Tolbutamide Controls Glucagon Release From Mouse Islets Differently Than Glucose: Involvement of K_ATP Channels From Both α-Cells and δ-Cells

Several mouse models were used: Sur1^−/− (lacking functional K_ATP channels) (20) and C57BL/6 (Sur1^+/+) mice, Sst^−/− (21) and Sst^+/+ mice (CBA/Ca × C57BL/10 F1 mice used as controls of Sst^−/− mice to have the same genetic background) (19), and ZnT8^−/− and ZnT8^+/+ mice (both strains obtained from heterozygous ZnT8^+/− mice) (22). The study was approved by our Commission d'Ethique d'Experimentation Animale.

Islets were isolated with collagenase and cultured overnight in RPMI 1640 medium containing 7 mmol/L glucose (G7) and 10% heat-inactivated fetal calf serum. The medium (pH 7.4) used for all experiments contained (in mmol/L): 120 NaCl, 4.8 KCl, 2.5 CaCl₂, 1.2 MgCl₂, 24 NaHCO₃, 1 mg/mL BSA, and various test agents as indicated. It was gassed with O₂:CO₂ (94:6%). To stimulate glucagon release, a 6 mmol/L amino acid mixture (2 mmol/L alanine, 2 mmol/L glutamine, and 2 mmol/L arginine) was present in most perifusion experiments. RO280450 was from Axon Medchem (the Netherlands), SST-14 was from Bachem, and pertussis toxin was from Tocris.

Batches of 100 to 500 islets were perifused at 37°C, at a flow rate of 0.5 mL/min, with various test solutions. Insulin (homemade assay) (23), glucagon (Millipore), and SST (Euro Diagnostica) were measured by radioimmunoassay.

The results are presented as mean traces (±SE) of experiments with islets obtained from at least three different preparations. Statistical significance of differences was evaluated by paired or unpaired Student t test.

Except for the experiments illustrated in Fig. 8, all perifusion experiments were performed in the presence of a 6 mmol/L amino acid mixture to stimulate glucagon secretion (23). This allows an easier detection of an inhibitory effect of glucose.

In the presence of 2 mmol/L glucose, glucagon secretion was high. Increasing the concentration to G7 reversibly inhibited glucagon release and stimulated insulin secretion (Fig. 1A). Addition of RO280450, a glucokinase activator (24), to a medium containing 2 mmol/L glucose mimicked the glucagonostatic and insulinotropic effects of G7 (Fig. 1B). By contrast, addition of 6 mmol/L 3-O-methyl-d-glucose, a nonmetabolizable glucose analog, to a medium containing 1 mmol/L glucose (G1) did not reproduce the glucagonostatic effect of glucose (Fig. 1C).

To test whether the glucagonostatic effect of glucose requires a modulation of K_ATP channels, we compared the effects of G7 on islets from Sur1^+/+ and Sur1^−/− mice. In Sur1^+/+ islets, switching from G1 to G7 reversibly inhibited glucagon release (Fig. 2A). In Sur1^−/− islets perifused with G1, glucagon secretion was much lower than in Sur1^+/+ islets (Fig. 2A; P < 0.05). This difference was not attributable to a difference in the glucagon content of the islets, which was similar in Sur1^+/+ and Sur1^−/− islets (726 ± 75 vs. 628 ± 116 pg/islet, respectively). Application of G7 to Sur1^−/− islets inhibited glucagon release. Expression of secretion as a percentage of release in G1 revealed that the extent of the inhibition was ∼50% lower in Sur1^−/− than in Sur1^+/+ islets (Fig. 2B).

We next tested the effect of glucose on glucagon secretion from Sur1^+/+ islets in conditions in which K_ATP channels were rendered pharmacologically insensitive to glucose after their maximal closure or opening with, respectively, 500 µmol/L Tolb or 250 µmol/L diazoxide (Dz). In the presence of Tolb and G1 (Fig. 2C), glucagon secretion was lower than in the absence of the K_ATP channel blocker (Fig. 2A; 0.32 ± 0.02 [n = 3] vs. 0.68 ± 0.1 pg/islet/min [n = 5]). Applying G7 induced a small and reversible inhibition of secretion (Fig. 2C). In the presence of Dz and G1, glucagon secretion was drastically reduced (0.043 ± 0.002 pg/islet/min; n = 3; Fig. 2C). Surprisingly, G7 still was able to reversibly suppress glucagon release (Fig. 2C, inset). As expected, neither Tolb nor Dz affected glucagon secretion from Sur1^−/− islets in the presence of G1 or G7 (not shown).

These experiments suggest that at least part of the glucagonostatic effect of glucose does not require K_ATP channels.

Addition of 500 µmol/L Tolb to a medium containing G1 reversibly inhibited glucagon secretion (Fig. 3A). The effect of the sulfonylurea was dose-dependent, being modest at 10 µmol/L (26% of inhibition; P < 0.01) and strong at 50 µmol/L (66% of inhibition; P < 0.01; Fig. 3C). Surprisingly, 500 µmol/L Tolb stimulated glucagon secretion when applied in G7 (Fig. 3B). A concentration of at least 50 μmol/L Tolb was required to see this effect (48% of stimulation; P < 0.05; Fig. 3D). It has been suggested that glucose concentrations higher than G7 paradoxically stimulate glucagon secretion (25). If this results from an additional closure of K_ATP channels, then this could be compatible with the stimulatory effect of Tolb. However, we found that an increase from G7 to 30 mmol/L glucose inhibited glucagon secretion (Fig. 3G) and did not reproduce the stimulatory effect of Tolb. Therefore, glucose and Tolb exert distinct effects on glucagon secretion.

We tested the effect of Dz. Addition of 250 µmol/L Dz in G1 strongly inhibited glucagon release (Fig. 3A). The inhibitory effect was already robust at 50 µmol/L of the drug (48% of inhibition; P < 0.01; Fig. 3E). We also checked whether any of the tested Dz concentrations could reverse the glucagonostatic effect of glucose; 250 µmol/L Dz strongly inhibited glucagon secretion in G7 (Fig. 3B), and a weak, but nonsignificant, inhibition was observed at 50 µmol/L (Fig. 3F). Importantly, lower concentrations of Dz never reversed the inhibitory effect of G7 (Fig. 3F), suggesting that glucose inhibited glucagon secretion independently from K_ATP channel closure.

The role of SST in the control of glucagon secretion by glucose was studied using Sst^+/+ and Sst^−/− mice. As expected, Sst^−/− islets lack immunoreactive SST (Supplementary Fig. 1) and do not secrete detectable amounts of SST. In G1, glucagon secretion was significantly (P < 0.05) higher in Sst^−/− than in Sst^+/+ islets (Fig. 4A). The difference was larger when secretion was expressed as percentage of content (0.24% ± 0.06 vs. 0.056% ± 0.01; P < 0.05) because the glucagon content was lower in Sst^−/− than in Sst^+/+ islets (664 ± 53 pg/islet [n = 18] vs. 820 ± 53 pg/islet [n = 23], respectively; P < 0.05). This suggests that SST exerts a strong tonic inhibition on glucagon release. Switching from G1 to G7 inhibited glucagon secretion from both Sst^+/+ and Sst^−/− islets, which demonstrates that SST alone is not responsible for the inhibition of glucagon secretion by glucose (Fig. 4A). However, the inhibition was less sustained in Sst^−/− than in Sst^+/+ islets, supporting a possible involvement of SST in the glucagonostatic effect of glucose (Fig. 4B). G7 stimulated SST release from Sst^+/+ islets (Fig. 4C) and triggered a larger insulin secretion from Sst^−/− than Sst^+/+ islets (Fig. 4D; 23.43 ± 5.12 [n = 3] vs. 8.28 ± 0.45 pg/min/islet [n = 5]). This latter observation was not attributable to reduced insulin content (47 ± 16 vs. 58 ± 19 ng/islet in Sst^+/+ and Sst^−/− islets, respectively) and suggests that SST exerts an inhibitory paracrine control on insulin release during glucose stimulation.

Addition of 500 µmol/L Tolb to G1 did not affect glucagon secretion from Sst^+/+ islets, whereas it stimulated that of Sst^−/− islets (Fig. 5A), demonstrating that Tolb did not reproduce the glucagonostatic effect of glucose. Because Tolb strongly stimulated SST release from Sst^+/+ islets (Fig. 5B) and because SST-14 potently inhibited glucagon secretion from Sst^−/− islets (Fig. 5D), it is most likely that in Sst^+/+ islets, Tolb-induced SST secretion has counteracted the direct stimulatory effect of Tolb on α-cells. By contrast, in Sst^−/− islets, glucagon secretion would be enhanced as a result of the direct stimulatory effect of Tolb on α-cells. It is worth noting that Tolb much more potently (nine-fold; P < 0.05) stimulated SST release by Sst^+/+ islets than did G7 (compare Figs. 4C and 5B). The sulfonylurea equally increased insulin secretion of both types of islets (Fig. 5C). 250 µmol/L Dz in G1 reversibly inhibited glucagon secretion of both types of islets (Fig. 5A) and had no detectable effect on insulin and SST secretion (Fig. 5B, C).

In the presence of G7, glucagon secretion was significantly (P < 0.05) higher in Sst^−/− than in Sst^+/+ islets (Fig. 5E) and, for both types of islet, was lower than in G1 (0.49 ± 0.11 [n = 3] vs. 1.31 ± 0.3 pg/min/islet [n = 6]; P = 0.11; and 0.15 ± 0.002 [n = 3] vs. 0.62 ± 0.04 pg/min/islet [n = 7]; P < 0.01; compare Fig. 5E and 5A). Also, 500 µmol/L Tolb stimulated glucagon and insulin secretion from Sst^−/− and Sst^+/+ islets (Fig. 5E, G) and increased SST secretion from Sst^+/+ islets (Fig. 5F). Increasing the glucose concentration from 7 to 30 mmol/L inhibited glucagon release by Sst^+/+ islets (Fig. 5H). Dz reversibly suppressed glucagon secretion from Sst^−/− and Sst^+/+ islets (Fig. 5E) but was ineffective on insulin and SST secretion (Fig. 5F, G).

To verify the involvement of SST in the control of islet hormone secretion, C57Bl/6 islets were pretreated for 18 h with 200 ng/mL pertussis toxin (PTX), which, by ADP-ribosylating the α-subunit of the G_i/o proteins, locks it in a GDP-bound inactive state and blocks the effect of SST. Control experiments showed that the pretreatment stimulated glucagon release approximately two-fold (0.79 ± 0.18 [n = 7] vs. 1.66 ± 0.22 pg/min/islet [n = 7]; P = 0.01; Fig. 6A, B) and prevented the glucagonostatic effect of SST (Fig. 6A). Increasing the glucose concentration from 1 to 7 mmol/L strongly and reversibly suppressed glucagon release in control and PTX-treated islets (Fig. 6B). Subsequent application of Tolb inhibited glucagon secretion of control islets but stimulated that of PTX-treated islets, as attested by the rapid decline in secretion on removal of the sulfonylurea. PTX treatment largely increased the stimulation of insulin release elicited by G7 and Tolb (Fig. 6C). Similar results were obtained in experiments in which Tolb was applied before G7 (Supplementary Fig. 2). These last experiments also show that G7 exerted a sustained glucagonostatic effect without concomitant sustained insulinotropic effect (insulin increases slightly and transiently only during the first application of G7), suggesting that insulin is not responsible for the glucagonostatic effect of glucose. Control and PTX-treated islets had similar glucagon (1.43 ± 0.12 ng/islet [n = 21] vs. 1.52 ± 0.17 ng/islet [n = 14], respectively) and insulin contents (148 ± 16 ng/islet [n = 17] vs. 145 ± 24 ng/islet [n = 11], respectively). These results confirm those obtained on Sst^−/− mice.

To test whether glucose could inhibit glucagon secretion independently of K_ATP channels and SST, Sur1^−/− islets were or were not pretreated with PTX. PTX treatment stimulated glucagon secretion four-fold (P = 0.04) in the presence of G1 (Fig. 7A), which is twice more than in control C57BL/6 islets (Fig. 6A, B). Again, it did not affect the glucagon content of the islets (1.12 ± 0.36 ng/islet [n = 3] vs. 1.19 ± 0.34 ng/islet [n = 4] for Sur1^−/− and Sur1^−/−-PTX islets, respectively). Switching from G1 to G7 strongly inhibited glucagon release from PTX-treated Sur1^−/− islets (Fig. 7A). Other series of experiments were performed on Sst^+/+ and Sst^−/− islets perifused with 500 µmol/L Tolb or 250 µmol/L Dz and showed that G7 decreased glucagon release under these conditions (Fig. 7B). These experiments indicate that glucose can inhibit glucagon secretion independently of K_ATP channels and SST.

Additional experiments were performed in amino acid–free media to verify key observations that were made in the presence of amino acids. The absence of amino acids dramatically reduced glucagon release. Dz (from 2 to 50 µmol/L) did not increase glucagon secretion of C57Bl/6 islets at G7, whereas a decrease in the glucose concentration to 1 mmol/L strongly stimulated glucagon release (Fig. 8A). Dz dose-dependently inhibited insulin release at G7 (Fig. 8B). Switching from G7 to G1 stimulated glucagon release of Sst^−/− islets, whereas the subsequent addition of Dz tended to decrease glucagon secretion, as attested to by the reacceleration of secretion on Dz removal (Fig. 8C). Again, Dz dose-dependently inhibited insulin secretion (Fig. 8D). Tolb stimulated both glucagon and insulin release (Fig. 8C, D). At G7, Dz did not affect glucagon secretion of PTX-pretreated C57Bl/6 islets, whereas Tolb potently stimulated their glucagon and insulin release (Fig. 8E, F). The observations that in the complete absence of influence of SST (Fig. 8C, E), Dz did not reverse the glucagonostatic effect of G7 whereas Tolb strongly stimulated glucagon release suggest that glucose inhibits glucagon release independently from α-cell K_ATP channels.

The effect of Tolb was also tested at G1. Under these conditions, Tolb did not affect glucagon secretion of Sst^+/+ islets but strongly stimulated that of Sst^−/− islets (Fig. 8G). Tolb also triggered a much larger insulin release by Sst^−/− than by Sst^+/+ islets (32.31 ± 5.61 [n = 4] vs. 7.08 ± 2.31 ng/min/islet [n = 4]; P < 0.05; Fig. 8H) and potently stimulated SST secretion from Sst^+/+ islets (Supplementary Fig. 3). Similar experiments were performed in C57Bl/6 islets pretreated or not with PTX. Tolb did not affect glucagon secretion of control islets, whereas it strongly stimulated that of PTX-treated islets (Fig. 8I). PTX treatment also potently increased the effect of Tolb on insulin release (Fig. 8J).

The role of Zn²⁺ in the inhibition of glucagon secretion by glucose was studied using ZnT8^+/+ and ZnT8^−/− mice. Incubation experiments showed that 10 mmol/L glucose decreased glucagon secretion to the same extent in both types of islets, demonstrating that Zn²⁺ is not responsible for the inhibition of glucagon secretion by glucose (Supplementary Fig. 4).

In the current study, we used pharmacological tools and three different genetically modified mouse strains to reassess the much controverted roles of K_ATP channels, paracrine SST, and paracrine Zn²⁺ in the glucagonostatic effect of glucose. We provide evidence that glucose and the K_ATP channels blocker, Tolb, have distinct effects, and that glucose can control glucagon release independently of α-cell and δ-cell K_ATP channels, SST, and Zn²⁺. We also show that Tolb influences glucagon secretion by two mechanisms, a direct stimulation of α-cells and an indirect inhibition by somatostatin released from δ-cells.

The α-cells express GLUT1 (but not GLUT2) (26) and the high K_m hexokinase, glucokinase (27). It has been reported that in α-cells, glucose slightly increases the free cytosolic [ATP] (17,28) and NAD(P)H fluorescence (18,29), but it does not affect the ATP-to-ADP ratio (30), suggesting that glucose is poorly metabolized. However, experiments with radioactive tracers have demonstrated a substantial rate of uptake (26) and anaerobic glycolysis (31), but limited oxidative metabolism and anaplerosis when compared with β-cells (31). By using 3-O-methyl-d-glucose, a nonmetabolizable glucose analog that is taken-up by α-cells at a similar rate as glucose (32), we showed that the glucagonostatic effect of glucose requires its metabolism. Moreover, activation of glucokinase by RO280450 strongly inhibited glucagon release. It is, however, unknown whether this last effect is direct or indirect.

The α-cells possess K_ATP channels (6,7,9,33). However, whether these channels are essential for the glucose control of glucagon secretion is still disputed (6,13,33). Our observations that, in the presence of G1 and amino acids, Tolb mimics the glucagonostatic effect of G7 on Sur1^+/+ islets and that glucagon secretion of Sur1^−/− islets was lower than that of Sur1^+/+ islets support, at the first sight, an involvement of α-cell K_ATP channels in the glucagonostatic effect of glucose. However, this conclusion is challenged by observations demonstrating that glucose and Tolb exert distinct effects. 1) In Sst^−/− islets or PTX-treated C57Bl/6 islets perifused with amino acids, Tolb stimulated glucagon secretion, whereas G7 inhibited it. 2) In Sst^−/− islets perifused without amino acids, glucagon secretion was stimulated on switching from G7 to G1 and by Tolb. 3) Glucagon secretion from control islets (C57Bl/6, Sst^+/+) was stimulated by Tolb at G7, whereas it was inhibited by an increase of the glucose concentration from 7 to 30 mmol/L. Abrogation of the effect of Tolb by knockout of Sur1 demonstrates that the stimulatory effect of Tolb did not result from an action on a mitochondrial-like K_ATP channel conductance (34) or the activation of Epac2 (35). The distinct effects of glucose and Tolb on glucagon secretion is compatible with previous reports showing that glucose slightly decreased, whereas Tolb increased [Ca²⁺]_c in single α-cells (13,33). The observation that, at G1, Tolb inhibited glucagon secretion from Sur1^+/+ islets but was ineffective on Sst^+/+ islets is probably attributable to strain differences. Other observations from our study suggest that glucose can inhibit glucagon secretion independently of α-cell K_ATP channels. 1) If the glucagonostatic effect of glucose results from a closure of α-cell K_ATP channels, then it should be reversed by Dz concentrations in a low and narrow range as previously suggested (6). We instead found that in the presence of amino acids or in their absence and without influence of paracrine somatostatin signaling (Sst^−/− and PTX-treated islets), none of the tested Dz concentration reversed the glucagonostatic effect of G7. On the contrary, the drug either was ineffective at low concentrations or suppressed glucagon secretion at higher concentrations, as expected from its hyperpolarizing action on α-cells (10,33). 2) Glucose could still inhibit glucagon secretion from Sur1^−/− islets or control islets perifused with high concentrations of Tolb or Dz. This last observation may seem surprising. However, it has been reported by others (13,25) and is consistent with the [Ca²⁺]_c-lowering effect of glucose in Dz-exposed α-cells (10). Our finding that glucose inhibited glucagon secretion without the intervention of α-cell K_ATP channels is compatible with a recent report (36) and our previous observation demonstrating that, in α-cells, glucose did not significantly affect I_KATP (13,33). From all the above-discussed observations, it is clear that glucose and Tolb exert distinct effects in α-cells and that glucose can inhibit glucagon secretion independently from α-cell K_ATP channels. Nevertheless, we cannot fully exclude the involvement of K_ATP channels in the glucagonostatic effect of glucose because Tolb mimicked the inhibitory effect of glucose in Sur1^+/+ islets. Previous reports using different Sur1 or Kir6.2 knockout models showed that disruption of the K_ATP channels did not affect (37), reduced (38), or completely prevented (6,9,39) the glucagonostatic effect of glucose.

The δ-cells also possess K_ATP channels (40). One model proposes that the glucose-induced inhibition of glucagon secretion is mediated by SST (2,19). Of the two endogenous bioactive forms of SST (SST-14 and SST-28), SST-14 is the predominant form in pancreatic islets. SST is a potent inhibitor of glucagon secretion (2), as confirmed here. Five SST receptor subtypes have been described (SSTR1–SSTR5). Although experiments using SSTR2-selective agonists and antagonists as well as Sstr2 knockout mice have suggested that SSTR2 is the main mediator of SST-induced inhibition of glucagon release (41,42), some reports suggest that α-cells also express other SSTRs (43,44). Because of this diversity in SSTR expression, the lack of selective and potent SSTR antagonists (44), and the reported unspecific effects of some antagonists (2), we have used islets of Sst^−/− mice and islets of C57Bl/6 mice pretreated with PTX (which, by invalidating G_i/o proteins, impairs SST signaling) to investigate the role of SST in the glucagonostatic effect of glucose. In all conditions, we observed a higher rate of glucagon release in islets devoid of SST paracrine influence. This is compatible with the higher rate of glucagon release found in Sstr2 knockout mice (42) and after blockade of SSTR2 receptors (13) or immunoneutralization of SST by antibodies (45), and it suggests that SST exerts a tonic inhibition on glucagon release. Importantly, glucose efficiently inhibited glucagon release of islets devoid of SST paracrine influence, indicating clearly that the glucagonostatic effect of glucose does not require SST. However, we cannot fully exclude a small participation of SST in the inhibitory action of glucose because glucose stimulated SST secretion and because the glucagonostatic effect of glucose was less sustained in Sst^−/− than in Sst^+/+ islets (Fig. 4B). Conflicting results have been reported in the literature. Thus, the glucagonostatic effect of glucose was found to be preserved or even increased in the presence of a SSTR2 antagonist, a somatostatin antibody (13,45), or after pretreatment with PTX (46). However, a previous study using the same mouse model as we used here reported that the suppressive effect of glucose on glucagon secretion was lost in Sst^−/− islets (19). The reasons for these discrepancies are unknown and might be related to differences in experimental conditions.

The involvement of SST in the effect of Tolb on glucagon secretion is much more obvious. Thus, in the presence of G1, Tolb did not affect or inhibited glucagon release of Sst^+/+ or control C57Bl/6 islets, whereas it stimulated glucagon secretion of Sst^−/− or PTX-treated C57Bl/6 islets (even very potently in the absence of amino acids). This clearly indicates that SST is involved in the control of glucagon release by Tolb. The stronger involvement of SST in the glucagonostatic effect of Tolb than that of glucose is compatible with our observation that the sulfonylurea stimulated SST secretion much more potently than glucose. The difference in the effect of Tolb between islets with or without paracrine SST signaling suggests that Tolb modulates glucagon secretion by two distinct mechanisms, a direct stimulatory effect on α-cells that is detected in the absence of SST (i.e., in Sst^−/− or PTX-treated islets) and an indirect inhibitory effect that is caused by the stimulation of SST release. The net effect of Tolb on glucagon secretion would thus result from a balance between the stimulatory and the inhibitory effects. That Tolb directly stimulates α-cells is supported by previous reports showing that the sulfonylurea increases [Ca²⁺]_c (13,33) and glucagon secretion from isolated α-cells (8,16). Although the α-cell K_ATP current is already small even at low glucose, a tiny additional reduction of the current elicited by Tolb would strongly affect α-cell electrical activity because of the high resistance of its plasma membrane.

The opposite effects of α-cell and δ-cell K_ATP channel modulation for the control of glucagon secretion would also explain why both Tolb and Dz, which have opposite effects on channel activity, inhibit glucagon secretion at G1. On the one hand, the glucagonostatic effect of Tolb would mainly be mediated by the dominating effect of SST as explained above. On the other hand, the glucagonostatic effect of Dz would essentially result from its dominating direct hyperpolarizing effect on α-cells.

The strong paracrine influence of SST released from K_ATP channel–deficient δ-cells might explain the reduced glucagon secretion of Sur1^−/− islets. It is compatible with the observation that pretreatment of these islets with PTX potently stimulated glucagon secretion because of the relief of the inhibitory effect of SST.

At G7, Tolb stimulated glucagon secretion from islets with or without paracrine SST signaling. It is possible that in conditions in which glucagon secretion is already inhibited, the direct stimulatory effect of Tolb on α-cells overwhelms the indirect inhibitory effect caused by the stimulation of SST release. The glucose dependency of the effects of Tolb and its two mechanisms of action, directly on α-cells and indirectly through δ-cells, might explain why sulfonylureas have been reported to exert variable effects on glucagon release. Thus, glucagon secretion was stimulated (8,16), unaffected (47), or inhibited (38,39,48) by sulfonylureas.

Paracrine SST also influences insulin secretion. Thus, in most tested conditions, glucose and Tolb induced a larger insulin secretion in islets without paracrine SST signaling than in control islets, confirming previous reports (19).

Experiments on islets with genetic or pharmacological disruption of both the K_ATP channels and SST signaling (Fig. 7) revealed that glucose can inhibit glucagon secretion independently from K_ATP channels and SST. This is compatible with our previous observation demonstrating that in isolated α-cells devoid of paracrine influence, glucose decreased [Ca²⁺]_c in the presence of a high concentration of Tolb (13,33). The nature of the underlying mechanism is, however, unknown.

It has been hypothesized that Zn²⁺ released from β-cells could be responsible for the glucagonostatic effect of glucose. By monitoring Zn²⁺ exocytosis from ZnT8^+/+ and ZnT8^−/− mice, we previously showed that ZnT8 is the main transporter responsible for Zn²⁺ accumulation in insulin granules because its ablation reduced the zinc exocytotic events by 99% (22). Here, we showed that glucose similarly inhibited glucagon secretion of ZnT8^+/+ and ZnT8^−/− islets. This confirms previous reports (49,50) and excludes Zn²⁺ as an inhibitory paracrine signal mediating the glucagonostatic effect of glucose.

SST exerts a tonic inhibition on insulin and glucagon secretion. Glucose can inhibit glucagon release independently of Zn²⁺ released from β-cells, K_ATP channels, and SST. Participation of these last two factors in the glucagonostatic effect of glucose, however, cannot be excluded. Closure of K_ATP channels controls glucagon secretion by two mechanisms, a direct stimulation of α-cells and an indirect inhibition via SST released from δ-cells. The net effect on glucagon release results from a balance between both effects. This might explain why Tolb reproduces the glucagonostatic effect of glucose in some conditions, whereas it stimulates glucagon release in others. This latter situation should be considered during treatment of type 2 diabetic patients by sulfonylureas because stimulation of glucagon secretion by the drugs could contribute to the unwanted hyperglucagonemia found in diabetes. Our study also calls for a careful examination of δ-cell function in diabetes.

This work was supported by Grant 3.4554.10 from the Fonds de la Recherche Scientifique Médicale (Brussels, Belgium), by Grant ARC (05/10-328) from the General Direction of Scientific Research of the French Community of Belgium, by the Interuniversity Poles of Attraction Programme (PAI 6/40) from the Belgian Science Policy, by Juvenile Diabetes Research Foundation Project Grant 2007-685, and by a European Foundation for the Study of Diabetes/Boehringer Ingelheim grant.

P.G. is Research Director of the Fonds National de la Recherche Scientifique, Brussels. M.A.R. is Chargé de Recherches at INSERM, Paris, France. F.C. is employed by Mellitech SAS, Grenoble, France. No other potential conflicts of interest relevant to this article were reported.

R.C.-X. and P.G. wrote the manuscript. R.C.-X. A.G.-R., N.A., L.A.N., H.-Y.C., M.A.R., and P.G. researched data. M.A.R. contributed to discussion. M.A.R. and F.C.S. reviewed the manuscript. M.A.R., F.C., and F.C.S. edited the manuscript. P.G. is the guarantor of this work, had full access to all the data, and takes full responsibility for the integrity of data and the accuracy of data analysis.

The authors thank F. Knockaert, V. Massé, and S. Godecharles (all three from Université Catholique de Louvain, Brussels, Belgium) for technical assistance, J. Bryan (Pacific Northwest Diabetes Research Institute, Seattle, Washington) for the gift of Sur1^−/− mice, J.C. Henquin (Université Catholique de Louvain, Brussels, Belgium) for providing access to his homemade insulin assay, and I. Robinson (National Institute for Medical Research, London, U.K.) and M. Low (Oregon Health and Science University, Portland, Oregon) for the gift of Sst^−/− mice.