The energy source that drives vesicle fusion with a target organelle in vivo has been unclear. It emerges that proteins that tether fusing structures together also decrease the energy needed for the final fusion step. See Letter p.634
Cells communicate with each other by releasing and taking up molecular 'messages', which are often proteins or hormones. These signalling components are delivered to the cell membrane through a pathway in which vesicles containing these molecules transit through different organelles by a series of vesicle budding and fusion events. However, many questions remain about the mechanisms driving the regulation and execution of the membrane fusion (in which the lipid bilayers on two separate membranes fuse to form one bilayer) underlying this process. D'Agostino et al.1 reveal on page 634 that proteins that act at an early stage of membrane fusion also have a role in the key final event of the fusion process.
SNAREs are a group of membrane proteins present on the surface of vesicles and organelles. They help to ensure that vesicles fuse with the correct target compartment through a lock-and-key-like process in which α-helical structural regions of SNAREs on different membranes intertwine and zipper together to form a structure called a trans-SNARE complex2. However, SNARE proteins have only a small cytoplasmic domain that is probably too short to bring a vesicle close to the organelle that is its fusion target (although SNARE proteins are sufficient to drive membrane fusion in reconstituted cell-free in vitro systems3). Instead, the process called docking establishes the first physical contact between a vesicle and the target membrane. This step requires a group of proteins called tethers, which can be extended coiled-coil proteins or multi-subunit tether complexes. Tethers determine whether a potential fusion interaction should occur, through recognition of specific proteins found on a vesicle surface, such as Rab proteins or coat proteins4.
After the initial recognition step, tethers bring the vesicle into closer contact with the target membrane until interaction between SNAREs leads to trans-SNARE-complex formation and membrane fusion2. This mechanism applies not only when vesicles fuse with organelles, but also when organelles such as endosomes or vacuoles fuse with each other.
Although this general mechanism is well accepted in the field, there are some unanswered questions. For example, why does the cell have two different types of tether, both of which often function on the same compartment? And how do the tethers bring the membranes of the two fusing compartments closer together before the SNAREs interact with each other? Some previous reports indicated that multi-subunit tether complexes have an active role in SNARE-complex assembly, and potentially have a role in fusion, yet the precise mechanism remained elusive2. Most multi-subunit tether complexes have two business ends: one interacts with Rab proteins and the other interacts with SNAREs through a tether subunit protein that is a member of the Sec1/Munc18 (SM) family. Thus, multi-subunit tether complexes probably act in the initial assembly of a trans-SNARE complex.
The number of trans-SNARE complexes actively engaged can vary substantially depending on the system studied. Between 1 and 15 have been reported5,6,7, and there are models in which SNARES alone can provide the energy for fusion8. The precise number of trans-SNARE complexes needed for fusion remains unclear, and might indeed vary depending on the specific fusion process.
D'Agostino and colleagues investigated fusion between vacuoles in the yeast Saccharomyces cerevisiae to test whether a multi-subunit tether complex called the HOPS complex has a role beyond aiding the docking process. This is an attractive model system for studying fusion because the key components are known, and the process has been reconstituted in vitro in a cell-free system and can be assessed in vivo. The authors provide evidence that SNAREs alone can achieve a state of only partial fusion between the two fusing membranes, known as hemi-fusion, in which only one of the lipid layers of each of the fusing membranes has fused (Fig. 1), and that HOPS is required to fully complete the fusion process.
The authors show that the function of HOPS in this final stage of fusion is independent of its function in the docking process, implying that HOPS needs to interact only with the SNARE complex for this fusion step. Consistent with this model, when the authors tested another multi-subunit tether complex called CORVET, which usually functions in endosomal-organelle fusion, and which differs from HOPS by only two protein subunits, they found that CORVET could also stimulate the late stages of vacuolar membrane fusion. Even more surprisingly, when the authors targeted a large enzyme called phosphofructose kinase that is not known to act in membrane fusion to bind to SNAREs, this could also drive full fusion.
D'Agostino and colleagues also carried out a computational analysis that simulated their biological systems at an atomic level. This analysis supported their experimental observations that the size, rather than the type, of protein that associates with SNAREs is the key factor driving full fusion. This is because a bulky protein can induce changes at the hemi-fusion site through molecular interactions known as steric effects. Their work gives rise to a model in which multi-subunit tether complexes are required not only to promote the initial docking, but also to lower the energy barrier to fusion. They presumably achieve this, at least partly, by increasing the membrane-curvature stress, so that the energy generated by trans-SNARE zippering can enable full fusion to proceed.
Many questions remain, however. In particular, how general is this membrane-fusion mechanism? For example, another multi-subunit tether complex called DSL1 tethers vesicles to the Golgi-complex organelle9, but DSL1 is less than half the molecular weight of HOPS or CORVET. If this is a general mechanism, are additional proteins recruited to the DSL1 complex, or are more DSL1 complexes needed per fusion event? Perhaps the protein NSF might step in10,11, or else membrane fluidity and rigidity might influence the minimal size requirements of the SNARE-binding protein. Moreover, if this is a widespread mechanism, how is the energy barrier for fusion lowered for multi-subunit tether complexes such as the TRAPP complexes, which do not contain the SM proteins needed to bind SNAREs?
Questions also remain about the specificity of this phenomenon. In yeast, CORVET and HOPS contain the same SM protein. So why does CORVET not act to fuse vesicles directly to the vacuole rather than to the endosome, and why does HOPS not drive vesicle fusion to the endosome? Finally, a recent report12 suggests that the final stage of vacuolar fusion is faster in the absence of HOPS. So a different model emerges, in which HOPS function might instead inhibit the final step of membrane fusion. Although many issues remain to be resolved, D'Agostino and colleagues' work provides an interesting and thought-provoking framework in which to investigate membrane fusion.