Rosiglitazone (BRL 49653) Enhances Insulin Secretory Response via Phosphatidylinositol 3-Kinase Pathway

Male Sprague-Dawley rats obtained from the Animal Center of the National Science Council of the Republic of China and weighing between 250 and 350 g were used in these experiments. The rats were kept at room temperature (∼25°C) in plastic cages under a 12-h cycle of light. The rats were given free access to tap water and were fed ad libitum with commercial diet (Fwusow; Sha-Lu, Taichung, Taiwan). After an overnight fast (>12 h), the rats were anesthetized with a 65 mg/kg intraperitoneal injection of pentobarbital sodium (MTC Pharmaceutics, Cambridge, Canada). Access to the pancreata was gained through a ventral midline incision, and the animals’ celiac arteries and portal veins were cannulated with polyvinyl tubings of 0.625 and 1.2 mm (internal diameter), respectively. The rats were maintained at 37°C throughout the experiments. Krebs-Ringer bicarbonate buffer, supplemented with 10 mmol/l N-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid (HEPES), 6 mmol/l glucose, 0.1% dextran, and 0.2% bovine serum albumin, was used as the perfusate (basal medium). This solution was continuously aerated with 95% O₂/5% CO₂, and the pH value was maintained at 7.4. All chemicals were purchased from Sigma Chemical (St. Louis, MO), except for RSG and LY294002, which were kindly provided by SmithKline Beecham Pharmaceuticals (Essex, U.K.) and Eli Lilly (Indianapolis, IN), respectively.

Using a method of Grodsky and Fanska (23), the in situ rat pancreatic perfusion with an open system was performed at 37°C. The pancreatic perfusion was maintained at a flow rate of 1 ml/min, which did not cause noticeable edema or impair insulin release. After the cannulation of the celiac artery and the portal vein and the establishment of the rat pancreatic perfusion preparation, the first 10 min of perfusion were considered as an equilibration period. Subsequently, the effluent fluid was collected every minute from the cannula of the portal vein for 50 or 55 min. In experiment 1, after a baseline period of 15 min, the perfusate containing RSG (0.045, 0.45, 1.5, or 4.5 μmol/l) was administered for 30 min, followed by the basal medium for 10 min. In experiment 2, after a baseline period of 10 min, the medium containing a PI3K inhibitor (LY294002 at 3.9 μmol/l) was administered for 5 min and was followed by LY294002 (3.9 μmol/l)/RSG (4.5 μmol/l) for another 30 min and a basal medium washout for the last 10 min. In experiment 3, after a baseline period of 10 min, the perfusate containing 10 mmol/l glucose with or without RSG (4.5 μmol/l) was administered for 30 min, followed by a basal medium washout for the last 10 min. In experiment 4, after a baseline period of 10 min, the medium containing LY294002 (3.9 μmol/l) was administered for 5 min as pretreatment and was followed by 10 mmol/l glucose with or without RSG (4.5 μmol/l) for another 30 min and a basal medium washout for the last 10 min. The collected effluent fluid was kept at 4°C and subsequently assayed within 12 h for insulin by using radioimmunoassay (RIA), as previously described by Hale and Randle (24). Rat insulin was used as standard for the RIA.

Data of effluent insulin concentrations were expressed as a percentage of the baseline level (mean of 12 baseline values) in means ± SE. Data were analyzed by using analysis of variance (ANOVA) to determine the significance of treatment and time. The treatment multiplied by time interaction was used as an error term to determine the effect of treatment. The significance of treatment was determined from the conservative F value. Tukey’s highly significant difference test was used to determine the differences between treatments for which the ANOVA indicated a significant (P < 0.05) F ratio. For analyzing the first (from 2 to 7 min) and second (from 8 to 30 min) phases of insulin secretion during glucose perfusion, the areas under the curve (AUCs) of the percentage increase over baseline (Figs. 2 and 5) were calculated and compared using Student’s t tests. P < 0.05 was considered statistically significant.

To demonstrate the direct effect of RSG on insulin secretion, various concentrations of RSG (0.045, 0.45, 1.5, and 4.5 μmol/l) within basal media with 6 mmol/l glucose were infused via in situ pancreatic perfusion. As shown in Fig. 1, insulin secretion was induced by RSG in a time- and dose-dependent manner (P < 0.01 vs. control). Insulin secretion was observed within 2 min but reached a statistical significant level at 15 min (RSG 4.5-μmol/l group) of continuous perfusion of RSG. This effect of RSG on insulin secretion still did not reach a plateau before the washout period. The induction of insulin secretion was monophasic, which was different from the typical glucose-induced biphasic insulin secretion (Fig. 2). Insulin secretory response declined immediately after discontinuation of RSG perfusion (Fig. 1).

To evaluate the effect of RSG on glucose-induced insulin secretion, glucose alone (10 mmol/l) and glucose containing RSG (4.5 μmol/l) were perfused through isolated rat pancreata. As shown in Fig. 2, 10 mmol/l glucose induced a biphasic response in insulin secretion, with an early peak at 2–7 min (first phase) and a continuous increase thereafter (second phase). After the administration of RSG in glucose perfusate, both the glucose-induced first and second phases of insulin secretion were further enhanced compared with those found in the perfusion experiments with glucose alone (P < 0.01). Insulin secretory response declined immediately after discontinuation of glucose and RSG perfusion (Fig. 2).

To address the effect of RSG on different phases of insulin secretion after glucose stimulation, we compared the AUCs of the percentage increase over baseline of the first and second phases of insulin secretion. As shown in Table 1, both the first and second phases of insulin secretion were enhanced by simultaneous RSG perfusion (by 80.7 and 52.4%, respectively).

To determine the possible signaling pathways leading to the effect of RSG on insulin secretion, a PI3K inhibitor, LY294002 (25,26), was included in the perfusate. As shown in Fig. 3, the addition of LY294002 at 3.9 μmol/l decreased RSG-potentiated insulin release dramatically (P < 0.01), indicating a possible role of the PI3K pathway in the acute effect of RSG. In contrast, both the first and second phases of insulin secretion after glucose stimulation were not affected by the PI3K inhibitor (Fig. 4). As depicted in Fig. 5, the administration of LY294002 blocked the potentiation of glucose-induced insulin secretion by RSG and yielded a time-response curve that was almost superimposed to that produced by the perfusion experiments with glucose alone (Fig. 4). Both the first and second phases of insulin secretion were reduced by LY294002 (by 34.3 and 33.0%, respectively) (Table 2).

Previous studies have reported that troglitazone, a member of the TZD family, can protect β-cells from lipoapoptosis and restore the insulin-secretory capacity of β-cells (20,21). In addition, a novel N-aryl tyrosine activator of PPAR-γ (GW1929) can restore glucose-induced biphasic insulin secretion in the perfused pancreas of ZDF rats after oral administration for 14 days (22). However, these effects of PPAR-γ activators on the islets are secondarily attributed to the amelioration of glucose toxicity and/or lipotoxicity (22). In the present study, we first detected the direct effect of RSG on glucose-induced insulin secretion by using an isolated rat pancreas perfusion system. We found that insulin secretion was induced by RSG in a time- and dose-dependent manner in the presence of 6 or 10 mmol/l glucose, indicating that RSG can potentiate glucose-stimulated insulin secretion. In the absence of glucose, perfusion of RSG alone did not stimulate insulin secretion (data not shown). Moreover, the effect of RSG on insulin secretion was blocked by a PI3K inhibitor, LY294002, indicating an involvement of PI3K in RSG-induced insulin secretion.

The direct effect of troglitazone on insulin secretion had been observed in the isolated islets and in a β-cell line, HIT (27). Masuda et al. (27) reported that 1 μmol/l troglitazone stimulated insulin release, whereas at a higher dose (100 μmol/l), troglitazone inhibited insulin release from these cells. To explore the mechanisms by which troglitazone triggers insulin release, a putative membrane binding site was postulated due to a noncompetitive displacement of the binding of [³H]glibenclamide by troglitazone on the β-cell membrane. However, the insulinotropic effect of troglitazone was different from that of glibenclamide because 1 μmol/l troglitazone increased insulin secretion but failed to reduce ATP-sensitive K⁺-channel activity (27). Thus, the signaling pathway of the insulinotropic effect of troglitazone on pancreatic islets and HIT cells is still unclear. Our study also showed that RSG could induce insulin secretion and potentiate both the first and second phases of glucose-stimulated insulin secretion with only short-term administration. Importantly, we found that a PI3K inhibitor inhibited both the direct and potentiation effects of RSG on insulin secretion. Treatment with this inhibitor did not reduce the glucose-induced insulin secretion. These results unequivocally indicated that the direct and potentiation effects of RSG on insulin secretion are mediated by the PI3K pathway (Fig. 6). The doses that we used in our present study were quite comparable with the blood concentrations after a single dosing of RSG at 1–8 mg. Interestingly, a recent study showed that insulin-induced insulin secretion also involved insulin receptors, insulin receptor substrate-1, and PI3K activation, which led to insulin secretion via an increase in intracellular calcium concentration (28,29).

Although PPAR-γ proteins, members of the nuclear receptor family, are expressed in the rat islet and in a β-cell line, MIN6 (L-M.C., M-W.L., unpublished observations), it is unlikely that the activation of these proteins mediates the potentiation effect of RSG on the first phase of glucose-induced insulin secretion, because the onset of the RSG-induced increase is too fast to be attributable to a genomic action. Because the second-phase secretion was potentiated by RSG within 30 min of exposure, it is likely that even the stimulation of this phase was not an effect at the genomic level. However, a recent study reported that a peroxisomal proliferator response element (PPRE) exists in the rat GLUT2 promotor. In HIT-T15 cells, promotor activity of the rat GLUT2 gene was increased by troglitazone and 9-cis retinoic acid, and mutations of GLUT2-PPRE resulted in a reduction of the promoter activity. In addition, they also observed an increase in the GLUT2 trascription by troglitazone and 9-cis retinoic acid in rat islets (30). Therefore, the potentiation effect of RSG on the second phase of glucose-induced insulin secretion might be explained by the increased expression of GLUT2 on pancreatic β cells, which play a key role in β-cells for glucose signaling to insulin secretion and biosynthesis (31). In a longer-term treatment, troglitazone was found to enhance the expression of p85α, a subunit of PI3K (11). Therefore, it is possible for a potentiation of RSG on the second phase of glucose-induced insulin secretion via an increase in PI3K protein synthesis and/or GLUT2 expression. Further studies are needed to elucidate the mechanism(s) involved in this pathway.

In conclusion, RSG has a potentiation effect on insulin secretion in the presence of glucose, which is mediated by the PI3K pathway and is clearly distinct from that of glucose-stimulated insulin secretion. From this point of view, TZD derivatives could become a promising category of therapeutic agents for type 2 diabetes, in addition to their well-known insulin-sensitizing effect on the peripheral tissues.

This work was supported in part by the Research Fund of Diabetes Mellitus of the National Taiwan University Hospital, Taipei, Taiwan, and a grant from the National Science Council (NSC89-2313-B-005-331) of the Republic of China.