SNAP-25, a “ARE Which Binds to Both Syntaxin and Synaptobrevin via Domains That May Form Coiled Coils”

The membrane proteins SNAP-25, syntaxin, and synaptobrevin (vesicle-associated membrane protein) have recently been implicated as central elements of an exocytotic membrane fusion complex in neurons. Here we report that SNAP-25 binds directly to both syntaxin and synaptobrevin. The SNAP-25-binding domain of syn- taxin lies between residues 199 and 243, within the region previously shown to mediate synaptobrevin binding (Calakos, N., Bennett, M. K., Peterson, K. E., and Scheller, R H. (1994) Science 263, 1146-1149). The syn-taxin-binding domain of SNAP-25 encompasses most of the amino-terminal half of SNAP-25, including its puta- tive palmitoylation sites. Truncation of the carboxyl-ter-minal 9 residues of SNAP-25, which yields a fragment corresponding to that generated by botulinum neurotoxin A, diminishes the interaction of SNAP-26 with syn- aptobrevin, but not with syntaxin. Sequence analysis revealed that the regions that mediate the interaction between SNAP-25 and syntaxin contain heptad repeats characteristic of certain classes of a-helices. Similar re- peats are also present at the carboxyl terminus of SNAP-25 and in synaptobrevin. These domains have a moderate to high probability of forming coiled coils. We conclude that SNAP-25 can interact with both syntaxin and synaptobrevin

The membrane proteins SNAP-25, syntaxin, and synaptobrevin (vesicle-associated membrane protein) have recently been implicated as central elements of an exocytotic membrane fusion complex in neurons. Here we report that SNAP-25 binds directly to both syntaxin and synaptobrevin. The SNAP-25-binding domain of syntaxin lies between residues 199 and 243, within the region previously shown to mediate synaptobrevin binding (Calakos, N., Bennett, M. K., Peterson, K. E., and Scheller, R H. (1994) Science 263, 1146-1149). The syntaxin-binding domain of SNAP-25 encompasses most of the amino-terminal half of SNAP-25, including its putative palmitoylation sites. Truncation of the carboxyl-terminal 9 residues of SNAP-25, which yields a fragment corresponding to that generated by botulinum neurotoxin A, diminishes the interaction of SNAP-26 with synaptobrevin, but not with syntaxin. Sequence analysis revealed that the regions that mediate the interaction between SNAP-25 and syntaxin contain heptad repeats characteristic of certain classes of a-helices. Similar repeats are also present at the carboxyl terminus of SNAP-25 and in synaptobrevin. These domains have a moderate to high probability of forming coiled coils. We conclude that SNAP-25 can interact with both syntaxin and synaptobrevin and that binding may be mediated by a-helical domains that form intermolecular coiled-coil structures.
Neurons release their neurotransmitter by regulated exocytosis of synaptic vesicles. Recently, major progress has been made in our understanding of the molecular mechanisms that underlie exocytotic membrane fusion (reviewed by Rothman and Warren (1994) and Siidhof et al. (1993)). A complex consisting of membrane constituents of both the synaptic vesicle as well as the synaptic plasma membrane is thought to act as the core of a n exocytotic docking and fusion machine. These components include the synaptic vesicle protein synaptobrevin (also referred to as vesicle-associated membrane protein (Trimble et al., 1988)) and the plasmalemmal proteins syntaxin and SNAP-25 (Bennett et al., 1992;Oyler et al., 1989;Sollner et a l . , 1993a).
Several lines of evidence indicate that these proteins are essential for exocytosis. First, tetanus and botulinum neurotoxins irreversibly block neuronal exocytosis by selectively cleaving synaptobrevin, syntaxin, or SNAP-25 (reviewed by * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked LLaduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. for intracellular fusion events, selectively interact with the same set of membrane proteins. These soluble proteins are NSF' and dp,y-SNAPs. Since SNAPs must bind to membrane receptors before NSF, the receptors have been designated as SNAREs, with the vesicular protein (synaptobrevin) designated as a v-SNARE and the plasmalemmal proteins (SNAP-25 and syntaxin) designated as t-SNARES. ATP cleavage by NSF appears to dissociate this complex, an event proposed to underlie membrane fusion (Sollner et al., 199313;Rothman and Warren, 1994). Third, homologues of all three membrane proteins, as well as of NSF and a-SNAP, have been identified in yeast, where genetic evidence has implicated each of these proteins in the final steps of intracellular membrane fusion. Each trafficking pathway appears to require its own specific set of SNAREs, whereas the yeast homologues of NSF and a-SNAP are thought to operate in all steps. Together, these findings suggest that the principal mechanisms of exocytotic membrane fusion have been highly conserved during evolution (reviewed by Bennett and Scheller (1993) and Ferro-Novick and Jahn (1994)).
Both synaptobrevin and syntaxin are small integral membrane proteins with single putative transmembrane domains located at the carboxyl-terminal ends of their sequences. SNAP-25 does not contain a transmembrane domain, but is palmitoylated, probably at cysteine residues present in the middle of the molecule (Oyler et al., 1989;Hess et al., 1992). In nervous tissue, two isoforms of syntaxin (1A and 1B (Bennett et al., 1992)), synaptobrevin (I and I1 (Elferink et al., 1989)), and SNAP-25 (A and B (Bark, 1993;Bark and Wilson, 1994)) have been characterized. In addition, one synaptobrevin homologue (cellubrevin (McMahon et al., 1993)) and several syntaxin homologues  have been identified in nonneuronal cells, indicating that the SNAREs comprise multimember protein families.
Recent experiments have demonstrated that the neuronal SNARE proteins are complexed in the absence of NSF and SNAPS (Sollner et al., 1993b). Furthermore, a direct interaction between syntaxin and synaptobrevin has been characterized (Calakos et a l . , 1994). This interaction exhibits a moderate to low affinity, but is selective for individual isoforms of syntaxin: synaptobrevins I and I1 bind t o syntaxins 1 and 4, but not to syntaxin 2 or 3. Therefore, these interactions may contribute to the selective targeting and fusion of membrane-bound vesicles with the appropriate acceptor membrane. However, the role of SNAP-25 in exocytotic membrane fusion, while essential (Blasi et al., 1993a1, has yet to be defined. Is SNAP-25 a subunit of the "ARE syntaxin, perhaps modulating the interaction of syn- taxin with V-SNARES, or can SNAP-25 itself function as a t-SNARE by directly interacting with v-SNARES? To address these issues, we have investigated potential interactions between SNAP-25 and other components of the SNARE complex. We demonstrate that SNAP-25 binds to both syntaxin and synaptobrevin and have examined the structural requirements that underlie these interactions.

MATERIALS AND METHODS
Plasmid Construction and Purification of Recombinant Proteins-cDNA encoding rat SNAP-25 (Blasi et at., 1993a;Binz et al., 1994) and cDNAs encoding synaptobrevin I1 (Elferink et ul., 1989) and syntaxin 1A  were kindly provided by T. C. Siidhof and R. H. Scheller, respectively. All full-length and truncated coding sequences were amplified using the polymerase chain reaction with oligonucleotides containing appropriate restriction sites for subsequent subcloning into the desired plasmids. For in vitro transcription, SNAP-25 and syntaxin were subcloned into CDM8 as described previously (Blasi et al., 1993a(Blasi et al., , 1993b. The systematic truncation of syntaxin was carried out using a sense PCR primer 80 residues upstream of the T7 promoter of CDM8 and 3"antisense primers corresponding to the end of the truncation. PCR products were extracted, precipitated, and added directly to the coupled transcription-translation reaction (described below).
For the production of glutathione S-transferase fusion proteins, synaptobrevin as well as full-length and various regions of syntaxin and SNAP-25 were subcloned into pGEX-1 or pGEX-2T (Pharmacia Biotech Inc.). JM109 cells (500 ml) expressing glutathione S-transferase or glutathione S-transferase fusion proteins were grown to A,, = 0.4, induced by the addition of 0.4 m M isopropyl-P-D-thiogalactopyranoside, and harvested after 2-4 h. Cells were pelleted; resuspended in 10 ml of 20 m M Tris, pH 7.2, 150 m M NaCl (TS buffer) containing protease inhibitors (1 m~ phenylmethylsulfonyl fluoride, 2 pg/ml pepstatin, and 20 pg/ml aprotinin); and sonicated on ice 3 x 30 s with a probe sonicator. Triton X-100 was added to the lysate (0.5%) and mixed for 20 min at 4 "C, and the suspension was centrifuged twice at 12,000 rpm for 20 min in an SS34 rotor. The supernatant was mixed with 0.5 ml of a 50% slurry of glutathione-Sepharose 4B (Pharmacia Biotech Inc.) for 30 min at 4 "C with mixing, batch-washed three times with 50 ml of TS buffer, and stored as a 50% slurry in TS buffer at 4 "C. Aliquots were subjected to SDS-PAGE and visualized by Coomassie Blue staining, and the concentration of the fusion proteins was determined by comparison with glutathione S-transferase standards.
Full-length synaptobrevin I1 was obtained by treating the glutathione-Sepharose-immobilized fusion protein, derived from pGEX-ST, with 2.5 units of thrombin (Sigma)/100 pl of beads for 1 h at room temperature in TS buffer plus 0.5% Triton X-100 (Guan and Dixon, 1991). The supernatant containing the purified synaptobrevin was collected, and the thrombin was inhibited by the addition of 1 m~ phenylmethylsulfonyl fluoride.
Full-length recombinant syntaxin 1A was prepared by subcloning into pTrcHis (Invitrogen), resulting in a fusion protein that contains a 6-histidine tag at its amino terminus. The fusion protein was expressed and harvested as described above for the glutathione S-transferase fusion proteins with the following differences. Expression was in TOP-10 cells (Invitrogen); bacteria were lysed in 20 m~ HEPES, pH 7.4, 500 r m KC1; and the fusion protein was purified using immobilized metal affinity chromatography using ProBond (Invitrogen) nickel resin in 20 m~ HEPES, 200 m~ KC1 with 0.5% Triton X-100. Purified syntaxin fusion protein was eluted with an imidazole gradient (0-500 m~ imidazole; with the majority of the fusion protein eluting at 120 rm), and the pure fractions were pooled and dialyzed against 10 m~ HEPES pH 7.4, 140 m~ KCl, and 0.5% Triton X-100.
In Vitro IFunscription-TkansZation-cDNAs encoding wild-type and mutant syntaxin and SNAP-25 sequences were placed under control of the T7 promoter by subcloning into the EcoRI site of a modified version of CDM8 (Seed, 1987) as described previously (Blasi et al., 1993a(Blasi et al., , 1993b. Radiolabled proteins were generated by coupled in vitro transcription-translation using the TnT system (Promega), in the presence of [36S]methionine, according to the manufacturer's instructions. Fulllength syntaxin was translated in the presence of microsomes (Promega) as described previously (Blasi et al., 1993b).
Binding Assays-Binding assays were carried out by adding radiolabeled or purified recombinant proteins to immobilized recombinant fusion proteins. Binding was monitored by fluorography or immunoblot analysis, respectively. In the former assay, 5 pl of the 135Slmethioninelabeled in vitro transcription-translation reaction mixtures was added t o 15 pl (dry bed volume) of glutathione-Sepharose beads containing 30 pg of immobilized glutathione S-transferase or GST-SNAP-2Wsyntaxin fusion proteins in a final volume of 100 pl in 10 m M HEPES pH 7.4,140 rm KC1, 2 m~ MgCl,, and 0.5% Triton X-100. Samples were mixed overnight at 4 "C and centrifuged; the supernatants were decanted; and the pellets were washed three times with 1 ml of binding buffer. Samples were solubilized in SDS sample buffer (final concentrations: 60 m~ Tris, pH 6.8,2% SDS, 10% sucrose, and 3% P-mercaptoethanol) and boiled for 3 min. Equal portions of the supernatant and pellets were subjected to SDS-PAGE, stained, destained, treated with Amplify ( A mersham Corp.) for 15 min, dried, and exposed to film for 4-12 h.
In the second assay, purified full-length recombinant syntaxin (5 pg) or synaptobrevin (9 pg) was mixed with 15-30 pg of various GST-SNAP-25 fusion proteins immobilized on glutathione-Sepharose for 2 h at 4 "C in a final volume of 100 pl in 10 m~ HEPES, pH 7.4,140 m~ KCl, 2 rm MgCl,, and 0.5% Triton X-100 or in TS buffer and 0.5% Triton X-100. Samples were washed three times with 1 ml of binding buffer, solubilized by boiling in SDS sample buffer, and subjected to SDS-PAGE and immunoblot analysis using lZ6I-protein A.
Antibodies-A mouse monoclonal antibody (10.1) directed against synaptobrevin has been described previously (Baumert et al., 1989). A monoclonal antibody directed against syntaxin (also referred to as HPC-1) was kindly provided by C. Barnstable. Rabbit antiserum directed against the carboxyl-terminal 12 residues of SNAP-25 was kindly provided by T. Chilcote and P. De Camilli. Monoclonal antibody 71.1 was raised against recombinant SNAP-25 using standard procedures (Kohler and Milstein, 1975;Jahn et al., 1985) and will be described in detail elsewhere.
Blots probed with lZ6I-protein A were incubated with mouse monoclonal primary antibodies as indicated above, washed five times, incubated for 1 h with rabbit anti-mouse IgG antibodies (Cappel) diluted 1:1000, washed five times, incubated with 0.1 pCi/ml '251-protein A (DuPont NEN) for 1 h, washed five times, dried, and exposed to x-ray film.
Miscellaneous Procedures-Coiled-coil predictions were carried out as described by Cam and Kim (1993) using a revised version of the computer program described by Lupas et al. (1991) (based on Parry (1982)) using a window size of 28 residues. Binding assays were quantitated by scanning densitometry using a Visage 2000 scanner (BioImage Products, MilligedBioresearch Division of Millipore).

RESULTS
To assay for a direct interaction between syntaxin and SNAP-25, we prepared full-length [36SlMet-labeled syntaxin 1A by in vitro transcription-translation in the presence of microsomes. Labeled syntaxin was mixed with glutathione S-transferase or a GST-SNAP-25 fusion protein immobilized on glutathione-Sepharose beads in the presence of detergent. Binding was analyzed by SDS-PAGE and fluorography following separation of the beads from the unbound material. Syntaxin 1A quantitatively bound to immobilized SNAP-25, but not to the control beads, which contained only glutathione S-transferase (Fig. hi). Similar results were obtained using translated syntaxin 1B (data not shown). To map the domain of syntaxin 1A that mediates this interaction, we generated a series of carboxyl-terminal truncation mutants (Fig. 1, top) and tested them for binding using the same procedure. As shown in was assayed by mixing the radiolabeled syntaxin with glutathione Stransferase beads (control) or GST-SNAP-25 beads overnight a t 4 "C in the presence of detergent (see Fig. 3A for Coomassie Blue-stained gel of the immobilized fusion proteins). The beads and supernatants were separated by centrifugation; the pellets were washed three times; and equal fractions of the supernatants (s) and pellets ( p ) were subjected to 12% SDS-PAGE and analyzed by fluorography. B, a series of carboxylterminal syntaxin truncation mutants, shown diagrammatically at the top, were generated by in vitro translation and assayed for SNAP-25 binding as described above. Truncation to residue 243 did not appear to affect binding. However, the interaction with SNAP-25 was reduced (by 60%) and abolished by truncations to residues 221 and 199 of syntaxin, respectively. TMR, transmembrane domain. loss of binding activity. However, further truncation of the carboxyl terminus to residue 221 reduced binding by 60%, and truncation to residue 199 abolished binding.
These data show that the region between residues 199 and 243 of syntaxin is important for binding to SNAP-25. However, these findings do not distinguish whether this domain is sufficient for binding, whether it contains only a part of the binding domain, or whether the deletions caused misfolding of the truncated molecule, resulting in loss of binding activity. To address this issue, we generated a glutathione S-transferase fusion protein that contained residues 199-243 of syntaxin 1A (Fig.  2 A ) . As a control, we also fused the entire cytoplasmic domain of syntaxin 1A to glutathione S-transferase (Fig. 2 4 ) . As shown in Fig. 2 B , translated SNAP-25 bound, as expected, to the cytoplasmic domain of syntaxin 1A. Furthermore, this interaction was preserved in the 45-amino acid segment of syntaxin.
Together, these data demonstrate that SNAP-25 binds to syntaxin 1A in the absence of other synaptic proteins. Binding was confined to a relatively short linear sequence of syntaxin lA, within residues 199-243. In the following experiments, we used a similar approach to identify the region of SNAP-25 that mediates its interaction with syntaxin. For this purpose, we carried out systematic carboxyl-and amino-terminal deletion analyses of SNAP-25. In these experiments, the deletion mutants were fused to glutathione S-transferase and immobilized using glutathione-Sepharose. Binding was assayed as described for Fig. 2  We first examined the effect of carboxyl-terminal deletions of SNAP-25 on syntaxin 1A binding. As shown in Fig. 3B, syntaxin binding was preserved in a fusion protein composed of the amino-terminal 100 residues of SNAP-25, demonstrating that the binding activity resides in the amino-terminal half of SNAP-25. To further define this binding domain, we examined syntaxin 1A binding to shorter segments of SNAP-25. A carboxyl-terminal deletion to residue 81 diminished binding by 38%, and further truncation to residue 61 abolished binding (Fig. 3B). We then prepared amino-terminal deletion mutants of SNAP-25 that were fused to glutathione S-transferase in order to determine the amino-terminal border of the syntaxinbinding domain (Fig. 4A). Syntaxin binding was preserved in a SNAP-25 mutant lacking the first 20 residues (Fig. 4B), a stretch that is poorly conserved between different species (Risinger et al., 1993). Further amino-terminal deletions of SNAP-25 resulted in the loss of detectable binding activity (Fig. 4B).
The data described so far indicate that the amino-terminal half of SNAP-25 is crucial for its interaction with syntaxin 1A. To confirm this assignment, we compared the abilities of SNAP-25 fusion proteins, consisting of either the aminoor carboxyl-terminal half of the protein (Fig. 5A), to bind syntaxin. For these experiments, we purified recombinant syntaxin 1A from Escherichia coli and carried out immunoblot analysis to assay for binding. As expected, syntaxin 1A bound to the full-length and the amino-terminal half of the molecule, but not to the carboxyl-terminal region (Fig. 5C).
Interestingly, the syntaxin-binding domain of SNAP-25 identified in the experiments described above does not contain the region of SNAP-25 that is cleaved by BoNTIA. BoN'l'IA, a potent inhibitor of exocytosis, has been shown to act by hydrolyzing  1-197 lanes); B, immunoblot analysis of the same immobilized SNAP-25 fusion proteins using antibodies directed against the carboxyl terminus of SNAP-25 or a monoclonal antibody that recognized the amino-terminal region (immunoreactive bands were visualized by enzymatic staining); C, binding of full-length recombinant syntaxin 1A or synaptobrevin I1 to the GST-SNAP-25 fusion proteins, measured by SDS-PAGE, and immunoblotting of the bound material. Bacterially expressed syntaxin 1A and synaptobrevin I1 were purified and mixed with glutathione S-transferase or GST-SNAPS5 fusion proteins immobilized on glutathione-Sepharose for 2 h at 4 "C. The beads were washed three times, and the bound proteins were solubilized with SDS sample buffer and subjected to SDS-PAGE and immunoblot analysis using monoclonal antibodies directed against syntaxin or synaptobrevin. Immunoreactive bands were visualized with '*'II-protein A. Synaptobrevin bound to full-length SNAP-25, but did not interact with either of the individual halves of SNAP-25. Deletion of the last 9 amino acids of SNAP-25 diminished binding of synaptobrevin by 73% but did not affect syntaxin binding. As expected, syntaxin bound to full-length and the amino-terminal (but not the carboxyl-terminal) half of SNAP-25.

SNAP-25 Binds to Both Syntaxin and Synaptobrevin
taxin binding activity, we prepared a SNAP-25 fusion protein that lacked the last 9 amino acids. As shown in Fig. 5C, residues 1-197 of SNAP-25 bound recombinant syntaxin 1A as well as full-length SNAP-25. What, then, is affected by BoNTIA cleavage of SNAP-25? It has been shown previously that the third membrane component of the SNARE complex, synaptobrevin, interacts directly with syntaxin, although with a relatively low affinity (Calakos et al., 1994; see the Introduction). Since there is some similarity between the synaptobrevin-binding domain of syntaxin and the carboxyl-terminal domain of SNAP-25 (Risinger et al., 19931, we examined whether synaptobrevin is also capable of binding to SNAP-25 and whether this interaction involves the carboxylterminal 9 residues of SNAP-25. As shown in Fig. 5C (lower  panel), recombinant synaptobrevin I1 purified from E. coli bound to full-length GST-SNAP-25 fusion protein immobilized on Sepharose beads. This binding was reduced by 73% when the last 9 amino acids of SNAP-25 were removed even though the amount of this mutant used in the binding experiment was higher than that of the full-length protein (Fig. 5, A and B ) . These findings show that the carboxyl terminus of SNAP-25 is involved in interactions with synaptobrevin and suggest that the perturbation of these interactions by BoNT/A may underlie C. An analysis of each set of heptad repeats is given in the table. The scores reflect the propensity of these regions to form coiled-coil structures. Scores below 1.3 generally correspond to amphipathic a-helices in globular proteins, and scores above 1.3 have a high probability of forming coiled-coil structures (Lupas et al., 1991). The cleavage site of BoNT/A is also indicated. TMR, transmembrane domain. The 4 cysteine residues in SNAP-25 that may serve as palmitoylation sites are also indicated.

SNAP-25 Binds to Both Syntaxin and Synaptobrevin
the effects of the toxin on exocytosis.
We have attempted to map the region of SNAP-25 that binds synaptobrevin in order to understand the structural basis for the interaction. However, synaptobrevin failed to bind to the isolated amino-and carboxyl-terminal halves of SNAP-25 (Fig.  5C). To confirm this finding, we tested the integrity of the SNAP-25 fusion proteins by immunoblot analysis of the same samples used in the binding assay using antibodies directed against the carboxyl-terminal 12 residues or the amino-terminal domain of SNAP-25. As shown in Fig. 5C, the fusion proteins were of the predicted sizes and patterns of immunoreactivity. Therefore, the inability of synaptobrevin to bind to the isolated halves of SNAP-25 indicates that the interaction requires multiple domains and/or a complex folded structure in SNAP-25. Since residues 41-206 of SNAP-25 bind synaptobrevin (data not shown) while residues 101-206 fail to bind (Fig.  5C), the amino-terminal border of the binding domain appears to lie between residues 41 and 101.

DISCUSSION
In this study, we have demonstrated that SNAP-25 binds directly to both syntaxin and synaptobrevin. The region of syntaxin that binds to SNAP-25 was localized to residues 199-243. This domain is contained within the region previously shown to mediate the interaction of syntaxin with synaptobrevin (residues 194-267 (Calakos et al., 1994)). Therefore, it is possible that SNAP-25 may modulate the interaction of syntaxin with synaptobrevin and vice versa (discussed in more detail below).
In contrast, the domain of SNAP-25 that binds to syntaxin or synaptobrevin (described in more detail below) could not be confined to such a small region. GST-SNAP-25 fusion proteins containing either residues 1-100 or 21-206 bind syntaxin (Figs. 3B and 4B). However, a fusion protein composed of residues 21-100 interacts only weakly with syntaxin (data not shown). These data indicate that while SNAP-25 residues 21-100 contain crucial determinants for syntaxin binding, additional residues adjacent to these borders influence the binding affinity. Interestingly, the binding domain contains a set of 4 cysteine residues thought to be modified by palmitoylation (Fig.  6) (Hess et al., 1992). While our work with bacterially expressed proteins clearly demonstrates that palmitoylation is not required for SNAP-25 to bind to syntaxin, it is possible that acylation may modulate the interaction. For instance, palmitoylation of GAP-43 regulates its interaction with the a-subunit of Go (Sudo et al., 1992), and palmitoylation of the &-adrenergic receptor is required for efficient coupling to G, (O'Dowd et al., 1989).
What types of interactions underlie the binding between syntaxin and SNAP-25? It has been previously reported that syntaxin contains heptad repeats characteristic of a-helices that form coiled-coil structures (Inoue et al., 1992;Spring et al., 1993;Calakos et al., 1994). Coiled coils consist of two or more right-handed a-helices wrapped around one another with a left-handed superhelical twist and are found in a variety of proteins. For instance, the leucine zipper motif, found in a number of DNA-binding proteins, forms intermolecular coiled coils that mediate dimerization (Landschulz et al., 1988;Ellenberger et al., 19921, and coiled coils bundle the a-helices of a number of fibrous proteins, such as tropomyosin and keratin (reviewed by Cohen and Parry (1990)). Amino acids residues within the heptad repeats that form coiled coils are assigned positions a through g. While there are preferences for each residue, the most conspicuous elements in these repeats are hydrophobic amino acids at the a and d positions (Parry, 1982;Lupas et al., 1991). Interestingly, the SNAP-25-binding domain of syntaxin contains six heptad repeats in register (Fig. 6). Analysis of the propensity of these repeats to form coiled coils revealed an average score of 1.25 (Fig. 6). For comparison, a mean score of 1.44 has been reported for keratins, and a score of 1.91 was obtained for the GCN4 leucine zipper sequence (Lupas et al., 1991). According to Lupas et al. (1991), scores above 1.1 generally correspond t o regions that form extended amphipathic a-helices in globular proteins, and scores above 1.3 have a high probability of forming coiled-coil structures. Therefore, the heptad repeats in the SNAP-25-binding domain of syntaxin are likely to form an amphipathic a-helix with a modest probability of forming coiled coils. Syntaxin contains two additional sets of five and six heptad repeats in register at its amino terminus, with average scores of 1.33 and 1.64, respectively (Fig. 6). While this region is not required for the interaction of syntaxin with synaptobrevin or SNAP-25, it may be involved in oligomerization or binding to other proteins such as Munc-18 (also referred to as rbSeclhSec1 (Hata et al., 1993;Garcia et al., 1994;Pevsner et al., 1994)) or synaptotagmin (Bennett et al., 1992).
The presence of heptad repeats in the SNAP-25-binding domain of syntaxin prompted an evaluation of the primary sequence of SNAP-25. Indeed, the amino-terminal half of SNAP-25, i.e. the region that interacts with syntaxin, contains two SNAP-25 Binds to Both Syntaxin and Synaptobrevin sets of six heptad repeats interrupted by a break in frame. These heptad repeats have high probabilities of forming coiled coils, yielding average scores of 1.60 and 1.70 (Fig. 6). Therefore, the association of SNAP-25 with syntaxin may involve the interaction of a-helical domains, perhaps by forming intermolecular coiled-coil structures.
In addition, Dascher et al. (1991) have noted that synaptobrevin and the yeast SLY2 and SLY12 gene products contain regions that, when arranged in an a-helical conformation, may form coiled-coil structures. Applying the algorithm of Lupas et al. (19911, we identified five heptad repeats in register between residues 51 and 88 of synaptobrevin 11. The average score of this segment was 1.48 (Fig. 6). Interestingly, these repeats (residues 51-88) are immediately preceded by a region (residues 37-55) with properties similar to those of fusion peptide sequences from viral fusion proteins (Jahn and Siidhof, 1994).
A similar arrangement between a n amphiphilic fusion peptide and a coiled-coil motif has been observed in the influenza hemagglutinin glycoprotein, the best structurally understood viral fusion protein (Cam and Kim, 1993). In the hemagglutinin glycoprotein, low pH induces a conformational change thought to expose the fusion peptide domain to the hydrophobic core of the target membrane (reviewed by White (1992)). Recent evidence indicates that this change is mediated by the transition of a looped region of hemagglutinin into a coiled coil (Cam and Kim, 1993). It is intriguing to speculate that synaptobrevin may operate by a similar mechanism whereby the exposure of its putative fusion peptide is regulated by its interactions with the t-SNARES SNAP-25 and syntaxin and/or with NSF and SNAPS. Further experiments are clearly needed to address this possibility.
SNAP-25 contains additional heptad repeats at its carboxyl terminus (Fig. 6). Our findings indicate that this domain is required for the binding of SNAP-25 to synaptobrevin. Truncation of this domain by 9 residues reduced this interaction, suggesting that BoNT/A may act by uncoupling a t-SNARE/ V-SNARE interaction. This domain, however, is not sufficient to mediate synaptobrevin binding since no binding was observed to the isolated carboxyl-terminal half of SNAP-25 (Fig. 5C).
The ability of SNAP-25 and syntaxin to bind to each other and to individually make direct contacts with synaptobrevin raises the issue as to why t-SNARES are composed of two subunits. Sollner et al. (1993131 demonstrated that the SNARE complex disassembles when ATP is hydrolyzed by NSF. Under these conditions, binding is abolished not only between synaptobrevin and the t-SNARES, but also between the t-SNARES themselves. Since the individual interactions of syntaxin or SNAP-25 with synaptobrevin appear to be relatively weak, it is possible that SNAP-25-syntaxin heterodimers form high affinity v-SNARE-binding sites that are modulated by cyclical association and disassociation of the t-SNARES. In addition, or alternatively, hetero-oligomerization of SNAP-25 and syntaxin may contribute to the specificity of t-SNARE/v-SNARE interactions. For instance, different combinations of SNAP-25s and syntaxins may bind to different sets of v-SNARES, reducing the absolute number of t-SNARES required for selective targeting.