The Neuromuscular Junction Revisited: Ca²⁺ Channels and Transmitter Release in Cholinergic Neurones in Xenopus Nerve and Muscle Cell Culture

The senior author (TPF) regrets that owing to unforeseen visa problems he was not able to take part, as he had intended, in the Batam Discussion Meeting on Synapse Formation that, being initially conceived by John Treherne shortly after he attended a similar meeting in Beijing University in 1988, has now sadly become a memorial to him. We are glad still to be able to contribute a paper to this volume which is, of course, also in honour of his memory. To its senior author this paper marks the beginning of a revisit to the NMJ, once his major research field (Feng, 1988). There are threads of ideas which in his mind quite naturally connect the past with the present. One thread originates from his early studies on various effects of calcium on neuromuscular transmission that pointed, for the first time, to a specific relationship between calcium and transmitter release by the motor nerve endings (Feng, 1937), and led to the present work. As a result of the basic research of Bernard Katz, Ricardo Miledi, Rodolfo Llinas and others during the last several decades, it is now universally recognized that the entry of calcium ions through voltage-gated Ca²⁺ channels in the plasma membrane of the presynaptic nerve endings when they are depolarized by action potentials is an essential step in evoked transmitter release at the NMJ and, indeed, in chemical synapses generally. However, although we now know a great deal about voltage-gated Ca²⁺ channels in many cells, including nerve cells (see Tsien et al. 1988), we have practically no direct knowledge of the Ca²⁺ channels of motor nerve endings or other presynaptic terminals (Smith and Augustine, 1988). Their small size and the inaccessibility of their synaptic surface are severe obstacles to any attempt at a direct study of their ion channels. Even so, with the various new and powerful techniques, such as the patch-clamp, that have recently become available, it now seems worthwhile to attempt to gain more direct knowledge of the Ca²⁺ channels of the motor nerve endings and to understand further the links between calcium entry and transmitter release. We report here some initial results of our efforts in this direction.

As yet there is no known NMJ that allows direct study of the Ca²⁺ channels of their presynaptic terminals with the patch-clamp technique, and we have to resort to more or less artificial preparations. In 1987 when one of us (TPF) was visiting the Department of Anatomy and Neurobiology in Washington University, St Louis, and the Jerry Lewis Neuromuscular Research Center in UCLA, he had, with the help of friends in these places, looked in a preliminary way into the possibility of using motor nerve endings, freed by enzymatic treatment, from snake and frog nerve-muscle preparations. Motor nerve endings freed or exposed in this way are useful for a variety of investigations, and a good example of their use has been given by Grinnell et al. (1989), but the possibility of making a direct study of the ion channels of the freed nerve endings using the patch-clamp technique needs further exploration. For the present work we turn first to Xenopus embryonic nerve and muscle cell culture and make use of the soma of an identified cholinergic neurone and a myoball manipulated into contact with it, the system used by Mu-ming Poo and his collaborators (Chow and Poo, 1985; Xie and Poo, 1986). Such a cell pair provides a model of the neuromuscular junction that allows the application of a patch-clamp on both the pre- and postsynaptic components and so enables the Ca²⁺ channels of the presynaptic neurone to be studied and the acetylcholine (ACh) release to be monitored simultaneously. Through the whole-cell recording patch-clamp micropipette it is also possible to manipulate the intracellular chemical composition of the presynaptic cell. The potential usefulness of this experimental design is quite obvious. In addition, this culture system seems to provide attractive opportunities for studying possible developmental changes of Ca²⁺ channels and transmitter release accompanying the formation and maturation of the NMJ.

Cultures of Xenopus nerve and muscle cells were prepared essentially as described by Spitzer and Lamborghini (1976), Young and Poo (1983) and Xie and Poo (1986). The neural tube and the associated myotomal tissue of 1-day-old Xenopus embryos (stage 19–22, Nieuwkoop and Faber, 1967) were dissociated in Ca²⁺- and Mg²⁺-free saline supplemented with EGTA. The cells were plated on clean glass coverslips. The culture medium contained 10% (vol/vol) Leibovitz medium L15, 2% (vol/vol) foetal bovine serum and 200 i.u.ml^-1 gentamycin sulphate. A whole-cell patch-clamp electrode carrying a myoball was used first as a probe to identify cholinergic neurones and then to monitor transmitter release when a neurone soma and a myoball were brought together and formed a NMJ. For the study of the Ca²⁺ channels of identified cholinergic neurones, both whole-cell recording and single-channel recording from cell-attached patches were carried out. Two patch-clamp amplifiers (V. Pantani, Yale University, USA) were used. An Atari computer served both to control the execution of the experiments and to handle the data. A tape recorder was also available for storing some of the data for playing back onto a storage oscilloscope.

We began by studying the Ca²⁺ channels of the soma of identified cholinergic neurones by using whole-cell patch-clamp recording to investigate whether there was more than one type of voltage-gated Ca²⁺ channel and, if so, what types they were. Fig. 1A shows a series of Ca²⁺ current records in response to 200 ms depolarizing voltage pulses from a holding potential of —100 mV, increasing the depolarization in steps of 10 mV, with the first pulse from –100 mV to –80 mV and the last from —100 mV to +70 mV. From Fig. 1A two points are clear. (1) With depolarizations up to —40 mV, there is practically no activation of the Ca²⁺ channels; opening of the channels only begins to be noticeable when depolarization reaches —30 mV, maximum current occurring at 0 to +10 mV, after which the currents become smaller with further depolarization, undergoing a reversal in sign at about +50 mV. The corresponding I-V curve is shown in Fig. IB. (2) Once activated, the currents are well maintained for the duration of the 200 ms long pulses, i.e. the inactivation is uniformly slow. Comparing the above characteristics of calcium currents in the cholinergic neuronal soma we are studying with the calcium currents of the dorsal root ganglion (DRG) neurones described by Fox et al. (1987a,b), the differences are obvious: the kinetic characteristics of the calcium currents of the DRG neurones, as given by Fox et al., clearly show the presence of more than one type of calcium channel, while the current characteristics of our neurone give no evidence of multiple channel types. In particular there is no sign of the presence of calcium channels in our neurone with properties resembling the T-type channels of the DRG neurones. Fig. 2 shows a series of Ca²⁺ current records produced by longer depolarizing voltage pulses. The holding potential is –80 mV. The successive pulses are from –10 mV to +50 mV, spaced at 10 mV intervals as indicated. The purpose of this figure is to show that, although the time course of inactivation varies with the depolarizing voltage, in each record the decay time course can be fitted by an exponential with a single time constant. This provides additional evidence that the Ca²⁺ currents of the neurone under study are given by a single type of Ca²⁺ channel.

Fig. 3 shows the effects of replacing Ca²⁺ with Ba²⁺ and also compares the effects of two different concentrations of the two cations. The larger current with Ba²⁺ instead of Ca²⁺ and the relative shifts in the positions of the peak current are all to be expected for Ca²⁺ channels activated by high voltage, which the Ca²⁺ channels under study evidently are. We also examined the I-V curve obtained with 90 mmol I^-1 Ba²⁺, instead of 5 mmol I^-1 Ba²⁺, in the bath. With the much larger Ba²⁺ concentration the peak current became much larger and its position shifted further in the direction of more positive voltage. We mention this observation here as in our single-channel recordings the higher Ba²⁺ concentration was used. An example of our single-channel recordings is shown in Fig. 4. Here the same characteristics of the Ca²⁺ channels, i.e. activation by high voltage and slow inactivation, are equally as evident as in whole-cell recordings.

Judging only by the above electrophysiological observations, it is possible to classify the Ca²⁺ channels of the neuronal soma we are studying as either the L or the N type. For a definite identification, it is also necessary to take into consideration their pharmacological properties.

So far we have studied the effects of two cations, Cd²⁺ and Ni²⁺, two dihydropyridines, nifedipine and Bay K 8644, ω -conotoxin (ω -CgTX) and met-enkephalin. The results are collected in Table 1. The previously known actions of these agents on the neuronal N- and L-type Ca²⁺ channels are also quoted in Table 1 for comparison. Special attention is called to the lack of effect of nifedipine and Bay K 8644 and the striking blocking effect of met-enkephalin on the Ca²⁺ channels of the Xenopus neurone soma. This combination of pharmaco-logical properties seems to be a clear indication that the Ca²⁺ channels under study can be classified as the N type.

In summary, we conclude that the Ca²⁺ channels of the soma of the Xenopus cholinergic neurones we are studying are all of the same type, the N type. Barish (1989) recently reported that ‘Three components of Ca current corresponding to T-, N-and L-types are found in neurons that differentiate in cultures of dissociated Xenopus neural plate cells.’ The reason for this apparent discrepancy between Barish’ s observations and ours remains to be explained. We plan to go into this matter later when taking up developmental studies of Ca²⁺ channels in culture Xenopus neurones.

It has been amply shown by Poo and his collaborators (Chow and Poo, 1985; Xie and Poo, 1986; Evers et al. 1989) that the different regions of a cultured Xenopus spinal neurone - the growth cone, the neurite and the soma - are all capable of spontaneous transmitter release, giving rise to what they call spontaneous synaptic current (SSC) in the myoball contacting any of the three regions. Usually the frequency and the mean amplitude of the SSCs found at the soma are significantly lower than at the neurite or growth cone (Evers et al. 1989). The evoked release giving rise to evoked synaptic current (ESC) at the soma-myoball junction, however, has not previously been studied. As we are interested in correlating Ca²⁺ current recorded in the soma with its transmitter release, we have made a special study of the ESC at the soma-myoball junction. We found that, to get ESCs reliably from the soma-myoball junction, the choice of neurone was important. Our experience was that the somas of neurones that had already grown long neurites were generally less capable of responding to stimuli, i.e. of giving ESCs. For the study of evoked release from the soma we therefore chose neurones that had only just begun to grow processes.

An example of ESCs elicited by stimuli (1V and 0.5 ms square pulses delivered through a polished micropipette with a tip diameter of 2–4 μm. in direct contact with the soma) given once every 5 s for several minutes after establishing the soma-myoball contact is given in Fig. 5. The ESCs are irregular in size, very similar to those given by the growth cone-myoball or neurite-myoball junction described by previous workers (Xie and Poo, 1986; Sun and Poo, 1987).

To conduct correlated studies on calcium currents and transmitter release requires simultaneous whole-cell patch-clamp recordings from the pre- and postsynaptic cells. This demands very careful experimentation, involving first forming a whole-cell recording patch-clamp microelectrode on a myoball, then manipulating the myoball into contact with the soma of a neurone, and finally, after verifying that the neurone is cholinergic from its ability to give rise to SSCs, forming another whole-cell patch-clamp microelectrode on the soma. The completion of this whole chain of delicate manoeuvres remained, even after considerable practice, quite a strenuous effort with frequent failures. From the successful experiments we have made, a special note may be made of the following two points.

(i) Within the first 1–2 min after starting the usual whole-cell recording, there was, with a series of depolarizing pulses of different amplitudes, a general correlation between the size of the Ca²⁺ current and the size of the ESC elicited each time, but there were many irregularities. This is to be expected, as the ESCs elicited by constant stimuli are irregular in size, as already shown in Fig. 5. An example of correlated observation on Ca²⁺ currents and transmitter release evoked by 20 ms depolarizing pulses is given in Fig. 6, which shows similar irregularities. The basis of such irregularities is still a puzzle. The optimal experimental conditions for conducting correlated studies on Ca²⁺ currents and evoked transmitter release remain ta be worked out. Attention may first be called to an unexpected phenomenon interfering with such studies.

(ii) With the usual whole-cell recording method, within 2–3 min after rupturing the cell-attached patch during the formation of the whole-cell recording mode, while the depolarizing voltage pulses still produced the same Ca²⁺ currents, the correlated evoked transmitter release began to decline. It disappeared altogether within 4–5 min. When this happened, the spontaneous transmitter release giving rise to SSCs continued without significant change. There is thus a rapid differential run-down of the evoked release, with unchanged calcium current and spontaneous release. Recording the Ca²⁺ current required the elimination of the potassium current, so the micropipette was filled with a solution containing caesium instead of potassium. We wondered whether the caesium entering the cell might be responsible for the run-down of the evoked release and so, in a number of experiments, we forsook the recording of Ca²⁺ currents. We filled the whole-cell Recording micropipette with a solution containing KC1 to replace CsCl as the main constituent, and followed the change of the evoked release with time. It was found that there was the same rapid run-down of the evoked release as when caesium was in the micropipette. Thus, it appeared that the run-down was probably due to the loss of some cytoplasmic compounds from the cell rather than to the introduction of some harmful substance into the cell. This idea was further tested by using a modified whole-cell patch-clamp recording method introduced by Horn and Marty (1988). In this method, nystatin, a pore-forming antibiotic, is added to the micropipette solution. After the micropipette has established a gigaohm seal on the cell, the patch of membrane separating the micropipette from the cell interior is not ruptured by suction, as is the usual procedure, but is made permeable to cations by the antibiotic to provide the electrical continuity required for whole-cell recording. The permeabilized patch, while allowing the free flow of electric current, prevents the loss of cytoplasmic constituents. We have made preliminary Observations with this method of whole-cell recording. Fig. 7 compares the change with time of the ESCs elicited by depolarizing pulses applied to the soma through the whole-cell recording micropipette containing caesium, potassium or caesium plus nystatin. It shows that with the nystatin method of whole-cell recording the quick run-down of the evoked release could be prevented.

We hope that with the nystatin whole-cell recording method we shall be able to carry out more detailed studies on the relationship between Ca²⁺ currents and transmitter release. It should also be noted that the wash-out effect, apparently specifically affecting the evoked transmitter release, seems to be an important phenomenon in itself. Its analysis might help to disclose or define some essential link in the chain of molecular interactions leading from calcium entry to transmitter release.

We are grateful to Drs R. W. Tsien and F. J. Sigworth for valuable help in the course of equipping our patch-clamp laboratory, to Professor H. H. Chuang for a constant supply of Xenopus embryos from his aquarium and to Professor Kunitaro Takahashi for a gift of nifedipine and ω -conotoxin.