Skeletal muscles in the trunk and limbs are derived from myogenic progenitors (myoblasts) that originate in the somites. Commitment of these mesodermal precursors to the myogenic cell lineage is characterized by expression of the MyoD family of myogenic regulatory factors (MRFs), namely Myf5, MyoD, Myogenin, and Myf6/MRF4 (Weintraub et al.,1991; Arnold and Braun,2000). During myogenesis, the muscles of the extremities (e.g., limb, neck, and some tongue muscle) are formed from myoblasts that migrate from the somites to their location. Upon reaching their final destination, the myoblasts stop dividing and fuse to form myotubes which differentiate into either slow-twitch or fast-twitch fibers. The fiber type patterning established during fetal muscle differentiation undergoes further adjustment after birth, regulated by mechanisms such as motoneuron stimulation and muscle usage (Buller et al.,1960; Lomo et al.,1974; Goldspink,1998; Pette and Vrbova,1999) and thyroid hormone levels (van der Linden et al.,1996; Adams et al.,1999). Because fiber types of adult skeletal muscle are highly plastic and driven by the activity of muscle, the adaptability of adult muscle is termed fiber type plasticity (Pette,2001).
Research of fiber type plasticity in mammalian skeletal muscle has in recent years identified many of the key regulatory factors. These factors include mediators that convert motoneuron stimulus to cellular signals in skeletal muscle cells. Examples include Ras (Murgia et al.,2000), calcineurin (Chin et al.,1998; Naya et al.,2000; Serrano et al.,2001), PGC1-α (Lin et al.,2002), PPARδ (Wang et al.,2004), MusTRD/GTF3 (Calvo et al.,2001; Polly et al.,2003), and MRFs (Hughes et al.,1993,1997; Voytik et al.,1993).
In contrast to the myogenic lineage commitment regulated by the MRFs, and fiber type plasticity that is driven by muscle activities, the mechanisms regulating fiber type differentiation in mammalian fetal skeletal muscle remain unknown. Calcineurin and MusTRD/GTF3, regulators of slow fiber-specific gene expression in adult muscle, have been tested for their possible roles in fetal skeletal muscle differentiation. Neither calcineurin, which stimulates slow fiber gene transcription in adult (Chin et al.,1998; Naya et al.,2000; Serrano et al.,2001), nor the MusTRD transcription factor, which represses slow fiber transcription (Calvo et al.,2001; Polly et al.,2003), affected differentiation of slow fibers in fetal skeletal muscle (Oh et al.,2005; Issa et al.,2006). These results indicate that, in mammals, the mechanisms regulating fiber type differentiation in fetal muscle and adult muscle may be different. In zebrafish embryos, the Sonic hedgehog (Shh) signaling pathway determines the slow muscle fiber lineage (Blagden et al.,1997; Norris et al.,2000; Barresi et al.,2000; Baxendale et al.,2004). However, in higher vertebrates such as chickens, Shh does not initiate slow muscle fiber differentiation (Duprez et al.,1998; Cann et al.,1999; Bren-Mattison and Olwin,2002).
In chickens and mammals, fiber type diversity emerges through two developmentally distinct waves of myogenesis, embryonic and fetal, which sequentially form the building blocks of the distinct skeletal muscle groups (Kelly and Zacks,1969; Rubinstein and Kelly,1981; Whalen et al.,1981; Stockdale,1992; Van Swearingen and Lance-Jones,1995). Myoblasts generated during these myogenic waves appear to be committed to a specific default fiber type, because the fiber types that are initially generated from embryonic myoblasts and fetal myoblasts in vivo are replicated in vitro (Stockdale,1992). In mammalian muscle, embryonic myogenesis initially generates slow muscle fibers whereas fetal myogenesis generates fast muscle fibers both in vivo (Kelly and Rubinstein,1980; Harris et al.,1989; Condon et al.,1990a; Gunning and Hardeman,1991) and in vitro (Vivarelli et al.,1988; Smith and Miller,1992; Pin and Merrifield,1993; Cho et al.,1994). In avian muscle, embryonic myogenesis generates either fast, fast/slow, or slow fibers, whereas fetal myogenesis initially generates fast fibers (Miller and Stockdale,1986a,b). Fiber type differentiation in fetal skeletal muscle has been investigated with the focus on determining the role of both intrinsic commitment of the myoblast and extrinsic signals that regulate eventual fiber types. To delineate the two, myoblast transplantation experiments have been performed by several laboratories. In chicken–quail transplantation experiments, it was shown that the embryonic myoblasts (which form primary myotubes) are committed to a specific fiber type independent of the environment (DiMario et al.,1993; Nikovits et al.,2001). A similar result was obtained by limb bud transplantation experiments showing that myoblasts migrating early (embryonic myoblasts) and late (fetal myoblasts) into limb buds have mainly slow and fast fiber fate, respectively (Van Swearingen and Lance-Jones,1995). On the other hand, in a mouse–chicken transplantation experiment, the mouse fetal myoblasts (which form secondary myotubes) transplanted in chicken limbs differentiated according to the local fiber composition (Robson and Hughes,1999). The importance of the extrinsic local environment was also demonstrated by lineage tracing of somitic myoblasts in chicken (Kardon et al.,2002). Taken together, these results suggest that intrinsic myogenic programs and the local environment both play roles in fiber type-specification of fetal skeletal muscle.
In this report, we present data indicating that the transcription factor Sox6 plays an important role in regulating the initial fiber type differentiation in fetal skeletal muscle, probably as part of intrinsic myogenic programs. Sox6 is a member of the Sox transcription factor family, a family that is evolutionarily highly conserved in vertebrates (Wegner,1999). The Sox proteins share the Sry-related HMG box domain that mediates sequence-specific DNA binding (Berta et al.,1990; Kamachi et al.,2000; Koopman et al.,1990; Wegner,1999). The Sox6 protein is expressed in multiple tissues during embryogenesis as well as in adult (Connor et al.,1995; Lefebvre et al.,1998; Hagiwara et al.,2000; Cohen-Barak et al.,2001; Murakami et al.,2001; Stolt et al.,2006); however, in adult tissues, skeletal muscle exhibits the highest expression along with testis (Hagiwara et al.,2000; Cohen-Barak et al.,2001). Despite a high level of expression, the role of Sox6 in skeletal muscle development has not been investigated at the cellular level. We have previously reported that the fetal skeletal muscle of the Sox6 mutant mouse, p100H, shows a systematic change in fiber type-specific mRNA expression: slow fiber type-specific gene expression is significantly increased, whereas fast fiber type-specific gene expression is significantly decreased (Hagiwara et al.,2005). Based on this observation, we hypothesized that Sox6 functions as a transcriptional repressor of slow fiber-specific genes in the developing fetal skeletal muscle (Hagiwara et al.,2005).
Here we report a more detailed investigation of Sox6 null-p100H fetal skeletal muscle development using immunohistochemistry and quantitative reverse transcriptase-polymerase chain reaction (RT-PCR). To elucidate how Sox6 regulates the transcription of slow fiber type-specific genes, we chose the myosin heavy chain β (MyHC-β, slow isoform) as a representative and analyzed (1) MyHC-β protein expression in the Sox6 null-p100H mutant fetal skeletal muscle, (2) mRNA expression of the MyHC-β gene in Sox6-siRNA–treated wild-type myotube cultures, (3) expression of MyHC-β promoter driven reporter constructs in wild-type and mutant myotubes in culture, and (4) chromatin immunoprecipitation (ChIP) assay of the MyHC-β promoter sequence using Sox6 antibody. The results indicate that Sox6 works as a transcriptional repressor of the MyHC-β gene in fetal skeletal muscle cells. We propose that the Sox6 transcription factor plays a critical role in the terminal differentiation of fetal skeletal muscle when fast and slow fibers first emerge during myogenesis.
MyHC-β Protein Expression in the Sox6 Null-p100H Fetal Skeletal Muscle Remains at a High Level Throughout Development
We previously reported that slow and fast fiber type-specific genes are differentially expressed between the Sox6 null-p100H mutant and wild-type muscles at the mRNA level beginning at approximately embryonic day (E) 15.5 through early postnatal stages (Hagiwara et al.,2005). The mutant skeletal muscle continues to express significantly higher levels of slow fiber-specific genes and lower levels of fast fiber-specific genes than wild-type as late as postnatal day (P) 9 (no p100H mutants survive past 2 weeks after birth). To confirm this previous observation at the protein level, we examined the expression of slow and fast MyHC isoform proteins in fetal skeletal muscle. We chose to examine MyHC protein expression because MyHC isoform (slow or fast) expression is used as the standard to classify slow and fast twitch skeletal muscle fiber types (Pette and Staron,2000). We investigated MyHC expression in fetal hindlimb muscles (E15.5 and E18.5), tongue muscles (E15.5 and E18.5), and body wall muscles (E18.5) by immunohistochemistry using MyHC isoform-specific monoclonal antibodies. Hindlimb and body wall muscles consist of both slow and fast fibers, whereas tongue muscles consist mainly of fast fibers.
As shown in Figure 1 A, as a whole, the intensities of slow and fast MyHC isoform staining at E15.5 were very similar between wild-type and mutant hindlimb muscles. By E18.5, however, the number of MyHC-β (slow) -positive fibers becomes more abundant in the mutant hindlimb muscles compared with wild-type. As depicted in Figure 1 Bb, in wild-type E18.5 hindlimb muscles, superficial muscle groups such as tibialis anterior (TA) and extensor digitorum longus (EDL), which are to become fast fibers, showed a reduced number of MyHC-β positive fibers as expected for normal skeletal muscle development at this developmental stage (Wigmore and Evans,2002). In contrast to wild-type, the myotubes in the mutant TA and EDL maintained evenly high expression of MyHC-β at E18.5 (Fig. 1 Bf). The future slow fiber muscle, soleus, showed uniform MyHC-β staining in both wild-type and the mutant (Fig. 1 Ba,e), indicating that slow muscle development is proceeding equally in wild-type and the mutant. Spatial expression of the MyHC-neo isoform, on the other hand, is similar between wild-type and the mutant (Fig. 1 Bc,g). Our MyHC-neo immunostaining data do not allow us to make quantitative assessment, although we previously observed that, at E18.5, MyHC-neo mRNA level in the mutant decreased to approximately, 20% of the wild-type level (Hagiwara et al.,2005). The reason that this sizable difference in MyHC-neo mRNA expression between wild-type and the mutant is not reflected on MyHC-neo staining may be due to a long half-life reported for the MyHC proteins in skeletal muscle (54 days in adult muscle; Papageorgopoulos et al.,2002).
Differential MyHC-β protein expression between the mutant and wild-type is even more striking in tongue muscle. The tongue is a fast muscle organ and its myogenesis is completed a few days ahead of the other skeletal muscles of the body (Dalrymple et al.,1999; Yamane et al.,2003). We, therefore, anticipated that the differential MyHC-β expression would be observed at an earlier stage in tongue muscles compared with hindlimb muscles. As expected, wild-type tongue muscles have already differentiated into fast fibers at E15.5 as indicated by the almost nonexistent expression of the MyHC-β isoform (Fig. 1 Ca), and the high expression of the neonatal fast MyHC isoform (Fig. 1 Cb). The mutant tongue muscles, on the other hand, maintain high levels of the slow MyHC-β isoform expression at E15.5 (Fig. 1 Ce) as well as at E18.5 (Fig. 1 Cg). This result indicates that the mutant tongue muscles fail to obtain the characteristic MyHC isoform expression pattern of fast fibers, that is, a high level of fast MyHC expression accompanied by the lack of slow MyHC expression. MyHC-β expression in the body wall muscles (E18.5) in the mutant reproduces the pattern observed in tongue and hindlimb muscles. In the mutant, uniform expression of slow MyHC-β in the entire body wall was observed (Fig. 1 Dc,d), in contrast to the sparse distribution of MyHC-β positive fibers in wild-type (Fig. 1 Da,b).
Combined with our previous report on the mRNA expression of fiber type-specific genes in the Sox6 null muscle, our current results of MyHC protein expression confirm the importance of Sox6 in regulation of fiber type differentiation in fetal skeletal muscle. In the absence of Sox6, the slow MyHC isoform gene (MyHC-β) fails to be silenced, thus, the Sox6 null-p100H mutant skeletal muscle maintains slow fiber characteristics beyond the normal developmental time frame.
Sox6 Null-p100H Myotubes Differentiated In Vitro Reproduce the Mutant Skeletal Muscle Phenotype Observed In Vivo
Here, we asked whether the mutant skeletal muscle phenotype observed in vivo is reproduced in the mutant myotubes formed in vitro. Because myotubes differentiated in vitro are formed independent of the in vivo environment, this experiment allows us to examine the default fiber type formed by fetal myoblasts (Vivarelli et al.,1988; Smith and Miller,1992; Pin and Merrifield,1993; Cho et al.,1994). Myoblasts were isolated from E18.5 hindlimbs of the p100H mutant and wild-type littermates and induced to form myotubes in culture. As described in the Introduction section, the myoblasts isolated from E18.5 hindlimb muscle are generated by fetal myogenesis, with the bulk of them becoming fast-twitch fibers (Kelly and Rubinstein,1980; Harris et al.,1989; Condon et al.,1990a; Gunning and Hardeman,1991). It has been shown that embryonic myoblasts (∼E11–E15) and fetal myoblasts (∼E15 to birth), when differentiated into myotubes in vitro, reproduce their initial in vivo MyHC isoform expression patterns. In culture, embryonic myoblasts form myotubes expressing MyHC-β (slow), but not neonatal fast MyHC (MyHC-neo), whereas fetal myoblasts form myotubes expressing MyHC-neo, but not MyHC-β (Smith and Miller,1992; Stockdale,1992; Pin and Merrifield,1993; Torgan and Daniels,2001).
In the wild-type myotube cultures, MyHC-β staining stayed negative (Fig. 2 Ac,i). This observation corroborates reports from other laboratories showing that wild-type fetal myoblasts do not produce myotubes expressing MyHC-β (slow) in vitro even after being kept in differentiation medium for a week (Pin and Merrifield,1993; Torgan and Daniels,2001). On the other hand, the p100H mutant myotube cultures expressed the MyHC-β protein as shown in Figure 2 Af,l. The MyHC-β protein in the mutant myotubes is detected starting at 48 hr in differentiation medium (DM) and becomes more markedly expressed in later hours (72 and 96 hr, Fig. 2 Af,l), likely reflecting an increase of MyHC-β mRNA expression in the mutant myotubes in culture (Fig. 3 A). Because myotubes formed in vitro differentiate in the absence of innervation and other environmental in vivo factors, our result suggests that the Sox6 null fetal myoblasts are committed to express the slow MyHC.
Next, we investigated whether calcineurin, a calcium-calmodulin–dependent phosphatase, is required for this high level of MyHC-β expression in Sox6 null-p100H myotubes. Calcineurin is known to be required for increased transcription of slow fiber-specific genes in adult skeletal muscle (Chin et al.,1998; Naya et al.,2000; Serrano et al.,2001) and the inhibition of calcineurin activity results in decreased expression of slow fiber-specific genes in adult muscle (Chin et al.,1998; Dunn et al.,1999). Because the identity of the signaling molecules regulating fiber type-specific gene expression in fetal skeletal muscle is not yet known, we decided to test the effect of calcineurin. The Sox6 null-p100H fetal mutant myoblasts were induced to differentiate in the presence of cyclosporin A, a calcineurin inhibitor, and MyHC-β protein expression was examined. As shown in Figure 2 B, MyHC-β staining of the mutant myotubes was not affected by increasing concentration of cyclosporin A. Although the fact that cyclosporin A also affects a calcineurin-independent pathway (Lo Russo et al.,1997) necessitates further experiments using calcineurin-specific inhibitors such as CAIN (e.g., Lai et al.,1998), our result agrees with the previous report of Oh et al. (2005), showing that calcineurin is not necessary for slow isoform gene expression in fetal skeletal muscle.
Temporal Expression of Fiber Type-Specific Genes in the Sox6 Null-p100HMyotubes In Vitro Replicates Developmental Expression of Fiber Type-Specific Genes in the Mutant Skeletal Muscle In Vivo
To investigate the temporal expression of fiber type-specific genes in differentiating p100H mutant myotubes in culture, we first performed a quantitative PCR assay of the three MyHC isoform genes: MyHC-β, embryonic MyHC (MyHC-emb), and neonatal fast MyHC (MyHC-neo). During the wild-type skeletal muscle development, MyHC-emb and MyHC-β are the first MyHC isoforms to be expressed, followed by MyHC-neo (Condon et al.,1990a; Lyons et al., 1990). Expression of MyHC-β or MyHC-neo in the myotube is indicative of the future fiber type, slow or fast, respectively, whereas expression of MyHC-emb is neutral to the future fiber type (Gunning and Hardeman,1991). Temporal mRNA expression of the three MyHC isoforms was compared between the p100H mutant and wild-type myotube cultures. In Figure 3 A, the relative mRNA expression levels (mutant to wild-type) of MyHC-β, MyHC-emb, and MyHC-neo are presented. After 48 hr in DM, MyHC-β mRNA expression in the mutant myotubes was significantly higher than in the wild-type, replicating its in vivo expression of the mutant fetal skeletal muscle (Hagiwara et al.,2005). As expected, mRNA expression of MyHC-emb, which is neutral to fiber types, was equivalent between the mutant and wild-type. Of interest, we did not observe differential expression of fast MyHC-neo mRNA between the mutant and wild-type myotubes in vitro. Because MyHC-neo mRNA expression in the mutant fetal muscle in vivo shows a significant decline compared with wild-type (Hagiwara et al.,2005), the current result suggests that additional in vivo factors missing in the in vitro condition may be required for precise developmental regulation of MyHC-neo expression.
In addition to MyHC isoforms, we also examined temporal expression of troponin I (TnI) slow and fast in the mutant and wild-type myotubes in vitro. In wild-type in vivo, TnI fast expression becomes greater than TnI slow expression as the fetal muscle matures, but this does not happen in the mutant (Hagiwara et al.,2005). We wanted to test whether this finding could be reproduced in myotubes differentiated in vitro. As shown in Figure 3 B, at the beginning of culture (0 hr) the TnI slow/ TnI fast expression ratio is comparable between the mutant and wild-type cultures. As differentiation of myotubes progresses, however, a clear difference emerged. In wild-type myotubes, the expression of TnI fast increases as indicated by decreasing TnI slow/TnI fast ratios. On the other hand, the mutant myotubes maintain higher levels of TnI slow expression over TnI fast, sustaining their “slowness” (Fig. 3 B). These data indicate that the lack of a functional Sox6 gene in the mutant myotubes results in sustained high levels of the slow fiber isoform genes, MyHC-β and TnI slow, compared with wild-type.
The results summarized in Figure 3 A,B demonstrate that in the absence of Sox6, mRNA expression of the two slow fiber type-specific genes remains high in the mutant myotubes. This observation led us to hypothesize that Sox6 functions as a transcriptional repressor of slow fiber-specific genes. To test this hypothesis, we first examined temporal expression of Sox6 mRNA in wild-type myotube cultures using TaqMan real-time PCR. According to our hypothesis, Sox6 expression is expected to be increased during wild-type myotube differentiation to suppress transcription of the slow fiber-specific genes. As shown in Figure 3 C, Sox6 mRNA expression in the wild-type myotubes sharply increased with time and was sustained at high levels as late as 120 hr in DM. This result, albeit indirectly, supports our hypothesis that Sox6 functions as a negative regulator of slow fiber-specific genes, thus, its absence in the mutant myotubes allows their continuing expression.
Knockdown of Sox6 mRNA Expression Results in Increased Expression of the MyHC-β Gene
Based on the observations described above, (1) the Sox6 null fetal myotubes express greater levels of MyHC-β and TnI slow than wild-type both in vivo and in vitro, and (2) Sox6 is highly expressed in wild-type fetal myotube cultures where no appreciable MyHC-β expression is detected, we hypothesized that Sox6 functions as a transcriptional repressor of the slow fiber type-specific genes in fetal skeletal muscle. To test this hypothesis directly, we applied RNA interference (RNAi) to reduce Sox6 expression specifically in wild-type myotubes. Three Sox6-siRNA constructs expressing one of the short sequences (Fig. 4 A) were generated using the mU6pro siRNA expression vector (Yu et al.,2002). Because expressing a single Sox6-siRNA only gave moderate levels of reduction in Sox6 mRNA expression (∼50% reduction in Sox6 mRNA level, unpublished data), we co-transfected two or three Sox6-siRNA constructs to obtain more efficient Sox6 mRNA reduction in wild-type myotubes. As shown in Figure 4 B, combined expression of Sox6-siRNA constructs successfully reduced the Sox6 mRNA level to as low as ∼20% of untreated wild-type cultures. To test the effect of reduced Sox6 expression in wild-type myotubes, MyHC-β mRNA expression level was also examined. As shown in Figure 4 C, in the Sox6-siRNA–treated wild-type myotubes, MyHC-β mRNA levels increased two- to threefold. This result suggests that knockdown of Sox6 expression relaxes transcriptional suppression of the MyHC-β gene, supporting our hypothesis that Sox6 functions as a transcriptional suppressor of MyHC-β, a slow fiber type-specific gene, in fetal skeletal muscle.
MyHC-β 5′ Upstream Sequence Contains a Negative cis-element Regulated by Sox6
To investigate how transcription of the MyHC-β gene is regulated by Sox6, we next analyzed the rat MyHC-β promoter sequence. The 3.5-kb rat MyHC-β 5′-upstream region has been reported to contain proximal cis-regulatory elements (the 350-bp immediate upstream region; Thompson et al.,1991; Shimizu et al.,1992; Huey et al.,2002,2003) and a distal enhancer (between −3.5 kb and −2.5 kb) that are necessary for the optimal MyHC-β expression in skeletal muscle (Giger et al.,2000). Based on these reports, we tested whether the sequence(s) responsible for suppression of MyHC-β in wild-type myotubes exists within this 3.5-kb upstream sequence. To test this, the Sox6 null-p100H mutant and wild-type fetal myoblasts were transfected with the MyHC-β promoter driven luciferase reporter constructs and then induced to form myotubes. As summarized in Figure 5 A, the full-length MyHC-β luciferase construct (3,500-bp 5′-upstream sequence) showed a considerably higher luciferase activity in the mutant myotubes compared with wild-type myotubes (∼ninefold), indicating that a Sox6-regulated negative sequence exists in this 3.5-kb 5′-upstream region. We next conducted deletion analysis postulating that eliminating a putative negative element would abolish differential luciferase expression between the mutant and wild-type myotubes. The deletion constructs −408 and −215 still expressed significantly higher luciferase activities in the mutant myotubes (five- to sixfold higher than wild-type). However, when the upstream sequence between −215 and −171 was removed, the difference in luciferase activities between the mutant and wild-type disappeared (fold difference = 1.4 ± 0.42; average ± se, Fig. 5 A). The DNA sequence between −215 and −171 contains the βe3 element, one of three previously identified muscle nuclear protein-binding sites in the proximal MyHC-β regulatory region (Thompson et al.,1991). As summarized in the diagram in Figure 5 A, the βe3 element contains a canonical Sox binding sequence, 5′-(A/T)(A/T)CAA(A/T)G-3′, a suitable candidate for a Sox6-regulated negative cis-element. The βe3 element also contains a MCAT enhancer element, CATACCA, which has been shown to be bound by the transcription enhancer-1 (TEF-1) family of proteins (Kariya et al.,1993,1994). As summarized in Figure 5 B, both the Sox and MCAT elements in βe3 are evolutionarily well conserved; especially the Sox consensus sequence, which is 100% conserved among rats, mice, and humans (Fig. 5 B). This sequence conservation suggests their functional importance in transcriptional regulation of the MyHC-β gene.
Because the region between −215 and −171 of the MyHC-β promoter sequence contains both positive (MCAT) and putative negative (Sox) cis-elements, we decided to delineate the effect of these two sequences. Four luciferase reporter constructs (Fig. 5 C) were tested: (1) WT −408 MyHC-β luciferase construct; (2) mMCAT, −408 with three nucleotide substitutions in the MCAT sequence (ACC to CGG); (3) mSox, −408 with three nucleotide substitutions in the Sox sequence (ACA to CAG); and (4) mMCAT/mSox, −408 with both the MCAT and Sox mutations. As summarized in Figure 5 D, disrupting the MCAT element in the MyHC-β βe3 element resulted in a major reduction of luciferase activity in both the p100H mutant and wild-type myotubes. Therefore, this MCAT element functions as a positive cis-element in both myotubes. On the other hand, disruption of the Sox consensus sequence led to an increase of luciferase activity in wild-type myotubes, but no appreciable change in the mutant myotubes. These observations support our hypothesis that Sox6 functions as a transcriptional repressor and the βe3 Sox consensus sequence is a likely target for the Sox6 protein. Interestingly, disruption of the MCAT and Sox consensus sequences abolished transcription in both the mutant and wild-type myotubes, which may indicate that optimum expression of the MyHC-β gene requires the sequence encompassing both the MCAT and Sox elements.
We next conducted ChIP experiments to test whether the βe3 Sox consensus sequence is bound by the Sox6 protein. C2C12 myoblast cells were transfected with 408-WT or 408-mSox along with a Sox6 protein expression vector, and the rat MyHC-β promoter DNA bound by the Sox6 protein was precipitated using Sox6 antibody. As summarized in Figure 6, in the presence of Sox6 antibody, the DNA fragment containing the βe3 Sox sequence was enriched, suggesting that the Sox6 protein is binding this region. On the other hand, no enrichment of the 408-mSox MyHC-β promoter (with the mutated Sox sequence) was observed, indicating that the Sox consensus sequence is required for the binding of the Sox6 protein. This result combined with the results summarized in Figure 5 support our hypothesis that Sox6 functions as a transcriptional repressor of the MyHC-β gene through the Sox consensus sequence in the MyHC-β βe3 element.
In this study, we have shown that the fetal skeletal muscle of the Sox6 null-p100H mutant maintains the characteristics of slow fibers both in vivo and in vitro. The slow MyHC isoform, MyHC-β, continues to be uniformly expressed in the mutant fetal skeletal muscle in vivo (Fig. 1). This finding is in stark contrast to wild-type fetal muscle development, where the majority of myotubes differentiate into fast isoform-expressing fibers and a limited number of deep muscle fibers remain slow (Narusawa et al.,1987; Sutherland et al.,1991; Calvo et al.,2001). In the mutant myotubes differentiated from fetal myoblasts in vitro, both MyHC-β and TnI slow remain highly expressed (Figs. 2 A, 3 A,B). This finding is in contrast to wild-type myotubes that differentiate into fast fibers in vitro, as indicated by the lack of MyHC-β expression (Fig. 2 A) and a rapid increase in TnI fast expression (Fig. 3 B). These results suggest that the absence of a functional Sox6 protein leads to sustained expression of slow isoform genes in the mutant fetal muscles. Based on these observations, we propose that Sox6 is required for normal fiber type differentiation in developing fetal skeletal muscle. We have also presented data indicating that Sox6 negatively regulates the transcription of the MyHC-β gene (Fig. 4 B,C). We, therefore, propose that Sox6 functions as a transcriptional repressor of slow fiber isoform genes in the future fast fibers being formed during late fetal myogenesis.
Myogenesis begins in the somites where the mesodermal precursors commit to become myoblasts. After this initial commitment, the subsequent migration and differentiation of myoblasts are regulated by the MRFs and other factors expressed in myogenic cells and factors in the surrounding environment (Buckingham,2001). Once myoblasts reach their destination, myotubes form and terminally differentiate into fast or slow fibers, establishing a basic pattern of fast and slow muscle groups in fetal skeletal muscle (Wigmore and Dunglison,1998). In the adult, the fetal fiber patterning is further adjusted in response to muscle activities and signals from motoneurons (fiber type plasticity, reviewed in Pette,2001). Among these myogenic differentiation stages, we believe that Sox6 plays a vital role in the terminal differentiation of fetal skeletal muscle when the basic pattern of fast and slow fibers initially forms, because in the absence of Sox6, all fetal myotubes maintain expression of slow fiber isoform genes (Figs. 1 B–D, 2A).
It remains unknown how fast and slow fibers are originally delineated in mammalian fetal skeletal muscle; what is known is that fetal skeletal muscle is assembled by waves of sequentially emerging myoblasts and the myotubes subsequently generated by them. In mice, embryonic myoblasts fuse to give rise to primary myotubes at ∼E11 to E15, followed by fetal myoblasts, which give rise to secondary myotubes (Kelly and Zacks,1969; Wigmore and Dunglison, 1998; Stockdale,1992). There is a good correlation between the types of myotubes (primary or secondary) generated by embryonic or fetal myoblasts and their initial fiber types; the majority of primary myotubes are initially slow (MyHC-β–positive), and secondary myotubes are initially fast (MyHC-β–negative and MyHC-neo–positive; Narusawa et al.,1987; Condon et al.,1990a; Gunning and Hardeman,1991; Cho et al.,1994). These initial fiber types of primary and secondary myotubes are reproduced in vitro (Vivarelli et al.,1988; Smith and Miller,1992; Pin and Merrifield,1993; Cho et al.,1994), which most likely is capturing the default phenotypes of primary and secondary myotubes. These default fiber types suggest the existence of regulatory mechanisms intrinsic to muscle cells. During in vivo fetal muscle development, however, the local environment (extrinsic to muscle cells) plays a deciding role in modulating the final fiber types eventually displayed by the myotubes (Dunglison et al.,1999; Hughes,1999; Robson and Hughes,1999). Innervation is undoubtedly an important part of the extrinsic regulatory mechanism, which has been shown both in vivo (McLennan,1994; Lefeuvre et al.,1996; Washabaugh et al.,1998), as well as in vitro using co-cultures of the nerve and myotubes (Jiang et al.,2004; Jordan et al.,2005). In addition, maturation and survival of both primary and secondary myotubes are largely dependent on innervation (Hughes and Ontell,1992; Ashby et al.,1993a,b; Wilson and Harris,1993). However, initial induction of fiber type-specific MyHC isoform expression seems to operate in the absence of innervation (Butler et al.,1982; Condon et al.,1990b; Robson and Hughes,1999). The identity of innervation-independent signals in mammalian fetal muscle is yet unknown. In chickens, it has been shown that the balance in expression of the Wnt family of proteins regulates the pattern of fast and slow fibers in the limb (Anakwe et al.,2003). The role of Wnt signaling in fiber type differentiation of mammalian fetal skeletal muscle remains unknown. Although the innervation-dependent extrinsic signals that determine fetal muscle fiber types have started to be uncovered (Jiang et al.,2004; Jordan et al.,2003,2005), the intrinsic and innervation-independent extrinsic factors that regulate initial fiber types of fetal muscle have not been identified.
Based on the fiber phenotype of Sox6 mutant fetal skeletal muscle observed both in vivo and in vitro, we propose that Sox6 plays a part in the intrinsic regulation of fiber type differentiation. Specifically, we postulate that the Sox6 protein functions as a transcriptional repressor of slow isoform genes in the future fast fibers. To support this hypothesis, we have shown that a knockdown of Sox6 expression by RNAi in wild-type secondary myotubes in vitro results in transcriptional up-regulation of the slow MyHC-β isoform gene (Fig. 4 B). Further supporting this hypothesis are the data obtained by transfection of the MyHC-β luciferase reporter constructs, indicating that Sox6 likely represses transcription of the MyHC-β gene through a specific cis-element in the MyHC-β promoter (Fig. 5). To support this result, we have also demonstrated the binding of the Sox6 protein to the Sox consensus sequence in the MyHC-β promoter using a ChIP assay (Fig. 6). To date, the genes reported to be repressed by Sox6 in a wide variety of tissues include the insulin II (Iguchi et al.,2005), cyclin D1 (Iguchi et al.,2007), fgf-3 (Murakami et al.,2001), epsilon globin (Yi et al.,2006), and myelin genes (Stolt et al.,2006). Our current report adds MyHC-β to this list as a muscle specific gene repressed by Sox6.
Another possible explanation for the sustained slow fiber isoform gene expression in the Sox6 null skeletal muscle is that the secondary myotubes (initially fast MyHC isoform-positive) fail to form in the Sox6 mutant fetal muscle. As a result, reduced expression of the fast isoform gene is observed, by default, in the mutant. We think this unlikely because (1) comparable numbers of secondary myotubes were observed between wild-type and the mutant limb skeletal muscle in vivo (Hagiwara et al.,2005), and (2) the mutant myoblasts isolated from E18.5 limb muscle (the fetal myoblast stage) formed long multinucleated myotubes in vitro (Fig. 2 A), which is the typical morphology of secondary myotubes formed by fetal myoblast fusion (Pin and Merrifield,1993; Robson and Hughes,1999). These observations make it unlikely that the high levels of slow isoform gene expression in the mutant muscle are a result of the failure of the mutant secondary myotube formation. We, therefore, favor the hypothesis of Sox6 functioning as a repressor of slow fiber isoform genes in the future fast muscle fibers.
Sox6 has been shown to suppress transcription of the insulin II gene in pancreatic cells (Iguchi et al.,2005). The authors suggest that the Sox6 protein functions as a repressor by blocking an activator on the promoter to initiate transcription. Similarly, it is reasonable to postulate that Sox6 suppresses MyHC-β transcription by interfering with an activator that binds to the MCAT enhancer element. As shown in Figure 5 B, the MCAT element, to which the TEF-1 protein has been shown to bind (Kariya et al.,1993,1994), and the Sox consensus sequence we propose as a negative element are located next to each other. Because of its proximity, it is possible that Sox6 represses MyHC-β transcription by blocking TEF-1 from binding to the DNA or by inactivating TEF-1 through a physical interaction. We are testing these possibilities by electrophoretic mobility shift assays and co-immunoprecipitation assays.
Because differentiation of fiber types in fetal skeletal muscle appears to be regulated by both the intrinsic genetic programs within myoblasts and the extrinsic local environment (DiMario and Stockdale,1997; Robson and Hughes,1999; Pin et al.,2002), we want to determine whether Sox6 is involved in intrinsic myogenic commitment or/and in regulating the extrinsic signals. To begin to answer this question, we have obtained Sox6 conditional knockout mice (Dumitriu et al.,2006a) to inactivate the Sox6 gene in a myogenic cell-specific manner. This strategy will allow us to define the temporal and spatial requirement of Sox6 for the fiber type specification in fetal muscle. Another issue is how Sox6 suppresses expression of slow fiber-specific genes only in fast fibers, but not in slow fibers, even though the Sox6 protein is expressed in both fast and slow muscles (Hagiwara et al.,2005). The Six1 protein, a transcriptional activator of fast fiber-specific genes, has been shown to be specifically localized in the nuclei of fast fibers, thus, activating fast fiber-specific genes in a fiber type-specific manner (Grifone et al.,2004). We are currently investigating the cellular localization of the Sox6 protein in both fetal and adult skeletal muscle to see if Sox6 exerts its effect by means of nuclear localization.
In summary, we have presented data indicating that Sox6 is necessary for fetal skeletal muscle to terminally differentiate into a fast or slow fiber. The proposed function of Sox6 in myogenic development, that Sox6 plays an important role in terminal differentiation, has also been reported in other cell types. These types include chondrocytes (Smits et al.,2001; Lefebvre and Smits,2005), erythrocytes (Dumitriu et al.,2006b; Yi et al., 2006), and oligodendrocytes (Stolt et al.,2006). Because of its broad expression during embryogenesis, Sox6 may regulate the terminal differentiation of many other tissues. In skeletal muscle development, because of the systematic change in fiber type-specific gene expression, the Sox6 mutant serves as an ideal model to investigate the yet unknown molecular mechanisms of the terminal differentiation of fetal skeletal muscle.
E15.5 and E18.5 embryos were obtained from timed matings of heterozygous (p100H/+) intercrosses, designated E0.5 on the day a vaginal plug was detected. Wild-type and p100H homozygotes were collected from the same litters. Embryonic tissues were dehydrated and embedded in paraffin and sectioned at 10 μm. Sections were then re-hydrated and treated with the low concentration of trypsin solution for 5 min at room temperature (0.002% trypsin with 0.1% CaCl2 in 20 mM Tris pH 8.0). Subsequently, sections were incubated in 0.3% H2O2 in water for 5 min to quench endogenous peroxidase activity. Blocking and incubation with primary and secondary antibodies were performed following the protocol of the Vector M.O.M. (mouse on mouse) immunodetection kit (Vector Laboratories). Antibody staining was visualized using Vectastain Elite ABC kit (Vector Laboratories) and diaminobenzidine (Sigma) as a substrate for horseradish peroxidase. Primary monoclonal antibodies used for myosin heavy chain (MyHC) isoforms were MF20 (Developmental Studies Hybridoma Bank, the University of Iowa), MY32 (Sigma), and NOQ7.54D (Sigma). The dilutions used were 1:20, 1:2,000, and 1:2,000, respectively. When myotubes differentiated in vitro were used for immunostaining, cultures were fixed in methanol containing 0.6% hydrogen peroxide for 20 min on ice and processed for immunostaining.
Primary Myoblast Cultures
All cell cultures were performed at 37°C with 5% CO2 in a humidified condition. Fetal myoblasts from E18.5 mouse limbs were collected using a protocol modified from the previously published methods (Springer et al.,1997; Machida et al.,2004). Briefly, E18.5 embryonic limbs were collected in phosphate buffered saline (PBS) on ice, skinned, and muscle tissues were pooled. Muscle tissues were minced and digested in collagenase/dispase solution (2.4 U/ml dispase II, 10 mg/ml collagenase D, Roche Applied Science) for 20 min at 37°C. Cell suspensions were filtered through 70 μm nylon mesh (BD Sciences). Cells were collected by centrifugation and suspended in DMEM with 10% horse serum (HS), 1 U/ml penicillin/streptomycin (Invitrogen). To enrich for myoblasts, cells were plated on uncoated plastic culture dishes and unattached cells were collected. This step was repeated twice: first, cell suspension was incubated in an uncoated dish for 30 min, and second, cell suspension was incubated for 17 hr before unattached cells were collected. Cells enriched for myoblasts were cultured on collagen-coated dishes in F10 medium with 20% fetal bovine serum (FBS), 1 U/ml penicillin/streptomycin, and 2.5 ng/ml recombinant human basic FGF (bFGF; Chemicon International). The myoblast cultures were visually inspected for fibroblasts with flat morphology, and cultures with no clear fibroblast contamination were used for experiments. Enriched myoblasts were routinely grown in growth medium (GM), which is DMEM/F10 (1:1) with 20% FBS, 1U/ml penicillin/streptomycin, and 2.5 ng/ml bFGF. To induce myotube formation, myoblast cultures were rinsed twice with PBS, and incubated in DM, which is DMEM with 2% HS. The medium was changed every 24 hr.
For cyclosporin A treatment of the p100H mutant myotube cultures, myoblasts were plated and incubated in GM for 24 hr, and then switched to DM in the presence (200 nM to 1 μM) or in the absence (vehicle only, ethanol) of cyclosporin A (Sigma). Cultures were kept in DM for 72 hr in the presence of cyclosporin A, then fixed and processed for immunostaining.
Total RNA was prepared from myotubes using Trizol (Invitrogen). cDNA was synthesized using oligo-d(T) primer and SuperScript II (Invitrogen). TaqMan real-time PCR was performed using Assays-on-Demands gene expression probes and TaqMan universal PCR mix, following the manufacturer's protocols (Applied Biosystems). Reactions were monitored by the ABI Prism 7900HT. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) expression was used for normalization of cell type-specific genes among samples. A few experiments were performed using β-actin as a standard and the same results were obtained when the experiments were repeated using GAPDH as a standard. Therefore, data presented here were analyzed using GAPDH. To compare expression levels of specific genes between two samples, the comparative cycle threshold (Ct) method was used. By this method, expression differences between two samples are calculated as a ratio of the cycle numbers necessary to reach an arbitrary amplification value (formula = 2(−ΔΔCt)).
Transient Transfection of Primary Myoblasts
Transfection of myoblasts was performed as follows. A total of 4–4.5 × 105 fetal myoblasts were plated on a collagen coated six-well plate (diameter, ∼3.5 cm) in GM. Twenty-four hours after plating, transfection was performed using FuGENE6 (Roche Applied Sciences) with ratio of DNA to FuGENE6 being 1 to 3 (1 μg of DNA to 3 μl of FuGENE6). Twenty-four hours after transfection, cells were rinsed with PBS twice, then switched to DM and incubated. Cell lysates and RNA were prepared for enzyme assays and real-time PCR, respectively.
Knockdown of Sox6 Expression by Sox6-siRNA
Three Sox6-siRNA expression vectors were constructed using the mU6pro vector which generates small hairpin RNA under the control of the mouse U6 small nuclear RNA promoter (a gift from Dr. Turner at University of Michigan; Yu et al.,2002). Sequences used for generating Sox6-siRNA are as follows: Sox6-siRNA 1, nucleotides 585-602; Sox6-siRNA 2, nucleotides 923-941; Sox6-siRNA 3, nucleotides 2037-2054 (nucleotide numbers are based on mouse Sox6 mRNA, GenBank MMU32614). The Sox6-siRNA expression vectors were transfected to wild-type myoblast cultures as combinations of Sox6-siRNA's 1 and 2, 2 and 3, 1 and 3, and 1, 2, and 3. We chose to transfect a combination of Sox6-siRNA constructs because a single construct transfection gave only ∼50% reduction of Sox6 mRNA, however, double or triple transfection achieved ∼80% reduction of Sox6 mRNA expression (see the Results section). Twenty-four hours after transfection, wild-type myoblasts were switched into DM for 48 hr and total RNA was prepared using Trizol (Invitrogen).
Analysis of the MyHC-β Promoter Using the Luciferase Reporter Gene Constructs
Luciferase expression vectors driven by the MyHC-β promoter were kindly provided by Dr. K.M. Baldwin at University of California, Irvine (Huey et al.,2002,2003). Each of these constructs contains 3,500 bp, 408 bp, 215 bp, and 171 bp of the 5′-upstream sequence of the rat MyHC-β gene. The MyHC-β expression vector (408 bp) containing base substitutions (ACC to CGG) in the MCAT sequence in the βe3 element (−215/−188; Thompson et al.,1991) was also obtained from Dr. Baldwin, which was originally termed 408 Δβe3 (Huey et al.,2002,2003). We re-named this construct to 408-mMCAT to distinguish it from other mutant MyHC-β luciferase constructs we generated. In 408-mSox, ACA to CAG substitutions were introduced to disrupt the canonical Sox sequence (−202/−196) in the βe3 element. The 408-mMCAT/mSox (double mutant) contains both mutations in the MCAT and Sox sequences. Mutations in the two constructs, 408-mSox and 408-mMCAT/mSox, were introduced by PCR amplification using the following primers: 5′-CATGCCATACCACACAGATGACG-3′ (forward primer for mSox), 5′-CATGCCATCGGACACAGATGACG-3′ (forward primer for mMCAT/mSox), and 5′-TGGCGTCTTCCATGGTGGC-3′ (reverse primer for the both, located in the luciferase gene sequence in pGL3 Basic, Promega). The CMV promoter driven lacZ expression vector pCMV-SPORT-β-gal (Invitrogen) was used to normalize transfection efficiencies.
Transfection of myoblasts with the luciferase expression vectors and pCMV-SPORT-β-gal was performed as described above. Luciferase and β-galactosidase enzyme assays were performed using Luciferase Assay System and β-Galactosidase Enzyme Assay System (Promega), respectively. β-Galactosidase activities were used to normalize transfection efficiencies among plates. Three independent transfection experiments (using different myoblast preparations), each with duplicate plates, were performed.
Transiently transfected C2C12 myoblast cells (ATCC CRL-1772) grown in DMEM with 10% FBS were used for ChIP assay. Three expression vectors, the wild-type Sox6 protein expression vector, 408-WT MyHC-β luciferase construct, and 408-mSox MyHC-β luciferase construct were used for transfection. The Sox6 expression vector was constructed by inserting the full-length Sox6 cDNA (GenBank U32614) in the multiple cloning site of pcDNA3.1 (Invitrogen). C2C12 cells (2 × 107) were transfected using FuGENE6 with either (1) a Sox6 expression vector and 408-WT MyHC-β luciferase construct, or (2) a Sox6 expression vector and 408-mSox MyHC-β luciferase construct. Cell lysates were prepared 36 hr after transfection. A total of 107 cells were used for one ChIP experiment following the protocol described by O'Geen et al. (2006) except that the protein G Sepharose (Amersham) was used for antibody precipitation, and following reversal of cross-links, the DNA was further treated with Proteinase-K before purification. A total of 2.5 μg of Sox6 antibody (ab12054; Abcam, Inc.) was used for one ChIP experiment. Purified DNA was collected in 50 μl of TE. Five microliters of the DNA was used for a 20-μl PCR reaction, and 8 μl of it was separated on a 2% agarose gel. To amplify the DNA purified from the 408-WT transfected cells, the forward primer of the rat wild-type MyHC-β promoter, 5′-CATGCCATACCACAACAATGACG-3′, was used along with the reverse primer in the luciferase gene in pGL3 Basic as described above. To amplify the DNA purified from the 408-mSox transfected cells, the same forward–reverse primer set used for generating the mSox mutation (as described in the section above) was used. Both primer sets amplify 295-bp amplicons. The amplification program was 94°C for 5 min, 35 cycles of 94°C for 30 sec, 58°C for 30 sec, and 72°C for 1 min.
We thank Drs. Richard Tucker, Paul Fitzgerald, and Mr. Adam Jenkins for helpful discussion. We also thank Quan Nguyen for his technical assistance. N.H. was funded by American Heart Association, Beginning Grant in Aid, and March of Dimes Birth Defects Foundation, Basil O'Connor Starter Scholar Research Award.