DNA hypomethylation traits define human regulatory T cells in cutaneous tissue and identify their blood recirculating counterparts

Abstract

CD4⁺ regulatory T (T_reg) cells in tissues play crucial immunoregulatory and regenerative roles. Despite their importance, the epigenetics and differentiation of human tissue T_reg cells are incompletely understood. Here, we performed genome-wide DNA methylation analysis of human T_reg cells from skin and blood and integrated these data into a multiomic framework, including chromatin accessibility and gene expression. This analysis identified programs that governed the tissue adaptation of skin T_reg cells. We found that subfamilies of transposable elements represented a major constituent of the hypomethylated landscape in tissue T_reg cells. Based on T cell antigen receptor sequence and DNA hypomethylation homologies, our data indicate that blood CCR8⁺ T_reg cells contain recirculating human skin T_reg cells. Conversely, differences in chromatin accessibility and gene expression suggest a certain reversal of the tissue adaptation program during recirculation. Our findings provide insights into the biology of human tissue T_reg cells, which may help harness these cells for therapeutic purposes.

Fig. 1. DNA methylation separates skin T_reg and blood CCR8⁺ T_reg cells from blood CD45RA⁺ T_reg cells.

a, Representative flow cytometry plot showing CD45RA⁻CCR8⁺ T_reg cells among CD4⁺CD127⁻CD25⁺ T_reg cells in human blood, fat and skin (left) and percentage of CD45RA⁻CCR8⁺ T_reg cells among CD4⁺CD127⁻CD25⁺ T_reg cells in human blood compared to skin (right; n = 6 healthy female donors; FACS Donors 1–6; median age: 46 years; range: 31–61 years). The P value was determined by two-tailed Wilcoxon signed-rank test. b, Sort layout showing isolation of blood CD45RA⁺ T_reg, blood CD45RA⁺ T_conv, blood CCR8⁺ T_reg, skin T_conv and skin T_reg cells in one healthy female donor (FACS Donor 7; age unknown). Not all gates are shown. c, Principal component analysis of DNA methylation in blood CD45RA⁺ T_reg cells (RA⁺ T_reg cells), blood CD45RA⁺ T_conv cells (RA⁺ T_conv cells), blood CCR8⁺ T_reg cells, skin T_conv cells and skin T_reg cells isolated from blood and skin from nine healthy female donors (Tissue Donors 7, 10 and 11 and Blood Donors 3–8). d, Methylation level by genomic position and corresponding differences with respect to blood CD45RA⁺ T_conv cells in blood CD45RA⁺ T_reg, blood CCR8⁺ T_reg, skin T_conv and skin T_reg cells in human donors as in c. Numbers in brackets indicate the average methylation level on autosomes and chromosome X. e, Schematic showing the extraction of cell-type signatures based on the largest numeric methylation gap between any two cell types (red arrows), which was required to be at least 0.15 and at least 1.5 times as large as the second-to-largest methylation gap (blue arrows) and resulted in the selection of signature regions; RA, CD45RA. f, DNA methylation in regions belonging to selected cell-type signatures as in e. Rows correspond to signature regions, and columns correspond to donors (n = 3 donors per cell type as in c). Numbers on the right indicate the number of regions per signature category. Signature nomenclature is based on the cell types that were hypomethylated in the corresponding signature regions. g, Distribution of cell-type signature regions from the signatures in e in genomic intervals defined by genes (top) and CpG islands (bottom). Numbers at the top indicate the number of regions per signature category. Data are representative of three or more independent experiments with three or more individual donors; TSS, transcription start site.

Fig. 2. A multiomics comparison defines a core T_reg cell signature.

a, Methylation, chromatin accessibility and expression of DMRs (left), differentially accessible peaks (middle) and differentially expressed genes (right) between blood CD45RA⁺ T_reg cells (RA⁺ T_reg cells) and blood CD45RA⁺ T_conv cells (RA⁺ T_conv cells). Rows represent genomic regions (left and middle) or genes (right). Columns indicate samples from individual donors (left and right) or mean accessibility across cells from a single donor (middle); n = 3 donors (methylation, Blood Donors 3–5), 2 donors (accessibility, ATAC Donors 1 and 2) and 3 blood CD45RA⁺ T_conv donors and 5 blood CD45RA⁺ T_reg donors (expression, Blood Donors 3–5, 9 and 10). b, Correlation between differential methylation, accessibility and expression for DMR–peak–gene links (n = 151 DMR–peak–gene links). Positive values indicate larger methylation, accessibility or expression in blood CD45RA⁺ T_reg cells. Donors are as in a; gex., gene expression. c, Smoothed methylation (top left), raw methylation (top right), chromatin accessibility (bottom left) and expression (bottom right; mean ± s.d.) for selected DMR–peak–gene links in blood CD45RA⁺ T_reg and blood CD45RA⁺ T_conv cells. Highlighted regions mark DMRs (methylation tracks) and differential peaks (accessibility tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites. Gene expression P values were determined by two-tailed Wald test implemented in DEseq2 (n = 5 T_reg cell donors and 3 T_conv cell donors) with a Benjamini–Hochberg correction. Donors are as in a; P_adj, adjusted P value. d, Amplicon-based bisulfite sequencing showing methylation differences between blood CD45RA⁺ T_reg and blood CD45RA⁺ T_conv cells in selected regions (n = 6 donors, Amplicon Donors 1–6). Each circle corresponds to a CpG site. P values were determined by two-tailed Wilcoxon signed-rank test (n = 6 donors) with a Benjamini–Hochberg correction; Acc., accessibility; adj., adjusted; Meth., methylation; RPKM, reads per kilobase per million; TPM, transcripts per million. Data are representative of two or more independent experiments with two or more individual donors.

Fig. 3. DMR–peak–gene links define a multiomic signature of skin T_reg cell development.

a, Methylation, chromatin accessibility and expression of DMRs (left), differentially accessible peaks (middle) and differentially expressed genes (right) between skin T_reg cells and blood CD45RA⁺ T_reg cells (blood RA⁺ T_reg cells). Rows represent genomic regions (left and middle) or genes (right). Columns indicate samples from individual donors (left and right) or mean accessibility across cells from a single donor (middle); n = 3 donors per cell type (methylation, Tissue Donors 7, 10 and 11 and Blood Donors 3–5), 2 donors per cell type (accessibility, ATAC Donors 1, 2, 4 and 5), 4 skin T_reg donors and 5 blood CD45RA⁺ T_reg donors (expression, Tissue Donors 7, 8, 10 and 11 and Blood Donors 3–5, 9 and 10). b, Correlation between differential methylation, accessibility and expression for DMR–peak–gene links (n = 1,203 DMR–peak–gene links). Positive values indicate larger methylation, accessibility or expression in skin T_reg cells. Donors are as in a; FC, fold change. c, Smoothed methylation (top left), raw methylation (top right), chromatin accessibility (bottom left) and gene expression (bottom right; mean ± s.d.) in skin T_reg and blood CD45RA⁺ T_reg cells for selected DMR–peak–gene links. Highlighted regions mark DMRs (methylation tracks) and differential peaks (accessibility tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites. Gene expression P values were determined by two-tailed Wald test implemented in DESeq2 (n = 4 skin donors and 5 blood donors) with a Benjamini–Hochberg correction. Donors are as in a. d, Enrichment of hallmark gene sets among genes belonging to the multiomic skin T_reg cell signature. All gene sets displaying a P_adj of <0.01 are shown. P values were determined by one-tailed Fisher’s exact test (n = 408 genes from the multiomic signature) with a Benjamini–Hochberg correction; DN, down; OR, odds ratio; r, Pearson correlation. Data are representative of two or more independent experiments with two or more individual donors.

Fig. 4. Activity of bZIP and bHLH transcription factors during skin T_reg cell development is associated with binding site hypomethylation.

a, Enrichment of bZIP motifs in DMRs and differential peaks between skin T_reg cells and blood CD45RA⁺ T_reg cells (blood RA⁺ T_reg cells). P values were determined by one-tailed binomial test implemented in homer (n = 1,742 DMRs, 298,457 DMRs, 5,722 peaks and 3,192 peaks for columns 1–4) with a Benjamini–Hochberg correction. b, Chromatin accessibility in skin T_reg and blood CD45RA⁺ T_reg cells (four donors, ATAC Donors 1, 2, 4 and 5) around genomic sites for a selected bZIP motif. c, Enrichment of bHLH motif sites in DMRs and differential peaks between skin T_reg cells and blood CD45RA⁺ T_reg cells. P values were determined as in a. d, Chromatin accessibility in skin T_reg and blood CD45RA⁺ T_reg cells around genomic sites for a selected bHLH motif. Donors are as in b. e, Transcriptomic footprint of transcription factors (that is, enrichment of target genes) among genes differentially expressed between skin T_reg and blood CD45RA⁺ T_reg cells. Labels correspond to transcription factor gene names. P values were determined by two-tailed permutation test (n = 28,078 genes) with a Benjamini–Hochberg correction. f, Expression of transcription factor genes corresponding to relevant motifs (mean ± s.d.; n = 4 skin T_reg cell donors and 5 blood CD45RA⁺ T_reg cell donors, Tissue Donors 7, 8, 10 and 11 and Blood Donors 3–5, 9 and 10). g, Methylation (top left, Tissue Donors 7, 10 and 11 and Blood Donors 3–5) and chromatin accessibility (middle left, ATAC Donors 1, 2, 4 and 5) in skin T_reg and blood CD45RA⁺ T_reg cells together with c-Myc ChIP–seq signal from three cell lines (bottom left) around c-Myc motif sites associated with MLPH and GNA11 and quantitative PCR with reverse transcription showing relative gene expression (mean) of MLPH (guide RNA A or B) and GNA11 in primary blood T_reg cells from six donors (CRISPR Donors 1–6, three donors per guide RNA) after CRISPR-mediated activation originating from regions shown in track plots (right). Highlighted regions correspond to selected DMRs (methylation tracks), scATAC-seq peaks (chromatin accessibility tracks) and c-Myc motif sites (motif ± 100 bases; ChIP–seq and gene tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites. P values were determined by two-tailed paired t-test (n = 3 donors). Data are representative of two or more independent experiments with two or more individual donors.

Fig. 5. DNA methylation patterns reflect similarity of skin T_reg and blood CCR8⁺ T_reg cells.

a, Methylation (Tissue Donors 7, 10 and 11 and Blood Donors 3–8), chromatin accessibility (ATAC Donors 1, 2, 4 and 5) and gene expression (Tissue Donors 7, 8, 10 and 11 and Blood Donors 3–12) of features that are different between skin T_reg and blood CD45RA⁺ T_reg cells (blood RA⁺ T_reg cells). Annotation bars indicate blood CCR8⁺ T_reg cell positioning. b, Proportions of features placing blood CCR8⁺ T_reg cells closer to skin T_reg cells or blood CD45RA⁺ T_reg cells. Donors are as in a. c, Smoothed methylation (top left), raw methylation (top right), chromatin accessibility (bottom left) and gene expression (bottom right, mean ± s.d.) in skin T_reg, blood CCR8⁺ T_reg and blood CD45RA⁺ T_reg cells for regions where blood CCR8⁺ T_reg cells are closer to blood CD45RA⁺ T_reg cells than to skin T_reg cells regarding chromatin accessibility. Highlighted regions mark DMRs and differential peaks. P values were determined by two-tailed Wald test implemented in DESeq2 (n = 4, 5 and 5 donors for skin T_reg cells, blood CCR8⁺ T_reg cells and blood CD45RA⁺ T_reg cells, respectively) with a Benjamini–Hochberg correction. Donors are as in a. d, Enrichment of bZIP motifs in feature sets defined by blood CCR8⁺ T_reg cell positionings. P values were determined by one-tailed binomial test implemented in homer (n = 6,424 DMRs, 292,033 DMRs, 1,128 peaks and 2,064 peaks for columns 1–4) with a Benjamini–Hochberg correction. e, Methylation around genomic sites (±200 bases) for bZIP motifs overlapping with DMRs (mean across three biological replicates) in skin T_reg, blood CCR8⁺ T_reg and blood RA⁺ T_reg cells. Annotation bars indicate whether blood CCR8⁺ T_reg cells are closer to skin T_reg cells or blood CD45RA⁺ T_reg cells as in a. Donors are as in a. Heat maps are ordered as in d. f, Chromatin accessibility in skin T_reg, blood CCR8⁺ T_reg and blood CD45RA⁺ T_reg cells around genomic sites for a selected bZIP motif (see Fig. 4). Donors are as in a. g, Module scores (with cell numbers, median and bottom/top quartile) of genes upregulated in blood CD45RA⁺ T_reg or skin T_reg cells in TCR-matched blood CCR8⁺ T_reg, skin T_reg and blood CD45RA⁺ T_reg cells from two donors (TCR Donors 6 and 7). h,i, Concatenated flow cytometry plot showing CCR8 expression (h) and percentage of CCR8⁺ cells among Kaede-red⁺ T_reg cells and Kaede-green⁺ T_reg cells (i) in skin dLN T_reg cells of Kaede mice 6 days after skin photoconversion from Kaede-green to Kaede-red (data concatenated from seven mice; mean ± s.d.). The P value was determined by two-tailed Wilcoxon signed-rank test. Data are representative of two or more independent experiments with two or more individual donors.

Fig. 6. TEs are hypomethylated in skin T_reg cells and blood CCR8⁺ T_reg cells.

a,b, Enrichment of TE classes and subfamilies in regions hypomethylated (a) or hyperaccessible (b) in skin T_reg cells. P values were determined by one-tailed permutation test (n = 298,457 DMRs and 3,192 differential peaks) with a Benjamini–Hochberg correction; RC, rolling circle. c, Methylation of annotated insertion sites (with median) for three TE classes in skin T_reg and blood CD45RA⁺ T_reg cells (blood RA⁺ T_reg cells; left; three donors per cell type; Tissue Donors 7, 10 and 11 and Blood Donors 3–5) and distribution of differences in mean methylation (across the three donors) between skin T_reg and blood CD45RA⁺ T_reg cells for each insertion site (right). The numbers indicate how many insertion sites displayed differences below and above 0; AU, arbitrary units. d, Methylation (donors are as in c) and chromatin accessibility (ATAC Donors 1, 2, 4 and 5) in skin T_reg and blood CD45RA⁺ T_reg cells around selected TE insertion sites. Highlighted regions mark DMRs (methylation tracks), differential peaks (accessibility tracks) and TE insertion sites overlapping with DMRs (only TE subfamilies enriched among the DMRs; gene tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites (CpG sites overlapping with TEs are shown in red). TE labels correspond to TE subfamilies (partly overlaps with Figs. 3 and 5). e,f, Enrichment of TE classes and subfamilies in regions that are hypomethylated (e) or hyperaccessible (f) in skin T_reg cells and defined by specific positionings of blood CCR8⁺ T_reg cells. P values were determined by two-tailed Fisher’s exact test (n = 292,033/2,064 DMRs/peaks ‘closer to skin T_reg’, 6,424/1,128 DMRs/peaks ‘closer to blood CD45RA⁺ T_reg’). g, Methylation of TE insertion sites overlapping with regions hypomethylated in skin T_reg cells (with median) in skin T_reg, blood CCR8⁺ T_reg and blood CD45RA⁺ T_reg cells (top; three donors per cell type; Tissue Donors 7, 10 and 11 and Blood Donors 3–8) and distribution of differences in mean methylation (across the three donors) between any two cell types for each insertion site (bottom). Numbers indicate how many insertion sites displayed differences below and above 0. Data are representative of two or more independent experiments with two or more individual donors.

Fig. 7. Several TE subfamilies are transcribed in skin T_reg cells.

a, Expression levels of TE subfamilies in skin T_reg, skin T_conv, blood CCR8⁺ T_reg, blood CD45RA⁺ T_reg (blood RA⁺ T_reg) and blood CD45RA⁺ T_conv (blood RA⁺ T_conv) cells with within-cell-type mean based on 4, 4, 5, 5 and 3 donors, respectively (Tissue Donors 7, 8, 10 and 11 and Blood Donors 3–12). Values equal to 0 are shown in light beige. TPM normalization reflects the expression level of a particular TE compared to the whole collection of TEs. b, Differential expression of TE subfamilies between skin T_reg cells and blood CD45RA⁺ T_reg cells. P values were determined by two-tailed Wald test implemented in DESeq2 (skin T_reg cell donors, n = 4; blood CD45RA⁺ T_reg cell donors, n = 5; donors are as in a) with a Benjamini–Hochberg correction. c, Methylation of annotated HERVIP10F-int and LTR45B insertion sites (with median) in skin T_reg, blood CCR8⁺ T_reg and blood CD45RA⁺ T_reg cells (top; n = 3 donors per cell type; Tissue Donors 7, 10 and 11 and Blood Donors 3–8) and distribution of differences in mean methylation (across the three donors) between any two cell types for each insertion site (bottom). Numbers indicate how many insertion sites displayed differences below and above 0.

Extended Data Fig. 1. Quality control and methylation-level cell type signatures.

a, Frequency of CD45RA⁻CCR8⁺ T_reg cells among CD4⁺CD127⁻CD25⁺ T_reg cells in human blood, fat and skin, 4 donors. b, Distribution of methylation at the first 3,000,000 CH sites of chromosome 1, stratified by sample (3 samples from three donors per cell type) and sequencing library. c, Distribution of CpG-wise coverage values (3 samples from three donors per cell type). d, DNA Methylation in regions belonging to methylation-level cell type signatures with at least 10 regions. Rows correspond to signature regions; columns correspond to samples from individual donors (n = 3 donors per cell type). Numbers on the left indicate the number of regions in each signature category. Parts of this panel are duplicated from Fig. 1f and included for the sake of completeness. Data are representative of three or more independent experiments with three or more individual donors.

Extended Data Fig. 2. Details and examples of the ‘core naïve T_reg ’ signature.

a, Schematic of the study design, describing that blood CD45RA⁺ T_conv cells, blood CD45RA⁺ T_reg cells, blood CCR8⁺ T_reg cells, skin T_conv cells and skin T_reg cells were analyzed with respect to their DNA methylation, chromatin accessibility and gene expression. Schematic generated using BioRender.com. b, UMAP showing scATAC-seq data of the analyzed cell types (encompassing 2 donors per cell type). c, Volcano plots for differential analyses on the chromatin accessibility and gene expression level (corresponding to heat maps in Fig. 2a). Features shown in red are considered statistically significant. Numbers at the top indicate the numbers of features displaying statistical significance. P values for chromatin accessibility, two-tailed likelihood ratio test (see Methods, n = 5,687 blood CD45RA⁺ T_reg cells, 6,531 blood CD45RA⁺ T_conv cells), adjusted using Bonferroni correction; P values for gene expression, two-tailed Wald test based on the raw count matrix as implemented in DESeq2, using donor ID as additional covariate (see Methods; n = 5 T_reg cell donors, 3 T_conv cell donors), adjusted using Benjamini-Hochberg correction. d, Methylation (left), chromatin accessibility (middle) and gene expression (right) of features in DMR-peak-gene links, stratified by quadrant (see Fig. 2b). Labels on the right indicate the genes in each DMR-peak-gene link. Parts of this panel are duplicated from Fig. 2a as the same features are shown. e, Methylation for selected genomic regions. Highlighted regions mark DMRs. Vertical lines at the bottom of the methylation tracks mark CpG sites. f, Smoothed methylation and chromatin accessibility for selected DMR-peak-gene links together with corresponding raw methylation for each shown CpG site (top right heat maps; rows indicate donors; columns indicate CpG sites; gray fields indicate missing values) and expression of the associated gene (bottom right bar charts; mean ± s.d.). Highlighted regions mark DMRs (methylation tracks) and differential peaks (accessibility tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites. Dashed boxes mark the locations of amplicons used for validation of methylation differences. Gene expression P values, two-tailed Wald test based on the raw count matrix as implemented in DESeq2, using donor ID as additional covariate (see Methods; n = 5 T_reg cell donors, 3 T_conv cell donors) with Benjamini-Hochberg correction. Data are representative of two or more independent experiments with two or more individual donors.

Extended Data Fig. 3. Details and examples of the ‘skin T_reg ’ signature.

a, Volcano plots for differential analyses on the chromatin accessibility and gene expression level (corresponding to heat maps in Fig. 3a). Features shown in red are considered statistically significant. Numbers at the top indicate the numbers of features displaying statistical significance. P values for chromatin accessibility, two-tailed likelihood ratio test (see Methods, n = 4,091 skin T_reg cells, 5,687 blood CD45RA⁺ T_reg cells), adjusted using Bonferroni correction; P values for gene expression, two-tailed Wald test based on the raw count matrix as implemented in DESeq2 (see Methods; n = 4 skin T_reg cell donors, 5 blood CD45RA⁺ T_reg cell donors; unpaired sample structure), adjusted using Benjamini-Hochberg correction. b, Methylation (left), chromatin accessibility (middle) and gene expression (right) of features in DMR-peak-gene links, stratified by quadrant (see Fig. 3b). Labels on the right indicate the genes in each DMR-peak-gene link. Parts of this panel are duplicated from Fig. 3a as the same features are shown. c, Smoothed methylation and chromatin accessibility for a selected DMR-peak-gene link together with corresponding raw methylation for each shown CpG site (top right heat map; rows indicate donors; columns indicate CpG sites; gray fields indicate missing values) and expression of the associated gene (bottom right bar chart; mean ± s.d.). Highlighted regions mark DMRs (methylation track) and differential peaks (accessibility track). Vertical lines at the bottom of the methylation track mark CpG sites. Dashed boxes mark the locations of amplicons used for validation of methylation differences. Gene expression P value, two-tailed Wald test (see Methods; n = 4 skin donors, 5 blood donors) with Benjamini-Hochberg correction. Expression bar charts are duplicated from Fig. 5 as they correspond to the same gene. d, Methylation for selected genomic regions. Highlighted regions mark DMRs. Vertical lines at the bottom of the methylation tracks mark CpG sites. e, DMRs (left), differentially accessible peaks (middle) and differentially expressed genes (right; row-scaled) between blood CD45RA⁺ T_reg cells, blood CD45RA⁺ T_conv cells, shown together with corresponding values in skin T_reg cells and fat T_reg cells (‘core naïve T_reg’ signature, upper panel), and between blood CD45RA⁺ T_reg cells, fat T_reg cells and skin T_reg cells (‘skin T_reg ’ signature, lower panel). Rows indicate genomic regions (left, middle) or genes (right). Columns indicate samples from individual donors, with n = 3 donors for methylation data, n = 2 skin T_reg / blood CD45RA⁺ T_reg / blood CD45RA⁺ T_conv donors and 3 fat T_reg donors for accessibility data, or n = 3 blood CD45RA⁺ T_conv donors, 4 skin T_reg donors and 5 blood CD45RA⁺ T_reg / fat T_reg donors for gene expression data. Parts of this panel are duplicated from Figs. 2a and 3a as the same features are shown. Data are representative of two or more independent experiments with two or more individual donors.

Extended Data Fig. 4. Details for transcription factor results with respect to skin adaptation of T_reg cells.

a, Logos for the bZIP motifs highlighted in Fig. 4a (obtained from homer). b, Chromatin accessibility around genomic sites for additional bZIP motifs. c, Methylation (each row corresponds to a motif site ± 200 bases, each column corresponds to one of 3 donors, donors for skin Treg cells and blood CD45RA⁺ T_reg cells are unpaired) around genomic sites for bZIP motifs (n = 29,813 motif sites, 29,546 motif sites, 28,220 motif sites, 28,693 motif sites, 26,628 motif sites, 30,426 motif sites, 27,920 motif sites, 23,868 motif sites, 28,863 motif sites and 18,907 motif sites, respectively, from top to bottom). d, Logos for the bHLH motifs highlighted in Fig. 4c (obtained from homer). e, Chromatin accessibility around genomic sites for additional bHLH motifs. f, Methylation (each row corresponds to a motif site ± 200 bases, each column corresponds to one of 3 donors, donors for skin Treg cells and blood CD45RA⁺ T_reg cells are unpaired) around genomic sites for bHLH motifs. n = 12,050 motif sites, 11,186 motif sites, 9,358 motif sites, 11,415 motif sites and 3,004 motif sites, respectively, from top to bottom). g, Methylation, chromatin accessibility and USF1 ChIP-seq signal from three cell lines around interesting motif sites for USF1. Highlighted regions correspond to selected DMRs (methylation tracks), scATAC-seq peaks (chromatin accessibility tracks) and USF1 motif sites (motif ± 100 bases; ChIP-seq and gene tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites. Parts of this panel are duplicated from Fig. 4g as they display the same genomic regions but focus on a different transcription factor. Data are representative of two or more independent experiments with two or more individual donors.

Extended Data Fig. 5. Targeting of GNA11_A guide and MLPH_A guide.

a, Sanger sequencing of GNA11 locus in human CD4⁺ T cells after CRISPR-Cas9 targeting with guide GNA11_A (top), scrambled control (middle) and not-CRISPRed genomic DNA (bottom) of human T cells. Blue boxes indicate position of GNA11_A guide. Lower part, percentage of aberrant sequences (not matching to genomic control sample). Dashed blue line indicates position of GNA11_A guide. b, Sanger sequencing of MLPH locus in human CD4⁺ T cells after CRISPR-Cas9 targeting with guide MLPH_A (top), scrambled control (middle) and not-CRISPRed genomic DNA (bottom) of human T cells. Blue boxes indicate position of MLPH_A guide. Lower part, percentage of aberrant sequences (not matching to genomic control sample). Dashed blue line indicates position of MLPH_A guide.

Extended Data Fig. 6. The ‘Skin T_reg ’ signature across many cell types and the molecular positioning of blood CCR8⁺ T_reg cells.

a, Feature numbers (that is numbers of DMRs, differentially accessible peaks and differentially expressed genes) underlying blood CCR8⁺ T_reg cell positionings shown in Fig. 5a. b, Methylation, chromatin accessibility and gene expression for the ‘skin T_reg ’ signature in skin T_reg cells, skin T_conv cells, blood CCR8⁺ T_reg cells, blood CD45RA⁺ T_reg cells and blood CD45RA⁺ T_conv cells. Parts of this panel are duplicated from Fig. 3a as the same features are shown. c, Methylation (left), chromatin accessibility (middle) and gene expression (right) of features that belong to DMR-peak-gene links with hypomethylation in skin T_reg cells (top) or blood CD45RA⁺ T_reg cells (bottom). Columns indicate samples from different donors (left, right) or donor-wise mean accessibility across cells (middle). n = 3 donors on the methylation level, 2 donors on the accessibility level, 4 (skin T_reg) and 5 (blood CCR8⁺ T_reg, blood CD45RA⁺ T_reg) donors on the expression level. Annotation bars on the right of each heat map indicate whether blood CCR8⁺ T_reg cells are closer to skin T_reg cells or closer to blood CD45RA⁺ T_reg cells with respect to a feature (based on the mean across donors; see Methods for details). Parts of this panel are duplicated from Fig. 5a as the same features are shown. d, Smoothed methylation and chromatin accessibility in regions where chromatin accessibility of the overlapping peak places blood CCR8⁺ T_reg cells closer to skin T_reg cells (top) and closer to blood CD45RA⁺ T_reg cells (bottom). These regions are shown together with corresponding raw methylation for each displayed CpG site (top right heat maps; rows indicate donors; columns indicate CpG sites; gray fields indicate missing values) and expression of the associated gene (bottom right bar charts; mean ± s.d.). Highlighted regions mark DMRs (methylation tracks) and differential peaks (accessibility tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites. P values, two-tailed Wald test based on the raw count matrix as implemented in DESeq2 (see Methods; n = 4 skin T_reg donors, 5 blood CCR8⁺ T_reg donors, 5 blood CD45RA⁺ T_reg donors; unpaired sample structure) with Benjamini-Hochberg correction. Data are representative of two or more independent experiments with two or more individual donors.

Extended Data Fig. 7. Details of motif analysis regarding the positioning of blood CCR8⁺ T_reg cells.

a, Logos for motifs highlighted in Fig. 5d (obtained from homer). b, Chromatin accessibility around genomic sites for additional bZIP motifs. Parts of these plots are duplicated from Extended Data Fig. 4 as they correspond to the same motifs. c, Enrichment of bHLH motifs in methylation-level region sets defined by different blood CCR8⁺ T_reg cell positionings. P values, one-tailed binomial test implemented in homer assessing enrichment of motif sites among the region sets compared to background sets of random regions (see Methods; n = 6,424 DMRs and 292,033 DMRs for columns 1 and 2, respectively) with Benjamini-Hochberg correction. d, Motifs highlighted in c (obtained from homer). Parts of this figure are duplicated from Extended Data Fig. 5 as they correspond to the same motifs. Data are representative of two or more independent experiments with two or more individual donors.

Extended Data Fig. 8. Details of scRNA-seq/scTCR-seq data analysis and Kaede mice experiments regarding the positioning of blood CCR8⁺ T_reg cells.

a–d, UMAP representations of cells from scRNA-seq/scTCR-seq, showing clusters and cell type annotation next to gene expression and signature values underlying the annotation. Data correspond to blood T_reg cells from TCR Donor 6 (sample 28; a), blood CD45RO⁺ T_reg cells from TCR Donor 6 (sample 29, b), blood T_reg cells from TCR Donor 7 (sample 24, c) and blood CD45RO⁺ T_reg cells from TCR Donor 7 (sample 25, d). Blood CCR8⁺ T_reg signature 1 defines blood CCR8⁺ T_reg cells against blood CD45RA⁺ T_reg cells. Blood CCR8⁺ T_reg signature 2 defines blood CCR8⁺ T_reg cells against blood CD45RO⁺ (memory) T_reg cells. For details regarding the annotation, see the Methods section. e, Stacked bar charts showing TCR clonotype sharing between blood CD45RA⁺ T_reg cells, skin T_reg cells and blood CCR8⁺ T_reg cells. Bar sections indicate the percentage of cells belonging to a particular clonotype. Connections between bar sections indicate that a clonotype was found in two cell types. All clonotypes that appear in skin T_reg cells and contain at least one TCR-α and TCR-β chain are shown (271 clonotypes from TCR Donor 6, 602 clonotypes from TCR Donor 7). f, Percentages of skin T_reg cell clonotypes that were also found among blood CD45RA⁺ T_reg cells and blood CCR8⁺ T_reg cells. g, Kaede-green and Kaede-red expression in skin-draining-lymph-node-derived T_reg cells of Kaede mice 6 days after photoconversion of skin. Cells were pre-enriched for CD25 and gated on size, viability, TCR-β, CD4, CD25 and GITR. Concatenated data of 7 mice. h, CCR8 expression was measured in T_reg and T_conv cells 6 days after light exposure of skin of Kaede mice. Cells were harvested from LN and enriched for CD25 expression. Shown here are the T_conv cells present in CD25-enriched fractions, results for T_reg cells are shown in the main figure. Concatenated data of 7 mice. Data are representative of two or more independent experiments with two or more individual donors/mice.

Extended Data Fig. 9. The contribution of TEs to the epigenetic landscape of skin T_reg cells and blood CCR8⁺ T_reg cells.

a, b, Enrichment of TE classes (left) and subfamilies (right) in regions hypomethylated (a) or hyperaccessible (b) in blood CD45RA⁺ T_reg cells. P values: one-tailed permutation test assessing enrichment of TE insertion sites among DMRs or peaks (see Methods) based on 1,742 DMRs (a) and 5,722 differential peaks (b) displayed in Fig. 3 with Benjamini-Hochberg correction. c, Left, Chromatin accessibility of peaks overlapping with annotated LTR-TE, DNA-TE and SINE insertion sites (n = 2 donors per cell type). White horizontal lines inside violins indicate medians across all peaks. Right, Distribution of differences in mean accessibility (across the two donors) between skin T_reg cells and blood CD45RA⁺ T_reg cells for each peak. Numbers indicate how many insertion sites displayed differences below and above 0. d, Left, Methylation of annotated LTR, DNA-TE and SINE insertion sites as in c with skin T_reg, fat T_reg and blood CD45RA⁺ T_reg (n = 3 donors per cell type). White horizontal lines inside violins indicate medians across all insertion sites. Right, Distribution of differences in mean methylation (across the three donors) between any two cell types for each insertion site. Numbers indicate how many insertion sites displayed differences below and above 0. e, Smoothed methylation and chromatin accessibility around selected TE insertion sites. Highlighted regions mark DMRs (methylation tracks), differential peaks (accessibility tracks) and TE insertion sites overlapping with DMRs (only TE subfamilies enriched among the DMRs; gene tracks). Vertical lines at the bottom of the methylation tracks mark CpG sites (CpG sites overlapping with TEs are shown in red). TE labels correspond to TE subfamilies. Parts of these plots are duplicated from Fig. 3 and Extended Data Fig. 3 as they show the same genomic regions. f, Top, Chromatin accessibility of peaks displaying hyperaccessibility in skin T_reg cells and overlapping with DNA-TE, LINE, LTR-TE and SINE insertion sites (n = 2 donors per cell type). White horizontal lines inside violins indicate medians across all peaks. Bottom, Distribution of differences in mean accessibility (across the two donors) between any two cell types for each peak. Numbers indicate how peaks displayed differences below and above 0. Data are representative of two or more independent experiments with two or more individual donors.

Extended Data Fig. 10. Methylation and accessibility of HERVIP10F-int and LTR45B insertion sites.

a, Left, Methylation of annotated HERVIP10F-int and LTR45B insertion sites in skin T_reg, fat T_reg and blood CD45RA⁺ T_reg cells (n = 3 donors per cell type). White horizontal lines inside violins indicate medians across all insertion sites. Right, Distribution of differences in mean methylation (across the three donors) between any two cell types for each insertion site. Numbers indicate how many insertion sites displayed differences below and above 0. b, Chromatin accessibility of peaks overlapping with annotated HERVIP10F-int and LTR45B insertion sites in skin T_reg, blood CCR8⁺ T_reg and blood CD45RA⁺ T_reg cells. White horizontal lines inside violins indicate medians across all peaks. Data are representative of two or more independent experiments with two or more individual donors.