Recombinant v- and t-SNARE proteins reconstituted into separate lipid bilayer vesicles assemble into SNAREpins-SNARE complexes linking two membranes. This leads to spontaneous fusion of the docked membranes at physiological temperature. Docked unfused intermediates can accumulate at lower temperatures and can fuse when brought to physiological temperature. A supply of unassembled v- and t-SNAREs is needed for these intermediates to form, but not for the fusion that follows. These data imply that SNAREpins are the minimal machinery for cellular membrane fusion.

Establishment and maintenance of proper architecture is essential for endoplasmic reticulum (ER) function. Homotypic membrane fusion is required for ER biogenesis and maintenance, and has been shown to depend on GTP hydrolysis. Here we demonstrate that Drosophila Atlastin--the fly homologue of the mammalian GTPase atlastin 1 involved in hereditary spastic paraplegia--localizes on ER membranes and that its loss causes ER fragmentation. Drosophila Atlastin embedded in distinct membranes has the ability to form trans-oligomeric complexes and its overexpression induces enlargement of ER profiles, consistent with excessive fusion of ER membranes. In vitro experiments confirm that Atlastin autonomously drives membrane fusion in a GTP-dependent fashion. In contrast, GTPase-deficient Atlastin is inactive, unable to form trans-oligomeric complexes owing to failure to self-associate, and incapable of promoting fusion in vitro. These results demonstrate that Atlastin mediates membrane tethering and fusion and strongly suggest that it is the GTPase activity that is required for ER homotypic fusion.

Membrane-enveloped vesicles travel among the compartments of the cytoplasm of eukaryotic cells, delivering their specific cargo to programmed locations by membrane fusion. The pairing of vesicle v-SNAREs (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) with target membrane t-SNAREs has a central role in intracellular membrane fusion. We have tested all of the potential v-SNAREs encoded in the yeast genome for their capacity to trigger fusion by partnering with t-SNAREs that mark the Golgi, the vacuole and the plasma membrane. Here we find that, to a marked degree, the pattern of membrane flow in the cell is encoded and recapitulated by its isolated SNARE proteins, as predicted by the SNARE hypothesis.

SUMMARY Macroautophagy mediates the degradation of long-lived proteins and organelles via the de novo formation of double-membrane autophagosomes that sequester cytoplasm and deliver it to the vacuole/lysosome; however, relatively little is known about autophagosome biogenesis. Atg8, a phosphatidylethanolamine-conjugated protein, was previously proposed to function in autophagosome membrane expansion, based on the observation that it mediates liposome tethering and hemifusion in vitro. We show here that with physiological concentrations of phosphatidylethanolamine, Atg8 does not act as a fusogen. Rather, we provide evidence for the involvement of exocytic Q/t-SNAREs in autophagosome formation, acting in the recruitment of key autophagy components to the site of autophagosome formation, and in regulating the organization of Atg9 into tubulovesicular clusters. Additionally, we found that the endosomal Q/t-SNARE Tlg2 and the R/v-SNAREs Sec22 and Ykt6 interact with Sso1-Sec9, and are required for normal Atg9 transport. Thus, multiple SNARE-mediated fusion events are likely to be involved in autophagosome biogenesis.

. We find that the isolated core of a SNARE complex efficiently fuses artificial bilayers and does so faster than full length SNAREs. Unexpectedly, a dramatic increase in speed results from removal of the N-terminal domain of the t-SNARE syntaxin, which does not affect the rate of assembly of v-t SNARES. In the absence of this negative regulatory domain, the half-time for fusion of an entire population of lipid vesicles by isolated SNARE cores (Ϸ10 min) is compatible with the kinetics of fusion in many cell types. Many genetic and biochemical experiments have implicated SNAREs in the overall process of membrane fusion (1-6). The finding that isolated SNARE proteins can efficiently fuse lipid bilayers directly establishes that they are the basic machinery that merges membranes (7), a conclusion recently confirmed in an elegant study using permeabilized cells (8) and by the demonstration of contents mixing during SNARE-dependent fusion [see the accompanying paper by Nickel et al. (9)]. This principle is underscored by the internal architecture of the SNARE complex, whose ''core'' consists of four parallel ␣-helices packed into a single bundle with their membrane anchors emerging together from one end of the assembled SNARE complex (10-14).The structure of the cytoplasmic core domain of a SNARE complex is reminiscent of the structure of the extracellularly localized core region of many viral fusion proteins in what is thought to correspond to a postfusogenic conformation (15,16). This suggests a general principle for membrane fusion, in which pin-like structures bridging two membranes promote their fusion. However, no core complex, whether cellular or viral, has actually been shown to be fusogenic. Here, we report such evidence with a core domain of SNARE proteins. Materials and MethodsPlasmid Construction, Protein Expression, and Purification. The following v-SNARE and t-SNARE complexes were bacterially expressed and purified by nickel affinity chromatography: VA MP2His 6 ; syntaxin1͞His 6 SNA P-25 (SYN͞SNA P-25); thrombin-cleavable syntaxin1͞His 6 SNAP-25 (tcSYN͞SNAP-25); syntax in1͞thrombin-cleavableHis 6 SNA P-25 (SYN͞ tcSNAP-25); and thrombin-cleavable syntaxin1͞thrombin-cleavableHis 6 SNAP-25 (tcSYN͞tcSNAP-25). For detailed information, see the supplemental material on the PNAS web site, www.pnas.org.Protein Reconstitution into Liposomes and Thrombin Cleavage of t-SNARE Liposomes. Both VAMP2 and t-SNARE complexes were reconstituted into liposomes as described (7). Once proteoliposomes were harvested, tcSYN͞SNAP-25 liposomes were first treated with human thrombin (Sigma, catalogue number T-1063) at 0.02 units͞l for 2 hours at room temperature and subsequently were inhibited with 2 mM 4-(2-aminoethyl)benzenesulfonyl f luoride (Calbiochem). SYN͞tcSNA P-25, tcSYN͞ tcSNAP-25, or SYN͞SNAP-25 containing proteoliposomes were treated with 0.04 units͞l thrombin (Sigma) for 4 hours at 37°C and subsequently were inhibited with 2 mM 4-(2-aminoethyl)benzenesulfonyl fluoride. As a control, thrombin was preinactivated by fir...