Adipose glucocorticoid action influences whole-body metabolism via modulation of hepatic insulin action

Abstract

The connection between adipose glucocorticoid action and whole-body metabolism is incompletely understood. Thus, we generated adipose tissue–specific glucocorticoid receptor–knockout (Ad-GcR^−/−) mice to explore potential mechanisms. Ad-GcR^−/− mice had a lower concentration of fasting plasma nonesterified fatty acids and less hepatic steatosis. This was associated with increased protein kinase B phosphorylation and increased hepatic glycogen synthesis after an oral glucose challenge. High-fat diet (HFD)–fed Ad-GcR^−/− mice were protected against the development of hepatic steatosis and diacylglycerol-PKCε–induced impairments in hepatic insulin signaling. Under hyperinsulinemic-euglycemic conditions, hepatic insulin response was ∼10-fold higher in HFD-fed Ad-GcR^−/− mice. Insulin-mediated suppression of adipose lipolysis was improved by 40% in Ad-GcR^−/− mice. Adipose triglyceride lipase expression was decreased and insulin-mediated perilipin dephosphorylation was increased in Ad-GcR^−/− mice. In metabolic cages, food intake decreased by 3 kcal/kg per hour in Ad-GcR^−/− mice. Therefore, physiologic adipose glucocorticoid action appears to drive hepatic lipid accumulation during stressors such as fasting. The resultant hepatic insulin resistance prevents hepatic glycogen synthesis, preserving glucose for glucose-dependent organs. Absence of adipose glucocorticoid action attenuates HFD-induced hepatic insulin resistance; potential explanations for reduction in hepatic steatosis include reductions in adipose lipolysis and food intake.—Abulizi, A., Camporez, J.-P., Jurczak, M. J., Høyer, K. F., Zhang, D., Cline, G. W., Samuel, V. T., Shulman, G. I., Vatner, D. F. Adipose glucocorticoid action influences whole-body metabolism via modulation of hepatic insulin action.

Glucocorticoids coordinate myriad physiologic adaptations to stressors (e.g., prevention of hypoglycemia); however, in excess, glucocorticoids profoundly distort lipid and glucose metabolism, leading to central obesity, dyslipidemia, fatty liver, and hyperglycemia (1, 2). Cellular glucocorticoid action depends both on circulating hormone concentration and local steroid metabolism (3). Both the physiological and pharmacological actions of glucocorticoids are mediated by the glucocorticoid receptor (GcR) (NR3C1), a ubiquitously expressed member of the nuclear receptor superfamily of ligand-dependent transcription factors (4). GcR-ligand binding transactivates some genes controlled by glucocorticoid response elements, transrepresses other (particularly proinflammatory) genes, and may also influence cellular biology at a post-translational level (5).

Glucocorticoids play several important roles in the regulation of adipose tissue function, and as such, glucocorticoid action in adipose tissue is likely to have a critical physiologic role impacting whole-body metabolism.

White adipose tissue (WAT) plays a variety of important roles as an energy storage depot, an endocrine organ that secretes adipokines (e.g., leptin, adiponectin, and retinol binding protein 4), and a source of inflammatory cytokines. WAT-derived adipokines and cytokines modulate eating behavior and energy expenditure (6, 7). Glucocorticoids alter adipocytokine production (8–12) and suppress WAT inflammation by preventing adipose macrophage recruitment (13). In cell culture systems, glucocorticoids induce preadipocyte differentiation and promote adipocyte lipolysis (14, 15). In general, glucocorticoids stimulate adipose lipolysis, although they also have antilipolytic effects (16, 17). The prolipolytic actions of glucocorticoids can in part be explained by induction of the lipolytic enzymes hormone-sensitive lipase (HSL) and adipose triglyceride lipase (ATGL) (18–20). Furthermore, the adipocyte cAMP-PKA signaling pathway regulates the induction of adipocyte lipolysis, and cAMP is elevated in glucocorticoid treated adipocytes, a finding potentially explained by a reduction in phosphodiesterase 3B activity (21). In vivo, glucocorticoid-stimulated adipose lipolysis can both increase hepatic gluconeogenesis, and promote liver and muscle insulin resistance through increased ectopic lipid deposition (22–26).

Adipose glucocorticoid action may impact whole-body metabolism by regulating the availability of lipids. There are several lines of evidence suggesting that adipose tissue–specific glucocorticoid action regulates whole-body energy metabolism. Augmenting adipose glucocorticoid action by overexpressing 11β-hydroxysteroid dehydrogenase type 1 in adipose tissue impairs whole-body glucose and lipid metabolism (27). Reducing hepatic and adipose GcR expression with antisense oligonucleotides improves hyperglycemia and hyperlipidemia in diabetic rats (28). Studies with adipose tissue–specific GcR-knockout (Ad-GcR^−/−) mice, developed independently of our present work, suggested that the GcR may be important for the control of systemic fuel partitioning and energy metabolism. One such study suggests that deficiency of the GcR in adipose is associated with the prevention of dyslipidemia and insulin resistance in aged fat-fed mice (29). In contrast, other studies suggest that the adipocyte GcR plays a negligible role in the regulation of whole-animal energy metabolism, with the absence of the GcR in adipose tissue failing to alter the course of metabolic disorders (30, 31). Yet another such study indicated that the adipose GcR has neither an important nor a reproducible role in the progress of metabolic dysfunction after high-fat feeding (32). In the setting of such conflicting results, the specific glucocorticoid-regulated functions of adipose tissue that are important for regulation of whole-body metabolism are still unclear. Therefore, to better understand the mechanisms by which alterations in adipose glucocorticoid action may alter whole-body glucose and lipid homeostasis, we performed a comprehensive set of studies in high-fat diet (HFD)–fed Ad-GcR^−/− mice.

MATERIALS AND METHODS

Animal care

All experimental procedures were approved by and conducted in accordance with the Institutional Animal Care and Use Committee guidelines of Yale University School of Medicine. Ad-GcR^−/− mice on a C57BL/6 background were generated by crossing Nr3c1 floxed mice (Jax stock 021021) with Adipoq-cre mice (Jax stock 010803; The Jackson Laboratory, Bar Harbor, ME, USA). Adipoq-cre was used because it is a well-established system for efficient targeting of genes in mature adipocytes (33). In all studies, wild-type (WT) littermates served as controls. Mice were individually housed under controlled temperature (23°C) and lighting (12-h light/dark cycle; lights on at 7:00 am) with free access to water and food. Mice were maintained with regular chow (Envigo 2108S: 24% protein, 58% carbohydrate, 18% fat calories; Envigo, Huntington, United Kingdom) or given an HFD (Research Diets D12492: 20% protein, 20% carbohydrate, 60% fat calories; Research Diets, New Brunswick, NJ, USA) for 3 wk. Male mice were used in all studies. For studies of 20-h food-withdrawn mice (vs. mice fed ad libitum), mice were euthanized in the morning in the ad libitum fed state or after having food withheld for 20 h; both cohorts were euthanized at the same time in the same experiment. For studies of mice with food withheld for 20 h (vs. refed state), blood was obtained by tail bleed, mice were refed with regular chow pellets, and after 1 h, blood was again obtained by tail bleed. Mice for oral glucose tolerance tests (OGTTs) had food withheld for 20 h prior to study and studied in the morning. Mice had food withheld overnight for infusion studies and studied in the morning, and food was withheld for 6 h for basal measurements and studied in the afternoon (except where otherwise indicated). Body composition was assessed by ¹H magnetic resonance spectroscopy using a Bruker BioSpin Minispec Analyzer (Bruker, Billerica, MA, USA). Energy expenditure, respiratory quotient, Vo₂, Vco₂, locomotor activity, and food intake were measured using a comprehensive laboratory animal metabolic system (Columbus Instruments, Columbus, OH, USA). Drinking in the metabolic cages was measured as previously described by Birkenfeld et al. (34). Mice were 11–20 wk old during metabolic studies.

OGTTs

After 20 h of having food withheld, mice received 2 mg glucose per gram by oral gavage of a 20% glucose solution. Plasma was collected from the tip of the tail at 0, 20, and 60 min. After the 60-min blood collection, mice were anesthetized with isoflurane and tissues were taken, snap frozen in liquid nitrogen, and stored at −80°C for subsequent use.

Hyperinsulinemic-euglycemic clamps and measurement of fatty acid turnover

Hyperinsulinemic-euglycemic clamps were performed in awake mice as previously described by Jornayvaz et al. (35). A jugular venous catheter was implanted 6–7 d before the studies were performed. To assess basal whole-body glucose turnover, [3-³H]-glucose (HPLC purified; PerkinElmer, Waltham, MA, USA) was infused at a rate of 0.05 μC_i/min for 120 min into the jugular catheter. After the basal period, hyperinsulinemic-euglycemic clamps were conducted for 140 min with a 3-min–primed infusion of insulin (10 mU/kg/min) and [3-³H]-glucose (0.24 μC_i/min) followed by a continuous (3 mU/kg/min) infusion of human insulin (Novo Nordisk, Bagsvaerd, Denmark) and [3-³H]-glucose (0.1 μC_i/min) and a variable infusion of 20% dextrose to maintain euglycemia. Plasma samples were obtained from the tip of the tail at 0, 25, 45, 65, 90, 100, 110, 120, 130, and 140 min. The tail cut was made at least 2 h before the first blood sample was taken to allow for acclimatization according to standard operating procedures. Also, mice received an intravenous artificial plasma solution [115 mM NaCl, 5.9 mM KCl, 1.2 mM MgCl₂-6H₂O, 1.2 mM NaH₂PO₄-H₂O, 1.2 mM Na₂SO₄, 2.5 mM CaCl₂ -H₂O, 25 mM NaHCO₃, and 4% BSA (pH 7.4)] at a rate of 4.2 μl/min during the insulin-stimulated period of the clamp to compensate for volume loss secondary to blood sampling. At the end of the clamps, mice were anesthetized with sodium pentobarbital injection (∼4 mg/mouse), and all tissues taken were snap frozen in liquid nitrogen and stored at −80°C for subsequent use. [¹³C]₁₆ palmitate enrichment was measured by GC/MS, and turnover was calculated as previously reported by Perry et al. (23).

Plasma assays

Plasma glucose was measured using a YSI 2700D glucose analyzer (YSI, Yellow Springs, OH, USA). Standard kits were used to measure plasma nonesterified fatty acids (NEFAs) (Fujifilm, Tokyo, Japan), triglycerides (Sekisui, Tokyo Japan), and β-hydroxybutyrate (EKF Diagnostics, Cardiff, United Kingdom). Insulin and glucagon concentrations were determined by radioimmunoassay (MilliporeSigma, Burlington, MA, USA). Leptin concentrations were determined by ELISA (Abcam, Cambridge, United Kingdom).

Liver lipid measurements

Tissue triglycerides were extracted using chloroform:methanol (2:1) and measured using a standard kit (Sekisui). For diacylglycerol (DAG) extraction, livers were homogenized in a buffer solution (20 mM Tris-HCl, 1 mM EDTA, 0.25 mM EGTA, 250 mM sucrose) containing a protease inhibitor mixture (Roche, Basel, Switzerland); samples were centrifuged at 100,000 g for 1 h. The supernatants containing the cytosolic fraction were collected. DAG levels were then measured by liquid chromatography–tandem mass spectrometry as previously described by Jornayvaz et al. (35). DAG content is expressed as the sum of individual species. Ceramide was measured as previously described by Yu et al. (36).

Liver histology

Mouse livers were fixed overnight with 10% neutral buffered formalin at 4°C. Livers were washed with PBS, incubated for 24 h in 30% sucrose, embedded in Optimal Cutting Temperature (OCT) compound, and frozen. Yale Pathology Tissue Services (New Haven, CT, USA) sectioned the liver with a cryostat and performed staining with Oil Red O for visualization of neutral lipids.

Liver glycogen measurements

Hepatic glycogen was assessed as previously described by Samuel et al. (37). Liver tissue was homogenized in 0.6 N perchloric acid, an aliquot was taken to proceed with hydrolysis, and a separate aliquot was taken to assess background free glucose. The sample taken for glycogen hydrolysis was neutralized with potassium bicarbonate and incubated with aminoglucosidase (MilliporeSigma) at 40°C for 2–4 h. Glycogen content is reported as millimole glucose incorporated in glycogen per kilogram liver tissue.

Urinary corticosterone

For urine collection studies, mice fed regular chow ad libitum were housed in wire-bottom cages designed to separate and collect urine and feces. The mice acclimated to the cages for 1 d; urine collected during this acclimation period was discarded. Urine collected over the following 24 h was used for measurements of corticosterone and creatinine. Corticosterone concentration was measured by ELISA (Abcam). Creatinine was measured by liquid chromatography–tandem mass spectrometry (38), with [d3]-creatinine added as an internal standard detected in single reaction monitoring mode as the positive ion transitions of 114/86, 117/86, 114/44, and 117/47. Urinary corticosterone is reported as the ratio of corticosterone: creatinine concentrations.

Immunoblotting analysis

Tissues were homogenized in lysis buffer supplemented with protease and phosphatase inhibitor cocktails (Roche) for protein isolation. Proteins from homogenized liver tissue (30 μg of protein extracts) were separated by 4–12% SDS-PAGE (Thermo Fisher Scientific, Waltham, MA, USA) and then transferred to PVDF membranes (MilliporeSigma) using a semidry transfer cell (Bio-Rad, Hercules, CA, USA) for 120 min. After blockade of nonspecific sites with 5% nonfat dry milk Tris-buffered saline and Tween 20 (10 mM Tris, 100 mM NaCl, and 0.1% Tween 20) solution, membranes were incubated overnight at 4°C with primary antibodies. Membranes were thoroughly washed and incubated with the appropriate secondary antibody (Cell Signaling Technology, Danvers, MA, USA) at a dilution of 1:5000, and immune complexes were detected using a luminol ECL system (Thermo Fisher Scientific) and exposed to photographic film. Signals on the immunoblot were quantified by optical densitometry.

For PKCε translocation, cytoplasm and plasma membrane were separated by ultracentrifugation as previously described (39, 40) prior to Western blotting.

Antibodies used are the following: glycogen synthase (3893S), phosphorylated glycogen synthase (S641, 3891S), IRK (insulin receptor β, 3020S), phosphorylated IRK (Tyr 1162, 3918S), protein kinase B (Akt) (2920S), phosphorylated Akt (Ser473, 9721S), perilipin (3467S), ATGL (2138S), and HSL (4107S) (Cell Signaling Technology); phosphorylated perilipin (4856; Vala Sciences, San Diego, CA, USA); sodium potassium ATPase (Ab7671; Abcam); glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (sc25778; Santa Cruz Biotechnology Dallas, TX, USA); heat shock protein 90 (HSP90) (610419) and PKCε (610086) (BD Biosciences Billerica, MA, USA); and liver glycogen phosphorylase (NBP2-32246; Novus Biologicals, Centennial, CO, USA). Primary antibodies were prepared at a dilution of 1:1000 with the following exceptions: sodium potassium ATPase (1:2000) and GAPDH (1:3000).

Quantitative PCR

Total RNA was isolated from tissue using standard kits (Qiagen, Hilden, Germany). The abundance of transcripts was assessed by real-time PCR on an Applied Biosystems 7500 Fast Real-Time PCR System (Thermo Fisher Scientific) with a SYBR Green Detection System (Bio-Rad). The expression data for each gene of interest were against a housekeeping gene as a control, and relative expression was determined using amplification efficiencies (41). Primer sequence for angiopoietin-like 4s (ANGPTL4) was downloaded from the PrimerBank Database (https://pga.mgh.harvard.edu/primerbank/; PrimerBank identification number: 255308872c3). All primer sequence information is provided in Table 1.

TABLE 1.

Quantitative PCR primer sequences

	Primer sequence, 5′–3′
Gene	Forward	Reverse
TBP	GAGCTGTGATGTGAAGTTTCC	TCTGGGTTTGATCATTCTGTAG
ATGL	AACGCCACTCACATCTACGG	GCCTCCTTGGACACCTCAATA
HSL	GGAGTCTATGCGCAGGAGTG	GCTTCTTCAAGGTATCTGTGCC
MGL	GGCTGGACATGCTGGTATTT	ACTTGGAAGTCCGACACCAC
ANGPTL4a	CCCCACGCACCTAGACAATG	GCCTCCATCTGAAGTCATCTCA

ANGPTL4, angiopoietin-like 4; MGL, monoglyceride lipase; TBP, TATA-binding protein.

^aFrom the PrimerBank Database.

Statistical analysis

All data are expressed as means ± sem. Results were assessed using a 2-tailed, unpaired Student’s t test or 2-way ANOVA followed by Tukey’s Multiple Comparison Test (Prism 7; GraphPad Software, La Jolla, CA, USA). A value of P < 0.05 was considered significant.

RESULTS

Food withdrawal–induced physiologic hepatic steatosis is attenuated in Ad-GcR^−/− mice

To evaluate the role of adipose glucocorticoid action in the development of fasting-induced hepatic steatosis, we compared Ad-GcR^−/− mice and WT littermates in the 20-h unfed and ad libitum fed states (Fig. 1A, B, D, E and Table 2), or in the 20-h unfed and refed states (Fig. 1C and Table 2). Body weight was matched between genotypes (Table 2). There was no difference in plasma glucose between Ad-GcR^−/− mice and WT mice in the ad libitum fed state or the refed state. In the fasting state, plasma glucose was significantly different between genotypes, although the direction of change was inconsistent between groups (glucose was higher in conscious Ad-GcR^−/− mice and lower in isoflurane-anesthetized mice, Table 2). Plasma insulin levels were matched between genotypes (Table 2). Plasma glucagon was unaffected by genotype in the unfed or refed states (Table 2). Plasma NEFA concentration, a reflection of WAT lipolytic activity, was significantly decreased in Ad-GcR^−/− mice (Fig. 1A). Fatty acids liberated from adipose tissue may be reassembled as triglyceride in the liver for secretion in very low-density lipoprotein and storage in hepatocytes increasing plasma or hepatic triglyceride content; or oxidized by the liver, thereby raising plasma ketones. Thus, we evaluated plasma triglyceride concentration, plasma ketone body concentration, and hepatic triglyceride content. Plasma triglyceride concentration was not affected by genotype in either the fed or unfed states (Fig. 1B). Fasting plasma β-hydroxybutyrate concentration was reduced in Ad-GcR^−/− mice (Fig. 1C). The physiologic hepatic steatosis induced by the unfed state was significantly decreased in Ad-GcR^−/− mice (Fig. 1D, E). Taken together, these results suggest that adipose glucocorticoid action is important for the promotion of fasting WAT lipolysis, fasting ketone body production, and fasting hepatic steatosis.

Figure 1.

Deletion of the adipose GcR prevents fasting hepatic steatosis and reduces fasting circulating NEFAs and ketone bodies. A, B) Right panels: mice fed ad libitum; left panels: 20-h food-withheld mice. A) Plasma NEFAs. B) Plasma triglyceride concentration. C) Right panel: refed mice; left panel: 20-h food-withheld mice. Plasma β-hydroxybutyrate concentration. D, E) Hepatic triglyceride concentration per gram liver tissue, with corresponding mouse liver histology with Oil Red O staining for lipid. Measured in the ad libitum fed condition (D) and the 20-h food-withheld condition (E). GcR^−/−, Ad-GcR^−/−; TG, triglyceride. Original magnification, ×40 (D, E). A, B, D, E) Data are represented as means ± sem; n = 8–11/group. C) Data are represented as means ± sem; n = 6/group. *P < 0.05, ***P < 0.001.

TABLE 2.

Regular chow–fed Ad-GcR^−/− mice in the 20-h unfed and fed states

Measurement	WT	Ad-GcR−/−	P
Unfed 20 h vs. fed ad libitum
Body weight (g)
Unfed	27.2 ± 1.6	27.1 ± 0.6	0.94
Fed	30.3 ± 1.6	30.5 ± 1.0	0.94
Plasma glucose (mg/dl)
Unfed	162 ± 5	144 ± 7	0.04
Fed	226 ± 10	215 ± 14	0.54
Plasma insulin (µU/ml)
Unfed	6.4 ± 0.2	5.9 ± 0.2	0.08
Fed	18.2 ± 1.6	18.9 ± 3.1	0.84
Unfed 20 h vs. refed
Unfed body weight (g)	21.1 ± 1.0	22.1 ± 0.8	0.44
Plasma glucose (mg/dl)
Unfed	102 ± 4	118 ± 5	0.03
Refed	223 ± 8	224 ± 10	0.99
Plasma insulin (µU/ml)
Unfed	7.1 ± 1.3	5.0 ± 1.4	0.31
Refed	30.3 ± 7.3	38.7 ± 6.0	0.39
Plasma glucagon (pg/ml)
Unfed	38.6 ± 4.7	41.8 ± 7.5	0.72
Refed	30.4 ± 2.5	32.9 ± 3.9	0.62

Ad-GcR^−/− vs. littermate WT controls. Unfed mice and mice fed ad libitum are separate groups of mice; plasma parameters are from blood drawn from inferior vena cava under isoflurane anesthesia. Unfed and refed parameters are from the same mice under both conditions; plasma parameters are from blood drawn by tail bleed. Data are presented as means ± sem; n = 8–11/group (unfed vs. fed ad libitum); n = 6/group (unfed vs. refed, measurements other than glucagon); n = 4–6/group (unfed vs. refed, glucagon measurements).

Refed-state hepatic glycogen storage is accelerated in Ad-GcR^−/− mice

We hypothesized that a physiologic role of adipose glucocorticoid action is to promote hepatic steatosis and hepatic insulin resistance in times of stress, which would curtail hepatic glycogen synthesis after meals during periods of food scarcity, thereby defending glucose for the CNS. To test this hypothesis, we performed OGTTs in mice subjected to 20-h food withdrawal. We measured hepatic glycogen content in mice unfed for 20 h (basal), and in mice withheld from food for 20 h, then euthanized 60 min after receiving 2 mg/g glucose by gavage. During the OGTT, glucose excursions were matched in WT and Ad-GcR^−/− mice (Supplemental Fig. S1A), and post–OGTT, the Ad-GcR^−/− mice retained the previously observed difference in hepatic triglyceride content (Supplemental Fig. S1B). Glycogen content did not increase after administration of glucose in WT mice but increased by 50 mmol/kg (∼1% glycogen/liver tissue, w/w) in Ad-GcR^−/− mice (Fig. 2A). To evaluate whether changes in insulin-mediated regulation of glycogen synthesis could explain this result, we assessed glycogen synthase phosphorylation and Akt phosphorylation (Fig. 2B, C). WT mice did not demonstrate Akt phosphorylation or glycogen synthase dephosphorylation after the GTT, whereas Ad-GcR^−/− mice demonstrated a robust increase in Akt phosphorylation and reduction in glycogen synthase phosphorylation after the GTT. These findings are consistent with greater hepatic insulin action in Ad-GcR^−/− mice than in WT mice. Hepatic glycogen phosphorylase abundance was not different between genotypes (Supplemental Fig. S1C).

Figure 2.

Refeeding hepatic glycogen accumulation after an extended unfed period is accelerated in Ad-GcR^−/− mice. A) Hepatic glycogen concentrations. B) Hepatic Akt phosphorylation assessed by immunoblotting. Ratio of phosphorylated (p)–Akt:total Akt represented in graph. C) Hepatic glycogen synthase phosphorylation assessed by immunoblotting. Ratio of p–glycogen synthase:total glycogen synthase represented in graph. Basal: liver from mice with food withdrawn for 20 h. Post-GTT: liver from mice with food withdrawn for 20 h, status post–1-h OGTT (2 mg/g). B, basal; GcR^−/−, Ad-GcR^−/−; GS, glycogen synthase; PG, Post-GTT. Data are represented as means ± sem: n = 8/group [basal (A)]; 7–12/group [post-GTT (A)], or 6/group (B, C). *P < 0.05 vs. basal, **P < 0.01 vs. basal, ^#P < 0.05 vs. WT.

Basal parameters after 3 wk HFD

Taken together, attenuated fasting hepatic steatosis, reduced fasting plasma glucose, and disinhibited fasting-state glycogen synthesis in Ad-GcR^−/− mice are consistent with a critical link between adipose glucocorticoid function and physiologic lipid-induced hepatic insulin resistance. To further probe the connection between adipose glucocorticoid action and lipid-induced alterations in whole-body insulin action, we performed a series of studies in a well-characterized model of pathologic lipid-induced insulin resistance: the HFD-fed mouse. GcR deficiency in adipose tissue of 3-wk-old high-fat–fed mice did not alter body weight, fat mass, or muscle mass (Table 3). The Ad-GcR^−/− mice were not different from WT mice with respect to 6-h fasting plasma glucose, insulin, and NEFA concentrations (Table 3). In contrast, plasma triglyceride concentrations were significantly reduced in adipose Ad-GcR^−/− mice.

TABLE 3.

Basal parameters of 6-h unfed HFD-fed Ad-GcR^−/− mice

Measurement	WT	Ad-GcR−/−	P
Body weight	37.1 ± 1.8	35.1 ± 1.3	0.37
Fat mass (%)	28.6 ± 1.8	25.1 ± 1.6	0.17
Muscle mass (%)	58.7 ± 0.8	60.9 ± 1.1	0.13
Plasma glucose (mg/dl)	270 ± 11	261 ± 13	0.61
Plasma insulin (µU/ml)	25 ± 2	19 ± 3	0.12
Plasma triglyceride (mg/dl)	63 ± 4	48 ± 3	0.01
NEFA (mM)	0.47 ± 0.03	0.38 ± 0.04	0.12

Ad-GcR^−/− vs. littermate controls. Data are presented as means ± sem; n = 8–9/group.

Ad-GcR^−/− mice are protected from HFD-induced hepatic insulin resistance

Systemic glucocorticoids influence insulin action by a variety of mechanisms (42). However, the role of adipose glucocorticoid action in the development of insulin resistance is not clear, as prior metabolic studies in mice with disrupted adipose glucocorticoid action have had conflicting results (29–32). We assessed whole-body insulin action by hyperinsulinemic-euglycemic clamps in 3 wk HFD-fed mice. There was a 25% increase in the glucose infusion rate (GINF) required to maintain euglycemia (Fig. 3A, B). This was primarily attributed to an improvement in insulin-mediated suppression of endogenous glucose production (EGP) in HFD-fed Ad-GcR^−/− mice, indicating improved hepatic insulin responsiveness (Fig. 3C, D). In contrast, whole-body insulin–stimulated glucose disposal was not changed in HFD-fed Ad-GcR^−/− mice (Fig. 3E).

Figure 3.

Hepatic insulin sensitivity was improved in Ad-GcR^−/− mice. A) Plasma glucose concentrations during the hyperinsulinemic-euglycemic clamp. B) Glucose infusion rate (GINF) required to maintain euglycemia. C) EGP under basal (food withheld overnight) and hyperinsulinemic clamped conditions. D) Insulin-mediated suppression of EGP. E) Whole-body insulin–stimulated glucose disposal (Rd). GcR^−/−, Ad-GcR^−/−. Data are represented as means ± sem; n = 7–8/group. *P < 0.05, ***P < 0.001.

HFD-fed Ad-GcR^−/− mice are protected against DAG-induced hepatic insulin resistance

Hepatic steatosis is strongly associated with hepatic insulin resistance in both human subjects and rodent models of type 2 diabetes (39, 43, 44). As regular chow–fed Ad-GcR^−/− mice had decreased hepatic steatosis after a 20-h food withdrawal, we hypothesized that Ad-GcR^−/− mice may also be protected from HFD-induced hepatic steatosis. We assessed hepatic lipid accumulation in 3 wk HFD-fed adipose Ad-GcR^−/− mice. Adipose GcR^−/− mice had ∼30% lower hepatic triglycerides as compared with WT (Fig. 4A–C). Concentrations of the lipid metabolites DAG and ceramide, 2 important putative mediators of hepatic insulin resistance, were measured. Consistent with the triglyceride reduction seen, total hepatocellular DAG content was reduced in Ad-GcR^−/− mice, largely reflecting a decrease in plasma membrane DAG (Fig. 4D–F and Supplemental Fig. S2A). In contrast, total hepatic ceramide concentrations were not changed in Ad-GcR^−/− mice (Fig. 4G); furthermore, the C16 and C18 ceramide species that are thought to induce insulin resistance were also unchanged (Supplemental Fig. S2B).

Figure 4.

Ad-GcR^−/− mice are protected from hepatic lipid accumulation. Hepatic lipid studies performed in mice with food withheld for 6 h. A) Liver triglyceride (TG) content. B) WT liver histology with Oil Red O staining for lipid. C) Ad-GcR^−/− liver histology with Oil Red O staining for lipid. D) Liver total DAG content. E) Liver cytoplasmic DAG content. F) Liver plasma membrane DAG content. G) Liver ceramide content. GcR^−/−, Ad-GcR^−/−. Original magnification, ×40 (B, C). Data are represented as means ± sem; n = 7–9/group. *P < 0.05.

DAG accumulation causes hepatic insulin resistance by activating PKCε, which impairs insulin signaling by decreasing insulin receptor tyrosine kinase activity (25). HFD-fed adipose GcR^−/− mice displayed reduced hepatic PKCε membrane translocation (a measure of PKCε activation) in comparison with WT mice (Fig. 5A). Consistent with protection against DAG-PKCε–mediated insulin resistance, insulin-stimulated insulin receptor kinase (IRK) phosphorylation was greater in HFD-fed Ad-GcR^−/− mice than WT mice; insulin-stimulated Akt phosphorylation was increased in Ad-GcR^−/− mice as well (Fig. 5B, C).

Figure 5.

Ad-GcR^−/− mice are protected from lipid-mediated disruption of hepatic insulin signaling. A) PKCε translocation in livers of 6-h food-withdrawn mice. Sodium potassium ATPase was used as a housekeeping control in plasma membrane fraction; GAPDH was used as a housekeeping control in cytosolic fraction. Translocation expressed as ratio of ratios normalized to the WT membrane:cytosol ratio. B) Liver IRK tyrosine phosphorylation in basal (6-h food-withheld) and insulin-stimulated (hyperinsulinemic-euglycemic clamped) conditions quantitated as ratio of phosphorylated (p)–IRK:total IRK normalized to WT basal condition. C) Liver Akt phosphorylation in basal and insulin-stimulated conditions. Ratio of p-Akt:total Akt represented in graph. Data are normalized to WT basal. GcR^−/−, Ad-GcR^−/−; KO, knockout; phospho, phosphorylated. Data are represented as means ± sem; n = 7–9/group. *P < 0.05 vs. WT, ^†P < 0.05 vs. WT, ^##P < 0.01 vs. basal, ^###P < 0.001 vs. basal.

WAT lipolysis is reduced in Ad-GcR^−/− mice

An important mechanism by which the GcR in WAT may regulate whole-body insulin action and hepatic lipid content is through modulation of adipose tissue lipolysis (22, 23, 45). We quantified WAT lipolysis under both basal and hyperinsulinemic-euglycemic conditions. Differences in NEFA concentrations are an indirect measure of differences in WAT lipolysis, whereas fatty acid rate of appearance (equal to fatty acid flux at steady state) is a direct in vivo assessment of net WAT lipolysis (WAT lipolysis − WAT reesterification). Plasma NEFAs were significantly reduced in overnight-unfed basal-state HFD-fed Ad-GcR^−/− mice as compared with WT mice, and insulin-suppressed plasma NEFAs were also reduced (Fig. 6A). Furthermore, whole-body insulin–mediated suppression of fatty acid turnover was significantly improved in the knockout mice (Fig. 6A).

Figure 6.

Adipose lipolysis is attenuated in Ad-GcR^−/− mice. A) Left panel: plasma NEFAs under basal (food withheld overnight) and hyperinsulinemic clamped conditions. Right panel: whole-body fatty acid turnover expressed as fraction of WT basal fatty acid turnover. B) Glucocorticoid-regulated genes of adipose lipolysis as assessed by quantitative PCR. Expression is reported relative to the housekeeping gene TATA-binding protein. ANGPTL4, angiopoietin-like 4; MGL, monoglyceride lipase. C) Adipose tissue lipolytic enzyme (ATGL and HSL) protein abundance assessed by immunoblotting. Quantitation was performed relative to housekeeping protein (HSP90) abundance. D) Basal and insulin-stimulated perilipin phosphorylation. GcR^−/−, Ad-GcR^−/−; KO, knockout; p-Perilipin, phosphorylated Perilipin. Data are represented as means ± sem; n = 6–8/group. *P < 0.05 vs. WT, ****P < 0.0001 vs. WT, ^##P < 0.01 vs. basal.

The expression of ATGL and HSL were assessed in epidydimal adipose tissue at both the mRNA and protein levels. Both ATGL gene expression and ATGL protein abundance were reduced in Ad-GcR^−/− mice as compared with WT (Fig. 2B). Although HSL is known to be regulated by GcR activation, neither HSL gene expression, nor HSL protein abundance, were altered by the loss of the GcR in the WAT (Fig. 6B, C).

Phosphorylation of the lipid droplet coat protein perilipin by PKA facilitates the translocation of intracellular lipases to the lipid droplet to initiate lipolysis (46). Insulin-mediated perilipin dephosphorylation is amplified in adipose GcR^−/− mice as compared with WT (Fig. 6D), consistent with an improvement in nongenomic regulation of lipolysis in Ad-GcR^−/− mice.

Food intake decreased in Ad-GcR^−/− mice

Adipose tissue participates in the humoral control of food intake and energy expenditure through the secretion of adipokines and cytokines such as leptin and bone morphogenetic protein 7 (6). Hence, food intake and energy expenditure assessed by indirect calorimetry were both measured in metabolic cages (Fig. 7A–D). Food intake and energy expenditure were decreased by knockout of the adipose tissue GcR. Food intake was decreased in adipose GcR^−/− mice by a mean of just over 3 kcal/kg/h (73 kcal/kg/d), whereas energy expenditure was decreased by a mean of 1 kcal/kg/h (29 kcal/kg/d). In the physiologic range leptin can suppress appetite, and we observed an increase in plasma leptin concentration by ∼70% in Ad-GcR^−/− mice compared with the WT mice (Fig. 7E). Circulating glucocorticoids may directly stimulate appetite, raising the question of whether altered glucocorticoid action in adipose tissue alters the hypothalamic-pituitary-adrenal axis sufficiently to alter circulating corticosterone. Thus, 24-h urinary corticosterone concentrations were measured, and no difference was seen between the 2 genotypes (Supplemental Fig. S3).

Figure 7.

Food intake and energy expenditure is reduced in Ad-GcR^−/− mice. A) Food intake in metabolic cages. B) Energy expenditure in metabolic cages. C) Hourly food intake. D) Hourly energy expenditure. E) Fed-state early morning leptin levels. GcR^−/−, Ad-GcR^−/−. Data are expressed as means ± sem; n = 9–15/group. *P < 0.05, **P < 0.01.

DISCUSSION

Circulating glucocorticoids increase plasma glucose by coordinating decreased peripheral glucose uptake, increased hepatic gluconeogenic capacity, and increased lipolysis (47). Lipolysis may be critical to provide glycerol as a gluconeogenic substrate and to provide fatty acids, which can both provide energy for gluconeogenesis and increase allosteric activation of gluconeogenesis by increasing hepatic acetyl Coenzyme A. In this study, the combination of findings of reduced fasting plasma NEFA, attenuated fasting hepatic steatosis, and reduced fasting β-hydroxybutyrate concentrations in Ad-GcR^−/− mice suggest that adipose tissue glucocorticoid action is in large part responsible for coordinating several of the physiologic responses to extended fasting. Adipose glucocorticoid action, driving increased fatty acid flux, leads to increased ketone body production and increased hepatic steatosis.

What is the role of this adipose glucocorticoid action–driven physiologic hepatic steatosis? These studies suggest the development of hepatic insulin resistance may serve to inhibit mealtime glycogen synthesis during periods of food scarcity. Normal, healthy animals demonstrate blunted glycogen synthesis upon refeeding after an unfed period (48). Increased hepatic glycogen accretion and reduced hepatic triglyceride accumulation in the refed state of Ad-GcR^−/− mice are consistent with an important role for lipid-induced hepatic insulin resistance in the prevention of inappropriate glucose storage after the stress of an extended fast. Overall, this study supports a model in which adipose glucocorticoid action initiates defense mechanisms against inappropriate energy storage during food scarcity: increased fasting glucocorticoids drive physiologic hepatic steatosis and hepatic insulin resistance, thereby inhibiting hepatic glycogen synthesis. In concert with this, increased fatty acid flux to the liver drives increased ketone body production, providing another substrate source that can further spare glucose utilization.

These same pathways that help to defend plasma glucose concentrations under physiologic stress when chronically or excessively activated might be predicted to contribute to the development of hepatic steatosis and insulin resistance. Consistent with this idea, prior studies in mice with adipose-specific augmentation (27) or attenuation (49) of glucocorticoid action have demonstrated significant changes in whole-body glucose and lipid metabolism. Additionally, in aged and high-fat–fed mice loss of the adipose GcR prevented hepatic steatosis and improved hepatic insulin signaling (29). However, metabolic analyses from this study were confounded by a divergence in body weight with aging and high-fat feeding: The aged and high-fat–fed Ad-GcR^−/− mice were studied at a lower body weight than the WT controls, and the lower body weight may have accounted for the improvements seen in hepatic metabolism. Indeed, several studies from other groups showed that loss of the adipose tissue GcR did not alter the impact of chronic HFD or high-fructose diet on whole-body metabolism (30–32). To determine whether adipose glucocorticoid action plays an important role in the deposition of ectopic lipid in dietary lipid–induced insulin resistance, we studied HFD-fed Ad-GcR^−/− mice and HFD-fed WT littermate controls. It is important to note that because Ad-GcR^−/− mice have been shown to have reduced HFD-induced weight gain (29) and body weight divergence can confound analyses of insulin sensitivity, these studies were performed in animals prior to body weight divergence between the genotypes.

HFD-induced hepatic lipid accumulation was reduced and lipid-induced insulin resistance was prevented in Ad-GcR^−/− mice as compared with WT. Assessment of hepatic bioactive lipid species and hepatic insulin signaling demonstrate that DAG-PKCε–induced insulin resistance was reduced in these animals. Although prior reports of glucose tolerance in the setting of ablation of adipose glucocorticoid action are conflicting, using hyperinsulinemic-euglycemic clamps, we saw that reduction in physiologic adipose glucocorticoid action specifically protects against hepatic lipid–induced insulin resistance. These findings suggest that adipose glucocorticoid action exerts impact on the development of hepatic steatosis under both physiologic and pathologic conditions.

In vitro, activation of the GcR can drive differentiation of preadipocytes and can increase lipolytic enzymes. In our in vivo investigation, we did not observe a significant difference in adipose tissue by ¹H magnetic resonance spectroscopy between genotypes. In contrast, a critical difference between HFD-fed WT mice and Ad-GcR^−/− mice was the difference in insulin-suppressed rates of WAT lipolysis. Transcriptional regulation of WAT lipolytic genes by the GcR is well described; HSL is directly under the control of a glucocorticoid response elements, whereas ATGL expression is indirectly controlled by GcR activation (16, 21, 50). In the Ad-GcR^−/− model, in which adipose glucocorticoid action is ablated, ATGL gene expression and protein abundance were significantly decreased, whereas HSL was unaffected. Furthermore, glucocorticoids increase adipocyte cAMP content through reduced phosphodiesterase 3B (21), thereby dampening insulin-mediated suppression of lipolysis. We observed increased insulin-stimulated perilipin dephosphorylation, consistent with a reduction in adipocyte cAMP tone. Maintenance of normal HSL levels in Ad-GcR^−/− suggests a degree of compensation by the WAT to ablated glucocorticoid action, but the effects on ATGL and perilipin dephosphorylation are enough to decrease WAT lipolysis, leading to reduced plasma NEFA and enhanced insulin-suppressed fatty acid turnover.

Excess circulating glucocorticoids are associated with increased appetite in humans, increasing the central regulation of appetite (51). We studied mice prior to any age-associated divergence in body weight (weight-matched mice) in order to avoid the confounding effects body weight difference would have on metabolic studies. However, even in this weight-matched cohort, we observed a reduction in food intake. Decreased dietary fat consumption could help to explain the observed reduction in hepatic lipid. Additionally, this would perhaps predict a decrease in body weight in older mice, consistent with a prior report in a different strain of Ad-GcR^−/− mice (29). The reduction in food intake in adipose GcR^−/− mice implies the existence of an alteration to the humoral regulation of appetite. Leptin secretion is a mechanism by which adipose tissue can communicate with the CNS, one of many inputs that help to regulate feeding behaviors. Because leptin secretion is induced by glucocorticoids, usually a positive correlation between leptin and glucocorticoids is observed (11, 52, 53). However, there are examples of leptin levels failing to correlate with the glucocorticoid stimulus (12, 54). The increased fed-state leptin in Ad-GcR^−/− mice is consistent with altered adipocytokine regulation contributing to the observed metabolic phenotype.

Our findings suggest that the role of adipose glucocorticoid action in the regulation of whole-body metabolism is a modulatory role. Under most normal physiologic circumstances, the impact of adipose glucocorticoid action on insulin action may be too subtle to detect. The impact of knocking out the adipose GcR is only revealed under specific circumstances, such as prolonged unfed periods or (in HFD-fed mice) using a technique like the hyperinsulinemic-euglycemic clamp that can specifically measure changes in hepatic insulin action. This would explain the lack of consensus in prior work with Ad-GcR^−/− mice in studies that relied on techniques that cannot reliably assess hepatic insulin action (29–32).

One caveat that applies when genetically manipulating adipose tissue is that there is no available model that exclusively targets white adipocytes. The Adipoq-cre used in our study not only efficiently drives Cre recombinase expression in white adipocytes, it also also drives Cre recombinase in brown adipocytes (33). An increase in brown adipose tissue function could certainly explain protection against hepatic steatosis through increased fatty acid oxidation. However, we did not observe an increase in energy expenditure in Ad-GcR^−/− mice, and thus it is unlikely that the knockout of the GcR from brown adipocytes is contributing to the observed reduction in either unfed- or high-fat feeding–induced hepatic steatosis.

In summary, the present study addresses the mechanisms by which WAT glucocorticoid action serves to alter whole-body glucose and lipid metabolism. Deletion of the GcR in WAT interferes with the development of lipid-induced hepatic insulin resistance: both preventing the development of physiologic fasting steatosis with a corresponding reduction in fasting glucose levels and increase in refeeding-induced glycogen accumulation, and it prevents the development of pathophysiological HFD-induced DAG-PKCε–mediated hepatic insulin resistance. The changes in WAT physiology that can explain this protection against hepatic lipid accumulation include improved insulin suppression of WAT lipolysis, and reduced food intake out of proportion to an accompanying reduction in energy expenditure. Both genomic (ATGL expression) and nongenomic (perilipin phosphorylation) regulation of WAT lipolysis were altered by the knockout of the WAT GcR, and changes in adipocytokines may explain the alterations in food intake. Hence, the key mechanisms by which adipose glucocorticoid action modulate whole-animal physiology are through regulation of WAT lipolysis and regulation of energy consumption.

Supplementary Material

This article includes supplemental data. Please visit http://www.fasebj.org to obtain this information.

Glossary

Ad-GcR^−/−

adipose tissue–specific GcR knockout

Akt

protein kinase B

ATGL

adipose triglyceride lipase

DAG

diacylglycerol

EGP

endogenous glucose production

GAPDH

glyceraldehyde-3-phosphate dehydrogenase

GcR

glucocorticoid receptor

GTT

glucose tolerance test

HFD

high-fat diet

HSL

hormone-sensitive lipase

NEFA

nonesterified fatty acid

OGTT

oral GTT

WAT

white adipose tissue

WT

wild type

AUTHOR CONTRIBUTIONS

A. Abulizi conceived the study, performed experiments, analyzed data, supervised the study, and wrote and edited the manuscript; J.-P. Camporez performed and supervised experiments and analyzed data; M. J. Jurczak planned and supervised experiments and edited the manuscript; K. F. Høyer and D. Zhang performed experiments; G. W. Cline developed methodology and performed experiments; V. T. Samuel and G. I. Shulman conceived the study, supervised the study, and edited the manuscript; D. F. Vatner conceived the study, performed experiments, analyzed data, supervised the study, and edited the manuscript; and all authors reviewed the manuscript and approved the final version.