The COOH Terminus of Synaptotagmin Mediates Interaction with the Newexins*

The interaction of the synaptic vesicle protein, synaptotagmin, and the presynaptic a-latrotoxin receptor, a neurexin, has been proposed to be involved in docking of synaptic vesicles at active sites or modulation of neu- rotransmitter release. Here I report the investigation of the domain of synaptotagmin responsible for this inter- action. Pieces of synaptotagmin containing the carboxyl terminus are capable of purifying neurexins from solu- bilized brain homogenates. Pieces as small as a synthesized peptide corresponding to the COOH-terminal 34 amino acids are capable of enriching neurexins 100-fold. The binding of neurexins to synaptotagmin is calcium-independent and of moderate affinity. This COOH-termi- nal segment of synaptotagmin is conserved in all species characterized to date. Reflective of this, a synthetic pep- tide corresponding to the carboxyl terminus of Drosophila synaptotagmin is capable of purification of rat neurexins, suggesting the possibility that this interaction may also exist in Drosophila. I propose that the carboxyl terminus of synaptotagmin binds to the car- boxyl terminus of the neurexins and that this interaction may mediate docking of synaptic vesicles or modu- lation of neurotransmitter release.

The interaction of the synaptic vesicle protein, synaptotagmin, and the presynaptic a-latrotoxin receptor, a neurexin, has been proposed to be involved in docking of synaptic vesicles at active sites or modulation of neurotransmitter release. Here I report the investigation of the domain of synaptotagmin responsible for this interaction. Pieces of synaptotagmin containing the carboxyl terminus are capable of purifying neurexins from solubilized brain homogenates. Pieces as small as a synthesized peptide corresponding to the COOH-terminal 34 amino acids are capable of enriching neurexins 100-fold. The binding of neurexins to synaptotagmin is calciumindependent and of moderate affinity. This COOH-terminal segment of synaptotagmin is conserved in all species characterized to date. Reflective of this, a synthetic peptide corresponding to the carboxyl terminus of Drosophila synaptotagmin is capable of purification of rat neurexins, suggesting the possibility that this interaction may also exist in Drosophila. I propose that the carboxyl terminus of synaptotagmin binds to the carboxyl terminus of the neurexins and that this interaction may mediate docking of synaptic vesicles or modulation of neurotransmitter release.
Synaptotagmin has been hypothesized to be involved in calcium-dependent exocytosis of synaptic vesicles at the synapse (1). This was first based on molecular characterization of synaptotagmin showing that the cytoplasmically exposed part of the protein contained two repeats with homology to a domain involved with calcium-dependent membrane interaction. This domain was first identified in protein kinase C (21, but has subsequently been found in an arachidonic acid-specific phospholipase A2, a phospholipase C, and GTPase-activating protein (GAP) (3). Synaptotagmin has been shown to bind calcium in the presence of acidic phospholipids with a n affinity in the range of 10-50 p~ (41, a concentration range hypothesized to be involved in calcium-dependent exocytosis of synaptic vesicles at synapses (5). Recently, genetic analysis of synaptotagmin in Drosophila and Caenorhabditis elegans has shown that neurotransmission is dramatically altered in the absence of synaptotagmin (6)(7)(8)(9). Electrophysiology of partial lack of function mutants point towards a direct role of synaptotagmin in calcium activation of exocytosis of synaptic vesicles (7).
Additional evidence on the involvement of synaptotagmin in neurotransmitter release has centered on the co-purification of synaptotagmin and a presynaptic receptor for the black widow spider venom toxin a-latrotoxin. Purification of this receptor on a column of immobilized latrotoxin yields two proteins of 200 * This work was supported by United States Public Health Service Grant R01 NS30541 and an Alfred P. Sloan Fellowship. The costs of charges. This article must therefore be hereby marked "aduertisement" publication of this article were defrayed in part by the payment of page in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. and 160 kDa. Partial amino acid sequences from these proteins led to the molecular cloning of a family of proteins that have been named the neurexins (10). Also eluting off this column is synaptotagmin (11). This co-purification suggests that synaptotagmin, a synaptic vesicle protein, and the a-latrotoxin receptorheurexins, a presynaptic membrane protein, interact. This potentially has additional significance because a-latrotoxin causes massive exocytosis of synaptic vesicles even in the absence of external calcium. This raises the possibility that the interaction of synaptotagmin and the a-latrotoxin receptor may mediate synaptic vesicle docking or mediate modulation of neurotransmitter release.
The neurexin family is complex due to multiple genes with extensive alternative splicing (10, 12). To date it has not been determined which forms of the neurexins represent receptors for a-latrotoxin. All transmembrane forms of the neurexins have extensive homology in their short carboxyl-terminal cytoplasmic domain. This domain of the neurexins has been shown to be capable of mediating binding to synaptotagmin (13).
We have investigated the domain of synaptotagmin responsible for binding neurexins based on evolutionary conservation of synaptotagmin between greatly separated species. Synaptotagmin has now been cloned from species as evolutionarily distant as fruit flies and man (14)(15)(16). A comparison between these species is revealing. Overall these proteins are 55% identical but this conservation is not evenly distributed. Sequence identity is restricted to the C2 repeats (79% identity) and the COOH-terminal43 amino acids (65% identity). This pattern of conservation suggests that these domains are functional modules. The C2 domains, based on their homology to a similar domain in other proteins, have been suggested to be involved in calcium-dependent membrane interactions, potentially fusion (1). In this paper, we show that the COOH terminus is sufficient for substantial purification of neurexins from solubilized brain membranes. This implicates this domain in binding neurexins, again suggesting a potential role for synaptotagminneurexin binding in docking or modulation of neurotransmitter release.
EXPERIMENTAL PROCEDURES Materials-Restriction and DNA modifying enzymes were from New England Biolabs. Peptides were chemically synthesized. All other chemicals were of reagent-grade and used without further purification.
Antibodies-Three neurexin antibodies were raised against synthetic peptides coupled to keyhole limpet hemocyanin as described (17) or synthesized directly as a polymeric resin (18). The synthetic peptides had the following sequences: LR1, CSANKNKKNKDKEYYV residues 1493-1507; LR2 YRNRDEGSYHVDES residues 1455-1468; LR3, CGLEFPGAEGQWTRFPKW residues 30-46 (residue numbers as in neurexin l a (10)). LR4 was kindly provided by Dr. Martin Geppert and Dr. Thomas Sudhof, and was raised to a recombinant bacterial portion of the cytoplasmic domain of neurexin la. The amino-terminal cysteine in each peptide is not present in the protein sequence and was introduced to allow efficient coupling. Synaptotagmin 1 antibodies were as previously described (19). Monoclonal antibody that recognizes syntaxin A and B was kindly provided by Dr. Colin Barnstable. from chromatographies using affinity columns of recombinant pieces of synaptotagmin or synthesized peptides ( b ) were run on 7% SDS-PAGE gels, transferred to nitrocellulose, and blotted with LR1 antibody ( a ) or run on 10% gels and blotted with synaptotagmin antibody (c). Latrotoxin receptorheurexin ( L R ) control (2 pg) was purified on a microcolumn of a-latrotoxin as described by Petrenko et al. (23). STD, molecular weight standards.
Bacterial Expression-Bacterial expression of cytoplasmic sequences of synaptotagmin in the PET vector has been described previously (1,14). Briefly sequences corresponding to the appropriate protein region were amplified using the polymerase chain reaction with synthetic oligonucleotides containing restriction sites to insert each fragment inframe. Polymerase chain reactions were performed as described (14,20), and the single resulting fragment of each reaction that had the predicted size was purified by PAGE,' digested with NcoI and BglII, and cloned into the NcoI and BamHI sites of pET8c or pETlld (21). Recombinant protein expression was induced by adding 0.4 mM isopropyl-athiogalactopyranoside (IPTG) to a growing culture of BL21 (DE3) cells transformed with synaptotagmin expression vectors. After 2 h of induction, cells were harvested and analyzed by SDS-PAGE followed by Coomassie Blue staining or immunoblotting. PET l coded for amino acids 78-421 of rat synaptotagmin 1, PET 4 for amino acids 265-421, PET 8 for amino acids 135-421, PET 6 for amino acids 78-263, and PET 5 for amino acids 78-134. Recombinant proteins were purified by conventional chromatography or by SDS-PAGE on a Bio-Rad Prep Cell. PET 5 and PET 6 were purified by ion exchange and hydroxyapatite chromatography. PET 1, PET 4, and PET 8 were purified by SDS-PAGE. All recombinant proteins gave single bands on Coomassie-stained gels. Constructs were verified by sequencing. Recombinant proteins were checked for the correct amino acid sequence by Western analysis with antibodies to the p65-9 peptide (residues 387-421, rat sequence) for PET 1, PET 4, and PET 8 or antibodies to p65-6 peptide (linking domain, residues 100-120, rat sequence) for PET 5 and PET 6.
Purification on Synaptotagmin Affinity Columns-Synaptotagmin affinity columns were prepared from bacterially expressed pieces of synaptotagmin or chemically synthesized peptides. Insoluble protein (PET 1, PET 4, and PET 8 ) was purified by SDS-PAGE using a Bio-Rad Prep Cell. Protein in SDS was adjusted to pH 7.0 with HEPES and coupled to activated CH-Sepharose (Pharmacia

LKB Biotechnology
Inc.). After coupling, SDS was removed by extensively washing the column with 0.29 Triton X-100. Soluble proteins (PET 5 and PET 6) were directly coupled to activated CH-Sepharose. Peptides were coupled to thiopropyl-Sepharose using their introduced amino-terminal cysteine. Triton X-100-solubilized rat brain membranes were used for purification of neurexins on synaptotagmin columns. Briefly 10-20 rat brains were homogenized in 0.32 .M sucrose, 1 m M EGTA. Brain membranes were pelleted by centrifugation a t 30,000 xg,, for 40 min. Brain membranes were solubilized in 20 mv HEPES, pH 7.4, 1 mM EGTA, 1% Triton X-100. Unsolubilized material was removed by a 1% h, 100.000 x The abbreviations used are: PAGE, polyacrylamide gel electrophoresis; IPTG, isopropyl-1-thio-/3-D-galactopyranoside. g,,. centrifugation and discarded. For chromatography, solubilized brain membranes were adjusted to 100 m M KCI, 1 my free CaCI, and loaded onto columns over a period of several hours. The column was washed extensively with 100 mv KCI, 20 m M HEPES, 1 m M CaCI,, 0.29 Triton X-100. In most cases, protein was eluted in two-steps consisting of a first elution ( E l ) with 100 m M KCI, 20 mv HEPES, 10 mhl EDTA, 0.2% Triton X-100, then a second elution (E2) of 1 M NaCI, 20 mv HEPES, 10 m M EDTA, 0.2% Triton X-100.
SDS-PAGE and immunoblotting were performed as described (17). Proteins reactive with antibodies were visualized with peroxidase-labeled secondary antibodies. Protein assays were performed according to Bradford (22) using commercial reagents (Bio-Rad) and bovine immunoglobulins as standards.

RESULTS
Synaptotagmin has been shown to co-purify with the a-latrotoxin receptor (a neurexin) upon chromatography of solubilized rat brain membranes on immobilized latrotoxin columns (11,23). Similarly, i t has been found that proteins immunoreactive for antibodies to the latrotoxin receptorheurexin family can be greatly enriched on columns of immobilized recombinant synaptotagmin (11). These results suggest the possibility that a specific domain of synaptotagmin is responsible for interaction with the latrotoxin receptor and potentially other members of the neurexin family. Based on this data, we investigated whether sequence elements of synaptotagmin that mediate neurexin binding might be identified by enrichment of neurexins via chromatography on small recombinant pieces of synaptotagmin.
Synaptotagmin has at least six identifiable domains based on structural considerations or sequence conservation between evolutionarily distant species. These include a short aminoterminal intravesicular sequence, a single membrane spanning sequence, a linker domain, two C2 repeats, and a short carboxyl-terminal sequence. Recombinant pieces of synaptotagmin were constructed to include most combinations of these domains, excluding the luminal amino terminus and the membrane spanning sequence (Fig. l b ) . Our ideas directing this search were influenced by the conservation of synaptotagmin between evolutionarily distant species. When comparing synaptotagmin sequences between species as different as man and  (14), or different synaptotagmin genes within a species (15,24), three domains of the protein have high percentages of sequence identity. These are the two C2 domains and the carboxyl-terminal sequence that follows the second C2 domain. DNA coding for these pieces was generated by subcloning or by polymerase chain reaction using oligos containing convenient restriction sites. This DNA was ligated in-frame into PET vectors (21) transformed in bacteria, and recombinant protein was produced after induction with IPTG as has been previously described (1,14). Recombinant proteins were purified using standard chromatographies (PET 5 and PET 6) or purified by preparative PAGE on a Bio-Rad Prep Cell (PET 1, PET 4, and PET 8). In addition, two peptides, corresponding to part of the linking domain (residues 100-120, rat sequence) and to the last 34 carboxyl-terminal residues of synaptotagmin (residues 387-421, rat sequence), were chemically synthesized.

S T -G A E L R H W S D M L A N P R R P I A Q W H T L Q V E E E V D A M L A V K K S T -G A E L R H W S D M L A N P R R P I A Q W H T L Q V E E E V D A M L A V K K S T -G A E L R H W S D M L A N P R R P I A Q W H T L Q V E E E V D A M L A V K K L R H W S D M L A N P R R P I L R H W S D M L A S P R R P
The recombinant pieces of synaptotagmin and the two peptides were coupled to separate columns and tested for the ability to purify neurexins from solubilized rat brain membranes ( Fig. la). I followed a similar chromatography as was used for the purification of the latrotoxin receptor on immobilized latrotoxin columns (23). As has been reported (23), neurexins are greatly enriched via chromatography on columns of recombinant protein corresponding to the cytoplasmic sequence of synaptotagmin (PET 1, Fig. la). I specifically followed the purification of the neurexins using antibodies raised to the carboxylterminal 15 amino acids of neurexin la. This domain is highly conserved between neurexins I, 11, and 111 (three to four changes in 15 amino acids; Refs. 10 and 12). Immunoreactivity to neurexins is barely detectable in brain membranes when assayed by Western blotting (Fig. la). After chromatography, neurexin immunoreactivity is easily detectable. This enrichment represents a t least a 50-fold increase in neurexin immunoreactivity as measured by dilution of the elute necessary to get a comparable staining to starting membranes (data not shown). Using this approach, I investigated whether smaller pieces of synaptotagmin could yield similar purification of the neurexins. The bacterially produced recombinant proteins that were tested are diagrammed in Fig. 16. Peptides tested included a 34-amino acid COOH-terminal peptide (p65-9) and a 21-amino acid peptide in the linking domain (p65-6). As shown in Fig. l a , PET 1, PET 4, PET 8, and the COOH-terminal peptide support enrichment of neurexins. PET 6, PET 5, and the linking peptide do not enrich neurexins. In addition, uncoupled activated CH-Sepharose, thiopropyl-Sepharose, or these matrices coupled to bovine serum albumin or 2-mercaptoethanol yield no enrichment of neurexins (data not shown). All synaptotagmin pieces containing the COOH-terminal sequence were capable of enriching neurexins; all synaptotagmin pieces that did not contain the COOH-terminal sequence were not capable of neurexin enrichment. The result that the COOHterminal 34-amino acid peptide was sufficient for substantial enrichment of neurexins is particularly dramatic. The p65-9 column consistently gave the highest enrichment of the neurexins, usually to a 4-fold higher level that the next best column, PET 1. Aconcern of this chromatography approach is that synaptotagmin exists in the synaptic vesicle as a tetramer (19). Because of this, we wanted to be able to exclude the possibility that this enrichment of neurexins is mediated by prior enrichment of native synaptotagmin on these columns. To test this, we assayed for native synaptotagmin in the starting and elution fractions from chromatographies over each column. This is shown in Fig. IC for chromatographies from the COOH-terminal synaptotagmin peptide (p65-9) and PET 6 (linking domain and first C2 domain) columns. For both columns (and the other tested columns including controls), native synaptotagmin is carried through the chromatography. Importantly, enrichment of synaptotagmin in eluates is not greater in a column (~65-9) that enrichs neurexins compared to one (PET 6) that does not.
Additional evidence that the COOH terminus of synaptotagmin is functionally important is its conservation between evolutionarily distant species. This is shown in Fig. 2 where the COOH-terminal sequences (approximately last 40 amino acids) of cloned synaptotagmins (1, 14-16) and a synaptotagmin-like protein rabphilin (25) are aligned. Residues in the beginning and middle of these sequences are particularly conserved, while the carboxyl-terminal end of the sequence is variable in conservation and length. Much like a comparison of the sequences of the entire protein, there is more conservation of synaptotagmin 1 between species than between synaptotagmin 1 and 2 within the same species. Even a more distant synaptotagmin-like protein, rabphilin has some homology in this domain. This could suggest that different homologs of synaptotagmin in one animal may interact with different latrotoxin receptorlneurexins.
The recombinant pieces of synaptotagmin and the COOHterminal peptide purify a neurexin immunoreactive protein of 200 kDa that runs at the same molecular mass as the 200-kDa latrotoxin receptor purified from latrotoxin columns (Fig. la) FIG. 3. Protein purified from the COOH-terminal synaptotagmin (p65-9) column is immunoreactive to numerous neurexin antibodies. 25 pg of eluate 2 ( E 2 ) was run on a 7%-gel, transferred, and blotted with appropriate antibodies. For LR1 peptide competition, 50 pg/ml peptide was added.
neurexin immunoreactivity, we tested whether the reactivity of this band was blocked by preincubation with the COOH-terminal neurexin peptide. As shown in Fig. 3, peptide preincubation completely blocks the reaction of this antibody with this 200-kDa band. In addition, we tested three other antibodies raised against either peptides (LR2 residues 1456-1469, LR3 residues 30-46, of neurexin l a ) or recombinant protein (LR4) of the neurexins. All recognized the same 200-kDa band. A 160-kDa latrotoxin receptorheurexin purified from a-latrotoxin columns was not apparent in our elutions when brain membranes were chromatographed over these columns in 100 mM KCI. In contrast, when solubilized brain membranes were chromatographed over the same columns in 200 mM KCl, both 160-and 200-kDa bands were present in the elutions, although a t less than one-fifth the amount of the 200-kDa band with chromatography a t 100 mM KC1 (data not shown).
Neurexins are not eluted from the COOH-terminal synaptotagmin peptide column by removal of calcium. Similarly, enrichment of neurexins on synaptotagmin columns occurs in the absence of calcium in loading and washing buffers (data not shown), suggesting that the interaction of synaptotagmin and neurexins does not require calcium. In addition, we investigated the conditions for elution of neurexins from the COOHterminal peptide column. Solubilized rat brain membranes were chromatographed over the COOH-terminal peptide column in 100 mM KCl, 20 mM HEPES, 1 m M calcium chloride. The column was extensively washed with the same buffer. The column was then eluted with increasing concentrations of KC1 in 50 mM steps in the presence of 10 mM EDTA. As shown in Fig.  4, Neurexins elute from the COOH-terminal peptide column from 150 to 400 mM KCI. During this elution, I also assayed for enrichment of synaptotagmin and the presynaptic protein, HPClkyntaxin. I investigated the potential purification of HPCl/syntaxin because it has been reported to co-immunoprecipitate with synaptotagmin (26)(27)(28)(29). HPClkyntaxin is present in eluate fractions but it is not enriched by chromatography on the COOH-terminal peptide column. This is shown in Fig. 4b where eluate fractions have been loaded a t twice the protein concentration of the starting solubilized membranes. Under these conditions, HPCUsyntaxin immunoreactivity is barely visible. This suggests that synaptotagmin does not bind to HPCUsyntaxin via synaptotagmin's COOH-terminal sequence.
The conservation of the COOH-terminal sequence of synaptotagmin between evolutionarily distant species suggests that the interaction between synaptotagmin and neurexins may also be conserved between species. Because neurexins have a s yet only been identified in mammals, we investigated whether the COOH-terminal peptide from divergent species could be sufficient to enrich mammalian neurexins. As an initial test of this, we assayed whether mouse neurexins could be purified on rat COOH-terminal synaptotagmin peptide. As shown in Fig. 5, a 200-kDa mouse protein, immunoreactive to rat neurexin antibodies, is enriched by chromatography on a column of the rat COOH-terminal synaptotagmin peptide. For a further test of the evolutionary conservation of synaptotagmin-neurexin binding, we synthesized a peptide to the COOH terminus of Drosophila synaptotagmin (residues 451-474, Ref. 14). When coupled to a gel matrix, it also is sufficient for enriching rat neurexin. Chromatography of solubilized Drosophila head membranes over this column resulted in a prominent enrichment of a 200-kDa Drosophila protein (data not shown). We are currently investigating whether this Drosophila protein may be a neurexin. DISCUSSION The potential for a n interaction between synaptotagmin and the latrotoxin receptorheurexins was first noted because of the co-purification of synaptotagmin, a synaptic vesicle protein, during chromatography for the latrotoxin receptor, a presynaptic membrane protein. This suggested that the latrotoxin receptor and potentially many of the neurexins may be presynaptic membrane receptors for synaptotagmin. We have investigated the localization of sequence elements of synaptotagmin 1 that mediate its interaction with the latrotoxin receptorheurexins by assaying enrichment of neurexins from solubilized membranes by chromatography over columns of smaller pieces of synaptotagmin. This approach has identified the extensive (50-100-fold) enrichment of neurexins on a 34amino acid peptide corresponding to the COOH-terminal sequence of synaptotagmin 1 and on all recombinant pieces that contain this element. The position of a n interaction element a t the end of the protein is consistent with the structural and functional organization of synaptotagmin. Synaptotagmin has four domains in the cytoplasmically exposed part of the protein.
These are the linking domain, the first and second C2 repeat, totagmin allows the purification of mammalian neurexins from mouse brain membranes wrre chromatographed on p65-9 columns or rat brain membranes were chromatographed on D65-3 column. 50 pg of eluate 2 (E21 was run on Trh gels, transferred to nitrocellulose, and blotted with LR1 antibody. and the carboxyl-terminal sequence. Proteolysis of synaptotagmin reveals that trypsin or Pronase only cuts in the linking domain, even though synaptotagmin has many lysine residues (19,30). This suggests that the two C2 domains and carboxylterminal sequence form a tight structure. Given that the two C2 repeats have been shown to bind acidic phospholipids (1,14) and be involved in calcium-dependent membrane association (4), the carboxyl-terminal sequence would be ideally situated to orient or modulate the C2 domains. Potentially reflecting this, the COOH-terminal sequence is strongly conserved between such evolutionarily distant species a s worms, fruit flies, squid, and human. Such conservation suggests that the synaptotagmin-neurexin interaction may be conserved, and perhaps directly involved in neurotransmitter release. The data that a Drosophila COOH-terminal peptide can enrich mammalian neurexins is suggestive of the functional conservation of this interaction and suggestive that neurexins may exist in Drosophila.
The enrichment of neurexins on synaptotagmin columns can be used to gain insight into if and how this binding may function in vivo. First, this interaction is specific. We have shown that binding is mediated by a small 34-amino acid sequence element on the carboxyl terminus of synaptotagmin. Hata et al. (13), in complementary work, have shown that a small carboxyl-terminal domain on the neurexins is sufficient for binding to synaptotagmin. Thus, short cytoplasmic domains on each protein appear to be responsible for mediating their interaction. The identification of the domains responsible for binding on both proteins, the correct orientation of these domains in the nerve terminal, and the strong enrichment of neurexins on synaptotagmin columns and the enrichment of synaptotagmin on immobilized neurexins strongly suggest that this is a interaction that occurs in vivo. The calcium independence of this binding suggests that it functions before calcium influx and neurotransmitter release. Another property of this interaction is its apparent moderate strength. Both in this study and the study by Hata et al. (13), moderate concentrations (150400 mM) of salt elute neurexins from synaptotagmin columns. This suggests that synaptotagmin and neurexins have an intermediate affinity for each other. This would be reasonable for a transient interaction mediating one step in the synaptic vesicle pathway. Conceptually, a high affinity, specific interaction could be problematic for separating these proteins after fusion, when synaptic vesicle protein is recycled into new synaptic vesicles.
There is emerging evidence from study of other secretory pathways and analysis of the targets of proteolysis by clos-tridial neurotoxins that docking and fusion of synaptic vesicles may involve complexes of synaptic vesicle, cytosolic and presynaptic cell membrane proteins (31). In addition to binding latrotoxin receptorheurexins, synaptotagmin has been suggested to be part of a large synaptic vesicle-protein complex (32), to bind to the presynaptic membrane proteins, HPCl/syntaxin (26,28), and voltage-gated calcium channels (26)(27)(28), and perhaps indirectly to such cytoplasmic proteins a s N-ethyl-maleimide-sensitive fusion protein, a-soluble NSF attachment protein and synaptosome-associated protein 25 to form a large fusion complex (31). Because an interaction of synaptotagmin with HPClIsyntaxin has also been suggested to possibly mediate docking, we tested whether HPClkyntaxin might also bind to the carboxyl terminus of synaptotagmin. We did not get enrichment of HPClkyntaxin on p65-9 columns. This suggests that the carboxyl-terminal domain of synaptotagmin is not mediating an interaction with HPCl/syntaxin and that HPCl/ syntaxin is not required for the interaction of synaptotagmin with the neurexins. This is also supported by the work of Hata et al. (13). In their investigation, the binding of purified recombinant pieces of neurexins to a synaptotagmin column occurred in the absence of other proteins such as syntaxin. This does not rule out that HPClkyntaxin interacts with other domains of synaptotagmin. It seems likely that HPCkyntaxin or other nerve terminal proteins form a large complex, perhaps one that may be centered on the binding of synaptotagmin and neurexins.
The specificity of binding of neurexins to synaptotagmin, the orientation of both proteins, and the potential association of both proteins with neurotransmitter release suggests that their interaction may also be directly involved with neurotransmitter release. I would suggest two hypotheses on how this interaction would function in release. One hypothesis is that this interaction is involved in docking synaptic vesicles at the active site. Such a function would be attractive given synaptotagmin's role in calcium activation of neurotransmitter release (7). This would fit with the orientation of the two proteins. It also could fit with the calcium independence of this interaction because docking presumably occurs before calcium influx and fusion. Acaveat to this hypothesis is that multiple proteins may mediate docking (31). A second possibility is that binding by neurexins modulates synaptotagmin function. If synaptotagmin mediates calcium-dependent neurotransmitter release, then neurexins may be a focal point for modulation of neurotransmission. That at least some neurexins are receptors for the black widow spider venom toxin, latrotoxin (23, lo), and that latrotoxin causes massive exocytosis of synaptic vesicles even in the absence of external calcium (33) suggests the possibility for the second hypothesis. Unfortunately, it is presently difficult to directly test these hypotheses. Docking or modulation must be intimately entwined with release. The recent demonstration of genomic and minigene rescue of synaptotagmin null mutants in C. elegans and Drosophila (7-9) raises the possibility of in vivo tests of rescue with synaptotagmin constructs missing the carboxyl-terminal sequence.