Regulation of Slow and Fast Muscle Myofibrillogenesis by Wnt/β-Catenin and Myostatin Signaling

Deviation from proper muscle development or homeostasis results in various myopathic conditions. Employing genetic as well as chemical intervention, we provide evidence that a tight regulation of Wnt/β-catenin signaling is essential for muscle fiber growth and maintenance. In zebrafish embryos, gain-of-Wnt/β-catenin function results in unscheduled muscle progenitor proliferation, leading to slow and fast muscle hypertrophy accompanied by fast muscle degeneration. The effects of Wnt/β-catenin signaling on fast muscle hypertrophy were rescued by misexpression of Myostatin or p21CIP/WAF, establishing an in vivo regulation of myofibrillogenesis by Wnt/β-catenin signaling and Myostatin. Epistatic analyses suggest a possible genetic interaction between Wnt/β-catenin and Myostatin in regulation of slow and fast twitch muscle myofibrillogenesis.


Introduction
Understanding muscle development is crucial for generating novel regenerative therapies for muscle diseases and treating muscle injuries. Extensive research has contributed to the current understanding of various aspects of somitogenesis and myogenesis. The periodicity of rostral-caudal somite formation [1] as well as their differentiation into the axial skeleton, skeletal muscle and dorsal dermis are similar in all vertebrates [2]. Furthermore, the zebrafish determinate muscle growth by hyperplasia, the increase in muscle fiber number, and by hypertrophy, the increase in muscle fiber size, are comparable to mammalian muscle growth making it a suitable model system to study myofibrillogenesis and various myopathies [3][4][5].
Wnt/b-catenin pathway plays a crucial role in early somitogenesis and myogenesis in birds [6,7], mice [8][9][10] as well as in zebrafish [11] by affecting skeletal muscle development at several levels, including mesodermal patterning, segmentation clock and myoblast differentiation [12,13]. The Wnt/b-catenin signaling regulates Lef/Tcf-mediated transcription of downstream target genes via the transcriptional coactivator b-catenin [14]. In the absence of Wnt ligand, b-catenin is targeted for proteosomal degradation by a ''destruction complex'' comprising of CK1, GSK3b, Axin1 and Apc1. The fine balance between proliferation and differentiation required for proper development and growth of the myotome depends on signaling cues originating from tissues surrounding the somites [15,16], including Wnt ligands. Cumulative evidence implicates Hedgehog and Fgf8 signaling in specification and differentiation of slow and fast twitch muscle fibers respectively, during the first wave of myogenesis [17][18][19][20]. Although recent work has shown the role of Hedgehog signaling in differentiation of a subset of secondary slow twitch muscle fibers [19], the precise molecular mechanism underlying specification and maintenance of secondary fast twitch muscle fibers as well as the Hedgehog independent slow twitch muscle fibers remains to be elucidated.
This study shows that upward deviation from the tightly controlled physiological level of Wnt/b-catenin activity by genetic and chemical intervention in zebrafish embryos leads to compromised growth and maintenance of slow and fast muscle fibers. This phenotype derives from hyperproliferation of the Pax3/7+ pre-myogenic precursors. Hence, misexpression of p21 CIP/WAF or mstn in the embryos with gain-of-Wnt/b-catenin function restores the integrity as well as morphology of the fast muscle fibers. We further discuss the possibility that this tight and opposing regulation of myofibrillogenesis by Wnt/b-catenin and Myostatin in zebrafish could operate through their genetic interaction.

Wnt/b-catenin hyperactivity causes loss of somites and aberrant muscle fibers
Wnt/b-catenin gradient has been shown to be important for somite segmentation [9]. Importantly, it has been suggested that Wnt/b-catenin is downregulated in the somite following skeletal muscle differentiation [21]. We investigated the expression of Wnt/b-catenin reporter TOPdGFP [22] during post-segmentation corresponding to the second wave of myogenesis. Consistent with previous studies [21], we observed only faint expression in the trunk and tail of wild-types at 28 hours post-fertilization (hpf) (Fig. 1A). Next, by employing the homozygous compound zebrafish mutants of axin1 [23,24] and apc1 [25] (hereafter referred to as axin1/apc1), we investigated whether Wnt/b-catenin is hyperactivated in the somites. Indeed, there was strong ectopic expression of the TOPdGFP reporter in axin1/apc1 mutants in a rostro-caudally rising gradient (Fig. 1A). This corresponded to enhanced expression of the Wnt transcription factor and its direct target gene lef1 [26] throughout the somites (Fig. 1B), showing that Wnt pathway is overactivated in the somites of axin1/apc1 mutants.
At completion of segmentation and the first wave of myogenesis at 24 hpf, axin1/apc1 embryos had a normal number of somites, size of somites (Fig. S1), as well as normal muscle fiber formation (data not shown). The earliest clear somite phenotype in axin1/ apc1 mutants was at 36 hpf, with a slight decrease in somite number from approximately 31 to 29 (Fig. 1, C and D). Strikingly, at 54 hpf, there was a severe tail truncation due to loss of approximately 10 somites (Fig. 1, C and D). The formation of normal somites at 24 hpf (Fig. S1) eliminates somite fusion and abnormal initiation of segmentation as an underlying cause of somite loss. Hence, this late and gradual somite-loss strongly suggests that the underlying mechanism does not entail a defect in somite induction and/or patterning.
Next, we examined whether upregulation of the Wnt/b-catenin signaling would affect the fast-and slow-twitch muscles that make up the myotome. The slow muscle fibers appeared to be hypertrophic, as well as hyperplastic with an additional 2-4 fibers per somite (n = 4) ( Fig. 2A). Strikingly, the fast muscle fibers were disorganized, with some muscle fibers detaching from the vertical myoseptum, forming small lesions, while becoming hypertrophic only at 54 hpf ( Fig. 2A). Confirming the distinct effects of hyperactive Wnt/b-catenin on slow versus fast muscle fibers, quantification by RT-qPCR of myosin heavy chain specific for slow or fast twitch muscle fibers revealed an increase in fast muscle myosin at 54 hpf (Fig. 2B). However, there is no significant difference observed for slow muscle myosin (Fig. 2B).
We confirmed the specific role of Wnt/b-catenin by employing a chemical activator of the Wnt/b-catenin pathway, lithium chloride (LiCl). The wild-type embryos treated with LiCl prior to completion of somitogenesis, at tailbud and mid-somitogenesis were truncated and curled albeit no detached muscle fibers were present (Fig. 2C, early treatment). In contrast, the embryos treated after 24 hpf showed severe muscle fiber detachment and hypertrophy (Fig. 2C, mid-treatment and late-treatment), resembling axin1/apc1 mutants. Hence, the muscle fiber defect in the axin1/apc1 mutants is likely also caused by late Wnt/b-catenin hyperactivation.
The fast muscle fiber degeneration beginning at 36 hpf, corresponded with gradual increase in apoptosis (Fig. 3A), with apoptosis occurring within the myotome (Fig. 3B) and along the vertical myoseptum (Fig. 3, A and C).

Hyperactive Wnt/b-catenin drives muscle progenitors into unscheduled proliferation
The ability of Wnts to enhance proliferation in the dermomyotome [27] led us to hypothesize that unscheduled proliferation in the somites might lead to muscle hypertrophy in axin1/apc1 mutants. While at 16 hpf there was no difference in proliferation between mutants and wildtypes (data not shown), from 28 hpf onwards, BrdU pulse experiments identified a sharp increase in number of cells in S-phase (Fig. 4A), which was confirmed by increased labeling of phosphohistone H3 (PH 3)+ mitotic cells (data not shown) and their quantification by FACS (Fig. S2B). To investigate whether this unscheduled increased proliferation caused muscle hypertrophy, we partially inhibited cell division with a combination of aphidicolin [28] and hydroxyurea (HUA) [29] from 24 hpf until fixation at 54 hpf. Strikingly, the fast muscle hypertrophy (Fig. 4B; compare to Fig. 2A, right panels) and degeneration (Fig. S3A) as well as the number of somites were partially rescued (Fig. 4B) confirming that hyperproliferation leads to the fast muscle hypertrophy and degeneration.
Pre-myogenic embryonic muscle progenitor cells expressing Pax3/7 transcription factors give rise to myoD+ myoblasts, myogenin+ myocytes and myotubes, that terminally differentiate into muscle fibers [30]. To determine whether these Pax3/7+ progenitors were stimulated by Wnt/b-catenin to hyper-proliferate, we performed co-labeling of PH 3 and Pax3/7 in the axin1/ apc1 mutants. There was a significant increase in co-localization of PH 3+ mitotic nuclei and Pax3/7+ nuclei in the axin1/apc1 mutants (Fig. 4C) as compared to wild-types, suggesting that Wnt/ b-catenin drives unscheduled proliferation of pre-myogenic progenitors. Surprisingly, the absolute number of Pax3/7+ cells was unaltered in axin1/apc1 mutants (data not shown), suggesting that the newly born progeny of dividing Pax3/7+ cells was not maintained in a Pax3/7+ precursor state, but instead was instructed to differentiate. Hence, we examined myogenic differentiation in the mutants. Consistent with the unperturbed Pax3/7+ muscle progenitors at 16 hpf, myoD was unaltered in axin1/apc1 mutants during initial myogenesis (data not shown). Importantly, later in development, its timely downregulation failed and its expression was sustained (Fig. S3B). Consistent with prolonged myoD expression in mutants, myogenin expression was also extended in axin1/apc1 (Fig. S3C). As myoD+ myoblasts are known to proliferate it is possible that the ectopic and extended myoblast maintenance also contributes to the hyperproliferation in the mutants. To test this, we performed co-labeling of anti-MyoD with anti-BrdU antibody (with BrdU incorporation chased for 2 hours). We observed substantial increase in MyoD+ cells in axin1/apc1 embryos (Fig. S3D), confirming the increased myoD RNA expression in the mutants. However, we do not observe an increase in proliferating (BrdU+) myoD+ cells (data not shown). The data is consistent with a positive role of Wnt/b-catenin signaling in driving myogenic differentiation [13].
Myotomal proliferation and hypertrophy in axin/apc mutants are counteracted by misexpression of Mstn and its downstream target p21 CIP/WAF Reportedly, Wnt/b-catenin through its direct target c-myc, can downregulate p21 CIP/WAF (also known as cyclin-dependent kinase inhibitor 1A) [31]. We hypothesized that sustained myotomal proliferation in axin1/apc1 mutants works through Wnt/b-cateninmediated inhibition of p21 CIP/WAF . We tested the hypothesis by examining the capacity of p21 CIP/WAF mRNA injected into axin1/ apc1 mutant to rescue muscle fiber phenotype. Employing misexpression with p21 CIP/WAF mRNA concentration that only subtly affected the wildtypes, muscle fiber hypertrophy was rescued in injected axin1/apc1 embryos (Fig. 5A), suggesting that muscle fiber degeneration is due to hyperproliferation caused by failure of timely p21 CIP/WAF -dependent cell cycle exit. However, we cannot exclude the possibility that forced cell cycle exit mediated by p21 CIP/WAF misexpression in itself, and independently of its postulated positioning downstream of the Wnt pathway, may have brought about the rescue. p21 CIP/WAF has been implicated in muscle differentiation as the downstream target of Mstn, a TGF-b family member. Mstn is a key negative regulator of muscle growth that promotes terminal differentiation of embryonic muscle progenitors through the activation of p21 CIP/WAF [32]. Decreased levels of Mstn, context dependently lead to muscle hypertrophy [32]. Based on the hypertrophic muscle phenotype in the compound mutants and upon late LiCl treatment of wildtype embryos, we hypothesize that p21 CIP/WAF may be epistatic to Mstn and that overactive Wnt pathway may through inhibition of Mstn downregulate p21 CIP/WAF . We tested this possibility by performing rescue of muscle fiber degeneration in the axin1/apc1 mutants via misexpression of mstn mRNA. While at this particular concentration of mstn, wild-types showed slightly hypotrophic muscle fibers, the axin1/apc1 embryos showed a partial rescue of the hypertrophic muscle fibers (Fig. 5A). Consistently, misexpression of mstn rescued the number of somites (n = 8) and length of mutant embryos (Fig. 5, B and C). Importantly, cell survival in axin1/apc1 mutants was also rescued (Fig. 5C, right panels).
To further explore opposing effects of Wnt/b-catenin and Myostatin on phenotypic aspects of myofibrillogenesis we asked whether morpholino (MO)-mediated knock-down of Mstn would result in a similar hypertrophic phenotype as hyperactive Wnt/bcatenin in zebrafish embryos. Injection of 5 ng mstn MO resulted in hypertrophic muscle fibers, while injection of 2 ng Lef1 MO resulted in hypotrophic muscle fibers (Fig. 6). We further coinjected 2 ng Lef1 MO with 10 ng Mstn MO, and asked whether muscle fibers would be hypertrophic or hypotrophic. The slow and fast muscle fibers appear hypertrophic (Fig. 6). As a loss of Mstn signaling would be expected to lead to hyperproliferation, and loss of Wnt/b-catenin signaling to reduction of proliferation of premyogenic precursors, the data suggests that loss of Mstn is dominant over the loss of Wnt/b-catenin signaling.
As expression profiling in Mstn loss-of-function (LOF) identified modulation of Wnt-pathway components [33], we examined for possible genetic interaction between the two pathways, by using gain and loss of Wnt/b-catenin signaling. To establish a suitable genetic means for analysis of mstn mRNA upon loss of Wnt/bcatenin function, we first tested whether morpholino (MO)mediated knock-down of Lef1 [22], which is upregulated in axin1/apc1 mutants (Fig. 1B) would rescue their aberrant somitogenesis. Knockdown of Lef1 with 2 ng MO in wild-types resulted in loss of a number of somites (59%, n = 54) at 54 hpf (Fig. 7A), suggesting that Lef1 is required for normal somitogenesis. Notably, in 50% of Lef1-MO-injected axin1/apc1 mutants, the normal number of somites was restored, establishing a mechanistic link between Lef1 hyperactivity and somite loss (Fig. 7A). To test whether GOF and LOF of Wnt/b-catenin signaling affected the levels of mstn RNA, we employed quantitative real-time PCR to quantify expression of mstn mRNA in axin1/apc1 mutants versus Lef1-morphants at different developmental times. The data showed that reduced Wnt/b-catenin signaling in Lef1 morphants resulted in upregulation of mstn mRNA (Fig. 7B, C), whereas hyperactivity of the pathway in the axin1/apc1 mutants led to a slight mstn downregulation at 36 hpf (Fig. 7B), and a more significant downregulation at 54 hpf (Fig. 7C). We then analyzed the level of Myostatin protein at 54 hpf. Predictably and in agreement with the qPCR data, the level of processed mature monomeric Myostatin protein was significantly induced in Lef1-morphants suggesting that reduced Wnt signaling through Lef1 leads to de-repression of Myostatin. Surprisingly, in axin1/apc1 embryos, the level of Myostatin protein is similar to that of wild-type embryos, suggesting that subtle increase of myostatin mRNA in the Wnt/b-catenin GOF context, does not translate into an increase in the stable protein (Fig. 7D). Why and how this mechanistically takes place need further investigation. As lef1 was upregulated in the somites of axin/ apc mutants at 30 hpf, we tested whether misexpression of mstn would alter this Wnt/b-catenin downstream target gene. Misexpression of mstn downregulated ectopic lef1 in the mutants as well as slightly downregulating lef1 expression in wildtype siblings (Fig. 7E) showing genetic interaction between Mstn and Wnt/b-catenin signalling and probably underlying the mechanism of muscle fiber rescue. To corroborate possible interaction between the two pathways we performed in silico analysis to identify putative TCFbinding elements (TBE) in myostatin promoter, as have been identified in the promoters of Wnt target genes [34]. Indeed, we found 3 putative TBE (NNCAAAG) within a 2.8 kb sequence upstream of the myostatin gene at positions -2790, -2389 and -1578 (data not shown) opening up the possibility of a direct interaction between Wnt and Myostatin.
Together, these data suggest an existence of a genetic interaction between Wnt/b-catenin and Mstn in myofibrillogenesis possibly existing as a negative feedback loop. We propose a model describing regulation of fast muscle fiber growth and maintenance during secondary myofibrillogenesis with respect to Mstn and Wnt signaling (Fig. S4).

Discussion
Most zebrafish mutants such as after eight (aei) and deadly seven (des) that harbor mutations in genes of the Delta-Notch pathway [35] display a reduction in somite numbers secondary to patterning. Together with Delta-Notch signaling, Wnt/b-catenin signaling contributes to somite clock establishment, mediating early somite patterning [36,37]. The unique phenotype of the axin1/apc1 mutant is characterized by normal somite patterning followed by a gradual loss of approximately 10 most caudal somites. Our data show that this somite loss, resulting from sustained, ectopic and hyperactive Wnt/b-catenin, is secondary to cell fate alterations, ultimately leading to hypertrophic and degenerative fast muscle fibers. The hyperactive Wnt signal is transduced mainly through Lef1 and leads to an imbalance between proliferation and differentiation in the myotome. The late activation of Wnt/b-catenin in wild-type embryos by treatment with LiCl phenocopies the fast muscle fiber hypertrophy and degeneration observed in axin1/apc1 embryos. This corroborates the notion that the mutant muscle fiber phenotype arises independently and is subsequent to normal somite establishment, providing an opportunity to decouple roles of Wnt/b-catenin in myofibrillogenesis versus somite patterning.
High Wnt/b-catenin activity is required for somite patterning, as well as for proliferation i.e. expansion of the Pax3/7+ premyogenic progenitor compartment. Pax3 and Pax7 transcription factors, that mark the pre-myogenic progenitors in the developing dermomyotome [38] and satellite cells in the adult muscle, positively regulate cell proliferation [39,40]. Wnt/b-catenin has been implicated in induction of Pax3/7+ precursors in cell culture systems [41,42]. A high Wnt/b-catenin activity is probably also required for proliferation of differentiating myoblasts. Our data suggest that the sustained upregulation of myoD and myogenin in axin1/apc1 reflects propensity towards differentiation of hyperproliferating pre-myogenic Pax3/7+ precursors being consistent with the known role of Wnt/b-catenin signaling in myogenic differentiation in several systems [43,44].
We show that hyperproliferative fast muscle fibers underlie the fast muscle fiber degeneration in the axin1/apc1 mutants, as partial inhibition of proliferation restored to near normal impaired cell survival and fast muscle fiber hypertrophy. The hypertrophic muscle fibers in axin1/apc1 embryos indicate that a myotomal cell population(s) hyperproliferates and differentiates, thus resulting in an increase in the mass of the muscle fiber. Simultaneously, conflicting instructions to myoblasts to undergo premature differentiation likely leads to apoptosis. To our knowledge, Wnt/ b-catenin has as yet not been implicated in muscle hypertrophy in vivo. Ex vivo studies of the adult muscle reveal the synergistic effect of insulin and Wnt/b-catenin in causing myotube hypertrophy [45]. In addition, it has been shown that Wnt/b-catenin is upregulated in overload-induced hypertrophy of the adult muscle [46]. As a conserved transcriptional hierarchy is thought to regulate the myogenic differentiation in embryos and adults [47], these reported data may be extrapolated to the developing myotome.
Several in vitro and in vivo studies showed that Mstn overexpression prevents proliferation and differentiation of muscle precursors by inducing expression of the cell cycle inhibitor p21 CIP/WAF , while endowing muscle progenitors with competence to respond to signals favoring muscle differentiation [32]. We showed that simultaneous knockdown of Lef1 and Myostatin, resulted in a hypertrophied muscle fiber, similar to knockdown of Myostatin. This suggests that Wnt/b-catenin signaling could lie upstream of the Mstn regulatory pathway, as knockdown of Lef1 is unable to rescue the myofiber phenotype (Fig. 6), while misexpression of mstn partially rescues the fast muscle hypertrophy in axin1/apc1 embryos (Fig. 5A). Thus, Wnt/b-catenin might mediate sustained proliferation of muscle progenitors by repressing mstn. However, there is also the possibility that the rescue of myofiber growth is non-specific and Myostatin might work independently of and/or in parallel with Wnt/b-catenin signaling in regulating myoblasts proliferation and differentiation. We favor the possibility of a Wnt/b-catenin-Mstn negative feedback loop, as our experimental evidence points towards a specific interaction, direct or indirect, between Wnt/b-catenin signaling and Myostatin as follows: (1) We observe an upregulation of mstn RNA transcripts and protein upon Lef1 knock-down which may reflect release from repression of mstn by Wnt/b-catenin (2) The downregulation of lef1 mRNA expression upon mstn misexpression in axin1/apc1 and wildtype embryos suggests a negative feedback loop between Wnt/ b-catenin and mstn, likely reflecting the mechanism that underlies phenotype-rescuing capacity of Lef1; (3) The identification of 3 putative TBE within a 2.8 kb region upstream of the Myostatin ATG start site opens up a possibility of a molecular interaction between Mstn and Wnt/b-catenin signaling. However, Wnt/bcatenin could also mediate repression of Mstn indirectly, through induction of its direct target follistatin [48] that is a known negative regulator of Mstn [49]. Whether and how this genetic hierarchy regulating myofibrillogenesis translates into direct molecular interactions is an important avenue for further research.
Unlike mice expressing dominant negative Mstn, which equally affects both fast and slow muscle fibers, the axin1/apc1 embryo exhibits different phenotypes with both slow and fast muscle fibers. Although hypertrophy and hyperplasia is observed in the slow muscle fibers at 36 hpf, there is a reduction in the total amount of slow muscle myosin RNA. We speculate that the lack of quantitative differences observed in the slow muscle fibers could be due to the fact that slow muscle fibers only make up a small portion of the myotome. Therefore, a small increase of slow muscle myosin is not quantifiable by qPCR. On the contrary, the fast muscle fibers exhibit muscle fiber degeneration and disorganization at 36 hpf, and at 54 hpf, they appear hypertrophic ( Fig. 2A). This is confirmed by a significant increase of the total amount of fast muscle myosin at 54 hpf (Fig. 2B). The fast muscle hypertrophy is likely to reflect a compensatory response to decreased muscle stability. Significantly, it is only the fast muscle fibers that degenerate in response to hyperactive Wnt/b-catenin signaling even prior to overt hypertrophy. In agreement to the upregulation of fast muscle myosin in axin1/apc1 embryos, it has recently been shown in cattle that knock-out mutations in myostatin result in preferential downregulation of fast 26 myosin heavy chain [50]. Consistently, mstn, which we showed is affected by Wnt/b-catenin pathway is predominantly found in fast twitch muscle [51]. This study opens up a prospect to unravel the poorly understood difference in regulation of maintenance and growth of secondary slow versus fast muscle fibers.
Although our work showed that Mstn negatively regulates Wnt/ b-catenin, it is very likely that there is an involvement of other signals that mediate timely and dosage-regulated restriction of the Wnt/b-catenin pathway, thereby safeguarding myofibrillogenesis and regulated muscle growth. The pathogenetic mechanism of the muscle hypertrophy in muscle degenerative diseases is still unclear. Our data, implicating a possible role of Wnt/b-catenin signaling in interaction with Mstn and p21 CIP/WAF , which have been shown to be important in muscle diseases, might pave a way to approaching muscle diseases from a novel angle.

Zebrafish embryos
Zebrafish embryos were raised and staged as previously described [52]. apc CA50a/CA50a is a lethal recessive zygotic mutation identified in a three generation forward mutagenesis screen [25]  according to standard mutagenesis protocol. axin/mbl(tm13) is a recessive lethal zygotic mutant obtained in the large scale Tubingen screen [24]. axin1/apc1 compound mutants were generated from crossing apc CA50a/CA50a with axin/mbl(tm13).

Phalloidin staining
Phalloidin-TRITC (Sigma) staining (1:50) was performed at room temperature overnight. Muscle fibers were analyzed at the level of the yolk extension where there is minimal muscle degeneration, as caudal to the yolk extension there is massive apoptosis.

Microinjection of mRNAs and morpholinos (MO)
Morpholino antisense oligonucleotides were obtained from Gene Tool (Philomath, OR): zflef1 (ATG) 59-CTCCTCCACCT-GACAACTGCGGCAT-39 [22] and zMstn (ATG) 59-TGCATGTTCCAAGGCGTGCTAAAGG-39. We validated effectiveness of Mstn morpholino by showing its capacity to knockdown the protein (Fig. 7D). Capped synthetic mRNA was prepared from pCS2+ constructs encoding zebrafish mstn (gift from L.D. Valle) or human p21 CIP/WAF (gift from C.J. Weijer) using the mMessage mMachine kit (Ambion), and injected into one-cell stage embryos using a microinjector (World Precision Instruments). A concentration range of 2.5-100 pg of mRNA was injected into one-cell stage embryos to test for viability and effect, and the concentration which had only a subtle effect on wild-type embryos was selected. For axin1/apc1 rescue experiments, 5 pg of mstn or p21 CIP/WAF mRNA was used.

Cell quantification and imaging
Fluorescent labelings were imaged using a Leica TCS SPE confocal microscope. For each set of experiments, all laser and software settings were standardized. Images from each embryo were cropped in Volocity (Improvision) to exclude the neural tube. Cell counts in the somites were done manually from a z-stack of the whole somite. For each set of experiments, cells were counted and imaged at the first four somites of the yolk extension, unless stated otherwise. For quantification of Pax3/7+ pre-myogenic progenitor cells, only weakly labeled Pax3/7+ nuclei were counted as previously described [38]. Digital pictures of WISH embryos were obtained using the Zeiss Axioplan Stereomicroscope (comparable available microscope is Zeiss Axio Imager) equipped with a Leica digital camera and were adjusted for brightness and contrast using Adobe Photoshop 7.0.

Lithium chloride treatment
LiCl treatment (0.3 M) was repeated twice on the same clutch of embryos for each of the 3 developmental intervals: (1) Early: LiCl treatment pulse for 40 minutes at tailbud and again at 16 hpf, (2) Mid: LiCl treatment pulse for 40 minutes at 16 hpf and again at 24 hpf, and (3) Late: LiCl treatment pulse at 24 hpf and again at 30 hpf. Embryos were washed 3 times in between treatments. Upon treatments, embryos were fixed at 36 hpf, and stained with Phalloidin to visualize all muscle fibers.

HUA treatment
Embryos were cultured in both 75 mM aphidicolin with 0.25% DMSO (Sigma-Aldrich) and 20 mM hydroxyurea (Sigma-Aldrich) from 24 hpf to 54 hpf. Embryos were then fixed for further experiments.

BrdU labeling
For BrdU labeling experiments, embryos (16 hpf, 28 hpf, 36 hpf) were dechorionated and placed in 10 mM BrdU with 15% DMSO on ice for 1 hour. After pulsing, embryos were washed in embryo medium several times and incubated at 28uC for 12 hours. Embryos were then fixed with 4% PFA and immunohistochemistry was performed as above, with incubation in 2 N HCl for 1 hour prior to blocking.

Statistical analysis
Shapiro-Wilk normality test was performed with SPSS 16.0. All data followed normal distribution, with the exception of wild-type DMSO controls in Fig. 3d. Unpaired two-tailed student's t-test was performed using SPSS 16.0. For wild-type DMSO controls in Fig. 3d, where no normal distribution was observed, nonparametric Mann-Whitney test was used. All significant differences (p,0.05) are marked with an asterisk (*) and highly significant differences (p,0.005) are marked with two asterisks (**). All bars in graphs depict mean values with error bars depicting standard deviations.