Characterization of a Ca²⁺-activated K⁺ current in insulin-secreting murine βTC-3 cells

Abstract

1. The whole-cell perforated-patch recording mode was used to record a Ca²⁺-dependent K⁺ current (I_K(Ca)) in mouse βTC-3 insulin-secreting cells.

2. Depolarizing voltage steps (to potentials where Ca²⁺ currents are activated) evoked a slowly activating, outward current, which exhibited a slow deactivation (in seconds) upon subsequent hyperpolarization.

3. This current was shown to increase with progressively longer depolarizing voltage steps. It could be reversibly abolished by the removal of Ca²⁺ from the external medium or by application of Ca²⁺ channel blockers, such as Cd²⁺ and nifedipine. It was concluded that the depolarization-evoked current was activated by Ca²⁺.

4. Variations in external K⁺ concentration led to shifts in the reversal potential of the Ca²⁺-dependent current as predicted by the Nernst equation for a K⁺-selective current.

5. The Ca²⁺-activated K⁺ current was insensitive to external TEA (10 mM), a concentration sufficient to block the large-conductance Ca²⁺-dependent (maxi-K_Ca) channel in β-cells. It was also insensitive to apamin, tubocurarine and scyllatoxin (leiurotoxin I), specific blockers of small-conductance K_Ca channels.

6. The current was blocked by quinine, a non-specific K_Ca channel blocker and, surprisingly, by charybdotoxin (ChTX; 100 nM) but not iberiotoxin, a charybdotoxin analogue, which blocks the maxi-K_Ca channel. It was sensitive to block by clotrimazole and could be potently and reversibly potentiated by micromolar concentrations of niflumic acid. Thus, the current exhibited unique pharmacological characteristics, not conforming to the known K_Ca channel classes.

7. The ChTX-sensitive K_Ca channel was permeable to Tl⁺, K⁺, Rb⁺ and NH₄⁺ but not Cs⁺ ions.

8. The ChTX-sensitive I_K(Ca) could be activated by the muscarinic agonists in the presence or absence of external Ca²⁺, presumably by releasing Ca²⁺ from internal stores.

9. Acutely isolated porcine islet cells also exhibited a slow I_K(Ca) resembling that described in βTC-3 cells in kinetic properties, insensitivity to TEA (5 mM) and sensitivity to quinidine, an analogue of quinine. The porcine I_K(Ca), however, was not sensitive to block by 100-200 nM ChTX. It is likely, that species differences account for pharmacological differences between the mouse and porcine slow I_K(Ca).

Calcium-dependent K⁺ (K_Ca) channels have been studied widely in various tissues, including endocrine cells (for review, see Latorre, Oberhauser, Labarca & Alvarez, 1989). They can be classified depending on their sensitivity to the toxins charybdotoxin (ChTX, from scorpion) and apamin (from bee venom) (Blatz & Magleby 1987; Marty, 1989). The ChTX-sensitive channels (maxi-K_Ca channels) have been the most studied, due to their ubiquity and large conductance (up to 200 pS). Their properties vary from tissue to tissue, but in most cases they are characterized by fast opening rates (in the order of several milliseconds) following an increase in Ca²⁺ concentration. In most tissues they have been demonstrated to possess a moderate voltage dependence (McManus, 1991) and to mediate the repolarization of the action potential, as in a neuroendocrine cell line (Lang & Ritchie, 1990). The apamin-sensitive channels have been studied in less detail. They generally activate and deactivate at slower rates, have higher sensitivity to Ca²⁺ and have been implicated in the slow after-hyperpolarization (Goh & Pennefather, 1987) and the generation of the after-spike hyperpolarization at high frequencies of firing (Lang & Ritchie, 1990).

Ca²⁺-dependent K⁺ currents (I_K(Ca)) can be activated by Ca²⁺ influx through voltage-dependent and -independent Ca²⁺-selective channels. They can also be activated by Ca²⁺ release from internal stores, for example IP₃-sensitive stores. This is the case in lacrimal gland cells where acetylcholine acts to release Ca²⁺ from the endoplasmic reticulum (Petersen & Maruyama, 1984). In pituitary cells and neurones, hormones have been shown to activate the Ca²⁺-dependent K⁺ channel (e.g. Ritchie, 1987; Fagni, Bossu & Bockaert, 1991).

In pancreatic β-cells, the first indication of the existence of K_Ca channels was provided by Atwater and colleagues (Atwater & Beigelman, 1976; Atwater, Dawson, Ribalet & Rojas, 1979). They demonstrated that a quinine-sensitive K⁺ conductance underlay the interburst silent phases in mouse islets. Similarly quinine could reverse, in part, the hyperpolarization induced by mitochondrial uncouplers such as DNP (dinitrophenol) and CCCP (carbonyl cyanide m-chlorophenyl hydrazone), which increase [Ca²⁺]_i and reduce internal ATP. Furthermore, addition of calcium ionophores or raising the external Ca²⁺ concentration was shown to induce a quinine-blockable K⁺ conductance (Ribalet & Beigelman, 1980; Atwater, Rosario & Rojas, 1983).

Using the patch-clamp technique, it was directly demonstrated that rodent β-cells possess maxi-K_Ca channels (Cook, Ikeuchi & Fujimoto, 1984; Tabcharani & Misler, 1989). This channel was blocked by TEA (2 mM) and ChTX (10 nM) but not by apamin (Lebrun, Atwater, Claret, Malaisse & Herschuelz, 1983; Tabcharani & Misler, 1989). However, the maxi-K_Ca channel was not found to play a role in the burst activity of the β-cell (Kukuljan, Goncalves & Atwater, 1991).

Rorsman's group has suggested that there is an apamin-insensitive channel in mouse β-cells, which is unlike the reported maxi-K_Ca channel (Ämmälä, Bokvist, Larsson, Berggren & Rorsman, 1993). However, this channel has not been characterized further and it remains unknown whether it represents the Ca²⁺-dependent K⁺ conductance underlying the bursting.

A Ca²⁺-dependent K⁺ conductance blocked by quinine but not TEA has been suggested to underlie the adrenergic modulation of β-cell activity (Santana de Sa, Ferrer, Rojas & Atwater, 1983). Recently Bordin, Boschero, Carneiro & Atwater (1995) showed that cholinergic activation of islet cells also results in an increase in the ChTX-sensitive K_Ca conductance.

In summary, these studies provide evidence for the existence of a quinine-sensitive K_Ca conductance, distinct from the maxi-K_Ca current and sensitive to activation by hormonal stimulation, which may be important in the silent phase of β-cell electrical activity. In this study we have characterized a slow Ca²⁺-dependent K⁺ current (I_K(Ca)) in insulin-secreting (βTC-3) cells which fits these characteristics and is reminiscent of an apamin-insensitive Ca²⁺-dependent K⁺ current described previously (Ämmäläet al. 1991). It is distinct from the fast, TEA-sensitive maxi-K current, and although it can also be blocked by ChTX it requires higher concentrations than the maxi-K_Ca current. We show that muscarinic agonists activate this current in the absence of external Ca²⁺, in agreement with the results of a previous report (Bordin et al. 1995). A similar current could be found in freshly dissociated porcine β-cells. A preliminary report has appeared in abstract form (Kozak & Logothetis, 1996).

METHODS

Cell culture

βTC-3 cells were kindly provided by Drs N. Fleischer and S. Efrat (Albert Einstein School of Medicine, Bronx, NY, USA). The cells (passages 33-50) were grown in 10 ml plastic dishes (Beckton-Dickinson) containing DMEM (Gibco) supplemented with 10 % fetal bovine serum (FBS) and 12.5 mM glucose, 100 U ml⁻¹ penicillin and 0.1 mg ml⁻¹ streptomycin (Gibco or Sigma). The cells were fed twice a week, passaged using 0.05 % trypsin-0.53 mM EDTA, and kept in an incubator with humidified 5 % CO₂-95 % air at 37°C. For recording purposes cells were plated at low density on acid-washed glass coverslips and grown for at least 1 day before recordings were performed. No difference in expression levels or properties of the currents were seen when cells were grown in the presence of 5 mM glucose.

Islets of Langerhans from adult pigs were isolated and purified by collagenase digestion and Ficoll-gradient separation techniques. They were generously provided by the Islet Transplantation Laboratory of Washington University (St Louis, MO, USA). The islets were dispersed into single cells according to a protocol modified from Lernmark (1974). Briefly, islets were incubated in trypsin-EDTA twice with a wash between incubations and triturated with a pipetter using 200 μl siliconized microtips. Subsequently the isolated cells were plated at various densities in 3 ml culture dishes containing sterile glass coverslips. The islet cells would attach to the coverslips within 2 h and were used to obtain recordings for up to two weeks. The cells were maintained in a 37°C incubator in CMRL medium (Sigma) with 10 % FBS, 11.1 mM glucose and penicillin-streptomycin.

Solutions

In most experiments a standard internal solution was used with K⁺ as the main cation. The composition of the solution was (mM): 50 K₂SO₄, 60 KCl, 1 MgCl₂, 10 Hepes, pH 7.2. In some experiments the internal solution also contained 10 mM KF substituted for KCl. Another intracellular solution had the following composition (mM): 25 KCl, 125 potassium gluconate, 1 MgCl₂; 10 Hepes, pH 7.2. The properties of the currents were indistinguishable with both solutions. Several different external solutions were used. The 5 mM K⁺ solution contained (mM): 5 KCl, 10 TEA, 2 CaCl₂, 140 NaCl, 10 Hepes. The 30 mM K⁺ solution contained (mM): 30 KCl, 10 TEA, 2 CaCl₂, 115 N-methyl-D-glucamine (NMDG), 10 Hepes. The 100 mM K⁺ solution contained (mM): 100 KCl, 10 TEA, 2 CaCl₂, 45 NMDG, 10 Hepes. The 30 mM Tl⁺ solution contained (mM): 30 TlNO₃, 125 NaNO₃, 2 CaCl₂, 10 Hepes. In solutions containing Rb⁺, Cs⁺ and NH₄⁺, 30 mM KCl was replaced by an equimolar concentration of the chloride salt of the given cation. In all solutions the pH was adjusted to 7.2 with HCl.

All salts were obtained from Sigma and Acros Organics (Pittsburgh, PA, USA).

Perforated-patch recording and separation of currents

Recordings were performed using the perforated-patch whole cell mode using antibiotics amphotericin B or nystatin as described previously (Kozak & Logothetis, 1997). Briefly, amphotericin B/nystatin (Sigma) stock was prepared in DMSO (6 mg ml⁻¹) and stored at -20°C up to 1 week. Before the experiment, 10 μl stock solution was dissolved in 2.5 ml of the internal solution, yielding a final antibiotic concentration of 240 mg ml⁻¹.

Micropipettes were made of borosilicate glass (WPI) using a programmable puller (Sutter). The pipette resistances ranged from 1 to 3 MΩ when filled with the recording solution. The bath was earthed through a 2 M KCl-agar bridge. The recording chamber was continuously perfused with a gravity-driven perfusion system. The access resistance drop was monitored during the perforation process and recording was started after values below 15 MΩ were reached. Typical access resistances were 4-15 MΩ and would be stable for up to 1.5 h. Series resistance compensation provided by the PULSE software (Heka Electronik, Lambrecht, Germany) was used for access resistances above 10 MΩ.

Ionic currents were recorded using the EPC-9 patch-clamp amplifier and the PULSE/PULSEFIT (v. 7.6 and v. 8.0) data acquisition software (Heka). Data were stored on the hard disk of a Macintosh Quadra computer or on Syquest cartridges for further analysis using the Igor Pro software (WaveMetrics, Lake Oswego, OR, USA). The sampling rate was 1 kHz for most recordings. At this sampling rate the contamination from inward Na⁺ currents was minimal, since such currents inactivate with a time constant (τ) of 2 ms in β-cells (Plant, 1988). The protocols used to elicit I_K(Ca) were spaced at 10 s or longer intervals to avoid Ca²⁺ overloading artefacts. All described experiments were performed at room temperature (22-26°C).

Possible offsets during the recording were estimated in the following way. Under the same recording conditions, cells were chosen which displayed a significant inwardly rectifying K⁺ current. This current was sensitive to Ba²⁺ and Cs⁺ but insensitive to 100 μM Cd²⁺ (a concentration sufficient to block the Ca²⁺-dependent conductance). The reversal potential of the inward rectifying current was assessed at various external potassium concentrations by applying Ba²⁺ or Cs⁺. The reversal potential was close to the calculated K⁺ equilibrium potential. In some cells the reversal potential was shifted to more positive potentials by up to +10 mV. Since this shift did not occur in all the cells tested, we did not correct our results to account for this offset.

The βTC-3 cell line is a non-clonal cell population. Over the course of this study, a great variability was observed in the degree to which I_K(Ca) was encountered, depending on the duration of cell growth after splitting. No strict correlation could be deduced, however.

In pig islet cells, I_K(Ca) amplitude was substantially lower than in βTC-3 cells. In many cells, therefore, the amplitude of the recorded I_K(Ca) was increased by increasing [Ca²⁺]_o from 2 to 4 mM. This increase in Ca²⁺ concentration did not have noticeable effects on the delayed rectifier and A-type K⁺ current in porcine cells. The current could be detected in the porcine cells a week or more after plating. The lability of I_K(Ca) correlated with that in βTC-3 cells.

RESULTS

Depolarizing voltage steps activate an outward current dependent on extracellular Ca²⁺

In the presence of 10 mM external TEA, depolarizing steps from a holding potential of -60 mV elicited slow outward currents (Fig. 1A, left panel). Upon repolarization to -60 mV, slow outward tail currents were observed. We recorded the tail currents at -60 mV to distinguish the tails from those of the Ca²⁺-dependent Cl⁻ current (I_Cl(Ca)), which can also be found in βTC-3 cells (Kozak & Logothetis, 1997) and at this potential would be inward. Addition of 100 μM Cd²⁺ to the bath (Fig. 1A, middle panel) abolished both the onset and the tail current in a reversible manner (Fig. 1A, right panel). Removal of external calcium or its replacement with Mg²⁺ had the same effect (data not shown). Current-voltage (I-V) relationships of the peak onset (Fig. 1B) and tail (Fig. 1C) currents are shown. The I-V curves are bell shaped and reminiscent of the Ca²⁺ current I-V found in these cells (Kozak & Logothetis, 1997). The correlation of the I-V shape of this current with I_Ca and the dependence on Ca²⁺ influx indicate that this outward current is activated by Ca²⁺.

Figure 1. Depolarizing voltage steps activate an outward current which requires influx of external Ca²⁺.

Left panel in A, 800 ms depolarizing voltage steps were applied from a holding potential of -60 mV and the evoked outward currents are shown in normal external solution (2 mM Ca²⁺). Middle panel in A, application of 100 μM Cd²⁺ in the presence of 2 mM Ca²⁺ abolished the slowly activating outward currents, as well as the slow tail currents. The right panel in A shows the reversal of Cd²⁺ block after washout. B and C, the I-V plots of ‘onset’ (B) and tail (C) currents generated from the traces shown in A. Both ‘onset’ and tail current amplitudes are plotted against the test potential.

The Ca²⁺-activated outward current increases with longer depolarizing steps

If the depolarization-activated ‘onset’ current and the tail current are activated by Ca²⁺ elevation, one would expect to see larger outward currents when increasingly longer depolarizing steps are applied by providing more Ca²⁺ influx. To test this we used the ‘envelope’ protocol shown in Fig. 2. Progressively longer steps to +10 mV were applied and the voltage was returned to -50 mV after each step to record a tail current. The +10 mV step value was chosen to maximize I_Ca and the -50 mV repolarization value was chosen to increase the outward tail current size. We showed previously (see Kozak & Logothetis, 1997) that no significant inactivation of I_Ca occurs during voltage steps up to 2 s long, and therefore in such an envelope experiment Ca²⁺ influx continues during all steps. Longer voltage steps elicit progressively larger outward ‘onset’ and tail currents whose peak amplitudes are correlated linearly (not shown). Such linear correlation strongly suggests that both ‘onset’ and tail currents are carried by the same ion channel species. The inset shows that in the presence of Cd²⁺ the Ca²⁺-activated current is abolished.

Figure 2. Progressively longer depolarizing voltage steps activate larger Ca²⁺-activated outward currents.

Voltage steps of increasing durations to +10 mV (‘envelope protocol’) evoke corresponding larger outward currents followed by larger tail currents. Inset: the same voltage protocol applied in the presence of 100 μM Cd²⁺. The outward currents are completely abolished.

Sr²⁺ but not Ba²⁺ can substitute for Ca²⁺ in activating the Ca²⁺-dependent outward current

We have shown previously, in agreement with other studies, that the L-like calcium channels in βTC-3 cells are permeable to both Ba²⁺ and Sr²⁺ (Kozak & Logothetis, 1997). We tested whether any of these permeant divalent ions would mimic Ca²⁺ in activating the outward current by replacing the external 2 mM Ca²⁺ with equimolar Sr²⁺ and Ba²⁺. As seen in Fig. 3A a depolarizing step in external Sr²⁺ activated a sizeable outward current (Fig. 3A) but failed to do so in external Ba²⁺ (data not shown). Increasing external Ba²⁺ to 20 mM remained ineffective (Fig. 3B). As with the I_Cl(Ca) (Kozak & Logothetis, 1997), Sr²⁺ but not Ba²⁺ could substitute for Ca²⁺ in activating the outward current. Sr²⁺ seems to be a more potent activator for I_K(Ca) than for I_Cl(Ca), which under similar conditions showed a relatively reduced amplitude (Kozak & Logothetis, 1997).

Figure 3. The effects of replacing external Ca²⁺ with Sr²⁺ or Ba²⁺ on the Ca²⁺-activated current.

A, outward currents were evoked by steps to +10 mV in the presence of external solutions containing either 2 mM Ca²⁺ or 2 mM Sr²⁺. B, outward currents were not supported by a solution containing 20 mM Ba²⁺; although a clear inward current carried through Ca²⁺ channels was seen.

The slow Ca²⁺-activated current is selective for K⁺

In order to determine the nature of the outward Ca²⁺-dependent current we proceeded to measure its reversal potential as a function of the external K⁺ concentration (Reale, Hales & Ashford, 1994). As shown in Fig. 4A, the membrane potential was stepped to a depolarizing voltage (+ 5 mV) followed by a series of steps to various hyperpolarized potentials (-110 to -40 mV). The same protocol was repeated in the presence of 100 μM Cd²⁺ and traces were subtracted from the Cd²⁺ control traces. The peak of the tail current was measured 15 ms following the repolarization step to avoid contamination by Ca²⁺ current tails and plotted against the tail voltage value (Fig. 4B). The reversal potential of the current in 5 mM external K⁺ as determined from the I-V plot was -80 mV (n= 5). At an external K⁺ concentration of 30 mM (Fig. 4A, right), as compared with 5 mM, there was a positive shift of the reversal potential of the Ca²⁺-dependent current (+ 45 mV), as predicted by the Nernst equation for a K⁺-selective channel.

Figure 4. The Ca²⁺-activated current reversal potential value is sensitive to changes in external K⁺ concentration; the ionic selectivity sequence.

A, tail currents were evoked at different membrane potentials following a depolarizing step to +5 mV. External K⁺ was 5 mM (left) and 30 mM (right). B, the peak tail current amplitude is plotted against the tail voltage value to generate a tail I-V relation, from which the reversal potential values can be obtained in the three K⁺ external concentrations (5, 30 and 100 mM). C, the ionic selectivity sequence of the slow I_K(Ca) derived from the reversal potential measurements. The reversal potential of the Cd²⁺-sensitive current was measured in an external solution containing 30 mM K⁺ and compared with that in solutions containing 30 mM of the monovalent cation in question (Tl⁺ or Rb⁺). The individual points are fitted with a 3rd degree polynomial.

In the experiments in which external K⁺ was increased we either reduced the Na⁺ concentration or used a high-K⁺ solution which contained NMDG instead of Na⁺. No differences in the reversal potential values were found when Na⁺ concentrations were altered (n= 3). The reversal potential did not vary when the internal solution contained gluconate instead of sulphate as the anion (see Methods). Based on these results we concluded that this current constitutes a Ca²⁺-dependent K⁺ current.

The selectivity sequence of permeant ions through I_K(Ca)

In order to determine the ionic selectivity sequence of this current, we replaced the external 30 mM K⁺ with an equimolar concentration of another cation. The reversal potential was then determined in various external solutions and compared with that in 30 mM K⁺ solution. As seen in Fig. 4C, the channel was permeable to K⁺, Rb⁺ and Tl⁺, but not to Cs⁺ (data not shown). The selectivity sequence was determined using a modified Nernst equation (Hille, 1992) for biionic conditions, which gave the following permeability sequence: Tl⁺≈ Rb⁺ > K⁺ > NH₄⁺. Na⁺ was not measurably permeant. The reversal potential values for various permeant ions were not affected by substitution of external Na⁺ with NMDG.

The blockers of I_K(Ca)

Rat pancreatic β-cells have been shown previously to contain high-conductance Ca²⁺-dependent K⁺ (maxi-K_Ca) channels, which are sensitive to external TEA block (Tabcharani & Misler, 1989). We also observed TEA block of the maxi-K_Ca channels in βTC-3 cells (data not shown) and have used 10 mM TEA in our external solutions in order to block these channels. The slow time course of activation and deactivation of our I_K(Ca) suggested an SK-like current described extensively in other secretory cells (Park, 1994). This current has been shown to be sensitive to block by the bee venom apamin and tubocurarine. Apamin (200 nM) and D-tubocurarine (2 μM) had no effect on I_K(Ca) in βTC-3 cells (n= 4). Scyllatoxin (leiurotoxin I) (Castle & Strong, 1986; Castle, Haylett & Jenkinson, 1989), which is known to block the apamin-sensitive channel, was also ineffective in blocking this current (up to 200 nM). The insensitivity of our I_K(Ca) to the above-mentioned drugs is consistent with previous studies in islets and single β-cells (Lebrun et al. 1983; Henquin, Garrino, Nenquin, Paolisso & Hermans, 1985; Ämmäläet al. 1993), which demonstrate the lack of an apamin-sensitive K⁺ conductance in these cells.

There are reports of charybdotoxin blocking the small-conductance Ca²⁺-activated K⁺ current in various tissues (reviewed by Latorre et al. 1989). As shown in Fig. 5A, increasing concentrations of charybdotoxin blocked I_K(Ca). The highest concentration tested (100 nM) blocked, on average, 80 % of the current. Figure 5B shows the effect of charybdotoxin on the I-V curve for I_K(Ca). The block was reversible and not voltage dependent.

Figure 5. The slow Ca²⁺-activated K⁺ current in βTC-3 cells is sensitive to block by charybdotoxin.

A, the TEA-insensitive Ca²⁺-activated K⁺ current was blocked by increasing concentrations of charybdotoxin (10 and 100 nM). B, the effects of 100 nM ChTX on the peak outward current I-V. The inset shows a current trace evoked by a step to 0 mV (1) and block by 100 nM ChTX (2).

Iberiotoxin, another scorpion toxin, is closely related to charybdotoxin and selectively blocks the maxi-K_Ca channel. At 100 nM, iberiotoxin did not affect I_K(Ca) (n= 3). At 100 nM, kaliotoxin, which is also known to block the maxi-K_Ca channel (Crest et al. 1992) did not affect the TEA-insensitive I_K(Ca) in βTC-3 cells (n= 4).

It has been shown previously that quinine blocks a Ca²⁺-dependent K⁺ conductance in β-cells (Atwater et al. 1979; Ashcroft & Rorsman, 1991). In our hands, 100 μM quinine abolished I_K(Ca) completely (n= 4). However, quinine is known to inhibit other K⁺ conductances in β-cells such as the inward rectifier and the delayed rectifier, and this was also seen in βTC-3 cells (J. A. Kozak and D. E. Logothetis, unpublished observations).

Cs⁺ ions are known to block various K⁺ channels, such as intermediate-conductance Ca²⁺-activated channels (Schwarz & Passow, 1983). In βTC-3 cells, external Cs⁺ ions (1-2 mM) although not permeant were ineffective in blocking I_K(Ca).

Clotrimazole, an inhibitor of cytochrome P450-dependent enzymes, was recently shown to be a blocker of erythrocyte K_Ca and maxi-K_Ca channels in smooth muscle (Rittenhouse, Parker, Brugnara, Morgan & Alper, 1997). At a concentration of 1 μM, it reversibly blocked I_K(Ca) in βTC-3 cells. The extent of block at this concentration was estimated to be > 60 %.

Niflumic acid is a potent activator of I_K(Ca)

We have shown previously that niflumic acid reversibly blocks the Ca²⁺-dependent Cl⁻ current in βTC-3 cells at micromolar concentrations (Kozak & Logothetis, 1997). Initially, we tried to separate I_K(Ca) from I_Cl(Ca) in cells containing both currents, by adding niflumic acid to the bath. Unexpectedly we discovered that 10-20 μM niflumic acid could activate I_K(Ca) in addition to blocking the Cl⁻ current. Figure 6A shows the effect of 10 μM niflumic acid on a cell exhibiting I_K(Ca) which was activated by a depolarizing step to +10 mV. Niflumic acid markedly potentiated both the ‘onset’ and the tail currents, in a fully reversible manner and its effect did not desensitize throughout its application. The effect of niflumic acid on the I-V is shown in Fig. 6B. There was no apparent voltage dependence of potentiation of the current at voltages where the current could be activated by Ca²⁺ influx through the L-like channels. Niflumic acid failed to activate I_K(Ca) when the external solution contained 100 μM Cd²⁺.

Figure 6. Micromolar concentrations of niflumic acid potentiate the slow Ca²⁺-activated K⁺ current.

A, 10 μM niflumic acid increases the depolarization-evoked ‘onset’ and the repolarization-accompanied tail current. B, 10 and 20 μM niflumic acid applications progressively increase the I_K(Ca) at various membrane potentials. The niflumic acid effect is not voltage dependent. The I-V curves were generated from the same cell as in A.

The kinetics of activation and deactivation: the Bay K effect

As seen in the previous figures, the time course of activation and deactivation (the tail current) of I_K(Ca) is very slow (seconds). We hypothesized that both of these rates may reflect the speed of Ca²⁺ accumulation caused by Ca²⁺ influx through voltage-dependent Ca²⁺ channels (VDCCs) (activation) and the speed of Ca²⁺ buffering (deactivation), rather than the behaviour of the channel gate itself. We used Bay K 8644, a dihydropyridine activator of L-type Ca²⁺ channels (Lebrun & Atwater, 1985), to test this hypothesis. Bay K 8644 is expected to increase the rate of the channel activation, because it will increase the size of the inward Ca²⁺ current and therefore the amount of Ca²⁺ influx. It is also expected to slow down the tail current (the deactivation), as more time will be required to buffer, sequester and pump out increased amounts of Ca²⁺. As shown in Fig. 7A, Bay K 8644 (1 μM) increased the amplitude of I_K(Ca) and the tail current. When the control peak outward currents are scaled appropriately to match those in the presence of Bay K 8644 (Fig. 7B), it can be seen that Bay K increases the activation rate and slows the tail current (deactivation). These results suggest, therefore, that the speed of activation and deactivation are dependent on intracellular Ca²⁺. The dihydropyridine Ca²⁺ channel blocker nifedipine (100 nM), as well as the phenylalkylamine verapamil (100 μM), reversibly blocked the activation of I_K(Ca) by depolarizing voltage steps (n= 4, data not shown).

Figure 7. The speed of onset and deactivation of the current are affected by the L-type Ca²⁺ channel agonist Bay K 8644.

A, Ca²⁺-activated K⁺ current was evoked by a long depolarizing pulse in the presence and absence of Bay K 8644 (1 μM). The amplitude of the ‘onset’ and the tail current is increased. B, the same recording as in A with the control trace scaled to match the maximum amplitude of that in the presence of Bay K 8644. The faster time course of the activation (onset) and slower time course of deactivation (tail) in Bay K 8644 are observed.

Muscarinic activation of I_K(Ca) in βTC-3 cells

Acetylcholine (acting through muscarinic receptors) has been shown to activate Ca²⁺-activated K⁺ channels in lacrimal cells (Marty, 1989) by releasing Ca²⁺ from intracellular stores. We tested whether carbachol and muscarine could also evoke I_K(Ca). As shown in Fig. 8A, carbachol evoked a large current which reversed at -40 mV (close to E_K in 30 mM K⁺). The response desensitizes rapidly (data not shown). Following washout of carbachol, [K⁺] was lowered to 5 mM and carbachol was applied again (Fig. 8B). This time the evoked current reversed at -85 mV, close to E_K. In the same cell, voltage steps also evoked the slow I_K(Ca) (data not shown).

Figure 8. Carbachol activated the slow I_K(Ca) in the presence or absence of external Ca²⁺.

All panels show recordings from the same cell. A, a ramp from -110 to +50 mV was applied to a βTC-3 cell which exhibited I_K(Ca) in normal external Ca²⁺ (2 mM) and 30 mM K⁺. At 50 μM, carbachol activated a current which crossed the steady-state ramp I-V at -40 mV, close to the calculated E_K (-42.1 mV). B, upon washout of carbachol, the external K⁺ was reduced to 5 mM and 100 μM carbachol was applied again. The carbachol-sensitive current reversed at around -85 mV, close to E_K. Carbachol was washed off afterwards. C, the steady-state ramp I-V in the presence and absence of external Ca²⁺. D, in the absence of external Ca²⁺, carbachol (200 μM) still elicited a linear current which reversed at -85 mV. (The experiments shown in A, B, C and D were performed sequentially.)

Figure 8C shows that upon removal of external Ca²⁺ (substituted with Mg²⁺) I_K(Ca) was abolished. Carbachol could also activate the channels when the external Ca²⁺ was removed (Fig. 8D). The evoked current reversed close to the E_K and had a linear shape over the whole voltage range. This result is consistent with the interpretations that carbachol exerts its actions by releasing Ca²⁺ from the internal stores by a mechanism independent of external Ca²⁺ influx.

We next examined the carbachol-induced increase in I_K(Ca) under conditions where the latter was activated by a long depolarizing voltage step. In Fig. 9A, carbachol (100 μM) increased the outward current activated by depolarization. The time course of activation changed, as seen by the shorter rise time. To test whether carbachol activates the same I_K(Ca) as the voltage step, we applied ChTX (200 nM) after washing out carbachol. Figure 9B shows that ChTX effectively blocked the outward current and revealed a contaminating I_Cl(Ca) current as judged from the inward tail. Subsequent application of carbachol (100 μM) in the presence of the toxin resulted in a very modest increase in the outward current, which was as expected if carbachol acts on the ChTX-sensitive current. The increase in the outward current can probably be attributed to the activation of I_Cl(Ca), which is shown to be present in this cell by the inward tail current at -60 mV. After washing out carbachol and ChTX, carbachol (100 μM) was applied again (Fig. 9D) and produced a much greater effect on the outward current than that demonstrated in panel C. Therefore the small carbachol effect shown in panel C is most probably not due to desensitization of the muscarinic receptor.

Figure 9. The current activated by carbachol is sensitive to ChTX.

A, carbachol increases the amplitude of the I_K(Ca) elicited by a depolarizing voltage step. Note that both the ‘onset’ and the outward tail currents are affected. B, after washing off carbachol, 200 nM charybdotoxin blocked most of the outward current (voltage-activated) and revealed an inward tail. It is most likely that this cell has a contaminating I_Cl(Ca) conductance which is revealed when most of the I_K(Ca) is blocked by ChTX. Hence the change in the direction of the tail from outward (I_K(Ca)+I_Cl(Ca)) to inward (mainly I_Cl(Ca)). C, in the presence of ChTX a second application of carbachol (100 μM) causes a small increase in the outward ‘onset’ current. D, after washing off carbachol and ChTX, carbachol (100 μM) is applied for the third time, causing a larger increase in the outward current. E, time course of carbachol-induced activation of I_K(Ca) at a holding potential (V_h) of -30 mV. At this potential I_K(Ca) is expected to be outward whereas the driving force for Cl⁻ is minimal (E_Cl≈ -20 mV) and therefore possible contamination from I_Cl(Ca) is minimized.

Figure 9E shows a response to carbachol in another cell which expressed I_K(Ca). The cell was held at -30 mV. Carbachol (100 μM) elicited an outward peak followed by oscillations. This pattern of oscillations occurred in most cells exposed to carbachol (7 out of 10 tested). The I_K(Ca) oscillations are likely to reflect oscillations in intracellular Ca²⁺ released from IP₃-sensitive stores, since they occur at a frequency similar to that described previously for such Ca²⁺ release (Gromada & Dissing, 1996).

Muscarine (50 μM) activated I_K(Ca) in a manner similar to carbachol (n= 2; data not shown).

Existence of a slow I_K(Ca) in acutely isolated porcine islet cells

We next tested whether a slow K_Ca conductance appears also in acutely dissociated, ‘normal’β-cells. We used conditions for pig islet cells similar to those described for βTC-3 cells. Figure 10 shows representative recordings from isolated porcine cells. In panel A, a 6 s depolarizing step evoked a slowly activating outward current followed by a characteristic slow tail. As is the case for βTC-3 cells, the slow I_K(Ca) of porcine islet cells could be blocked reversibly by 100 μM Cd²⁺. The inset shows the early part of the record and the block of I_Ca by Cd²⁺.

Figure 10. The slow Ca²⁺-activated current is present in acutely dissociated porcine islet cells.

A, long depolarizing pulses to 20 mV evoked a slowly activating outward current which was blocked by 100 μM Cd²⁺. The inset shows the initial part of the recording where the block of the I_Ca is seen. B, a similar experiment in another porcine islet cell, which exhibited a large transient outward (A-type) current. At 100 μM, Cd²⁺ failed to affect the A-type current but completely abolished the slowly activating current. In the presence of Cd²⁺, 6.4 mM 4-AP blocked the transient current. The effect of Cd²⁺ was reversible. The inset shows the initial part of the record demonstrating the 4-AP effect on the transient current.

The non-specific K⁺ channel blocker quinidine (200 μM) completely abolished the Ca²⁺-sensitive current (n= 4; data not shown). I_K(Ca) was seen in sixteen porcine cells and was insensitive to external TEA (5 mM; n= 3; data not shown). Pig cells consistently displayed a fast outward A-like K⁺ current which was sensitive to 4-AP. In panel B, a recording from a cell is shown which displayed a pronounced A-like K⁺ current. First, when Cd²⁺ was applied, it affected only the slow current but not the spike of the A-current. In the presence of 100 μM Cd²⁺, 6.4 mM 4-AP completely abolished the A-type current, as shown in the inset. The effect of Cd²⁺ was reversible. Thus, the porcine I_K(Ca) appeared similar to I_K(Ca) in βTC-3 cells (Fig. 1) with respect to kinetics and sensitivity to TEA and quinidine. However, charybdotoxin at concentrations of 100-200 nM was ineffective in blocking the current (n= 5), suggesting species differences in the pharmacological characteristics of these Ca²⁺-dependent K⁺ currents.

DISCUSSION

In the present study we describe a Ca²⁺-activated K⁺ current in insulin-secreting cells which displays slow activation and deactivation kinetics, is insensitive to apamin and is distinct from other K_Ca channels, such as the maxi-K_Ca or SK channels.

βTC-3 cells (Efrat et al. 1988) have served as a convenient model to study β-cells, since they resemble ‘normal’β-cells in several important properties. They secrete insulin when glucose concentration is increased and produce significantly higher levels of insulin than RIN (rat islet insulinoma) or HIT (hamster islet cell line) cells. βTC-3 cells have also been studied electrophysiologically: they were shown to possess, for example, the ATP-dependent K⁺ (K_ATP) channel, a hallmark of β-cell electrical activity. In addition, they have recently been used for studies of muscarinic elevation of cytoplasmic Ca²⁺ via release from internal stores (Gromada & Dissing, 1996).

We have previously shown that βTC-3 cells possess ‘L-like’ VDCCs and a Ca²⁺-dependent Cl⁻ current (I_Cl(Ca)) (Kozak & Logothetis, 1997). In the present study we show that a TEA-insensitive Ca²⁺-activated K⁺ current, sensitive to block by ChTX can be activated by influx of Ca²⁺ through these channels. We also show that the same K⁺ channels can be activated by muscarinic agonists, even when the VDCCs are blocked, presumably through muscarinic release of Ca²⁺ from internal stores.

The Ca²⁺-activated K⁺ currents in insulin-secreting cells

Earlier studies in β-cells have demonstrated the existence of a large-conductance or maxi-K_Ca channel (Findlay, Dunne, Ullrich, Wollheim & Petersen, 1985; Tabcharani & Misler, 1989) and a small-conductance K_Ca channel (Marty, 1989; Ämmäläet al. 1991, 1993). The maxi-K_Ca channel, which has been studied extensively, was shown to be blocked by ChTX (20 nM) and TEA (1 mM) from the external side (Kukuljan et al. 1991). It was also shown to be sensitive to micromolar concentrations of quinine. This channel is insensitive to apamin, a bee venom toxin.

The small-conductance K_Ca channel could be activated by intracellular GTP and carbachol (Ämmäläet al. 1993). This current was shown to be insensitive to apamin and charybdotoxin (100 nM) (Ämmäläet al. 1991).

The I_K(Ca) we describe has unique pharmacology. It is insensitive to D-tubocurarine, apamin and scyllatoxin (leiurotoxin I), specific blockers of small-conductance K_Ca channels found in other tissues (Castle et al. 1989). Surprisingly, ChTX at high nanomolar concentrations was found to reversibly inhibit this current (100 nM inhibits 80 %). Charybdotoxin has been shown to block various Ca²⁺-dependent and -independent K⁺ channels (Garcia, Galvez, Garcia-Calvo, King, Vazquez & Kaczorowski, 1991), in addition to its high affinity block of the maxi-K_Ca channel (Blatz & Magleby, 1987). Since all our experiments were carried out in the presence of 10 mM TEA, the participation of the maxi-K_Ca channel can be ruled out. Iberiotoxin (IbTX), a peptide closely related to ChTX, is known to be a specific blocker of the maxi-K_Ca channel at picomolar concentrations. IbTX (100 nM) was ineffective in blocking the βTC-3 cell K_Ca channel. Kaliotoxin, a scorpion toxin which blocks the neuronal K_Ca channels, was also ineffective. Quinine (100 μM), on the other hand, completely abolished this I_K(Ca). The Cl⁻ channel blockers niflumic, mefenamic and flufenamic acid have been shown to potentiate maxi-K_Ca channels (Ottolia & Toro, 1994; Gribkoff et al. 1996) and the recombinant human I_SK (non Ca²⁺-dependent) channel. We found that niflumic acid (micromolar concentrations) potently and reversibly activated our slow, apamin-insensitive I_K(Ca). The effect of niflumic acid was not voltage dependent and was abolished in the presence of Cd²⁺. We have demonstrated previously that I_Cl(Ca) in βTC-3 cells is blocked by 100 μM niflumic acid. Thus, niflumic acid can serve as a useful tool for distinguishing the physiological roles of the I_K(Ca) and I_Cl(Ca) in β-cells.

We found that the most permeant cation for the K_Ca channel was Tl⁺, followed in decreasing order of permeance by K⁺, Rb⁺ and NH₄⁺. The selectivity sequence is similar to that of the maxi-K_Ca channel studied in β-cells (Tabcharani & Misler, 1989). Cs⁺ and Na⁺ were virtually impermeant through the channel. This contrasts with the small-conductance K⁺ channel, which in other secretory cells (Park, 1994) is permeable to Cs⁺.

Muscarinic agonists activate I_K(Ca)

In pancreatic β-cells, acetylcholine exerts its effects through muscarinic receptors of the M3 subtype (Boschero et al. 1995), which are coupled to the phospholipase C-IP₃ pathway (Malaisse, 1986). IP₃ triggers a large release of Ca²⁺ from internal stores (Petersen, Petersen & Kasai, 1994; Gromada & Dissing, 1996), which in turn can act as a second messenger on various effectors. In β-cells, carbachol and other muscarinic agonists increase glucose-induced insulin release. Ämmäläet al. (1993) showed that in single β-cells micromolar concentrations of carbachol activate the small-conductance K_Ca channel through oscillations in Ca²⁺ levels. Recently Atwater's group (Bordin et al. 1995) demonstrated that in islets carbachol induces a Ca²⁺-dependent K⁺ conductance which is blocked partially by 50 nM ChTX. In βTC-3 cells, we show that carbachol and muscarine activate I_K(Ca) in the presence or absence of Cd²⁺. This current can therefore be activated both by Ca²⁺ influx through the ‘L-like’ channels and by Ca²⁺ release from internal stores (ER) triggered by hormonal activation. The current activated by carbachol does not exhibit the bell-shaped I-V relationship seen when current is activated by voltage steps, but rather appears to be linear. The bell-shaped I-V relation of the K_Ca channel is therefore caused indirectly by the I-V characteristics of Ca²⁺ channels.

As seen in Fig. 9A, the slow time course of activation of I_K(Ca) is changed when carbachol acts to release Ca²⁺ from internal stores. The current activates much faster, presumably because the massive release of Ca²⁺ from the stores rapidly increases the Ca²⁺ concentration in the vicinity of the channels. This is consistent with data shown in Fig. 7, where increased Ca²⁺ influx due to Bay K action causes faster activation of I_K(Ca).

As shown recently (Gromada & Dissing, 1996), application of acetylcholine to βTC-3 cells elicits a peak rise in internal [Ca²⁺]_i caused by release from the internal stores. The internal Ca²⁺ elevation reaches a maximum within 5 s and declines to prestimulatory levels after approximately 40 s. The IP₃ levels rise and fall in parallel with those of [Ca²⁺]_i. Our findings of activation of I_K(Ca) by carbachol are in agreement with this study. Figure 9E shows the time course of I_K(Ca) activation at -30 mV which is similar to that of the [Ca²⁺]_i level fluctuations in βTC-3 cells. We also frequently observe oscillations of the current in response to muscarinic stimulation in the presence and absence of external Ca²⁺. The lack of such oscillations in the study by Gromada & Dissing (1996) may result from the concentration of ACh not being high enough. We observed oscillations when the carbachol concentrations used were 100 μM or higher.

How does our I_K(Ca) relate to the other K_Ca conductances described in β-cells? It resembles the small-conductance channel described in mouse β-cells (Ämmäläet al. 1991) in that it is apamin-insensitive and has slow kinetics, but differs from it in its sensitivity to ChTX. Ämmälä and colleagues tested the effects of ChTX at a concentration as high as 100 nM at various voltages (n= 3). At 100 nM, ChTX blocked only 8 % of the outward current at -40 mV, a voltage level at which most of the maxi-K_Ca channels are not active. However, TEA, in the same experiment, blocked 23 % of the total current. ChTX was not applied in the presence of 5 mM TEA to determine whether ChTX and TEA blocked the same current. The reason for the disparity between block by TEA and ChTX (23 vs. 8 %) is difficult to discern, if we assume that both agents at these high concentrations block the maxi-K_Ca current alone. The evidence for insensitivity of the small-conductance K_Ca channel to ChTX needs to be clarified further. The possibility exists that the I_K(Ca) in the βTC-3 cell line has a higher sensitivity to ChTX than its counterpart in ‘normal’ cells.

Kukuljan et al. (1991) showed that crude scorpion venom or application of 20 nM ChTX had no effect on β-cell bursting behaviour. This finding strongly suggested that the maxi-K_Ca channels are not involved in the pacemaking conductance in the bursting cells. Bordin et al. (1995) showed that 50 nM ChTX reduced the carbachol-induced hyperpolarization and concluded that the maxi-K_Ca channels may be involved in the muscarinic-induced hyperpolarization but not in that induced by glucose. If the I_K(Ca) we have described was present, then its block by ChTX could have explained these results. It is unlikely that the maxi-K_Ca channel would be involved, since the probability that it would be open at negative membrane potentials is low (Tabcharani & Misler, 1989). It may be the case that the I_K(Ca) we have described is involved in setting the interburst hyperpolarizing intervals and carbachol-induced hyperpolarizations.

We also attempted to test whether the slow Ca²⁺-dependent K⁺ current is also present in ‘normal’, acutely isolated β-cells. In porcine dissociated islet cells, a current similar to that in βTC-3 cells appeared to be sensitive to Cd²⁺ and quinidine. However, this current was not sensitive to ChTX. Indeed β-cell electrical behaviour has revealed specific differences between various species. For example, in dog and human β-cells, there is a large voltage-dependent Na⁺ current generating trains of action potentials which in turn activate Ca²⁺ channels (Pressel & Misler, 1991; Barnett, Pressel & Misler, 1995; but see also Kelly, Sutton & Ashcroft, 1991). In mouse and rat β-cells, on the other hand, Na⁺ channels are thought not to play a significant role in the firing pattern of the cell. Differences have been noted even between mouse and rat β-cell Ca²⁺ currents; rat cells are thought to posses both L- and T-type VDCCs, whereas mouse cells are believed to possess only slow L-type channels (Ashcroft & Rorsman, 1991). This provides the possibility for interesting species differences in the pharmacology of this channel.

Generally, single β-cells do not exhibit bursting behaviour (Rorsman, Bokvist, Ämmälä, Eliasson, Renström & Gäbel, 1994). In our hands, βTC-3 cells fire regular action potentials. Interestingly, we find that the presence of the I_K(Ca) in these cells is necessary and sufficient for their firing (n= 10; data not shown). These preliminary results provide some evidence for this current having the role of a feedback conductance which can restart the firing after the plateau depolarization.