The regeneration pattern of two identified central neurones was studied in the leech Hirudo medicinalis. Anterior pagoda (AP) and mechanosensory touch-sensitive (T) neurones were stained in adult segmental ganglia, maintained in culture for 6–10 days. AP neurones, which normally project only to the contralateral nerve roots, sprouted extensively in all the available nerve paths during regeneration. Mechanosensory T cells, in the same experimental conditions, showed only a moderate growth and did not change their normal pattern of axonal projections. The observed differences in the growth pattern might account for the different electrophysiological responses to axotomy exhibited by the two types of neurone. Interruption of interganglionic connectives induced a moderate and stereotyped remodelling of the morphology of intact AP neurones, which was reminiscent of that transiently exhibited during embryonic development. This response was observed in 25 % of the AP neurones we examined.
Axotomy affects the electrical properties of the soma–dendritic membrane in a number of different types of neurones from several animal groups (Pitman et al. 1972; Goodman and Heitler, 1979; Kuwada and Wine, 1981; Faber, 1984; Pellegrino et al. 1984; Sernagor et al. 1986; Gilly and Brismar, 1989). Since these alterations occur during axon regeneration, it is tempting to assign a functional role to them. However, the significance of the electrophysiological changes is still unclear, mainly because different neurones display a wide variety of responses to axotomy.
In the central nervous system of the leech, the mechanosensory neurones, which have spiking somata, do not show obvious changes in the electrical properties of their cell bodies after axotomy (Bannatyne et al. 1989), whereas other neurones, such as the anterior pagoda cells (AP), develop excitable cell bodies (Pellegrino et al. 1984; Matteoli et al. 1986; Bigiani and Pellegrino, 1990; Simoni et al. 1990). Since both mechanosensory and AP neurones sprout after injury, the function of axotomy-induced alterations in axon repair has been questioned (Bannatyne et al. 1989). We have as yet no explanation for the disparity in the responses of these different cell types. The response to axotomy of AP cells is to convert non-spiking neurones into spiking neurones, suggesting that it is depolarization of the cell body rather than the changes in electrical properties themselves that are significant. However, some spiking neurones, such as the dorsal unpaired median (DUM) neurosecretory cells of the cockroach, increase the density of their Na+ currents after axotomy (Tribut et al. 1991).
Various factors, such as the cell type, the specific synaptic connections, the cell geometry and metabolic features might account for such different cellular responses. To clarify this point, we studied the morphology of both AP and T cells after axotomy, by examining the responses of the cells in segmental leech ganglia maintained in culture (Miyazaki and Nicholls, 1976).
Here we report that, although the mechanosensory T cells, as previously described (Miyazaki and Nicholls, 1976; Bannatyne et al. 1989), follow a conservative scheme of regeneration, the AP neurones send sprouting axonal branches into all the available nerve paths. This differential behaviour is discussed in terms of the strategies of embryonic growth exhibited by the two types of neurone.
A surprising result of this study is that, after crushing of the anterior and posterior connectives, intact AP neurones undergo a moderate and stereotyped remodelling and take on a morphology that is reminiscent of some stages of embryonic development.
Materials and methods
All procedures were performed on adult (2-to 4-year-old) specimens of Hirudo medicinalis L. obtained from a commercial supplier (Ricarimpex, France).
In the first series of experiments, single midbody (G7–G18) segmental ganglia were isolated and maintained in culture for a maximum of 3 weeks, in Leibowitz-15 medium (Gibco) supplemented with 6 % foetal calf serum (Gibco), 0.6 % glucose and 100 µg ml−1 gentamicin (Sigma), at 16 °C. Intracellular staining was performed with glass microelectrodes pulled from ultrathin capillary tubing (Clark) and filled with either Lucifer Yellow (Sigma) 5 % in 0.1 mol l−1 LiCl (Stewart, 1981) or horseradish peroxidase (Type VI, Sigma) 20 mg ml−1 in 0.2 mol l−1 KCl (Muller et al. 1981). AP and mechanosensory cells were identified on the ventral side of segmental ganglia on the basis of their position and their electrical properties. Identified neurones were stained by pressure injections through bevelled electrodes. Ganglia stained using Lucifer Yellow were kept in the dark for at least 30 min at 4 °C, in order to allow diffusion of the dye and to prevent photo-killing. Cells were observed and photographed in live preparations. Ganglia stained with horseradish peroxidase were incubated with 0.4 % benzidine and 0.15 % hydrogen peroxide, after fixation, and were then clarified using xylene so that they could be observed as whole-mount preparations.
The same procedures were applied to segmental ganglia freshly isolated from either normal or operated leeches. Operations, performed under chlorobutanol (0.15 %) anaesthesia 5–30 days in advance of staining, consisted of crushing the anterior and posterior connectives of three adjacent ganglia. In five animals, three adjacent ganglia were completely isolated.
Morphology of AP cells in control ganglia
Anterior pagoda cells of freshly isolated midbody ganglia send axonal branches only through the two contralateral nerve roots (Fig. 1A). This morphology is typical of leech motoneurones innervating the body wall muscles, but AP cells cannot be classified as motor cells since their targets are still unidentified. Although the extensions of secondary branches display detectable variations in different segments, the main pattern of projections is strictly repeated in the ganglia of all the segments selected for this study (G7–G18). We screened more than 100 AP cells in normal ganglia and found no exceptions to the described morphological pattern. In 40 % of cells in the normal ganglia, a very short branch enters the contralateral anterior (about 80 %) or posterior (about 20 %) connectives. This branch, illustrated in Fig. 1B, is 1–40 µm long, with a single tip ending with a knob-like structure.
Morphology of AP and T cells in cultured ganglia
The morphology of AP cells changed dramatically in ganglia that had been cultured for a few days. Sprouting was observed which pushed newly formed projections through every available nerve path, even those that normally contained no branch of the adult AP axon. Fig. 2 illustrates a typical growth pattern of an AP cell in a ganglion cultured for 6 days, in which sprouting collaterals run through the anterior and posterior connectives on both sides (Fig. 2A,B,D) and through the ipsilateral nerve roots (Fig. 2C). These alterations in the pattern of axonal projections were observed in about 80 % of the 48 cells injected. Three AP cells, stained with horseradish peroxidase, gave similar results.
AP cells of ganglia that had been isolated within the animal 7 days prior to staining, displayed the same pattern of regeneration as that observed in culture.
Under parallel experimental conditions, the morphology of mechanosensory T cells was studied. In all 17 cells that were injected, we confirmed the pattern of regeneration reported by other authors (Miyazaki and Nicholls, 1976; Bannatyne et al. 1989). T cells are conservative in their projections during regeneration: they increase varicosities in the neuropile, but never cross the ganglion midline and always grow in routes that the cell processes normally occupy. The only exception we found was a lateral T cell, which normally innervates the dorsal portion of the skin by sending an axonal branch only through the posterior nerve root (Nicholls and Baylor, 1968). As shown in Fig. 3, in our experimental conditions, this cell also sends an axonal collateral to the anterior root (arrow).
Remodelling of AP cells after crushing of interganglionic connectives
This experiment was designed to dissociate the effects of axotomy from those induced by the separation of adjacent ganglia. Crushing of both anterior and posterior connectives interrupted all the interganglionic axons without directly damaging AP cells. By 11–16 days after the operation approximately 25 % of AP cells exhibited stereotyped growth.
A typical example is illustrated in Fig. 4; two branches ran along the anterior and posterior contralateral connectives and always originated from the same region of the AP neurone arborization (see arrow in Fig. 1). These branches often showed ramified endings with growth cones at their tips (Fig. 4D). Similar results were obtained when horseradish peroxidase was used for staining and when the interruption of the connectives was achieved by cutting instead of by crushing. The longitudinal projections often ran along half the length of the connectives. They persisted for more than 1 month in our experiments.
Two major findings are reported here: the first is that AP cells and T cells differ qualitatively and quantitatively in the regenerative growth they undergo in isolated ganglia; the second is that, after crushing interganglionic connectives, adult intact AP cells extend new projections that are reminiscent of those transiently expressed by embryonic AP cells (Wolszon and Macagno, 1992).
Cultured leech ganglia provide a useful preparation for studying the regenerative growth of identified neurones in vitro (Miyazaki et al. 1975; Miyazaki and Nicholls, 1976; Wallace et al. 1977). Isolation of a segmental ganglion results in the axotomy of AP and T cells, the interruption of interganglionic and peripheral connections and a redistribution of sprouting-promoter molecules such as laminin (Masuda-Nakagawa et al. 1993). Both AP and T cells begin to sprout after a ganglion is isolated, but extension of the newly formed branches is consistently greater in AP cells. Assuming that in the leech, as in other preparations (Tribut et al. 1991; Brismar and Gilly, 1987; Gilly and Brismar, 1989), the alterations of the electrical properties of the cell body described by Bannatyne et al. (1989) and Pellegrino et al. (1984) are due to the insertion of newly synthesized ion channels, a metabolic reaction might underlie the dramatic changes in electrical properties observed in AP cells after axotomy. However, other factors, such as the ratio between the remaining and the excised portions of the neuronal arborization and the size of metabolically stable pools of ion channels precursors (Schmidt and Catterall, 1986), may be different in the two cell types.
The growth of AP cells in isolated ganglia did not seem to follow a path-specific pattern of repair and also differed qualitatively from that exhibited by T cells. The two different growth strategies might be related to the number or quality of central synaptic connections.
This different behaviour of the two cell types has an interesting counterpart in the growth strategies that mechanosensory and motor cells use during development. In Hirudo medicinalis, mechanosensory neurones extend to the periphery only those projections that are present in the adult, whereas motor cells initially grow extra processes that then retract after contacting their targets (Baptista and Macagno, 1988). It is worth considering whether other neurones that use the same strategy, such as AE motor cells (Wallace, 1984), show the same pattern of regeneration in isolated ganglia. Van Essen and Jansen (1975) reported abnormal but functionally appropriate connections between the L motoneurone and longitudinal muscle established through a segment of nerve connective implanted into denervated body wall approximately 4 months earlier. These authors suggested the possibility that the L motor cell, normally without longitudinal processes, could be triggered to sprout into the connective.
The second result reported in this paper is the surprising effect of crushing the interganglionic connectives. Adult, undamaged AP cells grow longitudinal projections and their pattern of growth is reminiscent of that of AP cells during embryonic development (Wolszon and Macagno, 1992). Both peripheral and central sprouting of undamaged leech neurones has previously been reported (Blackshaw et al. 1982; Gu and Muller, 1990; Muller and Gu, 1991; Masuda-Nakagawa et al. 1993). Our experiment also provides evidence that extrinsic factors induce an intact neurone to grow. According to Masuda-Nakagawa et al. (1993), the triggering factor could be the redistribution of microglia and laminin following the crushing of connectives.
Our results show three peculiar aspects: (1) unlike the pattern of growth in P cells reported by Masuda-Nakagawa et al. (1993), AP cells undergo a remodelling of their projections, in that they are induced to run in the interganglionic connectives, which normally contain no AP axonal branches; (2) the new branches always originate from the same portion of the cell arborization; (3) although the connectives of the two sides have been crushed, growth is expressed only in those contralateral to the AP cell body. These observations suggest that the newly formed projections may run along a pathway defined during development.
As previously reported during development of the AP cells in the embryo, the cell initially sustains both longitudinal (in the contralateral connectives) and lateral (in the contralateral nerve roots) projections (Macagno et al. 1990). Then, from embryonic day 14 (E14) to embryonic day 30 (E30), these two projections undergo very different fates, the longitudinal projection is retracted, while the lateral projection is stabilized. Two factors must coincide for retraction to occur: connection with the peripheral targets and the inhibitory effect of homologous neurones in adjacent ganglia (Gao and Macagno, 1987, 1988; Macagno et al. 1990). The latter is probably mediated by transient gap junctions between the longitudinal projections of adjacent homologous AP cells (Wolszon et al. 1994a,b). If one of the two above-mentioned factors is lacking at this stage, the retraction of longitudinal projections is blocked. After E30, the longitudinal processes become irreversibly retracted.
According to this scheme, an adult AP cell, undamaged and still connected with its peripheral targets, should remain quiescent and should not grow longitudinal projections even in the absence of homologues. The results of this study indicate that this is not the case. However, we have to take into account that the crushing of connectives may have multiple effects and does not simply remove AP homologues.
Our findings suggest that the signals that are sufficient to stop growth in the connectives after E30 become insufficient in adult neurones, when strong sprouting promoters, such as laminin, are redistributed. The growth of AP neurones in the crushed connectives appears to be attracted by a powerful stimulus and to be assisted by a recognition of landmarks left during embryonal development. Moreover, in many normal AP cells in different segmental ganglia, we found very short branches which originate from the region of the cell arborization that is involved in the growth of new projections. These branches could be the result of incomplete retraction of processes during development due to fluctuations in the efficacy of ‘stop’ signals exchanged between homologous AP cells. We cannot rule out the possibility that the new longitudinal projections that develop following the crushing of interganglionic connectives originate from the short branches that we observed in normal cells.
Interesting questions which remain unresolved are whether the newly formed projections make synaptic connections and what their targets are.
The authors wish to thank Dr Monica Pellegrini for her help and advice on Lucifer Yellow and horseradish peroxidase injections, Dr Florence Tribut for critical comments on the manuscript and Edoardo Biagetti for his excellent technical assistance.