Presynaptic calcium currents and facilitation of serotonin release at synapses between cultured leech neurones

At neuromuscular junctions of frogs and crustaceans, several lines of evidence point to ‘residual calcium’ as an underlying mechanism for facilitation (Dudel & Kuffler, 1961a,b; Katz & Miledi, 1965, 1968; Dodge & Rahamimoff, 1967; Rahamimoff & Yaari, 1973; Parnas et al. 1986). If each impulse in a pair of impulses were to allow the same amount of Ca²⁺ to enter but removal was slow, the second Ca²⁺ transient would cause greater release, since the initial baseline level would be higher. Alternatively, the second depolarization could give rise to a larger inward Ca²⁺ flux that would produce a larger response.

There are few preparations in which a direct correlation can be made between Ca²⁺ entry and transmitter release, apart from the giant synapse of the squid (Lunas et al. 1981a,b; Charlton et al. 1982; Smith et al. 1985; Augustine & Charlton, 1986). At the squid synapse, the transmitter is not known and facilitation in normal Ca²⁺ is small (Charlton et al. 1982; Stanley, 1986). The primary aim of the present experiments has been to determine by voltage-clamp analysis whether presynaptic facilitation of Ca²⁺ entry occurs at chemical synapses formed between Retzius and P sensory cells isolated from the CNS of the leech and maintained in tissue culture. The advantages of this synapse are (1) that serotonin has been identified as the transmitter and (2) that the degree of facilitation in normal Ca²⁺ is large, up to 400 %, and persists for 250 ms or more (Henderson, 1983; Henderson et al. 1983; Dietzel et al. 1986).

With voltage-clamp it is possible to record inward Ca²⁺, Sr²⁺ and Ba²⁺ currents in Retzius cells after K⁺ currents have been blocked by TEA⁺ and 4-AP and Na⁺ currents are reduced (Stewart et al. 1989). Since tetrodotoxin does not block Na⁺ currents in leech neurones, they are reduced by lowering external Na⁺ concentration (Nicholls & Kuffler, 1964; Kleinhaus & Prichard, 1983; Beleslin, 1985). An important feature of this preparation is that the cells are virtually isopotential (Ross et al. 1987, 1988). Synaptic potentials in the P cell are chloride-dependent and are still present under the conditions used for measuring Ca²⁺ currents (Fuchs et al. 1982). Moreover, with pairs of impulses the facilitated synaptic potentials recorded in the P cell can be attributed to increased numbers of quanta (Henderson et al. 1983).

Using these synapses in culture the following measurements have been made. (1) The time course and amplitude of facilitation in solutions containing low Na⁺ and TEA⁺ in comparison to those recorded in normal L-15 medium. (2) The relationship between presynaptic divalent cation currents and serotonin release. (3) Successive divalent cation currents evoked by two presynaptic depolarizations following in rapid succession. In addition, we have compared the effectiveness of Ca²⁺, Sr^24- and Ba²⁺ for releasing transmitter. A brief report of some of these observations has been published elsewhere (Stewart et al. 1987).

The methods for removing and culturing leech neurones have already been described (Fuchs et al. 1981; Dietzel et al. 1986; Stewart et al. 1989). In brief, individual Retzius and P sensory cells were removed from the CNS of the leech by looping or by suction after mild enzyme treatment. Pairs of cells were plated in microwell dishes coated with polylysine or concanavalin A, and maintained at 20°C in Leibowitz 15 (L-15) culture medium supplemented with 2% foetal calf serum (FCS), 0·6% glucose, and 2 mmoll^-1 glutamine. Under these conditions recordings with voltage-sensitive dyes indicate that depolarizing and hyperpolarizing pulses spread to the ends of the processes without detectable attenuation (Ross et al. 1987,1988). Synapses formed within a few days and were tested between days 2 and 12 in culture.

Two-electrode voltage-clamp was used to measure Ca²⁺ currents in Retzius cells that had formed synapses with P cells. Electrodes were filled with 4 mol l^-1 potassium acetate and had resistances of 15–30MΩ. Electrode resistances within this range allowed rapid delivery of large currents and produced small capacitative coupling. Electrode resistances lower than 15 MΩ damaged cells and caused large leak currents.

Divalent cation currents were recorded after blocking K⁺ currents with TEA⁺, or with TEA⁺ and 4-AP, and reducing Na⁺ currents with a low external Na⁺ concentration (see solutions in Table 1). For measuring Ca²⁺ currents during short, 5–10 ms stimulation pulses, the addition of 4-AP was necessary since in 100 mmol l^-1 TEA⁺ a residual, fast-activating and fast-inactivating outward K⁺ current of 5–10 nA remained (I_A; see Stewart et al. 1989). 4-AP reduced this residual current so that it was barely measurable. In experiments in which stimuli of longer duration (20 ms or greater) were used, measurements of the divalent cation current were taken at the end of the current trace. By this time I_A had inactivated and the Ca²⁺ current-voltage relationships in solutions containing TEA⁺ were essentially the same with or without 4-AP. Even in the presence of TEA⁺, 4-AP and Cd²⁺, to block divalent cation currents, a slowly increasing outward current remained that increased in amplitude with progressively larger depolarizations of +20 to +50 mV. This caused our measurement of the Ca²⁺, Ba²⁺ or Sr^2-1- reversal potential to be lower than expected. Ca²⁺ reversal potentials measured in single, internally perfused Retzius cells in which K⁺ currents were more completely blocked were higher by 15 mV or more (R. J. Bookman & Liu, unpublished results).

For analyses of synaptic function, single or twin depolarizing pulses of fixed duration (5–120 ms) were delivered to the Retzius cell from a holding potential of –50 mV. We waited for 45 s between tests of synaptic function to avoid depression. In some experiments, three hyperpolarizing pulses, each one-third of the amplitude of the depolarizing pulse (P/3), were delivered before each trial. In others, identical depolarizing and hyperpolarizing pulses were used. The currents generated during the hyperpolarizing pulses were used later for leak subtraction. The postsynaptic P cells, which were not voltage-clamped, were impaled with a single KC1 electrode of 20–40 MΩ to reverse IPSPs and maintain the membrane potential at a constant value during the experiment. Control experiments have shown that such reversal potentials remain constant in amplitude over periods of hours while the KC1 electrode remains in the P cell (Fuchs et al. 1981; Henderson et al. 1983). Transmission tended to fail after a few trials in the low-Na⁺, high divalent cation, TEA⁺ and 4-AP solutions. Several variables were tested but failed to improve stability. These included: (1) adding glucose to the recording medium, (2) reducing the Na⁺ concentration from 125 to 30 mmol l^-1 instead of 10 mmol l^-1 to keep the Na⁺/Ca²⁺ exchanger functioning and (3) adding 5-hydroxytryptophan (0·2mmoll^-1), the immediate precursor of serotonin (5-HT). No information is available about effects of Ba²⁺ and Sr²⁺ on postsynaptic 5-HT receptors. Changes in Ca²⁺ concentration do not affect the amplitudes of miniature potentials in P cells (Henderson et al. 1983).

‘Almost Perfect Electronic’ voltage amplifiers, voltage-clamps and stimulators were used. Records were stored on a Hewlett-Packard tape recorder and analysed for leak subtraction and averaging on a Minc-Declab computer as described by Stewart et al. (1989). Records were filtered with a Bessel filter having a bandwidth of 2·5 kHz. Capacitative and leak currents were subtracted from current records produced by depolarizing voltage pulses. Holding currents were then subtracted to give a zero baseline.

Cells were cultured in supplemented L-15 as described above. Shortly before beginning an experiment, the L-15 medium bathing the cells was replaced with one of the solutions shown in Table 1. Most of the experiments were done in solution 4.

Cultured Retzius cells gave prolonged action potentials in low external Na⁺ concentrations, with K⁺ channels largely blocked. In the presence of 10 mmol l^-1 Ca²⁺, Sr²⁺ or Ba²⁺ the duration of action potentials could be extended from the normal value of about 5 ms to as long as 45 s (Fig. 1; solution 1, Table 1). Ba²⁺ produced action potentials of the longest duration followed by Sr²⁺ then Ca²⁺. Ca²⁺, Sr²⁺ or Ba²⁺ action potentials in the Retzius cell all triggered transmitter release and evoked synaptic potentials of large amplitude and long duration in the postsynaptic P cell. Whereas Ca²⁺ and Sr²⁺ solutions were able to sustain synaptic release for many trials, release failed within a few trials in solutions containing Ba²⁺. When recordings were made in these solutions with KC1 electrodes, synaptic potentials in P cells appeared as depolarizing events.

Fig. 2 shows examples of transmitter release under voltage-clamp in normal L-15 and with Ca²⁺, Sr²⁺ and Ba²⁺ as charge carriers. In normal L-15 and in Ca²⁺ and Sr²⁺ solutions, release occurred with voltage-clamp pulses as short as 5 ms (Fig. 2A-C). In solutions containing Ba²⁺, release only occurred occasionally with pulses of very long duration (120 ms or greater; see Fig. 2D) and with total integrated currents of 1336 ± 345 pC (S.E.M., N = 6 trials). In Ca²⁺ the integrated current was 522 ± 60pC (N = 25 trials), and for Sr²⁺ it was 319 ± 54pC (N = 16 trials). These results indicated that the potency for promoting release was considerably greater for Sr²⁺and Ca²⁺ than for Ba²⁺.

In the presence of TEA⁺, 4-AP and reduced Na⁺, transmission between Retzius and P cells failed more rapidly than in normal L-15 solution. As a result, only a liiyiited number of trials could be made in any one experiment. In some pairs pf cells, Ca²⁺ currents in the Retzius cell diminished with repeated trials, but in others, Ca²⁺ currents persisted unchanged for minutes after transmitter release had been depressed.

In Fig. 3A the inward Ca²⁺ current measured in the presynaptic Retzius cell reached its maximum amplitude at a depolarization of +15 mV and reversed at +35 mV. In six experiments, the amplitude of the Ca²⁺ current reached a maximum at +13 ±2-5 mV (S.E.M.) and with larger depolarizations decreased to zero at +33 ±2·0 mV. This decrease was presumably the result in part of a reduced driving force for Ca²⁺. Part of the decline, however, could be attributed to outward K⁺ currents not completely blocked by TEA⁺ and 4-AP; such residual K⁺ currents became more pronounced as the driving force increased (see Materials and methods). Correlated with the larger inward Ca²⁺ current following depolarizations up to +15 mV was a steep increase of transmitter release (Fig. 3B). At larger depolarizations, transmitter release in the absence of net inward current is probably the result of masking of Ca²⁺ currents by residual outward K⁺ currents. After normalization, the relationship between integrated Ca²⁺ current and transmitter release was similar to that for Ca²⁺ current amplitude measured at the end of 20 ms pulses. The experiment shown in Fig. 3 was unusual in the large number of points we were able to obtain in the solutions before depression became apparent.

The synaptic potentials in P cells also depended critically on the duration of the pulse delivered to the Retzius cell, as shown in Fig. 4 in which Sr²⁺ was the charge carrier. Pulses shorter than 5 ms were ineffective. Similar results were obtained with Ca²⁺ in the bathing fluid. After a pulse longer than 20 ms, a second depolarization of the Retzius cell evoked only a small synaptic potential. Such depression could persist for several minutes and precluded detailed examination of the effects of long depolarizing pulses on the Retzius cell.

In solutions containing TEA⁺, 4-AP and reduced Na⁺, facilitation was still apparent (solution 4 of Table 1). With two brief voltage-clamped pulses to the presynaptic Retzius cell, the second gave rise to a facilitated synaptic potential in the P cell. An example is shown in Fig. 5. The amount of facilitation reached values as high as 460 % and persisted for over 250 ms. Fig. 6 shows the time course of facilitation under voltage-clamp conditions with currents other than Ca²⁺ largely blocked. The same graph shows points obtained in normal L-15 with paired action potentials and with paired voltage-clamp pulses. Although the recordings were made in solutions of markedly different composition, facilitation was remarkably similar.

When sequential Ca²⁺ currents were measured with twin voltage-clamp pulses, their amplitudes and time courses were indistinguishable (Fig. 7). Fig. 7A shows two superimposed traces of presynaptic voltage pulses (below) in the Retzius cell and synaptic potentials (above) in the P cell. The second pulse, delivered after an interval of 45 ms, produced a synaptic potential facilitated by 410 ± 25 % (S.E.M., N= 4 trials). Such facilitation has been shown to be due to increased numbers of serotonin quanta released from the presynaptic Retzius cell (Henderson et al. 1983). The two inward Ca²⁺ currents are shown in Fig. 7B and are superimposed in Fig. 7C. Identical Ca²⁺ currents during twin pulses delivered to the Retzius cell were observed in six pairs of cells. In 42 pairs of cells, the second Ca²⁺ current was pmaller by 2·8 ± 1·1 % (N = 48 trials). For the four pairs of cells of Fig. 6 in which clear facilitation was observed, the mean amplitude of the second Ca²⁺ current was reduced slightly by 0·05 ± 1·4% (N = 11 trials). Thus, the same quantity of Ca²⁺ or slightly less, but not more, entered the Retzius cell during facilitated transmitter release.

In the present series of experiments, the abilities of Ca²⁺, Sr²⁺ and Ba²⁺ to produce release of transmitter were determined under current- and voltage-clamp. Moreover, direct measurements of inward Ca²⁺ currents were made during twin stimulation pulses in which strong facilitation of transmitter release occurred. Our results indicate that Ca²⁺ and Sr²⁺ were more effective than Ba²⁺ at releasing transmitter. These results are similar to those found by Callaway & Stuart (1987) in that Ca²⁺ and Sr²⁺, but not Ba²⁺, were able to sustain transmitter release at a synapse made by barnacle photoreceptors. By contrast, Augustine & Eckert (1984) determined at the squid giant synapse that Ba²⁺ and Sr^24- were less able to produce significant transmitter release.

Several lines of evidence suggest that residual calcium plays a role in two-pulse facilitation of transmitter release (Katz & Miledi, 1968; Charlton et al. 1982; Ross et al. 1987). Thus, the time course and amplitude of facilitation, as well as the effects of alterations in external Ca²⁺ and Mg²⁺concentration can be explained in terms of Ca²⁺ entry and the steep sigmoidal relationship between Ca²⁺ concentration and transmitter release. In addition, measurements of intracellular Ca²⁺ with voltage-sensitive dyes indicate that successive Ca²⁺ transients with paired presynaptic impulses are similar and that Ca²⁺ concentration remains raised (Charlton et al. 1982; Ross et al. 1987). In the present experiments, direct measurements of inward Ca²⁺ currents showed that the second voltage-clamp pulse did not give rise to a larger inward current than the first. The two currents were identical in their time courses and amplitudes. These results support the idea that it takes a significant time for intracellular Ca²⁺ levels to fall after the first pulse; and that while the level remains elevated, the second identical increase in Ca²⁺ can release more transmitter.

There remain problems in the interpretation of our results, however. As with voltage-clamp studies made at other synapses, we cannot rule out the presence of a distinct population of Ca²⁺ channels with different characteristics localized to presynaptic release sites. Current recordings made from an entire cell would not be expected to reveal the properties of such channels or the currents they contribute. Experiments with patch-clamp and loose patch-clamp should enable such regional differences in Ca²⁺ channels to be analysed (Bookman & Dagan, 1987; Bookman et al. 1987). Another problem is that serotonin, to which Retzius cells are sensitive, may have presynaptic effects on the facilitation process following Ca²⁺ entry. Presynaptic application of serotonin should allow us to determine if changes in release occur that resemble facilitation. A technical problem was that K⁺ currents could not be blocked completely by TEA⁺ and 4-AP in Retzius cells and so the inward currents were contaminated with outward currents, especially with large depolarizations. The sharp decline in inward currents without abolition of release can probably be attributed to this effect (Fig. 3). A second technical difficulty was to maintain transmission constant for long enough to make reproducible observations in the abnormal solutions required for Ca²⁺ current measurements. Nevertheless, findings such as those shown in Fig. 7 provide convincing evidence for similar inward calcium transients during facilitation. The alternative explanation of similar net inwardly directed current traces, attributable to complex but precise summation of both increased inward and outward currents following the second pulse, seems unlikely.

Certain quantitative problems indicate that residual Ca²⁺ on its own may not account for various aspects of facilitation observed at neuromuscular synapses. It has been suggested, for example, that presynaptic voltage changes might activate molecules with which Ca²⁺ interacts to produce transmitter release (Parnas et al. 1986; Parnas & Parnas, 1986). The size of the leech neurones and the proximity of the cell bodies to transmitter release sites should permit a direct test of the independent effects of presynaptic depolarization and raised intracellular Ca²⁺ levels.

We thank Dr R. J. Bookman for his help and encouragement throughout these experiments and Mrs H. Niederer and P. Lüdin for their skilful preparation of cell cultures. We are grateful to Ms J. Wittker for her excellent secretarial assistance and to Mr P. Baettig for photographing the figures. This work was supported by grants from the Swiss National Fund and the US Navy to JGN; RRS was supported by an NIH Postdoctoral Fellowship.