This paper reviews recent experiments which attempt to gain more understanding about the recognition processes involved in the formation of neuronal connexions by studying the degree of specificity with which sensory neurons form their central connexions. This is done by generating ectopic neurons (either by transplantation or by genetic mutation) whose axons grow into novel regions of the central nervous system, and then examining their projections and synapses.

The sensory systems reviewed are: the Antennapedia, spineless-aristapedia, proboscipedia, and bithorax homeotic mutants of Drosphila melanogaster; the cercus-to-giant interneuron system of crickets, and the wind-sensitive hair system of locusts.

The results show that ectopic neurons form projections that are discrete and characteristic, not random and chaotic. In those cases where single classes of sensilla have been studied, they follow either their normal CNS pathways or those pathways normally used by their segmental homologues.

Ectopic sensory neurons can also form appropriate functional connexions in some cases but not in others. Possible reasons are discussed, but detailed understanding of the underlying events requires further experimentation.

The nervous system is an outstanding example of spatial organization, from the level of its individual neurons each with their appropriate biochemical properties and characteristic morphologies, to the level of complex networks of many neurons with their precise patterns of synaptic connexions.

In insects the basic framework of the central nervous system (CNS) is laid down in the embryo to form a fully functioning nervous system at the time of hatching. Sensory neurons continue to be added during postembryonic life and must navigate through this complex framework to appropriate regions of the CNS and there form synapses with appropriate target neurons.

In the establishment of these precise patterns of sensory connexion it is thought that specific recognition of pre-established pathways and targets is very important. One way to discover more about these recognition processes is to define the degree of specificity shown by sensory neurons for growing along particular pathways and for synapsing with particular central neurons. This may be done by generating ectopic neurons whose axons grow into novel regions of the CNS, and examining their resulting projections and synapses.

Recent experiments using this approach have employed extracellular and intracellular electrophysiological recording techniques, and have benefitted enormously from the development of intracellular staining techniques using horseradish peroxidase (HRP) and cobalt; the projections of groups of neurons and individual neurons can now be observed anatomically at the level of fine tertiary branches within wholemount preparations and sectioned material. In this way it has been possible to map with precision those pathways within the CNS that can be followed by different classes of sensory neurons. Three experimental systems have been studied at this new level of resolution: the homeotic mutants of Drosphila, the cercus-to-giant interneuron system of crickets, and the wind-sensitive hairs of locusts.

In insects, a sensory receptor (a sensillum) consists of a group of cells produced by the mitotic divisions of epidermal cells. Within the group are cells which secrete the external cuticular structures of the sensillum and one or more sensory neurons with axons extending from the periphery into the CNS. The ectopic neurons examined in these studies were generated in one of two ways. The first is by surgically grafting pieces of epidermis (either before or after the sensilla have developed) from one location to another. This method has been used in the locust Schistocerca gregaria and the cricket Acheta domesticas. An alternative method for the fruitfly Drosophila melanogaster is the use of homeotic mutants. In these mutants one set of body structures is replaced by another set, e.g. in the mutant Antennapedia imesothoracic leg structures develop in place of antennal structures. Thus sensory neurons on the body part in question are ‘genetically grafted’ to an ectopic site.

This review will consider in turn each of the experimental systems from the perspective of recently published work or unpublished observations.

In the homeotic mutant Antennapedia (Antp), the antenna is replaced by a mesothoracic leg, whilst in spineless-aristapedia (ssa), the distalmost segment of the antenna, the arista, and parts of the third antennal segment, are replaced by the distalmost segment of a leg, the tarsus. The location of these body parts is shown in Fig. 1. The arista contains several sensory neurons (Stocker & Lawrence, 1981) but the sensilla from which they originate have not yet been described. Scanning electron micrographs show that the normal tarsus bears contact chemoreceptors (also called taste receptors) and mechanoreceptive bristles (Green, 1979). The homeotic appendages bear several types of sensillum including the contact chemoreceptors and mechanoreceptors characteristic of the tarsus (Green, 1979) and they have a nerve which is characteristic of the mesothoracic leg nerve in number and arrangement of axons (Stocker, 1979).

Fig. 1

Diagram of Drosophila melanogaster to show the location of sensilla studied in homeotic mutants. Large identifiable bristles (macrochaetes) on the mesothoracic notum are individually drawn, for further details see Ghysen (1980). The notum is also covered with many small bristles (microchaetes) which are not shown. Small bristles are shown on the haltere end-knob. Large black dots on the wing veins represent large campaniform sensilla. Small dots on the proximal wing and haltere stalk represent small campaniform sensilla.

Fig. 1

Diagram of Drosophila melanogaster to show the location of sensilla studied in homeotic mutants. Large identifiable bristles (macrochaetes) on the mesothoracic notum are individually drawn, for further details see Ghysen (1980). The notum is also covered with many small bristles (microchaetes) which are not shown. Small bristles are shown on the haltere end-knob. Large black dots on the wing veins represent large campaniform sensilla. Small dots on the proximal wing and haltere stalk represent small campaniform sensilla.

Deak (1976) and Stocker (1977) have shown that stimulation with sugar solutions of the chemoreceptors on the homeotic legs regularly elicits the proboscis extension reflex, a behaviour normally elicited by stimulation of thoracic legs but not of antennae. Stimulation of a single sensory bristle is sufficient to elicit this reflex in other flies (Dethier, 1955), so a cautious conclusion is that in the mutants at least one, but not necessarily all or many, of the homeotic receptors form functional connexions within this reflex pathway.

With Antp flies, degeneration staining revealed that this response was not achieved by the growth of neurons from the homeotic leg to the thoracic ganglion where normal leg neurons project. In fact the homeotic neurons projected only to the antennal centres of the brain (Stocker, Edwards, Palka & Schubiger, 1976). The projection within the antennal glomeruli was somewhat different from that of a normal antenna: it was restricted to an area near the site of entry of the antennal nerve into the ipsilateral antennal glomerulus, whereas the normal antennal projection spreads to the periphery of the antennal glomeruli on both ipsilateral and contralateral sides of the brain.

A more detailed analysis has now been made for ssa flies using cobalt filling to compare the projections from homeotic tarsi, normal tarsi and normal aristae (Stocker & Lawrence, 1981). The results are shown in Fig. 2. As with Antp, the homeotic ssa tarsi never project to the thoracic ganglion areas used by normal tarsal neurons. They do project to normal antennal areas: ipsilateral and contralateral antennal glomeruli (but with a different pattern of terminals), ipsilateral posterior antennal centre and suboesophageal ganglion, and the ventrolateral protocerebrum. They also form a novel tract passing from the ipsilateral antennal glomerulus to the proboscis centre of the suboesophageal ganglion (Fig. 2 c). Since the motor neurons innervating the proboscis are also located here (Stocker & Schorderet, 1981), it seems likely that it is this part of the homeotic projection which activates the proboscis extension reflex. However, in Antp flies, which also show this reflex, a projection was not detected in this region.

Fig. 2

Sensory projections from normal, ssa and pbts homeotic appendages in D. melanogaster. (a) Schematic diagram of the fly brain and thoracic ganglion showing the centres to which the normal arista and the tarsi of the three thoracic legs project. (b-f) Schematic diagrams of the fly brain showing the projections from: (b) the normal third antennal segment, (c) the ssa homeotic tarsus, (d) the pbt3 homeotic arista, (e) the pbt3 homeotic tarsus, (f) the normal proboscis. AR, arista; AG, antennal glomerulus; PAC, posterior antennal centre; SG, suboesophageal ganglion; N1, N2, N3, prothoracic, mesothoracic, and metathoracic neuromeres respectively of the thoracic ganglion. T1, T2, T3, prothoracic, mesothoracic and metathoracic tarsi respectively; 3AS, third antennal segment; VLP, ventrolateral protocerebrum; PC, proboscis centre of the suboesophageal ganglion (after Stocker & Lawrence, 1981; Stocker & Schorderet, 1981; Stocker, 1981).

Fig. 2

Sensory projections from normal, ssa and pbts homeotic appendages in D. melanogaster. (a) Schematic diagram of the fly brain and thoracic ganglion showing the centres to which the normal arista and the tarsi of the three thoracic legs project. (b-f) Schematic diagrams of the fly brain showing the projections from: (b) the normal third antennal segment, (c) the ssa homeotic tarsus, (d) the pbt3 homeotic arista, (e) the pbt3 homeotic tarsus, (f) the normal proboscis. AR, arista; AG, antennal glomerulus; PAC, posterior antennal centre; SG, suboesophageal ganglion; N1, N2, N3, prothoracic, mesothoracic, and metathoracic neuromeres respectively of the thoracic ganglion. T1, T2, T3, prothoracic, mesothoracic and metathoracic tarsi respectively; 3AS, third antennal segment; VLP, ventrolateral protocerebrum; PC, proboscis centre of the suboesophageal ganglion (after Stocker & Lawrence, 1981; Stocker & Schorderet, 1981; Stocker, 1981).

In proboscipedia (pbts) flies, the proboscis is transformed into a tarsus if the flies are reared at 29°C or into an arista if reared at 17°C. The normal proboscis bears several taste bristles which can be selectively filled with cobalt. The bristles probably have a mixed mechanosensory and chemosensory function; they have the appearance of trichoid sensilla, possess terminal pores characteristic of chemoreceptors, and are innervated by three to five neurons (Stocker & Schorderet, 1981). The sensilla on the homeotic pbts tarsus and arista have not been described in detail. The projections from these appendages have been investigated by Stocker (1981), and are shown in Fig. 2.

Sensory axons from the normal proboscis project through the labial nerve to a specific region, the ‘proboscis centre’ of the suboesophageal ganglion (Stocker & Schorderet, 1981). Axons from homeotic tarsi and homeotic aristae that replace the proboscis may take the labial nerve and/or certain other nerves to the CNS. The central projection pattern is generally not affected by the site of entrance into the brain, although central tracts may be followed in opposite directions. Fibres reach the proboscis centre but do not go to that part of the suboesophageal ganglion normally innervated by the arista, nor to the tarsal regions of the thoracic ganglion. In this respect the neurons from homeotic tarsi and aristae behave like proboscis neurons. However, unlike proboscis neurons they also project into the antennal glomerulus where they terminate in a pattern not like that of a normal arista, but like that of a homeotic ssa tarsus. It would be interesting to know whether these various appendages have some sensilla in common and if these form the common elements of the projections.

Several studies on the bithorax mutants have examined the projections from identified receptors into different neuromeres of the thoracic ganglion (Ghysen, 1978; Palka, Lawrence & Hart, 1979; Ghysen, 1980; Ghysen & Janson, 1980; Palka & Schubiger, 1980; Strausfeld & Singh, 1980).

In normal flies the mesothoracic notum bears a characteristic pattern of innervated bristles, and the mesothoracic dorsal appendage - the wing - bears 4 – 5 large campaniform sensilla on the third vein, several rows of bristles on the margin of the wing, and many small campaniform sensilla near the base of the wing (Fig. 1). The metathoracic notum is very small and devoid of bristles. The metathoracic dorsal appendage - the haltere - has no large campaniform sensilla, but does have scattered bristles on the end-knob and many small campaniform sensilla on the stalk (Fig. 1). The end-knob and stalk are considered to be homologous, respectively, to the blade and the base of the wing (Ouweneel, 1973; Morata, 1975; Morata & Garcia-Bellido, 1976). Each class of sensillum has a characteristic projection within the CNS (Ghysen, 1978; Palka et al. 1979), those of receptors in homologous regions being very similar (Ghysen, 1978,1980).

The homeotic mutant bithorax postbithorax (bx3pbx) has metathoracic structures transformed into mesothoracic structures. The transformation is not always complete and may result in sensilla which appear untransformed or intermediate between wing and haltere type (Palka et al. 1979). The sensilla on the homeotic notum are indistinguishable from those on the normal notum (Ghysen, 1978). On the homeotic wing, the marginal bristles and large campaniform sensilla look wing-like, but the small campaniform sensilla show a range of morphologies from wing-like to haltere-like (Cole & Palka, 1980; Palka & Schubiger, 1980).

Projections from specific bristles on the normal mesothorax and homeotic mesothorax of bx3pbx flies were studied by filling with HRP (Ghysen, 1978). The projections from the two pairs of scutellar bristles of the notum, from the many microchaetes covering the notum, and from the large campaniform sensilla of the third vein of the wing, all showed the same result: although the neurons entered the CNS posteriorly through a metathoracic nerve, they formed projections like those of normal mesothoracic receptors developing in situ and entering the CNS more anteriorly through mesothoracic nerves. Fig. 3 a illustrates this point for the large campaniform sensilla projection. Ghysen proposed that there are specific ‘trails’ within the CNS which may be recognized by appropriate axons at any point of encounter and then followed in either direction (Ghysen, 1978).

Fig. 3

Diagrams of thoracic ganglia of bx3pbx flies showing the ventral (a, c) and dorsal (b) sensory projections from: (a) large campaniform sensilla of the normal wing (open arrow) and homeotic wing (closed arrow); (b) small campaniform sensilla of the normal wing (open arrow), homeotic wing (closed arrow) and normal haltere (double arrow); (c) marginal bristles of the normal wing (open arrow) and homeotic wing (closed arrow). N1, N2, N3, prothoracic, mesothoracic and metathoracic neuromeres respectively of the thoracic ganglion (after Palka et al. 1979; Ghysen & Jansen, 1980).

Fig. 3

Diagrams of thoracic ganglia of bx3pbx flies showing the ventral (a, c) and dorsal (b) sensory projections from: (a) large campaniform sensilla of the normal wing (open arrow) and homeotic wing (closed arrow); (b) small campaniform sensilla of the normal wing (open arrow), homeotic wing (closed arrow) and normal haltere (double arrow); (c) marginal bristles of the normal wing (open arrow) and homeotic wing (closed arrow). N1, N2, N3, prothoracic, mesothoracic and metathoracic neuromeres respectively of the thoracic ganglion (after Palka et al. 1979; Ghysen & Jansen, 1980).

Palka et al. (1979) did not fill specific receptors in the mutants but filled entire projections from the homeotic wings, thus including projections from the marginal bristles and small campaniform sensilla, i.e. those sensilla which may have homologues on the metathoracic haltere. The projections from the homeotic wings had some elements that were unlike anything found in normal projections of wing or haltere but also had three prominent and consistently recognizable components. One component looked just like that formed by the large campaniform sensilla of the wing. Palka et al. (1979) concluded, as did Ghysen from his specific fills of large campaniform sensilla, that these receptors can form a normal projection within the CNS even when they enter the CNS at an abnormal position.

The second component was reminiscent of that produced by the marginal bristles of the normal wing, although instead of being in the mesothoracic segment of the CNS, it was in the metathoracic segment (Fig. 3 c). Specific fills of the homeotic wing bristles have since confirmed the origin of this projection (M. Schubiger, unpublished observations). Since these bristles have been considered homologous to those on the end-knob of the haltere, and they project to the same area of the metathoracic neuromere as do end-knob bristles (Ghysen, 1978), it is difficult to define to what extent this projection is ‘normal’ or ‘abnormal’. These neurons may be able to recognize homologous areas in the mesothoracic and metathoracic neuromeres. An alternative interpretation is that the bristle neurons are not fully transformed (the degree of transformation of the externally visible cuticular structures of the sensillum and its underlying neuron need not be the same). Evidence against this interpretation is provided by the projections from flies also bearing the mutation wingless (wg) (Palka & Schubiger, 1980). This mutation results in the replacement of the wings and halteres with mirror-image duplicates of the mesonotum and metanotum respectively. The mutation is expressed variably so that wg flies may either be normal in appearance, have a duplicate mesonotum, a duplicate metanotum, or both. An additional effect is that in some flies with an apparently normal haltere, the haltere nerve is misrouted so that it joins the ganglion at the level of the mesothoracic neuromere at a place appropriate for the wing nerve. When the haltere is replaced by a homeotic wing in wg/wg; bx3pbx/Ubx flies, the homeotic nerve may also be misrouted. The projections from marginal bristles on homeotic wings, when misrouted into the mesothoracic nerve, terminate in the mesothoracic neuromere just as do those from normal wings, suggesting that the homeotic transformation is complete.

The third component of the homeotic projection looked like that of the small campaniform sensilla of the haltere (Fig. 3 b). In this case homeotic wing sensilla appeared to be forming a haltere-like projection rather than projecting to areas appropriate to normal wing. This projection has now been further traced into the suboesophageal ganglion of the brain, where the projections of small campaniform sensilla from the wing and haltere can be more readily distinguished than in the thoracic ganglion, and confirmed to be haltere-like (Ghysen, unpublished observations). One interpretation is that the small campaniform sensilla of the wing can also recognize homologous haltere pathways. It also seems plausible that these projections arise from the incompletely transformed sensilla observed in the scanning electron micrographs. Evidence supporting this view is that in wg flies when these fibres are rerouted into the mesothoracic neuromere, they do not form a wing-like projection (Palka & Schubiger, 1980).

An important point, discussed by all the authors of papers using Drosophila homeotic mutants, is the possibility that the mutations also affect the CNS. This would complicate the interpretations considerably. The metathoracic neuromere of bithorax flies is usually enlarged but most authors agree that the internal organization of mutant thoracic ganglia appears normal when viewed in wholemount preparations, although this offers only a low level of resolution.

Better evidence has been obtained from the use of genetic mosaics (Palka et al. 1979; Stocker & Lawrence, 1981; Stocker, 1981). In mosaic flies only a small patch of sensilla-bearing cuticle is homozygous for the mutation, the CNS remaining heterozygous and thus phenotypically normal. Palka et al. (1979) found that the projections from mosaic bithorax appendages differed slightly from those of wholly mutant flies, but not in any important details. Stocker & Lawrence (1981) and Stocker (1981) found no significant difference between mosaic and mutant flies in their studies of the ssa and pbts mutants.

Strausfeld & Singh (1980) examined the projections of sensory antennal fibres and certain ascending and descending interneurons in normal and mutant flies. Neurons that normally terminate in the mesothoracic neuromere did not project further to the metathoracic neuromere in bithorax mutants, nor was there any duplication in the metathoracic neuromere of neurons which normally develop only in the mesothoracic neuromere. They suggested that the enlargement of the metathoracic neuromere in bithorax flies is in part due to increased sensory input from the homeotic wing, and in part due to the corresponding increase in branching of interneurons which they observed in this region.

Green (1980) backfilled thoracic leg motor neurons with HRP. The arrangement of motor neurons in the mesothoracic and metathoracic neuromeres was quite normal in bx3pbx/Ubx flies, suggesting no direct genetic effect on the CNS. However, flies with another genotype, abx bx3pbx/Ubx, sometimes showed a strikingly different result: motor neurons in the metathoracic neuromere were now arranged like those of the mesothoracic neuromere. If this change indeed arises from a direct genetic effect (and this is being further tested in mosaic flies - S. H. Green, personal communication), it provides a powerful tool for examining the determination and development of the CNS.

Crickets possess a pair of large posterior appendages - the cerci - that bear hundreds of sensory filiform hairs. The hairs are highly sensitive to air movements and their activity triggers the animals’ escape response. The sensory neurons project to the terminal abdominal ganglion where they make synapses with several giant ascending interneurons. The regeneration of the cercus-to-giant interneuron system has been studied after transplantation of the cercus to the mesothorax (Edwards & Sahota, 1967).

Cerci were transplanted to the stump of a severed mesothoracic leg where they regenerated their sensory projections to the CNS. Silver-stained sections of the mesothoracic ganglion showed the neurons entering the CNS via the leg nerve (nerve 5) and passing in the vicinity of a major bundle of neurons that included the giant interneurons. Stimulation of the grafted cercus with air puffs produced a burst of activity, recorded extracellularly, in the anterior mesothoracic connective which was characteristic of giant interneuron activity. This activity was observed in the presence of normal posterior cerci and also after isolation of the thoracic ganglia, i.e. in the absence of the normal input region in the terminal ganglion. These results suggested that cereal afferents can recognize a target neuron at a point other than the normal region of input and this can occur even when normal cereal inputs are present (Edwards & Sahota, 1967).

The head of the locust bears a population of wind-sensitive hairs whose inputs contribute to the initiation, maintenance, and control of flight. The hairs are grouped into five fields (Fl – F5) on each side of the head (Fig. 4a). Neurons associated with hairs in each of the fields form characteristic projections within the CNS as a function of their position on the head (Tyrer, Bacon & Davies, 1979). Examples of projections from F1 and F 2 on top of the head and F 3 on the side of the head are shown in Figs. 4 (b, c); F1 and F 2 neurons form distinctive contralateral branches in the suboesophageal ganglion and prothoracic ganglion, while F 3 neurons form only an ipsilateral projection in these ganglia.

Fig. 4

Neuronal projections in the locust, (a) Diagram of the locust head to show the location of the five fields of head-hairs, F1-F 5. (b) Sensory projections from F1 and F 2 head-hairs, (c) Sensory projections from F 3 head-hairs, (d) Projection of the TCG interneuron. BR, brain; SG, suboesophageal ganglion; Pro, prothoracic ganglion; Meso, mesothoracic ganglion; Meta, metathoracic ganglion.

Fig. 4

Neuronal projections in the locust, (a) Diagram of the locust head to show the location of the five fields of head-hairs, F1-F 5. (b) Sensory projections from F1 and F 2 head-hairs, (c) Sensory projections from F 3 head-hairs, (d) Projection of the TCG interneuron. BR, brain; SG, suboesophageal ganglion; Pro, prothoracic ganglion; Meso, mesothoracic ganglion; Meta, metathoracic ganglion.

Experiments in which epidermis was transplanted between different positions on the head showed that the type of anatomical projection formed was determined by the origin of the graft epidermis and not by the location of the hairs on the head at the time of their differentiation (Anderson & Bacon, 1979).

The wind-sensitive neurons make connexions with many interneurons in the CNS. One interneuron, the Tritocerebral Commissure Giant (TCG) has been studied in some detail. Its morphology is shown in Fig. 4(d). Extracellular recording of the activity of this neuron has shown that, in the brain, it receives excitatory input from all ipsilateral fields except F 3 (Bacon & Tyrer, 1978). A more detailed intracellular study has now shown that inputs from F 3 hairs strongly inhibit the TCG (Bacon & Anderson, 1981).

Exchange of epidermis between F 2 and F 3 has shown that the polarity of the connexions formed between wind-sensitive hairs and the TCG also depends upon the origin of the epidermis and not the final location of the differentiated hairs on the head (Bacon & Anderson, 1981).

To investigate further the degree of specificity with which wind-sensitive hairs form particular projections and connexions, similar epidermal grafts were made to other body segments (the posterior head, prothorax, mesothorax, and metathorax) to cause the sensory neurons developing from the different grafts to enter the CNS at different levels (Fig. 5; Anderson, 1981 a).

Fig. 5

Sensory projections from Fl neurons developing at 4 different ectopic locations: the posterior head (a), the prothorax (b), the mesothorax (c), and the metathorax (d). Closed arrows indicate the site of entry of the neurons into the CNS. At the foot of each column is a diagram of a transverse section through the ganglion which the neurons first enter. Within each section the neuropil is delimited by a dotted line, the major longitudinal tracts are indicated, the ventral neuropil area is cross-hatched, and the Median Ventral Tract is indicated by an open arrow. SG, suboesophageal ganglion; Pro, prothoracic ganglion; Meso, mesothoracic ganglion; Meta, metathoracic ganglion.

Fig. 5

Sensory projections from Fl neurons developing at 4 different ectopic locations: the posterior head (a), the prothorax (b), the mesothorax (c), and the metathorax (d). Closed arrows indicate the site of entry of the neurons into the CNS. At the foot of each column is a diagram of a transverse section through the ganglion which the neurons first enter. Within each section the neuropil is delimited by a dotted line, the major longitudinal tracts are indicated, the ventral neuropil area is cross-hatched, and the Median Ventral Tract is indicated by an open arrow. SG, suboesophageal ganglion; Pro, prothoracic ganglion; Meso, mesothoracic ganglion; Meta, metathoracic ganglion.

The projections in the CNS were stained by cobalt backfilling. They were found to be remarkably conservative; in all cases the neurons passed directly to the ventral neuropil where they formed a discrete projection (Fig. 5). Indeed, it has not yet proved possible to misdirect these neurons into a dorsal location (unpublished observations). In those cases where the sensory neurons also formed ascending and/or descending branches to other ganglia, they always did so within a single tract, the Median Ventral Tract (Tyrer & Gregory, 1981), which runs in a similar position in each ganglion (Fig. 5). The same projection area in the ventral neuropil and the same tract are also used by sensory neurons from normal wind-sensitive hairs and other mechanoreceptive hairs of the body surface (Anderson, 1981 b).

Another feature of the projections was also conservative. F 3 neurons formed only an ipsilateral projection in all ganglia, whereas F1 or F 2 neurons formed additional contralateral branches, although the projections of Fl and F2 neurons in ganglia distant from the ganglion of entry were sometimes only ipsilateral (e.g. compare the projections in the mesothoracic and prothoracic ganglia from F1 neurons developing on the mesothorax, Fig. 5(c)).

The only variability shown by the projections was in their extent. Grafts to the posterior head that entered the suboesophageal ganglion all showed the same projection as in Fig. 5(a). There was no variation between individual preparations. Grafts to the prothorax showed some variability. Some examples projected up to the suboesophageal and down to the mesothoracic ganglion (Fig. 5 b) but many formed only descending or ascending branches or were entirely local. Even more variation occurred in projections from grafts to the mesothorax and metathorax, where all combinations of local, ascending and descending components were observed. Fig. 5(c, d) show two of the more extensive projections. In contrast, the projections from surrounding host hairs were constant in extent from animal to animal. In all grafts the neurons never extended more anteriorly than the suboesophageal ganglion (Fig. 5).

Extracellular recording of the TCG failed to reveal any excitatory or inhibitory input to the TCG from the graft neurons, even though the TCG passes through these ganglia.

Choice of pathways within ‘foreign’ ganglia

Ectopic neurons form projections that are discrete and characteristic, not random and chaotic. The pathways taken are not merely a result of following surrounding host neurons since the host and graft projections often differ in clear and precise ways.

In those cases where the CNS route used by the normal sensilla is present, it is always taken, even if entered at an abnormal point, and it may be followed in either direction. Where the normal route is absent, but a homologous one is present, then this is taken.

These simple statements cannot yet be applied to the Antp, ssa andpbts systems where the complex organization of the brain makes it difficult to assign homologies to particular pathways and tracts. Analysis is also complicated by the fact that it is not known which sensillum types are present on all the homeotic appendages, e.g. mechanoreceptors, olfactory receptors, taste receptors, or multimodal receptors, nor which components they each contribute to the observed projections. The incomplete transformation of campaniform sensilla seen in bithorax flies indicates that it may not be sufficient to assume, for example, that a homeotic tarsus bears only ‘tarsal’ receptors.

Recognition of target neurons

Ectopic tarsal chemoreceptors in Antp and ssa flies form functional connexions within the proboscis extension reflex pathway. Whilst cobalt-filling of projections from the homeotic tarsi has provided details of the pathways taken within the brain, it has not revealed the region in which the reflex is mediated since it has not been possible to fill specifically the tarsal chemoreceptors. A behavioural comparison with homeotic pbts tarsi might be illuminating since they have only two projection areas in common - the antennal glomerulus, and the proboscis centre. Trans-neuronal filling from normal tarsal neurons in the thoracic ganglion might also help with the identification of interneurons in the normal reflex pathway. One could then ask, can ectopic sensory neurons recognize different neurons in the pathway or only the normal interneuron but at a different location?

In crickets, sensory neurons on ectopic cerci also succeed in forming functional connexions with appropriate interneurons. Again, the neural pathways underlying this response have not yet been elucidated, but since these early experiments, considerable advances have been made in understanding the details of the normal cercus-to-giant interneuron system. Two major classes of filiform hairs are now recognized: the dorsally and ventrally located T-hairs whose hair shafts vibrate transversely to the long axis of the cercus, and the medially and laterally located L-hairs whose shafts vibrate longitudinally (Palka, Levine & Schubiger, 1977). Four interneurons have been identified and their responses to T- and L-hair stimulation examined both extracellularly and intracellularly (Palka & Olberg, 1977; Matsumoto & Murphey, 1977; Levine & Murphey, 1980): each has a specific pattern of excitatory and inhibitory inputs from ipsilateral and contralateral T- and L-hairs. It seems likely that the excitatory pathways are monosynaptic, but the inhibitory pathways involve other interneurons (Levine & Murphey, 1980).

These details of interconnexion raise interesting questions about the earlier cereal transplantation study: 1. How are the regenerating afferents distributed in the mesothoracic ganglion ? Do they follow a limited number of pathways ? If so, are they homologous to those normally taken in the terminal ganglion ? 2. What proportion of afferents regenerate connexions with these ascending interneurons? What proportion form connexions with other inappropriate interneurons? 3. Are the regenerated connexions of the appropriate sign and strength for the source of input? 4. If inhibitory responses are observed, are they mediated by interneurons which are homologous to those in the terminal ganglion? These questions are for future investigation.

Another promising system to study at this level might be the equilibrium detecting system of the cricket. At the base of each cercus is an orderly arrangement of clavate hairs that act as equilibrium detectors (Bischof, 1975). Individual clavate hairs have been backfilled and their CNS projections described in detail; they form a topographic arrangement of branches within a particular region of neuropil in the terminal ganglion, where the termination site of each afferent is related to its position and/or birthdate along and around the cercus (Murphey, Jacklet & Schuster, 1980). Furthermore, two pairs of large bilaterally symmetrical ascending interneurons receiving clavate hair input have been described (Sakaguchi & Murphey, 1980). This system is now being examined following transplantation of cerci to the mesothorax (R. K. Murphey, personal communication).

In contrast to the systems described earlier in this section, ectopic windsensitive hairs of the locust fail to form functional connexions with one of their appropriate interneurons, the TCG. What might account for this difference ? One reason could be that only parts of an interneuron may be recognized as input regions. Those of the TCG might be restricted to its arborization in the tritocerebrum where it normally receives head-hair input. Since the ectopic hair neurons never ascend more anteriorly than the suboesophageal ganglion (Fig. 5), they would not encounter this input region. On the other hand, the definitive tests for monosynapticity of the pathway between head-hairs and TCG have not been done, so it remains possible that local interneurons might be required to mediate the input. These interneurons, or their homologues, might be absent in other ganglia, thus preventing the formation of a functional connexion. It also seems likely that sensory neurons require close proximity to an interneuron before recognition can occur. Since the TCG passes through a dorsal tract, the Dorsal Intermediate Tract, and has a dorsal arborization in all of the ventral cord ganglia (Bacon & Tyrer, 1978), while the sensory neurons have an exclusively ventral projection, the two may never come close enough for recognition. The cricket giant ascending interneurons on the other hand have some ventral branches, and run in a ventral tract, the Ventral Intermediate Tract (J. L. D. Williams, personal communication), and may be seen in close proximity to the sensory fibres from grafted cerci (Edwards & Sahota, 1967).

The picture presented here is derived only from observation of the end-result following a perturbation. It is possible that these projections are far more extensive during their establishment, but such a dynamic view remains to be obtained.

I thank Drs Duncan Byers, Steven Green, Brian Mulloney, Dick Nassel, Margrit Schubiger and Rheinhard Stocker for valuable comments on the manuscript.

Anderson
,
H.
(
1981a
).
Development of a sensory system: the formation of central projections and connections by ectopic wind-sensitive hairs in the locust
Schistocerca gregaria (in prep
.).
Anderson
,
H.
(
1981b
).
The organization of mechanosensory inputs within the locust central nervous system (in prep.)
.
Anderson
,
H.
&
Bacon
,
J.
(
1979
).
Developmental determination of neuronal projection patterns from wind-sensitive hairs in the locust, Schistocerca gregaria
.
Devi Biol
.
72
,
364
373
.
Bacon
,
J.
&
Anderson
,
H.
(
1981
).
Developmental determination of central connections from wind-sensitive hairs in the locust
,
Schistocerca gregaria (in prep
.).
Bacon
,
J. P.
&
Tyrer
,
N. M.
(
1978
).
The tritocerebral commissure giant (TCG): a bimodal interneurone in the locust
.
J. comp. Physiol
.
126
,
317
325
.
Bischof
,
H. J.
(
1975
).
Die keulenformigen Sensillen auf den Cerci der Grille Gryllus bimaculatus ais Schrewe receptoren
.
J. comp. Physiol
.
98
,
277
288
.
Cole
,
E. S.
&
Palka
,
J.
(
1980
).
Sensilla on normal and homeotic wings of Drosophila
.
Am. Zool
.
20
,
740
.
Deak
,
I.I.
(
1976
).
Demonstration of sensory neurons in the ectopic cuticle of spinelessaristapedia, a homeotic mutant of Drosophila
.
Nature
260
,
252
254
.
Dethier
,
V. G.
(
1955
).
The physiology and histology of the contact chemoreceptors of the blowfly
.
Q. Rev. Biol
.
30
,
348
371
.
Edwards
,
J. S.
&
Sahota
,
T. S.
(
1967
).
Regeneration of a sensory system: the formation of central connections by normal and transplanted cerci of the house cricket, Acheta domesticas
.
J. exp. Zool
.
166
,
387
396
.
Ghysen
,
A.
(
1978
).
Sensory neurones recognize defined pathways in Drosophila central nervous system
.
Nature
274
,
869
872
.
Ghysen
,
A.
(
1980
).
The projection of sensory neurons in the central nervous system of Drosophila; choice of the appropriate pathway
.
Devi Biol
.
78
,
521
541
.
Ghysen
,
A.
&
Janson
,
R.
(
1980
).
Sensory pathways in Drosophila central nervous system
.
In Development and Neurobiology of Drosophila
(eds
O.
Siddiqi
,
P.
Babu
,
L. M.
Hall
&
J. C.
Hall
), pp.
247
265
.
New York
:
Plenum
.
Green
,
S. H.
(
1979
).
Neuroanatomy of flies with abnormal appendages
.
Rep. Div. Biol. Calif. Instit
.
Green
,
S. H.
(
1980
).
Innervation of antennae and legs in Drosophila; wild-type and homeotic mutants
.
Soc, Neurosci. Abstr
.
6
,
677
.
Levine
,
R. B.
&
Murphey
,
R. K.
(
1980
).
Pre- and postsynaptic inhibition of identified giant interneurons in the cricket (Acheta domesticas)
.
J. comp. Physiol
.
135
,
269
282
.
Matsumoto
,
S. G.
&
Murphey
,
R. K.
(
1977
).
The cercus-to-giant interneuron system of crickets. IV. Patterns of connectivity between receptors and the Medial Giant Interneuron
.
J. comp. Physiol
.
119
,
319
330
.
Morata
,
G.
(
1975
).
Analysis of gene expression during development in the homeotic mutant Contrabithorax of Drosophila melanogaster
.
J. Embryol. exp. Morph
.
34
,
19
31
.
Morata
,
G.
&
Garcia-Bellido
,
A.
(
1976
).
Developmental analysis of some mutants of the bithorax system of Drosophila
.
Wilhelm Roux Arch, devl Biol
.
179
,
125
143
.
Murphey
,
R. K.
,
Jacklet
,
A.
&
Schuster
,
L.
(
1980
).
A topographic map of sensory cell terminal arborizations in the cricket CNS: correlation with birthday and position in a sensory array
.
J. comp. Neurol
.
191
,
53
64
.
Ouweneel
,
W. J.
(
1973
).
Replacement patterns of homoeotic wing structures in halteres
.
Dros. Inf. Serv
.
50
,
102
.
Palka
,
J.
,
Lawrence
,
P. A.
&
Hart
,
H. S.
(
1979
).
Neural projection patterns from homeotic tissue of Drosophila studied in bithorax mutants and mosaics
.
Devl Biol
.
69
,
549
575
.
Palka
,
J.
,
Levine
,
R.
&
Schubiger
,
M.
(
1977
).
The cercus-to-giant interneuron system of crickets. I. Some attributes of the sensory cells
.
J. comp. Physiol
.
119
,
267
283
.
Palka
,
J.
&
Olberg
,
R.
(
1977
).
The cercus-to-giant interneuron system of crickets. III. Receptive field organization
.
J. comp. Physiol
.
119
,
301
317
.
Palka
,
J.
&
Schubiger
,
M.
(
1980
).
Formation of central patterns by receptor cell axons
.
In Development and Neurobiology of Drosophila
(eds
O.
Siddiqi
,
P.
Babu
,
L. M.
Hall
&
J. C.
Hall
), pp.
223
246
.
New York
:
Plenum
.
Sakaguchi
,
D. S.
&
Murphey
,
R. K.
(
1980
).
Re-establishment of functional neuronal connections following transplantation of a sensory appendage in the equilibrium detecting system of the cricket
.
Soc. Neurosci. Abs
.
6
,
678
.
Stocker
,
R. F.
(
1977
).
Gustatory stimulation of a homeotic mutant appendage, Antennapedia, in Drosophila melanogaster
.
J. comp. Physiol
.
115
,
351
361
.
Stocker
,
R. F.
(
1979
).
Fine structural comparison of the antennal nerve in the homeotic mutant Antennapedia with the wild-type antennal and second leg nerves of Drosophila melanogaster
.
J. Morphol
.
160
,
209
222
.
Stocker
,
R. F.
(
1981
).
Homeotically displaced sensory neurons in the proboscis and antenna of Drosophila project into the same identified brain regions though by different pathways (in prep.)
.
Stocker
,
R. F.
,
Edwards
,
J. S.
,
Palka
,
J.
&
Schubiger
,
G.
(
1976
).
Projection of sensory neurons from a homeotic mutant appendage, Antennapedia, in Drosophila melanogaster
.
Devl Biol
.
52
,
210
220
.
Stocker
,
R. F.
&
Lawrence
,
P. A.
(
1981
).
Sensory projections from normal and homeotically transformed antennae in Drosphila
.
Devl Biol
.
82
,
224
237
.
Stocker
,
R. F.
&
Schorderet
,
M.
(
1981
).
Cobalt filling of sensory projections from internal and external mouthparts in Drosophila
.
Cell Tiss. Res
.
216
,
513
523
.
Strausfeld
,
N. J.
&
Singh
,
R. N.
(
1980
).
Peripheral and central nervous system projections in normal and mutant (bithorax) Drosophila melanogaster
.
In Development and Neurobiology of Drosophila
(eds
O.
Siddiqi
,
P.
Babu
,
L. M.
Hall
&
J. C.
Hall
), pp.
267
291
.
New York
:
Plenum
.
Tyrer
,
N. M.
,
Bacon
,
J. P.
&
Davies
,
C. A.
(
1979
).
Sensory projections from the windsensitive head hairs of the locust Schistocerca gregaria
.
Cell Tiss. Res
.
203
,
79
92
.
Tyrer
,
N. M.
&
Gregory
,
G. E.
(
1981
).
A guide to the neuroanatomy of locust suboesopha-geal and thoracic ganglia
.
Proc. Roy. Soc. Lond. B (in press)
.