Brain and muscle Arnt-like 1 is a key regulator of myogenesis

The circadian clock is a cell-intrinsic molecular mechanism that generates ∼24-hour daily cycles of biological processes, which responds to external timing cues and imparts a temporal regulation to key physiological functions (Bass and Takahashi, 2010; Panda et al., 2002). These daily rhythms are driven by interlocking feedback loops of transcriptional and translational regulation (Harmer et al., 2001) that exists in essentially all tissue or cell types of the organism (Yoo et al., 2004). The central clock in the suprachiasmatic nuclei is responsible for receiving light input, whereupon it synchronizes peripheral clocks in other tissues, including the skeletal muscle (Schibler and Sassone-Corsi, 2002). The molecular clocks residing in skeletal muscle has been shown to control the daily oscillations of gene expression, with ∼40% of rhythmically expressed transcripts belonging to metabolic processes (McCarthy et al., 2007; Miller et al., 2007).

Brain and muscle Arnt-like protein 1 (Bmal1) is a transcriptional activator of the molecular clock feedback network and is highly expressed in skeletal muscle (Hogenesch et al., 1997; Ikeda and Nomura, 1997). Bmal1 binds to canonical E-box motifs (CACGTG) as a heterodimer with circadian locomotor output cycles kaput (CLOCK) on its target promoters to activate transcription (Gekakis et al., 1998; Takahashi et al., 2008), which includes the negative regulators of their activity, the Pers (Period1, 2 and 3) and Crys (Chryptochrome 1 and 2). These transcriptional repressors can in turn associate with the Bmal1/CLOCK complex and suppress gene activation. Thus, this transcriptional negative feedback loop, further coupled with transcriptional and posttranscriptional mechanisms, elicits the robust daily oscillations of the molecular clock and its output genes (Takahashi et al., 2008). As an essential component of the molecular clock network, inactivation of Bmal1 in mice leads to complete loss of circadian rhythm under constant dark conditions (Bunger et al., 2000). Interestingly, these mice exhibited severely compromised skeletal muscle strength and reduced survival, which can be restored by its targeted overexpression in the muscle (Kondratov et al., 2006; McDearmon et al., 2006). These studies suggests a critical role of the molecular clock in the maintenance of mature post-mitotic skeletal muscle functions (Andrews et al., 2010; Kondratov et al., 2006; McDearmon et al., 2006). But, whether it can directly regulate myogenic differentiation, a key process not only in muscle development but also postnatal muscle growth and maintenance, is not known.

Differentiation of myogenic precursors into mature myotubes, or myogenesis, is a highly orchestrated process that involves myogenic precursor cell determination, myoblast proliferation and eventual terminal differentiation and fusion to become mature multinucleated myotubes (Braun and Gautel, 2011; Buckingham, 2006). This complex yet precise myogenic program requires the concerted effort of a cascade of myogenic regulatory factors in a tightly-regulated temporal and spatial manner during development (Pinney et al., 1988), with Myf5 and MyoD specifying myoblasts from their precursors and subsequent activation of myogenin driving terminal differentiation. Thus an intriguing question is, could the molecular clock, as an evolutionarily conserved timing mechanism, play a role in the temporal regulation of this process?

Recently, components of the Wnt signaling pathway were found to be potential Bmal1 targets in genome-wide ChIP-Seq in the liver (Rey et al., 2011) and epidermal stem cells (Janich et al., 2011). As Wnt is a major instructive signal that drives skeletal muscle development (Cossu and Borello, 1999; Tsivitse, 2010), we hypothesized that Bmal1 may critically impact myogenesis through this signaling cascade. Using cellular systems of myocyte differentiation and genetic models, we demonstrate that Bmal1 is a key positive regulator of myogenesis and identify critical steps along the canonical Wnt signaling pathway as its direct transcriptional targets.

Bmal1 is highly expressed in skeletal muscle with functions in metabolic regulation (Hogenesch et al., 1997; Ikeda and Nomura, 1997), but whether it participates in skeletal muscle growth is not known. To investigate the role of Bmal1 in postnatal skeletal muscle growth, we used young adult mice (8–12 weeks old) before the onset of aging-related pathologies, as Bmal1-null (Bmal1^−/−) mice display an early aging phenotype with reduced survival starting at 26 weeks of age (Kondratov et al., 2006). NMR analysis revealed a significant reduction of total muscle mass in Bmal1^−/− mice compared to wild-type littermate controls (Fig. 1A), in addition to increased fat mass as we reported previously (Guo et al., 2012). The reduction in global lean mass was also evident in individual muscle weights of muscles of mixed fiber-type gatrocnemius, and predominantly slow-twitch soleus muscle (Fig. 1B). In contrast, heart weights were similar in Bmal1-null mice and controls, suggesting that the function of Bmal1 in muscle growth is specific to skeletal muscle. Consistent with muscle weights, we observed reduced myofiber diameter in Bmal1-deficient hindlimb muscles (Fig. 1C), with a significant shift toward smaller muscle fibers (Fig. 1D). These in vivo findings suggest that Bmal1 is important for postnatal muscle growth or maintenance of muscle mass.

We thus tested whether the reduced muscle mass in Bmal1^−/− animals could be a result of impaired function of resident myogenic precursor cells, a major source of skeletal muscle growth and maintenance postnatally (Wagers and Conboy, 2005). We first established that the isolated myoblasts from Bmal1^−/− and wild-type (WT) littermates are homogeneous populations, as wild-type myoblasts prepared by this method uniformly differentiated into myotubes with almost 100% efficiency (Fig. 2A) when subjected to differentiation in culture by 2% horse serum for 4 days. In contrast, genetic deletion of Bmal1 markedly impaired the formation of organized myotubes, and a significant number of myoblasts remained unincorporated into multi-nucleated myotubes by 2 days of differentiation (Fig. 2A,B). This was particularly evident by immunostaining of the myocyte-specific structural protein, myosin heavy chain (MHC; Fig. 2B), which showed significantly reduced percentage of MHC-positive myonuclei in Bmal1^−/− cells (Fig. 2B, right). In line with this phenotype, mRNA expression levels of myogenic markers Myf5, Mrf4 and myogenin, as well as MHC3 (Fig. 2C) (supplementary material Fig. S1, left panel) were substantially suppressed in Bmal1-null myoblasts relative to WT cells. We observed similar reductions in Myf5, MyoD and Troponin I protein in Bmal1^−/− myoblasts (Fig. 2D), although Pax7 level was comparable to WT cells. Taken together, these findings demonstrate that Bmal1 is required for the proper myogenic differentiation of primary myoblasts, a major source of postnatal skeletal muscle myogenesis.

We further tested whether Bmal1 is important for myogenic differentiation in a commonly used C2C12 myoblast cell line (Blau et al., 1985; Yaffe and Saxel, 1977). We generated C2C12 cell lines expressing Bmal1-specific shRNA to achieve stable Bmal1 knockdown (KD), and one construct (VGM5520-99342254, out of three constructs tested) with ∼60% Bmal1 mRNA knockdown efficiency (Fig. 3A), was used in our subsequent experiments. Interestingly, consistent with a role for Bmal1 in normal myogenic differentiation, both Bmal1 transcript (Fig. 3A) and protein (Fig. 3B) were markedly induced during differentiation of C2C12 myoblasts (approximately fivefold at day 6 compared to day 0 at mRNA level). In the KD cell line, Bmal1 mRNA was effectively reduced to 40% of the level of the companion scrambled control (SC) cell line and maintained throughout differentiation to day 6 (Fig. 3A). Silencing of Bmal1 was more striking at the protein level (Fig. 3B). In stark contrast to fully differentiated morphology of the scrambled control cells, very few cells expressing Bmal1-shRNA were able to differentiate into mature myotubes (∼6.7%; Fig. 3C,D). SC cells differentiated with a similar efficiency as the naïve C2C12 cells, excluding potential effects of stable selection. Conversion of naïve C2C12 myoblasts to organized myotubes under low serum condition normally requires 6 days. We then extended the differentiation to day 9 when normal myoblasts are converted to organized myotubes, Bmal1-deficient cells still failed to differentiate (Fig. 3C), indicating that this phenotype is not a result of delayed differentiation. Consistent with the defect in morphological differentiation, immunostaining of MHC further confirmed the profound defect in myogenic conversion with KD of Bmal1 (∼6.7% MHC+ myonuclei versus 48.2% in SC, Fig. 3D, right panel). In addition, mRNA expression of mature myocyte markers, myosin heavy chain 3 (MHC3) and myosin light chain 1 (MLC1) was nearly abolished, as compared to their robust induction in the control cells during myogenic differentiation (Fig. 3E). This defect in myogenesis by silencing of Bmal1 occurred early during differentiation, as both the transcript (Fig. 4A) and protein levels (Fig. 4B) of the early myogenic regulatory factors, Myf5, MyoD and Mrf4 (supplementary material Fig. S1) were markedly suppressed in C2C12 cells expressing Bmal1-specific shRNA. Consistent with the impaired terminal differentiation of these cells, myogenin expression was substantially downregulated as well.

To test whether Bmal1 is also sufficient to drive the myogenic program, we used stable overexpression of Bmal1 (BM cDNA) in C2C12 myoblasts. The cells stably expressing Bmal1 had ∼10-fold higher Bmal1 transcript levels (Fig. 5A) and substantially elevated Bmal1 protein (Fig. 5E), as compared to cells transfected with empty vector control (pcDNA3). In contrast to the effects observed with inhibition of Bmal1 function, its forced expression drastically accelerated myogenesis, with majority of the cells exhibiting differentiated myotube morphology by day 3, compared to very few in the control population at this early stage of differentiation (Fig. 5B). At day 4 of differentiation in the BM cDNA-expressing myoblasts, MHC immunostaining revealed already markedly advanced myogenic differentiation with many enlarged, fused myotubes (Fig. 5C), which are typically observed in the control cells at day 9. Forced expression of Bmal1 resulted in a significant ∼60% increase in the percentage of MHC+ (Fig. 5D, left panel) as well as fused myonuclei (Fig. 5D, right panel) compared to the vector control. At the molecular level, these morphological changes were accompanied by stronger induction of both the mRNAs encoding the myogenic terminal differentiation factor, myogenin, as well as MLC1 (Fig. 5A) in Bmal1-expressing C2C12 myoblasts relative to control cells. Correspondingly, protein levels of Myf5 and MyoD were increased (Fig. 5E). Consistent with the early acquisition of differentiated morphology, the mature myocyte marker Troponin I was also expressed earlier and to a greater extent in C2C12 myocytes expressing Bmal1 cDNA (Fig. 5E). Thus, these data indicate that Bmal1 is required for myoblast differentiation, it is also sufficient to promote this process.

We next investigated the molecular basis of the effect of Bmal1 on myocyte differentiation. A genome-wide ChIP-Seq study in the liver (Rey et al., 2011) revealed that several genes of the canonical Wnt signaling pathway are potential Bmal1 targets and have E-box elements in their promoter regions. Our previous study (Guo et al., 2012) identified that these and additional components of the Wnt cascade are under Bmal1 regulation in adipocytes. As Wnts are major morphogenic signals controlling skeletal muscle development (Cossu and Borello, 1999; Logan and Nusse, 2004), we investigated whether this pathway mediates Bmal1 action during myogenic differentiation. We first examined whether loss of Bmal1 alters expression of genes of the Wnt pathway that were identified as putative Bmal1 targets (Rey et al., 2011). Consistent with our hypothesis, mRNA levels of many genes involved in canonical Wnt signaling were significantly reduced in Bmal1^−/− primary myoblasts, including those of the ligand Wnt10a, signaling mediators Dishevelled 2 (Dvl2) and β-catenin, and the transcription factor T-cell Factor 3 (TCF3; Fig. 6A). Importantly, Wnt signaling activity is also attenuated in Bmal1^−/− myoblasts. The known Wnt target gene, Axin2, is robustly induced during normal differentiation, while this induction was nearly abolished in Bmal1-deficient cells (Fig. 6A). To directly assess Wnt pathway activity, we determined both the cytoplasmic accumulation and nuclear abundance of β-catenin in stable C2C12 Bmal1 KD cells under basal conditions and upon Wnt3a addition. β-catenin is the key signal transducer of canonical Wnt signaling. In response to Wnt ligands, β-catenin is stabilized and accumulates in the cytosol, ultimately leading to its nuclear translocation and activation of Wnt target genes (Clevers, 2006). β-catenin cytoplasmic as well as its nuclear abundance in Bmal1 KD cells was markedly lower as compared to the SC cells, under either basal conditions or upon Wnt3a stimulation (Fig. 6B,C). And the Wnt3a-induced nuclear β-catenin accumulation in the control cells was completely abrogated with silencing of Bmal1 (Fig. 6C), indicating a profound defect in canonical Wnt signaling transduction. In addition, β-catenin protein in the total lysate of Bmal1-deficient cells was reduced as well, likely reflecting Bmal1-dependent regulation of β-catenin transcription. On the other hand, although forced expression of Bmal1 led to only slightly higher β-catenin level in the cytoplasm (Fig. 6B), both basal and Wnt-activated nuclear abundance are increased compared to vector control (Fig. 6C). As assessment of β-catenin stabilization in the cytosol only reflects Wnt signaling upstream of this event and Bmal1 regulates a number of genes that signal downstream of β-catenin, we analyzed a distal readout of Wnt pathway activity using TOPFlash luciferase reporter assays. This reporter contains optimized TCF binding sites (Veeman et al., 2003) and is widely used to monitor canonical Wnt signaling. Consistent with reduced cytoplasmic β-catenin accumulation in Bmal1-KD cells, Wnt reporter activity was also attenuated relative to controls (Fig. 6D, left). Whereas, in C2C12 cells overexpressing Bmal1, this activity was augmented (Fig. 6E, left). There was an even more pronounced attenuation of Wnt reporter activity upon Wnt3a treatment in the Bmal1-KD cells (∼30% of control level, Fig. 6D, right). In contrast, forced expression of Bmal1 potentiated Wnt3a-stimulated TOPFlash reporter activity (Fig. 6E, right). Thus, we identified Bmal1 as a new regulator of canonical Wnt signaling in myoblasts.

Based on above findings, we tested whether Bmal1 directly regulates the identified Wnt pathway genes. Owing to the circadian regulatory nature of Bmal1 on its direct targets, we predicted that this mechanism may also impart circadian expression profiles to these genes. To reveal the potential circadian time-dependent association of Bmal1 with its target promoters, we synchronized C2C12 myoblasts by serum shock, an established method of eliciting circadian profiles in cell culture systems (Balsalobre et al., 1998; Chalmers et al., 2008), followed by chromatin immunoprecipitation (ChIP)-qPCR analysis. We previously (Guo et al., 2012) established that 8 and 20 hours after serum shock (CT8 and 20, CT: circadian time) corresponds to the highest and lowest Bmal1 binding activity on its target promoters in C3H10T1/2 cells (Guo et al., 2012). Consistent with this, in C2C12 myoblasts, Bmal1 strongly occupies most of the candidate target promoters of the Wnt pathway analyzed (>20-fold enrichment over IGg) at CT8 (Fig. 7A), including Dvl2, β-catenin and TCF3. In contrast, this association was near completely abolished at CT20, corresponding to the lowest Bmal1 activity. The known Bmal1 target gene in the core molecular clock loop, the Rev-erbα gene, displayed a similar pattern of enriched Bmal1 binding at CT8 and nearly undetectable Bmal1 occupancy at CT20. Interestingly, the Wnt ligand, Wnt10a, which harbors an identified E-box element in its promoter (Rey et al., 2011), exhibited a different pattern of Bmal1 occupancy. Bmal1 associated with the Wnt10a locus with relatively less enrichment than other Bmal1 targets at CT8, but there was significant residual Bmal1 binding at CT20, suggesting that Bmal1 occupancy is target promoter specific. We then determined whether the direct regulation by Bmal1 would confer circadian expression patterns on these genes. As expected, serum shock elicited cyclic expression of Bmal1 protein with robust induction at 12 hours and a second cycle peak at 36 hours (Fig. 7B). Remarkably, β-catenin follows the same oscillation pattern as Bmal1, with peaks at 12 and 36 post serum shock. Furthermore, qPCR analysis, performed every 4 hours continuously for 48 hours after serum shock, revealed detailed circadian expression patterns of the Wnt pathway genes that are direct Bmal1 targets. Wnt10a, Dvl2, β-catenin and TCF3 all displayed robust rhythmic expression profiles in phase with Bmal1 peaks and troughs (Fig. 7C), and these oscillatory patterns were largely abolished in the Bmal1-null cells. MyoD also exhibited a circadian oscillatory pattern, consistent with the report that its core enhancer is under the transcriptional regulation of Bmal1, as reported previously (Andrews et al., 2010) and confirmed by our ChIP study (supplementary material Fig. S2). In contrast, level of TBP does not oscillate across these circadian time points examined. Taken together, these analyses provided direct evidence that multiple steps of the canonical Wnt signaling pathway are under the coordinated circadian control of Bmal1.

Above findings suggests that the effect of Bmal1 on myogenesis may be mediated by its transcriptional regulation of the Wnt pathway. We thus postulated that activation of Wnt pathway may rescue the myogenic defects in Bmal1^−/− primary myoblasts. In wild-type cells, either Wnt3a or the GSK-3β inhibitor, SB-216763, significantly enhanced myogenic differentiation as demonstrated by increased number of MHC-positive myotubes or their fusion (Fig. 8A,B), though the effect of Wnt3a is more robust than SB-216763. Interestingly, Wnt3a was most effective in promoting myoblast fusion to form multi-nucleated myotubes at this differentiation stage, and this effect was lost in the Bmal1-null cells (Fig. 8B). In contrast, although Wnt activation was able to augment myogenesis in Bmal1-deficient cells to that of the untreated wild-type cell levels, the terminal differentiation in response to Wnt3a nevertheless remains blunted, based on the morphology and quantification of MHC immunostaining (Fig. 8A,B). Importantly, the increment of percentage of cells positive for MHC in response to Wnt activation was nearly two (Wnt3a) or threefold higher (SB-216763) in the absence of Bmal1 (Fig. 8C, left panel), indicating that Wnt pathway indeed corrects, albeit partially, the deficits in differentiation. However, the efficacy of Wnt activation on fusion was comparable in these cells (Fig. 8C, right panel). Further analysis of expression levels of myogenic markers indicated that the myogenic response induced by Wnt3a is either abolished (Myf5, MyoD) or significantly attenuated (MHC3, MLC1) in Bmal1-deficient myoblasts (Fig. 8D). We also examined expression of the Wnt target gene Axin2 as an indicator of Wnt activity in these cells. Compared to its approximately threefold induction by Wnt3a in WT cells, both the basal and Wnt-stimulated level of Axin2 expression were markedly lower with near complete loss of the Wnt responsiveness in Bmal1^−/− myoblasts. A similar tendency of reduced effectiveness of SB-216763 was observed in Bmal1-deficient cells (Fig. 8E), as shown by consistently lower level of myogenic gene expression compared to wild-type controls. Thus, activation of Wnt signaling by either Wnt3a or GSK-3β inhibition only partially restored myogenic differentiation in Bmal1-null conditions. These results indicate that correcting individual components of the canonical Wnt signaling is not sufficient to fully rescue Bmal1 loss-of-function, as numerous steps of this cascade are downregulated in Bmal1-null myoblasts (Figs 6, 7). These observations further raise the possibility that additional mechanisms under Bmal1 regulation may contribute to the myogenic defects.

As previous studies (Andrews et al., 2010; Zhang et al., 2012) indicated that the key myogenic factor, MyoD, is a Bmal1 target, we investigated whether forced expression of MyoD, alone or in concert with Wnt signaling, is sufficient to restore myogenesis in Bmal1-deficient myoblasts. As shown by MHC immunostaining (Fig. 9A,B), in normal primary myoblasts, MyoD alone robustly enhanced the formation of MHC+ myotubes, while Wnt3a addition in MyoD-overexpressing cells primarily promoted the fusion process as indicated by increased fusion index (Fig. 9A,C). In contrast, in Bmal1-null cells, forced expression of MyoD or in combination of Wnt3a resulted in a similar ∼1.5-fold and 2-fold increase of MHC+ myocytes (Fig. 9B, right panel), although the total number of these cells remained to be lower (Fig. 9B, left panel) and their morphology were drastically different than wild-type controls (Fig. 9A). Further quantification revealed that not only the percentage of fused cells were reduced (Fig. 9C, left panel), the fold increment of fusion in response to MyoD or MyoD and Wnt3a in the Bmal1-deficient myoblasts was ∼50% lower (Fig. 9C, right panel). These data indicate that although MyoD can enhance the maturation of MHC+ myocytes in the absence of Bmal1, its action on fusion is significantly diminished. When assessed by the transcript levels of mature myocyte markers MHC3 and MLC1, MyoD1 effect on promoting differentiation in Bmal1-null cells were similarly deficient as compared to its robust action in WT myoblasts (Fig. 9D). In both the WT and Bmal1-null cells, MyoD cDNA achieved comparable approximately fourfold overexpression over its respective vector alone control (pcDNA3), and initial transfection efficiency in both cell types were similar at ∼60% as assessed by co-transfected GFP. On the other hand, as an integral response to myogenic induction during differentiation, the consistently downregulated MyoD expression in Bmal1-deficient myoblasts was additional evidence of the myogenic deficits present in these cells. As expected, MyoD does not significantly affect Myf5 expression, which acts upstream of MyoD in the myogenic induction cascade. Collectively, these results indicate that although MyoD largely corrects the defect in formation of MHC+ myocytes with Bmal1 deficiency, it fails to fully restore differentiation at the myocyte fusion stage.

Peripheral circadian clocks exist in skeletal muscle and regulate genes involved in its metabolic activity (Andrews et al., 2010; McDearmon et al., 2006; Miller et al., 2007), but their potential function in myogenesis have not been explored. Here we show for the first time that Bmal1, the essential molecular clock activator, promotes myogenic differentiation through coordinated transcriptional control of genes of the Wnt cascade, a critical myogenic signal during embryogenesis (Cossu and Borello, 1999) and postnatal skeletal muscle maintenance (Brack et al., 2008; Tsivitse, 2010). Our demonstration of this cell-autonomous function of Bmal1 in driving myogenesis was corroborated by the in vivo phenotype of Bmal1^−/− mice, while identification of its transcriptional regulation of Wnt pathway genes provides the molecular basis of these findings.

We identified the canonical Wnt cascade as a molecular pathway mediating the action of Bmal1 on myogenic differentiation. Since Wnt signaling is a critical regulator of stem cell behaviors in many tissue types including the resident myogenic precursors of skeletal muscle, the satellite cells (Clevers, 2006; Ling et al., 2009; Nusse, 2008; Tsivitse, 2010), our findings raise the possibility that Bmal1 may modulate satellite cell properties through Wnt signaling to maintain postnatal muscle growth and remodeling. In adult muscle, myogenic precursors are the major source for skeletal muscle remodeling process elicited by exercise that critically determines muscle mass (Shi and Garry, 2006; Wagers and Conboy, 2005). Our results revealed presence of muscle atrophy in Bmal1^−/− early at 2–3 months of age, supporting a hypothesis that Bmal1-dependent modulation of differentiation of adult myogenic progenitors participates in maintenance of muscle mass, in addition to its impact on mature skeletal muscle physiology. Future studies of the function of Bmal1 in the skeletal muscle regenerative response to injury would provide further evidence.

During embryonic development, Wnt activity is a major instructive signal that drives lineage specification, patterning and eventual differentiation of skeletal myocytes (Cossu and Borello, 1999). Therefore, an intriguing aspect, although out of the scope of the current study, is what potential consequence of loss of the temporal control by Bmal1 may have on skeletal muscle development in vivo. In adult stages, Wnt signaling contributes to skeletal muscle remodeling and repair (Brack et al., 2008; Wagers and Conboy, 2005). Although the precise mechanisms of Wnt action have not been fully elucidated, known molecular targets of Wnt signaling in myogenesis include the key myogenic factors, Myf5 and MyoD (Borello et al., 2006; Cossu and Borello, 1999; Tajbakhsh et al., 1998). Interestingly, Wnt action displays embryonic origin specificity. Although Myf5 was identified as a direct target of Wnt through a TCF-binding site upstream of its early epaxial enhancer (Borello et al., 2006), Wnt activity predominantly regulates MyoD in mesoderm-derived myoblasts (Tajbakhsh et al., 1998). Compelling in vivo evidence suggests that Myf5 and MyoD, although both critical for myoblast differentiation, share functionally redundant roles during embryonic muscle development (Rudnicki et al., 1992; Rudnicki et al., 1993). Furthermore, the expression level of another myogenic factor, Mrf4 in Myf5:Myod double-null mice is also involved in establishing myogenic cell fate (Kassar-Duchossoy et al., 2004). Our results indicated that Bmal1 effect on myogenesis mediated by the Wnt pathway could converge on Myf5, MyoD and Mrf4 (supplementary material Fig. S1), which were markedly downregulated with attenuation of Bmal1 function. Additionally, our ChIP result (supplementary material Fig. S2) indicates that Bmal1 can directly control the MyoD core enhancer, as reported previously (Zhang et al., 2012), as well as the Myf5 promoter. Therefore, conceivably in Bmal1-deficient myoblasts in our study, both the attenuated Wnt signaling and loss of direct Bmal1 transcriptional activation contributed to the lack of sufficient induction of these critical myogenic determination factors. And as a result, these early changes led to subsequently blunted myogenic events, such as activation of the myogenin gene, and eventual impairment of myogenic differentiation.

So far, evidence regarding the role of molecular clock in skeletal muscle mainly concerns its regulation of metabolic functions (Andrews et al., 2010; McDearmon et al., 2006; Miller et al., 2007). Interestingly, although previous reports (Andrews et al., 2010; Zhang et al., 2012) indicated that Bmal1 may activate MyoD promoter through a distal enhancer region, they examined involvement of Bmal1 in the regulation of muscle strength and mitochondrial function in mature post-mitotic myofibers (Andrews et al., 2010). In contrast, we investigated how Bmal1 exerts temporal control of myogenesis, a fundamentally different aspect of the function of circadian clock in skeletal muscle formation. It is possible that Bmal1 direct regulation of the MyoD and Myf5 promoters are additional mechanisms responsible for its effects on myogenesis as we observed, working in concert with its transcriptional regulation of the Wnt pathway. Our finding that attenuation of Bmal1 function impaired MyoD mRNA and protein levels during myogenic differentiation, and that forced expression of MyoD can partially rescue formation of MHC+ myocytes but fails to fully restore the fusion process, is certainly in line with this notion. Interestingly, possibly due to partially overlapping effects on the downstream myogenic pathways or the presence of Bmal1-regulated additional myogenic mechanisms, the combination of Wnt activation in the presence of MyoD was not able to fully rescue myogenic deficits of the Bmal1-null cells.

In primary myoblasts devoid of Bmal1, activation of Wnt signaling by either Wnt3a treatment or GSK-3β inhibition achieved only partial rescue of their myogenic defect. The markedly lower response of Bmal1-deficient cells to Wnt3a indicates that the impaired Wnt signaling was not fully corrected by these interventions alone. As we demonstrated, genes encoding many components downstream of Wnt ligand in the canonical Wnt pathway, such as Dvl2, β-catenin and TCF3, are direct Bmal1 transcriptional targets; and thus, the partial rescue effect is likely due to the attenuated function of these additional steps involved in Wnt signaling. However, it may also be possible that this occurred as a result of failure to restore additional pathways involved in myogenesis that are under Bmal1 regulation. In epidermal stem cells, besides the Wnt cascade, genes of the signaling transducers of Notch (Hes1, Hes7) and TGF-β pathways (TGF-β1, TGFbr2) were identified as putative Bmal1-regulated genes (Janich et al., 2011). As these are known signals participating in skeletal muscle development (Ge et al., 2012; Kollias and McDermott, 2008; Tsivitse, 2010), alterations in activities of these pathways may have also contributed to the myogenic deficits observed with loss of Bmal1 function. In particular, Notch signaling is known to cooperate with Wnt in muscle development (Buas and Kadesch, 2010) and it precedes Wnt signal to activate satellite cell proliferation in muscle regeneration (Brack et al., 2008), whereas the key signal transducer of the TGF-β pathway, Smad3, also promotes myogenesis (Ge et al., 2012; Kollias and McDermott, 2008; Tsivitse, 2010). How these additional Bmal1-regulated developmental signaling pathways coordinately modulate skeletal muscle development awaits investigation in our future studies.

In summary, we demonstrate that the core clock regulator, Bmal1, through its transcriptional control of the canonical Wnt signaling pathway, confers a temporal regulatory element to fine-tune myogenic differentiation. Given the important roles of the Wnt cascade in stem cell biology and tissue development, we anticipate that this cell-intrinsic timing mechanism may modulate developmental processes and stem cell behaviors in a myriad of tissue types.

Animals were maintained in the Methodist Hospital Research Institute vivarium under a constant 12∶12 light dark cycle, with lights on at 7:00 am (ZT0). All experiments were approved by the IACUC committee of Methodist Hospital Research Institute. Bmal1^−/− mice (Bunger et al., 2000) were obtained from the Jackson Laboratory. Mice, 8–12 weeks of age, were used throughout the study to exclude effects of aging-related pathologies (Kondratov et al., 2006).

The C2C12 myoblast cell line was obtained from ATCC and maintained in 10% FBS DMEM. 2% horse serum supplemented DMEM was used for differentiation of 80–90% confluent cultures. All shRNA constructs for Bmal1 (VGM5520-99941526, VGM5520-99342254, VGM5520-99211363) and scrambled control vector (RHS4346) were purchased from Open Biosystems. pcDNA3-Bmal1 cDNA construct (Fu et al., 2002) was used for overexpression. Stable C2C12 cell line expressing Bmal1 shRNA or cDNA were constructed by transient transfection using FUGENE 6 (Roche). Antibiotic selection (1.5 µg/ml Puromycin or 0.8 mg/ml Neomycin, respectively) following transfection were maintained for 7–10 days to obtain stable pools, with sub-culturing as necessary prior to confluency to preserve differentiation efficiency. MyoD cDNA (Origene), together with GFP control, were transiently transfected into primary myoblasts by electroporation using Amaxa Nucleofactor transfection kit (Lonza), and cells were allowed to recover overnight before induction of differentiation.

Primary myoblasts were isolated from hind limb muscle of 4-week-old mice as described (Rando and Blau, 1994). Briefly, muscles were minced into small pieces and subjected to 1% collagenase digestion at 37°C for 30 minutes with agitation. Cells were seeded on collagen coated-plates with preplating for 15 minutes to deplete fibroblasts. Subsequent selective growth and serial passaging over five to six passages were performed to enrich for myoblasts. Purity of myoblasts obtained by this method was confirmed by their uniform differentiation into myosin heavy chain (MHC)-positive myotubes.

Trizol (Invitrogen) and RNeasy miniprep kits (Qiagen) were used to isolate total RNA from snap-frozen muscle tissues and cells, respectively. cDNA was generated using q-Script cDNA Supermix kit (Quanta Biosciences) and quantitative PCR was performed using a Roche 480 Light Cycler with Perfecta SYBR Green Supermix (Quanta Biosciences). Relative expression levels were determined using the comparative Ct method to normalize target genes to β-actin or 36B4 internal controls. Primer sequences used are included in supplementary material Table S1.

For total protein, 20–50 µg was resolved on SDS-PAGE gels followed by immunoblotting after PVDF membrane transfer, and developed by chemiluminescence (Supersignal; Pierce Biotechnology). Myf5, MyoD, myogenin, β-actin and TBP antibodies were obtained from Santa Cruz Biotechnology, β-catenin from BD Biosciences, MHC clone A4.1025 antibody from Millipore, and Bmal1 antibody AB93806 from Abcam. For immunostaining, cells were grown and differentiated on chamber slides coated with poly-D-lysine (BD Biosciences).

Serum shock was used to synchronize primary and C2C12 myoblasts as described. To rule out potential effect of cell cycle, we differentiated primary myoblasts in 2% horse serum for 2 days, cultured in serum free medium overnight and induced serum shock (20% FCS) for 1 hour. RNA samples were collected every 4 hours for 48 hours after serum treatment. For ChIP assays, C2C12 cells were fixed with formaldehyde and sonicated to shear the chromatin. Immunoprecipitation was performed with Bmal1 antibody or control rabbit IgG plus protein A/G beads. The immunoprecipitated chromatin fragments were eluted and purified (Qiaquick PCR purification kit, Qiagen). Real-time PCR was carried out with an equal volume of each reaction, with primers flanking E-box binding sites identified by ChIP-Seq annotation of (Rey et al., 2011). Specific primer sequences were as published previously (Guo et al., 2012) with additional primers included in supplementary material Table S2. Negative control primers amplified the TBP promoter and the known Bmal1 target gene Rev-erbα served as a positive control. Values were normalized to 1% of input, expressed as fold enrichment over IgG.

Assessment of cytosolic amount of β-catenin was carried out as described (Young et al., 1998) after treatment with Wnt3a or control media for 8 hours. Wnt3a-conditioned media was obtained from cultures of Wnt3a producing L cells (Wnt3a-L cell) as described (Jackson et al., 2005), and used in a 1∶1 ratio. Treated cells were lysed in hypotonic buffer (20 mM Tris, pH 7.5, 25 mM sodium fluoride, and 1 mM EDTA) containing protease inhibitor (Roche), followed by 30 strokes in a Dounce homogenizer. Membranous and cytosolic material was obtained by ultracentrifugation at 100,000 g for 40 minutes. The supernatant was collected as the cytosolic fraction and total cellular protein was obtained separately. Nuclear protein was extracted as described (Peterson et al., 2008). Briefly, cells were harvested in hypotonic buffer containing 50 mM NaCl and the nuclear pellet was collected, washed and subsequently lysed in buffer containing 500 mM NaCl. The soluble fraction was used after centrifugation at 14,000 rpm. 12 µg of cytosolic, 4 µg of total protein and 2 µg of nuclear protein were used for Western blot analysis for β-catenin antibody (BD Biosciences). For TOPFlash luciferase assay, C2C12 cells were seeded to ∼80% confluency overnight followed by transient transfections (FuGENE 6, Roche) in four replicates as described (Ma et al., 2009). The transfection mixture contains 200 ng of Super TOPFlash luciferase reporter (Veeman et al., 2003), together with 50 ng of Renilla luciferase (pRL-TK, Promega) as an internal control. 16 hours after transfection, media was replaced by Wnt3a or control media. Luciferase activity was measured using Dual-Glo luciferase assay system (Promega) 24 hours after treatment, and values were normalized to Renilla luciferase readings.

Data are expressed as means ± s.e.m. Statistical significance between null and controls were tested using Student’s t-test, and one-way ANOVA was used for comparison of gene expression at different days of myogenic differentiation and circadian time points after serum shock.

We thank the Center for Diabetes Research at the Methodist Hospital Research Institute for technical assistance.