Obesity reshapes regulatory T cells in the visceral adipose tissue by disrupting cellular cholesterol homeostasis

Abstract

Regulatory T cells (T_regs) accumulate in the visceral adipose tissue (VAT) to maintain systemic metabolic homeostasis but decline during obesity. Here, we explored the metabolic pathways controlling the homeostasis, composition, and function of VAT T_regs under normal and high-fat diet feeding conditions. We found that cholesterol metabolism was specifically up-regulated in ST2^hi VAT T_reg subsets. T_reg-specific deletion of Srebf2, the master regulator of cholesterol homeostasis, selectively reduced ST2^hi VAT T_regs, increasing VAT inflammation and insulin resistance. Single-cell RNA/T cell receptor (TCR) sequencing revealed a specific loss and reduced clonal expansion of ST2^hi VAT T_reg subsets after Srebf2 deletion. Srebf2-mediated cholesterol homeostasis potentiated strong TCR signaling, which preferentially promoted ST2^hi VAT T_reg accumulation. However, long-term high-fat diet feeding disrupted VAT T_reg cholesterol homeostasis and impaired clonal expansion of the ST2^hi subset. Restoring T_reg cholesterol homeostasis rescued VAT T_reg accumulation in obese mice, suggesting that modulation of cholesterol homeostasis could be a promising strategy for T_reg-targeted therapies in obesity-associated metabolic diseases.

Fig. 1.. VAT Tregs exhibit enhanced cholesterol metabolism at steady state.

(A to B) Transcriptome analysis of VAT vs lymphoid-tissue Foxp3⁺ Tregs from 30-week-old Foxp3^YFP-cre mice at steady state (n=3). (A) GSEA showing top enriched (NES>1.5, P<0.05) or suppressed (NES<−1.5, P<0.05) hallmark pathways in VAT Tregs. NES, normalized enrichment score; FDR, false discovery rate. (B) Volcano plot comparing VAT vs lymphoid-tissue Tregs. Genes involved in cholesterol biosynthesis and uptake (Red) and those not directly related to cholesterol metabolism (Black) are highlighted. FC, fold change. (C) Relative cholesterol content in Foxp3⁺ Tregs and Foxp3⁻ Tconvs from 20–25-week-old male vTreg53⁺ Foxp3^{thy1.1 or GFP} Pparg^Tdt reporter mice. Each data point represents cells pooled from 2–4 mice (n≥4). (D) Relative uptake of CholEsteryl BODIPY^™ 542/563 C₁₁ in Foxp3⁺ Tregs and Foxp3⁻ Tconvs from the spleen or VAT of 25-week-old Foxp3^YFP-cre reporter mice fed a normal chow diet (NCD) (n=6). (E) CH-related gene expression in Foxp3⁺ PPARγ⁺ VAT Tregs isolated from 15–25-week-old male Foxp3^thy1.1 Pparg^Tdt reporter mice and cultured with or without anti-CD3/CD28 Dynabeads^™ and/or 1uM pioglitazone (Pio) for 3 days. NT, no treatment (n=3). (F to H) Effect of Srebf2 (G) or Srebf1 (H) ablation on the steady state accumulation of adoptively transferred vTreg53⁺ Tregs. (F) Experimental scheme. (G, H) Top: Representative flow cytometry plots; Bottom: Proportion of mRFP⁺ (Ctl sg) or GFP⁺ (Srebf2 sg or Srebf1 sg) cells normalized to the input. Cells were gated on transduced clonotype⁺ (Vα2⁺ Vβ4⁺) donor-derived (CD45.1⁻Thy1.1⁺) Tregs in the spleen and VAT of recipient mice (pooled from 2–3 independent experiments; n≥7). Plots show mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by unpaired (C, D) or paired Student’s t test (G, H).

Fig. 2.. Srebf2 is required for the homeostatic accumulation of Tregs specifically within the VAT.

(A) qPCR assessment of Srebf2 mRNA in CD4⁺ Foxp3⁻ Tconvs and CD4⁺ Foxp3⁺ Tregs from pooled spleen and lymph nodes of 8-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice. (n≥3). A.U., arbitrary unit. (B to G) Splenic and VAT Treg analysis in 25-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice fed a NCD (n≥6). (B) Bodyweight. (C) VAT weight. (D) Left: Representative flow cytometry plot. Gated on CD4⁺ T cells. Right: Summary plot showing the frequencies of Foxp3⁺ Tregs among CD4⁺ T cells in the spleen and VAT. (E) Total Foxp3⁺ Tregs in the spleen. (F) Total Foxp3⁺ Tregs per gram of VAT. (G) Frequency of KLRG1⁺ Tregs among total Foxp3⁺ Tregs. (H) Frequency of Foxp3⁺ Tregs among CD4⁺ T cells in various other nonlymphoid tissues from 25-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice. SAT, subcutaneous adipose tissue (n≥5). (I to J) Frequency (I) and number (per gram VAT) (J) of Foxp3⁺ VAT Tregs from 8-, 14–16, and 25-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice fed a NCD (pooled from 2–4 independent experiments; n≥5) (Mean ± SEM). Data from 25-week-old mice also summarized in Figures 2D and 2F. Plots show mean ± SD, unless otherwise stated. *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001, ns (not significant) by unpaired Student’s t test (B, C, E, and F) or by two-way ANOVA with Šidák multi-comparison correction (A, D, G, H, I, and J).

Figure 3.. Loss of Srebf2 impairs the proliferation, but not survival or recruitment, of VAT Tregs.

(A to B) Analysis of Treg cell death in spleen and VAT of 14–16-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice (pooled from 2–3 independent experiments; n≥5). (A) Left: Representative flow cytometry plot. Right: Frequencies of Annexin V⁺ Tregs among total Foxp3⁺ Tregs. (B) Frequencies of PI⁺ Tregs among total Foxp3⁺ Tregs. (C) Analysis of Treg proliferation in 12-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice 12 hours following i.p. EdU injection (n≥3). Left: Representative flow cytometry plot. Right: Frequencies of Ki67⁺ EdU⁺ Tregs among total Foxp3⁺ Tregs in the spleen and VAT. (D to E) CD25⁺ Foxp3⁺ thymus Tregs from 6-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice (n≥4). (D) Left: Representative flow cytometry plot. Gated on CD4 single positive (SP) thymocytes. Right: Frequencies of CD25⁺ Foxp3⁺ Tregs among total CD4 SP thymocytes. (E) Total CD25⁺ Foxp3⁺ Tregs in the thymus. (F) Transcriptome comparison of PPARγ^lo precursor vs PPARγ⁻ Tregs in the spleen of 8-week-old Pparg^Tdt Foxp3^GFP male mice (n=4). Key genes involved in cholesterol biosynthesis and uptake are highlighted (Red). (G to J) Spleen and VAT Tregs from 8-week-old Foxp3^YFP-cre Pparg^Tdt Srebf2^WT/WT or Srebf2^flox/flox mice (pooled from 4 independent experiments; n≥5). (G) Left: Representative flow cytometry plot. Gated on CD4⁺ T cells. Right: Frequencies of PPARγ⁺ Tregs among total Foxp3⁺ Tregs. (H) Total PPARγ⁺ Foxp3⁺ Tregs in the spleen. (I) Total PPARγ⁺ Foxp3⁺ Tregs per gram of VAT. (J) CCR2 protein expression in PPARγ⁺ Foxp3⁺ Tregs. Mean Fluorescence Intensity (MFI) values were normalized to average MFI values in wildtype splenic Tregs within individual experiments. Plots show mean ± SD. *p < 0.05, **p < 0.01, ns (not significant) by unpaired Student’s t test (D, E, H, I) or by two-way ANOVA with Šidák multi-comparison correction (A, B, C, G, and J).

Fig. 4.. Treg-specific disruption of Srebf2 enhances obesity-associated pathology.

(A to O) 10–12-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice were fed a high fat diet (HFD) for 8 weeks (pooled from 3 independent experiments; n≥11). (A) Bodyweight. (B) VAT weight. (C) Left: Representative flow cytometry plot. Gated on CD4⁺ T cells. Right: Frequencies of Foxp3⁺ Tregs among CD4⁺ T cells in the spleen and VAT. (D) Total Foxp3⁺ Tregs in the spleen. (E) Total Foxp3⁺ Tregs per gram of VAT. (F) Frequency of ST2⁺ Tregs among total Foxp3⁺ Tregs. (G) Total CD45⁺ cells per gram of VAT. (H) Total CD45⁺ CD68⁺ CD11b⁺ macrophages per gram of VAT. (I) Total CD45⁺ CD11b⁺ SiglecF⁺ eosinophils per gram of VAT. (J to O) qPCR quantification of Tnfa (J), Il6 (K), Saa3 (L), Il1b (M), Col1a1 (N), and Pdgfra (O) expression in VAT. (P to R) Insulin tolerance test (ITT) of Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice fed a HFD for 7 weeks (data representative of 2 independent experiments; n≥4). (P) Blood glucose levels after 4 hours of fasting. (Q) Blood glucose levels measured 0-, 20-, 40-, 90-, and 120-minutes post insulin administration and normalized to pre-insulin, fasting glucose levels (0 minutes). Mean ± SEM. (R) Calculated area under the curve for ITT. Plots show mean ± SD, unless otherwise stated. *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001, ns (not significant) by unpaired Student’s t test (A, B, D, E, G, H, I, J, K, L, M, N, O, P, and R) or by two-way ANOVA with Šidák multi-comparison correction (C and F) or two-stage Benjamini, Krieger, & Yekutieli FDR procedure (Q).

Fig. 5.. Cholesterol homeostasis is required for the steady state VAT Treg transcriptional program.

(A to B) Bulk RNA-seq analysis of Foxp3⁺ VAT Tregs from 25-week-old Foxp3^YFP-cre Srebf2^flox/flox vs Foxp3^YFP-cre Srebf2^WT/WT male mice (n=3). (A) GSEA showing the most enriched (NES>1.3, P<0.05) or suppressed (NES<−1.3, P<0.05) hallmark pathways in Srebf2^flox/flox vs Srebf2^WT/WT VAT Tregs (n=3). (B) Volcano plot comparing the transcriptome of Srebf2^flox/flox vs Srebf2^WT/WT VAT Tregs. Key genes involved in cholesterol biosynthesis and uptake (Red) and those not directly related to cholesterol metabolism (Black) are highlighted. (C to E) Analysis of donor-derived Tregs following Fdft1 (C), Dhcr24 (D), or Ldlr (E), ablation. Experimental set up was the same as Fig 1F. Left: Representative flow cytometry plots; Right: Proportion of mRFP⁺ (Ctl sg) or GFP⁺ (Fdft1 sg, Dhcr24 sg, or Ldlr sg) cells normalized to the input. Cells were gated on transduced clonotype⁺ (Vα2⁺Vβ4⁺) donor-derived (CD45.1⁻Thy1.1⁺) Tregs in the spleen and VAT of recipient mice (pooled from 2 independent experiments; n≥7). (F to G) Volcano plot comparing the transcriptome of Srebf2^flox/flox vs Srebf2^WT/WT VAT Tregs (n=3). VAT Treg signature genes (21) (F) and HFD signature genes(23) (G) are highlighted in red (up-signature) or blue (down-signature) (gene lists shown in data file S1). The number of genes from each signature preferentially expressed by one or the other population are shown at the top of the plot. Plots show mean ± SD. *p < 0.05, **p < 0.01, ns (not significant) by paired Student’s t test (C-E) or by Chi-square test (F, G).

Fig. 6.. Srebf2 controls the composition and TCR repertoire of distinct Treg subsets.

(A to H) scRNA/TCR sequencing analysis of VAT Tregs isolated from male 25-week-old Foxp3^YFP-cre Srebf2^WT/WT and Srebf2^flox/flox mice fed a NCD (n=4). (A) UMAP showing distinct VAT Treg subsets in pooled Foxp3^YFP-cre Srebf2^WT/WT and Srebf2^flox/flox VAT Tregs. (B) Heatmap showing top 10 genes differentially expressed across various VAT Treg clusters. Key cluster-defining genes are listed on the left. (C) Table highlighting key cluster-defining markers (left) and feature plots showing expression of key markers across clusters (right). (D) Comparison of cell density between Foxp3^YFP-cre Srebf2^WT/WT (top) and Srebf2^flox/flox (bottom) VAT Tregs. (E) Comparison of cluster composition between Foxp3^YFP-cre Srebf2^WT/WT and Srebf2^flox/flox VAT Tregs. (F) Clone size distribution of pooled VAT Tregs from Foxp3^YFP-cre Srebf2^WT/WT and Srebf2^flox/flox mice. (G) UMAPs (left) and bar graphs (right) comparing overall clonal expansion between Foxp3^YFP-cre Srebf2^WT/WT and Srebf2^flox/flox VAT Tregs. Colors indicate individual cells (left) or the proportion of cells (right) containing TCR sequences represented 1, 2–9, or ≥10 times. (H) Comparison of clonal expansion between Foxp3^YFP-cre Srebf2^WT/WT and Srebf2^flox/flox VAT Tregs within individual clusters.

Fig. 7.. Strong TCR signaling, potentiated by cholesterol, promotes accumulation of ST2⁺ VAT Tregs.

(A to B) Enriched lymphocytes from VAT of 25-week-old male Foxp3^YFP-cre Srebf2^WT/WT and Srebf2^flox/flox mice were activated with 1μg/mL plate-bound anti-CD3ε and 0.5μg/mL anti-CD28 antibodies in vitro for 10 minutes (pooled from 3 independent experiments; n≥4). (A) Left: Representative flow cytometry plots; Right: Frequencies of pERK⁺ cells among total Foxp3⁺ Tregs with or without anti-CD3/28 stimulation. (B) Frequency of pERK⁺ cells among ST2⁻ or ST2⁺ Tregs following 10-minute of anti-CD3/28 activation. (C to E) Sorted Foxp3⁺ CXCR3⁻ ST2⁻ CD62L⁺ naïve Tregs from the pooled lymph nodes and spleens of 7–8-week-old Foxp3^YFP-cre Srebf2^WT/WT or Srebf2^flox/flox mice were activated for 5 days in vitro in 96 well plates pre-coated with 0.5μg/mL anti-CD28 and varying levels of anti-CD3ε antibody under ST2⁺ or CXCR3⁺ Treg polarizing conditions (data representative of 2 independent experiments; n=3). (C) Total ST2⁺ Tregs following activation under ST2⁺ Treg inducing conditions. Water-soluble cholesterol was supplemented at 25μg/mL. (D) Frequency of CD62L^lo cells out of total Tregs under ST2⁺ Treg inducing conditions. (E) Total CXCR3⁺ Tregs following activation under CXCR3⁺ Treg inducing conditions. (F to G) Effect of TCR signal strength on generation/expansion of ST2^hi VAT Tregs in vivo. (F) Experimental design scheme showing the adoptive transfer of vTreg53⁺ CD4⁺ T cells followed by immunization with low-affinity (Fat7) or high-affinity (Fat1562) vTreg53-specific surrogate peptide ligands. s.q, subcutaneously; i.p., intraperitoneally. (G) Left: Representative flow cytometry plots; Right: Frequencies of ST2^hi Tregs among total clonotype⁺ (Vα2⁺ Vβ4⁺) donor-derived (CD45.1⁻ Thy1.1⁺) Tregs within the VAT of recipient mice following immunization with surrogate peptide ligands (n=5). Plots show mean ± SD. *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001, ns (not significant) by unpaired Student’s t test (G) or by two-way ANOVA with Šidák (A, B) or Tukey (C, D, and E) multi-comparison corrections.

Fig. 8.. Obesity impairs the accumulation of ST2^hi VAT Tregs by disrupting cholesterol homeostasis.

(A) Expression of CH-related genes in VAT Tregs from 28-week-old male vTreg53⁺ Foxp3^GFP mice following 16 weeks NCD or HFD feeding (n≥2). (B) Relative cholesterol content in splenic and VAT Tregs from male vTreg53⁺ Foxp3^{thy1.1 or GFP} Pparg^Tdt reporter mice fed NCD or HFD for 16–20 weeks. Each data point represents pooled Tregs from 2–4 mice (n≥4). NCD data also shown in Figure 1C. (C to G) scRNA/TCR sequencing analysis of Foxp3⁺ VAT Tregs from 23–27-week-old male Foxp3^thy1.1 Pparg^Tdt mice fed a NCD or HFD for 18–22 weeks (n=5). (C) UMAP of total VAT Tregs and feature plots displaying genes defining distinct VAT Treg subsets. (D) UMAPs showing VAT Treg composition. (E) Bar graphs showing VAT Treg composition. (F) UMAPs showing the clonal expansion of VAT Tregs. Colors indicate individual cells containing TCR sequences represented 1, 2–9, or ≥10 times. (G) Cholesterol gene signature (gene list shown in data file S1) expression among VAT Treg clusters following NCD or HFD feeding. (H) Effect of Abcg1 ablation on the accumulation of adoptively transferred vTreg53⁺ Tregs during HFD-induced obesity. Following cell transfer, recipients were fed a HFD for 10–12 weeks prior to analysis (see Figure S6E for experimental scheme). Left: Representative flow cytometry plot; Right: Proportion of mRFP⁺ or GFP⁺ cells normalized to the input. Cells were gated on transduced clonotype⁺ (Vα2⁺ Vβ4⁺) donor-derived (CD45.1⁻ Thy1.1⁺) Tregs in the spleen and VAT of recipient mice (n=4). (I) Expression of cholesterol gene signature (gene list shown in data file S1) in human VAT Tregs from individuals across various ranges of body mass index (BMI). Plots show mean ± SD. *p < 0.05, **p < 0.01, ns (not significant) by paired Student’s t test (H) or by two-way ANOVA with Šidák multi-comparison correction (B).