Characteristics of Five Types of K⁺ Channel in Cultured Locust Muscle

Studies of voltage-sensitive K⁺ currents have revealed a diversity of K⁺ channels which are present in the membranes of many cell types and which frequently co-exist in the same membrane. Single-channel conductances range from the 2 pS delayed rectifier of snail (Helix pomatia) neurones (Lux et al. 1981) to the 250 pS maxi-K⁺ channel found in rabbit sarcoplasmic reticulum (Latorre, 1986). K⁺ channels differ from one another in their current-voltage (I–V) relationships, gating kinetics, structure, pharmacology and function (for reviews see Latorre and Miller, 1983; Rudy, 1988; Kolb, 1989).

Particularly detailed studies of K⁺ channels have been undertaken in insect muscle, both in vivo and in vitro, using the dorsal longitudinal flight muscles of larval and adult Drosophila melanogaster. Studies of wildtype and mutant individuals have disclosed several types of K⁺ current. First, a Ca^2+-sensitive, depolarisation-activated, transient, K⁺ current (I_A) which exhibits rapid activation and inactivation kinetics (Salkoff, 1983; Salkoff and Wyman, 1983; Wu et al. 1983). This current is present in both adult Drosophila and cultured myotubes (Sole et al. 1987; Wu and Ganetzky, 1988; Zagotta et al. 1988). A K⁺ channel with a conductance of 14 pS has been shown to be responsible for the I_A current in Drosophila myotubes (Zagotta et al. 1988). Second, an outward K⁺ current (I_K), exhibiting slow inactivation, which appears in adult Drosophila muscle at about 96 h post-pupation. This I_K current is also present in cultured myotubes after 16 h in vitro. Two K⁺ channels, designated K_D (14pS) and K_o(40pS), contribute to this current, with the K_D channel contributing the largest fraction (Zagotta et al. 1988). Third, a Ca^2+-dependent K⁺ current (I_Kca) found in the muscles of third-instar Drosophila larvae. Fourth, a Ca^2+-activated K⁺ current (I_Acd) found in the dorsal longitudinal flight muscles of adult Drosophila, which becomes active after I_KCa matures in young adults. This current exhibits steady-state inactivation kinetics and is blocked by Ba²⁺ (Wu, 1988; Wu and Ganetzky, 1988). Fifth, a stretch-sensitive K⁺ channel (90pS), designated K_ST, which is present in myotubes and is voltage-and Ca^2+-independent (Zagotta et al. 1988).

K⁺ and Cl^- are responsible for the resting membrane permeability of locust skeletal muscle. The concentration gradients of these ions are maintained by metabolic pumps (Usherwood, 1967, 1969; Piek and Njio, 1979; Pichón and Ashcroft, 1985; Leech, 1986). Depolarisation of the locust muscle fibre leads to a voltage-dependent increase in the Na⁺ permeability, possibly accompanied by an increase in Ca²⁺ permeability and a delayed increase in K⁺ permeability (Usherwood, 1969). Current-voltage studies of the locust retractor unguis muscle undertaken in environments with varying K⁺ concentrations reveal the presence of inward and delayed rectifying K⁺ currents (Usherwood, 1969). The following is a report on the voltage-sensitive K⁺ channels present in cultured locust muscle.

Primary cultures containing skeletal myofibres were prepared from eviscerated, mechanically dissociated 11-to 12-day-old embryos of the locust Schistocerca gregaria. The cells were plated out in ‘5+4’ medium (Chen and Levi-Montalcini, 1969) using a modified version of the hanging-column technique (Shields et al. 1975). Cultures were incubated in a humidified atmosphere at 28°C. Full details of the methodology are contained in a paper by Duce and Usherwood (1986).

Locust muscle cultures were used for single-channel studies employing giga-ohm patch-clamp techniques (Hamill et al. 1981). Recordings were made in the cell-attached, inside-out and outside-out modes. For the recordings the growth medium was replaced by standard locust saline [180mmoll⁻¹ NaCl; 10mmoll⁻¹ KC1; 2mmoll⁻¹ CaClj; 3mmoll⁻¹ Hepes buffer, pH6.8; Usherwood and Grundfest (1965)]. The culture dish was placed on the stage of an inverted, phasecontrast microscope with a long-range condensor (Olympus C.K.). The indifferent electrode, a Ag/AgCl wire, was immersed in the saline and connected to the headstage of the patch-clamp amplifier (List LM-EPC7). All experiments were conducted at room temperature (20–24°C).

Patch pipettes, made from micro-haematocrit tubing (Hawksley and Sons Ltd), were pulled in two stages using a vertical microelectrode puller (Kopf Instruments, model 700C). Pipettes with a broad shaft and a fine tip were used, with resistances in the 3–8 MΩ range when filled with saline. The tips and shoulders of the pipettes were coated with Sylgard resin (Dow Corning). The pipettes were filled with a high-K⁺ saline in which the concentrations of Na⁺ and K⁺ in the standard saline were reversed, i.e. the pipette saline contained 180 mmol l⁻¹ KC1 and 10 mmol l⁻¹ NaCl.

Currents were recorded on a video cassette recorder (Sony β-max SI F30) using a pulse-code modulation unit (PCM 701 ES) modified to give a uniform frequency response from d.c. to 20kHz (Lamb, 1985).

Data were low-pass filtered at either 3 or 10 kHz using a Tektronix differential amplifier. Data reduction was undertaken using a MassComp MC5500 computer system, using procedures which are standard in our laboratory (Kerry et al. 1987a-c; Ball and Sansom, 1988). For analysis, the data were digitised at intervals of 50 μs (20 kHz) for current amplitude analysis and at intervals of 20 μs (50 kHz) for dwell time analysis. The reduced data were stored on hard disc.

For every 50 μs sample point of the digitised data the amplitude of a channel current (pA) was calculated and used to construct frequency histograms. Gaussian distributions were fitted to the frequency histograms using a non-linear leastsquares algorithm (NAG subroutine E04FDF), the means of which were taken to represent the currents of the closed and open states of the channel. The probability of a channel being in its open state (P_o) was determined from the integral of the points lying under the curves.

Channel dwell times were measured using a single cursor set halfway between the closed and open channel current levels of the digitised data (Colquhoun and Sigworth, 1983). Any open-state transitions which crossed this half-amplitude threshold were detected. Open and closed events longer than 60 μs were readily recognised using this approach.

Dwell time distributions were expressed as histograms. These histograms were normalized to give estimates of dwell time probability density functions (PDFs) (Colquhoun and Sigworth, 1983). Dwell time PDFs providing information on the number of open and closed states of the channel (Fredkin et al. 1985; Ball and Sansom, 1987; Kerry et al. 1987a) were displayed on logarithmic plots (Blatz and Magleby, 1986). To include channel dwell times ranging over several orders of magnitude, the bin widths were exponentially increased (Kerry et al. 1987b). The dwell time PDFs were fitted with one or more exponential functions using a simplex algorithm (NAG subroutine E04CCF), which used a method of maximum likelihood to determine the most probable parameters for the proportion (α) and the time-constant (τ) of each exponential component. These initial fits were then optimized using a Gauss-Newton-Marquardt algorithm (Marquardt, 1963; Kerry et al. 1987b) until a best fit was obtained. The minimum number of components required for a fit was determined using the asymptotic information criterion (AIC) method (Akaike, 1974; Landaw and DeStefano, 1984). The autocorrelation function was used to examine the correlation between the durations of successive openings or successive closings (Ball et al. 1985; Fredkin et al. 1985; Labarca et al. 1985; Kerry et al. 1987a,b).

The longevity of patches obtained from the cultured myofibres was dependent upon the age of the cultures. With cultures maintained for less than 2 months in vitro giga-ohm seals were often unstable and the patches usually lasted only a few minutes. With cultures maintained for more than 3 months in vitro giga-ohm seals were less easy to obtain.

In preliminary studies, it was found that there was no difference in the distribution of K⁺ channels along the length of a cultured myofibre. However, throughout this study most membrane patches were obtained from the middle region of a myofibre. About 60% of seals formed spontaneously upon contact between the patch pipette and the myofibre. The remainder were formed by applying a slight negative pressure to the patch pipette interior. K⁺ channels were usually seen as soon as a seal was formed and were observed in almost 50 % of the patches that were obtained.

102 recordings of K⁺ channel activity, each from a separate membrane patch, were analysed. Five types of K⁺ channel were identified in cell-attached and excised membrane patches, and have been designated types 1-5. In each case the channel type was characterised with respect to its open-channel conductance and voltage-sensitivity. All five types of K⁺ channel were operative in cell-attached patches at the resting membrane potential of the myofibre, i.e. at a pipette potential of 0 mV. Using the whole-cell recording mode the mean resting potential measured for the myofibres was −39±7mV (±S.D., N=2). Several pieces of evidence point to the K⁺ selectivity of these channels. First, the reversal potential and conductances of the single-channel currents were unaffected by the Cl^- channel blocker picrotoxin (1 mmol l⁻¹) or by the substitution of SO₄²⁻ for Cl-in the pipette and/or the bathing saline (Miller and Usherwood, 1988; Duce et al. 1988; Miller, 1988). Second, the reversal potential for the single-channel currents was shown to follow the K⁺ reversal potential (EK) as estimated using the Nernst equation for a pure K⁺ electrode. For cell-attached patches the internal K⁺ concentration was taken as 140 mmol l⁻¹ (Leech, 1986). With the cultured muscle bathed in standard saline and the patch pipette containing a high-K⁺ saline, E_K had an estimated value of −6mV, which corresponded to a pipette potential of −33 mV. Elevation of the K⁺ concentration in the bathing saline from the normal 10 to 95 mmol l⁻¹ shifted the single-channel I–V relationships to the right, i.e. the currents reversed near to a pipette potential of −20 mV. When the muscle was bathed in the high-K⁺ saline (180 mmol l⁻¹ K⁺) the currents reversed near to a pipette potential of OmV (N=33). The type 1 and type 2 K⁺ channels were observed most frequently, accounting for about 90 % of the channels seen in both cell-attached and excised patches.

The type 1 K⁺ channel was that most frequently observed. Recordings from twenty-one cell-attached, two outside-out and four inside-out patches containing type 1 channels were analysed. The I–V relationship for this channel in cell-attached patches was non-linear with marked inward rectification. From the I–V curve of Fig. 1A,B, estimated chord conductances of 43 pS for inward current over the pipette potential (V_pip) (range +100 mV to −40 mV) and 8pS for outward current between −40 mV and −60mV are obtained. The channel current reversed at a pipette potential of −40 mV (an approximate membrane potential of 0mV), which was close to E_K. Only outward currents were observed for the type 1 channel in excised, outside-out patches, to give a chord conductance of 5pS between pipette potentials of 0mV and +120 mV. In inside-out patches, only inward currents were observed, to give a chord conductance of 20pS between pipette potentials of +40 mV and +100 mV. There was no evidence of subconductance states in any of the recordings of type 1 channel currents. Block of the type 1 channel currents by Ba²⁺, 20 mmol l⁻¹ in the pipette saline, was not observed.

In Fig. 1C, ln[(1–P_o)/P_o] for the type 1 channel is plotted as a function of applied pipette potential (V_pip). The values of P_o were obtained by pooling data from twenty-one cell-attached patches. From this plot E_1/2 was calculated to be a pipette potential of −120 mV, which corresponded to a membrane potential of +80 mV, and z was calculated to be −0.4. P_o became progressively larger as the pipette potential was made more negative (Fig. 1D). Thus, the increase in P_o as V_pip was made more negative could result in the total current flow through the channel being ohmic, despite the obvious inward rectification indicated by the I–V relationship. To test this idea, the single-channel current at each V_pip was multiplied by P_o and the product was plotted as a function of the V_pip. This showed that the relationship between the total current flowing through the type 1 K⁺ channel and V_pip was not linear, but that P_o was considerably higher for all negative pipette potentials than for positive pipette potentials. P_o did not increase further when patches were excised (inside-out or outside-out for a variety of pipette potentials) into the bathing saline, which contained 2 mmol l⁻¹ Ca²⁺.

Most recordings of type 1 channels were too brief to enable a rigorous dwell time analysis to be undertaken. However, in one long-duration recording from a cell-attached patch, which appeared to contain only one channel, 2687 transitions between closed and open states were observed. The closed time PDF for these data was best fitted by the sum of four exponentials (Fig. 2A); the open time PDF was best-fitted by the sum of three exponentials (Fig. 2B). Autocorrelation function analysis of the channel closed times produced significant correlation, but the autocorrelation function for the open times showed that these were not correlated (Fig. 2C,D).

Recordings of type 2 K⁺ channel currents obtained from twenty-four cell-attached, nine outside-out and five inside-out patches were analysed. The type 2 channel currents exhibited inward rectification in cell-attached patches (Fig. 3), with a chord conductance of 115 pS for pipette potentials between +100 mV and −35mV and 29pS for pipette potentials between −30mV and −100mV. The channel current reversed at a pipette potential of −35 mV (a membrane potential of approximately −5 mV) which was near to the estimated E_K. In two cell-attached patches, subconductance states one-quarter, half and three-quarters of the amplitude of the full conductance level were seen. These subconductance states occurred in isolation, as well as preceding or following transitions to the full conductance state of the channel. Only outward currents were observed in outside-out patches. Between pipette potentials of 0mV and +100mV a chord conductance of 11 pS was obtained. In inside-out patches, the channel currents exhibited inward rectification, with a chord conductance of 28pS between pipette potentials of −140 mV and −70mV and 38pS between −70mV and +60mV.

Type 2 channel activity consisted of bursts of brief openings interrupted by periods of inactivity. Because of the fast kinetics of this channel and the problem of recording bandwidth-limitation, no attempt was made to measure the dwell times for this channel. P_o for this channel was not altered when inside-out patches were excised into bathing saline containing 2 mmol l⁻¹ Ca²⁺.

With 20 mmol l⁻¹ Ba²⁺ in the pipette saline there was clear evidence of channel block in cell-attached patches (Fig. 4). Both outward and the inward currents were affected by Ba²⁺, although inward currents were more affected than outward currents.

The following procedure was employed to determine whether the type 2 channel was ATP-sensitive. An inside-out patch containing at least one type 2 channel was held at a pipette potential of −90 mV. Single-channel currents were recorded from this patch prior to its perfusion with standard saline containing 1 mmol l⁻¹ ATP (Fig. 5). Within a few seconds of ATP perfusion the channel activity declined to zero. The ATP was then removed by washing the patch with standard saline. After ATP removal, channel activity was at a much higher level than before, with two channels now being active in the membrane patch. Similar results were obtained with three separate inside-out membrane patches. The appearance of two channel openings after wash-out may represent some form of post-inhibitory activation of the type 2 K⁺ channels (Findlay and Dunne, 1986).

Recordings of type 3 K⁺ channel activity obtained from eight cell-attached, nine outside-out and four inside-out patches were analysed. In cell-attached patches the I–V relationship for this channel was linear, with a conductance of 130pS (Fig. 6A,B). The channel current reversed at a V_pip of −30mV (a membrane potential of −10mV), which was close to the estimated E_K. Gating of the type 3 channel was voltage-dependent. In cell-attached patches, z was estimated to be −0.33 and E_1/2 to be +80mV (V_pip −119mV), respectively (Fig. 6C,D). The type 3 channel was sensitive to Ca²⁺. In cell-attached patches it exhibited only very brief openings, i.e. P_o was low, but in excised patches (inside-out) bathed in standard locust saline containing 2 mmol l⁻¹ Ca²⁺, P_o was markedly higher. The type 3 channel was not blocked by 20 mmol l⁻¹ Ba²⁺, in the pipette saline.

It was observed during the course of these experiments that the type 3 channel was sensitive to stretching of the membrane patch: the channel activity increased with suction. However, this stretch-activation was not followed up experimentally.

Only outward currents were observed in outside-out patches. The I-V relationship showed rectification and gave a maximum slope conductance of 135 pS (Fig. 7A,B). Single-channel currents recorded from outside-out patches exhibited subconductance states one-quarter, half and three-quarters of the amplitude of the full conductance level (Fig. 7C,D). Dwell time PDFs were fitted to data obtained from a recording of type 3 channel currents made from an outside-out patch held at a range of pipette potentials. The closed time PDFs constructed from these data were best fitted with three exponentials and the open time PDFs were best fitted by two exponentials (Table 1). As the pipette potential was made more positive the channel exhibited longer closings. The PDFs for 12391 close-open transitions recorded at a pipette potential of +40 mV are shown in Fig. 8A,B. Significant correlations were found between successive openings and successive closings (Fig. 8C,D). In inside-out patches, the type 3 channel had a maximal chord conductance of 75 pS and the channel currents reversed at a pipette potential of −70 mV.

Nine recordings containing the type 4 K⁺ channel, seven from cell-attached patches and two from inside-out patches, were obtained. This channel was not observed in outside-out patches. In cell-attached patches the I–V relationship for the type 4 channel was linear, with a conductance of 207 pS (Fig. 9). The channel current reversed at a pipette potential of −30 mV (a membrane potential of 0 mV), Much was near to E_K. The channel exhibited a subconductance state which was three-quarters of the amplitude of the full conductance level. Estimates of E_1/2 and z are +22 mV (V_pip −62 mV) and −0.61, respectively. The type 4 channel exhibited little voltage-dependency, it was not Ca^2+-dependent and did not appear to be blocked by 20 mmol l⁻¹ Ba²⁺ present in the pipette saline.

In inside-out patches the type 4 channel had a maximum conductance of 140pS. Fig. 10A,B shows the dwell time PDFs for data obtained from an inside-out patch held at a pipette potential of −120 mV. The closed time PDF was best fitted by the sum of three exponential components. The open time PDF was best fitted by a single exponential. There was little correlation between successive closings, but there was significant correlation between successive openings (Fig. 10C,D).

This channel was observed only in cell-attached patches. The I–V relationship for the type 5 K⁺ channel was curvilinear (Fig. 11), with a maximal slope conductance of 5pS. This channel appeared to be an inward rectifier. Because of the poor signal-to-noise ratio, the reversal potential for the channel current could not be determined accurately.

The results presented here demonstrate that five types of K⁺ channel co-exist in cultured embryonic locust muscle. There appeared to be three inward rectifiers, the type 1, type 2 and type 5 channels, and two channels with large linear conductances, the type 3 and type 4 channels. The presence of inward rectifiers in this preparation was not surprising, since an inwardly rectifying current is present in adult locust muscle (Usherwood, 1969). The apparent lack of a delayed rectifier channel is surprising, especially since excitatory postsynaptic potentials and spontaneous contractions of the myofibres have been reported previously (Miller, 1988). However, this may reflect the developmental stage of the cultured embryonic muscle where synaptic transmission may either be absent or at an early stage in its development.

There are few similarities between the K⁺ currents in Drosophila muscle and those of the embryonic locust muscle. Both preparations have Ca^2+-sensitive K⁺ currents, but the Ca^2+-dependent currents of Drosophila (I_Kca and I_Acd) rapidly inactivate and are blocked by Ba²⁺ (Wu and Ganetzky, 1988), whereas the Ca^2+- sensitive type 3 channel of locust muscle does not inactivate and is not blocked by Ba²⁺. Both preparations have a large-conductance, stretch-sensitive channel, but the Drosophila K_ST channel is a Ca^2+-insensitive, voltage-independent channel (Zagotta et al. 1988) and the locust type 3 channel is Ca^2+-sensitive and voltagedependent. Rapidly activating and inactivating K⁺ channels, such as those responsible for the I_A current of Drosophila, are unlikely to have been seen with the experimental method employed. Patch pipette holding potentials were shifted manually, not by voltage-jumps, so any transient voltage-sensitive channel present would probably have inactivated during the time taken to alter the membrane potential. P_o for the types 1 and 3 K⁺ channels was voltage-dependent, whereas P_o for the type 4 channel appeared to be voltage-independent. It was not possible to measure P_o for channel types 2 and 5, but for both of these channels the open times appeared to shorten with hyperpolarisation. All the channels had a P_o above zero at the resting potential of the myofibre. The values for E_1/2 were as follows: type 1, +80mV; type 3, +80mV; and type 4, +22mV. Using a simple two-state model, the constant voltage-dependence for the type 1, 3 and 4 channels could be accounted for by a Boltzmann-type equilibrium function with values for z of −0.40, −0.33 and -0.61, respectively. Reported values of z range from 0.40 for Ca^2+-activated K⁺ channels in human red blood cells to 1.00 for the rabbit sarcoplasmic reticulum K⁺ channel (Moczydlowski, 1986). The types 2 and 3 channels appear to be driven by chemical modulators, the type 2 channel by ATP and the type 3 channel by Ca²⁺.

Dwell time analysis for the type 1, type 3 and type 4 channels revealed multiple closed and open states, with the autocorrelation functions having values greater than zero. This suggests that the gating of these channels may be best described by branched or cyclic models rather than by simple linear systems. No attempt has yet been made to fit models to these data.

Our results are not sufficient to allow precise physiological roles to be ascribed to each channel type. We can only speculate about their contribution to the total current record. The type 1 and type 2 inward rectifiers accounted for about 90 % of the channels observed. These channels may be involved in the long-term regulation of the resting membrane conductance. However, the type 5 inward rectifier channel was not as ubiquitous and was of such low conductance that it would be unlikely to contribute substantially to the resting conductance. In fact, it is difficult to postulate what role a channel such as the type 5 channel would have. The Ca^2+-sensitive type 3 channel may serve to repolarise or even hyperpolarise the cell during periods of raised internal free Ca²⁺ concentration. Perhaps depolarisation of the myofibre membrane leads to an increase in its Ca²⁺ permeability, as it does in adult locust skeletal muscle (Usherwood, 1969); such an occurrence would certainly activate type 3 channels. Additionally, the stretchsensitivity of the type 3 channel suggests that this channel may also be involved in regulating membrane excitability during contractions of the myofibre. A K⁺ channel with similar properties to the type 4 K⁺ channel has been observed in cell-attached patches on adult locust muscle (Huddie et al. 1986). In the same preparation, a large proportion of the membrane conductance at the resting potential can be accounted for by an ohmic K⁺ conductance, which is reduced by FMRFamide-like peptides and proctolin in a G-protein-dependent manner (Walther and Zittlau, 1989; Zittlau et al. 1989).

We are grateful to Dr C. F. Walther for his comments on the manuscript, and to Drs R. L. Ramsey and M. S. P. Sansom for the provision of computer programs. This work was funded by the British Science and Engineering Research Council.