Mice and cells

CD45.1 + congenic mice (B6. SJL −PtprcaPepcb/BoyJ), Foxp3YFP − Cre (B6.129(Cg) − Foxp3tm4(YFP/icre)Ayr/J), Foxp3eGFP (B6. Cg − Foxp3tm1Mal/J), and Notch2flox/flox mice (B6.129S − Notch2tm3Grid/J) were purchased from Jackson Laboratory (Bar Harbor, Maine, USA). All the mice were kept under pathogen − free conditions in the Laboratory Animal Resource Center at the Gwangju Institute of Science and Technology and the Animal Center for Pharmaceutical Research at Seoul National University. Notch2flox/flox mice were bred with Foxp3YFP − Cre transgenic mice, and Foxp3YFP − Cre:Notch2flox/flox mice were bred with Foxp3YFP − Cre:Notch2flox/wt mice to generate Foxp3YFP − Cre:Notch2flox/flox and Foxp3YFP − Cre:Notch2flox/wt mice, respectively. All the mice used in the experiments were aged 6−10 weeks and were used in accordance with protocols approved by the Animal Care and Ethics Committees of Seoul National University. Mouse primary CD4+ T cells were isolated from the spinal cords, spleens, and inguinal lymph nodes of the mice via a FACSAria III Cell Sorter (BD Biosciences, Franklin Lakes, New Jersey, USA) or an EasySep Mouse CD4+ T cell isolation kit (Stem Cell, Vancouver, Canada). All animal experiments were approved by the Institutional Animal Care and Use Committee at Seoul National University and the Gwangju Institute of Science and Technology.

Study design and participants

The study was approved by the Seoul National University Hospital’s Institutional Review Board (protocol number H−2012−115−1183). All patients provided informed written consent. The ages, sexes, and disease characteristics of all participants are provided in Supplementary Fig. 13A.

Antibodies and proteins

Anti−mouse CD3 (17A2) and anti−mouse CD28 (145−2C11) antibodies were purchased from Bio X Cell (Seoul, Korea). The following were purchased from eBioscience (San Diego, California, USA): FITC−conjugated anti−mouse TCRβ (H57−597); Alexa Fluor 488−conjugated anti−mouse Foxp3 (FJK−16s), anti−human CD4 (OTK4), anti−human GITR (eBioAITR), and anti−human PD−1 (MIH4); PE−conjugated anti−human/mouse Notch2 (16F11) and anti−mouse CD25 (PC61.5) and CD69 (H1.2F3); Peridinin chlorophyll protein cyanine 5.5 (PerCP−Cy5.5)−conjugated anti−mouse CD3ε (145 − 2C11), IFN−γ (XMG1.2), CD62L (MEL−14), CD45.2 (104), and anti−human Foxp3 (RCH101); Allphycocyanin (APC)−conjugated anti−mouse Notch1 (22E5), anti−mouse/rat IL−17a (ebio17B7), anti−mouse CTLA4 (UC10−4B9) and ICOS (7E.17G9); anti−human/mouse CD44 (IM7) and CD45.1 (A20), and anti−human/mouse CTLA4 (UC10−4B9); PE−cy7−conjugated anti−mouse GITR (DTA−1), OX40 (OX−86), PD−1 (J43), CD4 (GK 1.5), and anti−human/mouse LAP (TW7−16B4); efluor450−conjugated anti−mouse CD4 (GK 1.5) and anti−human/mouse TNFα (MP6−XT22); and Brilliant Ultraviolet 737−conjugated anti−human CD8a (RPA−T8). The following antibodies were purchased from BioLegend (San Diego, California, USA): PE−conjugated anti−mouse Tim3 (5D12), Brilliant Ultraviolet 395−conjugated anti−human CD3ε (UCHT1), and PerCP−Cy5.5−conjugated anti−mouse Lag3 (C9B7W). The following antibodies were purchased from BD Pharmingen: APC−conjugated anti−mouse Notch2 (HMN2−35); Brilliant Violet 785−conjugated anti−human PD−1 (NAT105); Brilliant Violet 605−conjugated anti−human CD336 (Tim3, F38−2E2); Brilliant Violet 711−conjugated anti−human CD223 (Lag3, 11C3C65); and Alexa Fluor 700−conjugated anti−human CD4 (RM4−5). The following antibodies were purchased from Santa Cruz (Dallas, Texas, USA): Alexa 488−conjugated anti−mouse/rat/human Jag1 (E−12).

Isolation of mononuclear leukocytes (MNLs) from the spinal cords of EAE mice

Spinal cords were isolated from EAE mice after perfusion with PBS. Each spinal cord sample was chopped in PBS containing collagenase D (2.5 mg/ml; Sigma‒Aldrich, Burlington, USA) and DNase I (1 mg/ml; Roche, Grenzacherstrasse, Basel, Switzerland) and then incubated in a 37 °C shaking incubator for 1 h. After incubation, spinal cord homogenates were produced by passing the tissue (in PBS) through a 70 μm nylon strainer, followed by centrifugation. The supernatants were discarded, and the remaining cells were purified on a Percoll (GE Healthcare, Chicago, Illinois, USA) gradient. Briefly, the cell pellets were suspended in 4 ml of 37% (v/v) Percoll, and a layer was carefully formed beneath the mixture using 4 ml of 70% (v/v) Percoll. This sample was then centrifuged at 1500 × g for 30 min with no brake. After centrifugation, the layer containing the cells was harvested and washed with PBS.

Isolation of MNLs from the colons of colitis model mice

The colons were removed from the mice with DSS−induced colitis, chopped in 1× HBSS containing 5 mM EDTA and 1 mM DTT, and then incubated for 30 min in a 37 °C shaking incubator. After incubation, the tissues (in 1× HBSS) were passed through a 100 μm nylon strainer, and this process was repeated twice. The remaining colon pieces were incubated in 1× PBS containing collagenase D, DNase I, and dispase II, followed by incubation for 30 min in a 37 °C shaking incubator. After incubation, the homogenates were produced by passing the tissue (in 1× PBS) through a 70 μm nylon strainer, and this process was repeated twice. After centrifugation, the supernatants were discarded, and the remaining cells were purified on a Percoll gradient. Briefly, the cell pellets were suspended in 40% (v/v) Percoll. Next, 4 ml of 80% (v/v) Percoll was carefully layered beneath this solution. The sample was then centrifuged at 1500 × g for 30 min with no brake. After centrifugation, the layer containing the cells was harvested and washed with PBS.

Isolation of MNLs from the colon of IBD patients

Colon biopsies in 1× PBS were immediately transported to the laboratory. Next, the tissues were chopped in 1× PBS containing collagenase D (2.5 mg/ml; Sigma‒Aldrich, Burlington, St. Louis, USA) and DNase I (1 mg/ml; Roche), followed by incubation for 30 min at 37 °C in a shaking incubator. After incubation, the tissues (in 1× PBS) were passed through a 100 μm nylon strainer, and this process was repeated twice. After centrifugation, the supernatants were discarded, and the remaining cells were purified on a Percoll gradient as described in the previous section.

siRNA electroporation

Notch2 siRNA (1320001) was purchased from Thermo Fisher Scientific (Waltham, Massachusetts, USA). The control siRNA (ss−1011) was purchased from Bioneer (Daejeon, Korea). The siRNA (1 μM) was electroporated via a 4D−Nucleofector (Lonza, Muenchensteinerstrasse, Basel, Switzerland). After 48 h, the electroporated cells were harvested and analyzed via flow cytometry.

MOG38 − 49 −IAb tetramer staining

MOG38 − 49 −IAb monomers were obtained from the NIH Tetramer Facility. PE−conjugated streptavidin was purchased from BioLegend. The tetramers were formed by individually incubating class II molecules with labeled streptavidin for 12–18 h at room temperature at a molar ratio of 8:1. PE−conjugated MOG38 − 49 −IAb tetramer staining was performed at RT for 30 min.

Confocal immunofluorescence analysis

The samples were fixed in 10% formalin in PBS and then washed with 1× PBS. The samples were then placed in 1× PBS containing 30% sucrose and incubated overnight at 4 °C. Next, the samples were embedded in either optimal cutting temperature compound or formalin−fixed paraffin−embedded (FFPE) blocks. FFPE blocks were subjected to antigen retrieval and permeabilization with Triton X−100. After sample preparation, images were acquired via a TCS8 (Leica, Wetzlar, Germany) confocal microscope or a superresolution confocal microscope (LSM900, ZEISS, Oberkochen, Germany).

Confocal live and 3D imaging analysis

Labeled cells were cultured in cover glass−bottom confocal dishes (SPL, Gyeonggi−do, Korea) coated with anti−CD3ε (2 μg/ml) and anti−CD28 (2 μg/ml) antibodies. After 1 day, the confocal dish was placed on a confocal scope TCS8 stage under temperature−, CO2-, and humidity−controlled conditions. For 3D imaging, a 4% formaldehyde−fixed plate was placed on a superresolution LSM900 confocal microscope.

Quantitative RT‒PCR

Total RNA was extracted from naïve CD4+ T cells and infiltrated CD4+ T cells isolated from the spleens and spinal cords of EAE mice via a Qiagen RNeasy Micro Kit (QIAGEN, Hilden, Germany). Complementary DNAs were generated via TOPscript RT DryMIX (Daejeon, Korea). The primers used were designed via PrimerBank (https://pga.mgh.harvard.edu/primerbank/) and synthesized by Macrogen (Seoul, Korea). Quantitative RT‒PCR was performed using SYBR Green Master Mix (Enzynomics, Daejeon, Korea) and an Mx3005p quantitative PCR system (Stratagene, La Jolla, California, USA). Glyceraldehyde−3−phosphate dehydrogenase or β−actin served as internal controls. Analysis was performed via the 2−ΔΔCT method, except for Hes5 mRNA (which was not detected in control cells), for which the 2−ΔCT method was employed.

Flow cytometry analysis

MNLs were isolated from the spinal cord, splenocytes, and lymphocytes of 6– to 10−week−old mice. To detect cell surface antigens, the cells (2 × 105) were labeled with fluorochrome−conjugated primary antibodies (30 min at RT). To identify the IFN−γ, IL−17a, and Foxp3−expressing cell populations, isolated CD4+ T cells (2 × 105) were cultured for 4 h with phorbol myristate acetate (PMA; 50 ng/ml), ionomycin (1 μg/ml), and brefeldin A (1×). To detect intracellular antigens, surface−stained cells were fixed and permeabilized with permeabilization buffer (eBioscience) or Foxp3/Transcription Factor Staining Buffer (eBioscience). The stained cells were analyzed on a FACSCanto II instrument (BD Biosciences) or a high−end performance flow cytometer (LSR Fortessa X−20; BD Biosciences).

EAE model and adoptive transfer

Mice (8–10 weeks old) were immunized subcutaneously (s.c.) with 200 μg of MOG p35−55 peptide (Peptron, Daejeon, Korea) mixed with supplemented complete Freund’s adjuvant (Sigma‒Aldrich), followed by intraperitoneal (i.p.) injection of Bordetella pertussis toxin (400 ng per mouse; List Biological Laboratories, Campbell, California, USA) on Days 0 and 2. Clinical EAE was graded on a scale of 1−5 on the basis of established criteria54: 0, no observable symptoms; 1, tail paralysis; 2, partial hind limb weakness; 3, hind limb paralysis; 4, hind paralysis and forelimb weakness; and 5, moribund or dead. For adoptive transfer, spinal cord−infiltrated CD4+ T cells were isolated from EAE−induced Foxp3eGFP mice (CD45.2) as previously described. Spinal cord lymphocytes were sorted on the basis of the expression of Notch2 and Foxp3 (i.e., CD4+Notch2+ T cells and CD4+Notch2+Foxp3lo T cells) via a FACS Aria III Cell Sorter (BD Biosciences). At 10−12 days post−EAE induction, the mice (CD45.1) were injected with PBS, CD45.2+CD4+Notch2− T cells, or CD45.2+ CD4+Notch2+Foxp3lo T cells. The number of cells transferred through the retro-orbital vein was 5 × 105 per mouse. Disease severity was scored on a scale of 0 to 5 (0, no disease; 1, loss of tail tone; 2, hind limb weakness; 3, hind limb paralysis; 4, hind limb paralysis and forelimb paralysis or weakness; and 5, moribund/death). The infiltrating Th population was counted via FACS analysis, cytokine production was measured (see below), and histological analysis was performed.

DSS model

The mice (8–10 weeks old) were exposed to 2.5% DSS in the drinking water for 5 days, followed by access to regular drinking water until the end of the experiment. Body weight was scored daily.

Cytokine analysis

To measure the levels of secreted cytokines, CD4+ T cells were isolated from the spinal cords of EAE mice. Isolated CD4+ T cells (2 × 105) were cultured for 6 h with PMA (50 ng/ml) and ionomycin (1 μg/ml). Secreted IFN−γ, IL−17a, IL−2, IL−6, and TGF−β levels were measured via a BD Cytometric Bead Array (BD Biosciences). The levels of intracellularly stained cytokines were analyzed via a FACSCanto II or LSRFortessa flow cytometer (BD Biosciences).

Proliferation assay

To analyze the proliferation of activated cells, a CellTrace Violet (CTV) or CFSE Cell Proliferation Kit (Thermo Fisher Scientific) was used. CD4+ T cells were isolated from the spleens and spinal cords of EAE mice, and the isolated cells were stained with CTV (5 μM). To analyze the suppressive effect on proliferation, CTV−stained CD4+ T cells (1 × 105) were cocultured at a 1:1 ratio with CD4+Notch2− T cells, CD4+Notch2+Foxp3− T cells, CD4+Notch2+Foxp3lo T cells, or Treg cells. Activation was performed via anti−CD3/CD28 Dynabeads at a 1:1 bead-to-CD4+CD25− T cell ratio. Treg cells were sorted by using CD25 and Foxp3−eGFP. Treg cell expansion was analyzed by CTV or CFSE under activation conditions, and Treg cells were cocultured with CD4+Notch2− T cells, CD4+Notch2+Foxp3− T cells, and CD4+Notch2+Foxp3lo T cells at a 1:1−to−3:1 ratio and activated with anti−mouse CD3ε (2 μg/ml) and anti−mouse CD28 (2 μg/ml) antibodies for 72 h. The intensity of CTV on the cells was measured via a FACSCanto II flow cytometer (BD Biosciences) or an LSRFortessa flow cytometer (BD Biosciences).

iTreg cell differentiation with CD4+ T cells infiltrating the CNS in EAE

CD4+ T cells isolated from the spleen were activated with anti−CD3ε (2 μg/ml) and anti−mouse CD28 (2 μg/ml) antibodies. Under these conditions, CD4+ T cells were cocultured for 72 h at a 1:1 ratio with CD4+ Notch2− T cells, CD4+Notch2+Foxp3− T cells, CD4+Notch2+Foxp3lo T cells, and Tregs, which were isolated from the CNS of EAE patients. Foxp3 expression was measured via a FACSCanto II flow cytometer (BD Biosciences) or an LSRFortessa flow cytometer (BD Biosciences).

CRISPR/Cas9 gene editing of NEDD4

Two single−guide RNAs (sgRNAs) were used to target mouse Nedd4: crRNA1, ATGGGTATGGGAGTTTTGCC; crRNA2, CACCATCTTCTGTCTTATCC (PhileKorea, Inc.). Cas9−GFP (Merk) and sgRNAs were incubated at room temperature for 10 min in a 3:1 reaction. A total of 1 × 106 cells were electroporated with the RNP complex. The KO efficiency of the genes was confirmed via genomic PCR and the TIDE assay. PCR was performed via the following primers: TCCACAGCTTGGAGACATTACC forward primer, GTAACATCACTTCCGGGGGA reverse primer (Macrogen) with the Q5 Hot Start High−Fidelity 2X Master Mix (New England Biolabs).

Abseq surface labeling and single − cell RNA − sequencing analysis

Isolated spinal cord mononuclear cells (1 × 106) were labeled with mouse CD4 Olgo AMM2002 (RM4−5) and mouse Notch2 Olgo AMM2129 (16F11) for 30 min at 4 °C and washed three times before the experiments were performed. The cells were resuspended in BD Rhapsody Cartridge reagent kit sample buffer, and 4 × 104 cells were captured with the BD Rhapsody single−cell system following the manufacturer’s instructions. Additionally, antibody tag libraries, multiplexing libraries and targeted mRNA gene expression libraries were generated following the manufacturer’s instructions. Single−cell transcriptomics analysis was performed via the Python package Scanpy (v.1.9.1). Cell clustering and UMAP visualization were performed on statistically significant principal components.

Quality control and data processing

Cells with fewer than 500 unique transcripts and more than 10% mitochondrial gene content were excluded. Additionally, cells displaying a unique gene counting over 7500 were considered outliers and removed, resulting in a final dataset of 16,417 cells after quality control. The count data were log − normalized using pp.normalize, and highly variable genes were selected with the pp.highly_variable_gene function (flavor = “seurat_v3”, incorporating 2000 genes). Principal component analysis (PCA) was performed, and 30 components were retained for downstream analysis. To correct batch effects across samples, the Harmony algorithm was applied.

CD4 + T-cell subset analysis

To investigate CD4+ T cell subsets, we subset the CD4+ T cell clusters from the dataset for focused analysis. Using Leiden clustering (resolution = 1.0), we identified 10 distinct clusters, each annotated on the basis of a unique gene marker profile. This classification enabled the categorization of CD4+ T cells into eight subclusters, providing deeper insights into their functional diversity.

scFates for trajectory analysis

For trajectory inference within CD4+ T cells, we first computed a PCA representation of the data matrix using sc.pp.neighbors. A neighborhood graph was then embedded with UMAP after computing the diffusion maps (sc.tl.diffmap). By identifying key transition nodes between clusters, we reconstructed six major trajectories within the CD4+ T cell subpopulations. This analysis delineated distinct transcriptional pathways and molecular signatures associated with each trajectory, offering insights into CD4+ T cell differentiation and functional specialization.

Statistical analysis

Statistical analyses of the differences between two groups were performed via Origin2021 software (OriginLab, Northampton, Massachusetts, USA) and Graph Prism v.10.0.3 (GraphPad, Boston, Massachusetts, USA). The statistical tests used to determine significance in each analysis are described in the respective figure legends of the corresponding figures. The results are displayed as the means ± SDs. Significance is indicated as *p < 0.05, **p < 0.01, and ***p < 0.001.