Specificity of Connections Formed by Nociceptive Cells of the Leech in Tissue Culture

Isolated, identified leech neurones in culture retain their membrane properties and are able to form chemical and electrical synaptic connections with some targets but not others (Fuchs, Nicholls & Ready, 1981 ; Aréchiga, Chiquet, Kuffler & Nicholls, 1986). Retzius cells, anterior pagoda cells, heart motor neurones (HE cells) and L motor cells all become connected by non-rectifying electrical junctions in the dish. In contrast, P sensory cells have been shown to form electrical or chemical connections but only with some cells and not with others. Moreover, all the electrical synapses so far observed in P cells in the dish and in the ganglion show rectification (Aréchiga et al. 1986; Nicholls, 1987). Thus, P cells make rectifying electrical junctions with L cells, HE cells and AP cells. They are never electrically connected to Retzius cells but are the postsynaptic element in a chemical synapse involving 5-hydroxytryptamine (5-HT) as a transmitter (Fuchs, Henderson & Nicholls, 1982; Henderson, 1983). On the one hand these results demonstrate specificity; on the other hand they show that cells in the dish are able to form connections not normally present in the ganglion.

The aim of the present series of experiments has been to extend information about synapses formed in culture by different neurones. Interesting candidates for exploration are the nociceptive N cells. Within the ganglia there are two N cells on each side: one medial, one lateral. They both form excitatory chemical connections upon the L cell and AE motor cells in the same ganglion. The connection to the L cell exhibits a weak, rectifying electrical component (Nicholls, 1987). In addition, the medial and lateral N cells on one side of a ganglion show differences in their membrane properties, in their chemosensitivity and in the connections they make on postsynaptic targets (Johansen & Kleinhaus, 1985).

From earlier work it was known that N cells retained their membrane properties and formed mixed chemical and electrical synapses upon L cells in culture, the electrical component showing rectification (Fuchs et al. 1981). We have now paired N cells with a variety of identified cell targets to observe whether the electrical connections are rectifying or non-rectifying, whether chemical synapses occur and whether connections in the dish resemble those observed in the animal. The question of how Retzius cells become connected to N cells was of interest in relation to specificity. Would they make pure chemical synapses without an electrical com-ponent (as on P cells) or electrical synapses or neither?

The methods for identifying cells in leech ganglia, Hirudo medicinalis, removing them and maintaining them in culture have been described previously (Fuchs et al. 1982). All the experiments were performed by removing cells with loops of nylon monofilament and plating them on polylysine. Recent experiments have shown that this procedure enhances synapse formation compared to the technique of removing cells by suction after enzyme treatment (Dietzel, Drapeau & Nicholls, 1986). Cells were plated on polylysine to ensure that the sprouts were short. Under these conditions the cells are isopotential (Ross, Aréchiga & Nicholls, 1987) and current spread from cell to cell can provide a reliable index of junctional properties (Davis, 1986). Only pairs of cells with comparable, high input resistance (>50MΩ) were used. Culture medium consisted of L-15 with 2% foetal calf serum and 0·1 mg ml⁻¹ Gentamycin. The cells were cultured in microwells as closely apposed pairs. Electrical recordings were made with microelectrodes filled with potassium acetate of resistances 20–40 MΩ. Amplification was through Almost Perfect Electronics amplifiers, signals being recorded on a Hewlett Packard tape recorder. The cells were identified by visual inspection in ganglia before removal. In culture each type of cell retains its characteristic membrane properties and action potential configuration by which it can be recognized. Occasionally T cells were isolated instead of N cells but this became apparent immediately upon recording. One source of ambiguity is that both medial and lateral N cells have very similar action potentials in culture and in the animal. Thus, the identification of these cells as medial or lateral N cells must be made in the ganglion, where their positions can show some variability. In some instances medial N cells may have been mistaken for lateral N cells.

In all experiments, except where specifically stated, the membrane potentials of both cells in the pair were held at about –50 mV by passing d.c. through the microelectrodes after balancing the bridge. (The normal resting potentials of these cells in culture range from –45 to –55 mV.) This was important since junctions with rectifying properties could appear to pass current in both directions if one cell was held at a different potential to the other (see Fig. 2; also Davis, 1986). Tests for synaptic connections were made at times ranging from 6 to 13 days. By this time the rectifying or non-rectifying properties of electrical junctions have become established (Davis, 1986).

The connections made by medial and lateral N cells with individual target neurones were not random. Lateral N cells never became connected by electrical or chemical synapses to each other. Action potentials, as well as large hyperpolarizing or depolarizing pulses, applied to one N cell produced no discernible effect on the other N cell in either direction (16 pairs). Similarly, N cells did not become coupled electrically or chemically to P sensory cells (eight pairs). In the following sections the connections made by lateral N cells on AP, AE and L cells are described first, followed by those made by medial N cells.

In earlier experiments by Fuchs et al. (1981) it was observed that lateral N cells could make rectifying electrical junctions with L cells in culture. Depolarization spread better from N to L cells, hyperpolarization from L to N cells. In the present experiments 11 closely apposed lateral N and L pairs showed similar rectification (three additional pairs showed no electrical or chemical interactions). Lateral N cells also formed rectifying electrical junctions with AP and AE cells (20 pairs). Again, the rectification was such that depolarization spread better from N cells. Fig. 1 shows that depolarizing current spread far better in one direction (from N to AP) and hyperpolarization in the other (AP to N). All junctions between lateral N cells and AP, AE and L cells continued to show clear rectification for 2 weeks, the longest time tested. Hence, the rectification does not represent a transient stage in the formation of non-rectifying junctions, such as those Davis (1986) observed for L cell pairs, AP cell pairs or Retzius cell pairs. In those experiments, current could spread in both directions by 6 days.

Apparent asymmetry of current spread between two cells could, in principle, result from damage caused by microelectrodes. In the present experiments both cells had similar high input resistances (see Materials and Methods); moreover, rectification was always in the same direction. Rectifying junctions established between P and AP cells in culture have been analysed by Davis (1986) using the voltage-clamp technique. The rectification characteristics with voltage-and with current-clamp techniques were similar and resembled those in the present experiments.

A possible pitfail in analysing current spread between pairs of cells coupled by rectifying junctions is shown in Fig. 2. As in Fig. 1, hyperpolarization spread better than depolarization from AP cell to N cell with both cells at their normal resting potential of about –50 mV. When the AP cell was hyperpolarized by steady current to –70mV, the junctional characteristics appeared to be ‘non-rectifying’. Such behaviour is inherent in the current-voltage relationship of the junctions: ‘depolariz-ation’ of the AP cell from –70 to –50mV is equivalent to a reduction in hyperpolarization by 20mV. As expected, when both cells were held at –70 mV coupling was again rectifying. In previous experiments by Davis (1986) and Fuchs et al. (1981) both cells were always held at their normal resting potential ( – 50 or –45 mV) to assess whether coupling was rectifying or non-rectifying to avoid such problems in interpretation.

Of 37 pairs of lateral N cells with various targets, 36 showed rectifying junctions. The one exception in which a non-rectifying junction was found between a lateral N cell and a Retzius cell may be due to incorrect identification of the lateral N cell during removal (see below).

Unlike lateral N cells, medial N cells made non-rectifying as well as rectifying connections with various targets. Figs 3, 4 show examples of both types of junctions made between medial N and AE cells in culture. The rectifying or non-rectifying properties of these junctions were not influenced by the duration of the time in culture beyond 6 days. The presence of rectification or non-rectification was not associated with the general condition of the cells as assessed by input resistance, electrical characteristics or microscopic appearance. Out of 12 medial N cells paired with AP and AE cells, four showed non-rectifying junctions and eight showed rectification. It seems likely that the single example of a non-rectifying synapse made by a lateral N cell in our experiments was due to an error in identification during or after removal and that this cell was in fact a medial N cell. Such errors can occur on occasion for N cells, whose position in the ganglion may vary.

Retzius cells in hundreds of trials have been shown never to make electrical connections with P sensory cells. Even large voltage-clamp pulses to the Retzius cells produce no coupling potentials in P cells (Dietzel et al. 1986). Instead, Retzius cells make chemical inhibitory synapses upon P cells with 5-hydroxytryptamine as the transmitter. In contrast, electrical connections were observed in 11 out of 11 pairs of Retzius and lateral N cells. In two other pairs, in which Retzius and medial N cells were used, no rectification was observed, as shown in Fig. 5.

Only rarely, in five pairs of cells out of all the trials we made of N cells, did we observe interactions that appeared to be chemically mediated. Fig. 6 shows synaptic potentials in a pair of N and L cells in which coupling was weak and in which there was a clear synaptic delay. In one other pair there was evidence of facilitation. However, depression was a more obvious feature and this, together with the usual strong electrical coupling, precluded a detailed analysis of chemically mediated potentials.

That N cells can make either rectifying or non-rectifying electrical junctions is an unexpected finding. Previous work in the animal and in culture had suggested that certain leech neurones could make only rectifying electrical junctions. For example, T, P and N sensory cells in the ganglion all make rectifying junctions with L motor neurones, such that depolarization spreads better from sensory cell to motor neurone. In the ganglion, T cells are interconnected by doubly rectifying junctions -depolarization spreads in both directions, hyperpolarization in neither (Nicholls, 1987).

A consistent picture has emerged from studies made on cultured neurones: whenever P cells make an electrical connection it is in the same direction whatever the target (Aréchiga et al. 1986). Thus, P cells always make similar, rectifying connections with AP, AE, HE and L cells. In contrast, these cells all make non-rectifying junctions with each other and with Retzius cells in culture.

From these results it seemed that some types of cells (such as P, N or T cells) could make only rectifying junctions; others that made non-rectifying junctions could also receive depolarizing signals from the rectifying cell type. Additional findings that extended these conclusions were obtained from experiments on pairs of cells in culture at different stages (Davis, 1986). At the earlier stages, P cell junctions with AP cells showed rectifying properties qualitatively similar to those at more mature, stronger junctions. In contrast, pairs of cells that make non-rectifying junctions (e.g. L–L, AP–AP, Retzius–Retzius, etc.) at early stages showed asymmetrical current flow. For example, identical L cells at 2 days showed depolarization spreading better in one direction, hyperpolarization in the other. However, by 6 days in the same pair of cells current spread symmetrically in both directions.

As shown in Figs 3, 4, medial N cells differ from lateral N cells and constitute a first exception: medial N cells can make both rectifying and non-rectifying junctions with the same target. Lateral and medial N cells had previously been shown to differ in their connections in the ganglion, in their electrical properties and in their antibody staining (Johansen & Kleinhaus, 1985; Johansen, Hockfield & McKay, 1984).

What determines whether a medial N cell in culture makes non-rectifying or rectifying junctions with its targets? Time does not seem to be a critical variable, neither do the membrane potentials nor the passive electrical properties of the cells in culture. One possible source of error that we cannot rule out is that the medial N cells from different leeches and from different segments were pooled in these experiments and they may vary in their properties. To remove identified medial N cells from ganglia is technically difficult; and we did not keep track of the segmental origin of each cell. This has not been a significant variable for other connections studied in culture. It would be of interest, albeit a daunting task, to make such experiments. Another approach would be to arrange cells in triplets. For example, one medial N cell could be placed between an AP cell and a Retzius cell. Would both connections be non-rectifying, rectifying or mixed?

Only occasionally were signs obtained of chemically mediated transmission with N cells as pre-or postsynaptic elements. The presence of electrical coupling would make small EPSPs or IPSPs hard to discern. Electron micrographs of N cells paired with L cells (D. Kuffler, unpublished data) have revealed morphological specializ-ations resembling synapses with a widened cleft and accumulations of clear vesicles apposed to the presynaptic membrane. The frequency and extent of these structures have not yet been established.

Unlike P cells, N cells became coupled electrically to Retzius cells. This result further emphasizes the specificity of connections formed in culture. Somehow Retzius cells and P cells are provided with cues that enable chemical but not electrical synapses to form. The cues for Retzius and N cells do allow them to make electrical synapses. With more detailed information about such hard and fast ground-rules, it may become possible to search, using modern techniques, for underlying mechan-isms involved in cell-cell recognition in this system.

We wish to thank Drs W. Adams and R. Stewart for many helpful and critical discussions, Mr P. Bättig for his unfailing technical assistance and Ms J. Wittker for skilled secretarial help. This work was supported by grants from the Swiss Nationalfonds (3.525-1.83/3.556-0.86) and by the US Navy to JN.