A role for the Myoblast city homologues Dock1 and Dock5 and the adaptor proteins Crk and Crk-like in zebrafish myoblast fusion

Skeletal muscle is composed of bundles of syncytial myotubes that arise during embryonic development by the fusion of precursor cells known as myoblasts. As well as underpinning muscle development, myoblast fusion also plays an essential role in post-embryonic life in the repair of damaged muscle fibres. Despite its fundamental importance, the molecular basis of myoblast fusion in vertebrates is still poorly understood. In Drosophila, by contrast, genetic analysis has provided a detailed inventory of the molecular components that underpin the process in embryonic development. Heterophilic interactions between members of the immunoglobulin superfamily - Dumbfounded(Duf; also known as Kirre), Roughest (Rst; also known as IrreC) and Sticks and stones (Sns) (Bour et al.,2000; Ruiz-Gomez et al.,2000; Strunkelnberg et al.,2001) - mediate the initial contact between fusing cells(Ruiz-Gomez et al., 2000; Galletta et al., 2004),triggering intracellular pathways that orchestrate the cytoskeletal rearrangements and cell shape changes that accompany fusion (reviewed in Taylor, 2003; Chen and Olson, 2004). A key component of the intracellular network is the protein encoded by the myoblast city (mbc) gene(Rushton et al., 1995; Erickson et al., 1997), a close homologue of the vertebrate protein dedicator of cytokinesis 1 (DOCK1,formerly DOCK180) and the Caenorhabditis elegans protein CED-5,which, together with Mbc, define the CDM family of proteins(Wu and Horvitz, 1998).

Myoblast fusion is much less readily amenable to experimental analysis in higher vertebrates, such as the mouse; for this reason, most of what is known about the process has been gleaned from in vitro studies using various murine and human muscle cell lines. These studies have identified a large set of proteins that appear to be required for fusion(Rosen et al., 1992; Zeschnigk et al., 1995; Schwander et al., 2003; Horsley and Pavlath, 2004; Doherty et al., 2005; Lafuste et al., 2005; Charrasse et al., 2006; Charrasse et al., 2007; Comunale et al., 2007; Georgiadis et al., 2007). Given the striking similarities between the cellular events that underpin myoblast fusion in vertebrates and invertebrates(Knudsen and Horwitz, 1977; Knudsen and Horwitz, 1978; Wakelam, 1985; Doberstein et al., 1997),there is surprisingly little, if any, overlap between these proteins and those identified by genetic analysis in Drosophila.

In contrast to the mouse, the zebrafish embryo has a number of qualities that makes it particularly well-suited to the in vivo analysis of myogenesis(Devoto et al., 1996; Blagden et al., 1997; Currie and Ingham, 1998; Roy et al., 2001; Henry and Amacher, 2004; Hughes, 2004). The embryonic myotome develops during the first 24 hours postfertilisation (hpf) and is comprised of two kinds of fibre: mononucleate slow-twitch fibres and multinucleate fast-twitch fibres, the latter of which arise via the fusion of precursor myoblasts (Roy et al.,2001). Muscle fibres and their progenitors can be readily imaged using confocal microscopy, and gene function can be manipulated using morpholino antisense oligonucleotides (morpholinos)(Ekker, 2000). We have taken advantage of these properties to investigate whether orthologues of the components that control myoblast fusion in Drosophila have a conserved function in vertebrate myogenesis. Here, we describe the establishment of a method for the quantitative analysis of fast-twitch myoblast fusion during zebrafish embryogenesis and its application to the analysis of the function of CDM family members closely related to Mbc. Because CDM proteins form part of multiple signalling pathways, the best-characterised being the Crk-CDM-Rac pathway, we have also used our assay to investigate a postulated role for the adaptor proteins Crk and Crk-like (Crkl) in myoblast fusion.

Zebrafish embryos were collected at 24 hpf, washed in water and RNA was extracted with Trizol (Invitrogen). Reverse transcription (RT) was performed using the Superscript RT kit (Invitrogen), and PCR was carried out using the following primers (all sequences are shown 5′-3′):

Dock1: AGAGAAGCGAGAAAAAGTGTGTG and GTCATAGTTGTAGACGGCCACTC,GAGTGGCCGTCTACAACTATGAC and ACACTGGATTGAAATGATGGAAC, ACACTGGATGCTCTCTTCAACAT and TTGTCCATCAGGTTTCCATAGTC; Dock5: ATTCCACTTCACTCGAGTTTTCC and TTTCAGGGTCTAAAGTGTTTCCA, GTGCCTACAGAATTGCCTTTAGTT and TGTTGATGTATTGTGCGTAATGTG, CGGAGGGTCAATCGTACAGT and CATTGGCCAGCTCCATAGTT,TTCCAGGAATCCTCAAGTGG and GTTTTCCTCCAGCCTCACAG.

Full-length sequences of dock1 and dock5 were cloned by PCR using the following primers:

Dock1: AGAGAAGCGAGAAAAAGTGTGTG and GAGGTGTCAGCTGCTTTTCC; Dock5:ATTCCACTTCACTCGAGTTTTCC and CACAGCATGCAGTTGTTTGA.

A full-length construct for crk was cloned using the following primers: AGGATCCACCATGGCCGGAAATTTTGA and TTCTAGAAGGGCTGCTAGGTAAGACAG.

All PCR products were ligated into the TOPO-TA pCRII vector with dual promoters (Invitrogen) and sequenced.

Anti-sense probes for dock1 and dock5 were transcribed from the RT-PCR products that were cloned into the TOPO-TA pCRII vector. RNAs from several sections of the genes were pooled. In situ hybridisation was performed using standard procedures(Westerfield, 2000).

Morpholino sequences were: crk ATG, GCCTTCCCTTCGCCTCTCCTTATCC; crk splice, GGAGCAAGCCCTGCGGGATGACATT; crkl ATG,AGGAGTCGAACCGTGCAGACGACAT; crkl splice, TCATTTAGCCACTTACCTCCAGTGC; dock1 ATG, TTTTTGTAGGCACCCAGCGCGACAT; dock1 splice,CATCACCTGCAAACACACAACACAC; dock5 splice 1, TGTTGATGTCTTACTATGTAGGGAG; dock5 splice 2, AAAACAGCGCTCACCTTCTGGAATG.

Morpholinos (Gene Tools) were injected at the one- to two-cell stage as previously described (Wolff et al.,2003) at 0.1-2.0 mM, in solution containing 40 ng/μl mylz2:GFP plasmid. This plasmid contains EGFP (Clontech)preceded by 2239 bp of the mylz2 upstream region(Ju et al., 2003) amplified from genomic DNA using the following primers: TCAATCAAATATCACCCCATATGTC and TGTGAAGTCTAAGAAGATCAAGAAGAGA. Embryos expressing the mylz2:GFPplasmid were washed in ice-cold PBS then fixed in 4% paraformaldehyde/PBS at 4°C overnight. Subsequently, embryos were washed in PBS + 0.1% Tween20 and mounted laterally in 75% glycerol with yolk removed. An Olympus BX61 microscope with Fluoview FV1000 confocal system and 60× water lens was used. Counting nuclei was easiest when embryos were viewed without any additional staining procedure through the microscope eyepieces at high magnification (600× total).

For morpholinos targeting splice junctions, groups of 20 injected embryos were collected at 26-28 hpf, RNA extraction and RT were performed as above,then PCR was carried out using the following primers: crk,GTGCTGTCTGTGTCGGAGAA and TGGTCACCTTTACCATGTCG; crkl,TTGGATACGACCACCCTGAT and CCTTCCCACTGACCACTGAT; dock1,GCTGAAGCTTCTTCCAGGTG and CTTGAAGGTGAAACGCAGGT; dock5 splice 1,CGGAGGGTCAATCGTACAGT and TTCTCAGGCGAGAAGGACTC; dock5 splice 2,CGGCTTCAAGCTGAGAGATA and CGGATATAAATGTCCTCCCTCTT.

PCR products were ligated into the TOPO-TA pCRII vector with dual promoters(Invitrogen), and sequenced.

For in vitro transcription, a full-length crk cDNA amplified by PCR (see above) was subcloned into pCS2+ with BamHI and XbaI. The crkl full-length construct was made by subcloning crkl from the plasmid IRAKp961A10118Q (RZPD clone) with SpeI and XbaI into the XbaI site of pCS2+. Full-length constructs were linearised with NotI and transcribed in vitro using SP6 and the mMessage Machine kit (Ambion), and the mRNA was injected into newly fertilised eggs as previously described (Krauss et al., 1993).

Antibodies used were F59 [developed by Frank E. Stockdale(Miller et al., 1985) and obtained from the Developmental Studies Hybridoma Bank] (1:50) and rabbit anti-β-catenin (Sigma; 1:50). Secondary antibodies were Alexa-Fluor-488 goat anti-rabbit IgG (Molecular Probes; 1:500), goat anti-mouse Cy3 (Jackson ImmunoResearch; 1:500). DAPI (Sigma; 1:1000) was also used. Staining was performed using standard procedures.

The process of fast-twitch myoblast fusion was analysed both by the global staining of fixed zebrafish embryos and by scatter labelling of fast fibres and their precursors using a mylz2:GFP transgene(Xu et al., 2000; Ju et al., 2003). Prior to the onset of fusion, the fast-fibre precursors were seen as small rounded cells that extended thin protrusions (Fig. 1A,D). These protrusions could best be seen when isolated cells were labelled by the mylz2:GFP transgene(Fig. 1D). At the 18-somite stage (18 hpf), mononucleated slow-twitch fibres were already elongated and spanned the entire length of the somite(Fig. 1B). Fast-twitch fibres elongated later; elongated fibres could be seen as fusion began, at approximately 19 hpf (20 somites) (Fig. 1E). By 26 hpf, fusion had largely finished and most fast fibres were multinucleated (Fig. 1C,F).

We performed a quantitative analysis of the process, counting nuclei in individual fast-twitch fibres highlighted by the transient mosaic expression of the mylz2:GFP transgene. Because varying numbers of fibres were labelled in each embryo, the mean number of nuclei/fibre for each embryo was treated as one sample value, weighting each embryo equally. Overall means of these samples are shown in Table 1, together with the total numbers of fibres scored, total number of nuclei and number of embryos analysed. From the distributions of the individual embryo means shown in Fig. 1G, it can be seen that the majority of fusion events occurred between 20 somites and 24 hpf. Mann-Whitney rank tests confirmed significant increases in the mean number of nuclei/fibre between embryos at 15 and 20 somites (P<0.02), 20 and 22 somites (P<0.01), and 22 somites and 24 hpf (P<0.02). Further fusion events occurred over the next 24 hours; there was a significant increase in nuclei/fibre between 30 hpf and 48 hpf (P<0.04). The raw data upon which this analysis is based can be found in Tables 1-6 in the supplementary material.

Given the key role of mbc in myoblast fusion in Drosophila, we searched for orthologues of the human DOCK180 sub-family (Cote and Vuori,2002; Meller et al.,2005) in zebrafish. One member of this family, dock2, has previously been annotated in zebrafish and shown to be expressed specifically in macrophages (Thisse et al.,2001), a pattern reminiscent of the haematopoietic restriction of human DOCK2 (Nishihara et al.,1999). We therefore focussed our attention on the identification of the other two sub-family members, dock1 and dock5. Using BLAST searches of the Sanger Ensembl zebrafish database, we identified nucleotide sequences whose conceptual translation products showed significant levels of similarity to the mammalian DOCK1 and DOCK5 proteins. By assembling overlapping database sequences, an RT-PCR strategy was devised to amplify sections of the putative zebrafish dock1 and dock5 genes,allowing the sequence of the full-length coding region of each gene to be determined.

Both genes encode predicted proteins with CZH1 and CZH2 domains. The CZH1 domain of CDM proteins has been shown to interact with PtdIns(3,4,5)P₃ (Cote et al.,2005), whereas the CZH2 domain is a novel guanine-nucleotide exchange factor (GEF) domain that is implicated in the regulation of the small GTPase Rac1 (Reddien and Horvitz,2000; Brugnera et al.,2002; Cote and Vuori,2002; Meller et al.,2005). Each predicted protein also contains one or two C-terminal proline-rich sequences conforming to the SH3 consensus binding motif (P)PXLPXK(Schumacher et al., 1995; Matsuda et al., 1996), which has been shown to mediate interaction with Crk; PPPLPVK and PPPLPSK in the case of Dock1 and MPPLPAK in the case of Dock5. These are signature features of CDM proteins, although in humans, only DOCK1 contains exact (P)PXLPXK motifs. On the basis of this domain organisation and of the high level of protein similarity between the human DOCK1 (Ensembl accession ENSP00000280333)and DOCK5 (ENSP00000276440) and the corresponding zebrafish proteins (84.1%and 70.8% identity respectively), we conclude that these transcripts encode the zebrafish Dock1 and Dock5 orthologues (see Fig. S1 in the supplementary material). Exon boundaries were inferred using homology to the human proteins,and confirmed with the trace search facility on the Sanger website, which searches the genomic shotgun sequence database. RT-PCR analysis showed expression of both genes throughout development, from the two-cell stage to 27 hpf (Fig. 2A). In situ hybridisation analysis revealed that both genes are ubiquitously expressed during the early somitogenesis stages (Fig. 2B,D), with some modulation of expression becoming apparent by 22 somites (20 hpf) (Fig. 2C,E).

To investigate a possible requirement for either of the Dock proteins in myoblast fusion, we used the injection of morpholino oligonucleotides into newly fertilised wild-type embryos to block the function of these proteins. Simultaneously, we injected plasmid DNA carrying the mylz2:GFPreporter gene in order to scatter-label fast-twitch myoblasts. The injected embryos were fixed at 26-28 hpf and analysed by confocal microscopy.

To assess the function of dock5, we injected splice morpholinos targeting the junctions between exons 27 and 28 (splice 1) and between exons 33 and 34 (splice 2) of this gene. The majority of labelled myocytes in the morphants were elongated and spanned the somite width(Fig. 3A,B,E), although some failed to elongate and appeared more rounded(Fig. 3A). In both cases, a substantial proportion were mononucleate. The dock5 splice 1 morpholino caused a decrease in overall mean nuclei/fibre from 2.63 (wild-type control) to 1.36 at 0.5 mM, and the dock5 splice 2 morpholino overall mean nuclei/fibre was 1.88 at 0.5 mM (see Table 2 for mean nuclei/fibre and Fig. 4 for distributions of individual embryo means). Using the Mann-Whitney test, the distributions of embryo means were shown to be significantly different from wild type for both groups (P<0.0001). Higher doses of the second splice morpholino gave a greater effect on fusion but led to a higher frequency of early arrest.

To confirm that mylz2:GFP-labelled mononucleate fibres are not of slow-twitch type, we stained the injected embryos with the F59 antibody, which specifically labels slow fibres at this stage(Devoto et al., 1996). None of the GFP-positive cells were labelled with the antibody. In addition, this staining revealed that the differentiation of superficial slow fibres occurs normally and on schedule in morphant embryos(Fig. 3F), although, in the more severely affected morphants, the somite shape was often disrupted such that the characteristic chevron-shape was lost(Fig. 3C). In either case,however, fusion was affected to an equal extent.

To abrogate Dock1 activity, we injected morpholinos directed against either the start ATG or the splice junction between exons 13 and 14 of the transcript. The resultant embryos were morphologically similar to the dock5 morphants described above. Injection of the splice morpholino at two different concentrations resulted, in both cases, in a reduction in the mean number of nuclei/fibre to 1.72 (0.3 mM) or 1.31 (0.5 mM); because the higher dose resulted in a high frequency of early arrest, only the 0.3 mM data were used for statistical analysis (see Table 2 and Fig. 4). The highest number of nuclei observed in a single fibre in morphant embryos was, respectively, five and three for the two different doses, compared with a maximum of seven in wild-type controls. The dock1 ATG morpholino had a less pronounced effect, with the number of nuclei/fibre ranging from one to six, with a mean of 2.35 for the highest dose of morpholino used (1.5 mM). Further tables of raw data are presented in Tables 1-6 in the supplementary material. Using the Mann-Whitney test, the distributions of embryo means were shown to be significantly different from wild type for all these groups(P<0.0001 for dock1 splice at 0.3 mM, P<0.003 for dock1 ATG at 1.5 mM).

The efficacy of each of the splice morpholinos was assayed by RT-PCR analysis of RNA extracted from the injected embryos(Fig. 5). Sequence analysis of the amplification products showed that the dock1 splice morpholino caused the deletion of 40 bp of exon 14, leading to the introduction of a premature stop codon. The dock5 splice 1 morpholino led to the inclusion of 273 bp of intron 27, introducing a stop codon, and dock5splice 2 morpholino caused deletion of 75 bp of exon 33, which also introduced a stop.

CDM proteins interact physically with Crk and have been implicated in the same signalling pathways both in mammalian cell lines and in C. elegans embryos (Hasegawa et al.,1996; Reddien and Horvitz,2000). A role for Crk in myoblast fusion in Drosophilahas been hypothesised based on its biochemical interaction with Mbc, which mirrors that shown to be important in regulating the mammalian DOCK proteins(Galletta et al., 1999), but has not been tested directly by genetic inactivation. The presence of putative Crk-binding motifs in zebrafish Dock1 and Dock5 is consistent with both proteins interacting with Crk and/or the related protein Crkl. We therefore used our in vivo fusion assay to investigate the involvement of Crk and Crkl in vertebrate myoblast fusion.

The zebrafish crk and crkl orthologues have previously been identified in a cDNA expression screen and have been shown to be expressed ubiquitously (Thisse and Thisse,2004). By RT-PCR, we confirmed that both genes are expressed throughout embryogenesis (Fig. 2A). To inactivate their function, we designed morpholino oligonucleotides that target the translation initiation codons and splice sites. The efficacy of the splice morpholinos was confirmed by RT-PCR and sequencing (Fig. 5). The crk morpholino caused some deletion of exon 2 and thus a frame shift in exon 3, whereas the crkl morpholino caused either inclusion of 57 bp of intron 2 and thus the introduction of a stop codon, or, more rarely,caused the deletion of exon 2. As with the Dock morphants, a large proportion of the fast myoblasts in injected embryos remained unfused at 26-28 hpf. The overall mean number of nuclei per mylz2:GFP-labelled cell following injection of the highest dose of the crk ATG or splice morpholino was 1.56 and 1.57, respectively (see Table 2 and Fig. 4). In the case of the crkl ATG and splice morpholinos, the mean nuclei/fibre was 1.49 and 1.77, respectively, at the highest doses. Each of these values is significantly different from those of the wild-type controls,as tested by the Mann-Whitney test (P<0.0001). The range of phenotypes caused by the crk and crkl morpholinos was very similar to that of the Dock morphant embryos described above; in most cases,mononucleate cells extended across the length of the somite, but, in some instances, many of the cells remained rounded but extended processes(Fig. 3D).

To confirm the specificity of the crk and crklmorpholinos, we assessed whether simultaneous injection of synthetic mRNA encoding either gene product could rescue the effects of the respective morpholinos. Embryos injected with crk mRNA at 0.3 μg/μl plus 1 mM crk splice morpholino had a mean number of nuclei/fibre of 2.10(P>0.06 compared to wild type, i.e. not significantly different,but P<0.02 compared to splice morphants); embryos injected with crkl mRNA at 0.3 μg/μl plus 1.5 mM crkl splice morpholino gave a mean number of nuclei/fibre of 2.93 (P>0.2 compared to wild type, i.e. not significantly different, but P<0.0002 compared to splice morphants). Distributions of the embryo means for these groups injected with mRNA are shown in Fig. 6E, overall mean numbers of nuclei/fibre are in Table 3and raw data can be found in Tables 1-6 in the supplementary material.

Significantly, we found that injection of crk mRNA alone (at 0.3μg/μl) increased the overall mean number of nuclei/fibre to 3.00(P<0.004 when compared with wild-type controls), and that crkl mRNA at 0.1 μg/μl also caused an increase to 3.42 nuclei/fibre (P<0.0002). Some fibres in the injected embryos contained eight or more nuclei, exceeding the maximum seven observed in wild-type fibres at the same stage. An example of fibres with an excessive number of nuclei is shown in Fig. 6C. This promotion of fusion was confined to fast-twitch progenitors; staining of slow-muscle fibres showed them to be mononucleated,as in wild type.

Taken together, these data reveal a requirement for Dock1, Dock5, Crk and Crkl during myoblast fusion in the zebrafish embryo. This could imply a partial redundancy of function between the pairs of related proteins;simultaneous knockdown of either Dock1 and Dock5 or of Crk and Crkl, however,resulted in only a slight enhancement of the phenotype relative to that of each single morphant (Table 2). Notably, the highest level of fusion suppression that we observed occurred in embryos injected with both dock5 and crkl morpholinos (see Figs 3 and 4). The failure to suppress fusion completely most probably reflects the fact that, in no cases, was the activity of the various genes completely eliminated, as indicated by our RT-PCR analysis of splicing. Injection of higher combined doses of morpholinos also tended to increase mortality, precluding analysis of the effects of total inhibition of gene function.

Two types of skeletal muscle fibres are formed during zebrafish embryogenesis: mononucleated slow-twitch fibres, the majority of which form a superficial layer on the surface of each somite, and multinucleated fast-twitch fibres, which constitute the bulk of the embryonic myotome. Previous studies have shown that these latter fibres arise by the fusion of myoblasts (Roy et al., 2001),a process that is specifically suppressed in the slow-twitch myoblasts by the Prdm1 transcription factor encoded by the ubo gene(Roy et al., 2001; Baxendale et al., 2004). Here,we have analysed the temporal progression of the fusion process and have begun to dissect its genetic control. At 18 hpf, fast-twitch myoblasts remain unfused but can be seen to be probing their environment by sending out processes; over the next 2 hours, fusion begins to occur and, by 26 hpf, the majority of myoblasts have fused to form multinucleated fibres, containing between two and seven nuclei.

In the Drosophila embryo, multinucleated muscle fibres similarly form by the fusion of myoblasts. In this organism, two distinct types of myoblasts have been identified: fusion-competent cells, which represent the majority of the muscle precursors, and founder cells, which play a key role in directing the type of muscle arising from the differentiation of the fusing myoblasts. When fusion is blocked by mutations that inactivate components of the fusion pathway, the founder cells proceed to differentiate into mononucleated muscle fibres whereas the unfused fusion-competent cells eventually undergo apoptosis and are engulfed by macrophages(Rushton et al., 1995; Doberstein et al., 1997). The first mutation affecting myoblast fusion to be recovered in Drosophila was myoblast city (mbc), which encodes the DOCK180 protein (Rushton et al.,1995; Erickson et al.,1997). Subsequent molecular and genetic analysis has revealed that DOCK180 occupies a pivotal role in a molecular pathway that induces the rearrangement of the actin cytoskeleton in response to interactions between muscle founder and fusion-competent cells mediated by the transmembrane proteins Duf, Rst and Sns.

To investigate whether the molecular control of myoblast fusion is conserved between Drosophila and vertebrates, we identified homologues of DOCK180 in the zebrafish and used morpholinos to block their function. We found that knocking down two members of the DOCK180 sub-group of CDM proteins, Dock1 and Dock5, significantly compromised myoblast fusion in the zebrafish embryo. Interestingly, many of the unfused myoblasts elongated to form mononucleate muscle fibres, reminiscent of the mini-muscles formed by founder cells in Drosophila fusion mutants. However, we have no evidence to suggest that a similar subdivision of myoblasts into founder and fusion-competent cells occurs in the zebrafish embryo. The fact that the suppression of fusion is not complete might well reflect the failure of the morpholinos to achieve total inhibition of gene function: although those directed against splice junctions appeared more-effective than that targeting the translation start site, it seems unlikely that any of these achieved complete abrogation of function, because both dock1 and dock5 transcripts appeared to be maternally supplied. Indeed, given the uniform expression and probable multiple roles of both proteins in a variety of cellular processes, it would not be surprising for their complete inactivation to result in much-earlier developmental arrest. This could also explain why the simultaneous injection of dock1 and dock5morpholinos resulted in only a marginal increase in the severity of the fusion phenotype.

DOCK180 was originally identified on the basis of its physical interaction with CRK, an adaptor protein that has been implicated in coupling DOCK proteins to upstream regulatory molecules(Hasegawa et al., 1996; Galletta et al., 1999; Reddien and Horvitz, 2000). On this basis, Crk has been postulated to play a role in myoblast fusion in Drosophila (Chen and Olson,2004), but, for technical reasons, such a role has not been investigated experimentally. Using the same morpholino-based approach, we asked whether Crk and the related Crkl protein are involved in myoblast fusion in the zebrafish. In both cases, we found that inhibition of expression of the gene products had a significant effect on myoblast fusion. Moreover, we found not only that loss of Crk or Crkl expression inhibited myoblast fusion, but also that overexpression of the mRNAs encoding either protein resulted in an enhancement of fusion. Taken together, these data represent definitive evidence of a role for Crk and Crkl in myoblast fusion.

Whether the requirement for Crk and Crkl in myoblast fusion that we have uncovered involves interaction with Dock1 and/or Dock5 remains to be determined. Based on its role in coupling DOCK180 proteins to upstream regulators in other signalling contexts, one possibility is that Crk and/or Crkl act to couple the transmembrane proteins required for myoblast fusion to the Mbc/DOCK1/5 proteins via other molecules, such as Rolling pebbles (also known as Antisocial; human orthologue known as TANC2)(Chen and Olson, 2001). Against this, however, a recent study has demonstrated that, in Drosophila,mutated forms of Mbc that lack Crk-binding motifs can rescue the mbcmutant phenotype, implying that, at least in flies, direct interaction between Crk and Mbc is not necessary for myoblast fusion(Balagopalan et al., 2006). Similarly, the deficiency in engulfment of apoptotic cells seen in C. elegans lacking the Dock1 orthologue CED-5 can be rescued by mutated forms of the protein that are unable to bind the Crk orthologue CED-2(Tosello-Trampont et al.,2007). Nevertheless, functional analysis has demonstrated that both Crk and Dock1 are required in C. elegans for this phagocytic activity, even if direct physical interaction between the two proteins is dispensable (Wu and Horvitz,1998; Reddien and Horvitz,2000). Thus, by analogy, the requirement for Crk and Crkl in myoblast fusion might be independent of a direct interaction with Mbc/DOCK1/5 proteins. In this context, it is notable that human CRKL and Drosophila Crk have been shown to interact directly with WIP(WASP-interacting protein) (Sasahara et al., 2002; Kim et al.,2007). WIP has recently been shown to have a role in myoblast fusion in Drosophila (Kim et al.,2007; Massarwa et al.,2007), leading to the proposal that Crk acts as a link between the transmembrane protein Sticks and stones and the WIP-WASP-Arp2/3 actin nucleation complex (Kim et al.,2007) in a pathway parallel to that mediated by Mbc. In this regard, it is notable that we found that simultaneous knockdown of Crkl and Dock5 blocked fusion almost completely in the zebrafish embryo. Further analysis of the function of these and other orthologues of the Drosophila myoblast fusion genes using the zebrafish model that we have established will help to elucidate further the functional relationships between these proteins.

While this paper was under review, a related study describing the role of a homologue of Duf/Kirre in zebrafish myoblast fusion was published(Srinivas et al., 2007).

We thank S. Roy and B. P. Srinivas for helpful discussions; S. Elworthy for providing the mylz2:GFP transgenic construct; and the staff of the CDBG Zebrafish Aquaria for animal husbandry. Online statistics help was provided by http://faculty.vassar.edu/lowry/VassarStats.html. This research project was funded by a Medical Research Council (MRC) Programme Grant (G0100151) awarded to P.W.I. The CDBG Zebrafish Aquaria and Confocal facility are supported by an MRC Centre Development Grant (G0400100).