ladybird determines cell fate decisions during diversification of Drosophila somatic muscles

Specification of distinctive myoblast lineages appears universal in the generation of the complex muscle pattern in both invertebrate and vertebrate embryos (Bate, 1993; Abmayr et al., 1995; Cossu et al., 1996). To date, little is known about the mechanisms that underlie the diversification of muscle precursors. Recent work in Drosophila has revealed some components of the genetic pathways governing early steps of myogenesis (Carmena et al., 1995; Ruiz-Gomez et al., 1997; Ruiz-Gomez and Bate, 1997; Nose et al., 1998; Carmena et al., 1998) showing that the Drosophila embryo is a powerful model to study the specification and patterning of muscle fibers.

Surprisingly, the mechanisms involved in the execution of early myogenic programme are similar to those of neurogenesis. In the somatic mesoderm, as in the neuroectoderm, the expression of a proneural gene lethal of scute (l’sc) delimits clusters of equivalent cells (Carmena et al., 1995). Due to lateral inhibition, involving coordinate action of neurogenic genes (Corbin et al., 1991; Bate et al., 1993), a single cell segregates from each proneural and promuscular cluster and becomes, respectively, a neuroblast or a muscle progenitor. During subsequent division, both these cell types use a common mechanism of asymmetric transmission of the Numb (Nb) protein, which seems crucial for the final diversification of both the neural and the muscular fates (Rhyu et al., 1994; Spana et al., 1995; Ruiz-Gomez and Bate, 1997, Carmena et al., 1998). Daughters of the Drosophila muscle progenitors, the founder cells, are thought to represent a specialized class of myoblasts able to recruit neighbouring mesodermal cells, fuse with them and produce a diversity of syncytial muscle fibers (Bate, 1990; Dohrmann et al., 1990; Abmayr et al., 1995). It has been proposed that the distinct size, shape and epidermal insertion sites of muscle fibers result from the expression, in a given founder cell, of one or a combination of putative muscle identity genes such as: S59 (Dohrmann et al., 1990; Carmena et al., 1995), even skipped (eve) (Frasch et al., 1987; Ruiz Gomez and Bate, 1997), vestigial (vg) (Wiliams et al., 1991), nautilus (nau) (Michelson et al., 1990), apterous (ap) (Bourgouin et al., 1992), Krüppel (Kr) (Ruiz Gomez et al., 1997) and muscle segment homeobox (msh) (D’Alesio and Frasch, 1996; Nose et al., 1998). Recent studies appear to confirm this hypothesis showing that the gain or loss of function of ap (Bourgouin et al., 1992), nau (Keller et al.,1997), Kr (Ruiz Gomez et al., 1997) and msh (Nose et al., 1998) alter the characteristics of at least a subset of muscle precursors in which they are expressed. However, low expressivity of observed phenotypes indicates that the majority of muscle fibers are specified by a combination of transcription factors rather than a single one (Ruiz Gomez et al., 1997; Nose et al., 1998). Since most putative muscle identity genes display lineage-specific expression in the differentiating central nervous system (CNS), it suggests that they may represent a further common element between the myogenic and neurogenic pathways. Indeed, the phenotypes of msh− embryos (Ishiki et al., 1997; Nose et al., 1998) indicate that this gene functions to specify positional identities of both muscle and neural precursors.

In addition to the intrinsic mesodermal programme, the formation of the correct muscle pattern requires inductive signals from the epidermis. The secreted molecule encoded by the segment polarity gene wingless (wg), required for the formation of a subset of muscle founder cells (Baylies et al., 1995), is a good candidate for this function. Wg gene product can act across germ layers (Baylies et al., 1995; Jagla et al., 1997b) and influence the myogenic cascade but the mesodermal targets of Wg signaling are not yet determined.

In this study, we analyse the muscle identity functions of the ladybird (lb) genes, which belong to a distinct family encoding homeodomain-containing transcription factors (Jagla et al., 1994, 1997a,b). We show that lb are expressed in the promuscular cluster, progenitor, founder cell and syncytial precursor of only one larval somatic muscle, the segmental border muscle (SBM, muscle 8). We characterise the requirements for the segregation of the lb-positive progenitor and its particular features including the generation of two non-differentiating myoblasts, the lateral adult muscle precursors (LaPs). From the analysis of gain- and loss-of-function phenotypes, we infer that lb acts in concert with the myogenic pathway and is indispensable for the identity of both the SBM and the LaPs.

The following fly strains were used: tin^EC40, tin^Df(3R)GC14 (Azpiazu and Frasch, 1993; Bodmer, 1993), Df(1)sc¹⁹ (Carmena et al., 1995), nb¹ (Uemura et al., 1989), hh^9K (Heemskerk and DiNardo, 1994), wg^IL114 (Bejsovec and Martinez-Arias, 1991) and a set of neurogenic mutants, N^55e11, bib^ID05, mam^Z3, Df(3R)E(spl)^Bx22, Dl^RevF10 and neu^IF65. hh^9K and wg^IL114 are temperature-sensitive alleles that behave as wild type at 18°C and as null at 29°C. To avoid the influence of S59 and other genes located distally to lbe, as lb− background, we used transheterozygous embryos, referred to as lb-deficiency and obtained from the cross of Df(3R)e-D7; tin rescue males with females carrying a shorter Df(3R)e-F1 deficiency (Azpiazu and Frasch, 1993; Jagla et al., 1997b). A Df(3R)eD7; tin rescue stock in which both Lbe and Lbl gene products are absent and tin activity is restored was kindly provided by M. Frasch. The balancer chromosomes were marked with a twi-lacZ or a ftz-lacZ P-element allowing distinction between mutant and wild-type embryos.

Uniform ectopic lb expression was induced in double transgenic hs-lbe;hs-lbl embryos (Jagla et al., 1997b) by the heat-shock treatment (15 minutes at 37°C in water) at the time when the muscle progenitors segregate (5 hours AEL). Alternatively, the UAS-lbe-16-1 line, which exhibits a high level of ectopic expression and an effector line 24B-Gal4 (Brand and Perrimon, 1993), were used to express lb ectopically in all mesodermal cells. Temperature shifts on hh and wg thermosensitive mutant embryos were performed at 4 or at 6 hours AEL as described previously (Jagla et al., 1997b).

The following primary antibodies were used: monoclonal anti-Lbe (1:1), rabbit anti-Lbl (1:5000), monoclonal anti-En (1:1), rabbit anti-

β-galactosidase (1:8000), rabbit anti-dMEF2 provided by H. Nguyen (1:2000), rabbit anti-MHC provided by D. Kiehart (1:2000), rabbit anti-β3 tubulin, provided by A. Paululat (1:200); rabbit anti-Kr (1:2000) provided by P. Carrera and G. Vorbrüggen, rabbit anti-Twi provided by F. Perrin-Schmitt (1:1000) and rabbit anti-Human Phospho-Histone H3 (Upstate Biotechnology) (1:200). Detections of antibody stainings were performed using ABC-Elite-peroxidase or ABC-alkaline phosphatase kits (Vector Laboratories). In the case of fluorescent stainings, we have used secondary antibodies conjugated with Cy3, Cy5, Cy2 (Jackson Immuno-Research) or Oregon Green Protein (Molecular Probes). Colour reactions were developed using diaminobenzidine (for peroxidase) or NBT (for alkaline phosphatase) as substrates. Whole-mount embryos were photographed on the Axiophot microscope under Nomarski optics or scanned on the Leica confocal microscope.

ladybird early (lbe) and ladybird late (lbl) genes are located in the 93E homeobox gene cluster together with tinman (tin), bagpipe (bap) (Azpiazu and Frasch, 1993; Bodmer, 1993) and S59 (Dohrmann et al., 1990) genes known to be involved in the mesoderm differentiation programme (Jagla et al., 1997a). We have previously shown that lbe and lbl are co-expressed in segmentally repeated populations of cardioblasts and pericardial cells and play an important role in the diversification of heart precursors (Jagla et al., 1997b). The analysis of the lbe and lbl expression patterns also showed that both genes are expressed in the same group of somatic mesodermal cells that give rise to a syncytial muscle fiber (Jagla et al., 1997a,b).

To determine the identity of lb-positive muscle, we used confocal microscopy of late-stage embryos stained with anti-Mef2 antibody, which labels all myoblast nuclei (Bour et al., 1995) or with anti-Myosin (Myo) (Leiss et al., 1988), which labels the muscle fiber architecture (Fig. 1). Double staining revealed that, in the stage 16 embryo, lb is co-expressed with mef-2 in the 6 to 7 myoblast nuclei per hemisegment (Fig. 1C). Spatial location of these nuclei coincides with the segmental borders indicating that they belong to the SBM. This observation is fully confirmed by lbe/myo double staining (Fig. 1D) clearly showing that lb is expressed in only one muscle fiber, the SBM.

To examine the spatial relationship between the lb-expressing progenitor/founders and other myoblast lineages, we have used three markers: twist (twi) (Thisse et al., 1991), Kr (Ruiz Gomez et al., 1997) and an enhancer trap line rP298 (Nose et al., 1998) (Fig. 2). twi labels segregating muscle progenitors (Fig. 2A) but is no longer expressed in muscle

founder cells (Fig. 2F; see also Carmena et al., 1995). The only myoblasts with persistent twi expression are the non-differentiated precursors of adult muscles (Bate et al., 1991; see also Fig. 2B). The SBM progenitor, which co-expresses twi and lb, in comparison to other progenitors, shows some particularities. Unlike S59 and Kr progenitors (Carmena et al., 1995), it divides giving three progeny (Fig. 2A,B,C,G,F): a twi-negative SBM founder cell (arrow in Fig. 2F,C,G) which recruits neighbouring myoblasts to built the syncytial SBM fiber and two adult muscle precursors with persistent twi expression (arrowheads in Fig. 2B,C,G). The position of the latter cells, close to the SBM, indicates that they correspond to lateral adult muscle precursors (LaPs) (Bate et al., 1991). The distinct fates of lb-positive SBM and LaP myoblasts are already apparent during late stage 12 (Fig. 2D,H). In embryos from the enhancer trap line rP298, known to express lacZ in the progenitors and founders of all larval muscle fibers (Nose et al., 1998), the SBM lineage co-expresses lb and lacZ while the LaPs do not (Fig. 2H). In contrast, neither of the lb-positive myoblasts express Kr, which labels neighbouring lateral and ventral muscle precursors (Fig. 2E,I).

Taking advantage of the restricted expression pattern of lb genes, we followed the somatic mesodermal cells that give rise to SBM. We found that lb activity is associated with all stages of SBM formation, namely the promuscular cluster (Fig. 3A,F), progenitor cell (Fig. 3B,G), founder cell (Fig. 3C,H), fusing myoblasts (Fig. 3D,I) and syncytial fiber (Fig. 3E,J). The SBM arises from a cluster of 6-7 mesodermal cells, each of which weakly expresses lb (Fig. 3A,F). During early extended germ band stage (about 5 hours AEL), lb expression becomes restricted to, and upregulated in, only one large cell, the SBM progenitor (Fig. 3B,G). This cell, as detected by double staining with a marker of mitosis, undergoes two divisions (Fig. 3G,H). per segment, the SBM (arrows).The first division (right panel of Fig. 3G) gives rise to the SBM founder and is morphologically asymmetric (left panel of Fig. 3H; compare also with Fig. 2F) while the second, most likely symmetric, results in two LaPs (right panel of Fig. 3H; see also Fig. 2B,C,G). At the end of germ band retraction (9 hours AEL), the SBM founder cell starts to migrate dorsally along the segmental border whereas the LaPs remain at their initial position (Fig. 3D,I). The migration of the SBM founder prefigures the final location of SBM syncytial fiber formed by the progressive integration of neighbouring myoblasts (Fig. 3I). At the onset of dorsal closure fusion is completed and the SBM contains 6-7 lb-positive nuclei (Fig. 3E,J).

The SBM progenitor corresponds to the most superficial cell from the promuscular cluster, thus suggesting a role for the overlying ectoderm during its segregation. To investigate this possibility we first determined the position of the SBM promuscular cluster with respect to the epidermal anterior and posterior compartments. As shown on Fig. 4A,F, this cluster is located ventrolaterally below the epidermal posterior compartment. After segregation, the SBM progenitor migrates to a more lateral and posterior position so that, by late stage 11 (7 hours AEL), it is detected at the segmental border (Fig. 4G,L). Since epidermal Wg and Hedgehog (Hh) signaling was shown to influence muscle formation (Baylies et al., 1995; Azpiazu et al., 1996), we analyzed the SBM-associated lb expression in embryos carrying hh and wg thermosensitive mutations. Temperature shifts administered during early extended germ band stage (at 4 hours AEL) revealed that the Wg and Hh signalings, mutually dependent at this time (Perrimon, 1994), are required for the promuscular lb activity and/or the segregation of SBM progenitors (Fig. 4B,C,F). The initial influence of these signals is no longer observed later in development (Fig. 4H,I).

In addition to signals from the epidermis, the activity of the mesodermal gene, tin (Azpiazu and Frasch, 1993; Bodmer, 1993), initially expressed in the whole trunk mesoderm, is involved in the early events of myogenesis. In tin⁻ embryos, the formation of SBM promuscular clusters and segregation of lb-positive progenitor cells are strongly affected (compare Fig. 4D,F) leading to the absence of the majority of SBM fibers (compare Figs 3E and 4J). Since, during promuscular cluster formation (about 5 h AEL), tin expression becomes restricted to the dorsal mesoderm (Azpiazu and Frasch, 1993; Bodmer, 1993), its influence on ventrolaterally located SBMs is likely to be indirect and mediated via an unknown factor (see Fig. 7). The lack of neurogenic gene function, known to be involved in cell-cell interactions during lateral inhibition (Bate et al., 1993) generates an opposite phenotype. Mastermind (Mam⁻) (data not shown) and Enhancer of split (E(spl)⁻) embryos fail to restrict promuscular lb expression to only one cell (Fig. 4E) and, in consequence, display a hyperplastic lb pattern in later stages (Fig. 4K). In strong neurogenic mutants such as Notch and Delta, the SBM progenitors do not segregate at all or do not divide properly (data not shown). In contrast, the loss of function of a proneural gene, lethal of scute (l’sc), which is specifically expressed in promuscular clusters and segregating muscle progenitors (Carmena et al., 1995), has no significant influence on SBM formation (data not shown). Altogether our analysis of how the segregation of SBM progenitor is regulated indicates that cell commitment in the somatic mesoderm is conferred by a coordinate input of mesodermal and epidermal cues.

To investigate the role of lb activity in the specification of SBM and LaP myoblast lineages, we examined the pattern of larval and adult muscle precursors in embryos ectopically expressing lb and in embryos lacking lb activity (see Materials and Methods). The comparison of SBM formation in the wild-type (Fig. 5A), the hs-lb (Fig. 5B) and UAS-lb embryos (Fig. 5H) revealed that, in about 70% of hemisegments, the ectopic lb expression leads to the formation of enlarged (Fig. 5B,H) or duplicated (Fig. 5H) SBMs. Similarly, in 24B-Gal4/UAS-lbe embryos, the number of LaPs with persistent twi expression is significantly increased (Table 1; compare Fig. 5J and K. The overproduction of SBM (Fig. 5B,H) and LaPs (Fig. 5K) is frequently accompanied by the loss of some neighbouring lateral muscle fibers (compare Fig. 5G,H) suggesting that the ectopic expression of lb may change the identity of a subset of early myoblasts (progenitors/founder cells) and modify the muscle pattern. Indeed, in hs-lb (data not shown) and 24B-Gal4/UAS-lbe embryos, the number of Kr-expressing muscle precursors just adjacent to the SBM (see Fig. 2I) is dramatically reduced (compare Fig. 5D and E) indicating lb-induced transformation of myoblast identities.

The lb⁻ embryos display opposite muscle phenotypes. In 57% of hemisegments, we note the absence of muscle fibers at the SBM positions (Fig. 5C) and instead, within segmental borders or around the normal SBM location, we detect some unfused myoblasts (Fig. 5C,I). In the remaining 43% of hemisegments, the muscle fibers lying within or close to the segmental borders display abnormal shapes and insertion sites clearly distinct from those of SBM (Fig. 5I). Similarly, the specification of LaPs is affected in lb⁻ embryos. In 55% of hemisegments, they are absent (Fig. 5L) and, in 38% of hemisegments, we observe only one LaP (not shown) instead of two (Fig. 5J). Surprisingly, alteration of lb activity also affects lb-negative precursors of ventral adult muscles (VaP) (see Table 1), which are duplicated in lb-deficient embryos (Fig. 5L) and absent when lb is ectopically expressed (Fig. 5K). Thus, we speculate that this feature is linked with the lb expression in a mesodermal cell (?) lying close to the VaP (Fig. 5J) and reflect the functions of lb in cell fate specification. Altogether these findings imply that lb is required for the acquisition of specific muscle properties and that the presence or absence of Lb proteins is sufficient to switch a subset of muscle founders between alternative fates.

segmentally repeated pattern of 30 different body wall muscles. Compared with vertebrate myogenesis, it appears much less complex, nevertheless an obvious similarity can be observed: (i) in both invertebrate and vertebrate embryos, somatic muscle primordia are specified as a population of multipotent mesodermal cells, respectively, within segmental units or somites, (ii) these cells become committed to a muscle fate under the influence of evolutionarily conserved intrinsic MyoD-like and MADS-box regulators and extrinsic inductive Wg/Wnts signals (for review see Yun and Wold, 1996) and (iii) in both systems, the Notch/Delta cell-cell signaling pathway mediates lateral inhibition involved in the segregation of muscle progenitors (Corbin et al., 1991; Bate et al., 1993; Lindsell et al., 1995). Extensive conservation of these early steps of the myogenic cascade makes Drosophila a model of choice to address one of the most important and as yet unanswered questions: how distinct characteristics of otherwise similar muscles are determined? Here we investigate the genetic pathways that lead to the diversification of Drosophila larval muscles showing that the identity of one of them is specified by the homeobox genes lbe and lbl.

During myogenesis, the lb genes display a highly restricted expression pattern suggesting that they require a precise regulatory apparatus. Indeed, both mesodermal information and inductive signals from the epidermis are involved in the proper regulation of lb activity in myoblast lineages leading to the formation of SBM and LaPs. Initially, tin function is required for activation of promuscular lb expression. However, this tin influence is likely to be indirect and mediated via an unknown factor, since tin expression at this time (5 hours AEL) becomes restricted to the dorsal mesoderm (Azpiazu and Frasch, 1993; Bodmer, 1993) while the lb-positive promuscular cells are located ventrolaterally. The indirect activation of lb by tin seems confirmed by the presence of about 20% of lb-expressing SBM fibers in tin⁻ embryos (see Fig. 4I). Interestingly, tin-dependent lb activation does not occur, in embryos lacking early wg or hh functions, providing further evidence for the role of epidermal signalings in the early events of myogenesis (see also Baker and Schubiger, 1995; Baylies et al., 1995, Ranganayakulu et al., 1996). Our analysis of mutants of neurogenic genes also confirm that the cell-cell interactions that lead to lateral inhibition are indispensable for the segregation of a proper number of SBM muscle progenitors (see also Corbin et al., 1991; Bate et al., 1995).

In contrast to S59 (Dohrmann et al., 1990; Carmena et al., 1995), Kr (Ruiz Gomez et al., 1997), msh (Nose et al., 1998) and other muscle identity genes, lb expression is restricted to only one muscle progenitor and its progeny. Recently, Ruiz Gomez and Bate (1997) and Carmena et al. (1998) have shown that the Nb protein is asymmetrically located in the progenitors and during their division transmitted to only one of the two daughter cells. As indicated by the loss of Kr-, S59- and eve-expressing muscles and the overproduction of adult precursors in nb mutant embryos (Ruiz Gomez and Bate, 1997), nb may determine the different fates of the sibling myoblasts.

After division of the progenitors expressing Kr, S59 and eve, the activity of these genes disappears from founders devoid of Nb before they start to fuse with neighbouring myoblasts and is maintained in muscle precursors that arise from Nb-containing founders (Ruiz Gomez and Bate, 1997). The lb-positive progenitor displays distinct features. After its division, lb expression is maintained in both the SBM and LaP lineages although, as indicated by the nb⁻ phenotype (lack of SBM; see Fig. 6B), the Nb protein is rather transmitted to the SBM founder than to the LaPs. These data indicate that lb expression may play an important role in the determination of LaP identity and is maintained in these myoblasts in a nb-independent manner.

The observation that the lb progenitor, unlike S59 and Kr progenitors (Carmena et al., 1995; Ruiz Gomez et al., 1997), produces three progeny (one SBM founder and two LaPs) instead of two, suggests a distinct pathway for the division. Indeed, as indicated by confocal microscopy (see Fig. 3G,H), the lb-positive progenitor divides two times. Surprisingly, its first division is morphologically asymmetric giving rise to a smaller SBM founder cell (see Figs 3H, 6). This finding indicates that the muscle progenitors can produce two daughter cells of the same size, like neural midline precursors (Carmena et al., 1995), and also divide like neuroblasts to generate smaller progeny. The second division of the lb-positive progenitor produces two LaPs, which do not contribute to the larval muscles and, most likely due to the persistent twi expression, remain nondifferentiated throughout embryogenesis. As was previously shown by Bate (1991), the LaPs proliferate during larval life and give rise to the majority of abdominal adult lateral muscles. Unexpectedly, the second myoblast lineage specified by lb expression, the SBM myoblasts, also contributes to the adult body wall muscles. During pupation, unlike other larval somatic muscles, the SBM fibers resist on histolysis and serve as a template for the formation of adult muscles (Bate et al., 1991). The role of lb in this ‘long life’ phenomenon remains to be elucidated.

Interestingly, lb genes are also expressed in a group of promuscular cells from which the lb-positive progenitor segregates. This early lb activity overlaps that of l’sc expressed in all promuscular clusters (Carmena et al., 1995) suggesting that lb may exert its own promuscular function and/or cooperate with l’sc in the specification of the SBM promuscular cluster. The unaffected development of SBM in l’sc⁻ embryos (data not shown) seems to confirm this possibility.

According to our observations, cell commitment in the somatic mesoderm involves three steps: (i) formation of promuscular clusters; (ii) segregation of progenitors and (iii) specification of founder cells or adult muscle precursors. Since, lb expression is associated with all these steps of SBM and LaPs commitment, it strongly suggests that lb genes may determine the identity of these two myoblast lineages. Indeed, in the embryos lacking lb function, the SBM and LaP formation is disturbed and we detect near segmental borders unfused or supernumerary Kr-expressing myoblasts. Thus, we conclude that, in the absence of lb activity, the prospective SBM myoblasts behave as noncommitted cells unable to form syncytial fibers or are recruited to built enlarged neighbouring muscles. Opposed muscle transformations occur in the embryos with ectopic lb expression. In the majority of segments, Kr-expressing progenitor/founder cells from the lateral domain are redefined, become lb-positive and contribute to the formation of giant SBMs or supernumerary LaPs. These data clearly show that lb functions to specify the cell fate of myoblasts in which it is expressed. Thus, we infer that lb is sufficient to determine the identity of the SBM founder and required, together with twi, to define distinct features of LaPs. Taking into consideration the requirements for the specification of lb-expressing myoblast lineages, we propose a model of cell commitment where the sequential activation or repression of defined, or as yet unknown, genes allows a pluripotent mesodermal cell to become the SBM founder or a LaP (Fig. 7). During gastrulation, the mesodermal cells of the trunk, from which body wall muscles arise, in addition to twi, express tin. The tin activity becomes restricted to the dorsal mesoderm, thus suggesting that another gene specifies the fate of ventrally located cells. The dorsal mesoderm, under the influence of epidermal signals, subdivides into the cardiac, visceral and fat body primordia while the ventral mesoderm mainly gives rise to the somatic musculature. Most likely, due to the action of l’sc (Carmena et al., 1995), a small portion of dorsal and the majority of ventral mesodermal cells become committed to promuscular clusters from which the progenitors are selected. The segregation of muscle progenitors requires a coordinate action of neurogenic genes as well as epidermal Wg and Hh signalings. The progenitors and, subsequently, their progeny are committed to different myoblast lineages by a network of identity genes such as S59, Kr, msh and lb. In the case of the lb-expressing progenitor, two myoblast lineages are produced: the twi-negative SBM founder and two LaPs with persistent twi activity. Similar to the co-expression of lb and twi in LaPs, several other myoblast lineages have been shown to express more than one identity gene. For example, in the DA1 founder, Kr is co-expressed with eve and in VA2 with S59 (Ruiz Gomez et al., 1997), thus suggesting that the combined activation of identity genes may play an important role in the acquisition of individual muscle properties.

In this work, we describe the myogenic functions of lb homeobox genes showing that altered lb expression leads to the transformation of myoblast identities. Since lb provides positional information to a subpopulation of cardiac cells (Jagla et al., 1997b), we conclude that lb exerts similar functions during muscle and heart formation. The role of lb genes in the determination of identity is reminiscent of that of homeotic genes but exerted at the ‘micro’ levels where a cell lineage not a body part is concerned.

The finding that the Drosophila lb determines cell fate decisions during diversification of somatic muscles suggests a similar role for the vertebrate lb homologs (Jagla et al., 1995, Dietrich et al., 1998) thus allowing the investigation of diversification of vertebrate skeletal muscles.

We are grateful to P. Chambon for the support. We thank P. Blader, M. Frasch, M. Bourouis and J. Bosher for valuable comments on the manuscript. We are indebted to P. Heitzler for discussions and M.-L. Nullans for technical assistance. This work was supported by the Institut National de la Santé et de la Recherche Médicale, the Centre National de la Recherche Scientifique, the Hôpital Universitaire de Strasbourg, the Association pour la Recherche sur le Cancer and the grant from Human Frontier Science Program.