Foxp3^YFP-cre mice (stock number 016959) and Foxp3^{eGFP-Cre-ERT2} mice (stock number 016961) were purchased from the Jackson Laboratory. Rosa26^Stat5b-CA mice were provided by Dr. Alexander Rudensky at Memorial Sloan Kettering Cancer Center (10). Ogt^fl/fl Mice (Jackson Laboratory stock number 004860) were kindly provided by Dr. Xiaoyong Yang at Yale University. All mice were on C57BL/6 background. Mice were free to access water and fed on a regular chow, tamoxifen food (Teklad, TD.130860), or 60% high fat diet (ResearchDiets, D12492) as indicated. All procedures were approved by the Institutional Animal Care and Use Committee at the University of Minnesota. All relevant ethical regulations for animal testing and research were complied with.

Body weights were monitored every week. Body composition was assessed using an EchoMRI system (39). For food intake measurement, mice were individually housed and food consumption was weighed every morning for 7 consecutive days. For the metabolic cage study, mice were first acclimated in metabolic chambers (Columbus Instruments), and then physical activity and energy expenditure were measured continuously for at least 2 days (40). For oral glucose-, and insulin- tolerance tests. 16 h fasted mice were given glucose (1.5 g/kg body weight) by gavage; 6 h fasted mice were injected with insulin (1 U/kg body weight) intraperitoneally. Tail-vein blood collected at the designated times was used to measure blood glucose level using a Contour Glucometer (Bayer). For lipid tolerance test, mice were fasted for 4 h, gavaged with 200 ul of olive oil, and tail-vein blood was collected every 1.5 h for quantification of triglyceride using a colorimetric assay kit (Cayman #10010303). Fecal energy was determined with a bomb calorimeter at Dr. Pedro Urriol’s lab. Fecal lipid was extracted with chloroform: methanol mix (41), weighed and calculated as percentage to fecal weight.

For surface markers, cells were stained in PBS containing 0.5% BSA with relevant antibodies at 4°C for 30 min. For analysis of intracellular markers, cells were first fixed with Fixation/Permeabilization buffer (ThermoFisher, catalog no. 00-5123) at 4°C for 30 min and then stained in Permeabilization Buffer (ThermoFisher, catalog no. 00-8333) with relevant antibodies at 4°C for 30 min. Flow cytometry data were acquired on BD Fortessa H0081 or X-20 and analyzed with Flowjo.

Mouse tissues were dissected and fixed in 10% buffered formalin. Sectioning and haematoxylin & eosin (H&E) staining were performed by the Comparative Pathology Shared Resource of UMN.

RNA was isolated with Trizol and reverse transcribed into cDNA with the iScript™ cDNA Synthesis Kit. Real-time RT-PCR was performed using iTaq™ Universal SYBR^® Green Supermix and gene-specific primers on a Bio-Rad C1000 Thermal Cycler. All data were normalized to the expression of 18s or 36b4.

Tissues were lysed in buffer containing 1% Nonidet P-40, 50 mM Tris 3 HCl, 0.1 mM EDTA, 150 mM NaCl, proteinase inhibitors and an OGT inhibitor, TMG. Equal amounts of protein lysates were electrophoresed on 4-20% SDS-PAGE gels and transferred to Nitrocellulose membranes. Primary antibodies were incubated at 4°C overnight. Western blotting was visualized by peroxidase conjugated secondary antibodies and ECL chemiluminescent substrate with a Bio-Rad ChemiDoc Imaging System or imaged by fluorescent IgG secondary antibodies with a LI-COR Odyssey 9120 imager.

Results are shown as mean ± SEM. The comparisons were carried out using two-tailed Student’s t test, one-way ANOVA followed by multiple comparisons with the Tukey adjustment, two-way ANOVA using GraphPad Prism 7.

Both the Foxp3 and Ogt genes are located on the mouse X Chromosome; about 40 centimorgans apart ( Figure 1A ). When we generated Treg cell-specific OGT knockout (KO) using the Foxp3^YFP-Cre mice with YFP-Cre knocked into the endogenous Foxp3 locus ( Figure S1A ), hemizygous male KO mice developed scurfy phenotypes and died of autoimmune disease by the age of 4 weeks (38), thus preventing us to study their whole-body metabolism. Due to random inactivation of the X chromosome in females, Foxp3^YFP-Cre/wtOgt^F/F female mice possess both OGT-sufficient (WT) and -deficient (KO) Treg cells ( Figure 1B ). Despite lineage instability and suppressor dysfunction of KO Treg cells, the remaining WT cells were sufficient to avert autoimmunity in mice (38). Therefore, Foxp3^YFP-Cre/wtOgt^F/F females and their littermate controls were used to investigate metabolic outcomes. When fed with normal chow (NC), no difference in body weight was observed in 12-week-old mice ( Figure 1C ). Compared to controls, Foxp3^YFP-Cre/wtOgt^F/F females had higher blood glucose levels during insulin tolerance test, indicative of impaired insulin sensitivity ( Figure 1D ). Glucose tolerance test revealed similar blood glucose levels between the two genotypes ( Figure S1B ), suggesting that pancreatic insulin secretion might be increased to compensate insulin resistance in KO mice. We then treated mice with insulin to assess its downstream signaling, and found similar AKT phosphorylation in gonadal white adipose tissue (gWAT) and liver ( Figure S1C ). However, insulin-stimulated Ser473 phosphorylation of AKT in brown adipose tissue (BAT) was diminished in Foxp3^YFP-Cre/wtOgt^F/F females ( Figure 1E ). These results demonstrate that OGT in Treg cells controls insulin pathway activation in BAT and whole-body insulin sensitivity.

We then fed animals with 60% high fat diet (HFD) to induce obesity. Female Foxp3^YFP-Cre/wtOgt^F/F mice gained more weight than controls ( Figure 2A ). Body composition analysis showed that weigh gain was due to increased fat mass ( Figures 2B, C ). Similar to animals fed with chow, HFD Foxp3^YFP-Cre/wtOgt^F/F females displayed impaired insulin tolerance ( Figure 2D ), but normal glucose tolerance ( Figure 2E ).

We went on to determine what caused diet-induced adiposity in Foxp3^YFP-Cre/wtOgt^F/F mice. Their daily intake of HFD was similar to control mice, when adjusted to body weight using ANCOVA analysis ( Figure 2F ). Metabolic cage study found no difference in locomotor activity ( Figure 2G ) or respiratory exchange ratio ( Figure 2H ). Rather, heat production, when adjusted to body weight (42), was significantly reduced in Foxp3^YFP-Cre/wtOgt^F/F mice during both light and dark periods ( Figures 2I, J ). Moreover, when challenged with cold, Foxp3^YFP-Cre/wtOgt^F/F animals showed reduced heat production than the controls ( Figure 2K ), indicative thermogenic defects. Together, these data demonstrate that OGT-controlled Treg cell function enables metabolic adaptation to positive energy balance.

The regulation of Treg cell suppressive program by OGT is dependent on STAT5 O-GlcNAcylation and activation (38). We then sought to determine whether STAT5 overactivation, to the opposite of OGT knockout, would drive Treg cell action in systemic metabolism. Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt male mice were generated to specifically overexpress a constitutively active (CA-) form of STAT5B in Treg cells (10). Flow cytometric analyses confirmed that STAT5B-CA not only increased the number of FOXP3⁺ Treg cells in the lymph nodes and the activity of the Foxp3 gene promoter ( Figure 3A ), but also drastically expanded the population of effector Treg cells that were CD44⁺KLRG1⁺ ( Figure 3B ) or CD44⁺CD103⁺ ( Figure 3C ). When fed with a normal chow, even though with comparable body weight ( Figure 3D ), Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice showed a slight, but statistically significant reduction in fat mass ( Figure 3E ). There was a declining trend of blood glucose levels during the insulin tolerance test ( Figure 3F ). Similar locomotor activity ( Figure 3G ), energy expenditure ( Figure 3H ), and energy or lipid excretion into the feces ( Figures 3I, J ) were observed between control and Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice.

We went on to challenge the mice with HFD. The resistant to HFD-induced weight gain could be readily seen in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice as early as 3 weeks after diet-switching ( Figure 4A ). The reduction in weight gain was solely because of reduced fat mass ( Figure 4B ). Both BAT and gWAT depots were significantly smaller in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice, while liver weight remained the same ( Figure 4C ). We then asked what led to the lean phenotype after STAT5B-CA expression in Treg cells. Continuous metabolic monitoring revealed a substantial reduction of food intake in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice ( Figure 4D ), while energy expenditure ( Figure 4E ) and locomotor activity ( Figure 4F ) were both comparable to wildtype controls. Moreover, there was a tendency that fecal energy excretion was increased after Treg cell hyperactivation ( Figure 4G ). We then measured lipid content in the feces and found more lipids were excreted in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice ( Figure 4H ). As expected, these HFD-fed but relatively lean Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice show much better glucose handling during the glucose tolerance test ( Figure 4I ). These data demonstrate that Treg cell expansion and activation stimulated by tonic STAT5 signaling restrain dietary fat intake/absorption and render mice resistant to diet-induced metabolic disturbances.

Due to stochastic activity of the Foxp3 promoter (43), constitutive overexpression of STAT5B-CA led to progressive splenomegaly and hematopoietic neoplasia in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice (data not shown), similar to that after STAT5 activation in other hematopoietic compartments (44–46). We therefore used the inducible Foxp3^{eGFP-Cre-ERT2} line to generate Foxp3^{eGFP-Cre-ERT2/Y}Rosa26^Stat5b-CA/wt male mice, fed them with tamoxifen-containing food to induce STAT5-CA expression in Treg cells, and challenged them with HFD feeding ( Figure 5A ). A trending reduction in weight gain could be observed in Foxp3^{eGFP-Cre-ERT2/Y}Rosa26^Stat5b-CA/wt mice ( Figure 5B ). When compared to their control counterparts, Foxp3^{eGFP-Cre-ERT2/Y}Rosa26^Stat5b-CA/wt mice possessed significantly less BAT ( Figure 5C ) and slightly smaller WAT ( Figure 5D, E ). No change in muscle mass was noticed ( Figure 5F ). Moreover, Foxp3^{eGFP-Cre-ERT2/Y}Rosa26^Stat5b-CA/wt animals also displayed much improved insulin tolerance ( Figure 5G ). Concomitantly, insulin signaling activity, reflected by AKT Ser473 phosphorylation, was augmented in BAT of Foxp3^{eGFP-Cre-ERT2/Y}Rosa26^Stat5b-CA/wt mice ( Figure 5H ). The more profound protection observed in the former constitutive model might stem from stronger and broader expression of STAT5B-CA. Nonetheless, using two complementary models, we demonstrate that STAT5 activation in Treg cells is sufficient to protect mice from diet-induced adiposity and insulin resistance.

We then inquired into the causes of metabolic protection after STAT5-dependent Treg cell overactivation. We first interrogated the involvement of BAT thermogenesis. BAT histology clearly revealed smaller adipocytes and less lipid deposition in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice ( Figure 6A ). However, the expression of thermogenic genes remained unchanged in BAT ( Figure 6B ), or even reduced in inguinal WAT (iWAT) ( Figure 6C ). This suggests that reduced weight gain by Treg cell activation was not a result of increased thermogenesis.

Not only was gWAT mass reduced in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice ( Figures 4B, C ), but adipocyte size was also much smaller ( Figure 6D ). Crown-like structures were more evident in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt gWAT ( Figure 6D ), suggesting active clearing of dying adipocytes by macrophages (47). Interestingly, the expression of both Cd36 ( Figure 6E ) and Lpl ( Figure 6F ), two important genes required for cellular uptake of free fatty acid (48), were substantially ablated in gWAT and iWAT of Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice. There seemed to be no defect in dietary intake of lipid in the intestine, as intestinal Cd36 expression was intact ( Figure 6G ) and blood triglyceride levels after an oral gavage of olive oil were even higher in Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice ( Figure 6H ). Taking in consideration of the elevated fecal lipid content ( Figure 4H ), we postulate that Treg cell activation suppresses lipid uptake into adipose tissues, leading to more energy excretion and less weight gain during HFD feeding.

Iron is critically important for adipose tissue function and systemic metabolism. Dietary iron supplementation augments thermogenesis, reduces fat mass, and protects mice from diet-induced weight gain (49, 50). On the other hand, nonanemic iron deficiency impairs thermogenesis and promotes visceral obesity in mice (51). Lowering adipocyte iron uptake via genetic deletion of transferrin receptor 1 (TFR1) leads to impaired thermogenesis, increased insulin resistance, and inflammation (52). We then inquired if the metabolic protection after STAT5-dependent Treg cell activation involves iron. Ferritin is a protein complex that stores cellular iron and its protein levels were elevated in BAT, gWAT and iWAT of HFD-fed Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt mice, compared to Cre-only controls ( Figures 7A–C ), indicative of excess iron storage in fat tissues. In Foxp3^YFP-Cre/YRosa26^Stat5b-CA/wt adipose tissues, there was a trending increase in Tfr1 gene expression ( Figure 7D ) and TFR1 protein levels ( Figures 7A–C ). Conversely, expression of the Fpn1 gene (encoding iron export protein Ferroportin) was significantly downregulated ( Figure 7E ). In contrast to adipose tissue, both liver and spleen showed substantial reduction in Ferritin protein levels ( Figures 7F, G ), accompanied with increased TFR1 expression. The presumptive impairment in hepatic iron storage was correlated with reduced expression of iron-responsive Hamp and Hfe genes ( Figure 7H ), which encode Hepcidin and Homeostatic iron regulator, respectively. As a result of reduced Hepcidin, intestinal genes involved in iron uptake including Dmt1, Cybrd1, Fn1 displayed varying degrees of upregulation ( Figure 7I ). Collectively, STAT5B-CA-mediated Treg cell overactivation redistributes the iron pool from liver and spleen to adipose tissues, which may contribute to the alleviation of metabolic stress associated with diet-induced obesity.