Conserved Prefusion Protein Assembly in Regulated Exocytosis

Plasmids and Protein Preparation

Plasmids encoding glutathione S-transferase (GST) fusion proteins with syntaxin1A (aa 1–261), synaptobrevin2 (aa 1–96), SYT1 C2AB (aa 95–421), C2B (aa 248–421), and complexin2 were described previously (Hu et al., 2002b; Rickman et al., 2004a). The full-length SNAP-25A and SNAP-25B were kind gifts of Drs. R. Jahn (Max-Planck-Institute for Biochemical Chemistry, Gottingen, Germany) and Y.-K. Shin (Iowa State University, Ames, Iowa) (Fasshauer et al., 1999; Kweon et al., 2003). GST-SNAP-25, carrying cysteine-to-alanine mutations, exhibits improved expression in bacteria with no changes detected in biochemical or biophysical properties (Fasshauer et al., 1999). His-tagged versions of syntaxins 1A and 1B were a kind gift of Dr. J. Blasi (University of Barcelona, Barcelona, Spain) (Perez-Branguli et al., 2002). Mammalian plasmids encoding C2AB of SYT1 (aa 96–421) and SNAP-25^R carrying a BoNT/E-resistant mutation [R179W,I180E,E182I] in the C-terminal end of SNAP-25 were described previously (Washbourne et al., 1999; Rickman et al., 2004a). Mutagenesis was performed using the QuikChange system (Stratagene, La Jolla, CA) with mutants identified by DNA sequencing by Cytomyx (Cambridge, UK). Native SYT was purified as a by-product of the SNARE isolation from bovine brains (Rickman and Davletov, 2003). To isolate the brain syntaxin, the native SNARE complex was disrupted by heating for 3 min at 95°C, and the SNARE proteins were subjected to electroelution (Hu et al., 2002a). The eluted syntaxin was refolded by replacing SDS with 0.8% β-octylglucoside on a gel-filtration column. The isolated syntaxin was active in SNARE, synaptotagmin, and complexin binding reactions and also in reconstituted fusion assays (Hu et al., 2002b; Rickman et al., 2004a, 2005). The recombinant syntaxin1A (1-261) fails to substitute for the brain-purified syntaxin in the constitutive SYT binding even when subjected to the same electroelution procedure.

Protein Binding Assays and Circular Dichroism (CD) Spectroscopy

All binding reactions were performed for 30 min in the absence of calcium (2 mM EDTA) at 22°C. GST-tagged proteins (3 μg each) were immobilized on glutathione-Sepharose beads and incubated in 100 μl of buffer A (100 mM NaCl, 0.1% Triton X-100, 2 mM EDTA, and 20 mM HEPES, pH 7.2) in the presence of 3 μg of partner proteins. Beads were washed with buffer A, and bound protein was analyzed by SDS-PAGE followed by Coomassie staining (50% of reaction) or Western immunoblotting (1% of reaction). For immunoblotting, protein bands were detected using syntaxin (clone HPC-1; Sigma-Aldrich, St. Louis, MO) and SNAP-25 (clone 71.2; Synaptic Systems, Goettingen, Germany) antibodies and chemiluminescence (West Dura; Pierce Chemical, Rockford, IL). The chemiluminescent signal was imaged using a ChemiDoc XRS (Bio-Rad, Hercules, CA) and quantified using Quantity One software (Bio-Rad). Real-time binding of C2B, without the GST tag, to the SNAP-25/syntaxin complex was performed using a Biacore 2000 system (Biacore, Uppsala, Sweden). A single channel of CM5 chip (Biacore) was used to immobilize GST-SNAP-25 protein that was then loaded with brain-derived syntaxin before C2B binding. Syntaxin and C2B binding to immobilized GST-SNAP-25 was highly reproducible between cycles (SD <2%). C2B (2 μM) was applied to the chip at a rate of 2 μl/min in buffer A. In circular dichroism experiments, far UV-CD spectra were obtained by averaging five scans with a step size of 0.2 nm on a Jasco J-810 CD spectrometer (Jasco, Tokyo, Japan) at 25°C. Measurements were performed in a quartz cuvette (0.2 cm; Hellma, Plainview, NY) using normalized protein amounts. The CD spectra measurements of the t-SNARE assembly were performed at equilibrium. In thermal denaturation experiments, the ellipticity was measured at 222 nm. Formation of the SDS-resistant SNARE complex was analyzed in 30-min reactions at 22°C as described previously (Hu et al., 2002a).

Computational Protein Docking

A set of in silico complexes was generated, on the basis of shape complementarity, by a systematic search for possible, rigid body, docking orientations of the atomic coordinates of the SNARE ternary complex (Protein Data Bank [PDB] code: 1sfc) with the C2A or C2B domains of SYT1 (PDB codes: 1byn and 1k5w, respectively) using the fast Fourier transform-based FTDOCK program (Gabb et al., 1997). Further refinement was performed using a MULTIDOCK program to minimize atom clashes and to improve bonding of the side chains of each interacting component (Jackson et al., 1998). In the resulting complexes, the electrostatic and shape complementarities of the individual molecules upon complexation were specifically assessed as 1) the difference of electrostatic potentials (E_elect) between the bound-unbound states using a finite difference Poisson–Boltzmann solver (Honig and Nicholls, 1995; Grant et al., 2001); and 2) the amount of buried area (BA) upon complexation calculated using the NACCES program (Hubbard et al., 1991). For the electrostatic calculations, the relative dielectric constants for protein and solvent were taken to be 2.0 and 78.54, respectively, considering an electrolyte concentration of 100 mM. The energetic contributions were combined into a total binding energy (E_total) defined as E_total = E_elect + γ× BA, where γ is equal to –0.005 kcal × mol^–1 × Å^–2. Favorable complexes, i.e., those with total binding energies less than –10 kcal/mol (Erickson, 1989), were subsequently filtered using the mutagenesis data. The presented model was chosen based on its very low energy (–27 kcal/mol) and the interfacial bonding of the C2B K326,K327 and SNAP-25 D51,E52,E55 residues. Small rotations and translations of the individual interacting components are allowed and will be valid within the accuracy of our calculations.

Analysis of Protein Mutants in Release Assays

The effect of the SNAP-25 [D51K,E52K,E55K] mutation on the release of growth hormone (GH) was investigated in PC12 cells. LipofectAMINE 2000 (Invitrogen, Carlsbad, CA) was used to transiently transfect cells with plasmids expressing human growth hormone (pXGH5) and the botulinum neurotoxin E light chain (BoNT/E) in the presence of either control (pcDNA3) or BoNT/E-resistant SNAP-25^R-encoding vectors (Graham et al., 2002). Seventy-two hours later, transfected cells were challenged with 0 or 10 μM calcium in the presence of 20 μM digitonin-containing Krebs-Ringer buffer (145 mM NaCl, 5 mM KCl, 1.3 mM MgCl₂, 1.2 mM NaH₂PO₄, 10 mM glucose, and 20 mM HEPES, pH 7.4). GH release over a 15-min period was assayed using a GH assay kit (Roche Diagnostics, Lewes, United Kingdom).

The ability of rat SYT1 C2AB, mutated or not, to compete with the native vesicular form was probed in bovine chromaffin cells by amperometry as described previously (Rickman et al., 2004a). Briefly, 3 d after electroporation, cells transfected with C2AB were identified through coexpression of enhanced green fluorescent protein. A carbon fiber electrode was positioned in contact with fluorescent, or control, cells, and exocytosis was triggered by pressure ejection of a buffer containing 20 μM digitonin and 10 μM calcium. Amperometric responses were collected at 4 kHz and digitized with a Digidata 1322A acquisition system (Molecular Devices, Sunnyvale, CA). Similar levels of expression of control and mutant proteins in mammalian cells were confirmed by immunoblotting using either PC12 cells (SNAP-25^R) or HeLa cells (C2AB).

Complementary Binding of Two Vesicular Proteins to t-SNAREs

We have recently identified the residues on the C2B domain of SYT1 involved in the constitutive association with syntaxin and SNAP-25, the latter two being purified from a native source, the bovine brain (Rickman et al., 2004a). To experimentally map the binding interface on the SNAREs, recombinant proteins amenable to mutagenesis were required. We found that bacterially expressed SNAP-25 can be used for the constitutive SYT binding. As shown in Figure 1A, the cytoplasmic part of SYT (C2AB) constitutively binds brain-derived syntaxin1 and the recombinant SNAP-25B when they are together in a binary complex, being unable to bind to individual proteins. Both forms of SNAP-25, A and B, were equally effective, whereas neither recombinant syntaxin1A nor syntaxin1B, both lacking their transmembrane regions, could efficiently support calcium-independent C2AB binding (our unpublished data). The deficiency of bacterially expressed syntaxin1 may be because of either the lack of the transmembrane region or the absence of yet uncharacterized posttranslational modifications. Therefore, we used the brain-purified syntaxin in all subsequent reactions and focused on SNAP-25 mutagenesis.

Before analyzing SNAP-25 residues, we investigated the relationship between SYT and synaptobrevin binding to the two t-SNAREs. Because SYT can bind the t-SNARE assembly in the absence of calcium, C2AB may act as a molecular clamp, physically interfering with synaptobrevin engagement of t-SNAREs. We tested whether C2AB could block formation of the ternary SNARE complex in defined reactions. Figure 1B shows that preincubation of syntaxin and SNAP-25 with C2AB had no effect on the ability of the two t-SNAREs to form, upon the addition of synaptobrevin, the tight SDS-resistant ternary SNARE complex implicated in membrane fusion (Hayashi et al., 1994; Sutton et al., 1998). Next, the two t-SNAREs were first prebound to the immobilized GST-C2AB, and then the triple protein assembly— C2AB/syntaxin/SNAP-25—was incubated with synaptobrevin. Synaptobrevin bound the C2AB/t-SNARE triple assembly with a stoichiometry similar to that found in the native SDS-resistant SNARE complex purified from bovine brain (Figure 1C). Therefore, SYT by itself does not prevent the soluble synaptobrevin engagement with the two t-SNAREs providing an explanation for copurification of all four proteins from the brain detergent extracts (Sollner et al., 1993). Because the SYT/t-SNARE interaction is not diminished after synaptobrevin binding (Figure 1C), the two vesicular proteins do not compete with each other and may have independent binding sites. Indeed, whereas the SYT interaction with t-SNAREs is primarily electrostatic (Rickman et al., 2004a), synaptobrevin binds to a hydrophobic groove formed by syntaxin and SNAP-25 (Sutton et al., 1998).

The SYT Binding Epitope on SNAP-25

The constitutive interaction of SYT with the preassembled t-SNAREs is mediated by a conserved polybasic motif on the C2B domain and principally involves a pair of lysines, 326 and 327 (Rickman et al., 2004a). The ability of the recombinant SNAP-25 to replicate this function allowed us to map the binding site on SNAP-25 through site-directed mutagenesis. First, we analyzed the two helices of SNAP-25 for the presence of conserved acidic amino acids in evolutionarily distant organisms. Such an analysis is now used, in the postgenomic era, to uncover potential areas for protein interactions (Lichtarge and Sowa, 2002). In the two helices, there are four conserved double acidic residue sites, labeled I to IV and highlighted on the ribbon representation of the helical crystal structure of the ternary SNARE complex (Figure 1, D and E). To test for involvement of these four sites in SYT binding, the relevant aspartate and glutamate residues were mutated to lysines. After protein normalization, SNAP-25 mutants were incubated, in the presence of syntaxin, with immobilized GST-C2B. For initial screens, bound material was analyzed by Western immunoblotting. Only mutation of site III [D51K,E52K] resulted in a reduction in bound t-SNAREs (Figure 2). Notably, this mutation reduced binding of not only SNAP-25 but also syntaxin, supporting the fact that t-SNAREs bind to SYT as a singular molecular entity.

Closer inspection of site III reveals two additional glutamate residues next to D51 and E52 (Figure 3A). This site is at the ionic layer (layer 0) observed in the crystal structure of the ternary SNARE complex (Sutton et al., 1998). To analyze whether these additional residues, E55 and E58, also play a role in the SYT binding, triple mutants were generated. Quantification of bound t-SNAREs showed that the further mutation of [E55K] but not [E58K] had an additive effect on the SNAP-25[D51K,E52K] mutant (Figure 3B), highlighting the specificity of the SYT interaction. Finally, we tested the ability of SYT to interact with either the binary or ternary SNARE complexes containing SNAP-25[D51K,E52K,E55K]. The triple mutation severely disrupted C2B binding of both t-SNAREs and of the ternary SNARE complex as judged by Coomassie staining (Figure 3C). Figure 3D illustrates the effect of the [D51K,E52K,E55K] mutation on the electrostatic potential of SNAP-25 in the context of the SNARE complex.

The SYT Binding Epitope on SNAP-25 Is Essential for Regulated Exocytosis

Before testing the impact of the SNAP-25 [D51K,E52K,E55K] mutation on regulated exocytosis, it was important to determine whether these changes were specifically disrupting the SYT/t-SNARE link or were simply compromising the ability of SNAP-25 to associate with syntaxin. In a direct binding reaction, SNAP-25[D51K,E52K,E55K] readily bound to GST-syntaxin in a similar manner to the nonmutated molecule (Figure 4A). Next, CD spectroscopy demonstrated that syntaxin and SNAP-25, when mixed, exhibit a significant increase in α-helicity compared with the calculated spectra for noninteracting molecules, in agreement with a previous study (Fasshauer et al., 1997). Comparison of CD spectra for syntaxin with either the nonmutated or SNAP-25[D51K,E52K,E55K] demonstrated a similar increase in α-helical spectra (Figure 4B), indicating that the two forms of SNAP-25 generate comparable helical structures. To assess the relative stability of the t-SNARE assembly, thermal denaturation was used with α-helical content monitored by CD (Figure 4C). There was no significant difference between the t-SNARE assemblies containing SNAP-25 with or without the triple mutation. Furthermore, we tested the ability of the t-SNAREs to bind synaptobrevin. Syntaxin, preincubated with either the wild-type SNAP-25 or SNAP-25[D51K,E52K,E55K], was added to synaptobrevin immobilized on Sepharose beads (Figure 5A). The triple mutation had no detectable effect on the synaptobrevin/t-SNARE interaction. Complexin, a soluble protein that specifically interacts with the fully folded ternary SNARE complex (Chen et al., 2002), exhibited comparable binding of the SNARE complexes made with SNAP-25, carrying the triple mutation or not (Figure 5B). Although this experiment provided strong evidence that folding of the SNARE helices is not affected by the mutation, it was essential to test the ability of the mutated SNAP-25 to form the SDS-resistant SNARE complex implicated in the fusion process (Sutton et al., 1998). Titration of SNAP-25 demonstrated that the mutated SNAP-25 was fully capable of entering into the “tight” SNARE complex (Figure 5C). We conclude therefore that the triple [D51K,E52K,E55K] mutation of SNAP-25 disrupts specifically the constitutive binding of SYT with the assembled t-SNAREs but not interactions between the SNAREs themselves.

To investigate the impact of the triple SNAP-25 mutation on regulated exocytosis, we sought to express the mutant SNAP-25 in the absence of endogenous SNAP-25. This was accomplished by 1) using botulinum neurotoxin E light chain (BoNT/E LC) that cleaves endogenous SNAP-25, leading to its removal from the plasma membrane syntaxin (Bajohrs et al., 2004); and 2) expressing a BoNT/E-resistant form of SNAP-25, SNAP-25^R, to restore exocytosis (Graham et al., 2002). Mutagenesis of this SNAP-25^R molecule thereby permits to assess the contribution of specific SNAP-25 residues in regulated exocytosis. We used neuroendocrine PC12 cells that serve as a popular model for studies of regulated exocytosis (Chen et al., 1999). Transfection of PC12 cells with the plasmid expressing BoNT/E LC alone resulted in an efficient blockade of exocytosis, as measured by the growth hormone release assay, whereas its cotransfection with SNAP-25^R resulted in restoration of exocytosis (Figure 5D). The triple mutation [D51K,E52K,E55K] was then introduced in SNAP-25^R. The extent of exocytosis upon expression of SNAP-25^R[D51K,E52K,E55K] was greatly reduced, despite the same level of protein expression as for SNAP-25^R (Figure 5D). Evidently, the SYT-binding epitope on SNAP-25 helix1 does play a critical role in regulated exocytosis. Because this epitope does not disrupt the interaction between the SNAREs themselves, it is reasonable to conclude that it affects the SYT/SNARE link, which is vital for efficient exocytosis (Sudhof and Scheller, 2001; Koh and Bellen, 2003).

In Silico Modeling of the SYT/SNARE Complex

The requirement for the full-length brain-purified syntaxin, mentioned above, to form the stoichiometric SYT/SNARE supramolecular assembly precludes crystallization trials. However, the available crystal structures of the SYT1 C2 domains and of the ternary SNARE complex offered an alternative way to gain further insight into the constitutive SYT/SNARE interaction, by using an in silico docking approach. This approach has recently been designed to derive protein complexes from known atomic coordinates of the individual components with no prior knowledge of their interacting surfaces (Smith and Sternberg, 2002; van Dijk et al., 2005). We used the FTDOCK program, which was developed for docking native structures and was shown to successfully predict protein complexes in test conditions (Gabb et al., 1997). An initial set of ∼27,000 binding geometries was generated by systematically docking different orientations of the atomic coordinates of the SNARE complex (Sutton et al., 1998) to individual C2 domains (Sutton et al., 1995; Fernandez et al., 2001). Total binding energies for each possible complex were then derived from their buried area and the electrostatic complementarity. Considering that ∼10 kcal/mol is required to overcome the entropy loss upon complexation (Erickson, 1989), the number of favorable complexes between the C2B domain and the SNARE complex was still relatively large, 296 (Figure 6A). By contrast, none of the C2A-involving complexes could fulfill such a criterion consistent with the inability of C2A to bind t-SNAREs in the absence of calcium (Rickman et al., 2004a).

The systematic docking of C2B to the SNARE complex surface residues, with the set electrostatic and buried area constraints, indicated amino acids that are likely to participate in energetically favorable complexes (Figure 6B). In the SNAP-25 helices, the residues centered on E52 exhibited the highest calculated frequency to take part in favorable complexes, thereby lending a structure-based support for our experimental data. The closely located syntaxin residues E224 and E228 also showed a strong tendency for interaction with the C2B domain; however, as evident from our experimental data (Figures 3C and 5D), the triple mutation of SNAP-25 is sufficient to disrupt the SYT/SNARE association in vitro and regulated exocytosis in vivo. Rotation of the SNARE complex by 180° highlights the overall selectivity of the in silico docking approach (Figures 6B). The face of the SNARE complex with the lowest frequency for the C2B interaction (Figure 6B, blue color) corresponds to the synaptobrevin helix (Sutton et al., 1998).

Taking into account the limitations of computational docking (van Dijk et al., 2005), we used our mutagenesis data of SNAP-25 (Figures 2 and 3) and of C2B (Rickman et al., 2004a) to determine a complex with interfacial bonding for all experimentally identified residues. The selected C2B/t-SNARE complex, shown in Figure 6C, represents the highest ranked prediction based on 1) the lowest total binding energy, –27 kcal/mol; and 2) participation of the C2B lysines 326 and 327 and the SNAP-25 three acidic residues D51, E52, E55 in the interaction. It presents both a very favorable electrostatic energy (–17 kcal/mol) and fulfills the essential requirement for shape complementarity (Lo Conte et al., 1999). In this near perpendicular arrangement of the two structures, the concave face of C2B partially envelops the SNARE cylinder, resulting in a buried area of almost 2000 Ų. In this configuration, the bound C2B domain is on the opposite face to the synaptobrevin-binding groove, consistent with the complementary interaction of the two vesicular proteins with syntaxin and SNAP-25 (Figure 1).

Examination of the AD3 Mutation in the SYT C2B Domain

Analysis of the literature for mutations residing at the SYT/SNARE binding interface revealed that a well characterized mutation, named AD3 in the original Drosophila study (DiAntonio et al., 1993), is in proximity to the interacting residues (Figure 6C, right). This mutant carries only a single amino acid substitution of tyrosine to asparagine but exhibits a pronounced decrease in evoked release (DiAntonio et al., 1993; Yoshihara and Littleton, 2002). Intriguingly, this mutation, corresponding to Y311 in the rat protein, affects neither the calcium-triggered phospholipid binding nor the reported calcium-dependent interactions with recombinant SNAREs (Littleton et al., 2001). We hypothesized that the Y311N mutation may impact upon the constitutive SYT/SNARE interaction providing an immediate test for the relevance of our model. Comparative CD spectroscopy of C2B demonstrated that the overall β-sheet content and its stability are not affected by this single mutation (Figure 7A). We then immobilized GST-SNAP-25 with brain-derived syntaxin on a surface plasmon resonance chip and tested constitutive binding of the C2B domain, mutated or not. The Y311N mutation strongly decreased the ability of the untagged C2B domain to associate with the t-SNAREs (Figure 7B). It was essential to determine whether the Y311N mutation, even in the context of the tandem C2A-C2B structure, would be sufficient to disrupt the constitutive SYT/SNARE interaction. Normalized amounts of the wild-type and mutated GST-C2AB domain were immobilized on Sepharose beads and incubated with the t-SNAREs. The Y311N mutation severely disrupted C2AB binding to the t-SNARE assembly in the quantitative pull-down assay (Figure 8A). In a control experiment, mutation of a neighboring tyrosine (Y364, highlighted in Figure 6C) to asparagine had no impact on the t-SNARE binding, indicating the specific importance of Y311. Mutation of tyrosine 311 to a different hydrophobic amino acid—phenylalanine [Y311F]— did not affect the constitutive t-SNARE binding (Figure 8A), arguing that the deleterious effect of the AD3 mutation is not due to disrupting a specific amino acid interaction but is likely a result of a local structural change in the C2B domain. Indeed, this amino acid is packed between neighboring side chains and is largely buried, indicating a structural role (Fernandez et al., 2001). The direct staining of pull-down reactions with Coomassie stain confirms the disruptive effect of the AD3 mutation on the constitutive SYT/t-SNARE interaction (Figure 8B).

To test the impact of the Y311N mutation in a mammalian system, we used mature bovine chromaffin cells, which strictly rely on the SYT1 isoform in regulated exocytosis (Burgoyne and Morgan, 2003). Expression of the soluble C2AB fragment in the cytoplasm is known to impede catecholamine release because this soluble C2AB competes with the native vesicular SYT for its molecular target. Mutation of this inhibitory C2AB allows analysis of the residues that contribute to this competition. By inference, this provides insight into functionally important amino acids in the native vesicular form of SYT as has previously been shown for the C2B lysines 326 and 327 (Rickman et al., 2004a). Amperometric recordings demonstrated that introduction of the Y311N mutation in the soluble C2AB is sufficient to relieve the blockade of exocytosis (Figure 8C). Therefore, the soluble C2AB carrying an equivalent of the Drosophila AD3 mutation cannot compete with the vesicular SYT function in exocytosis. This highlights the functional importance of the Y311 tyrosine residue at the concave face of the C2B domain, further indicating an essential role of the constitutive SYT/SNARE link in regulated exocytosis.