Temperature and the expression of myogenic regulatory factors (MRFs) and myosin heavy chain isoforms during embryogenesis in the common carp Cyprinus carpio L.

The myogenic regulatory factors (MRFs) are a family of basic helix-loop-helix (bHLH) transcription factors essential to the specification and determination of the muscle cell lineage. The four members of this protein family, MyoD, Myf-5, myogenin and MRF4, are characterised by their ability to induce myogenic conversion in a variety of cell types, including fibroblasts,neurons, adipocytes, chondrocytes and melanocytes (reviewed by Edmondson and Olsen, 1993; Arnold and Braun, 2000). The bHLH domain is central to the role of transcriptional activation and is highly conserved, with the four proteins sharing approximately 80% amino acid sequence identity in this region within species(Edmondson and Olsen, 1993). The HLH region is characterised by two amphiphatic α-helices, separated by an unstructured intervening loop. HLH regions are mutually attractive and facilitate the formation of functionally active protein dimers(Maleki et al., 2002). The basic region forms an extension of one of the α-helices of the HLH region and facilitates DNA binding. bHLH dimers specifically bind E-box elements (CANNTG) found in the promoters and enhancers of most, if not all,muscle-specific genes (Apone and Hauschka,1995; Spinner et al.,2002) although it is likely that nucleotide variation in the flanking regions and within the motif imparts some specificity(Ludolph and Konieczny,1995).

During myogenesis, the transcription factors Myf-5 and MyoD are required for the initial determination of the myogenic lineage. Gene knockout studies in mice show that lack of MyoD and Myf-5 results in failure of myoblast formation, and a consequent lack of all head and trunk skeletal muscle(Rudnicki et al., 1993). In zebrafish, targeted knockdown with a Myf-5 morpholino has been shown to induce defects in myogenesis and brain formation(Chen and Tsai, 2002). The expression of myogenin and MRF4 is activated during myoblast differentiation (Rhodes and Konieczny,1989; Wright et al.,1989; Miner and Wold,1990; Edmondson and Olson,1993; Pownall et al.,2002), and myogenin and MRF4 probably have cooperative functions with MyoD and Myf-5 as transcription factor regulators for the activation of muscle contractile protein genes (Lassar et al., 1991). In myogenin-knockout mice, myoblasts form in the correct place but do not fuse into muscle fibres(Hasty et al., 1993; Nabeshima et al., 1993; Venuti et al., 1995). The function of MRF4 is less clear because in all three mutants constructed to inactivate it, Myf-5 production is also affected(Olson et al., 1996;Summerbell et al., 2000, 2002).

In the zebrafish, Danio rerio, Myf-5 and MyoD transcripts are initially seen at approximately 7.5 h at 28.5°C (80% epiboly) in bilateral bands of cells flanking the presumptive notochord(Weinberg et al., 1996; Chen et al., 2001; Coutelle et al., 2001). The expression patterns of these two genes overlap considerably, incorporating the adaxial cells as they form. Expression of Myf-5 extends further into the presomitic mesoderm than that of MyoD but, strikingly, as the adaxial cells become incorporated into the somites, Myf-5transcription dramatically declines. Expression of MyoD persists in the differentiated somites until much later, after they become chevron-shaped,whereupon it is downregulated. Expression of myogenin begins at 10.5 h (at 28.5°C) in a subset of the MyoD/Myf-5-expressing cells (Weinberg et al., 1996; Chen et al., 2000). The myogenin transcripts first appear in bands of cells extending laterally away from the adaxial cells. However, this lateral extension of expression is narrower than in the case of MyoD and, due to its later onset, first expression is within the somites rather than the presomitic mesoderm. Transcription of myogenin is also transient, and persists until shortly after the disappearance of MyoD transcription. Furthermore, there are some differences in MRF gene expression between fish species. In the rainbow trout, Oncorhynchus mykiss, for instance, MyoD expression, rather than spreading laterally, remains confined to the medial domain of the somite for a prolonged period(Delalande and Rescan, 1999). In the herring Clupea harengus, myogenin mRNA shows a more transient expression pattern than that seen in zebrafish(Weinberg et al., 1996) and trout (Delalande and Rescan,1999), disappearing from the somites before the downregulation of MyoD (Temple et al.,2001). A number of species including the trout(Rescan and Gauvry, 1996),gilthead seabream Sparus aurata(Tan and Du, 2002) and Xenopus laevis (Scales et al., 1990; 1991; Charbonnier et al., 2002) also possess multiple copies of one or more MRF-encoding genes.

Temperature has been shown to influence many aspects of development in teleosts, including muscle cellularity(Stickland et al., 1988; Vieira and Johnston, 1992; Nathanailides et al., 1995; Johnston and McLay, 1997; Matschak et al., 1998;Galloway et al., 1998, 1999; Hall and Johnston, 2003) and the relative timing of myofibrillogenesis (Johnston et al., 1995, 1996, 1997). There is also a small body of evidence to suggest the timing and extent of MRF gene expression varies with temperature. Xie et al.(2001) detected MyoDand myogenin mRNAs in a greater number of somites in trout embryos of the same developmental stage, reared at 12°C compared with 4°C. This change in expression was apparently concomitant with a `relatively advanced'state of muscle development at 12°C compared with 4°C. Similarly,Wilkes et al. (2001) used quantitative northern blots to show that MyoD and myogeninmRNA levels in trout and sea bass Dicentrarchus labrax were highest at temperatures close to those of the usual environmental spawning temperatures for the species. By contrast, Temple et al.(2001) found no difference in the timing of MyoD or myogenin expression in herring embryos reared at 5, 8 and 12°C. Hall et al.(2003) also found no difference in the timing of MyoD expression between Atlantic cod Gadus morhua embryos reared at 4, 7 and 10°C, although the timing of blastopore closure relative to somite stage was relatively delayed at 7 and 10°C when compared with 4°C, and the number of deep fibres at hatching in the 10°C group was significantly higher than in the lower temperature groups.

Fishes from cold environments express myosin heavy chain (MyHC) protein isoforms with a higher specific myofibrillar ATPase activity and a lower thermal stability than those from warmer environments (Johnston et al., 1973, 1975a,b),and there is an apparent trade off between these traits. Species with a broad temperature tolerance, such as the goldfish Carassius auratus and the common carp Cyprinus carpio, can alter their Mg²⁺Ca²⁺ ATPase activity depending on the ambient temperature by differential expression of multiple MyHC genes(Goldspink et al., 1992; Watabe et al., 1995; Imai et al., 1997; Cole and Johnston, 2001). The control of such acclimation responses is unknown and, to date, has not been demonstrated in embryos, which express many of their own developmental stage-specific isoforms of muscle proteins(Scapolo et al., 1988; Crockford and Johnston, 1993; Johnston et al., 1997). In mammals, there is evidence for involvement of the MRFs in the determination of contractile protein isoform expression and fibre typing(Voytik et al., 1993; Hughes et al., 1999) along with other influences, such as hormones and innervation(Hughes et al., 1993; Lefeuvre et al., 1996).

The common carp is a eurythermal species commonly inhabiting waters that fluctuate between near freezing and 30°C seasonally(Michaels, 1988). Spawning occurs in the summer months at a minimum temperature of ∼18°C, and the eggs and larvae develop normally between temperatures of 18 and 25°C(Penáz et al., 1983; Balon, 1995). In the presnt study, the spatial and temporal expression patterns of MyoD,myogenin, and Myf-5 were characterised, and the hypothesis that temperature influences expression of the MRFs within the normal limits of thermal tolerance was investigated by comparing embryos and larvae reared at 18 and 25°C. The in situ expression pattern of Myf-5 was of particular interest because within the Teleostei, to date, it has only been described in the zebrafish and has never been investigated in relation to temperature. In addition, the expression of five different MyHCtranscripts (two embryonic types, Ennion et al., 1999; and three temperature-specific types, Imai et al., 1997) were characterised and compared between temperature groups. The aims of the present study were to investigate the initial expression of temperature-specific MyHC isoforms in larvae, and whether embryonic isoforms are differentially expressed in response to rearing temperature, and to characterise the timing of expression switching from embryonic to adult isoforms. Finally, since many MRF cDNAs from teleosts have been cloned in recent years and parologous genes have been identified, a comprehensive phylogeny of vertebrate MRFs was also undertaken. Neighbour-joining and parsimony analyses were used to generate phylogenies to elucidate evolutionary relationships between the genes and the relative timing of gen(om)e duplication events.

Carp spawning and egg incubation were carried out according to Michaels(1988). Briefly,over-wintering adult carp were brought into the laboratory in early January 2000. The water temperature was raised by 3°C per day from 4 to 25°C,where it was held for a further six weeks. Female fish were given 0.6 mg kg^-1 of carp pituitary acetone powder (Sigma, Poole, UK) by intramuscular injection, followed by 3 mg kg^-1 12 h later. Males were given a single injection of 1.5 mg kg^-1. After a further 12 h,eggs and milt were stripped into separate dry containers. They were mixed in the ratio 1:100 (v/v) milt:eggs, and activated with an equal volume of 0.3%urea, 0.3% NaCl. The fertilization reaction was allowed to proceed for 1 h,after which the eggs were washed three times in 0.5% (v/v) tannic acid to prevent aggregation. Fertilized eggs were transferred to Zuger jars and incubated under constant aeration at 18 and 25°C±1°C (range). Embryos were sampled every 6 h by anaesthetizing in 0.1% (m/v) tricaine(MS-222; Sigma, Poole, UK), puncturing the chorion with a hypodermic needle,and fixing in 4% (m/v) paraformaldehyde in phosphate-buffered saline (PBS). After 12 h of fixation the embryos were washed once in PBS and stored at-80°C in 100% methanol.

The MyoD, myogenin and Myf-5 clones used were as previously described by Kobiyama et al.(1998). 10°C-type,intermediate-type, and 30°C-type MyHC were as described by Imai et al. (1997). The two embryonic-type MyHC clones (Eggs22 and Eggs24) were generously supplied by Geoff Goldspink and are described by Ennion et al.(1999). DIG-labelled cRNA probes were constructed from linear plasmids according to Hall et al.(2003). Details of plasmids,restriction endonucleases and transcriptases are shown in Table 1.

Five embryos of equivalent developmental stages from each sample were selected per cRNA probe. In situ hybridization was carried out using the procedure described by Hall et al.(2003). Photographs were taken on a Leica MZ7.5 binocular microscope (Leica, Milton Keynes, UK) using darkfield illumination and a Zeiss Axiocam imaging system (Zeiss, Welwyn Garden City, UK).

Total RNA was extracted from the trunk muscle of hatched larvae (the head,tail and yolk sac were removed) using Tri-reagent (Sigma, Poole, UK). RNA dot-blots were performed by spotting 2.5 μg of total RNA in 0.5 μl water onto nitrocellulose (Hybond-N⁺; Amersham-Pharmacia, Little Chalfont, UK), and fixing at 120°C in an oven for 30 min. A 30 min prehybridization was carried out in 50% (v/v) formamide, 0.1% (m/v)N-lauroylsarcosine, 0.02% (m/v) SDS, 2% (v/v) blocking reagent (Roche, Lewes,UK) at 65°C, before addition of probe at 100 ng ml^-1. After hybridization overnight at 65°C, the blots were washed 2×15 min in 2× SSC, 0.1% (m/v) SDS at room temperature, followed by 2×15 min in 0.5× SSC, 0.1% SDS at 65°C. Membranes were blocked in 2% (v/v)blocking reagent, 100 mM maleic acid, 150 mM NaCl, pH 7.5 for 1 h, before addition of an alkaline-phosphatase-conjugated anti-DIG antibody, Fab fragments (Roche, Lewes, UK) at a dilution of 1/100,000. After a 30 min incubation in the antibody solution, membranes were washed 2×15 min in 100 mM maleic acid, 150 mM NaCl, pH 7.5, 0.3% (v/v) Tween-20. Detection was achieved using a 1:100 dilution of the chemiluminescent substrate CSPD (Roche,Lewes, UK), in 100 mM Tris-HCl, 100 mM NaCl, pH 9.5 followed by exposure to X-ray film.

A phylogenetic analysis was undertaken using full-length amino acid sequences of vertebrate MRFs taken from the GenBank database (NCBI, Bethesda,USA). Five additional sequences were predicted from Ensembl(www.ensembl.org)and Genoscope(www.genoscope.cns.fr)genome assemblies (see Data 1 in supplementary material), using Blast2(v2.2.6) (Altschul et al.,1997) and Genewise (Birney et al., 2004). An initial multiple alignment was constructed using the Clustal algorithm in Lasergene (DNAstar Inc., Madison, USA), which was then improved by eye. A neighbour-joining (NJ) tree was constructed in PHYLIP(Felsenstein, 1995) and bootstrapped 1000 times to provide statistical support. Parsimony analysis was carried out using PAUP (Swofford,2002) (see Data 2 in supplementary material).

Somitogenesis began almost immediately following epiboly as described by Verma et al. (1970) and Penáz et al.(1983), at 22 h in the 18°C group, and 12 h in the 25°C group. Somites were formed at ∼1 per hour (18°C) and ∼2 per hour (25°C) to a final number of 38 or 39 (both groups). Time until 50% hatching was 120 h at 18°C and 55 h at 25°C. At hatching, embryos measured 3.72±0.39cm (s.d.),and there was no significant difference between temperature groups (Student's t-test, P>0.05, N=20 fish per group).

Expression of MyoD and Myf-5 occurred simultaneously following epiboly in the pre-somitic mesoderm. MyoD was expressed in a pair of bilaterally symmetrical strips corresponding to the position of the adaxial cells (Fig. 1c,d),adjacent to the notochord. Myf-5 was also expressed in the adaxial cells, but as development proceeded, transcripts spread further laterally into the mesoderm (Fig. 1a,b). Before the appearance of the first somite furrows, Myf-5 expression could be seen very faintly in two bands corresponding to the cellular fields of the first somites (Fig. 2a). As soon as each somite formed, however, expression of Myf-5 was downregulated. By contrast, expression of MyoD persisted as the somites were formed (Figs 1c,d). The dynamics of expression were such that at any time during somitogenesis, the newest ∼12 somites stained positive for MyoD mRNA (Fig. 2b).

Expression of myogenin was switched on in the somites later than Myf-5 and MyoD (Fig. 1e,f). The extent of staining lagged behind that of MyoDby ∼5 somites, and ∼12 were stained at any one time(Fig. 1c). The expression patterns of all three transcripts gave the appearance of a rostral-caudal wave, initiated by Myf-5, and followed by MyoD and myogenin, respectively (Fig 1a-f). No differences were seen between 18 and 25°C groups relative to developmental stage.

The embryonic forms of MyHC, Eggs22 and Eggs24 were first seen at the 25-30 somite stage, beginning in the anterior-most somites and progressing caudally (Fig 1g-j). After the completion of somitogenesis, Eggs22transcripts became concentrated in the caudal somites, whereas Eggs24predominantly stained the anterior somites. Expression persisted post-hatch,but was much reduced. No differences were seen between the 18 and 25°C groups with respect to developmental stage(Fig. 1g-j). No expression of mRNA for the 10°C-type, intermediate-type and 30°C-type MyHC isoforms were seen at any stage. Positive dot-blots using RNA isolated from fast muscle of 10 and 30°C acclimated adult carp (10 cm total length), alongside negative blots from the 18°C and 25°C incubated post-hatch larvae provided a positive control for the in situ results(Fig. 3).

The neighbour-joining tree separated the four MRFs in relation to the outgroup Ascidian sequence (Fig. 4; for accession numbers see Data 2 in supplementary material). Within genes, clades broadly reflected evolutionary relationships, and the majority of the bootstrap values were high (>90%). Further support was given by comparison with the tree from parsimony analysis, which was almost identical. Importantly, the Xenopus MyoD and myogeninparalogues clustered together, as did the trout MyoD paralogues. By contrast, the seabream amino acid sequences were more highly divergent.

The expression patterns of carp MyoD and myogeninmore-closely resembled those of the zebrafish than those of other teleosts studied to date. In the trout, MyoD expression extends laterally outwards from the adaxial cells relatively late in development, after the somites acquire their chevron shape, whereas in the other fish species studied(zebrafish, Weinberg et al.,1996; herring, Temple et al.,2001; seabream, Tan and Du,2002), adaxial cell expression of MyoD occurs across the somites soon after their formation. Furthermore, expression of myogenin persisted for longer in the somites than expression of MyoD, unlike in the herring, where the reverse is the case(Temple et al., 2001). Expression of carp Myf-5 also resembled that seen in the zebrafish,although in the zebrafish a more clearly defined banding pattern is seen in the presomitic mesoderm prior to somite formation, with at least five Myf-5-positive presomitic bands(Coutelle et al., 2001). Similarities between expression patterns might be expected between the carp and zebrafish given that they are taxonomically closely related, both belonging to the family Cyprinidae. MRF4 expression has not been studied in any teleost to date, although a genomic clone has been isolated from the pufferfish, Fugu rubripes(Carvajal et al., 2001; Fig. 4), and a cDNA sequence exists for the zebrafish (Fig. 4; see Data 1 in supplementary material).

The expression patterns of the genes encoding the embryonic MyHC isoforms(Eggs22 and Eggs24) also showed no difference in timing between temperature groups, and the timing of transcription was broadly similar to that described by Ennion et al.(1999). However, the finding that the adult 10°C-type, intermediate-type and 30°C-type MyHCs were not expressed, even as the embryonic forms disappeared,was significant although not altogether unexpected. Other MyHC isoforms must be present to bridge the gap, either further embryonic forms, adult forms, or forms specific to the larval stages. Embryonic MyHC isoforms have been described in a variety of other species including human(Eller et al., 1989; Karsch-Mizrachi et al., 1989)rat (Strehler et al., 1986),chicken (Molina et al., 1987; Hofmann et al., 1988) and Xenopus (Radice and Malacinski,1989). However, the myosin heavy chain multigene family in the carp is particularly large. Kikuchi et al.(1999) isolated 29 different genomic clones, more than twice the number present in humans(Soussi-Yanacostas et al., 1993; Kikuchi et al., 1999). Such diversity in carp myosin genes probably reflects the need for different molecular characteristics during the life cycle, as a result of allometric scaling relationships and temperature acclimation (Imai et al.,1997; Ennion et al.,1999; Kikuchi et al.,1999; Cole and Johnston,2001).

The neighbour-joining tree for the MRF family is shown in Fig. 4. The topology supports the notion, proposed by Atchley et al.(1994), that all four members evolved from a common ancestor by gene duplication. After an initial duplication, each lineage divided again, one giving rise to Myf-5 and MyoD, and the other giving rise to myogenin and MRF4. However, despite the fact that MRF4 is most-closely related to myogenin, in the human and pufferfish the MRF4gene is most-closely associated spatially with Myf-5. In human, MYF5 and MRF4 are located on chromosome 12, with their start codons only 8.5 kb apart (Patapoutian et al., 1993) and in pufferfish they are even closer together, with their start codons differing by less than 5 kb (genomic clone encoding Myf-5 and MRF4, NCBI accession no. AJ308546). It is possible that the functions of the two genes demand that they respond to the same control regions, or that their close proximity is essential for their autoregulation, a hypothesis that is supported by the fact that in all of the three Mrf4-knockout mice constructed, Myf5 function is also affected (Summerbell et al.,2002).

Recently, the view of the MRFs as a discrete family of four transcription-factor-encoding genes has been clouded by the discovery of parologous forms, which have diverged in function in some species. Rescan and Gauvry (1996) isolated a second form of MyoD from the trout, and demonstrated different expression patterns using in situ hybridization. MyoD1 was expressed in the adaxial cells of the unsegmented mesodermal plate and in the developing somites. MyoD2 expression, however, was initiated later and was limited to the posterior compartment of the somite. Similarly, in Xenopus, paralogous forms of MyoD and myogenin have been isolated. One MyoD transcript (xlmf25) is expressed as a maternal mRNA in the early embryo, while the other (xlmf1) is activated from the zygotic genome near to the beginning of somitogenesis(Scales et al., 1990, 1991). Of the myogenin transcripts, one (XmyogU2) is expressed during embryogenesis, while the other (XmyogU1) is exclusive to the adult skeletal muscle (Charbonnier et al.,2002).

The expression of parologous genes is common in some organisms, such as trout and Xenopus, both of which have undergone recent genome duplication events and are in a state of pseudotetraploidy(Allendorf and Thorgaard, 1984; Hughes and Hughes, 1993; Rescan, 2001). However, the non-tetraploid gilthead seabream also differentially expresses two parologous forms of MyoD (Tan and Du,2002). In this case, the sequence identity of the two forms is lower than for the tetraploid organisms(Fig. 4), suggesting a more-ancient duplication event. Interestingly, a cDNA that clustered with seabream MyoD2 (Fig. 3) was recently isolated from the Atlantic cod(Hall et al., 2003).

No parologous forms of MRF family genes have been isolated from any of the tetrapod lineage, with the exception of the tetraploid Xenopus, and,paradoxically, despite the availability of whole genome shotgun sequences, in the zebrafish or pufferfish. The dynamics of teleost genome evolution is extremely complex, with evidence for specific genome duplication events remaining a contentious issue (Meyer and Malaga-Trillo, 1999; Meyer and Schartl, 1999; Robinson-Rechavi et al., 2001a,b;Taylor et al., 2001a,b). In any case, whether at the whole-genome or more-regional level, teleost genomes are characterised by a high rate of duplication followed by substantial gene loss (Robinson-Rechavi et al., 2001c; Sibthorpe,2002; Smith et al.,2002). Further characterizing the molecular evolution of the MRF family in relation to function remains a challenging, but potentially rewarding, task.

Since going to press, important new evidence has arisen regarding genome duplication in the teleost lineage. Jaillon et al.(2004) present near definitive evidence from Tetraodon nigroviridis of an ancient full-scale genome duplication. They demonstrate firstly, that every chromosome was involved in large-scale duplication, and secondly, a striking pattern of double synteny,with one chromosomal region in humans matching two in the pufferfish, across the whole genome.

The authors would like to thank Dean Sibthorpe and Robert Bryson-Richardson for valuable advice and discussion on phylogenetic analysis. T.E.H., N.J.C. and M.A.C. are grateful for research grants from the Natural Environment Research Council. C.I.M. is grateful for a studentship from the Biotechnology and Biological Sciences Research Council.