Insulin and Leptin Resistance With Hyperleptinemia in Mice Lacking Androgen Receptor

All animal procedures were approved by the animal care and use committee of the University of Rochester School of Medicine, in accordance with National Institutes of Health guidelines. Construction of targeting vectors and generation of the chimera founder mice have been described previously (15). The strains of the mosaic founder mice were C57BL/6 and 129Sv background. β-Actin is a housekeeping gene and is universally expressed in every tissue; therefore, the β-actin promoter–driven Cre (ACTB-Cre; Jackson Laboratories, Bar Harbor, ME) will express and delete floxed AR fragments in all of the tissues. The AR^−/y mice were genotyped by PCR, as described previously (15). Animals were housed in pathogen-free facilities, maintained on a 12-h light/dark schedule (light on at 0600), and had free access to standard laboratory chow (no. 5010; PMI Lab Diet, St. Louis, MO) and water.

Tissues were fixed in 10% neutral buffered formalin and embedded in paraffin. Nonadjacent sections, separated by 70–80 μm, were obtained from perigonadal fat pads and analyzed systematically with respect to adipocyte size and number. Staining of the sections was performed with hematoxylin/eosin. Images were acquired using an E800 microscope (Nikon, Melville, NY) and a SPOT camera (Diagnostic Instruments, Sterling Heights, MI) and were analyzed using SigmaScan Pro software (version 5.0; SPSS, Chicago, IL).

Fasting blood samples were taken from mice 14 h after withdrawal of food. Blood samples designated as random-fed state were drawn 6 h after introducing food into the cages of mice that had been subjected to a preceding 14-h fast. Blood glucose concentrations were measured using a glucometer (One Touch Ultra; Lifescan, Milpitas, CA). Insulin levels and leptin levels were determined in duplicate 5-μl serum samples, using a mouse insulin and leptin enzyme-linked immunosorbent assay kit (Crystal Chem, Downers Grove, IL) according to the manufacturer’s protocol. Serum tumor necrosis factor-α (TNF-α) levels were determined in duplicate in 20-μl serum samples using a mouse TNF-α enzyme-linked immunosorbent assay kit (eBioscience, San Diego, CA) according to the manufacturer’s protocol. For the glucose tolerance test (GTT), after a 14-h fast, mice were given an oral bolus of d-glucose (2 g/kg body wt), and the blood glucose concentration was measured in samples taken at 0, 30, 60, 90, and 120 min after the glucose bolus. Insulin tolerance test (ITT) was performed on 6-h fasting mice by intraperitoneal injection of 1 unit/kg body wt human insulin (Sigma Aldrich, St. Louis, MO). Blood glucose concentration was determined at 0, 30, 60, 90, and 120 min after insulin administration. Triglyceride levels in serum from fasting animals were determined using a GPO-Trinder assay (Sigma Aldrich). Serum free fatty acid levels in fasting animals were measured using a NEFA-Kit-U (Wako Pure Chemical, Richmond, VA). For determination of tissue triglyceride content, 50–100 mg tissue pieces were homogenized on ice in pH 7.3 extraction buffer (20 mmol/l Tris, 1 mmol/l β-mercaptoethanol, 1 mmol/l EDTA). After centrifugation, the glycerol content of the supernatants was determined using a GPO-Trinder assay (Sigma Aldrich) according to the manufacturer’s instructions.

Mice were subjected to 14-h fast, injected with saline or insulin (10 units/kg body wt i.p.) and sacrificed 3 min after injection. Tissues were collected and frozen. Phosphoinositide-3-OH kinase (PI3K) activity was measured in phosphotyrosine immunoprecipitates (p-Tyr, Ab-4; EMD Biosciences, San Diego, CA) from white adipose tissue (WAT), skeletal muscle, and liver lysates, as previously described (16).

Mice were divided into two groups and were treated once daily with equal volumes of intraperitoneally injected saline or mouse leptin (R&D Systems, Minneapolis, MN) dissolved in saline at doses of 5 μg/g body wt for 6 days. Food intake and changes in body weight were measured to estimate the effects of exogenous leptin administration.

Mouse WAT, skeletal muscles, and livers from wild-type and AR^−/y were dissected and total RNA isolated, using a TRIzol reagent (Invitrogen, Carlsbad, CA). cDNA synthesis and PCR were performed using Superscript RNase H⁻ reverse transcriptase and cDNA cycle kit (Invitrogen) according to the manufacturer’s instructions. Real-time PCR was performed using iCycler real-time PCR amplifier (Bio-Rad Laboratories, Hercules, CA). Each PCR contained 1 μl cDNA, 50 μmol/l primers, and 12.5 μl iQ SYBR Green Supermix reagent (Bio-Rad Laboratories) and was triplicated, and 18s was used as an internal control. A list of the primer sequences for real-time PCR is available in Table 1.

A 60-day time-release 5α-dihydrotestosterone (5 mg) or placebo pellet (Innovative Research of America, Sarasota, FL) was inserted into a subcutaneous pocket (posterior neck region) of male wild-type and AR^−/y mice, respectively. At 8 weeks after implantation of the pellet, the mice were killed to study serum hormones and metabolic parameters.

All values are the means ± SE. Differences between two groups were assessed using the unpaired two-tailed Student’s t test or among more than two groups by ANOVA. In all statistical comparisons, P < 0.05 was defined as a significant difference. SigmaStat statistics software (version 2.0; SPSS) was used for all calculations.

Using Cre/loxP-mediated recombination by crossing AR-loxP mice with ACTB-Cre mice, we generated mice lacking functional AR in mixed C57BL/6/129Sv/FVB background (15). Male wild-type and knockout AR^−/y mice were used for the present studies. Because the AR^−/y mice were phenotypically female in appearance, we also used female wild-type mice as an additional comparison group for phenotypic analysis. The growth curves of male wild-type, female wild-type, and AR^−/y mice showed that AR^−/y mice exhibited a significant decrease in the body weight gain throughout the 20th week compared with male wild-type mice (Fig. 1A). The body weight of male wild-type and AR^−/y mice was significantly greater at all time points after 8 weeks compared with female wild-type littermates. However, body weight gain was significantly greater in AR^−/y mice after 22 weeks compared with their male and female wild-type littermates throughout the 40-week study period (male wild-type: 36.0 ± 1.3 g; female wild-type: 26.35 ± 0.59 g; AR^−/y: 40.87 ± 1.0 g; P = 0.009 for male wild-type vs. AR^−/y; n = 9). This was accompanied by significantly enlarged gonadal and perirenal fat pads compared with male and female wild-type counterparts (Figs. 1B and C, Table 2). Aging obesity profiles are consistent with previous findings (17,18). Circulating triglyceride levels of AR^−/y mice was increased by twofold compared with male and female wild-type mice (Table 2). Notably, histological analysis of WAT after fixation (Fig. 1B) and quantification of adipocyte size (Fig. 1C) revealed that adipocytes from 35-week-old AR^−/y mice were significantly larger than those in wild-type counterparts. The comparison data for muscle and bone mass showed no significant difference in those tissue masses between wild-type and AR^−/y mice (data not shown). Although obvious enlargement of adipose tissues can be seen from intra-abdominal appearance in AR^−/y mice, this only accounts for ∼15% increased body weight in AR^−/y mice compared with wild-type mice. These results indicate that AR^−/y mice had marked increased adiposity at an older age.

Given the excess lipid deposition in WAT, we examined blood glucose and insulin levels in male wild-type, female wild-type, and AR^−/y mice. AR^−/y mice start to show elevated blood glucose at 20 weeks (Table 3), which persisted to at least 35 weeks under both the fasting and fed conditions (Table 2). The hyperglycemia occurred despite a marked increase in serum insulin levels in the fasting (∼60%) and fed (∼67%) states, indicating that AR^−/y mice were more resistant to insulin than their wild-type counterparts under ambient conditions (Table 2).

To assess whole-body glucose homeostasis, we next performed oral bolus GTTs and intraperitoneal ITTs on these mice. Oral bolus GTTs demonstrated marked glucose intolerance and distinguishable area under the curve in AR^−/y mice (Figs. 2A and B). Blood glucose levels of AR^−/y mice were greater at all times during the test, and hyperglycemia was still apparent 2 h after the glucose bolus. ITTs showed that AR^−/y mice were slightly resistant and highly resistant to the hypoglycemic effect of exogenous insulin at 25 and 35 weeks of age, respectively (Figs. 2C and D). The defect in whole-body insulin sensitivity was not caused by female-like phenotype in AR^−/y mice because we detected no differences in response patterns of male and female wild-type mice.

Because insulin resistance can be correlated to the activity of PI3K, a signaling mediator needed for many metabolic effects of insulin, we examined insulin-stimulated PI3K activity using wild-type mice as controls. The PI3K activity on insulin stimulation in AR^−/y mice was reduced by 60–63% in insulin target organs, such as skeletal muscle and liver (Fig. 2E), suggesting that AR^−/y mice did have skeletal muscle and hepatic insulin resistance and hyperinsulinemia with obesity.

Serum free fatty acid levels in the fasting state are elevated in AR^−/y mice (Table 2). Moreover, skeletal muscle and hepatic triglyceride content markedly increased by 2.6- and 1.9-fold, respectively, indicating that insulin resistance was associated with increasing triglyceride deposition in the skeletal muscle and liver (Fig. 2F). As expected from the increased WAT mass, fed serum leptin concentrations were higher in AR^−/y mice at both 25 and 35 weeks of age (Fig. 3A). Moreover, serum leptin levels show elevated linear relationships with body weight in AR^−/y mice compared with wild-type littermates (Fig. 3B). Surprisingly, fed serum leptin levels were also higher, even though AR^−/y mice gained significantly less weight before 20 weeks of age (Fig. 3A), indicating that loss of AR may cause increased leptin earlier than increased body weight. Adiponectin, an insulin-sensitizing adipokine, was reduced in AR^−/y mice (Fig. 3C), whereas there was no difference in TNF-α levels (Fig. 3D).

We also found that food intake and body weight was significantly reduced after exogenously administrated leptin in wild-type mice, but not in AR^−/y mice (Figs. 4A and B). Furthermore, food intake was not significantly different in AR^−/y mice compared with wild-type mice before exogenous leptin administration, despite elevated leptin levels, indicating AR^−/y mice were leptin resistant at 35 weeks of age. However, food intake and body weight were reduced after an exogenous load of leptin in both 20-week-old AR^−/y mice and wild-type mice, whereas two control groups had similar body weight and adiposity (Figs. 4C and D). These results suggest that AR^−/y mice develop leptin resistance that may be caused by the development of adiposity and the long-term absence of AR.

To determine the mechanism of the increased lipid deposition in WAT, as well as skeletal muscle and hepatic triglyceride accumulation, mRNA from these tissues were further analyzed. The mRNA levels of four lipid metabolism genes, peroxisome proliferator–activated receptor (PPAR)-γ (PPARγ), CCAAT/enhancer–binding protein-α (C/EBPα), adipocyte fatty acid–binding protein/adipocyte P2 (aP2), and sterol regulatory element–binding protein 1c (SREBP1c), were higher in WAT of AR^−/y compared with wild-type mice (Table 4), suggesting that loss of AR may contribute to the increase of adipogenesis and lipogenesis via stimulation of several genes. Moreover, consistent with triglyceride accumulation, loss of AR reduced mRNA levels of PPAR-α (PPARα) in skeletal muscle and liver (Table 4). These results suggest that AR is directly or indirectly involved in lipid metabolism.

Because serum testosterone levels were markedly decreased as a result of atrophic testes in AR^−/y mice, we could not exclude the possibility that the insulin resistance and metabolic abnormalities in AR^−/y mice simply reflected the low levels of androgens. To address this issue, nonaromatizable androgen, dihydrotestosterone, was given to both 26- and 12-week-old AR^−/y and wild-type littermates. Several serum hormones and metabolic parameters were assessed after 8 weeks of pellet implantation. Dihydrotestosterone replacement restored serum dihydrotestosterone levels to within the physiological range (0.6–0.9 ng/ml) in AR^−/y mice. It is known that estradiol is converted not only from estrone but also testosterone; therefore, we could not exclude the possibility that the abnormalities in AR^−/y mice simply reflected less estrogen converted from testosterone, although we found that serum estradiol levels, as well as levels of the prohormone androstenedione, remained unchanged in AR^−/y mice compared with wild-type mice (Tables 3 and 5). Dihydrotestosterone replacement couldn’t reverse the metabolic abnormalities and insulin resistance in AR^−/y mice in either age study group (Tables 3 and 5), which suggests that androgen actions directly mediated via AR are significant in insulin sensitivity.

Cross-sectional epidemiological studies have indicated a direct correlation between serum testosterone concentrations and insulin sensitivity (19). Low testosterone levels are associated with an increased risk of type 2 diabetes in men (20,21). Because most androgens need to bind to AR to exert their androgenic biological functions, AR have been thought to function as a modulator of insulin sensitivity. Our current results demonstrate that mice lacking AR develop insulin and leptin resistance at an advanced age. Late onset of obesity, as we observed in AR^−/y mice, was reported to be associated with insulin resistance (22). The marked hyperinsulinemia and hyperglycemia in AR^−/y mice clearly demonstrate that loss of AR may reduce insulin sensitivity. A relatively small increase in body weight (∼15%) is associated with a remarkable reduction in insulin sensitivity (∼65%) in AR^−/y mice and with insulin resistance that occurs as early as 20 weeks of age in nonobese AR^−/y mice, suggesting that loss of AR may directly reduce insulin sensitivity in target tissues without first increasing body weight significantly.

The phenotypically female appearance of AR^−/y mice is similar to that of Tfm (testicular feminized) mice, in which AR is functionally deficient by introduction of the Tfm mutation in the AR gene (23). Therefore, we added female wild-type mice for comparison, and these mice remain smaller and have less adiposity than both male wild-type and AR^−/y mice. In addition, we found no significantly distinguishable metabolic pattern between female and male wild-type mice, whereas AR^−/y mice exhibited severe insulin resistance and obesity. A previous study has showed that db/db Tfm/Y males develop severe diabetes. In contrast, female db/db littermates only exhibit mild hyperglycemia (24).

The excess fat mass in AR^−/y mice may be caused by an impaired ability of skeletal muscle to use lipids as a fuel substrate and of liver to catabolize lipids, leading to a shunting of lipids to adipose tissue. In fact, the unchanged food intake in AR^−/y mice suggests that the excess weight gain and adiposity may be caused by normal energy input coupled with reduced lipid oxidation and increased lipid storage. Indeed, impairment in hepatic lipid oxidation resulted in elevated circulating free fatty acid and hepatic steotosis in PPAR-α null mice (25). Activation of PPAR-α by fibrates reduces adiposity in fa/fa ZDF (Zucker diabetic fatty) rats via activation of several peroxisomal and mitochondrial fatty acid oxidation genes (26). Given that skeletal muscle and liver are the major sites of lipid oxidation and deposition, respectively, it is possible that loss of PPAR-α could produce such an effect. Consistent with this possibility, decreased PPAR-α expression in skeletal muscle, decreased hepatic lipid metabolism, and subsequent alterations in expression of genes that stimulate adipocyte differentiation (PPARγ [22], C/EBPα [27], and SREBP1c [28]) and lipid accumulation (aP2 [29]) in WAT were found in AR^−/y mice. Furthermore, liver-specific AR knockout mice, which are more susceptible to high-fat diet–induced insulin resistance, showed lower hepatic PPAR-α expression (H.-Y.L., C.C., unpublished observations). Recent studies using hyperinsulinemic-euglycemic clamps have shown that PPAR-α null mice are not protected against high-fat diet–induced insulin resistance (30). Thus, loss of PPAR-α expression in AR^−/y mice may be a potential mechanism contributing to the decreased insulin sensitivity observed.

An alternative explanation is that skeletal muscle and hepatic insulin resistance in AR^−/y mice might be secondary to altered release of adipokines. Leptin increased insulin sensitivity and concomitantly reduced triglyceride content by promoting lipid oxidation both in animal models of insulin resistance and in humans with lipodystrophic diabetes (31). The AR^−/y mice with elevated leptin, however, have significant leptin resistance when exogenous leptin is administered. Related results have been observed in ZDF rats, ob/ob mice that lack leptin, db/db mice that have an inactive leptin receptor, and the livers of such mice are all steatotic (32,33). In ZDF rats, the skeletal muscle, pancreas, and heart are also steatotic (34). A strong correlation between intracellular triglyceride content and insulin resistance has been established in both human and animal studies of obesity-related insulin resistance and type 2 diabetes (35). Consistent with this possibility, a substantial ectopic deposition of triglycerides in nonadipocytes, such as skeletal muscle and liver, in AR^−/y mice is likely a central element. Triglyceride overload of nonadipocytes causes insulin resistance, lipotoxic heart disease, and adipogenic type 2 diabetes (36). We have also demonstrated, through PI3K activity assay after injection of insulin, that skeletal muscle and liver in AR^−/y mice are insulin resistant. Interestingly, PPAR-α is necessary for the lipopenic action of hyperleptinemia on white adipose and liver tissues. In PPAR-α null mice infused with adenovirus-leptin, upregulation of carnitine palmitoyl transferase-1 mRNA in fat, downregulation of acetyl CoA carboxylase in liver, and upregulation of PPAR-γ coactivator-1α mRNA in both tissues are abolished, as is the reduction in their triglyceride content, suggesting that leptin action may be mediated through PPAR-α (37). Thus, loss of PPAR-α in AR^−/y mice may reduce leptin action, despite elevated leptin levels. In contrast, we found decreased adiponectin and no elevation of serum TNF-α (another adipokine that contributes to insulin resistance) (38) in AR^−/y mice at 35 weeks of age, whereas mice lacking adiponectin exhibited severe insulin resistance (39). However, it is known that other adipokines also mediate the metabolism of lipids and contribute to the pathogenesis of insulin resistance. Further studies are needed to determine the mechanism of lipid profile elevation in AR^−/y mice.

Although there is no evidence that androgen-AR can directly affect PPAR-α, indirect interactions of AR and PPAR-α through nuclear receptor coregulators are possible. ARA70 (AR-associated coregulator 70), a coactivator of AR, was able to physically interact with PPAR-α, as determined by coimmunoprecipitation. In the adrenal cell line Y1, ARA70 behaved as a repressor of PPAR-α (40). The mechanism of how leptin is regulated by androgen-AR remains elusive. An early study indicated that in primary rat adipocytes expressing AR, testosterone and dihydrotestosterone were able to suppress leptin mRNA and leptin secretion (41), suggesting a direct effect of testosterone on the regulation of leptin secretion in adipocytes. AR^−/y mice not only have an AR deficiency, but they also have decreased serum levels of androgens. Thus, in AR^−/y mice, treatment with nonaromatizable dihydrotestosterone restored the physiological serum dihydrotestosterone levels, whereas estrogen levels remained unchanged compared with wild-type mice. Therefore, the possibility that a of lack of estrogen receptor-α activation results in increased WAT, as seen in male mice deficient in either estrogen receptor-α or aromatase, can be excluded. Importantly, we found no reversed effect after dihydrotestosterone treatment, indicating that insulin resistance in AR^−/y mice, as seen in the present study, is mediated directly via AR.

Taken together, these data suggest that loss of AR may contribute to an increase of leptin levels and leptin resistance, which may play important roles for the development of obesity and insulin resistance. A previous study using transgenic mice overexpressing leptin demonstrated that loss of leptin effectiveness in older transgenic mice might contribute to the accumulation of adipose mass (42). Consistent with this possibility, the AR^−/y mice were progressively desensitized to leptin because of long-term elevated circulating leptin throughout the 40-week study period. In rats, testosterone was reported to be associated negatively with serum leptin, independent of BMI (43). Leptin levels are higher in aging men with lower testosterone (44). Furthermore, low testosterone may lead to the accumulation of visceral fat (45). As total body fat mass increases, resistance develops to both leptin and insulin. Elevated leptin fails to stimulate weight loss, and the hypogonadal-obesity cycle ensures further visceral obesity and insulin resistance. In our present study, AR^−/y mice progressively develop leptin resistance and insulin resistance, resulting in obesity. This differs from the ob/ob and db/db mice that display early onset genetic obesity and have leptin inactivity from a very early stage (46). AR^−/y mice have milder obesity because of the late onset of leptin and insulin resistance. However, leptin resistance in AR^−/y mice is similar to that of db/db mice, and having dysfunctional leptin receptors is likely to be a major contributing factor. Progressive insulin resistance can then lead to type 2 diabetes and other diseases, such as high triglyceride–low HDL pattern dyslipidemia and metabolic syndrome X, as well as an increased risk of cardiovascular diseases (45). It has been reported that hypotestosteronemia may be a risk factor for coronary artery disease in men (47).

Interestingly, along with loss of AR, for which we demonstrated a resulting metabolic syndrome, both loss of estrogen receptor-α (48) and aromatase deficiency, which causes loss of the ability to synthesize estrogen (49), result in metabolic syndromes. These observations indicate the possibility that loss of both androgen and estrogen responsiveness disrupts energy homeostasis.

In summary, AR^−/y mice provide an in vivo model showing that loss of AR increases serum leptin concentration and skeletal muscle/hepatic triglyceride content, which may result in the development of obesity, leptin resistance, and insulin resistance. Because obesity and progressive insulin resistance may lead to type 2 diabetes and an increased risk of cardiovascular disease (50), a better understanding of the molecular mechanisms involved and dissection of the roles of androgen-AR in insulin and leptin resistance may help in the development of better therapeutic approaches to type 2 diabetes, obesity, and cardiovascular diseases.

This work was supported by grants from the George Whipple Endowment, the National Institutes of Health (DK60948), and the Dr. Shu-Shi Chang Neuroendocrinology Fund.

We thank K. Wolf, L. Collins, and T. Chang for manuscript help.