Critical role of cAMP-GEFII--Rim2 complex in incretin-potentiated insulin secretion.

Incretins such as glucagon-like peptide-1 and gastric inhibitory polypeptide/glucose-dependent insulinotropic peptide are known to potentiate insulin secretion mainly through a cAMP/protein kinase A (PKA) signaling pathway in pancreatic beta-cells, but the mechanism is not clear. We recently found that the cAMP-binding protein cAMP-GEFII (or Epac 2), interacting with Rim2, a target of the small G protein Rab3, mediates cAMP-dependent, PKA-independent exocytosis in a reconstituted system. In the present study, we investigated the role of the cAMP-GEFII--Rim2 pathway in incretin-potentiated insulin secretion in native pancreatic beta-cells. Treatment of pancreatic islets with antisense oligodeoxynucleotides (ODNs) against cAMP-GEFII alone or with the PKA inhibitor H-89 alone inhibited incretin-potentiated insulin secretion approximately 50%, while a combination of antisense ODNs and H-89 inhibited the secretion approximately 80-90%. The effect of cAMP-GEFII on insulin secretion is mediated by Rim2 and depends on intracellular calcium as well as on cAMP. Treatment of the islets with antisense ODNs attenuated both the first and second phases of insulin secretion potentiated by the cAMP analog 8-bromo-cAMP. These results indicate that the PKA-independent mechanism involving the cAMP-GEFII--Rim2 pathway is critical in the potentiation of insulin secretion by incretins.

Blood glucose levels are precisely controlled by insulin release from pancreatic ␤-cells. Insulin secretion from the ␤-cells is regulated positively and negatively by many intracellular signals generated by various factors, including nutrients, hormones, and neurotransmitters (1)(2)(3)(4). cAMP is thought to be a most critical intracellular signal in the mechanism of potentiation of insulin secretion (5)(6)(7)(8). In fact, insulin secretion varies significantly with treatment by adenylyl cyclase activators and inhibitors and phosphodiesterase inhibitors that alter cAMP levels in pancreatic islets (5,6,9). In addition, phosphorylation of many regulatory proteins by cAMP-dependent protein ki-nase (protein kinase A; PKA) 1 in the ␤-cells has been suggested in the regulation of insulin secretion (10). However, despite its importance as an intracellular signal in the ␤-cells, cAMP is not considered a primary signal in the insulin secretion process (5,9). Rather, cAMP is thought to be a potentiating signal that modulates nutrient-induced insulin secretion, especially the potentiation of glucose-induced insulin secretion (5,9).
Until recently, PKA was the only molecule identified as a direct target of cAMP in pancreatic ␤-cells (21,26). Activation of PKA upon stimulation is assumed to phosphorylate regulatory proteins associated with the process of insulin secretion (21,27). However, the substrates for PKA activated by cAMP or incretins and the roles of PKA-phosphorylated proteins in cAMP-potentiated insulin secretion are not clear. Although GLUT2 (a low K m glucose transporter), Kir6.2, and SUR1 (subunits of the pancreatic ␤-cell ATP-sensitive potassium channel) and ␣-SNAP (a vesicle-associated protein), all of which are known to be expressed in pancreatic ␤-cells, have been shown to be phosphorylated by PKA (28 -31), the roles of these phosphorylations in insulin secretion are unknown. Furthermore, using electrophysiological measurements, cAMP has been found to promote exocytosis in the pancreatic ␤-cells by a PKAindependent mechanism as well as by a PKA-dependent mechanism (32). We reported recently that the cAMP-binding protein cAMP-GEFII (33), also referred to as Epac2 (34,35), is a direct target of cAMP in regulated exocytosis and that cAMP-GEFII, interacting with Rim2 (26), a target of the small Gprotein Rab3, mediates cAMP-dependent, PKA-independent exocytosis in a reconstituted system (26). However, the role of cAMP-GEFII in insulin secretion in native pancreatic ␤-cells is not known.
In the present study, we have investigated the role of cAMP-GEFII⅐Rim2 in incretin-potentiated insulin secretion. Our data indicate that a PKA-independent mechanism involving cAMP-GEFII⅐Rim2 is critical in the potentiation of insulin secretion by both GLP-1 and GIP.
Isolation of Mouse Pancreatic Islets and Batch Incubation Experiments-All animal procedures were approved by the Chiba University Animal Care Committee. Mouse pancreatic islets were isolated by collagenase digestion method as described previously (36) and were cultured in RPMI medium 1640 (Invitrogen Corp., Carlsbad, CA) containing 10% (v/v) fetal bovine serum, 60.5 mg/liter penicillin, 100 mg/liter streptomycin, and 11.1 mM glucose under a humidified condition of 95% air and 5% CO 2 . The islets then were cultured with the medium containing 4 M of antisense phosphorothioate-substituted ODNs against mouse cAMP-GEFII (5Ј-CAACGGCCTTTTATCC-3Ј) or control ODNs (5Ј-ACCTACGTGACTACGT-3Ј) (BIOGNOSTIK, Göttingen, Germany) for 96 h. Batch incubation experiments were performed as described previously (36). After preincubation (30 min) of isolated islets with Hepes-Krebs buffer containing 2.8 mM glucose, five size-matched islets were collected in each tube and incubated in 500 l of the same buffer containing glucose and test substances at the indicated concentrations for 30 min. GLP-1, GIP, H-89 as a PKA inhibitor, MDL 12330A as an adenylyl cyclase inhibitor, and 3-isobutyl-1-methylxanthine were added in the preincubation period, and 8-Br-cAMP was added in the incubation period. Insulin released into the medium was measured by radioimmunoassay (Eiken Chemical, Tokyo, Japan) (37).
Measurements of cAMP Content in the Islets-The cAMP content of the islets was measured according to the manufacturer's instructions for the cAMP enzyme immunoassay system (Amersham Pharmacia Biotech) in the presence of 20 mM glucose. 3-Isobutyl-1-methylxanthine (250 M) was always added to the incubation buffer. Twenty islets were used for cAMP measurements in each batch. The cAMP levels were normalized to the protein concentration.
Glutathione S-Transferase Pull-down Assay-The isolated mouse pancreatic islets, treated with control ODNs or antisense ODNs as described above, were homogenized and incubated at 4°C for 12 h with 2 g of glutathione S-transferase-Rim2 (residues 1466 -2453) or glutathione S-transferase alone immobilized on glutathione beads. The complexes were washed and then separated by SDS-PAGE and immunoblotted with the IgG-purified anti-cAMP-GEFII antibody (26).
MIN6 cells were transfected with human preproinsulin expression vector (pCMV-hproins) plus pCMV-luciferase, pCMV-HA-Rim2⌬A, pFLAG-CMV-2-mutant cAMP-GEFII (G114E,G422D), pCMV-HA-Rim2, or pCMV-GFP-and Myc-cAMP-GEFII. The deletion mutant Rim2 (Rim2⌬A) lacks the zinc finger and C 2 domains but retains the cAMP-GEFII binding region and has a dominant negative effect on interaction between WT cAMP-GEFII and WT Rim2 (26). The mutant cAMP-GEFII (G114E,G422D), in which both cAMP binding sites are disrupted, was also used (26). As control, luciferase was used. Three days after transfection, the C-peptide secretory response to 8-Br-cAMP (1 mM) in the presence of glucose (16.7 mM) for 60 min was evaluated by human C-peptide released into medium. Human C-peptide was measured by a human C-peptide radioimmunoassay kit (Linco Research Inc., St. Charles, MO).
Perifusion Experiment-Perifusion of pancreatic islets was performed as described previously (36). Briefly, groups of 100 isolated mouse islets treated for 96 h with 4 M control or antisense ODNs as described above were loaded onto filters in columns with Bio-Gel P-2 (Bio-Rad) and continuously perifused with Hepes-Krebs buffer at a constant flow rate of 1.0 ml/min. After a 30-min stabilization period with 2.8 mM glucose, the groups of islets were successively stimulated with 16.7 mM glucose with or without 100 M 8-Br-cAMP. Perifusate solutions were gassed with 95% O 2 and 5% CO 2 and maintained at 37°C.

GLP-1-and GIP-potentiated Insulin Secretion and cAMP
Effects of MDL12330A on Incretin-potentiated Insulin Secretion-To determine whether cAMP is an essential signal in incretin-potentiated insulin secretion, we examined the effects of MDL12330A (10 M) on the GLP-1-and GIP-potentiated insulin secretions in the presence of 11.1 mM glucose using the batch incubation method (Fig. 2, C and D (Fig. 2D). Together, these data indicate that both the GLP-1and GIP-potentiated insulin secretions depend critically on cAMP production in pancreatic ␤-cells.
Effects of the Antisense ODNs against cAMP-GEFII on Incretin-potentiated Insulin Secretion-We have shown recently that the cAMP-binding protein cAMP-GEFII is a direct target of cAMP in regulated exocytosis (26). In the present study, we investigated with the purpose of evaluating cAMP-GEFII involvement in the potentiation of insulin secretion by incretins with native pancreatic ␤-cells. We first ascertained if treatment of the islets with antisense ODNs could suppress the level of endogenous cAMP-GEFII protein. Treatment with antisense ODNs markedly decreased the cAMP-GEFII protein level in the islets (Fig. 3A). To further confirm the specificity of the antisense ODNs, we checked the level of other proteins (i.e. the PKA regulatory subunit II␣, small G-protein Rab3A, and VAMP-2 (vesicle-associated membrane protein-2)). There were no differences in expression of protein levels between control ODN-treated and antisense ODN-treated pancreatic islets (Fig. 3A), indicating that the antisense ODNs are specific for cAMP-GEFII. We then examined the effect of antisense ODNs on the insulin secretory responses to GLP-1 and GIP. Glucoseinduced insulin secretion was not affected by antisense ODNs treatment (Fig. 3B). The GLP-1 (100 nM)-potentiated insulin secretion in the presence of 11.1 mM glucose from pancreatic islets treated with antisense ODNs decreased significantly, compared with that treated with control ODNs (control ODNtreated, 8.51 Ϯ 0.63 ng/islet/30 min; antisense ODN-treated, 6.26 Ϯ 0.57 ng/islet/30 min, n ϭ 5, p Ͻ 0.0001) (Fig. 3C). The GIP (100 nM)-potentiated insulin secretion also decreased sig-nificantly (control ODN-treated, 7.01 Ϯ 0.79 ng/islet/30 min; antisense ODN-treated, 5.10 Ϯ 0.44 ng/islet/30 min, n ϭ 6, p Ͻ 0.0005) (Fig. 3D). These results indicate that cAMP-GEFII is involved in the potentiation of insulin secretion by both GLP-1 and GIP in pancreatic islets. with H-89 and the antisense ODNs caused a further reduction in GLP-1-potentiated insulin secretion (4.27 Ϯ 0.12 ng/islet/30 min, n ϭ 5, p Ͻ 0.001). The insulin secretion potentiated by GIP (100 nM) also was measured in the presence of 11.1 mM glucose (Fig. 4B). Similarly, H-89 partially blocked GIP-potentiated insulin secretion (GIP alone, 7.27 Ϯ 0.18 ng/islet/30 min; GIP plus H-89, 4.24 Ϯ 0.13 ng/islet/30 min, n ϭ 5, p Ͻ 0.0001), and combination treatment with H-89 and the antisense ODNs caused a further reduction in GIP-potentiated insulin secretion (2.99 Ϯ 0.14 ng/islet/30 min, n ϭ 5, p Ͻ 0.0005). To confirm the involvement of cAMP-GEFII in cAMP-dependent, PKA-independent insulin secretion, we used the cAMP analog 8-Br-cAMP in the presence of 16.7 mM glucose. Similar results were obtained with 8-Br-cAMP (Fig. 4C). These results suggest strongly that both GLP-1-and GIP-potentiated insulin secretions are mediated by PKA-independent as well as PKA-dependent mechanisms and that cAMP-GEFII participates in a PKA-independent mechanism.
cAMP-potentiated Insulin Secretion Is Mediated by the cAMP-GEFII⅐Rim2 Complex-cAMP-GEFII has been shown to interact with Rim2, a target of the small G-protein Rab3 (26). To determine whether the effect of cAMP-GEFII on cAMPpotentiated insulin secretion requires its direct interaction with Rim2, we used two mutants (26) (Fig. 5A). We first examined the effect of Rim2⌬A on 8-Br-cAMP-potentiated exocytosis from MIN6 cells in which endogenous cAMP-GEFII and Rim2 are expressed. For this purpose, we utilized MIN6 cells transfected with human preproinsulin cDNA (25,42). Proinsulin is converted into insulin and C-peptide during the secretory process in pancreatic ␤-cells (43). Since antibodies against human insulin cross-react with endogenous mouse insulin, we monitored secretion by measuring the human C-peptide release from MIN6 cells transfected with human preproinsulin and Rim2⌬A (26). Overexpression of Rim2⌬A in MIN6 cells signif-icantly inhibited the 8-Br-cAMP-induced C-peptide secretion in the presence of 16.7 mM glucose (Fig. 5B). Coexpression of WT cAMP-GEFII with Rim2⌬A in MIN6 cells significantly restored inhibition of the C-peptide secretion by Rim2⌬A, suggesting that the effect of cAMP-GEFII on cAMP-potentiated insulin secretion requires interaction with Rim2. Similarly, we also assessed the effect of the mutant cAMP-GEFII (G114E,G422D). An in vivo binding experiment shows that overexpression of cAMP-GEFII (G114E,G422D) inhibits interaction between the WT cAMP-GEFII and WT Rim2 (Fig. 5C), indicating that the mutant acts as a dominant-negative inhibitor of the interaction. We reasoned that the double mutant (G114E,G422D), when overexpressed in MIN6 cells, might trap endogenous Rim2 to inhibit cAMP-potentiated C-peptide secretion. While overexpression of WT cAMP-GEFII did not alter 8-Br-cAMP-potentiated C-peptide secretion (data not shown), overexpression of the double mutant (G114E,G422D) significantly inhibited it (Fig. 5D). This inhibition of the C-peptide secretion was mostly restored by coexpression of WT Rim2, due probably to its interaction with endogenous WT cAMP-GEFII. These results indicate that cAMP-potentiated insulin secretion is mediated by the cAMP-GEFII⅐Rim2 complex.
The Effects of cAMP-GEFII Are Dependent on Intracellular Calcium as Well as cAMP-To determine whether the effect of cAMP-GEFII on cAMP-potentiated insulin secretion requires a rise in intracellular calcium concentrations ([Ca 2ϩ ] i ), we examined the effects of 8-Br-cAMP (1 mM), high K ϩ (60 mM), and their combination, in the presence of 2.8 mM glucose, on insulin secretion in pancreatic islets treated with control ODNs or antisense ODNs. There were no differences in the insulin secretions stimulated by 8-Br-cAMP alone or high K ϩ alone between control ODN-and antisense ODN-treated islets. In contrast, the insulin secretion stimulated by a combination of 8-Br-cAMP plus high K ϩ in antisense ODN-treated islets was significantly lower than that in control ODN-treated islets (control ODNs, 8.81 Ϯ 0.60 ng/islet/30 min; antisense ODNs, 6.24 Ϯ 0.40 ng/islet/30 min, n ϭ 10, p Ͻ 0.005) (Fig. 6A). We also examined the effect of carbachol (50 M) on insulin secretion in the islets treated with control ODNs or antisense ODNs. While there were no differences in insulin secretion stimulated by 8-Br-cAMP alone or carbachol alone, the insulin secretion stimulated by a combination of 8-Br-cAMP plus carbachol in the antisense ODN-treated islets was significantly lower than that in the control ODN-treated islets (control ODNs, 3.48 Ϯ 0.17 ng/islet/30 min; antisense ODNs, 2.45 Ϯ 0.08 ng/islet/30 min, n ϭ 16, p Ͻ 0.0001) (Fig. 6B). These results indicate that the effects of cAMP-GEFII on insulin secretion depend on intracellular Ca 2ϩ as well as cAMP in pancreatic ␤-cells.
cAMP-GEFII Is Involved in Both the First and Second Phase of cAMP-potentiated Insulin Secretion-We examined the involvement of cAMP-GEFII in the insulin secretory phase using perifused mouse pancreatic islets. No significant difference was found between control ODN-treated islets and antisense ODN-treated islets in the absence of 8-Br-cAMP (Fig. 7A). When the islets were treated with antisense ODNs, both the first phase and second phase potentiations by 8-Br-cAMP were suppressed (Fig. 7B), clearly showing that cAMP-GEFII is involved in both phases of insulin secretion.

DISCUSSION
Incretins such as GLP-1 and GIP play an important role in the potentiation of insulin secretion (44 -47). It has generally been thought that both GLP-1 and GIP potentiate glucoseinduced insulin secretion primarily by cAMP/PKA signaling, which leads to phosphorylation of regulatory proteins associated with the secretory process in pancreatic ␤-cells (15,48,49). A study by capacitance measurements has suggested that cAMP also promotes exocytosis in pancreatic ␤-cells in a PKAindependent mechanism (32). In the present study, we show that MDL 12330A, an inhibitor of adenylyl cyclase, completely blocks both the GLP-1-and the GIP-stimulated cAMP production in pancreatic islets and that MDL 12330A remarkably inhibits both the GLP-1-and the GIP-induced insulin secretions. These results confirm that the effects of both GLP-1 and GIP on insulin secretion depend critically on the intracellular cAMP elevation due to activation of adenylyl cyclase. It is interesting that MDL 12330A did not completely inhibit either GLP-1-or GIP-potentiated insulin secretion under the conditions in which cAMP production was blocked. This suggests that the potentiating effects of the incretins on insulin secretion are mediated at least in part by a cAMP-independent mechanism, although the effects are small.
We recently found that the cAMP-binding protein cAMP-GEFII, by interacting with Rim2, a target of Rab3, participates in cAMP-dependent, PKA-independent exocytosis in a reconstituted system (26). In the present study, we investigated in order to find whether the cAMP-GEFII in native pancreatic ␤-cells is involved in GLP-1-and GIP-potentiated insulin secretions and if such action is PKA-independent. Treatment of islets with antisense ODNs reduced both GLP-1-and GIPpotentiated insulin secretion, clearly indicating that the effects of the incretins are mediated in part by cAMP-GEFII. Ten M of H-89, a widely used specific inhibitor of PKA phosphorylation in intact cells (24,28,41,50), was then used to block the phosphorylation of GLUT2, a substrate of PKA in pancreatic ␤-cells (28), to evaluate incretin-potentiated insulin secretion. Interestingly, although treatment of pancreatic islets with H-89 reduced (about 50%) both GLP-1-and GIP-potentiated insulin secretions, treatment of the islets with H-89 plus antisense ODNs further reduced the insulin secretions (80 -90%), suggesting strongly that the potentiation of insulin secretion by both GLP-1 and GIP is mediated by PKA-independent as well as PKA-dependent mechanisms and that cAMP-GEFII is involved in the PKA-independent mechanism.
We then determined whether or not the potentiating effects of cAMP on insulin secretion are mediated by Rim2. Overexpression of a dominant negative mutant, Rim2 (Rim2⌬A) or cAMP-GEFII (G114E,G422D double mutant), inhibited the potentiating effect of 8-Br-cAMP on C-peptide secretion from human preproinsulin-transfected MIN6 cells. In addition, the inhibitory effect of Rim2⌬A or the cAMP-GEFII double mutant on C-peptide secretion was mostly restored by coexpression of WT cAMP-GEFII or WT Rim2, respectively, suggesting that the potentiating effects of the incretins are mediated by the cAMP-GEFII⅐Rim2 complex.
Because intracellular Ca 2ϩ is essential in triggering insulin secretion, we investigated to find if the mechanism of potentiation by the cAMP-GEFII⅐Rim2 complex is also Ca 2ϩ -dependent. The effects of high K ϩ and carbachol, which triggers Ca 2ϩ influx and mobilizes intracellular Ca 2ϩ (51), respectively, on insulin secretion were examined in islets treated with antisense ODNs. While there were no differences in the insulin secretions stimulated by 8-Br-cAMP alone, high K ϩ alone, or carbachol alone between control ODN-treated and antisense ODN-treated islets, insulin secretion stimulated by a combination of 8-Br-cAMP plus high K ϩ or 8-Br-cAMP plus carbachol was significantly reduced in antisense ODN-treated islets. These findings indicate that the potentiation of insulin secretion through the cAMP-GEFII⅐Rim2 pathway depends on intracellular Ca 2ϩ as well as cAMP. Since Rim2 has two C 2 domains, Ca 2ϩ might modulate interaction between cAMP-GEFII and Rim2.
cAMP potentiates both phases of insulin secretion at high glucose concentrations in isolated perifused pancreas (2). Similarly, GLP-1 and GIP are both known to potentiate both the first and second phases of glucose-induced insulin secretion (15,16). To determine the involvement of cAMP-GEFII in each phase of insulin secretion, we evaluated the potentiation of insulin secretion by 8-Br-cAMP in perifused pancreatic islets with antisense or control ODNs treatment. Both the first and second phase enhancement by 8-Br-cAMP was significantly suppressed in islets treated with antisense ODNs compared with control, showing that both the first and second phases are potentiated by cAMP in a PKA-independent mechanism.
Considering these findings together, we propose that incretins potentiate glucose-induced insulin secretion primarily by two mechanisms: the pathway involving phosphorylation of regulatory proteins by PKA activation (PKA-dependent) and the pathway involving the cAMP-GEFII⅐Rim2 complex (PKAindependent). The affinity for cAMP is quite different in PKA and cAMP-GEFII, with K d of ϳ100 nM (52) and ϳ10 M (26), respectively. cAMP at basal state in pancreatic islets has been reported in a range of micromolar concentrations (52), suggesting that many substrates for PKA already are maximally phosphorylated in pancreatic islets. This is the case with GLUT2 (28) and the sulfonylurea receptor SUR1, a subunit of the ␤-cell K ATP channel (29). Accordingly, PKA and cAMP-GEFII have distinct roles in cAMP-potentiated insulin secretion. The mechanism involving the cAMP-GEFII⅐Rim2 complex might operate upon a rise in local cAMP concentrations by stimulation. The mechanism involving PKA phosphorylation might also be controlled by PKA-anchoring protein (44,53,54). Since cAMP-GEFII has guanine exchange factor activity toward the small G-protein Rap1 (33), it is also tempting to speculate that Rap1, which is activated by cAMP-GEFII through incretins, might also be involved in insulin secretion.
Further elucidation of the regulation of the cAMP-GEFII⅐Rim2 complex by incretins should both clarify the mechanism of the potentiation of insulin secretion and suggest novel anti-diabetic drug therapy.