An Essential Role of Small Ubiquitin-like Modifier (SUMO)-specific Protease 2 in Myostatin Expression and Myogenesis*

Background: It is unknown whether SUMOylation regulates myogenesis in skeletal muscle. Results: SENP2 de-SUMOylates MEF2A, promoting myostatin expression and suppressing myogenesis in skeletal muscle. Conclusion: SENP2 plays a critical role in the regulation of myostatin-induced inhibition of myogenesis. Significance: SENP2 is a potential therapeutic target in skeletal muscle regeneration. Sentrin/small ubiquitin-like modifier (SUMO)-specific protease 2 (SENP2) has broad de-SUMOylation activities in vitro, which is essential for embryonic heart development. Here, we show that myostatin, a key factor in skeletal muscle development, is markedly reduced in Senp2−/− mouse embryonic fibroblast cells and embryos. SENP2 regulates the transcription of myostatin mainly through de-SUMOylation of MEF2A. Silencing SENP2 can reduce myostatin expression and, therefore, promote myogenesis of skeletal muscle. These results reveal the important role of SENP2 in the regulation of myostatin expression and myogenesis.

The small ubiquitin-like modifier (SUMO) 3 covalently modifies a large number of cellular proteins and regulates the localization, function, and protein-protein interaction of proteins (1)(2)(3). The transcription factors are the most popular targets for SUMO conjugation, which suggests an important role of SUMOylation in the regulation of gene expression (4,5). SUMOylation is catalyzed by SUMO-specific E1s, E2s, and E3s, and is reversed by a family of Sentrin/SUMO-specific proteases (SENPs) (6,7). Studies have shown that SENPs are important determinants of the SUMO modification status in cells (8 -10). Although the biochemical properties of SENPs have been well documented, their targets with involved biological processes are largely unknown.
There are six mammal SENPs with different subcellular locations and substrate specificities (3,9,10). SENP2 was originally reported to be associated with the nuclear envelope through the binding to Nup153 (11,12). We have shown previously that SENP2 contains both nuclear import and export signals that can shuttle between the nucleus and cytoplasm (13). SENP2 is the key regulator of Pc2/CBX4 function through the regulation of its SUMOylation status, which is critical for embryonic heart development (14).
Myostatin (growth and differentiation factor 8, GDF8) is a member of the TGF-␤ superfamily of secreted growth/differentiation factors (15)(16)(17). Myostatin is primarily expressed by skeletal muscle and acts in an autocrine manner to inhibit myoblast proliferation, differentiation, and protein synthesis (18,19). Disruption of the myostatin gene in mice causes a large and widespread increase in skeletal muscle mass because of cell hyperplasia and hypertrophy (20 -22). An inactivating mutation of the myostatin gene in a child has been reported to be hypermuscular (23). However, the overexpression or lack of active myostatin does not affect cardiac systolic function, although it reduces cardiac mass and cardiomyocyte proliferation (24).
Myostatin promoter activity is upregulated by differentiation. Both the myogenic regulatory factor and the myocyte enhancer factor 2 family of muscle transcription factors increase myostatin promoter activity (25,26). Myostatin transcription is also upregulated by dexamethasone treatment (27). However, it is not currently known whether SUMOylation regulates myostatin expression and myogenesis. In this study, we found that the deletion of the Senp2 gene in mice diminished myostatin expression. We further confirmed that MEF2A is a target for SENP2 de-SUMOylation activity and that SENP2 promotes myostatin expression and inhibits myogenesis through MEF2A. These results reveal the critical role of SENP2 in muscle development through the MEF2A-myostatin pathway.

EXPERIMENTAL PROCEDURES
Cell Culture-The generation of MEF cells has been described previously (14). The C2C12 cell line was purchased from the ATCC. Satellite stem cells were isolated from the hind limb muscles of 6-week-old C57BL/6 mice. All cell lines were cultured in DMEM (Invitrogen) supplemented with 10% FBS, 100 units/ml penicillin, and 100 g/ml streptomycin.
Real-time Quantitative PCR-Total RNA was isolated by an RNeasy kit (Qiagen) and treated with DNase (Promega). Complementary DNA was synthesized using the cDNA synthesis kit (Clontech) according to the instructions of the manufacturer. Fluorescence real-time PCR was performed with the TaqMan probe using the ABI Prism 7300 system (PerkinElmer Life Sciences). PCR was done in triplicate, and standard deviations representing experimental errors were calculated. All data were analyzed using ABI Prism SDS 2.0 software (PerkinElmer Life Sciences).
Mystatin Promoter-Luciferase Plasmids-A myostatin bacterial artificial chromosome (BAC) clone (BACPAC Resources Center) was used to amplify the myostatin promoter ( Fig. 2A) with specific primers (the sequences of primers will be provided upon request). The primers included a cleavage recognition site for either KpnI (forward primer) or XhoI (reverse primer) to facilitate subcloning into the multiple cloning sites of the pGL3-basic vector.
Skeletal Myogenesis-C2C12 cells were cultured in growth medium (DMEM supplemented with 10% FBS). When cells reached about 80 -90% confluence, the growth medium was aspirated, and cells were switched to differentiation medium (DMEM supplemented with 2% horse serum). Refeeding with fresh differentiation medium was done every 48 h. After 6 days of differentiation, the cells were stained with myosin heavy chain antibody to show myotube formation. DAPI was used to stain the nuclei.
Satellite Stem Cell Differentiation-Satellite stem cells were isolated from C57BL/6 mouse skeletal muscle. Cells with round morphological characteristics were selected and defined as primitive muscle satellite stem cells. The genetic marker PAX7 was used to identify satellite stem cells. To induce satellite stem cell differentiation, the cells were grown to 80 -90% confluence, growth medium was aspirated, and cells were switched to differentiation medium (DMEM supplied with 2% horse serum). Refeeding with fresh differentiation medium was done every 48 h. After 6 days of differentiation, the cells were stained with myosin heavy chain antibody to show myotube formation. DAPI was used to stain the nuclei.
Cachexia Mouse Model-10 6 LLC cells in 0.1 ml PBS and PBS control were injected into the tail veins of 8-week-old C57BL/6 mice. Six weeks after LLC cell injection, when tumors appeared in the lungs, body weight was measured, and the skeletal muscle was harvested. The SENP2 and myostatin mRNA expression levels were measured via real-time PCR.
Statistical Analysis-Results were presented as mean Ϯ S.D. Student's t test was used to compare the differences between two groups.

SENP2 Modulates Myostatin
Expression-Because SENP2 is highly expressed in the dermomyotome in mouse embryos (data not shown), we reasoned that SENP2 might be an important regulator in myogenesis. Microarray profiles from Senp2 ϩ/ϩ and Senp2 Ϫ/Ϫ embryos at embryonic day 10.5 showed that myostatin mRNA in Senp2 Ϫ/Ϫ embryos was reduced significantly compared with that in Senp2 ϩ/ϩ embryos. We evaluated this observation in Senp2 Ϫ/Ϫ MEF cells and embryos by using real-time PCR, and the evaluation showed that in both MEFs and embryos, myostatin mRNA was decreased in comparison with that in the Senp2 ϩ/ϩ controls (Fig. 1A). To further determine the role of SENP2 in myostatin expression, we overexpressed SENP2 in murine myoblast C2C12 cells, which showed that the overexpression of SENP2 increased myostatin expression (Fig. 1B). Interestingly, the SENP2 catalytic mutant did not have such an effect (Fig. 1B), suggesting that the de-SUMOylation activity is essential for SENP2 to regulate myostatin expression. We also confirmed the role of SENP2 in the regulation of myostatin expression by using the SENP2 siRNA approach. As shown in Fig. 1C, silencing SENP2 significantly reduced myostatin expression. Collectively, these results show a crucial role of SENP2 in myostatin expression.
MEF2A Mediates the Effect of SENP2 on Myostatin Expression-To explore how SENP2 regulates myostatin expression, we generated a myostatin promoter (Ϫ2858 to ϩ152 bp of myostatin genomic DNA)-luciferase report gene ( Fig. 2A, pMSTN6). Cotransfection of SENP2 significantly induced the transcription activity of the myostatin promoter (Fig. 2B). To further map the SENP2 response element of the myostatin promoter, we tested a series of truncated forms of myostatin promoter-luciferase plasmids and found that the SENP2-responsive element was located between Ϫ1938 to Ϫ1358 bp of the myostatin promoter ( Fig. 2A). Consistent with the observation in Fig. 1B, we found that SENP2 catalytic activity was also required to activate myostatin promoter transcription (Fig. 2B). The TESS program predicted a MEF2 binding site located in this area (Ϫ1336 to Ϫ1326 bp of the myostatin promoter) (Fig. 2A). We tested whether this MEF2 binding site might mediate SENP2 action on myostatin promoter activity by a mutation approach. As expected, we found that the SENP2 activation on the myostatin promoter was almost abolished when the MEF2 binding site was mutated (Fig. 2C). The

SENP2 Regulates Myostatin Expression and Myogenesis
of myostatin in a reporter gene assay (Fig. 2D). Thus, we speculated that MEF2A was essential for SENP2 in the regulation of myostatin expression. Indeed, silencing only MEF2A, no other MEF2 genes blocked SENP2 activation of the myostatin promoter. These data suggest a critical role of MEF2A in SENP2 activation of myostatin transcription. SENP2 De-SUMOylates MEF2A-Given that MEF2A is a SUMOylated protein, the induction of myostatin might be through the de-SUMOylation of MEF2A by SENP2. Therefore, we first assessed whether SUMOylation of MEF2A would have an effect on myostatin transcription. Indeed, the SUMOylation site mutant of MEF2A induced more activity of the myostatin promoter than the MEF2A wild type (Fig. 3A). We further showed that the overexpression of SUMO-1 suppressed the activity of the MEF2A wild type, not the MEF2A mutant, on the myostatin promoter (Fig. 3A). These results suggest that SUMOylation can negatively regulate MEF2A transactivation on the myostatin promoter.
We next determined whether SENP2 de-SUMOylated MEF2A in vivo and in vitro. In the in vitro assay, MEF2A SUMOylation was detected in C2C12 cells transfected with MEF2A and HA-SUMO-1. The overexpression of SENP2, not the SENP2 mutant, deconjugated SUMO-MEF2A (Fig. 3B). The role of SENP2 in de-SUMOylation of MEF2A was also confirmed in Senp2 ϩ/ϩ and Senp2 Ϫ/Ϫ MEF cells. SUMO-MEF2A was detected easily detected and accumulated in Senp2 Ϫ/Ϫ MEF cells (Fig. 3C). These data indicate that SENP2 can de-SUMOylate MEF2A and enhance MEF2A transactivation on the myostatin promoter.
SENP2 Inhibits Skeletal Myogenesis-Because SENP2 promotes myostatin expression, we speculated that SENP2 would repress myogenesis as myostatin does. We tested this hypothesis in two cell models. We first used C2C12 cells transfected with SENP2 siRNA, MEF2A, or myostatin siRNA (Fig. 4A). Myostatin expression was reduced significantly in these cells compared with si-NS control cells (Fig. 4, A and B). These cells   FEBRUARY 7, 2014 • VOLUME 289 • NUMBER 6 were induced into myogenesis by 2% horse serum in culture. Myotube formation, as a marker for myogenesis, was checked in these cells by using immunofluorescence staining with antimyosin heavy chain antibody. As shown in Fig. 4C, silencing of MEF2A or SENP2 enhanced myotube formation in C2C12 cells as myostatin siRNA did. Another cell model used was satellite stem cells isolated from C57BL/6 mouse skeletal muscle. The isolated satellite stem cells were confirmed by PAX7 immunostaining (data not shown). Satellite stem cells transfected with MEF2A, SENP2, or myostatin siRNA were induced into differentiation by addition of 2% horse serum. As observed in the C2C12 cell model, myostatin expression was decreased significantly after MEF2A or SENP2 knockdown in satellite stem cells (Fig. 4B). Differentiation was shown by immunostaining with anti-myosin heavy chain antibody in myostatin siRNA-or SENP2 siRNA-transfected satellite stem cells (Fig. 4C). Like myostatin, SENP2 is a negative regulator in myogenesis.

The Correlation of Senp2 and Myostatin Expression in
Cachexia-Because myostatin expression is increased and related to weight loss and muscle atrophy during cachexia, we reasoned that SENP2 might be involved in cachexia through the regulation of myostatin expression. To test this possibility, we created a cancer cachexia mouse model by injecting LLC cells via the tail vein. Six weeks after LLC cell injection, the mice not only grew lung tumors (data not shown) but also showed a body weight loss of 45%, which is a typical cachexia condition (Fig.  5A). Interestingly, the expression of both Senp2 and myostatin mRNA showed a remarkable increase in the skeletal muscle of the cachexia mice (Fig. 5, B and C), suggesting that SENP2 might promote cachexia through the regulation of myostatin expression in cancer-bearing mice.

DISCUSSION
In this study, we found that SENP2 contributes to the regulation of myostatin expression in muscle cells. We also observed SENP2 acting as a negative regulator in myogenesis, like myostatin. We further observed that MEF2A, a specific transcription factor of myostatin, mediates SENP2 promotion for myostatin expression through de-SUMOylation. We detected the increase of both the expression of SENP2 and the expression of myostatin in a tumor-bearing cachexia model, suggesting that SENP2 plays a critical role in cancer-induced cachexia.
Senp2-null mice died from embryonic heart failure at around embryonic day 10 (14,30). Further evidence showed that the knockout of Senp2 decreased cardiomyocyte proliferation and  heart formation in mouse embryonic hearts (14). Interestingly, here we found that silencing SENP2 would promote the differentiation and myogenesis in skeletal muscle progenitor cells. The two phenotypes in Senp2 deficiency cells look contradictory. However, SENP2 acts with different mechanisms in these two different processes. In cardiomyocytes, SENP2 regulates Gata4/6 expression through the PcG complex, which is essential for heart development (14). However, the function of SENP2 in skeleton muscle is through the induction of myostatin, which suppresses myogenesis. These observations suggest that SENP2 has multiple functions in regulating biological processes involving different targets. In a future study, it will be interesting to identify by which mechanism SENP2 targets specific proteins in different signaling pathways.
Satellite stem cells are quiescent muscle stem cells (31)(32)(33). How to activate the differentiation process of satellite stem cells into new myofibers is still a critical issue for skeletal muscle regeneration. It seems that SENP2 functions in maintaining the quiescent status of satellite stem cells. When SENP2 expression is silenced or SENP2 activity is inactivated, satellite stem cells can differentiate into myocytes. Thus, SENP2 is a potential therapeutic target in skeletal muscle regeneration.