Overexpression of Suppressor of Cytokine Signaling 3 in Adipose Tissue Causes Local but Not Systemic Insulin Resistance

Rabbit SOCS3 antiserum was generated as previously described (8). Rabbit polyclonal anti–insulin receptor, –IRS-1, and –IRS-2; mouse monoclonal anti–phosphotyrosine 4G10; and polyclonal anti-p85 antibodies were kindly provided by Dr. Ronald Kahn’s laboratory (Joslin Diabetes Center, Boston, MA) or purchased from Upstate (Lake Placid, NY). Protein A agarose, rabbit polyclonal anti-hemagglutinin (HA) antibody, and goat polyclonal anti-SOCS3 antibody was purchased from Santa Cruz Biotech (Santa Cruz, CA). NADH, NADPH, acetyl-CoA, and malonyl-CoA were all purchased from Sigma (St. Louis, MO). ¹⁴C-deoxy-glucose was from Amersham Pharmacia Biotech (Piscataway, NJ).

The 5.4-kb aP2 promoter/enhancer was used to drive fat-specific expression of the mouse SOCS3 transgene (13). Mouse SOCS3 cDNA with a 3′ sequence encoding two HA tags and a sequence for the bovine GH polyadenylation signal were derived by PCR, using pcDNA 3.0 SOCS3 expression vector (7) as a template with a 5′ primer sequence including an SmaI site and a Kozak consensus sequence, paired with a 3′ primer sequence including an SalI site and a stop codon. The PCR product was ligated into the SmaI and SalI sites of the 5.4-kb aP2 promoter. The aP2-SOCS3-HA construct was linearized, using NotI and SalI digestion and injected into pronuclei of fertilized eggs from FVB and C57BL/6 mice. Two independent lines with SOCS3 overexpression in adipose tissue were obtained.

This study was approved by the institutional animal care and use committee of the Beth Israel Deaconess Medical Center. Animals were weaned at age 3 weeks and housed individually in our animal facilities on a 12-h light/dark cycle. We placed 5-week-old aP2-SOCS3 transgenic mice and their nontransgenic littermates on a high- fat/high-sucrose diet (TD-88137; Harlan Tekland, Madison, WI) or standard chow diet with water ad libitum, and body weights and food intake were measured weekly. Resting metabolic rate and locomotor activity were measured, using a comprehensive laboratory animal monitoring system (CLAMS; Columbus Instruments, Columbus, OH) (14). Glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs) were performed as previously described (8). At the end of study, all mice were killed by CO₂, and blood was obtained via cardiac puncture for insulin, leptin, adiponectin, TNF-α, and other measurements of metabolic parameters. Fat pads were dissected, immediately weighed, frozen in liquid nitrogen, and stored at −80°C. For adipose tissue histology, subcutaneous fat pads were fixed in Bouin’s solution (Sigma), imbedded in paraffin for multiple sectioning, and stained with hematoxylin/eosin. The average area size of adipocytes was measured, using ImageJ image analysis software (available from http://rsb.info.nih.gov/ij/).

Blood glucose was measured with a OneTouch Ultra glucose meter (Lifescan, Mulpitas, CA). Serum insulin and leptin levels were measured, using rat insulin enzyme-linked immunosorbent assay (ELISA) and mouse leptin ELISA kits (Crystal Chem, Downers Grove, IL), respectively. Serum adiponectin and TNF-α levels were determined, using a mouse adiponectin ELISA kit (Linco Research, St. Charles, MO) and a mouse TNF-α ELISA kit (R&D Systems, Minneapolis, MN). Serum triglycerides, free fatty acids, and cholesterol levels were measured, using an L-type TG H kit, NEFA C kit, and Cholesterol E kit (Wako Chemicals, Richmond, VA), respectively.

Fatty acid synthase (FAS) and glycerol-3-phosphate dehydrogenase (GPDH) activity was determined spectrophotometrically in crude cytosolic extracts of mouse adipose tissue (15). Lipolysis assay was conducted as previously described (16).

Adipocyte total RNA was extracted, using an RNeasy Mini Kit (Qiagen, Valencia, CA). Mouse adipocyte mRNA expression was quantitatively measured, using an Mx 4000 Mutiplex quantitative PCR system (Stratagene, Cedar Creek, TX) with a Brilliant Single-Step quantitative RT-PCR kit (Stratagene), as previously described (17). The sequence of primers and probes for the mouse adipocyte genes will be provided on request.

For regular RT-PCR detection of the SOCS3 transgene, total RNA was reverse transcribed to the first-strand cDNA with a primer designed in the HA-tag region, and it was then amplified by PCR.

For ex vivo insulin signaling experiments, epididymal fat pads were dissected, minced, and cultured in DMEM medium containing 1% fetal bovine serum and 1% BSA for a 4-h recovery, and then they were incubated in Krebs-Ringer bicarbonate buffer containing 1% BSA with or without 100 nmol/l insulin for 5 min. Adipose tissue was then harvested for immunoprecipitation and immunoblotting analysis of insulin signaling protein, as previously described (12).

In vivo insulin signaling experiments were performed on mice after a 16-h fast. Mice were intravenously injected with 10 units/kg body wt of human insulin (Eli Lilly) or saline. At 5 min after injection, epididymal fat pads were dissected and frozen in liquid nitrogen for immunoprecipitation and immunoblotting analysis of insulin signaling proteins.

Adipocytes were isolated from epididymal fat pads by collagenase digestion (15), and glucose transport assays were conducted as previously described (18).

All data are the means ± SE. Data were evaluated for statistical significance by one-way ANOVA, and significantly different group means were then separated by the least significant difference test, using SPSS (Chicago, IL). P < 0.05 is considered significant.

We generated a transgenic mouse (aP2-SOCS3 mouse) overexpressing SOCS3 specifically in adipose tissue under control of the 5.4-kb aP2 promoter/enhancer. Two independent lines overexpressing the SOCS3 transgene were obtained. aP2-SOCS3 mice were born at the expected Mendelian frequency and showed normal behavior and fertility. All mice used in this study were hemizygous for the SOCS3 transgene or were their nontransgenic littermates on the inbred FVB strains.

To assess the overexpression of the SOCS3 transgene in adipose tissue, we measured the SOCS3 mRNA by real-time RT-PCR. Figure 1A shows that SOCS3 mRNA expression was increased in epididymal, subcutaneous, and brown fat by >10-fold, whereas SOCS3 mRNA was increased to a lesser extent in mesenteric and retroperitoneal fat. Similar results were observed when SOCS3 protein levels were measured by immunoblotting (Fig. 1B). Using an anti-SOCS3 antibody, the SOCS3 protein was barely detectable in epididymal and subcutaneous fat and not detectable in mesenteric fat of nontransgenic littermates. However, the SOCS3 protein was significantly increased in epididymal and subcutaneous fat and detectable in mesenteric fat of aP2-SOCS3 mice (Fig. 1B, lower panel). Using an anti-HA tag antibody, the transgene-encoded protein was detectable in fat depots of transgenic mice but not in their nontransgenic littermates (Fig. 1B, upper panel). Moreover, using RT-PCR, we detected SOCS3 transgene mRNA exclusively in white and brown adipose tissues at 25 cycles of PCR (Fig. 1C).

We first monitored the body weight development of transgenic and nontransgenic littermates fed either a standard chow or high-fat/high-sucrose diet. Body weights of aP2-SOCS3 and nontransgenic mice on a chow diet slightly diverged after 10 weeks, with aP2-SOCS3 mice gaining less weight compared with nontransgenic littermates (Fig. 2A). When challenged with a high-fat/high-sucrose diet, the aP2-SOCS3 mice exhibited resistance to diet-induced weight gain, with an 11% reduction compared with nontransgenic littermates at age 18 weeks (Fig. 2A). Consistent with the body weight, decreased fat mass was also observed in aP2-SOCS3 mice. Although transgenic mice on a chow diet showed a slightly decreased fat mass, aP2-SOCS3 mice on a high-fat diet exhibited a significantly decreased mass of epididymal, mesenteric, and subcutaneous fat, with a 40% reduction of total fat mass, compared with nontransgenic littermates (Fig. 2B and C). In addition, histological examination of white adipose tissue revealed that fat tissue from aP2-SOCS3 mice on chow or high-fat diets contained smaller adipocytes than their nontransgenic littermates (Fig. 2D). These data show that aP2-SOCS3 mice are resistant to diet-induced obesity.

We also measured food intake and parameters of energy expenditure in these aP2-SOCS3 mice. On a chow diet, cumulative food intake over 40 h was decreased by 29% in transgenic mice compared with wild-type littermates (Fig. 3A). Locomotor activity was monitored during a 24-h light/dark cycle. aP2-SOCS3 mice exhibited a tendency toward increased nocturnal activity, although the increase was not statistically significant (Fig. 3B). The metabolic rate was determined by measuring oxygen consumption and carbon dioxide production. Analysis of oxygen consumption revealed no alteration in transgenic animals (Fig. 3C). However, the transgenic mice showed a trend toward decreased respiratory quotient (respiratory exchange ratio) compared with wild-type mice, although this decrease did not reach statistical significance (Fig. 3D).

We assessed the blood metabolic profile of aP2-SOCS3 mice. Serum triglycerides were decreased by 50% in transgenic mice on a high-fat diet, although free fatty acids and cholesterol levels were unchanged in these mice (Table 1). Consistent with the decreased fat mass, transgenic mice on a high-fat diet showed lower serum leptin levels (Table 1). Although serum TNF-α did not change, serum adiponectin levels were increased by 80% in transgenic mice on a high-fat diet compared with nontransgenic littermate controls (Table 1). These data suggest that along with resistance to a high-fat diet, aP2-SOCS3 mice show improved metabolic parameters.

Although aP2-SOCS3 mice on a chow diet did not show any alterations in fed or fasting glucose and insulin levels, transgenic mice on a high-fat diet exhibited lower fasting glucose and insulin levels and lower fed insulin levels compared with nontransgenic mice (Table 1). These data suggest that aP2-SOCS3 mice are protected against insulin resistance associated with diet-induced obesity. To further confirm this, we performed glucose and ITTs on mice fed chow or high-fat diets. GTTs and ITTs showed no difference between aP2-SOCS3 and nontransgenic mice on a chow diet (Fig. 4A and B). However, the hyperglycemic response to intraperitoneal injection of glucose during the GTT was attenuated in transgenic mice on a high-fat diet compared with controls (Fig. 4C). Consistent with this finding, the hypoglycemic response to intraperitoneal injection of insulin during an ITT was greater in transgenic mice on a high-fat diet (Fig. 4D). Therefore, along with resistance to diet-induced obesity, aP2-SOCS3 mice are protected against insulin resistance associated with diet-induced obesity.

To determine the physiological consequence of adipocyte-specific overexpression of SOCS3 on adipocyte insulin action, we examined adipocyte insulin signaling in vivo. Mice were fasted overnight and injected with human insulin (10 units/kg body wt i.v.) or saline. As shown in Fig. 5A, insulin-stimulated phosphorylation of the insulin receptor did not change in adipose tissue of aP2-SOCS3 mice. Although insulin injection markedly stimulated phosphorylation of IRS-1 in adipose tissue of nontransgenic mice, this signaling event was attenuated in aP2-SOCS3 mice (Fig. 5C, upper panels). The decreased insulin-stimulated phosphorylation of IRS-1 in transgenic mice was accompanied by reduced IRS-1 protein levels (Fig. 5B). Quantitation of the blots showed a 50% reduction of both IRS-1 phosphorylation and protein levels in adipose tissue of aP2-SOCS3 mice (Fig. 5B and C), suggesting that the IRS-1 degradation caused by SOCS3 overexpression may account for the decreased IRS-1 phosphorylation and impaired insulin signaling in adipose tissue of aP2-SOCS3 transgenic mice. This is consistent with data we and others have observed that SOCS3 targets IRS proteins for proteasomal degradation (11,12). Similar results were observed in IRS-2 phosphorylation and protein levels (Fig. 5D).

To determine whether the defect of insulin signaling caused by overexpression of SOCS3 in adipocytes is intrinsic to adipose tissue or caused by the systemic metabolic milieu, we examined insulin signaling in primary-cultured adipose tissue derived from aP2-SOCS3 mice and nontransgenic controls. Figure 6A shows that insulin-stimulated phosphorylation of insulin receptor was slightly decreased in adipose tissue of transgenic mice. Moreover, insulin-stimulated phosphorylation of IRS-1 was decreased in transgenic mice, with a correspondingly decreased IRS-1 protein level (Fig. 6B). Furthermore, insulin-stimulated association of p85 with IRS-1 was reduced in transgenic mice (Fig. 6D). A similar result was observed in insulin-stimulated phosphorylation of IRS-2 (Fig. 6C). To determine the physiological consequence of this impaired insulin signaling on downstream insulin actions, we measured insulin-stimulated glucose uptake in isolated adipocytes from aP2-SOCS3 mice and nontransgenic controls. Insulin-stimulated glucose uptake was markedly decreased in isolated adipocytes of transgenic mice at different doses of insulin from 0.1–100 nmol/l, although basal glucose transport was not altered in transgenic mice (Fig. 6E).

To investigate the physiological consequence of impaired insulin signaling on adipocyte lipid metabolism, we measured the activity of two lipogenic enzymes, FAS and GPDH. The expression and activity of FAS, a key enzyme involved in de novo lipogenesis, have been demonstrated to be upregulated by insulin signaling (19). Figure 7A and B shows that FAS and GPDH activity were decreased by 40–50% in transgenic mice on a high-fat diet compared with nontransgenic controls. On the other hand, because insulin signaling has an antilipolytic effect, we determined whether the impaired insulin signaling caused by SOCS3 overexpression would attenuate insulin’s antilipolytic action in the adipose tissue of transgenic mice. Figure 7C shows that although isoproterenol stimulated lipolysis by 3.5-fold in adipose tissue from nontransgenic mice on a high-fat diet, pretreatment with 100 nmol/l insulin inhibited isoproterenol-stimulated lipolysis by 40%. However, pretreatment of adipose tissue from aP2-SOCS3 mice with 100 nmol/l insulin was not able to inhibit isoproterenol-stimulated lipolysis. Similar results were observed in mice fed a chow diet (Fig. 7D). These data suggest that the impaired insulin signaling caused by SOCS3 overexpression inhibits de novo lipogenesis and blocks insulin’s antilipolytic effects, and these changes may contribute to impaired energy partitioning and decreased fat mass in aP2-SOCS3 mice.

Using real-time RT-PCR, we measured the expression of genes involved in adipocyte differentiation, glucose and lipid metabolism, and secretory factors in white adipose tissue (Table 2). Expression of several genes whose expression is characteristic of mature adipocytes such as peroxisome proliferator–activated receptor (PPAR)γ, aP2, and GLUT-4 did not change in aP2-SOCS3 mice, suggesting that adipose tissue development was not affected by overexpression of SOCS3. FAS and sterol response element–binding protein 1c (SREBP1c) mRNA, encoding two genes involved in lipogenesis, were decreased in white adipose tissue of transgenic mice. This is consistent with the decreased FAS and GPDH activity in adipose tissue of aP2-SOCS3 mice and may contribute to the decreased de novo lipogenesis in transgenic mice. In agreement with the serum adiponectin protein level, adiponectin mRNA was also increased in white fat of transgenic mice. Increased adiponectin protein levels may increase the insulin sensitivity of skeletal muscles and liver and contribute to increased systemic insulin sensitivity.

In addition, we measured the expression of genes that act on fatty acid oxidation and utilization in skeletal muscle. The mRNA levels of CPT1 (carnitine palmitoyl transferase-1), ACO (fatty acyl-CoA oxidase), and PPARα were increased in skeletal muscle of aP2-SOCS3 mice on a high-fat diet. The increased expression of fatty acid oxidative genes in skeletal muscle may be caused by the increased adiponectin levels in fat and circulation.

We have previously observed in vitro that SOCS3 deficiency increases insulin-stimulated IRS-1 and -2 phosphorylation and enhances downstream phosphatidylinositol 3 kinase activity, leading to increased insulin-stimulated glucose uptake in adipocytes. Moreover, lack of SOCS3 blocks TNF-α’s inhibition of insulin signaling in adipocytes, which is largely attributed to the suppression of TNF-α–induced IRS-1 and -2 protein degradation (12). We therefore hypothesized that adipocyte-specific overexpression of SOCS3 would cause impaired insulin signaling in adipose tissue in vivo. To test this hypothesis, we generated transgenic mice that overexpress SOCS3 specifically in adipose tissue. We demonstrate that overexpression of SOCS3 in adipocytes decreases IRS protein levels and insulin-stimulated IRS phosphorylation and causes decreased insulin-stimulated glucose uptake in adipocytes. This is consistent with data we and others have previously observed in vitro (11,12). Therefore, overexpression of SOCS3 in adipocytes causes insulin resistance in adipose tissue in vivo.

Although adipocyte-specific overexpression of SOCS3 induces insulin resistance in adipose tissue, whole-body glucose homeostasis and systemic insulin sensitivity are not impaired. Instead, aP2-SOCS3 mice have improved systemic insulin sensitivity and reduced adiposity when exposed to a high-fat diet. How can this paradox be explained? aP2-SOCS3 mice have reduced adipocyte cell size and total body fat mass. It has been recognized for years that excessive fat (obesity) is causally associated with insulin resistance and is the leading risk for the development of type 2 diabetes (20,21,22). Likewise, leanness is generally associated with enhanced systemic insulin sensitivity. There are several interesting examples wherein leanness is associated with increased systemic insulin actions, despite insulin resistance within the adipocyte. The most striking example is the adipose-specific insulin receptor knockout (FIRKO) mouse (23). These mice have reduced fat mass and are protected against systemic insulin resistance associated with obesity, despite a complete loss of insulin action and abolished insulin-stimulated glucose uptake in adipocytes (23). A similar phenotype has also been observed in a transgenic mouse overexpressing the transmembrane TNF-α specifically in adipose tissue (18). These transgenic mice have decreased adipocyte cell size and whole-body adipose mass. Although overexpression of transmembrane TNF-α in adipocytes causes local insulin resistance, these mice show a tendency to improved systemic insulin sensitivity in GTT and ITT studies (18). Another example is the heterozygous PPARγ-deficient mouse model, which shows protection against high-fat diet–induced adipocyte hypertrophy and systemic insulin resistance, although PPARγ deficiency is expected to decrease insulin sensitivity (24). Taken together, these studies support a notion that a moderate reduction in adipocyte size and fat mass can, under certain circumstances, exert a dominant effect on systemic insulin sensitivity, overcoming local adipocyte insulin resistance and decreased glucose uptake in adipose tissue.

Indeed, regarding glucose uptake in vivo, adipose tissue only accounts for ∼10% of whole-body glucose uptake, whereas muscle accounts for 80–90% (1). Therefore, insulin resistance in adipose tissue may not necessarily cause systemic insulin resistance. In contrast, molecules secreted by adipose tissue may play a more important role in systemic insulin sensitivity than adipose tissue’s direct capability for glucose disposal. These secretory molecules can act on muscle and liver to either enhance or decrease glucose uptake, and they thereby override the capacity for glucose uptake by adipose tissue itself. Adipocyte size and fat mass may determine the levels of molecule secretion, and, as we have stated above, a moderate reduction in adipocyte size and fat mass may improve systemic insulin sensitivity by increasing the secretion of molecules such as adiponectin.

How might adipocyte insulin resistance induced by SOCS3 confer protection against adipocyte hypertrophy? Adipocyte hypertrophy characterized by increased storage of triglyceride may result from increased de novo lipogenesis and/or decreased lipolysis. Insulin, a potent lipogenic and antilipolytic hormone, plays a crucial role in triglyceride storage and adipocyte hypertrophy through coordinated control over lipogenesis and lipolysis. Insulin stimulates lipogenesis by inducing the expression and activity of key lipogenic enzymes, including FAS (25). Moreover, insulin stimulation of FAS is mediated by SREBP1c (26). Indeed, one consequence of impaired insulin signaling in adipocytes of aP2-SOCS3 mice is the downregulation of FAS expression and activity as well as reduced SREBP1c expression. On the other hand, insulin inhibits lipolysis by activating phosphodiesterase 3B, mediated by the downstream insulin signal protein kinase B (27). We also observed an attenuation of the antilipolytic action of insulin in adipose tissue of aP2-SOCS3 mice. Therefore, decreased lipogenesis coupled with increased lipolysis, both caused by impaired insulin signaling, may contribute to the protection against adipocyte hypertrophy in aP2-SOCS3 mice on a high-fat diet. These data indicate that insulin signaling in adipose tissue is permissive for increased energy storage in adipose tissue in response to high-fat diet and that SOCS3 expression in fat can resist energy storage.

The limited flow of energy to adipose tissue in aP2-SOCS3 mice may therefore regulate whole-body energy homeostasis through a feedback circuit. Indeed, our data show that food intake is decreased in aP2-SOCS3 mice, whereas physical activity is slightly increased in these transgenic mice. This suggests that energy equilibrium is reestablished in these transgenic mice to accommodate the restricted deposition of energy in adipocytes. Although the mechanisms underlying this deficit of energy balance are not entirely clear, the alteration of adipocyte insulin signaling and lipid metabolism caused by SOCS3 overexpression in adipocytes most likely affects systemic energy balance through molecules secreted by adipocytes.

Adipocyte hypertrophy alters the metabolic and endocrine function of adipose tissue and results in abnormal secretion of hormones and cytokines, one of which is the adipocyte-derived hormone adiponectin. Adiponectin has been shown to increase systemic insulin sensitivity by increasing fatty acid oxidation and reducing triglyceride content in muscle as well as by inhibiting gluconeogenesis in liver (28,29). Expression and circulating levels of adiponectin correlate inversely with obesity and insulin resistance (30). Indeed, aP2-SOCS3 mice have elevated adiponectin expression in adipose tissue and a corresponding increase in serum adiponectin levels. Increased adiponectin levels may act on muscle and liver to enhance insulin action in these tissues and may therefore account, at least in part, for the improved systemic insulin sensitivity of aP2-SOCS3 mice on a high-fat diet, despite the insulin-resistant glucose uptake in adipocytes.

In conclusion, SOCS3 expression is induced in adipose tissue in obesity, and here we have used a transgenic approach to clarify the biological consequence of this overexpression. SOCS3 overexpression in fat suppresses several elements of insulin signaling in fat, suppresses insulin-stimulated glucose uptake in fat, and reduces fat storage, at least in part, by suppressing lipogenesis and increasing lipolysis. aP2-SOCS3 mice resist obesity induced by high-fat diets, and, despite insulin-resistant glucose uptake in adipocytes, systemic glucose tolerance is enhanced, as is systemic insulin sensitivity. The latter could be attributable to increased adipocyte expression and circulating levels of adiponectin.

This study was supported by grants to J.S.F. from the National Institutes of Diabetes and Digestive and Kidney Diseases (grant R37 28082) and Takeda Pharmaceuticals.