In fish, the skeletal muscle of the trunk and the tail derives from somites/myotomes that form in the paraxial mesoderm in a rostrocaudal sequence. The development of the fish myotome begins with the onset of myogenic regulatory factor expression (Weinberg et al., 1996; Delalande and Rescan, 1999) and continues with the formation of a distinct superficial layer of slow muscle fibers covering the bulk of fast muscle fibers located in the deep portion of the myotome. The superficial slow muscle fibers originate from adaxial presomitic cells flanking the notochord that migrate radially to reach the lateral surface of the myotome (Devoto et al., 1996; Stoiber et al., 1998; Rescan et al., 2001). The fast muscle, which is predominant in the fish myotome, forms from the differentiation of lateral presomitic cells that do not move during somite maturation. It has been proposed that the differentiation of fast muscle cells occurs in a mediolateral progression immediately after the differentiating wave of adaxial cell migration (Blagden et al., 1997). A subset of adaxial cells that fails to undergo the lateral migration toward the outermost domain of the myotome remains juxtaposed to the notochord. These cells, known as pioneer cells, express the Engrailed proteins as do the fast fibers near the horizontal myoseptum. Graded exposure to notochord-derived hedgehogs accounts at least in part for this muscle cell diversity (Wolff et al., 2003).
Although previous studies have reported the early temporal sequence of muscle-specific gene activation in developing fish embryos (Xu et al., 2000; Hall et al., 2003), little is known regarding the temporal evolution of gene expression during the maturation and the diversification of fish embryonic muscle cells. In a large-scale, trout 3′ and 5′ends cDNA sequencing project (AGENAE-INRA research programs), we identified several expressed sequence tag (EST) clones encoding putative muscle-specific proteins involved either in the cytoarchitecture (tropomyosins, myosins, troponins, and so on) or the metabolism (muscle aldolase, creatine kinase, enolase, and so on) of the muscle fibers. Using these EST cDNAs, we generated muscle-specific riboprobes and examined by in situ hybridization the developmental expression pattern of the corresponding genes to describe the differentiation and the late maturation of fish muscle fibers. We found that a medial to lateral wave of muscle gene expression spreading laterally throughout the width of the whole myotome marks the onset of muscle differentiation in fish embryos. This mediolateral progression of muscle gene expression is coordinated with the lateral migration of the slow muscle precursors toward the outermost domain of the somite. Subsequently, slow and fast muscle isoforms expressed in the course of the early mediolateral wave of muscle gene activation are progressively down-regulated in maturing fast and slow muscle fiber, respectively. Moreover, the transcription of several muscle genes was found to be initiated only at a late stage of myogenesis.
Muscle-Specific cDNA Clones
In this work, we report the comparative developmental expression pattern of 25 muscle-specific genes selected among the EST cDNA clones produced in the course of the AGENAE program. When compared with other vertebrate proteins, the deduced amino acid sequence of the selected trout EST cDNA clones generally showed high conservation (Table 1). The genes whose developmental expression was examined in this study included genes encoding proteins involved in muscle metabolism (creatine kinase, β enolase, aldolase A, and pyruvate kinase muscle isoforms), genes encoding components of the striated muscle cytoarchitecture (tropomyosins, troponins, myosin light chains, alpha actinin, nebulin, capping proteins such as capZ and tropomodulins, and myomesins, which are filament-linking molecules), genes encoding proteins associated with myofibrillar components (myosin binding protein C and parvalbumin), or genes involved in the membrane linkage of the cytoskeletal filaments (CRP1 and SLIM1/FHL1). We also considered the gene encoding the unconventional myosin X, whose developmental expression pattern is unknown in vertebrates.
Table 1. Summary of the Muscle-specific cDNA Clones Used for In Situ Hybridization Experimentsa
Most homologous cDNA
aa sequence identity
The common myogenic program affecting the slow and fast muscle cells includes the activation of the 13 cDNAs presented in the upper part of the table.
Gallus gallus myosin X
Oreochromis mossambicus muscle type Creatine Kinase
Oncorhynchus mykiss Desmin
Danio rerio Aldolase A
fast Protein C
Homo sapiens fast Myosin Binding Protein-C
Danio rerio Enolase 3 (β, muscle)
Salmo salar fast myotomal muscle Tropomyosin
fast Troponin C
Danio rerio fast skeletal muscle Troponin C
slow Troponin I
Danio rerio Troponin I
Salmo trutta slow myotomal muscle Tropomyosin
Mus musculus Tropomyosin slow
Danio rerio Actinin
Gallus gallus Nebulin
Danio rerio four and a half LIM domains protein 1
slow Myosin Light Chain 1
Caranx delicatissimus Myosin Light Chain 1
Mus musculus Myomesin 1
Mus musculus skeletal muscle myomesin2
Danio rerio skeletal muscle Tropomodulin 4
fast Troponin I
Salmo salar fast myotomal muscle Troponin I
Salmo salar Parvalbumin β
Tetraodon nigroviridis 2 capping protein muscle Z-line
slow Protein C
Homo sapiens slow myosin binding protein-C
slow Troponin T
Salmo trutta slow myotomal muscle Troponin T isoform 1S
Xenopus tropicalis Cysteine Rich protein 1
Danio rerio muscle Pyruvate Kinase
A medial-to-lateral wave of muscle-specific gene activations, including the expression of several slow and fast muscle isoforms, marks the onset of muscle differentiation.
The first event of muscle differentiation in the trout somite was the expression, at approximately the 25-somite stage (stage 12 of Ballard, 1973), of the unconventional myosin X and a variety of muscle-specific contractile protein genes, including slow and fast muscle isoforms (see the list in the upper part of Table 1). In agreement with the rostrocaudal development of the somite, the expression of these muscle genes swept along the embryo from the anterior to the posterior (Fig. 1A–C). Transverse and frontal sections indicated that the expression of all these genes always started in the medial part of the somite where the slow muscle precursors are initially located (Fig. 2A–F) and then progressed laterally in fast muscle cells (Fig. 2G–P). Therefore, we can conclude that, in early trout embryos, a common myogenic program progresses mediolaterally, affecting both slow and fast muscle cells.
The expression patterns of slow MyLC1 and SLIM1/FHL1 reveal an apparent migration of adaxial cells from the notochord to the lateral surface of the myotome.
Around the 30-somite stage (stage 14), adaxial cells initially present in the medial part of the somite specifically started to express the slow myosin light chain l (sMLC1) isoform as well as the Lim protein SLIM1/FHL1 (Fig. 3A,D). As somitogenesis proceeded along an anteroposterior axis, the labeling for these two genes appeared progressively in adaxial cells of the more caudal somites. During the maturation of the somite, the slow MyLC1- and SLIM1/FHL1-positive cells appeared to move in a lateral direction until they formed the outermost domain of the myotomes (Fig. 3B,C,E,F). This apparent lateral migration of the slow MyLC1- and SLIM1-positive medial cells is in agreement with the template for slow muscle development reported in zebrafish.
The lateral migration of the slow myogenic cells is coordinated with the early mediolateral wave of muscle differentiation.
Double in situ hybridization using a MyLC1 probe specific to adaxial cells and probes corresponding to several genes expressed in the course of the early mediolateral activation, namely a fast tropomyosin, a fast troponin C, and two distinct slow tropomyosins, indicated that the expression of these muscle isoforms of the common myogenic program starts in medial slow muscle cells before these cells specifically express the slow-specific MyLC1 gene (Fig. 4A). Double in situ hybridization also revealed that the lateral migration of the slow MyLC1-positive cells coincides with the mediolateral activation of genes of the common myogenic program (Fig. 4B,C).
Slow and fast muscle isoforms of the common myogenic program are progressively down-regulated in maturing fast and slow muscle fibers, respectively.
The expression, during the early stage of myogenesis, of several slow and fast muscle isoforms throughout the width of the trout myotome was unexpected. To learn more about the evolution of the expression pattern of these isoforms, we performed in situ hybridization of the corresponding transcripts in late embryos. We observed that the slow α-tropomyosin (Fig. 5A,B), the slow tropomyosin (Fig. 5C,D), and the slow troponin I transcripts that were present in the whole myotome soon after the end of the segmentation (Fig. 5A,C), were down-regulated, at the eyed stage, in the fast deep muscle to become specifically expressed in the superficial slow fibers (Fig. 5B,D). On the other hand, the merged image of double in situ hybridization with fluorescent markers revealed that the fast tropomyosin (Fig. 6A–F) and the fast troponin C (Fig. 6G–L) transcripts present throughout the whole myotome of eyed-stage embryos (Fig. 6C,I) became undetectable in superficial slow MyLC1-positive fibers around hatching (Fig. 6F,L). Taken together, these observations show that the maturation of the fish myotome involves the down-regulation of the fast and slow muscle isoforms of the common myogenic program in slow and fast muscle cells, respectively.
Late Expression of Muscle-Specific Genes
Several muscle-specific genes were found to be transcribed only at a late stage of myogenesis. For example, the slow myosin binding protein C (Fig. 7A), the slow troponin T (Fig. 7B), the fast troponin I (Fig. 7C), the myomesin 2, and the tropomodulin transcripts were only detected around the eyed stage (stage 20). Noticeably, the late expression of parvalbumin in eyed-embryos was found to be restricted to the ventral fast fibers of the myotome (Fig. 7D). Transcripts for alpha2 capZ and pyruvate kinase muscle isoform were not detected at the embryonic stages studied, although readily detected in muscle fibers from fry (not shown) and transcripts for the Lim domain containing protein CRP1 and myomesin 1 were never detected in either embryos or larvae. Taken together, these observations indicate that the differentiation of embryonic fast and slow fibers in the developing myotome of fish is not completed with the lateral migration of slow cells toward the outermost part of the somite but requires subsequent muscle gene activation. Moreover, the restricted expression of parvalbumin to the ventral fast myofibers indicates that further fiber diversification takes place in the developing fish myotome.
In this study, we have described some aspects of the developmental gene activations associated with fish myotome differentiation. We have shown that a mediolateral wave of gene activations, including the expression of several slow and fast muscle isoforms, spreads throughout the whole width of the myotome. This mediolateral progression of muscle gene expression coincides with the lateral migration of the slow muscle precursors toward the outermost domain of the somite. These early events are followed by sequential gene activations and repressions in fast and slow muscle cells.
Recently, our understanding of teleost muscle fiber origins has been expanded by the striking work of Devoto et al. (1996). This work convincingly demonstrated that the slow superficial fibers originate from a group of adaxial cells next to the notochord. The adaxial cells, which are the first to express the myogenic regulatory factor MyoD, undergo radial migration to form a monolayer at the lateral surface of the myotome, whereas other cells differentiate into fast muscle. In trout embryos, we observed that somitic medial cells, which are specifically labeled with slow MyLC1 and SLIM1/FHL antisense riboprobes exhibit an apparent migration toward the outermost domain of the somite. Thus, our observations are in agreement with the template for slow muscle development in zebrafish. On the basis of immunolocalization studies with slow and fast MyHC isoform-specific antibodies, Bladgen et al. (1997) proposed that the differentiation and the migration of slow cells precedes the expression of fast characteristics by the rest of the cells of the myotome. Somewhat differently to these observations but in agreement with our previous in situ hybridization of fast MyHC (Rescan et al., 2001), we show here that a mediolateral wave of muscle gene activation that included the expression of several fast and slow muscle isoforms was initiated in adaxial cells and then progressively spread laterally throughout the width of the myotome. Therefore, a common myogenic program is shared by slow and fast muscle cells meanwhile genes such as the slow MyLC1 and the SLIM1/FHL1 are transcribed only in migrating slow cells. On the basis of previous studies showing that notochord-derived hedgehog proteins are essential for the differentiation of the zebrafish slow muscle fibers (Blagden et al., 1997; Du et al., 1997), it can be speculated that the common myogenic program taking place in all muscle cells does not depend on hedgehog activity, whereas the expression of the early slow-specific markers, including the slow MyLC1, SLIM1/FHL1, and the slow MyHC (Rescan et al., 2001), is triggered by hedgehog signalling. The discrepancies that have been observed between the zebrafish and the trout models regarding the early expression of fast and slow-specific markers in the developing fish myotome may result from the fact that a limited repertoire of antibodies raised against avian myosins was used for the characterization of the slow and fast muscle differentiation in the zebrafish (Blagden et al., 1997), whereas in this study, we took advantage of a large repertoire of fish muscle-specific riboprobes. In an attempt to conciliate the two fish models, it would be of interest to re-examine muscle differentiation in the zebrafish using orthologous probes similar to those used in this study. On the other hand, we cannot formally exclude the possibility that some differences do exist between fish species regarding myotome differentiation. Nevertheless, it is interesting to note that, in the trout as well as in zebrafish, the adaxial slow muscle cells do differentiate before the fast ones in so far as they were found, in both species, to be the first to express muscle-specific genes. That the slow adaxial cells are the first to differentiate in the developing fish somite is not surprising given that they are the first to express the basic helix–loop–helix myogenic regulator MyoD and myogenin (Weinberg et al., 1996; Delalande and Rescan, 1999).
Our observations showing a precocious expression of several slow and fast muscle isoforms in both slow and fast muscle cells may appear surprising. However, our paradoxical observations can be explained if we consider the late expression of these isoforms in fish embryos. Indeed, using double in situ hybridization with fluorescent markers, we found that the fast muscle isoforms expressed in the whole myotome as a result of the mediolateral wave of muscle gene activation, are no longer expressed in the superficial slow fibers of hatching embryos. On the other hand, around the eyed stage, the slow muscle isoforms of the common myogenic program are down-regulated in the deep fast fibers to become restrictedly expressed in the superficial slow fibers. Therefore, the slow and fast differentiation of muscle cells in fish embryo involves developmentally regulated gene repressions that follows the activation of the common myogenic program.
In addition to gene repression taking place in late stage myogenesis, a late activation of several muscle-specific genes was also found to occur in the maturing fish myotome. Thus, we have observed that genes encoding slow troponin T, fast troponin I, tropomodulin, and myomesin 2 were transcribed only around the eyed stage. This finding reinforces the notion that muscle differentiation and maturation in fish is not completed with the lateral migration of the slow adaxial cells toward the outermost part of the myotome but requires subsequent gene repression and activation in both slow and fast fibers.
As reported in zebrafish (Xu et al., 2000), parvalbumin transcription in the trout embryo was found to be initiated at late stages of myogenesis. Somewhat different to what has been reported in zebrafish, we observed that parvalbumin expression was confined to ventral subdomains of the myotome. The restricted expression of parvalbumin in certain myofibers suggests that a late diversification of fiber phenotype occurs in the developing fish myotome. This raises the question of the mechanisms that regulate a complex series of genes expression within individual muscle fibers. Extrapolating the work by Wolff et al. (2003) on hedgehog activity and muscle cell identities, we can speculate that distinct levels and timing of morphogen activity within the myotome account for muscle cell maturation and diversification, also functional innervation may be necessary for the initiation of expression of several muscle genes, as shown in avian muscle fibers (Sacks et al., 2003).
Among the genes of interest specifically expressed in the developing fish embryo myotome are myosin X and skeletal muscle LIM protein 1 (SLIM1/FHL1). Myosin X is an unconventional myosin that associates with the cytoplasmic domains of integrins and actins and is involved in cell adhesive structure (Homma et al., 2001). The developmental expression pattern of Myosin X is unknown in vertebrates. Here, we show an early and specific expression of myosin X in the fish myotome, suggesting that myosin X is probably linked to key events in early vertebrate myogenesis. SLIM1/FHL1, a LIM domain protein, has been shown to regulate integrin-mediated myoblast spreading and migration (Robinson et al., 2003). In developing mouse embryos, SLIM1/FHL1 transcript localizes to the cardiac outflow tract of the developing heart and to somites (Chu et al., 2000). In developing fish embryos, we observed a specific labeling restricted to migrating slow cells. This surprising observation raises the question of the functional role of the SLIM1/FHL1 gene in regulating the differentiation and/or the migration of the slow myogenic cells in fish embryos. To answer this question, it would be of interest to examine the phenotype of fish embryos injected with SLIM1/FHFL1 antisense morpholino nucleotide.
In conclusion, in situ hybridization of a large repertoire of muscle-specific transcripts shows that a myogenic program common to both slow and fast muscle cells is activated in a mediolateral progression paralleling the lateral migration of slow MyLC1- and SLIM1/FHL1-positive slow muscle cells. During the maturation of the fish myotome, slow and fast muscle isoforms expressed in the course of the common myogenic program activation become down-regulated in fast and slow muscle cells, respectively. Although, hedgehog proteins undoubtedly have a major role in the early patterning of the fish myotome (Du et al., 1997; Blagden et al., 1997; Wolff et al., 2003), the detailed mechanisms regulating this complex sequence of gene expression in the maturing fish myotome in large part remain to be elucidated.
Whole-Mount In Situ Hybridization
Muscle-specific cDNAs have been identified from a large-scale, rainbow trout 3′ and 5′ sequencing project (AGENAE research program). Digoxigenin-labeled antisense RNA probes were synthesized from a polymerase chain reaction–amplified template using T3 RNA polymerase. The embryos were dechorionated with fine forceps and fixed overnight at 4°C in paraformaldehyde in phosphate buffered saline (PBS). Specimens were dehydrated and stored in methanol at −20°C. After rehydration in graded methanol/PBS baths, embryos were processed according to established procedures (Joly et al., 1993), with minor modifications. Depending on the embryonic stage, different times, temperatures, and concentrations were chosen for proteinase K treatment. Double, whole-mount in situ hybridizations were performed as previously described (Rescan et al., 2001).
Double In Situ Hybridization of Sections With Fluorescent Markers
Double in situ hybridizations were performed on transverse sections of rainbow trout embryos according to Gabillard et al. (2003), with minor modifications. The green fluorescence was obtained, once the hybridization was performed, by incubation of sections with mouse anti-digoxigenin antibody (Roche) followed by an incubation with Alexa Fluor 488–conjugated rabbit-derived anti-mouse IgG antibodies (Molecular probes). The red fluorescence was obtained by incubation of sections with goat anti-fluorescein antibodies (Vector) followed by an incubation with Alexa Fluor 594–conjugated donkey-derived anti-goat IgG antibodies (Molecular probes). Fluorescein isothiocyanate (revealing Alexa 488) and Texas Red (revealing Alexa 594) filters were then used for confocal microscopy. The confocal microscope system used in this study was a Leica TCS NT equipped with Kr/Ar laser and mounted on a Leica DMB microscope
We thank Roselyne Primault for her help in the observation of sections by confocal microscopy (Service de Microscopie Electronique, Faculté de Médecine, Université de Rennes I). We thank Dr. G.S. Butler-Browne and Dr. W. Stoiber for critical reading of this manuscript.