|Home | About | Journals | Submit | Contact Us | Français|
Mrf4 (Myf6) is a basic helix-loop-helix (bHLH) myogenic regulatory transcription factor (MRF) family which also contains Myod, Myf5 and myogenin. Mrf4 is implicated in commitment of amniote cells to skeletal myogenesis and is also abundantly expressed in many adult muscle fibres. The specific role of Mrf4 is unclear both because mrf4 null mice are viable, suggesting redundancy with other MRFs, and because of genetic interactions at the complex mrf4/myf5 locus. We report the cloning and expression of an mrf4 gene from zebrafish, Danio rerio, which shows conservation of linkage to myf5. Mrf4 mRNA accumulates in a subset of terminally differentiated muscle fibres in parallel with myosin protein in the trunk and fin. Although most, possibly all, trunk muscle expresses mrf4, the level of mRNA is dynamically regulated. No expression is detected in muscle precursor cell populations prior to myosin accumulation. Moreover, mrf4 expression is not detected in head muscles, at least at early stages. As fish mature, mrf4 expression is pronounced in slow muscle fibres.
Mrf4 shows biphasic expression in rodent muscle development, being primarily expressed in terminally differentiated muscle fibres (Bober et al., 1991; Hinterberger et al., 1991). However, early expression is in the mouse ventrolateral dermomyotome, prior to muscle terminal differentiation (Summerbell et al., 2002). This early expression has been suggested to permit myogenesis of a subset of myotomal muscle fibres in Myf5;Myod double knockouts (Kassar-Duchossoy et al., 2004). Such a myogenic function has been associated with the presence of a serine-rich chromatin-remodelling region shared with Myod and Myf5 (Bergstrom and Tapscott, 2001). Mrf4 null mice are viable, although genomic manipulations cause suppression of the linked Myf5 gene, making interpretation of phenotypes difficult (Braun and Arnold, 1995; Olson et al., 1996; Patapoutian et al., 1995; Zhang et al., 1995). Mrf4 null mice have defects in neural control of adult gene expression (Thompson et al., 2005). Regulation at early and late stages is controlled by distinct genetic elements (Carvajal et al., 2001; Fomin et al., 2004; Pin et al., 1997). In adult muscle, Mrf4 expression persists as the most abundant MRF and appears to be differentially regulated between different fibre types (Miner and Wold, 1990; Pin and Konieczny, 2002; Walters et al., 2000). To gain insight into the ancestral role of Mrf4, we analysed the sequence and expression of mrf4 in zebrafish.
We cloned zebrafish mrf4, sequenced both a BAC and cDNA clones and assembled an open reading frame (Fig. 1A). The sequence agreed with the latest release of the zebrafish genome project (Zv6), which shows mrf4 linked to myf5. Zebrafish mrf4, like other MRF genes, has three exons separated by two introns. An open reading frame encoding 239 amino acids was identified (as in Fugu Mrf4 and compared with 240 amino acids in Xenopus Mrf4 and 242 amino acids in chicken, mouse, rat and human MRF4). Zebrafish Mrf4 is highly homologous to MRF4 from other vertebrates (Fig. 1B; ~56% similarity, comparable to the 55% and ~73% similarity between zebrafish myf5 and myoD, respectively, with these genes in amniotes (Coutelle et al., 2001; Weinberg et al., 1996).
Homology is especially high in the basic helix-loop-helix domain that functions in the heterodimerisation and DNA binding of MRFs and in the serine-rich area which may be associated with ability to remodel chromatin (Fig. 1B; Gerber et al., 1997). The divergence between proteins generally matches known evolutionary relationships (Fig. 1C). No duplicate of the mrf4 or myf5 genes is apparent in the current genome release. Moreover, despite the proposed genome duplication in teleosts, myod and myogenin also do not show duplication.
During 3′RACE a second mrf4 transcript was also obtained, which showed an alternate third exon (Fig. 1D). Analysis of the genomic sequence revealed the insertion of a fish-specific DANA retrotransposon element (Izsvak et al., 1996) into the second intron of mrf4, presumably providing an alternative splice acceptor site. This transcript predicts a truncated Mrf4 protein lacking the third exon and with five novel amino acids VPLLI at the C terminus. In situ mRNA hybridisation with a probe against the canonical third exon gives muscle-specific signal. A probe to the alternative third exon was not made due to predicted cross-reaction to widespread DANA transcripts.
Two muscle fibre types form in the first few hours after somite formation: slow fibres arise first from adaxial cells adjacent to the notochord and fast fibres form later from more lateral somitic cells (Devoto et al., 1996). Mrf4 mRNA is not detected by in situ hybridisation at the 2 cell, 64 cell, 1k cell, dome or tailbud stages, consistent with the lack of expression at 1-4 somite stage (www.zfin.org). Mrf4 mRNA is first detected in cells adjacent to the notochord in nascent somites at the 5 somite stage, after actin (5s; Fig. 2A). As these rostral somites mature, mrf4 expression is initially strong for around 3 hours. and then declines (Fig. 2B-F). From about the 6s stage, one somite is formed every 30 minutes, so six somites continue to show strong adaxial mrf4 expression (Fig. 2E-H). Within the presomitic mesoderm (PSM), mrf4 mRNA is absent from adaxial cells, which express myod and the slow muscle determination factor prdm1 (Fig. 2B,F,H). Mrf4 mRNA accumulates much later than early differentiation markers such as actin (actc; Fig. 1C), but also after later ones, such as myosin mRNA (data not shown). Co-localization of myosin heavy chain protein (MyHC) and mrf4 mRNA at 13s reveals that mrf4 is detected exclusively in terminally differentiated slow fibres at this stage (Devoto et al., 1996; Fig. 2G). Similarly, in frogs, Xmrf4 is co-localised with MyHC and absent from the PSM (Della Gaspera et al., 2006).
Early muscle differentiation is ablated by antisense morpholino knockdown of myf5 and myod (Hammond et al., 2007). Such treatment prevents both mrf4 mRNA and MyHC accumulation, consistent with expression of mrf4 only in differentiated muscle (Fig. 2S). This contrasts with the ability of murine Mrf4 to express in Myf5:Myod double mutants (Kassar-Duchossoy et al., 2004).
That early mrf4 mRNA is in differentiated slow muscle cells is supported by its dependence on Hedgehog (Hh) signalling, which drives slow myogenesis (Blagden et al., 1997; Du et al., 1997). In the Hh signalling mutant gli2a/yot (Karlstrom et al., 1999), mrf4 mRNA is missing at 12s (Fig. 3A). Loss of slow muscle in the prdm1 mutant (Baxendale et al., 2004) is also accompanied by loss of mrf4 expression. Thus, mrf4 appears to be transiently expressed for around three hours in slow muscle cells of each somite as it forms.
The transient expression of mrf4 led us to ask whether all slow muscle fibres express mrf4 similarly. Flatmounts of wild type embryos show that mrf4 expression is only in some nascent fibres, those that are largest and located nearest to the midline (Fig. 2G inset). This suggests that muscle pioneer (MP) fibres express mrf4, because MPs were originally described as the first fibres to differentiate in the somite and remain close to notochord (Felsenfeld et al., 1991). In contrast, the slow fibres that translocate to the lateral myotome surface, precursors of the superficial slow fibres (SSF; Devoto et al., 1996), do not appear to express mrf4 as highly as MPs, at least at these early developmental stages.
To test the hypothesis that MPs alone express mrf4 at early stages, we examined sonic hedgehog a (shha; syu) mutants that lack MPs but have residual SSF differentiation. At 15s, shha mutant embryos contain medial slow muscle but lack eng2a expression, a marker of MPs (Fig. 3B; Coutelle et al., 2001; Schauerte et al., 1998). Despite the residual SSF precursors in shha mutants, no mrf4 expression was detected prior to 18s (Fig. 3C; 43/159 embryos from a shha heterozygote cross). Similar results were observed in spadetail (tbx16 mutants, 23/77 embryos) and floating head (flh mutants, 11/32 embryos) that also lack MPs but retain SSF cells (data not shown). In the hypomorphic prdm1 mutant, prdm1tp39(ubo), which lacks MPs but has residual slow fibres at 10s (Baxendale et al., 2004; Roy et al., 2001), the high level mrf4 in nascent MPs is also missing (Fig. 3D). During late somitogenesis, when MPs are forming in the caudal most somites, strong mrf4 signal is absent from shha, prdm1 and gli2a mutants, but visible in siblings (Fig. 3E,G,H arrowheads). More anteriorly in 24hpf wild type embryos, mrf4 mRNA is absent from the location of the MPs at the dorsoventral midline, and this gap in expression is missing in shha, prdm1 and gli2a mutants (Fig. 3E and data not shown). In contrast mrf4 expression in slow fibres is not affected in fgf8 morphants (Fig. 3J; 27/27 embryos) or mutants (17/17 embryos; Fig. 3K). Taken together, these data suggest that mrf4 is transiently strongly expressed in MP fibres, in contrast to SSF precursors, as they begin to accumulate muscle MyHC protein adjacent to the midline.
Mrf4 is expressed in SSF at later stages. SSFs differentiate adaxially and then trasverse the somite to form a monolayer of mononucleate fibres at the lateral myotome surface (Devoto et al., 1996). SSF and fast precursors express little, if any, mrf4 prior to 18s when mrf4 has declined in MPs of rostral somites (Fig. 2A-H). However, in prdm1 hypomorphic mutants, which have some residual slow muscle (Roy et al., 2001), very low levels of mrf4 mRNA are detected (Fig. 3D). Interestingly, heterozygous gli2a mutants also show reduced mrf4 mRNA despite the presence of slow muscle, raising the possibility that Hh signalling can drive mrf4 expression in SSFs (Fig. 3A). By 21s, mrf4 mRNA is up-regulated in differentiated fibres of anterior somites (Fig. 2I,). At this stage, differentiated SSF precursors have spread dorsoventrally to flank the neural tube and hypochord and commence lateral migration. Mrf4 is detected co-localized with MyHC in migrating SSF precursors of shha embryos, which lack MPs (Fig. 3F). Such mutants have no fast myosin protein at this stage. When the SSFs reach the lateral surface of the somite, mrf4 is readily detected in the SSF monolayer at 24 hpf and beyond (Fig. 2J,M,R). High level mrf4 mRNA is not observed in MPs at 24 hpf (Fig. 2J). Thus, the SSFs express mrf4 later than MPs, starting as they begin migration.
Mrf4 is also expressed in fast fibres. Mrf4 mRNA is not detected in fast fibre precursors marked by myod or myog mRNA in the lateral somite at early stages (Fig. 2B,D,H). However, at 21s, mrf4 mRNA is detected in cells of the lateral somite (Fig. 2I). Mrf4 mRNA in these cells accumulates shortly after differentiation markers such as myosin light chain mylz2 mRNA (data not shown). By the end of somitogenesis, mrf4 mRNA is clearly detected in dorsal and ventral somitic cells that are not MPs (Figs (Figs2K2K and and3E).3E). Both wild type, and also gli2a mutant embryos which entirely lack slow muscle fibres, show mrf4 expression in fast muscle (Figs (Figs2J2J and 3H,I). Thus, mrf4 is expressed in fast fibres. Fgf8 signalling is required for differentiation of fast but not slow fibres (Groves et al., 2005; Hammond et al., 2007). In fgf8/ace mutants, the amount of mrf4 mRNA and differentiated fast fibre is substantially reduced compared to their siblings (17/56 embryos; Fig. 3K). After somitogenesis is complete around 24hpf, levels of detectable mrf4 mRNA decline, although low levels of mrf4 mRNA remain in muscle fibres at 48hpf and 5d of development, more prominently in slow fibres (Fig. 2L,M,Q,R).
Mrf4 mRNA is first detected in pectoral fin muscles at 48hpf (long pec stage) in two muscle masses (dorsal and ventral; Fig. 2N). Myogenin (myog) mRNA and MyHC protein are also expressed at this stage in a similar pattern (Fig. 2O; Neyt et al., 2000), whereas myod mRNA is detected in the fin buds earlier, from around 30hpf (Neyt et al., 2000; Weinberg et al., 1996). Mrf4 expression in fin muscles remains strong at 3d of development (Fig. 2P). In rodents, Mrf4 is expressed after other MRFs, during secondary fibre myogenesis (Bober et al., 1991; Hinterberger et al., 1991). In contrast to fin expression, we found no mrf4 mRNA in head muscles at least until 5d of development (Fig. 2N). On the contrary, myog (Fig. 2O) and muscle structural genes (Schilling and Kimmel, 1997; Xu et al., 2000 and data not shown) express in head muscle from around 48hpf. Our inability to detect mrf4 mRNA in head muscle, contrasts with a recent report of mrf4 mRNA in a similar pattern to myogenin mRNA in head muscles at 3d (Lin et al., 2006). Lack of early Mrf4 expression in cranial muscle has also been described in embryonic mice (Bober et al., 1991). At no stage did we observe expression of mrf4 in heart, smooth muscle or any other tissue except for skeletal muscle.
We have recently shown that muscle differentiation involves two steps: initial expression of early muscle genes followed by a second step regulating myofibril assembly (Hinits & Hughes, 2007). Zebrafish mrf4 expression parallels assembly of myosin protein into myofibrils. Expression is not detected in presomitic mesoderm, even though adaxial cells are terminally differentiating in this location as indicated by actin expression. Mrf4 is dynamically expressed in subsets of terminally differentiated muscle fibres in both trunk somites and fins, but is not detected in head or fin muscle precursors, or in somitic cells lacking myosin. Transient expression in a subset of nascent slow fibres is followed by expression in both fast fibres and, more strongly, superficial slow. All manipulations examined that reduce myosin expression also diminish mrf4 mRNA accumulation.
Mutant lines shhatbx392 (Schauerte et al., 1998), gli2aty119 (van Eeden et al., 1996), flhn1(Blagden et al., 1997), sptb104 (Kimmel et al., 1989), prdm1tp39 (van Eeden et al., 1996), fgf8ti282a (Whitfield et al., 1996) were maintained on King’s wild type background and staging and husbandry were as described (Westerfield, 1995).
A zebrafish BAC library was screened with a zebrafish myf5 cDNA probe (Coutelle et al., 2001) and a single clone was isolated. Sequence analysis revealed no mrf4 sequence in the approximately 10kb upstream of myf5 (O. Coutelle and P.W.J.R., unpublished data). A probe generated from the end of this BAC was used to rescreen the library a a single new clone isolated. Multiple restriction digests of this BAC were analysed by Southern blotting using mouse and Fugu rubripes mrf4 probes in order to locate mrf4; an appropriate fragment was subcloned and sequenced. More sequence reads were added by sequencing a PCR product containing exon 1 and traces from the ongoing zebrafish genome project. DNA from 15s and 25s zebrafish embryos was extracted as described (Westerfield, 1995). Total RNA was extracted by using TRIZOL (Invitrogen) according to the manufacturer’s protocol. Genomic PCR products were amplified using primers made according to the primary genomic reads obtained from the BAC library (FW: 5′TGAATCTGAAGCCCCGCAAC3′ and REV: 5′ATTGCTCTGCTCCTGCTCATCC3′). A full-length cDNA was obtained by performing 3′ and 5′ RACE using Gene Racer kit (Invitrogen). Other genomic sequences were obtained from the zebrafish genome project and a sequence assembly mRNA made resolving ambiguity by reference to the original traces (GenBank: AY335193). Upon the release of Zv2 genome assembly all bases matched the genome sequence.
In situ mRNA hybridisation was conducted as previously described (Groves et al., 2005). A mrf4 riboprobe was made by cloning the first exon PCR product into pGEM-T easy (Promega). For antisense probe plasmid was linearised with SpeI and T7 RNA polymerase used. Sense probe gave no signal above background. A second mrf4 probe was made by amplifying the full sequence encoding Mrf4 with the reverse primer harbouring a T3 RNA polymerase binding site. This probe and a mixture of the two probes gave similar results. Other probes used were myod, myogenin (Weinberg et al., 1996), eng2a (Ekker et al., 1992), mylz2 (Xu et al., 2000), prdm1 (Baxendale et al., 2004), actc (I.M.A.G.E clone 7284336). Probes were fluorescein or digoxigenin tagged by using Roche labelling mix. Antibody labelling of wholemount or sectioned embryos was performed as described (Groves et al., 2005), using primary antibodies: A4.1025 (IgG2a, recognises all MyHC, used 1:5; Dan-Goor et al., 1990), F59 (IgG1, recognises slow MyHC, used 1:5; Devoto et al., 1996). Antisense morpholinos to myf5, myod and fgf8 were used as described (Groves et al., 2005; Hammond et al., 2007).
Thanks to Oliver Coutelle and Carlos Moreno de Barreda, who contributed to the cloning of zebrafish mrf4. We thank M. Westerfield, Z. Gong, F. Stockdale and E. Bandman for reagents. This work was supported by the MRC and the Institute of Cancer Research. YH received additional support from B’nai Brith, London, UK.