Organelle docking: R-SNAREs are late.

F usion of cellular compartments is a central aspect of cellular homeostasis and intercellular communication in virtually all eukaryotic cells. Although the general principles of the fusion reaction are beginning to be firmly established, it is also becoming clear that several distinct molecular events need to occur first, before the actual fusion reaction can. Coming from different angles, different research groups have developed different ideas about these upstream steps. A new study in a recent issue of PNAS (1), using reconstituted vacuole fusion in yeast, establishes important elements of consensus. One exciting aspect is that a well-known component of fusion reactions, the R-SNARE, appears to be universally dispensable for the initial docking step. For organelle inheritance in budding yeast, the mother cell projects tubular and vesicular structures into the growing bud, which then fuse to generate a new vacuole in the daughter cell (2). This vacuole fusion reaction has been reproduced in vitro, initially using an isolated vacuole preparation and recently also with purely synthetic components. The vacuole system has been highly instrumental to address the question of whether genes are necessary and sufficient to orchestrate the fusion and upstream reactions, using a combination of genetic studies in intact cells and isolated vacuoles and in vitro fusion in synthetic proteoliposomes. Because vacuole fusion can be arrested experimentally in vitro at different stages, this system has also been exploited to investigate the sequence of events leading up to fusion. This option is particularly valuable given the current state of the field where new proteins are rarely identified, but the way known proteins do the job is a matter of intense debate. The upstream molecular events are referred to as priming, tethering, and docking steps (see ref. 3 for a review). Here, considerable difference of ideas exists among researchers working on different systems, in addition to vacuole fusion, for instance, mammalian endosome fusion (4, 5) and fusion of different secretory vesicles in mammalian cells (6–9). Adding to the confusion is the fact that the nomenclature describing the different steps in different systems evolved to be dissimilar and confusing (see Fig. 1). All of this has precluded the establishment of a universal model for the initial steps upstream of the membrane fusion reaction. The new study by Stroupe et al. (1) now generates more consensus among fusion and upstream reactions in different systems. First, several alreadyestablished conclusions are further strengthened, for instance, the essential role of specific lipids such as sterols and phosphoinositides, the universal role of SNARE proteins, the general configuration of three Q-SNAREs on one side and one R-SNARE on the other together engaging in a four-helical bundle to fuse two membranes (transSNARE bundles; see ref. 10), and the fact that dissociation of SNARE bundles in one membrane (cisSNARE bundles) is actually the first essential step in setting up the fusion reaction. This idea has been initially proposed in the vacuole system (the actions of Sec17p/18p; see ref. 11) and has been long disputed for other systems (expressing the orthologs Nethylmaleimide sensitive factor/ sensitive factor attachment protein), but is now generally accepted (see Fig. 1). Second, more consensus is now also reached on the essential role of a Raslike GTPase, in the case of vacuole fusion Ypt7p. With an improved lipid preparation (the direct method, see ref. 1), both docking and fusion are strictly Ypt7p dependent. With this new preparation, reconstituted vacuole docking/ fusion is now consistent with reconstituted mammalian endosome fusion, probably the best-characterized in vitro fusion reaction to date. In this system, the Ras-like GTPase Rab5 was shown to coordinate the establishment of a specific local lipid milieu and the recruitment of a large protein complex that orchestrates homotypic endosome fusion (5). Third and most importantly, the work by Stroupe et al. (1) strengthens some new and unexpected conclusions about the docking step. Using a proteoliposome clustering assay, they conclude that the R-SNARE Nyv1p is dispensable for vacuole docking, whereas the three Q-SNAREs are required. A very similar conclusion was recently reached for secretory vesicle docking in intact mammalian cells (12). This is an exciting convergence of new findings, which together suggest that establishing/stabilizing a cis3Q-SNARE complex may be a central step for docking to occur (see Fig. 1) and that conditions/factors that favor the existence of such complexes or promote their stability promote docking. Moreover, other factors than the R-SNARE may perform the initial docking reaction and bind to the cis3QSNARE complex. In mammalian secretory vesicles this is the vesicular protein synaptotagmin-1 (12). It would be very interesting to see whether the vacuole SM-protein Vps33p promotes the existence/stability of cis3Q-SNARE complexes in a similar manner as the mammalian SM-protein Munc18-1 does for the mammalian cis3Q-SNARE complex (12) and whether artificial stabilization of the 3Q SNARE complex using a short C-terminal R-SNARE peptide (see ref. 13) might bypass the require-