BK channels mediate a novel ionic mechanism that regulates glucose-dependent electrical activity and insulin secretion in mouse pancreatic β-cells

Abstract

BK channels are large unitary conductance K⁺ channels cooperatively activated by intracellular calcium and membrane depolarisation. We show that BK channels regulate electrical activity in β-cells of mouse pancreatic islets exposed to elevated glucose. In 11.1 mm glucose, the non-peptidyl BK channel blocker paxilline increased the height of β-cell action potentials (APs) by 21 mV without affecting burst- or silent-period durations. In isolated β-cells, paxilline increased AP height by 16 mV without affecting resting membrane potential. In voltage clamp, paxilline blocked a transient component of outward current activated by a short depolarisation, which accounted for at least 90% of the initial outward K⁺ current. This BK current (I_BK) was blocked by the Ca²⁺ channel blockers Cd²⁺ (200 μm) or nimodipine (1 μm), and potentiated by FPL-64176 (1 μm). I_BK was also 56% blocked by the BK channel blocker iberiotoxin (100 nm). I_BK activated more than 10-fold faster than the delayed rectifier I_Kv over the physiological voltage range, and partially inactivated. An AP-like command revealed that I_BK activated and deactivated faster than I_Kv and accounted for 86% of peak I_K, explaining why I_BK block increased AP height. A higher amplitude AP-like command, patterned on an AP recorded in 11.1 mm glucose plus paxilline, activated 4-fold more I_Kv and significantly increased Ca²⁺ entry. Paxilline increased insulin secretion in islets exposed to 11.1 mm glucose by 67%, but did not affect basal secretion in 2.8 mm glucose. These data suggest a modified model of β-cell AP generation where I_BK and I_Kv coordinate the AP repolarisation.

Introduction

Elevated ambient glucose triggers action potentials (APs) in β-cells of murine pancreatic islets (Dean & Matthews, 1968). It is thought that voltage-dependent Ca²⁺ current (I_Cav) underlies the AP upstroke, and delayed rectifier K⁺ current (I_Kv, mediated mostly by K_v2.1 channels (Roe et al. 1996; MacDonald & Wheeler, 2003; Herrington et al. 2006)) drives the downstroke (Ashcroft & Rorsman, 1989; Houamed et al. 2004).

BK channels are K⁺ channels of large unitary conductance that are activated cooperatively by elevated [Ca²⁺]_i and membrane depolarisation (Marty, 1981; Kaczorowski et al. 1996; Vergara et al. 1998). This dual sensitivity allows BK channels to subserve disparate physiological functions in many cell types (Orio et al. 2002). BK channels vary widely in their biophysical and pharmacological properties as a result of multiple splicing of the ion-conducting α-subunit mRNA (Adelman et al. 1992; Butler et al. 1993), differential incorporation of four modulatory β-subunits (Orio et al. 2002), and multiple phosphorylation patterns (Yan et al. 2008).

Rodent β-cells and certain insulinomas express BK channels (Cook et al. 1984; Findlay et al. 1985; Satin et al. 1989; Tabcharani & Misler, 1989; Bokvist et al. 1990; Mancilla & Rojas, 1990; Smith et al. 1990b; Kukuljan et al. 1991; Li et al. 1999), but the role of these channels in β-cell physiology has been controversial. Early studies and theoretical models proposed that I_BK regulated glucose-dependent electrical excitability in β-cells (Atwater et al. 1979, 1983; Chay, 1986). However, Smith et al. (1990b) observed that I_BK constituted only a minor and variable component of the depolarisation-activated I_K, concluding that it was thus unlikely to play a major role in shaping APs. Similarly, Kukuljan et al. (1991) reported that the I_BK blocker charybdotoxin (ChTx) did not affect glucose-induced APs or islet bursting, and concluded that I_BK did not participate in glucose-induced electrical activity. Thus, there emerged a consensus that BK channels do not significantly participate in the glucose-mediated islet electrical activity or in glucose-stimulated insulin secretion (GSIS) in mouse. However, β-cells from a K_v2.1 knockout mouse lacking I_Kv are still capable of AP repolarisation, and moreover, express a transient outward K⁺ current of unknown molecular identity (Jacobson et al. 2007). Similarly, toxin block of β-cell K_v2.1 channels slowed, but did not inhibit, AP repolarisation. Further, human β-cells have recently been shown to express a robust I_BK whose pharmacological modulation affects both AP shape and GSIS (Braun et al. 2008).

The present study examines whether I_BK is activated in primary mouse β-cells under physiological conditions, and investigates its possible role in the control of islet electrical activity and GSIS. Preliminary results have been described previously in abstract form (Houamed & Satin, 2009).

Methods

Isolation of islets and single β-cells

Islets were isolated from adult male Swiss Webster mice as previously described (Goforth et al. 2002; Zhang et al. 2005), using a protocol approved by the University of Michigan University Committee on the Use and Care of Animals, and following published standards (Drummond, 2009). Animals were killed by neck dislocation. The pancreas was perfused through the bile duct with cold Krebs solution containing 1 mg ml⁻¹ collagenase P (Roche Diagnostics, Indianapolis, IN, USA) and 0.1% bovine serum albumin (BSA), then rapidly excised, incubated at 37°C for 15 min, triturated and washed. Krebs solution contained (in mm): 137 NaCl, 5 KCl, 1.2 MgCl₂, 1 CaCl₂, 5 NaHCO₃, 5 glucose, 10 Hepes; pH 7.35. Single β-cells and islets were dispersed in Spinner's salts, washed with culture medium and plated onto glass coverslips. Islets and single β-cells were maintained in RPMI 1640 medium supplemented with 10% fetal calf serum and antibiotics (Invitrogen, Carlsbad, CA, USA) in 5% CO₂ at 37°C. Most islets and single cells were used within 24 h and 72 h, respectively. As we have noted previously (Zhang et al. 2003), these culture periods did not appear to affect the electrical properties or the glucose sensitivity of islets and individual β-cells.

Electrophysiology

Single islets or coverslips with attached β-cells were placed in a ∼300 μl recording chamber constantly superfused, at 2 ml min⁻¹ and 33.5°C, with a saline solution containing (in mm): 140 NaCl, 10 Hepes, 1 MgCl₂, 2.5 CaCl₂, 3.6 KCl, pH 7.35. Drugs and glucose were dissolved directly into the saline solution; when drugs were diluted from DMSO stocks, the highest final concentration of the solvent was <0.1%. Iberiotoxin solutions contained 0.1% BSA to prevent the toxin from sticking to the perfusion system. Glucose (11.1 mm) was added for membrane potential recording; for voltage-clamp recordings, 5 mm glucose and 200 μm tolbutamide were present. The two glucose concentrations, which have been used in β-cell studies involving measurement of membrane potential and current, Ca²⁺_i dynamics, and insulin secretion (Gopel et al. 1999b; Zhang et al. 2005), were used to facilitate comparison with previous studies. For I_Cav measurements, 30 mm TEA was added to block I_K, and NaCl reduced accordingly to maintain the osmolarity of the saline solution.

Recordings utilized the perforated patch technique, using the pore-forming compounds nystatin or amphotericin-B (Falke et al. 1989). Pipette resistance was 1.6–4.4 MΩ when filled with a solution composed of (in mm): 76 K₂SO₄, 5 NaCl, 10 KCl, 1 MgCl₂, 10 Hepes, pH 7.35. For I_Cav measurements, Cs₂SO₄ replaced K₂SO₄. The pore formers were dissolved in DMSO at 50 μg ml⁻¹ then diluted in pipette solution to a final concentration of 250 μg ml⁻¹. Pipette tips were dipped in pipette solution, then back-filled with pipette solution containing the pore former. Recordings commenced when the series resistance (R_s) was <40 MΩ; most of the remaining R_s was compensated electronically so that the voltage error generated by the largest currents was <6 mV. All salts and drugs were from Sigma (St Louis, MO, USA), except paxilline, which was from Enzo Life Sciences (Plymouth Meeting, PA, USA), and iberiotoxin, which was from AnaSpec (Fremont, CA, USA).

Functional criteria used to identify β-cells in situ on the surface of an islet were presence of AP bursts in 11.1 mm glucose, membrane capacitances >5.7 pF, and the absence of an early Na⁺ current that remains activatable at physiological membrane potentials (Gopel et al. 1999a). Similar criteria were used to identify dissociated β-cells; in addition, we visually selected for larger cells, which were more likely to be β-cells (Leung et al. 2005). Data reported in the present study are based on recordings obtained from 51 isolated β-cells whose average membrane electrical capacitance was 7.86 ± 0.32 pF. The average capacitance of β-cells in situ in islets was 8.34 ± 0.39 pF (n = 18).

An EPC9 amplifier and PULSE software (HEKA, Lambrecht/Pfalz, Germany) provided stimulation and data acquisition. This study utilized four pulse protocols (P1, AP1, AP2 and IV1). P1 consisted of 40 ms depolarising pulse from −70 mV to −10 mV. AP1 and AP2 were synthetic APs patterned on β-cell APs recorded in 11.1 mm glucose without and with 1 μm paxilline, respectively (Fig. 1D). Each AP command consisted of a −65 mV holding potential segment followed by 14 ramps (Fig. 5A and B). IV1 consisted of 500 ms depolarising test pulses from −70 mV holding potential; test potentials were −60 mV to +10 mV in 10 mV increments. Linear leak and capacitive components of membrane current were subtracted by a combination of analog and digital (p/n) techniques; data traces shown in Figs 2, 5 and 6 are averages of several consecutive recordings. Data were analysed using PULSEFIT (HEKA), Igor (Wavemetrics, Lake Oswego, OR, USA), Excel (Microsoft, Redmond, WA, USA) and Prism (Graphpad, La Jolla, CA, USA). APs were analysed using Synaptosoft (Decatur, GA, USA). Recordings were corrected for a +10.8 to +12.3 mV calculated junction potential (Neher, 1992).

Figure 1. Effects of BK channel blocker paxilline on the electrical activity of glucose-stimulated islets and single β-cells.

A, membrane potential recording in an islet exposed to 11.1 mm glucose, showing typical bursting. Paxilline (1 μm; horizontal bar) dramatically increased AP height. B, data from A on an expanded time scale. The arrows show approximate location of traces in A. C, effect of paxilline on APs of a single β-cell exposed to 1 μm paxilline in 11.1 mm glucose (denoted by the horizontal bar). D, two APs extracted from the trace in C are shown superimposed; dotted and continuous traces were recorded in the absence and presence of paxilline, respectively. Dashed lines denote 0 mV level; all panels have the same voltage calibration.

Figure 5. Activation of I_BK and I_Kv by waveform commands patterned on β-cell APs.

The waveform in A: low amplitude AP1 command (continuous line) patterned on the AP (dotted line), recorded in 11.1 mm glucose, shown in Fig. 1D. B, high amplitude AP2 command (continuous line) patterned on the high amplitude AP, recorded in 11.1 mm glucose and 1 μm paxilline, shown in Fig. 1B. Voltage calibration bar refers to A and B. C, membrane currents activated by AP1. ‘Control’ (black) was recorded in 5 mm glucose and 200 μm tolbutamide. ‘Pax’ (green) was recorded after 1 μm paxilline. ‘Pax/Cd’ (blue) was recorded in 1 μm paxilline and 200 μm Cd²⁺. D, comparison of I_BK and I_Kv activated by AP1. ‘BK’ (red) is the paxilline-sensitive I_BK component, obtained by subtracting ‘Pax’ from ‘Control’ in A. ‘Kv’ (blue) is I_Kv, the current remaining after paxilline and Cd²⁺ have blocked I_BK and I_Cav; it is identical to the ‘Pax/Cd’ trace in C. E, comparison of I_BK and I_Kv activation and deactivation time courses. I_BK (‘BK’; red) and I_Kv (‘Kv’; blue) have been normalized to their respective peaks (arbitrary amplitude scale). F, AP2 command enhances I_Kv activation. ‘Kv’ (blue) is the same I_Kv trace marked ‘Pax/Cd’ and ‘Kv’, in C and D, respectively. ‘Kv AP2’ (purple) is I_Kv elicited by AP2 command. Current calibration bar refer to C, D and F; vertical calibration of E is arbitrary; the time calibration bar refers to all the panels.

Figure 2. Identification of I_BK in voltage-clamped β-cells.

A, membrane currents recorded in response to P1 command. a was recorded in 5 mm glucose and 200 μm tolbutamide. b was recorded in the added presence of 1 μm paxilline; c was recorded in the added presence of 200 μm Cd²⁺. d shows full recovery of inward current and partial recovery of the outward current upon return to ‘Control’ solution. e shows the paxilline-sensitive I_BK, obtained by subtraction of b from a. f shows Cd²⁺-sensitive current, obtained by subtraction of c from a. B, time course of the experiment illustrated in A. Circles and squares denote the peak outward I_K and inward I_Cav currents, respectively. The numbers above the graph depict the approximate positions for a, b, c and d shown in A. The horizontal bars indicate periods of paxilline and Cd²⁺application. C, membrane currents, recorded in a different β-cell. ‘Control’ trace was recorded in solution containing 5 mm glucose and 200 μm tolbutamide. ‘Cd’ trace was recorded in the presence of 200 μm Cd²⁺ to block I_Cav and Ca²⁺-dependent currents. ‘Wash’ trace shows full recovery of inward and outward current components upon return to Control solution. ‘Cd-sens’ trace shows Cd²⁺-sensitive current (I_BK, I_Cav and potentially other Ca²⁺-sensitive currents) obtained by subtraction of Cd from Control traces. The calibration bar is the same as in A. D, time course of experiment in C. Circles and squares denote peak outward and inward currents, respectively. The numbers above the graph depict positions of the traces shown in C; the horizontal bar indicates duration of Cd²⁺ application. Both inward and outward currents were blocked by Cd²⁺ and recovered rapidly upon washout. Parallel vertical lines depict a 255 s break in the recording while other protocols were executed. Dashed lines in A–D indicate zero current level. E, correlation between I_Cav and I_BK in individual β-cells. The continuous line shown is a least-square fit to the data with slope −0.9763. F, dynamics of the contributions of I_BK and I_Kv to current elicited by a short depolarizing pulse. The proportion of total outward I_K carried by I_BK and I_Kv, respectively, are shown as a function of time after the pulse commenced at 5 ms. The continuous lines represent the mean of 12 cells; the dotted lines represent the s.e.m.

Figure 6. I_Cav activation by AP-waveform commands.

I_Cav activated by AP1 (A) and AP2 (B) commands. C and D show the corresponding cumulative Ca²⁺influx (Q_Cav). The dashed lines indicate the zero I_Cav or Q_Cav. E, a family of I_Cav traces activated by IV1 protocol, recorded in the same cell as panels A and C. F, comparison of I_Cav waveforms activated by AP commands and step depolarisations. Circles represent the voltage dependence of peak I_Cav traces activated by the depolarizing steps in E. Blue and red curves represent the I_Cavs activated by AP1 and AP2, plotted as a function of their voltage waveforms. The red arrows illustrate the direction of the current progression with time in response to AP2 command. Both currents loop clockwise, beginning and ending at the top left corner of the panel; the current activated by AP2 (red trace) has an additional anticlockwise loop, covering ∼−5 to +15 mV potential range, nested within the main current loop.

Insulin secretion measurements

Insulin secretion measurements were as previously described (Jung et al. 2009). Islets were preincubated for 60 min in Krebs–Ringer bicarbonate solution composed of (in mm): 98.5 NaCl, 4.9 KCl, 2.6 CaCl₂, 1.2 MgSO₄, 1.2 KH₂PO₄, 25.9 NaHCO₃, 20 Hepes, 2.8 glucose and 0.1% BSA. Batches of 10 islets were incubated for 60 min in 200 μl of RPMI 1640 media containing either 2.8 or 11.1 mm glucose, in the presence or absence of 1 μm paxilline; supernatant insulin was determined by radioimmunoassay (RIA; Linco Research Inc., St Charles, MO, USA). Determinations were in triplicate, and the experiments were repeated on three different islet preparations.

Statistics

Error bars denote s.e.m.; Student's t test was used to determine statistical significance.

Results

The effect of BK channel block on glucose-dependent electrical activity in islets

Islets exposed to elevated glucose generate AP bursts separated by silent intervals (Fig. 1A) (Henquin & Meissner, 1984; Gopel et al. 1999b; Zhang et al. 2003, 2005). Slow-, fast- and mixed-bursting patterns, ranging in frequency from ∼2 to ∼0.2 min⁻¹, have been observed in individual islets (Zhang et al. 2003). We investigated the effect of 1 μm paxilline on the membrane potential of superficial β-cells (n = 22; average recording per cell ∼25 min) on intact islets stimulated by 11.1 mm glucose (Fig. 1A). Paxilline is a non-peptidyl tremourogenic BK channel inhibitor with nanomolar potency (Knaus et al. 1994; Sanchez & McManus, 1996). As the effects of paxilline were poorly reversible, each islet or single β-cell was exposed to the drug only once.

Analysis of burst- and silent-interval durations in pooled data revealed that the pattern of alternating bursts and silent intervals was unaffected by addition of paxilline, while AP amplitude was increased dramatically. The mean burst duration was 178.29 ± 39.80 s (n = 18 islets) in 11.1 mm glucose, and 162.37 ± 32.14 s (n = 17) after paxilline addition; these values were not significantly different (P = 0.76). Likewise, the mean silent interval duration in 11.1 mm glucose was 106.31 ± 18.85 s (n = 16), and 94.80 ± 14.55 s (n = 16) in paxilline; this difference was also not statistically significant (P = 0.63). In terms of duty cycle, islets spent on average 62.7% of the time bursting in 11.1 mm glucose alone, and 63.1% when paxilline was present.

Paxilline changed the baseline potential immediately preceding each AP by a statistically insignificant −0.5 ± 0.9 mV, but significantly potentiated the AP peak by 20.3 ± 3.0 mV (n = 6; P < 0.0001; Fig. 1B). Consequently, BK channel blockade in islets exposed to 11.1 mm glucose resulted in increased AP height, but did not alter burst dynamics.

The effect of BK channel block on glucose-dependent APs in single β-cells

In contrast to the bursting behaviour seen in intact islets, elevated glucose triggers continuous AP firing in most single β-cells, with a minority of cells showing a variety of bursting behaviours (Smith et al. 1990a; Ammala et al. 1991; Kinard et al. 1999). To test whether the effects of paxilline on APs in whole islets were recapitulated in isolated single β-cells, membrane potential was recorded in 11.1 mm glucose in the absence and presence of 1 μm paxilline (Fig. 1C). Analysis of 751 APs in 11.1 mm glucose and 352 APs in the added presence of paxilline showed that paxilline enhanced AP peak height by an average of 15.6 ± 0.5 mV, a difference which was highly significant (P < 0.0001) in each of the four β-cells tested. In contrast, paxilline had no significant effect on baseline potential (0.1 ± 1.0 mV).

Figure 1D shows two exemplar APs, selected from the trace in Fig. 1C, plotted on an expanded time scale and manually aligned so that their peaks coincided temporally. The dotted-line AP was recorded in 11.1 mm glucose, and the continuous-line AP was recorded after 1 μm paxilline. Paxilline potentiated AP height, resulting in a positive overshoot, did not affect baseline potential, and did not appear to affect AP width, nor did it inhibit the afterhyperpolarisation (AHP) phase that often follows the AP.

I_BK mediates a pharmacologically and kinetically distinct component of the outward current activated by a brief depolarisation

Membrane current recorded in single mouse β-cells elicited by a P1 protocol consisted of an early inward component that relaxed outwards (Fig. 2Aa). The outward component has been considered to be mainly delayed rectifier I_Kv (Smith et al. 1990b; Roe et al. 1996). However, paxilline block revealed that I_BK constituted a substantial component of the outward I_K that was larger than the paxilline-insensitive I_Kv component (Fig. 2Ab).

The early inward current was blocked by 200 μm Cd²⁺ (Fig. 2Ac), confirming its identity as I_Cav (Satin & Cook, 1985; Rorsman & Trube, 1986; Smith et al. 1993; Jing et al. 2005). Addition of Cd²⁺ in the presence of paxilline did not decrease I_Kv further (compare Fig. 2Ab and 2Ac). Upon removal of paxilline and Cd²⁺, I_Cav recovered rapidly and fully, whereas I_BK recovered slowly and partially (Fig. 2Ad). The difference current traces in Fig. 2Ae and 2Af (Fig. 2Ab minus 2Aa, Fig. 2Ac minus 2Aa, respectively) represent I_BK and the paxilline- and Cd²⁺-sensitive current.

The time course of the experiment shown in Fig. 2A is plotted in Fig. 2B, illustrating that paxilline blocks the I_BK component of outward current in a slowly reversible manner, without affecting the inward, Cd²⁺-sensitive, I_Cav. In this experiment, the poor reversibility of paxilline block did not allow us to distinguish whether Cd²⁺ and paxilline were blocking the same I_K component. In an independent set of experiments Cd²⁺, applied alone, blocked a current grossly indistinguishable from that seen when Cd²⁺ was combined with paxilline (Fig. 2C). However, unlike with paxilline, Cd²⁺ block recovered readily upon drug washout (Fig. 2D). In pooled data, Cd²⁺ alone blocked 60.0 ± 11.6% of the peak total outward current (n = 11), and in combination with paxilline blocked 60.2 ± 3.6% (n = 17); the two values were not significantly different. I_Kv accounted for 38.3 ± 3.7% of the peak total outward current (n = 17). Therefore, 1 μm paxilline blocks the β-cell I_BK selectively and completely.

Figure 2E depicts peak I_BK as a function of the corresponding peak I_Cav, determined in12 separate β-cells, spanning ∼5.7- and ∼5.0-fold ranges of I_Cav and I_BK, respectively. The continuous line fit to the data has a slope of −0.9763; I_Cav and I_BK are significantly correlated (P = 0.0267).

Additional pharmacological evidence for a Ca²⁺-activated I_K was provided by the L-type I_Cav blocker nimodipine (1 μm) and the enhancer FPL-64176 (1 μm; McDonough et al. 2005), which reduced and enhanced outward current by 15.6 ± 3.3% (P = 0.020; n = 3) and 72.2 ± 38.1% (P = 0.099; n = 3), respectively.

The scorpion-toxin peptide iberiotoxin (IbTx) is a ChTx analogue that is equally potent at blocking BK channels; unlike ChTx, which was used in an earlier study of BK channels in β-cells (Kukuljan et al. 1991), IbTx does not block K_v channels (Galvez et al. 1990). In the current study, IbTx blocked 56.2 ± 12.6% of the paxilline-sensitive I_BK current (n = 4; individual-cell block ranged from 22.3 to 81.2%).

To compare their activation time courses quantitatively, I_BK and I_Kv in 12 cells were normalized to the total outward I_K (i.e. I_BK+I_Kv; Fig. 2F). I_BK carries most of the early repolarizing current (∼100%, 2.6 ms into the pulse). Concurrent I_BK inactivation and I_Kv activation result in either current approaching 50% by the end of the 40 ms pulse.

Taken together, these observations show that I_BK constitutes a sizeable component of the outward I_K activated by depolarizing pulses in mouse β-cells, in agreement with our observation that paxilline increased the amplitude of β-cell APs.

Membrane potential- and time-dependence of I_BK in β-cells

Time- and voltage-dependence of I_BK activation and inactivation were characterized using the IV1 protocol. The multiphasic and noisy currents in Fig. 3A were recorded in 5 mm glucose and 200 μm tolbutamide. Paxilline blocked I_BK, resulting in outward currents with slower activation, less inactivation and less noise (Fig. 3B). I_Kv recorded in paxilline and Cd²⁺ activated exponentially (continuous lines superimposed on the current traces; Fig. 3C). Subtracting traces in Fig. 3B from those in Fig. 3A revealed I_BK (Fig. 3D), characterized by rapid activation, partial inactivation and prominent fluctuations; I_BK inactivated monoexponentially (continuous lines superimposed on current traces). Figure 3E shows the paxilline- and Cd²⁺-sensitive membrane current component obtained by subtracting traces in Fig. 3C from those in Fig. 3A; addition of Cd²⁺ following paxilline had little effect on outward currents but, as expected, blocked I_Cav, so that the subtracted current shown in Fig. 3E closely resembles the current in Fig. 3D, with an added early inward I_Cav component. This early I_Cav is shown on an expanded timescale in Fig. 3G.

Figure 3. Voltage- and time-dependence of I_BK and I_Kv.

Family of membrane currents activated by IV1 protocol. A, control (5 mm glucose + 200 μm tolbutamide). B, in 1 μm paxilline. C, I_Kv recorded in the presence of Cd²⁺ and paxilline. The continuous lines are single exponential fits to the rising phase of the current. D, paxilline-sensitive I_BK, obtained by subtracting the currents in B from A. The continuous lines are monoexponential fits of I_BK decay. E, Cd²⁺+ paxilline-sensitive current, obtained by subtracting traces in C from A. F, I_BK activation. The early phase of I_BK, shown in D, is plotted on an expanded time scale. The continuous lines are double exponential fits to I_BK rising phase. G, early inward I_Cav current. The traces were obtained by plotting the early stages of the traces in E on an expanded time scale. The leftmost calibration bars refer to A, B and C; the middle bars refer to D and E, and the rightmost bars refer to F and G.

Normalized conductance–voltage (G–V) curves for I_BK and I_Kv show that both currents activate over a similar membrane potential range (Fig. 4A). The dual time constants (τs) for I_BK activation and its inactivation τ, have U-shaped membrane potential dependence (Fig. 4B); I_BK activates at least an order of magnitude faster than I_Kv over all but the most depolarised potentials. Figure 4C shows the ratio of steady-state I_BK, recorded at the end of the 500 ms pulse, to peak I_BK. The line fit through the data in the range of −20 to +30 mV has a slope of 7.8 × 10⁻³ mV⁻¹, suggesting that this parameter is slightly voltage dependent.

Figure 4. Voltage dependence of amplitude and kinetics of I_BK and I_Kv.

A, normalized conductance–voltage (G–V) curves for I_BK (open circles) and I_Kv (open squares). Each point is the mean ±s.e.m. of 3–4 determinations in different cells normalised to the value at −10 mV. B, voltage dependence of the time constant of I_Kv activation (Kv_{activation τ}; filled squares), the time constant of I_BK inactivation (BK_{inactivation τ}; filled circles), the first time constant of I_BK activation (BK_{activation τ1}; upright open triangles), and the second time constant of I_BK activation (BK_{activation τ2;} inverted open triangles). Each point is mean ±s.e.m. of 3–4 determinations. Note logarithmic axis for τs. C, voltage dependence of ratio of steady-state I_BK current (average of I_BK over the last 100 ms of the 500 ms pulse) to peak I_BK. Each point is the average of 3–4 measurements in separate cells, except the leftmost point, which is a single determination. The dotted line is a linear fit to the data between −20 and +30 mV; its slope is 0.0078 mV⁻¹. The membrane potential axis is the same for all panels; the vertical dashed line denotes 0 mV.

Therefore, I_BK activates with the same voltage dependence, but ∼10-fold faster, as I_Kv, and inactivates rapidly to a steady-state component that is ∼30 to ∼60% of the peak.

K⁺ currents activated by voltage waveforms patterned on APs

To characterize the roles of I_BK and I_Kv in AP repolarisation, currents were evoked by two voltage commands (AP1 and AP2) patterned on APs recorded in 11.1 mm glucose alone and with paxilline, respectively (Figs 1D, and 5A and B). AP1 evoked a biphasic current (Fig. 5C, black trace) comprising an early transient inward component, followed by a transient outward component. Paxilline (green trace) enhanced the inward component and reduced the outward component. Subsequent Cd²⁺ addition blocked the inward I_Cav component, revealing a small transient outward I_Kv (blue trace). A transient outward paxilline-sensitive I_BK component was obtained by subtraction of the traces. The AP1-activated I_BK component was substantially larger than I_Kv (red and blue traces in Fig. 5D). In five cells, peak I_BK was 7.01 ± 2.29-fold larger than I_Kv. Normalized I_Kv and I_BK (Fig. 5E) show that I_Kv activated ∼2 ms slower than I_BK, a delay much shorter than observed with the P1 step voltage command (Figs 2–4).

The AP2 command was applied in the presence of paxilline and Cd²⁺ to investigate the I_Kv-dependent repolarizing mechanisms in the higher amplitude APs observed in paxilline (Fig. 5B). AP2 activated a dramatically larger (4.11 ± 0.76-fold; n = 5) I_Kv than AP1 (Fig. 5F). In a paired comparison, peak amplitude of I_Kv activated by AP2 was not significantly different from the peak amplitude of I_BK activated by AP1 (n = 5; P = 0.1440), indicating that the I_Kv increase observed with AP2 suffices in repolarizing the AP in the absence of I_BK.

AP2 peak is 13 mV more positive than AP1. To determine the contribution of increased conductance (G_Kv) or K⁺ electromotive driving force in the I_Kv increase, a point-by-point subtraction of the K⁺ reversal potential (determined from I_Kv tail current protocol; data not shown) from AP commands resulted in driving-voltage (ΔV) waveforms. I_Kv was divided by ΔV to produce the G_Kv waveform. Comparison of the peak amplitudes of the G_Kv and I_Kv waveforms activated by AP1 and AP2 revealed that increased G_Kv underlay 91.28 ± 2.17% (n = 8) of the I_Kv increase, with the remaining ∼8% increase due to increased ΔV. This is considerably larger than the ∼20% expected from the steady-state G_Kv calculated from the step activation experiments illustrated in Fig. 4A.

Thus, because the change in peak AP height observed in paxilline occurs over a potential range where the G_Kv is steeply voltage dependent, the enhanced activation of I_Kv compensates for the loss of I_BK as a repolarizing mechanism.

Ca²⁺ currents activated by voltage waveforms patterned on APs

AP1 and AP2 were applied in the presence of K⁺ channel blockers to characterize the waveform of I_Cav underlying the AP upstroke and to investigate whether the increase in AP height seen in paxilline was associated with changes in Ca²⁺ influx. AP1 and AP2 evoked I_Cavs with multiple peaks (Figs 6A and B, and 7) that had similar maximum amplitudes. I_Cav activated by AP1 had a broad staircase-like peak; AP2-evoked I_Cav had twin peaks separated by a prominent trough. AP2 elicited more Ca²⁺ influx, calculated as Q_Cav, the time integral of I_Cav (Fig. 6C and D). In paired recordings, AP2 significantly increased Q_Cav (11.91 ± 4.29%; n = 4; P = 0.037). Additionally, the traces in Fig. 6C and D show that a substantial fraction of I_Cav activated during the slow depolarisation preceding the AP threshold. This slow I_Cav component carried 27.03 ± 5.64% and 33.78 ± 6.02% of the total Q_Cav activated by AP1 and AP2, respectively (n = 4). Thus, inhibition of I_BK enhances Ca²⁺ influx per AP, which may subsequently raise [Ca²⁺]_i and potentiate insulin exocytosis.

Figure 7. Temporal alignment of membrane voltage and normalized currents activated by AP1 (A) and AP2 (B).

Colour coding for the I_K traces is as in Fig. 5. In both panels, black traces marked V (AP1) and V (AP2) represent membrane voltage and green traces marked Cav represent I_Cav. Dashed lines show zero level for I_Cav and maximum level for I_Kv, I_BK and membrane voltage. The calibration bar refers to both panels.

Figure 6E shows a set of I_Cavs activated by the IV1 step protocol recorded in the same cell as panels A and C. I_Cavs activated by IV1 were biophysically similar to those previously reported for β-cells (Satin & Cook, 1985; Rorsman & Trube, 1986). Figure 6F compares trajectory of I_Cavs evoked by AP1 and AP2 plotted as a function of voltage to the I–V relation of peak I_Cavs obtained from the IV1 protocol. I_Cavs activated by AP commands show more complex time- and voltage-dependence than expected from the I–V relation. In all four cells examined, both AP1 and AP2 evoked I_Cavs that displayed prominent hysteresis; thus, at any membrane potential, more I_Cav was activated on the AP downstroke than the upstroke.

BK channel block enhances glucose-stimulated insulin secretion in islets

As I_BK enhanced Ca²⁺ entry during APs, we determined whether paxilline enhanced GSIS. Basal insulin secretion rate, measured in 2.8 mm glucose, was not significantly affected by paxilline (0.028 ± 0.005 and 0.032 ± 0.006 ng (100 islets)⁻¹ min⁻¹ in the absence and presence of 1 μm paxilline, respectively; P > 0.05). In contrast, in the presence of a stimulatory glucose concentration (11.1 mm), 1 μm paxilline significantly enhanced insulin secretion by approximately 67%, from 0.576 ± 0.110 to 0.960 ± 0.211 ng (100 islets)⁻¹ min⁻¹ (P = 0.0148).

Discussion

We report that BK channels activate under physiological conditions and regulate AP amplitude, Ca²⁺ entry and GSIS in mouse pancreatic β-cells. I_BK activates over a similar membrane potential range as I_Kv; however, it activates more rapidly and inactivates partially; thus, it carries most of the outward current during the initial phases of the AP. Compared to previous studies (Smith et al. 1990b; Kukuljan et al. 1991), our results suggest a more important role for I_BK, and a reappraisal of the role of I_Kv, in β-cell function, and posit that I_BK serves a similar role in modulating AP shape and GSIS in mouse as it does in human β-cells (Braun et al. 2008).

While the I_BKs in the present study are kinetically similar to previous reports (Smith et al. 1990b; Kukuljan et al. 1991), our observation that I_BK constituted >50% and >85% of outward I_K activated by P1 and AP1, does not support the previous conclusion that BK channels play at most a minor role in AP regulation (Smith et al. 1990b). Similarly, our observation that paxilline enhanced AP height by ∼21 mV contrasts with the report that ChTx did not affect glucose-stimulated electrical activity in individual β-cells and whole islets (Kukuljan et al. 1991).

I_BKs and APs recorded in the present study are somewhat larger than reported previously (Smith et al. 1990b; Kukuljan et al. 1991). Mouse strain-dependent differences in BK- and Ca_v-channel expression, or the elevated recording temperature and intracellular milieu-preserving perforated patch technique, may have facilitated I_Cav and contributed to enhanced I_BK amplitudes in the present study (Fig. 2E). In addition, use of ChTx by Kukuljan et al. (1991) may have led these authors to underestimate the amplitude and role of I_BK. ChTx and IbTx are large charged peptides whereas paxilline is a small lipid-soluble molecule. Therefore, paxilline may have accessed the islet interior better and exerted a bigger effect on APs and GSIS, as has been reported for other tissues (Hu et al. 2001; Imlach et al. 2010). Moreover, use of ChTx and IbTx in previous studies may have led to underestimation of I_BK, as we found that IbTx only partially blocked the paxilline-sensitive I_BK in mouse β-cells. Parenthetically, the variable (22% to 81%) IbTx-mediated blockade of β-cell I_BK, as well as BK inactivation, may reflect the heterogeneous stoichiometry of β-subunits of the underlying BK channels. Inclusion of β1-, β2- and β4-subunits have been shown to reduce ChTx/IbTx potency by ∼11-, ∼50- and >1000-fold, respectively (Wallner et al. 1999; Lippiat et al. 2003), whereas paxilline appears able to equally block all known BK channel stoichiometries (Hu et al. 2001; Imlach et al. 2010). BK channels containing β2- or β3-subunits also inactivate rapidly (Xia et al. 1999, 2003). Therefore, if a β-cell expresses multiple β-subunits, the resulting macroscopic I_BK may exhibit both inactivating and IbTx-insensitive components (Ding et al. 1998). The BK channel β-subunit expression profile of mouse primary β-cells is unknown; however, an insulinoma expresses mRNA of the β3 subunit (Xia et al. 1999), and human β-cells express mRNAs for both β2 and β3 (Braun et al. 2008). In summary, quantitative technical issues may help account for the discrepancies between earlier studies and the conclusions of the present study.

The multiple peaks in I_Cavs activated by AP commands (Figs 6 and 7) are qualitatively similar to those reported in an insulinoma cell line (Li et al. 1999). Differences in activation and deactivation rates of the underlying Ca_v channel subtypes may have caused time-dependent changes in the composition of the macroscopic I_Cavs, resulting in the hysteresis shown in Fig. 6F (Chan et al. 2005).

BK channel block increased Ca²⁺ influx into the β-cells by ∼12% and enhanced GSIS by ∼67%. A single AP may raise [Ca²⁺] at the secretory apparatus sufficiently to trigger exocytosis (Ammala et al. 1993). Insulin secretion is steeply dependent on [Ca²⁺]_i in the physiological range (Hill coefficient = 3.4; Renstrom et al. 1997). This is also reflected in the observation that secretion rate increased 3-fold from −20 to 0 mV (Gopel et al. 2004), a similar voltage range over which the AP peaks are enhanced by paxilline. Thus, the increase in Q_Cav could in principle account for the ∼67% increase in GSIS associated with I_BK block. Moreover, the substantial Ca²⁺ influx that precedes AP threshold may stimulate insulin vesicle priming or exocytosis from the highly Ca²⁺-sensitive pool (Gromada et al. 1999; Neher & Sakaba, 2008).

The results of the present study suggest a modification of the current model of AP generation in the β-cell. Figure 7A shows normalized I_BK, I_Kv and I_Cav, activated by AP1, averaged from four to five β-cells and aligned with voltage trace. Figure 7B shows the corresponding I_Kv and I_Cav activated by AP2 and aligned with voltage trace.

In 11.1 mm glucose, a depolarisation beyond the AP threshold triggers a sharp increase in I_Cav that depolarizes the membrane rapidly. The depolarisation and Ca²⁺ influx activate I_BK, beginning ∼4 ms after the AP threshold and coinciding with the last ∼3 ms of the AP upstroke (Fig. 7A). I_BK activation slows the depolarization rate and truncates the AP. The slower I_Kv activates with a further ∼2 ms delay, coinciding with the last ∼1 ms of the AP upstroke. As I_Kv is slower and 7-fold smaller than I_BK (Fig. 5D), it minimally affects the AP upstroke and peak amplitude. I_BK and I_Kv peak ∼3 ms after the AP peak and underlie the downstroke. The larger I_BK dominates early repolarisation; however, I_BK turns off faster than I_Kv, which is consistent with the latter contributing preferentially to the later stages of downstroke and the AHP.

In paxilline (Fig. 7B), I_BK block eliminates a major part of the early repolarising drive, allowing I_Cav to depolarise the membrane further, and resulting in a taller AP. However, increased depolarisation increases G_Kv and K⁺ driving force, and accelerates I_Kv activation, resulting in a 4-fold increase in peak I_Kv than that activated by AP1 command. Along with I_Cav inactivation and decreased Ca²⁺ driving force (Fig. 6E), this larger I_Kv serves to repolarise the AP and underlies the AHP.

Our results and model for the distinct roles of I_BK and I_Kv in AP repolarisation and AHP generation are consistent with previous studies of I_K in β-cells. Thus, while the I_Kv in β-cells is mostly carried by K_v2.1 channels (Roe et al. 1996; MacDonald et al. 2002; Herrington et al. 2006; Herrington, 2007; Jacobson et al. 2007) absence of this current does not prevent AP repolarisation (Jacobson et al. 2007). Instead, a specific K_v2.1 blocker broadens (>2-fold) the AP and inhibits the AHP, while only marginally affecting the AP amplitude (Herrington et al. 2006; Herrington, 2007). Similarly, genetic ablation of K_v2.1 reveals a transient outward current that is remarkably similar in its amplitude and kinetics to the I_BK described in the present study (see Figs 2 and 3 of Jacobson et al. (2007)). Thus, it is conceivable that the remaining outward current in K_v2.1-knockout β-cells is carried by BK type channels. This remains to be determined in future studies.

In conclusion, this study demonstrates that BK channels in the mouse β-cell activate under physiological conditions, and modulate glucose-dependent electrical activity and insulin secretion, raising the possibility that inherited BK channel defects (Du et al. 2005) may have metabolic consequences. As BK channels are modulated by multiple signalling mechanisms (Hou et al. 2009), this mechanism has the potential to dynamically integrate, on an AP-by-AP basis, multiple cellular signalling pathways, such as those triggered by neurotransmitters or incretin hormones, to modulate GSIS by up to 67%. In addition, the wide variety of drugs that modulate BK channel function (Nardi et al. 2003; Nardi & Olesen, 2008) could provide a novel approach for modulating GSIS in diabetic and hyperinsulinaemic patients.

Glossary

Abbreviations

AHP

afterhyperpolarisation

AP

action potential

BK

large-conductance Ca²⁺-activated K⁺

ChTx

charybdotoxin

G_Kv

voltage-activated K⁺ conductance

GSIS

glucose-stimulated insulin secretion

I_BK

large-conductance Ca²⁺-activated K⁺ current

IbTx

iberiotoxin

I_Cav

voltage-activated Ca²⁺ current

I_K

total outward K⁺ current

I_Kv

voltage-activated K⁺ current

Q_Cav

voltage-activated Ca²⁺ charge influx

RIA

radioimmunoassay

R_s

series resistance

Author contributions

All authors contributed to the conception, design, analysis and/or interpretation of the data presented in this paper. The electrophysiological experiments were designed by K.M.H. and L.S.S. and carried out by K.M.H. Analysis of the electrophysiological experiments was done by K.M.H. The insulin secretion experiment was designed by K.M.H., L.S.S. and I.R.S., and carried out and analysed by I.R.S. The paper was written by K.M.H., initially in collaboration with L.S.S., and revised with the critical appraisal of I.R.S. All authors approved the final version of the paper for publication.