New Role for Serum Response Factor in Postnatal Skeletal Muscle Growth and Regeneration via the Interleukin 4 and Insulin-Like Growth Factor 1 Pathways

Generation of mutant mice.Mice homozygous for SRF floxed alleles (Srf-flex2neo, abbreviated to S^f/S^f) and HSA-Cre transgenic mice have been described elsewhere (20, 26). These two mouse strains were backcrossed onto the C57BL/6J genetic background and then crossed to generate HSA-Cre:S^f/+ mice. The crossing of HSA-Cre:S^f/+ mice and S^f/S^f mice generated mutant HSA-Cre:S^f/S^f mice. Mice were genotyped by PCR, using DNAs extracted from tail biopsies, and Cre-mediated SRF recombination was detected in various tissues as previously described (26). In all experiments, sex- and aged-matched S^f/S^f mice were used as controls. All studies were conducted in accordance with European guidelines for the care and use of laboratory animals and were approved by the institutional animal care and use committee.

Muscle histology, immunohistochemistry, and morphometric measurements.All experiments involved comparisons of control and mutant littermates. Hind limb muscles from newborn to 8-week-old mice were removed and embedded in Cryomatrix, frozen in isopentane cooled in liquid nitrogen, and sectioned in a microtome cryostat (Leica). For assessment of tissue morphology, 5-μm-thick transverse sections were stained with hematoxylin and eosin (H&E) and examined under a light microscope. For fiber type analysis, serial sections were processed with a set of antibodies against the various myosin heavy chain (MyHC) isoforms, as previously described (4). We analyzed fiber size and determined the number of nuclei per myofiber by incubating muscle sections with mouse anti-dystrophin Dys2 antibody (Novocastra) and staining them with Hoechst stain (24). Hoechst-stained nuclei within the dystrophin-positive sarcolemma were counted in each myofiber of the entire muscle section. The fiber cross-sectional area (CSA) and the number of nuclei per myofiber were determined for three consecutive sections from five animals in each group, using Metamorph, version 2.56, software. For SRF immunodetection, paraffin-embedded longitudinal hind limb muscle sections were incubated with a 1:100 dilution of a rabbit polyclonal antibody directed against the carboxy-terminal domain of SRF (Santa Cruz Biotechnology), as described elsewhere (25).

Single myofiber isolation and primary muscle cell culture.Single myofibers with associated satellite cells were isolated from the tibialis anterior (TA) muscles of 2-month-old control and mutant mice, as described by Rosenblatt et al. (28). Primary cultures were derived from TA muscles, as described by Ohanna et al. (24).

Regeneration.Regeneration experiments were performed as described elsewhere (23). Briefly, four control, four heterozygous HSA-Cre:S^f/+, and four mutant mice were anesthetized with isoflurane (0.75 to 1.0% in oxygen) and received a single injection of cardiotoxin (0.25 μg/g of body weight) into the TA muscle. Mice were killed 9 days after injection, and the TA muscle was removed and processed for histological analysis.

Electron microscopy.Electron microscopy was performed on TA muscles from mutant and control adult mice, as previously described (30).

Western blot analysis.Western blotting was performed as previously described (26), using anti-SRF (1:200) and anti-glyceraldehyde-3-phosphate dehydrogenase (anti-GAPDH; 1:400) antibodies (Santa Cruz Biotechnology).

Northern blot analysis.Northern blots were carried out with 10 μg of total RNA from hind limb muscles, as described by Bertrand et al. (4). Membrane blots were successively hybridized at 65°C with murine cDNA probes for SRF, skeletal alpha-actin, MCK, desmin, MyoD1, myogenin, MRF4, the nuclear factor of activated T cells c2 isoform (NFATc2), follistatin, myostatin, β1-integrin, and IGF-1 labeled with [α-³²P]dCTP using the Megaprime DNA labeling system (Amersham). These double-stranded probes were synthesized by reverse transcription-PCR (RT-PCR), using previously described primers (4, 26), and were inserted into the pCR2.1 TOPO vector (Invitrogen). A human NFATc2 probe spanning 927 bp (PstI/PstI) was also used (11). Sample normalization was assessed by hybridization with an 18S ribosomal probe.

Quantitative and semiquantitative RT-PCR analysis.Total RNAs were extracted from proliferating myoblasts from control and mutant TA muscles by using Trizol reagent (Invitrogen) and were reverse transcribed with Moloney murine leukemia virus reverse transcriptase (Invitrogen) and random hexamers (Promega) to generate cDNAs. Semiquantitative PCRs were then performed, using primers for SRF, IL-4, and GAPDH. Quantitative PCR analysis was performed as already described (25).

ChIP assays.C2C12 myoblasts were cultured in Dulbecco's modified Eagle's medium supplemented with 20% fetal bovine serum and were allowed to differentiate into myotubes by decreasing the concentration of serum to 2% for 48 h. Chromatin immunoprecipitation (ChIP) assays were performed by a modified version of the protocol described by Shang et al. (34). Myotubes were cross-linked by incubation in 1% formaldehyde for 10 min at 37°C, rinsed in phosphate-buffered saline, and lysed. Nuclei were collected by centrifugation and lysed in nuclear lysis buffer. Chromatin was subjected to sonication, and the mixture was then centrifuged to remove cellular debris. Supernatants were precleared with protein G-agarose and then immunoprecipitated with 2 μg of rabbit anti-SRF antibody or 2 μg of preimmune immunoglobulin G (Santa Cruz Biotechnology) overnight at 4°C. Antibody-protein-DNA complexes were isolated by immunoprecipitation with blocked protein G-agarose. Bound DNA fragments were eluted and analyzed by PCR. As a positive PCR control, we amplified 10 ng of genomic DNA. The H4 promoter was used as a negative control for ChIP assays. PCR primer sequences are available upon request.

Statistical analysis.Results are expressed as means ± standard errors of the means (SEM). The significance of differences between means was assessed with Student's t test. P values of <0.05 were considered statistically significant.

Generation of mice with skeletal muscle-restricted disruption of SRF.Recombination of the SRF floxed allele exclusively in the skeletal muscle cell lineage was achieved by breeding S^f/S^f mice with transgenic HSA-Cre mice. The crossing of HSA-Cre:S^f/+ mice, which were viable and fertile, with S^f/S^f mice resulted in the generation of HSA-Cre:S^f/S^f mice (referred to hereafter as mutants) at the expected Mendelian frequency of 1/4, indicating an absence of fetal lethality. In all the experiments reported below, we compared mutant and S^f/S^f mice (referred to hereafter as controls).

Exon 2 of SRF was efficiently excised, as shown by PCR analysis of genomic DNAs extracted from the skeletal muscles of 8-week-old HSA-Cre:S^f/S^f mice (Fig. 1A). In contrast, no excision was observed in other mutant mouse tissues. Western blotting showed that SRF protein levels were much lower in mutant than in control skeletal muscles (Fig. 1B). Cells other than myofibers are present in muscle samples and may account for the low level of nonexcised S^f/S^f alleles and the detection of residual amounts of SRF protein in mutant muscles. Consistent with these findings, SRF-immunoreactive nuclei in the control myofibers were clearly detected, whereas mutant myofibers uniformly lacked SRF-stained nuclei, demonstrating that the recombination efficiency was high and that SRF had been depleted successfully from all mutant skeletal muscle cells (Fig. 1C). We observed only a few cells with SRF-stained nuclei in histological tissue sections, and most of these cells were endothelial and smooth muscle cells surrounding blood vessels (not shown).

• Open in new tab
• Download powerpoint

FIG. 1.

Generation of skeletal muscle-specific knockout mice. (A) Genotyping of mice by PCR analysis in the presence of a mixture of primers—SF1, SF2, and SF3—as described by Parlakian et al. (26), showing that the Cre-mediated excision of SRF exon 2 is restricted to skeletal muscle in adult mutant mice. SF1 and SF2 were used to detect floxed alleles (S^f/S^f) by amplification of a 492-bp fragment. SF1 and SF3 amplified a 310-bp DNA fragment only after exon 2 excision by Cre. (B) Evaluation of SRF protein levels by Western blot analysis of control and mutant muscle protein lysates. Much weaker signals were observed for the 67-kDa SRF band for the skeletal muscles of HSA-Cre:S^f/S^f mutant mice (Mu) than for S^f/S^f controls (Co). GAPDH served as an internal standard for loading. (C) Immunostaining of longitudinal muscle sections with the anti-SRF antibody indicated the absence of nuclear SRF protein in the myofibers of mutant mice, whereas this protein was present in the control. Arrows indicate SRF-positive nuclei. Bars, 100 μm.

HSA-Cre:S^f/S^f mice display severe growth retardation, reductions of muscle mass, and myopathy.At birth, HSA-Cre:S^f/S^f neonates were indistinguishable from their control littermates, displaying normal gross morphology. However, within 3 days of birth, mutant mice became easy to distinguish from controls before DNA genotyping, based on their smaller body size (Fig. 2A). Moreover, although they appeared healthy at birth, about 30% of the mutant mice died within the first 5 days, and only ≈51% were still alive at 6 weeks of age (Fig. 2D). A comparative study of growth from birth to adulthood revealed that both male and female surviving mutant mice developed severe growth retardation, as demonstrated by total body weight measurements (Fig. 2A and B). Heterozygous HSA-Cre:S^f/+ mice were healthy and displayed normal growth (not shown). In addition, all surviving mutant mice rapidly displayed scoliosis, a sign of muscle weakness (Fig. 2C). As shown in Fig. 2E for various hind limb muscles, adult mutant mice displayed severe and generalized decreases in muscle mass. On average, muscle weights in mutants were only about 25% those in controls. This large decrease in muscle mass in mutants is illustrated in Fig. 2F for the diaphragm and in Fig. 4A for the extensor digitorum longus (EDL), demonstrating that the entire CSA of the muscle was affected. Mutant mice may die prematurely as a result of respiratory insufficiency, due to the particularly thin diaphragms in these animals. In accordance with the results of the growth curves, mutant mice also presented generalized dwarfism affecting all organs, as illustrated for the liver. However, the decrease in muscle mass was much more severe than the decreases in size for other organs.

• Open in new tab
• Download powerpoint

FIG. 2.

Phenotypic analysis of HSA-Cre:S^f/S^f mice. (A) HSA-Cre:S^f/S^f mice were indistinguishable from controls at birth but displayed growth retardation as early as day 3, with differences in growth becoming accentuated with age. Three- and 15-day-old male mutant (Mu) and control (Co) littermates are shown. (B) Body growth curves from birth to 10 weeks for mutant and control mice. Only males were included in this study. Values are means ± SEM for 3 to 12 mice per group. (C) Muscle weakness in a 1-month-old mutant mouse, as demonstrated by visible scoliosis (arrow). (D) Life span analysis of mutant mice, showing a high level of postnatal mortality (n = 60). (E) Skeletal muscle hypotrophy in adult HSA-Cre:S^f/S^f mice, as demonstrated by comparison with the control. Results are mean wet masses of TA, gastrocnemius, and EDL muscles (n = 8) and livers (n = 3) ± SEM. ***, P < 0.001. (F) Representative images of H&E-stained diaphragms of adult control versus mutant mice. Bars, 100 μm.

Myofiber hypotrophy in HSA-Cre:S^f/S^f mice.We characterized the myopathy and nature of muscle hypotrophy due to SRF loss by carrying out histological examinations of transverse sections of various hind limb muscles from 3-day-old and adult mutant and control mice subjected to H&E staining (Fig. 3A) or to a combination of dystrophin immunostaining of the sarcolemma and Hoechst staining of the myonuclei for adults (Fig. 3B). Only small differences in myofiber size and interstitial space were observed between the muscles of mutant neonates and those of their littermate controls.

• Open in new tab
• Download powerpoint

FIG. 3.

Strongly reduced myofiber size in adult HSA-Cre:S^f/S^f mutant mice. (A) Representative H&E-stained transverse sections through hind limb muscles from 3-day-old (upper panel) and adult (lower panel) control and mutant mice. Bars, 10 μm. (B) Representative sections of TA muscles from adult control and mutant mice stained with an antibody against dystrophin (green) and with Hoechst dye (blue) or incubated with secondary antibody alone and stained with Hoechst dye. Bars, 10 μm. Data for myofiber CSA measurements are means ± SEM (n = 4 for control group and n = 6 for mutants). ***, P < 0.001. (C) Frequency histogram showing distributions of myofiber CSA in control (n = 864) and mutant (n = 1,321) TA muscles.

In contrast, in adults, as illustrated in Fig. 3A and B for the TA muscle, while control muscles contained regularly shaped myofibers, mutant muscles were characterized by myofibers with a drastically reduced size and much larger interstitial spaces, suggestive of immaturity. Such major changes in muscle structure were already visible in the longitudinal mutant muscle sections presented in Fig. 1C, in which mutant myofibers appeared much thinner and disjointed than control myofibers. Using dystrophin immunostaining, we compared the myofiber CSA for SRF mutant mice with that for control mice (Fig. 3B). The mean CSA of myofibers was ≈80% lower for mutant than for control TA muscles, and the fiber size distribution was profoundly altered, with the vast majority of fibers having CSAs of <800 μm² (Fig. 3C). Moreover, both H&E and dystrophin/Hoechst staining revealed no evidence of regenerating myofibers in adult mutant muscles, as no increase in the number of fibers with central nuclei was observed. Such a low percentage of centrally nucleated myofibers, identical to that observed in controls, was also observed in 3-day-old mutants (see Fig. S1 in the supplemental material). However, while in control muscles the percentage of centrally nucleated fibers increased between days 3 and 12, probably reflecting active recruitment of satellite cells by preexisting fibers for growth, no such increase was observed in mutants.

Altogether, these data indicate that the decrease in muscle mass in HSA-Cre:S^f/S^f mutant mice results from severe myofiber hypotrophy, which settles during the postnatal period.

Fiber type composition shift in adult HSA-Cre:S^f/S^f skeletal muscles.The maturation of skeletal muscles is accompanied by the transcriptional activation of an array of muscle-specific genes, the products of which confer unique contractile (fast/slow) and metabolic (glycolytic/oxidative) properties on the various types of myofibers, reflecting the physiological specialization of muscles (29). The CSA of myofibers depends on their contractile and energy metabolism status, with the slow/oxidative fibers that express MyHC 1 having the smallest CSA. Moreover, it has been suggested that SRF regulates MyHC 2B gene expression (1). We therefore investigated whether the distribution of fast/slow myofiber subtypes was modified in mutant muscles, using antibodies against the various MyHC isoforms. Immunostaining of serial sections of mutant and control EDLs showed that mutant EDLs contained a much smaller number of fast/glycolytic type 2B myofibers, whereas the levels of fast/oxidative type 2A myofibers and slow/oxidative type 1 myofibers were much higher in all muscles of the mutants than in those of the controls (Fig. 4A). The control soleus muscle typically contained 50 to 60% MyHC 1 fibers, whereas the mutant soleus displayed an almost complete shift to MyHC 1 fibers, with only a very few MyHC 2A fibers remaining (Fig. 4B). As illustrated with the mutant soleus, even mutant type 1 fibers were much smaller than control type 1 fibers. These results indicate a major fast-to-slow fiber type transition in mutant skeletal muscles which is associated with a switch in oxidative metabolism, as revealed by NADH staining (not shown). Thus, the observed fiber type shift accounts for only a small proportion of the overall decrease in the mutant myofiber CSA, and myofiber hypotrophy affects all mutant fiber types. In addition, embryonic MyHC continued to be expressed strongly in mutant neonate muscles (4-day-old animals) but was not detected in the corresponding control muscles (not shown). Taken together, our data highlight the importance of SRF for the acquisition of a mature fiber type-specific pattern in adult muscle.

• Open in new tab
• Download powerpoint

FIG. 4.

Adult muscles from HSA-Cre:S^f/S^f mutant mice display a switch towards a slow/oxidative phenotype. (A) Immunohistochemical detection of MyHC 2B, MyHC 2A, and MyHC 1 in 5-μm-thick transverse serial sections of control and mutant EDLs. Bars, 250 μm. (B) Adult HSA-Cre:S^f/S^f soleus muscle fibers express only MyHC 1. Note that type 1 fibers are much smaller in the mutant than in the control muscles. Bars, 100 μm.

Muscle-restricted SRF depletion results in an impaired gene expression program.We further investigated the phenotype of mutant muscles by studying the expression of well-known SRF target genes and genes whose expression varies according to muscle maturity by Northern blotting of RNAs from adult mutant and control limb muscles. Consistent with the efficient excision of the SRF gene and loss of the SRF protein, only very low levels of SRF transcripts were detected in mutant muscles (Fig. 5A). The loss of SRF in mutant muscles led to a drastic downregulation of the transcription of SKA and MCK, both of which are well-characterized CArG-dependent genes, thereby demonstrating that SRF plays a crucial role in the transcriptional regulation of these two genes in vivo in the context of the adult myofiber (Fig. 5A and C). In contrast, the expression of other described SRF target genes was either moderately increased by the loss of SRF, as for desmin (see Fig. 6C), or strongly increased, as for MyoD (Fig. 5A). Consistently, as a consequence of MyoD gene reexpression, expression of the p21 gene, which is MyoD dependent, was also upregulated in mutant myofibers. Interestingly, myogenin expression, which was barely detectable in adult control limb muscles, was also strongly upregulated in SRF-depleted muscles, whereas MRF4 expression increased only moderately (not shown). These changes to the muscle gene expression program evoke defects in the maturation of adult mutant muscles. Moreover, the decrease in SKA expression may itself contribute to the mutant myofiber growth defect by impairing the construction of new sarcomere units.

• Open in new tab
• Download powerpoint

FIG. 5.

Altered SKA gene expression and impaired sarcomere ultrastructure in HSA-Cre:S^f/S^f skeletal muscles. (A) Transcripts were detected by Northern blotting of total RNAs isolated from limb skeletal muscles of 2-month-old HSA-Cre:S^f/S^f mutant (Mu) and control (Co) mice. We loaded 10 μg of total RNA into each slot, and membranes corresponding to three blots were sequentially hybridized with the indicated probes and finally with a ribosomal 18S rRNA probe to control for loading and RNA integrity control. (B) Representative electron micrographs of longitudinal sections of TA muscles. Well-structured sarcomere units are observed in control myofibers. Major changes to the myofiber ultrastructure in the mutant TA muscle are illustrated by irregular arrangements of myofilaments, Z disk disorganization similar to nemaline rods (black arrows), accumulation of glycogen granules, and abnormal sarcoplasmic reticulum throughout the sarcomeres (black arrowhead). (C) Quantitative real-time PCR was performed on RNAs prepared from gastrocnemius muscles of adult mutant and control mice. Mean expression levels for SRF, skeletal alpha-actin (sk-actin), beta-tropomyosin (Tpm2), nebulin (Neb), and slow skeletal muscle troponin T (TnnT1) mRNAs were normalized using 18S ribosomal transcripts as the reference. Ratios for RNA levels are means ± SEM (n = 4 for controls and n = 3 for mutants). ***, P < 0.001; NS, not significant.

• Open in new tab
• Download powerpoint

FIG. 6.

HSA-Cre:S^f/S^f mutant myofibers have reduced numbers of myonuclei and impaired IL-4 expression. (A) Representative control myofiber immunostained with an antibody against dystrophin (green) and stained with Hoechst stain (blue), illustrating the number of myonuclei. The numbers of Hoechst-stained nuclei within the dystrophin-positive sarcolemma (arrows) were counted in TA muscles from control and mutant mice and expressed per 100 myofibers. Arrowheads indicate nuclei outside the myofiber. Bar, 10 μm. (B) Numbers of myonuclei per 100 myofibers and cytoplasmic CSA-to-myonucleus ratios. Data are means ± SEM (n = 4 for controls and n = 6 for mutants). ***, P < 0.001. (C) Northern blots were performed on total RNAs isolated from limb skeletal muscles of 2-month-old HSA-Cre:S^f/S^f mutant (Mu) and control (Co) mice, as described in the legend to Fig. 5. (D) Semiquantitative RT-PCR analysis of SRF and IL-4 transcripts from hind limb skeletal muscles of 12-day-old HSA-Cre:S^f/S^f mutant and control mice. Samples from two animals with each genotype are shown.

SRF-depleted myofibers display extensive sarcomere disorganization.Electron microscopy analysis of mutant TA muscle samples revealed major pleomorphic changes to the ultrastructure of mutant myofibers (Fig. 5B). Whereas age-matched control muscles displayed regularly oriented actin/myosin filaments and well-structured Z disks, mutant muscle fibers displayed sarcomere units oriented in different directions and irregular arrangements of myofibrils. In addition, Z disks were disorganized, and numerous fragments of Z disks resembling rods that are observed in nemaline myopathies (NEM) and are thought to be accumulations of sarcomeric proteins were frequently observed. The mitochondria had accumulated in foci in some myofibers. The accumulation of glycogen was evidenced by the presence of electron-dense granules throughout the mutant myofiber and was subsequently confirmed by periodic acid-Schiff staining (not shown). The accumulation of abnormal sarcoplasmic reticulum, a sign of degeneration, was observed repeatedly. Interestingly, these defects resemble the phenotype of patients suffering from NEM caused by mutations in the genes encoding thin filament proteins, such as SKA, beta-tropomyosin (Tpm2), slow skeletal muscle troponin T (TnnT1), and nebulin (Neb) (31, 37). Among these genes, SKA, Tpm2, and TnnT1 contain functional CArG boxes in their promoters. As shown in Fig. 5A, SKA expression was strongly impaired in mutant muscles. By using quantitative RT-PCR, we confirmed this result and showed that Tpm2 mRNA levels were also decreased in mutants, but to a lesser extent, while TnnT1 mRNAs were clearly increased (Fig. 5C). No change in the expression of the nebulin gene was observed. Therefore, SRF loss led to altered expression in three components of thin filaments in mutant myofibers. These data demonstrate that SRF is crucial for the establishment of sarcomere unit integrity and stability.

Reduced number of myonuclei and impairment of the postnatal IL-4 pathway in myofibers lacking SRF.Postnatal myofiber growth involves cytoplasmic growth, through the addition of new contractile filaments to the preexisting sarcomere, and the incorporation of new nuclei through myoblast recruitment and fusion with nascent myotubes. We wondered whether the postnatal myofiber growth defect in mutants could also result from the defective activation and/or recruitment of satellite cells for fusion into the preexisting SRF-depleted fibers. For this purpose, we counted the Hoechst-stained myonuclei inside the dystrophin-stained sarcolemma in transverse TA and EDL sections (Fig. 6A and B). A major defect in the cytoplasmic growth of mutant myofibers was reflected by an approximately 60% decrease in the cytoplasmic domain regulated by a single myonucleus. However, the number of myonuclei was significantly lower (about 40%) in both SRF mutant muscles than in control muscles, suggesting that defective cytoplasmic growth is not the only mechanism accounting for mutant myofiber hypotrophy. Since satellite cells were unaffected by the SRF mutation in mutant muscles (see below and Fig. 7A and B), alterations in the ability of SRF-depleted myofibers to activate and/or recruit satellite cells for postnatal growth may also play a significant role in decreasing their CSA.

• Open in new tab
• Download powerpoint

FIG. 7.

Impaired regeneration and IGF-1 expression in mutant myofibers. (A) Dissociation, proliferation, and differentiation of satellite cells from myofibers isolated from control and mutant mouse TA muscles. Mutant satellite cells recapitulate normal myogenesis in vitro. Images were taken 6, 10, and 14 days after myofiber isolation. Bars, 100 μm. (B) RT-PCR analysis of SRF expression in proliferating myoblast cultures established from adult control and mutant mouse TA muscles. (C) Representative H&E-stained transverse sections through TA muscles from control and mutant mice taken 9 days after cardiotoxin injection, showing regenerating myofibers in the control muscle, with very few centronucleated myotubes detected in the injured mutant muscle. Bars, 10 μm. (D) Northern blot analysis of IGF-1 expression, using total RNAs isolated from limb skeletal muscles of 2-month-old HSA-Cre:S^f/S^f mutant (Mu) and control (Co) mice.

We investigated mRNA levels for several factors known to be involved in the recruitment of myoblasts by preexisting myofibers. We carried out Northern blots on mRNAs isolated from adult muscles (Fig. 6C) and RT-PCR on RNAs isolated on day 12, during postnatal growth (Fig. 6D). During myogenesis, follistatin promotes myoblast recruitment by inhibiting myostatin (12). Surprisingly, myostatin expression decreased in the absence of SRF, whereas follistatin expression was strongly upregulated. The expression of β1-integrin, which is important for myoblast fusion with the preexisting fibers, was also upregulated (32). IL-4 secretion by nascent myofibers was recently implicated in myoblast recruitment during muscle growth (11). As shown in Fig. 6D, IL-4 expression was strongly downregulated in mutant muscles during postnatal muscle growth. Since IL-4 transcription in muscle has been shown to be regulated by the NFATc2 transcription factor isoform (11), we searched for possible changes in NFATc2 expression in SRF mutant myofibers. NFATc2 expression was not impaired in the absence of SRF, but instead was clearly enhanced (Fig. 6C). Thus, in the absence of SRF, IL-4 expression is strongly downregulated, suggesting that alterations in this pathway could participate in the postnatal growth defects observed in muscles lacking SRF by impairing normal myoblast recruitment.

Regeneration and IGF-1 expression are impaired in adult HSA-Cre:S^f/S^f muscle.Surprisingly, although they clearly developed myopathy and muscle weakness, adult HSA-Cre:S^f/S^f mice presented no signs of muscle regeneration, such as centrally located myonuclei. In several ways, muscle regeneration recapitulates certain steps in developmental myofiber growth. We assessed the ability of myofibers lacking SRF to regenerate by subjecting the TA muscles of adult control, HSA-Cre:S^f/+ heterozygous, and SRF-depleted mice to cardiotoxin injury. Nine days after cardiotoxin injection, new myofibers had formed in the injured control TA muscles and in heterozygous mice (not shown), as demonstrated by the presence of centrally located myonuclei, whereas the injured mutant TA muscles were clearly atrophied and reproducibly failed to undergo normal regeneration (Fig. 7C). In injured mutant muscles, only very small numbers of regeneration foci and newly formed myotubes were observed, and muscle tissue was replaced by interstitial tissue composed of inflammatory and fibroblastic cells. This failure to regenerate was not due to a cell-autonomous defect of satellite cells, as the HSA-Cre transgenic line used in this study does not allow Cre-mediated recombination in satellite cells (23), and myoblasts isolated from mutant muscles contained normal amounts of SRF (Fig. 7B). In addition, similar numbers of myogenic progenitor cells migrated from single myofibers isolated from mutant muscles and from control myofibers (Fig. 7A), indicating that the poor regeneration in mutant muscles was not linked to alterations in the satellite cell pool. In differentiation medium, myoblasts from mutant muscle underwent normal differentiation into myotubes. This result is consistent with the observation that in muscle cell cultures, the Cre recombinase, under control of the HSA promoter, becomes active only in late myotubes (23). Indeed, we failed to detect efficient SRF exon 2 excision in cultured mutant myotubes at day 14 after satellite cell plating (see Fig. S2 in the supplemental material).

Downregulation of the IL-4 pathway is not sufficient in itself to account for the lack of mutant myofiber regeneration, as IL-4^−/− myofibers do regenerate (11). IGF-1, a growth hormone known to play a key role in muscle growth and regeneration (13), has been implicated in almost all stages of muscle regeneration in adults (35). Northern blotting demonstrated that the 7.5-kb IGF-1 transcript level was strongly and reproducibly downregulated in mutant muscles lacking SRF (Fig. 7D). This downregulation of IGF-1 expression may account for the impaired regeneration observed in muscles lacking SRF.

SRF binds the promoter regions of both IL-4 and IGF-1 in vivo.We investigated whether the downregulation of IL-4 and IGF-1 expression observed in myofibers lacking SRF could result from direct transcriptional regulation via the binding of SRF to the corresponding promoter regions. Examination of 10-kb promoter regions for the mouse IL-4 and IGF-1 genes led to the identification of four putative SRF-binding sites in the IL-4 promoter region and two putative SRF-binding sites in the IGF-1 promoter region (Fig. 8A). Although these motifs were not strictly conserved at the same place in rat and human genes, several CArG-like sequences were also found in these regions. CArG-like sequences often constitute functional SRF-binding sites (19), and the most proximal IL-4 CArG-like sequence, with a 1-bp deviation (underlined) (CCATTCTTGG), was considered, as a similar sequence has been shown to be essential for MCK expression in vivo in muscle tissues (38). We investigated whether SRF bound these putative CArG motifs in the IL-4 and IGF-1 promoters in intact mouse myofibers by carrying out ChIP assays on differentiated mouse C2C12 myotubes. As expected, SRF bound specifically to one of the known consensus CArG boxes of the SKA promoter sequence, whereas no PCR signal was obtained if immunoprecipitation was performed with an irrelevant antibody (Fig. 8B). The CArG box-deficient histone H4 promoter was used as a negative control. Interestingly, three of the four putative CArG boxes within the IL-4 mouse promoter region were clearly enriched in SRF immunoprecipitates, including the nonconsensus proximal CArG-like sequence CCATTCTTGG, which is conserved at an identical location in the rat IL-4 promoter (Fig. 8B). The anti-SRF antibody gave specific enrichment of the most proximal CArG box 1 element of IGF-1 but not of the distal motif. Our ChIP assays indicate that SRF can bind the promoters of IL-4 and IGF-1 in differentiated myotubes, suggesting that these two genes are probably directly regulated by SRF.

• Open in new tab
• Download powerpoint

FIG. 8.

ChIP analysis of SRF binding to murine genomic SKA, IL-4, and IGF-1 CArG elements in vivo. (A) The sequences of the putative SRF-binding sites found in the IL-4 and IGF-1 promoters are indicated. (B) ChIP experiments were performed as described in Materials and Methods, using differentiated C2C12 myotubes. Lane 1 shows PCR amplification of 10 ng of the total input DNA for immunoprecipitation, lane 2 shows amplification with an irrelevant antibody, and lane 3 shows amplification of ChIP products with anti-SRF antibody.

Development of a myopathy characterized by myofiber hypotrophy and impaired maturation in skeletal muscles lacking SRF.All adult mutant mice developed generalized growth retardation and scoliosis linked to muscle weakness and myopathy, characterized by a decrease in total skeletal muscle mass and severe myofiber hypotrophy. However, mutant muscles did not display the characteristic pathological features of muscle necrosis/regeneration processes, such as macrophage infiltration and centrally nucleated fibers. Unexpectedly, the MyoD and myogenin genes, which are generally reexpressed in regenerating fibers and following denervation injury, were strongly expressed in adult mutant muscles. In the absence of signs of regeneration, this may reflect mutant muscle immaturity and/or changes in motor neuron transmission at the neuromuscular junction, suggesting a role for SRF in regulation of the genetic program for adult skeletal muscle differentiation. Indeed, adult skeletal muscles lacking SRF also displayed an overall increase in the number of slow/oxidative fibers, indicating a disruption of the early postnatal maturation of the various types of fiber, which is also partly dependent upon innervation (29). It was recently shown that SRF activates the fast/glycolytic fiber-specific MyHC 2B gene promoter, whereas a CArG-like element in the MyHC 2A promoter may act as a repressor (1). Activation of the MyHC 2B and MCK genes during muscle differentiation has been shown to involve chromatin remodeling and SRF recruitment (5). The absence of SRF in mutant muscles led to lower levels of MyHC 2B and MCK expression and higher levels of MyHC 2A expression. SRF may therefore play an essential role in the terminal differentiation of myofibers into the fast/glycolytic type 2B myofiber phenotype. Another, nonexclusive hypothesis is that the slow/oxidative phenotype of mutants could also be related to a physiological adaptation of muscles to the overload endogenously generated by a disproportionate body mass. NEM muscles also display such an increase in the number of slow/oxidative fibers generated by endogenous overload (14).

Different roles for SRF depending on skeletal muscle developmental stage.The absence of SRF protein in adult mutant skeletal muscles was associated with the strong downregulation of known target genes, such as SKA and MCK. In contrast, in experiments using a disruption of SRF occurring earlier in muscle development and resulting in the death of mutant neonates, Li et al. observed only mild effects on SKA transcription and no effect on MCK transcription at birth (17). These apparent discrepancies probably reflect differences in SRF function according to the skeletal muscle developmental stage. SRF seems to be particularly important for the transcription of SKA and MCK during postnatal muscle development. These observations are consistent with our previous observations concerning the cardiac alpha-actin gene. Expression of the cardiac alpha-actin gene is unaffected in embryonic hearts lacking SRF (26) but strongly decreased if SRF inactivation is triggered in the adult heart (25).

The absence of SRF in adult muscles led to an increase in mRNA levels for the desmin, troponin T, and MyoD genes, which have also been described as SRF-regulated target genes (16, 18, 39). These results reveal an unexpected complexity with regard to in vitro studies and show that the behavior of CArG box-dependent genes may depend on the cellular context and the developmental stage of the skeletal muscle tissue. Thus, the CArG box present in the distal regulatory region of the mouse MyoD gene, which is required for expression in myoblasts, is no longer necessary in the context of the differentiated myofiber.

Crucial role for SRF in various muscle growth mechanisms.Why do mice lacking SRF in their skeletal muscles display such severe postnatal myofiber growth defects? Around birth, mutant skeletal muscle mass and myofiber CSA seemed only very mildly affected, indicating that the mutant phenotype results primarily from the defective postnatal growth of skeletal myofibers. Using an earlier expressing Cre mouse model, Li et al. observed that SRF is necessary for muscle growth during late embryogenesis (17). In this study, the phenotype observed for HSA-Cre:S^f/S^f mice reflects the timing of SRF gene deletion, which mostly occurs during the postnatal period due to the use of the HSA promoter to direct Cre expression. Collectively, our results suggest that several superimposed defects may contribute to the postnatal muscle growth defect of HSA-Cre:S^f/S^f mice, as follows.

(i) SRF loss led to an important decrease in SKA expression and, to a lesser extent, Tpm2 expression, which could account for a large proportion of the postnatal cytoplasmic growth defect and the smaller size of the myonuclear domain. SKA expression is upregulated at birth, and this isoform predominates in adult skeletal muscle, in which it accounts for up to 90% of the total muscle actin present in the adult sarcomere structure. The profound disorganization of the sarcomere ultrastructure observed in SRF mutant myofibers is also consistent with the impaired stoichiometry of contractile thin filament gene expression (SKA, Tpm2, and TnnT1). Thus, it is not surprising that these ultrastructural alterations resemble those observed in human NEM, which are caused by mutations affecting the same genes (31, 37). Moreover, like our SRF mutants, SKA-null mice appear normal at birth but within 3 days display marked growth retardation, reduced muscle strength, and scoliosis. However, all SKA-null mice die early in the neonatal period (7).

(ii) SRF controls the expression of IL-4 and IGF-1, two locally secreted factors essential for muscle growth and regeneration. Skeletal myofiber atrophy due to defective cytoplasmic growth, as observed in mice deficient for the mTOR substrate S6K1, is not systematically accompanied by alterations in the number of myofiber nuclei (24). In contrast, myofibers lacking SRF displayed a decrease in the number of myonuclei similar to that observed in mice homozygous for an IL-4 null mutation (11). Indeed, IL-4 expression was strongly downregulated in mutant muscles during the period in which postnatal growth and satellite cell recruitment normally occur. The very low percentage of centrally nucleated myofibers observed in 12-day-old mutants compared to that observed in controls also argues for such defective satellite cell recruitment during postnatal growth. Our ChIP assays with myotube chromatin also revealed that SRF bound three CArG box sequences present in the IL-4 promoter in vivo, strongly suggesting that SRF regulates IL-4 transcription directly. Other transcription factors, such as NFATc2, have been shown to be involved in the control of IL-4 production (11). Unexpectedly, we found that NFATc2 expression was clearly upregulated in mutant muscle. As recently described for regulation of the smooth muscle α-actin gene by an NFATc3/SRF complex (8), we cannot exclude the possibility that NFATc2 and SRF interact cooperatively to regulate IL-4 gene expression. The increased level of NFATc2 transcripts may reflect a compensatory attempt to counterbalance mutant myofiber hypotrophy. Similarly, follistatin mRNA levels were very high in mutant muscle, whereas myostatin mRNA levels were low. This balance in the pattern of gene expression would favor myoblast recruitment and fusion (12). Similar abortive compensatory hypertrophy mechanisms have also been observed in adult SRF-depleted cardiomyocytes (25). These data identify SRF as a crucial regulator of IL-4 transcription in growing myofibers.

The failure of myofibers lacking SRF to regenerate suggests that these cells are unable to promote the efficient activation and/or recruitment of satellite cells for their growth and regeneration. Although IL-4 expression is upregulated in damaged adult muscle, downregulation of the expression of this gene in myofibers lacking SRF is not sufficient in itself to explain this phenotype because IL-4^−/− myofibers do regenerate, but with a subsequent defect in size (11). In contrast, the downregulation of the predominant IGF-1 mRNA variant expressed in skeletal muscle (35) observed in mutant myofibers may contribute to impairments in both growth and regeneration. Indeed, IGF-1 acts as an autocrine/paracrine mediator, promoting myofiber hypertrophy and regeneration through myoblast proliferation and myogenic differentiation. Targeted expression of IGF-1 in skeletal muscle has been shown to promote myofiber hypertrophy and to accelerate muscle regeneration (22, 27). We also found that SRF bound to a CArG element of the endogenous IGF-1 gene promoter in intact chromatin. Very little is known about the sequences controlling IGF-1 expression in skeletal muscle, but our data suggest an important role for SRF. Interestingly, liver-specific downregulation of IGF-1 expression has also been observed in a model of conditional disruption of SRF in the liver (M. U. Latasa, D. Couton, C. Mitchell, C. Charvet, A. Lafanechère, J. E. Guidotti, Z. Li, D. Tuil, D. Daegelen, and H. Gilgenkrantz, submitted for publication), consistent with SRF playing a direct role in the transcriptional regulation of IGF-1. Another, nonexclusive explanation is that the lack of IL-4 expression in mutant muscles also contributes to IGF-1 downregulation. Indeed, IL-4 has been shown to stimulate IGF-1 expression in macrophages, and a similar pathway may operate in muscle cells (40).

We show here that SRF may play a new role in muscle growth and regeneration by controlling local IL-4 and IGF-1 expression. Further promoter dissection experiments should be carried out to confirm our initial hypothesis that IL-4 and IGF-1 are target genes for SRF.

Our data demonstrate a central role for SRF in postnatal myofiber hypertrophy, maturation, and regeneration through control both of the building of new sarcomere units and of two crucial pathways for skeletal muscle physiology. These findings open up several lines of future research. The generation of mice with inducible skeletal muscle-restricted inactivation of the SRF gene would make it possible to elucidate further the relationships between the SRF, IL-4, and IGF-1 pathways.