Evolution of the insect body plan as revealed by the Sex combs reduced expression pattern

Homeotic (HOM/Hox) genes are a group of regulatory loci that control elaborate biological processes and have been implicated in the evolution of animal body plans (Lewis, 1978; Patel, 1994; Carroll, 1995). The examination of homeotic mutant phenotypes and their apparent segment-specific patterns of expression has led to the proposal that the HOM/Hox genes control segment identity through a kind of ‘genetic address’ system (Lawrence, 1992; Lawrence and Morata, 1983). A large set of experiments on Drosophila melanogaster and other metazoans suggest that the HOM/Hox genes control specific cell fates (e.g. Botas, 1993; Peifer et al., 1987) and that specific identity is a unique combination of cell types achieved by a unique mosaic of HOM/Hox gene expression patterns (Castelli-Gair and Akam, 1995). The products of the HOM/Hox genes are a set of transcription factors (Affolter et al., 1990) that control cell fates by binding DNA via the homeodomain and activating and repressing specific sets of targets, or what have been called ‘realizator genes’ (Garcia-Bellido, 1977). By examining the expression pattern of the Sex combs reduced ortholog in non-drosophilid insects, we can test the generality of the identified functions of Scr that have been determined from the analysis of Drosophila.

The proper development of Drosophila requires the ectodermal expression of Scr in the most posterior segment of the head, the labial segment, and the most anterior segment of the thorax, the prothorax (T1) (Riley et al., 1987; Carroll et al., 1988; Mahaffey et al., 1989; Pattatucci, 1991; Pattatucci and Kaufman, 1991; Gorman and Kaufman, 1995). Scr has quite different roles in these segments. In the embryo, it is required for the formation of the T1 denticle belt and beard (Pattatucci et al., 1991). It is also required for ventral migration and subsequent fusion of the labial lobes (Pattatucci et al., 1991), a general feature of insects, and is the only HOM/Hox gene capable of providing this function (B. Rogers and T. Kaufman, unpublished). In the adult, Scr is necessary for the specification of the labial palps (Pattatucci et al., 1991), for the development of the sex combs on the T1 legs of males (Kaufman et al., 1980; Pattatucci et al., 1991) and, as we show here, for the suppression of wing formation on the prothorax.

Previous work has implicated Scr in the repression of wings on the prothorax (Carroll et al., 1995). These authors showed that in the primordia of the dorsal discs Scr is capable of repressing the expression of vestigial and snail, which are thought to be necessary for wing formation. In addition, the authors proposed that Scr suppresses the formation of wing primordia in other pterygotes. Here we report that certain hypomorphic alleles of Scr result in partial wing formation on the prothorax of adult Drosophila and that the expression of Scr is in the correct location to suppress prothoracic wing formation in other pterygotes.

The early embryonic expression pattern of Scr in Drosophila is complex. It initiates early (stage 5) in a jagged dorsolateral stripe around the border of the maxillary and labial segment primordia which, as judged by stripes of engrailed expression, is neither segmental nor parasegmental in register (Gorman and Kaufman, 1995). Soon after, it resolves into a parasegmental register (PS2) ventrally (primarily progenitors of the central nervous system) and what is largely a segmental register in its dorsolateral domain (the labial epidermis) (Mahaffey et al., 1989). This early expression includes a small number of cells in the lateral epidermis of the posterior maxillary compartment (Riley et al., 1987; Carroll et al., 1988). Later, the expression of Scr expands into the prothorax. This expansion begins in the anterior prothorax and eventually fills the entire prothoracic epidermis but does not expand into the most ventral region of the ectoderm (Riley et al., 1987 ; Carroll et al., 1988; Gorman and Kaufman, 1995). Our observations of Scr expression in other insects reveals that the expression pattern seen in Drosophila is not entirely conserved. Although the anterior (early) expression pattern of Scr in the labium and posterior maxilla is well conserved, expression in the prothorax, the more posterior (late) pattern, shows considerable variation.

The collection of Drosophila, milkweed bug, and cricket embryos was performed according to Rogers and Kaufman (1996). Firebrats (Thermobia domestica) were raised in large (>4 l) glass containers, with a beaker full of water inside, in an incubator kept at 40% humidity and 32°C. Animals of all stages of development are raised together in the same cage and fed dry cat food. Females lay eggs once per molt cycle in cotton placed in the cages (Watson, 1964). Eggs were collected from the cotton every 7 to 10 days and kept in glass Petri dishes at 32°C with moistened tissue paper to keep humidity elevated. Embryogenesis lasts about 12 days. The Scr8 and ScrWrv5 alleles of Drsophila have been decribed previously (Lindsley and Zimm, 1992).

RT-PCR was performed using the GeneAmp (Perkin Elmer Cetus) reagents following the manufacturer’s recommended protocol. Oligo dT was used to prime cDNA synthesis and degenerate primer pairs made to the amino acid sequences PQIYPWM-MNIVPYHM or PQIYPWM-WFQNRR were used to amplify the target cDNAs. PQIYPWM = 5’ CCR CAR ATH TAY CCR TGG ATG 3’ and MNIVPYHM = 5’ CAT RTG GYA NGG NAC RAT RTT CAT 3’ using IUPAC codes. The design of these primers was based upon Scr cDNA sequences of Drosophila (Lemotte et al., 1989), honey bee (Waldorf et al., 1989), Artemia (Averof and Akam, 1993) and the mouse HoxA5 gene (Fibi et al., 1988); they target the Scr-specific YPWM motif upstream of the homeobox and the 3’ end of the homeobox. The WFQNRR primer is the same as described by Averof and Akam (1993). The initial five thermo-cycles used a 50°C annealing temperature with a 90 second ramp time. The final 35 stepcycles were a 50°C annealing temperature with 30 seconds allowed for each annealing, extension and denaturation step. Potential clones were verified by sequencing using Sequenase 2.0 (U. S. Biochemicals) according to the manufacturer’s instructions.

Similarity searches were performed using BLAST (Altschul et al., 1990) to search all currently available data bases. Initial alignments and all sequence manipulations were performed using the MacVector (Kodak) software package. The final alignment was determined by the authors.

In situ hybridization detection was performed as described by Panganiban et al. (1994) except that 0.2% glutaraldehyde was added to the fixative and protease treatment was performed for up to 1 hour at a concentration of 50 µg/ml. Immunohistochemistry was used to detect SCR and engrailed (EN) protein as described by Gorman and Kaufman (1995) and Rogers and Kaufman (1996).

Embryos were mounted on microscope slides in AquaPolymount (PolySciences Inc.) or methyl salicylate. Slides were examined on a Zeiss axiophot microscope and photographed with Kodak ASA100 print film at 50-200× magnification. Scanning electron micrographs were taken of ethanol preserved specimens as described by Gorman and Kaufman (1995). The accumulation of chromagen was considered to be the signal only if it fulfilled the following criteria: (1) was dependent on specific probes, (2) was confined to the cytoplasm and (3) was reproducible.

In order to evaluate the model of homeotic gene function suggested by previous work on Drosophila and to examine the potential role of homeotic genes in insect evolution, we used RT-PCR to clone partial cDNAs of the Scr ortholog from three insects: milkweed bugs (Oncopeltus fasciatus, Hemiptera), crickets (Acheta domestica, Orthoptera) and firebrats (Thermobia domestica, Thysanura). We observed the accumulation of Scr RNA in embryos by in situ hybridization detection and compared these patterns to the expression of Scr protein in D. melanogaster (Diptera) (Riley et al., 1987; Carroll et al., 1988; Mahaffey et al., 1989; Pattatucci, 1991; Pattatucci and Kaufman, 1991; Gorman and Kaufman, 1995; this work). As an aid in determining the domains of Scr expression we also compared these patterns with the engrailed protein accumulation pattern of each insect (see also Rogers and Kaufman, 1996).

Fig. 1 shows the alignment of the conceptually translated Scr partial cDNAs from the milkweed bug, the cricket, and the firebrat with D. melanogaster Scr, Deformed (Dfd), and Antennapedia (Antp). The partial cDNAs encompass three conserved regions of the HOM-C genes. In all three, the non-drosophilid insect sequences resemble Drosophila Scr much more strongly than Dfd, Antp or any other HOM-C gene. In the homeodomain, complete or nearly complete identity is found among the partial cDNAs relative to each other and to Drosophila Scr, while they show several differences when compared to Dfd and Antp.

Further evidence that these partial cDNAs are true Scr orthologs is found by comparison of the sequences immediately flanking both sides of the homeodomain. On the amino-terminal end lies the YPWM motif and what we call the ‘variable region,’ so named because of its variability among members of the HOM-C. The ‘variable region’ sequence, however, is highly conserved among orthologs of each member of the HOM-C, including Scr (Fig. 1 and unpublished data). The ‘C-Conserved Domain’ lies just downstream of the homeodomain and, like the variable region, is conserved among HOM-C orthologs, but differs among individual members of the HOM-C. In addition, a cross species in situ hybridization of milkweed bug Scr (OFScr) to Drosophila embryos specifically detected endogenous Scr expression in a normal pattern (Fig. 2).

The earliest expression of Scr occurs in an anterior domain that corresponds primarily to the labial segment, but also includes ventral and lateral portions of the maxillary segment. In milkweed bugs, crickets and firebrats, this expression initiates in domains that are neither entirely segmental nor parasegmental (Fig. 3D,E, not shown), but soon resolves into a pattern that is parasegmental in register ventrally (central nervous system neuromere) and segmental in its dorsolateral domain (labial epidermis) (Figs 3F-H, 4A,E,I,J).

In all three new species examined, as in Drosophila, there is a shift of register of Scr expression in dorsolateral versus ventral domains at both the anterior and posterior limits of expression. The posterior-most limit of dorsolateral expression is at the labial-T1 segment border (Fig. 3F-H; arrow in 4A,E,I,J). The posterior-most border of the ventral expression, however, lies a few cell lengths anterior to this, at the apparent compartmental border (double arrow in Figs 3F-H, 4A,E,I,J) or what in Drosophila is called the border between parasegments (PS) 2 and 3. Anteriorly, the shift is the same. The ventral expression extends into the posterior maxillary segment (PS 1-2 border; white arrow in Figs 3F-H,K, 4A,E,I,J), while most of the dorsolateral domain stretches only to the maxillary-labial segment border (Figs 3F-H, 4A,E,I,J). In addition, some maxillary cells in the lateral epidermis express Scr (white arrowheads, Figs 3G,I, 4A,F,I,J). As in flies, the maxillary expression in the lateral epidermis of these insects does not fill the entire posterior compartment, but is limited to a subset of cells that become the posterior region of the lacinia, a maxillary structure (see below and Fig. 4).

In addition to the conserved epidermal pattern, the pterygote insects, but not firebrats, also have mesodermal expression of Scr. In crickets this expression appears broadly over the mesodermal primordia that are located ventrally and subepidermally in the maxilla, labium and thorax (white arrows, Fig. 3E). Later in the development of cricket and milkweed bug embryos, Scr expression accumulates in the characteristically spherical cells of the somatic mesoderm (smeso) which underly the columnar epithelial cells of the appendages (white arrows, Fig. 3J; solid white arrowheads, 4A,B,G) and include the progenitors of the leg muscles.

As shown in Figure 4, the mature expression of Scr in the firebrat, cricket and milkweed bug includes small portions of the prothorax, as well as the labial and maxillary expression described above. The prothoracic expression of Scr in milkweed bugs, crickets and firebrats differs from that of Drosophila (see Discussion).

In firebrats, milkweed bugs and crickets, Scr is detectable in a dorsal anterior region of the prothorax near the apparent base of the prothoracic legs (dT1; Fig. 4C,G,K). In milkweed bugs and crickets, Scr expression is also detected in a patch on the anterior side of the leg (large arrowhead, Fig. 4A-C,E-G). These Scr-expressing cells appear to lie just proximal to the tibial-tarsal junction (Fig. 4C,G) and are specific to the prothorax. No such patch is seen on the second or third thoracic legs (T2, T3; Fig. 3B,C). In milkweed bugs, a comb strikingly similar to the Drosophila sex comb is produced at this location on the prothoracic legs of both first instars (Fig. 4D) and adults (and in both sexes), but not on T2 or T3 legs (not shown). No comb-like entity is present on any of the legs of crickets or firebrats (Fig. 4H,L, and data not shown).

Weak expression has also been observed in the CNS extending from the labial segment posteriorly through the abdomen (Fig. 3B,C). In most cases this expression is weak and may be hybridization to other homeobox-containing RNAs. Although apparent low levels of Scr protein accumulation have been detected in the posterior CNS of Drosophila (Gorman and Kaufman, 1995), this expression is dependent on the activity of other homeotic genes (ibid; B. Rogers, unpublished) and is likely to be due to cross-reactivity of the antibody used to other homeodomain proteins.

During the later stages of embryogenesis, Scr continues to be expressed in the labium and cells of the maxilla. Two dynamic aspects of Scr expression in Drosophila that are conserved in milkweed bugs are the retraction of Scr expression from the distal tip of the labium and from the salivary gland (SG; white double arrowheads, Fig. 3K).

Although null alleles of Scr cause embryonic lethality in Drosophila, certain allelic combinations allow animals to live to larval, pupal, and even adult stages. The combination of the hypomorphic allele Scr⁸ in trans with the null allele Scr^Wrv5 allows some survival to the pupal stage. An examination of imagos extracted from pupal cases reveals numerous defects of the prothorax. The most striking of these defects is the production of wing tissue. Fig. 5 is a scanning electron micrograph showing wing growth out of the pleuron dorsal to the prothoracic leg of a pharate adult. The presence of ectopic wings in Scr⁸/Scr^Wrv5 mutants is a clear demonstration that Scr is required for wing suppression on the prothorax of D. melanogaster.

Comparison of the early embryonic ectodermal expression of Scr and the mature derivatives of the domains among the four orders studied here reveals a well conserved pattern. Moreover, due to the structural similarity of the adult Drosophila to the larval stages of hemimetabolous and ametabolous insects, e.g., the presence of seta, thoracic and gnathal appendages, and an exposed head with eyes, a comparison of the Scr pattern in the anlage of the adult, the imaginal discs, is also informative. The segment-wide expression of Scr over the labial (Lab) epidermis and the parasegmental expression in the CNS is observed in all embryos examined (‘+’ in Fig. 6). Additionally, it would appear that Scr expression in the ectoderm of the Drosophila labial disc, and a portion of the maxillary anlagen of the eyeantennal disc (Pattatucci, 1991; Pattatucci and Kaufman, 1991), has counterparts in the species studied here. The retention of the labial expression is consistent with a conserved role in labium formation, specifically the control of migration and fusion of the labial lobes, a feature that sets the labial segment apart from the other gnathal segments.

The maxillary expression of Scr in firebrats and crickets is clearly homologous, as it is confined to the posterior lacinia of the maxilla. Although the milkweed bug and fruit fly maggot lack lacinia because of the modified structure of the mouthparts, we propose that the observed posterior maxillary expression of Scr in a small cluster of cells in these more derived insect embryos is homologous to that seen in the firebrat and cricket. Thus, we conclude that the firebrat and all three pterygotes, including embryonic and adult Drosophila, have homologous expression patterns of Scr in the maxillary (Max) and labial segments (Fig. 6).

Although the expression of Scr in the prothorax of Drosophila differs from the other insects examined, we interpret some individual aspects of the total pattern in the fly as being conserved among either pterygotes or all insects. In addition to being expressed in the embryonic prothorax of Drosophila, Scr is also expressed in the prothoracic leg disc, the dorsal prothoracic (humeral) disc and in the mesoderm of all of the thoracic leg discs (Pattatucci, 1991; Pattatucci and Kaufman, 1991). Unlike the less derived insects, Drosophila embryos do not exhibit patches of dorsal or ventral prothoracic expression. Rather, Scr is expressed across the entire prothoracic ectoderm and presumably overlaps both subregions of expression seen in the other insects. This dissimilarity extends to the prothoracic leg discs in which Scr is expressed over the epidermis of the anlagen of the entire prothoracic leg rather than in a tibial-tarsal patch. Since the broader domains of expression seen in both embryonic and adult Drosophila include the smaller domains of other insects, we interpret expression in the leg patch (LT1) as being conserved in all pterygotes (Fig. 6).

Scr expression in the thoracic somatic mesoderm (smeso) of pterygotes is also conserved (Fig. 6), as demonstrated by the expression seen in the developing legs of milkweed bugs and crickets and in the leg mesoderm of all thoracic leg discs of Drosophila (Pattatucci and Kaufman, 1991). Because the Drosophila maggot lacks appendages, expression of Scr in leg mesoderm is not apparent in the embryo. We did not attempt to examine the visceral mesoderm surrounding the anterior midgut for the expression of Scr. This mesodermal expression is known to be required for formation of the gastric caeca in Drosophila (Reuter and Scott, 1990). At early stages we have no reliable marker for identifying cells of the visceral mesoderm in the non-drosophilid insects and later when midgut morphogenesis occurs, and the mesodermal cells are identifiable, the deposition of cuticle makes the in situ hybridization detection of expression difficult.

The dorsal prothoracic (dT1) expression of Scr differs in embryonic and adult Drosophila. In the embryo, Scr epidermal expression covers the entire prothorax, but in the dorsal prothoracic disc expression is restricted to the anterior compartment of the disc (Pattatucci, 1991). The Drosophila disc pattern more closely parallels the pattern seen in the other insect embryos. This higher degree of similarity between the embryos of less derived species and the imaginal discs that give rise to the adult Drosophila is consistent with the idea that the maggot is more highly derived than the adult (Snodgrass, 1953). We interpret the dorsal prothoracic patch (dT1) as being conserved in all insects (Fig 6).

All modern winged insects lack wings on the prothorax, but examination of Paleozoic fossil insects has shown that the absence of wings on the prothorax is likely a secondary event following the evolution of wing-like appendages on all abdominal and thoracic segments (Kukalova-Peck, 1978; 1987). Kukalova-Peck (1987) has proposed that wings evolved from side lobes that grew out of the exite of the basal-most limb segment, the epicoxa, on all thoracic and abdominal segments. This hypothetical epicoxal exite would be located just ventral to the tergites and dorsal to the leg (telopodite), exactly where we observe Scr expression in modern insects (Fig. 4).

The evidence that Scr suppresses wing development comes from three sources: (1) the demonstration by Carroll et al. (1995) that in the embryonic primordia of the dorsal prothoracic discs, Scr represses vestigial and snail, two genes involved in promoting wing development; (2) the formation of wing tissue on the prothorax of flies carrying a hypomorphic allele heterozygous with a null allele of Scr (Fig. 5); (3) the formation of elytra (T2 wing derivatives) on the prothorax of adult Tribolium carrying a hypomorphic allele heterozygous with a null allele of Cephalothorax, the Scr ortholog (Beeman et al., 1989). Taken together, the ability of Scr to repress wing formation in Drosophila and Tribolium and the expression of Scr in crickets and milkweed bugs just dorsal to the leg argue that in pterygotes Scr functions to suppress the production of prothoracic wings.

As described, a critical step in the evolution of modern winged insects involved the repression of prothoracic wings, apparently mediated by Scr activity. In addition to this role for Scr, there are two further cases in which an expansion of Scr expression within the prothorax of pterygotes has played a role in insect evolution. Since Scr is expressed in the labial and maxillary segments and in the dorsal prothorax of both pterygotes and a Thysanuran apterygote, we infer this to be the Scr expression pattern of the last common ancestor of both groups (Fig. 6). However, neither the ventral (vT1) nor lateral (LT1) epidermal prothoracic expression pattern of Scr is seen in the firebrat; this expression is observed only in the winged insects (Fig. 6). Further, the Scr expression in lateral and ventral prothoracic epidermis is apparently associated with the development of structures uniquely formed in those regions. Thus, the absence of Scr ventral T1 epidermal expression in the presumed insect ancestor and the control of the development of T1-specific structures by Scr must have involved an expansion of the Scr expression pattern in the acquisition of this new role.

The milkweed bug has a patch of Scr expression in the epidermis of the tibia in a position where a leg comb forms. Both the patch of Scr expression and the comb are specific to the prothoracic leg. In Drosophila, proper development of the sex comb requires Scr function and ectopic Scr expression leads to the production of ectopic sex combs on the T2 and T3 legs (Pattatucci and Kaufman, 1991). This suggests that Scr in milkweed bugs may perform a similar role in specifying the prothoracic leg comb. This argument is strengthened by the observations that the combs of Drosophila and the milkweed bug form in non-homologous positions along the proximal-distal axis of the leg and that the milkweed bug comb correlates with the position of Scr expression, not with the position of the Drosophila comb. The Drosophila comb forms on the basitarsus, whereas the milkweed bug comb forms on the tibia. One possible test of the relationship between Scr and leg combs would be to examine Scr expression in other hemipteran species, several of which have multiple combs (Schuh and Slater, 1995). Because the lateral accumulation of Scr in the region of the leg patch is not seen in the firebrat, we conclude that this novel domain seen in the other insects represents an expansion of expression that occurred during the evolution of pterygotes (‘b’ on Fig. 6).

The expression of Scr in the epidermis of the prothorax is unique to Drosophila. We infer that expression has expanded during the evolution of Drosophila to include the ventral and posterior prothorax of the embryo and the ventral epidermis and entire leg of the adult. The ventral expression in the embryo has been shown to be required for the proper development of ventral structures, including specifying denticle types and the elaboration of the prothoracic beard (Pattatucci et al., 1991). Thus, this expansion of the Scr expression domain was necessary for Scr to evolve control of the development of ventral T1 epidermis.

It is commonly accepted that apterygotes, modern Thysanurans in particular, are primitively wingless (Boudreaux, 1979), which makes the dorsal prothoracic Scr expression in the firebrat curious. However, unlike the evolution of wings, which is thought to have occurred once in the progenitor of all pterygotes, the suppression of wings may have occurred multiple times. Kukalova-Peck (1978, 1987) has reported fossil Neopteran and Paleopteran specimens with wings or proto-wings on the prothorax and has argued for separate loss of this T1 appendage in the two lineages independently. Furthermore, the T1 wings of Paleopteran and Neopteran fossils are morphologically distinct from the T2 and T3 wings (Kukalova-Peck, 1978, 1987). We infer from the existence of dT1 expression of Scr in all insects examined that the last common ancestor of Thysanurans and all pterygotes, Paleoptera and Neoptera, also had dT1 expression of Scr. Therefore, the Scr expression predated the evolution of wings and was in the correct location, perhaps initially, to control the unique development of primitive prothoracic wings. It was subsequently recruited to suppress wing formation in both the Paleoptera and Neoptera. An examination of Scr expression in Paleopterans would help to confirm this hypothesis. At present the function of the dT1 Scr expression in firebrats is unknown. There are no obvious morphological structures specific to the prothoracic pleuron that distinguish it from the second and third thoracic segments (data not shown) that might shed light on any possible role of Scr in dT1 of apterygotes.

Scr is expressed in the tibia of the prothoracic legs of both crickets and milkweed bugs in a patch or hemi-stripe and we infer that this homologous expression existed in the common ancestor of Hemiptera and Orthoptera. Although in milkweed bugs this spot marks the position of the developing comb, no comb is produced on the prothoracic leg (or any other leg) in crickets and the Scr-expressing leg spot does not seem to correlate with any morphological structure unique to the prothoracic leg at the tibialtarsal junction (Fig. 3). Thus, the role of the leg spot in crickets remains unclear. However, because leg combs exist in Diptera, Hemiptera and Coleoptera, but not in Orthoptera or more primitive lineages, we propose that Scr expression in the prothoracic leg, like dorsal T1 expression, preceded the adoption of a novel patterning function. We do not have enough data points to infer precisely when leg spot expression evolved. It may have been coincident with the evolution of pterygotes (‘b’ Fig. 6), with only some pterygote lineages subsequently developing the new regulatory networks necessary to produce leg combs (‘c’ Fig. 6). Alternatively, the LT1 expression could have evolved several times in the different lineages. Parsimony argues for the former model, but a resolution of the point will require an examination of Scr expression in the Paleoptera and other more primitive pterygotes as well as additional apterygotes.

Our observations provide insight into how the activity of Scr provides segment identity. Most recent evidence from Drosophila supports the idea that HOM/Hox genes control cell-specific functions that depend upon the developmental history of each cell (Castelli-Gair and Akam, 1995). Unique segment identity is provided by a mosaic of cell types. However, an examination of HOM/Hox gene expression patterns in species of different phyla has revealed a striking conservation of their primary anterior to posterior domain of expression throughout the metazoa. The order of expression (with respect primarily to the anterior-most ectodermal expression domain) is generally collinear with the known order of the genes within the homeotic complexes of both chordates and insects (Garcia-Fernandez and Holland, 1994; Manak and Scott, 1994; McGinnis and Krumlauf, 1992). This provides further evidence that the HOM/Hox genes supply positional information and that this function is conserved.

The continuous requirement for HOM/Hox gene activity for certain cell fates (Morata and Garcia-Bellido, 1976) has been used as evidence that they function as a ‘genetic address’. However, the branching and irreversible nature of developmental pathways would require that the addresses only be checked early to set in motion pathways that would provide a segment-specific developmental history. In this model, positional information, or the ‘genetic address’, could be one of the earliest cell-specific functions of HOM/Hox genes that helps to set the unique developmental background upon which future cell-specific functions will depend.

The intense early expression of Scr in all the cells of the labial segment in all of the insects examined is consistent with it providing this early positional information. The expression covers all the cells of a single domain, the labial segment epidermis, and provides a common developmental background that by itself or in concert with other gene products could set all the cells of the labium apart from every other segment. This expression of Scr could immediately specify functions characteristic of this anterior-posterior position, such as labial migration, or initiate developmental pathways that could lead to the production of unique structures even in the later absence of the Scr protein, such as at the tip of the labium.

In contrast, the later expression of Scr in the thorax seems incompatible with providing positional information but instead is required to modulate developmental pathways common to the thorax. Once thoracic identity is established, Scr can suppress common features, such as the wing, or promote the production of novel features, such as combs, in a manner that is relatively independent of a cell’s recent history of HOM/Hox expression. The production of a unique thoracic identity by mosaic HOM/Hox expression is similar to the situation reported for the abdomen of Drosophila (Castelli-Gair and Akam, 1995). However, reports of HOM/Hox gene expression patterns suggest that the degree of mosaicism is much greater in Drosophila than in other insects. We have shown that with the exception of Drosophila, Scr is not expressed broadly over the prothorax, but is generally limited to one or two clusters of cells. This difference is similar to that observed by Hayward et al. (1995) who reported little of the modulation of grasshopper Antennapedia expression that is characteristic of the expression of this gene in the thorax of Drosophila. Also, in the grasshopper, abdominal-A is expressed much more uniformly over each abdominal segment than is Drosophila abdominal-A (Tear et al., 1990).

In our study of the Scr expression pattern in three non-dipteran insects, we have shown that although there are conserved elements of expression, including the anterior-most ectodermal domain, expansion of the expression domain posteriorly into the prothorax has directly affected the morphological evolution of insects by allowing specialization of unique prothoracic characters. It is important to establish the degree of variability of HOM/Hox gene expression in a wider group of insects and arthropods in general to determine the relative contribution of this type of diversity of expression on the evolution of segment identity and the arthropod body plan.

We would like to thank S. Brown and R. Denell for unpublished information concerning the expression pattern of Scr in Tribolium embryos; J. Kukalova-Peck, A. Popadic and D. Rusch for helpful discussions and analysis of the manuscript; G. Panganiban and L. Nagy for help with in situ hybridization protocols; D. Verostko, for administrative assistance; R. Turner for the S.E.M.s, and A. Kalkbrenner for help with engrailed antibody preparations and techniques. This work was supported by the Howard Hughes Medical Institute (HHMI). B. T. R. is a HHMI associate; M. D. P. is a HHMI pre-doctoral fellow; T. C. K. is a HHMI investigator.