How Could SNARE Proteins Open a Fusion Pore?

Abstract

The SNARE (Soluble NSF Attachment protein REceptor) complex, which in mammalian neurosecretory cells is composed of the proteins synaptobrevin 2 (also called VAMP2), syntaxin, and SNAP-25, plays a key role in vesicle fusion. In this review, we discuss the hypothesis that, in neurosecretory cells, fusion pore formation is directly accomplished by a conformational change in the SNARE complex via movement of the transmembrane domains.

The SNARE Complex

The mechanisms of vesicle fusion are of ubiquitous importance in cell biology and neuroscience. Such fusion events mediate intracellular trafficking, viral infection, and exocytotic secretion of neurotransmitters, hormones, and many other mediators from a wide variety of cell types. This release occurs from the interior of the secretory vesicle to the outside of the cell via formation of a fusion pore. The mechanism by which this fusion pore is formed is still very poorly understood. The neuronal soluble NSF attachment protein receptor (SNARE) complex (49) has three components. The v-SNARE synaptobrevin 2 (syb2 also called VAMP2) is a 116-amino acid protein anchored in the vesicle membrane by a single transmembrane (TM) domain. Syntaxin (stx) is correspondingly anchored in the plasma membrane via a single TM helix. The third component, SNAP-25, has lipid anchors in the plasma membrane. SNAP-25 and stx are called t-SNAREs, being in the target membrane for fusion of secretory vesicles. The key importance of these proteins in the fusion mechanism has been demonstrated by the finding that proteolytic cleavage of the SNARE proteins by specific neurotoxins results in strong inhibition of transmitter release in neurons as well as in chromaffin cells (42, 44). One example for the medical relevance of the SNARE complex is the BoTox treatment, which exerts its function through inhibition of transmitter release by specific cleavage of the SNARE protein SNAP-25.

The cytosolic so-called SNARE domains of these three proteins form a coiled coil, which incorporates one helix each from syb2 and stx and two helices (named SN1 and SN2) from SNAP-25. The coiled coil structure of the SNARE domains has been solved by X-ray crystallography several years ago (57). Based on this structure, the configuration shown in FIGURE 1A was proposed for a trans configuration, in which a SNARE complex links the vesicle membrane and plasma membrane. The arrows indicate the cleavage sites for the neurotoxins mentioned above. More recently, a crystal structure of the SNARE complex, including the syb2 and stx TM helices, was solved (56), which shows helical extension from the SNARE domains through the linkers into the TM domains (FIGURE 1B). This structure is thought to resemble a post-fusion state in which all components of the SNARE complex are in a cis configuration, located in the same fused membrane.

FIGURE 1.

Structure of the SNARE complex

A: hypothetical model for the structure of the trans SNARE complex as suggested by Ref. 57. The TM domains (yellow) and the SNAP-25 linker (brown) were not part of the crystal structure. B: structure of the post fusion cis SNARE complex, including the TM domains of syb2 and stx, shows helical extension of the SNARE domains into the membrane (56). Figure adapted from Refs. 56 and 57 and used with permission from Nature.

Synaptic vesicles contain ∼70 copies of syb2 (59), but it is unknown and controversial how many of these copies are acting together in the formation of a fusion pore (43). Interestingly, neurosecretory vesicles in PC 12 cells recruit t-SNARE clusters containing a similar number of stx and SNAP-25 molecules (26). It has been shown that in reconstituted systems a single SNARE complex is sufficient to promote lipid vesicle fusion (50, 62). However, at least three SNARE complexes appear to be required for fast membrane fusion kinetics (8, 17, 41, 50). In cultured hippocampal neurons, two copies of syb2 were found to be sufficient for fusion, but the experiments were performed with low time resolution (52).

The Triggering Mechanism

Transmitter release from neurosecretory vesicles is triggered by intracellular Ca²⁺ ions, and vesicles in the primed state are able to undergo fusion with rapid kinetics in response to stimulation. There is very strong evidence that assembly of the primed state and triggering of rapid fusion are mediated by the proteins synaptotagmin 1 and complexin (12, 37, 47, 53, 60). Synaptotagmin 1 is anchored in the vesicle membrane by a single transmembrane helix, which is connected to a C2A domain through a flexible linker. Another linker connects the C2A domain to its C2B domain (46). The C2 domains bind multiple Ca²⁺ ions and negatively charged lipids, which also contribute to the Ca²⁺ binding (5). Complexin is a small cytosolic protein that interacts with the SNARE complex (39).

It is widely believed that, in the primed state, the SNARE complex is assembled in the NH₂-terminal part of the SNARE domains, whereas COOH terminal zippering is inhibited (55), possibly through an interaction of the SNARE complex with complexin (30, 33). Structural data suggested that complexin may interact with one SNARE complex via its central helix, whereas its accessory helix interacts with the COOH-terminal part of a second SNARE complex, thereby cross-linking two SNARE complexes. It has been suggested that this cross-linked structure clamps fusion, preventing COOH-terminal zippering of the SNARE complex (30). The clamping of fusion by complexin can then be removed in response to Ca²⁺ binding by synaptotagmin (28, 37). The vectorial assembly of the SNARE complex, which proceeds from the NH₂ terminus to the COOH terminus toward the membranes (23, 54, 55, 65), can now generate a force that is transferred to the apposed membranes, leading to fusion pore formation.

Although there is wide consensus that synaptotagmin 1 is the Ca²⁺ sensor for triggering of fusion, the mechanism of synaptotagmin function is controversial. Alternative to a function of disinhibition of COOH-terminal zippering as described above, it has been suggested that synaptotagmin promotes fusion by inducing membrane curvature through the shape of its C2 domains (19, 38). Independent of Ca²⁺, the C2B domain binds to PI(4,5)P₂ that is concentrated in t-SNARE clusters on the plasma membrane through the cationic charges of syntaxin's juxtamembrane domain (3, 18, 24). A recent study suggested that this mechanism may be essential for vesicle docking and that Ca²⁺ binding to the C2B domain may promote simultaneous binding of the C2B domain to phosphatidylserine on the vesicle membrane, pulling the membranes closer together (18). This finding supports the hypothesis that the formation of a trans SNARE complex is enabled only after arrival of the Ca²⁺ stimulus by cross-linking the vesicle and plasma membranes via the C2B domain (22, 63).

The Force Generation and Force Transfer

The energy barrier for fusion of lipid vesicles is very high, and only very rough estimates exist. Due to this energy barrier, fusion of vesicles is an extremely slow process. It has been estimated that energies roughly in the range of ∼15–50 k_BT must be provided by a protein machinery to lower the activation energy such that fusion can occur on a physiological millisecond time scale (14, 27, 51). This energy is likely provided by one or more SNARE complexes and possibly the interaction of synaptotagmin with the membranes. The energy from SNARE complex assembly must be transferred to the membrane via the generation of corresponding forces. The forces that zippering of the SNARE complex could produce have been investigated in a few experiments. Force spectroscopy experiments performed on the SNARE complex yielded rupture forces of ∼250 pN (36), but this value depends on the loading rate. Very recently, optical tweezers were applied to determine the forces and energies of SNARE complex zippering (11). These experiments suggested that zippering of a single SNARE complex may output a free energy of ∼65 k_BT. The contributions from the COOH-terminal half of the SNARE domains and those from the linkers connecting the SNARE domains of syb2 and stx to the TM domains were found to be 28 k_BT and 8 k_BT, respectively, with estimates for the respective maximum force output in the COOH-terminal and linker domain zippering stages of ∼17 pN and ∼12 pN (11). In similar experiments using magnetic tweezers, COOH-terminal unzipping was observed in the same force range (40).

When a force is generated by zippering of the COOH-terminal part of the SNARE complex that pulls the two membranes further together and if this force is doing work to overcome the energy barrier of membrane fusion, then the force transfer must occur via the syb2 and stx TM domains in the vesicle and plasma membrane, generating the structural changes that lead to fusion pore formation. The relation between a force acting on the syb2 TM domain and its position in the membrane has recently been investigated in coarse-grain molecular dynamics simulations (35). From these simulations, the activation energy for tilting the syb2 TM domain through the membrane from its prefusion transmembrane orientation to a position parallel to the membrane plane in the cytoplasmic membrane-water interface was estimated to be ∼27 k_BT. It therefore appears that generation of the energy by COOH-terminal SNARE complex zippering will lead to considerable changes in syb2 TM domain orientation.

The COOH-Terminal Part of the SNARE Complex Determines Fusion Pore Properties

The molecular mechanism of fusion pore formation is still highly controversial. In the lipid-stalk-hemifusion hypothesis, the outer and the inner leaflets of the two membranes merge via formation of a hemifusion intermediate in response to forces exerted by proteins surrounding the fusion site (71). An alternative proteinaceous fusion pore model was proposed where the fusion pore is lined by several TM domains of stx (17) and possibly syb2 (21). However, the TM domains of syb2 and stx are hydrophobic, and it is unclear how an aqueous fusion pore, through which transmitter molecules and inorganic ions permeate by electrodiffusion (13), could be formed between them.

When the COOH-terminal SNARE domain interactions are reduced by mutating or deleting the COOH terminus of SNAP-25, or when flexible linkers are introduced between the syb2 TM domain and its SNARE domain, the rate of exocytosis is reduced (6, 7, 25, 67, 69) and the flux of transmitter through the early fusion pore is decreased (9, 25). Chromaffin cells expressing the SNAP-25 mutant named SNAP-25Δ9 (lacking the last nine COOH-terminal residues) display smaller amperometric “foot-current” currents, reduced fusion pore conductances, and lower fusion pore expansion rates. These results suggested that SNARE complexes involving the SNAP-25Δ9 mutant produce a lower force, are less tightly zipped, and therefore generate structurally different fusion pores. These mutant fusion pores are presumably longer, consistent with the observed decrease of fusion pore conductance (9). Based on these findings, a proteolipidic fusion pore structure was proposed that is formed by a molecular complex of both lipids and SNARE proteins (9).

At the extravesicular end of the syb2 TM domain, there are two tryptophans (W89/W90). When these are mutated to alanines in the W89A/W90A double mutant [also named WA mutant (37)], catecholamine release from chromaffin cells measured by carbon fiber amperometry showed a higher flux of catecholamine through the initial fusion pore as well as through the expanded fusion pore (10). From these results, it may be concluded that fusion pores formed by SNARE complexes involving the WA mutant are initially wider or shorter and subsequently expand more readily. Unexpectedly, the number of catecholamine molecules released per vesicle from syb2 WA-expressing chromaffin cells was considerably increased, indicating increased quantal size, although vesicle size measured by capacitance measurements and total transmitter stored per vesicle were unchanged (10). The larger quantal size from syb2 WA mutant-expressing cells was consistent with complete release of the vesicular contents, suggesting that release mediated by wild-type SNARE complexes may be incomplete with fusion pore closure before the vesicle contents are fully discharged. These results suggest that the TM domain of syb2 may have a direct role in determining early fusion pore structure as well as fusion pore expansion.

The Role of SNARE Protein TM Domains in Fusion

The NH₂-to-COOH terminal SNARE complex zippering produces a force that is transferred to the apposed vesicle and plasma membranes via the syb2 and stx TM domains. Evidence for the significance of the TM domains in SNARE-mediated fusion events as well as viral fusion events has come from many studies (reviewed in Ref. 32). It has been suggested that tilting of the TM domains may be an essential step in facilitating membrane fusion pore formation or fusion pore expansion by inducing changes in membrane curvature (20, 61). In the prefusion state, the COOH terminus of the syb2 TM domain is located in the intravesicular membrane water interface (FIGURE 1A).

The ability of syb2 to support exocytosis is inhibited by addition of one or two residues to the syb2 COOH terminus, depending on their energy of transfer from water to the membrane interface (45). These results suggested that, following stimulation, the force transfer generated by the SNARE complex pulls the COOH terminus of syb2 deeper into the vesicle membrane and that such a movement of the syb2 COOH terminus may be a prerequisite for fusion pore formation (45). Evidence that penetration of the COOH terminus into the hydrophobic membrane core triggers fusion pore formation was also obtained in CG-MD simulations of a SNARE complex bridging two membranes (48). To investigate more directly how such mutations may inhibit the movement of the syb2 TM domain in the membrane, coarse-grain molecular dynamics simulations were performed using a model of a COOH-terminal fragment of syb2 comprised of residues Q71-T116 (35). These simulations revealed that application of piconewton forces to the TM helix in the simulation pulls the syb2 COOH terminus deeper into the membrane. The energy for this COOH-terminal movement was increased in the syb2-KK construct, with two lysines added at the COOH terminus, which could explain the experimentally observed inhibition of fusion.

The two tryptophans (W89/W90) of syb2 located at the cytoplasmic end of the syb2 TM domain have a particularly low energy in the membrane water interface (68) and thereby stabilize the syb2 TM domain position. Accordingly, the WA mutant showed a decrease of the energy for the TM domain movement in CG-MD simulations (35). Experimentally, the WA mutant increased the rate of spontaneous fusion events and reduced evoked release in neurons (37). Increased spontaneous release and decreased stimulated release were also observed in chromaffin cells expressing the WA mutant, suggesting that the tryptophans W89/W90 act as a fusion clamp, making release stimulation dependent (10). It therefore appears that the WA mutant allows fusion to occur spontaneously with higher frequency, leaving fewer vesicles in a primed state available for rapid release in response to stimulation. These results are consistent with the hypothesis that fusion is facilitated by a movement of the syb2 TM domain. In another study using chromaffin cells, a marked reduction in rapid release following stimulation was also observed, but an increase in spontaneous events was not detected (4). This apparent discrepancy is still to be resolved.

SNARE complex zippering will bring about fusion of the membranes in a transition from a trans state (FIGURE 1A) to a cis state (FIGURE 1B). The syb2 TM domain movement during this transition will likely occur in a tilting motion. Such a tilting motion of the syb2 TM domain through the membrane could be produced in CG-MD simulations by application of forces to the COOH-terminal end of the TM domain (FIGURE 2). The activation energy of tilting the TM domain from the transmembrane orientation (FIGURE 2A) to an orientation parallel to the membrane on the membrane-water interface (FIGURE 2C) was estimated to be ∼27 k_BT (FIGURE 2E). Applying a constant 80-pN force to the syb2 COOH terminus lowered the energy barrier to ∼3 k_BT (FIGURE 2E, dashed line), and in the simulation this barrier was overcome in ∼300 ns. Based on this result, the energy barrier that would allow the tilting process to occur on the physiological time scale of fusion (∼1 ms) can be estimated (in units of k_BT) as

$$\Delta G(1\hspace{0.17em}\text{ms)}=\Delta G(300\hspace{0.17em}\mathrm{n}\text{s)}+\ln \left(\frac{1\hspace{0.17em}\text{ms}}{\text{300}\hspace{0.17em}\text{ns}}\right)\sim 11k_{B}T$$

FIGURE 2.

Energetics of syb2 tilting movement in the membrane

Harmonic potentials were applied to S115/T116 (ST group) in state (A), generating directional movement through the hydrophobic core (B) to the extravesicular membrane-water interface (C). D: the change of the ST group position is accompanied by a corresponding change of the TM domain tilt angle. E: the free energy profile shows a maximum when the COOH terminus is located near the membrane center (black dots with error bars). The energy barrier is lowered to ∼3 k_BT when an 80-pN force is applied in z direction (broken line). Image adapted from Ref. 35 and used with permission.

This estimate suggests that the SNARE complex zippering needs to lower the activation energy by ∼16 k_BT to enable the tilting motion of the syb2 TM domain with a time constant of ∼1 ms. A similar estimate may be expected for the stx TM domain. As mentioned before, experiments measuring the forces and energies of SNARE complex zippering using optical tweezers (11) indicate that the COOH-terminal SNARE complex zippering may output a free energy of ∼28 k_BT, which may be close to the value needed to obtain ms kinetics for tilting both the syb2 and the stx TM domains through the membranes.

The Mechanism by Which the Force Transfer Leads to Fusion Pore Formation

Based on the experimental results and computational studies, the hypothesis for the molecular mechanism of SNARE-mediated fusion pore opening illustrated in FIGURE 3 has been proposed (35). In this model, fusion pore formation is induced by movement of the charged syb2 COOH terminus within the membrane in response to pulling and tilting forces generated by COOH-terminal zippering of the SNARE complex. From the prefusion state (state i), the SNARE complex pulls the membranes together while the COOH termini of syb2 and presumably syntaxin maintain contact with the exoplasmic membrane leaflet lipid head groups (state ii). Once the COOH termini are detached, they move toward the endoplasmic leaflets (state iii). Rearrangement of the lipids leading to fusion pore formation could be induced directly in state iii, as suggested by CG-MD simulations (48).

FIGURE 3.

Model of TM domain movements

Model of TM domain movements leading to fusion pore formation (A) and corresponding states of the syb2 TM domain from CG-MD simulations (35). The zippering of the SNARE complex will lead to helical extension, leading to force transfer and TM domain, tilt as indicated in states i–iv. Fusion pore formation (state vi) could proceed via the stalk state (state v) or could be induced via the transition from state iii to iv. Image adapted from Ref. 35 and used with permission.

The model of FIGURE 3 proposes an alternative role for the syb2 TM domain that is reminiscent of the role of the hemagglutinin fusion peptide in viral fusion. The viral fusion peptide becomes exposed in response to a decrease in pH and inserts into the apposed leaflet of the endosomal membrane in an inverted V- or boomerang-shaped conformation but essentially parallel to the membrane with a shallow tilt angle (16). It is thought that formation of a fusion pore proceeds from this state with the fusion peptide located between viral and endosomal membrane (15, 34). In the model of FIGURE 3, syb2 and syntaxin assume an extended state parallel to the membranes in state iv, and the TM domains come into contact between the membranes, thereby linking the two membranes. FIGURE 2 shows the energy landscape for TM domain tilting. If forces from SNARE complex zippering can lower the energy of the parallel state such that the state iv is assumed, this is not a low-energy state for the TM domains. A rearrangement of the lipids around the TM domains that leads to a fusion pore reestablishes a low-energy transmembrane orientation of the TM domains (state vi). In this model, the formation of the fusion pore is thus driven by lowering the energy of the TM domains. It has been shown that synthetic SNARE TM domain peptides can act as fusion peptides and drive liposome fusion depending on their conformational flexibility (31). It seems possible that, in such systems, fusion may be mediated by a small population of peptides that translocate to the vesicle surface and can thereby link and fuse the membranes of two vesicles. Alternatively, from state iv, fusion pore formation may occur via a stalk intermediate (state v). Such an intermediate will be unstable due to penetration of the charged syb2 COOH terminus into the hydrophobic membrane core (48), which could trigger fusion pore formation (state vi).

The hypothesis for the fusion mechanism shown in FIGURE 3 is consistent with a large body of experimental findings. Its validity may be tested by comparing the predictions from molecular dynamics simulations with experimental data that show how specific mutations in the syb2 and stx TM domains, the linker regions, and the COOH-terminal part of the SNARE domains affect SNARE complex force generation, the energy of tilting the TM domain through the membrane, and fusion pore formation. However, more direct methods are needed to determine the precise conformational changes in the SNARE complex that lead to fusion pore formation. One method to probe protein conformational changes utilizes fluorescence resonance energy transfer (FRET). In this approach, a FRET donor and a FRET acceptor are incorporated into the protein. For FRET, the emission spectrum of the donor needs to overlap with the absorption spectrum of the acceptor, and the distance between donor and acceptor must be sufficiently small, on the order of a few nanometers. Under these conditions, excitation of the donor leads to fluorescence emission from the acceptor with reduced emission from the donor. The FRET efficiency changes when the distance and/or relative orientation between donor and acceptor change. The fluorescent proteins YFP and CFP (and their derivatives with similar spectra) form a suitable FRET pair and have been incorporated at the NH₂-terminal ends of the SNAP-25 SNARE domains (2, 58, 66). Such constructs report formation and conformational changes of the SNARE complex.

In a recent study, imaging of FRET changes using such a FRET-based SNARE complex reporter was combined with electrochemical detection of single fusion events at high time resolution. These experiments revealed a rapid conformational change in SNAP-25 that occurred specifically at the sites of fusion events and preceded fusion pore formation by ∼90 ms (70). This FRET change may result from the formation of trans SNARE complexes that form after cross-linking the vesicle and plasma membranes via the synaptotagmin C2B domain, as recently proposed (22, 63). Alternatively, it could reflect structural changes in preformed trans SNARE complexes that occur in response to stimulation. The development of new fluorescence probes that could monitor movements of the TM domains of syb2 and stx could provide more direct evidence regarding their role in fusion pore formation. However, GFP-derived tags would be much too bulky for such experiments. One approach using a much smaller fluorescent tag involves fluorescent labeling of SNAREs with the biarsenical dye FlAsH using the tetracysteine system (1, 64). Recent progress in the incorporation of fluorescent unnatural amino acids in proteins expressed in mammalian cells (29) suggests that this method may be another possible tool to provide more detailed insight into the conformational changes of the SNARE complex that lead to fusion pore formation. These may also be able to provide direct evidence regarding the controversy if a stable, primed trans SNARE complex exists.