IL-10 and ICOS Differentially Regulate T Cell Responses in the Brain during Chronic Toxoplasma gondii Infection

Immune responses have intricately evolved to protect hosts from a wide range of potentially harmful pathogens (1, 2), yet these same inflammatory responses can often cause host damage themselves. The importance of a balanced immune response is apparent in models of infection, in which inflammation is required for pathogen control, yet amplified immune responses observed after depletion of regulatory T cells (T_regs) or immunosuppressive cytokines often lead to exacerbated tissue damage and increased mortality (3–8). One such immunosuppressive cytokine, IL-10, has been broadly studied in the context of both tissue homeostasis and during infection and has been shown to play a key role in suppressing many aspects of an immune response. Production of IL-10 during immune responses to infection has been attributed to a wide variety of cell types, including T cells, dendritic cells (DCs), macrophages, NK cells, and B cells (9). IL-10 also acts on a wide range of cell types, with one of its main roles being the downregulation of MHC and costimulatory molecules on APCs, thereby preventing their full activation capacity and limiting T cell responses (10–12). IL-10 also has direct effects on T cells, limiting IFN-γ and IL-2 production as well as T cell proliferation in vitro (13, 14).

Infection with the eukaryotic parasite Toxoplasma gondii leads to widespread activation of the immune system and systemic inflammation that is required for host survival (15). The generation of a parasite-specific adaptive immune response clears the parasite from most peripheral tissues; however, the parasite is able to encyst in the CNS and establish a chronic infection (16, 17). This chronic infection requires ongoing activation and infiltration of highly polarized Th1 cells into the brain to prevent extensive parasite replication and fatal disease (18, 19). However, like in other models of infection, regulation of this immune response is also required to promote host survival. In particular, IL-10 production is required for host survival during acute infection. IL-10 knockout mice succumb to CD4⁺ T cell–mediated immunopathology and excessive inflammatory cytokine production in the periphery early in the course of infection (3, 20). Similarly, continued IL-10 production in the chronic phase of infection is also necessary for host survival. IL-10 knockout mice given the antibiotic sulfadiazine early in the infection to limit parasite replication survive the acute phase of infection but later present with similar CD4⁺ T cell–mediated fatal immunopathology in the brain (21). Despite demonstrating the requirement for IL-10 signaling over the course of T. gondii infection, previous studies have not addressed what additional signals promote immune regulation in the context of chronic neuroinflammation.

ICOS is a costimulatory molecule expressed on activated T cells (22, 23). ICOS signaling is important in a wide variety of immune responses, including the development of T follicular helper cells and the formation of germinal centers, as well as effector T cell inflammatory cytokine production (23–26). The primary function attributed to ICOS is the amplification of effector T cell responses by serving as a costimulatory molecule similar to its family member CD28 (27). More recently, ICOS has been shown to also promote immune regulation through potent induction of IL-10 both in vitro and in mouse models of acute inflammation (28–30). In addition to promoting production of IL-10, ICOS is also important for maintaining effector T_reg populations. During homeostasis, blockade of ICOS signaling results in a loss of CD44^hiCD62L^lo effector T_regs in the spleen (31, 32). In mouse models of diabetes and helminth infection, a similar loss of T_regs is seen with a lack of ICOS signaling, in addition to decreased IL-10 production (33, 34). During acute T. gondii infection, ICOS signaling has been reported to amplify T cell inflammatory responses by promoting increased IFN-γ production early in the infection (35, 36). ICOS appears to play a redundant role with CD28 in this setting as mice lacking ICOS are only more susceptible to infection on a CD28^−/− background (35). Together, these reports highlight the context-dependent role of ICOS signaling, contingent on variables ranging from the type of inflammatory environment to the stage of infection (22–36).

The role of ICOS, and its relationship to IL-10–mediated regulation of immune responses, in the context of a chronic neuroinflammatory response to a pathogen is not well understood. In this study, we first characterized what role IL-10 plays in promoting regulation of immune responses in the chronic stage of infection by blocking signaling through the IL-10R. This IL-10R blockade during chronic infection led to an expansion of CD4⁺ effector T cells correlating with increased expression of CD80 on APCs, along with widespread immunopathology. Based on previous reports implicating a role for ICOS in stimulating IL-10 production, we then addressed the question of whether ICOS signaling can promote suppression of chronic T cell responses in the CNS through induction of IL-10. Surprisingly, we found that blockade of ICOS signaling during chronic infection with T. gondii did not lead to decreased IL-10 production from either regulatory or effector T cells, nor did it lead to impaired inflammatory cytokine production. In fact, despite maintaining IL-10 levels in the brain, blockade of ICOS ligand (ICOSL) still resulted in a loss of T cell regulation, with 2–3-fold more effector T cells found in the inflamed brain. Interestingly, this increase in effector T cells occurred without a loss of T_regs in the brain and did not affect parasite burdens. We found this increase in T cell number in the brain correlated with increased levels of CD25 and p-STAT5 expression in effector T cells following ICOSL blockade, suggesting increased responses to IL-2. Along these lines, ICOSL blockade increased T cell proliferation and expression of the survival factor Bcl-2 among the effector CD4⁺ and CD8⁺ T cell populations in the brain, and we also observed decreased Annexin V–positive T cells in the brain following anti-ICOSL treatment. Interestingly, IL-10R blockade did not result in the same increases in IL-2–associated signaling molecules CD25 and Bcl-2; rather, IL-10R blockade increased the activation state of APCs. Taken together, our results suggest that ICOS signaling on T cells can suppress STAT5-induced survival signals, providing a mechanism of local suppression in the context of chronic inflammation in the brain that is distinct from IL-10–mediated regulation.

C57BL/6 and B6.129S6-Il10^tm1Flv/J (Tiger) mice were purchased from The Jackson Laboratory. All animals were kept in University of Virginia specific pathogen-free facilities and were age- and sex-matched for all experiments. All experimental procedures followed the regulations of the Institutional Animal Care and Use Committee at the University of Virginia. Infections used the avirulent type II T. gondii strain Me49, which were maintained in chronically infected Swiss Webster mice (Charles River Laboratories) and passaged through CBA/J mice (The Jackson Laboratory) before experimental infections. For experimental infections, the brains of chronically infected (4–8 wk) CBA/J mice were homogenized to isolate tissue cysts. Experimental mice were then injected i.p. with 10–20 Me49 cysts.

For IL-10R blockade studies, chronically infected mice (28–35 d postinfection) were treated i.p. with 200 μg of an mAb blocking the IL-10R (BioXCell) or a control rat IgG (polyclonal) Ab (BioXCell). Treatments were given every 3 d, and mice were euthanized when neurologic symptoms developed between 7 and 10 d post–Ab treatment. For ICOSL blockade studies, chronically infected mice (28–35 d postinfection) were treated i.p. with 150 μg of an anti-ICOSL (CD275)–blocking Ab (BioXCell) or a control rat IgG (polyclonal) Ab (BioXCell). Ab treatments were given every 3 d for 14 d, totaling five treatments.

Mice were sacrificed and perfused with 40 ml 1× PBS, and perfused brains, spleens, and lymph nodes (pooled deep and superficial cervical) were put into cold complete RPMI media (cRPMI) (10% FBS, 1% nonessential amino acid, 1% penicillin/streptomycin, 1% sodium pyruvate, 0.1% 2-ME). Brains were then minced with a razor blade and enzymatically digested with 0.227 mg/ml collagenase/dispase and 50 U/ml DNase (Roche) for 1 h at 37°C. After enzyme digestion, brain homogenate was passed through a 70-μm filter (Corning). To remove myelin, filtered brain homogenate was resuspended with 20 ml 40% Percoll and spun for 25 min at 650 × g. Myelin was then aspirated, the cell pellet was washed with cRPMI and then resuspended, and the cells were counted.

Spleens and lymph nodes were homogenized and passed through a 40-μm filter (Corning) and pelleted. Lymph node cells were then resuspended in cRPMI and counted. Spleen cells were resuspended with 2 ml RBC lysis buffer (0.16 M NH₄Cl). Following RBC lysis, spleen cells were washed with media and then resuspended with cRPMI for counting.

For experiments in which peripheral blood was taken, mice were sacrificed, and the right atrium was cut in preparation for perfusion. Before perfusion, 300 μl of blood was collected from the chest cavity. For isolation of circulating leukocytes, collected blood was put in 1 ml heparin (100 USP/ml) to prevent clotting. Samples were then pelleted and resuspended in 2 ml RBC lysis buffer for 2 min. Samples were washed once with cRPMI, and a second RBC lysis step was performed. Finally, blood cells were resuspended in cRPMI for staining and counting. For serum isolation, blood was allowed to clot overnight at 4°C. Samples were then spun at 14,000 rpm × g for 10 min to separate clotted blood from serum. After spinning, serum (supernatant) was transferred to a clean tube and stored at −80°C.

ELISAs for parasite-specific IgG were performed as previously described (37). Briefly, Immunolon 4HBX ELISA plates (Thermo Fisher Scientific) were coated with 5 μg soluble Toxoplasma Ag diluted in 1× PBS overnight at 4°C. After Ag coating, plates were washed in 1× PBS with 0.1% Triton and 0.05% Tween and then blocked with 10% FBS for 2 h at room temperature. After washing, serial dilutions of collected serum were added to plate wells overnight at 4°C. After incubation with serum samples, plates were washed, and wells were incubated with goat anti-mouse IgG and human ads-HRP (SouthernBiotech) for 1 h at room temperature. ABTS peroxidase substrate solution (KPL) was then added to wells, and immediately after a color change, plates were read on an Epoch BioTek plate reader using Gen5 2.00 software.

Single-cell suspensions from collected tissues from C57BL/6 mice were plated in a 96-well plate. Cells were initially incubated with 50 μl Fc block (1 μg/ml 2.4G2 Ab [BioXCell], 0.1% rat γ globulin [Jackson ImmunoResearch]) for 10 min at room temperature. Cells were then surface stained for CD3 (145-2C11), CD19 (eBio1D3), NK1.1 (PK136), ICOS (7E.17G9), ICOSL (HK5.3), MHC class II (MHCII) (M5/114.15.2), CD25 (PC61), CD8 (S3-6.7), CD11c (N418), CD4 (GK1.5), CD80 (16-10A1), CD86 (GL1), CD45 (30-F11), CD11b (M1/70), Ly6G (1A8), and a LIVE/DEAD stain for 30 min at 4°C. Parasite-specific cells were identified using a PE-conjugated MHCII tetramer (AVEIHRPVPGTAPPS) (National Institutes of Health Tetramer Facility). After surface staining, cells were washed with FACS buffer (1% PBS, 0.2% BSA, and 2 mM EDTA) and fixed at 4°C overnight with a fixation/permeabilization kit (eBioscience) or 2% paraformaldehyde (PFA). Following overnight fixation, cells were permeabilized and stained for intracellular markers Bcl-2 (3F11), Ki67 (SolA15), and Foxp3 (FJK-16S) for 30 min at 4°C. Cells were then washed, resuspended in FACS buffer, and run on a Gallios flow cytometer (Beckman Coulter). For Annexin V staining, surface staining of cells was performed as above, with the addition of a nuclear stain (Hoechst 33342; Life Technologies). Following surface staining, cells were washed with Annexin V–binding buffer (BD Biosciences) and incubated with Annexin V-FitC (BD Biosciences) diluted in Annexin V–binding buffer for 15 min at room temperature. Cells were then immediately run on a Gallios flow cytometer. All analysis was done using FlowJo software v.10.

Single-cell suspensions from C57BL/6 mice were plated into a 96-well plate. For T cell cytokine restimulation, cells were incubated at 37°C with 20 ng/ml PMA and 1 μg/ml ionomycin (Sigma-Aldrich) in the presence of brefeldin A (Alfa Aesar). After incubation for 5 h, cell suspensions were washed, surface stained, and fixed. Cells were then incubated with an Ab against IFN-γ (XMG1.2) for 30 min at 4°C. To determine IL-10 production, Tiger mice expressing enhanced GFP (eGFP) under the IL-10 promoter were used. Cells from these mice were stimulated with PMA/ionomycin and stained for surface molecules as described above and then lightly fixed with 2% PFA for 30 min at room temperature. Cells were then permeabilized with permeabilization buffer (eBioscience) and incubated with a biotin-conjugated anti-GFP Ab (BD Biosciences) for 30 min at 4°C. Cells were then washed and incubated with a PE-conjugated streptavidin secondary Ab (eBioscience) for 30 min at room temperature. Cells were then fixed with a fixation/permeabilization kit (eBioscience) overnight at 4°C before other intracellular staining was performed. For myeloid cell cytokine restimulation, single-cell suspensions were plated into a 96-well plate and incubated with brefeldin A for 5 h at 37°C before surface and intracellular staining for IL-12 (C17.8) (eBioscience).

Approximately 100 mg brain tissue was put into 1 ml TRIzol (Ambion) in bead beating tubes (Sarstedt) containing 1 mm zirconia/silica beads (BioSpec). Using a Mini-bead beater (BioSpec), tissue was homogenized for 30 s, and RNA extraction was completed using the manufacturer’s instructions (Ambion). For cDNA synthesis, a High Capacity Reverse Transcription Kit (Applied Biosystems) was used. Quantitative RT-PCR (qRT-PCR) was set up using a 2× Taq-based Master Mix (Bioline) and TaqMan gene expression assays (Applied Biosystems) and reactions were run on a CFX384 Real-Time System (Bio-Rad Laboratories). HPRT was used as a housekeeping gene for all reactions, and relative expression compared with control-treated animals was calculated as 2^(−ΔΔCT).

After mincing with a razor blade, brain tissue was passed through an 18- and 22-gauge needle. Thirty microliters of brain homogenate was then placed onto a microscope slide (VWR), and cysts were counted manually on a brightfield DM 2000 LED microscope (Leica Biosystems).

Brains from mice were embedded in OCT and flash frozen on dry ice. Samples were then stored at −20°C until cutting. For Ki67 immunofluorescence staining, fresh-frozen sections were lightly fixed in 25% EtOH/75% acetone solution for 15 min at −20°C. Sections were then blocked in 1% PBS containing 2% goat serum (Jackson ImmunoResearch), 0.1% triton, and 0.05% Tween 20 for 1 h at room temperature and then incubated with primary Abs at 4°C overnight. After primary staining, sections were washed with 1% PBS and incubated with secondary Abs for 30 min at room temperature. Finally, sections were nuclear stained with DAPI (Thermo Fisher Scientific) for 5 min at room temperature. Sections were then covered with AQUAMONT (Lerner Laboratories) and coverslips (Thermo Fisher Scientific).

For p-STAT5 immunofluorescence staining, fresh-frozen sections were fixed in 3.2% PFA for 20 min at room temperature and then permeabilized with ice-cold 90% methanol for 10 min at −20°C (31). Sections were then incubated with blocking buffer and an anti–p-STAT5 (D47E7) (Cell Signaling Technologies) primary Ab as described above. After primary Ab staining, the p-STAT5 signal was amplified using an anti-rabbit biotinylated Ab (Jackson ImmunoResearch) before using a streptavidin secondary Ab. All images were captured using a DMI6000 B widefield microscope with a Hamamatsu C10600 Orca R² digital camera (Leica Biosystems) and visualized using MetaMorph software. Images were then analyzed using ImageJ software.

For quantification of the number of Ki67- or p-STAT5–expressing cells, 12–15 equal-sized pictures were taken (40×) within each brain slice, and the numbers of Ki67⁺ or p-STAT5⁺CD4⁺ or CD8⁺ cells were counted in each image. Numbers from each picture were then averaged and reported as the average number of Ki67⁺ or p-STAT5⁺ per 100 or 500 μm² per mouse, respectively.

Statistical analysis comparing two different groups at a single time point was performed in Prism software (v. 7.0a) using a Student t test. In some cases, multiple experiments from different infection dates were combined to show natural biological variation between infections. When data from multiple experiments were combined, a randomized block ANOVA was used in R v.3.4.4 statistical software. This statistical test accounted for natural variability between experimental dates by modeling the treatment (IgG versus anti–IL-10R or anti-ICOSL) as a fixed effect and the experimental date as a random effect. The particular test used for each individual panel shown is specified in the figure legend. The p values are displayed as follows: *p < 0.05, **p < 0.01, and ***p < 0.001. All data were graphed using Prism software, and the number of mice per group is indicated in the figure legend. All data are presented as the mean ± SD unless otherwise noted in the figure legend.

During chronic infection with T. gondii, both effector T cells and T_regs recruited to the brain are capable of producing IL-10 (38). Previously published results have shown a requirement for IL-10 signaling to limit fatal immunopathology in both the acute and chronic stages of infection (3, 21, 39, 40). Despite the necessity for IL-10 over the course of infection with T. gondii, previous studies addressing the role of IL-10 in the chronic phase of infection relied on total IL-10 knockout mice, which succumb to fatal immunopathology during the first 2 wk of infection (41). For these mice to survive to the chronic stage of infection, the antiparasitic drug sulfadiazine must be administered for the first 2 wk of infection to limit parasite replication and dissemination. These previously published studies reported that, after sulfadiazine treatment, IL-10 knockout mice subsequently presented with CD4⁺ T cell–dependent lethal immunopathology and died late in the chronic stage of disease (21). It is still unknown, however, how a loss of IL-10 only in the chronic phase of infection influences immune responses. Thus, we treated mice with an anti–IL-10R blocking Ab beginning 4 wk postinfection. Mice treated with an anti–IL-10R–blocking Ab during the chronic stage of infection presented with overt disease and became moribund between 7 and 10 d posttreatment. H&E staining of tissue sections from brains of anti–IL-10R–treated mice showed increased leukocyte infiltration and associated areas of necrosis not seen in control-treated animals (Fig. 1A, 1B). Similar to previously published results using IL-10 knockout mice (21), the increased numbers of immune cells in the brains of anti–IL-10R–treated mice included increases in both CD4⁺Foxp3⁻ T cells and infiltrating macrophages (Fig. 1C, Supplemental Fig. 1A). Although the increase in T cell numbers in the brains of anti–IL-10R–treated mice predominantly came from an increase in the CD4⁺Foxp3⁻ T cell compartment and not the CD8⁺ T cell compartment, an increased frequency of both Ki67⁺CD4⁺ and CD8⁺ effector T cells was observed (Fig. 1D). Interestingly, using an MHCII tetramer reagent to measure CD4⁺ T cells specific for the parasite, we observed no significant increase in the number of tetramer-positive CD4⁺Foxp3⁻ T cells in the brain (Fig. 1E). Although this analysis of Ag specificity was done using only a single parasite peptide, this result suggests that IL-10R blockade in the chronic phase of infection leads to the expansion of CD4⁺ effector T cells with potentially different Ag specificities, possibly through the expansion of other parasite-specific T cell clones or self-antigen–specific T cell clones.

The increased proliferation of effector T cells in the brain correlated with an increase in the expression of costimulatory molecule CD80 on infiltrating DCs and macrophages in anti–IL-10R–treated mice (Fig. 1F). The increased numbers of immune cells infiltrating the brain was associated with increased mRNA levels of many proinflammatory cytokines and chemokines, including IFN-γ, IL-6, TNF-α, IL-17, and CXCL1, demonstrating a widespread increase in inflammation in the brain in the absence of IL-10 signaling (Fig. 1G, 1H). An increase in IL-17 production is notable in this model as infection of wild-type mice with T. gondii typically leads to a robust Th1-polarized immune response, characterized by IL-12 and IFN-γ production that persists throughout the chronic stage, with very little production of Th17-associated cytokines (42, 43). Indeed, production of IL-17 has been linked to a loss of IL-27–mediated regulation of immune responses during infection with T. gondii (43, 44), suggesting a pathogenic role for IL-17 in this context.

Production of IL-17 and CXCL1 has previously been shown to enhance recruitment of neutrophils to inflamed tissues and enhance their activity in certain disease contexts (45–50). Neutrophil recruitment to the CNS, however, has been shown to be detrimental in many cases (50–52). With the increased mRNA levels of both IL-17 and CXCL1 seen in the brain after IL-10R blockade, we wanted to assess whether more neutrophils were recruited to the inflamed brain in this context. Indeed, anti–IL-10R–treated mice had a nearly 3-fold increase in neutrophil numbers in the brain in comparison with control-treated mice (Fig. 1I, 1J, Supplemental Fig. 1B, 1C). Despite the increased T cell response in the brain with IL-10R blockade, no change in the number of parasite cysts was observed (Fig. 1K).

The effects of IL-10R blockade during the chronic phase of infection were not limited to the inflamed brain as increased CD4⁺Foxp3⁻ effector T cells were found in the spleens of anti–IL-10R–treated mice (Supplemental Fig. 1D). Similar to what was observed in the brain, although total CD4⁺ effector T cells were increased in the spleen after IL-10R blockade, there was no increase in parasite tetramer-specific CD4 T cells in this tissue (Supplemental Fig. 1E). We also observed an increase in myeloid cells in the spleen following IL-10R blockade (Supplemental Fig. 1F). These myeloid cells were also highly activated, with higher levels of CD80 expression on APCs in the spleens of anti–IL-10R–treated mice as opposed to controls (Supplemental Fig. 1G). Large areas of necrosis were also observed in the livers of anti–IL-10R–treated mice (Supplemental Fig. 1H, 1I), suggesting that the increased inflammatory response seen following IL-10R blockade can contribute to extensive immunopathology in not only the brain but peripheral tissues as well. Overall, a loss of IL-10 signaling specifically during the chronic infection led to widespread immune cell activation that rapidly resulted in fatal immunopathology.

Despite clear evidence that IL-10–mediated regulation of immune responses during T. gondii infection is required for host survival, little is known about what signals can induce IL-10 production in activated T cells during the chronic inflammation in the brain associated with the later stages of infection. Several studies have implicated a role for ICOS in inducing IL-10 production in cases of acute inflammation (29, 33). We found ICOS-expressing T cells in the brain during chronic T. gondii infection as well as infiltrating APCs expressing the ICOSL (Fig. 2A, 2B). Additionally, ICOS was most highly expressed on the activated T cells in the brain compared with the secondary lymphoid organs (Supplemental Fig. 2A). To address whether ICOS signaling is important for T cell production of IL-10 in the brain during chronic infection, we treated chronically infected IL-10–eGFP reporter (Tiger) mice with the ICOSL-blocking Ab. Surprisingly, following ICOSL blockade in the chronic stage of infection, we did not observe decreases in IL-10 production from either CD4⁺Foxp3⁻ effector T cells or CD4⁺Foxp3⁺ T_regs in the brain (Fig. 2C), and IL-10 mRNA expression from whole brain homogenate was also not decreased (Fig. 2D). Although IL-10 levels were not decreased after anti-ICOSL treatment, we unexpectedly observed a 2–3-fold increase of both CD4⁺Foxp3⁻ and CD8⁺ effector T cells, respectively (Fig. 2E). This increase in effector T cell number was not due to a loss of the local T_reg population because their numbers in the brain were also increased, although to a lesser degree than the effector T cell populations (Fig. 2E). The increased numbers of T cells in the brains of anti-ICOSL–treated animals could be seen dispersed throughout the brain parenchyma (Fig. 2F, 2G), although unlike anti–IL-10R–treated animals, ICOSL blockade was not lethal in the observed time frame. Furthermore, we observed an increase in the number of tetramer-positive CD4⁺ effector T cells in the brain (Fig. 2H), suggesting that ICOSL blockade can lead to an expansion of parasite-specific CD4⁺ effector T cells. We also found an increase in the number of IFN-γ–producing effector T cells in the brain (Fig. 2I), whereas mRNA levels of other proinflammatory cytokines and chemokines were not increased (Fig. 2J). Infiltrating myeloid cell numbers were also assessed, revealing almost a 2-fold increase in DCs in the brain compared with control-treated animals (Fig. 2K). Although increased numbers of DCs were found in the brain following anti-ICOSL treatment, there was no effect on IL-12 production from either the DCs or macrophages isolated from the brain (Supplemental Fig. 2B), suggesting that continued production of IL-12 is not reliant on ICOS–ICOSL interactions. An important distinction between IL-10R blockade and ICOSL blockade in the chronic phase of infection was a difference in neutrophil recruitment. Whereas an increase in neutrophil numbers in the brain was seen in anti–IL-10R–treated mice (Fig. 1I, 1J), no significant increase was seen in anti-ICOSL–treated mice (Fig. 2L, 2M).

Blockade of ICOSL primarily affected the T cell populations in the CNS and did not affect T cell numbers in the spleen, draining lymph nodes, or blood of anti-ICOSL–treated mice (Supplemental Fig. 2C–E). Interestingly, although we observed a significant increase in the number of tetramer-positive CD4⁺ effector T cells in the brain following ICOSL blockade, we did not see an increase in tetramer-positive CD4⁺ effector T cells in the spleen at the same time point (Supplemental Fig. 2F). ICOS signaling has also been shown to be crucial for primary Ab responses to infection (26, 53, 54). To assess whether the increased T cell numbers found in the brains of anti-ICOSL–treated animals were merely a result of a change in parasite burden because of decreased Ab production, both parasite-specific serum IgG titers and brain cyst burden was measured. No change was seen in either circulating parasite-specific IgG (Supplemental Fig. 2G) or parasite burden in the brain (Fig. 2N). Representative H&E-stained brain sections from control- or anti-ICOSL–treated mice reflected the increased density of infiltrating immune cells seen by flow cytometry and immunohistochemistry but lacked severe tissue damage in comparison with those treated with anti–IL-10R–blocking Ab (Supplemental Fig. 2H, 2I). Compared with anti–IL-10R–treated mice, fewer areas of necrosis and tissue destruction were evident following anti-ICOSL treatment (Supplemental Fig. 2H, 2I). Together, these results suggest that ICOS signaling limits excessive T cell responses in the brain during chronic neuroinflammation independent of changes in either IL-10 production or parasite burden.

We next wanted to determine how lack of ICOS–ICOSL interaction during chronic inflammation leads to increased numbers of T cells in the CNS. Using immunofluorescence staining for Ki67, increased numbers of proliferating CD4⁺ and CD8⁺ effector T cells were found throughout the brain after ICOSL blockade during chronic infection (Fig. 3A, 3B, 3D, 3E). Using flow cytometry, we confirmed this increase in the number of proliferating effector T cells in the brain (Fig. 3C, 3F). Distinct from IL-10R blockade, the increase in Ki67⁺ effector T cells in the brain following ICOSL blockade was not correlated with an increase in CD80 or CD86 expression on infiltrating APCs (Fig. 3G). Rather, ICOSL blockade led to the upregulation of the prosurvival factor Bcl-2 in both CD4⁺Foxp3⁻ and CD8⁺ effector T cells isolated from the brain (Fig. 3H, 3I). Bcl-2 has been shown to increase T cell survival by preventing cytochrome c release from mitochondria that would normally lead to apoptosis (55, 56). The upregulation of Bcl-2 following ICOSL blockade suggested that effector T cells could have increased survival in the brain following anti-ICOSL treatment. To assess this, we isolated T cells from the brain and stained them with Annexin V as a marker of apoptosis. Interestingly, we observed a significant decrease in the frequency of Annexin V–positive CD4⁺ and CD8⁺ T cells in the brain after anti-ICOSL treatment, suggesting that fewer T cells in the brain are undergoing early apoptosis following ICOSL blockade (Fig. 3J, 3L). Together, these results suggest that ICOS limits effector T cell proliferation and survival during chronic neuroinflammation.

IL-2 signaling in T cells has been shown to support both their proliferation and survival (57, 58). Thus, we wanted to determine if the increased effector T cell proliferation and survival factor Bcl-2 expression could be a result of an increased response to IL-2 following ICOSL blockade. After ICOSL blockade, both CD4⁺Foxp3⁻ and CD8⁺ effector T cells isolated from the brain expressed higher levels of CD25, and there was a 2–3-fold increase in the total number of CD25⁺ effector T cells in the brain compared with control-treated animals (Fig. 4A–D). To assess whether more of the infiltrating T cells in the brain may be responding to IL-2, we used immunofluorescence staining for p-STAT5, the main signaling molecule downstream of the IL-2R. Correlated with the increased number of CD25⁺ effector T cells found in the brain with ICOSL blockade, we observed increased numbers of p-STAT5–positive CD4⁺ and CD8⁺ T cells in the brains of anti-ICOSL–treated animals (Fig. 4E–J). The increases in IL-2–associated signaling molecules occurred independently of increased IL-2 production from effector T cells (Supplemental Fig. 2J), suggesting that similar amounts of IL-2 are available in the brain environment after anti-ICOSL or control treatment. Interestingly, T_regs were increased in number in the brain following ICOSL blockade (Fig. 1E), but unlike effector T cells, T_reg expression of CD25 and Bcl-2 was unaffected after anti-ICOSL treatment (Supplemental Fig. 3A, 3B). In contrast, we did observe increases in Ki67⁺ T_regs in the brain after anti-ICOSL treatment (Supplemental Fig. 3C–E). These results indicate that ICOSL blockade may differentially affect regulatory and effector T cell populations in the inflamed brain. Taken together, these data suggest that after ICOSL blockade, effector T cells may maintain higher levels of CD25 and p-STAT5 which in turn could support increased proliferation, Bcl-2 expression, and decreased apoptosis of effector T cells during chronic inflammation in the brain.

Whereas increased expression of both Bcl-2 and CD25 was seen on effector T cells in the brain following ICOSL blockade, no change in Bcl-2 expression was seen in effector T cell populations isolated from the brains of anti–IL-10R–treated mice (Fig. 5A, 5B). Additionally, increased levels of CD25 were not observed to the same degree with anti–IL-10R treatment as there were no changes in the levels of CD25 expression on CD4⁺Foxp3⁻ T cells and only a slight increase in CD25 expression on the CD8⁺ T cell population (Fig. 5C). Overall, although both IL-10R blockade and ICOSL blockade resulted in increased numbers of effector T cells in the brain, IL-10R blockade correlated with increased APC activation and inflammatory cytokine mRNA expression, whereas ICOSL blockade correlated with upregulation of IL-2–associated signaling molecules and decreased apoptosis in effector T cells. These data further support that ICOS and IL-10 signaling pathways can differentially promote the regulation of effector T cell responses in the brain during chronic infection.

Inflammation is required to promote clearance of pathogens but preventing excessive inflammation during immune responses to infection is necessary to avoid immune-mediated tissue damage (3, 7, 15). In cases of chronic infection, this balance between inflammation and regulation must be maintained over long periods of time (21, 43), but many of the signals required to maintain control of ongoing immune responses are not well understood. Our results identify one such signal, ICOS, whose expression on activated T cells in the chronically infected brain provides a regulatory signal, preventing overabundant T cell accumulation in the inflamed CNS. We show that ICOSL blockade is associated with an increased expression of the IL-2–associated signaling molecules CD25, p-STAT5 and Bcl-2 as well as effector T cell accumulation in the brain.

ICOS, similar to its family member CD28, was initially characterized for its ability to amplify both B cell Ab production and T cell inflammatory cytokine production in cases of acute inflammation (24, 25, 53). The proinflammatory role for ICOS signaling was subsequently supported by human data, as humans carrying mutations in the ICOS gene are included in the class of mutations known as common variable immunodeficiency (CVID) (59–61). Patients with CVID have increased susceptibility to bacterial infections and are diagnosed based on severely decreased class-switched Ab production, further implicating ICOS as important for T cell–dependent B cell responses (60). Interestingly, further data from patients with CVID began to emerge highlighting widespread immune system abnormalities outside of the B cell compartment, particularly splenomegaly, a loss of naive CD4⁺ T cells and an expansion of activated CD4⁺ T cells and increased T cell inflammatory cytokine production (59, 62, 63). Additionally, despite being defined as an immunodeficiency disease, ∼20% of CVID patients also present with autoimmune complications, although the pathogenesis of this autoimmunity remains unclear (64).

Studies regarding the role of ICOS in promoting the regulation of immune responses largely come from mouse models, in which ICOS has been shown to both support T_reg populations and promote IL-10 production (29, 31, 33, 34). Our observation of increased effector T cell numbers in the brain following ICOSL blockade suggests a loss of regulation that could occur if the T_reg population is decreased or defective after anti-ICOSL treatment. Surprisingly, we found no evidence of a local T_reg defect following ICOSL blockade, suggesting that T_regs in the brain during chronic infection rely on signals other than ICOS to support their survival, proliferation, and accumulation in the tissue as well as their production of IL-10. Many activated effector T cells in the brain continue to produce IL-2 in the chronic phase of infection with T. gondii, so it is possible that the largely CD25⁺ T_regs in the CNS rely mainly on IL-2 signals for their maintenance in an inflamed tissue, whereas other signals are required during homeostasis when lower levels of IL-2 would be present in the absence of an ongoing effector T cell response. We also observed an increase in DC numbers in the brain after anti-ICOSL treatment that could be potentiating the increased T cell response. Although this could be a contributing factor, we do not anticipate that the increased DC number in the brain is the primary driver of the T cell expansion we observe with ICOSL blockade because the activation status of the DCs, including MHC and costimulatory molecule expression, is not increased.

One of the main differences observed between IL-10R blockade and ICOSL blockade in the chronic phase of infection was the disparity in the magnitude of the immune response. IL-10R is expressed more broadly than ICOS, which may explain the widespread inflammation and lethality of IL-10R versus ICOSL blockade. IL-10R is expressed by most hematopoietic cells but can also be induced on nonhematopoietic cells, such as fibroblasts and endothelial cells, rendering them also able to respond to IL-10 in inflammatory settings (65–67). In the context of chronic T. gondii infection, this could explain why changes were seen in myeloid and T cell subsets in both the brain and peripheral tissues following anti–IL-10R blockade. In contrast, ICOS is only expressed on activated T cells during chronic T. gondii infection, and its highest levels of expression were found in the inflamed brain. This more limited expression pattern of ICOS compared with IL-10R could potentially explain the more local and specific response to ICOSL blockade.

Another interesting aspect of the immunological phenotype seen with IL-10R blockade during chronic infection was the differential effects on the accumulation of CD8⁺ and CD4⁺ effector T cell populations in the brain. The accumulation of CD4⁺ effector T cells, but not CD8⁺ T cells, is in agreement with a previously published study examining T cell responses in IL-10 knockout mice (21). In the current study, we also examined the phenotype of APCs and the proliferative capacity of T cells in the brain. Following IL-10R blockade, local APCs in the brain upregulate CD80, so it is possible that the CD4⁺ effector T cells infiltrating the brain are interacting with the highly activated local MHCII⁺ APCs more so than the CD8⁺ effector T cells, leading to their increased accumulation. It is interesting to note that, of the brain-infiltrating APCs isolated during chronic T. gondii infection, only a small fraction of them are classic CD8α⁺ cross-presenting DCs that could interact in a TCR-dependent fashion with infiltrating CD8⁺ T cells (68). This could suggest that the activated CD8⁺ effector T cells in the brain rely less on local restimulation through TCR–MHC and costimulatory interactions and are therefore less affected by extrinsic mechanisms of suppression through APCs in this context. Overall, it is largely unknown what kinds of secondary signals, such as TCR–MHC or costimulatory interaction, are required for activated T cells to carry out effector function and survive at a distal site of inflammation after initial priming in secondary lymphoid organs, although it is likely these requirements differ in some way for CD8⁺ and CD4⁺ T cells. Although the overall number of CD8⁺ effector T cells in the brain is not increased, we still observed an increased frequency of Ki67⁺ CD8⁺ T cells in the brain after IL-10R blockade, suggesting that this population is still responsive to IL-10–mediated suppression, perhaps through an intrinsic response to IL-10 that limits their proliferative capacity. IL-10 has been shown to be able to directly inhibit proliferation of CD8⁺ T cells in vitro (14) as well as control the threshold of their response to Ag upon initial activation (69). Much is still unknown about the direct effects of IL-10 on activated CD8⁺ T cells that could contribute to their regulation, although these data further suggest that CD8⁺ effector T cells may be differentially regulated from the CD4⁺ effector T cell compartment in the brain.

Two well-characterized inhibitory coreceptors are CTLA-4 and PD-1, both of which have been shown to carry out their inhibitory effect at least partially through inhibition of the PI3K/Akt signaling pathway (70, 71). In this light, it is interesting to consider ICOS, which is a potent activator of the PI3K/Akt pathway (72, 73), as also providing an inhibitory signal to T cells during chronic inflammation. Initial Akt activation induced during priming has been associated with increased T cell responses, both through promotion of T cell proliferation and survival (74–76). However, more recent reports have shown that constitutive Akt activation in CD8⁺ T cells is associated with decreased expression of CD122 and Bcl-2 and can promote the development of short-lived effector T cells over the development of memory T cells, whereas constitutive STAT5 signaling can maintain Bcl-2 expression and favor the development of memory precursor cells (77, 78). These results emphasize that the fate of T cells is extremely sensitive to both the level and duration of signaling cascades like PI3K/Akt. During the chronic neuroinflammation seen with T. gondii infection, continued activation of Akt downstream of ICOS in activated T cells in the brain could potentially serve as an intrinsic mechanism of controlling T cell responses by downregulating IL-2–associated signaling molecules and driving effector T cells in the brain to be short-lived effectors rather than memory precursors. Indeed, our data support the hypothesis that increased IL-2 signaling in activated effector T cells in the inflamed brain provides survival signals, like upregulation of Bcl-2, which limits T cell apoptosis and allows for T cell accumulation in the tissue. Continued signaling through ICOS then might normally promote short-lived effector T cells by providing a strong PI3K/Akt signal that inhibits expression of IL-2–associated signaling molecules, leading to apoptosis.

Overall, we provide evidence that ICOS costimulation provides an inhibitory signal to Ag-experienced effector T cells in the chronically inflamed brain. These data postulate a regulatory role for ICOS in T cells during chronic neuroinflammation that is distinct and more specific than the suppressive role of IL-10. Altogether, although we show that IL-10 is absolutely required for preventing exaggerated immune responses and immunopathology, other regulatory signals like ICOS are also likely at play during chronic inflammation that provide more fine-tuned suppression of ongoing immune responses without affecting IL-10–mediated suppression.

We thank all the members of the Harris laboratory, the Center for Brain Immunology, and Glia for insightful comments and discussion during the preparation of this work. We acknowledge support from the Research Histology Core Facility at the University of Virginia. We thank Marieke K. Jones for helpful input regarding relevant statistical analysis and coding and Dr. Ken Tung for assistance with histopathological examination of tissue sections.

The authors have no financial conflicts of interest.