Muscular G9a Regulates Muscle-Liver-Fat Axis by Musclin Under Overnutrition in Female Mice

Muscle is the most abundant tissue in the human body and plays a pivotal role in maintaining whole-body metabolic homeostasis (1). As a locomotive apparatus, exercise-enhanced muscle functions have been shown to reduce the rates of all-cause mortality, including metabolic diseases (2,3). Muscle dysfunctions due to ectopic storage of fat or insensitivity to insulin contribute to the onset and development of multiple metabolic diseases (4,5).

Mounting evidence suggests that muscle is a secretory organ that secretes proteins known as myokines (6,7). Myokines function in autocrine, paracrine, and endocrine ways to regulate muscle development, remodeling, and repair by affecting nearby vascular and neuronal systems or to maintain systemic metabolic homeostasis via interorgan cross talk with adipose tissues, liver, or other organs (6,8). Thus, the myokine-mediated interaction among tissues has become a hotspot in metabolism.

Critical roles of epigenetic changes in metabolic diseases are increasingly recognized (9). EHMT2-encoded histone methylase G9a belongs to the Su(var), enhancer-of-zeste, trithorax (SET) domain–containing Su(var) 3–9 family (10). Usually, G9a can form a heterodimeric complex with G9a-like protein (GLP; encoded by EHMT1) to regulate gene expression by increasing H3K9 monomethylation (H3K9me1) and dimethylation (H3K9me2) in euchromatin (10,11). Both G9a and GLP are essential for early embryogenesis, since whole-body deletion of Ehmt2 or Ehmt1 results in a lethal phenotype (11,12). Tissue-specific functions of G9a or GLP in adipocytes and neurons have been reviewed (13). Links between G9a and metabolic diseases have been reported recently. Under continuous high-fat diet (HFD) stress, the offspring gradually develop severe metabolic syndromes associated with a gradual downregulated hepatic G9a level over generations (14). In hepatic cells, G9a improves insulin signaling via regulating HMGA1 (15). Furthermore, G9a negatively regulates adipogenesis, and thus increased adipose tissue weights have been found in adipocyte-specific G9a knockout (KO) mice, even under normal chow (NC) feeding (16). However, the role of muscular G9a in metabolic diseases remains unclear.

In this study, muscle-specific G9a KO (Ehmt2^Ckmm KO or Ehmt2^HSA KO) mice were created. Surprisingly, Ehmt2^Ckmm KO or Ehmt2^HSA KO protected against obesity under HFD stress in female but not in male mice. Furthermore, increased muscular and circulating levels of musclin, a myokine, was found in muscle-specific G9a KO female mice under HFD stress. A methylase activity–dependent regulation of musclin by G9a was demonstrated. Importantly, exogenous musclin inhibited HFD-induced obesity and hepatic steatosis in wild-type (WT) mice, independent of sex. Collectively, our study demonstrates a sex-dependent epigenetic regulation of musclin by muscular G9a and suggests a critical role of musclin in regulating metabolic changes.

Ehmt2^flox/flox mice (17), Ckmm-Cre mice (JAX, no. 006475), or HSA-Cre mice (JAX, no. 006139) were used to create muscle-specific G9a KO (Ehmt2^Ckmm or Ehmt2^HSA) mice. Genotyping was performed (primers listed in Supplementary Table 1) as previously reported (17). Mice were maintained in a specific-pathogen-free, temperature-controlled (22°C ± 1°C) animal facility with a 12-h light/dark cycle, with free access to water and food. Animals were handled according to the Guidelines of the China Animal Welfare Legislation, as approved by the Committee on Ethics in the Care and Use of Laboratory Animals of the College of Life Sciences, Wuhan University. Animals were usually maintained on NC. Four-week-old mice were fed an HFD (60% kcal fat; Research Diets, New Brunswick, NJ) for an indicated period of time. BIX01294 (BIX) and A366, two structurally different inhibitors of G9a activity (18,19), were obtained from TargetMol (Boston, MA). C57BL/6 female mice were intraperitoneally injected with BIX (50 mg/kg body weight [BW]) or A366 (2 mg/kg BW) once a day for 2 successive days. Tibialis anterior (TA) muscle was used unless otherwise specified.

Mouse musclin cDNA was amplified by PCR and cloned into a pET-15b plasmid. Full-length musclin (Mus-F) was expressed in Escherichia coli and purified with Ni²⁺-NTA agarose (Qiagen, Valencia, CA) (20). Mouse musclin core peptide (Mus33) (21), corresponding to residues 80–112 (#AAS87598.1; GenBank), was synthesized by GenScript (Nanjing, China). Mice were subcutaneously injected with Mus-F or Mus33 (both at 25 μg/mouse) or vehicle twice a day for 7 consecutive days.

Body composition and metabolic studies were analyzed by a minispec LF-50 analyzer (Bruker, Hamburg, Germany) and CLAMS (Columbus Instruments, Columbus, OH), respectively, as previously described (22). Energy expenditure was calculated as follows: energy expenditure = (3.815 + 1.232VO₂/VCO₂) × VO₂.

Serum levels of insulin, leptin, and musclin were measured with ELISA kits for rat/mouse insulin (Millipore, Billerica, MA), mouse leptin (Millipore), or mouse musclin (Cusabio, Wuhan, China), respectively. Concentrations of hepatic/serum triglyceride (TG) and total cholesterol (TC) were measured using kits for TG (Jiancheng Bio., Nanjing, China) or TC (KeHua Bio., Shanghai, China), respectively. Blood glucose was measured with OneTouch blood glucose meter (LifeScan, Milpitas, CA). Insulin tolerance tests were performed as previously described (22).

Hematoxylin-eosin (H&E) staining and Oil Red O staining were performed (14,23). The mean cross-sectional area (CSA) of muscle fibers and the area/distribution of adipocytes were manually traced and analyzed. For immunohistochemical study, anti-Ucp1 (Supplementary Table 2) was used, and positive staining was visualized by 3,3′-diaminobenzidine substrate (Vector, Burlingame, CA). For immunofluorescent studies, muscle cryosections were labeled with antimyosin heavy chain I, IIA, or IIB (Supplementary Table 2). Positively stained myofibers were determined by manual counting and quantified using Image-Pro Plus 6.0 software.

C2C12 myoblasts, originally derived from the thigh muscle of female C3H mouse (24), were cultured. pSuper-shRNA targeting G9a (Supplementary Table 1) or scramble was packed into a lentivirus expression system, and infected, stable knockdown cell lines were selected. PAdeno-mG9a-3*Flag (Ade-mG9a), PAdeno-mG9aΔset-3*Flag (Ade-mG9aΔset) that encodes a SET domain–deleted G9a and the control plasmid were constructed, packed into adenovirus, and purified (Obio, Shanghai, China). C2C12 myoblasts were infected with these adenoviruses at a multiplicity of infection of 50, or treated with 0-4 μmol/L BIX or 0-1 μmol/L A366 or vehicle (DMSO), for 24 h. Mouse musclin (phage-mMusclin) was packed into lentivirus and infected 293T cells to establish a stable musclin-overexpressing cell line, which was treated with 2 μmol/L BIX for 24 h, and followed by transfection with pRK-Flag-mFoxo1 or pRK-Flag-mFoxo1 (S253A) plasmid for another 24 h.

Stromal vascular fraction cells were isolated from brown adipose tissue (BAT) of C57BL/6 mice as described (22). Brown adipogenesis was induced by treating confluent cells (day 0) with 0.5 mmol/L isobutylmethylxanthine, 1 μmol/L dexamethasone, 850 nmol/L insulin, 1 nmol/L T3, and 1 μmol/L rosiglitazone. At 2 days after induction, cells were switched to differentiation medium with 850 nmol/L insulin, 1 nmol/L T3, and 1 μmol/L rosiglitazone until day 7. Then, cells were treated with Mus-F (1 μmol/L) or Mus33 (1 and 5 μmol/L) for another 48 h.

Quantitative real-time PCR (qPCR), Western blots, and chromatin immunoprecipitation (ChIP) assays were performed as previously described (14,17,23). Primers, antibodies, and dilutions used are provided in Supplementary Tables 1 and 2. Regions of musclin promoter used for ChIP assay ranged from −1,500 base pair to the transcription start site.

Data are expressed as the average ± SD. Statistical significance was determined by analyzing the data with the nonparametric Kruskal-Wallis test followed by the Mann-Whitney test. Differences were considered statistically significant at P < 0.05.

All data generated or analyzed during this study are included in this published article and Supplementary Material.

Ehmt2^Ckmm mice were born at a normal Mendelian ratio and showed no obvious abnormal phenotype. KO efficiency was verified by significantly reduced Ehmt2 (but not Ehmt1) level in different muscle parts, but not in the liver, in both sexes (Fig. 1A and B and Supplementary Fig. 1A and B). In TA, G9a and H3K9me2 levels, but not Glp level, were significantly reduced in both sexes (Fig. 1C and Supplementary Fig. 1C).

To investigate the role of muscular G9a in metabolic diseases, WT and Ehmt2^Ckmm mice were fed NC or HFD starting at 4 weeks old (Supplementary Fig. 1D). No differences in BW between NC-fed WT and Ehmt2^Ckmm mice were found for both sexes; however, the Ehmt2^Ckmm females, but not males, showed a significantly lower BW after 14 weeks on HFD, as well as significantly lower fat mass and higher lean mass (Fig. 1D and E and Supplementary Fig. 1E and F). Anatomical studies further revealed significantly reduced epididymal white adipose tissue (eWAT) and inguinal white adipose tissue (iWAT) weights in the Ehmt2^Ckmm females under HFD stress but not in the other tissues examined (Fig. 1F–H and Supplementary Fig. 1G and H). Furthermore, the amount of intermuscular fat (IMF) was significantly reduced in Ehmt2^Ckmm females under HFD stress (Fig. 1I). A similar food intake was observed in both sexes of NC-fed conditions; however, an unexpected higher food intake without significantly affecting serum leptin level was observed in Ehmt2^Ckmm females, but not males, in HFD-fed mice (Fig. 1J, Supplementary Fig. 1I, and Supplementary Tables 3 and 4). Meanwhile, the 6-h fasting blood glucose (FBG) levels were significantly lower in Ehmt2^Ckmm females compared with WT controls under NC-fed conditions (Supplementary Fig. 2A). Insulin tolerance tests demonstrated that NC-fed Ehmt2^Ckmm mice were more sensitive to insulin stimulation compared with their controls (Supplementary Fig. 2B).

The size of individual myofiber and the CSA of myofibers were comparable between WT and Ehmt2^Ckmm mice under NC- or HFD-fed conditions in both sexes (Fig. 2A and Supplementary Fig. 3A and B). Furthermore, mRNA levels of the skeletal muscle atrophy markers (Atrogin-1 and MuRF-1) and the myogenic markers (Myostatin and Myf5) were similar between NC-fed WT and Ehmt2^Ckmm mice in both sexes, suggesting unaffected muscle development by muscular G9a deficiency (Fig. 2B and Supplementary Fig. 3C). Among the lipid metabolism genes that were examined, only the Fasn level was significant decreased in the muscle of HFD-fed Ehmt2^Ckmm females (Fig. 2C). Furthermore, Oil Red O staining demonstrated reduced lipid accumulation in the muscle of HFD-fed Ehmt2^Ckmm females (Supplementary Fig. 4).

Skeletal muscle is composed of distinct myofiber subtypes defined by antimyosin heavy chain isoforms and metabolic activity. Among them, type I and IIa myofibers are oxidative, type IIb and IIx myofibers are glycolytic (3). Significantly upregulated Myh7 and Myh2 levels (type I and IIa myofiber markers, respectively) were found in HFD-fed Ehmt2^Ckmm females, whereas Myh4 level (type IIb myofiber marker) was comparable (Fig. 2D). Furthermore, immunostaining suggested a dramatic increase in type I, but not type IIa and IIb, myofibers in HFD-fed Ehmt2^Ckmm females (Fig. 2E).

Consistent with the markedly decreased fat mass, significantly smaller adipocyte sizes were found in eWAT, iWAT, and IMF of HFD-fed Ehmt2^Ckmm females (Fig. 3A and C and Supplementary Fig. 5A). Moreover, HFD-fed Ehmt2^Ckmm females showed markedly elevated levels of key thermogenic genes in iWAT (Ucp1 and Prdm16) (Fig. 3D) and eWAT (Ucp1 and Cidea) (Supplementary Fig. 5B). However, no obvious changes were found for genes involved in lipolysis (Pnpla2, Lipe, Mgll, and Ppargc1a) or β-oxidation (Acadl, Cox4, and Cox7a) in eWAT or iWAT (Fig. 3D and Supplementary Fig. 5B). Although there was no obvious change in adipocyte size, significantly increased Ucp1 immunostaining (Fig. 3E) and upregulated thermogenic genes (Ucp1 and Cidea) (Fig. 3F) were detected in BAT of HFD-fed Ehmt2^Ckmm females. Furthermore, Ehmt2^Ckmm females, but not males, showed significantly increased VO₂ and energy expenditure under HFD stress (Fig. 3G and H and Supplementary Fig. 6A and B).

Although no obvious difference in liver weight between WT and Ehmt2^Ckmm mice was observed under either diet in both sexes, HFD-fed Ehmt2^Ckmm females, but not males, displayed a significant reduction in hepatic fat deposition (Fig. 3I and Supplementary Fig. 6C). Meanwhile, significantly reduced hepatic TG levels, but not TC levels, were found in HFD-fed Ehmt2^Ckmm females (Fig. 3J). Regardless of sex, similar serum TG and TC levels were observed in NC- or HFD-fed WT and Ehmt2^Ckmm mice, whereas similar serum insulin levels were found in HFD-fed WT and Ehmt2^Ckmm mice (Supplementary Tables 3 and 4). We speculated that altered hepatic TG levels may be due to changes in hepatic lipid metabolism. As expected, several key genes involved in fatty acid transport and lipogenesis, such as Cd36, Srebp1c, Acly, and Fasn, were downregulated in the liver of HFD-fed Ehmt2^Ckmm females (Fig. 3K).

Emerging evidence has shown that muscle-released myokines mediate interorgan metabolic communications (6,25). To explore this possibility, transcriptional levels of several previously reported myokines were evaluated (6), and musclin was found to be significantly increased in the TA of Ehmt2^Ckmm females under either diet (Fig. 4A and Supplementary Fig. 7A). Interestingly, musclin levels were similar in TA and gastrocnemius of WT and Ehmt2^Ckmm males (Supplementary Fig. 7B). Consistently, the muscular musclin levels were significantly increased in Ehmt2^Ckmm females (Fig. 4B and Supplementary Fig. 7C). Moreover, increased serum musclin levels were found in HFD-fed Ehmt2^Ckmm females (Fig. 4C). Thus, musclin may act as a circulating factor that affects metabolism.

To investigate how G9a regulates musclin, we performed a loss-of-function study using C2C12 myoblasts. Knockdown of G9a was confirmed, and shG9a cells exhibited markedly upregulated mRNA and protein levels of musclin (Fig. 4D and E). Furthermore, structurally different G9a inhibitors, BIX and A366, both reduced H3K9me2 levels and increased musclin levels (Fig. 4F and G and Supplementary Fig. 8A and B). Furthermore, C2C12 myoblasts overexpressing full-length G9a, but not a SET domain–deficient G9a (G9aΔSET), exhibited reduced musclin levels (Fig. 4H and I). To confirm that G9a methylase activity also regulates musclin expression in vivo, BIX or A366 was injected into WT females. Significantly reduced H3K9me2 levels in TA and elevated musclin levels in TA and serum were observed in mice treated with either inhibitor (Fig. 4J and K and Supplementary Fig. 8C). Since G9a represses gene expression by adding the repressive marker H3K9me2 on promoters (12), we hypothesized that G9a may directly regulate musclin levels. ChIP assays demonstrated that G9a and H3K9me2 were significantly decreased in the musclin promoter of Ehmt2^Ckmm females, while G9a and H3K9me2 were significantly increased in the musclin promoter in C2C12 myoblasts overexpressing full-length G9a (Fig. 5A and B).

FOXO1 is known to inhibit musclin transcription (26), and phosphorylation of FOXO1 at S253 causes nuclear exclusion that suppresses its activity (27). We speculated that altered level of FOXO1 or phosphorylated-FOXO1 (p-FOXO1) may contribute to increased musclin transcription in Ehmt2^Ckmm females. Under either diet, significantly reduced FOXO1 as well as increased p-FOXO1 and p-FOXO1/FOXO1 were found in the TA of Ehmt2^Ckmm females (Fig. 5C and D). Consistently, these changes were also found in G9a knockdown or in BIX/A366-treated C2C12 myoblasts (Fig. 5E and F and Supplementary Fig. 8D). Furthermore, a significant elevation in p-FOXO1 was found in the TA of WT female mice injected with either inhibitor (Fig. 5G and Supplementary Fig. 8E).

To investigate the roles of FOXO1 or p-FOXO1 in regulating musclin expression, we performed gain-of-function studies in 293T cells. No significant changes in G9a levels were found in musclin-overexpressed 293T cells treated with BIX or transfected with FOXO1 (WT or S253A mutant) (Fig. 5H). Similar to C2C12 cells, BIX also upregulated musclin, while it reduced FOXO1 and increased p-FOXO1 and p-FOXO1/FOXO1 levels in 293T cells overexpressing musclin (Fig. 5H). Overexpression of S253A, but not WT FOXO1, prevented BIX-induced upregulation of musclin and p-FOXO1/FOXO1 (Fig. 5H), indicating that inhibition of G9a activity increases musclin level by upregulating p-FOXO1/FOXO1.

Muscle-specific deletion of G9a using Ckmm-Cre also resulted in decreased mRNA and protein levels of G9a, but not of musclin, in the heart (cardiomyocytes) (Supplementary Fig. 9A–D). Importantly, compared with skeletal muscle, musclin protein levels were barely detectable in the heart (Supplementary Fig. 9E), which is in agreement with a report stating that musclin is a skeletal muscle–enriched myokine (28).

To rule out possible cardiomyocytic G9a-mediated metabolic effects on above-observed phenotypes, another skeletal muscle–specific G9a deficiency mouse strain (Ehmt2^HSA) was created. As expected, G9a levels were dramatically decreased in multiple muscle parts but not in liver and heart; however, G9a and H3K9me2 levels were also significantly reduced in both sexes (Fig. 6A–C and Supplementary Fig. 10A–C). No difference in body and tissue weights between NC-fed WT and Ehmt2^HSA mice were found for both sexes (data not shown). However, HFD-fed Ehmt2^HSA females, but not males, showed significantly lower BW and fat mass (Fig. 6D and E and Supplementary Fig. 10D and E). Anatomical studies revealed significantly reduced eWAT and iWAT weights in HFD-fed Ehmt2^HSA females, but not in other tissues examined; also, similar amounts of food intake were found in either sex or diet of the WT and Ehmt2^HSA mice (Fig. 6F and G and Supplementary Fig. 10F–H). As expected, significantly reduced FOXO1, as well as significantly increased p-FOXO1 and p-FOXO1/FOXO1 levels, were found in the TA, while significantly increased musclin levels were found in the TA and serum of HFD-fed Ehmt2^HSA females (Fig. 6H–J).

To address whether the obese-resistant phenotype of muscular G9a KO females is related to musclin-dependent muscle-adipose/liver cross talk, we administered Mus-F or Mus33 to 8-week-old HFD-fed C57BL/6 females for 7 days (Supplementary Fig. 11A). Both forms of musclin effectively prevented weight gain in HFD-fed mice without affecting food intake (Fig. 7A and B), while significantly increasing circulating musclin (Fig. 7C). Mus-F, but not Mus33, significantly decreased the weights of eWAT and iWAT (Fig. 7D) without affecting weights of other tissues examined (data not shown). Furthermore, Mus-F but not Mus33 significantly reduced HFD-induced serum TG elevation (Fig. 7E). Compared with vehicle-treated controls, both forms of musclin significantly reduced lipid deposition in the liver of HFD-fed mice (Fig. 7F and G). Additionally, both musclin types significantly reduced adipocyte size and increased immunostaining of Ucp1 in eWAT, iWAT, and BAT of HFD-fed mice (Supplementary Figs. 7H and 11B and C). Moreover, HFD-fed mice treated with either musclin showed markedly elevated Ucp1 in iWAT, and Mus-F also significantly upregulated Prdm16 in iWAT (Fig. 7I). Consistently, both musclin types upregulated Ucp1 and Prdm16 levels in primary brown adipocytes (Supplementary Fig. 12).

Interestingly, both musclin types also effectively prevented HFD-induced weight gain without affecting food intake in C57BL/6 males (Fig. 8A and B and Supplementary Fig. 11A). Compared with vehicle-treated HFD-fed mice, only Mus-F significantly decreased the weights of eWAT and iWAT of HFD-fed males, while a significant decrease in hepatic TG level was achieved by both musclin types (Fig. 8C and D). Moreover, both musclin types significantly reduced the adipocyte sizes and upregulated Ucp1 staining of eWAT, iWAT, and BAT of HFD-fed males (Fig. 8E).

Skeletal muscle acts as a key metabolic and endocrine mediator of systemic metabolism (1,6). Understanding how muscle communicates with other tissues, such as adipose tissues and liver, to coordinate systemic metabolism remains an integral question. Here, we found that increased circulating musclin, an exercise-induced myokine (28), confers resistance to HFD-induced obesity and hepatic steatosis upon muscular G9a deficiency in female mice. Musclin is initially identified as osteocrin, a bone-derived peptide expressed only during the embryonic stage; later, musclin was found to be exclusively expressed in skeletal muscle of adults (28). Musclin has been demonstrated to enhance physical endurance by promoting mitochondrial biogenesis; therefore, whole-body musclin-deficient mice exhibit low exercise tolerance (21), and they lose more muscle mass during cancer cachexia (29). Here, we report novel effects of short-term administration of musclin on systemic metabolism.

Musclin contains regions and a putative serine protease cleavage site highly homologous to members of natriuretic peptides (NPs). It is thus proposed that musclin may signal through NPs and their receptors (NPRs) (30). Three kinds of NPs (atrial NP [ANP], B-type NP [BNP], and C-type NP), have been identified so far, and they are reported to function as endocrine/paracrine factors in regulating blood pressure, fat metabolism, and skeletal muscle development by binding to NPR1, 2, or 3, among which NPR3 is a scavenger receptor that lacks cytosolic enzymatic domain (31). Musclin specifically binds to NPR3 (30), thus increasing the odds of NPs binding to their functional receptors. Here, the effects of musclin in preventing HFD-induced obesity and hepatic steatosis (Figs. 7 and 8 and Supplementary Fig. 11) may be due to competition with NPs, which affects NP-NPR signaling and regulates metabolism. However, the different effects of Mus-F and Mus33 on serum lipid level and WAT weights (Figs. 7D, 7E, and 8C) suggest that specific receptor(s) may exist for musclin.

As an exercise-induced myokine, the circulating level of musclin increases 1.6-fold in mice after 5 days of exercise (21). Increased musclin levels, possibly due to musclin resistance, have been found in the serum of patients with type 2 diabetes (32) and HFD-fed mice (Fig. 7C). In our study, the level of circulating musclin was found to be increased 1.3- to 1.6-fold in HFD-fed muscle-specific G9a-deficient females or musclin-treated mice (Figs. 4C, 6J, and 7C). Although musclin was discovered years ago, there is no direct evidence demonstrating its role in metabolism. Here, we showed effects of musclin on Ucp1 levels of the different adipose tissues and primary brown adipocytes, as well as in preventing HFD-induced hepatic steatosis (Figs. 7 and 8 and Supplementary Figs. 11 and 12). Since ANP and BNP can stimulate lipolysis in adipocytes through a cGMP-dependent protein kinase signaling pathway (33), musclin may blockade NPR3 to enhance the effects of ANP and BNP on adipocytes. However, further research is needed to elucidate how musclin acts on adipose tissue or liver.

The present study identified G9a as a novel negative regulator of musclin. In vivo and in vitro studies showed that G9a transcriptionally inhibited musclin expression by maintaining H3K9me2 level in its promoter (Fig. 5). Loss of G9a also elevated musclin expression by upregulating p-FOXO1/FOXO1, a process that depends on the methylase activity of G9a (Fig. 5). The dual roles of G9a in biological processes were previously described. For example, G9a represses adipogenesis by inhibiting peroxisome proliferator–activated receptor γ expression in a methylase activity–dependent manner and by facilitating Wnt10a expression of methylase independently (16). Thus, in addition to exercise, we revealed a G9a-dependent epigenetic pathway that regulates musclin.

Substantial evidence highlights the importance of epigenetic regulation in skeletal muscle. Skeletal muscle–specific ablation of MLL4 (a H3K4 methylase) in mouse results in decreased type I myofibers and diminished mitochondrial respiration, which impairs exercise endurance by reducing fat utilization in muscle (34). In addition, skeletal muscle–specific depletion of histone deacetylase 3 causes severe systemic insulin resistance in mice (35). However, reports on G9a in muscle differentiation are controversial. Methylase activity–dependent inhibition of muscle development of G9a has been found in C2C12 cells (36), while G9a is dispensable for skeletal muscle development and regeneration using MyoD-Cre to KO G9a in muscle (37). Here, muscular G9a deficiency driven by Ckmm-Cre did not alter muscle mass and morphology (Figs. 1H, 2A, and 2B), which is consistent with a previous in vivo study (37). Moreover, we revealed that muscular G9a function in metabolic diseases, especially for females, may be partially due to increased type I myofibers and musclin levels.

It is well known that sex differences exist in the regulation of glucose and lipid metabolism (38,39), but the exact mechanism remains poorly defined. Herein, a sexually dimorphic role for muscular G9a in the regulation of metabolic homeostasis is suggested from an epigenetic angle. Sex hormones, such as estrogens and androgens, are major mediators of sex differences in metabolism (39). Estrogen in skeletal muscle is essential to promote glucose and lipid homeostasis (40). However, the direct binding sequences by sex hormones have not been found on the promoters and enhancers of musclin, and future investigations are needed to define how muscular G9a mediates sex differences in metabolism.

In summary, our results revealed a novel muscle-dependent G9a/musclin signaling for maintaining lipid homeostasis, particularly in females. Targeting this pathway may provide a potential therapeutic approach against obesity and related complications.

Acknowledgments. The authors thank the Analytical and Testing Core of College of Life Sciences of Wuhan University as well as the Analytical and Testing Center of Huazhong University of Science and Technology for technical assistance.

Funding. This work is supported by the National Natural Science Foundation of China (91957114 and 32021003) and the National Key R&D Program of China (2018YFA0800700 and 2019YFA0802701).

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

Author Contributions. W.Z., K.H., and L.Z. designed the research. W.Z., D.Y., Y.Y., C.L., H.C., Y.Z., and Q.W. performed the experiments. W.Z., R.B.P., K.H., and L.Z. analyzed the data and wrote the manuscript. L.Z. 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.