Communication between cells relies on hormone release from secretory granules, but how these vesicles fuse with cell membranes is unclear. An imaging study provides in vivo evidence for a stable intermediate fusion step. See Letter p.548
Cells communicate with each other by secreting small messenger molecules such as hormones or growth factors, many of which are stored in vesicles called secretory granules. To release these messengers to the cell exterior, secretory granules fuse with the cell membrane through a process called exocytosis. On page 548, Zhao et al.1 show that exocytosis occurs through the reversible formation of a hemi-fused intermediate, in which only one of the two leaflets (lipid layers) of the cell membrane has merged with the secretory granule's membrane. These results answer the long-standing question of whether membrane fusion involves a hemi-fused intermediate and also provide in vivo evidence for the 'kiss-and-run' model of secretory-granule exocytosis.
Examples of hormones that are released by secretory granules include insulin in pancreatic β-cells and adrenaline, found mainly in chromaffin cells of the adrenal glands. Hormone secretion requires the granules to dock with the cell membrane and then partially or completely merge with it2,3, allowing hormone release into the extracellular space. However, the regulatory mechanism that underlies this type of membrane fusion3, and the nature of the intermediates involved, have long been debated4,5.
Zhao et al. used a high-resolution optical-microscopy approach6 to observe membrane fusion directly in living chromaffin cells in real time and in three dimensions. To trigger secretory-granule release, the chromaffin cells were electrically stimulated while being bathed in a cell-impermeable fluorescent marker dye. The granule's lipid layer maintains a tight seal, so that the dye can gain access to the interior of a docked secretory granule only when full fusion of the granule and cell membrane occurs. Use of this dye, along with simultaneous monitoring of the cytoplasmic leaflet of the cell membrane using a lipid-bound fluorescent protein, enabled the authors to analyse the membrane changes that occur during fusion.
As the granules begin to fuse, diffusion of the fluorescent marker protein from the inner leaflet of the cell membrane to the outer leaflet of the granule membrane results in an increase in fluorescence in the granule that serves as a reporter for changes in the fusing membranes. In more than half of all fusion events, the authors observed a simultaneous influx of fluorescent dye into the secretory-granule lumen and rise in membrane fluorescence on the docked secretory granule. This type of fluorescence change would be expected if secretory granules fuse completely with the cell membrane.
The authors also frequently observed events in which a rise in fluorescence of the cell-membrane marker on the secretory granules preceded dye influx by several seconds, or in which no dye influx into the secretory granule was detected during 40 seconds of observation. It was surprising that so many such events were detectable and that they are stable over many seconds. From these and further control experiments, the authors concluded that these events must correspond to hemi-fusion, thought to be a metastable state in which the outer leaflet of the secretory granule and the cytoplasm-facing leaflet of the cell membrane have merged, while the inner leaflet of the secretory granule and the extracellular-facing leaflet of the cell membrane remain separate4.
What might be the advantages of this type of fusion mechanism? Unlike neuronal synaptic vesicles, which release neurotransmitter molecules in an all-or-nothing, 'quantal' fashion, secretory granules can partially secrete their hormone content. Such partial release has been postulated to be mediated by kiss-and-run exocytosis, which is a model for how secretory granules open and close a fusion pore through which molecules pass between the granule and the cell membrane2. Whether the fusion pore is made of lipid or protein, or both, is not known. A stable hemi-fused intermediate might indicate the existence of a reversible fusion process that would enable partial release during secretory-granule exocytosis, and might underlie fusion-pore opening and closing2.
An identical structure to a hemi-fused intermediate could also arise if a fully fused granule underwent fusion-pore closure through fission (the splitting of a membrane into two separate entities). In this context, it would be called a hemi-fission intermediate. To probe whether fusion-pore opening and closing are reversible processes that proceed through a common intermediate, Zhao et al. tracked the movement of the fluorescent dye and membrane markers over time. They frequently observed fluorescence dynamics consistent with closure of the fusion pore through a hemi-fission state.
The authors then investigated the role of dynamin, a protein involved in endocytosis — a cell-membrane-dependent process in which materials are transported into cells. Dynamin can bind7 to narrow lipid 'necks' at places where membrane pinching occurs, and Zhao et al. found that its depletion or inhibition tipped the balance towards full fusion of both layers of the granule and cell membranes at the expense of hemi-fusion events. Conversely, the hemi-fused state seemed to be stabilized by a high influx of calcium into the cytoplasm. Overall, the authors' data indicate that secretory-granule fusion and fission are reversible processes, at least in chromaffin cells, with the transition from hemi-fusion to full fusion being counteracted by dynamin and regulated by cytoplasmic calcium (Fig. 1). Consistent with this model, Zhao et al. occasionally observed reversible opening, closing and reopening of fusion pores in the same docked secretory granules.
Although hemi-fusion has previously been observed and characterized in reconstituted systems in vitro8,9,10, Zhao and colleagues' work is the first demonstration of this process in living cells. The new results indicate that this intermediate fusion state is a physiologically relevant and surprisingly stable intermediate en route to the exocytic release of hormones and related molecules. It has been suggested that hemi-fusion underlies the fusion of yeast membrane-bound structures called vacuoles11, and that it is also probably responsible for the delayed fusion pathway in reconstituted vesicles in vitro8. However, the authors' model of exocytic membrane fusion is difficult to reconcile with the idea that the fusion pore is lined with transmembrane proteins — as has been postulated from mutational analysis of the transmembrane segments of key exocytic proteins5,12 — because transmembrane proteins span both layers of the membrane and therefore would be excluded from the centre of the hemi-fused intermediate.
Whether a mechanism that involves hemi-fused intermediates operates in neurons to release neurotransmitters also remains an open question. However, the direct visualization of membrane fusion during neurotransmission presents special challenges, such as the speed of exocytosis, the small size of synaptic vesicles and the complex architecture of neurons in the brain. Finally, the observation that dynamin regulates the partitioning between hemi- and full fusion or fission events lends further support to the idea that membrane fission during endocytosis and other vesicle-budding events proceeds through hemi-fission intermediates13.