IFN-γ-induced Th1-Treg polarization in inflamed brains limits exacerbation of experimental autoimmune encephalomyelitis

To find brain-specific events for T cells during EAE, we collected CD3⁺ cells from the spleens and brains of MOG-induced EAE mice at the peak of the disease and subjected them to scRNA-seq analysis. First, each cell was unbiasedly clustered and visualized on a t-stochastic neighbor embedding (t-SNE) map. As a result, 10 clusters were automatically defined (Fig. 1A and SI Appendix, Fig. S1). Among them, cluster 4 exhibited high Foxp3 expression and, vice versa, Foxp3 expression was exclusively in cluster 4 (Fig. 1 A and B and SI Appendix, Fig. S1). Considering the results of clustering (Fig. 1 A and B and SI Appendix, Fig. S1) and the tissue of origin (Fig. 1C), automatic unbiased clustering was insufficient to distinguish between splenic Tregs and brain Tregs, or to further subcluster Tregs. Next, based on the tissue of origin (Fig. 1C) and the expression of CD4 and Foxp3 (Fig. 1B), the CD3⁺ cells were classified into splenic conventional Tconvs, splenic Tregs, brain Tconvs, and brain Tregs (Fig. 1D). Among them, brain Tregs significantly up-regulated the expression of 248 genes and 161 genes compared to brain Tconvs and splenic Tregs, respectively (Fig. 1E). Of these up-regulated genes, 64 genes were overlapped (Fig. 1E and Dataset S1) and identified as characteristic genes of brain Tregs during EAE. To elucidate the biological and functional relevance of these genes, we conducted a Gene Ontology (GO) term enrichment analysis using Metascape (41). The top-ranked GO cluster “defense response to protozoan (GO: 0042832)” and components of the cluster “cellular response to interferon-gamma (GO: 0071346)” and “response to interferon-gamma (GO: 0034341),” all of which are related to IFN-γ-dependent responses (Fig. 1F). In addition, 9 genes including Actg1, Socs1, Gbp2, Igtp, Stat1, Gbp3, Gbp5, Gbp6, and Gbp7 were annotated to terms related to IFN-γ, representing the highest number of annotations within the GO terms of this cluster (Fig. 1F). Additionally, “Type II interferon signaling (IFNG) (WP1253)” was also identified as another GO cluster (Fig. 1F), suggesting that brain Tregs express IFN-γ-stimulated genes during EAE.

To validate the gene expression found in the scRNA-seq analysis, we performed quantitative-PCR to compare the expression of the IFN-γ-inducible genes such as Gbp2, Gbp3, Gbp4, Gbp5, Igtp, Iigp1, Irgm1, Irgm2, and Tgtp in Treg from brains and spleens of EAE-induced FDG mice, in which Foxp3⁺ cells are labeled by GFP (42) (Fig. 2 A and B and SI Appendix, Figs. S1A and S2B). GFP⁺ CD4⁺ T cells were collected from spleens and brains at the peak of the disease and were evaluated IFN-γ-inducible gene expression (Fig. 2A and SI Appendix, Fig. S2A). Of the genes evaluated, Gbp3, Gbp5, Iigp1, Irgm1, Irgm2, and Tgtp were significantly more highly expressed in the brain GFP⁺ CD4⁺ T cells (brain Tregs in FDG mice) compared to the splenic GFP⁺ CD4⁺ T cells (splenic Tregs in FDG mice) (Fig. 2A and SI Appendix, Fig. S2A). To assess whether IFN-γ is involved in the expression of these genes in brain Tregs during EAE, we next administered an IFN-γ neutralizing antibody to FDG mice a day before EAE induction and evaluated IFN-γ-related gene expression at the peak of the disease (Fig. 2B and SI Appendix, Fig. S2B). Although brain Tregs expressed higher expression levels of Gbp3, Gbp5, Iigp1, and Irgm2 than splenic Tregs (Fig. 2A), the IFN-γ neutralizing antibody treatment significantly reduced the expression levels (Fig. 2B). In addition, expression of Gbp2, Gbp4, Irgm1, Igtp, and Tgtp in brain Tregs also showed IFN-γ dependency, whereas expression of these genes was not significantly higher than that in splenic Treg or brain GFP⁻ CD4⁺ T cells (brain Tconvs in FDG mice) (SI Appendix, Fig. S2 A and B). Collectively, these data suggest that brain Tregs have signs of IFN-γ stimulation during EAE. In addition, to assess the differential sensitivity of Tregs in the CNS compared to lymphoid tissues, we isolated Tregs (GFP⁺ cells) from the CNS and spleen of FDG mice and stimulated them with IFN-γ. We then examined the phosphorylation of STAT1 (SI Appendix, Fig. S2C). Approximately 20% of CNS Tregs responded to IFN-γ with phosphorylation of STAT1, whereas in the spleen, 40% of Tregs exhibited phosphorylation of STAT1. These data indicate that the sensitivity of Tregs to IFN-γ in the naïve state may differ between the CNS and lymphoid tissues.

Next, we investigated which cell types serve as the source of IFN-γ that can stimulate brain Tregs during EAE (Fig. 3). Major sources of IFN-γ are T cells, particularly the Th1 subset (43) and CD8⁺ T cells (44), as well as NK cells (45, 46). In addition, other immune cell types such as B cells, dendritic cells, and macrophages can also produce IFN-γ (47–52). EAE was induced in a strain of IFN-γ reporter mice, where YFP is expressed in the endogenous IFN-γ locus (53). Subsequently, the brain cells were harvested at the peak of the disease, restimulated with PMA/ionomycin, and analyzed for YFP frequency in each immune cell type (Fig. 3A). Among CD3⁺ cells, CD11b⁺ cells, NK1.1⁺ cells, B220⁺ cells, and other immune cells of the EAE-induced IFN-γ reporter mice, CD3⁺ cells and NK1.1⁺ cells exhibited approximately 30% positivity for YFP expression (Fig. 3B), indicating that T cells and NK cells have the highest ability of IFN-γ production among the immune cell types tested in the inflamed brains. Considering the cell number of each population, we found that over half of YFP⁺ cells were CD3⁺ cells (Fig. 3C). In addition, to address which CD4⁺ T cells produce IFN-γ, including autoreactive Tconvs and Tregs, and whether this phenotype precedes disease, we investigated IFN-γ expression in brain T cell subsets at the preonset and peak stages of EAE (Fig. 3 D and E). At these time points, we separated brain T cells into Tconv and Treg based on Foxp3 expression. Additionally, we used MOG_35-55-tetramer to distinguish autoreactive T cells and nonautoreactive T cells. Due to the low number of MOG_35-55-tetramer binding (Tet⁺) cells at day 10, we isolated Tetramer-positive cells only at the peak time point. Both Tconvs and Tregs showed higher IFN-γ expression at the peak stage compared to the preonset stage (Fig. 3E). Notably, Tet⁺ Tconvs exhibited the highest IFN-γ expression (Fig. 3E). While Tet⁺ Tregs also expressed higher levels of IFN-γ compared to Tet^- Tregs, their expression levels were not as high as those observed in Tet⁺ Tconvs (Fig. 3E). These data suggest that IFN-γ expression from T cells increases along with the rise in autoreactive T cells as the disease progresses. Taken together, these data suggest that T cells are the predominant source of IFN-γ in the brains during EAE.

We next examined whether the expression of IFN-γ-inducible genes in brain Tregs during EAE requires T cell-derived IFN-γ using CD4-Cre/Ifng^fl/fl mice (Fig. 3 F–H), in which IFN-γ production of T cells was specifically disrupted (SI Appendix, Fig. S3) (54). EAE was induced in CD4-Cre/Ifng^fl/fl mice and the control Ifng^fl/fl mice. Subsequently, we conducted scRNA-seq analysis on brain T cells at the peak of the disease to compare expression levels of the 64 characteristic genes in brain Tregs of EAE-induced CD4-Cre/Ifng^fl/fl mice and the control Ifng^fl/fl mice (Figs. 1E and 3 F–H). We found that 18 out of the 64 characteristic genes of brain Tregs were significantly reduced in the CD4-Cre/Ifng^fl/fl mice (Fig. 3H and Dataset S2). The 18 genes included the IFN-γ-inducible genes such as Gbp2, Gbp3, Gbp5, Gbp6, Gbp7, Igtp, Iigp1, Socs1, and Stat1 (Fig. 3H and Dataset S2), indicating that T cell-derived IFN-γ stimulates brain Tregs during EAE.

Next, we assessed the role of IFN-γ in EAE pathology. After the peak of EAE, the disease scores decreased in wild-type (WT) mice (Fig. 4A). As described previously (9), mice conventionally lacking IFN-γ receptor (Ifngr1^−/− mice) did not display recovery after the peak of the disease (Fig. 4A), suggesting that IFN-γ plays an important role in the disease amelioration. To test the role of T cell-derived IFN-γ, we induced EAE in CD4-Cre/Ifng^fl/fl mice and the control Ifng^fl/fl mice (Fig. 4B). Similar to conventional Ifng^−/− mice, CD4-Cre/Ifng^fl/fl mice also did not show recovery after the peak of the disease (Fig. 4B), suggesting the importance of T cell-derived IFN-γ in EAE amelioration. It remained unanswered as to what extent IFN-γ signaling in Tregs contributes to the EAE recovery. To test this directly, we generated Treg-specific IFNγR conditional KO mice (SI Appendix, Fig. S4A). Foxp3-Cre/Ifngr1^fl/fl mice specifically lost Ifngr1 mRNA expression in Tregs (SI Appendix, Fig. S4B). Then, EAE was induced in Foxp3-Cre/Ifngr1^fl/fl mice (Fig. 4C). It was noteworthy that Foxp3-Cre/Ifngr1^fl/fl hardly exhibited the recovery of the disease (Fig. 4C), which was similar to that observed in Ifngr1^−/− mice and CD4-Cre/Ifng^fl/fl mice (Fig. 4 A and B). In addition, in the EAE-developed brain, Foxp3-Cre/Ifngr1^fl/fl mice exhibited increased production of IL-17 and GM–CSF from CD4+ T cells compared to ctrl Ifngr1^fl/fl mice (SI Appendix, Fig. S4C). Given that IL-17 and GM–CSF are shown to be effector molecules involved in the progression of EAE (55, 56), these data indicate that the selective loss of the IFN-γ receptor in Tregs enhances the CD4 effector response that exacerbates EAE.

We next explored the biological significance of IFN-γ-stimulation of Tregs in the brains of EAE. It is known that IFN-γ stimulation on Tregs induces their polarization into Th1-Tregs that express the Th1-lineage transcription factor T-bet in addition to the Treg-specific transcription factor Foxp3 (23, 29). Therefore, we hypothesized that IFN-γ stimulation in brain Tregs leads to their polarization into Th1-Tregs. To assess this possibility, we examined the role of IFN-γ in Treg polarization into Th1-Tregs (hereafter called Th1-Treg polarization) in the brains of WT or Ifngr1^−/− FDG mice (Fig. 5 A and B). The chemokine receptor CXCR3 is highly expressed on Th1-Treg (23, 29). We observed CXCR3 expression on brain Tregs (GFP⁺ CD4⁺ T cells) in EAE-induced WT FDG mice (Fig. 5A). In sharp contrast, the CXCR3 expression on brain Tregs in the Ifngr1^−/− FDG mice was strikingly reduced (Fig. 5A). Notably, the CXCR3 expression in GFP⁻ CD4⁺ cells (brain Tconvs) from EAE-induced Ifngr1^−/− FDG mice was comparable to that in brain Tconvs of WT FDG mice (Fig. 5A). Furthermore, intracellular T-bet expression in brain Tregs of Ifngr1^−/− FDG mice was significantly less than that in those of WT FDG mice (Fig. 5B), suggesting that IFN-γ is specifically involved in Th1-Treg polarization but not in Tconv polarization into Th1.

To directly prove the presence of Th1-Tregs in brans during EAE, we utilized Foxp3-Cre/Tbx21-Flp/VeDTR mice and investigated whether Th1-Tregs were induced in the brains during EAE (Fig. 5 C and D). YFP⁺ cells in the brain were increased in EAE-developed mice (Fig. 5C). To confirm that these YFP⁺ cells in the brains during EAE can be regarded as Th1-Tregs, we examined Foxp3 and T-bet expression in YFP⁺ cells in the brain of EAE-developed mice. YFP⁻ cells were primarily composed of Foxp3 single positive, T-bet single positive, and double negative cells, whereas YFP⁺ cells were predominantly Foxp3 and T-bet double positive (Fig. 5D). Next, when comparing the frequency of YFP⁺ cells in the brain, spleen, inguinal lymph nodes (ILN), liver and lung of naïve and EAE-developed Foxp3-Cre/Tbx21-Flp/VeDTR mice, the frequency of YFP⁺ cells in brains of the EAE-developed mice were significantly increased but not those in other tissues (Fig. 5E). Moreover, the absolute numbers of YFP⁺ cells in the brains robustly increased during EAE (Fig. 5F), indicating that the frequency and number of Th1-Tregs are specifically increased in the inflamed brains during EAE. We next analyzed the localization of Th1-Tregs in the brains of EAE-induced Foxp3-Cre/Tbx21-Flp/VeDTR mice by immunohistochemistry (Fig. 5F). White matters (WMs) contain abundant myelin proteins in the healthy cerebellum of naïve mice (57). In contrast, the cerebellum of EAE-developed brains becomes the sites of demyelinating lesions (58, 59). We found massive CD4⁺ T cell infiltration into the demyelination area (Fig. 5G). It was noteworthy that Th1-Treg, which can be stained by anti-DTR (28), was also found in the demyelination area (Fig. 5G). Moreover, Th1-Tregs were localized within CD4⁺ cell clusters near the vasculatures (Fig. 5G), similar to the previous study suggesting that Tregs are primarily localized within the perivascular cluster of T cells in the EAE brain (60). We next assessed whether the Th1-Treg polarization depends on IFN-γ by comparing Th1-Treg frequency in the inflamed brains of WT and Ifngr1^−/− Foxp3-Cre/Tbx21-Flp/VeDTR mice (Fig. 5H). The frequency of YFP⁺ CD4⁺ T cells from EAE-induced Ifngr1^−/− Foxp3-Cre/Tbx21-Flp/VeDTR mice were significantly reduced compared with those from the WT mice (Fig. 5H). Collectively, these data indicate that IFN-γ-dependent Th1-Treg polarization in the inflamed brains during EAE.

Next, we assessed the effect of Th1-Treg depletion on EAE severity (Fig. 6). DT treatment in EAE-developed Foxp3-Cre/Tbx21-Flp/VeDTR mice effectively depleted YFP⁺ cells in the brains (Fig. 6A). When evaluating the disease recovery, EAE scores significantly increased post DT treatment (Fig. 6B), indicating the importance of Th1-Tregs in ameliorating EAE. Moreover, IL-17 and GM–CSF production in CD4⁺ T cells were also increased by DT treatment (Fig. 6C), which was similar to that observed in Foxp3-Cre/Ifngr1^fl/fl mice (SI Appendix, Fig. S4C). During EAE, WMs are shown to be inflamed due to the infiltration of CD11b⁺ cells (61). Compared with WMs from naïve mice, those from EAE-induced Foxp3-Cre/Tbx21-Flp/VeDTR mice were infiltrated by CD11b⁺ cells (Fig. 6D). It was noteworthy that the area of inflammatory WM lesions in the cerebellum was further expanded in DT treated-Foxp3-Cre/Tbx21-Flp/VeDTR mice (Fig. 6D), suggesting Th1-Treg-dependent suppression of CD11b⁺ cell infiltration into WMs during EAE.

Finally, we investigated the effect of Th1-Treg depletion on the status of CD11b⁺ cells in inflamed brains (Fig. 7). We analyzed which type of CD11b⁺ cells increased by Th1-Treg depletion during EAE. (Fig. 7 A–C). We compared cell numbers of neutrophils (CD11b⁺Ly6G⁺), infiltrated macrophages/monocytes (Mfs/Mos) (CD11b⁺Ly6G⁻CD45^highP2RY12⁻) and microglia (CD11b⁺Ly6G⁻CD45^lowP2RY12⁺) in the brains of PBS- or DT-treated EAE-developed Foxp3-Cre/Tbx21-Flp/VeDTR mice (Fig. 7A). DT treatment significantly increased the number of neutrophils and infiltrated Mfs/Mos. On the other hand, the number of microglia (CD11b⁺Ly6G⁻ CD45^lowP2RY12⁺) was not changed by DT treatment (Fig. 7A). Given that neutrophils and Mfs/Mos are known to promote EAE pathology progression (62–64), Th1-Tregs may suppress the increase of these cells in inflamed brains. Regarding activation status, macrophages that contribute to EAE development or neurotoxicity express proinflammatory phenotype markers such as CD40, CD80, and CD86 (65, 66). When we assessed the levels of CD40, CD80, and CD86 on Mfs/Mos and microglia, all surface activation markers tested were unchanged in microglia between both conditions (Fig. 7B) On the other hand, we found that CD40 and CD80 expression on Mfs/Mos was significantly increased by DT treatment (Fig. 7B). The increments of CD40 and CD80 were also observed in the mRNA expression levels (Fig. 7C). In addition, expression of inflammation mediators encoded by Il6 and Tnf genes in Mfs/Mos was increased by DT treatment (Fig. 7C). Collectively, these data suggest that Th1-Tregs suppress the accumulation of neutrophils and activated Mfs/Mos into WMs to ameliorate EAE.

C57BL/6 N mice were purchased from Japan SLC. Foxp3-Cre/Tbx21-Flp/VeDTR mice, Foxp3-Cre mice, CD4-Cre mice, IFN-γ reporter mice, Ifng^fl/fl mice, and FDG mice were described as previously (28, 42, 53). All animal experiments were approved by the Animal Research Committee of Research Institute for Microbial Diseases in Osaka University.

The T7-transcribed Ifngr1_gRNA1 and gRNA2 PCR products, which were amplified by using KOD FX NEO (Toyobo) and the primers (Ifngr1flox_gRNA1 5′- TTAATACGACTCACTATAGGcaatcaccgaggcagagtgtGTTTTAGAGCTAGAAATAGCAAGTTAAAAT-3′; Ifngr1flox_gRNA2 5′- TTAATACGACTCACTATAGGactattgaagtgacccaaaaGTTTTAGAGCTAGAAATAGCAAGTTAAAAT-3′) were used as the subsequent generation of Ifngr1_gRNA1 and Ifngr1_gRNA2. MEGAshortscript T7 (Life Technologies) was used for the generation of these gRNAs. Cas9 mRNA was generated by in vitro transcription (IVT) using the mMESSAGE mMACHINE T7 ULTRA kit (Life technologies) and the template that was amplified by PCR using pEF6-hCas9-Puro and the primers T7Cas9_IVT_F and Cas9_R (28), and gel-purified. The synthesized gRNA and Cas9 mRNA were purified using the MEGAclear kit (Life Technologies). For generation of the targeting fragment for the floxed Ifngr1 allele, the Ifngr1 gene was isolated from genomic DNA that was extracted from C57BL/6N embryonic fibroblasts by PCR using KOD FX NEO (Toyobo) and primers (Ifngr1_flox_LA_F 5′- gtcgaCCCAGAATATGTCCAGACTACTTGAGTCTGCAGTTCTGGTTTTCAGAGCAAAGAAGTGAACCATT-3′; Ifngr1_flox_LA_R 5′- gaattctctgcctcggtgattgAACTTCACCTCTGACTTTGTCTTTAG GAAAGTATGATTAGTGATAAAT -3′; Ifngr1_flox_MA_F 5′- gaattcATAACTTCGTATAGCATACATTATACGAAGTTATgtgtAGGTAAAGAGCTTCTTGTCTGAGATT-3′; Ifngr1_flox_MA_R 5′- acgcgtgggtcacttcaatagtACTTATGTTGGTTTTTTTCTTATCAG AATCTATGACACATGAGTCTTA -3′; Ifngr1_flox_RA_F 5′- acgcgtATAACTTCGTATAGCATACATTATACGAAGTTATaaaaTGGACTTAATTGCCAACACTGGCCAG-3′; Ifngr1_flox_RA_R 5′- gcggccgcGAAGTAACCTGCCCAGTGTGGAAAGACTACGAAGAGGAGGTGACAACCCCTGACCCGAAGAA-3′). The targeting fragment was constructed from a 0.5-kb fragment of Ifngr1 genomic DNA containing exon 4 and loxP site-containing 0.5-kb subfragments using restriction enzymes in pBluescript. The vectors were amplified and coinjected into the embryos with the Cas9-encoding mRNA, Ifngr1_gRNA1, and Ifngr1_gRNA2 to obtain Ifngr1^fl/+ pups. Ifngr1^fl/+ mice were further crossed with Foxp3-Cre mice to generate Foxp3-Cre Ifngr1^fl/fl mice.

Antibodies and tetramer used in this study are described in Dataset S3. Human HB-EGF (DTR) antibody (R&D systems) was conjugated with Alexa Fluor® 488 using Fluorescent Protein Labeling Kits (Invitrogen).

9 to 16 wk old Age- and sex-matched mice were subcutaneously immunized on the back with 200 μL emulsion consisting of 100 μL PBS containing 2 mg/mL MOG peptide _35-55 (ProSpec) and 100 μL complete Freund’s adjuvant (CFA) containing 5 mg/mL heat-killed Mycobacterium tuberculosis H37 RA (Chondrex). The mice received an i.p. injection of 500 μg Pertussis toxin (Millipore) immediately after the MOG/CFA immunization, and again 2 d later. The clinical scores were assessed using Mouse EAE Scoring Guide in Hooke Laboratories, LLC (https://hookelabs.com/).

Cell preparation from the spleen, lymph node, lung, and liver was described previously (28). To prepare cells from the brain, mice were killed and transcardially perfused with 10 mL PBS. Subsequently, the brain was minced to 2 to 3 mm with scissors in C tubes (Miltenyi) containing 2.5 mL HBSS and mechanically dissociated using MACS Octo Dissociator (Miltenyi), sequentially employing preset programs B and D. After dissociation, the cells were centrifuged for 5 min at 2,000 rpm. The cell pellet was resuspended in ACK buffer and incubated for 2 min at room temperature and then washed with HBSS. The washed cells were resuspended in HBSS with 40% Percoll (Sigma-Aldrich) and centrifuged for 20 min at 2,380×g to eliminate floating debris. Finally, the resulting pellet was rinsed with HBSS. Prepared cells were resuspended in the appropriate buffer for each experiment.

For surface staining, cells were stained with antibodies in 2% BSA in PBS for 15 min on ice and then washed twice in 2% BSA in PBS. For cytokine staining, Cytofix/Cytoperm Fixation/Permeabilization Kit (BD) was used following the manufacturer’s instruction. For intranuclear staining, Foxp3/Transcription Factor Staining Buffer Set (eBioscience) was used following the manufacturer’s instruction. For STAT1 staining, BD Cytofix Fixation Buffer and BD Phosflow Perm Buffer III were used following the manufacturer’s instruction for fixation and permeabilization, respectively. FACS Aria III (BD) was used for data acquisition and cell sorting. For live/dead cell discrimination, DAPI (NacaraiTesque), 7AAD (BD Pharmingen), or Fixable Viability Dye eFluor 450 (Invitrogen) were used. Some experiments used Precision Count Beads (BioLegend) to obtain cell numbers. The acquired data were analyzed using FlowJo ver. 10.8.0 (BD).

Single-cell suspensions of each sample from mice were stained with anti-Lin (B220, CD11b, and CD11c), anti-CD3 antibodies, Totalseq-C Hashtag (BioLegend), and 7-AAD. Subsequently, living Lin⁻ CD3⁺ cells were sorted using FACS Aria III. After sorting, the Hashtag-labeled single cells of six samples were pooled together. For library preparation, Chromium Next GEM Single Cell Chromium Next GEM Single Cell 5′ Library and Gel Bead Kit v2 (10× Genomics) were used following the manufacturer’s instructions. The libraries were sequenced on NovaSeq 6000 (Illumina) in a 28+91 base paired-end mode to yield a minimum of 20,000 reads per cell for 10,000 cells. The resulting raw data were processed by Cell Ranger 5.0.0.

The processed data obtained from Cell Ranger were loaded into SeqGeq software ver 1.6 (BD), normalized to count per 10,000, and performed the quality control following the instruction (https://docs.flowjo.com/seqgeq/quality-control/). Subsequently, Seurat plugin ver. 2.6 was used for unsupervised clustering and t-SNE visualization. Metascape (https://metascape.org) was used for the Gene Ontology enrichment analysis of differentiallyexpressed genes (DEGs).

For quantitative PCR, total RNA was extracted using the RNeasy Mini Kit (Qiagen), after which RNA was reverse transcribed using the Verso cDNA synthesis Kit (Thermo Scientific) following the manufacturer’s instructions. Quantitative PCR was performed using GoTaq qPCR Master Mix (Promega) and CFX Connect (Bio-Rad). The expression of mRNA was normalized to that of Actb mRNA. The primer sequences are described in Dataset S4.

For PMA/ionomycin stimulation, cells were resuspended in RPMI1640 (Nacalai Tesque) supplemented with 10% heat-inactivated FCS (Gibco), 100 U/mL Penicillin/Streptomycin (Nacalai Tesque), and 50 mM 2-ME (Nacalai Tesque), and stimulated with 50 ng/mL PMA (Nacalai Tesque) and 1 μg/mL ionomycin (Nacalai Tesque) in the presence or absence of 0.67 μL/mL GolgiStop (BD) for 4 h at 37 °C. For IFN-γ stimulation, cells were resuspended in PBS containing 5% heat-inactivated FCS (Gibco), and stimulated with 100 ng/mL recombinant mouse IFN-γ (PeproTech) for 30 min at 37 °C.

For immunohistochemical staining, mice were transcardially perfused with 10 mL PBS. Their brain was then embedded in FSC 22 Frozen Section Media (Leica) and the frozen specimen was sectioned at 10 µm using the CM1860 UV (Leica). The sections were fixed in cold acetone for 10 min at −20 °C. After fixation, they were incubated in Blocking Buffer (PBS with 0.1% BSA, 1% mouse serum, 1% donkey serum, and 1% goat serum) for 1 h at RT. They were then stained with the antibodies in Blocking Buffer overnight at 4 °C and subsequently washed three times with PBS. The stained sections were mounted using ProLong Diamond Antifade Mountant (Invitrogen). For myelin staining, FluoroMyelin Red (Invitrogen) was applied before the antibody staining following the manufacturer’s instruction. The images were observed using FV3000 (Olympus), and the lesion area was quantified with ImageJ software.

For statistical significance assessment, unpaired two-tailed Student’s t test was performed to compare the two groups. One-way ANOVA with a post Tukey’s test was performed to compare multiple groups. One-way ANOVA with a post Dunnett’s test was performed for many-to-one comparison. Statistical significance values are indicated as follows: ns; not significant, *P < 0.05, **P < 0.01, and ***P < 0.001.

M.O. and M.Y. designed research; M.O., A.K., and D.O. performed research; D.O. and M.S. contributed new reagents/analytic tools; M.O., A.K., D.O., N.K., T.K., and M.Y. analyzed data; and M.O., M.S., and M.Y. wrote the paper.

The authors declare no competing interest.