Morphological methods were used to examine injury-induced growth of peripheral and central axons of nociceptive mechanosensory neurones in the ventrocaudal (VC) clusters of the pleural ganglia of Aplysia californica. Pedal nerve crush transected all axons in the nerve while leaving the overlying sheath largely intact. Immunohistochemical staining was performed with an antibody to a sensory-neurone-specific peptide, sensorin-A. Following bilateral crush of pedal nerve p9, which innervates the tail, sensorin-A immunofluorescence was lost distal to the crush site within 2 days. Fine immunopositive fibres began to invade the crush region within 5 days. These fibres arborized in the crush region and gradually extended down the crushed nerve. Immunopositive fibres were found near the tail within 3 weeks. Similar results were obtained after injecting individual sensory neurone somata in the tail/p9 region of the VC cluster with biocytin. Biocytin injections and horseradish peroxidase injections 3 weeks after ipsilateral pedal nerve crush revealed new fibres projecting rostrally from the tail/p9 region of the VC cluster and entering the pleural–cerebral and pleural–abdominal connectives. Such projections were never observed in control, uncrushed preparations. These results demonstrate that nerve injury triggers extensive growth of both peripheral and central processes of the VC sensory neurones.
Axonal growth is a common reaction to peripheral nerve injury. Injury-induced growth includes regeneration of severed axons (e.g. Fawcett and Keynes, 1990) and sprouting of neurites from either damaged or undamaged neurones at a distance from the site of injury (e.g. Weddell et al. 1941; Rotshenker, 1988; Bannatyne et al. 1989). The presence of large, identified neurones in gastropod molluscs, such as the snails Helisoma trivolvis and Lymnaea stagnalis, has proved useful for examining cellular mechanisms that control regeneration and sprouting, especially using in vitro preparations (e.g. Bulloch and Ridgway, 1989; Carrow and Levitan, 1989; Ridgway et al. 1991; Rehder et al. 1992). The gastropod nervous system also provides advantages for linking injury-induced neuronal growth to recovery of function in the whole animal (e.g. Allison and Benjamin, 1985; Moffett and Snyder, 1985; Cohan et al. 1987; Kruk and Bulloch, 1992; Syed et al. 1992).
The occurrence of apparently similar neuritic outgrowth in the central nervous system (CNS) or central ganglia after nerve injury (Kelly et al. 1989; Cameron et al. 1992; Woolf et al. 1992) and learning (reviewed by Bailey and Kandel, 1993) raises interesting questions about potential mechanistic and evolutionary relationships between these traditionally separate classes of plasticity (Walters, 1994; Walters and Ambron, 1995). A marine gastropod, Aplysia californica, has been utilized extensively to study neural mechanisms underlying behavior and learning (Kandel, 1979; Walters, 1994; Krasne and Glanzman, 1995). Among these studies have been several showing that sensory neurones display growth of new neurites, varicosities and synapses following one form of learning, general sensitization, or following the application of extracellular or intracellular signals associated with sensitization (Bailey and Chen, 1983, 1988; Nazif et al. 1991; Bailey et al. 1992). Although growth of neurites from A. californica neurones has been described in dissociated cell culture (e.g. Schacher and Proshansky, 1983; Ambron et al. 1985; Glanzman et al. 1989), there have been remarkably few studies of injury-induced morphological alterations in the whole animal. Fredman (1988) used retrograde staining and intracellular dye injection to show regeneration of axons in the pleural–cerebral connectives correlated with recovery of escape locomotion. Recently Ross et al. (1994) provided a detailed analysis of axonal regeneration and functional recovery in an identified buccal motor neurone of A. californica. However, injury-induced morphological alterations have not yet been reported in sensory neurones of A. californica or of any other mollusc.
Wide-dynamic-range, nociceptive mechanosensory neurones in the ventrocaudal (VC) clusters of A. californica (Walters et al. 1983) offer a particularly interesting opportunity to examine morphological reactions to nerve injury. These cells are known to express dramatic alterations after nerve injury, including long-term synaptic facilitation and increased excitability of their somata (Walters et al. 1991; Clatworthy and Walters, 1994; Gunstream et al. 1995). In addition, they display morphological alterations after learning-related treatments (Glanzman et al. 1990; Nazif et al. 1991). Moreover, their peripheral receptive fields display persistent enlargement following noxious stimulation, suggesting the occurrence of injury-induced growth in their peripheral arborization (Billy and Walters, 1989). As described in the preceding paper (Dulin et al. 1995), pedal nerve crush transiently abolishes the tail-evoked siphon reflex. Within 3 weeks, some axotomized VC sensory neurones re-establish receptive fields, which led us to hypothesize that one mechanism underlying recovery of the reflex is regeneration of injured VC cell axons. Furthermore, because axotomized VC sensory neurones could be activated antidromically by stimulation of connectives and nerves into which they do not normally project, we hypothesized that axonal injury causes these cells to sprout new fibres within the CNS that grow into aberrant pathways. Here, we test these hypotheses with morphological methods and show that nerve injury triggers both regenerative growth of the peripheral axons of VC sensory neurones and sprouting of neurites in central ganglia and connectives.
Materials and methods
Aplysia californica Cooper (Gastropoda; Opisthobranchia) (40–250 g) were obtained from Marinus (Long Beach, CA), Alacrity Biological Services (Redondo Beach, CA) and the University of Miami Aplysia Resource Facility. Animals were maintained in aerated artificial sea water (ASW; Instant Ocean) at 16–19 °C and fed a diet of romaine lettuce. One or more peripheral nerves leaving one or both pedal ganglia (Fig. 1A) were crushed using previously described procedures (Walters et al. 1991; Clatworthy and Walters, 1994; Dulin et al. 1995). In brief, nerve crushes were performed after anaesthetizing the animal by injecting ice-cold isotonic MgCl2 while the animal was cooled to 1–2 °C. For unilateral crush, most of the pedal nerves, including p7, p8 and p9, were crushed on one side of the animal (100–250 g) approximately l cm from the pedal ganglion (Dulin et al. 1995). The side crushed was determined randomly. For bilateral crush (40–75 g animals), both p9 nerves were crushed with fine forceps approximately 1 cm from the pedal ganglia (Fig. 1A). After an animal had recovered from anaesthesia, both sides of the tail were gently pinched. Failure to elicit a siphon response confirmed that one or both sides of the tail was disconnected from the CNS (Dulin et al. 1995). At various times following crush, the animal was anaesthetized with a large injection of isotonic MgCl2 (>50 % of the animal’s body mass) and dissected for histological processing.
A. californica tissue was fixed overnight at 4 °C in 4 % paraformaldehyde in artificial sea water with the composition (in mmol l-1): NaCl, 460; KCl, 10; MgCl2, 55; CaCl2, 11; Hepes, 10 (pH 7.4). Following fixation, tissue was rinsed three times in phosphate-buffered saline (PBS; 10 min for each rinse) and then immersed overnight in PBS with 13 % sucrose to cryoprotect the tissue. Samples were frozen in liquid isopentane at -40 to -60 °C, and 10 μ,m frozen cryostat sections were cut and collected on gelatin-coated glass slides.
Frozen sections were incubated overnight at 4 °C in a blocking solution consisting of 10 % normal goat serum (Jackson Immunoresearch, West Grove, PA), 1 % bovine serum albumin (BSA) and 0.4 % Triton X-100 in PBS. After a single quick rinse in washing buffer (0.1 % Triton X-100 in PBS), sections were incubated overnight at 4 °C with a rabbit polyclonal antiserum raised against sensorin-A (Brunet et al. 1991; a gift from E. Kandel and R. Hawkins) at 1:400 in PBS with 0.4 % Triton X-100. Following the primary antibody incubation, ganglia were rinsed four times in washing buffer (10 min each rinse) and then incubated overnight at 4 °C with a dichlorotriazinylaminofluorescein (DTAF)-conjugated goat anti-rabbit secondary antibody (Jackson Immunoresearch) at 1:100 in PBS with 0.4 % Triton X-100. After three rinses in washing buffer (10 min each rinse), sections were mounted in a 3:1 solution of glycerol in PBS with 1 % n-propyl gallate, a reagent used to reduce photobleaching. Ganglia were viewed with a Zeiss Universal microscope equipped with epifluorescence optics (filter set H7909 for fluorescein) and photomicrographs were taken on Ilford XP2–400 film.
For ionophoretic injection into sensory neurones, a 4 % solution of biocytin (Sigma, St Louis, MO) was drawn into the tip of an electrode; the electrode was then backfilled with a solution of 0.33 mol l-1 KCl and 10 mmol l-1 Tris, pH 7.5. Sensory neurone somata in the tail/p9 region of the VC cluster in surgically desheathed pleural ganglia (Fig. 1B) were alternately hyperpolarized and depolarized by 30 mV from resting potential (at 0.5 Hz) by passing current through the injection electrode for 30–40 min. Ganglia were then placed in isotonic L15 culture medium (Schacher and Proshansky, 1983) at 4 °C for 20–42 h to allow time for the biocytin to move down the axons. Tissue was fixed in 4 % paraformaldehyde in ASW overnight at 4 °C, rinsed three times in PBS (10 min each rinse) and then cryoprotected in PBS with 15 % sucrose. Frozen cryostat sections were prepared as above. Sections were incubated overnight in blocking solution. Following a single quick rinse, sections were incubated in Biodipy (Molecular Probes, Eugene, OR) at 1:100 in PBS with 0.4 % Triton X-100. After three rinses in washing buffer, sections were mounted, viewed and photographed as above.
Horseradish peroxidase fills
Sensory neurones in the tail/p9 area of the VC cluster (Fig. 1B) were pressure-injected with a solution of 4 % horseradish peroxidase (HRP) in 1 mmol l-1 KCl and 6 mmol l-1 Fast Green, using procedures similar to those described by Nazif et al. (1991). Injection electrodes (3–5 Mil) were connected to a Picospritzer II (General Valve, Fairfield, NJ) which delivered 2–3 pulses (10–20 ms each) at 140–170 kPa. The Fast Green allowed visual monitoring of each injection. Injected ganglia were left in culture medium at 4 °C for 20–24 h. Preparations were then fixed in a solution of 2.5 % glutaraldehyde and 30 % sucrose in 0.1 mol l-1 phosphate buffer (PB), pH 7.3, for 1 h. The ganglia were washed with PB and placed in a solution containing diaminobenzidene (Vector Laboratories). Tissue was then reacted using 0.003 % peroxide for 15 min, dehydrated in a graded series of ethanol concentrations (20, 50, 75 and 95 %) and cleared in methyl salicylate. Whole-mount preparations were placed on a glass slide in Permount, viewed on an Olympus BH-2 microscope, and photographed with Kodak TMAX 100 (black and white) or Ektachrome Tungsten.
General effects of nerve crush
Nerve crush caused the evident transection of all fibres at the crush site, while leaving the overlying sheath apparently intact, as illustrated by the phase micrograph of Fig. 2B. Transection of the fibres was also indicated by visual inspection of crushed nerves under the dissecting microscope (not shown). Furthermore, in every case that the nerve was tested electrophysiologically within 1 week after nerve crush, sensory neurone action potentials failed to be conducted through the crush site (Dulin et al. 1995). Approximately 1–2 weeks after nerve crush, the signs of fibre transection at the crush site disappeared, sensory neurone conduction through the crush site began to be restored (Dulin et al. 1995) and the crushed region consistently swelled to about twice the diameter of the adjacent nerve regions. The swelling persisted for as long we examined the nerves, up to 60 days post-crush. This swelling provided a useful marker for the crush site in both morphological and electrophysiological experiments (see Dulin et al. 1995). Large inclusions (up to 100 μm) were observed in the region of the swelling (Fig. 2D) and distally (Fig. 5B–D,H). Failure of the inclusions to stain with the fluorescent nuclear dye Hoechst 33,258 indicated that these profiles were acellular (Fig. 2E).
Immunohistochemical evidence for sensory neurone regeneration
Brunet et al. (1991) discovered a peptide, sensorin-A, that appears to be expressed in identified clusters of mechanosensory neurones and in no other neurones in A. californica. An antibody to this sensory-neurone-specific peptide provided a useful tool for examining injury-induced changes in anatomical properties of the pleural VC sensory neurones after nerve crush. Because the normal anatomical features and behavioural functions of the VC sensory neurones innervating the tail are particularly well defined, we focused on these cells. The somata of the tail sensory neurones are restricted to one margin of the cluster (Fig. 1B). Each neurone in this region has a single axon that projects exclusively through nerve p9 (Walters et al. 1983; Billy and Walters, 1987; Zhang et al. 1993; Dulin et al. 1994). Sensorin-A immunofluorescence in control animals (N=9) revealed fibres in the pleural and pedal ganglia and in all pedal nerves, including p9 (Figs 3, 4). Immunoreactive fibres typically travelled along the ventral side of the pleural–pedal connective, arborizing (Figs 3A, 4A) in the pedal ganglion in the vicinity of identified tail motor neurones that receive strong monosynaptic inputs from tail sensory neurones in the ipsilateral VC cluster (Walters et al. 1983). Numerous fibres exited the ganglion in nerve p9, where they could be seen to be restricted to the ventral part of the nerve in longitudinal sections (Fig. 4B) and cross sections (not shown). The high signal-to-background ratio obtained with the sensorin-A antibody permitted the resolution of numerous individual fibres in the nerve (Fig. 4C).
The effects of p9 crush on sensorin-A immunofluorescence was investigated in 17 animals. Axonal transection did not eliminate the immunofluorescence distal to the crush site 6 h after p9 crush, indicating that the severed distal fibres still contained immunoreactive peptide at this time (Fig. 5A). However, by 48 h post-crush (Fig. 5B) immunoreactive fibres distal to the crush site were no longer visible. Using the characteristic nerve swelling as a marker for the site of nerve crush, we examined the recovery of sensorin-A immunofluorescence in nerve p9 distal to the crush. By day 5 post-crush, immunopositive fibres had invaded the site of injury and the region just distal to the crush site (Fig. 5C). These new fibres were finer than those observed in control animals and finer than the original fibres seen distal to the crush site before they degenerated (e.g. Fig. 5A). It was common for the new fibres to have varicosities and to be highly arborized in the region of the crush (as determined by focusing through longitudinal sections). The number of new fibres appearing just distal to the crush site increased progressively from day 5 to day 12 (Fig. 5C–F). With increasing time after the crush, immunopositive fibres were found at greater distances distal to the crush site (Fig. 5E–H). Three weeks after the crush, immunopositive fibres were found near the insertion of nerve p9 into the tail (Fig. 5H).
Whereas tail sensory fibres in uncrushed nerves were confined to the ventral part of p9, regenerating sensory fibres were found throughout the crush site and emerged distally with an essentially random distribution in p9. We found no evidence in any of these experiments for preferential growth of regenerating sensorin-A-containing fibres in the ventral part of p9. At the ‘regenerating front’, sensory fibres were fine and highly arborized, but behind this front they became less arborized and thicker with time. At the swollen crush site, however, fine extensively arborized fibres remained even 3 weeks after the crush. It was not possible to determine whether these represented persistent, stable structures or processes that were still actively extending and retracting more than 2 weeks after their initial regrowth into the crush site.
Dye-fills of individual sensory neurones demonstrate axonal regeneration
Biocytin is a particularly effective dye for intracellular fills in molluscs (Ewadinger et al. 1994). We used biocytin to examine the morphology of individual sensory neurones 3 weeks after crushing nerve p9. In two animals, we identified sensory neurones with presumptive axons in nerve p9 by electrically stimulating the nerve and recording action potentials in the soma (see Dulin et al. 1995). One to three cells activated by p9 stimulation were then filled with biocytin in each pleural ganglion in each animal (N=8 cells in total). Each filled sensory neurone had a single axon in the pleural–pedal connective on both the control (not shown) and nerve-crushed sides (Fig. 6A), which passed through the pedal ganglion and entered nerve p9. On the nerve-crushed side, the biocytin-containing fibres traversed the crush site and continued through p9 for another 1–2 cm to where the nerve had been cut during dissection (Fig. 6B–E). In the region of the crush, the sensory fibres branched and meandered (Fig. 6B,C). Once they emerged from the swollen region, they followed a straight path and arborizations were no longer observed (Fig. 6D,E).
Morphological evidence for central sprouting by injured sensory neurones
As described in the preceding paper (Dulin et al. 1995), we have obtained electrophysiological evidence consistent with the growth of new neurites into central connectives and peripheral nerves by injured VC sensory neurones. To seek anatomical evidence of injury-induced sprouting in the CNS by sensory neurones, we injected either HRP or biocytin into sensory neurone somata in the tail/p9 region of the VC cluster (Fig. 1B). Previous studies indicated that sensory neurones in this region send single axons directly into the pleural–pedal connective and never send axons into the pleural–cerebral or pleural–abdominal connectives (Walters et al. 1983; Zhang et al. 1993; Dulin et al. 1994; A. Billy and E. T. Walters, unpublished observations). In initial experiments, we injected HRP into 5–10 sensory neurone somata in each pleural ganglion 3 weeks after unilateral pedal nerve crush. All stained cells on the uncrushed side had a single fibre that travelled directly into the pleural–pedal connective; no filled fibres were found rostral to the group of injected somata, projecting towards either of the more anterior pleural connectives (Fig. 7A). In contrast, in three of four animals, HRP-containing cells had more than one major fibre, one of which travelled towards, and sometimes entered, the pleural–cerebral or pleural–abdominal connective (Fig. 7B). In these animals, the arborization of the filled sensory neurones in the neuropile beneath the somata appeared denser and the organization of the fibres much less regular on the crushed than on the uncrushed side.
Biocytin was injected into tail/p9 sensory neurones that displayed electrophysiological responses to stimulation of the pleural–cerebral or pleural–abdominal connectives 3 weeks after ipsilateral pedal nerve crush (Dulin et al. 1995). In four animals, 1–3 VC sensory neurone somata were successfully filled in each pleural ganglion. On the uncrushed side, all seven filled cells sent a single fibre posteriorly into the pleural–pedal connective. No neurites projected anteriorly or entered the pleural–cerebral or pleural–pedal connectives. In contrast, all seven filled cells on the nerve-crushed side displayed one or more fibres projecting anteriorly (Fig. 7C) towards the pleural–cerebral and/or the pleural–abdominal connectives. Each also sent a fibre posteriorly into the pleural–pedal connective. As shown in Fig. 7D, some of the anteriorly projecting fibres entered an anterior connective. Two cells had fibres entering both of the anterior connectives, two cells had fibres entering the pleural–cerebral connective and one cell had a fibre entering the pleural–abdominal connective. The two remaining cells had fibres that projected anteriorly, but were not observed to leave the pleural ganglion.
Because biocytin passes through at least some electrical synapses in A. californica (Hickie, 1994) and other molluscs (Ewadinger et al. 1994), we also looked in these same animals for evidence of injury-induced formation of electrical synapses in the CNS. Biocytin was not found in any neuronal somata in the pleural or pedal ganglia that had not been injected. This suggests that, if novel electrical synapses are formed by injured VC sensory neurones, they are formed outside the CNS.
Regeneration of severed sensory neurone axons
We have found that the VC sensory neurones display regenerative growth of peripheral axons following nerve crush that transects their axons but leaves the overlying sheath essentially intact. Axonal regeneration under similar conditions has been described in other mechanosensory neurones having centrally located somata and peripheral receptive fields, notably those in the leech (Van Essen and Jansen, 1977; Bannatyne et al. 1989) and mammals. In mammals, sensory neurones with somata in dorsal root ganglia have been utilized to examine a variety of cellular and molecular concomitants of axonal regeneration (e.g. Wujek and Lasek, 1983; Oblinger and Lasek, 1984; Bisby and Keen, 1986; Kanje et al. 1986; Jenkins et al. 1993; Mearow et al. 1994). The VC sensory neurones in A. californica, like the N, P and T sensory neurones in the leech, offer special opportunities to examine molecular and functional aspects of regeneration because the functions of individual cells are well defined and intracellular manipulations (e.g. injection of putative molecular signals or blockers – see Walters and Ambron, 1995) can be performed readily on one or more identified cells during various stages of regeneration.
Although we have not yet conducted a systematic analysis of the time course of axonal regeneration in VC sensory neurones, the appearance of new fibres at the crush site and their extension into the distal segment of nerve p9 roughly parallel the reappearance of spike conduction through the crush site and the recovery of functioning receptive fields on the tail (Dulin et al. 1995). Thus, some regenerating fibres that label with the sensorin-A antibody traverse the crush site 5 days after injury, extend farther into p9 as time goes by and are found near the tail by 3 weeks after injury, when receptive fields on the tail are being re-established. Recent results obtained with sensorin-A immunostaining have revealed characteristic helical coil endings in the muscle layer of A. californica body wall in the tail and siphon (I. Steffensen and C. E. Morris, in preparation). It will be interesting to examine changes in sensorin-A immunoreactivity in body wall disconnected from the CNS by nerve injury and in the density or appearance of these putative mechanosensory structures during loss and recovery of function.
Demonstrations by sensorin-A immunofluorescence and biocytin injections of the extension of VC sensory neurone axons well beyond the crush site suggest that regenerating tail sensory neurones from the VC cluster contribute to recovery of the tail-evoked siphon reflex described in the preceding paper (Dulin et al. 1995). However, we pointed out that the failure of the centrally conducted axon responses to nerve shock to be restored earlier than the behavioural responses (even though the crush site was relatively close to the VC somata) indicates that the earliest stages of reflex recovery are not due to regeneration of VC sensory neurone axons into the tail. Two mechanisms that, in principle, could produce relatively rapid restoration of sensory input to the CNS from the denervated half of the tail are (1) regeneration of the central axons of peripheral sensory neurones (or sensory interneurones) across the relatively short distance between the central nerve crush site and their central targets and (2) reconnection of regenerating VC neurites to surviving axons in the distal stump of nerve p9. The first mechanism seems likely to contribute, but has not yet been tested. Some of our data are relevant to the second mechanism.
Do proximal and distal stumps of severed axons reconnect?
The possibility that regenerating neurites form electrical synapses with surviving distal stumps of severed VC sensory neurone axons needs further examination. The disappearance of sensorin-A immunoreactivity from axon segments distal to the crush within several days might be due to degeneration of these segments (see below). However, the disappearance of sensorin-A does not necessarily indicate that the axon segments are lost. It might simply be that the peptide is lost – perhaps by transport to the periphery. This peptide is sufficiently large (Brunet et al. 1991) that it is unlikely to pass through gap junctions and so should not be replenished in distal segments if reconnection occurs via electrical synapses. Our experiments with biocytin, which readily passes through at least some electrical synapses in molluscs (Ewadinger et al. 1994; Hickie, 1994), argue against novel coupling with other neurones in the CNS, but have not yet demonstrated either the presence or the absence of electrical synapses between regenerating fibres and surviving distal stumps. Ross et al. (1994) found no evidence for reconnection of proximal and distal axonal segments in an A. californica buccal motor neurone after nerve crush, but they did find that the distal stumps survived for many weeks. It will be important to perform a systematic series of biocytin injections into VC sensory neurones during the period when reflex responses first start to recover (our biocytin injections were all 3 weeks post-crush) to see whether there is a sudden appearance of biocytin-filled distal axons extending far beyond the crush site. Such tests have demonstrated dye-coupling of proximal and distal axon stumps in the leech (Carbonetto and Muller, 1977; Camhi and Macagno, 1991).
If the formation of electrical synapses plays a role in reconnecting the tail to the CNS, this mechanism for functional recovery is likely to be temporary. Injury-induced formation of novel electrical synapses has been shown to be transient in other molluscs (e.g. Cohan et al. 1987). Furthermore, the large globular inclusions we observed distal to the crush resemble those seen in other injured A. californica axons and these have been linked to axonal degeneration, which occurs over a period of weeks (Ross et al. 1994). Regenerative growth of new axonal segments and slow degeneration of old axonal segments distal to the crush would indicate that over several weeks new growth into the receptive field gradually takes over sensory functions that might be temporarily maintained by electrical synapses between proximal and distal segments of severed sensory axons.
Central sprouting by sensory neurones after peripheral axon injury
In the preceding paper, we report the unexpected discovery that previously injured VC sensory neurones can sometimes be activated by stimulating central connectives or peripheral nerves that do not normally contain axons of the tested sensory neurones (Dulin et al. 1995). As shown in Fig. 7, dye-fills of sensory neurone somata in the tail/p9 region of the VC cluster ipsilateral to crushed pedal nerves reveal fibres entering tracts (the pleural–abdominal and pleural–cerebral connectives) into which the sensory neurones in this region do not normally project. These observations, and the lack of fibres entering these tracts on the uncrushed side, indicate that peripheral axon injury can trigger the sprouting of new fibres from these cells within the CNS. We do not yet know the destination of the new fibres, but it will be interesting to test (1) whether they follow alternative paths back to the original receptive field, and (2) whether they form synapses with other neurones that may be involved in recovery of function or protection of the peripheral region affected by the nerve injury (see Dulin et al. 1995).
Central sprouting of mechanosensory neurones in response to peripheral axotomy has also been described in the leech (Bannatyne et al. 1989) and mammals (Cameron et al. 1992; Woolf et al. 1992). In the latter case, the new growth was near the region of the synaptic terminals of the sensory neurones, suggesting the formation of new synapses in response to injury (see also Walters, 1994). Interestingly, in A. californica, unilateral nerve crush causes facilitation of sensory neurone synapses that lasts for weeks (Walters et al. 1991; Clatworthy and Walters, 1994). Long-term behavioural sensitization (a simple form of learning) in A. californica and long-term synaptic facilitation induced by application of 5-hydroxytryptamine or cyclic AMP are associated with an increased arborization and growth of new synapses by mechanosensory neurones (Bailey and Chen, 1983, 1988; Nazif et al. 1991; Bailey et al. 1992). To what extent do the morphological and physiological similarities in cellular reactions to peripheral injury and cellular responses during learning reflect shared mechanisms? Explicit comparative studies of injury-induced and learning-induced alterations in mechanosensory neurones in A. californica and other animals may provide insight, not only into mechanisms of regeneration but perhaps also into fundamental mechanisms of plasticity that are also utilized in some forms of learning and memory.
The authors are grateful to J. Pastore and K. Hensley for aid in the preparation of illustrations. This work was supported by an NSERC Canada research grant to C.E.M., and grants IBN-9210268 from the NSF and MH38726 from the NIMH to E.T.W. I.S. was the recipient of an Ontario Graduate Scholarship.