Highly Activated Cytotoxic CD8 T Cells Express Protective IL-10 at the Peak of Coronavirus-Induced Encephalitis

Interleukin-10 is a potent anti-inflammatory cytokine with a crucial role in limiting proinflammatory responses and preventing autoimmune diseases. Although several factors contribute to the anti-inflammatory repertoire, the role of IL-10 is nonredundant. Thus, a spontaneous enterocolitis occurs in mice that are deficient in IL-10, because of florid proinflammatory T cell responses to normal bacterial flora (1). In addition, a deficiency in IL-10 results in exaggerated proinflammatory responses during bacterial, protozoal, and viral infections (2, 3). In most of these studies, IL-10 is produced by CD4 T cells (including effector and regulatory T cells) or by macrophages or dendritic cells, although NK cells have also been identified as an important source in systemic Toxoplasma gondii infections (2, 4, 5). Production of IL-10 by IFN-γ⁺CD4 (Th1) T cells is dependent on strong Ag stimulation; in vitro, these cells required greater dosages of peptide to express maximal levels of IL-10 compared with IFN-γ (6). In addition, IL-10 expression in CD4 T cells is dependent on stimulation via IL-12 and STAT-4 signaling, and requires activation of ERK-dependent pathways.

IL-10 production before complete pathogen clearance contributes to persistent infection. Neutralization of IL-10 in persistent leishmania or lymphocytic choriomeningitis virus infections results in pathogen clearance, with concomitant immunopathological disease in some instances (7–9). Although it is clear that IL-10 has an important role in prolonging pathogen persistence during chronic infections, only recently has its role in nonpersisting virus infections begun to be appreciated. Unlike chronic infections, CD8 T cells, in addition to CD4 T cells, have been identified as major producers of IL-10 in experimental respiratory infections caused by influenza A virus, SV5, and respiratory syncytial virus (10–13); little is known about IL-10 expression by CD8 T cells.

Control of the pathogen and consequent inflammatory response is especially important in infections of the CNS, including viral encephalitis, to minimize tissue damage. Therefore, to investigate the role of IL-10 in acute encephalitis, in particular IL-10 expressed by CD8 T cells, we infected mice with the neurotropic JHM strain of mouse hepatitis virus (14, 15). The J2.2-V-1 variant of this virus produces encephalomyelitis, with progression to a chronic demyelinating disease. Infectious virus is cleared by 10–14 d postinfection (p.i.), although viral Ag and RNA can be detected for months after infection accompanied by low levels of demyelination (14, 16). Inflammation is maximal and extensive throughout the acute phase of the infection (14).

In this article, we demonstrate that expression of IL-10 by CD8 T cells during peak inflammation is not limited to respiratory infections, and that ∼50% of IL-10–producing cells in the infected brain are murine hepatitis virus-specific cytotoxic CD8 T cells. In addition, this CD8 T cell-derived IL-10 ameliorates clinical disease in acute encephalitis and subsequent demyelination. IL-10 expression by CD8 T cells requires strong Ag stimulation and signaling through MAPK pathways. Furthermore, using microarray profiling, quantitative RT-PCR (qRT-PCR), and protein analyses, we demonstrate that the fraction of CD8 T cells that produce IL-10 are hyperactivated CD8 T cells. Thus, cytotoxic CD8 T cells may regulate their inflammatory effects during acute infections by producing IL-10, and thereby minimize immunopathological disease.

Specific pathogen-free 6-wk-old Thy1.2 and Thy1.1 C57BL/6 mice were purchased from the National Cancer Institute (Bethesda, MD) and The Jackson Laboratory (Bar Harbor, ME), respectively. Vert-X (C57BL/6 bicistronic IL-10/eGFP) mice were kindly provided by C. Karp (Cincinnati Children’s Hospital) (17). IL-10^−/− mice (B6.129P2-Il10tm1Cgn/J) were bred at the University of Iowa. Mice were examined and weighed daily. All animal studies were approved by the University of Iowa Animal Care and Use Committee.

Recombinant J2.2-V-1, used in all experiments, was propagated and titered as described previously (18). Mice were inoculated intracerebrally with 500 or 1500 PFU J2.2-V-1 in 30 μl DMEM.

Brain mononuclear cells were isolated as described previously (16).

All Abs were purchased from BD Pharmingen (San Diego, CA) or eBioscience (San Diego, CA) unless indicated later. NK cells were CD45⁺CD3⁻NK1.1⁺. Macrophages were CD45⁺CD11b⁺Ly6C⁺Ly6G⁻. Neutrophils were CD45⁺CD11b⁺Ly6C⁺Ly6G⁺. MHC class II was detected with anti–I-A/anti–I-E. For detection of intracellular protein expression, CD8 or CD4 T cells were stimulated with 1 μM peptide S510 or 5 μM peptide M133, respectively, as previously described (16). IFN-γ, IL-10, TNF, CD107a/b, and granzyme B were detected using Abs purchased from BioLegend (San Diego, CA). For detection of p-ERK, mAb clone D13.14.4E was purchased from Cell Signaling Technology (Beverly, MA). FcεR1 was detected using anti-DNP IgE (Sigma-Aldrich, St. Louis, MO) followed by rat anti-mouse IgE. M133-specific and S510-specific cells were also detected using PE-conjugated I-A^b/M133 and H-2D^b/S510 tetramers, respectively, obtained from the National Institutes of Health Tetramer Core Facility (Atlanta, GA). Cells were analyzed using a FACSCalibur (BD Biosciences).

Mononuclear cells were isolated from J2.2-V-1–infected Vert-X/Thy1.2 mice brains at day 7 p.i. Cells from three mice were pooled and CD8 T cells were positively selected or Thy1.2⁺ (CD8 and CD4) T cells were negatively selected using magnetic beads from STEMCELL Technologies (Vancouver, British Columbia, Canada), according to the manufacturer’s protocol. GFP⁺, GFP⁻, CD69⁺, and CD69⁻ CD8 and/or CD4 T cells were then sorted by a FACSDiva or a FACSAria (BD Biosciences). Postsort analysis showed that cells were >99% pure.

Thy1.1 B6 lymph node cells were labeled with 2 μM CFSE, and 2 × 10⁶ cells were placed in the lower compartment of anti-CD3 (clone 145-2C11; eBioscience) coated wells. A 0.4-μm filter separated the two compartments. Thy1.1 B6 splenocytes were pulsed with varying concentrations of S510 peptide (1 μM to 0 μM) for 1 h before irradiation with 3000 rad. A total of 2 × 10⁶ cells was placed in the upper compartment. Thy1.2 CD8⁺GFP⁺ and CD8⁺GFP⁻ cells were sorted as described earlier, and 2.5 × 10⁴ cells were placed in the upper compartment with the irradiated splenocytes. A blocking anti–IL-10 mAb (clone JES5-2A5) or an isotype-matched control mAb at 10 μg/ml (both from BioLegend) was added to some wells. Some cultures were treated with the MEK1/2 inhibitor PD 184161 (ERK1/2) or the p38 inhibitor PD169316 (Cayman Chemical, Ann Arbor, MI). For proliferation studies, lower compartment lymphocytes were stained with anti–CD4-PE and anti–CD8-PerCP, and examined by flow cytometry for CFSE dilution after 48 h. Culture supernatants were harvested for cytokine detection by ELISA.

Thy1.2⁺GFP⁺ and Thy1.2⁺GFP⁻ (CD4 and CD8) cells from d7 J2.2-V-1–infected Vert-X mouse brains were sorted as described earlier. Either 1 × 10⁶ GFP⁺ or 1 × 10⁶ GFP⁻ cells were transferred in 300 μl PBS i.v. into J2.2-V-1–infected Thy1.1⁺ B6 mice at day 1 p.i. Six days later, lymphocytes were harvested, stained, and analyzed by flow cytometry.

J2.2-V-1–infected brains were weighed and homogenized directly into 50 mM Tris, 150 mM NaCl, 5 mM EDTA, 1 mM Na₃VO₄, 1% NP-40, and a protease inhibitor mixture (Complete; Roche, Mannheim, Germany). IL-10 and IFN-γ ELISA were performed using reagents and protocols provided by the manufacturer (eBioscience and BioLegend, respectively). Samples were plated in duplicate.

Blinded quantification of demyelination was performed using Luxol fast blue-stained sections as previously described (19).

GFP⁺CD8⁺ and GFP⁻CD8⁺ T cells harvested from infected Vert-X mouse brains were sorted at day 7 p.i., as described earlier. RNA was purified using RNeasy columns (Qiagen) according to manufacturer instructions. RNA samples were verified for purity spectroscopically, and the quality of the intact RNA was assessed using an Agilent 2100 Bioanalyzer. RNA for the microarray was processed using a NuGEN WT-Ovation Pico RNA Amplification System together with a NuGEN WT-Ovation Exon Module. Samples were hybridized and loaded onto Affymetrix GeneChip Mouse GENE 1.0 ST arrays. Arrays were scanned with an Affymetrix Model 7G upgraded scanner, and data were collected using GeneChip Operating Software. Complete microarray data have been deposited at the Gene Expression Omnibus under accession number GSE25846 (http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE25846).

Data from the Affymetrix Mouse Exon 1.0 ST arrays were first quantile normalized and median polished using Robust Multichip Average background correction with log₂ adjusted values. GFP⁺ and GFP⁻ samples were obtained from the same mice, allowing for a paired t test. Partek batch correction was used to remove variation caused by the hybridization batch. Probe sets for exons were then summarized for a specific gene using the median value. After obtaining log₂ expression values for genes, significance testing was performed comparing the two different CD8⁺ T cell groups (IL-10⁺ versus IL-10⁻). False discovery rate correction was applied to all of the p values to correct for multiple testing. Significance was assessed by false discovery rate q < 0.25 and 2-fold change. Analysis and visualization were done in PartekGS software. Functional assignment of the genes was performed using the “Functional Annotation Tools” in DAVID bioinformatics resources (http://david.abcc.ncifcrf.gov) following recommended protocols. Additional analysis was performed using Ingenuity Pathways Analysis (Ingenuity Systems, Redwood City, CA).

GFP⁻ and GFP⁺ CD8 T cells were sorted as described earlier, centrifuged, and resuspended in TRIzol Reagent (Invitrogen/Life Technologies, Carlsbad, CA). RNA was extracted and reverse-transcribed using Superscript II (Invitrogen). Relative cytokine transcript quantities were determined by qRT-PCR using SYBR Green (SA Biosciences, Frederick, MD). Primers used for qRT-PCR were as follows: IL-10: forward, 5′-GCGTCGTGATTAGCGATGATG-3′; reverse, 5′-CTCGAGCAAGTCTTTCAGTCC-3′; TNF: forward, 5′-TCAGCCGATTTGCTATCTCA-3′; reverse, 5′-CGGACTCCGCAAAGTCTAAG-3′; TLR2: forward, 5′-TTTCGTTCATCTCTGGAGCA-3′; reverse, 5′-GAGTCCGGAGGGAATAGAGG-3′; TLR3: forward, 5′-ACCTTTGTCTTCTGCACGAACCT-3′; reverse, 5′-AGTTCTTCACTTCGCAACGCA-3′; TLR4: forward, 5′-TCAGAACTTGAGTGGCTGCA-3′; reverse, 5′-GAGGCCAATTTTGTCTCCAC-3′; TLR7: forward, 5′-CAGCCATAACCAGCTGACAA-3′; reverse, 5′-TTGCAAAGAAAGCGATTGTG-3′; CCL2: forward, 5′-AGCACCAGCCAACTCTCACT-3′; reverse, 5′-TCATTGGGATCATCTTGCTG-3′; IL-1α: forward, 5′-CACAACTGTTCGTGAGCGCT-3′; reverse, 5′-TTGGTGTTTCTGGCAACTCCT-3′; IL-1β: forward, 5′-ACTGTTTCTAATGCCTTCCC-3′; reverse, 5′-ATGGTTTCTTGTGACCCTGA-3′; STAT-4: forward, 5′-CCTACTGGCAGAGAGTCTTTTCC-3′; reverse, 5′-GGTTGTAGATCAGGAAGGTAGC-3′; perforin: forward, 5′-TTGGCCCATTTGGTGGTAAG-3′; reverse, 5′-AGTCTCCCCACAGATGTTCTGC-3′; prosaposin: forward, 5′-ACTTTCCCACCACCTGTAGC-3′; reverse, 5′-TAACTGCACAGGCTGTCTCC-3′; HEXA: forward, 5′-GTGCTCATGAAGTCGTAGGTGC-3′; reverse, 5′-CAGGATGTGAAGGAGGTCATTG-3′; hypoxanthine-guanine phosphoribosyltransferase: forward, 5′-GCGTCGTGATTAGCGATCATC-3′; reverse, 5′-CTCGAGCAAGTCTTTCAGTCC-3′. Data were analyzed as previously described (18).

GFP⁻ and GFP⁺ CD8 T cells were sorted as described earlier. EL4 target cells were peptide pulsed with 1 μM S510 peptide at 37°C for 1 h and washed. Cytolytic activity was determined using a Cytotox 96 nonradioactive kit (Promega, Madison, WI) according to the manufacturer’s instructions. Cytotoxicity assays were performed in triplicate.

JHMV-immune splenocytes were prepared from B6 or IL-10^−/− mice immunized with 3 × 10⁵ PFU wild-type JHMV i.p. 7 d before adoptive transfer. Donor splenocytes were depleted of CD4 T lymphocytes, as previously described (20). Greater than 99% depletion was achieved, as assessed by FACS.

Two-tailed unpaired Student t tests were used to analyze differences in mean values between groups. Fisher’s exact tests were used to analyze differences in survival. One-way ANOVAs were used to analyze differences in multiple groups. All results are expressed as means ± SEs of the means. Differences of p values <0.05 were considered significant.

Previous work demonstrated that IL-10 protein is expressed in the CNS of J2.2-V-1–infected mice (5, 21, 22). To identify the cellular sources of IL-10 during the acute and recovery stages of viral encephalitis, we infected IL-10–GFP reporter (Vert-X) mice (5, 13, 17) with J2.2-V-1 and monitored brain, draining cervical lymph nodes (CLN), spleen, and peripheral blood for GFP⁺ cells from days 0–42 p.i. These time points were chosen because virus is cleared by day 14 p.i. and inflammation is mostly resolved by day 42 p.i. (14, 23). We observed the highest frequency of IL-10–producing cells at day 7 p.i., coincident with the onset of the adaptive immune response (Fig. 1A, 1B). Uninfected mice (Fig. 1B) and mice at days 2 and 4 p.i. (data not shown) had very low levels of GFP⁺ cells. At all time points, the highest frequency of GFP⁺ cells was found in the brain, and not in the CLN, spleen, or peripheral blood (Fig. 1B–D, peripheral blood not shown). In addition, T cells accounted for ∼90% of IL-10 production (Fig. 2A). Thus, IL-10 is produced by T cells specifically at the site of infection during peak inflammation.

During the peak of J2.2-V-1 infection, more CD8 T cells than CD4 T cells infiltrate the CNS (24, 25). Therefore, although a greater fraction of CD4 than CD8 T cells expressed IL-10, ∼50% of the IL-10–producing cells were CD8⁺ (Fig. 2B, 2C). The frequency of IL-10⁺CD8 T cells diminished from 40% at day 7 p.i. to <5% at day 21 p.i., whereas the frequency of IL-10⁺CD4 T cells declined more slowly over the course of infection (Fig. 2C). Consistent with the numbers of IL-10–producing T cells, IL-10 protein levels peaked in the brain at day 7 p.i. and diminished thereafter (Fig. 2D). Of note, at all time points, the mean fluorescence intensity (MFI) of GFP expression was greater in CD4 compared with CD8 T cells, demonstrating greater IL-10 production on a per cell basis (Fig. 2E) (17).

Because IL-10 production is generally indicative of a T cell with suppressor function (26), we next determined whether the IL-10⁺CD8 T cells were as potent effectors as IL-10⁻CD8 T cells. Using MHC class I/peptide tetramer staining, we observed that approximately half of the GFP⁺ cells were specific for S510, the immunodominant CD8 T cell epitope recognized in J2.2-V-1–infected B6 mice (Fig. 3A) (23, 25). In addition, we examined IFN-γ and TNF expression by IL-10⁺CD8 T cells, using intracellular cytokine staining after stimulation with peptides corresponding to S510 and the subdominant epitope, S598. In these assays, we identified IL-10⁺ cells by intracellular cytokine staining because GFP diffused from cells under conditions necessary for IFN-γ and TNF detection. Nearly all of the virus epitope-specific IL-10⁺CD8 T cells expressed IFN-γ and 60–70% also produced TNF (Fig. 3C). Furthermore, cells recognizing the two peptides accounted for almost all of the IL-10 produced by CD8 T cells (Fig. 3A, 3C). Interestingly, IFN-γ and TNF were expressed by virus-specific IL-10⁺CD8 T cells at greater levels per cell than IL-10⁻CD8 T cells (MFI for S510-specific CD8 T cells: IFN-γ⁺ cells: 418 ± 8 versus 217 ± 4, p = 0.0001; TNF⁺ cells: 305 ± 23 versus 169 ± 11; p = 0.0007 for IL-10⁺ and IL-10⁻ cells, respectively) consistent with a more highly activated phenotype.

Most IL-10⁺CD4 T cells in the brains of J2.2-V-1–infected mice were effector Th1 T cells, with the majority coproducing IFN-γ and TNF when stimulated by immunodominant (M133) and subdominant (S358) peptides (Fig. 3A, 3B). Of note, a lower percentage of M133-specific CD4 T cells were identified using tetramer than by intracellular cytokine expression, consistent with low affinity of TCR to MHC class II/peptide binding (27). Approximately 10–15% of IL-10⁺CD4 T cells expressed Foxp3, expressed by regulatory T cells, whereas 85–90% expressed T-bet, consistent with a Th1 phenotype (Fig. 3D). J2.2-V-1 infection induces a strong Th1 cytokine response (21, 28, 29). Only a very low number of cells in the J2.2-V-1–infected brain expressed IL-4 or IL-17; consequently, few, if any, of the IL-10⁺ cells also expressed either of these cytokines (Fig. 3D).

High doses of stimulatory Ag are required for IL-10 compared with IFN-γ production by CD4 T cells (6). CD8 T cells produced IL-10 at the site of infection and production was maximal at the peak of the inflammatory response (Fig. 1). Furthermore, a greater frequency of IL-10⁺CD8 T cells expressed CD69, a marker of recent antigenic stimulation, compared with IL-10⁻CD8 T cells (Fig. 4A). In addition, CD127 expression was slightly higher and KLRG-1 expression modestly lower on IL-10⁺CD8 T cells, suggestive of a memory progenitor phenotype (Fig. 4A) (30). This raised the possibility that IL-10 production in CD8 T cells was also dependent on strong Ag stimulation. To assess this possibility, we examined IL-10 production by stimulating GFP⁺CD8 T cells with irradiated splenocytes pulsed with varying S510 peptide doses directly ex vivo for 48 h. Indeed, ∼100-fold more peptide was required to produce 50% maximal levels of IL-10 compared with IFN-γ (Fig. 4B).

Strong signaling through the TCR activates the MAPK pathway, consisting of ERKs (ERK1/2), JUN N-terminal kinases, and p38 (4, 31). Because IL-10 production was dependent on stimulation with high Ag concentrations, we next examined whether the MAPK pathway was activated under conditions of high Ag stimulation. As shown in Fig. 4C, the level of p-ERK was greater in IL-10⁺CD8 T cells than in IL-10⁻CD8 T cells when cells were stained directly ex vivo for p-ERK1 and p-ERK2 (p44/42).

To demonstrate a functional role for ERK, we stimulated GFP⁺CD8 T cells with irradiated splenocytes pulsed with S510 peptide in the presence of varying concentrations of the ERK inhibitor PD184161 for 48 h in an in vitro assay. PD184161 is a potent, selective inhibitor of MEK1/2, the upstream kinases of ERK1/2 (32). As seen in Fig. 4D, selective inhibition of ERK phosphorylation decreased IL-10 production by ∼50%. Another MAPK, p38, was also involved in IL-10 expression because stimulation of GFP⁺CD8 T cells with S510 peptide in the presence of a selective p38 inhibitor (PD169316) (33, 34) decreased IL-10 production by 75% (Fig. 4E). Decreased amounts of IL-10 did not reflect a loss of cell viability because only low levels of apoptosis (∼10%), as assessed by Annexin V and 7-AAD staining, were detected in all cultures (data not shown). Therefore, inhibition of either ERK1/2 and p38 MAPK pathways decreased IL-10 expression by CD8 T cells.

Unexpectedly, when we stimulated GFP⁻CD8 cells for 48 h with high concentrations of peptide S510, we detected IL-10 in the cultures, suggesting that the sorted GFP⁻CD8 T cells had converted to IL-10–producing cells after high antigenic stimulation (Fig. 4F). Consistent with this result, some of the cells that were initially GFP⁻ expressed GFP after in vitro culture (data not shown). IL-10 expression did not occur in the absence of peptide S510 or in cultures without sorted CD8 T cells (Fig. 4F).

To investigate whether this phenotypic plasticity also occurred in the infected animal, we transferred Thy1.2 GFP⁺ T cells or Thy1.2 GFP⁻ T cells into J2.2-V-1–infected Thy1.1 mice. Mice were sacrificed 6 d after transfer, and Thy1.2 CD8 and CD4 T cells were examined for GFP expression. Mice were analyzed at day 7 p.i. to allow sufficient time for proliferation of transferred cells. The highest frequency of transferred cells, whether initially GFP⁺ or GFP⁻, was detected in the infected brain compared with the CLN or spleen (Fig. 5B, 5C). In addition, transferred GFP⁺ and GFP⁻ T cells were detected in the brain at similar frequencies, suggesting that IL-10⁺ cells were not more prone to cell death. Remarkably, IL-10 production was transient in both CD8 and CD4 T cells, so that some cells that were initially IL-10⁺ (GFP⁺) became GFP⁻ and some of the IL-10⁻ (GFP⁻) T cells became GFP⁺ (Fig. 5A, 5D). Of note, T cells that were originally IL-10⁺ or IL-10⁻ appeared to proliferate equally, even though IL-10 inhibited proliferation in vitro. This equivalent proliferative ability may reflect IL-10 plasticity, because similar frequencies of cells were IL-10⁻ after 6 d in mice.

Because J2.2-V-1–specific IL-10⁺CD8 T cells coproduced IFN-γ and TNF, we assessed whether the IL-10 produced by these cells was immunosuppressive or, alternatively, immunostimulatory (35, 36) in an in vitro proliferation assay. Using a transwell assay system, we placed GFP⁺CD8 T cells in the upper compartment with peptide-stimulated irradiated splenocytes. CFSE-labeled naive lymphocytes were incubated in the lower compartment of an anti-CD3–coated plate. As expected, naive CD4 and CD8 T lymphocytes proliferated in the presence of anti-CD3 stimulation as demonstrated by CFSE dilution (Fig. 6). Proliferation was inhibited if GFP⁺CD8 T cells were present in the culture. Inhibition was mediated by IL-10, because the presence of a blocking IL-10 Ab largely reversed the inhibitory effect.

IL-10⁺CD8 T cells were most abundant during the peak of inflammation (Fig. 1B) and coproduced inflammatory cytokines (Fig. 3C), and IL-10 expression was dependent on MAPK activation and high Ag load (Fig. 4). These results, coupled with data showing that these cells were immunosuppressive (Fig. 6), raised the possibility that the IL-10⁺CD8 T cells were more highly activated than IL-10⁻CD8 T cells. To evaluate this possibility, we compared patterns of global transcription in IL-10⁺ and IL-10⁻ CD8 T cells by microarray analyses.

For this purpose, CNS-derived CD8 T lymphocytes were pooled from three mice and sorted for recent IL-10 production as detected by GFP expression to high purity (>99%) at day 7 after J2.2-V-1 infection. Genome-wide microarray studies were performed on mRNA purified from three pools of lymphocytes. We performed statistics on the gene expression data and found 384 genes statistically significantly increased and 25 genes decreased in IL-10⁺ cells compared with IL-10⁻ cells (Supplemental Table I). Using DAVID and Ingenuity Pathways Analysis to evaluate the gene sets identified, we detected significant changes in several pathways of immune activation and regulation (Table I, Supplemental Table I). Notably, there was increased expression of genes encoding cytokines (Il1a, Il1b, Il10, Il18, Tnf), chemokines (Ccl2, Ccl3, Ccl6, Ccl7, Ccl8, Ccl9, Ccl12, Cxcl2, Cxcl4 Cxcl6, Cxcl9), immune sensor molecules (Tlr1, Tlr2, Tlr3, Tlr4, Tlr6, Tlr8, Tlr13, Fcer1g, Fcgr1), and lysosomes/endocytosis (Abca1 [ATPeV], Gla, Hexa/b, Lamp1/2 [CD107a/b], Acp2, Prf1, Psap, Lgmn, ctsb/c/h/l/z [cathepsin B/C/H/L/Z]), suggesting that cells expressing IL-10 also exhibited a highly activated proinflammatory transcriptional profile.

To validate these results, we analyzed a subset of these differentially regulated genes by qRT-PCR and for protein expression. As shown in Fig. 7A, IL-1α, IL-1β, IL-10, CCL2, TLR2, TLR3, TLR4, prosaposin, hexosaminidase A, and perforin mRNA were increased and STAT-4 mRNA was decreased in IL-10⁺CD8 compared with IL-10⁻CD8 T cells, confirming our microarray analysis. As shown in Fig. 4A, some IL-10⁻CD8 T cells were CD69⁺, indicating recent Ag exposure. To confirm that these cells were not hyperactivated, we sorted CD69⁺IL-10⁻, CD69⁺IL-10⁺, and CD69⁻IL-10⁻ CD8 T cells, and compared expression levels of a subset of genes shown in Fig. 7A. CD69⁺IL-10⁺CD8 T cells had increased IL-10, IL-1α, CCL2, TLR3, and perforin, and decreased STAT-4 mRNA levels compared with CD69⁺IL-10⁻CD8 T cells (Fig. 7B). In addition, we investigated protein expression in IL-10⁺ cells by surface staining for MHC class II and the high-affinity IgE receptor (FcεR1), and by intracellular staining for LAMP1/2 (CD107a/b) and granzyme B after stimulation with peptide S510. LAMP1/2, granzyme B, MHC class II, and FcεR1 were all increased in epitope S510-specific IL-10⁺CD8 compared with IL-10⁻CD8 T cells (Fig. 7C), confirming our microarray data. Most importantly, IL-10⁺CD8 T cells were more cytotoxic than IL-10⁻CD8 T cells when examined directly ex vivo for the ability to lyse peptide S510-coated EL4 target cells (Fig. 7D).

Demyelination in J2.2-V-1–infected mice is largely immunopathological, with myelin damage occurring during virus clearance (20). The observation that IL-10⁺CD8 T cells were highly activated and cytolytic raised the question whether the IL-10 expressed by these cells would be able to diminish immunopathological changes in the J2.2-V-1–infected CNS.

First, we confirmed a role for IL-10 in J2.2-V-1–infected mice by monitoring survival, weight, and demyelination in age- and sex-matched IL-10^+/+ and IL-10^−/− B6 mice infected with 1500 PFU J2.2-V-1. Infected IL-10^−/− mice died at a greater rate than IL-10^+/+ mice and had increased weight loss (Fig. 8A; days 7–12 p.i.; p ≤ 0.05). We detected more rapid virus clearance in IL-10^−/− mice, showing that the mice had increased ability to control the infection. Diminished IL-10 expression is often associated with increased numbers of inflammatory cells present at sites of inflammation (2). Consistent with these results, we found more innate immune effector cells (macrophages, neutrophils, and NK cells) at day 4 p.i. and increased virus-specific CD4 and CD8 T cells at day 8 p.i. (Fig. 9). Thus, increased cellular infiltration in the CNS of IL-10^−/− mice enhanced viral clearance, but at the cost of diminishing survival. To assess the role of IL-10 in demyelination, we examined infected spinal cords for myelin destruction at day 21 p.i., the peak of demyelination (37). For these experiments, we infected mice with 500 PFU J2.2-V-1 to enhance survival (Fig. 8B). Demyelination, as assessed by Luxol fast blue staining, was significantly increased in IL-10^−/− mice when compared with B6 mice (Fig. 8B).

To directly examine the individual contribution of IL-10 expressed by CD8 T cells, we adoptively transferred virus-immune undepleted or CD8 T cell-enriched splenocytes from IL-10^+/+ or IL-10^−/− mice to IL-10^−/− mice 1 d p.i. with 1500 PFU J2.2-V-1. Similar numbers of epitope S510-specific T cells were present in the transferred IL-10^+/+ and IL-10^−/− splenocytes (data not shown). Although there was no difference in survival between mice receiving CD8 T cell-enriched splenocytes from B6 or IL-10^−/− mice, there was diminished weight loss and demyelination in mice receiving IL-10–replete compared with IL-10^−/− CD8 T cells (Fig. 8C, 8D; days 9–12 p.i.; p ≤ 0.05). Thus, IL-10 expressed by CD8 T cells was protective in the context of infection with J2.2-V-1.

In this study, we demonstrate that virus-specific CD8 T cells that are most cytotoxic and hyperactivated in mice with acute encephalitis express IL-10 during the height of the inflammatory response. One interpretation of these results is that IL-10 functions in a self-suppressing or localized manner by tempering the immune response induced by these highly activated CD8 T cells. In this scenario, CD8 T cells would self-regulate when viral Ag concentration (Fig. 4B) and inflammatory response reach a certain level, to minimize immunopathological disease (e.g., demyelination) (Fig. 8). Because only the most activated virus-specific cells express IL-10, immunosuppression may not be generalized, but localized to sites where bystander damage would be maximal. IL-10 expression by CD8 T cells was also detected during the peak of inflammation in mice infected with influenza A virus or SV5 (10, 13). Although there is not agreement in all studies (12), inflammation is increased and clinical disease worsens in the absence of IL-10 in both infections. Furthermore, we show that IL-10 expression solely by CD8 T cells was protective in an adoptive transfer system, even though expression was highly elevated for only a short period (Figs. 1, 8). In addition, expression of IL-10 by these highly activated CD8 cells was dynamic, as shown by the high rate of interconversion of IL-10⁻ and IL-10⁺ populations in vitro and in vivo after adoptive transfer. This plasticity may reflect recent exposure to virus Ag because the IL-10⁺CD8 T cells preferentially expressed CD69 (Fig. 4A). Variable IL-10 expression suggests that exposure to Ag is transitory, possibly reflecting focal distribution of virus Ag in the brain (38). Furthermore, virus is cleared from the brain in a rostral to caudal manner (39, 40), with clearance of inflammatory cells lagging behind that virus (S. Perlman, unpublished observations); cells remaining at sites of virus clearance would have diminished viral Ag exposure resulting in downregulation of IL-10 production. The high activation state of these cells was not a precursor to cell death, because IL-10⁺ cells could be detected for at least 6 d after adoptive transfer.

Until recently, IL-10 has been considered to be most important in chronic infections and autoimmune disease. In animals persistently infected with lymphocytic choriomeningitis virus or Leishmania or HCV-infected humans, IL-10 produced by CD4 T cells has been implicated in pathogen persistence (7–9, 41). In these infections, IL-10 was produced by CD4 and not CD8 T cells; this is consistent with our results because IL-10 expression by CD8 T cells rapidly diminishes after day 7 p.i. and is nearly at background levels by 21 d p.i. in the brain, whereas CD4 T cells continue to express IL-10 for at least 42 d p.i. (Fig. 1).

To our knowledge, this is the first report to demonstrate that IL-10 expression in CD8 T cells is dependent on high Ag stimulation and signaling through the ERK1/2 and p38 MAPK pathways (Fig. 4C–E). p38, but not ERK1/2 signaling, was found to be important in IL-10 production in monocytes (42). Conversely, ERK1/2 signaling but not p38 is required for IL-10 production in all CD4 Th subsets (6, 31). Therefore, IL-10 expression is dependent on MAPK pathways in all tested cells, but exact requirements are cell specific. Although we did not examine the role of JNK of the MAPK pathway in IL-10 expression, its signaling may also be needed for maximal IL-10 production as IL-10⁺ cells showed increased jun expression as assessed by gene expression array (Supplemental Table I).

Our results showed that IL-10⁺CD8 T cells were more cytolytic than IL-10⁻CD8 T cells (Fig. 7D). Consistent with these data, genes involved in cytolytic function were increased in IL-10⁺CD8 compared with IL-10⁻CD8 T cells in gene expression analyses (Fig. 7, Table I). Other upregulated genes in IL-10⁺CD8 T cells included several that are involved in the innate immune response and are most commonly expressed by monocytes and macrophages. Some of these proteins, such as the chemokines CCL2, CCL8, CCL12, and CXCL4 (platelet factor 4) and cytokines IL-1α, IL-1β, and IL-18 are predicted to enhance the inflammatory response. However, further work will be required to understand why some genes, such as TLRs, C-type lectins, and FcRs, are increased in expression in IL-10⁺CD8 T cells.

Equally interesting is diminished expression of STAT-4 in IL-10⁺CD8 T cells; STAT-4 is required for maximal IL-12–induced IL-10 expression in CD4 T cells (6). CD8 T cell expression of IL-10 is likely to be STAT-4 and thus, IL-12, independent because we found the same frequency of IL-10⁺CD8 T cells in the brains of IL-12Rβ2^−/− and wild type B6 mice (data not shown). Recently, the transcription factors c-Maf and Ahr (aryl hydrocarbon receptor) were demonstrated to be essential for IL-10 expression by Tr1 CD4 T cells and c-Maf for IL-10 production in macrophages (43, 44). The expression of these transcription factors was not increased in IL-10⁺CD8 T cells, possibly because they may be upregulated only during the early induction phase. Alternatively, different pathways of IL-10 induction may exist in CD4 and CD8 T cells.

In summary, IL-10 is produced by the most highly activated, cytotoxic CD8 T cells in the infected CNS at the peak of inflammation and is protective against morbidity. This study is consistent with the notion that cytotoxic CD8 T cells self-regulate themselves by producing IL-10, and thus that enhanced activation does not invariably result in the development of cells with greater potential for immunopathological disease. Manipulation of IL-10 expression by CD8 T cells might afford another approach to diminishing tissue destruction in animals or humans with acute encephalitis.

We thank Drs. John Harty, Hai-Hui Xue, and Jon Houtman for stimulating discussion and critical review of the manuscript. We thank Dr. Tom Bair (University of Iowa DNA facility) for help with gene expression analysis and the National Institutes of Health Tetramer Core Facility (Atlanta, GA) for providing tetramers.

The authors have no financial conflicts of interest.