Prmt7 Deficiency Causes Reduced Skeletal Muscle Oxidative Metabolism and Age-Related Obesity

With skeletal muscle aging (called sarcopenia), one exhibits decreased muscle mass and exercise capacities accompanied by increased fatigability and muscle atrophy leading to a reduced quality of life (1–3). Skeletal muscle exhibits a remarkable plasticity in energy metabolism and contractile functions in response to various stimuli like exercise, hormones, or nutritional states (4). Defects in the oxidative metabolism and function of skeletal muscle have been implicated in numerous metabolic pathologies, including insulin resistance, glucose intolerance, and obesity (5–8). Skeletal muscle adaptation toward oxidative metabolism promoted by stimuli like endurance exercise involves multiple signaling pathways, including Ca2+/calmodulin-dependent protein kinases, AMPKs, and p38 mitogen-activated protein kinase (p38MAPK), which induce the expression and activation of a key transcriptional coactivator, PGC-1α (9–13). PGC-1α is a critical regulator for mitochondrial function and oxidative muscle metabolism (9,14,15). Consistently, mice lacking PGC-1α exhibit declined muscle oxidative metabolism and exercise capacity (16,17). Furthermore, PGC-1α expression in skeletal muscle is reduced in obesity, diabetes, or denervation associated with the fiber-type switch toward glycolytic fibers (15,18–21). PGC-1α is preferentially expressed in oxidative fibers, and its ectopic expression in skeletal muscle drives oxidative fiber formation, at least in part through MEF2 interactions thereby directly activating oxidative fiber-specific gene expression (22). Among signaling pathways, p38MAPK (p38) plays a critical role for metabolic adaptations of skeletal muscle induced by exercise-triggered muscle contraction, which is mediated through regulation of PGC-1α expression and activity. p38 can phosphorylate the transcription factors, such as ATF2 and MEF2c, to stimulate the transcription of the PGC-1α gene (23–25). In addition, p38 can derepress PGC-1α by phosphorylation and in turn PGC-1α can autoregulate via interaction with MEF2c (24,26). The ectopic activation of p38 by expression of a constitutive active form of MAPK kinase 6 (MKK6EE) enhanced oxidative metabolism by the upregulation of PGC-1α and mitochondrial proteins in skeletal muscles (24). ATF2 binds to the CRE (cAMP response element) sequence in the PGC-1α promoter region in addition to CREB transcription factors to stimulate gene expression (23,24). Recent studies suggest the cross talk and redundancy of signaling pathways aforementioned in regulation of fiber-type switch and oxidative metabolism in response to exercise. Even though PGC-1α was long believed to be indispensible for the muscle adaptation in response to exercise (27,28), mice lacking PGC-1α still showed the exercised-induced fiber-type switch with the reduced expression of mitochondrial genes, suggesting that PGC-1α is specifically required for mitochondrial function involved in muscle adaptation (29). These studies suggest that the signaling-transcription network promoting skeletal muscle adaptation induced by various stimuli including exercise remains to be fully defined.

Protein arginine methyltransferases (Prmts) catalyze symmetric or asymmetric dimethylations of arginine residues on both histone and nonhistone substrates linked to gene regulation in diverse biological processes including glucose metabolism and myoblast differentiation (30,31). Based on dimethylation characteristics, Prmts can be classified as either the type I catalyzing asymmetric arginine dimethylation (Prmt1, Prmt2, Prmt3, Prmt4, Prmt6, Prmt8) or the type 2 subfamily generating symmetric dimethyl-arginine residues (Prmt5 and Prmt7) (30,32). Among these, Prmt1, -4, and -5 are relatively well studied and known to play critical roles in various biological processes. Recent studies have shown that Prmt1 and Prmt5 are involved in regulation of hepatic gluconeogenesis through inhibition of FOXO (33) or through activation of CREB (34). In addition, Prmt1 has been shown to methylate PGC-1α and thereby stimulates the nuclear receptor–mediated expression of mitochondrial genes in nonmuscle cells (35). Prmt4 and Prmt5 have been implicated in promotion of myoblast differentiation through interaction with Mef2c or MyoD, respectively (36,37). Recent studies with muscle stem cell–specific knockout mouse models have suggested that Prmt4 and Prmt5 play critical roles in control of muscle stem cell functions during muscle regeneration (38,39). In this study, we characterized the role of Prmt7 in skeletal muscle metabolism by using Prmt7-deficient mouse models. Prmt7 is highly expressed in skeletal muscle, and its expression declines in skeletal muscles with age and obesity. Interestingly, Prmt7 deficiency results in reduced oxidative metabolism and diminished endurance exercise capacity. Furthermore, Prmt7-deficient mice exhibit age-related obesity with excessive body fat accumulation and hyperglycemia. This function of Prmt7 appears to be partly mediated through regulation of PGC-1α expression by interaction with and activation of p38. Taken together, our data support that Prmt7 is important for muscle function maintenance, thereby contributing to balanced body metabolism.

Male and female Prmt7^{<tm1a(EUCOMM)Wtsi>} mice were purchased from the Sanger Institute. All animal experiments were approved by the Institutional Animal Care and Research Advisory Committee at Sungkyunkwan University School of Medicine Laboratory Animal Research Center. Mice were backcrossed onto C57BL/6J background for at least six generations and maintained on C57BL/6J background, and littermate wild-type controls were used for comparison with Prmt7^−/− mice in all experiments. To assess the age-related effect of Prmt7 deficiency on obesity, Prmt7^+/+ or Prmt7^−/− mice (n = 20–25) were used. The blood glucose levels were measured after fasting for 16 h with free water. For the glucose and pyruvate tolerance test, mice were fasted for 16 h prior to injection with 1.5g/kg body wt i.p. 20% D-(+)-glucose or 20% pyruvate (Sigma-Aldrich), followed by measurement of blood glucose levels. For the insulin tolerance test, mice were fasted for 5 h and injected with 0.5 IU/kg body wt i.p. insulin (Sigma), followed by measurement of blood glucose levels.

For assessment of metabolic parameters, 5-month-old Prmt7^+/+ and Prmt7^−/− mice were analyzed with metabolic cages (Panlab Harvard Apparatus). Measurements were performed for 48 h, during which animals had free access to food and water.

For assessment of muscle endurance capacity, Prmt7^+/+ and Prmt7^−/− mice (n = 10 per group) were tested for the treadmill running and grip strength. For the treadmill running, a Columbus Exer-6M treadmill was used. Prior to exercise, mice were accustomed to the treadmill with a 5-min run at 7 m/min once per day for 5 days. The exercise test was performed on a 10% incline for 8 m/min for 20 min, followed by increase toward 9 m/min until exhaustion. The grip strength of forelimb was measured by a grip strength meter (Bioseb). All grip strength readings (measured in grams) were normalized to body weight. The exercise was assessed every 15 min for a total of 1 h.

Mice tissues and whole blood were collected at the end of experiments. For assessment of metabolite content, blood serum was separated and plasma insulin was measured by mouse insulin ELISA kits (U-type; Shibayagi Corp.). Triglyceride and nonesterified fatty acid (NEFA) in blood serum were measured by colorimetric assay kits (Wako). For measurement of the hepatic triglyceride content, frozen livers were digested in ethanolic KOH at 55°C overnight followed by the extraction of organic layer with 50% ethanol and centrifugation. After addition of 1 μmol/L MgCl₂, the collected organic layer was used for analysis of the triglyceride content.

The blood lactate levels were measured with Prmt7^+/+ and Prmt7^−/− mice (n = 9) before or after a single bout of treadmill running at 10 m/min for 2 h with use of a Lactate Pro2 kit (Arkray).

Liver was processed through a fixation with 4% paraformaldehyde and sucrose series followed by cryo-embedding and sectioning with 10-μm thickness on a cryostat microtome (Leica). Freshly dissected muscles were snap frozen in optimal cutting temperature and sectioned with 7-μm thickness. NADH dehydrogenase activity was determined by incubation for 30 min with 0.9 mmol/L NADH and 1.5 mmol/L Nitro Blue Tetrazolium (Sigma-Aldrich) in 3.5 mmol/L phosphate buffer (pH 7.4). Succinate dehydrogenase activity was determined by incubation for 1 h with 50 μmol/L sodium succinate and 0.3 mmol/L Nitro Blue Tetrazolium in 114 mmol/L phosphate buffer containing K-EGTA (Sigma-Aldrich). For myosine heavy-chain (Myh) immunostaining, muscle sections were fixed, permeabilized, and processed for incubation with primary antibodies against Myh type I (MyhI), MyhIIa, and MyhIIb (Developmental Studies Hybridoma Bank) and secondary antibodies. Images were captured under Nikon ECLIPS TE-2000U and NIS-Elements F software (Nikon). Myofibers were traced, and their area was measured using ImageJ software. For hematoxylin-eosin staining, cryosections were processed for staining with Mayer hematoxylin-eosin (BBC Biomedical). For periodic acid Schiff staining, liver sections were fixed with Carnoy fixative for 10 min, followed by incubation with 0.5% periodic acid solution (Sigma-Aldrich) for 10 min and counterstaining with Mayer hematoxylin-eosin. For Oil Red O staining, liver sections were fixed and washed with 60% isoprophanol, followed by staining with Oil Red O working solution.

For electron microscopy, tibialis anterior (TA) muscle samples were fixed overnight with 2.5% glutaraldehyde/4% paraformaldehyde solution at 4°C. After incubation for 1 h in 1% OsO₄, the specimens were dehydrated in ethanol series, passed through propylene oxide, and embedded in epoxy resin (Epok). Ultrathin sections (70 nm) were collected on 200 mesh nickel grids and stained for 20 min in 2% uranyl acetate and Reynolds lead citrate. The specimens were observed with a Hitachi HT7700 electron microscope at 100 kV. Electron microscopy was performed by the Research Electron Microscopy Core at Samsung Biomedical Institute.

Microcomputed tomography was performed on a preclinical scanner (Inveon Preclinical CT; Siemens Healthcare) at 200-μm slice thickness, with exposure time of 500 ms, photon energy of 60 keV, and current of 400 μA. The projection images were reconstructed into a three-dimensional image with IRW software. Total abdominal fat volumes including subcutaneous and visceral fat were using Siemens Inveon software.

Quantitative RT-PCR (qRT-PCR) analysis was carried out as previously described (40). Tissues were homogenized by FastPrep^R-24 (MP Biomedicals) and extracted with an easy-spin Total RNA Extract kit (iNtRON). All data are normalized to expression of ribosomal gene L32. The primer sequences are shown in Supplementary Table 1.

Western blot analysis was performed as previously described (41,42). Briefly, cells were lysed in cell extraction buffer (10 mmol/L Tris-HCl, pH 8.0; 150 mmol/L NaCl; 1 mmol/L EDTA; and 1% Triton X-100) containing complete protease inhibitor cocktail (Roche), followed by SDS-PAGE and incubation with primary and secondary antibodies. Tissue extracts were prepared by cryo-pulverization with liquid nitrogen and lysed in cell extraction buffer. Primary antibodies used were Prmt7, ATF2, pATF2, Hsp90 (Santa Cruz Biotechnology), CREB, pCREB, p38MAPK, pp38MAPK, pFOXO1, GAPDH, Prmt5, Prmt1 (Cell Signaling Technology), pFoxo1 (Cell Signaling Technology), PGC-1α, Prmt1, SYM10 (Millipore), β-tubulin (Zymed), and hemagglutinin (HA) (AbFrontier).

Immunoprecipitation was performed as previously described (43). Briefly, precleared cell extracts were incubated with primary antibodies overnight at 4°C, followed by incubation with protein G-agarose beads (Roche) for 1 h and washing three times with cell extraction buffer. Precipitates were analyzed by Western blotting.

For mitochondrial DNA contents, total DNA was extracted from TA muscles, hearts, or brown adipose tissue using a DNeasy Blood and Tissue kit (QIAGEN). The amount of mitochondrial DNA was quantified by the ratio of Mtco2 to β-actin by quantitative PCR.

C2C12 myoblasts (44) and 293T and 10T1/2 cells were cultured as previously described (41). For the generation of stable cell lines, C2C12 cells were transfected with pSuper or pSuper/Prmt7 short hairpin (sh)RNA and selected with 1 μg/mL puromycin followed by pooling the colonies and analyzed. Five different shRNAs against Prmt7, as listed in Supplementary Table 1, were tested and screened for the knockdown effectiveness in P19 embryonal carcinoma cells (Supplementary Fig. 7). Prmt7-shRNA1 and -2 are cloned into pSuper-puro vector based on reproducibility and used interchangeably. Luciferase assays were performed as previously described (45). All experiments were carried out as triplicates and repeated at least three times.

Constructs used in this study are as following: pCMV6-Prmt7 (Origene), CMV–β-galactosidase (33), CRE-Luc (46), and PGC-1α–Luc (Addgene). Chromatin immunoprecipitation (ChIP) assay was carried out as previously described (46). For generation of HA-tagged Prmt7 construct, Prmt7 was subcloned into a pCDNA3.1-HA vector (Clonetech). Antibodies used for ChIP included rabbit IgG (Millipore), Prmt7 (GeneTex), and ATF2 (Cell Signaling Technology). The primers used to amplify the PGC-1α promoter region between −260 to −54 are listed in Supplementary Table 1. All primer sequences are listed in Supplementary Table 1.

Oxygen probe analysis was performed as previously described (47). C2C12/pSuper and C2C12/shPrmt7 were induced to differentiate for 2 days and incubated with a Mito-ID O₂ Probe kit (Enzo) for 4 h followed by washing and analysis by using a fluorescent plate reader in 100 μL DMEM without serum or phenol red. For induction of or inhibition of ATP production, cells were preincubated for 30 min in serum-free DMEM containing 1 μmol/L carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) (Sigma-Aldrich) or 1 μmol/L antimycin A (Sigma-Aldrich) and covered in mineral oil, followed by analysis using a fluorescent plate reader.

Values are expressed as means ± SD or ± SEM, as indicated in the figure legends. Statistical significance was calculated using paired or unpaired two-tailed Student t test. Differences were considered statistically significant at or under values of P < 0.05. For comparison between multiple groups, statistical significance was tested by ANOVA test using SPSS (version 12.0; SPSS, Chicago, IL). To control for the influence of body size variation on energy expenditure (EE), we adjusted group comparisons involving this outcome for total body mass using ANCOVA.

To investigate the in vivo role of Prmt7, we analyzed the level of Prmt7 transcripts in various adult mouse tissues by qRT-PCR. Prmt7 exhibited the highest expression in skeletal muscle, while a modest expression of Prmt7 was detected in heart, brown adipose tissue, kidney, liver, lung, stomach, spleen, white adipose tissue, and small intestine (Fig. 1A). It is of note that Prmt7 transcripts were progressively decreased in aging skeletal muscles analyzed with quadriceps muscles from 5-, 10-, or 28-month-old mice (Fig. 1B). The age-dependent decrease of Prmt7 correlated well with decreased expression of MyhIIa (oxidative fast-twitch fibers), myoglobin (an oxygen transporter), succinate dehydrogenase subunit b (Sdhb) (a mitochondrial enzyme), and PGC-1α (Fig. 1B). In contrast, the expression of MyhI (oxidative slow-twitch fibers) and MyhIIb (glycolytic fast-twitch fibers) was not altered greatly in aging quadriceps. In addition, Prmt7, PGC-1α, and myosin heavy chain (MHC) type IIa (MHCIIa) levels were significantly reduced in gastrocnemius muscles (Gas) from 3-month-old obese mouse models, ob/ob and db/db, compared with the respective control mice (Fig. 1C). These data suggest that Prmt7 expression is decreased in skeletal muscles from aging or obese mouse models, correlated with decreased oxidative muscle metabolism. Hindlimb, Gas, TA, and extensor digitorum longus (EDL) and soleus (Sol) muscles that exhibit distinct metabolic characteristics did not show any specific difference in Prmt7 protein expression (Fig. 1D and E). The role of Prmt7 in skeletal muscle function was investigated by using 4-month-old Prmt7^+/+ and Prmt7^−/− mice that were obtained from the Sanger Institute and backcrossed onto a C57BL/6J background for at least six generations. Prmt7^+/+ and Prmt7^−/− mice did not differ in body weights or exterior appearance (Fig. 1F and K, upper panel). The LacZ transgene in Prmt7^−/− mice can be readily detected in Gas and TA muscles (Fig. 1G). Consistently, Prmt7 proteins were readily detected in Gas muscles of Prmt7^+/+ mice and were absent in Gas muscles of Prmt7^−/− mice (Fig. 1H). The organ weights from 4-month-old Prmt7^−/− mice did not differ from the wild-type littermates (Fig. 1I and Supplementary Table 2). In addition, the weight of Prmt7^−/− muscles did not show any significant difference, except for Sol muscles, with a slight but significant decrease (Fig. 1J and Supplementary Table 2). Interestingly, skeletal muscles of Prmt7^−/− mice appeared pale compared with the wild-type littermates (Fig. 1K and L). Four-month-old Prmt7^−/− muscles did not show any signs of structural abnormalities such as centrally localized nuclei in the histological analysis (Supplementary Fig. 1), suggesting that Prmt7 is dispensable for the gross muscle development. In addition, the histological analysis of white and brown adipose tissue of 4-month-old Prmt7^−/− mice showed no obvious difference in the appearance and the cross-sectional area from that of the wild-type mice (Supplementary Fig. 2).

The qRT-PCR analysis of TA muscles from Prmt7^+/+ and Prmt7^−/− mice revealed that Prmt7 deficiency caused a significantly reduced expression of oxidative fiber markers MyhI and –IIa, accompanied by a substantial increase in glycolytic fiber markers MyhIIx and -IIb, compared with wild type (Fig. 2A). Prmt7^+/+ and Prmt7^−/− muscles were cryosectioned and immunostained for Myh types. Similarly to the qRT-PCR analysis, Prmt7^−/− TA muscles had fewer and smaller MyhIIa-immunopositive fibers, while these contained more and larger MyhIIb-positive fibers, compared with controls (Fig. 2B and C). In addition, Prmt7^−/− Sol muscles showed significantly reduced MyhI-positive fibers in number and size. Similarly, Prmt7 deficiency in EDL muscles resulted in diminished MyhI-positive fibers and increase of MyhIIb fibers in number and size (Supplementary Fig. 3). Skeletal muscles were further assessed by the enzymatic staining for two oxidative enzymes, NADH tetrazolium (NADH-TR) and succinate dehydrogenase (SDH), in TA and EDL muscles. The number of NADH-TR– and SDH-positive fibers and the staining intensities were decreased in Prmt7^−/− TA and EDL muscles (Fig. 2D and E). In addition, Prmt7^−/− Gas muscles had more weakly stained myofibers for both NADH-TR and SDH staining; however, Prmt7^−/− Sol muscles did not exhibit any significant difference from the wild-type muscles (Supplementary Fig. 4A and B). Taken together, Prmt7-deficient muscles exhibit the oxidative to glycolytic fiber-type switch.

The expression of genes involved in the regulation of various muscle functions, Mef2c, Fndc5, myoglobin (muscle), Mcad, Mtco1, Sdhb, Cox7a1, Ucp2 (mitochondria), Fabp3, VLDLR, CideA (lipid metabolism and mitochondria), Vegfα, Vegfβ (angiogenesis) and Glut4, and PEPCK (glucose metabolism), was examined by qRT-PCR analysis (Fig. 3A and B). Myoglobin expression was reduced roughly to 60% of the control level in both Prmt7^−/− TA and EDL muscles, and this likely contributed to the pale appearance of Prmt7^−/− muscles seen in Fig. 1L. Among mitochondrial genes, Mtco1, Sdhb, and Ucp2 and angiogenic factors Vegfα and Vegfβ, which are implicated in oxidative muscle functions, were expressed significantly less in Prmt7^−/− muscles. In addition, the expression of PEPCK was also reduced in Prmt7^−/− muscles, compared with wild type, while a glucose transporter, Glut4, was not greatly changed. Among lipid binding and metabolism regulators, VLDLR was increased in both muscle types, while Fabp3 and CideA expression was decreased. Since many of these genes are targets of PGC-1α, a key regulator of the oxidative metabolism and mitochondrial function (9,14), we assessed the expression of PGC-1α and related genes (PGC-1β, PPARα, and PPARγ) (Fig. 3C and D). The expression of PGC-1α, PPARα, and PPARγ was greatly decreased in both EDL and TA muscles from Prmt7^−/− mice, while PGC-1β was reduced only in TA but not in EDL muscles compared with wild types. In the transmission electron microscopy analysis, Prmt7^−/− TA muscles displayed reduced mitochondrial content compared with wild type (Fig. 3E and F). Consistently, the relative mitochondrial DNA content was decreased in Prmt7^−/− TA muscles compared with the Prmt7^+/+ muscles (Fig. 3G). In contrast, Prmt7^−/− hearts and brown adipose tissues exhibited no difference in mitochondrial DNA content compared with the Prmt7^+/+ tissues (Supplementary Fig. 5A and B).

The functional consequence of this switch was assessed by treadmill running and the repeated grip tests at every 15 min for a 1-h period. On the treadmill, Prmt7^−/− mice ran a significantly shorter distance compared with the wildtype littermates (Fig. 3H). However, these mice had stronger grip strengths at the initial time point, which then declined significantly during the repeated training time points, whereas wild-type mice sustained their muscle strength (Fig. 3I). In addition, the basal lactate level in pre-exercise Prmt7^+/+ and Prmt7^−/− mice showed no significant difference. After a single bout of treadmill exercise, it increased in both mice groups, but Prmt7-deificent mice displayed significantly higher lactate levels compared with the wild-type mice (Fig. 3J). Taken together, these results indicate that Prmt7 deficiency results in reduced oxidative muscle metabolism and impaired endurance muscle capacity.

We then have assessed the whole-body metabolic rates with 5-month-old Prmt7^+/+ and Prmt7^−/− mice. Prmt7^+/+ and Prmt7^−/− mice did not exhibit any significant difference in body weights, food and water intake, or spontaneous locomotive or rearing activity relative to Prmt7^+/+ mice (Fig. 4A–E). Prmt7^−/− mice exhibited a significant decrease in total EE, respiratory quotient, VO₂, and VCO₂ compared with wildtype mice (Fig. 4F–J). These results suggest that Prmt7 deficiency causes decreased EE.

The alteration of metabolic characteristics and the decreased PGC-1α level in skeletal muscle are associated with various metabolic pathologies such as obesity (17,21). The shift from oxidative to glycolytic fibers and the decreased expression of PGC-1α in Prmt7-deficient mice predicted a metabolic phenotype such as obesity. Four-month-old Prmt7^+/+ and Prmt7^−/− mice showed no difference in body weights (Fig. 1F); however, thereafter Prmt7^−/− mice gained progressively more weight relative to wild-type littermates. At ∼10–12 months of age, Prmt7^−/− mice became obviously fatter than littermates, and this obese phenotype was observed in both female and male mice (Fig. 5A and B). The excessive body fat accumulation was observed in 13-month-old Prmt7^−/− female mice compared with the wild-type littermates (Fig. 5A). This was further confirmed by micro–computed tomography scan analyses of 12-month-old Prmt7^+/+ and Prmt7^−/− male mice that showed roughly 1.5-fold more body fat accumulation compared with wild-type littermates (Fig. 5B and C). Livers from three 12-month-old Prmt7^+/+ and Prmt7^−/− littermates were assessed, and two Prmt7^−/− mice exhibited enlarged livers with excessive lipid accumulation analyzed by Oil Red O staining compared with that of a wild-type littermate (Fig. 5D). The histological analysis of white adipose tissues and brown adipose tissues from 8-month-old Prmt7^+/+ and Prmt7^−/− mice revealed that Prmt7 deficiency resulted in enlarged white adipocytes and enhanced lipid droplet accumulation in brown adipocytes (Supplementary Fig. 6). For assessment of the metabolic consequence in Prmt7^−/− mice, the glucose, insulin, and pyruvate tolerance tests were carried out with 12-month-old Prmt7^+/+ and Prmt7^−/− mice. Prmt7^−/− mice displayed impaired glucose and insulin sensitivity (Fig. 5E and F), without alterations in pyruvate tolerance, suggesting that the hepatic gluconeogenesis is normal (Fig. 5G). In addition, Prmt7^−/− mice displayed increased fasting blood glucose levels relative to Prmt7^+/+ littermates (Fig. 5H) without any significant difference in the food intake (Fig. 5I) or in circulating leptin or fasting insulin level (Fig. 5J and K). However, the plasma NEFA, plasma, and hepatic triglyceride content increased significantly in Prmt7^−/− mice relative to the wild-type control (Fig. 5L–N). Taken together, these data suggest that Prmt7 deficiency in mice causes age-dependent obesity and hyperglycemia without altered energy intake, leptin, or insulin levels.

The phenotypic characteristics of PGC-1α (21) and Prmt7-deficient mice are similar, and PGC-1α was decreased considerably in Prmt7-deficient muscles. Thus, we examined whether Prmt7 regulates oxidative muscle metabolism via regulation of PGC-1α expression. C2C12 myoblasts were used to assess the effect of Prmt7 depletion. We tested five Prmt7 shRNAs (shPrmt7) in p19 embryonal carcinoma and all five showed efficient knockdown effects (Supplementary Fig. 7), and shPrmt7-1 or -2 were subcloned into pSuper-puro vector. C2C12 cells were transfected with the control pSuper or shPrmt7 vector, and the transfectants were selected with puromycin. C2C12/pSuper or C2C12/shPrmt7 were induced to differentiate by transferring cells into 2% horse serum containing medium for a total of 2 days, and we analyzed the expression of Prmt7, MHC, and myogenin protein (Fig. 6A) and the mRNA expression of PGC-1α, Mef2c, myoglobin, Mtco1, and Sdhb (Fig. 6B and E). C2C12/shPrmt7 cells exhibited decreased Prmt7 levels relative to the control C2C12/pSuper cells, and the knockdown effect was strongest at day 1 (D1), while it was weaker at D2. C2C12/shPrmt7 cells showed reduced levels of myogenin and MHC expression at D2 compared with the control cells, suggesting a role of Prmt7 for myoblast differentiation. PGC-1α, Mef2c, and myoglobin transcription was markedly increased in differentiating control C2C12/pSuper cells (Fig. 6B–D). Consistent with the results obtained from Prmt7-deficient muscle analysis, Prmt7 knockdown C2C12 cells exhibited significantly reduced expression of PGC-1α at D1, which was further decreased at D2 (Fig. 6B). Furthermore, the expression of Mef2c, myoglobin (Fig. 6C and D), Mtco1, and Sdhb (Fig. 6E) was significantly reduced in these cells at D2 relative to control shRNA-expressing cells. We then examined whether these decreased levels of PGC-1α, Mtco1, and Sdhb have any effect on basal respiration (Fig. 6F) using control and Prmt7 knockdown C2C12 cells at D2. Prmt7 knockdown led to reduced oxygen consumption under basal DMSO-treated conditions. However, the maximal mitochondrial oxidative capacity assessed by FCCP-mediated uncoupling did not differ between two groups of cells, suggesting that the mitochondrial functions are normal.

We tested whether Prmt7 regulates PGC-1α transcription by using a luciferase reporter construct containing the PGC-1α promoter region ranging from 43 to −870 (PGC-1α–Luc). Prmt7 depletion reduced luciferase activities roughly to 60% of that in the control shRNA cells (Fig. 6G), while Prmt7 overexpression markedly enhanced the PGC-1α–Luc activities—approximately sevenfold above control levels (Fig. 6H). This promoter region of PGC-1α-Luc contains a CRE sequence and a binding site for transcription factors CREB and ATF2, which can be activated through phosphorylation by protein kinases Ca2+/calmodulin-dependent protein kinase and p38, respectively. Consistently, Prmt7 overexpression enhanced a reporter activity controlled by CRE (CRE-Luc) in C2C12 cells (Fig. 6I). These data suggest that Prmt7 regulates PGC-1α expression in C2C12 myoblasts during differentiation.

To further determine the molecular mechanism of PGC-1α regulation by Prmt7, we have examined the activation status of CREB, FOXO1, ATF2, and p38 by immunoblotting with antibodies recognizing active phosphorylated forms of these proteins in control or Prmt7 shRNA-expressing C2C12 cells at D1. The level of an active phosphorylated form of CREB (pCREB) and Foxo1 (pFOXO1) or total CREB was unaltered by Prmt7 depletion in C2C12 cells, while the level of ATF2 phosphorylation (pATF2) was decreased without changes in total ATF2 levels (Fig. 7A and B). Furthermore, the level of an active phosphorylated form of p38 (pp38) was diminished without changes in total p38 levels (Fig. 7A and B). These data predicted that Prmt7 may regulate PGC-1α expression through a p38-ATF2 pathway. Previously, it has been shown that Prmt1 regulates p38 activation via interaction and methylation in regulation of megakaryocyte and erythroid differentiation (48). The level of Prmt1 protein was slightly elevated in Prmt7-depleted C2C12 cells (Fig. 7A), suggesting that the decreased p38 activation might be specific to Prmt7 depletion. Furthermore, Prmt7 overexpressing C2C12 cells at D1 exhibited elevated levels of pATF2 and pp38 compared with the control C2C12/pcDNA3.1 cells (Fig. 7C and D).

To determine whether ATF2 and Prmt7 are recruited to CRE sites of the PGC-1α promoter region in C2C12 cells during differentiation, we carried out ChIP analyses. In agreement with previous reports (23,24), the enrichment of ATF2 was greatly enhanced at CRE sites in C2C12 cells at D2 compared with undifferentiated cells (D0). Prmt7 recruitment at CRE sites was also enhanced in differentiating C2C12 cells (Fig. 7E). However, the ATF2 enrichment to CRE sites of the PGC-1α promoter region was reduced in C2C12/shPrmt7 cells at D2 (Fig. 7F). These results suggest that Prmt7 regulates PGC-1α expression through p38 and ATF2.

We then tested the possibility of an interaction of Prmt7 with p38, thereby activating p38 in myoblast differentiation. p38 was immunoprecipitated specifically with a Prmt7 antibody in C2C12 cells (Fig. 8A). To corroborate this notion, we evaluated the potential role of Prmt7 in p38 activation. Prmt7 was coimmunoprecipitated with a p38 antibody, and the level of pp38 concomitantly increased in Prmt7-overexpressing 293T cells compared with the control cells (Fig. 8B). Previous reports have shown that p38 can be dimethylated at arginine residues, whereby its activity is modulated (49). Thus, we tested whether p38 can be symmetrically methylated by Prmt7 by using an antibody recognizing symmetric dimethylation of arginine (Sym10). Prmt7-overexpressing 293T cells showed enhanced Sym10-postive p38 proteins, and this enhancement occurred without alterations in the expression of Prmt1 or Prmt5 (Fig. 8B). For assessment of whether the methyltransferase activity of Prmt7 is required for p38 activation, control or Prmt7-overexpressing 293T cells were treated with the vehicle DMSO or a methyltransferase inhibitor, adenosine dialdehyde (Adox), at a concentration of 20 μmol/L for 12 h and subjected to immunoblotting (Fig. 8C and D). The control 293T cells treated with Adox exhibited a slight decline of pp38 and pATF2, while the robust increase of pp38 and pATF2 levels in Prmt7-overexpressing cells was abrogated by Adox treatment. These data suggest that the methyltransferase activity is required for p38 activation by Prmt7.

Previously, it has been shown that Prmt1 methylates and activates PGC-1α, thereby inducing its target genes, like ERRα and cytochrome c (35). To examine whether, like Prmt1, Prmt7 might activate PGC-1α through interaction and methylation, we performed coimmunoprecipitation analysis between PGC-1α and Prmt7 in 293T cells (Supplementary Fig. 8A–C). Prmt7 and PGC-1α failed to form complexes, while Prmt1 interacted with PGC-1α as shown in a previous study (35). In summary, Prmt7 regulates PGC-1α expression through activation of p38/ATF2, thereby enhancing mitochondrial biogenesis and oxidative muscle metabolism (Fig. 8E). Prmt7 deficiency causes a shift toward glycolytic muscle metabolism, blunting maximum muscle endurance capacity. These alterations in skeletal muscle cause impaired EE, resulting in the excessive energy storage associated with exacerbation of the age-related obesity.

In this study, we demonstrate that Prmt7 is a critical regulator for the oxidative metabolism and the endurance function of skeletal muscle. Mice lacking Prmt7 develop an age-associated obesity. A direct link of Prmt7 to human obesity has not been found; however, a recent genome-wide association study has identified eight single nucleotide polymorphisms in the CTCF-PRMT7 region of the human chromosome 16q22.1 linked to low serum HDL and dyslipidemia (50). In addition, the human chromosomal region 16q22–23 contains genes, such as FTO and MAF, that are associated with obesity (51). A recent study on genes associated with recessive developmental disorders has identified a rare missense mutation in the Prmt7 gene in three families (52). Consistently, Prmt7 mutant mice phenocopy the associated clinical phenotypes, such as pseudohypoparathyroidism and mild intellectual disability, with obesity. This study further support a role of Prmt7 in metabolic control; however, the molecular mechanism underlying such clinical phenotypes is unclear. The whole-body metabolism analysis of Prmt7-deficient mice suggests that the obesity observed in Prmt7-deificent mice likely is linked to the decreased EE. The decrease in oxidative fibers found in TA and EDL muscles of Prmt7-null mice is accompanied by an increased proportion of glycolytic fiber type IIx and IIb at the expense of oxidative fiber type I and IIa. Growing evidence supports that a shift from oxidative to glycolytic muscle metabolism is associated with an imbalance in energy homeostasis and causes exacerbation of diet-induced glucose intolerance, insulin resistance, and obesity in mice (5–8). Furthermore, the proportion of oxidative fibers correlates positively with whole-body insulin sensitivity in humans (19). In obesity and type 2 diabetes, skeletal muscle exhibits a reduced oxidative capacity and increased glycolytic activities. Notably, Prmt7-deficient mice at a young age exhibit the fiber-type switch without any signs of obese phenotypes. Therefore, the fiber-type switch appears to be the cause and not the consequence of obesity in Prmt7-deficient mice.

Consistent with the decreased oxidative metabolism, Prmt7-deficient muscles exhibit decreased expression of various regulatory genes for the oxidative muscle function including PGC-1α and PPARs. The decrease in oxidative metabolism is accompanied by the reduction in mitochondrial contents and the endurance muscle function. Several studies have shown that PGC-1 coactivators play critical roles in the control of muscle oxidative metabolism, mitochondrial biogenesis, and energy homeostasis (9,14,15). The muscle-specific ablation of PGC-1α results in a fiber-type switch toward less oxidative fibers along with impaired endurance exercise and age-associated obesity (16,17). The ability of Prmt7 to regulate the metabolic capacity in skeletal muscle may be achieved, at least in part, through PGC-1α transcription in response to various stimuli like exercise. This is further supported by the fact that Prmt7 depletion in differentiating myoblasts caused decreased expression of PGC-1α and its target genes (Fig. 6). Prmt7 appears to regulate PGC-1α expression through CRE by the activation of ATF2 and p38 without affecting the expression of these genes. Prmt7 can be recruited to the CRE site in the PGC-1α promoter region with ATF2 and pp38. Since Prmt7 interacts with p38, it is possible that Prmt7-mediated methylation of p38 may play a critical role for its activation. A recent study has shown that Prmt1 associates with p38, leading to its methylation and activation required for erythroid differentiation (48). Using an antibody recognizing symmetric dimethylation of arginine (Sym10), we proved that Prmt7 specifically increased the symmetric methylarginine-specific p38 without alteration of Prmt5 levels (Fig. 8B). Another possibility is that Prmt7 interacts with active p38 in nucleus and stabilizes the active form, thereby prolonging its downstream events. Depending on cell types or contexts, PGC-1α expression is regulated differentially by distinctive signaling pathways and transcription regulators. For example, FOXO1 and CREB play critical roles in the regulation of PGC-1α expression in hepatic cells, while MEF2c, ATF2, and CREB have been shown to be important in skeletal muscles (13). In support of the differential mode of PGC-1α gene regulation, Prmt7 deficiency did not alter PGC-1α levels in livers of HFD-fed mice, and PGC-1α expression levels were slightly elevated in white adipose tissue from Prmt7-deficient mice (H.-J.L. and J.-S.K., unpublished data). An alternative but not exclusive mechanism is that Prmt7 may activate transcription factors, like MEF2c or MyoD, in addition to PGC-1α through p38/ATF2, which play critical roles in expression of fiber-type switch–related genes. We showed that MEF2c and MyoD were reduced in Prmt7^−/− mice. This is further supported by the recent studies showing that PGC-1α may be required for exercise-induced mitochondrial biogenesis but not for fiber-type switch (53).

In addition to skeletal muscle, Prmt7 is also expressed in various energy-expanding tissues, including heart and brown adipose tissues. In contrast to skeletal muscle, which developed a fiber-type switch from oxidative to glycolytic muscle at the age of 4 months, however, heart and brown adipose tissues did not show any gross change at the age of 4 months, including organ weights, the size and appearance of adipocytes, and the mitochondrial DNA. Given that decreased EE and exercise capacity were already observed in 4- to 5-month-old Prmt7^−/− mice (Figs. 3 and 4), the muscle function plays a key role in defective energy metabolism of Prmt7-deficient mice. However, we do not rule out the possible contribution of tissues other than skeletal muscle, and we are currently generating tissue-specific knockout mice to further address the specific contribution of skeletal muscle to this phenotype in future.

In conclusion, the results presented here provide the first direct evidence of functional roles for Prmt7 in the skeletal muscle at the gene expression, muscle fiber-type switch, and whole-body metabolism levels. Recent studies have suggested the PGC-1α activation pathway as a therapeutic target to circumvent mitochondrial dysfunction and muscle atrophy in pathological states or normal muscle aging (54,55). Like PGC-1α, Prmt7 appears to be critical for maintenance of muscle mass in aging, and Prmt7-null mice exhibit premature loss of muscle mass. Thus, it is conceivable that modulation of Prmt7 levels in muscle may be an important target to increase mitochondrial biogenesis and to treat muscle atrophy and obesity-related metabolic conditions.

Acknowledgments. The authors thank Drs. Ruth Simon (University of Ulm), Robert S. Krauss (Mount Sinai School of Medicine), and Dario Colleti (Sapienza University of Rome) for critical reading of the manuscript.

Funding. This research was supported by the National Research Foundation of Korea, funded by the Korean Government, Ministry of Science, ICT and Future Planning (NRF-2011-0017315 and NRF-2013M3A9B1069776).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. H.-J.J., H.-J.L., T.A.V., K.-S.C., D.C., and S.C.C. contributed to the experimental design, research, and data analysis. S.-H.K., H.C., and J.-S.K. contributed to the experimental design and data analysis. J.-S.K. wrote the manuscript. J.-S.K. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.