Musclin, a novel skeletal muscle-derived secretory factor.

Skeletal muscle is involved in the homeostasis of glucose and lipid metabolism. We hypothesized that the skeletal muscle produces and secretes bioactive factor(s), similar to adipocytokines secreted by fat tissue. Here, we report the identification of a novel secretory factor, musclin, by signal sequence trap of mouse skeletal muscle cDNAs. Musclin cDNA encoded 130 amino acids, including NH(2)-terminal 30-amino acid signal sequence. Musclin protein contained a region homologous to natriuretic peptide family, and KKKR, a putative serine protease cleavage site, similar to the natriuretic peptide family. Full-length musclin protein and KKKR-dependent cleaved form were secreted in media of musclin cDNA-transfected mammalian cell cultures. Musclin mRNA was expressed almost exclusively in the skeletal muscle of mice. Musclin mRNA levels in skeletal muscle were markedly low in fasted, increased upon re-feeding, and were low in streptozotocin-treated insulin-deficient mice. Musclin mRNA expression was induced at late stage in the differentiation of C2C12 myocytes. In myocytes, insulin increased, while epinephrine, isoproterenol, and forskolin reduced musclin mRNA, all of which are known to increase the cellular content of cyclic AMP, a counter-regulator to insulin. Pathologically, overexpression of musclin mRNA was noted in the muscles of obese insulin-resistant KKAy mice. Functionally, recombinant musclin significantly attenuated insulin-stimulated glucose uptake and glycogen synthesis in myocytes. In conclusion, we identified musclin, a novel skeletal muscle-derived secretory factor. Musclin expression level is tightly regulated by nutritional changes and its physiological role could be linked to glucose metabolism.

Skeletal muscle is involved in the homeostasis of glucose and lipid metabolism. We hypothesized that the skeletal muscle produces and secretes bioactive factor(s), similar to adipocytokines secreted by fat tissue. Here, we report the identification of a novel secretory factor, musclin, by signal sequence trap of mouse skeletal muscle cDNAs. Musclin cDNA encoded 130 amino acids, including NH 2 -terminal 30-amino acid signal sequence. Musclin protein contained a region homologous to natriuretic peptide family, and KKKR, a putative serine protease cleavage site, similar to the natriuretic peptide family. Full-length musclin protein and KKKR-dependent cleaved form were secreted in media of musclin cDNA-transfected mammalian cell cultures. Musclin mRNA was expressed almost exclusively in the skeletal muscle of mice. Musclin mRNA levels in skeletal muscle were markedly low in fasted, increased upon re-feeding, and were low in streptozotocin-treated insulin-deficient mice. Musclin mRNA expression was induced at late stage in the differentiation of C2C12 myocytes. In myocytes, insulin increased, while epinephrine, isoproterenol, and forskolin reduced musclin mRNA, all of which are known to increase the cellular content of cyclic AMP, a counter-regulator to insulin. Pathologically, overexpression of musclin mRNA was noted in the muscles of obese insulin-resistant KKAy mice. Functionally, recombinant musclin significantly attenuated insulin-stimulated glucose uptake and glycogen synthesis in myocytes. In conclusion, we identified musclin, a novel skeletal muscle-derived secretory factor. Musclin expression level is tightly regulated by nutritional changes and its physiological role could be linked to glucose metabolism.
Muscle-specific glucose transporter GLUT4 1 (3) and peroxisome proliferator-activated receptor-␥ (4, 5) knock-out mice exhibited the alterations in insulin sensitivity in fat and liver. Muscle-specific insulin receptor knockout mice showed adipocyte hyperproliferation (6). These findings suggest that the skeletal muscle may release bioactive factors (myokines), like adipocytokines, to target fat, liver, and potentially skeletal muscle itself.
In the present study, we attempted to identify skeletal muscle-derived secretory factors using an efficient signal sequence trap (SST) method (7). Here, we report a novel skeletal musclederived secretory factor, musclin, whose mRNA was dynamically regulated by nutrition and hormonal factors.

EXPERIMENTAL PROCEDURES
Cloning of Mouse Musclin cDNA-Poly(A) ϩ RNAs were extracted from gastrocnemius muscles of 10-week-old male C57BL/6J mice under ad libitum, 24-h fasting or 24-h refeeding after 24-h fasting condition and ad libitum db/db mice. Equal amounts of poly(A) ϩ RNA from each group were pooled to synthesize cDNA. To selectively clone the genes that possess signal sequence at the NH 2 -terminal end of cDNAs, SST-REX system (signal sequence trap by retrovirus-mediated expression screening system) was introduced as we reported previously (7). Interleukin-3-independent Ba/F3 cells were harvested, and the integrated cDNAs were isolated from the cells by genomic PCR and sequenced.
To identify full-length cDNA sequence of musclin, the original fragment obtained by SST-REX system was subjected to 5Ј-and 3Ј-RACE, using SMART TM RACE cDNA amplification kit (BD Biosciences). Rat and human full-length cDNA were cloned, using homology search for rat and human genome (GenBank TM ) with mouse musclin sequence, RACE, and reverse transcription-PCR.
Animals and Experimental Protocol-Male C57BL/6J mice (10 weeks old) were divided into three groups (n ϭ 5, each): mice fed ad libitum with standard chow, mice fasted for 48 h, and mice refed 24 h after 48-h fasting.
Male C57BL/6J mice (10 weeks old) received one-shot intraperitoneal injection of streptozotocin (STZ, Sigma) (100 mg/kg body weight) every day for 3 days (total three times). The mice were sacrificed 4 days after the final injection.
Musclin mRNA expression in various tissues of mice was analyzed in 10 -12-week-old male C57BL/6J mice. Musclin mRNA expression was also analyzed in muscles of 13-week-old female C57BL/6J and KKAy mice.
All experimental protocols were approved by the Ethics Review Committee for Animal Experimentation of Osaka University.
Northern Blot and Quantitative Reverse Transcription-PCR Analysis-RNA extraction and Northern blotting were conducted as described previously (8). First strand cDNA was synthesized using Ther-* This work was supported in part by the Suzuken Memorial Foundation; The Nakajima Foundation; the Kanae Foundation for Life and Socio-Medical Science; the Takeda Medical Research Foundation; the Takeda Science Foundation; a grant-in-aid from the Japan Medical Association; The Naito Foundation; by a grant from the Ministry of Health, Labor and Welfare, Japan; and by grants from the Ministry of Education, Culture, Sports, Science and Technology, Japan. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY573932, AY573933, and AY573934.

Effects of Various Hormones on Musclin mRNA Expression in C2C12
Myocytes-Mouse myoblast cell line C2C12 cells were grown and induced for differentiation, as described previously (9). Cells were harvested at the indicated time. The effects of insulin, IGF-1, epinephrine, isoproterenol, forskolin, and vehicle on musclin mRNA expression were examined in differentiated C2C12 myocytes (day 7). After 24-h incubation with serum-free Dulbecco's modified Eagle's medium containing the test hormone, the cells were harvested for measurement of mRNA.
Stable Cell Lines Expressing FLAG-tagged Musclin-A retrovirus carrying full-length mouse musclin with FLAG epitope (DYKDDDDK) at the C terminus or green fluorescent protein was generated as described previously (7). The virus was added to C2C12 cells, and after puromycin selection, virus-infected stable C2C12 cells were differentiated to myocytes. Seven days after differentiation induction, the medium was collected, and cells were lysed with extraction buffer as described previously (8) and subjected to immunoprecipitation with anti-FLAG antibody (anti-FLAG M2-agarose gel, A2220, Sigma), followed by Western blotting using anti-FLAG M2 monoclonal antibody (A8592, Sigma).
Secretion of Wild-type and Mutant Musclin-The cDNA encoding mouse musclin protein fused to COOH-terminal FLAG epitope or mutant musclin protein fused to COOH-terminal FLAG (the 76 KKKR 79 was changed to 76 A, using a QuikChange site-directed mutagenesis kit, Stratagene) was subcloned into pEF-BOS expression plasmid (pEF-BOS-WT musclin FLAG and pEF-BOS-Mut musclin FLAG, respectively) (10). HEK293 cells were transfected with pEF-BOS-WT musclin FLAG or pEF-BOS-Mut musclin FLAG. After 72 h, the culture medium was subjected to immunoprecipitation with anti-FLAG antibody, followed by Western blotting using anti-FLAG antibody.
Preparation of Recombinant Musclin Protein-Recombinant musclin protein was purified from culture medium of stable C2C12 myocytes expressing FLAG-tagged musclin using agarose beads column conjugating anti-FLAG antibody (anti-FLAG M2-agarose gel, A2220, Sigma).
Statistical Analysis-Statistical analyses were performed with unpaired t tests, except for data shown in Fig. 3, B and D-F. Statistical analyses of the data shown in Fig. 3, B and D-F, were performed with analysis of variance (Fisher's projected least significant difference). A p value less than 0.05 denoted the presence of a statistically significant difference.

RESULTS
We previously developed an efficient SST method using retrovirus-mediated gene transfer (7). To identify skeletal musclederived nutritionally regulated secretory factor(s), we conducted this SST method using the pooled poly(A) ϩ RNA from gastrocnemius muscles of mice under ad libitum, fasted and refed conditions, and obese diabetic db/db mice. We screened and sequenced 1812 clones. 18 clones were selected as unknown putative secreted proteins with a signal sequence and no putative transmembrane region. The mRNA of one clone was almost exclusively expressed in the skeletal muscle. The original fragment obtained from SST-REX screening was subjected to 5Ј-and 3Ј-RACE to obtain the non-coding region of musclin. The encoded protein was previously unidentified and was named musclin (Fig. 1A). Musclin consisted of 130 amino acids, including an amino-terminal 30-amino acid signal sequence, with a predicted molecular mass of 11 kDa. Rat and human coding regions of cDNAs were cloned using homology search for rat and human genome (GenBank TM ) with mouse musclin sequence, RACE, and RT-PCR (Fig. 1B). Amino acid sequence identity between mouse and human was 75.9% and that between mouse and rat was 90.2%. COOH terminus of the protein contained a region highly homologous to mouse natriuretic peptide including ANP, BNP, and CNP (12). A KKKR putative serine protease cleavage site also existed prior to this homologous region, similar to the NPs. These motifs suggested that musclin protein might also excise cleaved form at KKKR site, similar to natriuretic peptide.
To investigate whether musclin is a myosecretory protein, we constructed a mouse myoblast C2C12 cell line stably expressing a FLAG epitope-tagged musclin with a retrovirus. The cell lysate and medium of these cells were immunoprecipitated with anti-FLAG antibody and subjected to Western blotting with anti-FLAG antibody. Differentiated C2C12 myocytes synthesized the musclin protein as seen in cell lysate, and the synthesized protein was robustly secreted into the medium (Fig. 1C), indicating that musclin is a secretory protein. Three bands were detected in blots of musclin. Two bands positioned between 15 and 20 kDa. These sizes were different from the deduced size of 11 kDa, suggesting that musclin protein undergoes post-translational modification. The third band, whose abundance was extremely low in this cell system expressing FLAG-tagged musclin protein, positioned at about 12-13 kDa, suggesting that musclin protein was cleaved at the 76 KKKR 79 site as described above. To test this, expression plasmid carrying wild-type or mutant musclin ( 76 KKKR 79 3 76 A) fused to FLAG at the COOH terminus was transiently transfected in HEK293 cells, and the culture medium was subjected to immunoprecipitation and Western blotting with FLAG antibody (Fig. 1D). The mutated musclin did not produce the cleaved form, suggesting that some amount of wild-type musclin protein was cleaved at the 76 KKKR 79 site.
When various tissues of C57BL/6J mice were subjected to Northern blotting, musclin mRNA was almost exclusively expressed in skeletal muscles ( Fig. 2A), and the mRNA length was about 1.4 kb. By real-time RT-PCR analysis, musclin mRNA was also expressed in brown adipose tissue, spleen, testis, and bone to a much lesser extent than skeletal muscle (Fig. 2B). Rat musclin mRNA was also expressed almost exclusively in skeletal muscle (data not shown).
We examined regulation of musclin mRNA expression in vivo (Fig. 2, C-E). Musclin mRNA expression in gastrocnemius muscle was almost eliminated by 48-h fasting and reversed by 24-h refeeding (Fig. 2C). This nutritional regulation of musclin mRNA was much more dramatic than those of GLUT4 and lipoprotein lipase mRNAs (Supplemental Fig. 1). To determine the effect of insulin on musclin mRNA expression, mice were treated with STZ. Musclin mRNA expression was significantly decreased in the gastrocnemius muscles of STZ-treated insulindeficient mice (Fig. 2D). With regard to pathophysiological significance, musclin mRNA expression was augmented in gastrocnemius muscle of obese KKAy mice, at both diabetic (13-week-old) (Fig. 2E) and non-diabetic (7-week-old) stages (data not shown). A similar increase of musclin mRNA was observed in the skeletal muscles of obese db/db mice (data not shown).
Next, we analyzed musclin mRNA expression in mouse myoblast cell line, C2C12. Musclin mRNA was not detected before induction (day 0) and markedly induced after differentiation of C2C12 myotubes, reaching a peak level at a later stage of differentiation (Fig. 3A), compared with mRNAs of other myogenic proteins, such as myogenin, GLUT4, PGC-1␣, and UCP3 (Supplemental Fig. 2A). Recombinant musclin and retrovirusmediated expression of musclin protein had no effect on proliferation and differentiation of C2C12 myoblasts, suggesting musclin had no significant effect on myogenesis (data not shown). These results suggested that the physiological significance of musclin is related to some yet unknown function of fully differentiated myocytes.
We also examined the regulation of musclin mRNA in fully differentiated C2C12 myocytes (day 7). Insulin markedly and dose-dependently augmented musclin mRNA expression (Fig.  3B), but not that of myogenin, a myocyte differentiation marker gene (Supplemental Fig. 2B), suggesting that insulin-stimulated musclin mRNA expression in myocytes was not due to overdifferentiation of myocytes but rather to direct induction of musclin mRNA. IGF-1 also induced musclin mRNA expression (Fig. 3B). On the other hand, epinephrine and ␤-adrenergic receptor agonist, isoproterenol, counter-regulators to insulin in glucose disposal in skeletal muscles, markedly decreased musclin mRNA expression (Fig. 3C) but not GLUT4 and myogenin (Supplemental Fig. 2C). Forskolin, which increases the intracellular cAMP contents similarly to epinephrine and isoproterenol, markedly and dose-dependently suppressed musclin mRNA expression (Fig. 3D) but not those of GLUT4 and myogenin (data not shown).
Finally, we assessed the biological role of musclin on insulinstimulated glucose uptake and glycogen synthesis in differentiated C2C12 myocytes using recombinant musclin protein purified from media of stable C2C12 expressing FLAG-tagged musclin (Fig. 3, E and F). The cells were pretreated with 0.5 g/ml recombinant musclin or FLAG peptide for 5 h and stimulated with phosphate-buffered saline or 100 nM insulin for 30 min (for 2-DG uptake) or 60 min (for glycogen synthesis). Treatment with insulin significantly increased 2-DG uptake and glycogen synthesis (Fig. 3, E and F). Musclin significantly attenuated insulin-stimulated 2-DG uptake (Fig. 3E) and glycogen synthesis at both basal and insulin-stimulated status (Fig. 3F). This 5-h treatment of musclin did not alter the expression levels of GLUT4, GLUT1, and hexokinase II mRNAs (data not shown).

DISCUSSION
Based on the hypothesis that skeletal muscles produce nutritionally regulated secretary factors, we identified musclin cDNA using an efficient SST technique. Although the precise physiological significance of musclin has yet to be determined, several important features were defined in the present study. 1) Musclin protein is actively secreted; 2) musclin mRNA is almost exclusively expressed in skeletal muscle; 3) musclin mRNA expression is dynamically regulated by nutritional changes and hormonal factors, especially insulin; 4) musclin mRNA expression is augmented in the skeletal muscle of obese insulin-resistant mice; and 5) recombinant musclin inhibited insulin-stimulated glucose uptake and glycogen synthesis in myocytes. These observations suggest that musclin could func- tion as an autocrine and paracrine factor in skeletal muscles and that increased production of musclin might be related to insulin resistance in the skeletal muscle of obese mice.
Musclin protein contained a homologous region to the functional 17 amino acids of the NP family, and basic KKKR sequence, a putative serine protease cleavage site. Proteins of the NP family are cleaved at this site and produce COOH-terminal active forms (12). Indeed, in cultures of mammalian cells transfected with musclin cDNA, full-length musclin protein and a trace amount of 76 KKKR 79 -dependent cleaved form were secreted. Different from the NP family, the conserved 17 amino acids of musclin are not positioned between two cysteine residues, crucial for physiological protein folding to produce natriuretic activity. Therefore, it is unlikely that musclin has a natriuretic activity.
The size of the recombinant musclin protein purified from conditioned medium of mammalian cells expressing musclin cDNA was 15-20 kDa, although the recombinant protein extracted from Escherichia coli was 11 kDa (data not shown). This size difference suggests that musclin protein undergoes protein modification in mammalian cells. However, there is no distinct N-glycosylation site in musclin, and treatment with neuraminidase, which cuts out O-glycosylation, did not a produce size-different band in Western gel (date not shown), suggesting that musclin protein modification cannot be explained by typical glycosilation. Other possible forms of modifications including fatty acidation (i.e. grelin) (13) and amidation (i.e. adrenomedulin) (14) remain to be examined, since such modifications could be major regulators of the protein activity (13,14).
We still do not know whether musclin is secreted into blood as an endocrine factor. Bruning et al. (6) reported that musclespecific insulin receptor knock-out mice exhibited adipose hyperplasia and increased mass, suggesting insulin-mediated production of muscle-derived secretory factor that decreases adiposity. In preliminary experiments, we noted that adenovirus-mediated production of musclin protein in plasma significantly decreased fat mass in mice. Musclin could be the missing link from skeletal muscles to fat, receiving intense insulin regulation and controlling adiposity, although musclin mRNA expression was low in skinny STZ-treated mice and high in obese KKAy mice.
Conceivably, musclin may transmit important signal(s) to the skeletal muscle itself or to remote organs. Thus, establishment of a reliable method to measure musclin protein level in blood and knockout mice should enhance our understanding of the physiological significance of this muscle-derived secretory factor. . The expression level was expressed relative to that of cyclophlin mRNA (n ϭ 3, each). E and F, effect of recombinant musclin on 2-DG uptake and glycogen synthesis. After 5-h pretreatment with FLAG peptide (0.5 g/ml) as control or FLAG-tagged musclin (0.5 g/ml), 2-DG uptake (E) and glycogen synthesis (F) were determined as described under "Experimental Procedures." Data are mean Ϯ S.E. values (n ϭ 6). **, p Ͻ 0.01. Similar results were obtained in two other independent experiments.