Minute-to-minute control of the release of insulin by pancreatic β-cells in response to glucose or other stimuli requires the precise delivery of large dense-core vesicles to the plasma membrane and regulated exocytosis. At present, the precise spatial organization at the cell surface and the nature of these events (‘transient’ versus ‘full fusion’) are debated. In order to monitor secretory events simultaneously over most of the surface of clusters of single MIN6 β-cells, we have expressed recombinant neuropeptide Y-Venus (an enhanced and vesicle-targeted form of yellow fluorescent protein) as an insulin surrogate. Individual exocytotic events were monitored using Nipkow spinning disc confocal microscopy, with acquisition of a three-dimensional complete image (eight to twelve confocal slices) in <1 s, in response to cell depolarization. Corroborating earlier studies using TIRF (total internal reflection fluorescence) microscopy, this approach indicates that events occur with roughly equal probability over the entire cell surface, with only minimal clustering in individual areas, and provides no evidence for multiple events at the same site. Nipkow disc confocal imaging may thus provide a useful tool to determine whether event types occur at different sites at the cell surface and to explore the role of endocytic proteins including dynamin-1 and -2 in terminating individual exocytotic events.
Defective insulin secretion underlies the pathology of both Type 1 (insulin-dependent) and Type 2 (non-insulin-dependent) diabetes, and the incidence of both diseases, currently affecting 4–8% of most westernized populations, is set to double in the next 20 years . Exocytotic release of insulin is triggered by enhanced glucose uptake and metabolism as the extracellular concentration of the sugar rises, followed by increases in intracellular ATP/ADP ratio, closure of ATP-sensitive K+ channels  and Ca2+ influx . In addition, glucose ‘amplifies’ these effects through mechanisms that involve changes in the apparent sensitivity of the secretory machinery to Ca2+ [4,5], possibly mediated by changes in lipid metabolism . Importantly, inhibition of AMP-activated protein kinase  is also essential for the normal stimulation of insulin secretion in response to glucose in isolated cells , islets  or in living mice , a change which appears to be essential for the recruitment of secretory vesicles to the cell surface  prior to exocytosis.
Sites of insulin release at the β-cell surface
Are secretory events spatially organized at the cell surface, and does their localization affect the type of event observed? Two-photon imaging of the soluble extracellular dye sulforhodamine B suggested that most of the events occur at the ‘abvascular’ (i.e. cell–cell interfaces away from blood vessels) of β-cells in the intact islet . ‘Zooming in’ further, evidence for the involvement in β-cells of the CAST (cytomatrix of the active-zone-associated structural protein) homologue ‘Elks’, an active-zone-enriched protein in neurons, has recently been provided by Nagamatsu and co-workers . In a similar vein, it has been suggested that exocytosis in PC12 cells occurs principally in regions poor in ‘lipid rafts’, given that SNARE (soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptor) proteins with enhanced association with these regions were less active . In contrast, through rapid confocal imaging of the insulin surrogate NPY (neuropeptide Y)–Venus release, we have so far seen little evidence for tight organization or of the ‘queuing’ of vesicles to fuse to the plasma membrane (Figure 1). In this approach, a near complete three-dimensional image of the cell is captured at >1 Hz, involving stacks of eight to twelve individual scans at approx. 0.5 μm spacing. Subsequent subtraction of consecutive images allows the position and timing of individual NPY–Venus release events (which cause a localized peak and fall in fluorescence, which is almost complete in 1–2 s) to be determined (Figure 1). Importantly, this approach provided no compelling evidence for compound exocytosis, i.e. the fusion of ‘incoming’ vesicles to those already resident and fused to the plasma membrane .
Rapid three-dimensional spinning disc confocal imaging of individual excytotic events in MIN6 β-cell clusters
Nature of the exocytotic event: full fusion or transient (kiss-and-run/cavity recapture)?
Whether LDCV (large dense-core vesicle) fusion involves a ‘complete’ merging of the vesicle and plasma membranes has been a matter of considerable contention over the past few years. Imaging of molecular LDCV-targeted and other probes at the single-cell (vesicle) level by ourselves and others has provided a growing body of evidence against this view, proposing instead that this interaction is usually transient, with a variable duration of <1 s to >10 s . Thus, using TIRF (total internal reflection fluorescence) microscopy  to image the behaviour of recombinant vesicle-targeted fluorescent probes immediately beneath the plasma membrane, we have defined three principal types of event [11,18]: (i) ‘pure’ kiss and run, involving the formation of a narrow pore and the selective release of low-molecular-mass species (ATP, GABA etc.) from the LDCV; (ii) ‘mixed’ kiss and run or ‘cavity recapture’ involving the further release of selected vesicle membrane proteins, possibly including ion channels etc. destined for localization on the plasma membrane, but not other cargo proteins; (iii) ‘quasi-full’ exocytosis involving the release of all the above, plus constituents of the crystalline dense core, including insulin. Importantly, the extent of release of the latter during ‘quasi-full’ fusion appears to be variable, i.e. an individual LDCV may release all or just a proportion of its hormone cargo, depending on the strength of stimulation (notably local concentrations of Ca2+ or possibly cAMP) as well as the levels of key regulators of the exocytotic process. Indeed, changes in the ‘completeness’ of release may contribute to insulin-secretory deficiency in some forms of Type 2 diabetes. Thus external factors, including lengthy exposure to elevated glucose or fatty acids, as may occur in patients with Type 2 diabetes and which may lead to ‘gluco/lipo-toxicity’, affect the extent of ‘quasi-full’ fusion, with the proportion of complete cargo release (monitored with NPY–Venus) decreasing in primary rat β-cells from approx. 25% after culture at near normal glucose concentrations to <5% after 48 h at 30 mM glucose .
Evidence has also been obtained by imaging other (non-recombinant) probes such as the lipophilic dye FM-1-43 , and recently by electron microscopy  for transient interactions between LDCVs and the plasma membrane. However, electrophysiological recordings of capacitance changes appear currently to be somewhat at variance with this model. The reasons for the discordance are unclear but variability including temperature (30°C for electrophysiology and 37°C for imaging) may be one factor, as well as the duration and strength of stimulation, and finally the ability of the latter technique to resolve the formation of very-large pores is capable of allowing the release of proteins including insulin. In any case, parallel measurements of capacitance alongside the imaging of single-vesicle fusion events are eagerly awaited.
Molecular mechanisms underlying cavicapture of insulin-containing vesicles
Reversal of SNARE pair formation seems unlikely to be favoured energetically so what drives fusion pore closure and vesicle recapture? Simultaneously imaged with a fluorescent vesicle cargo (NPY–mRFP), dynamin-1–EGFP was observed to arrive at sites of exocytosis synchronously with the onset of release events, and to linger for 2–3 s after the peak of release, consistent with a role for this small GTPase, well known for its role in ‘rapid’ exocytosis in neurons , in closing a fusion pore and hence terminating release of insulin from a single vesicle .
Although the above experiments demonstrated that dynamin-1 is (i) recruited to exocytotic sites and (ii) required for pore closure , the signals that recruit this GTPase remain unclear. Thus local increases in PtdIns(4,5)P2 may play a limited role in as much as other pleckstrin homology domain-containing proteins associated with (clathrin-mediated) endocytosis such as amphiphysin and epsin were not recruited to these sites. Whether local increases in the concentration of other soluble messengers such as Ca2+ [22,23] might be involved in recruiting dynamin-1, perhaps via binding partner(s) that form the components of a ‘capturesome’ involved in retrieving exocytosing LDCVs, remains to be determined, as does the nature and composition of such a structure. Potential binding partners including members of the SNX (sorting nexin) family, including SNX9  as well as phospholipase D  and syndapin-1. Thus dynamin-1 has been shown recently to interact with syndapin-1 (also called PACSIN for protein kinase C and CK2 substrate in neurons)  in a phosphorylation-dependent manner, whereby reversal of Cdk5 (cyclin-dependent kinase 5) mediated phosphorylation at two residues (Ser-774 and Ser-778) increases the affinity of dynamin-1 for syndapin-1. There is evidence that a similar mechanism may operate in β-cells. Thus both dynamin-1 (the ‘neuronal’ form of this protein) and syndapin-1 are present in mouse β-cells at the mRNA and, albeit in low amounts, at the protein level (M.K. Loder and G.A. Rutter, unpublished work). An interaction between dynamin-1 and syndapin-1 may thus be an important signalling event controlling the termination of exocytotic events (Figure 2).
Possible roles of syndapin-1 in dynamin-1 recruitment during cavity recapture of LDCVs
Of note, ‘capturesome’ structures are likely to form transiently and be highly localized on the endocytosing vesicle. Since only a tiny proportion of intracellular vesicles (fewer than 0.3% of total granules per minute during strong stimulation) undergo exocytosis in β-cells, the isolation of this complex is likely to be difficult such that its characterization will require a candidate approach and systematic inactivation/silencing strategy. Definitive identification of binding partners for dynamin-1 (e.g. through immunoisolation/MS analysis or yeast two-hybrid analysis) may provide useful clues. Finally, it should also be mentioned that dynamin has also been proposed to play a positive role in vesicle release, based on the function of the Saccharomyces cerevisiae homologue Vps1p in vacuole formation , as well as studies of the association between dynamin-2 and syntaxin-1 in PC12 cells  and a recent report  of a stimulatory effect of a dynamin dominant-negative on insulin release from MIN6B cells.
Pathophysiology and event types
Are there changes in the nature of release events in pathological states including Type 2 diabetes? Consistent with this view, extended culture of rat β-cells at elevated glucose concentrations (30 mM) led to a decrease in the proportion of events, which led to essentially complete release of the vesicle cargo (from ∼25 to 5%) . This change was fully reversible upon lowering the culture glucose concentration to near normal levels for a further 24 h. Interestingly, in an in vivo model of mild hyperglycaemia and severe lipidaemia, the ZDF (Zucker diabetic fatty) rat, analysis of oligonucleotide microarrays  revealed changes in the expression of a variety of genes likely to be associated with exocytosis including VAMP2 (vesicle-associated membrane protein 2), rab3, RIM2 and rab27a, whereas dynamin-1 mRNA levels were identical in ZDF and non-diabetic fa/+ animals, consistent with similar changes in human Type 2 diabetes . Thus a relative slowing of vesicle fusion, compared with dynamin-1-mediated vesicle recovery, may contribute to a decrease in the proportion of ‘full’ release events in this model, and hence to defective overall insulin secretion.
The use of advanced microscopy approaches is allowing the nature and organization of insulin secretory events to be analysed in fine detail at the single-cell level. Changes in these parameters, including the premature termination of single exocytotic events, are demonstrated as a possible contributor to the overall deficiency in insulin secretion observed in Type 2 diabetes .
Co-ordination of Cellular Processes: A Focus Topic at BioScience2006, held at SECC Glasgow, U.K., 23–27 July 2006. Edited by P. Clarke (Dundee, U.K.), P. Coffer (Utrecht, The Netherlands), M. Cousin (Edinburgh, U.K.), I. Dransfield (Edinburgh, U.K.), S. High (Manchester, U.K.) and G. Rutter (Imperial College London, U.K.).
Supported by grants from the Wellcome Trust, Juvenile Diabetes Research Foundation, Medical Research Council and the National Institutes of Health (Bethesda, MD, U.S.A.). G.A.R. is a Wellcome Trust Research Leave Fellow. We thank Ken Salisbury (Improvision, Coventry, U.K.) for assistance with data analysis.