Homologous patterns in the embryonic development of the peripheral nervous system in the grasshopper Schistocerca gregaria and the fly Drosophila melanogaster

Investigations on the fly Drosophila melanogaster have shown that the development of the peripheral nervous system is a multistep process, which includes several genetic levels acting on the undifferentiated embryonic ectoderm in a hierarchical manner (see Giangrande and Palka, 1990, Ghysen and Dambly-Chaudière, 1989; Jan and Jan, 1990). Once segment-specific positional information is determined (see Akam, 1987; Ingham, 1988), neurogenic and proneural genes generate sensory cell precursors at defined locations within the embryonic ectoderm of each segment (Hartenstein and Campos-Ortega, 1986; Dambly-Chaudière and Ghysen, 1987; Caudy et al. 1988). Thereafter other genes control the development of sensory neurons from these precursors (Bate, 1978; Bodmer et al. 1987, 1989; Blochhnger et al. 1988, 1990; Uemura et al. 1989). Finally, sensory neurons initiate axonogenesis and project their axons along stereotyped pathways that were initially established by pioneer neurons with the help of guidepost neurons (Bentley and Keshishian,1982; Ho and Goodman, 1982; Blair et al. 1987; Klose and Bentley, 1989). The overall result of segmentally reiterated neurogenesis and axonogenesis in the peripheral nervous system is a homologous pattern of sense organs and peripheral nerves in the segments of the thorax and abdomen (Jan and Jan, 1982; Campos-Ortega and Hartenstein, 1985; Ghysen et al. 1986; Bodmer and Jan, 1987; Hartenstein, 1988).

How general are these developmental mechanisms? The dipteran flies, to which Drosophila belongs, are among the most specialized and recently evolved insects (Chapman, 1982). Embryogenesis in Drosophila results in a highly specialized larva with similar thoracic and abdominal segments and no appendages. The larval peripheral nervous system is transient; most of the adult peripheral nervous system is derived from imaginal discs. In contrast, embryogenesis in more primitive insects such as orthopterans results in a larva which is a miniature replica of the adult. This type of larva has specialized thoracic segments which bear legs and wingbuds, and several specialized abdominal segments. In these insects, embryonic components of the peripheral nervous system are retained in the adult (Bate, 1978; Sehnal, 1985).

Given these and other basic differences in development (Sander, 1976), the embryogenesis of the peripheral nervous system in Drosophila might be radically different from that of other, evolutionarily less advanced insects. To investigate this, we have characterized the embryonic development of the peripheral nervous system in a more primitive insect, the grasshopper Schistocerca gregaria. Our results show that the overall pattern of peripheral nervous structures in the two insect embryos is remarkably similar. Homologous spatiotemporal sequences of nerve pathway formation and of sense organ neurogenesis can be documented in the two species, often at the level of single identified cells. Since the more primitive orthop-teran Schistocerca is separated from the highly specialized dipteran Drosophila by at least 300 million years of evolution, these findings indicate that fundamental mechanisms involved in the development of the peripheral nervous system in different insect embryos are highly conserved.

Schistocerca gregaria (Acrididae, Cyrtacanthacridinae) eggs were collected from a crowded laboratory culture and kept in moist aerated containers in an incubator at 30°C. Staging of the embryos of Schistocerca was performed according to the entena established by Bentley et al. (1979). The youngest embryos used were at 33 % of total embryonic development and the oldest were at 55 % of embryonic development At the 55 % stage most of the internal sense organs have formed. The external mechanoreceptors of the epidermis, such as the hair receptors and the campaniform sensilla, which start to differentiate in a subsequent wave of peripheral neurogenesis (Kutsch, 1989), have not been studied. For investigations of sex-specific neurogenesis, the criteria established by Roonwal (1937) for Locusta were found to apply to Schistocerca as well and were used to determine the sex of embryos.

Embryos were dissected out of the egg and embryonic membranes in a 4% formaldehyde solution in phosphate-buffered saline (PBS-FA) (pH=7.35) and fixed for 45 – 60 min. Either the whole embryo or isolated segments were then washed several times in PBS.

To follow the embryonic development of the peripheral nervous system, we used a neuron-specific anti-HRP antibody (Cappel) in whole-mount preparations and serial sections. This antibody has been shown to label the peripheral and central neurons along with their axons, dendrites and growth cones in developing Drosophila and grasshopper embryos by recognizing a specific set of membrane glycoproteins (Jan and Jan, 1982; Snow et al 1987).

For immunocytochemistry, whole-mount preparations were preincubated first in PBS with 0.2% bovine serum albumin (BSA) and 0.4 % Triton X-100 for 30 min and then in PBS with 5% normal goat serum (NGS), 0 2% BSA and 0.4% Triton for 60min at room temperature. The anti-HRP antibody was diluted 1:300 in PBS with 5 % NGS, 0.2 % BSA and 0 4% Triton and the preparations were incubated in diluted antibody overnight in the cold with agitation. When a fluorescein isothiocyanate (FITC)-conjugated anti-HRP (goat) antibody was used, the preparation was washed after antibody incubation for 3-6 h in PBS and then cleared Staining of 5 μm serial sections was earned out after rehydration and preincubation of the sections in PBS with 1 % NGS and 0 1 % Trton for 30 mm followed by incubation for 3h with the FITC-conjugated anti-HRP antibody diluted 1:300 in the same solution. To keep FITC-treated matenal from fading, whole mounts and sections were cleared and mounted in a mixture of glycerol and PBS (9:1) with p-phenylenediamine added (Johnson and Nogueira Araujo, 1981) and stored at – 20°C. For whole mounts in which an anti-HRP antibody (rabbit) with subsequent diaminobenzidine tetrahydrochlonde (DAB) staining was used, the preparations were incubated for several hours with a goat anti-rabbit antibody (Nordic) in a 1 · 20 dilution of PBS with 10% NGS, 0.2% BSA and 0.025% Triton. Then they were washed in PBS with 0.025 % Triton and incubated for 3 h together with a peroxidase anti-peroxidase complex (Nordic) at a concentration of 1:40 in a solution of PBS with 1 % NGS and 0.025 % Triton. After a final washing, the DAB reaction was performed under standard conditions (Bourne, 1983).

Immunostained matenal was viewed in a Zeiss UEM compound microscope equipped for epifluorescence and differential interference contrast For documentation, FITC-stained preparations were photographed using epifluorescence optics, and DAB-stained preparations were photographed using differential interference contrast optics. The film used was Ektachrome 200 or TMAX 100 (Kodak) Additionally drawings were made from DAB-stained matenal with a drawing tube.

For comparison, a brief description of the early development of the peripheral nervous system in the thoracic and abdominal segments of the Drosophila embryo shall be given here. These developmental processes have been characterized by Jan and Jan (1982), Campos-Ortega and Hartenstein (1985), Ghysen et al. (1986), Bodmer and Jan (1987), Hartenstein (1988), Johansen et al. (1989) and Ghysen and O’Kane (1989). The nomenclature used here is that of Campos-Ortega and Hartenstein (1985).

In the developing thoracic and abdominal ganglia, each of the peripheral nerves that exit a given hemiganglion is made up of an anterior and a posterior fascicle. The anterior fascicle is pioneered by efferent axons from the anteriorly adjacent neuromere and by afferent axons from sensory cells in the dorsal body wall of the same segment. Efferents and afferents make contact in the body wall and fasciculate. The posterior fascicle is also pioneered by efferent and afferent components. The efferent axons originate in the homotopic neuromere. They are contacted by afferent axons from a ventral cluster of sensory cells in the body wall. Thus, in the early embryo pioneer neurons generate a simple, serially reiterated nerve scaffold, that is used by subsequently developing peripheral sensory cells for axonal guidance.

The development of the sense organs in Drosophila results in patterns of sensory cells that are slightly different in the different thoracic and abdominal segments. However, in most segments, a dorsal, lateral and ventral cluster of developing sense organs can be identified. Homology among groups of sense organs has been established for the thoracic and abdominal segments.

Two pairs of axon fascicles are formed in each of the developing thoracic (T1 –3) and pregenital abdominal segments (A1 –7). The anterior fascicle forms the intersegmental nerve; the posterior fascicle forms the segmental nerve (Thomas et al. 1984).

The intersegmental nerves in most of the thoracic and pregenital abdominal segments are formed by efferent and afferent pioneers. The efferent pioneers are the U1 and U2 axons which together with the aCC, RP1 and RP2 axons form the U-fascicle. The U-fascicle establishes the proximal part of the intersegmental nerve (Goodman et al. 1984; Bastiani et al. 1986, du Lac et al. 1986).

As is shown in Fig. 1A, the distal part of the intersegmental nerve is pioneered by a group of peripheral cells, which we call the dorsal body wall (dBw) cells because they are located in the dorsal body wall region of each segment. In most cases, the dBw cell group consists of 6 cells, although in some cases 8 – 9 cells were observed (Fig. 1C,D). At approximately the same time as the U-fascicle leaves the CNS in a posterolateral direction (35 – 37% stage), the axons of the dBw cells start to grow and project ventrally along the segmental body wall. From the outset, these afferent axons are oriented toward the outgrowing U-fascicle. After having travelled approximately the same distance, the axons of the dBw cells meet those of the U-fascicle at the level of the segmental spiracle. The two axon bundles fasciculate and the growth cones of the two groups overgrow each other in opposite directions (Fig. 1B). Immunostained cells that might have guidepost function (Bentley and Keshishian, 1982; Ho and Goodman, 1982) were never observed in this process. Following fasciculation, the proximally growing dBw axons extend into the CNS, whereas the distally growing axons of the U-fascicle extend into the dorsal body wall region where they subsequently ramify. There is a rostrocaudal temporal gradient for nerve formation. The intersegmental nerve is established in the mesothoracic, metathoracic and anterior abdominal ganglia by the 40 % stage; it is established in the remaining abdominal segments by the 43 % stage. In a given segment, the total time between onset of axonogenesis in the efferent U-neurons and afferent dBw cells until they fasciculate is approximately 3 – 5 % of total embryonic development.

In the prothoracic and labial (posterior head) segments, the intersegmental nerve is pioneered exclusively by the efferent U-fascicle and dBw cells do not form. In the labial segment, cells located in a position that is more dorsal and anterior than the dBw cells in other segments were observed. However, these cells are generated after the intersegmental nerve has been established (after 40 % stage). An intersegmental nerve is not formed in the mandibular (first mouthpart) segment, the maxillary (second mouthpart) segment or the terminal abdominal segment. The development of the cereal nerve in the terminal abdominal segment has been studied by Shankland and Bentley (1983) and is not further considered here.

The development of the segmental nerves has been described in detail in the leg-bearing thoracic segments (Bentley and Keshishian, 1982; Ho and Goodman, 1982; Keshishian and Bentley, 1983a,b,c;Caudy and Bentley, 1986a,6; Myers et al. 1990; Whitington, 1989). In each of these segments, two major segmental nerve branches (N3 and N5) innervate each leg. The posterior nerve branch (N5) is pioneered by limb bud afferents that project via guidepost neurons into the CNS The anterior nerve branch (N3) is pioneered by afferents and efferents. Other sensory cells project their axons onto these two nerve branches and thus create secondary branches of the segmental nerves

In the mouthpart-bearing gnathal segments, segmental nerve branches are also formed. The major nerve branches are pioneered by afferent nèurons with the help of guidepost neurons; the minor nerve branches are pioneered by afferent neurons fasciculating with efferent neurons (Meier and Reichert, 1991). In contrast to the thoracic segmental nerves, however, the two segmental nerve branches in the gnathal segments do not exit the ganglion separately. Instead they both segregate from a single main segmental nerve near the ganglion edge.

The development of the segmental nerves in the abdominal segments involves the transient expression of small stump-like limb buds which start to develop at the 35 % stage and start to become reduced at the 55 % stage (Bentley et al. 1979). Although these transient structures never differentiate into limbs, they play a role in segmental nerve formation that is comparable to that of the other segmental appendages. In each of these abdominal limb buds, a single pioneer neuron, which we call the limb bud pioneer (LbP) cell, differentiates from the ectoderm and undergoes axonogenesis (Fig. 2A). This occurs at the 35-37% stage in the anterior abdominal segments and is somewhat more retarded in the posterior abdominal segments due to a general developmental gradient. The axon of the LbP cell projects the 50 – 70 microns into the adjacent neuromere without the help of guidepost cells (Fig. 2B). Subsequently, efferent axons from the neuromere leave the ganglion at the level of the anterior commissure and fasciculate with the afferent LbP axon. By this time the LbP cell has been displaced from the luminal face of the limb bud ectoderm into the lumen of the body wall. The efferent axons follow the LbP axon peripherally and extend into more distal parts of the body wall (Fig. 2C). The axon pathway that is pioneered by the LbP cell becomes the anterior fascicle of the segmental abdominal nerve. The morphology of the LbP cell along with its location in the ventral body wall and its pattern of innervation indicate that this cell becomes the stretch receptor of the ventral diaphragm (Hustert, 1974, 1975).

The posterior fascicle of the segmental abdominal nerve is of purely efferent origin and is pioneered approximately 1 % later than the anterior fascicle by a bundle of central neurons which grow dorsally and laterally along the body wall without orienting along guidepost cells. These two main nerve fascicles remain separated as independent branches until katatrepsis (50% stage). In later embryonic stages, the anterior and posterior fascicle of a given abdominal segment fuse at the edge of the ganglion to form the single adult segmental nerve (N2; Campbell, 1961; Hustert, 1974), thus masking the original bipartite origin of this nerve.

In contrast to the other abdominal segments, the genital segment A10 shows sex-specific variations in the development of the segmental nerves. In the A10 segment of male embryos, only one segmental nerve fascicle develops. We were not able to determine if this is the anterior or the posterior fascicle, because we could not identify a LbP cell in this segment. In female embryos neither of the two segmental nerve branches develops in A10.

Following the formation of a simple stereotyped scaffold of embryonic nerves, sensory neurons differentiate in a well-defined spatiotemporal pattern along the dorsoventral axis of the body. In most segments, this early differentiation of sensory neurons occurs in three distinct clusters, a dorsal, a lateral and a ventral cluster. The axons of the sensory cells in the dorsal and lateral clusters project onto the intersegmental nerve. In the abdominal segments, the axons of the sensory cells in the ventral cluster join the anterior fascicle of the segmental nerve. For a given segment, a temporal gradient of development results in more dorsal clusters developing earlier than more ventral clusters. An overview of these developing sensory cells for different segments is shown in Fig. 6.

In most of the thoracic and abdominal segments, the dorsal cluster derives from the dBw cell group. We have studied the differentiation of one of the cells in this group into an identified single cell sense organ. In the pterothoracic segments T2 and T3, a single cell begins to separate from the dBw cluster shortly after formation of the intersegmental nerve at the 42 % stage (Fig. 3A). Subsequently this cell becomes enlarged, sends out numerous filopodia in all directions and begins moving posteriorly. During the cell’s movement over several hundred microns, its axon remains connected to the intersegmental nerve. At the 45 % stage the cell reaches another developing sense organ, which will become the wing hinge chordotonal organ (Gettrup, 1962; Pearson et al 1989). After reaching a position near the posterior segment border, the cell stops moving and differentiates further. This cell will become the wing hinge stretch receptor of the adult animal (Gettrup, 1963, Heathcote, 1981; Pfau, 1983). The axon of the developing wing hinge stretch receptor helps guide the axons of the differentiating wing hinge chordotonal organ to the intersegmental nerve via the dBw cells (Fig. 3B).

In the abdominal segments Al – 10, a cell differen-, tiates at approximately the same time (42—45 % stage), in a similar way and at the segmentally equivalent position (Fig. 3C,D). It derives from the dBw cluster, enlarges, moves posteriorly, stops near the posterior segment boundary and remains connected with its axon to the intersegmental nerve. At the 50 % stage this cell, which we consider to be the segmental homolog of the wing hinge stretch receptor (‘SR’), remains connected to the dBw cells by fine filopodia-like processes. However, these processes soon begin to disappear, the result being that at the 55 % stage this cell and the dBw cell groups have separated almost entirely and have axons that form different distal branches in the intersegmental nerve (Fig. 4). The homolog of the wing hinge stretch receptor subsequently differentiates distal T-shaped dendrites that probably serve to register muscle tension during ventilatory movements (Hustert, 1974, 1975).

Although we were able to follow the development of the other remaining cells in the dBw cluster up to the 55 % stage, it was not possible to determine which sense organs these cells differentiate into. Neither dBw cells nor stretch receptor homologs develop in the labial and prothoracic segments or in the terminal abdominal segment All.

The lateral sensory cell clusters give rise to the wing hinge chordotonal organs in the pterothoracic segments, the auditory organs in abdominal segment Al and the pleural chordotonal organs in the abdominal segments A2-9. All three types of sense organs differentiate at similar developmental stages by epithelial invagination from the body wall ectoderm near the posterior segment boundary, migrate anteriorly towards the intersegmental nerve and project their axons onto a fascicle of the intersegmental nerve and via this fascicle into the CNS (Meier and Reichert, 1990). The cells of the developing auditory and pleural chordotonal organs project their axons directly onto the intersegmental nerve without the help of guidepost cells (Figs 5 and 6). In contrast, the axons of the developing wing hinge chordotonal organ first contact the wing hinge stretch receptor cell and from there project to the dBw cell group before fasciculating with the intersegmental nerve (Figs 3B and 6A).

In the abdominal segments, a second group of sensory cells also develops in the lateral cluster and projects its axons onto the intersegmental nerve (Figs 6B,C, asterisks). We were not able to determine which sense organs these cells differentiate into and thus did not study these cells further. A lateral cluster was not observed in the labial or prothoracic segments or in abdominal segment A10 even though intersegmental nerves are formed in these segments.

In the thoracic segments Tl – 3 and the abdominal segments Al – 9, a group of sense organs develops in the ventral part of the body wall (Figs 5, 6). In the abdominal segments, these sense organs of the ventral cluster, which can be further divided into a distal and a proximal subgroup, project their axons onto the anterior fascicle of the segmental nerve. The distal subgroup develops dorsal to the LbP cell at the base of the transient abdominal limb bud at the 43 – 45 % stage.

It is composed of approximately six neurons and probably forms the sternal chordotonal organ of the adult (Hustert, 1974). (Since the segmental limb bud homolog in abdominal segment Al, the pleuropodium, is located more dorsally than the other limb buds, the sternal chordotonal organ as well as the other sense organs of the ventral cluster in this segment are situated slightly more dorsally than in the other abdominal segments.) In a given abdominal segment, the proximal subgroup develops ventral to the LbP cell near the lateral ganglionic border slightly later than the distal subgroup. It is usually composed of approximately six cells, but in some cases the formation of an adjacent second set of approximately four cells was also observed. On the basis of position and innervation, we assume that the proximal cell group differentiates into the chordotonal-organ-like sensory cell group of nerve N2n2 of the adult (Hustert, 1974), which we refer to as the ventral sensory organ. The proximal subgroup does not develop in abdominal segment A9. The entire ventral cluster fails to develop in abdominal segment A10.

In the thoracic segments, groups of sensory cells also develop in the ventral part of the body wall. However, these cells project their axons onto the intersegmental nerve. (The segmental nerves of the thoracic segments innervate the legs and not the body wall.) In the prothoracic segment, this ventral group of cells differentiates into a single chordotonal organ. In the meso- and metathoracic segments, two chordotonal organs are formed (Fig. 6A). The functional role of these thoracic chordotonal organs in the adult, as well as possible homologies with other segmental sense organs, remain obscure.

This paper describes the spatiotemporal sequence of neurogenesis and nerve pathway formation in the grasshopper peripheral nervous system at the cellular level. Two remarkable findings emerge from our investigations. The first is the degree to which serial homology is preserved during the early neuronal development of the peripheral nervous system in the different segments of the grasshopper. The second is the evolutionary conservation that is expressed in the generation of the embryonic peripheral nervous system in insects as divergent as a grasshopper and a fly. In the following, we shall discuss some of the implications of these findings.

From an evolutionary standpoint, serial homology is a reflection of the homonymous segmentation of primitive forms, in which comparable structures were present in most body segments (Dobzhansky et al. 1977). In the peripheral nervous system of postembryonic animals, segmental homology is often obscured by differentiation phenomena such as fusion, fragmentation and shift of position (Matsuda, 1976). However, segmental homologies are in many cases revealed clearly by studies of embryonic development.

Serial homology in the embryonic peripheral nervous system of the grasshopper is a case in point. On the basis of classical morphological criteria for homology such as position, innervation and structural organization as well as developmental criteria such as mode of neurogenesis, relative timing of differentiation and trajectory of cell migration and axonal navigation, we have been able to identify serially homologous peripheral nerves and sensory cells.

Clearly, there are distinct patterns in the developing peripheral nervous system that are different in the gnathal and, to a lesser degree, in the thoracic segments as compared to the abdominal segments. Also, the genital and terminal abdominal segments have segment- or sex-specific features in their peripheral nervous organization. Nevertheless, the extent of overall serial homology in the peripheral nervous system is striking. In most embryonic segments, an intersegmental and a segmental nerve branch can be identified. The intersegmental nerve in segments T2 through A10 is established by serially homologous central and peripheral pioneers. The formation of the originally bipartite segmental nerve in virtually all segments involves afferent pioneers located in limb buds, irrespective of whether these limb buds later form mouthparts, legs or are only transiently expressed as in the abdominal segments. The diversity of sense organs in the different segments derives from a basic, serially reiterated framework of dorsal, lateral and ventral clusters. The sensory structures in each cluster share similar mechanisms of neurogenesis, morphogenesis and axonogenesis. In several cases, the same is true for individual cells within a given cluster such as for the thoracic wing hinge stretch receptor and its abdominal homologs. Within a given cell cluster, serially homologous sense organs differentiate to perform surprisingly different functions in behaviors as diverse as flight, hearing and respiration.

The remarkable degree to which serial homology is preserved in the segments of the grasshopper embryo may indicate that the formation of the peripheral nervous system relies on evolutionarily conserved developmental mechanisms (Thomas et al. 1984; Tear et al. 1988). Such mechanisms would reflect the origin of modem insects from ancestral mynapod-like animals in which the general body plan was that of very similar metameric repeats (Raff and Kaufmann, 1983). If this is so, one might predict that similar patterns of serial homology should occur in the development of the peripheral nervous system in other insects, even in those that are separated from the grasshopper by hundreds of millions of years of divergent evolution. This is indeed the case.

The same criteria that allow us to postulate serial homology in the grasshopper embryo can also be used to investigate interspecies homologies between the grasshopper Schistocerca and the fly Drosophila. Although we cannot identify the Drosophila homologs of all of the nerves and neuron clusters in the embryonic peripheral nervous system of the grasshopper, we can, on the basis of the work described in this paper on Schistocerca and in other reports on Drosophila, postulate the existence of a number of homologous structures in the two insects. A summary diagram of these postulated interspecies homologies for thoracic and abdominal segments is shown in Fig. 7.

The basic pattern of peripheral nerves in the body wall of Drosophila and grasshopper embryos is extremely similar. Peripheral nerve pathways corresponding to an intersegmental nerve and a segmental nerve are found in most of the segments of both species and in many cases homologies among the individual cells that pioneer these nerves can be established. For example, in both animals the intersegmental nerve is pioneered by the efferent axons of the U-fascicle (Thomas et al. 1984) and by the afferent axons of a dorsal cell cluster, the dBw cells in Schistocerca and the dc/dh cells in Drosophila (Hartenstein and Campos-Ortega, 1986; Ghysen et al. 1986; Campos-Ortega and Hartenstein, 1985). Likewise, in both animals, the formation of the segmental nerves in the abdominal segments involves identified pioneer neurons of a ventral cell cluster, the LbP cells in the abdominal limb buds of Schistocerca and the vcl/vc2 cells in Drosophila (Campos-Ortega and Hartenstein, 1985; Hartenstein, 1988). Interestingly, the vcl/vc2 cells are thought to be serially homologous to the Keilin organs, which are themselves sensory organs associated with the leg remnants in the thoracic segments of the Drosophila larva (Ghysen and Dambly-Chaudière, 1990).

Obvious similarities also exist in the development of the peripheral sense organs in the homologous segments of both species. Similar dorsal, lateral and ventral cell clusters, similar axonal projection patterns and similar dorsoventral temporal gradients of differentiation are found in both Schistocerca and Drosophila. For example, both animals have comparable complex abdominal chordotonal organs that derive from the lateral cluster, are located near the posterior segmental border, and project their axons directly onto the intersegmental nerve. These are the lch5 chordotonal organs in Drosophila and the pleural chordotonal organs and auditory organs in Schistocerca. Likewise, in both animals, a single sensory cell in the dorsal cluster of the thoracic segments has the dual developmental function of co-pioneenng the intersegmental nerve and of guiding the afferent axons of an adjacent chordotonal organ to the intersegmental nerve. This cell is the wing hinge stretch receptor in Schistocerca and is thought to be the dhl cell in Drosophila (Hartenstein, 1988). Moreover, both cells, which we postulate to be interspecific homologs, have a similar set of serial homologs in the abdominal segments.

In some cases, the interspecific homologies between grasshopper and fly can even help uncover hidden intraspecific serial homologies that might otherwise have remained obscure. For example, in Schistocerca, a number of cellular and molecular critieria indicate that the thoracic homolog of the abdominal pleural chordotonal organs and auditory organ is the wing hinge chordotonal organ (Meier and Reichert, 1990). If we postulate a similar serial homology in Drosophila, the thoracic homolog of the abdominal lch5 chordotonal organs should be the dch3 chordotonal organ. Two gross morphological features are at odds with this hypothesis. Namely the fact that the dch3 chordotonal organ is associated with the dorsal and not with the lateral sensory cluster and the fact that the dch3 chordotonal organ has 3 sensory neurons instead of 5 like the lch5 chordotonal organ However, on closer inspection a number of developmental criteria for homology such as precursor identity, lineage, timing of differentiation and axon pathway trajectory support the hypothesis of lch5 and dch3 homology (Campos-Ortega and Hartenstein, 1985; Hartenstein, 1988; Ghysen and O’Kane, 1989). Moreover, there are genetic criteria that also support the proposed homology. For example, a mutation in the gene rhomboid transforms the 5-cell chordotonal organ lch5 in the abdominal segments into a 3-cell chordotonal organ (Bier et al. 1990). Additionally, a number of mutations that result in parasegmental transformations are in accordance with the proposed homology. A mutation in the engrailed locus deletes the dch3 and the lch5 chordotonal organs without affecting other sensory cells in the lateral cluster (Hartenstein, 1987). In mutants lacking the abd-A gene, which is part of the segment-identity controlling bithorax complex, the 5-cell chordotonal organ in the lateral cluster of abdominal segments Al – 7 is replaced by a 3-cell chordotonal organ in the dorsal cluster of the same segments (Pfeifer et al. 1987; Duncan, 1987; Karch et al. 1990). Considering the summed evidence, we postulate that dch3 and lch5 in Drosophila are serial homologs and that their interspecies homologs in Schistocerca are the wing hinge chordotonal organs and the pleural chordotonal organs and auditory organs. (In Fig. 7 we accordingly attribute the dch3 cells to the lateral cluster.) Thus, the predictions on serial homology in Drosophila, which were obtained on the basis of comparative structural and developmental data in Schistocerca, are supported by a large body of developmental and genetic data obtained using Drosophila wild type and mutants. Indeed, a combined comparative developmental and genetic analysis of this type may give us a more profound insight into the way that the peripheral nervous system in insects evolved.

On the basis of the extensive interspecies homologies described here, we conclude that the construction of the embryonic peripheral nervous system, even in a specialized holometabolous insect like Drosophila, relies on conserved developmental mechanisms that were already operative in ancestral insect forms over 300 million years ago. Moreover, we postulate that the development of the peripheral nervous system in the highly derived fly larva is actually based on mechanisms that much more primitive hemimetabolous insects use to generate their adult peripheral nervous system. Finally, the obvious interspecies homologies that we describe here, taken together with the results of other comparative studies (e.g. Thomas et al. 1984; Patel et al. 1989; Tear et al. 1990), add further support to the hypothesis that the molecular genetic control systems that operate in the Drosophila embryo also direct the development of the nervous system in other insects.

We wish to thank Marianne Fnedli for excellent technical support and Alain Ghysen, Francois Karch and Hugh Rowell for helpful comments on the manuscript This work was supported by the Swiss NSF.