Double NPY motifs at the N-terminus of Sso2 synergistically bind Sec3 to promote membrane fusion

Exocytosis is an active vesicle trafficking process by which eukaryotes secrete materials to the extracellular environment and insert membrane proteins into the plasma membrane. The final step of exocytosis in yeast involves the assembly of two t-SNAREs, Sso1/2 and Sec9, with the v-SNARE, Snc1/2, on secretory vesicles. The rate-limiting step in this process is the formation of a binary complex of the two t-SNAREs. Despite a previous report of acceleration of binary complex assembly by Sec3, it remains unknown how Sso2 is efficiently recruited to the vesicle-docking site marked by Sec3. Here we report a crystal structure of the pleckstrin homology (PH) domain of Sec3 in complex with a nearly full-length version of Sso2 lacking only its C-terminal transmembrane helix. The structure shows a previously uncharacterized binding site for Sec3 at the N-terminus of Sso2, consisting of two highly conserved triple residue motifs (NPY: Asn-Pro-Tyr). We further reveal that the two NPY motifs bind Sec3 synergistically, which together with the previously reported binding interface constitute dual-site interactions between Sso2 and Sec3 to drive the fusion of secretory vesicles at target sites on the plasma membrane. Significance SNARE assembly, which involves one v-SNARE with two t-SNARE proteins, drives the fusion of vesicles to target compartments. The rate-limiting step in SNARE assembly is the assembly of the two t-SNARE proteins on the target membrane. Previous studies in yeast showed that Sec3, a component of the exocyst vesicle tethering complex, directly interacts with the t-SNARE protein Sso2 to promote fast assembly of an Sso2-Sec9 binary t-SNARE complex. This paper presents a new crystal structure of the Sec3 PH domain in complex with a nearly full-length version of Sso2, which reveals a previously unknown binding site for Sec3 at the N-terminus of Sso2. Our work demonstrates that the dual-site interactions between Sso2 and Sec3 plays an essential role in promoting the fusion of secretory vesicles at target sites on the plasma membrane.


Abstract
Exocytosis is an active vesicle trafficking process by which eukaryotes secrete materials to the extracellular environment and insert membrane proteins into the plasma membrane.
The final step of exocytosis in yeast involves the assembly of two t-SNAREs, Sso1/2 and Sec9, with the v-SNARE, Snc1/2, on secretory vesicles. The rate-limiting step in this process is the formation of a binary complex of the two t-SNAREs. Despite a previous report of acceleration of binary complex assembly by Sec3, it remains unknown how Sso2 is efficiently recruited to the vesicle-docking site marked by Sec3. Here we report a crystal structure of the pleckstrin homology (PH) domain of Sec3 in complex with a nearly fulllength version of Sso2 lacking only its C-terminal transmembrane helix. The structure shows a previously uncharacterized binding site for Sec3 at the N-terminus of Sso2, consisting of two highly conserved triple residue motifs (NPY: Asn-Pro-Tyr). We further reveal that the two NPY motifs bind Sec3 synergistically, which together with the previously reported binding interface constitute dual-site interactions between Sso2 and Sec3 to drive the fusion of secretory vesicles at target sites on the plasma membrane.

Introduction
The cytoplasm in eukaryotic cells is compartmentalized into distinct membrane bound organelles. Inter-organelle material exchange is carried out primarily through membrane traffic in which membrane bound transport vesicles bud from a donor compartment and are delivered to a specific acceptor compartment. Upon arriving at the destination, cargo-packed vesicles are first recognized and caught by tethering factors situated on the target membrane, which then hand the captured vesicles over to the soluble N-ethylmaleimide-sensitive factor-attachment protein receptor (SNARE) proteins that drive membrane fusion [1][2][3][4].
There are several types of SNARE proteins, one of which is attached to the membrane of vesicles (v-SNARE), and the others are on the target membrane (t-SNARE).
In neuronal exocytosis, fusion of synaptic vesicles to the presynaptic plasma membrane is driven by the assembly of a four-helix bundle containing two t-SNAREs, syntaxin-1 and SNAP-25, on the target membrane and the v-SNARE, synaptobrevin, on synaptic vesicles [5,6]. Extensive studies have established that SNARE assembly is tightly regulated by multiple auxiliary proteins, including the Sec1/Munc18 (SM) family of proteins, tethering factors, and small GTPases [7][8][9][10][11][12]. Munc18 is a chaperone protein that maintains syntaxin-1 in an activated conformation and passes it to its cognate SNARE partners for assembly to catalyze membrane fusion [13][14][15][16][17].
The exocytic vesicle-docking site in yeast is marked by the octameric exocyst complex, which belongs to the CATCHR family of multi-subunit tethering proteins [18][19][20].
The main function of the exocyst is to capture secretory vesicles at sites of cell surface growth, which include the tip of the daughter cell early in the cell cycle, and the mother-daughter cell junction late in the cycle [21]. The two t-SNAREs for exocytosis in yeast are Sso1/2 and Sec9, which are homologs of syntaxin-1 and SNAP-25, respectively. The v-SNARE attached to secretory vesicles in yeast is Snc1/2, which is equivalent to synaptobrevin in neuronal exocytosis.
Our previous work showed that one of the exocyst components, Sec3, promotes SNARE assembly by interacting with the t-SNARE Sso2 [22]. Sec3 consists of an Nterminal pleckstrin homology (PH) domain, a central putative coiled coil, and a C-terminal helical domain. Like syntaxin-1 and other related t-SNAREs, Sso2 consists of four helices, with the first three (Habc) forming an inhibitory domain and the last (H3) serving as the SNARE motif that interacts with the other two SNAREs during membrane fusion. We have shown that the Sec3 PH domain binds to the auto-inhibited four-helix bundle of Sso2 and promotes a conformational change of the linker between Hc and H3 of Sso2 via an allosteric effect [22]. This change promotes the release of the SNARE motif (H3) of Sso2 and substantially accelerates the formation of the initial binary complex between H3 of Sso2 and the two helices of the other t-SNARE, Sec9. However, it remains unclear how Sso2 is initially recruited to the vesicle target sites marked by the exocyst to drive the efficient fusion reaction between secretory vesicles and the plasma membrane.
Here we report our structural studies of Sec3-PH in complex with a nearly fulllength construct of Sso2 (aa1-270), which lacks only its C-terminal transmembrane region (aa271-295). Our crystal structure of this Sec3/Sso2 complex reveals a previously unknown binding site for Sec3 on Sso2 in addition to the one on its four-helix bundle as reported in our previous work [22]. This extra binding site is located at the N-terminal end of Sso2 and consists of two highly conserved NPY (i.e. Asn-Pro-Tyr) motifs. These NPY motifs are connected to the helical core of Sso2 (i.e. Habc and H3) via a long variable linker. In the two heterodimeric complexes present in the crystal structure, the two NPY motifs of the two Sso2 molecules bind individually to a similar conserved hydrophobic pocket on the two Sec3 molecules. Interestingly, however, our in vitro interaction studies using synthetic polypeptides and recombinant Sec3-PH protein demonstrated that each NPY motif alone bound Sec3 much more weakly than the two NPY motifs together. The importance of the interaction between the NPY motifs of Sso2 and Sec3 was confirmed by a series of in vivo assays in yeast.
Overall, our work has uncovered a new interaction interface and thus establishes dual-site interactions between Sec3 and Sso2, which also suggests potentially an extra regulatory step in exocytic membrane fusion. Binding of the NPY motifs of Sso2 allows efficient recruitment of the t-SNARE protein to the vesicle-docking site on the plasma membrane to facilitate vesicle fusion.

Crystal structure reveals two NPY motifs at the N-terminus of Sso2 bound individually to the Sec3 PH domain
We previously reported the structure of Sso2-HabcH3 (aa36-227) in complex with the PH domain of Sec3 (aa75-320) [22]. Recently we crystallized another complex of the two proteins using a longer version of Sso2 (aa1-270), which contains all Sso2 sequence except for its C-terminal transmembrane part, together with a shorter Sec3 PH domain (aa75-260) ( Fig. 1A and Fig. S1A). A stable binary complex was obtained via size exclusion chromatography (SEC) (Fig. S1B) (Fig. 1B). 2F o -F c electron density maps have a high quality, and sidechains of most residues in both Sec3 and Sso2 can be confidently built (Fig. 1C).
Primary sequence alignments of Sso2 homologs from various yeast species reveal two conserved three-residue motifs toward the N-terminus of Sso2, which we name as NPY motifs. These double NPY motifs are connected to the highly conserved core of Sso2 (i.e. Habc and H3) via a nonconserved linker with variable lengths in different homologs ( Fig. 2A; Fig. S2). These NPY motifs bind individually to Sec3 in the two structural copies (Fig. 2B, E). In one complex structure residues 4-8 of Sso2 show clear densities in the 2F o -F c map (Fig. 2C); in the other structure we could unambiguously trace sidechains of residues 10-14, but for residues 6-9 we could build only the backbone atoms ( Fig. 2F). Despite variations in the flanking residues of these NPY motifs, the NPY cores adopt essentially the same conformation and share similar contacts with ordered solvent molecules and neighboring residues from Sec3 (Fig. S3). Particularly, the tyrosine residues in both cases (i.e. Y7 and Y13) are docked into a conserved hydrophobic pocket on the concave surface on Sec3 (Fig. 2D, G). Overall, the NPY motif adopts a T-shaped conformation, with its broad top part shaped by the asparagine (N) and the proline (P) residues, which is stabilized by a hydrogen bond between the carboxyal group of the asparagine and the amide proton of the proline (Fig. 2H). The tryosine residue sticks out to form a pin-like structure that fits neatly in the pocket on Sec3.

The two NPY motifs of Sso2 binds synergistically to Sec3
To determine how the two NPY motifs of Sso2 interact with Sec3, we carried out isothermal titration calorimetry (ITC) assays using synthetic polypeptides of Sso2 and recombinant Sec3-PH purified from bacteria (Fig. 3). Wild type (WT) double NPY motifs of Sso2 (aa1-15) bound Sec3 with a dissociation constant (Kd) of 21.1 µM (Fig. 3A).

Mutation of Sso2 NPY motifs inhibits cell growth as well as secretion of Bgl2 and invertase
To assess the in vivo role of the interactions between Sec3 and the two NPY motifs of Sso2, we constructed various sso2 alanine substitution mutations in a yeast integrating vector and introduced them into a yeast strain deleted for the paralogous gene, SSO1.
The mutations were incorporated into the endogenous SSO2 locus by the loop-in/loopout method, leaving the surrounding sequence entirely unaltered ( Fig. 4A; Fig. S4A) [23].
The sso1Δ sso2 mutants were tested for growth at both 25 o C and 37 o C ( Fig. 4B; Fig.   S4B). No effect was observed with single, double or triple mutations in the first NPY domain (sso1Δ sso2M1-M4), however changing all four residues to alanine (sso1Δ sso2M5) resulted in reduced growth at 37 o C (Fig. S4B). Mutating all four residues of the second NPY domain to alanine (sso1Δ sso2M6) did not affect growth at any temperature, however changing all residues of both the first and second NPY domain to alanine (sso1Δ sso2M7) resulted in significantly reduced growth at both 25 o C and 34 o C and severely impaired growth at 37 o C (Fig. 4B). The synergistic effects of eliminating the first and second NPY motif of Sso2 suggest that both motifs are functional and at least partially redundant.
We next assayed the export of two different cell surface enzymes, Bgl2 and invertase. Bgl2 is synthesized and secreted constitutively, while the synthesis of invertase is under hexose repression. Both enzymes become trapped at the cell surface by the cell wall glucan and this external pool can be released by treatment of cells with exogenous glucanase, while any internal pool remains associated with the resulting spheroplasts [24].
Using western blot analysis to measure the internal and external pools of Bgl2, we found that the secretory efficiency at 37 o C generally paralleled growth: sso1Δ sso2M1-M6 showed only a modest accumulation of an internal pool, while sso1Δ sso2M7 accumulated a significantly larger internal pool (Fig. 4C, D; Fig. S4C, D).
To assay invertase secretion we started with cells grown at 25 o C in 5% (w/v) glucose to repress synthesis and then shifted to 0.1% (w/v) glucose to derepress synthesis and simultaneously shifted the cells to 37 o C. Using these conditions we found that sso1Δ sso2M1-M4 and sso1Δ sso2M6 were not significantly different from the sso1Δ SSO2 control, while sso1Δ sso2M5 showed a minor defect in invertase secretion and sso1Δ sso2M7 showed a more substantial defect ( Fig. 4E; Fig. S4E).

Mutations of the Sso2 NPY motifs cause polarized accumulation of secretory vesicles
Defects on the secretory pathway are typically associated with the accumulation of membrane bound intermediates [25]. Loss of function of exocytic SNAREs, including Sso1 and Sso2 leads to the accumulation of secretory vesicles [26]. Secretory vesicles are normally delivered to sites of polarized cell surface growth, such as the tip of the bud, early in the cell cycle and the neck separating the mother cell and bud, late in the cell cycle. Thin section electron microscopy revealed an accumulation of vesicles in the mutants that mirrored their growth: the sso1Δ SSO2 control and the sso1Δ sso2M6 mutant had relatively few vesicles per cell section, while sso1Δ sso2M5 and sso1Δ sso2M7 had many more ( Fig. 5A-E). The vesicles were similar in size in the different strains ( Fig. 5F) and were found preferentially within the buds of small budded cells ( Fig.   5A-D).

Mutations in the NPY motifs of Sso2 affect Snc1 recycling
In addition to the export of newly synthesized cell surface proteins, secretory vesicles are also important for recycling certain plasma membrane proteins back to the cell surface after they have been internalized by endocytosis. The Snc1 v-SNARE has been shown to rapidly cycle from secretory vesicles to the plasma membrane, and then into endocytic vesicles from which it is recycled through the Golgi into a new round of secretory vesicles [27]. Under normal growth conditions, Snc1 is predominantly found on the plasma membrane, with only a minor pool in internal structures. Impeding any step in the cycle leads to a shift in the steady state distribution of GFP-Snc1. We examined the distribution of GFP-Snc1 in various sso1Δ sso2 mutants. In the sso1Δ SSO2 control and the sso1Δ sso2M6 mutant, GFP-Snc1 was mostly on the plasma membrane with a small number of internal, patch-like structures apparent (Fig. 6A, C). In contrast, the sso1Δ sso2M5 and sso1Δ sso2M7 mutants showed a much greater fraction of cells with internal patches of GFP-Snc1, presumably representing concentrations of secretory vesicles, and a greatly reduced localization to the plasma membrane ( Fig. 6B, D, E).

The sso2 mutations have no effect on the actin-independent localization of Sec3
Secretory vesicles are delivered to sites of cell surface growth by the type V myosin, Myo2, moving along polarized actin cables [28]. Loss of actin or Myo2 function leads to the rapid depolarization of a vesicle marker, such as the Rab GTPase Sec4 [29]. Sec3, in contrast remains associated with the tips of small buds and the necks of large buds even after actin polymerization has been blocked by addition of Latrunculin A (LatA) [30].
We determined if the interactions between Sec3 and the NPY motifs of Sso2 are required for the actin-independent localization of Sec3. GFP-Sec4 and Sec3-3×GFP were expressed in sso1Δ SSO2, sso1Δ sso2M5 and sso1Δ sso2M7 cells. Localization was evaluated after treatment with either LatA or DMSO for 15 min. The polarized localization of GFP-Sec4 was lost after treatment with LatA ( Fig. 7A, B). While the localization of Sec3-3×GFP was largely resistant to LatA treatment in the control and sso1Δ sso2M5 mutant cells, the polarization of Sec3-3×GFP in sso1Δ sso2M7 was only slightly reduced by LatA treatment (Fig. 7C, D). These results demonstrate that the NPY motifs of Sso2 do not play a major role in the recruitment of Sec3 to sites of polarized growth. Notably, prior studies have shown that the actin-independent localization of Sec3 involves its interaction with the Rho1 and Cdc42 GTPases [31,32].

The NPY motifs play an essential role in stabilizing the interaction between Sso2 and Sec3
In the structure with the second NPY motif bound to Sec3 we see extra electron densities beyond the bound NPY motif. However, the quality of the map in that part was poor and we could only build the main chains for residues 6-9 (Fig. 8A). We found that this "fuzzy" part upstream of the bound NPY motif is in close contact with the C-terminal tip of the SNARE motif (i.e. H3) of Sso2 (Fig. 8B). To find out whether the NPY motifs are essential in stabilizing the interaction between Sso2 and Sec3, we carried out ITC experiments using either WT or M7 mutant of Sso2 with the Sec3 PH domain. Our results show that WT Sso2 bound Sec3 robustly, with Kd ~2.7 µM (Fig. 8C). However, the M7 mutant of Sso2 (aa1-270) did not interact with Sec3 at all (Fig. 8D).
We further examined their interactions using two other independent methods. Our electrophoresis mobility shift assays (EMSA) show that more and more WT Sso2 shifted up to the complex band with increasing amounts of Sec3 in the mixtures. However, no complex was formed when we mixed the M7 mutant of Sso2 with Sec3 ( Fig. S5A, B).
Consistently, our further test using size exclusion chromatography also shows that Sec3 could form complex with only WT but not the M7 mutant of Sso2 (Fig. S5C, D).
To investigate whether the M7 mutant of Sso2 also affects its interaction with Sec3 in vivo, we carried out co-immunoprecipitation of Sec3-3×Flag and Sso2 using yeast cell lysate. Our results show that, in contrast to the clear signal of WT Sso2 pulled down by Sec3-3×Flag, the pull-down band for the M7 mutant was much weaker and near the level of the negative control in which the Sec3-3×Flag was absent (Fig. S6).

Discussion
Sec3 is a subunit of the octameric exocyst complex, a tethering factor that marks the docking site for secretory vesicles in exocytosis. Sec3 is recruited to these sites by binding to the membrane-anchored small GTPases Rho1/Cdc42 and phosphoinositides on the plasma membrane [31][32][33][34]. Sso2, similar to its homologous t-SNARE in neuronal exocytosis, bears a C-terminal transmembrane helix. We have previously reported crystal structures of the Sec3 N-terminal PH domain in complex with the closed form of Sso2 as a four-helix bundle [22]. We found that Sec3 destabilizes the linker between Hc and H3 of Sso2 via an allosteric effect, which promotes the assembly of the binary complex between Sso2 and the other t-SNARE, Sec9, and thus drastically accelerates subsequent full SNARE assembly with the v-SNARE Snc1/2 to drive fusion of secretory vesicles with the plasma membrane. However, it remains unknown whether the conserved N-terminal extension of Sso2 also participates in the Sso2/Sec3 interaction and how Sso2 is effectively recruited to Sec3.
Here we report a new crystal structure of the Sec3 PH domain in complex with a nearly full-length version of Sso2 that contains all sequence except for the C-terminal transmembrane helix. In addition to the interaction between the helical bundle of Sso2 and Sec3 observed in the previous study, we found an extra interaction interface between two conserved NPY motifs at the N-terminal end of Sso2 and a hydrophobic pocket on Sec3 (Fig. 1). The NPY motifs are connected to the highly conserved helical core of Sso2 via a non-conserved linker with variable lengths (roughly 15-40 residues), which is invisible in our crystal structure, suggesting that it is mobile within the crystal structure.
There are two copies of the Sso2/Sec3 complex in our crystal structure. The two NPY motifs were independently bound to Sec3 in these two complex structures (Fig. 2).
Despite variations in the flanking residues, the cores of the two motifs, i.e. residues Asn, Tyr and Pro, adopt essentially the same conformation and have similar hydrogen bond networks with the neighboring Sec3 residues and water molecules (Fig. S3). Our ITC experiments show that either of the two NPY motifs alone could bind Sec3 tightly, although with binding affinities 3-4 fold lower than the polypeptide with both motifs (Fig.   3). Similarly, mutation of either motif to alanine also substantially reduced the interaction of the N-terminal part of Sso2 with Sec3, whereas simultaneous mutation of both motifs completely abolished the interaction. Together these data suggest that the two NPY motifs bind Sec3 synergistically.
To further understand the function of the NPY motifs in exocytosis, we carried out a series of in vivo assays using yeast strains carrying mutations of Sso2 in its NPY motifs.
While mutation of either NPY motif alone only slightly reduced cell growth, simultaneous mutation of both NPY motifs severely impaired cell growth, particularly at 37 o C (Fig. 4B).
We also explored how mutations of the NPY motifs of Sso2 influence protein secretion using Bgl2 and invertase as reporters. Our data show that mutation of both NPY motifs simultaneously substantially affects cell secretion efficiency, with a significantly larger internal pool of Bgl2 failing to reach the cell surface, whereas mutation of either motif alone yielded only a modest accumulation of an internal pool (Fig. 4C, D; Fig. S4C, D).
Similar effects were observed in invertase secretion, where the double NPY mutant M7 showed a more substantial defect than all other mutants ( Fig. 4E; Fig. S4E).
We further checked how the sso2 NPY mutations affect fusion of secretory vesicles to the target sites on the plasma membrane. Mutation of both NPY motifs resulted in the accumulation of many vesicles within the cell. Similar results were also observed for the mutation of the first NPY motif, whereas that of the second NPY motif showed no significant effect (Fig. 5). Consistently, we found that both the double NPY mutation and mutation of the first NPY motif caused accumulation of GFP-Snc1 patches within the cytoplasm (Fig. 6). Taken together, we conclude that the NPY motifs of Sso2 play an essential role in the secretory pathway. Notably, however, the interaction between Sec3 and the NPY motifs of Sso2 is dispensable for the actin-independent localization of Sec3 ( Fig. 7), which is consistent with previous reports that recruitment of Sec3 to the plasma membrane is determined by its interaction with the small GTPases Rho1 and Cdc42 as well as phosphoinositides on the membrane [31][32][33][34].
Although the M7 mutant in the synthetic N-terminal part of Sso2 (aa1-15) disrupts its interaction with Sec3, the major interaction interface between the helical bundle of Sso2 and the Sec3 PH domain remains unchanged (Fig. 1B). Therefore, it was intriguing to us why the M7 mutant showed such a strong deleterious effect in vesicle trafficking.
Our ITC data reveal that the M7 mutant of Sso2 (aa1-270) completely abolished its interaction with the Sec3 PH domain (Fig. 8C, D). The disrupted binding of the M7 mutant with Sec3 was further confirmed by two other in vitro experiments using purified recombinant proteins (Fig. S5), as well as by our co-immunoprecipitation data (Fig. S6).
All these demonstrate that the NPY motifs play an essential role in stabilizing the interaction between Sso2 and Sec3. This might be explained by what was observed in the crystal structure of the Sso2/Sec3 complex, where the NPY motifs are in close contact with the C-terminal tip of the SNARE motif (i.e. H3) of Sso2 and may thus stabilize their interaction as explained below (Fig. 8A, B).
The dual interaction interfaces between Sso2 and Sec3 are reminiscent of what has been seen in the structures of syntaxin-1 and Tlg2 in complex with their partner SM proteins Munc18 and Vps45, respectively (Fig. 9A-C). In the latter two complex structures, a short peptide at the N-terminus of Munc-18 and Vps45, which was named "N-peptide", forms a short helix and binds distally to the backside of the first domain of Munc18 or Vps45, opposite to their interaction sites with the helical bundles of the SNARE proteins [35,36]. In contrast, the binding site for the NPY motifs of Sso2 is on the same side of the Sec3 PH domain as the aforementioned direct contact with the C-terminal part of H3 of Sso2 (Fig. 9D). Furthermore, the Sec3 PH domain is a single globular domain and much smaller than Munc18 and Vps45, both of which contain three domains that are folded into a horseshoe-like conformation. The helical bundle of syntaxin-1 and Tlg2 is clamped between the two tips of the "horseshoe". Additionally, in those two complex structures, the C-terminal extensions of the SNARE motifs (i.e. H3) form a short helix and insert into a hydrophobic pocket deeply inside the "horseshoe" (Fig. 9B, C). These together ensure a strong interaction between the two t-SNAREs and their partner SM proteins. In contrast, the interaction interface between Sec3 and Sso2 is very small, and thus their interaction is presumably much weaker than that in the Munc18/syntaxin-1 and Vps45/Tlg2 complexes. Our ITC data show that double mutation of both NPY motifs completely abolished the interaction between Sso2 and Sec3 (Fig. 8C, D). Consistently, our in vivo data demonstrate that simultaneous mutation of both NPY motifs substantially impaired protein secretion as well as recycling of cell surface proteins. We therefore think that the NPY motifs play an important role in stabilizing the interaction of Sso2 with Sec3 by providing an additional binding site, which ensures effective recruitment of Sso2 to vesicle docking sites on the plasma membrane.
Given that binding of Sec3 destabilizes the linker connecting H3 to Hc and prepares H3 for its subsequent interaction with Sec9, we think that another possible role for the NPY motifs, which are packed tightly against the C-terminal tip of the SNARE motif H3 (Fig. 8A, B), might be to serve as a stopper to hold the destabilized SNARE motif in place to prevent a premature release of H3 before the right time comes for it to form a binary complex with the other t-SNARE protein Sec9.
Notably, the N-peptide motif is ubiquitously present in all syntaxins that interact with SM proteins [37][38][39][40]. It serves as an initiator to recruit SM proteins to their SNARE partners to facilitate their subsequent assembly [41]. Given that the NPY motifs are connected to the rest of Sso2 via a long variable linker, we hypothesize that they may similarly function like fishing hooks to search for Sec3 around the membrane-anchored Sso2 (Fig. S7). Once the "hooks" find and bind to Sec3, Sso2 would be locally restrained, which would promote the binding of Sec3 to the helical bundle of Sso2. This would in turn lead to the destabilization of the linker between Hc and H3 of Sso2 and thus promote the assembly of the binary complex between Sso2 and Sec9 to facilitate the final formation of the full SNARE complex with the v-SNARE Snc1/2, which eventually drives the fusion of secretory vesicles with the plasma membrane.

Molecular cloning of expression constructs for in vitro assays
The Sso2 sequence excluding only its C-terminal transmembrane domain (aa1-270) and the Sec3 PH domain sequence (aa75-260) were both sub-cloned into the pET-15b vector (Novagen) between the NdeI and BamHI sites. This plasmid provides an Nterminal His6 tag followed by a thrombin cleavage site prior to the target proteins. All constructs were validated by DNA sequencing.

E. coli strain BL21(DE3) cells harboring the expression plasmids for Sso2 and
Sec3 were cultured in Luria Broth (LB) medium containing 50 mM ampicillin at 37°C to an OD600 of 0.6-0.8. Over-expression of the target proteins was induced using 0.5 mM isopropylthio-β-d-galactoside (IPTG) and cultures were incubated at 18°C overnight.
Cells were harvested by centrifugation (6,000×g, 12   To generate protein complex for crystallization, purified domains of Sec3 and Sso2 were mixed in a molar ratio of 2:1. After incubation at 4°C for one hour, the mixture was loaded on a Superdex S-200 16/60 column (GE Healthcare) pre-equilibrated with the same running buffer as above. Elution fractions were checked on a 15% (w/v) SDS-PAGE gel. The first elution peak containing both proteins were pooled and concentrated to 10-weight cutoffs.

Crystallization, data collection, and structure determination
Concentrated protein of the Sec3-Sso2 complex (~12 mg/ml) was subjected to extensive crystallization screening trials using commercial crystallization kits. Initial France. Data reduction was carried out using the XDS program [42].
For structure determination, the maximum-likelihood molecular replacement by PHASER [43] was conducted using our previously determined structures of Sec3 and Sso2 (PDB code: 5M4Y) as the searching model [22]. The structural models were carefully checked; all different regions as well as extra parts in the models where electron densities were clearly visible were manually built using the program COOT [44]. Refinement was carried out by Phenix.refine [45] using data of 20-2.19 Å. All subsequent structure analyses and figure generations were carried out using Pymol (http://www.pymol.org). The details of data collection and refinement statistics are summarized in Supplementary Table 1.

Isothermal titration calorimetry (ITC)
All ITC measurements were conducted on a MicroCal PEAQ-ITC microcalorimeter  with a flow rate of 0.5 ml/min. Samples from each elution peak were checked on an SDS PAGE gel to visualize protein content in that peak. Plasmids carrying the mutant alleles were linearized by digestion with the MscI enzyme to promote integration into the SSO2 locus and introduced into yeast strains (an sso1 strain for mutants sso2M1-M6 or wt for sso2M7) by the lithium acetate method.

Construction of strains and plasmids for in vivo assays
Transformants were selected on SC-Ura plates. Multiple independent transformants were grown in SC-Ura medium overnight. To select for Ura-"loop-out" segregants, cells (100 µl) were collected, washed in sterile water, and then plated on YPD+5-FOA plates.
Integration and loop-out events leading to the genomic expression of the sso2 mutations were verified by sequencing PCR products. Due to genetic instability issues, we used a two-step procedure to construct an sso1 sso2M7 strain. A wt strain was transformed with plasmid NRB1659 and then Ura-cells were selected by growth on 5-FOA plates.
The selected sso2M7 strain was crossed with an sso1SSO2 strain and, after dissection haploid sso1 sso2M7 spores were identified by PCR analysis.

Growth test
Mutants were grown in yeast extract peptone dextrose (YPD) medium overnight to

Bgl2 secretion assay
The Bgl2 secretion assay was carried out as previously described by Yuan et al [24]. Briefly, 15 mL of yeast cells were grown at 25°C in YPD medium overnight to early log phase (∼0.5 OD600/mL), then shifted to 37°C for 90 min. 6 OD600 units of cells from each strain were harvested by centrifugation at 900 × g for 5 min. Cell pellets were resuspended in 1 mL of ice-cold 10 mM NaN3 and 10 mM NaF, and then incubated on ice for 10 min. The cell suspension was transferred to 1.5 ml microfuge tubes, pelleted, and resuspended in 1 mL of fresh prespheroplasting buffer consisted of 100 mM Tris-HCl (pH 9.4), 50 mM β-mercaptoethanol, 10 mM NaN3, and 10 mM NaF, then incubated on ice for 15 min. Cells were then pelleted and washed with 0.5 mL of spheroplast buffer (50 mM KH2PO4-KOH (pH 7.0), 1.4 M sorbitol, and 10 mM NaN3). Cells were resuspended in 1 mL of spheroplast buffer containing 167 µg/mL zymolyase 100T (Nacasai Tesque) and incubated at 37°C in a water bath for 30 min. Spheroplasts were spun down at 5,000 × g for 10 min, and 100 µL of the supernatant from each tube was transferred into a new 1.5ml tube and mixed immediately with 34 µL of 4× SDS sample buffer (the external pool).
All of the remaining supernatant was discarded. The pellets (spheroplasts) were resuspended in 100 µL 1×SDS sample buffer (the internal pool). Samples were boiled for 10min and proteins were separated on a 10% SDS/PAGE gel. Bgl2 was visualized by Western blotting with anti-Bgl2 rabbit polyclonal antibody at 1:5,000 dilution (provided by the laboratory of Randy Schekman, University of California, Berkeley). For quantitation of Bgl2, images of western blots were captured using the ChemiDoc system (Bio Rad) and multiple images were collected to ensure an unsaturated signal. Serial dilutions of a control sample were run in parallel to establish a standard curve. The electrophoretic bands were quantitated using ImageJ software (https://imagej.nih.gov).

Invertase Secretion Assay
The invertase assays were performed as previously described by Yuan et al [24].
Yeast strains were grown at 25 o C in YP medium containing 5% (w/v) glucose overnight to early log phase (∼0.5 OD600/ml). Then, 1OD600 unit of cells was transferred to 2 sets of 15 ml centrifuge tubes and pelleted at 3500 rpm for 5 min. The first set of cells were

Electron Microscopy
Control (sso1/SSO2) or sso2 mutant cells were grown at 25°C in YPD to an OD600 of ∼0.5 and then processed for EM study as previously described [24]. In brief, ∼10 OD600  Quantitation analysis was done as described above. Three separate experiments were used to calculate the SD.

Co-immunoprecipitation
Strains were grown at 25°C overnight to an OD600 around 1.0. 70 OD600 units of cells were collected from each strain. Cell lysates were prepared as described previously with modifications [22,46]. Briefly, cell pellets were washed once in cold water and resuspended in ice-cold lysis buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 1 mM EDTA, 5 mM NaF, 1mM sodium pyrophosphate, and 1 mM DTT) and a protease inhibitor cocktail (Roche). Cells suspensions were transferred to 2-ml screw cap tubes containing prewashed 2 mg zirconia/silica 0.5 mm beads. Cell were lysed using a bead beater. 1% (v/v) Triton X-100 was added to the cell lysates and incubated for 15 min at 4°C. The cell lysates were then spun at 20,000×g for 30 min and supernatants were incubated with 10 µl prewashed anti-Flag agarose beads (Sigma, A2220) at 4°C for 3 h. Beads were washed five times with lysis buffer containing 0.1% (v/v) Triton X-100. Proteins bound on the beads were eluted with 1× sample buffer. Proteins were detected with anti-Flag antibody (Sigma, F1804, monoclonal, 1:1,000) or anti-Sso antibody (rabbit antiserum, 1:2,000).

Data availability
Coordinates and structure factors for the Sec3-Sso2 complex have been deposited in the Protein Data Bank with accession code (https://www.rcsb.org/structure/7Q83).