Synergistic activation of G- and L-type Ca²⁺ channels by D-glucose and membrane depolarization in pancreatic B-cells

 

Walters, Rhodri J.

 

Department of Pharmacology, Catholic University of Louvain, B-1200, Brussels

 

Changes in free intracellular Ca²⁺ couple changes in D-glucose concentration to insulin release within the pancreatic b-cell, although the specific role of Ca²⁺ channel subtypes in the regulation of insulin secretion is unclear. Under pseudo-physiological conditions individual b-cells and small b-cell clusters responded to membrane depolarization with the appearance of two populations of regenerative Ca²⁺-dependent current deflections at elevated D-glucose concentrations, an activation that was synergistic in nature.  The first population appeared as a train of repetitive current deflections which were abolished upon incubation with TEA⁺ or the L-type inhibitor nifedipine.  A second ‘G-type’ Ca²⁺ current deflection, of shorter latency and higher threshold, was also activated co-dependently by membrane depolarization and D-glucose.  The kinetic properties of the L- and G-type Ca²⁺ channels allude to distinct roles in the regulation of insulin secretion.

Introduction

Much controversy remains regarding the nature, number and distribution of the calcium channels present in the pancreatic b-cell.  An understanding of their distribution, regulation and voltage-dependence is of critical importance in developing therapeutic strategies towards the treatment of type I and II insulin-dependent diabetes.  Whilst there may be many other modulators and effectors governing insulin release (1), the concentration of intracellular free calcium is believed to be the major determinant (2, 3), and therefore a detailed characterization of the properties and regulation of the different populations of calcium channels present within the b-cell is of primary importance with respect to drug design.

The central involvement of a voltage-dependent L-type Ca²⁺ channel in the regulation of insulin release has been accepted for some time (4). Further, L-type currents have been shown to be augmented by glucose in the presence of the L-type activator Bay K8644 and K⁺ channel inhibitors (5), and their defective regulation by glucose in a rat model of diabetes has been demonstrated (6).  In addition L-type channels have previously been shown to co-localize with insulin secretory vesicles (7), strong evidence for a central role in governing insulin secretion.  However, fundamental discrepancies exist between observations of the kinetic behavior of currents elicited from b-cells in response to glucose in whole islets and cultured single-cells.  D-glucose elicits regenerative action potentials in islets at a steady state plateau potential (8, 9), but currents routinely recorded from single isolated b-cells until now have been transient in nature (5, 7, 10, 11, 12).  What then happens to these regenerative currents in patch-clamp recordings from single-cells, and what experimental procedures may account for their disappearance?

Another apparent caveat is the absence of a temporal correlation between the high frequency of L-type action potentials recorded from intact islets and the frequency of pulsatile insulin release (circa 0.47 events/min) observed both in vivo (13) and in isolated islets with either stable or oscillatory cytoplasmic Ca²⁺ concentrations (14, 15, 16).  Such a low frequency would suggest that L-type currents alone may not fully explain the central phenomenon of pulsatile insulin release.  Might another class of Ca²⁺ channel be present within the b-cell membrane with properties that could account for the pulsatile nature of insulin secretion?

Free Ca²⁺ activity is known to oscillate in temporal synchrony with the membrane potential of pancreatic islets in response to glucose (17, 18, 19).  Prior work has suggested a linear correlation between insulin release and both free ATP levels (18) and intracellular Ca²⁺ activity in the b-cell (2).  However, the observation that glucose strongly enhances insulin release from islets whose membrane potential has already been effectively ‘clamped’ at depolarized potentials by high external K⁺ in the presence of the K_ATP channel opener diazoxide (20, 21) is apparently at odds with the observation that it is only the initial rise in free Ca²⁺ which is modestly enhanced by glucose in measurements employing the ratiometric Ca²⁺ indicator fura-II (18).

Intriguingly, small clusters of pancreatic b-cells do not exhibit slow oscillations in membrane potential in response to either saturating or intermediate glucose concentrations (22), yet otherwise respond to increases in D-glucose with action potentials and conductance changes which are reminiscent of those seen in membrane potential recordings from intact islets (23, 24).  Thus we have employed small clusters of b-cells as a model system through which to study the regulation of calcium channels by glucose, minimising the complications of global oscillatory changes in input and gap junctional resistance.  The perforated-patch clamp technique has been used in conjunction with voltage-step and paired-pulse protocols, ion substitutions and inhibitors to identify two kinetically distinct calcium channel populations within murine pancreatic b-cells, whose properties are suggestive of specialized roles in the regulation of insulin secretion.

Results

D-Glucose activates voltage-dependent current deflections in small clusters of islet cells

To separate the constituent voltage-dependent membrane conductances of pancreatic islet cells, a simplified linear voltage step protocol was applied to small clusters of islet cells from a holding potential of –70mV (the ‘IK1’ protocol).  This formed the basis of a current ‘dissection’ protocol to elucidate the nature and mechanism of regulation of the voltage-gated currents present within the pancreatic b-cell.  After a prolonged 40-60 min pre-incubation in Krebs media containing 1 or 3mM D-glucose (G1-G3), only outward currents were observed with pseudo-physiological ion concentrations using the perforated-patch recording technique (25, Fig.1).  However, after a 2 min pre-incubation with 20mM glucose, small clusters of islet cells maintained at 30-30°C responded to membrane depolarization under the IK1 protocol with one of three representative patterns.  'Type I' cell clusters responded to membrane depolarization following a 2 min incubation in 20mM glucose (G20) with a train of regenerative  inward current deflections that were concomitant with a suppression of the time-independent component of the outward current (Fig.1).  'Type II' cells in contrast typically responded to G20 with only a single 'fast' inward current deflection, responding to depolarization in G3 with a train of smaller, higher frequency inward current deflections.  A third class of small cell clusters, termed 'type III' showed no evidence of current deflections in response to increases in D-glucose concentration.

Figure 1. A comparison of type I and type II cell responses to changes in glucose concentration in small clusters of islet cells.

A. Upper panels show the current responses elicited by a 300ms depolarizing step to -28mV from a holding potential of -70mV for a type I cell cluster (left) and a type II cell cluster (right) after pre-incubation in G3 (upper panels) and from a subsequent IK1 protocol step to -19mV after a 2 minute incubation in G20 (lower panels).  In the simplified IK1 protocol, 300ms depolarizing steps were applied in 3mV increments from a holding potential of –70mV with a 2s interpulse interval. All recordings were obtained at 30-33°C.  Holding currents were typically between -5 and -25pA after correction for offsets. Current and time scale bars are representative of all recordings illustrated.  The Cs150 pipette solution was used.

B. Distribution histogram of the mean recurrent deflection amplitudes elicited by a V_step to -28mV from a holding potential of –70mV derived from both small cluster recordings  (hollow circles) and from four identified single-cells and one cell pair (filled circles). 

A representative comparison of type I and type II cell responses to voltage steps is shown in Figure 1A.  Type I clusters exhibited a large outward current with a chord conductance of between 0.5 and 2nS whether 150mM K⁺ or 150mM Cs⁺ was used as the intracellular cation, although the outward current showed a substantial time-dependent activating component above -20 mV with K⁺ as the charge carrier.  The time-independent component of the outward current present in G3 was reversibly inhibited after a 2 min incubation with either 20mM glucose (G20) or tolbutamide (100mM, not shown), and also by bath application of 10mM TEA⁺ or Ba²⁺, consistent with the known pharmacology of the K_ATP conductance (26, 27).  However, equimolar replacement of intracellular K⁺ with Cs⁺ did not appear to suppress the K_ATP conductance or alter the kinetics of inward current deflections activated by G20 (Fig.2B).

Following a 2-5 min pre-incubation in G20, 'type I' clusters responded with a regenerative train of current deflections of 10-15 ms duration whose threshold, latency and deflection probability all appeared strongly dependent upon both step potential and the ambient glucose concentration.  Activation of these recurrent current deflections was completely reversible within 12-16 min of returning to G3, and the threshold for these recurrent deflections was concentration dependent, with a mean threshold of -47.2 ± 1.5 mV (n=30) after 2 min incubation in G20, and -26.5 (± 1.5) mV after 2 min incubation in G5 (n=4, P<0.005 ANOVA), suggesting that threshold potential for deflection activation exhibits a dose-dependent modulation in the range of glucose concentrations G3-G20 as deflections were rarely, if ever seen in clusters after prolonged pre-incubation in G1 or G3.  Further, these recurrent deflections showed no evidence of accommodation during a 60s step depolarization to –30mV (n=2, not shown), and were of lower threshold a second transient voltage-activated deflection (-35.5 ± 2.3 mV (n=6) G20 + 5min) which could be readily distinguished by its shorter latency and greater amplitude (Fig.1A).  In contrast, type II cell clusters exhibited a train of fast regenerative current deflections upon depolarization in G3, with a mean threshold of -24 ± 6 mV (n=10), responding to G20 with a larger deflection of short latency.  The nature and identity of these type II clusters is not considered further.  

Of all small clusters tested, 76% (175/230) responded to G20 with type I kinetics, 10% (23/230) with a type II pattern and 14% (32/230) showed no current deflections at any concentration of glucose tested (type III), a proportionality approaching that classically described within pancreatic islets by Hedeskov (28). The overwhelming predominance of type I clusters, the activation of current deflections in these clusters by G20 with the concomitant suppression of a time-independent and tolbutamide-sensitive outward current, and further kinetic and pharmacological considerations described below, strongly support the contention that these responses representative those of pancreatic b-cells.

Glucose-activated current deflections exhibit strongly voltage-dependent kinetics

The threshold, frequency, amplitude and latency of the glucose-activated recurrent inward deflections, typically of 10-15 ms in duration, appeared to be strongly influenced by the step potential.  Figure 2A shows how the frequency of these regenerative inward current deflections increases as a function of step potential, the sensitivity being steepest between -50 and -20mV, with a half-maximal frequency elicited at -43mV.  These recurrent deflections show no requirement for membrane repolarization for reactivation, indicating the involvement of other regulatory factors in the cyclical inactivation and reactivation that underlies their oscillatory behavior (29, 30). The mean activation threshold derived for these recurrent deflections was -47.2 ± 1.5 mV (G20 + 2 min, n=30) is consistent with intracellular recordings from micro-dissected whole islets (31).  The absence of any evidence for distant 'unclamped' current deflections, or ‘echoes’, which might be expected if unclamped neighboring cells were being depolarized to threshold by the metabolic action of D-glucose is consistent with the argument that an effective spatial voltage-clamp had been obtained.

The latency between the onset of depolarization and the appearance of the first, second and third recurrent deflections is shown plotted as a function of step potential (Fig.2D).  Evidently, the rate of activation of the intrinsic voltage sensor within the channel complex is strongly voltage dependent and can be described by a fit to the Boltzmann equation.  The fits to the 1^st, 2^nd and 3^rd latencies also shows how the inter-deflection interval decreases as a function of voltage. There is no evidence to suggest that the inward current deflections incorporate a substantial K⁺ component that masks or distorts their true kinetics, as (a) there was little current ‘overshoot’ from baseline following the inactivation phase of the current deflection; (b) the current deflections were fast inactivating, minimizing attenuation of current amplitude through any co-localized Ca²⁺-activated K⁺ channels, and (c) the kinetics of the spikes were not substantially affected by intracellular Cs⁺ (Fig.2B),  an inhibitor of voltage-, but not Ca²⁺-activated K⁺ currents (32).

  

Figure 2.  The activation of recurrent deflections by glucose is voltage-dependent.

 (A) The frequency of inward current deflections activated by glucose is strongly voltage-dependent. Upper left panel. Currents elicited by a 300ms depolarizing step (V_step) from a V_hold of -70mV to -55mV (threshold) in a small b-cell cluster perfused with G20 Krebs for 5 min prior to commencement of the IK1 pulse protocol. Lower left panel. Currents elicited from the same cluster by subsequent V_steps to -40mV and (upper right panel) -22mV within the same IK1 sequence.  The pipette solution was K150.

(B) Glucose-elicited inward current deflections are not apparently modulated upon intracellular dialysis with Cs⁺. Voltage step to -46mV in another small b-cell containing cluster with Cs⁺ (Cs150) as the charge carrier. Holding currents recorded were typically of between -5 and -25pA after correction for offsets. Current and time scale bars are given for all recordings illustrated. All clusters were perfused with a Krebs HCO₃^--buffered Ringer gassed with 94%O₂/6% CO₂ at 30-33°C throughout all recordings.  The pipette solution used was Cs150. All illustrated traces have been leak subtracted. 

(C) The frequency of inward current deflections evoked by G20 is steeply voltage-dependent.  Mean frequency of inward current deflections plotted as a function of V_step for regenerative inward current deflections.  Frequency values are presented as means ± SE from 7 separate b-cell clusters. The fit is to the Boltzmann equation.

(D) The latencies of inward current deflections are voltage-dependent.  The interval (latency) between the onset of the depolarization step from a holding potential of -70mV and the appearance of the 1st, 2nd and 3rd regenerative current deflections are presented as a function of step potential.  The first latency (from the commencement of the voltage step to the point of deflection from baseline) and the second and third latencies (defined as the interval between successive returns and departures from baseline) were determined for each step potential tested. Values are presented as means ± S.E. from 15 cell clusters. The pipette solution used in (C) and (D) was K150. Fits are to the Boltzmann equation.

Repetitive inward current deflections are carried by Ca²⁺ ions

The ionic selectivity of these glucose-activated recurrent deflections should be readily apparent from shifts in the amplitude of these current deflections upon ionic replacement.  Figure 3A shows representative recordings from a small b-cell cluster where the external Na⁺ was reduced from 144 to 59 mM by equimolar substitution of NaCl with NMGCl (n=4), or else external Ca²⁺ was reduced from 2.5 to 0.25mM by equimolar replacement of CaCl₂ with MgCl₂ (n=6).  The amplitudes of these recurrent deflections were shown to be reversibly diminished by Ca²⁺, but not Na⁺ replacement.  This clearly indicates that Ca²⁺ is the dominant permeant cation through this conductance pathway with pseudo-physiological gradients. Electrophysiological experiments were also performed in parallel with measurements of free Ca²⁺ from small clusters that had been loaded with the ratiometric indicator Fura II (see methods).  However surprisingly this resulted in an inhibition of the recurrent deflections. 

Figure 3. The glucose-activated recurrent deflections are carried by Ca²⁺   

(A) Upper left panel. Current responses elicited from a small voltage-clamped b-cell cluster bathed in Krebs medium containing 144mM Na⁺ and 2.5mM Ca²⁺ in response to a 300ms depolarizing step to -46mV from a holding potential of -70mV after pre-incubation in G20 for 2 min. Lower left-hand panel.  Currents recorded from the same cluster in response to an isopotential sweep 2 min after the external Ca²⁺ had been reduced to 0.25mM by equimolar substitution with Mg²⁺ and then (Upper right panel) 2 min after restoring external Ca²⁺ to 2.5mM.  All currents illustrated were elicited by a V_step to -46 mV. (Lower right panel)  Absence of a change in amplitude or duration of recurrent deflections upon equimolar replacement of extracellular Na⁺ from 144 to 59mM with the impermeant cation NMG⁺.  Clusters were perfused throughout all recordings with a Krebs HCO₃^--buffered Ringer gassed with 94%O₂:6% CO₂ at 30-33°C. Holding currents were typically between -5 and -25pA after correction for offsets. Current and time scale bars are given for all recordings illustrated. The pipette solution used was Cs150. All traces illustrated have been leak subtracted.

(B)  Permeability ratios of peak deflection amplitudes of regenerative current deflections following cationic replacement.  Summary of all ion replacement experiments where extracellular Na⁺ was reduced from 144mM to 59mM (filled squares, n=4), or else the external Ca²⁺ was reduced to 0.25mM from 2.5mM  (filled circles, n=6).  Data are presented as the ratios of paired peak deflection amplitudes, wherein the ratio of the peak deflection amplitude in the reduced ionic environment was divided by its peak amplitude in the control ionic environment. Ratios are presented as means (± SE) for the number of separate experiments indicated.

(C)  The peak conductance of regenerative current deflections is modulated both as a function of step potential and external Ca²⁺ concentration.  The peak deflection current was plotted as a function of potential for both ‘high’ (2.5mM) and ‘low’ extracellular Ca²⁺ (0.25mM). The mean deflection (base-line to peak) amplitude was calculated for L-type deflections at each step potential tested, and the resulting mean deflection amplitude was plotted as a function of the calculated electrochemical gradient for recordings from small b-cell clusters incubated in either low (n=6) or high (n=18) external Ca²⁺ assuming an intracellular free Ca²⁺ activity of 200nM (from (17), n=6).

(D) Reducing extracellular Ca²⁺ modulates the frequency of recurrent inward deflections at depolarized potentials. Deflection frequencies are presented as a function of step potential under control conditions (2.5 mM Ca²⁺, n=6, hollow circles), or where either extracellular Na⁺ was reduced from 144 to 59mM by equimolar replacement with N-Methyl-D-Glucamine⁺ in G20 Krebs (n=4, hollow squares), 100nM TTX was added (n=6, filled squares), or else external Ca²⁺ was reduced from 2.5mM to 0.25mM (n=6, filled circles).

Kinetic analysis of the recurrent deflections

Current deflections permit an isolated channel population to be studied within a population of macroscopic currents without the necessity for the use of unphysiological ionic gradients or inhibitors.  Peak amplitudes of current deflections were measured across a range of step potentials to yield macroscopic conductance values.  As the peak current deflection amplitude (I) will be given by n.P_o.i, where n is the number of functional channels, i the unitary conductance, and P_o is the mean open probability of a given channel within the macroscopic population.  However, as exocytotic and endocytotic processes are known to occur sequentially within the b-cell after depolarization at higher glucose concentrations (7, 11), none of these parameters can be held to be constant.

An example of such an analysis is shown in Figure 3C, where the relationship between mean current deflection amplitude and step potential is presented as a function of the electromotive force for Ca²⁺ entry calculated according to the Nernst relation (assuming a perfectly selective Ca²⁺ conductance). Figure 3C reveals that the regenerative current deflections in b-cell clusters exhibit a biphasic slope conductance in 2.5mM, but not 0.25mM Ca²⁺ across an equivalent range of step potentials. Between threshold and -30mV the mean deflection amplitude unexpectedly decreases as the electrochemical gradient for Ca²⁺ increases (mean slope conductance of -0.55nS), whereas at more depolarized step potentials the peak deflection current amplitude decreases as a function of the electromotive force for Ca²⁺ entry (0.18 nS), suggesting that either channel mean open probability and/or unitary conductance are modulated as a function of step potential.

The recurrent deflections are also clearly calcium-dependent in duration, their duration increasing as the external Ca²⁺ concentration is reduced, indicative of a Ca²⁺-dependent feedback mechanism in channel gating (29, 30). Changing the extracellular Ca²⁺ concentration thus alters a range of channel properties underlying this recurrent deflection, including the slope of the macroscopic current-voltage relation and the relationship between deflection frequency and step potential (Figure 3D), as the deflection frequency first increases, and then decreases with membrane depolarization.

The demonstration that a ten-fold reduction in external calcium to 0.25mM reduces deflection amplitude by only 40% is perhaps not surprising, as the estimated E_Ca calculated according to the Nernst relation changes from +137 to +107mV (see methods). Moreover there is the suggestion of a voltage-dependent component in the apparent permeability ratio, as the 0.25/2.5mM deflection amplitude ratio decreases from 60% to only 40% as the potential changes from -50 to -25mV.

Effects of Ca²⁺ channel inhibitors and chelators upon the recurrent deflections

The variations in kinetic forms of the current deflections observed indicate that there might be more than one voltage-dependent current activated by glucose.  All current deflections were inhibited by 100mM Cd²⁺ and 1mM Co²⁺ (n=2) and also by the Ca²⁺ channel antagonist D-600 (100mM) in a voltage-dependent and reversible manner (n=4, not shown).  Further evidence that the recurrent deflections represent the opening of Ca²⁺ channels is given by the observation that the voltage-gated Na⁺ channel inhibitor tetrodotoxin (TTX) at 100nM has no effect upon the frequency (Fig.3D), amplitude or threshold. Neither the addition of 100nM TTX (n=6), nor a reduction of external Ca²⁺ to 0.25mM (n=6), nor the acute addition of 10mM TEA⁺ affected the mean threshold of these regenerative current deflections in response to G20.  Surprisingly, these regenerative current deflections were also abolished following a 60 minute preincubation with BAPTA-AM (5-50mM) or Fura-II-AM (1mM), yet were only partly diminished after 15 min incubation in 5mM BAPTA-AM and were apparently unaffected following a 30 minute pre-incubation with 1mM BAPTA-AM.  These data suggest that this channel activity is acutely sensitive to the degree of calcium buffering in the cell, possibly due to a calcium-dependent priming mechanism (30) or change in phosphorylation state.

Nifedipine and phentolamine discriminate between two populations of glucose-activated Ca²⁺ currents

Figure 4A shows pairs of current traces recorded from a b-cell cluster held at threshold (-49mV in this cluster) and subsequently at -10mV after 2 min pre-incubation in G20, before and after the addition of 10mM nifedipine to the bath. From 6 experiments it was clear that although the recurrent inward deflections were reversibly abolished by 10mM, but not 1mM, nifedipine, a transient inward current deflection persisted in the presence of nifedipine.  By virtue of their kinetic resemblance to the L-type spike trains evoked by D-glucose in whole islets, and their sensitivity to dihydropyridines, these recurrent Ca²⁺ current deflections were designated L(G)-type (glucose-activated L-type Ca²⁺ current).

The high threshold current deflection activated by G20 could be readily discriminated from the L(G)-type deflection population by its shorter latency, higher threshold, complete inactivation and greater amplitude. This deflection could also be isolated after a 2min pre-incubation with 1-10mM phentolamine (n=4, Fig.7B). 

Figure 4.  Pharmacology reveals two distinct glucose-activated Ca²⁺ currents

(A) Nifedipine inhibits the recurrent, but not the high threshold, glucose-activated current deflection in small b-cell clusters. Paired recordings of deflections elicited after a 2min pre-incubation in G20 using the IK1 protocol to threshold (-49mV, left-hand trace pairing) and subsequently to -10mV (right-hand trace pairing), both before (lower trace) and after (upper trace) the addition of 10mM nifedipine to the bathing medium (n=6). All clusters were perfused with a Krebs HCO₃^--buffered Ringer gassed with 94%O₂:6% CO₂ at 30-33°C throughout all recordings.  Holding currents were typically between -5 and -25pA after correction for offsets. Traces are shown after leak current subtraction. Current and time scale bars are representative of all recordings illustrated. The pipette solution used was Cs150.

(B) G-type current deflections are carried by Ca²⁺ ions. G-type deflection amplitudes are presented as ratios of peak amplitudes before and after cation replacement. Ion replacement experiments were performed, wherein either extracellular Na⁺ was reduced from 144mM to 59mM by equimolar substitution with N-Methyl D-Glucamine (filled squares, n=4), or else when external Ca²⁺ was reduced to 0.25mM from 2.5mM by the equimolar substitution of MgCl₂ for CaCl₂ (filled circles, n=6). Ratios were calculated by dividing the isopotential peak deflection amplitude in the reduced ion environment by the paired amplitude in the control cationic environment. Data are presented as means (±  SE) for the number of experiments indicated.

(C) Effect of TTX upon G-type deflection amplitude. Ratios of peak G-type amplitude before and after the addition of 100nM TTX to the bathing medium plotted as a function of step potential. Data are presented as the ratios of paired deflection amplitudes, where the ratio of the peak deflection amplitude in TTX is divided by the paired control amplitude. Ratios are presented as means (± SE) from 3 separate experiments.

(D) The macroscopic G-type current deflection shows a sigmoidal relationship between conductance and step potential.  The mean G-type deflection amplitude is presented as a function of step potential from 6 recordings.  The sigmoidal fit is to Boltzmann.

Kinetic properties of the G-type current

The high threshold, nifedipine-resistant G-type current deflection was ‘robust’, in that it showed no evidence of ‘run-down’ in its amplitude over 10 min of recording in G20, and its amplitude could be fully restored upon repolarization to -70mV after the step depolarization.  This novel current was designated as the ‘G-type’ current deflection, by virtue of its properties as a voltage- and glucose-activated and dihydropyridine resistant Ca²⁺ current.

Analysis of the G-type current deflections revealed distinctive kinetic properties and a Ca²⁺ selectivity in the presence of 10mM nifedipine (Fig. 4B).  The threshold for G-type activation was significantly lower after 5 min pre-incubation in G20 (-35.5 ± 2.3 mV, n=6 in nifedipine) than after 2 min (-26.5 ± 5 mV), suggestive of a slower rate of activation than for the L(G)-type deflection.  The G-type current exhibited a maximal slope conductance of 2.6nS between -25 and -20 mV (Figure 4D) and a peak current amplitude at -10mV (thereafter the current declined, but measurement was complicated by the activation of an outward current).  These data are suggestive of a strong voltage-dependent rectification of the G-type current, the macroscopic current increasing with a decreasing electrochemical gradient for Ca²⁺ entry.  Further evidence for Ca²⁺ selectivity is given by the temporal coincidence  of the inward current deflection and elevations of intracellular free Ca²⁺ concentration with Fura-II as observed before (7). However, despite its evident Ca²⁺ selectivity, the G-type current was partially inhibited by 100nM TTX in a voltage-dependent manner between -19 and -25mV (n=3, Figure 4C) and abolished at 1mM TTX (n=2, not shown), although the threshold was affected neither by cation substitution, nor by the addition of 100nM TTX.

Glucose activation of G- and L(G)-type currents is concentration dependent

G-type deflections exhibited a time-dependent activation by G5, the probability of observing a current deflection increasing between 2 and 5 min after increasing glucose concentration from G3 to G5 (not shown). The frequency of the L(G)-type deflections showed a marked dependence upon glucose concentration as well as potential in the range -60 to -30mV (not shown).  The probability of observing an L(G)-type current deflection increased incrementally with both time and glucose concentration, appearing to reach a maximal deflection probability in G15.  In contrast, G-type deflection probability was maximal in G7 (not shown).  The response to G20 was neither mimicked nor inhibited by pre-incubation in G1 + 19mM 3-O-methyl glucose (n=3), suggesting that neither an osmotic nor an allosteric mechanism of action underlies the activation of these currents by glucose, rather that the current response arises due to a direct metabolic action of glucose.  This observation is further supported by the reversible inhibition of the recurrent L(G)-type deflections following pre-incubation in G20 by the addition of 2mM azide (not shown).  Therefore glucose and membrane depolarization constitute synergistic and interdependent facilitators of Ca²⁺ channel activation and hence of insulin secretion.

Deconstructing differences with previous findings

Smith and colleagues (5) classically reported an enhancement of the integral of the whole-cell current evoked by a step depolarization to 0mV, 6 min after increasing glucose from 0 to 20mM at room temperature, but as others observed no recurrent deflection patterns as reported here.  Although we were fully able to reproduce this observation under identical conditions to those published (Fig. 5A), both at 25°C and at 30-33°C, the augmentation of the inward current by glucose appeared to be concomitant with a general 'run-up' of inward current.  Further, pre-incubation with 10mM TEA⁺ appeared to have a biphasic action upon the inward 'calcium' current.  Following a 20 min pre-incubation with the G0 Smith external solution containing 10mM TEA⁺, the integral of the leak-subtracted inward current was observed first to 'run-up' and then to 'run-down' in the continued presence of TEA⁺ (n=4, not shown).  These observations are summarized in Figure 5. 

Figure 5. Tetraethylammonium is an inhibitor of the recurrent L-type deflections.

(A) Actions of glucose upon the whole-cell inward current after pre-incubation with TEA⁺. 

Left panel. Currents elicited by a 100ms voltage-step from a holding potential of -70mV to 0 mV in a small b-cell cluster pre-incubated for 20 min in the Smith extracellular medium containing 0 mM glucose (G0), 10mM TEA⁺ and 10mM HEPES at room temperature (20-21°C). The upper current trace shows the leak-subtracted current immediately upon addition of G20 to the medium (t=0), and the lower trace the isopotential current elicited after 6 min. Traces are shown after leak current subtraction for all recordings illustrated. The Smith pipette solution was used.

Right panel. Average of leak-subtracted inward currents from a voltage-step from -70 to 0 mV using a protocol identical to that shown opposite. After recording currents elicited by a continuous train of depolarizing pulses for 6 min in G0, the perfusing medium was first changed to one containing G20 for 6 min, and then finally to G0 for 8 min. Mean integrals of the leak-subtracted inward current were measured using Heka on-line analysis software and are plotted (± S.E.) as a function of time for 8 separate recordings.

(B) Prolonged incubation with TEA⁺ abolishes current deflections in the pancreatic b-cell. Current responses elicited by a sequential ‘three-step’ voltage pulse train (steps 300ms in duration) from a holding potential of -70mV first to -40mV (upper traces),  subsequently to -20mV (center traces) and finally to 0 mV (lower traces) in a small b-cell cluster perfused first for 6 min with G0 Smith external solution (without TEA⁺, traces in left column), then with G20 for 2 min (traces in center column) and finally with Smith external solution to which 10mM TEA⁺ was added (right-hand column). The Smith pipette solution was used.

(C) Acute exposure to TEA⁺ protracts the G20-activated current deflections.  Left panel. Paired recordings of currents elicited after pre-incubation for 4 min in G3 Krebs + 10mM TEA⁺ (upper trace),  and 2 min after application of G20 + 10mM TEA⁺ at 30-33°C (lower trace) after depolarization from a holding potential of -70mV to -22mV (n=5). Right panel. Recording elicited from the same small cluster in response to a step depolarization to -46mV. Holding currents were typically between -5 and -25pA after correction for offsets. Current and time scale bars are representative of all recordings illustrated. The pipette solution used was Cs150.

An augmentation upon incubation with G20 of the integral of the inward current elicited at 0 mV was observed after first pre-incubating small clusters for 20 min at room temperature in a glucose-free medium containing 10mM TEA⁺. However the inward current was either absent, or else was small and infrequently observed after 30 min pre-incubation in G0 in the absence of TEA⁺  (Fig.5B, left-hand column).  Thus in G0, as in G3 and G1, no sizeable inward current was evoked at 0mV, whether the hypotonic Smith pipette solution or the more isotonic Cs150 pipette solution was used.  Neither pipette solution prevented the appearance of current deflections upon incubation in G20. Figure 5B shows the current deflections elicited by a simplified voltage protocol stepping to -40, -20 and 0mV in a small b-cell cluster using the conditions and solutions employed by Smith et al. 1989, excepting for the omission of TEA⁺ from the bathing medium.  Within 2 min of addition of 10mM TEA⁺ to the G20 bathing solution, a complete abolition of recurrent inward deflections was observed, concomitant with the development of an inward current that resembled that observed upon prolonged pre-incubation with TEA⁺.

In summary, after recording from 29 small cell clusters, pre-incubation with TEA⁺ was found to be primary explanation for the difference in findings with Smith et al. (5), despite differences in the osmolarity of the pipette solutions, their ionic composition (see methods), the extracellular  buffer system employed or in the pre-incubation conditions.

Differences between current deflection patterns observed in single-cells and small clusters

TEA⁺ alone may be insufficient to explain all the differences in findings between our recordings from small voltage-clamped clusters and those previously observed in recordings from single-cells maintained in primary culture. Thus recordings were taken from acutely isolated single-cells within 8 hours of isolation.  Figure 6 illustrates the marked differences in the patterns of responses to G20 between acutely-isolated single-cells and small clusters.  Only 24% of 230 small clusters showed no type I current deflections in response to G20, whilst 53% (8/15) of acutely isolated single-cells exhibited no current deflections in response to G20.  Five single-cells however, did express a fast inactivating inward current deflection reminiscent of the G-type current observed in clusters.  However, in contrast to the G-type current observed in small clusters, this current was constitutively active in single-cells pre-incubated in G3, and in four of these cells it was further augmented by G20 (Figure 6D).  Moreover, 2 single-cells responded to G20 with the appearance of recurrent inward deflections alone reminiscent of the L(G)-type current expressed in clusters (Figure 5C), whilst two single-cells responded to G20 by exhibiting both current patterns (Figure 5A & 5B), and 3 other cells exhibited current deflections reminiscent of the G-type alone (not shown), indicating that L(G)- and G-type currents are not necessarily always co-expressed within the same cell, further supporting a distinct molecular identity.

Figure 6.  Acutely isolated single-cells express L(G)- and G-type currents independently.

(A) Cell 1. Paired recordings of current deflections elicited in an identified single-cell by 300ms depolarizing steps from -70mV to threshold (at -43mV, upper traces) and to -19mV (lower traces) in G3 (left-hand trace pair) and 5 min after the addition of G20 (right-hand trace pair).

(B) Cell 2. Paired recording of current deflections elicited by sequential 300ms depolarizing steps from -70mV to threshold (also -43mV, upper trace) and to -19mV (lower trace) from a second identified single-cell elicited after 5 min pre-incubation in G20.

(C) Cell 3. Recurrent deflections elicited by stepping to –34mV after 5 min in G20. Holding currents were typically between -5 and -25pA after correction for offsets.  All traces shown are leak-subtracted excepting Cell 3.  Current and time scale bars are representative of all recordings illustrated. The pipette solution used was Cs150.   All recordings were performed at 30-33°C.

(D) Transient inactivating currents are constitutively active in acutely-isolated single-cells in the presence of low glucose.

Upper left panel. Mean current integrals (± SE) of leak-subtracted currents obtained from 5 identified single-cells presented as a function of step potential after pre-incubation in G3 (hollow circles), and then 2 (hollow squares) and 5 (filled squares) min after increasing the extracellular glucose concentration from 3 to 20mM. Lower left panel. Means (± SE) of current integrals for the single-cells plotted after normalizing the current integral elicited at each step potential to the peak current integral for each recording in G20 (+5 min). Upper right panel. Mean integrals (± SE) of leak-subtracted inward currents plotted as a function of step potential for 11 separate experimental recordings from small cell clusters after pre-incubation in G3 (hollow circles), and 2 min after increasing the glucose concentration to G20 (filled circles). Lower right panel. Means (± SE) of current integrals for the small cell clusters normalized as for single-cells. All experiments were performed using the Cs150 pipette solution at 30-33°C.

Voltage-dependent reactivation properties of the G-type current revealed by paired-pulse protocols

Reactivation of the G-type current requires a repolarization step, after the classical kinetic properties of voltage-gated Na⁺ channels. Thus the voltage-dependent properties of the G-type Ca²⁺ current must consist not only of a defined threshold associated with an intramolecular voltage sensor, but also of a half-maximal repolarization potential for reactivation (designated V_0.5) suggestive of a second voltage sensitive gate whose kinetic parameters might be measured by means of paired voltage pulse protocols.  Figure 7 shows a paired-pulse protocol designed to determine these half-maximal repolarization potentials.  Employing a 300ms repolarization step, the V_0.5 for the G-type in small clusters using a paired-pulse protocol after an initial V_step to -30mV was -45.0 ± 0.9 mV, with a near complete recovery of peak current observed following a 300ms repolarization to -80mV.  The pattern of repolarization-dependent reactivation is more clearly demonstrated in the presence of 10mM phentolamine at which concentration the recurrent deflections are inhibited (Fig.7B).  Thus the extent of reactivation of the G-type current increases with the magnitude of the repolarization step, providing for a simple putative ‘on-off’ switch during the slow wave oscillations which occur in the islet in response to intermediate glucose concentrations.

Figure 7. G-type current deflections exhibit voltage-dependent reactivation kinetics

(A) Current traces elicited by a paired-pulse protocol (top left-hand panel) shown together with the corresponding voltage protocol (below). A 100ms step pulse (V_step) to -30mV is given from a holding potential of -70mV followed by an initial 300ms repolarization pulse to -80mV followed by a paired 100ms V_step to -30mV. The sequence is then ended by returning to -70mV for a 2 second interval before a subsequent paired-pulse was applied wherein the repolarization step interval was diminished by 5mV for each successive and otherwise identical paired-pulse sequence. Upper panel. Seven selected superimposed sweeps showing the decay in the amplitude of the 2nd deflection elicited by a paired step depolarization to -30mV concomitant with the diminution in magnitude of the repolarization step. The corresponding voltage protocol is given (lower panel).

Right-hand panel. Mean ratios of peak currents elicited plotted as a function of repolarization potential.  Ratios are determined by the dividing the peak amplitude of the current elicited by the 2^nd ‘paired’ depolarizing step of the paired-pulse protocol shown beneath, by the peak amplitude of the current elicited by the first control depolarization pulse.  Data are presented  (± SE) from 7 separate experiments.  Note that the repolarization pulse was of 300ms duration. Holding currents were typically between -5 and -25pA after correction for offsets. Traces are shown after leak current subtraction and current and time scale bars are as given. The pipette solution used was Cs150 and experiments were performed at 30-33° C.

(B) The kinetics of recovery of the G-type current are readily apparent in a paired-pulse protocol following a 2 min pre-incubation with 10mM phentolamine. A 100ms pulse (V_step) to -30mV from a holding potential of -70mV was given, followed by an initial 300ms repolarization to -80mV, followed by a paired V_step to -30mV. Eight selected sweeps have been superimposed showing the decrease in the amplitude of the 2nd current deflection elicited by a Vstep to -30mV that follows a progressive diminution of the repolarization potential from -65mV to -30mV in 5mV increments. The repolarization step was of 300ms duration. Holding currents were typically between -5 and -25pA after correction for offsets. Traces are shown after leak current subtraction and current and time scale bars are as given. The pipette solution used was Cs150 and all experiments were performed at 30-33°C.

The transient current deflection observed in acutely isolated single-cells is the G-type current

The transient inward current that is constitutively active in G3 within acutely-isolated single-cells closely resembles the G-type current activated in clusters by G20 by virtue of its high threshold and rapidly inactivating kinetics. This current was augmented only slightly by G20 in acutely isolated single cells after 5 min incubation (Fig.6D). Thus the paired-pulse protocol was used to determine whether its voltage-dependent reactivation properties were equivalent to those of the G-type current previously identified in clusters.  Figure 8 shows currents elicited by paired-pulse protocols applied both to an acutely isolated single-cell and a small cluster pre-incubated in G20 for 2 min with two different paired-pulse activation protocols which differ only in the magnitude of the paired depolarization step.  Aside from illustrating that the rapidly-inactivating current in single-cells exhibits reactivation kinetics that are indistinguishable from those of the G-type current observed in small clusters, this protocol is further used to demonstrate that the V_0.5 is also modulated as a function of the depolarization step potential (Figure 8B, central panel).

A summary of the data from the paired-pulse protocols is given in Figure 8B.  Note that the V_0.5 values obtained from isopotential steps in single-cells and small clusters are not significantly different.  The plot of V_0.5 as a function of step potential also indicates a steep voltage-dependence over the narrow physiological range of potentials over which the G-type current is active (-35 to 0 mV).

Figure 8. Clusters and single-cells both express G-type currents with voltage-dependent reactivation kinetics.

(A)  Left-hand panels.  Current deflections elicited by a paired-pulse protocol recording from an identified single-cell following a 5 min pre-incubation in G20. A 100ms pulse (V_step) to -20mV from a holding potential of -70mV was given, followed by an initial 500ms repolarization to -80mV and then a paired V_step to -20mV. The step sequence is then ended by returning to -70mV for a 2 s interval before subsequent paired-pulses were applied, wherein the magnitude of the step repolarization was diminished in 5mV increments from -80mV for each successive paired-pulse application to -20mV. Upper panel. Six selected superimposed sweeps are shown revealing the decay of the amplitude of the 2nd deflection elicited at -20mV upon diminution of the repolarization potential. The corresponding voltage pulses are given to the same time scale (central panel). Lower panel. Mean ratios of peak currents determined by dividing the amplitude of the 2^nd current deflection by that of the 1^st for each paired step potential (± SE), and plotting the ratio as a function of repolarization potential for 4 single-cell experiments using the GPP-20 protocol, part of which is illustrated (center panel). 

Right-hand panels. Paired-pulse protocol recording from a small b-cell cluster after a 2 min pre-incubation in G20. A 100ms pulse (V_step) to 0mV from a holding potential of -70mV was given followed by an initial 500ms repolarization pulse to -80mV, followed by a paired V_step to 0mV. The step sequence was ended by returning to -70mV for a 2s interval before a subsequent paired-pulse was applied wherein the repolarization was diminished by 5mV for each successive paired-pulse sequence as far as 0mV. Upper panel. Six selected superimposed sweeps are shown revealing the decay of the amplitude of the 2nd deflection at -20mV with the progressive diminution in repolarization potential. The corresponding voltage pulses are shown on the same time scale (central panel). Lower panel. Mean ratios of peak currents determined by dividing the amplitude of the 2^nd current deflection by that of the 1^st for each paired step potential.  Data are plotted as a function of repolarization potential from 4 separate small cluster experiments using the GPP0 protocol, part of which is shown (center). Note that the repolarization pulse was of 500ms duration. Holding currents were typically between -5 and -25pA after correction for offsets. Traces are shown after leak current subtraction and current and time scale bars are given. The pipette solution used was Cs150 and all experiments were performed at 30-33°C.

(B)  Dependence of G-type current reactivation upon the repolarization and step potentials. Ratios of the peak amplitude of the 2nd current elicited by the paired depolarization step, divided by the 1st are plotted as a function of repolarization potential for 4 single-cell recordings stepped to -20mV (upper left-hand panel), 3 single-cell recordings stepped to -30mV (lower left-hand panel); for 3 recordings from small b-cell clusters stepped to -20mV(upper right-hand panel) and also for 4 small cluster recordings stepped to 0 mV (lower right-hand panel).  All data were fitted to the Boltzmann equation (see methods).  Corresponding V_0.5 values are shown inset for all data sets.  Center panel. Summary of V_0.5 values obtained presented as a function of the paired step potential for the four data sets given, and also for the data set presented in Figure 7.  The only agreeable fit was to a third order polynomial function.

Discussion

Relevance to the prevailing model of stimulus-secretion coupling in the pancreatic b-cell

Elevations in ATP concentration within the vicinity of the plasma membrane that result from an increased metabolism of D-glucose are thought to stimulate insulin secretion as a direct consequence of the closure of ATP-dependent K⁺ (K_ATP) channels (33, 34, 35), via the binding of ATP to its 'primary sensor' located upon the K_ATP channel protein (Kir6.2, 36), rather than its site upon the closely associated sulphonylurea receptor (37, 38, 39). The subsequent closure of K_ATP channels leads to a membrane depolarization (8), associated with an initial transient decrease in both membrane resistance and in the P_K/P_Na ratio (40) and the opening of voltage-gated Ca²⁺ channels (23). The resulting influx of Ca²⁺ triggers exocytosis by the fusion of insulin-containing granules with the plasma membrane (7, 41, 42), which may well be localized to a specific region of the plasma membrane (43) that possesses a high density of L-type Ca²⁺ channels (7).

At intermediate glucose concentrations slow membrane potential oscillations are observed, whose genesis is explained by a gradual rise in the ATP/ADP ratio which in turn causes an incremental progression of membrane depolarization towards the threshold for 'LVA' Ca²⁺ channels.  The opening of L-type channels is hypothesized to further promote depolarization towards the plateau potential upon which the regenerative L-type action potentials are superimposed.  Whether the LVA class of Ca²⁺ channels, which are implicated in the generation of the slow wave oscillations of membrane potential, represent another class of L-type Ca²⁺ channels in addition to those which underlie the regenerative train of action potentials which are seen to superimpose upon the plateau depolarization is unclear.  However the L(G)-type Ca²⁺ currents resolved here resemble, both kinetically and pharmacologically, the regenerative trains of action potentials observed in islets. 

Advantages of the cluster model

In recordings obtained from two day primary cultures of murine pancreatic b-cells, all but three from more than thirty membrane potential recordings from large clusters failed to show slow wave oscillations at intermediate concentrations of glucose (not shown).  Those few large clusters that did oscillate exhibited classical behavior, with the frequency of oscillations dependent upon the external calcium concentration (unpublished observations).  None of the small clusters tested exhibited membrane potential oscillations in response to glucose, in agreement with the previous findings of Rorsman et al (22), with only a sustained plateau depolarization being observed (approximately -20mV in G20).  As for the intact islet, the plateau potential was again characterized by the superimposition of Ca²⁺-dependent action potentials (22). This confers another advantage upon the cluster model, in that complications presented by slow, global and oscillating changes in input resistance and membrane conductance are apparently avoided in small clusters which appear to 'function better as electrical rather than as biochemical syncytia' (44). 

More importantly for the interpretation of the data from the small cluster preparation, gap junctional conductance has also been shown to oscillate synchronously between the active (plateau) and silent (repolarization) phases in pairs of acutely dissociated b-cells (45, 46), the extent of gap-junctional coupling increasing markedly during the active phase (46).  However, actual coupling parameters may be even higher within the core of the islet of Langerhans, where parallel arrangements of gap junctional conductances are present within the endogeneous clustered arrays of b-cells. The coupling coefficients and gap junctional conductances measured between acutely isolated pairs of b- cells at room temperature are almost certainly underestimates of the ‘true’ values within small clusters of b-cells (see appendix), as rates of metabolism increase non-linearly with temperature, and gap junctional conductance increases with glucose metabolism.

An active rather than passive role for Ca²⁺ channels in glucose-evoked stimulus-secretion coupling?

An exclusive role for the K_ATP channel in mediating the ionic basis of the action of glucose in stimulus-secretion coupling has previously been questioned (9, 18, 21).  This contention derives from a number of observations, including the biphasic concentration-dependence of the action of glucose in evoking insulin release from islets (18), by the glucose-dependent augmentation of insulin release in islets depolarized by high K⁺ in the presence of the K_ATP channel opener diazoxide (47), and by the original observation that glucose enhances both inward currents and L-type Ca²⁺ single-channel activity in the presence of Bay K8644 (5).  Collectively these data have suggested that, at least in the pancreatic b-cell, the regulation of Ca²⁺ channel activity is under direct metabolic influence.  These results confirm and extend these observations, and help to explain the differences between observations in intact islets and the findings of Smith et al (5) and those of other groups in the light of the presence of TEA⁺ in their experiments.  Moreover, these results effectively lay to rest the contention that voltage-gated Ca²⁺ channels are activated only as a secondary consequence of the membrane depolarization that is evoked by the closure of K_ATP channels.

Evidence for a diversity of Ca²⁺ channels in the regulation of insulin release

Both low (LVA) and high (HVA) threshold voltage-activated Ca²⁺ channels have previously been observed in primary cultures of human (10) and rat (48) pancreatic b -cells, Barnett et al (10) reporting thresholds of -55mV and -40mV for 'LVA' and 'HVA' channels respectively.  Only the HVA current was reported to be dihydropyridine-sensitive (L-type) and to be important in mediating sustained insulin release at high glucose concentrations (10), as the LVA 'T-type' channel rapidly inactivated. However, some contend that there may be as many as four different types of Ca²⁺ channel in the pancreatic b-cell; an HVA 'L-type' Ca²⁺ channel, an LVA 'T-type' channel (absent from murine b -cells?); a low threshold, slowly inactivating Ca²⁺ channel, and finally a low threshold, non-inactivating Ca²⁺ current (49). Here we present evidence for a novel high-threshold, rapidly inactivating 'G-type' Ca²⁺ current which is synergistically activated by both membrane depolarization and elevations in D-glucose concentration. The synergistic activation of ion channels by multiple effector pathways has been shown previously for both second messenger operated pathways (50) and for ligand-gated ion channels through heterologous binding sites for multiple ligands (51), but not by the convergent and interdependent action of membrane depolarization and an increase in cellular metabolism.

The L-type channel as a key signal integrator in stimulus-secretion coupling in the b-cell

L-type channels are now thought to co-localize with insulin secretory granules (7) and to be augmented by glucose in ‘normal’ b -cells (5), but not in b-cells obtained from a rat model of non-insulin dependent diabetes mellitus (NIDDM, 6), suggesting the dysfunctional regulation of this Ca²⁺ permeability pathway in at least one form of the diabetic condition. Although many neurotransmitters and hormones may act to regulate insulin secretion through mechanisms other than by modulating intracellular Ca²⁺ release and extracellular Ca²⁺ entry, for example by altering the efficacy of the action of Ca²⁺ in evoking exocytosis (1, 8, 20), many modulators are now known to act directly at the level of the 'HVA' L-type channel(s).  Thus L-type channel(s) clearly function as key cellular integrators of secretory stimuli and thus serve as a primary determinant of the secretory status of the b-cell.   As demonstrated here, not only are the frequency, amplitude, threshold and latency of L(G)-type channels modulated by glucose in a voltage and concentration-dependent manner, but they are also regulated by the external Ca²⁺ concentration in addition to the intracellular action of Ca²⁺ in binding to a cytoplasmic calmodulin sensor thought to be located upon the L-type channel (29, 30, 52).

Acetylcholine activates L-type Ca²⁺ channels through muscarinic receptors in a PKC-dependent manner (53), but also inhibits them via muscarinic receptors in a G-protein dependent manner (12).  The evidence presented here would also suggest that they are inhibited by the a₁/a ₂ antagonist phentolamine at low concentrations, further supporting a role for adrenergic receptors in the modulation of L-type currents and in the regulation of insulin secretion.  Cardiac a ₁ L-type Ca²⁺ channels, which are co-expressed in b-cells together with the ‘beta/endocrine’-subtype, are phosphorylated by cAMP-dependent protein kinase A (PKA) in a manner that is closely correlated with increases in glucose-evoked insulin secretion (54). Indeed, cAMP has been shown to act in synergy with glucose or tolbutamide in increasing cytosolic Ca²⁺ in b-cells in a manner that is functionally linked to L-type Ca²⁺ channel activity (55).

A potential role for the G-type Ca²⁺ channel in pulsatile insulin secretion

Although L-type Ca²⁺ channels are central elements in the regulation of Ca²⁺ entry and insulin secretion in response to glucose, the kinetics of the macroscopic currents previously obtained from single b-cells are not consistent with the ensemble averaged L-type single-channel currents, as these clearly lack a fast-inactivating component (7). Moreover, both dihydropyridine-sensitive and insensitive components of insulin release have previously been shown to co-exist (56).  These data are consistent with a potential role for the high threshold G-type current in the regulation of insulin secretion.  Further, these properties might serve to explain the phenomenon of pulsatile insulin secretion, especially if the G-type channel were found to co-localize with the exocytotic or vesicular recruitment zones.  The apparent conundrum in proposing a role for 'L-type' channels in both depolarizing the b-cell and in underlying the action potentials that superimpose upon that resulting depolarization, could well be explained by the existence of multiple L-type isoforms or splice variants. The respective contributions of the various L-type channel isoforms to the electrical phenomena of the b-cell might thus be better addressed by selective ‘knock-down’ experiments using antisense oligonucleotides against previously cloned subunits, as dihydropyridines are insufficiently discriminating for these purposes.

It remains to be established whether L-type Ca²⁺ channels mediate the recruitment of insulin-containing granules to the active exocytotic zone, or the entry of some or all of the Ca²⁺ that is responsible for vesicle fusion and exocytosis, or indeed both. However the pulsatile nature of insulin secretion leaves room for the involvement for a G-type channel in mediating a transient surge of Ca²⁺ entry that might serve to explain the key phenomenon of pulsatile insulin secretion.  In contrast to the ribbon synapse of the retinal On-bipolar cell where exocytosis has been shown to be proportional to the fourth or fifth power of the free synaptic Ca²⁺ concentration (57), in the pancreatic b-cell insulin secretion may be linearly correlated with the intracellular free Ca²⁺ concentration, although this remains controversial, even within the same laboratory (2, 17).  Indeed this may be a general feature of neuroendocrine cells (58).  The relative weight of the various Ca²⁺ channels present within the b-cell in evoking insulin secretion will be related in part to their proximity to the active zones of vesicle recruitment and exocytosis. However, as there now appears to be a clear correlation between NIDDM and a dysfunctional regulation of Ca²⁺ channels (6) which may specifically involve a change in the nature of the a_1C and a_1D (dihydropyridine-sensitive) L-type isoforms (59) which are co-expressed with the a_1E isoform in the b-cell (60).  Moreover, mice overexpressing calmodulin CaM-8, a known regulator of the oscillatory behavior of L-type Ca²⁺ channels (30, 52), have been shown to possess an ‘impaired’ Ca²⁺ current and to exhibit a deficit in their capacity to release insulin in response to glucose elevations (61), although whether this deficit results from an impairment of vesicular recruitment or exocytosis is not clear.  The molecular identity of the G-type current has yet to be determined, but Ligon et al (62) have found that w-agatoxin IVA sensitive a_1A-type channels are present within rat pancreatic b-cells that contribute to the dihydropyridine insensitive component of insulin secretion (56).

Differences in Ca²⁺ currents observed between clusters and single-cells

Another important anomaly between recordings from single-cells and small clusters, other than the demonstration that G- and L-types show a differential, but not mutually exclusive pattern of expression within acutely isolated cells, is that the G-type current deflection is constitutively active in acutely isolated single-cells at low glucose concentrations, but not in small clusters. Moreover, in the single-cell preparation the activation of the recurrent inward deflections by glucose appears to have been conserved, but their activation kinetics were dissimilar in both their lower frequency and higher threshold.  This consistent absence of an inward current within clusters incubated in low glucose suggests that cell-to-cell coupling and/or other intercellular contacts may be necessary for the ‘normal’ expression and regulation of the Ca²⁺ currents of the pancreatic b-cell, as b-cells are intrinsically gregarious.

Experimental Procedures

Cell culture

Islets of Langerhans were isolated using a modification of the methodology of Rorsman et al. (22). Two to five fed female NMRI mice (25-30g) were injected with 0.7-1ml pilocarpine and after 45min-1 hr incubation mice were cervically dislocated, decapitated and briefly immersed in 70% ethanol. The pancreas was surgically exposed and injected with a sterile-filtered  Rorsman isolation buffer (from frozen) containing (mM); NaCl 135, KOH 4.8, CaCl₂ 2.5, MgCl₂ 1.2, HEPES 10, glucose 10, bovine serum albumin (BSA) 1% (spontaneous pH 7.29), dissected and then incubated briefly in 10ml of isolation buffer supplemented with 50mg/ml Fungizone (Gibco).  Pancreatic tissue was then finely chopped and digested with 5-7mg collagenase, in proportion to tissue yield, and incubated for 12 min with gentle shaking in a water bath at 37°C. The resulting digest was washed first with DNAse-containing isolation buffer at 4° C to quench the reaction, and then washed twice with chilled Rorsman isolation buffer. Islets (typically 100-200) were sorted from the exocrine tissue and transferred to and incubated for eight min in a Ca²⁺-free saline containing BSA and EGTA at 4° C. The islets were then centrifuged at <100g for 5 min before decanting the supernatent, which was  replaced with 2ml RPMI 1640 medium (Gibco BRL, Paisley, Scotland, U.K.) supplemented with 10 mmol/l glucose, 100 IU/ml penicillin, 100 mg/ml streptomycin, 10% heat-inactivated fetal bovine serum and L-glutamine. The islets were then triturated with a heat-polished, sialinised pasteur pipette until only a fine cell suspension remained. The resulting suspension was plated at 0.3 ml/cover slip and allowed to attach for 7-20 hours before supplemention with 3ml RPMI 1640 medium.

b-cells make up approximately 80% of islet cells in normal mice (28) and have a larger diameter (11-14mm) than a-cells (8-11mm, (63)) and were thus co-cultured as a heterogeneous population of single-cells and small clusters.  Recordings were routinely made two days after trituration, except in recordings from identified single-cells, which were performed using an acutely isolated preparation wherein the RPMI trituration medium was additionally supplemented with 5mM HEPES and the pH adjusted to 7.4 with 1N NaOH in order to prevent cellular acidosis during isolation. Cells from selected cover slips were then transferred directly into an appropriate Krebs recording media after pre-incubation for 40-60 min in low glucose (G3) Krebs medium within a 37°C CO₂ incubator.  Cells at the center of the clusters were selected to optimize the space-clamp.

Solutions.

All cells and clusters were perfused in a heated chamber prior to recording with Krebs ringer containing (mM) NaCl 120, KCl 4.8, CaCl₂ 2.5, MgCl₂ 1.2, NaHCO₃^- 24, HEPES 5 (pH 7.35 with 1M NaOH) after gassing for 15 min with 95%O₂ + 5%CO₂, with a D-glucose concentration appropriate to the experiment. Solutions were continuously gassed during the recording with 94%O₂ + 6% CO₂. The K150 pipette solution contained (mM) KOH 140, KCl 10, N-methyl-D-glucamine 60, MgCl₂ 1.2, HEPES 10 (pH 7.1 with 1M H₂SO₄), and the Cs150 pipette solution contained (mM) CsOH 140, CsCl 10, N-methyl-D-glucamine 60, MgCl₂ 1.2, HEPES 10 (pH 7.1 with 1M H₂SO₄).

Ion substitution experiments were performed by equimolar substitutions of constituents within the Krebs medium. The low (59mM) Na⁺ Krebs contained (mM) NaCl 35, KCl 4.8, CaCl₂ 2.5, MgCl₂ 1.2, NaHCO₃^- 24, N-methyl-D-glucamine 85, HEPES 5 (pH 7.35 with 1M NaOH), and the low (0.25mM) Ca²⁺ Krebs contained (mM) NaCl 120, KCl 4.8, CaCl₂ 0.25, MgCl₂ 3.45, NaHCO₃^- 24, HEPES 5 (pH 7.35 with 1M NaOH).

Experiments performed to replicate and explain the experiments of Smith et al., (5) were identical to those published, the ‘Smith external’ buffer containing (mM) NaCl 138, KCl 5.6, CaCl₂ 2.6, MgCl₂ 1.2, TEACl 10, HEPES 10 (pH 7.4 with 1M NaOH) and the TEA-free buffer contained (mM) NaCl 138, KCl 5.6, CaCl₂ 2.6, MgCl₂ 1.2, HEPES 10 (pH 7.4 with 1M NaOH). The Smith hypotonic pipette solution contained (mM) KCl 10, CsCl 10, Cs₂SO₄ 70, MgCl₂ 7, HEPES 10, pH 7.4 with CsOH.  Measurements from the air-stone gassed reservoirs confirmed that the pH of the solution was maintained in the range of 7.35-7.4 by continuous gassing with the 94%O_2:6% CO₂ mixture.

Voltage-clamp recordings

Recordings were made with an EPC-9 patch-clamp amplifier using HEKA v8.11 voltage-clamp software (HEKA Electronics, Lambrecht/Pfalz, Germany). Cells plated upon cover slips were maintained in a purpose-designed perspex chamber equilibrated at 30-33°C with a heated stage (Intracell, Royston, Herts.).  The temperature was confirmed by measurement with a thermistor in the bath and at the mouth of the perfusion pipette before and after each recording.  Clusters were viewed under phase-contrast optics at 400x on an Axiovert 100 microscope, and for single-cell recordings using non-phase optics and an oil-immersion objective (x400).

Perforated-patch recordings were performed after the original method of Horn and Marty (25) using WPI glass electrodes with a 2-4 MW resistance in saline after fire-polishing. After coating with Sylgard elastomer (Dow Corning, Wiesbaden, Germany) to reduce stray capacitance and filling the tip with antibiotic-free intracellular solution, pipettes were back-filled with K150 or Cs150 solution containing Amphoteracin B. The principal recording solution was a HCO₃^--buffered Krebs solution supplemented with 5mM HEPES which was adjusted, after gassing for 10-15 min with 94% 0₂ + 6% CO_2, to pH 7.35 with 1N NaOH.  All Amphoteracin B containing pipette solutions were made freshly every hour with 250mg/ml Amphoteracin B (Sigma), from a fresh stock of Amphoteracin B dissolved by sonication in Hybrimax DMSO (Sigma). Fresh Amphoteracin B-containing pipette solutions were sonicated for 5min in 0.5ml pipette solution in an eppendorf and were maintained wrapped in foil in a syringe applicator on ice. All recordings were performed in the dark with only a minimum illumination intensity allowed during seal formation and perfusion alignment, due to the known light sensitivity of the polyene antibiotics.

The fundamental voltage-clamp pulse protocol used throughout these experiments, after the original methods of Hamill et al. (64), was termed IK1, and comprised a series of 21 voltage steps applied in 3mV increments from –70 to –10mV, each 300ms in duration from a holding potential of -70mV with a 2s inter-pulse interval. All other protocols were as described in the figure legends. The command voltage was checked weekly using an oscilloscope in parallel to the command output of the amplifier.  All recordings were commenced 10-15 min after seal formation at the central point of the small cluster to allow permeabilization of the patch and equilibration between the intracellular milieu and the pipette solution. Recordings were monitored continuously until both currents in G3 and the access conductance had stabilized at a minimum (typically >50nS). All recordings were acquired at a frequency of 10 kHz and presented after low-pass filtering at 1-2 kHz by an internal 8-pole Bessel filter.

In all recordings (except those replicating the experiments of Smith et al. (5), clusters and cells were perfused directly at 0.7-1.0 ml/min with solutions passed through a heated block to equilibrate at 30-33°C into an estimated chamber volume of 0.8-1.2ml constantly perfused with Krebs HEPES/HCO₃^---buffered medium gassed continuously with 94%O₂: 6%CO₂. All solutions contained 24mM NaHCO₃ and 5mM HEPES with a measured pH in the range 7.35-7.4, kept constant by continuously gassing all reservoirs with 94%O₂:6%CO₂ using miniature air stones. Solutions were changed using an electronic valve-regulated (24V Isolatch valves, General Valve Corporation, N.J.) perfusion system directed via a heating block into a fused silica capillary delivery flow pipe (O.D.740mm, Composite Metal Services Ltd., Worcs, U.K.,), ensuring that rapid and direct solution chang