Epac: A New cAMP-Binding Protein in Support of Glucagon-Like Peptide-1 Receptor-Mediated Signal Transduction in the Pancreatic β-Cell

To appreciate the role of Epac in β-cell function, it is necessary to define the molecular basis for cAMPGEF signal transduction. The cAMPGEFs are cAMP-binding proteins first identified on the basis of their ability to stimulate the exchange of GDP for GTP at the guanyl nucleotide binding sites of the Rap1 and Rap2 small molecular weight G proteins (13,14). Two isoforms of cAMPGEFs exist and they are referred to as Epac1 and Epac2—the Exchange Proteins directly Activated by cAMP (13). Epac1 and Epac2 are encoded by two distinct genes, and mRNAs corresponding to both isoforms of Epac are detectable in pancreas (14), islets (15), and insulin-secreting cell lines (15). The binding sites for cAMP located near the NH₂-terminus of cAMPGEFs bear structural resemblance to the B-type cAMP recognition site of PKA regulatory subunit I (13,14). They contain the Pro-Arg-Ala-Ala motif that confers selectivity for binding of cAMP over cGMP. Epac1 contains a single cAMP-binding domain, whereas Epac2 contains two—a low affinity cAMP-binding domain designated as “A” and a higher affinity cAMP-binding domain designated as “B” (Fig. 2). The K_d for cAMP binding to full-length Epac1 is 2.8 μmol/l (16). For Epac2 the “A” and “B” binding sites exhibit a K_d of 87 and 1.2 μmol/l, respectively, when each binding site is studied in isolation (17). It may be concluded that cAMP exhibits a lower affinity for Epac than it does for PKA (K_d 0.1–1.0 μmol/l) (16). Such observations suggest that Epac is activated in a region of the cell where cAMP formation is strongly stimulated or in a subcellular compartment where the ability of cAMP to activate Epac is not inhibited by the activity of cAMP phosphodiesterase.

Analysis of the Epac2 cDNA indicates that the two cAMP-binding domains are connected via a linker sequence to a GEF catalytic domain responsible for activation of Rap1. In the GEF domain are found three structurally conserved regions diagnostic of GEFs that interact with G proteins homologous to Ras. Binding of cAMP to cAMPGEFs activates Rap1 because it relieves an autoinhibitory function exerted by the cAMP-binding domain at the GEF’s catalytic domain (18). Both isoforms of Epac contain a Ras exchange motif (REM) thought to be responsible for stabilization of the GEF. Also present is a DEP domain where homologies to disheveled, Egl-10, and pleckstrin are found. The DEP domain may play a role in conferring to Epac its cell cycle-dependent targeting to intracellular membranes including the nuclear envelope (19). In this regard, a potential role for Epac in the regulation of mitosis is indicated because Cheng et al. (19) report that Epac localizes to the mitotic spindle and centrosomes during metaphase.

The interaction of Epac with Rap1 is of interest because this G protein is expressed in β-cells, where it is suggested to play a stimulatory role in glucose-dependent insulin secretion (20). Although the signal transduction properties of Rap1 are poorly understood, it is established that Rap1 acts as a molecular switch: it is active in its GTP-bound form and inactive in its GDP-bound form. Rap1 possesses an intrinsic GTPase activity so that the Epac-mediated activation of Rap1 is terminated by GTP hydrolysis. The hydrolysis of GTP is accelerated by GTPase Activating Proteins (GAPs), such as Rap1GAP, thereby completing a full cycle of activation and inactivation.

Exactly how activated Rap1 influences cellular function remains something of a mystery. Rap1 will antagonize Ras-mediated signal transduction, and a role for Rap1 in the integrin-mediated stimulation of cell adhesion is demonstrable (21). Furthermore, activation of Rap1 and/or Rap2 might play a role in the stimulation of protein kinase B (PKB, Akt) (22), phospholipase C-ε (PLC-ε) (23), and the ERK1/2 mitogen-activated protein kinases (MAPKs) (24). Interestingly, Rap1 has been reported to be located on secretory granule membranes, as expected if it plays a role in exocytosis (25). It should be noted that not all actions of Epac are necessarily Rap mediated. For example, Epac is reported to interact directly with the insulin granule-associated proteins Rim2 and Piccolo (6,7) and also the β-cell sulfonylurea receptor (SUR1) (5). Epac may also activate c-Jun kinase (JNK) independently of Rap1 (26). Table 1 summarizes signaling pathways that are proposed to be regulated by Epac. Note that because this is a new area of study, not all citations listed refer to β-cells.

Analysis of cAMPGEF signal transduction has been complicated by the lack of suitable tools with which to activate or inhibit Epac selectively. For example, the cAMP agonist Sp-cAMPS activates PKA and Epac, whereas the cAMP antagonist Rp-cAMPS is a poor blocker of Epac but an effective blocker of PKA (16). The limited ability of Rp-cAMPS to prevent activation of Epac may explain previous reports that cAMP exerts Rp-cAMPS-insensitive stimulatory effects on β-cell Ca²⁺ signaling and insulin secretion (27). With these points in mind, it is clear that previous studies demonstrating Rp-cAMPS-insensitive and apparently PKA-independent actions of GLP-1 must now be considered in a new light (28–30). It seems likely that at least some “PKA-independent” actions of GLP-1 may in fact be Epac-mediated.

An important new advance is the development of a cAMP analog (8-pCPT-2′-O-Me-cAMP) that activates Epac in a selective manner (31). The selectivity with which 8-pCPT-2′-O-Me-cAMP acts is explained by the fact that interactions of cAMP with PKA are contingent on the presence of a 2′-OH group on the ribose moiety of the cyclic nucleotide. Substitution of the 2′OH with a 2′-O-Me group generates a cAMP analog exhibiting high affinity for Epac (K_d 2.2 μmol/l for Epac1) and a reduced affinity for PKA (K_d 20–30 μmol/l) (31). New studies demonstrate that in β-cells, 8-pCPT-2′-O-Me-cAMP increases [Ca²⁺]_i and stimulates exocytosis (10,11,32). These effects of 8-pCPT-2′-O-Me-cAMP are not blocked by the cAMP antagonist 8-Br-Rp-cAMPS or by the PKA inhibitors H-89 and KT5720 (10,11). Conversely, studies examining PKA-mediated signal transduction in the β-cell are likely to be facilitated by the use of N6-benzoyl-cAMP, a cAMP analog that preferentially activates PKA but not Epac (16).

It is also possible to evaluate the role of Epac in β-cell function by use of a “dominant negative” mutant Epac2 in which inactivating G114E and G422D amino acid substitutions are introduced into the “A” and “B” cAMP-binding domains (6). Similarly, a dominant negative Epac1 incorporating an R279E substitution within the single cAMP-binding domain has been described (19). Dominant negative Epac2 blocks stimulatory actions of GLP-1 and 8-pCPT-2′-O-Me-cAMP on Ca²⁺ signaling and exocytosis in β-cells (6,8,10–12). It is also possible to generate constitutively active Epac1 or Epac2 by truncation of the cAMPGEF NH₂-terminus to remove autoinhibitory regulatory domains that bind cAMP (17). By use of transient transfection and overexpression of dominant negative or constitutively active Epac, efforts are now under way to evaluate the potential role of Epac1 or Epac2 in the regulation of β-cell stimulus-secretion coupling.

Seino and colleagues (5,6) first reported that Epac2 interacts directly with the insulin granule-associated protein Rim2. This finding is of significance because the closely related isoform Rim1 is a Rab3-interacting molecule that is demonstrated to play a central role in Ca²⁺-dependent exocytosis. Studies of neurons indicate that Rim1 promotes the priming of synaptic vesicles, thereby rendering them release competent (33). Such findings are of significance because the studies of Sharp et al. (34) indicate that insulin granule priming is a necessary step for glucose-dependent first- and second-phase secretion. Therefore, by interacting directly with Epac2, Rim2 may accelerate insulin granule priming in a cAMP-regulated manner. The net effect is expected to be a facilitation of insulin secretion (Fig. 3A).

Epac2 is also reported to interact with Piccolo, a CAZ protein (Cytoskeletal matrix protein that associates with the Active Zone) originally described in studies of presynaptic nerve endings. Seino et al. (7) report that Epac2, Rim2, and Piccolo form a macromolecular complex, the stoichiometry of which has yet to be determined. Given that the binding of Ca²⁺ to Piccolo stimulates the formation of this complex, it may be speculated that the Epac2-Piccolo-Rim2 macromolecular complex may confer stimulatory effects of cAMP and Ca²⁺ on insulin secretion (Fig. 3A). This model of Epac signal transduction is consistent with the known ability of cAMP to facilitate Ca²⁺-dependent exocytosis in the β-cell (11,27).

The potential role of Epac as a regulator of insulin secretion is emphasized by studies demonstrating an interaction of Rim2 with the Ras-related G protein Rab3A (5). Rab3A is located on the cytoplasmic surface of insulin secretory granules, and it plays a role in the recruitment and docking of granules at the plasma membrane. Although the precise role of Rab3A in the regulation of insulin secretion has yet to be defined, Rab3A knockout (KO) mice exhibit a defect of insulin secretion, are glucose intolerant, and develop fasting hyperglycemia (35). However, unlike its role in the regulation of Rap1, Epac2 does not appear to stimulate the exchange of guanyl nucleotides at Rab3A. Instead, Epac2 appears to act downstream of Rab3A by virtue of its ability to form a complex with Rim2. It may be speculated that by interacting with Rim2, Epac2 is targeted to docked secretory granules containing Rab3A. Once situated at the docked secretory granules, Epac2 might then mediate stimulatory effects of cAMP on exocytosis (Fig. 3A).

Mechanistically, a stimulatory effect of cAMP on insulin secretion might result from an Epac-mediated increase in the size of the readily releasable pool (RRP) of secretory granules. Rorsman et al. (8) propose that such an effect of Epac2 is achieved by virtue of its ability to promote insulin granule acidification, a step necessary for granule priming. It is suggested that the binding of cAMP to Epac2 promotes the opening of ClC-3 chloride channels that are located in the secretory granule membrane. cAMP-induced influx of Cl⁻ into the secretory granule lumen then creates an electromotive force that facilitates ATP-dependent H⁺ uptake mediated by a V-type H⁺-ATPase (Fig. 3B). How this takes place is not certain but it may involve sulfonylurea receptors expressed in the granule membrane (gSUR or SUR1) and/or plasma membrane (SUR1). These sulfonylurea receptors may act as transmembrane conductance regulators of ClC-3 channels so that the interaction of gSUR or SUR1 with ClC-3 might be Epac regulated. One attractive feature of this model is that it explains not only the ATP dependence of granule priming, but also the potential role of the Epac2-Rim2-Piccolo macromolecular complex as a determinant of the RRP size.

Epac appears to mediate stimulatory actions of GLP-1 on Ca²⁺-dependent insulin secretion. Evidence in support of this contention is provided by studies demonstrating that the action of GLP-1 is reduced by downregulation of the expression of Epac2 in mouse islets (6). In these same studies, the action of GLP-1 is shown to be reproduced by 8-Br-cAMP, and the effect of 8-Br-cAMP is inhibited by overexpression of dominant negative Epac2. Stimuli that produce an increase of [Ca²⁺]_i (e.g., KCl and carbachol) magnify the effect of 8-Br-cAMP, but this interaction is diminished under conditions of reduced Epac2 expression. Such findings suggest that Epac2 promotes Ca²⁺-dependent exocytosis, a conclusion supported by recent electrophysiological studies of mouse β-cells. Rorsman et al. (11) report that GLP-1 acts via Epac2 to increase the size of the RRP of secretory granules by as much as 2.3-fold. This action of GLP-1 facilitates fast Ca²⁺-dependent exocytosis measured as an increase of membrane capacitance occurring within 200 ms of membrane depolarization. Facilitation of fast exocytosis is reduced by downregulation of Epac2 expression or by transfection with dominant negative Epac2. It is also reproduced by 8-pCPT-2′-O-Me-cAMP in an Rp-cAMPS-insensitive manner, thereby confirming the likely involvement of Epac2 in this signaling pathway (11).

A substantial body of evidence indicates that Epac is unlikely to be the sole cAMP-binding protein subserving stimulatory effects of GLP-1 on insulin secretion. Overexpression of AKAP18 (an A-Kinase Anchoring Protein) in RINm5F cells facilitates insulin secretion measured in response to GLP-1 (36). This is significant because AKAP18 interacts with the RII regulatory subunit of PKA and targets the kinase to the plasma membrane. In contrast, a sequestration of PKA and a concomitant inhibition of GLP-1-stimulated insulin secretion is observed after overexpression of a mutant AKAP18 that fails to target PKA to the plasma membrane (36). Inhibition of GLP-1-stimulated insulin secretion is also observed after exposure of islets to inhibitory peptides that disrupt the interaction of PKA with AKAPs (37). These findings demonstrate that under appropriate experimental conditions, the subcellular distribution of PKA dictates the effectiveness of GLP-1 as an insulin secretagogue. Recently, Wolf et al. (38) elaborated on these findings by providing evidence that it is the β isoform of the PKA catalytic subunit (Cβ) that mediates stimulatory actions of GLP-1 at the plasma membrane. Interestingly, GLP-1 is shown to stimulate the translocation of Cβ from the cytosol to the plasma membrane in betaTC6 insulin-secreting cells. Therefore, the subcellular distribution of activated catalytic subunit is not dictated solely by the distribution of AKAPs.

Seino et al. (3) report that treatment of mouse islets with Epac2 antisense oligodeoxynucleotides reduces but does not abrogate the insulin secretagogue action of GLP-1. In contrast, exposure of mouse islets to the PKA inhibitor H-89 in the presence of Epac2 antisense oligodeoxynucleotides results in a complete failure of GLP-1 to act. These findings are understandable in terms of electrophysiological studies of mouse β-cells demonstrating that cAMP acts via PKA to recruit secretory granules from a reserve pool to an RRP (11,27). This action of cAMP is responsible for a facilitation of slow Ca²⁺-dependent exocytosis because it allows replenishment of the RRP under conditions in which a sustained increase of [Ca²⁺]_i occurs. Slow exocytosis is defined by Rorsman et al. (27) as an increase of membrane capacitance that occurs in response to repetitive membrane depolarization, but only after a delay of ∼500 ms. The delay that precedes the appearance of slow exocytosis presumably reflects the time required to recruit reserve granules to the plasma membrane where they may then undergo docking and ATP-dependent priming in order to become release competent. The facilitation of slow exocytosis by cAMP is reproduced by GLP-1, is independent of Epac, and is blocked by Rp-cAMPS, thereby confirming the involvement of PKA. Importantly, the action of cAMP is not observed in the complete absence of Ca²⁺ (27). This key observation contrasts with one prior report that cAMP stimulates Ca²⁺-independent exocytosis (39). However, the significance of Ca²⁺-independent exocytosis is uncertain given that a Ca²⁺ channel blocker (nifedipine) abrogates stimulatory effects of GLP-1 on islet insulin secretion (30). Furthermore, Henquin et al. (40) report that under conditions of mild or stringent Ca²⁺ deprivation, GLP-1 does not potentiate glucose-dependent insulin secretion from mouse or rat islets.

One difficulty associated with the interpretation of prior patch clamp electrophysiological studies is that the measurements of exocytosis obtained in this manner do not necessarily reflect insulin secretion. This uncertainty results from the fact that β-cells contain not only large dense core secretory granules, but also synaptic vesicle-like structures that undergo Ca²⁺-dependent exocytosis. Therefore, measurements of membrane capacitance might reflect vesicular rather than insulin granule exocytosis. If this were to be the case, it might be erroneously concluded that fast exocytosis plays a significant role in β-cell insulin secretion.

In an attempt to avoid these complications, Kasai and colleagues (41,42) measure exocytosis in mouse β-cells by means of carbon fiber amperometry. This method of analysis relies on the loading of β-cells with serotonin (5-HT), a monoamine that is selectively sequestered in large dense core secretory granules. By measuring the release of 5-HT, it is possible to detect exocytosis of single granules, thereby providing an indirect assay of insulin release. Amperometric determinations of exocytosis are not without interpretive difficulty because exposure of β-cells to 5-HT is known in some cases to interfere with glucose-dependent insulin secretion (43,44). However, using this approach, Kasai and colleagues (41,42) provide the surprising finding that PKA increases the release probability of docked secretory granules, an effect mediated at a step subsequent to ATP-dependent priming. This model of Kasai and colleagues is in contrast to that proposed by Rorsman and colleagues (11,27) because it suggests that fast Ca²⁺-dependent exocytosis is regulated not only by Epac, but also by PKA.

One interesting aspect of the model proposed by Kasai and colleagues is that activation of PKA occurs in response to an elevation of [ATP]_i secondary to glucose metabolism. It is proposed that the availability of ATP is a limiting factor for cAMP production and that adenylyl cyclase acts as a low-affinity ATP sensor, transducing a glucose-dependent increase of [ATP]_i into an increase of [cAMP]_i (41,42). In this model, the basal activity of adenylyl cyclase is suggested to be sufficiently high so as to allow significant cAMP production under conditions in which β-cells are exposed to glucose in the absence of cAMP-elevating agents such as forskolin. Kasai and colleagues then propose that glucose-dependent insulin secretion is contingent on the activation of PKA, which exerts a “postpriming” stimulatory effect at the docked insulin secretory granules. This model is in accordance with the role of ATP as a precursor for cAMP. It is also consistent with the ATP dependence of cAMP action because ATP serves as a phosphate donor in support of PKA-mediated phosphorylation.

What remains to be demonstrated is that glucose-dependent ATP production is a direct stimulus for cAMP biosynthesis. Although previous studies demonstrate that glucose increases the level of cAMP in the β-cell, this effect is usually attributed to Ca²⁺-dependent activation of adenylyl cyclase (45). It is also uncertain whether activation of PKA is an absolutely necessary step in support of glucose-dependent insulin secretion. Kasai and colleagues. (41,42) report that Rp-cAMPS abrogates stimulatory effects of Ca²⁺ on exocytosis in mouse β-cells. On the basis of this observation, it is suggested that the postpriming action of PKA at insulin secretory granules is an essential determinant of stimulus-secretion coupling. However, prior studies directly measuring glucose-induced insulin secretion from intact islets offer no support for an absolute PKA dependence of secretagogue action (46,47). Furthermore, Kang et al. (9) demonstrate that under conditions in which PKA signaling is abrogated, exocytosis is readily observed in response to stimuli that mobilize Ca²⁺ from intracellular Ca²⁺ stores.

Whether it is Epac or PKA that plays a dominant role as a determinant of insulin secretion remains to be elucidated. In studies of mouse islets, Nakazaki et al. provide the surprising finding that PKA plays little or no role in the stimulation of insulin secretion by GLP-1 (30). Similarly, in studies of INS-1 insulin-secreting cells, inhibition of PKA by use of 8-Br-Rp-cAMPS fails to block exocytosis in response to the GLP-1 receptor agonist exendin-4 (9). In contrast, Gromada and colleagues provide clear evidence for a PKA-dependent mechanism by which GLP-1 stimulates insulin secretion in human and mouse β-cells (48–50). Perhaps one clue that resolves this quandary is provided by new studies examining the action of GLP-1 in SUR1 KO mice. When exposed to GLP-1, these mice secrete unusually small amounts of insulin despite the fact that GLP-1 remains effective as a stimulator of cAMP production (30,51). This loss of secretagogue action might be explained by the failure of β-cells to express ATP-sensitive K⁺ channels (K_ATP) because SUR1 is a subunit of K_ATP channels and because prior studies demonstrate that K_ATP channels are inhibited by GLP-1 (29,48–50,52–54, but see ref. 55). Since the inhibition of K_ATP channels by GLP-1 is mediated at least in part by PKA (54), these studies of SUR1 KO mice hint at a predominant role for PKA as the principal effector by which GLP-1 acts.

Such a conclusion is tempered, however, by one recent study demonstrating that the signaling properties of Epac are downregulated in SUR1 KO mice (11). Although a facilitation of fast exocytosis by 8-CPT-2′-O-Me-cAMP is observed in wild-type mice, it is absent in the KO mice (11). To explain these findings, Eliasson et al. (11) propose that plasma membrane-associated SUR1 recruits Epac2 to the inner cell surface where it mediates stimulatory effects of cAMP on secretory granule priming. Because SUR1 is also reported to be expressed in secretory granules (56) and because evidence exists for a sulfonylurea receptor-like protein (gSUR) in the granule membrane (11), the recruitment of Epac2 might not be limited to the plasma membrane (Fig. 3B). If the model of Eliasson et al. should prove to be correct, it may be concluded that the stimulation of fast Ca²⁺-dependent exocytosis by GLP-1 is an Epac-regulated event that might play a significant role in stimulus-secretion coupling.

Recent studies of Kang and colleagues demonstrate that Epac is also likely to play a significant role in the regulation of β-cell Ca²⁺ signaling. Exposure of human β-cells or INS-1 insulin-secreting cells to the Epac-selective cAMP analog 8-pCPT-2-O-Me-cAMP produces a transient increase of [Ca²⁺]_i that reflects the mobilization of Ca²⁺ from intracellular Ca²⁺ stores (10). In both cell types this action of 8-pCPT-2-O-Me-cAMP is accompanied by exocytosis as measured by amperometric detection of released 5-HT (8,10). Two lines of evidence indicate that the Ca²⁺-mobilizing action of 8-pCPT-2-O-Me-cAMP is Epac-mediated. First, 8-pCPT-2-O-Me-cAMP remains effective under conditions in which INS-1 cells are treated with a cAMP antagonist (8-Br-Rp-cAMPS) or a selective inhibitor of PKA (H-89) (10). Second, overexpression of dominant negative Epac2 blocks the action of 8-pCPT-2-O-Me-cAMP in transfected INS-1 cells (10). Interestingly, the Ca²⁺-mobilizing action of 8-pCPT-2-O-Me-cAMP in INS-1 cells is not accompanied by a detectable increase of inositol trisphosphate (IP₃) production (10). This key finding demonstrates that the action of 8-pCPT-2-O-Me-cAMP in INS-1 cells is unlikely to be explained by Epac-mediated activation of PLC-ε with concomitant IP₃ production and the release of Ca²⁺ from IP₃ receptor (IP₃-R)-regulated Ca²⁺ stores.

Prior studies of β-cells offer competing hypotheses as to how cAMP mobilizes an intracellular source of Ca²⁺. Liu et al. (57) and Tengholm et al. (58) report that cAMP promotes PKA-dependent sensitization of the IP₃-R, thereby facilitating CICR from the endoplasmic reticulum. Although this signaling mechanism may exist in β-cells, it is unlikely to explain the Ca²⁺-mobilizing action of 8-pCPT-2-O-Me-cAMP. Instead, attention has recently focused on the possible involvement of an alternative source of intracellular Ca²⁺, one that is regulated by ryanodine receptor (RYR) intracellular Ca²⁺ release channels. Kang et al. (10) report that pretreatment of human β-cells or INS-1 cells with ryanodine effectively blocks the increase of [Ca²⁺]_i measured in response to 8-pCPT-2-O-Me-cAMP. Because GLP-1 also stimulates ryanodine-sensitive CICR in β-cells (59–61), it may be speculated that a functional coupling exists between the GLP-1-R, Epac, and RYR (Fig. 4). Indeed, our most recent studies demonstrate that the GLP-1-R agonist exendin-4 stimulates CICR under conditions in which INS-1 cells are treated with 8-Br-Rp-cAMPS or H-89 (8). Furthermore, this action of exendin-4 is abrogated by overexpression of dominant negative Epac2 (G.G.H. and G. Kang, unpublished observations).

What remains to be determined is exactly how activation of Epac is translated into the release of Ca²⁺ from Ca²⁺ stores. Such an effect might require direct interactions of Epac with Ca²⁺ release channels or, alternatively, Epac might act via an intermediary to promote kinase-mediated phosphorylation of the channels. Either mechanism might increase the sensitivity of Ca²⁺ release channels to cytosolic Ca²⁺, thereby gating the channel from a closed to an open conformation (10). Such a mechanism of intracellular Ca²⁺ channel modulation might complement the previously reported ability of cAMP to sensitize Ca²⁺ release channels in a PKA-dependent manner (62–65).

One promising avenue of future investigation concerns the possible importance of Epac as a determinant of mitochondrial metabolism. Recent studies of MIN6 insulin-secreting cells demonstrate that Ca²⁺ is a direct stimulus for glucose-dependent mitochondrial ATP production and that an increase of mitochondrial [ATP] is measurable under conditions in which CICR is triggered by cAMP (12). Although previous studies of β-cells offer competing interpretations as to how Ca²⁺ influences ATP production (66,67), the findings obtained with MIN6 cells hint at a direct stimulatory effect of Ca²⁺ on mitchondrial dehydrogenases important to oxidative phosphorylation (Fig. 5). Indeed, Tsuboi et al. (12) report that GLP-1 acts via Epac to trigger CICR from ryanodine-sensitive Ca²⁺ stores, thereby raising the intramitochondrial concentrations of Ca²⁺ and ATP. These actions of GLP-1 are reproduced by forskolin, are inhibited by dominant negative Epac2, and are abrogated by pretreatment of MIN6 cells with a Ca²⁺ chelator (12).

Because GLP-1 stimulates free fatty acid production in β-cells (68), fatty acids might in theory serve as substrates for mitochondrial ATP production, thereby suggesting that the cAMP-dependent increase of mitochondrial [ATP] reported by Tsuboi et al. originates not only from glucose metabolism but also from lipid metabolism. It also remains to be determined to what extent GLP-1 acts via Epac to stimulate ATP production in authentic β-cells and, if so, whether the observed alterations of [ATP]_i are translated into major alterations of K_ATP channel function, insulin granule transport, priming, and exocytosis (Fig. 5). If these criteria are met, the findings of Tsuboi et al. offer one simple explanation as to why all reported insulin secretagogue actions of GLP-1 are absolutely contingent on simultaneous exposure of β-cells to glucose.

This article focuses on the actions of GLP-1 that are likely to be Epac mediated and which might play a significant role in the regulation of β-cell stimulus-secretion coupling. Future studies will undoubtedly expand on this avenue of investigation to examine the possibility that Epac subserves stimulatory effects of glucagon, glucose-dependent insulinotropic peptide, or pituitary adenylyl cyclase activating polypeptide on β-cell function. Moreover, it will be of interest to ascertain whether such effects complement the previously reported actions of GLP-1 to produce long-term alterations in the pattern of gene expression important to β-cell neogenesis, differentiation, and survival (69–71). Should this prove to be the case, interest within the diabetes research community might spur efforts to develop new pharmacological agents that activate Epac in a direct and selective manner.

Shibasaki et al. (Shibasaki T, Sunaga Y, Fujimoto K, Kashima Y, Seino S: Interaction of ATP sensor, cAMP sensor, Ca²⁺ sensor, and voltage-dependent calcium channel in insulin granule exocytosis (J Biol Chem. In press) have recently provided additional biochemical evidence for a direct protein-protein interaction between Epac2 and the nucleotide binding fold-1 (NBF-1) of SUR1.

G.G.H. thanks the National Institutes of Health (grant nos. DK45817 and DK52166); the American Diabetes Association (Research Grant Award); the Marine Biological Laboratory, Woods Hole, Massachusetts (Summer Research Fellowship); and the human islet distribution program of the Juvenile Diabetes Research Foundation International.

Special thanks to O.G. Chepurny for critical reading of the manuscript.