Regional specification of muscle progenitors in Drosophila: the role of the msh homeobox gene

The Drosophila embryo provides a simple model system for genetically dissecting the molecular mechanisms that control muscle patterning (reviewed by Bate, 1993; Abmayr et al., 1995). Each abdominal hemisegment contains 30 muscle fibers with unique characteristics in terms of position, shape, epidermal attachment, and pattern of innervation. It has been proposed that the muscle diversity is specified in a special class of myoblasts, called muscle founders (Bate, 1990, 1993). These cells are found at stereotypic positions in the somatic mesoderm prior to the onset of fusion and initiate the formation of syncytial precursors of individual muscles by fusing with neighboring myoblasts. The best evidence in support of the muscle founder hypothesis comes from the studies on the expression of S59, a homeobox-containing protein (Dohrmann et al., 1990; Rushton et al., 1995; Carmena et al., 1995). S59 is initially expressed in a small number of myoblasts prior to the onset of fusion. The S59-expressing cells fuse with each other and with neighboring myoblasts to form the precursors of specific muscles. During this process, the neighboring myoblasts also initiate expression of S59. Furthermore, such recruitment to S59 expression is not observed in myoblast city (mbc) mutants which do not undergo muscle fusion (Rushton et al., 1995). These observations led to the suggestion that the information necessary for muscle patterning is first conferred to a small number of founder cells, and is then transferred to neighboring unspecified myoblasts as they fuse with the founders. In addition to S59, several putative transcription factors have been identified that are expressed in subsets of somatic mesoderm, further supporting the notion that a special class of myoblasts are specified very early during development (reviewed by Bate, 1993; Abmayr et al., 1995).

How are the founders segregated during mesodermal development? A recent study (Carmena et al., 1995) showed that expression of the proneural gene, lethal of scute (l’sc), prefigures the formation of the muscle founders. l’sc is expressed in clusters of cells in the somatic mesoderm, from which cells with higher levels of l’sc expression are singled out by a process involving neurogenic gene function. The cells with higher levels of l’sc expression appear to be the progenitors of muscles, since a subset of them divide to give rise to S59-expressing founders. Based on these observations, Carmena et al. (1995) summarized the early phases of myogenesis as follows. (1) l’sc expression in clusters of mesodermal cells with potential to form muscle progenitors. (2) Singling out of the muscle progenitors by competitive interaction. (3) Production of the founders of individual muscles by the division of the progenitors.

Although these studies have provided a framework for considering early aspects of muscle patterning, several important questions remain unanswered. (1) Where in the mesodermal anlagen, are the progenitors or founders of the 30 individual muscles initially formed? Carmena et al. (1995) mapped the location of the 19 l’sc expressing clusters in the mesoderm, which is likely to correspond to the origin of most, if not all, of the muscle progenitors. However, with the exception of the S59-expressing cells, the ultimate fate of these putative progenitors has not been determined, due to the transient nature of l’sc expression. Thus, for a majority of the muscles, the position and identity of their progenitors/founders are currently unknown. (2) How are the individual founders located and determined? What molecular mechanisms underlie the information that confers specific characteristics on each muscle founder? As noted above, several putative transcription factors are known to be expressed in subsets of cells in the somatic mesoderm, and thus are implicated in the specification of muscles. These include the proteins encoded by the S59, apterous (ap), vestigial (vg) and even-skipped (eve) genes (reviewed by Bate, 1993; Abmayr et al., 1995). However, the functions of these genes in muscle development are largely unknown, since the mutant phenotype of these genes has not been reported with the exception of ap. The absence or overexpression of ap leads to the loss or duplication of ap-expressing muscles (Bourgouin et al., 1992). The expressivity of these phenotypes, however, is rather low, suggesting that ap plays a minor or redundant role in muscle patterning.

In this study, we first used a novel marker, rP298-LacZ, to map the location of muscle progenitors/founders and follow their development to give rise to the 30 muscle fibers. We show that the 30 somatic muscles originate from four domains of progenitor/founders (called dorsal, dorsolateral, lateral and ventral domains) located at specific positions in a hemisegment. We then describe the genetic analysis of the function of the msh gene in muscle patterning. msh encodes a homeobox-containing protein, vertebrate homologues of which are known as Msxs (Robert et al., 1989; Lord et al., 1995; D’Alessio and Frasch, 1996; Isshiki et al., 1997; reviewed by Davidson, 1995). In a previous study (Isshiki et al., 1997), we showed that msh is regionally expressed in a subset of neural progenitors (neuroblasts), and is required for their correct development to form specific neurons and glia. We show in this study that msh is also expressed in subsets of muscle progenitors that form in the dorsal and lateral domains, and is required for the realization of their correct myogenic pathways leading to the formation of specific muscles. Furthermore, ectopic expression of msh in the entire mesoderm inhibits the proper development of the msh-negative progenitors/founders that normally give rise to the dorsolateral muscles. Severe and highly selective phenotypes seen in the loss-of-function and gain-of-function mutants suggest that msh plays a pivotal role in specifying the regional identities of muscle progenitors/founders.

rP298 is a P[lacZ, ry⁺] insertion on the X chromosome and was isolated in an enhancer trap screening conducted in Corey S. Goodman’s laboratory (Klämbt et al., 1991; Nose et al., 1992). In addition to the somatic muscle expression described in this study, rP298-LacZ also marks the visceral mesoderm and some mesodermal glia from stage 11, but not the cardiac mesoderm. The msh-lacZ line, rH96, and the Connectin-lacZ line, rF400, were described by Isshiki et al. (1997) and Nose et al. (1992), respectively. Alleles of msh used were, msh⁶⁸ and msh^lacZ-Δ89 (Isshiki et al., 1997). They were produced by imprecise excision of a P-element insertion and contain deletions in the 5′ end of the gene. In msh^lacZ-Δ89, the lacZ gene in the P-element was left intact and is expressed in the msh pattern. Connectin-lacZ was meiotically recombined with the chromosome carrying msh⁶⁸ to yield Connectin-lacZ, msh⁶⁸.

The GAL4-UAS system (Brand and Perrimon, 1993) was used to express msh in the mesoderm. Isolation of the UAS-msh reporter lines was described previously (Isshiki et al., 1997). Two of the five UAS-msh lines obtained, UAS-msh-m25-m1 and UAS-msh-m29-m1, which exhibit a high and moderate level of ectopic expression, respectively, were used. Both contain the transgene on the third chromosomes. Ectopic expression of msh in the entire mesoderm was achieved by crossing the UAS-msh reporter lines with the 24B-GAL4 effector line (Brand and Perrimon, 1993). Similar muscle phenotypes were observed when either of the two UAS-msh reporter lines was used. However, the severity of the phenotype was more pronounced when UAS-msh-m25-m1 was used. rP298-LacZ expression in the misexpressor was studied by crossing rP298-lacZ; msh-m25-m1 females with 24B-GAl4 males. For the rescue experiments, 24B-GAL4 and UAS-msh-m29-m1 were each meiotically recombined with the msh⁶⁸ chromosome to yield 24B-GAL4, msh⁶⁸/TM2 and UAS-msh-m29-m1, msh⁶⁸/TM2 lines. Rescue of the lateral muscle phenotypes was examined in the embryos obtained by crossing these two lines.

RNA in situ hybridization, antibody staining and dissection of embryos were carried out as previously described (Lehmann and Tautz, 1994; Patel, 1994). The following primary antibodies were used: rabbit sAb against β-galactosidase (β-gal, Capell), mouse mAb against Engrailed (EN; Patel, 1994), and rabbit sAb against muscle Myosin (Kiehart and Feghali, 1986).

We utilized the reporter gene expression in an enhancer trap line, rP298-LacZ, to follow the development of the somatic mesoderm. rP298-LacZ expression is first detected in a small number of large cells that arise sequentially during stage 11/12 (Fig. 1A,B). These cells then divide to give rise to smaller cells which occupy specific positions in the somatic mesoderm by mid-stage 12 (Fig. 1C). Shortly after, these cells fuse with the neighboring myoblasts to form the muscle precursors (Fig. 1D,E). During this process, additional cells begin to express rP298-LacZ upon fusion with the cells that had earlier expressed the reporter gene. This results in a rapid increase in the number of the rP298-positive nuclei towards the end of the germ band shortening. At stage 16, when the mature pattern of muscle fibers is achieved, rP298-LacZ is detected in the nuclei of the 30 differentiated muscle fibers (Fig. 1F). These and other features of rP298-LacZ expression (see below and Discussion) suggest that it marks the progenitors and founders of all muscle fibers during early development of the mesoderm. The early onset of rP298-LacZ expression and its persistence to later stages when individual muscle fibers can be uniquely identified, allowed us to trace the origin of the majority of muscle fibers in the abdominal segments A2-A7. Based on such analysis and the data previously described by others (Bate, 1993; Dunin Borkowski et al., 1995; Carmena et al., 1995), we subdivide the progenitors/founders into four domains based on their origin (summarized in Fig. 1G-J).

Dorsal (D) domain: This domain is located at the dorsal-most region of the somatic mesoderm and just anterior to the ectodermal engrailed (en) expressing stripe and includes rP298-LacZ-positive putative progenitors/founders that give rise to the four dorsal muscles (1, 2, 9 and 10; nomenclature of muscles according to Bate, 1990; see Fig. 1J). As will be described below, some of the rP298-positive cells express msh and contribute to muscles 9 and 10 that occupy the external layer of the dorsal musculature.

Dorsolateral (DL) domain: This domain includes the progenitors/founders of the six dorsolateral muscles (3, 4, 11 and 18-20) and is located between the tracheal placode and the en stripe. The progenitors in this domain are among the first to exhibit rP298-LacZ expression (see Fig. 1A).

Lateral (L) domain: The progenitors in this group originate in the lateral region of the somatic mesoderm near the boundary of the neuroectoderm. One of them divides to form four smaller cells that migrate ventrally and contribute to ventral external muscles 26, 27 and 29. The characteristic patterns of cell division and migration suggest that this progenitor is identical to the l’sc- and S59-expressing progenitor l’sc10/S59II (Dohrmann et al., 1990; Carmena et al., 1995). The other two progenitors in this domain divide to produce four founders that will later form the lateral muscles 21-24.

Ventral (V) domain: The progenitors/founders in this group are found interior to the developing central nervous system (CNS), and give rise to the ventral internal muscles (6, 7, 12-17 and 30; we use internal to include both the internal and intermediate muscles described by Bate, 1990). Progenitors of three other muscles (5, 8, 25) also arise in this domain but follow distinct fates. One of them is likely to be identical to the S59-expressing l’sc3/S59I progenitor (Dohrmann et al., 1990; Carmena et al., 1995) which migrates in a characteristic manner and produces two founders that contribute to the lateral internal muscle 5 and ventral muscle 25. Another progenitor appears to migrate dorsally and forms the segment border muscle 8.

It should be noted that a majority of the muscles differentiate near the position where their progenitors are initially formed. Thus, in general, their final location corresponds to the domain of their origin (e.g. dorsal muscles are derived from the D domain). However, several muscles migrate during their formation, and their final location differs from that of their origin (e.g. ventral external muscles 26, 27 and 29 originate from the L domain).

It has previously been reported that msh is expressed in specific regions of ectodermal and mesodermal cells during stage 10/11 (Lord et al., 1995; D’Alessio and Frasch, 1996). In this study, we describe more detailed analyses of msh expression in correlation with the development of the somatic mesoderm. Beginning at stage 10, msh is expressed in two patches of epidermal cells in the body wall (Fig. 2A, see also Fig. 7). The more dorsal patch is located at a position dorsal and posterior to the developing trachea, whereas the other patch lies in the lateral region at the boundary of the neuroectoderm. Following the ectodermal expression, some mesodermal cells underlying these epidermal patches also begin to express msh (Fig. 2B). In the dorsal region, a large mesodermal cell expresses msh (Fig. 2C). At the beginning of the germ band shortening, it divides to produce two smaller msh-positive cells, the posterior one of which appears to further divide shortly thereafter (Fig. 2E,F). As will be described below, two of the resultant three msh-positive cells are the founders for muscles 9 and 10. Double staining with rP298-LacZ showed that these cells also express rP298-LacZ, supporting the notion that rP298-LacZ marks the muscle progenitors/founders (Fig. 2E,F). When fusion of the myoblasts is initiated towards the end of germ band shortening, MSH expression is seen in the precursors of muscles 9 and 10 (Fig. 2G,H).

Under the lateral ectodermal patch, several msh-expressing cells are seen, including some dorsal neuroblasts (e.g., NBs 6-4 and 7-4, and the longitudinal glioblast; Isshiki et al., 1997) and fat body precursors (D’Alession and Frasch, 1996). In addition, we noted expression in approx. 3 cells in the somatic mesoderm that are found anterior to the en stripe and just lateral to the edge of the developing CNS (Fig. 2D). Unlike the dorsal muscle progenitor, however, expression in these mesodermal cells is transient and disappears by late stage 11 when individual progenitors can be identified by rP298-LacZ expression. It was thus impossible to assign the ultimate fate of these cells. However, from their location and the mesodermal requirement of msh described below, these cells are likely to be the progenitors in the L domain that will later form the lateral and ventral external muscles.

The persistent expression of msh in the progenitor of dorsal muscles 9 and 10 serves as a marker to follow the development of these muscles. To analyze in more detail the events occurring during the formation of these muscles, we studied the reporter gene expression in the msh enhancer trap line (msh-LacZ; Fig. 3A-F, summarized in Fig. 3J). The nuclear localization and high sensitivity of the LacZ reporter makes it possible to monitor msh expression in individual myoblasts as they fuse to form the muscle syncytium. During stage 11/12, msh-LacZ is expressed in a similar pattern to msh RNA, initially in the single progenitor and then in its three progeny (Fig. 3A,B). Two of them fuse with the surrounding myoblasts to form the precursors of muscles 9 and 10, and thus are the founders for these muscles (Fig. 3C-E). During this process, neighboring msh-negative myoblasts are recruited to msh-LacZ expression (white arrowheads in Fig. 3C). This results in the rapid increase in the number of msh-LacZ-expressing nuclei towards the end of germ band shortening. The msh-LacZ-positive nuclei arrange themselves in a characteristic pattern and finally align along each of the ends of muscles 9 and 10 by stage 14 (Fig. 3F).

The muscle phenotype of msh loss-of-function mutants, which were isolated by imprecise excision of a P-element from the gene (Isshiki et al., 1997), was determined by staining for muscle Myosin. In accordance with the normal expression pattern of msh, we observed severe defects in the formation of specific muscles (two dorsal muscles, 9 and 10, derived from the D domain; four lateral muscles, 21-24, and three ventral external muscles, 26, 27 and 29, derived from the L domain), but not in others (e.g. dorsolateral muscles and ventral internal muscles; Fig. 4).

The dorsal-most region of the somatic musculature is composed of four muscles. Muscles 1 and 2 form the internal layer, whereas muscles 9 and 10 lie in the external layer closer to the epidermis (Fig. 4A). In msh embryos, instead of the two external muscles, only one muscle fiber is present that is located at a position halfway between those normally occupied by the two muscles (Fig. 4B,C). This phenotype is highly penetrant and was observed in all hemisegments examined (n=41). The other two dorsal muscles, 1 and 2, also display somewhat disorganized morphology, such as the presence of protrusions extending towards the region normally occupied by muscles 9 and 10 (arrow in Fig. 4C), possibly as secondary consequences of the defects in the neighboring muscles 9 and 10.

Lateral external muscles 21-24 run in parallel in the lateral region of the body wall (Fig. 4A). In most of the mutant hemisegments, one or more of the four lateral muscles are missing (Fig. 4B, D; summarized in Table 1). On average, the mutant hemisegments contain only 1.74 instead of four muscle fibers in this region. Furthermore, many of the remaining muscles are abnormally shaped, often extending beyond their normal territories (black thin arrow in Fig. 4D). Taken together, 72% of the lateral muscles are either missing or abnormally formed. Similar but less frequent abnormalities are also seen for the ventral external muscles 26 and 27 (21.9% missing and 22.6% misshaped; total of 44.5% abnormalities; Fig. 4D).

Given the defects observed for subsets of differentiated muscles, we wanted to determine when in muscle development msh is required. To this end, we investigated the expression of early markers in msh embryos. As described above, msh-LacZ can be used to visualize the early development of muscles 9 and 10. In mutants with an allele of msh, msh^lacZ-Δ89, the lacZ gene in the P-element is left intact thus allowing the monitoring of msh-LacZ expression in the mutant background (Isshiki et al., 1997). In msh^lacZ-Δ89 embryos, msh-LacZ is expressed initially in the dorsal muscle progenitor and then in its three progeny as in the case of wild-type embryos, suggesting that the formation of the progenitor/founders occurred normally (Fig. 3G). However, their subsequent development was found to be markedly altered (Fig. 3H,I; summarized in Fig. 3J). As described above, the two msh-positive founder cells normally contribute to two precursors of dorsal muscles, 9 and 10 (see Fig. 3D). In the mutant embyros, however, the two founders form only one syncytium (Fig. 3H). Furthermore, many fewer neighboring myoblasts are recruited to express msh-LacZ. The arrangement of the recruited nuclei in the precursor or mature muscle is also abnormal (Fig. 3I).

The lacZ reporter in the Connectin enhancer trap line (called Connectin-LacZ; Nose et al., 1992) was utilized to investigate the early development of lateral and ventral external muscles, 21-24, 27 and 29 (Fig. 5A,C). In msh embryos, significant reduction in the number of Connectin-LacZ-positive nuclei is evident as early as stage 13 (Fig. 5B). Furthermore, the positioning of the remaining nuclei is abnormal and they are often seen clustered near the region where they initially formed (Fig. 5D). Thus, in all muscles whose formation is dependent on msh function, the defects can be traced to early myogenic pathways leading to the formation of muscle precursors.

To further analyze the role of msh in muscle specification, we studied the effect of its ectopic expression using the GAL4-UAS system. We used the 24B line, which promotes homogeneous GAL4 expression in the entire mesoderm from stage 10, to drive the expression of msh from two UAS-msh lines, m25-m1 and m29-m1 (see Materials and Methods for details). A higher level of ectopic msh expression is induced when m25-m1 is used as the effector line.

When either of the two effector lines is utilized, the muscle morphology is severely altered (Fig. 6A,B). Although the entire musculature is affected with considerable variation, we note a clear tendency for particular muscles to be more severely affected. Similar observations were made by Lord et al. (1995); however, the identities of muscles affected have not been determined. We find that the most severe defects occur in the four dorsolateral muscles 3, 11, 19 and 20. When msh is misexpressed by using the stronger effector, m25-m1, abnormalities in these muscles are seen in every hemisegment scored (n=124, Fig. 6A). In 44% of the cases, the four muscles are completely missing, leading to a dorsolateral gap in the musculature (thick arrows in Fig. 6A), whereas in the remaining 56%, the region is occupied by muscle(s) with abnormal shape and orientation, which cannot be assigned to any of the muscles in this region (thin arrows in Fig. 6A). Similar but less frequent alterations in these muscles are seen when the weaker effector line, m29-m1, is used (Fig. 6B). The other two dorsolateral muscles, 4 and 18, are also often missing or abnormally shaped, in particular when the stronger effector is used.

Less severe phenotypes are seen in other regions of the body wall muscles (Fig. 6A,B). The dorsal region contains the four muscles that are thicker than normal, and also occasionally contains additional muscle(s) with similar morphology to dorsal muscles at the boundary with the dorsolateral region (arrowhead in Fig. 6B), suggesting that some mesodermal cells that normally contribute to the dorsolateral muscles may be directed towards a dorsal fate. The ventral internal muscles are also disorganized when the stronger effector line is used; however they form largely normally when the weaker effector line is used. The four lateral muscles (21-24) form normally, although their morphologies are somewhat irregular.

To characterize the earlier events that lead to the severe defects in the dorsolateral muscles, we studied the expression of rP298-LacZ. In 24B-msh embryos, rP298-LacZ-positive progenitors of the dorsolateral muscles form normally in the DL domain (data not shown). However, their subsequent development to form founders and precursors fails to occur properly. This results in a clear reduction in the number of rP298-positive nuclei in the DL domain, which is evident by the completion of the germ band retraction (Fig. 6C,D). Thus, ectopic msh expression interferes with the correct specification of the progenitors/founders.

Since lateral muscles 21-24 largely form normally in 24B-UAS-msh embryos, we also used the GAL4-UAS system to determine if the mesodermal expression of msh can rescue the loss-of-function phenotypes in these muscles. As summarized in Table 1, expression of msh by 24B-UAS-msh-m29-m1 substantially rescues the lateral muscle phenotype in msh⁶⁸ embryos, thus indicating a requirement for msh in the mesoderm. As described above, msh is expressed transiently in the mesodermal cells in the L domain where the rP298-positive progenitors of the lateral muscles are later seen. msh is not expressed in other phases of the development of the lateral muscles. Thus, the results strongly suggest that the msh-expressing mesodermal cells are the progenitors of these muscles and msh is required in these cells for their correct development to form the lateral muscles.

We have used a novel marker, rP298-LacZ, to chart the origin and development of somatic muscles in the Drosophila embryo. rP298-LacZ is initially expressed in a small number of large mesodermal cells which divide to produce smaller cells prior to myoblast fusion. Several lines of evidence suggest that these cells are the progenitors and founders of the somatic muscles. Firstly, they form at stereotypic positions in the somatic mesoderm very early during development and appear to seed the formation of the muscle precursors by fusing with neighboring myoblasts. Secondly, the rP298-positive cells contain the progenitors/founders of the dorsal muscles 9 and 10 that specifically express msh. Furthermore, developmental profiles of some of the other rP298-expressing cells suggest that they are identical to the l’sc/S59-expressing progenitors/founders (Dohmann et al., 1990; Carmena et al., 1995).

Detailed analysis of rP298-LacZ expression at different stages allowed us to follow the development of progenitors that form in four domains in the body wall. Although the complexity of the expression pattern makes it difficult to determine the lineage of individual progenitors, we could monitor what groups of muscles are formed by the progenitors in each of the four domains. We showed that the D domain of the progenitors gives rise to the four dorsal muscles, the DL domain to dorsolateral muscles, the L domain to the lateral and ventral external muscles, and the V domain to ventral internal muscles. Although these analyses provide a large scale description of the regional specification of muscle progenitors/founders, we do not claim that the postulated subdivision of the progenitors/founders into four domains defines the precise limits of distinct populations of these cells. It also remains to be determined if the subdivision is of functional significance, namely, if it represents an actual organization that functions to specify the progenitors into distinct muscle fates. However, the results described in this study suggest that the expression and function of at least one gene, msh, is closely correlated with the subdivision of the muscle progenitors/founders.

msh is expressed in the D and L domains of muscle progenitors. Expression in the D domain is initially seen in a single progenitor, then in founders and precursors of dorsal muscles 9 and 10. As in the case of S59, msh expression can be seen to be initiated in neighboring myoblasts as they fuse with the msh-positive founders. In contrast, expression in the L domain is transient, and disappears by the time the founders are formed. In msh loss-of-function mutant embryos, severe and highly penetrant defects are seen in the formation of muscles derived from the msh-positive progenitors. Analysis of early markers such as msh-LacZ and Connectin-LacZ showed that early aspects of the development of these muscles are perturbed. In all cases examined, the progenitors and founders form normally. However, their subsequent development to form precursors of specific muscles is found to be abnormal. Furthermore, recruitment of the surrounding myoblasts to developing precursors also fails to occur properly. These results suggest that msh function is required for the realization of the myogenic program leading to the formation of particular muscle precursors.

A role of msh in muscle patterning was also suggested by the effects of its ectopic expression. The pan-mesodermal expression of msh severely disrupts the correct formation of somatic muscles, indicating that restricted expression of this gene is critical for muscle development. Although the entire musculature is affected, the most severe defects are seen in the formation of the dorsolateral muscles. Analysis of rP298-LacZ expression showed that the progenitors of these muscles in the DL domain initially form, but fail to give rise to the appropriate number of founders, suggesting that ectopic msh expression interferes with the normal development of the progenitor cells. It has been reported that the same sets of dorsolateral muscles affected by the ectopic expression of msh are also missing in embryos lacking rhomboid or pointed, members of the spitz group genes that are implicated in the Drosophila EGF-receptor (DER) signaling pathway (Bier et al., 1990; Klämbt, 1993; reviewed by Perrimon and Perkins, 1997). Furthermore, we observed that as in the case of ectopic msh expression, rP298-LacZ-positive muscle progenitors in the DL domain initially form but fail to produce appropriate founders and precursors in rhomboid mutant embryos (A. N., unpublished observations). These results may suggest that msh and the DER-signaling pathway function as part of two opposing genetic systems that specify the regional identities of muscle progenitors.

msh is also expressed in specific regions of the ectoderm, preceding mesodermal expression. Recent evidence suggests that inductive signals from the ectoderm play critical roles in the specification of mesodermal cells (Lewis and Crews, 1994; Staehling-Hampton et al., 1994; Frasch, 1995; Baylies et al., 1995; Lawrence et al., 1995; Wu et al., 1995; Ranganayakulu et al., 1996; Azpiazu et al., 1996). It thus is an interesting possibility that in addition to its role in the mesoderm, msh has an earlier function in the ectoderm to determine the precise locations of the muscle progenitor cells. Correlation with inductive epithelial-mesenchymal interactions in a variety of tissues (e.g. limb bud, facial structures, tooth bud) has also been suggested for the vertebrate homologues, Msx 1 and Msx 2, based on their expression and the results of grafting and tissue combination experiments (reviewed by Davidson, 1995).

The expression and function of msh during myogenesis and during neurogenesis show a number of similarities (see Fig. 7). During neurogenesis, msh is expressed in neural progenitors (neuroblasts) that form in the dorsal portion of the developing neuroectoderm (D’Alessio and Frasch, 1996; Isshiki et al., 1997). In msh embryos, msh-positive dorsal neuroblasts fail to produce their appropriate lineage (Isshiki et al., 1997). Similarly, during myogenesis, msh is expressed and required in the muscle progenitors that form in specific regions of the somatic mesoderm. In both cases, msh is not required for the initial formation of the progenitors, but for the subsequent events leading to the formation of particular neural or muscular cells. The expression of msh in the progenitors is preceded by its expression in the overlying ectoderm; the dorsal neuroectoderm during early neurogenesis and in two ectodermal patches during late neurogenesis and myogenesis. Furthermore, ectopic expression of msh interferes with the proper development of normally msh-negative neural or muscular progenitor cells (the ventral neuroectoderm and the DL domain of muscle progenitors; Lord et al., 1995; Isshiki et al., 1997; this study). Thus, msh appears to function in a similar manner during both neurogenesis and myogenesis to determine the positional identities of the progenitor cells.

We are indebted to C. S. Goodman in whose laboratory A. N. isolated the muscle enhancer trap lines described in this study. We also thank Y. Takahashi and colleagues in our lab for helpful discussions; D. P. Kiehart for Myosin antibodies; S. Hayashi and the Bloomington stock center for fly strains; and E. Shishido for comments on the manuscript. T. I. is a postdoctoral research fellow of the Japan Society for the Promotion of Science. This work was supported by research grants to A. N. and M. T. from the Ministry of Education, Science, and Culture of Japan.