myoblasts incompetent encodes a zinc finger transcription factor required to specify fusion-competent myoblasts in Drosophila

Myogenesis in Drosophila is initiated by the local segregation of founder myoblasts within muscle forming mesoderm. The founder myoblasts have the capacity to implement terminal differentiation through the expression of Dmef2 (Bour et al., 1995; Lilly et al., 1995) and, by fusing with neighbouring fusion-competent myoblasts they seed the formation of multinucleate muscle fibres (Baylies et al., 1998). In the absence of fusion, fusion-competent cells cannot complete myogenesis (Rushton et al., 1995), so founder myoblasts act as local gates on the formation of muscles. The specific characteristics of individual myotubes are determined by the particular combination of transcriptional regulators such as Krüppel (Kr) and S59 (also known as slouch) (Knirr et al., 1999) that each founder expresses (Baylies et al., 1998). Interestingly, however, fusion-competent myoblasts do express some aspects of terminal myogenic differentiation in the absence of fusion. In wild-type embryos, the contractile protein myosin is expressed in fusion-competent cells before they fuse with founders and in mutants where fusion is blocked, myosin is strongly expressed in clusters of unfused myoblasts. This, together with the expression of genes such as sticks and stones (sns), that are specific to fusion-competent cells (Bour et al., 2000) and essential for normal myogenesis, suggests that the fusion-competent myoblasts are a distinct population of muscle-forming cells and that there are regulatory factors that lead to their selective differentiation within the general pathway of myogenesis.

Here we report the first such factor, encoded by the gene myoblasts incompetent (minc). minc is expressed in fusion-competent cells and required for these cells to differentiate. In the absence of minc the fusion-competent cells remain as an immature population of mesodermal cells that cannot contribute to muscle formation. Thus in these mutants, fusion fails and no multinucleate muscles are formed. Interestingly our evidence suggests that minc not only controls the specific properties of fusion-competent myoblasts but that it is also required for myogenic differentiation itself, presumably acting in combination with genes such as Dmef2.

Most of the deficiencies and lethals in the 94CD region used in this work were obtained from the Drosophila stock center at Bloomington and the Szeged stock center in Hungary (Flybase Consortium, 1999). Df(3R)EB6 and Df(3R)r90B were provided by J. Mohler (Mohler and Vani, 1992) and Df(3R)hhE23 was obtained from K. Basler (Basler and Struhl, 1994). The minc[A388] EMS-induced allele was generated on a multiply marked third chromosome following standard techniques. We used Twi-GAL4 (Baylies and Bate, 1996), En-GAL4 (Tabata et al., 1995) and Sca-GAL4 (Budnik et al., 1996) to drive ectopic gene expression.

Molecular biology experiments were performed following standard procedures (Sambrook et al., 1989). LD28538, LD47926 and LD0474 are expressed sequence tag clones from the Berkeley Drosophila Genome Project (BDGP). Cosmids in the 94CD region were obtained from the European Drosophila Genome Project (EDGP). Selection of minc[A388] mutant embryos used for genomic DNA sequencing was based on the use of a third chromosome balancer carrying a Kr-GFP construct, provided by I. Guerrero. For sequencing of minc[A388] genomic DNA, independent PCR amplification products covering the 3 exons of the gene minc were generated and subcloned using the TopoTA cloning kit (Invitrogen Corporation). Three independent PCR reactions were performed to confirm the transition responsible for the generation of a stop codon in the mutant DNA. DNA sequencing was performed with ABI chemistry in an automatic DNA sequencer and by the sequencing facility of Cambridge Bioscience. Custom synthesised oligonucleotides were obtained from ISOGEN (Bioscience BV, Maarsesen, The Netherlands). The conceptual protein sequence was analysed using MOTIF (http://www.motif.genome.ad.jp). Similarity searches for the Minc predicted protein were performed using the BLAST algorithm (Altschul and Lipman, 1990).

Minc protein was ectopically expressed by means of the UAS/GAL4 system. The UAS-minc construct was made by subcloning a 2924n EcoRI-BglII fragment from LD47926 containing the entire minc ORF into the pUAST transformation vector (Brand and Perrimon, 1993). This construct was injected into yw embryos under standard conditions (Spradling, 1986). For ectopic expression of minc, males containing the UAS-minc transgene were crossed to females of different GAL4 lines at 29°C. For rescue experiments females containing Twi-GAL4 combined with minc[A388] were crossed to males UAS-minc; minc[A388] or UAS-Mef2; minc[A388] at 29°C.

Immunocytochemistry was as described previously (Ruiz-Gómez et al., 1997). The following primary antibodies were used: anti-muscle myosin, a gift from Dan Kiehart (Kiehart and Feghali, 1986), anti-Krüppel, a gift from Dave Kosman (Kosman et al., 1998), anti-β-galactosidase (Cappel), anti-Twist (Thisse et al., 1988), monoclonal anti-Hairy 24.1, kindly provided by D. Ish-Horowicz and anti-MEF2 (Bour et al., 1995). minc RNA localisation by means of in situ hybridisations using digoxigenin-labelled RNA probes was performed as described previously (Taylor, 2000). Double RNA in situ hybridisation and antibody staining was done sequentially with minor modifications of the previous procedures.

In situ hybridisations to polytene chromosomes using biotinylated labelled probes were performed as described previously (Ashburner, 1989).

In an EMS mutagenesis and screen for genes affecting myogenesis and muscle patterning in the Drosophila embryo, we isolated a recessive embryonic lethal mutation (A388) in a gene required for myoblast fusion. We mapped the allele meiotically to the interval between ebony (93D2-6) and claret (99B5-9) on the third chromosome. Subsequent analysis of deficiencies showed that A388 fails to complement Df(3R)hhGR2, Df(3R)EB6, Df(3R)M95A, Df(3R)GW2 and Df(3R)23D1, but complements Df(3R)hhE23, Df(3R)5C1 and Df(3R)r90B. A388 complements all known lethals in this region and identifies a new gene required for myoblast fusion, which is located proximal to hedgehog (hh), in the 94D3-13 region of the third chromosome.

A detailed analysis of the mutant phenotype in A388 (see below) revealed that the failure in myoblast fusion reflects a specific defect in the differentiation of fusion-competent myoblasts, while there is no apparent requirement for the gene in the differentiation of founder myoblasts. For this reason we have called the new gene identified by the mutation, myoblasts incompetent (minc).

The GenBank accession number for the minc gene discussed in this paper is AJ311850.

The mutagenesis and screen were designed to incorporate a marker for founder myoblasts (rP298) (Nose et al., 1998) in the genetic background. rP298 is an enhancer trap insertion in the region of dumbfounded (duf) that drives the expression of β-galactosidase in all founder cells, their progenitors and the muscles they give rise to. Thus putative mutants could be assayed rapidly for failures of myogenesis or altered patterns of muscles, by scoring the expression of β-galactosidase. In minc[A388] we found that there was a complete lack of myoblast fusion and all muscles were mononucleated.

To show whether founder cells are specified normally in the mutant embryos, we used antibodies to a variety of transcription factors (e.g. Kr; Fig. 1A-C) that are expressed by subsets of founder myoblasts in normal embryos (Ruiz-Gómez et al., 1997). In all cases, these expression patterns in founder cells were normal. To characterise the phenotype in more detail, we used anti-myosin antibody to visualise the process of myogenesis in wild-type (Fig. 1D) and mutant embryos (Fig. 1E,F). Myosin staining showed that some mesodermal derivatives, including the heart, developed normally (Fig. 1F), but that multinucleate somatic muscles failed to form (Fig. 1E). Fusion also failed to occur in the muscles of the pharynx (Fig. 1F inset). Founder myoblasts appeared to differentiate normally, developing at appropriate positions, expressing myosin, elongating and attaching to the epidermis at their correct insertion sites (Fig. 1E). However, with myosin staining, fusion-competent myoblasts could not be detected at any stage (Fig. 1E,F) and this is in striking contrast to other non fusion mutants in which unfused myoblasts express myosin at high levels (Doberstein et al., 1997; Rushton et al., 1995).

The failure to detect fusion-competent myoblasts with myosin staining in the mutant embryos could reflect either the complete absence of these cells or, alternatively, it could be that the cells are present but have failed to differentiate. Because founder cells form normally in minc embryos, we assume that the segregation process that leads to the separation of a population of founders and fusion-competent cells has probably taken place in these embryos. Thus it seemed likely that the fusion-competent cells would be present but in some undifferentiated state. To explore this possibility further, we used an antibody to Dmef2, a gene that is expressed and required for normal differentiation in all myogenic cells (Bour et al., 1995; Lilly et al., 1994). Staining with this antibody revealed once again a normal pattern of founder cells, but an absence of expression in fusion-competent cells (compare Fig. 1G,H, and insets). This again is in marked contrast to other non-fusion mutants such as myoblast city (mbc) (Fig. 1I and inset).

Because staining with an antibody to Dmef2 did not resolve the issue of the fate of the fusion-competent cells in mutant embryos, we used instead an antibody to twist, which is a general marker for early undifferentiated mesodermal cells (Thisse et al., 1988). In normal embryos twist expression is lost from fusion-competent myoblasts during stages 11 and 12. In the mutant embryos, we find a striking persistence of twist expression in a population of cells whose arrangement is similar to the normal distribution of fusion-competent cells (compare Fig. 2A,B) (Bour et al., 2000). However there are fewer such twist-expressing cells in a minc mutant embryo than those detected by Dmef2 expression in a non fusion mutant such as mbc (compare Fig. 2B and Fig. 1I). Interestingly therefore, it appears that in minc mutant embryos, twist expression persists in only a subset of the putative fusion-competent myoblasts. This expression is maintained throughout the normal period when fusion would be expected to occur. Thus, while in wild-type embryos founder cells are surrounded by fusion-competent myoblasts, with which they fuse, in minc embryos, normally differentiating founder cells are surrounded by an abnormal population of apparently undifferentiated mesodermal cells expressing twist.

To examine the extent to which these twist-expressing cells have been specified, we looked at the expression of two further genes, hairy (h) (Ingham et al., 1985) and sticks and stones (sns), whose products are characteristic of fusion-competent cells in the mesoderm of normal embryos. sns, in particular, is an early marker for this population and a critical determinant of the function of these cells in fusion (Bour et al., 2000). In both cases, we fail to detect any expression in the putative fusion-competent cells surrounding the founder myoblasts in mutant embryos (Fig. 2C,D, h; Fig. 2E,F, sns). sns is also expressed in a second population of fusion-competent cells (Bour et al., 2000) (Fig. 2G), that will contribute to the visceral muscles of the midgut (San Martin et al., 2001). However, we find that these cells do not lose sns expression in minc mutant embryos (Fig. 2H). Nonetheless, minc mutant embryos fail to form syncytial visceral muscles (data not shown).

It seems, therefore, that the process of segregation that generates two populations of somatic myoblasts, founders and fusion-competent cells, in normal embryos, has occurred in the mutant, but while founders develop normally, there is a complete failure of the fusion-competent cells to differentiate. The putative fusion-competent cells remain as an undifferentiated population that continues to express twist and does not express any of the normal markers for the differentiation of this class of myoblasts. It appears that in minc embryos an essential step in the specification of fusion-competent cells has failed to occur, and that they therefore fail to differentiate.

The phenotype that we observe in minc mutant embryos indicates that the novel gene identified by this mutation has an important function in myogenesis. To investigate the nature of the defect in this mutation we analysed the mesodermal phenotype of those deficiencies (see above) that fail to complement A388. In all cases we find a comparable mesodermal defect namely a complete lack of fusion and maintained expression of twist in the somatic mesoderm. These results suggest that minc[A388] is a null allele for minc in agreement with the finding that the transheterozygote minc[A388] with Df(3R)hhGW2 has a phenotype that is indistinguishable from the A388 homozygote.

All the known lethals in the region defined by noncomplementation with deficiencies complement minc. To identify the region of DNA corresponding to minc we obtained cosmid clones spanning 94D3-13. Using in situ hybridisations to polytene chromosomes from deficiencies in the region, we identified cosmids partly or wholly removed by the deficiencies. We looked for transcription units with mesodermal patterns of expression within the genomic region encompassed by such cosmids and identified a fragment 3 kb long (present in cosmids 12A20, 47F9 and 43K19) with such expression. However, this fragment corresponds to DNA that is not removed by the smaller deficiencies that remove minc. The fragment was sequenced and used in BLAST searches of the Drosophila genome to place a proximal limit on the region containing minc (http://www.fruitfly.org). Using the location of hh (see above) as a distal limit, we concluded that minc was within a region of DNA where 34 predicted genes map.

Because the phenotype of minc[A388] suggests a role in regulating patterns of gene expression in fusion-competent myoblasts, we chose to concentrate on two of these predicted genes, CG4677 and CG4639, both of which correspond to predicted transcription factors, (both with putative zinc finger domains). We used the EST (expressed sequence tag) clones identified by The Berkeley Drosophila Genome Project (BDGP) for both genes (LD28538 and LD47926 for CG4677 and LD05474 for CG4639) in in situ hybridisations to embryos, and found that one of them, LD47926 (CG4677), has a mesodermal pattern of expression whose profile coincides with sites where minc function is required. Furthermore, LD47926 is not expressed in embryos homozygous for deficiencies that remove minc (data not shown). Probing a northern blot containing RNA samples from embryos, larvae and adults with the LD47926 clone we identified a unique transcript 3.0 kb long, which is expressed in embryos (Fig. 3B). The putative minc cDNA sequence was determined by sequencing the EST clone LD47926. The cDNA is 3192 nucleotides long, contains an open reading frame (ORF) of 866 amino acids and untranslated regions (UTR) on the 5′ and 3′ ends of 256 and 335 nucleotides respectively. The sequence of the cDNA differs from the predicted translation for the CG4677 gene. Thus the gene CG4677 has 3 exons instead of the 5 predicted, as predicted introns 1 and 4 are included in the cDNA sequence. The putative minc cDNA encodes a transcription factor that contains four C2H2-type zinc finger motifs (Fig. 3A) whose closest homologues are the zinc finger protein Gli1 (Kinzler et al., 1988) and the Cubitus interruptus protein (Orenic et al., 1990).

To verify that cDNA LD47926 corresponds to minc, we sequenced genomic DNA encoding the gene CG4677 obtained from minc[A388] embryos and found a change G to A in nucleotide 959 of the coding region that changes tryptophan 320 into a stop codon (Fig. 3A, shaded box, underlined). The truncated protein that is produced in minc mutant embryos lacks the four C2H2 finger domains and most probably its ability to bind DNA. This result is in agreement with minc[A388] being a null allele for minc as suggested (see above) from its mutant phenotype.

The earliest expression of minc that we detect is in cells of the visceral mesoderm – initially at stage 10 in a small patch of cells at the posterior tip of the embryo, that probably corresponds to the primordium of the caudal visceral mesoderm (Kusch and Reuter, 1999) and then in segmentally repeated clusters of cells that correspond to the bagpipe (bap)-expressing progenitors of the trunk visceral mesoderm (Azpiazu and Frasch, 1993) (Fig. 4A). These patches are joined ventrally by expression in the mesodermal cross bridges (Bate, 1993) (Fig. 4A, arrowheads). Expression in the visceral mesoderm includes both founders and fusion-competent myoblasts (Fig. 4B). By stage 11, expression starts in cells of the somatic mesoderm and begins to decline in cells of the visceral mesoderm (Fig. 4C). Expression is predominantly in cells that do not express the rP298 marker for founder cells. There is probably residual expression of minc in rP298-expressing cells but it is clearly at a lower level than in the adjacent cells with which they will fuse (Fig. 4D). There is strong expression in the somatic mesoderm until stage 13 (Fig. 4E). Expression begins to decline in stage 14 and, apart from residual expression in a few cells, has disappeared by late stage 15. At this stage, however, minc is detectable in mesodermal cells of the gonad (Fig. 4F).

Mutations in twist, which eliminate all mesodermal derivatives also completely lack expression of minc (Fig. 5A,B), so confirming its exclusively mesodermal pattern of expression. Previous work has shown that founder cells segregate from the somatic mesoderm by a process of lateral inhibition that is mediated by the neurogenic genes and the activation of the Notch signalling pathway in neighbouring cells, that are thereby prevented from becoming founders themselves. In Notch mutant embryos, there is an overproduction of founder cells at the expense of other adjacent cells (Corbin et al., 1991). It seems likely that the cells that are inhibited from becoming founders are the cells that will enter the other myoblast class, namely the fusion-competent cells. This view is reinforced by the finding that sns expression, which is characteristic of these cells, is reduced in Notch mutant embryos (Bour et al., 2000). If this is so, and if minc is involved in the specification of fusion-competent cells, then we would expect that its expression too would be reduced in neurogenic mutant embryos. Accordingly, we assayed the expression of minc in mutants for Notch and Delta. In both cases there is a striking reduction in the expression of minc (Fig. 5C,D, N; E,F, Dl). There is some persistent minc expression in these embryos which is likely to be in cells of the visceral mesoderm.

To verify that the product encoded by the LD47926 cDNA corresponds to the missing function in minc mutant embryos, we used the GAL4/UAS system (Brand and Perrimon, 1993) to drive expression of this cDNA in the mesoderm of minc[A388] embryos. Using Twi-GAL4 (Baylies and Bate, 1996) to drive expression throughout the mesoderm, we found substantial rescue of all aspects of the minc mutant phenotype, including fusion to form multinucleate muscle fibres, expression of myosin (compare Fig. 6A,B,C) and reduction of twist expression in fusion-competent cells (compare Fig. 6D,E,F). The rescue is incomplete and the reason for this is likely to be the use of the Twi-GAL4 driver. The expression of minc itself downregulates twist in those cells that express it and therefore the effectiveness of the driver is reduced.

We used En-GAL4 (Tabata et al., 1995) and Sca-GAL4 (Budnik et al., 1996) to drive ectopic expression of minc in the ectoderm. Ectodermal Minc leads to the expression of Dmef2 in the CNS (En-GAL4Fig. 6G) and sns, which is normally confined exclusively to fusion-competent cells, in the epidermis (En-GAL4Fig. 6H) and CNS (Sca-GAL4Fig. 6I).

From the rescuing effects of driving minc expression in the mesoderm and the ectopic activation of genes normally expressed in fusion-competent myoblasts we conclude that cDNA LD47926 is responsible for functions that are lost in minc[A388] mutant embryos.

Dmef2 is required for the normal differentiation of all myogenic cells (Bour et al., 1995; Lilly et al., 1995). However, we know that sns expression is not lost in Dmef2 mutants (own observations) (Bour et al., 2000) and therefore the early segregation and specification of fusion-competent cells appears to be independent of Dmef2. In addition, the downregulation of twist in fusion-competent cells occurs normally in Dmef2 mutants (data not shown). Nonetheless, in the absence of Dmef2, fusion fails and myosin is not expressed (Bour et al., 1995; Lilly et al., 1995). Thus the failure of late steps in the differentiation of fusion-competent cells in minc mutant embryos might solely be caused by the absence of Dmef2 from these cells. To test this idea, we drove expression of Dmef2 in the mesoderm of minc mutants using Twi-GAL4. In this experiment we find that there is no rescue of fusion or of myosin expression and that twist expression is maintained at a high level in the fusion-competent cells (data not shown). Thus in the absence of minc, Dmef2 is incapable of eliciting myogenic differentiation from this population of mesodermal cells. We conclude that Minc is not only required for the initial specification of fusion-competent myoblasts but is also essential for the activation of terminal differentiation in these cells.

There is an essential subdivision of the population of muscle-forming cells during myogenesis in Drosophila (Fig. 7) (Baylies et al., 1998). Specific cells are selected locally to seed the assembly of particular muscles. These founder myoblasts endow muscles with characteristic properties such as size, innervation and orientation, by the operation of the particular constellation of transcription factors that they express. The other population of cells that contribute to the muscles are the fusion-competent myoblasts, which fuse with the founder myoblasts so that the local initiation of muscle formation by a founder leads to the development of a specific multinucleate muscle fibre. This subdivision of the muscle-forming cells into myoblasts of two kinds is essential for fusion. There is an intrinsic asymmetry to the fusion process that enables the founders to act as local gates on the myogenic process: founders will not fuse with each other, but only with fusion-competent myoblasts, fusion-competent cells will only fuse with founders. This asymmetry is driven by the differential expression of genes encoding two Ig domain proteins that are essential for fusion: founder myoblasts express duf (Ruiz-Gómez et al., 2000), whereas fusion-competent cells express sns (Bour et al., 2000). The diversification of myoblasts applies not only to the formation of somatic (body wall) muscles, but also to the myoblasts that contribute the visceral muscles lining the midgut (San Martin et al., 2001).

The specialisation of the fusion-competent cells revealed by their exclusive expression of sns, suggests that these myoblasts are not simply a naïve population of muscle-forming cells but that they have a specific programme of differentiation which must be implemented if myogenesis is to proceed. This programme includes the expression of sns and hairy, the down-regulation of twist, the maintenance of Dmef2 expression and the synthesis of contractile proteins such as myosin. Some of these steps are specific to fusion-competent cells: the expression of sns and hairy; others such as the down regulation of twist, the maintained expression of Dmef2 and the synthesis of contractile proteins, appear to be steps in a pathway of differentiation that is common to all myogenic cells.

In this paper we have identified a gene, minc, that is required for the proper specification of the population of fusion-competent myoblasts that contributes to the body wall muscles. minc is upregulated in fusion-competent cells and downregulated in founders and the key to this differential expression is the activation of the N signalling pathway in the fusion-competent cells as the progenitors of founder myoblasts segregate in the somatic mesoderm (Fig. 7). Our experiments show that N signalling leads directly or indirectly to the maintained expression of minc in fusion-competent cells. minc in turn is required for the expression of genes that are specific to the fusion-competent cells, such as sns and hairy, and it is also required for the implementation of myogenic differentiation in these cells. In the absence of minc, Dmef2 is not maintained in fusion-competent cells but the loss of Dmef2 expression is not solely responsible for the failure of myogenic differentiation in fusion-competent cells that we observe in minc mutant embryos. If Dmef2 alone were sufficient to activate myogenic differentiation in these cells then it would be possible to rescue this aspect of the minc phenotype simply by providing the potential fusion-competent cells with Dmef2. However, myosin expression in the fusion-competent cells is not rescued by Dmef2 in the absence of minc and this shows that minc is an essential component in implementing the terminal differentiation of this population of muscle-forming cells, presumably in combination with Dmef2.

The fact that minc is required not only for specific patterns of gene expression but also for the implementation of what might be considered the common pathway of myogenesis in the somatic fusion-competent myoblasts is interesting and prompts us to speculate as to why there should be such a specific control mechanism for terminal myogenesis in these cells. We know that specific muscle properties within the somatic muscle lineage are regulated by the expression of particular transcription factors in subsets of muscle founder cells. However, it is likely that there are general characteristics of the somatic muscles that are not shared with other differentiated myogenic cell types such as the visceral and cardiac muscles. It might be that the implementation of these characteristics in the somatic lineage could depend, at least in part, on a particular programme of myogenic differentiation implemented only in the somatic fusion-competent myoblasts under the control of minc.

However, minc is also expressed in the visceral mesoderm. Here too it is expressed at relatively higher levels in fusion-competent cells than in founders, but curiously it is not required for sns expression in these cells, nor for the expression of Dmef2 or markers of terminal differentiation such as myosin. In addition, the control of minc expression is probably exerted through a different mechanism. Instead of a response to N activation it seems likely that minc expression is controlled by general regulators of visceral mesoderm differentiation such as bap. Thus, minc expression appears in the bap-expressing clusters and in the region of bap expression in the caudal visceral mesoderm, and in mutations that reduce bap expression [such as loss of hedgehog function (Azpiazu et al., 1996)], there is a concomitant reduction in minc (data not shown). However, in the absence of minc, hairy expression is lost from the visceral fusion-competent myoblasts and visceral myoblasts fail to fuse to form either longitudinal or circular syncytial muscles (data not shown). Clearly minc has some function in the development of the visceral mesoderm but in these cells it does not operate as an essential regulator of terminal myogenic differentiation as it does in the somatic mesoderm. The fact that fusion fails in the visceral mesoderm, despite the expression of both duf and sns in founders and fusion-competent myoblasts respectively indicates that additional factors, regulated by minc are required, perhaps specifically in the fusion-competent myoblasts if fusion is to occur.

minc encodes a putative zinc finger transcription factor whose closest homologues are Gli proteins in vertebrates (Kinzler et al., 1988) and the Cubitus interruptus protein (Orenic et al., 1990) in Drosophila. Proteins of this class can act both as activators and repressors of transcription (Ruiz i Altaba et al., 1997). Our experiments indicate that Minc acts to down regulate the expression of twist and to activate the expression of sns, hairy and Dmef2. Since ectopic expression of minc can drive expression of both sns and Dmef2 in ectodermal derivatives and rescues the mutant phenotype in fusion-competent myoblasts we conclude that minc acts specifically within the myogenic lineage to endow fusion-competent myoblasts with their essential properties. Although minc appears to act at a high level in the hierarchy of regulators that control myogenesis it does not on its own convert cells of a different lineage – the ectoderm – to the myoblast cell fate. Some aspects of myoblast differentiation are initiated, such as sns and Dmef2 expression but it is clear that additional factors, presumably present only in the mesoderm, are necessary for terminal muscle differentiation. This accords with the view that minc acts as a regulator that is intrinsic to the myogenic pathway, modifying the fate of cells in the muscle lineage to form a particular kind of myoblast.

Furlong et al. (Furlong et al., 2001) have reported a gene gleeful, which is identical to the gene minc described in this paper.

We are grateful to Susan Abmayr, Hanh Nguyen, David Ish-Horowicz and Dan Kiehart for the gift of cDNAs and antibodies and K. Basler, J. Mohler, I. Guerrero and The Bloomington and Szeged stock Centres for fly stocks. We thank Susan Rolfe and Eva Caminero for their expert technical assistance, Encarna Madueño for genomic sequences, and BDGP and EDGP for EST and cosmid clones. We are grateful to our colleagues in Cambridge and Madrid for their advice and their many helpful comments on the manuscript. This work is supported by a grant from the Wellcome Trust (052879) to M. B. and grants from the Dirección General de Investigación Científica y Técnica (PB98-0682) to J. Modolell; Dirección General de Investigación, Ministerio de Ciencia y Tecnología (BMC2000-1133) to M. R.-G. and by an Institutional grant from Fundación Ramón Areces to the Centro de Biología Molecular Severo Ochoa.