SNAREing Voltage-Gated K (cid:1) and ATP-Sensitive K (cid:1) Channels: Tuning (cid:2) -Cell Excitability with Syntaxin-1A and Other Exocytotic Proteins

The three SNARE (soluble N -ethylmaleimide-sensitive factor attachment protein receptor) proteins, syntaxin, SNAP25 (synaptosome-associated protein of 25 kDa), and synaptobre- vin, constitute the minimal machinery for exocytosis in secretory cells such as neurons and neuroendocrine cells by forming a series of complexes prior to and during vesicle fusion. It was subsequently found that these SNARE proteins not only participate in vesicle fusion, but also tether with voltage-dependent Ca 2 (cid:1) channels to form an excitosome that precisely regulates calcium entry at the site of exocytosis. In pancreatic islet (cid:2) -cells, ATP-sensitive K (cid:1) (K ATP ) channel clo- sure by high ATP concentration leads to membrane depolarization, voltage-dependent Ca 2 (cid:1) channel opening, and insulin secretion, whereas subsequent opening of voltage-gated K (cid:1) (Kv) channels repolarizes the cell to terminate exocytosis. We have obtained evidence that syntaxin-1A physically interacts with Kv2.1 (the predominant Kv in (cid:2) -cells) and the sulfonylurea receptor subunit of (cid:2) -cell K ATP channel to modify their gating behaviors. A model has proposed that the conformational changes of syntaxin-1A during exocytosis induce dis- tinct functional modulations of K ATP and Kv2.1 channels in a manner that optimally regulates cell excitability and insulin secretion. Other proteins involved in exocytosis, such as Munc-13, tomosyn, rab3a-interacting molecule, and guanyl nucleotide exchange factor II, have also been implicated in direct or indirect regulation of (cid:2) -cell ion channel activities and excitability. This review discusses this interesting aspect that exocytotic proteins not only promote secretion per se , but also fine-tune (cid:2) -cell excitability via modulation of ion channel gating. ( Endocrine Reviews 28: 653–663, 2007)

A process termed "priming" precedes the formation of the trans-SNARE complex. Before priming, Syn-1A exists in the closed conformation whereby its N-terminal H ABC domain folds back onto its C-terminal H3 domain (2, 9 -12). This blocks the active sites (SNARE motif) within H3 domain from interacting with SNARE motifs within SNAP-25 and synaptobrevin to form the SNARE complex. Activation of Syn-1A from the closed conformation to the open form is therefore a key event required to enable SNARE complex assembly (therefore, priming), and Munc13 and Munc18 have been implicated in the Syn-1A conformation changes (see Refs. 2 and 9 -12; also see Section V).
In islet ␤-cells, opening of voltage-dependent Ca 2ϩ channels (VDCCs) provides the Ca 2ϩ influx required to trigger exocytotic fusion of insulin granules (13). VDCC is a complex of several subunits, namely ␣ 1 , ␣ 2 , ␦, ␤, and in some VDCC members a ␥-subunit (for a review, see Ref. 14). The ␣ 1 -subunit forms the ion-conducting channel pore, whereas other subunits may serve to enhance channel expression and/or modify channel gating. The ␣ 1 -subunit is a single polypeptide consisting of four homologous repeats (I-IV) clustering around the center pore, with each repeat having six transmembrane helices (S1-S6). S4 is the voltage sensor for channel activation, whereas the P-loop is between S5 and S6; the P-loops from the four repeats are believed to constitute the Ca 2ϩ ion selectivity filter (14).
SNARE proteins were subsequently shown to be tethered to the cytoplasmic domain between the second and third repeats of various VDCCs, including VDCCs found in islet ␤-cells (15)(16)(17)(18). There are two obvious strategic advantages of such a molecular arrangement. First, it ensures rapid release response because the secretory machinery (SNARE complex) is immediately exposed to a high local Ca 2ϩ concentration permeating through the VDCCs to trigger membrane fusion (15)(16)(17). Second, the SNARE proteins can, depending on the status of the exocytotic cycle, provide feedback to the VDCC to control the rate and amount of Ca 2ϩ influx (15-17, 19 -22). Such an intimate physical and functional coupling among the secretory vesicle-SNARE protein-Ca 2ϩ channel complex has been termed excitosome in ␤-cells (16) and neurons (19).
VDCC gating is finely regulated by changes in membrane potential. In pancreatic ␤-cells, membrane potential is modulated mainly by ATP-sensitive K ϩ (K ATP ) and voltage-gated K ϩ (Kv) channels (13,23,24). When pancreatic islet ␤-cells are exposed to high glucose, the raised ATP/ADP ratio in the cytosol causes closure of K ATP channels. Cessation of K ϩ efflux due to shutting down of K ATP channels, together with a continuous depolarizing current of yet unidentified nature, results in ␤-cell membrane depolarization. The latter activates VDCC, leading to Ca 2ϩ influx and ensuing exocytosis (13,23). As a result of ␤-cell membrane depolarization, Kv channels also become activated, providing K ϩ efflux necessary to repolarize the cell membrane. Subsequently, repolarization closes VDCC, leading to cessation of Ca 2ϩ influx and termination of exocytosis (24).
It would be of interest to ask whether and how these K ϩ channels (Kv, K ATP ) in the ␤-cell could be intimately coupled to the exocytotic cycle. Toward this, substantial insights have been obtained over the last several years that demonstrated that SNARE proteins, mainly Syn-1A, physically and functionally interact with ␤-cell Kv2.1 (the predominant Kv in ␤-cells) and K ATP channels (25-31). These reports support the notion that SNARE proteins do not only participate in exocytosis per se, but also regulate excitability by modulating the ion channels involved in secretion. It is of particular importance to note that SNARE protein levels are severely reduced in islets of diabetic rodents and humans (32-35), which would be expected to contribute to the dysregulation of these ␤-cell secretory steps in diabetes. Remarkably, restoration of these deficient SNARE proteins in islets of type 2 diabetic rodent models could normalize insulin secretion (32, 36), which has therapeutic implications.
In this review, we explain how Syn-1A modulates the gating of Kv2.1, VDCC, and K ATP channels to fine-tune ␤-cell excitability. Munc-13, tomosyn, rab3a-interacting molecule (RIM) and guanyl nucleotide exchange factor II (GEFII) are also known to modulate vesicle fusion per se, but they could also be involved in direct or indirect regulation of ␤-cell ion channel activities and excitability, which we also discuss in this review.

A. Syn-1A inhibits Kv currents and modulates Kv channel gating
A Kv channel is composed of ␣and ␤-subunits. Although the ␣-subunit is the conducting pore, the ␤-subunit is a modifier of channel gating (37). The Kv ␣-subunit comprises four polypeptides clustering around a central pore. The T1 region is the domain in the N terminus responsible for tetramerization of the polypeptides (37). Each polypeptide subunit has six transmembrane helices (S1-S6). Similar to VDCC, S4 is the voltage sensor, whereas the P-loop is between S5 and S6; the P-loops from the four polypeptide subunits form the K ϩ selectivity filter (38). In all K ϩ channels, there is a wellconserved "sequence signature," TVGYG, which lines the selectivity filter (38).
Kv1.1, a Kv channel abundant in neurons, was the first Kv channel shown to interact with Syn-1A (25). Physical interaction between Kv1.1 and Syn-1A was shown in rat brain synaptosomes using reciprocal coimmunoprecipitation. Kv1.1 gating (existing as Kv1.1/Kv␤ channel complex) was shown to be modulated by Syn-1A, which caused an increase in the extent of inactivation, possibly by binding to the ␤-subunit and increasing its efficacy in causing inactivation. Syn-1A decreases Kv1.1 current magnitude by binding to the N terminus of Kv1.1 (26). The expression of Kv1.1 is, however, very low in insulinoma HIT-T15 cells and undetectable in rat islet ␤-cells, arguing against an involvement of Kv1.1 in ␤-cell excitability regulation (39). It was further found that Syn-1A and SNAP-25 modulated Kv channels in cardiac myocytes (40) and esophageal smooth muscles (41), respectively. These reports establish a new paradigm that SNARE proteins can directly modulate the Kv channel superfamily.
In pancreatic islet ␤-cells, Kv2.1 is the major (70%) delayed rectifier K ϩ channel (39, 42). Syn-1A and SNAP25 colocalized with both Kv2.1 and Ca v 1.2 (L-type VDCC) in lipid raft microdomains within HIT-T15 ␤-cell plasma membranes, and these lipid raft domains contain other components of the SNARE exocytotic machinery (43). This raised the possibility that SNARE proteins could be forming additional excitosomes with Kv channels (43). In support, coimmunoprecipitation experiments on a heterologous expression system demonstrated a molecular complex consisting of Syn-1A, SNAP25, and Kv2.1 (44). In fact, in neuroendocrine PC12 cell line, it was just reported that direct association of Syn-1A to Kv2.1 promoted exocytosis (45). It would be of interest to examine whether such a Kv2.1-containing excitosome also occurs in islet ␤-cells. It has been reported that SNAP-25 binds to the N terminus of Kv2.1 to inhibit current amplitude (44). This role of SNAP-25 on Kv2.1 was confirmed in ␤-cells whereby SNAP-25 was cleaved by botulinum neurotoxins (46).
When the structure-functional interaction of Kv2.1 with Syn-1A was examined, Syn-1A was shown to bind strongly to the cytoplasmic C-terminal fragments (C1, amino acids 412-633; C2, amino acids 634 -853), and only weakly to the cytoplasmic N terminus (amino acids 1-182), of Kv2.1 (27). Dialysis of recombinant Syn-1A into ␤-cells or HEK cells expressing Kv2.1 caused inhibition of currents, which could be prevented by codialysis of C1 or C2 fragments, but not N-terminus protein (27). This would suggest that Syn-1A inhibition of Kv2.1 is mediated through the C terminus, which is in contrast to Kv1.1, where Syn-1A inhibition is mediated via the channel N terminus (26). This demonstrates that SNARE proteins interact with distinct domains within the different Kv members, which is in stark contrast to the VDCC superfamily, where SNARE proteins interact with a common conserved cytoplasmic domain linking transmembrane repeats II and III, termed the synprint site, which is present in all high voltage-activated VDCC members (15)(16)(17).
Syn-1A not only reduces Kv2.1 current amplitude, but also changes its gating behavior (27). Coexpressing Syn-1A with Kv2.1 in HEK cells increases the voltage sensitivity of steadystate inactivation of Kv2.1 over a range of physiologically relevant membrane potential (Ϫ40 to 0 mV), suggesting that Syn-1A decreases channel availability upon steady-state depolarization. Furthermore, in the presence of Syn-1A, Kv2.1 channel activation becomes slower (27). This is, however, not accompanied by a significant change in voltage-dependence of activation.
Pharmacological blockade of Kv2.1 can effectively potentiate glucose-stimulated insulin secretion in ␤-cells, suggesting that Kv2.1 is a potential drug target in the treatment of diabetes (24, 47). Insights into SNARE-Kv2.1 interaction may therefore reveal potential novel strategies to optimize insulin secretion.

B. Syn-1A inhibition of Kv2.1 depends on the conformation of Syn-1A
More recently, efforts have been made to examine whether the conformational changes of Syn-1A from closed form to open form occurring during exocytosis would have distinct actions on Kv2.1 gating (31), perhaps in a manner that would dynamically modulate the rate of membrane repolarization conducive to optimizing exocytosis. To examine the specific actions of the activated open form Syn-1A, the mutant Syn-1A, which is constitutively open by the introduction of two mutations at the linker region (L165A, E166A) (9, 12), was employed. A note of caution is that in previous reports showing Syn-1A regulation of Kv channels and VDCC, the form of Syn-1A used was the wild type (WT) (20,(25)(26)(27)(28) Open form Syn-1A also reduced Kv2.1 channel availability to a larger extent than WT Syn-1A upon steady-state depolarization (31). Because WT Syn-1A "flip-flops" between the closed and open forms (48), it is reasonable to speculate that closed form Syn-1A inhibits Kv2.1 less than WT Syn-1A. Unfortunately, there is so far no constitutively closed form Syn-1A reported to verify this point.

C. Reciprocal regulation of Kv2.1 and VDCC by closed/open forms of Syn-1A: implications
L-type VDCC has been accepted to be the predominant VDCC involved in both the first and second phase of glucosestimulated insulin secretion in human, rat, and mouse islets and insulin-secreting ␤-cell lines (16,49,50). It mediates Ca 2ϩ entry required for triggering insulin exocytosis in K ATP channel-dependent first phase insulin secretion. It also regulates intracellular Ca 2ϩ signaling to replenish releasable insulin granules during K ATP channel-independent second phase insulin secretion. Interaction of VDCC with SNARE proteins has been well documented (19). Specifically, Syn-1A has been shown to modulate L-type VDCC activities in pancreatic ␤-cells and ␤-cell lines (18,20).
In a species-specific fashion, N-type VDCC has also been implicated in ␤-cell secretion (50). For example, N-type VDCC has been found in INS-1 rat insulinoma cells and rat ␤-cells; and inhibition of N-type VDCC with -conotoxin blocks glucose-stimulated insulin release (51-55). N-type VDCC inhibition by -conotoxin also blocks human ␤-cell hormonal secretion (52). Zamponi's group (22) has shown that WT and open forms of Syn-1A have differential effects on N-type VDCC gating. Thus, it was demonstrated that whereas both WT and open form Syn-1A equally support G protein inhibition of N-type VDCC, the availability of N-type VDCC was only decreased by WT Syn-1A, but not by open form Syn-1A. This is in contrast to the stronger inhibition of Kv2.1 by open form Syn-1A than WT Syn-1A (31).
What is the biological implication of the above-described reciprocal modulation of VDCC and Kv channels by Syn-1A conformational changes? We hypothesize that such reciprocal modulation optimally regulates cell excitability and thus exocytosis (Fig. 1). When insulin granules are docked onto the plasma membrane and the ␤-cell is at resting state, Syn-1A assumes the closed form (Fig. 1A). Closed form Syn-1A reduces VDCC (N-type or L-type) availability and The open form of Syn-1A strongly inhibits Kv2.1 to limit K ϩ efflux, thus slowing down repolarization and consequently augmenting Ca 2ϩ influx during exocytosis. The SNARE core complex disassembles after exocytosis, and Syn-1A resumes its closed form (Fig. 1D). VDCC availability now becomes much reduced because of interaction with closed form Syn-1A. Ca 2ϩ influx is then substantially attenuated. Meanwhile, the closed form Syn-1A inhibits Kv2.1 only weakly, permitting more K ϩ efflux and thus speeding up repolarization. Eventually, both the VDCC and Kv2.1 channels close, because the ␤-cell repolarizes to its resting membrane potential (Fig. 1A).

III. Syn-1A Interaction with K ATP Channels
A. WT Syn-1A inhibits K ATP channels through SUR1 NBF1 K ATP channels play an important role in coupling cell metabolism and membrane excitation in ␤-cells (13,23). In low glucose, low ATP concentration in ␤-cells allows K ATP channels to open partially, thus keeping ␤-cells from depolarizing. In high glucose, the raised ATP/ADP ratio closes K ATP channels, leading to membrane depolarization, which opens VDCC to allow Ca 2ϩ influx to evoke exocytosis.
A K ATP channel has two dissimilar components, namely, Kir6 [a member of inward rectifier K ϩ (Kir) channels] and sulfonylurea receptor (SUR). Functionally mature K ATP channels in ␤-cells are heterooctamers, composed of four Kir6.2 subunits and four SUR1 subunits (57, 58). The Kir6 subunit is the ion-conducting pore, whereas SUR is the subunit regulating channel gating (for reviews, see Refs. 59 -61). ATP inhibits the K ATP channel by directly binding to Kir6. SUR comprises of an N-terminal TMD0 domain, and the TMD1 and TMD2 domains. For TMD1 and TMD2, each has six transmembrane helices; and there are two cytoplasmic nucleotide-binding folds (NBFs), namely, NBF1 (between TMD1 and TMD2) and NBF2 (at TMD2 C terminus). Convincing evidence has suggested that NBFs are sites for regulation by adenine nucleotides (e.g., MgADP activation), sulfonylurea drugs (e.g., inhibition by glibenclamide), K ATP channel openers (such as activation by diazoxide) (57, 59 -61), and Syn-1A (see below).
Considerable insights into the structure-function of K ATP channels (SUR1/Kir6.2) in controlling ␤-cell insulin secretion have arisen from reports of genetic mutations causing either loss-of-function (defective K ATP channel opening) or gainof-function (persistent K ATP channel opening) of the channel, which, respectively, result in congenital hyperinsulinemia or neonatal diabetes mellitus (reviewed in Ref. 61).
It could be pondered whether auxiliary proteins may modulate K ATP channel gating. Because Syn-1A has been shown to affect Kv channels and VDCC, whether Syn-1A could also affect ␤-cell K ATP channel functions to even more exquisitely modulate membrane excitability during exocytosis was subsequently examined. Thus, it has been shown in insulinoma HIT-T15 cells and rat ␤-cells that dialysis of recombinant Syn-1A inhibited K ATP channel opening in the presence of low cytosolic ATP concentration (29), indicating that Syn-1A could further enhance ␤-cell excitability by closing K ATP channels. In concordance, cleavage of endogenous Syn-1A by  Inhibition of K ATP channels by Syn-1A is of importance because Syn-1A levels are severely reduced in islets of diabetic rodents and humans (32-35). Islet ␤-cell K ATP channel activities of these diabetic rodents and humans may therefore be less inhibited, contributing to the reduced glucose-stimulated insulin secretion (32-35). Insulin secretion in type 2 diabetic patients is characterized by elevated basal release, abrogated glucose-stimulated first phase release, and reduced glucose responsiveness of second phase release (62). Despite much work, the precise mechanisms underlying this discrepant insulin secretory response remain undefined. This could be due to dissimilar exocytotic pathways for basal (constitutive) and glucose-stimulated release, with Syn-1A playing no role or only a very minor role in basal release. Strong evidence for this notion comes from a recent study showing that basal insulin release is unaffected in Syn-1A knockout mice, which nevertheless have severely impaired first-and second-phase insulin secretion (63).
Syn-1A binds SUR1 at its NBF1 and NBF2 with equal affinity (29). However, codialysis of NBF1, but not NBF2, blocked Syn-1A inhibition, suggesting that Syn-1A inhibition of K ATP channels is transduced via NBF1. Syn-1A does not bind the N terminus of SUR1 (Y. Kang (27), it is possible that Syn-1A may also affect the trafficking of K ATP channels, which remains to be examined.

B. Open form Syn-1A inhibits K ATP channels through NBF1 and NBF2
More recently, whether the conformation status of Syn-1A affects its ability to inhibit ␤-cell K ATP channels was investigated (30). Open form Syn-1A inhibits K ATP channels with similar efficacy as WT-Syn-1A; the H3 domain, not the H ABC domain, is responsible for the inhibitory action. Interestingly, open form Syn-1A and H3 domain differ from WT-Syn-1A in their mode of inhibition (30). Codialysis of NBF1 or NBF2 is equally effective in blocking open form Syn-1A (or H3) inhibition of K ATP channels, indicating that open form Syn-1A transduces its inhibition of K ATP channels via both NBF1 and NBF2, in contrast to WT Syn-1A, which seems to act mainly on NBF1. Therefore, the way Syn-1A interacts with SUR1 also depends on Syn-1A conformation, which led us to propose a model in Fig. 2. Because ADP is known to interact with NBF2 (but not NBF1) to activate ␤-cell K ATP channel (64), inhibition of K ATP channels by open form Syn-1A via binding to NBF2 might interfere with ADP action.
C. Syn-1A modulates ␤-cell K ATP channel sensitivity to K ATP channel openers By hyperpolarizing ␤-cell membrane potential, SUR1-selective K ATP channel openers induce "␤-cell rest" and have been developed as a novel therapeutic approach to counter the effects of ␤-cell overstimulation in diabetes (65)(66)(67)(68)(69). The potential therapeutic implication of Syn-1A-SUR1 interaction was most recently demonstrated by studying the effects of Syn-1A on the actions of a potent ␤-cell-selective K ATP channel opener, NNC 55-0462 (70). In this study, WT Syn-1A was found to decrease the potency and efficacy of NNC 55-0462 and its parent compound diazoxide in activating K ATP channel in rat islet ␤-cells. These inhibitory effects of Syn-1A on NNC 55-0462 drug action could be mimicked by the Syn-1A H3 domain (70). In agreement with our results, in Zucker obese diabetic rats whose islets have severely reduced Syn-1A levels (34), SUR1-selective K ATP channel opener NN414 is more effective in attenuating hyperinsulinemia than in their lean controls (71). Thus, the severely reduced islet Syn-1A levels found in diabetes may indeed enhance the therapeutic efficacy of SUR1-selective K ATP channel openers to induce ␤-cell rest.

A. Rab3A-interacting molecule (RIM) and guanyl nucleotide exchange factor (GEF)
Glucagon-like peptide-1 (GLP-1), a potential antidiabetogenic agent, is capable of bypassing the secretory defects of the diabetic islets to correct the insulin secretory response (72,73). These defects include a reduction in islet levels of SNARE proteins (32-34). GLP-1 also inhibited Kv2.1 in ␤-cells to attenuate membrane repolarization (74,75). These GLP-1 effects on Kv2.1 are mediated by dual activation of cAMP/ protein kinase A (PKA) and epidermal growth factor receptors [and subsequent phosphoinositol-3 kinase and protein kinase C signaling] (75). The precise molecular substrates for the cAMP/PKA pathways are, however, unclear but may involve RIM and GEFII.
RIM proteins have a critical role in vesicle priming. RIM1knockout mice exhibited a reduction in the probability of neurotransmitter release (76,77). However, RIM2 (1590 amino acids), which possesses 62% identity to RIM1, is more abundant than RIM1 in islet ␤-cells (78,79). RIM2 contains all of the functional domains of RIM1 and exhibits similar interactions with RIM1-binding proteins, which are also present in islet ␤-cells (78 -80). PKA phosphorylates RIM1 at Ser413 (between zinc finger and PDZ domain) and Ser1548 at the very C terminus (Fig. 3) (81). The very N terminus of RIM1 binds Rab3a, which serves to tether the vesicle to the plasma membrane (82)(83)(84). An adjacent domain (zinc finger domain) binds Munc13-1 to create a link between synaptic vesicle tethering and priming (85, 86) (Fig. 3). A central PDZ domain binds GEFII, a presynaptic substrate directly activated by cAMP (78) (Fig. 3). The roles of RIM2 and GEFII in insulin secretion have been actively explored. Antisense treatment to reduce RIM2 or GEFII expression, or expression of their dominant-negative constructs, inhibited GLP-1 (and cAMP/PKA)-stimulated insulin secretion (78 -80, 87, 88). RIM2 binds islet ␤-cell rab3a (87) and Munc13-1 (a major diacylglycerol receptor) (89) to prime insulin exocytosis. Taken together, RIMs act as scaffolding proteins that serve as platforms for RIM-binding proteins to effect their actions or to form complexes with other proteins (Fig. 3). Thus, GLP-1-activated cAMP/PKA signaling effects on ␤-cell targets, including VDCC, Kv2.1, and K ATP channels, may in part be mediated by independent or sequential actions on RIM2 and GEFII, which could potentially bypass the defective SNARE exocytotic machinery in diabetes mellitus.

B. Do RIM/GEFII regulate ␤-cell ion channels?
RIM1-C 2 A and RIM1-C 2 B bind to ␣1B (pore-forming subunit of N-type VDCC) and ␣1C (pore-forming subunit of L-type VDCC) (90). In support, in giant calyx-type synapse of chick ciliary ganglion, it is shown by immunostaining that RIM and N-type VDCC (Ca v 2.2) strongly covary in expression at the presynaptic transmitter release site, suggesting that they are closely associated (91). However, in immunoprecipitation experiments, Ca v 2.2 does not coprecipitate with RIM or its binding partner Munc-13 (91). These suggest that Ca v 2.2 and RIM do not form a stable molecular complex. Whether RIM affects Ca 2ϩ channel current amplitude or gating is hitherto unknown.
cAMP-GEFII has also been implicated as a signaling pathway downstream of Ca 2ϩ influx (92). In rat insulin-secreting INS-1 cells, Ca v 1.2-mediated glucose-stimulated insulin secretion was activated by both GEFII-selective agonist 8-pCPT-2Ј-O-Me-cAMP and PKA-selective agonist N6-benzoyl-cAMP (their effects are additive), indicating involvement of both signaling pathways (92). In contrast, Ca v 1.3mediated glucose-stimulated insulin secretion was only activated upon PKA activation, with GEFII activation only providing a potentiating role (92). However, there is no evidence showing direct binding of GEFII to VDCC, nor are there data indicating modulation of VDCC gating by cAMP-GEFII.
GEFII has been reported to directly bind to the NBF1 of SUR1, but not SUR1-NBF2 or SUR2A-NBF2 (80,93). Recently, K ATP channel activities in human ␤-cells and rat INS-1 insulin-secreting cells have been shown to be inhibited by cAMP and GEFII-selective agonist 8-pCPT-2Ј-O-Me-cAMP, but not by 2Ј-O-Me-cGMP (cGMP analog) or PKA-selective agonist N6-benzoyl-cAMP (93). These data suggest direct modulation of the ␤-cell K ATP channel by cAMP-GEFII but not PKA-mediated phosphorylation. Intriguingly, it was shown that GEFII dissociates from SUR1 upon an elevation of cAMP concentration (80). This finding may indicate that GEFII on its own has a constitutively activating effect on the ␤-cell K ATP channel. The extent to which cAMP-GEFII-mediated inhibition of K ATP channel activities affects insulin secretion remains to be investigated, but such inhibition presumably causes ␤-cell membrane depolarization, promotes Ca 2ϩ entry, and triggers exocytosis (93). Given our finding that Syn-1A also binds to NBF1 of SUR1 (29, 30), it would be interesting to dissect the interrelation of NBF1-Syn-1A-GEFII binding.
We have also tested whether RIM2/GEFII have any effect on Kv2.1. However, in HEK 293 cells expressing Kv2.1, coexpression of RIM2 and/or GEFII did not cause a significant effect on Kv2.1 channel current amplitude or gating (either in the absence or presence of 100 m intracellular cAMP) (Y. M. Leung and H. Y. Gaisano, unpublished observations).

V. Regulation of Syn-1A Conformational Changes by Munc13 and Tomosyn
Munc13-1 and its orthologs (Unc13 and Dunc13 in Caenorhabditis elegans and Drosophila, respectively) have been recently suggested as priming proteins by virtue of their hypothesized ability in displacing Syn-1A from Munc18-1, which would induce Syn-1A to switch from the closed to open form, which is then capable of interacting with its cognate SNARE partners to form a stable complex (9,11,12). The family of Munc13 isoforms belongs to the superfamily of C2 domain proteins (94,95). Studies showed that overexpression of Munc13-1 increases the size of the readily re- FIG. 3. Interaction between Munc13, RIM, and GEFII and potential modulation of VDCC and K ATP channels, respectively, by RIM and GEFII. RIM is under dual control by PKA (two phosphorylation sites) and GEFII (protein-protein interaction). GEFII is directly activated by cAMP. RIM also interacts with Munc13, which serves multiple roles in exocytosis including the opening up of Syn-1A. GEFII binds to SUR1 NBF1 of the K ATP channel to regulate channel activities. The C terminus of RIM binds loosely to VDCC, and whether such binding regulates VDCC activities is hitherto unknown. For full explanation, see text.  (96,97), whereas neurons deficient of Munc13 or its orthologs exhibit a profound impairment in neurotransmitter release due to a near absence of RRP of synaptic vesicles (98 -100). These results suggest a role for Munc13-1 in priming secretory vesicles for fusion. In fact, recent evidence reveals that Munc13-1-deficient mice exhibited severely reduced first and second phase insulin secretion accounted for by reduced RRP size and mobilization of insulin granules, leading to an abnormal glucose tolerance (101,102). Remarkably, the level of Munc13-1 has been reported to be dramatically reduced in islets from type II diabetes models (33). Of note, the priming action of Munc13-1 is hypothesized to be based on three observations. First, Munc13-1/Unc13 interacts with the H ABC domain of Syn-1A (103,104), presumably preventing the domain from folding back onto its H3 SNARE motif (9). Second, overexpression of open form Syn-1A restores fusion of synaptic vesicles in C. elegans Unc13-deletion mutants (12). Third, the interaction between Munc13-1/Unc13 and Syn-1A is essential for Munc13-1/ Unc13 priming activity (104,105). In light of reports from our group and others that Syn-1A regulates activities and/or availabilities of Kv2.1, K ATP , and N-type VDCC in a conformation-dependent manner, it is tempting to hypothesize that the specific Syn-1A forms required for affecting the ion channels are actively regulated by the priming activity of Munc13-1.
It is noteworthy that there is no evidence so far that Munc13-1 directly interacts with any ion channel (at least with regard to exocytosis triggered by Ca 2ϩ alone), based on several observations. First, Kv channel and VDCC activities are normal in pancreatic ␤-cells from heterozygous Munc13-1 knockout mice (101). Second, maximal Ca 2ϩ rise induced by ionomycin or photolysis of caged Ca 2ϩ was unable to normalize fusion of synaptic vesicles and insulin granules (99,102). Furthermore, Ca 2ϩ -dependent fusion and granule replenishment are unaltered in Munc13-1 deletion mutants (105).
Nonetheless, it should be remembered that the results mentioned above were obtained from protocols where Ca 2ϩ was used as the sole trigger for exocytosis. Second messengers such as cAMP have been known to act synergistically with Ca 2ϩ to potentiate release of neurotransmitters and exocytosis of granules (106). Similarly, GLP-1, and its mimetics being promising antidiabetogenic agents, have been well known to act via the cAMP signaling pathway to potentiate insulin exocytosis by severalfold, resulting in a partial bypass of insulin secretory deficiency in type II diabetes models (107). Remarkably, the potentiation is underscored by a marked increase in the size of RRP and replenishment of insulin granules and has been thought to be in part explained by alteration of ion channel activities in the ␤-cells, such as those of Kv2.1 (74), L-type VDCC (108), and K ATP channels (93). Alteration of activities of these channels is collectively associated with lower threshold for action potential and prolonged Ca 2ϩ influx, hence suggestive of greater magnitude of insulin exocytosis.
Activation of PKA by cAMP has been shown to cause PKA-catalyzed phosphorylation of RIM, which interacts with the N-terminal domain of Munc13-1 via its Zn 2ϩ finger region and modulates Munc13-1's priming activity (85,86). Importantly, PKA-catalyzed phosphorylation of RIM has recently been implicated to profoundly increase recruitment of Munc13-1/Unc13 to exocytotic sites (109,110). These observations suggest that cAMP/GLP-1 stimulation may lead to a dominant presence of open form Syn-1A in exocytotic sites.
Recent studies have suggested that besides Munc13-1, Syn-1A conformation may be affected by another protein called tomosyn. Tomosyn binds to the H3 SNARE motif of Syn-1A via its C-terminal VAMP homology region to form an inhibitory SNARE complex with Syn-1A and SNAP-25, reducing the availability of open form Syn-1A for assembly of functional SNARE complex (111,112). Indeed, several studies have suggested that tomosyn negatively regulates catecholamine exocytosis of PC12 cells (111), adrenal chromaffin cells (113), and insulin exocytosis (114). Moreover, tomosyn-deletion C. elegans mutants have enhanced synaptic fusion of neurotransmitters due to increased number of RRP of primed vesicles (112). Remarkably, like the expression of open form Syn-1A, deletion of tomosyn nearly restores fusion of synaptic vesicles in Unc13 null nematodes (112). These results suggest that tomosyn acts antagonistically to Unc13/Munc13 to regulate fusion competence of secretory granules. Tomosyn may therefore be important in the indirect regulation of ␤-cell ion channels by modulating Syn-1A conformations.
Interestingly, it has been reported that tomosyn is directly phosphorylated by PKA, resulting in decreased interaction with Syn-1A, increased formation of the functional SNARE complex, and increased size of RRP of primed synaptic vesicles (115). These observations thus shed light on an additional mechanism for the cAMP/GLP-1 pathway to alter ion channel activities in the ␤-cells.
Of note, it was recently reported that reduction of tomosyn by RNA interference leads to impaired insulin exocytosis in a ␤-cell line, INS-1E, suggesting that tomosyn positively regulates insulin secretion (116). However, it is not known in the study whether deficient tomosyn might indirectly cause reduced insulin secretion through perturbation of other exocytotic proteins that have direct positive roles in insulin secretion. This is in part supported by evidence in the study that overexpression of tomosyn did not potentiate insulin secretion, suggesting that the expression/function of tomosyn is not limiting (116). Therefore, the temporal patterns and possible cell-specific regulation of recruitment of Munc13-1 by RIM, alteration in tomosyn-Syn-1A interaction, and the resulting balance between closed and open form Syn-1A warrant further investigation.

VI. Conclusion and Future Perspectives
Earlier works on Syn-1A-Ca 2ϩ channel interaction, taken together with our works examining how Syn-1A inhibits Kv2.1 and K ATP channels, strongly suggest that Syn-1A changes its conformation not only to prepare for and promote vesicle exocytotic fusion, but also to modulate the gating of these ion channels in a manner that optimally regulates membrane excitability. Thus, during priming for exocytosis in ␤-cells, open form Syn-1A promotes K ATP channel closing  118) may suggest that this complex could also somehow modulate VDCC and K ATP channels as part of the priming process for secretion. Alternatively, Munc13-1 and tomosyn may modulate VDCC and K ϩ channels (Kv and K ATP ) via their ability to alter the conformation of Syn-1A.
In diabetic rodent and human ␤-cells, the reduced level of Syn-1A (32-35) would be expected to affect channel functions. For example, it is expected that in such ␤-cells, K ATP channels have a lower sensitivity to ATP inhibition, and dialyzing Syn-1A to these cells would enhance the ATP sensitivity. Similarly, Kv2.1 channels may have higher current density and a depolarizing shift in inactivation in those diabetic ␤-cells.
It would be of interest to determine whether Syn-1A-Kv interaction or Syn-1A-K ATP interaction is more important in modulating insulin secretion. The use of Kir6.2 genetically engineered mice (119) and a Kv2.1 knockout (yet unavailable) may give a partial clue to the relative importance of K ATP channels and Kv channels in ␤-cell insulin release, but could not provide an insight into the relative importance of Syn-1A-Kv interaction and Syn-1A-K ATP interaction. A feasible attempt would be to map out the "minimum-possible" binding domains of Kv2.1 and K ATP channels that interact with Syn-1A. Indeed, a smaller (half) segment within C1 of Kv2.1 C terminus has already been defined, which binds strongly to Syn-1A (120). Further efforts need to be made to define subdomain(s) within SUR1-NBFs that interact with Syn-1A. These minimum-possible binding domains would desirably not prevent channel interaction with other molecules (such as RIM/GEF) but only block Syn-1A binding. It would be informative to introduce these peptide domains in pancreatic islets and examine which one would more potently impair insulin secretion.