Differences in the actions of some blockers of the calcium-activated potassium permeability in mammalian red cells

Abstract

1. The actions of some inhibitors of the Ca²⁺-activated K⁺ permeability in mammalian red cells have been compared.

2. Block of the permeability was assessed from the reduction in the net loss of K⁺ that followed the application of the Ca²⁺ ionophore A23187 (2 μM) to rabbit red cells suspended at a haematocrit of 1% in a low potassium solution ([K]₀ 0.12–0.17 mM) at 37°C. Net movement of K⁺ was measured using a K⁺-sensitive electrode placed in the suspension.

3. The concentrations (μM±s.d.) of the compounds tested causing 50% inhibition of K⁺ loss were: quinine, 37±3; cetiedil, 26±1; the cetiedil congeners UCL 1269, UCL 1274 and UCL 1495, ∼150, 8.2±0.1, 0.92±0.03 respectively; clotrimazole, 1.2±0.1; nitrendipine, 3.6±0.5 and charybdotoxin, 0.015±0.002.

4. The characteristics of the block suggested that compounds could be placed in two groups. For one set (quinine, cetiedil, and the UCL congeners), the concentration-inhibition curves were steeper (Hill coefficient, n_H, ⩾2.7) than for the other (clotrimazole, nitrendipine, charybdotoxin) for which n_H≈amp;1.

5. Compounds in the first set alone became less active on raising the concentration of K⁺ in the external solution to 5.4 mM.

6. The rate of K⁺ loss induced by A23187 slowed in the presence of high concentrations of cetiedil and its analogues, suggesting a use-dependent component to the inhibitory action. This was not seen with clotrimazole.

7. The blocking action of the cetiedil analogue UCL 1274 could not be overcome by an increase in external Ca²⁺ and its potency was unaltered when K⁺ loss was induced by the application of Pb²⁺ (10 μM) rather than by A23187.

8. These results, taken with the findings of others, suggest that agents that block the red cell Ca²⁺-activated K⁺ permeability can be placed in two groups with different mechanisms of action. The differences can be explained by supposing that clotrimazole and charybdotoxin act at the outer face of the channel whereas cetiedil and its congeners may block within it, either at or near the K⁺ binding site that determines the flow of K⁺.

Introduction

An increase in the cytosolic concentration of Ca²⁺ causes the erythrocytes of most mammalian species to become much more permeable to K⁺, as first shown by Gárdos (1958; see Sarkadi & Gárdos, 1985 for a review). This action is mediated by the opening of Ca²⁺-activated K⁺ channels (‘Gárdos channels') that are voltage-insensitive and have a single channel conductance of ∼15 pS (Grygorczyk & Schwarz, 1983; Christophersen, 1991; Leinders et al., 1992a,1992b; Dunn, 1998). Since this lies between that of the small conductance (SK_Ca) and large conductance (BK_Ca) varieties of Ca²⁺-activated K⁺ channel, they are regarded as belonging to an intermediate conductance (IK_Ca) subtype (for reviews see Cook & Quast, 1990; Haylett & Jenkinson, 1990). The recent cloning and sequencing by Ishii et al. (1997) of a human intermediate conductance Ca²⁺-activated K⁺ channel with the properties of the Gárdos channel has shown it to be structurally related to, but distinct from, the SK_Ca subtype.

This erythrocyte Ca²⁺-activated K⁺ permeability (P_K(Ca)) has been known for some years to be blocked by quinine and quinidine (Armando-Hardy et al., 1975; Reichstein & Rothstein, 1981), carbocyanine dyes (Simons, 1979), cetiedil (Berkowitz & Orringer, 1982, 1984; Christophersen & Vestgergaard-Bogind, 1985; Roxburgh et al., 1996) and charybdotoxin (Wolff et al., 1988; Brugnara et al., 1993a) though not by apamin (Burgess et al., 1981; for additional references see Schwarz & Passow, 1983; Sarkadi & Gárdos, 1985). More recently, nifedipine (Kaji, 1990), nitrendipine (Ellory et al., 1992) and clotrimazole (Alvarez et al., 1992; Brugnara et al., 1993b) have also been shown to be potent inhibitors. The present paper is concerned with the characteristics of the blockade caused by these diverse substances which are also of interest because of their potential value in the treatment of sickle cell disease (Benjamin et al., 1986; for reviews see Joiner, 1993; Bunn, 1997). Our findings suggest that quinine, cetiedil and three close analogues of cetiedil (UCL 1269, UCL 1274 and UCL 1495) act in a different way to clotrimazole, nitrendipine and charybdotoxin. Some of the results have been published in a preliminary form (Benton et al., 1994).

Methods

Preparation of red blood cells

Blood (2–5 ml) was withdrawn from the ear vein of adult New Zealand White rabbits and mixed with heparin (20 units ml⁻¹ whole blood). It was stored at 4°C for up to 2 h until used. The blood was centrifuged (3 min at 1600×g) and the supernatant and buffy coat aspirated and discarded. The packed cells were then resuspended in five volumes of a solution containing (in mM): NaCl  145, KCl 0.1, MgSO₄ 1, EDTA 1, TRIS 10, inosine 10. The pH was adjusted to 7.4 by adding 1 M NaOH. This suspension was centrifuged and the pellet re-suspended twice. After the final centrifugation, the cells were stored as a pellet in this solution at 4°C for up to 3 days.

Construction and use of K⁺sensitive electrodes

The application of the Ca²⁺-ionophore A23187 to a suspension of red cells at 37°C causes a net K⁺ loss which is attributable to the opening of Ca²⁺-activated K⁺ channels. The amount of K⁺ lost can be measured by means of a K⁺-sensitive electrode placed in the suspension. This provides a convenient if indirect measure of the increase in K⁺ permeability, and of the effect of drugs thereon (see e.g. Burgess et al., 1981; Cook & Haylett, 1985). The K⁺-sensitive electrodes were prepared using a minor modification of the method described by Hill et al. (1978). Valinomycin (4 mg), sodium tetraphenylborate (1 mg) and high molecular weight polyvinyl chloride (PVC, 25 mg) were dissolved in 1 ml tetrahydrofuran and 75 μl dibutyl sebacate. This solution was stored at −20°C. Membranes were formed by applying 5 μl of the solution to filter paper discs inserted into the end of a short length (∼2 cm) of PVC tubing (2.0 mm, i.d.). After the solvent had evaporated a further 5 μl was applied and the membranes were allowed to dry for at least 2 h. They could then be stored in a dessicator at 4°C for several weeks. When required for use the PVC tube carrying the membrane was filled with a solution containing NaCl (145 mM) and KCl (5.4 mM) prior to connection to a second closely fitting tube housing an Ag/AgCl bead. The reference electrode consisted of a PVC tube containing a similar bead and filled with the salt solution used to suspend the cells. The potential difference between the electrodes was measured by means of a high impedance amplifier (WPI F-223 A) and displayed on a potentiometric pen recorder (Kipp & Zonen BD 100). The performance of the electrode was checked before each experiment by applying standard concentrations of K⁺ covering the range to be used. The recording chamber comprised a water jacketed glass bath containing 2 ml of the red cell suspension. The suspension was stirred by a small magnetic ‘flea' driven by a rotating magnet placed below the chamber.

Measurement of K⁺ loss induced by the calcium ionophore A23187

Most of the experiments were performed using a low K⁺ bathing solution, for two main reasons. First, as is shown, cetiedil and its congeners become more effective when the external K⁺ is reduced so that it is then possible to use lower concentrations of the blocking agents, some of which were rather inactive and available in limited quantity. Second, small increases in the K⁺ content of the external solution resulting from loss of K⁺ from the cells are more easily detected by means of a K⁺-sensitive electrode when [K]₀ is initially low.

The standard low K⁺ bathing solution contained (mM): NaCl 145, KCl 0.1, MgSO₄ 1, CaCl₂ 1, inosine 10, TRIS 10. The pH was adjusted to 7.4 by addition of 1 M NaOH. 2 ml of this solution were placed in the recording chamber and 20 μl of packed red cells were added to give a final haematocrit of just under 1%. Because of the K⁺ content of the fluid adhering to the cells, their addition caused the concentration of K⁺ to rise to 0.12–0.17 mM, as indicated by the signal from the K⁺ sensitive electrode.

When the reading from the electrode had become steady, K⁺ loss from the cells via the Ca²⁺-activated K⁺ permeability was initiated by the addition of a supramaximal concentration of A23187 (2 μM, added as 4 μl of a 1 mM solution in DMSO) to the cell suspension. This caused a rapid release of K⁺. Three minutes after the application of A23187, digitonin (100 μM) was added to produce cell lysis and so allow the total available K⁺ to be estimated (Cook & Haylett, 1985). The magnitude of the K⁺ loss initiated by A23187 could then be calculated as the increase in [K]₀ 3 min after addition of A23187, expressed as a percentage of the total increase after addition of digitonin. This is equivalent to the quantity of K⁺ released by A23187 as a percentage of total K⁺ content of the cells.

The inhibitory effects of P_K(Ca)-blocking drugs were tested by adding a small volume (usually <5 μl) of a concentrated stock solution to the cell suspension for a preincubation period (usually 3 min) before applying A23187 to initiate K⁺ loss via Ca²⁺-activated K⁺-channels. The loss of K⁺ in the presence of the drug was then compared with that in its absence, so that the inhibition caused by the drug could be expressed as a percentage. Much longer preincubation periods (up to 2 h) were explored in some experiments. In these instances, packed red cells (20 μl) were added to a glass vial containing 2 ml of the standard low K⁺ solution containing the drug. The vial was gently shaken in a water bath at 37°C for the time required. Its contents were then transferred to the recording chamber prior to the application of A23817.

Because the K⁺ content of the incubation fluid was continuously monitored, the rate at which the cells lost K⁺ when treated with A23187 could also be determined. This was done by expressing the amount (ΔQ) of K⁺ lost during successive 20 s periods as a fraction of the K⁺ content (Q) of the cells midway in that period. Dividing this fraction by the time (Δt, normally 20 s) over which the loss occurred provided an estimate of the rate coefficient (k) for K⁺ loss:

Changes in this coefficient provided a second index of the action of drugs that block P_K(Ca). Both indices were used in the present work though most of the results presented are based on measurements of the amount of K⁺ lost after a standard 3 min application of A23187.

Activation of P_K(Ca) by Pb²⁺

The cells were prepared and handled as in the experiments with A23187. Lead II nitrate was added to the bath to give nominal Pb²⁺ concentration of 10 μM. The concentration of free Pb²⁺ is likely to have been lower because lead ions are extensively bound by red blood cell proteins (see e.g. Simons, 1993). Since the action of lead was slower than that of A23187, and had a greater latency, the K⁺ loss was measured after 5 min. It was noticed that lengthy exposure to lead had a deleterious effect on the K⁺ sensitive electrodes. This could be avoided by washing the bath and the electrodes with a solution containing 10 μM EDTA.

Preparation of buffered calcium solutions

Buffered calcium solutions were prepared in a solution containing (in mM) NaCl 145, KCl 0.1, MgSO₄ 1, inosine 10, HEPES 10, EGTA 5, pH 7.4. The desired Ca²⁺ concentration was attained by adding the appropriate volume of 1 M CaCl₂ and re-adjusting pH to 7.4 with 1 M NaOH. Free Ca²⁺ concentrations were calculated using the program REACT (written by G.L. Smith and kindly provided by him) based on the affinity constants listed by Smith & Miller (1985).

Data analysis

Values are given as the mean±s.e.mean except when the Hill equation was fitted to concentration-response data. This was done using a weighted least-squares minimization program (CVFIT) written by Professor D. Colquhoun (Department of Pharmacology, University College London). It provided estimates of the molar concentration (the IC₅₀) of the compounds that caused 50% inhibition of A23187-induced K⁺ loss, and of the Hill coefficient (n_H), ± an approximate standard deviation (see Colquhoun et al., 1974, for details). When comparing the effects of changes in external K⁺ on the potency of the compounds, the same program was used to estimate the factors (equivalent to the equi-effective molar ratios) by which the concentration of the compound had to be altered to achieve the same inhibition under the new experimental conditions.

Drugs and reagents

Charybdotoxin, nitrendipine, high molecular weight polyvinyl chloride (PVC) and cetiedil (the (±)-2-hexahydro-1H-azepin-1-yl) ethyl ester of 2-cyclohexyl-2-(3-thienyl) ethanoic acid) were generous gifts from Dr P.N. Strong, Bayer, ICI, and Innothéra (Arceuil, France), respectively. The compounds UCL 1274, UCL 1269 and UCL 1495 were synthesized in the Department of Chemistry at University College London by standard methods to be described elsewhere. UCL 1269 (‘hydroxy cetiedil') is the (±)-2-(hexahydro-1H-azepin-1-yl) ethyl ester of 2-cyclohexyl-2-(3-thienyl)-2-hydroxy ethanoic acid, UCL 1274 is the 2-(hexahydro-1H-azepin-1-yl) ethyl ester of 2,2,2-triphenylethanoic acid and UCL 1495 is the 2-[N-(5-ethyl-2-methyl-piperidino)] ethyl ester of 2,2,2-triphenylethanoic acid. Dibutyl sebacate, dimethyl sulphoxide (DMSO), inosine, valinomycin, clotrimazole, A23187 (calcimycin), EGTA (1,2-di(2-aminoethoxy)ethane-tetra-acetic acid), quinine and heparin were purchased from Sigma, TRIS (tris (hydroxymethyl)aminomethane) from Calbiochem and sodium tetraphenyl borate from Koch-Light. Standard salts were of Analar quality and were obtained from BDH, who also provided EDTA (ethylene diamine tetra-acetic acid), digitonin and tetrahydrofuran.

Results

Preliminary experiments showed that the blocking action of cetiedil and several of its congeners was strongly dependent on the time for which the red cells had been exposed to the compound. To study this, rabbit red cells were preincubated for different periods with a concentration of the compound that produced 50–80% inhibition of the P_(K(Ca)-mediated increase in K⁺ loss initiated by the Ca²⁺ ionophore A23187. It can be seen in Figure 1 that the action of cetiedil and hydroxy cetiedil (UCL 1269) developed only slowly so that a relatively long preincubation was necessary in order to obtain a true measure of their effectiveness. In contrast, the actions of nitrendipine (Figure 1, lower right) and clotrimazole were complete within 60 s, the shortest preincubation period practicable with the technique used.

Figure 1.

The dependence on preincubation time of the P_K(Ca) blocking action of nitrendipine, cetiedil and two analogues of cetiedil, UCL 1269 and UCL 1274. Each compound was incubated with a suspension of rabbit red blood cells for the periods indicated (abscissa) prior to adding the Ca²⁺-ionophore A23187 to the suspension in order to initiate K⁺ loss through Ca²⁺-activated K⁺ channels. The amount of K⁺ lost was measured using a K⁺-sensitive electrode placed in the suspension (see also Figure 4a). The total K⁺ loss 3 min after applying A23187 in the presence of the test compound was expressed as a percentage of the loss in its absence (ordinate). Each point is the mean of 3–6 observations and the error bars indicate the s.e.mean. For cetiedil, UCL 1269 and UCL 1274 the curves have been drawn by fitting the expression y=y_max (1−e^−kt), where y is percentage inhibition, k is a rate constant and t is time (see also Table 1). The onset of the action of nitrendipine was too rapid to be resolved by present technique and the broken line has been constructed using a value of k of 7 min⁻¹, to indicate a lower limit.

Though the factors that underlie the slow onset of action of the cetiedil series have not been studied in any detail, the onset was noted to be approximately exponential in time course, with a rate constant that increased with the activity of the compound. Table 1 lists the rate constants for cetiedil, UCL 1269 and UCL 1274 together with the concentrations causing half maximal inhibition (IC₅₀). Because the potency of these substances is strongly correlated with their lipophilicity (Benton et al., 1994) a measure of lipophilicity (Σf) is also given. This was estimated from the hydrophobic fragmental constants (f) introduced by Nys & Rekker (1973) and refined by Rekker & de Kort (1979). It is clear that the most potent (and lipophilic) of the three compounds, UCL 1274, acts rapidly, at a rate such that a 3-min incubation period was adequate. This has also been seen in subsequent work with more active analogues of cetiedil and UCL 1274, including UCL 1495 (D.C.H. Benton, M. Malik, C.R. Ganellin and C.J. Roxburgh, unpublished observations).

Table 1.

Values of the rate constants (k) obtained on fitting the expression y=y_max (1−e^−kt) to the results of Figure 1

The concentration-dependence of the blocking action

The next aim was to study the concentration dependence of the blocking action. Figure 2 presents inhibition-concentration curves for six inhibitors of the Ca²⁺-activated K⁺-permeability of red cells. The curves are clearly steeper for four of the compounds (quinine, cetiedil, UCL 1274 and UCL 1495) than for clotrimazole and nitrendipine. Table 2 lists the parameters obtained on fitting the Hill equation to these data and to the results of similar experiments with charybdotoxin. The values for the Hill coefficient, n_H, suggest that the compounds can be placed in two categories. For one set (quinine, cetiedil, UCL 1274 and UCL 1495) the value of n_H is considerably greater than for the second (nitrendipine, clotrimazole and charybdotoxin).

Figure 2.

The concentration dependence of the inhibitory actions of clotrimazole, nitrendipine quinine, cetiedil, UCL 1274 and UCL 1495 on A23187-induced K⁺ loss from rabbit red cells bathed in a low K⁺ solution. The preincubation times were 60 min for cetiedil, UCL 1274 and UCL 1495, and 3 min for quinine, clotrimazole and nitrendipine. Each point is the mean of 3–8 observations and the error bars show the s.e.means. The lines have been drawn according to the Hill equation, using the parameter values listed in Table 2.

Table 2.

Hill slopes (n_H) and IC₅₀ values for fits of the Hill equation to the data shown in Figure 2

Two limitations of these findings deserve mention. First, a satisfactory concentration-response relationship could not be obtained for the cetiedil analogue UCL 1269, probably because of its relative inactivity and slow onset of action. The prolonged incubation with high concentrations needed to establish the upper region of the curve is likely to have adversely affected the cells. Second, and more important, a puzzling and as yet unexplained feature of the present observations was that although individual experiments, and sets of experiments on successive days, gave reproducible results, the IC₅₀ values for the compounds were found to vary, by as much as 3 fold on occasion, when determined at longer intervals. The variation was seen even with cells from the same animal. This could reflect differences in the state of the cells which may in turn influence the complex equilibria known to exist between closed and open states of the Gárdos channel (see e.g. Leinders et al., 1992a). Further work would be needed to identify the underlying cause. In view of this uncertainty, and bearing in mind that the measures of channel block are rather indirect, the IC₅₀ values listed in Tables 1 and 2 clearly can not be regarded as equivalent to the dissociation equilibrium constants for the combination of the blocking compounds with their sites of action.

The influence of external K⁺

A study of the K⁺-dependence of the blocking action confirmed a second difference between the two groups of compounds. The work of Armando-Hardy et al. (1975) and of Reichstein & Rothstein (1981) has shown that the inhibitory action of quinine on the Ca²⁺-activated K⁺ permeability of red cells is reduced when the concentration of K⁺ in the external solution is increased. In contrast, the blocking action of nitrendipine is little affected by large changes in external K⁺ (Ellory et al., 1992). Figure 3 shows the consequences of raising [K]₀ from between 0.12 and 0.17 mM to 5.4 mM on the blocking action of cetiedil, UCL 1495, charybdotoxin, nitrendipine and clotrimazole. It is evident that cetiedil and UCL 1495 resemble quinine in becoming less effective when [K]₀ is increased. In contrast, the inhibitory actions of nitrendipine, charybdotoxin and clotrimazole are not significantly affected (see also Table 3).

Figure 3.

The effect of increasing the external concentration of K⁺ on the blocking action of (a) cetiedil, (b) UCL 1495, (c) charybdotoxin and nitrendipine, (d) clotrimazole. Each point is the mean of 3–8 observations and the error bars show the s.e.mean. The cells were incubated with the blocking agents for 1 h before the application of A23187 to initiate K⁺ loss, except for nitrendipine and clotrimazole which were applied 3 min beforehand.

Table 3.

The factor by which the concentration of each of the compounds listed had to be increased in order to produce the same inhibition of P_K(Ca) when the concentration of K⁺ in the external fluid was raised from 0.12–0.17 to 5.4  mM (see also Figure 3).

Differences in the time course of A23187-stimulated K⁺loss in the presence of cetiedil and clotrimazole

Inspection of the time course of the K⁺ loss induced by A23187 in the presence of the blocking agents suggested a third difference in the actions of the two groups of compounds identified above. Figure 4a shows sample records and the combined results from several experiments to compare cetiedil and clotrimazole have been plotted in Figure 4b. It can be seen that although clotrimazole at 3 μM greatly reduced the rate of loss of K⁺, this loss continued throughout the presence of the drug. Similar results were obtained with nitrendipine and charybdotoxin. In contrast, with a high concentration of cetiedil or its congeners (e.g., UCL 1274, UCL 1495), the rate of loss rapidly fell (see Figure 4a, ii) so that by the end of the 3-min application of A23187 in the presence of cetiedil (50 μM), it had almost ceased.

Figure 4.

(a) Records showing the loss of K⁺ from a suspension of rabbit erythrocytes exposed to A23187 (2 μM) either alone (i) or in the presence of either 50 μM cetiedil (ii) or 3 μM clotrimazole (iii). A23187 was applied at the first of each pair of arrows, followed ∼3 min later (second arrow) by digitonin (100 μM) to cause lysis and so allow the total K⁺ content of the cells to be determined. Cetiedil and clotrimazole were added 3 min prior to A23187. Because the absolute value of the initial and final K⁺ concentration varied a little, the traces have been scaled to the same amplitude and accordingly the ordinate shows K⁺ loss as a percentage of the total K⁺ content, as measured following the addition of digitonin. Typically, the concentration of K⁺ in the suspension rose from ∼0.15 (before A23187) to 0.8 mM (after digitonin). The nonlinearity of the ordinate scale reflects mainly the logarithmic relationship between [K⁺] and the potential measured by the K⁺-sensitive electrode used to detect the net movement of K⁺ between the erythrocytes and the solution in which they were suspended. (b) Averaged results. The values are the means of six observations for the control and three for the test compounds. Vertical bars indicate the s.e.mean.

To quantitate this difference, the amount of K⁺ leaving the cells in successive 20-s intervals was measured from the results shown in Figure 4b. A rate coefficient, k, for K⁺ loss was then calculated as described in the Results section. The values obtained are shown in Figure 5a. The initial rate of loss is reduced almost equally (by ∼85%) by cetiedil and clotrimazole at concentrations of 50 and 3 μM respectively. However, with increasing time, the rate falls steeply in the presence of cetiedil (squares) but not clotrimazole (triangles). Figure 5b presents the ratios of the rate constants at 70 and 10 s. There is a clear difference in the patterns seen with increasing concentrations of clotrimazole and cetiedil.

Figure 5.

(a) The time course of changes in the rate coefficient (ordinate: note log scale) for A23187-induced K⁺ loss from erythrocytes in the absence (n=6) or presence of either 50 μM cetiedil (n=3) or 3 μM clotrimazole (n=3). Abscissa: time after the introduction of A23187. (b) The ratios of the rate coefficients at 70 and 10 s after applying A23187. Results are shown for clotrimazole (clot, μM concentrations) and cetiedil (cet, μM concentrations). This ratio becomes smaller as the concentration of cetiedil is increased. ^* denotes significance at the 5% level as indicated by an unpaired t-test.

An incidental observation was that the loss of K⁺ caused by A23187 in the absence of blocking agents did not follow simple first order kinetics, as evidenced by the decline in the ‘control' rate coefficients (Figure 5a, circles). This was not surprising because of the concomitant changes in membrane potential and concentration gradients that would be expected to occur: after 3 min exposure to A23187, the cells will have lost ∼70% of their K⁺ content and their volume will have fallen because of the accompanying loss of Cl⁻ and water. These changes can be expected to be much smaller in the presence of a large concentration of clotrimazole and accordingly the loss of K⁺ is then closer to first order, as indicated by the relatively small decline with time in the rate coefficient (Figure 5a, triangles).

The finding that cetiedil and clotrimazole differ in their effects on the time course of A23187-induced K⁺ loss has two implications. First, it suggests a difference in the mechanism of the blocking action. The inhibition caused by cetiedil but not clotrimazole appears to be use-dependent in the sense that it becomes greater in the continued presence of A23187. This would be expected were cetiedil to have some preferential affinity for open ion channels (see Discussion). Second, it is clear that the apparent magnitude of the block caused by the P_K(Ca) blockers will depend on the time at which the measurement is made. Figure 6 shows concentration response curves constructed using the reduction in the rate coefficients measured at 10 and 170 s after A23187 had been added to the cell suspension. At 170 s the IC₅₀'s estimated from Hill plots were 16±2 and 1.4±0.3 μM for cetiedil and clotrimazole respectively. When measured at 10 s the corresponding values were 20±2 for cetiedil and 0.47±0.06 μM for clotrimazole. Thus whereas the IC₅₀ for clotrimazole increased if the measurements were taken later in the response, that for cetiedil became a little lower if anything. An increase with time in the potency of cetiedil is consistent with use-dependent inhibition, as already discussed. The apparent decrease in the effectiveness of clotrimazole is likely to arise because, as already noted, when the K⁺ flux in the presence of the blocker is greatly reduced, the loss of K⁺ approximates more closely to first order kinetics, in contrast to the loss from the control cells (see also Figure 5a).

Figure 6.

The concentration dependence of the effect of clotrimazole and cetiedil on the rate coefficients for K⁺ loss measured at 10 s (filled symbols) and 170 s (open symbols) after the application of A23187. The values plotted are the coefficients in the presence of the blocker expressed as a percentage of the corresponding values in its absence. Vertical bars indicate the s.e.mean (n=3). The lines have been drawn according to the Hill equation (see text).

It is also worth noting that the slopes of the concentration-inhibition curves constructed in this way were consistently greater for cetiedil than for clotrimazole, in keeping with the results presented in Figure 2 and Table 2. Thus the estimates of the Hill slope, n_H, were 1.0±0.1 and 1.4±0.4 clotrimazole at 10 s and 170 s respectively, as compared with 1.9±0.2 and 2.2±0.2 for cetiedil at the same times.

Additional experiments

A common feature of the work so far described was the use of the Ca²⁺ ionophore A23187 to activate the P_K(Ca) mechanism. A possible if unlikely complication was that the blockers might have interfered with the action of the ionophore. To rule this out, Pb²⁺ rather than A23187 was employed in a few experiments. Pb²⁺ ions are known to enter erythrocytes via the anion exchanger (Simons, 1984) and to activate the Ca²⁺-dependent K⁺ channels by a direct effect not involving Ca²⁺ (Shields et al., 1985). In keeping with this, the addition of Pb²⁺ to rabbit erythrocytes suspended in the standard low K⁺ solution caused a loss of K⁺ comparable to, though a little slower than, that seen with A23187. In six such experiments, the mean K⁺ loss in response to a 5-min application of Pb²⁺ at 10 μM (a maximal concentration) was 53±1%, as compared with 58.4±5% (n=6) with the standard application of A23187 (2 μM, also maximal). As figure 7a shows, the cetiedil congener UCL 1274 was as effective in blocking K⁺ loss induced by Pb²⁺ as by A23187. The IC₅₀'s observed in this set of experiments were 5.4±0.4 μM (with Pb²⁺) and 5.3±0.4 μM (A23187).

Figure 7.

(a) Inhibition by UCL 1274 of K⁺ loss from rabbit erythrocytes exposed to either A23187 (2 μM) or Pb²⁺ (10 μM). Each point is the mean of 3–4 observations and the vertical bars indicate the s.e.mean. (b) The relationship between the concentration of external Ca²⁺ and the release of K⁺ in response to A23187 (2 μM) applied in the absence or presence of UCL 1274 (6 μM). The erythrocytes were suspended in solutions containing buffered Ca²⁺ at the concentrations shown on the abscissa. The lines have been drawn according to the Hill equation (E_max=69±3 and 39±2%, and n_H=1.9±0.3 and 2.6±1.1, in the absence and presence respectively of UCL 1274).

UCL 1274 was also used in a final set of experiments to determine whether its blocking action could be surmounted by an increase in Ca²⁺ concentration. A direct competitive interaction between the blocker and Ca²⁺ at its recognition site at the inner face of the K⁺ channel seemed unlikely for several reasons. For example, an intracellular site of action would not be in keeping with the rapidity of the effect of UCL 1274 (see Figure 1). Also, cetiedil, to which UCL 1274 is chemically closely related, is much less active when applied to the inner as compared with the outer face of the membrane (Dunn, 1998, and unpublished observations). To test the surmountability of the block, the effect of UCL 1274 on A23187-induced K⁺ loss was re-examined using external solutions containing different buffered concentrations of Ca²⁺. The results are shown in Figure 7b and as expected are in keeping with noncompetitive rather than simple reversible competitive block. Fitting the Hill equation to the data provided an estimate of 69±3% for the maximum K⁺ loss in the absence of the blocker and an EC₅₀ of 2.2±0.4 μM for the action of Ca²⁺, which is in keeping with previous values (see e.g. Simons, 1976). The Hill coefficient was 1.9±0.3 which is also consistent with earlier work. In the presence of 6 μM UCL 1274, the maximum K⁺ loss was reduced to 39±2%. This corresponds to 44% inhibition which is in keeping with the results presented in Figure 2. Similar findings for the action of the P_K(Ca) blocker nifedipine on human red cells have been reported by Kaji (1990).

Discussion

The main new finding is of differences in the characteristics of the block of the red cell Ca²⁺-activated K⁺ permeability by clotrimazole and by cetiedil and its congeners, as well as by quinine. Thus the inhibitory effect of cetiedil, but not clotrimazole, is increased by a reduction in external K⁺ (Figure 3); the concentration-inhibition curves for the two sets of compounds have very different slopes (Figures 2 and 6) and there appears to be a use-dependent component of the blocking action of cetiedil but not clotrimazole (Figures 4 and 5).

These differences can be discussed in relation to the properties of the Gárdos channels. It is known from single channel recordings (Grygorczyk & Schwarz, 1983, 1985) and from flux measurements (Vestergaard-Bogind et al., 1985) that they have the characteristics of long pores that can contain more than one K⁺ ion (probably 2–3) at the same time. Movement of K⁺ through such a multi-ion channel involves the successive binding and dissociation of individual K⁺ ions and this binding underlies the permeation characteristics of the channel (Hille & Schwartz, 1978; Hille, 1992). The present findings with cetiedil could be accounted for by supposing that it either acts at these sites or influences their properties. Thus, were it to compete with K⁺ for them, or alter their affinity for K⁺ through an allosteric effect, the observed K⁺ dependence of the blocking action could be explained. The steepness (n_H>2: see Table 2) of the concentration-inhibition curves observed with cetiedil and its congeners UCL 1274 and UCL 1495 is also in keeping with this since Hill slopes of greater than unity are expected for the action of blockers that act at the binding sites within multi-ion channels (see e.g. Adelman & French, 1978; Hille, 1992).

Blockade of the Gárdos channel by a carbocyanine dye (3,3′-diethylthiadicarbocyanine; diS-C₂-(5)) is similarly reduced by an increase in external K⁺, as shown by Simons (1979) who already suggested that displacement of the dye by K⁺ within the channel might be involved. Simons also noted that the dependence of the block on the extent to which the Ca²⁺-activated K⁺ permeability had been activated suggested that diS-C₂(5) binds either only or preferentially to open channels. It was found in the present work that in the presence of a relatively high concentration of cetiedil, the rate at which red blood cells lost K⁺ in response to indirect activation of the Gárdos channel by A23187 declined sharply with time though it was clearly already reduced at the earliest point at which measurements could be made (Figure 5). This can be explained by supposing that cetiedil, like the carbocyanine dyes, blocks open channels but also has some affinity for the shut form. A preference for open channels could be relevant to the possibility that potent blockers of the Gárdos channel may prove useful in the treatment of sickle cell disease. The potential value of clotrimazole in this condition is under study (Brugnara et al., 1996; see also Bunn, 1997).

The correlation between the potency of cetiedil and its analogues and their lipophilicity (Table 1; Benton et al., 1994) suggests that lipophilicity is an important factor in determining the concentration reached at the site of action. One possibility is that these compounds must first enter the cell, as suggested by Grygorczyk & Schwarz (1985) for the Gárdos channel blocking action of quinine (though see Reichstein & Rothstein, 1981, for a contrary view). This seems unlikely to hold for cetiedil in view of single channel recordings (Dunn, 1997, 1998) that show that cetiedil is only weakly active when applied to the cytoplasmic surface of inside-out patches isolated from human red cells. An action within the channel rather than at either its inner or outer surface seems more probable, particularly in view of our findings (i) that quaternization of cetiedil greatly reduces its activity (Benton et al., 1994), and (ii) that the action of cetiedil and of hydroxy cetiedil (UCL 1269), a less lipophilic derivative of cetiedil, increased with the period of preincubation with the cells (Figure 1). The observation (Roxburgh et al., 1996) that the enantiomers of cetiedil are equi-active as blockers of the Gárdos channel also suggests that the lipophilicity of the molecule is more important that its shape as a determinant of activity. However, in recent studies with more potent analogues of cetiedil, we have obtained evidence that structural features can also play a role (unpublished observations by D.C.H. Benton, M. Malik, C.R. Ganellin, S. Athmani, Z. Miscony & D.H. Jenkinson).

To summarize to this point, the evidence suggests that cetiedil and its congeners either act at or affect the K⁺ binding sites within the channel that determine its selectivity. Access is strongly influenced by the lipophilicity of the compound and does not occur from the inner surface of the membrane. Interestingly, the same conclusions have been reached for the Ca²⁺ channel blocking action of the dihydropyridines, as discussed by Hockerman et al. (1997).

Our findings with clotrimazole, taken in conjunction with the work of others (Brugnara et al., 1993a, 1995) suggest that it inhibits in a quite different way. The Gárdos channel is blocked by charybdotoxin (Wolff et al., 1988; Brugnara et al., 1993a) and the binding of labelled charybdotoxin is inhibited by clotrimazole (Brugnara et al., 1995). This suggests that both substances act at the outer region of the channel, rather than within it. A common mechanism of action of clotrimazole and charybdotoxin (and one distinct from that of cetiedil) is also suggested by the finding that the block is little affected by a reduction in external K⁺ (Figure 3). Nor does the block by clotrimazole increase with time in the manner seen with cetiedil (Figures 4 and 5). Finally, the similarity of the Hill coefficients observed for the inhibitory effects of clotrimazole, charybdotoxin and nitrendipine is also consistent with a common mechanism of action, whereas the coefficients for cetiedil, quinine and the cetiedil analogues UCL 1274 and UCL 1495 were consistently greater (Table 2).

Though the evidence strongly suggests that clotrimazole, like charybdotoxin, acts at the outer face of the channel, this may not be the only site involved. Recent single channel recordings by Dunn (1997, 1998) have shown that clotrimazole is a potent blocker of the Gárdos channel even when applied to the inner face of the red cell membrane. The contribution that this makes to the overall inhibition seen when clotrimazole is applied to intact cells remains to be determined.

In our work the IC₅₀ value for the inhibitory action of clotrimazole on Ca²⁺-activated K⁺ loss from rabbit erythrocytes was found to be 1.2±0.1 μM. Alvarez et al. (1992) and Brugnara et al. (1993b) using human cells have reported a much greater potency (IC₅₀≈amp;50 nM). Two factors seem likely to contribute to the discrepancy. First, the IC₅₀'s listed in Table 2 are based on measurements taken at the end of a 3-min exposure to A23187. This time point was chosen for consistency with our studies with cetiedil whose blocking action increases with time (Figure 4). On recalculating the potency of clotrimazole but now from the reduction in the rate of loss when A23187 was first applied, an IC₅₀ of 470 nM was obtained (Figure 6). Second, we have found that rabbit erythrocytes are less sensitive to clotrimazole than human cells, by a factor of ∼2 (M. Malik & D.H. Jenkinson, unpublished data). The same factors are likely to account for the somewhat lower potencies observed for charybdotoxin and nitrendipine in the present work.

To conclude, we have found differences in the effects of Gárdos channel blockers that are likely to reflect different modes of interaction with the channel. Cetiedil and compounds related to it seem likely to act within the channel, at or near the K⁺ binding sites that determine permeation, whereas the evidence so far available suggests that clotrimazole and charybdotoxin become attached to its outer surface.