Kiss-and-run, fuse-pinch-and-linger, fuse-and-collapse: The life and times of a neurosecretory granule

The fate of secretory membrane after fusion with the cell surface has been a subject of interest and considerable debate for almost 30 years. At nerve terminals, the limited number of synaptic vesicles and often high rates of action potential firing require that efficient machinery be in place to recover and replenish the vesicle pool to sustain the flow of information across the synapse over time. Since the first description of vesicle recycling at synapses (1, 2), a debate about the nature of the machinery responsible for this task has persisted.

One of the hallmarks of this debate has been the discussion of whether secretory vesicles (donor), after fusion with the plasma membrane (acceptor), are retrieved via endocytosis after a complete intermixing with the acceptor membrane or in a more intact form via a mechanism that directly pinches off the vesicle. This latter pathway, often referred to as kiss-and-run, circumvents the need for endocytic resorting of donor from acceptor membrane components during the process of resculpting a secretory vesicle. Two new studies by Holroyd et al. (3) and Taraska et al. (4) provide some of the most convincing evidence to date that in PC-12 cells, a model neurosecretory cell line, a direct retrieval route is manifest for a sizeable fraction of secretory events. This was achieved by clever imaging methods that can visualize individual secretory vesicles and some of their proteins before, during, and after exocytosis in intact cells (4), as well as semiintact membrane sheets (3) that offers complete access to the cytoplasmic space.

The debate about the mechanism of vesicle retrieval began over differing interpretations of images of ultrastructure of synaptic terminals after intense action potential firing (1, 2). Although clathrin-coated pits and membrane-associated clathrin lattices were clearly observed peripheral to sites of exocytosis during stimulation, the kinetic efficiency of such processes was thought to be much too slow to keep up with the demands of neuronal circuitry. Evidence for another retrieval route was first obtained when electrical capacitance measurements of single secretory granules fusing with the plasma membrane revealed that the continuity between vesicle lumen and extracellular space was often very transient, occasionally displaying flickering-like properties. These fusion pore dynamics were taken as a signature of kiss-and-run-type events, as well as evidence that the fusion process itself (and therefore the underlying molecular machinery) might be reversible. These flicker-fusion pores were originally only detectable for larger secretory granules; however, they have also recently been observed for small synaptic-like secretory microvesicles (5).

Is the vesicle retrieved intact, or is there continuity between the plasma membrane and the vesicle after fusion?

Although electrophysiological evidence provided new insights into the dynamics of fusion, they did not directly address the fate of the postfusion granule. In the studies by Holroyd et al. (3) and Taraska et al. (4), fluorescence imaging of GFP-labeled secretory peptide (neuropeptide Y, NPY) provided a means to visualize single secretory granules, measure exocytosis of peptide, and allow the observation that the vesicle itself frequently remained in place after exocytosis, by virtue of the fact that some peptide remained in the lumen of the vesicle. The fact that GFP is also a convenient pH sensor allowed the authors of the studies to verify that, after exocytosis, the vesicle lumen reacidified, and thus that either the fusion pore had closed or that fission had occurred. In the study by Holroyd et al., it was also possible to show that the ability to take up and trap external tracers in secretory vesicles (HRP or sulforhodamine) could be blocked by interfering with dynamin function, a critical player needed to carry out the mechanochemistry of membrane fission at the plasma membrane, as well as in other intracellular membrane systems.

Although it is possible that dynamin might interact directly with fusion machinery and regulate fusion pore dynamics, there is no biochemical indication that such a direct (or even indirect) interaction takes place. Instead, it seems likely that three potentially distinct mechanisms may be in play with regard to the fate of postfusion granule membranes (Fig. 1). For intact PC-12 cells, fusing granules appear to follow conventional fusion leading to complete merger of donor and acceptor membranes (fuse-and-collapse) about one-third of the time (4). In the semiintact PC-12 membrane sheets studied by Holroyd et al. (3), the number of fusion events leading to uptake and retention of external fluid tracers is reduced 60% by interfering with dynamin function. Thus, a sizeable fraction of transient fusion events appears to be best described as evidence for regulation of the fission pore rather than the fusion pore (Fig. 1, fuse-pinch-linger). The remaining fusion events that led to retention of an intact vesicle but were resistant to dynamin block may represent pure membrane retrieval by simple fusion pore closure (referred to here as kiss-and-run).

Figure 1.

Three types of membrane retrieval mechanisms may coexist for neurosecretory vesicles. (Top) A conventional intermixing of the membrane follows the fusion event eventually leading to a clathrin-mediated retrieval of the vesicle components (fuse-and-collapse). (Middle) A new mechanism whereby dynamin serves to cause fission of the intact fused vesicle before collapse and intermixing with the plasma membrane (fuse-pinch-linger). (Bottom) The fusion event is terminated by a direct closure of the fusion pore (kiss-and-run).

The specific pathway that a postfusion granule adopts may thus depend on the relative balance of the kinetics of fusion pore dilation and reclosure compared with the kinetics of dynamin-induced fission. The direct closure of the fusion pore on its own may occur only relatively rarely, because electrophysiological measures of rapid fusion pore conductance flicker indicate that under standard conditions it occurs in only 5% of fusion events for both large granules and synaptic-like microvesicles (5, 6). Perhaps under conditions where the kinetics of dynamin assembly or GTP hydrolysis is limited, the fused vesicle proceeds to a complete intermixing of membranes.

A number of important considerations naturally arise concerning the implications of a dynamin-mediated, but presumably nonclathrin-mediated, retrieval pathway. Is the vesicle really retrieved intact, or is there continuity between the plasma membrane and the vesicle after fusion that would allow intermixing of components? If membranes intermix, then a direct retrieval route could be futile, because the vesicle might loose critical molecular components onto the plasma membrane. In addition to imaging GFP-tagged NPY, Taraska et al. (4) also determined that phogrin (a secretory granule transmembrane protein) does not appear to disperse into the plasma membrane after fusion. Lateral dispersion might be prevented if a proteinaceous fusion pore, acting as a barrier to lateral diffusion, persists for the entire time between fusion and fission. Previous measurements of fusion events with giant secretory granules in mast cells (7) indicate that lipid markers freely gain access to the vesicle membrane after fusion and before fission. Thus, the retention of vesicle proteins during a fuse-pinch-linger-type event may simply be a question of timing. The time for a protein to effectively escape by lateral diffusion is given by τ_diff = (r_v²/4D), where r_v is the radius of the vesicle and D is the lateral diffusion coefficient. For membrane proteins, this would correspond to a τ_diff of a few seconds for the larger secretory granules (such as dense core vesicles) but only tens of milliseconds for small synaptic vesicles. Thus, the fusion-to-fission time scale might be critical in preserving the molecular identity of the secretory vesicle.

A second and physiologically compelling question is whether a fuse-pinch-linger retrieval mode might serve to regulate the amount of exocytosis from a given granule. Although the Holroyd and Taraska studies both showed that the fusion events frequently led to only incomplete release of very large proteins (NPY-GFP or GFP-labeled tissue plasminogen activator), recent amperometric studies showed that molecular interference with dynamin function clearly changes the amount of exocytosis of even a small molecule such as catecholamine (8, 9). Thus, regulation of fission pore dynamics might prove important in the physiological regulation of secretion.

The new studies (3, 4) indicate that in addition to a kiss-and-run pathway that is mediated by fusion pore dynamics, a pathway that uses dynamin (pinch-fuse-linger), and therefore is mediated by fission pore dynamics, can also lead to intact retrieval of secretory vesicles. It is unclear at this point how these two pathways are intermingled. In the future, it will be critical to perform high-resolution capacitance measurements that allow for the measurement of fusion pores and membrane addition while simultaneously imaging dynamin, preferably under conditions of both normal and perturbed states of dynamin function.

At synaptic terminals, where the teleological argument for direct and fast retrieval of secretory vesicles is clearer than for chromaffin cells or PC-12 cells, the evidence for the existence of a kiss-and-run pathway is more mixed. Most of the experimental evidence has come from measurements of the relative efficiency of amphipathic compounds with differing membrane dissociation rates (FM 1–43 and related probes) to be released from synaptic terminals during repetitive stimulation. These have been used to argue both for (10) and against (11) the existence of kiss-and-run fusion events at rat hippocampal synapses. In retinal bipolar terminals, where direct measurements of the fusion of single small synaptic vesicles have been possible (12), FM 1–43 appears to destain completely after fusion, implying either that kiss-and-run is not occurring, or that it is in a form that might resemble fuse-pinch-linger and allow the escape of lipids during the fusion-to-fission time window. A kiss-and-run-like mechanism that completely prevents either uptake of FM 1–43 or its loss during a fusion event has also been proposed for hippocampal terminals (13), as well as at Drosophila neuromuscular junctions, where it appears to also be critically dependent on dynamin function (14). The first ever capacitance measurements that could resolve the fusion of single synaptic-like microvesicles at nerve terminals indicated, surprisingly, that transient fusion pores were so small they would limit the useful release of even a small neurotransmitter and occurred in only ≈5% of fusion events (5). Thus, at synapses, the debate remains open as to the molecular nature of synaptic vesicle retrieval mechanisms, as well as their biophysical properties; however, kiss-and-run has been posited to play a role in several forms of synaptic plasticity, including long-term potentiation (15) and depression (16).

In clathrin-mediated endocytosis (such as with the fuse-and-collapse scenario in Fig. 1), the recruitment of dynamin is thought to occur as a natural downstream consequence of an initial recruitment of adaptor complexes and clathrin coats to the cell surface, and the formation of an invaginated coated pit. An important molecular question for the fuse-pinch-linger events reported by Holroyd and Taraska is: what triggered the recruitment of dynamin? Is it simply the appearance of a fusion neck of appropriate diameter that allows the assembly of dynamin into a collar, or is there a molecular trigger that appears immediately on successful fusion? Answers to these questions will bring us much closer to a detailed understanding of the necessarily intimate coupling of exocytosis and endocytosis at synaptic terminals.