The C terminus of SNAP25 is essential for Ca(2+)-dependent binding of synaptotagmin to SNARE complexes.

The plasma membrane soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) proteins syntaxin and synaptosome-associated protein of 25 kDa (SNAP25) and the vesicle SNARE protein vesicle-associated membrane protein (VAMP) are essential for a late Ca(2+)-dependent step in regulated exocytosis, but their precise roles and regulation by Ca(2+) are poorly understood. Botulinum neurotoxin (BoNT) E, a protease that cleaves SNAP25 at Arg(180)-Ile(181), completely inhibits this late step in PC12 cell membranes, whereas BoNT A, which cleaves SNAP25 at Gln(197)-Arg(198), is only partially inhibitory. The difference in toxin effectiveness was found to result from a reversal of BoNT A but not BoNT E inhibition by elevated Ca(2+) concentrations. BoNT A treatment essentially increased the Ca(2+) concentration required to activate exocytosis, which suggested a role for the C terminus of SNAP25 in the Ca(2+) regulation of exocytosis. Synaptotagmin, a proposed Ca(2+) sensor for exocytosis, was found to bind SNAP25 in a Ca(2+)-stimulated manner. Ca(2+)-dependent binding was abolished by BoNT E treatment, whereas BoNT A treatment increased the Ca(2+) concentration required for binding. The C terminus of SNAP25 was also essential for Ca(2+)-dependent synaptotagmin binding to SNAP25. syntaxin and SNAP25.syntaxin.VAMP SNARE complexes. These results clarify classical observations on the Ca(2+) reversal of BoNT A inhibition of neurosecretion, and they suggest that an essential role for the C terminus of SNAP25 in regulated exocytosis is to mediate Ca(2+)-dependent interactions between synaptotagmin and SNARE protein complexes.

Regulated neurotransmitter secretion is a specialized version of a general membrane fusion mechanism in which exocytotic fusion is strictly Ca 2ϩ -regulated. Studies of this process have yielded insights into universal mechanisms for intracellular membrane fusion and the identity of core components of the fusion machinery (1)(2)(3). VAMP, 1 syntaxin, and SNAP25 are the neural protein substrates for clostridial neurotoxins, a family of highly selective proteases that potently inhibits neurose-cretion (4,5). These proteins are soluble N-ethylmaleimidesensitive factor attachment protein receptors (SNAREs) that mediate the membrane association of N-ethylmaleimide-sensitive factor, a protein required for membrane fusion (6). The SNARE proteins are capable of assembling into extremely stable heterotrimeric complexes (7)(8)(9)) that consist of a four-helix bundle in parallel alignment (10 -13). A current hypothesis suggests that the formation of SNARE complexes in trans across apposing membranes promotes intimate bilayer interactions and provides the energy to drive membrane fusion (10,(13)(14)(15)(16). Studies with donor and acceptor proteoliposomes containing VAMP or syntaxin/SNAP25 indicate that trans-SNARE complexes can mediate bilayer phospholipid mixing and possibly fusion (16).
In contrast to the slow fusion mediated by reconstituted SNAREs in vitro (16), neurosecretion is rapid and Ca 2ϩ -dependent suggesting that Ca 2ϩ may regulate SNARE complex formation (1,17). Genetic studies have established an important role for the Ca 2ϩ -binding vesicle protein synaptotagmin I in rapid neurosecretion (18 -21). Synaptotagmin I exhibits Ca 2ϩdependent interactions with acidic phospholipids as well as with syntaxin (22)(23)(24)(25)(26)(27) leading to its suggested role as a Ca 2ϩ sensor for exocytosis (1,18,22,23). The precise role of synaptotagmin and its Ca 2ϩ -dependent interactions with phospholipids or syntaxin in regulated neurosecretion remains to be determined.
Studies of regulated exocytosis in membrane preparations from neuroendocrine cells demonstrated that SNAREs are required for a late Ca 2ϩ -dependent step that occurs after vesicle docking and ATP-dependent priming and immediately before fusion (28,29). Tetanus toxin and botulinum neurotoxin (BoNT) B, C1, and E completely inhibited the Ca 2ϩ -dependent triggering of exocytosis, which implied that VAMP, syntaxin, and SNAP25 participate in steps close to or at fusion. Paradoxically BoNT A, which like BoNT E cleaves SNAP25, was only partially inhibitory for triggered fusion despite efficient proteolysis of SNAP25 (28). Because these toxins cleave SNAP25 at distinct C-terminal sites (Arg 180 -Ile 181 for BoNT E and Gln 197 -Arg 198 for BoNT A; see Refs. 30 and 31), it was inferred that the domain within the C terminus of SNAP25 between the toxin cleavage sites plays a distinct role in Ca 2ϩ -dependent membrane fusion events (28). In the present study, this domain of SNAP25 is revealed to be essential for Ca 2ϩ -dependent interactions with synaptotagmin. This finding clarifies classical observations that BoNT A inhibition of neurosecretion is partially reversed by elevating neuronal Ca 2ϩ levels (32). Moreover, the results suggest that an important role for the C terminus of SNAP25 in regulated exocytosis is to mediate Ca 2ϩ -dependent interactions between synaptotagmin and SNARE protein complexes.

EXPERIMENTAL PROCEDURES
Assays for Ca 2ϩ -activated Exocytosis-PC12 cells grown in Dulbecco's modified culture medium supplemented with 5% horse serum and 5% iron-supplemented calf serum were incubated with 0.5 Ci/ml [ 3 H]norepinephrine (Amersham Pharmacia Biotech) plus 0.5 mM sodium ascorbate for 16 h at 37 o C. Cells were washed as described previously (28,33) and homogenized by multiple passes through a stainless steel ball homogenizer for preparation of a plasma membrane fraction that contains docked dense-core vesicles (29). Secretion assays were conducted in 5-min incubations at 30 o C in 0.05 M HEPES, pH 7.2, 0.12 M potassium glutamate, 0.02 M potassium acetate, 0.002 M EGTA, 0.1% bovine serum albumin, 0.002 M MgATP, and free Ca 2ϩ adjusted to indicated values. [ 3 H]Norepinephrine was detected in supernatants following centrifugation of reaction mixtures. Results are plotted as percent norepinephrine release by normalization to the total content of [ 3 H]norepinephrine per incubation. BoNT A, B, and E holotoxins (generously provided by B. R. DasGupta) and BoNT C1 holotoxin (generously provided by S. Kozaki) were activated by preincubation with 5 mM dithiothreitol. Toxin treatment of membrane preparations (at 1-100 nM BoNT) was conducted for 3 min at 30 o C prior to adding Ca 2ϩ for secretion incubations. Mouse SNAP23 is susceptible to proteolysis by BoNT A and E but only at concentrations or in incubations that exceed those used here to cleave SNAP25 by at least 1000-fold (34,35). It is unlikely that the rat SNAP23 protein is the functional target of BoNT inhibition in PC12 cell membranes.
Protein Binding Assays-Glutathione S-transferase (GST) fusion proteins encoded by pGEX vectors were produced in Escherichia coli by standard methods. Vectors encoding synaptotagmin I C2AB, C2A, C2B, SNAP25B, and VAMP were kindly provided by R. H. Scheller (36,37). The vector encoding synaptotagmin III was kindly provided by M. Fukuda and K. Mikoshiba (38). The vector encoding syntaxin 1A was kindly provided by E. Chapman (24). GST fusion proteins were purified by glutathione-agarose chromatography (Amersham Pharmacia Biotech) using either glutathione for elution or thrombin cleavage for removing the GST. GST-SNAP25 was cleaved by BoNTs by incubation with ϳ1 M toxin for 1 h at room temperature. Standard conditions for binding studies employed 200-l reactions containing 0.02 M HEPES, pH 7.2, 0.15 M KCl, 1 mM CaCl 2 or 2 mM EGTA, 0.5% Triton X-100, and 2.5% bovine serum albumin or 1% cold fish skin gelatin (Sigma). Binary and ternary SNARE complexes were formed by overnight incubation at 4 o C of SNAP25, syntaxin, or VAMP each at 5 M. Glutathione-agarose beads containing 5-10 g of GST-synaptotagmin (C2AB) were incubated with 1 M SNAP25, binary complex or ternary complex. Glutathione-agarose beads containing 5-10 g of GST-SNAP25 were incubated with 0.5 M synaptotagmin (C2AB). Incubations were for 1 h at 4 o C following which beads were recovered by centrifugation and washed twice in 400 l of assay buffer prior to solubilization in sample buffer for SDS-polyacrylamide gel electrophoresis. Detection of SDSresistant heterotrimeric SNARE complexes was by described methods (8). Electrophoresed samples were transferred to nitrocellulose for immunoblotting with the following antibodies: a polyclonal rabbit antibody generated with the C2AB regions of synaptotagmin I or III, a polyclonal rabbit antibody generated with a 15-residue N-terminal VAMP peptide sequence conjugated to bovine serum albumin, a polyclonal rabbit antibody generated to a C-terminal SNAP25 peptide (residues 195-206, generously provided by M. Wilson or obtained commercially from StressGen Biotechnologies Corp.), a rabbit polyclonal antibody generated to GST-SNAP25 (Alomone Labs), a mouse monoclonal antibody to syntaxin 1 (HPC-1, Sigma), or a mouse monoclonal antibody to VAMP (Cl 69.1, generously provided by R. Jahn). 125 I-Protein A or goat anti-mouse IgG (NEN Life Science Products) were used as secondary reagents. Autoradiograms were quantitated with a Molecular Dynamics SI Densitometer using ImageQuaNT software.
Ca 2ϩ -dependent binding of native synaptotagmin I to SNAP25 was assessed in binding studies with rat brain detergent extracts incubated Inhibition observed with these toxins was reversed to only a small extent at higher Ca 2ϩ concentrations. Each data point shown represents the mean of duplicate determinations that differed by less than 3%.
with GST-SNAP25 immobilized on glutathione-agarose as described above. Rat brain extracts were prepared by homogenizing rat brains in 0.32 M sucrose using 10 strokes of a motor-driven Teflon glass homogenizer. Following clarification at 5,000 rpm for 2 min in an SS34 rotor, crude synaptosomes were collected by centrifugation at 11,000 rpm for 12 min and solubilized in 1% Triton X-100 in 0.05 M HEPES, pH 7.2, 0.1 M NaCl with protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 2 g/ml pepstatin, and 20 g/ml aprotinin). Detergent extracts were adjusted to 1 mM EGTA or 1 mM CaCl 2 for binding or immunoprecipitation studies. SNAP25 immunoprecipitations were conducted by overnight incubation of detergent extracts at 4 o C with SNAP25 monoclonal antibodies (Sternberger Monoclonals Inc.) or mouse immunoglobulins immobilized on protein A-Sepharose.

Ca 2ϩ Reversal of Toxin Inhibition Is Restricted to BoNT
A-Studies of the late Ca 2ϩ -dependent step of exocytosis in PC12 cells showed that BoNT E completely inhibited Ca 2ϩtriggered fusion, whereas BoNT A was only partially, if at all, inhibitory despite effective cleavage of SNAP25 (28,33). An explanation for this difference between toxins was suggested by earlier studies demonstrating that BoNT A inhibition of neurosecretion was unique in being reversed by elevating Ca 2ϩ levels in synapses (32). We therefore examined the effects of Ca 2ϩ on the BoNT inhibition of regulated norepinephrine secretion in a cell-free assay that utilizes purified PC12 cell plasma membranes containing docked dense-core vesicles (29).
BoNT E treatment strongly inhibited Ca 2ϩ -dependent norepinephrine release at all Ca 2ϩ concentrations tested (Fig. 1, upper panel). In contrast, BoNT A treatment inhibited Ca 2ϩdependent norepinephrine release at intermediate Ca 2ϩ concentrations, but inhibition was strongly reduced at high Ca 2ϩ concentrations. BoNT E and BoNT A were present at maximally effective concentrations and catalyzed similar extents of SNAP25 proteolysis (ϳ80%, not shown). Treatment with BoNT A essentially increased the apparent EC 50 for Ca 2ϩ in triggering exocytosis (from 2.3 Ϯ 0.3 to 4.4 Ϯ 1.1 M; mean Ϯ S.D. for six experiments). This increase was not simply the result of a partial attenuation of exocytosis at all Ca 2ϩ concentrations since the percent inhibition by BoNT A decreased with increasing Ca 2ϩ concentrations unlike that for other BoNTs (Fig. 1, middle panel).
The Ca 2ϩ reversal of inhibition by toxin was unique for BoNT A and was not observed for BoNT B, which cleaves VAMP (39), or for BoNT C1, which preferentially cleaves syntaxin (40). These toxins strongly inhibited Ca 2ϩ -dependent norepinephrine release at all Ca 2ϩ concentrations ( Fig. 1, lower panel). The unique Ca 2ϩ reversal of BoNT A inhibition of secretion observed in this purified PC12 cell membrane preparation is similar to that reported previously for BoNT inhibition in more complex systems such as the neuromuscular junction or hippocampal and chromaffin cells (32,(41)(42)(43).
Synaptotagmin Interacts with the C Terminus of SNAP25 in a Ca 2ϩ -dependent Manner-Because cleavage of SNAP25 by BoNT A increased the EC 50 for Ca 2ϩ activation of exocytosis, we considered the possibility that SNAP25 was involved in the Ca 2ϩ regulation of exocytosis. The Ca 2ϩ -binding vesicle protein synaptotagmin has been proposed to function as a Ca 2ϩ sensor for exocytosis (1,18,22), so we determined whether synaptotagmin interacts directly with SNAP25. In the absence of Ca 2ϩ , recombinant SNAP25 was found to bind an immobilized recombinant synaptotagmin I cytoplasmic domain ( Fig. 2A, upper panel) as was previously reported (44). However, the inclusion of Ca 2ϩ in the binding reaction markedly enhanced SNAP25 binding to synaptotagmin I ( Fig. 2A, upper panel). In the reverse orientation, binding of the synaptotagmin I cytoplasmic domain to immobilized SNAP25 was also observed, and the binding was strongly stimulated by inclusion of Ca 2ϩ ( Fig

. Distinct Ca 2؉ -independent and Ca 2؉ -dependent interactions between synaptotagmin I (Syt I) and SNAP25. A, Upper,
binding of SNAP25 to immobilized synaptotagmin I. Glutathione (GSSG)-agarose beads containing 5 g of GST-cytoplasmic domain fusion protein of synaptotagmin I or without bound protein were incubated with 5 g of SNAP25 either in the absence or presence of 1 mM Ca 2ϩ as indicated. Beads were washed and solubilized in sample buffer for analysis of the bound fraction by Western blotting with a SNAP25 antibody. Lower, binding of synaptotagmin I cytoplasmic domain to immobilized SNAP25. GSSG-agarose beads containing 5 g of GST-SNAP25 or without bound protein were incubated with 5 g of synaptotagmin I cytoplasmic domain in the absence or presence of 1 mM Ca 2ϩ . Beads were processed for Western blotting with synaptotagmin I antibody. B, summary of synaptotagmin I-SNAP25 binding studies. Binding studies similar to those of A and B were analyzed by Western blotting and quantified by densitometry. Inclusion of 1 mM Ca 2ϩ in the incubations stimulated binding 2.5-4-fold above that observed with 2 mM EGTA. C, synaptotagmin I binding to SNAP25 exhibits saturation for Ca 2ϩ -independent and Ca 2ϩ -dependent interactions. Binding of the indicated concentrations of synaptotagmin I cytoplasmic domain to 5 g of GST-SNAP25 immobilized on GSSG-agarose beads was determined. In four similar experiments, the stoichiometry of binding at saturation (ϩCa 2ϩ ) was 0.3-0.8 mol of synaptotagmin I bound per mol of SNAP25. both formats showed that Ca 2ϩ stimulated the interaction of SNAP25 with synaptotagmin I by 2.5-5-fold (Fig. 2B). Synaptotagmin I binding to SNAP25 in the absence of Ca 2ϩ was of relatively low affinity (K D ϳ1.2 M), but Ca 2ϩ increased the affinity of the interaction about 6-fold to a K D ϳ0.2 M (Fig.  2C). The stoichiometry of the binding at saturation in the absence or presence of Ca 2ϩ approached 1:1.
High Ca 2ϩ concentrations were required to stimulate synaptotagmin I binding to SNAP25 with half-maximal stimulation observed at ϳ200 M Ca 2ϩ (Fig. 3A). When tested at 1 mM, Sr 2ϩ , Ba 2ϩ , and Mg 2ϩ did not stimulate the interaction (not shown). Synaptotagmin I interactions with syntaxin were previously reported to exhibit a very similar Ca 2ϩ dependence and cation specificity (24 -27). Synaptotagmin contains two C2 domains that mediate Ca 2ϩ -dependent activities of the protein (45). To define the region of synaptotagmin I required for the Ca 2ϩ -stimulated interaction with SNAP25, recombinant proteins representing the full cytoplasmic domain (C2AB), a membrane-proximal region containing the C2A domain (C2A), or a membrane-distal region containing the C2B domain (C2B) were used as immobilized ligands. The C2B protein exhibited virtually no SNAP25 binding in the presence or absence of Ca 2ϩ (Fig. 3B). In contrast, the C2A fusion protein exhibited Ca 2ϩ -stimulated interactions with SNAP25 although it did not possess the full activity of the complete cytoplasmic domain (C2AB). These results for SNAP25 were similar to those for synaptotagmin I interactions with syntaxin where the C2A domain exhibited partial Ca 2ϩ -dependent binding activity (24 -27). The syntaxin binding domain of synaptotagmin I has recently been shown to consist of the C2A domain with a linker region plus a portion of the C2B domain (62).
To define the region of SNAP25 essential for synaptotagmin I binding, immobilized SNAP25 was quantitatively cleaved with BoNT A or BoNT E. SNAP25 truncated by cleavage with either toxin retained its capacity to interact with synaptotagmin I in the absence of Ca 2ϩ (Fig. 3C, closed bars) as previously reported (44). In contrast, the Ca 2ϩ -dependent interaction of SNAP25 with synaptotagmin I was completely abolished by treatment with BoNT E and partially inhibited by treatment with BoNT A (Fig. 3C, open bars). The extent of inhibition of synaptotagmin I binding observed with BoNT A-treated SNAP25 was dependent upon the Ca 2ϩ concentration in the binding reaction with the inhibition strongly reduced at high Ca 2ϩ concentrations (Fig. 3A). BoNT A treatment essentially increased the EC 50 for Ca 2ϩ stimulation of synaptotagmin I binding to SNAP25. interaction. Binding studies were conducted at the indicated Ca 2ϩ concentrations with 5 g of synaptotagmin I cytoplasmic domain and 10 g of GST-SNAP25 (control) or BoNT A-treated GST-SNAP25 (BoNT A) immobilized on GSSG-agarose beads. SNAP25 was quantitatively cleaved by BoNT A treatment. Cleavage by BoNT A shifted the Ca 2ϩ dependence of the binding to the right. B, domains of synaptotagmin I required for SNAP25 binding. GST fusion proteins consisting of the complete synaptotagmin I cytoplasmic domain (C2AB), the membraneproximal domain (C2A), or the membrane-distal domain (C2B) were immobilized in equimolar amounts on GSSG-agarose beads and incubated with 1 M SNAP25 in the absence (Ϫ) or presence (ϩ) of 1 mM Ca 2ϩ . SNAP25 binding to the C2B fusion protein was not different from binding to protein-free agarose beads. The C2A fusion protein exhibited significant SNAP25 binding, but it was quantitatively less than that for the C2AB fusion protein. C, effect of cleavage by BoNT A and BoNT E on SNAP25 interactions with synaptotagmin I. Binding reactions were conducted with immobilized intact GST-SNAP25 or with GST-SNAP25 preparations that were cleaved by treatment with BoNT A or BoNT E. Gel electrophoresis and Western blotting confirmed that cleavage by toxins was quantitative. Binding in the absence of Ca 2ϩ (filled histograms) was not affected by BoNT treatment, whereas binding stimulated by Ca 2ϩ (open histograms) was inhibited by BoNT treatment.
Additional studies on native synaptotagmin I and SNAP25 in brain detergent extracts revealed Ca 2ϩ -dependent interactions between these proteins similar to those observed with recombinant proteins (Fig. 4). The specific immunoprecipitation of SNAP25 from rat brain detergent extracts resulted in the co-isolation of synaptotagmin I (Fig. 4A, lane 2), which was markedly enhanced by inclusion of Ca 2ϩ (lane 3 versus 2). The Ca 2ϩ -stimulated retention of synaptotagmin I in the SNAP25 immunoprecipitates was largely eliminated by treatment with BoNT E (lane 4). Neither condition altered the recovery of syntaxin in the SNAP25 immunoprecipitates, and neither condition promoted the retention of CAPS, an abundant cytosolic protein. In a similar manner, synaptotagmin I in brain detergent extracts exhibited Ca 2ϩ -dependent retention on immobilized GST-SNAP25 (Fig. 4B). Ca 2ϩ promoted synaptotagmin I binding over the same concentration range (EC 50 ϳ300 M) as was observed in binary protein binding studies with the recombinant proteins (Fig. 4B, lanes 3-6; compare with Fig. 3A).
Cleavage of the C terminus of SNAP25 with BoNT A or E markedly reduced the Ca 2ϩ -dependent binding of native synaptotagmin I to SNAP25 (Fig. 4B, lane 6 versus 7 and 8).
Neither Ca 2ϩ nor toxin treatment markedly affected the binding of native syntaxin to SNAP25.
Ca 2ϩ Dependence of the Interaction with SNAP25 Varies with the Synaptotagmin Isoform-There were considerable similarities between the effects of Ca 2ϩ in reversing BoNT A inhibition of exocytosis (Fig. 1, upper panel) and in reversing BoNT A inhibition of synaptotagmin I-SNAP25 interactions (Fig. 3A) except that the effective Ca 2ϩ concentrations differed. Several synaptotagmin isoforms that differ in Ca 2ϩ sensitivity have been characterized (26,27,45). Ca 2ϩ -dependent interactions of isoforms such as synaptotagmin I with syntaxin occur at high (ϳ200 M) Ca 2ϩ concentrations, whereas isoforms such as synaptotagmin III exhibit Ca 2ϩ -dependent interactions with syntaxin at lower (ϳ10 M) Ca 2ϩ concentrations (26,27). Synaptotagmin III was found to exhibit Ca 2ϩ -stimulated interactions with SNAP25 at much lower Ca 2ϩ concentrations than those required for synaptotagmin I (Fig. 5A). The stimulation of binding exhibited an EC 50 for Ca 2ϩ of ϳ8 M, which is closer to the EC 50 for Ca 2ϩ -activated dense-core vesicle exocytosis (2.3 M, Fig. 1). The Ca 2ϩ -dependent binding of synaptotagmin III to SNAP25 required C-terminal regions of SNAP25 since cleavage by BoNT A and E inhibited the binding (Fig. 5B). High Ca 2ϩ concentrations were partially effective in restoring synaptotagmin III binding to BoNT A-cleaved SNAP25 and to a lesser extent to BoNT E-cleaved SNAP25 (data not shown).
C-terminal Regions of SNAP25 Are Required for Ca 2ϩ -dependent Synaptotagmin Interactions with SNARE Complexes-SNAP25 and syntaxin form relatively stable heterodimeric complexes consisting of a 2:1 molar ratio of syntaxin to SNAP25 (9). Such binary complexes may be the prevalent form of SNAP25 in cellular membranes, so we examined SNAP25 binding to synaptotagmin I in the presence of syntaxin under con-

FIG. 4. Native SNAP25 and synaptotagmin I exhibit Ca 2؉ -stimulated interactions dependent upon the C terminus of SNAP25.
A, Ca 2ϩ stimulates the interaction of native SNAP25 and synaptotagmin I. Rat brain detergent extracts were incubated with BoNT E for 30 min at 30 o C where indicated, adjusted to either 1 mM EGTA (ϪCa 2ϩ ) or 1 mM CaCl 2 (ϩCa 2ϩ ) with pH adjustment to 7.2, and incubated for 16 h at 4 o C with immobilized control mouse IgGs or SNAP25 immune IgGs. Washed immunoprecipitates were analyzed by immunoblotting for CAPS, synaptotagmin I, syntaxin, and SNAP25. Rat brain detergent extract (10% of starting material) was analyzed in parallel (6th lane). Immunoprecipitations were specific for the SNAP25 IgGs (compare 1st and 5th lanes). Ca 2ϩ stimulated the co-immunoprecipitation of synaptotagmin I (2nd lane versus 3rd lane), which was reduced by BoNT E treatment (4th lane). In contrast, the co-immunoprecipitation of syntaxin was unaffected by Ca 2ϩ or toxin treatment. CAPS was not coimmunoprecipitated with SNAP25. B, Ca 2ϩ stimulates the binding of native synaptotagmin I to immobilized SNAP25. Rat brain detergent extract was incubated for 2 h on ice with GSSG-agarose beads or 20 g of GST-SNAP25 immobilized on GSSG-agarose beads at indicated concentrations of Ca 2ϩ (2 mM EGTA indicated as 0 mM Ca 2ϩ ). Parallel incubations were conducted with immobilized GST-SNAP25 quantitatively cleaved with BoNT A or E. The washed beads were eluted for SDS gel analysis and immunoblotting with synaptotagmin I or syntaxin antibodies. Little binding of synaptotagmin I to control beads was detected (1st and 2nd lanes). Ca 2ϩ stimulated the binding of synaptotagmin I to SNAP25 with an EC 50 ϳ300 M (2nd to 5th lanes). Cleavage of SNAP25 with BoNT A or E markedly reduced the Ca 2ϩ -dependent binding of synaptotagmin I (7th and 8th lanes). Syntaxin binding to SNAP25 was largely unaffected by Ca 2ϩ or BoNT treatment.

FIG. 5. Synaptotagmin III interactions with SNAP25 are stimulated at lower Ca 2؉ concentrations.
A, Ca 2ϩ stimulates SNAP25 binding to synaptotagmin III. 10 g of GST-cytoplasmic domain fusion protein of synaptotagmin III immobilized on GSSG-agarose beads was incubated with 10 g of SNAP25 in the presence of 2 mM EGTA without or with adequate CaCl 2 to generate the indicated concentrations of free ionic Ca 2ϩ . The mean of three experiments is plotted with range indicated. B, BoNT A and BoNT E inhibit Ca 2ϩ -stimulated interactions between synaptotagmin III and SNAP25. 10 g of GST-cytoplasmic domain fusion protein of synaptotagmin III immobilized on GSSGagarose beads was incubated with 10 g of intact SNAP25, BoNT A-cleaved SNAP25, or BoNT E-cleaved SNAP25 as indicated either in the absence or presence of 1 mM Ca 2ϩ . For 1st lane, SNAP25 was incubated with empty GSSG-agarose beads. ditions where heterodimers form. Under our conditions, the binding of syntaxin alone to immobilized synaptotagmin I was not detected in either the absence or presence of Ca 2ϩ (Fig. 6A,  1st two lanes). However, inclusion of SNAP25 promoted Ca 2ϩdependent syntaxin binding to synaptotagmin I, which increased progressively with increasing SNAP25 concentration (Fig. 6A). The results indicate that Ca 2ϩ -dependent syntaxin binding under these conditions is mediated by the formation of syntaxin/SNAP25 heterodimers and the Ca 2ϩ -dependent binding of the heterodimers to synaptotagmin I. Syntaxin/SNAP25 heterodimers were as effective as ligands for Ca 2ϩ -dependent interactions with synaptotagmin I as were SNAP25 monomers. The Ca 2ϩ -dependent binding of syntaxin to synaptotagmin I was strongly inhibited by cleavage of the C terminus of SNAP25 with BoNT E and to a lesser extent BoNT A (Fig. 6B). Since the C terminus of SNAP25 is not required for the formation of syntaxin/SNAP25 heterodimers (46), the results indicate that Ca 2ϩ -dependent binding of binary SNARE complexes to synaptotagmin I requires the C terminus of SNAP25.
It has been suggested that Ca 2ϩ may act in exocytosis to trigger the formation of ternary SNARE complexes in trans to drive bilayer fusion (14). To determine whether synaptotagmin I also exhibited Ca 2ϩ -stimulated interactions with ternary SNARE complexes, VAMP was included in the preincubations. Under these conditions, the Ca 2ϩ -dependent binding of VAMP to synaptotagmin I was observed, whereas little or no VAMP binding was detected in the absence of SNAP25 or syntaxin (i.e. in the absence of ternary complex formation) (Fig. 7A). The results indicate that Ca 2ϩ -dependent VAMP binding is mediated through ternary SNARE complex binding to synaptotagmin I, which was confirmed by finding that synaptotagmin I-bound SNARE proteins were present in SDS-resistant complexes (data not shown). Binary and ternary SNARE complexes exhibited similar Ca 2ϩ -dependent binding to synaptotagmin I as detected by bound SNAP25 (Fig. 7B). Inclusion of VAMP to form heterotrimeric complexes reduced syntaxin binding relative to that of SNAP25 (Fig. 7B), which is consistent with the different stoichiometries of ternary (syntaxin⅐SNAP25⅐VAMP as 1:1:1) and binary (syntaxin⅐SNAP25 as 2:1) SNARE complexes (9).
The Ca 2ϩ -dependent binding of VAMP to synaptotagmin I was strongly inhibited when BoNT E-or BoNT A-treated SNAP25 was used to assemble ternary complexes (Fig. 7C). The inhibition of VAMP binding could result either from destabilization of ternary complexes by BoNT treatment (8) or from inhibition of the binding of ternary complexes to synaptotagmin I by the removal of essential portions of SNAP25 by toxin cleavage. Under our conditions, BoNT E treatment was found to destabilize ternary SNARE complexes (Fig. 7D) as previously reported (8), and this likely accounts for the inhibition of Ca 2ϩ -dependent VAMP binding to synaptotagmin I (Fig. 7C). In contrast, BoNT A treatment had virtually no effect on the stability of SDS-resistant ternary SNARE complexes (Fig. 7D) as previously reported (8). Therefore, the reduction in SNARE complex binding to synaptotagmin I resulting from BoNT A treatment (Fig. 7C) implies that SNAP25 C-terminal residues removed by BoNT A cleavage are essential for the Ca 2ϩ -dependent interaction of synaptotagmin I with ternary SNARE complexes.

DISCUSSION
The data in this study indicate that synaptotagmin exhibits a high affinity Ca 2ϩ -dependent interaction with SNAP25 monomers, with syntaxin/SNAP25 heterodimers, and with syntaxin/SNAP25/VAMP heterotrimers. In each case, an essential determinant for Ca 2ϩ -dependent synaptotagmin binding is the C terminus of SNAP25. The results provide an explanation for the reversal of BoNT A inhibition of neurosecretion at elevated Ca 2ϩ concentrations observed in classical studies. They also suggest a plausible basis for the Ca 2ϩ regulation of exocytosis.
In 1977 Thesleff and co-workers (32) reported that treatment with a K ϩ channel blocker that increases synaptic accumulation of Ca 2ϩ could partially overcome the paralytic effects of BoNT A in the neuromuscular junction. Similarly Simpson (47) reported that BoNT A decreased the Ca 2ϩ cooperativity of neurotransmitter release and suggested that BoNT A acted close to the site of Ca 2ϩ action in neurotransmitter release to cause a decreased affinity or efficacy of Ca 2ϩ . Reversal of inhibition by Ca 2ϩ is relatively unique to BoNT A and has not been observed for tetanus toxin, BoNT B, BoNT F, or BoNT D (41, 48 -50) and only to a lesser extent for BoNT E (51,52). These classical studies, conducted before the molecular targets of BoNT action were known, foreshadowed the conclusion of the present study that the target of BoNT A action, SNAP25, is closely linked to the Ca 2ϩ regulation of neurotransmitter release.
Studies of toxin action in permeable cell assays have found BoNT A to be far less effective in inhibiting Ca 2ϩ -dependent secretion than BoNT E or other toxins in contrast to its effectiveness in intact nerve cells (28,33,53). The present results clarify this distinctive feature of BoNT A by finding that the high Ca 2ϩ concentrations used to optimally activate exocytosis in permeable cells results in a reversal of BoNT A inhibition. Because such relatively high Ca 2ϩ concentrations are rarely achieved in intact cell studies, these findings readily account FIG. 6. Synaptotagmin I binds binary syntaxin⅐SNAP25 complexes in a Ca 2؉ -dependent manner. A, syntaxin binding to synaptotagmin I is mediated through heterodimer formation. 5 g of GSTsynaptotagmin I immobilized on GSSG-agarose beads was incubated with 1 M syntaxin in the presence or absence of 1 mM Ca 2ϩ with increasing concentrations of SNAP25 from 0 -1 M. Binding was assessed by Western blotting with syntaxin monoclonal and SNAP25 polyclonal antibodies. Ca 2ϩ -dependent syntaxin binding was not detected in the absence of SNAP25. B, Ca 2ϩ -dependent binding of binary complexes to synaptotagmin I requires the C terminus of SNAP25. Binding incubations similar to those of A were conducted with 1 M SNAP25, BoNT A-treated SNAP25, or BoNT E-treated SNAP25 in the absence or presence of 1 M syntaxin. Bound SNAREs were detected by Western blotting with syntaxin and SNAP25 antibodies. Syntaxin binding to synaptotagmin I (upper panel), mediated through formation of binary syntaxin⅐SNAP25 complexes, was inhibited by treatment of SNAP25 with BoNT A or BoNT E. The 15% SDS-acrylamide gels resolved SNAP25 and co-migrating BoNT A-cleaved SNAP25 from higher mobility BoNT E-cleaved SNAP25 (lower panel). An artifact migrated immediately behind SNAP25 (see 11th and 12th lanes). BoNT E-cleaved SNAP25 bound to synaptotagmin I was equivalent to SNAP25 bound in the absence of Ca 2ϩ . BoNT A-cleaved SNAP25 bound to synaptotagmin I was equivalent to that observed in Fig. 3A at 1  for reported discrepancies between intact cell and permeable cell studies on the efficacy of BoNT A as an inhibitor of regulated secretion (54). Decreased inhibition by BoNT A has also been observed in intact cell studies when high Ca 2ϩ concentrations are used to elicit exocytosis (43).
Because BoNT A cleavage of SNAP25 decreased the Ca 2ϩ sensitivity of regulated exocytosis, it seemed possible that SNAP25 was involved in the Ca 2ϩ sensing reactions of exocytosis. This idea motivated direct binding studies of SNAP25 with synaptotagmin I, a proposed Ca 2ϩ sensor for exocytosis (1,18,22). A Ca 2ϩ -independent interaction between these proteins, which does not require the C terminus of SNAP25, was described previously (44). The binding studies reported here, in contrast, revealed a novel Ca 2ϩ -dependent interaction between SNAP25 and synaptotagmin that requires the C terminus of SNAP25. Truncation of SNAP25 from Ile 181 to Gly 206 by BoNT E treatment abolished Ca 2ϩ -dependent synaptotagmin I binding, which correlates with the strong inhibitory effect of this toxin on Ca 2ϩ -activated exocytosis. Truncation of SNAP25 from Arg 198 to Gly 206 by BoNT A treatment, in contrast, modified the Ca 2ϩ dependence for the binding of synaptotagmin I to SNAP25 by shifting it to higher Ca 2ϩ concentrations, an effect that was similar to that of BoNT A on Ca 2ϩ -activated exocytosis. The qualitatively similar inhibition by BoNT A and reversal at elevated Ca 2ϩ concentrations for both Ca 2ϩ -dependent synaptotagmin-SNAP25 binding and for Ca 2ϩ -dependent exocytosis provides a striking correlation. This correlation is consistent with a role for SNAP25 as an important effector for synaptotagmin in the Ca 2ϩ -triggering reactions of exocytosis. Another possible basis for the Ca 2ϩ reversal of BoNT A inhibition of exocytosis is the cation-binding site in the ternary SNARE complex located near the BoNT A cleavage site (13,55); however, Ca 2ϩ binding to this site remains to be demonstrated.
Previous studies identified a number of Ca 2ϩ -dependent interactions for synaptotagmin including with acidic phospholipids, SV2, synaptotagmin itself, and syntaxin (23, 24 -27, 56, 57, 58), but it has been unclear which, if any, of these mediates the essential role of synaptotagmin in exocytosis. The characteristics of the synaptotagmin I-SNAP25 interaction reported here bear a striking resemblance to those described for the synaptotagmin I-syntaxin interaction (24 -27). Both are stimulated at high Ca 2ϩ concentrations, are not supported by Sr 2ϩ or Ba 2ϩ , and involve in part the membrane-proximal C2A domain of synaptotagmin I (45). Different synaptotagmin isoforms exhibit characteristic Ca 2ϩ dependences for interactions with both syntaxin and SNAP25. Synaptotagmin I interacts with both SNAREs at Ca 2ϩ concentrations similar to those required for triggering synaptic vesicle exocytosis (ϳ200 M), whereas synaptotagmin III requires the lower Ca 2ϩ concentrations characteristic of triggering dense-core vesicle exocytosis (ϳ10 M) (1)(2)(3)17). The similarity of Ca 2ϩ -dependent synaptotagmin binding to both SNAREs suggests the possibility that both plasma membrane SNAREs or complexes containing them are the Ca 2ϩ -dependent effectors for synaptotagmin.
Ca 2ϩ -dependent synaptotagmin binding to SNAP25 appeared to be of higher affinity than the binding to syntaxin. The FIG. 7. Synaptotagmin I binds ternary SNARE complexes in a Ca 2؉ -dependent manner. A, VAMP binding to synaptotagmin I is mediated by binding of ternary SNARE complexes. Overnight incubations were conducted with 1 M VAMP and 1 M syntaxin and/or 1 M SNAP25 as indicated. Preassembled complexes were incubated in the presence or absence of 1 mM Ca 2ϩ with 5 g of GST-synaptotagmin I immobilized on GSSG-agarose beads. Bound VAMP was assessed by Western blotting with a VAMP monoclonal antibody. Ca 2ϩ -stimulated VAMP binding to synaptotagmin I was detected when syntaxin and SNAP25 were present to form ternary complexes. B, comparison of synaptotagmin I binding by binary and ternary SNARE complexes. Syntaxin and SNAP25 were preincubated to form heterodimers or preincubated with VAMP to form heterotrimers. Binding of the SNARE complexes to synaptotagmin I in the presence or absence of 1 mM Ca 2ϩ was assessed. Similar amounts of SNAP25 were bound in a Ca 2ϩ -dependent manner for binary and ternary complex binding. Ternary complex binding resulted in reduced Ca 2ϩ -dependent syntaxin binding in accord with the reduced molar content of syntaxin in heterotrimeric compared with heterodimeric complexes. C, Ca 2ϩ -dependent interactions of ternary SNARE complexes with synaptotagmin I require the C terminus of SNAP25. Heterotrimeric complexes were assembled in overnight incubations with full-length, BoNT A-, or BoNT E-treated SNAP25. Binding to synaptotagmin I was determined in the absence (2 mM EGTA) or presence of Ca 2ϩ at 0.1, 0.3, and 1.0 mM as indicated. BoNT A treatment partially inhibited (by ϳ56%) and BoNT E treatment strongly inhibited (by ϳ72%) Ca 2ϩ -dependent ternary complex binding to synaptotagmin I. D, BoNT E treatment inhibits ternary SNARE complex assembly, whereas BoNT A treatment has no effect. SDS-resistant SNARE complexes were detected following assembly of complexes with intact, BoNT A-, and BoNT E-treated SNAP25. Samples were split for gel electrophoresis following incubation in sample buffer at 30 o C (Ϫ) or boiling (ϩ). SDS-resistant SNARE complexes of ϳ60 and 90 kDa, similar to those previously reported (9), were detected by Western blotting with syntaxin monoclonal antibodies. BoNT A treatment had little if any effect (Ͻ10%) on SDS-resistant complexes in three similar experiments. In contrast, BoNT E treatment reduced SDS-resistant complexes either partially (as shown) or completely in three similar experiments. Residual complexes following BoNT E treatment exhibited an altered electrophoretic mobility. apparent K D reported for Ca 2ϩ -dependent synaptotagmin Isyntaxin interactions is ϳ0.5-2 M (24, 25), whereas we estimated an apparent K D ϳ0.2 M for Ca 2ϩ -dependent synaptotagmin I-SNAP25 binding. More directly, when compared at the same concentration, Ca 2ϩ -dependent SNAP25 binding to immobilized synaptotagmin I was readily detected, whereas that of syntaxin was not (Fig. 5A). Syntaxin binding to synaptotagmin I was, however, evident in the presence of SNAP25 because syntaxin⅐SNAP25 binary complexes formed and bound to synaptotagmin I in a Ca 2ϩ -dependent manner. The C terminus of SNAP25 was required for optimal Ca 2ϩ -dependent interactions between synaptotagmin I and binary SNARE complexes. A similar high affinity Ca 2ϩ -dependent binding of ternary SNARE complexes to synaptotagmin I was observed, which also required the C terminus of SNAP25. These results indicate that Ca 2ϩ -dependent interactions between synaptotagmin and SNAP25 or SNARE complexes are mediated in part through the C terminus of SNAP25.
It was surprising that Ca 2ϩ -dependent synaptotagmin binding to SNAP25 or to binary or ternary SNARE complexes was so similar since SNAP25 may be largely unstructured as a monomer but acquires ␣-helicity in binary and ternary SNARE complexes (9). Possibly monomeric SNAP25 becomes structured in complexes with synaptotagmin. The C-terminal region of SNAP25 likely represents a direct site for Ca 2ϩ -dependent synaptotagmin binding since Ca 2ϩ -dependent binding to a Cterminal SNAP25 fragment can be detected, and a C-terminal SNAP25 peptide inhibits binding. 2 The region Ile 181 -Gln 197 may constitute the core of this binding site since BoNT E cleavage abolishes Ca 2ϩ -dependent interactions with synaptotagmin. The region Arg 198 -Gly 206 may indirectly affect binding since cleavage by BoNT A reduces synaptotagmin binding in a manner that is partially compensated by elevating the Ca 2ϩ concentration.
The recent structural elucidation of a protease-resistant core of the ternary SNARE complex as consisting of a four-helix bundle in parallel orientation (13) reveals a potential site for Ca 2ϩ -dependent interactions of synaptotagmin. The C-terminal membrane-proximal region of syntaxin (residues 220 -266), reported to mediate Ca 2ϩ -dependent interactions with synaptotagmin I (Refs. 24, 25, and 62 but also see Ref. 59), is juxtaposed to the BoNT-sensitive C terminus of SNAP25 (residues 181-206) (e.g. syntaxin Ile 230 aligns with SNAP25 Ile 178 of SNAP25; Ref. 13). Residues from these regions contributed by SNAP25 and syntaxin in binary or ternary SNARE complexes may constitute the essential Ca 2ϩ -dependent binding site for synaptotagmin.
The Ca 2ϩ triggering of membrane fusion could be mediated by the Ca 2ϩ -dependent binding of synaptotagmin on the vesicle to syntaxin and SNAP25 on the plasma membrane. Ca 2ϩ could promote synaptotagmin binding to syntaxin/SNAP25 heterodimers, which would bring VAMP on the vesicle into proximity with syntaxin/SNAP25 heterodimers to form ternary SNARE complexes that could initiate fusion (16). Alternatively, Ca 2ϩ could stimulate the binding of synaptotagmin to partially "zippered" pre-assembled ternary SNARE complexes bridging the vesicle and plasma membrane to promote full zippering and the initiation of fusion (10,14,15,60). Finally, Ca 2ϩ -independent interactions might drive the assembly of pre-fusion complexes between synaptotagmin and SNARE proteins (7). The Ca 2ϩ -dependent interaction between synaptotagmin and SNAP25 described here could mediate molecular interactions within the complex that promote a fusion-competent conformation.
The C terminus of SNAP25 is targeted for proteolysis by three of the clostridial neurotoxins (BoNT E, A, and C1) indicating its importance, but its precise role in Ca 2ϩ -activated neurotransmitter secretion has not previously been identified. Studies have shown that the C-terminal 9 residues of SNAP25 removed by BoNT A are required for VAMP interactions (46) and that removal of the C-terminal 26 residues by BoNT E cleavage compromises ternary SNARE complex formation (8).
Our results indicate that the C-terminal 9 residues of SNAP25 affect the Ca 2ϩ dependence of exocytosis and the Ca 2ϩ dependence of synaptotagmin binding to SNAP25. The C-terminal region of SNAP25 between BoNT A and E cleavage sites (Ile 181 -Gln 197 ) is essential for Ca 2ϩ -triggered membrane fusion (28) and for Ca 2ϩ -dependent synaptotagmin binding. The recent finding that a C-terminal SNAP25 fragment is capable of promoting Ca 2ϩ -dependent rescue of BoNT E-inhibited exocytosis further reinforces the view that the C terminus of SNAP25 participates directly in the Ca 2ϩ triggering of exocytosis (61). By mediating Ca 2ϩ -dependent interactions between synaptotagmin and SNARE complexes, the C terminus of SNAP25 may constitute a key element of a Ca 2ϩ switch for regulated exocytosis.