per animal the average numbers of (A) apoptotic (casapse3+, Casp3+) and (B) proliferating (Ki67+) cells as well as of (C) T lymphocytes (CD3+) and (D) B lymphocytes were assessed in immunohistochemically stained hepatic paraffin sections. Medians (black bars), levels of significance (p-values) as determined by one-way ANOVA test followed by Tukey post-correction for multiple comparisons and numbers of analyzed mice (in parentheses) are indicated. Data were pooled from four independent experiments.

induced hepatic inflammatory cell responses were accompanied by enhanced secretion of pro-inflammatory mediators measured at day 10 p.i. (p < 0.001 vs. naive; **Figures 6A–C**). At day 36 p.i., hepatic anti-inflammatory IL-10 concentrations were lower as compared to those obtained at day 10 p.i. (p < 0.01; **Figure 6D**).

Enhanced inflammatory responses upon low-dose T. gondii infection could also be observed in the kidneys. In fact, at day 10 following T. gondii infection mice exhibited multifold increased numbers of apoptotic cells in their kidneys, but not later-on (**Figure 7A** and **Figure S4A**), whereas renal T cell numbers were elevated at either time point p.i. (p < 0.05–0.001), but with lower counts at day 57 vs. day 10 p.i. (p < 0.001; **Figure 7B** and **Figure S4B**). Pathogen induced increases in renal B cell numbers could be observed rather late in the course of infection (i.e., days 36 and 57 p.i.; p < 0.001; **Figure 7C** and **Figure S4C**). When assessing pro-inflammatory mediator secretion in renal ex vivo biopsies, nitric oxide, TNF and IFN-γ increased as early as day 10 p.i. (p < 0.001; **Figures 8A–C**). Conversely, IL-10 concentrations were lower at days 10 and 36 p.i. when compared to naive mice (p < 0.001; **Figure 8D**).

We further expanded our inflammatory survey to the lungs. Apoptotic pulmonary cells dramatically increased until day 10 p.i. (p < 0.001), but decreased back to naïve levels later-on (p < 0.001; **Figure 9A** and **Figure S5A**). Low-dose T. gondii infected mice further exhibited increased numbers of F4/80+ cells in the lungs (p < 0.001) with maximum counts at day 36 p.i. (**Figure 9B** and **Figure S5B**). In addition,

indicated. Data were pooled from four independent experiments.

pulmonary T lymphocytes were elevated at either time point following T. gondii infection (p < 0.001; **Figure 9C** and **Figure S5C**), whereas T. gondii induced increases in B lymphocytes could be assessed in the later stages of the infection (i.e., days 36 and 57 p.i.; p < 0.001) with highest counts at the end of the survey; (p < 0.001; **Figure 9D** and **Figure S5D**).

Interestingly, increased secretion of pro-inflammatory cytokines such as TNF and IFN-γ (p < 0.001) as well as of anti-inflammatory IL-10 concentrations (p < 0.05) could be exclusively determined in lungs taken at the earliest time point p.i. (**Figure 10**).

Remarkably, overt T. gondii induced inflammatory changes could also be assessed in the heart as indicated by increased numbers of apoptotic cells at days as well as of T lymphocytes in the cardiac muscle of infected mice (p < 0.01–0.001; **Figures S6A,B**, **S7A,B**), whereas highest apoptotic cell numbers could be counted at the end of the survey (p < 0.05 vs. d10 p.i.; **Figures S6A**, **S7A)**. Furthermore, cardiac B cell numbers were higher on days 36 and 57 p.i. as compared to earlier time points (p < 0.05–0.01; **Figures S6C**, **S7C**).

Hence, pronounced inflammatory responses following peroral low-dose T. gondii infection were not restricted

to the intestinal tract, but could also be observed in extra-intestinal compartments.

#### Systemic Inflammatory Responses Upon Peroral Low-Dose *T. gondii* Infection

We next assessed T. gondii induced immunopathological responses in systemic compartments. Whereas, increased proinflammatory cytokines such as IFN-γ and IL-12p70 could be measured in splenic ex vivo biopsies at day 10 p.i. only (p < 0.001; **Figures 11A,B**), anti-inflammatory IL-10 levels were deprived in the spleen at either time point p.i. (p < 0.001; **Figure 11C**). Strikingly, elevated TNF, IFN-γ, MCP-1 and IL-6 concentrations were obtained in serum samples taken from T. gondii infected mice at day 10 p.i. (p < 0.001), but declined back to baseline levels later-on (p < 0.001; **Figures 12A–D**). In case of IL-10, slightly increased serum concentrations could be exclusively observed at day 36 p.i. (p < 0.05; **Figure 12E**).

Hence, inflammatory responses following peroral low-dose T. gondii infection could be assessed even in systemic compartments that were most pronounced within the first 10 days p.i.

#### DISCUSSION

The peroral high-dose T. gondii infection mouse model has been proven to be well-suited for investigating the underlying molecular mechanism of pathogen-host interactions and subsequently induced acute small intestinal inflammation (McLeod et al., 1989a,b; Liesenfeld et al., 1996, 1999; Liesenfeld, 2002; Vossenkamper et al., 2004; Heimesaat et al., 2006; Munoz et al., 2009, 2011, 2015). In this context one needs to take into consideration, however, that the severity of T. gondii induced ileitis does not meet a linear causal relationship to the amount of parasite cysts the mice had been challenged with. In fact, it is rather a critical threshold of cyst number that needs to be reached to induce the full-blown inflammatory scenario, further depending on the parasitic strain and the genetic background and sex of the murine host (Liesenfeld et al., 1999; Munoz et al., 2011; Heimesaat et al., 2018a). Given the lethal outcome within 7–10 days, information regarding long-term sequelae of T. gondii ME49 strain induced intestinal inflammation are scarce, however.

We have established a far less acute T. gondii induced ileitis model by decreasing the numbers of the perorally applied parasitic infection dose from >50 cysts to only one cyst per recipient mouse (Dunay et al., 2008; Heimesaat et al., 2018a). One can imagine that an accidental increase in the parasitic inoculum may yield an unwantedly more severe ileitis with fatal outcome within <2 weeks. Besides the critically low infection dose, distinct host characteristics such as age and sex of the infected mice might impact the severity of induced immunopathology (Liesenfeld et al., 1999; Dunay et al., 2008; Munoz et al., 2011; Heimesaat et al., 2018a). Hence, the reproducibility and reliability of the peoral low-dose T. gondii infection model critically depends on highly standardized conditions and the experiences of the handling researchers (Heimesaat et al., 2018a). However, with the subacute ileitis model at hand, we are now able to survey the interactions within the triangle relationship ("ménage-à-trois") of the parasitic pathogen, the host immunity and the commensal gut microbiota during intestinal inflammation for 2 months (like in the present study) or even longer. In our very recent low-dose T. gondii infection studies mice were sacrificed at day 9 p.i. (Escher et al., 2018; Heimesaat et al., 2018a).

In our actual immunopathological survey for more than 8 weeks, perorally low-dose T. gondii infected mice developed subacute ileitis within 10 days that was characterized by a non-lethal outcome, rather mild symptoms and mild-tomoderate histopathological changes of the ileal mucosa, but no intestinal necrosis at all. Furthermore, ileal epithelial apoptosis was accompanied by a concomitant increased abundance of proliferating/regenerating epithelial cells in the terminal ileum counteracting the inflammatory damage. Remarkably, T. gondii induced inflammatory changes did not exclusively affect the

terminal ileum which is considered the predilection site following peroral high-dose T. gondii infection (Liesenfeld et al., 1996; Munoz et al., 2011), but also the large intestinal tract given that numbers of apoptotic epithelial cells as well as of T and B lymphocytes in the mucosa and lamina propria of both, the ileum and the colon were highest at day 10 post-infection, but declined thereafter. These inflammatory cellular responses were accompanied by enhanced pro-inflammatory mediator secretion in ileum, colon and mesenteric lymph nodes that was also most pronounced during the early phase of infection. In support, in our previous acute ileitis studies in (with respect to their gut microbiota) "humanized" mice we could demonstrate the colonic co-affection upon peroral high-dose T. gondii infection (von Klitzing et al., 2017a,b). Furthermore, Suzuki and colleagues reported inflammatory changes within the large intestines following peroral infection of IL-10−/<sup>−</sup> mice with 20 cysts of the ME49 strain (Suzuki et al., 2000).

Remarkably, both peroral high-dose and low-dose T. gondii infection resulted in pro-inflammatory sequelae that were not restricted to the intestinal tract, but could also be observed in extra-intestinal and even systemic compartments as shown in our previous reports (von Klitzing et al., 2017a,b; Heimesaat et al., 2018a,b) and actual study. Overall, T. gondii induced and T cell driven pro-inflammatory immune responses were observed in

representative high power fields (HPF, 400x magnification) per animal the average numbers of (A) apoptotic (casapse3+, Casp3+) cells, of (B) macrophages and monocytes (F4/80+), of (C) T lymphocytes (CD3+) and (D) B lymphocytes were assessed in immunohistochemically stained pulmonary paraffin sections. Medians (black bars), levels of significance (p-values) as determined by one-way ANOVA test followed by Tukey post-correction for multiple comparisons and numbers of analyzed mice (in parentheses) are indicated. Data were pooled from four independent experiments.

extra-intestinal organs such as liver, kidneys, lungs, and heart during the entire observation period, but were overall most pronounced during the early phase (i.e., around day 10 p.i.). Depending on respective parasitic infection (i.e., amount and strain of cysts, application via the peroral vs. intraperitoneal route) of mice with a defined genetic background, sampling and detection techniques varying information regarding extraintestinal parasitic infection are available in the literature. For instance, in one study exclusively assessing the parasitic dissemination following peroral infection with 20 cysts of the avirulent T. gondii C strain, parasitic loads had rapidly increased in the lungs until days 7 to 10 and resolved until day 50 p.i. (Derouin and Garin, 1991). Compared to pulmonary parasitic detection, parasitemia was rather delayed during the acute phase of infection (i.e., first 7 days p.i.), whereas low-level parasitemia was shown to occur transiently later-on (Derouin and Garin, 1991). In another infection study, T. gondii could be detected in liver and lungs of mice as early as 7-10 days following peroral challenge with 20 cysts of the avirulent C strain (Sumyuen et al., 1995). Dubey reported parasitemia within 24 h post peroral infection and parasitic invasion of extra-intestinal organs including liver, lungs, kidneys, skeletal and heart muscle

FIGURE 10 | Pulmonal inflammatory mediator secretion over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.; closed circles). Naive (N) mice served as uninfected controls (open circles). (A) TNF, (B) IFN-γ, and (C) IL-10 concentrations were measured in ex vivo biopsies taken from the lungs at respective time points. Medians (black bars), levels of significance (p-values) as determined by one-way ANOVA test followed by Tukey post-correction for multiple comparisons and numbers of analyzed mice (in parentheses) are indicated. Data were pooled from four independent experiments.

between days 11 and 15 p.i. (Dubey, 1997). In these studies, however, no further information regarding immunopathological sequelae within infected compartments are available. Following challenge of C57BL/6 mice with 10 cysts of the ME49 strain via the intraperitoneal route, Liesenfeld and colleagues reported rather mild infiltration of lungs with inflammatory mononuclear cells within 7 days p.i., whereas no inflammatory foci could be assessed in other organs such as the heart or the large intestines (Liesenfeld et al., 1996).

In our present low-dose T. gondii infection study, parasitic DNA were detectable only in ileal ex vivo biopsies taken at day 10 p.i., but were below the detection limit in any other intestinal or extra-intestinal ex vivo biopsies taken during the observation period further underlining that the observed inflammatory changes are rather due to the immunological responses of the initial T. gondii challenge with one single cyst only. One needs to take into consideration that the (parasitic) pathogen does not necessarily need to be permanently in the infected body compartment in order to induce inflammatory sequelae. It is rather the initial hit set by the pathogen to the host subsequently tipping the balance toward immunopathology (Alutis et al., 2015; Heimesaat et al., 2016a,b,c; Grunau et al., 2017).

It is well known that the fine-tuned interactions between the complex intestinal bacterial ecosystem and the host are critical for immune homeostasis, cell physiology and resistance to morbidities (Ekmekciu et al., 2017a). Hence, disturbances in the orchestrated commensal gut microbiota composition are associated with the development and outcome of immunopathological diseases including gastrointestinal inflammation (Heimesaat et al., 2006; Erridge et al., 2010; Bereswill et al., 2011; Fiebiger et al., 2016; Ekmekciu et al., 2017a). Our culture-independent survey of the gut bacterial changes during low-dose T. gondii infection revealed that until day 10 p.i. (when the severity of induced ileitis was maximum) Gramnegative bacterial commensals such as enterobacteria including E. coli had increased in the inflamed ileal lumen, whereas

(p-values) as determined by one-way ANOVA test followed by Tukey post-correction for multiple comparisons and numbers of analyzed mice (in parentheses) are indicated. Data were pooled from four independent experiments.

conversely, Gram-positive lactobacilli and clostridia as well as Mouse Intestinal Bacteroides had decreased. These observations are well in line with results derived from our lethal highdose T. gondii infection studies given that ileitis development was accompanied by dramatic shifts in the ileal microbiota compositions toward an overgrowth with enterobacterial species, whereas bacterial commensal diversity was reduced and particularly lactobacilli and clostridia were virtually undetectable (Heimesaat et al., 2006, 2007a; Erridge et al., 2010). T. gondii induced ileitis was further aggravated by Toll-like receptor (TLR)−4 dependent signaling of lipopolysaccharide derived from the cell walls of the overgrowing Gram-negative species (Heimesaat et al., 2007a; Erridge et al., 2010). During the chronic phase of low-dose T. gondii infection (i.e., at days 36 and 57 p.i.), mice harbored fewer bifidobacteria but also lactobacilli in the small intestines as compared to uninfected controls. Particularly bifidobacteria and lactobacilli are considered "health-beneficial" commensals with immuno-modulatory properties and sharing important functions during immune homeostasis (Bereswill et al., 2017; Ekmekciu et al., 2017b,c; Heimesaat et al., 2017). Dysbiosis with deprived intestinal bifidobacterial loads has been shown to be associated with celiac disease, irritable bowel syndrome, inflammatory bowel disease (IBD) and atopic diseases, for instance (Tojo et al., 2014). Furthermore, acute ileitis was more pronounced in high-dose T. gondii infected conventional mice that were gene deficient for either TLR-9 or nucleotide-binding oligomerization domain (NOD) 2 and were both lacking intestinal bifidobacterial commensals (Bereswill et al., 2014; Heimesaat et al., 2014).

The murine low-dose T. gondii infection model has most commonly been used in order to investigate parasitic inflammation of the central nervous system given that upon intraperitoneal challenge with <10 T. gondii cysts of the ME49 strain susceptible mice develop chronic cerebral inflammation (Dunay et al., 2010; Biswas et al., 2015; Mohle et al., 2016; Lang et al., 2018; Dusedau et al., 2019). However, for investigating subacute/chronic intestinal and extra-intestinal (besides neurological) immunopathological changes the low-dose infection model has only been applied in single studies so far (Dunay et al., 2008; Escher et al., 2018; Heimesaat et al., 2018a).

In conclusion, the here provided long-term kinetic survey of immunopathological sequalae in intestinal, extra-intestinal and systemic compartments following peroral low-dose T. gondii infection provides valuable corner stones for a better understanding of the complex interactions within the triangle relationship of (parasitic) pathogens, the host immunity and the commensal gut microbiota during subacute and chronic intestinal inflammation. Furthermore, the low-dose T. gondii infection model may be applied as valuable gut inflammation model in future pre-clinical studies in order to test potential treatment options for intestinal inflammatory conditions in humans (Bereswill et al., 2019).

#### ETHICS STATEMENT

Following approval by the local authorities for animal experiments (für Gesundheit und Soziales, LaGeSo, Berlin, registration numbers G244/98), mouse experiments were performed in accordance with the European Guidelines for animal welfare (2010/63/EU). Clinical conditions of mice were assessed at least once daily.

# AUTHOR CONTRIBUTIONS

MH designed and performed experiments, analyzed data, wrote paper. ID critically discussed experimental design and results, co-edited paper. SB provided advice in experimental design, critically discussed results, co-edited paper.

#### FUNDING

This work was supported from the German Federal Ministries of Education and Research (BMBF) in frame of the zoonoses research consortium PAC-Campylobacter to MH and SB (IP7/ 01KI1725D). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript.

### ACKNOWLEDGMENTS

We thank Alexandra Bittroff-Leben, Ines Puschendorf, Ulrike Fiebiger, Anne Grunau, Ulrike Escher, Gernot Reifenberger, and the staff of the animal research facility at Charité - University Medicine Berlin for excellent technical assistance and animal breeding. We further thank Dr. Anja A. Kühl (Department of Medicine I for Gastroenterology, Infectious Diseases and Rheumatology/Research Center ImmunoSciences (RCIS), Charité – Universitätsmedizin Berlin) for taking representative photomicropgraphs of immunhistochmically stained paraffin sections. We acknowledge support from the German Research Foundation (DFG) and the Open Access Publication Fund of Charité – Universitätsmedizin Berlin.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00098/full#supplementary-material

Figure S1 | Representative photomicrographs illustrating inflammatory changes and distinct immune cell populations in the ileum over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.). Naive (N) mice served as uninfected controls. Representative photomicrographs taken from the ileum illustrate (A) histopathological changes (H&E staining) and (B) apoptotic (caspase3+, Casp3+) cells, (C) proliferating (Ki67+) cells, (D) T lymphocytes (CD3+), and (E) B lymphocytes (B220+) in immunohistochemically stained paraffin sections at respective time points (100 x magnification, scale bar 100µm).

Figure S2 | Representative photomicrographs illustrating inflammatory changes and distinct immune cell populations in the colon over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.). Naive (N) mice served as uninfected controls. Representative photomicrographs taken from the colon illustrate (A) histopathological changes (H&E staining) and (B) apoptotic (caspase3+, Casp3+) cells, (C) proliferating (Ki67+) cells, (D) T lymphocytes (CD3+), and (E) B lymphocytes (B220+) in immunohistochemically stained paraffin sections at respective time points (100 x magnification, scale bar 100µm). Figure S3 | Representative photomicrographs illustrating inflammatory changes and distinct immune cell populations in the liver over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.). Naive (N) mice served as uninfected controls. Representative photomicrographs taken from the liver illustrate (A) apoptotic (caspase3+, Casp3+) cells, (B) macrophages/monocytes (F4/80+), (C) T lymphocytes (CD3+), and (D) B lymphocytes (B220+) in immunohistochemically stained paraffin sections at respective time points (100 x magnification, scale bar 100µm).

Figure S4 | Representative photomicrographs illustrating inflammatory changes and distinct immune cell populations in the kidneys over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.). Naive (N) mice served as uninfected controls. Representative photomicrographs taken from the kidneys illustrate (A) apoptotic (caspase3+, Casp3+) cells, (B) T lymphocytes (CD3+), and (C) B lymphocytes (B220+) in immunohistochemically stained paraffin sections at respective time points (100 x magnification, scale bar 100µm).

Figure S5 | Representative photomicrographs illustrating inflammatory changes and distinct immune cell populations in the lungs over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii

on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.). Naive (N) mice served as uninfected controls. Representative photomicrographs taken from the lungs illustrate (A) apoptotic (caspase3+, Casp3+) cells, (B)

#### REFERENCES


macrophages/monocytes (F4/80+), (C) T lymphocytes (CD3+), and (D) B lymphocytes (B220+) in immunohistochemically stained paraffin sections at respective time points (100 x magnification, scale bar 100µm).

Figure S6 | Microscopic inflammatory sequelae in the heart muscle over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.; closed circles). Naive (N) mice served as uninfected controls (open circles). Out of six representative high power fields (HPF, 400x magnification) per animal the average numbers of (A) apoptotic (casapse3+, Casp3+) cells, of (B) T lymphocytes (CD3+) and (C) B lymphocytes were assessed in immunohistochemically stained cardiac paraffin sections. Medians (black bars), levels of significance (p-values) as determined by one-way ANOVA test followed by Tukey post-correction for multiple comparisons and numbers of analyzed mice (in parentheses) are indicated. Data were pooled from four independent experiments.

Figure S7 | Representative photomicrographs illustrating inflammatory changes and distinct immune cell populations in the heart over time following peroral low-dose T. gondii infection. Mice were perorally infected with one cyst of T. gondii on day 0 and surveyed at days (d) 10, 36, and 57 post-infection (p.i.). Naive (N) mice served as uninfected controls. Representative photomicrographs taken from the heart illustrate (A) apoptotic (caspase3+, Casp3+) cells, (B) T lymphocytes (CD3+), and (C) B lymphocytes (B220+) in immunohistochemically stained paraffin sections at respective time points (100 x magnification, scale bar 100µm).


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

Copyright © 2019 Heimesaat, Dunay and Bereswill. 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.

# Mechanisms of Human Innate Immune Evasion by *Toxoplasma gondii*

#### Tatiane S. Lima and Melissa B. Lodoen\*

*Department of Molecular Biology and Biochemistry and the Institute for Immunology, University of California, Irvine, Irvine, CA, United States*

*Toxoplasma gondii* is an intracellular protozoan parasite of global importance that can remarkably infect, survive, and replicate in nearly all mammalian cells. Notably, 110 years after its discovery, Toxoplasmosis is still a neglected parasitic infection. Although most human infections with *T. gondii* are mild or asymptomatic, *T. gondii* infection can result in life-threatening disease in immunocompromised individuals and in the developing fetus due to congenital infection, underscoring the role of the host immune system in controlling the parasite. Recent evidence indicates that *T. gondii* elicits a robust innate immune response during infection. Interestingly, however, *T. gondii* has evolved strategies to successfully bypass or manipulate the immune system and establish a life-long infection in infected hosts. In particular, *T. gondii* manipulates host immunity through the control of host gene transcription and dysregulation of signaling pathways that result in modulation of cell adhesion and migration, secretion of immunoregulatory cytokines, production of microbicidal molecules, and apoptosis. Many of these hostpathogen interactions are governed by parasite effector proteins secreted from the apical secretory organelles, including the rhoptries and dense granules. Here, we review recent findings on mechanisms by which *T. gondii* evades host innate immunity, with a focus on parasite evasion of the human innate immune system.

#### *Edited by:*

*Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France*

#### *Reviewed by:*

*Masahiro Yamamoto, Osaka University, Japan David Allan Christian, University of Pennsylvania, United States*

#### *\*Correspondence:*

*Melissa B. Lodoen mlodoen@uci.edu*

#### *Specialty section:*

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

*Received: 01 February 2019 Accepted: 25 March 2019 Published: 16 April 2019*

#### *Citation:*

*Lima TS and Lodoen MB (2019) Mechanisms of Human Innate Immune Evasion by Toxoplasma gondii. Front. Cell. Infect. Microbiol. 9:103. doi: 10.3389/fcimb.2019.00103* Keywords: *Toxoplasma gondii*, immune evasion, innate immunity, IFN-γ, pro-inflammatory cytokines, apoptosis

#### HOST INNATE IMMUNITY TO *T. GONDII* INFECTION

Toxoplasma gondii is an obligate intracellular protozoan parasite that infects an estimated onethird of the global human population. Although most infections are asymptomatic, T. gondii can cause life-threatening infections in immunocompromised individuals and the developing fetus (Montoya and Liesenfeld, 2004). During infection, T. gondii disseminates via the circulatory system and establishes chronic infection in several organs, including the heart and brain (Harker et al., 2015).

Although both humans and rodents are hosts for T. gondii, there are key differences in the innate immune responses between these species. In the mouse, innate immunity is mediated by TLR11 and TLR12 recognition of T. gondii profilin, which is the dominant mechanism driving IL-12 production (Yarovinsky et al., 2005; Koblansky et al., 2013). Notably, TLR11 is non-functional in humans, and TLR12 does not exist in the human genome, indicating alternative mechanisms of parasite sensing in human cells. Although these mechanisms have not been completely defined, it is Lima and Lodoen Immune Evasion by *T. gondii*

known that IL-12 is produced by human neutrophils and monocytes in response to T. gondii (Bliss et al., 1999; Aldebert et al., 2007). In human monocytes, unlike in mouse macrophages (Robben et al., 2004), phagocytosis of the parasite drives this IL-12 response (Tosh et al., 2016). IL-12 induces the production of IFN-γ, a key mediator of immunity in humans and mice (Suzuki et al., 1988; Ceravolo et al., 1999) that initiates protective type 1 immunity (Gazzinelli et al., 1994; Däaubener et al., 1995).

In addition to activating T cell-mediated immunity, IFN-γ functions in a cell-autonomous manner to control intracellular parasites. IFN-γ increases tryptophan degradation in human fibroblasts, inhibiting parasite replication (Pfefferkorn, 1984). More recently, IFN-γ inducible proteins were found on the parasitophorous vacuole membrane (PVM). Interestingly, these immunity-related GTPases (IRGs) play an important role in cellintrinsic antimicrobial defense in the mouse (Zhao et al., 2009), but this locus is considerably smaller in humans and does not appear to be involved in immune defense. A parallel IFN-γdependent mechanism of resistance in humans and mice consists of the guanylate binding proteins (GBPs), which are recruited to the parasitophorous vacuole membrane and cause vacuolar membrane disruption and parasite clearance (Yamamoto et al., 2012; Degrandi et al., 2013; Selleck et al., 2013). Human GBP1 restricts replication of type II T. gondii in epithelial cells without targeting the parasitophorous vacuole (Johnston et al., 2016), suggesting that GBPs can participate in host defense without causing classical vacuolar membrane disruption. In human cells, ubiquitination at the parasite vacuole has also emerged as a key mechanism of parasite control, leading to non-canonical autophagy and parasite growth stunting in HeLa cells (Selleck et al., 2015) or endolysosomal fusion and parasite clearance in umbilical vein endothelial cells (Clough et al., 2016).

As an obligate intracellular parasite, T. gondii has evolved strategies to successfully manipulate the host immune system to establish a productive infection and maintain an optimal replicative niche. Here we review recently described strategies by which T. gondii specifically evades human innate immunity, with brief mention of related studies in the mouse.

#### Modulation of Host Signaling Pathways

The manipulation of signaling pathways leading to cytokine production is an effective strategy to impair immune responses that compromise pathogen survival. Although T. gondii resides within a vacuole in infected cells, effector proteins released through the parasite's specialized secretory organelles, the rhoptries or dense granules, are instrumental in manipulating host cell signaling and transcriptional responses (**Figure 1**).

The three dominant clonal lineages of T. gondii (types I, II, III) notably differ in their effects on host cells. Type I and III, but not type II strains activate the signal transducer and activator of transcription 3 and 6 (STAT3 and STAT6) in human and mouse cells, thereby down-regulating IL-12 (Saeij et al., 2007). Similarly, in mouse macrophages, constitutive activation of STAT3 by type I strains prevents LPS-induced IL-12p40 production (Butcher et al., 2005). The rhoptry kinase ROP16 is responsible for these effects, by phosphorylating and activating STAT3 and STAT6 in human and mouse cells (Saeij et al., 2007; Yamamoto et al., 2009; Ong et al., 2010).

T. gondii infection induces a robust IFN-γ-driven immune response that is critical to resolve acute infection and control chronic infection (Suzuki et al., 1988, 1989). IFN-γ stimulation induces a vast transcriptional program (Platanias, 2005), and genome-wide microarray analysis in human foreskin fibroblasts (HFFs) revealed that T. gondii infection blocks up-regulation of all 127 genes that were induced by IFN-γ treatment in this study (Kim et al., 2007). Subsequent research determined that type I, II and III strains inhibit STAT1 transcriptional activity through mechanisms independent of ROP16 or GRA15, a dense granule protein that activates sustained NF-κB signaling (Rosowski et al., 2011; Rosowski and Saeij, 2012). IFN-γ stimulation initiates JAK/STAT signaling, whereby STAT1 homodimers translocate to the nucleus and bind to gamma-activated sequences (GAS) in DNA to activate transcription (Sadzak et al., 2008). Notably, T. gondii inhibits the expression of IFN-γ responsive genes by preventing the dissociation of STAT1 from DNA, hampering its recycling and further cycles of STAT1-mediated transcription (Rosowski et al., 2014). Recent studies identified a parasite factor conserved among T. gondii strains that is required for blocking transcription of IFN-stimulated genes in HFFs: T. gondii inhibitor of STAT1-dependent transcription (TgIST) is a dense granule protein that binds to activated STAT1 dimers in the nucleus of IFN-γ-treated cells and also to the chromatinmodifying Mi2/NuRD complex, resulting in altered chromatin and blockade of IFN-γ-dependent transcription (Gay et al., 2016; Olias et al., 2016). Ectopic expression of TgIST in human cells demonstrated that it is sufficient to repress STAT1-dependent promoter activity (Gay et al., 2016). Moreover, in IFN-γ-treated mouse macrophages, TgIST blocks IRG-mediated clearance of type II T. gondii (Gay et al., 2016).

Another major signaling cascade dysregulated by T. gondii is the NF-κB pathway, which leads to the production of proinflammatory cytokines involved in host immunity. In infected HFFs, type I T. gondii limits NF-κB activation by reducing p65/RelA phosphorylation and translocation to the nucleus (Shapira et al., 2005). Type I T. gondii also inhibits LPSinduced IL-1β production in primary human neutrophils, and this effect is associated with inhibition of NF-κB signaling. In T. gondii-infected neutrophils, IκBα degradation and p65/RelA phosphorylation are reduced, as are transcripts for IL-1β and the inflammasome sensor NLRP3. T. gondii also inhibits caspase-1 cleavage and activation in infected neutrophils (Lima et al., 2018), but not in infected human monocytes (Gov et al., 2013, 2017), representing different human cell type-specific mechanisms of IL-1β regulation.

Recently, GRA18 was identified as a dense granule protein that reprograms inflammatory responses. GRA18 forms complexes with regulatory elements of the βcatenin destruction complex, which includes β-catenin, GSK3α/β, and the PP2A-B56 holoenzyme, promoting stabilization and nuclear translocation of β-catenin, and inducing β-catenin-dependent gene expression (He et al., 2018). β-catenin is the main effector of the Wnt pathway, functioning as a coactivator of the T-cell factor/lymphoid

enhancer factor (TCF/LEF) transcription factors (Cadigan and Waterman, 2012). In murine macrophages, GRA18 induces β-catenin-dependent genes associated with antiinflammatory responses, including CCL17 and CCL22 (He et al., 2018), which may counterbalance type I inflammatory responses.

#### Inhibition of Apoptosis

Although cell death caused by infection can be detrimental to the host, apoptosis is also an important means of eliminating intracellular pathogens (Williams, 1994). Perhaps unsurprisingly, viruses, bacteria, and parasites have evolved strategies to inhibit this programmed cell death (Friedrich et al., 2017). Indeed, T. gondii can arrest both cell-intrinsic (mitochondrial) and extrinsic (death receptor-mediated) pathways of apoptosis (**Figure 2**) in the cells it has invaded. This may help the parasite to preserve its intracellular niche, replicate, and avoid clearance by humoral immunity.

The initial observations that T. gondii inhibits host cell apoptosis were in mouse cell lines (Nash et al., 1998); however, over the last 20 years, multiple studies have revealed these effects in human cell lines and primary cells. Collectively, these data show that both type I and II T. gondii inhibit the extrinsic and intrinsic apoptotic pathways through similar mechanisms. The first study on human cells demonstrated that T. gondii-infected HL-60-derived macrophages are protected from actinomycin D-induced apoptosis (Goebel et al., 1999). This effect on the mitochondrial apoptotic pathway is associated with inhibition of cytochrome c release, which in turn reduces cleavage of apoptotic caspase-9 and caspase-3. In addition, Mcl-1, an antiapoptotic factor from the Bcl-2 family is up-regulated by T. gondii infection (Goebel et al., 2001). T. gondii inhibition of

UV-induced apoptosis of infected HeLa cells is also associated with decreased cytochrome c release and apoptotic caspase activity (Carmen et al., 2006). This pathway is known to rely on c-Jun NH2-terminal kinase (JNK) signaling (Tournier et al., 2000), and indeed, JNK activity was repressed in T. gondiiinfected cells. Subsequent studies of staurosporine-treated HeLa cells and human Jurkat T cells provided evidence for how T. gondii impairs cytochrome c release. The oligomerization of the Bcl-2 pro-apoptotic proteins Bax and Bak permeabilizes the mitochondrial membrane, allowing the release of apoptogenic proteins, including cytochrome c (Jürgensmeier et al., 1998; Annis et al., 2005). Although T. gondii infection does not affect Bax or Bak expression, it inhibits conformational changes in these proteins, translocation of Bax from the cytosol to the mitochondria, and oligomerization of Bax, which contributes to decreased cytochrome c release (Hippe et al., 2009). Similarly, in arsenic trioxide-treated THP-1 macrophages, T. gondii increases expression of Bcl-2 and the anti-apoptotic chaperone heat-shock protein 70 (HSP70), which in turn reduces cytochrome c release and caspase-3 activation (Hwang et al., 2010). In staurosporinetreated cells, the mechanism is associated with induction of the serine protease inhibitors B3 and B4 (SERPIN B3/B4) via STAT6 activation (Song et al., 2012). In Jurkat cells, T. gondii inhibits apoptosis mediated by granzyme B, a death-inducing serine protease, by inhibiting granzyme B activity (Yamada et al., 2011).

The anti-apoptotic effects of T. gondii in diverse cell types appear to converge on inhibition of cytochrome c and apoptotic caspases. Interestingly, in a cell-free system with Jurkat cell extracts, the parasite can directly affect cytochrome cinduced caspase activation, independent of cytochrome c release from host cell mitochondria or upregulation of antiapoptotic molecules (Keller et al., 2006). Notably, parasite lysates mediated this effect, suggesting that a soluble parasite molecule specifically interferes with cytochrome c-induced caspase activation (Keller et al., 2006). Binding of cytochrome c and of dATP or ATP to the protease activating factor 1 (Apaf-1) allows the formation of a wheel-like heptameric complex, the apoptosome, which in turn activates caspase-9 (Reubold et al., 2009). Interestingly, T. gondii inhibits the binding of caspase-9 to Apaf-1, which prevents caspase-9 activity and subsequent caspase-7 and caspase-3 activation (Graumann et al., 2015).

In addition to blocking the intrinsic pathway, T. gondii also inhibits the extrinsic pathway of apoptosis. T. gondii prevents apoptosis in infected U937 monocytic cells treated with TNFα and cycloheximide (Goebel et al., 2001). Fas/CD95-induced apoptosis is blocked in the human B cell line SKW6.4 by T. gondii interference with the initiator caspase-8, in the absence of a mitochondrial amplification loop (Vutova et al., 2007). Reduced levels of pro-caspase-8 decrease its association with the death-inducing signaling complex (DISC) and impair activation of effector caspases (Vutova et al., 2007). In HeLa cells, in which Fas/CD95-ligation is amplified via the mitochondrial amplification loop, T. gondii inhibits cleavage of the proapoptotic BH3-only protein Bid, the release of cytochrome c, and the activity of the initiator caspase-8 and caspase-9 and the effector caspase-3 and caspase-7 (Hippe et al., 2008).

All of the previously noted human studies characterize the anti-apoptotic effect of T. gondii in human cell lines; however, more recently, this effect has been demonstrated in primary human cells. T. gondii infection of human peripheral blood mononuclear cell (PBMCs)-derived macrophages blocks staurosporine-induced apoptosis via increased expression of the miR-17-92 gene cluster (Cai et al., 2014). The promoter of this cluster contains two putative STAT3 binding sites, and T. gondii TgCtwh3 with atypical genotype China 1, activates STAT3, similar to type I T. gondii. STAT3 activation leads to increased miR-17-92 expression and decreased expression of Bim, a BH3-only member of the Bcl-2 family that contributes to pore formation in the mitochondrial membrane and cytochrome c release (O'Connor et al., 1998; Cai et al., 2014). The miRNA miR-20a is a member of the miR-17-92 gene cluster and its expression is up-regulated in human macrophages infected with type I T. gondii. Inhibition of this miRNA reverses the effect, resulting in apoptosis of human macrophages (Rezaei et al., 2018).

Glycosylphosphatidylinositols (GPIs) are glycolipids that link proteins to eukaryotic cell membranes. GPI anchors are abundantly expressed on many protozoan parasite surfaces, including T. gondii (Lekutis et al., 2001). Since exposing macrophages to Trypanosoma cruzi GPIs enhances expression of the anti-apoptotic A1 and Bcl-2-like genes (Ropert et al., 2002), a similar mechanism for T. gondii GPIs was investigated; however, highly purified T. gondii GPIs do not affect apoptosis of HL-60, Jurkat, or SKW6.4 cells (Debierre-Grockiego et al., 2007). Despite the many studies describing the anti-apoptotic effect of T. gondii in human cells, the parasite factor(s) that trigger this response remain unknown.

#### Evading Intracellular Death

In phagocytes, such as neutrophils and macrophages, ROS production is an important antimicrobial response for the elimination of pathogens. Interestingly, however, ROS is not induced in T. gondii-infected human macrophages (Wilson et al., 1980), potentially due to an immune evasion mechanism, as noted below. Non-phagocytic cells also generate low levels of ROS (Bedard and Krause, 2007), and in ARPE-19 cells, T. gondii targets the main NADPH oxidase by reducing Nox4 at the transcript and protein levels, resulting in decreased intracellular ROS. The effect on Nox4 expression was associated with activation of PI3K/AKT signaling in infected cells (Zhou et al., 2013). Proliferation of type I T. gondii in murine inflammatory macrophages was also associated with decreased ROS production (Shrestha et al., 2006). Recent studies in mouse macrophages showed that clearance of type III, but not type I, T. gondii relies on NADPH activity, increased ROS production, and induction of GBP5, suggesting that virulent strains may block ROS production, enabling parasite survival (Matta et al., 2018).

Microbicidal enzymes also contribute to destroying intracellular and extracellular pathogens. Neutrophil granule enzymes are secreted into the phagolysosome and released during NETosis. A Kazal family serine protease inhibitor, T. gondii protease inhibitor 1 (TgPI-1), is secreted by the dense granules and inhibits neutrophil elastase activity (Morris et al., 2002). It is known that both tachyzoites and bradyzoites are resistant to physiological concentrations of trypsin (Sharma and Dubey, 1981), which the parasite encounters in the intestine. TgPI-1 also inhibits trypsin and chymotrypsin, two proteolytic enzymes of the small intestine (Pszenny et al., 2000; Morris et al., 2002). Together, these data suggest a role for TgPI-1 in evading neutrophils and in protecting the parasite in the gut.

#### Establishment of a Replicative Niche

T. gondii also affects the cell cycle in infected human cells. In HFFs, the parasite induces progression from G0/G1 to S phase and an arrest toward G2/M (Molestina et al., 2008). This response is associated with sustained activation of extracellular signal-regulated kinase (ERK) signaling, which may act as a positive feedback to maintain HFFs in S phase (Molestina et al., 2008). Similar G2 arrest was observed in the human BeWo trophoblast cell line and in primary normal dermal human fibroblasts (NHDFs). T. gondii infection induces expression of the E3 ubiquitin-protein ligase UHRF1 and down-regulates the cyclin B1, which may cause the cell cycle arrest (Brunet et al., 2008). GRA16 is a dense granule protein that is exported from the PV into the cytoplasm and accumulates in the host nucleus. This protein binds to two host enzymes, the deubiquitinase HAUSP and the PP2A phosphatase, which regulate p53 and cell cycle, suggesting that GRA16 controls host cell arrest in G2/M phase (Bougdour et al., 2013). Modulation of the host cell cycle may influence how T. gondii controls its own replication and suggests a preference for proliferation in G2/M phase.

The first microarrays performed on T. gondii-infected cells revealed up-regulation of host genes involved in nutrient scavenging and metabolism, which the parasite requires for replication (Blader et al., 2001). Interestingly, the hypoxiainducible factor-1 alpha (HIF-1α) transcription factor, which is required for parasite replication, is stabilized and activated in T. gondii-infected HFFs (Spear et al., 2006). HIF-1 stabilization occurs because PHD2, a prolyl hydroxylase that targets HIF-1 for proteasomal degradation, is down-regulated during infection via activin-like receptor kinase signaling (Wiley et al., 2010).

Poly-adenosine-binding proteins (PABPs) are RNA binding proteins that bind to polyadenylated RNA and are involved in metabolic pathways of the mRNA, and their sub-cellular distribution changes in response to cellular stress (Gray et al., 2015). Nuclear granulation of PABPs is induced in T. gondiiinfected HFFs (Fischer et al., 2018), which may enable the parasite to influence the host cell transcriptome. Quantitative proteomic analysis of HFFs also indicates that T. gondii globally reprograms key metabolic pathways in the host cell, including glycolysis, lipid, and sterol metabolism, mitosis, apoptosis, and structural-protein expression (Nelson et al., 2008). Together, these processes may facilitate T. gondii establishment of its replicative niche.

#### CONCLUDING REMARKS

In the last decade, significant progress has been made in characterizing mechanisms of immune evasion by T. gondii. Rodents are a natural host for T. gondii and a relevant model for studying many aspects of parasite immunity.

#### REFERENCES


However, the extension of these studies to human cells, which differ in key innate immune pathways, will be critical for understanding determinants of human disease. Future work on the contribution of parasite effector proteins to host-pathogens interactions in both hematopoietic and non-hematopoietic human cells will be of particular interest, as will studies investigating the synergistic effects of these proteins or their role in establishing chronic infection, potentially by altering pathways in brain cells (Schlüter et al., 2001; Xiao et al., 2011; Mammari et al., 2014). Ultimately, elucidation of the molecular mechanisms governing human immune evasion by T. gondii may provide new insights into potential therapeutic targets that contribute to reduced disease and improved outcomes for human health.

#### AUTHOR CONTRIBUTIONS

TSL wrote the first draft of the manuscript. TSL and MBL edited and revised the manuscript. Both authors read and approved the submitted version.

#### FUNDING

MBL is supported by NIH R01AI120846.


chromatin repressors dampening STAT1-dependent gene regulation and IFNγ-mediated host defenses. J. Exp. Med. 213, 1779–1798. doi: 10.1084/jem. 20160340


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

Copyright © 2019 Lima and Lodoen. 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.

# *Toxoplasma gondii:* CD8 T Cells Cry for CD4 Help

#### Imtiaz A. Khan\*, SuJin Hwang and Magali Moretto

*Department Microbiology, Immunology and Tropical Medicine, The George Washington University, Washington, DC, United States*

*Toxoplasma gondii*, an apicomplexan parasite, is a pathogenic protozoan that can infect the central nervous system. In pregnant women, infection can result in congenital problems of the fetus, while in immunocompromised individual it can lead to severe neurological consequences. Although CD8 T cells play an important effector role in controlling the chronic infection, their maintenance is dependent on the critical help provided by CD4 T cells. In a recent study, we demonstrated that reactivation of the infection in chronically infected host is a consequence of CD8 T dysfunction caused by CD4 T cell exhaustion. Furthermore, treatment of chronically infected host with antigen-specific non-exhausted CD4 T cells can restore CD8 T cell functionality and prevent reactivation of the latent infection. The exhaustion status of CD4 T cells is mediated by the increased expression of the transcription factor BLIMP-1, and deletion of this molecule led to the restoration of CD4 T cell function, reversal of CD8 exhaustion and prevention of reactivation of the latent infection. In a recent study from our laboratory, we also observed an increased expression of miR146a levels by CD4 T cells from the chronically infected animals. Recent reports have demonstrated that microRNAs (especially miR146a) has a strong impact on the immune system *of T*. *gondii* infected host. Whether these molecules have any role in the BLIMP-1 up-regulation and dysfunctionality of these cells needs to be investigated.

#### *Edited by:*

*Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France*

#### *Reviewed by:*

*Bellisa Freitas Barbosa, Federal University of Uberlandia, Brazil Gaoqian Feng, Burnet Institute, Australia Ellen A. Robey, University of California, Berkeley, United States*

> *\*Correspondence: Imtiaz A. Khan imti56@gwu.edu*

#### *Specialty section:*

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

> *Received: 10 January 2019 Accepted: 15 April 2019 Published: 01 May 2019*

#### *Citation:*

*Khan IA, Hwang S and Moretto M (2019) Toxoplasma gondii: CD8 T Cells Cry for CD4 Help. Front. Cell. Infect. Microbiol. 9:136. doi: 10.3389/fcimb.2019.00136*

Keywords: toxoplasma, CD8T cells, CD4T cells, IL-12, IL-21, BLIMP-1, exhaustion

# INTRODUCTION

Toxoplasma gondii, an Apicomplexan with broad geographical representation, can cause severe infection of the central nervous system (Torgerson et al., 2014). Also, maternal infection during pregnancy can result in congenital infection with serious neurological and ocular complications (Peyron et al., 2011). In immune compromised individuals, reactivation of latent neurologic foci can result in encephalitis (TE) (Saadatnia and Golkar, 2012). T. gondii is an important foodborne pathogen found in many domesticated animals used for human consumption in United States (Hill and Dubey, 2013; Dubey et al., 2014) and human infection often develops after the ingestion or handling of undercooked or raw meat containing tissue cysts (Hoffmann et al., 2012). Alternatively, it can result from direct contact with cat, soiled cat litter or from the consumption of water or food contaminated by oocysts excreted in the faces of infected cats (Torrey and Stanley, 2013). Toxoplasmosis continues to be a significant public health problem worldwide and while more than a million people are believed to be infected with this parasite in the United States, about 3,000 individuals develop symptomatic disease annually. The cost of illness caused by T. gondii in the United States of America is estimated to be nearly 3 billion with an annual loss of 11,000 quality adjusted life years (QALY) worldwide (Batz et al., 2012). Recent studies have also linked T. gondii infection to mental illness like schizophrenia and suicidal tendencies in apparently asymptomatic Khan et al. CD4 Help for CD8 T Cells

individuals (Pedersen et al., 2012; Torrey et al., 2012; Bhadra et al., 2013a,b). Reactivation of latent toxoplasmosis can have severe consequences not only in individuals infected with human immunodeficiency virus, but also in patients who have undergone allogeneic hematopoietic stem cell transplant (HSCT) or received solid organ transplant (Tenter et al., 2000; Bhopale, 2003). Although toxoplasmosis presents most often as a localized central nervous system infection, severely immunocompromised patients like those receiving HSCT also exhibit disseminated infection involving lungs, heart, and liver (Tenter et al., 2000). In recent studies, it was reported that severe disseminated toxoplasmosis in patients undergoing HSCT and leading to Intensive care Unit admission had a poor prognosis, necessitating strategies aimed at preventing this fatal opportunistic infection (Schmidt et al., 2013; Voegele et al., 2013). As far as HIV infected population is concerned, despite combination antiretroviral therapy (cART) many patients continue to suffer from toxoplasmosis. Furthermore, even after full cART introduction, 65% of these patients died within the first year of diagnosis with TE (Mayor et al., 2011). Overall, T. gondii infection induces a strong immune response in the infected host that restricts the infection to latency (Jordan and Hunter, 2010). However, in case this immunity is compromised it can pose a severe risk to infected individuals and lead to reactivation of infection (Shearer et al., 1986).

### PROTECTIVE IMMUNE RESPONSES AGAINST *T. GONDII* INFECTION

Innate immune responses that include NK cells, neutrophils and dendritic cells are important for the resistance against the parasite (Yarovinsky, 2014). Original work revealed that antibodies transferred from infected to naïve hamsters provided little protection to the recipient (Frenkel, 1967). Also, in this study, intact or lysed spleen cells were transferred from infected to naïve animals which were subsequently challenged. It was observed that only intact cells could confer protective immunity to the recipient animals, emphasizing the role of cell mediated immunity in this response. A later study using a vaccine strain model of infection demonstrated that both CD4 and CD8 T cells are important for controlling the infection, even though CD8 played a more dominant role (Suzuki and Remington, 1988). Further studies, using an antibody depletion method, demonstrated a pivotal role for IFNγ, a major mediator of protective immunity against the disease (Suzuki et al., 1988). Shortly thereafter, Khan and colleagues developed for the first time, antigen-specific CD8 T cell clones capable of responding and killing T. gondii tachyzoites via cytotoxic activity in vitro (Suzuki et al., 1988; Khan et al., 1994). These reports suggest that two effector mechanisms, including IFNγ mediated activation of macrophages and cytotoxicity mediated by CD8 T cells, play a role in controlling T. gondii infection. Studies conducted years later suggest that while IFNγ plays a critical role in controlling T. gondii infection during the acute phase of infection (Suzuki et al., 1988; Gazzinelli et al., 1992), chronicity of the infection is contained by cytotoxic CD8 T cells (Suzuki et al., 2010). Subsequently, Suzuki's group reported that antigen-specific CD8 T cells are capable of removing cysts from immunodeficient animals infected with T. gondii (Sa et al., 2017). Moreover, the importance of CD8 T cells in the protection against T. gondii infection was also demonstrated by other laboratories. In one of these reports, CD8 CTL (cytotoxic T lymphocytes) generated against a vaccine strain of the parasite protected the animals against a lethal challenge with a virulent strain of the parasite (Gazzinelli et al., 1991). In another study conducted by Brown et al., the importance of CD8 T cells in controlling toxoplasma cyst burden was demonstrated (Brown et al., 1994). A number of other studies have confirmed the dominant role of CD8 T cells in long-term immunity against T. gondii infection, which is important to keep the chronic infection under control (Gazzinelli et al., 1992; Khan et al., 1994; Khan and Kasper, 1996). The mechanism of CD8 T cell mediated protection during the late stages of the acute infection can be attributed to their ability to produce IFNγ (Gazzinelli et al., 1992), which plays a pivotal role in immune-protection against T. gondii infection (Suzuki et al., 1988). However, during chronic toxoplasmosis, the perforin dependent cytotoxic ability of CD8 T cell population is involved in restricting the parasite to chronic state (Suzuki et al., 2010).

# CD8 T Cell Immunity During *T. gondii* Infection

Effector CD8 T cells are one of the important sources of IFNγ, which is responsible for controlling both the acute as well as the chronic phase of the infection (Suzuki and Remington, 1988; Gazzinelli et al., 1992; Khan et al., 1999). The importance of CD8 T cells in maintaining chronic toxoplasmosis was demonstrated by antibody depletion studies. In these experiments, treatment of chronically infected mice with anti-CD8 antibody led to the reactivation of the latent infection (Gazzinelli et al., 1992; Bhadra et al., 2011a). However, the biggest challenge in restricting the infection to the chronic phase is the fact that CD8 T cells need to be maintained in a functional state for a sustained period of time. Studies conducted in an TE model have demonstrated that during the later stages of the chronic infection, CD8 T cells exhibit a graded increase in the expression of inhibitory receptor PD-1 resulting in their dysfunctionality/exhaustion (Bhadra et al., 2011a). CD8 T-cell exhaustion has been reported in several chronic viral infections, like LCMV infection, and are characterized by persisting high levels of viremia (Mueller and Ahmed, 2009; Shin et al., 2009). Similar to these viral infections, blockade of the PD-1– PDL-1 pathway in mice carrying chronic toxoplasma infection reinvigorates the suboptimal CD8 T-cell response, resulting in the control of parasites reactivation and prevention of mortality (Bhadra et al., 2012). Interestingly, in continuation of these studies, it was observed that PD-1 is preferentially expressed by polyfunctional memory CD8 T cells, which leads to their loss of functionality and renders them susceptible to apoptosis (Bhadra et al., 2012). The selective dysfunctionality in the memory CD8 population could be an impediment for the development of a robust CD8 T cell response needed for the long-term protection against the infection. Another question that needs to be addressed is if apoptosis of the CD8 T cell population is dependent on the strain of parasites as previously reported (Nishikawa et al., 2007; Hippe et al., 2009). In that case, the role of different parasite strains in the induction of CD8 T cell dysfunctionality will need to be ascertained. However, as stated in our previous manuscript (Bhadra et al., 2011b) a major hurdle in investigating these questions thoroughly was the lack of information regarding T. gondii dominant CD8 epitopes. Nevertheless, the decapeptide HPGSVNEFDF (HF10) from the dense granule protein GRA6 as a naturally processed peptide recognized by CD8 T cells during T. gondii infection in BALB/c (H-2d) mice has been identified (Blanchard et al., 2008). This was subsequently followed by the discovery of two more H2-Ldrestricted epitopes, SPMNGGYYM and IPAAAGRFF, from the dense granule protein GRA4 and rhoptry protein ROP7 (Frickel et al., 2008). Subsequently, Wilson et al. identified a novel H-2Kbrestricted epitope, SVLAFRRL, derived from TGD057, a protein of unknown function (Wilson et al., 2010). However, as we have stated earlier (Bhadra et al., 2011b) with the discovery of as MHC class I tetramers, T-cell receptor transgenic mice and ovalbumin expressing transgenic parasites will enable the investigation of the effector mechanisms from various CD8 subsets and their correlation with immune protection during acute and chronic phases of the infection with much greater clarity.

# Role of CD4 T Cells in the Induction of CD8 T Cell Response

CD4 T cells are critical for the induction of primary CD8 T cell response (Bennett et al., 1997). CD8 T cell immunity generated in the absence of CD4 T cells cannot be maintained and respond poorly to secondary challenge (Bennett et al., 1997; Laidlaw et al., 2016). CD4 T cells help CD8 T cell response primarily by facilitating antigen-presentation and up-regulation of costimulatory molecules on the dendritic cells to optimal levels that induce a robust CD8 T cell response (Bennett et al., 1997; Schoenberger et al., 1998). In addition to their role in primary CD8 T cell immunity, CD4 T cells are also essential for the robust expansion of memory CD8 T cell population (Williams et al., 2006a). Helpless CD8 T cells upon re-stimulation undergo activation induced cell death and memory response is severely impeded (Janssen et al., 2005). CD4 T cells are a critical source of IL-2 which is important for CD8 T cell development (Williams et al., 2006b). Regulatory CD4 T cells (Treg) have been reported to modulate IL-2 exposure of effector CD8 T cells during the primary phase and are essential for generation of functional memory CD8 population (McNally et al., 2011). Similarly role of T follicular helper cells (Tfh) population, which are the primary source of IL-21 (Hale and Ahmed, 2015) in the maintenance of CD8 T cell functionality is well documented (Yi et al., 2009).

# HELPER ROLE OF CD4 T CELL IN THE MAINTENANCE OF CD8 FUNCTIONALITY

Toxoplasma gondii induces a strong CD4 T cell response that is a major source of IFNγ during both acute, as well as chronic infection (Gazzinelli et al., 1992; Liesenfeld et al., 1996), and is similar to what has been observed in other infections (Green et al., 2013). Earlier studies, including those from our laboratory, have demonstrated the importance of CD4 T cells for the maintenance of the CD8 T cell response against intracellular pathogens (Carvalho et al., 2002; Casciotti et al., 2002; Williams et al., 2006a). It is believed that CD8 T cells play an important effector role during T. gondii infection, while the CD4 T cell subset provides the essential help needed for their maintenance. It has been reported that depletion of both CD4 and CD8 T cell populations results in the reactivation of latent toxoplasmosis and, as a consequence, the susceptible animals succumb to TE (Casciotti et al., 2002). Similarly, the emergence of severe toxoplasmosis in patients infected with HIV is concomitant with a decline in CD4 T cell numbers (Shearer et al., 1986). Although depleted CD4 numbers in HIV patients lead to increased susceptibility to TE, most cases of toxoplasmosis in HIV patients occur during the late stage of HIV infection (advanced AIDS), when a deficiency in CD8 T cells is also evident (Shearer et al., 1986). Overall, the depleted CD4 population during the late stages of HIV infection compromises the CD8 T cell immunity against the chronic (latent) toxoplasmosis leading to reactivation of the infection.

As stated above, several studies conducted in our laboratory have demonstrated a severe CD8 T cell dysfunctionality during chronic T. gondii infection (Bhadra et al., 2011c, 2012; Gigley et al., 2012). In recent publications, we reported that the dysfunction observed in the CD8 T cell population from chronically infected mice is due to inadequate help caused by CD4 exhaustion (Hwang et al., 2016; Moretto et al., 2017). This was established by our findings which showed that transfer of non-exhausted antigen-specific CD4 T cells to chronically infected mice reversed the CD8 dysfunctionality and prevented the reactivation of the latent infection (Hwang et al., 2016). Comparably, a recent report emphasized that a durable CD4 T cell response is more efficient in promoting a robust CD8 T cell immunity against tumors than targeting the CD8 T cells directly (Melssen and Slingluff, 2017). It is believed that the "strong cytokine producing help" from CD4 T cells facilitates the maintenance of CD8 functionality (Bhadra et al., 2011a) (Phan-Lai et al., 2016). Moreover, understanding the mechanism(s) by which CD4 T cells ensures the maintenance of a robust CD8 T cell immunity will provide important insights into the development of new therapeutic strategies that will allow the maintenance of functional CD8 T cell memory, therefore preventing TE in immunosuppressed patients with latent T. gondii infection.

# BLIMP-1 Mediated CD4 T Cell Exhaustion During Chronic Toxoplasmosis

Exhausted CD4 T cell population expressed elevated levels of multiple inhibitory receptors concomitant with reduced functionality and up-regulation of BLIMP-1, a transcription factor. Although BLIMP-1 has been intrinsically linked with CD8 T cell exhaustion (Shin et al., 2009), studies from our laboratory have demonstrated that ablation of this transcription factor in CD4 T cells reverses their exhaustion status and allow

them to provide essential help to CD8 T cells. This help from recovered CD4 T cells restores the long-term functionality of the CD8 T cells and prevents the reactivation of the latent infection (Hwang et al., 2016). One of the important questions that need attention is why CD4 T cells during chronic toxoplasmosis become exhausted/dysfunctional? Also, the mechanism of BLIMP-1 mediated CD4 T cell exhaustion that leads to CD8 T cell dysfunctionality needs to be investigated.

# Role of microRNA in Immune Response to *T. gondii* Infection

Recently, studies have demonstrated a pivotal role for microRNAs (miRNAs) in controlling the differentiation as well as functionality of various immune cells (Escobar et al., 2014; Lin et al., 2014). miRNAs are crucial post-transcriptional regulators of hematopoietic cell fate decisions (Oliveto et al., 2017). miR146a and miR 125 have been reported to control the inflammatory response and the outcome of pathogenic infections (Lee et al., 2016). miRNAs were also reported to be the regulators of the host response to infection by apicomplexan parasites (Cai and Shen, 2017). Brain miRNAs changed in abundance in response to T. gondii infection (Hu et al., 2018). It has been reported that two immuno-modulatory miRNAs, miR146a and miR155, were co-induced in the brains of mice challenged with T. gondii in a strain specific manner (Cannella et al., 2014). We performed a real time PCR on the antigen specific CD4 T cells and found that (a) among the miRNAs tested, miRNA146a was significantly up regulated at week 7 post-infection (p.i.) (**Figure 1**) a time point at which increased BLIMP-1 expression peaks in these cells (Hwang et al., 2016). Conversely, levels of miRNA9, which has been shown to enhance IL-2 production (Thiele et al., 2012), were increased at week 2 p.i. but were reduced to background levels at week 7 p.i. Based on this information, the role of miR146a in CD4 T cell exhaustion/dysfunctionality clearly needs to be investigated further. As TRAF-6 has been reported to be the direct target of miR146a (Stickel et al., 2014), we suspect that the downregulation of this molecule by antigen-specific CD4 T cells during chronic toxoplasmosis may lead to increased BLIMP-1 expression. However, before drawing any conclusions, predicted targets of miR146a (www.targetscan.org), including but not limited to TRAF-6, IRAK-1, CD28, Rel, and TLR3 should be investigated even though our data suggests an important role for miR146a in BLIMP-1 mediated CD4 exhaustion. Whether this role is targeted via TRAF-6 or other molecules still need to be determined.

## ROLE OF CO-STIMULATORY MOLECULES IN MAINTAINING CD4 T CELL FUNCTIONALITY

It has been postulated that the interplay between positive signals from co-stimulatory molecules and negative inhibitory receptors play an important role in T cell activation, differentiation, effector function, and exhaustion (Chen and Flies, 2013). In this regard, it is important to report that, as stated in a recent review, a robust therapeutic response against infectious diseases or cancer requires not only releasing the brakes (blocking the inhibitory receptors), but also stepping on the gas (targeting the appropriate T cell co-stimulatory molecules) to promote the expansion and functionality of T cells (Linch et al., 2015). Recently, a very important report demonstrated that rescue of exhausted CD8 T cells by PD-1 blockade is exclusively dependent on the interaction between a single co-stimulatory molecule CD28 and its receptor B7 (Hui et al., 2017; Kamphorst et al., 2017). In these studies, it was reported that anti-PD-1 treatment during chronic LCMV infection was ineffective if CD28: B7 interaction was blocked. Thus, the efficacy of inhibitory receptor blockade in reversing T cell exhaustion depends on the upregulation of important co-stimulatory molecules. However, the involvement of co-stimulatory receptors in the restoration of CD4 functionality due to BLIMP1 ablation has not been studied either in infectious diseases or cancer. These are very important studies and data obtained from these experiments will enable the target of co-stimulatory molecules needed to reverse

important insights into the CD4 T cell exhaustion.

CD4 T cell dysfunctionality which, in case of Toxoplasmosis and other chronic infections, is important for the maintenance of CD8 T cell immune response. To identify co-stimulatory molecules involved in the rescue of CD4 T cells in the context of BLIMP-1 ablation, we examined a panel of 5 such receptors in CD4 T cells from chronically infected BLIMP-1 conditional

knockout mice and wild type littermate controls. 4-1BB, a member of the TNF-TNFR superfamily, was identified as one of the most important co-stimulatory molecules showing high levels of differential expression as compared to wild type littermates (Hwang et al., 2016). A small difference in the expression of another TNFα receptor family member, OX40, was also

noted in BLIMP-1 ablated mice while expression of other costimulatory molecules (CD27, ICOS, and CD40 L) did not show any significant difference (Hwang and Khan, manuscript in preparation). Treatment of chronically infected animals with 4- 1BB agonist antibody prevented host mortality, highlighting the critical importance of this co-stimulatory pathway in the reversal of CD4 T cells exhaustion (Khan et al., 2019). Significantly, treatment of infected animals with a 4-1BB agonist increased the ability of these animals to control parasite multiplication and also increased the ability of infected animals to release IL-21 (Khan et al., 2019), a cytokine produced predominantly by CD4 T cells (Suto et al., 2008) and known to be important for the maintenance of CD8 T cell functionality (Shin et al., 2009). These findings suggest the importance of co-stimulatory molecules in the maintenance of CD4 T cell functionality during a chronic infection, which can be compromised as result of increased BLIMP-1 expression. Obviously, the molecular mechanism(s) involved in BLIMP-1/4-1BB mediated CD4 T cell dysfunction during chronic toxoplasmosis that leads to CD8 exhaustion in chronically infection needs to be studied further.

#### CONCLUSIONS

Toxoplasma gondii infection induces a strong innate and adaptive immune response. While the innate immunity is important for controlling the early stages of the infection (Yarovinsky, 2014), the adaptive immunity is critical for restricting the parasite replication during the later stages (Gazzinelli et al., 1992). Amongst the adaptive immune subsets, CD8 T cells are the primary effector cells while CD4 T cells play an essential helper role to maintain long-term immunity (Casciotti et al., 2002). Notwithstanding, a robust CD8 T cell immunity induced during acute phase of infection, does not result in the total eradication of parasites and the pathogen persists in a chronic state (Bhadra et al., 2011c). Studies conducted in our laboratory have shown that during chronic toxoplasmosis CD8 T cells exhibit increased expression of inhibitory receptors,

#### REFERENCES


especially PD-1 that leads to their dysfunction/exhaustion (Bhadra et al., 2011c). In more recent studies, we have demonstrated that CD8 T cell dysfunction is a sequelae of CD4 T cell dysfunction mediated by increased expression of the transcription factor BLIMP-1 (Hwang et al., 2016) In very recent and studies from our laboratory, we observed an increased expression of microRNA146a by antigen-specific CD4 T cells from T. gondii (**Figure 1**), infected animals at the time point at which elevated BLIMP-1 expression in these cells is noted. It will be very interesting and important to determine if there is a BLIMP-1/miR146a axis responsible for CD4 exhaustion during chronic toxoplasmosis. With the use of conditional knock outs for this microRNA it will be essential to determine if the strategies to block miR146a could have a profound effect on BLIMP-1 mediated CD4 T cell functionality that should ensure the persistence of functional CD8 T cell immunity resulting in substantial decreased in chronic parasitic burden (**Figure 2**). The approach could prove highly beneficial for individuals carrying the infection and chances of reactivation would significantly decrease especially in immunocompromised subjects.

### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of the George Washington University Institutional Animal Care and Use Committee under Animal Use Protocol A052.

# AUTHOR CONTRIBUTIONS

SH conducted and analyzed the experiments. MM and IK wrote the manuscript.

# FUNDING

This work was supported by NIH grant AI33325 awarded to IK.

during rescue of exhausted CD8 T cells. J. Immunol. 187, 4421–4425. doi: 10.4049/jimmunol.1102319


dysfunction during chronic toxoplasmosis. J. Exp. Med. 213, 1799–1818. doi: 10.1084/jem.20151995


perforin-mediated activity of CD8<sup>+</sup> T cells. Am. J. Pathol. 176, 1607–1613. doi: 10.2353/ajpath.2010.090825


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

Copyright © 2019 Khan, Hwang and Moretto. 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.

# *Toxoplasma* Effector GRA15-Dependent Suppression of IFN-γ-Induced Antiparasitic Response in Human Neurons

Hironori Bando1,2, Youngae Lee1,2, Naoya Sakaguchi <sup>1</sup> , Ariel Pradipta<sup>1</sup> , Ryoma Sakamoto<sup>1</sup> , Shun Tanaka<sup>1</sup> , Ji Su Ma1,2, Miwa Sasai 1,2 and Masahiro Yamamoto1,2 \*

*<sup>1</sup> Department of Immunoparasitology, Research Institute for Microbial Diseases, Suita, Japan, <sup>2</sup> Laboratory of Immunoparasitology, WPI Immunology Frontier Research Center, Osaka University, Osaka, Japan*

#### *Edited by:*

*Nicolas Blanchard, Centre de Physiopathologie de Toulouse Purpan (INSERM), France*

#### *Reviewed by:*

*Yasuhiro Suzuki, University of Kentucky HealthCare, United States Sandra K. Halonen, Montana State University, United States*

#### *\*Correspondence:*

*Masahiro Yamamoto myamamoto@biken.osaka-u.ac.jp*

#### *Specialty section:*

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

> *Received: 31 January 2019 Accepted: 16 April 2019 Published: 01 May 2019*

#### *Citation:*

*Bando H, Lee Y, Sakaguchi N, Pradipta A, Sakamoto R, Tanaka S, Ma JS, Sasai M and Yamamoto M (2019) Toxoplasma Effector GRA15-Dependent Suppression of IFN-*γ*-Induced Antiparasitic Response in Human Neurons. Front. Cell. Infect. Microbiol. 9:140. doi: 10.3389/fcimb.2019.00140* *Toxoplasma gondii* is an important human and animal pathogen that causes life-threatening toxoplasmosis. The host immune system produces interferon-γ (IFN-γ) to inhibit *T. gondii* proliferation. IFN-γ-inducible indole-2,3-dioxygenase 1 (IDO1), which mediates tryptophan degradation, has a major role in anti-*T. gondii* immune responses in various human cells. In response to the host's immune system, *T. gondii* secretes many virulence molecules into the host cells to suppress IFN-γ-dependent antiparasitic immune responses. The GRA15-induced proparasitic mechanism for suppressing IDO1-dependent immune responses has previously been tested only in human hepatocyte and monocyte co-cultures. Thus, whether human cells other than hepatocytes contain this virulence mechanism remains unclear. Here, we show that the GRA15-dependent virulence mechanism for suppressing the IDO1-dependent anti-*T. gondii* response operates in human neuronal cell lines and primary human neurons. Analysis of various human cell lines revealed that IL-1β-induced iNOS-dependent reduction of IDO1 mRNA expression occurred in brain cell lines (A172; glioblastoma, IMR-32; neuroblastoma, and T98G; glioblastoma) and liver cell lines (Huh7 and HepG2), but not in other cell lines. Moreover, co-culturing type II *T. gondii*infected THP-1 human monocytes with the brain cell lines inhibited the IDO1-mediated anti-*T. gondii* response in a GRA15-dependent manner. These data suggest that a GRA15-dependent virulence mechanism antagonizes the IDO1-dependent host immune response in human brain cells.

Keywords: IFN-γ, IDO1, GRA15, human, neuron, toxoplasmosis, *Toxoplasma gondii*

# INTRODUCTION

Toxoplasma gondii is a widespread protozoan that can infect most warm-blooded vertebrates. Infection with T. gondii causes toxoplasmosis in humans and animals (Boothroyd, 2009; Dubey, 2010). Nearly one-third of the human population is estimated to be infected with T. gondii. Although T. gondii infections in healthy individuals remain mostly asymptomatic, immunocompromised individuals often experience damage to their liver, brain, eyes, and other organs, thus resulting in lethal toxoplasmosis (Weitberg et al., 1979; Frenkel and Remington, 1980). In addition, T. gondii infections potentially lead to congenital toxoplasmosis in fetuses and newborn children via their primarily infected pregnant mothers (Montoya and Remington, 2008). Furthermore, the World Health Organization (WHO) and the Food and Agriculture Organization (FAO) have recently established toxoplasmosis as a foodborne infection of global concern (FAO/WHO, 2014). Thus, T. gondii is a common and important zoonotic pathogen.

Interferon-γ (IFN-γ) and the subsequent induction of IFN-stimulated genes (ISGs) are essential in anti-T. gondii host immune responses. Among ISGs, IFN-γ-inducible GTPases, such as p65 guanylate-binding proteins (GBPs), and p47 immunity-related GTPases (IRGs), have been shown to be important for clearing T. gondii in mice (Yamamoto et al., 2009; Gazzinelli et al., 2014). In addition, inducible nitric oxide synthase (iNOS) plays an important role in suppressing T. gondii growth in mice (Scharton-Kersten et al., 1997). In human cells, IFN-γ-inducible indoleamine 2,3 dioxygenase 1 (IDO1), rather than IFN-γ-inducible GTPases, and iNOS, is reported to play a major role in inhibiting T. gondii growth by degrading tryptophan, which is an essential amino acid for intracellular parasitic growth (Pfefferkorn et al., 1986a,b) in many human cell types (Bando et al., 2018b).

When T. gondii infects host cells, various effector molecules are secreted from dense granules to resist the IFN-γ-induced antiparasitic host immune responses in the human cells (Hunter and Sibley, 2012). A T. gondii dense granule protein TgIST directly inhibits STAT1-mediated IDO1 expression (Rosowski et al., 2014; Olias et al., 2016; Bando et al., 2018b). In addition, we recently found that another T. gondii dense granule protein GRA15 indirectly inhibits IDO1-dependent anti-T. gondii responses in human hepatocytes co-cultured with monocytes (Bando et al., 2018a). In detail, T. gondiiinfected monocytes secrete interleukin-1β (IL-1β) in a GRA15 dependent manner. Subsequently, the secreted IL-1β mediates iNOS expression and nitric oxide (NO) production in IFNγ-stimulated hepatocytes. Because iNOS and NO are strong negative regulators of IDO1, NO reduces IDO1 mRNA, and protein levels (Nathan and Xie, 1994; Thomas et al., 1994). Thus, T. gondii can proliferate in co-cultures of monocytes and hepatocytes in a GRA15-dependent manner. Because the GRA15-dependent virulence mechanism relies on iNOS induction in human hepatocytes in response to IL-1β and IFN-γ, other human cell types that can induce iNOS in response to IL-1β may allow GRA15-dependent T. gondii proliferation. However, which cell types are sensitive to GRA15-dependent functions when co-cultured with human monocytes remains unclear.

In the present study, we found iNOS-dependent IDO1 degradation in human brain cell lines (A172, IMR-32, and T98G) and human primary neurons. We further showed that GRA15 effectors play key roles in proparasitic functions in human brain cells when co-cultured with type II T. gondii-infected monocytes. These data demonstrate that T. gondii uses a GRA15-dependent virulence mechanism to suppress the IDO1-dependent anti-T. gondii immune responses in brain cells and hepatocytes.

### RESULTS

# IL-1β Stimulation in Human Brain Cells Downregulates IDO1 mRNA Expression Levels

We previously showed that costimulating IFN-γ and IL-1β significantly reduced IDO1 mRNA expression in the human hepatocyte cell line, Huh7 (Bando et al., 2018a), but not in the leukocyte cell line, HAP1 (**Figure 1A**). To assess whether IL-1β-dependent reduction of IDO1 mRNA is specific to hepatocytes or occurs in other human cell types, we first tested the effect of IL-1β stimulation on downregulation of IFN-γ-induced IDO1 mRNA expression levels in various human cell types (**Figure 1B**). We confirmed that IL-1β stimulation in liver cell lines (Huh7 and HepG2) strongly suppressed the IFN-γ-induced IDO1 mRNA expression levels (**Figure 1B**). Interestingly, IL-1β stimulation in all tested brain cell lines (A172; glioblastoma, IMR-32; neuroblastoma, and T98G; glioblastoma) severely reduced IFN-γ-induced IDO1 mRNA expression levels in a manner similar to or greater than that of the liver cell lines (**Figure 1B**). In contrast, adding IL-1β had either no or a lesser suppressive effect on IFN-γ-induced IDO1 mRNA expression levels in colonic (HCT116, CCK-81), leukocytic (HAP1, THP-1), lung (A549, PC-3), breast (MCF7, YMB-1), pancreatic (MIA PaCa-2, KP-2), kidney (293T, KMRC-1), ovarian (RMG-I, RKN), placental (BeWo), splenic (OVTOKO), cervical (Ca Ski, HeLa), osteosarcoma (U2OS), foreskin fibroblastic (HFF), and retinoblastic (Y79, WERI-Rb-1) cell lines (**Figure 1B**). These results suggest that IL-1β regulation reduces IDO1 mRNA in human brain cells as well as hepatocytes.

### Costimulating IFN-γ and IL-1β in Human Brain Cells Induces iNOS mRNA and NO

NO induced by iNOS expression downregulates IDO1 expression transcriptionally, translationally, and posttranslationally (Alberati-Giani et al., 1997; Daubener et al., 1999). In addition, we confirmed that costimulating IFN-γ and IL-1β significantly induced iNOS mRNA expression and NO production in Huh7 (**Figure S1**) as previously reported (Bando et al., 2018a). We found that IFN-γ stimulation alone did not induce iNOS mRNA expression or NO production in A172 glioblastoma, IMR-32 neuroblastoma, or T98G glioblastoma human brain cell lines (**Figures 1C,D**). Conversely, stimulating both IL-1β and IFN-γ strongly induced iNOS mRNA expression and NO production in all brain cell lines tested (**Figures 1C,D**). These results suggest that costimulating IL-1β and IFN-γ leads to iNOS expression and NO production in human brain cells.

# IL-1β Stimulation in Human Brain Cells Downregulates IDO1-Dependent Anti-*T. gondii* Immune Responses

We previously showed that IFN-γ-induced IDO1 plays a major role in anti-T. gondii responses in various human cells (Bando et al., 2018b). To examine whether IL-1β-induced iNOS expression and NO production are involved in proparasitic functions in A172 glioblastoma, IMR-32 neuroblastoma, and

T98G glioblastoma cells, we compared the IFN-γ-mediated reduction of T. gondii numbers in the presence or absence of IL-1β (**Figure 2A**). IFN-γ alone strongly suppressed T. gondii numbers in all brain cell lines tested (**Figure 2A**). In contrast, costimulating IFN-γ and IL-1β impaired the IFN-γ-mediated anti-T. gondii responses in all brain cell lines tested (**Figure 2A**). Next, we examined IDO1 and iNOS protein levels in the presence or absence of IL-1β (**Figure 2B**). IFN-γ stimulation alone resulted in high IDO1 expression levels, while iNOS protein expression was undetected in A172 glioblastoma, IMR-32 neuroblastoma, and T98G glioblastoma cells (**Figure 2B**). In contrast, costimulating IFN-γ and IL-1β induced iNOS protein expression and conversely reduced IDO1 protein expression levels (**Figure 2B**). Next, to test whether iNOS expression is important for IL-1β-dependent IDO1 reduction, we assessed the effect of the selective iNOS inhibitor, aminoguanidine, on proparasitic functions (**Figure 2C**). T. gondii numbers were significantly reduced in the presence of aminoguanidine (**Figure 2C**) in the brain cell lines tested. In addition, aminoguanidine treatment inhibited NO production (**Figure S2**) and prevented IDO1 protein level reductions in all IFN-γ- and IL-1β-costimulated A172 glioblastoma, IMR-32 neuroblastoma, and T98G glioblastoma human brain cell lines (**Figure 2D**). These data suggest that iNOS expression and NO production are important for IL-1β-dependent proparasitic functions in human brain cells.

#### iNOS Is Critical for GRA15-Dependent Proparasitic Functions in Human Brain Cells Co-cultured With Monocytes

We previously showed that Toxoplasma effector GRA15 has a virulence function in co-cultures of the human monocyte cell line, THP-1, and the human hepatocyte cell line, Huh7 (Bando et al., 2018a). Next, we examined whether GRA15 is involved in proparasitic functions in human brain cell lines co-cultured with THP-1 cells. A172 glioblastoma, IMR-32 neuroblastoma, and T98G glioblastoma cell lines produced NO when cocultured with wild-type Pru T. gondii type II strain-infected THP-1 cells (**Figure 3A**). In contrast, the brain cell lines cocultured with GRA15-knockout (KO) Pru T. gondii-infected THP-1 cells did not produce NO (**Figure S3A**). We next assessed parasite numbers in wild-type or GRA15-KO Pru T. gondiiinfected THP-1 cells co-cultured with A172 glioblastoma, IMR-32 neuroblastoma, and T98G glioblastoma cell lines (**Figure 3A**). Parasite numbers in the GRA15-KO Pru T. gondii co-cultures with human brain cell lines and THP-1 cells were significantly lower than those that included wild-type Pru T. gondii-infected THP-1 (**Figure 3A**). Furthermore, iNOS protein expression and IDO1 protein reduction occurred in co-cultures with THP-1 cells infected with wild-type Pru T. gondii but not GRA15- KO T. gondii (**Figure 3B**). On the other hand, survival of wild-type T. gondii infecting neuronal cell lines alone was comparable to that of GRA15-KO T. gondii (**Figure S3B**). Next we examine whether iNOS inhibition restores GRA15 dependent reduction of parasite growth in wild-type T. gondii infected co-cultured conditions (**Figure 3C**). Aminoguanidine treatment significantly reduced numbers of wild-type T. gondii but not those of GRA15-KO parasites (**Figure 3C**). These results indicate that GRA15 is specifically required for iNOSdependent IDO1 reduction and parasitic growth in human brain cell line and THP-1 cell co-cultures in the presence of IFN-γ.

# Strain-Specific IL-1β-Induced IDO1 Suppression Facilitates *T. gondii* Survival

GRA15 is involved in sustained NF-κB activation in cells infected with the type II T. gondii strain, but not the type I or type III strains (Gov et al., 2013). In addition, GRA15's amino acid sequence varies among T. gondii strains (Rosowski et al., 2011; Gay et al., 2016). Therefore, we tested whether the GRA15 mediated virulence mechanism is strain-specific. When IL-1β production was tested in THP-1 cells, infection with the type II T. gondii strain, but not the type I or type III strains, induced IL-1β production in THP-1 cells (**Figure 4A**) as previously reported (Gov et al., 2013). In addition, NO was produced when THP-1 cells were infected with the type II strain, but not the type I or type III strains, when co-cultured with Huh7 cells (**Figure 4B**). Moreover, GRA15-dependent proparasitic functions were observed with the type II strain infection but not the type I or type III strain infections (**Figure 4C**). Furthermore, iNOS protein expression and IDO1 protein reduction occurred in co-cultures with THP-1 cells infected type II T. gondii strains, but not the type I or type III strains (**Figure 4D**) These results suggest that type II GRA15-mediated virulence mechanisms are strain-specific under co-culture conditions.

#### Type II GRA15-Dependent IDO1 Reduction by iNOS-Mediated NO in Co-cultures of Primary Human Monocytes and Neurons

To examine whether type II GRA15-dependent proparasitic functions exist in primary human neurons, we tested the IDO1 and iNOS mRNA expression levels (**Figure 5A**) as well as NO production (**Figure 5B**) in the presence or absence of IL-1β in primary human neurons. IFN-γ alone stimulated high IDO1 expression levels, whereas iNOS protein expression and NO production were undetected (**Figures 5A,B**). In contrast, IFNγ and IL-1β costimulation led to iNOS protein expression and NO production and conversely reduced IDO1 protein levels (**Figures 5A,B**). Next, to assess whether type II GRA15 dependent proparasitic functions exist in primary human neurons when co-cultured with primary human monocytes, we compared the number of wild-type or GRA15-KO Pru T. gondii in these co-cultures (**Figure 5C**). The number of GRA15- KO parasites in the co-cultures was significantly lower than the number of wild-type parasites (**Figure 5C**). In addition, iNOS expression, NO production and IDO1 reduction were observed in the co-cultures containing wild-type Pru T. gondii but not in those containing GRA15-KO parasites (**Figures 5D** and **Figure S3C**). These results indicate that type II Toxoplasma effector type II GRA15-dependent virulence mechanisms operate in co-cultures of human primary neurons and human primary monocytes and are required for counter defense against IFN-γinduced IDO1-dependent anti-T. gondii responses in humans.

# DISCUSSION

We previously showed that IL-1β stimulation inhibits IFN-γ-induced IDO1 mRNA expression in the hepatocyte cell line, Huh7, and primary human hepatocytes

FIGURE 2 | iNOS suppresses IDO1-dependent anti-*T. gondii* response in human brain cell lines. (A) Luciferase assay of the parasite survival rate at 24 h post infection in IFN-γ- and/or IL-1β-stimulated the brain cell lines (A172, IMR-32, or T98G). (B) Western blot analysis showing the expression of iNOS, IDO1, and β-Actin in the indicated cells after stimulation with IFN-γ and/or IL-1β for 24 h. (C) Luciferase assay of the parasite survival rate at 24 h post infection in the cells treated with the indicated cytokines and/or aminoguanidine. (D) Western blot analysis showing the expression of IDO1 and β-Actin in the presence of aminoguanidine. Each western blot image is representative of three independent experiments (B,D). Indicated values are means of ± s.d. (three biological replicates per group from three independent experiments) (A,C). \**p* < 0.05; \*\**p* < 0.01, (Student's *t*-test).

were infected with WT or GRA15-KO Pru *T. gondii*. The infected THP-1 cells were co-cultured with A172, IMR-32, or T98G cells and then left untreated or treated with IFN- and/or aminoguanidine. In the presence or absence of IFN-γ for 48 h. The parasite survival rates were measured by luciferase assay. Each western blot image is representative of three independent experiments (B). Indicated values are means of ± s.d. (three biological replicates per group from three independent experiments) (A,C). \**p* < 0.05; \*\**p* < 0.01, N.S., not significant; (Student's *t*-test).

(Bando et al., 2018a). In this study, we found that IL-1β also inhibits IDO1 mRNA expression in the A172 glioblastoma, IMR-32 neuroblastoma, and T98G glioblastoma human brain cell lines and in primary human neurons. Interestingly, the 5 cell lines, Huh7 hepatoma, HepG2 hepatoma, A172 glioblastoma, IMR-32 neuroblastoma, and T98G glioblastoma cell lines, which

are derived from human liver and brain tissue, were ranked in the top 5 for IL-1β-dependent IDO1 reduction, suggesting that IL-1β suppresses IDO1-dependent host immunity in the human brain and liver.

T. gondii infects its host by using migratory immune cells, such as neutrophils, dendritic cells (DCs), and monocytes, to spread throughout the body via a mechanism known as the Trojan horse mechanism (Coombes et al., 2013). This enables infected immune cells to make contact with various cells, tissues, and organs. Although T. gondii can potentially infect all nucleated cells, the parasite is often isolated from specific organs such as the liver and brain (Robert-Gangneux and Darde, 2012). Here, we found that the Toxoplasma effector GRA15-dependent virulence mechanism operates in human liver and brain cells. Our findings suggest that GRA15-dependent virulent mechanisms may define the liver and brain specificity. We previously reported that IDO1 plays a major role in anti-T. gondii responses in various human cells (Bando et al., 2018b), whereas IDO1-independent anti-T. gondii immune responses involve ATG16L1 and GBP1 in some human cell lines (Selleck et al., 2015; Clough et al., 2016). However, the GRA15 virulence mechanism requires secondarily infected cells to express iNOS in response to IFN-γ and IL-1β, thus reducing IDO1 mRNA, and protein expressions. Therefore, although an IDO1-dependent anti-T. gondii response is observed in various cells, the GRA15-dependent virulence mechanism targeting IDO1 operates in specific cells such as neurons and hepatocytes. Moreover, we have shown that the GRA15 dependent virulence mechanism to suppress IDO1-mediated anti-T. gondii immune response requires IL-1β production in the primarily infected monocytes and subsequent iNOS expression in

infected monocytes were co-cultured with primary human neurons in the presence or absence of IFN-γ for 48 h. (C) Luciferase assay of the parasite survival rate. (D) Western blot analysis showing the expression of iNOS, IDO1, and β-Actin. Each Western blot image is representative of three independent experiments (D). Indicated values are means of ± s.d. (three biological replicates per group from three independent experiments) (A–C). \**p* < 0.05; (Student's *t*-test).

the secondarily infected hepatocytes in the previous study (Bando et al., 2018a) or brain cells in this study. In the co-culture of monocytes and hepatocytes or neurons, IL-1β derived from T. gondii-infected monocytes in a manner dependent on GRA15 stimulates hepatocytic or neuronal production of iNOS and NO in the presence of IFN-γ, resulting in suppression of IDO1 mRNA and protein expressions. On the other hand, since there is no source of IL-1β, which is essential for neuronal induction of iNOS, in the single culture, the brain cells can robustly induce IDO1, and remain the expression levels, enabling suppression of T. gondii growth. Taken together, presence of IL-1β from T. gondii-infected monocytes characterizes the GRA15-dependent virulence mechanism that is observable in the co-culture system but not in the single culture of brain cells. Although we found a virulent role of GRA15 to suppress immune response in the coculture of human monocytes and neurons is found in this study, we failed to find such an immune-suppressive function in the single culture of neurons. Whether GRA15 possesses a unique function in neurons alone will be assessed in the future.

In the present study, we demonstrated that primary neurons exhibit the GRA15-dependent iNOS expression and IDO1 reduction in a manner similar to various neuronal cell lines that tend to be considered to be distinct from physiological neurons and therefore underestimated as fibroblasts with neuronal cell markers such as GFAP. Given neuronal cell lines are materially inexhaustible and can be genetically modified, they might be relatively more useful than primary neurons for further investigation of the detailed molecular mechanisms of GRA15 dependent T. gondii virulence in humans.

T. gondii is classified into three major clonal lineages known as types I, II, and III (Howe and Sibley, 1995a; Darde, 1996). Type I is a highly virulent strain in mice, while type II exerts intermediate virulence, and type III exerts low virulence. The current study demonstrates that type II T. gondii strain can survive more than types I or III parasites in co-culture of human neuronal cells and monocytes, suggesting that T. gondii virulence might be different between humans and mice, probably among host species. Notably, type II T. gondii is the most prevalent cause of both human congenital and acquired toxoplasmosis in North America and Europe (Howe and Sibley, 1995b; Ajzenberg et al., 2002, 2009). Our previous and current findings involving GRA15 can be one of the mechanisms that might account for T. gondii tropism in human brains and livers. However, the co-culture system in the previous and current studies have observed for only 48 h that is insufficient to assess bradizoite transformation and cyst formation. Obviously, further investigations to examine whether the presence of type II GRA15 affects bradyzoite transformation in the co-culture system may be of interest in the future.

In summary, we demonstrated that the T. gondii effector, GRA15, plays an important role in inhibiting IFN-γ-inducible IDO1-dependent anti-T. gondii responses in human brain and liver cells when co-cultured with human monocytes. Because GRA15-dependent virulence mechanisms are important for T. gondii infections in humans, further elucidation may contribute to developing advanced therapeutic strategies to treat human toxoplasmosis.

# MATERIALS AND METHODS

#### Cells and Parasites

All T. gondii strains (type I; RH strain, type II; Pru strain, type III; CTG strain) were maintained in Vero cells in RPMI (Nacalai Tesque) supplemented with 2% heat-inactivated FBS (JRH Bioscience), 100 U/mL penicillin/streptomycin (Nacalai Tesque), as previously described (Ma et al., 2014). GRA15-deficient Pru T. gondii was generated our previous study (Bando et al., 2018a). HAP1 (myelogenous leukemia) cells were maintained in IMDM (Nacalai Tesque) containing 10% heat-inactivated FBS, 100 U/mL penicillin/streptomycin. BeWo (choriocarcinoma), PC-3 (lung adeno carcinoma), HepG2 (hepatoma), OVTOKO (ovarian carcinoma), YMB-1 (breast carcinoma), KP-2 (tubular adenocarcinoma), THP-1 (acute monocytic leukemia), WERI-Rb-1 (Retinoblastoma), RKN (leiomyosarcoma), Ca Ski (epidermoid carcinoma), Y79 (Retinoblastoma), HFFs (fibroblast), and Huh7 (hepato cellular carcinoma) cells were maintained in RPMI (Nacalai Tesque) containing 10% heatinactivated FBS, 100 U/mL penicillin/streptomycin. A172 (glioblastoma), T98G (glioblastoma), IMR-32 (neuroblastoma), CCK-81 (adenocarcinoma), MCF-7 (breast adenocarcinoma), KMRC-1 (renal carcinoma), A549 (lung carcinoma), U2OS (bone osteosarcoma), RMG-I (Ovarian mesonephroid adenocarcinoma), MIA PaCa-2 (pancreatic cancer), and HeLa (cervix epitheloid carcinoma) cells were maintained in DMEM (Nacalai Tesque) containing 10% heat-inactivated FBS, 100 U/mL penicillin/streptomycin. Cryopreserved primary human monocytes (c-12909) were obtained from TAKARA. Cryopreserved primary human neurons (#1520) were obtained from ScienCell. According to the manufacture, the primary human neurons are cortical neurons isolated from embryonic tissue. The neurons are characterized by neurofilament, MAP2, and β-tubulin III. Primary cells were maintained in neuronal medium (ScienCell).

#### Reagents

Rabbit anti-IDO1 polyclonal antibody (13268-1-AP) was obtained from Proteintech. Mouse anti-iNOS monoclonal antibody (NOS2; sc-7271) was obtained from Santa Cruz Biotechnology. Mouse anti-β-Actin monoclonal antibody (A1978) was obtained from Sigma. Recombinant human IFN-γ and IL-1β were obtained from Peprotech. Aminoguanidine hydrochloride (396494) was obtained from Sigma. Purified antihuman IL-1β (511601) and biotin anti-human IL-1β antibodies (511703) were obtained from BioLegend.

### Quantitative RT-PCR

Total RNA was extracted, and cDNA was synthesized using the Verso Reverse transcription kit (Thermo Fisher Scientic). Quantitative RT-PCR was performed with a CFX connect realtime PCR system (Bio-Rad Laboratories) using the Go-Taq Real-Time PCR system (Promega). The values were normalized to the amount of glyceraldehyde 3-phosphate dehydrogenase (GAPDH) for human cells. Sequences of all primers are listed in **Table S1**.

### Western Blot Analysis

Cells were lysed in a lysis buffer (0.5% Nonidet P-40, 150 mM NaCl, and 20 mM Tris-HCl, pH 7.5) containing a cocktail of protease inhibitors (Roche). The cell lysates were separated by SDS-PAGE and transferred to polyvinylidene difluoride membranes (Immobilon-P, Millipore). Western blot analysis was performed using the indicated antibodies as described previously in detail (Yamamoto et al., 2009).

#### Luciferase Assay

Luciferase activities of total cell lysates were measured using standard protocols as described previously (Yamamoto et al., 2009). Cells were untreated or treated with 10 ng/mL IFN-γ and/or 20 ng/mL IL-1β before 24 h or at the same time of the luciferase-expressing T. gondii infection (MOI = 0.5–1). To measure the number of T. gondii, all infected cells were collected for the indicated periods and lysed by 100 µl of lysis buffer (Promega), followed by sonication. After centrifugation at 20,000 × g at 4◦C, the luciferase activity of the supernatant was measured using the Dual Luciferase Reporter Assay System (Promega) and a GLOMAX 20/20 luminometer (Promega). The percentages of the luciferase activities in cytokines-stimulated cells compared to those in unstimulated cells were shown as "Relative T. gondii numbers" in figures.

### Measurement of the Production of NO<sup>2</sup>

5 × 10<sup>5</sup> A172, T98G, Huh7 cells, or 1 × 10<sup>6</sup> IMR-32 or HAP1 were cultured in 12- or 24-well plates with 10 ng/mL IFN-γ and/or 10 ng/mL IL-1β for the indicated periods. The concentration of NO in the culture supernatant was measured using a NO2/NO<sup>3</sup> Assay Kit-FX (Dojindo).

## Stimulation of Cell Lines With IFN-γ and IL-1β

1 × 10<sup>6</sup> HAP1, PC-3, HepG2, YMB-1, THP-1, IMR-32, RMG-I, Ca Ski, or CCK-81, or 5 × 10<sup>5</sup> BeWo, OVTOKO, KP-2, WERI-Rb-1, RKN, HFFs, Huh7, A172, U2OS, T98G, MCF-7, HeLa, HCT116, KMRC-1, MIA PaCa-2, A549, Y79 cells were untreated or treated with 10 ng/mL IFN-γ and/or 20 ng/mL IL-1β for 24 h. Then, untreated or treated cells were used for quantitative RT-PCR, Western blot or parasite infection.

#### Infection of the Cell Lines With *T. gondii* in Single-Culture System

5 × 10<sup>5</sup> A172, T98G, or 1 × 10<sup>6</sup> IMR-32 were untreated or treated with 10 ng/mL IFN-γ and/or 20 ng/mL IL-1β for 24 h, then the luciferase-expressing T. gondii were infected (MOI = 0.5). At 24 h after parasite infection, the cell lysates and cell culture supernatants were collected and used for each experiment. Luciferase activities of total cell lysates in the single-culture condition were measured.

## Infection of the Cell Lines With *T. gondii* in Co-culture System

5 × 10<sup>5</sup> A172, T98G, 1 × 10<sup>6</sup> IMR-32, or 1 × 10<sup>5</sup> primary human neurons were cultured in 12- or 24-well plates for 24 h and washed twice with PBS before co-culture. 5 × 10<sup>4</sup> Primary human monocytes or 2.5 × 10<sup>5</sup> or 5 × 10<sup>5</sup> THP-1 cells were plated in 24-well plates and infected or uninfected with the luciferaseexpressing T. gondii (MOI = 1). At 24 h post infection, the cell culture supernatants containing either uninfected or infected primary human monocytes or THP-1 cells were added directly on top of the 24-well plates containing the human brain cell lines or primary human neurons in the presence or absence of 10 ng/mL IFN-γ. At 48 h after treatment with IFN-γ, the cell lysates, and cell culture supernatants were collected and used for each experiment. Luciferase activities of total cell lysates in the co-culture condition were measured.

#### Inhibitor Treatment

5 × 10<sup>5</sup> A172 or T98G and 1 × 10<sup>6</sup> IMR-32 were treated with 10 ng/mL IFN-γ and/or aminoguanidine hydrochloride (500µM) for 24 h and then infected or uninfected with T. gondii as described above.

#### Statistical Analysis

All statistical analyses were performed using Graphpad Prism 7 (GraphPad Software) or Excel (Microsoft). All the experimental data represent the average of three biological replicates (three independent experiments). The statistical significance of differences in mean values was analyzed by using an unpaired two-tailed Student's t-test. p < 0.05 were considered to be statistically significant.

#### AUTHOR CONTRIBUTIONS

HB and MY designed the study. HB, YL, NS, AP, RS, and ST performed the experiments. HB, JM, MS, and MY analyzed the data. HB, YL, and MY wrote the paper.

#### ACKNOWLEDGMENTS

We thank M. Enomoto (Osaka University) for secretarial and technical assistance. Supported by the Research Program on Emerging and Re-emerging Infectious Diseases (JP18fk0108047) and Japanese Initiative for Progress of Research on Infectious Diseases for global Epidemic (JP18fm0208018) from Agency for Medical Research and Development (AMED), Grant-in-Aid for Scientific Research on Innovative Areas (17K15677, 19K16628 and 19H00970) from Ministry of Education, Culture, Sports, Science and Technology, Cooperative Research Grant of the Institute for Enzyme Research, Joint Usage/Research Center, Tokushima University, Takeda Science Foundation, Ohyama Health Foundation, Heiwa Nakajima Foundation, the Cell Science Research Foundation, Mochida Memorial Foundation on Medical, and Pharmaceutical Research, Uehara Memorial Foundation, and Research Foundation for Microbial Diseases of Osaka University.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00140/full#supplementary-material

Figure S1 | IL-1β-induced iNOS expression and NO production in Huh7 cells. Huh7 cells were either untreated or treated with the indicated cytokines for 24 h. (left panel of the figure) Quantitative RT-PCR analysis of iNOS mRNA level in the cells. (right panel of the figure) Level of NO2 released into the culture supernatant was measured by ELISA. Indicated values are means of ± s.d. (three biological replicates per group from three independent experiments).

Figure S2 | The effect of aminoguanidine on NO production in human brain cell lines. A172, IMR-32, or T98G cells were either untreated or treated with the indicated cytokines in the presence of aminoguanidine for 24 h. Level of NO2 released into the culture supernatant was measured by ELISA. Indicated values are means of ± s.d. (three biological replicates per group from three independent experiments).

Figure S3 | NO production in co-culture conditions. (A) THP-1 cells were infected with WT or GRA15-KO Pru *T. gondii* for 24 h. The infected THP-1 cells were co-cultured with A172, IMR-32, or T98G cells in the presence or absence of IFN-γ for 48 h. Level of NO2 released into the culture supernatant was measured by ELISA. (B) A172, IMR-32, or T98G cells were left untreated or treated with IFN-γ for 24 h and then infected with wild-type or GRA15-KO Pru *T. gondii*. The parasite survival rates were measured by luciferase assay. (C) Primary human monocytes were infected with WT or GRA15-KO Pru *T. gondii* for 24 h. The infected monocytes were co-cultured with primary human neurons in the presence or absence of IFN-γ for 48 h. Level of NO2 released into the culture supernatant was measured by ELISA. Indicated values are means of ± s.d. (three biological replicates per group from three independent experiments) (A–C) <sup>∗</sup>*p* < 0.05; (Student's *t*-test).

Table S1 | List of primers used in this study. The information for primer names, restriction enzymes, sequences and the resulting plasmids is shown.

# REFERENCES


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

Copyright © 2019 Bando, Lee, Sakaguchi, Pradipta, Sakamoto, Tanaka, Ma, Sasai and Yamamoto. 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.

# Toxoplasma Hypervirulence in the Rat Model Parallels Human Infection and Is Modulated by the Toxo1 Locus

Corinne Loeuillet <sup>1</sup> , Anais Mondon<sup>1</sup> , Salima Kamche<sup>1</sup> , Véronique Curri <sup>2</sup> , Jean Boutonnat 2,3, Pierre Cavaillès <sup>1</sup> and Marie-France Cesbron-Delauw<sup>1</sup> \*

<sup>1</sup> BNI Team, Grenoble Alpes, CNRS, Grenoble INP, TIMC-IMAG, Grenoble, France, <sup>2</sup> Therex Team, Grenoble Alpes, CNRS, Grenoble INP, TIMC-IMAG, Grenoble, France, <sup>3</sup> Unit of Anatomopathology, Institute of Biology and Pathology, Grenoble Alpes Hospital, Grenoble, France

#### Edited by:

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### Reviewed by:

William Harold Witola, University of Illinois at Urbana-Champaign, United States Rima McLeod, University of Chicago, United States

#### \*Correspondence:

Marie-France Cesbron-Delauw marie-france.delauw@ univ-grenoble-alpes.fr

#### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

> Received: 31 January 2019 Accepted: 15 April 2019 Published: 01 May 2019

#### Citation:

Loeuillet C, Mondon A, Kamche S, Curri V, Boutonnat J, Cavaillès P and Cesbron-Delauw M-F (2019) Toxoplasma Hypervirulence in the Rat Model Parallels Human Infection and Is Modulated by the Toxo1 Locus. Front. Cell. Infect. Microbiol. 9:134. doi: 10.3389/fcimb.2019.00134 Toxoplasmosis is considered as an opportunistic parasitic disease. If post-natally acquired in children or adults, it may pass unnoticed, at least with strains of European origin. However, in the wild biotopes especially in South America, Toxoplasma gondii strains display a greater genetic diversity, which correlates to higher virulence for humans, particularly along the Amazon River and its tributaries. In French Guiana, several atypical strains have been associated with severe clinical forms: ocular toxoplasmosis and acute respiratory distress syndrome both of which can result in death. Among these, the GUY008-ABE strain was responsible for an epidemic of severe disseminated toxoplasmosis in Suriname, which led to the death of one immunocompetent individual. To better understand the mechanism underlying the hypervirulence of the GUY008-ABE strain, we have tested the rat model which compared to the mouse, better reflects the immune resistance of humans to Toxoplasma infection. Here we compare the outcome of toxoplasmosis in F344 rats infected either by the GUY008-ABE strain or the type II Prugniaud strain. We show that the GUY008-ABE strain displays a higher virulence phenotype leading to the death of all infected rats observed in this study. GUY008-ABE infection was characterized by an increase of the parasite load in several organs, especially the heart and lung, and was mainly associated with severe histological changes in lungs. Moreover, correlating with its hypervirulence trait, the GUY008-ABE strain was able to form cysts in the LEW rat model otherwise known to be refractory to infection by other Toxoplasma strains. Together, these results show that the rat is a discriminating experimental model to study Toxoplasma virulence factors relevant to the pathogenesis of human infection and that the degree of virulence is linked to the Toxo1 locus.

Keywords: Toxoplasma, virulence, GUY008-ABE, Toxo1, resistance, rat model

# INTRODUCTION

Toxoplasma gondii is one of the most widespread parasites in the world. It is an obligate intracellular parasite which can infect all warm-blooded animals including humans (Hill and Dubey, 2002). Infection is acquired by ingesting either tissue cysts (in undercooked meat) or oocysts excreted in the environment by felidae (cats in particular) (Hill and Dubey, 2002).

The T. gondii population structure is clonal with most of the isolates found in Europe and North America belonging to three major haplotypes (I, II, and III), and recently the newly described haplotype 12 which is present in North America (Khan et al., 2011). In these countries, the type II is responsible of the majority of human cases of toxoplasmosis. While the infection may lead to severe or life-threatening illness in immunocompromised patients (Porter and Sande, 1992) and in congenitally infected newborns (McAuley et al., 1994), it is typically asymptomatic in healthy individuals, present as dormant cysts in brain and heart muscle (Tenter et al., 2000).

However, the genomic diversity of T. gondii strains is far higher in other parts of the world, especially in tropical South America (Shwab et al., 2014; Lorenzi et al., 2016) and is associated with a higher pathogenicity in healthy individuals (Dubey et al., 2012; Carneiro et al., 2013). In the French Guiana, some strains like the GUY008-ABE, have provoked outbreaks with severe forms including acute respiratory distress syndrome, sometimes with a lethal outcome (Demar et al., 2012). The discovery of strains highly virulent for humans has enlarged the possibility of finding new virulence factors of T. gondii and the corresponding interacting-host pathways (Melo et al., 2013).

To explore complex virulence traits, we require animal models representative of human infection for experimentation. One recent study described the infection outcome of seven French Guiana Toxoplasma strains in CD1 mice and showed a correlation between the variability of the mouse chromosome 1a (chr1a) and mice lethality (Khan et al., 2014). However, while mice are among the most highly susceptible species to toxoplasmosis (Zenner et al., 1998), the rat model by its natural resistance to Toxoplasma infection is more relevant for human toxoplasmosis and opportunity's better model for understanding mechanisms controlling human infection (Dubey and Frenkel, 1998). Most rat strains develop asymptomatic toxoplasmosis with the persistence of cysts in brain and muscles. Other strains of rat like the Lewis (LEW) are refractory and efficiently prevent the parasite dissemination into the body. This innate resistance response to Toxoplasma infection is mainly controlled by a single locus called Toxo1 located on the rat chromosome 10 (Sergent et al., 2005; Cavaillès et al., 2006). In humans, the Toxo1 locus is associated with the susceptibility to congenital toxoplasmosis (Witola et al., 2011, 2014).

In this work, we examined the outcome of Toxoplasma infection in rats infected with the GUY008-ABE strain responsible for an epidemic episode in Suriname with severe disseminated, even lethal toxoplasmosis (Demar et al., 2007). By contrast to the type II Prugniaud strain, the GUY008-ABE strain displayed high virulence in infected F344 rats leading to the death of the animals. We also showed that in the LEW rat, the GUY008- ABE strain could establish a silent chronic infection, indicating that the Toxo1 locus may modulate virulence without blocking parasite dissemination.

# MATERIELS AND METHODS

#### Toxoplasma Strains and Culture

The highly virulent Guiana GUY008-ABE strain was characterized and compared to Prugniaud type II strain. All the parasite strains were maintained by serial passages in primary human foreskin fibroblasts (HFF) at 37◦C with 5% CO<sup>2</sup> in a humidified atmosphere. HFFs were grown in Dulbecco's modified Eagle medium (DMEM, GibcoBRL) supplemented with 10% FCS, 2 mM glutamine, 100 U/ml penicillin and 100 U/ml streptomycin (Invitrogen).

## Ethics Statement

Breeding and experimental procedures were carried out in accordance with national and international laws for laboratory animal welfare and experimentation (EEC Council Directive 2010/63/EU, September 2010). Experiments were performed under the supervision of C.L. (agreement 38 10 38) in the "Plateforme de Haute Technologie Animale (PHTA)" animal care facility (agreement C3851610006 delivered by the "Direction Départementale de la Protection des Populations") and were approved by the ethics committee of the PHTA and by the French government (APAFIS#7617-2016111710364203 v3).

### Rat Strains, Inoculation and Disease Monitoring

Lewis (LEW) and Fischer (F344) rats were purchased from Janvier and Charles Rivers laboratories, respectively. Congenic LEW.BN.c10-F strains were bred in PHTA platform. All these rats were housed under specific pathogen-free conditions.

To study the acute phase of infection, F344 rats were anesthetized with isofluorane (Abbot Laboratories) and subsequently inoculated intraperitoneally (i.p.) with 10<sup>7</sup> tachyzoites of the GUY008-ABE or the Prugniaud strains. At days 0, 4, and 14, rat's weights were checked and rat survival monitored. Then, to characterize parasite dissemination, animals were infected and organs were sampled at days 4 and 14. Heart, lungs, mesenteric lymph nodes, brain, liver and spleen were separated in two parts: one part was fixed in formol for subsequent histological study, the other was placed directly in liquid nitrogen and then preserved at −80◦C before DNA extraction and parasite load determination by quantitative PCR.

To look for influence of the Toxo1 locus on infection outcome, resistant LEW and susceptible congenic line LEW.BN.c10-F rats were i.p. inoculated with 10<sup>7</sup> tachyzoites of each strain (GUY008-ABE and Prugniaud). Then, 2 months post-infection, quantification of brain cysts was assessed as described previously (Aldebert et al., 2011; Cavailles et al., 2014).

### DNA Extraction and Parasite Load Determination

DNA was extracted from each organ sample (20 mg) by using the Purelink Genomic DNA mini kit (Invitrogen) and following the manufacturers' recommendations. A final elution of 50 µL for each sample was performed.

The parasite load in heart, lungs, brain, liver and spleen was estimated by quantitative PCR on a Step One Plus real time PCR system apparatus (Applied Biosystems).

Quantitative standard curve was obtained by 8 fold-dilutions ranging from 5,26 × 10<sup>5</sup> -5.26 × 10−<sup>2</sup> tachyzoïtes.

DNA samples were analyzed using 300 nM of primers (Eurogentec) targeting the 529 bp repeated sequence: 5′ - GCGTCGTCTCGTCTAGATCG-3′ for the forward and 5′ - AGGAGAGATATCAGGACTGTAG-3′ for the reverse, 5 µL of Fast SyBr-Green (Life Technologies) and 1 µL of 1/50 diluted DNA, qsp H20 up to 10 µL. Forty PCR cycles were: 95◦C, 20 s; 40 × (95◦C, 3 s; 62◦C, 30 s; 72◦C, 15 s), finally a melting curve was performed.

For each sample, the final value of the parasite load was given by the following equation: Parasite load = (Nx50xorgans' weight (mg))/(samples' weight (mg)), with N is the number of parasites obtained by the qPCR.

#### Tissue Histological Analysis

The brains, lungs, spleens, livers, hearts samples were collected, fixed in 10% buffered formalin and processed routinely for paraffin embedding and sectioning. Tissue sections with 4µm thickness (40µm distance between sections) of each organ were cut with microtome (Leica, RM2245) and mounted on slides for histopathological study. Tissue sections were stained with Haematoxylin and Eosin and they were observed under light microscope. These manipulations have been made on the histological platform of Jean Roget Institute, La Tronche.

#### Parasites Proliferation in Rat Peritoneal Macrophages

Rat resident peritoneal cells were obtained by injection of sterile PBS into the peritoneal cavity. Collected cells were centrifuged, resuspended in Serum-Free Medium (SFM, Life Technologies, Inc) and counted. Macrophages were obtained by adhering cells for 1 h at 37◦C and 5% CO2. After 1 h, non-adherent cells were removed by gentle washing with SFM and SFM medium supplemented with 20% of L929-conditioned media added. After 24 h, parasites were added to macrophages at a ratio of 3:1 for 1 h. After washing to eliminate extracellular parasites, cells were cultured for 40 h in the presence of [3H] uracil (5 µCi per well, Ci = 37 GBq) as previously described (Pfefferkorn and Pfefferkorn, 1977). Monolayers were washed three times in PBS, disrupted with 500 µl of lysis/scintillation solution (Optiphase Supermix, Perkin Elmer) and radioactivity measured by liquid scintillation counting using a Wallac MicroBeta TriLux (Perkin Elmer).

#### RESULTS

## The Infection of F344 Rats by the French Guianan Toxoplasma Strain GUY008-ABE Is Lethal

We first analyzed rat survival following intraperitoneal inoculation of F344 rats by the GUY008-ABE strain or the avirulent type II Prugniaud strain. As expected, F344 rats infected with the Prugniaud strain, did not present clinical signs and all animals survived until 2 months post-infection (**Figure 1**). By contrast, while no clinical signs were noticeable in rats infected by the GUY008-ABE strain during the first seven days-post infection, a rapid decline of their health status was

observed after 10 days with symptoms including respiratory distress, asthenia, emaciation and orbital hemorrhage. All the GUY008-ABE-infected F344 rats died within 22 days post-infection (**Figure 1**). These results demonstrate that the F344 rat model discriminates between virulent and avirulent Toxoplasma strains.

#### GUY008-ABE Infection Is Responsible for Rat Weight Reduction as Well as Organomegalies

We found that GUY008-ABE infection resulted in deaths of 75% of the rats after 15 days post-infection (**Figure 1**). In order to define more precisely the pathological processes leading to the animal death, the F344 rats were infected intraperitoneally and sacrificed at day 4 (D4) or day 14 (D14) post-infection. Between D4 and D14, un-infected (NI) or Prugniaud-infected rats gained 10% of their weight load (**Figure 2**, p < 0.05 for Prugniaud) whereas GUY008-ABE-infected rats lost 20% of their weight load (**Figure 2**, p < 0.0005).

Guianese toxoplasmosis is known in humans to be associated with respiratory distress, hepato- and spleno-megalies (Carme et al., 2002), we decided to study the weight evolution of liver, spleen, lung, and heart after infection by Prugniaud or GUY008-ABE strains (**Figure 3**). Because of cyst formation in the brain, we also analyzed the weight of this organ. Since the GUY008-ABE-infected rats presented a global weight loss, the weights of organs were normalized to the rat weight before infection. Compared to NI rats and independent of the parasite strain, weight increase was observed for all organs of infected rats at both days 4 and 14 (**Figures 3A,B**, respectively) after infection. Additionally, inspection of GUY008- ABE-infected organs at day 14 showed a highly significant weight increase for the lungs compared to Prugniaud infected lungs. This lung swelling is therefore a signature of the GUY008- ABE hypervirulence.

\*\*\*\*p < 0.0005).

#### The Parasite Load Is More Important in the GUY008-ABE-Infected Rat and Is Associated With Severe Histological Changes

We further compared the parasite load at days 4 and 14 after infection. Parasites were detected within all the tested organs of both Prugniaud- and GUY008-ABE-infected rats (**Figure 4**). At day 4 post-infection, significant difference was noticed between GUY008-ABE and Prugniaud infections for the lungs only (**Figure 4A**). At day 14 post-infection, a high increase of the parasite load was observed in both the hearts (2,419 times increased by comparison to day 4) and lungs (117 times increased by comparison to day 4), specifically in GUY008-ABE-infected

rats. The higher parasite load in heart and lungs is thus associated with the strain virulence.

Tissues observations were realized at days 4 and 14 postinfection in both infected rats (**Figure 5**). No tissue perturbations were observed at day 4 post-infection when comparison was done regarding the tissue of NI rats and whatever the parasite strain used (GUY008-ABE or Prugniaud) for rat infection (not shown). At day 14 post-infection, comparison to the NI rat (**Figures 5A–H**) revealed that for the Prugniaudinfected rats, there was no abnormality in heart (**Figures 5I,J**), mesenteric lymph nodes, lung (**Figures 5K,L**) or spleen. In the liver, a slight inflammation was observed surrounding the portal vein sinus (**Figure 5N**). Parasites were found in the brain macrophages (**Figure 5P**) and a perivascular lymphocyte infiltration was also observed. For the GUY008-ABE-infected rats, no histological change was detected in spleen, thymus, or mesenteric lymph nodes. However, damage could be seen in other organs. In the liver, we observed a reaction of lymphohistiocytic vessels in response to infection (**Figure 5V**) and the presence of parasite granuloma. In the heart, as in the brain, focal inflammatory reactions were detected as well as parasite granuloma (**Figures 5R,X**). The lungs are also greatly affected with a histiocytic inflammatory reaction and the presence of parasite foci (**Figures 5S,T**). These latter observations (inflammation and presence of parasites in both lungs and heart) highly correlated with the respiratory distress observed in GUY008-ABE-infected animals.

#### Influence of the Toxo1 Locus on Guianese Toxoplasmosis

In order to analyze the involvement of the Toxo1 locus in the control of Guianese toxoplasmosis, resistant LEW and susceptible LEW congenic rats with Toxo1 from BN origin (LEW.BNc10-F) were intraperitoneally infected with the GUY008-ABE strain (**Figure 6**). As expected, no cyst was found in the brain of LEW rats infected with Prugniaud strain while cyst burden was found in the LEW.BNc10-F (Cavaillès et al., 2006). By contrast, GUY008-ABE-infected LEW rats displayed brain cysts (mean = 45) indicating that GUY008-ABE parasites were

able to bypass Toxo1-mediated refractoriness. Moreover, in the susceptible LEW.BN.c10-F rats, the number of brain cysts was significantly higher (an average of 482 cysts per brain). These results demonstrated that the hypervirulence of GUY008-ABE strain is associated with its capacity to modulate the Toxo1 mediated resistance and conversely, that the Toxo1 locus controls GUY008-ABE strain hypervirulence, in negatively impacting cyst burden.

# Parasite Proliferation in Peritoneal Macrophages

The Toxo1-mediated in vivo resistance has been correlated to the in vitro inhibition of parasite proliferation in LEW peritoneal macrophages and to both macrophages and parasite death (Cavailles et al., 2014). Here, we have examined the capacity of GUY008-ABE parasites to proliferate either in resistant LEW or susceptible LEW.BNc10-F peritoneal macrophages by comparison to Prugniaud strain (**Figure 7**). We showed that Prugniaud and GUY008-ABE strains do not proliferate in LEW macrophages (**Figure 7A**) and this correlated with the death of infected cells (**Figure 7B**). These results reveal that for the GUY008-ABE strain, there is a discrepancy between the observed in vitro and in vivo phenotypes.

# DISCUSSION

Successful pathogenicity depends on both host and parasite genotypes and environmental factors which can be controlled in experimental animal models. Using the F344 rat model of Toxoplasma infection, we showed here that the atypical GUY008- ABE strain, by contrast to the Prugniaud strain, exhibits a highly virulent phenotype in leading to the death of animals. In this model, the symptoms and pathophysiology of infection parallel that found in patients infected by the GUY008-ABE strain (Demar et al., 2007). We also demonstrated that the virulence of this strain is dependent on the genetic background of the rat. Specifically, we showed that Toxo1, a locus known to control the outcome of toxoplasmosis in rats infected with archetypal strains (Cavaillès et al., 2006), is involved in the modulation of the GUY008-ABE virulence.

The clinical signs described for the Guianese toxoplasmosis are pleiotropic including patients presenting pneumonia, cardiac or ophthalmological abnormalities. One patient was found to suffer of one organ failure while another died due to multiple organ failure (Demar et al., 2007, 2012). During the acute infection with GUY008-ABE parasites, F344 rats displayed signs of major general decrease in health. At day 14 post-infection, they were emaciated and underweight (20% loss). All rats also displayed orbital hemorrhage. Analysis of organ weights revealed, as in patients, hepato- and spleno-megaly. However, since both of these signs were also found in Prugniaud-infected F344 rats, these are likely more associated with Toxoplasma infection rather than to strain virulence. Only the lung weight was significantly higher in the GUY008-ABE-infected rats. Lung involvement was observed for all patients, with unilateral or bilateral crackles, unilateral bronchial breathing or areas of dullness (Demar et al., 2012). Thus, the lung weight increase is clearly related specifically to GUY008-ABE virulence.

Another major observation was the very high parasite load found in some organs. We have analyzed the parasite burden in organs known to be infected in the rat model (Zenner et al., 1998). In this work, the RH acute infection was monitored in F344 rats and low parasite numbers were found in the spleen and the lung at day 4 post-infection, and in the brain only at day 16 post-infection. Parasites were never detected in the heart and liver during the course of infection. On the contrary, here, we observed parasites in all tested organs and whatever the strain used. For the Prugniaud-infected rats, organ parasite loads were smaller than those observed with the GUY008-ABE strain and were almost stable between day 4 and day 14 postinfection. Results were different for the GUY008-ABE infection, organ parasite loads increased between day 4 and day 14 postinfection with a profile more closely aligned to the profile of parasite dissemination found in RH i.p. inoculated mice (Zenner et al., 1998). The RH strain is extremely virulent in mice as one parasite allows animal death within 10 days of infection (Howe and Sibley, 1995). At day 4 post-infection, mice present high parasite burden in all organs (10<sup>5</sup> for the brain, 10<sup>6</sup> for the heart,

10<sup>7</sup> for the liver and 10<sup>8</sup> for the spleen and the lungs) (Zenner et al., 1998). These numbers are close to those we found in the GUY008-ABE-infected rats especially for the heart and the lungs (10<sup>5</sup> and 10<sup>6</sup> parasites, respectively). Thus, the parasite loads found in the organs of GUY008-ABE-infected F344 rats were similar to that found in RH-infected OF-1 mice, and is likely to be associated with parasite hypervirulence.

The Guianese toxoplasmosis was associated with multiple organs failure. In the GUY008-ABE-infected F344 rats, lungs were found to be the most disrupted organ with histiocytic inflammation, alveoli full of serous fluid and presence of parasites foci. On the contrary, no tissue disruption was observed for the Prugniaud-infected rats. It has already been shown that the lung architecture can be modified by Toxoplasma infection in the rat (Foulet et al., 1994). Indeed, plurifocal fibrin alveolitis or acute bronchiolitis were observed in lungs of F344 nude rats after RH or Prugniaud infection, respectively (Foulet et al., 1994). Here, the GUY008-ABE acute infection was performed in the immunocompetent rat F344. Therefore, despite an active immune system, atypical strain led to similar lung failure, as in the case of RH infection of immunocompromised nude rat. Altogether, in GUY008-ABE-infected F344 rats, the observed physiopathology (loss of rat weight, lungs weight increase associated with high parasite burden and tissue destruction) suggests a severe respiratory failure leading to the death of animals and is fully concordant with the pathology observed in Guianese patients.

The severity of toxoplasmosis observed in the F344 rats infected with the atypical hypervirulent GUY008-ABE (LD100 = 10<sup>7</sup> ) provides evidence that the lethal effect is due to the parasite genotype. Indeed the same route and inoculum of infection by Prugniaud parasites did not cause F344 rat mortality nor acute clinical signs. Up until now, Toxoplasma virulence and its association with parasite genotype have been widely investigated in the mouse model using the archetypal I, II and III clonal lineages (Saeij et al., 2006; Taylor et al., 2006). Their virulence in mice differs substantially, with the type I

strain being acutely virulent (LD100: 1 to 10 viable organisms), type II strains exhibiting intermediate virulence (LD50 >10<sup>3</sup> in inbred mice and >10<sup>5</sup> in outbred mice), and type III being avirulent (Howe and Sibley, 1995; Sibley and Ajioka, 2008). However, compared to humans, mice are naturally more sensitive to Toxoplasma infection and recent studies have shown that they employ distinct innate pathways to control the infection (Gazzinelli et al., 2014). By contrast, in the F344 rats, the clinical course of infection between rats and humans is similar, with infection by the Prugniaud strain (type II) leading to an asymptomatic chronic infection (Darcy and Zenner, 1993; Zenner et al., 1998). Moreover, infection with the RH strain (type I) is totally controlled (not lethal) even with high inoculum (>10<sup>8</sup> parasites) (unpublished data), supporting the notion that divergent mechanisms of resistance between rat and mice are at work (Gazzinelli et al., 2014). Hence, surprisingly, when inoculated in CD1 mice, the GUY008-ABE strain was responsible for only 83% of animal death at 30 days post-infection (Khan et al., 2014). However, the lack of any defined LD50 or LD100 in this study hampers valuable comparison.

In the rat model, the outcome of toxoplasmosis is directed by the Toxo1 locus (Cavaillès et al., 2006; Cavailles et al., 2014). Indeed, all described refractory rat strains bear a highly conserved Toxo1-LEW haplotype while susceptible rats like the F344 or the BN strains bear divergent Toxo1 haplotypes. Rats harboring the Toxo1-LEW haplotype block do not develop anti-Toxoplasma serology nor brain cysts when infected with Prugniaud or RH strains (Cavailles et al., 2014). Our data revealed that, unexpectedly, LEW rats infected by GUY008-ABE parasites developed positive serologies (not shown) and brain cysts. To our knowledge, this is the first time that a parasite strain is described as being able to bypass the LEW refractoriness in vivo, demonstrating that the Toxo1-control of Toxoplasma infectivity is dependent on the parasite genotype. However, even if Toxo1 is not able to prevent GUY008-ABE infection, it modulated the parasite burden, since 20-fold reduction in the number of cysts was found in the brain of LEW rats compared to the susceptible

LEW.BN-c10-F rats (BN Toxo1-haplotype). In rats of the Toxo1- LEW haplotype, the lack of serological response together with the undetectable local parasite burden (Sergent et al., 2005) indicated that parasite elimination is very efficient and occurs at the level of natural barriers. Moreover, using RH type I parasites, it has been shown that Toxo1 controls very rapidly the local spreading of the parasite after i.p. infection (Cavaillès et al., 2006). It seems therefore that depending on the parasite strain, Toxo1 could modulate the parasite infectivity and spreading.

Up to now, the in vivo Toxo1-mediated resistance has been strictly correlated to both the inhibition of parasite proliferation in vitro and the death induction of both macrophage and parasites (Cavailles et al., 2014). Similar observation was described in LEW bone marrow-derived macrophages (Cirelli et al., 2014). As the GUY008-ABE strain is able to bypass the LEW resistance in vivo, we expected to observe in vitro parasite proliferation and decrease of cell death in the peritoneal LEW macrophages. Since this was not observed, it appears that depending of the parasite genotype, the inhibition of parasite proliferation within the macrophages in vitro, does not strictly correlate with the rat resistance to infection in vivo. This suggests that the Toxo1 resistance is a complex trait, which may involve other cells and mechanisms of innate immunity.

In conclusion, we have established that the F344 is a pertinent model in reproducing the pathophysiology of infected patients by hypervirulent T. gondii strains. We also provided evidence that virulence of Toxoplasma strains is highly modulated by the Toxo1 haplotype. Therefore, the rat model opens new avenues to discover new host-parasite interacting genes involved in the virulence of Toxoplasma gondii.

#### REFERENCES


#### ETHICS STATEMENT

This study was carried out in accordance with the recommendations of Breeding and experimental procedures were carried out in accordance with national and international laws for laboratory animal welfare and experimentation (EEC Council Directive 2010/63/EU, September 2010). Experiments were performed under the supervision of C.L. (agreement 38 10 38) in the Plateforme de Haute Technologie Animale (PHTA) animal care facility (agreement C3851610006 delivered by the Direction Départementale de la Protection des Populations) and were approved by the ethics committee of the PHTA (ComEth) and by the French government (APAFIS#7617-2016111710364203 v3).

#### AUTHOR CONTRIBUTIONS

CL designed and performed experiments, wrote article. AM and SK performed in vivo experiments. VC did the histological part. JB did the histological analysis. PC wrote article. M-FC-D designed experiments and wrote article.

#### ACKNOWLEDGMENTS

This work was supported by CNRS (Center National de la Recherche Scientifique), UGA (Université Grenoble, Alpes), Fondation pour la Recherche Médicale- and Labex Parafrap-ANR-11-LABX-0024. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.


strains of Toxoplasma gondii in French Guiana correlates with a monomorphic version of chromosome1a. PLoS Negl. Trop. Dis. 8:e3182. doi: 10.1371/journal.pntd.0003182


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

Copyright © 2019 Loeuillet, Mondon, Kamche, Curri, Boutonnat, Cavaillès and Cesbron-Delauw. 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.

# Interplay Between *Toxoplasma gondii*, Autophagy, and Autophagy Proteins

Carlos S. Subauste1,2 \*

*<sup>1</sup> Division of Infectious Diseases and HIV Medicine, Department of Medicine, Case Western Reserve University, Cleveland, OH, United States, <sup>2</sup> Department of Pathology, Case Western Reserve University, Cleveland, OH, United States*

Survival of *Toxoplasma gondii* within host cells depends on its ability of reside in a vacuole that avoids lysosomal degradation and enables parasite replication. The interplay between immune-mediated responses that lead to either autophagy-driven lysosomal degradation or disruption of the vacuole and the strategies used by the parasite to avoid these responses are major determinants of the outcome of infection. This article provides an overview of this interplay with an emphasis on autophagy.

Keywords: autophagy, parasite, CD40, IFN-gamma, cell signaling

#### *Edited by:*

*Jeroen P. J. Saeij, University of California, Davis, United States*

#### *Reviewed by:*

*Louis Weiss, Albert Einstein College of Medicine, United States Eric Denkers, University of New Mexico, United States*

> *\*Correspondence: Carlos S. Subauste carlos.subauste@case.edu*

#### *Specialty section:*

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

*Received: 26 February 2019 Accepted: 16 April 2019 Published: 01 May 2019*

#### *Citation:*

*Subauste CS (2019) Interplay Between Toxoplasma gondii, Autophagy, and Autophagy Proteins. Front. Cell. Infect. Microbiol. 9:139. doi: 10.3389/fcimb.2019.00139* Lysosomal degradation is an important mechanism of defense against numerous pathogens. This can be accomplished not only through the endocytic pathway but also through macroautophagy (called herein autophagy) (Levine et al., 2011). Autophagy is a homeostatic mechanism whereby large portions of cytosol or entire organelles are encircled by a double membrane (isolation membrane) leading to the formation of an autophagosome (Klionsky and Emr, 2000; Yoshimori, 2004; Mizushima et al., 2010). This structure fuses with lysosomes resulting in the formation of an autolysosome and cargo degradation (Mizushima et al., 2010).

Autophagy is dependent on a cascade of autophagy proteins (ATG). However, these proteins can have functions independent of autophagosome formation and lysosomal degradation (Subramani and Malhotra, 2013). This led to the use of the terms canonical and non-canonical autophagy where the latter was frequently used for processes that are non-degradative and/or not dependent on a core component(s) of the autophagy cascade [ATG3, ATG5, ATG7, Unc-51-like kinase 1 (ULK1), Beclin 1, and/or Phosphatidylinositol 3-kinase catalytic subunit type 3, PI3KC3, also known as VPS34] (Galluzzi et al., 2017). To avoid confusion, an expert panel recommended against the use of the terms "canonical"/"non-canonical," and advised that the term autophagy be used solely for processes dependent on autophagosomes where cytosolic material (either endogenous or exogenous) is directed to a process that culminates with and is strictly dependent on lysosomal degradation (Galluzzi et al., 2017). The processes can be further characterized by stating the autophagy proteins they are dependent on Galluzzi et al. (2017). In this review, we summarize studies on the interplay between T. gondii and host autophagy as well as non-degradative processes controlled by autophagy proteins.

# INVASION OF HOST CELLS BY *T. GONDII*

Tachyzoites of T. gondii infect virtually any nucleated cell and survive by residing in a compartment called the parasitophorous vacuole (PV). This vacuole is formed during active invasion of host cells, a process dependent on the parasite actin-myosin motor and sequential secretion of proteins from micronemes and rhoptries (Bradley and Sibley, 2007; Besteiro et al., 2011; Santos and Soldati-Favre, 2011). Once secreted from micronemes, T. gondii micronemal proteins (MICs) are expressed on the parasite surface and function as adhesins that interact with host cell membrane receptors (Carruthers and Tomley, 2008). MICs are expressed as multiprotein complexes that include MIC1/4/6, MIC3/8, MIC2/M2AP, and a complex of the microneme protein TgAMA1 with rhoptry neck proteins (Cerede et al., 2005; Huynh and Carruthers, 2006; Sheiner et al., 2010). MICs contain domains such as type I thrombospondin repeats, apple domains, epidermal growth factor (EGF) repeats, and integrin A domains (Tomley and Soldati, 2001; Anantharaman et al., 2007). The connection between transmembrane MICs and the parasite actin-myosin motor together with binding to host cell receptors enables the parasite to penetrate host cells (Soldati-Favre, 2008; Sibley, 2011). Following the release of MICs, rhoptries secrete a complex of neck proteins (RONs) containing RON2 that associates with the host cell membrane, plus RON4, RON5, and RON8 that are exposed to the host cell cytoplasm (Bradley and Sibley, 2007; Besteiro et al., 2011; Santos and Soldati-Favre, 2011). The complex forms a structure called moving junction that anchors the parasite to the host cell cytoskeleton during invasion (Bradley and Sibley, 2007; Besteiro et al., 2011; Santos and Soldati-Favre, 2011). Tachyzoites penetrate the host cell through the moving junction causing invagination of the host cell membrane. The moving junction also appears to function as a sieve that excludes host type I transmembrane proteins from entering the membrane that encircles the parasite as it penetrates the host cell (Mordue et al., 1999; Besteiro et al., 2011). Once invasion is completed, T. gondii resides within the PV. While host endocytic structures are delivered intact into the vacuolar space, there is no release of endosomal contents into the vacuole (Coppens et al., 2006). The lack of fusion with the endocytic compartment would be explained by the absence of host type I transmembrane proteins in the PV membrane (PVM) (Mordue et al., 1999; Besteiro et al., 2011).

# AUTOPHAGY OVERVIEW

Formation of the isolation membrane is dependent on recruitment of ATG proteins (Mizushima et al., 2010). Activation of both ULK1 and the complex that contains Beclin 1 and PI3KC3 drive the recruitment of ATG proteins to the isolation membrane promoting autophagosome formation and maturation (Chan et al., 2009; Itakura and Mizushima, 2010; Mizushima et al., 2010; Russell et al., 2013). ULK1 is the upstream kinase that triggers autophagy (Itakura and Mizushima, 2010). ULK1 is regulated by AMP-activated protein kinase (AMPK) and mechanistic target of rapamycin complex 1 (mTORC1), kinases that sense nutrient and energy status. In response to falling energy status, AMPK activates ULK1 and autophagy is stimulated (Chang et al., 2009; Egan et al., 2011; Kim et al., 2011; Mack et al., 2012). In contrast, ULK1 is inhibited by mTORC1 under nutrient rich conditions, leading to inhibition of autophagy (Chang et al., 2009). ULK1 undergoes membrane translocation upon activation by AMPK (Chang et al., 2009; Egan et al., 2011; Kim et al., 2011; Mack et al., 2012). Autophagosome biogenesis begins with the formation and activation of a ULK1-containing protein complex on membranes that express ATG9 (Papinski et al., 2014). ULK1 activates and recruits the Beclin 1–PI3KC3 complex to the membrane (Itakura and Mizushima, 2010). PI3KC3 causes production of PI3P at the membrane (Liang et al., 1999) and recruitment of PI3P-binding proteins that would act as scaffold for proteins that mediate membrane remodeling (Nascimbeni et al., 2017). Active Beclin-PI3KC3 triggers recruitment of ATG proteins that in turn function as two ubiquitin-like conjugation systems. In one cascade, ATG7 and ATG10 promote the conjugation of ATG5 to ATG12 (Mizushima et al., 1998). In the other cascade, ATG3 and ATG7 together with the ATG12-ATG5-ATG16L1 complex allow lipidation (phosphatidylethanolamide conjugation) of LC3 (ATG8) (Mizushima et al., 1998). Lipidated LC3 (LC3-II) is recruited to the autophagosome membrane (Kabeya et al., 2000) and allows substrate uptake by binding to several autophagy receptors (Stolz et al., 2014; Wild et al., 2014). Once the cargo is sequestered by the autophagosomes and through the effect of proteins that include Rab7 (Gutierrez et al., 2004), these structures fuse with lysosomes leading to the formation of an autolysosome.

Beclin 1 is regulated through protein-protein interactions. Beclin 1 binds proteins that promote autophagy (e.g., ATG14L) whereas binding to other proteins (e.g., Bcl-2 family members) inhibits autophagy (Pattingre et al., 2005; Sun et al., 2008; Matsunaga et al., 2009). Under basal conditions Bcl-2 binds to the BH3 domain of Beclin 1 preventing the association of Beclin 1 to PI3KC3 and the initiation of autophagy (Pattingre et al., 2005). Starvation stimulates autophagy in part because it triggers JNK1 dependent phosphorylation of Bcl-2 that releases Beclin 1 from Bcl-2 (Wei et al., 2008).

Autophagy proteins can be involved in cellular processes activated during intracellular infections that do not represent bona fide autophagy. LC3-associated phagocytosis (LAP) represents an example of such a process. LAP consists in the recruitment of LC3 and some other components of the autophagy pathway to single-membrane phagosomes that contain pathogens or dead cells that have been actively phagocytosed (Sanjuan et al., 2007). While LAP requires proteins that include ATG3, ATG5, ATG7, ATG12, ATG16L, Beclin 1, and PI3KC3, other proteins notably ULK1 are not involved in this process (Martinez et al., 2015). LAP is believed to result in faster fusion with lysosomes and plays a protective role against various pathogens (Sanjuan et al., 2007; Martinez et al., 2015). Autophagy proteins can be involved in additional mechanisms of anti-microbial activity that occur independently of the formation of autophagosomes and are not mediated by lysosomal degradation of the pathogen. In this regard, IFN-γ induces autophagy protein-dependent recruitment of GTPases that disrupt the integrity of the PV membrane (see below).

# AUTOPHAGY DURING *T. GONDII* INFECTION

#### CD40 Stimulates Autophagy and Triggers Autophagic Targeting of *T. gondii*

Autophagy can be stimulated by innate and adaptive immune mechanisms to degrade various pathogens (Levine et al., 2011). Pattern recognition receptors including TLR and NOD2 as well as cytokines such IFN-γ and type I interferon can stimulate autophagy (Shi and Kehrl, 2010; Gade et al., 2012; Matsuzawa et al., 2012; Chauhan et al., 2015). CD40 is a stimulator of autophagy that confers resistance against cerebral and ocular toxoplasmosis. CD40 is a member of the TNF receptor superfamily that is expressed on antigen presenting cells and various non-hematopoietic cells (Van Kooten and Banchereau, 2000). CD40 ligand (CD154) is expressed primarily on activated CD4<sup>+</sup> T cells but is also present in activated platelets and plasma (Van Kooten and Banchereau, 2000). Studies in patients with congenital lack of functional CD154 (X-linked Hyper IgM syndrome) uncovered the central role of the CD40- CD154 pathway in protection against toxoplasmosis (Subauste et al., 1999). While the CD40-CD154 pathway promotes Th1 type cytokine response against T. gondii (Subauste et al., 1999; Reichmann et al., 2000), toxoplasmacidal activity induced by CD40 ligation in infected cells likely contributes to protection against the parasite (Reichmann et al., 2000; Portillo et al., 2010). Using peritoneal cells from T. gondii-infected mice, it has been proposed that CD40 plays a secondary role in parasite elimination in macrophages, although the CD4<sup>+</sup> T cell– macrophage ratio (extent of CD40-CD154 interaction in vitro) and whether macrophages infected in vitro had undergone prior CD40-CD154 signaling in vivo were unclear (Zhao et al., 2007). Other studies demonstrated that CD154<sup>+</sup> T. gondii-reactive CD4<sup>+</sup> T cells induce anti-T. gondii activity in macrophages even if CD40 ligation occurs in cells already infected with T. gondii (Andrade et al., 2006). Importantly, studies in CD154−/<sup>−</sup> and CD40−/<sup>−</sup> mice established that this pathway is central for restricting parasite load in the brain and retina, and protecting against cerebral and ocular toxoplasmosis (Reichmann et al., 2000; Portillo et al., 2010), the two main forms of disease in humans.

Several lines of evidence indicate that CD40 stimulates autophagy and induces killing of T. gondii through autophagic targeting of the parasite, a phenomenon that occurs in hematopoietic and non-hematopoietic cells from both human and mice. CD40 ligation increases conversion of the autophagy protein LC3-I to LC3-II, as well as increases formation of autophagosomes and autolysosomes (autophagy flux). These events are dependent on ULK1, ATG5, ATG7, and Beclin 1 (Andrade et al., 2006; Portillo et al., 2010; Ogolla et al., 2013; Van Grol et al., 2013; Liu et al., 2015). In cells infected with T. gondii, CD40 ligation induces accumulation of mannose 6 phosphate receptor, Rab7, LAMP-1, LAMP-2, CD63, and cathepsin D around the PV as well as co-localization of these vacuoles with the acidotropic dye Lysotracker (Andrade et al., 2006; Portillo et al., 2010; Ogolla et al., 2013; Van Grol et al., 2013). Accumulation of lysosomal markers occurs around vacuoles that contain proteins secreted by parasite dense granules within the vacuolar lumen (Andrade et al., 2006). This indicates that the events triggered by CD40 represent fusion of the PV with lysosomes rather than being a consequence of phagocytosis since secretion of dense granule contents takes place during formation of PV but not during phagocytosis of T. gondii. Moreover, vacuolelysosomal fusion (VLF) still occurs even if CD40 is engaged 18 h after infection (Andrade et al., 2006). VLF is preceded by accumulation of LC3 around the PV. Autophagy mediates VLF and killing of T. gondii since knockdown of ULK1, Beclin 1, PI3KC3, ATG5, or ATG7, expression of dominant negative Rab7 or pharmacologic inhibition of vacuolar ATPase or PI3K, and importantly incubation with lysosomal enzyme inhibitors ablate killing of T. gondii induced by CD40 (Andrade et al., 2006; Portillo et al., 2010; Van Grol et al., 2013). CD40 triggers VLF via autophagy rather than LAP since ULK1 is required for autophagy while LAP takes place independently of ULK1 (Martinez et al., 2015). In addition, the events triggered by CD40 ligation do not represent phagocytosis of the parasite.

CD40 stimulates autophagy via four mechanisms (**Figure 1**). First, CD40 induces CaMKKβ-mediated Threonine-172 AMPK phosphorylation, a marker of AMPK activation (Liu et al., 2016). In turn, AMPK signaling causes Serine-555 ULK1 phosphorylation and ULK1-mediated autophagy (Liu et al., 2016). Second, CD40 induces autocrine secretion of TNF-α that causes JNK1/2-dependent phosphorylation of Bcl-2 at Serine 87 and dissociation of Bcl-2 from Beclin 1 (Subauste et al., 2007; Liu et al., 2016). The latter process is known to allow binding of Beclin 1 to PI3KC3 and initiation of autophagy (Pattingre et al., 2005). Third, CD40 upregulates Beclin 1 protein levels in vitro and in vivo (Portillo et al., 2010). This effect appears to occur through downregulation of p21, a protein that degrades Beclin 1 (Portillo et al., 2010). Consistent with the evidence that the level of Beclin 1 expression is linked to autophagic activity (Liang et al., 1999), CD40-induced Beclin 1 upregulation facilitates autophagic killing of T. gondii triggered by CD40 (Portillo et al., 2010). These three events act in synchrony, likely optimizing the ability of CD40 to stimulate autophagy and induce toxoplasmacidal activity. Finally, CD40 also promotes autophagy by activating PKR (Ogolla et al., 2013), a serine-threonine kinase that stimulates autophagy (Talloczy et al., 2002, 2006). These events are relevant to T. gondii since autophagic targeting and/or killing of the parasite induced by CD40 is dependent on CaMKKβ, AMPK, TNF-α, JNK1/2, Beclin 1 upregulation, p21 downregulation, and PKR (Subauste et al., 2007; Portillo et al., 2010; Ogolla et al., 2013; Liu et al., 2016).

Animal studies support the importance of autophagy for control of T. gondii in the brain and eye. Autophagy-deficient BECN1+/<sup>−</sup> mice, mice with deficiency of the autophagy protein ATG7 targeted to microglia/macrophages (Atg7flox/flox-Lyz-M Cre mice) and PKR−/<sup>−</sup> mice are susceptible to cerebral and ocular toxoplasmosis (Portillo et al., 2010; Ogolla et al., 2013). This susceptibility is not explained by defects in cellular or humoral immunity against the parasite. Moreover, macrophages/microglia from these mice exhibit impaired killing of T. gondii in response to CD40 but not IFN-γ stimulation (Portillo et al., 2010; Ogolla et al., 2013).

### *T. gondii* Manipulates Host Cell Signaling to Avoid Targeting by Autophagy

Avoidance of the lysosomal compartment is essential for T. gondii survival. Autophagy is a constitutive process in eukaryotic cells. Moreover, a fraction of host cells is unable to exhibit autophagic targeting and VLF of intracellular tachyzoites despite activation through CD40. These findings suggest that T. gondii

uses strategies to avoid targeting by autophagosomes. Indeed, the parasite activates host cell signaling pathways that achieve this purpose.

Epidermal growth factor receptor (EGFR) is expressed in various cells various cell types (including epithelial cells, endothelial cells, microglia, and certain neurons) and can inhibit autophagy (Sobolewska et al., 2009). EGFR is composed of extracellular (ligand binding), transmembrane, intracellular tyrosine kinase and carboxyl-terminal tail domains (Purba et al., 2017). Ligand binding causes a conformational change in the kinase domain leading to activation of EGFR through autophosphorylation of tyrosine residues in the carboxylterminal tail (Purba et al., 2017). These phosphorylated residues recruit signaling molecules downstream of EGFR (Purba et al., 2017). T. gondii induces phosphorylation of the tyrosine residues 1,068, 1,148, and 1,173 of EGFR during infection of human or rodent cells (Muniz-Feliciano et al., 2013). T. gondiiinduced EGFR signaling leads to activation of PI3K (Muniz-Feliciano et al., 2013), a molecule that triggers production of phosphatidylinositol 3,4,5 trisphosphate (PIP3). Transfection of host cells with a plasmid encoding GFP-tagged amino-terminal pleckstrin homology (PH) domain of Akt that binds PIP3 revealed PIP3 accumulation around the PV (Muniz-Feliciano et al., 2013). Consistent with the fact that PIP3 production is a major trigger of Akt activation, T. gondii induces Akt activation (Muniz-Feliciano et al., 2013) (**Figure 2A**). While parasite-induced Akt activation in macrophages is impaired by Pertussis Toxin (PTx) suggesting that Akt signaling can be dependent on G-protein coupled receptors (GPCR) (Kim and Denkers, 2006), genetic and pharmacologic blockade of EGFR in various cell types including macrophages/microglia revealed that EGFR is an important driver of Akt activation triggered by T. gondii (Muniz-Feliciano et al., 2013).

Inhibition of the EGF:Akt pathway results in spontaneous recruitment of LC3 and formation of a double membrane structure around the PVM followed by VLF (Muniz-Feliciano et al., 2013). In both human and mouse cells, ensuing killing of type I and II strains of T. gondii is dependent on ULK1, Beclin 1, ATG7, and lysosomal enzymes (Muniz-Feliciano et al., 2013). These results are likely explained by the fact that Akt is a negative regulator of autophagy via activation of mTORC1 (Menon et al., 2014). Given that Akt activation is linked to inhibition of apoptosis of T. gondii-infected cells (Kim and Denkers, 2006), parasite-induced EGFR-Akt signaling may promote parasite survival by preserving the non-fusogenic nature of the PV and by avoiding death of infected cells subjected to pro-apoptotic signals.

EGFR can be activated by transmembrane proteins that are shed from the plasma membrane as a consequence of the ADAM (a disintegrin and metalloprotease) family of zinc-dependent metalloproteases (Yarden and Sliwkowski, 2001). This process is stimulated by GPCR (Yarden and Sliwkowski, 2001). However, treatment with GM6001, a broad-spectrum ADAM inhibitor, or with PTx fails to inhibit T. gondii-induced EGFR activation (Muniz-Feliciano et al., 2013). The parasite causes EGFR phosphorylation at tyrosine 1,148 (Muniz-Feliciano et al., 2013), a site that appears to be phosphorylated only by ligand binding to EGFR (Moro et al., 2002). In this regard, MIC3, MIC6, MIC8 have multiple domains with homology to EGF (Meissner et al., 2002). Recombinant MIC3 and MIC6 but not MIC4 or M2AP induce EGFR-Akt signaling in mammalian cells and impair the ability of CD154 to induce LC3 accumulation around the parasite (Muniz-Feliciano et al., 2013). In addition, MIC1 ko (deficient in MIC6), MIC3 ko and especially MIC1/3 ko parasites are defective in induction of EGFR-Akt activation (Muniz-Feliciano et al., 2013). While cells infected with MIC1/3 ko T. gondii do not exhibit spontaneous targeting by LC3<sup>+</sup> structures, there is increased recruitment of LC3 and susceptibility to killing after incubation with stimulators of autophagy (CD154, Rapamycin) (Muniz-Feliciano et al., 2013). The likely explanation for these results is that MIC1/3 ko T. gondii have residual ability to induce EGFR-Akt signaling (Muniz-Feliciano et al., 2013). Although MIC8 has EGF-like domains, MIC8 ko parasites show no defect in EGFR activation (Muniz-Feliciano et al., 2013). These findings are likely explained by the fact that MIC8 ko parasites are not defective in host cell attachment and secrete MICs (Kessler et al., 2008). Whether simultaneous deficiency in MIC3, MIC6 and MIC8 ablates EGFR autophosphorylation or whether there is another mechanism that contributes to EGFR autophosphorylation remains to be determined. Taken together, in addition to being key for invasion of host cells, these studies indicate that MIC3 and MIC6 promote EGFR-Akt signaling to avoid lysosomal degradation of the parasite (**Figure 2A**).

Another mechanism operative in both human and murine cells that enables type I, II, and atypical strains of T. gondii to avoid targeting by autophagosomes is dependent on activation

of Focal Adhesion Kinase (FAK), a cytoplasmic molecule that links extracellular signals to intracellular signaling cascades. T. gondii induces FAK activation at the level of the moving junction, an effect that is largely mediated by β integrins, presumably in the form of mechano-transduction-induced integrin clustering at the site of penetration of host cells (Portillo et al., 2017) (**Figure 2B**). Src becomes activated as a consequence of T. gondiiinduced FAK activation (Portillo et al., 2017). Src can bind EGFR and transactivate this receptor even in the absence of ligand binding (Biscardi et al., 1999). EGFR transactivation is characterized by phosphorylation of a unique tyrosine 845 in the kinase domain of EGFR that recruits alternate signaling cascades downstream of EGFR including STAT3. Indeed, T. gondii triggers Src dependent phosphorylation of tyrosine 845 of EGFR followed by activation of STAT3 (Portillo et al., 2017) (**Figure 2B**), a negative regulator of autophagy (Van Grol et al., 2010; Shen et al., 2012). In the case of T. gondii infection, STAT3 activation prevents autophagic targeting of the parasite by impairing activation of the pro-autophagy protein PKR and its downstream signaling molecule eIF2α (Portillo et al., 2017). Genetic or pharmacologic blockade of any component of the FAK:Src:p845Y EGFR:STAT3 pathway causes recruitment of LC3 around the parasite, VLF and parasite killing dependent on ULK1, Beclin 1, and lysosomal enzymes (Portillo et al., 2017). Thus, T. gondii activates an Akt- and a STAT3-dependent signaling pathway in both human and mouse cells to avoid autophagic targeting, and these pathways appear to function independently (Portillo et al., 2017) (**Figure 2**).

Animal studies have recently demonstrated the in vivo relevance of T. gondii-induced manipulation of host cell signaling in the pathogenesis of cerebral and ocular toxoplasmosis. The CNS is invaded via the blood stream when tachyzoites present in circulating infected leukocytes or extracellular tachyzoites reach the brain (Courret et al., 2006; Konradt et al., 2016). CNS invasion is preceded by infection of endothelial cells (Konradt et al., 2016). Expression of a dominant negative EGFR in endothelial cells ablates T. gondii-induced autophosphorylation and transactivation of EGFR (Lopez Corcino et al., 2019). Transgenic mice whose endothelial cells express DN EGFR exhibit diminished parasite load and histopathology in the brain and retina after T. gondii infection (Lopez Corcino et al., 2019). Mice with DN EGFR have reduced parasite load in these organs after i.v. administration of infected leukocytes or extracellular tachyzoites (Lopez Corcino et al., 2019). This protective effect is not explained by enhanced immunity or reduced leukocyte recruitment into the CNS. Rather, the effect of DN EGFR is to reduce the foci of infection in neural endothelial cells (Lopez Corcino et al., 2019). DN EGFR in these cells results in the spontaneous recruitment of LC3 around T. gondii, VLF and parasite killing dependent on ULK1 and lysosomal enzymes (Lopez Corcino et al., 2019). Moreover, in vivo administration of autophagy inhibitor 3 methyl adenine prevents DN EGFR mice from exhibiting reduced CNS invasion (Lopez Corcino et al., 2019). Altogether, EGFR is a novel regulator of T. gondii invasion of neural tissue, enhancing invasion likely by promoting survival of the parasite within endothelial cells through avoidance of autophagic targeting.

Although T. gondii activates signaling molecules that can inhibit autophagy, T. gondii does not prevent autophagosome formation in infected cells (Wang et al., 2009). In fact, T. gondii increases LC3-II levels and autophagosome formation in host cells at 24 h post-infection (Wang et al., 2009). These studies together with the demonstration that T. gondii induces lipophagy in host cells to obtain fatty acids (Pernas et al., 2018) would support that the parasite co-opts host cell autophagy to gain access to nutrients for its growth (Wang et al., 2009; Pernas et al., 2018). A model has been proposed whereby T. gondii utilizes the autophagy machinery of permissive (non-activated) host cells for its own benefit, whereas host cell autophagy would lead to parasite killing only in immune-activated host cells (Latre De Late et al., 2017). However, the signaling studies described above revealed an additional layer of complexity. They begin to indicate that CD40 and T. gondii have opposing effects on signaling molecules that regulate autophagy (i.e., PKR). The balance between these opposing effects may determine whether autophagic targeting of the parasite takes place. Host cell autophagy would cause parasite killing not only in CD40 activated host cells but also in resting cells if the effects of T. gondii on EGFR:Akt or FAK/Src:p-Y845 EGFR:STAT3 signaling are blocked. This model has therapeutic implications since, for example, addition of EGFR tyrosine kinase inhibitors to resting cells restricts T. gondii growth (Muniz-Feliciano et al., 2013; Yang et al., 2014). Host cell autophagy would be beneficial to the parasite in resting host cells as long as the parasite is able to activate negative regulators that prevent autophagosomes from targeting and killing the parasite. In these cells, activation of the regulatory signaling cascades would not appear to inhibit global autophagy but rather would impair targeting of the parasite by autophagosomes.

#### AUTOPHAGY-INDEPENDENT EFFECTS OF AUTOPHAGY PROTEINS DURING *T. GONDII* INFECTION

#### IFN-γ Restricts *T. gondii* Through Autophagy-Independent Effects of Autophagy Proteins

IFN-γ is a major activator of effector responses against T. gondii in mammalian cells. IFN-γ causes vesiculation and rupture of the PVM in mouse cells leading to parasite release into the cytoplasm and parasite death (Martens et al., 2005; Zhao et al., 2008, 2009). While autophagosome-like structures can be noted around the parasites (Ling et al., 2006), the function of these structures is not to kill the parasite but likely to clear parasite/PVM fragments (Zhao et al., 2008; Choi et al., 2014; Ohshima et al., 2014). Indeed, parasite killing in IFN-γ-activated mouse cells is not mediated by lysosomal activity since expression of DN Rab7 that would inhibit lysosomal fusion and/or autolysosome maturation (Andrade et al., 2006) or incubation with lysosomal protease inhibitors fail to impair the anti-T. gondii activity induced by IFN-γ (Andrade et al., 2006; Van Grol et al., 2013; Choi et al., 2014). Thus, bona fide autophagy does not mediate the anti-T. gondii effects of IFN-γ.

Interestingly, while autophagosomes are not involved in killing of susceptible strains of T. gondii within IFN-γ-activated cells (Zhao et al., 2008), selected autophagy proteins are required for parasite death in mouse cells. These autophagy proteins function by promoting recruitment of Immunity Regulated GTPases (IRGs) and Guanylate Binding Proteins (GBPs) to the PVM (**Figure 3A**). IFN-γ induces recruitment and loading of effector GKS subfamily of IRGs (Irga6, Irgb6, Irgb10, Irgd) onto the PVM causing its disruption (Martens et al., 2005; Ling et al., 2006; Khaminets et al., 2010). IRGs promote ubiquitin deposition on the PVM followed by p62-dependent recruitment of GBPs (Haldar et al., 2015) (**Figure 3A**). Disruption of the PVM enables GBPs to bind and kill the parasite (Kravets et al., 2016). ATG5 is required for recruitment of GKS IRGs (Irga6, Irgb6, Irgd) and mGBP1 to the PVM in IFN-γ-activated mouse macrophages and fibroblasts (Zhao et al., 2008; Khaminets et al., 2010; Selleck et al., 2013). Similarly, ATG7 and ATG16L1 are necessary for Irgb6 and mGBP1-5 recruitment (Choi et al., 2014; Ohshima et al., 2014) while ATG3 is required for recruitment of Irgb6, Irb10 and mGBP1-5 in fibroblasts (Choi et al., 2014; Haldar et al., 2014). Consistent with the fact that ATG3, ATG7, and the ATG12-ATG5-ATG16L1 complex mediate LC3 conjugation, LC3 is recruited to the PVM in IFN-γ-activated mouse macrophages and fibroblasts (Choi et al., 2014; Park et al., 2016). The LC3 homologs gamma-aminobutyric acid-Areceptor-associated proteins (GABARAPs) are also recruited to the PVM in a conjugation-dependent manner (Park et al., 2016). These proteins target IRGs to the PVM in mouse cells (Park et al., 2016). In another study, GABARAPL2 (Gate-16) but not LC3 was required for recruitment of Irga6 and GBP1-5 to the vacuole of IFN-γ-treated mouse fibroblasts (Sasai et al., 2017).

As stated above, ATG proteins do not function through bona fide autophagy to restrict T. gondii in IFN-γ activated cells. Indeed, lysosomal degradation does not mediate the effects of these proteins (Choi et al., 2014). Moreover, ULK1 and ATG14L are not required in order for IFN-γ to control T. gondii (Choi et al., 2014). Similarly, ATG9, ATG14L and FIP200 are not required for recruitment of LC3, Irg6a and GBP to the vacuole (Choi et al., 2014; Ohshima et al., 2014; Sasai et al., 2017). While the mechanism of action of ATG proteins remains to be fully elucidated, these proteins may activate IRGs (Haldar et al., 2014), LC3 may target IRGs to the membrane (Park et al., 2016), and Gate-16 associates with the small GTPase ADP-ribosylation factor 1 (Arf1) to mediate IRG recruitment (Sasai et al., 2017).

IRGs represent the major mechanism by which IFN-γ protects mice during the acute phase of T. gondii infection (Martens et al., 2005). Consequently, ATG proteins not only mediate the anti-T. gondii activity in mouse cells activated by IFN-γ but they are also required for in vivo protection. Mice with deficiency in ATG5, ATG7, or ATG16L targeted to phagocytes exhibit marked susceptibility to acute infection with T. gondii (Zhao et al., 2008; Choi et al., 2014). In contrast, ATG14L deficiency does not increase susceptibility to acute infection (Choi et al., 2014). In addition, Gate-16−/<sup>−</sup> mice succumb to

acute infection with T. gondii in a manner that mimics IFN-γ −/− mice (Sasai et al., 2017).

The effector mechanisms activated by IFN-γ to restrict T. gondii growth in human cells are less well-characterized and differ from those in mouse cells. Mechanisms in human cells appear to be cell-type specific and are reported to include induction of indoleamine 2,3-dioxygenase that deprives the parasite from tryptophan (Pfefferkorn, 1984) and host cell death that results in early parasite egress without replication (Niedelman et al., 2013). In contrast to mice, humans express only 2 IRGs that cannot be induced by IFN-γ explaining why IRGs do not mediate the effects of IFN-γ in human cells (Bekpen et al., 2005). Human cells express GBPs (Ohshima et al., 2014). Moreover, hGBP1-5 are recruited to the parasite in an ATG16L-dependent manner in IFN-γ-activated human HAP1 cells (Ohshima et al., 2014). However, GBPs are not required for restriction of T. gondii (Ohshima et al., 2014). Studies in human epithelial (HeLa) cells identified a mechanism for control of type II and type III T. gondii induced by IFN-γ that is dependent on ubiquitination and some ATG proteins (Selleck et al., 2015). IFN-γ induces ubiquitination of the PV and recruitment of the ubiquitin adaptor proteins p62 and Nuclear Domain 10 Protein 52 (NDP52), as well as LC3 (Selleck et al., 2015). These vacuoles become surrounded by multiple layers of host membrane that would restrict parasite growth (Selleck et al., 2015). While this process is dependent on ATG16L and ATG7, it occurs independently of Beclin 1 and does not cause VLF, indicating that it does not represent autophagy (Selleck et al., 2015).

IFN-γ was reported to restrict T. gondii growth within human endothelial cells through a mechanism that remained to be identified (Woodman et al., 1991). A recent study revealed that vacuoles containing type II T. gondii within human endothelial cells are targeted by K63-linked ubiquitin in response to IFNγ (Clough et al., 2016). This is followed by recruitment of p62 and NDP52, acidification of the vacuole and parasite death. This process does not represent autophagy since it is not accompanied by recruitment of ATG16L, GABARAP, and LC3 (Clough et al., 2016). Moreover, in contrast to IFN-γ-activated HeLa cells, the ability of IFN-γ to restrict parasite growth in human endothelial cells is not dependent on ATG16L (Clough et al., 2016). Taken together, there are two novel cell-type specific mechanisms by which IFN-γ restricts the parasite growth in human nonhematopoietic cells. These mechanisms involve ubiquitination of the vacuole followed by either the formation of a multilayer structure around the vacuole or vacuole acidification.

### *T. gondii* Impairs Recruitment of IRGs to the PVM

The ability of IRGs to protect mice against T. gondii depends on the parasite strain. While virulent type I T. gondii prevents recruitment of IRGs to the PVM and avoids eradication, low virulence type II strains and avirulent type III strains of T. gondii cannot avoid IRG recruitment and are thus eradicated (Zhao et al., 2009; Khaminets et al., 2010). Evasion of IRG recruitment is mediated by parasite proteins released within host cells during invasion. The rhoptry protein ROP18 is a polymorphic protein kinase and a major determinant of parasite virulence in mice (Saeij et al., 2006; Taylor et al., 2006). Type I T. gondii secretes a catalytically active form of ROP18 that phosphorylates IRGs at two threonine residues in the nucleotide-binding domain (Fentress et al., 2010; Steinfeldt et al., 2010) (**Figure 3B**). As a result, the GTPase function of IRGs is inhibited and their oligomerization and loading into the PVM are impaired (Fentress et al., 2010; Steinfeldt et al., 2010). The ability of ROP18 to phosphorylate IRGs is dependent on the presence of virulent alleles of ROP5. ROP5 are a group of catalytically inactive kinases (pseudokinases) that control parasite virulence in mice (Behnke et al., 2011; Reese et al., 2011). ROP5 proteins bind a conserved surface of IRG and promote that IRG remain in an inactive GDP-bound conformation (Fleckenstein et al., 2012; Reese et al., 2014). As a result, GTP-dependent activation of IRG is prevented, and threonines in the nucleotide-binding domain become exposed, enabling their phosphorylation by ROP18 and permanent inactivation of IRG (Fleckenstein et al., 2012; Reese et al., 2014). Thus, ROP5 appears to act as an allosteric regulator of ROP18 (Reese et al., 2014) and is required for the catalytic activity of ROP18 (Behnke et al., 2012). Indeed, virulent forms of both ROP5 and ROP18 are required to prevent IRG recruitment to the PVM. ROP18 and ROP5 largely explain the differences in virulence in mice among type I, II, and III strains (Saeij et al., 2006; Taylor et al., 2006). Despite encoding ROP18 that is likely catalytically active, type II strains cannot prevent IRG recruitment because they carry alleles of ROP5 that do not assist IRG phosphorylation by ROP18. In addition, type III strains carry a high-virulence allele of ROP5 but are avirulent because of their minimal expression of ROP18. The allelic combination of ROP18 and ROP5 genes also determines the virulence of atypical strains of T. gondii (Niedelman et al., 2012).

In addition to ROP18 and ROP5, ROP17 also contributes to T. gondii virulence in mice (Etheridge et al., 2014). ROP17 associates with ROP5 and phosphorylates threonine residues of

#### REFERENCES


IRG (Etheridge et al., 2014). However, in contrast to ROP18, the in vitro activity of ROP17 does not require ROP5 (Etheridge et al., 2014). Finally, the dense granule protein GRA7 is another component of the ROP18-ROP5 complex and modulates IRG recruitment to the PVM (Hermanns et al., 2016) (**Figure 3B**). GRA7 appears to associate with ROP5 and functions by allowing efficient ROP18 kinase activity (Hermanns et al., 2016).

In summary, important advances have been achieved in our understanding of how autophagy proteins and autophagy attack T. gondii-containing vacuoles within host cells. Given that maintaining the integrity of this niche is essential to parasite survival, it is not surprising that T. gondii utilizes various strategies to counteract the effects of autophagy proteins and autophagy. Pharmacologic approaches to enhance autophagy for therapeutic purposes may be complicated by the homeostatic role of autophagy in various cellular processes, the complexity of autophagy cascades, and the specificity of pharmacologic agents. Strategies to prevent T. gondii from blocking autophagic targeting may represent a more feasible avenue to develop novel ancillary approaches to improve the treatment of toxoplasmosis.

#### AUTHOR CONTRIBUTIONS

The author confirms being the sole contributor of this work and has approved it for publication.

#### FUNDING

CS is funded by NIH-R01 EY018341 and NIH-R01 EY019250.

#### ACKNOWLEDGMENTS

The author thanks all the members of the Subauste lab for their feedback on this manuscript.


and nutrient regulation of mTORC1 at the lysosome. Cell 156, 771–785. doi: 10.1016/j.cell.2013.11.049


**Conflict of Interest Statement:** The author declares 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 Subauste. 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.

# Impact of *Toxoplasma gondii* Infection on Host Non-coding RNA Responses

#### Kayla L. Menard\*, Breanne E. Haskins and Eric Y. Denkers\*

Department of Biology, Center for Evolutionary and Theoretical Immunology, University of New Mexico, Albuquerque, NM, United States

#### *Edited by:*

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### *Reviewed by:*

Mohamed Ali Hakimi, Institut National de la Santé et de la Recherche Médicale (INSERM), France Xing-Quan Zhu, Lanzhou Veterinary Research Institute (CAAS), China

#### *\*Correspondence:*

Kayla L. Menard kmenard@unm.edu Eric Y. Denkers edenkers@unm.edu

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

> *Received:* 07 March 2019 *Accepted:* 12 April 2019 *Published:* 14 May 2019

#### *Citation:*

Menard KL, Haskins BE and Denkers EY (2019) Impact of Toxoplasma gondii Infection on Host Non-coding RNA Responses. Front. Cell. Infect. Microbiol. 9:132. doi: 10.3389/fcimb.2019.00132 As an intracellular microbe, Toxoplasma gondii must establish a highly intimate relationship with its host to ensure success as a parasite. Many studies over the last decade-and-a-half have highlighted how the host reshapes its immunoproteome to survive infection, and conversely how the parasite regulates host responses to ensure persistence. The role of host non-protein-coding RNA during infection is a vast and largely unexplored area of emerging interest. The potential importance of this facet of the host-parasite interaction is underscored by current estimates that as much as 80% of the host genome is transcribed into non-translated RNA. Here, we review the current state of knowledge with respect to two major classes of non-coding RNA, microRNA (miRNA) and long non-coding RNA (lncRNA), in the host response to T. gondii infection. These two classes of regulatory RNA are known to have profound and widespread effects on cell function. However, their impact on infection and immunity is not well-understood, particularly for the response to T. gondii. Nevertheless, numerous miRNAs have been identified that are upregulated by Toxoplasma, and emerging evidence suggests a functional role during infection. While the field of lncRNA is in its infancy, it is already clear that Toxoplasma is also a strong trigger for this class of regulatory RNA. Non-coding RNA responses induced by T. gondii are likely to be major determinants of the host's ability to resist infection and the parasite's ability to establish long-term latency.

Keywords: non-coding RNA, microRNA, miRNA, long non-coding RNA, lncRNA, *Toxoplasma gondii*, parasite

# INTRODUCTION

#### *Toxoplasma gondii* and the Immune Response

Toxoplasma gondii is one of the most prevalent human parasites in the world. It infects a wide range of species and establishes latent infection in brain and muscle tissue. In immune compromised individuals, as well as in the developing fetus, infection can result in severe disease (McLeod et al., 2013). Toxoplasma initiates strong protective Th1 immunity through induction of dendritic cell IL-12, while also inducing the activity of counter-regulatory cytokines such as IL-10 (Dupont et al., 2012; Sasai et al., 2018). In mouse models, parasite profilin functions as a pathogen-associated molecular pattern molecule triggering IL-12 through host Toll-like receptors 11 and 12 (Andrade et al., 2013; Raetz et al., 2013; Gazzinelli et al., 2014; Yarovinsky, 2014). From within the cell, Toxoplasma directly injects host-directed effector proteins such as ROP16, TgIST, GRA18, and GRA24 (Olias et al., 2016; Hakimi et al., 2017; He et al., 2018).

These proteins seize control of host signaling responses through respective activation of STAT3/6, NFκB, and p38 MAPK molecules (Ong et al., 2010; Butcher et al., 2011; Rosowski et al., 2011; Braun et al., 2013). It is likely that a major factor in the success of Toxoplasma lies in its ability to produce host-directed effectors that act to ensure a balance between pro-inflammatory and anti-inflammatory responses. Host noncoding RNA responses are now emerging as important regulators of cell function. Regarding the host response to Toxoplasma, several microRNAs and a growing list of long non-coding RNAs are known to be triggered by infection. Here, we survey the current state of our knowledge of this important but still little understood class of host responses during infection with T. gondii.

# Non-coding RNA

Only a small percentage (<3%) of the genome codes for proteins, yet the majority (∼80%) is actively transcribed (Spurlock et al., 2016; Atianand et al., 2017). The transcripts that do not code for proteins are collectively referred to as non-coding RNAs (ncRNAs) and they are separated into two general categories: shorter non-coding RNAs and longer non-coding RNAs (Xue et al., 2017; Zhang et al., 2017). Shorter non-coding RNAs are defined as being <200 nucleotides in length and include microRNAs (miRNAs), short interfering RNAs (siRNAs), transfer RNAs (tRNAs), piwi-interacting RNAs (piwiRNA), and small nucleolar RNAs (snoRNAs). Non-coding RNAs <200 nucleotides include ribosomal RNAs (rRNAs) and long noncoding RNAs (lncRNAs). Here, we focus on two types of noncoding RNAs: microRNAs and long non-coding RNAs. While several other classes of RNAs exist, miRNAs and lncRNAs are the two major classes of host ncRNA that have been examined during Toxoplasma infection.

## microRNA

MicroRNAs are ∼18 to 22 nucleotides in length. While examples of miRNAs acting transcriptionally exist, they primarily function post-transcriptionally by directly binding to mRNAs through direct base pair interactions (Xue et al., 2017). This interaction leads to mRNA cleavage, mRNA degradation, or blocking of translation (**Figure 1A**). MicroRNAs play important roles in regulating both innate and adaptive immunity. For example, the miR-17-92 cluster regulates B-cell, T-cell, and monocyte development through downregulation of the proapoptotic protein Bim (Xiao et al., 2008). The miR-146 family is a negative regulator of the innate immune response and may target TRAF6 and IRAK1 (Taganov et al., 2006; Lindsay, 2008; Cannella et al., 2014). miR-155 is a regulator of T-cell and B-cell maturation, as well as the innate immune response (Lindsay, 2008; Cannella et al., 2014).

# lncRNA

The largest group of RNA produced is long non-coding RNAs (lncRNAs), and it accounts for up to 68% of the transcriptome, not including ribosomal RNAs (Iyer et al., 2015; Chen et al., 2017). Compared to microRNAs, lncRNAs are much longer and more complex in structure and function. Thus, lncRNAs have multiple operational units and extensive functional diversity through their ability to interact with RNA, DNA, and protein (Guttman and Rinn, 2012; Fitzgerald and Caffrey, 2014; Chen et al., 2017). lncRNAs are widely involved in gene regulation at both transcriptional and post-transcriptional levels. Known Menard et al. T. gondii Impact on Host Non-coding RNA

functions of lncRNAs include transcriptional co-activation, recruitment of chromatin modifiers, miRNA sponges, regulation of splicing, and mRNA stabilization (**Figure 1B**; Fitzgerald and Caffrey, 2014; Szcze´sniak and Makałowska, 2016). The study of lncRNAs in the immune system is a relatively new field. In fact, the first study identifying a function for a particular lncRNA involved in the innate immune response was published as recently as 2013 (Carpenter et al., 2013). This lncRNA, lncRNA-Cox2, is broadly involved in both activation and repression of immune genes by performing functions both in cis at nearby genes and trans at genes on different chromosomes (Elling et al., 2018). In addition to lncRNA-Cox2, several other lncRNAs are now known to play a role in the immune response (Spurlock et al., 2016; Atianand et al., 2017; Chen et al., 2017). Recently, lncRNA responses were examined in epithelial cells infected with the intestinal pathogen Cryptosporidium parvum, an apicomplexan related to Toxoplasma. One lncRNA in particular, NR\_045064, was found to regulate selected host defense genes through modifications in chromatin structure (Li et al., 2018).

#### Host microRNA in the Response to *T. gondii*

Numerous studies have surveyed global host miRNA responses during T. gondii infection in different anatomical regions (liver, spleen, brain, plasma), cell types (primary human foreskin fibroblasts, neuroepitheliomal cells, peripheral blood mononuclear cell-derived macrophages), and mammalian species (mouse, human, cat, pig). Several different parasite strains and infectious stages of T. gondii have also been employed. These studies are summarized in **Table 1**. Here, we highlight some key findings as well as underscore some of the commonalities between the studies.

#### miR-17-92 Gene Cluster

The first study profiling microRNAs differentially expressed during T. gondii infection was published almost a decade ago (Zeiner et al., 2010). It reported that 14% of the miRNAs on the microarray were differentially regulated, including members of the miR-17-92 and miR-106b-25 family that are upregulated after infection with RH in primary human foreskin fibroblasts (HFFs). Other studies have confirmed this finding in other cell types. The miR-17-92 polycistronic gene cluster was also found to be upregulated in human macrophages derived from peripheral blood mononuclear cells after infection with TgCtwh3 strain parasites (Cai et al., 2013, 2014). He et al. found that mir-17- 5p (a member of miR-17-92 gene cluster) is upregulated in the mouse spleen after infection with the RH strain (He et al., 2016). Nevertheless, in an initial analysis knockdown of miR-17-92 in HFF had no observable effect on infection, parasite replication or host cell lysis (Zeiner et al., 2010).

Despite lack of an overt effect of miR-17-92 on the intracellular growth cycle of T. gondii, more recent work suggests a connection of this miRNA cluster with inhibition of host cell apoptosis. Blocking apoptosis is a well-documented survival strategy of Toxoplasma in the host cell (Lüder and Gross, 2005; Carmen and Sinai, 2007). Knocking down expression of miR-20a (a member of the miR-17-92 cluster) in infected macrophages resulted in increased sensitivity to apoptosis (Cai et al., 2013; Rezaei et al., 2018). The model proposed for how Toxoplasma induces miR-17-92 to inhibit apoptosis revolves around the parasite secretory kinase ROP16 (**Figure 2**). This protein is injected into the host cell cytoplasm where it tyrosine phosphorylates STAT3 (Saeij et al., 2007; Yamamoto et al., 2009). The miR-17- 92 cluster contains STAT3 binding sites in its promoter and is therefore upregulated by STAT3 phosphorylation. Evidence for this comes from the observation that knockdown of STAT3 using siRNA prevents miR-17-92 upregulation. Furthermore, luciferase constructs with the binding site for STAT3 in the promoter region of miR-17-92 led to increased luciferase expression (Cai et al., 2013). STAT3-induced miR-17-92 then binds to the 3′ UTR of the Bim transcript, reducing BIM levels (Cai et al., 2014). The BIM protein is pro-apoptotic and, therefore, deficiency leads to the inhibition of apoptosis observed in host cells following infection with T. gondii.

#### miR-132

The miRNA miR-132 has both neural and immune functions and dysregulation has been associated with numerous neurological disorders (Soreq and Wolf, 2011; Miller et al., 2012; Wanet et al., 2012). Xiao et al. surveyed miRNA expression profiles in neural cells (SK-N-MC cells) acutely infected with Type I, Type II, and Type III Toxoplasma strains (Xiao et al., 2014). They found that miR-132 was the only miRNA upregulated by >2-fold by all three T. gondii strains, and this upregulation was confirmed in the peritoneal cavity and striatum of infected mice. Among the predicted targets of miR-132, the strongest pathway enriched was that for dopamine signaling. Three genes (Drd1, Drd5, and Maoa) in the dopamine signaling pathway displayed decreased transcription and protein expression in T. gondii-infected mice. HPLC data also demonstrated increased production of dopamine, serotonin, and 5-hydroxyindoleacetic acid in the striatum of Toxoplasma-infected mice. miR-132 is known to be upregulated by LPS and several viruses (Taganov et al., 2006; Lagos et al., 2010). These combined results suggest that miR-132 may be a common target of a broad range of pathogens and may represent a general response to infection. It is also worth noting that other researchers have reported dysregulation of dopamine pathways during Toxoplasma infection, suggesting possible links to nervous system abnormalities (Syn et al., 2018; Alsaady et al., 2019).

In a recent study, expression of miR-132 in the mouse brain during chronic (5 month) infection with PRU strain parasites was investigated (Li et al., 2015). Contrary to acute infection, chronic infection resulted in decreased expression of miR-132 relative to non-infected mice. Correlation between levels of miR-132 in different brain regions and the number of parasites was weak, suggesting that the effects of Toxoplasma on miRNA expression were indirect.

#### miR-146a and miR155

Cannella et al. surveyed microRNAs differentially expressed between T. gondii high and low virulence strains to determine if differences in host pathogenesis correlated with specific microRNAs (Cannella et al., 2014). This resulted in identification TABLE 1 | Studies surveying host microRNAs and lncRNAs differentially expressed during T. gondii infection.


<sup>a</sup>Tachyzoites used unless otherwise stated.

<sup>b</sup>Functional data.

<sup>c</sup>Human foreskin fibroblasts.

<sup>d</sup>Peripheral blood mononuclear cells.

<sup>e</sup>Human neuroepithelial cell line.

<sup>f</sup>S-NSC are adult human neural stem/progenitor cells; S-NDC are a differentiated form of S-NSC; MM6 are in vitro culture-adapted human monocytes.

<sup>g</sup>alveolar macrophages.

of miR-146a and miR155. These investigators found that miR-146a was induced by Type II but not by Type I tachyzoites in several non-hematopoietic cell types. In addition, levels of miR-146a increased in the central nervous system of mice chronically infected with Type II Toxoplasma and were correlated with the presence of cysts. An independent study also found miR-146a was induced in the brain during chronic infection initiated with PRU oocysts (Hu et al., 2018).

In contrast to Type II T. gondii, Type I Toxoplasma strains were found to repress miR-146a levels. Furthermore, when ROP16 was deleted in Type I strains, expression of miR-146a was greatly enhanced. ROP16 deletion in Type II strains had no effect on miR-146a expression as predicted given the lack of kinase activity of this rhoptry protein. Interestingly, miR-146a−/<sup>−</sup> mice displayed increased resistance to Type II T. gondii (Cannella et al., 2014). The lower levels of IFN-γ in miR-146a knockout mice suggests that increased resistance results from avoidance of cytokine-induced inflammation that can lead to pathology and death when not appropriately controlled (Gazzinelli et al., 1996; Gavrilescu and Denkers, 2001; Mordue et al., 2001).

The miRNA miR-155 is known to target SOCS1 mRNA to promote inflammation (Rao et al., 2014). Cannella et al. reported that miR-155 was induced after T. gondii infection (Cannella et al., 2014). Induction was strain-independent during in vitro infection of stromal and phagocytic cells. In vivo, miR-155 was induced in the central nervous system of mice infected with Type II Toxoplasma strains. Another study found that miR-155 was induced in mouse spleens after infection with RH tachyzoites (He et al., 2016). However, it is worth noting that in a study in pigs, miR-155 was downregulated during chronic infection (Hou et al., 2019). The reason for this conflicting result is not clear and could be due to host species or T. gondii strain differences. Nevertheless, taken together the data indicate that both miR-146a and miR-155 are triggered by Toxoplasma. At least for the case of miR-146a, this appears to impact the outcome of infection.

#### MicroRNAs and Exosomes

Parasites exploit exosomes to influence host responses during infection (Gavinho et al., 2018; Wu et al., 2018; Atayde et al., 2019). It is, therefore, of interest that HFF infected with T. gondii release exosome-like vesicles containing several distinct miRNA species (Pope and Lässer, 2013). Thus, 10 miRNAs (miR-92a, miR-595, miR125b, miR-199a-3p, miR-125a-5p, miR-503, miR-320d, miR-1183, miR-99a-star, miR-23b) were identified within exosomes isolated from infected cells. The miRNA miR-23b is particularly of interest because it is a known negative regulator of IL-17 synthesis (Hu and O'Connell, 2012; Zhu et al., 2012). A more recent study identified as many as 64 distinct microRNAs that were present within exosomes derived from cells infected with RH vs. exosomes derived from uninfected cells (Kim et al., 2016). Additionally, the Cannella et al. (2014) study identified expression of a specific miRNA, miR-146a, in exosomes of Type II infected cells. Together, these data suggest that nearby uninfected cells (and not just infected cells) experience altered microRNA expression during infection. The influence of miRNA on exosome infection biology is likely to be a fruitful area of investigation in the near future.

## Host lncRNAs in the Response to *Toxoplasma*

To date, we know very little regarding expression of host long non-coding RNAs during Toxoplasma infection. Here, we highlight two studies that provide some interesting first glances at regulation of this class of RNA during T. gondii infection. While these studies are intriguing, direct comparison of the results is complicated by the small degree of sequence conservation between lncRNAs of different species (Spurlock et al., 2016).

#### lncRNA Responses in Human Cells

One group of investigators examined lncRNAs differentially regulated in HFF cell lines during infection with Type II (ME49) T. gondii (Liu et al., 2018). They found that 1,206 lncRNAs were upregulated after infection, including 996 that were only induced by active infection relative to heat-inactivated parasites. Interestingly, inactivated parasites upregulated 392 lncRNAs when compared to uninfected cells, demonstrating that simply the presence of dead parasites can elicit effects on lncRNA expression. It was found that one particular lncRNA, NONHSAT022487, is highly upregulated after Type II infection. Knock down of NONHSAT022487 resulted in increased expression of cytokines IL-12, TNF-α, IL-1β, and IFNγ in T. gondii-infected THP-1 monocytic cells. Overexpression of NONHSAT022487 resulted in decreased expression of these four cytokines. This lncRNA would therefore appear to be a prime candidate to further explore how it exerts its effects on proinflammatory cytokine gene expression.

#### lncRNAs Regulated in Mouse Cells

We surveyed the expression of lncRNA in mouse bone marrowderived macrophages during infection with both Type I (RH) and Type II (PTG) strains (Menard et al., 2018). We found that hundreds of putative lncRNAs were differentially regulated, and a substantially greater number were differentially regulated with RH rather than PTG infection (1,522 and 528, respectively). The total number of lncRNAs upregulated by infection was

#### REFERENCES


similar to the number down-regulated. Interestingly, we also found that Toxoplasma likely directly controls expression of host lncRNAs, as upregulation of two lncRNAs was ablated in parasite strains deleted for the STAT3/6-directed kinase ROP16. Functional studies are now required to identify biologically important lncRNAs in the response to infection.

# CONCLUSIONS

We have only a nascent understanding of the role of noncoding RNA in host defense against infection. However, we now know that there is a large number of miRNA that are up or downregulated during T. gondii infection, including the miR-17-92 gene cluster, miR-132, miR-146a, miR-155, and miR-23b. At least some of these miRNAs play important roles in the response to T. gondii. Clearly, there are many miRNA yet to be characterized, and it is likely that some will prove to significantly impact host defense largely in ways yet to be discovered. Our knowledge of lncRNA responses to Toxoplasma is even more rudimentary. However, the finding that thousands of lncRNAs that are up or downregulated during T. gondii infection indicates that host lncRNA responses are extensive, and most likely biologically important. Future work is required to determine those regulatory RNAs that are paramount in the response to infection with Toxoplasma and other microbial pathogens.

#### AUTHOR CONTRIBUTIONS

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

#### FUNDING

Our lncRNA work was supported by the National Institute of General Medical Sciences of the National Institutes of Health under award number P30 GM110907.

and sustained host p38 MAPK activation. J. Exp. Med. 210, 2071–2086. doi: 10.1084/jem.20130103


suppression of the p300 transcriptional co-activator. Nat. Cell Biol. 12, 513–519. doi: 10.1038/ncb2054


novel Toxoplasma gondii dense granule protein. J. Exp. Med. 208, 195–212. doi: 10.1084/jem.20100717


has implications for dopamine signaling pathway. Neuroscience 268, 128–138. doi: 10.1016/j.neuroscience.2014.03.015


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

Copyright © 2019 Menard, Haskins and Denkers. 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.

# Immunomodulatory Effects of the Neuropeptide Pituitary Adenylate Cyclase-Activating Polypeptide in Acute *Toxoplasmosis*

#### Caio Andreeta Figueiredo<sup>1</sup> , Henning Peter Düsedau<sup>1</sup> , Johannes Steffen<sup>1</sup> , Nishith Gupta<sup>2</sup> , Miklos Pal Dunay <sup>3</sup> , Gabor K. Toth<sup>4</sup> , Dora Reglodi <sup>5</sup> , Markus M. Heimesaat <sup>6</sup> and Ildiko Rita Dunay 1,7 \*

*Edited by:*

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### *Reviewed by:*

Gaoqian Feng, Burnet Institute, Australia Dumith Chequer Bou-Habib, Oswaldo Cruz Foundation (Fiocruz), Brazil

> *\*Correspondence:* Ildiko Rita Dunay ildikodunay@gmail.com

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

> *Received:* 31 January 2019 *Accepted:* 26 April 2019 *Published:* 28 May 2019

#### *Citation:*

Figueiredo CA, Düsedau HP, Steffen J, Gupta N, Dunay MP, Toth GK, Reglodi D, Heimesaat MM and Dunay IR (2019) Immunomodulatory Effects of the Neuropeptide Pituitary Adenylate Cyclase-Activating Polypeptide in Acute Toxoplasmosis. Front. Cell. Infect. Microbiol. 9:154. doi: 10.3389/fcimb.2019.00154 <sup>1</sup> Medical Faculty, Institute of Inflammation and Neurodegeneration, Otto-von-Guericke University Magdeburg, Magdeburg, Germany, <sup>2</sup> Faculty of Life Sciences, Institute of Biology, Humboldt University, Berlin, Germany, <sup>3</sup> Department and Clinic of Surgery and Ophthalmology, University of Veterinary Medicine, Budapest, Hungary, <sup>4</sup> Department of Medical Chemistry, University of Szeged, Szeged, Hungary, <sup>5</sup> Department of Anatomy, MTA-PTE PACAP Research Team, University of Pecs Medical School, Pecs, Hungary, <sup>6</sup> Department of Microbiology and Hygiene, Charité - University Medicine Berlin, Berlin, Germany, <sup>7</sup> Center for Behavioral Brain Sciences - CBBS, Magdeburg, Germany

Pituitary Adenylate Cyclase-Activating Polypeptide (PACAP) is an endogenous neuropeptide with distinct functions including the regulation of inflammatory processes. PACAP is able to modify the immune response by directly regulating macrophages and monocytes inhibiting the production of inflammatory cytokines, chemokines and free radicals. Here, we analyzed the effect of exogenous PACAP on peripheral immune cell subsets upon acute infection with the parasite Toxoplasma gondii (T. gondii). PACAP administration was followed by diminished innate immune cell recruitment to the peritoneal cavity of T. gondii-infected mice. PACAP did not directly interfere with parasite replication, instead, indirectly reduced parasite burden in mononuclear cell populations by enhancing their phagocytic capacity. Although proinflammatory cytokine levels were attenuated in the periphery upon PACAP treatment, interleukin (IL)-10 and Transforming growth factor beta (TGF-β) remained stable. While PACAP modulated VPAC1 and VPAC2 receptors in immune cells upon binding, it also increased their expression of brain-derived neurotrophic factor (BDNF). In addition, the expression of p75 neurotrophin receptor (p75NTR) on Ly6Chi inflammatory monocytes was diminished upon PACAP administration. Our findings highlight the immunomodulatory effect of PACAP on peripheral immune cell subsets during acute Toxoplasmosis, providing new insights about host-pathogen interaction and the effects of neuropeptides during inflammation.

Keywords: pituitary adenylate cyclase-activating polypeptide (PACAP), *Toxoplasma gondii*, acute infection, monocytes, macrophages, innate immunity, neurotrophins

# INTRODUCTION

Pituitary adenylate cyclase-activating polypeptide (PACAP) is a 38-amino-acid neuropeptide in the glucagon superfamily together with secretin and vasoactive intestinal peptide (VIP) (Sherwood et al., 2000). PACAP is widely expressed in the peripheral and central nervous systems (CNS) and functions as a neurotransmitter, neuromodulator and neurotrophic factor (Waschek, 2002; Zhou et al., 2002; Dejda et al., 2005; Botia et al., 2007; Armstrong et al., 2008; Abad and Tan, 2018). The nervous and immune systems participate in a complex bidirectional crosstalk through neuropeptides such as PACAP/VIP in multiple organs (Abad and Tan, 2018). Correspondingly, PACAP strongly improves the outcome of inflammatory disorders, such as rheumatoid arthritis, septic shock, inflammatory bowel disease, and multiple sclerosis in rodent models (Abad et al., 2001, 2003; Martinez et al., 2002; Gonzalez-Rey et al., 2006; Tan et al., 2009).

PACAP acts by binding to three specific membrane receptors from the G protein-coupled receptors (GPCR) family: PAC1, VPAC1, and VPAC2. The PAC1 receptor is highly expressed in the nervous system and possesses the highest affinity for PACAP, while the receptors VPAC1 and VPAC2 have the same lower affinity for PACAP, and are expressed among different cell types (Pozo et al., 1997). The majority of immune cells express one or more PACAP receptors. For example, PAC1 is expressed on peritoneal macrophages, microglia and pulmonary dendritic cells (DCs) (Delgado et al., 2004a). VPAC1 is constitutively expressed in T cells, macrophages, monocytes and DCs (Delgado et al., 2004a). VPAC2 is rarely expressed in these cells during a resting state, but its expression is induced following lipopolysaccharide (LPS) stimulation in vitro (Delgado et al., 1996b). Recently, the expression of VPAC1 and VPAC2 was found in innate lymphoid cells (ILC) 2 (Nussbaum et al., 2013), and has been implicated in the resolution of inflammation (Talbot et al., 2015).

Many studies have demonstrated that PACAP acts as a neuronal growth factor during development and regeneration (Waschek, 2002; Deguil et al., 2007; Watanabe et al., 2007). The neuromodulatory properties of PACAP were shown to be involved with the neurotrophin signaling of brainderived neurotrophic factor (BDNF), promoting neuronal survival and synaptic plasticity (Frechilla et al., 2001). The family of neurotrophins is comprised of four secreted proteins, characterized by their ability to modulate survival, differentiation, and apoptosis of neurons (Bothwell, 2016). BDNF, nerve growth factor (NGF), neurotrophin-3 (NT-3), and neurotrophin-4 (NT-4) exert their functions via interaction with tropomyosin receptor kinases (Trk) TrkA, TrkB, TrkC, and the p75 neurotrophin receptor (p75NTR). The effect of PACAP on neurotrophin signaling was presented in human monocytes, where the exposure to the neuropeptide resulted in proinflammatory cell activation with increased Ca2<sup>+</sup> mobilization (El Zein et al., 2006, 2007, 2008, 2010).

As an immunomodulator, PACAP exerts a dual role in regulating innate immunity depending on the activation status of cells and their environment. Several studies reported that PACAP is a potent immunomediator for both innate and adaptive immunity, primarily assuming an anti-inflammatory role. Exposure to PACAP inhibits the pro-inflammatory response of macrophages, such as the production of tumor necrosis factor (TNF) (Delgado et al., 1999b) and interleukin 6 (IL-6) (Martinez et al., 1998a), as well as the chemokines MCP-1 (CCL2), MIP1-α (CCL3) and RANTES (Delgado et al., 2003). Additionally, PACAP treatment leads to polarization of T helper cells to a Type 2 (Th2) phenotype (Delgado, 2003). It also promotes the development of tolerogenic DCs and favors the generation of regulatory T cells (Tregs), suppressors of immune responses (Delgado et al., 2005). In contrast, PACAP was able to stimulate the phagocytic activity, adhesion and mobility of resting macrophages as well as release of free radicals and IL-6 (Delgado et al., 1996a; Garrido et al., 1996; Martinez et al., 1998b), associating PACAP with a crucial mechanism to pathogen elimination.

Antiparasitic effects of PACAP were first described against the protozoa Trypanosoma brucei, showing a membranelytic effect, closely associated with autophagy and apoptosislike cell death (Delgado et al., 2009). More recently, we showed that administration of PACAP ameliorated acute small intestinal inflammation and extra-intestinal sequelae caused by Toxoplasma gondii (T. gondii) infection (Heimesaat et al., 2014; Bereswill et al., 2019). Toxoplasma gondii is an obligate intracellular parasite acquired by oral ingestion of contaminated food or water. The parasite infects the small intestine and then differentiates to its rapidly replicating stage (tachyzoite), which is able to infect all nucleated cells through active penetration (Dobrowolski and Sibley, 1996). After crossing the intestinal barrier, the parasites encounter both resident and recruited immune cells, resulting in parasite elimination, antigen presentation and cytokine production (Buzoni-Gatel et al., 2006).

The successful dissemination of T. gondii within the host is highly dependent on invading migratory immune cells and the ability of these immune cells to phagocyte the parasite. Elimination of T. gondii involves a complex recruitment of immunity-related GTPases (IRGs) and guanylate-binding proteins (GBPs) (Zhao et al., 2009; Fentress et al., 2010). After infection, these host defense factors are known to accumulate in the membrane of the parasitophorous vacuole (PV) culminating with its disruption and parasite elimination (Macmicking, 2012). Moreover, a particular preference for myeloid cells has been explored as a key mechanism for parasite dissemination into the CNS (Weidner and Barragan, 2014; Blanchard et al., 2015).

Our group has previously demonstrated the critical importance of myeloid cells controlling T. gondii infection in the periphery as well as in the CNS (Dunay et al., 2008, 2010; Biswas et al., 2015; Mohle et al., 2016). Besides, we recently showed the innate immune response and the influence of neurotrophin signaling upon T. gondii-induced neuroinflammation. Particularly, neurotrophin signaling via p75NTR altered innate immune cell behavior and changed the structural plasticity of neurons (Dusedau et al., 2019).

Here, we set out to evaluate the immunomodulatory effects of PACAP on the innate immune response during T. gondii acute infection. We show that PACAP is able to reduce immune cell recruitment and enhance phagocytic capacity of mononuclear cells, promoting parasite elimination. At the same time, the neuropeptide attenuated pro-inflammatory mediators while upregulating its own receptors. Interestingly, we detected altered expression of BDNF and p75NTR in peritoneal cells, pointing toward the contribution of PACAP to the parasite elimination and neurotrophin signaling in immune cells upon acute Toxoplasmosis.

### MATERIALS AND METHODS

#### Animals

Experiments were conducted with female C57BL/6JRj mice (8 weeks old, purchased from Janvier, Cedex, France). All animals were group-housed in a 12 h day/night cycle at 22◦C with free access to food and water under specific-pathogen-free conditions, according to institutional guidelines approved by the Animal Studies Committee of Saxony-Anhalt.

# *T. gondii in vitro* Culture

Tachyzoites of ME49 and PTG-GFP type II strain of T. gondii were grown in monolayers of human foreskin fibroblast (HFF) cells cultured in DMEM medium (FG0435, Biochrom, Germany), supplemented with 10% fetal bovine serum (FBS) (Thermo Fisher, Germany), 1% Penicillin/Streptomycin (Pen/Strep; Sigma, USA) and 1% non-essential amino acids (NEEA) (Thermo Fisher, Germany) (Morisaki et al., 1995). HFF cells and tachyzoites were scrapped from culture flasks, spinned down at 500 × g for 10 min and passed through 20 and 22G needles to liberate intracellular parasites. To obtain a host cell-free parasite suspension, the solution was filtered through a 5µm Millex-SV syringe filter (Millipore, Germany). The parasite suspension was pelleted at 800 × g for 20 min, resuspended in 1 ml sterile phosphatebuffered saline (PBS) and the number of living tachyzoites was determined by counting under a light microscope using Trypan Blue 0.4%. Subsequently, the freshly egressed parasite suspension was used for infections and plaque assay experiments.

#### Experimental Acute *T. gondii* Infection and PACAP Administration

In order to investigate acute T. gondii infection in the peritoneal cavity, all mice were infected by intraperitoneal (i.p.) injection of 1 × 10<sup>4</sup> ME49 or PTG-GFP tachyzoites, freshly harvested from HFF cultures, in a final volume of 200 µl with PBS. For PACAP treatment, 50 µg (11 nmol/mouse) of the synthesized neuropeptide PACAP38 dissolved in 200 µl of PBS was administrated i.p. on days 2 and 4 post infection. Non-treated control mice received PBS only. At day 5 post infection (dpi) peritoneal exudate cells were collected by peritoneal lavage (Fentress and Sibley, 2011) for further analysis. Spleens were collected and stored in Allprotect Tissue Reagent (Qiagen, Germany) at −80◦C until further processing.

### Flow Cytometric Analysis

Single cell suspensions were first incubated with ZOMBIE NIRTM fixable dye (BioLegend, San Diego, CA) or 7-AAD Viability Staining Solution (BioLegend) for live/dead discrimination. To prevent unspecific binding of antibodies, anti-FcγIII/II receptor antibody (clone 93) was applied to cells before staining with fluorochrome-conjugated antibodies against cell surface markers in FACS buffer (PBS, supplemented with 2% FBS and 0.1% sodium azide). CD11b (M1/70), Ly6C (HK1.4), MHCII I-A/I-E (M5/114.15.2), and F4/80 (BM8) were all purchased from eBioscience (San Diego, USA). Ly6G (1A8) and CD11c (N418) were purchased from BioLegend and p75NTR (MLR2) from Abcam (Germany). Cells were incubated for 30 min at 4 ◦C, washed and subsequently analyzed. Fluorescence Minus One (FMO) controls were used to determine the level of autofluorescent signals for each conjugated antibody. Data was acquired using a BD FACS Canto II (BD Biosciences, USA) or Attune NxT flow cytometer (Thermo Fisher, Germany) and analyzed using FlowJo (v10, FlowJo Inc., USA). A minimum of 2 × 10<sup>5</sup> cells per samples were acquired.

#### Plaque Assay

All experiments were conducted using fresh syringe-released extracellular tachyzoites. As previously described (Arroyo-Olarte et al., 2015), 200 parasites per well were used to infect HFF monolayers in six-well plates at different concentrations of PACAP. Briefly, parasitized cells were incubated for 7 days, fixed with cold methanol, and then stained with crystal violet. Plaques were imaged and scored for their sizes and numbers using the ImageJ software (NIH, US).

### Generation of Bone Marrow-Derived Macrophages

For generation of bone marrow-derived macrophages (BMDMs), femurs and tibias of 8 to 12 weeks old C57BL/6JRj mice were collected. Bones were flushed with a syringe filled with DMEM (FG0435, Biochrom, Germany) containing 10% FBS and 1% Pen/Strep to extrude bone marrow onto a 40µm cell strainer. The obtained cell suspension was then spinned down for 10 min at 400 × g, 4◦C and the cells were seeded into 6 well/plate using DMEM supplemented with 10% FBS, 1% Pen/Strep, 10 ng/ml recombinant murine GM-CSF (315-03, PeproTech, USA) and incubated at 37◦C, 5% CO2. After 10 days, cells were primed to M1 macrophage phenotype with 150 Units/ml recombinant murine IFN-γ (315-05, Peprotech, USA) for 10 h, and then with 20 ng/ml of LPS (L2630, Sigma-Aldrich, USA) for another 12 h. Following stimulation, cells were washed and subsequently used for phagocytosis assay.

#### *In vitro* Phagocytosis Assay

Phagocytosis assay was performed with M1 macrophages generated as described above and assessed in triplicates by incubation of cells with carboxylated, yellow-green fluorescent FluoSpheresTM (F8823, Fisher Scientific, Germany) in serumfree DMEM (with 1% Pen/Strep) in the presence of PACAP (0.1, 1, and 10µM) for 4 h at 37◦C, 5% CO2. Negative controls were established by keeping cells at 4◦C throughout the experiment. After incubation, cells were washed twice with PBS to remove remaining microspheres before detachment with Accutase <sup>R</sup> solution (423201, BioLegend). Finally, detached cells were washed with PBS and directly stained with fluorochromeconjugated antibodies for flow cytometric analysis. In vitro phagocytosis of microspheres was assessed by the median fluorescence intensity (MFI) of the positive population in the FITC-fluorescence channel. Further, percentages of cells in the FITC-positive fraction where divided according to the amount of microspheres internalized.

#### DNA and RNA Isolation

DNA and RNA samples were isolated from peritoneal exudate cells and spleens of acutely infected mice. Spleen samples were homogenized in lysis buffer using BashingBeads Lysis tubes (Zymo Research, Germany) and isolated using AllPrep DNA/RNA Mini Kit (Qiagen, Germany) according to the manufacturer's instructions. For peritoneal exudate cells, parts of the cell suspensions were pelleted down, resuspended in lysis buffer and processed as describe above. The concentration and purity of DNA and RNA samples was determined using NanoDrop 2000 spectrophotometer (Thermo Fisher; Germany).

#### qPCR

Parasite burden was assessed in triplicates using 30 ng of isolated DNA, FastStart Essential DNA Green Master and LightCycler <sup>R</sup> 96 System (both Roche, Germany), as described previously (Biswas et al., 2017). Thermal-cycling parameters were set as follows: initial activation (95◦C, 10 min), 45 amplification cycles consisting of denaturation (95◦C, 15 s), annealing (60◦C, 15 s) and elongation (72◦C, 15 s). The DNA target was the published sequence of the highly conserved 35-fold-repetitive B1 gene of T. gondii (Burg et al., 1989; Lin et al., 2000). Murine argininosuccinate lyase (Asl) was used as reference gene for normalization and relative DNA levels were determined by the ratio gene of interest / reference gene and subsequently normalized to mean values of control group (Butcher et al., 2011). Primers were synthetized by Tib MolBiol (Germany) and used at 300 nM final concentration.

#### RT-qPCR

Expression levels of cytokines, inflammatory mediators, hostdefense factors, neurotrophins, and neurotrophin receptors were assessed in triplicates using 30 ng isolated RNA, TaqMan <sup>R</sup> RNA-to-C<sup>T</sup> TM 1-Step Kit (Applied Biosystems, Germany) and LightCycler <sup>R</sup> 96 (Roche, Germany) as previously described (Mohle et al., 2016). Thermal-cycling parameters were set as follows: reverse transcription (48◦C, 15 min), inactivation (95◦C, 10 min) followed by 45 cycles of denaturation (95◦C, 15 s) and annealing/extension (60◦C, 1 min). Utilized TaqMan <sup>R</sup> Gene Expression Assays (Applied Biosystems, Germany) are listed in **Supplementary Table 1**. Hprt was chosen as a reference gene and relative mRNA levels were determined by the ratio gene of interest/reference gene and subsequently normalized to mean values of control group.

The expression of PACAP receptors were evaluated using Power SYBR <sup>R</sup> Green RNA-to-CTTM 1-Step Kit (Applied Biosystems, Germany). Samples were analyzed in triplicates (30 ng of isolated mRNA per reaction) using LightCycler <sup>R</sup> 96 and the following parameters: reverse transcription (48◦C, 30 min), inactivation (95◦C, 10 min) followed by 55 cycles of denaturation (95◦C, 15 s) and annealing/extension (60◦C, 1 min) and melting curve analysis. The primer sequences are listed in **Supplementary Table 1** and were synthetized by Tib MolBiol and used at 100 nM final concentration. Expression of Hprt was chosen as reference gene and relative mRNA levels were determined by the ratio gene of interest / reference gene and subsequently normalized to mean values of control group.

#### Statistical Analysis

Results were statistically analyzed using GraphPad Prism 7 (STATCON, Germany) and two-tailed unpaired t-test was used on flow cytometry, qPCR and RT-qPCR data, and considered significant for p ≤ 0.05. Statistical analysis of phagocytosis assay data was carried out by applying one-way ANOVA with post-hoc Holm-Sidak test. For plaque assay, data was analyzed by one-way ANOVA followed by post-hoc Bonferroni test. All data are presented as arithmetic mean ± standard error of the mean (SEM) and are representative of two to three independent experiments.

#### RESULTS

#### Immune Cell Recruitment Is Reduced Following PACAP Administration

Upon acute infection with T. gondii, neutrophil granulocytes, inflammatory monocytes and DCs are recruited to the site of infection (Robben et al., 2005; Dunay et al., 2008; Dunay and Sibley, 2010). To assess the effect of PACAP on cell recruitment and activation upon acute infection, mice were infected with tachyzoites, followed by administration of PACAP or PBS (control). The peritoneal exudate cells were collected and characterized by flow cytometry (**Figure 1A**). As our previous studies demonstrated the critical importance of myeloid cells in the control of T. gondii infection (Dunay et al., 2010; Biswas et al., 2015; Mohle et al., 2016), we analyzed the mononuclear compartment based on the expression of CD11b and Ly6G. While the CD11b+Ly6G<sup>+</sup> subset defined neutrophil granulocytes, the fraction of CD11b+Ly6G<sup>−</sup> cells was further discriminated into CD11chiMHCIIhi DCs as previously described (Dupont et al., 2014). Subsequently, remaining immune cells were then defined according to Ly6C expression as Ly6Chi inflammatory monocytes and Ly6C<sup>−</sup> peritoneal macrophages.

In general, the PACAP-treated group presented less recruited cells in the peritoneal cavity (control: 1.67 × 10<sup>5</sup> ± 0.10 × 10<sup>5</sup> vs. PACAP: 0.95 × 10<sup>5</sup> ± 0.07 × 10<sup>5</sup> ; p = 0.0012) (**Figure 1B**). When compared to the controls, administration of PACAP significantly reduced the recruitment of all analyzed myeloid cell subsets. Ly6Chi inflammatory monocytes appeared to be the most affected cell population (control: 4.34 × 10<sup>4</sup> ± 0.10 × 10<sup>3</sup> vs. PACAP: 2.99 × 10<sup>4</sup> ± 2.41 × 10<sup>3</sup> ; p = 0.00003), followed by Ly6C<sup>−</sup> (control: 3.19 × 10<sup>4</sup> ± 2.75 × 10<sup>3</sup> vs. PACAP: 2.08 × 10<sup>4</sup> ± 2.15 × 10<sup>3</sup> ; p = 0.0003) and neutrophils (control: 1.62 × 10<sup>4</sup> ± 2.09 × 10<sup>3</sup> vs. PACAP: 7.96 × 10<sup>3</sup> ± 7.7 × 10<sup>2</sup> ; p = 0.0043). No evident difference was found for DCs (control: 4.02 × 10<sup>3</sup> ± 6.10 × 10<sup>2</sup> vs. PACAP: 1.92 × 10<sup>3</sup> ± 2.29 × 10<sup>2</sup> ; p = 0.4301) (**Figure 1C**). As the ability to present antigens is dependent on MHCII expression, we evaluated whether this was modulated by PACAP treatment upon acute T. gondii infection. Our data indicated that PACAP

was able to increase the expression of MHCII on peritoneal DCs (control: 2.74 × 10<sup>5</sup> ± 5.58 × 10<sup>3</sup> vs. PACAP: 3.06 × 10<sup>5</sup> ± 8.25 × 10<sup>3</sup> ; p = 0.0012) (**Figure 1D**) but not on the other myeloid subsets (Ly6C−, control: 5.36 × 10<sup>4</sup> ± 3.23 × 10<sup>3</sup> vs. PACAP: 6.35 × 10<sup>4</sup> ± 1.92 × 10<sup>3</sup> ; p = 0.2448; Ly6Chi, control: 1.11 × 10<sup>5</sup> ± 4.91 × 10<sup>3</sup> vs. PACAP: 1.15 × 10<sup>5</sup> ± 8.21 × 10<sup>3</sup> ; p = 0.6767). Thus, PACAP reduced the recruitment of mononuclear cells to the peritoneal cavity upon T. gondii infection and increased MHCII expression on peritoneal DCs.

#### Antiparasitic Effect of PACAP Is Immune Cell-Mediated

Infected migratory immune cells, such as DCs, monocytes and macrophages are responsible for the parasite dissemination throughout host tissues, including lung, spleen, and also the peritoneal cavity (Ueno et al., 2014). As previous reports demonstrated the anti-parasitic effect of PACAP, we set out to analyze whether PACAP administration is able to affect the presence of T. gondii in myeloid-derived cell subsets. By infecting mice with a GFP-fluorescent reporter parasite, we were able to elucidate T. gondii occurrence in cell subsets isolated from the peritoneal cavity via flow cytometry. Here, the experimental group receiving PACAP showed a marked reduction of infected cells in all myeloid populations (**Figures 2A**′**–C**′ ). Alongside with Ly6G<sup>+</sup> neutrophil granulocytes, Ly6Chi monocytes represented the cell subset with the strongest reduction of parasitic GFP signal, while DCs and Ly6C<sup>−</sup> macrophages were less affected (neutrophils, control: 32 ± 5.6% vs. PACAP: 6.2 ± 0.34%; p = 0.027; Ly6Chi, control: 26 ± 3% vs. PACAP: 5.5 ± 0.3%; p = 0.0004; Ly6C−, control: 15 ± 1.5% vs. PACAP: 3.4 ± 0.19%; p = 0.0003; DCs, control: 21 ± 2.5% vs. PACAP: 3.6 ± 0.65%; p = 0.0006) (**Figures 2A–C**). In order to address if this was mediated by the anti-parasitic effect of PACAP, we first performed an in vitro plaque assay using HFF culture infected with T. gondii in the presence of different PACAP concentrations (0.1, 1, and 10µM) (**Figure 2D**). In contrast to our in vivo data in the peritoneal cavity, these results revealed that PACAP did not directly affect the size (**Figure 2D**′ ) or the number of plaques (**Figure 2D**′′). Thus, PACAP has no direct effect on parasite elimination and/or impairment of the parasite development.

As infections with intracellular pathogens promote the expression of host defense factors such as IRGs and GBPs in myeloid cells, we hypothesized that these pathways were modulated by PACAP possibly explaining the reduced parasite burden in vivo. To this end, expression of IRGs (IRGM1, IRGM3) and GBP2b on peritoneal cells isolated from acutely

FIGURE 2 | Anti-parasitic effect of myeloid peritoneal immune cells upon PACAP-treatment. Peritoneal cells of acutely-infected mice were isolated and analyzed by flow cytometry. A GFP-fluorescent T. gondii reporter was used to track the presence of the parasite in myeloid peritoneal subsets selected as described above. (A–C) Contour plots show the presence of GFP<sup>+</sup> cells for each cell subset in control (left column) and PACAP-treated (right column) group. Numbers represent the mean percentage of parent population for each group from a representative experiment (n = 4). (A ′–C′ ) Bar charts compare the frequency of GFP<sup>+</sup> cells. Control (black bars) and PACAP-treated (white bars). (D–D′′) Representative images show in vitro replication of T. gondii by plaque assays in the presence of PACAP. Plaques from three independent experiments were scored for size and numbers using ImageJ. (D) Images of the formed plaques initially infected with 200 tachyzoites/well; numbers indicate different PACAP concentration. (D′ ) Scored number of plaques under different concentrations of PACAP. (D′′) Scored plaque sizes shown as arbitrary units (a.u.) in the presence of different PACAP concentrations. Data are expressed as mean ± SEM, \*\*p < 0.01, \*\*\*p < 0.001 (two-tailed unpaired t-test).

infected mice was assessed by RT-qPCR (**Figure 3A**). Both IRGs were upregulated in the group that received PACAP (IRGM1: p = 0.0329; IRGM3: p = 0.0220), suggesting that the observed decrease in parasite burden might be associated with an improved phagocytic ability, also reported to be modulated by PACAP (Delgado et al., 1996a). Therefore, we performed, a phagocytosis assay with bone marrow-derived macrophages (BMDMs) primed to a classically activated M1 phenotype and stimulated with different concentrations of PACAP (**Figure 3B**). When compared to the non-treated group all PACAP-treated groups showed an increased phagocytosis as displayed by the MFI, with the highest effect observed at 1µM concentration (control vs. 0.1µM PACAP: p < 0.0001; control vs. 1µM PACAP: p < 0.0001; control vs. 10µM PACAP: p < 0.0001). With the signal of FITC-fluorescent microspheres being well distinguishable by flow cytometry, we were further able to subdivide the composition of each experimental group based on the number of beads being phagocyted (**Figure 3B**′ ). In line with the MFI data, all PACAP-treated groups showed an increased number of internalized microspheres when compared to the non-treated control. Moreover, treatment with 1µM PACAP resulted in the largest fraction of BMDMs with more than 3 beads engulfed when compared to other groups (control: 68.3 ± 1.007 %; 0.1µM PACAP: 79.33 ± 0.9939; 1µM PACAP: 88.9 ± 1.0044 %; 10µM PACAP: 81.3 ± 0 %). These findings were further supported by analysis of BMDMs with respect to their expression of the macrophage marker F4/80 (**Figure 3C**). Also here, PACAP treatment resulted in upregulation of F4/80 that was most prominent in the group treated with 1µM PACAP (control vs. 0.1µM PACAP: p < 0.0045; control vs. 1µM PACAP: p < 0.0004; control vs. 10µM PACAP: p < 0.004). Altogether, these results demonstrate that the neuropeptide PACAP was able to reduce the presence of T. gondii in peritoneal myeloid cells not by a direct anti-parasitic effect but by modulating their phagocytic capabilities.

#### Decreased Parasite Burden and Expression of Inflammatory Mediators

Previous studies have shown that PACAP inhibits the production of pro-inflammatory cytokines (Martinez et al., 1998a; Delgado et al., 1999a,b). In order to evaluate how PACAP affects the acute inflammation caused by T. gondii, we assessed the parasite burden and the gene expression of inflammatory mediators (**Figure 4**). The results show a reduced parasite load (p = 0.0171) in the PACAP-treated group and reduced expression of IFN-γ (p = 0.0022), TNF (p = 0.0069), IL-6 (p = 0.0009), CCL-2 (p = 0.0009), and iNOS (p = 0.0451). Expression levels of IL-12, IFN-β, IL-10, and TGF-β did not differ between PACAP and the control group. In summary, PACAP was able to reduce parasite burden while diminishing the robust Th1 response characteristic for T. gondii infection without affecting anti-inflammatory mediators.

### Immune Cells Upregulate PACAP Receptors and Neurotrophin Expression

PACAP-mediated effects on immune cells are elicited by binding of the neuropeptide to its receptors PAC1, VPAC1, and VPAC2, whereby the main anti-inflammatory effect is primarily exerted through VPAC1 resulting in activation of the cAMP/PKA pathway (Delgado et al., 2004a). This signaling pathway regulates the activity of a range of transcription factors critical for expression of the most inflammatory mediators (Delgado et al., 1998, 1999a; Delgado and Ganea, 1999, 2001). Therefore, we evaluated whether PACAP modulates the expression of its intrinsic receptors on peritoneal immune cells upon T. gondii infection (**Figure 5A**). Our results show that the neuropeptide was able to increase the expression of VPAC1 and VPAC2 ∼3 fold when compared to the control group (VPAC1: p = 0.00001; VPAC2: p = 0.035). Thus, while PACAP binds to its own receptor, it also regulates their expression on innate immune cells.

Previous studies have shown the involvement of VPAC1 in myeloid cells and the PACAP-mediated signal transduction into neurotrophin signaling (El Zein et al., 2007). In general, neurotrophins are associated with the growth, development and survival of neuronal cells in the CNS (Mitre et al., 2017). However, besides neuronal tissue, neurotrophin receptors of the Trk superfamily and the p75 neurotrophin receptor (p75NTR) were detected in a variety of immune cells (Frossard et al., 2004; Fischer et al., 2008; Minnone et al., 2017), thus emphasizing the interdependency of the neuronal and immune system. In accordance with that, our previous work indicates that neurotrophin signaling via the p75NTR affects innate immune cell behavior upon T. gondii-induced neuroinflammation (Dusedau et al., 2019). Therefore, to further explore the influence of PACAP-mediated signaling on the neurotrophin signaling pathway as another modulator of the innate immune response, we analyzed the expression of BDNF and neurotrophin receptors (p75NTR, TrkA, TrkB, and TrkC) in peritoneal exudate cells upon acute infection (**Figure 5B**). Surprisingly, BDNF expression level was found to be elevated ∼6 fold in PACAP-treated animals when compared to the control group (BDNF, p = 0.0129; p75NTR , p = 0.0584; TrkA, p = 0.9990; TrkB, p = 0.9992; TrkC, p = 0.9674). Altogether, our results indicate an upregulation of BDNF expression upon PACAP treatment, suggesting a modulation of neurotrophin pathways in innate immune cells upon acute T. gondii infection.

# PACAP Reduce p75NTR Expression on Ly6Chi Monocytes

We have recently highlighted the upregulation of p75NTR on innate immune cells in the blood and brain of infected animals in T. gondii-induced neuroinflammation (Dusedau et al., 2019). In the periphery, the infection increased p75NTR expression on myeloid cell subsets, which are essential for control of toxoplasmosis (Dunay et al., 2010; Biswas et al., 2015). To this end, we analyzed expression of p75NTR on Ly6Chi inflammatory monocytes and Ly6C<sup>−</sup> resident macrophages isolated from the peritoneal cavity upon acute infection. In the PACAPtreated group, the p75NTR expression on Ly6Chi monocytes was significantly reduced (Ly6Chi , p < 0.00001; Ly6C−, p = 0.29) (**Figure 5C**) with no changes in frequency of p75NTR<sup>+</sup> cells for Ly6Chi (p = 0.276) or Ly6C<sup>−</sup> (p = 0.5181) (**Figure 5D**). Taken together, our results point toward the anti-inflammatory effect of

FITC-fluorescent microspheres in the presence of different PACAP concentrations (0.1, 1, and 10µM); phagocytic capability was evaluated by the MFI of the positive populations. (B') Histograms and bar charts show frequency cells fractioned according to the amount of microspheres internalized. (C) Histograms and bar charts show activation marker F4/80 expressed on BMDMS in the phagocytosis assay. Data are expressed as mean ± SEM (one-way ANOVA with post-hoc Holm-Sidak test).

PACAP on Ly6Chi monocytes, and suggest the involvement of p75NTR in the acute inflammatory response against T. gondii.

#### DISCUSSION

We hypothesized that application of the neuropeptide PACAP might modulate the behavior of myeloid-derived mononuclear cells and potentially contribute to the resolution of the infection and parasite elimination. Accordingly, we investigated the immunomodulatory effect of PACAP on innate immune cells isolated from the peritoneal cavity during experimental acute T. gondii infection. We detected an interaction of PACAP and neurotrophin signaling that suggests a contribution to the resolution of acute Toxoplasmosis.

Previously, we have reported that PACAP administration ameliorates acute small intestinal inflammation and extraintestinal sequelae during acute ileitis caused by T. gondii infection (Heimesaat et al., 2014; Bereswill et al., 2019). Besides, we demonstrated the critical importance of myeloid cells to control T. gondii infection in the periphery as well

as in the CNS (Dunay et al., 2008, 2010; Biswas et al., 2015; Mohle et al., 2016). Furthermore, our recent studies have revealed the emerging role of neurotrophin signaling via p75NTR that affects innate immune cell behavior and influences structural plasticity of neurons upon T. gondiiinduced neuroinflammation (Dusedau et al., 2019). Therefore, we analyzed the effects of PACAP on the course of acute T. gondii infection in mice.

Initially, our data revealed that in the peritoneal cavity, the number of recruited myeloid cells, especially Ly6Chi inflammatory monocytes, was reduced by the administration of exogenous PACAP. PACAP has been shown to modulate chemokines produced by activated macrophages and adhesion molecules expressed by granulocytes, thereby affecting the recruitment of different immune cell subsets (Ganea and Delgado, 2002; El Zein et al., 2008). Under steady-state conditions, two types of macrophages are found in the peritoneal cavity: large peritoneal macrophages (LPMs) and small peritoneal macrophages (SPMs) (Ghosn et al., 2010). During inflammation, the peritoneal cell composition dramatically changes, with a massive recruitment of Ly6Chi monocytes that give rise to new SPMs, while LPMs leave the peritoneal cavity and migrate to the omentum for antigen presentation (Cassado Ados et al., 2015). Upon in vitro LPS stimulation, SPMs developed a pro-inflammatory profile as indicated by TNF, CCL3 and RANTES production (Cain et al., 2013). These findings align with our previous results where Ly6Chi monocytes contributed to T. gondii removal via production of TNF, iNOS

and reactive oxygen species (ROS) (Dunay et al., 2008, 2010; Dunay and Sibley, 2010; Karlmark et al., 2012). Analyzing the gene expression of selected chemokines, cytokines and inflammatory mediators, we detected that exposure to PACAP diminished the levels of TNF, IL-6, and iNOS similarly to our previous observations in the intestine (Heimesaat et al., 2014). Moreover, the expression of the chemokine CCL2, important for the recruitment of Ly6Chi monocytes (Biswas et al., 2015), was also significantly downregulated further supporting our observations of reduced myeloid cell recruitment. Levels of the anti-inflammatory cytokine IL-10 were not elevated, but remained balanced upon PACAP administration in our experimental setup. These data are in line with a previous finding in a study with experimental autoimmune encephalomyelitis (EAE), where PACAP was able to reduce IFN-γ levels but had no effect on production of IL-10 by spleenocytes (Kato et al., 2004).

Interestingly, we observed a decreased systemic parasite burden despite IFN-γ being negatively affected by the administration of PACAP. Generally, downregulation of IFN-γ, the main driving force against T. gondii infection, would result in an uncontrolled parasite replication (Suzuki et al., 1988). Contrary to previous reports with Trypanosoma (Delgado et al., 2009), we did not detect direct antiparasitic effects of PACAP on the T. gondii replication, which was assessed by monitoring changes in the size or the number of plaques. Instead, IFN-β was found to be upregulated by tendency after exposure to PACAP. The expression of the inflammatory cytokine IFN-β is upregulated following T. gondii infection (Mahmoud et al., 2015). Anti-parasitic effects of type I IFNs in myeloid cells are independent from iNOS and IFN-γ-induced effects. IFN-β acts via induction of IRGM1 that accumulates on the parasitophorous vacuole (PV), in order to disrupt it (Mahmoud et al., 2015). Therefore, we further analyzed the expression of IRGs (IRGM1, IRGM3, GBP2b) and other host defense factors associated with PV disruption. Here, we observed upregulated expression of IRGM1/IRGM3, which implies that PACAP modulates immune cell-mediated parasite elimination rather than direct anti-parasitic effects. Our results align with previous studies (Delgado et al., 1996a), where macrophages exhibit an increased F4/80 expression and enhanced phagocytic capacity.

In contrast to Ly6Chi monocytes and Ly6G<sup>+</sup> neutrophils, CD11c<sup>+</sup> DC recruitment was not affected by PACAP. However, the expression of MHCII in the PACAP-treated group was increased, pointing toward a promotion of antigen recognition and subsequent activation of lymphocytes. In line with previous studies, PACAP-mediated effects on DCs have shown to be mainly induced by signaling through VPAC1, which is the same receptor-mediated pathway utilized for macrophages (Delgado et al., 2004b). Indeed, PACAP administration upregulated the expression of VPAC1 and VPAC2 but not PAC1 by peritoneal exudate cells from acutely infected mice. All three receptors result in the activation of cyclic adenosine monophosphate (cAMP) and the subsequent activation of protein kinase A (PKA) (Delgado et al., 2004a). Specific studies using agonists and antagonists for PACAP receptors have established VPAC1 as the major mediator of the immunomodulatory effects from PACAP, both in vitro and in vivo, with moderate involvement of VPAC2, and minimal or none from PAC1 (Delgado et al., 2004a). Interestingly, in human neutrophils and monocytes, PACAP interaction with VPAC1 and the NGF receptor TrkA resulted in calcium mobilization and subsequent pro-inflammatory activation (El Zein et al., 2006, 2007). Even though PACAP exposure had no effect on the gene expression of Trk receptors, we detected an upregulation of BDNF expression by immune cells upon PACAP administration.

It was previously reported that PACAP upregulated BDNF expression in primary neuronal cultures from rat cerebral cortex, as well as in human neuroblastoma cells upon injury (Frechilla et al., 2001; Shintani et al., 2005; Brown et al., 2013). BDNF is able to influence the immune system via modulation of cytokine expression in peripheral blood mononuclear cells (Vega et al., 2003). BDNF was shown to also be produced by immune cells (Kruse et al., 2007), to modulate monocyte chemotaxis, participate in tissue-healing mechanisms (Samah et al., 2008), and enhance macrophage phagocytic activity (Hashimoto et al., 2005). Recently, we have described a specific role of neurotrophins and their receptors during neuroinflammation. We described that the BDNF receptor p75NTR has a functional impact on the activation status of innate immune cells during T. gondii-induced neuroinflammation (Dusedau et al., 2019).

Here, PACAP was able to reduce the overall recruitment of myeloid-derived mononuclear cells to the peritoneal cavity; in particular, Ly6Chi monocytes. Interestingly, the same cell subset presented higher p75NTR expression than Ly6C<sup>−</sup> peritoneal macrophages, and the administration of PACAP exclusively reduced p75NTR expression on Ly6Chi monocytes. In previous studies (Lee et al., 2016), the use of an antagonist blocker of p75NTR reduced the recruitment of inflammatory monocytes to the CNS, further suggesting that neurotrophin signaling is involved in immune cell migration. Moreover, in a model of in EAE model, the induction of the inflammation resulted in expression of p75NTR in endothelial cells. Although they focus on endothelial p75NTR expression, the study reports a differential immune cell recruitment to the CNS of p75NTR knockout mice, with reduced numbers of cells from the monocyte-macrophage lineage (Kust et al., 2006). In other studies, p75NTR expression by immune cells was reported to increase by a factor of 10 in response to injury (Ralainirina et al., 2010). Our previous work also showed p75NTR upregulation on resident microglia cells and myeloid-derived mononuclear cell subsets in T. gondii-infected brains (Dusedau et al., 2019). These studies supported our findings, where immune cell recruitment and the expression of p75NTR upon inflammation were reduced upon PACAP treatment.

Alongside BDNF, p75NTR signaling in immune cells can be modulated by levels of neurotrophin precursors (proneurotrophins). proBDNF has been shown to negatively affect neuronal plasticity and cell death, reinforced by elevated levels of proBDNF detected in peripheral macrophages (Wong et al., 2010; Luo et al., 2016; Dusedau et al., 2019). Here we observed a downregulation of p75NTR, implying a reduced influence of pro-neurotrophins on Ly6Chi monocytes and thus having a beneficial effect on the resolution of inflammation. The increase of BDNF gene expression in response to PACAP treatment and the involvement of neurotrophins/receptors with VPAC1, especially on neutrophils and monocytes, suggest a possible interaction between p75NTR, BDNF and PACAP within myeloid cells during an inflammatory response. However, future experiments should investigate whether the downregulation of p75NTR is directly PACAP-mediated or a result of an overall reduced inflammation.

In our experiments, the effect of PACAP on parasite elimination may correlate with the synergic, indirect effect of elevated BDNF expression by peritoneal immune cells. However, the role of p75NTR signaling in relation to the immune system remains poorly understood due to the complex interplay of mature vs. pro-neurotrophins, and heterodimeric interactions with Trk receptors (Meeker and Williams, 2014). In summary, our results indicate different routes of PACAP-mediated regulation of the innate response during acute T. gondii infection. As a potent immunomodulator, PACAP has been shown to contribute to the resolution of acute inflammation and parasite elimination by innate immune cells. Furthermore, our findings point toward a potential connection between PACAP and neurotrophinmediated signaling in Ly6Chi inflammatory monocytes. Taken together, these results contribute to the understanding of

the interaction between the nervous and immune systems through neuropeptides.

#### AUTHOR CONTRIBUTIONS

CF and HD performed experiments and analyzed data. JS, NG, MD, GT, DR, and MH critically discussed experimental design, provided material, and co-edited the manuscript. ID conceived experimental design. CF and ID wrote the manuscript.

#### FUNDING

This work was supported by the SFB 854, TP25 to ID and 20765-3/2018/FEKUTSTRAT, GINOP-2.3.2-15-2016-00050

#### REFERENCES


PEPSYS, MTA-TKI 14016, NAP 2017-1.2.1-NKP-2017-00002 to DR.

#### ACKNOWLEDGMENTS

The neuropeptide PACAP38 was synthesized at the Department of Medical Chemistry, university of Szeged (Hungary) kindly provided by DR. We thank Petra Grüneberg, Dr. Sarah Abidat Schneider, and Dana Zabler for excellence assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00154/full#supplementary-material


growth in murine macrophages and embryonic fibroblasts: role of immunityrelated GTPase M1. Cell Microbiol. 17, 1069–1083. doi: 10.1111/cmi.12423


rat cortical neurons. Regul. Pept. 126, 123–128. doi: 10.1016/j.regpep.2004. 08.014


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

Copyright © 2019 Figueiredo, Düsedau, Steffen, Gupta, Dunay, Toth, Reglodi, Heimesaat and Dunay. 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.

# Indoleamine 2,3-Dioxygenase Activity During Acute Toxoplasmosis and the Suppressed T Cell Proliferation in Mice

#### *Edited by:*

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### *Reviewed by:*

Jason Paul Gigley, University of Wyoming, United States Neide Maria Silva, Federal University of Uberlandia, Brazil

#### *\*Correspondence:*

Walter Däubener daeubene@uni-duesseldorf.de

†These authors have contributed equally to this work

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 30 January 2019 *Accepted:* 13 May 2019 *Published:* 05 June 2019

#### *Citation:*

Ufermann C-M, Domröse A, Babel T, Tersteegen A, Cengiz SC, Eller SK, Spekker-Bosker K, Sorg UR, Förster I and Däubener W (2019) Indoleamine 2,3-Dioxygenase Activity During Acute Toxoplasmosis and the Suppressed T Cell Proliferation in Mice. Front. Cell. Infect. Microbiol. 9:184. doi: 10.3389/fcimb.2019.00184

Christoph-Martin Ufermann1†, Andreas Domröse1†, Timo Babel <sup>1</sup> , Anne Tersteegen<sup>1</sup> , Sevgi Can Cengiz <sup>2</sup> , Silvia Kathrin Eller <sup>1</sup> , Katrin Spekker-Bosker <sup>1</sup> , Ursula Regina Sorg<sup>1</sup> , Irmgard Förster <sup>2</sup> and Walter Däubener <sup>1</sup> \*

1 Institute of Medical Microbiology and Hospital Hygiene, Heinrich-Heine-University Düsseldorf, Düsseldorf, Germany, 2 Immunology and Environment, Life and Medical Sciences (LIMES) Institute, University of Bonn, Bonn, Germany

Toxoplasma gondii (T. gondii) is an obligate intracellular parasite and belongs to the phylum Apicomplexa. T. gondii is of medical and veterinary importance, because T. gondii causes the parasitic disease toxoplasmosis. In human cells, the interferon-gamma inducible indoleamine 2,3-dioxygenase 1 (IDO1) is an antimicrobial effector mechanism that degrades tryptophan to kynurenine and thus limits pathogen proliferation in vitro. Furthermore, IDO is described to have immunosuppressive properties, e.g., regulatory T cell differentiation and T cell suppression in humans and mice. However, there is only little known about the role of IDO1 in mice during acute toxoplasmosis. To shed further light on the role of mIDO1 in vivo, we have used a specifically adjusted experimental model. Therein, we infected mIDO1-deficient (IDO−/−) C57BL/6 mice and appropriate wild-type (WT) control mice with a high dose of T. gondii ME49 tachyozoites (type II strain) via the intraperitoneal route and compared the phenotype of IDO−/<sup>−</sup> and WT mice during acute toxoplasmosis. During murine T. gondii infection, we found mIDO1 mRNA and mIDO1 protein, as well as mIDO1-mediated tryptophan degradation in lungs of WT mice. IDO−/<sup>−</sup> mice show no tryptophan degradation in the lung during infection. Even though T. gondii is tryptophan auxotroph and rapidly replicates during acute infection, the parasite load was similar in IDO−/<sup>−</sup> mice compared to WT mice 7 days post-infection. IDO1 is described to have immunosuppressive properties, and since T cell suppression is observed during acute toxoplasmosis, we analyzed the possible involvement of mIDO1. Here, we did not find differences in the intensity of ex vivo mitogen stimulated T cell proliferation between WT and IDO−/<sup>−</sup> mice. Concomitant nitric oxide synthase inhibition and interleukin-2 supplementation increased the T cell proliferation from both genotypes drastically, but not completely. In sum, we analyzed the involvement of mIDO1 during acute murine toxoplasmosis in our specifically adjusted experimental model and found a definite mIDO1 induction. Nevertheless, mIDO1 seems to be functional redundant as an antiparasitic defense mechanism during acute toxoplasmosis in mice. Furthermore, we suggest that the systemic T cell suppression observed during acute toxoplasmosis is influenced by nitric oxide activity and IL-2 deprivation.

Keywords: *Toxoplasma gondii*, IDO, T cell suppression, mouse, kynurenine

## INTRODUCTION

The apicomplexan parasite Toxoplasma gondii (T. gondii) is, due to the fact that it can infect nearly all warmblooded animals, considered to be the most successful parasite worldwide. A primary T. gondii infection in humans is usually asymptomatic, but can cause congenital toxoplasmosis and can thus lead to fatal consequences for the fetus or newborn. In immunocompetent individuals, T. gondii establishes a chronic infection and will thus persist lifelong in the host. Reactivation of a chronic T. gondii infection in humans—e.g., during immunosuppression—results in cerebral toxoplasmosis in most cases (Schlüter et al., 2014). A recent study by Wilking et al. (2016) showed that T. gondii infection, while depending on demographic factors, is highly prevalent in Germany; about 55% of the representative cohort was seropositive for T. gondii.

Defense mechanisms directed against T. gondii are intensively studied. In addition, extensive data were obtained analyzing murine toxoplasmosis since the mouse model is the preferred animal model to study toxoplasmosis in vivo (Gazzinelli et al., 2014; Yarovinsky, 2014; Sasai et al., 2018).

Many different effector mechanisms are described to be involved in the defense against T. gondii, including iron depletion (Dimier and Bout, 1998) and enhanced autophagy (Krishnamurthy et al., 2017). However, the most frequently studied mechanisms directed against intracellular parasites in mice are the enhanced production of nitric oxide (NO) by inducible nitric oxide synthase (iNOS) (Adams et al., 1990; Khan et al., 1997) and the activity of GTPases, enzymes that can hydrolyze guanosine triphosphate (Hunn et al., 2011; Degrandi et al., 2013; Sasai et al., 2018). In humans, tryptophan depletion by indoleamine 2,3-dioxygenase (IDO) is the most frequently described defense mechanism (Pfefferkorn, 1984; MacKenzie et al., 2007).

iNOS was found to be effective against T. gondii in cell cultures, e.g., murine macrophages (Adams et al., 1990) or murine mesenchymal stem cells (Meisel et al., 2011), and more importantly, in in vivo studies using iNOS-deficient mice (Khan et al., 1997). However, in contrast to these findings, NO production favored the growth of T. gondii in cytokineactivated human uroepithelial cells (Däubener et al., 1999), human hepatocytes (Bando et al., 2018), and human retinal pigment epithelial cells (Spekker-Bosker et al., 2019).

Another important antiparasitic effector mechanism directed against T. gondii is the induction of GTPases such as immunity-related GTPases (IRGs) (Hunn et al., 2011) or murine guanylate binding proteins (mGBPs) (Degrandi et al., 2013; Sasai et al., 2018). In humans, there is only one IRG present, and this human IRG is not interferon inducible (Bekpen et al., 2005). Furthermore, human GBP-mediated antiparasitic mechanisms differ from those of murine GBPs (Hunn et al., 2011; Johnston et al., 2016).

The role of the interferon-gamma (IFN-γ) inducible IDO in the defense against T. gondii was first recognized in vitro using human fibroblasts (Pfefferkorn, 1984) and has been confirmed in other human cell lines (e.g., epithelial and endothelial cells) (MacKenzie et al., 2007). In contrast, in murine cells, mIDO does not mediate defense against intracellular T. gondii tachyzoites after IFN-γ stimulation as shown in macrophages and mesenchymal stroma cells (Schwartzman et al., 1990; Meisel et al., 2011). Interestingly, another isoform of IDO has been reported, named indoleamine 2,3-dioxygenase 2 (IDO2). IDO2 has a lower tryptophan affinity than IDO1 in vitro, and its induction, expression, as well as distribution are described to be different from IDO1 (Yeung et al., 2015).

Despite the abovementioned differences in the IDO-mediated antiparasitic effects between human and murine cells, IDOmediated immunoregulatory effects have been described in both murine and human cells. For example, the group of Munn and coworkers found an important function for IDO in the development of immune tolerance in allogeneic pregnancy in mice and described a profound immunosuppression mediated by IDO-positive human macrophages (Munn et al., 1998, 1999). Furthermore, they described the tryptophan depletion as a possible reason for the inhibited T cell proliferation (Munn et al., 1998, 1999); thus, we suggest that this local reduction of tryptophan could also cause a local antimicrobial environment. The immunosuppressive activity of IDO has been confirmed by several groups and is of interest in transplant medicine as well as in tumor immunology and autoimmunity as reviewed previously (Löb et al., 2009). In sum, it was found that IDOpositive dendritic cells (DCs) are able to induce tolerance during T cell activation, while within the tissue, IDO-positive non-professional antigen-presenting cells such as fibroblasts and

**Abbreviations:** ConA, concanavalin A; CpG B, class B phosphate linked cytosine and guanine oligonucleotide; DC, dendritic cell; dpi, days post-infection; EDTA, ethylenediaminetetraacetic acid; FBS, fetal bovine serum; GBP, guanylate binding protein; GTPase, enzyme that hydrolyzes guanosine triphosphate; HFF, human foreskin fibroblasts; HPLC, high-performance liquid chromatography; i. p., intraperitoneal; IDO, indoleamine 2,3-dioxygenase; IL-2, interleukin-2; IMDM, Isocove's Modified Dulbecco's Medium; inf, infected; IFN-γ, interferon gamma; iNOS, inducible nitric oxide synthase; IRG, immunity-related GTPase; MLN, mesenteric lymph node; n.d., not detectable; n.s., not significant; NGMMA, NGmonomethyl-L-arginine; NO, nitric oxide; PBS, phosphate-buffered saline; qRT PCR, quantitative real-time polymerase chain reaction; SD, standard deviation; SEM, standard error of the mean; TDO, tryptophan 2,3-dioxygenase; WT, wild type; 1-L-MT, 1-methyl-L-tryptophan.

endothelial cells could inhibit the effector function of T cells (Lee et al., 2017).

Since the aforementioned effector mechanisms directed against T. gondii are differentially regulated in different species as well as cell types, we decided to investigate the influence of IDO on acute toxoplasmosis in a specifically adjusted murine experimental in vivo model.

#### MATERIALS AND METHODS

#### Cell Line and Parasite Strain Cultivation

Human foreskin fibroblasts (HFF; ATCC <sup>R</sup> SCRC-1041TM, Wesel, Germany) and the murine macrophage cell line (RAW 264.7; ATCC <sup>R</sup> TIB-71TM , Wesel, Germany) were cultured in Iscove's modified Dulbecco's medium (IMDM; Life Technologies, Carlsbad, USA), supplemented with 5% (vol/vol) heat-inactivated fetal bovine serum (FBS; BioWhittaker <sup>R</sup> , Lot N◦ : 9SB003, Lonza, Basel, Switzerland). Cells, as well as isolated cells for ex vivo cultivation, were kept in a humidified Heraeus BB 6220 CO<sup>2</sup> incubator (Thermo Fisher Scientific, Waltham, USA) (37◦C, 5% CO2). HFF cells were passaged after confluency was reached using 0.05% trypsin/ethylenediaminetetraacetic acid (EDTA), (Life Technologies, Carlsbad, USA). Confluent HFF monolayers were used as host cells.

T. gondii strain ME49 tachyzoites (ATCC <sup>R</sup> 50611, Wesel, Germany) were maintained in vitro by serial passages in HFF. For infection experiments, parasites were propagated in HFF (for 42–48 h). Parasites were harvested by scraping off parasitized HFFs in phosphate-buffered saline (PBS) (Life Technologies, Carlsbad, USA). Intracellular parasites were syringe-released and dissociated from host cells debris by differential centrifugation [85×g, room temperature (RT), 5 min; 780×g, RT, 5 min]. Parasites were resuspended in PBS, counted, and adjusted to 5 × 10<sup>5</sup> tachyzoites/ml.

#### Animals and Infection Experiments

mIDO1-deficient (IDO−/−) mice (B6.129-Ido1tm1Alm/J) were originally obtained from the Jackson Laboratory (Bar Harbor, Maine, USA) and had a C57BL/6 genetic background. IDO−/<sup>−</sup> mice were bred and kept under specific pathogen-free (SPF) conditions in the Central Unit for Animal Research and Animal Welfare Affairs of the Heinrich-Heine-University Düsseldorf. C57BL/6 (C57BL/6JRj) mice purchased from Janvier Labs (Le Genest-Saint-Isle, France) were used as wild-type (WT) controls. All experiments were performed with age- and sex-matched cohorts. Mice were infected intraperitoneally (i.p.) with 10<sup>5</sup> T. gondii ME49 tachyzoites in 200 µl of PBS. Naïve control mice and infected mice were kept under SPF conditions and were checked daily. For sample collection, mice were euthanized by cervical dislocation 7 days post-infection (dpi). This study was performed in strict compliance with the German Animal Welfare Act. The experiments were authorized by the North Rhine-Westphalia State Agency for Nature, Environment and Consumer Protection (Permit# 84-02.04.2013.A271, 84- 02.04.2013.A495, and 84-02.04.2016.A508). All efforts were made to minimize animal suffering during the experiments.

#### Sample Collection

Blood samples were taken by cardiac puncture, and the sera were generated from clotted blood samples (4◦C overnight) in two centrifugation steps (20,000 × g, 4◦C, 10 min). Organs [lung, brain, liver, spleen, and mesenteric lymph nodes (MLNs)] were collected and washed in PBS. Whole lung, brain, and liver were homogenized in PBS using the Percellys <sup>R</sup> lysing kit CK28 and the Percellys <sup>R</sup> Minilys <sup>R</sup> tissue homogenizer (Bertin Instruments, Montigny-le-Bretonneux, France). All samples were stored at −80◦C for further processing.

#### Western Blot Analyses

The protein contents of supernatants from centrifuged tissue homogenates or cell lysates generated by freeze–thaw were determined via the Bradford assay (Bio-Rad Laboratories, Hercules, USA). Electrophoretic separation of proteins (30 µg protein per lane) was done with 10% NuPAGE Novex Bis-Tris Mini gels in the appropriate electrophoresis system (Thermo Fisher Scientific, Waltham, USA). Proteins were semi-dry blotted on nitrocellulose membranes (CarboGlas, Schleicher & Schüll, Dassel, Germany). Membranes were blocked in 5% (w/v) skim milk powder in PBS for 1 h at RT. For specific protein detection, the primary antibodies for murine β-actin (1:10,000) (AC-15, Sigma-Aldrich, Munich, Germany), murine iNOS (1:1,000) (1131-1144, CalBiochem <sup>R</sup> , Munich, Germany), or murine IDO (1:500) (AB9900, Chemicon, Merck Millipore, Billerica, MA, USA) were diluted in 0.5% (w/v) skim milk powder in PBS. Membranes were incubated for 1.5 h at RT and were washed three times with PBS (5 min each). The peroxidase-conjugated, secondary antibodies goat anti-mouse IgG (for mβ-actin) or goat anti-rabbit IgG (for mIDO and miNOS) (1:10,000–70,000, Jackson ImmunoResearch Laboratories, Dianova, Hamburg, Germany) were diluted in 0.5% (w/v) skim milk powder in PBS. Membranes were incubated for 2 h at RT and were washed three times with PBS (5 min each). Labeled proteins were detected by enhanced chemiluminescence (Amersham Pharmacia Biotech, Freiburg, Germany).

#### qRT-PCR Analysis of Transcript Levels

Total RNA was extracted according to the TRI Reagent protocol (Merck, Darmstadt, Germany). Briefly, total RNA was extracted from 50 µl of lung tissue homogenate with 500 µl of TRI Reagent and 100 µl of chloroform followed by precipitation with isopropyl alcohol. Extracted RNA was dissolved in 40 µl of UltraPureTM distilled water (Thermo Fisher Scientific, Waltham, USA) and RNA concentration was determined via NanoDrop (Thermo Fisher Scientific, Waltham, USA). Reverse transcription of 1.5 µg of total RNA to cDNA was performed with M-MLV reverse transcriptase and oligo(dT) 12–18 primers according to the manufacturer's instruction (Thermo Fisher Scientific, Waltham, USA). PCR primers to amplify the genes of interest were designed using the Universal ProbeLibrary Assay Design Center (Roche, Basel, Switzerland) and are listed in **Supplementary Table S1**. Real-time PCR was performed with the Takyon NoRox Probe MasterMix dTTP (Eurogentec, Lüttich, Belgium) on a BioRad CFX96 Touch Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, USA). Quality of qPCR analysis was verified by technical replicates for each sample in each run. Each well of a multiplate 96-well PCR plate contained 5 µl of cDNA template, 12.5 µl of Takyon NoRox Probe Master Mix dTTP, 0.3 µl of primer (10µM each), 0.5 µl of probe (10µM), and 6.4 µl of H2O for a total reaction volume of 25 µl. The PCR conditions were 7 min at 95◦C and 40 cycles of 95◦C for 20 s and 60◦C for 1 min.

#### Tryptophan and Kynurenine Quantification

We used high-performance liquid chromatography (HPLC) analysis to quantify total free tryptophan and kynurenine in mice serum and lung tissue. To precipitate existing proteins within the samples, they were mixed with trichloroacetic acid (2.5% final concentration; Sigma-Aldrich, Munich, Germany). To monitor measurement quality, all samples were mixed with 3-nitro-L-tyrosine (Sigma-Aldrich, Munich, Germany) with final concentrations of 2.5 or 10µg/ml for lung tissues or sera, respectively, as internal standard. All samples were filtered (pore size 0.22µm) prior to injection.

Analysis was performed with a System Gold <sup>R</sup> HPLC system (Beckman Coulter, Krefeld, Germany) under usage of a module 166 UV/VIS detector. For separation, a reverse-phase C18 column cartridge (Purospher <sup>R</sup> STAR RP-18 endcapped, Sorbent Lot No. FC095368, 3-µm particle size, 55-mm length, 2-mm diameter; Merck, Darmstadt, Germany) with an adequate guard column (Purospher <sup>R</sup> STAR RP-18 endcapped, Sorbent Lot No. HX435803, 5µm particle size, 4 mm length, 4 mm diameter; Merck, Darmstadt, Germany) was used in a manuCART <sup>R</sup> 55 mm cartridge holder (Merck, Darmstadt, Germany). The mobile phase consisted of 50 mM sodium acetate (Merck, Darmstadt, Germany) adjusted to pH 4.2 with acetic acid (Merck, Darmstadt, Germany) with 5 or 2% acetonitrile (Merck, Darmstadt, Germany) for tryptophan and kynurenine analysis, respectively, using a flow of 0.5 ml/min. All eluents were purchased at least as gradient grade and underwent a vacuum degassing as well as a filtration with a 2µm filter. The absorbance was measured at 280 nm for tryptophan and 360 nm for kynurenine; calculation occurred on the basis of previously measured calibration curves with purchased highly pure L-tryptophan and L-kynurenine (Sigma-Aldrich, Munich, Germany).

#### qPCR Analysis of the Parasite Load

DNA was extracted from lung tissue homogenate by proteinase K digestion. In brief, 500 µl of digestion buffer (1% proteinase K (200µg/mL; Qiagen, Venlo, Netherlands) in lysis buffer [100 mM Tris/HCl (pH 8.5), 5 mM EDTA (pH 8), 0.2% SDS, and 200 mM NaCl] was added to 20 µl of lung tissue homogenate and was incubated at 56◦C and 1,100 rpm on a thermo-shaker for 90 min. DNA was precipitated with 500 µl of isopropyl alcohol and washed with 500 µl of 70% ethanol. Extracted DNA was dissolved in 50 µl of UltraPureTM distilled water and the DNA concentration was adjusted to 100 ng/µl. Quantitative real-time PCR (qPCR) was performed with the abovementioned detection system. For parasite quantification, a standard curve with adjusted T. gondii genomic DNA concentrations was established. The oligonucleotides and template-specific probe that were used are listed in **Supplementary Table S1**. These oligonucleotides bind to a sequence segment of the 35-fold repetitive B1 gene of T. gondii that is commonly used in diagnostics (Burg et al., 1989; Pelloux et al., 1998). Quality of qPCR analysis was verified by technical replicates for each sample in each run. Each well of a multiplate 96-well PCR plate contained 5 µl of DNA template, 12.5 µl of Takyon NoRox Probe Master Mix dTTP, 2.5 µl of primer (3µM each), and 2.5 µl of probe (2µM) for a total reaction volume of 25 µl. The PCR conditions were 10 min at 95◦C and 45 cycles of 95◦C for 15 s and 60◦C for 1 min.

#### Isolation and Cultivation of Murine Cells From Spleen and Mesenteric Lymph Nodes

For ex vivo lymphocyte proliferation experiments, cells from spleen and MLN tissues were digested using 1 mg/ml collagenase (C2139, Sigma-Aldrich, Munich, Germany) and 180 U/ml DNase (Roche, Basel, Switzerland) in PBS for 30 min at 37◦C. Digested tissues were passed through 70-µm nylon sieves (Falcon <sup>R</sup> Corning Inc.; Corning, New York, USA) followed by erythrocyte lyses (MORPHISTO GmbH; Frankfurt am Main, Germany). Cells were resuspended in medium [IMDM with 5% FBS and 100 U/ml penicillin/100µg/ml streptomycin (Biochrom GmbH, Berlin, Germany)] and counted using trypan blue (0.4%; Sigma-Aldrich, Munich, Germany). Cells were seeded in low-evaporation lid 96-well flat-bottom plates (Corning Inc., Corning, New York, USA) at 3 × 10<sup>5</sup> cells per well.

#### Lymphocyte Proliferation Assay

Lymphocyte proliferation was stimulated with the mitogens concanavalin A (ConA; 1µg/ml; Sigma-Aldrich, Munich, Germany) and the class B phosphate-linked cytosine and guanine oligonucleotide ODN1826 (CpG B; 0.1µM; Invivogen; San Diego, CA, USA) as indicated. Additional supplementation with recombinant human interleukin-2 (IL-2; 5 ng/mL; R&D Systems, Minnesota, USA) and the NOS inhibitor NG-monomethyl-Larginine (NGMMA; 100µg/ml; Merck, Darmstadt, Germany) was performed in the concentrations indicated.

Lymphocyte proliferation was determined by the <sup>3</sup>Hthymidine incorporation method. In brief, <sup>3</sup>H-thymidine (74 kBq per well; GE Healthcare Buchler GmbH & Co. KG, Braunschweig, Germany) was added 48 h post-stimulation. Lymphocyte proliferation was stopped after additional 24 h of cultivation by freezing. Lymphocyte proliferation was determined by measuring incorporated <sup>3</sup>H-thymidine using liquid scintillation spectrometry (1205 Betaplate, PerkinElmer, Jugesheim, Germany).

#### Indirect Nitric Oxide Estimation

NO production was measured via the Griess assay (Ding et al., 1988). Here, nitrite—a stable breakdown product of NO—is measured. In brief, 100 µl of cell culture supernatant was used after 72 h of in vitro cultivation. The Griess assay was performed as described before (Meisel et al., 2011). The nitrite content was calculated by extrapolation from a sodium nitrite standard curve assayed parallel to each measurement.

## Statistical Analysis

Results are indicated as means ± SD or ± SEM as indicated in figure legends. Statistical significances of differences in mean values were analyzed by using the unpaired two-tailed Student's t-test (GraphPad Prism). Significant differences were indicated with asterisks (∗p ≤ 0.05; ∗∗p ≤ 0.001; ∗∗∗p ≤ 0.0001).

#### RESULTS

#### Expression of mIDO1 mRNA and Protein in Lungs of Infected Mice

To clear up potentially different murine indoleamine 2,3 dioxygenase 1 (mIDO1) distribution among various murine tissues within WT and IDO−/<sup>−</sup> mice, Western blot analyses were performed. Here, no mIDO protein was detectable in liver, brain, or lung tissue of naïve WT mice, while infection induced strong mIDO1 expression in lung tissue as well as a slight expression in liver tissue. As suggested in IDO−/<sup>−</sup> mice, in all tested conditions and tissues, no mIDO protein was detectable (**Figure 1A**). Further quantitative real-time PCR experiments with the lung tissues were conducted to detect mIDO1 mRNA. Shown data represent the relative gene expression of infected to naïve mice in WT and IDO−/−, respectively. Expression of mGBP2 was equally strong in WT and IDO−/<sup>−</sup> mice 7 dpi, as expected. Upon T. gondii infection, mIDO1 mRNA expression is strongly upregulated in WT mice. As expected, we did not detect any mIDO1 expression in infected IDO−/<sup>−</sup> mice. Murine mIDO2 was measured as well to exclude mIDO2 as a responsible candidate for results shown further on. mIDO2 is only marginally increased during T. gondii infection in a few infected mice and there is no significant difference between WT and IDO−/<sup>−</sup> mice (**Figure 1B**). Additionally, we measured miNOS expression in lungs of WT and IDO−/<sup>−</sup> mice at different time points post-infection via quantitative real-time PCR and Western blot analysis (**Supplementary Figure S1**). Relative miNOS expression in infected WT mice was increased in a time-dependent manner (**Supplementary Figure S1A**). However, differences between WT and IDO−/<sup>−</sup> on day 7 post-infection were not significant (**Supplementary Figure S1B**). miNOS protein was absent in lungs of naïve WT and IDO−/<sup>−</sup> mice. Infected IDO−/<sup>−</sup> mice were positive for miNOS protein at 7 and 9 dpi and infected WT mice only at 9 dpi (**Supplementary Figure S1C**).

#### Comparison of mIDO1-Based Tryptophan Degradation and Parasite Loads of Naïve and Infected WT and IDO−/<sup>−</sup> Animals

We explored the antiparasitic properties influenced by mIDO1 during T. gondii infection by comparing the tryptophan degradation as well as the parasite load.

Therefore, we determined tryptophan and kynurenine concentrations in sera via HPLC analyses to analyze the systemic distribution of these metabolites. Furthermore, we analyzed lung tissue via HPLC, since we previously identified lung tissue as one center of mIDO1 protein and mRNA expression.

With 15.3 and 17.7µg/ml, naïve WT and IDO−/<sup>−</sup> mice exhibit no significant differences in mean tryptophan

mIDO1, and mIDO2 in lung tissue homogenates of infected mice relative to their expression in naïve control samples (B). Data were normalized to the housekeeping gene <sup>β</sup>-actin and were represented as 2−11CT (naïve vs. infected) in scattered dot plots and means ± standard deviation. The Student's t-test (unpaired, two-tailed) was used to determine statistical differences marked with asterisks (n.s., not significant; \*\*\*p ≤ 0.0001).

concentrations in serum. On day 7 after T. gondii infection, serum tryptophan concentrations drop significantly to 7.5 and 10.2µg/ml in WT and IDO−/<sup>−</sup> animals, respectively. Concomitantly with this serum tryptophan drop, the serum kynurenine concentration in the WT rises significantly from 0.23 to 0.94µg/ml. Even though the serum tryptophan concentration drops in the IDO−/<sup>−</sup> animals similar to the WT, the serum kynurenine concentration is unaltered (<0.1µg/ml) in the IDO−/<sup>−</sup> animals (**Figure 2A**). The lung tryptophan concentrations of naïve WT and IDO−/<sup>−</sup> mice behaved like the serum tryptophan concentrations without significant differences, but with 7.7 and 7.3µg/ml, they are overall lower. Infected WT animals show a significant drop in lung tryptophan concentration (from 7.7 to 2.4µg/ml) paired with a significant increase in the lung kynurenine concentration (from 0.4 to 6.7µg/ml). In contrast, infected IDO−/<sup>−</sup> animals exhibit no significant difference in lung tryptophan concentrations compared with the naïve group. Kynurenine concentrations in lungs of naïve and infected IDO−/<sup>−</sup> animals are likewise low (<0.1µg/ml) as in sera (**Figure 2B**).

To draw a conclusion regarding the previously mentioned potential antiparasitic properties of mIDO1, we determined

statistical differences marked with asterisks (n.s., not significant; \*p ≤ 0.05, \*\*p ≤ 0.001, and \*\*\*p ≤ 0.0001).

the parasite load in lung tissues via real-time PCR from the same samples we analyzed beforehand. Here, we used specific oligonucleotides to detect the 35-fold repetitive B1 gene of T. gondii. As expected, there was no detection of T. gondii in naïve WT and IDO−/<sup>−</sup> mice. We measured a significant amount of parasites in the lungs of WT and IDO−/<sup>−</sup> mice 7 dpi; however, the parasite load in WT and IDO−/<sup>−</sup> mice was not significantly different (**Figure 2C**).

### Suppressed T Cell Proliferation Responses During Acute *T. gondii* Infection

Splenocytes were isolated from naïve and infected WT mice to analyze the proliferative responses of lymphocytes during acute toxoplasmosis. We performed initial T cell proliferation experiments to analyze the suitability of our specifically adjusted experimental model **(Supplementary Figure S2**). Therefore, we infected WT mice i.p. with in vitro cultivated tachyzoites or bradyzoites isolated from lysed brain cysts propagated in vivo. Here, we could not detect any differences in mitogen-stimulated T cell proliferation responses (**Supplementary Figure S2A**). Furthermore, we tested the time-dependent mitogenstimulated T cell proliferation responses. The results shown in **Supplementary Figure S2B** clearly illustrate that the proliferation responses are not impaired at 3 dpi, are reversibly impaired at 7 dpi, and are irreversibly impaired at 10 dpi (**Supplementary Figure S2**). The proliferation of T cells and B cells was induced by stimulation with the mitogens concanavalin A (ConA) and the class B CpG oligonucleotide ODN1826, respectively (**Figure 3A**). Untreated splenocytes from infected WT mice show a weak basal proliferation compared to splenocytes from naïve mice. Mitogen stimulation of naïve splenocytes induced a potent lymphocyte proliferation response, whereas splenocytes from infected mice have a low proliferation response. In more detail, CpG B stimulation showed that the B cell proliferation response was slightly reduced by approximately 28% during acute T. gondii infection. However, T

cell stimulation with ConA showed a very prominent impaired T cell proliferation response (>90% reduction). ConA stimulation of MLN cells from naïve and infected WT mice resulted in the same phenotype (**Figure 3B**). In detail, MLN-derived T cells also showed a very weak proliferative response to ex vivo ConA stimulation during acute T. gondii infection (>90% reduction) as observed in the stimulation of splenic T cells. Thus, further experiments were conducted with splenocytes to perform more profound analyses of the T cell responses during acute toxoplasmosis.

≤ 0.001 and \*\*\*p ≤ 0.0001).

#### IL-2 Availability and NOS Activity, but Not mIDO1 Influence T Cell Proliferation Responses During Acute Toxoplasmosis

The role of mIDO1 in the suppressed T cell proliferation responses during acute toxoplasmosis is not known. Thus, splenocytes were isolated from naïve and T. gondii-infected WT and IDO−/<sup>−</sup> mice 7 dpi.

Splenic T cells from naïve WT and IDO−/<sup>−</sup> mice respond comparably strong to mitogen stimulation (**Figure 4A**). During acute T. gondii infection, the mitogen-induced proliferative responses were highly reduced in WT splenocytes (>90%) and IDO−/<sup>−</sup> splenocytes (>92%) (**Figure 4A**). This indicates that the suppressed T cell responses are affected independently of mIDO1.

IL-2 deprivation (Khan et al., 1996) as well as iNOS activity (Patton et al., 2002) have previously been described to be involved in the impaired T cell proliferation response during acute toxoplasmosis. Thus, we performed ex vivo mitogen stimulation experiments with supplementation of IL-2 and the NOS inhibitor NGMMA to elucidate their interplay in the proliferation of splenocytes from WT and IDO−/<sup>−</sup> mice. Supplementation of IL-2 alone did not significantly improve the proliferation of T cells from either infected genotype (WT: from 9.7 to 14.1%; IDO−/−: from 7.6 to 10.9%) (**Figure 4A**). NOS inhibition via ex vivo NGMMA treatment resulted in a small but significant elevation of T cell proliferation, which was equally strong for cells from both infected genotypes (WT: from 9.7 to 29.7%; IDO−/−: from 7.6 to 30.2%) (**Figure 4A**). Combining the supplementation of IL-2 and NGMMA increased the proliferation of mitogen-treated T cells from both infected genotypes even further, without however reaching the level of the naïve proliferation response (WT: from 9.7 to 59.5%; IDO−/−: from 7.6 to 61.8%) (**Figure 4A**), thus resulting in a highly significant elevation of the proliferative response compared to mitogen stimulation alone.

Stimulated as well as untreated splenocytes isolated from T gondii-infected WT and IDO−/<sup>−</sup> mice showed NOS activity as measured indirectly via nitrite accumulation in the supernatant (**Figure 4B**). Here, splenocytes from IDO−/<sup>−</sup> mice produce significantly more nitrite compared to the equally treated WT splenocytes (untreated: 4.6µM for WT and 8.1µM for IDO−/−; ConA stimulated: 5.9µM for WT and 9µM for IDO−/−) (**Figure 4B**). NOS inhibition via ex vivo NGMMA treatment reduced the nitrite concentration in supernatants strongly to 1.4 and 2.2µM for WT and IDO−/−, respectively (**Figure 4B**).

# DISCUSSION

Indoleamine 2,3-dioxygenase (IDO) is described as a potent antimicrobial factor in in vitro systems using human, porcine, and bovine cells. In this context, IDO activity has been shown to inhibit pathogens like bacteria (e.g., group A streptococci, Staphylococcus aureus), viruses (e.g., Herpes simplex virus 1, Cytomegalovirus), and parasites (e.g., T. gondii, Neospora caninum) (Däubener et al., 2009). However, the role of IDO as a potent antimicrobial factor in vivo remains controversial. Here, we used C57BL/6 mice deficient for mIDO1 (IDO−/−) to investigate acute toxoplasmosis with specific regard to the general systemic proinflammatory reaction and the local parasite burden within the lung, a strong IDO-expressing organ. We have adjusted our experimental model by comparing the infection with tachyzoites and bradyzoites. Furthermore, we have

evaluated the optimal time point for our objectives. In this specifically adjusted experimental model, using a high dose of tachyzoites via the intraperitoneal route, C57BL/6 mice develop a more intense acute toxoplasmosis compared to BALB/c mice (own preliminary data not shown). The type II strain T. gondii ME49 was chosen for our infection experiments, since type II strains are the most frequently found T. gondii strains in human toxoplasmosis (Schlüter et al., 2014). To ensure standardized infection inoculums, we infected mice with a high dose of T. gondii ME49 tachyzoites via i.p. injection, thereby circumventing the use of brain-derived cysts that vary in size and parasite number (Dubey and Frenkel, 1998). However, it has to be taken into account that oral cyst uptake is a natural route of infection, whereas i.p. injection of tachyzoites into inbred mice is a strictly experimental setup. Additionally, infection via the natural route results in a slower but more natural course of disease compared to our specifically adjusted experimental model. Finally, it has to be considered that our model does not represent the natural course of toxoplasmosis, but is ideally adjusted for the herein analyzed objectives. Furthermore, it cannot be excluded that the high infection dose used in our model might mask a role of IDO during the natural course of a T. gondii infection.

Here, we show that infection of mice with T. gondii tachyzoites results in a strong mIDO1 induction in lungs. In detail, we found high amounts of mIDO1 mRNA and mIDO protein in the lungs of T. gondii-infected WT animals. Similar observations have been obtained during allergic diseases (Hayashi et al., 2004) or allogeneic stem cell transplantation (Lee et al., 2017) in mice. In both publications, mIDO immunoreactivity was found especially in lung epithelial cells (Hayashi et al., 2004; Lee et al., 2017). Furthermore, published data in the context of other murine infections have shown similar mIDO expression. mIDO protein and/or mRNA was found in the lungs of mice experimentally infected with the influenza A virus (Gaelings et al., 2017), the pathogenic fungus Paracoccidioides brasiliensis (Araújo et al., 2014), and the pathogenic bacterium Mycobacterium tuberculosis (M. tuberculosis) (Blumenthal et al., 2012), thus indicating that mIDO does function as an antimicrobial effector mechanism in murine lungs in vivo.

During T. gondii infection, we found reduced tryptophan concentrations in sera of WT animals, which were even more pronounced in lung tissue, confirming previously published data (Silva et al., 2002; Murakami et al., 2012). The same samples were tested for their kynurenine concentration, as kynurenine is a degradation product of tryptophan. The decreased tryptophan concentrations were accompanied with an increase in kynurenine concentrations. However, we also detected a drop in tryptophan concentrations in infected IDO−/<sup>−</sup> mice. This observation is unlikely due to possible mIDO2 activity, since we did not detect mIDO2 mRNA in the majority of samples. Furthermore, the tryptophan drop in the IDO−/<sup>−</sup> mice is not accompanied by an increase of kynurenine. Therefore, we suggest that enhanced protein biosynthesis by host cells and the rapidly proliferating T. gondii tachyzoites during the acute phase of toxoplasmosis are responsible for the decreased tryptophan concentration in the serum of infected animals. Our finding supports this hypothesis, since there is no evidence for an enhanced tryptophan cleavage in the lungs of infected, IDO−/<sup>−</sup> mice.

We can confirm the observation by Divanovic et al. that IDO−/<sup>−</sup> mice show no phenotype compared to the WT during acute toxoplasmosis, but rather behave similarly (data not shown). Furthermore, they reported that treatment of chronically infected WT mice with the IDO inhibitor 1-methyl-D-tryptophan (1-D-MT) resulted in T. gondii encephalitis (Divanovic et al., 2012). In a previous publication, Murakami et al. reported reduced mRNA expression of the T. gondii surface antigen 2 in lungs of T. gondii-infected IDO−/<sup>−</sup> mice compared to WT mice 7 dpi, indicating a lower parasite load or a reduced metabolic activity. Therefore, we analyzed the parasite load in the lungs of the infected animals by detecting a T. gondiispecific DNA sequence. Here, we did not detect a significant difference in the T. gondii load in lungs of WT or IDO−/<sup>−</sup> mice. Again, a possible involvement of mIDO2 to compensate for the lack of mIDO1 is unlikely, since mIDO2 mRNA was rare and detectable only at low levels in infected IDO−/<sup>−</sup> and WT mice. Another tryptophan-degrading enzyme—tryptophan 2,3-dioxygenase (TDO)—might, however, be involved. Human TDO has been described by us to mediate antimicrobial and immunoregulatory effects similar to human IDO (Schmidt et al., 2009). Human TDO has been identified by Hsu et al. (2016) as the main tryptophan-degrading enzyme in human lung cancerassociated fibroblasts. Due to these findings, we have recently established a mIDO1 and mTDO double-deficient mouse strain to further elucidate the involvement of mIDO1 and mTDO during murine infections.

We have shown that a tryptophan concentration of <1µg/ml is necessary to inhibit bacterial (S. aureus) growth as well as human T cell proliferation in vitro (Müller et al., 2009). Despite our current finding that the tryptophan concentration in murine lung tissue is strongly reduced during T. gondii infection, we could not detect increased parasite loads in lungs of IDO−/<sup>−</sup> animals, even though T. gondii is tryptophan auxotroph. Thus, the tryptophan depletion 7 dpi might not be sufficient to mediate antiparasitic effects in vivo. Detailed information concerning the minimal tryptophan concentration for T. gondii growth in vivo is not available. Our data clearly showed a time-dependent increase of miNOS in lung tissue of infected WT mice on transcript level but could not detect a difference between WT and IDO−/<sup>−</sup> mice 7 dpi. However, in lungs of IDO−/<sup>−</sup> mice, we could detect miNOS protein earlier post-infection compared to the WT. Thus, iNOS expression in murine tissues might mediate parasite control during acute toxoplasmosis. This might be another reason why we did not find mIDO1 to be involved in the control of the rapidly replicating tachyzoites during acute toxoplasmosis. iNOS is a previously described antimicrobial defense mechanism and is induced in T. gondii-infected mice (Khan et al., 1997). However, mice deficient for iNOS showed prolonged survival in comparison to WT mice (Khan et al., 1997). Detailed analyses showed that enhanced liver degeneration, extensive ulceration, and necrosis in the small intestine were responsible for the earlier death of iNOS-expressing WT mice (Khan et al., 1997). We found higher nitrite accumulation in supernatants of ex vivo splenocyte cultures from infected IDO−/<sup>−</sup> mice compared to infected WT mice. Ye et al. have reported a similar observation in a stem cell transplantation model. They showed that 1 methyl-DL-tryptophan mediated inhibition of mIDO resulted in increased NO concentration in the supernatant of mixed lymphocyte cultures with lymphocytes isolated from BALB/c and C57BL/6 mice (Ye et al., 2017). This indicates, on the one hand, that IDO is influencing NO production. On the other hand, we (Däubener et al., 1999) and others (Bando et al., 2018) found that iNOS can block IDO-mediated antimicrobial effects. Thus, we suggest that mIDO1 and iNOS interact during acute toxoplasmosis and that mIDO1 activity might be required for the regulation of iNOS activity during acute toxoplasmosis in WT mice. Higher iNOS activity might compensate for the missing mIDO1 in IDO−/<sup>−</sup> mice, whereby potential antiparasitic effects of mIDO1 were not detectable in our experimental setup. The herein mentioned detection of miNOS protein supports this suggestion. Thus, infection experiments with mice deficient for mIDO1 and iNOS might be of interest, since Scharton-Kersten et al. (1997) reported that iNOS-deficient mice can survive acute toxoplasmosis and control parasite growth at the site of infection via NO-independent mechanisms. This observation might be due to other aforementioned defense mechanisms (e.g., GTPases) or due to IDO activity. However, that remains to be shown.

Experimental evidence that IDO mediates antimicrobial effects directly via tryptophan depletion in mice came from in vivo experiments with bacterial infections. For example, Peng and Monack published that genes in the tryptophan biosynthesis pathway are essential for the pathogenic bacterium Francisella novicida (F. novicida) to multiply in lungs of C57BL/6 mice (Peng and Monack, 2010). Thereafter, bacteria deficient in tryptophan synthesis were constructed, and it was found that this strain had lost its capacity to replicate in the lungs of C57BL/6 mice. In lungs of IDO−/<sup>−</sup> mice, this tryptophan auxotrophic F. novicida strain was able to replicate, thus suggesting that tryptophan depletion via mIDO1 did protect the WT mice from the bacterial infection. Comparable data were obtained with a pharmacologic blockage of tryptophan synthesis in M. tuberculosis. Zhang et al. showed that a blockage of tryptophan synthesis by halogenated anthranilate analogs disrupted tryptophan biosynthesis in M. tuberculosis. Treatment of infected mice with this compound resulted in an inhibition of bacterial growth (Zhang et al., 2013). Inhibition of IDO in macaques during experimental infections with M. tuberculosis led to reduced bacterial burden, indicating a better control of the M. tuberculosis infection in treated animals

(Gautam et al., 2018). However, Gautam et al. used 1-D-MT, which is not an IDO inhibitor but is rather described to inhibit IDO-mediated immunoregulatory functions (Metz et al., 2012). Therefore, the observed effects in macaques might be due to an enhanced immune reaction against M. tuberculosis.

In mice, mIDO expressing plasmacytoid DCs have been reported to suppress T cell responses in tumor-draining lymph nodes (Munn et al., 2004). Furthermore, DCs that express IDO have been linked to several other immunoregulatory functions, for example, the differentiation of regulatory T cells (Grohmann et al., 2017). In addition, tolerance toward self-antigens is regulated by mIDO in the marginal zones of the murine spleen (Ravishankar et al., 2012). Therefore, it was of interest to analyze whether mIDO1 is involved in the T cell suppression, seen during an acute T. gondii-infected mouse.

We measured T cell responses from in vitro mitogenstimulated splenocytes, isolated from T. gondii-infected mice. Here, we observed a strong suppression of the T cell proliferation in splenocytes from infected compared to naïve mice. However, there was no difference between IDO−/<sup>−</sup> and WT mice in our experiments, indicating that mIDO1 is not a major factor that regulates the observed T cell suppression. Previous experiments by Chan et al. (1986) have indicated that IL-2 availability as well as macrophages (as potential NO producers) are involved in the T cell suppression observed during acute toxoplasmosis.

Supplementation of IL-2 alone did not influence the proliferation of T cells in our setup, as reported by Khan and coworkers. They observed an increase in the T cell proliferation upon in vitro supplementation of IL-2 during mitogen stimulation of purified CD4<sup>+</sup> T cells (Khan et al., 1996). In our setup, we stimulated splenocytes consisting not only of T cells but rather of a broad variety of cell types, including macrophages. T cell proliferation has also been reported to be influenced by NO derived from activated macrophage before (Albina et al., 1991; Patton et al., 2002). Inhibition of NOS in our experiments increased the proliferation of T cells derived from both infected mouse strains significantly. IL-2 supplementation and NOS inhibition in combination further increased T cell proliferation, but it did not reach the proliferation level of naïve T cells.

Salinas et al. (2014) have demonstrated that conventional T cells compete with regulatory T cells for available IL-2 in purified T cells isolated during acute toxoplasmosis, induced by infection of C57BL/6 mice orally with 50 T. gondii ME49 cysts. With our finding that splenocytes from IDO−/<sup>−</sup> mice

#### REFERENCES


behave like splenocytes from WT mice during ex vivo mitogen stimulation, we suggest that the T cell suppression during acute toxoplasmosis is mediated by NOS activity and might even be mediated by IL-2 deprivation as described by Salinas et al. (2014). However,in that case, induction of regulatory T cells would then be independent of mIDO1 in the described T cell suppression during acute toxoplasmosis.

#### ETHICS STATEMENT

This study was performed in strict compliance with the German Animal Welfare Act. The experiments were authorized by the North Rhine-Westphalia State Agency for Nature, Environment and Consumer Protection (Permit# 84 02.04.2013.A271, 84 02.04.2013.A495 and 84 02.04.2016.A508). All efforts were made to minimize animal suffering during the experiments.

#### AUTHOR CONTRIBUTIONS

WD and IF conceived and supervised the study. C-MU, AD, and WD designed the experiments, prepared the figures, and wrote the manuscript. C-MU and AD performed the majority of experiments and analyzed the data. TB, AT, SC, SE, and KS-B performed the experiments. US supervised animal experiments. All authors reviewed the manuscript.

#### FUNDING

WD and IF acquired funding from The Jürgen Manchot Foundation. C-MU, AT, and SC are scholarship holders as part of the Manchot Graduate School Molecules of infection III. IF is a member of the DFG-funded cluster of excellence ImmunoSensation2 (Project-ID: 390873048).

#### ACKNOWLEDGMENTS

We gratefully thank Claudia Woite and Winfried Schwippert for technical assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00184/full#supplementary-material


of the cell autonomous resistance mechanism in the human lineage. Genome Biol. 6:R92. doi: 10.1186/gb-2005-6-11-r92


Germany: a representative, cross-sectional, serological study. Sci. Rep. 6, 1–9. doi: 10.1038/srep22551


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

Copyright © 2019 Ufermann, Domröse, Babel, Tersteegen, Cengiz, Eller, Spekker-Bosker, Sorg, Förster and Däubener. 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.

# Disease Tolerance in Toxoplasma Infection

#### Stephanie J. Melchor and Sarah E. Ewald\*

Department of Microbiology, Immunology and Cancer Biology and the Carter Immunology Center, University of Virginia School of Medicine, Charlottesville, VA, United States

Toxoplasma gondii is a successful protozoan parasite that cycles between definitive felid hosts and a broad range of intermediate hosts, including rodents and humans. Within intermediate hosts, this obligate intracellular parasite invades the small intestine, inducing an inflammatory response. Toxoplasma infects infiltrating immune cells, using them to spread systemically and reach tissues amenable to chronic infection. An intact immune system is necessary to control life-long chronic infection. Chronic infection is characterized by formation of parasite cysts, which are necessary for survival through the gastrointestinal tract of the next host. Thus, Toxoplasma must evade sterilizing immunity, but still rely on the host's immune response for survival and transmission. To do this, Toxoplasma exploits a central cost-benefit tradeoff in immunity: the need to escalate inflammation for pathogen clearance vs. the need to limit inflammation-induced bystander damage. What are the consequences of sustained inflammation on host biology? Many studies have focused on aspects of the immune response that directly target Toxoplasma growth and survival, commonly referred to as "resistance mechanisms." However, it is becoming clear that a parallel arm of the immune response has evolved to mitigate damage caused by the parasite directly (for example, egress-induced cell death) or bystander damage due to the inflammatory response (for example, reactive nitrogen species, degranulation). These so-called "disease tolerance" mechanisms promote tissue function and host survival without directly targeting the pathogen. Here we review changes to host metabolism, tissue structure, and immune function that point to disease tolerance mechanisms during Toxoplasma infection. We explore the impact tolerance programs have on the health of the host and parasite biology.

#### Edited by:

Nicolas Blanchard, INSERM U1043 Centre de Physiopathologie de Toulouse Purpan, France

#### Reviewed by:

Carsten Lüder, Institut für Medizinische Mikrobiologie, Universitätsmedizin Göttingen, Germany Elizabeth Wohlfert, University at Buffalo, United States Ira Blader, University at Buffalo, United States

> \*Correspondence: Sarah E. Ewald se2s@virginia.edu

#### Specialty section:

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

Received: 02 February 2019 Accepted: 14 May 2019 Published: 06 June 2019

#### Citation:

Melchor SJ and Ewald SE (2019) Disease Tolerance in Toxoplasma Infection. Front. Cell. Infect. Microbiol. 9:185. doi: 10.3389/fcimb.2019.00185 Keywords: tolerance, Toxoplasma gondii, resistance, cachexia, parasite, innate immunity, chronic infection

# INTRODUCTION: DISEASE TOLERANCE vs. RESISTANCE

A successful immune response requires two distinct, but complementary components: "resistance mechanisms" and "tolerance mechanisms." Resistance (or restriction) mechanisms directly target pathogens to limit microbial replication or dissemination. Examples of resistance mechanisms include antimicrobial peptides, complement, and degranulation by neutrophils or cytotoxic T cells. Tolerance mechanisms target host cell biology to improve tissue integrity and function in the setting of damage caused by the pathogen or the inflammatory response. Examples include extracellular matrix remodeling, the DNA damage response, antioxidant production, and shifts in cell metabolism (Ayres and Schneider, 2009, 2012; Medzhitov et al., 2012; Soares et al., 2014). Although tolerance mechanisms are often induced by and intimately related to an immune response, many disease tolerance programs are carried out by non-immune (non-hematopoietic) cell types. In fact, disease tolerance mechanisms were first described in plants, which lack a distinct cellular immune system, where they have been associated with tissue integrity, growth, and reproductive capacity (Caldwell et al., 1958).

Historically, immunologists have used the word "tolerance" to describe two processes that limit lymphocyte autoreactivity. In central tolerance, B or T lymphocytes with an affinity for self-antigens are deleted in the bone marrow or thymus before entering circulation. In peripheral tolerance, auto-reactive lymphocytes that arise later in development are deleted or rendered unresponsive to antigen (anergic) in peripheral tissues or in lymph nodes. Regulatory T cells (Tregs) play an important role in peripheral tolerance. Central and peripheral lymphocyte tolerance are included under the broader umbrella of so-called "disease tolerance" strategies because they are mechanisms of limiting inflammation and preventing auto-immunity but do not directly influence the pathogen. In this review, we refer to the broader definition of "disease tolerance" as mechanisms that support host fitness and survival during infection, not by directly targeting pathogen biology, but by shifting homeostasis to maintain tissue function during infection and inflammation (Martins et al., 2019).

It is important to note that tolerance biology, like resistance mechanisms, represent a shift from normal homeostasis. The costs of overactive resistance mechanisms are well-established. Examples include bystander damage from reactive oxygen species, delayed wound healing, and recognition of autoantigens like DNA- or RNA-binding proteins. The negative effects of excessive tolerance programs are comparatively underappreciated, but include mutations due to imperfect DNA repair, fibrotic wound healing and, potentially, metabolic disorders like Type II diabetes. We can assume that long-term reliance on tolerance mechanisms must be maladaptive, otherwise they would be selected for homeostatic function. This is an important consideration, as tolerance programs are receiving attention as therapeutic targets to interrupt the maladaptive consequences of acute inflammation (for example, sepsis or influenza infection) without increasing susceptibility to pathogens. Host survival during infection depends having both tolerance and resistance mechanisms, otherwise the host will succumb either to pathogen overgrowth or lethal immunopathology. As discussed in the following sections, it is also important to consider how promoting tolerance biology may benefit the life cycle of pathogens, particularly those that have evolved a strategy for long-term persistence within a host (**Figure 1**).

## REINTERPRETING THE IMMUNE RESPONSE TO TOXOPLASMA IN THE FRAMEWORK OF DISEASE TOLERANCE

Toxoplasma gondii is an obligate intracellular protozoan parasite that establishes life-long infection in a wide range of warmblooded intermediate hosts, including rodents and humans.

infection (left quadrants). Without tolerance mechanisms the host is susceptible to inflammation-induced pathology, even if parasite replication effectively restricted (lower right quadrant). Both resistance and tolerance mechanisms are necessary for host survival (upper right quadrant). This balance also benefits Toxoplasma by ensuring that the host survives long enough to enable bradyzoite cyst differentiation, a requirement for transmission to another host.

Three major haplotypes, types I–III, dominate Europe and North America which differ in virulence by several logs. Feline definitive hosts support Toxoplasma sexual recombination and shed millions of environmentally stable, highly infectious oocysts in feces. This confers a tremendous benefit for the parasite in terms of genetic diversity and dissemination potential, which suggests that rodents may be an important intermediate host for Toxoplasma given the predator-prey relationship between the two host organisms.

Intermediate hosts are infected by ingesting Toxoplasma oocysts or tissue cysts, the so-called bradyzoite form, from a previously infected intermediate host. Toxoplasma invades the small intestine, converting to the rapidly dividing tachyzoite form and inducing inflammation (Dubey et al., 1998). This process activates an immune response that is critical to the parasite's life cycle in two major ways. First, Toxoplasma infects and replicates within infiltrating immune cells (Courret et al., 2006). The parasite uses immune cells to disseminate systemically, reaching tissue sites that support chronic infection including the brain, muscle, and other tissues (Bierly et al., 2008). Second, chronic infection is defined by parasites shifting to a bradyzoite transcriptional program and clearance of most tachyzoites. This involves synthesis of a saccharide-rich parasite cyst wall which is necessary for the parasite to survive transit through the gastrointestinal tract of the next host (Bierly et al., 2008; Gregg et al., 2013). Thus, without a robust immune response, the parasite kills the host before this shift occurs and the opportunity for transmission is limited.

In the mouse, Toxoplasma is recognized by innate immune sensors including Toll-like receptors, and the NLRP1 and NLRP3 inflammasomes. Mice deficient in tlr11 or its signaling adaptor myd88 fail to control parasite replication after intraperitoneal (i.p.) infection (Yarovinsky et al., 2008). Similarly, mice lacking the inflammasome effector caspases-1 and -11 die of parasite overgrowth early in chronic infection (Ewald et al., 2014; Gorfu et al., 2014). Disabling the IFN-γ-regulated GTPase (IRG) system, comprised of GTPases which identify modified intracellular membranes and target them for degradation, leads to rapid parasite expansion and host death in inbred laboratory mice in a manner that depends on parasite genetics. Type II Toxoplasma, the most frequently isolated type in Europe and North America, expresses alleles of the effector proteins ROP5 and ROP18 which make it susceptible to IRG-mediated clearance. Hypervirulent type I and hypovirulent type III Toxoplasma express alleles of rop5 and rop18 that inactivate IRGs at the parasite vacuole improving intracellular replication in many inbred strains of mice (Reese et al., 2011; Etheridge et al., 2014). However, some wild-derived mouse strains are able to restrict growth of hypervirulent type I parasites in an IRG-dependent manner, suggesting that the IRG restriction pathway has imparted an important selective pressure on both host and parasite evolution (Lilue et al., 2013).

Innate immune recognition of Toxoplasma promotes IL-12 release, NK cell-mediated IFN-γ production and a Th1 polarized adaptive immune response (Sher et al., 1993; Hunter et al., 1994). The cellular mechanisms of these responses and the importance of CD8<sup>+</sup> cytotoxic T cell response for host survival have been reviewed extensively elsewhere (Dupont et al., 2012; Sasai et al., 2018). Importantly, many cytokines and effector molecules central to this response have been shown to be necessary for parasite resistance. Specifically, mice deficient in IL-12 (Yap et al., 2000), IFN-γ (Suzuki et al., 1988; Deckert-Schlüter et al., 1996), IL-6 (Jebbari et al., 1998), TNF-α (Schlüter et al., 2003), iNOS (Scharton-Kersten et al., 1997), or their receptors die from parasite overgrowth in acute infection or early chronic infection. Host haplotype also plays a dominant role in parasite resistance. BALB/c mice are resistant to infection and express the H-2L<sup>d</sup> MHC haplotype, which presents an immunodominant peptide from the dense granule protein GRA6 (Blanchard et al., 2008). In contrast, C57BL/6 mice have the H2b haplotype which presents ROP5, a lower abundance rhoptry protein that elicits a weak CD8<sup>+</sup> T cell response, resulting in worse parasite restriction (Grover et al., 2014).

A number of immune cell-intrinsic pathways have been identified that mitigate Th1-mediated immunopathology during Toxoplasma infection which are relevant to disease tolerance (**Table 1**). At acute infection, mice deficient in IL-10 had a similar liver parasite burden as wildtype mice, but died from a cytokine storm of IL-12, TNF-α, and IFN-γ (Gazzinelli et al., 1996; Neyer et al., 1997). This phenotype could be reversed by depleting CD4+ T cells (Gazzinelli et al., 1996), or by depleting IFN-γ (Suzuki et al., 2000). More recently, Jankovic et al. showed that T-bet+Foxp3<sup>−</sup> Th1 cells were a major source of IL-10 during chronic i.p. infection with 20 ME49 cysts. These IL-10-producing cells were required to limit fatal immunopathology during both acute and chronic infection (Jankovic et al., 2007). In 2005, Wilson et al. found that lethal Toxoplasma encephalitis in chronically infected IL-10−/<sup>−</sup> mice is not due to higher parasite loads, implicating IL-10 as a tolerance effector in the central nervous system (CNS) during Toxoplasma infection (Wilson et al., 2005). Similarly, mice lacking IL-4 were more susceptible to Toxoplasma infection despite having slightly reduced cyst burdens and microglial nodules in the brain at chronic infection (Roberts et al., 1996). These data suggest that tolerance mechanisms induced during infection are critical to the survival of host. Although the mechanistic basis for the effect of IL-4 is currently unclear, it has been wellstudied in other systems. One example is the rodent helminth Nippostrongylus brasiliensis. N. brasiliensis burrows through the skin then migrates to the lung where it is coughed up and swallowed to complete its life cycle in the gastrointestinal tract. N. brasiliensis chitin promotes IL-4, driving alveolar macrophages to limit lung pathology by producing insulin-like growth factor 1, resistin-like molecule-α and arginase-1 (Chen et al., 2012). These effectors dampen inflammation and promote the fibroblast wound repair response (Knipper et al., 2015). Importantly, the parasite's egg-laying stage occurs in the gut, after trans-lung migration. There may be significant selective pressure for N. brasiliensis to activate tolerance programs that promote wound healing and host survival so that the parasite has sufficient time for transmission.

In addition to their well-characterized roles in pathogen clearance and adaptive immune response activation during Toxoplasma infection, innate immune cells also play a critical role in mitigating tissue pathology. A 2013 study from Yasmine Belkaid's laboratory found that lamina propria-infiltrating neutrophils generated "casts" containing cells and extracellular DNA over regions of small intestine damaged by per oral (p.o.) Toxoplasma infection. These casts which appeared to limited bacterial translocation into host circulation. Depleting neutrophils with an α-Gr-1 antibody increased mortality without increasing parasite burden, indicating that the immunecommensal axis also plays an important role in disease tolerance during Toxoplasma infection (Molloy et al., 2013). During p.o. Toxoplasma infection, lamina propria-infiltrating Ly6Chi monocytes were also shown to release IL-10 and prostaglandin E<sup>2</sup> (PGE2) which limited neutrophil-induced pathology. Administration of a PGE<sup>2</sup> analog during infection was sufficient to reduce intestinal tissue pathology and immune infiltration in the absence of monocytes without affecting parasite burden (Grainger et al., 2013).

T cells are important players in disease tolerance mechanisms during Toxoplasma infection. 5-lipoxygenase is the rate-limiting enzyme in the synthesis of lipoxin A4, an eicosanoid mediator of the resolution phase of an inflammatory response (Colgan et al., 1993). The importance of this pathway in disease tolerance is underscored by the observation that 5-lipoxygenase deficient mice succumbed to i.p. infection with 20 ME49 cysts by 35 days post-infection, despite harboring significantly fewer brain cysts TABLE 1 | Summary of literature describing immunoregulatory pathways during Toxoplasma gondii infection.


than wild type mice. Death was associated with encephalitis, increased T cell infiltration into the brain and elevated circulating IL-12, IFN-γ, and TNF-α (Aliberti et al., 2002). IL-27 is part of the IL-12 family of JAK-STAT signaling cytokines with an emerging role in tolerance to Toxoplasma infection. WSX-1 knockout mice (which lack the IL-27 receptor) i.p. infected with 20 ME49 cysts succumbed by 13 days post-infection. No change in parasite burden was detected between knockout and wild type mice in peritoneal lavage fluid at 7 days post-infection, but knockout mice had elevated circulating IL-12, IFN-γ, increased inflammation and necrosis in the liver and lungs, and hyperactive and hyperproliferative CD4<sup>+</sup> T cells in the spleen. Survival was partially rescued when CD4<sup>+</sup> T cells were depleted (Villarino et al., 2003). Subsequently, the same group found that IL-27 limited production of IL-2 by activated CD4<sup>+</sup> T cells, implicating IL-27 receptor as a negative regulator of T cell responses during Toxoplasma infection (Villarino et al., 2006). This conclusion was also supported by a 2012 study from the Hunter Laboratory where mice were p.o. infected with 100 ME49 cysts or i.p. infected with 20 cysts. IL-27 promoted development of Tregs at primary sites of infection, the small intestine or peritoneal cavity, respectively. IL-27−/<sup>−</sup> mice succumbed to acute infection, which could be partially rescued by adoptive transfer of Treg cells (Hall et al., 2012). Although this result is consistent with a tolerance strategy, parasite burden was not directly measured.

IL-2 is also important for Treg biology in Toxoplasma infection. Using p.o. infection, several groups have shown a transient, but significant loss of Treg function in the small intestine (Oldenhove et al., 2009; Benson et al., 2012). Addition of IL-2 promoted Treg survival and prevented liver pathology, consistent with a role in disease tolerance in the liver. However, IL-2 treatment also resulted in a higher brain cyst burden (Oldenhove et al., 2009). Similarly, Benson et al. found that IL-2 treated mice had more Tregs and significantly higher brain cyst burdens causing lethality (Benson et al., 2012). Whether the inability to restrict cerebral Toxoplasma infection in these studies was due to Treg-mediated suppression of effector T cells or a direct effect of IL-2 on naïve or effector T cell activity is unclear. However, directly targeting effector T cell responses has been shown to reduce tissue damage in a mouse model of ocular toxoplasmosis. Specifically, in the retina both infiltrating antigenpresenting cells (MHCII expressing) and tissue resident retinal cells expressed PD-L1, the ligand for the T cell inhibitory receptor PD-1. PD-L1 expressing retinal cells were able to suppress splenic T cell proliferation ex vivo using a co-culture system with Toxoplasma-antigen loaded dendritic cells (Charles et al., 2010). These results suggest that retinal cells may be able to moderate local immune responses and reduce tissue damage by directly suppressing T cell activation in the eye. Cumulatively, these studies underscore the importance of targeting the T cell response to limit tissue damage and restrict parasite growth.

#### HOST METABOLIC DYSREGULATION IN TOXOPLASMA INFECTION

There is a growing appreciation that immune responses are intimately linked with shifts in metabolic homeostasis. This is a critical arm of pathogen resistance, for example, T cell activation requires a glycolytic burst; anorexia during infection mobilizes glycogen and lipid stores to support gluconeogenesis for the immune system; metabolic shifts are also used to sequester trace metals or nutrients to limit pathogen growth, often referred to as nutritional immunity (Hood and Skaar, 2012; Núñez et al., 2018). Shifts in nutrient utilization are also important mediators of disease tolerance. For example, diet restriction improves fruit fly survival during S. typhimurium infection without affecting bacterial load (Ayres and Schneider, 2009).

There is a growing body of literature showing metabolic shifts in the host during acute Toxoplasma infection. However, a role for these shifts in disease tolerance has not been addressed specifically. Using Swiss-Webster mice infected p.o. with 8 cysts of the human-derived Type II strain BGD-1, one group reported reduced circulating cholesterol and HDL at 14 days post infection (Milovanovic et al., ´ 2017). Untargeted proteomic analysis of sera isolated from BALB/c mice infected per orally with 10 Type II Pru cysts showed increased circulating amino acids and reduced choline levels in the first 21 days of infection (Zhou et al., 2016). The same group observed an increase in cholinederived phosphatidylcholine and phosphatidylethanolamine in the brains of infected mice (Zhou et al., 2015). Choline is an essential dietary nutrient as the precursor of phosphatidylcholine and the neurotransmitter acetylcholine. Phosphatidylcholine is a major component of cell membranes and is necessary for the packaging and export of very-low-density lipoproteins (VLDL) from the liver (Corbin and Zeisel, 2012). Together, these studies suggest that Toxoplasma infected mice shift toward amino acid and fat metabolism and away from glycolytic metabolism during Toxoplasma infection. This is consistent with a well-described, although transient, period of anorexia during acute infection in Toxoplasma infected mice which would induce such programs (Arsenijevic et al., 1997; Weiss and Dubey, 2009; Jin et al., 2017; Hatter et al., 2018). In mice infected with L. monocytogenes, anorexia-induced ketogenesis protected tissues from oxidative stress, whereas glucose supplementation increased mortality. Interestingly, the opposite effect was observed with influenza infection, suggesting that the effectiveness of tolerance programs, like restriction mechanims, depend on the pathogen (Wang et al., 2016). Similar experiments to look at the role of glycolytic and beta-oxidative host metabolism in Toxoplasma will be necessary to determine if anorexia respresents a host tolerance strategy or if the parasite capatilizes on altered host metabolism for infection.

It is well-established that scavenging host lipids is necessary for Toxoplasma survival in vivo and in vitro. For example, most of the cholesterol in Toxoplasma is derived from scavenged host LDL (Bisanz et al., 2006). Metabolic labeling with <sup>14</sup>C-acetate showed that Toxoplasma must scavenge fatty acid precursors from its host to synthesize its full range of lipids (Sehgal et al., 2005). Recent studies indicate that Toxoplasma competes with the host cell for lipids at the level of lipid droplet recruitment, mitochondrial interaction, and vesicular transport for intracellular survival (Hu et al., 2017; Nolan et al., 2017; Pernas et al., 2018). Further studies are warranted to determine whether host metabolic shifts that mobilize lipid stores at a systemic (rather than a cellular) level benefit Toxoplasma. By contrast, there is growing evidence that altered metabolism has a long-term negative impact on the host during Toxoplasma infection. A series of studies by Arsenijevic et al. demonstrated that mice orally infected with 10 Me49 cysts undergo anorexiaassociated hypermetabolism during acute infection. Nearly half of the infected cohort could not regain weight as they progressed to chronic infection. These "non-gainers" harbored a higher parasite load than mice that regained weight and sustained the hypermetabolic phenotype along with elevated circulating inflammatory cytokines (Arsenijevic et al., 1997, 2001). A subsequent study showed that non-gainers challenged with LPS had a more severe inflammatory response, worse pathology, and a longer rebound period than infected mice that gained weight (Arsenijevic et al., 1998). Related, a study by Kugler et al. showed that Toxoplasma infection led to long-term defects in thymus and lymph node structure, hindering naïve T cell responses to subsequent viral challenge (Kugler et al., 2016). More recently, the Wohlfert lab showed that oral infection with 5 Me49 cysts causes acute weight loss in mice and inability to regain weight as chronic infection progresses (Jin et al., 2017). Surprisingly, muscle inflammation and myositis were driven by Tregs, a population classically associated with tolerance, suggesting that sustained tolerance programs may negatively impact the host. In 2018 Hatter et al. similarly reported that C57BL/6J mice orally infected with Me49 cysts develop chronic cachexia defined by lean muscle and fat wasting and chronic elevation of circulating innate cytokines. Although cachectic mice recovered from acute ileitis, dysbiosis in the intestinal commensal microbiota population was sustained (Hatter et al., 2018). Loss of total gut commensal diversity and a shift toward "pathobiotic" Gram negative species is well-established in acute Toxoplasma infection (Heimesaat et al., 2006; Benson et al., 2009; Molloy et al., 2013; Hatter et al., 2018). Although the precise composition of outgrowth species is animal colony-dependent, several groups have reported E. coli outgrowth (Benson et al., 2009; Raetz et al., 2012; Molloy et al., 2013; Fonseca et al., 2015). This is interesting in the context of the Zhao Lab metabolomics data because these bacteria are major metabolizers of the choline derivative ethanolamine (Garsin, 2012). It is highly plausible that shifts in commensals influence nutrient availability, in addition to playing a better-described role in influencing host immunity during Toxoplasma infection

Toxoplasma interactions with major organs involved in nutrient regulation may have important implications for host metabolism as well. The liver coordinates dietary nutrient uptake (bile acid recycling), availability and storage (fat and glycogen), and detoxifies the blood. Liver resident macrophages called Kupffer cells screen incoming blood for pathogens and intestinal microbes that leak from the gut during intestinal inflammation, including during Toxoplasma infection (Molloy et al., 2013). Toxoplasma has been detected in the livers of mice at acute infection using bioluminescence, histology, and PCR in a number of studies using Type II strains (Silva et al., 2002; Di Cristina et al., 2008; Zhou et al., 2015; He et al., 2016a,b). In the first week of infection, Toxoplasma has been observed replicating in hepatocytes near regions of inflammatory infiltrate and focal necrosis in both Swiss-Webster and BALB/c mice (Atmaca et al., 2013; Bottari et al., 2014). Interestingly, Atmaca et al. also reported an expansion of hepatic stellate cells during Toxoplasma infection (Atmaca et al., 2013). During inflammation, hepatic stellate cells differentiate into myofibroblasts and produce extracellular matrix which is consistent with the induction of a tissue remodeling tolerance program during Toxoplasma infection. However, an important caveat with these studies is that they were performed with lethal doses of the hypervirulent Type I strain RH raising the question of relevance to hepatic infection with strains that are more commonly found in mice or humans. Using clinically isolated Type II strains, Toxoplasma cysts were found in the livers of infected Swiss-Webster mice as late as 33 weeks post-infection, indicating that the liver may be a reservoir for chronic infection (Autier et al., 2018). This is consistent with clinical reports of Toxoplasma-negative transplant recipients who have developed toxoplasmosis after receiving livers from sera positive donors (Assi et al., 2007; Galván-Ramírez et al., 2018). While there are no studies directly implicating Toxoplasma in the development of liver disease, approximately 30% of patients with chronic liver disease test sera-positive for Toxoplasma B1 compared to 6% in control populations (El-Sayed et al., 2016). These Toxoplasma-infected patients had significantly elevated circulating ALT and AST, clinical markers of liver damage, compared to uninfected patients with liver disease. Although these data do not imply causality, they are consistent with the interpretation that chronic Toxoplasma infection may negatively impact host fitness in diseases of co-occurrence.

Toxoplasma infection may also change the metabolic landscape of the liver. In a recent liver proteomics study, BALB/c mice intraperitoneally infected with 200 type II PYS tachyzoites had reduced signatures of fatty-acid oxidation proteins and an upregulation of cell death, inflammatory, and stress response pathways at 6 days post-infection (He et al., 2016b). A parallel liver transcriptomics study reported down-regulation of gene families involved in lipid metabolism, cholesterol and bile synthesis, and amino acid metabolism, with an increase in inflammatory transcripts (He et al., 2016a). Together, these studies provide evidence that Toxoplasma occupies a liver niche in acute and chronic infection and may directly contribute to shifts in liver metabolic homeostasis.

Adipose tissue depots are another important site of calorie storage and immune regulation. Fat tissues have long been described as an anti-inflammatory environment that become inflamed in diseases of the adipose tissue, including obesity and diabetes. As better tools have become available to survey low pathogen loads in tissue, researchers have begun to appreciate the frequency of microbial translocation to and persistence in this nutrient-rich environment. M. tuberculosis, T. brucei, and facultative pathogen strains of E. coli have been detected in this niche, indicating that it serves as a reservoir of infection for many pathogens (Neyrolles et al., 2006; Schieber et al., 2015; Trindade et al., 2016). More recently, Toxoplasma has been reported in the visceral fat following intraperitoneal infection and oral infection by bioluminescence assay and PCR (Di Cristina et al., 2008). Interestingly, using a stage-specific luciferase reporter system, di Cristina et al. showed that parasites in visceral fat expressed GFP driven by the SRS9 promoter, a gene product enriched during bradyzoite differentiation (Di Cristina et al., 2008). Future studies

#### REFERENCES


will be necessary to determine if fat is a long-term reservoir for Toxoplasma, and whether colonization of the fat is related to host metabolic shifts.

#### CONCLUSIONS

Toxoplasma gondii has evolved sophisticated mechanisms to evade sterilizing immunity, yet activating a robust immune response is necessary to ensure host survival long enough for Toxoplasma encystation and transmission (Bohne et al., 1993, 1994). Disease tolerance programs are adaptations to cell biology and metabolism that allow tissues to function in the harsh environment of an inflammatory response (Tzelepis et al., 2018). However, tolerance adaptations must come at a cost to the host, otherwise they would be selected for homeostatic use. The literature reviewed here are consistent with a model where tolerance programs initiated in acute Toxoplasma infection, including immune-microbiota interactions, T cell-mediated responses and metabolic shifts fail to return to homeostasis in chronic infection. Emerging studies suggest that these shifts in homeostasis have sustained negative consequences for the host, including muscle wasting, and impaired responses to secondary immune stimuli (Arsenijevic et al., 1997; Kugler et al., 2016; Jin et al., 2017; Hatter et al., 2018). Whether or not these shifts in homeostasis confer a benefit to the parasite is an open question. Compromising rodent fitness in these ways would likely enhance the opportunity for predation by felines, the parasite's definitive host. Passage through a cat is extremely advantageous for Toxoplasma because feline hosts facilitate genetic diversity and range expansion. In this way, Toxoplasma may benefit from promoting tolerance programs that ensure host survival during the acute phase of infection, during the tachyzoite to bradyzoite transition, but ultimately impair host fitness in the long term.

#### 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 supported by NIH NIAID K22 AI116727-01 (SE) and T32 AI7496-23 (SM).


in IFN-γ-dependent elimination of Paneth cells. Nat. Immunol. 14, 136–142. doi: 10.1038/ni.2508


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

Copyright © 2019 Melchor and Ewald. 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.

# Long-Term Impact of *Toxoplasma gondii* Infection on Human Monocytes

#### Hauke G. Ehmen and Carsten G. K. Lüder\*

Institute for Medical Microbiology, University Medical Center Goettingen, Georg-August-University, Goettingen, Germany

Toxoplasma gondii is a prevalent parasite of mammals and birds including up to 30% of humans world-wide. Primary infection of immunocompetent hosts leads to a robust cell-mediated immune response, which controls but does not clear the infection, thus enabling long-term parasite persistence in brain and muscle tissues. Chronic toxoplasmosis in mice is associated with resistance to heterologous pathogens and this has been related to increased numbers of inflammatory monocytes. Here we have analyzed whether chronic T. gondii infection impacts the subset distribution and the phenotype of peripheral human monocytes in vivo and their responses to parasite infection in vitro. CD14<sup>+</sup> monocytes from T. gondii-seropositive blood donors expressed significantly less FcγRIII (CD16) than those from seronegative controls, but they did not show a shift in the distribution of classical, intermediate and non-classical monocyte subpopulations. Percentages of CD62L<sup>+</sup> and CD64<sup>+</sup> monocytes were however decreased and increased, respectively, in chronically infected individuals as compared to naïve controls. Infection of monocyte-enriched PBMCs from both seropositive and seronegative individuals with T. gondii led to an increase of CD14+CD16<sup>−</sup> classical monocytes and a decrease of CD14+CD16<sup>+</sup> double positive monocytes. Remarkably, after in vitro parasite infection, expression of the chemokine receptor CCR2 was severely impaired in monocytes from both, individuals with chronic toxoplasmosis and seronegative controls. In contrast, only monocytes from chronically infected humans but not those from controls dose-dependently up-regulated HLA-DR, DP, DQ expression following in vitro infection. Furthermore, monocyte-enriched PBMCs from seropositive individuals up-regulated IL-12 mRNA more vigorously after in vitro infection than cells from naïve controls. Collectively, our results establish that infection of humans with T. gondii exerts long-term effects on the phenotype and responsiveness of blood monocytes. This may have important implications for innate immune responses to T. gondii and unrelated pathogens.

Keywords: toxoplasmosis, humans, monocytes, monocyte subsets, cytokines, surface markers, chronic infection

# INTRODUCTION

The intracellular parasite Toxoplasma gondii is widespread in birds and mammals including an estimated 30% of humans world-wide. Infections of immunocompetent hosts are mostly asymptomatic or benign but they lead to parasite persistence for months to years and possibly even for the hosts' life. Reactivation of chronic toxoplasmosis in immunocompromised individuals,

#### *Edited by:*

Eva Frickel, Francis Crick Institute, United Kingdom

#### *Reviewed by:*

Björn Felix Caesar Kafsack, Cornell University, United States Renato Augusto DaMatta, Universidade Estadual do Norte Fluminense Darcy Ribeiro, Brazil Bellisa Freitas Barbosa, Federal University of Uberlandia, Brazil

> *\*Correspondence:* Carsten G. K. Lüder clueder@gwdg.de

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology *Received:* 04 February 2019

*Accepted:* 14 June 2019 *Published:* 28 June 2019

#### *Citation:*

Ehmen HG and Lüder CGK (2019) Long-Term Impact of Toxoplasma gondii Infection on Human Monocytes. Front. Cell. Infect. Microbiol. 9:235. doi: 10.3389/fcimb.2019.00235 e.g., those with AIDS or under immunosuppressive therapy, can however lead to necrotizing tissue damage and lifethreatening Toxoplasma encephalitis (Montoya and Liesenfeld, 2004). Furthermore, primary infection of pregnant women can lead to intrauterine transmission with severe symptoms or even death of the fetus or significant sequelae after birth. T. gondii is also an important cause of posterior uveitis after infection of otherwise healthy adolescents or adults (Pleyer et al., 2014). Together, human toxoplasmosis has been recognized as one of the leading food-borne infectious diseases in the USA based on annual costs and loss of quality-adjusted life years (Hoffmann et al., 2012).

After primary infection, infectious sporozoites or bradyzoites penetrate enterocytes and transform within cells of the lamina propria to metabolically highly active tachyzoites. Tachyzoites disseminate throughout the host by several rounds of fast replication within various host cells including monocytes and dendritic cells (DCs). These innate immune cells sense the parasite by Toll-like receptor (TLR)-dependent (Yarovinsky et al., 2005; Debierre-Grockiego et al., 2007; Andrade et al., 2013) and –independent mechanisms (Witola et al., 2011; Gov et al., 2013; Ewald et al., 2014; Gorfu et al., 2014) in a partially host speciesdependent manner (Sher et al., 2017). They subsequently secrete interleukin (IL)-12 that is critical to stimulate early interferon (IFN)-γ production by natural killer (NK) cells (Suzuki et al., 1988; Gazzinelli et al., 1993, 1994). Priming and differentiation of CD4<sup>+</sup> and CD8<sup>+</sup> T lymphocytes further contribute to robust IFN-γ secretion (Suzuki and Remington, 1988; Gazzinelli et al., 1991) which in turn activates hematopoietic and nonhematopoietic cells (Yap and Sher, 1999) to exert various effector mechanisms in a host cell type- and host speciesspecific manner (Yarovinsky, 2014; Krishnamurthy et al., 2017). Importantly, some tachyzoites are able to evade killing by stage differentiation to latent and slowly to non-replicating bradyzoites predominantly within neurons and muscle cells. Bradyzoites rebuild the PV to an intracellular tissue cyst that is surrounded by a robust cyst wall and that marks the chronic phase of infection. The immune response to the bradyzoite stage is largely unexplored, but it is clear that IL-12 (Yap et al., 2000) and IFN-γ secretion by CD8<sup>+</sup> and CD4<sup>+</sup> T lymphocytes (Gazzinelli et al., 1992) are critical to control the chronic phase of infection. What remains unknown however is whether such parasite control is mediated by immune responses to the bradyzoite stage or rather by killing of tachyzoites which emerge after occasional re-differentiation of the latent stage and tissue cyst rupture.

Remarkably, mice infected with T. gondii are more resistant to infections with Listeria monocytogenes or Salmonella enterica ser. Typhimurium than non-infected controls (Ruskin and Remington, 1968; Neal and Knoll, 2014). Resistance against heterologous pathogens persists for several months and appears to be mediated by macrophages (Ruskin and Rengton, 1968; Ruskin et al., 1969). Mechanistic studies recently showed that Ly6C<sup>+</sup> "inflammatory" monocytes which are recruited after stimulation of TLR11 by T. gondii profilin (TgPRF) in a CCR2 dependent manner are able to confer resistance against bacterial infection in mice (Neal and Knoll, 2014). Ly6C<sup>+</sup> monocytes are recruited to the site of infection during infections with several intracellular pathogens including L. monocytogenes and T. gondii (Robben et al., 2005; Serbina and Pamer, 2006) and are critical to control these pathogens (Dunay et al., 2008; Serbina et al., 2012), presumably by directly exerting antimicrobial effector mechanisms. TgPRF is a soluble protein functioning in actin remodeling in T. gondii (Plattner et al., 2008). It is a major IL-12 inducer after being bound to TLR11/TLR12 on DCs from mice (Yarovinsky et al., 2005; Plattner et al., 2008) but not humans who lack these receptors. Together, these results suggest that in T. gondii-infected mice distinct monocyte populations can confer lasting resistance to heterologous infections.

In humans, three distinct monocyte subpopulations have been described based on their expression of CD14 and CD16 (Ziegler-Heitbrock et al., 2010). The so-called classical monocytes account for ∼85% of the circulating monocyte pool under steady state conditions, they are CD14<sup>+</sup> and CD16−, and appear to be the human counterparts of Ly6C+CD43lowCCR2high monocytes from mice (Ziegler-Heitbrock et al., 2010). The remaining human monocytes are the non-classical ones which are CD14dimCD16<sup>+</sup> and those with an intermediate phenotype (CD14+CD16+). Additional surface markers have been proposed to distinguish between these monocyte subsets (Ingersoll et al., 2010; Ziegler-Heitbrock et al., 2010; Patel et al., 2017). Of note, after emigration of classical monocytes from the bone marrow in a CCR2 dependent manner, they appear to sequentially develop into intermediate and then non-classical monocytes, indicating that monocytes are quite dynamic (Patel et al., 2017) and subset boundaries not always easy to define (Ziegler-Heitbrock et al., 2010). Importantly, the distribution of monocyte subsets and their functional properties change during distinct pathologies including infectious and non-infectious inflammatory diseases (Fingerle et al., 1993; Fingerle-Rowson et al., 1998; Horelt et al., 2002; Patel et al., 2017).

Here, we have analyzed the subset distribution of peripheral monocytes and their phenotypic and functional properties from chronic toxoplasmosis patients and from naïve control individuals. Results for the first time indicate differences in the repertoire of monocyte surface markers expressed during human chronic toxoplasmosis but not a shift in the distribution of the three major subsets as compared to control individuals. In vitro infection of monocyte-enriched PBMCs led to an expansion of the classical monocyte subset from both chronically infected and naïve individuals, and to an up-regulation of HLA-DR,DP,DQ, and a more vigorous IL-12 response specifically in cells from chronic toxoplasmosis patients. Long-term effects of T. gondii infection on innate immune cells of humans can have important consequences on their reactivity to homologous and heterologous pathogens.

# MATERIALS AND METHODS

#### Study Population and Ethics

Buffy coats and plasma from heparinized blood of healthy volunteers were obtained from the Blood Donation Service Center of the University Medical School Göttingen, Germany. Study subjects were excluded from donating blood when presenting pathologies indicative for transmissible infectious diseases, being at risk of having blood products-transmitted infectious diseases, or presenting severe non-infectious diseases which precluded regular blood donation. All donors gave written informed consent, and the study was approved by the Ethics commission of the University Medical School Göttingen (Project numbers 28/3/08 and 1/3/13). Plasma was obtained by centrifugation of blood aliquots at 3,000 × g for 5 min. Serological tests on plasma samples for detection of T. gondiispecific total immunoglobulins, T. gondii-specific IgG, IgM and IgA and for measuring the avidity of anti-T. gondii-specific IgG were performed at the Institute for Medical Microbiology (Göttingen, Germany) using standardized routine procedures.

#### Isolation of PBMCs and Enrichment of Monocytes

Peripheral blood mononuclear cells (PBMCs) were isolated by Ficoll-Paque Plus (GE Healthcare Life Sciences, Freiburg, Germany) density centrifugation of blood diluted in RPMI 1640 medium at 900 × g for 30 min. They were extensively washed and monocytes were then enriched by allowing them to adhere to polysterene cell culture dishes (Greiner Bio-One, Frickenhausen, Germany) in RPMI 1640 medium, 100 U/ml penicillin and 100µg/ml streptomycin for 2 h at 37◦C and subsequent removal of non-adherent cells. The monocyte-enriched PBMCs were then directly analyzed by fluorescence-activated cell sorting (FACS), or they were incubated in RPMI 1640, 10% heat-inactivated fetal calf serum and antibiotics as above for infection assays at 24 h after isolation.

#### *T. gondii* and Parasite Infection

The mouse-avirulent T. gondii type II strain NTE (Gross et al., 1991) was used throughout this study. Tachyzoites were propagated in L929 fibroblasts as host cells. For infection assays, freshly egressed parasites were separated from host cells by differential centrifugation as described previously (Lang et al., 2006). After having been extensively washed, they were added to monocyte-enriched PBMCs at multiplicities of infection (MOI) of 3:1 and 6:1 for 24 h.

# Flow Cytometry

Expression of cell surface markers by human monocytes was quantified by FACS. To this end, freshly isolated monocyteenriched PBMCs or those cultivated and/or infected with T. gondii in vitro were detached from tissue culture dishes using cell scrapers and were washed twice in PBS, pH 7.4. For each staining, 500,000 cells were transferred into the wells of a 96-well-V-bottom microplate and unspecific binding sites were blocked with 1% human AB serum, 1% bovine serum albumin (BSA), 0.1% NaN<sup>3</sup> in PBS, pH 7.4 during 30 min at 4◦C. Cells were then co-stained for 30 min at 4◦C using FITC-conjugated mouse anti-human CD14 (clone M5E2) and PE-conjugated mouse anti-human CD16 (clone 3G8) or appropriate isotype control antibodies (clones G155-178 and MOPC-21; all antibodies from BD Biosciences, Heidelberg, Germany; diluted at 1:5 in 1% BSA, 0.1% NaN<sup>3</sup> in PBS). Alternatively, they were labeled with 2µg/mL of mouse monoclonal antibodies directed against human CD62L (clone DREG-56), human CD64 (clone 10.01.13), human HLA-A,B,C (clone G46-2.6), human HLA-DR,DP,DQ (clone Tu39; all antibodies from BD Biosciences), or directed against human CCR2 (R&D Systems, Wiesbaden-Nordenstadt, Germany), or they were incubated with appropriate isotype control antibodies (clones MOPC-21, 27-35 and G155-178; BD Biosciences) for 30 min at 4◦C. After having been washed three times in 1% BSA, 0.1% NaN<sup>3</sup> in PBS, pH 7.4, immune complexes were labeled with R-PE-conjugated goat F(ab')<sup>2</sup> fragment antimouse IgG (diluted at 1:50 in 1% BSA, 0.1% NaN<sup>3</sup> in PBS for 30 min at 4◦C). After immunolabeling, cells were washed (as above) and were then fixed using 1% paraformaldehyde in PBS, pH 7.4. Ten thousand cells per sample were analyzed using a FACSCalibur (BD Biosciences).

#### RNA Isolation and Quantitative Reverse Transcriptase-PCR

Total RNA was isolated from monocyte-enriched PBMCs after their isolation or after cultivation and/or infection with T. gondii in vitro, using the GenElute Mammalian Total RNA Miniprep kit (Sigma-Aldrich, Taufkirchen, Germany). Contaminating genomic DNA was digested with DNase I (amplification grade; Sigma-Aldrich) as recommended by the manufacturer. After reverse transcription of mRNA using the Omniscript RT kit (Qiagen, Hilden, Germany) and oligo(dT) primers, cDNAs were amplified by LightCycler quantitative PCR using the LightCycler FastStart DNA MasterPLUS SYBR Green I kit as recommended (Roche, Mannheim, Germany). Human transcripts were amplified using primer pairs specific for il-12b (p40) (forward: 5′ -ATGCCCCTGGAGAAATGGTG-3 ′ ; reverse: 5′ -GAACCTCGCCTCCTTTGTGA-3′ ), il-10 (forward: 5′ - GGCGCTGTCATCGATTTCTTC−3 ′ ; reverse: 5 ′ - TAGAGTCGCCACCCTGATGT−3 ′ ), tnfa (forward: 5 ′ - GCCCATGTTGTAGCAAACCC−3 ′ ; reverse: 5′ - GGAGGTTGACCTTGGTCTGG−3 ′ ) and the reference gene β-actin (forward: 5′ - TATCCAGGCTGTGCTATCCC−3 ′ ; reverse: 5′ - CCATCTCTTGCTCGAAGTCC−3 ′ ). The relative gene expression levels were calculated as the fold changes between freshly isolated (0 h) and in vitro-cultivated, noninfected cells (48 h, non-infected) or between freshly isolated (0 h) and T. gondii-infected cells (48 h, T. gondii) using the formula Ratio = (Etarget) 1CPtarget(0hours−48hours/±T. gondii) / (Eref) 1CPref(0hours−48hours/±T. gondii) , where E denotes real-time PCR efficiencies and 1CP the crossing point (CP) differences (Pfaffl et al., 2002).

# Statistical Analyses

Outlyers within datasets were identified by Dixon's Q-test and were not further considered. Data are presented as means ± S.E.M. or as box-whisker plots indicating the mean, median and percentiles as indicated. Individual data points have also been included within graphs. Significant differences were identified by Student's t-test or by ANOVA of between groups and repeated measures matrices using the General ANOVA/MANOVA module of Statistica (Version 13.3.; TIBCO Software, Palo Alto, CA, USA). P-values of <0.05 were considered significant.

# RESULTS

#### Study Subjects

Buffy coats from 21 healthy blood donors were included into this study. Screening of plasma from blood samples using the VIDAS <sup>R</sup> Toxoplasma IgG/IgM competition kit (bioMerieux, Nürtingen, Germany) identified specific antibodies in 5 out of 21 individuals (23.8%; **Supplementary Table 1**). Subsequent differentiation of the positive samples revealed Toxoplasmaspecific IgG in all samples (VIDAS <sup>R</sup> TOXO IgG II; 24–>300 IU/ml), Toxoplasma-specific IgM in two of them (VIDAS <sup>R</sup> TOXO IgM; index values 0.94 or 1.95), and no specific IgA in any sample (Platelia Toxo IgA (TMB); BioRad, München, Germany). Furthermore, high avidities were found for Toxoplasma IgG antibodies from the two IgM-positive samples (VIDAS <sup>R</sup> TOXO IgG Avidity; **Supplementary Table 1**). Serological data thus indicated chronic (> 4 months) T. gondii infections in 5 blood donors whereas no signs of infection were found in 16 controls.

# Expression of Surface Markers Differs Between Monocytes From Chronic Toxoplasmosis Patients and Naïve Controls

Monocytes from human peripheral blood are classified into three subpopulations based on surface expression of CD14, i.e., the co-receptor for lipopolysaccharide (LPS), and CD16, i.e., the type III Fcγ receptor (FcγRIII) (Ziegler-Heitbrock et al., 2010). In order to distinguish between these subsets, plastic-adherent PBMCs were fluorescently labeled using FITC-conjugated anti-CD14 and PE-conjugated anti-CD16 reagents or appropriate isotype control antibodies and analyzed by FACS. Results confirmed three monocyte subsets, i.e., CD14+CD16<sup>−</sup> classical monocytes (R2 in **Figure 1A**), CD14+CD16<sup>+</sup> intermediate monocytes (R3) and CD14dimCD16<sup>+</sup> non-classical monocytes (R4) among the PMBCs from chronically infected toxoplasmosis patients and non-infected controls (**Figures 1A–D**). In contrast, isotype control antibodies did not specifically bind to these cells (**Figure 1A**). CD14+CD16<sup>−</sup> monocytes clearly predominated, with ∼85% of the total CD14<sup>+</sup> cells from T. gondii seronegative and seropositive individuals belonging to this subset (**Figure 1B**). CD14+CD16<sup>+</sup> intermediate monocytes accounted for 7.0 to 8.1% (means) and CD14dimCD16<sup>+</sup> non-classical monocytes for ∼4.75% of the total CD14<sup>+</sup> cells and this did not grossly differ between T. gondii-infected and non-infected individuals (**Figures 1C,D**). CD14+/dim monocytes were further analyzed for expression levels of CD16 and CD14 (see **Supplementary Figure 1** for a representative FACS analysis). Remarkably, expression levels of CD16 were significantly lower on monocytes from chronically infected toxoplasmosis patients than those from naïve controls (p = 0.009, Student's t-test; **Figure 1F**). We also recognized a trend toward lower CD14 levels on monocytes from T. gondii-positive individuals as compared to controls, but this did not reach statistical significance (p = 0.133; **Figure 1E**).

To further unravel phenotypic differences between monocytes from chronically infected toxoplasmosis patients and noninfected controls, we stained freshly isolated plastic-adherent PBMCs also for other cell surface proteins which have been considered as additional markers of monocyte subsets (Ingersoll et al., 2010; Ziegler-Heitbrock et al., 2010; Patel et al., 2017) or which are crucial for cellular functions. To this end, CD14<sup>+</sup> and CD14dim cells were backgated onto FSC vs. SSC to identify the monocyte population, and these cells were then analyzed for expression of L-selectin (CD62L), high affinity FcγRI (CD64), chemokine receptor CCR2 (CD192), MHC class I (HLA-A,B,C) and class II (HLA-DR,DP,DQ) (see **Figure 2A** and **Supplementary Figure 2** for representative examples of the gating strategy and labeling of different surface markers, respectively). Staining of the cells with isotype control antibodies were used to set thresholds for positive cells whereas expression levels were determined for R2-gated cells (**Figure 2A** and **Supplementary Figure 2**). Results revealed significantly less CD62L<sup>+</sup> monocytes in chronically infected toxoplasmosis patients as compared to naïve controls (**Figure 2B**). Expression levels of CD62L were also considerably lower on monocytes from seropositive individuals than on those from control individuals although this did not reach statistical significance (**Table 1**). CD62L is involved in adhesion of leukocytes to endothelial cells and facilitates early stages of emigration from the circulating blood toward the extravascular tissue (Rzeniewicz et al., 2015). It is more prevalent on mouse Ly6C<sup>+</sup> and human CD16<sup>−</sup> monocytes as compared to the respective Ly6Clow and CD16<sup>+</sup> counterparts (Ingersoll et al., 2010). Mean expression levels of CCR2, i.e., a surface marker of CD14+CD16<sup>−</sup> cells (Ingersoll et al., 2010) did not significantly differ between monocytes from seropositive individuals and those from seronegative controls (**Table 1**). Likewise, the proportion of CCR2<sup>+</sup> cells was similar among monocytes from both groups (**Figure 2D**). CD64<sup>+</sup> cells were to a small extent but significantly expanded within the monocyte population of patients with chronic toxoplasmosis as compared to that of seronegative controls (**Figure 2C**), but expression levels did not differ between both groups (**Table 1**). CD64 confers strong phagocytic activity and is mainly expressed on CD14+CD16<sup>−</sup> classical and CD14+CD16<sup>+</sup> intermediate but not on CD14dimCD16<sup>+</sup> non-classical monocytes (Grage-Griebenow et al., 2001; Ingersoll et al., 2010). Finally, monocytes from both T. gondii seropositive and seronegative individuals did not differ with respect to percentages or expression levels of HLA-A,B,C and HLA-DR,DP,DQ (**Figures 2E,F**; **Table 1**). Together, these results establish that monocytes of chronically infected toxoplasmosis patients differ from those of control individuals by surface expression or positivity of CD16, CD62L, and CD64, but that the distribution of the three major monocyte subsets was not altered.

### *Ex vivo* Infection With *T. gondii* Strongly Alters Monocyte Subpopulations and Expression of Surface Markers

Monocyte subsets fulfill distinct functions during steady state and pathological conditions (Grage-Griebenow et al., 2001; Auffray et al., 2009; Ziegler-Heitbrock, 2015). We therefore wondered how human primary monocytes respond to direct exposure to T. gondii and whether reactivity to infection differs between

and non-classical (R4; i.e., CD14dimCD16+) monocyte subsets. Staining with isotype control antibodies is shown in the right panel. (B–D) Percentages of CD14+CD16<sup>−</sup> (B), CD14+CD16<sup>+</sup> (C), and CD14dimCD16<sup>+</sup> (D) subsets among monocytes from T. gondii seropositive or seronegative individuals. Solid and dashed lines in the box-whisker plots indicate median and mean values, respectively; circles indicate individual data points. (E,F) Expression levels of CD14 (E) and CD16 (F) on CD14+/dim monocytes, i.e., cells within R2+R3+R4 from T. gondii seropositive or seronegative blood donors. Data are from 5 T. gondii seropositive and 16 seronegative blood donors; outlyers were excluded. Significant differences between both groups were identified by Student's t-test (\*\*p < 0.01).

monocytes from individuals who were previously exposed to the parasite or not. To this end, monocyte-enriched PBMCs were incubated in vitro for 48 h and were infected with T. gondii during the final 24 h or left non-infected (**Figure 3A**). They were then labeled with FITC-conjugated anti-CD14 and PEconjugated anti-CD16 and FACS-analyzed as above. Infection with T. gondii led to a dramatic expansion of CD14+CD16<sup>−</sup> classical monocytes (p < 0.001; ANOVA) with a simultaneous strong decrease of CD14+CD16<sup>+</sup> intermediate (p < 0.001) and a moderate decrease of CD14dimCD16<sup>+</sup> non-classical monocytes (p < 0.01), as compared to non-infected controls (**Figures 3B–D**). Importantly, the changes in subset distribution did not differ between monocytes from chronically infected toxoplasmosis patients and naïve controls, indicating a general response to the parasite. Furthermore, the parasite dose had no impact on the level of subset redistribution, since MOIs of 3:1 and 6:1 yielded almost identical results. An expansion of CD14+CD16<sup>−</sup> monocytes (p < 0.001) and a concomitant reduction of CD14+CD16<sup>+</sup> and CD14dimCD16<sup>+</sup> (p < 0.05 and p < 0.001, respectively) monocytes after infection became also

T. gondii-infected or non-infected using plasma from the blood samples. (A) CD14<sup>+</sup> monocytes (R1) were back-gated and identified (R2) among FSC/SSC-analyzed total cells. R2-gated cells were then analyzed for expression of cell surface markers as indicated, and positive cells were identified after specific (anti-CCR2 in (A); see Supplementary Figure 2 for the other surface markers) and isotype control labeling. (B–F) Percentages of monocytes from T. gondii seropositive or seronegative individuals positive for cell surface markers as indicated. Solid and dashed lines in the box-whisker plots indicate median and mean values, respectively; circles indicate individual data points. Data are from 5 T. gondii seropositive and 16 seronegative blood donors; outlyers were excluded.\*p < 0.05 (Student's t-test).

evident when results were compared to cells freshly isolated from the buffy coats (i.e., at time point 0 in **Figures 3B–D**). With respect to CD14+CD16<sup>−</sup> and CD14+CD16<sup>+</sup> subsets, these changes were however much lower when compared to those between non-infected controls and infected cells at 48 h of infection, since the mere in vitro incubation of monocytes for 2 days in the absence of T. gondii significantly impacted subset distribution (**Figures 3B–D**). It is interesting to note, that the changes in CD14dimCD16<sup>+</sup> non-classical monocytes between freshly isolated cells and parasite-infected cells by contrast even exceeded those observed between non-infected monocytes at 0 and 48 h (**Figure 3D**). Consistent with changes



\*Data are means ± S.E.M.

in subset distribution, expression levels of CD14 and CD16 were also altered in response to the parasite, with both markers being decreased as compared to non-infected controls (p < 0.001 or p < 0.01, respectively; ANOVA; **Figures 3E,F**). This response abolished the increase in CD14 and CD16 expression as observed after in vitro incubation for 48 h without T. gondii (p < 0.01). Again, CD14 and CD16 surface expression were similarly modulated in monocytes from both, chronic toxoplasmosis patients and from naïve controls. Of note however, those from T. gondii seropositive individuals generally expressed less CD14 than those from control individuals (p < 0.05; ANOVA; **Figure 3E**), thus confirming a trend that we already recognized in freshly isolated cells (see above and **Figure 1E**). In contrast, CD16 levels were only significantly lower on monocytes from chronically infected toxoplasmosis patients directly after isolation (time point 0 in **Figure 3F**; p < 0.05; also see **Figure 1F**).

The impact of in vitro infection with T. gondii was further examined by also immunolabeling other surface proteins as above and by applying a gating strategy as indicated in **Figure 2A**. Parasite infection did not alter percentages of CD62L<sup>+</sup> monocytes or their CD62L expression levels from both, chronically infected toxoplasmosis patients and naïve controls as compared to monocytes incubated without T. gondii (**Figures 4A,F**). CD62L expression levels however increased on cells from both groups during ex vivo cultivation for 48 h irrespective of being parasite-infected or not (**Figure 4F**; p < 0.001; ANOVA). Furthermore, monocytes from T. gondii seropositive individuals consistently showed a trend toward reduced percentages of CD62L<sup>+</sup> and lower levels of CD62L than those from seronegative controls although this did not reach statistical significance (**Figures 4A,F**). This extended results as observed with cells directly labeled after isolation (**Figure 2B**, **Table 1**). Expression levels of CD64 increased during in vitro infection of monocytes with T. gondii as compared to noninfected controls (p < 0.05; ANOVA), and this did not significantly differ between cells from chronic toxoplasmosis patients and naïve individuals (**Figure 4G**). Such increase did however not lead to a change in the proportion of CD64<sup>+</sup> monocytes after infection as compared to non-infected controls (**Figure 4B**). The proportion of CD64<sup>+</sup> cells after parasite infection in vitro instead even decreased among monocytes from both groups of individuals when compared to freshly isolated cells (i.e., those analyzed at 0 h) but this was not specific to parasite infection since it also occurred by in vitro cultivation for 48 h in the absence of T. gondii (**Figure 4B**; p < 0.001; ANOVA). In sharp contrast, expression of the chemokine receptor CCR2 and the percentages of CCR2<sup>+</sup> monocytes strongly diminished during in vitro infection when compared to non-infected control cells irrespective of whether monocytes originated from T. gondii seropositive or seronegative individuals (**Figures 4C,H**; p < 0.001; ANOVA). HLA-A,B,C was consistently up-regulated and the proportion of HLA-A,B,C<sup>+</sup> cells increased during 48 h of in vitro cultivation (**Figures 4D,I**; p < 0.001 or p < 0.01, respectively; ANOVA). These changes were however similar on monocytes infected in vitro with T. gondii and non-infected control cells, and it did also not differ between monocytes from chronically infected individuals or noninfected controls (**Figures 4D,I**). Finally, whereas percentages of HLA-DR,DP,DQ<sup>+</sup> monocytes did not significantly differ between monocytes from both groups of individuals or after in vitro parasite infection (**Figure 4E**), expression levels of these molecules clearly increased from 0 to 48 h of in vitro cultivation of cells and parasite infection (p < 0.001; ANOVA; **Figure 4J**). Furthermore, monocytes from both groups differed markedly in their responses to parasite infection; whereas those from chronically infected toxoplasmosis patients dosedependently up-regulated HLA-DR,DP,DQ as compared to noninfected control cells, those from naïve control individuals dosedependently down-regulated such expression (**Figure 4J**; p < 0.01). Consequently, monocytes from both groups of blood donors differed significantly in HLA-DR,DP,DQ expression following in vitro cultivation and parasite infection (p < 0.05). Together, the results revealed a dramatic expansion of CD14+CD16<sup>−</sup> monocytes after parasite infection in vitro with a concomitant increase of CD64 expression but a strong decrease of CCR2. Whereas, these changes occurred irrespective of a chronic T. gondii infection of the blood donors, we also unraveled distinct differences in the in vitro reactivity of monocytes from chronically infected toxoplasmosis patients and T. gondii naïve individuals with respect to CD14 and HLA-DR,DP,DQ. It has to be stressed that in vitro cultivation of monocytes in the absence of T. gondii significantly impacted their phenotypes (**Figures 3**, **4**). Depending on the surface marker under investigation, direct exposure to the parasite reversed (**Figures 3B,C,E,F**, and **Figure 4G**), augmented **Figure 3D**, and **Figures 4C,H** or did not alter (**Figures 4B,D,F,I**) these changes.

#### Monocyte-Enriched PBMCs From Chronic Toxoplasmosis Patients Express High IL-12b mRNA Levels After *in vitro* Parasite Infection

Monocytes are an important source for various cytokines under steady state conditions (Auffray et al., 2009) and in response to microorganisms including T. gondii (Yarovinsky, 2014; Sher et al., 2017). We therefore compared cytokine mRNA levels in monocyte-enriched PBMCs from chronic toxoplasmosis patients with those from naïve controls following in vitro infection with T. gondii or in non-infected control cells. Remarkably, cells isolated from T. gondii seropositive individuals up-regulated IL-12b (i.e., IL-12p40) mRNA significantly stronger in response to in vitro

CD14+CD16<sup>+</sup> (C) and CD14dimCD16<sup>+</sup> (D) subsets among monocytes from T. gondii seropositive (gray bars) or seronegative (open bars) individuals. (E,F) Expression levels of CD14 (E) and CD16 (F) on monocytes from T. gondii seropositive or seronegative blood donors. Data represent means ± S.E.M. from 5 T. gondii seropositive and from 13 out of 16 seronegative blood donors which had been randomly selected for in vitro infection assays; outlyers were excluded. Individual data points are also indicated. Significant differences between groups were identified by ANOVA (\*\*\*p < 0.001; \*\*p < 0.01; \*p < 0.05).

infection than those from seronegative control individuals (p < 0.001; ANOVA; **Figure 5A**). Relative IL-12b mRNA production was also slightly higher in cells from chronic toxoplasmosis

patients than those from controls when cultivated for 48 h in the absence of the parasite but this was not statistically significant. Up-regulation of IL-12b mRNA in response to the parasite

(Continued)

FIGURE 4 | (0 h) or were cultivated in vitro for 48 h and infected or not with T. gondii during the final 24 h as indicated and then FACS-analyzed. Expression of cell surface markers was determined for CD14-positive monocytes as outlined in Figure 2. (A–E) Percentages of cells from T. gondii seropositive (gray bars) or seronegative (open bars) individuals with expression of surface markers above background staining. (F–J) Expression levels of surface markers as indicated on monocytes from T. gondii seropositive or seronegative blood donors. Data represent means ± S.E.M. from 5 T. gondii seropositive and from 13 out of 16 seronegative blood donors which had been randomly selected for in vitro infection assays; outlyers were excluded. Individual data points are also indicated. Significant differences between groups were identified by ANOVA [\*\*\*p < 0.001; \*\*p < 0.01; \*p < 0.05; a and b indicate dose-dependent increase or decrease, respectively, of HLA-DR,DP,DQ after parasite infection (p < 0.01)].

coincided with a significant down-regulation of mRNA of antiinflammatory IL-10 irrespective of a chronic T. gondii infection of the blood donors (**Figure 5B**). Finally, TNFa mRNA decreased similarly during in vitro cultivation or parasite infection as compared to freshly isolated cells, and this occurred likewise in cells from both, T. gondii seropositive and seronegative blood donors (**Figure 5C**). Thus, in vitro infection of monocyteenriched PBMCs with T. gondii induced a pro-inflammatory cytokine milieu with higher il-12b gene expression in cells from chronically infected toxoplasmosis patients.

# DISCUSSION

A hallmark of T. gondii infections of humans and other intermediate hosts is the parasites' persistence for months to years or even for the hosts' life. Chronic human infection underlies lifethreatening reactivated toxoplasmosis in immunocompromised patients (Montoya and Liesenfeld, 2004), recrudescent ocular toxoplasmosis in immunocompetent individuals (Pleyer et al., 2014), and it has been associated with several psychiatric disorders (Sutterland et al., 2015) and behavioral changes (Flegr et al., 2002). Not surprisingly, it is accompanied by potent memory B and T cell responses of immunocompetent humans to the parasite, with the latter ones being particularly important to control the parasite (Canessa et al., 1988; Curiel et al., 1993; Prigione et al., 1995; Montoya et al., 1996). Here, we provide the first evidence that peripheral monocytes from humans with chronic toxoplasmosis differ from those of non-infected controls by their repertoire of cell surface markers and their reactivity to T. gondii in vitro. This is of major interest since it suggests a long-term impact of the parasite on cells of the innate immune system which is reminiscent to a form of innate immune memory (Quintin et al., 2012; Lachmandas et al., 2016; Schrum et al., 2018), referred to as trained immunity (Netea et al., 2011). Such reprogramming of monocytes during toxoplasmosis might not only influence the host responses to T. gondii but possibly also to heterologous pathogens.

Ex vivo analyses revealed that monocytes from humans with chronic toxoplasmosis expressed less CD16 and were less frequently CD62L+, but more frequently CD64<sup>+</sup> than monocytes from sero-negative controls. They showed also a trend toward lower expression of CD62L. This repertoire of cell surface markers is remarkable, since it indicates phenotypic changes during chronic toxoplasmosis which are typical for both, CD16<sup>−</sup> classical monocytes and CD16<sup>+</sup> intermediate or non-classical monocytes. Low CD16 and high CD64 expression indeed argue for expansion of CD16<sup>−</sup>

monocytes whereas lower expression of CD62L is indicative for CD16<sup>+</sup> monocytes (Auffray et al., 2009; Ingersoll et al., 2010; Ziegler-Heitbrock et al., 2010). Consistent with this conclusion is our finding that the distribution of the three major human monocyte subsets, i.e., CD14+CD16<sup>−</sup> classical, CD14+CD16<sup>+</sup> intermediate and CD14dimCD16<sup>+</sup> non-classical monocytes (Ziegler-Heitbrock et al., 2010) did not differ between chronic toxoplasmosis patients and naïve controls. Thus, during chronic human toxoplasmosis monocytes from peripheral blood present a distinct phenotype rather than showing expansion of one of the bona fide subpopulations. Peripheral blood monocytes are versatile und dynamic cells (Grage-Griebenow et al., 2001; Auffray et al., 2009) which consecutively develop from rather short-lived classical monocytes to intermediate and finally non-classical monocytes (Patel et al., 2017). The results presented here with monocytes from chronic toxoplasmosis patients highlight the plasticity of human monocytes.

Beside phenotypic differences as observed directly after cell isolation, monocytes from T. gondiiseropositive humans partially also presented different responses to direct exposure to parasite infection than those from T. gondii naïve individuals. Though not specific to the in vitro infection, they expressed less CD14 after encountering T. gondii, as compared to seronegative humans. Remarkably, they specifically up-regulated HLA-DR,DP,DQ in response to parasite infection and up-regulated IL-12b mRNA more vigorously than cells from seronegative humans. It must be stressed that we did not FAC-sort monocytes before analyzing cytokine transcript levels, and we can therefore ascribe the higher IL-12b expression only to monocyte-enriched PBMCs. After exposure to live parasites, CD16<sup>+</sup> monocytes and CD1c<sup>+</sup> DCs are major producers of IL-12 in humans ex vivo (Tosh et al., 2016). The contribution of each of these cell populations to the vigorous IL-12b mRNA as observed in monocytesenriched PBMCs from chronically infected toxoplasmosis patients therefore needs to be further investigated.

We also uncovered several responses to in vitro parasite infection by both, monocytes from seropositive and seronegative humans. Thus, CD14+CD16<sup>−</sup> classical monocytes clearly increased with a concomitant decrease of the CD14+CD16<sup>+</sup> and CD14dimCD16<sup>+</sup> subsets, and they generally expressed less CD14 and CD16 than non-infected control cells. Furthermore, CCR2 expression was almost completely down-regulated whereas CD64 expression increased. Finally, IL-10 mRNA levels also clearly decreased in cells from both groups of humans, although this can again only be ascribed to monocyte-enriched PBMCs (see above). Expansion of CD14+CD16<sup>−</sup> monocytes, downregulation of IL-10 mRNA, and up-regulation of IL-12b mRNA

as observed with cells from both groups of humans (albeit significantly higher in those from T. gondii seropositive ones; see above) are indicative for an inflammatory cell phenotype. CD14+CD16<sup>−</sup> classical monocytes are considered the human equivalent of murine Ly6c<sup>+</sup> (Gr1+) monocytes (Auffray et al., 2009; Ingersoll et al., 2010; Ziegler-Heitbrock et al., 2010). The latter ones are often referred to as "inflammatory" monocytes although this attribute may be too simplistic (Ziegler-Heitbrock et al., 2010). Although the cytokine profile indeed argues for a pro-inflammatory function, it is interesting to note that CCR2 levels and positivity were strongly decreased in response to parasite infection in vitro. This chemokine receptor is among the bona fide monocyte subsets restricted to human CD14+CD16<sup>−</sup> or murine Ly6c<sup>+</sup> classical monocytes. It is required for the emigration of monocytes from the bone marrow into the blood circulation and for the subsequent recruitment to sites of injury or infection (Shi and Pamer, 2011), including acute infections of mice with T. gondii (Robben et al., 2005; Dunay et al., 2008). Expansion of CD14+CD16<sup>−</sup> classical monocytes and concomitant down-regulation of CCR2 in response to T. gondii thus point toward development of monocytes with a distinct phenotype that do not resemble the three main human subsets. It is reminiscent to a versatile monocyte reprogramming after they encounter site-specific or signal-specific environments (Arnold et al., 2007; Avraham-Davidi et al., 2013).

The functional consequences of monocyte reprogramming during chronic human toxoplasmosis and in response to in vitro parasite infection yet remain to be uncovered. Due to the distinct cell phenotypes with characteristics of different monocyte subsets they are also not easy to predict. The increase in positivity or expression of the high affinity IgG receptor (FcγRI, i.e., CD64) nevertheless suggests increased phagocytic activity of monocytes from T. gondii seropositive humans and after direct exposure to the parasite in vitro. The up-regulation of HLA-DR,DP,DQ as specifically observed in response of monocytes from chronic toxoplasmosis patients to parasite infection in vitro additionally argues for increased capacities to present antigens to CD4<sup>+</sup> T lymphocytes. These findings are remarkable since human monocytes in contrast to their murine counterparts need to phagocytose T. gondii in order to subsequently induce cytokine production (Tosh et al., 2016). Finally, the decrease in CD62L<sup>+</sup> cells and the trend toward reduced CD62L expression levels suggest reduced capacities of monocytes from humans with chronic toxoplasmosis to adhere to endothelial cells and to transmigrate into the surrounding tissue (Rzeniewicz et al., 2015). The severe down-regulation of CCR2 after direct exposure of monocytes from both seropositive and seronegative humans to the parasite might further contribute to a reduced recruitment of human monocytes to sites of infection or injury. However, these assumptions clearly need to be confirmed in humans with chronic toxoplasmosis although this will be a challenging task.

It is generally assumed that humans chronically infected with T. gondii harbor a restricted number of tissue cysts predominantly in brain and muscle tissues, though only few studies mostly with AIDS patients have addressed these issues (reviewed in Mcconkey et al., 2013; Wohlfert et al., 2017). Recent findings indicate that bradyzoites within tissue cysts may be more active than previously thought (Watts et al., 2015), and tissue cysts might even rupture occasionally with the majority of parasites being rapidly cleared by the host's immune response. Even on such occasion the inflammatory responses remain however locally confined to the respective sites of parasite reactivation. It is therefore rather unlikely that a limited number of more or less dormant intracellular parasites in brain and muscle tissues induces those changes in phenotypes and reactivity of monocytes from peripheral blood as described here. We instead propose that these changes are signs of a monocyte reprogramming during the acute phase of infection. Human monocytes are readily infected and permissive to rapid tachyzoite division (Channon et al., 2000), and studies in mice show that blood monocytes are critical for parasite dissemination (Courret et al., 2006). Acute infection with T. gondii consequently induces a robust innate immune response in humans and their monocytes are readily activated after phagocytosis of live parasites (Sher et al., 2017). Remarkably, accumulating evidence suggests that sensing of distinct pathogen-associated molecular patterns from bacteria, fungi and parasites through pattern recognition receptors can alter functionality of monocytes in the long term (Quintin et al., 2012; Lachmandas et al., 2016; Schrum et al., 2018). This form of innate immune memory, i.e., trained immunity, seems to be regulated by epigenetic mechanisms and metabolic reprogramming of monocytes and their progenitor cells (Quintin et al., 2012; Bekkering et al., 2018; Mitroulis et al., 2018; Schrum et al., 2018). Importantly, the reprogrammed monocytes can respond to secondary stimulation by the same but also by heterologous stimuli with enhanced reactivity. This resembles previous reports on chronically T. gondii-infected mice which are more resistant to challenge infections with bacterial pathogens (Ruskin and Remington, 1968; Ruskin et al., 1969; Neal and Knoll, 2014). Thus, the altered phenotypes and reactivity of human monocytes as reported herein suggest that innate immune training may also operate during infections with T. gondii but this awaits future confirmation.

#### REFERENCES


## ETHICS STATEMENT

This study was carried out in accordance with the recommendations of Ethics commission of the University Medical School Göttingen with written informed consent from all subjects. The protocol was approved by the Ethics commission of the University Medical School Göttingen (Project numbers 28/3/08 and 1/3/13).

### AUTHOR CONTRIBUTIONS

CL conceived the study and the experimental setup, prepared the figures, and wrote the manuscript. HE conducted the experiments and analyzed the data. Both authors read and approved the final version of the manuscript.

# FUNDING

We acknowledge support by the German Research Foundation and the Open Access Fund of the University of Göttingen for covering publication fees.

# ACKNOWLEDGMENTS

The authors thank the Blood Donation Service Center of the University Medical School Göttingen, Germany for providing the human peripheral blood samples. They are also grateful to Philipp Stalling, Münster, Germany for his preparatory analyses of human-derived monocytes.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00235/full#supplementary-material

CD4+ monoclonal T cells and macrophages results in killing of trophozoites. J. Immunol. 140, 3580–3588.


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

Copyright © 2019 Ehmen and Lüder. 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.

# An Open-Format Enteroid Culture System for Interrogation of Interactions Between *Toxoplasma gondii* and the Intestinal Epithelium

Lisa Luu1†, Luke J. Johnston1†, Hayley Derricott <sup>1</sup> , Stuart D. Armstrong<sup>1</sup> , Nadine Randle<sup>1</sup> , Catherine S. Hartley <sup>1</sup> , Carrie A. Duckworth<sup>2</sup> , Barry J. Campbell <sup>2</sup> , Jonathan M. Wastling1‡ and Janine L. Coombes <sup>1</sup> \*

*<sup>1</sup> Department of Infection Biology, Faculty of Health and Life Sciences, School of Veterinary Science, Institute of Infection and Global Health, University of Liverpool, Liverpool, United Kingdom, <sup>2</sup> Department of Cellular and Molecular Physiology, Institute of Translational Medicine, University of Liverpool, Liverpool, United Kingdom*

When transmitted through the oral route, *Toxoplasma gondii* first interacts with its host at the small intestinal epithelium. This interaction is crucial to controlling initial invasion and replication, as well as shaping the quality of the systemic immune response. It is therefore an attractive target for the design of novel vaccines and adjuvants. However, due to a lack of tractable infection models, we understand surprisingly little about the molecular pathways that govern this interaction. The *in vitro* culture of small intestinal epithelium as 3D enteroids shows great promise for modeling the epithelial response to infection. However, the enclosed luminal space makes the application of infectious agents to the apical epithelial surface challenging. Here, we have developed three novel enteroid-based techniques for modeling *T. gondii* infection. In particular, we have adapted enteroid culture protocols to generate collagen-supported epithelial sheets with an exposed apical surface. These cultures retain epithelial polarization, and the presence of fully differentiated epithelial cell populations. They are susceptible to infection with, and support replication of, *T. gondii*. Using quantitative label-free mass spectrometry, we show that *T. gondii* infection of the enteroid epithelium is associated with up-regulation of proteins associated with cholesterol metabolism, extracellular exosomes, intermicrovillar adhesion, and cell junctions. Inhibition of host cholesterol and isoprenoid biosynthesis with Atorvastatin resulted in a reduction in parasite load only at higher doses, indicating that *de novo* synthesis may support, but is not required for, parasite replication. These novel models therefore offer tractable tools for investigating how interactions between *T. gondii* and the host intestinal epithelium influence the course of infection.

Keywords: enteroid, organoid, *Toxoplasma gondii*, intestine, cholesterol, statin, monolayer

# INTRODUCTION

Toxoplasma gondii infection is commonly acquired following oral ingestion of cyst-containing meat, or oocyst-contaminated water and produce. As a result, the first encounter between parasite and host occurs in the small intestinal epithelium. Subsequently, the parasite travels from the intestine to the brain and other tissues, where it forms cysts that persist for the lifetime of the

#### *Edited by:*

*Jeroen P. J. Saeij, University of California, Davis, United States*

#### *Reviewed by:*

*Laura Knoll, University of Wisconsin—Madison, United States Frank Seeber, Robert Koch Institute, Germany Michael S. Behnke, Louisiana State University, United States*

*\*Correspondence:*

*Janine L. Coombes jcoombes@liverpool.ac.uk*

*†These authors have contributed equally to this work as joint first authors*

#### *‡Present address:*

*Jonathan M. Wastling, Faculty of Natural Sciences, Keele University, Staffordshire, United Kingdom*

#### *Specialty section:*

*This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology*

> *Received: 29 March 2019 Accepted: 05 August 2019 Published: 28 August 2019*

#### *Citation:*

*Luu L, Johnston LJ, Derricott H, Armstrong SD, Randle N, Hartley CS, Duckworth CA, Campbell BJ, Wastling JM and Coombes JL (2019) An Open-Format Enteroid Culture System for Interrogation of Interactions Between Toxoplasma gondii and the Intestinal Epithelium. Front. Cell. Infect. Microbiol. 9:300. doi: 10.3389/fcimb.2019.00300* infected individual. This can have serious consequences to human health: reactivation of cysts in people whose immune systems are compromised can result in severe encephalitis and death. Furthermore, spontaneous abortion, stillbirth, and severe birth defects can occur if the infection is caught during pregnancy and transmitted to the fetus. Treatment for toxoplasmosis is available, but it can cause severe side effects and is ineffective against brain cysts. Thus, the development of novel vaccines and therapeutics remains an important research goal. An obvious target is the initial interaction between the parasite and intestinal epithelium, which is critical not only in controlling initial invasion and replication, but also in shaping the quality of the systemic immune response.

Surprisingly little is known about how T. gondii interacts with the small intestinal epithelium of orally infected hosts. In vivo, it has been reported that dividing parasites can be observed in intestinal tissue as early as 1 day post infection (dpi), although our own work shows that parasites do not become readily detectable in the intestine until 5 dpi or later (Dubey, 1997; Kobayashi et al., 1999; Coombes et al., 2013). This makes it almost impossible to study the earliest interactions between parasite and host in a whole animal model. Experiments in cell line and explant culture models have shown that paracellular transmigration may play a significant role in early traversal of the epithelial barrier by T. gondii (Barragan et al., 2005). Tachyzoites cluster at intercellular junctions and transmigrate, often without significant disruption of the epithelial barrier (Barragan et al., 2005; Weight and Carding, 2012; Jones et al., 2016). This allows the parasite to rapidly disseminate, likely following invasion of motile immune cells (Barragan and Sibley, 2002; Courret et al., 2008; Weidner et al., 2016). Despite this rapid transmigration, there is clear evidence of communication between the parasite and the host epithelium. T. gondii uses host intercellular adhesion molecule 1 (ICAM-1), and the tight-junction protein, occludin, as receptors for transmigration, resulting in redistribution of occludin from tight junctions to an intracellular compartment (Barragan et al., 2005; Weight and Carding, 2012). Furthermore, a proportion of tachyzoites are seen to invade, rather than bypass, cultured epithelial cells. Nevertheless, we understand remarkably little about how epithelial cells respond to contact with the parasite, or if the parasite targets a specific cell type or location along the crypt-villus axis. To do this, we require sophisticated models of the intestinal epithelium.

The continuous division of intestinal epithelial stem cells in in vitro 3-dimensional (3D) culture, results in the formation of intestinal organoids ("enteroids") consisting of a fully differentiated, polarized epithelium, surrounding a central lumen (Sato et al., 2009; Yui et al., 2012). Enterocytes, enteroendocrine cells, goblet cells, Paneth cells, and tuft cells are all represented in these cultures, and enteroids are proving to be a valuable resource for the study of enteric infections, including norovirus, rotavirus, Salmonella, and Escherichia coli (Sato et al., 2009; Barker et al., 2012; Finkbeiner et al., 2012; Zhang et al., 2014, 2016; Wilson et al., 2015; Walsh et al., 2016). The most commonly used methods of enteroid culture produce an enclosed luminal space, meaning that pathogens need to be microinjected if invasion is to be restricted to the physiologically relevant apical surface of the epithelium (Zhang et al., 2014; Wilson et al., 2015; Heo et al., 2018). While this method has been used for viruses and bacteria, T. gondii is significantly larger, making microinjection technically challenging. Therefore, modification of enteroid culture techniques to generate an exposed apical surface would provide a more practical solution for the largescale analysis of parasitic infections. One such avenue that shows promise is the passage of 3D enteroids from Matrigel <sup>R</sup> onto the surface of type 1 collagen gels (Jabaji et al., 2013; Wang et al., 2017). The enteroid fragments grow out across the surface of the gel to form epithelial sheets with the apical surface open to the culture media.

Here, we have developed three enteroid-based techniques for modeling T. gondii infection: fragmentation, micro-injection and open-format enteroid cultures. We show that open-format enteroid cultures support invasion and replication of T. gondii. Infection resulted in modulation of host cholesterol pathways, possibly providing a source of cholesterol to support parasite invasion and growth. For T. gondii infection, we conclude that enteroid models bridge the gap between simplistic in vitro cell line models, and in vivo murine models that yield unworkably low invasion events during the first hours of infection.

# MATERIALS AND METHODS

#### Animal Tissues

Murine tissues used within this study were harvested from female specific-pathogen-free, C57B1/6J mice, aged between 6 and 12 weeks (Charles River, Margate, United Kingdom). In some experiments, mT/mG mice (Gt(ROSA)26Sortm4(ACTB−tdTomato,−EGFP)Luo, The Jackson Laboratory) were used to visualize epithelial cell membranes, and parasite invaded cells. Prior to tissue harvest, mice were culled by cervical dislocation as outlined in Schedule 1 of the Animals (Scientific Procedures) Act 1986. Tissue use was approved by the UK Home Office (project license) and the University of Liverpool Animal Welfare and Ethical Review Body.

# Murine Enteroid Culture

Murine jejunal and ileal tissues were collected by dissection, and abdominal fat was removed. Tissues were sliced longitudinally, washed thrice in PBS and cut into 0.5 cm<sup>3</sup> sections. Crypt units were dissociated from tissues by four cycles of incubation in ethylenediamine tetraacetic acid (EDTA) (30 mM in PBS) (Corning, Loughborough, United Kingdom) for 5 min followed by 15 s of vigorous shaking in PBS. Crypt fractions were examined by microscopy and fractions containing intact crypts largely free of contaminating villi were selected for further processing. Isolated crypts were counted and concentrated by centrifugation (300xg, 10 min, 4◦C). Crypts were re-suspended at 10 crypts/µL in 50% Growth Factor Reduced, Phenol Red Free, Matrigel <sup>R</sup> (Corning), and 50% complete enteroid medium [Dulbecco's Modified Eagle Medium: Nutrient mixture F12 (Gibco; Loughborough, United Kingdom) supplemented with 500 ng/mL human R-spondin 1, 50 ng/mL human epidermal growth factor (EGF), 100 ng/mL murine Noggin (all PeproTech; London, United Kingdom), 1X B-27 (Gibco), 1X N-2 (Gibco), 1% w/v Penicillin/streptomycin (Sigma), and 10 mM HEPES (Sigma Aldrich, Dorset, United Kingdom)] (Sato et al., 2009). Crypts in Matrigel <sup>R</sup> were polymerised at 37◦C in 5% v/v atmospheric CO<sup>2</sup> for 30 min, and overlaid with complete enteroid medium. Enteroid medium was replenished every 4 days and cultures were split 1:3 or 1:4 every 7–10 days. In some experiments, mouse IntesticultTM (StemCell; Cambridge, United Kingdom) was substituted for complete enteroid medium.

# Preparation of Collagen-Support Matrix

Type 1 rat-tail collagen gels (Life Technologies) were prepared following manufacturer's instructions to a final concentration of 2 mg/mL. NaOH and water were added to 10X Minimum Essential Media (MEM) (Sigma) and pipetted carefully to ensure thorough mixing. Rat-tail collagen (3 mg/mL stock) was added to the MEM solution and pH strips were used to determine pH. Collagen solutions at pH of 7.2 were accepted as suitable, and were polymerised into 30 µL domes at 37◦C and 5% v/v atmospheric CO<sup>2</sup> for 60 min.

# Establishment of Collagen-Supported Epithelial Sheets

To generate collagen-supported epithelial sheet cultures, media was removed from 3D Matrigel <sup>R</sup> -embedded enteroids and cultures were disrupted through pipetting with cold PBS, followed by centrifugation (300xg for 10 min). Enteroid fragments were re-suspended in PBS, overlaid onto polymerised collagen gel domes and allowed to adhere during a 15 min incubation at 37◦C and 5% v/v atmospheric CO2. Enteroid fragments on collagen gels were overlaid with complete enteroid medium, and cultured at 37◦C and 5% v/v atmospheric CO2, with medium replenished every 3–4 days.

### *T. gondii* Cultivation, Purification, and Infection

Type I RH, type II Prugniaud (PRU) and type III VEG strains of T. gondii were used in this study. The PRU strains were genetically modified: PRU parasites expressing GFP (PRU-GFP) were a kind gift from Jeroen Saeij (UC Davis) and Eva Frickel's (Francis Crick Institute) laboratories, and PRU parasites expressing tdTomato and a toxofilin-cre fusion protein (PRUtdTom-cre) were a kind gift from Anita Koshy (University of Arizona) (Koshy et al., 2010; Coombes et al., 2013). All T. gondii types were cultivated in Vero (African green monkey kidney epithelial) cell line in basal media Dulbecco's Modified Eagle's Medium with high glucose (Sigma Aldrich) supplemented with 5% v/v fetal bovine serum and 1% v/v Penicillin/Streptomycin, incubated at 37◦C in a 5% v/v CO<sup>2</sup> atmosphere. To isolate parasites for enteroid infections, infected Vero cells were scraped from flasks and lysed by blunt-end needle syringing to release tachyzoites. A PD-10 desalting column (GE HealthCare Life Sciences; Buckinghamshire, United Kingdom) was equilibrated by releasing and discarding the storage fluid. The column was then washed twice with 5 mL PBS, and the flow through discarded. The parasite cell suspension was added to the column, collecting the flow through. An additional 5 mL of PBS was added to flush the column. Parasites were counted and re-suspended in complete enteroid media or IntesticultTM (StemCell) for infection studies. To infect enteroids by fragmentation, day 5–7 enteroids were resuspended in PBS, vigorously pipetted to expose the apical surface, and centrifuged at 300xg for 5 min. The supernatant was removed, and enteroids resuspended in 50 µL of IntesticultTM medium containing 1 × 10<sup>6</sup> or 1 × 10<sup>7</sup> T. gondii tachyzoites. For each infection, enteroids from two wells of a 48 well plate were resuspended per 50 µL of parasite-containing medium, then re-plated into a single well. Samples were incubated at 37◦C for the indicated periods before washing the samples twice with PBS by centrifugation at 300xg for 5 min. Infected enteroids were cultured in Matrigel <sup>R</sup> and IntesticultTM medium on glass coverslips and incubated at 37◦C in 5% v/v CO2. For microinjection, microneedles (TW 100-4, World Precision Instruments) were generated with a bore size of 8–10µm to prevent blockages at the needle tip. Glass capillaries were pulled using a micropipette puller (Sutter instruments) at 50◦C with 355 g of weight and microneedle tips broken in a glass-bottom culture dish to provide the suitable bore size. Parasites were loaded into microneedles at 4 × 10<sup>9</sup> /mL and the microneedle lowered onto the enteroid to press down on the enteroid surface. The microneedle was moved laterally to pierce the enteroid and the contents of the loaded microneedle were injected into the enteroid lumen. To infect collagen-supported epithelial sheets, enteroid media was carefully removed and replaced with 200 µL of complete enteroid medium containing purified T. gondii tachyzoites. Cultures were incubated at 37◦C in a 5% v/v CO<sup>2</sup> atmosphere. For atorvastatin (Cayman Chemical via Cambridge Biosciences; Cambridge, United Kingdom) treated infections, collagen-supported epithelial sheets were infected, incubated for 15 min then atorvastatin (30 µM) was added and remained in the medium for the remaining duration of the experiment.

# Immunofluorescent Staining

Enteroids were cultured on round glass coverslips in 48 well plates, and fixed and stained in the wells before mounting on glass slides. Enteroids were washed with PBS, and fixed with 4% w/v paraformaldehyde (in PBS). A general staining protocol was used for antibody staining at room temperature on a rocker; block for 1 h with blocking buffer (10% v/v Donkey serum (Sigma) and 1% v/v Triton X-100 in PBS), primary antibody at 1:200 in blocking buffer, secondary antibody at 1:200 with wash buffer (1% v/v serum and 0.1% v/v Triton X-100 in PBS) with three PBS washes between each step. Primary antibodies used in this study include; monoclonal anti-mouse epithelial cell adhesion molecule (EpCAM; G8.8; Affymetrix eBiosciences), monoclonal anti-mouse lysozyme (LYZ; BGN/06/961; AbCam, Cambridge, UK), monoclonal anti-mouse e-cadherin (E-Cad; 24E10; BD Transduction Laboratories), polyclonal anti-mouse mucin-2 (MUC2; H-300; Insight Biotechnology), mouse anti-Toxoplasma gondii SAG1 membrane protein (TP3; AbCam), and anti-Toxoplasma gondii GRA7 (a kind gift from David Sibley, Washington University, St. Louis). Secondary antibodies were donkey anti-rabbit conjugated with either tetramethylrhodamine isothiocyanate (TRITC) or fluorescein isothiocyanate (FITC) (both Jackson Immuno via Stratech Scientific; Cambridge, United Kingdom) and anti-mouse Alexa Fluor 488 (AbCam). Actin filaments (F-actin) were labeled with phalloidin conjugated with either rhodamine or Alexa Fluor 647 and nuclei were stained with 4′ ,6-diamidino-2-phenylindole (DAPI) (all from Life Technologies). Coverslips were removed from wells and inverted onto small washers on glass slides, filled with Hydromount (National Diagnostics).

#### Confocal Microscopy

Confocal images were acquired using Zen Black software (Zeiss; Darmstadt, Germany) on a Zeiss LSM880 upright confocal microscope with laser lines Diode (405), Argon (488), DPSS-5610 (561) and HeNe633 (633) and W-Plan Apochromat 40x/1.0 Dic (water immersion) objective (Zeiss) or Plan-Apochromat 63x/1.0 oil DIC M27 (oil immersion) objective (Zeiss).

### Two-Photon Microscopy

Enteroids derived from ROSAmT/mG mice were infected with T. gondii Pru-tdTom-Cre in solution for 1 h at 37◦C before plating in Matrigel <sup>R</sup> in a 35 mm culture dish in warm phenol-red-free DMEM/F12 (ThermoFisher Scientific) medium. Z-stack images were acquired over a 50 min period at 2 min intervals using Zen Black software on a Zeiss LSM880 MP microscope (Zeiss) and a two-photon Chameleon laser set to 920 nm (Coherent). Emission light was separated with 490 or 555 dichroics with bandpass filters 525/25 and 590/20 M used to minimize spectral overlap.

#### Image Analysis

Images were processed using Imaris x64 v9.0.1 (BitPlane AG; Zürich, Switzerland). In all cases, automated cell counting was manually checked for accuracy. The proportion of differentiated epithelial cell populations, or of T. gondii infected cells, was analyzed using the spots function in Imaris, with DAPI staining used to determine the total number of cells. For quantification of infected cells, intra- and extra-cellular parasites were distinguished using the surfaces function in Imaris, applied to the channel containing the pan-epithelial cell label. For 3D enteroids, quantification of goblet and Paneth cells was performed on a single Z-stack slices using the same methods as described above. Quartile and median slices of any given enteroid were used for quantification, and an average of the three slices were taken.

#### Sample Preparation for Mass Spectrometry

Matrigel <sup>R</sup> -grown enteroids were harvested by removal of culture medium, followed by pipetting in PBS. Enteroids were pelleted by centrifugation (300xg for 10 min at 4◦C) then washed three times in PBS. For collagen-supported epithelial sheets, culture medium was removed, cultures were washed three times with PBS and incubated with equal volumes of 0.1 mg/mL type VIII collagenase from Clostridium histolytica (Sigma) for 30 min at 37◦C in a 5% v/v CO<sup>2</sup> atmosphere. An equal volume of complete media was added, and cells were washed three times with PBS. For both the characterization of non-infected collagen-supported epithelial sheets, and for the elucidation of host-pathogen interactions of infected collagen-supported epithelial sheets, 10 wells (per condition and experimental replicate) were processed, pooled and stored at -80◦C for label-free mass spectrometry. For both proteomic experiments, three experimental replicates were performed.

#### Protein Digest for Mass Spectrometry

Buffer [50 mM ammonium bicarbonate, 0.2% w/v RapiGest SF surfactant (Waters; Elstree, United Kingdom)] was added to washed enteroid pellets before sonication using sonicating waterbath (Jencons Scientific; Leighton Buzzard, United Kingdom) for 3 × 10 min on ice. Samples were then heated at 80◦C for 10 min. Protein content was determined using the PierceTM Coomassie plus protein assay kit (ThermoFisher) according the manufacturer's instructions. Protein content was normalized between samples using 50 mM ammonium bicarbonate. Protein were then reduced with 3 mM dithiothreitol (Sigma) at 60◦C for 10 min then alkylated with 9 mM iodoacetimde (Sigma) at room temperature for 30 min in the dark. Proteomic grade trypsin (Sigma) was added at a protein: trypsin ratio of 50:1 and samples incubated at 37◦C overnight. RapiGest was removed by adding tricholoroacetic acid to a final concentration of 1% (v/v) and incubating at 37◦C for 2 h. Peptide samples were centrifuged at 12,000xg for 60 min (4◦C) to remove precipitated RapiGest.

# NanoLC-MS/MS

NanoLC-MS/MS was performed as previously described in Derricott et al. (2019), with the exception of the proteome reference databases used for peptide identification. Peptides were analyzed by on-line nanoflow LC using the Ultimate 3000 nano system (Dionex/Thermo Fisher Scientific). Samples were loaded onto a trap column (Acclaim PepMap 100, 2 cm × 75µm inner diameter, C18, 3µm, 100 Å) at 5 µL min−<sup>1</sup> with an aqueous solution containing 0.1 % v/v TFA and 2% v/v acetonitrile. After 3 min, the trap column was set in-line an analytical column (Easy-Spray PepMap <sup>R</sup> RSLC 50 cm × 75µm inner diameter, C18, 2µm, 100 Å) fused to a silica nano-electrospray emitter (Dionex). The column was operated at a constant temperature of 35◦C and the LC system coupled to a Q-Exactive mass spectrometer (Thermo Fisher Scientific). Chromatography was performed with a buffer system consisting of 0.1% v/v formic acid (buffer A) and 80 % v/v acetonitrile in 0.1% v/v formic acid (buffer B). The peptides were separated by a linear gradient of 3.8–50% buffer B over 90 min at a flow rate of 300 nL/min. The Q-Exactive was operated in data-dependent mode with survey scans acquired at a resolution of 70,000 at m/z 200. Up to the top 10 most abundant isotope patterns with charge states +2 to +5 from the survey scan were selected with an isolation window of 2.0Th and fragmented by higher energy collisional dissociation with normalized collision energies of 30. The maximum ion injection times for the survey scan and the MS/MS scans were 250 and 50 ms, respectively, and the ion target value was set to 1E6 for survey scans and 1E5 for the MS/MS scans. MS/MS events were acquired at a resolution of 17,500. Repetitive sequencing of peptides was minimized through dynamic exclusion of the sequenced peptides for 20 s.

Thermo RAW files were imported into Progenesis LC–MS (version 4.1, Nonlinear Dynamics). Runs were time aligned using default settings and using an auto selected run as reference. Due to poor alignment one sample was removed from the dataset (RH 40 h infected). Therefore, the data for this experimental condition represents duplicate samples. Peaks were picked by the software and filtered to include only peaks with a charge state of between +2 and +6. Peptide intensities were normalized against the reference run by Progenesis LC-MS and these intensities are used to highlight differences in protein expression between uninfected and infected samples with supporting statistical analysis (ANOVA p-values) calculated by the Progenesis LC-MS software. Pairwise comparisons between RH-infected and control enteroids, and between VEG-infected and control enteroids were performed. Using exclusion criteria of p < 0.05 and fold change > 2, significantly up- and down-regulated proteins were identified. Spectral data were transformed to .mgf files with Progenesis LC–MS and exported for peptide identification using the Mascot (version 2.3.02, Matrix Science) search engine. Tandem MS data were searched against the murine (Mus musculus predicted proteome UP000000589, Feb 2017), T. gondii predicted proteomes (T. gondii ME49 V13 and T. gondii VEG V13, EuPathDb) and a contaminant database (common Repository of Adventitious Proteins database, "cRAP," The Global Proteome Machine Organisation, 2011). Mascot search parameters were as follows; precursor mass tolerance set to 10 ppm and fragment mass tolerance set to 0.05 Da. One missed tryptic cleavage was permitted. Carbamidomethylation (cysteine) was set as a fixed modification and oxidation (methionine) set as a variable modification. Mascot search results were further processed using the machine learning algorithm Percolator. The false discovery rate was <1%. Individual ion scores > 13 indicated identity or extensive homology (p < 0.05). Protein identification results were imported into Progenesis LC–MS as .xml files. The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD013306.

#### RESULTS

#### Fragmentation of Enteroids Allows *T. gondii* to Invade Enteroids Through the Apical Epithelial Surface

T. gondii infection can be acquired through the oral route following consumption of oocyst-contaminated water or produce, or consumption of cyst-containing meat from other infected hosts. Following ingestion, the parasite invades the small intestinal epithelium, but we know remarkably little about this initial interaction. As a first approach to achieving infection of enteroids with T. gondii, we removed the enteroids from Matrigel <sup>R</sup> and fragmented them by vigorous pipetting to expose the apical epithelial surface. Enteroids (C57Bl/6 or RosamT/mG) were then incubated with T. gondii tachyzoites (PRU-GFP or PRU-tdTom-cre) in a small volume of media, before being re-plated in Matrigel <sup>R</sup> . In preliminary experiments, successful invasion events were rare (data not shown). To increase the probability of T. gondii invading the epithelium, we tested different infectious doses, the introduction of a centrifugation step to increase contact between the enteroids and parasites, and increasing the contact time between epithelium and parasites before re-plating in Matrigel <sup>R</sup> (**Figure 1A**). As expected, increasing the infectious dose increased the proportion of infected cells in each enteroid (**Figure 1B**). Addition of a centrifugation step did not result in any significant increase in parasite invasion (**Figure 1B**). Finally, increasing the time the enteroid/parasite mixture spent in solution with the parasites from 30 min to 1 h increased parasite invasion at higher doses, although this was not statistically significant (**Figure 1B**). Beyond 1 h, enteroids needed to be re-plated in Matrigel <sup>R</sup> to avoid loss of viability (data not shown). Although the fragmentation technique does not restrict parasite invasion solely to the apical epithelial surface, we were able to confirm by two-photon microscopy that viable, motile, parasites were present in the enteroid lumen (**Figure 1C**, corresponds to **Supplementary Video 1**). By 24 hours post infection (hpi), replicating parasites were visible in epithelial cells, and by 72 hpi extensive replication led to an increase in the number of parasites observed in each enteroid (**Figures 1D,E**). Parasites were observed in both Paneth cells (Lysozyme+) and Goblet cells (Muc2+), though not exclusively (**Figure 1F**). Therefore, enteroid fragmentation results in successful invasion of T. gondii, in agreement with previously published preliminary studies (Klotz et al., 2012; Delgado Betancourt et al., 2019).

# Use of a Cre-Secreting Parasite Confirms Invasion of Enteroid Epithelial Cells by *T. gondii*

Previous studies in explant and cell line cultures have suggested that T. gondii takes a paracellular route across the intestinal epithelium (Barragan et al., 2005; Weight and Carding, 2012; Jones et al., 2016). We therefore wanted to confirm that the parasites we observed in enteroids had productively invaded the epithelium. During cell invasion, T. gondii establishes a parasitophorous vacuole (PV) in which it resides and replicates. Formation of the PV is influenced by the secretion of parasite proteins from micronemes and rhoptries upon invasion, some of which enter the host cell cytosol. Therefore, the detection of rhoptry proteins (e.g., toxofilin) within enteroid cells would verify successful infection by T. gondii. To achieve this, we infected enteroids from ROSAmT/mG mice with a genetically modified T. gondii line, which expresses tdTomato and a toxofilin-cre fusion protein that is released into host cells upon invasion (PRU-tdTom-cre) (Koshy et al., 2010; Coombes et al., 2013). ROSAmT/mG mice constitutively express membrane tdTomato (mT) in all cells, but when exposed to Crerecombinase, the tdTomato gene is cleaved and membrane eGFP (mG) is expressed in its place. Directly invaded epithelial cells of ROSAmT/mG enteroids should therefore express mG, while bystander uninfected cells should express mT (**Figure 2A**). Infections were performed by fragmentation, as described above.

Enteroids infected for 24, 48, and 72 h were analyzed by confocal microscopy (**Figures 2B,C**). At each time point, numerous epithelial cells were observed that both contained parasites (visualized by parasite tdTomato expression and/or staining with an antibody to a T. gondii surface protein, SAG1), and expressed mG (**Figures 2B,C**). Some neighboring uninfected

*(Continued)*

FIGURE 1 | Enteroids were incubated for either 30 m or 1 h in suspension, before being plated in Matrigel and growth medium for 4 h. To quantify infection, enteroids were stained and imaged by confocal microscopy. (B) Graph depicts the number of infection events per 10<sup>6</sup> µm<sup>3</sup> of enteroid, as defined by the presence of a parasite enclosed within host cell f-actin (phalloidin) staining. Data are pooled from three independent experiments (for each experiment, one data point representing the mean of two wells per condition, and three images captured per well, is plotted). Mean ± SEM is shown. (C) Enteroids from ROSAmT/mG mice were infected with *T. gondii* Pru-tdTom-Cre for 24 h and imaged by 2-photon microscopy. The images depict three time points from a time-lapse movie, showing that viable *T. gondii* were present within the enteroid lumen (green arrows). White dotted line represents apical (lumenal) surface. Scale bar 10µm. Corresponds to Supplementary Video 1. Note that *T. gondii* PRU-tdTom-cre and ROSAmT/mG epithelium both express tdTomato. Parasites, defined by morphology and higher fluorescence intensity, have been pseudocoloured white in Imaris. (D) ROSAmT/mG enteroids were infected with *T. gondii* Pru-tdTom-Cre and incubated for 72 h before fixing and staining with an antibody to the SAG1 membrane protein of *T. gondii* (clone TP3). Z-stack images were acquired by confocal microscopy and analyzed using Imaris. (i) Schematic representation of enteroid. White dotted line represents apical surface, yellow dotted line represents the basal surface. Asterisk represents crypt base. E is epithelium, L is lumen. (ii) The surfaces and mask functions in Imaris were used to separate the SAG1 signals derived from intracellular (white) and extracellular (green) parasites into two separate channels. (iii–v) The inset shows a rosette of replicating parasites E is epithelium, L is lumen. (E) Graph depicts intracellular parasites per enteroid volume over a 72 h period of infection. Pooled data from two independent experiments (for each experiment, one data point representing the mean of two wells per condition, and three images captured per well, is plotted). Mean ± SEM is shown. (F) C57Bl/6 Enteroids were infected with *T. gondii* RH (green) and stained for Paneth cells (Lysozyme, white) and goblet cells (Muc2, white). Examples of uninfected Paneth or Goblet cells are indicated with asterisks. White dotted lines indicate apical surfaces of the epithelium throughout.

cells also expressed mG (**Figures 2D,E**). Since Cre-recombinase causes a permanent switch to mG expression, daughter cells will also express mG, with one daughter cell inheriting the PV. This division of infected mG<sup>+</sup> cells has been directly observed in fibroblasts in a previous study (Koshy et al., 2010).

More interestingly, we also observed uninfected mG<sup>+</sup> cells at locations distant to sites of infection ("D-U mG<sup>+</sup> cells") (**Figure 2F**). Previous studies using the same reporter system revealed that parasites could inject effector proteins into cells they did not subsequently invade (Koshy et al., 2010). Since enteroids contain a variety of differentiated epithelial cell types, the presence of D-U mG<sup>+</sup> cells may indicate a probing mechanism by T. gondii to select one epithelial cell type over another. D-U mG<sup>+</sup> cells could also result from the migration of an uninfected daughter cell away from the paired daughter cell carrying the PV. However we feel this is unlikely, since lineage tracing experiments of the small intestinal epithelium in vivo show clonal expansion of epithelial cells up the crypt-villus axis (Snippert et al., 2010). Finally, D-U mG<sup>+</sup> cells could occur when the paired daughter cell carrying the PV is shed into the enteroid lumen, as observed in **Supplementary Video 2**.

It was evident that not all infected cells expressed mG. This is more apparent at early time-points post-infection (24 hpi), suggesting that this may simply represent the expected time lag between secretion of cre into the epithelial cell, and expression of mG. Alternatively, these parasites may be using a paracellular route to cross the epithelium. However, infected cells that did not express mG were also evident at later time points (48–72 hpi) even within cells containing large rosettes of replicating parasites (**Figure 2G**).

### Microinjection of 3D Enteroids Results in Infection of Epithelial Cells

Enteric infections occur at the luminal surface of the intestinal epithelium. Although fragmentation of enteroids exposes the luminal surface for infection, this method does not restrict infection via this physiologically relevant route. Luminal infections can be achieved through the microinjection of pathogens into the luminal space of enteroids and have been established with bacteria, viruses, and only recently achieved with the parasite Cryptosporidium parvum (Heo et al., 2018). This method of enteroid infection has not been established with T. gondii and may provide a relevant model of infection. The microinjection technique was optimized to provide consistent injections of enteroids (**Figures 3A,B**) and injected enteroids were analyzed using two-photon microscopy, showing T. gondii present within the lumen (**Figure 3C**). Injected enteroids cultured for 48 h showed successful infection as indicated by mG expression in ROSAmT/mG enteroids (**Figure 3D**). The rate of infection was however low, suggesting possible defensive mechanisms of intact enteroids to prevent invasion, or an adverse response of the parasites to injection pressure. These defensive mechanisms might include the presence of an intact mucus layer, and the release of anti-microbial peptides.

# Generation of Enteroid-Derived Collagen-Supported Epithelial Sheets

To generate enteroid cultures with an accessible luminal surface, 3D enteroids were fragmented by pipetting, washed free of Matrigel <sup>R</sup> , and the fragments overlaid onto 2 mg/mL rat tail collagen (**Figure 4A**). As the epithelial cells proliferated, they spread outwards from the fragments, and across the surface of the gel to form large epithelial sheets by day 7–8 of culture (**Figure 4B**). The epithelial sheets exhibited a cobblestone morphology (**Figure 4B**), but with the appearance of distinct micro-domains containing epithelial cells of differing sizes and morphologies (**Figure 4B**) (Jabaji et al., 2013). This suggested that the collagen-supported epithelial sheets retained some organizational features of the 3D epithelium, such as crypt- and villus-like domains, but with an accessible luminal surface.

## Collagen-Supported Epithelial Sheets Retain Features of a Fully Differentiated Epithelium

3D enteroids are favored as in vitro models of the small intestine, as they faithfully recapitulate many features of the in vivo gut environment, including the presence of a polarized epithelium containing multiple fully differentiated epithelial cell types. To test that the collagen-supported epithelial sheets were appropriately polarized, we stained day 7 sheets with

*(Continued)*

FIGURE 2 | upon exposure to Cre-recombinase, the tdTomato gene is cleaved and membrane eGFP (mG) is expressed in its place. Pru-tdTom-cre parasites secrete cre into the host cell upon invasion. *T. gondii* infected epithelial cells should therefore express mG. (B) Graph depicts volume of mG-expressing epithelial cells as a percentage of enteroid volume. Data are pooled from two independent experiments (for each experiment, one data point representing the mean of two wells per condition, and three images captured per well, is plotted). Mean ± SEM is shown. (C) 20µm orthoslices (or maximum intensity projections for insets) of z-stack images showing ROSAmT/mG enteroids infected with *T. gondii* Pru-tdTom-Cre over a 72 h period. Parasite tdTomato signal (red) is distinguished from host tdTomato signal (mT; red) by morphology and fluorescence intensity. Parasites appear as bright puncta. Infected cells also express membrane eGFP (mG; green). \*Indicates crypt base. (D) An epithelial cell expressing mG, but not containing a parasite is indicated with a white arrow. (E) Graph depicts the proportion of infected (parasite-containing) mG<sup>+</sup> cells among total mG<sup>+</sup> cells. Data are pooled from two independent experiments (for each experiment, one data point representing the mean of two wells per condition, and three images captured per well, is plotted). Mean ± SEM is shown. (F) An infected (parasite-containing) epithelial cell expressing mG is indicated with a blue arrow, and distant uninfected Cre-exposed mG<sup>+</sup> cells (D-U mG+) are indicated with white arrows. (G) An epithelial cell containing replicating parasites, but not expressing mG, is indicated with a white arrow. White dotted lines indicate apical surfaces of the epithelium throughout.

phalloidin (to label F-actin) and an antibody to E-cadherin (**Figure 5A**). Enrichment of F-actin on the apical surface of the epithelium, coupled with basolateral expression of E-cadherin, confirmed normal polarization of the monolayer (**Figure 5A**). Lateral expression of E-cadherin also suggested the presence of functional adherens junctions. EpCAM, which regulates adherens and tight junctions, was also expressed laterally, in agreement with its in vivo localization (**Figure 5B**).

Positive staining for both Lysozyme (LYZ) and Mucin 2 (MUC2) was observed, indicating the presence of Paneth and goblet cells, respectively (**Figures 5C,D**). By mass spectrometry, multiple proteins associated with the differentiation and function of goblet and Paneth cells (together with enterocytes and enteroendocrine cells) were detected (**Table 1**). Paneth cell associated proteins included anti-microbial alpha-defensins, while goblet cell associated proteins included major components of the mucus layer. Together, these data confirm the presence differentiated epithelial cell types with host defensive function in our monolayer cultures, making them suitable for infection studies. Interestingly, Paneth cells were observed in tight clusters in the central regions of epithelial sheets (36.84% of images analyzed), whereas goblet cells were more widely dispersed (**Figure 5C**). This is reminiscent of their in vivo distribution, where Paneth cells cluster at crypt bases, and goblet cells are distributed along the crypt-villus axis. Collagen-supported epithelial sheets therefore retain elements of the structural organization of their 3D counterparts (**Figure 5C**).

# *T. gondii* Successfully Invades, and Replicates Within, Collagen-Supported Epithelial Sheets

Unlike 3D Matrigel <sup>R</sup> grown enteroids, the collagen-supported epithelial sheet model exhibits an exposed luminal surface accessible to pathogens. Collagen-supported epithelial sheets were exposed to 1 × 10<sup>6</sup> T. gondii RH strain tachyzoites at the apical surface. Parasites were observed between the apical and basal surface of the monolayer (indicating invasion) as early as 1 h post-infection. Previous studies have shown that T. gondii can use a paracellular route to cross the intestinal epithelium. Therefore, to confirm that parasites had actively invaded host cells, we stained the monolayers with an antibody to the T. gondii dense granule protein, GRA7, which marks the parasitophorous vacuole (Bonhomme et al., 1998). Parasites located between the apical and basal surfaces of the epithelium stained positively for GRA7, while those located outside of the monolayer did not (**Figure 6A**).

Having established that this model was capable of supporting infection, we identified the time-points at which invasion and replication of the parasite occurred. For this analysis we compared a virulent type I strain (RH) with an avirulent type III strain (VEG). In both cases, parasites were observed in the monolayer at 1 hpi. However, the proportion of epithelial cells containing parasites was low with large number of parasites observed extracellularly (**Figure 6B**). By 4 hpi, the number of parasites present in the monolayer had increased for both strains, despite there being no evidence for any replication having taken place. We therefore conclude that new invasion events continued to occur over the first few hours of culture (**Figures 6B,D**).

The first instance of parasite replication was detected between 16 and 24 hpi in both VEG and RH infected cultures. The proportion of epithelial cells harboring replicating parasites had further increased by 40 hpi. To maximize our chances of detecting host responses to invading parasites, we focused on the 40 hpi time-point to study the host response to infection (**Figures 6C,D**).

#### Infection of Collagen-Supported Epithelial Sheets by Virulent (RH) and Avirulent (VEG) Strains of *T. gondii* Induces Differential Host Cell Protein Responses

To perform an unbiased analysis of the host response to infection, we subjected T. gondii infected collagen-supported epithelial sheets to quantitative label free mass spectrometry. For this analysis, three biological replicates of collagen-supported epithelial sheets were infected for 40 h with either RH or VEG tachyzoites. After removal of single peptide hits, 1,909 proteins were identified, of which 67 were T. gondii proteins. Pairwise comparisons between RH-infected and control enteroids, and between VEG-infected and control enteroids were performed. Using exclusion criteria of p < 0.05 and fold change > 2, significantly up- and down-regulated proteins were identified.

Twenty-five proteins changed in abundance following infection with RH (11 upregulated and 14 downregulated, compared to uninfected controls), and 30 proteins changed in abundance following infection with VEG (8 upregulated and 22 downregulated, compared to uninfected controls). A subset of six host cell proteins were similarly modulated in response to either strain at 40 hpi (**Figure 7A**). Across both

FIGURE 3 | Microinjection of *T. gondii* into the enteroid lumen results in successful invasion of epithelial cells. (A) Schematic of the optimized technique for inserting microneedles into the lumen of enteroids. (i) the microneedle is lowered onto the enteroid to, (ii) press down on the enteroid surface. The microneedle is moved laterally to pierce the enteroid. (iii) The contens of the loaded microneedle are injected into the enteroid lumen. (iv) The microneedle is removed from the enteroid lumen by moving up and laterally away. (B) Bright field images corresponding to the technique described in (A). White arrow indicates the microneedle. Scale bar: 100µm. (C) ROSAmT/mG enteroids were microinjected with *T. gondii* Pru-GFP. Image depicts a single time-point from a two-photon time-lapse movie. Parasites (green) are present in the enteroid lumen (red). White arrow indicates injection site. Scale bar 30µm. (D) *T. gondii* Pru-tdTom-Cre were loaded into microneedles at 4 × 10<sup>9</sup> /mL, microinjected into ROSAmT/mG enteroids, and incubated for 48 h. Samples were fixed and z-stack images obtained by confocal microscopy. Images were analyzed by Imaris with *T. gondii* pseudo-colored for clarity. A cross-section of the infected enteroid showing intracellular parasites within mG<sup>+</sup> cells. Scale bar 15µm. 63x objective.

strains, there was an increase in the abundance of apolipoprotein A1 (APOA1), apolipoprotein A4 (APOA4) and chitinaselike protein 4 (CHIL4), and a decrease in the abundance of canalicular multispecific organic anion transporter 2 (ABCC3) and pleioptropic regulator 1 (PLRG1) (**Figure 7A**). Although these proteins were similarly modulated at 40 hpi, the response to each clonal lineage appears to differ in other important ways. For example, mevalonate kinase (MVK), a key enzyme in both

sterol and isoprenoid synthesis, was upregulated following infection with T. gondii RH (**Figure 7B**). On the other hand, in VEG infected enteroids, we observed an increase in SPTLC1, which negatively regulates cholesterol efflux by biding to the ABCA1 transporter (**Figure 7C**).

The online bioinformatics tool, DAVID, was used to characterize the differentially expressed proteins according to biological process (GOTERM\_BP), cellular component (GOTERM\_CC) and molecular function (GOTERM\_MF) (**Supplementary Figures 1, 2**). For this analysis, exclusion criteria of p < 0.05 and fold change > 1.5 were used. The low fold change cut off reflects the low proportion of invaded cells in our model system, and the likelihood of a high background level of unperturbed host cells.

In both RH and VEG infected enteroids, we observed an enrichment of upregulated proteins assigned to GO terms related to extracellular exosomes (GO:0070062), and to cholesterol absorption, synthesis and transport. In addition, in VEG infected enteroids, we observed an enrichment of upregulated proteins assigned to GO:0090675: intermicrovillar adhesion, and GO:0030054: cell junction (**Supplementary Figure 2**). The latter category may be related to earlier findings suggesting that parasite-mediated disruption of cellular junctions aids in paracellular migration across the intestinal epithelium (Barragan et al., 2005; Weight and Carding, 2012; Weight et al., 2015; Jones et al., 2016).

## Atorvastatin Treatment Attenuates *T. gondii* Invasion and Replication in Enteroids

The ability to synthesize sterols and other isoprenoids via the mevalonate pathway is absent in T. gondii, which scavenges cholesterol from the cells it invades. Consequently, the observed upregulation of MVK, an enzyme acting early in the isoprenoid/sterol biosynthesis pathway, may support parasite growth. To test this, we targeted the host mevalonate pathway with Atorvastatin (an HMG-CoA reductase inhibitor) and assessed the effect of limiting host isoprenoid/sterol biosynthesis on parasite replication in the intestinal epithelium. When collagen-supported epithelial sheets were treated with 30µM Atorvastatin directly after exposure to RH tachyzoites, we noted a marked inhibition of parasite replication (**Figures 8A,B**). At this concentration, staining of F-actin revealed no gross morphological differences between control and Atorvastatin

FIGURE 5 | orthoslices of a single plane are shown to demonstrate staining localization. (A) Confocal images showing apical enrichment of F-actin (red) and basolateral localization of E-cadherin (green; scale bars = 20 µm). (B) Confocal images showing lateral expression of EpCAM (green; scale bar = 30µm). (C) Confocal images showing localization of Paneth cells (LYZ; green) and goblet cells (MUC2; green) in collagen supported-epithelial sheets and Matrigel®-grown enteroids (scale bars = 30µm). (D) Graphs depict the proportion of enteroid epithelial cells expressing LYZ (Paneth cells; *P* < 0.05 by Student's unpaired *t*-test) or MUC2 (Goblet cells; *P* < 0.05 by Student's unpaired *t*-test) in 3D enteroids and collagen-supported epithelial sheets, respectively. Each data point indicates the mean of an individual experiment. Pooled data from at least three independent experiments per condition are shown. Mean ± SEM is shown. \*indicates *p*<0.05.

TABLE 1 | Differentiated cell markers detected in collagen-supported epithelial sheets by label free mass spectrometry.


<sup>+</sup>*Detected in two of three biological samples only.*

<sup>∼</sup>*Detected in one of three biological samples only.*

treated epithelial sheets (**Supplementary Figure 3**). Although not statistically significant, we did observe an increase in LDH release upon Atorvastatin treatment, and therefore cannot rule out the possibility that the drug affected parasite replication indirectly, via host cell stress. In addition, a lower dose of Atorvastatin (5µM) produced variable effects on parasite replication (**Supplementary Figure 3**), leading us to conclude that de novo synthesis of cholesterol/isoprenoids by intestinal epithelial cells may support, but is not required for, parasite replication. These findings nevertheless provide evidence that

replication of *T. gondii* at 40 hpi (scale bars = 30µm).

point indicates the mean of an individual experiment) (D) Representative z-stack confocal images showing *T. gondii* invasion of the intestinal epithelium at 4 hpi, and

collagen-supported epithelial sheets can be used as an alternative in vitro model to study the effect of various perturbagens on enteric host-pathogen interactions.

# DISCUSSION

To study early interactions of T. gondii with the intestinal epithelium, we optimized and validated three enteroid based infection techniques: fragmentation of enteroids, microinjection, and development of collagen-supported epithelial sheets with an exposed apical surface. Label-free mass spectrometry identified key biological processes utilized by T. gondii during active replication within intestinal epithelial cells, demonstrating the applicability of enteroid-based models as alternatives to primitive in vitro culture systems for the study of enteric pathogens. These infection techniques have the potential to be adapted to enteroid cultures of other host species

(such as livestock), for which suitable infection models are lacking.

Since the establishment of enteroid cultures, there have been several attempts to adapt these cultures to 2-dimensional (2D) formats primarily using membrane supports (Moon et al., 2013; Noel et al., 2017; Moorefield et al., 2018; van der Hee et al., 2018), scaffold supports (Kim et al., 2014; Sims et al., 2017), or gel supports (Jabaji et al., 2013; Wang et al., 2017; Thorne et al., 2018). Here we validated a collagen-supported epithelial culture, as a T. gondii infection model. In our model, Paneth cells appeared in clusters within central regions of the epithelial sheet, reminiscent of a crypt-like niche. This agrees with a previous study, where Paneth cells were located adjacent to LGR5<sup>+</sup> stem cells in proliferative zones of enteroids cultured on ECM coated surfaces (Thorne et al., 2018). The sporadic distribution of goblet cells throughout the sheet was also reminiscent of in vivo patterning, where active Notch signaling inhibits secretory cell lineage in neighboring cells, and so clusters of secretory cell types are not usually seen (Peignon et al., 2011; Vandussen et al., 2012, 2015). The characteristics reported here are suggestive of microdomains of differentiated cell types in a polarized epithelium in vitro, making the collagen-supported epithelial model comparable to 3D enteroids in terms of complexity, and an advancement on current in vitro cell line models. In addition, the exposed lumen allows for practical large-scale application of pathogens, and therefore, the high-throughput screening of small molecule or protein pertubagens that may be challenging in whole enteroid or micro-injection culture systems. This could be exploited to screen the anti-parasitic and off-target effects of novel drugs in a relevant organ system.

Enteroids possess an architecture and cellular complexity reminiscent of the intestinal epithelium in vivo, and are therefore emerging as useful infection models for a variety of enteric pathogens (Finkbeiner et al., 2012; Klotz et al., 2012; Zhang et al., 2014; Vandussen et al., 2015; Wilson et al., 2015; Yin et al., 2015; Co et al., 2019). Two notable breakthroughs that have been achieved through the use of enteroids are the success of human norovirus and Cryptosporidium culture in vitro, relieving some of the challenges associated with these highly specialized pathogens (Ettayebi et al., 2016; Heo et al., 2018). Other adaptations of enteroids to establish infection include the reversal of epithelial polarity to generate "apical-out" cultures (Co et al., 2019) and microinjection into the luminal space (Wilson et al., 2015; Williamson et al., 2018). Moreover, 3D enteroids are now being generated from other sources such as feline, porcine and bovine stem cells or tissues (Powell and Behnke, 2017; Hamilton et al., 2018; Derricott et al., 2019). Developing these animal enteroids into a tractable infection models would allow for species-matched infection models for other relevant enteric pathogens such as Neospora and Cryptosporidium, as well as for studying sexual reproduction of T. gondii in the definitive feline host.

Although a variety of in vitro and in vivo models have been used to study T. gondii infection, host-pathogen interactions at the intestinal epithelium are still poorly understood (reviewed in Delgado Betancourt et al., 2019). The importance of this interaction should not be underestimated–it is crucial to controlling initial invasion and replication, as well as shaping the systemic immune response. Enteroid models are likely to be key to addressing this knowledge deficit, and as a consequence, the design of novel vaccines and adjuvants (Klotz et al., 2012; Delgado Betancourt et al., 2019).

We report an up-regulation in host MVK, a key enzyme in both sterol and isoprenoid synthesis, following infection of enteroids with T. gondii RH. In VEG infected enteroids, we observed an increase in SPTLC1, which negatively regulates cholesterol efflux by biding to the ABCA1 transporter (Tamehiro et al., 2008). Both modifications might be predicted to result in an increase in host cell cholesterol levels. T. gondii lacks the ability to synthesize sterols via the mevalonate pathway, and must scavenge cholesterol from the host cell (Coppens et al., 2000; Nolan et al., 2017). Similarly, while T. gondii can synthesize isoprenoid precursors in the apicoplast, optimal survival, and growth relies on the use of host cell isoprenoids (Li et al., 2013).

Intestinal epithelial cells can acquire cholesterol either by synthesis through the mevalonate pathway, absorption of dietary cholesterol, or by uptake of low-density lipoprotein (LDL) particles. The strategies used by Apicomplexa to salvage host cholesterol may depend on the identity of the host cell. In T. gondii infected Chinese hamster ovary cells, uptake of LDL particles is increased upon infection, in the absence of any increase in the activity of HMG-coA reductase (a key enzyme in the cholesterol biosynthesis pathway) (Coppens et al., 2000). On the contrary, T. gondii infected Human Foreskin Fibroblasts (HFF) show increased expression of key genes involved in the melavonate pathway of cholesterol biosynthesis, including HMG-CoA reductase (Blader et al., 2001). Finally, de novo cholesterol biosynthesis supports growth of the parasite in macrophages and HeLa cells, while provision of exogenous LDL has minimal effect (Cortez et al., 2009; Nishikawa et al., 2011).

It is not known whether T. gondii modulates uptake or de novo biosynthesis of cholesterol to invade and replicate in the intestinal epithelium, or whether it has a preferred source. Our enteroid models can help to address this question, providing physiologically relevant data to refine a growing literature on the potential role of therapeutic targeting of cholesterol pathways in T. gondii infection. In our experiments, blockade of de novo synthesis with the HMG-CoA reductase inhibitor, Atorvastatin, reduced parasite replication only at higher doses. It will be important to determine if provision of dietary cholesterol or excess LDL cholesterol affects the reliance on synthesis. Finally, it is worth noting that in addition to providing an essential resource for the parasite, increased cholesterol biosynthesis may also benefit the host by potentiating toll-like receptor signaling and the generation of a host-protective immune response (Tall and Yvan-Charvet, 2015).

In response to both strains of T. gondii we also observed an increase in host APOA1 and APOA4. APOA1 is a component of high-density lipoprotein (HDL). It promotes efflux of cholesterol from cells, and as a component of HDL, mediates the transport of excess cholesterol to the liver (Francis et al., 1995). APOA4 is a component of chylomicrons, a class of lipoprotein particle responsible for transporting dietary lipids away from the intestinal epithelium (Green et al., 1980). APOA1 is also associated with chylomicrons, but is transferred to HDL in blood. The observed increase in proteins associated with cholesterol efflux is seemingly at odds with the increase in enzymes related to the biosynthetic pathway, but could represent an attempt by the host cell to clear excess cholesterol produced in response to parasite invasion. Increased expression of APOA1 and APOA4 may also represent a more general response to infection. Intestinal epithelial cells upregulated cholesterol efflux proteins, including APOA1, in response to Citrobacter rodentium infection (Berger et al., 2017). Interestingly, this study also revealed an apparently contradictory response of intestinal epithelial cells to infection; increased cholesterol biosynthesis coupled with increased cholesterol efflux. The authors suggest that this may reflect the competing interests of host and pathogen. APOA1 can also regulate host immune responses and promotes tight junction formation to resolve allergen-induced airway inflammation (Park et al., 2013), which could indicate multiple roles for these proteins in modulating the host-pathogen interaction in the intestinal epithelium.

In summary, intestinal enteroids provide a physiologically relevant cellular landscape, for modeling the interaction between T. gondii and the host intestinal epithelium. This will allow us to better understand the dialogue between parasite and host early in infection, contributing to vaccine development and nextgeneration adjuvant development. Furthermore, they can be used to test the anti-parasitic and off-target effects of novel drugs, which may be host cell-type specific in nature. In this regard, collagen-supported epithelial sheet model we describe will be crucial in allowing for high-throughput infection studies.

# DATA AVAILABILITY

The proteomic datasets generated for this study can be found in the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD013306. Other raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

# ETHICS STATEMENT

Murine tissues used within this study were harvested from female specific-pathogen-free, C57B1/6J mice, aged between 6 and 12 weeks (Charles River, Margate, United Kingdom). In some experiments, mT/mG mice (Gt(ROSA)26Sortm4(ACTBtdTomato,-EGFP)Luo, The Jackson Laboratory) were used to visualize epithelial cell membranes, and parasite invaded cells. Prior to tissue harvest, mice were culled by cervical dislocation as outlined in Schedule 1 of the Animals (Scientific Procedures) Act 1986. Tissue use was approved by the UK Home Office (project license) and the University of Liverpool Animal Welfare and Ethical Review Body.

# AUTHOR CONTRIBUTIONS

BC, CD, NR, JW, and JC obtained funding. LL, LJ, SA, NR, CD, BC, JW, and JC designed the experimental work. LL, LJ, HD, SA, NR, CH, and JC performed the experiments. LL, LJ, SA, NR, and JC analyzed the data. LJ contributed sections of the manuscript. LL and JC wrote the manuscript. All authors reviewed, edited, and approved the final manuscript.

# FUNDING

The work was funded by Biotechnology and Biological Sciences Research Council (BBSRC) Doctoral Training Partnership Studentships (LL and LJ) and a BBSRC Tools and Resources Development Fund award (BB/M019071/1). The Zeiss LSM 880 multiphoton confocal upright microscope housed in the Centre for Cell Imaging at the University of Liverpool was funded by MRC grant number MR/M009114/1.

# ACKNOWLEDGMENTS

The authors gratefully acknowledge: Emily Lees and Gordon Dougan, Wellcome Trust Sanger Institute, for assistance with micro-injection technique, Marco Marcello, Jennifer Adcott and the Centre for Cell Imaging, University of Liverpool, for assistance with image capture, the staff of the Biomedical Services Unit (BSU), and Cathy Glover, Jenna Lowe and the Institute of Infection and Global Health (IGH) technical team for expert technical assistance.

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00300/full#supplementary-material

#### REFERENCES


Supplementary Video 1 | Enteroids from ROSA mT/mG mice were infected with *T. gondii* Pru-tdTom-Cre for 24 h and imaged by 2-photon microscopy. The movie shows that motile *T. gondii* were present within the enteroid lumen.

Supplementary Video 2 | Enteroids from ROSA mT/mG mice were infected with *T. gondii* Pru-tdTom-Cre and imaged by 2-photon microscopy. The movie shows an infected epithelial cell (Epithelial cell in green, *T. gondii* in red) being extruded into the organoid lumen.


investigate mucosal gut physiology and host-pathogen interactions. Sci. Rep. 7, 1–14. doi: 10.1038/srep46790


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

Copyright © 2019 Luu, Johnston, Derricott, Armstrong, Randle, Hartley, Duckworth, Campbell, Wastling and Coombes. 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.

# Multi-Omics Studies Demonstrate *Toxoplasma gondii*-Induced Metabolic Reprogramming of Murine Dendritic Cells

Kerrie E. Hargrave\*, Stuart Woods, Owain Millington, Susan Chalmers, Gareth D. Westrop and Craig W. Roberts\*

Strathclyde Institute of Pharmacy & Biomedical Sciences, University of Strathclyde, Glasgow, United Kingdom

#### *Edited by:*

Jeroen P. J. Saeij, University of California, Davis, United States

#### *Reviewed by:*

Cyrille Botté, INSERM U1209 Institut pour l'Avancée des Biosciences (IAB), France Martin Blume, Robert Koch Institute, Germany

#### *\*Correspondence:*

Kerrie E. Hargrave kerrie.hargrave@glasgow.ac.uk Craig W. Roberts c.w.roberts@strath.ac.uk

#### *Specialty section:*

This article was submitted to Parasite and Host, a section of the journal Frontiers in Cellular and Infection Microbiology

*Received:* 16 April 2019 *Accepted:* 12 August 2019 *Published:* 11 September 2019

#### *Citation:*

Hargrave KE, Woods S, Millington O, Chalmers S, Westrop GD and Roberts CW (2019) Multi-Omics Studies Demonstrate Toxoplasma gondii-Induced Metabolic Reprogramming of Murine Dendritic Cells. Front. Cell. Infect. Microbiol. 9:309. doi: 10.3389/fcimb.2019.00309 Toxoplasma gondii is capable of actively invading almost any mammalian cell type including phagocytes. Early events in phagocytic cells such as dendritic cells are not only key to establishing parasite infection, but conversely play a pivotal role in initiating host immunity. It is now recognized that in addition to changes in canonical immune markers and mediators, alteration in metabolism occurs upon activation of phagocytic cells. These metabolic changes are important for supporting the developing immune response, but can affect the availability of nutrients for intracellular pathogens including T. gondii. However, the interaction of T. gondii with these cells and particularly how infection changes their metabolism has not been extensively investigated. Herein, we use a multi-omics approach comprising transcriptomics and metabolomics validated with functional assays to better understand early events in these cells following infection. Analysis of the transcriptome of T. gondii infected bone marrow derived dendritic cells (BMDCs) revealed significant alterations in transcripts associated with cellular metabolism, activation of T cells, inflammation mediated chemokine and cytokine signaling pathways. Multivariant analysis of metabolomic data sets acquired through non-targeted liquid chromatography mass spectroscopy (LCMS) identified metabolites associated with glycolysis, the TCA cycle, oxidative phosphorylation and arginine metabolism as major discriminants between control uninfected and T. gondii infected cells. Consistent with these observations, glucose uptake and lactate dehydrogenase activity were upregulated in T. gondii infected BMDC cultures compared with control BMDCs. Conversely, BMDC mitochondrial membrane potential was reduced in T. gondii-infected cells relative to mitochondria of control BMDCs. These changes to energy metabolism, similar to what has been described following LPS stimulation of BMDCs and macrophages are often termed the Warburg effect. This metabolic reprogramming of cells has been suggested to be an important adaption that provides energy and precursors to facilitate phagocytosis, antigen processing and cytokine production. Other changes to BMDC metabolism are evident following T. gondii infection and include upregulation of arginine degradation concomitant with increased arginase-1 activity and ornithine and proline production. As T. gondii is an arginine auxotroph the resultant reduced cellular arginine levels are likely to curtail parasite multiplication. These results highlight the complex interplay of BMDCs and parasite metabolism within the developing immune response and the consequences for adaptive immunity and pathogen clearance.

Keywords: dendritic cells, immunometabolism, *Toxoplasma gondii*, LPS stimulation, Warburg effect, arginine metabolism, multi-omics

#### INTRODUCTION

Toxoplasma gondii is an obligate intracellular parasite that actively invades mammalian cells where it multiplies rapidly before actively egressing to infect other cells. It can infect almost any mammalian cell type including phagocytes such as dendritic cells (Dubey, 1998; Sibley, 2011). As an intracellular parasite T. gondii has evolved a close relationship with mammalian host cells and is known to rely on them for a number of nutrients (Silva et al., 2002; Dubey, 2004; Sibley et al., 2012). Conversely the host has evolved immune mechanisms that directly kill T. gondii or restrict the availability of certain nutrients and thus control the growth of the parasite (Denkers, 2010; Pifer and Yarovinsky, 2011). In addition, a considerable body of literature now recognizes that activation of phagocytes during infection not only results in alteration of canonical immune markers and mediators, but causes fundamental changes to metabolism with potential to direct the quality of the immune response. Therefore, the nature and combination of these early events in these cells may play a pivotal role in establishing T. gondii infection, initiating host immunity and determining the outcome of acute toxoplasmosis.

An important immune-mediated metabolic change known to occur in phagocytes is energy metabolism which is responsible for the production of ATP used in all energy dependent processes within cells (Hinkle and McCarty, 1978). In the most simplistic form, ATP can be formed through glycolysis in the cytosol and yields two ATP molecules. Pyruvate is made as a byproduct of this process and in anaerobic conditions is fermented to lactate. In aerobic conditions, pyruvate can be fed into the TCA cycle located within the mitochondria to produce thirty-six ATP molecules. This is dependent on a functioning electron transport chain (ETC) and this entire process is termed, oxidative phosphorylation. Unsurprisingly, ∼95% of a naïve immune cells' energy is generated via oxidative phosphorylation under normoxic conditions. However, it has been shown under certain circumstances, immune cells favor ATP production by glycolysis, even in an normoxic environment. This is known as aerobic glycolysis or the "Warburg effect" (Warburg et al., 1958; Reviewed by Palsson-McDermott and O'Neill, 2013). Changes in the TCA cycle can also affect a large variety of biosynthetic processes for which it provides essential intermediates. Thus, modulation of energy metabolism has proximal and distal functional benefits for immune cells including macrophages and dendritic cells as it provides the precursors for protein manufacture and fatty acid synthesis required for cytokine production and membrane remodeling.

The vast majority of work described thus far in the literature has used LPS, IFN-γ, or IL-4 as surrogate immune activators rather than live infections. Within the literature, metabolic reprogramming of LPS (innate activation), LPS + IFN-γ (classical activation) or IL-4 (alternative activation) treated macrophages has been well-documented (Pearce et al., 2013; Murray et al., 2014). Generally, in the literature, LPS stimulation is associated with numerous metabolic changes culminating in a shift from oxidative phosphorylation to aerobic glycolysis ("Warburg effect"). In contrast, IL-4 stimulation is associated with augmented oxidative phosphorylation and ATP production (Newsholme et al., 1986; O'Neill and Hardie, 2013; Jha et al., 2015). BMDCs are known to be responsive to many of the same mediators as macrophages and depending on their activation status, share a number of functions (O'Neill and Pearce, 2016). As the main antigen presenting cells in the body, DCs have the unique ability to initiate T cell activation by delivering antigen from the periphery to secondary lymphoid organs. In addition, DCs that encounter pathogens generate cytokines and chemokines attracting or activating other cell types at the site of infection (Austyn, 2016). In vitro granulcoyte (GM-CSF) differentiated BMDCs have been likened to in vivo murine iDCs (Guilliams et al., 2016) and studies have shown that stimulation with LPS induces increased glucose uptake, aerobic glycolysis and augmentation of the pentose phosphate pathway while suppressing oxidative phosphorylation (encompassing both the TCA and ETC) (Jantsch et al., 2008; Krawczyk et al., 2010). This leads to "enzymatic breaks" in the TCA cycle and intermediate accumulation which is used for other immune functions (Rubic et al., 2008; Infantino et al., 2011; Tannahill et al., 2013; Everts et al., 2014). In addition, a dichotomy exists between LPS stimulated and IL-4 treated BMDCs in terms of arginine metabolism. LPS stimulation of BMDCs, like macrophages, was found to promote increased arginine metabolism as determined by an increase in both inducible nitric oxide synthase (iNOS) and the production of citrulline and arginosuccinate. In contrast, IL-4 treatment induces the conversion of arginine into ornithine and proline via arginase (Corraliza et al., 1995; Munder, 1997; Van den Bossche et al., 2012; Reviewed by Thwe and Amiel, 2018).

Host cells recognize T. gondii through a number of pathogen associated molecular patterns (PAMPS) that are known to activate toll-like receptors (TLR2, 4 and 11) in mice and in theory have the ability to induce similar changes as seen with LPS stimulation of phagocytes (Butcher et al., 2005) Investigating how host cells reprogram their metabolism in response to intracellular pathogens is complex. Firstly, it is challenging to separate host from pathogen metabolism in many assays. Secondly, these pathogens have a number of secretory molecules that can also alter host cell metabolism for their own benefit. To overcome this, the transcripts can be aligned to the mouse genome and although there is a degree of similarity between host and parasite enzymes at the amino acid level, they are very different at the DNA level and can therefore be identified unambiguously. This strategy has been used previously for the study of a variety of different intracellular and extracellular pathogens (Blader et al., 2001; Chaussabel et al., 2003).

Herein, we used a multi-omics approach comprising transcriptomics and metabolomics on murine GM-CSF differentiated dendritic cells which we validated with functional studies to better understand how T. gondii metabolically reprograms dendritic cell metabolism in vitro and how this compares with the effect of LPS. We show that some observed changes are consistent with parasite evolved mechanisms to subvert the host immune response, while other are consistent with host derived mechanisms to control parasite dissemination.

# MATERIALS AND METHODS

#### Mice

Eight to ten-week-old, male BALB/c mice were bred and maintained in house at the Strathclyde Institute of Pharmacy and Biomedical Sciences, Glasgow, UK. All animal procedures including Schedule 1 conformed to guidelines from The Home Office of the UK Government.

#### Generation of Murine Dendritic Cells

Bone marrow derived dendritic cells were cultured by flushing femurs and tibia of 8 to 10-week-old BALB/c mice with RPMI 1640 supplemented with 10% heat inactivated FCS, 10% Granulocyte-monocyte colony stimulating factor (conditioned from the supernatant of x63 cells), 5 mM L-glutamine, 100 U/ml penicillin, 100 <sup>µ</sup>g/ ml streptomycin and incubated at 37◦ C for 7 days. After this time, adherent and semi-adherent cells were harvested and then seeded into either 24 well plates (1 × 10<sup>6</sup> cells/ ml) or 96 wells plates (1 × 10<sup>5</sup> cells/100 ul).

# Maintenance of Transfected *T. gondii* Prugniaud Strain

Tachyzoites were routinely maintained in confluent human foreskin fibroblasts (HFFs) grown in DMEM complete medium comprising of 10% fetal calf serum, 5 mM L-glutamine, 100 U/ml penicillin, 100 µg/ ml streptomycin and 50 U/ml amphotericin B at 37 ◦ C in 5% CO2.

#### Dendritic Cell Infection With *T. gondii*

For infection, 1 × 10<sup>6</sup> BMDCs were seeded into 24 well-plates in complete RPMI 1640 medium. The cells were infected with 1 × 10<sup>6</sup> type II Prugniaud strain T. gondii tachyzoites. In some experiments, dendritic cells were also treated with E.coli LPS (1 µl/ml).

#### Measuring Nitric Oxide in Cell Supernatant

Cell supernatant nitric oxide levels were determined by Griess assay. Cell supernatant or standards were added in equal volumes with Griess Reagent (1:1 ratio of 2% sulphanilamide in 5% H3PO<sup>4</sup> and 0.2% Napthylene diamine HCL in ddH2O) in a 96 well-plate and incubated in the dark for 10 min. Absorbance was read at 540 nm on a Spectromax 190 plate reader and serum nitrite concentrations were calculated against a standard curve.

#### Measurement of Arginase-1 Expression

Arginase activity was measured using an assay based on a reaction with α-isonitrosopropiophenon (ISPF) as described previously (Al-Mutairi et al., 2010). Briefly, BMDCs were grown on 24-well plates, exposed to agonist as appropriate and harvested in 50 ul lysis buffer (50 mM Tris-HCL, 10 mM MnCl2, 0.1% Triton X-100, 5 ug/ml pepstatin A, 5 ug/ml aprotinin, and 5 ug/ml antipain hydrochloride, pH 7.4). Arginine hydrolysis was carried out by incubating cell lyates with 25 ul of 0.5M L-arginine (pH 9.7) at 37 ◦ C for 60 min. The reaction was terminated by adding 400 ul of an acid solution (H2SO4, H3PO4, and H2O in a ratio of 1:3:7 and 25 ul of 9% solution of ISPF). Samples along with known urea standards were incubated at 95◦ C for 45 min and then allowed to cool for 10 min in the dark. Aliquots were added to wells of a 96 well-plate and absorbance read at 540 nm on a Spectromax 190 plate reader.

#### Flow Cytometry

Cells were harvested for staining and incubated with Fcblock for 10 min at room temperature. Cells were stained for 1 h at 4◦ C with CD11c—PE or FITC, CD40—APC, CD80— FITC, CD86—APC-cy7 (BD bioscience). Cells were fixed using Fix & Perm (Life Technologies) following manufacturer's instructions. Intracellular staining was performed using iNOS— PE (ebioscience). For glucose uptake assays, 50 ul fluorescent glucose analog (2-NBDG-FITC) was added to the cells 120 min before harvest. CFSE labeled T. gondii was used for uptake assays. A total of 30, 000–50, 000 events per sample was acquired on a BD FASCanto and data analysis was carried out using FlowJo software.

### Measurement of Lactate Dehydrogenase Activity

BMDCs were co-cultured or stimulated with LPS as previously described before the cells were harvested and lysed. Intracellular lactate dehydrogenase activity was measured following manufacturer's instructions. The lactate dehydrogenase activity assay kit was supplied by Sigma-Aldrich (Catlog number: MAK066). This kit reduces NAD to NADH, which is specifically detected as a colorimetric assay.

# Preparation of Extracts for LCMS

BMDCs were co-cultured with T. gondii or stimulated with LPS for 24 h before metabolite extraction. The extraction process involved cooling the cells for 10 min on ice to suppress further metabolic changes before the removal of the supernatant. The cells were washed twice with ice cold PBS (Thermo Fisher, UK) and an extraction mixture of cold methanol (VWR Chemicals, Leicestershire, UK), ddH20 and chloroform (VWR Chemicals) in a 60: 20: 60 ratio was added. The cells were scraped thoroughly, and the extracts were shaken at 1,400 rpm for 60 min at 4◦C in a thermomixer. After centrifugation at 12,000 × g for 15 min at 4 ◦C, the supernatant from these samples were then transferred to LCMS vials (Sigma-Aldrich/ Merck, Germany) and stored at −80◦C.

LCMS was carried out by Glasgow Polyomics, University of Glasgow, UK. Separation was performed using a zwitterionic ZIC pHILIC column (150 × 4.6 mm; 5 uM, Merck) on a Dionex Ultimate 3000 RSLC system (Thermo Fisher) with an injection volume of 10 ul and a flow rate of 0.3 ml/ min. The column was eluted on a gradient of mobile phase A, 20 mM ammonium carbonate pH 9.2 and mobile phase B, acetonitrile (ACN). Mass detection was carried out using an Orbitrap QExactive mass spectrometer (Thermo Fisher) operated in polarity switching mode.

#### Data Processing

Raw data files of metabolite standard solutions were processed at the University of Strathclyde using ToxID 2.1 (Thermo Fisher Scientific Inc., Hemel Hempstead, UK) with ± 3 ppm (parts per million) mass accuracy of both ESI positive and negative modes. After checking the appearance of the ion chromatograms with respect to peak shape, the standards were used to calibrate IDEOM v19. Raw files of the sample metabolites from the LCMS were processed by converting the data files to a universally accepted mxXML open file format using msConvert (ProteoWizard). Chromatograms were extracted using a detection algorithm from XCMS and stored in PeakMLfiles before aligning replicate peaks and combining them using mzMatch.R. A CSV file was generated after noise filtering and gap filling (quality control), which was then imported into IDEOM v19 for metabolite identification (Creek et al., 2012). This was based on accurate mass (± 3 ppm) and matching the observed retention time to a database of predicted retention times, based on physicochemical properties of metabolites calculated from their chemical structure. Chromatograms for individual metabolites were examined manually to check peak shape. All metabolites assigned an arbitrary confidence level ≤6 by IDEOM, representing a predicted retention time match of <35% (± 21 s), were rejected. Lipids and peptides were also excluded from the putatively identified metabolites in IDEOM.

#### Confirmation of Metabolite Identity

The identity of the metabolites was confirmed by accurate mass and matching the sample retention time to that of an authentic standard (± 0.3 min). The confirmed metabolites correspond to the metabolic standards initiative (MSI) level 1 whilst metabolites putatively identified by accurate mass and predicted retention time correspond to MSI level 2.

#### RNA-Sequencing

BMDCs activated with LPS or co-cultured with T. gondii were harvested after 6 h and the mRNA was extracted from the cell pellets using the RNeasy mini kit (Qiagen) with QIAshredder column (Qiagen, Manchester, UK). An agilent 2100 bio analyser was used to assess the quality of the RNA extracted (Agilent, Cheshire, UK). RNA samples with a concentration >20 ng/µl and RIN >8 was sent to be processed for RNA-seq using the Illuminia sequencing technology at Eurofins GATC Biotech, Konstanz, Germany. Bowtie was used to generate the reference alignments transcriptome alignments to align the RNA-seq reads to the reference transcriptome (mouse). Potential exon-exon splice junctions of the initial alignment were identified by Tophat. Cufflinks (part of CummeRbund software) was then used to identify and quantify the transcripts from a pre-processed RNA-seq alignment assembly. After this, Cuffmerge merges the pieces of the transcripts into full length transcripts and annotated the transcripts. Finally, merged transcripts from two (or more) samples were compared using Cuffdiff to determine the differential expression at transcript level. This includes giving a measure of significance (Benjamini-Hochberg correction) between the samples using fragment per kilobase per million mapped reads (FPKM) for each transcript. For interpretation purposes, each transcript is shown as Log<sup>2</sup> (fold change) compared to the control and illustrated in a heat map. The quality of the data obtained is outlined in the **Supplementary data**.

#### Mitochondrial Staining

BMDCs were plated at 1 × 10<sup>5</sup> into a Lab-tek chambered 1.0 borosilicate coverglass system (Catalog no: 155411; Thermo Scientific). Briefly, the cells were washed with PBS before the addition of Tetramethylrhodamine, Methyl Ester, Perchlorate (TMRM) and Mitotracker Green (Thermo fisher). All dyes were added at a final concentration of 100 nM each for 45 min in complete phenol red free RPMI. Cells were washed with PBS to remove extracellular Mitotracker Green, but TMRM was retained in the imaging PBS (as it is freely permeable and equilibrates across membranes in a Nernstian-manner). The cells were placed in a 37◦C stagetop chamber (OKO Labs H301-mini) for imaging. BMDCs were imaged on a Nikon Eclipse Ti inverted epifluorescence microscope with a 100x 1.3 NA oil immersion objective lens plus a 1.5x internal microscope adapter lens (Nikon) and Flash 4.0 CMOS camera (Hamamatsu). The samples were imaged with an excitation light of 470nm (Mitotracker Green) and 550 nm (TMRM) (pE4000 LED light source, CoolLED). Camera recording and excitation light controlled by Win Fluor v3.9.1 software (John Dempster, University of Strathclyde). Image analysis was performed using FIJI (Schindelin et al., 2012). All MitoTracker Green and TMRM images were set to the same intensity scale and then converted from 16 to 8-bit to allow application of a mean intensity threshold. Regions of interest were drawn manually around each cell periphery and the threshold set such that only the mitochondria were selected, equally for each pair of images. The "Multi Measure" function of FIJI was then used to measure the mean fluorescence intensity within each cell for both MitoTracker Green and TMRM and the ratio of (mean TMRM)/(mean MitoTracker Green) taken as a measure of relative mitochondrial membrane potential (membrane-potential-dependent fluorophore/membrane-potential-independent fluorophore).

# Cytokine Bead Array (CBA)

Concentrations of different cytokines were determined in the supernatant of naïve BMDCs, LPS stimulated BMDCs and BMDCs co-cultured with T. gondii using the LegendplexTM Mouse Inflammation Panel (Cat no; 740150; Biolegend, UK). The assay was carried out as per the manufacturer's instructions.

#### Data Analysis

Prior to Principal component analysis (PCA) and Orthogonal partial least squares discriminant analysis (OPLS-DA), the data were mean-centered and Pareto (Par) scaled. PCA, OPLS, and VIP (Variable importance in the projection) plots were performed using SIMCA-P13 (Umetrics, Sweden). For metabolomics data, a non-parametric one-tailed Mann Whitney test was used for all individual putatively identified metabolites. p < 0.05 was considered significant. Graph Pad prism 7 was used for plotting the graphs and heat maps. Heat-maps show the fold change in each metabolite compared to the control. All functional data was performed by a one-way ANOVA with either a Bonferroni or Dunnett post-test unless otherwise stated. Throughout this thesis, <sup>∗</sup>p < 0.05, ∗∗p < 0.01, and ∗∗∗p < 0.001.

# RESULTS

#### *T. gondii* Actively Invades BMDCs

CFSE labeled T. gondii parasites (MOI 1:1) were co-cultured with BMDCs to establish the extent these parasites were internalized. Flow cytometry was used to identify CFSE labeled T. gondii in CD11c+ cells. Paraformaldehyde fixed parasites were used as a comparator to determine the potential of BMDCs to internalize dead parasites. CFSE labeled live T. gondii were found in approximately 80% of CD11c<sup>+</sup> cells while PFA fixed T. gondii were found in significantly fewer BMDCs (approximately 40%) (**Figure 1**).

#### *T. gondii* Augments CD40, CD80 and CD86 Transcript and Protein Levels, but Down Regulates MHC Transcripts

CD80 and CD86 work in concert with MHCI and MHCII to activate T cells. The activation profiles of BMDCs cocultured with T. gondii or stimulated with LPS were determined by flow cytometry and transcriptomics. In comparison to naïve BMDCs, those co-cultured with T. gondii or stimulated with LPS had significantly increased CD40 and CD80 protein levels (**Figures 2B,C**). In contrast CD86 was only seen to be upregulated in BMDCs co-cultured with T. gondii (**Figure 2C**). Many of these changes also were evident in the transcriptomic data including cd80 and cd86 mRNA transcripts which were upregulated in BMDCs co-cultured with T. gondii. Similarly, transcripts for cd40, cd80, and cd86 were significantly upregulated in LPS stimulated BMDCs. (**Figure 2A**). A number of transcripts for MHC class I and II molecules were significantly downregulated in BMDCs co-cultured with T. gondii. In contrast, transcripts for MHC class I and II molecules were generally significantly upregulated in BMDCs stimulated with LPS compared with control cells (**Figure 2A**).

# *T. gondii* Influences the Cytokine Profile of BMDCs at Both the mRNA and Protein Level

to determine significance (P < 0.01) where \*\*\*p < 0.001.

Cytokine production by BMDCs co-cultured with T. gondii or stimulated with LPS were measured in culture supernatants by cytometric bead array (CBA) and compared with the transcriptomic data to validate the RNA sequencing process. The cytokine protein levels of IL-1α, IL-12p70, IL-6, and IL-10 were significantly increased in T. gondii co-cultured BMDCs. This was largely consistent with the transcriptomic data which found significant upregulation of Il1a, Il12a, and Il6 transcripts. Similarly, LPS significantly increased the mRNA and protein levels compared to naïve BMDCs, for of all cytokines measured (**Figure 3**).

#### Dendritic Cells Co-cultured With *T. gondii* or Stimulated With LPS Undergo Distinct but Overlapping Transcript and Metabolic Changes

Principal component analysis (PCA) demonstrated distinct separation between T. gondii infected and naïve BMDCs (**Figure 4**). Furthermore, Orthogonal partial least squares discriminant analysis (OPLS-DA) was used to generate a ranked table of variable importance in projection (VIP) table for metabolites and transcripts (**Supplementary Tables 1–4**). The transcripts were aligned to a Mus musculus reference gene ensuring there is no T. gondii specific transcripts and all raw

shows no changes from the control cells. In addition, cells were analyzed by flow cytometry after 24 h to determine the proportion of CD11c+ that cells expressed the activation markers (B) CD40, (C) CD80, or (D) CD86. Results are representative of one experiment for transcriptomics and three independent experiments for flow cytometry and show mean ± SEM. Statistical analysis was performed (A) using Cuffdiff (Eurofin) to determine the differential expression levels at the transcript level including a measure of significance between samples. (B–D) one-way ANOVA with Bonferroni post-test was performed to determine significance (P < 0.05) where \*p < 0.05, \*\*p < 0.001 and \*\*\*p < 0.0001 are significant compared to unstimulated.

signal values as well as fold- change values for this entire data set are publicly available (**Supplementary Table 5**).

As expected, most transcripts scoring high in the pair wise OPLS-DA analyses comparisons between control cells and T. gondii co-cultured or LPS stimulated cells were immune related or involved with signaling molecules/transcription factors. Interestingly, many transcripts associated with energy metabolism including glycolysis, TCA cycle and OXPHOS as well as arginine metabolism were also ranked within the initial 100 transcripts of interest (**Supplementary Tables 1**, **2**). Similarly, multiple metabolites from these pathways were highlighted within the first 50 metabolites of the metabolome VIP tables and heatmaps (**Supplementary Tables 3**, **4** and **Supplementary Figure 1**). From both VIP tables, it is clear that T. gondii induces metabolic changes in BMDCs.

FIGURE 3 | mRNA and protein levels of cytokines expressed by BMDCs when co-cultured with T. gondii or stimulated with LPS. Bone marrow derived DCs were either co-cultured with T. gondii or stimulated with LPS as indicated for 6 or 24 h. After 6 h, (A) mRNA was subsequently extracted and then quantified using RNA-seq. Data shows the FPKM values. Red indicates an increase in mRNA expression, green indicates a decrease in transcript expression and gray shows no changes from the control cells. (B) A cytokine bead array was performed on the supernatant to obtain cytokine protein levels. Results are representative of one experiment for transcriptomics and three independent experiments for flow cytometry and show mean ± SEM. Statistical analysis was performed (A) using Cuffdiff (Eurofin) to determine the differential expression levels at the transcript level including a measure of significance between samples. (B) One-way ANOVA with Dunnett post-test. Significant differences are p < 0.05 where \* is compared to unstimulated.

Within this paper, only transcript and metabolic changes associated with live T. gondii infection will be discussed in great detail (**Supplementary Figure 1**). A limitation of investigating the metabolic profile of BMDCs infected with live parasites is distinguishing whether deviations of the metabolites described herein are associated with changes in BMDC function or reflect metabolites of the parasite. In an attempt to understand the host contribution to the observed metabolic changes in the absence of live parasites, paraformaldehyde (PFA) fixed T. gondii or Toxoplasma lysate antigen (TLA) were also used to stimulate BMDCs. Results found either a marked reduction or only relatively minor changes in cultures incubated with PFA fixed parasites or TLA compared unstimulated BMDCs. This implies that activation of parasite specific TLR receptors alone is insufficient to modify the metabolism of BMDCs and that an active infection is required (**Supplementary Figure 1.7**). Consequently, we used both metabolomic data and transcriptomic analysis in concert to gain insight into the likely contribution of the host metabolism during T. gondii infection.

## *T. gondii* Infection Induces Aerobic Glycolysis of BMDCs

Glycolysis is the first step in the breakdown of glucose to produce the high-energy molecules, ATP and NADH. Within this pathway, glucose is metabolized into pyruvate (under aerobic conditions) or lactate (in an anaerobic environment) (**Figure 5A**).

Distinct differences in glycolytic transcripts and metabolites were observed in BMDCs following co-culture with T. gondii or following LPS stimulation for 24 h. LCMS analysis detected increased levels of the glycolytic metabolites, glucose, glucose 6 phosphate, fructose 6 phosphate, 2 phosphoglycerate, pyruvate and lactate in BMDCs co-cultured with T. gondii compared with control BMDCs (**Figure 5B**). These differences were similar to those observed in LPS-stimulated BMDCs. Transcript levels generally did not reflect this increase in glycolysis as BMDCs co-cultured with T. gondii had decreased levels of key transcripts encoding glycolytic enzymes phosphoglycerate mutase (pgam2) and enolase (eno3). Transcription of genes for the glucose transporters GLUT1 (slc2a1) and GLUT3 (slc2a3) was also reduced but there were modest increases in monocarboxylate transporters MCT1 (slc16a1) and MCT6 (slc16a6) (**Figure 5C**). MCT1 is a proton dependent transporter for lactate, pyruvate and ketone bodies (Fisel et al., 2018). MCT6 has the similar substrate specificity in vitro but its function is unknown. BMDCs stimulated with LPS had significantly increased levels of glucose transporters GLUT1, GLUT6 (slc2a1 and slc2a6), hexokinases (hk1, hk2, hk3), 6 phosphofructokinase b3 (pfkfb3), lactate dehydrogenase a (ldha) and monocarboxylate transporters MCT1 (slc16a1) and MCT10 (slc16a10). There were significantly decreased levels of glucose transporters GLUT3, GLUT8 and GLUT9 (slc2a3, slc2a8, slc2a9) glycolytic enzymes glucose 6 phosphate isomerase (gpi), 6 phosphofructokinase m (pfkm), aldolase c (aldoc), phosphoglycerate mutase (pgam) and monocarboxylate transporters MCT6, MCT7, and MCT13 (slc16a6, slc16a7, and slc16a13) (**Figure 5C**). MCT10 is required for transport of aromatic amino acids, GLUT9 has been implicated in urate transport but the physiological roles of

then analyzed by SIMCA and a Principal component analysis plot was generated. Key: Blue, unstimulated; Green, LPS and Red, T. gondii.

GLUT6, GLUT8, MCT6, MCT7, and MCT13 have not been determined (Cura and Carruthers, 2012; Fisel et al., 2018). Overall, observed transcripts and metabolites associated with glycolysis trend toward an induction of this pathway upon T. gondii co-culture and LPS stimulation.

# *T. gondii* Increases Glucose Uptake and Up-Regulates LDH Activity in BMDCs

To further confirm the observed increase in glycolysis highlighted in the transcriptomics and metabolomic data, the ability of T. gondii to alter BMDC uptake of 2-NBDG (as a proxy for glucose) and influence lactate dehydrogenase (LDH) activity was used. Co-culture of BMDCs with T. gondii significantly increased glucose uptake and elevated LDH activity compared with control BMDCs. An increase in glucose uptake and up-regulation in LDH activity was also observed in BMDCs stimulated with LPS (**Figures 6A,B**). These findings confirm that upon T. gondii infection, BMDCs increase their capability to take up glucose favoring glycolysis.

# *T. gondii* Infection Alters the TCA Cycle in BMDCs

Once metabolized from pyruvate (the end product of glycolysis), acetyl-CoA can enter the TCA cycle. The key reaction of the cycle is the reduction of NAD+ into NADH, which can then be introduced into the electron transport chain (i.e., OXPHOS) to generate ATP (**Figure 7A**). LCMS analysis from

FIGURE 5 | Glycolysis differences in T. gondii co-cultured or LPS stimulated BMDCs. Bone marrow derived DCs were co-cultured with T. gondii or stimulated with LPS as indicated for 6 or 24 h. After this time, the metabolites of the BMDCs were (B) extracted and analyzed by Liquid Chromatography Mass Spectroscopy (LCMS) (Continued)

FIGURE 5 | or (C) mRNA was extracted and then quantified using RNA-seq. Data shows the log2 (fold change) as normalized to unstimulated controls. Red indicates an increase in transcript expression, green indicates a decrease in transcript expression and gray shows no changes from the control cells. For clarity, the diagram shown in (A) focuses on relevant metabolites (shown in the graphs) and enzymes [depicted (a) through to (l)] for glycolysis only where (a) hexokinase; (b) phosphoglucose isomerase; (c) phosphofructokinases; (d,e) aldolase; (f) triose phosphate isomerase; (g) glyceraldehyde 3 phosphate dehydrogenase; (h) phosphoglycerate kinase; (i) phosphoglycerate mutase (j) enolase; (k) pyruvate kinase and (l) lactate dehydrogenase. Statistical analysis was either performed using (B) a non-parametric Mann Whitney test, (C) Cuffdiff (Eurofin) to determine the differential expression levels at the transcript level including a measure of significance between samples. Significant differences are p < 0.05 where \* is compared to unstimulated. Results are representative of three independent runs.

co-culturing T. gondii with BMDCs for 24 h demonstrated increased levels of citrate and malate in T. gondii co-cultured BMDCs (**Figure 7B**). A decrease in the majority of mRNA transcripts in these cultures could be observed although only succinate dehydrogenase B (sdhb) was statistically significant in itself (**Figure 7C**). Similarly, many TCA intermediates were up-regulated in LPS stimulated BMDCs including citrate, alpha-ketoglutarate, fumarate and malate. Itaconate was observed to be significantly down-regulated compared to naïve cells (**Figure 7B**). Significant down-regulation of transcripts for isocitrate dehydrogenase A, B and G (idh3g, idh3b, idha), alpha ketoglutarate dehydrogenase (ogdh), succinate-coA ligase (suclg1) and fumarate hydratase (fh) were evident in LPS activated BMDCs in comparison to unstimulated BMDCs. Significant upregulation of dihydrolipoamide Ssuccinyltransferase (dlst) were observed in LPS stimulated BMDCs (**Figure 7C**).

Transcripts for citrate synthase (cs) were upregulated in LPS stimulated, but not T. gondii infected BMDCs. Similarly, the citrate carrier (slc25a1) that transports citrate from the mitochondria to the cytoplasm was markedly upregulated in LPS stimulated BMDCs, but only modestly raised in T. gondii infected BMDCs. Examination of potential routes for citrate usage in the cytoplasm of LPS stimulated cells revealed an increase in ATP citrate lyase (acly) which is responsible for the conversion of citrate to acetyl coA. However, transcripts for acetyl-CoA carboxylase (acaca) and fatty acid synthase (fasn) which consecutively convert acetyl coA to fatty acid were downregulated in these cells. In contrast in T. gondii infected cells no change in ATP citrate lyase (acly) transcripts were noticed, but an increase in acetyl-CoA carboxylase (acaca) and fatty acid synthase (fasn) were observed.

Generally, from our multi-omics data-set, we observed a decrease in the majority of TCA transcripts, but an increase in many metabolites associated with the TCA cycle. This is in agreement with current metabolomic studies in LPS stimulated macrophages and dendritic cells that show blockage of specific TCA cycle enzymes leading to the increase of certain intermediates. It has been shown that these metabolites have an immunoregulatory role within host cell cells.

#### *T. gondii* Effects on the Electron Transport Chain and Mitochondrial Membrane Potential

Due to the direct link between the TCA cycle and electron transport chain, the effect of T. gondii infection on transcripts encoding components of the electron transport chain (ETC) was assessed. Down regulation of components of the ETC was observed in T. gondii infected and LPS stimulated BMDCs. Generally, complex I (NADH dehydrogenase), complex II

FIGURE 7 | TCA changes in T. gondii co-cultured or LPS stimulated BMDCs. Bone marrow derived DCs were co-cultured with T. gondii or stimulated with LPS as indicated for 6 or 24 h. After this time, the metabolites of the BMDCs were (B) extracted and analyzed by Liquid Chromatography Mass Spectroscopy (LCMS) or (Continued) FIGURE 7 | (C) mRNA was extracted and then quantified using RNA-seq. Data shows the log2(fold change) as normalized to unstimulated controls. Red indicates an increase in transcript expression, green indicates a decrease in transcript expression and gray shows no changes from the control cells. For clarity, the diagram shown in (A) focuses on relevant metabolites (shown in the graphs) and enzymes [depicted (a) through to (j)] for the TCA cycle only. (a) citrate synthase; (b) aconitase 2; (c) isocitrate dehydrogenase 3; (d) alpha-ketoglutarate dehydrogenase; (e) succinyl-CoA synthetase; (f) succinate dehydrogenase; (g) fumarase; (h,i) malate dehydrogenase; (j) cis-aconitate decarboxylase; (k) mitochondrial citrate carrier; (l) ATP citrate lyase; (m) acetyl CoA carboxylase; (n) fatty acid synthase; (o) aconitase 1 and (p) isocitrate dehydrogenase 1. Statistical analysis was either performed using (B) a non-parametric Mann Whitney test, (C) Cuffdiff (Eurofin) to determine the differential expression levels at the transcript level including a measure of significance between samples. Significant differences are p < 0.05 where \* is compared to unstimulated. Results are representative of three independent runs (n = 3).

(succinate dehydrogenase) and complex 3 (cytochrome bc1 complex) were unaffected or downregulated (of note, ndufa6 and sdhb were significantly (p<0.05) down-regulated). Previous literature shows that a downregulation of the electron transport chain is expected upon BMDC activation, as limiting ATP production for energy allows TCA metabolites to be used in other biosynthetic pathways e.g., fatty acid biosynthesis. However, interestingly, transcripts encoding subunits of complex IV and V (ATP synthase) were generally found to be upregulated in BMDCs following exposure to T. gondii. Statistically significantly upregulated subunits were atp6v0b and atp2b4. This is a relatively new concept with studies demonstrating that subunits IV and V of the ETC have roles other than energy generation including heme A biosynthesis (cox11), transport of different ions and maintenance of calcium homeostasis (atp2b4), mediating acidification (atp6v0b) (**Figure 8A**).

Moreover, important differences in the mitochondrial membrane potential (an indicator of oxidative phosphorylation) between naïve and T. gondii-infected BMDCs were observed using epifluorescence microscopy. After 24 h of co-culture between BMDCs and T. gondii, Mitotracker Green was used to stain mitochondria in combination with TMRM to visualize mitochondrial membrane potential. TMRM accumulates in active mitochondria with an intact membrane potential. A color change from red/orange to yellow/green when Mitotracker green and TMRM signals were merged demonstrates that T. gondii and LPS reduce mitochondrial membrane potential (**Figures 8B–D**). Furthermore, significant downregulation in the relative mitochondrial membrane potential (TMRM/MTG) mean per cell of T. gondii infected and LPS stimulated BMDCs could be observed compared to naïve BMDCs (**Figure 8E**). Overall, this indicates a decrease in oxidative phosphorylation in T. gondii infected and LPS stimulated BMDCs.

#### *T. gondii* Infected BMDCs Direct Arginine Metabolism to Polyamines and Proline

Independent of energy metabolism in T. gondii infected BMDCs, RNA sequencing and LCMS analysis detected a number of transcripts and metabolites associated with arginine metabolism in T. gondii co-cultured BMDCs (**Figure 9A**). The metabolism of arginine is associated with increased production of toxic mediators deleterious to many pathogens. Notably, in T. gondii infected BMDC cultures co-cultured for either 6 h (RNA) or 24 h (metabolites), arginine levels were similar to control BMDCs and only a small increase in citrulline levels was evident. However, significant increases in L-ornithine, L-proline, L-1-pyrroline-3-hydroxy-5-carboxylate and L-glutamate were observed in T. gondii co-cultured BMDCs. This differed from LPS stimulation which induced significantly increased levels of citrulline, N- (omega)-hydroxyarginine and arginosuccinate as well as Lproline and L-1-pyrroline-3-hydroxy-5-carboxylate (**Figure 9B**). Transcripts for a number of enzymes involved in arginine conversion to proline and glutamate were also seen to be collectively augmented, but none other than spermidine synthase (srm) were individually significantly increased in T. gondii infected BMDC cultures (**Figure 9C**).

In keeping with the omics data, intracellular iNOS was detected in 80% of CD11c+ BMDCs stimulated with LPS, but in <10% of cells from T. gondii infected cultures. Furthermore, nitrite levels in LPS stimulated, but not T. gondiiinfected BMDCs were significantly increased (**Figures 10A,B**). Arginase enzyme activity was upregulated in BMDCs cultures infected with T. gondii or stimulated with LPS. Collectively, these observations support that arginine is primarily converted to ornithine and polyamines in T. gondii infected cells (**Figure 11**). In contrast, LPS stimulation primarily induces arginine conversion to citrulline and recycling through N- (L-arginosuccinate) with accompanying NO production. By combining metabolomic, transcriptomic data with conventional functional studies reveals that T. gondii is able to divert arginine favorably toward ornithine and proline via arginase.

# DISCUSSION

Previous studies using microarray analyses have examined global changes in gene expression for human foreskin fibroblasts, macrophages and DCs in response to T. gondii (Blader et al., 2001; Chaussabel et al., 2003). These studies were conducted before the development of RNASeq (which allows almost complete coverage of the transcriptome) and non-targeted metabolomic analyses LCMS techniques. The current studies were initiated to make use of both these modern techniques in concert with conventional functional studies to validate the findings. We elected to use murine cells to complement and enrich the datasets available and to allow extrapolation to the many murine studies already published.

Recent studies have described how macrophages and to some extent BMDCs undergo metabolic changes when exposed to environmental stimuli. Generally, LPS stimulation culminates in a shift from oxidative phosphorylation to aerobic glycolysis, whereas augmented oxidative phosphorylation and ATP production occurs in those stimulated with IL-4 (Newsholme et al., 1986; Jantsch et al., 2008; Rubic et al., 2008; Krawczyk et al., 2010; Infantino et al., 2011; Everts et al., 2012; O'Neill and Hardie,

(Continued)

FIGURE 8 | normalized to unstimulated controls. Red indicates an increase in transcript expression, green indicates a decrease in transcript expression and gray shows no change from control cells. For Clarity, relevant gene abbreviates are shown in the diagram where Complex I is NADH dehydrogenase (ndufs1-bdufb6), Complex II is Succinate dehydrogenase (sdhc-sdha), Complex III is Cytochrome bc1 complex (uqcrc1-uqcrfs1), Complex IV is Cytochrome C oxidase (cox7b–cox15) and Complex V is ATP synthase (Atp6ap2–atp5k). Statistical analysis was performed using Cuffdiff (Eurofin) to determine the differential expression levels at the transcript including a measure of significance between samples. (B–D) After 24 h, the cells were stained with 100 nM of Mitotracker Green and 100 nM TMRM for 30 min to determine mitochondrial morphology. The images show separate images for Bright field (BF), Mitotracker green (MtG), TMRM and then MtG and TMRM merged (TMRM/MTG). Examples of areas used for quantification of relative mitochondrial potential are shown zoomed in yellow dashed boxes. Images are representative of at least 3 independent runs per group, with excitation light intensity and image intensity kept constant throughout. (E) Relative mitochondrial membrane potential for control BMDCs, BMDCs stimulated with LPS or BMDCs infected with T. gondii (determined as described in the material and methods section). Statistical analysis performed using non-parametric Kruskal-Wallace ANOVA with Dunn's post-hoc test (OriginPro), with \*\*\*p < 0.001; individual cell values overlaid with median ± 95% confidence interval.

2013; Tannahill et al., 2013; Jha et al., 2015; O'Neill and Pearce, 2016). However, the effect of intracellular pathogens on cellular metabolism is less well-studied and indeed more complex as host cells undergo metabolic changes that have evolved to directly kill pathogens or restrict access to essential nutrients (Denkers, 2010; Pifer and Yarovinsky, 2011). Conversely pathogens have evolved mechanisms to subvert host cell metabolism for their own benefit. As these early events are important for establishment of infection, initiating host immunity and determining the eventual success of infection we have employed a multi-omics approach to inform confirmatory functional assays.

Initial studies determined that using the procedures outlined, a T. gondii infection rate of approximately 80% could be achieved in BMDCs with concurrent upregulation of costimulatory molecules (CD40, CD80, and CD86) and their transcripts cd80 and cd86. A similar pattern of changes was also observed by BMDCs stimulated with LPS. This is consistent with the literature (Verhasselt et al., 1997; Morgado et al., 2014). In keeping with the literature that demonstrates the ability of T. gondii infection to inhibit MHCI and II expression in macrophages, the majority of MHC Class I and II transcripts were down-regulated in T. gondii co-cultured BMDCs (Lüder et al., 2001). In contrast, LPS stimulated BMDCs had elevated levels of many transcripts associated with MHC I and II complex. Additionally, T. gondii infected BMDCs had congruent elevated expression of both mRNA and protein levels of proinflammatory cytokines IL-1α, IL-12, and IL-6 (Nam et al., 2011). These data validate the model of infection and the multiomics approach where metabolomic, transcriptomic data and functional studies can complement, be related to, and add value to each other.

Principal component analysis and Orthogonal projection to latent structure for discriminant analyses (OPLS-DA) was used to determine the major changes identified through RNA sequencing and non-targeted liquid chromatography mass spectroscopy (LCMS). From the VIP tables generated from our OPLS-DA analysis, metabolites associated with energy metabolism including glycolysis, the TCA cycle and oxidative phosphorylation were determined as major discriminants (i.e., the first 100 transcripts/ metabolites) between control uninfected and T. gondii infected cells.

Glycolysis is metabolic pathway that converts glucose into pyruvate. The energy released in this process is used by downstream metabolic processes to form ATP and NADH. From transcriptomic and metabolic analyses, T. gondii cocultured with BMDCs for 24 h demonstrated an increase in glycolysis intermediates (glucose 6 phosphate, fructose 6 phosphate, 2 phosphoglycerate, pyruvate and lactate) when compared with control BMDCs. Our results also found that LPS stimulation of dendritic cells had a similar effect on glycolysis intermediates. In keeping with these results, we noted increased levels of transcripts for some glycolytic enzymes including hk and pfkfb3, the highly active isoform of phosphofructokinase. We noted a reduction in phosphoenolpyruvate in infected cultures, which could be linked to the downregulation of Eno3 or alternatively due to parasite utilization of this resource. Similarly, up-regulation of ACCase and FAS may allow T. gondii to take advantage of the increased production of pyruvate and acetyl-CoA from TCA intermediates that is then used to increase glycolytic demand or FA synthesis (in the apicoplast of the parasite). Conversely, the parasite could contribute to the observed increase in pyruvate we report in infected cultures. Thus, T. gondii has the opportunity to contribute to metabolites measured and to exploit host cell resources. Again, LPS stimulated BMDCs underwent similar, but more profound changes to their transcripts encoding glycolytic intermediates.

Through functional assays we were able to demonstrate that even though T. gondii infected BMDC had reduced transcript expression of the glucose transporter GLUT1 (slc2a1), they had increased uptake of 2-NBDG as a surrogate for glucose and had increased levels of LDH activity compared with control BMDCs. This apparent discrepancy could be due to temporal differences in RNA expression and GLUT1 activity or reflect increased efficiency of transport of glucose due to increased utilization, recruitment of intracellular sugar transporters to the plasma membrane or activation of GLUT1 by reduction in ATP levels (Cura and Carruthers, 2012). Increased levels of lactate and LDH activity was accompanied by increased expression of MCT1, a protein responsible for lactate efflux required to prevent intracellular acidification (Fisel et al., 2018). Interestingly, LDH activity also correlates with an increase metabolites (3-(4- hydroxyphenyl) lactate and indolelactate) produced by LDH from products of aromatic amino acid catabolism. Collectively, these results demonstrate that both T. gondii infection of BMDCs and LPS stimulation of BMDCs increase glucose uptake, glycolysis and lactate production in

(Continued)

FIGURE 9 | Spectroscopy (LCMS) or (C) mRNA was extracted and then quantified using RNA-seq. Data shows the log2(fold change) as normalized to unstimulated controls. Red indicates an increase in transcript expression, green indicates a decrease in transcript expression and gray shows no changes from the control cells. For clarity, the diagram shown in (A) focuses on relevant metabolites (shown in the graphs) and enzymes [depicted (a) through to (h)] for arginine metabolism only: (a) NOS; (b) argininosuccinate synthase; (c) argininosuccinate lyase; (d) arginase; (e) ornithine decarboxylase; (f) ornithine aminotransferase; (g) spermidine synthase; (h) spermine synthase; and (i) △1-pyrroline- 5-carboxylate reductase. Statistical analysis was performed using (B) non-parametric Mann Whitney test, (C) Cuffdiff (Eurofin) to determine the differential expression levels at the transcript including a measure of significance between samples. Significant differences are p < 0.05 where \* is compared to unstimulated. Results are representative of three independent runs (n = 3).

cells even under normoxic conditions, a phenomenon sometimes referred to as aerobic glycolysis or the Warburg effect (Warburg et al., 1958; Reviewed by Rodríguez-Prados et al., 2010; Kelly and O'Neill, 2015).

The Warburg effect is not only associated with aerobic glycolysis, but also by a down-regulation of the TCA cycle and this has already been reported in macrophages stimulated with LPS (Jha et al., 2015). The majority of mRNA transcripts associated with the TCA cycle were down-regulated in LPS activated and T. gondii infected BMDCs compared to naïve BMDCs. As a predicted physiological consequence of such circumstances we also detected increased levels of many TCA intermediates in T. gondii infected BMDCs and LPS stimulated BMDCs including citrate and malate. In the literature, a buildup of specific TCA intermediates and reduction in TCA cycle for ATP production has been reported in LPS stimulated macrophages. It has been suggested that this facilitates a variety of biosynthetic processes that depend on these intermediates (O'Neill, 2011; Tannahill et al., 2013; Lampropoulou et al., 2016; Mills et al., 2016). For example, citrate itaconic accumulates due to breaks in the TCA cycle between the isocitrate and alpha-ketoglutarate conversion step (catalyzed by isocitrate dehydrogenase) (Infantino et al., 2013; Jha et al., 2015). Citrate can also be replenished by the anaplerotic reactions of the TCA cycle (conversion of aspartate into oxaloacetate by aspartate aminotransferase). Export of citrate by the mitochondrial citrate carrier is accompanied by uptake of malate from the cytoplasm (converted to citrate by enzymes of the TCA cycle). In support of this possibility, a marked increase in transcripts for the citrate carrier was noted in LPS stimulated BMDCs. Furthermore, studies of macrophages demonstrate that LPS stimulation increased expression of the citrate carrier which increased cytosolic citrate levels which through the action of citrate lyase produces acetyl CoA and oxaloacetate. Oxoalocetate is converted to malate via malate dehydrogenase then pyruvate to generate NADPH required for ROS and NO production. Using gene silencing to ablate the citrate carrier gene in this system was therefore found to reduce NO and ROS production (Infantino et al., 2011). The concurrently generated aceyl-CoA can augment fatty acid synthesis (Everts et al., 2012; O'Neill and Pearce, 2016). Consequently, these changes to energy metabolism with accompanying biosynthetic reprogramming are likely to have a widespread influence on proximal and distal immune cell function including cytokine production and membrane remodeling. We suggest that these changes, similar to those seen following LPS stimulation (in the absence of infection) are host evolved mechanisms that occur during T. gondii infection are likely to provide host benefits in terms of shaping the immune response but can potentially aid parasite growth.

As the TCA cycle is linked to the ETC in mitochondria we examined the effect of T. gondii infection on transcripts for components of the ETC. We found that a number of these transcripts encoding complex I, II and III were down regulated in T. gondii infected BMDCs and LPS stimulated BMDCs. However, transcripts encoding complex IV and V subunits were increased in BMDCs infected T. gondii or stimulated with LPS. It has been observed in the literature that specific components of the ETC have additional roles other than generating energy including heme A biosynthesis, ion transport, calcium homeostasis, vesicle formation and lipid signaling (Antonicka et al., 2003; Sharma et al., 2019). Interestingly, there was an increase in gamma-glutamyl cysteine and cystine but not glutathione. These results suggest a response to oxidative stress (e.g., increased cysteine and glutathione synthesis to maintain intracellular glutathione levels. Additionally, to further determine the functional implications of these transcriptional changes we exploited the membrane potential-sensitive dye, TMRM. Our results indicate that mitochondria membrane potential is similarly down regulated in BMDCs infected with T. gondii or stimulated with LPS (Zhu et al., 2015; Van den Bossche et al., 2016). This is likely due to a reduced TCA cycle.

Changes to arginine metabolism in T. gondii infected BMDCs were also clearly evident in these studies described. Notably T. gondii infection favored the expression of arginase and production of L-ornithine, L-proline, L-1-pyrroline-3-hydroxy-5-carboxylate, and L-glutamate. Concomitant increases in the levels of transcript for srm which encodes spermidine synthesis was elevated in these infected cells suggesting further metabolism of ornthine to polyamines. In contrast LPS activation of BMDCs favored the expression of iNOS and the production of nitric oxide and citrulline. Opposing hypotheses can be formed as to whether these changes benefit the host or the parasite. Viewing the alteration in arginine metabolism as a host evolved strategy, arginine depletion has the potential to restrict T. gondii growth as it is an arginine auxotroph (Fox et al., 2004). Alternatively, directing arginine metabolism toward ornithine, proline and polyamines can also be viewed as a parasite evolved strategy subverting arginine metabolism toward arginase degradation reducing the ability of iNOS to produce nitric oxide which has been shown to have limit parasite multiplication. Additionally, diverting arginine toward ornithine and polyamines and ultimately to glutamate may provide T. gondii this resource for downstream use in the TCA cycle. In support of this studies have demonstrated that ROP16 (released from the rhoptry organelles during invasion) induces Arg expression in a STAT6 dependent manner (El Kasmi et al., 2008; Butcher et al., 2011; Marshall et al., 2011). In reality, as parasites and hosts evolve together both these hypotheses can coexist. Importantly, our previously published work demonstrates that T. gondii growth can be curtailed in vivo both through arginase dependent and iNOS dependent mechanisms (Woods et al., 2013).

Overall the studies described herein, demonstrate that the metabolism of BMDCs are profoundly affected by T. gondii infection in a manner similar to the Warburg effect. The multi-omics approach used here provides a wealth of metabolic and transcriptomic indicators of other pathways that are affected including arginine metabolism and provide insight in to the potential interactions of the host and parasite biochemistry. Understanding the interplay of these host and parasite interactions could provide insight into novel antimicrobial therapies including host directed interventions to limit parasite multiplication and survival.

#### DATA AVAILABILITY

The RNASeq datasets generated for this study can be found in Figshare, https://doi.org/10.6084/m9.figshare.9751904.v1.

#### ETHICS STATEMENT

All animal care and experimental procedures were conducted in accordance with relevant guidelines and regulations with the approval of the University of Strathclyde Animal Welfare and Ethical Review Body (AWERB), under UK Home Office regulations (Animals (Scientific Procedures) Act 1986, UK). Animals are housed according to or above the standard of the Home Office Code of Practice for the housing and care of animals bred, supplied or used for scientific purposes. The Appendix D Schedule 1 procedure dislocation of the neck was used to obtain tissue. This is followed by a confirmation method.

#### AUTHOR CONTRIBUTIONS

KH, OM, and CR conceived the study and participated in its design. KH with help from SW, co-cultured BMDCs with Toxoplasma gondii and coordinated all experiments described herein. SW performed the CBA assay. GW processed all the raw metabolomics and transcriptomics data-sets whilst KH analyzed and interpreted them. Technical help from SC was necessary to image the mitochondrial membrane potential of the T. gondii co-cultured BMDCs. KH performed statistical analysis and interpreted the results alongside CR. KH and CR wrote the manuscript. All authors read and approved the final manuscript.

#### REFERENCES


#### FUNDING

KH is supported by doctoral training partnerships (DTP) studentship from the Biotechnology and Biological Sciences Research Council (BBSRC), BB/J013854/1 BBSRC DPT Studentship (with Glasgow University).

#### SUPPLEMENTARY MATERIAL

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fcimb. 2019.00309/full#supplementary-material


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

Copyright © 2019 Hargrave, Woods, Millington, Chalmers, Westrop and Roberts. 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.

digital media

of impactful research

article's readership