Essential role of Epac2/Rap1 signaling in regulation of insulin granule dynamics by cAMP

Abstract

cAMP is well known to regulate exocytosis in various secretory cells, but the precise mechanism of its action remains unknown. Here, we examine the role of cAMP signaling in the exocytotic process of insulin granules in pancreatic beta cells. Although activation of cAMP signaling alone does not cause fusion of the granules to the plasma membrane, it clearly potentiates both the first phase (a prompt, marked, and transient increase) and the second phase (a moderate and sustained increase) of glucose-induced fusion events. Interestingly, all granules responsible for this potentiation are newly recruited and immediately fused to the plasma membrane without docking (restless newcomer). Importantly, cAMP-potentiated fusion events in the first phase of glucose-induced exocytosis are markedly reduced in mice lacking the cAMP-binding protein Epac2 (Epac2^ko/ko). In addition, the small GTPase Rap1, which is activated by cAMP specifically through Epac2 in pancreatic beta cells, mediates cAMP-induced insulin secretion in a protein kinase A-independent manner. We also have developed a simulation model of insulin granule movement in which potentiation of the first phase is associated with an increase in the insulin granule density near the plasma membrane. Taken together, these data indicate that Epac2/Rap1 signaling is essential in regulation of insulin granule dynamics by cAMP, most likely by controlling granule density near the plasma membrane.

Exocytosis, the final vesicular-transport step in the secretory pathway, is a crucial event in secretory cells including neuronal, endocrine, neuroendocrine, and exocrine cells (1, 2). Peptide hormones, neuropeptides, and digestive enzymes are stored in secretory vesicles called dense-core granules, the fusion of which to the plasma membrane is tightly controlled by various signals (1). Although a rise in the intracellular Ca²⁺ concentration is the primary signal in the regulation of exocytosis, other intracellular signals also are critical, cAMP being especially important in many secretory cells (1, 3, 4). The pancreatic beta cell is an endocrine cell containing a typical dense-core granule that secretes insulin, a key hormone in the maintenance of glucose homeostasis, and dysfunction of the insulin secretory process causes diabetes mellitus. Glucose-induced insulin secretion and its potentiation manifest the fundamental regulatory mechanisms of insulin secretion. cAMP is physiologically the most important potentiator of insulin secretion (5). cAMP signaling in pancreatic beta cells is activated by incretins such as glucose-dependent insulinotropic polypeptide (GIP) and glucagon-like peptide-1 (GLP-1), gut hormones secreted upon meal ingestion from enteroendocrine K and L cells, respectively (6, 7). cAMP is now known to potentiate insulin secretion in both a protein kinase A (PKA)-dependent and a PKA-independent manner, the latter involving cAMP-GEFII (Epac2) (8–10). However, the mechanism of cAMP action in the exocytotic process, including recruitment, docking, and fusion of secretory vesicles, is still unclear. Clarification of the role of cAMP signaling in the dynamics of insulin granules may provide insight into the basis of the regulatory mechanism of dense-core granule exocytosis.

In the present study, we have investigated the mechanism underlying regulation of insulin granule exocytosis by cAMP signaling. Our data indicate that Epac2/Rap1 signaling is essential in the regulation of insulin granule dynamics by cAMP, most likely by controlling granule density near the plasma membrane.

Results

Insulin Granule Dynamics Occur Primarily in Three Modes.

To monitor the dynamics of insulin granule exocytosis, we constructed insulin fused with enhanced yellow fluorescent protein (Venus) (insulin-Venus), which has advantages of pH insensitivity and a longer life span in comparison with green fluorescent protein (GFP) (11). A previous study using the insulin-secreting cell line MIN6 reported two modes of fusion events of insulin granules to the plasma membrane based on the dynamics of the granules (12). In one mode, fusion events are caused by granules that are predocked to the plasma membrane [referred to as previously docked granules (12) (called old face in this article)] (Fig. 1A). In the other mode, fusion events are caused by granules that are newly recruited to the plasma membrane (newcomer in ref. 12). However, our detailed analyses of insulin granule dynamics induced by various stimuli using primary cultured pancreatic beta cells show that newcomer can be classified into two groups: (i) granules that are newly recruited and immediately fused to the plasma membrane without docking [a docking state can barely be detected by total internal reflection fluorescence microscopy (TIRFM)] (restless newcomer); and (ii) granules that are newly recruited, docked, and then fused to the plasma membrane (resting newcomer) [Fig. 1A and supporting information (SI) Movie 1].

Fig. 1.

Characteristics of insulin granule dynamics in pancreatic beta cells. (A) Modes of fusion events of insulin granules. Old face: granules that are predocked to the plasma membrane and fused to the membrane by stimulation. Restless newcomer: granules that are newly recruited and immediately fused to the plasma membrane by stimulation. Resting newcomer: granules that are newly recruited, docked, and fused to the plasma membrane by stimulation. Sequential images (1 μm × 1 μm) were acquired every 300 msec. (B and C) Histograms and distributions of fusion events at 30-sec intervals in pancreatic beta cells stimulated by 60 mM K⁺ (B) or 16.7 mM glucose (C) in a cell surface area of 200 μm². Primary cultured pancreatic beta cells were preincubated with Hepes-KRB containing 2.8 mM glucose at 37°C for 30 min. Thirty seconds after acquisition of image, the primary cultured cells were stimulated by 60 mM K⁺ (B) or 16.7 mM glucose (C). “2.8” indicates 2.8 mM glucose. Blue, red, and green bars indicate old face, restless newcomer, and resting newcomer, respectively. Data were obtained from five to six independent experiments (n = 1–5 for each) and expressed as means ± SE.

K⁺ Stimulation and Glucose Stimulation Cause Distinct Modes of Insulin Granule Dynamics.

Ca²⁺-triggered insulin granule fusion events, as assessed by K⁺ stimulation (60 mM) alone, occurred almost immediately after stimulation but were only transient (Fig. 1B and SI Movie 2). More than 60% of the granules responsible for fusion events induced by K⁺ stimulation were old face (Fig. 1B). However, only ≈17% of predocked granules were fused to the plasma membrane. In contrast, fusion events induced by glucose stimulation (16.7 mM) occurred in two phases, the first phase of a prompt, marked, and transient increase (beginning 50 sec or more after stimulation) followed by the second phase of a moderate and sustained increase (SI Movie 1). The first phase was counted as fusion events during the first 5 min after stimulation, and the second phase was counted as fusion events occurring thereafter. Interestingly, both phases of glucose-induced fusion events involved mostly restless newcomer (Fig. 1C). These results indicate that the dynamics of insulin granules differs in K⁺ stimulation and glucose stimulation. Because K⁺ stimulation elicits Ca²⁺ influx, and glucose stimulation generates various metabolic signals such as ATP, followed by Ca²⁺ influx in pancreatic beta cells (13, 14), the difference in intracellular signal may underlie these distinct modes of exocytosis. It has been proposed that the first phase and the second phase of insulin secretion induced by glucose result from fusion of insulin secretory granules to the plasma membrane from a readily releasable pool (RRP) comprising predocked granules and a reserve pool comprising granules farther away, respectively (15). However, we find here that insulin granule dynamics in the first phase of glucose-induced exocytosis differs from that involved in Ca²⁺-triggered exocytosis. This finding indicates that the RRP involved in the first phase does not necessarily comprise predocked granules, and that some granules a distance away from the plasma membrane are immediately releasable in pancreatic beta cells, as has been found in the neuromuscular junction in frogs (16). Because our TIRFM system can detect insulin granules only within ≈40 nm below the cell surface, some of the RRP involved in the first phase may well be located >40 nm from the surface.

cAMP Potentiates Insulin Granule Exocytosis by Increasing the Number of Restless Newcomer.

To clarify the role of cAMP signaling in the process of insulin granule exocytosis, we first examined the effect of cAMP-increasing agents on fusion events. The cAMP-analog 8-Bromo-cAMP alone did not cause either significant docking or fusion events of insulin granules (Fig. 2A Right and SI Movie 3). However, 8-Bromo-cAMP clearly enhanced the frequency of glucose-induced fusion events in both the first and the second phases (Fig. 2A). The sum of the fusion events (per 200 μm²) in the first phase by glucose stimulation alone and that by glucose plus 8-Bromo-cAMP stimulation were 27.1 ± 3.5 and 50.6 ± 5.2, respectively, whereas the sum of fusion events in the second phase by glucose stimulation alone and that by glucose plus 8-Bromo-cAMP stimulation were 30.9 ± 7.6 and 54.6 ± 7.3, respectively (Fig. 2B). Interestingly, 8-Bromo-cAMP promoted fusion events by increasing only restless newcomer. These results indicate that activation of cAMP signaling enhances fusion events by increasing the number of restless newcomer. To clarify the involvement of cAMP signaling in spatial regulation of exocytosis, we compared the fusion sites induced by glucose stimulation and those induced by 8-Bromo-cAMP stimulation (Fig. 2C Left). New fusion sites appeared upon 8-Bromo-cAMP stimulation (Fig. 2C Right), indicating that cAMP signaling also participates in spatial regulation of insulin granule exocytosis.

Fig. 2.

Effects of cAMP on insulin granule dynamics. (A) (Left) Histogram of fusion events induced by stimulation of 16.7 mM glucose alone. Primary cultured pancreatic beta cells were preincubated with Hepes-KRB containing 2.8 mM glucose at 37°C for 30 min and then incubated with 16.7 mM glucose for 19.5 min. (Right) Histogram of fusion events induced by stimulation of 16.7 mM glucose plus 1 mM 8-Bromo-cAMP. Primary cultured pancreatic beta cells were preincubated with Hepes-KRB containing 2.8 mM glucose at 37°C for 30 min and then incubated in the presence of 2.8 mM glucose plus 1 mM 8-Bromo-cAMP for 4.5 min followed by the addition of glucose (16.7 mM in a final concentration) for 15 min. Blue, red, and green bars indicate old face, restless newcomer, and resting newcomer, respectively. Data were obtained from six independent experiments (n = 2–4 for each) and expressed as means ± SE. (B) Comparison of fusion events in the first and second phases between glucose stimulation alone and glucose plus 8-Bromo-cAMP stimulation. Sum of fusion events in the first phase (the first 5 min after glucose stimulation) are shown on the left. Sum of fusion events in the second phase (from 5 to 15 min after glucose stimulation) are shown on the right. Open and filled bars indicate fusion events by glucose stimulation alone and those by glucose plus 8-Bromo-cAMP stimulation, respectively. *, P < 0.01; **, P < 0.05 (Student's unpaired t test). Data were expressed as means ± SE. (C) Comparison of fusion sites induced by glucose stimulation and those induced by cAMP stimulation. (Left) Experimental protocol and histogram of fusion events. For comparison of fusion sites, pancreatic beta cells were first incubated in the presence of 16.7 mM glucose for 9.5 min followed by the addition of 1 mM 8-Bromo-cAMP for 10 min. Blue, red, and green bars indicate old face, restless newcomer, and resting newcomer, respectively. Data were obtained from five independent experiments (n = 2 or 3 for each) and expressed as means ± SE. (Right) Image of fusion sites after stimulation. Yellow boxes (1 μm × 1 μm) indicate fusion sites induced by 16.7 mM glucose stimulation alone. Light blue boxes (1 μm × 1 μm) indicate new fusion sites induced by 1 mM 8-Bromo-cAMP stimulation. A representative image from five independent experiments is shown. (Scale bar, 5 μm.)

Knockout of Epac2 Diminishes the First Phase of cAMP-Potentiated Insulin Granule Exocytosis.

We recently found that Epac2, a cAMP-binding protein (9, 17, 18), is a direct target of cAMP in regulated exocytosis (9), and that it is responsible for cAMP-dependent, PKA-independent insulin secretion (9, 10). Glucose-induced insulin secretion potentiated by 8-Bromo-cAMP, a cAMP analog that activates both PKA and Epac2, was partially blocked by the PKA inhibitor H-89 in isolated pancreatic islets (Fig. 3A), as reported in ref. 10. In contrast, glucose-induced insulin secretion potentiated by 8-pCPT-2′-O-Me-cAMP, a cAMP analog specific for activation of Epacs, was not blocked by H-89 (Fig. 3A). Because Epac2 but not Epac1 is expressed in pancreatic islets (SI Fig. 5), these results further confirm that Epac2 mediates cAMP-dependent, PKA-independent insulin secretion (9, 10). To directly ascertain the role of Epac2 in insulin granule dynamics, we generated Epac2 knockout (Epac2^ko/ko) mice (SI Fig. 6). Insulin granule fusion events induced by glucose stimulation alone in primary cultured pancreatic beta cells of Epac2^ko/ko mice (Fig. 3B Left) were similar to those of wild-type mice (Fig. 2A Left). However, potentiation of the first phase of glucose-induced fusion events by 8-Bromo-cAMP, which is seen in wild-type mice (Fig. 2 A Right and B), was markedly reduced in pancreatic beta cells of Epac2^ko/ko mice (Fig. 3B Right and SI Movie 4): the sum of fusion events (per 200 μm²) in the first phase by glucose stimulation alone and that by glucose plus 8-Bromo-cAMP stimulation were 25.0 ± 3.0 and 32.4 ± 4.7, respectively (Fig. 3C). In contrast, potentiation of the second phase of fusion events by 8-Bromo-cAMP was not affected in pancreatic beta cells of Epac2^ko/ko mice: the sum of fusion events in the second phase by glucose stimulation alone and that by glucose plus 8-Bromo-cAMP stimulation were 29.4 ± 4.2 and 51.5 ± 6.9, respectively (Fig. 3C). These results demonstrate that Epac2 is essential in the potentiation of insulin granule exocytosis by cAMP, primarily in the first phase of cAMP-potentiated exocytosis, by increasing the number of restless newcomer.

Fig. 3.

Epac2-mediated insulin secretion and insulin granule dynamics. (A) Effects of cAMP analogs and PKA inhibitor on insulin secretion. Pancreatic islets were preincubated for 30 min with Hepes-KRB containing 2.8 mM glucose alone or 2.8 mM glucose together with 10 μM H-89. The islets were then incubated for an additional 30 min with Hepes-KRB containing 11.1 mM glucose alone or 11.1 mM glucose plus 1 mM 8-Bromo-cAMP or 100 μM 8-pCPT-2′-O-Me-cAMP with or without 10 μM H-89. NS, no significant difference. *, P < 0.0001; **, P < 0.01 (Student's unpaired t test). Data were obtained from two independent experiments (n = 5 or 6 for each) and expressed as means ± SE. (B) Insulin granule dynamics in pancreatic beta cells of Epac2^ko/ko mice. (Left) Histogram of fusion events induced by stimulation of 16.7 mM glucose alone. (Right) Histogram of fusion events induced by stimulation of 16.7 mM glucose plus 1 mM 8-Bromo-cAMP. Primary cultured pancreatic beta cells were stimulated as described in Fig. 2A. Blue, red, and green bars indicate old face, restless newcomer, and resting newcomer, respectively. Data were obtained from six independent experiments (n = 2–5 for each) and expressed as means ± SE. (C) Comparison of fusion events in the first and second phases between glucose stimulation alone and glucose plus 8-Bromo-cAMP stimulation. Sum of fusion events in the first phase (the first 5 min after glucose stimulation) is shown on the left. Sum of fusion events in the second phase (from 5 to 15 min after glucose stimulation) is shown on the right. Open and filled bars indicate fusion events by glucose stimulation alone and those by glucose plus 8-Bromo-cAMP stimulation, respectively. *, P < 0.05 (Student's unpaired t test). Data were expressed as means ± SE.

Activation of Rap1 by cAMP Through Epac2 Mediates PKA-Independent Insulin Secretion.

Epac2 has been shown to exhibit guanine nucleotide exchange factor (GEF) activity toward Rap1 in a cAMP-dependent manner (17, 18). Although Rap1 is involved in the regulation of various cellular functions (19), its role in exocytosis is not known. To address this, we first examined the subcellular localization of Rap1 in MIN6 cells. Interestingly, most Rap1 was found to be colocalized with insulin granules in these cells (Fig. 4A). We then examined activation of Rap1 by cAMP in MIN6 cells. Rap1 was found not to be activated by glucose alone at 2.8 or 16.7 mM but was activated by 2.8 or 16.7 mM glucose plus GIP, GLP-1, or 8-Bromo-cAMP (Fig. 4 B and C). Rap1 also was activated by 8-pCPT-2′-O-Me-cAMP (Fig. 4D). The effect of 8-Bromo-cAMP or 8-pCPT-2′-O-Me-cAMP on activation of Rap1 was not inhibited by H-89 at 10 μM, a concentration sufficient to block PKA phosphorylation of CREB (Fig. 4E). These results demonstrate that Rap1 is activated in insulin-secreting cells in a cAMP-dependent, PKA-independent manner. Using pancreatic beta cells of Epac2^ko/ko mice, we intended to directly show that activation of Rap1 by cAMP is mediated by Epac2. However, the limited number of isolated pancreatic beta cells of these mice made such an experiment impossible. We therefore established clonal pancreatic beta cells lacking Epac2 (Epac2^ko/ko beta cell line) by crossbreeding Epac2^ko/ko mice and IT6 mice expressing simian virus 40 (SV40) large T antigen under human insulin promoter (20). No activation of Rap1 by 8-Bromo-cAMP was detected in this beta cell line, but activation was rescued by the introduction of wild-type Epac2 into the cell line by adenovirus-based gene transfer (Fig. 4F). These results indicate that Rap1 is activated by cAMP signaling specifically through Epac2 in pancreatic beta cells.

Fig. 4.

Activation of Rap1 through Epac2 and its effect in insulin-secreting cells. (A) Localization of Rap1 in MIN6 cells. After fixation, MIN6 cells were coimmunostained with anti-Rap1 antibody and anti-insulin antibody. (Scale bar, 10 μm.) (B) Activation of Rap1 in MIN6 cells by 8-Bromo-cAMP. After preincubation with 2.8 mM glucose, MIN6 cells were incubated with 2.8 or 16.7 mM glucose in the absence or presence of 1 mM 8-Bromo-cAMP for 15 min. GTP-bound Rap1 was affinity-purified by GST pull down assay by using GST-RalGDS-RID and detected with anti-Rap1 antibody. (C) Activation of Rap1 in MIN6 cells by GIP or GLP-1. After preincubation with 2.8 mM glucose, MIN6 cells were incubated with 2.8 mM glucose plus 10 nM GIP, 10 nM GLP-1, or 1 mM 8-Bromo-cAMP for 15 min. (D) Activation of Rap1 in MIN6 cells by 8-pCPT-2′-O-Me-cAMP. After preincubation with 2.8 mM glucose, MIN6 cells were incubated with 2.8 mM glucose plus 100 μM 8-pCPT-2′-O-Me-cAMP or 1 mM 8-Bromo-cAMP for 15 min. (E) Effects of H-89 on activation of Rap1 by cAMP analogs. MIN6 cells were preincubated for 30 min with Hepes-KRB containing 2.8 mM glucose alone (lanes 1, 2, and 4) or 2.8 mM glucose plus 10 μM H-89 (lanes 3 and 5). The cells were then incubated for an additional 15 min with 2.8 mM glucose plus 100 μM 8-pCPT-2′-O-Me-cAMP or 1 mM 8-Bromo-cAMP with or without 10 μM H-89. Phosphorylation of CREB in MIN6 cells by cAMP signal was detected with anti-phospho-CREB antibody. (F) Activation of Rap1 by 8-Bromo-cAMP in clonal pancreatic beta cells lacking Epac2 (Epac2^ko/ko beta cell line). Clonal cells were infected with adenovirus carrying β-galactosidase (Ad-β-gal) or Myc-Epac2 (Ad-Epac2) at a multiplicity of infection (MOI) of 200. After a 2-day culture, the infected cells were preincubated with 2.8 mM glucose and then incubated with 2.8 mM glucose plus 1 mM 8-Bromo-cAMP for 15 min. Myc-Epac2 was detected with anti-Myc antibody (Santa Cruz Biotechnology). (G) Effect of Rap1 siRNA on 8-Bromo-cAMP-potentiated insulin secretion in MIN6 cells. MIN6 cells were treated with control siRNA or Rap1 siRNA for 2 days. After preincubation with 2.8 mM glucose for 30 min, the cells were incubated with 16.7 mM glucose alone (open bars) or 16.7 mM glucose plus 1 mM 8-Bromo-cAMP (filled bars) for 1 h. Rap1 siRNA knocked down both Rap1A and -B (Upper). *, P < 0.05 (Student's unpaired t test). Similar results were obtained in three independent experiments (n = 4 or 5 for each) and expressed as means ± SE. (H) Inhibition of 8-Bromo-cAMP-induced Rap1 activation by Rap1GAP1. (Left) MIN6 cells were infected with adenovirus carrying β-galactosidase (Ad-β-gal) or mRFP-Rap1GAP1 (Ad-Rap1GAP1) at an MOI of 200. After a 2-day culture, the infected cells were preincubated with 2.8 mM glucose and then incubated with 2.8 mM glucose plus 1 mM 8-Bromo-cAMP for 15 min. mRFP-Rap1GAP1 was detected with anti-DsRed antibody (Takara Bio). (Right) Effects of Rap1GAP1 on 8-Bromo-cAMP-potentiated insulin secretion in isolated pancreatic islets. Isolated pancreatic islets were infected with adenovirus carrying β-galactosidase (open bars) or mRFP-Rap1GAP1 (filled bars) at an MOI of 200. After a 2-day culture, the infected islets were preincubated with 2.8 mM glucose and then incubated for an additional 30 min with 11.1 mM glucose alone or 11.1 mM glucose plus 1 mM 8-Bromo-cAMP. NS, no significant difference. *, P < 0.005 (Student's unpaired t test). Data were obtained from two independent experiments (n = 5 or 6 for each) and expressed as means ± SE.

We then evaluated Rap1 participation in cAMP-potentiated insulin secretion by using small interfering RNA (siRNA) against mouse Rap1. Both Rap1A and -B were expressed in MIN6 cells (SI Fig. 7), but Rap1 siRNA almost completely suppressed the expression of both Rap1A and Rap1B proteins (Fig. 4G). Under this condition, whereas glucose-induced insulin secretion was similar in control and Rap1 knockdown MIN6 cells, potentiation of insulin secretion by 8-Bromo-cAMP in Rap1 knockdown cells was decreased by ≈40% compared with control (Fig. 4G). We also investigated the effect of Rap1-specific GTPase-activating protein (Rap1GAP), which enhances the GTPase activity of endogenous Rap1 and blocks downstream signals of Rap1 on insulin secretion. The activation of Rap1 by 8-Bromo-cAMP was inhibited by the introduction of Rap1GAP1 (Fig. 4H Left). As shown in Fig. 4H Right, 8-Bromo-cAMP-potentiated insulin secretion was markedly decreased in pancreatic islets infected by adenovirus carrying Rap1GAP1 compared with control islets. These results demonstrate that Rap1 is required in cAMP-dependent, PKA-independent potentiation of insulin secretion.

Discussion

Epacs (Epac1 and Epac2) are recently discovered cAMP-binding proteins (9, 17, 18). By using specific analogs, Epacs have been shown to control integrin-mediated cell adhesion and cadherin-mediated cell junction formation (19). We recently found that Epac2 mediates cAMP-induced exocytosis in a PKA-independent manner (10), but the precise role of Epac2/Rap1 signaling in the exocytotic process is unknown. The present study indicates that Epac2 is essential in the first phase of insulin granule exocytosis potentiated by cAMP. This finding complements the observation that the PKA-resistant component of exocytosis gives rise to a rapid capacitance increase (21). Our data also show that Rap1 is activated by cAMP through Epac2 in pancreatic beta cells at a low concentration of glucose (2.8 mM), but that activation of Rap1 does not cause fusion events of insulin granules at that concentration of glucose. Thus, Epac2/Rap1 signaling is not involved in triggering exocytosis but may increase the size of the nondocked granule pool and/or facilitate recruitment of the granules to the plasma membrane.

To explore the possible mechanisms of Epac2/Rap1 signaling in the regulation of insulin granule dynamics, we developed a simulation model of insulin granule movement based on a random walk process (SI Fig. 8, SI Table 1, and SI Movie 5). The simulation suggests that a combination of density of insulin granules in a putative RRP near the plasma membrane and facilitation of recruitment of these granules to the plasma membrane is critical for glucose-induced fusion events with a biphasic pattern. Because our experimental data show that activation of cAMP signaling potentiates both the first and second phases of glucose-induced fusion events, cAMP signaling could well regulate both the density of the insulin granules in a putative RRP and the recruitment of these granules to the plasma membrane (SI Fig. 8E). The recent findings that cAMP promotes insulin granule exocytosis by both increasing the size of the RRP and accelerating the refilling of the RRP, as assessed by capacitance measurements (8, 21, 22), and that cAMP persistently increases the frequency of granule movement in a pancreatic beta cell line, as assessed by fluorescence image by quinacrine (23), accord with this notion. Because knockout of Epac2 blocks the potentiation by cAMP of fusion events in the first phase of glucose-induced fusion events specifically, Epac2/Rap1 signaling might well participate in the regulation of the density of insulin granules near the plasma membrane, likely reflecting the size of the RRP contribution to the potentiation of the first phase (SI Fig. 8F). Recent studies have shown that GLP-1 improves primarily the first-phase insulin response to glucose in type 2 diabetic patients (24). Thus, the effect of GLP-1 may well be mediated by Epac2/Rap1 signaling.

Rap1 is known to activate Vav2 and Tiam, guanine nucleotide exchange factors toward Cdc42 and Rac, members of the small G protein Rho family (25). Rho has been shown to stimulate trafficking of secretory vesicles by regulating actin dynamics (26). Because Cdc42 is activated by cAMP in insulin-secreting cells (SI Fig. 9), it may well be a downstream target of Epac2/Rap1 signaling in the regulation of insulin granule exocytosis. It has been shown that cAMP mediates ryanodine receptor-regulated Ca²⁺ mobilization independently of PKA, and that Epac2 is involved in this process (27, 28). In addition, 8-pCPT-2′-O-Me-cAMP has been shown to mobilize Ca²⁺ from intracellular Ca²⁺ stores (29). It is possible, therefore, that Epac2/Rap1 signaling is also involved in Ca²⁺ mobilization in the regulation of insulin granule exocytosis.

cAMP signals are known to be compartmented in different regions of a cell (30). Such cAMP compartmentation contributes to diverse cAMP-mediated cellular functions (31). Because Epac2 binds to cAMP with a much lower affinity than PKA does (3), a cAMP compartment containing Epac2 should be distinct from one containing PKA (3). Thus, because Rap1 is activated specifically through Epac2 in a cAMP-dependent, PKA-independent manner, Rap1-mediated exocytosis is closely associated with an Epac2-containing cAMP compartment. In addition to pancreatic beta cells, Epacs have been shown to mediate cAMP-dependent, PKA-independent exocytosis in various secretory cells such as certain neurons, pituitary cells, and acinar cells in parotid glands (32–34). Because Rap1 is expressed widely in tissues, Epac/Rap1 signaling might also regulate exocytosis in these cells by controlling the dynamics of synaptic vesicles or dense-core granules.

Materials and Methods

Recombinant Proteins.

Construction of recombinant proteins was performed as described in SI Methods.

Pull-Down Assay for GTP-Rap1.

MIN6 cells were cultured in growth medium as described in ref. 20. Pull-down assay for GTP-Rap1 was performed as described in ref. 35, with minor modification 35. See SI Methods for details.

Insulin Secretion Experiments.

Insulin secretion experiments using mouse pancreatic islets and MIN6 cells were performed as described in SI Methods.

Analysis by TIRFM.

Primary cultured beta cells isolated from mouse pancreatic islets were subjected to analysis by TIRFM as described in SI Methods.

Immunocytochemical Analysis.

Immunocytochemical analysis was performed as described in SI Methods.

Targeted Disruption of the Epac2 Gene.

Epac2 knockout mice (Epac2^ko/ko) were generated by placing tandemly oriented lox P sites around a 1.3-kb DNA fragment containing a part of exon 1 and intron 1 of the mouse Epac2 gene, and subsequently removing them with Cre recombinase (SI Fig. 6A). The lack of expression of Epac2 in the pancreatic islets of Epac2^ko/ko mice was confirmed by reverse transcriptase PCR analysis (SI Fig. 6B).

Generation of Epac2-Deficient Pancreatic Beta Cell Line.

To investigate the role of Epac2 at the cellular level, pancreatic beta cell lines lacking Epac2 (Epac2^ko/ko beta cells) were established by crossbreeding Epac2^ko/ko mice and IT6 mice expressing SV40 large T antigen under human insulin promoter (20) that develop highly differentiated beta cell tumors. Twenty-three lines of Epac2^ko/ko beta cells were generated from a 7-week-old mouse lacking Epac2 and carrying a large T antigen load (Epac2^ko/ko; IT6 mouse) by a method modified from the original protocol (20). See SI Methods for further details.