Increased glucose metabolism in Arid5b−/− skeletal muscle is associated with the down-regulation of TBC1 domain family member 1 (TBC1D1)

Background Skeletal muscle has an important role in regulating whole-body energy homeostasis, and energy production depends on the efficient function of mitochondria. We demonstrated previously that AT-rich interactive domain 5b (Arid5b) knockout (Arid5b−/−) mice were lean and resistant to high-fat diet (HFD)-induced obesity. While a potential role of Arid5b in energy metabolism has been suggested in adipocytes and hepatocytes, the role of Arid5b in skeletal muscle metabolism has not been studied. Therefore, we investigated whether energy metabolism is altered in Arid5b−/− skeletal muscle. Results Arid5b−/− skeletal muscles showed increased basal glucose uptake, glycogen content, glucose oxidation and ATP content. Additionally, glucose clearance and oxygen consumption were upregulated in Arid5b−/− mice. The expression of glucose transporter 1 (GLUT1) and 4 (GLUT4) in the gastrocnemius (GC) muscle remained unchanged. Intriguingly, the expression of TBC domain family member 1 (TBC1D1), which negatively regulates GLUT4 translocation to the plasma membrane, was suppressed in Arid5b−/− skeletal muscle. Coimmunofluorescence staining of the GC muscle sections for GLUT4 and dystrophin revealed increased GLUT4 localization at the plasma membrane in Arid5b−/− muscle. Conclusions The current study showed that the knockout of Arid5b enhanced glucose metabolism through the downregulation of TBC1D1 and increased GLUT4 membrane translocation in skeletal muscle. Electronic supplementary material The online version of this article (10.1186/s40659-020-00313-3) contains supplementary material, which is available to authorized users.


Background
AT-rich interactive domain 5B (ARID5B) is a member of the AT-rich interactive domain (ARID) transcriptional factor family, which consists of seven subfamilies, ARID1 through ARID5, Jumonji ARID 1 (JARID1) and JARID2. The ARID proteins play diverse roles in cell proliferation, cell cycle control, differentiation, senescence and development [1]. Several previous reports have shown that Arid5b regulates cell differentiation of the mesenchymal stem cell (MSC) lineage; in vitro studies demonstrated that adipogenesis was significantly suppressed in 3T3-L1 preadipocytes by knockdown of Arid5b and in mouse embryonic fibroblasts (MEFs) isolated from Arid5b knockout (Arid5b −/− ) mice [2]. In addition, recent studies have shown that Arid5b promotes differentiation in skeletal muscle satellite cells (SCs) and chondrocytes; Murray et al. demonstrated that the deletion of Arid5b in SCs impaired prostaglandin (PG) I 2 (PGI 2 ) production through the downregulation of cyclooxygenase-1 (COX-1) and PGI 2 synthase (PTGIS). Decreased production of PGI 2 increased cell migration and inhibited myotube formation and fusion in Arid5b −/− SCs, which were rescued by supplementation with the PGI 2 analog, iloprost [3]. Hata et al. identified ARID5B as a co-regulator of SRYbox transcription factor 9 (SOX9) that recruits plant homeodomain (PHD) finger protein 2 (PHF2), a histone demethylase, to the promoters of Sox9-target genes, such as collagen type II alpha 1 chain (Col2a1) and aggrecan. The authors showed that the ARID5B-PHF2 complex removes the dimethylated lysine 9 on histone 3 (H3K9Me2) repression mark and activates transcription of these target genes to facilitate chondrogenesis in mice [4].
Interestingly, we found that Arid5b −/− mice had reduced white adipose tissue mass relative to Arid5b wild-type (Arid5b +/+ ) mice and were resistant to high-fat diet (HFD)-induced obesity [5]. Additionally, knock-down of Arid5b in 3T3-L1 adipocytes activated free fatty acid recycling into cellular triglyceride by increasing both lipolysis and triglyceride synthesis [5,6]. Moreover, in hepatocytes, it has been demonstrated that ARID5B forms a complex with PHF2 on the promoters of its target genes, phosphoenolpyruvate carboxykinase (Pepck) and glucose-6 phosphatase (G6Pase), in response to protein kinase A (PKA) activation. The ARID5B-PHF2 complex activates transcription of these genes by removing the repressive H3K9Me2 mark from their promoters [7]. These findings suggest Arid5b may play a role in energy metabolism in various cell types.
Skeletal muscle is one of the main tissues utilizing glucose and fatty acids for ATP generation and maintains whole-body energy homeostasis by responding to energy demands and nutrient availability [8]. Skeletal muscle takes up glucose from the blood in response to insulin stimuli and maintains blood glucose levels [9]. In fact, impairment of glucose transport into skeletal muscle due to insulin resistance in type II diabetes is a serious condition that eventually disrupts functions of other tissues [10]. In addition, the activity of mitochondrial oxidative metabolism determines the efficiency of energy production from both glucose and fatty acids [8]. Since the consequence of Arid5b deletion in skeletal muscle metabolism has not been studied, we investigated the potential role of Arid5b in skeletal muscle metabolism using Arid5b −/− mice. In this study, we report that basal glucose uptake and glucose oxidation were enhanced in Arid5b −/− skeletal muscle and were associated with the downregulation of TBC domain family member 1 (TBC1D1) expression.

Animal studies
Arid5b −/− mice were established by homologous recombination with a PGK promoter-driven neomycin cassette as described previously [5]. Mice were fed standard chow ad libitum and maintained under controlled lightdark cycles (light cycle 6AM to 6PM). 10 to 21 week-old, standard chow-fed male mice were used for all experiments and euthanized between 11AM-2PM unless otherwise specified. Movements, O 2 consumption and CO 2 production, and food intake of mice were recorded using the TSE PhenoMaster V5. 9 were isolated from mice. Whole SoL or EDL muscles were used for oxidation assays and glucose uptake assays.

Quantitative real-time PCR (qRT-PCR)
Skeletal muscle tissue powder (30 mg) was homogenized with the POLYTRON ® homogenizer (Kinematica, #PT2100) in 700 μL of QIAzol ® Lysis Reagent. Total RNA was purified from the homogenate using the miRNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. Reverse transcription was performed using 0.75 μg RNA following the instructions of the iScript ™ cDNA synthesis kit (Bio-Rad). Experimental transcript levels were analyzed using iQTM SYBR Green Supermix (Bio-Rad) on the CFX96 ™ Real-Time System. All qRT-PCR data was normalized to ribosomal protein L13a (Rpl13a) expression analyzed in separate reactions. Primer sequences are listed in Table 1.

Western blotting analysis
Whole-tissue protein lysate was prepared from powdered GC muscle by sonication with the Q700 sonicater For fatty acid oxidation assays, isolated skeletal muscles were pre-incubated in KHB buffer supplemented with 10 mM HEPES, 5 mM glucose, and 4% fatty-acid free BSA (hereafter referred to as fatty acid oxidation assay buffer) on ice and transferred to 37 °C with 5% CO 2 for 30 min before the assay. Palmitic acid-[9, 10-3 H]-BSA complex was prepared before dissection of skeletal muscles by following steps. 1.6 × 10 -6 μmol of palmitic acid-[9, 10-3 H] (Moravek, Inc., #MT-845) and 4 μmol of cold palmitic acid were mixed with 3 μmol of KOH and the mixture was dried under nitrogen gas. The dried palmitic acid-[9, 10-3 H] was complexed to BSA by resuspension in 25% BSA solution to give a final concentration of 1.6 × 10 -6 mM palmitic acid-[9, 10-3 H] and 4 mM cold palmitic acid (fatty acid-BSA stock). Fifty microliters of fatty acid-BSA stock was added to 1 mL of fatty acid oxidation assay buffer in round bottom culture tubes (12 × 75 mm, #110428, Globe Scientific Inc.), and skeletal muscle was incubated in the buffer for two hours at 37 °C under 5% CO 2 . After incubation, 400 μL of supernatant was transferred into a new 15 mL tube, and 2 mL of chloroformmethanol (2:1, vol/vol) was added in order to separate 3 H 2 O from palmitic acid-[9, 10-3 H] in the supernatant. The solution was vortexed for 10 s and centrifuged briefly at 3000 × g at room temperature (RT). 0.8 mL of 2 M KCl-2 M HCl was added to the solution, vortexed again and centrifuged at 3000 × g for 5 min at RT. Radioactivity of 0.5 mL of the aqueous phase was measured for [ 3 H] for 5 min per sample with the Beckman Liquid Scintillation Counter ™ (#LS6500). Skeletal muscles used for glucose and fatty oxidation assay were washed in dH 2 O, placed on aluminum foil and dried at 60 °C for 20-24 h. Dry tissue weights were measured and used for normalization of radioactivity of each sample.

Quantification of metabolites and enzyme activity
Powdered GC muscles (10-20 mg per assay) were sonicated in the appropriate assay buffer (Q700 ice-water bath sonicator, QSONICA) for six cycles of amplitude 35 for 30 sec followed by 1 min on ice, centrifuged at 10,000 × g for 10 min at 4 °C and the lysates were subjected to the following assays: measurement of the metabolites and citrate synthase (CS) activity. Lactate Colorimetric Assay (Biovision, #K607-100), Glycogen Colorimetric Assay (Biovision, #K648-100), and Citrate Synthase Activity Colorimetric Assay (Biovision, #K318-100) were performed according to the manufacturers' instructions.

Nucleotide extraction
Powdered GC muscles (10-15 mg) were sonicated in 3% trichloroacetic acid (TCA)/PBS (Q700 ice-water bath sonicator, QSONICA) for six cycles at amplitude 35 for 30 s followed by 1 min on ice and centrifuged at 12,000×g for 5 min at 4 °C. The supernatant was transferred into a new ice-cold tube, diluted with an equal volume of ddH 2 O, and pH was adjusted to neutral with 1 M KOH. The amounts of ATP, ADP and AMP in each sample were measured by UV-HPLC in the Analytical Pharmacology Core at City of Hope (Duarte, CA). The pellet after centrifugation was dissolved in 5% sodium dodecyl sulfate (SDS)/ 0.1 N NaOH, and protein concentration was measured by BCA assay (Thermo Fisher Scientific).

Intraperitoneal glucose tolerance test (IP-GTT) and plasma insulin concentration
The mice were fasted for 7-8 h before the glucose tolerance test was conducted. Whole blood samples were collected from tails of mice before (0 min), and 10, 20, 30, 60 and 120 min after an i.p. injection of glucose (2 mg D-glucose/g body weight) for the measurement of blood glucose levels with a Clarity Diagnostics BG1000 glucose meter. The plasma was prepared by centrifugation at 1,000×g for 10 min. The plasma insulin levels were determined using Insulin (mouse, rat) EIA kits (Cayman Chemical, #589501) by following the manufacturers' instructions. The plasma samples collected during IP-GTT were spiked with 1 ng/mL of insulin standard provided in the kit. The data were analyzed using the analysis workbook (ELISADouble) available in Cayman's website for the determination of the insulin concentration of standards and the spiked plasma samples. The insulin concentration of the plasma samples was calculated by subtracting the average concentration of 1 ng/mL of insulin standard samples (duplicate) from the spiked plasma samples. The area under the curve (AUC) was calculated from the IP-GTT curve using the trapezoidal rule.

Coimmunofluorescence staining
GC muscle was frozen immediately after isolation in isopentane cooled with liquid nitrogen. GC muscle was cryosectioned at a thickness of 10 µm and affixed to Tissue Path Superfrost Plus Gold microscope slides (Fisher Scientific). Sections were fixed in 4% formaldehyde in PBS and then permeabilized in 0.15% Triton-X100 (Fisher Scientific) in PBS. Antibodies to GLUT4 (#MA5-17176, 1:500, Thermo Fisher Scientific) and dystrophin (#ab15277, 1:400, Abcam) were applied simultaneously to the sections for 2 h at RT. After incubation with primary antibodies, sections were washed 3 times in PBS for 5 min each time. Secondary antibodies were applied to the sections for 45 min at RT at a dilution of 1:500. For GLUT4 we used goat anti-mouse Alexa Fluor 488 (Thermo Fisher Scientific), and for dystrophin we used goat anti-rabbit Alexa Fluor 555 (Thermo Fisher Scientific) secondary antibodies. Sections were then washed 3 times in PBS for 5 min each time and mounted in ProLong Glass Antifade Mountant (Thermo Fisher Scientific). Images (n > 10 per mouse) were taken on the Zeiss LSM 880 inverted confocal microscope (Carl Zeiss, Inc.). All the images were captured with the same laser intensities. The Pearson's correlation coefficient for quantitating colocalization of GLUT4 and dystrophin was calculated for each image using Zen Black software (Carl Zeiss, Inc.).

Statistical analysis
Data were expressed as the means ± SD. Statistical significance between Arid5b +/+ mice and age-matched Arid5b −/− mice was assessed by Student's t test (unpaired, two tailed). A P value of < 0.05 was considered significant.

Deletion of Arid5b in mice did not alter normalized skeletal muscle weight
To confirm that the Arid5b gene is knocked out in skeletal muscle, its mRNA expression was examined in GC muscle with a primer set designed in the deletion region of Arid5b gene (Table. 1). Arid5b mRNA expression was undetectable by qRT-PCR in Arid5b −/− GC muscle (Fig. 1a). We previously reported that the body weight of the Arid5b −/− mice was significantly reduced [5]. Weights of GC, QC, and SoL were significantly decreased in Arid5b −/− mice compared to Arid5b +/+ mice because of the reduction in body weight (Fig. 1b, c), while the weight of EDL did not change. However, the normalized weight of individual skeletal muscles was similar between the genotypes (Fig. 1d, e). Notably, the absolute food intake and normalized food intake relative to body weight did not differ between the genotypes (Fig. 1f ).

Glucose oxidation was enhanced in Arid5b −/− skeletal muscle
To characterize the metabolic phenotype of skeletal muscle in Arid5b −/− mice, we first measured oxygen consumption and found that it was significantly increased in Arid5b −/− mice compared to Arid5b +/+ mice (Fig. 2a). We also measured the basal locomotor activity of the mice and found no significant change in total distance traveled (Fig. 2b). Since oxygen consumption was increased in Arid5b −/− mice, we investigated the substrate oxidation rates in mitochondria using the isolated skeletal muscles, SoL and EDL. In both types of skeletal muscles, the glucose oxidation rate was significantly elevated by 1.3-to 1.8-fold in Arid5b −/− skeletal muscle compared to Arid5b +/+ skeletal muscle (Fig. 2c). The fatty acid oxidation rate did not change in the SoL or EDL of Arid5b −/− mice relative to Arid5b +/+ mice (Fig. 2d). We analyzed lactate content as an end product of anaerobic glycolysis and ATP content in GC muscle and found that lactate content was decreased in Arid5b −/− GC muscle (Fig. 2e) while ATP content was increased (Fig. 2f ). The amount of AMP and ADP and the ratio of AMP/ATP did not change (Additional file 1: Fig. S1). Additionally, the expression and phosphorylation levels of AMPKα1/2 (Thr172) did not change (Fig. 2g). Therefore, the changes in the glucose metabolism in the GC skeletal muscle were not due to AMPK activation.
Next we investigated the potential cause of the increased glucose oxidation in skeletal muscle. Previous studies have shown that enhanced oxidative metabolism in skeletal muscle and other cell types is associated with highly fused mitochondria [12,13]. Therefore, we analyzed the morphology of mitochondria and the integrity of sarcomeres in myofibers using TEM imaging. There were no noticeable changes in the sarcomere arrangement or cristae morphology of mitochondria in white GC muscle between the two genotypes (data not shown). It has been shown that multiple mitochondrial fusion and fission factors, such as mitofusin 1 (MFN1), mitofusin 2 (MFN2), optic atrophy 1 (OPA1) and dynamin-related protein 1 (DRP1), regulate mitochondrial morphology [14,15]; however, we did not observe significant alteration in the expression levels of MFN1, MFN2, OPA1 and DRP1 in Arid5b −/− GC muscle (Additional file 1: Fig. S2). These data suggest that the increase in oxidative metabolism in Arid5b −/− skeletal muscle was not related to mitochondrial dynamics.

The expression of PGC-1α and CS activity did not change in Arid5b −/− GC muscle
To determine if Arid5b −/− skeletal muscle showed an increase in mitochondrial biogenesis, we analyzed the expression levels of PGC-1α and electron transport chain (ETC) subunits in Arid5b −/− skeletal muscle and assessed CS activity in Arid5b −/− skeletal muscle. PGC-1α is a major regulator of mitochondrial biogenesis [16,17], and CS activity is used as a marker of mitochondrial content [18]. The protein expression of PGC-1α was similar between the genotypes (Fig. 3a); moreover, CS activity did not show a difference between the genotypes (Fig. 3b). The expression of the ETC complexes did not change relative to Arid5b +/+ GC muscle (Additional file 1: Fig. S3). These data indicate that the increased oxygen consumption and glucose oxidation in Arid5b −/− skeletal muscle are not due to changes in mitochondrial biogenesis.

The expression levels of myosin heavy chain (Myh) isoforms were not altered in Arid5b −/− GC muscle
To assess whether a fiber-type switch from fast-glycolytic to slow-oxidative could be a potential cause of the increased glucose oxidation in Arid5b −/− skeletal muscle, we analyzed mRNA expression of the Myh isoforms, Myh7, Myh2, Myh1 and Myh4, in GC muscle. We found that Myh1 expression was significantly decreased in Arid5b −/− GC muscle relative to Arid5b +/+ GC muscle (Additional file 1: Fig. S4). The expression of Myh7, Myh2, and Myh4 in Arid5b −/− GC muscle was comparable to Arid5b +/+ GC muscle (Additional file 1: Fig. S4). At the protein level, the expression of MYH1 was not significantly decreased (Fig. 4). The protein expression of MYH7 was not altered in Arid5b −/− GC muscle (Fig. 4). Taken together, although the expression of Myh1 was perturbed at the mRNA level, the myofiber type did not change in Arid5b −/− GC muscle.

Whole-body glucose clearance and glucose transport in skeletal muscle were increased in Arid5b −/− mice
We further characterized systemic glucose metabolism by assessing glucose tolerance in mice and glucose uptake in isolated skeletal muscles. The glucose tolerance test revealed increased glucose clearance in Arid5b −/− mice relative to Arid5b +/+ mice while a b c d e f Fig. 1 Analyses of body composition and food intake of Arid5b +/+ and Arid5b −/− mice. a Arid5b expression in GC muscle. Total RNA was isolated from GC muscle and qRT-PCR analysis was performed. Gene expression was normalized to Rpl13a mRNA expression. Arid5b +/+ , n = 7; Arid5b −/− , n = 3. *, P < 0.05. b, c Absolute muscle weights in Arid5b +/+ and Arid5b −/− mice. GC, QC, SoL and EDL muscles were dissected from male Arid5b +/+ mice and Arid5b −/− mice, and the weight was measured. d, e Normalized skeletal muscle weight was determined by taking the ratio of each skeletal muscle weight to the body weight of the mice. Results are presented as the means ± SD. *, P < 0.05; ***, P < 0.001. GC and QC: Arid5b +/+ mice, n = 18; Arid5b −/− mice, n = 9. SoL and EDL: Arid5b +/+ mice, n = 3; Arid5b −/− mice, n = 4. f Absolute food intake and normalized food intake relative to body weight (BW) in each genotype (n = 5 per group). Data are presented as the means ± SD. *, P < 0.05; **, P < 0.01 (dark period vs light period) insulin concentration significantly decreased at only the 10 min time point in Arid5b −/− mice (Fig. 5a, b). We observed an increase in basal glucose uptake in SoL isolated from Arid5b −/− mice, which was consistent with the increased glucose clearance (Fig. 5c). Glucose uptake was significantly stimulated by insulin treatment in SoL from Arid5b +/+ mice as expected, but it was not stimulated in SoL from Arid5b −/− mice. Because the basal level of glucose uptake in SoL from Arid5b −/− mice was similar to the insulin-stimulated level in SoL from Arid5b +/+ mice, it is possible that uptake in Arid5b −/− SoL is maximally-stimulated in an insulinindependent manner. Additionally, we measured the amount of glycogen in Arid5b −/− GC muscle since its levels are associated with the levels of glucose utilization in skeletal muscle [19]. We found that glycogen content was greater in Arid5b −/− GC muscle compared to Arid5b +/+ GC muscle (Fig. 5d), suggesting increased glucose uptake.

The expression of TBC1D1 was downregulated and GLUT4 translocation to the plasma membrane was increased in Arid5b −/− GC muscle
Since basal glucose uptake was increased in Arid5b −/− skeletal muscle, we speculated that Akt activation and GLUT4 translocation may be altered in Arid5b −/− skeletal muscle at basal levels. We first analyzed the protein expression levels of Akt and its phosphorylation at Ser473 by the mammalian target of rapamycin complex 2 (mTORC2) [20]. We found that the levels of expression and phosphorylation at Ser473 of Akt remained similar in Arid5b −/− GC muscle compared to Arid5b +/+ GC muscle (Additional file 1: Fig. S5), suggesting that the changes in glucose metabolism in Arid5b −/− muscle are not due to alterations in Akt signaling. Several publications suggest that TBC1D1 and TBC1D4 (also known as AS160) negatively regulate GLUT4 translocation from the intracellular space to the plasma membrane through their Rab-GTPase activity. Since these factors are key regulators of glucose uptake in skeletal muscle [21,22], we evaluated their mRNA and protein expression levels in Arid5b −/− GC muscle. Interestingly, protein expression levels of TBC1D1 but not mRNA expression levels were significantly downregulated in Arid5b −/− GC muscle (Fig. 6a and Additional file 1: Fig. S6a) compared to Arid5b +/+ GC muscle whereas expression of TBC1D4 did not change ( Fig. 6b and Additional file 1: Fig. S6b). Since the deletion or downregulation of Tbc1d1 in mice has been shown to influence the protein expression of GLUT4 [11,22,23], we evaluated the basal protein expression of GLUT4 and GLUT1 in Arid5b −/− GC muscle. However, no significant difference was observed in the expression of either protein between the genotypes (Fig. 6c, d).
To determine whether there was an increase in GLUT4 membrane translocation in Arid5b −/− muscle, we performed coimmunofluorescence staining for GLUT4 and dystrophin, a plasma membrane marker, on GC muscle sections from Arid5b +/+ and Arid5b −/− mice. We observed an increase in GLUT4 translocation to the membrane in Arid5b −/− GC relative to Arid5b +/+ GC muscle (Fig. 7a). We then calculated the Pearson's correlation coefficient, which is a commonly used method to quantitate the extent of colocalization of two proteins [24]. The Pearson's correlation coefficient was significantly increased in Arid5b −/− GC compared to Arid5b +/+ GC (P < 0.05) (Fig. 7b). These results indicate that there is enhanced GLUT4 translocation to the plasma membrane in Arid5b −/− GC muscle.
To further confirm our results, we used primary skeletal muscle satellite cells isolated from Arid5b +/+ and Arid5b −/− skeletal muscle [3]. These cells were differentiated for four days, and the assays were performed on day 4 myotubes. First, we evaluated the expression of TBC1D1, GLUT4, and GLUT1 in Arid5b +/+ and Arid5b −/− primary day 4 myotubes. Similar to the Arid5b −/− skeletal muscle, we found that only TBC1D1 expression was down-regulated in Arid5b −/− primary myotubes compared to Arid5b +/+ myotubes while the expression level of GLUT4 and GLUT1 proteins was similar between the genotypes (Additional file 1: Fig. S7a). To confirm the increase in membrane GLUT4 content, we evaluated membrane GLUT4 in primary myotubes from each genotype using sulfo-NHS biotinylation method and immunoprecipitation of the biotinylated proteins with (See figure on next page.) Fig. 2 Glucose oxidation was enhanced in Arid5b −/− skeletal muscles. a Oxygen consumption rates were measured and normalized to body weight (n = 5). *, P < 0.05. b Average distance traveled in two dark vs light periods of 24 h and total distance traveled in 48 h (n = 5) is shown. c Glucose oxidation rates and d fatty acid oxidation rates were measured in isolated skeletal muscles from non-fasted mice. Arid5b +/+ mice (n = 13-15) and Arid5b −/− mice (n = 5-6). *, P < 0.05. e Lactate levels and f ATP content was analyzed in GC muscles. Arid5b +/+ mice (n = 10) and Arid5b −/− mice (n = 5). *, P < 0.05; **, P < 0.01. g The phosphorylation levels of AMPKα1/2 at Thr172 and protein levels of total AMPKαα1/2 were normalized to loading control (HSP70), and the ratio of pAMPKα1/2 (Thr172) to total AMPKα1/2 was calculated. Values represent means ± SD. Arid5b +/+ mice, n = 7; Arid5b −/− mice, n = 5. #, HSP70 was detected in the same membrane as pAMPKα1/2 (Thr172) Okazaki et al. Biol Res (2020) 53:45 avidin beads, as detailed in the methods. We detected membrane and cytosolic GLUT4 and calculated the ratio of membrane GLUT4 content to cytosolic GLUT4 content. As shown in Additional file 1: Fig. S7b, the ratio was higher in Arid5b −/− primary myotubes compared to Arid5b +/+ primary myotubes, suggesting an increase in GLUT4 membrane content. In addition, we conducted IHC for GLUT4 expression in GC muscle sections from mice of each genotype, and the result supports the increased membrane GLUT4 content in Arid5b −/− skeletal muscle (Additional file 1: Fig. S7c). Collectively, these results are in agreement with the coimmunoflurescence data suggesting that the elevated basal glucose uptake in Arid5b −/− skeletal muscle could be due to the downregulation of TBC1D1, which leads to an increase in translocation of GLUT4 to the cell membrane.

Discussion
In this study, we characterized skeletal muscle metabolism in Arid5b −/− mice and showed that glucose metabolism was upregulated in Arid5b −/− skeletal muscle. Our results suggest that the enhancement of glucose clearance in Arid5b −/− mice and glucose uptake in skeletal muscle was insulin-independent and associated with the downregulation of TBC1D1, an inhibitor of GLUT4 translocation.
In Arid5b −/− skeletal muscle, we observed increases in oxygen consumption, glucose oxidation, and ATP content. While these metabolic changes can be associated with morphological changes to the mitochondria, we did not observe alterations in mitochondrial morphology. Also, the expression of factors regulating mitochondrial biogenesis, such as PGC-1α, or factors regulating a b blood glucose concentration. Upon insulin stimuli, GLUT4 localizes to the plasma membrane to incorporate glucose into the cells. Perturbation of glucose uptake into skeletal muscle causes hyperglycemia, which eventually damages vessels and organs, such as kidney and heart [10]. Intracellular GLUT4 localization is controlled by small G protein Rabs, such as Rab8a and Rab14, which are physically associated with GLUT4 vesicles, and GTP-bound active Rabs are necessary to facilitate translocation of GLUT4 [25,26]. TBC1D1 and TBC1D4 are Rab-GTPase-activating proteins (GAPs) that negatively control GLUT4 trafficking through their Rab-GAP activities under basal conditions [25,27,28]. Their activity is inhibited upon stimuli, such as insulin and exercise, through phosphorylation at multiple sites by Akt, AMPK, and Ca 2+ /calmodulin-dependent protein kinase kinase (CaMKK) [29]. Our results show an enhancement of glucose clearance in vivo and glucose uptake ex vivo in Arid5b −/− mice without changes in plasma insulin levels or phosphorylation levels of Akt, which transduces signals from the insulin receptor to activate glucose transport. Instead of altering the insulin signaling pathway, the decreased expression of TBC1D1 in Arid5b −/− skeletal muscle may influence glucose metabolism through GLUT4 translocation without altering GLUT4 expression. TBC1D1 downregulation has been shown to increase GLUT1 expression in cultured adipocytes [30]. However, no changes in GLUT1 expression were observed in Arid5b −/− GC muscle, suggesting that the increased basal glucose uptake was not , and the data are presented as the means ± SD (n = 6). ***, P < 0.001. c, d The expression of GLUT4 and GLUT1 protein in GC muscles is shown. Quantification of GLUT4 and GLUT1 protein in GC muscles is shown. (bottom) Quantitation of GLUT4 protein expression was performed, and the data are presented as the means ± SD (n = 6) due to the induction of GLUT1 expression. Our coimmunofluroescence data demonstrated that GLUT4 localization to the plasma membrane was significantly increased in Arid5b −/− skeletal muscle. These results suggest that enhanced GLUT4 translocation through TBC1D1 downregulation is the potential mechanism for the increased glucose uptake and oxidation in Arid5b −/− skeletal muscle.
Our data are consistent with a previous report in L6 muscle cells showing that the downregulation of TBC1D1 increased basal GLUT4 translocation to the plasma membrane without altering GLUT4 expression [21]. In this regard, the downregulation of TBC1D1 has been reported in Bmal1 skeletal muscle-specific knockout (Bmal1 −/− ) mice; however, insulin-stimulated glucose uptake but not basal glucose uptake was reduced Fig. 7 GLUT4 is localized to the plasma membrane in Arid5b −/− GC muscles. a Coimmunofluorescence analysis was performed for GLUT4 (green) and dystrophin (red) in GC muscle sections, and representative confocal images are shown. Merged images show increased colocalization of GLUT4 and dystrophin at the plasma membrane in Arid5b −/− GC muscles relative to Arid5b +/+ GC muscles. b The Pearson's correlation coefficient quantifying the degree of colocalization was calculated (n > 10 images per mouse, n = 3 mice per genotype). Data are presented as the means ± SD. *, P < 0.05 in Bmal1 −/− skeletal muscle, which was accompanied by a decrease in GLUT4 protein expression [11]. Similarly, Tbc1d1-null mice showed a decrease in insulinstimulated glucose uptake in skeletal muscle and reduced GLUT4 protein expression [23,31]. In contrast, basal glucose uptake and the expression of GLUT4 were increased in skeletal muscle from mice with muscle-specific double knockout of insulin-like growth factor and insulin receptor (MIGIRKO) [22]. The discrepancy in glucose uptake in skeletal muscles from these knockout mice may be explained, at least partially, by the differences in GLUT4 protein levels. The potential mechanism for the regulation of GLUT4 protein levels by TBC1D1 remains to be established.
Moreover, in Arid5b −/− whole skeletal muscle, we observed the downregulation of TBC1D1 at the protein level but not at mRNA level, suggesting that TBC1D1 protein level is decreased by translational regulation or reduced protein stability in Arid5b −/− mice. However, we further analyzed Tbc1d1 mRNA expression in primary myotubes from Arid5b −/− and Arid5b +/+ mice, and found that Tbc1d1 mRNA level was significantly decreased in Arid5b −/− primary myotubes (Additional file 1: Fig. S8). Based on a Target Genes search in ChIP-Atlas for human ARID5B, TBC1D1 is identified as a target gene of ARID5B (data not shown). These observations suggest that there is a possibility that TBC1D1 expression is regulated at mRNA level in myotubes from Arid5b −/− mice.
Skeletal muscle is highly plastic and its proper metabolic functions are important for whole-body energy homeostasis [32]. Due to its high plasticity, skeletal muscle is sensitive to metabolic alterations in other tissues, such as adipose tissues and liver, and systemic factors, such as hormones and cytokines, all of which can affect the characteristics of skeletal muscle [8,33]. In fact, in Arid5b −/− mice, white adipose tissue mass was significantly reduced [5], and we cannot rule out the possibility that the characteristics of Arid5b −/− skeletal muscle we showed in this report are non cell-autonomous. Nevertheless, this study provides additional information that the expression of TBC1D1 is critical for glucose uptake in skeletal muscle. The potential involvement of Arid5b in glucose metabolism through TBC1D1 would provide a new insight in the treatment of diabetes and obesity.

Conclusions
Our results demonstrated that glucose metabolism is enhanced in the skeletal muscle from Arid5b −/− mice, which was associated with the downregulation of TBC1D1. We provided additional information regarding the levels of TBC1D1 expression, glucose uptake, and GLUT4 expression in skeletal muscle. However, more research is necessary to elucidate the potential role of Arid5b in the regulation of TBC1D1 and GLUT4 expression.