Absence of IFN-γ Increases Brain Pathology in Experimental Autoimmune Encephalomyelitis–Susceptible DRB1*0301.DQ8 HLA Transgenic Mice through Secretion of Proinflammatory Cytokine IL-17 and Induction of Pathogenic Monocytes/Microglia into the Central Nervous System

Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the CNS of presumed autoimmune origin (1). Genetic predisposition to MS is associated with certain MHC class II genes, whereas environmental factors also contribute to disease (1–6). High polymorphism, linkage disequilibrium, and heterogeneity of the human population made it difficult in the past to decipher the exact role of the human HLA class II genes HLA-DQ and HLA-DR in disease pathogenesis. The generation of HLA class II transgenic (Tg) mice has helped resolve some of these questions. Studies done in HLA class II Tg mice indicate that Tg mice expressing human HLA-DR2, DRB1*0301, and DR4 molecules are susceptible to CNS Ag-induced experimental autoimmune encephalomyelitis (EAE), an experimental model used to study MS (7, 8). Using these HLA Tg mice, we showed that HLA-DRB1*0301 Tg mice were susceptible to proteolipid protein (PLP) (91–110)–induced EAE (9), whereas DQ6 (DQB1*0601) and DQ8 (DQB1*0302) Tg mice were resistant. Surprisingly, DQ6/DRB1*0301 double-Tg mice were resistant (10), whereas DQ8/DRB1*0301 mice showed higher disease incidence and severity than DRB1*0301 mice (11).

CNS Ag-specific CD4⁺ T cells secreting proinflammatory cytokines are responsible for disease onset, as well as the chronic phase. Elegant studies in murine/rodent EAE have documented that encephalitogenic T cells are CD4⁺, Th1-type cells secreting TNF-α/β and IFN-γ (12–14). However, recent studies have indicated that another T cell phenotype, Th17 secreting IL-17, IL-17F, IL-21, IL-22, and IL-23, plays an important role in the immunopathogenesis of EAE (15). Thus, the current hypothesis of EAE indicates that Th1 and/or Th17 cytokines play important roles in the immunopathogenesis of EAE. However, the exact role of IFN-γ and IL-17 in disease pathogenesis is poorly understood because IFN-γ is pathogenic in some models and protective in others. Further, the Th1/Th17 ratio might influence the CNS pathology because a high Th17/Th1 ratio leads to a severe brain pathology, whereas a high Th1/Th17 ratio leads to a severe spinal cord pathology (16, 17).

Our previous data indicate that MHC class II molecules modulate immune responses through activation of specific cytokine profiles (10, 11, 18). The protective effect of DQ6 in DRB1*0301.DQ6 mice was mediated by IFN-γ (10), whereas the disease-exacerbating effect of DQ8 molecule was mediated by IL-17 (11). Thus, our studies indicated a pathogenic role for IL-17 and a protective role for IFN-γ in the pathogenesis of EAE in certain HLA class II Tg mice. To further decipher the role of IFN-γ in inflammation and demyelination, we generated HLA-DRB1*0301.DQ8 Tg mice lacking IFN-γ (DRB1*0301.DQ8.IFN-γ^−/−). We observed that DRB1*0301.DQ8.IFN-γ^−/− mice developed atypical EAE characterized by ataxia, abnormal gait, and dystonia without hind-limb paralysis. Mice with EAE showed brain-specific pathology with minimal or no spinal cord pathology. Brain pathology was characterized by severe demyelination in the cerebellum, brainstem, and subcortex/striatum region. Demyelinating regions showed relative preservation of axon and neurons with extensive presence of CD68⁺ cells, suggesting an important pathogenic role for monocytes/microglial cells in demyelination. Thus, our study indicates that IFN-γ plays a protective role in HLA-DRB1*0301.DQ8.IFN-γ^−/− HLA class II Tg mice.

The HLA-DQ8/DRB1*0301 double-Tg mice [DQ8 (DQA1*0103, DQB1*0302).-DRB1*0301 (DRA1*0101, DRB1*0301)] were produced as previously described (11, 19). In brief, HLA class II transgenes were introduced into (B6 × SWR)F₁ fertilized eggs. Positive offspring were backcrossed to B10.M mice for several generations. HLA class Tg mice were mated with MHC class II–deficient mice (AE^−/−) on B6 background (7) and intercrossed to generate the HLA-DRB1*0301.AE^−/− and HLA-DQ8.AE^−/− Tg lines. HLA-DRB1*0301.AE^−/− and HLA-DQ8.AE^−/− Tg mice were separately mated with IFN-γ–deficient mice on B6 background (The Jackson Laboratory) to generate HLA-DRB1*0301.AE^−/−.IFN-γ^−/− and HLA-DQ8.AE^−/−.IFN-γ^−/−, respectively. DRB1*0301.DQ8.AE^−/−.IFN-γ^−/− Tg mice were produced by intercrossing HLA-DRB1*0301.AE^−/−.IFN-γ^−/− with HLA-DQ8.AE^−/−.IFN-γ^−/− mice. The majority of HLA class II Tg mice have mixed B6/B10 background. All mice were bred and maintained in the pathogen-free Immunogenetics Mouse Colony of Mayo Clinic according to National Institutes of Health and institutional guidelines. All experiments were approved by the Institutional Animal Care and Use Committee at Mayo Clinic, Rochester, MN.

Expression of HLA-DR and HLA-DQ molecules on PBLs, lymph node cells, and splenocytes were analyzed by flow cytometry using mAbs L227 and IVD12 specific for HLA-DR (20) and HLA-DQ (21), respectively. Surface expression of CD4 (GK1.5), CD8 (53.6.72), B cells (RA3-6B2), dendritic cells (DCs; HL3), monocytes/macrophages (M1/70), NK cells (PK136), and neutrophils (7/4) were analyzed using fluorescent-conjugated mAbs from BD Biosciences (San Jose, CA). Intracellular levels of IL-6, IL-17, TNF-α, GM-CSF, and IFN-γ were analyzed using directly conjugated Abs from BD Biosciences.

The 20-aa-long synthetic peptide PLP_91–110 (YTTGAVRQIFGDYKTTICGK) was synthesized at the peptide core facility of Mayo Clinic (Rochester, MN).

For disease induction, 12- to 16-wk-old Tg mice were immunized s.c. in both flanks with 100 μg PLP_91–110 emulsified in CFA containing Mycobacterium tuberculosis H37Ra (400 μg/mice) (11). Pertussis toxin (100 ng; Sigma Chemicals, St. Louis, MO) was injected i.v. at days 0 and 2 after immunization. Mice of both sexes were used. Because IFN-γ–sufficient mice develop classical EAE characterized by ascending paralysis, these mice were scored for disease severity using standard EAE scores: 0, normal; 1, loss of tail tone; 2, hind-limb weakness; 3, hind-limb paralysis; 4, hind-limb paralysis and forelimb paralysis or weakness; 5, morbidity/death. Atypical EAE was graded as described previously (22) with slight modification: mild head tilting/ataxia = 1; severe ataxia = 2; severe ataxia and/or gait problem = 3; spasticity and/or dystonia = 4. As the classical and atypical EAE were scored on different scales, mice were weighed as a common and quantitative measure of the disease in both groups.

Splenocytes and lymph nodes were collected from immunized mice and challenged with Ag in vitro (9). The results are presented as stimulation indices (cpm of test sample/cpm of the control). For in vitro inhibition experiments, mAbs specific for CD4 (GK1.5), CD8 (TIB 105), HLA-DQ (IVD12), and HLA-DR (L227) were added to lymph node cells challenged in vitro with human PLP_91–110 (20 μg/ml). All of the neutralizing Abs were generated in-house using the Mayo Monoclonal Hybridoma core facility.

To study the Ag-presentation function, CD4⁺ T cells, C11b⁺ monocytes/macrophages, CD19⁺ B cells, and CD11c⁺ DCs were isolated from splenocytes and draining lymph nodes of PLP_91–110–immunized HLA-DRB1*0301.DQ8 or DRB1*0301.DQ8.IFN-γ^−/− Tg mice by magnetic sorting with a cell-specific positive isolation kit according to manufacturer’s protocol (Miltenyi Biotec). CD4⁺ T cells were plated at 1 × 10⁵ cells/well in presence or absence of 20 μg/ml PLP_91–110. Magnetically sorted C11b⁺ monocytes/macrophages, CD19⁺ B cells, and CD11c⁺ DCs from DRB1*0301.DQ8 or DRB1*0301.DQ8.IFN-γ^−/− Tg mice were irradiated and added at 0.2 × 10⁵ cells/well to CD4 T cells cultures in 96-well plates. Two sets of experiments were run in parallel, with one set used for T cell proliferation measurement and the other to collect supernatant for cytokine analysis.

Draining LNs were collected 10 d after immunization and stimulated with PLP_91–110 peptide as mentioned earlier in the Immunization and T Cell Proliferation Assay section. Supernatants were collected from the culture 48 h after peptide stimulation. The concentration of cytokines was measured using the mouse cytokine 23-plex protein bead array system as per the manufacturer’s instructions and analyzed with Bio-Plex manager 2.0 software (Bio-Plex; Bio-Rad Laboratories, Hercules, CA). Some cytokines were measured by sandwich ELISA using pairs of relevant anti-cytokine mAbs according to manufacturer’s protocol (BD Biosciences, San Jose, CA).

Expression of various cytokines, chemokines, and chemokine receptors (Supplemental Table 1) were analyzed by real-time PCR using commercial primer pairs (Realtimeprimrs.com, Elkins Park, PA). RNA was extracted from cells using RNAeasy columns (Qiagen), and cDNA was prepared using RNase H-reverse transcriptase (Invitrogen). cDNA was analyzed by real-time quantitative PCR in triplicates by using SYBR GreenER qPCR reagent system (Invitrogen). The expression level of each gene was quantified using the threshold cycle (C_t) method normalized for the housekeeping genes β-actin, GADPH, and HPRT (11).

Mice were perfused via intracardiac puncture with 50 ml Trump’s fixative (4% paraformaldehyde + 0.5% glutaraldehyde). The spinal cords and brains were removed and postfixed for 24–48 h in Trump’s fixative in preparation for morphologic analysis. All grading was performed without knowledge of the experimental group.

Spinal cords were cut into 1-mm coronal blocks and every third block postfixed in osmium and embedded in glycol methacrylate. Two-micrometer sections were stained with a modified eriochrome/cresyl violet stain. Morphologic analysis was performed on 12–15 sections per spinal cord. In brief, each quadrant from every coronal section of each spinal cord was graded for the presence or absence of inflammation and demyelination. The score was expressed as the percentage of pathologic abnormality in the spinal cord quadrants examined. A maximum score of 100 indicated a particular pathologic abnormality in each quadrant of each spinal cord section. Brain pathology was assessed after perfusion. Two coronal cuts in the intact brain (one section through the optic chiasm and a second section through the infundibulum) resulted in three paraffin-embedded blocks. This allowed analysis of the cortex, corpus callosum, hippocampus, brainstem, striatum, and cerebellum. The resulting slides were stained with H&E. Each area of brain was graded on a four-point scale: 0 = no pathology; 1 = no tissue destruction but minimal inflammation; 2 = early tissue destruction, demyelination, and moderate inflammation; 3 = moderate tissue destruction (neuronal loss, demyelination, parenchymal damage, cell death, neurophagia, neuronal vacuolation); 4 = necrosis (complete loss of all tissue elements with associated cellular debris). Meningeal inflammation was graded as follows: 0 = no inflammation; 1 = one cell layer of inflammation; 2 = two cell layers of inflammation; 3 = three cell layers of inflammation; 4 = four or more cell layers of inflammation. The area with maximum tissue damage was used to assess each brain region.

Brain and spinal cord tissues from animals were collected in optimal cutting temperature compound (Sakura Finetek Tissue Tek, Leiden, the Netherlands) and immediately frozen at −80°C freezer. Ten-micrometer sections were cut using cryostat, transferred to positively charged slides, fixed in cold acetone and stained with either fluorochrome-conjugated Abs or primary Ab, and followed by corresponding secondary Ab as per standard techniques. Stained sections were mounted using VECTASHIELD HardSet Mounting Medium with DAPI (Vector Lab, Burlingame, CA) and were analyzed using an Olympus Provis AX70 microscope (Leeds Precision Instruments, Minneapolis, MN) fitted with a DP70 digital camera.

The statistical significance of the differences in functional and histologic scores between groups was assessed by a one-way ANOVA on ranks (Kruskal–Wallis test) when comparing more than two groups and by the Mann–Whitney rank-sum test when comparing only two groups. The differences in proliferation or in cytokine levels between groups was assessed by a one-way ANOVA with multiple comparisons of the means when more than two groups were analyzed or by Student t test when only two groups were analyzed and their data were normally distributed.

All Tg lines developed normally and showed no gross phenotypic abnormalities. Both HLA-DR and DQ were expressed on 35–50% of cell population in PBLs and splenocytes. No endogenous class II expression was observed (data not shown). Thus, both HLA-DR and HLA–DQ molecules were expressed at similar levels in DRB1*0301.DQ8 and DRB1*0301.DQ8.IFN-γ^−/− Tg mice.

To investigate the effect of IFN-γ on disease susceptibility, we induced EAE in IFN-γ– sufficient and –deficient strains using PLP_91–110 as an Ag. The susceptibility and clinical features of IFN-γ–sufficient and –deficient strains to PLP_91-110–induced EAE are presented in Table I. Administration of PLP_91–110 to DRB1*0301.DQ8 Tg mice led to chronic progressive disease in 96% (24/25) of Tg mice. The disease was characterized by weight loss and ascending paralysis (limp tail followed by hind-limb weakness followed by complete hind-limb paralysis). Interestingly, DRB1*0301.DQ8.IFN-γ^−/− Tg mice developed atypical EAE with disease incidence rate of 92% (23/25) characterized by ataxia, gait problems, shivering and tremors (dystonia), spasticity, moderate hind-limb weakness, and weight loss, features suggestive of brain-specific pathology. These mice also maintained their tail tonicity during the course of the disease. The ascending paralysis observed in IFN-γ–sufficient DRB1*0301.DQ8 Tg mice has been associated with both brain- and spinal cord–specific pathology. The disease onset (10.3 ± 1.5 versus 13 ± 1.5, p = ns) and incidence (96 versus 92%, p = ns) was similar between IFN-γ–sufficient and –deficient DRB1*0301.DQ8 Tg mice. Both DRB1*0301.DQ8.IFN-γ^−/− with atypical EAE and DRB1*0301.DQ8 with classical EAE showed similar weight loss (data not shown). DRB1*0301.DQ8 Tg mice showed progressive weight loss, and there was a direct correlation between the clinical disease score and weight loss. Although DRB1*0301.DQ8.IFNg^−/− mice also showed progressive weight loss that correlated with disease score, animals started to recover clinically and gained weight around the fourth week (data not shown). The major difference between EAE observed in IFN-γ–sufficient and –deficient mice was the course of EAE. IFN-γ–sufficient DRB1*0301.DQ8 mice developed chronic progressive EAE characterized by ascending paralysis and never recovered from their symptoms. These mice were sacrificed after 3–4 wk because of their severe paralysis and significant weight loss. In contrast, IFN-γ–deficient DRB1*0301.DQ8.IFN-γ^−/− mice develop a resolving disease with disease onset at day 13 (± 1.5), disease peak at weeks 3 or 4, and resolution by weeks 7 or 8. We followed these mice for 90 d postimmunization, and they remained in remission. We used DRB1*0301.IFN-γ^−/− DQ8.IFN-γ^−/− and AE°.IFN-γ^−/− (lacking mouse class II and IFN-γ) as controls. In contrast with atypical EAE in DRB1*0301.DQ8.IFN-γ^−/− Tg mice, DRB1*0301.IFN-γ^−/− single-Tg mice develop both classical and atypical EAE (Table I). IFN-γ–sufficient DRB1*0301 Tg mice develop classical EAE characterized by ascending paralysis and weight loss; however, it was less severe compared with DRB1*0301.DQ8 double-Tg mice (data not shown). The majority of the DRB1*0301 Tg mice remain paralytic until end of the study but did not become quadriplegic as observed in DR3.DQ8 Tg mice. No disease was observed in DQ8.IFN-γ^−/− or AE°.IFN-γ^−/− mice (Table I). Thus, our data indicate that disease-susceptible human HLA class II transgene DRB1*0301 is necessary for the development of disease, whereas absence of IFN-γ in DRB1*0301.DQ8 double-Tg mice leads to atypical EAE disease characterized by head tilting/ataxia, gait, dystonia, and spasticity.

The presence of atypical EAE in the absence of IFN-γ in DRB1*0301.DQ8 Tg lines (DRB1*0301.DQ8.IFN-γ^−/−) suggested a brain-centric pathology (23, 24) compared with classical EAE observed in IFN-γ–sufficient mice, which is characterized by both brain- and spinal cord–specific pathology (11). To confirm this, we analyzed the brain and spinal cord tissue of mice with EAE from all single- and double-Tg mice sufficient or deficient for IFN-γ. Pathological analysis of CNS tissue showed that, whereas IFN-γ–sufficient DRB1*0301 single-Tg and DRB1*0301.DQ8 double-Tg mice developed both spinal cord and brain pathology, IFN-γ–deficient DRB1*0301.DQ8 mice showed only brain-specific pathology. Although both IFN-γ–sufficient and –deficient strains with EAE showed brain pathology, the pathology was more severe in DRB1*0301.DQ8.IFN-γ^−/− Tg mice compared with DRB1*0301.DQ8 Tg mice and was characterized by a higher degree of inflammation and demyelination, especially in the cerebellum, brainstem, striatum, and septal nuclei regions (Fig. 1A–P). IFN-γ–deficient DRB1*0301.DQ8 Tg mice with disease showed extensive inflammation and demyelination in the septal nuclei area of forebrain regions (Fig. 1E), whereas no pathology in this region was observed in IFN-γ–sufficient DRB1*0301 and DRB1*0301.DQ8 mice with EAE (Fig. 1A). Similarly, severe inflammation and demyelination were observed in the striatum region of IFN-γ–deficient DRB1*0301.DQ8 Tg animals with atypical disease (Fig. 1F) compared with IFN-γ–sufficient DRB1*0301.DQ8 Tg animals with classical EAE (Fig. 1B). Although DRB1*0301.DQ8 mice with EAE showed inflammation and demyelination in cerebellum (Fig. 1C), the extent of inflammation and demyelination in the cerebellum of DRB1*0301.DQ8.IFN-γ^−/− Tg mice (Fig. 1E) was more extensive. IFN-γ–deficient DRB1*0301.DQ8 animals with disease also showed a higher degree of inflammation in the brainstem region (Fig. 1D) compared with the mild inflammation in IFN-γ–sufficient DRB1*0301.DQ8 animals with disease (Fig. 1H). Brain sections from DRB1*0301.DQ8.IFN-γ^−/− showed more parenchymal white matter loss compared with DRB1*0301.DQ8 mice. HLA-DRB1*0301.IFN-γ^−/− single-Tg mice also showed increased brain pathology (Fig. 1I–L) compared with IFN-γ–sufficient HLA-DRB1*0301 Tg mice (Fig. 1M–P). However, the extent of pathology in IFN-γ–deficient HLA-DRB1*0301 single-Tg mice was less severe compared with DRB1*0301.DQ8.IFN-γ^−/− double-Tg mice, indicating importance of DQ8 molecule in the brain pathology. Interestingly, despite severe brain pathology, IFN-γ–deficient DRB1*0301.DQ8 mice with EAE showed a very mild or no pathology in the spinal cord. As expected, DRB1*0301.DQ8 mice with EAE showed significant inflammation and demyelination in the spinal cord compared with very mild pathology in DRB1*0301.DQ8.IFN-γ^−/− mice (Fig. 2A). Quantitative analysis of spinal cord tissues showed that, on average, 38 ± 10% of the spinal cord quadrants from DRB1*0301.DQ8 showed inflammation compared with only 2 ± 3% in DRB1*0301.DQ8.IFN-γ^−/− mice (p < 0.01; Fig. 2B). Similarly, IFN-γ–sufficient DRB1*0301.DQ8 showed demyelination in 37.8 ± 6% of spinal cord quadrants, compared with only 2 ± 3% of the quadrants from IFN-γ–deficient Tg mice (p < 0.001; Fig. 2B). Both IFN-γ–sufficient and –deficient DRB1*0301 single-Tg mice showed a similar level of spinal cord inflammation (26.8 ± 7.2 versus 20.8 ± 7.2%; p = ns) and demyelination (20.3 ± 5.4 versus 15.7 ± 6.6%; p = ns; Fig. 2B). Thus, the absence of IFN-γ leads to brain-restricted inflammation and demyelination in DRB1*0301.DQ8.IFN-γ^−/−, whereas both brain and spinal cord pathology was observed in DRB1*0301IFN-γ^−/− Tg mice.

To confirm demyelination of axons, we performed Luxol fast blue (LFB) staining, which stains lipoproteins in myelin and gives them a blue appearance under the microscope. LFB staining of paraffin-embedded brain tissues showed severe loss of LFB staining, indicating myelin loss in DRB1*0301.DQ8.IFN-γ^−/− compared with DRB1*0301.DQ8 Tg mice. Severe inflammation and demyelination was also observed in the brainstem of DRB1*0301.DQ8.IFN-γ^−/− mice with EAE compared with only mild inflammation in DRB1*0301.DQ8 mice with EAE (Fig. 3A, 3B). As shown in Fig. 3C, whereas DRB1*0301.DQ8 mice with EAE had perivascular inflammation and some demyelination in the cerebellum, DRB1*0301.DQ8.IFN-γ^−/− mice had myelin loss in significantly larger portions of the cerebellum (Fig. 3D) and brainstem (Fig. 3B). All of the myelin was destroyed within some lobules of the cerebellum (Fig. 3D). In addition to the cerebellum and brainstem, loss of myelin was also observed in the striatum, fornix, and subcerebrum (data not shown).

We also performed Bielschowsky staining to investigate whether demyelination was associated with axon loss in DRB1*0301.DQ8.IFN-γ^−/− mice. DRB1*0301.DQ8 mice with EAE showed preserved axon tracts in the brainstem (Fig. 3E) and cerebellum in demyelinated areas (Fig. 3G). IFN-γ–deficient DRB1*0301.DQ8 animals with severe demyelination showed some axonal loss in the brainstem (Fig. 3F), whereas axons in the cerebellum were preserved except in areas with extensive inflammation. Thus, the data indicate that absence of IFN-γ leads to severe brain pathology characterized by inflammation and demyelination but relatively preserved axons.

Ag-specific CD4⁺ T cells of both Th1 and Th17 subtype have been shown to be responsible for initiation of disease in the EAE model. Therefore, we analyzed T cell proliferative responses and levels of different inflammatory cytokines in an attempt to identify the mechanism for this unique form of EAE in IFN-γ–deficient mice. Although the numbers of total cells were similar between the two groups, we observed increased PLP_91–110 peptide–specific CD4 T cell proliferation in IFN-γ–deficient DRB1*0301.DQ8 mice (Fig. 4A). HLA-DRB1*0301.IFN-γ^−/− single-Tg mice also showed increased T cell proliferation compared with IFN-γ–sufficient HLA-DRB1*0301 Tg mice (Supplemental Fig. 1A). However, among all the groups, the strongest T cell proliferative response was observed in DRB1*0301.DQ8.IFN-γ^−/− double-Tg mice (Supplemental Fig. 1A). We have previously shown that DRB1*0301.DQ8 mice with EAE produce both Th1 and Th17 cytokines; however, IL-17 is the major cytokine as neutralization of IL-17 abrogated disease in DRB1*0301.DQ8 Tg mice (11). Cytokine analysis showed that both IFN-γ–deficient DRB1*0301.DQ8.IFN-γ^−/− and IFN-γ–sufficient DRB1*0301.DQ8 Tg mice produced Th17 cytokines such as IL-17 (Fig. 4B), GM-CSF (Fig. 4B), IL-6, and IL-12 (data not shown). However, levels of these proinflammatory cytokines were significantly higher in DRB1*0301.DQ8.IFN-γ^−/− compared with DRB1*0301.DQ8 mice (Fig. 4B, 4D), indicating that the absence of IFN-γ caused increased levels of these proinflammatory cytokines. HLA-DRB1*0301.IFN-γ^−/− single-Tg mice also showed greater levels of IL-17 and GM-CSF compared with IFN-γ–sufficient HLA-DRB1*0301 Tg mice (Supplemental Fig. 1B, 1D). However, the levels of both IL-17 and GM-CSF were highest in DRB1*0301.DQ8.IFN-γ^−/− double-Tg mice (Supplemental Fig. 1B, 1D). Among the different strains, levels of IFN-γ were as expected (Supplemental Fig. 1C). To confirm that CD4⁺ T cells are the main source of IL-17 and IFN-γ, we analyzed intracellular levels of these cytokines. Splenocytes from PLP_91–110–immunized mice were stimulated in vitro in the presence or absence of the same Ag. After 3 d, cells were stimulated with PMA and ionomycin for 5 h and followed by addition of GolgiStop after 1 h. Intracellular levels of IL-17 and IFN-γ on CD4⁺ gated T cells were analyzed by flow cytometry. As shown in Fig. 4E, CD4⁺ T cells from DRB1*0301.DQ8.IFN-γ^−/− mice produced greater levels of IL-17 (4.8 ± 1.0 versus 2.2 ± 0.7; p < 0.05) compared with DRB1*0301.DQ8 mice, but DRB1*0301.DQ8 mice produced significantly greater levels of IFN-γ (6.0 ± 1.6 versus 1.0 ± 0.1; p > 0.01).

To further characterize the CD4⁺ T cell–specific cytokines and chemokines responsible for the atypical EAE and brain-specific pathology, we isolated splenic CD4⁺ T cells from IFN-γ–deficient mice with atypical EAE and IFN-γ–sufficient mice with classical EAE. CD4⁺ T cells were stimulated in in vitro cultures with PLP_91–110, and levels of various cytokines, chemokines, chemokine receptors, and transcription factors were analyzed at protein level (Bio-Plex) or mRNA level (real-time RT-PCR). CD4⁺ T cells from DRB1*0301.DQ8.IFN-γ^−/− Tg mice with atypical EAE had greater protein levels of IL-1β, IL-3, IL-5, IL-6, IL-13, IL-17, G-CSF, GM-CSF, KC, and TNF-α (Fig. 4F), whereas CD4⁺ T cells from DRB1*0301.DQ8 Tg mice with classical EAE produced greater levels of IL-4, IL-12p40, IFN-γ, and MCP-1 (Fig. 4F). Analysis of mRNA showed that CD4⁺ T cells from IFN-γ–deficient mice showed greater levels of the earlier cytokines, as well as an increase in IL-11, IL-12p40, IL-22, CCR4, CCR6, CXCR2, and CXCR4, whereas CD4⁺ T cells from IFN-γ–sufficient mice with classical EAE showed an increase in IL-15, CCR5, and CXCR3 beside the inflammatory mediators observed at protein level (Fig. 4G and Supplemental Table 1). Thus, these data indicate that the severe brain-specific disease in DRB1*0301.DQ8.IFN-γ^−/− mice is due to the increased ability of Ag-specific CD4⁺ T cells to proliferate, as well as its ability to produce proinflammatory cytokines and express specific chemokine receptors.

CD4⁺ T cells from DRB1*0301.DQ8.IFN-γ^−/− mice with EAE showed increased proliferation of and higher Th17 cytokines, which can be either because of increased Ag presentation of IFN-γ–deficient mice or increased proliferative and IL-17/GM-CSF–producing capacity of CD4⁺ T cells. To identify the cell population responsible for the increased encephalitogenic CD4⁺ T cells, we isolated CD4⁺ T cells, CD11b⁺ monocytes, CD19⁺ B cells, and CD11c⁺ DCs from IFN-γ–sufficient or –deficient DRB1*0301.DQ8 Tg mice and performed an in vitro Ag-presentation assay in a crisscross manner. As shown in Fig. 5A, when CD4⁺ T cells from DRB1*0301.DQ8 or DRB1*0301.DQ8.IFN-γ^−/− Tg mice with EAE were cultured with CD11c⁺ as APCs from either of these two strains, CD4⁺ T cells from IFN-γ–deficient DRB1*0301.DQ8 mice (DRB1*0301.DQ8.IFN-γ^−/−) showed higher proliferation in response to recall Ags compared with those from IFN-γ–sufficient mice. As monocytes/macrophages have been shown to play an important role in EAE pathogenesis, we isolated CD11b⁺ monocytes from IFN-γ–sufficient or IFN-γ–deficient mice and performed Ag presentation. As shown in Fig. 5B, culturing of CD11b⁺ monocytes with CD4⁺ T cells from either IFN-γ–sufficient or –deficient mice induced similar T cell proliferation. We did not observe any difference in CD4⁺ T cell proliferation when B cells from IFN-γ–sufficient or –deficient mice were used as APCs (data not shown). Analysis of cytokines showed that CD4⁺ T cells from DRB1*0301.DQ8.IFN-γ^−/− mice produced greater levels of IL-17 and GM-CSF when cultured with CD11c⁺ DCs from either strain compared with CD4⁺ T cells from DRB1*0301.DQ8 mice (Fig. 5C, 5D). Similar difference in the cytokine levels of IL-17 and GM-CSF between the two groups were observed, when CD11b⁺ monocytes or CD19⁺ B cells were used as APCs (data not shown). Thus, our Ag-presentation studies indicate that increased proliferative and IL-17/GM-CSF–producing capacity in IFN-γ–deficient mice were due to the proliferating ability of CD4⁺ T cells rather than the increased Ag-presenting function of APCs such as B cells, CD11b⁺ monocytes, or CD11c⁺ DCs. To further confirm that atypical EAE and brain-specific pathology in IFN-γ–deficient mice is due to inability of immune cells to produce IFN-γ, we performed bone marrow chimera experiments. IFN-γ–sufficient DRB1*0301.DQ8 mice and IFN-γ–deficient DRB1*0301.DQ8 mice were lethally irradiated (600 rad 4 h apart) and reconstituted with BM cells from IFN-γ–deficient or –sufficient mice, respectively. Six weeks after adoptive transfer, reconstitution was confirmed and EAE was induced in mice using standard protocol. IFN-γ–sufficient mice reconstituted with BM from IFN-γ–deficient mice showed both classical and atypical EAE (Table II) with both severe brain and spinal cord pathology (data not shown). DRB1*0301.DQ8.IFN-γ^−/− Tg mice reconstituted with BM from IFN-γ–sufficient mice develop classical EAE (Table II) with severe spinal cord pathology and mild brain pathology (data not shown). These data demonstrate that absence of IFN-γ leads to atypical EAE, and IFN-γ–induced pathways protect brain from severe pathology, whereas presence of IFN-γ leads to the development of classical EAE with severe spinal cord pathology.

CD4⁺ Th17 cells producing IL-17 attract polymorphic nuclear cells such as brain neutrophils, leading to increased CNS pathology (25). Therefore, we determined the cellular profile of immune cells in both the spleen and CNSs to identify the cell population responsible for the brain-restricted pathology in DRB1*0301.DQ8.IFN-γ^−/− mice. Both IFN-γ–sufficient and –deficient mice with EAE had similar numbers of cells in the spleen (Fig. 6A); however, DRB1*0301.DQ8.IFN-γ^−/− mice had more brain-infiltrating leukocytes (BILs) than IFN-γ–sufficient DRB1*0301.DQ8 mice (Fig. 6B). Of the immune cells, we specifically observed increased frequency of CD4⁺ T cells, CD11b⁺ cells, and 7/4⁺ neutrophils in IFN-γ–deficient mice compared with IFN-γ–sufficient mice (Fig. 6C). DRB1*0301.DQ8.IFN-γ^−/− mice had fewer B220⁺ B cells and CD8⁺ T cells than DRB1*0301.DQ8 mice with EAE (Fig. 6C). Spinal cord–infiltrating cells (SCILs) from DR3.DQ8 mice with classical EAE had a cellular profile similar to BILs from the same strain except an increase in number of CD11b⁺ cells. The major difference between two groups was reduced number of B cells and CD8 T cells in IFN-γ–deficient mice with atypical EAE. We were not able to get enough cells from DRB1*0301.DQ8.IFN-γ^−/− mice with atypical EAE to perform cellular profile analysis in SCILs. Cytokine analysis of BILs showed DRB1*0301.DQ8.IFN-γ^−/− mice with EAE have greater levels of IL-17 (Fig. 6D) and GM-CSF (Fig. 7E) than DRB1*0301.DQ8 mice with EAE (Fig. 6D). We also performed intracellular analysis of BILs to confirm that CD4⁺ T cells are the main source of IL-17 and GM-CSF. BILs from PLP_91–110–immunized mice were stimulated with PMA and ionomycin for 5 h and followed by addition of GolgiStop after 1 h. Intracellular levels of IL-17, GM-CSF, IL-6, and TNF-α on CD4⁺ gated T cells were analyzed by flow cytometry. As shown in Fig. 6F, BILs-CD4⁺ T cells from DRB1*0301.DQ8.IFN-γ^−/− mice produced greater levels of IL-17 (18 ± 4.0 versus 3.0 ± 1.0; p < 0.05), GM-CSF (20 ± 3.6 versus 3.5 ± 1.5; p < 0.05), and IL-17⁺GM-CSF⁺ double-positive cells (11 ± 2.6 versus 0.8 ± 1.0; p < 0.01) compared with IFN-γ–sufficient mice. CD4⁺ T cells from BILs of IFN-γ–deficient mice also showed greater levels of IL-6 and TNF-α (data not shown). However, SCILs from IFN-γ–sufficient DRB1*0301.DQ8 mice produced greater levels of IFN-γ (5.0 ± 1.2 versus 1.0 ± 0.4; p > 0.05) and IL-17 (6.0 ± 1.6 versus 1.5 ± 0.6; p > 0.05) compared with IFN-γ–deficient mice (data not shown). Thus, the earlier data indicate that the severe brain-specific disease in DRB1*0301.DQ8.IFN-γ^−/− mice is due to the increased ability of Ag-specific CD4⁺ T cells to proliferate, as well as their ability to produce greater levels of both IL-17 and GM-CSF.

To decipher the mechanism responsible for localization of brain-specific pathology characterized with atypical EAE in IFN-γ–deficient mice versus classical EAE with spinal cord pathology in IFN-γ–sufficient mice, we performed mRNA expression analysis of various chemokine ligands and cytokines in brain and spinal cord tissue. As shown in Fig. 6G, greater levels of MIP-1α, CCL7, CCL8, CCL17, and CCL19 chemokines were observed in brain tissue of IFN-γ–deficient mice compared with the brain tissue of IFN-γ–sufficient mice. Whereas spinal cord tissue from DRB1*0301.DQ8 mice with classical EAE showed a relative increase in the expression levels of CCL5, CXCL9, CXCL10, and CXCL12 compared with IFN-γ–deficient mice (Fig. 6H). Levels of CXCL12 were similar between two groups within the brain tissue. We also analyzed BILs and SCILS from both groups and observed a relative increase in mRNA levels of IL-6, IL-11, IL-17, IL-22, IL-23, and TNF-α in BILs from IFN-γ–deficient mice compared with IFN-γ–sufficient mice, whereas the latter had increased levels of IL-1β, IL-15, and IFN-γ (Supplemental Table 1). SCILs from IFN-γ–sufficient mice showed higher transcript levels of IL-1β, IL-10, IL-12p35, IL-17, and IFN-γ (Supplemental Table 1). These data demonstrate that the presence of specific cytokines and chemokines within the brain or spinal cord might determine the differential disease in IFN-γ–sufficient versus deficient mice.

To further characterize the cells present in demyelinating areas in the brain, we stained frozen sections from both IFN-γ–sufficient and –deficient mice with EAE with Abs specific for CD4, CD8, CD68 (monocyte/microglia), and A7/4 (neutrophil). CD4⁺ and CD8⁺ T cells were present around the perivascular inflammatory region and meningeal lining in both strains (data not shown). However, demyelinating regions in the cerebellum were specifically populated by CD68⁺ cells, suggesting an important role for monocytes/microglia in inducing demyelination (Fig. 7A). We observed the presence of neutrophils along the borders of demyelinated regions; however, they were significantly less than CD68⁺ cells. To further analyze whether CD68⁺ cells are peripheral monocytes or brain resident glial cells, we stained brain tissue with CD68 and CD45. Whereas IFN-γ knockout (KO) mice with atypical EAE had a large area with inflammation and demyelination, IFN-γ–sufficient mice had small areas with pathology (Fig. 7B). Within the brain tissue with inflammation and demyelination, both CD45⁺CD68⁺ and CD45⁻CD68⁺ cell population were present, indicating importance of both monocytes and microglia in the pathology (Fig. 7B). In the brain of DR3DQ8 mice with classical EAE, the majority of CD68⁺ cells were positive for CD45, indicating a role for only monocyte population. Thus, our immunohistochemistry data indicate that CD68⁺ monocytes/microglia are the major cells in the demyelinating lesions of IFN-γ–deficient DRB1*0301.DQ8 mice.

We have previously shown that IFN-γ plays a protective role in PLP_91–110–induced EAE in disease-susceptible HLA-DRB1*0301 mice (10). In this study, we have extended the findings by generating IFN-γ–deficient HLA-DRB1*0301.DQ8 (HLA-DRB1*0301.DQ8.IFN-γ^−/−) Tg mice and showed that the absence of IFN-γ leads to atypical EAE characterized by ataxia and gait abnormalities. HLA-DRB1*0301.DQ8.IFN-γ^−/− mice with EAE showed inflammation and demyelination restricted to brain, compared with the brain and spinal cord pathology observed in IFN-γ–sufficient DRB1*0301.DQ8 Tg mice. The inflammation and demyelination in the brain was also more severe and widespread in HLA-DRB1*0301.DQ8.IFN-γ^−/− than in HLA-DRB1*0301.DQ8 Tg mice. The demyelination in our model correlated with relative preservation of axons. Ag-presentation assays revealed that severe pathology in HLA-DRB1*0301.DQ8.IFN-γ^−/− Tg mice was due to increased encephalitogenicity of IFN-γ–deficient Ag-specific CD4⁺ T cells. These CD4⁺ T cells caused an increase in inflammation and demyelination through production of proinflammatory cytokines IL-17, GM-CSF, IL-6, and TNF-α, which helped in the recruitment of inflammatory cells into CNS and/or the activation of resident microglia. The role of IFN-γ in inflammatory and demyelinating diseases is not well understood and is hypothesized to have both a pathogenic and a protective role. Our study indicates that: (i) IFN-γ is not required for development of disease; (ii) its absence leads to development of atypical EAE characterized by ataxia, gait problem, dystonia, and weight loss; (iii) IFN-γ–deficient mice with EAE develop brain-specific inflammation and demyelination; and (iv) CD68⁺ monocytes and microglia plan an important role in inducing the brain pathology.

Previous studies have shown that absence of either the IFN-γ or IL-12 signaling pathway leads to more severe neurologic disease (26, 27). Ferber et al. (26) and Krakowski and Owens (27) used IFN-γ–deficient mice with B10.PL (H-2u) and BALB/c (H-2d) backgrounds to show that the absence of IFN-γ leads to the development of severe classical EAE. In contrast with these studies, we observed development of atypical EAE characterized by ataxia, circling (ataxia-rotation), gait problem, and dystonia in our IFN-γ–deficient double-Tg strains. Abromson-Leeman et al. (22) reported the development of similar ataxia-rotation, atypical disease in BALB/c IFN-γ KO animals on immunization with MBP Exon 2 peptide. Wensky et al. (24) also reported atypical EAE in MBP-specific, TCR Tg mice lacking IFN-γ. However, in both studies, atypical disease was attributed to extensive inflammation in the lateral medullary regions of brain. None of these studies mentioned or showed demyelination. To our knowledge, our study is the first to show that atypical EAE in IFN-γ–deficient DRB1*0301.DQ8 mice correlates with demyelination in cerebellum, brainstem, as well as striatum and septal nuclei region of the forebrain. The maximal demyelination was observed in the cerebellum region as LFB staining revealed the destruction of approximately half of the myelin sheath within some lobules of the cerebellum. Myelin within the simple lobule and inferior semilunar lobule region of cerebellum showed complete loss in IFN-γ–deficient mice compared with mild perivascular inflammation in IFN-γ–sufficient mice. Despite the severe demyelination observed in IFN-γ–deficient DRB1*0301.DQ8 Tg mice, the majority of the axons were preserved, and only the area with extensive inflammation showed axonal degeneration. Thus, our study confirms a strong neuroprotective role of IFN-γ, whereas its absence leads to severe inflammation and demyelination in the unique parts of the brain. IFN-γ–deficient DRB1*0301 single-Tg mice developed both atypical EAE and classical EAE, and showed higher degree of inflammation and demyelination compared with IFN-γ–sufficient DRB1*0301 single-Tg mice. However, the pathology was more severe in IFN-γ–deficient DRB1*0301.DQ8 double-Tg mice compared with DRB1*0301.IFN-γ^−/− single-Tg mice. This together with failure of the HLA-DQ8 Tg mice to develop EAE either in IFN-γ–sufficient or –deficient mice indicate that whereas HLA-DR3 is required for susceptibility of disease, the presence of DQ8 molecule worsens the disease. We have previously shown that HLA-DQ8–restricted Th17 cells exacerbate EAE in the HLA-DRB1*0301.DQ8 Tg mice (11). As IFN-γ is known to regulate Th17 response (28), absence of IFN-γ might cause an increase in the DQ8-restricted Th17 cell response. The increased expansion of DQ8-restricted Th17 subset together with expansion of DR3-restricted encephalitogenic CD4 T cells could lead to increased trafficking of inflammatory cells to CNS, as well as production of proinflammatory cytokines and chemokines, resulting in severe brain pathology in DRB1*0301.DQ8.IFN-γ^−/− double-Tg mice. Thus, in the presence of disease-susceptible HLA class II molecule (e.g., HLA-DRB1*0301), other class II molecules (e.g., HLA-DQ8 in this study) can influence the outcome of the disease through epistatic interaction in trans.

DRB1*0301.DQ8.IFN-γ^−/− with EAE had more brain-infiltrating cells (BILs) than IFN-γ–sufficient DRB1*0301.DQ8 mice. Flow-cytometry analysis of BILs showed increased frequency of monocytes and neutrophils in IFN-γ–deficient mice compared with IFN-γ–sufficient DRB1*0301.DQ8 Tg mice. Increased levels of IL-17 and GM-CSF in BILs of DRB1*0301.DQ8.IFN-γ^−/− mice suggest an important role for these cytokines in recruiting monocytes and neutrophils to the brain. Recent studies have shown that high IL-17/IFN-γ ratio leads to brain-specific pathology, whereas high IFN-γ/IL-17 ratio leads to spinal cord pathology (17, 23). The increased brain-specific pathology in DRB1*0301.DQ8.IFN-γ^−/− mice was accompanied with the increased levels of IL-17, thus supporting a major role for IFN-γ in localizing pathology in brain versus spinal cord. Previous studies have shown that neutrophil recruitment to inflammatory sites is a major function of the IL-17 cytokine (29). However, we observed an increased presence of CD68⁺ inflammatory monocytes in demyelinating areas, whereas neutrophils were present only around perivascular areas, which suggests that CD68⁺ monocytes/microglia might be the major pathogenic cells in our demyelinating model. Our CD45 and CD68 dual staining of the brain tissues indicated that both infiltrating monocytes and brain resident microglia play a role in inducing severe pathology in IFN-γ–deficient mice with atypical EAE. Both infiltrating monocytes and microglia have been shown to be involved in the CNS demyelination with monocyte-derived inflammatory macrophages causing the destruction of myelin, whereas microglia-derived macrophages are responsible for clearing the debris (30). Presence of inflammatory monocytes in demyelinating region is also in agreement with an earlier study from Lafaille’s group (24), who showed that adoptive transfer of MBP-specific TCR Tg T cells lacking IFN-γ into RAG-deficient mice leads to development of atypical EAE characterized with increased presence of monocytes in brain. However, because they did not analyze levels of IL-17 and GM-CSF in their system, we cannot compare requirement of IL-17 and GM-CSF between these two models.

IFN-γ also regulates T cell response, specifically IL-17 levels, in the peripheral nervous system (28). Indeed, we observed higher T cell proliferation and increased levels of IL-17 in vitro in response to PLP_91–110 in IFN-γ–deficient DRB1*0301.DQ8 Tg mice compared with IFN-γ–sufficient DRB1*0301.DQ8 Tg mice. The increased proliferation and IL-17 levels observed in IFN-γ–deficient DRB1*0301.DQ8 mice were due to the ability of CD4⁺ T cells to proliferate more vigorously as Ag presentation by APCs from both IFN-γ–sufficient and –deficient mice were similar. PLP_91–110–specific CD4⁺ T cells from IFN-γ–deficient mice also produced greater levels of GM-CSF. IFN-γ can regulate peripheral immune response by keeping the Ag-specific CD4⁺ T cell response in check through activation-induced death of T cells (31). Previous studies have shown that IFN-γ loss results in the expansion of autoreactive CD4 T cells and development of severe EAE (32). The increased IL-17 observed in the absence of IFN-γ supports previous reports showing that IFN-γ downregulates the number and function of Th17-specific CD4⁺ T cells (28). Thus, the absence of IFN-γ leads to increase in the frequency of encephalitogenic Th17 cells producing IL-17 and GM-CSF, which results in increased recruitment of inflammatory cells to the brain. We observed a high number of CD4⁺ T cells producing both IL-17 and GM-CSF (IL-17⁺GM-CSF⁺ CD4⁺ T cells) in the brain infiltrate from IFN-γ–deficient mice. Recently, it has been suggested that CD4⁺ T cells producing both IL-17 and GM-CSF are more encephalitogenic than single IL-17⁺ or GM-CSF⁺ cells (33). GM-CSF is required for induction of IL-6 and IL-23, which plays an important role in the generation and maintenance of encephalitogenic Th17 cells. We also observed higher levels of IL-3 and IL-12 expression in brain tissue, as well as in BILs. Tg expression of CNS-specific IL-3 (34) or IL-12 (35) has been shown to cause ataxia-like symptoms with cerebellum-specific pathology in animals. Thus, in the absence of IFN-γ–induced protective factors, increased expression of GM-CSF, IL-3, IL-6, and IL-12 play an important role in activation of the inflammatory cascade in the brain of IFN-γ–deficient mice, which leads to severe demyelination. In addition, we also observed higher of levels of IL-5 and IL-13 in CD4⁺ T cells from IFN-γ–deficient mice, indicating their importance in disease pathogenesis. Both of these cytokines have been shown to have proinflammatory characteristics and can cause CNS-specific inflammatory disease (24, 36).

Our data further demonstrate that presence of specific chemokines in the brain versus spinal cord determines the regional localization of pathology. We observed that brain tissue from IFN-γ–deficient mice with atypical EAE had greater levels of chemokines such as CCL7, CCL8, CCL19, and CCL21, whereas spinal cord from IFN-γ–sufficient mice with classical EAE showed greater levels of CCL5, CXCL9, and CXCL10. CCL7 and CCL8 are known to bind CCR2 and attract inflammatory cells such as monocytes and/or polymorphic nuclear cells (37). CCL19 and CCL21 attract memory CD4⁺ T cells especially Th17 cells through CCR7 receptors (38). CD4⁺ T cells and BILs from IFN-γ–deficient mice had higher expression levels of CCR2, CCR4, and CCR7. In contrast, CD4⁺ T cells and SCILs from IFN-γ–sufficient mice showed higher levels of CXCR3 and CCR5. IFN-γ is known to induce CXCR3 (39), whereas IL-12/IFN-γ induce CCR5 (40). Both these receptors are considered to be specific for Th1 cells. CCL5 attracts immune cells to inflammatory region by binding to CCR5, whereas CXCL9 and CXCL10 do so by binding to CXCR3 receptors. Thus, distinct chemokines/chemokine receptor combinations direct regional pathology in EAE. Once recruited to their specific tissue, cytokines and other chemical mediators such as reactive oxygen species and NO determine the final outcome through modulation of the inflammatory cascade, leading to demyelination and tissue injury.

In normal circumstances, IFN-γ interacts with brain-resident cells to induce neurotrophins (brain-protective factors) such as glial cell line–derived neurotrophic factor, insulin-like growth factor-1, nerve growth factors, brain-derived neurotropic factors, and neurotrophic cytokines, such as ciliary neurotrophic factors and LIF (leukemia tissue factor). These factors protect the myelin sheaths and axons from inflammation-induced injuries (41–43). Previously, in a spinal cord injury model, IFN-γ treatment has been shown to enhance myelination through induction of neurotrophins such as glial cell line–derived neurotrophic factor or insulin-like growth factor-1 (41). However, in the absence of IFN-γ, IL-17–producing autoreactive CD4⁺ T cells expand and recruit monocytes, neutrophils, and other inflammatory cells from the periphery, which leads to increased brain-infiltrating inflammatory cells. These cells secrete inflammatory mediators such as reactive oxygen species and NO, which, in the absence of brain-specific protective factors, attack and destroy the myelin coating on axons. This results in severe demyelination and the onset of neurologic deficits.

In summary, we describe an animal model with severe inflammation and demyelination in DRB1*0301.DQ8.IFN-γ KO mice. Our data suggest a protective role for IFN-γ in peripheral immune cells, as well as in the brain. In the periphery, IFN-γ might regulate expansion of encephalitogenic IL-17/GM-CSF–producing CD4⁺ T cells, whereas in the CNS, IFN-γ might induce glial cells to produce neurotrophin and neurotrophic cytokines that protect the myelin sheath and axons. Thus, our studies indicate an important protective role of IFN-γ in brain inflammation and demyelination.

We thank Julie Hanson and staff for breeding and maintaining the various HLA class II Tg mice used for this study. We also thank Laurie Zoecklein and Mabel Pierce for technical assistance and Lea Dacy for proofreading and editing the manuscript.

The authors have no financial conflicts of interest.