Transport vesicles continuously shunt molecules around the cell, ensuring delivery to the correct intracellular destination. Essential to this process are the SNARE proteins, which mediate docking and attachment of vesicles with their target membrane. But how, once a vesicle contacts the target membrane, are the repulsive forces between the membranes overcome to allow fusion? Scientists remain divided on this point. Whereas some believe that SNAREs, by bringing membranes into close proximity, are sufficient for this step, others propose a fusion pore model, in which a proteineous channel forms. Now, reporting in Nature, Peters and colleagues reveal the identity of such a channel, and it turns out to be a familiar one.

Calcium and calmodulin can trigger the final step of fusion, but their downstream targets were not known. So, to identify calmodulin targets, Peters and colleagues used label-transfer analysis to isolate nearby factors on vacuolar membranes. Intriguingly, when they then purified these factors, mass spectrometry identified them as components of the V0 integral membrane sector of V-ATPase ? a membrane complex that functions as a proton pump. The relevant calmodulin-binding partner was then found to be the V0 proteolipid.

So can these proteolipids trigger fusion? To test this, the authors monitored their ability to trigger choline release in reconstituted liposomes. And they found, consistent with the channel model, that they triggered release in a calcium/calmodulin-dependent manner.

Membrane docking requires formation of an electrochemical membrane potential, which is mainly generated by the V-ATPase. But if V0 is important for fusion itself then it should have a direct role in fusion, independent of proton pump activity. The authors confirmed this, showing that fusion could occur once they had knocked out proton pump activity.

So where do SNAREs come in? By inhibiting different stages of fusion, the authors showed that trans-SNARE pairing mediates channel formation, but not maintenance. One possibility, then, is that SNAREs concentrate V0 sectors at the site of close membrane contact.

What remains to be seen is how, once formed, these channels allow bilayer mixing. The authors' bets are hedged on a model in which channels expand radially to form an aqueous pore but, as is often the case, more specific analytical tools are needed before we can dig any deeper.