Membrane fusion is a pivotal process for eucaryotic cells. It facilitates communication between different cellular compartments and the secretion of diverse molecules such as neurotransmitters and hormones. The underlying cellular machinery mediates the recognition and subsequent attachment of the merging membranes in the course of endocytic and secretory pathways. One of the cellular systems that are thought to be involved in fusion between a vesicle and its target membrane are the SNARE proteins, where vesicle (v-) SNAREs and SNARES in the target membranes (t-SNAREs) interact specifically to form inter-membrane complexes, which subsequently trigger fusion. Although numerous other proteins may be involved in regulating membrane mergers, the SNAREs are currently the best candidates for universal fusion mediators.

In order to fortify the SNARE hypothesis, Jim Rothman and his team tried to obtain the same level of experimental evidence for SNARE-mediated membrane fusion as already exists for viral fusion proteins such as the influenza hemagglutinins. For this purpose they simply asked: Are SNARE proteins, which mediate membrane fusion within cells under natural conditions, sufficient to fuse cells together if they are expressed on the cell's surface?

To answer this question they first had to design a cellular system that allows functional expression of the SNARE proteins on the extracellular surface. SNARE complexes are formed by three subunits that belong to different protein families. The VAMP proteins comprise the v-SNAREs, while syntaxins and SNAP-25s assemble to form the t-SNAREs. During membrane fusion, v- and t-SNAREs associate via their cytoplasm-facing interaction domains, so Rothman's team needed to express the SNARE proteins so that the interaction domains were placed on the extracellular side of the plasma membrane, to get cells to fuse.

The team engineered expression vectors to generate SNARE proteins with an N-terminal signal sequence that directed the proteins to the plasma membrane and exposed the protein's interacting domains on the cell's external surface. They also introduced some genetic mutations into the SNAREs to prevent unwanted post-transcriptional modifications to the protein. When Rothman and his coworkers expressed the proteins in monkey COS cells and determined the protein's orientation by means of antibodies, they found significant amounts of flipped SNAREs on the extracellular surface.

To test whether the flipped SNAREs could mediate cell-cell fusion, the scientists expressed the v-SNARE (VAMP) in one cell population and the t-SNAREs (syntaxin and SNAP-25) in another. The cytoplasm of the v-SNARE expressing cells was labeled with a red fluorescent protein, and the nuclei of the cells producing the t-SNAREs were labelled with a cyan-coloured fluorescent protein. After combining both cell populations and incubating for a while, the team found numerous merged cells, with multiple cyan nuclei surrounded by red cytoplasm. So the v- and t-SNARE expressing cells had fused within minutes. A series of elaborate control experiments left no doubt that the observed cell-cell fusions were strictly dependent on the presence of functional SNAREs.

Until Rothman and colleagues demonstrated SNARE-dependent cell-cell fusion, evidence for the involvement of SNAREs in membrane fusion was lacking at the cellular level. Now there is strong evidence that SNARE core complex formation is sufficient for physiological membrane fusion. Defining the core of the fusion machinery is also a central issue in neurophysiology and will help us to understand how neurotransmitters are released into the synapse.