SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) proteins are a highly conserved set of membrane-associated proteins that mediate intracellular membrane fusion. Cognate SNAREs from two separate membranes zipper to facilitate membrane apposition and fusion. Though the stable post-fusion conformation of SNARE complex has been extensively studied with biochemical and biophysical means, the pathway of SNARE zippering has been elusive. In this review, we describe some recent progress in understanding the pathway of SNARE zippering. We particularly focus on the half-zippered intermediate, which is most likely to serve as the main point of regulation by the auxiliary factors.
Many vital life processes in eukaryotic cells, such as trafficking of proteins or membranes and secretion of hormones or neurotransmitters, require membrane fusion. Intracellular membrane fusion must happen in a specific and regulated manner. For this, highly specialized proteins called ‘fusogens’ mediate the merging of two otherwise stable membranes to a single bilayer. It is now established that widely conserved soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) proteins are the primary fusogen, responsible for nearly all intracellular membrane fusion [1–3].
Since the discovery of SNAREs in late 80s through early 90s significant progress has been made towards understanding the mechanism by which SNAREs drive membrane fusion. Vesicle-anchored SNARE (v-SNARE) associates with the target membrane-anchored SNARE (t-SNARE) to form a SNARE complex. More precisely, individual SNAREs contain SNARE motifs that are essentially coiled coil sequences of 60–70 residues [4,5]. Cognate coiled–coiled interactions between v- and t-SNAREs are the basis for SNARE complex formation [6,7]. The complex is however believed to assemble in multiple steps, each of which may be mechanically-coupled to a membrane remodelling step . Eventually, the SNARE complex ends up a four-stranded coiled coil [9–14]. This post-fusion conformation has been extensively studied with biochemical and biophysical means.
In contrast, the pathway of SNARE complex formation has been elusive. Thus, the mechanistic details of how the SNARE conformational changes are coupled to membrane remodelling steps are poorly understood. Furthermore, the regulatory interventions of auxiliary factors on SNARE complex formation are not well understood (for reviews see ref. [2,15–18]). However, it has long been speculated that SNAREs might zipper, from the membrane distal N-terminal region towards the membrane proximal C-terminal region . Previously, several research groups have made progress in trapping the partially-zippered intermediate, independently using the advanced biophysical methods [20–24]. These discoveries are major because the results shed lights on to the mechanism of SNARE zippering. The outcomes open up exciting possibilities of studying the regulation of SNARE zippering as mechanisms to control intracellular membrane fusion.
In this review, we describe some recent progress in understanding SNARE zippering and the characterization of the half-zippered intermediate. Additionally, although the data are limited at this early stage we discuss how the half-zippered intermediate might interact with the auxiliary factors to regulate vesicle fusion, particularly for Ca2+-triggered membrane fusion at the synapse.
PRELUDE TO SNARE ZIPPERING
One of the best characterized SNARE families is the neuronal one involved in synaptic vesicle fusion, which is required for neurotransmitter release into the synapse cleft. We will focus on the structure and the function of neuronal SNAREs throughout the review unless otherwise noted. In the present study, vesicle-associates membrane protein 2 or synaptobrevin 2 (VAMP2) is v-SNARE whereas syntaxin 1A and synaptosomal-associated protein of 25 kDa (SNAP-25) are two t-SNARE entities (Figure 1). These neuronal SNAREs were first individually identified in the nervous system [25–29] and later identified together as a complex, the soluble NSF-attachment protein (SNAP) receptor . VAMP2 and syntaxin 1A both contain one SNARE motif each connected to the C-terminal single transmembrane (TM) helix by a short linker regions whereas SNAP-25 contains two SNARE motifs and is attached to plasma membrane by lipid anchors [31,32]. The individual SNARE proteins are partially or fully unstructured as monomers [24,33,34] although there have been some debates if the SNARE motif of VAMP2 interacts with the vesicle membrane [35,36].
Primary structures of all three SNAREs in a bar representation, showing different domains
Prior to the interaction with VAMP2, t-SNAREs, syntaxin 1A and SNAP-25 are believed to form a 1:1 binary complex which serves as the receptor for v-SNARE for vesicle docking and fusion . The structure of the 1:1 binary complex of yeast SNAREs was characterized by NMR and it was found that the N-terminal region is well-structured whereas the C-terminal region is frayed . For neuronal SNAREs, syntaxin 1A and SNAP-25 prefer a non-functional 2:1 complex [38,39] in vitro instead of the functional 1:1 complex, making it difficult for the structural characterization. However, based on the structure of the 2:1 complex and others [38,39], one might speculate that the 1:1 complex could form an extended three-stranded coiled coil  although it remains to be verified experimentally.
EPILOGUE TO SNARE ZIPPERING: THE ALL PARALLEL FOUR-STRANDED COILED COIL
Ideally, it would be best if one could follow the process of SNARE zippering in the chasm of two membranes whereas v- and t-SNAREs are anchored to apposing membranes. However, this is a tall order and alternatively, one could study the interaction between soluble recombinant SNARE motifs out of context with membranes although there is a serious caveat with this approach that we will discuss later.
The early EM and the FRET works suggested that syntaxin 1A and VAMP2 align parallel in the SNARE complex, consistent with the general idea that SNARE complex formation would bring about the close apposition of two membranes [41,42]. Later, EPR and X-ray crystallography showed that SNARE motifs assemble as an all parallel four-stranded coiled coil [9,10] (Figure 2). The SNARE core contains 15 layers of interacting hydrophobic side chains, and right at the centre there is a central ionic layer consisting of one arginine (R) residue from VAMP2 and three glutamine (Q) residues from syntaxin 1A and SNAP-25. Accordingly, SNARE motifs are often classified into R, Qa, Qb and Qc types [43,44]. This highly conserved feature appears to play an important role in SNARE zippering (see below). Recently, the X-ray structure of the neuronal SNARE complex that includes the TM regions of both syntaxin 1A and VAMP2 has been determined . The structure showed that both syntaxin 1A and VAMP2 extend their helical structures of SNARE motifs through the TM helices  (Figure 2C). Apparently, these structures are most likely to represent the post-fusion SNARE conformation.
Structures of four-helix bundles
SNARE ZIPPERING AND THE DISCOVERY OF THE HALF-ZIPPERED SNARE INTERMEDIATE
It has long been speculated that the SNARE complex assembles in a zipper-like fashion, proceeding from the N-terminal region towards the C-terminal region, which would progressively narrow the gap between two membranes. Consistently, there is evidence that SNARE complex formation takes place in multiple steps. Firstly, because the SNARE core is stabilized by the hydrophobic layers, the disrupting mutations at the C-terminal hydrophobic layers affect the fast phase of exocytosis in vivo. These mutations result in two-step thermal unfolding in vitro . Secondly, the force compared with distance measurement using a surface force apparatus (SFA) reveals that the SNARE complex assembles through a series of intermediates . Thirdly, a partially-zippered SNARE complex with a frayed C-terminal region was trapped by intercalating a small hydrophobic molecule myricetin into the SNARE core .
More direct characterization of the partially-zippered intermediate in a single molecule level was achieved using high-resolution optical tweezers and also, independently, using magnetic tweezers [21,22]. These experiments were made possible by attaching one handle at the C-terminal end of v-SNARE and the other handle at that of t-SNARE respectively. Optical tweezers reveals that SNARE unzipping proceeds through three distinct stages with two transitions, the first occurring near the juxtamembrane region and the second at the C-terminal half. The half-zippered intermediate could be stabilized by external force and can release ∼36 kBT by transitioning to the fully zippered state . On the other hand, the magnetic tweezers reveals that single SNARE complex can be unzipped with 34 pN force and rezipping is achieved by lowering the force below 11 pN. Here, a half-zippered state could be stably held under the constant force of 11 pN. Thus, the results detail the energy landscape of SNARE zippering . The valuable information from these studies would eventually help correlating the mechanics of SNARE zippering and the energetics of membrane fusion.
Furthermore, some structural details of the partially-zippered SNARE intermediate have been obtained with EPR using a ‘nanodisc sandwich’ . Experimentally, two nanodiscs which bear single VAMP2 and single t-SNARE respectively are prepared and SNARE complex formation is allowed between two nanodiscs, creating a nanodisc sandwich that harbours a single trans-SNARE complex in the middle (Figure 3A). Due to the rigid structure of nanodiscs membrane fusion does not occur, and the transient SNARE intermediate is captured and studied with SDSL (site-directed spin labelling) EPR as well as single molecule FRET (smFRET) . The nitroxide-scanning EPR study shows that an apparent structural hinge for SNARE zippering is located exactly at the 1RQ3 ionic layer, potentially revealing the structural role of this highly conserved feature. Furthermore, smFRET with the acceptor and donor pairs near the C-terminal ends of v- and t-SNAREs respectively show that the half-zippered intermediate is energetically balanced with the fully zippered state, exhibiting two well-defined low FRET and high FRET populations (Figure 3B).
A structural model of the half-zippered intermediate
BLESSINGS AND CURSES OF THE FOUR-STRANDED COILED COIL
The determination of the four-stranded coiled coil structure has been clearly one of the most important blessings towards understanding the mechanism of intracellular membrane fusion. Very interestingly, the SNARE core shares many important structural features with viral fusogens , strongly arguing for the possibility that common biophysical and biochemical principles do exist and are shared by many biological membrane fusion systems, if not all. However, what distinguishes SNARE-dependent membrane fusion from other membrane fusion systems is the sophistication in its regulation.
One of the remarkable features of synaptic membrane fusion is its capacity to synchronize fusion of nearly all vesicles in the readily releasable pool (RRP) to the presynaptic membrane in less than 1 ms upon Ca2+ influx . It is believed that such tight regulation is orchestrated by a series of exquisitely coordinated interactions of auxiliary factors with SNAREs. A major Ca2+-sensor synaptotagmin 1 (syt1), a clamping factor complexin (cpx) and a chaperon Munc18-1 are considered as the major regulatory components for the synchronization. According to a current mechanistic model [49–51], membrane fusion is clamped by cpx prior to the Ca2+ influx. But upon Ca2+ influx, the Ca2+-bound syt1 knocks off the cpx clamp from the SNARE complex, which frees the SNARE complex to be able to drive membrane fusion. Now, the important question is whether we could test and verify this mechanistic model structurally.
An easy access to the coiled coil structure of the SNARE core let us attempt to address this question by examining the impact of auxiliary factors to the SNARE core. But the outcomes of this approach have been confusing at best. For example, the structure of the SNARE core bound to cpx reveals that cpx binds to the surface groove on the coiled coil without the anticipated disruption of the SNARE core structure . Furthermore, when VAMP2 SNARE motif is shortened at the C-terminal region, cpx cross-links neighbouring SNARE cores in a zigzag fashion , which initiated contested debates in the fields for the biological validity of the structure [54,55]. Likewise, two structural models for the syt1–SNARE core interactions paint very different pictures from each other [56,57]. Thus, it still remains to be seen if these structures truly represent the action of syt1 in triggering vesicle fusion.
HALF-ZIPPERED SNARE INTERMEDIATE AS A POTENTIAL TARGET OF REGULATION
A caveat of studying the interaction between the SNARE core and the auxiliary factors is that the four-helix bundle may well be the post-fusion conformation. It is more than likely that auxiliary factors interfere with SNARE zippering at early stages to clamp, decelerate or accelerate zippering . Since the half-zippered SNARE intermediate is now accessible [21–23,58], one could explore the possibility that it is indeed the point of the regulation.
This hypothesis could be tested experimentally. For example, with optical or magnetic tweezers one could ask if auxiliary factors affect the distance compared with force relationship for the SNARE core . Although the experiments appear to be straightforward one intrinsic difficulty with these approaches is the absence of membranes, particularly because it is known that syt1 and cpx both interact with the membrane [60,61].
An alternative but promising approach is to use the nanodisc sandwich that harbours the half-zippered intermediate in the middle . There the half-zippered intermediate is energetically balanced and thus, is in equilibrium with the fully zippered SNARE complex [23,62] (Figure 4A). The conformational changes or the shift of equilibrium induced by the auxiliary proteins may be detected by placing nitroxide probes or fluorescence labels at strategic positions in the SNARE complex. Such efforts are already underway and start to produce some interesting results. For example, Munc18 is shown to stimulate SNARE-dependent membrane fusion . However, the proposition was relied heavily on a simplified in vitro proteoliposome fusion assay . Consistent with this notion, smFRET with the FRET dye pair placed at the C-terminal ends of v- and t-SNARE respectively shows the shift of the equilibrium towards fully zippered complex in expense of the half-zippered intermediate  (Figure 4B).
Munc18-1 stimulates SNARE zippering
For cpx, however, to be consistent with its fusion clamping role we expect that cpx either shifts the equilibrium towards the half-zippered intermediate from the fully zippered complex or it might lock the SNARE intermediate at a yet unknown state that can prevent it from progressing towards the fully zippered state.
Ultimately, structures of the trans-SNAREs complexed with individual auxiliary proteins in the nanodisc sandwich must be determined to fully comprehend the mechanisms. EPR or cryo EM may be useful for these challenging tasks.
SNARE proteins, which are widely conserved from yeast to human, are the core machinery for intracellular membrane fusion. Vesicle-associated v-SNARE associate with target membrane t-SNARE to drive the fusion of two membranes. There is now sufficient evidence that the SNARE complex assembles in a zipper-like fashion, initially at the membrane distal N-terminal region and subsequently at the membrane proximal C-terminal region. Using single molecule manipulation techniques such as optical and magnetic tweezers, the energy landscape of the multistep folding/unfolding transitions of the SNARE complex have been determined in an unprecedented accuracy. Furthermore, it was found, with EPR that SNARE zippering hinges precisely at the conserved ‘zero’ layer. The half-zippered SNARE intermediate can be trapped in trans between two nanodiscs, where the half-zippered state is in equilibrium with the fully zipped state. This conformational trap inside the nanodisc sandwich provides exciting opportunities to investigate the intervention of auxiliary proteins on to SNARE zippering as means of regulating intracellular membrane fusion such as Ca2+-triggered synaptic vesicle fusion.
This work was supported by the National Institutes of Health [grant number R01 GM051290]; the Membrane protein structure and dynamics consortium; and the Roy J. Carver Professorship.
single molecule FRET
synaptosomal-associated protein of 25 kDa
soluble N-ethylmaleimide-sensitive factor attachment protein receptor
target membrane-anchored SNARE
vesicle-associates membrane protein 2