A muscle-specific knockout implicates nuclear receptor coactivator MED1 in the regulation of glucose and energy metabolism

Abstract

As conventional transcriptional factors that are activated in diverse signaling pathways, nuclear receptors play important roles in many physiological processes that include energy homeostasis. The MED1 subunit of the Mediator coactivator complex plays a broad role in nuclear receptor-mediated transcription by anchoring the Mediator complex to diverse promoter-bound nuclear receptors. Given the significant role of skeletal muscle, in part through the action of nuclear receptors, in glucose and fatty acid metabolism, we generated skeletal muscle-specific Med1 knockout mice. Importantly, these mice show enhanced insulin sensitivity and improved glucose tolerance as well as resistance to high-fat diet–induced obesity. Furthermore, the white muscle of these mice exhibits increased mitochondrial density and expression of genes specific to type I and type IIA fibers, indicating a fast-to-slow fiber switch, as well as markedly increased expression of the brown adipose tissue-specific UCP-1 and Cidea genes that are involved in respiratory uncoupling. These dramatic results implicate MED1 as a powerful suppressor in skeletal muscle of genetic programs implicated in energy expenditure and raise the significant possibility of therapeutical approaches for metabolic syndromes and muscle diseases through modulation of MED1–nuclear receptor interactions.

The transcriptional control of gene regulation in response to environmental cues plays a major role in energy homeostasis (1, 2). Prominent among the transcription factors involved in the associated metabolic processes are nuclear receptors, which are generally activated by ligands in various signaling pathways. The actual function of promoter-bound nuclear receptors, either as activators or repressors, depends on their interactions with various coregulators (coactivators and corepressors), which, like the receptors themselves, may also be modulated by various signaling pathways (1–4). Apart from coregulators that are involved mainly in the modulation of chromatin structure, the multisubunit Mediator also serves as a major coactivator for diverse nuclear receptors (5).

Mediator is a large 25- to 30-subunit complex that interacts directly with diverse transcriptional activators and with RNA polymerase II (5, 6), thus serving as a bridge between DNA-bound activators and the general transcription machinery to facilitate formation and/or function of the preinitiation complex. In the case of nuclear receptors, this was established mainly through in vitro biochemical and cell-based assays (5) following identification of the human TRAP/Mediator complex through a ligand-dependent intracellular association with thyroid hormone receptor (7). In particular, the in vitro studies established ligand-dependent Mediator coactivator functions that involve ligand-dependent binding of Mediator, through the MED1 subunit, to nuclear receptor AF2 domains.

Toward establishment of in vivo functions, early mouse genetic (knockout) studies indicated that MED1 is essential for early embryonic development, but not for cell viability per se, and for optimal functions of ectopic nuclear receptors in mouse embryonic fibroblasts (MEFs) (8, 9). A subsequent study showed that MED1 is essential for peroxisome proliferator-associated receptor γ (PPARγ)–mediated differentiation of MEFs to adipocytes and for associated PPARγ-dependent transcription events (10). More physiological in vivo assays of MED1 function have involved conditional Med1 knockout and mutant Med1 knock-in analyses. As examples, a recent study of a mutant Med1 knock-in has demonstrated a cell-specific role for MED1 in ERα-dependent mammary gland development (11), whereas an earlier study of a liver-specific Med1 knockout mouse demonstrated a role for MED1 in PPARα-mediated oxidation of fatty acids (12).

In extending the in vivo analyses of MED1 function, the current study is focused on its role in muscle. As a major site for the regulation of fatty acid and glucose metabolism, skeletal muscle plays a significant role in the control of obesity, diabetes, and cardiovascular disease, and nuclear receptors (1, 13) and coregulators (2, 4) play key roles in the regulation of normal muscle maintenance and function. In relation to coregulators, deletion of corepressor RIP140 has been shown to enhance the formation of oxidative slow twitch muscle fibers and expression of genes involved in fatty acid oxidation (14), whereas the muscle-specific ablation of PGC-1α, a coactivator with a major role in mitochondrial biogenesis and oxidative phosphorylation (4), effects a slow-to-fast muscle fiber switch and muscle fiber damage (15). In contrast, ectopic expression of PGC-1α in muscle enhances formation of slow twitch fibers (16), affords protection to denervation/disuse-associated muscle atrophy (17), and ameliorates symptoms of muscular dystrophy (18). Such studies have suggested previously undescribed therapeutical opportunities for metabolic and muscle diseases involving coregulator manipulation.

Here, we report that muscle-specific Med1 knockout (Med1 MKO) mice exhibit enhanced insulin sensitivity and improved glucose tolerance and are resistant to high-fat diet–induced obesity. Med1 MKO mice also exhibit increased mitochondrial density and increased expression of genes specific to type I and type IIA fibers in white muscle, indicating a switch toward slow fibers, as well as increased expression of brown adipose tissue (BAT)-specific genes (UCP-1 and Cidea) involved in energy expenditure through uncoupled respiration. This profound and somewhat surprising phenotype is indicative of a strong suppressive function of MED1 on energy expenditure pathways in muscle and also has significant implications for therapeutical approaches to metabolic and muscle diseases.

Results

Generation and Characterization of Muscle-Specific Med1 MKO Mice.

Med1 conditional knockout mice carrying the Med1 floxed allele (Med1^fl/fl) mice were generated and characterized as described in Figs. S1 and S2. These mice were mated with MCK-Cre transgenic mice to generate Med1 MKO mice that were characterized as described in Fig. S2. Of note, the muscle-selective knockout of Med1 was evident from the presence of the mutant Med1 gene, encoding a truncated MED1 protein, in skeletal muscle, but not liver, from mice that carry both the Med1 flox and Cre alleles. Because MCK is expressed after myocyte differentiation and MyoD-stimulated myogenesis is normal in Med1^−/− MEFs (10), the effect of Med1 deletion in muscle may be restricted only to nuclear receptor-dependent responses of differentiated cells.

Med1 MKO Mice Exhibit Increased Glucose Tolerance and Insulin Sensitivity and Are Resistant to High-Fat Diet–Induced Obesity.

Despite muscle-related functions for several nuclear receptors (13) with demonstrated or anticipated Mediator interactions through MED1, Med1 MKO mice did not show any abnormalities in gross appearance, behavior, or fertility. On a standard chow diet, these mice also had normal body weight and fat content, as determined by a PIXImus densitometer (GE Medical Systems), compared with control mice (Fig. S3 A and B). Histological analysis of the skeletal muscle tissues also did not reveal any differences between WT and Med1 MKO mice.

The apparent lack of an effect of muscle MED1 deficiency under normal conditions raised the question of possible conditional requirements for MED1 function. To study the role of muscle MED1 in dietary fat metabolism, 8-week-old WT and Med1 MKO mice were placed on a high-fat (60%) diet. Significantly, Med1 MKO female mice showed far less weight gain on the high-fat diet compared with control littermates (Fig. 1A). The weight gain of male Med1 MKO mice was not distinguishable from that of control male littermates on the high-fat diet (Fig. S3C), with the difference between male and female mice on the high-fat diet possibly reflecting different hormonal effects. To determine whether the resistance to weight gain reflected an alteration in feeding behavior, food intake in control and mutant animals was monitored daily for a period of 3 weeks. The results indicated no significant difference in food intake between Med1 MKO and control mice for either gender (Fig. S3D). Altogether, these results indicate that Med1 MKO mice are resistant to diet-induced obesity and, thus, that muscle MED1 plays a role in energy homeostasis.

Fig. 1.

Med1 MKO mice have altered glucose and energy metabolism. (A) Med1 MKO female mice are resistant to high-fat diet–induced obesity. Four-week-old mice of three different genotype groups (WT, Med1^fl/fl, and Med1^fl/fl/Cre) were fed a high-fat diet containing 60% fat, and body weights were checked every week. (B) Med1^fl/fl/Cre mice have similar levels of plasma glucose compared with those of Med1^fl/fl mice at fed and fasted states. (C) GTT. The test was performed as described in Experimental Procedures. (D) ITT. The test was performed as described in Experimental Procedures. Data values are mean ± SEM. *P < 0.1; **P < 0.05; ***P < 0.01 (Student's t test).

In a further analysis of the function of muscle MED1 in energy homeostasis, we first analyzed blood glucose and insulin levels in mice in both fed and fasted states and found no significant differences in these parameters between Med1 MKO and control mice on a normal chow diet (Fig. 1B). We then performed i.p. glucose tolerance tests (GTTs) and insulin tolerance tests (ITTs). As shown in Fig. 1C, Med1 MKO mice showed significantly lower glucose levels than did WT controls in the GTT test. In parallel, Med1 MKO mice showed a greater hypoglycemic response to exogenous insulin in the ITT test than did WT controls (Fig. 1D). These results thus indicate a clear role for MED1, through its action in muscle, in the regulation of glucose metabolism. Furthermore, the resistance of Med1 MKO mice to diet-induced obesity is likely attributable to these beneficial effects of muscle MED1 ablation on glucose tolerance and insulin sensitivity.

Microarray Analysis of Gene Expression in Skeletal Muscle of Med1 MKO Mice.

Given the established roles of MED1 as a transcriptional coactivator for nuclear receptors (5), we performed a microarray analysis with Illumina mouse chips (Illumina Inc.) to compare the gene expression profiles of control and Med1 null skeletal muscle. Among 22,000 genes analyzed, 243 were down-regulated at least 1.5-fold and 156 were up-regulated at least 1.5-fold. A pathway analysis showed that the down-regulated genes fall into diverse pathways that include cell organization and biogenesis, cell differentiation, muscle development, regulation of muscle contraction, cytoskeleton organization, cytoskeleton biogenesis and energy pathways (Fig. S4A). However, half of the genes that are up-regulated at least 1.5-fold are involved in metabolic pathways (Fig. S4B). Most strikingly, two mitochondrial protein genes (UCP-1 and Cidea) that are specific to brown adipose tissues and implicated in uncoupled respiration showed the highest levels of up-regulation. Other up-regulated genes of note are PEPCK, C/EBPα, and PPARγ, which have been strongly implicated in energy metabolism (Fig. S5) and further indicate a potential role for MED1 in energy and glucose metabolism.

To validate the microarray results, we performed quantitative real-time RT-PCR with isolated quadriceps (white muscle) and soleus (red muscle) from mouse hind legs. The results in Fig. 2 A and B indicate that UCP-1 (∼40-fold) is the most up-regulated gene in the white muscle of Med1 MKO mice, followed by Cidea (∼20-fold), PEPCK (∼10-fold), PPARγ (∼3-fold), and C/EBPα (∼3-fold), consistent with the microarray analysis. Other relevant genes, including PGC-1α, PGC-1β, UCP-2, UCP-3, Dio2, ERRα, Cox4i, and CytC, showed no change in expression levels in the white muscle of Med1 MKO mice compared with control white muscle. The expression levels in red muscle of all the above genes showed little or no change in Med1 MKO mice compared with control mice (Fig. 2C). These results not only validate our microarray results but are consistent with a report that MCK is expressed highly in white muscle. The source (cell type) of increased UCP-1 expression in white muscle tissue remains to be established. However, this very prominent enhancement of UCP-1 expression, which should lead to a large increase in uncoupled respiration, may account for the enhanced glucose tolerance and insulin sensitivity of the Med1 MKO mice (Discussion). Consistent with MCK expression in cardiac muscle, a secondary knockout of Med1 in the heart was also observed (Fig. S6A). However, H&E staining of the heart revealed no apparent defects (Fig. S6B). Moreover, a gene expression analysis of several relevant genes in heart tissue revealed no significant differences between WT and Med1 MKO mice (Fig. S6C). Of note, UCP-1 expression in heart tissue was undetectable, and thus not indicated in the figure. Therefore, it is unlikely that ablation of Med1 in the heart influences the phenotypes that were observed with Med1 MKO mice.

Fig. 2.

Gene expression analyses of white and red muscles. (A and B) Real-time RT-PCR analysis of total RNA from quadriceps (white muscle) of Med1^fl/fl and Med1^fl/fl/Cre mice. (C) Real-time RT-PCR analysis of total RNA from soleus (red muscle) of Med1^fl/fl and Med1^fl/fl/Cre mice. Each bar represents the mean ± SD of four mice.

White Muscle of Med1 MKO Mice Has Enriched Mitochondria.

The microarray analysis indicated that the most up-regulated genes in the white muscle of Med1 MKO mice are mitochondrial protein genes such as UCP1 and Cidea. This led us to compare the mitochondrial density of white muscle from WT and Med1 MKO mice. As shown in Fig. 3, an electron microscopic analysis of the quadriceps muscle revealed that Med1 MKO mice have significantly enriched mitochondria relative to control mice. This correlates well with the increased expression of UCP-1 and Cidea in the Med1 MKO white muscle, and also with the switch toward more oxidative types of muscle fibers.

Fig. 3.

Electron microscopy of muscle cells from the quadriceps of Med1^fl/fl (A) and Med1^fl/fl/Cre (B) mice. Lower magnification (Left) and higher magnification (Right) are shown. The arrows point to mitochondrial structures.

Med1 Deletion in White Muscle Causes a Switch Toward Slow Fibers.

Skeletal muscles are composed of two classes of fibers, the slow twitch type I and the fast twitch type II, that differ in their contractile and metabolic properties. Type II fibers are further classified as IIA, IIX, and IIB based on the expression of myosin heavy chain isoforms. Type IIA is the slowest fiber among the three fiber type II subtypes and has characteristics close to those of type I fiber. Type I and type IIA fibers generally have high oxidative capacities and corresponding high mitochondrial densities, whereas type IIX and type IIB have high glycolytic capacities and relatively low mitochondrial densities (19). Because we observed a significantly increased mitochondrial density in the white muscle of Med1 MKO mice, we investigated the possibility that the fiber composition of skeletal muscle is altered in Med1 MKO mice. To that end, we first analyzed expression of slow and fast fiber-specific genes from skeletal muscle by quantitative real time RT-PCR. As shown in Fig. 4A, expression levels of genes that are usually enriched in slow fibers, including troponin I 1 (Tnni 1), troponin T 1 (Tnnt 1), troponin C 1 (Tnnc 1), and myoglobin, were significantly increased in quadriceps muscle from Med1 MKO mice compared with the levels in control mice. In contrast, genes specific to fast fibers, such as Tnni 2 and parvalbumin, showed comparable expression levels in white muscle from Med1 MKO and control mice. We then analyzed the expression of the above-mentioned genes in soleus muscle and found that all were expressed at similar levels in Med1 MKO and control mice (Fig. 4B). These results strongly suggest that deletion of Med1 in white muscle causes a switch toward slow fibers.

Fig. 4.

Real-time RT-PCR analysis of type I- and type II-specific gene expression in white muscle (A) and red muscle (B) from Med1^fl/fl and Med1^fl/fl/Cre mice. Each bar represents the mean ± SD of four mice.

To investigate fiber subtype composition in white muscle further, we analyzed the relative gene expression levels of fiber subtype-specific myosin heavy chain isoforms (MyHC7, type I; MyHC2, type IIA; MyHC1, type IIX; and MyHC4, type IIB). As shown in Fig. 5, expression levels of type I and type IIA MyHC were significantly increased in Med1 MKO white muscle compared with control white muscle. Given that type I and type IIA fibers use oxidative phosphorylation as their energy source, they thus have relatively higher mitochondrial densities than the type IIX and type IIB fibers that are glycolytic in nature. Therefore, there is an overall strong correlation between increased expression levels of myoglobin and troponins, increased expression levels of type I and type IIA MyHC, and increased mitochondrial density in the white muscle of Med1 MKO mice.

Fig. 5.

Real-time RT-PCR analysis of type I- and type II-specific myosin heavy chain gene expression levels in white muscle from Med1^fl/fl and Med1^fl/fl/Cre mice. Each bar represents the mean ± SD of four mice.

Discussion

Skeletal muscle is a major site for the regulation of fatty acid and glucose metabolism, and thus energy homeostasis, and nuclear receptors and associated cofactors are emerging as critical regulators of transcriptional programs that govern these processes (1, 2, 4). Here, studies of a mouse muscle-specific knockout of nuclear receptor coactivator MED1 have established a role for this coactivator as a powerful suppressor in muscle of genetic programs implicated in energy expenditure. Thus, our results show that Med1 deletion elicits an increased mitochondrial density, a switch toward slow fibers, and increased expression of brown adipose tissue-specific genes (UCP-1 and Cidea) in skeletal muscle. These striking changes may account, through increased energy expenditure, for the improved systemic insulin and glucose responses observed in Med1 MKO mice and have implications for possible therapeutical intervention both in type 2 diabetes and in muscle diseases associated with mitochondrial dysfunction.

MED1 Plays Important Roles in Regulating Glucose and Energy Metabolism in Skeletal Muscle.

Although MED1 is essential for embryonic development (8, 9), Med1 MKO mice do not show obvious morphological abnormalities. They are viable, with normal growth and body fat content, and fertile, and they exhibit normal glucose and insulin levels in both fed and fasted states. However, GTTs and ITTs revealed enhanced glucose tolerance and insulin sensitivity in Med1 MKO mice. Moreover, on challenge with a high-fat diet, and despite a similar food intake compared with control mice, the Med1 MKO mice show resistance to diet-induced obesity. These results strongly suggest that MED1 acts in the muscle to regulate glucose and energy metabolism.

Relevant to the underlying molecular mechanisms, gene expression profiles revealed 146 up-regulated genes in the skeletal muscle of Med1 MKO mice, with almost half of these being involved in metabolic pathways. Most strikingly, a set of genes that are specific to brown adipose tissues, including, especially, UCP-1 and Cidea, is highly up-regulated in Med1 null skeletal muscle. Consistent with these results, there are several reports of increased UCP-1 expression in skeletal muscle following deletion of nuclear receptors or nuclear receptor cofactors. Thus, knockouts of LXRα/β (20) and RIP140 (21) result in ectopic expression of UCP-1 in muscle as well as metabolic phenotypes similar to those described here for Med1 MKO mice. The increased expression of UCP-1 and associated energy expenditure (through uncoupled respiration) in muscle tissue may account, at least in part, for the enhanced glucose tolerance and insulin sensitivity of the knockout mice as well as the associated protection from high-fat diet–induced obesity. In strong support of this view, transgene-mediated UCP-1 overexpression in skeletal muscle was found to protect mice from diet-induced obesity and insulin resistance attributable to increased skeletal muscle respiratory uncoupling (22). Moreover, because transgene-mediated expression of ectopic UCP1 appears to be sufficient to induce mitochondrial biogenesis in other cell types (23), the dramatic induction of the endogenous UCP1 gene in Med1 MKO muscle may also be important, perhaps through retrograde signaling (24), for the increased mitochondrial content of this tissue.

Although the effects of UCP-1 overexpression are clear, the issue of whether UCP-1 is expressed in muscle cells or in ectopic BAT (intermuscular BAT) in muscle tissue in the knockouts is still controversial (25). In this regard, intermuscular BAT has been found in muscle tissues, and these ectopic BAT depots are more sensitive to β3-adrenergic agonists, which increase induction of UCP-1, than is intrascapular BAT (25). Although we have not observed any obvious BAT deposition in muscle tissues, we have not yet been able to establish that UCP-1 is expressed within muscle cells per se. This interesting possibility is raised by recent demonstrations that brown fat and skeletal muscle cells arise from a common progenitor and that muscle cells can be induced by ectopic expression of PRDM16 to express brown fat-specific genes (26, 27). It is also relevant to note that our previous study has demonstrated cooperativity between MED1 and PGC-1α in activation of UCP-1 expression by TRα in differentiated MEF-derived adipocytes (28). Therefore, our current observation that deletion of Med1 in muscle, where UCP-1 usually is silenced, results in activation of UCP-1 presents somewhat of a paradox. One possible explanation is that MED1 acts indirectly in muscle by coactivating (with nuclear receptors or other activators) genes whose products repress expression of UCP-1 and other genes such as Cidea and slow fiber-specific genes. Another possibility is that MED1/Mediator may also work as a corepressor, potentially with RIP140, to keep some nuclear receptor target genes silent. In this regard, RIP140 has been reported to repress UCP-1 expression through interactions with LXRα and ERRα (24, 29, 30), and RIP140 null mice, like Med1 MKO mice, show, in addition to UCP-1 induction in muscle, improved glucose tolerance and insulin sensitivity, resistance to diet-induced obesity, and an increase in type I (oxidative) muscle fibers. Consistent with a possible cooperativity between MED1 and RIP140, the effect of Med1 ablation on derepression of specific genes was observed in the absence of any changes in the level of RIP140 (Fig. S6D). In relation to the possibility of both coactivator and corepressor functions for MED1, there is precedent for other well-established nuclear receptor coactivators exhibiting context-dependent corepressor functions as well. This has been demonstrated, for example, for GRIP1/SRC2 (31), for PRDM16 (32), and also for the Mediator acting in conjunction with other corepressors and the G9a methyltransferase (33).

MED1 Is Involved in the Regulation of Fiber Composition and Mitochondrial Biogenesis in Muscle.

Skeletal muscles are composed of type I, IIA, IIX, and IIB fibers according to the type of MyHC gene expressed in fibers. Type I and type IIA fibers are rich in mitochondria and have a high oxidative capacity and a long duration, whereas type IIX and type IIB fibers are relatively poor in mitochondria and have a high glycolytic capacity and a short duration (19). Importantly, muscle fibers have enormous plasticity, and the fiber composition can be changed in association with various physiological and pathological conditions. Our finding that Med1 deletion in muscle promotes oxidative fiber type I and IIA formation clearly indicates a role for MED1, through a yet unknown mechanism, in regulating muscle fiber composition. Consistent with the slow fiber enrichment, Med1 MKO mice muscle also has a higher mitochondria density than does the muscle of control mice. The increased oxidative fiber composition in Med1 MKO mice may contribute to the enhanced glucose tolerance and insulin sensitivity attributable to increased fatty acid oxidation and mitochondrial respiration.

As reported here for MED1, previous studies have established a major role for PGC-1α, a primary regulator of mitochondrial biogenesis and oxidative metabolism (4), in fiber switching in mice. Thus, PGC-1α is preferentially expressed in type I fibers, and its ectopic expression in skeletal muscle drives type I fiber formation, at least in part through MEF2 interactions that, in turn, directly activate the transcription of type I myofiber genes (16). In contrast, PGC-1α deletion in skeletal muscle elicits a slow-to-fast fiber switch (15). These results, along with those of the RIP140 knockout study (14) and our observations on MED1, demonstrate important roles of diverse nuclear receptor cofactors in the regulation of muscle fiber composition and mitochondrial density. The mechanisms by which MED1 ablation elicits the changes in gene expression that lead to fiber type switching and increased mitochondrial density as well as enhanced UCP1 expression in skeletal muscle are currently unknown and will be the subject of future investigation. However, it is significant that these changes occur without an increase in the level of muscle PGC-1α, indicating that the normal level of PGC-1α is sufficient for enhanced activation of PGC-1α target genes in the absence of MED1. This, in turn, suggests the possibility of an active MED1-dependent suppression mechanism that acts directly on PGC-1α target genes and that normally imposes a requirement for elevated levels of PGC-1α for further activation of specific target genes.

In relation to muscle-related disorders, it is important to note that whereas exercise stimulates a switch from fast to slower fibers (34), type II diabetes and obesity are often associated with a reduced slow and elevated fast fiber composition (35) and with mitochondrial dysfunction related to a lower mitochondrial density (36). Mitochondrial oxidative phosphorylation also is impaired in muscular dystrophy (37, 38), consistent with the observation that mitochondrial-rich slow muscle fibers are more resistant to the adverse effects of dystropin (19). Hence, the ability of MED1 ablation to enhance oxidative fiber formation, as well as UCP1 expression, offers unique prospects for therapeutical approaches to metabolic and muscle-related diseases through down-regulation of MED1 function. Similar possibilities have been suggested in relation to the up-regulation of PGC-1α activity in muscle (18–20).

Experimental Procedures

Animal Protocols.

Mice were maintained on a standard rodent chow and 12-h light/dark cycle in The Rockefeller University institutional animal facility. The guidelines for the care and use of animals at The Rockefeller University were followed for all animal experiments. Med1 conditional knockout mice were generated by standard gene-targeting protocols. The ES cell used for gene targeting was derived from the Sv129 strain. Mice carrying the Med1 floxed allele (Med1^fl/fl or Med1^fl/+) were mated with MCK-Cre mice (39) to obtain Med1^fl/fl/Cre (Med1 MKO) mice. Med1 MKO mice were maintained in a mixed genetic background of C57BL/6: Sv129. Med1^fl/fl or WT mice were used as control mice. The primers used for genotyping are available on request. Body fat content was measured with a dexascanner. For high-fat diet–induced obesity, 4-week-old mice were fed a high-fat diet containing 60% fat (Research Diets). Body weight was measured weekly for 16 weeks. For food intake measurements, mice were singly housed and food was weighed every day for 3 weeks. Plasma glucose levels were measured from tail blood using a glucometer. Insulin levels were measured using an ELISA kit (catalog no. 34-05105; Mercodia, Inc.).

GTTs and ITTs.

Male mice aged 6 to 8 weeks were fasted overnight (for GTT) or 4 h (for ITT) and then injected i.p. with a glucose solution at a concentration of 2 g/kg of body weight or with an insulin solution at 0.75 U/kg of body weight. Blood glucose levels were then measured using a glucometer from tail blood taken at indicated time points over a 3-h period.

Real-Time RT-PCR.

Quadriceps (white muscle) and soleus (red muscle) were collected from mice hind legs, cut into small pieces, and homogenized in TRIzol reagent (Invitrogen) solution using a Tissue Tearor (Biospec Products, Inc.). Total RNA was then sequentially purified with TRIzol and an RNeasy kit (Qiagen). Equal amounts of total RNA were subjected to first-strand cDNA synthesis using SuperScriptIII First Strand cDNA Synthesis for the RT-PCR kit (Invitrogen) following the manufacturer's instruction. RNA levels were analyzed by quantitative real-time PCR in 25-μL reactions with SYBR Green (Applied Biosystems). For normalization, 18S RNA was used as an internal control. Fold inductions are shown relative to control RNA. The figures show the mean and SD of four mice of the same genotype in each group. Primer sequences used in real-time PCR are available on request.

Microarray Analysis.

RNA was isolated from total skeletal muscle of hind legs of 10- to 12-week-old mice. The RNA was provided to the Genomics Resource Center at The Rockefeller University to perform Illumina genome-wide expression microarrays analysis for mice according to the manufacturer's instructions. Microarray data analysis was done using GeneSpring software (Agilent Technologies). Four mice in each group were included in the microarray experiments.

Electron Microscopy.

Quadriceps muscle samples were collected from 12-week-old mice, fixed, and submitted to the Electron Microscopy Resource Center at The Rockefeller University for electron microscopic analysis.