A novel function for the second C2 domain of synaptotagmin. Ca2+-triggered dimerization.

Synaptotagmin serves as the major Ca sensor for regulated exocytosis from neurons. While the mechanism by which synaptotagmin regulates membrane fusion remains unknown, studies using Drosophila indicate that the molecule functions as a multimeric complex and that its second C2 domain is essential for efficient excitation-secretion coupling. Here we describe biochemical data that may account for these phenomena. We report that Ca causes synaptotagmin to oligomerize, primarily forming dimers, via its second C2 domain. This effect is specific for divalent cations that can stimulate exocytosis of synaptic vesicles (Ca Ba, Sr > Mg) and occurs with an EC value of 3-10 μM Ca. In contrast, a separate Ca-dependent interaction between synaptotagmin and syntaxin, a component of the fusion apparatus, occurs with an EC value of 100 μM Ca and involves the synergistic action of both C2 domains of synaptotagmin. We propose that Ca triggers two consecutive protein-protein interactions: the formation of synaptotagmin dimers at low Ca concentrations followed by the association of synaptotagmin dimers with syntaxin at higher Ca-concentrations. Our findings, in conjunction with physiological studies, indicate that the Ca-induced dimerization of synaptotagmin is important for the efficient regulation of exocytosis by Ca.

Rapid chemical signaling between neurons relies on the regulated secretion of neurotransmitters stored in synaptic vesicles. In the resting nerve terminal, a population of synaptic vesicles is tightly bound, or docked, to the presynaptic plasma membrane. When the nerve terminal is depolarized, Ca 2ϩ enters via voltage-gated Ca 2ϩ channels. High local concentrations of Ca 2ϩ then rapidly trigger the fusion of a subpopulation of docked vesicles with the plasma membrane, resulting in the release of neurotransmitters into the synaptic cleft.
While the mechanism(s) by which Ca 2ϩ ions trigger membrane fusion remains obscure, strong evidence indicates that the synaptic vesicle protein, synaptotagmin I, plays an essential role in excitation-secretion coupling. To date, nine isoforms of synaptotagmin have been cloned and characterized (Perin et al., 1990;Geppert et al., 1991;Mizuta et al., 1994;Hilbush and Morgan, 1994;Hudson and Birnbaum, 1995;Li et al., 1995). The members of this family are integral membrane proteins that span the vesicle membrane once and possess a short ami-no-terminal intravesicular domain and a large cytoplasmic domain (Perin et al., 1990. The cytoplasmic domain contains two repeats homologous to the C2 domains found in Ca 2ϩ -dependent but not Ca 2ϩ -independent isoforms of protein kinase C, suggesting that this conserved motif comprises a Ca 2ϩ binding domain (reviewed by Nishizuka (1988)). Homologous domains have also been identified in other proteins, at least some of which interact with lipids in a Ca 2ϩ -dependent manner (Clark et al., 1991;Stahl et al., 1988;Vogel, et al., 1988;Shirataki et al., 1993). Subsequent studies demonstrated that synaptotagmin indeed binds Ca 2ϩ and negatively charged phospholipids in a mutually dependent manner (Brose et al., 1992) via its first C2 domain (Davletov and Sü dhof, 1993;Fukuda et al., 1994).
The localization of synaptotagmin on synaptic vesicles and its ability to bind Ca 2ϩ ions suggested that the molecule may serve as a Ca 2ϩ sensor in exocytosis (Brose et al., 1992). This hypothesis has been the subject of numerous studies. For example, the synaptotagmin gene has been disrupted in Drosophila, Caenorhabditis elegans, and mice DiAntonio and Schwarz, 1994;Broadie et al., 1994;Nonet et al., 1993;Geppert et al., 1994). In all cases, perturbation of synaptotagmin expression results in decreased excitation-secretion coupling. In fact, disruption of the synaptotagmin I gene in mice virtually abolishes the fast component of Ca 2ϩ -dependent exocytosis, while both spontaneous synaptic vesicle fusion and Ca 2ϩ -independent exocytosis triggered by ␣-latrotoxin are unaffected (Geppert et al., 1994). Therefore, the fusion apparatus, while fully functional in the synaptotagmin I-deficient mice, is not appropriately regulated by Ca 2ϩ . These studies suggest that synaptotagmin either acts as the Ca 2ϩ sensor for exocytosis or is essential in some other capacity to maintain efficient excitation-secretion coupling, for example by docking vesicles close to the site of Ca 2ϩ influx (Neher and Penner, 1994). The former interpretation is strengthened by genetic studies in Drosophila, which demonstrated that discrete mutations in the gene encoding synaptotagmin can affect the cooperativity of the Ca 2ϩ dependence of exocytosis . Furthermore, the phenotypes resulting from crosses between Drosophila carrying distinct mutant synaptotagmin alleles suggest that the molecule functions as a multimeric complex and point to an essential function for its second C2 domain.
The precise mechanism by which synaptotagmin functions in excitation-secretion coupling remains ill-defined. To address this issue, efforts have been made to identify effectors of Ca 2ϩsynaptotagmin action. Synaptotagmin was shown to form Ca 2ϩ -independent complexes with neurexins (Petrenko et al., 1991), a family of neuronal cell surface proteins, and with the clathrin adaptor protein AP-2 (Zhang et al., 1994). Synaptotagmin also interacts with the plasma membrane protein syntaxin 1 (Bennett et al., 1992), an essential component of the synaptic vesicle fusion apparatus (Söllner et al., 1993a, b;Blasi et al., 1993;Schulze et al., 1995). Recently, we and others Li et al., 1995) have reported that the interaction between synaptotagmin and syntaxin is promoted by Ca 2ϩ . The Ca 2ϩ dependence and divalent cation specificity of binding were consistent with those observed for exocytosis, suggesting that this interaction may comprise a step in transducing Ca 2ϩ transients to membrane fusion.
We have continued to look for additional effectors of Ca 2ϩsynaptotagmin action and report here that the second C2 domain of the protein is involved in the Ca 2ϩ -triggered formation of homodimers. These findings may begin to provide the biochemical basis for the defects in synaptic transmission imparted by mutant synaptotagmins in vivo.

EXPERIMENTAL PROCEDURES
Binding Assays-Rat brain detergent extracts were prepared by solubilizing a crude synaptosome fraction using Triton X-100. Two rat brains were homogenized in 30 ml of 320 mM sucrose with 10 strokes at 900 rpm, using a Teflon-glass homogenizer. All manipulations were carried out on ice. The homogenate was centrifuged at 5,000 rpm for 2 min in an SS34 rotor, and a crude synaptosome fraction was collected by centrifugation of the supernatant at 11,000 rpm for 12 min in an SS34 rotor. Synaptosomes were solubilized for 45 min at a detergent:protein ratio of 10:1 (w/w) with 1% Triton X-100 in 50 mM HEPES, pH 7.6, 100 mM NaCl containing protease inhibitors (1 mM phenylmethlysulfonyl fluoride, 2 g/ml pepstatin, 20 g/ml aprotinin). Insoluble material was removed by centrifugation at 50,000 rpm in a TLA 100.3 rotor for 15 min. One-mg aliquots of the rat brain detergent extract (1 mg/ml) were incubated with (0.5 nmol) various regions of synaptotagmin, immobilized to glutathione-Sepharose as GST 1 fusion proteins, in EGTA or EGTA-buffered Ca 2ϩ , as indicated, for 2.5 h. Samples were washed three times in binding buffer, boiled in SDS-sample buffer and subjected to SDS-PAGE and immunoblot analysis as described . Native synaptotagmin was detected using a monoclonal antibody directed against the luminal domain of the protein (Cl 604.1; ) that is absent from all of the recombinant immobilized synaptotagmin constructs. Monoclonal antibodies directed against syntaxin (HPC-1; Barnstable et al. (1985)), synaptophysin (7.2; Jahn et al. (1985)), SNAP-25 (71.2;Hanson et al. (1995)), rab 3A (42.1; Matteoli et al. (1991)), and synaptobrevin (69.1; Edelmann et al. (1995)), have been described previously.
The association of recombinant regions of synaptotagmin with recombinant syntaxin was assayed by co-immunoprecipitation using antisyntaxin antibodies. Recombinant full-length His 6 -syntaxin and His 6 -C2AB, or C2AB, C2A L , and C2B L generated by thrombin cleavage of their respective GST-fusion proteins, were incubated at the indicated concentrations in 50 mM HEPES, pH 7.6, 100 mM NaCl, and 0.5% Triton X-100 supplemented with EGTA or CaCl 2 for 2.5 h. His 6 -syntaxin was immunoprecipitated by incubating the samples with purified HPC-1 IgG (5 g) for 1.5 h and 12 l of protein G-Sepharose Fast-flow (Pharmacia Biotech Inc.) for 1 h. The immunoprecipitates were washed three times, subjected to SDS-PAGE, and immunoblotted, as described above, using rabbit antisera (kindly provided by T. C. Sü dhof) directed against either the first C2 domain (antisera W855; used to detect C2AB and C2A S ) or the second C2 domain of synaptotagmin (antisera Y940; used to detect C2B L ).
Sucrose Density Gradients-One-half-ml aliquots (l mg/ml) of the rat brain Triton X-100 extract described above were supplemented with 2 mM EGTA or 1 mM CaCl 2 and resolved on sucrose gradients (5-15% sucrose; 12 ml total volume) in 50 mM HEPES, pH 7.5, 100 mM NaCl, 0.5% Triton X-100, and 2 mM EGTA or 1 mM CaCl 2 using a Beckman SW41 rotor. Samples were centrifuged at 35,000 ϫ rpm (150,000 ϫ g) for 16 h at 4°C. Twenty-four 0.5-ml fractions were collected, and aliquots of the first 12 fractions (corresponding to the top of the gradient) were subjected to SDS-PAGE and immunoblot analysis.
Recombinant Proteins-Soluble forms of full-length rat syntaxin 1A (His 6 -syntaxin) and the cytoplasmic domain of rat synaptotagmin 1 (residues 97-421; His 6 -C2AB) were prepared by subcloning into pTrcHisA (InVitrogen), resulting in fusion proteins with T7 and His 6 tags at their amino termini . Immobilized syn-aptotagmins were prepared as GST fusion proteins by amplifying the indicated regions of the cDNA using the polymerase chain reaction and subcloning the products into pGEX-2T (Pharmacia). Recombinant proteins were expressed in Escherichia coli and purified as described . Soluble forms of the first (C2A L ) and second (C2B L ) C2 domains of synaptotagmin were prepared by thrombin cleavage of the corresponding GST fusion protein as described  Miscellaneous Procedures-Quantitation of the immunoblots was carried out using a Molecular Dynamics PhosphorImager and Image-Quant software. For the Ca 2ϩ dose-response analyses, samples were buffered with 1 mM EGTA, and the total Ca 2ϩ added to yield the indicated free Ca 2ϩ concentrations was determined as described previously .

RESULTS
To determine the effect of Ca 2ϩ ions on the interaction of synaptotagmin with other synaptic proteins, we immobilized the cytoplasmic domain of the protein as a GST fusion protein (GST-C2AB) bound to glutathione-Sepharose and used it as an affinity matrix. When incubated with rat brain detergent extracts in the presence of Ca 2ϩ or EGTA, we observed a striking Ca 2ϩ -dependent association of native synaptotagmin with GST-C2AB (Fig. 1). Binding was specific for synaptotagmin since no detectable binding of other synaptic proteins, including syntaxin, synaptophysin, SNAP-25, and synaptobrevin, was observed. To determine the region of synaptotagmin that mediates Ca 2ϩ -dependent oligomerization, different domains of the molecule were immobilized as GST fusion proteins and analyzed for native synaptotagmin binding as described above.
FIG. 1. Ca 2؉ -induced self-association of synaptotagmin is mediated through the second C2 domain. Domains of synaptotagmin (upper panel) fused to GST (0.5 nmol), were immobilized using glutathione-Sepharose and incubated for 2.5 h with rat brain detergent extracts in 2 mM EGTA (ϪCa 2ϩ ) or 1 mM Ca 2ϩ (ϩCa 2ϩ ). Beads were washed with binding buffer, and bound proteins were analyzed by SDS-PAGE and immunoblotting using 125 I-protein A as described . One-tenth of the samples were analyzed. For comparison, 4 g of total rat brain detergent extract were analyzed in parallel (total). Native synaptotagmin was detected using an antibody directed against the luminal domain (604.1; ). The amino acid residues of each recombinant synaptotagmin construct are listed in the upper panel (right). TMR, transmembrane region. Boxed regions indicate sequence conservation with the C2 domains of protein kinase C, as defined by Perin et al. (1990). The C2 domain closest to the transmembrane segment is designated A; the more distal, carboxyl-terminal C2 domain, is designated B.
Long (with flanking sequences; GST-C2A L ) and short (without flanking sequences; GST-C2A S ) versions of the isolated first C2 domain did not bind any of the synaptic proteins analyzed. In contrast, a long version of the isolated second C2 domain (GST-C2B L ) bound native synaptotagmin as efficiently as the fragment containing most of the cytoplasmic domain of the protein (GST-C2AB). This property was preserved in the minimal second C2 domain sequence (GST-C2B S ) originally defined by its homology to protein kinase C (Perin et al. (1990) ; Fig. 1). Therefore, the second C2 domain mediates Ca 2ϩ -dependent self-association of synaptotagmin.
It has been reported previously that the first C2 domain of synaptotagmin mediates the Ca 2ϩ -dependent component of phospholipid binding by the molecule (Fukuda et al., 1994;Chapman et al., 1995), while the second C2 domain specifically mediates Ca 2ϩ -independent binding to AP-2 and IP 4 (Zhang et al., 1994;Fukuda et al., 1994). Our finding that C2B mediates Ca 2ϩ -dependent oligomerization demonstrates yet another functionally divergent property of the individual C2 domains of synaptotagmin and represents the first Ca 2ϩ -dependent interaction that directly involves the second C2 domain of the protein.
We and others have recently demonstrated that synaptotagmin binds to syntaxin in a Ca 2ϩ -dependent manner Li et al., 1995). Therefore, the failure of native syntaxin to bind to the immobilized GST-C2AB shown in Fig. 1 was surprising. However, when C2AB is removed from the GST-glutathione-Sepharose by thrombin cleavage it then associates with immunoprecipitated syntaxin in a Ca 2ϩ -dependent manner (Fig. 2). Ca 2ϩ stimulated the binding of C2AB to syntaxin ϳ6 -8-fold, in agreement with our previous observations . This effect is not due to the Ca 2ϩ -dependent oligomerization of C2AB associated with syntaxin since recombinant C2AB only weakly oligomerizes in a Ca 2ϩdependent manner (described in detail below). Therefore, fusion of C2AB to GST and/or its immobilization onto glutathione-Sepharose diminishes its ability to bind syntaxin in a Ca 2ϩdependent manner. To further compare Ca 2ϩ -dependent synaptotagmin oligomerization and syntaxin binding, we examined the abilities of the isolated C2 domains, C2A L and C2B L , to co-immunoprecipitate with syntaxin in the presence and absence of Ca 2ϩ . As shown in Fig. 2, under the conditions of our assay, C2A L did not co-immunoprecipitate with syntaxin either in the presence or absence of Ca 2ϩ . Ca 2ϩ -independent binding of C2B L to syntaxin was observed, which was only weakly enhanced (ϳ1.5-fold) by Ca 2ϩ . These results indicate that both C2 domains act in a synergistic manner to confer Ca 2ϩ -dependent synaptotagmin binding to syntaxin.
Our data demonstrating that the isolated C2A region of synaptotagmin does not form a stable complex with syntaxin do not agree with the findings of Li et al. (1995), who reported a Ca 2ϩ -dependent interaction between immobilized GST-C2A S and native syntaxin. We observed low levels of Ca 2ϩ -dependent syntaxin binding to immobilized GST-C2A S only after overnight incubation periods or by using very high Ca 2ϩ concentrations (i.e. 2.5 mM Ca 2ϩ ). These findings, in conjunction with the data described above, indicate that both C2 domains of synaptotagmin are required for efficient binding to syntaxin.
We next tested the ability of different divalent cations to stimulate synaptotagmin oligomerization. Binding assays using immobilized GST-C2AB and rat brain detergent extracts were carried out as described in Fig. 1 in the presence of EGTA, Mg 2ϩ , Ca 2ϩ , Ba 2ϩ , or Sr 2ϩ . Self-association of synaptotagmin was specifically promoted by Ca 2ϩ ions (Fig. 3). These data agree with physiological studies demonstrating that Ca 2ϩ selectively triggers exocytosis (reviewed by Augustine et al. (1987)). The inability of Sr 2ϩ to promote oligomerization is of particular interest since this divalent cation is unable to support the fast component of Ca 2ϩ -dependent exocytosis . This finding is consistent with the notion that the Ca 2ϩ -dependent oligomerization of synaptotagmin may be an essential step in mediating this rapid component.
In parallel experiments, we found that recombinant His 6 -C2AB also readily binds to immobilized GST-C2AB. However, this interaction was weakly (Ͻ Ͻ2-fold) and inconsistently stimulated by Ca 2ϩ (data not shown). The more tightly regulated Ca 2ϩ -dependent interaction between native synaptotagmin derived from rat brain detergent extracts and GST-C2AB does not appear to be due to other factors (i.e. proteins or lipids) present in the extracts since the addition of the extract to the binding assay mixture does not restore Ca 2ϩ responsiveness to recombinant C2AB (data not shown). We also addressed the possibility that residues 1-96 of synaptotagmin, which are not present in C2AB, are essential for Ca 2ϩ -dependent synaptotagmin oligomerization. However, full-length synaptotagmin derived from transfected COS-7 cells or generated by in vitro transcription and translation also failed to consistently exhibit significant Ca 2ϩ -dependent GST-C2AB binding (data not shown). However, these full-length forms of synaptotagmin have greater electrophoretic mobilities than the native protein, suggesting that they are not properly processed or folded. In summary, Ca 2ϩ -dependent oligomerization of synaptotagmin FIG. 2. Ca 2؉ -dependent synaptotagmin-syntaxin binding involves the synergistic action of both C2 domains of synaptotagmin. Soluble forms of C2AB, C2A L , and C2B L were prepared by thrombin cleavage of the GST-fusion proteins described in Fig. 1  . Ca 2ϩ -dependent binding of these fragments (0.5 M) to His 6syntaxin (0.7 M) was analyzed by co-immunoprecipitation using antisyntaxin antibodies in the presence of 2 mM EGTA or 0.5 mM Ca 2ϩ , as described under "Experimental Procedures." Bound recombinant synaptotagmins were detected using rabbit antisera directed against the individual C2 domains of the protein. Immunoreactive bands were visualized using enhanced chemiluminescence (Amersham Corp.). Onetenth of the immunoprecipitates were analyzed for recombinant synaptotagmin binding. total corresponds to 1 ⁄100 of the starting material.
FIG. 3. Ca 2؉ selectively promotes synaptotagmin oligomerization. Binding of native synaptotagmin to GST-C2AB was assayed as described in Fig. 1 with the following additions: 2 mM EGTA, 0.5 mM Ca 2ϩ , 2.5 mM Mg 2ϩ , 0.5 mM Sr 2ϩ , or 0.5 mM Ba 2ϩ . The Mg 2ϩ sample also contained 2 mM EGTA. Binding was detected by SDS-PAGE and immunoblotting as described in Fig. 1. can only be measured when one of the binding partners is brain-derived native protein.
We therefore investigated the Ca 2ϩ dependence of synaptotagmin oligomerization by monitoring the binding of native synaptotagmin, derived from brain extracts, to immobilized GST-C2AB (Fig. 4, upper panel). Bound synaptotagmin was analyzed by immunoblotting as described and quantitated by PhosphorImager analysis, yielding an EC 50 value of 3-10 M Ca 2ϩ (Fig. 4, bottom panel, closed circles).
In a previous study we observed that the Ca 2ϩ dependence for the association of native synaptotagmin with immobilized syntaxin was composed of two components . The minor component required Ca 2ϩ concentrations in the low micromolar range (EC 50 ϳ1 M Ca 2ϩ ), very similar to Ca 2ϩ dependence reported here for the oligomerization of synaptotagmin (EC 50 ϭ 3-10 M Ca 2ϩ ). Thus it is possible that this minor component reflected the Ca 2ϩ -dependent oligomerization of native synaptotagmin associated with the immobilized syntaxin rather than being due to a direct increase in binding between the two different proteins. If this assumption is correct, only the second (and major) binding component, with an EC 50 value of approximately 180 M Ca 2ϩ , would reflect the direct association of native synaptotagmin with the immobilized recombinant syntaxin. To address this possibility, we reexamined the interaction of synaptotagmin with syntaxin using recombinant His 6 -C2AB, which does not exhibit significant Ca 2ϩ -dependent oligomerization. In principle, the Ca 2ϩ dependence for the interaction of His 6 -C2AB with syntaxin should be composed of a single component without a significant contribution due to His 6 -C2AB oligomerization. We mixed purified His 6 -C2AB with purified His 6 -syntaxin at different Ca 2ϩ concentrations and immunoprecipitated the resultant complexes with monoclonal antibodies directed against syntaxin. As expected (Fig. 4, middle and lower panels, open circles), the Ca 2ϩ dependence for binding of His 6 -C2AB to His 6 -syntaxin was composed of a single component with an EC 50 value of ϳ100 M Ca 2ϩ . This value is in reasonable agreement with the previously reported value of 180 M Ca 2ϩ described above. Therefore, synaptotagmin oligomerization (EC 50 ϭ 3-10 M Ca 2ϩ ) and syntaxin binding (EC 50 ϳ100 M Ca 2ϩ ) are clearly FIG. 5. Ca 2؉ induces the formation of native synaptotagmin dimers. Rat brain Triton X-100 extracts were prepared as described under "Experimental Procedures" and supplemented with either 2 mM EGTA or 1 mM Ca 2ϩ . Five hundred-l aliquots were subjected to density gradient centrifugation on 5-15% sucrose gradients. Aliquots from 12 0.5-ml fractions corresponding to the top of the gradient were analyzed by SDS-PAGE and immunoblotting using monoclonal antibodies directed against the indicated proteins and visualized by enhanced chemiluminescence (Amersham). The intensities of the synaptotagmin immunoreactive bands were determined by densitometry and are plotted in the bottom panel (closed circles, EGTA; open circles, Ca 2ϩ ). In the presence of Ca 2ϩ , the peak of synaptotagmin immunoreactivity shifts from ϳ50 to ϳ100 kDa. Molecular mass markers were as follows: 29 kDa (carbonic anhydrase), fraction 4; 45 kDa (ovalbumin), fraction 5; 67 kDa (bovine serum albumin), fraction 6; 150 kDa (alcohol dehydrogenase), fraction 9.
FIG. 4. Synaptotagmin oligomerization and syntaxin binding have distinct Ca 2؉ dependences. Upper panel, the association of native synaptotagmin with GST-C2AB was assayed as described in Fig.  1. Samples were buffered with 2 mM EGTA (EGTA) or 1 mM EGTA plus Ca 2ϩ to yield the indicated free Ca 2ϩ concentration. Middle panel, the association of His 6 -C2AB (0.4 M) with His 6 -syntaxin (0.7 M) was assayed by co-immunoprecipitation using anti-syntaxin antibodies as described under "Experimental Procedures." Samples were buffered with 2 mM EGTA (EGTA) or 1 mM EGTA plus Ca 2ϩ to yield the indicated free Ca 2ϩ concentration. Both His 6 constructs contain the T7 tag and were visualized using a T7 tag monoclonal antibody (Novagen) and 125 I-protein A. H and L denote the heavy and light chains of the anti-syntaxin antibody used for immunoprecipitation. Bottom panel, data were quantified by PhosphorImager analysis, and the percentage of maximal binding was plotted versus the free Ca 2ϩ concentration. Closed circles represent native synaptotagmin binding to GST-C2AB; open circles represent His 6 -C2AB binding to His 6 -syntaxin. E, 2 mM EGTA.
Finally, we addressed the stoichiometry of the Ca 2ϩ -induced synaptotagmin oligomers using sucrose density gradient centrifugation. Previous reports had shown that synaptotagmin was broadly distributed on density gradients when solubilized in CHAPS or Zwittergent 3-10 (Brose et al., 1992;Perin et al., 1991). Using Triton X-100-solubilized rat brain membranes, Garcia et al. (1995) reported a more compact distribution, with a major peak corresponding to the dimeric form of synaptotagmin (ϳ100 kDa). It is notable that in their study (Garcia et al., 1995), the free Ca 2ϩ was unbuffered and can therefore be assumed to be present at M levels. We therefore examined the effect of Ca 2ϩ and EGTA on the migration of native synaptotagmin on similar gradients, using Triton X-100-solubilized rat brain membranes as a protein source. As shown in Fig. 5, synaptotagmin peaks at a position corresponding to its monomeric molecular mass in EGTA (ϳ50 kDa) but exhibits a dramatic shift in the presence of Ca 2ϩ , peaking at a position corresponding to the molecular mass of its dimeric form (ϳ100 kDa). This shift was highly specific for synaptotagmin; no shift was observed for synaptophysin, rab 3A (Fig. 5), or other synaptic proteins, including the Ca 2ϩ and phospholipid binding protein rabphilin (data not shown). Under these dilute conditions, little of the total synaptotagmin is bound to syntaxin (Ͻ10% in Ca 2ϩ , as determined by immunoprecipitation of syntaxin), and as expected, syntaxin did not shift in the gradient (Fig. 5). Since this approach does not rely on the use of recombinant proteins, it demonstrates that Ca 2ϩ can induce the oligomerization of native synaptotagmin.

DISCUSSION
In order to understand how the Ca 2ϩ -binding protein synaptotagmin functions in regulated exocytosis it is essential to identify and characterize effectors for Ca 2ϩ -synaptotagmin action. Previous work has shown that these effectors include negatively charged phospholipids (Brose et al., 1992;Davletov and Sü dhof, 1993; and syntaxin Li et al., 1995). Here we report that Ca 2ϩ induces the formation of synaptotagmin dimers. Domain mapping analysis demonstrated that the second C2 domain (C2B) is involved in Ca 2ϩ -dependent oligomerization, while both C2 domains are required for efficient Ca 2ϩ -dependent synaptotagmin-syntaxin binding. Interestingly, the second C2 domain of synaptotagmin is essential for efficient Ca 2ϩ -regulated exocytosis in Drosophila . Few larvae with a mutant allele lacking the second C2 domain are viable. Of the surviving larvae, the cooperativity of excitationsecretion coupling is reduced 2-fold. In addition, an allele carrying a point mutation in the second C2 domain confers a higher requirement for extracellular Ca 2ϩ , while the cooperativity of release remains comparable with that observed in wild type flies. These studies suggest that the Ca 2ϩ -induced forma-tion of synaptotagmin dimers via the second C2 domain is essential for efficient excitation-secretion coupling.
We have determined the Ca 2ϩ dependences for the oligomerization of synaptotagmin and the interaction of synaptotagmin with syntaxin and found that they differ by an order of magnitude. These findings suggest a sequential order of events, which are summarized in the model shown in Fig. 6. As Ca 2ϩ levels rise from resting (0.1 M) to low M levels (3-10 M), synaptotagmin forms dimers via its second C2 domain. As Ca 2ϩ levels rise to concentrations capable of triggering exocytosis (20 -200 M;Heidelberger et al. (1994)), synaptotagmin dimers bind to syntaxin via the synergistic action of both C2 domains.
As described above, nine isoforms of synaptotagmin have been identified thus far. The findings presented in this study raise the possibility that low levels of Ca 2ϩ may trigger the association of different synaptotagmin isoforms with one another. Future experiments will be directed at determining whether such complexes form and whether these complexes exhibit distinct Ca 2ϩ -dependent interactions with downstream effectors. FIG. 6. Model depicting the sequential effects of Ca 2؉ on synaptotagmineffector interactions. Light shading denotes the first C2 domain of synaptotagmin (C2A), darker shading indicates the second C2 domain (C2B); syntaxin is shown in black. Arrows in the right panel denote the Ca 2ϩ -dependent interaction of C2A and the Ca 2ϩ -independent interaction of C2B of synaptotagmin with negatively charged phospholipids in either the synaptic vesicle or the plasma membrane.