Single-Molecule Measurements of Dissociation Rates and Energy Landscapes of Binary trans SNARE Complexes in Parallel versus Antiparallel Orientation

Abstract

Interactions between synaptobrevin 2 (Sb2) and syntaxin 1A (Sx1A) can be readily isolated and studied with the use of force spectroscopy single-molecule measurements. We studied interactions between Sx1A and Sb2 in two different orientations (parallel and antiparallel) using four different terminus configurations of these proteins. Force-loading experiments indicated that protein pairs in any configuration/orientation are zippered. We measured the extension and force for disassembly of these interactions, calculated the spontaneous dissociation lifetimes, and determined their free energies, enthalpies, and entropies. Although the free energies were very similar for all four configurations (∼28 k_BT (Eyring model) and ∼20 k_BT (Kramers model)), the enthalpy changes of binary Sx1A-Sb2 interactions varied between 24.7 k_BT and 33.1 k_BT. This variation is consistent with the conformation changes that occur during disassembly of the various protein terminus configurations, as verified by alterations in the extension. The parallel interactions appear to be energetically somewhat advantageous over antiparallel configurations/orientation, especially when the N-termini of Sx1A-Sb2 are left to interact freely.

Introduction

The soluble N-ethylmaleimide-sensitive fusion protein attachment protein receptor (SNARE) ternary complex, which is composed of the vesicular protein synaptobrevin 2 (Sb2) and plasma membrane proteins syntaxin 1A (Sx1A) and SNAP25 (synaptosome-associated protein of 25 kDa), is crucial for the process of exocytosis, whereby secretory vesicles fuse to the plasma membrane (1). Sb2 and Sx1A each have one SNARE domain within their cytoplasmic tails, and SNAP25 has two SNARE domains. The ternary complex is a four-helix bundle formed by the assembly of these SNARE domains (2,3). It has been proposed that in the cellular milieu, Sx1A and SNAP25 form a cis complex at the plasma membrane, which then interacts with the vesicle protein synaptobrevin to form the ternary complex, leading to fusion (4,5). The assembled ternary complex has 15 hydrophobic leucine zipper layers sandwiched perpendicularly along its backbone, with a highly conserved ionic zero layer buried at the center of the structure providing registry for the complex (6,7). It is thought that the leucine zipper layers surrounding the zero layer help to seal the ionic layer from water, thereby decreasing the local dielectric constant and enhancing the electrostatic force. The formation of such a complex could release the energy required to overcome the barrier to bilayer lipid membrane fusion. Conversely, if the hydrophobic seal is broken and the ionic layers are exposed to the solvent, the complex will disassemble, as observed in single-molecule experiments by atomic force microscopy (AFM) (8).

The ternary SNARE complex is a very stable structure that is resistant to sodium dodecyl sulfate and cleavage by clostridial neurotoxins, has a high melting temperature, and undergoes much stronger interactions among its three constituents than the corresponding binary interactions between the SNARE proteins (9–12). In recent studies, investigators measured the interaction energies of SNARE proteins using AFM, a surface force apparatus, and isothermal titration calorimetry (13–16).

Förster resonance energy transfer (FRET) experiments have shown that both parallel (both N-termini of Sx1A and Sb2 on the same side of the complex) and antiparallel (N-terminal of Sx1A and Sb2 at opposite ends of the complex, with SNAP25 remaining parallel to Sx1A) ternary complexes can be formed in solution (17). However, these complexes have different stabilities, with the parallel ternary complex being substantially more stable. Consequently, these two orientations of the ternary complex may have different roles in the fusion process. It has been reported that the Sx1A-Sb2 binary trans complex may also lead to fusion under some experimental conditions (18–20). However, the role of the orientation of SNARE proteins in this binary complex has not been studied. Investigators have studied the mechanics of the Sx1A-Sb2 binary interaction in parallel orientation by AFM to obtain the interaction force and extension, spontaneous dissociation lifetime, and energies (8,14,16). Here we extend such measurements, using the single-molecule force spectroscopy mode of AFM, to antiparallel binary trans complexes of Sx1A-Sb2, and compare the results with those obtained from the complex oriented in parallel.

Materials and Methods

Protein preparation

Plasmids encoding for the recombinant proteins used in this study were generated as described previously (8). The plasmids were generously provided by Dr. Edwin R. Chapman (University of Wisconsin, Madison, WI). Briefly, cytoplasmic domains of rat Sb2 (amino acids (aa 1–94) and Sx1A (aa 1–266) were tagged with six histidines (H6) at their C-termini (Sb2H6 and Sx1AH6) or N-termini (H6Sb2 and H6Sx1A). Plasmids encoding for tagged proteins were used in recombinant protein production. Proteins were purified with Ni-NTA agarose beads (Qiagen, Valencia, CA). Protein solution concentrations (0.17 mg/ml for Sb2H6, 0.1–0.2 mg/ml for Sx1AH6, 0.1 mg/ml for H6Sb2, and 0.1 mg/ml for H6Sx1A) were measured with the Bradford reagent (Pierce Biotechnology, Rockford, IL) using bovine serum albumin as a standard. Expression of the recombinant proteins was confirmed by Western blotting probed with specific antibodies against Sb2 (clone 69.1, 1:1000 dilution, catalog No. 104 201; Synaptic Systems, Göttingen, Germany; note that this product has been replaced by the manufacturer with catalog No. 104 211), Sx1 (clone 78.2, 1:10,000 dilution, catalog No. 110 001, Synaptic Systems; or clone HPC-1, 1:1000 dilution, catalog No. S0664, Sigma-Aldrich) and His-tag (1:500 dilution, catalog No. MAB3114; Chemicon). Immunoreactivity of the bands was detected by enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ). All proteins showed single immunoreactive bands with appropriate molecular mass.

Force spectroscopy

Single-molecule AFM experiments were conducted with the use of a Nanoscope E (Digital Instruments, Santa Barbara, CA). Each cantilever was calibrated individually and had typical spring constants of ∼10–20 mN/m. The bending of the cantilever was taken into account in the calculation of the extension (21). The piezoelectric tube extension, including nonlinearities, was calibrated interferometrically for all force-loading rates and temperature set-points used (22). Nickel-coated AFM tips and coverslips were prepared as described in detail elsewhere (8,23). The high affinity, strength, and specificity of coordinative bonding between H6 and Ni²⁺ allow directional deposition of proteins with only their H6 being attached to the stratum, whereas the remaining part of the molecule retains conformational flexibility (8,24). After they were functionalized with recombinant proteins, the tips and coverslips were kept separately submerged in an internal solution (potassium gluconate, 140 mM; sodium chloride, 10 mM; and Hepes, 10 mM; pH 7.35), in a humidified chamber at +4°C (277 K) until they were used in experiments, for up to 36 h. Before experiments were conducted, the glass coverslips were mounted on metal disc AFM sample holders. All experiments were done in a fluid cell filled with an internal solution to maintain the hydration and osmotic properties of the sample. SNARE protein interaction bonds were formed by extending the tip onto the coverslip. Then the tip was retracted from the coverslip until the bond was broken and the cantilever went back to the equilibrium position. The rupture force and extension were recorded from such events. These so-called pull-off experiments were carried at different force-loading rates ranging from 500 pN/s to 70,000 pN/s, and three set-point temperatures (277 K, 287 K, and 297 K) as previously described (14,16). When reporting on interactions between proteins, we list (left to right) the protein deposited on the tip hyphenated by the protein deposited on the coverslip. For example, H6Sx1A-Sb2H6 implies that interactions were recorded with N-terminally bound H6-tagged Sx1A functionalized tips and with C-terminally bound H6-tagged Sb2 on the coverslip. All extension and force measurements are expressed as the mean ± standard error (SE). Lifetime errors were calculated based on the SE of fitting parameters.

Results

We studied the interactions of parallel and antiparallel orientations of Sx1A and Sb2 by single-molecule force spectroscopy (8). Both Sx1A and Sb2 cytoplasmic tails have a six consecutive histidine molecules (H6) tag at either the C- or N-terminus that is used to attach the proteins to the surfaces. In our experiments, we measured interactions between Sx1A and Sb2 of two different orientations using four different terminus configurations of these proteins: We studied parallel orientations using Sx1AH6 (tip)-Sb2H6 (glass) (16) and H6Sx1A-H6Sb2 (Fig. 1, left column) configurations. For the antiparallel orientations, the proteins were paired as Sx1AH6-H6Sb2 (Fig. 1, middle column) and H6Sx1A-Sb2H6 (Fig. 1, right column). First, we coated the AFM tips and glass coverslips with a thin nickel film by thermal evaporation. After allowing Ni to partially oxidize by exposure to air, we applied Sx1A molecules and Sb2 molecules to the tip and coverslip, respectively. The protein molecules were directionally immobilized on the surfaces through the steric interaction between Ni²⁺ and H6 (8,24). Note that here the free end of the molecule varied depending on the terminal H6 position (Fig. 1 A). The presence of proteins on the functionalized tips and coverslips was verified by indirect immunochemistry as described previously (8). After the proteins were attached, interactions were measured at the single-molecule level by AFM. The Sb2-functionalized coverslip was mounted on top of the piezoelectric tube of the AFM, and the Sx1A functionalized microcantilever was mounted on the fluid cell holding the internal solution. A periodic triangular voltage was then applied to the piezoelectric tube through the AFM controller, so that the coverslip moved toward and away from the cantilever.

Figure 1.

Sx1A and Sb2 interactions in parallel and antiparallel orientations. (A) Orientation schemes of two proteins. H6, the six-histidine tag; N, N-terminus; C, C-terminus. Drawing is not to scale. (B) Typical retraction parts of force-distance curves. Note the boxed part of real protein-protein interaction; nonspecific binding is indicated by an arrow. (C and D) Histograms of rupture forces (C) and extensions (D). Average forces and extensions are indicted by arrowheads.

This directional approach of bringing interacting proteins together allows them to form bonds with their free ends first, and assembly then progresses as the proteins are brought in succession into closer proximity. Thus, directional deposition and probing of proteins ensure the formation or appropriate orientations of protein pairs and homogeneity of their interactions (8,14,16,24). When the cantilever was brought into contact with the coverslip, Sx1A and Sb2 molecules would form a bond with a certain probability, as revealed during the retraction of the tip (Fig. 1 B). Hence, this bond would bend the cantilever when the coverslip was retracted away from the cantilever. By recording the deflection of the cantilever as a function of the coverslip position, we were able to obtain the interaction force and extension between the stretched proteins. Representative force-distance curves are shown in Fig. 1 B. These curves correspond to interactions of different combinations and protein pairs with (anti)parallel orientation as sketched in Fig. 1 A. We recorded two types of bindings: a nonspecific binding that always exists even when proteins are absent, as we described previously (8) (Fig. 1 B, left, arrow), and the real protein-protein interaction (Fig. 1 B, left, box). The various configurations showed similar probabilities of interaction: 32.7% (54 of 165 tested for Sx1AH6-Sb2H6 (16)) and 36.4% (60 of 165 tested) for H6Sx1A-H6Sb2, which are in parallel orientation; and 33.3% (55 of 165 tested) for Sx1AH6-H6Sb2 and 33.8% (54 of 160 tested) for H6Sx1A-Sb2H6, which are in antiparallel orientation. The value of the rupture force was recorded from the maximum cantilever deflection before the bond was broken. For example, in the case of H6Sx1A-H6-Sb2 (Fig. 1 A, left), we measured a force of 244 pN and a corresponding extension of 23 nm. Note that here the intermolecular extension is measured as the vertical z-axis movement of the piezoelectric tube subtracted by the deflection of the cantilever. Similarly, we obtained the forces and extensions for antiparallel Sx1A-Sb2 orientations (Fig. 1, C and D). Because of thermal fluctuations, one can never obtain the same rupture force even for the same interacting molecules, and therefore we obtained an average force and extension from the histograms.

From these measurements, we found that the mean interaction forces of the various Sx1A-Sb2 orientations of molecules were not significantly different. For example, when the piezoelectric actuator retracting speed was 1.6 μm/s (corresponding to ∼20 nN/s force-loading rate), the rupture force values were 252 ± 10 pN (Sx1AH6-Sb2H6, n = 54; values obtained from our previous study (16)) and 224 ± 11 pN (H6Sx1A and H6Sb2, n = 60; Fig. 1 C, left) for parallel orientations; and 252 ± 14 pN (Sx1AH6-H6Sb2, n = 55; Fig. 1 C, middle) and 226 ± 13 pN (H6Sx1A-Sb2H6, n = 54; Fig. 1 C; right) for antiparallel orientations (one-way analysis of variance, F_(3,219) = 0.175). However, there were significant differences (∼3 nm, corresponding to ∼20 amino acids considering 0.15 nm per amino acid (25)) in the mean extensions between parallel (22.2 ± 1.0 nm for Sx1AH6-Sb2H6 (16), and 22.9 ± 0.9 nm for H6Sx1A-H6Sb2) and antiparallel (25.5 ± 0.8 nm for Sx1AH6-H6Sb2, and 20.2 ± 0.9 nm for H6Sx1A-Sb2-H6) orientations (one-way analysis of variance, followed by Fisher's least significant difference test, p < 0.05; Fig. 1 D). The difference in extension measurements between the two antiparallel orientations (in different terminus configurations) was more pronounced (least significant difference test, p < 0.01), whereas the extensions obtained for two parallel orientations were not significantly different from each other. Note that all of these experiments were done at a temperature of 297 K. These results indicate that due to the different configurations of the terminus, the stretched parts of the Sx1A and Sb2 might be different, leading to the significant differences in extensions. Although the bound regions may vary, the strength of the bond between them is similar because the protein regions are brought together by the binding of their hydrophobic residues, leading to similar rupture forces (Fig. 1 C). Because force and extension measurements at a single pulling speed cannot tell the real intrinsic bond strength, it is necessary to extend our experiments to various pulling speeds to obtain more complete information about the Sx1A and Sb2 interaction.

Consequently, we carried out force-loading rate experiments to investigate the nature and stability of the interactions for each intermolecular pair (Fig. 2). In our previous work (8), we demonstrated that Sx1A and Sb2 interact via their SNARE domains by bonding, which we characterized as zippering. Consistent with that finding, in the work presented here, we found that as the force-loading rate increased, the mean rupture force and extension required to take apart Sx1A-Sb2 pairs in any orientation displayed an exponential relationship (r = 0.95–0.99 for various pairs), indicating a zippering type of interaction, i.e., the formation of coiled-coils. Having determined the nature of bonding, we focused on the force versus force-loading-rate experiments to study the stability of intermolecular interactions as detailed below.

Figure 2.

Dynamic-force experiments. Rupture extension (top graphs) and force (bottom graphs) versus force-loading rate are plotted. As the force-loading rate is increased, the mean extension and rupture force required to take apart Sx1A-Sb2 pairs in all orientations/configurations display an exponential relationship pointing to the zippering nature of the interaction. By extrapolating the force to zero, we calculated the spontaneous lifetimes to be 0.21 ± 0.05 s for H6Sx1A-H6Sb2 (left), 0.26 ± 0.08 s for Sx1AH6-H6Sb2 (middle), and 0.16 ± 0.04 s for H6Sx1A-Sb2H6 (right) interactions, respectively.

According to chemical reaction rate theory, if we measure the rupture forces at different force-loading rates, by fitting the force versus the force-loading rate (which is the rate of change of applied force), we will obtain the spontaneous lifetime of the bond. In these measurements, we changed the moving speed v of the coverslip using the piezoelectric actuator, so that the rate of applied force on the bond, f = k × v (where k is the spring constant of a cantilever), changed accordingly. By single-molecule chemical reaction rate theory (26), the rupture force will display an exponential change with the force-loading rate. From the slope and x-intercept of the force versus the natural logarithm of the force-loading rate, we can obtain the spontaneous lifetime τ of the bond (26). Fig. 2 shows the rupture force as a function of the force-loading rate for the different configurations of Sx1A and Sb2 at 297 K. We measured the rupture forces at five different force-loading rates ranging from 500 to 70,000 pN/s. The rupture force is clearly seen to increase exponentially with an increase of the force-loading rate for all of the termini configurations (Fig. 2). The differences between calculated lifetimes τ₀ for the various protein combinations—0.18 ± 0.05 s (Sx1AH6-Sb2H6 (16)), 0.21 ± 0.05 s (H6Sx1A-H6Sb2), 0.26 ± 0.08 s (Sx1AH6-H6Sb2), and 0.16 ± 0.04 s (H6Sx1A-Sb2H6)—at 297 K are an order of magnitude smaller than the 2.1 s lifetime for the ternary complex containing SNAP25B in addition to Sx1AH6 and Sb2H6 (8,23). Because the spontaneous lifetime is closely associated with the free energy of the bond, the interaction free energy of these SNARE protein bonds can be obtained from these force-loading experiments. According to reaction rate chemistry,

$$\frac{1}{{\tau }_0}=k_{\text{a}}\text{exp}\left(\frac{-\Delta G}{k_{\text{B}}T}\right),$$ (1)

where k_a is a constant depending on the system studied and the model chosen (Eyring or Kramers) (26,27),τ₀ is the spontaneous lifetime, and ΔG is the interaction free energy. Because ΔG = ΔH − TΔS (where ΔH and ΔS are the corresponding changes in the enthalpy and entropy, respectively), we can rewrite Eq. 1 as

$$\frac{1}{{\tau }_0}=k_{\mathrm{a}}\times \text{exp}\left(\frac{\Delta S}{k_{\mathrm{B}}}\right)\times \text{exp}\left(\frac{-\Delta H}{k_{\mathrm{B}}T}\right).$$ (2)

After obtaining the spontaneous lifetime of a bond at different temperatures, we can plot the lifetime dependence as a function of temperature. Using Eq. 2, we can obtain changes in entropy ΔS and enthalpy ΔH of the bond leading to the ΔG. We repeated the force-loading rate experiments of Sx1A and Sb2 proteins at three different temperatures: 277 K, 287 K, and 297 K. We observed that the lifetime increased with the decrease in temperature, as expected. After we obtained these lifetimes, we plotted the natural logarithm of lifetime versus the inverse of the temperature as shown in Fig. 3. Using Fig. 3 and Eq. 2, we can find the ΔS and ΔH corresponding to the interaction. Table 1 lists the energy parameters of the various interactions. We found that the free energies were about the same for all four configurations. The ΔG value is ∼28 k_BT (Eyring model) or 20 k_BT (Kramers model). This number is reasonably consistent with that obtained for similar leucine zipper structures (28). Of interest, the enthalpies of binary Sx1A-Sb2 interactions vary between 24.7 k_BT and 33.1 k_BT. This variation is consistent with the conformation changes (also verified by alterations in the extension) that occurred during disassembly of the various protein terminus configurations.

Figure 3.

Temperature-dependent dynamic-force experiments. Spontaneous lifetimes (τ) at three different temperature set-points were measured and plotted as a natural logarithm function against the inverse temperature. The energy parameters were obtained by fitting the data (see Table 1).

Table 1.

Energies for dissociation of Sx1A and Sb2 interactions

Interaction	ΔH(kcal mol−1/kBT)	ΔS(cal mol−1 K−1/kBT)		ΔG(kcal mol−1/kBT)
		Eyring	Kramers	Eyring	Kramers
Sx1AH6-Sb2H6	19.8/33.1	10.8/5.4	24.9/12.4	16.6/27.7	12.4/20.7
H6Sx1A-H6Sb2	18.3/30.5	5.5/2.7	20.6/10.3	16.7/27.9	12.2/20.3
Sx1AH6-H6Sb2	17.7/29.6	3.4/1.7	18.6/9.3	16.7/27.9	12.2/20.3
H6Sx1A-Sb2H6	14.8/24.7	−5.5/−2.8	9.7/4.8	16.4/27.5	11.9/19.9

Change in enthalpy (ΔH), entropy (ΔS), and free energy (ΔG) from the bound state to the transition state is shown for various binary Sx1A-Sb2 interactions. The data for Sx1AH6-Sb2-H6 were adapted from Liu et al. (14).

Discussion

Our single-molecule measurements of Sx1A-Sb2 interactions, with the proteins held at different ends, indicate that although these interactions are similar in some aspects, they also have many differences. Our AFM approach to study Sx1A-Sb2 interactions uses directional protein deposition and probing to ensure conformational flexibility of the proteins while allowing their homogeneous interactions in each case/population of parallel and antiparallel orientations and configurations (8,24). The homogeneity of the population is also aided by the fact that Sx1A and Sb2 progressively zipper starting at their free ends toward their anchoring sites (8,24). These statements are supported by our findings of significant differences between measured extensions for parallel and antiparallel orientations, as well as between two antiparallel configurations (Fig. 1 C), while all orientations/configurations zippered (Fig. 2). Hence, the extension measurements obtained using a retraction speed of 1.6 μm/s show significant differences, with averages ranging from 20.2 nm to 25.5 nm for different terminal combinations (Fig. 1 C). Of more interest, as shown in Table 1, the free energies ΔG of all four configurations are about same, with values of ∼28 k_BT (Eyring model) or ∼20 k_BT (Kramers model), but the enthalpies ΔH vary from 33.1 k_BT to 24.7 k_BT. This indicates that extension is closely associated with enthalpy, which is consistent with the fact that the enthalpy change includes the energy involved in conformation changes. This enthalpy change is balanced out by a change of the entropy, leading to the same free energy. These findings imply that the orientation of Sx1A-Sb2 pairs does not grossly affect the interaction of the proteins, even though structurally there should be differences.

It is worth remembering the characteristic structure of the parallel ternary SNARE complex of Sx1A-SNAP25-Sb2 obtained from x-ray crystallography data (6), i.e., four α-helices aligned in parallel with a zero layer buried in the center. Also, the H_abc region of Sx1A does not affect the Sx1A-Sb2 interaction through their SNARE domains, as we described elsewhere (8). With these two considerations in mind, and given the fact that Sx1A-Sb2 proteins zipper (with an initiation site at their free ends, which then progresses toward their anchoring/H6 tag sites) in any orientation/configuration (Fig. 2), an obvious assumption for the bonds here would be that the Sx1A-Sb2 binary interaction may also be aligned at the zero layer equivalent (Arg-56 of Sb2 and Gln-226 of Sx1A interactions are buried by a number of hydrophobic pairs) as shown in Fig. 4. Given this assumption, it is easy to understand that despite the position of the H6 tag used to immobilize the proteins on the nickel surface, the two parallel configurations basically zippered, as we determined from force-loading rate experiments, and perhaps aligned in similarity to the constituents of the parallel ternary complex. Thus, they could be opened from either end depending on the configuration. Note that in the case of H6Sx1A-H6Sb2, we find that pulling of the extra N-terminal part of Sx1A is not bound with Sb2. However, a noticeable difference can be observed when we look at the measurements obtained using a retraction speed of 1.6 μm/s for the two antiparallel configurations, where a rupture force of 252 ± 14 pN for Sx1AH6-H6Sb2 is associated with a 25.5 ± 0.8 nm extension, and a statistically similar force of 226 ± 13 pN for H6Sx1A-Sb2H6 is associated with a significantly shorter extension of 20.2 ± 0.9 nm. Consistent with the shorter extension, H6Sx1A-Sb2H6 also has a smaller enthalpy change than Sx1AH6-H6Sb2 (24.7 k_BT vs. 29.6 k_BT, respectively).

Figure 4.

Amino acid sequence of partial Sx1A and Sb2 aligned in parallel and antiparallel orientations. (A) Parallel alignments of either Sx1AH6-Sb2H6 or H6Sx1A-H6Sb2. (B and C) Antiparallel alignments: Sx1AH6-H6Sb2 in B and H6Sx1A-Sb2H6 in C. The amino acid sequence of partial Sb2 (upper: GenBank accession number BC074003) and Sx1A (lower: GenBank accession number AF217191). Numbers in parentheses denote the position of amino acids in the sequence. The proteins in pairs are aligned using Arg-56 of Sb2 and Gln-226 of Sx1A, which otherwise contribute to the zero layer of the ternary SNARE complex. The putative interactive hydrophobic residues/bonds are depicted in bold.

Based on the previous assumption of alignment of the proteins at the zero layer, the two antiparallel configurations should also have similar extensions instead of the difference observed. A possible explanation for this apparent discrepancy is as follows: the WL pair at either end of two configurations plays an important role. When the zipper of H6Sx1A-Sb2H6 is initialized, the starting WL pair is relatively far away from the center (Fig. 4 C). There is a possibility that this hydrophobic pair is not formed due to slight misalignment owing to strain from the proximity of the surface. An additional possibility is that this lone bond at the end is easily ruptured by thermal fluctuations. If this were true, then ∼30 amino acids (∼4.5 nm) would be effectively unbound compared with Sx1AH6-H6Sb2 (Fig. 4 B), resulting in the ∼5 nm difference in the extension measured. This also could explain the small difference in free energy and the substantially lower enthalpy change, as shown in Table 1. Indeed, future experiments using WL mutants of truncated Sb2 (aa 1–94) and Sx1A (aa 176–266) may be able to corroborate this explanation. Nonetheless, a similar explanation could be used when the free energy of parallel and antiparallel configurations are compared. As shown in Fig. 4 A, parallel configurations actually have more hydrophobic pairs than do antiparallel configurations. For example, besides the core SNARE region, there are a few more pairs at the N-terminal of the parallel configurations. However, the large distance from center and possible misalignment result in a similar number of hydrophobic pairs for all of the configurations, leading to similar free energies.

The estimated energy required to tether/dock the vesicle on the plasma membrane is ∼13 k_BT, and the energy needed to merge vesicular and plasma membrane bilayers, leading to fusion, is ∼40–50 k_BT (29–32). The activation energy of a single binary trans Sx1A-Sb2 pair (∼33 k_BT maximum obtained using the parallel Sx1H6-Sb2H6 pair, and ∼25 k_BT minimum obtained using the antiparallel H6Sx1A-Sb2H6) is sufficient to let the vesicle dock on the membrane, but is not enough to lead to fusion. Thus, if the binary complex could lead to fusion, the probability would be low and require the involvement of two or more binary complexes. In contrast, single ternary SNARE complexes can lead to successful vesicular fusion (33). Perhaps this is not surprising, given that an activation energy of ∼43 k_BT for (dis)assembly of a parallel ternary SNARE complex has been reported (13,14) (for reviews, see Liu and Parpura (34,35)), which is comparable to the energy levels necessary to merge vesicular and plasma membrane bilayers. However, how the energy is transferred from the bundling of the SNARE domains within the ternary complex into the lipid bilayers, which ultimately need to reshape in membrane mergers, remains a key unknown. It has been suggested that the linker regions between the Sb2 and Sx1A SNARE domains, within the ternary complex, and their transmembrane domains may play a role in this process (36,37).

Single-molecule FRET experiments in solution recorded a mixture of ternary SNARE complexes in (anti)parallel orientations, with the ternary SNARE complex with parallel orientation displaying substantially more stability then the ternary complex in which Sb2 had antiparallel orientation with its acceptor, the cis binary Sx1A-SNAP25 complex, to which was bound (17). Consistent with this notion, the native ternary SNARE complex extracted from the brain displays parallel orientation as revealed by electron microscopy (3). This raises the possibility that accessory molecules present in the cellular milieu are important in the selection of the parallel ternary SNARE complex for mainstream regulated exocytosis, whereas the antiparallel ternary SNARE complex may represent an off state that can reduce the probability of release (17).

In recent models for vesicle tethering, docking, and fusion to/with the plasma membrane, investigators have proposed several (non)productive serial/parallel pathways, which can include the interactions of the ternary SNARE complex with various lipids (e.g., sphingosine (38)) and accessory proteins such as synaptotagmins, complexins, and Munc18 (39–44). For instance, in single-molecule FRET experiments, Munc18 stabilized the binary cis Sx1A-SNAP25 complex with 1:1 stoichiometry, which represents an acceptor for Sb2 binding with subsequent formation of a parallel ternary SNARE complex (44). Similarly, a study using single-molecule AFM experiments with isolated proteins combined with whole-cell electrophysiology indicated that Munc18-1 can tune interactions between proteins constituting the ternary SNARE complex (20). The results of that study suggest that several modes of vesicle-plasma membrane interactions may exist within a cell depending on the cytosolic amount of Munc18-1 and its interactions with various binary/ternary SNARE complexes (20). For instance, Munc18-1 interacting with Sx1A could annul binding of vesicular Sb2 to Sx1A, thus preventing formation of the binary trans Sx1A-Sb2 complex and instead favoring the formation of a 1:1 Sx1A-SNAP25B cis complex at the plasma membrane. Of course, the acceptor complex would then follow a classic pathway of interacting with vesicular Sb2 to form the ternary SNARE complex. However, a parallel alternative pathway that relies on trans Sx1A-Sb2 interactions also seems possible in the cellular environment under conditions in which Munc18-1 has a reduced ability to interact with Sx1A. One likely display of such a dichotomy of tethering/fusion events was evident in findings of two distinct fusion pore states. Consequently, in addition to a variety of possible interactions, with respect to the core SNARE proteins, it is possible that two vesicular tethering/fusion modes occur in the cell: one that is less stable and based on trans Sb2-Sx1A interaction alone, and one that is substantially more stable and based on the formation of the ternary SNARE complex.

Whether the orientation of the binary trans Sb2-Sx1A complex plays any additional role in this process is unclear. On the basis of our energy measurement, one might speculate that the orientation of the binary complex is irrelevant. However, it should also be noted here that the parallel complex zippering brings the vesicle much closer to the membrane than the formation of the antiparallel binary complex, which would leave the vesicle at a distance corresponding to the length of the order of half the measured extension (∼10–13 nm). In this instance, the likelihood of vesicle-plasma membrane fusion would be remote. Nonetheless, the energetic requirements for the binary trans Sx1A-Sb2 and other complexes that contribute to various (non)productive exocytotic pathways need to be determined before we can gain a detailed understanding of this critical process for cell-cell communication in eukaryotes.

Conclusion

We have determined the energy requirements for disassembly of the single trans binary Sx1A-Sb2 complex in various orientations/configurations. This complex has substantially lower activation energy and stability than the blue-chip parallel ternary SNARE complex. Even though the trans binary Sx1A-Sb2 complex would be less utilized in the processes of vesicular tethering, docking, and fusion, giving its competing nature to form the ternary SNARE complex, its (dis)assembly could represent an additional site for regulation of exocytosis.