Increased Systemic Glucose Tolerance with Increased Muscle Glucose Uptake in Transgenic Mice Overexpressing RXRγ in Skeletal Muscle

Background Retinoid X receptor (RXR) γ is a nuclear receptor-type transcription factor expressed mostly in skeletal muscle, and regulated by nutritional conditions. Previously, we established transgenic mice overexpressing RXRγ in skeletal muscle (RXRγ mice), which showed lower blood glucose than the control mice. Here we investigated their glucose metabolism. Methodology/Principal Findings RXRγ mice were subjected to glucose and insulin tolerance tests, and glucose transporter expression levels, hyperinsulinemic-euglycemic clamp and glucose uptake were analyzed. Microarray and bioinformatics analyses were done. The glucose tolerance test revealed higher glucose disposal in RXRγ mice than in control mice, but insulin tolerance test revealed no difference in the insulin-induced hypoglycemic response. In the hyperinsulinemic-euglycemic clamp study, the basal glucose disposal rate was higher in RXRγ mice than in control mice, indicating an insulin-independent increase in glucose uptake. There was no difference in the rate of glucose infusion needed to maintain euglycemia (glucose infusion rate) between the RXRγ and control mice, which is consistent with the result of the insulin tolerance test. Skeletal muscle from RXRγ mice showed increased Glut1 expression, with increased glucose uptake, in an insulin-independent manner. Moreover, we performed in vivo luciferase reporter analysis using Glut1 promoter (Glut1-Luc). Combination of RXRγ and PPARδ resulted in an increase in Glut1-Luc activity in skeletal muscle in vivo. Microarray data showed that RXRγ overexpression increased a diverse set of genes, including glucose metabolism genes, whose promoter contained putative PPAR-binding motifs. Conclusions/Significance Systemic glucose metabolism was increased in transgenic mice overexpressing RXRγ. The enhanced glucose tolerance in RXRγ mice may be mediated at least in part by increased Glut1 in skeletal muscle. These results show the importance of skeletal muscle gene regulation in systemic glucose metabolism. Increasing RXRγ expression may be a novel therapeutic strategy against type 2 diabetes.


Introduction
The skeletal muscle, known as the largest organ in the human body, plays an important role in exercise, energy expenditure, and glucose metabolism. It is a major site of glucose disposal [1,2]. In type 2 diabetic subjects, glucose uptake in the skeletal muscle is impaired [2]. Blood glucose is taken up by the skeletal muscle via insulin-dependent and independent glucose transporters (Glut4 and Glut1, respectively) [3], where it is converted into glycogen. Increasing capacity of glucose uptake in the skeletal muscle is considered beneficial for the prevention and treatment of type 2 diabetes [3,4]. On the other hand, glucose metabolism in the skeletal muscle may affect the whole body metabolism; the skeletal muscle-specific inactivation of Glut4 has resulted in defect in insulin action in the adipose tissue (glucose uptake) and liver (suppression of gluconeogenesis) [5]. Thus, identification of the molecular mechanisms involved in skeletal muscle glucose metabolism should help clarify the pathophysiology of diabetes.
Nuclear receptors are part of a large superfamily of transcription factors that includes receptors for steroids, retinoic acid, and thyroid hormones [6]. RXRs are heterodimeric partners of many nuclear receptors, such as retinoic acid receptors (RARs), thyroid hormone receptors (T3Rs), liver X receptors (LXRs), peroxisome proliferator activated receptors (PPARs), and RXRs themselves [6]. Among RXR heterodimer partners, activation of PPARd in the skeletal muscle increases insulin sensitivity [7]. The RXR subfamily consists of RXRa, RXRb, and RXRc [6,8]. Although RXRc is preferentially expressed in the skeletal muscle, its functional role is poorly understood. We have recently found that expression of retinoid X receptor c (RXRc) is changed in the skeletal muscle under nutritional conditions; RXRc mRNA expression is down-regulated by fasting and recovered by refeeding [9]. In an attempt to explore the role of RXRc in the skeletal muscle, we established transgenic mice overexpressing RXRc in the skeletal muscle (RXRc mice) and found that they exhibit increased triglyceride contents in the skeletal muscle as a result of increased expression of sterol regulatory element binding protein 1c (SREBP1c), a transcriptional master regulator of lipogenesis [9]. Indeed, RXRc has been shown to enhance SREBP1c gene expression in C2C12 myocytes in vitro at least in part by heterodimerization with LXR [9]. On the other hand, we also found that blood glucose levels are lower in RXRc mice than in control mice [9]. These observations, taken together, suggest that RXRc plays a critical role in glucose and lipid metabolism in the skeletal muscle. However, the molecular mechanism involved in RXRc regulation of glucose metabolism in the skeletal muscle and how it affects systemic glucose metabolism are poorly understood.
Here, we demonstrate enhanced glucose metabolism with increased Glut1 expression and glucose uptake in RXRc mice. This study suggests that activation of the skeletal muscle RXRc is a novel therapeutic strategy to treat or prevent type 2 diabetes.

Animals
C57BL6 mice were purchased from Charles River Japan (Yokohama, Japan). Generation of RXRc mice under the control of the human a-actin promoter was described previously [9]. They were allowed free access to food (CRF-1; Charles River) and water, unless otherwise stated. All animal experiments were approved by Institutional Animal Care and Use Committee of Tokyo Medical and Dental University (approval ID: No. 0090041).

Blood analysis
Serum samples were obtained from mice, when fed ad libitum. Serum glucose levels were measured by the blood glucose test meter (Glutest PRO R; Sanwa-Kagaku, Nagoya, Japan). Serum concentrations of insulin were determined by the enzyme-linked immunosorbent assay (ELISA) kits (Morinaga Institute of Biological Science, Inc., Yokohama, Japan).

Glucose and insulin tolerance tests
For glucose tolerance test, D-glucose (1 mg/g of body weight, 10% (w/v) glucose solution) was administered by intraperitoneal injection after an overnight fast. For insulin tolerance test, human insulin (Humulin R; Eli Lilly Japan K.K., Kobe, Japan) was injected intraperitoneally (0.75 mU/g of body weight), when fed ad libitum.

Hyperinsulinemic-euglycemic clamp studies
We analyzed as described previously [10][11][12] with slight modification. Two days before the clamp studies, a catheter was inserted into the right jugular vein for infusion under general anesthesia with sodium pentobarbital. Studies were performed on mice under conscious and unstressed conditions after a 4-h fast. The clamp study began with a prime (4 mg/kg for 5 min) of [6,6-2 H] glucose (Cambridge Isotope Laboratories, Inc, Andover, MA) followed by continuous infusion at a rate of 0.5 mg/kg per minute for 2 hr to assess the basal glucose turnover. After the basal period, hyperinsulinemic-euglycemic clamp was conducted for 120 min with a primed/continuous infusion of human insulin (25 mU/kg prime for 5 min, 5 mU/kg/min infusion) and variable infusion of 20% glucose to maintain euglycemia (approximately 100 mg/dl). The 20% glucose was enriched with [6,6-2 H] glucose to approximately 2.5% as previously described [12]. To determine the enrichment of [6,6-2 H]glucose in plasma at basal and insulin stimulated state, samples were deproteinized with trichloroacetic acid and derivatized with p-aminobenzoic acid ethyl ester. The atom percentage enrichment of glucose m+2 was then measured by high-performance liquid chromatography with LTQ-XL-Orbitrap mass spectrometer (Thermo Scientific, CA). The glucose m+2 enrichment was determined from the m/z ratio 332.2: 330.2. The hepatic glucose production was calculated by using the rate of infusion of [6,6-2 H]glucose over the atom percent excess in the plasma minus the rate of glucose being infused. The insulinstimulated whole-body glucose uptake was calculated by adding the total glucose infusion rate plus the hepatic glucose production [12].

Muscle incubation and glucose transport
Glucose transport was measured as described [15]. Mice were fasted overnight and killed. The extensor digitorum longus (EDL) muscles were rapidly removed, each of which was mounted on the incubation apparatus and preincubated in Krebs-Ringer bicarbonate (KRB) buffer containing 2 mmol/l pyruvate for 30 min. The muscles were then incubated in KRB buffer in the absence or presence of 50 mU/ml insulin for 10 min. The buffers were kept at 37uC throughout the experiment and gassed continuously with 95% O 2 and 5% CO 2 . Immediately after incubation, the muscle was transferred to KRB buffer containing 1 mmol/l 2-

Measurement of skeletal muscle glycogen
The skeletal muscle glycogen content was measured as glycosyl units after acid hydrolysis [16]. The skeletal muscle samples were minced, 50 mg of which was added to 1 ml of 0.3 M percholic acid and homogenized on ice. One ml of 1 N HCl was added and incubated for 2 h at 100uC, thereafter, 1 ml of 1 N NaOH was added at room temperature. Glucose content was examined using the F-kit glucose (Roche Diagnostics, Mannheim, Germany).

Electroporation and in vivo luciferase reporter analysis
In vivo electroporation was performed according to the modified method of Aihara and Miyazaki [17]. Under the pentobarbital anesthesia (30 mg/kg), bilateral quadriceps muscles from C57BL6  mice (male, 12 weeks of age) were injected with 80 mg of plasmid DNA (25 ml) by using a 29-gauge needle attached to a 0.5-ml insulin syringe (Terumo Corporation, Tokyo, Japan). Square-wave electrical pulses (160 V/cm) were applied six times with an electrical pulse generator (CUY21EDIT, Nepa Gene Co. Ltd., Chiba, Japan) at a rate of one pulse per second, with each pulse being 20 ms. in duration. The electrodes were a pair of stainless steel needles inserted into the quadriceps muscles and fixed 5 mm apart. Seven days after gene delivery, the muscles were removed and subjected to analysis. Frozen muscle tissues were homogenized in ice-cold passive lysis buffer from Promega. The homogenate was centrifuged at 10,000 g for 10 min at 4uC. The supernatant was reserved for luciferase assay using Promega's dual luciferase assay kit. The luciferase activity was calculated as the ratio of firefly to Renilla (internal control) luciferase activity and represented as the average of triplicate experiments.

Computer-based DNA sequence motif search
We used MATCH software [18] (BIOBASE GmbH, Wolfenbuettel, Germany) to investigate transcriptional binding sites in the mouse Glut1 promoter. We investigated mouse genome in the region of -1500 to +100 relative to transcription start of Glut1 (Chr.4 11878131(+)).
cDNA microarray analysis RNA was isolated from the skeletal muscle of sex-and agematched RXRc mice (line 4-3) and non-transgenic control mice (females at 4 months of age, five samples from each group were combined). Each of the combined samples was hybridized to the Affymetrix MG430 microarray, which contains 45,102 genes, including expressed sequence tags (ESTs), and analyzed with the software Affymetrix Gene Chip 3.1. Of the 45,102 genes including ESTs analyzed, 8,054 (non-transgenic control mice) and 8,083 (transgenic) were expressed at a substantial level (absolute call is present and average difference is above 150). In order of fold changes in gene expression levels in skeletal muscle from RXRc mice relative to control mice, genes whose expression was increased more than 2 0.4 -fold in RXRc mice were listed (Dataset S1). Fold changes were calculated as an indication of the relative change of each transcript represented on the probe array. Differentially expressed genes were identified using the following criteria; 'absolute call' is present, and 'average difference' was above 250. 'Absolute call', which was calculated with this software using several markers, is an indicator of the presence or absence of each gene transcript. The 'average difference' value is a marker of the abundance of each gene, obtained by comparing the intensity of hybridization to 20 sets of perfectly matched 25-mer oligonucleotides relative to 20 sets of mismatched oligonucleotides using Affymetrix Gene Chip 3.1 software. All data of microarray is MIAME compliant and that the raw data has been deposited in a MIAME compliant database (GEO), whose accession number is GSE28448.

Gene Ontology Analysis
We used DAVID v6.7 [19] for gene ontology (GO) analysis. DAVID is a web application providing a comprehensive set of functional annotation tools to understand the biological meaning behind a large list of genes. Functional Annotation Clustering of DAVID was applied to the genes whose gene expression increased in RXRc mice. Our GO analysis produced 62 GO terms from gene sets from RXRc mice with increased expression compared to the wild-type mice, under the condition of P,0.05 (P value from Fisher's Exact Test). The obtained GO terms contained many similar functional concepts. In order to group similar GO terms, we applied the Functional Annotation Clustering tool provided by DAVID [19,20]. Twenty-three clusters were produced with genes showing increased expression. We showed one GO term of the lowest P value of all the members of an individual cluster.

Transcriptional factor binding sites analysis
We used MATCH software [18] with BKL TRANSFAC 2010.3 Release (BIOBASE GmbH, Worfenbuettel, Germany) to investigate transcriptional binding sites in the promoter regions of genes. The F-Match algorithm compares the number of sites found in a query sequence set against the background set. It is assumed, if a certain transcription factor (or factor family), alone or as a part of a cis-regulatory module, plays a significant role in the regulation of the considered set of promoters, then the frequency of the corresponding sites found in these sequences should be significantly higher than expected by random chance. We investigated the mouse genome in the region of -1000 to +100 relative to the transcription start of an individual gene. Statistical hypothesis testing was evaluated against housekeeping genes of mice. We investigated promoter regions of 15 genes: that is 14 'glucose metabolic process' genes with increased expression in RXRc mice, plus Glut1. In the GO analysis, Glut1 was categorized as a 'transporter' not 'glucose metabolic process' gene.

Statistical analysis
Statistical analysis was performed using the Student's t test and analysis of variance (ANOVA) followed by Scheffe's test. Data were expressed as the mean 6 SE. P,0.05 was considered statistically significant.

Increased glucose metabolism in RXRc mice
In our previous study, we established two lines of RXRc mice (named lines 4-3 and 5-3) with similar expression levels of the RXRc transgene and protein specifically in the skeletal muscle [9]. There was no significant difference in body weight, adipose tissue, and liver weight between RXRc and control mice in both lines 4-3 and 5-3, although the skeletal muscle weight was slightly lower in RXRc mice than in control mice (Tables 1 and 2). In this study, blood glucose in RXRc mice was lower than in control mice (fasting state, P,0.01 in lines 4-3 and 5-3; basal state, P,0.05 in line 4-3 and P = 0.06 in line 5-3), which is consistent with our previous report [9].
To elucidate the role of the skeletal muscle RXRc in systemic glucose metabolism, we performed glucose and insulin tolerance tests in RXRc mice. The glucose tolerance test revealed increased glucose disposal in RXRc mice relative to control mice (Fig. 1A). On the other hand, there was no significant difference in the insulin-induced hypoglycemic response between genotypes (Fig. 1B). These observations suggest that RXRc mice have a higher capacity for glucose disposal with no change in insulin sensitivity.
Increased Glut1 expression and glucose uptake in the skeletal muscle of RXRc mice As both lines of RXRc mice showed increased glucose metabolism, for the following experiments, we only utilized the line 4-3 of RXRc mice (hereafter just RXRc mice). Interestingly, mRNA expression of Glut1 was increased in the skeletal muscle from RXRc mice relative to control mice (P,0.01), whereas that of Glut4 was unchanged ( Fig. 2A). We also observed that Glut1 is significantly increased in RXRc mice at the protein level (P,0.05), with no significant difference in Glut4 between genotypes (Fig. 2B). Consistently, we also found increased glucose uptake in the skeletal muscle from RXRc mice relative to control mice (P,0.05), which was not further enhanced in the presence of insulin (Fig. 2C). We also observed increased glucose glycogen content in the skeletal muscle from RXRc mice relative to control mice (P,0.05, Fig. 2D).

Increased basal glucose disposal rate of RXRc mice
To gain further insight into the glucose metabolism in RXRc mice, we performed a hyperinsulinemic-euglycemic clamp study. Plasma insulin concentrations during the basal period were similar between genotypes ( Table 3). The basal glucose disposal rate was significantly increased in RXRc mice than in the controls (P,0.05, Fig. 3A), supporting that an insulin-independent increase in glucose uptake occurred, as observed in Fig. 2C. Meanwhile, we observed that the rate of glucose infusion needed to maintain euglycemia (glucose infusion rate) was similar between genotypes (Fig. 3B), which is consistent with the result of the insulin tolerance test (Fig. 1B). On the other hand, insulin-stimulated glucose disposal rate was higher in RXRc mice than in the controls (P,0.01, Fig. 3C), which probably reflects the increased basal glucose disposal rate (Fig. 3A). Also, the clamp hepatic glucose production (hepatic glucose production during the clamp period) was higher in RXRc mice than in the controls (P,0.05, Fig. 3D). Meanwhile, hepatic glucose production was similarly suppressed by insulin in both genotypes (Fig. 3E). Together, these data support the idea that Glut1, an insulin-independent glucose transporter, is involved in the increased glucose disposal in the skeletal muscle of RXRc mice.

Activation of the Glut1 promoter by combination of RXRc and PPARd in the skeletal muscle in vivo
To examine whether RXRc directly activates mRNA expression of Glut1, we performed the in vivo luciferase reporter analysis  Up-regulated genes in RXRc mice in the microarray, classified as 'glucose metabolic process' genes by GO analysis, as described in Method. Genes are listed in alphabetic order of gene symbol. Glut1, which appeared in the upregulated list in the microarray (Dataset S1), is not included in this list, as it was classified 'transporter' in the GO analysis. doi:10.1371/journal.pone.0020467.t005 using Glut1 promoter (Glut1-Luc). In this study, the activity of Glut1-Luc was marginally enhanced by RXRc alone (Fig. 4A). Moreover, we found no significant activation of the Glut1-Luc by PPARd, an important regulator of glucose as well as lipid metabolism in the skeletal muscle [7]. Interestingly, combination of RXRc and PPARd resulted in significant increase in Glut1-Luc activity in the skeletal muscle in vivo (P,0.01) (Fig. 4A). Motif search analysis revealed two putative PPAR-responsive elements (PPRE1 and PPRE2) (2657/2645 and 2520/2508, respectively) in the mouse Glut1 promoter. A series of deletion mutant analysis showed that combination of RXRc and PPARd activates the regions of 21500/+75, and 2647/+75, but not the region of 2152/+75 in the Glut1 promoter (Fig. 4B), suggesting that the second putative PPRE (2520/2508) is involved in the RXRc/ PPARd-induced Glut1 promoter activation.

Microarray and bioinformatics analyses of up-regulated gene in RXRc mice
In order to gain insight into the gene expression change in RXRc mice, we performed microarray analysis. As shown in Dataset S1, 738 genes were up-regulated in the analysis. As expected, Glut1 expression was increased in the microarray data. Also, SREBP1c expression, which we previously reported [9], was increased in the microarray data. Using the data, we performed GO analysis to determine if genes, up-regulated in RXRc mice, are associated with particular biological processes. Our GO analysis revealed genes with increased expression in the RXRc mice in various categories (Table 4), including 'glucose metabolic process' genes and 'fatty acid biosynthetic process' genes, which indicated that overexpression of RXRc affects the expression of many genes. The up-regulated genes categorized as glucose metabolism genes in GO term are listed in Table 5. Among them, we confirmed enhanced gene expression by quantitative real time PCR (Fig. 5), supporting the microarray data reliable. Moreover, we calculated the ratio of putative transcription factor binding motifs in glucose metabolism genes, which were up-regulated in RXRc mice. In the sample, several motifs showed statistical significance (Table 6) (P,0.05), including PPAR responsive elements. These data suggest that glucose metabolism genes upregulated in RXRc mice are possible target genes of the RXRc and PPAR heterodimer.

Discussion
RXRc is a nuclear receptor-type transcription factor that is expressed abundantly in the skeletal muscle and is regulated by nutritional conditions. Treatment of obese and diabetic mice with RXR pan-agonists (agonists for all the RXR isoforms) has improved glucose metabolism in mice [21][22][23], suggesting the beneficial effect of RXR on diabetes. However, which RXR isoform(s) are involved and where they work to improve diabetes has not been addressed. Here, we investigated glucose metabolism in RXRc mice.
We demonstrated increased glucose metabolism in RXRc mice relative to control mice, although there was no significant difference in the insulin-induced hypoglycemic effect between genotypes, based on insulin tolerance test and glucose clamp analysis. Because insulin level is similar between RXRc mice and control mice, it is unlikely that increased insulin secretion is responsible for increased glucose tolerance in RXRc mice. On the other hand, mRNA and protein expression of Glut1 is increased in the skeletal muscle from RXRc mice relative to control mice, with increased glucose uptake and glycogen content. Because transgenic overexpression of Glut1 in the skeletal muscle has resulted in increased glucose uptake [24], glycogen content [25] and lowering of blood glucose [24], it is likely that enhanced glucose tolerance in RXRc mice is mediated at least in part by increased Glut1 in the skeletal muscle.
As RXRc is a nuclear receptor-type transcription factor that heterodimerizes with many nuclear receptors [6], we examined whether the Glut1 gene is directly regulated by RXRc. In an in vivo luciferase reporter analysis, combination of RXRc and PPARd activates the Glut1 promoter activity, which is diminished by deletion of the putative PPREs. It is, therefore, RXRc/PPARd may activate Glut1 expression in the skeletal muscle in vivo. In this regard, we found that overexpression of RXRc or PPARd alone or both in C2C12 myocytes in vitro does not induce Glut1 gene expression (unpublished data). This may be because C2C12 cells lack other factor(s) that are present in the skeletal muscle in vitro and are required to activate Glut1 transcription. Whether the increased Glut1 gene in RXRc mice is mediated by RXRc/PPARd should be confirmed by additional experiments using PPARd knockout mice. Meanwhile, the transgene expression level in RXRc mice was very high, and appeared to be beyond the physiological level. Nonetheless, if RXRc can be enhanced in skeletal muscle, an increase in glucose metabolism can be expected.
We recently demonstrated that in addition to glycogen content, RXRc mice exhibit increased triglyceride in the skeletal muscle as a result of increased SREBP1c gene expression [9]. In obese and diabetic subjects, intramuscular lipid is high with insulin resistance [26]. Moreover, in athletes, the skeletal muscle is also high in lipid but with high insulin sensitivity, which is known as the athlete paradox [27,28]. A partial explanation is an increased fatty acid load in obese and diabetic subjects, but not in athletes; increased intramuscular lipotoxic fatty acid metabolites, such as diacylglycerol and ceramide, cause insulin resistance in skeletal muscles [29,30]. Taken together with our previous report [9], this study demonstrates that RXRc mice do not develop insulin resistance despite high intramuscular lipids, and appear to be protected against obesity-induced lipotoxicity in the skeletal muscle. Depositing glucose as both glycogen and triglycerides in the skeletal muscle may be effective in enhancing glucose metabolism.
Microarray analysis showed that various genes are up-regulated in the skeletal muscle of RXRc mice. As RXRc can heterodimerize several nuclear receptors, it is not surprising that expression of many genes was up-regulated by RXRc overexpression. Concerning glucose metabolism genes, in the skeletal muscle of RXRc mice, the expression levels of genes, such as glucose phosphate isomerase 1 and phosphoglycerate mutase 1 (stimulate glycolysis or gluconeogenesis), fructose bisphosphatase 2 (stimulates gluconeogenesis), pyruvate dehydrogenase kinase Table 6. Possible transcription factor binding sites in the 'glucose metabolic process' gene up-regulated in RXRc mice. The mouse genome in the region of 21000 to +100 relative to the transcription start of an individual gene classified as glucose metabolism gene by GO analysis (Table 5), was analyzed. Statistical hypothesis testing was evaluated against housekeeping genes of mice. Matrics names are based on the MATCH software (see Methods), which are listed in P-value order. Indicated binding sites of transcription factors appeared significantly more frequent occurrence in the promoter sets. doi:10.1371/journal.pone.0020467.t006 isoenzyme 1 (suppresses glycolysis), glycogen synthase 2 (stimulates glycogen synthesis) [31,32], were markedly increased (Fig. 5), suggesting the anabolic reaction of glucose. This is consistent with the observation that the glycogen content of skeletal muscle is higher in RXRc mice (Fig. 2C). It is of note that glycogen synthase 2 is a liver-type enzyme, not a skeletal muscle-type enzyme [32]. In addition, usually, gluconeogenesis is not considered to be a major metabolic pathway in skeletal muscle. Chronic transgenic overexpression of RXRc may have caused a super-physiological gene expression change in the skeletal muscle. Meanwhile, our bioinformatics analysis showed that glucose metabolism genes upregulated in RXRc mice contain several transcription factor binding motifs including PPRE in their promoter region. These observations support the concept that the RXRc/PPAR heterodimer contributes to activation of these gene sets, although this needs to be confirmed by further experiments.
In summary, we demonstrated enhanced glucose metabolism with increased Glut1 expression and glucose uptake in RXRc mice. This study suggested that activation of the skeletal muscle RXRc is a novel therapeutic strategy to treat or prevent type 2 diabetes.

Supporting Information
Dataset S1 List of genes up-regulated in RXRc mice compared with wild-type control mice by microarray. (XLS)