Glucose-Dependent Modulation of Insulin Secretion and Intracellular Calcium Ions by GKA50, a Glucokinase Activator

Male Sprague-Dawley rats (230–280 g) were humanely killed according to Home Office–approved procedures. The pancreas was then removed and partially digested by incubating at 37°C in HEPES-balanced Krebs-Ringer phosphate buffer (KRH), composed of 129 mmol/l NaCl, 5 mmol/l NaHCO₃, 4.8 mmol/l KCl, 1.2 mmol/l KH₂PO₄, 1.2 mmol/l MgSO₄, 10 mmol/l HEPES, 2.5 mmol/l CaCl₂, 5.6 mmol/l glucose, and 0.1% BSA (pH 7.4, NaOH) supplemented with 1.38 mg/ml liberase Rodent Islet isolation formulation (Roche, Welwyn Garden City, U.K.) and 1.83 mg/ml each of egg white and soybean trypsin inhibitors. The pancreas was further digested by incubating at 37°C in KRH supplemented with 0.72 mg/ml liberase RI and 1.55 mg/ml each of egg white and soybean trypsin inhibitors until islets were dissociated from the exocrine tissue. Islets were handpicked and cultured for 1–3 days in RPMI 1640 (5 mmol/l glucose; Invitrogen, Paisley, Scotland, U.K.), 10% FCS (Invitrogen), and penicillin/streptomycin (100 units/ml, 0.1 mg/ml; Invitrogen) at 37°C in a 5% CO₂ humidified atmosphere.

Islets from C57BL6 mice were isolated using a similar protocol except that volumes were reduced accordingly and culture was in 11 mmol/l glucose RPMI 1640. Human islets were obtained on separate occasions from two donors and provided by United Kingdom Human Tissue Bank (Leicester, U.K.) and isolated in the Islet Research Laboratory (Worcester, U.K.) in accordance with current ethical guidelines pertaining to the use of human tissue in research. Cold ischemia time was 10 h. Islets were isolated as previously described (33) and, upon receipt, were cultured in RPMI (11 mmol/l glucose; Invitrogen), 10% FCS, and penicillin/streptomycin (100 units/ml, 0.1 mg/ml) at 30°C in a 5% CO₂ for at least 2 h before glucose-stimulated insulin secretion (GSIS) experiments.

MIN6 cells (passages 25–40) (34,35) were maintained in Dulbecco's modified Eagle's medium (DMEM; 25 mmol/l glucose; Invitrogen) supplemented with 15% FCS, penicillin/streptomycin (100 units/ml, 0.1 mg/ml), 2 mmol/l/l l-glutamine, and 5 μl/l 2-mercaptoethanol. Cells were incubated at 37°C in a 5% CO₂ humidified incubator. The culture medium was changed every 48–72 h.

GSIS from MIN6 cells and islets was determined using a static incubation protocol. For MIN6, cells were cultured in 96-well plates until ∼80% confluent, and the medium was changed every 48–72 h. On the day of experiment, growth medium was removed, and the cells were washed twice with glucose-free KRH supplemented with 0.1% radioimmunoassay grade BSA. Cells were preincubated for 1 h in 5% CO₂ at 37°C in KRH (1 mmol/l glucose). Incubation medium was removed, and the cells were washed once in glucose-free KRH. The cells were then incubated for 1 h in KRH containing the appropriate treatment. There were two modifications to this standard protocol. First, when using a high extracellular concentration of K⁺, NaCl was replaced by equimolar KCl in the KRH buffer; second, when investigating the action of GKA50 in the presence of glucose analogs, cells were preincubated in a glucose-free KRH for 30 min. For all experiments, incubation medium was collected, and the amount of secreted insulin was determined using ELISA (standard and high range kits; Mercodia, U.K.). For those experiments in which protein content was measured, MIN6 cells were solubilized by addition of 2% Triton X-100 solution, and the protein content was determined using the Bradford method (Bio-Rad, Hemel Hempstead, U.K.).

Insulin secretion from isolated islets was determined using a similar protocol. Briefly, the islets (rat, mouse, and human) were handpicked in groups of three and incubated in KRH (1 and 3 mmol/l glucose for rodent and human islets, respectively) for 30 or 60 min at 37°C and in 5% CO₂. After the appropriate preincubation period, KRH containing the appropriate treatment was added, and the islets were incubated for 1–2 h at 37°C and in 5% CO₂. After incubation, the islets were centrifuged, the medium was collected, and the amount of secreted insulin was determined using standard ELISA or homogeneous time-resolved fluorescence assay (CisBio, Bognols/Cèze, France). MIN6 and rodent islet datasets were obtained from three to six replications and repeated a minimum of three times. Human islet experiments were obtained from six replicates and repeated twice.

To form small clusters of cells, MIN6 cells were cultured for 3–4 days in DMEM (as above) on poly-d-ornithine–coated coverslips (10 μg/ml) in 24-well plates. Before experimentation, cells were loaded with Fura-2AM and pluronic acid (Invitrogen, Cambridge, U.K.) at a final concentration of 6 μmol/l for 1 h at 37°C. Mouse islets were cultured for up to 36 h on fibronectin-coated coverslips (10 μg/ml) in RPMI 1640 (5 mmol/l glucose, as above). Islets were loaded with Fura-2AM and pluronic acid to a final concentration of 6 μmol/l and 0.06%, respectively, for 1 h at 37°C.

Estimated internal calcium concentration ([Ca²⁺]_i) was measured by mounting the coverslips to form the base of a perifusion chamber (Harvard Apparatus, Edenbridge, U.K.). The cells were excited by 340 and 380 nm with a cycle time of 1.3 s using a monochromator (TILL Photonics, Planegg, Germany), attached to an Axiovert epifluorescence microscope (Zeiss, Jena, Germany). Images were collected by a Quantix CCD camera (Roper Scientific, Marlow, Bucks, U.K.) using the Metafluor software (Roper, U.K.). An in vitro calibration procedure was performed to determine estimated changes in [Ca²⁺]_i using the K_d value of Fura-2 as 224 nmol/l (36).

The control perifusion solution contained 137 mmol/l NaCl, 5.36 mmol/l KCl, 0.81 mmol/l MgSO₄, 0.34 mmol/l Na₂HPO₄, 0.44 mmol/l KH₂PO₄, 1.26 mmol/l CaCl₂, 4.17 mmol/l NaHCO₃, 10 mmol/l HEPES, and 2.02 mmol/l glucose (pH 7.4, NaOH). In solutions containing a high extracellular concentration of K⁺, NaCl was replaced by equimolar KCl.

Measurements of recombinant enzyme activity of GKA50 for mouse and rat pancreatic glucokinase were performed using published procedures (25).

Data are presented as means ± SE. Comparisons were made using unpaired Student's t tests and one-way ANOVA, as appropriate. In Fig. 1B, data are plotted as fold change in insulin secretion against a log scale, upper and lower 95% confidence limits (CL) were calculated, and P values were derived from a two-sided Student's t test. Curves were fitted using the four-parameter Hill equation using Sigma Plot 2001 (Systat Software).

All reagents were of analytical grade and, except where noted, were from Sigma (Gillingham, U.K.). GKA50 was synthesized by AstraZeneca (30).

The effects of GKA50 on GSIS from rodent and human islets and MIN6 cells are summarized in Figs. 1 and 2. These experiments revealed that GKA50 caused a shift in the measured glucose sensitivity, resulting in insulin release at lower glucose concentrations when compared with control islets/cells. Figure 1A and B summarizes the actions of 1 μmol/l GKA50 on GSIS in mouse islets. These data show a clear leftward shift in the concentration response profile to glucose in the presence of GKA50. This was significant from sub-stimulatory glucose concentrations (Fig. 1B) to 10 mmol/l glucose (Fig. 1A). Thus at 1 mmol/l glucose, we found that GKA50 (1 μmol/l, all n = 7; Fig. 1B) evoked a 1.6-fold (95% CL 1- to 2.5-fold; P = 0.04) increase in insulin release; at 2 mmol/l glucose, this was 3.4-fold (1.8- to 6.5-fold; P = 0.002); at 3 mmol/l, this was 5.5-fold (2.3- to 12.9-fold; P = 0.002); and at 5 mmol/l glucose, this was 8.3-fold (4.4- to 15.5-fold; P = 0.0002). Thus, at these glucose concentrations, the magnitude of the GKA50 effect on insulin secretion was increased greatly at higher glucose concentrations. At 20 mmol/l glucose, although GKA50 was found to potentiate GSIS, this failed to reach statistical significance. In human islets, 1 μmol/l GKA50 enhanced insulin release at 3 mmol/l glucose, and this was significant at 5 mmol/l glucose (Fig. 1C, representative of two separate experiments with different donors). The glucose sensitivity of rat islets (Fig. 1D) and MIN6 cells (Fig. 2A) was also similarly increased by GKA50 and was found to be significantly different at 3 and 2 mmol/l glucose, respectively. GSIS half-maximal effective concentration (EC₅₀) values were 9.3 ± 0.1 mmol/l (n = 5, vehicle control) versus 5.9 ± 0.2 mmol/l (n = 5, 1 μmol/l GKA50 treated) for rat islets and 13.2 ± 0.4 mmol/l (n = 5, control) versus 2.5 ± 0.2 mmol/l (n = 7, GKA50 treated) for MIN6 cells. However, the maximal rate of insulin secretion from both isolated rat islets (2.5 ± 0.02 vs. 2.5 ± 0.02 ng · islet⁻¹ · h⁻¹, control and treated, respectively) and MIN6 cells (65 ± 2.6 vs. 65.21 ± 1.3 pmol · mg⁻¹ · h⁻¹, control and treated, respectively) was not significantly affected by the addition of 1 μmol/l GKA50 (Figs. 1D and 2A, respectively).

After our initial evaluation in islets, we used MIN6 cells to investigate the concentration responsiveness to GKA50 (1 nmol/l to 10 μmol/l) over a range of glucose concentrations (Fig. 2B). In the absence of glucose, GKA50, even at concentrations up to 30 μmol/l, was unable to modulate insulin secretion (not shown, n = 3). However, in the presence of glucose (2–5 mmol/l), GKA50 exerted a concentration-dependent increase in insulin secretion with an EC₅₀ at 5 mmol/l glucose of 0.3 ± 0.2 μmol/l (Fig. 2B, n = 4). Depolarization of MIN6 cells with 40 mmol/l KCl was also found to enhance GSIS at all glucose concentrations tested (1–5 mmol/l), but there was no further action of 1 μmol/l GKA50 on insulin release under these conditions (Fig. 2C, n = 4). These data suggest that modulation of voltage-dependent Ca²⁺ influx is the principal mechanism associated with the action of GKA50 on insulin-secreting cells.

Because glucose facilitates insulin secretion through changes in the free [Ca²⁺]_i and Ca²⁺-dependent exocytosis, we next examined the actions of GKA50 on [Ca²⁺]_i in rodent islets and MIN6 cells. In mouse islets, elevating the glucose concentration from 2 mmol/l to 8 and 15 mmol/l initiated an initial decrease in [Ca²⁺]_i, followed by a marked rise in [Ca²⁺]_i and then oscillations (Fig. 3A, n = 5). Although glucose was without effect on [Ca²⁺]_i at concentrations <5 mmol/l in control mouse islets, in the presence of 1 μmol/l GKA50, changes in [Ca²⁺]_i were seen (Fig. 3B–D). Thus, in the presence of 2 mmol/l glucose, 1 μmol/l GKA50 caused an initial decrease in [Ca²⁺]_i followed by a marked, transient increase in [Ca²⁺]_i and then an elevated, sustained rise (Fig. 3B, n = 3). In the presence of 5 mmol/l glucose, 1 μmol/l GKA50 typically caused an immediate increase in cytosolic Ca²⁺ followed by [Ca²⁺]_i oscillations (Fig. 3C, n = 4). Elevating glucose from 5 to 10 mmol/l in the presence of GKA50 further increased [Ca²⁺]_i (Fig. 3D, n = 5). GKA50 (1 μmol/l) was without effect on [Ca²⁺]_i in the absence of glucose (n = 4) (not shown). Fig. 3E shows the results of experiments to examine interactions between GKA50 and glucose using rat islets. In these experiments, 1 μmol/l GKA50 failed to elevate [Ca²⁺]_i at 2 mmol/l glucose but was effective at higher glucose concentrations (Fig. 3E, n = 3). Finally, we also found that GKA50 modulated glucose-dependent increases in [Ca²⁺]_i in MIN6 cells (n = 15 from 16 cells, Fig. 3F).

The sulfonylurea tolbutamide inhibits ATP-sensitive K⁺ (K_ATP) channels in β-cells and mediates insulin release through depolarization-dependent mechanisms. We therefore compared the actions of GKA50 and tolbutamide using MIN6 cells and rat islets. In MIN6 cells, both GKA50 (1 μmol/l, Fig. 2) and tolbutamide (200 μmol/l, Fig. 4A) stimulated insulin release at 2 mmol/l glucose. However, in rat islets (Figs. 1D and 4B), whereas tolbutamide initiated insulin release at 2 mmol/l glucose, GKA50 was without effect. Tolbutamide was unable to further enhance 1 μmol/l GKA50-stimulated insulin secretion from either MIN6 cells or rat islets (Fig. 4A and B). Figure 4C shows typical response times for the actions of tolbutamide and GKA50 in islets compared with responses to a 40 mmol/l KCl-induced depolarization. Although both compounds evoke similar magnitudes of [Ca²⁺]_i rises, in comparison with the effects of tolbutamide, the actions of GKA50 were significantly slower in onset. The average time to reach 50% of the maximal rise in [Ca²⁺]_i after stimulation was 44 ± 6 s (n = 24) for tolbutamide versus 392 ± 15 s (n = 15) for GKA50.

Metabolic-dependent modulation of K_ATP channels is responsible for glucose-dependent rises in [Ca²⁺]_i in β-cells. We therefore investigated how the direct activation of K_ATP channels by diazoxide and inhibition of glucose metabolism by mannoheptulose, 5-thioglucose (5-TG), and 3-O-methoxyglucose (OMeG) affected responses to GKA50. Figure 5A and B summarizes the actions of diazoxide (200 μmol/l) on the inhibition of glucose-induced and 1 μmol/l GKA50-induced insulin secretion in MIN6 cells (n = 4) and rat islets (n = 3), respectively. Because diazoxide hyperpolarizes the cell membrane potential via activation of K_ATP channels, these effects were also associated with a reversible diazoxide-induced inhibition of cytosolic Ca²⁺ (Fig. 5C, n = 8).

Inhibition of glucokinase by mannoheptulose and 5-TG was found to selectively inhibit 10 μmol/l GKA50-induced insulin release, whereas the actions of 50 μmol/l tolbutamide were not significantly affected (Fig. 5D, n = 3). Consistent with this data, measurements of [Ca²⁺]_i in mouse islets revealed that there was no significant action of the nonmetabolizable glucose analog, OMeG, on tolbutamide (100 μmol/l)-induced rises in [Ca²⁺]_i (average change in [Ca²⁺]_i = 111 ± 16 nmol/l, n = 10), whereas the actions of 1 μmol/l GKA50 were attenuated (average change in [Ca²⁺]_i, 15 ± 4 nmol/l, n = 11) (Fig. 6).

Glucokinase is the glucose sensor for glucose-stimulated insulin release from β-cells and is the rate-limiting step in the glucose responsiveness of glucosensitive cells. Glucokinase therefore plays a major role in regulating carbohydrate metabolism in man, and defects in the glucokinase gene cause either hyperglycemia as a result of “loss of function” mutations or hypoglycemia through “gain of function” mutations.

Recently, several activators of glucokinase have been described in the literature (24–32), and the potential use of these agents, GKAs, in the therapeutic management of type 2 diabetes has been reviewed (37–39). This group of compounds also includes the pyridine acid–based molecule, GKA50, which was recently shown to have a significant blood glucose–lowering efficacy in vivo in both control and insulin-resistant/hyperinsulinemic rats (32). Because the actions of GKA50 include the modulation of insulin release in vivo, we have used isolated islets and MIN6 cells to examine the effects of GKA50 on both the regulated release of insulin and the control of intracellular Ca²⁺ signaling events.

In this series of experiments, we used mouse, rat, and human islets to show that GKA50 is a novel modulator of insulin release. We then used rat islets and MIN6 insulin-secreting cells to quantify the action of GKA50 on EC₅₀ values and maximal rates of insulin secretion. Although there may be some differences in the action of GKA50 across species (discussed below), we were able to make the following general conclusions. First, our data show that GKA50 was unable to elevate changes in the cytosolic Ca²⁺ concentration ([Ca²⁺]_i) or to promote insulin release in the absence of glucose, even up to activator concentrations as high as 30 μmol/l. Second, GKA50-induced insulin release is dependent on an increase in [Ca²⁺]_i, and the threshold for this occurs at sub-stimulatory glucose concentrations. Collectively, these data indicate that GKA50 cannot initiate insulin release from β-cells and that its actions are strictly dependent on glucose availability. This is consistent with the modes of action of other GKAs (24,28) and distinctly different from that of the sulfonylureas, which act independently of glucose availability (40–45). Third, GKA50, like other GKAs, caused a leftward shift in the EC₅₀ value for GSIS without affecting the maximal rates of insulin release. Finally, we describe for the first time that metabolic inhibition selectively eliminated the actions of GKA50 but not tolbutamide on insulin release and changes in [Ca²⁺]_i. Activation of K_ATP channels by diazoxide also reversed the action of GKA50. Because glucose metabolism leads to a K_ATP channel–dependent depolarization of the membrane, voltage-dependent Ca²⁺ influx, and Ca²⁺-dependent exocytosis, these findings are consistent with the proposed actions of both agents. Through enhancing glucose metabolism, GKA50 acts in a glucose-dependent manner, and metabolic inhibition, resulting in the activation of K_ATP channels, will eliminate these responses. Under the same conditions, the actions of tolbutamide are preserved because it acts directly on K_ATP channels and is therefore independent of glucose availability. Consistent with this mechanism of action, we were also able to demonstrate that, in comparison with tolbutamide, GKA50-induced changes in [Ca²⁺]_i that were biphasic at low glucose concentrations and were significantly slower at reaching maximal responses (Fig. 4).

The profile of GKA50-induced potentiation of insulin release is similar to that recently reported for LY2121260 (28), but not that of Compound A, which has a profound effect on insulin release at low glucose concentrations and enhances insulin release at elevated glucose concentrations (31). In rat islets, 1 μmol/l GKA50, 1 μmol/l LY2121260, and 10 μmol/l RO-28-1675 had similar effects on insulin release at 4–5 mmol/l glucose, increasing insulin release by approximately threefold, which is approximately equal to the amount of insulin released by 8 mmol/l glucose (24,28). Each of the compounds at 1 μmol/l failed to further enhance insulin release in the presence of 15 mmol/l glucose or above.

Here, we also examined the actions of the GKA on mouse islets for the first time. The data in Figs. 1 and 2 appear to suggest differences in the sensitivity of rat and mouse islets to GKA50. In these experiments, whereas GKA50 failed to have any significant effect on insulin release at 2 mmol/l glucose in rat islets, a 3.5-fold increase in insulin release was induced by GKA50 at the same concentration in mouse islets. However, a closer review of the data reveals that this is not likely to be because of species differences in the sensitivity of GKA50 for glucokinase but rather a consequence of different [Ca²⁺]_i responses after stimulation. At 2 mmol/l glucose, mouse islets responded to GKA50 through a biphasic response in [Ca²⁺]_i, an initial decrease followed by a rise in cytosolic Ca²⁺. The first stage of this response is mediated by Ca²⁺ sequestration to cytosolic stores (46,47) and is then followed by Ca²⁺ influx leading to insulin release. In rat islets, GKA50 caused only a sustained decreased [Ca²⁺]_i without a subsequent rise (Fig. 3B and E). This suggests that there is little difference in the sensitivity of glucokinase for GKA50 in rat versus mouse islets, and alone, it would not account for the differences in GKA50-mediated insulin release at this concentration of glucose. Recombinant enzyme activity assays of GKA50 for mouse and rat pancreatic glucokinase were very similar with GKA50, showing an EC₅₀ of 56 nmol/l (n = 7, 95% CL 40–79 nmol/l) at the mouse glucokinase and 65 nmol/l (n = 23, 53–79 nmol/l) at the rat glucokinase. RO-28-1675 (24) and LY2121260 (28) also failed to evoke insulin release at 2 mmol/l glucose, although both agents were effective at sub-stimulatory glucose concentrations.

Given the powerful ability of GKAs to modulate insulin secretion, it is of benefit to understand how these molecules interact with, and subsequently activate, glucokinase and glucose metabolism. The exact mechanism is currently unknown. However, there is evidence to suggest that GKAs alter the kinetics by which glucokinase shifts between active and inactive conformations. Recently, the crystal structure of human glucokinase has been solved in the presence of glucose and a GKA (26). The crystal structures revealed that glucokinase can adopt three distinct conformations termed closed, open, and “super open,” which is inactive (26). This hypothesis allows for glucokinase to have two distinct catalytic cycles (termed “fast” and “slow,” respectively) and that the ratio of glucokinase in each of the two cycles is determined by the glucose concentration and accounts for the sigmoidal kinetics of glucokinase. Thus, at low glucose concentrations, the majority of glucokinase will enter the “slow” cycle that cycles through the inactive super open configuration; whereas at higher glucose concentrations, the majority of glucokinase will favor the fast cycle through the closed and open conformations and avoid the super open conformation (26). GKAs shift the equilibrium between the two cycles in favor of the fast kinetic cycle and stabilize glucokinase in the active form (48). This will account for a reduced S_0.5; however, the GKAs all have different activating properties. All of the known GKAs decrease the S_0.5 of glucokinase for glucose, and some, but not all, also increase V_max. It is therefore notable that GKA50 does not increase V_max. Hence, there must be various subtle effects on glucokinase conformation caused by the different GKAs. Regardless of the exact kinetic changes brought about by the different GKAs, the end result is a lowering of plasma glucose brought about by increased insulin secretion and/or increased hepatic glucose metabolism. With the data presented here, GKA50 has now been shown to significantly increase both parameters (25,32).

In conclusion, we have shown for the first time that an activator of glucokinase, GKA50, acts as a Ca²⁺-mediated glucose-dependent insulin secretagogue in human and rodent islets. No changes in intracellular Ca²⁺ or insulin release were seen in the absence of glucose, and GKA50 cannot alone initiate insulin release. In man, we would expect GKA50 to cause a similar leftward shift in the EC₅₀ value for glucose in insulin secretion, and in type 2 diabetic islets, this would translate into both an increase in GSIS and normalization of the rightward-shifted glucose dose-response profile seen in these islets. These data therefore further support a role for GKAs as novel utilities in the therapeutic management of diabetes.

M.J.D. has received a project grant from AstraZeneca.

We thank Heather Wightman (AstraZeneca) for her contributions to the isolated enzyme assays.