Specific SNARE complex binding mode of the Sec1/Munc-18 protein, Sec1p

Abstract

The Sec1/Munc-18 (SM) family of proteins is required for vesicle fusion in eukaryotic cells and has been linked to the membrane-fusion proteins known as soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs). SM proteins may activate the target-membrane SNARE, syntaxin, for assembly into the fusogenic SNARE complex. In support of an activation role, SM proteins bind directly to their cognate syntaxins. An exception is the yeast Sec1p, which does not bind the yeast plasma-membrane syntaxin, Sso1p. This exception could be explained if the SM interaction motif were blocked by the highly stable closed conformation of Sso1p. We tested the possibility of a latent binding motif using sso1 mutants in yeast and reconstituted the Sec1p binding specificity observed in vivo with purified proteins in vitro. Our results indicate there is no latent binding motif in Sso1p. Instead, Sec1p binds specifically to the ternary SNARE complex, with no detectable binding to the binary t-SNARE complex or any of the three individual SNAREs in their uncomplexed forms. We propose that vesicle fusion requires a specific interaction between the SM protein and the ternary SNARE complex.

Eukaryotic cell growth and organization depend on targeted membrane-fusion reactions between vesicles and other intracellular membranes. Fusion of the vesicle and target membranes is thought to be catalyzed by a family of membrane proteins called soluble N-ethylmaleimide-sensitive factor attachment protein receptors (SNAREs; refs. 1 and 2), which assemble between membranes in a conformation similar to the fusion-active (fusogenic) state of many viral membrane-fusion proteins (3, 4). Additional conserved protein families are required for vesicle transport and fusion throughout the cell (5, 6). For example, the Sec1/Munc-18 (SM) family of proteins is proposed to be essential for activation of SNARE-complex assembly at the vesicle-fusion step (7), and a general requirement of SM proteins for vesicle fusion is supported by studies in a variety of organisms (8).

Activation of SNARE-complex assembly is a plausible mechanism for SM-dependent vesicle fusion. SM proteins bind and potentially modulate the conformation of SNARE proteins that are homologous to the neuronal synaptic membrane protein, syntaxin (9). Syntaxins can adopt a fusion-inactive “closed” conformation (10), in which an N-terminal three-helix bundle motif (H_abc) binds the C-terminal α-helical SNARE motif (H3; ref. 11). The structure of the neuronal SM protein, nSec1, bound to the closed conformation of Syntaxin 1a has been interpreted as an intermediate required to convert syntaxin to the “open” conformation, which then assembles with the other SNAREs to form the fusogenic SNARE complex (12). Alternatively, the closed syntaxin-binding mode may represent a snapshot of nSec1 acting as an inhibitor of assembly, a function that has been attributed to SM proteins required for highly regulated exocytosis events, such as synaptic vesicle fusion (13, 14).

The universality of the closed-syntaxin-binding mode for vesicle fusion is called into question by the different syntaxin-binding modes that have been observed for other SM proteins (9, 15). The SM proteins, Sly1p and Vps45p, interact with an N-terminal peptide-finger motif of their cognate syntaxins, Sed5p and Tlg2p (16, 17). A crystal structure of Sly1p bound to the peptide-finger motif of Sed5p reveals that the surface used by Sly1p is distinct from that used by nSec1p to bind Syntaxin 1A (18). An indirect syntaxin-binding mode may describe the interaction of the yeast vacuolar SM protein, Vps33p, which is part of a large protein complex that binds to the cognate syntaxin, Vam3p (19, 20). In the most extreme case, Sec1p from yeast extracts has no observable affinity for the yeast syntaxin, Sso1p, but instead binds, perhaps indirectly, to the assembled SNARE complex (21, 22). It has been suggested that the absence of a detectable Sec1p–Sso1p interaction in yeast may be due to the stability of the closed conformation of Sso1p, which could result in an inaccessibility of Sec1p-binding sites (15).

We examined the specificity of endogenous Sec1p for the three exocytic SNAREs in yeast and report that Sec1p binds the SNAREs only when they are assembled as a ternary SNARE complex. The uncomplexed SNAREs and the binary t-SNARE complex do not coprecipitate with Sec1p from yeast lysates. When purified from yeast, Sec1p binds specifically and directly to the ternary SNARE complex, reconstituting the binding specificity observed in vivo for the endogenous full-length SNAREs. Purified Sec1p binds to the cytoplasmic domain of the exocytic SNARE complex with a 1:1 stoichiometry, and binding requires the loop region connecting the two SNARE motifs in the Sec9p SNARE. This interaction represents a SNARE-complex-binding mode that is not mediated through the peptide-finger motif or through a latent binding site exposed in the open conformation of the syntaxin SNARE. This particular binding mode may offer a snapshot of an SM protein activating membrane fusion through an interaction with the assembled SNARE complex.

Results

Endogenous Sec1p Binds Specifically to the SNARE Complex, Not to Individual SNAREs.

An interaction between endogenous Sec1p and the exocytic SNARE complex was observed in yeast lysates by using an immunoprecipitation (IP) protocol followed by immunoblot analysis (21). Recently, additional interactions between SM proteins and uncomplexed SNAREs have been reported (23, 24), including an interaction between recombinant Sec1p and the plasma membrane syntaxin, Sso1p (25).

To examine the binding specificity of endogenous Sec1p in greater detail, we used the IP protocol under conditions that maintain the individual SNAREs, Sec9p, Sso1p, and Snc2p, separated from each other. Mutations that block SNARE-complex assembly, such as in the exocyst mutant sec5–24, maintain the steady-state level of individual SNAREs but abolish the SNARE-complex interaction with Sec1p (22). At the restrictive temperature, none of the three uncomplexed SNAREs coprecipitates with the C-terminally triple Myc-tagged Sec1p (Sec1p-MYC₃) in sec5–24 (Fig. 1a). In the wild-type control (SEC+), there is no block in SNARE-complex assembly at the higher temperature, and the interaction with Sec1p-MYC₃ is preserved, as demonstrated by persistence of all three SNAREs in the IP.

Fig. 1.

In yeast, Sso1p, Sec9p, and Snc2p bind Sec1p as a SNARE complex, not as individual SNAREs. (a) All three SNAREs coprecipitate from lysate of the wild-type strain, SEC+ (NY1689), both at 25°C and after a shift to restrictive temperature (37°C). In the vesicle-tethering mutant strain, sec5–24 (NY2228), SNARE-complex assembly is blocked at 37°C, and none of the SNAREs are detected in the IP. Sec1p-MYC₃ was immunoprecipitated with the 9E10 monoclonal antibody, and coprecipitation of Sso1p, Sec9p, and Snc2p (arrows) was detected by using Western blot analysis. The input for each IP is shown (Left) and represents 1.5% of the total. The positions of molecular weight standards are indicated. (b) The interaction between Sec1p and SNAREs is abolished when SNARE complexes are disassembled. SEC+ (NY1689) and the SNARE-complex disassembly mutant strain, sec18–1 (NY1691), were lysed either in the presence (+) or absence (−) of ATP. Coprecipitated Sso1p, Sec9p, and Snc2p (arrows) were detected by using Western blot analysis, as in a. (c) SNARE complexes assembled with Sso1p[31–290] or Sso1p-Open1 coprecipitate with Sec1p-MYC₃. ATPase sensitivity of coprecipitation of SNAREs with Sec1p-MYC₃ was tested as in b. In the absence (−) of ATP, all three SNAREs coprecipitated with Sec1p-MYC₃ in lysates from the sso1 mutants, sso1[31–290] (CCY18), and sso1p-open1 (CCY17), as well as the isogenic wild-type control (CCY16). In the presence (+) of ATP, SNAREs coprecipitated only from lysates of the sso1p-open1 strain, presumably due to rapid reassembly of SNARE complexes.

In case binding of Sec1p to any individual SNARE requires disassembly of SNARE complexes, we repeated the IP under disassembly conditions. Fig. 1b shows coprecipitation of the SNARE complex with Sec1p-MYC₃ under the normal assay conditions (−ATP), which preserve SNARE complexes. Complex disassembly is activated upon addition of an ATP-regeneration system (+ATP; ref. 21), and under these conditions, no coprecipitation of disassembled SNAREs is observed for the SEC+ strain. To demonstrate that the ATP sensitivity is due to SNARE-complex disassembly, we repeated the experiment in a sec18–1 strain. The mutant gene, sec18–1, encodes an ATPase-defective version of NSF. The synaptic protein NSF (and its homologs) is a chaperone required for general SNARE-complex disassembly. When shifted to restrictive temperature, SNARE complexes are preserved in the sec18–1 strain, even in the presence of ATP. Therefore, the ATP sensitivity of SNARE coprecipitation with Sec1p-MYC₃ is due to disassembly by Sec18p. Once SNARE complexes are disassembled, the separated SNAREs do not bind to Sec1p.

Endogenous Sec1p Binds SNARE Complexes Containing Mutant Sso1p.

The IP protocol was used to test the possibility that Sec1p interacts with a latent motif that is accessible only in the open conformation of Sso1p (cartoon of conformations; Fig. 5, which is published as supporting information on the PNAS web site). Because wild-type Sso1p is stably folded in the closed conformation, binding to a latent motif was tested by using yeast strains in which the wild-type Sso1p (and Sso2p, the product of the duplicate gene, SSO2) has been replaced by either a truncation or conformational mutant of Sso1p.

Deletion of the first 30 amino acids of Sso1p removes a potential “peptide-finger” motif, described as an interaction domain between syntaxins and SM proteins known to function at other vesicle trafficking steps in the cell (16, 17). The result is a truncated protein, Sso1p[31–290], which functionally replaces the proteins encoded by the endogenous wild-type genes, SSO1 and the duplicate, SSO2, indicating that the putative peptide-finger motif is not required for the essential function of these proteins (data not shown). Fig. 1c shows ATP-sensitive coprecipitation of Sec9p, Snc2p, and the truncated Sso1p[31–290] with Sec1p-MYC₃. This result indicates that the first 30 amino acids of Sso1p are not required for SNARE-complex assembly or for the interaction of the SNARE complex with Sec1p.

The conformational mutant, Sso1p-Open1, contains three amino acid substitutions (V84E, K95E, and Y148A) in H_abc, the N-terminal autoinhibitory domain, which stabilize Sso1p in the open conformation (26). Coprecipitation of Sec9p, Snc2p, and Sso1p-Open1 with Sec1p-MYC₃ indicates that the mutations do not interfere with the interaction between Sec1p-MYC₃ and the Sso1p-Open1 SNARE complex (Fig. 1c). Because Sso1p-Open1 rapidly assembles after disassembly, accumulation of Sso1p-Open1 SNARE complexes is expected for this conformational mutant. However, reassembly is problematic for testing whether Sec1p can interact with the uncomplexed open conformation of Sso1p. We address this question using purified proteins (below).

Purification of Sec1p from Saccharomyces cerevisiae.

Previous studies using the recombinant cytoplasmic domains of Sso1p, Snc2p, and the SNAP-25 homologous domain of Sec9p demonstrated that this soluble SNARE complex, and not the uncomplexed Sso1p, binds to Sec1p from a yeast lysate (21). If the interaction between Sec1p and the SNARE complex is direct, we expected to be able to reconstitute that interaction in vitro. Binding reconstitution was performed with recombinant SNARE peptides purified from Escherichia coli and Sec1p purified from S. cerevisiae. Attempts to purify Sec1p using either E. coli or baculovirus systems failed due to aggregation of the Sec1p product in physiological buffers (M.M., data not shown). When we tried the E. coli system overexpressing GroEL/ES and His₆-Sec1p (25), the purified recombinant His₆-Sec1p did not remain in solution after centrifugation for 30 min at 300,000 × g (Fig. 6 Inset, which is published as supporting information on the PNAS web site).

Soluble Sec1p was purified from S. cerevisiae by using the pYES2/CT vector and host (Invitrogen, Carlsbad, CA). This expression vector adds to the C terminus of the SEC1 coding sequence a V5 epitope, plus a six-His tag (His₆), for affinity purification. The additional sequences do not disrupt Sec1p function in vivo, because there was no growth defect when the endogenous essential SEC1 gene was replaced with SEC1-V5-His₆ under the regulation of the SEC1 promoter (data not shown).

We used high-speed centrifugation to separate soluble Sec1p-V5-His₆ from aggregates. Centrifugation at 300,000 × g sediments particles larger than 13S (e.g., ribosomes and exocyst complexes), whereas soluble proteins smaller than 13S remain in the supernatant fraction. Sec1p-V5-His₆ purified from S. cerevisiae remained in the supernatant fraction (Fig. 5 Inset). Furthermore, Sec1p-V5-His₆ eluted from a Superdex 200 sizing column at the position predicted for the monomeric protein, M_r of 89 kDa (Fig. 5). The M_r observed by MS is 89,344 ± 11 Da (mean of five runs, ±SD; Synthesizing/Sequencing Facility, Department of Molecular Biology, Princeton University, Princeton, NJ). This mass is consistent with the calculated M_r of 89,353 Da for Sec1p-V5-His₆ with the N-terminal Met cleaved, and the Ser that follows acetylated, a common cotranslational modification in eukaryotes, including yeast (27).

Sec1p Binds Directly to the Cytoplasmic SNARE Complex.

The cytoplasmic domain of the exocytic SNARE complex, which includes Sso1p[1–265] and Snc2p[1–93] (without transmembrane domains), plus the SNAP-25 domain, Sec9p[416–651], was previously shown to bind Sec1p from a yeast lysate (21). Therefore, we used this cytoplasmic SNARE complex to test for a direct interaction with Sec1p-V5-His₆ in a pulldown assay (Fig. 2; see Materials and Methods). Bound proteins and peptides were quantitated by densitometry by using stocks of the proteins and peptides as concentration standards (Supporting Text, which is published as supporting information on the PNAS web site). The cytoplasmic SNARE complex bound directly to Sec1p-V5-His₆ (Fig. 3a). By incubating increasing concentrations of SNARE complexes with a constant amount of Sec1p-V5-His₆, we found that SNARE-complex binding saturated at 1 μM, with a stoichiometry of one cytoplasmic SNARE complex bound to each Sec1p-V5-His₆ (Fig. 7, which is published as supporting information on the PNAS web site). We detected no binding to the individual SNAREs, or the t-SNARE complex, Sec9p[416–651] plus Sso1p[1–265] (Fig. 3a, lanes 3–6) up to 10 μM (Fig. 7a). However, when 1 μM Snc2p[1–93] was added to the sample containing 1 μM t-SNARE complex (Fig. 3a, ∗), ternary complexes rapidly assembled and bound to Sec1p-V5-His₆ (compare lanes 2 and 7).

Fig. 2.

SNARE peptides used in this study. (a) SNARE peptides were mixed stoichiometrically to define the three types of SNARE complexes used for the binding studies. Lane 1: The cytoplasmic SNARE complex, Sso1p[1–265]:Sec9p[416–651]:Snc2p[1–93]. Lane 2: The core complex, Sso1p[179–265]:Sec9p[416–651]:Snc2p[1–93]. Lane 3: The four-helix bundle, Sso1p[179–265]:Sec9p[416–504]:Sec9p[571–651]:Snc2p[1–93]. Peptides are observed as Coomassie-stained bands, separated on a 16.5% Tris-tricine gel. Molecular weight markers are indicated. (b) A schematic of each of the parent proteins with SNARE peptides defined by amino-acid positions. H_abc: the N-terminal inhibitory domain of Sso1p. H₃: the SNARE motif helix. Sec9p H_a and H_b: the two SNARE-motif helices of the SNAP-25 domain of Sec9p. TM, transmembrane domains of Sso1p and Snc2p.

Fig. 3.

In vitro reconstitution of Sec1p-SNARE-complex binding. (a) Purified Sec1p-V5-His₆ binds to the cytoplasmic SNARE complex but not to individual SNAREs or the t-SNARE complex. For each binding reaction, SNAREs were added to a final concentration of 1 μM. Lane 1, Control IP: the cytoplasmic SNARE complex was added to V5 antibody immobilized on Protein G Sepharose in the absence of Sec1p-V5-His₆. Lanes 2–6: Assembled SNARE complexes or individual SNARE peptides were added, as indicated (+), to 0.4 μg of Sec1p-V5-His₆ immobilized on V5-Protein G Sepharose. Lane 7: The same t-SNARE complex used in lane 6, but with Snc2p added to the binding reaction (∗). Arrows indicate Coomassie-stained bands of the proteins and peptides, which were separated on a 16.5% Tris-tricine gel. Molecular weight markers, V5 antibody heavy chain (H_c) and light chain (L_c) are indicated. (b) The Sec9p loop region is required for interaction with Sec1p-V5-His₆. Binding of SNARE peptides to immobilized Sec1p-V5-His₆ was performed as in a. Lane 1: Control IP. Lane 2: The cytoplasmic SNARE complex. Lane 3: The cytoplasmic SNARE complex assembled with H_a and H_b of Sec9p, in the absence of the loop region, Sso1p[1–265]:Sec9p[416–504]:Sec9p[571–651]:Snc2p[1–93]. Lane 4: The four-helix bundle, Sso1p[179–265]:Sec9p[416–504]:Sec9p[571–651]: Snc2p[1–93].

A Sec9p Loop Region Is Required for SNARE-Complex Binding to Sec1p.

To determine the SNARE domains necessary for interaction with Sec1p-V5-His₆, the binding assay was performed with SNARE complexes assembled from shortened versions of the SNARE peptides (Fig. 2). In the core complex, Sec9p[416–651] and Snc2p[1–93] are assembled with Sso1p[179–265], which is truncated to remove the H_abc domain. In the four-helix bundle, Sso1p[179–265] and Snc2p[1–93] are assembled with Sec9p[416–504] and Sec9p[571–651], which represent Helix_a (H_a) and Helix_b (H_b), without the intervening loop region. For each combination of SNARE peptides, heterologously tagged SNARE constructs were used in a separate pulldown assay to confirm SNARE-complex assembly (Fig. 8, which is published as supporting information on the PNAS web site). The cytoplasmic SNARE complex, the t-SNARE complex, the four-helix bundle and each of the individual peptides behaved as soluble monomers, remaining in the supernatant fraction after centrifugation at 300,000 × g (Fig. 9, which is published as supporting information on the PNAS web site).

Although the cytoplasmic SNARE complex binds Sec1p-V5-His₆, the four-helix bundle does not (Fig. 3b), suggesting that either or both of the domains H_abc of Sso1p or the loop region Sec9p are required for binding. Two lines of evidence suggest that the H_abc domain is not sufficient for binding. First, the binary t-SNARE complex, which includes H_abc as part of Sso1p[1–265], does not bind Sec1p-V5-His₆ (Fig. 3a, lane 6). Second, the core complex, which retains the Sec9p loop region but lacks H_abc, binds to Sec1p-V5-His₆ suprastoichiometrically (Fig. 4, lane 6). The core complex behaves as a particle, sedimenting at 300,000 × g (Fig. 9, lane 4). Rotary-shadowing electron microscopy revealed oligomers containing six to eight core SNARE complexes (Fig. 10, which is published as supporting information on the PNAS web site). The dimensions of the individual components of the oligomer are in agreement with similar studies of yeast exocytic SNARE complexes (28). The oligomeric nature of the core complex complicates interpretation of the requirement of Sso1p H_abc for Sec1p-V5-His₆ binding. Nonetheless, we could test whether the Sec9p loop region is required by using a combination of SNARE peptides that lacks the Sec9p loop region, but retains H_abc: Sso1p[1–265], Snc2p[1–93], Sec9p[416–504] and Sec9p[571–651]. The resulting SNARE complex assembles (Fig. 8), yet it does not bind Sec1p-V5-His₆ (Fig. 3b, lane 3; Fig. 7a), suggesting that the Sec9p loop region is an important binding determinant.

Fig. 4.

Sec1p does not bind to the open conformation of Sso1p. SNARE-complex binding to purified Sec1p-V5-His₆ was performed as in Fig. 3. For each binding reaction, SNAREs were added to a final concentration of 1 μM. Lane 1: Uncomplexed Sso1p[1–265]. Lane 2: cytoplasmic SNARE complex (SC), assembled with Sso1p[1–265]. Lane 3: Uncomplexed Sso1p[1–265]-Open1. Lane 4: Cytoplasmic SC assembled with Sso1p[1–265]-Open1. Lane 5: Uncomplexed SNARE motif, Sso1p[179–265]. Lane 6: Cytoplasmic SC assembled with Sso1p[179–265], also called the core complex (Fig. 2). Arrows indicate Coomassie-stained bands of the proteins and peptides, which were separated on a 16.5% Tris-tricine gel. Molecular weight markers, the V5 antibody heavy chain (H_c) and light chain (L_c), are indicated.

Sec1p Does Not Bind to the Open Conformation of Sso1p.

To test for a latent Sec1p-interaction site, Sso1p mutants were tested for their ability to interact with purified Sec1p-V5-His₆. Wild-type Sso1p[1–265] is stably folded in the closed conformation (Fig. 5), and it does not bind Sec1p-V5-His₆ (Fig. 4, lane 1). The conformational mutant, Sso1p[1–265]-Open1, favors the open conformation, yet uncomplexed Sso1p[1–265]-Open1 does not interact with Sec1p-V5-His₆ (Fig. 4, lane 3). Complete removal of the H_abc domain, as in Sso1p[179–265], exposes the Sso1p SNARE motif (H₃), but Sso1p[179–265] does not bind to Sec1p-V5-His₆ (Fig. 4, lane 5). All three Sso1p peptides were tested up to 10 μM (Fig. 7a). SNARE complexes made with each of these Sso1p peptides do interact with Sec1p-V5-His₆ (Fig. 4, lanes 2, 4, and 6). Note that the suprastoichiometric binding of SNARE complexes made with Sso1p[179–265] is likely to be a consequence of the oligomeric nature of the core complex (described above and in). With that caveat, it appears that the H_abc truncation and mutations in Sso1p[1–265]-Open1 do not disrupt the Sec1p-binding site. Taken together, these results suggest that, rather than a latent binding motif on Sso1p, Sec1p binds specifically to a site on the assembled SNARE complex, a site that includes the Sec9p loop region.

Discussion

Direct Binding of Yeast Sec1p to the Exocytic SNARE Complex.

The results of these studies demonstrate that Sec1p purified from yeast binds specifically and directly to the exocytic SNARE complex. SNARE-complex binding was reconstituted in vitro with the purified cytoplasmic domain of the ternary SNARE complex, which consists of Sso1p[1–265], Sec9p[416–651], and Snc2p[1–93]. All three components of the SNARE complex are necessary to reconstitute the interaction, because Sec1p-V5-His₆ binds to none of the individual SNAREs or to the binary t-SNARE complex, in contrast to a recent observation (25). Binding studies with truncated SNARE domains revealed that the Sec9p H_a-H_b loop region is required to support an interaction with Sec1p.

A specific interaction between Sec1p and the ternary SNARE complex is consistent with what is observed in yeast. Using an IP protocol with lysates from mutants defective in SNARE-complex assembly, we observed none of the individual SNAREs or the t-SNARE complex coprecipitated with Sec1p. Using the same protocol with wild-type yeast, we found that disassembly of SNARE complexes abolishes Sec1p binding to any of the SNAREs, as expected if binding requires the fully assembled SNARE complex. Likewise, Sec1p concentrates at sites of SNARE-complex assembly, where secretory vesicles fuse with the plasma membrane (22, 26, 29). Furthermore, Sec1p is mislocalized in sec mutants defective for SNARE-complex assembly and shows an increase in polarized localization in mutants that accumulate SNARE complexes (21, 22, 26), supporting the conclusion that Sec1p binds to SNARE complexes. Our ability to reconstitute the specific interaction between Sec1p and the SNARE complex not only supports these in vivo observations, it demonstrates that this interaction does not require other factors, such as additional proteins (30, 31) or lipids.

Another Way for SM Proteins to Interact with SNAREs.

Using wild-type and mutant Sso1p constructs, our binding studies with yeast Sec1p indicate the following: (i) Sec1p has no observable affinity for Sso1p in the closed conformation, ruling out the closed-syntaxin-binding mode characterized for nSec1/Munc-18 (12, 32); (ii) the N-terminal 30 amino acids of Sso1p are not required for Sec1p binding or SNARE function, ruling out the syntaxin peptide-finger-binding mode characterized for Sly1p and Vps45p (16, 17); (iii) Sec1p purified from yeast binds directly and stoichiometrically to the purified SNARE complex, ruling out an indirect binding mode, as suggested for Vps33p (19, 20); and (iv) there is no evidence for an interaction between Sec1p and individual nonsyntaxin SNAREs, as has been observed for Sly1p (23, 33). We also explicitly tested the proposal that Sec1p can bind to the t-SNARE complex (25) and found no evidence, in vivo or in vitro, for an interaction between Sec1p and the t-SNARE complex formed by Sec9p and Sso1p. Notably, we did observe nonspecific interactions between Sec1p-V5-His₆ purified from yeast and recombinant SNARE peptides that were tagged with heterologous sequences (data not shown). We conclude that the interaction observed between Sec1p and the exocytic SNARE complex represents a specific mode of SNARE interactions: the SNARE-complex-binding mode.

Evidence for a SNARE-Complex-Binding Mode Raises New Questions.

The finding that at least one SM protein binds specifically to the assembled SNARE complex raises new questions about the role of SM proteins in vesicle trafficking and membrane fusion. One suggested function is protection of SNARE complexes from disassembly before membrane fusion (21). However, we see no evidence for protection from disassembly when over-expressing Sec1p in yeast (Fig. 11a, which is published as supporting information on the PNAS web site), or when using an in vitro disassembly assay with purified components (Fig. 11b). Alternatively, Sec1p may activate membrane fusion by zippering together the membrane proximal ends of the SNAREs (21, 25, 34, 35). This possibility remains to be tested.

Does the SNARE-complex interaction observed in yeast represent an intermediate common to all SM proteins? If so, we expect to find that other SM proteins use a similar SNARE-complex-binding mode to interact with their cognate SNARE complexes. Indeed, recent findings indicate that the SNARE-complex interactions observed for Sly1p and Vps45p are not simply the result of the SM protein binding to the syntaxin peptide-finger motif, as previously believed (23, 33).

In the cell, SM proteins may function with protein complexes that link vesicle tethering and SNARE-complex assembly. The vacuolar SM protein, Vps33p, is part of a large protein complex known as HOPS (19, 36). HOPS interacts with assembled SNARE complexes and is required for vacuole docking and fusion (37, 38). HOPS binds Vam7p, the vacuole-specific SNARE homologous to neuronal SNAP-25 (37). We find that the Sec1p interaction with SNARE complexes requires a binding site on the plasma membrane-specific SNAP-25 homolog, Sec9p, suggesting that interaction with this t-SNARE may be a common requirement for SM protein function. Like HOPS, the plasma membrane-specific vesicle tethering complex, known as the exocyst, is both a Rab effector (39) and required for SNARE-complex assembly (22). Although Sec1p is not a stable component of the exocyst, it may associate with low affinity to promote SNARE-complex assembly (30). Although we observe no activation of assembly with Sec1p alone (Fig. 12, which is published as supporting information on the PNAS web site), one exocyst protein, Sec6p, binds directly to Sec9p and inhibits assembly of SNARE complexes (40). If the Sec6p–Sec9p interaction represents a regulatory intermediate in SNARE-complex assembly, perhaps association of Sec1p with Sec9p and the exocyst is required for a subsequent step to activate SNAREs for membrane fusion.

Whether SM proteins share common functions will require a combination of mechanistic studies using purified proteins in SNARE complex-assembly and membrane-fusion reactions, plus introduction of mutations designed to test the predictions of a model for SM-dependent SNARE-complex assembly and membrane fusion.

Materials and Methods

Plasmid Construction.

For construction of expression vectors pKLN9 Sso1p[179–265], pKLN10 Sec9p[416–504] and pKLN11 Sec9p[571–651] (K. N. Tomishima and F.M.H., unpublished results), BamHI and EcoRI restriction sites were added to the ends of the SSO1 and SEC9 sequences by PCR, and the products were ligated into the pGEX4T-1 vector (GE Healthcare, Piscataway, NJ). For construction of pYES2/CT Sec1p, BamHI and HindIII restriction sites were added to the ends of the SEC1 sequence by PCR using pNB680 (21) as template. The PCR product was then ligated into the pYES2/CT vector (Invitrogen) and confirmed by sequencing (IDT, Piscataway, NJ).

Immunoprecipitation Protocol.

Seven S. cerevisiae strains were used for the IP experiments (Table 1). For CCY16, CCY17, and CCY18, FHY102 (26) was modified by integration of a triple-myc tag at the C terminus of SEC1, as described (21) and transformed with the plasmids indicated. Cell lysis and IP were performed as described, with at least 50% recovery of Sec1p-MYC₃ using the 9E10 antibody (21). The proteins that co-precipitated with Sec1p-MYC₃ were detected by Western blot analysis. Blots were probed with primary antibodies against Sso1p (16,371; ref. 26); Sec9p (CUMC938) and Snc2p (CUMC6; ref. 41) and with the secondary antibody, peroxidase-conjugated anti-rabbit IgG (Sigma, St. Louis, MO). Antibody recognition was detected by using a chemiluminescence reagent (Pierce, Rockford, IL).

Table 1.

Yeast strains used in this study

Strain	Genotype	Source
CCY16	MATa his3Δ1 leu2Δ0 ura3Δ0 sec1Δ::SEC1MYC URA3 sso1Δ::kanMX4 sso2Δ::kanMX4pMM250 (CEN LEU2 SSO1)	This study
CCY17	MATa his3Δ1 leu2Δ0 ura3Δ0 sec1Δ::SEC1MYC URA3 sso1Δ::kanMX4 sso2Δ::kanMX4pMM266 (CEN LEU2 sso1-open1)	This study
CCY18	MATa his3Δ1 leu2Δ0 ura3Δ0 sec1Δ::SEC1MYC URA3 sso1Δ::kanMX4 sso2Δ::kanMX4pFH24 (CEN LEU2 sso1[31–290])	This study
FHY102	MATa his3Δ1 leu2Δ0 ura3Δ0 sso1Δ::kanMX4 sso2Δ::kanMX4pMM250 (CEN LEU2 SSO1)	M. Munson
INVSc1	MATa/α his3Δ1/ his3Δ1 leu2/leu2 trp1–289/trp1–289 ura3–52/ura3–52	Invitrogen
NY1689	MATα his3-Δ200 leu2–3, 112 trp1 ura3–52 sec1Δ::SEC1MYC URA3	P. Novick
NY1691	MATα sec18–1 his3-Δ200 leu2–3, 112 trp1 ura3–52 sec1Δ::SEC1MYC URA3	P. Novick
NY2228	MATα sec5–24 ura3–52 sec1Δ::SEC1MYC URA3	P. Novick

Protein Purification.

Recombinant SNARE peptides Sso1p[1–265], Sso1p[1–265]-Open1 Snc2p[1–93] and Sec9p[416–651] were expressed and purified as described (10, 21, 26). Sso1p[179–265], Sec9p[416–504] and Sec9p[571–651] were expressed in E. coli BL21 cells as GST fusion proteins in a pGEX4T-1 expression vector as described by the manufacturer (GE Healthcare). After removal of GST, Sec9p[416–504] and Sec9p[571–651] were further purified by using Benzamadine Sepharose (GE Healthcare) and Sso1p[179–265] by anion exchange (MonoQ, GE Healthcare). Purified peptides were concentrated by ultrafiltration (Millipore Amicon, Billerica, MA; Ultra-15) and stored at −80°C.

His-tagged Sec1p (Sec1p-V5-His₆) was expressed in yeast by using the pYES2/CT Sec1p plasmid transformed into the S. cerevisiae strain, INVSc1, as described by the manufacturer (Invitrogen). After a 14-h induction in galactose, cells were harvested and resuspended in wash buffer (50 mM Hepes, pH 7.4/20 mM NaF/20 mM NaN₃). Washed cells were lysed in buffer A (50 mM Hepes, pH 7.4/150 mM KCl/10% glycerol/30 mM imidazole) plus 1 mM PMSF and yeast protease inhibitor mixture (21), using a microfluidizer, model M-110Y (Microfluidics, Newton, MA). The lysate was clarified by centrifugation; after 10,000 × g for 30 min (Sorvall, Guelph, ON, Canada; rotor SS-34), the supernatant fraction (S10) was centrifuged for another 45 min at 100,000 × g (Beckman, Fullerton, CA; rotor Ti-70). Sec1p-V5-His₆ from the S100 was purified on a nickel-charged chelating column (HiTrap, GE Healthcare), with a gradient of 10–100% buffer B (50 mM Hepes, pH 7.4/150 mM KCl/250 mM imidazole/10% glycerol). For storage, 2 mM DTT and 1 mM EDTA were added and the aliquots were frozen at −80°C.

SNARE Binding with Immobilized Sec1p-V5-His₆.

To standardize the amount of Sec1p-V5-His₆ in the binding reaction, ≈0.4 μg of nickel purified Sec1p-V5-His₆ was immunoaffinity-purified by using 1.8 μg of V5 monoclonal antibody (Invitrogen). Excess Sec1p-V5-His₆ was washed from the resin with SNARE-binding buffer (50 mM Hepes, pH 7.4/100 mM NaCl/0.5% IGEPAL/0.5 mM DTT/1 mM EDTA). Before addition to the binding reaction, SNARE complexes were assembled, as described (10). The time required for completion of SNARE-complex assembly was measured separately by using a SNARE-assembly assay (Supporting Text, which is published as supporting information on the PNAS web site). One micromolar SNARE or SNARE complex was added to immobilized Sec1p-V5-His₆ and incubated for 2 h at 4°C. The reactions were separated by SDS/PAGE and stained with Coomassie blue R250.

Abbreviation

IP

immunoprecipitation.