During metamorphosis, the adult muscles of the Drosophila abdomen develop from pools of myoblasts that are present in the larva. The adult myoblasts express twist in the third larval instar and the early pupa and are closely associated with nerves. Growing adult nerves and the twist-expressing cells migrate out across the developing abdominal epidermis, and as twist expression declines, the myoblasts begin to synthesize β3 tubulin. There follows a process involving cell fusion and segregation into cell groups to form multinucleate muscle precursors. These bipolar precursors migrate at both ends to find their correct attachment points, β3 tubulin expression continues at least until 51 h APF by which time the adult muscle pattern has been established.

Like other holometabolous insects, Drosophila makes two sets of muscles. The larval muscles form during embryogenesis whereas the development of the second set, the adult muscles, is postponed until metamorphosis at the end of larval life. In this and the accompanying papers, we are concerned with the questions of how the adult musculature develops in Drosophila, where the adult myoblasts come from and how the muscles are patterned. We concentrate on the abdomen as an accessible system in which to study the formation of adult muscles. The muscles of the abdomen develop de novo and they do so on a simple epithelium, which first begins to grow at the onset of metamorphosis.

There are many gaps in our understanding of adult myogenesis in the fly. The question of the origin of the adult muscle-forming cells has long been a matter of contention. By 1910 there had already been suggestions that they originated from the fat body, the cells of the imaginai discs, or from within the larval muscles themselves (reviewed in Lowne, 1890; Perez, 1910). More recently gynandromorph fate maps (Hotta and Benzer, 1972; Deak, 1977), clonal analysis (Lawrence, 1982) and nuclear transplantation experiments (Lawrence and Johnston, 1986a) have all indicated that the adult muscles have a mesodermal origin. Experiments involving the transplantation of discs show that the muscles that develop in association with them originate from cells within the discs (Ursprung et al. 1972) and these muscle-forming cells can be identified as the so- called adepithelial cells which are present in imaginai discs in the third larval instar (Lawrence, 1982; Reed et al. 1975). No such cells have been found in association with the abdominal histoblasts and the cells which make the abdominal muscles have not previously been identified.

Here we show that the abdominal myoblasts, like the adepithelial cells of the discs, are derived from distinct populations of mesoderm cells that can be identified in the larva. In the previous paper, we showed that cells with persistent twist expression are set aside in the embryo and divide to produce clusters of twistexpressing cells in the third instar larva (Bate et al. 1991). In this paper, we document the development of the abdominal muscles from these cell clusters. Each cluster gives rise to a specific part of the abdominal muscle set. As in the embryo, so in the adult, muscle differentiation is preceded by a phase of proliferation in a population of myoblasts that expresses twist. The expression of muscle-specific genes is initiated as twist expression declines and eventually disappears. Unlike the situation in the embryo, nerve pathways are crucially involved in the distribution of the proliferating myoblasts as they migrate to sites of myogenesis on the newly differentiating adult epidermis. Our demonstration of a direct role of nerves in adult myogenesis is paralleled by the experiments of Lawrence and Johnston (1986b) which show that, at least for one adult muscle, innervation is a decisive factor in the formation of the muscle pattern.

Collection and timing of pupae

Wild-type (Oregon-R) Drosophila melanogaster were raised on a standard medium al 25°C. Prepupae were collected from the tubes at the white prepupal stage. All times are given with this point as zero, i.e. Oh APF (after puparium formation). Prepupae were placed on moist filter paper in Petri dishes and incubated at 25°C.

A stock containing a j8-galactosidase gene linked to the myosin heavy chain promoter was used for Fig. 1. By staining for β-galactosidase activity, we were able to reveal the adult muscle pattern. This stock was generously provided by Dr S. Bernstein and Dr N. Hess.

Fig. 1.

(A) Flat preparation of the abdomen of an adult female Drosophila showing the three main groups of adult muscle, d, dorsal muscles: 1, lateral muscles; v, ventral muscles. Scale bar 100 μm. (B) Pattern of adult muscles in the abdomen at eclosion (after Miller 1950). d, dorsal muscles; pld, persistent larval dorsal muscles; I, lateral muscles; pl, persistent iarval lateral muscles; v, ventral muscles; s, spiracular muscles. (C) Position of abdominal histoblast nests in two segments of the larva (after Roseland and Reinhardt 1982). adh, anterior dorsal histoblast nest; pd, posterior dorsal histoblast nest; sp, spiracular nest; v, ventral nest. Arrows indicate direction of migration of cells from each nest during metamorphosis.

Fig. 1.

(A) Flat preparation of the abdomen of an adult female Drosophila showing the three main groups of adult muscle, d, dorsal muscles: 1, lateral muscles; v, ventral muscles. Scale bar 100 μm. (B) Pattern of adult muscles in the abdomen at eclosion (after Miller 1950). d, dorsal muscles; pld, persistent larval dorsal muscles; I, lateral muscles; pl, persistent iarval lateral muscles; v, ventral muscles; s, spiracular muscles. (C) Position of abdominal histoblast nests in two segments of the larva (after Roseland and Reinhardt 1982). adh, anterior dorsal histoblast nest; pd, posterior dorsal histoblast nest; sp, spiracular nest; v, ventral nest. Arrows indicate direction of migration of cells from each nest during metamorphosis.

Flat preparations of the abdomens were made by slitting along either the dorsal, ventral or lateral midlirte and pinning them out flat. The gut and any fat cells were then carefully removed.

Antibody staining

Flat preparations were fixed for 45-60 min in 4 % paraformaldehyde in phosphate-buffcred saline (PBS) and then washed three or four times in phosphate-buffered saline with 0.3% Triton-X 100 (PBS-TX). After incubation in PBS with 0.5% bovine serum albumin (Sigma) for one hour at room temperature, they were washed twice in PBS-TX before incubation in primary antibody (1:200 dilution) overnight at 4 °C. The primary antibodies used in this study included an anti-twist antibody, generously donated by Dr F. Perrin- Schmitt, an anti- β3 tubulin antibody generously donated by Dr D. Leiss, an anti-horseradish peroxidase (HRP) antibody and an anti- β-galactosidase antibody. After overnight incubation, the primary antibody was removed and the abdomens were washed four times in PBS-TX and then incubated for one hour in a 1:50 dilution of goat serum. After a quick wash, they were incubated in a 1:300 concentration of biotinylated secondary antibody for 1 h at room temperature. After four washes in PBS-TX they were exposed to a peroxidase-linked avidin/biotin complex (Vectalabs ABC Elite) for 30 min. The preparations were thoroughly washed and then the peroxidase label was revealed in a reaction medium consisting of PBS, diaminobenzidine (Sigma), NiCl2 and H2O2 under a microscope to determine the end point. The reaction was stopped by replacing the reaction medium with 30% alcohol. The preparations were dehydrated, cleared in cedar wood oil and mounted in DPX. These mounted specimens were examined using 16 ×, 25 ×, 40 ×, and 63 × oil immersion objectives on a Zeiss microscope fitted with Nomarski optics.

Sectioning

Material for sectioning was stained with antibodies as described above. The preparations were then dehydrated, incubated in propylene oxide for 1 h and taken through an increasing series of Araldite/propylene oxide mixtures. They were then embedded in Araldite and semi-thin sections (1/tm) were cut on a Reichert microtome. Sections were placed on a slide and alternate sections were then stained with toluidine blue. The slides were viewed on a Zeiss microscope using 63 × and 100 × bright-field and phase-contrast optics.

The adult muscle pattern

The pattern of muscles in the adult abdomen is shown in Fig. 1 and has been described previously by Miller (1950). In this paper, we concentrate on the musculature of A2-A7. The pattern in the terminal segments is complicated by the musculature of the genitalia and analia and the pattern in Al is slightly different from that of the other abdominal segments. In A2-A7, there are three major groups of muscles, dorsal, lateral and ventral, and a small muscle in each hemisegment that is associated with the lateral spiracle. In each hemisegment, the dorsal muscles consist of 17-22 parallel longitudinal fibres, while the lateral (tergosternal) muscles are composed of about 20 parallel fibres which run dorsoventrally and are attached at the dorsal and ventral limits of the pleura. There is one set of ventral muscles in each hemisegment, which consists of 5-8 fibres, just lateral to the ventral midline.

A few larval muscles persist through metamorphosis but degenerate 48 h after adult eclosion (Kimura and Truman, 1990). These are the temporary oblique dorsal muscles: the dorsal internal oblique muscles of the larva no.l, no.2 and no.3, (for muscle numbering see Crossley, 1978) which consist of at least two large muscles in each hemisegment, and the temporary internal lateral muscle no.8, a single large vertical fibre that spans the pleura in the intersegmental region.

The development of the new adult muscles is intricately linked with the development of the adult epidermis and adult innervation. To understand how the pattern of new muscles is established, it is necessary to begin by providing a brief description of these events.

Epidermal development

In Drosophila the adult abdominal epidermis develops from sets of abdominal histoblasts which are arranged in four nests in characteristic positions in each abdominal hemisegment (Fig. 1). There are two dorsal, one ventral and one spiracular nest, each contributing to a separate region of the adult epidermis (Roseland and Schneiderman, 1979). During larval life the histoblasts do not divide and remain diploid, unlike their increasingly polyploid larval neighbours (Pearson, 1972). After the onset of metamorphosis, the histoblasts divide rapidly and migrate outwards (Fig. 1) to replace the dying larval epidermis (Madhavan and Madhavan, 1980). In each hemisegment, the anterior and posterior dorsal nests are the first to meet and fuse. The ventral nest fuses with the spiracular nest by 24 h APF. The histoblasts continue to spread outwards so that by 28 h APF the dorsal and fused spiracular and ventral nests have met laterally. By 36 h APF the histoblasts from the two hemisegments meet at the dorsal and ventral midline. It is upon this dynamic, migrating group of cells that adult muscle development occurs.

The development of adult innervation

We studied the organization and development of the adult pattern of abdominal innervation using antibodies to horseradish peroxidase (HRP) which recognise glycoproteins in insect nerve cell membranes (Jan and Jan, 1982). The pattern of nerves and muscles in each larval segment is almost completely rearranged during metamorphosis. Here we concentrate on the restructuring of the abdominal innervation.

In the first 24 h APF, nearly all the larval muscles are broken down by histolysis and phagocytosis. The nerves that innervate these muscles also disappear. A single major nerve trunk remains in each hemisegment and maintains the innervation of those larval muscles that persist. This nerve remains in contact with the epidermis in four positions in each hemisegment (Fig. 2). It is most likely that these points of association with the epidermis correspond to positions where larval epidermal sensory structures were previously located. By 20 h APF, it is clear that two of these points of contact are situated at the ventral (a, in Fig. 2A) and lateral (b) sides of the ventral histoblast nest, while more dorsally there is a small branch from the nerve that contacts the epidermis at a point between the dorsal and ventral histoblast nests (c), close to the spiracular nest. The fourth point of contact is situated just dorsolateral to the dorsal histoblast nests (d).

Fig. 2.

Diagrammatic representation of the position and migration of nerves in one hemisegment of the pupal abdomen. (A) 10 h APF, (B) 30 h APF, (C) 90 h APF. vm, ventral midline; dm, dorsal midline; n, nerve. Points of attachment of nerve to epidermis, a-d. pm, persistent larval muscles; gc, growth cones from nerves; 8, muscle no.8 (pleural internal transverse); h, dorsal histoblast nests; v, ventral histoblast nest. Anterior is up.

Fig. 2.

Diagrammatic representation of the position and migration of nerves in one hemisegment of the pupal abdomen. (A) 10 h APF, (B) 30 h APF, (C) 90 h APF. vm, ventral midline; dm, dorsal midline; n, nerve. Points of attachment of nerve to epidermis, a-d. pm, persistent larval muscles; gc, growth cones from nerves; 8, muscle no.8 (pleural internal transverse); h, dorsal histoblast nests; v, ventral histoblast nest. Anterior is up.

As the larval nerves and muscles break down, the establishment of a new adult innervation begins with the initiation of nerve growth at three of the nerve/epidermis contact points in each hemisegment. Numerous fine filopodial extensions emanate from the nerve at each of these points. We have not observed the development of the nerve at the fourth (spiracular) attachment point (c). The positions of the apparent growth cones are illustrated diagrammatically in Fig. 2: one at the most ventral site of attachment (a), one at the ventrolateral site of contact (b) and one dorsolateral to the dorsal histoblast nest (d). The points of contact with the epidermis from which these apparent growth cones arise are originally associated with larval cells and, as the larval cells are replaced by proliferating histoblasts, so the growth cones continue to migrate out over the expanding adult epidermis.

By 26 h APF, each of the three growth cones is migrating a few cell diameters behind an advancing front of histoblast cells. The dorsal growth cones grow out on the epidermis in the middle of the histoblast nest. As they migrate medially, they make numerous short lateral projections (Fig. 3). The relevance of these projections to the establishment of the muscle pattern will be discussed later. By 36 h APF these dorsal growth cones have completed their migration from a lateral position in the abdomen to a point just short of the dorsal midline and by 48 h the lateral projections from the dorsal nerve remain only where they innervate new adult muscles (Fig. 3C). The lateral innervation develops in a slightly different fashion, forming an array of very fine, branching processes, which will innervate the numerous, parallel muscles in that part of each segment. The ventral growth cone moves ventrally with the migrating histoblast cells. It does not send out as many processes as the dorsal growth cone and the axons from this growth cone will finally innervate the adult ventral muscle which is situated medipventrally.

Fig. 3.

Anti-HRP-stained preparations to show growth of the dorsal nerve at three stages during pupal development. Anterior is up. (A) Growth zone at 22 h APF. Scale bar 10 μm. (B) Nerve growth at 35 h APF. There are numerous short projections emanating from the nerve (arrows). n, nerve; pld, persistent larval dorsal muscle. Scale bar 10 μm. (C) Pattern of nerve branches that innervate the developing adult dorsal muscles at 44 h APF. Also evident are the axons from bristles on the epidermis (a), b, nerve branch; d, adult dorsal muscle. Scale bar 20 μm.

Fig. 3.

Anti-HRP-stained preparations to show growth of the dorsal nerve at three stages during pupal development. Anterior is up. (A) Growth zone at 22 h APF. Scale bar 10 μm. (B) Nerve growth at 35 h APF. There are numerous short projections emanating from the nerve (arrows). n, nerve; pld, persistent larval dorsal muscle. Scale bar 10 μm. (C) Pattern of nerve branches that innervate the developing adult dorsal muscles at 44 h APF. Also evident are the axons from bristles on the epidermis (a), b, nerve branch; d, adult dorsal muscle. Scale bar 20 μm.

twist expression and muscle assembly

In the late third instar larva, there are six main groups of twist-expressing cells in each abdominal hemisegment, each group containing 8-15 cells (Bate et al. 1991). This arrangement persists in the early pupa and in each case the /ivrsr-expressing cells are still closely associated with a nerve (Fig. 4). The most ventral group is located on the nerve between muscle no.7 and muscle no.28. More laterally there are two groups associated with the anterior and posterior faces of muscle no.8 and the nerves running to it. The dorsal groups are associated with the nerves that pass close to the dorsal interna] oblique muscle no.3 and the dorsal external oblique muscle no. 10. In the thoracic segments, twistexpressing cells are also associated with the imaginai discs (Fig. 5).

Fig. 4.

Groups of twist-expressing cells (arrowheads) associated with nerves 5.5 h APF. 8, muscle no. 8; n, nerve. Scale bar 50 μm,

Fig. 4.

Groups of twist-expressing cells (arrowheads) associated with nerves 5.5 h APF. 8, muscle no. 8; n, nerve. Scale bar 50 μm,

Fig. 5.

twist-expressing cells (arrows) in an unevaginated leg disc in white prepupa Oh APF. ps, peripodial stalk. Scale bar 30 μm.

Fig. 5.

twist-expressing cells (arrows) in an unevaginated leg disc in white prepupa Oh APF. ps, peripodial stalk. Scale bar 30 μm.

From the onset of pupariation (Oh APF), the twistexpressing cells increase in number while the larval muscles are being broken down and by 13 h APF they have begun to spread out along the nerves. Although the ztv/.sz-expressing cells are now distributed over much of the length of the nerves associated with the remaining larval muscles (but not in the region between the ganglion and the first site of contact with the muscles), there are still four principal concentrations of twist -expressing cells in each hemi-segment. Of the previous six groups of rwAz-expressing cells in the third instar larva, the three dorsal groups have coalesced to form an apparently continuous cluster, while the ventral and lateral groups remain distinct, though with an increased number of cells in each.

By 20 h APF the two lateral clusters, attached to the nerves running to muscle no.8 at both ends of the segment, have spread along their associated nerves and formed a single group spanning the entire segment (Fig. 6). The other groups of ovist-expressing cells are also still attached to the nerves, most of them close to the growth cones al the sites of attachment to the epidermis (Fig. 6). Some twist cells can clearly be seen associated with nerves and growth cones, which lie over larval epidermal cells that have not yet been replaced by migrating adult histoblasts.

Fig. 6.

(A) anti-HRP staining of pupa 23 h APF showing 2 ventral points of contact with the epidermis (arrows). h, histoblast cells; I, larval epidermal cells; n, nerve. Scale bar 20μm. (B) twist-ex pressing cells (arrowheads) at a similar point of contact in a pupa 25 h APF. The twistexpressing cells are concentrated close to the growth cones at the site of contact with the epidermis, n, nerve; h, histoblast cells; I, larval-epidermal cells. Scale bar 20 μm. (C) Groups of twist-ex pressing cells in four adjacent segments of a pupa 25 h APF. Boxed region shown in Fig. 6B. Note the three main groups of twist -staining cells; v, ventral group; I, lateral group; d, dorsal group. Scale bar 50 μm.

Fig. 6.

(A) anti-HRP staining of pupa 23 h APF showing 2 ventral points of contact with the epidermis (arrows). h, histoblast cells; I, larval epidermal cells; n, nerve. Scale bar 20μm. (B) twist-ex pressing cells (arrowheads) at a similar point of contact in a pupa 25 h APF. The twistexpressing cells are concentrated close to the growth cones at the site of contact with the epidermis, n, nerve; h, histoblast cells; I, larval-epidermal cells. Scale bar 20 μm. (C) Groups of twist-ex pressing cells in four adjacent segments of a pupa 25 h APF. Boxed region shown in Fig. 6B. Note the three main groups of twist -staining cells; v, ventral group; I, lateral group; d, dorsal group. Scale bar 50 μm.

Fig. 7.

Section of a pupal abdomen 26 h APF viewed with phase contrast showing the nuclei of the fnvjz-expressing cells (arrows) that overlie the histoblast cells. This section was very lightly stained with methylene blue and so the nucleoli (n) of some of the histoblast cells have stained, h, histoblast cells; c, cuticle; Im, persistent larval muscle. Scale bar 10 μm.

Fig. 7.

Section of a pupal abdomen 26 h APF viewed with phase contrast showing the nuclei of the fnvjz-expressing cells (arrows) that overlie the histoblast cells. This section was very lightly stained with methylene blue and so the nucleoli (n) of some of the histoblast cells have stained, h, histoblast cells; c, cuticle; Im, persistent larval muscle. Scale bar 10 μm.

As the histoblasts and nerve growth cones migrate outwards, the twist-expressing ceils migrate with them. By 25 h APF. the course of this migration is clear. The ventral group migrates ventromedially with its nerve and histoblast cells and will give rise to the ventral muscles. The dorsal group migrates dorsomediahy with and along the nerve bundle on the epidermis and will give rise to the dorsal muscles. Sections through the histoblasts at this stage show that the twist-expressing cells only occur on the inner surface of the histoblast nest and not within the nest itself (Fig, 7). At this stage, the twist-expressing cells that will form the adult spiracular muscle, can also be seen, apparently derived from the lateral cell clusters. The association with nerves is clearest in anti- twist/anti-HRP double-stained preparations (Fig. 8). twist-ex pressing cells occur along the length of the nerves with high concentrations of cells in the developing dorsal, lateral and ventral regions of each hemisegment. At 31 h, the celts expressing twist are organising to form (he adult muscles. This is particularly striking in the lateral region where the cells are already aligned in characteristic parallel rows (Figs 9A,B).

Fig. 8.

Anti-HRP, anti- twist double-stained preparation of a pupa 31 h APF, showing close association of twist-ex pressing cells with the developing dorsal innervation, n, nerve; tw, nuclei staining positive for twist expression. Scale bar 10 μm.

Fig. 8.

Anti-HRP, anti- twist double-stained preparation of a pupa 31 h APF, showing close association of twist-ex pressing cells with the developing dorsal innervation, n, nerve; tw, nuclei staining positive for twist expression. Scale bar 10 μm.

Fig. 9.

(A) Anti-win stained preparation of a pupa 31 h APF to show the beginnings of parallel organisation among twistexpressing cells in the lateral region of the abdomen. Scale bar 20, μm. (B) β3 tubulin expression in a pupa 41 h APF pupa showing the characteristic parallel alignment of the developing lateral muscles. Scale bar 20 μm.

Fig. 9.

(A) Anti-win stained preparation of a pupa 31 h APF to show the beginnings of parallel organisation among twistexpressing cells in the lateral region of the abdomen. Scale bar 20, μm. (B) β3 tubulin expression in a pupa 41 h APF pupa showing the characteristic parallel alignment of the developing lateral muscles. Scale bar 20 μm.

β 3 tubulin expression is turned on in muscle precursors

Early in embryogenesis all mesodermal cells express twist (Thisse et al. 1988). In embryonic muscle precursors, twist expression declines and is followed by the synthesis of β3 tubulin. β3 tubulin expression continues until the muscles have been formed (Leiss et al. 1988). Staining with the β3 tubulin antibody in the pupa shows that by 26 h APF some cells organising to form the adult muscles have already started to express the product of the β3 tubulin gene. We find these cells on top of the histoblasts in the same position as the cells revealed by anti-twist staining. Double staining for twist (nuclear), and β3 tubulin (cytoplasmic), clearly shows that it is the twist-expressing cells that begin to express the β3 tubulin gene. These are the cells that will form the adult muscles. Fig. 10 is a 31 h twist/β3-tubulin stained preparation. The myoblasts that are stained for twist will go on to express β3 tubulin and form a single group of adult ventral muscle fibres. The nuclei of most of the cells that will fuse to form these muscle fibres are still expressing twist (tw) at this stage, although this is not the case for the celts at the most posterior end of the assembling muscle pritnordium where β3 tubulin expression can be seen (arrowed). It is important to emphasise that, although we know that β3 tubulin is expressed after twist in the cells that will form the adult muscles, we do not know whether there is a temporal overlap in the expression of twist and β3 tubulin in the same cell. Within the group of cells that are to form one set of ventral muscle fibres, there is an overlap of expression in that anterior cells clearly express twist while posteriorly twist is no longer being expresssed and these cells have already begun to make β3 tubulin (Fig. 10). In the embryo, twist expression does not switch off in all cells at the same time (Thisse et al. 1988). Indeed, there is still a considerable number of twist-expressing cells present at the end of germ band shortening (Bate et al. 1991) by which time β3 tubulin expression has already begun (Leiss et al. 1988) and al! muscle precursors are present with cells fusing to them (Bate, 1990).

Fig. 10.

Anti-β3 tubulin. anti- twist double stained preparation 31 h APF showing a group of myoblasts that will fuse to form ventral muscle fibres. The nuclei of myoblasts in the anterior portion of the group are still expressing twist while posteriorly ceils express β3 tubulin (arrow), a. anterior; p. posterior; tw. twist expressing nuclei. Scale bar 10 μm.

Fig. 10.

Anti-β3 tubulin. anti- twist double stained preparation 31 h APF showing a group of myoblasts that will fuse to form ventral muscle fibres. The nuclei of myoblasts in the anterior portion of the group are still expressing twist while posteriorly ceils express β3 tubulin (arrow), a. anterior; p. posterior; tw. twist expressing nuclei. Scale bar 10 μm.

The formation of adult muscles

Beginning at 26h APF β3 tubulin is expressed by muscle precursor cells and is a useful marker as the cells begin to assemble to form the adult muscles, initially the cells seem to be randomly oriented, but by 28 h APF they have begun to align themselves in the appropriate direction for the adult muscles (Fig. 11A). The assembling dorsal muscle cells now produce long anterior and posterior projections as they begin to locate their correct level in the segment for attachment to the epidermis. Fusion of the cells first becomes clear at 28 h APF. The myoblasts then form small groups which separate as the anlagen of individual muscle fibres over a period of a few hours (Fig. 11B, C) and align parallel to one another. At the ends of the muscles there are numerous fine projections al the points of contact with the epidermis (Fig. 11C), similar to those seen in developing muscles in the embryo (Bate. 1990) and in the embryonic grasshopper leg (Ball and Goodman. 1985). Laterally the β3 tubulin-expressing cells orient themselves in a dorsoventrai direction to form the lateral muscles and a small group in each hemisegment aligns anteroposteriorly to form the ventral muscles. It is interesting that the aggregating muscle cells begin to orient themselves on the adult epidermal cells while these cells are still actively migrating to replace the larval epidermis.

Fig. 11.

(A) tubulin expression in the dorsal myoblasts 28 h APF. The myoblasts are clearly beginning to orient in an anteroposterior direction. A number of cells send out fine processes as they begin to migrate to their attachment sites, a, anterior; p, posterior. Scale bar 20 μm. (B) β3 tubulin expression at 36 h APF showing the precursors of dorsal muscle fibres. Scale bar 20 μm. (C) Dorsal muscle fibres in a pupa 41 h APF. Note the more developed posterior attachment site and fine filopodia! extensions at the anterior ends of the fibres (inset), a, anterior; p, posterior. Scale bar 20μm.

Fig. 11.

(A) tubulin expression in the dorsal myoblasts 28 h APF. The myoblasts are clearly beginning to orient in an anteroposterior direction. A number of cells send out fine processes as they begin to migrate to their attachment sites, a, anterior; p, posterior. Scale bar 20 μm. (B) β3 tubulin expression at 36 h APF showing the precursors of dorsal muscle fibres. Scale bar 20 μm. (C) Dorsal muscle fibres in a pupa 41 h APF. Note the more developed posterior attachment site and fine filopodia! extensions at the anterior ends of the fibres (inset), a, anterior; p, posterior. Scale bar 20μm.

At 41 h the pattern of the adult muscles has been largely completed (Fig. 12A.B). The newly forming dorsal muscles are already developing a posterior attachment to the epidermis. Anteriorly, however, the same cells have numerous long, fine filopodia, which suggests that the developing muscles are still extending forwards. The same is true of the ventral muscles: posteriorly they have a broad region which will develop into an attachment site but at their anterior ends the cells focus to a narrow point from which fine filopodia project (Fig. 12B). At this stage there are still a few j53 tubul in-expressing ceils that have not yet fused with developing muscles and are associated with muscle groups or nerves. By 65 h the cells have completed fusion, twist expression has disappeared, and the final pattern of adult muscles has been established.

Fig. 12.

(A) dorsal and (B) ventral muscle pattern in a pupa 41 h APF revealed with the anti-β3 tubulin antibody, v, ventral muscles; 1, lateral muscles. Scale bars 50μm.

Fig. 12.

(A) dorsal and (B) ventral muscle pattern in a pupa 41 h APF revealed with the anti-β3 tubulin antibody, v, ventral muscles; 1, lateral muscles. Scale bars 50μm.

The origin of the adult myoblasts

In the developing embryo, the cells of the mesoderm express twist at the blastoderm stage prior to their invagination at gastrulation (Thisse et al. 1988). In the pupa, there is no event comparable to gastrulation, which pinpoints and defines the origin of the mesoderm in the embryo, and until now it has been impossible to identify the source of the adult myoblasts. It is only by looking in the pupa for the products of genes, such as twist, which are expressed in the embryonic myoblasts, that we have been able to identify with certainty the adult muscle-forming cells.

The adult abdominal muscles are formed from a population of twist-expressing cells that is already present in the third instar larva. These cells are set aside in the embryo and proliferate in larval life (Bate et al. 1991). It is significant that none of the cells is ever directly associated with the histoblasts until pupation. Until then they remain tightly linked to the larva! nerves. It has been suggested that the adult myoblasts are derived from the histoblasts (Madhavan and Schneiderman, 1977) or are at least closely associated with them (Robertson, 1936). Our observations show that this is not the case and are in agreement with previous work, which indicates that the adult abdominal musculature has an embryonic origin that is mesodermal and separate from the epidermis (Lawrence and Johnston, 1982; 1986a). This is further supported by the evidence from sections of histoblasts in the pupa, which show that twin-expressing cells are only found internal to the histoblast nests and not within them. Our results show that the myoblasts remain closely associated with nerves as adult development begins, and that during adult development the proliferation and movement of the myoblasts is channeled along these nerves. Thus the nerves may be an important factor in determining myoblast distribution and in dictating the separation of myoblasts into groups, each of which gives rise to a different muscle set. This idea is investigated further in the accompanying paper (Broadie and Bate, 1991).

Mechanisms of muscle formation

There are a number of significant similarities between adult and embryonic muscle formation. In both cases, the muscle-forming cells are of mesodermal origin and first express twist (Thisse et al. 1988) and subsequently pi tubulin (Leiss et al. 1988). In the Drosophila embryo (Bate, 1990) and the pupal abdomen, the muscles form from a large population of cells present as a layer that overlies ectodermal tissue (sec Fig. 11). In both cases cells within this layer go through a process of orientation and fusion to form muitinucleate groups. They thereby segregate from each other and sub- sequently go on to form the final muscle pattern.

What controls the correct spatial patterning of individual developing muscles? In the grasshopper embryo there is a class of mesodermal cell, the muscle pioneers, which develops from the mass of mesodermal cells in every segment. These ceils grow so that they span the position that the fully developed muscle will occupy (Ho et al. 1983). They then seed the formation of individual muscles by acting as a scaffold with which other mesodermal cells fuse (Ball et al. 1985). Although this study has not revealed a similar class of large mesodermal cells in the pupa, it remains possible that the individual adult muscles are seeded by ceils which would be analogous to the founder cells suggested to exist in the Drosophila embryo (Bate, 1990). It is not clear from this study whether founder cells like those postulated in the embryo exist for the developing adult muscles. If such founder ceils exist in the pupa, then, as in the embryo, they could not perform exactly the same role as the muscle pioneers in the grasshopper as they would be unlikely to be able to span the relatively huge distances required before and during fusion. This is further emphasised by the elongation of the muscle precursors in the pupa which continues after fusion has begun. The long extensions from the anterior and posterior edges of the aggregating myoblasts (Fig. 11 A) suggest that there are muitinucleate precursors which are migrating at both their anterior and posterior ends to span the region that the completed muscle will cover. Whether such precursors were seeded by an individual founder is unclear.

The role of the epidermis in establishing muscle pattern

The developing muscles migrate large distances in order to span their final territory and align themselves in the proper orientation. They could be following cues laid down by the developing nerves and it has been proposed that myoblasts themselves have some selforganizing capability (Hooper, 1986). However, it is also possible that they may be getting information about polarity and position from the epidermis as they migrate out in close contact with it. It seems unlikely that muscle insertion points are already specified by the epidermis since the muscle precursors are beginning to organise themselves on the adult epidermis by 28 h APF, well before the migrating histoblasts have replaced all the larval cells in each segment. In the case of the ventral muscles, one consequence of this is that, while myoblasts are still fusing and aligning, they must also migrate a considerable distance ventrolaterally in order to reach their final position near the ventral midline. This movement could be a passive one as they are dragged along by the migrating histoblast cells, but it could also be that they migrate actively across the developing epidermis, as in Tenebrio (Williams and Caveney, 1980), and that some interaction between them and the epidermis defines their final attachment site. The lateral nw.vi-expressing cells also begin to align very early, and they do so in the absence of any apparent alignment of the nerves. They clearly have access to some information about polarity and the most likely source is the epidermis which they directly overlie. In the light of this, it is also interesting to note that, as the histoblast cells migrate out to replace the dying larval cells, the integrity of the band of cells expressing engrailed in the posterior compartment of each segment is maintained right across the border between larval and histoblast epidermis (Currie unpublished observation). It is possible that other positional information is also being maintained as the new epidermis develops.

Nerves and growth cones

The migration of nerves from their points of contact with the epidermis across the surface of the developing histoblasts has not previously been described. The fact that the nerves, presumably the axons of motor neurons, begin to migrate long before axons have been produced by sensory neurons is unusual. In the embryo, sensory and motor axons begin growing at about the same time (Johansen et al. 1989). However, these pupal ‘growth cones’ have not unequivocally been identified as such. They could (at least initially) reflect the outward migration of glial cells and therefore we suggest that, for the moment, they should be thought of as growth zones, rather than clearly identified growth cones. From the present study, it is not possible to state how much their outward growth may be assisted by the migration of the developing histoblasts. However, it is these migrating growth zones that pioneer the path for the adult nerves and their associated myoblasts. The numerous filopodial projections that can be seen later at the points of nerve growth appear very similar to the growth cones seen in embryonic Drosophila and locust (Johansen et al. 1989, Ball et al. 1985).

Role of nerves in adult muscle formation

Lawrence and Johnston (19866) have demonstrated that, at least for the male-specific muscle in A5, the genotype of its innervation is decisive in determining the development of this element of the muscle pattern. The close association of twist cells with nerves rather than with histoblast nests in the 3rd instar larva and with migrating nerves in the pupa is an unexpected result and once again suggests that nerves may play a crucial part in the development of the adult musculature. This is in contrast to the situation in the embryo where muscle formation has begun, and the pattern is complete, well before the outgrowth or ingrowth of motor and sensory axons (Bate, 1990; Johansen et al. 1989). Lawrence and Johnston (19866) have shown that nerves play a decisive role in the development of one aspect of the abdominal muscle pattern. What is not clear is whether the malespecific muscle that they studied is a special case, or an example of a general requirement for nerves in adult muscle patterning. The tight interaction between the developing muscles and nerves in the pupa would certainly allow for such a requirement.

It may well be that nerves have a simple mechanical function in adult myogenesis, in that they provide a substratum for the proliferation and distribution of the adult myoblasts. In embryos, the mesodermal cells in vaginate at gastrulation, migrate over the inner surface of the ectoderm and so come to lie as a sheet in the region where muscles will form. In the pupa, there is no comparable event and the arrangement of the adult muscle-forming cells has its origins in an earlier embryonic distribution. Thus myoblasts that will form the adult thoracic muscles come to be associated with the imaginai discs early in embryonic development (Bate et al. 1991), but in the abdomen the myoblasts are not associated with the precursors of the adult epidermis, the histoblasts. The abdominal myoblasts arrive at the appropriate regions of the developing abdominal epidermis by their association with and migration along the nerves, which themselves develop in close association with the epidermis.

In addition, the nerves may have an active role in establishing the pattern of muscles. The lateral projections from the developing nerves do occur in regions where there are myoblast extensions from the cell mass. It is possible that there is a decisive interaction going on between nerves and muscles at these points. Such interactions have been documented in the grasshopper embryo (Ball et al. 1985) although in that case the nerves are following cues provided by the muscle pioneers rather than the reverse, which we would expect if the nerves were influencing the muscle pattern.

The association with nerves described here may also help to explain the results of previous clonal analyses of adult muscle development, which show that the muscle precursors in the thorax are divided into specific sets, in the case of the mesothorax, dorsal and ventral sets, which develop independently (Lawrence, 1982). Extrapolating from our observations on the abdominal muscles, it could be that the segregation of muscle lineages in the thorax is achieved initially by the restriction of precursors to specific discs and after évagination by an association with nerves specific to particular discs. Lawrence (1982) suggested that one mechanism for maintaining muscle lineages would be by the maintenance of mutual contact between mesodermal cells throughout metamorphosis. The association with nerves could function to maintain such a mutual contact and would have the effect of restricting the ability of cells to migrate between segments. We are unable to explain the observations of Lawrence and Johnston (1982) that occasional clones in the ventral abdominal muscles cross between segments. These authors have also provided evidence that points to a small number of free floating myoblasts occurring in the thorax (Lawrence and Johnston, 1986a). Given the association of the ívrái-expressing cells with nerves seen in this study it seems unlikely that there are many free floating adepithelial cells in the pupa but the possibility cannot be ruled out. What we have shown here is that in the normal development of the abdomen, the distribution of the adult myoblasts is closely linked to the growth and reorganisation of the peripheral nervous system. We have not seen any obvious restriction in the movement of myoblasts along nerves. Nonetheless, dorsal, ventral and lateral cells remain separately clustered. It is possible that occasional cells may pass between clusters, but this would be a relatively rare event. We now hope to show, by experiments in progress, whether the presence of nerves is an essential requirement for the formation of the muscle pattern.

We thank Helen Skaer and David Shepherd for their critical comments on earlier versions of this manuscript and also Alfonso Martinez-Arias, Emma Rushton, Kendal Broa- die and Dagmar Leiss for many stimulating discussions. We are grateful to F. Perrin-Schmitt and D. Leiss for providing antibodies. This work was supported by an SERC studentship to D.A.C. and a grant from the Wellcome Trust to M.B.

Ball
,
E. E.
and
Goodman
,
C. S.
(
1985
).
Muscle development in the grasshopper embryo II. Syncitial Origin of the extensor tibiae muscle pioneers
.
Devl Biol
.
111
,
399
416
.
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.
,
Rushton
,
E.
and
Currie
,
D. A.
(
1991
).
Cells with persistent twist expression are the embryonic precursors of adult muscles in Drosophila
.
Development
113
,
79
89
.
Broadie
,
K. S.
and
Bate
,
M.
(
1991
).
Development of adult muscles in Drosophila: Ablation of identified muscle precursor cells
.
Development
113
,
103
118
.
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
.
Deak
,
I.
(
1977
).
Mutations of Drosophila melanogaster that affect muscles
.
J. Embryol. exp. Morph
.
40
,
35
63
.
Ho
,
R. K.
,
Ball
,
E. E.
and
Goodman
,
C. S.
(
1983
).
Muscle Pioneers: Large mesodermal cells that erect a scaffold for developing muscles and motoneurons in grasshopper embryos
.
Nature
301
,
66
69
.
Hooper
,
J. E.
(
1986
).
Homeotic gene function in the muscles of Drosophila larvae
.
EMBO J
.
5
,
2321
2329
.
Hotta
,
Y.
and
Benzer
,
S.
(
1972
).
Mapping of behaviour in Drosophila mosaics
.
Nature
240
,
527
535
.
Jan
,
L. Y.
and
Jan
,
Y. N.
(
1982
).
Antibodies to horseradish peroxidase as specific neuronal markers in Drosophila and in grasshopper embryos
.
Proc. natn. Acad. Sci. U.S.A
.
79
,
2700
2704
.
Johansen
,
J.
,
Halpern
,
M. E.
and
Keshishian
,
H.
(
1989
).
Axonal guidance and development of muscle fibre-specific innervation in Drosophila embryos
.
J. Neurosci
.
9
(
12
),
4318
4332
.
Kimura
,
K.
and
Truman
,
J. W.
(
1990
).
Postmetamorphic cell death in the nervous and muscular systems of Drosophila melanogaster
.
J. Neurosci
.
10
(
2
),
403
411
.
Lawrence
,
P. A.
(
1982
).
Cell lineage of the thoracic muscles of Drosophila
.
Cell
29
,
493
503
.
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
.
Lawrence
,
P. A.
and
Johnston
,
P.
(
1986a
).
Observations on cell lineage of the internal organs of Drosophila
.
J. Embryol. exp. Morph
.
91
,
251
266
.
Lawrence
,
P. A.
and
Johnston
,
P.
(
1986b
).
The muscle pattern of a segment of Drosophila may be determined by neurons and not by contributing muscles
.
Cell
45
,
505
513
.
Leiss
,
D.
,
Hinx
,
U.
,
Gasch
,
A.
,
Mertz
,
R.
and
Renkawitz-Pohl
,
R.
(
1988
).
β3 tubulin expression characterizes the differentiating mesodermal germ layer during Drosophila embryogenesis
.
Development
104
,
525
531
.
Lowne
,
B. T.
(
1892
).
In The Anatomy, Physiology, Morphology and Development of the Blow-Fly
. vol
1
(ed.
R. H.
Porter
), pp.
333
338
.
London
.
Madhavan
,
M. M.
and
Madhavan
,
K.
(
1980
).
Morphogenesis of adult abdomen of Drosophila
.
J. Embryol. exp. Morph
.
60
,
1
31
.
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’s. Arch, devl Biol
.
183
,
269
305
.
Miller
,
A.
(
1950
).
The internal anatomy and histology of the imago of Drosophila melanogaster
.
In Biology of Drosophila
(ed.
M.
Demerec
), pp.
420
534
.
New York
:
Wiley
.
Pearson
,
M. J.
(
1972
).
Imaginai discs and the abdominal histoblasts of Caliphora erythrocephala
.
Nature
238
,
349
351
.
Perez
,
C.
(
1910
).
Recherches histologiques sur la metamorphose des muscides
.
Arch. Zool. Exp. Gen. 5es
.
4
,
1
275
.
Reed
,
C. T.
,
Murphey
,
C.
and
Fristrom
,
D.
(
1975
).
The ultrastructure of the differentiating pupal leg of Drosophila melanogaster
.
Wilhelm. Roux’s. Arch, devl Biol
.
179
,
285
302
.
Robertson
,
C. W.
(
1936
).
The metamorphosis of Drosophila melanogaster, including an accurately timed account of the principal morphological changes
.
J. Morph
.
59
,
351
399
.
Roseland
,
C. R.
and
Reinhardt
,
C.
(
1982
).
Abdominal development
.
In Handbook of Drosophila Development
, (ed.
R.
Ranson
.), pp.
215
235
.
Elsevier Biomedical Press
.
Roseland
,
C. R.
and
Schneiderman
,
H. A.
(
1979
).
Regulation and metamorphosis of the abdominal histoblasts of Drosophila melanogaster
.
Wilhelm Roux’s. Arch, devl Biol
.
186
,
235
265
.
Thisse
,
B.
,
Stoetzel
,
C.
,
Gorostiza-Thisse
,
C.
and
Perrin-Schmitt
,
F.
(
1988
).
Sequence of the twist gene and nuclear location of its protein in endomesodermal cells of early Drosophila embryos
.
EMBO J
.
7
,
2175
2183
.
Ursprung
,
H.
,
Conscience-Egli
,
M.
,
Fox
,
D. J.
and
Walliman
,
T.
(
1972
).
Origin of leg musculature during Drosophila metamorphosis
.
Proc. natn. Acad. Sci. U.S.A
.
69
,
2812
2813
.
Williams
,
G. J. A.
and
Caveney
,
S.
(
1980
).
Changing muscle patterns in a segmental epidermal field
.
J. Embryol. exp. Morph
.
58
,
13
33
.