Glucose or Insulin, but not Zinc Ions, Inhibit Glucagon Secretion From Mouse Pancreatic α-Cells

Adenoviruses were constructed and amplified, using the pAdEasy system (34) as previously described (31). The 1.6-kb rat PPG promoter was digested using SacI and subcloned into pEGFP-C2 (Clontech). The resulting plasmid was then digested with XhoI/HindIII and transferred into pShuttle (pShuttle-PPG). cDNA encoding cytosolic luciferase (pGL3-Basic; Promega) was digested with HindIII/SalI and transferred into pShuttle-PPG (pShuttle-PPG-Luc). cDNA encoding cytosolic ratiometric pericam (33) was digested with HindIII/XbaI, and the restricted fragment was subcloned into pGL3-Basic. The resulting plasmid was then partially digested using HindIII/SalI and transferred into pShuttle-PPG (pShuttle-PPG-ratiometric pericam). Recombination with pAdEasy-1, transfection into HEK 293 cells, and viral amplification were performed as described (31).

αTC1-9 cells (passage 35–45; American Type Culture Collection) were grown in Dulbecco’s modified Eagle’s medium containing (in mmol/l): 18 NaHCO₃, 16 glucose, 0.1 nonessential amino acids, 10% heat-inactivated FCS, 100 IU/ml penicillin, and 100 μg/ml streptomycin. Islets were aseptically isolated by collagenase digestion of the pancreas of female CDI mice and selected by hand-picking. Islets were then cultured for 1 day in RPMI 1640 medium containing 10 mmol/l glucose, 10% heat-inactivated FCS, 100 IU/ml penicillin, and 100 μg/ml streptomycin (35). To obtain clusters or single cells, islets were dissociated after incubation for 5 min in Ca²⁺-free medium. After brief centrifugation, this solution was replaced by culture medium, and the islets were disrupted by gentle pipetting through a siliconized glass pipette. Clusters and single cells were then cultured for 1 day on circular glass coverslips. The control medium (Krebs-Ringer bicarbonate buffer [KRBB]) used for islet isolation contained (in mmol/l) 120 NaCl, 4.8 KCl, 2.5 CaCl₂, 1.2 MgCl₂, 24 NaHCO₃, 10 glucose, and 1 mg/ml BSA and was gassed with O₂/CO₂ (95/5) to maintain a pH of 7.4. The same medium was used for all imaging experiments. Where the concentration of KCl was increased to 30 mmol/l, the concentration of NaCl was decreased accordingly to maintain iso-osmolarity. αTC1-9 cells and islets or dispersed islet cells (2nd day of culture) were infected with adenovirus at a multiplicity of infection of 50–100 infectious particles per cell for 4 h, and then the medium was changed. Imaging was performed 2–3 days after infection.

Changes in [ATP]_c were measured in mouse α-cells within intact islets or in αTC1-9 cells after infection with PPG-luciferase adenovirus. Cells were perifused in KRBB containing 100 μmol/l beetle luciferin and imaged at 37°C using a single-cell photon-counting imaging device (ICCD218 intensified camera; Photek, Lewes, U.K.) (31) mounted on an Olympus IX-70 microscope fitted with a 10× objective lens. Photon release was monitored continuously at a video rate (60 frames/s) and data pooled over 10 s followed by three points moving average were performed off-line to optimize the signal-to-noise ratio (30).

Changes in cytosolic concentration of Ca²⁺ ([Ca²⁺]_c) were measured in dispersed mouse α-cells and in αTC1-9 cells after infection with PPG-ratiometric pericam-encoding adenovirus. Cells were perifused in KRBB and imaged at 37°C using an Olympus IX-70 with an Imago charge-coupled device camera (Till Photonics, Grafelfing, Germany) controlled by TILLvisION software (Till Photonics). Cells were illuminated alternatively for 5 ms (mouse α-cells) or 15 ms (αTC1-9 cells) at 410 and 480 nm, and the emitted light was filtered at 535 nm. The ratio images were used to calculate [Ca²⁺]_c off-line, as previously described (33).

Islets (2nd day of culture) or αTC1-9 cells were preincubated for 1 h at 37°C in control KRBB medium containing 10 mmol/l glucose. Islets were then distributed in batches of six and incubated for 1 h in 400 μl of medium containing test substances, whereas αTC1-9 cells were incubated for 30 min in 700 μl in a 12-well plate. At the end of experiments, an aliquot (350 μl) of the medium was taken and saved until glucagon and insulin were measured by radioimmunoassay (Linco Research, St. Charles, MO).

Cells were fixed and incubated with primary and secondary antibodies as previously described (36). Glucagon was probed with polyclonal rabbit anti-glucagon antibody (1:250; DakoCytomation, Cambridgeshire, U.K.) and revealed with Alexa Fluor 568 goat anti-rabbit IgG (1:3,000; Molecular Probes, Eugene, OR). Images were captured on a Leica SP2 laser-scanning confocal microscope using a 63× oil immersion objective.

Data are the means ± SE for the number of observations indicated. Statistical significance and differences between means were assessed by Student’s t test (paired or unpaired, as appropriate) or by ANOVA followed by a Newman-Keuls test when more than two groups were compared.

Expression of luciferase under the control of the rat PPG promoter (PPG-Luc) led to efficient expression of the photoprotein in islets (Fig. 1A). Because we could not perform immunocytochemistry for luciferase on islets due to antibody cross-reactivity, we compared the ability of the PPG-Luc virus to infect clonal α- versus β-cells. In six independent experiments, 56% of αTC1-9 cells were infected with PPG-Luc and 56% with CMV (cytomegalovirus)-Luc (i.e., 100% efficiency after normalization). By contrast, 7% of MIN6 β-cells were infected with PPG-Luc versus 55% with CMV-Luc (13% efficiency after normalization), demonstrating α-cell selectivity, and this was consistent with α-cell–restricted expression of the pericam probe (see below).

We used photon-counting imaging to monitor the effect of glucose on luciferase luminescence, and hence [ATP]_c (30–32), within α-cells in individual islets. Although this technique did not allow resolution at the level of individual cells over short integration times (10–20 s), recordings could nevertheless be made from the surface of the whole islet with a resolution of ∼30 μm, corresponding to 1–6 single α-cells (Fig. 1A). Increasing the glucose concentration from 0 to 20 mmol/l increased the luminescence from α-cell–localized luciferase by 9% (100 ± 0.5 vs. 109 ± 3.0 units, P < 0.01) (Fig. 1B). Consistent with previous measurements in groups of islets (21), we also observed an increase of 7% in [ATP]_c when the glucose concentration was increased from 3 to 20 mmol/l (100 ± 0.4 vs. 107 ± 2 units, P < 0.01) (Fig. 1C). By contrast, an increase from 5 to 20 mmol/l glucose did not produce any further increase in [ATP]_c (100 ± 0.3 vs. 96 ± 3.0 units, NS) (Fig. 1D), suggesting that [ATP]_c reached a maximum between 3 and 5 mmol/l glucose.

To explore the potential role of β-cell secretory products in mediating the effects of glucose, we sought to measure changes in [ATP]_c in α-cells in the absence of other islet cell types. Because the luminescence signals were too weak to be recorded from individual α-cells after islet disruption (data not shown), we instead monitored the effects of the sugar on [ATP]_c in small populations of 30–100 αTC1-9 cells. In these clonal cells, an increase in glucose concentration from 0 or 1 mmol/l to 20 mmol/l induced an increase in [ATP]_c of 4 or 3%, respectively (P < 0.05) (Fig. 1E and F), whereas an increase from 3 to 20 mmol/l glucose did not produce any further increase in [ATP]_c (Fig. 1G). The addition of 17 nmol/l insulin either in the absence (Fig. 1H) or in the presence (not shown) of glucose (3 mmol/l) failed to elicit any change in [ATP]_c.

To determine whether glucose may be able to affect intracellular [Ca²⁺]_c in the absence of β-cell secretory products, we used a recombinant ratiometric pericam (33,37) expressed under the control of the PPG promoter (PPG-pericam). As expected, the ratiometric pericam was targeted efficiently and almost exclusively to glucagon-positive α-cells (26 of 28 cells examined, 93%) (Fig. 2A), allowing selective measurements of [Ca²⁺]_c in this cell type.

To infect individual α-cells, mouse islets were dispersed and then treated with the PPG-pericam–bearing adenovirus (Fig. 2). We then monitored [Ca²⁺]_c changes in single, well-dispersed primary α-cells. At the low cell densities used (<1 cell/mm²), we anticipated that, under constant perifusion, local paracrine effects were unlikely to mediate any effects of glucose on individual cells. Of the 26 single primary α-cells examined in seven separate preparations, 31% (8 of 26 cells) displayed [Ca²⁺]_c oscillations at the lower glucose concentration tested (0.5 mmol/l), with a frequency of 0.2–1 oscillations/min. At a high glucose concentration (20 mmol/l), single α-cells displayed either a steady, low [Ca²⁺]_c value (2 of 8 cells) (Fig. 2B) or detectable [Ca²⁺]_c oscillations (6 of 8 cells) (Fig. 2C). In nonoscillating α-cells, a decrease in glucose concentration from 20 to 0.5 mmol/l induced oscillations of [Ca²⁺]_c (2 of 8 cells) (Fig. 2B), a response similar to that observed in mixed clusters of α- and non–α-cells (see below) (Fig. 5D). By contrast, in cells already displaying oscillations, the frequency was either increased by ≥15% (4 of 6 cells) (Fig. 2C) or remained unchanged (2 of 6 cells) (Fig. 2D), whereas the amplitude of oscillations was not affected (Fig. 2C and D). The addition of 30 mmol/l K⁺ induced either a sustained (56%) or a transient (31%) increase of [Ca²⁺]_c in almost all (14 of 16, 88%) α-cells examined, even in non–glucose-responsive cells.

To investigate further the ability of glucose to influence Ca²⁺ oscillations in the α-cell in the absence of β-cells, we used clonal mouse αTC1-9 cells. As shown in Fig. 3A, ∼50% of αTC1-9 cells displayed clear oscillations in [Ca²⁺]_c in the complete absence of exogenous glucose, with a frequency of 0.1–2.4 oscillations/min. These [Ca²⁺]_c oscillations were either completely suppressed or markedly decreased in frequency when glucose concentrations were raised (Fig. 3 and 5A), demonstrating a direct effect of the sugar that did not require the release of β-cell products. Analysis of the dose response for these effects (Fig. 3B and 5A) revealed saturation at glucose concentrations >3–5 mmol/l.

We next sought to determine whether there may be a separate role for insulin, or cosecreted Zn²⁺ ions, to potentiate the above direct effects of glucose. Added to αTC1-9 cells in the absence of glucose, where [Ca²⁺]_c oscillations were maximally stimulated, low concentrations (1.7–17 nmol/l) of insulin led to complete or partial suppression of [Ca²⁺]_c transients (Fig. 4A and B). The effects of insulin were also apparent at 0.5 mmol/l glucose, but they were no longer evident as the glucose concentration was raised to 1 or 20 mmol/l (Fig. 4B). Consistent with the actions of insulin via classical insulin receptors, and signal transduction via phosphatidylinositol 3′ kinase, the effects of the hormone were not mimicked by IGF-1 but were suppressed by 100 nmol/l wortmannin (Fig. 4B) or LY294002 (data not shown). Figure 5A illustrates the decrease in frequency of [Ca²⁺]_c oscillations at a given glucose concentration in the presence or absence of insulin. Insulin decreased the frequency of [Ca²⁺]_c oscillations at 0 or 0.5 mmol/l to the same extent as was observed at 1 mmol/l glucose in the absence of insulin, indicating that the effects of insulin and glucose were not additive.

To determine whether the effects of insulin might largely be explained by the presence of cocrystallized Zn²⁺ ions (insulin contained ∼0.5% zinc), we added 0.7 nmol/l ZnCl₂, the concentration present after the dissolution of 1.7 nmol/l insulin. This maneuver exerted no effect on the frequency of Ca²⁺ oscillations in clonal αTC1-9 cells (100 ± 13 vs. 92 ± 15, NS) (Fig. 5B). Similarly, at either 0 or 20 mmol/l glucose, 30 μmol/l of ZnCl₂ (as used previously in the perfused pancreas) (21), exerted either no effect or had a minor stimulatory action on the frequency of Ca²⁺ oscillations in the presence or absence of 17 nmol/l insulin in αTC1-9 cells (Fig. 5C). Similar results were obtained in isolated primary mouse α-cells: insulin either completely suppressed (Fig. 5D) or decreased (not shown) the frequency of [Ca²⁺]_c oscillations. The stimulatory effect of Zn²⁺ ions was more evident in primary α-cells from dissociated mouse islets, where 30 μmol/l ZnCl₂ reversed the inhibitory effect of insulin (Fig. 5D).

Glucagon secretion from αTC1-9 cells was dose-dependently decreased by glucose (Fig. 6A) and, in common with the observed [Ca²⁺]_c oscillations, significant inhibition was observed at glucose concentrations ≥1.0 mmol/l (P < 0.05). It is important to note that the release of insulin from αTC1-9 cells was essentially undetectable (≤50 pg · ml⁻¹ · h⁻¹ · well⁻¹) and was not affected by an increase in glucose concentration (not shown), consistent with a well-preserved α-cell phenotype. Again, the addition of insulin decreased glucagon secretion in the absence of glucose to approximately the same extent as 1 mmol/l glucose (Fig. 6A, Table 1), and the effects of insulin on glucagon secretion were suppressed by 100 nmol/l wortmannin (Table 1). Note, however, that the αTC1-9 cells used in this study displayed a left shift in the responses to glucose compared with primary α-cells in terms of [ATP]_c, [Ca²⁺]_c, and glucagon secretion (see below).

The release of glucagon from primary mouse islets was dose-dependently decreased in the presence of glucose, although in this case the main inhibitory effect was observed between 3 and 5 mmol/l (Fig. 6B), a range of glu-cose concentrations that did not stimulate insulin secretion (Fig. 6D). These results are consistent with the range over which glucagon release is suppressed in vivo (18). As in αTC1-9 cells, insulin (17 nmol/l) suppressed glucagon secretion from intact islets, and these effects were completely reversed by the phosphatidylinositol 3′-kinase inhibitor wortmannin (Fig. 6B and C, Table 2). Wortmannin, in the absence of insulin, also tended to increase slightly the release of glucagon, possibly as a result of the suppression of an inhibitory effect of basal insulin release (66 pmol/l) at 0.5 mmol/l glucose (Fig. 6C and D). However, even in the presence of wortmannin, 10 mmol/l glucose still inhibited glucagon secretion, consistent with a direct, insulin-independent effect of the sugar on α-cells (Fig. 6C). In line with the absence of effect of ZnCl₂ on [Ca²⁺]_c changes in both αTC1-9 cells and mouse α-cells, 0.7 nmol/l (not shown) or 30 μmol/l ZnCl₂ (Tables 1 and 2) had no impact on glucagon secretion in either the absence or the presence of insulin and/or elevated glucose concentration.

We demonstrate here the feasibility of imaging recombinant probes delivered selectively to either primary or clonal mouse α-cells under the control of the PPG gene promoter. We show firstly that fluctuations in luciferase luminescence may be used as a guide to intracellular free ATP concentrations in α-cells within the intact islet, as well as in clonal α-cells, as previously reported for β-cells (30,32). In the present studies, no attempt was made to provide a calibration of absolute free ATP concentrations in the α-cell. When such calibrations are performed after cell permeabilization (30), there are in any case considerable uncertainties as to the intracellular concentrations of luciferin and other luciferase regulators. However, total ATP concentrations have been reported to be ∼2 mmol/l in αTC1-6 cells (38), compatible with free ATP concentrations (comprising ATP⁴⁻ and Mg-ATP²⁻) in the range of 1–2 mmol/l (22).

Luciferase imaging revealed that glucose is able, in the absence of changes in β-cell secretion, to increase free cytosolic ATP concentrations in α-cells. This result is in apparent contrast to measurements of total cellular ATP content made in fluorescence-activated cell-sorted rat α-cells (22), but it is consistent with measurements of α-cell–restricted luciferase luminescence in rat islet populations (21). The magnitude of the observed increases (7–9% in primary α-cells in the intact islet, ∼4% in αTC1-9 cells) was, however, significantly lower than those recorded in either β-cells within the intact islet or clonal MIN6 cells, possibly reflecting the poorer capacity for oxidative metabolism of α- versus β-cells (39,40).

Using an α-cell–targeted pericam, we analyzed here intracellular Ca²⁺ oscillations in primary and clonal mouse α-cells. Using this approach, we provide evidence that glucose is able to directly influence the activity of α-cells in the absence of β-cell secretory factors. Thus, although we were unable to compare the efficiency of glucose to suppress Ca²⁺ oscillations in isolated α-cells versus α-cells within the intact islet (because of interference from β-cell autofluorescence in the latter case) (data not shown), the finding that a decrease in glucose concentration accelerated oscillations in most isolated α-cells is in line with previous observations that Ca²⁺ oscillations are inhibited in ∼19% of all cells within the intact islet (41). Importantly, we demonstrate here for the first time that glucose and insulin modulate the frequency of Ca²⁺ oscillations in clonal mouse α-cells. This contrasts with earlier reports in lnR1-G9 glucagonoma cells (42), where nutrients were found principally to affect the amplitude of oscillations. Although the sample sizes in both the current and an earlier (41) study were not sufficiently large to quantify the changes in primary mouse α-cells, such frequency modulation nonetheless provides a potential mechanism through which glucose may regulate the pulsatile release of glucagon.

What intracellular signaling mechanisms may mediate the effects of glucose on α-cell Ca²⁺? Several are possible. First, K_ATP channels may close as ATP concentrations (or the ATP-to-ADP ratio) increase. This has been proposed (43,44) to lead to the inactivation of Na⁺ channels and thus the inhibition of action potentials. However, we observed little if any response to 500 μmol/l tolbutamide, the K_ATP channel blocker, in clonal α-cells (M.A.R and G.A.R., unpublished observations), which is consistent with previous reports that these channels are scarce (45) or insensitive to tolbutamide and diazoxide in mouse α-cells, at least at basal glucose concentrations (3 mmol/l) (15,46,47). On the other hand, electrical and secretory responses to glucose are lost in α-cells from SUR1^−/− mice (44), arguing for an important role for these channels. Nevertheless, it is conceivable that additional sensors of cellular energy charge, such as AMP-activated protein kinase (26,48), are involved, as recently described for glucose-inhibited hypothalamic neurons (49). Alternatively, increases in ATP may also lower [Ca²⁺]_c by activating Ca²⁺ pumps, which in turn inhibit store-operated Ca²⁺ channels (46).

The present studies considered the possibilities 1) that the effects of glucose on glucagon release are indirect and mediated largely by insulin (or another β-cell–derived factor) and 2) that they are direct, but that paracrine factors additionally affect glucagon release when glucose concentrations are low. With regards to the former, our findings indicate that glucose stimulates α-cell oxidative metabolism directly, elevating [ATP]_c and in turn suppressing glucagon release. By contrast, Ishihara et al. (21) proposed that in the absence of a factor(s) secreted from β-cells, enhancement of α-cell metabolism by glucose would stimulate glucagon release. Although the reasons for these divergent conclusions are unclear, the current data argue that the release of a β-cell–derived factor is unlikely to be essential to inhibit glucagon release in response to glucose. Thus, the inhibitory effects of glucose were observed here both in αTC1-9 cells, where an effect of β-cell factors is excluded, and in mouse islets when insulin signaling was blocked with wortmannin (Fig. 6C). Correspondingly, glucagon secretion was fully inhibited at glucose concentrations (<5 mmol/l) that were clearly lower than those that began to stimulate the secretion of insulin (>5 mmol/l) (Fig. 6B vs. D) and presumably other β-cell factors. Concerning the possibility that the effects of glucose on glucagon are direct but that paracrine factors additionally affect glucagon release, insulin was found here to suppress glucagon release in the absence of glucose. Whereas the physiological relevance of this phenomenon is uncertain, it may ensure that insulin released in response to metabolizable amino acids such as leucine prevents the simultaneous activation of glucagon secretion (50).

In further contrast to recent observations in the perfused rat pancreas (21), we observed no or only small stimulatory effects of ZnCl₂ on [Ca²⁺]_c transients and glucagon release from either mouse αTC1-9 cells or mouse islets. One possible explanation for this apparent difference between rat and mouse is that the effects of Zn²⁺ ions may be mediated in the former case through a K_ATP channel–dependent mechanism, given that these channels are relatively abundant in rat, but not mouse, α-cells (see above). It is also conceivable that other targets for Zn²⁺ action, including γ-aminobutyric acid receptors (16,51), are important: although these have been characterized in part in rat α-cells (52), data are currently unavailable for mouse α-cells.

This study was supported by Juvenile Diabetes Research Fund Project Grant 2003-235 (to G.A.R.). G.A.R. is a Wellcome Trust Research Leave Fellow.

We thank Dr. T. Tsuboi for advice on the use of the targeted pericam, Rebecca Rowe for assistance with the assay of glucagon, and Dr. J. Philippe (University of Geneva) for PPG promoter cDNA.