Distinct Ca-dependent Properties of the First and Second C-domains of Synaptotagmin I

Synaptotagmin I (SytI) is a synaptic vesicle protein that binds Ca and is essential for fast, Ca-dependent neurotransmitter release in the hippocampus, suggesting that it serves as a Ca sensor for exocytosis. Although SytI has two cytoplasmic C-domains, only the first C-domain was shown to exhibit Ca regulation; it binds phospholipids and syntaxin in a Ca-dependent manner. By contrast, the second C-domain is inactive in these assays and only binds putative interacting molecules in a Ca-independent manner. We have now discovered in a yeast two-hybrid screen for SytI-interacting molecules that the C-domains of SytI interact with themselves. Using immobilized recombinant C-domains from SytI and SytII, we found that only the second but not the first C-domains of these synaptotagmins are capable of affinity-purifying native rat brain SytI and that this binding is Ca-dependent, suggesting that only the second C-domain is capable of a Ca-triggered self-association. A relatively high Ca concentration (>100 μM) is required for binding in the presence of Mg; Sr and Ba but not Mg can substitute for Ca. Our data suggest that the second C-domain of SytI is also a Ca-regulated domain similar to the first C-domain but with distinct binding activities.


Synaptotagmin I (SytI) is a synaptic vesicle protein that binds Ca
and is essential for fast, Ca 2؉ -dependent neurotransmitter release in the hippocampus, suggesting that it serves as a Ca 2؉ sensor for exocytosis. Although SytI has two cytoplasmic C 2 -domains, only the first C 2 -domain was shown to exhibit Ca 2؉ regulation; it binds phospholipids and syntaxin in a Ca 2؉ -dependent manner. By contrast, the second C 2 -domain is inactive in these assays and only binds putative interacting molecules in a Ca 2؉ -independent manner. We have now discovered in a yeast two-hybrid screen for SytI-interacting molecules that the C 2 -domains of SytI interact with themselves. Using immobilized recombinant C 2 -domains from SytI and SytII, we found that only the second but not the first C 2 -domains of these synaptotagmins are capable of affinity-purifying native rat brain SytI and that this binding is Ca 2؉ -dependent, suggesting that only the second C 2 -domain is capable of a Ca 2؉ -triggered self-association. A relatively high Ca 2؉ concentration (>100 M) is required for binding in the presence of Mg 2؉ ; Sr 2؉ and Ba 2؉ but not Mg 2؉ can substitute for Ca 2؉ . Our data suggest that the second C 2 -domain of SytI is also a Ca 2؉ -regulated domain similar to the first C 2 -domain but with distinct binding activities.
Synaptotagmin I (SytI) 1 is a member of a family of neuronal proteins that is characterized by a short N-terminal intravesicular sequence, a single transmembrane region, and a large cytoplasmic sequence containing two C 2 -domains (reviewed in Sü dhof (1995)). Nine synaptotagmin isoforms have been described in mammals (Perin et al., 1990;Geppert et al., 1991;Wendland et al., 1991;Mizuta et al., 1994;Hilbush and Morgan, 1994;Li et al., 1995a;Craxton and Goedert, 1995;Hudson and Birnbaum, 1995), four of which are also found in non-neural tissues (Li et al., 1995a;Hudson and Birnbaum, 1995). At least three synaptotagmins (SytI, SytII, and SytIII) are synaptic vesicle proteins, of which SytIII is present throughout the brain whereas SytI and SytII show restricted complementary expression patterns with partially overlapping distributions (Ullrich et al., 1994). Mice in which the SytI gene has been mutated exhibit a lethal phenotype in which there is a selective loss of fast Ca 2ϩ -dependent neurotransmitter release in hippocampal synapses . Spontaneous neurotransmitter release and neurotransmitter release evoked by Ca 2ϩ -independent mechanisms are normal, suggesting an essential role for SytI only in the Ca 2ϩ -dependent last step of membrane fusion. Together with the Ca 2ϩ binding properties of SytI (Li et al., 1995a(Li et al., , 1995b, these data suggest that SytI may serve as an exocytotic Ca 2ϩ sensor. Based on the presence of C 2 -domains in SytI and the observation that the C 2 -domain confers Ca 2ϩ regulation onto protein kinase C, it was speculated early on that SytI may be a Ca 2ϩbinding protein (Perin et al., 1990). Indeed, experiments with purified SytI demonstrated that it binds Ca 2ϩ and phospholipids (Brose et al., 1992) and that it undergoes a conformational change as a function of Ca 2ϩ . Studies on recombinant C 2 -domains showed that the first C 2 -domain of SytI and of most but not all other synaptotagmins binds phospholipids as a function of Ca 2ϩ Chapman and Jahn, 1994;Ullrich et al., 1994;Li et al., 1995a). In addition, Ca 2ϩ triggers binding of syntaxin to the first C 2 -domain of synaptotagmins with a Ca 2ϩ dependence that is distinct from that of phospholipid binding, implying the presence of two Ca 2ϩ -binding sites in a single C 2 -domain (Li et al., 1995a(Li et al., , 1995b. Surprisingly, the second C 2 -domain of all synaptotagmins is inactive in these assays despite a high degree of sequence homology. Furthermore, the Ca 2ϩ -dependent binding properties of the native cytoplasmic domain of purified brain SytI containing both C 2 -domains has the same properties as the recombinant single first C 2 -domain (Li et al., 1995a(Li et al., , 1995b. Together these data suggest that the known Ca 2ϩ binding properties of SytI can be entirely accounted for by its first C 2 -domain alone and raised the possibility that the second C 2 -domain may not represent a Ca 2ϩbinding domain. This possibility was supported by the Ca 2ϩindependent interactions of the second C 2 -domains of synaptotagmins with clathrin AP2 Li et al., 1995a) and with polyanions such as polyinositol phosphates (Fukuda et al., 1994).
We now report the results of a yeast two-hybrid interaction screen for proteins binding to the C 2 -domains of SytI. Unexpectedly, SytI itself was identified as an interacting partner. In vitro binding assays demonstrated a Ca 2ϩ -dependent self-interaction of SytI that is specific for its second C 2 -domain. These data suggest that in addition to the first C 2 -domain, the second C 2 -domain of SytI is a Ca 2ϩ -regulated domain. However, the two C 2 -domains have distinct Ca 2ϩ -regulated properties, suggesting a functional diversification of C 2 -domains in synaptotagmins.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid (Y2H) Screens and Interaction Assays-A bait vector (pBTM116p65-8) encoding the cytoplasmic domains of SytI starting with the first C 2 -domain was constructed by cloning the 0.9-kilobase pair SmaI-SalI fragment from pGEX65-8  into the same sites of pBTM116 (Vojtek et al., 1993). This results in a vector * This study was partially supported by a grant from the Perot Family Foundation. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by a postdoctoral fellowship from the Muscular Dystrophy Association.
** To whom correspondence should be addressed. 1 The abbreviations used are: SytI and SytII, synaptotagmins I and II; Y2H, yeast two-hybrid; GST, glutathione S-transferase; PAGE, polyacrylamide gel electrophoresis. THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 271, No. 3, Issue of January 19, pp. 1262-1265, 1996 expressing a LexA-fusion protein with SytI starting at residue 120. A cDNA library was constructed in the NotI site of Y2H prey vector pVP16 (Vojtek et al., 1993) from poly(A) ϩ -enriched rat brain RNA using the Life Technologies, Inc. Choice system. Y2H screens were performed essentially as described (Fields and Song, 1989;Vojtek et al., 1993;Hata and Sü dhof, 1995) by sequentially transfecting yeast strain L40 with the bait vector and the cDNA library using the lithium acetate method (Schiestl and Gietz, 1989). Transformants were plated on selection plates lacking histidine, uracil, tryptophan, lysine, and leucine but containing 2.5 mM 3-amino triazol. Positive clones were picked after 4 -6 days of incubation at 30°C, and the ␤-galactosidase activity of the clones was assayed on a nitrocellulose filter. Extrachromosomal DNA from clones that grew in the absence of histidine and were ␤-galactosidase positive was isolated using the glass bead method (Ward, 1990). Prey plasmids were rescued in Escherichia coli HB101 cells by electroporation and selection on M9 plates containing 50 mg/liter proline and 0.1 g/liter ampicillin. 14 million yeast transformants with the cDNA library were screened, and 42 clones positive upon retransformation were isolated and sequenced using the dideoxy chain termination method.
SytI Binding to Recombinant Proteins-One frozen rat brain (Pelfreeze) was homogenized in 11 ml of 4 mM HEPES-NaOH pH 7.4 containing 0.1 g/liter phenylmethylsulfonyl fluoride. The homogenate was extracted for 4 h at 4°C after addition of 11 ml of 4 mM HEPES-NaOH pH 7.4, 0.1 g/liter phenylmethylsulfonyl fluoride, 0.2 M NaCl, 2% Nonidet P-40, 2 mM EDTA. Insoluble material was removed from the extract by centrifugation (30 min at 100,000 ϫ g), and MgCl 2 was added to the supernatant to 3.5 mM final concentration. 1-ml aliquots of the supernatant were incubated overnight at 4°C with glutathione-agarose beads (GSH beads) containing 5-10 g of immobilized GST-fusion proteins with either 3.5 mM CaCl 2 or 5 mM EGTA. The GSH beads were centrifuged, washed 5 times in the respective incubation buffers, and resuspended in 120 l of SDS-PAGE sample buffer, and 40 l were analyzed by SDS-PAGE and immunoblotting using antibodies against synaptotagmin I (Cl604.4; kind gift of Dr. R. Jahn, Yale University), syntaxin I (HPC-1), or AP2 (M11AC1; kind gift of M. Robinson, Cambridge). For the experiments testing the effects of various cations, identical procedures were used except that the extraction was performed in the absence of EDTA, no MgCl 2 was added to the extract, and the incubations with the GSH beads and washing steps were carried out with buffers containing 1 mM EGTA, CaCl 2 , MgCl 2 , BaCl 2 , or SrCl 2 . The Ca 2ϩ titration experiments were performed similarly except that the different indicated Ca 2ϩ concentrations were used in the incubation buffers.
Miscellaneous Procedures-SDS-PAGE and immunoblotting were performed using standard procedures and antibodies described previously (Laemmli, 1970;Johnston et al., 1989;Li et al., 1995a). Protein assays were performed with the Bio-Rad kit.

Yeast Two-hybrid Screens for Synaptotagmin I Interacting
Proteins-We screened 14 million yeast colonies transformed with a rat brain cDNA library with a bait construct encoding the cytoplasmic C 2 -domains of SytI. 42 clones were isolated that demonstrated ␤-galactosidase activation after retransformation of the prey plasmids into yeast. Sequencing revealed that most of these clones encoded either novel proteins or proteins unlikely to interact with SytI physiologically (such as mucin apoprotein; data not shown). One positive clone (pPrey820), however, encoded SytI itself, starting at residue 120 immediately N-terminal to the first C 2 -domain and containing its complete C terminus (data not shown). This clone resulted in high ␤-galactosidase levels upon co-transformation with the SytI bait construct, confirming an interaction in the Y2H assay.
Ca 2ϩ -dependent Binding of the Second C 2 -domain of SytI to Endogenous Brain SytI-The Y2H result suggested that the C 2 -domains of SytI can interact with themselves. To obtain independent evidence for such an interaction and to localize the interacting sequences, we purified GST-fusion proteins encoding either the first or the second or both C 2 -domains of SytI and SytII. Recombinant proteins were immobilized on glutathioneagarose beads, and binding of endogenous brain SytI to the immobilized fusion proteins was measured as a function of Ca 2ϩ . Fusion proteins containing both C 2 -domains or containing only the second C 2 -domain of either SytI or SytII efficiently affinity-purified SytI from total brain (Fig. 1). Strong binding was observed only in the presence of Ca 2ϩ , whereas in the absence of Ca 2ϩ weak binding was present. By contrast, the first C 2 -domains of SytI and of SytII were unable to bind. A point mutation in the second C 2 -domain of SytII that corresponds to a mutation in Drosophila synaptotagmin, which impairs synaptotagmin function (DiAntonio and Schwarz, 1994), had no effect on Ca 2ϩ -dependent binding. Together these results suggest that the C 2 -domains of synaptotagmins self-associate as a function of Ca 2ϩ via their second but not their first C 2 -domains. Since the low binding observed in the absence of Ca 2ϩ appears to be sufficient for an interaction observed in the Y2H assay, it is possible to identify Ca 2ϩ -dependent binding proteins for SytI using the Ca 2ϩ -independent Y2H screen.
Divalent Cation Specificity of Synaptotagmin Self-interaction-Previous studies on the first C 2 -domains of synaptotagmins demonstrated that Sr 2ϩ and Ba 2ϩ but not Mg 2ϩ can substitute for Ca 2ϩ in activating phospholipid binding, although with a much lower affinity Li et al., 1995b). We therefore tested the effects of different divalent cations on the binding of SytI to the second C 2 -domain of SytI (Fig. 2). Mg 2ϩ was unable to trigger binding and in fact inhibited binding compared with that observed in the absence of divalent cations (Fig. 2). By contrast, both Sr 2ϩ and Ba 2ϩ were capable of activating binding, with Ba 2ϩ having the lowest effect. Parallel incubations with GST alone demonstrated that the binding observed was dependent on the SytI-fusion protein and not due to divalent cation-dependent aggregation of SytI in the homogenate (lower panel in Fig. 2).
FIG. 1. Ca 2؉ -dependent binding of synaptotagmin I (SytI) from rat brain to recombinant C 2 -domains from SytI and SytII. GSTfusion proteins containing the first C 2 -domain (C 2 -A), the second C 2domain (C 2 -B), or both C 2 -domains (C 2 -A/B) of SytI or SytII were immobilized on glutathione beads. Immobilized recombinant proteins were incubated with solubilized rat brain homogenate containing 3.5 mM Mg 2ϩ in the presence or absence of Ca 2ϩ . Beads were washed 5 times, and bound proteins were analyzed by SDS-PAGE and immunoblotting with a monoclonal antibody to the N terminus of SytI that does not recognize the recombinant GST-fusion proteins. Note specific binding only to constructs containing the second C 2 -domain of SytI or SytII. For SytII, a mutant second C 2 -domain corresponding to a mutation observed in Drosophila (DiAntonio and Schwarz, 1994) was also analyzed (indicated by asterisks in protein names). In this mutant tyrosine 312 was changed to asparagine. Numbers on the left indicate positions of molecular weight markers; arrow points to SytI.

Ca 2ϩ Dependence of Second C 2 -domain Binding-To test the
Ca 2ϩ concentration dependence of the binding of brain SytI to the second C 2 -domains of SytI or SytII, we incubated brain homogenates with immobilized GST-C 2 -domain fusion proteins at different Ca 2ϩ concentrations and analyzed binding of SytI, syntaxin I, and clathrin AP2 to the immobilized proteins as a function of Ca 2ϩ (Fig. 3). As shown earlier Li et al., 1995a), binding of AP2 and syntaxin I to the second C 2 -domain was Ca 2ϩ -independent. By contrast, SytI binding was enhanced by Ca 2ϩ with an identical concentration dependence for the two C 2 -domains tested, with half-maximal binding at approximately 250 M free Ca 2ϩ . This Ca 2ϩ concentration dependence resembles that of syntaxin I binding to the first C 2 -domain of SytI and SytII (Li et al., 1995a). Note that these experiments were carried out in the presence of 3.5 mM Mg 2ϩ and the apparent Ca 2ϩ affinity may be higher in the absence of Mg 2ϩ .

DISCUSSION
SytI is a Ca 2ϩ -binding protein of synaptic vesicles that is essential for fast Ca 2ϩ -dependent neurotransmitter release from hippocampal neurons (Brose et al., 1992;Geppert et al., 1994), suggesting that it serves as the exocytotic Ca 2ϩ sensor. Previous studies demonstrated that the first C 2 -domain of SytI serves as a Ca 2ϩ -dependent phospholipid-and syntaxin-binding domain Sü dhof, 1993, 1994;Li et al., 1995aLi et al., , 1995b. The second C 2 -domain is inactive in these assays but binds AP2 and polyanions in a Ca 2ϩ -independent manner Fukuda et al., 1994). The phospholipid and syntaxin binding properties of a cytoplasmic fragment from SytI containing both C 2 -domains are identical to that of the single recombinant first C 2 -domain (Li et al., 1995a(Li et al., , 1995b. Together these results suggested that SytI may perform its Ca 2ϩ sensor function primarily via its first C 2 -domain whereas the second C 2 -domain may have a distinct function. We now demonstrate that the second C 2 -domain also has a Ca 2ϩ -dependent activity suggestive of a Ca 2ϩ -binding domain. It mediates the Ca 2ϩ -dependent aggregation of SytI and SytII with a Ca 2ϩ concentration dependence that mirrors the Ca 2ϩ de-pendence of neurotransmitter release if the experiments are performed in the presence of physiological concentrations of Mg 2ϩ . These data suggest a model of SytI whereby both C 2domains of SytI can serve as Ca 2ϩ binding modules with distinct functions (Fig. 4).
Even in the absence of Ca 2ϩ , SytI is not a monomer but a multimer (Perin et al., 1991). The basic unit of this multimer is an SDS-resistant dimer that can be detected by SDS-PAGE, and this multimerization is mediated by sequences N-terminal to the two C 2 -domains (Brose et al., 1992) (Fig. 4). Our current demonstration of a Ca 2ϩ -dependent binding of SytI to itself via its second C 2 -domain raises the possibility that during nerve terminal depolarization and Ca 2ϩ influx, SytI multimers may be cross-linked by Ca 2ϩ into large superstructures. The function of such a superstructure in fusion is unknown, but it is conceivable that it would aid in forming a pore that must occur during membrane fusion and probably involves assembly of a proteinaceous ring.
A considerable number of interactions has been described for synaptotagmins, not all of which may be physiologically important. Considering the point of action of SytI, it seems likely that Ca 2ϩ -regulated activities are more relevant than constitutive binding activities. Another criterion that supports the potential physiological relevance of an interaction is the colocalization of the binding partners. Based on these two criteria, the interactions of SytI with phospholipids, syntaxin, and itself appear to FIG. 4. Domain model of SytI and its binding activities. SytI binds phospholipids and syntaxin in a Ca 2ϩ -dependent manner via its first C 2 -domain and self-associates in a Ca 2ϩ -dependent manner via its second C 2 -domain. In addition, the N-terminal domains of synaptotagmin I self-associate in a Ca 2ϩ -independent manner via an undetermined sequence, and the C-terminal domains also bind AP2, polyanions such as polyinositol phosphates (IP 4 -6 ), and neurexins in a Ca 2ϩ -independent manner (see text for references). was immobilized on glutathione beads and incubated with solubilized brain homogenates containing either 1 mM EGTA, no additions, or 1 mM of the indicated divalent cations. Beads were washed in the incubation buffers, and bound proteins were analyzed by SDS-PAGE and immunoblotting. The asterisk identifies the position of full-length SytI; the 40-kDa band observed in lanes 3 and 6 containing high levels of bound SytI represents the major proteolytic product of SytI (Perin et al., 1991). Note that addition of Mg 2ϩ slightly suppresses binding. Numbers on the left indicate positions of molecular weight markers.
FIG. 3. Ca 2؉ dependence of SytI binding to the second C 2domains of SytI and SytII. Recombinant GST-fusion proteins of the second C 2 -domains of SytI and SytII (GSTSytIC 2 -B and GSTSytIIC 2 -B, respectively) were immobilized and incubated with brain homogenates in Mg 2ϩ -containing buffers with the Ca 2ϩ concentrations indicated on top of the panels (E ϭ EGTA). Beads were washed extensively in the incubation buffers. Bound proteins were analyzed by immunoblotting for the clathrin assembly protein AP2, SytI, and syntaxin I. Numbers on the left indicate positions of molecular weight markers.
be the most likely to be relevant. The recombinant second C 2 -domain of SytI also binds syntaxin (see Fig. 3) and phospholipids (Damer and Creutz, 1994) in a Ca 2ϩ -independent manner. However, when the complete double C 2 -domain fragment from native SytI prepared by partial proteolytic cleavage is analyzed, phospholipid binding and syntaxin binding are completely dependent on Ca 2ϩ , and little Ca 2ϩ -independent binding is observed (Li et al., 1995a(Li et al., , 1995b. Two other binding activities were demonstrated for synaptotagmin I that are Ca 2ϩ -independent and conceptually intriguing: binding of neurexins and AP2. Although no in vivo data exist to support the physiological significance of neurexin binding, AP2 binding may be physiologically significant since AP2 transiently associates with synaptic vesicles (Pfeffer and Kelly, 1985;Maycox et al., 1992) and synaptic vesicle recycling is severely impaired in synaptotagmin mutants in Caenorhabditis elegans (Jorgensen et al., 1995).