Mice

Wild-type C57BL/6J mice were purchased from KBT Oriental (Breeder: Jackson Laboratory Japan). Foxp3hCD2KI mice, IL-17a-GFP mice and DEREG mice were generated as described previously [26, 27]. All mice were housed in a specific pathogen-free animal facility at Kyushu University with a 12-h light cycle. All the experimental procedures were reviewed and approved by the Kyushu University Animal Experiment Committee, and the care of the animals was in accordance with institutional guidelines.

Mouse model of ischemic stroke

Male mice aged 8–12 weeks and weighing 20–30 g were used for focal brain ischemia experiments [5]. There was no significant difference in weight or age between the wild-type mice and any of the knockout groups. We used a transient MCAO model constructed using an intraluminal suture. This transient suture MCAO model was established as previously described [5]. The mice were anesthetized with isoflurane in a mixture of 70% nitrous oxide and 30% oxygen. During the MCAO procedure, the head temperature was maintained at 36 °C using a heat lamp. Sixty minutes after MCAO, the brain was reperfused by removing the intraluminal suture. Neurological function was evaluated using a previously described four-point scale neurological score method (0, no observable deficit; 1, forelimb flexion; 2, decreased resistance to lateral push without circling; 3, same behavior as grade 2, with circling) [28].

Mouse model of EAEImmunization of MOG33-55 peptides

Emulsions were prepared by mixing 200 μg of myelin oligodendrocyte glycoprotein peptide (MOG33-55) with complete Freund's adjuvant plus M. tuberculosis. For each mouse, 200 μg of the emulsion was injected subcutaneously into 3 sites on the back of Foxp3hCD2KI and WT mice (2–3 months old, female). Pertussis toxins (500 ng/body) were injected intraperitoneally twice, on days 0 and 2.

Clinical score scoring method

0: normal, 0.5: decreased tail tonus, 1: complete tail droop, 1.5: hind limb weakness, 2: incomplete hind limb paralysis, 3: complete hind limb paralysis, 4: complete hind limb paralysis including incomplete forelimb paralysis, 5: death.

Isolation of cells from the adult mouse brain and spinal cord

Mice were perfused with PBS transcardially, and the cerebrum, cerebellum and spinal cord were removed. For isolation of CNS Treg cells, the cerebrum, cerebellum and spinal cord were minced and enzymatically dissociated with 2 mg/ml collagenase D (Roche Applied Science) and 1 mg/ml DNase I in RPMI-1640 for 30 min at 37 °C. The digested tissue was centrifuged and resuspended in 37% Percoll (Cytiva) and centrifuged at 800 × g for 20 min, after which the cell pellets were collected. The isolated cells were stained with a fluorochrome-conjugated antibody. Fluorescein isothiocyanate (FITC), phycoerythrin (PE), peridinin chlorophyll protein-cyanine 5.5 (PerCP-Cy5.5), allophycocyanin (APC), PE-Cy7, and APC-Cy7-conjugated antibodies were purchased from BioLegend, BD or eBioscience. The primary antibodies used were as follows: anti-mouse CD45.2 antibody, anti-mouse CD11b antibody, anti-mouse CD3ε antibody, anti-human CD2 antibody, anti-mouse CD4 antibody, anti-mouse Areg antibody (R&D; BAF989), anti-mouse ST2 antibody, and anti-mouse HTR7 antibody (Novus; 56,352), anti-mouse Rorgt, anti-mouse Gata3, anti-mouse IL-1R, anti-mouse CD25, anti-mouse Lag3, anti-mouse CXCR6, anti-mouse CCR8, anti-mouse CXCR4, anti-mouse CD304 (Neuropilin1), and anti-mouse Helios. The fixable viability dye eFluor 780 (FVD780) was used to remove dead cells. The stained cells were sorted by FACSMelody, LSRFortessa (BD Biosciences) and FlowJo software (Tree Star). The gating strategy for Tregs was FVD−CD45.2+CD11b−CD3+CD4+Foxp3+. For Areg staining, cells were stained after stimulation with eBioscience™ Cell Stimulation Cocktail (Thermo Fisher Scientific) for 3 h.

Immunohistochemistry

Mice were anesthetized by inhalation with isoflurane and perfusion was performed by injecting PBS into the left ventricle. The lumber spinal cord segments were dissected and immersed in 4% paraformaldehyde (PFA) and fixed at 4 °C for 24 h. For paraffin-embedded sections, the spinal cords in cassettes were prepared by the Laboratory for Research Support, Medical Institute of Bioregulation, Kyushu University. Samples were embedded in paraffin and cut into 5 µm sections. Samples on slides were de-waxed in xylol and ethanol, antigen retrieval was performed using boiled citrate buffer and samples were blocked for 1 h at room temperature (25 °C) using Blocking One Histo (nacalai tesque). The sections were incubated with primary antibodies included in 5% BSA/PBS overnight at 4 °C: rabbit anti-MBP (1:300; Cell Signaling Technologies, 78,896), mouse anti-GFAP (1:500; Cell Signaling Technologies, 3670), goat anti-Iba1 (1:300; FUJIFILM Wako Pure Chemical Corporation, 011–27991). After primary antibody incubation, sections were washed twice with PBS containing 0.1% Triton X-100 (PBS-T) and once with PBS, and incubated with the secondary antibody and Hoechst for 30 min at room temperature: donkey anti-rabbit Alexa Fluor 647, donkey anti-mouse Alexa Fluor 546, donkey-goat Alexa Fluor 488 (all 1:300; Invitrogen), then washed with PBS-T and PBS. Sections were sealed with cover glasses using PermaFluor mounting medium (Thermo Fisher, TM-030-FM). The images were taken with BZ-X700 (Keyence) and image analysis was performed using ImageJ. Statistical analysis was performed in Prism9 (Graphpad).

In vivo depletion of Treg cells in EAE mice

DEREG or wild-type littermates were injected with diphtheria toxin (Merck) intraperitoneally at 20 μg/kg on days 13 and 15 after immunization with MOG. At day 17, these mice were sacrificed, and brains and spinal cords were analyzed by flow cytometry.

Transferring Tregs into mice with EAE

EAE was induced in Foxp3hCD2KI mice, and Tregs were sorted from the brain and spinal cord of EAE mice at day 23 after immunization. As a control, Tregs from the lymph nodes of untreated mice were sorted. These Tregs (2 × 104 cells) from the brain or spinal cord of 40 Foxp3hCD2KI mice were injected intravenously into WT recipients 5 days after immunization, and neurological symptoms were observed at the appropriate time points. (Recipients of LN Treg: N = 14, EAE brain Treg: N = 14, EAE SC Treg: N = 16).

For transfer experiment of Tregs after onset of EAE (Fig. S8), EAE was induced in Foxp3hCD2KI mice, and Tregs were sorted from the brain and spinal cord of EAE mice at day 23 after immunization. As control, Tregs from the spleens of 4 CFA-treated mice were sorted. These Tregs (2 × 104 cells) from the brain or spinal cord of 34 Foxp3hCD2KI mice were injected intravenously into WT recipients 17 days after immunization, and neurological symptoms were observed at the appropriate time points. (Recipients of PBS: N = 6, CFA-treated SP Treg: N = 6, EAE brain Treg: N = 7, EAE SC Treg: N = 6).

Chemokine array

EAE was induced in 2–3-month-old C57BL/6 J mice. On day 23, after the mice were euthanized and perfused with PBS, the cerebrums and spinal cords were extracted and crushed by a bead shocker for 20 s at 5,000 rpm three times. The samples were centrifuged for 10 min at 15,000 rpm and 4 °C, after which the supernatants were collected. A chemokine array was performed using the Proteome Profiler Mouse Chemokine Array Kit (R&D; ARY020) according to the manufacturer’s instructions. The chemiluminescence signals on the membranes were detected by ChemiDoc (Bio-Rad). The data analysis was performed with Image Lab (Bio-Rad).

Migration assay

A migration assay was performed using a CytoSelect™ 96-well cell of migration assay kit (Cell Biolabs; CBA-105) according to the manufacturer’s instructions. Tregs isolated from the brains or spinal cords of EAE mice were seeded at 2 × 104 cells/well in 10% FBS medium supplemented with 0.2 mg/ml CCL6, CCL12, or CCL1 (BioLegend). At 4 h after seeding, the migrating cells were lysed with lysis buffer/CyQuant® GR dye, and fluorescence was measured with an EnSpire® Multimode Plate Reader (PerkinElmer).

TCR cloning and IL-2 ELISA

TCR sequences were obtained from scTCR-seq information from brain or spinal cord Tregs and 2D2 or OT-II mice. The DNA was synthesized by eBlocks Gene Fragments (IDT). The α and β chains were linked by the P2A sequence and inserted into the pMX-IRES-GFP vector. The retroviruses generated by these pMX-TCR-IRES-GFP vectors were infected into TG40 cells. GFP-expressing cells were sorted by FACSMedoly and cocultured with bone marrow-derived dendritic cells in media supplemented with MOG33-55 peptide (5 μg/ml), OVA peptide (200 μg/ml) or anti-CD3ε/CD28 antibodies (2 μg/ml) for 24 h. IL-2 in the supernatants was detected by a mouse IL-2 ELISA kit (Invitrogen; 88–7024) according to the manufacturer’s instructions.

RNA sequencingCell isolation and RNA extraction

Bulk RNA-seq of cerebrum, cerebellum and spinal cord tissues from EAE mice: EAE was induced in 2- to 3-month-old C57BL/6J mice. On day 18 and 32 post immunization, after the mice were euthanized, the cerebrum, cerebellum and spinal cord were extracted from 12 mice (day 18: pooled 3 mice × 2 samples, day 32: pooled 3 mice × 2 samples) and homogenized after PBS perfusion. The tissues were homogenized in PBS by beads shocker (TOMY, MS-100). RNA was extracted from a portion of the tissues using an RNeasy Mini Kit (QIAGEN).

Bulk RNA-seq of Tregs: Foxp3hCD2KI mice were induced to experience ischemic stroke or EAE (stroke: male 4 mice × 2, EAE brain and spinal cord: female 4 mice × 2, spleen: 2 mice × 2). At day 28 after ischemic stroke onset or immunization, the mice were perfused with PBS, and the brains and spinal cords were homogenized. The immune cells were isolated using the methods described in “Isolation of cells from the adult mouse brain and spinal cord”. Treg sorting was performed using a FACSAriaIII, and RNA extraction was performed using an RNeasy Plus Micro Kit (QIAGEN).

RNA sequencing

Library preparation was performed using NEBNext PolyA stranded Ultra II Directional RNA Library Prep (New England BioLabs, MA) by Azenta. The libraries were sequenced on Illumina NovaSeq 6000 sequencers (Illumina, San Diego, CA). Sequence reads in FASTQ format were applied to FastQC (v0.11.4) to assess sequence quality. For adapter trimming, Trim Galore! (v0.6.0) was used. The sequence reads were mapped to the mouse reference genome GRCm38/mm10 from Ensembl (release 102) using HISAT2 (version 2.2.1). The mapped reads were counted for genes using featureCounts (v.2.0.0). The data analysis was performed by iDEP.96. Differentially expressed genes (DEGs) obtained between clusters were subjected to Gene Ontology (GO) analysis using Metascape (v3.5).

Single-cell RNA-seqCell isolation

Two- to three-month-old C57BL/6 J mice were subjected to EAE or cerebral infarction. After the mice were euthanized, the cerebrum, cerebellum, spinal cord and lymph nodes were removed by PBS perfusion and homogenized. Staining was performed with the following antibodies: anti-CD45.2, anti-CD11b, anti-hCD2, anti-CD3ε, and anti-CD4. Totalseq hashtag antibody (BioLegend) was then reacted with each sample, and Tregs were sorted with a FACSMelody (BD Biosciences).

Library preparation and NGS

Libraries were prepared according to the protocol of the Chromuim Next GEM Single Cell 5' Reagent Kits v2 (Dual Index, 10 × Genomics, Pleasanton, CA, USA) using a Chromium controller (10 × Genomics). Briefly, GEMs were prepared by loading barcoded Single Cell VDJ 5' Gel Beads, master mix containing cells and RT enzymes, and Oil onto Chromium Next GEM Chip K. Barcoded cDNA was subsequently generated from the GEM. The cDNA was amplified via PCR to generate V(D)J, 5' Gene Expression and Feature Barcode libraries. Sequencing was then performed using a NovaSeq 6000 (Illumina, San Diego, CA, USA) at the Laboratory for Research Support of the Institute of Bioregulatory Medicine, Kyushu University.

Preprocessing

Fastq files were generated from the NGS data using Cell Ranger (v6.1.2) (10 × Genomics). The c.loupe files (gene expression) and v.loupe files (TCR sequence) were obtained from Cell Ranger Multi (mouse genome; Genome Reference Consortium Mouse Build 38, mm10).

4Data analysis

The Loupe Browser (v7.0.1) and Loupe V(D)J Browser (v5.1.0) were used to analyze gene expression and the TCR repertoire. Differentially expressed genes (DEGs) obtained between clusters were subjected to Gene Ontology (GO) analysis using Metascape (v3.5).

Statistics

All the experiments were performed with at least two biological replicates and yielded comparable results except for the scRNA-seq experiment, which was conducted only once. Sample sizes were determined based on previous similar experiments [5] but were not predetermined by any statistical methods. The data are expressed as the mean ± s.e.m. Statistical significance was determined by one-way or two-way analysis of variance (ANOVA) followed by post hoc Tukey’s multiple comparisons test to analyze differences among three or more groups and by unpaired Student’s t test to analyze differences between two groups. P < 0.05 was considered to indicate a significant difference (*p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; ns, not significant). All the statistical analyses were performed with GraphPad Prism (version 9.5.1).