Positional Discrimination and re-development of Synapses in the Leech Whitmania Pigra

The leech preparation used in this investigation has a number of advantages: (1) the sensory and motor neurones are identifiable and innervate defined receptive fields (Nicholls and Baylor, 1968) or muscle groups (Bowling et al. 1978); (2) in Whitmania pigra the axons of sensory and motor neurones have a stable structural pattern and show symmetrical arborizations in the ganglia and connectives; (3) a pair of identified neurones, the anterior pagoda (AP) cells, can discriminate and integrate the PSPs evoked by mechanoreceptive signals from both sides of the body. These neurones express their ability for spatial discrimination both centrally and peripherally. Centrally, discrimination occurs at the level of the synapse, where the strength of excitation varies depending upon the position of the receptive field. The second site is located at the growth cone of the peripheral terminal which differentiates the position to be innervated. The two problems, that of forming connections and that of signalling, involve very different mechanisms and different experimental approaches.

The morphology of leech neurones was reviewed by Muller et al. (1981). Recently, the stability of axon architecture in embryonic leech has been recognized (Gao and Macagno, 1987). In order to assess the stability of axon architecture in Whitmania pigra, individual neurones were visualized by injection with horseradish peroxidase and the number of cells showing morphological variation was expressed as a percentage of the total number of neurones examined. In some cells no variation was found; 100 % of T cells (N=52), P cells (N =31) and N cells (N =177) had a stable morphology. Of the Retzius (R) cells, 88 % (71 of a sample of 81) showed a consistent axon pattern. The main variations of Retzius cell architecture occurred on the collateral branches towards the anterior connectives (Wang and Zhang, 1990). Of the 260 AP cells examined, 96 % had a stable morphology; from the second to the nineteenth segmental ganglion, the branching pattern was the same.

In the leech the T, P and N mechanoreceptive cells supply receptive fields in the periphery. Similarly, the innervation of the annuli erector muscles on both sides of the body by the annuli erector neurones (AE) is symmetrical. Several neurones also show symmetrical axonal arborization within the mature central nervous system. In Whitmania pigra, T, P, N, AE and Retzius cells all show symmetrical axonal branching (Wang and Zhang, 1990) and AP neurones show symmetrical axonal branching throughout the chain of segmental ganglia (Fig. 1).

The central nervous system (CNS) of the leech consists of a series of ganglia linked by connectives. Neurones send out their axons to the adjacent anterior and posterior ganglia. Retzius cells send out a collateral axon from each of these branches towards the anterior and posterior roots separately, and also to the connectives. These branches are always located in a particular part of the connectives. Retzius and N cell axons are located in the ventral-lateral area of connectives (Fig. 2). There is no obvious difference in the position of these two neurones.

The AP neurone contains both acetylchonine (ACh) (Wallace, 1981) and FMRFamide (Kuhlman et al. 1988); its function is unknown, and its connections with other cells have not been ascertained. In Whitmania pigra, rectifying electrical synapses occur between the receptive cells (T, P, medial N) and the AP neurones. When ganglia were bathed in normal Ringer’s solution no facilitation of excitatory postsynaptic potentials (EPSPs) could be found when paired electrical stimuli were delivered to medial P cells. When a depolarizing current was passed into a medial P cell bathed in a Ringer’s solution containing 10 mmol I^-1 Ca²⁺ or 15 mmol 1^-1 Mg²⁺, EPSPs recorded in the contralateral AP neurone were the same as those recorded in normal Ringer. When a train of depolarizing impulses was passed to the medial P cell, synaptic facilitation and suppression were not seen on AP neurones; the coupling ratio recorded from a medial P cell when the hyperpolarizing current was passed into an AP neurone was only 0.01. This indicates that there is a rectifying connection between medial P and AP cells. The inhibitory chemical synapse from a medial N cell to an AP cell seems to be a polysynaptic connection. The connections between other cells and the AP neurone are shown in Table 1. In culture, the synapses formed between medial N and AP cells were either rectifying or non-rectifying electrical connections, but only rectifying connections formed between lateral N and AP cells (Vyklicky and Nicholls, 1988).

To maintain a constant level of stimulus intensity from different sensory cells, depolarizing currents were passed into the somata of homologous dorsal, lateral and ventral T cells so that three spikes were produced. These spikes elicited EPSPs on the AP neurone showing obvious differences in amplitude. Thus, we can place in order of rank the strength of the excitation evoked by T cells with different receptive fields. In the three modalities of the mechanoreceptive cell, the rules are: (1) that the EPSPs evoked by the contralateral receptive cell are stronger than those evoked by the ipsilateral receptive cell; and (2) that the EPSPs elicited by the neurone with a dorsally situated receptive field (dorsal T cell) are stronger than those from the neurone with a ventrally situated receptive field (ventral T cell), while the neurone with a lateral receptive field (lateral T cell) produces EPSPs in the middle of the range.

The AP neurone can sum the EPSPs elicited by the bilateral medial P cells from both sides (Fig. 3). The contralateral medial N cell elicits either an electrically induced EPSP or a chemically induced IPSP on the AP neurone, but the ipsilateral medial N cell produces only an electrically induced EPSP. The passage of a train of Impulses simultaneously into the bilateral medial N cells resulted in the suppression of the spontaneous potentials of the AP neurone. The contralateral lateral N cell excited the AP cell to discharge action potentials. When a train of current impulses was applied simultaneously to the contralateral lateral and medial N cells, the frequency of impulse discharge from the lateral N cell was decreased because of the suppression exerted by the medial N cell. An increase in the frequency of the current passing through the medial N cell enhanced the suppression effect upon the lateral N cell (Fig. 4).

The peripheral innervation of the AP neurone is not fully known, so we can only comment on its central function. The arborization of the AP neurone is morphologically similar to that of the L and the AE motor neurones, with contralateral innervation and with few synaptic varicosities. The axon branches of the AP cell can be seen in the anterior root, in the posterior root and in its dorsal branch, indicating that it is probably a motor neurone. The function of the AP neurone is similar to that of the L and AE neurone, since stimulating the T, P and N cells can elicit excitatory monosynaptic potentials in the ipsilateral and contralateral AP, L and AE neurones (Nicholls and Purves, 1970). However, for the AP cell, T and P cells elicit synergistic potentiation, while the medial N cell produces an antagonistic depression. In addition, the AP neurone can discriminate and integrate the afferent signals from the T, P and N receptive cells. There are no electrical or chemical connections linking the two AP cells bilaterally (Sunderland, 1980), but the L and AE cells do have such a connection.

In a number of different animals changing or rotating the position of the target area has been used to study neural regeneration (Sperry, 1943; Miner, 1956; Jacobson and Backer, 1968; Bloom and Tompkins, 1976; Jacobson, 1978; Murphey, 1985; Koopowitz and Holman, 1988; Edwards, 1988). The results obtained from such experiments were variable. Nerves in the cricket or in Xenopus reinnervated their original target, but in the leech neurones kept their original spatial orientation after the body wall had been rotated in any one of three dimensions, even when the nerve roots had been crushed after rotation. This means that, in the course of reconstruction, regrowing nerves take the same pathway as they do in development and are uninfluenced by the position of the original target. Fig. 5 shows diagrammatically that the T cells reinnervate the original position of their receptive fields. This is also true of all the other mechanoreceptive cells and the AE motor neurone (Table 2). After the direction of the annulus erector (AE) muscles has been changed by 90°, the reinnervated muscles contract at right angles to the rest of the annuli. A more interesting phenomenon is that the three types of mechanoreceptive cells, T, P and N, are repaired synchronously. Similarly, the innervated areas of receptive and motor neurones on each side and the receptive and/or motor functions on each side are also repaired synchronously. Fig. 6 shows the bilateral reinnervation of a lateral P cell.

The specificity of synaptic connections between identified neurones in the leech has been convincingly demonstrated in situ and also in culture (Nicholls, 1987). Recently, a further report has shown that when a single neurone is implanted into a cultured ganglion, novel synapses are formed and are characterized by features such as the fluctuation of postsynaptic potentials and a lower electrical coupling ratio (Zhang, 1989). In dissociated cell culture it is possible to induce the formation of electrical synapses; the electrical coupling ratio starts at a weak level, becoming stronger within 2 days (Liu and Nicholls, 1989). In contrast, the coupling ratio of synapses formed in situ is much higher. For example, in the ganglia, the electrical coupling ratio is not less than 0.5 when depolarizing current is passed into R-R cell pairs, and ranges from 0.5 to 0.9 for both original and novel connections formed by the third day. However, in dissociated cell culture, the coupling ratio ranges from 0.03 on the fourth day to 0.07 by day 21 (Table 3). An obvious difference between the in situ and the dissociated cell culture situation is the presence of glial cells in situ. The coupling ratio of glial cells is about 0.3 under hyperpolarizing current (Kuffler and Potter, 1964). After glial cells have been killed, the reformed S-S cell connection has a coupling ratio of 0.04 with depolarizing current and 0.06 with hyperpolarizing current (Elliott and Muller, 1983).

Electrical synaptic connections have been reported between many neurones in the leech; for example, between pairs of Retzius, T, L, AE and S cells (Muller et al. 1981). However, until now there have been few electron micrographs of these areas in the leech.

The gap junction between nerve terminals in the leech displays a gap of approximately 2nm in width. Fig. 7 shows an electron micrograph of a putative gap junction, filled with lanthanum, between two terminals.

The entire central nervous system is composed of two kinds of cell, neurones and glial cells, but ‘what specialized structures or junctions (if any) occur between neurones and glia?’ (Nicholls, 1981). Recently a few articles have appeared that suggest arthropod-type gap junctions occur between neurones and glia, e.g. in houseflies (Saint Marie and Carlson, 1985) and in crayfish (Cuadras and Marti-Subirana, 1987).

The leech CNS is a classical model for research on the functional relationship between neurones and glia (Kuffler and Potter, 1964; Kuffler and Nicholls, 1966). Its outstanding advantage is that there are only 10 glial cells in each ganglion and these have defined positions. The six capsules forming each ganglion are distributed symmetrically, and there is one glial cell in each capsule, two glial cells envelop the neuropile lying longitudinally along the midline and two gliocytes occur on each side of the ganglion, one in the anterior and one in the posterior connective. Thus, the anatomical distribution of glial cells is also symmetrical.

The central nervous system of the leech contains identified neurones that have Symmetrical axonal arborizations. The anatomical distribution of glial cells is also symmetrical. This symmetry has been exploited by researchers for studying both the integrative mechanisms for localization of sensory stimuli and the regeneration of synaptic connections in the ganglion. Afferent signals from the receptor neurones result in synaptic inputs that are discriminated and integrated by the AP neurone.

Our results give rise to a number of questions including those of whether position perception depends upon a neuronal circuit in the CNS comparing input from both sides of the body, and how the bilateral synchrony of neuronal repair is achieved. Is regeneration dependent upon the rate of axoplasmic transport and are glial cells involved in the process, for example by the production of extracellular attachment or other molecules? The central nervous system of leeches should provide an ideal preparation for answering such questions.

This work was supported by grants from the Chinese Academy of Science, the Chinese Natural Science Foundation and the Chinese Education Commission.