In Situ Electrophysiological Examination of Pancreatic α Cells in the Streptozotocin-Induced Diabetes Model, Revealing the Cellular Basis of Glucagon Hypersecretion

Animal care and all procedures were approved by the Institutional Animal Care and Use Committee of the University of Toronto. GYY mice were previously described by P.G. (10). GYY mice (16–20 weeks old) used in this study were of similar age at which NOD mice develop overt diabetes (11) and have fully developed pancreas with adult-size islets to ensure that islet shrinkage from STZ-induced β-cell death (9) remains visible within the pancreas slices.

STZ (Bioshop, Burlington, Ontario, Canada) dissolved in sodium citrate buffer at a single daily dose of 40 mg/kg body mass was injected intraperitoneally into mice for 5 consecutive days and maintained for 4 weeks to allow animals to reach and surpass the peak incidence of β-cell apoptosis (9). Control animals were injected with citrate buffer. Mice were killed by cervical dislocation after the 4th week of STZ treatment for further experimentation.

For the intraperitoneal glucose tolerance test (IPGTT), 1.5 g glucose/kg body wt i.p. was injected into overnight-fasted (16 h) mice; blood glucose (glucometer, LifeScan, Burnaby, British Columbia, Canada) and glucagon were measured as indicated. For refed levels, mice fasted overnight (for fasting levels) were fed ad libitum with standard normal chow for 3 h. Blood samples were collected from femoral vein and loaded into capillary tubes coated with Kalium-EDTA (Microvette CB 3000; Sarstedt, Montreal, Quebec City, Canada); plasma was separated by centrifugation (10,000g, 10 min), and glucagon was assayed by RIA (Millipore, Etobicoke, Ontario, Canada).

Islet cell mass was measured as previously described (12). Briefly, resected pancreata were fixed in 4% paraformaldehyde, embedded in paraffin, and thinly sectioned (5 μm thin) and then dewaxed, rehydrated, and immunostained with rabbit polyclonal anti-insulin (1:200; Biomeda) or antiglucagon (1:150; Vision Biosystems) antibodies and counterstained with hematoxylin-eosin. Slides were digitized on a bright-field scanner (×20 magnification), and quantitative analysis of stained areas was performed with ImageScope software (Aperio Technologies, Vista, CA) using a positive-pixel count algorithm.

For hormone content measurements, isolated pancreata were placed in acid-ethanol mixture (1.5% HCl, 70% EtOH), incubated overnight (−20°C), and then homogenized; the tissue levels of insulin and glucagon extracted by centrifugation (5,000g, 15 min) were then determined by RIA kits (Millipore) and normalized to total pancreatic protein content.

Pancreas tissue slices were prepared as we recently reported (6). Electrophysiology was conducted an hour after preparation and completed within 5 h.

Immunofluorescence microscopy was performed as described in our recent report (6) with the following primary antibodies (1:100; 3 h): mouse anti-glucagon or anti-insulin (Sigma) or anti-somatostatin (GeneTex, Irvine, CA). Electron microscopy was performed as we previously described (13). Glucagon granules (thus α cells) were identified by the typical appearance (large black dense core with thin halo) (14,15). Glucagon dense core diameter was measured using NIH ImageJ software.

Electrophysiological recording was done as we described in our recent report using pancreatic slices (6), and here EYFP-expressing islet α cells in slices can be visualized and patched on. Standard whole-cell configuration was applied to all recordings, using a lock-in patch-clamp amplifier (EPC-9; HEKA Elektronik, Lambrecht/Pfalz, Germany). The signals were low-pass filtered and stored on the computer. Pulse (HEKA Electronik) was used for voltage pulse generation, data acquisition, and basic analysis, followed by further processing of the acquired data using Igor software. Voltage-clamp mode was used for recording all currents, and current-clamp mode was used for recording membrane potentials. The resting membrane potential was determined immediately after current-clamp mode was made, and spontaneous APs were observed without any current injection. The number of APs over recording periods of 30–120 s was counted using the event detection/threshold search program of Clampfit9 software (Axon), with a threshold range between −28.6 and −29.6 mV. The amplitude and firing frequency of APs were analyzed using Clampfit9 software. The firing frequency was normalized and presented as the number of APs per second (Hz). The extracellular solution has previously been described (6). The pipette solutions used were as follows: 1) for recording Ca²⁺ and Na⁺ currents, in millimoles, 127 Cs-methanosulfonate, 8 CsCl, 10 HEPES, 20 tetraethylammonium-Cl, 2 MgCl₂, 2 Na₂ATP, and 0.05 EGTA; 2) for recording K⁺ current and depolarization-evoked cumulative membrane capacitance (ΔC_m), in millimoles, 140 KCl, 10 HEPES, 2 MgCl₂, 2 Na₂ATP, and 0.05 EGTA; and 3) for membrane potential recording, the same pipette solution was used as in no. 2. pH of all solutions was 7.2; osmolality was 300 ± 10 mOsm kg⁻¹. All recordings were performed at 32°C. The pipette resistance ranged between 2 and 4 MΩ. All chemicals were from Sigma.

Fasting blood glucose level in STZ-treated GYY mice increased at 2 weeks (Fig. 1A) and refed blood glucose level at 1 week (Fig. 1B) posttreatment. Fasting (Fig. 1C) and refed (Fig. 1D) plasma glucagon levels rose to abnormal levels at 3 weeks and 1 week post-STZ treatment, respectively. The increased plasma glucagon levels (Fig. 1C and D) contributed in part to the observed hyperglycemia (Fig. 1A and B) in STZ-treated GYY mice. At 4 weeks post-STZ treatment, the mice showed the stable diabetes phenotype characterized as high and sustained fasting blood glucose (Fig. 1A) (control 6.1 ± 0.3 mmol/L, n = 10, vs. STZ 11.7 ± 1.7 mmol/L, n = 12), refed blood glucose (Fig. 1B) (control 10.1 ± 0.5 mmol/L vs. STZ 28.7 ± 1.5 mmol/L), fasting plasma glucagon (Fig. 1C) (control 70.3 ± 4.3 pg/mL vs. STZ 120.9 ± 24.2 pg/mL), and fed plasma glucagon (Fig. 1D) (control 113.9 ± 31.8 pg/mL vs. STZ 220.1 ± 32.7 pg/mL) compared with the controls. The STZ-GYY mice also experienced mild weight loss (Fig. 1E) and developed glucose intolerance (IPGTT) (Fig. 1F) at 4 weeks after treatment. We thus selected mice at 4 weeks post-STZ treatment for all the subsequent in vitro and electrophysiological experiments.

Confocal microscopy analysis of pancreas slices demonstrated a redistributed pattern of insulin (Fig. 2Bi and Bi’ [cells in red]) and glucagon (Fig. 2Bii and Bii’ [cells in both red and yellow]) staining in islets of STZ-treated mice compared with controls. Control islets showed glucagon-labeled α cells resided along the islet periphery (Fig. 2Ai and 2Ai’ [cells in yellow] and 2Aii and 2Aii’ [cells in both yellow and red]) and the larger mass of insulin-labeled β cells occupied the islet core (Fig. 2Ai and Ai’, red). Total pancreatic α- and β-cell masses were determined by immunohistochemical quantification of insulin- and glucagon-positive staining areas (see 1research design and methods), normalized to entire pancreas cross-sectional tissue area; the resulting values were then multiplied by the wet pancreas weight of each mouse. Mean pancreas wet weights were similar (control 0.216 ± 0.006 g vs. STZ mice 0.251 ± 0.05 g; P = 0.43). The relative ratio of α cell to total pancreatic tissue area was also similar (control 0.0029 ± 0.0005 vs. STZ 0.0024 ± 0.0002; P = 0.36). However, the ratio of β cell to total pancreatic tissue area was reduced in STZ-treated mice (control 0.0152 ± 0.0017 vs. STZ 0.0048 ± 0.0006; P = 0.0005). STZ treatment reduced β-cell mass by 71% (Fig. 2C), but there was no difference in α-cell mass between STZ and control mice (Fig. 2D).

Total pancreatic glucagon and insulin content normalized to the total protein content of each pancreas showed that STZ mice had increased mean pancreatic glucagon content (STZ mice 440,747 ± 55,473 pg vs. control 103,398 ± 5,999 pg; n = 5 mice/group, P = 0.0002) and reduced mean insulin content (STZ mice 4,091 ± 564 ng vs. control 68,518 ± 5579 ng; P = 0.000003). Mean pancreatic protein content was similar (STZ mice 23,590 ± 470 μg vs. control 22,621 ± 473 μg; P = 0.18). In the STZ-treated mice group, total pancreatic insulin content per protein content was severely reduced because of β-cell destruction (Fig. 2E) (control 3.02 ± 0.22 ng/µg vs. STZ 0.17 ± 0.02 ng/µg). Surprisingly, total pancreatic glucagon content per protein content in the STZ group was remarkably elevated to fourfold higher than in control mice (Fig. 2F) (control 4.59 ± 0.32 vs. STZ 18.71 ± 2.19 pg/µg).

The above results demonstrated that STZ induction of diabetes did not alter α-cell mass but increased glucagon content, which taken along with the increased plasma glucagon led us to postulate that each α cell in the STZ-treated mouse islet contains and releases more glucagon. We directly tested this hypothesis by two strategies: 1) single α-cell ∆C_m induced by serial membrane depolarization and 2) ultrastructural analysis of α-cell glucagon granules. Since GYY mice express EYFP in pancreatic α cells (10,16), this allows α cells to be visualized within pancreas slices under a fluorescence microscope (Fig. 2Gb and Gd) and be accurately reached with the patch pipette (Fig. 2Ga and Gc). We performed electrophysiological characterizations on every recorded α cell as previously reported (6) to further confirm their identity.

A train of depolarizing pulses trigger ΔC_m in α cells due to Ca²⁺ influx and subsequent glucagon granule exocytotic fusion, as an indicator of α-cell secretion (6). Figure 3A shows mean ΔC_m in α cells of control and STZ mice in response to 50 pulses depolarizing from −80 to 10 mV, which was higher in STZ-treated mice α cells. Figure 3B summarizes the cumulative ΔC_m triggered at the 50th pulse to be higher in α cells of STZ-treated mice (control 12.2 ± 1.9 fF/pF vs. STZ 19.9 ± 3.2 fF/pF; P < 0.05). Interestingly, the resting C_m of α cells (Fig. 3C) measured before the depolarization stimuli, which is proportional to cell membrane area, was also slightly increased in STZ α cells (5.33 ± 0.3 fF/pF) compared with control α cells (4.61 ± 0.18 fF/pF; P = 0.0496).

The increases in both evoked and resting C_m in α cells of STZ-treated mice suggest the possibility of an increase in the number and/or size of glucagon granules fusing with plasma membrane. To assess this possibility, we performed electron microscopy analysis of the islet α cells (Fig. 3D, control α cells; Fig. 3E, STZ-treated α cells). The corresponding enlarged views of the indicated areas (Fig. 3D’ and E’) show the glucagon dense core granules to be obviously larger in STZ α cells. Quantitative analysis shows that the mean diameter of granule dense core (Fig. 3I) was indeed larger in α cells of STZ-treated mice (228.50 ± 1.17 nm, N = 3,679 granules/32 cells; P < 0.0001) than control α cells (195.48 ± 0.94 nm, N = 2,642/18 cells). Analysis of granule distribution shows a shift in the overall sizes of glucagon granules of STZ-treated mice (Fig. 3G) compared with control α cells (Fig. 3F), with corresponding modes (peak of the curves relative to the vertical dotted line) of 209 ± 1.56 and 184 ± 1.5 nm, respectively. However, the number of glucagon granules per α cell determined as mean glucagon granules/cell area (Fig. 3H) was similar (control 3.0 ± 0.3 granules/μm² vs. STZ 2.3 ± 0.3 granules/μm²). If one assumes that dense core granules are perfect spheres and sizes of α cells are similar between the two groups, the volume of glucagon in α cells can be calculated by the equation V = 4/3 πr³, where r is the radius of the dense core. Accordingly, the volume of glucagon of an α cell of STZ-treated mice can be estimated to be 1.6 times larger than control α cells. Since resting C_m measurement (Fig. 3C) suggested an increased cell size in STZ-treated α cells, glucagon contained in α cells of STZ-treated mice should be >1.6 times larger than control α cells.

Nonetheless, this increased glucagon content in glucagon granules may not be sufficient to fully account for the four times increase in pancreatic glucagon content in STZ-treated mice. Taken along with the similar sizes of α-cell mass, our results suggest that the excess glucagon content per α cell is likely also contributed by accumulation of glucagon content in other compartments (endoplasmic reticulum, Golgi, etc.), which in turn suggests an increased in glucagon protein synthesis (which we currently don’t have the technical ability to determine).

Voltage-gated K⁺ (K_V) channels, shown to play a positive role in regulating glucagon secretion (17), were therefore assessed in α cells of control and STZ-treated mice (Fig. 4). In the control group (15 cells), a transient K_V current became detectable when depolarizing membrane potential to −30 mV and above (Fig. 4A [control]); this current inactivated rapidly (∼30 ms). Further depolarizing the membrane to −20 mV and higher voltages evoked an additional sustained K_V component (indicated in Fig. 4A). In the STZ group (14 cells), α-cell transient K_V current was notably suppressed (Fig. 4A [STZ]). Figure 4B summarized the transient K_V current density, which was significantly suppressed in STZ α cells when membrane potential was depolarized to 20 mV (control 281.8 ± 20.2 pA/pF vs. STZ 241.8 ± 11.4 pA/pF; P < 0.05) and higher voltages. Figure 4C summarized K_V-sustained current density, which was comparable between the two groups.

It is possible that α cells in STZ-treated mice might have larger Ca²⁺ influx to in part explain the larger C_m increase. We thus examined whether there is alteration in the properties of Ca²⁺ channels. In all the experiments, tetrodotoxin (0.1 μg/mL) was added in the bath solution to block voltage-gated Na⁺ current. A 300-ms ramp protocol running from −80 to 60 mV was applied to each α cell to trigger low voltage–activated (LVA) and high voltage–activated (HVA) Ca²⁺ currents. α cells from STZ-treated mice possessed LVA and HVA Ca²⁺ current density similar to that of control mouse α cells (Fig. 5A–H) (mean LVA control −3.14 ± 0.51 pA/pF vs. STZ −2.85 ± 0.25 pA/pF, P > 0.5; mean HVA control −7.10 ± 1.15 pA/pF vs. STZ −6.84 ± 0.82 pA/pF, P > 0.5; n = 8 control cells and 16 STZ cells). Since T-type current likely contributes to LVA Ca²⁺ currents, we added NiCl (100 μmol/L) to block T-type Ca²⁺ channels (Fig. 5A and B [control] and Fig. 5E and F [STZ]). As anticipated, NiCl reduced LVA Ca²⁺ current amplitude in α cells of controls from 3.14 ± 0.51 to 1.87 ± 0.42 pA/pF (Fig. 5A and B) (n = 6 cells) and that of STZ-treated mice from 2.51 ± 0.25 to 0.84 ± 0.29 pA/pF (Fig. 5E and F) (n = 11 cells). For confirmation of the HVA Ca²⁺ current component, CdCl₂, a broad-spectrum HVA Ca²⁺ channel blocker, was applied. The inward current component, peaked at 0–10 mV in both control (Fig. 5C and D [n = 6 cells]) and STZ (Fig. 5G and H [n = 6 cells]) cells, was completed abolished by Cd²⁺ (200 μmol/L). Taken together, our results indicate that Ca²⁺ current in α cells is contributed by predominantly HVA channels, consistent with previous reports (18,19). Both current amplitudes of HVA- and T-type channels were not significantly altered by STZ treatment.

Current-voltage dependence of voltage-gated Na⁺ inward current was assessed by using Cs²⁺-based tetraethylammonium-containing pipette solution to block K_V currents (Fig. 6A). Na⁺ current can be elicited in every α cell of both control and STZ groups from −80-mV holding potential. Na⁺ inward current became detectable when depolarizing membrane potential to higher than −30 mV (Fig. 6A). Dotted lines indicate zero current level. The peaked Na⁺ current amplitudes were −59.6 ± 6.2 pA/pF in controls and −78.4 ± 9.9 pA/pF in the STZ group upon depolarizing to 0 mV (Fig. 6B and C). Na⁺ channel I–V relationships of both control and STZ α cells exhibited a U-shaped voltage-dependent activation (Fig. 6B). Upon depolarizing to between −20 and 0 mV, α cells of the STZ group revealed significantly larger Na⁺ current density than controls (P < 0.05). Peak Na⁺ current was reached between −10 and 0 mV in both the STZ and control groups. Steady-state Na⁺-channel inactivation was examined by a depolarizing pulse to 0 mV from a set of conditioning pulses between −150 and 0 mV in 10-mV increments (Fig. 6D). In both control and STZ-group α cells, steady-state Na⁺ channel inactivation curve showed no difference; Na⁺ channel half-inactivation potential (V_1/2) in the control and STZ groups was −37.2 and −36.4 mV, respectively (Fig. 6E).

Voltage-gated Na⁺ and K⁺ currents are important components contributing to α-cell AP firing. We demonstrated that in the STZ-treated group, the Na⁺ and K_V current densities are, respectively, increased and decreased, and, thus, AP firing pattern in α cells could be subsequently altered. We performed current-clamp recordings to test this postulation.

After breaking the seal between pipette and cell membrane in voltage-clamp mode, current clamp mode was immediately switched to record resting membrane potential and regenerative AP firing. To confirm the α-cell type in addition to their cellular EYFP (6), a short step protocol (<4.5 s) was executed in voltage-clamp mode to determine the Na⁺-channel V_1/2.

The resting membrane potential in α cells was not significantly different between the two groups (control −56.3 ± 1.4 mV vs. STZ −54.0 ± 2.9 mV) (Fig. 7C). Robust AP firing was recorded in control (Fig. 7A) and STZ-treated mouse α cells (Fig. 7B). In control cells, AP fires spontaneously at a mean firing frequency of 0.68 ± 0.16 Hz (Fig. 7D): initiating at approximately −45 mV, giving rise to mean upstroke amplitude of 33.7 ± 1.8 mV, which then rapidly repolarized (Fig. 7A’). In α cells of STZ-treated mice, AP firing frequency was significantly higher (2.33 ± 0.27 Hz; P < 0.0001) (Fig. 7B and D), along with higher mean AP amplitudes (control 33.7 ± 1.8 mV vs. STZ 54.2 ± 2.1 mV, P < 0.0001).

α Cells in healthy GYY mice expressed voltage-gated Na⁺, Ca²⁺, and K⁺ currents and ΔC_m when depolarized. These electrophysiological properties are comparable with their background C57BL/6J and NMRI mice (6,10), validating GYY mice as an excellent model to study α-cell physiology and pathophysiology. STZ-treated GYY mouse α cells exhibited hyperglucagonemia and consequent hyperglycemia. Hyperglucagonemia in STZ-treated GYY mice resulted from elevated total pancreatic glucagon content partly contained in the larger glucagon granules, which are expected to effect more stimulated-glucagon granule release. Since total pancreatic α-cell mass of STZ-treated GYY mice remained unchanged while total pancreatic glucagon content was fourfold elevated, it can be predicted that the larger glucagon granules containing some of the excess glucagon are exocytosed in response to every stimulus (i.e., low glucose) on each α cell, culminating in the observed hyperglucagonemia. This is evident by the higher C_m increase in STZ-treated mouse α cells when electrically depolarized. Our results in the STZ-treated mouse are similar to a recent report showing that genetically induced near-complete β-cell ablation led to increased α-cell glucagon content and also consequent hyperglucagonemia (20).

What are the electrical and/or cellular properties of α cells in STZ-treated mice that contributed to the observed hyperglucagonemia? First, AP firing frequency and amplitude in α cells of STZ-treated mice were more intensified, which would be expected to lead to more glucagon secretion. Whereas these changes in AP are expected to increase Ca²⁺ influx, we showed that whole-cell Ca²⁺ current was not different between STZ-treated and control mice. The increased AP could be attributed to changes in Na⁺ and K_V currents; in fact, STZ-treated mouse α cells exhibited larger Na⁺ current density and lower K⁺ current density (discussed below). Second, the higher C_m increase in STZ-treated mouse α cells induced by serial membrane depolarization can be attributed to the larger glucagon granules exocytosing at the same Ca²⁺ stimulus, thus emptying a larger cargo of glucagon than control mouse α cells. We showed that K_V transient current within healthy α cells is rapidly activated and inactivated, which is similar to the A-type K current we (19) and Rorsman and colleagues (21) have previously demonstrated in rodent α cells. Consistent with the consensus thinking, this would suggest that the reduced K_V transient current in α cells of STZ-treated mice was likely the A-type K current, contributing to the observed increased AP firing rate and amplitude (21,22). However, recent reports showed that the K_V transient currents in both α- and β cells could also be contributed by large-conductance Ca²⁺-activated voltage-gated K⁺ currents (BK currents) (7,23,24), and, hence, reduced BK currents could also contribute to increased secretion and potentiated AP amplitude (7,20,21,25). In STZ-treated mouse α cells, we noted the increased Na⁺ current density coincided with an increase in AP amplitude. As expected (7,21), tetrodotoxin (an Na⁺-channel blocker) inhibited glucagon secretion in both STZ and control pancreas slices (data not shown). These findings are consistent with the notion that Na⁺ current influx-evoked AP trigger glucagon granule exocytosis.

What are the factors in STZ-treated mice that would cause suppression of K_V-transient current and enhancement of Na⁺ current in α cells? One possibility is the loss of paracrine regulation emanating from β-cell destruction, and these paracrine factors include insulin, γ-aminobutyric acid, and zinc (26–30) and possibly other factors. A second possibility is inflammatory cytokines secreted by cytotoxic T cells recruited during β-cell destruction induced by STZ, such as interleukin 1β, interferon-γ, and tumor necrosis factor-α (26), which could conceivably activate or sensitize some α-cell ion channels. In fact, several reports have shown that in colitis and ileitis animal models, release of inflammatory cytokine is associated with hyperexcitability of colonic dorsal root ganglion neurons, which were attributed to alterations in K_V and Na⁺ currents (31–35). These findings are consistent with our findings in α cells in the STZ-induced diabetes mouse model. Diabetes development in the STZ-treated mouse model (8) or in patients with T1D (36) is associated with similar inflammatory processes in the islets (i.e., insulitis), wherein cytotoxic T cells are recruited and release inflammatory cytokines (26) that can be predicted to perturb ion channels in α cells, perhaps in a manner similarly observed in this study.

The STZ-induced diabetes animal model in several respects mimics human T1D. STZ is transported into the β cell through Glut2 transporters, inducing cytotoxicity by oxidative damage and autoimmune reaction (8), including lymphocyte infiltration leading to insulitis and β-cell apoptosis (9). Because α cells express only Glut1 or Glut5 transporters (37,38), they are likely protected from STZ-mediated destruction, hence, the normal α-cell mass observed in this study. Furthermore, the complement of ion channel expression and functional properties of α cells in STZ-treated mice was comparable with α cells from control mice. Nevertheless, additional models of T1D, including noninflammatory (genetic ablation of β cells) (20) and autoimmune inflammatory (NOD mouse) models, should be tested in future studies to further validate the findings presented in this study and to explore the contributions of paracrine and inflammatory factors that may contribute to the perturbation of α cells in T1D. Insights from such studies could lead to strategies to correct such perturbation, thereby normalizing α-cell glucagon secretory responses (39).

This study was mainly supported by the Canadian Diabetes Association (OG-3-10-3020-HG to H.Y.G.) and by a Canadian Institutes of Health Research doctoral award (CGD 76317 to Y.-C.H.). Z.-P.F. holds a Heart and Stroke Foundation New Investigator Award. This work is also supported by Juvenile Diabetes Research Foundation Project Grant 2007-685 to P.G. and by Slovenian Research Agency Programme P3-0310-2334, grant J3-2290-2334, and MPG Partner Group Programme to M.R.

No potential conflicts of interest relevant to this article were reported.

Y.-C.H. performed all the experiments and data analysis, designed experiments, and wrote the manuscript. M.S.R. and Z.-P.F. contributed to the design of some experiments and to discussion and reviewed and edited the manuscript. N.K. provided technical assistance in glucagon secretion assay. P.L.H. and P.G. contributed to the discussion and provided the mice. H.Y.G. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.