P2Y Purinergic Potentiation of Glucose-Induced Insulin Secretion and Pancreatic Beta-Cell Metabolism

Experiments were performed in vitro in the isolated perfused pancreas from male Wistar albino rats (Charles River, Lyon, France). Rats weighing 340–360 g were housed in a light-controlled room and given free access to food and water in accordance with the national guidelines for the use and care of laboratory animals. After sodium pentobarbitone anesthesia (60 mg/kg i.p.), the pancreas was completely isolated from all the neighboring tissues, according to a technique previously described (15), and perfused in the presence of a 5 mmol/l glucose background. A 30-min adaptation period was allowed before the first sample was taken for insulin assay and was followed by a further 30-min control period before any change in glucose concentration, or in type of secretagogue, and ADPβS administration. Upon administration of various secretagogues in the presence or absence of ADPβS, pancreatic effluents were collected every minute during the first 5 min immediately after the beginning of the treatments. Additional samplings were performed after 10, 15, and 20 min. Effluents were measured in graduated test tubes and immediately frozen until insulin assay.

ADPβS [adenosine 5′-O-(2-thiodiphosphate)], glucose, sodium nitroprusside, tolbutamide, pyruvic acid methyl ester, succinic acid monomethyl ester, arginine hydrochloride, bovine serum albumin (fraction V), and diazoxide were purchased from Sigma Aldrich (St. Louis, MO). Other chemicals were obtained from various commercial sources at the greatest purity available.

Insulin concentrations were determined by the radioimmunological method of Herbert et al. (16) using purified rat insulin as standard (Linco Research, St. Charles, MO) and guinea pig antiporcine-insulin antiserum (ICN Biochemicals, Puteaux, France). The sensitivity of our assay was 0.06 ng/ml.

Insulin output was calculated by multiplying concentration in the effluent by pancreatic outflow rate measured during 1 min. Results are expressed as means ± SE. Integrated data were obtained by calculating the areas under the curves during the 20 or last 15 min (for arginine and KCl) of high glucose or of other secretagogue administration. Kinetics data were submitted to ANOVA followed by the multiple comparison test of Newman-Keuls and integrated data by the Student’s t test.

ADPβS (0.5 μmol/l), ineffective on the 5 mmol/l glucose background (data not shown), causes a threefold increase in insulin release induced by a rise in glucose concentration from 5 to 11 mmol/l (Fig. 1). The typical biphasic pattern of insulin response to glucose was retained with a mean integrated response reaching 845 ± 83 ng × 20 min vs. 251 ± 19 ng × 20 min in the absence of ADPβS (P < 0.001). As shown also in Fig. 1, the nitric oxide donor sodium nitroprusside (SNP) at a concentration of 300 μmol/l is able to drastically reduce the second phase of secretion in the presence of the P2Y receptor agonist; the last 15 min integrated response was reduced by 75%, reaching 171 ± 36 ng × 15 min vs. 678 ± 71 ng × 15 min in the absence of the metabolic inhibitor (P < 0.001).

A 5 mmol/l increase in the concentration of the biophysical depolarizing agent KCl provoked a biphasic response of insulin secretion. After a 5-min first phase, insulin secretion stabilized slightly over basal values. Addition of ADPβS resulted in the potentiation of the second phase (Fig. 2) with a 15-min area under the curve (AUC) reaching 210 ± 14 ng × 15 min vs. 93 ± 26 ng × 15 min in the absence of ADPβS (P < 0.05). Similar data could be observed with arginine as a depolarizing agent. When tolbutamide (185 μmol/l) was used as a nonmetabolic stimulus, β-cell membrane depolarization resulted also in a biphasic insulin response in the presence of the 5 mmol/l glucose concentration. Simultaneous administration of ADPβS markedly potentiated both phases of tolbutamide effect. As shown in Fig. 2, mean 20 min integrated insulin secretion reached 1,229 ± 121 ng × 20 min and 462 ± 51 ng × 20 min in the presence and the absence, respectively, of ADPβS (P < 0.001).

Methyl pyruvate (15 mmol/l) induced a biphasic insulin response with a 20-min AUC reaching 791 ± 102 ng × 20 min. Addition of ADPβS resulted in a slight but significant 45% increase (1,145 ± 99 ng × 20 min; P < 0.05; Fig. 2). In contrast, when methyl succinate (2.5 mmol/l) was used as a secretagogue, no further increase could be observed in the presence of the P2Y receptor agonist. Mean integrated data reached 603 ± 85 ng × 20 min and 681 ± 90 ng × 20 min in the presence and absence, respectively, of ADPβS. Similar data could be observed in the presence of 5 mmol/l methyl succinate (Fig. 2).

Administration of diazoxide and KCl, during a 15-min pretreatment on the 5 mmol/l glucose background, induced a slight increase in insulin secretion. Raising glucose concentration from 5 to 11 mmol/l provoked an increase in insulin secretion with a mean 20-min AUC reaching 366 ± 52 ng × 20 min (Fig. 3A). It is noteworthy that the first phase of insulin response was completely suppressed. Upon addition of ADPβS, β-cell response appeared markedly enhanced by 200%. It must be mentioned that the increase observed during the first 5 min is not due to restoration of a first phase since it also occurs with ADPβS in the absence of any change in glucose concentration (Fig. 3B). It may be ascribable to the potentation by ADPβS of the effect of a slight increase in intracellular Ca²⁺ during the diazoxide-KCl pretreatment, which also accounts for the moderate elevation of insulin secretion over basal values.

Our data demonstrate that the stable P2Y receptor agonist ADPβS potentiates glucose-induced insulin secretion during an acute glucose challenge that mimics a postprandial rise in the sugar concentration. This effect depends on glucose metabolism, is independent of a direct effect on K_ATP channels, but probably requires a rise in intracellular Ca²⁺ concentration. The involvement of glucose metabolism is clearly demonstrated by our experiments performed in the presence of SNP, a chemical NO donor. Indeed, NO has been reported to inhibit important glycolytic enzymes such as phosphofructokinase (17) and glyceraldehyde phosphate dehydrogenase (18) as well as mitochondrial cytochrome oxidase (19), and the NO donor drastically decreases the ADPβS-potentiating effect. Because glucose is known to trigger insulin secretion through its oxidative metabolism, rise in the ATP/ADP ratio, closure of K_ATP channels, membrane depolarization, opening of voltage-operated Ca²⁺ channels, Ca²⁺ influx, rise in cytoplasmic free Ca²⁺ concentration, and activation of the exocytotic machinery (20), we wondered whether ADPβS could be able to potentiate insulin secretion induced by a nonmetabolic membrane depolarization. Using tolbutamide, a sulfonylurea known as a K_ATP channel blocker, we show that the P2Y receptor agonist is still strongly effective, suggesting that a rise in cytosolic Ca²⁺ is probably involved in the potentiating effect. This is also supported by the data resulting from experiments using two other nonmetabolic depolarizing agents, KCl and arginine, but raises the question of a possible direct involvement of K_ATP channels. To investigate such a possibility, we first adapted the elegant method developed by Gembal et al. (21) to the “isolated perfused rat pancreas” model. Therefore, pancreatic β-cells were first hyperpolarized using the K_ATP channel opener diazoxide and then repolarized with an appropriate rise in KCl concentration. Because vascular smooth muscle cells are regulated by K_ATP channels (22), we used our results on pancreatic effluent output rate to determine the KCl concentration necessary to repolarize these cells (i.e., the concentration required to completely counteract the vasodilator effect of diazoxide). Under such conditions (i.e., bypass of K_ATP channels in the presence of clamped intracellular Ca²⁺ concentration), ADPβS still strongly potentiates the amplifying pathway of glucose-induced insulin secretion, which definitely excludes a possible direct effect on K_ATP channel activity in the effect of P2Y receptor activation on pancreatic β-cell function. As concerns the triggering pathway, the potentiation of glucose- and tolbutamide-induced first phases, but not of arginine- and KCl-induced first phases, raises the possibility of a modulation by the P2Y agonist, of K_ATP channel closure induced by either glucose or tolbutamide. Such an additional mechanism might be put together with the effect of GLP-1 on K_ATP channels mediated by protein kinase A (PKA) (23).

The importance of glucose metabolism prompted us to investigate the effect of ADPβS on insulin secretion induced by the cell permeant methyl esters of two tricarboxylic acid cycle intermediates: pyruvate and succinate. These compounds have both been reported to stimulate β-cell function, not exclusively through mitochondrial ATP production. Methyl pyruvate has first been proposed to initiate insulin release through an extramitochondrial effect or production of mitochondrial coupling factors different from ATP (24), and more recently through a direct inhibitory effect on K_ATP channels (25). In our conditions, such an action should however be of minor importance since a consistent response to the methyl ester was obtained only at the high 15 mmol/l concentration and could only be amplified by 45% in the presence of ADPβS. From the present data, we cannot conclude as to the precise mechanism involved; nevertheless, it probably differs from that of tolbutamide, which appears to be much more sensitive to the activation of P2Y receptors.

When compared with methyl pyruvate, the succinate methyl ester appears to be approximately four times more potent on insulin secretion. This probably results from the ability of succinate to react with both succinate dehydrogenase to provide metabolic energy and succinate thiokinase when present at high concentrations (26). As a consequence, succinyl CoA generated by the reversal of thiokinase reaction interacts with acetoacetate to produce succinate and acetoacetyl-CoA that would in turn react with acetyl-CoA to yield 3-hydroxy-3-methylglutaryl-CoA, a precursor of mevalonate that has been shown to modify GTP binding proteins by isoprenylation, a prerequisite for their association to membranes (27,28). Therefore, mevalonate or one of its metabolites has been proposed as a signal for insulin secretion. If the thiokinase reaction with succinate consumes either GTP or ATP, the cost in metabolic energy is probably low in the presence of succinate esters, due to the flux of high levels of succinate through succinate dehydrogenase, which generates two molecules of ATP per molecule of succinate. Considering this succinate mechanism on insulin release proposed by Fahien and McDonald (26), the inability of ADPβS to potentiate methyl succinate-induced insulin secretion might be due to the decrease in intracellular GTP levels resulting from the reversal of succinate thiokinase reaction. Indeed, GTPase activity is well known to play a central role in the signaling pathways that couple receptors to adenylyl cyclase.

We conclude that ADPβS, a stable P2Y receptor agonist, potentiates the insulin secretory effect of a glucose challenge that mimics a postprandial increase in the sugar concentration. This effect requires glucose metabolism, is probably related to an increase in intracellular Ca²⁺ concentration, and occurs independently of a direct effect on K_ATP channels, although different mechanisms are probably involved for the two phases of glucose-induced insulin secretion.

The authors thank Céline Jahannault, Michel Tournier, and Jacqueline Boyer for expert technical assistance.