Exocytosis of insulin-containing secretory vesicles in pancreatic β-cells is crucial to maintenance of plasma glucose levels. They fuse with the plasma membrane in a regulated manner to release their contents and are subsequently recaptured either intact or through conventional clathrin-mediated endocytosis. Here, we discuss these mechanisms in β-cells at the single-vesicle level.
Insulin regulates whole body energy balance. Its secretion from the endocrine pancreatic islet β-cells is stimulated by increases in blood glucose and modulated by neuronal and hormonal inputs (2, 24, 50). Insulin acts primarily at liver, muscle, and fat to promote energy storage. The major effects of insulin include the upregulation of glucose uptake and storage and the downregulation of glycogen breakdown. Along with insulin, β-cells secrete a host of peptides and small molecules of largely unknown physiological relevance (34, 87). Among these, the small transmitter molecules ATP and γ-aminobutyric acid (GABA) may mediate important auto- and paracrine communication within the islets of Langerhans (14, 77, 84, 101). Secretion of these various peptides and small molecules occurs via the regulated fusion of vesicles with the plasma membrane, a process termed exocytosis.
In diabetes, the insulin-producing β-cells are either destroyed (Type 1) or demonstrate a reduced functional capacity (Type 2). In either case, this leads to a relative reduction in circulating insulin and increased blood glucose levels. The pathophysiology of reduced insulin secretion in diabetes is not well understood, but some evidence implicates defects in the distal mechanisms of insulin release (56, 67, 98, 100). Thus an understanding of how individual vesicles fuse with the plasma membrane, release their content, and are ultimately recaptured may be of importance in determining causes and treatment of diabetes. Although several excellent recent articles have reviewed β-cell exocytosis at the single-cell level (12, 78, 79), in the present review, we will focus on exocytosis and vesicle recapture in the pancreatic β-cells at the single-vesicle level.
Insulin Release by Exocytosis
In pancreatic islet β-cells, insulin is stored in a crystalline (30), Zn2+-complexed (10, 23) form within dense-core vesicles (DCVs), named so due to their dense appearance in electron micrographs (FIGURE 1A⇓). When blood glucose is elevated, insulin-containing DCVs are stimulated to fuse with the plasma membrane and release their contents into circulation. The mechanism coupling the glucose stimulus to DCV exocytosis (summarized in FIGURE 1B⇓) has been extensively reviewed (50, 76) and will not be considered here. In order for DCVs to undergo exocytosis, they must be in close proximity to the plasma membrane. Indeed, a subset of insulin-containing DCVs within the β-cell appears physically “docked” with the plasma membrane, in close association with the voltage-dependent Ca2+ channels (VDCCs) that mediate Ca2+ entry (8, 9, 52). This results from a physical interaction of the channels with exocytotic SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) proteins (102). These well conserved and essential components of exocytosis are expressed in neuronal and endocrine cells, including β-cells, and have recently been reviewed in Physiology (80). This pool of vesicles, accounting for ≤1% of all β-cell DCVs, likely underlies the rapid first phase of insulin secretion (79). It is worth noting that in human Type 2 diabetes the first phase of insulin release is reduced (28), often before overt disease onset. Furthermore, the expression of several exocytotic proteins is downregulated in cultured islets exposed to hyperglycemia (1), animal models of diabetes (56, 68), and islets from humans with Type 2 diabetes (67), supporting a role for reduced exocytosis in diabetes.
Membrane Recapture by Endocytosis
By definition, exocytosis results in addition of vesicular membrane to the plasma membrane. To maintain membrane homeostasis, the added membrane must eventually be recaptured. Indeed, stimulated insulin secretion is coupled to the reuptake of vesicles (65). Whether this occurs rapidly at the site of exocytosis (by kiss-and-run; see below) or through the upregulation of conventional endocytosis is a matter of debate. The so-called “conventional” endocytosis (reviewed in Refs. 17, 83, 91) involves recruitment of clathrin, which, along with a host of adaptor proteins, initiates an inward curvature of the plasma membrane. The GTPase dynamin is then thought to form a ring structure that constricts to mediate the ultimate scission of the endocytotic vesicle from the plasma membrane. Although the exact mechanism by which dynamin catalyzes membrane fission has yet to be elucidated (91), the role for this enzyme in endocytosis is undisputed.
This conventional endocytotic pathway can be coupled to stimulated exocytosis through its upregulation by Ca2+ (17, 18, 86). Calcium acts through the Ca2+-dependent phosphatase calcineurin, which dephosphorylates a number of endocytotic proteins collectively termed the dephosphins (17). Dephosphrylation of several of these, including dynamin and the clathrin adaptor protein AP180, appears crucial to the upregulation of endocytosis (18). Recent work suggests that many of the dephosphins are subsequently rephosphorylated by Cdk5 (92, 96). Although little is known about the upregulation of conventional endocytosis in β-cells, several endocytotic proteins, including the dephosphins, are expressed in insulinoma cells and pancreatic islets (32, 39, 66).
Monitoring Single Vesicles in the β-Cell
An elegant way of monitoring single insulin release events is to develop electrodes that sense insulin. Such measurements have been performed (33, 38) and allow the millisecond resolution of individual secretory events (FIGURE 2A⇓; reviewed in Ref. 55). This work importantly demonstrated that frequency changes in single-vesicle exocytosis correspond well with known insulin secretory responses (33). Notably, transient insulin secretion in response to depolarization with high extracellular K+ evoked a brief burst of exocytotic activity, whereas the prolonged secretory response to glucose resulted from a sustained period of stimulated DCV exocytosis. Ultrastructural analysis has revealed that stimulation with high K+ alone leads to the depletion of the pool of docked granules, an effect that is prevented in the presence of glucose (64). Clearly, glucose has the capacity to maintain the size of the docked vesicle pool by recruitment of new granules (63). The amperometric detection of insulin itself has not been widely adopted, however, largely due to the difficulty in producing insulin-oxidizing electrodes. Improved methods for producing these will undoubtedly prove useful (69, 74, 85, 107).
The majority of amperometric studies of single-vesicle secretion from β-cells do not measure quantal insulin release but instead rely on preloading of cells with serotonin (5-HT) (108), which is much more easily oxidized than insulin. Although preloading with 5-HT may alter β-cell secretory activity (105), this methodology is thought to faithfully report the exocytosis of single DCVs (5, 6). This work has demonstrated the direct coupling between membrane depolarization (108), bursts of β-cell action potentials (11), and DCV exocytosis; an important role for ATP and cAMP in increasing the number of rapidly releasable vesicles (89); and compound or multi-vesicle exocytosis (11). The occurrence of the latter is supported by fluorescence imaging studies (42), although more recent work in intact islets suggests that this is a rare occurrence in β-cells (88). Comparison of unitary 5-HT and insulin release from β-cells has also revealed important aspects of membrane fusion kinetics. However, for obvious reasons, amperometic measurements of 5-HT release do not provide any information about what happens to vesicles after fusion with the plasma membrane.
Advances in fluorescence imaging approaches (FIGURE 2B⇑) have made it possible to track a single DCV within the β-cell before, during, and in some cases after exocytosis. These methods include the use of lipophilic dyes (42), DCV-targeted fluorescent proteins (61, 70, 71, 99), fluorescent extracellular probes (47, 88, 90), and fluorescent Zn2+ indicators (27, 72, 73). Combination of these markers with confocal microscopy, two-photon imaging (37), or total internal reflection fluorescence microscopy (TIRF-M) (60) has resulted in a number of seminal observations (80). These include evidence supporting the direct and intact recapture of insulin-containing DCVs following exocytosis (59, 70, 82, 97, 99); the compound or sequential exocytosis of DCVs (40, 42, 88); and a role for DCV recruitment to the membrane as an underlying mechanism for sustained exocytosis (57, 63). Perhaps more importantly, this recent work has raised numerous questions, and indeed the above conclusions are not universally accepted. Two major areas of contention remain to be resolved in our view: 1) what is the contribution and physiological role of the direct and intact recapture of vesicles, via a kiss-and-run mechanism, to the recycling of β-cell DCVs (47, 88, 90); and 2) does DCV exocytosis occur at physically stable “hot spots,” perhaps defined by SNARE protein clusters (62)? An elegant three-dimensional imaging approach has recently provided evidence against the latter (81).
More recently, we have employed two patch-clamp-based methodologies to approach the issue of single vesicle exo- and endocytosis. The first employs the overexpression of ionotropic P2X2 purinoreceptors or GABAA receptors to detect the unitary release of ATP and GABA, respectively (13, 59) (FIGURE 2C⇑). Thus ATP or GABA released from a single vesicle in the β-cell under study is detected as a current spike resulting from the autocrine activation of the ionotropic receptors. This technique is somewhat comparable to amperometry in that the quantal release of transmitter is detected but provides the advantage of measuring all events over the surface of the cell. Since ATP is stored in β-cell DCVs and co-secreted with insulin (41), ATP spikes in these cells are good indicators of DCV exocytosis. GABA is stored within β-cell SLVs (75, 93), and thus the detection of GABA spikes in β-cells expressing GABAA receptors may reflect single SLV exocytotic events (13), but it is important to mention that some β-cell DCVs also contain GABA (26) and that they are thus likely to contribute to the unitary GABA release events observed with this technique.
In a second approach, we utilized a variation of the capacitance technique (reviewed in Ref. 36) that monitors the capacitance (i.e., membrane surface area) of small cell-attached membrane patches rather than the entire cell (46) (FIGURE 2D⇑). This method increases the resolution of capacitance measurements by >100-fold, allowing single exo- and endocytotic vesicles to be observed as discrete capacitance steps. The diameter of single vesicles undergoing exo- or endocytosis can be calculated from the magnitude of the capacitance step (46). Using this method, we were able to monitor the exocytosis of vesicles as small as 50 nm in diameter in INS-1 and rat β-cells and thus could distinguish between the exocytosis of single SLVs and LDCVs based on their respective sizes (48, 51). Furthermore, the ability to record cell-attached patch capacitance in electrically and metabolically intact cells makes it possible to investigate the response of single vesicles to physiological stimuli such as glucose. Finally, this technique allows the simultaneous measurement of changes in membrane patch electrical conductance, and therefore the analysis of single small fusion pore kinetics in a manner similar to that of a single ion channel (19).
The Regulated Exocytosis and Rapid Recapture of β-Cell SLVs
In addition to DCVs, pancreatic β-cells contain small (~70 nm in diameter), clear SLVs that actively accumulate GABA (75, 93) and share many similarities with neuronal synaptic vesicles (94) (FIGURE 1⇑). Until recently, it was unclear whether endocrine SLVs were even capable of undergoing regulated exocytosis, and most evidence in support of this idea was indirect (44). We sought to address this issue through the overexpression of GABAA-receptors in β-cells and thus observed Ca2+-regulated exocytosis of GABA-containing vesicles (13). Further work demonstrated that the release of GABA via this pathway acts to inhibit glucagon secretion from neighboring α-cells (101) and to inhibit insulin secretion via an autocrine pathway (14). However, the recent ultra-structural finding (26) that some DCVs also contain GABA complicates the interpretation of the data.
Using single-vesicle on-cell capacitance measurements, we have obtained direct evidence for exocytosis of small SLVs. Like β-cell DCVs, the exocytosis of SLVs can be stimulated by glucose through a pathway that is dependent on L-type VDCCs (51). We further demonstrated glucose-stimulated SLV exocytosis at a rate slightly lower than that of DCV exocytosis (40 vs. 60 vesicles · cell−1 · min−1). Data from insulinoma cells and rat β-cells showed that the membrane docked pool of SLVs is larger than that for DCVs (100 vs. 67 vesicles/cell) (51) and that a greater fraction of SLVs (1.2% min−1) than DCVs (0.6% min−1) undergo exocytosis on stimulation.
The relatively smaller reserve pool of β-cell SLVs may be offset by the mechanism of SLV recycling. The so called “conventional” view of vesicle recycling is described above and shown in FIGURE 3A⇓. More recently, synaptic vesicle recycling by a “kiss-and-run” mechanism has been intensively investigated. In this mechanism (FIGURE 3B⇓), the vesicle fuses with the plasma membrane only transiently and is reinternalized intact. The vesicle can then be refilled directly with neurotransmitter, ready for another round of exocytosis. This requires the presence of transporters for these transmitters in the β-cell secretory vesicle membrane and such have indeed been documented in ultrastructural studies. However, whether β-cell DCVs can be refilled following partial release has not been studied. In chromaffin granules, ATP-dependent uptake of ATP has been documented and requires a positive intra-granular membrane potential set up by a bafilomycin-sensitive proton pump and appears to involve DIDS-sensitive Cl– channels (7). The relative contribution of the kiss-and-run and conventional recycling mechanisms is still a matter of debate (29, 31). However we recently demonstrated kiss-and-run as a mechanism for β-cell SLV recycling (48, 51). This occurred via the transient opening of small fusion pores averaging 0.8 nm in diameter and sufficient to allow the release of most of the GABA content (48). Furthermore, although kiss-and-run accounted for approximately 25% of the exocytotic events in the intact cell, this value approached 70% when we examined membrane-docked SLVs exclusively [using excised inside-out patches (51)], suggesting that predocked vesicles may preferentially undergo kiss-and-run recycling. It seems possible that these events are followed by the refilling of β-cell SLVs, since these can actively accumulate GABA in an ATP-dependent manner (93).
Recapture of Membranes by Conventional Endocytosis
Although kiss-and-run appears to account for the reinternalization of a relatively large fraction of exocytotic SLVs, particularly membrane docked SLVs, it does not account for the recycling of all SLVs. A bigger issue is the recapture of DCV membranes, which account for as much as 99% of the cell surface area increase during stimulated exocytosis (13). Although some DCVs may be recaptured through a kiss-and-run mechanism (discussed below), this is responsible at the very most for 50–60% of vesicles. Therefore, a mechanism must exist to retrieve exocytosed membranes following stimulation to maintain membrane homeostasis and cell size.
The stimulated upregulation of compensatory endocytosis is well documented and described above. Not surprisingly then, in β-cells, endocytosis is triggered in a stimulation- and Ca2+-dependent manner (25), involves a calcineurin-dependent step and the formation of phosphatidylinositol 4,5-bisphosphate (32), which may play an important role in the recruitment of endocytotic proteins and stimulated endocytosis (22). Endocytosis itself has received little attention at the single vesicle level in β-cells, in part due to the lack of high-resolution techniques to monitor it. We have found (49) that the vast majority (98%) of endocytotic events in INS-1 insulinoma cells are not the result of direct retrieval of exocytotic vesicles, and furthermore these were comparatively small (70-nm diameter). Although whole-cell membrane retrieval was indeed upregulated in a glucose- and Ca2+-dependent manner, the frequency of endocytotic events did not change significantly. The increase in membrane retrieval was due solely to a 60% increase in endocytotic vesicle size concomitant with an increased rate of fission pore constriction.
Few studies have addressed the regulation of endocytotic vesicle size. Certain manipulations, such as disruption of the actin cytoskeleton (16) and excision of membrane patches (20), are known to alter endocytotic vesicle size. More interesting perhaps, mutation of the AP180 homolog in C. elegans and Drosophila increases synaptic vesicle size as a result of altered vesicle biogenesis at the point of endocytosis (58, 106). AP180 is also known to regulate the size of clathrin cages in vitro (104) and thus represents an interesting potential target for the Ca2+-dependent regulation of endocytotic vesicle size, particularly since this clathrin adaptor protein is dephosphorylated in a Ca2+-dependent manner by calcineurin. The same is true for dynamin, which may contribute to the Ca2+-dependent increase in fission pore constriction observed. The specific mechanism mediating upregulation of endocytosis in β-cells remains largely unclear. Although we provide some evidence for a stimulated increase in endocytotic vesicle size, further work is needed to clarify the role and mechanism for the upregulation of conventional endocytosis in these cells.
Kiss-and-Run Exocytosis of β-Cell DCVs
Kiss-and-run as a mechanism for DCV exocytosis and recycling is controversial. Strong evidence suggests that chromaffin cell DCVs undergo kiss-and-run (20, 45) and, furthermore, that the transient opening of the fusion pore during these events allows catecholamine release (3, 21). The vesicles are presumably refilled directly with catecholamine following these events (95). The physiological rationale for kiss-and-run of peptide-containing DCVs, such as in β-cells, is less clear since these must return to an endosomal compartment to be refilled with peptide cargo. Thus it seems unlikely that the kiss-and-run of β-cell DCVs would serve the purpose of vesicle recycling at the membrane following insulin release. Other possible roles for the kiss-and-run of β-cell DCVs remain, including the maintenance of vesicular membrane protein and lipid composition, vesicle priming via alterations in luminal pH, and the selective release of small transmitter molecules through the fusion pore. It is the latter possibility that we have become the most interested in.
In β-cells, the frequency of kiss-and-run, its apparent mechanism, and indeed its definition depend on the method by which it is studied. The amperometric method has provided valuable information regarding formation and expansion of the fusion pore before full vesicle fusion (33, 108) and can report kiss-and-run exocytosis as the gradual release of transmitter through a restrictive fusion pore without a rapid full-release spike (4). However, this method has not been used to report the frequency of β-cell DCV kiss-and-run due to the difficulty in discriminating between kiss-and-run events and full fusion events occurring far from the carbon fiber electrode. Examination of cytosolic marker exclusion (47, 88, 90) generally reports a low frequency (>7%) of DCV kiss-and-run, whereas studies utilizing fluorescent DCV cargo or membrane markers have produced results in striking contrast (as much as 60%) (59, 82, 97, 99). Our own electrophysiological measurements present intermediate results, suggesting that DCV kiss-and-run occurs 20–30% of the time in response to glucose stimulation (48, 51). Furthermore, direct imaging of insulin crystal dissolution demonstrated that approximately 33% of insulin release events occurred via a slow release form (54), which may correlate to release by a kiss-and-run mechanism.
These approaches represent fundamentally differing measurements, however, making direct comparisons difficult. Studies examining tagged DCV membrane proteins provide the clearest evidence that the β-cell DCVs frequently remain intact following exocytosis and are often associated with the retention of peptide and closure of the fusion pore (59, 97). Monitoring the release of different sized cargoes, it was inferred that multiple forms of kiss-and-run exist (99), which may be related to different functional states and different stable diameters of the fusion pore (FIGURE 3, B AND C⇑). Indeed, kiss-and-run fusion pores of β-cell DCVs have been reported as either small (1.4 nm or less) (48, 90), of intermediate diameter (97), or so large as to be practically nonexistent (70, 82, 97, 99). These measurements are not without their drawbacks, however, and expression of bulky granule-targeted probes may alter the biophysical properties of the exocytotic release event (53).
If the relative occurrence of transient fusion of DCVs in the β-cell is not well established, the mechanistic basis for this is even less so. Recent work demonstrated the recruitment of dynamin to putative DCV kiss-and-run sites in insulin-secreting cells (97), suggesting a mechanism whereby dynamin restricts expansion of the fusion pore before the vesicle can fully fuse with the plasma membrane. Other work suggests a role for the Ca2+-sensing exocytotic protein synaptotagmin 1, since genetic knockdown of this resulted in a reduction in fast endocytosis often taken to represent kiss-and-run (103). Our own work suggests that a stepwise opening and closing of a fusion pore, not much larger than an ion channel (1.4-nm diameter), mediates transient DCV fusion and acts to slow the release of ATP and likely to restrict or even prevent the release of insulin (48). This may explain the different results obtained by imaging of extracellular markers vs. granule-targeted probes. The former rely on extracellular markers ≥1.4 nm in diameter, whereas the latter often depend on transient changes in vesicle pH. It seems clear that measurements relying on the flux of the smaller marker (that is H+: diameter ~0.1 nm) would detect a proportionally greater number of fusions mediated by small pores. Our own measurements of the early fusion pore demonstrate that permeation of the pore increases with decreasing transmitter size (14a). The sequence GABA > 5-HT > ATP echoes the size of these molecules, suggesting that their flux may be determined by the dimension of the pore.
Larger fusion pores have been associated with long-lasting “kiss-and-stay,” “kiss-and-glide,” or “cavicapture” events responsible for the release of peptides of 5- to 10-nm diameter (97). We ourselves have detected DCV fusion pores on this scale that failed to close over the course of an experiment (48). These became apparent at the expense of the smaller, transient fusion pores in the absence of forskolin, suggesting a role for cAMP as a regulator of the kiss-and-run fusion pore. The nature of the fusion pore, whether it is proteinaceous or lipidic, remains unclear (for a review of these, see Refs. 15, 35, 43). The answer may be “both”; i.e., that the protein pore accounts for the initial small (ion channel-like pore), whereas the cavicapture-type pore is predominantly lined by lipids.
Conclusion and Future Perspective
A consensus has yet to be reached in regard to specific details of the exocytotic and endocytotic process in β-cells of the pancreas. Several interesting lines of research have been generated by the study of single vesicle exocytosis through a variety of approaches. Many discrepancies remain to be investigated and may only be resolved through the combination of elecrophysiological and imaging approaches. Recent imaging experiments have revealed that the insulin secretion defect that develops following long-term exposure of β-cells to high glucose results from insufficient expansion of the fusion pore rather than a reduction of exocytosis per se (98). It is an exciting possibility that similar defects may contribute to the pathophysiology of Type 2 diabetes in humans.
We thank Dr. M. Braun and Anne Clark for providing some of the original data presented in the Figures.
This review was supported by grants from the Alberta Heritage Foundation for Medical Research and Canadian Institutes of Health Research to P. E. MacDonald (MOP 81350) and the Wellcome Trust to P. Rorsman. P. E. MacDonald is a Canadian Diabetes Association Scholar, Alberta Heritage Foundation for Medical Research Scholar, and the Canada Research Chair in Islet Biology. P. Rorsman is a Wolfson-Royal Society Merit Award Research Fellow.
- © 2007 Int. Union Physiol. Sci./Am. Physiol. Soc.