Delineation of the Oligomerization, AP-2 Binding, and Synprint Binding Region of the C2B Domain of Synaptotagmin*

Biochemical and genetic studies indicate that synaptotagmin I functions as a Ca 2 1 sensor during synaptic vesicle exocytosis and as a membrane receptor for the clathrin adaptor complex, AP-2, during endocytosis. These functions involve the interaction of two conserved domains, C2A and C2B, with effector proteins. The C2B domain mediates Ca 2 1 -triggered synaptotagmin oligomerization, binds AP-2 and is important for the interaction of synaptotagmin with Ca 2 1 channels. Here, we report that these are conserved biochemical properties: Ca 2 1 promoted the hetero-oligomerization of synaptotagmin I with synaptotagmins III and IV, and all three synaptotagmin isoforms bound the synprint region of the a 1B subunit of N-type Ca 2 1 channels. Using chimeric and truncated C2 domains, we defined a common region of C2B that mediates oligomerization and AP-2 binding. Within this region, two adjacent lysine residues were identified that were critical for synaptotagmin oligomerization, AP-2, and synprint binding. Competition experiments demonstrated that the synprint fragment was an effective inhibitor of synaptotagmin oligomerization and also blocked binding of synaptotagmin to AP-2. In a model for the structure of C2B, the common effector binding site localized to a putative Ca 2 1 -binding loop and a concave region formed by two b -strands. These studies provide the first structural in-formation regarding C2B target protein recognition and provide the means to selectively disrupt synaptotagmin-effector interactions for functional studies.

protein, synaptotagmin I (12)(13)(14), is essential for rapid and efficient Ca 2ϩ -triggered release of neurotransmitters. These findings are consistent with the proposed function of synaptotagmin as a Ca 2ϩ sensor for regulated exocytosis (13). In addition, biochemical studies demonstrated that synaptotagmin is a high affinity receptor for the clathrin adaptor protein complex, AP-2 (15). Genetic studies in Caenorhabditis elegans indicate that loss of synaptotagmin also results in defective synaptic vesicle recycling (16). Thus, synaptotagmin is thought to function in both the exo-and endocytotic limbs of the synaptic vesicle cycle.
Synaptotagmin is represented by a large gene family that currently contains 12 members (17)(18)(19). Each member spans the vesicle membrane once and has a short C-terminal intravesicular domain and a large cytoplasmic region that contains two C2 domains, designated C2A and C2B (14). C2A is the membrane proximal domain and mediates the Ca 2ϩ -dependent interaction of synaptotagmin with anionic phospholipids (20,21). The membrane distal C2B domain mediates Ca 2ϩ -dependent oligomerization of synaptotagmin, potentially clustering the release machinery into a collar or ring-like structure (22)(23)(24). The C2B domain has also been reported to mediate Ca 2ϩindependent interactions. These include the above described interaction with AP-2, as well as ␤-SNAP (25), inositol polyphosphates (26), and the II-III cytoplasmic loop or "synprint" region of N-and P/Q-type Ca 2ϩ channels (27,28). The direct interaction of synaptotagmin with Ca 2ϩ channels may contribute to the speed of exocytosis, which occurs with a lag time of 60 -200 s (29,30). Finally, both C2 domains are required for high affinity binding to the pre-synaptic proteins syntaxin 1A (22,31) and SNAP-25, 1 although some binding is preserved in the isolated C2A domain 1 (32,46). These latter interactions are promoted by Ca 2ϩ concentrations 1 (31) similar to those required for neuronal exocytosis (34). Syntaxin and SNAP-25 form a complex with the synaptic vesicle protein synaptobrevin that is capable of catalyzing membrane fusion in vitro (35) and in vivo (36). The mechanism by which synaptotagmin may regulate the fusion complex is not yet known.
To elucidate the mechanism by which synaptotagmin functions in exo-and endocytosis, it is essential to determine how synaptotagmin engages effector molecules. These studies will make it possible to address the function of specific interactions using site-directed inhibitory peptides or by genetic analysis based on loss-of-function mutations. Details concerning the molecular mechanisms by which C2A interacts with effector molecules have recently begun to emerge (37,38,53). However, very little is known concerning the molecular basis for the wide range of C2B target-protein interactions. Here, we have begun to address the structural basis for how the C2B domain of synaptotagmin recognizes and binds different effector proteins.
We demonstrate that distinct isoforms of synaptotagmin form hetero-oligomers and bind the synprint region of N-type Ca 2ϩ channels. Using chimeric and truncated C2 domains, we identified a common region of C2B that mediates oligomerization and AP-2 binding. Furthermore, we report point mutations that have graded effects on the ability of the C2B domain to engage effector proteins. Finally, we demonstrate that synprint can competitively inhibit the oligomerization of synaptotagmin and the binding of AP-2. These findings provide the means to selectively disrupt synaptotagmin-effector interactions for functional studies.
Recombinant DNA and Proteins-Vectors encoding GST 2 fusion proteins to express wild type, mutant, chimeric and truncated regions of synaptotagmin I, and the cytoplasmic domains of synaptotagmins III and IV were prepared by polymerase chain reaction and subcloned into pGEX-2T (Amersham Pharmacia Biotech) as described (31). Mutations and chimeras were prepared using the overlapping primer method as described (21) and were confirmed by DNA sequencing. All GST fusion proteins were purified using glutathione-Sepharose (Amersham Pharmacia Biotech) chromatography as described (22). Proteins were quantified by SDS-PAGE and staining with Coomassie Blue using bovine serum albumin as a standard. Full-length rat syntaxin 1A and ␣1Bsynprint were expressed using pTrcHis vectors (In Vitrogen) as described (22) resulting in fusion proteins that contain his6 and T7 tags at their N termini.
Binding Assays-Rat brain detergent extracts (1 mg/ml) were prepared as described previously (31) in HEPES-buffered saline (HBS; 50 mM HEPES, pH 7.4, 100 mM NaCl with 1% Triton X-100. 1-ml aliquots were incubated with immobilized GST fusion proteins as described in the figure legends. For synprint competition experiments, 4.2 M purified synprint was included in the rat brain detergent extract binding assays. Bound proteins were detected by immunoblot analysis as described (31,22) using enhanced chemiluminescence. We observed that purified recombinant proteins containing the C2B domain of synaptotagmin lost native synaptotagmin I binding activity after 3 days. Therefore, all binding experiments were carried out using purified recombinant proteins that were 2 days old.

RESULTS
The C2B domain of synaptotagmin I mediates Ca 2ϩ -triggered homo-oligomerization (22). In the first series of experiments, we determined whether oligomerization is a conserved property among distinct isoforms of synaptotagmin. To address this question, the cytoplasmic domains of synaptotagmins I, III, and IV were immobilized as GST fusion proteins and incubated with rat brain detergent extracts in the presence of Ca 2ϩ or EGTA. The binding of native synaptotagmin I to the immobilized fusion proteins was monitored by immunoblotting with an anti-synaptotagmin I-specific luminal domain monoclonal antibody. As shown in Fig. 1, native synaptotagmin I efficiently bound to recombinant synaptotagmin I, III, and IV in a Ca 2ϩdependent manner. These findings demonstrate that Ca 2ϩ not only triggers the self-association of synaptotagmin I but also promotes hetero-oligomerization of distinct synaptotagmin isoforms. These interactions are specific; immunoblotting with an anti-synaptophysin antibody demonstrated that other abundant synaptic vesicle antigens did not associate with the immobilized synaptotagmins. Consistent with previous results, oligomerization was mediated solely by the C2B domain of synaptotagmin I (22). As an additional control, we compared the anionic lipid binding properties of the immobilized synaptotagmin isoforms (Fig. 1). In agreement with previous reports (39), synaptotagmin I and III but not IV bound liposomes composed of phosphatidylserine/phosphatidylcholine in a Ca 2ϩdependent manner. Thus, whereas synaptotagmin IV cannot bind liposomes in a Ca 2ϩ -dependent manner via its C2A domain, this isoform is able to engage in Ca 2ϩ -dependent hetero-oligomerization.
We next tested the ability of recombinant synprint, which comprises the II-III loop of the ␣1B subunit of N-type Ca 2ϩ channels (40), to bind synaptotagmins I, III, and IV. Purified synprint was incubated with immobilized synaptotagmins I, III, and IV, and binding was detected by immunoblot analysis. Consistent with previous reports, synprint bound to the cytoplasmic domain of synaptotagmin I in a Ca 2ϩ -independent manner (27). Synprint also efficiently bound to synaptotagmin III and IV in a Ca 2ϩ -independent manner, demonstrating that the interaction with Ca 2ϩ channels is a conserved property among distinct synaptotagmin isoforms. We next examined the interaction of synprint with the isolated C2 domains of synaptotagmin I. As reported previously, in the absence of Ca 2ϩ , synprint bound to C2B but not to C2A (27). However, addition of 1 mM free Ca 2ϩ resulted in dramatic binding of synprint to C2A and modestly promoted synprint binding to C2B (Fig. 1). This effect was observed in four independent experiments. These findings demonstrate that the interaction of synaptotagmin with the synprint region from the ␣ 1B subunit of N-type Ca 2ϩ channels, unlike oligomerization (22) or binding to AP-2 (15), is not perfectly preserved within an isolated C2 domain. For this reason, ␣1B-synprint binding will not be included in the domain mapping studies described below. The C2B domain of synaptotagmin has also been reported to mediate synaptotagmin binding to SV2 (41) and ␤-SNAP (25). We found that the C2B-SV2 interaction is disrupted by physiological concentrations of Mg 2ϩ (1 mM), 3 and although we observe an interaction between GST-synaptotagmin and His 6 -␤-SNAP, we cannot detect binding between the native proteins under our experimental conditions. 4 Therefore, the analysis of the C2Beffector recognition site, described below, is focused on the self-association of synaptotagmin I and the binding of synaptotagmin to AP-2.
The major goal of this study was to delineate the region(s) of the C2B domain of synaptotagmin I (hereafter referred to as synaptotagmin) that interacts with effector proteins. To address this issue, we first opted to map effector binding sites using chimeric C2 domains composed of C2A and C2B. The rationale for this approach was based on two observations. First, C2A does not mediate oligomerization and also fails to bind AP-2. Second, C2 domains are eight-stranded ␤-sandwich structures in which the strands are connected in a complex manner (Refs. 42-44 and see Fig. 7). Thus, linear truncations would be expected to have drastic effects on the overall confor-mation of the domain. The most notable structural difference among C2 domains studied thus far lies in the two different connectivity patterns for a ␤-strand that results in two distinct topologies. The C2B domain of synaptotagmin is predicted to conform to the same type I topology as the C2A domain (45). Thus, it should be possible to generate correctly folded chimeras.
To determine the minimal region of C2B that confers oligomerization and AP-2 binding we "swapped" regions of C2B and C2A as shown in Fig. 2. To further ensure that the chimeras were properly folded, chimeric junctions were selected that corresponded to perfectly conserved regions within the ␤-strands. Chimeras were immobilized as GST fusion proteins and incubated with rat brain detergent extracts in the presence of EGTA or Ca 2ϩ . Binding of native synaptotagmin and AP-2 was assayed by immunoblot analysis. As shown in Fig. 2, Ca 2ϩ -dependent binding of synaptotagmin and Ca 2ϩ -independent binding of AP-2 were mediated by the C2B domain of synaptotagmin; C2A failed to bind in both the presence and absence of Ca 2ϩ . Binding of both synaptotagmin and AP-2 was preserved in a chimera (C2B/A) in which the N-terminal half of C2B was fused to the C-terminal half of C2A. Neither synaptotagmin nor AP-2 bound to the reciprocal chimeric C2 domain (C2A/B) that contained the N-terminal half of C2A and the C-terminal half of C2B. To further localize the region of C2B that confers effector binding, we subdivided the N-terminal half of C2B and generated chimeras in which either the first (C2B/A/A) or second quarter (C2A/B/A) of the chimeric se-quence corresponded to C2B, with the balance of the C2 domain composed of C2A sequence. Synaptotagmin and AP-2 bound the chimera that contained the second quarter of C2B (C2A/ B/A) but did not bind the chimera that contained the first quarter of C2B (C2B/A/A). We further divided C2A/B/A into C2A/B/A/A and C2A/A/B/A and observed trace levels synaptotagmin and AP-2 binding to the C2A/A/B/A chimera and no binding to the C2A/B/A/A chimera. In summary, the replacement of residues 165-195 of C2A with homologous residues 296 -328 from C2B, resulted in a C2 domain (C2A/B/A) with C2B character, i.e. a chimera that interacted with the C2Bspecific targets AP-2 and native synaptotagmin. Interestingly, the C2A/B/A chimera also exhibited Ca 2ϩ -dependent phospholipid binding activity (Fig. 2). Thus, it is possible to engineer gain-of-function C2 domains. It should be noted that fluorescence (37) and NMR (53) studies indicate that loop 2 and Ca 2ϩ -binding loop 3 of C2A contain the major determinants for mediating C2 domain-lipid interactions. These loops are preserved in the C2A/B/A chimera (see Fig. 7) and may account for the dual ability of this chimera to engage both C2A-and C2Bspecific effectors.
We next sought to determine whether residues 296 -328 of C2B were sufficient for binding native synaptotagmin. A fragment comprised of amino acids 293-328 was fused to GST and used as an affinity matrix. As shown in Fig. 3, this isolated region did not bind native synaptotagmin in the presence or absence of Ca 2ϩ . Failure to bind could arise from a number of causes including an inability to adopt an active conformation, To assay for native synaptotagmin I binding, beads were incubated for 2 h with 1 ml (1 mg/ml) rat brain detergent extract in the presence of 2 mM EGTA (Ϫ) or 1 mM Ca 2ϩ (ϩ) in HBS. Beads were washed three times, and bound proteins were eluted by boiling in SDS sample buffer. Samples were resolved by SDS-PAGE, and bound native synaptotagmin was detected by immunoblotting with an anti-synaptotagmin luminal domain antibody (604.4). Immunoreactive bands were visualized using enhanced chemiluminescence. 10 g of the extract (total) and 15% of the bound material were loaded onto the gel. As a control, samples were also analyzed for the presence of synaptophysin. For synprint binding assays, 30 g of the GST-synaptotagmin isoforms and isolated C2 domains of synaptotagmin I were immobilized on 30 l of glutathione-Sepharose and incubated with 2.6 g of recombinant synprint in 200 l of HBS with 0.1% Triton X-100 in the presence of either 2 mM EGTA (Ϫ) or 1 mM Ca 2ϩ (ϩ). Samples were processed as described above, and bound synprint was detected using an anti-T7 tag monoclonal antibody (Novagen) and enhanced chemiluminescence. 3% of the total binding reaction (total) and 12% of the bound material were loaded onto the gel. Bottom panel, 15 g of the cytoplasmic domains of GSTsynaptotagmin I (stg-I), III (stg-III), and IV (stg-IV) were immobilized on 10 l of glutathione-Sepharose. Phospholipid binding assays were carried out as described (20,37)  the loss of flanking contact sites and/or steric hindrance due to GST. To better define the minimal requirements for C2B-effector interactions, we carried out a systematic truncation analysis of C2B and analyzed the truncation mutants for native synaptotagmin binding activity. C-terminal truncations, which removed essentially the C-terminal half of C2B, were tolerated; a fragment comprised of residues 248 -328 was able to mediate Ca 2ϩ -dependent oligomerization. These data demonstrate the Binding of native synaptotagmin I and AP-2 to the immobilized fusion proteins was assayed as described in the legend to Fig. 1. 4 g of detergent extract (total) and 11% of the bound material was subjected to SDS-PAGE and immunoblot analysis using an anti-synaptotagmin I luminal domain antibody. To assay for AP-2 binding, 3 g of the extract (total) and 30% of the bound material was subjected to immunoblot analysis using anti-␣-adaptin mouse monoclonal antibodies. Right panel, a chimeric C2 domain mediates both C2A-and C2B-specific effector interactions. GST, GST-C2A, GST-C2B, and the GST-C2A/B/A chimera, which bound to AP-2 and mediated Ca 2ϩ -dependent binding of native synaptotagmin I, were assayed for Ca 2ϩ -dependent phospholipid binding as described in the legend to Fig. 1.   FIG. 3. Truncation analysis of the synaptotagmin self-association domain. 10 -20 g of C2A, C2B, and truncated versions of C2B were immobilized on 60 l of glutathione-Sepharose. The sequences of the truncated C2B domains are represented schematically with the domain corresponding to residues 296 -328, as described in the legend to Fig. 2, indicated with dashed lines. Binding of native synaptotagmin I was assayed as described in the legend to Fig. 1. 8 g of the extract (total) and 18% of the bound material was subjected to SDS-PAGE and immunoblot analysis using the anti-synaptotagmin I luminal domain antibody (604.4) and enhanced chemiluminescence. C-terminal truncation from residues 328 to 326 (⌬KKT) or N-terminal truncation from residues 321 to 329 (⌬KRLKKKKT) abolished synaptotagmin self-association. This region is shown in the bottom panel where the critical residues, 326 -328, are underlined and in bold. Note that the alignment of C2A and C2B breaks down between ␤-strands 3 and 4, necessitating the introduction of a gap in the sequence of C2A. feasibility of using a truncation approach to map the C2Beffector binding domain. Analysis of additional truncation mutants demonstrated that further C-terminal truncation to amino acid 325 abolished binding. Thus, amino acids 326 -328 are essential for oligomerization. Significant N-terminal truncations were also tolerated because binding was preserved in a fragment encoding 321-421. Further N-terminal truncation to residue 329 resulted in the complete loss of binding. In summary, C-terminal truncation from 328 to 325 or N-terminal truncation from 321 to 329 results in the loss of binding activity. Thus, residues 321-328 comprise an important region for mediating oligomerization. Within this region, amino acids 326 -328 appear to be essential. These data are consistent with the chimera studies, described above, in which we delineated 296 -328 as the region that confers C2B-specific effector protein interactions. It is notable that residues 326 -328 (amino acids KKT) lie in the region of C2B which exhibits a breakdown in its alignment with C2A. This region is shown at the bottom of Fig.  3 and corresponds to the only segment in which the alignment requires a two-amino acid gap in the sequence of C2A. Thus, this region appears to have significantly diverged during evolution, further supporting our findings that this region accounts for the distinct effector interactions exhibited by C2B.
The domain encoded by residues 293-328 also failed to bind AP-2 (Fig. 4). We therefore used the truncation mutants that corresponded to the borders of the oligomerization domain to determine whether AP-2 bound to a similar site on C2B, as suggested by the chimeric C2 domain analysis described above. Consistent with those results, AP-2 bound to 248 -328 but failed to bind the 248 -325 truncation mutant. Binding was also observed using 321-421 of C2B. Like oligomerization, AP-2 binding activity was lost by further truncation of C2B to 329 -421. These data demonstrate that oligomerization and AP-2 binding are mediated by a common region of C2B.
The analysis described above demonstrated that truncation of the 248 -328 fragment to 248 -325 resulted in the loss of AP-2-and native synaptotagmin binding to C2B. To further explore the role of lysines 326 and 327 in binding synaptotagmin effectors, these amino acids were replaced individually, or in tandem, with alanine residues. As shown in Fig. 5, substi-tution of lysine 327 with an alanine reduced oligomerization, AP-2 binding, and synprint binding by approximately 50%. Substitution of both lysines 326 and 327 with alanines essentially abolished all three interactions. CD spectra of the wild type and lysine mutant synaptotagmin were identical. 4 In addition, the double lysine mutant retained partial ability to co-immunoprecipitate with syntaxin. 3 These findings indicate that the mutations did not result in disruption of the overall structure of recombinant synaptotagmin. The domain mapping experiments described above indicate that a common region of the C2B-domain of synaptotagmin mediates interactions with a number of distinct effector molecules. To confirm these findings, we tested whether synprint could inhibit the oligomerization of synaptotagmin or the interaction of synaptotagmin with AP-2. As shown in Fig. 6, 4.2 M synprint completely blocked both of these interactions. These experiments do not involve the use of chimeric, truncated, or mutant proteins and independently support our finding that the C2B domain of synaptotagmin interacts with distinct effectors via the same or closely overlapping subdomains. These findings are discussed in more detail below.
Finally, we sought to localize the C2B-effector binding domain within the three-dimensional structure of C2B. The structure of C2B is not known but is predicted to be analogous to the structure of C2A. By aligning C2A and C2B and using the crystal structure of C2A (42) as a template, we generated a model for the three-dimensional structure of C2B (Fig. 7). In this model of C2B, amino acid residues 296 -328, which impart C2B-specific effector interactions, are shaded. Lysines 326 and 327, which are essential for binding AP-2, synprint, and for oligomerization, are shown in black. For comparison, the known structure of C2A is also shown with the residues that form contacts with lipids (37, 53) rendered in black.

DISCUSSION
C2 domains are conserved motifs of approximately 130 amino acid residues that are found in over 60 different proteins (45). These proteins include lipid-modifying enzymes (lipases and kinases), protein kinases, GTPase-activating proteins, regulators of membrane traffic, proteins involved in ubiquitinmediated protein degradation, and a pore-forming protein secreted by cytolytic T-cells. Many C2 domains bind Ca 2ϩ ions and interact with other molecules. To date, little is known FIG. 4. Truncation analysis of the synaptotagmin-AP-2 binding domain. Binding assays were carried out as described in the legend to Fig. 3. 2 g of the extract (total) and 30 -40% of the bound material was subjected to SDS-PAGE and immunoblot analysis as described. AP-2 was detected using an anti-␣-adaptin mouse monoclonal antibody and enhanced chemiluminescence. 5. Lysines 326 and 327 are essential for synaptotagmin oligomerization, AP-2 binding, and synprint binding. In the cytoplasmic domain of synaptotagmin I, both lysine residues 326 and 327, as well as lysine 327 alone, were substituted with alanine residues. Wild type (wt) and mutant (K326,327A and K327A) cytoplasmic domains (35 g) were immobilized using 22 l of glutathione-Sepharose and assayed for binding of native synaptotagmin I and AP-2 as described in the legend to Fig. 2. 3 g of the detergent extract (total) and 30% of the bound material were subjected to SDS-PAGE and immunoblot analysis. Synprint binding assays were carried out as described in the legend to Fig. 1 using 2.6 g of synprint and 20 g of each synaptotagmin immobilized on 20 l of beads in a 150-l reaction. 1% of the total synprint added to the binding reaction (total) and 10% of the bound material were subjected to SDS-PAGE and immunoblot analysis. concerning the molecular basis by which C2 domains recognize, bind, and regulate effectors. Elucidating the molecular mechanism by which synaptotagmin functions in exo-and endocytosis necessitates a detailed analysis of the structure and function of its two C2 domains.
The crystal structures of the C2A domain of synaptotagmin (42), phospholipase C␦1 (43), and cytoplasmic phospholipase A2 (44) have been reported and are compact eight-stranded ␤-sandwich structures. Three flexible loops that protrude from one end of the domain contain Ca 2ϩ ligands that coordinate two to three metal ions. Details regarding the interaction of C2A with effectors have recently begun to emerge. A fluorescence study revealed that Ca 2ϩ -binding loop 3 of the C2A domain of synaptotagmin directly penetrates into lipid bilayers, potentially providing movement or force to trigger neuronal exocytosis (37). Subsequent NMR studies demonstrated that residues in loops 2 and 3 form contact sites for monodisperse anionic lipids (53). NMR studies have also provided evidence that residues within each of the three flexible loops of C2A and in particular Ca 2ϩ -binding loops 1 and 3 make contacts with syntaxin (38). However, less is known concerning the basis for synaptotagmin C2B-effector interactions. The only interaction to be mapped in detail is with inositol polyphosphates. Fukuda et al. (47) mapped the inositol 1,3,4,5-tetraphosphate-binding site to residues 315-346 of the C2B domain of synaptotagmin II. Within this region, lysine residues 327, 328, and 332 were critical for binding. As described in more detail below, these residues correspond to lysines 326, 327, and 331 in the sequence of synaptotagmin I.
In this study we sought to determine the structural basis for the interaction of the C2B domain of synaptotagmin with AP-2 and synprint and to map the region of C2B that mediates Ca 2ϩ -dependent self-association. Analysis of chimeric C2 domains revealed that residues 296 -328 contain critical elements for C2B-mediated synaptotagmin oligomerization and AP-2 binding. Consistent with these data, independent truncation experiments demonstrated that residues 325-328 were essential for C2B-effector interactions. Furthermore, substitution of lysine 327 with an alanine residue reduced synaptotagmin oligomerization, AP-2 binding, and synprint binding by approximately 50%, whereas substitution of both lysine residues 326 and 327 with alanines abolished all three interactions. As noted above, these residues precisely correspond to lysines 327 and 328 in the inositol 1,3,4,5-tetraphosphate-binding domain of synaptotagmin II. Thus, a common region of C2B is involved in mediating interactions with a number of different molecules. This observation is supported by our findings that synprint is an effective inhibitor of C2B-mediated synaptotagmin interactions. Micromolar concentrations of synprint completely abolished Ca 2ϩ -triggered oligomerization of synaptotagmin as well as the interaction of synaptotagmin with AP-2 (Fig. 6). Interestingly, two previous studies reported that the introduction of synprint into neurons resulted in the inhibition of synaptic transmission (54,55). This effect was assumed to be due to inhibition of N-type Ca 2ϩ channel-syntaxin interactions. From the data reported here, it is clear that synprint can have multiple inhibitory effects on the molecular machinery that mediates exo-and endocytosis. For example, the inhibitory effects of synprint on synaptic transmission may be due to blockade of Ca 2ϩ -triggered synaptotagmin oligomerization. It will be interesting to determine whether synprint peptides can affect endocytosis via inhibition of the synaptotagmin-AP-2 interaction.
Lysines 326 and 327 lie in a region of C2B that yields a breakdown in its alignment with C2A, necessitating the introduction of a gap of two residues in C2A. This observation suggests that this region diverged during evolution such that each C2 domain evolved to carry out independent and distinct Competition experiments were carried out to test the effect of synprint on synaptotagmin oligomerization and on the interaction of synaptotagmin with AP-2. Immobilized GST-synaptotagmin fusion protein was pre-incubated with 4.2 M synprint in HBS with 0.5% Triton X-100 for 1.5 h and then assayed for native synaptotagmin and AP-2 binding as described in the legend to Fig. 1. 30% of the bound material and 9 g of the detergent extract (total) were subjected to SDS-PAGE and immunoblot analysis using an anti-synaptotagmin luminal domain antibody (604.4). To assay for AP-2 binding, 40% of bound material and 2 g of the detergent extract (total) were subjected to SDS-PAGE and immunoblot analysis using an anti-␣-adaptin antibody.
FIG. 7. Molecular models depicting the regions of the C2 domains of synaptotagmin that mediate effector interactions. A, structure of C2A. B, model of C2B. The crystal structure of C2A was rendered as described (37) according to a description file modified from Sutton et al. (42). Note that loops 1 and 3 form the Ca 2ϩ -binding jaws of C2A (42). The residues of C2A that interact with lipids (amino acids 198, 205, 206, 233, and 234 in loops 2 and 3) are shown in shading (37,53). The structure of C2B is not known and was modeled with a Silicon Graphics Indigo computer and Look 2.0/SYBYL 6.3 software using the crystal structure of C2A, in conjunction with alignments of C2A and C2B, as a template. Residues 296 -328, which confer C2B-specific effector interactions (corresponding to the C2A/B/A chimera in Fig. 1) are shaded. Lysines 326 and 327, critical for oligomerization, AP-2 binding, and synprint binding, are rendered in black.
functions. It should be noted, however, that in some cases both C2 domains of synaptotagmin cooperate to interact with effectors. This principle is illustrated in Fig. 1 in which isolated C2 domains fail to mediate ␣1B-synprint binding in the same manor as the intact cytoplasmic domain of synaptotagmin. Similar results have been observed for the interaction of synaptotagmin with phosphatidylinositol polyphosphates (26), SNAP-25, 1 and syntaxin (22,31,46).
Synaptotagmin is a member of a large gene family (17)(18)(19). The ability of distinct isoforms of synaptotagmin to bind to one another (Fig. 1) indicates that a substantial array of heterooligomers, each with distinct properties, may assemble in vivo. This property could provide a mechanism to fine tune the regulation of membrane traffic in distinct cell types and compartments. Future studies are required to determine whether individual secretory vesicles possess multiple isoforms of synaptotagmin and whether hetero-oligomers exhibit differences in their abilities to interact with and modulate effectors.
The finding that synaptotagmin IV bound the synprint peptide was somewhat surprising. Synaptotagmins I and III are thought to function as Ca 2ϩ sensors during exocytosis, whereas synaptotagmin IV, due the substitution of an aspartate that normally serves as a Ca 2ϩ ligand, does not bind Ca 2ϩ via its C2A domain (19). Why would a presumably Ca 2ϩ -insensitive isoform retain Ca 2ϩ channel binding properties? One possibility is that its C2B domain retains its ability to bind Ca 2ϩ and that this interaction functions in some capacity to couple Ca 2ϩ to the regulation of secretion. This hypothesis is supported by primary sequence analysis of C2B. Five acidic amino acid residues, which serve as Ca 2ϩ ligands in C2A (42), are present in the same positions in C2B (14). However, Ca 2ϩ binding has not yet been demonstrated for the C2B domain of any isoform of synaptotagmin.
An important outcome of this study is the generation of graded loss-of-function mutations. The K327A mutation reduced oligomerization, AP-2 binding, and synprint binding by 50%, whereas the double mutation (K326A,K327A) abolished these interactions. These mutations provide tools for gene replacement experiments to begin to address the function of these interactions in vivo. An additional principle that has emerged is the overlapping or coincident nature of the binding sites for different C2B-binding molecules. The ability of synprint to inhibit C2B-effector interactions suggests that synaptotagmin engages in these interactions in a sequential manner. Furthermore, the inhibitory effect of synprint provides yet another tool to address the physiological function of C2B-effector interactions.
Previous studies on the C2A domain of synaptotagmin demonstrated that residues within loops 2 and 3 form contacts with lipids (37,53), whereas residues within loops 1 and 3 interact with syntaxin (38). In the present study, we provide evidence that C2B makes contacts with effector proteins via Ca 2ϩ -binding loop 1 along with two ␤-strands that form a concave face of its putative three-dimensional structure (Fig. 7). From these data we can begin to build more detailed molecular models for the mechanism of C2 domain-effector recognition. Future studies will refine this model and, using loss-of-function mutations, will make it possible to address the role of specific effector interactions in exo-and endocytosis in vivo.