twist expression in the embryonic mesoderm of Drosphila declines during germ band retraction to leave a residual population of twisr-expressing cells in the late embryo. In the abdomen, the pattern of twist expression is a simple one: a single cell ventrally, pairs of cells laterally and three cells dorsally in each hemisegment. In the thorax, there are patches of cells associated with the imaginai discs and there are additional clusters in A8 and A9. During larval life, the twist-expressing cells proliferate and, in the abdomen, they form ventral, lateral and dorsal clusters, which are the precursors of the adult abdominal muscles, while in the thorax, they form populations of cells in the imaginai discs that correspond to the adepithelial cells described by previous authors. While most thoracic twist-expressing cells are associated with the discs, the abdominal cells are separate from the precursors of the adult abdominal epidermis, the abdominal histoblasts, and lie on branches of peripheral nerves. The distribution of these cells is tightly linked to the pattern of peripheral nerves, but they segregate normally in da/da embryos despite the absence of the peripheral nervous system.

During embryogenesis, Drosophila constructs a set of muscles that forms a complex pattern on the inner surface of the larval epidermis (Crossley, 1978; Campos Ortega and Hartenstein, 1985; Bate, 1990). At the end of larval life, most of these muscles are demolished and a new adult set forms in their place. The adult muscles assemble, differentiate and attach to adult epidermis, which is formed from the proliferating cells of the imaginai discs and abdominal histoblasts. The adult muscles are made in various ways: some, like the abdominal muscles, are newly constructed, whereas others, such as the dorsal longitudinal indirect flight muscles of the thorax, are built on a scaffold of preexisting larval fibres (Tiegs 1955; Crossley 1978; Fernandes et al. 1991). In both cases, however, the muscles develop from a population of adult myoblasts that appears in the metamorphosing pupa. Thus Drosophila makes two sets of muscles during its development and uses apparently distinct populations of cells to do so.

The relation between the larval and the adult muscleforming cells is important for our understanding of adult muscle differentiation as well as for any analysis of the mechanisms that specify and pattern the adult muscles. In transplant experiments, imaginai discs are a source of muscle-forming cells (Ursprung et al. 1972; Lawrence and Brower, 1982) and, during the late third larval instar, adepithelial cells (so-called for their close association with the inner surface of the disc epithelium; Poodry and Schneiderman, 1970) can be identified in the folds of the mature discs. It can been shown directly that these cells differentiate to form adult muscles (Reed et al. 1975) and transplant experiments show that only those fragments of the wing disc that carry associated adepithelial cells are a source of muscle-forming cells (Lawrence and Brower, 1982). No such cells are found with the abdominal histoblasts, however, even though, during adult development, myoblasts clearly migrate out over the proliferating epidermis of the adult abdomen and assemble into muscles (Currie and Bate, 1991). Various origins have been suggested for the myoblasts that make the adult muscles, including the possibility that they delaminate directly from the imaginai epithelium of the discs or histoblasts (Robertson, 1936; Madhavan and Schneiderman, 1977). The most compelling evidence, however, comes not from direct observations of adult muscle development, but by deduction from clonal analysis and gynandromorph fate mapping. All such studies indicate that the cells that make the adult muscles are in fact derived from the embryonic mesoderm and suggest that in some way, either during embryonic or larval life, these imaginai mesoderm cells become associated with the primordia of the adult epidermis on which they will later differentiate as adult muscles (Hotta and Benzer, 1972; Deak, 1977; Lawrence, 1982; Lawrence and Johnston, 1982).

In this and the following papers we set out to resolve the question of the embryonic origin of the adult muscle cells in Drosophila. A distinctive characteristic of the embryonic myoblasts in Drosophila is that they are derived from a mid-ventral population of cells in the blastoderm, which express the gene twist (Thisse et al. 1988). These zwAz-expressing cells invaginate during gastrulation and spread out over the surface of the embryonic ectoderm to form an inner layer of mesoderm. Later, twist expression declines (Thisse et al. 1988), the synthesis of muscle-specific gene products begins (Leiss et al. 1988, Dohrmann et al. 1990, Michelson et al. 1990) and cells fuse to form the larval muscles (Bate, 1990). Here we show that, as the larval muscles assemble after the retraction of the germ band, twist expression persists in a small, highly reproducible pattern of cells in every segment, and that these cells are the embryonic precursors of the cells that make the adult muscles.

Egg collection and embryo staging

Eggs and larvae were collected from stocks of wild-type (Oregon-R) flies on agar/apple juice plates (Wieschaus and Nüsslein Volhard, 1986) and kept at 25°C. Larvae were collected from the plates at hatching and maintained on yeasted agar plates at 25°C. We also used an amorphic allele of daughterless (da), l(2)IIP106, (Brand and Campos Ortega, 1988), which was kindly provided by Professor J. A. Campos Ortega.

Dissections

Embryos were dissected as previously described (Bate, 1990). Larvae were dissected in a Sylgard-filled Petri dish under Drosophila saline. After cutting along the dorsal mid-line, the sides were pinned out flat and the gut and fat body removed to reveal the nervous system, the muscles and the imaginai discs.

Immunocytochemistry

Embryos and larvae were fixed in 4% paraformaldehyde in phosphate buffer. Undissected embryos were dechorionated, fixed and devitellinised using methods described by Wieschaus and Nüsslein Volhard (1986). Dissected and undissected embryos and larvae were stained with antibodies using standard protocols and the Vectastain ABC Elite kit from Vector Labs. For anti-nv/óv staining, devitellinised embryos or dissected larvae were washed in phosphate-buffered saline (PBS) containing 0.3% Triton-X and 0.5% bovine serum albumin (Sigma), and incubated overnight in preabsorbed anti-Zwzsz antibody at 1:250 in PBS at 4°C. After washing in PBS-TX, the embryos were incubated in goat serum in PBS at a dilution of 1:50 for 1 h before being exposed to preabsorbed biotinylated secondary antibody at 1:200 in PBS-TX for 1 h at room temperature. The protocol for BUdR incorporation and antibody staining has been described previously by Truman and Bate (1988). The anti-zwAz antibody is a generous gift of Dr F. Perrin Schmitt, mAb 22C10 originally described by Zipursky et al. (1984) was kindly provided by Dr C. Cabrera. Anti-HRP was from Cappel.

The embryonic pattern of persistent twist expression twist expression is initiated in the mid-ventra] region of blastoderm stage embryos in a population of cells that invaginates at gastrulation (for a description of twist expression during early embryogenesis see Thisse et al. 1988). The invaginated cells continue to express twist and spread out over the inner surface of the ectoderm as a layer of mesoderm. During the extended germ band stage, twist expression begins to decline and, by the completion of germ band shortening, it only remains in a much smaller, residual population of cells in every segment. In 12 h embryos (Fig. IB), this population has been further reduced to produce a final, highly consistent pattern of persistent zwAz-expressing cells. Most conspicuously, the thoracic segments contain patches of cells while, in the abdomen, the pattern is a strikingly simple one. In each of the segments Al to A7, there is a single ventral cell, a pair of lateral cells just anterior to the segment border and three cells dorsally. All these zwAz-expressing cells lie internally in the embryo, although the ventral cell is closely apposed to the epidermis. In addition, there is another set of dorsal cells located along abdominal segment borders that continues to express twist and is associated with the alary muscles of the heart. We call these cells alary cells and we have not yet investigated their function in the embryo, larva or adult. In A8 and posteriorly, the pattern of persistent twist expression is distinctive. The single ventral cell is missing and there are two patches of 3-4 cells in each hemisegment, close to the invaginated hind gut. We have not so far discovered any sexual dimorphism in this embryonic pattern. In addition, there is a regular arrangement of a small number of ZwAz-expressing cells in the head and persistent twist expression in the visceral mesoderm lining the fore and hind gut. The pattern of twist expression, once established, persists until the embryo hatches.

The segregation of the persistent twist-expressing cells

At the onset of germ band retraction, although there has been a substantial decline in the proportion of mesoderm cells expressing twist, there is still a large population in every segment, twist expression is already segment specific, with distinctive patterns in the thorax, Al to A7 and in A8 and posterior. Some of the twistexpressing cells are still in division at this stage although the final general mesoderm division has occurred some 90 min earlier at 6h AEL (Campos Ortega and Hartenstein, 1985). Already at the onset of shortening, and more conspicuously as the germ band retracts, the pattern in each of the segments can be divided into segmental and intersegmental groups with the segmental cells falling into ventral and dorsal clusters (Fig. 1). The intersegmental cells lie as a small ventrolateral cluster, which straddles each of the segment borders from anterior T2 to posterior A7. At the end of retraction, it is clear that the persistent ventral and dorsal cells are emerging from the segmental population while the lateral cells will segregate from the intersegmental cluster. By 9h AEL, the single most dorsal cell is conspicuous, and usually already isolated from other zwAz-expressing cells. By contrast, the ventral and lateral cells are still emerging from groups of neighbouring cells which are expressing lower levels of twist. By 10 h, these neighbouring cells have ceased to express twist, with the exception of a single cell in the neighbourhood of the lateral pair, which now lies just anterior to the posterior margin of every segment from Al to A7. This additional cell usually disappears rapidly but occasionally persists into 12 h embryos. At the T2/T3 and T3/A1 boundaries, the arrangement of the lateral cells is distinctive. Here the lateral cells persist as groups of three cells, which, at this stage, straddle the segment borders (Fig. 2, Fig. 5A).

Fig. 1.

Embryos stained with anti-twist antibody to reveal persistent twist expression. (A) 8.5 h embryo during germ band retraction, showing large patches of twist expression in the thorax and segmental and intersegmental (arrowed) clusters in the abdomen. (B) 13 h embryo showing the late pattern of twist expression. Patches are present in the thoracic segments where discs have invaginatcd (Bate and Martinez Arias, 1991). In the abdomen (arrows in A2), there are single cells ventraily, pairs of cells laterally and three cells dorsally (not ail visible in this plane of focus). Faint staining dorsally on the abdominal segment margins (arrow heads in A3) reveals alary cells (see text).

Fig. 1.

Embryos stained with anti-twist antibody to reveal persistent twist expression. (A) 8.5 h embryo during germ band retraction, showing large patches of twist expression in the thorax and segmental and intersegmental (arrowed) clusters in the abdomen. (B) 13 h embryo showing the late pattern of twist expression. Patches are present in the thoracic segments where discs have invaginatcd (Bate and Martinez Arias, 1991). In the abdomen (arrows in A2), there are single cells ventraily, pairs of cells laterally and three cells dorsally (not ail visible in this plane of focus). Faint staining dorsally on the abdominal segment margins (arrow heads in A3) reveals alary cells (see text).

Fig. 2.

Dissections of 13 h embryos stained with antibodies to reveal the distribution of twist-expressing cells. (A) Flat preparation stained with anti-twist antibody, showing thoracic clusters in the region of the leg (arrowheads) and dorsal discs (not in this plane of focus). Abdominal segments show ventral and lateral cells (arrows in A3) and some dorsal ceils. Clusters (arrowed) are visible in A8. (B,C) Anti-HRP/anti-twist double stain. (B) Nerves (arrowed) close to twistexpressing cells dorsal to the leg disc (L). Compare Fig. 5A. (C) Posterior branches of SNs (arrowed; see also Fig. 3) contacting pairs of lateral cells (asterisks) in A2 and A3. (D) Anti-22C10/anti-twist double stain, showing association of ventral twist-expressing cells with a branch of the SN and a sense organ (arrowed). (A) Anterior to the left. (B,C,D) Anterior to the left, dorsal up. Scale bars: A=25 μm; B,C,D=5 μm.

Fig. 2.

Dissections of 13 h embryos stained with antibodies to reveal the distribution of twist-expressing cells. (A) Flat preparation stained with anti-twist antibody, showing thoracic clusters in the region of the leg (arrowheads) and dorsal discs (not in this plane of focus). Abdominal segments show ventral and lateral cells (arrows in A3) and some dorsal ceils. Clusters (arrowed) are visible in A8. (B,C) Anti-HRP/anti-twist double stain. (B) Nerves (arrowed) close to twistexpressing cells dorsal to the leg disc (L). Compare Fig. 5A. (C) Posterior branches of SNs (arrowed; see also Fig. 3) contacting pairs of lateral cells (asterisks) in A2 and A3. (D) Anti-22C10/anti-twist double stain, showing association of ventral twist-expressing cells with a branch of the SN and a sense organ (arrowed). (A) Anterior to the left. (B,C,D) Anterior to the left, dorsal up. Scale bars: A=25 μm; B,C,D=5 μm.

The attachment sites of the persistent twist-expressing cells

To locate the zwzsz-expressing cells exactly within the internal structure of the embryo, we made flat preparations of 12-13 h embryos and stained them with antibodies (Fig. 2). In anti-zwwz-stained preparations of this kind, the most conspicuous feature of the expression pattern is that most of the cells in the thorax are clustered around the invaginated primordia of the imaginai discs (Bate and Martinez Arias, 1991). There are 14-15 disc-associated cells in the prothorax and 17-18 in the mesothorax and metathorax. In T2 and T3, the cells are partitioned between the dorsal and ventral discs, with about 8 cells clustered about each of the leg discs and a strand of about 6-7 cells linking these to 2-3 cells attached to the proximal ends of the wing and haltere discs. In the prothorax, the cells are strung out dorsally from a cluster of about 5 cells attached to the leg disc. This pattern is less easy to interpret and may be associated with the dorsal invagination of the anterior spiracle. Other cells arc present laterally and dorsally in T1-T3 that are not obviously associated with the discs and among them are the three lateral cells on the segment borders of T2/T3 and T3/A1. The most dorsal of the cells in each of these lateral clusters is close to the proximal end of the wing and haltere discs. In A8 and A9, we suspect that the distinctive clusters of twistexpressing cells arc associated, like those in the thorax, with invaginated adult primordia, in this case those of the genital disc, but we have not been able to show this in our flat preparations.

Since the zwwz-expressing cells in the thorax are closely linked to the imaginai discs we might expect that if similar cells are also the progenitors of the abdominal muscles, they would be close to the anlagen of the abdominal histoblasts. We have not so far succeeded in locating the abdominal histoblasts in embryos, but it is quite clear that the zwzsZ-expressing cells are not associated with them. Earlier descriptions of the histoblasts in the larva, which we have confirmed, show that they form part of the epidermis close to the dorsal and ventral insertions of the pleural muscles (Madha- van and Schneiderman 1977). None of the twistexpressing cells is located here. The dorsal and lateral cells are internal to the larval muscles, while the ventral cell, although close to the epidermis, is many cell diameters ventral to the prospective site of the ventral histoblast nest.

Association of twist-expressing cells with nerves

Stained with antibodies to HRP or with mAb 22C10, flat preparations of embryos show that all of the abdominal zw/sz-expressing cells and many of those associated with the imaginai discs, are lying on parts of the peripheral nervous system (Fig. 2). The correlation with peripheral nerves is extremely tight and may account for small migrations of the zwsz-expressing cells as they form contacts with outgrowing nerve branches (see below and Discussion). In the abdomen, the ventral cell is associated with a ventral branch of the segmental nerve (SN) and a ventral sense organ described by Bodmer and Jan (1987) as consisting of cells with dendritic arborisations (DA cells). The dorsal cells lie along branches of the intersegmental nerve (ISN). The lateral cell pairs initially lie on a posterior branch of the segmental nerve that innervates muscle 8. Muscle 8 itself is closely associated with the nerve that runs out along the segment boundary from the pairs of mesodermally derived cells (Beer et al. 1987) on the dorsal face of the CNS. This nerve fails to form permanent contacts anterior to the A1/A2 border (unpublished observations; M. Caudy and Y. N. Jan personal communication) and the innervation of muscle 8 is different at the T2β and T3/A1 boundaries from that in more posterior segments. In segments A1-A7, muscle 8 on the posterior segment border is innervated by a posterior branch of the SN, and the lateral pairs of zwMt-expressing cells are associated with this branch (Figs 2C and 3). Later, this nerve grows across muscle 8, establishing a connection with nerves in the next segment posterior (Bodmer and Jan, 1987, see also Fig. 7), and one of the twist-expressing cells migrates to he on the posterior face of muscle 8 (Fig. 3). Uniquely in Al, a branch of the SN also innervates muscle 8 on the anterior, i.e. T3/A1, boundary (Fig. 3). Muscle 8 at the T2/T3 boundary is also innervated by an anterior branch of the SN. Thus at the T2/T3 and the T3/A1 boundaries there is no posterior branch of the SN innervating muscle 8 and the arrangement of the lateral twist-expressing cells is correspondingly different (Fig. 3). In these segments they lie posteriorly to muscle 8 on the anterior branch of the SN. At the A7/A8 boundary the arrangement is also unique: here no nerve connection forms across muscle 8, there is no migration of the twist-expressing cells and they both remain in A7 (Fig. 3).

Fig. 3.

Diagram to show the relation between nerves, muscles and rtrár-expressing cells in embryonic or larval hemisegments. Note unique innervation patterns in Al and A7 (sec also Fig. 7). Muscles arc numbered in Al according to Crossley (1978). Nerve branches shown in A2-A6 are according to Johansen et al. (1989). Nerve designations in A7: SN, segmental nerve; ISN, intersegmental nerve; DN (dashed line), dorsal nerve. Arrowheads denote segment borders. Dorsal up, ventral midline bottom of figure, anterior to the left.

Fig. 3.

Diagram to show the relation between nerves, muscles and rtrár-expressing cells in embryonic or larval hemisegments. Note unique innervation patterns in Al and A7 (sec also Fig. 7). Muscles arc numbered in Al according to Crossley (1978). Nerve branches shown in A2-A6 are according to Johansen et al. (1989). Nerve designations in A7: SN, segmental nerve; ISN, intersegmental nerve; DN (dashed line), dorsal nerve. Arrowheads denote segment borders. Dorsal up, ventral midline bottom of figure, anterior to the left.

Although the persistent twist-expressing cells are distinctively attached to parts of the peripheral nervous system, the nerves do not provide the signal for their initial segregation from the rest of the mesoderm. Segregation begins as the germ band shortens, before sensory and motor nerves have grown out (our observations and Ghysen et al. 1986; Johansen et al. 1989). However, the ventral cell in particular is associated with a larval sense organ and we have looked in homozygous da embryos to see if the cells segregate normally in the complete absence of the larval sensory system (Gaudy et al. 1988). Segregation is relatively normal (Fig. 4) but the final pattern of cells is deranged. None of our observations addresses the question of whether the nerves are required for the proper arrangement of the twist-expressing cells, although this seems very likely, or whether the nerves are a necessary substrate for continued twist expression and subsequent proliferation.

Fig. 4.

twist-expressing cells in a da/da embryo. (A) Superficial focus to show ventral and lateral cells in the abdomen and clusters of cells in the thorax. (B) Internal focus to show the fragmented CNS characteristic (Caudy et al. 1988) of daughterless embryos.

Fig. 4.

twist-expressing cells in a da/da embryo. (A) Superficial focus to show ventral and lateral cells in the abdomen and clusters of cells in the thorax. (B) Internal focus to show the fragmented CNS characteristic (Caudy et al. 1988) of daughterless embryos.

Fig. 5.

twist-expressing cells associated with the wing disc at different developmental stages. (A) twist-expressing cells strung out between the mesothoracic leg (L) and wing (W) discs in a 13 h flat dissected embryo. The three lateral twistexpressing cells on the T2/T3 border (see text) can also be seen out of focus (arrowed). (B) twist-expressing cells on the proximal end of a wing disc (arrowed) in a flat dissected 48 h larva. (C) twist-expressing cells in the wing disc of a flat dissected %h larva. Lightly staining cells are below the disc, not associated with it. Disc-associated cells arc darkly stained and clustered al the proximal end (arrows). (A, B) Dorsal tip. anterior to the left; (C) Distal up, proximal down. Scale bars: A=10 μm, B=25 μtm; C=50 μm.

Fig. 5.

twist-expressing cells associated with the wing disc at different developmental stages. (A) twist-expressing cells strung out between the mesothoracic leg (L) and wing (W) discs in a 13 h flat dissected embryo. The three lateral twistexpressing cells on the T2/T3 border (see text) can also be seen out of focus (arrowed). (B) twist-expressing cells on the proximal end of a wing disc (arrowed) in a flat dissected 48 h larva. (C) twist-expressing cells in the wing disc of a flat dissected %h larva. Lightly staining cells are below the disc, not associated with it. Disc-associated cells arc darkly stained and clustered al the proximal end (arrows). (A, B) Dorsal tip. anterior to the left; (C) Distal up, proximal down. Scale bars: A=10 μm, B=25 μtm; C=50 μm.

twist expression in the larva

In this and the following papers we have largely focussed our attention on the simple pattern of abdominal rn-At-ex pressing cells. However, we have also monitored twist expression in the imaginai discs of dissected larvae. The leg, wing, haltere and genital discs have associated twist-expressing cells from the first instar onwards. There is a dramatic increase in the number of these cells in the course of larval life, so that each disc has a large twist -expressing population attached to it at the end of the third larval instar. Fig. 5 shows for comparison a wing disc from an embryo, a late second instar (48 h after hatching) and from a late third instar (96 h). In this case, it is clear that the adult myoblasts are no longer associated with nerves, because they are clustered in a region of the wing disc that is not innervated at this stage (Murray et al. 1984).

In the abdomen, each of the embryonic cells persists as a single, twist -expressing cell into the first instar larva. From the end of the first larval instar, there is a steady increase in cell numbers so that at the end of iarval life each of the originally single cells is replaced by a group of cells expressing twist. Each segment from Al to A 7 now contains a ventral, a dorsal, and two lateral groups of twist-expressing cells (Fig. 6), the lateral groups sitting on either side of muscle 8, with special patterns in Ai and A7 (Fig. 7). The dorsal group can occasionally be subdivided into three groups, but the location of the cells varies and it is simpler to consider them as one. Typically, there are about 8-10 cells in the ventral group, 15-18 in each of the lateral groups, and some 15 cells dorsally. Because the larvae are dissected dorsally, the likelihood of damage to the dorsal cells is increased and may account for some of the variation that we see. In the accompanying paper (Currie and Bate, 1991), we show that these pools of cells are the precursors of the adult ventral, lateral and dorsal muscles of the abdomen.

Fig. 6.

twist-expressing cells in an abdominal segment of a 96 h larva. Larvae dissected to reveal muscles and stained with anti-iwwz antibody. (A) Dorsal group (muscle is no. 2); (B) lateral groups associated with muscle 8 and (C) ventral group (large biréfringent muscle-is no. 26) Associated larval nucleus is tracheal. Nomarski optics throughout. Scales: A.B,C=25 μm.

Fig. 6.

twist-expressing cells in an abdominal segment of a 96 h larva. Larvae dissected to reveal muscles and stained with anti-iwwz antibody. (A) Dorsal group (muscle is no. 2); (B) lateral groups associated with muscle 8 and (C) ventral group (large biréfringent muscle-is no. 26) Associated larval nucleus is tracheal. Nomarski optics throughout. Scales: A.B,C=25 μm.

Fig. 7.

Patterns of distribution of twist-expressing cells associated with larval muscle 8 and their relation to its innervation. (A) Anterior and posterior innervation associated with muscle 8 in A2-A6 (here A4) of a 96 h larva. (B) As A, lower focus to show nerve (arrowed) linking innervation on anterior and posterior faces of muscle 8. (C) Muscle 8 innervation in Al: left (anterior) muscle 8 is innervated by an anterior branch of SN on its posterior face only. Right (posterior) muscle 8 has A2-A6 type innervation (posterior component only visible here). (D) Muscle 8 innervation in A7: left (anterior) muscle 8 has A2-A6 innervation (link nerve only visible here, arrowed). Right (posterior) muscle 8 is innervated by posterior branch of SN on its anterior face only. There is no nerve linking the anterior and posterior faces. (E) Arrangement of lateral twist -expressing cells in anterior Al: one cluster only (arrows) is present, on the nerve shown (above) in C. (F) Arrangement of lateral twist -expressing cells in posterior A7: one cluster only (arrows) is present, on the nerve shown above in D. (A-D) mAb 22C10, (E-F) anti-twist. Nomarski optics throughout. Anterior to the left, ventral down. Scales: A,B,E,F=25 μm. C,D=50 μm.

Fig. 7.

Patterns of distribution of twist-expressing cells associated with larval muscle 8 and their relation to its innervation. (A) Anterior and posterior innervation associated with muscle 8 in A2-A6 (here A4) of a 96 h larva. (B) As A, lower focus to show nerve (arrowed) linking innervation on anterior and posterior faces of muscle 8. (C) Muscle 8 innervation in Al: left (anterior) muscle 8 is innervated by an anterior branch of SN on its posterior face only. Right (posterior) muscle 8 has A2-A6 type innervation (posterior component only visible here). (D) Muscle 8 innervation in A7: left (anterior) muscle 8 has A2-A6 innervation (link nerve only visible here, arrowed). Right (posterior) muscle 8 is innervated by posterior branch of SN on its anterior face only. There is no nerve linking the anterior and posterior faces. (E) Arrangement of lateral twist -expressing cells in anterior Al: one cluster only (arrows) is present, on the nerve shown (above) in C. (F) Arrangement of lateral twist -expressing cells in posterior A7: one cluster only (arrows) is present, on the nerve shown above in D. (A-D) mAb 22C10, (E-F) anti-twist. Nomarski optics throughout. Anterior to the left, ventral down. Scales: A,B,E,F=25 μm. C,D=50 μm.

Cell proliferation and the formation of the twist expressing cell groups

To establish whether the pools of twist-expressing cells that we identify in late larvae are produced by recruitment, or are clonally derived by proliferation from the original twist-expressing cells in the embryo, we monitored DNA synthesis in larval cells using BUdR and coupled this with an analysis of the increase in twist cell numbers. The number of abdominal twistexpressing cells remains constant throughout the first instar and only begins to rise at the beginning of the second (Fig. 8). This increase in cell number is strikingly consistent for the different cell groups: at 28 h there are variably 1 or 2 cells in each ventral cluster, 1 cell either side of muscle 8, and three single cells dorsally. At 32 h there are two cells ventrally, two cells on either side of muscle 8 and dorsally cells are present as pairs. Clearly there is a doubling of cell number between 28 and 32h with the ventral cells leading.

Fig. 8.

Increasing size of twist -exprcssing cell clusters during larval development. Each block represents the cells in single ventral, lateral and dorsal clusters averaged for 15 segments.

Fig. 8.

Increasing size of twist -exprcssing cell clusters during larval development. Each block represents the cells in single ventral, lateral and dorsal clusters averaged for 15 segments.

When larvae are fed for 24 h on a diet containing BUdR, cells that have entered S phase during this period, and their progeny, are revealed by antibody staining. BUdR fed from 24 to 48, from 48 to 72, and from 72 to 96 h after hatching labels the cells in each of the dorsal, ventral and lateral clusters (Fig. 9). At each stage, the number of cells labelled by BUdR incorporation matches the number labelled with the anti-twist antibody. We conclude that the twist-expressing cells are replicating their DNA as their numbers increase. More conclusively, we often see individual twistexpressing cells in mitosis during larval life (Fig. 9) confirming that they do indeed divide. In summary, we find (1) that each cell in a larval twist-ex pressin g cluster replicates its DNA; (2) the cells remain small, indicating that, unlike the neighbouring larva! cells, DNA replication leads to division, not to an increase in ploidy; (3) occasional twist-expressing cells are seen in division and (4) the increase in the twist-expressing cell number is completely accounted for by the observed pattern of BUdR labelling. While we cannot entirely exclude recruitment as a possible contribution to the twist-expressing clusters on the basis of these data alone, we feel that the evidence is strong that each group of adult myoblasts is formed by proliferation and that each is clonally derived from a single embryonic progenitor cell.

Fig. 9.

(A) Anti- twist staining of a 48 h larva dissected to show muscle 8 in A3. On the posterior face of the muscle there is a cluster of three twist-expressing cells. Of these, one (arrowed) is in division. (B) Ànti-BÙdR staining of muscle 8 region in A3 of a 72 h larva fed on a BUdR- containing diet from 48-72 h. Two clusters (arrows) of imaginai cells (small nuclei) on either side of the muscle have replicated their DNA during this period. Large nuclei are polyploid larval muscle nuclei, many of which have also replicated their DNA during the exposure to BUdR. Dorsal up, anterior to the left. Scales: A=10 μm; B=25 μm.

Fig. 9.

(A) Anti- twist staining of a 48 h larva dissected to show muscle 8 in A3. On the posterior face of the muscle there is a cluster of three twist-expressing cells. Of these, one (arrowed) is in division. (B) Ànti-BÙdR staining of muscle 8 region in A3 of a 72 h larva fed on a BUdR- containing diet from 48-72 h. Two clusters (arrows) of imaginai cells (small nuclei) on either side of the muscle have replicated their DNA during this period. Large nuclei are polyploid larval muscle nuclei, many of which have also replicated their DNA during the exposure to BUdR. Dorsal up, anterior to the left. Scales: A=10 μm; B=25 μm.

Homozygous twist embryos fail to gastrulate and lack mesodermally derived internal structures (Niisslein Volhard et al. 1984; Simpson, 1983). The gene twist encodes a DNA-binding protein of the HLH class, which is initially expressed in a mid-ventral population of blastoderm cells from which the mesoderm is derived (Thisse et al. 1987; 1988). It continues to be expressed in at least a proportion of the mesoderm long after gastrulation has occurred (Thisse et al. 1988) and until larval muscle differentiation begins. Because of the drastic gastrulation phenotype produced by loss of function at the twist locus little is known of twist’s role, if any, in the subsequent development and differentiation of mesodermally derived tissues. Nonetheless, twist expression, which is initially present in all ventral blastoderm cells, comes to be uniquely a property of those cells that are restricted (Beer et al. 1987) to mesodermal pathways of differentiation. Many of these cells continue to express twist until the onset of germ band retraction and the differentiation of the larval musculature. Here we have shown that twist expression then persists in a limited, highly consistent set of cells and that these cells proliferate during larval life, producing pools of twist-expressing cells, that are the myoblasts from which the adult muscles will be formed (Currie and Bate, 1991). The behaviour of the adult myoblasts highlights special characteristics of adhesion, migration and proliferation associated with twist expression and suggests that twist may either be required for cells to have access to mesodermal pathways of differentiation and/or to retain a set of quasi embryonic properties prior to the onset of differentiation. Since twist (so far as we know) is switched on only once during development by the local action of dorsal in the mid- ventral cells of the blastoderm (Thisse et al. 1987; Roth et al. 1989), the characteristic of being assigned to a mesodermal pathway of differentiation, plus the proliferative state, are propagated through the twistexpressing cells, which persist in the late embryo.

The relation between the adult muscles and their precursors in the embryo is like that of the adult epidermis and neurons to the imaginai discs and neuroblasts, respectively. In all three cases, elements of the adult pattern are laid down in the embryo but proliferation and differentiation are delayed until the larval and pupal stages. Properties that are characteristic of future adult cells are only manifested about half way through embryogenesis when the details of the embryonic pattern have already been laid down. Thus, the imaginai disc cells segregate from the rest of the epidermis and invaginate into the interior of the embryo after the completion of the segmental patterning of the epidermis (Bate and Martinez Arias, 1991). The neuroblasts produce a full complement of larval neurons and then reduce in size, before reappearing in larval life to produce adult neurons (Truman and Bate, 1988; Prokop and Technau, 1991). In no case are future adult cells derived as a distinct population that takes no part in embryonic development. This may be because the acquisition of specifically adult properties by a limited number of cells is a relatively late event both in evolution and in development. Thus the uniquely adult characteristics of segregation and delayed proliferation and differentiation are an adaptation that is imposed on an earlier, more generalised pattern-forming mechanism which lays out the fundamental structure of the fly.

It is likely that the specification of adult muscle cells is similarly integrated into the general patterning of the embryonic mesoderm. The special properties of the future adult myoblasts (i.e. their persistent twist expression, adhesion to nerves, continued proliferation) only appear after the patterning of the larval muscles is complete (although the ‘adultness’ of these cells may have been specified earlier than this). In this view, the fly delays the production of adult structures until larval and pupal stages, but, at least partly, specifies the elements of the adult pattern as part of a general patterning process that takes place in the embryo. In the case of the muscles, an essential part of the patterning process may be the specification of founder cells (Bate, 1990; Dohrmann et al. 1990), with which other cells fuse to form particular muscles. It may be that the embryonic patterning mechanism lays out a pattern of founder cells, both larval and adult: larval cells now begin to express muscle-specific genes, turn off twist and fuse to form muscles, whereas adult cells continue to express twist, retain a capacity for proliferation and delay differentiation until the end of larval life. Two points follow from this hypothesis: first, that we might expect genes such as S59, which is expressed in a subset of larval founder cells (Dohrmann et al. 1990), to be expressed in the embryonic progenitors of adult muscle cells, and second, that we do not know what, if any, restriction there may be on the developmental potential of the individual adult muscle cells that we identify in the embryo. We have made a detailed experimental analysis of this second issue in the accompanying paper (Broadie and Bate, 1991).

Although the early appearance of the adult myoblasts and the immunity of this process to loss of da function show that segregation is independent of the peripheral nervous system, the detailed distribution of the twistexpressing cells and the way in which this distribution changes are not. Thus, in A7, T2 and T3, the peculiar innervation of muscle 8 and associated sense organs is directly reflected in the subsequent arrangement of the lateral cell clusters. The lateral cells in posterior Al to anterior A7 undertake a characteristic redistribution late in embryogenesis. This coincides with the formation of a new nerve branch (Bodmer and Jan, 1987), which grows over muscle 8 and probably provides the pathway on which the posterior of the two lateral cells migrates. It is certainly the case that (a) this cell migrates and that a second cluster of iwwt-expressing cells develops on the nerve posterior to muscle 8 and (b) the migration and formation of the second cluster fail to occur at the A7/8 boundary where there is no nerve pathway across muscle 8 (Fig. 7). More tenuously, but probably just as significantly, the positions of the twistexpressing cells also change earlier in embryogenesis as sensory and motor nerves grow out. Thus the ventral abdominal cell moves from a posterior to a slightly more anterior position in the segment, where it comes to lie on a ventral branch of the segmental nerve. It is not inconceivable (although we have no evidence on this point), that the twist-expressing cells themselves offer a substrate or a target for nerve growth, and that it is this which ensures that nerves and myoblasts are intimately linked in subsequent development. In any event, if we view this linkage as part of a developmental sequence in which the differentiation of the adult muscles is delayed, then an association between growing axons and embryonic muscle cells is not unexpected. We know, for example, that, in the grasshopper embryo, muscle pioneers offer guidance cues to outgrowing motor axons (Ball et al. 1985). One way of viewing these myoblast/nerve associations is as evolutionary echoes of an earlier state in which the adult muscles were innervated by the branches of the nerves with which they are connected and that now provide a substrate on which the precursors of these muscles proliferate, migrate and differentiate.

We are left with the unsolved questions of mesodermal patterning in general and the segregation of the adult myoblasts in particular. There is as yet no convincing evidence either for the view that the patterning mechanism is intrinsic to the mesoderm, or for the alternative that it involves an inductive interaction with the ectoderm. It is, nonetheless an attractive possibility that it is the epidermis that triggers the development of local differences between mesoderm cells such that, for example, in the thorax, patches of twist -expressing cells persist over the anlagen of the imaginai discs. It is also possible that the mechanisms that lead to the local segregation of these cells, and the others that we have described, are the same as those that specify the larval muscle pattern. Persistent twist expression in a consistent, readily detectable set of embryonic cells, which we have described here, gives us another marker, like S59 (Dohrmann et al. 1990) and nau (Michelson et al. 1990), with which to resolve the issue of how patterns are formed in the mesoderm.

We are grateful to Fabienne Perrin Schmitt for the gift of the anti-twist antibody and to Alfonso Martinez Arias and Helen Skaer for their comments and advice. This work is supported by a grant from the Wellcome Trust.

Ball
,
E. E.
,
Ho
,
R. K.
and
Goodman
,
C. S.
(
1985
).
Development of neuromuscular specificity in the grasshopper embryo: guidance of motoneuron growth cones by muscle pioneers
.
J. Neurosci
.
5
,
1808
1819
.
Bate
,
M.
(
1990
).
The embryonic development of larval muscles in Drosophila
.
Development
110
,
791
804
.
Bate
,
M.
and
Martinez Arias
,
A.
(
1991
).
The embryonic origin of imaginai discs in Drosophila
.
Development (in press)
.
Beer
,
J.
,
Technau
,
G.
and
Campos Ortega
,
J. A.
(
1987
).
Lineage analysis of transplanted individual cells in embryos of Drosophila melanogaster. IV. Commitment and proliferative capabilities of individual mesodermal cells
.
Roux’ Arch, devl Biol
.
196
,
222
230
.
Bodmer
,
R.
and
Jan
,
Y. N.
(
1987
).
Morphological differentiation of the embryonic peripheral neurons in Drosophila
.
Roux’ Arch, devl Biol
.
196
,
69
77
.
Brand
,
M.
and
Campos Ortega
,
J. A.
(
1988
).
Two groups of interrelated genes regulate early neurogenesis in Drosophila melanogaster
.
Roux’ Arch, devl Biol
.
197
,
457
470
.
Broadie
,
K.
and
Bate
,
M.
(
1991
).
Development of adult muscles in Drosophila: Ablation of identified muscle precursor cells
.
Development
113
,
103
118
.
Campos Ortega
,
J. A.
and
Hartenstein
,
V.
(
1985
).
The Embryonic Development of Drosophila melanogaster
.
Berlin, Heidelberg, New York, Tokyo
:
Springer Verlag
.
Caudy
,
M.
,
Grell
,
E. H.
,
Dambly Chaudière
,
CH.
,
Ghysen
,
A.
,
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1988
).
The maternal sex determination gene daughterless has zygotic activity necessary for the formation of peripheral neurons in Drosophila
.
Genes Dev
.
2
,
843
852
.
Crossley
,
A. C.
(
1978
).
The morphology and development of the Drosophila muscular system
.
In The Genetics and Biology of Drosophila
2b
(eds.
M.
Ashburner
and
T.R.F.
Wright
) pp
499
560
.
London, New York, San Francisco
:
Academic Press
.
Currie
,
D. A.
and
Bate
,
M.
(
1991
).
The development of adult muscles in Drosophila’. Adult myoblasts express twist and are associated with nerves
.
Development
113
,
000
000
.
Deak
,
I.
(
1977
).
Mutations of Drosophila melanogaster that affect muscles
.
J.Embryol. exp. Morph
.
40
,
35
63
.
Dohrmann
,
C.
,
Azpiazu
,
N.
and
Frasch
,
M.
(
1990
).
A new Drosophila horneo box gene is expressed in mesodermal precursor cells of distinct muscles during embryogenesis
.
Genes Dev
.
4
,
2098
2111
.
Fernandes
,
J.
,
Bate
,
M.
and
Vijayraghavan
,
K.
(
1991
).
The development of the indirect flight muscles of Drosophila
.
Development
113
,
67
77
.
Ghysen
,
A.
,
Dambly Chaudière
CH.
,
Aceves
,
E.
,
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1986
).
Sensory neurons and peripheral pathways in Drosophila embryos
.
Roux’ Arch, devl Biol
.
195
,
281
289
.
Hotta
,
Y.
and
Benzer
,
S.
(
1972
).
Mapping of behaviour in. Drosophila mosaics
.
Nature
240
,
527
535
.
Johansen
,
J.
,
Halpern
,
M. E.
and
Keshishian
,
H.
(
1989
).
Axonal guidance and the development of muscle fiber-specific innervation in Drosophila embryos
.
J. Neurosci
.
9
,
4318
4332
.
Lawrence
,
P. A.
(
1982
).
Cell lineage of the thoracic muscles of Drosophila
.
Cell
29
,
493
503
.
Lawrence
,
P. A.
and
Brower
,
D. L.
(
1982
).
Myoblasts from Drosophila wing disks can contribute to developing muscles throughout the fly
.
Nature
295
,
55
57
.
Lawrence
,
P. A.
and
Johnston
,
P.
(
1982
).
Cell lineage of the Drosophila abdomen: the epidermis, oenocytes and ventral muscles
.
J. Embryol. exp. Morph
.
72
,
197
208
.
Leiss
,
D.
,
Hinz
,
U.
,
Gasch
,
A.
,
Mertz
,
R.
and
Renkawitz Pohl
,
R.
(
1988
).
B3 tubulin expression characterizes the differentiating mesodermal germ layer during Drosophila embryogenesis
.
Development
104
,
525
531
.
Madhavan
,
M. M.
and
Schneiderman
,
H. A.
(
1977
).
Histological analysis of the dynamics of growth of imaginai discs and histoblast nests during the larval development of Drosophila melanogaster
.
Wilhelm Roux’ Arch, devl Biol
.
166
,
1
44
.
Michelson
,
A.
,
Abmayr
,
S.
,
Bate
,
M.
,
Martinez Arias
,
A.
and
Maniatis
,
T.
(
1990
).
Expression of a MyoD family member prefigures muscle pattern in Drosophila embryos
.
Genes Dev
.
4
,
2086
2097
.
Murray
,
M. A.
,
Schubiger
,
M.
and
Palka
,
J.
(
1984
).
Neuron differentiation and axon growth in the developing wing of Drosophila melanogaster
.
Devl Biol
.
104
,
259
273
.
Nüsslein Volhard
,
C.
,
Wieschaus
,
E.
and
Kluding
,
H.
(
1984
).
Mutations affecting the pattern of the larval cuticle in Drosophila melanogaster. I. Zygotic loci on the second chromosome
.
Wilhelm Roux’ Arch, devl Biol
.
193
,
267
282
.
Poodry
,
C. A.
and
Schneiderman
,
H. A.
(
1970
).
The ultrastructure of the developing leg of Drosophila melanogaster
.
Wilhelm Roux’ Arch, devl Biol
.
166
,
1
44
.
Prokop
,
A.
and
Technau
,
G. M.
(
1991
).
The origin of postembryonic neuroblasts in the ventral nerve cord of Drosophila melanogaster
.
Development
111
,
79
88
.
Reed
,
C. T.
,
Murphy
,
C.
and
Fristrom
,
D.
(
1975
).
The ultrastructure of the differentiating pupal leg of Drosophila melanogaster
.
Wilhelm Roux’ Arch, devl Biol
.
178
,
285
302
.
Robertson
,
C. W.
(
1936
).
Metamorphosis of Drosophila melanogaster including an accurately timed account of the principle morphological changes
.
J. Morph
.
59
,
351
399
.
Roth
,
S.
,
Stein
,
D.
and
Nüsslein Volhard
,
C.
(
1989
).
A gradient of nuclear localization of the dorsal protein determines dorsoventral pattern in the Drosophila embryo
.
Cell
59
,
1189
1202
.
Simpson
,
P.
(
1983
).
Maternal-zygotic gene interactions during the formation of the dorsoventral pattern in Drosophila embryos
.
Genetics
105
,
615
632
.
Thisse
,
B.
,
Stoetzel
,
C.
,
El Messal
,
M.
and
Perrin Schmitt
,
F.
(
1987
).
Genes of the Drosophila maternal dorsal group conrol the specific expression of the zygotic gene twist in presumptive mesodermal cells
.
Genes Dev
.
1
,
709
715
.
Thisse
,
B.
,
Stoetzel
,
C.
,
Gorostiza
,
T. C.
and
Perrin Schmitt
,
F.
(
1988
).
Sequence of the twist gene and nuclear localization of its protein in endomesodermal cells of early Drosophila embryos
.
EMBO J
.
7
,
2175
2183
.
Tiegs
,
O. W.
(
1955
).
The flight muscles of insects
.
Phil. Trans. Roy. Soc. Lond. B
.
238
,
221
348
.
Truman
,
J. W.
and
Bate
,
M.
(
1988
).
Spatial and temporal patterns of neurogenesis in the central nervous system of Drosophila melanogaster
.
Devl Biol
.
125
,
145
157
.
Ursprung
,
H.
,
Concience-Egli
,
M.
,
Fox
,
D.
and
Walliman
,
T.
(
1972
).
Origin of leg musculature during Drosophila metamorphosis
.
Proc. natn. Acad. Sci. U.S.A
.
69
,
2812
2813
.
Wieschaus
,
E.
and
Nüsslein Volhard
,
C.
(
1986
).
Looking at embryos
.
In Drosophila a Practical Approach
(ed.
D. B.
Roberts
) pp.
199
227
.
Oxford, Washington D.C
.:
IRL Press
.
Zipursky
,
S. L.
,
Venkatesh
,
T. R.
,
Teplow
,
D. B.
and
Benzer
,
S.
(
1984
).
Neuronal development in the Drosophila retina: monoclonal antibodies as molecular probes
.
Cell
36
,
15
26
.