SNAREs are a superfamily of small membrane proteins catalyzing membrane fusion (1,2). They share as a common feature a conserved stretch of 60-70 amino acids, called the SNARE-motif. Most SNARE-motifs are membrane anchored by a C-terminal transmembrane region (TMR), and a few SNAREs possess two SNAREmotifs linked by a region containing a palmitoylated cysteine cluster for membrane attachment.Each fusion step requires a specific set of three to four SNAREs providing four SNARE-motifs. For instance, neuronal exocytosis is driven by formation of a complex between SNAP25 (containing two SNARE-motifs) and syntaxin 1A at the plasma membrane and vesicle associated synaptobrevin 2.The fusion reaction results in a ternary cis-SNARE complex in which the four SNARE-motifs adopt alpha-helical conformation, forming a parallel, twisted bundle stabilized by 16 layers of interaction (3). All layers are hydrophobic with the exception of a highly conserved central ionic layer to which each of the SNARE-motifs contributes either one out of three glutamines (consecutively named Q a , Q b and Q c ) or an arginine (R). According to their contributed amino acids, the SNAREs have been classified as Q a -(syntaxin), Q b -and Q c -(N-and C-terminal SNARE-motifs of SNAP25, respectively) and R-SNARE (synaptobrevin 2) (4). Retrospectively, the neuronal SNARE complex laid the ground for the structural Q/R-SNARE classification of all SNAREs discovered in the animal kingdom, fungi or plants.Although the post-fusion SNARE complex is well characterized, it is still debated by what sequence of interactions it assembles (5). Because Q-SNAREs form complexes but the R-SNARE-motif does not interact with any of the individual Q-SNAREs [for exception see e.g. (6)], it is generally believed that initially a Q-SNARE complex forms, which then functions as acceptor for the R-SNARE.However, in vitro soluble Q-SNARE-motifs (lacking their membrane anchor) do not form an intuitively expected stable Q a Q b Q c or Q x Q y complex, but have the intrinsic property to assemble into several four-helix bundles [e.g. (Q a ) 2 Q b Q c (7), (Q a ) 2 (Q b ) 2 (8) and (Q a ) 4 (8,9); for a SNAP23 (Q b ) 4 bundle see (10)].More recently, several groups accomplished to prevent the stabilization of Q a Q b Q c complexes into (Q a ) 2 Q b Q c complexes by different approaches. For instance, a Q a Q b Q c complex forms when SNAP25 is immobilized on sepharose beads by anti-SNAP25 antibodies (11), a C-terminal synaptobrevin fragment is used for stabilization (12), SNAP25 is available in large excess over syntaxin (12), or complexes are prepared in supported lipid bilayers containing syntaxin at a very low protein to lipid ratio (13).The latter study even allowed the characterization of single complexes and found several states. In addition to the Q a Q b Q c complex, other complexes were also observed with one of the SNAP25 SNARE-motifs dissociated (in principle representing a Q a Q b or Q a Q c complex). 394 www.traffic.dk