STAT1 Dissociates Adipose Tissue Inflammation From Insulin Sensitivity in Obesity

The obesity epidemic contributes to the increased health burden of chronic inflammatory conditions, including insulin resistance, type 2 diabetes mellitus (T2DM), fatty liver, and cardiovascular disease (1). Obesity reflects facultative white adipose tissue (WAT) expansion that occurs during prolonged dietary stress. Although some clinical relationships explain how excess body weight causes insulin resistance in most individuals (2), WAT inflammation remains an enigmatic complication of obesity.

Obesity cultivates persistent low-grade inflammation that likely impacts the metabolic functions of WAT. Many studies demonstrate WAT inflammation causes local and systemic insulin resistance in rodents (3–6). In humans, expression of inflammatory cytokines in WAT durably correlates with BMI and insulin resistance (6–9). However, causal relationships between obesity-mediated inflammation and insulin resistance are still unclear, hindering the development of immune therapies to enhance the treatment of obesity.

In addition to mature adipocytes, WAT includes a stromal vascular fraction (SVF) containing lymphocytes and macrophages, which interact with one another and distant tissues through paracrine and endocrine signaling (10). Local elevation of the inflammatory cytokine interferon-γ (IFN-γ) correlates with maladaptive WAT expansion and systemic insulin resistance (8,9). The IFN-γ receptor (IFNGR1) communicates the IFN-γ signal through phosphorylation of Janus kinases and subsequent activation of STAT proteins. STAT1 mediates IFN-γ signaling and transcription of immune regulatory genes, including IRF1, IRF9, and ISG15 (11,12). Previously, we and others (13–15) demonstrated inhibition of IFN-γ signaling in WAT prevents development of insulin resistance and fatty liver in obese mice. Additional evidence suggests both type I and type II IFNs signal to repress transcription factors essential for adipocyte differentiation and mitochondrial function (16–18). Therefore, while IFN-γ signaling plays an important role in the development of insulin resistance, the mechanistic underpinnings of this observation remain poorly defined.

In this study, we used STAT1-deficient mouse models to investigate how the inflammatory cytokine IFN-γ affects insulin resistance and WAT inflammation. We found that disruption of IFN-γ–STAT1 signaling in mouse and human adipocytes relieves WAT inflammation and replenishes tricarboxylic acid (TCA) metabolites. Mechanistically, STAT1 broadly represses gene programs involved in fatty acid metabolism and oxidative phosphorylation. As a result, STAT1 depletion in WAT reduces subcutaneous adipocyte size in obese mice. However, while adipocyte-specific STAT1 expression controls WAT inflammation, it is dispensable for systemic glucose control. Rather, complete deletion of IFN-γ–STAT1 signaling using IFNGR1^−/− mice couples reduced inflammation with improved insulin sensitivity. These findings reveal how local cytokine activity controls obesity-associated inflammatory signals in adipose tissues.

All animal procedures have been approved by the Baylor College of Medicine Institutional Animal Care and Use Committee. All experiments were conducted using littermate-controlled male mice aged 8–10 weeks. All experimental animals were maintained on a C57BL/6 background and housed in a barrier-specific pathogen-free animal facility with 12-h dark/light cycle at room temperature and free access to water and food. Stat1^fl/fl mice (19) were provided by Lothar Hennighausen (National Institute of Diabetes and Digestive and Kidney Diseases). Stat1^fl/fl mice were crossed with AdipoQ-Cre (#028020; The Jackson Laboratory) to generate adipocyte-specific Stat1-knockout (STAT1^a-KO) and littermate controls (STAT1^fl/fl). IFNGR1^−/− mice were purchased from The Jackson Laboratory (#003288) and bred in-house. C57BL/6J wild-type mice were obtained from the Baylor College of Medicine Center for Comparative Medicine. Mice were fed a 60% high-fat diet (HFD; Research Diets) for 12 or 18 weeks before experiments. Mouse body composition was examined by MRI (Echo Medical Systems, LLC).

Subcutaneous WAT biopsies were obtained from 15 obese subjects during gastric bypass surgery (13). Nine subjects were defined as normal fasting plasma glucose (fasting plasma glucose, 93.3 ± 3.3 mg/dL; BMI, 40.3 ± 3.8; and HOMA of insulin resistance, 5.2 ± 3.0), while six patients were defined as having prediabetes (fasting plasma glucose, 106.3 ± 5.4 mg/dL; BMI, 39.5 ± 2.9; HOMA of insulin resistance, 4.9 ± 2.4). Additional subcutaneous WAT biopsies were obtained from 14 obese subjects (20). Nine subjects were defined as normal fasting plasma glucose (fasting plasma glucose, 86.7 ± 10.4 mg/dL), while five patients were insulin resistant (fasting plasma glucose, 110.8 ± 4.1 mg/dL). Based on American Diabetes Association guidelines, normal fasting plasma glucose was defined as <100 mg/dL and prediabetes as fasting glucose of 100–125 mg/dL. One subject was excluded with a fasting plasma glucose of 152 mg/dL, identified as having diabetes. Samples were stored at −80°C until RNA extraction. Human studies were approved by the Ethics Committee at Karolinska Institutet (Dnr 2008/2:3; Stockholm, Sweden) and the New Mexico VA Health Care System (Institutional Review Board: 11.139; Albuquerque, NM). In total, we analyzed gene expression in subcutaneous WAT collected from 17 males and 12 females.

To determine glucose tolerance, mice were fasted 16 h, and glucose was administered (1.5 g/kg body weight) by intraperitoneal injection. To determine insulin tolerance, mice were fasted 4 h prior to intraperitoneal insulin injection (1.5 units/kg body weight). Blood glucose levels were measured by handheld glucometer. Fasting serum levels were quantified by ELISA for insulin (Millipore) and leptin (Crystal Chem).

Western blotting was performed as previously described (21). Antibodies are listed in Supplementary Table 1.

Total RNA was extracted using the Direct-zol RNA Miniprep kit (Zymo Research). Primer sequences and gene expression assays are detailed in Supplementary Table 2.

Sample quality control, mRNA library preparation, and RNA sequencing (RNA-Seq) were performed by the University of Houston Sequencing and Editing Core. mRNA libraries were prepared with the Ovation RNA-Seq System V2 (NuGEN) and Ovation Ultralow Library System V2 (NuGEN) using input RNA. Size selection for libraries was performed using SPRIselect beads (Beckman Coulter), and purities of the libraries were analyzed using the High Sensitivity DNA chip on the Bioanalyzer 2100 (Agilent Technologies). The prepared libraries were pooled and sequenced using an Illumina NextSeq 500, generating ∼20 million 76-base-pair paired-end reads per sample. Reads were mapped to the UCSC mouse reference genome mm10 using HISAT2. Stringtie was used to calculate the expression level as reads per kilobase per million. Hierarchical cluster analysis was performed by Euclidean distance using log-twofold change over STAT1^fl/fl controls of significantly altered (P < 0.05) genes. Gene set enrichment analysis was performed with the Molecular Signatures Database, and normalized enrichment scores were calculated for the Hallmark gene sets.

Inguinal WAT (iWAT) metabolites were extracted by water/methanol methods. Chloroform was used to extract aqueous and organic layers. Glycolytic and TCA intermediates were measured using negative ionization mode with an electrospray ionization voltage of −3,500 eV. Prior to mass spectrometry analysis, the dried extract was resuspended in water/methanol (1:1) containing formic acid and then analyzed. For all samples, 10 μL of sample was injected and analyzed using a 6495 triple-quadrupole mass spectrometer (Agilent Technologies) coupled to a 1290 series HPLC system. The data were normalized with internal standards and log2-transformed on a per-sample basis. Differential metabolites were identified by adjusting the P values for multiple testing at a false discovery rate threshold of <0.25. Detailed methods are provided in the Supplementary Material.

Subcutaneous human preadipocytes (#SL0065; Zen-Bio Inc.) were differentiated, transfected, and Seahorse experiments performed as previously described (21). Mature adipocytes were transfected with STAT1 siRNA (Dharmacon) or siRNA control (Dharmacon). Recombinant human and mouse IFN-γ were purchased from R&D Systems. FLAG-STAT1 (#71454; Addgene) or vector control lentiviral particles were introduced into human adipocytes for 48 h before experiments.

For CRISPR-Cas9 gene deletion experiments, single guide RNAs (gRNAs) targeting sequences in exon 7 (g1), exon 6 (g2), and exon 17 (g3) of Stat1 were designed using the Broad Institute GPP Web Portal. A nonmammalian targeting gRNA sequence with similar GC content was used as a control. Guide sequences (Supplementary Table 3) were cloned into the lentiCRISPR v2 plasmid (#52961; Addgene), and lentiviral particles were generated in 293T cells (ATCC) using packaging plasmids pMD2.G (#12259; Addgene) and psPAX2 (#12260; Addgene) for pooled transfection with iMFectin (GenDEPOT). 3T3-L1 cells were selected for vector incorporation using puromycin (Gibco).

SVFs were isolated from mouse iWAT. Fat depots were digested in PBS containing collagenase D (1.5 units/mL) (Roche) and dispase II (2.4 units/mL) (Sigma-Aldrich) supplemented with 10 mmol/L CaCl₂ at 37°C for 45 min. The primary cells were filtered twice through 70-μm cell strainers and centrifuged at 700 relative centrifugal force to collect the SVF. The SVF cell pellets were rinsed and plated. Adipocyte differentiation was induced by treating confluent cells in DMEM/F12 medium containing GlutaMAX (Thermo Fisher Scientific), 10% FBS, 0.250 mmol/L isobutylmethylxanthine (Sigma-Aldrich), 1 μmol/L rosiglitazone (Cayman Chemical Company), 1 μmol/L dexamethasone (Sigma-Aldrich), 850 nmol/L insulin (Sigma-Aldrich), and 1 nmol/L T3 (Sigma-Aldrich). Four days after induction, cells were switched to maintenance medium containing 10% FBS, 1 μmol/L rosiglitazone, 1 μmol/L dexamethasone, 850 nmol/L insulin, and 1 nmol/L T3. Experiments that tested IFN-γ effects on SVF-derived adipocytes occurred 8–10 days after induction of differentiation.

We used GraphPad Prism 8.4.2 (GraphPad Software) for graphs and statistical analysis. For comparisons between two independent groups, unpaired two-tailed t tests or Mann-Whitney U tests were used. One-way ANOVA followed by Tukey post hoc tests were used to compare more than two independent groups. Differences between genetic interventions and ligand treatments were determined by two-way ANOVA followed by Tukey post hoc tests. All data are presented as mean ± SD. Our primary threshold for statistical significance was P < 0.05.

All data generated or analyzed during this study are included in the published article (and the Supplementary Material). RNA-Seq data sets are deposited in the Gene Expression Omnibus under accession number GSE153337. Resources (STAT1^a-KO mice and gSTAT1 knockout cells) generated during the current study are available from the corresponding author upon reasonable request.

STAT1 regulates genes that drive proinflammatory responses (11,12) and impair mitochondrial function (13,22). To test the hypothesis that STAT1 expression in WAT inversely correlates with insulin sensitivity, we measured STAT1 levels in fat tissues collected from obese mice and humans. Among the primary WAT depots, Stat1 expression was exclusively higher in subcutaneous (inguinal) WAT isolated from obese mice compared with mice fed normal chow (Fig. 1A). Similarly, in obese humans, STAT1 was almost threefold higher (Mann-Whitney U, P = 0.092) in subcutaneous WAT isolated from subjects with prediabetes (fasting plasma glucose 100–125 mg/dL) compared with normoglycemic (fasting plasma glucose <100 mg/dL) counterparts (Fig. 1B). Likewise, STAT1 levels correlated with fasting plasma glucose (Pearson correlation coefficient = 0.322; P = 0.088). These results indicate increased STAT1 expression in subcutaneous WAT is associated with insulin resistance.

Adipocyte metabolism and insulin sensitivity correlate with expression of mitochondrial electron transport chain (ETC) components (23). Accordingly, enforced FLAG-STAT1 expression in primary human adipocytes repressed mitochondrial ETC proteins (COXIV and NDUFA12), suggesting impaired respiratory capacity (Fig. 1C). We next measured oxygen consumption rate (OCR) to establish the functional impact of these changes on metabolic activity. We found STAT1 overexpression suppressed maximal respiration in human adipocytes compared with control conditions (Fig. 1D). Together, these results demonstrate higher STAT1 expression impairs metabolic functions of adipocytes.

To further define the metabolic impact of STAT1 on adipocyte function, we leveraged gene editing and siRNA approaches to deplete STAT1 in primary human and 3T3-L1 mouse adipocytes. First, we transfected mature human adipocytes with scrambled control (scRNA) or STAT1 siRNA followed by treatment with vehicle or IFN-γ for 24 h (Fig. 2A). STAT1 knockdown blocked IFN-γ–mediated induction of STAT1 itself, as well as the STAT1 target gene IRF1. Remarkably, STAT1 depletion increased basal expression of genes that reflect mature adipocyte metabolic functions (ADIPOQ and UCP1) and mediated resistance to the effects of IFN-γ treatment. Likewise, IFN-γ treatment decreased maximal OCR in control cells (green lines), but STAT1 knockdown conferred resistance to this response (Fig. 2B, red lines).

To examine the impact of STAT1 signaling during adipocyte differentiation, we leveraged CRISPR-Cas9 to knock out Stat1 in 3T3-L1 cells. We screened three single gRNAs in 3T3-L1 cells targeting different exons of Stat1 and determined Stat1 guide #3 (g3) targeting exon 17 imposed complete depletion of Stat1 protein expression (Fig. 2C). Stat1 knockout (gSTAT1) enhanced 3T3-L1 adipocyte maturation as indicated by elevated expression of proteins classically associated with adipocyte differentiation (PPARγ, ADIPOQ, and ACLY). Similarly, COXIV, a subunit of the terminal enzyme of the mitochondrial ETC, was increased in Stat1 knockout cells, suggesting enhanced respiratory capacity (Fig. 2C). Stat1 knockout also increased the expression of PPARγ target genes, including Pparγ2, AdipoQ, and Pgc1a (Fig. 2D). Furthermore, gSTAT1 adipocytes contained more lipid droplets and densely packed mitochondria compared with differentiation controls (Fig. 2E). To test how Stat1 knockout affects the response to inflammatory stimulus, we treated control (ntCR1) and gSTAT1 adipocytes with IFN-γ for 24 h (Fig. 2F). IFN-γ treatment stimulated expression of Stat1 and Irf1 and suppressed the adipocyte marker genes Pparγ2, AdipoQ, and Acly in control cells. Stat2 and Stat3, but not other Stat family members, also showed robust sensitivity to IFN-γ in ntCR1 adipocytes (Supplementary Fig. 1). In contrast, Stat1 deletion increased expression of AdipoQ, Pgc1a, and the lipogenic enzyme Aspa and reversed IFN-γ effects, including repression of Pparγ2 and Acly. Collectively, these experiments performed in human and mouse adipocytes demonstrate STAT1 inhibition elevates mitochondrial function, accelerates adipocyte differentiation, and blunts responses to the obesity-related cytokine IFN-γ.

STAT1 regulates inflammatory responses in multiple tissues and cell types (24–26). However, the immune cell functions of STAT1 in obesity may mask dominant roles in WAT. Therefore, to examine the adipocyte-specific impact of STAT1 signaling on obesity-induced inflammation and insulin resistance, we crossed STAT1^fl/fl mice (19) with mice expressing Cre recombinase under control of the adiponectin promoter to generate Stat1 adipocyte-specific knockout mice (STAT1^a-KO). Immunoblotting confirmed loss of Stat1 expression in WAT depots from STAT1^a-KO mice (Supplementary Fig. 2A). Liver Stat1 levels and WAT expression of Stat2 and Stat3 were unchanged in STAT1^a-KO mice, validating specific deletion of Stat1 only in tissues that express adiponectin (Supplementary Fig. 2B). mRNA analysis confirmed adipocyte-specific knockout reduced Stat1 levels by 70% in primary WAT depots (Supplementary Fig. 2C). The mRNA expression of other Stat family members remained mostly unaffected by Stat1 conditional deletion in WAT (Supplementary Fig. 3). To model obesity and the metabolic stress resulting from excessive caloric intake, STAT1^a-KO mice and littermate controls were maintained on HFD for 18 weeks. Contrary to our expectations, weight gain (Fig. 3A), body composition (Fig. 3B), and tissue weights (Fig. 3C) were similar between groups after diet-induced obesity. Obese STAT1^fl/fl and STAT1^a-KO also displayed indistinguishable energy balance profiles and brown adipose tissue features (Supplementary Fig. 4) at conventional housing conditions. We observed nominal improvements in glucose (Fig. 3D) and insulin (Fig. 3E) tolerance tests, as well as fasted serum insulin levels (Fig. 3F) in STAT1^a-KO mice compared with controls.

Diet-induced obesity increases macrophage infiltration of epididymal WAT (eWAT) and iWAT adipose depots, which correlates with impaired WAT expansion (10). Based on our in vitro studies, adipocyte-specific deletion of Stat1 may decrease the production of proinflammatory signaling molecules in WAT, thereby restoring metabolic fitness in adipocytes. To examine this hypothesis, we collected iWAT and eWAT depots from STAT1^a-KO mice and littermate controls for immunohistochemistry and molecular analysis. Grossly, Stat1 deletion in the adipocytes of obese mice decreased proinflammatory macrophage infiltration into eWAT and iWAT (Fig. 3G) while reducing mRNA expression of several inflammatory Stat1 target genes in STAT1^a-KO WAT compared with controls (Fig. 3H). Last, we performed quantitative image-based histological analysis to measure adipocyte size in STAT1^a-KO WAT compared with controls. We observed adipocytes from STAT1^a-KO iWAT were significantly smaller (Fig. 3I) and more abundant (Fig. 3J), favoring reduced adipocyte hypertrophy (27). Together, these results suggest STAT1 disruption in adipocytes reduces inflammation in WAT, which facilitates healthy WAT remodeling and may in turn improve adipocyte glucose and lipid metabolism.

Our mouse and in vitro models suggest STAT1-deficient fat cells engage pathways that merge metabolic and differentiation genes to resist the inflammatory effects of persistent obesity. Therefore, we used RNA-Seq to identify the biologically cohesive gene programs of Stat1 depletion in the WAT of obese STAT1^fl/fl and STAT1^a-KO mice. These efforts uncovered clear signatures that explain the impacts of Stat1 knockout in the eWAT and iWAT (Fig. 4A). In STAT1^a-KO mice, iWAT and eWAT shared 22 suppressed genes, 11 of which were common IFN-γ–STAT1 targets (e.g., Stat1, Stat2, and Isg15). Consistent with its known roles in inflammation, gene set enrichment analysis (GSEA) indicated Stat1 deletion in WAT exerted broad anti-inflammatory effects, including evidence for suppression of IFN responses (Fig. 4B). Expression of STAT1 targets within these gene sets (Stat1, Irf1, Isg15, and Oas1) was reduced in STAT1^a-KO iWAT (Fig. 4C). Consistent with the idea that reduced inflammation may improve adipocyte function, GSEA also revealed STAT1^a-KO increased levels of genes found in central metabolic pathways, including oxidative phosphorylation, adipogenesis, and fatty acid metabolism. Notably, distinct targets of established metabolic transcription factors (Supplementary Table 4) were selectively enhanced in the iWAT of STAT1^a-KO compared with STAT1^fl/fl controls (Fig. 4D). Together, these findings explain how Stat1 deletion imparts gene changes that enhance substrate metabolism in adipocytes.

To further understand the metabolic impact of Stat1 deletion in obesity, we assessed the relative steady-state levels of metabolic intermediates in iWAT collected from obese STAT1^fl/fl and STAT1^a-KO mice (Fig. 4E). We observed depletion of several long-chain fatty acid carnitine species (myristoyl carnitine, octenyl carnitine, and lauroyl carnitine) from iWAT of STAT1^a-KO mice that reflect higher rates of β oxidation (28). Moreover, Stat1 loss caused accumulation of pyruvate and key metabolic intermediates associated with the TCA cycle, including the end-stage metabolites fumarate and oxaloacetate. These data suggest Stat1 loss increases fatty acid breakdown and flux through the TCA cycle. By contrast, TCA cycle intermediates are frequently depleted in the iWAT of mice fed HFD relative to normal chow controls (Fig. 4F and G, red). Thus, gene expression and metabolite profiles establish Stat1 deletion in the iWAT powers an integrated program that restores TCA cycle flux (Fig. 4G, green stars) to enhance adipocyte respiration and metabolic fitness in the face of nutrient stress.

The IFNGR1 protein imparts the IFN-γ signal to the transcription of unique proinflammatory genes. An important question is whether signals upstream of Stat1 oppose critical responses that couple WAT expandability to insulin sensitivity. To address this question, we placed IFNGR1^−/− and IFNGR1^+/+ mice on HFD for 12 weeks followed by comprehensive metabolic phenotyping and mechanistic studies. Body weight (Fig. 5A) and composition (Fig. 5B) studies demonstrated IFNGR1^−/− mice accumulated less iWAT (Fig. 5C) and resisted diet-induced obesity. Accordingly, Ifngr1 deletion improved insulin sensitivity (Fig. 5D) coupled with reduced fasting insulin levels (Fig. 5E). Fasting leptin levels reflected reduced fat mass in IFNGR1^−/− mice (Fig. 5F). Histological sections of iWAT (Fig. 5G) indicated IFNGR1^−/− impacted adipocyte hypertrophy reflected by reduced cell size (Fig. 5H) and a trend toward more fat cells (Fig. 5I). Consistent with the experiments in obese STAT1^a-KO mice, we observed suppressed IFN-γ–STAT1 proinflammatory target genes coupled with enhanced expression of lipid metabolism and mitochondrial genes in iWAT, suggesting enhanced metabolic function (Fig. 5J). In addition to Stat1 depletion, Ifngr1 deletion also reduced the gene expression of other Stat family members, including Stat2 and Stat3, in the iWAT (Supplementary Fig. 5A).

Disruption of IFN-γ activity improves mitochondrial function in adipocytes (13,22). Therefore, we performed metabolomics to examine the impact of complete ablation of IFN-γ signaling on the steady-state levels of glycolytic and TCA cycle metabolites in iWAT. While obese STAT1^a-KO mice exhibited partial restoration of TCA metabolite pools (Fig. 4G), Ifngr1 deletion rescued most TCA cycle metabolites that were sensitive to HFD, including oxaloacetate, citrate, α-ketoglutarate, and fumarate (Fig. 5K). Next, to test whether Ifngr1 deletion enabled metabolic resistance to IFN-γ in a cell-autonomous manner, we treated differentiated SVF derived from the iWAT of either IFNGR1^−/− mice or control mice with IFN-γ. IFNGR1^−/− adipocytes did not respond to IFN-γ at the level of phosphorylated STAT1 (Fig. 5L), canonical Stat1 target genes (Fig. 5M), or IFN-γ stimulation of Stat2 and Stat3 transcription (Supplementary Fig. 5B). Accordingly, the ability of IFN-γ to inhibit the expression of adipocyte marker (AdipoQ) and mitochondrial metabolism (Ucp1) genes was lost in cells lacking Ifngr1 (Fig. 5M). Metabolic activity of wild-type adipocytes remained IFN-γ sensitive, as demonstrated by reductions in maximal OCR (Fig. 5N). Consistent with metabolic resistance, IFN-γ did not affect maximal OCR in IFNGR1^−/− adipocytes. In summary, ablation of IFN-γ signaling reduces weight gain while preserving insulin sensitivity and physiologic adipocyte expansion. At the molecular level, ablation of IFN-γ signaling reduces WAT inflammation, increases levels of glycolytic and TCA metabolites, and improves respiratory capacity.

Our work sheds light on the enigmatic role of inflammation in obesity. Numerous studies implicate cytokines and soluble mediators in the response of WAT to overnutrition and the metabolic dysfunction that characterizes obesity in rodents and humans. In this study, we demonstrate depletion of STAT1 in adipocytes reduces WAT inflammation while enabling metabolic adaptations that promote healthy adipocyte remodeling. In general, obesity raises inflammation in rodent and human visceral WAT to a greater extent than subcutaneous WAT (10). Stat1 levels were elevated in the mouse subcutaneous WAT, but not eWAT, suggesting tissue-specific influences of inflammation and precursor cells (29) on adipose tissue responses to overnutrition. Nonetheless, our observations linking STAT1 expression in subcutaneous WAT to elevated plasma glucose in humans suggests IFN-γ signaling influences insulin sensitivity. However, adipocyte-specific Stat1 knockout exerts a nominal influence on insulin sensitivity in obese mice. Rather, we found complete disruption of IFN-γ signaling exerts antidiabetic effects and enables adipocytes to retain metabolic function in obesity.

Obesity and insulin resistance increase IFN-γ production (30) and IFN-γ–mediated activation of the transcription factor STAT1 blocks adipocyte differentiation (13,18,22). We extend these findings with genetic and RNA interference studies in vitro that demonstrate STAT1 depletion increases human and mouse adipocyte differentiation. To our surprise, Stat1 deletion in WAT marginally improved the metabolic criteria associated with obesity. STAT3 (31) and STAT5 (32) promote adipocyte differentiation in vitro, but we did not detect evidence of compensatory impacts on gene expression in Stat1-deficient cells. These observations suggest STAT1 governs unique metabolic signals in WAT. Although STAT1 mediates many IFN-γ–dependent actions, IFN-γ requires its cognate receptor and the kinase Jak1 to increase transcription of genes in STAT1^−/− cells (33,34), possibly through nuclear factor-κB (35). Nuclear factor-κB (36) regulates inflammation in WAT and other peripheral tissues, so it seems plausible that activation of this transcription complex in the fat cells of obese STAT1^a-KO mice does not allow coupling of subcutaneous adipocyte remodeling with broader metabolic effects. Indeed, complete elimination of IFN-γ activity at the level of its receptor improves insulin sensitivity in obese mice. Although whole-body Ifngr1 deletion exposes impacts on immunity across multiple endocrine tissues (15,30,37), IFNGR1^−/− adipocytes resist the detrimental effects of IFN-γ on metabolism.

Excess calorie intake evokes WAT expansion through both increased adipocyte size (hypertrophy) and number (hyperplasia) in depot-specific patterns. Increased hypertrophy is a hallmark of WAT enlargement in obesity and associated with metabolic alterations, proinflammatory response, and increased risk of developing T2DM independent of total fat mass (38–40). Experimental observations argue larger, hypertrophic fat cells behave differently than smaller adipocytes, namely in responses to lipolytic stimuli, secretory functions, and the anabolic effects of insulin (41). In male mice, the microenvironment dictates how HFD induces subcutaneous WAT hypertrophy (27). Although genetic depletion of Ifngr1 or Stat1 in vitro suggests some contributions of hyperplasia, our data indicate ablation of the IFN-γ signal gives rise to smaller adipocytes in iWAT that derive from reduced hypertrophy. Either way, these studies establish ways inflammatory signals must be separated to generate small, metabolically beneficial subcutaneous adipocytes that allow WAT expandability in healthy obesity.

The metabolic and anti-inflammatory effects of Stat1 and Ifngr1 knockout resemble activities of PPARγ agonists, including UCP1 expression and higher mitochondrial respiration in smaller, brown-like (“beige”) adipocytes (42). Not surprisingly, inflammatory transcription factors inhibit beige adipocyte formation and function (16,43). Despite no evidence of multilocular, beige adipocytes in histological sections, our in vitro studies suggest some metabolic benefits of IFN-γ–STAT1 blockade in subcutaneous WAT may partly derive from the recruitment of UCP1-positive cells that do not ultimately impact weight gain. Future studies will use thermoneutral or cold-challenge conditions to assess whether IFN-γ–STAT1 signals in brown and beige fat cells contribute to regulation of energy expenditure.

Inflammation and decreased mitochondrial oxidative capacity frequently copresent in obesity. IFNs broadly inhibit expression of several mitochondrial genes encoded within the mitochondrion (16,17,22), but the mechanisms by which IFN activation reduces mitochondrial function remain unclear. One possibility is that IFN-γ–activated STAT1 represses the transcriptional activity of integral regulators of lipid and glucose metabolism, as occurs with PPARγ, in obesity. IFN-γ causes STAT1 binding near PPARγ sites in multiple enhancer regions of mitochondrial and insulin sensitivity genes corresponding with reduced mRNA expression (13). These findings suggest a negative cross talk between the occupancy of PPARγ or new factors identified in our study (44,45) and STAT1 binding sites near genes important for WAT expansion and whole-body insulin sensitivity.

Hypertrophic obesity reflects persistent mitochondrial dysfunction and diminished lipid storage (41). Our study and others (46,47) highlight how obesity slows TCA cycle function and likely impacts anabolic functions of WAT. The TCA cycle provides intermediates and energy that drive lipid synthesis in adipose tissue. To this end, failure of integral TCA and lipid metabolism reactions limits WAT expandability in the face of obesity. IFNs lower cellular capacity to generate ATP (16,17,22) mirrored by decreased mitochondrial ETC activity. As a result, adipocytes cannot devote necessary ATP to biosynthesis of lipid and mitochondrial building blocks required for adipocyte expansion in response to chronic nutrient excess. We discovered elimination of IFN-γ action in adipocytes repletes critical TCA intermediates. For example, accumulation of α-ketoglutarate and fumarate allows generation of reducing equivalents to transfer electrons to the mitochondrial respiratory chain for ATP production. The elevation of citrate supplies carbon backbones to re-establish cytosolic acetyl-CoA and oxaloacetate pools to promote lipid and nucleotide synthesis. The consequence of these events is evidenced functionally by enrichment of oxidative phosphorylation genes, superior OCRs in vitro, and resistance to inflammation in obesity. Along these lines, it will now be important to leverage modern mass spectrometry methods to determine whether IFN-γ or other obesity-related cytokines impact metabolic flux of carbohydrate and lipid substrates in the adipose tissues of living animals.

Lastly, these findings describe a mechanism that allows insulin resistance to occur independent of WAT inflammation. Proinflammatory cytokines inhibit insulin signaling and mitochondrial function in adipocytes. However, broad anti-inflammatory strategies that impact STAT1 or STAT3 activity lack clinical efficacy in the treatment of obesity (48,49). Furthermore, recent provocative studies argue WAT inflammation responds to insulin resistance (50). Our studies and others (50) argue the causative relationship between chronic WAT inflammation and insulin resistance remains oversimplified. Although our findings do not rule out the idea that inflammation degrades metabolic fitness and insulin sensitivity, they bring into question whether anti-inflammatory monotherapies will be an effective strategy to improve all of the clinical phenotypes of obesity and T2DM.

Funding. This work was funded by American Diabetes Association grant 1-18-IBS-105 (to S.M.H.) and National Institute of Diabetes and Digestive and Kidney Diseases grants R01DK114356 (to S.M.H.) and R01DK121348 (to H.W.). This study was also funded (in part) by an award from the Baylor College of Medicine Nutrition and Obesity Pilot and Feasibility Fund. The Cellular and Molecular Morphology Core receives support from the Texas Digestive Diseases Center (National Institute of Diabetes and Digestive and Kidney Diseases grant P30DK056338). The Metabolomics Core was supported by the Cancer Prevention and Research Institute of Texas Core Facility Support Award RP170005, “Proteomic and Metabolomic Core Facility,” National Cancer Institute Cancer Center Support grant P30CA125123, and intramural funds from the Dan L. Duncan Comprehensive Cancer Center. This study was also supported, in part, by the Assistant Secretary of Defense for Health Affairs endorsed by the U.S. Department of Defense Peer Reviewed Medical Research Program Discovery Award (W81XWH‐18‐1‐0126 to K.H.K.) and U.S. Department of Veterans Affairs Clinical Sciences Research and Development Merit Review I01 CX00042403 (to R.A.-V.) and Eunice Kennedy Shriver National Institute of Child Health and Human Development grant R01 HD093047 (to R.A.-V.).

This content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health and the Department of Veterans Affairs or the U.S. Government.

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. A.R.C. and S.M.H. conceptualized the study. A.R.C., N.C., D.A.B., P.M.M., and S.M.H. designed experiments. A.R.C., D.A.B., and S.M.H. wrote the manuscript with editorial input from all authors. A.R.C. and S.M.H. performed all experiments with assistance as noted: P.K.S. and K.H.K. assisted with mouse phenotyping, V.P. and N.P. assisted with metabolomics analysis, N.C., P.M.M., J.B.F., and R.S. performed quantitative PCR and gene expression validation, K.R. and C.C. assisted with RNA-Seq data analysis and metabolomics data integration, Z.L. and H.W. guided immune cell assays, and D.T.V. and R.A.-V. provided clinical specimens. S.M.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.