Main

Sex-dependent differences in adipose tissue physiology and organismal metabolism are well documented across species10,11. Consistent with this notion, male and female mice display differences in body weight, ratio of lean mass to fat mass and rates of energy expenditure (Fig. 1a, b, Extended Data Fig. 1a–c). Males compared with age-matched females show relative glucose intolerance and concomitant hyperinsulinaemia, hallmarks of insulin resistance, but no differences in adipokines (Fig. 1c, Extended Data Fig. 1d, e). Immune cells have critical roles in VAT-mediated regulation of organismal metabolism12,13,14,15,16,17. Analysis of perigonadal VAT of lean male mice revealed much larger proportions and numbers of Treg cells compared with those from female mice (Fig. 1d–f). This difference was specific to Treg cells, as there were no significant differences between any other major adaptive and innate immune cell populations, including type 2 innate lymphocytes (ILC2s), which together with Treg cells have an important role in VAT homeostasis14,15,18,19 (Extended Data Fig. 1f, g). VAT Treg cells also displayed marked sex-dependent phenotypic differences. While Treg cells from VAT of both males and females had an activated phenotype (CD62L−CD44+), only those from males expressed high levels of the IL-33 receptor ST2, the maturation marker KLRG1 and the chemokine receptor CCR2 (Fig. 1g, Extended Data Fig. 1h). Likewise, the immune suppressive cytokine IL-10 was abundant in VAT Treg cells from males but not in those from females (Fig. 1h). Sex-specific differences in Treg cells were specific to the VAT, as abundance and IL-10 expression of Treg cells in other lymphoid or non-lymphoid tissues, such as small intestine lamina propria, colon, liver and lung were similar (Extended Data Figs. 1i, 2a). Furthermore, the differences were specific to the adipose tissue depot, as neither the subcutaneous nor the peri-nephric adipose tissue showed sex-specific differences in abundance or phenotype of Treg cells (Extended Data Fig. 2b, c).

Fig. 1: Treg cells show VAT-specific sexual dimorphism.
figure 1

a, Ratio of lean mass to fat mass. Female, n = 6; male, n = 6. b, Oxygen consumption. Female, n = 7; male, n = 8. c, Glucose tolerance in 25-week-old female and male mice under normal chow diet conditions. n = 4 mice of each sex. Graph on the right shows area under the curve (AUC) for the glucose-tolerance test. d, Proportions of FOXP3+ cells among T cell receptor β (TCRβ)-positive T cells in VAT of female and male C57BL/6 mice. Representative of n = 19 females and n = 16 males. e, FOXP3+ Treg cells as a proportion of CD4+ cells. Female, n = 19; male, n = 16. f, Treg cell numbers in the VAT of female and male mice. n = 7 mice of each sex. g, Expression of indicated cell surface markers on male and female VAT Treg cells. Representative of n = 19 females and n = 16 males. h, Il10 (GFP) expression in Treg cells in VAT from male and female Foxp3RFPIl10GFP mice. Representative of n = 6 mice of each sex. Two-tailed unpaired t-test (a–c, e, f); data are mean ± s.d. Data pooled or representative of two or three independent experiments.

Source data

Full size image

Sex-specific VAT Treg cell molecular profile

To gain insight into the molecular mechanisms that underpin the sex-dependent differences in VAT Treg cells, we compared transcriptional profiles of Treg cells isolated from the VAT and spleens of male and female mice by RNA sequencing (RNA-seq). Treg cells isolated from male VAT and spleen differed substantially, revealing a distinct VAT Treg cell transcriptional signature of almost 3,000 genes (false discovery rate 0.1, fold change >2) (Fig. 2a, Supplementary Information). By contrast, Treg cells from female VAT were similar to their splenic counterparts, with only 305 differentially expressed genes (Fig. 2b, Supplementary Information). Comparison of male and female VAT Treg cells revealed large differences in their transcriptional profiles, with >1,100 genes differentially expressed between the sexes; splenic Treg cells showed no such differences (Fig. 2c, Extended Data Fig. 2d, Supplementary Information). In line with our flow-cytometric data, male VAT Treg cells showed higher expression of Il1rl1 (encoding ST2), Il10, Ccr2 and Klrg1. Similarly, male but not female VAT Treg cells expressed high amounts of Pparg (required for VAT Treg cell differentiation20), Prdm1 (encoding BLIMP1, associated with effector Treg cell differentiation21) and Gata3. By contrast, female VAT Treg cells showed increased amounts of Sell (encoding CD62L), Cxcr5, Stat1, Foxo1 and Tcf7 (Fig. 2c). Far fewer genes, including the ubiquitous male-specific Ddx3y and female-specific Xist, were differentially expressed between male and female VAT-resident ILC2s and conventional CD4+ T cells (Extended Data Fig. 2e, f, Supplementary Information), indicating that the sex-dependent differences in gene expression were specific to VAT Treg cells. To test whether the distinct transcriptional profiles of male and female VAT Treg cells were reflected in differential chromatin accessibility, we performed assay for transposase-accessible chromatin using sequencing (ATAC-seq) (Fig. 2d). We found 3,833 loci with sex-dependent differential accessibility in VAT Treg cells. This included Il1rl1, Il10, Pparg and Klrg1, which were part of the male VAT Treg cell transcriptional signature and showed increased accessibility in male VAT Treg cells compared with both female VAT Treg and male splenic Treg cells (Fig. 2e, Extended Data Fig. 2g, h, Supplementary Information).

Fig. 2: Treg cells from male and female VAT exhibit distinct transcriptional profiles and chromatin accessibility.
figure 2

Treg cells were sorted from VAT and spleens of 25- to 32-week-old Foxp3RFP mice to perform RNA-seq and ATAC-seq. n = 2 samples, each sample contains Treg cells from n = 5 male and n = 12 female mice. a, Volcano plot shows genes differentially expressed between Treg cells from male VAT and spleen. Each dot represents a gene; genes in red are upregulated and genes in blue are downregulated in Treg cells from VAT of male mice. b, Volcano plot shows genes differentially expressed between Treg cells from female VAT and spleen. Genes in red are upregulated and genes in blue are downregulated in Treg cells from VAT of female mice. c, Heat map shows top-200 genes that are differentially expressed between male and female VAT and comparison with splenic Treg cells. d, Multi-dimensional scaling analysis of ATAC-seq data. Distances shown on the plot represent the leading log2 fold change between samples. e, ATAC-seq tracks show chromatin accessibility at the Klrg1 locus of male splenic Treg cells (green) and Treg cells from female (red) and male (blue) VAT. Arrows indicate regions of differential chromatin accessibility. Statistical methods and software packages are described in Methods.

Full size image

Treg cell extrinsic sex-hormonal function

Sex hormones are central to many developmental processes and are enriched in adipose tissue22. To test whether VAT Treg cells were regulated by sex hormones, we analysed mice that were deficient for either androgen or oestrogen receptors. VAT Treg cells from male mice lacking the androgen receptor (Ar−/−) were significantly reduced compared to their wild-type counterparts and displayed a phenotype similar to female wild-type VAT Treg cells, including diminished ST2, KLRG1 and CCR2 expression (Fig. 3a–c, Extended Data Fig. 3a). By contrast, female but not male mice lacking oestrogen receptor-α (Era−/−, also known as Esr1−/−) showed a significant increase in VAT Treg cells that displayed the phenotype of male VAT Treg cells with elevated expression of ST2, KLRG1 and CCR2 (Fig. 3d–f, Extended Data Fig. 3b, c). This phenotype was recapitulated when we treated female wild-type mice with ICI 182-780, an oestrogen receptor antagonist (Extended Data Fig. 3d, e). Consistent with an important role of sex hormones in VAT physiology, male Ar−/− mice had reduced VAT mass and improved glucose tolerance, whereas female Era−/− mice had increased VAT mass and decreased glucose tolerance (Extended Data Fig. 3f–i). Both Ar−/− and Era−/− mice showed modestly increased fasting plasma insulin compared with controls (Extended Data Fig. 3j, k).

Fig. 3: Sex differences in VAT Treg cells are linked to sex hormones.
figure 3

a, FOXP3 and ST2 expression in CD4+ T cells from VAT of male wild-type (WT) and Ar−/− mice. Wild type, n = 6; Ar−/−, n = 7. b, FOXP3+ cells among CD4+ T cells in VAT (n = 6 wild type, n = 7 Ar−/−), spleen and small intestine lamina propria (SI-LP) of wild-type and Ar−/− male mice. n = 3 mice per genotype. c, KLRG1 expression in Treg cells from VAT of wild-type and Ar−/− male mice. Wild type, n = 6; Ar−/−, n = 7. d, FOXP3 and ST2 expression in VAT CD4+ T cells from female wild-type and Era−/− mice. Wild type, n = 6; Era−/−, n = 7. e, FOXP3+ Treg cells among CD4+ T cells from wild-type and Era−/− mice; VAT (n = 6 wild type, n = 7 Era−/−), spleen (n = 7 wild type, n = 5 Era−/−) and SI-LP (n = 3 per genotype). f, KLRG1 expression inTreg cells from VAT of female wild-type and Era−/− mice. Wild type, n = 6; Era−/−, n = 7. g, Left, percentage of FOXP3+ Treg cells in VAT from mock-treated (n = 7) and oestrogen-treated (E-2) male wild-type mice (n = 13). Right, as left, but for mock-treated (n = 6) and testosterone (T)-treated (n = 9) female wild-type mice. h, i, Volcano plot showing genes differentially expressed between male VAT and subcutaneous adipose tissue (SC-AT) (h), or between total male and female VAT (i). Each dot represents a gene; genes in red are upregulated and genes in blue are downregulated in the respective comparison. j, Proportion of Treg cells in the wild-type and Ccr2−/− compartments of the VAT (n = 5), spleen (n = 4) and SI-LP (n = 3) of male mixed bone marrow chimeric mice. k, Quantitative PCR analysis of gene expression in age-matched male wild-type and Ar−/− and female wild-type and Era−/− mice. Two-tailed unpaired t-test (b, e, g, j); one-way analysis of variance (ANOVA) (k). Data are mean ± s.d., except in k, mean ± s.e.m. Data are pooled or representative of two independent experiments.

Source data

Full size image

Sex hormone receptors are widely expressed, including by cells of the immune system23,24. To test whether the functions of hormone receptors are intrinsic to Treg cells, we generated bone marrow chimeric mice by transferring congenically marked male wild-type bone marrow into lethally irradiated male Ar−/− or wild-type recipients (Extended Data Fig. 4a). Male wild-type VAT Treg cells that developed in an androgen receptor-deficient environment acquired the phenotype typical for female VAT Treg cells, whereas they adopted the male phenotype in a wild-type environment (Extended Data Fig. 4b, c), indicating that sexual dimorphism of VAT Treg cells is extrinsic. This notion was confirmed when we generated Arfl/flFoxp3cre mice, in which the androgen receptor was deleted specifically in Treg cells (Extended Data Fig. 4d, e). Consistent with a Treg cell-extrinsic activity of oestrogen, oestrogen receptor-deficient and wild-type Treg cells were represented equally in mixed bone marrow chimeric mice containing congenically marked wild-type and Era−/− haematopoietic cells (Extended Data Fig. 4f). Finally, treatment of male wild-type mice with oestrogen resulted in a decrease in Treg cells specifically in the VAT and reduced KLRG1 and ST2 expression, whereas female wild-type mice treated with testosterone showed an increase in these parameters (Fig. 3g, Extended Data Fig. 4g–l). Overall, our data show that a sex hormone-dependent niche enforces the VAT Treg cell-specific phenotype.

VAT inflammation recruits Treg cells

To examine the nature of the sex-specific Treg cell niche in the VAT, we performed RNA-seq analysis of total VAT and subcutaneous adipose tissue from male and female mice. We observed substantial differences between adipose tissue from different depots. Compared with subcutaneous adipose tissue, the VAT was enriched in proinflammatory genes including Ccl2, Tnf and Il1b as well as Il33 (Fig. 3h, Supplementary Information). VAT displayed substantial sex-dependent transcriptional differences, with nearly 1,300 genes differentially expressed between tissues isolated from male and female mice (Fig. 3i, Extended Data Fig. 5a, Supplementary Information). In particular, the expression of genes that contribute to inflammation (Tnf, Ccl2 and Il1b), tissue fibrosis (Col6a5) and prostaglandin metabolism (Hpgds) was elevated in male VAT compared with female VAT. As Treg cells from male VAT express high levels of CCR2, the high expression of its ligand, CCL2 (also known as MCP-1), in male VAT was of particular interest. To understand the Treg cell-intrinsic role of CCR2 in the VAT, we generated mixed bone marrow chimeric mice containing both CCR2-deficient (Ccr2−/−) and wild-type haematopoietic cells. In male chimeric mice, we observed a reduction of Ccr2−/− VAT Treg cells compared with wild-type cells as well as reduced expression of prototypical VAT Treg cell markers. This defect was restricted to the VAT, as spleens or small intestines contained similar proportions of Ccr2−/− and wild-type Treg cells (Fig. 3j, Extended Data Fig. 5b, c). The difference was specific to Treg cells, as ILC2s of either genotype were similarly represented in the VAT of chimeric mice (Extended Data Fig. 5d). By contrast, mice deficient in TNF, IL-1β and IFNγ showed no substantial loss or altered phenotype of VAT Treg cells (Extended Data Fig. 5e). Notably, a small fraction of Treg cells in the spleens of both male and female mice co-expressed KLRG1 and CCR2, and these cells could be expanded by administration of IL-33 (Extended Data Fig. 5f–h). This suggested that VAT Treg cells were recruited from splenic precursors, a notion consistent with recent work25 and long-term parabiosis experiments, which showed that Treg cells—unlike ILC2s—are continuously recruited to the VAT (Extended Data Fig. 5i–k). In line with a critical role for sex hormones in regulating the abundance of inflammatory mediators in the VAT, expression of Ccl2, Il1b and Il6 was higher in female Era−/− VAT compared with wild-type controls (Fig. 3k). Similarly, female mice treated with testosterone showed an increase in the expression of Il6, Ccl2 and Il1b in the VAT and increased VAT weight, whereas male mice treated with oestrogen showed reductions in these parameters (Extended Data Fig. 6a, b). Finally, treatment of male mice with celecoxib, a pharmacological inhibitor of cyclooxygenase-2 (COX2), reduced the amount of inflammatory mediators such as CCL2 and the abundance of VAT Treg cells in male mice (Extended Data Fig. 6c–e). Together, these results show that Treg cells use the same molecular cues as pro-inflammatory cells to populate the adipose tissue and are recruited to the male VAT in a manner dependent on CCL2 and limited by oestrogen.

Sex hormones control IL-33+ stromal cells

To characterize the tissue niche that imparts the phenotypic and molecular features of VAT Treg cells, we performed RNA-seq analysis of adipocytes, endothelial cells (CD31+Gp38−) and stromal cells (CD31−Gp38+) isolated from the VAT of male and female mice (Extended Data Fig. 6f, Supplementary Information). We first identified the source of IL-33, which acts as a critical growth factor that facilitates expansion of VAT Treg cells26,27. Consistent with recent reports13,28,29,30, Il33 expression was largely restricted to Gp38+ stromal cells, which were present in males and females in similar numbers (Extended Data Fig. 6g, i). However, Gp38+ cells of males and females showed marked differences in their transcriptional profiles, including elevated expression of Nt5e (encoding the ecto-nucleotidase CD73) in males, whereas Cd90—encoding a marker of mesenchymal stromal cells—was downregulated (Extended Data Fig. 6h). Differential expression of CD90 and CD73 segregated four distinct Gp38+ stromal cell populations, of which the two that expressed CD73+ were largely restricted to the male VAT and almost absent from other adipose tissue depots and from females (Fig. 4a, b, Extended Data Fig. 6j). Analysis of Il33GFP reporter mice revealed that IL-33 production was restricted to Gp38+ stromal cells, including the CD73+ fraction (Fig. 4c, d, Extended Data Fig. 7a). Indeed, we detected fewer IL-33-producing stromal cells in female VAT compared with male VAT (Extended Data Fig. 7b, c). Consistent with the idea that IL-33 is limiting in female mice, administration of IL-33 led to robust expansion of female VAT Treg cells, which proliferated locally in the VAT and upregulated ST2 (Extended Data Fig. 7d–h). Further supporting this model, Treg cells sorted from the spleens of transgenic mice expressing a VAT-specific T cell receptor25 and transferred into congenically marked mice, populated the VAT of male mice more efficiently and expressed higher amounts of ST2 compared with Treg cells transferred into female mice (Extended Data Fig. 7i). Male Ar−/− mice had significantly fewer CD73+ stromal cells, whereas female Era−/− mice had more CD73+ stromal cells compared with their wild-type counterparts, suggesting that the development of these cells is supported by sex hormones (Fig. 4e, f). Indeed, treatment of female mice with testosterone resulted in the induction of CD73+ stromal cells, whereas treatment of male mice with oestrogen resulted in a reduction in the number of these cells (Fig. 4g, h, Extended Data Fig. 8a, b). Administration of IL-33 to Ar−/− mice resulted in pronounced population expansion of VAT Treg cells, suggesting that IL-33 is limiting in these mice (Extended Data Fig. 8c). By contrast, treatment of male mice with celecoxib resulted in an increase in CD73+ stromal cells (Extended Data Fig. 8d), indicating that the reduction of VAT Treg cells in celecoxib-treated mice was not linked to the loss of IL-33-producing stromal cells. Together these experiments reveal that sex hormones regulate the development of specific IL-33-producing stromal cell populations and show that IL-33 availability controls the size of the VAT Treg cell niche.

Fig. 4: Hormone-dependent stromal cells and BLIMP1 underpin VAT Treg cell sexual dimorphism.
figure 4

a, CD90 and CD73 expression in Gp38+ VAT stromal cells from C57BL/6 female and male mice. b, Percentage of CD73+ cells of CD90+ (left) and CD90- (right) cells. n = 12 mice of each sex. c, Il33 (GFP) and Gp38 expression in CD45− VAT stromal vascular fraction from male wild-type and Il33GFP mice. Representative of n = 6 mice. d, CD73 and CD90 expression in Gp38+Il33 (GFP)+ VAT stromal cells from male and female Il33GFP mice. Representative of n = 8 of both sexes. e, CD73 and CD90 expression within Gp38+ stromal cells from male wild-type and Ar−/− VAT. f–h, Proportions of CD73+ cells within Gp38+ VAT stromal cells of wild-type (n = 7) and Ar−/− (n = 8) males (f, left), wild-type (n = 5) and Era−/− (n = 7) females (f, right), C57BL/6 mice treated with testosterone (T) (n = 9 females treated, n = 5 females untreated, n = 8 males treated, n = 6 males untreated) (g) and C57BL/6 mice treated with oestrogen (E2) (n = 12 males treated, n = 6 males untreated, n = 10 females treated, n = 8 females untreated) (h). i, Proportions of Treg cells in the VAT (n = 7 per genotype) and spleens (n = 4 per genotype) of Blimp1fl/flFoxp3cre and Foxp3cre mice. j, Expression of FOXP3, ST2 and KLRG1 in VAT CD4+ T cells from Blimp1fl/flFoxp3cre and Foxp3cre control mice. k, ATAC-seq tracks show chromatin accessibility (male spleen, green; female VAT, red; male VAT, blue) and ChIP–seq shows BLIMP1 occupancy (black) in the Il1rl1 locus. Arrows indicate differential accessibility and boxes indicate BLIMP1-occupied sites. l, Flow cytometry plots show expression of FOXP3 and CD25 in VAT CD4+ T cells from wild-type and Il6−/− mice. m, Proportions of splenic, VAT and small intestine lamina propria (SI-LP) Treg cells among VAT CD4+ T cells of wild-type and Il6−/− mice (n = 4 per genotype). Two-tailed unpaired t-test; data are mean ± s.d. Data are pooled or representative of two independent experiments.

Source data

Full size image

BLIMP1 controls VAT Treg cell genes

Finally, to elucidate how the VAT Treg cell-specific transcriptional landscape is shaped in a Treg cell intrinsic manner, we tested the function of the transcription factor BLIMP1, which was increased in VAT Treg cells from male mice compared with those from female mice (Extended Data Fig. 8e). Male Blimp1fl/flFoxp3cre mice, which lack BLIMP1 specifically in Treg cells, exhibited reduced numbers of VAT Treg cells, which showed lower ST2, KLRG1, TIGIT and CCR2 expression, and upregulated CD103 and CD62L expression (Fig. 4i, j, Extended Data Fig. 8f). Consistent with a loss of VAT Treg cells, male Blimp1fl/flFoxp3cre mice showed reduced glucose tolerance (Extended Data Fig. 8g). RNA-seq revealed more than 2,500 genes deregulated in BLIMP1-deficient VAT Treg cells compared with control cells, including downregulation of VAT Treg signature genes Il1rl1, Klrg1, Ccr2, Pparg and Il10, resulting in a transcriptional profile similar to that of female VAT Treg cells (Extended Data Fig. 9a, b, Supplementary Information). To identify direct targets of BLIMP1, we interrogated our ATAC-seq data and previously published BLIMP1 chromatin immunoprecipitation with sequencing (ChIP–seq) data31. In total, we detected 2,095 ChIP peaks that overlapped with open chromatin in VAT Treg cells. These sites were associated with BLIMP1 target genes such as Ccr7, Tcf7, Klf2, Cxcr5, S1pr1, Myc and Bcl6, and key genes of the VAT Treg cell signature, including Il10, Ccr2 and Il1rl1, suggesting that these genes were direct transcriptional targets of BLIMP1 in VAT Treg cells (Extended Data Fig. 9c, d, Supplementary Information). BLIMP1-binding sites were detected in multiple loci with differential accessibility in splenic Treg cells and male and female VAT Treg cells, respectively. This included sites in Il1rl1, which showed full accessibility in male VAT Treg cells but reduced accessibility in female VAT Treg cells and was fully closed in splenic Treg cells (Fig. 4k). Notably, CCR2+KLRG1+ splenic Treg cells also expressed BLIMP1 and Pparg, and loss of BLIMP1 resulted in a loss of CCR2 expression and an overall reduction of KLRG1+ Treg cells (Extended Data Fig. 10a–e). IL-6 and IL-4, both abundant in the VAT, potently induced BLIMP1 in Treg cells in vitro, whereas IL-33 did not (Extended Data Fig. 10f). Il6 was mainly expressed by VAT dendritic cells, macrophages and ILC2s (Extended Data Fig. 10g). Consistent with an important role for IL-6 in VAT physiology, CD73+ stromal cells, Treg cells and ILC2s were reduced in the VAT of male IL-6-deficient mice, and VAT Treg cells showed impaired expression of ST2, KLRG1 and CCR2 (Fig. 4l, m, Extended Data Fig. 10h–j). Administration of IL-33 to IL-6-deficient mice resulted in pronounced population expansion of VAT Treg cells, suggesting that IL-6 per se is not essential for the differentiation of VAT Treg cells but promotes the availability of IL-33 (Extended Data Fig. 10k).

The VAT constitutes an inflammatory environment, and low-grade inflammation—which increases in obesity—contributes to the development of metabolic disease and type 2 diabetes. Our data show that in females, inflammation is limited by oestrogen. In males, however, increased VAT inflammation and male-specific IL-33-producing stromal cells mediate the active recruitment and local expansion of Treg cell numbers in a BLIMP1-dependent manner. This pathway constitutes a male-specific feedback circuit that limits inflammation in the VAT (Extended Data Fig. 10l).

Methods

No statistical methods were used to predetermine sample size. The experiments were not randomized. The investigators were not blinded to allocation during experiments and outcome assessment.

Mice

Era−/− (Jax stock no. 026176), Ar−/− (ref. 32), Arfl/fl (ref. 33), Ccr2−/− (Jax stock no. 004999), Blimp1fl/fl (ref. 34), Blimp1GFP (ref. 35), Il33GFP (ref. 36), Il6−/− (Jax stock no. 002650), Tnf−/− (Jax stock no. 003008), Il1b−/− (ref. 37), Rag1−/− (Jax stock no. 002216) and C57BL/6 Ly5.1 mice were bred at the The Walter and Eliza Hall Institute animal facility. Foxp3cre (Jax stock no. 016961), Foxp3RFP (Jax stock no. 008374) and IL10GFP (Jax stock no. 008379) mice were purchased from the Jackson laboratory. All mouse lines were maintained on a C57BL/6J (Ly5.2) background, except Il33−/−, which is on a C57BL/6N background. Mice were analysed at 25–30 weeks of age unless specified. Mice were maintained and used in accordance with the guidelines of the WEHI Animal Ethics Committee.

Bone marrow chimaeras

Bone marrow chimeric mice were generated by lethally irradiating recipient mice with twice with 550 röntgen (R) and reconstituting them with 1 × 107 congenically marked bone marrow cells. Ccr2−/−/Ly5.1 mixed chimeric mice were made in Rag1−/− recipients irradiated twice with 350 R.

Antibodies and flow cytometry

Fluorochrome-conjugated antibodies directed against the following mouse antigens were used for analysis by flow cytometry. Antibodies, clone names and manufactures: Invitrogen or eBioscience, KLRG1 (2F1) FITC, BV711; Thy1.2 (30-H12) FITC; γδTCR (ebioGL3) FITC; CD25 (PC61.5) PEcy7; Gp38 (eBio8.1.1) PEcy7; KLRG1 (2F1) PEcy7; ST2 (RMST2-2) APC; Tigit (GIGD7) eFluor660; GITR APC; CD25 (PC61.5) APC; TCRβ (H57-597) PerCP Cy5.5; Foxp3 (FJK-16 s) eFluor450; Ly5.2 (104) eFluor450; Gata3 (TWAJ) PE; CD103 (2E7) PE and EOMES (Dan11mag) PerCPeFluor710; BD Biosciences, Ly5.2 (104) FITC; CD4 (GK1.5) BUV496; CD19 (1D3) BUV737; CD8a (53-6.7) BUV737; CD45.2 (104) BUV395; CD11b (M1/70) BUV737 and KLRG1 (2F1) BV711; BioLegend, CD45.1 (A20) FITC; F4/80 (BM8) FITC; CD45.2 (104) BV605; CD4 (GK1.5) APC/cy7; CD11c (N418) BV711; CD69 (H1.2F3) BV711; F4/80 (BM8) BV605; CD140a (APA5) APC; CD31 (PacBlue) and CD73 (TY/11.8) PE; R&D Systems, mCCR2 A700; WEHI, NK1.1 (PK136) Alexa594. For surface staining, antibodies were diluted at 1:200 and incubated with cells in PBS with 2% BSA on ice or 4 °C for 30 min. The cell pellet was resuspended in live/dead stain containing PBS with 2% BSA. For intranuclear staining, a FOXP3 staining kit was used (Invitrogen).

Preparation of lymphocytes from adipose tissue

Unless indicated otherwise, perigonadal VAT was collected from 30- to 32-week-old male or female mice, finely minced and suspended in 0.025% collagenase type IV (Gibco) (2 ml collagenase per gram fat). The suspension was incubated for 45 min at 37 °C in a shaker. After incubation the suspension was 10 times diluted with PBS + 2% FCS and spun at 800g for 15 min at 4 °C. The upper adipocyte fraction was discarded and the stromal vascular fraction that settled down was further purified to obtain lymphocytes by Histopaque (Sigma) gradient.

Isolation of intestinal lamina propria lymphocytes

Lamina propria lymphocytes (LPLs) were extracted from the small intestine. In brief, Peyer’s patches removed and intestines were opened longitudinally and cut into small pieces (<5 mm). Epithelial cells and intraepithelial lymphocytes were removed by washing with HBSS and incubating with 5 mM EDTA for 30 min at 37 °C. The intestinal pieces were washed with RPMI and 10% FCS, and LPLs were isolated by digestion with 1 μg ml−1 DNase (Sigma-Aldrich) and 200 μg ml−1 collagenase III (Worthington) for 40 min at 37 °C. The LPL fractions were purified with a 40–80% Percoll (GE Healthcare) gradient.

Parabiosis experiments

Mice were anaesthetized with ketamine and xylazine at 10 μg g−1. Skin was shaved and disinfected by wiping with alcohol prep pads and betadine three times. Matching incisions were made from the olecranon to the knee joint of each mouse and subcutaneous fascia were bluntly dissected to create 0.5 cm of free skin. The olecranon and knee joints were attached by a 5-0 silk suture, and dorsal and ventral skin were attached by continuous staples or sutures. Betadine was used to cover the entire incision following surgery.

Treg cell cultures

Treg cells were purified from Foxp3RFP mice by fluorescence-activated cell sorting. Sorted Treg cells were cultured in 96-well flat-bottom plates with a density of 100,000 cells per well in 250 μl medium. Plates were pre-coated with anti-CD3 (2C11) (5 μg ml−1) and Treg cells were cultured in complete IMDM medium with IL-2 (100 U) and soluble anti-CD28 (2 μg ml−1) along with other cytokines mentioned.

Adoptive transfer of Treg cells

Five million CD4+ T cells were purified from pooled spleen and lymph nodes of 6- to 8-week-old male or female CD45.2+ vTreg53 TCR Tg+ littermates, using the Dynabeads Untouched Mouse CD4 Cells Kit (Thermo Fisher Scientific), and injected intravenously into sex-matched 6–8-week-old B6.CD45.1+ mice. Engraftment and enrichment of donor-derived cells in the spleen, epididymal VAT, ovarian VAT or inguinal subcutaneous adipose tissue of recipient mice were analysed 12 weeks after transfer.

Glucose-tolerance tests

Glucose (1.75 g per kg of body weight) was injected intraperitoneally to mice fasted for 8 h. Blood samples were obtained from the tail tip at the indicated times, and blood glucose concentrations were measured using a handheld glucometer (Accu-Chek Performa, Roche).

Intracellular IL-33 staining

Intracellular IL-33 was detected by fixation and permeabilization using True-Nuclear Transcription Factor Buffer Set (BioLegend) as per the manufacturer’s instructions followed by incubation with a goat anti–IL-33 polyclonal primary antibody (no. AF326, R&D Systems) and a donkey anti-goat Cy3 secondary antibody (no. 705-166-147, Jackson Immunoresearch Laboratories).

Celecoxib treatment

Seven-week-old male C57BL/6 were fed with celecoxib containing diet 29 mg kg−1 day−1 for 15 weeks.

Hormone treatment

Sixty-day release 0.25 mg 17-β oestradiol pellets (SE-121, Innovation Research of America) were surgically implanted subcutaneously in 7–8-week-old male and female C57BL/6 mice. For testosterone treatment, 4 mg testosterone (Sigma Aldrich, T1500) powder was packed in 1-cm silastic tubing and surgically implanted subcutaneously in 7–8-week-old male and female C57BL/6 mice. Mice were analysed six weeks post-implantation.

Serum adipokine and insulin measurement

Bio-Plex Pro mouse diabetes immunoassay kit (Bio Rad, 171F7001M) was used to measure fasting serum concentrations of adipokines according to the manufacturer’s protocol. Adiponectin was measured with a separate kit from Bio Rad (Bio-Plex Pro Mouse Diabetes Adiponectin Assay Kit, 171F7001M).

ICI 182-780 treatment

For the oestrogen receptor antagonist experiments, adult C57BL/6 mice were ovariectomized and then either not treated any further or treated with 5 mg of ICI 182-780 (intraperitoneal injections, 1 per week for 6 weeks). At the end of the experiment, mice were collected and uterine weight used to confirm that ICI 182-780 treatment was successful.

Cytokine treatment

IL-33 (R&D systems) at 0.5 μg per head was administered intraperitoneally for 3 alternate days and mice were analysed on day 7. PBS was used as control.

High-fat diet

Male C57BL/6 mice were fed a high-fat diet in which 59% of the total energy is derived from lipids (59 kcal% fat, Specialty Feeds, SF03-002) for 40 weeks.

Echo MRI and indirect calorimetry

Body composition (fat and lean mass) was measured with a 4-in-1 EchoMRI body composition analyser (EchoMRI), and whole-body oxygen consumption was measured with a Comprehensive Laboratory Animal Monitoring System (Columbus Instruments) as previously described38.

RNA extraction

RNA was extracted from VAT using Qiagen RNeasy lipid tissue mini kit as per the manufacturer’s protocol. RNA from Treg cells and stromal cells was isolated using Qiagen RNeasy plus micro kit as per the manufacturer’s protocol. VAT was pooled from three mice per sample. VAT Treg cells and stromal cells were sorted from pooled VAT from five or more mice. Stromal cells were sorted by flow cytometry as Gp38+ or Gp38− from the CD45−Ter119− population. Adipocytes were purified after removal of the stromal vascular fraction.

RNA-seq and analysis

Samples were generated from a male and female mouse respectively. All samples were sequenced on an Illumina NextSeq 500 generating 75 bp paired end reads. Reads were aligned to the mouse reference genome GRCm38/mm10 using the Subread aligner (v.1.6.2)39. Mapped reads were assigned to NCBI RefSeq mouse genes and genewise counts were produced by using featureCounts40. Genes that failed to achieve a CPM (counts per million mapped reads) value of 0.5 in at least 2 libraries were excluded from downstream analysis. Read counts were converted to log2 CPM, quantile-normalized and precision-weighted with the voom function of the limma package41,42. A linear model was fitted to each gene, and empirical Bayes-moderated t-statistics were used to assess differences in expression43. Genes were designated differentially expressed if they achieved a false discovery rate (FDR) less than 0.1 and a fold change greater than 1.2.

ATAC-seq and analysis

ATAC-seq was performed as described44. In brief, 35, 000–50,000 Treg cells were resuspended in 50 μl of chilled ATAC lysis buffer (10 mM TrisHCL, 10 mM NaCl, 3 mM MgCl2 and 0.1% (v/v) Igepal CA-630) and centrifuged for 10 min at 500g at 4 °C. Nuclei of an equivalent of 25,000 cells were resuspended in 25 μl of Tagmentation buffer (2× Tagment DNA buffer, 20× Tagment DNA enzyme (Nexterea library preparation kit) and incubated at 37 °C for 30 min. Next, Tagmented DNA was isolated using the MinElute PCR purification kit (Qiagen). Samples were PCR-amplified with a common forward primer and unique barcoding reverse primers before tagmentation efficiency and concentration was assessed with a bioanalyzer high sensitivity DNA analysis kit (Agilent). Pooled libraries were cleaned from small (<150 bp) fragments and sequenced on a NextSeq 500 sequencer (Illumina) producing paired-end 75 bp reads. Sequence reads were mapped to mouse genome GRCm38/mm10 using Subread. Only uniquely mapped reads were retained. ATAC peaks were called using Homer (v.4.9) with a FDR cut-off of 10−5. Overlapping peaks from different samples were merged into a single peak region that covers all the overlapping peaks. Mapped reads were assigned to all the merged regions for each sample using featureCounts. Regions were removed from analysis if they failed to achieve a CPM value of 0.7 or higher in at least one sample. Regions were annotated by assigning them to the nearest gene. Read counts for regions were converted to log2 CPM and precision-weighted with the limma voom function. A linear model was fitted to each region, and empirical Bayes moderated t-statistics were used to assess differences in chromatin accessibility. A FDR cut-off of 0.2 was applied for calling differentially accessed regions.

ChIP–seq analysis

BLIMP1 ChIP–seq reads were mapped to the mouse reference genome GRCm38/mm10 using the Subread aligner. BLIMP1 binding peaks were called using Homer (v.4.9) with a FDR cutoff of 10−4. Peaks were assigned to their nearest gene. Overlapping peaks between ChIP–seq data and ATAC-seq data were identified as those that have at least 1 bp overlap.

Statistics

Unless specified otherwise, t-tests were performed to test for statistical significance and error bars denote mean ± s. d. unless specified.

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.