ATP-Sensitive K⁺ Channel Mediates the Zinc Switch-Off Signal for Glucagon Response During Glucose Deprivation

Male and female wild-type C57BL6/J mice (8–15 weeks old) were purchased from Jackson Labs (JAX no. 000664). SUR1 knockout mice (SUR1KO) were bred at PNDRI (10). Animals were maintained with a 12-h light-dark cycle and constant temperature and were given free access to food and water. All experiments were approved by the PNDRI Institutional Animal Care and Use Committee.

On the day of pancreas removal, animals were anesthetized with ketamine-xylazine and then killed by removing blood from the heart. Pancreata were cannulated for infusion of collagenase and then removed from the animal for processing. Mice pancreata were processed as described by Tanaka (12), using 0.75 mg/ml collagenase in wild-type mice and 1.5 mg/ml collagenase in SUR1KO mice. A higher concentration was used for SUR1KO mouse islets since they tended to adhere to ducts when the lower concentration was used. Once processed, islets were hand picked and cultured overnight in RPMI 1640 media with 11.1 mmol/l glucose, 10% FBS, and antibiotic/antimycotic (100 units/ml penicillin G and 0.1 mg/ml streptomycin sulfate and 0.25 μg/ml amphoteracin B). The pool of islets for each experiment was derived from at least eight animals. In some experiments STZ was used to render pancreatic islets devoid of β-cells. In these experiments, isolated islets were cultured for 2 h in RPMI media with 11.1 mmol/l glucose, 10% FBS, and antibiotic/antimycotic then for 30 min in RPMI media with STZ (5 mmol/l), 11.1 mmol/l glucose, 0.8% FBS, and antibiotic/antimycotic. Then, STZ was washed out of the islets and they were cultured overnight in RPMI media with 11.1 mmol/l glucose, 10% FBS, and antibiotic/antimycotic.

The day after isolation, islets were split into groups of 200 islets. Each group was placed into one of the individual perifusion chambers for simultaneous study. The islets were placed in the chambers on top of mesh filters (60 μm) small enough to prevent islets from escaping into the effluent. Each experiment was preceded by a 60-min stabilization period using a perifusion of Kreb's Ringer buffer (KRB) containing 0.1% BSA, 0.238% HEPES, 0.1 mmol/l isobutylmethylxanthine (IBMx), and 2.8 mmol/l glucose. Thereafter, experimental KRB buffers contained 0.1% BSA, 0.238% HEPES, 0.1 mmol/l IBMx, an amino acid mixture (3), and either 16.7 or 0 mmol/l glucose. When STZ-treated islets were used, 30 μmol/l zinc chloride was also included in the buffer. Samples for measurement of insulin and glucagon were collected at −10, −5, −2, −1, and 0 min before and 2, 5, 10, 15, 20, 25, and 30 min after endogenous insulin (Fig. 1) and exogenous zinc (Figs. 2, 3 and 4) switch-off signals were provided at time 0.

The rationale for choosing the concentration of 30 μmol/l for zinc chloride is provided in a previous publication (9). During exposure to physiologic or elevated glucose levels, the zinc-insulin complex is released from the β-cell into the periportal circulation that travels to α-cells. This design assumes that zinc dissociates from insulin when exposed to the higher pH of blood (13) and acts to open α-cell K_ATP channels, which causes tonic suppression of glucagon secretion. It also assumes that zinc is normally bound to protein in blood, as it dissociates from insulin in the periportal circulation and circulates to the α-cell, and that zinc concentration during glucose stimulation of insulin release is fourfold greater (9) than the 10 mmol/l concentration found in systemically circulating blood. This fourfold higher concentration of zinc is assumed to drop to the same levels found in systemically circulating blood during glucose deprivation and cessation of insulin secretion.

Nifedipine: a 1 μmol/l perifusion solution was prepared from a 1 mmol/l stock solution in 100% ethanol. Diazoxide: a 50 μmol/l perifusion solution was prepared from a 50 mmol/l stock solution in 0.1 mol/l NaOH. Tolbutamide: a 50 μmol/l perifusion solution was prepared from a 50 mmol/l stock solution in 100% ethanol. Epinephrine: a 1 μmol/l perifusion solution was prepared from a 1 mmol/l stock solution in 1 mol/l HCl. Epinephrine was infused in the absence of glucose and the amino acid mixture for 5–7 min at the end of each experiment as alternative stimulus for glucagon secretion to ensure that α-cells were still functional after STZ treatment. Nifedipine, diazoxide, tolbutamide, and epinephrine were purchased from Sigma.

Perifusion samples were collected into tubes containing 500 units of aprotinin to prevent degradation of glucagon. Insulin was measured by radioimmunoassay according to the method of Morgan and Lazarow (14). Glucagon was measured using a glucagon radioimmunoassay kit (Millipore, St. Charles, MO).

Results for each perifusion experiment were calculated as means ± SE. Measurements of at least three lanes were considered as n = 1. Comparisons were made by paired Student's t test, unpaired Student's t test with or without Welch correction, Wilcoxon matched-pairs signed-ranks test, and ANOVA, where appropriate. P values <0.05 were considered significant.

Wild-type mouse islets were perifused with 16.7 mmol/l glucose for 30 min to stimulate endogenous insulin secretion. At time 0, the perifusate was switched to a 0 mmol/l glucose buffer for 30 min so that endogenous insulin secretion would decrease. In control experiments, at time 0, 16.7 mmol/l glucose was continued rather than discontinued. Insulin levels in the effluent decreased when the 16.7 mmol/l glucose perifusion was discontinued (0 min = 108 ± 7 μU/ml, 30 min = 5 ± 1 μU/ml, n = 8; P < 0.001) (Fig. 1,A). When the 16.7 mmol/l glucose perifusion was not discontinued, insulin levels did not change significantly (0 min = 107 ± 10 μU/ml, 30 min = 90 ± 8 μU/ml, n = 7; P = NS) (Fig. 1,A). Glucagon levels increased significantly only when 16.7 mmol/l glucose buffer was discontinued (0 min = 61 ± 8 pg/ml vs. 30 min = 133 ± 26 pg/ml, n = 8; P < 0.01) (Fig. 1,B) and not when the glucose perifusate was continued for the duration of the experiment (0 min = 54 ± 9 pg/ml vs. 30 min = 71 ± 14 pg/ml, n = 7; P = NS) (Fig. 1 B).

STZ-induced diabetic wild-type islets were perifused with 16.7 mmol/l glucose and zinc (30 μmol/l) for 30 min. Then, at time 0, both glucose and zinc perifusions were discontinued. In controls, either glucose or zinc alone was discontinued. In other experiments, at time 0 nifedipine was added to the perifusate when zinc was switched off during glucose deprivation (Fig. 2). Glucagon levels increased only when zinc perifusion was switched off during glucose deprivation (0 min = 28 ± 6 pg/ml vs. 30 min = 50 ± 4 pg/ml, n = 5; P < 0.02) (Fig. 2). If the zinc perifusion was continued during glucose deprivation, glucagon levels did not rise significantly (0 min = 15 ± 2 pg/ml vs. 30 min = 23 ± 3 pg/ml, n = 4; P = NS) (Fig. 2). Similarly, if perifusion containing glucose was continued after the zinc perifusion was switched off, the glucagon levels did not rise (0 min = 18 ± 4 pg/ml vs. 30 min = 25 ± 7 pg/ml, n = 4; P = NS) (Fig. 2). In the presence of nifedipine, glucagon levels after the zinc switch-off signal during glucose deprivation did not increase but significantly decreased (0 min = 36 ± 5 pg/ml vs. 30 min = 11 ± 1 pg/ml, n = 3; P < 0.05) (Fig. 2). Insulin levels were undetectable throughout.

STZ-induced diabetic wild-type mouse islets were perifused with 16.7 mmol/l glucose and diazoxide (50 μmol/l) for 30 min (−60 to −30 min), followed by a perifusate containing 16.7 mmol/l glucose, diazoxide, and zinc (30 μmol/l) for another 30 min (−30 to 0 min). At 0 min, both glucose and zinc were discontinued and diazoxide infusion only was continued for 30 more minutes (Fig. 3). Glucagon secretion remained suppressed despite the absence of glucose and zinc in the perifusate (0 min = 10 ± 0 pg/ml vs. 30 min = 14 ± 2 pg/ml, n = 3; P = NS) (Fig. 3). Insulin levels were undetectable throughout.

STZ-induced diabetic wild-type mouse islets were perifused with 16.7 mmol/l glucose and tolbutamide (50 μmol/l) for 30 min (−60 to −30 min), followed by a perifusate containing 16.7 mmol/l glucose, tolbutamide, and zinc (30 μmol/l) for another 30 min (−30 to 0 min). At 0 min, both glucose and zinc were discontinued and tolbutamide infusion only was continued for 30 more minutes (Fig. 4). Glucagon secretion reached a higher baseline than in the previous STZ-induced diabetic islet experiments and it did not rise significantly when zinc was switched off during glucose deprivation (0 min = 53 ± 5 pg/ml vs. 30 min = 67 ± 3 pg/ml, n = 3; P = NS) (Fig. 4). Insulin levels were undetectable throughout.

An exogenous zinc switch-off signal during glucose deprivation was provided to STZ-induced diabetic SUR1KO mouse islets by infusing 16.7 mmol/l glucose and zinc (30 μmol/l) for 30 min and then switching to a perifusate without glucose and zinc at time 0 (Fig. 5). No significant glucagon increase was observed despite the provision of a zinc switch-off signal during glucose deprivation (0 min = 55 ± 17 pg/ml vs. 30 min = 61 ± 20 pg/ml; n = 4, P = NS). Insulin levels were undetectable throughout.

Glucagon levels before and after infusion of drugs that interfere with K_ATP channel activity in wild-type or SUR1KO islets are shown in Fig. 6,A. Compared with wild-type values, depolarizing conditions (tolbutamide, knockout) were associated with higher glucagon levels (P < 0.05), and the polarized condition (diazoxide) was associated with lower glucagon levels (P < 0.05) at baseline (time 0). Glucagon responses to changes in the perifusate in STZ-induced diabetic mouse islets are compared as area under the curve in Fig. 6 B. Glucagon response was statistically significant (P < 0.01–0.001) only when a zinc switch-off signal during glucose deprivation was provided to wild-type islets pretreated with STZ. No statistically significant glucagon responses were observed if zinc or glucose perifusions alone were discontinued or a zinc switch-off signal during glucose deprivation was provided in the presence of nifedipine, diazoxide, tolbutamide, or to SUR1KO islets.

α-Cell function was tested at the end of most experiments using a 5- to 7-min infusion of epinephrine as alternative stimulus. Glucagon responses were intact in all the conditions involving wild-type and knockout islets (Fig. 7).

These experiments were designed to identify the mechanism through which zinc provides a switch-off signal for glucagon release during glucose deprivation. We previously reported that switching off zinc and insulin, but not zinc-free insulin, triggered glucagon secretion during hypoglycemia in vivo in rats (9). The results from the first part of the current study are consistent with a role for cessation of zinc corelease with insulin as a switch-off signal to trigger glucagon secretion during glucose deprivation. Since our experiments used mouse islets, our data argue against the contention of Ravier and Rutter (11) that zinc does not regulate α-cells in mice and support the more recent findings by Gyulkhandanyan et al. (15).

The second part of our study was designed to identify the mechanism of action for zinc regulation of the α-cell. To achieve this goal, we used two approaches. The first approach was to pharmacologically manipulate the K_ATP channel with drugs prior to providing STZ-induced diabetic wild-type mouse islets with a zinc switch-off signal during glucose deprivation. The first drug was diazoxide that keeps the K_ATP channels open. The rationale was that if the zinc switch-off signal is mediated by a change in the K_ATP channel activity, then using diazoxide to prevent any further change in the K_ATP channel activity should blind the K_ATP channel to the zinc switch-off signal. Zinc removal from the perifusate during glucose deprivation induced no changes in glucagon secretion from diazoxide-treated islets. This suggests that by maintaining the K_ATP channels open, the drug masked and overrode the signal given by zinc removal. In other mechanistic studies, we interfered with the K_ATP channel activity using tolbutamide. This drug, which keeps K_ATP channels closed, led to the same result. Because of the tolbutamide-mediated prestimulation of K_ATP channel activity, the already closed K_ATP channels failed to respond to the signal provided by the switch-off of zinc during glucose deprivation. We also examined whether the post–K_ATP channel signal transduction pathway involved calcium channels. Nifedipine was used to prevent the calcium channels from opening. Since zinc does not act on calcium channels, no preinfusion (as with the diazoxide and tolbutamide experiments) with nifedipine was deemed necessary. In the presence of nifedipine, glucagon secretion failed to respond to the zinc switch-off signal during glucose deprivation. These results suggest that the pathway through which the zinc switch-off signal triggers glucagon secretion during glucose deprivation involves both K_ATP and calcium channels (Fig. 8). Our findings also demonstrate that the α-cell is responsive to the zinc switch-off signal only in the circumstance of glucose deprivation, consistent with the fact that glucose is widely recognized as a suppressant of glucagon secretion (16,17).

The second approach used a model derived from the mouse line C57BL6/J, whose SUR1 subunit of the K_ATP channel is knocked out (10). Islet K_ATP channels consist of two subunits: the K⁺ inward rectifier, Kir6.2, and the high-affinity sulfonylurea receptor, SUR1, combined in 4:4 stoichiometry to form the hetero-octomeric channel. The Kir6.2 subunit forms the conducting pore for K⁺ ions, while the SUR1 subunit acts as a regulatory subunit conferring sensitivity to openers and blockers (18,–20). A tight interaction between the two subunits is required for appropriate channel function. In our experiments, although SUR1KO mouse islets were exposed to an exogenous zinc switch-off signal during glucose deprivation, glucagon secretion failed to respond. Importantly, SUR1KO mouse islets, as well as wild-type islets, responded to the alternative stimulus epinephrine, a known stimulator of glucagon secretion (21). Glucagon responses to this stimulus indicate that α-cell glucagon synthesis and exocytotic mechanisms are intact.

Our data reinforce the hypothesis that an intercellular dialogue is critical within the pancreatic islet for optimal glycemic control (4) and that zinc participates to this dialogue and therefore to the regulation of glucagon secretion by acting through the SUR1 subunit of the K_ATP channels as reported by other groups (8,22,,–25). In this scenario, during exposure to physiologic or elevated glucose levels, the zinc-insulin complex is released from the β-cell into the periportal circulation that travels to α-cells. Zinc dissociates from insulin when exposed to the higher pH of blood (13) and acts to open α-cell K_ATP channels, which causes tonic suppression of glucagon secretion. However, during states of glucose deprivation, such as hypoglycemia, β-cells virtually cease releasing zinc-insulin. This results in cessation of zinc effects on α-cell K_ATP channels and closure of the channels, which in turn causes depolarization, voltage-dependent calcium channel opening, calcium entry, and glucagon exocytosis (Fig. 8).

In support of this mechanism is the observation that the glucagon values reached during the α-cell response to the zinc switch-off signal during glucose deprivation in wild-type mouse islets were very similar to the values of the glucagon baselines in wild-type mouse islets perifused with tolbutamide and in knockout mouse islets (Fig. 6 A). Thus, switching off zinc, closing the K_ATP channels by means of tolbutamide, and absent K_ATP channels (as occurs in the KO mice) depolarize the α-cell, causing Ca²⁺ channels to open and increase glucagon secretion. On the other hand, adding diazoxide to the perifusate keeps K_ATP channels open, supporting K⁺ efflux, which keeps the cell polarized and thus prevents Ca²⁺ channels from opening, resulting in less glucagon secretion. Different from previous studies by others, our observations were obtained from intact and STZ-induced diabetic mouse islets, using a dynamic system that specifically studied glucagon secretion during glucose deprivation. This technique more closely approximates the important clinical setting in which regulation of glucagon secretion is most critical (i.e., hypoglycemia). Moreover, our results support the hypothesis that the association of high basal glucagon in type 1 diabetes could at least in part be attributed to lack of insulin-bound zinc tonically suppressing α-cell function.

In conclusion, our results strongly implicate the α-cell K_ATP channel and calcium channel as obligate parts of the mechanism whereby zinc suppresses glucagon secretion from the α-cell and whereby a sudden absence of zinc secretion from the β-cell during glucose deprivation triggers glucagon secretion.

Supported by National Institutes of Health Grant NIDDK RO1 39994 (to R.P.R.) and RO1DK044311(to J.B.) and an American Diabetes Association Mentor-Based Grant (to R.P.R.).

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

Parts of this article have been presented in abstract form at the American Diabetes Association 69th Scientific Sessions, New Orleans, Louisiana, 5–9 June 2009.