Unperturbed islet α-cell function examined in mouse pancreas tissue slices

Abstract

Critical investigation into α-cell biology in health and diabetes has been sparse and at times inconsistent because of the technical difficulties with employing conventional strategies of isolated islets and dispersed single cells. An acute pancreas slice preparation was developed to overcome the enzymatic and mechanical perturbations inherent in conventional islet cell isolation procedures. This preparation preserves intra-islet cellular communication and islet architecture in their in situ native state. α-Cells within tissue slices were directly assessed by patch pipette and electrophysiologically characterized. The identity of the patched cells was confirmed by biocytin dye labelling and immunocytochemistry. α-Cells in mouse pancreas slices exhibited well-described features of I_Na (excitable at physiological membrane potential), I_KATP, small cell size, low resting membrane conductance, and inducible low and high voltage-activated I_Ca, the latter correlating with exocytosis determined by capacitance measurements. In contrast to previous reports, our large unbiased sampling of α-cells revealed a wide range distribution of all of these parameters, including the amount of K_ATP conductance, Na⁺ and Ca²⁺ current amplitudes, and capacitance changes induced by a train of depolarization pulses. The proposed pancreas slice preparation in combination with standard patch-clamping technique allowed large sampling and rapid assessment of α-cells, which revealed a wide distribution in α-cell ion channel properties. This specific feature explains the apparent inconsistency of previous reports on these α-cell ion channel properties. Our innovative approach will enable future studies into elucidating islet α-cell dysregulation occurring during diabetes.

Non-technical summary

Critical investigation into pancreatic islet α-cell biology in health and diabetes has been sparse and inconsistent because of technical difficulties in islet isolation and dispersion into single cells. We have circumvented these difficulties by employing the pancreas slice preparation. We functionally characterized (electrophysiologically) the α-cells in their in situ native state, then loaded the tested cells with biocytin dye to subsequently confirm the cell identities by immunocytochemistry. We characterized a very large number of α-cells, which showed a wide-range distribution in the electrophysiological parameters of several ion channels (ATP-sensitive K⁺, Na⁺ and Ca²⁺ currents) and capacitance changes as measure of exocytosis of glucagon granules. This could explain the apparent inconsistency of previous reports on α-cells that inadvertently showed skewed data due to insufficient sampling of α-cells. Our innovative approaches will enable future studies into elucidating α-cell dysregulation in diabetes.

Introduction

The assimilation of ingested nutrients, particularly glucose, from gastrointestinal tract into body tissues involves a complex interplay between regulatory and counter-regulatory endocrine mechanisms. These mechanisms are exemplified by the opposing actions of insulin and glucagon on target organs, including liver, adipose tissue and skeletal muscle, to very finely regulate glucose homeostasis. After glucose crosses from the intestinal tract into the bloodstream, insulin released from pancreatic β-cells would promote the transport of glucose into these tissues, whereas under high metabolic demand, glucagon secreted from pancreatic α-cells would mobilize glucose from these tissues to ensure adequate supply of circulating glucose particularly for vital organs (heart, brain). Distortion of insulin and glucagon secretion in diabetes can therefore cause severe perturbation in glucose homeostasis leading to an array of acute and chronic complications.

Much is known about regulation of insulin secretion from β-cells, and current therapy for diabetes has been directed at β-cells. In contrast, studies on α-cell biology have been relatively sparse and results have been inconsistent (Gopel et al. 2000; Ishihara et al. 2003; Franklin et al. 2005; Vignali et al. 2006; Zhou et al. 2007; Gyulkhandanyan et al. 2008). The latter has been attributed to technical difficulties in accurately identifying and examining a sufficient quantity of α-cells for study. The most commonly used islet α-cell isolation strategy involves collagenase digestion to free islets from exocrine tissues (Kanno et al. 2002; Speier et al. 2005). α-Cells are then examined in situ on the mantle surface of intact islets (Gopel et al. 2004) or as single cells by further dispersion of the isolated islets (Gopel et al. 2000). A major pitfall of these methods is that α-cells situated on islet mantle are inevitably exposed to the harsh enzymatic and mechanical injury. Furthermore, upon dispersion to single cells, the surviving α-cells deteriorate rapidly in culture conditions leaving only a small subpopulation of α-cells that are fit for study. Dispersed α-cells lose their normal contacts with neighbouring islet cells that are now known to exert important physiological paracrine inhibitory or stimulatory actions on α-cell function and secretion (Akesson et al. 2003; Barg, 2003; Wendt et al. 2004; Xu et al. 2006; Zhou et al. 2007). These inherent limitations of conventional approaches could contribute to observed alteration or loss of membrane channel protein expression or function in α-cells, including those previous reports from our own laboratories (Leung et al. 2005; Speier et al. 2005).

To circumvent the enzymatic and mechanical perturbations inherent in conventional islet isolation approaches, and which are further confounded by chronic culture conditions, we have reported an acute pancreas slice preparation as an alternative strategy to assess a number of properties of β-cell physiology (Speier & Rupnik, 2003). The pancreas slice preparation permits islet cells to be acutely prepared and examined, well-preserved in their natural cellulo-social environment within the intact islet micro-organ and unperturbed from destruction of adjacent exocrine tissue, thus ensuring a true in situ and healthy state. Using the pancreas slice preparation, we first demonstrated differences in K_ATP channel sensitivity to ATP between β-cells in slice and dispersed β-cells (Speier et al. 2005). Second, we showed perturbed β-cell Ca²⁺–secretion coupling in type 2 diabetic Goto–Kakizaki (GK) rats (Rose et al. 2007). Third, we found a disruption in gap junction coupling between β-cells caused by connexin (Cx36) deletion (Speier et al. 2007). And lastly, we employed this preparation to study perinatal pancreas (Meneghel-Rozzo et al. 2004) and islet development (Rozzo et al. 2009).

Here, we extend the use of this pancreas slice preparation to characterize normal α-cell physiology in an environment closely resembling intact tissue. The unique advantage of acutely examining α-cells in their in situ state is that it enables a large yield assessment of high quality α-cells that was previously not attainable. This work in the initial characterization of normal α-cells is required for subsequent work directed at examining the precise mechanisms underlying dysregulated glucagon secretion resulting from diabetes, which disrupts α-cells per se and also intra-islet integrative physiology (Gerich, 1988; Cryer, 2002).

Methods

Animal

Animal care and all the procedures were approved by the Institutional Animal Care and Use Committees of the Universities of Toronto and Maribor and have met the guidelines outlined in The Journal of Physiology (Drummond, 2009). C57BL/6 (The Jackson Laboratory, Bar Harbor, ME, USA) or Kir6.2 knockout (KO) mice (Miki et al. 1998) between 8 and 10 weeks old were used in this study.

Pancreas tissue slice preparation

Pancreas tissue slices were prepared as previously described (Speier & Rupnik, 2003). Briefly, after cervical dislocation, the mouse abdominal cavity was immediately opened to identify and clamp the common bile duct. 0.5 ml of 37°C low melting 1.9% agarose gel (Seaplaque GTG agarose, BMA, Walkersville, MD, USA) was infused into the pancreas via the proximal end of the common bile duct. The infused pancreas was cooled down by ice-cold extracellular solution, excised, trimmed into smaller blocks and embedded in 1.9% agarose. The hardened agarose blocks were glued onto the sample plate of a vibrating blade microtome (Vibratome, Leica Microsystems, Mannheim, Germany), covered with ice-cold extracellular fluid, and sliced into 140 μm-thick slices at a blade frequency of 70 Hz. Sliced tissues were kept in ice cold extracellular solution, bubbled with carbogen (5% CO₂, 95% O₂), and used immediately and for up to 8 h. Although the exocrine tissue remained intact for >24 h (data not shown), we took precautions to prevent any inadvertent degradation by digestive enzymes that could be released from exocrine tissues. Each glass beaker contained a maximum of five slices and filled with large volume (∼350 ml, 4° Celsius) of extracellular solution. Moreover, the agarose is porous, allowing ready exposure of embedded tissues to buffer nutrients and oxygenation (Speier & Rupnik, 2003), and preserves the exocrine tissue architecture and function, thus minimizing release of damaging digestive enzymes.

α-Cell labelling and confocal immunofluorescence microscopy

Alexa Fluor 594 biocytin conjugate (0.01 mg ml⁻¹, Invitrogen, Carlsbad, CA, USA) was dialysed (10–15 min) into a patched cell within the pancreas slice while that cell was being characterized electrophysiologically. Only one cell per slice was studied. After dialysis, the slices were immediately fixed in 4% paraformaldehyde (30 min), permeabilized with 0.1% Triton X-100 (30 min), then blocked with 10% normal goat serum. The slices were then stained with primary antibodies (1:100 dilution, 3 h): mouse monoclonal anti-glucagon or anti-insulin (Sigma-Aldrich Corp.), or anti-somatostatin (GeneTex Inc., Irvine, CA, USA); and later tagged with anti-mouse IgG FITC secondary antibody (AbD Serotec, Raleigh, NC, 1:100 dilution, 2 h). Finally, slices were mounted using anti-fade fluorescent mounting medium (DAKO Diagnostics, Ontario, Canada) and viewed under a confocal microscope (Carl Zeiss LSM510, Oberkochen, Germany).

Electrophysiology

Patching pipettes were pulled from borosilicate glass capillaries (GC150F-15; Harvard Apparatus Ltd, Edenbridge, UK) by a pipette puller (P-97; Sutter Instrument Co., Novato, CA, USA) to a resistance of 2–4 MΩ in K⁺-based solutions. The pancreas slice was fixed in the recording chamber by a U-shaped platinum frame with nylon fibres crossing the frame. The chamber was superfused with warmed (30 ± 2°C) carbogen-bubbled extracellular solution at 1 ml min⁻¹. Superfusion allows changes in solution and also washes out digestive enzymes that may be released from exocrine tissues. Islet cells in slices were viewed with an upright microscope (60× water, NA 1.0, Eclipse E600FN, Nikon). Standard whole-cell configuration was applied to all cell recordings performed in the present study, using a lock-in patch-clamp amplifier (SWAM IIC, Celica, Ljubljana, Slovenia or EPC-9, HEKA Electronik, Lambrecht/Pfalz, Germany), low-pass filtered and stored on standard PC. WinWCP (Strathclyde University, Glasgow, UK) and Pulse (HEKA Electronik) software were employed for voltage pulse generation, data acquisition and basic analysis, followed by further processing of the acquired data by Matlab (The Mathworks, Inc., Natick, MA, USA) and Matview software (Wise Technologies Ltd, Ljubljana, Slovenia). Data presented in this study are means ± standard deviation of the indicated number (n) of cells analysed.

Solutions

The extracellular solution bathing the pancreas slices was composed of (in mm): 125 NaCl, 2.5 KCl, 1 MgCl₂, 2 CaCl₂, 1.25 NaH₂PO₄, 26 NaHCO₃, 2 sodium pyruvate, 0.25 ascorbic acid, 3 myo-inositol, 6 lactic acid, 3 glucose. For recording Ca²⁺ and Na⁺ currents and depolarization-evoked cumulative capacitance change (ΔC_m), the pipette solution was as follows (in mm): 127 caesium methanosulfonate, 8 CsCl, 10 Hepes, 20 TEA-Cl, 2 MgCl₂, 2 Na₂ATP, 0.05 EGTA. The same pipette solution was used for Ca²⁺-induced capacitance change but excluding EGTA and including 5 NP-EGTA (Molecular Probes/Invitrogen, Eugene, OR, USA), 2.6 CaCl₂, 0.1 Fura-6F (Molecular Probes/Invitrogen). The pipette solution for K_ATP conductance measurement contained (in mm): 140 KCl, 10 Hepes, 2 MgCl₂, 0.05 EGTA. The pH of all solutions was adjusted to 7.2 with appropriate acid or base and osmolality was kept at 300 ± 10 mosmol kg⁻¹. All chemicals were purchased from Sigma-Aldrich Corp. (St Louis, MO, USA) unless otherwise indicated.

Ca²⁺ measurements

Cells loaded with Fura-6F pipette solution were subjected to monochromatic light (Polychrome IV, TILL Photonics, Victor, NY, USA) at 380 nm, short-pass filtered at 410 nm, reflected by a dichroic mirror (centred at 400 nm), and directed through a 60× water immersion objective focused on the vertical midline section of the patched and Fura-loaded cell. The emitted fluorescence was transmitted by dichroic mirror and through a 470 nm barrier filter. Images were obtained using a cooled emCCD camera (Ixon, Andor Technology, Belfast, UK). Intracellular Ca²⁺ concentration ([Ca²⁺]_i) was calculated using the equation derived by Grynkiewicz et al. (1985):

where K_d is the dissociation constant for Fura-6F (5.3 mm), F is the experimentally derived fluorescence, F_max is autofluorescence in the cell-attached configuration and F_min is fluorescence in resting whole-cell recording. Calibration of [Ca²⁺]_i measurements was performed in each cell as we and others previously described (Carter & Ogden, 1994; Sedej et al. 2004; Turner et al. 2005; Rose et al. 2007).

Results

Discrimination of islet cell types by biocytin labelling, immunocytochemistry and electrophysiology

Pancreatic α-cells in freshly prepared tissue slices are visually indistinguishable from their neighbouring islet cell types (Fig. 1). To increase the chance of hitting the cells of interest, cells on the islet edge just beneath the islet capsule were patch-clamped for electrophysiological characterization. Alexa Fluor 594 conjugated biocytin dye was included in the pipette solution to allow diffusion of the dye during the electrophysiological study. Upon completion of biocytin dialysis, the cell would have been labelled. Immunocytochemistry was then performed on these slices to identify the labelled cells and matched this to the identifying electrical properties of that cell type.

Figure 1. Immunocytochemistry confirming the identity of patched α-cells in pancreatic slices.

Each islet (Islets 1–7) is shown as a brightfield image (A) from which six confocal images were taken; the lower three images (B′, C′ and D′) are respective high magnifications of the indicated insets of the upper images (B, C and D). A in each panel shows a patch pipette, indicated by a white arrow, approaching cells on the edge of the islets. Alexa Fluor 594-conjugated biocytin was included in the pipette for dialysis into the patched-cells, thus labelling the cell subjected to electrophysiological and subsequent immunocytochemistry studies. Those biocytin-labelled cells (B and B′) were functionally characterized to be α-cells (Islets 1 to 5), β-cells (Islet 6) or δ cells (Islet 7) and then confirmed by immunostaining (C and C′) with glucagon (Islets 1, 2 and 3), insulin (Islets 5 and 6) or somatostatin (Islets 4 and 7) antibodies. D and D′ are the respective merged images which confirm the identities of the functionally categorized and biocytin-labelled islet cells. Specifically, merged images of Islets 1, 2 and 3 show abundance of glucagon in the α-cells surrounded the infused biocytin in the cytosol, whereas biocytin-loaded and functionally categorized α-cells were negative for somatostatin (Islet 4) and insulin (Islet 5). Biocytin-loaded and functionally categorized β- (Islet 6) and δ-cells (Islet 7) were confirmed to contain insulin or somatostatin respectively.

All cell types can be found on the islet edge (Fig. 1, islets 1–7). The electrophysiologically characterized α-cells (Fig. 1, islets 1–5; their functional properties are described in subsequent sections) were stained positive for glucagon (Fig. 1, islets 1–3, n = 15 electrophysiologically matched α-cells) and negative for insulin (Fig. 1, islet 5, n = 4 electrophysiologically matched α-cells) and somatostatin (Fig. 1, islet 4, n = 5 electrophysiologically matched α-cells), indicating that the functionally characterized cells were truly α-cells. Similarly, β-cells (Fig. 1, islet 6, n = 8 electrophysiologically matched β-cells) and δ-cells (Fig. 1, islet 7, n = 9 electrophysiologically matched δ-cells) were also identified using this approach, and found to not exhibit α-cell properties. For the interest of this study, we have focused only on islet α-cells. β-Cells in pancreas slices have been previously reported by us (Speier & Rupnik, 2003; Speier et al. 2005; Speier et al. 2007), and δ-cell electrophysiological characterizations will require a much larger sampling and an independent study. The combined approaches of biocytin labelling, hormone-specific immunocytochemistry and electrophysiology enabled us to refine the electrophysiological characterization of islet cells to the level where we can reliably use such ‘fingerprinted’ electrical properties alone to discriminate different cell types.

Voltage-gated Na⁺ channels

A major property distinguishing mouse α-cells from other islet cell types, including β- and δ-cells, is their Na⁺ channel half-maximal inactivation potential (V_1/2) (Gopel et al. 1999a, 2000). We therefore assessed the steady-state inactivation curve of Na⁺ channel in α-cells. Conditioning pulses between −150 and 0 mV in 10 mV increment were applied to the recorded cells (Fig. 2A inset). Membrane potential was then depolarized to 0 for 10 ms to evoke voltage-activated Na⁺ currents (Fig. 2A and B, α- and β-cells respectively). The evoked Na⁺ currents (I_Na) were normalized to the peak Na⁺ current elicited at −150 mV to derive the percentage peak Na⁺ currents, which were then used to plot against the applied step pulse voltages (Fig. 2C). As shown in Fig. 2C, the steady-state inactivation curves of α-, β-, and δ-cells were profoundly different. We found that α-cell Na⁺ channel V_1/2 was −37 ± 5.1 mV (n = 60 cells), similar to that reported previously (Gopel et al. 1999a, 2000), but in contrast to β-cells at −102 ± 8 mV and δ-cells at −66 ± 11 mV (n = 22 and 12 cells respectively, Fig. 2C). Note in Fig. 2A′ and B′ that when holding at −80 mV, I_Na can be fully activated in α-cells but not in β-cells. This I_Na is equivalent to previously reported early inactivating Na⁺ current, a major feature used to distinguish α-cells from β-cells (Gopel et al. 2004) and δ-cells, but has been notably diminished or even completely absent when α-cells were dispersed into single cells (Leung et al. 2005). Maximum Na⁺ current density in α-cells averaged −114 ± 75 pA pF⁻¹ when depolarizing cell membrane from −80 to −10 mV. Of note, analysis of this large population of α-cells showed that Na⁺ current amplitudes exhibited a broad distribution ranging from −15 to −316 pA pF⁻¹ (Fig. 2D).

Figure 2. α-Cell Na⁺ channel properties.

Representative α- (A) and β-cell (B) Na⁺ current traces elicited by a pulse protocol (inset). For clarity, only current traces triggered by −150, −80, −40, −30 and 0 mV step pulses are presented. Note that when V_m is hyperpolarized to below −80 mV, α-cells reveal close to zero inward current, denoted as ^∧∧ in A, indicating low resting conductance. This is in contrast to that of β-cells (denoted as ^∧∧ in B). The grey highlighted areas were expanded to a higher time scale shown in the lower boxes A′ and B′. Numbers in the graphs represent potential in mV. Note that when depolarizing cells from −80 mV membrane potential and above (A′, indicated by −80 arrow), Na⁺ currents in α-cells were elicited to the maximum amplitude but were completely inactivated in β-cells (B′, indicated by −80 arrow). The steady-state Na⁺ channel inactivation curve of α-, β- and δ-cells (labelled as α-, β-, and δ-cells, respectively) are shown in C. A histogram of I_Na density is shown in D.

We found that those biocytin-labelled, glucagon-positive α-cells exhibited consistent small cell size, averaging 4.1 ± 0.9 pF (n = 98 cells), which is in agreement with our previous report (Leung et al. 2005), in contrast to β-cells and δ-cells that are 6.2 ± 0.9 pF and 5.1 ± 1.7 pF, respectively. All α-cells displayed low resting conductance of 0.15 ± 0.08 nS, equivalent to 23 ± 17 pA of inward current at −150 mV membrane holding potential (n = 98 cells, Fig. 2A, denoted as ^∧∧), in contrast to β-cells (Fig. 2B, denoted as ^∧∧), which typically display >1 nS; the latter is due to the gap junction expression for electrical communication between adjacent β-cells (Gopel et al. 1999a; Speier & Rupnik, 2003).

α-Cell K_ATP channels

While Barg et al. (2000) reported relatively lower density (∼0.9 nS pF⁻¹) of K_ATP channels in α-cells than β-cells in isolated islets, we had reported higher K_ATP channel density (up to 1.5 nS pF⁻¹) in dispersed α-cells (Leung et al. 2005). We thus re-examined total K_ATP conductance (G_KATP) in α-cells within pancreas slices. K⁺-based ATP-free pipette solution was dialysed into α-cells to relieve K_ATP channels from their closed state. A ramp protocol running from −100 to −60 mV, the voltage range that avoids activation of any of the known voltage-activated channels, was applied to an α-cell at 1 s interval. The obtained ratio between current and voltage differences was calculated according to Ohm's law to derive G_KATP (Speier et al. 2005). Of 34 α-cells tested mean G_KATP was 1.96 ± 1.1 nS pF⁻¹, smaller than that of β-cells and δ-cells, which were 3.43 ± 0.51 (n = 7, P < 0.05) and 2.4 ± 2.0 nS pF⁻¹ (n = 9) respectively (Fig. 3A and E). Similar to the broad data distribution of Na⁺ current amplitude (Fig. 2D), we also observed a broad distribution of α-cell G_KATP densities, which ranged from 0 to 4.2 nS pF⁻¹ (Fig. 3D) and were independent of cell size (see Fig. 3C). This phenomenon of wide-spectrum K_ATP channel density in the α-cell population may in part explain the apparent inconsistency between our previous report and the study by Barg et al. (2000). Given that both studies examined only a small sampling of α-cells (n = 4 cells), these two studies might have inadvertently selected a small population of α-cells expressing K_ATP channels at either the higher (Leung et al. 2005) or lower end (Barg et al. 2000) of the wide spectrum of α-cell K_ATP channel density to have given rise to this apparent discrepancy.

Figure 3. K_ATP conductance.

A, representative α- and β-cell profiles showing development of G_KATP during the course of cytosolic ATP washout. Dotted lines in A, B and C indicate zero level G_KATP, and dashed lines in A indicate time to reach peak G_KATP. B, a representative Kir6.2 KO mouse α-cell G_KATP profile showing no development of G_KATP over time. C, the amount of G_KATP in α-cells is independent of cell size. Each dot represents one α-cell. No correlation between these two parameters was found in this graph. D, histogram of α-cell G_KATP density. Note the wide range of G_KATP density distribution. E, bar graph summarizing mean G_KATP of all islet cell types. *P < 0.05.

To ensure the measured current above were K_ATP currents, Kir6.2 knockout mice, which express non-functional K_ATP channels (Seino et al. 2000), were used as a negative control. Kir6.2 knockout α-cells were dialysed with ATP-free pipette solutions and showed no G_KATP development (Fig. 3B) during the entire course of ATP washout, suggesting that effects seen in wild-type mice were a result of K_ATP channel opening.

Since intracellular [ATP] in a cell decreases over the course of ATP washout, the time to reach maximum G_KATP can be an indicator of the sensitivity of K_ATP channels to ATP (Speier et al. 2005). For instance, the shorter the time required to reach the maximum conductance the less sensitive are the K_ATP channels to ATP and vice versa. We therefore examined the time to reach maximum G_KATP. Mean time to open α-cell K_ATP channels maximally was 277 ± 27 s (n = 29), slower than β-cells (177 ± 40 s, n = 9, Fig. 3A) but similar to δ-cells (299 ± 65 s, n = 8, data not shown). After maximum K_ATP conductance was reached, both α- and β-cell G_KATP showed a significant rundown (Fig. 3A), consistent with our previous reports (Speier et al. 2005; Rozzo et al. 2009). This phenomenon is likely to be a result of channel rundown largely attributed to prolonged ATP washout. All cells tested were constantly monitored to maintain consistent access resistance. This result suggests that α-cell K_ATP channels in pancreatic slices are more sensitive to ATP than those of β-cells.

Of note, three α-cells revealed almost no change in G_KATP throughout the entire course of ATP washout (Fig. 3C). It is likely that these cells were extremely sensitive to ATP so that a trace amount of ATP left in the cell was enough to block K_ATP channels. Alternatively, these cells can simply be ATP insensitive, or even expressed no K_ATP channels. This finding is similar to that reported by Gromada et al. (2004) showing a small population of α-cells to be consistently refractory to any experimental manipulation and reveals persistent glucagon secretion. Further work will be required to explain the lack of responsiveness in this population of α-cells.

α-Cells in pancreas slices possess LVA and HVA Ca²⁺ channels

Whereas healthy mouse β- and δ-cells express predominantly high voltage activated (HVA) Ca²⁺ channels (Kanno et al. 2002; Leung et al. 2005), mouse α-cells express both low voltage activated (LVA) and HVA Ca²⁺ channels (Barg et al. 2000; Gopel et al. 2000; Leung et al. 2006b). A cell determined as an α-cell in the previous section would be further examined for LVA and HVA I_Ca. A ramp protocol running from −80 to +60 mV was applied to α-cells to rapidly stimulate LVA and HVA I_Ca (Fig. 4A). The ramp was designed to incline at 0.47 mV ms⁻¹ to prevent activation of voltage-gated Na⁺ channels, and thus eliminate contamination of I_Na.

Figure 4. α-Cell LVA and HVA Ca²⁺ currents.

A and B, representative Ca²⁺ current profiles of α-cells (A) and β-cells (B) evoked by a ramp protocol (A top). Note the shape difference between α- and β-cell current traces. C, bar graph showing mean LVA and HVA I_Ca amplitudes of α-cells (n = 98 cells). The large standard deviations of the two bars are attributed by wide data distribution demonstrated in histograms D and E.

LVA and HVA I_Ca each peaked at −26 ± 3 mV and 0 ± 3 mV (Fig. 4A), with mean amplitudes of −8.5 ± 3.1 pA pF⁻¹ and −9.7 ± 3.1 pA pF⁻¹ (Fig. 4C), respectively. The LVA I_Ca clearly separated from HVA I_Ca in α-cells (Fig. 4A) was a unique feature observed in every α-cell examined and has been useful when distinguishing α- from β-cells in slices (Fig. 4B). Both LVA and HVA I_Ca amplitude distribution in α-cells ranged from −16 to −3 pA pF⁻¹ and −18 to −3 pA pF⁻¹, respectively (Fig. 3D–E). Again, a broad and non-parametric distribution of HVA and LVA I_Ca parameters was observed in the histograms that analysed large population of α-cells (n = 98).

α-Cell exocytosis is Ca²⁺ dependent

Since membrane depolarization leads to Ca²⁺ influx, this led us to question whether this Ca²⁺ influx would stimulate a rise in membrane capacitance (C_m), an indicator of cumulative exocytosis of dense core granules and synaptic-like microvesicles (Takahashi et al. 1997; Liu et al. 2005; Hatakeyama et al. 2006; Kasai et al. 2010). A train of 50 depolarizing pulses (Fig. 5A top), mimicking action potential firing, was applied to α-cells to stimulate voltage-gated Ca²⁺ channel opening and ensuing Ca²⁺ influx. The resulting C_m responses, observed at the 50th pulse, showed variable C_m changes (cumulative ΔC_m) ranging from 6 up to 214 fF increase (Fig. 5A, n = 32 cells). The mean cumulative ΔC_m evoked at the 50th pulse was 74 ± 54 fF (Fig. 5B). This variable ΔC_m was independent of cell size (Fig. 5C).

Figure 5. α-Cell exocytosis is a Ca²⁺-dependent process.

A, α-cell cumulative capacitance (ΔC_m) increase triggered by a train of 50 depolarization pulses (top) alternating between −80 mV (60 ms) and +10 mV (40 ms). The average cumulative ΔC_m is shown in B. C, the C_m increase elicited by the last (50th) pulse was taken as cumulative ΔC_m and plotted as a function of cell size. The grey area emphasizes the fact that ΔC_m is independent of cell size. D, a representative α-cell capacitance change (ΔC_m) in response to intracellular photo-release of caged Ca²⁺. The top graph demonstrates rise in intracellular Ca²⁺ concentration [Ca²⁺]_i and the bottom shows the corresponding C_m increase in response to [Ca²⁺]_i rise. UV light is turned on at 0 s (indicated by dotted lines). Dashed lines indicate the threshold [Ca²⁺]_i that initiates C_m increase. Arrow indicates a fast rising slope that is typically seen in many cells. The heights of this initial slope from all cells tested are scattered in E as grey dots. Average of the grey dots is shown as a black dot with standard deviation.

We further performed photo-release of caged-Ca²⁺ (NP-EGTA) experiment to examine α-cell exocytotic machinery per se, independent of Ca²⁺ channel activation. α-Cells were first loaded with NP-EGTA-contained pipette solution, and then subjected to UV exposure to free Ca²⁺ from NP-EGTA, thus increasing intracellular [Ca²⁺] ([Ca²⁺]_i). When [Ca²⁺]_i reached a threshold of 2.3 ± 0.8 μm (Fig. 5D, indicated by dashed lines), similar to the [Ca²⁺] induced by a train of depolarization pulses (Speier & Rupnik, 2003; Meneghel-Rozzo et al. 2004), a fast rise of C_m was initiated in α-cells. This initial slope (Fig. 5D arrow) increased to at least a minimum of 98 fF (Fig. 5E, scatter plot) and averaged to 171 ± 54 fF (Fig. 5E, black dot with standard deviation). The fact that no α-cell failed to show an increase in C_m in the photolysis experiment suggested the following: (1) α-cell exocytosis is a Ca²⁺-dependent process; and (2) α-cell exocytotic machinery per se is intact despite sometimes having reduced responsiveness to serial membrane depolarization – the latter could have been attributed to lack of coupling of Ca²⁺ channels to the exocytotic apparatus.

Discussion

In this study, we employed the novel pancreas tissue slice preparation (Speier & Rupnik, 2003; Speier et al. 2005) to acutely and efficiently examine a large number of α-cells preserved in its fresh, healthy and native cellular milieu, unperturbed by the damaging mechanical and enzymatic stresses inherent in the conventional islet cell isolation and dispersion procedures. We determined the electrophysiological identity of α-cells, and their identity was double-confirmed as true α-cells by biocytin-dye labelling and immunocytochemistry (Fig. 1). We showed that α-cells in pancreas slices were consistently small in size and expressed fast inactivating voltage-gated I_Na that can be fully elicited at physiological membrane potential (Fig. 2). ATP-sensitive K⁺ channels of α-cell in pancreas slices were more sensitive to ATP than β-cells (Fig. 3), which is consistent with our previous report (Leung et al. 2006a). In agreement with previous reports (Kanno et al. 2002; Leung et al. 2005), α-cell LVA and HVA I_Ca were easily distinguishable upon membrane depolarization (Fig. 4). α-Cell Ca²⁺ influx correlated with membrane capacitance increases, an indicator of Ca²⁺-dependent cumulative exocytosis of dense core granules and synaptic-like microvesicles (Kasai et al. 2010; Takahashi et al. 1997; Liu et al. 2005; Hatakeyama et al. 2006) (Fig. 5).

Throughout the study, we observed that many ion channel parameters in α-cells were distributed over a wider range than those previously reported. I_Na ranged from −15 to −316 nS pF⁻¹, LVA and HVA I_Ca ranged −16 to −3 nS pF⁻¹ and −18 to −3 nS pF⁻¹, respectively. K_ATP channel conductance in α-cells was expressed between 0 and 4.2 nS pF⁻¹. As α-cell electrical activity arises from orchestrated ion channel activities, this wide data distribution implies that each α-cell may be undergoing a dynamic electrical process operating in its own time frame unperturbed by coupling to neighbouring cells. This in part explains the observed low resting conductance of α-cells (Fig. 1). Each α-cell could thus be considered as a distinct or lone individual, as opposed to β-cells working in a syncytium (Rupnik, 2009). Another explanation for the wide distribution of α-cell parameters would be the different states of health of the α-cells examined. However, this possibility is unlikely for two reasons. First, all α-cells reported here revealed fast-inactivating voltage-gated Na⁺ current, which is prone to disappear in dispersed α-cells as a result of the enzymatic and mechanical damage from islet isolation and dispersion, as was also evident in our previous report (Leung et al. 2005). Second, there were no differences between data collected from different dates of the experiments or different times within a single experiment day (data not shown). Specifically, we did not see any trend of gradual decreasing or increasing channel activities when plotting the data between early (first hour) to late times (eighth hour) of channel recording after slice preparation. Thus, it is more likely that our postulate of α-cell individuality is the reason causing the wide-range data distribution. However, we do acknowledge that more work will be required to gain more insight into the underlying mechanism causing this pattern of wide data distribution in α-cells, which may in part be attributed to the intricate intra-islet neural and paracrine regulation.

Although our characterization of α-cells from pancreas slices is largely consistent with previous reports employing isolated islets or dispersed single cells, our preparation has distinct advantages for the study of α-cells. One major advantage is that all α-cells patched in pancreatic slices exhibited a complete set of electrophysiological fingerprints, in contrast to other preparations wherein some ion channel properties became altered or absent as a result of the harsher preparations. For example, one previous study (Vignali et al. 2006) reported that mouse α-cells do not express both readily activated (a.k.a. early inactivating) Na⁺ channels and LVA T-type Ca²⁺ channels. Paradoxically, they reported that the former was present in β-cells. These findings are contrary to the general consensus on the ion channel profile of α-cells (Gopel et al. 2000; Franklin & Wollheim, 2004; Wendt et al. 2004; Olsen et al. 2005; Zhou et al. 2007; Gyulkhandanyan et al. 2008) as well as the current study. Below, we attempt to explain the discrepancy.

First, our study employed Alexa Fluor 594-conjugated biocytin to label the patch-clamped cells during electrophysiological characterization. Immunocytochemistry was immediately performed on the slice to verify the identity of the biocytin-labelled patched cells. This was done with rigor demonstrating the presence of glucagon and absence of insulin and somatostatin. In contrast, the previous study (Vignali et al. 2006) employed single-cell RT-PCR to screen and identify α-cells and then functionally characterize these selected cells. Several recent reports showed that the use of single-cell RT-PCR technique to distinguish islet cell types is not appropriate because islet cells frequently express multiple islet hormone genes (Katsuta et al. 2010; Chiang & Melton, 2003). Employing a MIP (mouse-insulin promoter)-GFP mouse that lineage tracks islet beta-cells by GFP expression, it was found that with nested RT-PCR assay 45% of these beta-cells (confirmed by immunocytochemistry) expressed not only insulin gene but also other islet hormone genes (Katsuta et al. (2010). In our own hands, we also found that when single-cell RT-PCR was performed on functionally categorized α-cells, expression of not only the glucagon hormone gene but also the insulin or somatostatin genes was found in some of these α-cells (data not shown). Thus, we believe that the strategy of employing immunocytochemistry to identify abundant protein expression should be the more reliable one, and we have used it in our study.

Second, both our study employing pancreas slices and other reports examining mantle cells from intact isolated islets studied α-cells in their native environment wherein the natural paracrine communication between neighbouring islet cells was retained. The latter strategy however, exposes the mantle cells to damaging mechanical and enzymatic (collagenase) stresses, and this damage is amplified upon further dispersion to single cells. Dispersed α-cells would lose their normal contacts with neighbouring cells, thus disrupting normal paracrine influences (Akesson et al. 2003; Cejvan et al. 2003; Franklin & Wollheim, 2004; Xu et al. 2006; Zhou et al. 2007; Gyulkhandanyan et al. 2008). This could in part explain why dispersed single α-cells used by the previous study (Vignali et al. 2006) have altered electrophysiological properties with distorted expression of membrane ion channels. This previous study likely had selected surviving subpopulation(s) of alpha-cells for study, hence inadvertently introduced bias. Our own previous work (Leung et al. 2005) employing dispersed α-cells isolated from a MIP-GFP mice suffered from similar pitfalls, where we reported a loss of the fast-inactivating voltage-activated Na⁺ current, a key marker to discriminate α-cells from other islet cell types.

β-Cell ion channel properties were also found to behave differently between pancreas slices and isolated islets or single cells. Our previous report showed β-cells within intact islets embedded in pancreas slices expressed K_ATP channels that are less sensitive to ATP than that in dispersed single cells (Speier et al. 2005). Others have demonstrated that on glucose stimulation dispersed mouse β-cells in culture displayed stochastic spiking activity (Misler et al. 1992; Kinard et al. 1999) different from the regular oscillatory bursting of action potential seen in β-cells on intact isolated islets (Gopel et al. 1999b). Dispersed single β-cells exhibited lower capacitance increases compared to β-cells lodged in intact islets (Pipeleers et al. 1982). All these reports imply that both α- and β-cell functional properties could be significantly altered when the single islet cell approach was employed to study the biology of islet cells which normally operate in a cellulo-social environment within the islet micro-organ.

The pancreas slice preparation is well suited for imaging modalities that can precisely examine exocytotic events, particularly by multi-photon excitation microscopy (Takahashi et al. 2002). This imaging strategy can circumvent some of the limitations of the current electrophysiological strategy of capacitance measurement employed in this study. Specifically, cell membrane capacitance increases reflect the exocytosis of not only large dense-core vesicles (Oberhauser et al. 1996; Ninomiya et al. 1997), but also small synaptic-like micovesicles that can be substantial in islet cells (Liu et al. 2005; Hatakeyama et al. 2007) and are not easily distinguished using this electrophysiological strategy. Furthermore, membrane capacitance in islet cells is also profoundly affected by concurrent endocytosis (Smith & Betz, 1996; Liu et al. 2005). Multi-photon excitation microscopy can distinguish all the distinct exocytotic and endocytotic events (Takahashi et al. 1997; Liu et al. 2005; Hatakeyama et al. 2006; Kasai et al. 2010).

A final major advantage of this preparation is that the pancreas slice is by far the only preparation amenable and ideally suited to examine the precise islet cellular pathophysiology in diabetes. In type 1 and late-stage type 2 diabetes, the islet mass is severely reduced and any attempt to isolate the islets from these rodent models would cause severe islet damage, which along with the ongoing inflammatory disease processes would make any result from examining the islet cells not interpretable. These technical difficulties can be surmounted by the pancreatic slice preparation as islet cells are retained within their native disease environment (diabetes) and natural albeit diabetes-induced distorted paracrine communication, which could be deciphered by electrophysiology or imaging (i.e. multi-photon microscopy). Diabetes-induced perturbation of the remaining intact innervation (Ahren, 2000) and microcirculation can also now be pursued employing this preparation. This pancreas slice preparation would thus enable us and others to eventually elucidate the longstanding questions regarding how these pathological factors contribute to the perturbation of intra-islet α–β–δ cell communication to explain such clinical phenomena as glucose blindness (Gerich et al. 1973), which can be fatal.

Glossary

Abbreviations

C_m

membrane capacitance

G_KATP

K_ATP conductance

HVA

high voltage activated

I_Ca

calcium current

I_Na

sodium current

K_ATP channel

ATP-sensitive potassium channels

LVA

low voltage activated

NP-EGTA

nitrophenyl EGTA

V_1/2

half-maximal inactivation potential

Author contributions

Y.H., M.R. and H.G. are responsible for conception and design of the experiments. Collection, analysis and interpretation of the data and drafting of the manuscript were performed by Y.H., and M.R. and H.G. then provided critical review. All authors have no conflict of interest.