Regulatory T cells shape the resident memory T cell response to virus infection in the tissues.

Abstract

Regulatory T cells (Tregs) are well known for their role in dampening the immune responses to self-antigens and thereby limiting autoimmunity. However, they must also permit immune responses to occur against foreign infectious agents. Using a mouse model of West Nile virus (WNV) infection, we examined the role of Tregs in the generation of effector and memory T cell responses in the secondary lymphoid organs (SLO) as well as the infected tissues. We found that Treg numbers and activation increase in both the SLO and CNS after infection. Using the Foxp3^DTR knock-in mice, we found that Treg-deficient mice had increased antigen-driven production of IFN-γ from both CD4⁺ and CD8⁺ T cells in both the spleen and CNS during the effector phase. In mice lacking Tregs, there were greater numbers of short-lived effector CD8⁺ T cells in the spleen during the peak of the immune response, but the memory CD8⁺ T cell response was impaired. Specifically, we demonstrate that Treg-dependent production of TGF-β results in increased expression of CD103 on CD8⁺ T cells, thereby allowing for a large pool of resident memory T cells to be maintained in the brain after infection.

INTRODUCTION

Regulatory T cells (Tregs) are well known for their suppressive properties, which can reduce immune responses to self-antigens and prevent autoimmunity (1, 2). Recent work has also highlighted the role of Tregs in the immune response to microbial infection (3). Several groups have reported that Tregs limit vigorous immune responses that would assist in pathogen clearance at the expense of damaging healthy tissue (4). This in some cases leads to a severely diminished effector T cell response unable to adequately clear the infection (5). Additionally, Tregs have been shown to facilitate early immune responses to viral infection by coordinating a timely trafficking of lymphocytes to the infection site in an HSV-2 model (6). Thus, because Tregs have demonstrated roles in the suppression of as well as the generation of anti-microbial immunity, we hypothesized that Tregs could have distinct roles in anti-viral immunity dependent on the time post-infection as well as the tissue microenvironment.

Thus, in this study, we utilized a well-established mouse model of West Nile virus (WNV) infection to investigate a role for Treg cells in T cell responses to neurotropic virus infection at various times post-infection as well as in various tissues. WNV is a single-stranded RNA virus that cycles between mosquitos and birds, with humans and other mammals serving as incidental hosts. Approximately 20% of infected individuals experience a limited febrile illness, with 1% developing a more severe neuroinvasive disease characterized by encephalitis and meningitis (7). The immune response to WNV is known to involve both innate and adaptive responses, including humoral and cellular components. Upon infection in the skin following injection or mosquito bite, WNV replicates and is able to infect dendritic cells (DCs), including Langerhans cells that can subsequently migrate to the draining lymph nodes (dLN) where they then initiate immune responses. DCs and other cells sense the presence of RNA virus infection through TLR expressed within the endosomal compartment, as well as ubiquitously expressed cytoplasmic RNA sensors such as retinoic-acid-inducible gene I (RIG-I) and melanoma-differentiation associated gene 5 (MDA-5) (8). One key immune mediator downstream of this virus sensing mechanism is type I IFN, an important anti-viral molecule capable of eliciting multiple anti-viral pathways. Both T and B lymphocytes are involved in protection against WNV, and in mouse studies it has been demonstrated that humoral immunity is involved in peripheral clearance of WNV, whereas T cells are critical for viral clearance within the CNS. Specifically, the induction of virus-specific IgM early after infection with WNV limits viremia and spread to the CNS, thus helping to protect against lethal infection (9, 10). CTLs are also known to mediate immunity to WNV infection, as adoptive transfer of WNV-specific CD8⁺ T cells results in a reduction of mortality and prolonged survival after WNV infection of recipient mice. CD8⁺ T cells were found to infiltrate the infected brain, suggesting that they could be involved in recovery from encephalitis (11). CD4⁺ T cell responses are also strongly induced, and are required for the maturation of IgG responses as well as sustaining CD8⁺ T-cell responses, both in the periphery and the CNS. Nevertheless, absence of CD4⁺ T cells did not cause a significant difference in viral titers in the periphery (12). WNV likely traffics from peripheral tissues to the CNS via axonal spread or by hematogenous route across the blood-brain barrier (13). The generation of immune responses within the CNS are critical to clear the virus but at the same time must be regulated such that damage to non-renewing populations of neurons is limited. After infection, CD8⁺ T cells migrate to the brain, and their presence correlates with viral clearance (14). Indeed, in the absence of CD8⁺ T cells, WNV persists in the brain of infected mice (15). Recruitment of T cells to the CNS is mediated by both CCL5 and CXCL10 (16–18). Although it has been postulated that CD8⁺ T cells may have a pathologic role in terms of damaging infected neurons in addition to their protective role during WNV infection, it is certainly clear that CD8⁺ T cells are required for clearance of WNV from the CNS.

Studies in mice have shown that WNV can persist in the periphery and CNS, in spite of the presence of virus-specific immune cells and antibodies, for at least 16 weeks post-infection (19, 20). It is hypothesized that the virus might persist in certain tissues such as the brain for longer periods, due to the slow turnover of neuronal tissue and the need to limit immunopathology within the brain (21). Given the persistence of WNV in the CNS, and the role of Tregs in balancing an adequate but not overly robust immune response, we sought to investigate how Tregs might modulate the effector and memory T cell response to virus infection in both the secondary lymphoid organs as well as the infected neuronal tissues. Further, as depletion of Tregs is known to result in increased WNV disease, weight loss, and lethality (22), we hypothesized that there would be dramatic effects on the T cell response to virus infection upon Treg depletion.

Our results show that Tregs limit antigen-driven proliferation of effector T cells in the CNS, as well as the production of inflammatory cytokines. In mice depleted of Tregs, there is an expansion of short-lived effector cells in the spleen at the peak of the effector phase, but the retention of antigen-specific CD8⁺ T cells in the brain is impaired. Specifically, we found that Treg-dependent production of TGF-β leads to increased expression of CD103 on CD8⁺ T cells, allowing a greater number of these resident memory T cells to remain in the brain after infection. In sum, our study demonstrates the two-sided nature of Tregs; they can suppress but also potentiate the immune response to microbes, depending on the time post-infection as well as the location within the host.

MATERIALS AND METHODS

Virus

West Nile virus TX-2002-HC (WN-TX) was kindly provided by Dr. Michael Gale, Jr. (University of Washington) and propagated as previously described (23, 24). Working stocks were generated from supernatants collected from infected Vero cell lines, and stored at −80°C.

Mice and infection

C57BL/6 mice, 6–8 weeks of age, were purchased from The Jackson Laboratory (Bar Harbor, ME) and housed in filtered cages. Foxp3^DTR (25) and Foxp3^GFP (26) mice (kindly provided by Dr. Alexander Rudensky, Memorial Sloan-Kettering Cancer Center) were bred onsite at the animal facility at Fred Hutchinson Cancer Research Center. Age-matched six to twelve week old mice were subcutaneously inoculated in the left rear footpad with 100 PFU WN-TX. To ablate regulatory T cells, Foxp3^DTR or Foxp3^GFP mice were treated intraperitoneally with 30 µg/kg body weight of diphtheria toxin (DT) the day prior to infection and 10 µg/kg DT the same day as infection. All animal experiments were approved by the FHCRC and University of Washington Institutional Animal Care and Use Committee. The Office of Laboratory Animal Welfare of the National Institutes of Health (NIH) has approved the FHCRC’s Animal Welfare Assurance (#A3226-01), as well as the University of Washington (#A3464-01), and this study was carried out in strict compliance with the Public Health Service (PHS) Policy on Humane Care and Use of Laboratory Animals.

Cell preparation

Following euthanasia, mice were perfused with 10 ml PBS to remove any residual intravascular leukocytes. Popliteal draining lymph nodes were collected and processed to obtain single cell suspensions. Spleens were homogenized, treated with ACK lysis buffer to remove red blood cells, washed, and resuspended in FACS buffer (1X PBS, 2% FBS). To obtain lymphocytes from the CNS, brains were harvested into RPMI and a suspension created through mechanical disruption. The suspension was added to hypertonic Percoll to create a 30% Percoll solution, vortexed, and centrifuged at 1250 rpm for 30 minutes at 4°C. After aspirating the supernatant, any remaining red blood cells in the cell pellet were ACK lysed, washed, passed through a 70 µm nylon mesh, and resuspended in FACS buffer. Cells were counted by hemacytometer using trypan blue exclusion.

Flow cytometry analysis

Following preparation of single cell suspensions, cells were plated at 1 × 10⁶ cells/well and stained for surface markers for 15 minutes on ice. For tetramer staining, cells were stained with the WNV NS4b-H2^d tetramer (generated by the Immune Monitoring Lab, Fred Hutchinson Cancer Research Center) for 30 minutes on ice. Cells were subsequently fixed, permeabilized (Foxp3 Fixation/Permeabilization Concentrate and Diluent, Ebioscience) and stained with intracellular antibodies for 30 minutes on ice. Cells from C57BL/6 mice were also stained with Foxp3-FITC at this time as described previously (6). Flow cytometry was performed on a BD LSRII machine using BD FACSDiva software. Analysis was performed using FlowJo software. The following directly conjugated antibodies were obtained from Ebioscience: CD4-PE (RM4-5) or PerCPCy5.5 (RM4-5), CD8-PECy5 (53-6.7), CTLA-4-PE (UC10-4B9), CD29-PE Cy7 (EbioHMb1-1), ICOS-PE (7E.17G9), CD62L-PerCPCy5.5 (MEL-14), CD127-Alexa 700 (A7R34), KLRG-1-PECy7 (2F1), IFN-γ PerCPCy5.5 (XMG1.2), TNF-α-APC (MP6-XT22), CD103-PE (2E7), and Foxp3-FITC (FJK-16a). The Foxp3 intracellular staining kit (Ebioscience) was used for fixation/permeabilization and all intracellular staining.

Intracellular cytokine staining

Splenocytes or CNS lymphocytes were resuspended at 1 × 10⁶ cells/well in RPMI 1640 supplemented with 10% HI-FBS, 2.5mM HEPES, 100 U/mL penicillin, 100 mg/mL streptomycin, 2 mM L-glutamine, and 1 mM sodium pyruvate to be analyzed for intracellular cytokine production. Lymphocytes were stimulated with either media, the WNV NS4b peptide (SSVWNATTAI), heat-inactivated WNV (MOI=5) or polyclonal stimulation using anti-CD3/CD28, and incubated for 5 hours at 37°C. Cells were washed and stained with fixable viability stain (Fixable Viability Dye eFluor 780; Ebioscience) for 30 minutes on ice, followed by surface staining for CD4-PE and CD8-PECy5 for 15 minutes on ice. The cells were then fixed, permeabilized, and stained with IFN-γ-PerCPCy5.5 and TNF-α-APC according to the manufacturer’s protocol (Ebioscience). After staining, the cells were analyzed as described above.

RNA Extraction

Five brains per group were collected after whole body perfusion with 10mL cold PBS, snipped into pieces ~1mm in thickness and immediately placed into pre-weighed tubes containing 7 mL RNAlater Stabilization Reagent (Qiagen, Hilden, Germany). Samples were reweighed to determine weight of brain alone and placed in −80°C freezer. Brains were thawed at 4°C, homogenized using a handheld homogenizer and total RNA extracted following protocol instructions included with the RNeasy Lipid Tissue Midi Kit (Qiagen, Hilden, Germany). RNA was eluted into PCR-grade nuclease free water and RNA concentration measured using a Nanodrop 2000 UV-Vis spectrophotometer (Thermo Scientific).

qRT-PCR for WNV RNA

Competent E. coli cells transformed with kanamycin resistant plasmids containing the WNV PCR target region were kindly provided by Dr. Michael Gale, Jr of the University of Washington. Cells were grown in kanamycin containing media and plasmids isolated using the Qiagen Plasmid Mini kit (Qiagen, Hilden, Germany). The plasmid concentration was determined using a Nanodrop 2000 UV-Vis spectrophotometer (ThermoScientific) and 10 fold standard dilution series was generated spanning a concentration of 1e10 to 1e2 plasmids/5ul.

WNV primers and probe were used as previously described (27). The fluorogenic probe was synthesized with a 5’ reporter dye 6-carboxyfluorescien (6-FAM) and a 3’ quencher dye 6-carboxytetramethylrhodamine (5’-TAMRA). Primers and probe were generated as custom assays from Integrated DNA Technologies (IDTDNA, Coralville, Iowa). qRT-PCR assays were performed using the SuperScript® III Platinum® One-Step Quantitative RT-PCR System (Life Technologies, Grand Island, NY). Reactions were carried out in a total volume of 20 ul, containing 5 ul of template RNA, 1X reaction mix, 500nM final concentration for forward and reverse primers and 250nM final concentration for probe, 0.4ul ROX dye, 0.4 ul RT/Taq enzyme mix and brought up with nuclease free water. After adding the reaction mixture and template RNA to MicroAmp® Fast Optical 96-Well Reaction Plates (Applied Biosystems, Inc., Foster City CA), reverse-transcription and amplification were carried out on the ABI 7900HT Fast Real-Time PCR System (Applied Biosystems, Inc., Foster City CA) in standard mode. Cycling conditions were as follows: 50°C for 15 minutes hold (cDNA synthesis step), 95°C for 2 minutes hold, 40 cycles of 95°C for 15 seconds followed by 60°C, 1 minute.

TGF-β ELISA

Whole brain homogenates, or purified samples of Tregs or conventional CD4 T cells, were prepared and assayed for TGFβ by ELISA. Cell populations were isolated using the CD4+ CD25+ Regulatory T cell isolation kit (Miltenyi Biotec, Auburn, CA). Non-CD4+ cells were indirectly magnetically labeled and depleted over a MACS® Column, and the flow-through fraction of pre-enriched CD4+ T cells was then labeled with CD25 MicroBeads for subsequent positive selection of CD4+CD25+ regulatory T cells. Brain homogenates and purified cell populations were prepared and lysed using NP-40 lysis buffer containing complete protease inhibitor cocktail tablets (Roche, Indianapolis, IN) for protein quantification. Lysates were centrifuged and supernatants stored at −80°C until the ELISA was performed according to the manufacturer’s protocol (Mouse TGF-β ELISA kit; Ebioscience). Briefly, samples were acid-treated and then neutralized to activate the latent TGF-β1 to the immunoreactive form. Standards and samples were incubated for two hours at room temperature, followed by incubation with biotin and streptavidin-HRP. TMB substrate was used and the plate read by spectrophotometer at 450nm.

Statistical analysis

When comparing groups, two-tailed unpaired Student’s t tests were conducted, with p-values <0.05 considered significant. Error bars show +/− SEM.

RESULTS

Regulatory T cell numbers increase and become activated in the dLN and CNS after WNV infection

To determine if Tregs respond to West Nile virus infection, we first examined the kinetics and activation status of the Treg population following infection in both the footpad-draining popliteal lymph node and the brain. WNV infection was performed by subcutaneous injection in the rear footpad to mimic a mosquito bite, as WNV is a vector-borne viral infection. Importantly, Foxp3⁺ Tregs were detectable in the brain post-infection, though they were largely absent in the naïve brain (Fig. 1a), thereby demonstrating that Tregs are present and could play an important role in the immune response to WNV within the infected tissues. In the dLN, Tregs expanded rapidly, beginning by day 4 post-infection and peaking around day 15, with total Treg numbers then decreasing through day 20. In the brain, Treg expansion occurred more gradually, with high numbers maintained out to a memory time-point, day 60 post-infection (Fig. 1b). Treg expansion kinetics in the dLN and brain were similar to those observed for the effector CD4⁺ T cell population (data not shown). Likewise, activated effector T cell and Treg cell populations increased after infection in both the dLN and brain, as indicated by increased expression of ICOS (Fig. 1c and data not shown). The frequency of CTLA-4⁺ and CXCR3⁺ Treg cells was also increased in both the dLN and brain (data not shown).

Figure 1. Regulatory T cell numbers increase and become activated in the dLN and CNS after WNV infection.

Foxp3^gfp mice were inoculated s.c. in the footpad with 100 pfu West Nile virus and sacrificed at the indicated time points after infection. Data shown are representative of four to five mice per time point in three independent experiments. Error bars represent +/− SEM. A) Representative flow cytometric staining of CNS derived lymphocytes for CD4+ T cells and CD4+ Foxp3+ T cells at given time points after infection. B) Number of total lymphocytes (top panel) and Tregs (bottom panel) in the dLN and brain from 0 to 60 days after WNV infection. C) Number of Treg+ ICOS+ cells in the dLN (left) or brain (right) at given time points after infection.

Tregs limit CD4⁺ and CD8⁺ T cell activation and infection-induced cytokine production

Since we observed Treg expansion and activation during WNV infection, we next examined the immune response in the systemic absence of Tregs using a mouse model of Treg ablation (25). Foxp3^GFP (26) and Foxp3^DTR mice were treated with diphtheria toxin (DT) the day prior to and on the day of infection. The Treg population was successfully ablated in the Foxp3^DTR mice through early time points post-infection in both the secondary lymphoid organs (SLO) as well as the brain, although importantly, Treg frequency rebounded by day 14 post-infection in the SLO and by day 20 post-infection in the brain (Fig. 2a). In Treg-deficient, WNV-infected mice, we observed increased numbers of activated CD4⁺ and CD8⁺ T cells in the dLN and brain (Fig. 2b) compared to Treg-sufficient mice. These differences were most striking early after infection, at day 4 post-infection in the dLN and days 8 and 11 post-infection in the brain, likely due to restoration of Treg homeostasis as DT is cleared from the host and Treg numbers return to steady state (Fig. 2a).

Figure 2. Treg depletion results in increased CD4 and CD8 T cell activation and infection-induced cytokine production.

Foxp3^DTR mice (Treg−) or Foxp3^gfp littermate controls (Treg+) were treated with 30 µg/kg DT one day prior to infection, and 10 µg/kg DT on the day of infection. WNV infection, or mock infection with PBS, was performed as described above. Data shown are from three to five mice per group for each time point, and representative of three independent experiments. Error bars represent SEM. A) Frequency of Foxp3+ Tregs following Treg depletion in splenocytes or CNS-derived lymphocytes in DT-treated Foxp3^gfp mice (Treg+) or Foxp3^DTR mice (Treg−) at the indicated days following WNV infection. B) Number of CD4+ ICOS+ or CD8+ ICOS+ cells present in the dLN at 4 or 8 days post-infection, and brain at 8, 11, and 14 post-infection, in Treg sufficient or deficient mice. C) Mean frequency of CD4+ or CD8+ T cells expressing IFN-γ following a 5 hr stimulation with heat-inactivated WNV or NS4b peptide. D) Frequency of single or dual cytokine producers (IFN-γ and TNF-α) in CD4+ or CD8+ T cells in the spleen and brain at d12 post-infection.

Because increased activation of T cells could be due merely to activation of bulk, non-WNV specific cells in the absence of tight regulation by Tregs, we next determined the WNV-specific functionality of activated effector cells in the Treg-deficient mice in the context of infection. Twelve days after infection, we performed intracellular cytokine staining to evaluate the number of CD4⁺ and CD8⁺ T cells that were producing IFN-γ, TNF-α, or both cytokines. The frequency of both CD4⁺ and CD8⁺ T cells producing IFN-γ directly in response to virus was increased in both the spleen and the brain of Treg-deficient mice, reaching statistical significance in the peripheral site of virus infection, the brain (Fig. 2c). Additionally, of the WNV-specific CD4⁺ and CD8⁺ T cells in both the SLO and brain, Treg-ablated mice had an increased proportion of cells that were polyfunctional in their ability to produce both IFN-γ and TNF-α (Fig. 2d), likely leading to a more pro-inflammatory environment. In contrast, Treg-sufficient mice had more CD8 T cells producing IFNγ or TNFα alone, as has been demonstrated by others examining CD8 T cell functionality after WNV infection (Fig. 2D)(28–30). Finally, we quantified the levels of WNV present in the brain and spleen in Treg-sufficient or –deficient mice. At day eight post-infection, there was no difference in WNV RNA in the spleen regardless of the presence or absence of Tregs, and further, there was no statistically significant difference at days eight or eleven post-infection in the brain (Fig. 3), suggesting that the increased pro-inflammatory cytokine production observed in the brain upon Treg depletion is not due simply to enhanced virus replication upon Treg ablation.

Figure 3. Quantification of WNV viral copy number.

Foxp3^DTR mice (Treg−) or Foxp3^gfp controls (Treg+) were treated with 30 µg/kg DT one day prior to infection, and 10 µg/kg DT on the day of infection. WNV infection was performed as described above. On days 8 and 11 post-infection, WNV copy numbers in the brains and spleens of infected Treg+ or Treg− mice were measured by qRT-PCR and expressed as WNV amplicons per 250 ng of total RNA. Data shown are from four mice per group and representative of three independent experiments. Error bars represent SEM.

Tregs modulate the fate of WNV-specific effector CD8 T cells in distinct tissue compartments

Because we observed differences in effector T cell activation and cytokine production during WNV infection in the absence of Tregs, we next considered how Tregs could affect the fate of WNV-specific T cells. To evaluate the effect of Treg ablation on the effector-to-memory cell transition in the context of WNV infection, we examined the frequency of WNV-specific CD8⁺ T cells expressing the cell surface markers CD127 or KLRG-1 in Treg+ or Treg-infected mice (31). Similar to results previously published using a vaccinia virus vector (32), ablation of Tregs resulted in increased numbers of short-lived effector cells (SLEC), defined as CD127⁻KLRG1⁺tetramer⁺ CD8+ T cells, in the spleen 8 and 11 days-post WNV-infection (Fig. 4a–b). We did not, however, observe a difference in the number of SLECs present in the brain dependent on the presence of Tregs (Fig. 4a–b). While we did observe slightly elevated numbers of memory precursor effector cells (MPECs), defined as tetramer⁺ CD8⁺ T cells that are CD127⁺ KLRG-1⁻, in the spleen of Treg-sufficient mice 8 days post-infection, this difference was absent by day 11 post-infection, and indeed in the brain the numbers of MPECs were equal regardless of whether or not the mice had Tregs (Fig. 4a–b). Thus, the data suggest that Tregs can modulate the effector to memory T cell transition following virus infection in distinct ways dependent on the tissue microenvironment.

Figure 4. Tregs control the generation of virus-specific short-lived effector cells in lymphoid tissues and memory precursors in the infected tissues, and shape the generation of immunological memory to WNV.

Foxp3^DTR mice or Foxp3^GFP littermate controls were treated with 30 µg/kg DT one day prior to WNV infection, and 10 µg/kg DT on the day of infection. Data shown are from five mice per group, and representative of three to four independent experiments. Error bars represent +/− SEM. A) Eight or eleven days after WNV infection, mice were sacrificed and lymphocytes from spleen and brain were prepared for flow cytometric analysis. Cells were gated on CD8 and MHC class I tetramer, and further analyzed for markers of short-lived effector cells (KLRG-1+CD127−) or memory precursor cells (KLR-G-1-CD127+). Representative flow plots are shown in B. C) The number of CD8+ tetramer+ cells in the spleen and brain at day 12, 20, and day 60 post-infection are shown.

Tregs shape the generation of immunological memory to WNV

We next examined the kinetics of the WNV-specific CD8+ T cell response out to day 60 post-infection in both the spleen and brain in order to determine how the memory T cell response is modulated by Tregs. In the spleen, we observed an expected peak in WNV-specific T cell numbers at day 12 post-infection that declined by day 20 as the T cell population contracted to a stable memory pool at day 60 post-infection. The pattern of CD8 T cell dynamics was similar regardless of Tregs, but Treg ablation did result in an increase in the total number of antigen-specific CD8⁺ T cells in the spleen at the peak of the effector phase (Fig. 4c). Strikingly, in the brain we observed that in Treg+ animals there was a steady increase in WNV-specific CD8⁺ T cells following infection out to a memory timepoint. In the absence of Tregs, there were similar numbers of antigen-specific CD8+ T cells at the effector phase of the immune response (days 12 and 20 post-infection) as compared to Treg-replete mice. However, in mice depleted of Tregs at the onset of virus infection, there were significantly fewer memory CD8+ T cells present in the brain at day 60 post-infection, suggesting that Tregs are vital for the creation of memory T cells in the brain post-infection (Fig. 4c).

By virtue of time and location, it appears that the CD8+ memory T cells present in the brain at day 60 post-infection are resident memory T cells (T_RM). To further characterize the phenotype of these putative T_RM, we examined expression of CD103 on WNV-specific CD8+ T cells in the brain at day 60 post-infection, as expression of this integrin αE is known to be expressed by mucosal memory T cells and is suggested to be important for their retention in tissues such as the intestinal epithelium, skin, and central nervous tissues (33). Furthermore, it has been previously shown that CD103 expression not only phenotypically characterizes the majority of T_RM in the central nervous system, but additionally is important for their generation and accumulation (34). There were significantly fewer CD103+ tetramer+ CD8+ T cells in the brain at day 60 in the absence of Tregs as compared to Treg-replete mice (p=0.0284, Fig. 5), suggesting both that Tregs are vital in the establishment of a T_RM population in the brain following WNV infection as well as hinting at a mechanism by which Tregs could control recruitment and retention of these cells. Of note, WNV-infected Treg-deficient mice do have a sizable population of tetramer-negative CD8 T cells in the brain that express CD103, but as this population is similar in frequency to that observed in the brains of naïve mice, we hypothesize that these are not WNV-specific T cells but rather cells that are resident in the brain prior to infection with WNV (Fig. 5a).

Figure 5. Tregs assist in the retention of resident memory T cells in the CNS.

Treg-sufficient and -deficient mice were infected as previously described, and tissues were harvested at day 60 and prepared for flow cytometry. A) CNS-derived lymphocytes were gated on CD8+ T cells. Flow cytometric plots show representative staining of the frequencies of virus-specific resident memory T cells (tetramer+ CD103+) cells in each group of mice. B) Mean numbers of resident memory T cells in the CNS. Data shown are from five mice per group for each time point, and representative of three independent experiments, with error bars representing +/− SEM.

As the expression of CD103 is known to be positively regulated by TGF-β (33, 35), we next examined the levels of TGF-β protein in the brain at various times post-WNV infection. TGF-β levels were significantly lower in Treg-deficient mice 7 days post-infection compared to Treg+ mice (p=0.0423). Additionally, there was a trend for decreased TGF-β levels in Treg-deficient mice at days 9 and 12 post-WNV infection (Fig. 6a). Further, Tregs are a key cellular source of TGFβ in the brain after WNV infection, as they contain significantly higher levels of TGFβ protein as compared to conventional CD4 T cells (p=0.0003; Fig. 6b). Therefore, we conclude that Tregs directly modulate the T_RM response to WNV in the brain via production of TGF-β.

Figure 6. Treg-dependent production of TGFβ correlates with CD103 expression on T_RM and subsequent retention in the CNS.

A) Treg-sufficient and -deficient mice were infected with 100pfu WNV and brains were harvested at days 7, 9, and 12 post-infection. CNS cell lysates were prepared and TGF-β levels measured by ELISA. Each sample was analyzed in duplicate, with five to seven mice per group for each time point. Error bars represent +/− SEM. B) Foxp3^GFP mice were infected with 100 pfu WNV and brains were harvested 12 days post-infection. Single cell suspensions of brains were used to prepare purified samples of Tregs or conventional CD4 T cells by MACS bead isolation, and cell lysates were subsequently prepared and assayed for TGFβ by ELISA.

DISCUSSION

The generation of a robust anti-viral immune response involves a rapid and dramatic clonal expansion of antigen-specific effector CD4+ and CD8+ T cells, which then contract to form a small but potent pool of stable memory T cells. In non-SLO tissues, it has been shown that effector memory T cells are particularly poised for instant response to infection, as memory CD8+ T cells from non-lymphoid tissues display effector levels of lytic activity directly ex vivo (36). As it is known that circulating memory T cells do not access all tissues during immune surveillance, particularly the brain and intestinal lamina propria (37), the generation of T_RM populations may be of particular importance in these tissues. T_RM are characterized as being tissue-resident and self-renewing, as well as extremely protective against repeat infections. An elegant study of T_RM in the brain has demonstrated that these cells express CD103, shown to be not only a phenotypic marker but also critical for their generation and accumulation. Further, Wakim et al demonstrated that brain T_RM can persist in the absence of persistent antigenic stimulation, though they do require local DC-dependent antigen presentation to upregulate CD103 on the incoming T cells (34).

During infection with West Nile virus, a balance must exist between a robust immune response that clears the pathogen, yet limits destruction of non-renewing populations of neurons in the brain. Given the role of Tregs in regulating and maintaining a delicate balance in immune responses to viral infections, we investigated the role of Tregs in the generation of effector and memory T cells both systemically and in the infected central nervous system. Using a mouse model in which Tregs could be ablated prior to infection, we showed that Treg-deficient mice had greater numbers of short-lived effector CD8+ T cells in the spleen during the peak of the immune response, but the memory CD8+ T cell response in the brain was impaired. Specifically, we demonstrate that Treg production of TGF-β results in increased expression of CD103 on CD8⁺ T cells, thereby allowing for a large pool of resident memory T cells to be maintained in the brain after infection. In our model of WNV infection in mice, as opposed to the work done by Wakim et al using intranasal infection with VSV (34), it is possible that there is prolonged antigen presentation in the brain, even after viral clearance, as Appler et al have shown that WNV RNA persists in the CNS up to 3 months post infection (19). This prolonged antigen presentation could drive CD103 expression on T_RM, perhaps through induction of Treg-dependent TGF-β production.

In the absence of Treg regulation during the immune response to viral infection, it is possible that bulk expansion of non-specific CD4 and CD8 effector T cells contribute to the net increase in effector T cells observed in the CNS. However, we showed that Treg-deficient mice had increases in CD4 and CD8 cells specifically producing cytokine in response to WNV-stimulation. While these Treg-deficient mice had increased numbers of effector T cells and more robust cytokine production during the peak of infection, the number of memory T cells in the brain were significantly decreased compared to WT-infected mice at d60 post-infection. This demonstrates that Tregs are vital for the creation of memory T cells in the brain, and that certain mechanisms and instructional cues must be in place to allow T cells to remain in such a tightly regulated area. Alternatively, it is possible that elimination of Tregs could impact the trafficking of CD8 T cells into the brain, which could potentially explain the resultant differences that we observe in memory CD8 T cell numbers in the brain. While this is possible, we favor the hypothesis that it is a lack of retention of CD8 T cells in the brain when Tregs are absent at the time of infection that results in the deficit of T_RM since we see a decreased expression of CD103 on WNV-specific CD8 T cells in the brain upon Treg ablation (Fig. 5). Additionally, we do not observe a lack of WNV-specific CD8 T cells present in the brain at early time points (Fig. 4a&c), suggesting that these cells are able to traffick to the CNS but rather fail to persist.

Our finding that Tregs control the generation of SLEC in the context of virus infection is similar to findings by Kastenmuller et al, who demonstrated that this was mechanistically due to a limited availability of IL-2, required for optimal SLEC generation (32). However, our results differ in that we uncovered a role for Tregs in generation of immunological memory, likely due to the fact that we used a neurotropic virus capable of inducing CD8+ T cell memory responses in the brain, pointing to the different roles that Tregs can play in immunity dependent on tissue location. Indeed, we speculate that the presence of Tregs during neurotropic virus infection can restrain CD4+ and CD8+ T cell production of pro-inflammatory and anti-viral cytokines in the brain, thereby resulting in the incomplete viral clearance demonstrated by others (19). This continued antigenic presence could in turn be important for sustained T_RM populations needed to protect the host from recurrent infections threatening this vital tissue. Recently, Casey et al demonstrated that TGF-β-driven CD103 expression was required for antigen-specific T_RM maintenance within the intestinal epithelium, and furthermore, that intestinal T_RM cells were not dependent on prolonged cognate antigen stimulation (38). Our data similarly indicate that TGF-β-driven CD103 expression is likely required to allow T_RM to persist in the brain, and further, our work suggests a mechanism by which Tregs control and maintain the tissue-resident memory population.

In sum, our results suggest that Tregs are necessary to generate a pool of resident memory T cells, again highlighting the important role these regulatory cells play in potentiating anti-viral immunity. While a significant population of T cells persisting in the brain could present a dangerous situation to the host, it is necessary to defend such critical tissues from neurotropic infection. While we demonstrate that Tregs can indeed negatively regulate T cell cytokine function in response to virus infection in the brain, they also play a pivotal role in establishing an appropriate T_RM population with which to protect the host from re-infection. Finally, the differing roles of Tregs in the SLO compared to the neuronal tissues suggest that the tissue microenvironment provides vital cues that play a significant role in defining T cell fates and balancing the host immune response to microbial infection.

Abbreviations used in this paper

Treg

regulatory T cell

WNV

West Nile virus

SLO

secondary lymphoid organs

DC

dendritic cell

dLN

draining lymph node

DT

diphtheria toxin

SLEC

short lived effector cell

MPEC

memory precursor effector cell

T_RM

resident memory T cell