Assembly and Comparison of Plasma Membrane SNARE Acceptor Complexes

Abstract

Neuronal exocytotic membrane fusion occurs on a fast timescale and is dependent on interactions between the vesicle SNARE synaptobrevin-2 and the plasma membrane SNAREs syntaxin-1a and SNAP-25 with a 1:1:1 stoichiometry. Reproducing fast fusion rates as observed in cells by reconstitution in vitro has been hindered by the spontaneous assembly of a 2:1 syntaxin-1a:SNAP-25 complex on target membranes that kinetically alters the binding of synaptobrevin-2. Previously, an artificial SNARE acceptor complex consisting of 1:1:1 syntaxin-1a(residues 183–288):SNAP-25:syb(residues 49–96) was found to greatly accelerate the rates of lipid mixing of reconstituted target and vesicle SNARE proteoliposomes. Here we present two (to our knowledge) new procedures to assemble membrane-bound 1:1 SNARE acceptor complexes that produce fast and efficient fusion without the need of the syb(49–96) peptide. In the first procedure, syntaxin-1a is purified in a strictly monomeric form and subsequently assembled with SNAP-25 in detergent with the correct 1:1 stoichiometry. In the second procedure, monomeric syntaxin-1a and dodecylated (d-)SNAP-25 are separately reconstituted into proteoliposomes and subsequently assembled in the plane of merged target lipid bilayers. Examining single particle fusion between synaptobrevin-2 proteoliposomes and planar-supported bilayers containing the two different SNARE acceptor complexes revealed similar fast rates of fusion. Changing the stoichiometry of syntaxin-1a and d-SNAP-25 in the target bilayer had significant effects on docking, but little effect on the rates of synaptobrevin-2 proteoliposome fusion.

Introduction

The plasma membrane SNARE proteins syntaxin-1a and SNAP-25 can form unproductive 2:1 complexes that slow down the rate of synaptobrevin-2 binding (1) and account for the slow rates of traditional SNARE-mediated in vitro fusion assays (2, 3). In vitro fusion can be greatly accelerated by preventing the formation of a 2:1 complex by utilizing a 1:1:1 acceptor complex of syntaxin-1a, SNAP-25, and a short synaptobrevin-2 peptide (residues 49–96) that is termed the “ΔN complex” (4). Full-length synaptobrevin-2 readily binds this complex, replacing the shorter peptide. Upon using this acceptor complex in single liposome and single vesicle assays, average fusion times of 20 ms and faster have been observed (5, 6).

The ΔN complex has been a useful tool in studying effects of membrane composition and curvature on fusion (7, 8, 9, 10), but understanding how fast fusion, as observed in cells, can be reproduced without resorting to artificial SNARE complex components is of great interest. Traditionally, syntaxin-1a is purified in βOG (octyl-β-D-glucoside) (2, 11, 12, 13, 14) or sodium cholate (3, 4, 9), and allowed to assemble into an acceptor complex with soluble SNAP-25a during coexpression in Escherichia coli or postexpression in cholate, CHAPS (3-[(3-cholamidopropyl)dimethylammonio]-1-propanesulfonate), or βOG. Depending on input stoichiometries and other conditions, this results, to different degrees, in the formation of a 2:1 syntaxin-1a:SNAP-25a complex that does not readily bind to synaptobrevin-2 (4, 5). We found that the oligomeric state of syntaxin-1a is dependent on the purification and assembly detergent and that an exclusively monomeric syntaxin-1a can be produced in DPC (dodecylphosphocholine) (15). Here we present two (to our knowledge) new procedures to form SNARE acceptor complexes capable of fast in vitro fusion. In the first procedure, we mix monomeric syntaxin-1a with soluble SNAP-25 and purify the assembled SNARE acceptor complex by anion exchange in DPC. In the second procedure, we reconstitute monomeric syntaxin-1a into proteoliposomes and mix them to form a planar-supported bilayer with liposomes containing quadrupally dodecylated SNAP-25, which then spontaneously assemble in the plane of the membrane to form a fusion-competent SNARE acceptor complex.

Materials and Methods

See the Supporting Material.

Results

Syntaxin-1a (residues 183–288) was prepared in DPC as previously described in Liang et al. (15), and combined with Cys-less soluble SNAP-25 as described in the Supporting Material. Alternatively, we expressed wild-type SNAP-25 and used alkylation with dodecyl chains to the four native cysteines as described in the Supporting Material. This form of SNAP-25 (d-SNAP-25) closely resembles palmitoylated SNAP-25, which is the predominant form in neurons or SNAP-25-expressing insect cells (16, 17). SNARE acceptor complexes were formed from these products by assembly in DPC or in the plane of supported bilayers, respectively, and their activities were compared with SNARE acceptor complexes prepared, with or without the syb49–96 peptide, in more traditional fashion. Bulk ensemble lipid mixing assays show that 1:1 SNARE acceptor complexes prepared in DPC or with lipidated SNAP-25 in proteoliposomes fused almost as fast as ΔN complex, and much faster than syntaxin:SNAP-25 complexes prepared in CHAPS (Fig. S1). These bulk assays report the overall kinetics over docking and fusion without distinguishing between them. To separate the docking and fusion steps of the overall fusion reaction, we reconstituted the different SNARE acceptor complexes into planar-supported bilayers. As expected from previous studies, the ΔN complex showed a much higher docking efficiency than the syntaxin:SNAP-25(CHAPS) complex (4, 5), while the syntaxin:SNAP-25(DPC) complex purified in DPC docked synaptobrevin-2 vesicles docked almost as efficiently as the ΔN complex (Fig. 1 b; Fig. S2). The reasons for the different docking results are that in CHAPS, a predominantly 2:1 complex is formed (4), whereas in DPC, sytaxin-1a tends to be monodispersed (15) (Fig. S3), which results in a predominantly 1:1 complex (Fig. S4). When proteoliposomes containing DPC-purified syntaxin-1a and d-SNAP-25, respectively, were combined to form acceptor complexes by coassembly in the supported bilayer, the docking efficiency increased with increasing amounts of d-SNAP-25, reflecting the increased probability of forming active 1:1 SNARE acceptor complexes (Fig. 1 b).

Figure 1.

Interactions between synaptobrevin-2 proteoliposomes and planar-supported bilayers with different SNARE acceptor complexes. (a) Schematic diagram of preassembled ΔN complex (left), DPC-preassembled 1:1 syx:SN25 complex (middle), and bilayer-assembled syx:d-SN25 complex (right) (adapted from Liang et al. (15)). (b) Docking of synaptobrevin-2 proteoliposomes to ΔN complex, syntaxin-1a:SNAP-25 complex assembled in DPC and CHAPS (left), syntaxin-1a assembled with d-SNAP-25 at indicated ratios (middle), and syntaxin-1a and d-SNAP-25 only (right). (c) Fusion efficiencies (% fused of all docked) of synaptobrevin-2 proteoliposomes with different SNARE complexes. (d and e) Fusion kinetics of synaptobrevin-2 proteoliposomes with colors as indicated in bar graphs. The kinetic curves are cumulative distribution functions derived from hundreds of single vesicle fusion events. The statistics are given in Table S1. To see this figure in color, go online.

Fusion of single synaptobrevin-2 proteoliposomes to the different SNARE acceptor complex-containing supported bilayers was examined. Because docking to CHAPS-assembled complexes and single reconstituted SNAREs was so low, potential fusion events were not analyzed in these cases. Representative traces of hundreds of analyzed fusion events under each active complex condition (Table S1) are shown in Fig. S5. Under the conditions of this study, predominantly docking and full fusion events were observed, and hemi-fusion and two-step hemi-to-full fusion events (10) were much less frequent (<2%). Once docked, the fusion probabilities of synaptobrevin-2 proteoliposomes were 30–40%, independent of how the SNARE acceptor complexes were prepared (Fig. 1 c). The delay time between docking and fusion was measured for each individual fusion event. The cumulative distributions normalized to the fusion probability for each SNARE acceptor complex are shown in Fig. 1, d and e. The fusion kinetics of all events that fused within 5 s are very similar for the ΔN complex, the DPC-assembled syntaxin-1a:SNAP-25 acceptor complex, and the supported bilayer-assembled 1:1 syntaxin-1a:d-SNAP-25 complex (Fig. 1 d). As described in Domanska et al. (5), the initial fast component of the fusion kinetics was examined and fit with a parallel-activation mixed fusion-site model. Similar numbers of activation steps (∼7 ± 1) were found for each complex, and the fast rates of fusion ranged between 60 and 150 s⁻¹ for all investigated complexes (5).

Discussion

Millisecond timescale fusion of synaptobrevin-2 proteoliposomes was observed with SNARE acceptor complexes that no longer require a nonphysiological peptide to ensure stoichiometric 1:1 syntaxin1a:SNAP-25 acceptor complex formation. The efficiency and rate of fusion of the 1:1 syntaxin-1a:SNAP-25 and the 1:1 syntaxin-1a:d-SNAP-25 complexes closely match those of the previously employed ΔN acceptor complex. The use of DPC as the purifying and reconstitution detergent for syntaxin-1a appears to be the key ingredient to maintain the monomeric form of syntaxin-1a, presumably because of its tight association with DPC micelles (15). This form of syntaxin-1a can be assembled either with a soluble form of SNAP-25 in DPC or with a lipidated form of SNAP-25 in lipid bilayers. Although SNAP-25 is multipally palmitoylated in eukaryotic cells (16), this posttranslational modification has previously not usually been included in attempts to reconstitute SNARE-mediated fusion in vitro (12). The exact number of palmitates that are attached to each SNAP-25 may be less than four, and may vary depending on physiological conditions; our procedure quantitatively attaches four dodecyl chains, which, however, are shorter by four carbons than the native 16-carbon palmitates.

The similarity of the fusion efficiencies and kinetics of all fusion-competent complexes that were investigated in this study suggests that none of the preparations were limited by a minimal concentration of SNAREs in the membrane that is necessary for fusion (9). The results also suggest that fusion operates at the intrinsic efficiency of this particular SNARE system, and that this is not controlled by any accessory proteins or calcium. The most significant differences between different SNARE acceptor complex preparations are observed in their different docking probabilities. These likely reflect the concentration of active SNARE complexes with a 1:1 syntaxin-1a:SNAP-25 stoichiometry and support the notion that previous reports of limited efficiencies of overall bulk fusion reactions are the result of inactive oligomeric syntaxin-1a products that are produced by the most common previous SNARE reconstitution procedures. The reported methodological advances on SNARE acceptor complex preparation with the highest possible activities should have a major impact on future reconstitution studies that are aimed at illuminating the roles of effector proteins and calcium that ultimately control the activity of SNAREs in neurons.

Author Contributions

All authors designed the research, analyzed data, and wrote or edited the article; A.J.B.K. and B.L. performed the research; and V.K. contributed analytical tools.

Editor: Joseph Falke.

Supporting Citations

References (18, 19, 20, 21, 22, 23, 24, 25) are found in the Supporting Material.