Properties of potassium currents and their role in membrane excitability in Drosophila larval muscle fibers

The existence of mutations which affect specific ionic currents in Drosophila provides an opportunity to analyze the properties of individual currents and to study the roles of these currents in membrane excitability. The advantage of mutational analysis can be complemented with the use of pharmacological agents that specifically block certain currents. The larval body-wall muscles in Drosophila support four K⁺ currents which can be distinguished and separated from each other with the use of mutations and drugs. These currents include two voltage-activated K⁺ currents, the fast I_A and the delayed I_K, and two Ca²⁺-activated K⁺ currents, the fast I_CF and the delayed I_cs (Salkoff, 1983; Wu et al. 1983; Wu and Haugland, 1985; Gho and Mallart, 1986; Wei and Salkoff, 1986; Singh and Wu, 1989). Among these currents, I_A and 1er are eliminated by the Shaker (Sh) (Salkoff and Wyman, 1981; Wu et al. 1983; Wu and Haugland, 1985) and the slowpoke (slo) (Elkins et al. 1986; Singh and Wu, 1989) mutations, respectively, and I_K is blocked by quinidine at micromolar concentrations (Singh and Wu, 1989). We have used Sh, slo and quinidine to study properties of the K⁺ currents that have not been described previously, and to look at their role in membrane excitability in larval body-wall muscle fibers.

Different names have been used previously for the Ca²⁺-activated K⁺ currents in Drosophila. The fast current has been designated as I_Acd (Salkoff, 1983), I_Acd (Gho and Mallart, 1986; Wei and Salkoff, 1986) and I_c (Elkins et al. 1986), and the slow current as I_KC (Wei and Salkoff, 1986) and I_c (Gho and Mallart, 1986). To avoid confusion and to preserve consistency with the conventional names used in other species we use the present nomenclature, i.e. I_CF for the fast C current and Ics for the slow C current (Singh and Wu, 1989). Part of the results presented here have appeared in abstract form previously (Singh et al. 1986; Singh and Wu, 1987).

The wild-type strain Canton-S was used as the normal control. An extreme allele of the Shaker (Kaplan and Trout, 1969) locus, Sh^KS133 (Wu and Haugland, 1985), was used for eliminating I_A. Only one allele of slo (Elkins et al. 1986) is available and was used in these experiments. These two mutant alleles have been extensively characterized in larval muscle (Wu et al. 1983, 1989; Wu and Haugland, 1985; Singh and Wu, 1989; Haugland and Wu, 1990).

The body-wall muscle preparation of mature third-instar larvae (Jan and Jan, 1976) was used for recordings, as described previously (Wu and Haugland, 1985; Singh and Wu, 1989). Two-microelectrode voltage-clamp recordings were made with a holding potential of −50 or −80 mV, as specified in the figure legends. Both the voltage-clamp and the current-clamp recordings were made at 4°C. Recording saline contained (in mmol 1⁻¹) 128 NaCl; 35.5 sucrose; 2 KC1 and 5 Hepes, and was buffered at pH7.1. In addition, the Ca²⁺-free saline had 0.5mmoll⁻¹ EGTAand MgCl₂ and was adjusted to 20 mmol 1⁻¹ for improving membrane stability (Wu and Haugland, 1985). Unless otherwise specified, the saline with Ca²⁺ contained 20 mmol 1⁻¹ CaCl₂, 4 mmol 1⁻¹ MgCl₂ and no EGTA. Quinidine and 4-aminopyridine (4-AP) were obtained from Sigma Chemical Co. (St Louis, MO).

The muscle fibers used in these experiments are supercontracting muscles (Osborne, 1967). The contractions induced during recordings, especially during strong depolarizations, damage the fiber quickly. Although it is possible to make refiable recordings from a fiber, it is difficult to take a single fiber through repeated experiments using different experimental paradigms or recording solutions. Therefore, all individual experiments described here were performed on separate fibers.

Voltage or current pulses were generated at intervals of 10 s by a Master-8-cp pulse generator (A.M.P.I., Jerusalem, Israel). The voltage-clamp circuit was equipped with a high-voltage head stage (model HVHD, Almost Perfect Electronics, Basel, Switzerland). The current electrode was shielded up to a few millimeters from the tip with a coaxial double shield. The inner shield was driven at the potential of the electrode and the outer shield was grounded. The voltage electrode was filled with KC1 and the current electrode with a mixture of KC1 and potassium citrate. The electrodes were pulled to a resistance between 5 and 15 MΩ. The current signal was recorded through the virtual ground using the amplifier M701 (WPI, New Haven, CT). The current and the voltage signals were pulse-coded using a Neurocorder unit DR-384 (Neuro Data Instruments Corp., New York, NY), and recorded using a video cassette recorder. The data were digitized and subsequently analyzed on an Apple-Macintosh II computer. Current densities in voltage-clamp recordings were obtained by normalizing the current with respect to the fiber capacitance (Haugland, 1987; Haugland and Wu, 1990). Leakage current was subtracted digitally and the initial 3 ms of recordings which contain capacitive transients have been omitted in the figures for the sake of clarity, as described previously (Singh and Wu, 1989).

Voltage-clamp recordings made in 1.8 mmol 1⁻¹ Ca²⁺, the concentration generally used in Drosophila saline, showed considerable variation in the Ca²⁺-dependent currents. This problem was overcome by increasing the external Ca²⁺ concentration. Saline with 20 mmol 1⁻¹ Ca²⁺ was therefore used (Singh and Wu, 1989) for voltage-clamp measurements of Ca²⁺-dependent currents in this paper.

Early membrane currents recorded from different genotypes are shown in Fig. 1. The fast current peak seen in the normal fibers (Fig. 1A) consists of two outward currents, the voltage-activated (I_A) and the Ca²⁺-activated (I_CF) K⁺ currents, and an inward Ca²⁺-current, I_ca-As shown previously (Gho and Mallart, 1986; Singh and Wu, 1989), the outward currents mask the inward current which is therefore not apparent in the figure. Lack of I_CF in the slo mutation (Elkins et al. 1986; Singh and Wu, 1989) revealed the sum of I_A and lea, in which the outward peak was reduced but the inward current was still masked (Fig. 1B). In a similar way, the Sh mutant that lacks I_A (Salkoff and Wyman, 1981; Wu et al. 1983; Wu and Haugland, 1985) showed the sum of I_CF and Ica (Fig. 1C). The inward Ca²⁺ current can be seen during the short interval before it is masked by the outward ICF-Since these fibers have both I_A and I_CF, the total outward current is stronger than in Sh or in slo. Elimination of both the outward currents (I_A and I_CF) in the Sh;slo double mutant (Fig. 1D) unmasks the inward I_ca

Both I_A and I_CF in larvae (Wu and Haugland, 1985; Gho and Mallart, 1986) and in adults (Salkoff, 1983; Salkoff and Wyman, 1983) have been shown to inactivate with membrane depolarization. The ICF inactivation in larvae is almost complete at a membrane potential of −20mV (Gho and Mallart, 1986), a potential at which lea or ICF have still not been activated. This implies that the I_CF inactivation is voltage-dependent and not Ca²⁺-dependent. However, there is no study on the recovery of ICF from inactivation in larvae. This was studied by using different experimental protocols and recovery times in recordings from Sh (44 fibers) and Sh;slo (36 fibers). Fig. 2 shows the effect of a 140ms pre-pulse to +20mV on ICF and lea-Their recovery, after different times at the holding potential of −50 mV, was tested by an identical test pulse. The pulse paradigm is shown in Fig. 2A. The first traces in both the Sh and the Sh;slo recordings in Fig. 2B show the membrane current during the pre-pulse, which serves as a control. Since I_K and Ics rise slowly with a delay, the early phase (0–40 ms) of this current mainly consists of the inward lea in Sh;slo and the sum of lea and I_CF in Sh. The traces on the right show membrane currents in response to the test pulse, after the indicated recovery period at −50 mV. I_CF was considerably inactivated by the pre-pulse. Note that the first test pulse (after 0.1s recovery time) shows stronger inward current than the control current seen with the pre-pulse in Sh (Fig. 2B). Since the Sh fibers lack I_A, the inactivation of I_CF unmasks the inward current to a greater degree. Even if inactivation did occur in lea during the pre-pulse, it recovered during the first 100ms, as indicated by recordings in Sh;slo. In contrast, I_CF showed clear inactivation which did not recover completely within 2 s (compare Sh and Sh;slo in Fig. 2B). The recovery from inactivation of I_cF was slower than that of I_A, which shows substantial recovery within I_cs at a comparable temperature (Wu and Haugland, 1985). The main result is that the recovery of 1er from inactivation may not be linked to that of I_ca

The possibility of extracting I_CF with the use of mutations enabled us to look at the relationship between this current and the inward Ca²⁺ current that is responsible for its activation. The current-voltage relationship for the extracted ICF is given in Fig. 3A. The values for ICF were obtained by subtracting the averaged current densities recorded in Sh;slo from those recorded in Sh. The current for Sh is shown as the early peak of the outward current at different membrane potentials. For the purpose of subtraction, the inward I_Ca in Sh;slo was measured at the time when a peak occurred for the outward current in Sh at the corresponding voltage. This time was a few milliseconds after the actual inward peak in Sh;slo (see Fig. 1C and D), which gives values very similar to the ones shown in Fig. 3A.

Fig. 3B shows the relationship between I_CF and l_ca, determined as described for Fig. 3A. It shows a linear dependence of I_CF on lea-This is consistent with a one-to-one correspondence between the Ca²⁺ entering the fiber and the opening of individual I_CF channels. A linear relationship between the amount of Ca²⁺ injected directly into the cell and the value of the Ca²⁺-activated K⁺ current has also been reported in the dorsal unpaired median neurons in the cockroach (Thomas, 1984). However, the data cannot be taken as conclusive evidence for a one-to-one correspondence, because the amount of free Ca²⁺ in the cytosol depends on the influx of Ca²⁺, on the nature of the Ca²⁺ sequestering process and on a number of other factors.

In addition to the fast Ca²⁺-activated I_CF, the larval muscle fibers in Drosophila also show a slow Ca²⁺-activated outward Ics (Gho and Mallart, 1986; Singh and Wu, 1989). We performed experiments with 0 mmoll⁻¹ Ca²⁺ (20 fibers for Sh and 10 fibers for Sh;slo) or 20 mmol 1⁻¹ Ca²⁺ (19 fibers for Sh and 13 fibers for Sh;slo) in the saline. Ics was observed in recordings made in saline with 20 mmol 1⁻¹ Ca²⁺. Sh and Sh;slo gave similar results because I_CF is almost completely inactivated during the 500 ms pulse used in these experiments (Fig. 2 shows inactivation produced by a 140 ms pulse) and the contamination from residual I_CF is expected to be minor. Fig. 4A shows the voltage-activated I_K in Ca²⁺-free saline, which reaches a steady state before the end of the 500 ms pulse. In Fig. 4B,C, the delayed outward current recorded in saline with Ca²⁺ consisted of I_K and I_cs-I_cs can be seen as changes in the overall amplitude of the slow outward current and the tail currents after the voltage pulse has ended. Unlike I_CF, I_CS did not inactivate, and in fact continued to rise for at least 500 ms. This contrast between the fast and the slow Ca²⁺-activated outward currents, I_CF and I_cs, parallels the difference between the two voltage-activated outward currents, the fast transient I_A and the delayed sustained I_K

The tail currents were recorded by stepping the membrane potential to +20 mV and then repolarizing it to different voltages (Fig. 4). The tail currents immediately after the end of the voltage pulse are either inward or outward, depending on the membrane potential reached. Inward tail currents are expected if the membrane potential is repolarized below the reversal potential for the current, and outward tails are expected if it is above the reversal potential. Only outward tail currents were seen in recordings made in Ca²⁺-free saline (Fig. 4A). This is because I_K does not show inward tail currents as there is strong rectification (Wu and Haugland, 1985). Inward tails were seen when recordings were made in the presence of Ca²⁺. Fig. 4B shows inward tail currents for Ics and combined outward tail currents for I_K and I_Cs-Inward tails are seen for potentials up to −40 mV. At −30 mV, the tails were completely outward. At this potential, I_cs makes only a marginal contribution to the outward tail in Fig. 4B, as can be seen by comparing with the tail at −30 mV in Fig. 4A. The reversal potential for I_Cs in Fig. 4B thus lies between −40 and −30mV, near to −30mV. In different fibers, the reversal potential ranged between −30 and −50 mV.

In some fibers, depolarization by repeated test pulses showed a cumulative effect on the outward current and gave rise to more complex tail currents. Fig. 4C shows an example where inward tail current is seen up to −20 mV. The outward tail for −10 mV (Fig. 4C) is also not as strong as the I_K tail at the same voltage (Fig. 4A). The reversal potential for the current in addition to I_K in this case seems to be more positive than −10 mV. A possible interpretation of this is suggested in the Discussion.

The ability to eliminate individual K⁺ currents by means of mutations and pharmacological agents allows an analysis of the role of different currents in membrane excitability. Mutational elimination of I_A and I_CF has been used in adult dorsal longitudinal flight muscles (DLM) of Drosophila for this purpose (Elkins and Ganetzky, 1988). We used the lack of I_A in Sh and the lack of I_CF in slo to study the role of these currents in the excitability of larval muscle fibers. This was done by looking at the membrane potential in response to constant current injection.

Regenerative potentials have recently been demonstrated in larval preparations in which air was supplied through tracheoles to maintain the resting potentials up to −80mV (Yamaoka and Ikeda, 1988). Most previous reports, which used unaerated preparations, described lower resting potentials and only graded potentials in response to current injection in normal physiological saline, which contains 1.8mmoll⁻¹ Ca²⁺ (Suzuki and Kano, 1977; Wu and Ganetzky, 1988). We found that regenerative responses were easier to elicit when the resting potential was sufficiently negative, either in a freshly dissected preparation or because of injection of a hyperpolarizing current even in the absence of aeration. Recordings were made from the normal (4 fibers), Sh (9 fibers) and slo (13 fibers). Fig. 5 shows recordings made in fibers with the resting potential kept near −60 to −70 mV by injecting hyperpolarizing current. The regenerative peak in these recordings is more pronounced in slo and in Sh. A smaller amount of current was required to elicit regenerative potentials in slo as compared to Sh. This indicates that I_CT is more effective than I_A in repolarizing the regenerative potential supported by I_Ca.

The relative role of different outward currents can be seen more clearly by using saline containing 20mmoll⁻¹ Ca²⁺. With increased Ca²⁺ concentration, nounced regenerative potentials could be elicited when one or more outward currents were removed by mutations or drugs. This could be achieved without aeration in preparations with resting potentials as low as −20 mV. Fig. 6 shows recordings made in saline containing 20 mmol 1⁻¹ Ca²⁺ without injection of hyperpolarizing current. The normal fibers showed graded potentials in response to increasing amounts of current injection (Fig. 6A) and did not show action potentials. Elimination of I_A by Sh led to increased depolarization, which was still graded in nature (Fig. 6B). However, when I_CF was eliminated by the slo mutation, the fibers underwent all-or-none action potentials (Fig. 6C). This is consistent with the observation shown in Fig. 5 that I_CF plays a stronger role than I_A in repolarizing the membrane. As expected, slo fibers treated with 4-AP (100μmoll⁻¹), which preferentially blocks I_A at this concentration, also exhibit all-or-none action potentials, but requiring even less injection of current (Fig. 6D) than slo (Fig. 6C). The number of fibers tested was 11 for the normal, 10 for Sh, 16 for slo and 22 for slo with 4-AP.

Quinidine, which blocks different K⁺ currents in several systems (Fishman and Spector, 1981; Hermann and Gorman, 1984; Imaizumi and Giles, 1987), blocks I_K in Drosophila (Singh and Wu, 1989). The inset in Fig. 7 shows membrane currents in response to voltage steps from the holding potential of −80 mV to −10, +10 and +30mV in Ca²⁺-free saline with and without quinidine. I_A was removed by the use of Sh, and 0.1 mmol 1⁻¹ quinidine substantially blocked the only remaining current I_K-This concentration of quinidine has earlier been shown not to block I_A, I_CF or Ics (Singh and Wu, 1989). The double-reciprocal plot in Fig. 7 shows the effect of different concentrations of quinidine on the amplitude of I_K. Quinidine at 10 μmol 1⁻¹ reduced I_K by more than half. In this plot, the reciprocal of the fraction of inhibition is plotted against the reciprocal of quinidine concentration in micromoles per liter. A linear relationship would signify a simple saturation with one-to-one binding of quinidine molecules to the I_K channels. Fig. 7 shows a deviation from a linear relationship. Although I_K is blocked more than 50% by 10μmoll⁻¹ quinidine (as indicated by the upper-most point in the plot), even 1.0 mmol 1⁻¹ quinidine leaves some current intact (the lower-most point in the plot). It is not clear at this stage if this is a manifestation of sub-populations of I_K with different sensitivity to quinidine.

Some K⁺ channel blockers are known to show voltage-dependent action. The extent of inhibition by quinidine was examined at +10 and +30 mV in Drosophila. The plot shows small differences at the two potentials (Fig. 7). These differences could not be substantiated statistically, owing to the small sample sizes, but were far less pronounced than the voltage-dependence reported for 4-AP (Yeh et al. 1976) and tetraethylammonium (Armstrong, 1971).

Selective blockade of I_K by quinidine provided an opportunity to examine the role of this current in determining the membrane potential. Fig. 8 shows current clamp recordings in saline containing 20mmoll⁻¹ Ca²⁺ and 100μmoll⁻¹ quinidine. In the absence of quinidine, normal and Sh fibers did not show action potentials, as discussed above. Removal of I_K by quinidine did not show a strong effect except for a slight regenerative potential, as indicated by a shoulder in Sh after the end of the current injection (Fig. 8B). In the case of slo and Sh;slo, regenerative potentials were initiated even during weak current pulses. After the end of the pulse, the regenerative potential was sustained for hundreds of milliseconds (Fig. 8C,D). The duration of the prolonged depolarization was highly variable even between fibers of the same genotype and no consistent differences were found between slo and Sh;slo. Elimination of I_K thus prevents repolarization of Ca²⁺ regenerative potentials initiated either by the lack of ICF alone or by the elimination of both ICF and I_A (compare Fig. 8 with Fig. 6). The number of fibers tested was 9 for the normal, 8 for Sh, 10 for slo and 7 for Sh;slo.

The availability of the two mutations, Sh and slo, allows a comparison between I_A and ICF underlying the early phase of the membrane current. Fig. 1B,C shows that the generation of the Ca²⁺-activated I_CF is preceded by the voltage-activated I_A. Presumably, ICF requires Ca²⁺ influx following I_Ca activation, whereas I_A is activated by membrane depolarization without a lag following lea activation. As previously reported (Gho and Mallart, 1986), ICF, like I_A, was inactivated by membrane depolarization. ICF recovery from inactivation followed a time course much slower than that of Ic_a recovery (Fig. 2), suggesting a process independent of the availability of cytosolic free Ca²⁺.

The late outward currents consist of I_K and I_cs-I_K can be isolated by using Sh in Ca²⁺-free saline. This allows an analysis of the effect of quinidine on I_K. A double reciprocal plot between the fractional inhibition and the quinidine concentration did not indicate a linear relationship (Fig. 7). This observation suggests that there are different sub-populations of I_K channels with different sensitivities to quinidine, a possibility that requires further investigation.

As mentioned above, I_K does not show inward tail currents because of rectification (Fig. 4A). Outward currents are seen at potentials more positive than −60 mV but not at a potential of −70 mV. The reversal potential for this current is therefore between −60 and −70 mV, consistent with the earlier reported reversal potential for I_K (Wu and Haugland, 1985). Fig. 4B shows both outward and inward tail currents, recorded in saline containing Ca²⁺. In different fibers, the component contributed by Ics reversed at a potential between −50 and −30 mV (see Results). This is consistent with Ics being a K⁺ current in larvae (Gho and Mallart, 1986) as well as in adults (Wei and Salkoff, 1986). The currents observed in the presence of Ca²⁺ show a greater degree of variation in some fibers. Fig. 4C shows such a fiber with a reversal potential that is more positive than −10 mV. Such a reversal potential was generally associated with stronger tail currents as well as a cumulative effect on the variable outward currents during repeated voltage pulses (Fig. 4C). This type of behavior was generally elicited in fibers that were relatively leaky, which may be an indication of deterioration or damage. A more positive reversal potential may be accounted for by the presence of yet another set of channels or a non-selective conductance in addition to the Ics channels.

Yamaoka and Ikeda (1988) have recently shown that the tracheole aeration of the larval preparation is essential in maintaining the resting potential of the muscle fibers and in initiating a regenerative response. Lack of aeration results in a drop in the resting potential from about −80 mV to about −30 mV within l h at room temperature. It is known that the resting potentials in many insect muscles are maintained below the K⁺ equilibrium potential by electrogenic pumps (Huddart and Wood, 1966; Rheuben, 1972; Henon and Ikeda, 1981). The metabolic state, which would depend upon effective aeration, is thus expected to influence the resting potential. However, it is not clear if the effect of aeration on the regenerative response is due to the membrane potential, which would determine the state of channel inactivation, or to a direct deteriorative effect on channel machinery or to some other cellular factors. Recordings in this study and in other reports (Jan and Jan, 1976; Suzuki and Kano, 1977; Wu and Ganetzky, 1988) have been made in the larval preparation without tracheole aeration. Our recordings from muscle fibers with resting potentials at different levels indicated that the metabolic condition was probably not a prerequisite for regenerative responses. Regenerative responses were easier to elicit in fibers with resting potentials between −60 and −70 mV, maintained by injection of hyperpolarizing currents or by low temperature (4°C) in freshly dissected preparations. For example, regenerative potentials could be seen in Sh fibers only if the resting potential was more negative than −60 mV (compare Figs 5B and 6B; note the inflexion points in Fig. 5B). An effect of resting potential on the regenerative events is expected since the Ca²⁺ channels mediating inward current inactivate to different degrees depending upon the extent of membrane depolarization. The main aim of the present study was to compare the relative role of different K⁺ currents in membrane excitability. We found clear differences between the consequences of eliminating different currents on excitability, even without tracheole aeration. However, it is important to investigate these differences under normal physiological conditions with effective aeration.

An important consideration is the free Ca²⁺ concentration around the fibers as well as inside the fibers. This can be a source of variation in the Ca²⁺-activated K⁺ currents as it can affect the amount of Ca²⁺ influx and the Ca²⁺ available to trigger these currents. From electron microscopic observations (Jan and Jan, 1976; C.-F. Wu, unpublished data), the larval muscle fibers are enveloped by relatively little connective tissue. In addition, ion replacement and drug additions to the recording saline show immediate effects, indicating a lack of effective diffusion barriers (Wu and Haugland, 1985; Haugland, 1987; Singh and Wu, 1989; Wu et al. 1989). The level of intracellular free Ca²⁺ can conceivably vary in relation to metabolic state, e.g. a weakened sequestering mechanism, and affect the Ca²⁺ concentration gradient across the plasma membrane as well as the available free Ca²⁺ for activating ICF and Ics. It will be important to determine the free Ca²⁺ level directly, employing other available techniques in future investigations.

Interactions between the outward and the inward currents determine the membrane potential in response to a current stimulus. The inward current depolarizes the membrane and the outward current counteracts this by repolarizing the membrane. As in other arthropod species, the inward current in Drosophila muscles is carried by Ca²⁺ (Suzuki and Kano, 1977; Salkoff and Wyman, 1983; Elkins et al. 1986; Gho and Mallart, 1986). A tetrodotoxin-sensitive inward current was recently suggested by current-clamp experiments (Yamaoka and Ikeda, 1988). It remains to be determined by voltage-clamp experiments whether it is a Na⁺ inward current. The outward currents in Drosophila muscles consist of two voltage-activated (I_A and I_K) and two Ca²⁺-activated (I_CF and I_cs) K⁺ currents. The effectiveness of different K⁺ currents in repolarizing the membrane depends on their channel density (current amplitude) and their kinetic characteristics. Because of these differences, they give rise to a variety of excitability patterns.

An ability to manipulate different currents, by the use of mutations and drugs, provides an opportunity to examine the role of individual currents in excitability. The two fast outward currents (I_A and I_CF) in the normal fibers limit the membrane depolarization immediately after the initiation of the current injection. As compared to the normal muscles, therefore, a lack of I_A in the Sh muscles allowed a larger initial depolarization with the same amount of current injection (compare Fig. 6A and 6B). However, removal of ICF alone, in slo, had a greater effect, resulting in action potentials, even at a low resting membrane potential (see Fig. 6C). This implies that I_CF plays a more effective role in membrane repolarization than does I_A during prolonged current injections. It is known that I_A inactivates rapidly (Salkoff and Wyman, 1983; Wu and Haugland, 1985) and is effective in delaying membrane excitation (Salkoff and Wyman, 1983; Elkins and Ganetzky, 1988) but would not have a prolonged repolarizing effect. Although I_CF is similar in magnitude to I_A, its activation is coupled to the activation of Ic_a-In addition, it inactivates at more depolarized potentials (Gho and Mallart, 1986) and with a slower time course (S. Singh and C.-F. Wu, unpublished observations) than I_A. The effect of I_A removal became more apparent only when the resting potential was more negative (Fig. 5B) or when I_CF was also removed (action potentials in Fig. 6D require less current injection compared to Fig. 6C). These observations on the relative role of I_A and I_CF in the excitability of larval muscles are consistent with the earlier conclusions on their role in the adult flight muscles (Elkins and Ganetzky, 1988).

In contrast to the early currents, the slow outward currents activate with delay but show little inactivation. The role of I_K in muscle repolarization is revealed by the use of quinidine. In normal and Sh fibers, blockade of I_K by quinidine led to only marginal differences. However, when the membrane undergoes an action potential, as when I_CF is removed in slo or Sh;slo, I_K assumes an important role in repolarization. The action potentials lasted only a few tens of milliseconds when I_K was present (Fig. 6C,D) but lasted for several hundred milliseconds when I_K was blocked by quinidine (Fig. 8C,D). In the experiments shown in Figs 6 and 8, I_A and I_CF are expected to have inactivated to a great extent by the end of the current pulse. The repolarization of the action potential, once initiated, thus depends mainly on I_K and I_cs-With the further removal of I_K, the termination of the prolonged action potential after the end of the current pulse will depend primarily on the amplitudes of I_ca and I_cs, and how they change with time. The membrane during this prolonged depolarization is presumably in an unstable equilibrium and the time taken by the membrane to repolarize varied considerably, regardless of the genotype (data not shown). The sustained action potential is similar to, but less prolonged than that seen in the larval muscle fibers treated with tetraethyl ammonium (Suzuki and Kano, 1977; S. Singh and C.-F. Wu, unpublished observations), which blocks all four K⁺ currents in this system to different degrees (Wu and Haugland, 1985; Gho and Mallart, 1986). So far no mutations or drugs remove Ics specifically. Such a possibility would be very useful for determining the role of Ics in membrane excitability.

This work was supported by USPHS grants NS-15350, NS-18500 and NS-26528.