Actions of dendrotoxin on K⁺ channels and neuromuscular transmission in Drosophila melanogaster, and its effects in synergy with K⁺ channel-specific drugs and mutations

A diversity of K⁺ channels gives rise to a large repertoire of firing patterns in nervous systems, and a definitive classification of different K⁺ channels is not yet possible (Latorre & Miller, 1983; Wu & Ganetzky, 1986; Rudy, 1988). Specific toxins, drugs and mutations thus provide useful tools to compare the types and properties of the channels in different excitable membranes.

Dendrotoxin (DTX), a non-enzymatic venom protein from the mamba snake Dendroaspis angusticeps, is a potent convulsant neurotoxin when injected into the brain. DTX is also the first snake toxin found to facilitate neurotransmitter release at vertebrate neuromuscular junctions (Harvey & Karlsson, 1980; Harvey et al. 1984; Harvey, 1987) and central synapses (Dolly et al. 1984). Its action has been characterized in several vertebrate species by voltage-clamp experiments and a reduction in K⁺ conductances is correlated with the facilitatory effect of the toxin (Dolly et al. 1984; Weller et al. 1985; Benoit & Dubois, 1986; Halliwell et al. 1986). However, the specific types of K⁺ conductances that are affected by this toxin vary in different preparations. In frog peripheral nerve nodes of Ranvier, a component of slowly inactivating K⁺ current is decreased by DTX (Weller et al. 1985 ; Benoit & Dubois, 1986), whereas in the cell body of central hippocampal neurons a transient K⁺ current, I_A, is suppressed (Dolly et al. 1984; Halliwell et al. 1986). However, the transient K⁺ current in rat sympathetic ganglion neurons is reported to be insensitive to DTX (Dolly et al. 1987). The endplate potential (EPP) in the frog neuromuscular junction is enhanced by increased quantal content and by multiple firing of the motor axon, but the type of K⁺ channels affected by DTX is unknown (Harvey & Anderson, 1985; Harvey, 1987). Although these contrasting results may be attributed to species differences, a more intriguing possibility is that they reflect distinctions and similarities among different K⁺ channels.

The generality of a DTX blocking action on K⁺ channels has not been well established since only vertebrate species have been studied. In the present study, we examined the effects of DTX on neuromuscular transmission and on muscle K⁺ channels of an arthropod species, Drosophila melanogaster. Both the wildtype and K⁺ channel mutants were studied to examine whether DTX can be used as a probe for characterization of various K⁺ channels. The genetics of membrane excitability in Drosophila has been well studied. Several mutations that eliminate or alter specific types of K⁺ currents are available (Ganetzky & Wu, 1986; Wu & Ganetzky, 1986).

The larval neuromuscular preparation of Drosophila is suitable for electrophysiological studies because the properties of neuromuscular transmission and of various ion channels in the muscle membrane have been characterized in detail (Jan & Jan, 1976a,b; Jan et al. 1977; Wu et al. 1978, 1983; Ganetzky & Wu, 1982, 1983,1985; Wu & Haugland, 1985; Gho & Mallart, 1986; Haugland, 1987; Singh & Wu, 1989; F. N. Haugland & C.-F. Wu, in preparation). Voltage-clamp studies of these larval muscle fibers have demonstrated that mutations of three genes separately affect different types of K⁺ channels. The transient I_A is missing in Shaker (Sh) mutants (Wu et al. 1983; Wu & Haugland, 1985); the delayed rectifier, I_K, is reduced to about 50 % in an ether a go-go (eag) mutant (Wu et al. 1983); and a Ca²⁺-dependent K⁺ current, Ic_F, is eliminated in the slowpoke (slo) mutant (Elkins et al. 1986; Singh & Wu, 1987,1989). We used these mutants in conjunction with other K⁺ channel blocking agents to characterize the effect of DTX on neuromuscular transmission. The synergism between the effects of DTX and various mutations was studied to infer the types of presynaptic K⁺ channels affected by DTX. The action of DTX on muscle membrane currents was also directly monitored by the voltage-clamp technique. These experiments using Drosophila mutants not only extend our understanding of the action of DTX on invertebrate K⁺ channels but also offer a unique opportunity for comparing the types and properties of the K⁺ channels present in nerve terminals and muscle fibers.

The wild-type strain, Canton-S, of Drosophila melanogaster was used for characterization of normal physiology. Among available mutant alleles of the Sh and eag loci (Ganetzky & Wu, 1986), Sh^KSI33 and eag¹ were chosen for this study because they have been described in the most detail. They are also the most extreme alleles of these two loci; Sh^KS133 completely blocks I_A and eag¹ removes about 50% of I_K (Wu et al. 1983; Wu & Haugland, 1985; Y. Zhong & C.-F. Wu, unpublished results). The slo locus is represented by a single allele (Elkins et al. 1986), which was used in this study.

For intracellular recording of EJPs, the larval neuromuscular preparation (Jan & Jan, 1976û; WU et al. 1978) was used. Briefly, a late third-instar larva was pinned out in saline, cut along the dorsal midline and eviscerated. The nerves innervating the body wall muscles in the abdominal segments were severed near the ventral ganglion. The cut end of a segmental nerve was stimulated via a suction electrode and EJPs were recorded in the ventral longitudinal muscle fibers. Experiments were carried out in Drosophila saline containing 128 mmoll^-1 NaCl, 2 mmoll^-1 KC1, 35 mmol I^-1 sucrose, and buffered at pH71 with 5 mmoll^-1 Hepes. The concentrations of Ca²⁺ (0·1, 0·2 or 1·8 mmoll^-1) and Mg²⁺ (0·5, 2, 4 or 14 mmoll^-1) were varied as specified in each experiment. Other details of the experimental protocol have been described elsewhere (Jan & Jan, 1976a; Wu et al. 1978).

An identical preparation was used for conventional two-microelectrode voltageclamp measurements of muscle membrane currents. We followed the established paradigms and protocols (Wu & Haugland, 1985; Haugland, 1987; F. N. Haugland & C.-F. Wu, in preparation) and only a brief description is given here. Measurements of I_A and I_K were made in Ca²⁺-free saline containing 0·5 mmol I^-1 EGTA. The concentration of MgCl₂ varied in each experiment, as specified. The composition of the Ca²⁺-free saline was otherwise identical to that used for EJP recordings (see above). The large differences in inactivation and recovery kinetics between I_A and I_K were utilized to separate these two currents. At a holding potential (V_H) of —80 mV, step depolarizations elicited I_A, which peaked rapidly and then inactivated, and I_K, which rose to a sustained plateau with a delay (Figs 6, 8 and 10). When a depolarizing prepulse (to —20 mV) preceded the test pulse, I_A was inactivated, leaving I_K alone. Digital subtraction of the two measurements (with or without the prepulse) yielded the inactivating I_A in isolation from I_K (Figs 7 and 9). The duration of the prepulse (2 s) and the interval between the prepulse and test pulse at 5 °C (10 ms) and 16 °C (5 ms) were chosen from previously established values to maximize the separation of these two currents (Haugland, 1987; F. N. Haugland & C.-F. Wu, in preparation; F. N. Haugland, Y. Zhong & C.-F. Wu, unpublished data).

When Ca²⁺ currents and Ca²⁺-dependent K⁺ currents were examined, 20mmoll^-1 CaCl₂, 4mmoll^-1 MgCl₂ and no EGTA were present in saline. All voltage-clamp measurements were stored digitally and analyzed by a microcomputer (Macintosh, Apple Computer, Cupertino, CA). After subtraction of linear leakage currents, the density of active currents was determined by adjustment for the cell membrane area of the fiber and expressed in μA μF^-1 using membrane capacitance as an indicator. For each experiment, data collected from a number of fibers were averaged as shown in the traces and current-voltage (I–V) relationships in Figs 6−10. The numbers of fibers and larvae used in each experiment are specified in the legend of the figure.

Measured amounts of DTX from D. angusticeps, 4-aminopyridine (4-AP) and quinidine stock solutions were added to the bathing chamber (volume, 1ml) to achieve the final concentrations. Solutions were carefully mixed and enough time was allowed for equilibration. Quinidine and 4-AP were acquired from Sigma Chemical Co. (St Louis, MO) and DTX was a gift from Drs F. J. Joubert and F. F. H. Carlsson.

Intracellular recordings of EJPs were performed at 25−29 °C, except where stated otherwise. The results in each figure (Figs 1−5) were based on observations on 5−11 fibers in 3−8 larvae. In voltage-clamp experiments, preparations were cooled to either 5 or 16°C. At higher temperatures, the rapid kinetics of the membrane currents made it difficult to achieve satisfactory control of voltage jumps (Wu & Haugland, 1985).

The ventral longitudinal body-wall muscle fibers in the abdominal segments of Drosophila larvae are each innervated by two motor neurons; and only excitatory junctional potentials (EJPs), but not inhibitory potentials, can be evoked by nerve stimuli (Jan & Jan, 1976a). As in other arthropods, glutamate is the putative transmitter at the neuromuscular junction (Jan & Jan, 1976b; Johansen et al. 1989) and there are Ca²⁺ channels and various types of K⁺ channels, but no Na⁺ channels, in the muscle membrane (Wu & Haugland, 1985; Haugland, 1987; Singh & Wu, 1989). Normally, only graded potentials are evoked in these muscle fibers, although all-or-none Ca²⁺ action potentials may be generated when the K⁺ channels are blocked (Suzuki & Kano, 1977).

It has been reported that the activity, but not binding, of DTX is suppressed by increased Mg^2-1- concentrations (3 mmol1^-1) in isolated vertebrate nerve-muscle preparations (Harvey & Anderson, 1985). When this phenomenon was investigated in larval muscles, we found that the effects of DTX on neuromuscular transmission in Drosophila depended on Ca²⁺ as well as Mg²⁺ concentrations in the saline. Even at the normal Ca²⁺ concentration (1·8 mmol1^-1), the action of DTX was apparently inhibited since EJPs were not significantly altered in the presence of 4·2 μmoll^-1 (30 μgml^-1) DTX. This was true for preparations bathed in salines containing either normal (4 mmol1^-1, data not shown) or reduced (2 mmol1^-1, Fig. 1A) concentrations of Mg²⁺.

The action of DTX became apparent when the external Ca²⁺ concentration was lowered (Fig. 1B,C). At 0·lmmoll^-1 Ca²⁺, EJPs showed quantal fluctuations in size and miniature EJPs (mEJPs) were the only spontaneous activity observed, consistent with previous reports (Jan et al. 1977; Ganetzky & Wu, 1982). Upon addition of DTX, the amplitude of EJPs increased appreciably (on average to 164 % of the control value in two fibers) and repetitive EJPs were occasionally evoked by a single nerve stimulus (Fig. IB). The DTX effects were further enhanced if the Mg²⁺ concentration was also reduced from 4 to 2 mmol1^-1. At low Mg²⁺ and Ca²⁺ concentrations the neuromuscular junction became more excitable, often leading to more than one, and sometimes up to five, EJPs in response to a single stimulus (Fig. 1C). In the presence of DTX, prolonged trains of repetitive EJPs (up to 30) of increased size occurred spontaneously or in response to stimulation (Fig. 1C).

Further experiments showed that at very low cation concentrations (0·2mmoll^-1 Ca²⁺ and 0·5mmoll^-1 Mg²⁺), DTX caused clearly enhanced neuromuscular transmission at either high (20°C) or low (8°C) temperature. This effect was completely suppressed by 14 mmol1^-1 Mg²⁺ (data not shown). The suppressive effect of Mg²⁺ is apparently on the activity rather than on the binding of DTX to its target site, as previously shown in vertebrate preparations (Harvey & Anderson, 1985). Application of DTX (2·8 μmoll^-1) in high-Mg²⁺ saline (4 or 14 mmoll^-1 Mg²⁺ and 0·2 mmoll^-1 Ca²⁺) did not change larval EJPs at 20°C. Nevertheless, following a rapid wash with low-Mg²⁺ saline (containing 0·5 or 1 mmoll^-1 Mg²⁺ and 0·2 mmoll^-1 Ca²⁺, but no DTX), a single nerve stimulus evoked repetitive EJPs with increased amplitude. These facilitated EJPs were suppressed again by high-Mg²⁺ saline (without DTX) and reappeared in low-Mg²⁺ saline (data not shown).

Calcium ions may suppress the effect of DTX on EJPs by a mechanism similar to that of magnesium ions. But at least part of the suppression at high external Ca²⁺ concentrations can be attributed to the activation of a Ca²⁺-dependent K⁺ conductance in the nerve terminals (Jan et al. 1977; Ganetzky & Wu, 1982, 1983). This notion was supported by the results obtained in slo mutants, in which a Ca²⁺-dependent K⁺ current, I_CF, is eliminated (Elkins et al. 1986; Singh & Wu, 1987, 1989). In saline containing l·8mmoll^-1 Ca²⁺ and 4·0 mmoll^-1 Mg²⁺, EJPs in slo larvae were nearly normal. Following treatment with 4·2 μmoll^-1 DTX, the duration of EJPs showed an approximately twofold increase (three fibers in different larvae, data not shown). Thus, DTX might affect some conductances other than I_CF and its effects become more obvious when I_CF, and preferably all other K⁺ currents that are not affected by DTX, are removed. In the following experiments, EJP data were collected at low Ca²⁺ concentrations (0·1 or 0·2 mmoll^-1) to minimize the complications introduced by the Ca²⁺-dependent K⁺ currents.

The effects of DTX on larval neuromuscular junctions were relatively minor in comparison with the changes induced by the K⁺ channel blocker 4-AP or the Sh mutation. Voltage-clamp measurements in larval muscle have shown that the Sh mutation eliminates I_A and that 4-AP at 0·1 mmoll^-1 blocks nearly all I_A with a minimal reduction in I_K (Haugland, 1987; Haugland & Wu, 1987). In addition, quinidine at 0·1 mmoll^-1 suppresses I_K without affecting other K⁺ currents (Singh & Wu, 1987, 1989).

As previously described (Jan et al. 1977), treatment with 0·1 mmoll^-1 4-AP drastically increased both the amplitude and the duration of EJPs (Fig. 2A,B), reflecting enhanced presynaptic Ca²⁺ influx and repetitive firing (Ganetzky & Wu, 1982). Addition of 4-2/zmoll^-1 DTX to the bath did not further enhance the transmitter release (Fig. 2C). In contrast, further application of just 50 μ moll^-1 quinidine prolonged the duration of EJPs from tens to hundreds of milliseconds (Fig. 2D). The prolonged EJPs were similar to those previously reported in drug-treated or double-mutant preparations in which both I_A and I_K are reduced (Wu et al. 1983; Ganetzky & Wu, 1983, 1985). These phenomena parallel the observations in Sh larvae. Identical results were obtained by adding quinidine (50 μ moll^-1) alone to Sh larvae with or without DTX pre-treatment. The Sh phenotype, i.e. raised and prolonged transmitter release (Fig. 3A, also see Jan et al. 1977), was further enhanced by quinidine but not by DTX (Fig. 3B,C).

The above results raised the possibility that DTX acted mainly to reduce I_A in nerve terminals and thus was not able to exert additional effects when I_A was blocked by 4-AP or the Sh mutation. This interpretation was confirmed in a different set of experiments employing quinidine and the eag mutation, which blocks about 50 % of I_K in larval muscle (Wu et al. 1983).

Although the effects of DTX on transmitter release are relatively mild in normal larvae (Fig. 1), it produced strong enhancement of neuromuscular transmission in eag larvae or in normal preparations pre-treated with quinidine. A low dose (50 μmoll^-1) of quinidine alone only slightly increased the amplitude of EJPs (Fig. 4A,B) and sometimes produced repetitive discharges of motor axons (data not shown). Application of 4·2 μmoll^-1 DTX qualitatively altered the EJPs, increasing their amplitude and duration several times (Fig. 4C). A further increase in the concentration of DTX (8·4 μmoll^-1) enhanced transmitter release to an even greater extent. As shown in Fig. 4D, this resulted in large, prolonged EJPs similar to those caused by diminishing both I_A and I_K (see Figs 2D and 3C).

Consistent with the above finding, our experiments on eag larvae further demonstrated the facilitatory effect derived from the synergistic interaction of DTX with a defect in I_K. The EJPs in eag muscles were not substantially greater than normal (Fig. 5A, also see Ganetzky & Wu, 1983, 1985; Wu et al. 1983). Application of 4·2 μmoll^-1 DTX immediately increased the size of EJPs from 2 −4 mV to 5 − 10 mV (Fig. 5B) and in a few minutes caused repetitive discharges of EJPs of varying sizes (Fig. 5C), some reaching 30 mV. Adding 0·1 mmoll^-1 4-AP to the preparation transformed these repetitive EJPs into the typical prolonged giant EJPs (data not shown) observed when both I_A and I_K are blocked (see Figs 2D and 3C). In sum, our findings indicate that DTX has an effect on the neuromuscular junction similar to, but smaller than, those caused by the Sh mutation or 4-AP treatment.

Direct evidence for alterations in K⁺ currents caused by DTX requires measurements by voltage-clamp experiments. The Drosophila presynaptic nerve terminals are not easily accessible to this type of analysis and such measurements were performed in larval body-wall muscles. These large, identifiable muscle fibers meet isopotential conditions, are enveloped by very little connective tissue and are directly exposed to the external medium (Wu & Haugland, 1985). Experiments were first carried out in Ca²⁺-free saline to eliminate Ca²⁺ and Camdependent K⁺ currents. Under this condition only I_A and I_K remain, and they can be physiologically separated by their differences in inactivation properties (Wu & Haugland, 1985; Haugland, 1987; N. H. Haugland & C.-F. Wu, in preparation). The current-voltage (I-V) relationships in Figs 6 − 9 illustrate I_A and I_K, extracted as described in Materials and methods, and the effects of DTX at different Mg²⁺ concentrations and temperatures. The time courses of the total active currents (I_A plus I_K) are shown in Figs 6 and 8 and the extracted I_A currents are shown in Figs 7 and 9.

Previous studies of larval muscle currents were conducted at low temperature (4 − 15 °C) for the ease of voltage-clamp control and, to improve membrane stability, a high concentration of Mg²⁺ (10 − 15 mmoll^-1) was usually present in the Ca²⁺-free saline (Salkoff & Wyman, 1981; Wu & Haugland, 1985). The EJP experiments (Fig. 1) suggest that the DTX effect is suppressed by high Mg²⁺ concentration. We found that, in the presence of 14 mmoll^-1 Mg²⁺, DTX (1·4 μmoll^-1) did not exert any significant effect on either I_A or I_K at 5 °C (Figs 6 and 7, 14 mmoll^-1 Mg²⁺ data). Subsequently, the concentration of Mg²⁺ was lowered to examine the possible inhibitory effect of Mg²⁺ on DTX activity. At low temperature (5°C), we still could not detect a significant effect of DTX at the physiological concentration (4mmoll^-1) of Mg²⁺ (Figs 8 and 9, 5°C data). Not until the Mg²⁺ concentration was further lowered to a non-physiological level (0·5 mmoll^-1) could we observe a clear effect of DTX at 5°C. Long-term treatment (30 min) with DTX reduced both I_A and I_K (Figs 6 and 7, 0·5 mmoll^-1 Mg²⁺ data). The effect of DTX at low temperature and low Mg²⁺ concentration was not reversed by washing and could not be suppressed again if the Mg²⁺ concentration was raised (data not shown).

It should be noted that at 0 mmoll^-1 Ca²⁺ and 0·5mmoll^-1 Mg²⁺ the membrane underwent a characteristic sequence of changes even in the absence of DTX (see Discussion and Fig. 10 for details). At 5°C, the changes in membrane currents took about 50−70 min to reach a final steady state. Addition of DTX to saline at this time again initiated a similar sequence of events (see Discussion and Fig. 10) before reaching a new steady state 30 min later. The above results (Figs 6 and 7, 0·5 mmoll^-1 Mg²⁺ data) include only steady-state data.

The activity of DTX is known to be enhanced at high temperature in chick biventer cervicis and mouse diaphragm preparations (Harvey & Anderson, 1985). The effect of temperature on the action of DTX was therefore examined in saline containing a physiological concentration (4 mmoll^-1) of Mg²⁺. Raising the temperature to 16°C revealed a readily demonstrable effect of DTX. As previously reported (Wu & Haugland, 1985; Haugland, 1987), both I_A and 1_B approximately doubled in size for an increase in temperature of about 10 °C (see 5°C and 16°C data in Figs 8 and 9). At 16°C, DTX reduced the amplitude of I_A without any obvious change in its kinetics (Fig. 9). In contrast, I_K was largely unaltered (Fig. 8). The same experiment was performed at lower DTX concentrations. We found that at 470 nmoll^-1 DTX had nearly reached a plateau level in its inhibitory effect on I_A, with a reduction in I_A almost identical to that caused by 1·4 μmoll^-1 DTX. By extrapolation, the blockade of I_A by DTX might be larger at room temperature and could account for the observed changes in EJPs (Figs 1−5) at the physiological concentration of Mg²⁺ (4 mmoll^-1).

The same type of voltage-clamp analysis was carried out in the presence of 20 mmoll^-1 Ca²⁺ at both 5 and 16°C. Under these conditions, the inward Ca²⁺ current and outward Ca²⁺-dependent K⁺ currents could be examined (Singh & Wu, 1987, 1989) but, in more than 10 fibers, no apparent effects of DTX (l·4 μmoll^-1) on these currents could be demonstrated (data not shown). Even though this result is consistent with previous reports on several vertebrate preparations (Halliwell et al. 1986; Harvey, 1987), we cannot rule out the possibility of a minor effect on these channels in Drosophila. Since the Camdependent currents are intrinsically more variable than I_A and I_K among Drosophila muscle fibers (Singh & Wu, 1989), a small quantitative change in their amplitude would be difficult to detect.

On the basis of electrophysiological and pharmacological criteria, a variety of functional types of K⁺ channels have been identified in a large number of vertebrate and invertebrate species, but there is no information about their molecular distinctions and similarities (Latorre & Miller, 1983; Rudy, 1988). Mutations that alter K⁺ currents can provide insight into this problem (Wu & Ganetzky, 1986). A combination of genetic and pharmacological techniques may prove to be a powerful approach to the study of the functioning and mechanisms of this class of membrane channels.

So far, only a few specific, high-affinity toxins for K⁺ channels are known. Thus DTX represents an especially valuable probe for functional and biochemical analyses of these channels (Dolly et al. 1984, 1987; Rehm & Lazdunski, 1988). Previous voltage-clamp studies on vertebrate preparations have indicated that the sensitivity of K⁺ channels to DTX in different excitable cells does not necessarily reflect similarity in channel function. DTX suppresses the non-inactivating or slowly inactivating K⁺ currents in the node of Ranvier (Weller et al. 1985; Benoit & Dubois, 1986), in nodose ganglion neurons (Dolly et al. 1987), in dorsal root ganglion neurons (Penner et al. 1986) and in visceral sensory neurons (Stansfeld et al. 1986) but acts on the rapidly inactivating transient I_A in hippocampal neurons (Halliwell et al. 1986). However, a rapidly inactivating variant of the transient K⁺ current in superior cervical ganglion neurons is reportedly not sensitive to DTX (Dolly et al. 1987). Recent toxin binding studies using DTXj (Rehm & Lazdunski, 1988) and molecular genetic analyses of the Sh gene in Drosophila (Kamb et al. 1988; Pongs et al. 1988; Schwarz et al. 1988) have suggested the possibility of multiple subtypes of closely related K⁺ channel proteins.

Our experiments extended the study of DTX to an invertebrate preparation. The action of toxin on muscle membrane was analyzed by voltage-clamp experiments. The results showed that, at physiological concentrations of Mg²⁺, DTX affects mainly I_A without apparent effects on other currents. Although direct voltage-clamp measurements of the DTX effects at higher temperatures are not available for technical reasons (see Materials and methods), our data at 16°C and 4 mmoll^-1 Mg²⁺ (Figs 8 and 9) could be extrapolated to infer a similar, or a greater, DTX effect on I_A at 25−9°C to account for the observed changes in EJPs (Figs 1−5). At low divalent cation concentrations I_K, in addition to I_A, was reduced (Fig. 6).

The exact mechanisms of DTX actions on membrane channels remain unknown. It has been previously reported that the toxin potency of enhancing neuromuscular transmission decreases drastically when temperature is lowered or when Mg²⁺ concentration is raised (Harvey & Anderson, 1985). In Drosophila, the effect of DTX also critically depends on temperature and the concentration of Mg²⁺, and possibly Ca²⁺, in the saline. It is worth pointing out the intriguing interactions between DTX and Mg²⁺, as suggested by the experiments in Ca²⁺-free saline at 5 °C (Fig. 10). Decreasing the Mg²⁺ concentration to 0·5 mmoll^-1 initiated a sequence of changes in the amplitude and kinetics of membrane currents. Before reaching a steady state in 70 min, the voltage dependence of the total current was strikingly different (Fig. 10 A). Outward currents larger than normal were elicited by strong depolarizing steps but the current turned inwards at some intermediate levels of depolarization (Fig. 10 Aii, arrow). At the steady state, the extracted I_A was reduced in amplitude whereas I_K retained nearly normal amplitude but seemed to rise more rapidly (Figs 6 and 7, 0 μmoll^-1 DTX data, and Fig. 10Bi). Interestingly, when DTX was added at this time, the membrane currents again underwent characteristic changes similar to the above sequence of events (Fig. 10Bi,ii). At the new steady state there was a decrease in both I_A and I_K (Figs 6 and 7,0·5 mmoll^-1 Mg²⁺ data, and Fig. 10Biii). These non-physiological changes induced by low Mg²⁺ concentration were prevented by the presence of DTX. Replacing 14 mmol I^-1 Mg²⁺ saline by 0-5 mmol I^-1 Mg²⁺ saline containing DTX caused only a gradual reduction in the amplitude of I_A and I_K without evident alterations in their kinetics (Fig. 10C).

The basis of the non-physiological events initiated by reducing Mg²⁺ concentration (Fig. 10A) is not understood. Divalent cations can change the local electric potential profile in the membrane by interaction with fixed negative charges on its outer surface (Hille, 1984). Divalent cations may also directly affect the gating of certain ion channels (Dingledine, 1986; Armstrong & Lopez-Bameo, 1987; Stern et al. 1987). One possibility is that an inward current was induced transiently after reducing the Mg²⁺ concentration in the Ca²⁺-free saline (Fig. 10Aii, Bii, arrows). Permeation of Na⁺ through muscle Ca²⁺ channels in Ca²⁺-free saline has been observed in one insect species (Yamamoto, 1987) and some vertebrate species (McCleskey et al. 1986) but not in another insect species (Ashcroft & Stanfield, 1982). Whether permeation of Na⁺ is responsible for the phenomenon in Drosophila muscle at very low cation concentrations must await further investigation. No matter what exact mechanism underlies these changes, the surprising effect of DTX, preventing their initiation (Fig. 10C) and reiterating the sequence of changes once they are induced by Mg²⁺ reduction (Fig. 10B), must be considered in future physicochemical approaches to the mechanism of DTX actions.

The effects of several presynaptic snake toxins, such as β-bungarotoxin (β-BTX) and related toxins, are known to show a similar temperature and cation dependence (Chang, 1985). More interestingly, DTX, /3-BTX and another toxin, the mast cell degranulating peptide, have been shown to bind to common protein targets and inhibit K⁺ currents (Petersen et al. 1986; Bidard et al. 1987; Stansfeld et al. 1987). The structures of these toxins and DTX have been completely or partially determined (Lee, 1979; Joubert & Taljaard, 1980; Bidard et al. 1987). Such structural information may be important in designing further biophysical and biochemical experiments to study the nature of the observed temperature and cation dependence, which may in turn give clues to the mechanism of interaction of toxins with relevant membrane sites.

Direct measurements of the various ion currents in presynaptic nerve terminals by voltage-clamp are technically difficult. Therefore, Drosophila mutants of the genes that encode specific types of ion channels provide a unique opportunity to infer a defective current underlying the mutant phenotypes in nerve terminals. The altered currents can be directly analyzed by the voltage-clamp method in larger cells, such as muscle fibers.

It is worth noting that the DTX effect is specific but relatively weak compared with the actions of other channel-specific toxins or drugs. As in Drosophila larval muscle, DTX reduces but does not eliminate the K⁺ currents in different vertebrate preparations (Weller et al. 1985; Halliwell et al. 1986; Penner et al. 1986; Stansfeld et al. 1986). At mouse and frog neuromuscular junctions, DTX enhances transmission by a mild increase in the quantal content of EPPs and the occasional appearance of repetitive EPPs (Anderson & Harvey, 1985; Harvey & Anderson, 1985; Harvey, 1987). As in vertebrate preparations, the effect of DTX at Drosophila neuromuscular junctions is relatively mild (Fig. 1). Similar but more extreme effects are observed when I_A is eliminated by Sh mutations or treatment with 4-AP (Figs 2 and 3). Previous extracellular nerve recordings and focal EJP recordings in Sh and eag larvae and in normal larvae treated with 4-AP or tetraethylammonium show that the abnormal EJPs reflect predominantly presynaptic defects rather than a failure of postsynaptic membrane repolarization.

The nerve terminals fail to repolarize properly, resulting in repetitive firing of the axon and release of an abnormally large amount of transmitter (Jan et al. 1977; Ganetzky & Wu, 1982). The results of the present study indicate that DTX facilitates larval neuromuscular transmission, probably by reducing I_A but not other K⁺ currents in the nerve terminal.

We found that the relatively mild action of DTX on synaptic transmission can, nevertheless, be enhanced and clearly demonstrated through its synergistic interactions with mutations or drugs that reduce different K⁺ currents (Figs 4 and 5). The synergistic effect of DTX on Sh, eag and slo mutants could be readily studied to obtain information regarding the type of K⁺ currents (i.e. I_A, I_K and I_CF) in the nerve terminal affected by DTX. The results were also corroborated by examining the synergism of DTX with drugs blocking I_A or I_K. At the neuromuscular junction, DTX effectively augmented the effect of quinidine or the eag mutation but did not further increase the effect of 4-AP or the Sh mutation.

Two conclusions may be drawn from these results. First, the I_A channels in the nerve terminal and muscle membrane may be very similar or identical in their properties, as suggested by their similar sensitivity to DTX as well as to 4-AP and the Sh mutation. Second, the regulation of synaptic transmission by interplay of different K⁺ currents is intricate and can produce results in a wide dynamic range. Blocking I_K produced a relatively smaller effect on EJPs than blocking I_A (compare Figs 2B and 3 A with 4B). Partial blockade of I_A by DTX resulted in only mild effects on EJPs at physiological concentrations of Mg^2-1- (Fig. 1A,B) but could lead to explosive discharges of repetitive or prolonged giant EJPs (Figs 4D and 5C) when I_K was also blocked or reduced. Since multiple K⁺ conductances are also present in vertebrate nerve terminals (David & Yaari, 1985; Mallart, 1985), our understanding of the ionic mechanisms underlying the generation of repetitive nerve spikes and increased synaptic activity in certain human pathological conditions (Dichter & Ayala, 1987) may be improved by further studies of the functional consequences of K⁺ channel-specific toxins and mutations.

We thank Drs F. J. Joubert and F. H. H. Carlsson (National Chemical Research Laboratory, CSIR, Pretoria, South Africa) for kindly providing DTX, Mr C.-Y. Han for technical assistance, and Mr P. Taft, Mrs K. Howard and Mrs G. Folkins for helping with the preparation of this manuscript. This work was supported by a grant (NSC-76-0412-B002-103) from the National Science Council, ROC, and US Public Health Service grants NS00675, NS18500 and NS15797.