Interaction Between Munc13-1 and RIM Is Critical for Glucagon-Like Peptide-1–Mediated Rescue of Exocytotic Defects in Munc13-1–Deficient Pancreatic β-Cells

PCR and DNA electrophoresis were performed to determine the genotypes of the Munc13-1^+/+ wild-type and Munc13-1^+/− knockout mice (10). All experimental procedures are approved by the University of Toronto.

Isolated islets were solubilized in sample buffer (2% SDS) and loaded and separated on SDS-PAGE (Epac2, 30 μg protein, 6% PAGE; Rim2α, 30 μg protein, 10% PAGE; and β-actin, 30 μg protein, 10% PAGE). Separated proteins were identified with primary antibodies (goat anti-Epac2 [Santa Cruz Biotechnology, Santa Cruz, CA], 1:100; rabbit anti-Rim2α, 1:1,000 [Synaptic Systems, Gottingen, Germany]; monoclonal mouse anti–β-actin, 1:20,000 [Sigma, St. Louis, MO]) and appropriate peroxidase-labeled second antibodies, visualized by chemiluminescence (Pierce).

Mouse pancreatic islets were isolated as described previously (11), allowed to recover for at least 2 h (11 mmol/l glucose), and then cultured in 2.8 mmol/l glucose (1 h). Batches of ∼25 islets were placed in perifusion chambers with a capacity of ∼1.3 ml at 37°C and perifused at a flow rate of ∼1 ml/min with a Krebs-Ringer bicarbonate HEPES buffer (KRBH; 10 mmol/l HEPES, pH 7.4, and 0.07% BSA). Islets were equilibrated for 30 min in KRBH (2.8 mmol/l glucose) before being subjected to the following stimulation protocols: 1) 2.8 mmol/l glucose for 10 min followed by 16.7 mmol/l glucose plus 10 nmol/l GLP-1 (7-36)-amide (Bachem, Torrance, CA) for 40 min; 2) 2.8 mmol/l glucose for 10 min, then 16.7 mmol/l glucose for 40 min, and a final 16.7 mmol/l glucose plus 10 nmol/l GLP-1 for 40 min; and 3) 2.8 or 16.7 mmol/l glucose, in presence of KCl (30 mmol/l) plus diazoxide (250 μmol/l) as described by Henquin (31). Fractions were collected for insulin determination by radioimmunoassay (RIA) (Linco Research, St. Louis, MO). At the end of each perifusion, islets were collected from the chamber and lysed in acid-ethanol (0 mmol/l HCl in 75% ethanol) for assessment of insulin content. Results are presented as insulin secreted normalized to islet insulin content.

For cAMP measurements, batches of 10–15 islets were preincubated in KRBH (2.8 mmol/l glucose, 1 h, 37°C) with 1 mmol/l isobutylmethylxanthine (IBMX; prevents cAMP degradation by inhibiting cyclic nucleotide phosphodiesterase activity) and then stimulated with glucose (16.7 mmol/l) ± 10 nmol/l GLP-1 (1 h, 37°C). Islet cAMP levels were then determined by RIA (Biomedical Technologies, Stoughton, MA).

Single cells were obtained by dispersing mouse pancreatic islets (11). The intracellular pipette solution contained 125 mmol/l K-glutamate, 10 mmol/l KCl, 10 mmol/l NaCl, 1 mmol/l MgCl₂, 5 mmol/l HEPES, 0.05 mmol/l EGTA, and 3 mmol/l MgATP, pH 7.1. For cAMP dialysis experiments, 0.1 mmol/l cAMP, 10 μmol/l 8-(4-chloro-phenylthio)-2′-O-methyladenosine-3′,5′-cyclic monophosphate (8-pCPT-2′-O-Me-cAMP) (Biolog, Hayward, CA), or 100 μmol/l N⁶-benzoyladenosine-cAMP (N⁶-Bnz-cAMP) (Biolog) was included in the intracellular solution. For Ca²⁺ infusion experiments, the same pipette solution plus or minus 0.1 mmol/l cAMP was used except that EGTA was changed to 10 mmol/l, and 9 mmol/l CaCl₂ was added, resulting in a free Ca²⁺ concentration estimated as 1.5 μmol/l (32). For dialysis experiments with glutathione S-transferase (GST) fusion proteins, 1 μmol/l GST–Rim2α-PDZ or GST–Rim2α-PDZ(AAA) (gifts from S. Seino, Division of Cellular and Molecular Medicine, Kobe University Graduate School of Medicine, Kobe, Japan) was included in the intracellular solution. Extracellular solution consisted of 138 mmol/l NaCl, 5.6 mmol/l KCl, 1.2 mmol/l MgCl₂, 2.6 mmol/l CaCl₂, 5 mmol/l HEPES, and 5 mmol/l d-glucose, pH 7.4. For H89 experiments, β-cells were pretreated with 3 μmol/l H89 (Sigma) for ∼10 min, and this H89 concentration was maintained in the extracellular solution during recordings. Cell membrane capacitance (C_m) was estimated by Lindau-Neher technique (33), implementing the “Sine + DC” feature of the Lock-in module (40 mV peak-to-peak and a frequency of 500 Hz for depolarization-evoked exocytosis) in whole-cell configuration. Recordings were conducted using EPC10 patch clamp amplifier, Pulse, and X-Chart softwares (HEKA Electronik, Lambrecht, Germany). Exocytic events were elicited by a train of eight 500-ms depolarization pulses (1-Hz stimulation frequency) from −70 to 0 mV. For Ca²⁺ infusion experiments, a 40-mV peak-to-peak 800-Hz sine wave about the holding potential (−70 mV) was applied (32). All recordings were performed at 30°C.

Results are expressed as means ± SE where appropriate. Statistical comparisons were performed by t test in which an acceptable level of significance was considered at P < 0.05.

Using islet perifusion, we examined biphasic glucose-stimulated insulin secretion (GSIS) in islets from Munc13-1^+/+ and Munc13-1^+/− mice. Glucose (16.7 mmol/l) stimulated a biphasic secretory pattern in both groups of islets (Fig. 1A). However, Munc13-1^+/− islets displayed a markedly lower level of insulin secretion in both phases compared with Munc13-1^+/+ islets, culminating in 37% reduction in F-PIS (3.42 ± 0.33 vs. 2.17 ± 0.32) and 45% reduction in S-PIS (10.51 ± 1.12 vs. 5.79 ± 0.56) as quantified by area under the curve (AUC) analysis (Fig. 1A and B). Remarkably, subsequent addition of 10 nmol/l GLP-1 potentiated S-PIS in Munc13-1^+/− islets to the same extent as Munc13-1^+/+ islets (Fig. 1A). In a second protocol, GLP-1 potentiated F-PIS and S-PIS in Munc13-1^+/+ and Munc13-1^+/− islets to the same extent (Fig. 1C).

To more rigorously determine which phase(s) of insulin secretion GLP-1 was rescued, we used a well-established protocol (31) of depolarizing concentrations of KCl (30 mmol/l) plus K_ATP-channel opener diazoxide (250 μmol/l) in the presence of low glucose (2.8 mmol/l) to isolate F-PIS (triggering phase) or in the presence of high-concentration glucose (16.7 mmol/l) to isolate S-PIS (amplifying phase). In low glucose conditions (KCl plus diazoxide), Munc13-1^+/− islets displayed a markedly lower level of F-PIS compared with Munc13-1^+/+ islets (Fig. 1D), culminating in 38% reduction (13.30 ± 1.16 vs. 8.93 ± 1.08) as quantified by AUC analysis (Fig. 1E). In high glucose conditions (KCl plus diazoxide), Munc13-1^+/− islets displayed a lower level of S-PIS compared with Munc13-1^+/+ islets (Fig. 1D), culminating in 31% reduction (31.56 ± 2.66 vs. 21.77 ± 1.37) (Fig. 1E). GLP-1 was therefore able to rescue both F-PIS and S-PIS in Munc13-1^+/− islets (Fig. 1A and C). The relatively greater reduction in S-PIS in glucose alone (Fig. 1B) versus S-PIS in KCl plus diazoxide and high glucose (Fig. 1E) in Munc13-1^+/− islets indicates a permissive role of the triggering phase in the amplifying phase in GSIS (31). Of note, this normalization of insulin secretory defects in Munc13-1^+/+ islets is not due to any alteration of islet cAMP accumulation, which was similar between Munc13-1^+/+ and Munc13-1^+/− islets in either GSIS or GLP-1 stimulation conditions (Fig. 1F).

We examined the specific pool of insulin vesicles acted on by GLP-1 stimulation to normalize biphasic insulin secretion in Munc13-1^+/− islets (Fig. 1) by C_m measurements of single β-cells. C_m changes elicited by the first two pulses approximate the number of vesicles released from and therefore the size of the RRP of primed fusion-ready vesicles. Subsequent pulses estimate the rate of refilling or mobilization of insulin vesicles from reserve pool(s) to RRP. The size of RRP and rate of refilling of a releasable pool of vesicles have been shown to correlate favorably with F-PIS and S-PIS from whole pancreatic islets (1).

Using this strategy of C_m measurements of single β-cells, we observed that depolarization-evoked insulin exocytosis was severely reduced in Munc13-1^+/− compared with Munc13-1^+/+ β-cells (Fig. 2A–C), consistent with our previous report (11). The size of RRP of vesicles (ΣΔCm_1st-2ndpulse) was reduced by 52% in Munc13-1^+/− β-cells (2.79 ± 0.15 vs. 5.77 ± 0.87 femto Farad per pico Farad [fF/pF] in Munc13-1^+/+ β-cells; Fig. 2D). Rate of refilling of a releasable pool of vesicles (ΣΔCm_3rd-8thpulse) was reduced by 54% in Munc13-1^+/− β-cells (3.82 ± 0.72 vs. 8.36 ± 1.04 fF/pF in Munc13-1^+/+ β-cells; Fig. 2D). No changes was observed in whole-cell currents evoked during the depolarization pulses, suggesting that the interactions between the β-cell K_v channels and exocytotic SNARE proteins were unaltered in the Munc13-1^+/− β-cells (11). Like GLP-1, intracellular dialysis of 100 μmol/l cAMP to maximally activate cAMP signaling (29) greatly enhanced insulin exocytosis in both Munc13-1^+/+ and Munc13-1^+/− β-cells (Fig. 3A–C). Remarkably, increase in RRP size (ΣΔCm_1st-2ndpulse) by cAMP activation in Munc13-1^+/− β-cells (11.60 ± 1.88 fF/pF) was very similar to Munc13-1^+/+ β-cells (13.96 ± 1.84 fF/pF) (Fig. 3D). The increased rate of refilling of a releasable pool of vesicles (ΣΔCm_3rd-8thpulse) by cAMP activation in Munc13-1^+/− β-cells (31.64 ± 6.02 fF/pF) was only mildly reduced by 28% (P = 0.11) compared with Munc13-1^+/+ β-cells (44.21 ± 4.67 fF/pF) (Fig. 3D). Similarly, the normalized exocytosis in Munc13-1^+/− β-cells is not due to altered whole-cell currents as explained above. The full rescue of RRP in Munc13-1^+/− β-cells by maximal cAMP stimulation is commensurate to GLP-1–potentiated F-PIS in the Munc13-1^+/− islet perifusion study (Fig. 1). However, the less-effective rescue of refilling of a releasable pool of vesicles by cAMP in Munc13-1^+/− β-cells is in contrast to the normalized S-PIS.

Of note, depolarization protocols have some inherent limitations. First, Ca²⁺ channels induced to open by the depolarization protocols might be affected by either the Munc13-1 deficiency or the cAMP rescue. Second, Munc13-1 deficiency might alter the coupling of insulin vesicles to Ca²⁺ channels or influence Ca²⁺-induced Ca²⁺ release. Any of these secretory components could have been acted on by cAMP to rescue exocytosis (Fig. 2) in a manner independent of the residual Munc13-1. To negate the contributions from these confounding possibilities, we “bypassed” the plasma membrane Ca²⁺ channel by infusing into β-cells a constant stimulatory concentration of Ca²⁺ (1.5 μmol/l) to induce Ca²⁺-evoked exocytosis (Fig. 3E). Analysis of the rate of change of C_m (cAMP free) at a 60- to 120-s period, which approximates the refilling rate of a releasable pool of vesicles, showed a reduction of 41% in Munc13-1^+/− β-cells (3.32 ± 0.22 vs. 5.67 ± 0.51 fF · s⁻¹ in Munc13-1^+/+ β-cells; Fig. 3F). This magnitude of reduction is similar to that observed with the serial depolarization (54%) study (Fig. 3A–D). However, when we calculated the rate of change of C_m during the first 30 s, which approximates the RRP size, we saw a reduction of 27% in Munc13-1^+/− β-cells (6.70 ± 0.53 vs. 9.24 ± 0.85 fF · s⁻¹ in Munc13-1^+/+ β-cells; Fig. 3F). This reduction is milder than that seen when exocytosis was induced by serial depolarization (52%) (Fig. 3A–D), suggesting a reduction in Ca²⁺ sensitivity. These results indicate that while Munc13-1 deficiency caused exocytotic defects by reducing the initial size and refilling of the RRP, the Munc13-1 deficiency also seemed to have caused some reduction in the sensitivity of exocytosis to Ca²⁺, the latter also contributing to the priming defects. Remarkably, in presence of cAMP, C_m increase in Munc13-1^+/− β-cells was completely normalized to that in Munc13-1^+/+ β-cells (14.84 ± 0.89 vs. 14.95 ± 0.80 fF · s⁻¹ in Munc13-1^+/+ β-cells; Fig. 3E and F) during the first 30 s and was only mildly reduced by 22% in Munc13-1^+/− β-cells (10.30 ± 0.88 vs. 13.27 ± 1.49 fF · s⁻¹ in Munc13-1^+/+ β-cells; Fig. 3E and F) during the 60- to 120-s period, indicating that cAMP rescues the exocytotic defects caused by the Munc13-1 deficiency.

cAMP/GLP-1–mediated potentiation of insulin secretion has been attributed to its actions on two major cAMP receptors, cAMP/Epac and cAMP/PKA (18). We hypothesized that these two cAMP receptors play a distinct role in rescuing the insulin exocytotic defects in Munc13-1^+/− β-cells. We measured islet expressions of Epac2 and Rim2 (a major PKA substrate), which were identical between Munc13-1^+/− and Munc13-1^+/+ mice (Fig. 4A). These cAMP receptors are therefore equally intact in both Munc13-1^+/+ and Munc13-1^+/− β-cells for cAMP to exert its full actions.

We proceeded to examine the independent actions of Epac signaling. We performed intracellular dialysis of 10 μmol/l 8-pCPT-2′-O-Me-cAMP, a selective agonist for Epac but not PKA (17). Although 8-pCPT-2′-O-Me-cAMP enhanced depolarization-evoked exocytosis (cAMP-free) in both groups of β-cells (Fig. 4B and C), Epac-potentiated exocytosis was less than that potentiated by maximal cAMP stimulation (Fig. 3A–D), in keeping with previous reports (28,29). More importantly, absolute (Fig. 4B) and incremental (Fig. 4C) increases in C_m by serial depolarization were lower in every single pulse in Munc13-1^+/− β-cells compared with Munc13-1^+/+ β-cells, culminating in 43% reduction in size of RRP (ΣΔCm_1st-2ndpulse) in Munc13-1^+/− β-cells (5.48 ± 0.50 vs. 9.68 ± 0.97 fF/pF in Munc13-1^+/+ β-cells) and a remarkably similar reduction of 41% in rate of refilling of a releasable pool of vesicles (ΣΔCm_3rd-8thpulse, 9.81 ± 1.35 vs. 16.51 ± 1.47 fF/pF in Munc13-1^+/+ β-cells; Fig. 4D). These results suggest that although Epac signaling pathway is functional and contributes to cAMP-mediated rescue of exocytic defects in Munc13-1^+/− β-cells, Epac signaling alone is insufficient in fully restoring the insulin exocytotic defects caused by Munc13-1 deficiency, implying that Munc13-1 abundance is required for full potentiating effects of Epac activation. This led us to postulate that PKA signaling would also contribute to cAMP-induced rescue of exocytosis caused by Munc13-1 deficiency.

To examine the independent contribution of PKA signaling in rescuing exocytosis in Munc13-1^+/− β-cells, we blocked Epac signaling by dialyzing into β-cells 1 μmol/l GST-Rim2PDZ peptide while simultaneously including 100 μmol/l cAMP in the pipette solution to activate the PKA signaling pathway. GST-Rim2PDZ directly interacts with Epac2 in a manner that abolished Epac2-Rim2 binding, which is essential for Epac to potentiate exocytosis (18). C_m changes elicited by the first four depolarization pulses were similar between Munc13-1^+/− and Munc13-1^+/+ β-cells (Fig. 5A and B), including similar sizes of RRP (ΣΔCm_1st-2ndpulse) between Munc13-1^+/− (9.67 ± 1.02 fF/pF) and Munc13-1^+/+ (11.89 ± 1.26 fF/pF) β-cells (Fig. 5C), indicating little effect by Epac blockade on RRP size. C_m changes elicited by subsequent four pulses were lower in Munc13-1^+/− β-cells (Fig. 5A and B), such that overall refilling of a releasable pool of vesicles encompassing the third to eighth pulse (ΣΔCm_3rd-8thpulse) remained impaired by 30% in Munc13-1^+/− β-cells (19.16 ± 2.74 vs. 27.22 ± 2.24 fF/pF in Munc13-1^+/+ β-cells; Fig. 5C). These patterns are rather similar to those effected by maximal cAMP activation (Fig. 3). As control for Rim2PDZ, 1 μmol/l GST-Rim2PDZ(AAA), a mutant that cannot bind Epac2 (28), did not affect cAMP-mediated potentiation of exocytosis (online appendix Supplemental Fig. 1C [available at http://dx.doi.org/10.2337/db06-1207]). To confirm that the inhibitory effects of GST-Rim2PDZ were attributable to blockade of Epac signaling, we showed that GST-Rim2PDZ completely inhibited the potentiating effects of 8-pCPT-2′-O-Me-cAMP (Fig. 5D–F).

Because the PKA signaling pathway was discriminated only indirectly by inhibiting Epac signaling in Fig. 5, we examined direct PKA activation by dialyzing 100 μmol/l N⁶-Bnz-cAMP, an agonist selective for PKA but not Epac (34). N⁶-Bnz-cAMP stimulation increased the size of RRP (ΣΔCm_1st-2ndpulse) in Munc13-1^+/− β-cells (7.18 ± 0.51 fF/pF) to a similar extent as Munc13-1^+/+ β-cells (8.20 ± 0.72 fF/pF) (Fig. 6A–C). Refilling of a releasable pool of vesicles (ΣΔCm_3rd-8thpulse) remained mildly but not significantly impaired (∼12%) in Munc13-1^+/− β-cells (14.52 ± 1.08 vs. 16.57 ± 1.18 fF/pF in Munc13-1^+/+ β-cells; Fig. 6A–C). These results suggest that PKA activation alone has a tendency to overcome most of the exocytotic defects caused by Munc13-1 deficiency. Nonetheless, maximal PKA-potentiated exocytosis by N⁶-Bnz-cAMP was less than maximal cAMP potentiation (Fig. 6C), consistent with previous reports (28). When these results are further compared with 8-pCPT-2′-O-Me-cAMP (from Fig. 4), Epac signaling accounts for some of the cAMP rescue of the exocytotic defects in Munc13-1^+/− β-cells but seems to contribute somewhat less to the cAMP rescue than PKA signaling (Fig. 6D). Nonetheless, it also appears that synergistic actions of both Epac and PKA signaling are required for full cAMP rescue.

If PKA activation tends to account for much of the cAMP-mediated rescue of exocytotic defects in Munc13-1^+/− β-cells, then blockade of PKA signaling should block these potentiating actions of maximal cAMP stimulation. Here, we have used the potent and relatively selective PKA inhibitor H89 (3 μmol/l) plus 100 μmol/l cAMP, which would leave Epac2 signaling intact. In presence of H89, cAMP failed to fully amplify insulin exocytosis in both groups of β-cells (Fig. 7A and B). C_m changes were much lower in Munc13-1^+/− β-cells in a majority of depolarization pulses (Fig. 7A and B). RRP (ΣΔCm_1st-2ndpulse) was reduced by 43% in Munc13-1^+/− β-cells (3.21 ± 0.70 vs. 5.66 ± 1.27 fF/pF in Munc13-1^+/+ β-cells; Fig. 7C). Refilling of a releasable pool of vesicles (ΣΔCm_3rd-8thpulse) was similarly impaired by 42% in Munc13-1^+/− β-cells (10.40 ± 0.81 vs. 17.99 ± 2.14 fF/pF in Munc13-1^+/+ β-cells; Fig. 7C). These results confirm that PKA signaling is a major contributor to cAMP potentiation and rescue of the exocytotic defects in Munc13-1^+/− β-cells.

Rim2 has been suggested to mediate PKA-dependent potentiation of insulin secretion (18). Rim isoform requires PKA-dependent phosphorylation for activation (21) and is believed to interact with Munc13-1 to enhance the exocytic action of Munc13-1 (24), but the precise mechanism remains unclear. We hypothesized that functional interaction of Munc13-1 and Rim was critical for PKA signaling to overcome the Munc13-1 deficiency. We blocked endogenous Rim2–Munc13-1 interaction in β-cells by dialysis of 1 μmol/l GST-Munc13-1–Rim-binding domain (RBD) peptide (35). Here, cAMP rescue of RRP size and refilling of a releasable pool of vesicles in Munc13-1^+/− β-cells were greatly reduced (Fig. 8A and B), with RRP (ΣΔCm_1st-2ndpulse) reduced by 54% in Munc13-1^+/− β-cells (4.46 ± 0.74 vs. 9.65 ± 1.63 fF/pF in Munc13-1^+/+ β-cells; Fig. 8C) and refilling (ΣΔCm_3rd-8thpulse) impaired by 37% in Munc13-1^+/− β-cells (12.56 ± 0.57 vs. 19.95 ± 1.93 fF/pF in Munc13-1^+/+ β-cells; Fig. 8C). These results strongly support a putative role of Munc13-1–Rim interaction in cAMP-mediated rescue of Munc13-1 exocytotic defects.

Biphasic insulin secretion is attributed to exocytosis of distinct pools of insulin vesicles, which become severely perturbed in type 2 diabetes (3). GLP-1 mimetics, acting on cAMP- and PKA-signaling pathways, have emerged as promising therapeutic agents in treating type 2 diabetes (4), by normalizing insulin exocytotic defects. However, the molecular substrates and corresponding steps in exocytosis (vesicle priming and refilling) acted on by GLP-1 remain unclear. In this work, we have gained much insight into the functional interactions of Munc13-1, Epac, and PKA (including Rim), which compose the priming machinery for vesicle priming and refilling steps of exocytosis. We have used Munc13-1–haplodeficient mouse islets, which exhibit defective biphasic insulin secretion mimicking type 2 diabetes and whose β-cells exhibit major defects in priming and refilling steps in exocytosis. We examined how these defects caused by Munc13-1 deficiency are selectively rescued by agents acting on distinct cAMP and PKA signaling pathways. Below, we discuss our major findings.

First, we characterized the kinetics of insulin secretion and exocytosis of adult haplodeficient Munc13-1^+/− islets. Islet perifusion studies showed that Munc13-1^+/− islets had severe reduction in F-PIS and S-PIS. Consistently, C_m measurements revealed severe impairment in the size of RRP and rate of refilling of a releasable pool of vesicles. Using homozygous Munc13-1^−/− knockout mouse islets, Kang et al. (12) showed nearly abolished S-PIS and severely reduced F-PIS. Their C_m measurements showed Munc13-1^−/− β-cells to exhibit severe reduction in rate of refilling of a releasable pool of vesicles, but surprisingly, the size of RRP was normal. Of note, in the study by Kang et al., β-cells examined were from perinatal mice because homozygous Munc13-1^−/− mice die shortly after birth (7). The size of RRP of insulin vesicles in Munc13-1^−/− mice could be larger initially and may then undergo normal progressive reduction during development. Alternatively, progressive reduction of RRP and F-PIS could have been induced by the diabetic state because adult Munc13^+/− mice are glucose intolerant (11). A more prolonged period of Munc13-1 deficiency in adult Munc13^+/− mice could lead to reduced tightness of coupling between Ca²⁺ channels and insulin vesicles, resulting in less efficient excitosome complex to effect exocytosis (36).

Second, GLP-1 stimulation normalized biphasic insulin secretory defects caused by Munc13-1 deficiency. Consistently, examination of kinetics of exocytosis showed that maximal cAMP stimulation completely restored the size of RRP and almost completely recovered the rate of refilling of a releasable pool of vesicles. The restoration was not due to alteration in Ca²⁺ entry through voltage-gated Ca²⁺ channels, which we previously showed to be normal (11), but was in part due to rescue of impaired sensitivity to a Ca²⁺ trigger as demonstrated in the Ca²⁺ infusion experiments. The rescue of refilling of a releasable pool of vesicles was somewhat less efficient than the complete rescue of S-PIS because serial depolarization is “supraphysiological” compared with physiological GSIS. Nonetheless, such an acute supraphysiological stimulation may simulate the prolonged and high demand on pancreatic islets imposed by diabetes. Islet levels cAMP receptors Epac and PKA substrate Rim2, as well as cAMP production, were normal in Munc13^+/− mice, indicating that GLP-1 could activate these cAMP receptors to bypass Munc13-1 deficiency or interact with residual Munc13-1 proteins to promote insulin exocytosis. Below, we discussed the independent contributions of Epac and PKA signaling.

Third, Epac activation partially normalized the defects in priming and refilling of a releasable pool of vesicles caused by Munc13-1 deficiency. Epac2 and Rim2 interaction is essential for GLP-1–enhanced insulin secretion (18). Epac2/Rim2 complex was postulated to serve as a scaffold to bind and orchestrate additional exocytic proteins, including not only Munc13-1 (24,27) and SNARE proteins but also other proteins involved in exocytosis, including Piccolo (Ca²⁺ sensor of this complex), l-type Ca²⁺ channel α-subunit, and nucleotide-binding fold of sulfonylurea receptor 1 (SUR1) (18). Epac activation through SUR1 inhibits K_ATP channels (30) and enhances Cl⁻ influx from insulin vesicles (37), leading to enhanced insulin exocytosis, and was abolished in SUR1-null mice (29). Ca²⁺-induced Ca²⁺ release could be amplified by Epac activation (17). Epac2 therefore integrates ATP, cAMP, and Ca²⁺ signals in pancreatic β-cells to facilitate stimulus-exocytosis coupling. Our results showed that Epac signaling contributes to but is insufficient to account for the full cAMP rescue of exocytic defects in Munc13-1^+/− β-cells.

Fourth, PKA activation is a major contributor to cAMP rescue of exocytic defects caused by Munc13-1 deficiency and is in part mediated via Rim2–Munc13-1 interaction. We show that PKA activation fully amplified the size of RRP of insulin vesicles to the same extent as maximal cAMP stimulation and partially accelerated refilling of a releasable pool of vesicles. PKA-catalyzed protein phosphorylation is known to potentiate insulin secretion, but the precise molecular substrates remain largely unknown (18). Rim proteins are major phosphorylation targets of PKA and can directly bind Munc13-1 (7,21). We here demonstrated that Rim2–Munc13-1 interaction was critical for the rescue by PKA. Rim binds Munc13-1 even when it is not PKA-phosphorylated, and disruption of Rim-Munc13-1 binding reduces fusion-competent synaptic vesicles mimicking Munc13-1 knockout neurons (25). Although the NH₂-terminal Rim-binding l-region of Munc13-1 does not exert a priming reaction per se (25,38), this binding is thought to influence indirectly the priming activity of Munc13-1 by promoting recruitment of Munc13-1 or causing conformational changes in Munc13-1 that affect its priming activity (27). These observations suggest that decrease in number of primed insulin vesicles in Munc13-1 β-cells is contributed by reduction in Rim-Munc13-1 binding, thus indirectly impairing the priming activity of Munc13-1. Following this reasoning, improvement in priming and refilling of a releasable pool of vesicles in Munc13-1^+/− β-cells by PKA activation is likely due to more effective binding of PKA-phosphorylated Rim to Munc13-1, which would enhance the priming reaction of Munc13-1 (21). Conversely, increase in Rim-Munc13-1 binding could be facilitated by PKA-phosphorylated Munc13-1. In fact, it was recently shown that surfaces of Munc13-1 responsible for heterodimerization with Rim contain several putative phosphorylation sites (39). Although Rim-Munc13-1 binding is important for maintenance of normal RRP, it may not be the case for recovery of vesicles (26), implying that upregulation of Rim-Munc13-1 interactions in Munc13-1^+/− β-cells may not be sufficient to fully account for the rescue of the exocytotic defects by PKA activation. PKA potentiation of exocytosis is contributed by phosphorylation of other exocytotic proteins, including SNAP25, α1.2-subunit of voltage-dependent Ca²⁺ channels, K_ATP channel (18), α-SNAP (40), synapsin I (41), snapin (42), cysteine string protein (43), tomosyn (44), and syntaphilin (45), the majority of which are also involved in insulin secretion (46–48). Although few studies have attempted to explain the precise functional consequences of PKA-catalyzed phosphorylation of these substrates on insulin and neurotransmitter release (49,50), much is still unknown and remains to be further studied, including the possibility that they could functionally interact with Munc13-1.

E.P.K. has received graduate studentship awards from the Canada Graduate Scholarships Doctoral Awards from the Canadian Institutes of Health Research (CIHR) and support from the University of Toronto Banting and Best Diabetes Center (BBDC) and Unilever-Lipton Fellowship in Neuroscience. H.Y.G. has received grants from the CIHR (grant MOP-64465 and equipment grant MMA-66939) and the University of Toronto BBDC.