ABSTRACT
The paired abdominal cerci of the cricket Acheta domesticus are mechanosensory appendages which regenerate readily when amputated during larval life. Their peripherally-located sense cells form axons which project centrally as a purely sensory nerve to the terminal abdominal ganglion.
In an attempt to analyze some of the factors which guide a regenerating sensory nerve to correct central terminations, implants of homologous, supernumerary terminal ganglia were made in cricket larvae and the host cerci amputated. The possibility that implants with multiple nerve stumps might release an attracting substance was considered. Surgical procedures used were (1) implant in posterior abdomen; (2) implant in posterior abdomen, ipsilateral to chronic cereal deprivation; (3) implant in mesothoracic leg socket, adjacent to heterotopically-transplanted regenerated cercus; (4) implant in posterior abdomen, ipsilateral host cereal motor nerve sectioned; (5) implant in posterior abdomen, ipsilateral margin of host terminal ganglion wounded. Results were determined after the adult molt, by con-ventional histology or by cobalt chloride filling of regenerated cereal nerves.
In all procedures except (3) and (4), the regenerated afferent nerve bypassed the implant and terminated in the host terminal ganglion. In (3), the regenerated fibers from cereal grafts bypassed the implant; terminations were not found. In (4), some regenerated cereal axons connected with the implant and the majority terminated in the host ganglion.
It is suggested that regenerating cereal afferents may depend in a facultative way on the cereal motor nerve as a pathway guide but there is as yet no clear evidence for a trophic influence from the central nervous system.
INTRODUCTION
The development in arthropods of functional neural connexions between a regenerating sense organ and the central nervous system requires that the regenerating neurons first extend axons from the integument to the central ganglion, and then establish appropriate synaptic connexions within the neuropile. This study is concerned with the first of these processes, the establishment of a pathway between periphery and center.
The abdominal cerci of the cricket, Acheta domesticus, provide appropriate material for such studies since they regenerate well, and have been used in studies of central connexion formation (Edwards & Sahota, 1967; Edwards & Palka, 1974; Palka & Edwards, 1974).
Cerci are paired elongate mechanosensory appendages arising from the dorso-posterior angles of the abdomen. They lack intrinsic muscles but are capable of limited movement by means of extrinsic dorsal muscles. Their limited mobility is sufficient to facilitate grooming and the support of the female by the male during copulation; they execute no evident movement in relation to their sensory function. The cerci are densely clothed with sensilla of three basic types, which have been described in detail by Edwards & Palka (1974). The cereal sensory nerve (N 10d) in the adult contains about 10500 axons (Edwards, 1971) This nerve terminates almost entirely ipsilaterally in the terminal abdominal ganglion and provides a major input to the two giant interneurons, 8–1 and 9–1, whose axons traverse the length of the ventral cord (Murphey, Mendenhall, Palka & Edwards, 1975). According to present knowledge, the cereal sensory nerve (N 10d) is entirely afferent; although the possibility that it may contain a small efferent component cannot be dismissed (d’Ajello, Bettini & Casaglia,1972). This nerve is entirely separate from the smaller adjacent cereal motor nerve, which branches from N lOv. Essential features of the cereal sensory nerve-giant interneuron system are summarized in Fig. 1.
Schematic diagram of Acheta terminal ganglion (TG) and neural connexions of the cercus (CE). Cereal sensory axons (SC) with cell bodies located beneath the cereal integument (CI) send fibers through the cereal sensory nerve (CSN) to the terminal ganglion where they synapse principally with giant interneurons, of which two (MGI, LGI) with their cell bodies (8–1, 9–1) are shown. The cereal motor nerve (CMN) arises from nerve 10 v and innervates the extrinsic cereal musculature at the base of the cercus. GN -genital nerve.
Schematic diagram of Acheta terminal ganglion (TG) and neural connexions of the cercus (CE). Cereal sensory axons (SC) with cell bodies located beneath the cereal integument (CI) send fibers through the cereal sensory nerve (CSN) to the terminal ganglion where they synapse principally with giant interneurons, of which two (MGI, LGI) with their cell bodies (8–1, 9–1) are shown. The cereal motor nerve (CMN) arises from nerve 10 v and innervates the extrinsic cereal musculature at the base of the cercus. GN -genital nerve.
After repeated cereal amputation or nerve section, degeneration of the proximal stump sheath follows degeneration of the sensory nerve. Thus, there is no obvious physical substrate that might supply contact guidance for regenerating afferent fibers. In experiments where crickets were deprived of cereal input during development by repeated amputation of postmolt regenerate cerci (Palka & Edwards, 1974), the surface of the adult terminal ganglion at the N 10d locus was smoothly healed and covered with neural lamella. Nevertheless, if such crickets were permitted to regenerate cerci toward the end of postembryonic development, they did so rapidly and with apparently correct afferent terminations.
Working with heterotopically-grafted regenerate cerci in cricket larvae, Edwards & Sahota (1967) found that afferent fibers, arising from successful grafts at the mesothoracic leg socket, made connexions with the host’s central nervous system. These contacts were found in the vicinity of the giant interneuron axons, the same cells with which the cereal nerve normally synapses in the terminal ganglion. Moreover, extracellular recordings from abdominal connectives indicated that the giant axons were excited by stimuli applied to the heterotopic regenerate cercus.
Thus, regenerating cereal axons are capable of (1) routing a correct course without the guidance of a pre-established path, (2) penetrating a healed ganglion in the correct region, and (3) recognizing appropriate post-synaptic cells in an abnormal region.
These results suggest questions that might be asked to determine the nature and precedence of the cues that allow regenerating sensory axons to connect with their target organ. For example, can regenerating fibers be diverted from their normal course and termination by the presence of another potentially acceptable target? Their response when presented with a homologous, supernumerary ganglion might help to sort out the various possible cues operating during afferent regeneration. If the diffusion of ‘wound factors’ (Bodenstein, 1957) from damaged nerve occurs, this tissue could supply such a directional signal to the regenerating axons. The release of material from the multiple wound sites of an implanted ganglion might then be sufficient to override other cues that normally influence regenerating fibers. Another question can be posed: In the event that the nerve recognized the implant, would it make contact at the appropriate (N 10d) region, or simply enter the nerve stump first encountered? Finally, how would regenerating axons respond to the presence of a potentially increased central field size?
Working with Acheta domesticus, Rummel (1970) came to the conclusion that successful regeneration of a cercus depends upon an uninterrupted nerve supply, inferring the dependence of sensory regeneration on a trophic factor distributed by the motor nerve. This hypothesis brings up another question regarding the requirement for an intact cereal motor nerve in sensory regeneration.
Implants of healthy, conspecific tissues generally grow and become tracheated by the insect host. In the case of ganglia, such grafts often participate in patterns of reciprocal motor and interneuronal innervation with the host system (Bodenstein, 1955, 1957; Guthrie, 1966; Jacklet & Cohen, 1967; Guthrie & Banks, 1969). The relationships of host to graft in the case of afferent nerves are less well documented.
The observations to be described here are endpoint results; the outcome of experiments was not ascertained until after the adult molt. The supernumerary ganglion was never finally accepted as the appropriate target of entire regenerated cereal nerves, except that we found small populations of regenerated cereal fibers within the implanted ganglion in cases where the host’s ipsilateral cereal motor nerve trunk was cut at the time of implantation.
METHODS
Rearing and development of Acheta domesticus. By providing gravid females from mass cultures (Fluker’s Cricket Farm, Baton Rouge, La.) with receptacles of damp sand for oviposition, continuous supplies of young crickets were made available.
All animals were housed in a controlled-environment chamber at 27–28 °C, about 70% relative humidity, with a photoperiod of 16-h light and 8-h dark; they were fed on ‘Little Friskies’ dried cat food (Carnation Co.) and fresh lettuce. Stock animals were reared in small groups, experimental animals in isolation.
The number of instars required for development of Acheta domesticus varies according to culture conditions. Our animals invariably passed through nine instars before the final molt to the adult. The duration of stadia is summarized in Table 1.
Surgery
All operations described below were performed on 6th to 8th stage larvae at a time before the midpoint of the stadium, that is, before apolysis. Post-apolysis larvae were less able to withstand surgical trauma. The diagrams of Fig. 2 illustrate the five categories of surgery that comprised these experiments.
Surgical procedures used in this study. (A) Simple terminal ganglion (TG) implant with concurrent ipsilateral cercectomy. (B) TG implant, host reared without ipsilateral cercus, bilateral cercectomy concurrent with implantation.(C) Heterotopic TG implant and cereal graft on mesothoracic leg base. (D) TG implant, ipsilateral cereal motor nerve sectioned, concurrent bilateral cercectomy. (E) TG implant, ipsilateral host TG wounded, concurrent bilateral cercectomy.
Surgical procedures used in this study. (A) Simple terminal ganglion (TG) implant with concurrent ipsilateral cercectomy. (B) TG implant, host reared without ipsilateral cercus, bilateral cercectomy concurrent with implantation.(C) Heterotopic TG implant and cereal graft on mesothoracic leg base. (D) TG implant, ipsilateral cereal motor nerve sectioned, concurrent bilateral cercectomy. (E) TG implant, ipsilateral host TG wounded, concurrent bilateral cercectomy.
For ganglion transfer experiments, two animals were immobilized by chilling to about 4 °C, then supported venter up by means of filter paper strips pinned on each side to a thin wax plate attached to a covered plastic container of ice water. In all experiments except those illustrated in Fig. 2C, the host’s penultimate sternite was freed on three sides with fine sharp scissors and braced open with a pin. To prevent desiccation of tissues and coagulation of hemolymph, a saline based on cricket hemolymph (Levine, 1966) was applied. The terminal ganglion of the donor animal was rapidly excised and rinsed in Levine’s solution. It was then placed in the host’s hemocoel, posterolateral to the autogenous terminal ganglion, and toward the origin of a forthcoming cereal nerve re-generate. The flap of cuticle was then repositioned, excess fluid blotted off, and the wound allowed to seal by hemolymph coagulation.
Host animals for the experiments diagrammed in Fig. 2B were reared without one cercus by means of repeated extirpation up to the 7th or 8th stage. This was done by removing the regenerated cereal ‘button’ just after each molt.
Control experiments to determine the response to cereal motor nerve section were done by cutting the entire motor root, N 10v, at its exit from the terminal ganglion (see Fig. 1). This nerve divides about halfway down its length to supply genitalia with one branch and cereal base musculature with the other. Isolating and sectioning only the cereal motor branch in a larval animal is technically difficult and requires more extensive surgery than sectioning the entire root near the source.
In the implant experiments diagrammed in Fig. 2D, E, nerve section or wounding of the host ganglion was done just before inserting the implant. Host ganglia were wounded either by sectioning the roots of N 8 (see Fig. 1) or puncturing at the midlateral margin, ipsilateral to the implant.
For the heterotopic implant-cereal graft experiments (Fig. 2C), donors had been cercectomized two molts earlier, so that they bore small regenerate cerci. After amputating the host’s mesothoracic leg at the coxofemoral joint, the donor terminal ganglion was inserted through the wound. The donor cereal regenerate was amputated and applied to the host’s leg socket and held until it adhered by coagulated hemolymph.
No antiseptic precautions were found necessary, other than cleaning instruments in 70 % ethanol.
Host animals were cercectomized by grasping the cercus at the base with fine forceps and pulling it cleanly away. The cereal motor nerve was left intact by this operation.
Nerve tracing and histology
Animals were chilled to immobility and injected with paraformaldehyde-glutaraldehyde fixative (Edwards & Palka, 1974) through the lateral mesothoracic intersegmental membrane. After one-half to one hour, the animal was dissected and the nerve tissue immersed in fixative for 1 to 2 h at room temperature.
Simultaneous fixation and staining of tissues by immersion in a solution of 0·5% thionin in 10% formalin for a period of 1 to 4 days (Ehrlich, in Gurr, 1953) provided a method for nerve tracing in which cell bodies were stained blue and fibers pink. The distortion produced by this treatment was reduced by prefixing tissues as described above for 1–5 days, then transferring them to the thionin-formalin solution for 2 days. Tissues were embedded in paraffin and sectioned at 8–15 μm.
Sections pre-treated with thionin were dewaxed and cleared in toluene and mounted in Permount. Sections not previously stained were hydrated and stained in 0·1 % aqueous toluidine blue.
In some experiments, regenerated cereal nerves were filled with cobalt chloride in order to visualize their terminations. The cereal nerve was cut at the base of the cercus, then the terminal ganglion including about 1 mm of connectives and other associated nerve stumps was excised from the host. The distal cereal nerve stump was immersed in 200 mM COC12 (aq.) at 5 °C for 18 to 24 h. Levine’s saline (1966) was used for rinsing and sulfide development, and the part of the tissue not exposed to CoCl2 during treatment was immersed in Schneider’s Drosophila medium. Tissues were fixed for 1 to 3 h in aldehyde fixative, then prepared for paraffin histology. Sections were cut at 10 to 20 μm and processed with Timm’s intensification method as modified by Tyrer & Bell (1974).
RESULTS
The data from each of five categories of experiments are summarized in Table 2. All regenerated cereal sensory nerves grew to the autogenous (host) ganglion, except where the ipsilateral cereal motor nerve trunk was sectioned (category (4)).
Experimental animals grew and molted normally, although maturation times occasionally exceeded the normal range. Regenerated cerci were indistinguishable from those of control animals without implants.
Implanted ganglia invariably became attached, usually by means of scar or connective tissue, to the body wall near the point where they were originally inserted. They increased in size in parallel with the autogenous terminal ganglion, and were amply tracheated with outgrowths from the host system (Fig. 3).
Dissection of adult with simple ganglion implant (procedure A). Implant (Im), in host 36 days, is well tracheated, with multiple connexions to nearby tracheal trunks (Tr). Scale: 0·5 mm.
Although there were varying degrees of internal disorganization and local degeneration, implanted ganglia produced fiber outgrowth volumes consonant with their length of time in the host. These outgrowths came from motor or mixed nerve or connective stumps; cereal sensory nerve stumps, where recognizable, had degenerated. Attachments between fibers originating from implants and host muscle occurred in most cases.
Simple implants (Fig. 2 A)
In all eight cases, the regenerated cereal sensory nerve established connexions in the appropriate region of the host terminal ganglion. Even in situations where the base of the regenerated cercus was clearly nearer to the implant than to the host ganglion, we found no regenerated afferent fibers contacting the implant.
Implant in ‘deprived’ host (Fig. 2B)
In all four adults the ipsilateral cereal sensory nerve made contact only with the host terminal ganglion. No neural connexions linking the implanted and host ganglia were found.
Heterotopic implant with cereal graft (Fig. 2C)
Of the five animals subjected to this operation, three lost the grafted cercus before or during the subsequent molt. Two animals retained grafts, but the cereal form was maintained in only one (Fig. 4 A). Serial sections of the heteromorphic structure in the other animal revealed no sensory cells. Sections from the animal with the successful heterotopic cercus revealed that the afferent fiber bundle formed from sensory cells in the graft had bypassed the implanted ganglion and had grown toward the host’s ventral cord (Fig. 4B).
(A) Adult female with heterotopic TG implant (procedure C) and cereal graft (Ce) to mesothoracic leg base. Scale: 1 mm. (B) Horizontal section to mesothorax showing implanted ganglion (Ig) and base of cercus below, with sensory nerve (N) entering leg base. Toluidine blue. Scale: 100 μm. (C) Base of normal regenerate cercus showing separate cereal motor (CM) and cereal sensory (CS) nerves. Arrow indicates direction of cercus tip. Thionin stain. Scale: 100 μm. (D) Base of regenerate cercus in animal with cereal motor nerve cut three instars earlier (procedure D), showing mixed motor and sensory nerve (SMN). Thionin stain. Scale: 100 μm.
(A) Adult female with heterotopic TG implant (procedure C) and cereal graft (Ce) to mesothoracic leg base. Scale: 1 mm. (B) Horizontal section to mesothorax showing implanted ganglion (Ig) and base of cercus below, with sensory nerve (N) entering leg base. Toluidine blue. Scale: 100 μm. (C) Base of normal regenerate cercus showing separate cereal motor (CM) and cereal sensory (CS) nerves. Arrow indicates direction of cercus tip. Thionin stain. Scale: 100 μm. (D) Base of regenerate cercus in animal with cereal motor nerve cut three instars earlier (procedure D), showing mixed motor and sensory nerve (SMN). Thionin stain. Scale: 100 μm.
Motor nerve section
Preliminary operations were done on 16 animals to determine survival rates. In these animals both motor nerves (N 10v) were sectioned, both cerci removed, and a ganglion was implanted. Two reached the last molt; most dying within 4 days after surgery. A control group of 12 animals had only one N 10v sectioned, and one or both cerci removed. In animals that reached adulthood, we found both motor and sensory cereal nerves had regenerated and made anatomically correct connexions. The only abnormality found upon histological examination was a mixing of fibers from sensory and motor bundles near the cereal base (Fig. 4D). More proximally, the two major bundles separated and were attached at normal positions on the terminal ganglion. Cerci regenerated by these animals were indistinguishable from normal regenerates.
Implant with cut motor nerve (Fig. 2D)
Five out of ten operated crickets became apparently healthy adults. Tissues from two of these were processed with the thionin method, and cobalt preparations of the ipsilateral cereal sensory nerve were made with three.
Both cereal motor and sensory nerves regenerated, but the sensory nerves had attached to both the implanted and the host terminal ganglia. A sketch of the ganglia from one of these adults in Fig. 5 A represents a typical pattern of connexions seen in this category.
Sketches of host and implanted ganglion and their relationships. (A) Ipsi-lateral nerve 10v cut at time of ganglion implantation. Implant (Im) has massive neuroma to left. Regenerate cereal sensory nerve (CSN) reaches host ganglion (HG) via implant. (B) Ganglia from animal with lateral wound (W) on host ganglion (HG) ipsilateral to implanted ganglion (Im). Neural connexion formed between neuroma at wound site and implant, which is heavily tracheated (Tr). Cereal sensory nerves (CSN) from normal (left) and regenerate (right) cerci connect with host ganglion (HG).
Sketches of host and implanted ganglion and their relationships. (A) Ipsi-lateral nerve 10v cut at time of ganglion implantation. Implant (Im) has massive neuroma to left. Regenerate cereal sensory nerve (CSN) reaches host ganglion (HG) via implant. (B) Ganglia from animal with lateral wound (W) on host ganglion (HG) ipsilateral to implanted ganglion (Im). Neural connexion formed between neuroma at wound site and implant, which is heavily tracheated (Tr). Cereal sensory nerves (CSN) from normal (left) and regenerate (right) cerci connect with host ganglion (HG).
Regenerated fiber bundles were associated with small glial cells and were not as clearly delineated as bundles in normal material. In every case it was possible to trace the regenerated sensory nerve from the base of the ipsilateral cercus to contact first with the implant, and further to contact with the host terminal ganglion (Fig. 6A and B). The great majority of afferent fibers appeared to terminate in the host ganglion but examination of CoCl2-filled tissues revealed a few fiber terminations in the implant (Fig. 6C).
(A and B): Horizontal 10 μm paraffin sections, CoCl2 treatment, of host and implanted ganglion in host with previously cut N 10v (procedure C). Implant in host 33 days. Sections A and B separated vertically by about 20 μm. Pathway of regenerated cereal sensory nerve (CSN and arrow) connects with implanted ganglion (1G) by way of neuroma the remainder of the CSN diverts to the host ganglion (HG). Scale: 100 μm. (C) Projection of cobalt-filled regenerate cereal sensory fibers in 28-day implant ganglion; 20µm paraffin section. Scale: 100 μm. (D) Periphery of healthy 33-day implant ganglion. PN, intact perineurial sheath. CB, neuron cell body. 10 μm paraffin section, glutaraldehyde fixation, osmium staining. (E) Degenerating 25-day implant ganglion, perineurial-neuropile zone, invaded by hemocytes (H), which may be distinguished from glial cells (G) by the density of their nuclei. 10 μm paraffin section, toluidine blue stain. Scale for D and E: 10 μm.
(A and B): Horizontal 10 μm paraffin sections, CoCl2 treatment, of host and implanted ganglion in host with previously cut N 10v (procedure C). Implant in host 33 days. Sections A and B separated vertically by about 20 μm. Pathway of regenerated cereal sensory nerve (CSN and arrow) connects with implanted ganglion (1G) by way of neuroma the remainder of the CSN diverts to the host ganglion (HG). Scale: 100 μm. (C) Projection of cobalt-filled regenerate cereal sensory fibers in 28-day implant ganglion; 20µm paraffin section. Scale: 100 μm. (D) Periphery of healthy 33-day implant ganglion. PN, intact perineurial sheath. CB, neuron cell body. 10 μm paraffin section, glutaraldehyde fixation, osmium staining. (E) Degenerating 25-day implant ganglion, perineurial-neuropile zone, invaded by hemocytes (H), which may be distinguished from glial cells (G) by the density of their nuclei. 10 μm paraffin section, toluidine blue stain. Scale for D and E: 10 μm.
CoCl2 projections in host ganglia were not as complete as those typically seen in the case of normally developed sensory neurons. This would be expected, since afferent projections from regenerated cerci in control animals are less dense than those from normal cerci, the reduced quantity of fibers evidently reflecting the time available for regeneration (Fig. 6C).
Implanted ganglia appeared to be healthy, and in general their symmetry was well preserved (Fig. 6D). In one case, the motor nerve (N lOv) of the implant made contact with muscles of the host’s cereal base, ipsilateral and parallel to the afferent nerve from the regenerated cercus to the implant.
The striking results of this group of experiments indicated that the cereal motor nerve -or at least N 10v -influenced the growth direction of sensory fibers, but whether the influence was direct or secondary was still unclear.
Ganglion implant with simultaneous wounding of host terminal ganglion (Fig. 2E)
This operation gave rise to moderate distortions of overall morphology and neuropile patterns of the host ganglion. Fig. 5B is a diagrammatic sketch made from a dissection of one of these animals : the relationships were similar in the other two cases. In every instance, host and donor ganglia were complexly interconnected by way of an outgrowth between the initial wound site of the host ganglion and one or more motor or connective trunks of the implanted ganglion (Fig. 7C). However, tracing of serial sections revealed that all the regenerate cereal sensory fibers bypassed the implant and terminated in the ipsilateral portion of the host ganglion (Fig. 7D). Since there was a varying degree of disorganization of the host neuropile brought about by wounding, pattern abnormalities appeared in the CoCl2-filled afferent terminations.
(A and B) Cobalt preparations of cereal sensory nerves in terminal ganglion, whole mounts. Dashed line is midline. (A) Normal adult. (B) Adult with regenerated cercus. Scale: 100 μm. (C and D) Animals with wounded host ganglia (treatment E). (C) Nerve connexions (arrow) between host (HG) and implanted ganglion (IG). Wound region of host ganglion at left (W). Implant in host 32 days. Toluidine blue stain. Scale: 100 μm. (D) Projection of regenerated cereal sensory nerve into host TG. Implant in host 28 days. .10 µm paraffin sections. Scale: 100 μm.
(A and B) Cobalt preparations of cereal sensory nerves in terminal ganglion, whole mounts. Dashed line is midline. (A) Normal adult. (B) Adult with regenerated cercus. Scale: 100 μm. (C and D) Animals with wounded host ganglia (treatment E). (C) Nerve connexions (arrow) between host (HG) and implanted ganglion (IG). Wound region of host ganglion at left (W). Implant in host 32 days. Toluidine blue stain. Scale: 100 μm. (D) Projection of regenerated cereal sensory nerve into host TG. Implant in host 28 days. .10 µm paraffin sections. Scale: 100 μm.
Non-sensory host-implant connexions
As indicated in Table 2, neural contacts between host and implanted ganglia occurred in four out of five categories. These connexions were formed by way of cut nerve stumps or damaged areas of ganglion. Outgrowths from implants occasionally resulted in neuromas (Fig. 6 A) which often served as channels for interconnexion between host and implant.
Neural connexions between the implant and host muscle were observed in three categories. The quantity of these contacts appeared to be related positively to the health of the implant and to the amount of the time in the host. The only known specificity exhibited in this class of connexions was the single case mentioned above in which the regenerated N 10v of the implanted ganglion made contact with muscles of the host’s cereal base.
Cellular migrations, proliferations; degenerative reactions
Nerve cells within implants usually appeared normal histologically (Fig. 6D). In two cases from the first category of experiments (Fig. 2 A), and in one from the second category (Fig. 2B), degenerative changes in the implants were pronounced. In most animals of the first category variable numbers of small cells were dispersed throughout the implants and their outgrowths (Fig. 6E). These appeared to be mainly small glial cells, but on the basis of comparison with hemolymph smears, also included hemocytes.
Summary of results
In all surviving animals, normal regenerate cerci were produced.
With the single exception of cereal afferents in experiments in which N 10v was sectioned, regenerated cereal sensory fibers bypassed the implant to terminate in the host terminal ganglion.
When implantation was combined with section of host N 10v, regenerating cereal afferents were diverted to the implanted ganglion, and a few fibers terminated within it.
In no instance did regenerate afferent (host) cereal nerves terminate entirely in the implanted ganglion.
In instances where contacts were made with the implanted ganglion, afferent fibers from the regenerate cercus did so in the appropriate region, that is, through part of the old sensory nerve stump.
Implants were always attached to host tissues. They were supplied with tracheae, grew in size during maturation of the host, and put forth nerve out-growths.
Nerve-muscle contact between donor and host, respectively, was found frequently. Nerve contacts between donor and host ganglia occurred in over half the cases.
Nerve outgrowths from donor ganglia arose from motor or connective stumps or from other wounds; sensory cereal nerve stumps of implants, where recognizable, were degenerate.
Glial cells responded to the changed circumstances of these experiments by accumulating in greater than normal numbers at sites of trauma, regeneration, and abnormal growth, and by atypical migrations within implants.
DISCUSSION
Target discrimination in regenerating nerves
Experimental studies of neural regeneration in orthopteroid insects, exploiting their capacity for sensory and motor regeneration and for sustaining implanted ganglia, have demonstrated the vigor of nerve growth and specificity in the restoration of functional connexions. It is clear that severed motor nerves regenerate in immature and adult stages (Bodenstein, 1957) and can recognize their correct target (e.g. Jacklet & Cohen, 1967; Pearson & Bradley, 1972) or their homologue from another segment (Young, 1972), and that supernumerary grafted limbs will acquire motor innervation even though the normal system is intact (Sahota & Edwards, 1969).
Implanted ganglia have been shown to provide channels for motor innervation and can innervate muscle (Guthrie, 1966; Guthrie & Banks, 1969). None of these studies addressed the question of the factors that determine the pathway of axons between their origin and target organs.
With sensory systems, similarly, the capacity for regeneration has been amply demonstrated, but with the exception of Wigglesworth’s (1953) demonstration of the tendency of ingrowing sensory axons to follow pre-existing pathways, the mechanisms of pathway determination have received little attention.
In our experiments sensory and motor connexions were made in the appropriate locations, as evaluated by endpoint histological examination, although some pattern abnormalities developed. Regenerated afferent and efferent fibers intermingled in the distal pathway; however they apparently became sorted out and formed topographically normal terminations. Hamburger (1929) observed that in the development of amphibian limbs the tips of pioneering nerve fibers growing out of motor and sensory centers take different routes through the tissue matrix, diverging at a considerable distance before reaching muscles or skin, respectively.
If the situations in amphibians and insects are comparable with respect to separation of types of growing nerves, it would be difficult to imagine how a regenerating cereal afferent fiber might depend for directional cues upon a motor nerve which, although running parallel to the sensory pathway, is normally entirely separate from it throughout its course. Further, the regenerating motor and sensory axons linking two regions grow in opposite directions. Yet the findings of this report show that, in the absence of the normal motor nerve at the onset of sensory regeneration, afferent fibers innervate a supernumerary ganglion which they would bypass were the motor nerve intact.
The presence of an intact motor nerve is not required as a ‘local guidepost’ for the regenerating sensory nerve to penetrate the ganglion in the appropriate locus. Indeed, afferent fibers always entered the ganglion at the proper place, whether the ganglion was an implant with no normal CNS connexions, or an autogenous ganglion with the motor nerve severed at its origin. It may be that pioneering regenerate afferents could be guided by adjacent motor fibers without making permanent attachments to them. Once the first group of sensory fibers had terminated and separated from the motor nerve, then afferent fibers developing in the following molts would be guided by those already in place.
Exploratory behavior of pioneer fibers in growth and regeneration has long been known to occur in vertebrates. Since time-course observations were not made in our studies, we cannot state that there were no temporary contacts formed by regenerating axons which were subsequently aborted. Regenerating afferent fibers might have first contacted the implant and then dissociated to grow further until they reached the host ganglion. Why then were persistence of sensory contacts in the implant associated only with experiments involving cut and subsequently regenerated motor nerves ? Since motor fibers tend to make attachments between host and implanted nerve tissue, it seems possible that the regenerating host N 10v could have made temporary contact with the implant, thus providing a guidance cue for pioneering afferents.
In their work with developing optic systems in Daphnia, Lopresti, Macagno & Levinthal (1973) pointed out that the growth cone is not necessarily a feature of all growing axons. They found that the lead axons from retinal ganglion cells growing in to meet lamina neuroblasts were the only ones of the group to produce growth cones. Furthermore, they suggested that this transient structure may be functionally associated with the recognition of cell surfaces or spatial relations by pioneer fibers. This idea is supported by a more recent paper (Lopresti, Macagno & Levinthal, 1974), where transiently formed gap junctions were found at these contact points.
The question remains, if cereal afferents depend upon intact motor nerve for pathway guidance, how are they able to regenerate to their normal locations when motor nerves are cut? The postulated dependence on contact with motor nerves could be facultative and not absolute. We can suggest two possible mechanisms : regenerating motor axons may have reached the periphery before cereal afferents terminated in the ganglion, thus providing pathway guidance. Alternatively, it may be that there exists a hierarchy of cues available to a regenerating nerve, so that in the absence of a primary cue (in this case, the intact motor nerve) a secondary one provides the requisite factor. Such a secondary cue might be factors diffusing from the stumps of cut nerves.
Terminal contact formation
The evidence presented above suggests that once a regenerating axon has established a contact, then the fibers following are guided into the same pathway. However, at the level of synaptic terminals, other processes must be operating, for only a small proportion of cereal afferents in cut motor nerve experiments remained in the implant, as judged by CoCl2 techniques.
Reduction in dendritic fields after axotomy has been reported in work with vertebrates (e.g. Cerf & Chacko, 1958; Sumner & Watson, 1971). Another post-axotomy phenomenon is that of ‘somatic stripping’, the loss of synapses on cell bodies and proximal dendrites. This has been measured electrophysiologically (Kuno & Llinas, 1970a,b; Mendell, Munson & Scott, 1974), and observed ultrastructurally in the form of glial displacement of synaptic terminals on axotomized motoneurons (Blinzinger & Kreutzberg, 1968; Kerns & Hinsman, 1973). In our experiments, the damage in the form of loss of axon volume of giant interneurons of the implant could have resulted in a deficit or deactivation of synaptic sites for incoming fibers.
Trophic effects in cereal regeneration
A trophic role of efferent axons has been inferred by Drescher (1960) who reported that antennae do not regenerate in Periplaneta americana if deprived of all neural connexions with the brain, and by Schoeller (1964) who concluded that differentiation of transplanted antennal disks of antennae, compound eyes and palps of Calliphora erythrocephala required innervation from the host.
An extensive exploration of the trophic role of the central nervous system in the regeneration of cerci in Acheta domesticus is closely relevant to the present study and must be critically assessed in relation to our own findings.
Rummel (1970) reported the effects of surgical intervention on cereal regeneration in Acheta domesticus larvae. The object of the study was to determine the influence of the nervous system in the regeneration of sensory appendages.
Separation of the cercus from its nerve supply was accomplished by a deep incision through the body wall in the middle of the angle formed by the cercus and the adjacent abdominal margin. Presumably both motor and sensory nerves were severed by this operation, but Rummel did not distinguish between the two. Cerci degenerated disto-proximally until regeneration set in. The regenerated cercus was formed partly from the existing cereal base; in some cases a small regenerate was formed in the wound region, just medial to the normally-situated cercus. Rummel suggested that maintenance of a normal cercus depended on the presence of an uninterrupted nerve supply, and that production of a regenerate must await the arrival of regenerated nerves in the wound area.
A comparable operation, but with the insertion of barriers of mica, plastic film, or aluminum foil yielded more varied results. The aim was to prevent normal regeneration or induce supernumerary regenerates by diverting nerves from their normal course. Mortality of operated animals was higher, and cereal regeneration reduced. Various abnormalities included distal degenerative changes in the contralateral cercus. Rummel inferred the need for a centrifugally moving substance from the central nervous system, requiring neural connexion between the central ganglion and the site of regeneration following the well established vertebrate model. Several questions arise from this interpretation. First, the innervation of the cereal musculature, the only possible pathway for a neurogenic stimulus to cereal regeneration since sensory fibers arise from the cercus itself, is entirely extrinsic to the cercus and has no known terminations on the epidermis in the tissue from which the sensory regenerate arises.
Secondly, the diminution of cereal regeneration obtained by interposition of barriers could be due to the diversion of hemolymph and tracheation by the barrier and scar tissue. The cerci lack pulsatile organs at their base which ensure hemolymph circulation in other elongate appendages such as antennae, and may thus be sensitive to alterations of patterns of hemolymph flow which could lead to local stagnation and thus to tissue necrosis.
A simple method for testing the importance of the motor nerve in cereal regeneration would be to remove the host terminal ganglion, thereby preventing efferent regeneration. This experiment was attempted with 7th, 8th and 9th stage larvae but none survived beyond ten days and none molted. Presumably the loss of regulating functions such as control of water balance accounted for the mortality. Cutting all nerves to the host terminal ganglion and leaving it in situ gave the same negative results in four animals. A decisive test of the need for centripetal innervation by means of such simple surgical intervention thus seems to be precluded. The less drastic expedient of severing the cereal motor nerve alone at its exit from the terminal ganglion had no significant effect on the time span or quality of cereal regeneration. We conclude that firm evidence for a trophic role of central innervation in the development of cereal regenerates has not yet been demonstrated.
ACKNOWLEDGEMENTS
This work was supported by Developmental Biology Training Grant HD-00266 from NICHHD and by research grant NB 07778. We thank Drs John Palka, Eldon Ball and Robert Seecoff for critical reading of the manuscript and for helpful discussion.