The SNARE hypothesis was first postulated nearly a decade ago as a mechanistic account of membrane fusion. In its earliest form, this hypothesis stated that SNARE proteins present in the acceptor and donor membranes mediate the specificity of the interaction preceding fusion. The hypothesis was later modified to propose that SNARE proteins are actually involved directly in the fusion reaction. At synapses, the main fusion event takes place during transmitter release, as synaptic vesicles fuse with the presynaptic membrane. The SNARE hypothesis was quickly invoked to explain this process, and suggestions of relevant SNARE proteins were rapidly put forward from the growing list of molecules present at synaptic contacts. These developments led to the optimistic view that the intimate details of synaptic exocytosis would soon be unveiled. Now, almost 10 years later, some of our basic questions on this phenomenon remain unanswered, including that of the actual role of SNAREs in fusion. Are SNAREs the molecular components of the fusion pore? Do they act as catalysts of the fusion reaction, or are they only involved in setting up its specificity? Reporting in Science, Schoch et al. argue that SNAREs are not necessary for the fusion step per se, but are required instead for stabilizing fusion intermediates.

Schoch et al. generated mice that lack synaptobrevin 2 — a SNARE protein of synaptic vesicles — and studied the characteristics of transmitter release in neurons cultured from the forebrains of these animals. They found that synaptic vesicle exocytosis was markedly reduced, regardless of whether it was spontaneous or evoked by changes in osmolarity or calcium influx. However, not all forms of exocytosis were equally affected; spontaneous and osmolarity-evoked release were 10-fold lower than in wild-type neurons, whereas calcium-evoked release was reduced by 100-fold. Changes in osmolarity and calcium influx affect the same population of vesicles. So, if synaptobrevin were involved directly in fusion, it would be hard to account for this difference. The authors therefore argue that synaptobrevin is not required for fusion per se, but is necessary to achieve a normal rate of fusion on stimulation. In other words, they propose that SNARE proteins stabilize transition states that exist before vesicles are responsive to calcium, and that although they do not participate in actual fusion, they limit the rate at which fusion can occur. It will now be crucial to perform similar experiments on neurons from mice that lack the synaptobrevin partners SNAP-25 and syntaxin to test whether this intriguing proposal holds for other SNARE proteins.