IL-21 Is a Major Negative Regulator of IRF4-Dependent Lipolysis Affecting Tregs in Adipose Tissue and Systemic Insulin Sensitivity

Obesity-associated tissue inflammation is now recognized as a major cause of decreased insulin sensitivity (1,2). Obesity, insulin resistance, and type 2 diabetes are closely associated with chronic inflammation characterized by abnormal cytokine production, increased acute-phase reactants and other mediators, and activation of a network of inflammatory signaling pathways (3,4). Excessive triglyceride accumulation within adipocytes leads to adipocyte hypertrophy and a dysregulation of adipokine secretory patterns. Adipocytes as well as cells of the stromal vascular fraction (SVF), including preadipocytes, fibroblasts, mesenchymal stem cells, and immune cells, contribute to the production of proinflammatory cytokines in obesity (3–5), with a pivotal role played by macrophages and T lymphocytes (6–8). In lean adipose tissue, T-helper (Th) type 2 cells produce anti-inflammatory cytokines such as interleukin (IL)-4, -10, and -13, which promote alternative activated M2 macrophage polarization (9). M2 polarization is also induced by regulatory T cells (Tregs) and eosinophils via IL-4. Conversely, in obese adipose tissue, investigators have observed an increase in the number of Th1 cytokines, M1 polarized macrophages, mast cells, B cells, and CD8⁺ T cells, which contribute to insulin resistance and promote macrophage M1 accumulation and proinflammatory gene expression (9–13). Factors orchestrating the switch between M1 and M2 are still undefined. Loss of interferon regulatory factor 4 (IRF4) specifically in the myeloid cells evoked a constitutive M1 polarization in the adipose tissue, suggesting that IRF4 is a negative regulator of inflammation in diet-induced obesity, in part through regulation of macrophage polarization (14). Interestingly, IRF4 expression is nutritionally regulated by the actions of insulin and FoxO1, playing a significant role in the transcriptional regulation of lipid handling in adipocytes, promoting lipolysis, at least in part by inducing the expression of the lipases adipose triglyceride lipase (ATGL; Pnpla2) and hormone-sensitive lipase (HSL; Lipe) (15).

Naturally occurring Treg cells are a unique subpopulation of CD4⁺ T cells specifically adapted to the suppression of aberrant or excessive immune responses that are harmful to the host (16). Treg cells are abundant in visceral adipose tissue (VAT) and have a different T-cell receptor repertoire compared with Treg cells in other tissues, suggesting that they might be activated via the recognition of a fat tissue–specific antigen (13). A recent study reveals an important role for VAT-specific natural Treg cells in the suppression of obesity-associated inflammation in adipose tissue and consequently in reducing insulin resistance (17). The number of VAT Treg cells is strikingly and specifically reduced in insulin-resistant models of obesity, and the cells are characterized by the expression of the transcription factor Foxp3 and the nuclear receptor peroxisome proliferator–activated receptor (PPAR)-γ (17).

IL-21 is a member of the type-I cytokine family and is synthesized by a range of CD4⁺ Th cells, including Th1 and Th17 cells, activated NKT cells, and T follicular helper cells (18–20). IL-21 biological functions are mediated via the IL-21 receptor (IL-21R) and after activation of the Janus kinase (JAK) family protein tyrosine kinases JAK1 and JAK3 and, subsequently, the activation of Stat1, Stat3 and to a lesser degree Stat4, Stat5, and Stat6 (21–23). IL-21 expression in T cells can be regulated by IL-21 via an autocrine positive-feedback loop, involving the activation of STAT3 (24). This feedback loop is essential for the development of Th17 cells (25,26). IL-21–mediated T-cell activation relies partly on its ability to inhibit the differentiation of inducible Tregs and to make T cells resistant to the Treg-mediated immunosuppression (27,28).

Because IL-21 is known to exert negative effects on Treg activity, we hypothesized that it could play a role in obesity-induced insulin resistance.

Wild-type (WT) and IL-21 knockout (KO) (129S5-Il21^tm1Lex) male mice, both on the same genetic background (C57BL/6J), were purchased from Lexicon Genetics, Inc. IL-21 KO mice are viable and do not exhibit any phenotype. Mice were maintained in standard animal cages under specific pathogen–free conditions in the animal facility at the University of Rome “Tor Vergata.” Mice were maintained under a strict 12-h light cycle (lights on at 7:00 a.m. and off at 7:00 p.m.), genotyped, and divided in separate cages at the beginning of each experiment. For the diet-induced obesity model, individually caged mice from all groups were fed a high-fat diet (HFD) (60% of calories from fat; Research Diets, New Brunswick, NJ) or normal diet (ND) (10% calories from fat, GLP; Mucedola S.r.l., Settimo Milanese, Italy) for 18 weeks after weaning as indicated. Metabolic testing procedures were performed as previously described (29,30).

Hormone and metabolite levels were measured using commercial kits: insulin (Mercodia), nonesterified fatty acid (NEFA) (Wako), glycerol (Sigma-Aldrich, St. Louis, MO), and glucagon (Uscn Life Science, Inc).

After 12 weeks of HFD, we evaluated peripheral insulin sensitivity by the euglycemic-hyperinsulinemic clamp technique. Surgery for the positioning of catheters was performed 3–5 days prior to the insulin clamp procedure as previously described (31,32), and then mice were housed in individual cages. The euglycemic-hyperinsulinemic clamp was performed in the awake state after a 6-h fast. At time zero, a primed continuous (18.0 mU · kg⁻¹ · min⁻¹, Actrapid 100 IU/mL; Novo Nordisk, Copenhagen, Denmark) infusion of human insulin was started simultaneously with a variable infusion of 20% dextrose in order to maintain the plasma glucose concentration constant at its basal level (80–100 mg/dL). Fasting plasma glucose was measured at time 0. Subsequently, blood samples (∼2 μL) were taken from the tail vein at 10-min intervals for at least 2 h to measure glucose concentration and adjust dextrose infusion rates. Insulin sensitivity (rate of peripheral glucose uptake [mg·kg⁻¹·min⁻¹]) was calculated from average glucose concentrations and dextrose infusion rates during the last 30 min of the steady-state clamp period.

Epigonadal fat and liver were obtained from WT and IL-21 KO mice; specimens were fixed in 10% paraformaldehyde and embedded in paraffin. Ten-micrometer consecutive sections were then mounted on slides and stained with hematoxylin-eosin. Adipose cell size and density were calculated as previously described (33).

VAT was subjected to collagenase digestion (1 mg/mL collagenase type 1; Sigma-Aldrich) in Krebs-Ringer buffer, with shaking at 180 rpm for 30 min at 37°C. After digestion, adipocytes were allowed to separate by flotation and the infranatant solution was centrifuged for 5 min at 300g to pellet the SVF. The adipocyte fraction was washed three times with the Krebs-Ringer buffer. Subsequently, RNA was isolated from adipocytes and the SVF fractions and analyzed by real-time PCR. The profile of adiponectin mRNA expression was used to test the purity of the isolated fractions. The SVF was analyzed by flow cytometry techniques.

3T3-L1 cells (American Type Culture Collection) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) (Invitrogen) with 10% bovine calf serum (Invitrogen) in 5% CO₂. Two days postconfluence, cells were exposed to DMEM 10% FBS (Invitrogen) with 1 μmol/L dexamethasone (Sigma), 5 μg/mL insulin (Sigma), and 0.5 mmol/L isobutylmethylxanthine (Sigma). After 2 days, cells were maintained in medium containing FBS only. For IRF4 regulation experiments, fully differentiated 3T3-L1 adipocytes were incubated in serum-free DMEM containing 1% fatty acid–free BSA (Sigma) with isoproterenol (10 μmol/L; Sigma) and IL-21 (100 ng/mL; R&D) at the doses and times indicated.

Preparation of tissue lysates, quantification, and immunoblot analysis were performed as previously described (33). Antibodies to IRF4, actin, and total FoxO1 (Santa Cruz Biotechnology), phospho-Ser⁴⁷³ Akt, total Akt, phospho-Ser²⁵⁶ FoxO1 (Cell Signaling Technology) were used.

Total RNA was isolated and gene expression analysis was performed as previously described (34).

Cells from the SVF of adipose tissue were stained for surface antigens CD4 and CD25 and for the intracellular Foxp3⁺ transcription factor by using a mouse regulatory T-cell detection kit as directed by the manufacturer’s instructions (Miltenyi Biotec, Bergisch Gladbach, Germany). For intracellular lipids, cells were stained with Nile red (1 μg/mL; Sigma-Aldrich). Samples were analyzed using a FACSCalibur flow cytometer (Becton Dickinson, Heidelberg, Germany) and FlowJo software.

A total of 207 adipose tissue samples (112 visceral and 95 subcutaneous) were collected at the Endocrinology Service of Hospital Universitari of Girona “Dr. Josep Trueta” from a group of Caucasian subjects with BMI between 20 and 58 kg/m². All subjects reviewed that their body weight had been stable for at least 3 months before the study and gave written informed consent after the purpose, nature, and potential risks of the study were explained to them.

Adipose tissue samples were obtained from subcutaneous and visceral depots during elective surgical procedures (cholecystectomy, surgery of abdominal hernia, and gastric bypass surgery), washed, fragmented, and immediately flash frozen in liquid nitrogen before being stored at −80°C and used for gene expression analysis.

Results of the experimental studies are expressed as means ± SD. Statistical analyses were performed using the unpaired Student t test as indicated. Values of P < 0.05 were considered statistically significant.

A recent study demonstrated that Treg cells with a unique phenotype were highly enriched in the abdominal fat of lean mice but were strikingly and specifically reduced at this site in insulin-resistant models of obesity (13). It has also been reported that IL-21 can counteract the immune-suppressive properties of Tregs in several types of tissues both in vitro and in vivo (27,35). In order to determine whether this IL-21 effect on Treg cells could be extended to those residing in the adipose tissue, we firstly studied the amount of Tregs, marked as CD4⁺CD25⁺Foxp3⁺, in the SVF of IL-21 KO mice, finding a significant increase compared with WT littermates (Fig. 1A). Next, in visceral (epigonadal) adipose tissue of WT mice fed an HFD for 16 weeks compared with WT fed an ND, we found a significant increase of both IL-21 and IL-21R mRNA expression (Fig. 1B). More in detail, we observed an augment of IL-21 and IL-21R expression both in adipocyte fraction and SVF, although this was not significant (Fig. 1C). We also found IL-21 and IL-21R gene expression in fully differentiated 3T3-L1 adipocytes (Fig. 1D). This suggested an association between increased IL-21/IL-21R signaling and the progression of obesity.

Next, in order to understand the effects of this cytokine on diet-induced obesity, we conducted a complete metabolic characterization of IL-21 KO mice under ND and HFD conditions. IL-21 KO mice fed ND did not show differences in body weight or fasting plasma glucose levels compared with WT mice (Fig. 2A). We fed 6- to 7-week-old WT and IL-21 KO mice in a context of HFD for 18 weeks. Fasting and fed body weight and fasting glycemia were comparable at the beginning of treatment, but their curves significantly diverged from week 5 to the end of our observation at week 18 (Fig. 2B). Intraperitoneal glucose tolerance test (IPGTT), intraperitoneal insulin tolerance test, and serum insulin levels suggested that metabolic control was improved, on an HFD, by IL-21 deficiency (Fig. 2C and D). The relief from diet-induced insulin resistance was finally confirmed through the measurement of peripheral (skeletal muscle) insulin sensitivity by the euglycemic-hyperinsulinemic clamp (Fig. 2E).

Afterward, we conducted a morphological and molecular characterization of adipose tissue in order to better understand the mechanism by which IL-21 deficiency protects from metabolic injury caused by diet-induced obesity. IL-21 KO mice fed an ND did not show significant differences in adipose tissue morphometry compared with WT mice littermates (Supplementary Fig. 1A). On the other hand, the IL-21 KO–reduced body weight, observed with HFD, was characterized by lower adiposity associated with decreased fat pad mass (Fig. 3A), reduced adipocyte size, and a higher density of smaller adipocytes (Fig. 3B).

Gene expression analysis of adipose tissue revealed significantly reduced levels of macrophage markers such as F4/80 and CD68 in IL-21 KO mice. This may indicate a low grade of infiltration of proinflammatory macrophages. On the other hand, we have found increased levels of YM1 and Mgl2 mRNA, suggesting a high presence of alternatively activated macrophages M2 (Fig. 3C). Next, we analyzed genes involved in the regulation of glucose/lipid metabolism and mitochondrial function, finding significantly increased levels of adiponectin, FoxO1, SOCS3, Sirt1, ERRα, and Nrf1 (Fig. 3C). The improved metabolic state of IL-21 KO mice was also supported by increased phosphorylation of Ser⁴⁷³ Akt in the refeeding condition (Fig. 3D).

In adipose tissue, fasting induces IRF4-dependent lypolisis, and insulin, during refeeding, inhibits its expression via AKT/FoxO1. In adipose tissue of IL-21 KO mice, despite higher Akt phosphorylation (Fig. 3D), we observed significantly increased expression of IRF4 in the refeeding state at both mRNA and protein levels (Fig. 4A). In the fasting state, we found increased expression of IRF4 only in adipose tissue from IL-21 KO mice fed an HFD compared with WT. Expression of IRF4 targets pnpla2 and lipe confirmed increased expression of both lipolytic genes, particularly in the refeeding state (Fig. 4A). To control that this nutritional effect was specific for adipose tissue, we measured IRF4 expression in spleen from the same mice, finding no differences (Fig. 4C). Analysis of NEFA and glycerol in fasting sera confirmed a trend to increased lipolysis in IL-21 KO compared with WT littermates during both ND and HFD (Fig. 4D).

IRF4 mRNA expression rose significantly in 3T3-L1 adipocytes treated with isoproterenol for 2 h and decreased when adipocytes were pretreated with IL-21. Accordingly, we found decreased mRNA levels of IRF4 targets when the treatment was extended to 4 h (Fig. 5A).

A recent study demonstrates that IRF4 promotes M2 polarization of adipose tissue macrophages (14). In SVFs from IL-21 KO adipose tissue, we found significant increased IRF4 and M2, markers of mRNA expression (Fig. 5B). M2 macrophages and Treg cells are known to prevalently use fatty acids for ATP generation to maintain their functions (36). Recently, PPAR-γ was highlighted as a crucial molecular orchestrator of VAT Tregs and M2 macrophage accumulation, phenotypes, and functions (17). We found increased levels of PPAR-γ mRNA in SVF of IL-21 KO HFD compared with WT. The profile of adiponectin mRNA expression provides evidence of the purity of the fraction preparations (Fig. 5B).

To determine whether the degree of infiltration of Tregs in the adipose tissue in our KO model could be influenced by a treatment inducing obesity treatment, we quantified by flow cytometry the number of Treg cells in SVFs of the two experimental groups at the end of 18 weeks of HFD. As well as for the animal KO fed an ND, the amount of Tregs present in the adipose tissue of HFD IL-21 KO mice was significantly higher than the equivalent WT animals (Fig. 6A). Comparative analysis of Tregs in SVF of IL-21 KO and WT fed an ND or HFD confirmed that the obesity condition reduced Tregs in SVF of WT mice. It is interesting to note that IL-21 KO animals fed HFD maintained a high number of Tregs comparable with that of WT fed ND (Fig. 6B). Of note, loss of IL-21 is associated with increased lipolysis and increased Tregs and M2 macrophage markers, suggesting that a state in which IL-21 is reduced is possibly associated with increased lipid uptake from anti-inflammatory cells such as Tregs and M2 macrophages. Interestingly, we found increased lipid content in Tregs from IL-21 KO compared with WT (Fig. 6C).

The overall improvement of glucose tolerance was associated with absence of liver steatosis in IL-21 KO compared with WT mice during HFD (Fig. 7A), with was associated with reduced inflammatory markers such as F4/80 and CD68 and an unexpected mild increase in gluconeogenic enzymes such as Pck1 and G6pc (Fig. 7B). Consistently, we found reduced Ser²⁵⁶ phosphorylation of FoxO1 in the fasting IL-21 KO liver (Fig. 7C). IL-21 KO mice during HFD revealed a marked tendency to lower fasting glucose compared with WT littermates, with no differences in glucagon levels (Supplementary Fig. 2A). This suggests that the slight increase in gluconeogenic enzymes is a reactive response to maintain glucose at physiological levels. The concept of reactive response is also supported by the intraperitoneal pyruvate tolerance test showing increased glucose levels in IL-21 KO mice fed HFD (Fig. 7D). Furthermore, we analyzed Pck1 and G6pc expression also in livers from HFD refed mice and at the end of euglycemic-hyperinsulinemic clamp (Supplementary Fig. 2B); overall, the data suggest that during fasting or intense glucose uptake from the muscle, the absence of IL-21 increases Pck1 expression, possibly to compensate for lower peripheral glucose level.

To explore the involvement of IL-21 effects on human adipose tissue inflammation, we analyzed its expression in adipose tissue biopsies from patients with different degrees of obesity (Supplementary Table 1). We found a significant negative correlation between IL-21R and the CD206-to-CD68 ratio expression; this ratio is known to be higher in subjects with less body fat and lower fasting glucose concentrations (37). Moreover, PPAR-γ was also found to be significantly and negatively correlated to IL-21R expression in subcutaneous adipose tissue of obese subjects (Table 1). On the other hand, we found a significant positive correlation between IL-21R and tumor necrosis factor-α both in visceral and in subcutaneous adipose tissue (Table 1), indicating an involvement of IL-21 signaling in the development or persistence of adipose tissue inflammation.

IRF4 expression is highly restricted to immune cells and adipose tissue and is more abundant in mature adipocytes (38). IRF4 is nutritionally regulated by the action of insulin and FoxO1 and plays a significant role in the transcriptional regulation of lipid handling in adipocytes, promoting lipolysis. Interestingly, we found a strong relationship between IRF4 and ADRP gene expression, a lipolytic gene, in human VAT (r = 0.47, P < 0.0001, data not shown). During fasting, in adipocytes, mRNA and protein levels of IRF4 rise dramatically with subsequent downregulation after refeeding. IRF4 promotes lipolysis at least in part by inducing the expression of the lipases ATGL and HSL (15). We measured, in adipose tissue of IL-21 KO mice fed an ND, high levels of expression of the transcription factor IRF4 and its targets lipe and pnpla2, particularly in the refeeding conditions. Interestingly IRF4 expression is repressed in whole VAT of three different rodent models of obesity, a hyperinsulinemic state (15). This may seem paradoxical given that obesity is associated with insulin resistance and mice lacking insulin receptors in fat display elevated Irf4 expression (15). We found large induction of IRF4 and its targets also in HFD IL-21 KO adipose tissue both in the fasting and in the refeeding state. Therefore, it remains possible that other factors, such as IL-21, dominate the control of Irf4 gene expression in the context of obesity. Indeed, our in vitro studies confirm a role for IL-21 in reducing IRF4 and its target levels during lipolysis. Despite the high levels of IRF4, we unexpectedly found only mildly elevated NEFA levels in fasting sera of IL-21 KO animals. The reason could be attributed to the abundance of M2 macrophages and Tregs residing in the adipose tissue of these animals, immune populations with a strong ability to capture and oxidize fatty acids released by adipocytes (17,36). IRF4 is a well-known player in a variety of immune activities, including Tregs function and the development of inflammatory Th17 cells, and is absolutely required for the autocrine production of IL-21 in Th17 cells (39). Recently, IL-21 has emerged as a key cytokine for the maintenance of the mucosal immune system homeostasis by modulating the balance between Tregs and proinflammatory Th17 cells (28). Interestingly, the frequencies of Tregs and Th17 cells often show an inverse relationship, as their differentiation processes are also counterbalanced (40). It is worthy of noting that the in vivo presence of Th17 T cells in adipose tissue under normal chow conditions or an HFD has not yet been extensively reported. Few recent observations demonstrate that diet-induced obesity predisposes to IL-6–dependent Th17 expansion in adipose tissue (41). Given that IL-21 induces and amplifies Th17 development independently of IL-6 (39), it is possible that IL-21 is secreted in specific phases of adipose tissue expansion, eventually exacerbating early disease progression.

Tregs with a unique phenotype were highly enriched in the abdominal fat of lean mice, but their numbers were strikingly and specifically reduced at this site in insulin-resistant models of obesity (11). Recent studies reveal an important role for VAT-specific natural Tregs in the suppression of obesity-associated inflammation in VAT and consequently in reducing insulin resistance. The number of VAT Tregs decreases with obesity, and a boost in the number of these cells in obese mice can improve insulin sensitivity (11,13). Tregs expressing Foxp3 can secrete anti-inflammatory signals such as IL-10 and transforming growth factor β, inhibit macrophage migration, and induce M2-like macrophage differentiation (11). IL-21–mediated T-cell activation relies partly on its ability to inhibit the differentiation of Tregs and to make T cells resistant to the Treg-mediated immunosuppression (27,28). In SVF of IL-21 KO adipose tissue, we found a significant increase in the number of resident Tregs in both mice fed an ND than in those receiving an HFD. This may suggest the possibility that IL-21 regulates Treg number and differentiation even in the adipose tissue. What causes the decrease in Treg fraction in abdominal adipose tissue during obesity is still undefined; nevertheless, recent reports indicate a possible role for the hyperleptinemic state characterizing obesity to modulate Treg number and activity (42,43). Thus, a hypothesis that needs further exploitation in appropriate models is that leptin and IL-21 share some biological function and cooperate in causing Treg dramatic reduction in adipose tissue during obesity development. Since the adipose tissue of IL-21 KO mice is characterized by a lower degree of macrophage infiltration and increased expression of M2 polarization antigens (YM1, Mgl2), it is intriguing to hypothesize that a IL-21/Tregs axis might regulate the balance between macrophage M2 polarization and M1 infiltration in the context of obese adipose tissue. A recent study shows that VAT-resident Tregs and M2 macrophages specifically express PPAR-γ, an important factor controlling their accumulation, phenotype, and function (17,36). Consistently, we measured significant increased levels of PPAR-γ and M2 markers in SVF of IL-21 KO mice. In this context, in humans we found a negative association between IL-21R and PPAR-γ gene expression in SAT, suggesting that IL21 signaling runs in parallel to PPAR-γ. A newly discovered property was that in VAT, but not in lymphoid tissue, Tregs can take up lipids—an ability not shared by conventional T cells residing at the same site (17). It is also known that M2 macrophages and Tregs prevalently use fatty acids for ATP generation to maintain their functions (36). In IL-21 KO mice, we observed a high induction of IRF4-related lipolysis and at the same time increased lipid uptake by Tregs. This highlights the possibility that IL-21 regulates Treg activity in adipose tissue.

In conclusion, we relate for the first time the IL-21/IL-21R dyad to IRF4-dependent regulation of lipolysis and reduction of Tregs in adipose tissue. We hypothesize that IL-21 is a crucial player in this context, since we found an increase in mRNA levels of IL-21 and IL-21R in adipose tissue of obese animals and obese human subjects compared with their lean controls. Our data suggest that preventing IL-21 signaling might counteract obesity and the consequent metabolic defects in an experimental model—a finding with potential therapeutic implications in human subjects with metabolic syndrome and type 2 diabetes.

Funding. This study was funded in part by Fondazione Roma 2008, European Foundation for the Study of Diabetes/Lilly 2012, AIRC 2012 Project IG 13163, FP7-Health-241913-FLORINASH, FP7-Health-EURHYTHDIA, and PRIN 2012 (all to M.Fe.). T.M. is the recipient of the Albert Reynolds Travel Fellowship from the European Association for the Study of Diabetes and Fellowship Prize from Società Italiana di Diabetologia. G.P.S. is the recipient of a fellowship from Laboratori Guidotti, Pisa, Italy. A.G. has received support by grants from Università Cattolica del Sacro Cuore (Fondi Ateneo Linea D.3.2 Sindrome Metabolica); from the Italian Ministry of Education, Universities and Research (PRIN 2010JS3PMZ_011); and from Fondazione Don Gnocchi, Milan, Italy.

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

Author Contributions. M.Fa. performed experiments, analyzed data, drafted the manuscript, and wrote the final version of the manuscript. V.M. performed experiments, analyzed data, and reviewed the manuscript. M.M., A.M., V.C., M.C., T.M., G.P.S., and A.G. performed experiments and analyzed data. J.M.M.-N. performed experiments. L.F. and R.M. contributed to the discussion and drafted the manuscript. R.L. and G.M. contributed to the discussion and edited the manuscript. J.M.F.R. analyzed data and reviewed the manuscript. M.Fe. drafted the manuscript and wrote the final version of the manuscript. M.Fe. 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.