Twist and Notch negatively regulate adult muscle differentiation in Drosophila

In Drosophila, the mesoderm is specified by genes that allot fates along the dorsal-ventral axis and results in activation of twist (twi) (Thisse et al., 1987a,b, 1988). twi expression is retained in adult muscle progenitors set aside during embryogenesis but declines in embryonic muscle progenitors as they fuse to form muscle (Bate, 1990, 1991). twi-expressing adult myoblasts proliferate in the larva and contribute to adult muscles during metamorphosis and as in the embryo, twi expression is shut-off as differentiation begins (Currie and Bate, 1991; Fernandes et al., 1991).

twi down-regulation during differentiation suggests that its function could be required to maintain myoblasts in an undifferentiated state. In a systematic analysis, Baylies and Bate (1996) showed that this hypothesis is not valid in the embryo. Persistent twi expression, at high levels using the GAL4/UAS system (Brand and Perrimon, 1994) had no effect on the development of somatic muscles. Baylies and Bate (1996) also showed that Twi behaved like a ‘myogenic switch’. twi mis-expression in the ectoderm caused these cells to express muscle marker genes and, in the mesoderm, Twi levels help determine cell-fate choices between somatic and visceral mesoderm.

Vertebrate twi expression and function show similarities with Drosophila but there are significant differences. During murine myogenesis, twi expression is initially observed all over the epithelial somite (Wolf et al., 1991; Stoetzel et al., 1995; Fuchtbauer et al., 1995). Subsequently, Twi is excluded from the myotome upon initiation of skeletal myogenesis. Expression of twi in myogenic cell lines prevents the onset of muscle differentiation (Hebrok et al.,1994), a result different from that seen in the Drosophila embryo (Baylies and Bate, 1996).

We have examined the regulation and function of twi during development of the adult flight muscles. We show that decline in Twi levels coincides with the onset of expression, in indirect flight muscle (IFMs) progenitors, of the erect-wing (ewg) encoded transcription factor (DeSimone and White, 1993) and that twi regulation involves Notch signalling. Both reduction and constitutive activation of Notch affect the development of the IFMs but not the closely related direct flight muscles (DFMs). These effects on differentiation are very similar to that seen in hypomorphic-conditional-twi mutant flies and in animals with persistent twi expression respectively. High levels of Twi in adult myoblasts during larval life does not affect larval development, but persistent high levels of Twi in these cells during pupal development affects IFMs, but not DFMs. Myosin Heavy Chain (MHC) levels are reduced and the muscles degenerate. Our results suggest a regulatory mechanism for twi expression in myogenesis and demonstrate that persistent twi expression can, as in mouse, inhibit muscle differentiation in adult Drosophila.

Canton-S was used as wild-type. N^ts1(Shellenbarger and Mohler, 1975, 1978) is a conditional Notch allele. N^ts1 animals were grown at the permissive temperature, 22°C, and shifted to the non-permissive temperature, 31.5°C, for varying intervals during pupation. twi^ry50 and twi^v50 are homozygous lethal, viable in trans combination at 18°C but show a twi-phenotype when grown above 29°C (Thisse et al., 1987b). The MHC-lacZ (Hess et al., 1989) reporter gene is expressed in all muscles and twi-lacZ (Thisse et al., 1991), in all myoblasts. An X-chromosome UAS-twi transgenic strain was a gift from M. K. Baylies and M. Bate (Cambridge University, UK). The UAS-twi transgene in this strain was mobilised and for the experiments mentioned here, a second chromosome insert was used. The UAS-activated Notch (UAS-N-intra) line was a gift from Masahiro Go and S. Artavanis-Tsakonas (Yale, USA). N-intra is a truncated form of the Notch receptor and is constitutively active (Rebay et al., 1993). The 1151 GAL4 driver was obtained from L. S. Shashidhara (CCMB, India). In the third larval instar, expression is seen in myoblasts on the wing and leg imaginal discs (Fig. 4A), and in nerve associated myoblasts (Fig. 4B). The adepithelial cells on the wing imaginal disc give rise to the dorsal mesothoracic muscles, the IFMs and the DFMs (Lawrence, 1982). The 1151 driver continues to be expressed in the pupa as myoblasts fuse to form muscles (Fig. 4C) and also in the adult IFMs and DFMs. There is no expression seen in larval muscles (Fig. 4B) that function as templates for DLM formation though a low transient expression is seen in almost all embryonic muscles (not shown). This does not appear to have any effects on embryonic muscle development when activated-Notch, twi or several other control UAS constructs such as UAS-Ubx, UAS-Scr, UAS-Antp and UAS-p21 (a gift from I. Hariharan, MGH Cancer Centre, USA) are expressed with the 1151 driver (Roy and VijayRaghavan, 1997, Roy and VijayRaghavan, unpublished observations). These latter UAS-constructs cause lethality when expressed ubiquitously in the embryonic mesoderm or muscle but when expressed with the 1151 driver allow survival to pupal or adult stages and the examination of adult muscle development (Roy and VijayRaghavan, 1997, Roy and VijayRaghavan, unpublished data).

White prepupae (0 hours APF) at 22°C were collected on moist filter paper in a Petri-dish and transferred to a 31.5°C incubator for different intervals. To correlate developmental stages after heat shocks with that at 25°C, MHC-lacZ pupae were grown either entirely at 22°C or at 31.5°C for different intervals and dissected. IFM development was compared to that at 25°C. We found that the rate of development at 22°C was 0.75 times that at 25°C and that at 31.5°C approximately 1.3 times that at 25°C.

Staged, heat-shocked pupae were dissected in phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde in PBS and later stained with X-GAL (Fernandes et al., 1991) or with appropriate antibodies. Flies were cut sagitally to view IFMs, dehydrated in an alcohol series, cleared in methyl salicylate, mounted in Canada balsam and viewed under polarized light. To view the DFMs, the DLMs and in some cases the DVMs were dissected out for greater clarity. Pupal preparations were mounted in 70% glycerol or in GelTol (Immunon, USA).

Anti-Twi and mAb22C10 stainings were done according to the protocols described by Fernandes and VijayRaghavan (1991). Twi antibody was a gift from Siegfried Roth (Max Planck Institute, Tubingen, Germany) and mAb22C10 from Seymour Benzer (Caltech, USA). Anti-Ewg (a gift from Kalpana White, Brandeis University, USA) staining was done as described by DeSimone et al. (1995). For confocal microscopy, anti-rabbit secondary was conjugated to rhodamine to mark anti-Ewg or anti-Twi primary antibodies. Anti-mouse secondary conjugated to FITC was used against anti-β-galactosidase. Confocal analysis was done on a Bio-Rad Model 600 confocal microscope.

The development, innervation, differentiation and segmental properties of the IFMs have been well investigated (Fernandes et al., 1991; Fernandes and VijayRaghavan, 1993; Fernandes et al., 1996; Roy et al., 1997; Roy and VijayRaghavan, 1997). Early during pupal development three dorsal muscles in each mesothoracic hemisegment escape histolysis and serve as templates for the formation of one group of IFMs, the dorsal longitudinal muscles (DLMs; El Shatoury, 1956; Costello and Wyman, 1986; Fernandes et al., 1991)(Fig. 1A-C). Another group of IFMs, the dorsoventral muscles (DVMs) are also derived from progenitors on the wing imaginal disc. However, the formation of these muscles takes place by the de novo fusion of myoblasts at appropriate epidermal sites without the use of larval templates (Fig. 1D). Apart from the IFMs, there are other muscles, the DFMs (Fig. 1E), also derived from wing-imaginal disc-located progenitors (Lawrence, 1982). Although the IFMs and DFMs are clonally related and share progenitors at least till the late third larval instar (Lawrence 1982), they differentiate into very different muscle types in terms of molecular markers, anatomy and physiology (Bate, 1993).

We have shown earlier that ewg is first expressed in IFM development as myoblasts begin to fuse to form these muscles (DeSimone et al., 1995). We now demonstrate that ewg expression is seen in IFM myoblasts just prior to their fusion at a time when twi expression is declining. Fig. 2 shows the developing DLMs at 16 hours APF. Twi expression (green) is seen in myoblasts present over the larval templates. The developing DLMs are visualised by the presence of Ewg-positive nuclei, aligned in rows, inside the fibre (red). Outside the fibre, myoblasts overlying the templates have started expressing Ewg while Twi expression has declined (red spots). Few myoblasts express both Twi and Ewg (yellow) and myoblasts still at a distance from the developing DLMs express only Twi (green). The timing of ewg expression and mutant phenotypes suggest that ewg acts after myoblast proliferation at the earliest stages of differentiation (DeSimone et al., 1995).

The function of the Notch locus has been demonstrated in several developmental pathways in flies and other animals (For review see Artavanis-Tsakonas, 1995). We examined the role of Notch in adult myogenesis by using a conditional allele, N^ts1. In wild-type muscles, expression of a twi-lacZ reporter gene is seen in differentiating IFMs although twi expression is itself absent in multinucleate cells. This expression of β-galactosidase in the developing fibre is due to the ‘perdurance’ of the enzyme synthesized in myoblasts prior to fusion. When N^ts1 animals are shifted to non-permissive temperature during early pupal development β-galactosidase activity staining resulting from twi-lacZ reporter gene expression, is greatly reduced (Fig. 3A,C) in the developing DLMs. This result suggested the possibility that Notch function could be required for the maintenance, directly or indirectly, of twi expression till early during metamorphosis. Alternatively, the lowering of twi-lacZ expression could merely be a consequence of the possible failure of several myoblasts to fuse to the developing DLMs in N^ts1 animals that were shifted to non-permissive temperatures during pupation. To examine the relationship between Notch function and twi expression we used antibodies specific to Twi to examine myoblasts in Notch mutant animals. As can be seen by Twi antibody staining on N^ts1 animals (Fig. 3B,D) pulsed between 10–19 hours APF, Twi immunoreactivity is drastically lowered compared to wild-type animals at the corresponding developmental stage.

The lowering of twi expression upon reduction of Notch function led us to ask if constitutive activation of Notch can cause elevation of Twi levels. We used the GAL4-UAS system (Brand and Perrimon, 1994) for this purpose. The 1151 driver described earlier (Roy and VijayRaghavan, 1997) was used to drive the expression of UAS-N-intra in disc associated myoblasts throughout larval and pupal development (Fig. 4). The 1151 driver is not detectably expressed in the thoracic epidermis or its progenitor cells on the wing disc (Fig. 4A; Roy and VijayRaghavan, 1997). In wild-type pupae, twi expression is not seen within the muscle (Fernandes et al., 1991) and this is documented again in optical sections (Fig. 5A-C). Expression of N-intra in developing adult thoracic muscles using 1151, however, results in Twi expression continuing within the DLMs (Fig. 5D). Thus, expression of activated Notch in adult myoblasts does not eliminate fusion but Twi fails to be down-regulated. The phenotypes of Notch mutant animals in both loss- and gain-of-function situations is described below.

To examine the role of Notch in flight muscle development more closely, animals carrying the temperature sensitive allele, N^ts1, were shifted to the non-permissive temperature between 0-18 hours APF. Pupae dissected soon after, showed that the process of splitting of the three larval templates for the DLMs had not begun (Fig. 6A). Depletion of wing disc myoblasts prevents splitting and differentiation of the larval templates (Roy and VijayRaghavan, unpublished data) suggesting that myoblast fusion is required for DLM formation. However, some myoblast fusion must occur in Notch mutant animals because the DLMs differentiate as three ‘un-split’ fibres (Fig. 6B). The DVMs in such pupae are sometimes mis-aligned and attach to each other or to the DLMs (not shown). Several tests were done to establish the minimum time window required to observe an effect of Notch reduction on IFM development. We find that IFMs visualized at 24 hours APF after a brief pulse between 9-16 hours APF at the non-permissive temperature showed that the larval templates do not split (Fig 6B). When N^ts1 pupae, were grown at permissive temperature after the above pulse, the adults that emerged had three DLMs and not six (not shown).

N^ts1 pupae pulsed between 9 –18 hours APF were examined at 24 hours APF with the monoclonal antibody mAb 22C10 which marks motor innervation and also developing adult muscle (Fujita et al., 1982; Fernandes and VijayRaghavan, 1993). A large number of spindle-shaped cells (average=45, n=14 hemithoraces) over the 3 unsplit DLMs were seen in a N^ts1 pupa (Fig. 6D). Such cells are very rarely seen in heat shocked control animals (average=2, n=8 hemithoraces, an example is shown in Fig. 6C). The extended processes of these spindle-shaped cells suggest that they could be prematurely differentiating myoblasts. Adult N^ts1 mutant animals raised to non-permissive temperature during pupal development show normal DFMs.

We used 1151 to drive UAS-N-intra in the developing flight muscles. 1151 (Materials and methods, also Roy and VijayRaghavan, 1997) expresses in disc-associated myoblasts and expression continues through pupal development in all the muscles derived from these myoblasts. When activated Notch is expressed under the control of this GAL4 driver, both the DLMs and DVMs are missing (Fig. 7B). To determine in more detail the effect, upon differentiation, of activated Notch we examined the expression of Myosin heavy chain (MHC) and indirect flight muscle actin. MHC gene expression during wild-type IFM development at 24 hours APF, as seen with a specific antibody, is shown in Fig. 8A. When N-intra is expressed persistently during IFM development, MHC levels are considerably lowered (Fig. 8B). A similar lowering of levels of IFM-specific actin, assayed by Actin 88F-lacZ expression, is observed (data not shown). However, DFMs, which derive from the same pool of disc-associated myoblasts as IFMs, are not noticeably affected by activated Notch (Fig. 7B), although the GAL4 driver is expressed at high levels in DFMs (Fig. 4D) and in their progenitor myoblasts (Fig. 4A).

The effect of Notch activation and reduction on twi, and the phenotypes of Notch mutants led us to examine the effects of twi mutant combinations on flight muscle development. Null mutations in the twi gene cause embryonic lethality when homozygous, since the gene product is required for ventral furrow formation (Thisse et al., 1987a,b, 1988). However, a heteroallelic combination of two twi mutations, twi^ry50/twi^v50, is viable and shows a temperature-sensitive loss of twi function (Thisse et al., 1987b). Animals of the genotype twi^v50/twi^ry50 were grown at permissive temperature till the second larval instar and then raised to non-permissive temperature for the rest of their development. Such a treatment did not affect larval development. However, the development of IFMs was affected. In particular, the DLMs were severely affected and only three fibres are seen (Fig. 7C). DFM development is not significantly affected though individual muscles are slightly thinner that in the wild type (Fig. 7D).

The expression pattern of twi during wild-type development suggested that continued expression of the gene could prevent differentiation. To test this we expressed a UAS-twi transgene under the control of the 1151-GAL4 driver described above. Fig. 9A and B show the expression of ewg in wild-type animals at 24 hours APF. When twi is expressed in a persistent manner using the 1151-GAL4 driver, IFM development is aborted. Although ewg expression is seen at 24 hours APF in such animals, the larval templates fail to split and eventually degenerate (Fig. 9C,D). It appears that a substantial part of the ewg expression is in unfused myoblasts, although it is not possible to unequivocally state that no fusion has occurred. This is in contrast to the wild type at this stage where a substantial part of ewg expression is already inside the developing fibres (Figs 2, 9A,B). In order to examine further the effect on muscle differentiation of persistent Twi expression we studied the expression of Actin 88F-lacZ (not shown) and MHC. As in the situation where N-intra is persistently expressed during IFM development, persistent twi expression results in lowering of MHC levels in the IFMs (Fig. 8C). Curiously, some elongated myoblasts with long processes are seen over the IFMs. They express high levels of MHC (Fig. 8C inset). Ectopic MHC expressing multinucleate myotubes are also seen in this region. In the mature animals that develop upon persistent expression of twi in the flight muscles, DLMs and DVMs are absent, while DFMs are present and appear normal (Fig. 9E,F).

Recent studies on the mechanisms underlying specification of muscle identity and differentiation have suggested similarities and differences between Drosophila and vertebrates (Michelson, 1996; Spicer et al., 1996; Baylies and Bate, 1996) and even within individual organisms. The Twi protein is an example of this. In vertebrates, Twi regulates skeletal muscle differentiation negatively. In cell transfection studies, continued expression of murine Twi impaired the ability of myogenic cells to differentiate (Hebrok et al., 1994) as assayed by prevention of myoblast fusion and lowered induction of myogenic markers. Although Twi can negatively regulate MEF-2, a positive regulator of muscle differentiation, in mouse cells as well as in flies (Spicer et al., 1996; Nyugen et al., 1994; Lilly et al., 1994; Taylor et al., 1995), dissimilarities exist. In the developing Drosophila embryo (Baylies and Bate, 1996), Twi is not only involved in cell-fate choice between somatic and visceral mesoderm in a dose-dependent manner, but twi mis-expression in the ectoderm can give this germ layer mesodermal properties such as the expression of a muscle-specific gene. However, prolonged twi expression at high levels does not prevent differentiation of somatic muscle although, during normal muscle development, twi expression declines at differentiation.

Our observations on twi function in the developing flight muscles lead us to suggest that twi repression could be a requirement for the initiation of muscle differentiation in some muscles of the fly. Persistent twi expression aborts the development of these muscles and markers of differentiation such as myosin are greatly reduced (Fig. 8). Thus, in these muscles at least, Twi functions in a manner similar to that observed for murine Twist.

twi expression during IFM development declines as muscle differentiation begins (Fernandes et al., 1991). We have shown in this study that Ewg, a transcription factor required for indirect flight muscle differentiation (DeSimone et al., 1995) begins to be expressed as twi expression declines (Fig. 2). It is important to note here that ewg expression is seen in myoblasts, overlying the larval templates, prior to fusion. In addition expression is subsequently seen inside the fibre. ewg encodes a regulatory factor and not a structural component of the muscle. Thus, the onset of regulatory events that lead to differentiation, just prior to myoblast fusion, is strongly correlated with the lowering of twi expression in these myoblasts. Very few myoblasts express both twi and ewg. Is Twi a negative regulator of ewg expression? Extant reagents -a conditional twi allelic combination and ewg regulatory sequence information, do not provide an unequivocal answer to this. However, when twi is persistently expressed during adult flight muscle development, ewg expression is still seen, indicating that the mere presence of Twi is not sufficient to prevent the onset of ewg expression.

Temperature-sensitive twi mutant animals grown at the non-permissive temperature from the second larval instar appear not to be affected until pupal development, at which time the formation of the IFMs is severely affected. In such animals, the development of the DFMs appears normal. The interpretation of the phenotype from conditional loss-of-function mutants must be tempered by some unknowns. Although the conditional allelic combination appears to exhibit a null phenotype in the embryo, it is possible that the effect of raising larvae and pupae to the non-permissive temperature effective for embryogenesis does not result in an adult ‘null’ phenotype. Given this caveat, our results from the twi temperature-sensitive mutant experiment suggest that Twi function is not required for larval life, during and after the second instar. However, Twi is required during pupal development for the development of the IFMs, but not the DFMs. Does removal of twi function result in premature muscle differentiation? The defects seen in twi mutant animals are at the earliest stages of DLM development when myoblasts fuse to the larval templates. This is consistent with the interpretation that reduction of Twi levels could have caused premature differentiation and thus have left fewer myoblasts that are correctly positioned to contribute to DLM development. This interpretation would also be consistent with the Notch loss-of-function data.

The phenotypes resulting from persistent twi expression are also revealing about the mechanisms that could operate during flight muscle development. Persistent and high levels of twi expression in wing-disc associated myoblasts during larval life has no apparent effect on larval development. But persistent twi expression affects IFM differentiation in a manner very similar to persistent expression of activated Notch: the larval templates degenerate and often fail to split (Fig. 9C,D) suggesting that very early events in myogenesis such as fusion are affected. However, Ewg, a marker for the onset of differentiation is detected in this situation. Since muscle development is in the process of being aborted in animals with persistent twi expression at the stage when Ewg is initiated, it is not easy to determine whether levels of Ewg are reduced or not. It is clear though that the expression of markers of differentiation such as myosin (Fig. 8) and IFM-specific actin are reduced. In addition to its effect on muscle differentiation, continued expression of Twi causes myoblasts to ectopically differentiate and express high levels of MHC. A possible explanation of these results is that persistent expression of twi does delay differentiation; eventually, however, unfused myoblasts may differentiate at ectopic locations as the long MHC expressing cells shown in Fig. 8C. This raises the question as to what happens during normal development to unfused myoblasts. It is likely that unfused myoblasts die during normal development and the ectopic differentiation seen upon persistent twi expression may be suggestive of interference with cell-death programs, a hypothesis which though tantalising, needs substantial investigation. In addition, when twi expression persists, the development of DFMs is apparently not significantly perturbed, a result similar to that seen upon reduction of twi function. It could be argued that the lack of a DFM phenotype in conditional twi mutant animals could be because of remnant twi function in the allelic combination used. However, we know that the GAL4 driver used is expressed at high levels in the developing direct flight muscles (Fig. 4D). Therefore, the presence of DFMs when twi is expressed persistently during direct flight muscle development very likely reflects the fact that these muscles are relatively refractive to manipulations of Twi levels. We discuss this in relation to other manipulations of DFMs later in this section.

What are the signals that control the down-regulation of twi during normal development? Flies carrying a temperature-sensitive allele at the Notch locus show a flight muscle phenotype strikingly similar to twi mutant animals when grown during early pupal development at non-permissive temperature. Here too, IFMs are affected but DFMs are not (and here too, it can be argued that the absence of effect on DFMs is due to perdurance of wild-type Notch function or because developing DFMs may have a very different temperature sensitive period). However, persistent expression of an activated Notch protein using the 1151-GAL4 driver also affects the development of the IFMs but not the DFMs. Thus Notch loss- and gain-of-function animals show similar effects (Figs 6 and 7) as twi loss- and gain-of-function animals on flight muscle development (Figs 7 and 9). In addition, twi-lacZ reporter gene expression and Twi protein levels are significantly lowered in N^ts1 animals (Fig. 3), and persistent twi expression is seen when activated Notch is expressed constitutively during flight muscle development (Fig. 5).

There are at least two explanations for these results that are consistent with known functions of Notch. Notch could function in the development of adult muscle, in a manner similar to that in the embryo, by its involvement in choice of a founder cell by ‘lateral inhibition’ (Corbin et al., 1991; Bate et al., 1993; Baker and Schubiger, 1996). In this a ‘founder cell’ is chosen from a pool of myoblasts and the Notch receptor functions in other myoblasts to receive signals that inhibit these myoblasts from becoming founder cells. Founder cells, identified in embryonic muscle development (Rushton et al., 1995), prefigure the developing muscle fibre and are the earliest detectable event in muscle differentiation in the embryo. Thus, in this model of Notch function, similar to that well-characterized in neurogenesis (Simpson, 1990), loss of Notch function will result in ‘more’ founder cells and twi expression will decrease as a result of the onset of differentiation. Conversely, persistent expression of activated Notch will result in myoblasts being unable to take on a founder cell fate. Consequently twi expression persists. Experiments by Ruiz-Gomez and Bate (1997) have shown that Notch is required for the segregation of twi-expressing adult muscle progenitors in the embryo.

In the second model Notch signaling could play a direct role in maintaining the un-differentiated state till myoblasts are correctly positioned to receive appropriate environmental signals to differentiate. Similar roles for Notch during Drosophila compound eye development (Fortini et al., 1993) Xenopus (Dorsky et al., 1995) and chick (Austin et al., 1995, Henrique et al., 1997) eye development have been suggested. Thus, in the second model, Notch signalling maintains the undifferentiated state and reduction of Notch function results in the reduction of Twi levels, since differentiation begins as Notch function declines. Conversely, expression of activated Notch maintains the cells in an undifferentiated state, and twi expression persists as an indirect consequence or as a direct effect of Notch signalling.

In favour of Notch acting to select a founder cell, is the observation of a large number of spindle-shaped cells. Thus, if loss of Notch function in the adult were to result in an excess of founder cells, these spindle-shaped cells could mark events consequent to this. However, if loss of Notch function were to lead to premature differentiation, the spindle-shaped cells could well represent premature differentiation events. Persistent expression of activated Notch affects the DVMs more severely than the DLMs (all three DVM groups seldom form). However occasional formation of a DVM is initiated as seen by IFM actin-lacZ expression. This suggests that though segregation of DVM founders do occur, high levels of activated Notch interfere with the process. Until markers for founder cells in adult myogenesis are identified, it will be difficult to distinguish between these possibilities. In the embryo, Notch loss-of-function results in an excess of cells that express markers for founder cells (Corbin et al., 1991; Bate et al., 1993; Baker and Schubiger, 1996) and expression of activated Notch in the mesoderm prevents myoblast fusion, though some muscles are unaffected (Baker and Schubiger, 1996).

An intriguing result obtained by persistent expression of activated Notch and twi is the significant difference in effects on very closely related muscles, the IFMs and the DFMs. It demonstrates how very closely related muscles can develop in very different ways. Supporting evidence comes from muscle development in the ‘four-winged-fly’ in which the T3 ectoderm is homeotically transformed towards a T2-like phenotype. Of the T2 set of thoracic muscles, DFMs develop normally beneath the transformed T3, however IFM development is severely deranged (Fernandes et al.,1994). In addition, DFMs, in otherwise wild-type animals, show no effect of ectopic expression of the homeotic selector genes Antennapedia or Sex-combs-reduced whereas the IFMs fail to develop in this situation (Roy et al., 1997 and S. Roy, unpublished observations). DFMs and IFMs are affected very differently by perturbations in levels of expression of various patterning genes. It is likely that developing DFMs respond to different cues possibly originating from other tissues like the overlying epidermis or the nervous system. Transplantation experiments (VijayRaghavan et al., 1996; Roy and VijayRaghavan, 1997) in the pupa and experiments by Baker and Schubiger (1995) in the embryo show that the ectoderm does influence the expression of muscle specific genes. For IFM development, these cues result in the Notch pathway being involved in differentiation. Mosaic analysis suggests that Notch receptor function is required autonomously in the mesoderm at this stage (S. Anant, unpublished observations). A mosaic analysis of the function of candidate ligands for the Notch receptor will reveal if the Notch phenotype is in response to signaling from the ectoderm, or if another signaling pathway is involved. Factors that regulate Notch activity in the embryonic segregation of adult myoblasts might also function at later stages during development to delineate IFM myoblasts from DFM myoblasts (Ruiz-Gomez and Bate, 1997). The investigation of how differences arise within the dorsal muscle set of the adult thorax is now tractable and promises to be of general value.

We thank Mary Baylies and Mike Bate for the UAS-twi strain and Masahiro Go and S. Artavanis-Tsakonas for the UAS-N-intra strain. Stimulating discussion with Mike Bate, Bill Chia and Veronica Rodrigues are gratefully acknowledged and we thank Veronica Rodrigues and Bill Chia for comments on the manuscript. We also thank the referees for their very constructive suggestions which have contributed substantially to improving the quality of this study. This work was supported by the Department of Science and Technology, Government of India, a Human Frontier Science Program grant and funds from NCBS.