Introduction

Most invertebrates have sense organs distributed over the body surface. Adults (imagos) of the true flies or Diptera generally bear a large number of sensory bristles. These arise from single precursor cells, sensory organ precursors (SOPs) that are born in the imaginal epithelium and develop into a bristle organ complete with its cuticular shaft, socket and underlying neuron and sheath cell, at the time of metamorphosis (Hartenstein and Posakony, 1989). Movement of the shaft in its socket causes excitation of the underlying neuron. The precursors do not move from their site of origin and so the adult array of bristles indicates their spatial arrangement at birth. Development of the bristle pattern on the thorax of Drosophila is well described. There are two classes of bristles, large (macrochaetes) and small (microchaetes) that arise at two distinct temporal phases (Simpson et al, 1999) (Figures 1 and 3). Precursors of the macrochaetes are born early during the larval period. As a result of the long period of intervening growth before metamorphosis these bristles are widely spaced apart. Precursors of the microchaetes are born later during the pupal period and as a result are more closely spaced than macrochaetes. Macrochaetes differ from microchaetes in both structure and probable function. They are longer, thicker and stouter. They are arranged into a stereotyped array. Each macrochaete has a specific axonal projection pattern in the thoracic ganglion that depends upon the position at which the bristle precursor is born (Ghysen, 1980). The notal macrochaetes of Drosophila have also been shown to have directional sensitivity (their neurons respond to movement of the shaft in a preferential direction) (Walker et al, 2000). Although they display some regional specificity, microchaetes are variable in number and position and do not appear to have individual defined functions (Usui-Ishihara and Simpson, 2005).

Figure 1
figure 1

Schematic drawings of the thorax of species belonging to different groups of Diptera. Large green dots represent the macrochaetes and small blue dots the microchaetes. Species belonging to the Schizophora (see Figure 2) frequently bear stereotyped arrangements of macrochaetes. Of those mentioned in the text, D. melanogaster and Ceratitis capitata belong to the Acalyptrata, whereas Calliphora vicina and Phormia terranovae belong to the Calyptrata. All of these species have stereotyped bristle patterns. Nematoceran flies, in contrast, generally bear variable numbers of bristles rarely arranged into patterns. There are no clear macrochaetes in many members of this taxa (grey dots). Species of Nematocera mentioned in the text are Anopheles gambiae, Culex pipiens, Clunio marinus, Toxorhynchites utilus, Chironomus thummi, Aedes aegypti and Chaoborus critillinus. None of these have macrochaetes nor stereotyped bristle patterns.

Full size image
Figure 3
figure 3

A correlation between peaks of ecdysteroid activity and the formation of macrochaete precursors. The graph shows the pulses of 20E experienced by D. melanogaster during larval and pupal life. Adapted from (Riddiford, 1993) and (Warren et al, 2006). A small pulse during the third and last larval instar causes wandering of the larva and glue synthesis. Pupariation is initiated at five days of development during a peak of 20E (dotted line). The temporal windows during which macrochaete or microchaete precursors are specified are shown as black rectangles. Note that macrochaete precursors form before, and microchaete precursors after, the 20E pulse corresponding to pupariation.

Full size image

The true flies or Diptera are a very large insect order with a huge number of species. A simplified phylogenetic tree of the Diptera is shown in Figure 2. Among Brachycera a distinction between macro- and micro-chaetes can be made in many taxa. This is particularly clear in the Cyclorrapha where macrochaetes are consistently found in patterned arrangements and microchaetes are mostly present ubiquitously (McAlpine, 1981; Simpson et al, 1999). In addition, macrochaetes have been shown to have position-specific axonal projection patterns in some of these species (Murphey et al, 1989; Usui-Ishihara and Simpson, 2005). In all cyclorraphous species examined so far, macrochaete SOPs are born earlier in the development of the notum than microchaete SOPs (Huang et al, 1991; Simpson et al, 1999; Wülbeck and Simpson, 2000; Pistillo et al, 2002). Using criteria of spacing and size it appears that in some orthorraphous Brachyceran lineages there is a clear distinction between the two types of bristles, but in others and indeed in some basal Cyclorraphan taxa the distinction is blurred and a continuum of bristle sizes can be seen. The macrochaetes are thought to represent a derived feature the origins of which may be traced back to a common ancestor with the Nematocera (McAlpine, 1981). Flies of the Nematocera, considered to be basal, do not appear to bear macrochaetes (McAlpine, 1981; Simpson et al, 1999) (Figures 1 and 2). Firstly, although some taxa of Nematocera have long bristles these are generally thin and not thick and stout like the macrochaetes of the Cyclorrapha. Secondly the bristles in Nematocera are not arranged into stereotyped patterns but are often randomly positioned over the dorsal thorax. Thirdly the bristles, of whatever size, of many Nematocera are equally spaced from one another suggesting that their precursors arise simultaneously. Indeed the precursors for all sensory organs are born in a single short phase in the mosquito Anopheles gambiae (Wülbeck and Simpson, 2002). Thus, macrochaetes may have arisen in the lineage that led to the Brachycera. In this article, we discuss the possible origin of the macrochaetes themselves and of the stereotyped patterns.

Figure 2
figure 2

Phylogeny of the Diptera. The tree represents the phylogenetic relationships between the major taxa of Diptera. The Nematocera and the orthorraphous Brachycera are probably polyphyletic. The names of monophyletic groups are shown at the point of their emergence (filled black circles). The origin of the evolutionary novelties discussed in the text are indicated (grey boxes).

Full size image

Origin of macrochaetes

Heterochronic shift in proneural gene expression

In flies sensory bristles result from activity of the proneural genes achaete (ac) and scute (sc). These encode transcription factors of the basic-helix-loop-helix family that, together with daughterless, provide neural potential to cells (Villares and Cabrera, 1987; Alonso and Cabrera, 1988; Ghysen and Dambly-Chaudiere, 1988; Gonzalez et al, 1989). Two temporally separate phases of ac-sc expression precede the formation of bristle precursors in Calliphora vicina (Calliphoridae), Ceratitis capitata (Tephritidae) and Drosophila (Drosophilidae), an early one for macrochaetes and a later one for microchaetes (Cubas et al, 1991; Skeath and Carroll, 1991; Simpson et al, 1999; Wülbeck and Simpson, 2000; Pistillo et al, 2002). In the single species of Nematocera examined, Anopheles gambiae, there is only one phase of AgASH ( A chaete- S cute- H omologue) expression, correlating with the simultaneous birth of all SOPs (Wülbeck and Simpson, 2002). Although there are two types of sense organs in Anopheles, bristles and scales, SOPs for both arise at the same time. Thus, the origin of macrochaetes may be linked to an additional, earlier phase of proneural gene expression. How could such a heterochronic shift result in two different types of sensory bristles? Studies in Drosophila indicate that an excess of scute before puparium formation (BPF) always results in additional macrochaetes, whereas an excess of scute after puparium formation (APF), results in only additional microchaetes (Rodriguez et al, 1990). So there is a distinct response within the epithelium to the same proneural protein at different times. On the other hand, although there is a small correlation between time of birth and bristle size (Murphey et al, 1980; Skaer et al, 2002b; Usui-Ishihara and Simpson, 2005), the earlier birth of macrochaete SOPs is not sufficient to account for their significantly larger size. Macrochaete SOPs are born over a period of about 30 h, the last forming at 3–4 h APF (Huang et al, 1991). The earliest microchaete SOPs already appear at 8 h APF (Usui and Kimura, 1993). So the question is: a shift in timing relative to what?

Additional pulses of 20-H ecdysone

In cyclorraphous flies pupariation heralds the beginning of metamorphosis. During this process, the larval skin is chemically modified to form the puparium, inside of which the pupal and then adult moults take place. Important hormonal changes occur. Moulting is regulated by pulses of 20-hydroxy-ecdysone (20E); for review see (Riddiford, 1993). In Drosophila there is a peak of 20E at each larval moult and during the last larval instar there are three small peaks (Warren et al, 2006) (Figure 3). One of these commits to pupariation and a second initiates wandering of the larva and synthesis of glue from the salivary glands. These are followed by a much larger peak, which correlates with pupariation. A further small peak causes head eversion and pupal cuticle deposition, and then a very large prolonged peak is associated with adult development. The role of 20E for moulting is very ancient. However, a role in the regulation of patterning and metamorphosis has evolved more recently (Truman and Riddiford, 1999). The effects of 20E are transduced by the heterodimeric receptors Ultraspiracle (Usp) and tissue-specific isoforms of Ecdysone Receptor (EcR), for review see Thummel (2001). Primary response genes such as BR-C and E74 are initiated at all stages in response to 20E to regulate appropriate patterns of target gene activity (Ashburner et al, 1974). However, other transcription factors are stage-specific. For example bFTZ-F1, BHR3 and E75B function at metamorphosis defining the outcome of the prepupal pulse of 20E, for review see Thummel (2001). Expression of stage-specific factors may define temporal states that will then ensure that responses to the next pulse of 20E will be distinct. EcR/Usp can also function as repressors in the absence of ligand. For example it has been shown that in the absence of Usp the SOPs form but differentiate prematurely (Schubiger and Truman, 2000). So the activity of 20E lifts the repressive activity of Usp and regulates the time of differentiation.

It is not completely understood what initiates the expression of ac-sc for macrochaete SOPs on the notum of Drosophila. The upstream activators Pannier and Iroquois are present in the notal epithelium before expression of ac-sc, suggesting that other permissive factors are required. It has been shown that the atonal-dependent sense organs, eyes, chordotonal organs and Johnston's organ, which arise in the early third instar, are dependent on the presence of 20E at that time (Niwa et al, 2004). Whatever the genetic regulation of expression, it is possible that the timing of SOP formation, relative to the successive pulses of 20E, determines whether a macro- or a micro-chaete will form. Formation of macrochaete SOPs in Drosophila is completed shortly before the peak corresponding to pupation, but microchaete SOPs are born after this pulse (Figure 3). A recent comparison between two species of blowfly suggests that the time of SOP formation is critical for the formation of macrochaetes. Calliphora vicina and Phormia terranovae (Calliphoridae) are two closely related flies with differing macrochaete patterns (Figure 1). They display two phases of sc expression on the developing notum, an early patterned one that is identical for both species and a later (ubiquitous) one. In Calliphora these are separated by about 24 h and macrochaete SOPs arise from the first phase and microchaetes SOPs from the second. In Phormia, however, the first phase is slightly delayed and the second phase greatly accelerated so that there is a small overlap between the two. As a result the last SOPs to form from the first patterned phase arise at the same time as the first SOPs from the second phase and they display a morphology that is indistinguishable from the ‘true’ microchaetes (Skaer et al, 2002b). Interestingly the period of overlap occurs soon after pupariation, consistent with a hypothetical link with 20E levels.

So what about the lower flies that do not display macrochaetes? Very few families of Nematocera undergo pupariation. A few families do pupariate and carry out the pupal moult inside in the last larval cuticle: Perissommatidae, Scatopsidae, some Bibionidae, some Cecidomiidae (McAlpine, 1981). However, the pupae in all these cases are obtect, an ancestral feature where the appendages are fastened down to the body wall (see below). Note also that in these cases the puparium is little modified from the larva, unlike that of cyclorraphous flies. Obtect pupae are also characteristic of the Orthorrapha that similarly do not pupariate. Flies of a single superfamily of orthorraphous Brachycera, the Stratiomyoidea, do pupariate, but like the Nematocera do not have imaginal discs that are free from the body wall. In pupae of the Cyclorrapha, the appendages are free from the body wall, a state derived from the obtect condition. Pupariation must, therefore, have arisen several times independently (McAlpine, 1981).

Flies that do not pupariate moult directly from larva to pupa (McAlpine, 1981). Measurements of ecdysteroid titres have been made in a few such species: Clunio marinus and Chironomus thummi (Chironomidae), and Toxorhynchites rutilus (Culicidae) (Westbrook and Russo, 1985; Laufer et al, 1986; Neumann and Spindler, 1991). There is a probable commitment peak in the last (fourth) larval instar and then about a 10-fold increase corresponding to pupation. Therefore, the pulses that initiate larval wandering/glue synthesis and, in particular, pupariation in Drosophila are absent. Interestingly this difference is during the critical period that separates the formation of macro- and micro-chaete SOPs in Drosophila. We postulate that exposure to different pulses of 20E in the Cyclorrapha may predispose the epithelium to respond to ac-sc expression by making either macro- or micro-chaetes.

Early formation of imaginal discs that include the trunk

Many macrochaete precursors in Drosophila are born BPF. Furthermore they occupy precise stereotyped positions, as is the case for many other cyclorraphous species. The genetic control of the pattern in Drosophila has been shown to result from a cascade of activity of regulatory genes culminating in a precise spatial expression of ac-sc (Modolell and Campuzano, 1998; Calleja et al, 2002), see below. Sufficient time is required for these patterning events. In cyclorraphous flies the adult head and thorax develop from imaginal discs: invaginations of the epithelium that are continuous with the larval epithelium (Figure 4). The Cyclorrapha are the only holometabolous insects in which the discs form in the embryo. Thus, they can grow and be patterned during the entire larval period. A loss of sensitivity to juvenile hormone in the Cyclorrapha may allow this early growth (Truman and Riddiford, 1999). Furthermore the discs comprise not only the appendages but also the trunk. Thus, the notum is included in the wing imaginal disc. The wing disc is not associated with any sense organs present at the surface of the larva, although Keilin's organs and the dorsal organ remain associated with the leg and antennal discs, respectively. The appendages are free from the body wall and in Drosophila are deeply embedded in the larval body. In Phormia the leg discs become deeply invaginated only after the final larval moult (Lakes-Harlan et al, 1991). Although the imaginal discs probably secrete a thin endocuticle at each larval moult (Svacha, 1992; Fristrom and Fristrom, 1993), they develop relatively free of the larval moult cycles.

Figure 4
figure 4

The evolution of imaginal discs. The drawings show cross-sections through the body of the larvae. (a) In the basal species Anopheles gambiae the future appendages are present as pouches budding from the larval epidermis. The future dorsal notum develops from a group of cells at the border between the larval and wing bud epithelia. Reproduced from (Clements, 1992). (b) In cyclorraphous Brachycera, as shown here for D. melanogaster, the developing imaginal discs are embedded well inside the laval body and remain connected to the larval epithelium through long thin stalks. The thorax and the wing both originate from the imaginal disc.

Full size image

Appendages that are free of the body wall and develop during early larval stages represent a derived condition (McAlpine, 1981). In contrast in Nematocera, imaginal discs only form at late larval stages. For example in Aedes aegypti and Anopheles gambiae (Culicidae) epithelial cells for the imaginal legs appear at the second larval instar (Clements, 1992). They are found as long folded structures at the fourth larval instar. In Clunio marinus (Chironomidae) and Chaoborus cristallinus (Chaoboridae) the discs only develop at the last larval instar (Neumann and Spindler, 1991; Svacha, 1992; Melzer et al, 1999). In Nematocera, as well as in the Orthorrapha, the appendages are fastened down to the body wall, an ancestral condition. Another characteristic of the discs of Nematocera is that they include the appendages but not the trunk. A fate map of the leg disc of Culex pipiens (Culicidae) indicated only the presence of the leg itself (Spinner, 1969). In Anopheles gambiae the dorsal notum of the adult was found to derive from a group of epithelial cells situated at the border between the wing appendage and the larval epithelium that undergo division and replace the larval epithelium during the last larval instar (Wülbeck and Simpson, 2002). Therefore, while the imaginal discs of cyclorrphous flies have an extended period of time for patterning and development, those of Nematocera develop over a rather short time period.

Origin of stereotyped patterns

The four row bauplan

Macrochaetes are almost invariably arranged into species-specific patterns that, particularly in acalyptrate cyclorraphous flies, may be stereotyped (McAlpine, 1981; Simpson et al, 1999). Nearly all patterns, however, appear to be variations on a theme of four longitudinal rows on the scutum (Figure 5) (McAlpine, 1981; Simpson et al, 1999). This postulated ancestral pattern is independent of size and flies may display a complete or partial loss of rows but never additional rows. What is the basis for this apparent constraint in the bristle pattern? One possibility is that this trait is favoured by selection. Artificial selection experiments so far do not appear to support this view (Macdowell, 1915; Dominguez et al, 1987; Pineiro, 1992a, 1992b; Dominguez et al, 1993). Although a ‘wild type’ pattern of 11 macrochaetes per heminotum is assumed to be fixed in Drosophila melanogaster, natural variants of this pattern can be found and enough genetic variation exists in nature to select flies for an increased number of notal bristles (Macdowell, 1915; Plunkett, 1926; Sheldon and Milton, 1972; Pineiro, 1992a, 1992b; Dominguez et al, 1993). Selection experiments result in the presence of additional dorsocentral bristles, scutellar bristles and some bristles on the lateral scutum (Macdowell, 1915; Plunkett, 1926; Vreezen and Veldkamp, 1969; Pineiro, 1992a, 1992b; Dominguez et al, 1993). The dorsocentral bristles may become numerous with continued selection, but they are either positioned close to the two extant ones or extend anteriorly to form a longitudinal row reminiscent of the ancestral pattern. Alternatively, developmental constraints could restrict the emergence of macrochaetes to particular sites. Indeed the additional bristles resulting from selection are not located over the sites of attachment of the flight muscles (Usui et al, 2004). There appears to be a constraint in the form of the indirect flight muscles that lie just below the surface of the scutum (Figure 6). These attach to the cuticle through tendons whose precursors are selected from the same epithelium as the bristle precursors (Lee et al, 1995; Fernandes et al, 1996; Frommer et al, 1996; Volk, 1999). In contrast to the bristles, the pattern of flight muscles and their attachment sites is highly conserved throughout the Diptera (Tiegs, 1955; Usui et al, 2004). Remarkably the flight muscles and the leg jump muscle attach in longitudinal domains that are between the rows of macrochaetes (Usui et al, 2004). From an examination of several hundred species, none were found in which the macrochaetes were located over the sites of muscle attachment. Bristle precursors develop in areas of ac-sc expression; tendon precursors develop in areas of expression of stripe (sr), a gene encoding a transcription factor with zinc finger motifs (Lee et al, 1995; Frommer et al, 1996). The domains of expression of ac-sc and sr are spatially segregated in Drosophila and Calliphora vicina (Usui et al, 2004; Richardson and Simpson, 2006). Experimentally contrived mis-expression of sr at the sites where bristle precursors develop, results in a loss of bristles in Drosophila; similarly mis-expression of ac-sc at the sites of tendon development interferes with muscle attachment (Usui et al, 2004). The mutual antagonism between the products of these genes and their spatially separate expression domains in wild type Drosophila, would account for the spatial segregation of tendons and bristles. This molecular mechanism, if conserved, would account for the four row bauplan.

Figure 5
figure 5

Proneural clusters may be derived from stripes of sc expression. Pistillo et al (2002) suggested that the proneural genes were expressed in an pattern of four longitudinal rows on the scutum of the ancestor leading to the Cyclorrapha (top left). These rows, and the bristles to which they give rise in the adult (bottom left) are named and represented with alternate black and white colour code. It is proposed that in the lineage leading to the Acalyptrata, different rows were entirely or partially lost, which correlates with a proneural gene expression pattern of reduced proneural clusters. As a result these flies have fewer bristles, but frequently in stereotyped arrangements on the notum. Abbreviations: AC, acrostichal; DC, dorsocentral; IA, intraalar; SA, supraalar; SC, scutellar. Adapted from Pistillo et al (2002).

Full size image
Figure 6
figure 6

The muscle attachment sites on the thorax of Diptera. The indirect flight muscles of Diptera attach to tendons whose precursors arise in the same imaginal epithelium as the bristle precursors, shown for Drosophila on the left. Note that the macrochaetes form between the muscle attachment sites (small circles). On the right, the muscle attachment sites are named and shown in black. Abbreviations: DLM, dorso-longitudinal; DVM, dorso-ventral; TDT, trochanter depressor. Adapted from Levine and Hughes (1973).

Full size image

A conserved trans-regulatory landscape for sc expression in cyclorraphous flies

Among the cyclorraphous flies examined so far there is a correlation between the spatial expression of sc and bristle patterns. sc is expressed in discrete domains, either in stripes or in small clusters of cells called proneural clusters, at the sites of formation of the future bristle precursors (Cubas et al, 1991; Skeath and Carroll, 1991; Wülbeck and Simpson, 2000; Pistillo et al, 2002). Thus, changes in sc expression have accompanied evolution of the bristle patterns. Are these changes due to evolution of cis-regulatory sequences at the sc locus or to evolution of the upstream trans-regulatory factors, or both?

Studies in D. melanogaster have uncovered some of the transcriptional regulators of ac-sc. Pannier, a member of the GATA protein family, has been shown to directly activate ac-sc for development of the bristles on the medial half of the notum (Garcia-Garcia et al, 1999). The genes of the Iroquois (iro) complex are required for the bristles of the lateral half of the notum (Gomez-Skarmeta et al, 1996). Consistent with this pannier (pnr) is expressed in the dorso-medial notum and the iro genes araucan and caupolican (caup) in the dorso-lateral notum (Ramain et al, 1993; Gomez-Skarmeta et al, 1996). Downstream of Pnr and Iro, other transcription factors progressively subdivide the notum into smaller developmental units (Aldaz et al, 2003). Other factors act either downstream of Pnr and Iro, or modify their transcriptional activity through protein association. Thus, U-shaped (Ush) associates with Pnr modulating its activity, Wingless (Wg) is required for ac-sc expression in the medial notum, while Stripe (Sr) has been shown to negatively regulate bristle development (Phillips and Whittle, 1993; Cubadda et al, 1997; Haenlin et al, 1997; Ramain et al, 2000; Usui et al, 2004). ac-sc, sr and wg are all targets of Pnr and are expressed in discrete subsets of cells within the broad medial domain of pnr expression (Garcia-Garcia et al, 1999; Ghazi et al, 2003).

Expression of pnr in the dorso-medial notum is conserved between Drosophila, Ceratitis and Calliphora (Ramain et al, 1993; Wülbeck and Simpson, 2000; Pistillo et al, 2002). Furthermore, with only minor differences, expression of caup, ush, sr and wg is conserved between Drosophila and Calliphora (Richardson and Simpson, 2006). This suggests the existence of a system of interacting components that has been conserved across about 100 Myear of evolution. These components are also required for the development of many other structures that together make up the detailed morphology of the notum (Calleja et al, 2002). Expression of these genes may have evolved in concert since altering one component may affect others and modify many thoracic structures.

In contrast to the cyclorraphous flies, spatial expression of trans-regulatory factors in the basal species Anopheles reveals important differences. Expression of Ag-pnr is restricted to the medial notum but it does not occupy a single broad domain. Instead there are four separate domains of expression. Furthermore Ag-ASH (the Anopheles proneural gene) is expressed in domains that appear co-incident with those of Ag-pnr, rather than in subsets of cells within the Ag-pnr domains (Wülbeck and Simpson, 2002). Precursors for the sensory scales arise everywhere within each expression domain of Ag-ASH, but patterning of the bristles is quite different from anything seen in cyclorraphous flies. Bristles arise along the borders of Ag-pnr/Ag-ASH expression domains. It is not known how the bristle pattern is controlled but clearly there are no proneural stripes or clusters, and one does not observe a spatial correlation between the expression of the proneural gene and the positioning of bristles. The expression domains of Ag-ASH being co-incident with those of Ag-pnr, it is likely that Ag-pnr is involved in the regulation of the bristle pattern. This suggests a conserved role for a function of Ag-pnr in patterning the notum.

Gene duplication and evolution of cis-regulatory sequences

The apparent conservation of upstream regulators may mean that much of the change is to be found in cis-regulatory sequences of the ac-sc genes themselves. The ac-sc complex of Drosophila has been dissected in some detail. Expression of ac-sc in proneural clusters is regulated by discrete independently acting cis-regulatory enhancer elements scattered over about 100 kb (Ruiz-Gomez and Modolell, 1987; Gomez-Skarmeta et al, 1995) (Figure 7). Interaction between some of these and their upstream regulators has been examined in detail (Gomez-Skarmeta and Modolell, 1996; Garcia-Garcia et al, 1999; Ramain et al, 2000). When did these sequences arise and to what extent are they conserved? If stereotyped patterns are derived from an ancestral pattern of rows, then are proneural clusters derived from an ancestral expression pattern of longitudinal stripes like those seen in Calliphora (Figure 5)? Are such stripes of expression driven by homologous enhancers? The answers to these questions await the isolation of regulatory sequences from other species. In the meantime, however, there are some indications that enhancers for spatial expression may have arisen somewhere in the lineage leading to the cyclorraphous flies.

Figure 7
figure 7

Genomic organisation and function of the ac-sc complex in D. melanogaster. (a) Genomic diagram showing the position of ac, sc, lethal of scute (l'sc) and asense (ase) as small arrows. Known enhancers driving expression of ac and sc in the thorax and wing are represented as black or grey boxes respectively. The name(s) of the proneural cluster(s) where the enhancer is active is indicated below. The downstream limit of the SC enhancer is still uncertain (fading black filling). Adapted from Gomez-Skarmeta et al (1995). (b) A photograph of the thoracic region of a wing disc, showing the expression of sc mRNA in proneural clusters. The clusters whose enhancers are known have been named. (c) A dorso-lateral view of an adult thorax. The macrochaetes arising from the proneural clusters described in (b) are named.

Full size image

The four domains of Ag-pnr expression appear to be co-incident with the domains of expression of Ag-ASH (Wülbeck and Simpson, 2002). This suggests that regulation of Ag-ASH by Ag-Pnr may be much simpler and would not need recourse to numerous enhancer elements. Thus, Anopheles Ag-ASH may not have a complex modular promoter like that of Drosophila. If so, this implies an origin for these sequences after the divergence of Anopheles and Drosophila. Interestingly, the number of genes at the ac-sc complex has increased throughout the Diptera: Anopheles has only two genes whereas Drosophila has four. The four genes have arisen from three independent duplication events (Skaer et al, 2002a). The first duplication (giving asense, a neuronal precursor gene and a single proneural gene) occurred within the insect clade since the beetle Tribolium castaneum and probably the butterfly Precis coenia have these genes (Galant et al, 1998; Wheeler et al, 2003). The second two, however, took place within the Diptera. Phylogenetic analysis revealed that Ag-ASH, the proneural gene of Anopheles is homologous to three proneural genes in Drosophila: lethal of sc (essential for development of the central nervous system), sc and ac (essential for development of the peripheral nervous system) (Skaer et al, 2002a). If the duplications included not only coding sequences but also regulatory sequences, then this may have facilitated the emergence of numerous enhancer elements that would then be free to diverge. It is tempting to speculate that the duplication events are associated with the evolution of stereotyped patterns of the Cyclorrapha.

Conclusions and perspectives

We have discussed the possibility that macrochaetes may have arisen as a result of an additional, early phase of proneural gene expression before pupation. Two other factors may have contributed to their formation. First the Cyclorrapha is a monophyletic group characterised by the phenomenon of pupariation. This has probably incurred the introduction of an additional peak of 20E that regulates puparium formation prior to the peak that regulates pupation per se. The response of the epithelium to the presence of ac-sc before, at and after the extra 20E peak may allow a distinction between macro- and micro-chaetes. Second the early formation of imaginal discs free from the body wall that include the trunk, allows a prolonged period of time during larval life within which to pattern the macrochaetes. Have macrochaetes evolved more than once within Diptera? Macrochaetes are found throughout the Brachycera but many taxa are devoid of them. So either they have arisen more than once or they arose once and have been lost many times. The fact that most stereotyped patterns are variations on a theme of four longitudinal rows on the scutum, would seem to suggest an origin, in an ancestral species, of a four row bauplan that would be unlikely to have arisen multiple times. However, the discovery that macrochaete patterns are constrained by sites of attachment of an invariant pattern of flight muscles may indicate that early proneural gene expression in any lineage could have led to convergent patterns. Thus, perhaps macrochaetes have arisen independently a number of times.

The spatial expression pattern of sc correlates with the bristle patterns in the cyclorraphous species examined so far, but such a correlation is not seen for the ac-sc homologue of Anopheles, a basal Dipteran species. To better understand the steps leading to this transition, it is necessary to examine the expression of ac-sc homologues in phylogenetically intermediate families, such as those of the orthorraphous Brachycera (Figure 2). The differences in spatial expression patterns of sc between Calliphora and Drosophila appear to be mainly due to differences at the sc locus itself. There appears to be a conserved network of upstream regulators. Thus, the challenge for the future will be the isolation of cis-regulatory sequences of sc from other cyclorraphous species and their comparison with those of Drosophila. This may be a first step towards tracing the origins of these enhancers within the Dipteran lineage and their role in the evolution of stereotyped bristle patterns.