Absence of Hyperplasia in Gasp-1 Overexpressing Mice is Dependent on Myostatin Up-Regulation

Skeletal muscle is composed of heterogeneous populations of muscle fibers bundled together and which differ in their metabolism and contractile properties [1,2]. This type of organization confers to skeletal muscles remarkable levels of plasticity in the face of changing external conditions. During embryonic development, the number of myofibers and their size increase until birth [3]. Postnatal muscle growth is then achieved by hypertrophy of myofibers in mouse and can be divided into two distinct steps. Between birth and 3 weeks old, hypertrophy is supported by a rapid increase in the number of myonuclei within the myofiber via activation and fusion of satellite cells (myonuclear accretion) [4]. Then, satellite cells enter in quiescence and reach their adult level around the third week [5]. From 3 weeks old to adulthood, muscle mass regulation is dependent of protein synthesis mainly and the myofiber hypertrophy occurs without addition of new nuclei. During the past two decades, much progress has been made in unraveling the molecular mechanisms underlying either adult muscular hypertrophy (increase in myofiber size) or atrophy (decrease in myofiber size) [6,7]. The maintenance of the muscle homeostasis is finely regulated by the balance between protein synthesis and protein degradation. This protein turnover is induced in response to various stimuli such as exercise, inactivity or environmental factors (hypoxia, heat, nutrient availability, growth factors) [7]. Among these growth factors, the myostatin (Mstn), a member of the transforming growth factor-β (TGF-β) superfamily, is the most potent inhibitor of skeletal muscle mass [8,9].

Disruption of the Mstn gene in mice leads to a dramatic increase in skeletal muscle mass due to both hyperplasia and hypertrophy, and to an overall glycolytic muscle phenotype [8,10]. Moreover, Mstn-null mice (Mstn^-/-) have a reduced adipogenesis and their organ weights are lower [11]. Naturally occurring mutations in Mstn gene lead to a hypermuscular phenotype in several species, from mice to human [8,12,13,14,15].

Interestingly, the action of this growth factor on muscle mass is not restricted to the embryonic period, as myostatin is also able to regulate adult muscle growth and size. Mice carrying a postnatal deletion of the myostatin gene or adult mice injected with various inhibitors of myostatin exhibit also an increased muscle mass resulting only from hypertrophy, suggesting that myostatin regulates hyperplasia during embryonic development [16,17,18,19].

After binding to the activin receptor type IIB (ActRIIB), myostatin acts through its canonical Smad2/3 pathway to inhibit myoblast proliferation by up-regulation of p21 [20], and differentiation by inhibition of MyoD and myogenin [21]. The phosphorylation of Smad2/3 also reduces the Akt/mTOR/p70S6K protein synthesis pathway [22,23]. Myostatin is able to act independently of Smad2/3 via the mitogen-activated protein kinase (MAPK) pathway such as the extracellular signal-regulated kinase (Erk) 1/2 pathway leading to the down-regulation of Pax7 [24,25]. Notably, other members of TGF-β superfamily such as activin A or GDF-11 can bind to ActRIIB and stimulate the same intracellular signaling pathway [26].

Various myostatin-binding proteins have been identified that are able of inhibiting myostatin activity [27]. Among them, GASP-1 (growth and differentiation factor-associated serum protein-1) is a secreted glycoprotein that contains protease inhibitor domains such as the tandem Kunitz domains that are unique among the related myostatin-binding proteins, the follistatin domain and the netrin domain, which allow GASP-1 to bind both mature myostatin and myostatin propeptide respectively and thus inhibit myostatin activity [28,29,30]. It was also shown that GASP-1 exhibits a relatively high affinity for BMP-2, BMP-4 and TGF-β1 [31]. However, these interactions do not necessarily lead to inhibition of their signaling pathway. Only, the myostatin or GDF-11 signal transduction is inhibited by this interaction, suggesting an additional role for GASP-1 as a carrier protein. As we have recently shown, overexpressing Gasp-1 transgenic mice (called Tg(Gasp-1)) present an increased muscle mass due to myofiber hypertrophy rather than hyperplasia. No variation in fiber type proportions was observed. Furthermore, body fat deposition was not altered in the Tg(Gasp-1) mice [32]. All these data suggest that Gasp-1 overexpression has no effect on myostatin during prenatal muscle development. Body weight increase in Tg(Gasp-1) mice is visible at the third week of postnatal life and remains stable throughout adulthood, showing that Gasp-1 overexpression leads to the muscular fiber hypertrophy mainly during the first 21 days after birth and would later participate to skeletal muscle mass maintenance.

To investigate the cellular and molecular mechanisms regulating muscle mass in Tg(Gasp-1) mice, we examined the proliferation and differentiation capabilities in satellite cell-derived myoblasts from Tg(Gasp-1) mouse line compared to wild-type and Mstn^-/- myoblasts. As observed in Mstn^-/- cells, we showed that Tg(Gasp-1) myoblasts increase the Akt/mTORC/p70S6K signaling, enhancing myoblast differentiation and myotube size by protein synthesis. Moreover, we showed an increased number of myonuclei per myofiber in Tg(Gasp-1) skeletal muscle confirming a higher myonuclear accretion contributing to the muscle hypertrophy. Molecular analysis revealed a myostatin up-regulation associated to a Pax7 down-regulation through Erk1/2 signaling pathway. We demonstrated that this myostatin up-regulation counteracts Gasp-1 effect during embryonic muscle development leading to the absence of hyperplasia. At last, we showed that overexpression of an inhibitor of a TGF-β superfamily member leads to a global deregulation of these members and their related proteins.

Mstn^-/- mice, harboring a constitutive deletion of the third Mstn exon, and Tg(Gasp-1) mice, overexpressing Gasp-1 ubiquitously, have been described previously [16,32]. All mice were bred and housed in the animal facility of Limoges University under controlled specific pathogen free conditions (21°C, 12-h light/12-h dark cycle) with free access to standard mouse chow and tap water. All of the experimental procedures were carried out in accordance with the recommendations in the guidelines of the European Communities Council (Directive 2010/63/UE). Experiments were approved by the Committee on the Ethics of Animal Experiments of the Author's Institution, “Comité Régional d'Ethique de l'Expérimentation Animale” of the Limousin region (n°17-2013-17).

Primary myoblasts were obtained using mice between 4 and 6 weeks of age. Concisely, murine myoblasts were isolated from hind limb muscles after enzymatic digestion by pronase. Cells were plated at a density of 20,000 cells/cm2 on Matrigel®-coated Petri dishes (BD Biosciences) in Ham's F10 supplemented with 20% horse serum and 1% penicillin/streptomycin. Cells were maintained at 37°C in a water-saturated atmosphere containing 5% CO2 in air for 2 days. Then, cells were washed with Ham's F10 (Gibco) and placed in complete growth medium supplemented with 5 ng/ml basic fibroblast growth factor (bFGF, Invitrogen), 20% horse serum and 1% penicillin/streptomycin (Growth Medium, GM). The myoblast population was enriched by differential adhesion compared with fibroblasts by serial 30-min preplate procedures after trypsinization. To induce differentiation (0 h of differentiation), primary myoblasts were reached at 80% confluence and switched in Differentiation Medium (DM) consisting of Ham's F10 containing 15% horse serum and 1% penicillin/streptomycin.

Myoblasts were treated with a recombinant myostatin protein (788-G8-010/CF, R&D Systems) at the final concentration of 250 ng/ml or with the same amounts of 4 mM HCl as control for 16 h. For Erk1/2 pathway inhibition, cells were pre-treated for 2 h with 5, 10 or 20 µM of a selective ERK inhibitor FR180204 (R&D Systems) or with the same amounts of DMSO as control. For myostatin neutralizing, cells were incubated with a goat anti-myostatin antibody (sc-6884, Santa Cruz) at different concentrations (2.5 to 10 µg/ml) for 16 h or with the same amounts of an isotype control (normal goat IgG, sc-2028, Santa Cruz).

Primary myoblast proliferation was assessed as described previously [33]. Cells were seeded at 2,500 cells per well in GM in 96-well microtiter plates and fixed at regular 24 h periods, before the Methylene Blue staining. Absorbance at 590 nm was read using an ELISA plate reader. Fusion index was determined as the ratio of the nuclei number in the cells containing two or more nuclei to the total nuclei number in hematoxylin-stained primary myotubes. Cells were stained with hematoxylin/eosin as previously described [29].

Total RNA from cells at each kinetic point and from embryos and skeletal muscles of wild-type and Tg(Gasp-1) was obtained by anion exchange chromatography (RNeasy mini Kit, Qiagen Inc). The High Capacity cDNA Archive Kit (Applied Biosystems) was used to convert 2 µg of total RNA into single-stranded cDNA. Taqman primers and probe sets used in this study were as follows: 18S (Hs99999901_s1), Gapdh (Mm99999915_g1), Gasp-1 (Mm00725281_m1), myostatin (Mm03024050_m1). Quantitative PCR reaction was performed from 20 ng of cDNA in an ABI Prism 7900HT Sequence Detection System (Applied Biosystems) using 40 cycles of 95°C for 15 sec followed by 60°C for 1 min. Relative mRNA expression values were calculated by the ΔΔCt method with normalization of each sample to the average change in cycle threshold value of the controls. TLDA (Taqman low density array, Applied Biosystems) assays were performed based on the same above conditions, except from each sample, 200 ng cDNA were used per TLDA card. Two TLDA cards were used: in the first one, the 93 selected target genes involved in myogenesis were described previously [32]; in the second one, the 43 selected genes involved in TGF-β signaling pathway were listed in the Table 2. Each card features five control assays: 18S, Gapdh, Tbp (Mm00446973_m1), Dffa (Mm00438410_m1) and Fcgrt (Mm00438887_m1).

Cell supernatants were concentrated by centrifugation at 3,000 × g with Amicon Ultra-15 Centrifugal Filter Unit with Ultracel-3 membrane (Millipore). GASP-1 concentrations from concentrated cell supernatants or mouse plasmas were determined in a sandwich ELISA (GASP-1 DuoSet ELISA kit, R&D Systems), as described previously [29]. Myostatin concentrations were also determined in a sandwich ELISA (GDF-8/Myostatin Quantikine ELISA kit, R&D Systems). Following the manufacturer recommendation, the propeptide was removed from myostatin by an acid activation followed by neutralization. The final dilutions measured for concentrated cell supernatants or mouse plasmas were 1:10 and 1:40 respectively. All measurements were performed in triplicate and data for the standard curve were fitted to a logistic plot with the MARS Data Analysis Software (BMG Labtech) to determine the levels of GASP-1 or myostatin.

To investigate the pathways by which GASP-1 enhances myogenesis, we performed a determination of phospho-kinase and apoptosis-related proteins using Proteome Profiler Arrays (Human Apoptosis Array, Human Phospho-Kinase Array, Human Phospho-MAPK Array, R&D Systems), according to the manufacturer's instructions. Briefly, the cells were collected, rinsed in 1× PBS and lysed in lysis buffer. The arrays were incubated overnight with the diluted lysates (300 µg) at 4°C on a rocking platform shaker. The blots were detected using an enhanced chemiluminescence (BM Chemiluminescence Western Blotting Substrate (POD)) (Roche Applied Science) and exposed to a film (GE Healthcare Hyperfilm ECL, GE Healthcare). Densitometrie analysis of the array image files were performed using ImageQuant TL software (GE Healthcare).

Cells and mouse tissues were collected by centrifugation (12,000 × g, 4°C, 20 min) and then lysed in a RIPA buffer (50 mM Tris, pH 8, 150 mM NaCl, 0.1% SDS, 1% NP-40, 0.5% sodium deoxycholate, and protease inhibitors). Proteins were quantified at A595nm using the Bradford assay (Bio-Rad). Equal amounts of proteins (20-50 µg) were resolved on SDS-polyacrylamide gels (4-12% gradient, Invitrogen) and transferred onto 0.2 µm nitrocellulose membranes. Then, membranes were blocked using 5% non-fat dry milk (w/v) in TBST0.1% buffer (50 mM Tris-HCl, 150 mM NaCl, pH 7.4, 0.1% Tween-20) for 1 h at room temperature, followed by incubation with specific primary antibodies overnight at 4°C. The following primary antibodies were used for immunoblotting: 1:5000 dilution of anti-Akt antibody (MAB2055, R&D Systems), 1:500 dilution of anti-phospho-Akt antibody (AF887, R&D Systems), 1:400 dilution of anti-cdk2 (sc-6248, Santa Cruz), 1:400 dilution of anti-cyclinD1 (sc-8396, Santa Cruz), 1:1000 dilution of anti-Erk1/2 (AF1576, R&D Systems), 1:500 dilution of anti-phospho-Erk1/2 (MAB1018, R&D Systems), 1:2000 dilution of anti-GAPDH (AF5718, R&D Systems), 1:1000 dilution of anti-GASP-1 (AF2070, R&D Systems), 1:4000 dilution of anti-MyHC (M4276, Sigma-Aldrich), 1:1000 dilution of anti-MyoD (M3512, Dako), 1:1000 dilution of anti- Myogenin (M3559, Dako), 1:2000 dilution of anti-Smad2/3 (AF3797, R&D Systems), 1:500 dilution of anti-phospho-Smad3 (AB3226, R&D Systems), 1:200 dilution of anti-p21 (sc-53870, Santa Cruz) and 1:1000 dilution of anti-β-actin (sc-1615, Santa Cruz). After four washes inTBST0.1%, membranes were incubated for 1 h at room temperature with 1:1000 dilution of secondary antibodies: anti-goat IgG Horse-radish peroxidase (HRP) conjugate (P0449, Dako), anti-mouse IgG HRP conjugate (P0447, Dako), or anti-rabbit HRP conjugate (P0399, Dako). After four more washes in TBST0.1%, immunoblots were developed by enhanced chemiluminescence. The developed films were analyzed using ImageQuant TL software (GE Healthcare).

Primary myoblasts were fixed in paraformaldehyde (PFA) 4%/PBS, washed three times in PBS, and permeabilized with HEPES/Triton X-100 buffer, pH7.4, consisting of 20 mM HEPES, 300 mM sucrose, 50 mM NaCl2, 3 mM MgCl2 and 0.5% Triton X-100. Dissected skeletal muscles were frozen in liquid nitrogen-cooled isopentane and stored at -80°C for further analysis or sectioned (8 µm thick) for immunostaining. Cryosections were thawed at room temperature and air-dried, fixed in PFA 4%/PBS, washed three times in PBS before permeabilization with cold methanol at -20°C and processed for antigen retrieval in 10 mM citrate buffer, pH 6 at 90°C for 2 × 5 min in a microwave. Then, cells and cryosections were blocked for 1 h in blocking buffer consisting of 10% goat serum, 1% bovine serum albumin (BSA) and 0.1% Triton X-100 in PBS, at room temperature. Incubation with primary antibodies diluted in BSA 1%/PBS took place overnight at 4°C. After three washes in BSA 0.2%/Tween-20 0.1%/PBS, slides were incubated for 15 min at 37°C with secondary antibodies conjugated to a fluorescent dye (Alexa Fluor® 488 Goat Anti-Mouse IgG (H+L) (A10680, Invitrogen) for anti-MyHC and anti-Pax7 or Alexa Fluor® 546 F(ab')2 Fragment of Goat Anti-Rabbit IgG (H+L) (A11071, Invitrogen) for anti-laminin and anti-MyoD) diluted in BSA 1%/PBS. The staining was completed with three washes in PBS and incubation for 5 min at room temperature in DAPI solution to label cell nuclei. Primary antibodies used for these analyses were as follows: anti-laminin (L9393, Sigma-Aldrich), anti-MyHC (M4276, Sigma-Aldrich), anti-MyoD (sc-304, Santa Cruz) and anti-Pax7 (DSHB). Images were acquired with a Leica DMI6000B inverted epifluorescence microscope using the MetaMorph software (Molecular Devices, Sunnyvale, USA). Myofiber cross-sectional areas were calculated from laminin-stained cryosections from randomly chosen fields using ImageJ software v1.43 (http://rsbweb.nih.gov/ij/). Myonuclei were manually counted in images captured at 40X magnification using the MetaMorph software to determine the number of myonuclei per fiber. Myonuclei were selected as they were located within the laminin boundary and as they are Pax7^-. MyoD/Pax7 cell populations were counted manually from 20 fields per experiments using the same software.

Tg(Gasp-1) mice have an overall increase in body weight as a result of an increase in muscle mass [32]. To confirm whether the muscle mass increase is due to hypertrophy rather than hyperplasia of all skeletal muscles, we performed histological analysis on two skeletal muscles of 12-week-old mice, tibialis anterior (TA) and soleus, which stand out for their fiber metabolism, glycolytic or oxidative respectively (Fig. 1A). Whatever muscle tested, no significant difference in muscle fiber number was observed between wild-type (WT) or Tg(Gasp-1) mice, underlying a lack of hyperplasia in contrast to the observed phenotype of Mstn^-/- mice (Fig. 1C). However, Tg(Gasp-1) mice has an increased muscle fiber cross-sectional area (CSA) when compared to the WT mice, confirming that muscle mass increased is due to hypertrophy (Fig. 1D). As hypertrophy could be related to the addition of new nuclei from activated satellite cells within the myofiber and/or to the increased rate of protein synthesis, we analyzed the number of myonuclei per myofiber in TA and soleus skeletal muscles of 12-week-old mice. Satellite cell nuclei labelled by Pax7 were excluded from this counting and only myonuclei beneath the basal lamina labelled by laminin were was assessed by a Student's t test analysis when compared counted (Fig. 1B). We showed an increased number of myonuclei per myofiber in Tg(Gasp-1) skeletal muscle demonstrating a higher myonuclear accretion (Fig. 1E). We also compared the number of satellite cell (Pax7⁺) into the cross-sections of tibialis and soleus skeletal muscles of 12-week-old mice. As shown in Fig. 1F, the number of satellite cells is unaffected in overexpressing Gasp-1 skeletal muscles when compared to the WT, suggesting that GASP-1 promotes hypertrophy independently of the adult satellite cell pool. Thus, the increase of myonuclei number per myofiber seems to be related to an activation and fusion of satellite cells prior to the first 12 weeks of life.

To investigate the molecular mechanisms regulating muscle mass in Tg(Gasp-1) mice, myoblasts derived from Tg(Gasp-1) and wild-type WT satellite cells were compared in vitro. First, the proliferation rate was monitored during 72 hours and revealed that Tg(Gasp-1) myoblasts proliferated faster than the WT myoblasts (Fig. 2A). We also analyzed the expression of Cdks and their associated cyclins, especially Cyclin-dependent kinase 2 (Cdk2) and cyclin D1, which are known to positively regulate cell-cycle progression. Western blot analysis showed an increase of Cdk2 and cyclin D1 in Tg(Gasp-1) myoblasts compared to WT myoblasts, consistent with the increase of proliferation rate, while p21, a Cylin-dependent kinase inhibitor (CKI), decreased in Tg(Gasp-1) myoblasts (Fig. 2B and 2C). These results revealed that Gasp-1 overexpression promoted cell-cycle progression.

To analyze the effect of Gasp-1 overexpression during myoblast differentiation, fusion indexes of WT and Tg(Gasp-1) myotubes were determined. After 72 hours of differentiation, we observed that both WT and Tg(Gasp-1) myoblasts fused to form myotubes and Tg(Gasp-1) myotubes had a greater size (Fig. 3A). We showed that Gasp-1 overexpression in myotubes led to an increase of differentiation when compared to the WT cells (Fig. 3B). Average numbers of mononuclei per myotube revealed that Tg(Gasp-1) myotubes had more myonuclei than WT myotubes, especially in late stages of differentiation, suggesting an increase of myoblast fusion (Fig. 3C). Consistent with this increase of differentiation, some gene expression results showed that the myogenic regulatory factors (MRFs), MyoD and myogenin, were up-regulated in the Tg(Gasp-1) myoblasts when compared to the WT cells, as well as p21 and MyHC all along the time course (Fig. 3D and 3E). All these results correlated to an increase of differentiation when Gasp-1 is overexpressed.

In order to further characterize the molecular basis for the observed improvement of myogenic processes, we performed an expression array analysis of 91 genes involved in muscle development during proliferation, 48 h after plating (n = 5 independent experiments). Gasp-1 expression was increased more than 200-fold in the Tg(Gasp-1) myoblasts compared to the WT, confirming that Gasp-1 overexpression was significant in Tg(Gasp-1) satellite cell-derived primary myoblasts (Table 1). We also found that myostatin and Inhbb (Inhibin beta B chain) were also up-regulated in Tg(Gasp-1) cells. In contrast, genes that have been previously described as components of myostatin pathway - Fst (follistatin), Fstl3 (follistatin-like 3) and Smad7 - were down-regulated in Tg(Gasp-1) myoblasts (Table 1). Interestingly, Gasp-1 overexpression led to down-regulation of three TGF-β superfamily members - Bmp4,Bmp2 and Tgfb1 - known to interact with GASP-1 which doesn't inhibit their signaling activity (Table 1) [31]. Pax7 and Pax3, two transcription factors known to be involved in the regulation of skeletal muscle stem cells, were also down-regulated. Finally, we observed a decreased expression of Igf-1 (insulin-like growth factor-1) known to have a pro-hypertrophic activity [34]. Since expression array analysis revealed an up-regulation of myostatin expression in overexpressing Gasp-1 cells, we performed a qPCR of these two genes during the first 96 h of differentiation of Tg(Gasp-1) compared to the WT myotubes. As seen in Fig. 4A and 4B, Gasp-1 and myostatin were stably up-regulated all along the time course. The ELISA provided the evidence that more GASP-1 and myostatin were found secreted from primary Tg(Gasp-1) myoblasts (Fig. 4C and 4D).

Since Gasp-1 overexpression seems to deregulate expression of myostatin and TGF-β superfamily-related genes in Tg(Gasp-1) primary myoblasts, we investigated their expression levels at different steps of muscle growth during late embryonic starting at E9.5, fetal starting at E14.5 and at postnatal day 15 (P15) (Table 2) [35]. Similarly to primary myoblasts, Bmp4, Bmp2, Tgfb1, Fst, Fstl3 and Smad7 were down-regulated as Pax3 and Pax7 in Tg(Gasp-1) mice compared to the wild-type (Table 2). We also found an up-regulation of Inhbb, Mstn and Gasp-1 (Table 2). Western blotting analysis of GASP-1 confirmed the protein overexpression at all the studied stages of mouse development (Fig. 5A). We also showed that GASP-1 protein and active mature form of myostatin were more secreted in Tg(Gasp-1) mice serum compared to the WT (Fig. 5B-C). This up-regulation of myostatin in Tg(Gasp-1) myoblasts, confirmed in mice, led us to analyze the signaling pathway activated in these cells.

To characterize the signaling pathways affected by Gasp-1 overexpression in myoblasts, three types of microarrays directed against 84 phospho-proteins were used (Fig. 6A). Among them, 13 proteins were significantly affected as shown in Fig. 6B. Gasp-1 overexpression increased the phosphorylation of Akt1 (S473) and Akt2 (S474) in agreement with the inhibition of myostatin pathway (Fig. 6B, positions 1-3). It also increased mTOR and p70S6 kinase activation in Tg(Gasp-1) myoblasts when compared to the WT cells, suggesting an overall increase in Akt/mTOR/p70S6K protein synthesis pathway (Fig. 6B, positions 7, 10). Heat shock proteins 27 and 70 (HSP27 and HSP70), which are known to prevent skeletal muscle atrophy [36,37], were significantly more phosphorylated in Tg(Gasp-1) myoblasts as HSP60 (Fig. 6B, positions 4-6). Conversely, the phosphorylation of CKIs p21 and p27 had decreased in Tg(Gasp-1) myoblasts compared to the WT (Fig. 6B, positions 8-9). Surprisingly, the phosphorylation of Erk1/2 had increased in Tg(Gasp-1) primary myoblasts when compared to WT myoblasts (Fig. 6B, positions 11-13). To confirm that Gasp-1 overexpression led to the inhibition of the canonical myostatin pathway in Tg(Gasp-1) primary myoblasts, the phosphorylation rate of Smad3 was assessed between WT, Tg(Gasp-1) and Mstn^-/- myoblasts. Immunoblot analysis revealed a decrease of phospho-Smad3 in Tg(Gasp-1) and Mstn^-/- compared to the WT, confirming the inhibition of the myostatin-induced activation of Smad2/3 signaling pathway (Fig. 6C and 6D). We also compared the Akt and Erk1/2 phosphorylation rates in WT, Tg(Gasp-1) and Mstn^-/- myoblasts. Akt phosphorylation had increased in Tg(Gasp-1) and Mstn^-/- myoblasts compared to the WT, Mstn^-/- myoblasts having the highest rate of phospho-Akt (Fig. 6C and 6D). Nevertheless, the analysis of Erk1/2 phosphorylation rate revealed that only Tg(Gasp-1) myoblasts had increased levels of activated Erk1/2 (Fig. 6C and 6D).

Because Erk1/2 signaling pathway is involved in myostatin-induced Pax7 down-regulation [24], we analyzed the level of Pax7 in WT, Tg(Gasp-1) and Mstn^-/- primary myoblasts with two independent experiments, namely Pax7 western blotting and immunofluorescence. Western blot analysis during the first 48 hours of differentiation revealed that Pax7 level had increased in Mstn^-/- cells compared to the WT and Tg(Gasp-1) myoblasts. During proliferation, Tg(Gasp-1) myoblasts showed a slight decrease of Pax7 levels compared to the wild-type but it seemed to have increased during differentiation (Fig. 7A and 7B). Then, we analyzed the proportions of Pax7 and MyoD expressing cells between WT, Tg(Gasp-1) and Mstn^-/- proliferating myoblasts by immunofluorescence (Fig. 7C). We obtained the same results for WT and Tg(Gasp-1) myoblasts, although Tg(Gasp-1) myoblast cultures had a slightly decreased proportion of undifferentiated Pax7⁺/MyoD^- cells compared to the WT cultures (Fig. 7D). However, WT and Tg(Gasp-1) myoblasts were dramatically different from the Mstn^-/- myoblasts which had the highest proportion of undifferentiated Pax7⁺/MyoD^- cells and a reduced proportion of committed Pax7⁺/MyoD⁺ cells and differentiated Pax7^-/MyoD⁺ cells (Fig. 7D). Specifically, overexpression of Gasp-1 seemed to result in reduced level of Pax7 when compared to the Mstn^-/- cells, which may be responsible of a decrease of progenitor muscle cells in Tg(Gasp-1) mice compared to the Mstn^-/- mice. Therefore, we hypothesized that deregulation of Pax7 could be related to the myostatin up-regulation and with the activation of Erk1/2 signaling pathway.

To verify this hypothesis, we first analyzed the ability of myostatin to regulate Pax7 through Erk1/2 pathway in WT and Mstn^-/- myoblasts. When treated with a recombinant myostatin protein, both WT and Mstn^-/- myoblasts showed a significant decrease of Pax7 levels (Fig. 8A and 8B). In contrast, treatment with ERK inhibitor increased Pax7 expression in WT myoblasts (Fig. 8A). Mainly, the combination of treatments by myostatin and ERK inhibitor prevented the decrease of Pax7 expression, confirming that myostatin signals through Erk1/2 pathway to regulate Pax7 (Fig. 8A). This phenomenon was also observed in Mstn^-/- myoblasts, although a treatment with only ERK inhibitor didn't seemed to induce an increase of Pax7 in these cells (Fig. 8B). To determine whether Erk1/2 activation in Tg(Gasp-1) myoblasts could be related to Pax7 deregulation, we treated Tg(Gasp-1) myoblasts with increasing concentrations of ERK inhibitor (Fig. 8C). As described previously, untreated Tg(Gasp-1) myoblasts showed a slight decrease of Pax7 protein levels compared to WT myoblasts and Mstn^-/- myoblasts had the highest Pax7 expression (Fig. 8C and 8D). As shown in Fig. 8C, increasing concentrations of ERK inhibitor enhanced Pax7 levels in Tg(Gasp-1) myoblasts, close to those of Mstn^-/- myoblasts for the highest ERK inhibitor concentration, confirming that Erk1/2 activation in Tg(Gasp-1) myoblasts led to reduced Pax7 expression. Finally, we sought to determine whether neutralization of myostatin could increase Pax7 expression in Tg(Gasp-1) myoblasts. Tg(Gasp-1) myoblasts were treated with increasing concentrations of anti-myostatin antibodies for 16 hours (Fig. 8D). The more myostatin was neutralized, the more Pax7 expression was improved. Eventually, we showed that the highest concentrations of ERK inhibitor or anti-myostatin antibodies led to Pax7 proteins levels close to those observed in Mstn^-/- myoblasts, suggesting that Pax7 deregulation in Tg(Gasp-1) myoblasts is related to myostatin up-regulation.

Recently, we have generated different transgenic mouse lines that ubiquitously overexpress Gasp-1, called Tg(Gasp-1) lines, in order to better understand the role of Gasp-1 during development [32]. In this paper, we investigated the cellular and molecular mechanisms deregulated by Gasp-1 overexpression to provide a detailed muscle phenotype characterization. Previous studies have shown that GASP-1 is capable of binding both myostatin and GDF-11 to inhibit their canonical pathway in vitro [29,38]. Mice lacking Gasp-1 exhibit only a slight muscle atrophy without an axial skeletal phenotype. This reduced muscle mass was observed only in old Gasp-1^-/- deficient mice (> 8 months). No significant differences between wild-type and mutant mice at 10 weeks of age was noted [39]. In contrast, Tg(Gasp-1) mice have an increased muscle mass due to hypertrophy, visible from 3 months of age. The delivery of an AAV encoding Gasp-1 leads to myofiber hypertrophy in adult mice [17]. In our model, we confirmed that mice had an increased skeletal muscle mass due to an increase in myofiber size - hypertrophy equally affecting oxidative or glycolytic muscles - without a variation in the myofiber number (Fig. 1A and 1C). As this number in mice is established soon after birth [3], we hypothesized that the absence of hyperplasia in Tg(Gasp-1) mice may be due to the lack of Gasp-1 overexpression during prenatal muscle development. Analysis of Gasp-1 expression in E9.5 and E14.5 Tg(Gasp-1) embryos showed a 20-fold overexpression compared to the wild-type (Table 2). This result was confirmed by western blotting analysis of GASP-1 at these two stages of development (Fig. 5A). To go deeper in the regulation of muscle growth by Gasp-1, we investigated the signaling pathways in satellite cell-derived myoblasts from wild-type (WT) and Tg(Gasp-1) mice.

We first characterized the Tg(Gasp-1) myoblast phenotype, i.e. the proliferation and differentiation status. As shown in Fig. 2A and 2B, the Tg(Gasp-1) myoblasts proliferate faster than the WT, as previously described for the Mstn^-/- primary myoblasts in vitro [40]. We also observed increased levels of Cdk2 and Cyclin D1, which cooperate to control the cell cycle through phosphorylation and inactivation of the retinoblastoma (Rb) protein [41]. Cyclin-dependent kinase inhibitor (CKI) p21 level was decreased in the Tg(Gasp-1) myoblasts, confirming the cell cycle progression and the increase in proliferation rate [42]. We proposed that an increase in satellite cell-derived myoblasts proliferation could be associated with an increase in satellite cell number. However, no variation was found between 12-week-old wild-type and Tg(Gasp-1) mice (Fig. 1D and 1E). This result was not surprising considering the recent analysis of satellite cell number in Mstn^-/- mice [43]. There is no effect on activation and proliferation of satellite cells in Mstn^-/- mice in adulthood [43,44], although they have an increase in the progenitor cell number during embryonic development [45]. Since expression array analysis and ELISA revealed a significant overexpression of Gasp-1 in Tg(Gasp-1) satellite cell-derived myoblasts (Fig. 4A and 4C), we could speculate that GASP-1 acts on the pool of satellite cells independently of myostatin. However, the absence of variation in number of satellite cells invalidates this hypothesis. The Tg(Gasp-1) myoblasts also differentiated faster than the wild-type cells, with an increase in both fusion index and myonuclei number (Fig. 3B and 3C). Increased protein levels of MyHC, myogenin and p21 (Fig. 3D and 3E), three major differentiation-related genes known to be down-regulated by myostatin [21], confirmed this enhanced differentiation, suggesting that Gasp-1 overexpression promotes fusion, probably by inhibition of myostatin. The increase of both proliferation rate and myonuclei number in myotubes could be related to an enhancement of satellite cell activation and proliferation in vivo. The absence of variation in number of satellite cells at 12-week-old suggests that their activation and proliferation are taking place during the first 3 postnatal weeks and are leading to a higher myonuclear accretion consistent with the myofiber hypertrophy observed in Tg(Gasp-1) mice.

A comparison of activated signaling pathways between WT and Tg(Gasp-1) myoblasts revealed a decrease of Smad3 phosphorylation rate concomitant with an increase in phospho-Akt levels, correlating with the inhibition of myostatin (Fig. 6C and 6D) [22]. We also showed a global increase in the Akt/mTOR/p70S6K protein synthesis pathway (Fig. 6). All together, these results confirmed the activation of the pro-hypertrophic signaling pathway in Tg(Gasp-1) myoblasts.

Expression array analysis between WT and Tg(Gasp-1) myoblasts showed a significant up-regulation of myostatin and a down-regulation Igf-1 (Table 1, Fig. 4B). These results were previously observed in Smad3-null mice and led to decreased muscle mass and pronounced skeletal muscle atrophy [46]. The myostatin up-regulation was also observed at the protein level in primary Tg(Gasp-1) myoblasts and serum (Fig. 4F and 4H). In contrast to the Smad3-null mice, the secretion of GASP-1 in our transgenic model allows the inhibition of myostatin, even if this latter is more secreted. Thus, the Tg(Gasp-1) mice present an opposite phenotype of the Smad3-null mice - a myofiber hypertrophy - but intermediate to Mstn^-/- one [8,32,46]. As shown in [47,48], myostatin auto-regulates its expression by a feedback loop through Smad7 dependent mechanism, where Smad7 expression is stimulated by myostatin and dramatically reduces the myostatin-induced transcription. We demonstrate for the first time that the overexpression of an extracellular myostatin inhibitor leads to the up-regulation of myostatin through its feedback loop. Consistent with these data, we found a down-regulation of Smad7.

Expression array analysis also revealed an up-regulation of Inhbb in Tg(Gasp-1) myoblasts and mice compared to the wild-type (Tables 1 and 2). Inhbb is one of the four genes encoding inhibin-β subunits [49]. While inhibins are heterodimers of a unique alpha subunit and one of the various beta subunits, activins are homodimers or heterodimers of the various beta subunit isoforms. The activin B, which is a dimer of inhibin-βb subunits, contribute to the control of muscle development since the Inhbb^-/- mice exhibit an increase of muscle mass [50]. The observed up-regulation could be involved in the intermediate phenotype of the Tg(Gasp-1) compared to Mstn^-/- mice. However, in Inhbb^-/- mice, only two muscles (pectoralis and triceps) showed a significant increase [50] while Tg(Gasp-1) mice exhibit a global increase of skeletal muscle mass [32]. Thus, the effect of Inhbb up-regulation is insubstantial compared to the Mstn up-regulation.

Surprisingly, expression array analysis revealed that Fst and Fstl3 expressions (Table 1), two other myostatin and activin inhibitors [9], were down-regulated in overexpressing Gasp-1 myoblasts, such as Tgfβ1 known as an agonist factor of myostatin [51]. Moreover, expression array analysis in Tg(Gasp-1) mice compared to the wild-type revealed the same gene expression profiles. All these results suggest the existence of a gene expression regulatory network of TGF-β superfamily members and their inhibitors.

Finally, Taqman Low Density Array (TLDA) analysis also showed in primary Tg(Gasp-1) myoblasts and mice, a down-regulation of Pax3 and Pax7, two major factors regulating muscle progenitor cell functions during embryogenesis and adulthood (Tables 1 and 2) [52]. In Tg(Gasp-1) myoblasts, we showed a significant decrease of Pax7 protein levels compared to the Mstn^-/- myoblasts and a slight decrease compared to the WT (Fig. 7). As myostatin inhibits Pax7 expression through Erk1/2 signaling pathway and its inhibition leads to increased levels of Pax7 in primary myoblasts [24], we hypothesized that the myostatin up-regulation is associated with to this deregulation of Pax7. Inhibition of myostatin using neutralizing antibodies or inactivation of Erk1/2 signaling pathway rescue the Pax7 expression in Tg(Gasp-1) myoblasts (Fig. 8). At the highest inhibitor concentration, Pax7 expression in the Tg(Gasp-1) myoblasts is quite similar to Mstn^-/- myoblasts. To exclude the action of other TGF-β on Pax7 regulation, we have determined Pax7 expression after inhibition of Erk1/2 signaling pathway in Mstn^-/- myoblasts. We found no variation, suggesting that only myostatin regulates Pax7 expression via Erk1/2 pathway. In Mstn^-/- mice, Matsakas et al. [44] and Manceau et al. [45] showed an increase in muscle progenitor Pax7⁺ cells during embryogenesis responsible for the hyperplasia observed in adult, underlying the importance of myostatin in the regulation of muscle progenitor pool . The decrease of Pax7 related to myostatin up-regulation in Tg(Gasp-1) myoblasts could be responsible for the absence of hyperplasia in Tg(Gasp-1) mice. Although, in embryonic stage, Gasp-1 overexpression is weaker than in postnatal development and in adult, this is sufficient to induce myostatin up-regulation in Tg(Gasp-1) mice (Fig. 4) [32]. We propose this myostatin up-regulation counteracts Gasp-1 effect during embryonic muscle development.

During embryogenesis, we suggest that myostatin acts preferentially through Erk1/2 signaling pathway to regulate Pax7 expression controlling the determination and the number of embryonic muscle progenitors. Thus, the absence of myostatin leads to the increase of muscle progenitors and consequently, an increase of myofiber number [45]. The Smad2/3 signaling pathway could regulate the size of myofibers during this step as Mstn^-/- mice have a slight increase of their embryonic myofibers [44]. As we showed, overexpression of Gasp-1 does not block the activation of Erk1/2 signaling pathway induced by myostatin, preventing hyperplasia in Tg(Gasp-1) mice. However, inhibition of Smad2/3 could be responsible for the hypertrophy observed in adult mice. In adulthood, myostatin could act preferentially through its canonical Smad2/3 signaling pathway to regulate cell proliferation, protein synthesis and myofiber size [23]. Absence of myostatin leads to increase of protein synthesis and therefore myofiber hypertrophy [23]. In Tg(Gasp-1) mice, we observed an intermediate hypertrophic phenotype due to a weaker inhibition of the Smad2/3 signaling pathway than in Mstn^{-/ -} mice.

This work was supported by the French National Institute of Agricultural Research and by the Limousin Regional Council. The authors thank Drs Pascal Maire and Fabrice Lalloué for their helpful scientific discussions.

The authors have declared that no competing interests exist.