The synaptic vesicle protein synaptotagmin I has been proposed to serve as a Ca2+ sensor for rapid exocytosis. Synaptotagmin spans the vesicle membrane once and possesses a large cytoplasmic domain that contains two C2 domains, C2A and C2B. Multiple Ca2+ ions bind to the membrane proximal C2A domain. However, it is not known whether the C2B domain also functions as a Ca2+-sensing module. Here, we report that Ca2+ drives conformational changes in the C2B domain of synaptotagmin and triggers the homo- and hetero-oligomerization of multiple isoforms of the protein. These effects of Ca2+ are mediated by a set of conserved acidic Ca2+ ligands within C2B; neutralization of these residues results in constitutive clustering activity. We addressed the function of oligomerization using a dominant negative approach. Two distinct reagents that block synaptotagmin clustering potently inhibited secretion from semi-intact PC12 cells. Together, these data indicate that the Ca2+-driven clustering of the C2B domain of synaptotagmin is an essential step in excitation-secretion coupling. We propose that clustering may regulate the opening or dilation of the exocytotic fusion pore.
C2 domains are conserved motifs found in nearly a hundred proteins that function in cellular processes ranging from membrane traffic to blood coagulation. These domains are composed of ∼130 amino acids and fold into compact eight-stranded distinct β-sandwich structures. Three flexible loops protrude from the tip of the domain. In some cases, two of these loops form a pocket that mediates the binding of Ca2+ and other divalent cations (Sutton and Sprang 1995; Sutton et al. 1995, Sutton et al. 1999; Grobler et al. 1996; Perisic et al. 1998; Ubach et al. 1998a,Ubach et al. 1998b; Pratt et al. 1999; Macedo-Ribeiro et al. 1999). Ca2+ binding regulates the interaction of some C2 domains with target molecules including lipids and proteins, while other C2 domains do not appear to bind Ca2+. Thus, the emerging view is that C2 domains are not necessarily Ca2+ -binding motifs, but rather function as modules that interact with a broad range of target molecules (reviewed by Nalefski and Falke 1996). In most cases, the biochemical and physiological functions of C2 domains remain to be defined.
Synaptotagmin was the first integral membrane protein discovered that harbored C2 domains (Matthew et al. 1981; Perin et al. 1990). This protein is localized to synaptic vesicles and large dense core secretory vesicles (Matthew et al. 1981). It spans the vesicle membrane once and has a short luminal domain and a large cytoplasmic region largely comprised of two C2 domains, designated C2A and C2B. These tandem C2 domains prompted biochemical studies, which demonstrated that synaptotagmin I is a Ca2+-binding protein (Brose et al. 1992; Davletov and Südhof 1993; Chapman and Jahn 1994). Subsequent gene disruption studies in mice, Caenorhabditis elegans, and Drosophila demonstrated that synaptotagmin I is essential for rapid and synchronous Ca2+-triggered release of neurotransmitters (Littleton et al. 1993; Nonet et al. 1993; DiAntonio and Schwarz 1994; Geppert et al. 1994; Littleton et al. 1994), suggesting that synaptotagmin I may function as a Ca2+ sensor that regulates neuronal exocytosis (Katz 1969).
Seven isoforms of synaptotagmin have been identified in the Drosophila genome (Littleton et al. 1999; Littleton, J.T., and E.R. Chapman, unpublished observations), indicating that the vertebrate gene family may be much larger than the current set of 12 isoforms. Furthermore, alternatively spliced forms of synaptotagmin have been identified that lack the transmembrane anchor (Fukuda and Mikoshiba 1999). These findings suggest that synaptotagmins may have widespread functions in intracellular membrane traffic.
A key to understanding the physiological function of synaptotagmin is to understand the biochemical properties of its C2 domains. The C2A domain of synaptotagmin I binds multiple Ca2+ ions (Sutton et al. 1995; Ubach et al. 1998b), and becomes partially inserted into membranes that contain anionic phospholipids (Chapman and Davis 1998; Davis et al. 1999; Bai et al. 2000). Equilibrium and kinetic Ca2+ binding properties of C2A are consistent with the Ca2+ requirement and speed of secretion (Davis et al. 1999).
The C2A domain is connected to C2B by a short tether and the tandem C2 domains cooperate to form complexes with components of the soluble N-ethyl maleimide-sensitive factor attachment protein receptor (SNARE) complex (Chapman et al. 1995, Chapman et al. 1996; Davis et al. 1999; Gerona et al. 2000). This complex is composed of the target SNAREs (t-SNAREs) syntaxin and SNAP-25 and the vesicle SNARE synaptobrevin (Söllner et al. 1993); assembly of the complex is necessary (Littleton et al. 1998; Chen et al. 1999) and may be sufficient (Weber et al. 1998) for exocytotic membrane fusion. Synaptotagmin forms direct contacts with both t-SNAREs at all stages of the SNARE complex assembly (data not shown; Chapman et al. 1995; Schiavo et al. 1997; Davis et al. 1999; Gerona et al. 2000) and can bind SNAREs and interact with membranes at the same time (Davis et al. 1999). These interactions suggest that synaptotagmin may regulate SNARE-mediated membrane fusion by modulating the conformation and/or assembly of SNARE complexes.
The interactions of synaptotagmin with SNAREs and anionic phospholipids are regulated by Ca2+ ligands that reside within C2A (Bai et al. 2000). Here, we address the question of whether and how the C2B domain of synaptotagmin functions as a Ca2+-sensing module during exocytosis. This question was prompted by the finding that mutations within the C2B domain of Drosophila synaptotagmin impair excitation-secretion coupling (DiAntonio and Schwarz 1994; Littleton et al. 1993, Littleton et al. 1994). A number of C2B–effector interactions have been identified including the following: AP-2 (Zhang et al. 1994; Jorgensen et al. 1995), SV2 (Schivell et al. 1996), β-SNAP (Schiavo et al. 1995), Ca2+ channels (Kim and Catterall 1997; Sheng et al. 1997), inositol polyphosphates (Fukuda et al. 1995) and, finally, homo-oligomerization (Chapman et al. 1996; Sugita et al. 1996). Of these interactions, only oligomerization was promoted by Ca2+.
While C2B was implicated in mediating Ca2+-triggered synaptotagmin clustering, this domain has not been directly shown to sense Ca2+ or transduce Ca2+ binding to clustering activity. Here, we report that the C2B domains of multiple synaptotagmin isoforms undergo Ca2+-driven conformational changes and can homo- and hetero-oligomerize with one another. We have characterized the structural requirements for these Ca2+ effects, and have applied a dominant negative approach to determine the role of Ca2+-triggered synaptotagmin clustering in Ca2+-regulated exocytosis. Our findings indicate that Ca2+-driven clustering of synaptotagmin constitutes a novel step in excitation-secretion coupling.
Materials and Methods
cDNA encoding rat synaptotagmin Ia (Perin et al. 1990) and Ib (Osborne et al. 1999) were provided by T.C. Südhof (University of Texas Southwestern Medical Center, Dallas, TX) and G. Schiavo (Imperial Cancer Research Fund, London, UK; Osborne et al. 1999), respectively. pGEX expression vectors to produce the cytoplasmic domains of synaptotagmins II–XIII (Li et al. 1995) were also provided by T.C. Südhof. pGEX expression vectors to produce the C2B domains of mouse synaptotagmins III and IV (Fukuda et al. 1995) were provided by M. Fukuda (Brain Science Institute, Ibaraki, Japan) and the C2B domains for rat synaptotagmins V–VIII were generated by PCR and were subcloned via BamHI and EcoRI sites into pGEX-2T. pGEX expression vectors to produce the C2A and C2B domain of rabphilin (Ubach et al. 1998a) were provided by J. Rizo (University of Texas Southwestern Medical Center, Dallas, TX).
The cytoplasmic (residues 96–421) and C2A domains (residues 96–265) of synaptotagmin Ia and Ib (see Fig. 1) were prepared as previously described (Chapman et al. 1995, Chapman et al. 1996). A new C2B domain (residues 248–421) was generated from the synaptotagmin Ib sequence reported by Osborne et al. 1999 and subcloned into pGEX-2T as previously described (Chapman et al. 1996). A series of point mutations (see Fig. 3) were generated by PCR and also expressed as GST fusion proteins. Soluble synaptotagmin fragments were prepared by thrombin cleavage of GST fusion proteins using thrombin (Chapman et al. 1996). His6-synprint (Sheng et al. 1997) was prepared as previously described (Chapman et al. 1998).
All recombinant protein binding assays were carried out by immobilizing one protein on glutathione-Sepharose beads. Immobilized fusion proteins were incubated with soluble binding partners for 1–2 h at 4°C in TBS (20 mM Tris and 150 mM NaCl) plus 0.5% Triton X-100 with either 2 mM EGTA or 1 mM free Ca2+, Mg2+, Ba2+, or Sr2+. Beads were washed three times in binding buffer. Proteins were solubilized by boiling in SDS sample buffer and subjected to SDS-PAGE. In most cases, proteins were visualized by staining with Coomassie blue. However, synprint binding was detected by immunoblotting using a monoclonal antibody directed against the T7-tag (Novagen).
All GST pull-down assays using brain detergent extracts were carried out as described (Chapman et al. 1998). Bound proteins were detected by immunoblotting using mouse monoclonal antibodies directed against the luminal domain of synaptotagmin I (604.4) and α/β-SNAP (77.1; this antibody recognizes α and β SNAP equally well) that were kindly provided by S. Engers and R. Jahn (Max Planck Institute for Biophysical Chemistry, Goettingen, Germany). The antibody directed against α-adaptin (100.2) was obtained from Sigma-Aldrich.
All binding assays were carried out at least four times and representative experiments are shown in the figures.
Immobilized GST fusion proteins were incubated for 1 h at room temperature in TBS with the indicated concentrations of chymotrypsin or trypsin in the presence of either 2 mM EGTA or 1 mM free Ca2+. Reactions were stopped by boiling in SDS sample buffer and samples were subjected to SDS-PAGE. Proteolytic fragments were visualized by staining with Coomassie blue, and samples were analyzed for the differential accumulation of proteolytic fragments in EGTA versus Ca2+. Ca2+ dose–response assays were carried out in a similar manner using a fixed protease concentration. Free [Ca2+] was determined using a Ca2+ electrode as previously described (Davis et al. 1999). Accumulation of a fragment that is selectively protected in the presence of Ca2+ was monitored by staining SDS-PAGE gels with Coomassie blue, and the band of interest was quantified by densitometry. Pixel intensities were plotted as a function of free [Ca2+].
Norepinephrine Release Assays
Two Forms of Synaptotagmin I with Distinct Oligomerization Properties
The discovery that Ca2+ triggers the oligomerization of synaptotagmin raises the question of whether and how this interaction functions to regulate exo- and/or endocytosis. Efforts to address this issue were hampered by the inability to generate recombinant rat synaptotagmin I that possessed Ca2+-triggered clustering activity. In previous studies, the recombinant protein was able to serve as an acceptor for the Ca2+-triggered binding of native synaptotagmin I from brain detergent extracts, but was not able to homo-oligomerize in response to Ca2+ (Chapman et al. 1996; Fig. 1 B). Thus, we could not assay for direct C2B–C2B interactions. We then explored the possibility that this loss of function was determined by the primary sequence of the protein. This possibility was prompted by the independent cloning of a synaptotagmin I cDNA (Osborne et al. 1999). The cytoplasmic domain of this clone was reported to differ from the original clone (Perin et al. 1990). We have resequenced these two cDNAs, and found that they vary by a single amino acid at position 374 (regarding the other two reported differences: one was not detected in our sequence analysis and one corresponds to a sequence error in the original report).
For simplicity, we refer to the original cDNA as Ia (Perin et al. 1990) and the second version as Ib (Osborne et al. 1999). Synaptotagmin Ib harbors a glycine at position 374 (G374), which is precisely conserved among all synaptotagmins; the lone exception is synaptotagmin Ia, which harbors an aspartate at this position (D374; Fig. 1 A). To determine whether the presence of D374 underlies the loss of Ca2+-triggered self-association activity of synaptotagmin Ia, we incubated soluble cytoplasmic domains of synaptotagmin Ia or Ib with their counterparts that had been immobilized as GST fusion proteins on glutathione-Sepharose beads. As shown in Fig. 1 B, synaptotagmin Ia failed to oligomerize in response to Ca2+; only trace levels of binding were observed in EGTA and Ca2+. In marked contrast, soluble synaptotagmin Ib exhibited a pronounced binding to immobilized synaptotagmin Ib in response to Ca2+. This effect is selective for Ca2+: Mg2+ failed to trigger binding, and Ba2+ and Sr2+ were less potent than Ca2+ in driving clustering (Fig. 1 C). These data demonstrate that the inability of synaptotagmin Ia to cluster in response to Ca2+ is due to the presence of an aspartate residue at position 374. The precise conservation of the glycine residue at this position, with the lone exception of synaptotagmin Ia, suggests that multiple synaptotagmins may exhibit clustering activity, as borne out by data described below.
We have confirmed the expression of both the G- and D374 forms by sequencing synaptotagmin cDNA from a variety of cDNA libraries (data not shown). At present, it is unclear how the D374 form of synaptotagmin arises; the genomic sequence of the rat clone has not been reported. Future studies are required to determine whether variation occurs at position 374 via RNA editing or alternative splicing, and whether seizure activity or other factors can influence the relative levels of the Ia and Ib messages.
We reiterate that while synaptotagmin Ia cannot oligomerize with itself, it can serve as an acceptor for the clustering activity of native synaptotagmin (Chapman et al. 1996; Sugita et al. 1996). Thus, clustering appears to involve at least two separable determinants or domains, and synaptotagmin Ia is only defective in one of these domains.
Mechanism of Ca2+-triggered Synaptotagmin Oligomerization
The finding that synaptotagmin Ib exhibits robust Ca2+-triggered clustering activity made it possible to examine the structural features that drive oligomerization. Our first goal was to determine whether the C2B domain was necessary and sufficient to drive clustering, as inferred from earlier studies (Chapman et al. 1996; Sugita et al. 1996). To address this issue, we carried out oligomerization assays using the isolated C2 domains of synaptotagmin. Soluble C2A did not bind to immobilized C2A or C2B, and soluble C2B did not bind to immobilized C2A (Fig. 2). However, in the presence of Ca2+, soluble C2B bound to immobilized C2B (Fig. 2). These data demonstrate that C2B is necessary and sufficient to drive synaptotagmin oligomerization. Furthermore, C2B functions as a Ca2+-sensing module.
The putative Ca2+ ligands within the C2B domain of synaptotagmin are indicated in the model shown in Fig. 3 A. To assess the function of these putative ligands, they were neutralized by substitution with asparagine residues either individually or in combination, and the resulting mutant proteins were assayed for oligomerization activity as described in Fig. 1 B. As a negative control, the K326,327A mutant was assayed and failed to oligomerize (Fig. 3B and Fig. C), which is consistent with our previous report (Chapman et al. 1998). As an additional control, the D230,232N mutant, which abolishes Ca2+-triggered interactions of synaptotagmin with anionic phospholipids and the SNARE complex (Bai et al. 2000), failed to inhibit the Ca2+-triggered clustering activity (Fig. 3B and Fig. C). These data are consistent with Fig. 2, demonstrating that the C2B domain is both necessary and sufficient for Ca2+-triggered oligomerization. Surprisingly, individual substitutions of the putative Ca2+ ligands within C2B had little effect on clustering activity. Rather, a slight trend toward constitutive self-association was observed among the point mutants (Fig. 3 C). Moreover, simultaneous neutralization of two (D363,365N) or four (D303,309,363,365N) of the five acidic Ca2+ ligands clearly resulted in Ca2+-independent constitutive clustering activity (Fig. 3B and Fig. C). These data provide strong evidence that the conserved putative Ca2+ ligands are functional and regulate synaptotagmin clustering. The conversion from regulated to constitutive clustering activity is addressed in the Discussion.
The inability of single point mutations to significantly disrupt Ca2+-triggered clustering activity suggests that C2B binds multiple Ca2+ ions in a redundant manner. Loss of individual sites are tolerated since other sites remain intact. This is in striking contrast to the C2A domain of synaptotagmin, in which each of the Ca2+ ligands are essential for binding to anionic phospholipids (Zhang et al. 1998; Bai et al. 2000). The interaction of synaptotagmin with SNAREs exhibits an intermediate dependence on Ca2+ ligands within C2A: some mutations are tolerated, whereas other mutations have strong disruptive effects (Chapman et al. 1995; Bai et al. 2000). There appear to be multiple modes by which Ca2+ binding can regulate the interactions of the C2 domains of synaptotagmin with effector molecules.
Ca2+ Drives a Conformational Change in the C2B Domain of Synaptotagmin
The experiments described above provide strong evidence that the C2B domain of synaptotagmin binds Ca2+, which drives clustering of the protein. We next sought to determine whether clustering is achieved via a Ca2+-induced conformational change in C2B. Ca2+-triggered polymerization of synaptotagmin (secondary to oligomerization) made a number of biophysical methods (NMR, CD, fluorescence measurements, etc.) inapplicable. To circumvent this problem, we applied a limited proteolysis approach to detect Ca2+-driven conformational changes in the domain. For these experiments, the C2B domain of synaptotagmin Ib was immobilized as a GST fusion protein. In addition, the quadruple Ca2+ ligand mutant version of C2B, D303,309,363,365N, as well as the K326,327A mutant that fails to oligomerize (Fig. 3 B; Chapman et al. 1998), were analyzed in parallel. Fusion proteins were incubated with increasing concentrations of trypsin or chymotrypsin in the presence of EGTA or Ca2+, and the proteolysis patterns were analyzed by SDS-PAGE. As shown in Fig. 4 A, the degradation patterns of wild-type C2B in EGTA versus Ca2+ were distinct, suggesting that C2B undergoes a conformational change upon binding Ca2+. With both proteases, a protease resistant fragment accumulated in the presence of Ca2+, suggesting that Ca2+ binds to and stabilizes the domain, which is analogous to the limited proteolysis data using C2A as a substrate (Davletov and Südhof 1994). This differential proteolysis pattern was not observed using the Ca2+ ligand mutant C2B domain as a substrate, providing further evidence that these acidic residues (D303,309,363,365N) constitute functional Ca2+ ligands (Fig. 4 A). Interestingly, the K326,327A mutant that fails to oligomerize is still capable of sensing Ca2+ (Fig. 4 A, bottom). Because C2 domains bind Ca2+ via a bipartite motif in which distal regions of the domain must come together via folding into a compact tertiary structure, this finding strongly indicated that the K326,327A mutant is correctly folded. The loss of oligomerization activity of this mutant (Chapman et al. 1998; Fig. 3 B) is not due to a loss of Ca2+ sensitivity, but rather is likely due to perturbation of its oligomerization interface.
We exploited the Ca2+-dependent accumulation of a protease-protected fragment to determine the Ca2+ dependence of this conformational change. Limited proteolysis was carried out at a fixed protease concentration, and the [Ca2+]free was varied. The appearance of the protected fragment was quantified by densitometry and is plotted in Fig. 4 B. The [Ca2+]1/2 was ∼62 ± 20 μM, which is consistent with the Ca2+ dependence for exocytosis (Heidelberger et al. 1994). This value is similar to the [Ca2+]1/2 for the interaction of the C2A domain of synaptotagmin I with anionic phospholipids (Davis et al. 1999). We note that previous measurements of the Ca2+ dependence for synaptotagmin clustering are highly dependent on the methods and conditions in which oligomerization was measured (Chapman et al. 1996; Sugita et al. 1996; Osborne et al. 1999). While the limited proteolysis technique employed here may only report a subset of conformational changes, this assay system has the advantage that it circumvents the need to carry out the wash steps employed in C2B-synaptotagmin pull-down assays and, therefore, accurately reports the Ca2+ requirement for a conformational change within C2B.
We also compared the proteolysis patterns of the C2B domains of synaptotagmins V–VIII. As shown in Fig. 4 C, Ca2+ drove a conformational change in synaptotagmins V and VI, suggesting that the C2B domains of multiple synaptotagmin isoforms can sense Ca2+. The proteolysis patterns of the C2B domains of synaptotagmins VII and VIII were not influenced by Ca2+, indicating that these domains either do not sense Ca2+, or Ca2+-driven conformational changes within these particular domains cannot be detected using this method. We note that the C2B domain of synaptotagmin VIII lacks three of the five acidic Ca2+ ligands (Fig. 1 A) and would not be expected to bind Ca2+ (Fig. 4 C). In contrast, synaptotagmin VII possesses each of the conserved Ca2+ ligands. Further studies, using a different approach, are required to determine whether the C2B domain of synaptotagmin VII can sense Ca2+ and change conformation.
Finally, as a further control to determine the effects of the D303,309,363,365N and K326,327A mutations on the structure of synaptotagmin, we assayed for the interaction of these mutants with effectors that bind to the C2B domain. As shown in Fig. 4 D, the D303,309,363,365N mutant exhibited undiminished ability to bind native synaptotagmin in response to Ca2+. This result is not surprising since the oligomerization activity of native synaptotagmin is off in the absence of Ca2+ such that the native protein cannot bind the constitutively activated immobilized protein. Upon binding Ca2+, the native protein is activated and binds to the oligomerization competent (albeit constitutively) immobilized mutant protein. We observed that AP-2 and synprint bound to the quadruple Ca2+ ligand mutant to the same extent as the wild-type protein, indicating that the mutant protein is correctly folded. In contrast to a previous report (Schiavo et al. 1995), we did not detect binding of β-SNAP to any of our immobilized synaptotagmin constructs. Consistent with our previous report, the K326,327A exhibited markedly reduced binding to native synaptotagmin, AP-2, and synprint. However, this domain appears to be correctly folded, since, as described above, this mutant still functions as a Ca2+ sensor. Furthermore, this mutant is capable of efficiently coimmunoprecipitating with t-SNAREs and SNARE complexes (data not shown), an interaction that requires the C2B domain of the protein (Chapman et al. 1996; Davis et al. 1999). Finally the circular dichroism spectra of the wild-type and K326,327A mutant are identical (data not shown). Thus, both mutants appear to be correctly folded.
Ca2+-triggered Synaptotagmin Oligomerization Is a Conserved Property among Different Synaptotagmin Isoforms
A key issue raised by the data presented above is whether the clustering activity of synaptotagmin Ib functions in excitation-secretion coupling. First, we addressed the question of whether synaptotagmin Ia or Ib is typical of the synaptotagmin family of proteins by examining the homo- and hetero-oligomerization activity of synaptotagmins I–VIII using the approach described in Fig. 1 B. Ca2+-triggered homo- and hetero-oligomerization is a highly conserved function of nearly all synaptotagmins (Fig. 5). Notable exceptions appear to be synaptotagmin VII and VIII (note, because of proteolysis problems that obscured the region of interest on the gels, some of the data concerning these isoforms was omitted from Fig. 5; these data are summarized here). Synaptotagmin VIII lacks three of the five acidic Ca2+ ligands (Fig. 1 A) and cannot homo-oligomerize in response to Ca2+. However, immobilized synaptotagmin VIII can clearly serve as an acceptor for binding other synaptotagmin isoforms, as indicated by the robust binding of soluble synaptotagmin III to immobilized synaptotagmin VIII (Fig. 5). Again, these data indicate that clustering involves the interaction of two distinct sites on C2B.
The other exception, synaptotagmin VII, efficiently bound to each of the immobilized synaptotagmin isoforms tested. However, these interactions are not promoted by Ca2+; in some cases, Ca2+ triggered slight decreases in the level of binding (Fig. 5). As discussed above, the basis for this unusual behavior is not readily apparent from the primary structure of the protein. Like synaptotagmin VIII, immobilized synaptotagmin VII can serve as an acceptor for the Ca2+-dependent binding of soluble synaptotagmin isoforms. In summary, all synaptotagmin isoforms tested exhibit some form of homo- and/or hetero-oligomerization activity. The conservation of this biochemical function suggested that it may be tenable to carry out dominant negative experiments to investigate whether, and at what stage, synaptotagmin oligomerization functions in regulated exocytosis.
Inhibition of Synaptotagmin Clustering Blocks Ca2+-triggered Exocytosis
A variant PC12 cell line lacking synaptotagmin I exhibits normal-to-elevated secretory responses (Shoji-Kasai et al. 1992). However, this cell line expresses numerous other synaptotagmin isoforms including synaptotagmin III, IV, and V (Mizuta et al. 1994; Hudson and Birnbaum 1995; Vician et al. 1995). Our findings described above, demonstrating that recombinant synaptotagmin I can homo- and hetero-oligomerize with synaptotagmins I, III, IV, and V, suggested that it may be possible to apply a dominant negative approach, using mechanically permeabilized PC12 cells, to study the function of synaptotagmin clustering in Ca2+-triggered exocytosis.
Secretion from permeable PC12 cells involves two stages: an ATP- and cytosol-dependent priming stage, followed by a Ca2+- and cytosol-dependent triggering reaction (Hay and Martin 1992). We reasoned that recombinant fragments of synaptotagmin would, in the presence of Ca2+, oligomerize with native synaptotagmin in the permeable cell preparation, and thereby block native synaptotagmin oligomerization. If the Ca2+-triggered clustering of native copies of synaptotagmin functions in secretion, recombinant synaptotagmin should block exocytosis. As shown in Fig. 6, the cytoplasmic domain of synaptotagmin Ib (referred to as C2AB in the figure) potently inhibited exocytosis. To determine whether inhibition was mediated by the Ca2+-triggered binding of recombinant C2AB to native synaptotagmin, we tested the ability of the K326,327A mutant, which only weakly clusters in response to Ca2+ (Fig. 3 B; Chapman et al. 1998; Davis et al. 1999), to inhibit exocytosis. As shown in Fig. 6, this mutant is not an effective inhibitor of secretion. We note that the K326,327A mutant is able to bind to SNAREs (Chapman et al. 1998), and this interaction requires both C2A and C2B (Chapman et al. 1995; Davis et al. 1999). These data indicate that C2B is not simply misfolded. The K326,327A mutant is also able to bind anionic phospholipids (Bai et al. 2000), yet fails to inhibit secretion. The results of these experiments are consistent with the hypothesis that C2AB blocks exocytosis via its C2B-mediated clustering activity.
To further test this model, we compared the ability of the isolated C2 domains of synaptotagmin to inhibit release. As shown in Fig. 6, C2B inhibited release in a manner analogous to C2AB. In contrast, C2A failed to inhibit secretion. Furthermore, we generated a mutant C2B domain harboring the K326,327A mutation, and this version of C2B failed to cluster (data not shown) and failed to inhibit exocytosis (Fig. 6). Finally, we examined the ability of a mutant C2AB domain, which harbored two mutations, D230,232N, within the C2A domain that abolish Ca2+-triggered C2A–effector interactions (Davis et al. 1999; Bai et al. 2000). This double mutant retained all of the inhibitory activity of wild-type C2AB (Fig. 6), despite the fact that it fails to bind either lipids or SNAREs in response to Ca2+. Thus, all of the synaptotagmin fragments that exhibit Ca2+-triggered clustering activity are able to inhibit exocytosis in this assay system, and all fragments that fail to cluster in response to Ca2+ fail to inhibit secretion despite the fact that other effector interactions remain intact. We also tested the C2A and C2B domains of rabphilin for inhibitory activity. Neither of the isolated rabphilin C2 domains bind to synaptotagmin (data not shown), and neither fragment inhibited secretion (Fig. 6), further establishing the specificity of the synaptotagmin C2B domain effect. We have tested additional C2 domains, from other proteins, and observed that some are inhibitory in our secretion assay; these C2 domains can also cluster with synaptotagmin, which is consistent with a model in which perturbation of native synaptotagmin clustering inhibits exocytosis (data not shown). Clustering may be a shared biochemical property among C2 domains from distinct proteins.
To further explore the dominant negative effect of C2AB, we determined whether inhibition occurs during the ATP-dependent priming stage or the Ca2+-triggered stage of secretion. If the inhibition of exocytosis is mediated via Ca2+-triggered clustering of recombinant synaptotagmin with the endogenous native protein, the inhibitory effect should be restricted to the Ca2+-triggered step in secretion. Indeed, this was the case; C2AB added during priming, and then washed out before triggering, has only a limited effect on release (Fig. 7 A). In contrast, C2AB that was added during the Ca2+-triggered stage resulted in the potent inhibition of secretion, presumably because Ca2+ triggered the oligomerization of recombinant with endogenous synaptotagmin (Fig. 7 A). Again, C2A served as a negative control.
We next sought to address the function of synaptotagmin oligomerization in exocytosis using an independent approach. This was achieved by analyzing the effects of the synprint peptide on secretion. Synprint corresponds to the II-III loop of N-type Ca2+ channels and binds to t-SNAREs and to synaptotagmin (Kim and Catterall 1997; Sheng et al. 1997). A critical point is that synprint binds to synaptotagmin in a Ca2+-independent manner and can block Ca2+-triggered synaptotagmin-clustering (Chapman et al. 1998). Thus, if clustering is critical for exocytosis, synprint should serve as an effective inhibitor akin to C2AB and C2B described above. However, in the case of synprint, the inhibitory effect should be equally effective during the ATP-dependent priming step (in the absence of Ca2+) as in the Ca2+-triggered stage of exocytosis, again because synprint can bind synaptotagmin equally well in the presence or absence of Ca2+. Congruent with this model, synprint inhibited exocytosis with equal or greater activity during the Ca2+-independent priming stage of secretion (Fig. 7 B). These data provide independent confirmation that disruption of synaptotagmin oligomerization can disrupt Ca2+-triggered exocytosis.
We note that a caveat of our inhibition studies is the ability of recombinant synaptotagmin and synprint to interact with a number of distinct proteins. For example, a recent study indicated that synprint can inhibit the binding of synaptotagmin and cysteine string protein to syntaxin (Wu et al. 1999). However, synprint binds much more tightly to syntaxin in the presence, rather than the absence of Ca2+ (Kim and Catterall 1997; Littleton et al. 1998; and data not shown), yet in our secretion assays, synprint does not preferentially inhibit release during the triggering phase when Ca2+ is present. Therefore, we view it as unlikely that synprint exerts its inhibitory effects via binding to syntaxin.
In this study, we have demonstrated that the C2B domain of synaptotagmin functions as a Ca2+-sensing module. Ca2+ drives a conformational change in the C2B domain of synaptotagmin I with a [Ca2+]1/2 of 62 ± 20 μM, similar to the Ca2+ sensitivity of the C2A domain of this isoform (Davis et al. 1999) and consistent with the Ca2+ dependence of exocytosis (Heidelberger et al. 1994). Furthermore, we directly demonstrate that Ca2+ triggers the oligomerization of the C2B domain of synaptotagmin Ib, and this property is conserved among multiple isoforms of the protein. We propose that Ca2+ acts by driving a conformational change that stabilizes the multimeric form of synaptotagmin. The C2A domain exhibits a similar conformational change in limited proteolysis experiments (Davletov and Südhof 1994). These changes may reflect the movement of charged residues within or near the Ca2+-binding loops (Ubach et al. 1998b; Chae et al. 1998), as indicated by structural studies of the C2 domain of phospholipase C-δ1 (Grobler et al. 1996). Given the diffusion-limited kinetics of the C2 domains of synaptotagmin, these conformational changes are unlikely to involve large-scale structural rearrangements (Davis et al. 1999). Numerous C2 domains harbor a conserved set of five acidic residues that function as Ca2+ ligands. Via mutagenesis, we sought to determine whether these residues indeed function as Ca2+ ligands within C2B. We observed that the simultaneous substitution of four of the five putative acidic Ca2+ ligands within C2B inhibited Ca2+-driven conformational changes and converted the domain to a constitutively oligomerizing form. The basis for this conversion is unclear. However, structural studies of the C2 domain of phospholipase C-δ1 indicate that these Ca2+ ligands participate in intramolecular ionic bonds (Grobler et al. 1996). Binding of Ca2+ to these acidic ligands serves to release positively charged side chains, resulting in the opening of the Ca2+ binding jaws or loops such that they can interact with effectors. In this model, neutralization of the acidic Ca2+ ligands may result in a constitutively activated domain. Clearly, however, this is not the case with C2A. In C2A, neutralization of any of the acidic Ca2+ ligands abolishes phospholipid binding activity (Zhang et al. 1998; Bai et al. 2000). Additional structural studies are required to understand the differential effects of Ca2+ ligand mutations on C2A– and C2B–effector interactions. Nonetheless, the effects of the putative Ca2+ ligand mutations in C2B, on both Ca2+-driven conformational changes and clustering activity, demonstrate that these conserved Ca2+ ligands are functional.
To assess the generality of Ca2+-triggered clustering activity, we assayed homo- and hetero-oligomerization of synaptotagmins I–VIII and observed that oligomerization is a highly conserved biochemical property. These findings made it possible to employ a dominant negative approach to study the function of synaptotagmin clustering in permeable PC12 cells, which harbor multiple synaptotagmin isoforms. In all cases, fragments of synaptotagmin that were capable of clustering, in response to Ca2+, served as potent inhibitors of secretion. This inhibitory activity mapped to the C2B domain of recombinant synaptotagmin and required the presence of Ca2+, presumably to drive oligomerization with native synaptotagmin within the permeable cells. The inhibitory activity did not involve the binding of Ca2+ to the C2A domain of the protein, demonstrating that Ca2+-triggered binding to SNAREs or anionic phospholipids is not required for the inhibition of secretion observed in our experiments. In contrast, mutations or deletions that block the clustering activity of the C2B domain abrogated the ability of recombinant C2B to block exocytosis.
Finally, the ability of synprint to inhibit exocytosis in our assay system, when applied at either the priming or triggering stages of secretion, is also consistent with a model in which Ca2+-triggered synaptotagmin clustering is critical for exocytosis. In this model, the inhibitory effects of synprint are independent of the stage of secretion because synprint can bind to synaptotagmin in a Ca2+-independent manner and block clustering. Thus, synprint should be able to inhibit secretion when used to treat cells during either the Ca2+-independent (i.e., ATP-dependent priming) or the Ca2+-triggered stage of secretion, as borne out by the data. Others have reported that synprint has inhibitory effects on synaptic transmission. It was suggested that inhibition was due to uncoupling syntaxin from Ca2+ channels (Mochida et al. 1996; Rettig et al. 1997). However, chronic application of synprint (Rettig et al. 1997) can disrupt synaptotagmin–AP-2 interactions (Zhang et al. 1994; Chapman et al. 1998), resulting in a loss in the total number of vesicles (Jorgensen et al. 1995). Use of permeable PC12 cells circumvents a number of these problems since Ca2+-triggered secretion does not depend upon Ca2+ channels nor upon endocytosis.
In an earlier report, a microinjection approach was used to assess the effects of synaptotagmin fragments on secretion from intact PC12 cells (Elferink et al. 1993). While C2AB and C2B did not appear to inhibit secretion, it was noted that these fragments had a tendency to aggregate and were difficult to quantitatively introduce into cells; from these studies, the potential ability of C2AB and C2B to inhibit secretion was not ruled out (Elferink et al. 1993). The use of permeable PC12 cells overcomes these difficulties and demonstrates that these fragments indeed inhibit release. Furthermore, recent experiments from L.A. Elferink and A.E. Hawley (personal communication), using viruses to infect PC12 cells, indicate that C2B is indeed critical for secretion from intact cells. Thus, the only apparent discrepancy between the initial PC12 cell microinjection data and the data present in our study is the inability of C2A to inhibit secretion in our assay system. One possible explanation is that C2A inhibits an upstream step in the vesicle trafficking pathway in intact cells (Elferink et al. 1993), which is not recapitulated in the permeable cells.
The function of C2B-mediated synaptotagmin clustering during exocytosis is not clear. We hypothesize that C2B-mediated oligomerization may drive the assembly of the fusion machinery into a collar or ringlike fusion pore or may alter the stability/conformation of a weakly preassembled pore. This hypothesis stems from the observation that purified SNARE-complexes do not efficiently cluster into multimeric complexes akin to viral fusion proteins (Lindau and Almers 1995; Blumenthal et al. 1996; Danieli et al. 1996; Fasshauer et al. 1997). Also, a protein designated EEA1 recently has been reported to cluster syntaxin-13 into oligomeric structures, and this oligomerization is required for endosomal fusion (McBride et al. 1999). Indeed, genetic studies indicate that synaptotagmin functions as a multimeric complex (Littleton et al. 1994). We speculate that the Ca2+-driven clustering activity of synaptotagmin may regulate the initial Ca2+-triggered opening (Katz 1969) or subsequent dilation (Fernandez-Chacon and Alvarez de Toledo 1995) of the exocytotic fusion pore (Lindau and Almers 1995).
The oligomerization of synaptotagmin may have other functions as well. In a previous report, we demonstrated that a seizure-induced isoform of synaptotagmin, IV, can form hetero-oligomers with synaptotagmin I. Upregulation of synaptotagmin IV reduces synaptic transmission, potentially by hetero-oligomerizing to form synaptotagmin clusters with hybrid properties (Littleton et al. 1999). Hetero-oligomerization of synaptotagmins with unique Ca2+-sensing properties may provide a means to modulate release probability and, thus, modulate synaptic function.
In summary, we have identified a key function for the C2B domain of synaptotagmin in excitation-secretion coupling: C2B udergoes Ca2+-driven conformational changes and assembles synaptotagmin into oligomers. In this light, we reiterate that a point mutation within this domain increases the Ca2+ requirement for secretion (Littleton et al. 1994). Future studies will determine whether this phenotype results from an impaired ability of C2B to function as a Ca2+-sensing oligomerization module.
We thank T. Südhof, G. Schiavo, M. Fukuda, J. Rizo, and W. Catterall for cDNA clones, R. Jahn and S. Enger for antibodies, R. Roy for assistance with experiments, and J. Bai for discussions.
This study was supported by grants from the National Institutes of Health (GM 56827-01), American Heart Association (9750326N), the Milwaukee Foundation and NIH grants DK40428 and DK25861 (to T.F.J. Martin). E.R. Chapman is a Pew Scholar in the Biomedical Sciences, J.T. Littleton is a Merck fellow of the Helen Hay Whitney Foundation.
J.T. Littleton's current address is Center for Learning and Memory, Massachusetts Institute of Technology, Cambridge, MA 02139.
Abbreviations used in this paper: SNARE, N-ethyl maleimide-sensitive factor attachment protein receptor; t-SNARE, target SNARE.