Adipocyte Glucocorticoid Receptor Deficiency Promotes Adipose Tissue Expandability and Improves the Metabolic Profile Under Corticosterone Exposure

Glucocorticoids (GCs) and synthetic analogs are among the most prescribed anti-inflammatory treatments (1). However high doses of GCs can lead to adverse effects, including hyperglycemia, dyslipidemia, insulin resistance, and lipodystrophy (2), which is characterized by an enlargement of visceral adipose tissue (VAT) at the expense of subcutaneous adipose tissue (SCAT). Conversely to the protective effect of SCAT expansion on metabolic disorders, VAT hypertrophy is a major risk factor for the development of diabetes and cardiovascular diseases (3). Thus, patients with endogenous hypercorticism (Cushing syndrome) share similar features with metabolic syndrome, suggesting a potential role of GCs in the pathophysiology of visceral obesity and insulin resistance (4).

GCs exert pleiotropic effects on adipose tissue (AT) physiology. They promote adipogenesis (5,6) and regulate adipocyte lipid storage and mobilization as well as adipokine secretion (5,7,8). These effects rely on the nature of subcutaneous or visceral AT as well as on the GC concentration and duration of exposure (9–11). GC actions are mediated through two nuclear receptors, the GC receptor (GR) and the mineralocorticoid receptor (MR) (7,8), which play specific roles in the development and functions of adipocytes and also on AT pathophysiology (7,8).

Global human MR overexpression in transgenic mice leads to resistance to high-fat diet–induced obesity by regulating adipocyte differentiation and macrophage polarization (12). However, this protective effect is abolished when MR is specifically overexpressed in AT, leading to insulin resistance and metabolic syndrome under diet-induced obesity (13). Furthermore, in vivo MR pharmacological blockade by eplerenone reduces the expression of proinflammatory factors in genetically obese and diabetic mice, suggesting a role for MR as a key factor mediating obesity-related inflammation and insulin resistance (14). The implication of GR in the pathophysiology of metabolic syndrome has been suggested by the pharmacological blockade of GR (by RU486), which improved fasting hyperinsulinemia, insulin resistance, and glucose intolerance (15). More recently, the role of the adipocyte GR was investigated in vivo using constitutive murine models of adipocyte GR deletion (16–20). However, these studies led to contradictory results, where the adipocyte GR invalidation could play a minor role in diet-induced obesity (17,20) or have a metabolic effect protecting from age-related and diet-induced obesity (19). This metabolic improvement in adipocyte GR-deficient mice could involve the inhibition of negative feedback mechanisms in the hypothalamic-pituitary-adrenal axis (16). Furthermore, studying the role of adipocyte GR in the context of GC-induced metabolic dysfunction also led to divergent results on insulin resistance improvement (18,20). These discrepancies may rely on the murine model of constitutive GR invalidation that could profoundly modify the biology of adipose cells and tissue through compensatory mechanisms.

To circumvent the phenotypic effect of GR ablation on growth and developmental processes, we generated a mouse model of conditional GR invalidation specifically in AT of adult adipocyte GR knockout mice (AdipoGR-KO). To address the role of adipocyte GR implications under exogenous hypercorticism conditions, the AdipoGR-KO mice were administered a 4-week corticosterone (CORT) treatment. Interestingly, these mice exhibited an expansion in fat mass under CORT exposure. Despite increased adiposity, AdipoGR-KO mice displayed an improved metabolic profile, with a selective enhancement in AT insulin sensitivity. Furthermore, AdipoGR-KO mice harbored a normal plasma lipid profile and were protected from lipid ectopic deposition in the liver. Thus, our results demonstrate that adipocyte GR is a key determinant of AT expandability and insulin sensitivity, which in turn influences whole-body glucose and lipid homeostasis.

The AdipoGR-KO mice were generated by crossing C57BL/6J GR^Lox/Lox mice (gift of Dr. F. Tronche, INSERM U1130, CNRS UMR 8246, Department of Neurosciences Paris Seine, Institute of Biology Paris Seine, Sorbonne University, Paris, France [21]) with C57BL/6J mice, which express the CreERT2 recombinase under the adiponectin promoter (gift of Dr. S. Offermanns, Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany [22]). Cre recombinase was activated by tamoxifen intraperitoneal injection in 10-week-old male mice (60 µg/g/day) (MP Biomedicals, Illkirch-Graffenstaden, France) for 5 consecutive days. During a 4-week period, 14-week-old AdipoGR-KO and control (wild-type [WT]) mice were treated with CORT (100 µg/mL) (Sigma-Aldrich, St. Louis, MO) or vehicle (VEH) (1% ethanol) added to tap water as previously described (23). Mice were sacrificed at 18 weeks old. Tissues were dissected, rapidly frozen in liquid nitrogen, and kept at −80°C. Blood was collected by intracardiac puncture with heparin-moistened syringes. Plasma was obtained after centrifugation at 5,000g for 10 min at 10°C.

The animal housing facility was granted approval (C 75-12-01) by the French Administration. All experiments were conducted according to the European Communities Council Directive (2010/63/UE) and approved by the Regional Animal Care and Use Committee (Ile-de-France, Paris, no5; agreement number 00917.02 and 4625).

For oral glucose tolerance tests (OGTTs), mice fasted for 5 h received an oral glucose load (1 g/kg body wt). For insulin tolerance tests (ITTs), mice fasted for 5 h were injected intraperitoneally with 1 IU/kg insulin (Actrapid Penfill; Novo Nordisk, Paris, France). Whole-tail vein blood was collected at baseline (t₀) and 15 min (t₁₅), and centrifuged plasma samples were stored at −80°C. For both tests, blood glucose was measured at the tail vein using an automatic Accu-Chek Performa glucometer (Roche, Meylan, France) at indicated times. For insulin-signaling experiments, mice were intraperitoneally injected with 1.5 IU/kg of human insulin (Actrapid Penfill) and sacrificed 10 min later, and tissues were stored at −80°C.

Functional analyses were performed on inguinal SCAT and gonadal AT (GAT) explants collected from WT and AdipoGR-KO mice treated with VEH or CORT. Glucose uptake was adapted from previous studies (24,25). Lipogenic activity, free fatty acid (FFA) reesterification, and lipolysis were adapted from a previous study (26). Details are presented in the Supplementary Data.

Paraffin-embedded liver and AT were cut in 4-μm-thick sections and stained with hematoxylin and eosin. Adipocyte size distribution was determined on 5–10 fields of SCAT and GAT per section covering the entire tissue surface, at original magnification ×10 (optical microscope IX83; Olympus). Adipocyte quantification was performed using ImageJ software (http://rsbweb.nih.gov/ij/) on ∼1,000–5,000 cells per mouse. Considering that adipocytes constituted 90% of the AT volume, cell number was estimated by dividing the tissue mass by the mean volume of cells. The results are expressed as cells per gram of tissue.

Macrophages were isolated from SCAT and GAT from WT and AdipoGR-KO mice treated with CORT and stained as described previously (27). Stained cells were analyzed using a Gallios flow cytometer (Beckman Coulter, Villepinte, France), and data were processed using Kaluza software (Beckman Coulter).

All values are expressed as the means ± SEM. Comparisons between animal groups were performed using a parametric Student t test or by a nonparametric Mann-Whitney–Wilcoxon test using Prism 5.0 software (GraphPad Software, La Jolla, CA). χ² analysis was performed to study the statistical differences in adipocyte surface distribution. A P value <0.05 was considered statistically significant.

To validate our model, we measured the relative GR gene expression in SCAT, GAT, perirenal AT (PAT), and brown AT (BAT) of AdipoGR-KO and WT mice. We observed a partial decrease in GR transcript level (50–60%) restricted to all AT of AdipoGR-KO mice (Supplementary Fig. 1A and C). The partial reduction was due to the heterogeneity of AT, which is composed of mature adipocytes and also of a mixture of nonadipocyte cell populations (28). Analysis of the isolated adipocyte fraction from SCAT and GAT showed GR protein content was nearly extinct in AdipoGR-KO mice compared with their littermates, confirming the GR invalidation in mature adipocytes (Supplementary Fig. 1B). As expected, gene expression of two known GR target genes (angiotensinogen [AGT] and 11β-hydroxysteroid dehydrogenase type 1 [11β-HSD1]) was significantly decreased by 50–80%, in accordance with GR deficiency in AT of AdipoGR-KO mice (Supplementary Fig. 1A and D). Finally, MR mRNA content and two known MR target genes (serum and glucocorticoid-regulated kinase 1 [SGK1] and prostaglandin D2 synthase [PTGDS]) remained unchanged (Supplementary Fig. 1E and F), indicating that no compensatory mechanisms through MR occurred in AT of AdipoGR-KO mice.

AdipoGR-KO and WT mice were treated with CORT or VEH for 4 weeks. As expected, CORT levels measured at the light/dark transition (8 a.m. and 8 p.m.) were significantly elevated in the plasma of CORT-treated mice, with a maintained daily variation, but were not significantly modified between the two genotypes (Supplementary Fig. 2A). In agreement with the absence of adipocyte GR and in response to CORT, we detected a strong downregulation of GR-specific target genes (AGT, 11β-HSD1, and FKBP5) in all AT of AdipoGR-KO mice (Supplementary Fig. 3). CORT treatment led to a progressive and similar body weight gain in both genotypes throughout the 4-week exposure (Fig. 1A). Interestingly, DEXA analysis showed a marked enhancement of fat mass correlated to a decreased lean mass in CORT-treated AdipoGR-KO mice compared with CORT-treated WT mice (Fig. 1B) and confirmed by the increased SCAT, GAT, and PAT mass (Fig. 1C) and the reduced liver and skeletal muscle weight in CORT-treated AdipoGR-KO mice (Fig. 1D and E). Surprisingly, despite AT expansion, leptin plasma displayed a twofold decrease in CORT-treated AdipoGR-KO versus WT mice (Supplementary Fig. 2B). Analysis of food intake, locomotor activity, and energy expenditure did not show any differences between both CORT-treated genotypes (Supplementary Fig. 2C–E). Although the respiratory exchange ratio (RER) was not modified over 24 h, CORT-treated AdipoGR-KO mice preferentially used glucose as substrate during the dark period and switched toward fatty acid (FA) utilization during the light period (Fig. 1F, right panel), which indicated a lower FA oxidation during the night period and a higher FA utilization in the daylight period (Supplementary Fig. 2F).

We next evaluated the effect of adipocyte GR invalidation on glucose and insulin tolerance in CORT- and VEH-treated mice (Fig. 2A). As expected in WT mice, CORT treatment led to glucose intolerance and insulin resistance compared with VEH-treated mice (Fig. 2A and B, right panel). Interestingly, AdipoGR-KO mice showed a trend toward improved glucose tolerance and a significant increase in insulin sensitivity upon CORT treatment (Fig. 2A and B). In agreement with these results, CORT dramatically increased plasma insulin concentration in WT mice but at a much lower level in AdipoGR-KO mice at 0 and 15 min after glucose gavage (Fig. 2C). The HOMA of insulin resistance index was significantly decreased in both CORT- and VEH-treated AdipoGR-KO mice (Fig. 2D). Insulin signaling was also assessed upon insulin pulse experiments in tissues from both genotypes (Fig. 2E and F). Interestingly, in CORT-treated AdipoGR-KO mice, Akt phosphorylation was significantly increased in the GAT and to a lesser extent in the SCAT compared with CORT-treated WT. Although Akt phosphorylation was only modestly affected in AT and muscle from CORT-treated WT mice, it was markedly decreased in the liver. This impaired Akt phosphorylation was not rescued in AdipoGR-KO mice, suggesting that adipocyte GR invalidation specifically improves insulin signaling activation in SCAT and GAT (Fig. 2E and F).

In the context of obesity, insulin resistance is associated with a low-grade tissue-specific inflammation state (29,30). GC treatment during high-fat diet induces adipose mass expansion and insulin resistance without increasing inflammation and macrophage recruitment (31). We analyzed the mRNA content of key cytokines in isolated adipocyte fractions from SCAT and GAT of CORT-treated mice (Supplementary Fig. 4) and did not observe any significant changes in gene expression of anti-inflammatory cytokines (IL-10 and IL-1 receptor antagonist, known to antagonize IL-1 action) (Supplementary Fig. 4A) and of proinflammatory IL-6 and IL-1β in AT of both genotypes (Supplementary Fig. 4B) except for tumor necrosis factor-α [TNF-α], which showed a trend toward a decreased expression in GAT of AdipoGR-KO mice (Supplementary Fig. 4B). Interestingly, MCP-1 chemokine (also known as CCL2) expression was highly increased in adipocyte fractions of SCAT and GAT in AdipoGR-KO mice (Supplementary Fig. 4C), suggesting an enhanced monocyte/macrophage migration and infiltration into AT. This result prompted us to measure by flow cytometry the percentage of total macrophage population in SCAT and GAT of CORT-treated AdipoGR-KO and control mice. As observed in Fig. 3A, the percentage of F4/80+/CD11b+ macrophages was not significantly altered in AdipoGR-KO compared with WT mice. We next determined the M1-like macrophage population using CD11c and NOS2 markers and M2-like macrophages with the CD206 marker. Interestingly, deletion of GR in AT was associated with a significant reduction in M1-like markers and an increase in M2-like markers in SCAT of AdipoGR-KO (Fig. 3B and C, left panels). In GAT of AdipoGR-KO mice, only an induction of M2-like macrophages was observed (Fig. 3B and C, right panels). Interestingly, this macrophage polarization pattern was associated with a significant increase in the mRNA and protein content of adiponectin, which could exert anti-inflammatory and insulin-sensitizing properties in AT of AdipoGR-KO mice (Supplementary Fig. 5). Altogether, these results suggest that adipocyte GR invalidation may favor a M2-like macrophage phenotype in the expanded AT.

Increased adiposity and metabolic changes were mostly observed in mice under CORT exposure. We thus examined the regulation of key enzymes and transcription factors involved in lipid metabolism under nutritional manipulations in response to CORT. Upon refeeding, two main transcription factors, SREBP1c and ChREBP, mediate the insulin and glucose action on enzymes of the lipogenic pathway (32–34). SREBP1c gene expression was significantly increased in SCAT and GAT of AdipoGR-KO mice, whereas ChREBP expression decreased but not significantly (Fig. 4A and B). Interestingly, the expression of their target genes involved in esterification and triglyceride (TG) synthesis (mGPAT, AGPAT, and DGAT), rather than in lipogenesis, was significantly enhanced in both AT of AdipoGR-KO versus WT mice (Fig. 4A and B). Under fasting conditions, we measured the expression of ATGL (adipocyte triglyceride lipase) and of HSL (hormone-sensitive lipase), the two main enzymes of the lipolytic response. HSL protein content and phosphorylation levels were not modified in AT from AdipoGR-KO compared with WT mice (Fig. 4E and F). In contrast, ATGL protein content was significantly decreased in GAT of AdipoGR-KO mice (Fig. 4E and F), whereas gene expression remained unchanged in both AT of AdipoGR-KO mice (Fig. 4C and D). Altogether, these results suggest that these metabolic pathways could contribute to the increased adiposity of AdipoGR-KO mice.

The phenotype of AdipoGR-KO mice prompted us to investigate the major metabolic pathways in SCAT and GAT explants. Glucose uptake and de novo lipogenesis were assessed in the presence or the absence of insulin (Fig. 5A–D). As expected, CORT inhibited insulin-stimulated glucose uptake (Fig. 5A and B) and lipogenesis (Fig. 5C and D) in AT of WT mice. Interestingly, a partial to complete rescue of insulin-stimulated glucose uptake and de novo lipogenesis was observed in AT of AdipoGR-KO mice (Fig. 5A–D).

Isoproterenol-induced acute lipolytic activity measured by glycerol release was decreased in a similar manner by CORT in WT and AdipoGR-KO mice compared with VEH-treated mice (Fig. 5E and F). FFA released in the media was also measured to determine the FFA reesterified into newly synthesized TGs into adipocyte lipid droplets (Fig. 5G and H). The FFA-to-glycerol release ratio (also called glyceroneogenic index) is an indicator of reesterification and is equal to 3 in the absence of newly esterified TG, because three molecules of FFA and one molecule of glycerol are released during complete TG hydrolysis (Fig. 5I and J). Under CORT treatment, FFA reesterification was decreased by 40–60% in AT explants of WT mice compared with VEH-treated mice (Fig. 5G and H), and consequently, the glyceroneogenic index was close to 3 (Fig. 5I and J). In AdipoGR-KO mice AT explants, the FFA reesterification was strongly enhanced (Fig. 5G and H) and was therefore associated with a lower glyceroneogenic index (∼2) in AT explants (Fig. 5I and J).

To further investigate in vivo the role of adipocyte GR deficiency on lipid metabolism, we measured plasma levels of nonesterified FAs (NEFAs), TG, LDL, and HDL cholesterol (Table 1). Increased TG and NEFA plasma levels by CORT treatment was significantly reduced in GR-deficient mice. Furthermore, CORT similarly enhanced LDL cholesterol concentration in both genotypes but only significantly increased HDL cholesterol concentration in AdipoGR-KO mice (Table 1). Taken together, these results show that adipocyte GR deletion improves lipid synthesis and storage (through de novo FFA synthesis and reesterification) in SCAT and GAT of AdipoGR-KO mice.

The flexibility of AT mass is an adaptive response to overnutrition or obesiogenic conditions to prevent ectopic lipid deposition and lipotoxicity in non-ATs (35). We investigated the remodeling of AT by analyzing the adipose cell size distribution in SCAT and GAT of CORT-treated mice under fed and refed conditions, a condition with increased TG accumulation.

Despite the increased fat mass in AdipoGR-KO mice (Fig. 1C), the adipose cell size distribution was similar in the SCAT of AdipoGR-KO and control mice (Fig. 6A and C). A similar mean adipocyte surface area and number of cells per gram of SCAT was determined on fed or refed conditions (Supplementary Fig. 6A and B). Analysis of GAT cell size frequency revealed clear changes depending on the nutritional status in AdipoGR-KO mice (Fig. 6B and D). Although a similar cell size distribution was observed between both genotypes under the fed condition, AdipoGR-KO mice displayed a higher frequency of large adipocytes in the refed condition (Fig. 6B), which was correlated with an increase in the mean adipocyte area and, consequently, a lower number of adipocyte cells per gram of GAT compared with refed WT mice (Supplementary Fig. 6A and B). Finally, we did not observe any significant change in gene expression of key adipogenic transcription factors (C/EBPβ, C/EBPδ, C/EBPα, and PPARγ) in SCAT or GAT of both genotypes (Supplementary Fig. 7), indirectly suggesting that the differentiation process could be attenuated, allowing a massive enlargement of the adipocytes present in the GAT of the AdipoGR-KO refed mice.

Chronic exposure to GCs is strongly associated with fatty liver development (36). Accordingly, numerous and large lipid droplets were revealed in liver sections of CORT-treated WT mice and correlated with a strong TG content compared with VEH-treated mice (Fig. 7A and B). Interestingly, CORT-treated AdipoGR-KO mice displayed a sharp decrease in hepatic TG concentrations, similar to VEH-treated mice and in agreement with the reduced liver weight (Figs. 1D and 7A and B). Furthermore, no remarkable difference in the expression of key enzymes involved in lipid and glucose metabolism was observed, except for Elovl6, which was the only lipogenic gene significantly decreased in the liver of AdipoGR-KO mice (Fig. 7C). CPT1 (carnitine palmitoyltransferase 1), an enzyme of FA oxidation, was also markedly reduced in AdipoGR-KO mice and paralleled the decrease in plasma ketone bodies (Fig. 7D and Table 1). Thus, these results suggest that AdipoGR-KO mice are protected from GC-induced hepatic steatosis through a preferential storage of lipids into AT.

GCs are involved in pleiotropic effects in AT, including the control of adipogenesis, lipolysis, lipogenesis, insulin sensitivity, and thermogenesis (37). Endogenous or pharmacological GC excess is also associated with lipodystrophy, insulin resistance, and disturbances in glucose and lipid homeostasis (8). However, despite an increasing number of studies, adipocyte GR contribution to these metabolic adverse effects remains unclear. To uncover its role, we generated a conditional adipocyte-specific model of GR invalidation and demonstrated that adipocyte GR invalidation leads to an expansion in fat mass under CORT exposure associated with improved carbohydrate and lipid profiles. Moreover, AdipoGR-KO mice exhibited enhanced insulin sensitivity, selectively restricted to AT. The main finding of our study is that adipocyte GR is a major determinant of AT expansion and is a key mediator of GC-induced metabolic dysregulations.

To our knowledge, only two studies have investigated the effects of GC administration on the metabolic phenotype of mice with a constitutive adipocyte GR invalidation model using CORT (18) or dexamethasone (18,20). Although very similar animal models were used, the metabolic phenotype diverged between both studies. A CORT treatment (10 mg/kg body wt in tap water) was used by Bose et al. (18) to mimic hypercorticism conditions. Metabolic phenotyping of AdipoGR-KO mice showed no difference in weight gain, fat pad weight, hepatic TG content, and glucose tolerance as well as insulin resistance compared with WT mice. These mice were treated with three-times less CORT dose than our mice (30 mg/kg body wt in tap water), and treatment started at the age of 6 months against 3.5 months in our study. Thus, low dose of CORT in older animals may account for the lack of marked metabolic effects.

Bose et al. (18) and Shen et al. (20) also treated their mice with dexamethasone, a specific synthetic GR agonist, but leading to conflicting results. Whereas Bose et al. (18) showed no major effect of dexamethasone treatment (10 mg/kg daily for 2 weeks on 6-month-old mice) on adiposity and metabolic profile on adipocyte GR-deleted mice (18), Shen et al. (20) reported a trend toward increased adiposity with an improved glucose tolerance and insulin sensitivity in their GR-KO mice compared with WT mice after GC exposure. Akt phosphorylation was specifically enhanced in skeletal muscle of AdipoGR-KO mice. These mice were treated with lower doses of dexamethasone (3 mg/kg body wt for 2 months in 4-month-old mice) (20). The explanations for these discrepancies may rely on the treatment modalities (dose of dexamethasone, treatment duration, and age of animals), which differ between the two studies.

In the current study, AdipoGR-KO mice exhibited obvious increased adiposity of all fat depots associated with improved insulin sensitivity. In contrast to the study of Shen et al. (20), which reported an increased insulin sensitivity restricted to skeletal muscle, we clearly demonstrated that GR deletion in adipocytes improved this signaling pathway selectively in AT. This suggests that adipocyte GR could represent an important determinant of whole-body insulin sensitivity in AdipoGR-KO mice. A key explanation for these tissue-specific modulations in insulin sensitivity could be related to our choice of a conditional model of GR invalidation, which allowed us to circumvent the phenotypic effect of GR ablation during AT growth and development. Because several studies have documented that GCs and the GR are key players of adipose physiology (8), constitutive GR invalidation models used in the previously published works (16–20) could combine GR pleiotropic effects on both developmental and metabolic functions of adipocytes.

That GC actions are mediated though GR and MR in AT of rodents is well established. GC affinity for the MR is 10-fold higher than for GR (38), and MR gene expression is known to be elevated in AT of obese rodents (38,39). In agreement with Shen et al. (20), we did not find any upregulation of MR mRNA levels or any of its target genes (SGK1 and PTGDS) in AT of our mice, suggesting that no adaptive mechanism through the MR pathway was observed in our GR-deficient mice.

Rodent and human obesity are associated with a subinflammatory state of AT, with proinflammatory M1 macrophage infiltration (40,41). In this context, the effects of GCs seem paradoxical because they induce an AT expansion and disorders in glucose and lipid metabolism and also exert potent anti-inflammatory properties (31). In our study, adipocyte GR deletion did not influence the total number of macrophages but modulated macrophage polarization in a depot-dependent manner. The global shift from M1 to M2 macrophage polarization observed in SCAT and in a less pronounced manner in GAT fits with the improved metabolic profile of AdipoGR-KO mice and suggests that the absence of adipocyte GR protects from an inflammatory phenotype and the GC-induced related metabolic complications.

What could be the mechanisms underlying these changes in macrophage polarization in GC-treated AdipoGR-KO mice? Several arguments are in favor of an M2-like shift of macrophage polarization in the AT of AdipoGR-KO mice after CORT exposure. First, the enhanced local adiponectin mRNA and protein content in SCAT and GAT of AdipoGR-KO mice may contribute to the anti-inflammatory profile in AT and to the specific improvement in AT insulin sensitivity (42). Second, despite the expansion of AT, AdipoGR-KO mice exhibited a twofold decrease in leptin plasma levels. This is in agreement with the known direct effect of GCs to increase leptin synthesis and secretion in adipocytes (43). Leptin modulates a wide range of immune and inflammatory processes, activating monocyte proliferation and cytokine/chemokine expression (44,45), macrophage phagocytosis, and chemotaxis (46). Thus, under GC exposure, the lower leptin levels in AdipoGR-KO mice could mitigate macrophage classical activation and promote an M2 alternative phenotype. Alternatively, due to the increased lipid esterification and synthesis, AdipoGR-KO mice exhibited lower NEFA plasma levels, which are also known to promote M1 polarization (47). Consequently, decreased NEFA and leptin levels could act in synergy to favor M2 polarization. This illustrates the possibility that expansion of AT could be associated with a prominent M2-like macrophage phenotype, together with a healthier metabolic profile.

In our work, several results converge to demonstrate that the adipocyte GR is a major determinant of AT expandability. First, at a morphological level, the adipocyte GR deletion was associated with an increased fat deposition. Interestingly, a higher proportion of large adipocytes in the AdipoGR-KO fat mass was only detectable upon refeeding nutritional manipulation and in a depot-specific manner because GAT seemed more prone to these genotype-associated cell size variations. Second, these morphological changes were paralleled by biochemical changes that fit with the increased insulin sensitivity in AdipoGR-KO mice despite CORT exposure. Experiments performed on SCAT or GAT explants showed a total or partial rescue of glucose uptake and de novo lipogenesis in CORT-treated AdipoGR-KO animals, which was in agreement with the increased expression of genes targeted by the key lipogenic transcription factor SREBP1c. Surprisingly, despite unmodified isoproterenol-stimulated lipolysis in adipose explants from CORT-treated AdipoGR-KO mice, FFA reesterification was clearly induced, participating in lipid storage into adipocytes and improved plasma lipid profile. All of these biochemical events could reflect the combined effect of the increased insulin sensitivity in AT and also of the absence of GR-mediated actions on its target genes (48). Finally, at a systemic level, adipocyte GR deletion is also characterized by changes in fuel utilization, as assessed by circadian variations in RER. The increased preferential carbohydrate oxidation observed in AdipoGR-KO mice mainly occurred during the dark period. Consequently, this carbohydrate utilization improves glucose homeostasis. The reorientation of glucose utilization toward AT in AdipoGR-KO mice is sufficient to totally prevent liver steatosis induced by CORT and may protect other non-ATs from lipid spill over. Thus, adipocyte GR could play a key role in the onset and progression of GC-induced liver steatosis.

In conclusion, our results demonstrate that a conditional adipose-specific GR invalidation protects from insulin resistance induced by GCs. This is closely related to a specific improvement in AT insulin sensitivity that favors an expansion of fat depots. The AdipoGR-KO mouse thus represents a model of AT overexpansion associated with a healthier metabolic phenotype. Furthermore, this strongly suggests that adipocyte GR is a limiting factor for AT expansion, contributing to metabolic disorders under GC exposure. A selective modulation of adipocyte GR expression and/or function is thus a relevant target to counteract the metabolic side effects of GCs.

Acknowledgments. The authors thank Dr. S. Offermanns (Max Planck Institute for Heart and Lung Research, Bad Nauheim, Germany) and Dr. F. Tronche (INSERM U1130, CNRS UMR 8246, Department of Neurosciences Paris Seine, Institute of Biology Paris Seine, Sorbonne University, Paris, France) for their kind gift of AdipoqCRE-ERT2 and GR^Lox/Lox mice, respectively. The authors are grateful to Dr. F. Benhamed (INSERM 1016, UMR8104, Paris), M. Auclair, C. Kazazian, R. Tchuinkam, and G. Vacher (all from INSERM, Saint-Antoine Research Center, Sorbonne University, Paris) for their technical assistance. The authors thank L. Dinard, A. Guyomard, T. Coulais, and Q. Pointout (animal housing facility, INSERM, Saint-Antoine Research Center, Sorbonne University, Paris), S. Dumont, B. Solhonne, and F. Merabtene (Histomorphology Platform, INSERM, UMS 30, Sorbonne University, Paris), R. Morrichon (Cell Imaging and Confocal Microscopy Platform, INSERM, UMS 30, Sorbonne University, Paris), and A. Munier (Cytometry Platform, INSERM, UMS 30, Sorbonne University, Paris) for their excellent work, and C. Klein and K. Garbin for technical and imaging support (UMRS 1138, CHIC Platform, Centre de Recherche des Cordeliers). The authors thank Dr. Alexandra Grosfeld (INSERM, Saint-Antoine Research Center, Sorbonne University, Paris) for helpful discussion.

Funding. This work was supported by grants from INSERM, Sorbonne University, French Society of Diabetes (Société Francophone pour le Diabète [SFD]) and the Medical Research Foundation (FRM). H.D. was supported by a doctoral fellowship from Ministère de l’Enseignement Supérieur et de la Recherche and FRM. V.B. received support from the French Society of Endocrinology (SFE). T.T.H.D. received support from SFE and the French Embassy in Vietnam.

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

Author Contributions. H.D., M.G., B.A., V.B., T.T.H.D., M.B., A.L., C.M., C.P., and M.M. performed the research. H.D., M.G., B.A., B.F., and M.M. analyzed data. H.D., B.F., and M.M. wrote the manuscript. T.L. contributed to generate the AdipoGR-KO mouse model and animal experiments. R.G.D. and S.L. performed indirect calorimetry measurements and analyzed data. B.F. and M.M. designed the research. B.F. and M.M. are the guarantors of this work and, as such, had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.