γδ T Cell Regulation of IFN-γ Production by Central Nervous System-Infiltrating Encephalitogenic T Cells: Correlation with Recovery from Experimental Autoimmune Encephalomyelitis¹

The γδ T cells have emerged as a population of lymphocytes capable of regulating the immune response under a variety of inflammatory stimuli, including autoimmunity (1, 2, 3). Although the focus of much study, the nature of their physiological ligands and the exact function of γδ T cells in many disease states still remain enigmatic. They have been shown to exert effects throughout an immune response, able to both reduce or exacerbate inflammatory damage, depending on the experimental disease model examined (1, 2, 3). γδ T cells as a population are Th1-like, with the capacity to produce high levels of the proinflammatory cytokine IFN-γ following activation (4). Thus, γδ T cells have the capacity to influence the nature of the immune response by providing IFN-γ required for the differentiation of naive CD4 T cells into Th1 effector cells and by promoting inflammation through the production of IFN-γ required for the activation of macrophages. In the autoimmune disease experimental autoimmune encephalomyelitis (EAE),³ a role for γδ T cells in the regulation of the autoimmune response has been indicated (5, 6, 7, 8).

EAE is the animal model of the human autoimmune CNS-demyelinating disease multiple sclerosis (MS) (9, 10). EAE is primarily an inflammatory disease characterized by demyelination and the accumulation of cellular infiltrates in the CNS containing macrophages and αβ T cells, B cells, and γδ T cells (11). In addition, the production of inflammatory cytokines, including IFN-γ, is observed in the CNS at the time of disease onset (9). IFN-γ-producing Th1 T cells with specificity for a variety of self Ags, including myelin basic protein (MBP), have been shown to induce EAE. EAE can be induced in B10.PL mice (H-2^u) by either immunization with whole MBP or its NH₂-terminal immunodominant peptide or by the adoptive transfer of MBP-specific Th1 T cells (12). The nature of the EAE disease in the B10.PL mouse is typically an acute monophasic disease course, with most mice completely recovering from disease symptoms in the absence of relapse (12).

Although it is known that EAE induction requires T cells of the Th1 phenotype, which produce IFN-γ, the exact role of IFN-γ in EAE or MS pathogenesis is not known. In MS patients, the administration of IFN-γ resulted in exacerbations of disease (13). A detrimental role for IFN-γ in MS is further supported by two separate studies in which IFN-γ was constitutively expressed in the CNS by oligodendrocytes. In the first study by Horwitz et al. (14), unmanipulated mice showed signs of primary demyelination and hallmarks of CNS immune-mediated disease. In the second study by Renno et al. (15), unmanipulated mice showed no signs of disease before EAE induction, but were unable to recover from disease. In contrast, a protective role for IFN-γ has also been suggested in EAE as mice deficient in either IFN-γ or the IFN-γR exhibited a chronic EAE disease course (16, 17, 18). The mechanism of the chronic disease has been attributed to a lack of macrophage activation via IFN-γ, resulting in dysregulation of NO production (19), which may suppress the expansion of activated CD4 T cells (20). IFN-γ has also been shown to act on T cell proliferation and to direct the production of chemokines in the CNS, thus potentially playing a role in the onset and progression of disease (18). The conclusion from these cumulative studies is that IFN-γ is likely to have multiple functions during various stages of the EAE or MS disease course. In the studies in which IFN-γ was shown to be required for disease recovery, the cell that produces the relevant source of IFN-γ is not known and may include encephalitogenic T cells, CD8 T cells, and possibly γδ T cells.

γδ T cells have been found in both the brain (21, 22, 23) and CSF (24, 25) of MS patients, and have been suggested to be involved in the process of demyelination, as they have been shown to be cytotoxic to oligodendrocytes in vitro (26). A specific role for γδ T cells in MS is suggested by the limited diversity of the TCRγδ isolated from active MS lesions (27). γδ T cells are also found in the CNS of mice with EAE, which have been shown to exhibit an activated phenotype (28, 29). Little is known about the diversity of the TCRγδ in the CNS during EAE.

The role of γδ T cells in the regulation of EAE has not been elucidated because contradictory results have been reported by a number of studies that used different rodent strains and different strategies to deplete γδ T cells. Two studies reported a reduced severity of EAE. The first used MBP-immunized SJL/J mice that were depleted of γδ T cells with a mAb (7), and a second used C57BL/6 TCRδ knockout mice in which EAE was induced by both immunization and passive transfer (8). Two additional studies observed aggravation of disease: one used Lewis rats (6), and the second used spinal cord homogenate-immunized B10.PL mice depleted of γδ T cells by anti-TCR (5). In addition, a fifth study did not observe any alteration in EAE disease induced by passive transfer in TCRδ knockout C57BL/6 mice (30). Thus, these studies cumulatively suggest that γδ T cells may regulate EAE at multiple levels, able to affect both the initiation and resolution of EAE.

In the present study, we asked whether γδ T cells could control/regulate CNS inflammation associated with EAE by promoting the production of IFN-γ. We found that B10.PL mice deficient in γδ T cells have reduced levels of both IFN-γ message and protein in the CNS before the onset of EAE, as compared with control mice. In addition, the regulation was shown to be CNS specific, as no alteration in IFN-γ production was observed between control and γδ T cell-deficient mice in the spleen or lymph nodes during EAE. The reduction in IFN-γ production was observed not only in the encephalitogenic T cells, but in all IFN-γ-producing cells as well. In control mice, the only γδ T cells detectable in the CNS before disease onset expressed the ADV1.1 or DV105s1 TCRδ chain and exhibited a restricted phenotype as compared with intestinal intraepithelial lymphocyte (IEL) γδ T cells. We also found that γδ T cell-deficient mice were unable to recover from EAE, exhibiting a long-term chronic disease course, similar to that of IFN-γ-deficient mice. By generating mixed bone marrow (BM) chimeras, we were able to determine that γδ T cell regulation of IFN-γ in the CNS is independent of IFN-γ production by the γδ T cells. In addition, we show that γδ T cell-deficient mice do not have a systemic defect in the production of IFN-γ. These data suggest that a function of γδ T cells during CNS autoimmunity is to regulate the production of IFN-γ by T cells early in the EAE disease course, which is necessary for the recovery from EAE disease symptoms.

B10.PL (H-2^u), B6.129P2-Tcrd^tm1Mom, and B6.129S7-Ifng^tm1Ts mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The MBP-TCR transgenic (tg) mice expressing a TCR transgene specific for the acetylated NH₂-terminal peptide of MBP (Ac1–11) were generated, as previously described (31). B10.PL-TCRδ^−/− (TCRδ^−/−) and B10.PL-IFN-γ^−/− (IFN-γ^−/−) mice were produced in our breeding colony by backcrossing B6.129P2-Tcrd^tm1Mom and B6.129S7-Ifng^tm1Ts mice onto B10.PL for three generations, respectively, and then intercrossing to generate homozygous mice carrying the indicated gene disruptions.

The Ac1–11 MBP peptide (Ac-ASQKRPSQRSK) was generated, as described (12). The anti-mouse Abs specific for CD4, CD8, CD11b, and Vβ8.2 were purchased from eBioscience (San Diego, CA). Anti-mouse IFN-γ was purchased from BD Biosciences (San Diego, CA).

EAE was induced by the adoptive transfer of MBP-specific encephalitogenic T cells, generated as described, in which over 98% of the cells expressed TCRβ (12). In addition, these T cell lines do not stain positive for the TCRγδ (data not shown). Briefly, 1 × 10⁶ activated MBP-TCR T cells were i.v. injected into sublethally irradiated (360 rad) 5- to 8-wk-old age- and sex-matched B10.PL and TCRδ^−/− mice. Individual animals were assessed daily for symptoms of EAE and scored using a scale from 1 to 5, as follows: 0, no disease; 1, limp tail and/or hind limb ataxia; 2, hind limb paresis; 3, hind limb paralysis; 4, hind and fore limb paralysis; and 5, death.

Mononuclear cells were isolated from the CNS of B10.PL or TCRδ^−/− mice with EAE following intracardial perfusion with 25–30 ml of cold PBS. The brains and spinal cords were homogenized, and mononuclear cells were isolated using 40/70% discontinuous Percoll gradients. Total cell numbers were determined by counting on a hemocytometer, and viability was assessed by trypan blue exclusion. FcR were blocked with anti-mouse FcR (2.4G2), followed by staining with the following Ab combinations: 1) anti-CD4 FITC with and without anti-Vβ8.2 biotin, followed by streptavidin-allophycocyanin-Cy7, and with and without CD11b-PE-Cy5; or 2) anti-CD8 FITC with and without CD11b-PE-Cy5. The cells were then fixed and permeabilized using the Cytofix/Cytoperm kit (BD Pharmingen, San Diego, CA), and stained with anti-IFN-γ PE or a PE-conjugated isotype-matched control (eBioscience). Ab incubations were conducted on ice, and the cells were fixed in 1% paraformaldehyde and analyzed using a FACScan (BD Biosciences).

Total RNA was isolated from the spinal cords of PBS-perfused B10.PL and TCRδ^−/− mice on the indicated days following EAE induction and from IEL using TRIzol (Invitrogen Life Technologies, Carlsbad, CA). IEL were isolated, as previously described, with modifications (32). Briefly, the small intestine was flushed with PBS, and after removal of the Peyer’s patches, was divided longitudinally and cut into 2- to 3-cm sections. The sections were rinsed with PBS and incubated on a stirring platform for 20 min in PBS without Ca²⁺ and Mg²⁺ and containing 2% FCS and 1 mM EDTA. Supernatant containing epithelium and IEL was collected, and tissue debris and cell aggregates were removed by passage through 75-μm nylon mesh. The lymphocytes were isolated by centrifugation on a discontinuous 40/70% Percoll gradient. Genomic DNA was removed from RNA samples using the DNA-free kit (Ambion, Austin, TX). cDNA was synthesized, as previously described (33), using Superscript II reverse transcriptase (Invitrogen Life Technologies).

IFN-γ mRNA was quantitated by real-time PCR using SYBR Green as the detection agent. The PCR was performed with the iCycler iQ (Bio-Rad, Hercules, CA). All components of the PCR mix were purchased from Bio-Rad and used according to manufacturer instructions. Cycler conditions were one amplification cycle of denaturation at 95°C for 5 min, followed by 40 cycles of 95°C for 1 min, 60°C for 1 min, and 72°C for 1 min. Specificity of the RT-PCR was controlled by the generation of melting curves. Threshold values of IFN-γ expression were normalized to GAPDH expression using standard curves generated for each sample by a series of four consecutive 10-fold dilutions (1–1 × 10³) of the cDNA template. For all reactions, each condition was performed in triplicate, PCR efficiencies were 95–100%, and correlation coefficients were 0.97–0.99. The data were analyzed using iQ Cycler analyzing software.

The sense GAPDH primer was TTCACCACCATGGAGAAGGC; the antisense primer was GGCATGGACTGTGGTCATGA. The sense primer for IFN-γ was TCAAGTGGCATAGATGTGGAAGAA; antisense primer was TGGCTGTGCAGGATTTTCATG.

Mixed BM chimeras were generated by transferring 4 × 10⁶ total BM cells from B10.PL, IFN-γ^−/−, or TCRδ^−/− mice into sublethally irradiated (360 rad) age- and sex-matched TCRδ^−/− or B10.PL mice and allowed to reconstitute for 4–6 wk. EAE was induced in chimera mice at 4–6 wk post-BM transplantation by the adoptive transfer of 0.5 × 10⁶ MBP T cells into sublethally irradiated (360 rad) animals. IFN-γ mRNA expression in the spinal cord, spleen, and cervical lymph nodes was analyzed by real-time PCR on days 10 and 15 after EAE induction.

On days 7, 14, and 21 after EAE induction, the spinal cords were removed from perfused B10.PL and TCRδ^−/− mice and fixed in paraformaldehyde-lysine-periodate. After 24 h, the tissues were sucrose infused, embedded in Tissue Tek OCT, and frozen in isopentane. For immunohistochemistry, frozen sections 7 μm thick were blocked with newborn calf serum and stained with biotinylated anti-CD11b, followed by staining with streptavidin-alkaline phosphatase. The color was developed using the HistoMark Red Phosphatase System (Kirkegaard & Perry Laboratories, Gaithersburg, MD) and counterstained with hematoxylin.

Groups of two to three B10.PL and TCRδ^−/− mice were immunized with 100 μg of emulsified CFA in the footpad. After 7 days, the popliteal lymph nodes were removed, total RNA was isolated, and IFN-γ message was determined by real-time PCR. For analysis of IFN-γ production in the spleen by Ag-specific T cells, MBP-specific splenic T cells isolated from GFP MBP-TCR tg mice were activated in vitro by coculture with irradiated B10.PL splenocytes in the presence of 5 μg/ml Ac1–11. After 3 days, 10 × 10⁶ activated GFP⁺ MBP-TCR tg T cells were i.v. injected into B10.PL and TCRδ^−/− recipient mice. At the same time, the mice were i.v. administered 15 μg of Ac1–11 diluted in 200 μl of PBS. Three days later, total splenocytes were isolated and incubated with 5 μg/ml Ac1–11 in the presence of GolgiStop (BD Biosciences) for 6 h, followed by staining with anti-IFN-γ PE. Flow cytometry analysis was performed assessing IFN-γ expression in GFP-gated cells.

EAE was induced in B10.PL mice by adoptive transfer, and spectratype analysis of TCRδ V gene usage was done using total RNA isolated from spinal cord samples on the indicated days. Rearrangement analysis was performed by PCR amplification of the TCR CDR3 using V and TCRδ C region-specific primers. The V region primer specific for the DV7s6 family was 5′-CAACCAGACGATTCGGGAAAG-3′; the V region primer specific for DV105s1 was 5′-CCTTCCATCTGGTGATCTCTC-3′; and the TCRδ C region primer was 5′-CAGCCTCCGGCCAAACCATC-3′. The C region primer was labeled with FAM, and the PCR products were analyzed on denaturing polyacrylamide gels, and the fluorescent PCR products were visualized using a FluorImager (Molecular Dynamics, Sunnydale, CA). The analyses were performed using 2 μl of cDNA generated from individual spinal cords in 40 μl of PCR together with 30 pmol of each Vδ-specific primer and the Cδ-FAM-labeled primer for 40 cycles. The spectratype technique has been previously described in detail (34). All of the TCRδ gene primers are based on published sequences (35).

Although an important role for IFN-γ production in the resolution of EAE has been demonstrated, the cellular source of the IFN-γ has not been elucidated. In EAE, there are a number of T cells in the CNS capable of producing IFN-γ that include encephalitogenic CD4 T cells, nonencephalitogenic CD4 T cells, CD8 T cells, and γδ T cells. Because γδ T cells have the capacity to both produce and regulate IFN-γ production, we first asked whether mice deficient in γδ T cells have altered levels of IFN-γ production in the CNS during EAE. To investigate this possibility, we generated γδ T cell-deficient B10.PL mice (TCRδ^−/−) by crossing B10.PL with C57BL/6 mice carrying a disrupted TCRδ gene. EAE was induced by passive induction by the adoptive transfer of encephalitogenic T cells specific for the NH₂-terminal peptide of MBP (Ac1–11) into sublethally irradiated B10.PL and TCRδ^−/− mice, and we used real-time PCR to quantitate the levels of IFN-γ message in the spinal cord of B10.PL and TCRδ^−/− mice on days 7, 15, 22, and 36 after EAE induction. IFN-γ message was quantitated in relation to expression of the housekeeping gene GAPDH. After EAE induction in B10.PL mice, the expression of IFN-γ message in the spinal cord was dramatically increased on day 7, just before the onset of disease symptoms. IFN-γ levels then decreased steadily during the effector stage of disease, reaching a minimal level by day 22 during the recovery phase (Fig. 1,A). Day 15 is the peak of disease, with the B10.PL mice having a disease score of 2, which was decreased to 0.5 on day 22, indicating recovery from EAE (see Fig. 3). IFN-γ expression was still detectable in the CNS in recovered animals on day 36 (Fig. 1,A), even though no EAE clinical symptoms were observed. In comparison, the level of IFN-γ in TCRδ^−/− mice also increased following EAE induction, but the expression level was 10-fold lower on day 7 as compared with B10.PL mice and did not peak until day 15 (Fig. 1,A). The level of IFN-γ in the CNS was similar in both groups of mice on days 22 and 36 (Fig. 1,A). The EAE disease scores for the TCRδ^−/− mice for the day 7, 15, 22, and 36 time points were 0, 2, 2, and 1.5, respectively. The inability of the TCRδ^−/− mice to recover from EAE will be shown in Fig. 3 A. In both B10.PL and TCRδ^−/− mice, message for IFN-γ in the spinal cord of unmanipulated mice is at the detectable range for real-time PCR (10⁻⁶ to 10⁻⁷, as compared with GAPDH) (data not shown).

To determine whether the decreased expression of IFN-γ mRNA correlated with a subsequent decrease in the production of IFN-γ protein in the CNS, we performed intracellular cytokine staining using flow cytometry to examine IFN-γ production by mononuclear cells isolated directly from the CNS of B10.PL and TCRδ^−/− mice with EAE. Because EAE was induced by CD4⁺ encephalitogenic T cells, we first examined IFN-γ production in the CD4⁺ cell population, and found that the percentage of CD4⁺ cells producing IFN-γ was statistically reduced (p < 0.04) from 19 ± 4 in B10.PL mice to 8 ± 1 in TCRδ^−/− mice 10 days after the induction of EAE (Table I). In addition to a reduction in the percentage of CD4⁺ cells producing IFN-γ, we also found that the absolute number of these cells was reduced in the CNS of TCRδ^−/− mice. This is shown in Table I, in which the absolute number of CD4⁺IFN-γ⁺ cells was 12 × 10³ cells in B10.PL mice as compared with 2.6 × 10³ cells in TCRδ^−/− mice. This reduction was statistically significantly reduced (p < 0.05). To determine whether the reduction in CD4⁺ IFN-γ-producing cells in the CNS of TCRδ^−/− mice was due to reduced numbers of encephalitogenic T cells, we analyzed the CD4⁺ cell population for the expression of Vβ8.2, which is the TCRβ chain expressed by the tg encephalitogenic T cells (31). We found that in B10.PL mice, 64% of the CD4⁺ T cells expressed Vβ8.2, while the expression level was 69% in TCRδ^−/− mice (Table I), indicating that a deficiency in γδ T cells did not alter the migration pattern of encephalitogenic T cells into the CNS.

The data presented in Table I indicate that γδ T cells regulate the expression of IFN-γ by CD4⁺ encephalitogenic T cells in the CNS of mice with EAE. However, other T cells are present in the CNS of mice with EAE that have the capacity to produce IFN-γ. Thus, we next asked whether γδ T cell regulation of IFN-γ production in the CNS was specific for encephalitogenic T cells or was a global mechanism affecting other IFN-γ-producing cells. To facilitate this analysis, we conducted a similar study as outlined in Table I, but instead of gating on CD4⁺ T cells, we gated out the nonlymphoid cells by the expression of CD11b before analysis of IFN-γ production. This was required, as lymphoid light scatter gating cannot eliminate all cells of myeloid origin, and these cells have a high autofluorescence that interferes with the analysis of IFN-γ production. We found that on day 10 following EAE induction, 31% of the lymphocytes isolated from the CNS of B10.PL mice produced IFN-γ (Fig. 1,B). In contrast, only 4% of CNS-isolated lymphocytes expressed IFN-γ in TCRδ^−/− mice (Fig. 1,E). Because both CD4 and CD8 T cells with the capacity to produce IFN-γ are present in the CNS during EAE, we examined both T cell populations for the production of IFN-γ. In this study, we found that 36% of the CD4⁺ cells produced IFN-γ in B10.PL mice (Fig. 1,C), while this number was reduced to 9% in TCRδ^−/− mice (Fig. 1,F). This experiment is consistent with the data shown in Table I, which is an average of three separate experiments. A similar trend was observed for CD8 T cells, in which 42% of the cells in B10.PL mice (Fig. 1,D) and 4% in TCRδ^−/− mice (Fig. 1 G) produced IFN-γ. These data indicate that γδ T cells globally regulate the production of IFN-γ in the CNS by at least CD4⁺ and CD8⁺ IFN-γ-producing T cells.

As shown in Fig. 1, at least two different cell populations can produce IFN-γ in the CNS during EAE; thus, to determine the primary cell population responsible for IFN-γ production, we gated on the CD11b⁻ IFN-γ-producing cells and analyzed their expression of CD4 and CD8. We found that 79% of the IFN-γ-producing cells in B10.PL mice expressed CD4, while only 5.2% expressed CD8 (Table II). A similar result was observed in the TCRδ^−/− mice, with 70% of IFN-γ-producing cells expressing CD4 and 3.6% expressing CD8 (Table II). To determine the percentage of the CD4⁺IFN-γ⁺ T cells that originated from the adoptively transferred encephalitogenic T cell population, we further analyzed the cells for the expression of the TCR Vβ8.2 chain. We found that 74% of the total IFN-γ-producing cells expressed Vβ8.2, representing 94% of the CD4⁺IFN-γ⁺ T cells (Table II). These data show that the primary IFN-γ-producing cell in the CNS is the CD4⁺Vβ8.2⁺ encephalitogenic T cell.

In Table I, we showed that ∼84% of the IFN-γ-producing cells in B10.PL mice expressed either CD4 or CD8, indicating that other cell types in the CNS produce IFN-γ during EAE. A likely candidate is γδ T cells, which have been shown to produce large quantities of IFN-γ upon activation (36), and are present in the CNS of mice with EAE (see Fig. 5) (28, 29). The γδ T cells do not contribute to the CD4⁺IFN-γ⁺ or CD8⁺IFN-γ⁺ populations, as very few, if any, γδ T cells in the CNS express CD4 or CD8 (data not shown). Thus, to determine whether the decrease in IFN-γ production in the CNS of TCRδ^−/− mice was due directly to the loss of IFN-γ production by γδ T cells or due to a direct effect by γδ T cell regulation of IFN-γ production, BM irradiation mixed chimeras were generated by transplanting BM from either B10.PL (B10.PL→TCRδ^−/−), TCRδ^−/− (TCRδ^−/−→TCRδ^−/−), or IFN-γ^−/− (IFN-γ^−/−→TCRδ^−/−) mice into sublethally irradiated TCRδ^−/− mice (Fig. 2). In these chimera mice, the emerging γδ T cell populations will either be wild type (wt) (B10.PL donor BM) or deficient in IFN-γ production (IFN-γ^−/− donor BM), while the αβ T cells will be mixed with ∼50% of the cells of recipient in origin (data not shown). The TCRδ^−/− mice reconstituted with BM from TCRδ^−/− mice will be deficient in γδ T cells and act as a control for wt or IFN-γ^−/− chimeras. γδ T cell reconstitution was evident 4 wk posttransplant, as indicated by the presence of γδ T cells in IELs (data not shown). Additional control BM chimera mice were generated by performing the identical BM transfers into B10.PL mice (B10.PL→B10.PL, TCRδ^−/−→B10.PL, and IFN-γ^−/−→B10.PL) (Fig. 2). EAE was induced 4–6 wk post-BM transplantation, and the level of IFN-γ expression was assessed in the spinal cords of the chimeric mice by real-time PCR 10 and 15 days later. TCRδ^−/−→TCRδ^−/− BM chimera mice produced 100-fold less IFN-γ message in the CNS than the control B10.PL→TCRδ^−/− chimeras on day 10 (Fig. 2,A). However, by day 15, the difference was less pronounced (Fig. 2,A). These data are consistent with the results observed in Fig. 1. The IFN-γ^−/−→TCRδ^−/− chimera mice expressed similar levels of IFN-γ as the B10.PL→TCRδ^−/− chimeras (Fig. 2,A), indicating that γδ T cells regulate the expression of IFN-γ by cells in the CNS and do not contribute substantially to the total level of IFN-γ produced in the CNS. The B10.PL control chimeras all had similar levels of IFN-γ production in the CNS as that detected in the TCRδ^−/− chimeras, regardless of the source of the donor BM (Fig. 2 A), indicating that the B10.PL→TCRδ^−/− chimeras successfully reconstituted a functional γδ T cell population.

To determine whether the regulation of IFN-γ production by γδ T cells during EAE was specific to the CNS, we also examined the level of IFN-γ message in the cervical lymph nodes and spleen of the chimera mice. As shown in Fig. 2,B, there was no difference in the level of IFN-γ produced in the lymph nodes on day 10 following EAE induction in any of the BM chimeras. In addition, although the production of IFN-γ was more variable in the spleen between chimeras, there was not a reduction in the production of IFN-γ in the spleen of TCRδ^−/−→TCRδ^−/− chimera mice (Fig. 2 B). These data strongly suggest that regulation of IFN-γ production, a hallmark of a Th1 inflammatory response, is regulated by γδ T cells at the site of inflammation and not at peripheral sites, and is not due to the γδ T cell deficiency in the whole animal.

In the above studies, we show that mice deficient in γδ T cells have reduced levels of IFN-γ in the CNS in early EAE. In the studies conducted for Fig. 1,A, we observed that TCRδ^−/− mice were unable to recover from the disease symptoms of EAE as compared with the B10.PL mice (data not shown). We rationalized that this lack of recovery may be due to the IFN-γ deficiency in the TCRδ^−/− mice, because mice deficient in IFN-γ or the IFN-γR have been shown to exhibit a severe chronic disease course (16, 17). Thus, to confirm that γδ T cells were required for recovery, we reconstituted TCRδ^−/− mice with γδ T cells by BM transplantation using B10.PL or TCRδ^−/− donor BM. The TCRδ^−/− mice transplanted with B10.PL BM were reconstituted with γδ T cells 4 wk later, as determined by normal numbers of γδ T cells in the IEL population (data not shown). EAE was induced by passive induction, and the mice were observed daily for EAE disease symptoms. As shown in Fig. 3,A, the mice transplanted with B10.PL BM (B10.PL→TCRδ^−/−) were able to recover from EAE disease symptoms in a similar manner as the B10.PL control mice. However, TCRδ^−/− mice transplanted with TCRδ^−/− donor BM (TCRδ^−/−→TCRδ^−/−) were unable to recover from EAE (Fig. 3 A). Taken together, these cumulative results demonstrate that γδ T cells are required to promote the production of sufficient levels of IFN-γ in the CNS to allow for recovery from EAE, and show that γδ T cells are important regulators of CNS autoimmune inflammation.

In addition to an ascending paralysis, EAE is also characterized by an accumulation of inflammatory cells in discrete lesions throughout the length of the spinal cord and in the brain. This infiltrate is composed primarily of macrophages and, as shown in Fig. 3 B, both B10.PL and TCRδ^−/− mice have CD11b⁺ macrophages in spinal cord lesions that increase in number as the mice progress through the EAE disease course. The presence of granulocytes in the spinal cord lesions was detected by immunohistochemistry staining for Gr1, and no differences in the number of Gr1⁺ cells were detected in B10.PL vs TCRδ^−/− mice (data not shown). In addition, encephalitogenic T cells, B cells, and CD8 T cells are also present in the lesions (data not shown).

The data to date support a role for γδ T cell regulation of IFN-γ production specifically in the CNS. However, to further confirm that TCRδ^−/− mice do not have a deficiency in IFN-γ production, we measured the steady-state levels of IFN-γ in the spleen and cervical lymph nodes of unmanipulated B10.PL and TCRδ^−/− mice and found no differences in the level of IFN-γ production (Fig. 4,A). We then further confirmed that γδ T cell regulation of IFN-γ was specific to the CNS and did not occur in the peripheral lymphoid organs. First, we immunized B10.PL and TCRδ^−/− mice with CFA, and 7 days later harvested the draining lymph nodes and quantitated the level of IFN-γ message by real-time PCR (Fig. 4,B). As shown in Fig. 4,B, no difference was observed in the level of IFN-γ message detected in the B10.PL and TCRδ^−/− mice. Second, we adoptively transferred splenic activated CD4⁺ MBP-TCR tg T cells expressing GFP into B10.PL and TCRδ^−/− recipient mice, and following immunization with the Ac1–11 MBP peptide, examined IFN-γ production by the MBP-TCR tg GFP⁺ splenocytes. By examination of intracellular IFN-γ, we found no difference in the percentage of GFP⁺ T cells producing IFN-γ in the B10.PL (Fig. 4,C) and TCRδ^−/− (Fig. 4,D) mice. To confirm that the results obtained in Fig. 4 were not due to an absence of γδ T cells in B10.PL mice, we confirmed by flow cytometry that B10.PL mice have normal levels of γδ T cells in both the spleen and the IEL population (data not shown). Thus, these collective data indicate that TCRδ^−/− mice do not have an overt deficiency or dysregulation of IFN-γ production. Therefore, we conclude that regulation of IFN-γ production by γδ T cells during EAE occurs specifically in the CNS at the site of autoimmune manifestation and not in the periphery.

Our collective data to date show that γδ T cells have the capacity to specifically regulate IFN-γ in the CNS during EAE. To date, no candidate γδ T cells with the capacity to regulate IFN-γ have been identified. Thus, we used spectratyping to characterize candidate γδ T cell populations present in the CNS before EAE onset and during early disease. PCR primers were designed to specifically identify 14 TCRδ V gene families used by γδ T cells and to determine the TCR diversity as measured by differences in the length of the CDR3 region (34, 35). We found that only the DV7S6 and the DV105s1 gene families were present in the CNS before the induction of EAE. As a positive control, IEL isolated from B10.PL mice were examined for the usage of DV7S6 and DV105s1. As shown in Fig. 5, IEL γδ T cells using DV7S6 (lane 1) or DV105s1 (lane 6) are very diverse, as shown by at least 11 bands, indicating great variability in the length of CDR3. A similar pattern is observed in the spleen, blood, and thymus (data not shown), demonstrating that the population of γδ T cells using DV7S6 or DV105s1 is diverse and unrestricted. In contrast, a restricted pattern is observed in the CNS of unmanipulated wt mice, with only 3 bands being present for each V gene family (Fig. 5, lanes 2 and 7, respectively). On days 3, 7, and 10 following the induction of EAE, the DV7S6 and DV105s1 gene families were still the only TCRδ chain genes detectable in the CNS (Fig. 5, and data not shown). For the DV7S6 gene family, on days 3 (Fig. 5, lane 3) and 7 (Fig. 5, lane 4), the level of diversity of the V gene usage was variable from animal to animal, but maintained the limited diversity observed in unmanipulated mice (Fig. 5, lane 2). By day 10, the onset of EAE, the pattern of DV7S6 gene usage was more diverse and resembled the IEL population (Fig. 5, lane 5). As shown in Fig. 5, a similar pattern was observed for the MDV105s gene family with the restricted diversity apparent for days 3 (lane 8), 7 (lane 9), and 10 (lane 10). For both the DV7S6 and DV105s1 spectratypes, only one band was detectable at all time points (indicated by arrow). Thus, the only γδ T cell detectable in the CNS on day 7, when IFN-γ levels are the highest (Fig. 1,A), uses the DV7S6 or DV105s1 TCRδ V genes (Fig. 5). These data suggest γδ T cells expressing TCRδ chains from the DV7S6 and DV105s1 gene families have a tropism for the CNS and are potentially γδ T cells with the capacity to promote the production of IFN-γ in the CNS.

In this study, we addressed whether γδ T cells could regulate the EAE disease course and IFN-γ production in the CNS of mice with EAE. We observed that mice deficient in γδ T cells could not recover from EAE disease symptoms and had greatly reduced levels of IFN-γ in the CNS at both the message and protein levels early in disease, as compared with normal controls. The reconstitution of TCRδ^−/− mice with γδ T cells by BM transplantation restored both IFN-γ production and disease recovery. The regulation of IFN-γ production by γδ T cells was found to be by an IFN-γ-independent mechanism that resulted in the reduction of IFN-γ production by both CD4 and CD8 IFN-γ-producing T cells in the CNS, including the primary IFN-γ-producing cell in the CNS, the encephalitogenic T cell. In addition, our data suggest that γδ T cells exert their regulatory function specifically at the site of EAE inflammation, because no difference in the level of IFN-γ production by B10.PL or TCRδ^−/− mice was observed in the spleens or lymph nodes of unmanipulated mice, mice with EAE, or immunized mice. Furthermore, we identified γδ T cells using the DV7S6 and DV105s1 gene families as the only γδ T cells present in the CNS early in disease. Thus, these data indicate that γδ T cells specifically regulate the production of IFN-γ in the CNS during EAE.

It has been shown in a variety of infectious disease models that γδ T cells can influence the nature and extent of inflammation, but the mechanism of this regulation is not clear. However, an emerging observation is the modulation of IFN-γ production at the site of infection. A reduction in the production of IFN-γ in γδ T cell-deficient mice was consistently observed in a variety of bacterial infections (37, 38, 39, 40, 41). In these studies, the bacteria could activate the γδ T cells either directly or indirectly, resulting in the production of IFN-γ. In our studies, the reduction in IFN-γ that we observed in TCRδ^−/− mice was not due to bacterial products, as we induced EAE by passive transfer of encephalitogenic T cells and not by active immunization, which requires Ag emulsified in CFA containing mycobacteria in combination with pertussis toxin. Our study using TCRδ^−/− mice is consistent with a study by Rajan et al. (42) that also observed a reduction in IFN-γ in the CNS of SJL/J mice with EAE when γδ T cells were depleted with a mAb specific for the TCRγδ. Thus, our data suggest that the regulation of IFN-γ production is an intrinsic function of γδ T cells specific to inflammation and does not require the direct activation of γδ T cells by microorganisms. The specificity to inflammation in the CNS is indicated by our studies showing that IFN-γ production was not different in the spleen and lymph nodes of mice with EAE of TCRδ^−/−, as compared with control B10.PL mice. These data indicate that γδ T cell regulation of IFN-γ occurs specifically in the CNS during EAE and not in the periphery, suggesting that γδ T cells are local regulators of inflammation. This is consistent with their localization in epithelial surfaces of the small intestine, genital tract, skin, and tongue, all tissues directly in contact with microorganisms with the potential to cause inflammation (2).

Our data could also be explained by an inability of T cells to produce IFN-γ in γδ T cell-deficient mice. Thus, to confirm that TCRδ^−/− mice do not have a global defect in IFN-γ production, we compared the steady-state levels of IFN-γ message in the spleen and lymph nodes of TCRδ^−/− mice with that of B10.PL mice, and found no differences (Fig. 4,A). Because EAE requires primed T cells, we also wanted to ensure that activated T cells in TCRδ^−/− mice were able to produce IFN-γ. First, we immunized mice with CFA and measured IFN-γ message in the draining lymph node and found no alterations in the TCRδ^−/− mice (Fig. 4,B). Second, we transferred MBP Ac1–11-specific naive splenic T cells isolated from the MBP-TCR tg mice into B10.PL and TCRδ^−/− mice and examined IFN-γ production by the tg T cells following immunization with Ac1–11, and again found no differences between the two types of mice (Fig. 4, C and D). These cumulative data show that TCRδ^−/− mice do not have a global defect in IFN-γ production, and activated T cells are able to produce IFN-γ in TCRδ^−/− mice.

It is established that IFN-γ has multiple functions in inflammation, including the activation of macrophages and up-regulation of genes important for inflammation. Although γδ T cells can produce substantial levels of IFN-γ upon activation, it is not clear whether this source of IFN-γ is important for the regulatory role of γδ T cells in EAE. By generating mixed BM chimera mice in TCRδ^−/− mice by the transfer of IFN-γ^−/− BM, we were able to generate mice with γδ T cells deficient in IFN-γ production. IFN-γ production in these chimeras was similar to the B10.PL control chimeras (Fig. 2,A), indicating that IFN-γ production specifically by γδ T cells is not required for their regulation of the inflammatory response, nor does it contribute substantially to the overall level of IFN-γ produced in the CNS. This conclusion is consistent with the observation that the primary IFN-γ-producing cell in the CNS early in EAE is the encephalitogenic T cell (Table II). Our observation that γδ T cells do not contribute substantially to the level of IFN-γ in the CNS during early EAE is different from other studies, in which the ability of γδ T cells to produce IFN-γ was shown to be critical for tumor immunity and resistance to both West Nile virus and CMV (43, 44, 45). In these situations, it is possible that the foreign tumor cell or virus directly interacted with γδ T cells, resulting in IFN-γ production. This is supported by a study by Wang et al. (46), who showed that the live bacterial product isobutylamine stimulated the production of IFN-γ by human γδ T cells expressing the Vγ2Vδ2 TCR. Thus, our data in combination with the above studies demonstrate that γδ T cells can regulate immune responses by both IFN-γ-dependent and -independent mechanisms.

IFN-γ production is important for a normal EAE disease course, as demonstrated in IFN-γ-deficient mice that exhibited an exacerbated disease course accompanied with enhanced cellular infiltrates (16, 17, 20). In the TCRδ^−/− mice, we observed similar findings with the mice exhibiting a chronic disease course (Fig. 3). The relationship between γδ T cells and IFN-γ production is unclear, but γδ T cells have been shown to regulate the production of IFN-γ by a variety of cell types, including CD8 T cells (41) and NK cells (38). Our results showing a reduction in IFN-γ production by both CD4 and CD8 IFN-γ-producing T cells in the CNS during early EAE are consistent with these two studies (Fig. 1, E and F). The role for IFN-γ in the CNS during EAE has been suggested to control the inflammatory infiltrate in the CNS by a variety of mechanisms, including apoptosis of activated CD4 T cells, activation of macrophages to produce NO, and regulation of chemokines (18, 19, 20). Thus, our data suggest that a major function of γδ T cells is to control the production of IFN-γ by a variety of cell types, which is in turn important for regulation of the inflammatory response, in particular the resolution of inflammation. This is supported by a contact sensitivity model that demonstrated a role for γδ T cells in controlling the αβ effector cells, which includes the ability to down-regulate IFN-γ production in vitro (47).

It is becoming increasingly clear that γδ T cells play multiple roles during immunity, and the nature of their ligands is very diverse, but little is known about specific mechanisms of how γδ T cells regulate inflammation. This question is difficult to study because specific γδ T cell populations are difficult to isolate and grow. Thus, to identify target γδ T cell populations responsible for regulating inflammation induced by a variety of mechanisms, a variety of groups have examined TCRγδ gene usage using PCR. For MS, the identification of γδ T cells in the CNS of MS patients has been at the genetic level examining TCR gene usage and TCR diversity, with the overall conclusion that γδ T cells found in the CNS have a restricted repertoire (22, 23, 24). Only two studies to our knowledge have examined the TCRγδ repertoire in the CNS of mice. In both studies, four different TCRδ V genes were found in unmanipulated mice, with Vδ6 being the most common (29, 48). We also found this TCR V gene, which is now referred to as DV7s6 (Fig. 5) (35). In addition, the study by the Olive laboratory also detected Vδ5, which we also found, and is now named DV105s1 (29, 35). Interestingly, the study by Szymanska et al. (48), using SJL/J mice, did not detect this gene family in the CNS or in the spleen, suggesting a difference in mouse strains or a variability in the technique. Only the study by the Olive laboratory examined TCRδ usage during EAE, but it did not examine the diversity of either the DV7S6 or DV105s1 TCRδ repertoire (29). In our study, we found that both the DV7S6 and DV105s1 TCRδ repertoires were restricted in the CNS of unmanipulated mice as compared with IEL, but were variable from sample to sample, as indicated by a different banding pattern at the day 3, 7, and 10 time points (Fig. 5). However, for both the DV7S6 and DV105s1 TCRδ analysis, a single band was observed at all the EAE time points (Fig. 5). Thus, γδ T cells expressing a TCR composed of a DV7S6 or DV105s1 TCRδ chain could potentially represent a population of γδ T cells able to promote the production of IFN-γ.

Our data suggest that γδ T cells have the innate capacity to regulate IFN-γ production by IFN-γ-producing cells present in the CNS during early EAE. This dysregulated IFN-γ production is seen at both the message and protein levels and is associated with the lack of recovery from the symptoms of EAE. Thus, one possible function of γδ T cells during inflammation is to promote the production of IFN-γ, a cytokine known to have multiple functions, including being required for disease recovery in EAE. Because the control of inflammation in the CNS of MS patients is often a therapeutic strategy and many γδ T cells do not recognize ligands in a similar manner as the pathogenic CD4 T cells, treatment modalities directly targeting γδ T cells may not only be feasible, but effective therapies for MS.

We thank Vicki Boelter and Shelley Morris for assistance with the animal colony. We thank Elizabeth McDonough for technical support with the immunohistochemistry.