Synemin acts as a regulator of signalling molecules during skeletal muscle hypertrophy

Intermediate filaments (IFs) are filamentous components of the cytoskeleton of 10-nm diameter that mechanically connect various structures of the cytoplasmic space (Lazarides, 1980). The tissue- or cell-type-specific expression of IFs makes most of them useful cell identity markers. An exception is the protein synemin. First identified as a high-molecular-mass protein associated with desmin and vimentin filaments in muscle (Granger and Lazarides, 1980), synemin expression was later detected in different tissues and cell types, including the nervous system, endothelial cells, retinal cells and hepatic stellate cells (Hirako et al., 2003; Izmiryan et al., 2006; Izmiryan et al., 2009; Izmiryan et al., 2010; Schmitt-Graeff et al., 2006; Tawk et al., 2003; Uyama et al., 2006). Three synemin isoforms differing in their C-terminal tails – α-synemin (also known as synemin H, 180 kDa), β-synemin (also known as synemin M, 150 kDa) and synemin L (41 kDa) – are produced by alternative splicing and are differentially regulated during embryonic development (Titeux et al., 2001; Xue et al., 2004). β-synemin is expressed in pluripotent embryonic stem cells, whereas both α- and β-synemin are expressed in multipotent neural stem cells in the subventricular zone of the adult brain (de Souza Martins et al., 2011). Importantly, synemin expression is upregulated in pathological conditions such as neurotrauma or Alexandre disease, a neurodegenerative disorder, and loss of its expression is associated with aggressive forms of breast cancer (Jing et al., 2005; Jing et al., 2007; Noetzel et al., 2010; Pekny et al., 2014). These features highlight a potential clinical significance; however, much remains unknown about the contributions of synemin to human disease because little is known about its functions.

Synemin is unique in that it cannot homopolymerise to form filament networks, and therefore requires an appropriate co-polymerisation partner, such as desmin, vimentin or keratin, depending on the cell type, to form filamentous structures (Bellin et al., 1999; Chourbagi et al., 2011; Hirako et al., 2003; Khanamiryan et al., 2008). Consequently, synemin filaments are unstable and delocalised in the cells of mice lacking desmin or vimentin (Carlsson et al., 2000; Izmiryan et al., 2006; Izmiryan et al., 2009; Jing et al., 2007; Xue et al., 2004). Synemin also interacts with α-actinin, plectin 1, zyxin, and three components of the dystrophin-associated protein complex, dystrophin, utrophin and α-dystrobrevin (Bellin et al., 1999; Bhosle et al., 2006; Hijikata et al., 2008; Mizuno et al., 2001; Sun et al., 2010a). α-synemin, but not β-synemin, interacts with vinculin, metavinculin and talin, suggesting that these isoforms might have different roles (Sun et al., 2008a; Sun et al., 2008b; Sun et al., 2010a). Indeed, β-synemin is believed to mediate the association of desmin IFs with myofibrillar Z-lines in neonatal cardiomyocytes, whereas α-synemin stabilises junctional complexes between cardiomyocytes (Lund et al., 2012). Further evidence demonstrates that synemin participates in focal adhesion dynamics, and is essential for cell adhesion and migration (Jing et al., 2005; Jing et al., 2007; Sun et al., 2010a). Based on these various interaction profiles and activities, synemin is proposed to act as a bridging protein between IFs and other cellular structures, such as Z-lines and focal adhesions.

Synemin has an additional role as an A-kinase-anchoring protein (AKAP) and it can modulate temporal and spatial targeting of protein kinase A (PKA) in adult and neonatal cardiac myocytes (Russell et al., 2006). More recently, synemin has been shown to play a role in Akt signalling during glioblastoma cell proliferation (Pitre et al., 2012). Thus, synemin might have two important functions: one as a bridging cytoskeletal protein between IFs and other cellular structures, and another as a regulator of signal transduction in the PKA and Akt signalling pathways.

Evidence for these roles of synemin continues to emerge, but many important questions remain. Therefore, to better define the functional roles of synemin as a cytoskeletal protein and/or as an AKAP in skeletal muscle, we generated synemin knockout (Synm^−/−) mice. Synm^−/− mice were viable and fertile, enabling the investigation of molecular and phenotypic changes in adult mice. We demonstrate that synemin is involved in membrane integrity, maximal force production and fatigue resistance of skeletal muscle. Synm^−/− mice displayed a higher hypertrophic response to mechanical overload in skeletal muscle. This increased remodelling capacity was characterised by muscle fibres switching in type to become more oxidative, lower myostatin (also known as GDF8) and atrogin (also known as FBXO32) expression, higher follistatin expression, and higher levels of phosphorylated forms of several signalling molecules implicated in muscle mass control. In addition, the absence of synemin might affect the balance between the self-renewal and differentiation programmes in muscle satellite cells. Taken together, our results show that synemin, a cytoskeletal protein, acts as a regulator of signalling molecules during muscle remodelling, and that its absence affects the behaviour of muscle satellite cells.

To analyse the role of synemin in vivo, we generated synemin-knockout (Synm^−/− ) mice using the Cre-loxP-mediated excision method. We first generated the loxP-floxed synemin mice in which exon 1, encoding 85% of the synemin rod domain, was flanked by two loxP sequences (Fig. 1A). Heterozygous synemin-knockout (Synm^+/−) mice were obtained by using a mouse line expressing Cre in the germline. Homozygous Synm^−/− mice were obtained by interbreeding heterozygous mice (Synm^+/−). Synm^−/− mice were found to be viable and fertile, and were born in the expected Mendelian ratio to wild-type (Synm^+/+), Synm^+/− and Synm^−/− mice. In adult Synm^−/− mice, the absence of synemin was confirmed at the mRNA and protein levels in skeletal muscle, heart and brain (Fig. 1B,C).

We next introduced transgenes for desmin-promoter-driven LacZ (Li et al., 1997) or vimentin-promoter-driven LacZ (Colucci-Guyon et al., 1994) into Synm^−/− mice, to gain further insight on a potential role for synemin during embryonic development of skeletal muscle and the cardiovascular system. The desmin promoter drives specific expression of LacZ in skeletal and cardiovascular muscle; the vimentin promoter drives specific expression of LacZ in mesenchymal cells. No differences were observed between Synm^+/+, Synm^+/− and Synm^−/− embryos during different developmental stages, indicating that the absence of synemin does not affect the development of the cardiovascular system and the patterning of skeletal muscle (Fig. 1D,E).

Synm^−/− mice display no overt morphological abnormalities. Further analysis of 12-week-old male mice fed with standard chow diet revealed similar biochemical and functional parameters for all genotypes, but Synm^−/− male mice had a slight lean phenotype, an increased oxygen consumption and spontaneous activity only at around 6 am (for details see supplementary material Table S1).

Morphological analysis of sections of tibialis anterior, gastrocnemius, plantaris and soleus muscles stained with haematoxylin and eosin revealed no major differences in the shape and size of the muscles fibres between 12-week-old Synm^−/− and control mice (Fig. 2). However, we quantified 3%±0.6 of muscle fibres as being centronucleated in Synm^−/− muscles as compared to control (0.4%±0.05) (mean±s.e.m., see arrows in Fig. 2D,E), indicating the occurrence of focal muscle degeneration and regeneration. Moreover, the presence of neonatal myosin heavy chain (MHCnn, also known as myosin perinatal, MYH8) in immature, newly regenerated muscle fibres (8±2 MHCnn-positive fibres in whole soleus and plantaris muscles) confirmed that regeneration was occurring (Fig. 2F).

To verify whether the absence of synemin influences the expression of other IFs, we performed quantitative real-time PCR of skeletal muscle. Synemin and seven other IFs – desmin, vimentin, nestin, syncoilin, cytokeratin 8, cytokeratin 18 and cytokeratin 19 – are expressed in skeletal muscle at various stages of differentiation. No significant differences were observed in the expression level of these seven IFs in adult plantaris muscles between Synm^−/− and control littermates (supplementary material Fig. S1A), suggesting that the absence of synemin does not affect the expression of other IF proteins in Synm^−/− muscle.

To assess whether the absence of synemin affects the subcellular localisation of its associated proteins, we performed immunohistochemistry using antibodies directed against different synemin partners. Immunostaining of muscle fibres for synemin confirmed the absence of synemin in Synm^−/− muscle, and immunostaining for Z-line-associated (α-actinin and desmin) and actin filament (phalloidin) proteins gave the expected striated pattern with a regular periodicity in both Synm^−/− and control muscle fibres (supplementary material Fig. S1B). Similar localisation patterns were observed for other synemin-binding partners, such as dystrophin and vinculin, in Synm^−/− and control mouse muscle fibres (data not shown).

Using electron microscopy, we examined the ultrastructure of the soleus, a weight-bearing muscle, from young adult (3-month-old) to aged (18-month-old) Synm^−/− mice (Fig. 3). At 3 months old, a normal ultrastructure, in terms of sarcomere organisation, Z-line alignment and distribution of mitochondria, was observed in Synm^−/− muscle (Fig. 3A–D). However, at 18 months old, irregularities were observed in the organisation of the myofibres, with abnormal sarcomeres when compared with control littermates (see arrows in Fig. 3H). Contrary to the smooth and well-defined membranes bordering control muscles (Fig. 3A,E), the sarcolemma appeared wrinkled and irregular in Synm^−/− muscle fibres at both ages (Fig. 3C,G). In order to quantify this phenomenon, we assessed the frequency of membrane invaginations at both ages. Results showed that invaginations were more frequent in 18-month-old Synm^−/− mice as compared to 3-month-old Synm^−/− mice (0.51 invaginations/µm of sarcolemma in muscle from 18-month-old versus 0.31 invaginations/µm of sarcolemma in muscle from 3-month-old Synm^−/− mice, P<0.05).

To examine the sarcolemmal integrity of the skeletal muscle fibres in Synm^−/− mice, we injected Evans Blue dye (EBD) in mice intraperitoneally and analysed its uptake into the diaphragm and plantaris muscles to mark sarcolemmal damage. Injection of EBD resulted in greater uptake of the dye in the muscle fibres of Synm^−/− mice than in those of control mice (Fig. 4). A quantification of muscle sections from Synm^−/− and control mice identified a greater than three-fold increase in the number of EBD-positive fibres in Synm^−/− muscles compared with control (Fig. 4F,M). Similar results were observed in the plantaris muscle of mechanically overloaded Synm^−/− mice (Fig. 4J–M). However, a creatine kinase activity assay of mouse serum failed to detect this localised membrane damage (105.9±23.8 units in control versus 111.2±32.6 units in Synm^−/− at the baseline; 172.8±21.3 units in control versus 158.7±35.5 units in Synm^−/− at 1 month after overload).

To investigate the effect of synemin ablation on force generation capacity, we measured the maximal isometric force production by plantaris muscles in response to nerve stimulation (Fig. 5). Plantaris muscles from Synm^−/− mice exhibited a decrease in maximal forces (−19%, P<0.05) (Fig. 5B) as compared with control mice. The fatigue resistance of plantaris muscle was also analysed by continuously stimulating the muscle and measuring the time corresponding to a decrease of 20% in initial force. Our results demonstrated that the fatigue resistance of plantaris muscles from Synm^−/− mice was decreased as compared to control mice (−20%, P<0.05) (Fig. 5C).

To analyse the role of synemin in mechanical signal transduction pathways, we examined the gains in muscle mass, muscle fibre diameter, force generation capacity and fatigue resistance in response to overload. Plantaris muscles were overloaded by surgical ablation of gastrocnemius and soleus muscles. At 1 month after overload, we found that plantaris muscle masses were markedly increased both in Synm^−/− and control mice (Fig. 5A) (P<0.05). This muscle mass gain was higher in Synm^−/− mice as compared to control mice (+238.0%±12.4 Synm^−/− versus +189.1%±9.5 control, P = 0.006, mean±s.e.m.), and was accompanied by a larger mean muscle fibre diameter (min-Ferret) in Synm^−/− mice than control mice (32.8±1.6 µm Synm^−/− versus 29.1±1.7 µm control, P = 0.045). In a similar way, the increase in maximal force in response to overload was higher in Synm^−/− mice as compared to control mice (+271.7%±18.4 mutant versus +197.4%±17.8 control, P = 0.01) (Fig. 5B). The fatigue resistance of plantaris muscle also increased in response to overload in both genotypes (Fig. 5C, P<0.05). This increase was also higher in Synm^−/− mice as compared to control mice (+176.7%±5.7 Synm^−/− versus +144.2%±8.8 control, P = 0.004). Taken together, our results, demonstrating a higher level of increase in both maximal force and fatigue resistance in response to overload in Synm^−/− mice as compared to control, suggest that synemin is involved in performance gain (maximal force and fatigue resistance) during muscle hypertrophy.

Given that the changes in muscle force generation capacity and fatigue resistance are related to MHC composition of muscle fibres, we used immunohistochemistry to compare the MHC composition of plantaris muscles of Synm^−/− and control mice without overload and after overload. The proportion of fibres expressing the two major MHC isoforms, MHC-2a (also known as myosin IIA, MYH2; indicative of oxidative fibre) and MHC-2b (also known as myosin IIB, MYH4; indicative of glycolytic fibre), was examined using a custom macro to count MHC-positive cells (Fig. 5D–F). The MHC composition in plantaris muscle did not significantly differ between Synm^−/− and control mice without overload. At 1 month after overload, the proportion of MHC-2a-expressing fibres significantly increased and the proportion of MHC-2b-expressing fibres significantly decreased in response to overload in both genotypes. The level of decrease of MHC-2b-expressing fibres in response to overload was higher in Synm^−/− mice as compared to control mice (−45.6%±7.4 in Synm^−/− versus −21.9%±10.4 in control, P = 0.002).

To elucidate the molecular mechanisms implicated in increased muscle adaptive response of Synm^−/− mice following overload, we examined the mRNA and protein expression of intracellular signalling molecules involved in muscle growth and maintenance by quantitative real-time PCR (Fig. 6) and western blotting (Fig. 7). Analyses were performed on muscle samples at 7 days after overload to analyse changes occurring during the early phase of muscle remodelling. mRNA expression of the investigated molecules was similar between Synm^−/− and control samples without overload (data not shown); after overload, Synm^−/− muscle samples exhibited significant differences from controls in the expression of four genes. Reduced mRNA expression was detected for myostatin (GDF8), an mammalian target of rapamycin (mTOR) deactivator, in Synm^−/− samples (Fig. 6A). In line with this observation, follistatin, an antagonist of myostatin, exhibited higher expression after overload in Synm^−/− mice than in control mice (Fig. 6B). Expression of atrogin, an E3-ligase-encoding gene implicated in negative control of muscle mass, was lower in Synm^−/− muscles after overload (Fig. 6C), whereas a positive regulator of muscle mass, the muscle-specific isoform of insulin growth factor-1 (mIGF-1), was increased (Fig. 6D). It should be noted that markers of inflammation were similar between Synm^−/− and control at 7 days after overload (data not shown).

Western blot analysis revealed increased expression of PKA regulatory subunit IIα (PKARIIα), Akt1, p70S6K, CREB1 and ribosomal protein S6 (RPS6) following overload in both control and Synm^−/− mice (Fig. 7; supplementary material Fig. S2). Consequently, the level of phosphorylated forms was higher in Synm^−/− versus control after overload for PKARIIα (Ser96), p70S6K (Thr389), CREB1 and ribosomal protein S6 (Fig. 7). The level of phosphorylated ERK1/2 (Thr202/Tyr204) was also increased in control mice following overload, but was unchanged in Synm^−/− mice (Fig. 7). Taken together, these results demonstrate that signalling molecules implicated in muscle hypertrophy have higher expression in Synm^−/− mice than in control mice both at the transcriptional and post-translational levels as early as 7 days after overload.

Satellite cells are resident myogenic progenitors in postnatal skeletal muscle that are involved in muscle growth and regenerative capacity. Postnatal muscle growth and satellite cell self-renewal require the transcription factor Pax7 (Mitchell et al., 2010). As shown in supplementary material Fig. S3A, synemin is present in satellite cells as shown by its presence in M-cadherin-positive cells (a surrogate marker of satellite cells) both in vivo and in vitro. To assess the behaviour of satellite cells during muscle hypertrophy in Synm^−/− mice, immunohistochemistry for Pax7 was performed on plantaris muscle cryosections from Synm^−/− and control mice without overload and 7 days after overload (Fig. 8A; supplementary material Fig. S3B). The number of Pax7-positive cells per 100 fibres was determined. As expected, the number of Pax7-positive cells significantly increased in response to overload in control mouse muscles (non-overload, 2.90±0.18; overload, 5.16±0.41; P<0.05). However, Synm^−/− satellite cells failed to respond to overload: the number of Pax7-positive satellite cells in Synm^−/− mouse muscles was unchanged (non-overload, 3.50±0.27; overload, 3.63±0.14). Notably, Synm^−/− mice have significantly more Pax7-positive satellite cells than control mice before overload (3.50±0.27 in Synm^−/− versus 2.90±0.18 in control, P<0.05).

To monitor the myogenic differentiation potential of satellite cells, we interbred Synm^−/− mice with mice expressing LacZ fused to a nuclear localisation sequence under the control of the desmin promoter, which is typically expressed in myoblasts during early myogenic differentiation (Li et al., 1997). First, we isolated skeletal muscle cells from Synm^−/− or control mouse limb muscles at 10 days after birth and cultured them to clonal density. Despite a similar number of cells per myogenic colony between Synm^−/− and control mice, the percentage of LacZ-positive cells per colony in Synm^−/− cultures (44.53%±4.46) was on average 16% higher than control (28.36%±4.38) primary skeletal muscle cell cultures (mean±s.e.m.; P<0.05) (Fig. 8B). These results suggest that, in the absence of synemin, primary muscle cells tend to commit to myogenic differentiation earlier than wild-type cells.

To better characterise this phenomenon, we analysed the expression profile of myogenic transcription factors such as Pax7, MyoD and myogenin during the time course of satellite cell differentiation (0, 24 and 72 h) using the single-fibre culture model. Based on the expression and colocalisation pattern of the above-mentioned myogenic factors, this model allows detection of satellite cell activation, differentiation and auto-renewal rates (Zammit et al., 2004). Immunostaining of freshly isolated fibres showed, as observed on cryosections, a significantly higher number of Pax7-positive (Pax7+) satellite cells per fibre in Synm^−/− mice as compared to control (4.7±0.5 in Synm^−/− versus 2.3±0.3 in control, P<0.001) (supplementary material Fig. S3C). This average number remained higher throughout the time course of analysis. At 72 h post-isolation, the average number of Pax7+ satellite cells in Synm^−/− mice was also higher as compared to control (7.8±0.6 in Synm^−/− versus 3.6±0.4 in control, P<0.001) (Fig. 8C,E). Co-immunostaining of single fibres, at 24 h post-isolation, with MyoD and Pax7 showed a similar satellite cell activation capacity as evidenced by the percentage of the cell population that was double-positive for Pax7 and MyoD in Synm^−/− and control mice (87.0%±3.0 in Synm^−/− versus 87.0%±2.9 in control). At 72 h post-isolation, MyoD and myogenin co-immunostaining revealed a higher differentiation rate in mutant mice as shown by the percentage of satellite cells positive for myogenin only in Synm^−/− mice (9.8%±2.5) when compared to control (1.3%±0.8) (P<0.001) (Fig. 8D,F). In addition, detection of myogenin-positive satellite cells at 48 h post-isolation in Synm^−/− mice suggests that this differentiation process occurs earlier in mutant mice (Fig. 8D). These results are in line with the previous observation showing a higher percentage of desmin-promoter-driven LacZ-positive cells in mutants and further enhance the notion that Synm^−/− satellite cells show a higher level of commitment to myogenic differentiation.

Synemin has been proposed to be a bridging protein between IFs and cellular membranes. Our in vivo investigations into the roles of synemin in skeletal muscle have demonstrated new functions for this protein. We showed that synemin inactivation leads to a decrease in maximal force, fatigue resistance and alteration of membrane integrity associated with an occurrence of focal muscle degeneration and regeneration. Surprisingly, Synm^−/− mice exhibited a higher hypertrophic capacity, accompanied by increased isometric maximal force and a higher fatigue resistance than control mice following overload. This higher response to hypertrophic stimulation in Synm^−/− mice appears to be directly linked to changes in gene expression or post-translational modification of signalling molecules implicated in muscle mass control. Moreover, we demonstrated that an altered behaviour of muscle satellite cells also participates in this hypertrophic response. Our results strengthen the hypothesis that synemin has dual functions, acting as both a bridging cytoskeletal protein and a regulator of signal transduction pathways, such as PKA and Akt signalling.

Synm^−/− mice are viable and fertile, and expression patterns of a LacZ reporter gene under the control of desmin or vimentin promoters show that Synm^−/− embryos develop normally. These results indicate that synemin is not essential for myogenesis (proliferation, migration, fusion of myoblasts and subsequent organisation of the muscle fibres), development of the cardiovascular system or embryonic development. This finding is consistent with reports that other IF proteins, like desmin and vimentin, are not required for embryonic development (Colucci-Guyon et al., 1994; Li et al., 1997).

Although Synm^−/− mice have similar body weight compared to age-matched control littermates, we observed mild muscle defects in Synm^−/− mice, characterised by the occurrence of focal muscle degeneration and regeneration, compromised membrane integrity and a limited sarcomeric disorganisation that increases with aging. These defects are also associated with reduced muscle performance. The myopathic phenotype, however, is less severe than that of desmin-null mice, where absence of desmin results in more degenerated muscle fibres, severe perturbations of mitochondria localisation and extensive sarcomere disorganisation (Agbulut et al., 2001; Li et al., 1996; Li et al., 1997; Milner et al., 1996). This difference might be linked to the nature of synemin, which forms obligatory heteropolymers with desmin in mature skeletal muscle fibres (Chourbagi et al., 2011; Titeux et al., 2001). Indeed, localisation of synemin in muscle tissue depends on the presence of desmin (Carlsson et al., 2000; Titeux et al., 2001; Xue et al., 2004); thus, desmin-knockout mice would also lose synemin activity in muscle. One particular feature of Synm^−/− mice is the defects in sarcolemma membranes, which appear wrinkled and have irregular invagination that is not found in desmin-knockout mice. We propose that lack of desmin renders muscle fibres more susceptible to damage during contraction, whereas synemin deficiency makes muscle less vulnerable to such damage, leaving time to develop the observed membrane defect. Interestingly, this membrane defect could be indirectly related to the absence of synemin and be the result of perturbation of its interaction with proteins that might be directly involved in the integrity of membrane structure, such as dystrophin, utrophin, α-dystrobrevin, plectin 1, zyxin, vinculin and talin (Bellin et al., 1999; Bhosle et al., 2006; Hijikata et al., 2008; Mizuno et al., 2001; Sun et al., 2008a; Sun et al., 2008b; Sun et al., 2010a). It has been reported that β-synemin mediates the association of desmin IFs with Z-lines in neonate cardiomyocytes, whereas α-synemin stabilises junctional complexes between cardiomyocytes (Lund et al., 2012), and synemin L binds to neurofilaments associated with the membrane compartment in adult mouse spinal cord (Izmiryan et al., 2006). We cannot know yet which synemin isoform is more important for membrane integrity because our Synm^−/− model is null for all forms of synemin.

Although no obvious difference is detected in muscle mass and fibre type between Synm^−/− and control mice under normal physiological conditions, Synm^−/− mice develop lower maximal force and show decreased fatigue resistance. These findings could be linked to defects such as decreased membrane integrity and muscle fibre degeneration in the absence of synemin. Surprisingly, our results indicate that the increase in maximal force and in fatigue resistance in response to overload was higher in Synm^−/− plantaris muscle, accompanied by a larger muscle mass and fibre diameter increase, compared to control muscle; thus, overload elicits a higher muscle adaptive response in the absence of synemin. Moreover, the increase in fatigue resistance in response to overload could be explained by a switch of muscle to a more oxidative phenotype, as demonstrated by a decrease in MHC-2b-positive fibres. Lower expression of atrogin and myostatin, genes involved in protein synthesis (Goodman et al., 2011), and higher expression of follistatin, an antagonist of myostatin (Sun et al., 2010b), and mIGF1, a positive regulator of muscle mass (Musarò et al., 2001), in Synm^−/− plantaris muscle following overload supports the increase in muscle mass. Interestingly, increased expression levels of total protein and phosphorylated forms of signalling molecules (Akt1, CREB1, RPS6 and PKARIIα) implicated in muscle mass control were observed in Synm^−/− plantaris muscle in response to overload. Increased expression of phosphorylated signalling molecules could explain, at least in part, the more important increase in muscle mass in Synm^−/− mice compared to control mice. As synemin is also expressed in the nervous system, we cannot exclude that some features of the Synm^−/− phenotype could be due to changes in the neural input. Tissue-specific (neural or muscle) knockout of the synemin gene will help to evaluate the specific contribution of each tissue to the observed Synm^−/− phenotype.

cAMP signalling plays a central role in hypertrophy, metabolism and regeneration of skeletal muscle (Berdeaux and Stewart, 2012). The major effector of cAMP signalling in skeletal muscle is PKA. cAMP binding to PKA regulatory subunits permits release of catalytic subunits that phosphorylate numerous target proteins, including metabolic enzymes, structural proteins, ion channels, and transcription factors such as CREBs. PKA, through CREB, controls myogenesis induced by Wnt proteins during mouse embryo development (Chen et al., 2005) and promotes muscle regeneration in vivo after acute muscle injury (Stewart et al., 2011). cAMP-PKA signalling is spatially restricted by AKAPs in skeletal myofibres (Dessauer, 2009). AKAPs organise PKA and its substrates into macromolecular complexes at specific subcellular locales. Synemin was identified as an AKAP in adult and neonatal cardiac myocytes in which it interacts with regulatory subunits of PKARIIα (Russell et al., 2006). The phosphorylated form of PKARIIα in Synm^−/− muscle is expressed more highly than in control muscle in response to overload; the phosphorylated forms of PKA target protein CREB and RPS6 are also higher in Synm^−/− muscle, indicating a possible increase in PKA activity. Using an antibody against substrates phosphorylated by PKA (against RRxS*/T*, where the asterisk indicates the phosphorylated residue), we confirmed an increase of PKA activity as evidenced by the higher level of phosphorylated PKA substrates in the muscle of Synm^−/− mice in response to overload compared to the control mice (supplementary material Fig. S2B). We speculate that synemin could restrict PKA activity in skeletal muscle. We hypothesise that the absence of synemin results in an increase in PKA activity, which, in turn, increases the phosphorylation of PKA target proteins such as CREB1 and RPS6, ultimately leading to an increase in protein synthesis, muscle hypertrophy and muscle growth. Increased expression of phospho-CREB1 and RPS6 could also result from increased activation of Akt signalling molecules in Synm^−/− mice. Indeed, synemin plays a role in Akt signalling during glioblastoma cell proliferation (Pitre et al., 2012) by helping sequester protein phosphatase type 2A away from Akt, thereby favouring Akt activation. Our results support the hypothesis that synemin acts as a regulator of signal transduction, particularly in PKA and Akt signalling. Interestingly, in response to overload, the levels of phosphorylated ERK1/2 are increased in muscle from control mice but not from Synm^−/− mice. The reason for this absence of expression change in Synm^−/− mice is not clear and requires further study.

To determine whether satellite cells play a role in increased muscle hypertrophy observed in Synm^−/− muscle in response to overload, we examined the behaviour of satellite cells. We previously reported that β-synemin is expressed in pluripotent embryonic stem cells, and both α- and β-synemin are expressed in multipotent neural stem cells in the subventricular zone of the adult brain (de Souza Martins et al., 2011). We also found that synemin protein is present in satellite cells. The number of Pax7-positive satellite cells in Synm^−/− muscle is slightly higher than in control muscle, suggesting the potential influence of synemin in satellite cell behaviour. However, during muscle overload the number of Pax7-positive satellite cells increased in control mice, but not in Synm^−/− mice. In addition, we found that satellite cells from the Synm^−/− mice had an enhanced capacity to express early differentiation markers in primary cell culture. One hypothesis is that the absence of synemin might affect the balance between self-renewal and differentiation of muscle satellite cells. The reason for this difference is not yet clear and warrants further characterisation. Based on the involvement of PKA signalling in myogenesis (Stewart et al., 2011), synemin might influence satellite cell behaviour through its role in the regulation of PKA activity.

Taken together, the results obtained in this study show that synemin, a cytoskeletal protein, acts as a regulator of signalling molecules during muscle remodelling, and that its absence affects the behaviour of muscle satellite cells. Future studies might explore the role of synemin during physiopathological conditions such as muscle dystrophy, cachexia, and aging-associated sarcopenia.

The synemin-floxed mouse was obtained with the help of Mouse Clinical Institute (Illkirch, France). This transgenic mouse harboured loxP sites at the two ends of exon 1 (Fig. 1A). Synemin-floxed male mice were bred with female Cre-expressing transgenic mice (Gary-Bobo et al., 2008) to excise exon 1 of synemin in vivo and give rise to germline-transmitted heterozygous (Synm^+/−) mice. In-vivo-mediated excision of exon 1 was verified by PCR genotyping. Heterozygous mice were interbred to generate homozygous Synm^−/− mice, which was verified by PCR genotyping. All animal studies were approved by our institutional Ethics Committee and conducted according to the French and European laws, directives, and regulations on animal care (European Commission Directive 86/609/EEC). Our animal facility is fully licensed by the French competent authorities and has animal welfare insurance.

Control and Synm^−/− mice were interbred with either desmin-LacZ-expressing (Li et al., 1997) or vimentin-LacZ-expressing (Colucci-Guyon et al., 1994) mice for two generations to obtain double-transgenic embryos (desmin-LacZ/Synm^−/−, desmin-LacZ/Synm^+/−, vimentin-LacZ/Synm^−/− or vimentin-LacZ/Synm^+/−). Embryos were collected at different time points during development and stained in toto in X-gal-containing solution (Li et al., 1997).

Animals were anesthetised (pentobarbital sodium, 60 mg/kg), and the limbs were fixed with clamps. The distal tendon of the plantaris muscle was attached to a dual-mode lever arm system that measures muscle isometric force (300C; Aurora Scientific, Ontario, Canada). Great care was taken to ensure that the blood and nerve supply remained intact during surgery. Active force measurements were performed as described previously (Agbulut et al., 2009; Hourdé et al., 2013b; Joanne et al., 2012). The sciatic nerve was crushed proximally and stimulated distally by a bipolar silver electrode using supramaximal square wave pulses of 0.1 ms duration. All isometric measurements were made at L0 (muscle length at which maximal force was obtained during the tetanos). Force productions in response to tetanic stimulation were successively recorded (pulse frequency from 25, 50 and 100 to 143 Hz, 500 ms) and at least 1 min was allowed between each contraction. Fatigue resistance was then determined after a 5-min rest period. The plantaris muscle was continuously stimulated at 50 Hz for 2 min (submaximal continuous tetanus). The time taken for initial force to fall by 20% was then determined. At the end of the experiments, the animals were killed with an overdose of pentobarbital, and the muscles were dissected and weighed, then frozen in isopentane precooled in liquid nitrogen or fixed in paraformaldehyde for further analysis. Ten animals were used for each experimental point.

Transverse frozen sections (10-µm thickness) were prepared from all of the muscles to examine general morphology using hematoxylin and eosin staining; 8-µm transverse (morphometric analysis) and longitudinal (striation pattern analysis) frozen sections were prepared for immunostaining. The sections were incubated with primary antibodies against perlecan (1∶400, rat monoclonal, Millipore, Molsheim, France), desmin (1∶100, mouse monoclonal, Dako-Cytomation, Trappes, France), synemin (1∶300, rabbit polyclonal, (Titeux et al., 2001), α-actinin (1∶100, mouse monoclonal, Sigma-Aldrich, Saint-Quentin Fallavier, France), Pax7 (1∶20, mouse monoclonal, clone P3U1, Developmental Studies Hybridoma Bank, University of Iowa), MHCnn (1∶100, rabbit polyclonal; Launay et al., 2006), MHC-2a (1∶3, mouse monoclonal, clone SC-71, Developmental Studies Hybridoma Bank, University of Iowa) or MHC-2b (1∶5, mouse monoclonal, clone BF-F3, Developmental Studies Hybridoma Bank). After washing in PBS, sections were incubated 1 h with secondary antibodies (Alexa Fluor®, Life Technologies, Saint Aubin, France) or FITC-labelled phalloidin (Sigma-Aldrich, Saint-Quentin Fallavier, France). After washing in PBS, slides were finally mounted using mowiol with DAPI. Images were captured using a motorised confocal laser-scanning microscope (LSM 700, Carl Zeiss SAS, Le Pecq, France). Morphometric analyses were made using the ImageJ software and a custom macro as described previously (Hourdé et al., 2013a; Joanne et al., 2012). The smallest diameter (min-Ferret) of all muscle fibres of the whole muscle section was measured. The pattern of striations was analysed on longitudinal muscle sections using the plot profile function of ImageJ software. For satellite cell quantification, the total number of Pax7-positive cells and myofibres per section were counted (three sections per mice; and three or four animals per genotype) and expressed as the average number of Pax7-positive satellite cells/100 fibres (Mitchell et al., 2010).

Mice were injected intraperitoneally with 10 µl/g body weight of a sterile 1% EBD solution in PBS, and, after 9 h, killed by cervical dislocation. Dissected tibialis anterior and diaphragm muscles were immediately transferred to isopentane precooled in liquid nitrogen. Cross-sections (5-µm diameter) were prepared using a microtome (Leica Microsystems, Nanterre, France) and stored at −80°C until use. To check for EBD-positive fibres, cryosections were fixed for 1 min with acetone, mounted in mowiol, and visualised under fluorescence microscopy using a single-bandpass filter (575 to 640 nm). Whole cross-sectional areas of two cryosections from three different animals per genotype were scored for EBD-positive fibres.

Creatine kinase activity in the mouse serum was measured according to the manufacturer's instructions using a EnzyChromTM Creatine kinase assay kit (ECPK-100, BioAssay system, Hayward, USA).

Electron microscopy was carried out as described previously (Joanne et al., 2013). Briefly, the calf muscles of mice were fixed in 2.5% glutaraldehyde buffered in 0.1 M cacodylate at pH 7.4. After 1 h, the soleus muscle was dissected and separated into three by a short-axis section, then fixed overnight at 4°C in the same fixative. After washing, specimens were post-fixed for 1 h with 1% osmium tetroxide solution, dehydrated and embedded in epoxy resin. Ultrathin sections (70 nm) were cut with an ultramicrotome (Leica UC6, Leica Microsystems, Nanterre, France) and stained for 15 min with 4% uranyl acetate and for 2 min with Reynolds lead citrate before observation at 80 kV with a TEM Phillips Tecnai 12 Bio Twin equipped with an Olympus Keenview CCD camera.

Total RNA was extracted from the plantaris muscle using QIAzol® lysis reagent, TissueLyser II system, and Rneasy minikit (Qiagen France SAS, Courtaboeuf, France) following the manufacturer's instructions. Extracted RNA was spectrophotometrically quantified using NanoDrop 2000 (Thermo Fisher Scientific, Saint Herblain, France). From 500 ng of extracted RNA, the first-strand cDNA was then synthesised using the Transcriptor First Strand cDNA Synthesis Kit (Roche, Meylan, France) with anchored oligo(dT)₁₈ primer and according to the manufacturer's instructions. Relative quantification PCR analysis was performed with green intercalating dye PCR technology using the LightCycler® 1536 Real-Time PCR System (Roche) on the Plateforme of Génotypage et Séquençage (Centre de recherche de l'Institut du Cerveau et de la Moelle, UPMC, Paris, France). The reaction was carried out in duplicate for each sample in a 2-µl reaction volume made up of 1 µl of LightCycler 1536 DNA Green Master (Roche) containing 500 nM each forward and reverse primers, and 1 µl of diluted (1∶25) cDNA. 1536-well plates were handled using the Bravo Automated Liquid Handling Platform (Agilent Technologies, Les Ulis, France) and sealed using PlateLoc Thermal Microplate Sealer (Agilent Technologies). The thermal profile for BrightGreen Dye qPCR was 95°C for 1 min, followed by 50 cycles at 95°C for 2 s and 60°C for 30 s. Primers sequences used in this study are available on request. The expression of hypoxanthine guanine phosphoribosyl transferase 1 (Hprt) was used as a reference transcript. At least six animals were used for each experimental.

Immunoblotting was carried out as described previously (Hourdé et al., 2013a; Joanne et al., 2012) using muscles 7 days after overload or adult muscles from overnight-fasted mice (non-overload). Muscle tissues were snap-frozen in liquid nitrogen immediately after dissection. Frozen muscles were placed into an ice-cold homogenisation buffer containing 50 mM Tris-HCl pH 7.6, 250 mM NaCl, 3 mM EDTA, 3 mM EGTA, 0.5% NP40, 2 mM dithiothreitol, 10 mM sodium orthovanadate, 10 mM NaF, 10 mM glycerophosphate and 2% of protease inhibitor cocktail (Sigma-Aldrich, Saint-Quentin Fallavier, France). Samples were minced with scissors and homogenised using plastic pestles, incubated 30 min on ice, sonicated three times for 5 s with 30-s intervals on ice, then centrifuged at 12,000 g for 30 min at 4°C. Protein concentration was measured using the Bradford method with BSA as a standard. Equal amounts of protein extracts (25 µg) were separated by SDS-PAGE before electrophoretic transfer onto a nitrocellulose membrane (GE Healthcare, Velizy-Villacoublay, France). Western blot analysis was carried out using anti-ERK1/2 (1∶1000, mouse monoclonal, Ozyme, Saint Quentin Yvelines, France), anti-phospho-ERK1/2 (Thr202/Tyr204) (1∶1000, mouse monoclonal, Ozyme), anti-70S6K (1∶1000, mouse monoclonal, Ozyme,), anti-phospho-70S6K (Thr389) (1∶1000, mouse monoclonal, Ozyme), anti-phospho-RPS6 (Ser240/244) (1∶2000, Ozyme), anti-Akt (1∶2000, Ozyme,), anti-phospho-Akt (Ser473) (1∶2000, Ozyme), anti-phospho-CREB1 (Ser133) (1∶1000, rabbit polyclonal, Ozyme), anti-phospho-PKARIIα (Ser96) (1∶1000, mouse monoclonal, Millipore, Molsheim, France) and an anti-GAPDH antibody (1∶5000, mouse monoclonal, Santa Cruz Biotechnology, Heidelberg, Germany). Proteins bound to primary antibodies were visualised with horseradish peroxidase (HRP)-conjugated secondary antibodies (Thermo-Fisher Scientific, Brebières, France) and a chemiluminescent detection system (ECL-Plus, GE Healthcare, Velizy-Villacoublay, France). Bands were quantified by densitometric software (Multi Gauge, Fujifilm). At least six animals were used for each experimental point.

Primary skeletal muscle cell cultures from the limb muscles of 10-day-old mice were prepared by enzymatic digestion, as described previously (Mitchell et al., 2010). Cells were plated on gelatin-coated dishes at a density of 1000 cells/cm² in growth medium, DMEM (Life Technologies, Saint Aubin, France) supplemented with 20% fetal bovine serum (Life Technologies, Saint Aubin, France) and 100 U/ml penicillin-streptomycin (Life Technologies, Saint Aubin, France), for 4 days. Cells were fixed and stained for LacZ and the number of blue cells (desmin-LacZ-positive) per myogenic colony was determined.

Single myofibres were isolated from the extensor digitorum longus muscles as previously described (Le Grand et al., 2009). Isolated myofibres were cultured in suspension in 6-well plates coated with horse serum to prevent fibre attachment. Fibres were incubated in plating medium consisting of 20% fetal bovine serum (Hyclone) and 1% chick embryo extract (CEE, Accurate Chemicals) in DMEM containing 2% L-glutamine, 4.5% glucose, 110 mg/ml sodium pyruvate and 100 U/ml penicillin-streptomycin. Immunochemical labelling of myofibres were performed at 0, 24, 48 and 72 h post-isolation using antibodies against Pax7 (1∶20, mouse monoclonal, clone P3U1, Developmental Studies Hybridoma Bank), myogenin (1∶10, mouse monoclonal, clone F5D, Developmental Studies Hybridoma Bank) and MyoD (1∶100, mouse monoclonal, Santa Cruz Biotechnology) as described in Zammit et al. (Zammit et al., 2004).

Groups were statistically compared using analysis of variance. If necessary, a subsequent Bonferroni post-hoc test was also performed. For groups that did not pass tests of normality and equal variance, nonparametric tests were used (Kruskal–Wallis and Wilcoxon). For all analyses, a two-tailed P-value of <0.05 was considered as statistically significant. Analyses were conducted using SAS 9.3 (Statistical Analysis System, Cary, USA). Values are presented as means±s.e.m. It should be noted that in this study, Synm^+/+ or Synm^+/− mice are used as control.

We would like to thank Alexis Canette from the imaging facility (ImagoSeine) at the Jacques Monod Institute (Paris, France) for his work in the electron microscopy study, Nathalie Vadrot and Takouhie Mgrditchian (University Paris Diderot, Paris, France) for technical assistance, Philippe Noirez (University Paris Descartes, Paris, France) for developing ImageJ software macros, and the Mouse Clinical Institute (Illkirch, France) for generating synemin-floxed mice.