Domain Structure of Synaptotagmin ( ~ 6 5 ) ”

Synaptotagmin (p65) is an abundant and evolution- arily conserved protein of synaptic vesicles that contains two copies of an internal repeat homologous to the regulatory region of protein kinase C. In the cur-rent study, we have investigated the biochemical prop- erties of synaptotagmin, demonstrating that it contains five protein domains: an intravesicular amino-termi- nal domain that is glycosylated but lacks a cleavable signal sequence; a single transmembrane region; a se- quence separating the transmembrane region from the two repeats homologous to protein kinase C; the two protein kinase C-homologous repeats; and a conserved carboxyl-terminal sequence following the two repeats homologous to protein kinase C. Sucrose density gradient centrifugations and gel electrophoresis indicate that synaptotagmin monomers associate into dimers and are part of a larger molecular weight complex. A sequence predicted to form an amphipathic a-helix that may cause the stable dimerization of synaptotagmin is found in its third domain between the transmembrane region and the protein kinase C-homologous repeats. contains a single hypersensitive proteolytic site that is located immediately amino-termi- nal to the amphipathic a-helix, suggesting that synaptotagmin contains a particularly exposed region as the peptide emerges from the dimer. with different concentrations of pronase or trypsin in 0.13 M NaCl, 20 mM HEPES-NaOH, pH 7.4, for 30 min at 37 "C. Reactions were stopped by boiling in electrophoresis sample buffer (62 mM Tris-HC1, pH 6.8, 5% 2-mer- captoethanol, 2% SDS). Endoglycosidase F digestions were performed in 0.13 M NaCl, 20 mM HEPES-NaOH, pH 7.4, 1% CHAPS, 10 mM 2-mercaptoethanol, and 0.2 unit of endoglycosidase F (13). In the experiments in which synaptic vesicles were subjected both to endo- glycosidase F and protease digestion, synaptic vesicles were first digested with trypsin as described above. The reaction was stopped with 1 mg/ml N-tosyl-L-lysine chloromethyl ketone, and CHAPS and 2-mercaptoethanol were added to 1% and 10 mM, respectively. 0.2 unit of endoglycosidase F was added, and the mixture was incubated overnight at 37 'C. SDS-PAGE and immunoblotting were performed as described (4, 7, 14, 15). Proteins reactive with antibodies were visualized with peroxidase-labeled secondary antibodies. Protein assays were performed according Bradford containing synaptophysin from this supernatant by antibody bead (immunobeads) oligonucleotide (fourth lane, <2 pg of protein) results in a co-precip-itation of synaptotagmin and cytochrome bSs1 (none of which remains in the supernatant, third lane, 20 pg of protein), suggesting that the synaptotagmin and cytochrome b561 that are not on chromaffin gran- ules are on the synaptophysin-containing microvesicles.

Synaptotagmin (p65) is an abundant and evolutionarily conserved protein of synaptic vesicles that contains two copies of an internal repeat homologous to the regulatory region of protein kinase C. In the current study, we have investigated the biochemical properties of synaptotagmin, demonstrating that it contains five protein domains: an intravesicular amino-terminal domain that is glycosylated but lacks a cleavable signal sequence; a single transmembrane region; a sequence separating the transmembrane region from the two repeats homologous to protein kinase C; the two protein kinase C-homologous repeats; and a conserved carboxyl-terminal sequence following the two repeats homologous to protein kinase C. Sucrose density gradient centrifugations and gel electrophoresis indicate that synaptotagmin monomers associate into dimers and are part of a larger molecular weight complex. A sequence predicted to form an amphipathic a-helix that may cause the stable dimerization of synaptotagmin is found in its third domain between the transmembrane region and the protein kinase C-homologous repeats. Synaptotagmin contains a single hypersensitive proteolytic site that is located immediately amino-terminal to the amphipathic a-helix, suggesting that synaptotagmin contains a particularly exposed region as the peptide backbone emerges from the dimer. Finally, subcellular fractionation and antibody bead purification demonstrate that synaptotagmin co-purifies with synaptophysin and other synaptic vesicle markers in brain. However, in the adrenal medulla, synaptotagmin was found in both synaptophysin-containing microvesicles and in chromaffin granules that are devoid of synaptophysin, suggesting a shared role for synaptotagmin in the exocytosis of small synaptic vesicles and large dense core catecholaminergic vesicles. Synaptotagmin (p65) is an abundant integral membrane protein of synaptic vesicles whose primary structure contains two copies of a repeat homologous to the regulatory region of protein kinase C (1,2). In the preceding study, synaptotagmin was shown to be structurally and functionally conserved from Drosophila to humans (3). These experiments demonstrated that the protein kinase C-homologous repeats of synaptotagmin are conserved between vertebrates and invertebrates and that these repeats most likely mediate the ability of recom-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. binant synaptotagmin from humans, rats, and Drosophila to bind to phosphatidylserine (3).
The unique structural and functional properties of synaptotagmin suggest a central role of the protein in the exocytosis of synaptic vesicles. Therefore, the elucidation of its subcellular distribution with respect to small synaptic vesicles and large dense core vesicles and the determination of the transmembrane orientation, glycosylation, and other structural features of synaptotagmin is of great importance.
In order to study the biochemical properties of synaptotagmin, we have now raised antibodies against five different epitopes of the rat protein. In addition, the 5' end of the synaptotagmin message was mapped to ensure the validity of the proposed amino terminus of synaptotagmin. Sucrose density gradient centrifugations demonstrate that synaptotagmin is part of a high molecular weight complex in the synaptic vesicle membrane that contains synaptotagmin dimers as its basic unit. Our results demonstrate that rat synaptotagmin is a glycoprotein that contains an amino-terminal intravesicular sequence which is translocated without a cleaved signal sequence. We suggest that synaptotagmin dimerizes via an amphipathic a-helix that is located between the transmembrane region and the internal repeats homologous to protein kinase C. A single hypersensitive proteolytic site is found in synaptotagmin that maps to the amino-terminal end of the amphipathic a-helix, compatible with the notion that the synaptotagmin peptide backbone is particularly exposed at this point because it emerges from a rigidly interacting domain. Together, these results lead to a five-domain model of the structure of synaptotagmin in which the last two domains, the carboxyl terminus and the protein kinase C-homologous repeats, are the conserved and presumably functionally active parts of the protein.

EXPERIMENTAL PROCEDURES
Materials-Reverse transcriptase was obtained from Life Sciences, and proteases and endoglycosidase F from Boehringer Mannheim.
Restriction enzymes, DNA modifying enzymes, and DNA molecular weight markers were purchased from New England Biolabs. CHAPS' and Zwittergent 3-10 were obtained from Calbiochem. RNA molecular weight markers were from Bethesda Research Laboratories, and reagents for SDS-PAGE including protein molecular weight markers and protein assay components were from Bio-Rad. Antibodies against synaptophysin, synaptobrevin, rab3, and cytochrome bSG1 were described previously (4)(5)(6)(7). The antibody against the 116,000 subunit of the proton pump was raised against a synthetic peptide corresponding to the carboxyl-terminal 18 amino acids of the subunit.' Peroxidaselabeled secondary antibodies were obtained from Cappel. Eupergit C1Z methacrylate Microbeads were purchased from Rohm Pharma.
Production of Synaptotagmin Antibodies-Three antibodies were raised against synthetic peptides coupled to keyhole limpet hemocyanin as described (4). The synthetic peptides had the following sequences (residue numbers as in (2), the topography of synaptotagmin is shown in Fig. 1 for identification of the domains): CMVSASH-PEALA (residues 1 to 11, amino terminus), CAINMKDVKDLG-KTMKDQALKD (residues 100 to 120, between transmembrane region and A repeat), and CMDVGGLSDPYVKIHL (residues 302 to 316, B repeat). The amino-terminal cysteine in each peptide is not present in the protein sequence and was introduced to allow efficient coupling. In addition to these three antipeptide antibodies, antibodies were raised against two bacterial recombinant proteins incorporating sequences comprised of residues 78 to 421 (complete sequence carboxyl-terminal to the transmembrane region) or residues 265 to 381 (B repeat). The production of the recombinant proteins was as described previously (2,3,8), and they were used for immunization after partial purification of inclusion bodies containing recombinant protein.
Subcellular Fractionation and Immunobead Purification-Purification of rat brain cortex synaptic vesicles with controlled pore-glass chromatography as the final step was performed as described previously (9,10). For most biochemical experiments, vesicles of lesser purity obtained before the controlled pore-glass chromatography were used. Subfractionation of bovine adrenal medulla via centrifugation through a 1.6 M sucrose step gradient for the purification of chromaffin granules was performed as described (11). Immunobead purification of synaptophysin-containing vesicles was performed as described using Eupergit C1Z methacrylate Microbeads coated with a purified monoclonal antibody against synaptophysin (12).
Protease and Endoglycosidase F Digestion of Synuptotagmin-synaptic vesicles (75 pg of protein) were digested with different concentrations of pronase or trypsin in 0.13 M NaCl, 20 mM HEPES-NaOH, pH 7.4, for 30 min at 37 "C. Reactions were stopped by boiling in electrophoresis sample buffer (62 mM Tris-HC1, pH 6.8, 5% 2-mercaptoethanol, 2% SDS). Endoglycosidase F digestions were performed in 0.13 M NaCl, 20 mM HEPES-NaOH, pH 7.4, 1% CHAPS, 10 mM 2-mercaptoethanol, and 0.2 unit of endoglycosidase F (13). In the experiments in which synaptic vesicles were subjected both to endoglycosidase F and protease digestion, synaptic vesicles were first digested with trypsin as described above. The reaction was stopped with 1 mg/ml N-tosyl-L-lysine chloromethyl ketone, and CHAPS and 2-mercaptoethanol were added to 1% and 10 mM, respectively. 0.2 unit of endoglycosidase F was added, and the mixture was incubated overnight at 37 'C.
Primer extension analyses were performed as described (17) using three "P-labeled oligonucleotides complementary to the following cDNA sequences: oligonucleotide A, 898 to 918; oligonucleotide B, 615 to 641; oligonucleotide C, 390 to 424 (nucleotide numbers according to Ref 2). Reaction products were analyzed by PAGE followed by autoradiography of the dried gels.
RNA blotting experiments were performed by hybridizing 32Plabeled synthetic oligonucleotides (B and C as described above) to blots of poly(A')-enriched RNA from rat brain. Hybridization were performed at 42 "C and washed at 50 "C as described (18,19).

Antibodies against Different Domains of Synaptotagmin-
The primary structures of the rat, human, and Drosophila synaptotagmins as deduced from their cDNA sequences predict the biosynthesis of a highly conserved protein that con-tains a single transmembrane region and two copies of a repeat homologous to the regulatory region of protein kinase C (3). The domain structure of rat synaptotagmin is compared to that of protein kinase C in Fig. 1, illustrating the localization of the internal repeats in synaptotagmin with respect to its transmembrane region and the similarity between repeats and protein kinase C.
In order to investigate the localization of synaptotagmin to synaptic vesicles and the domain structure of synaptotagmin, antibodies were raised against five different epitopes from rat synaptotagmin. Three of these were directed against synthetic peptides corresponding to the amino-terminal 11 amino acids of synaptotagmin, the sequence in the region separating the transmembrane region from the A repeat, and to a conserved sequence in the B repeat of synaptotagmin. The other two antibodies were raised against recombinant proteins containing either the entire sequence of rat synaptotagmin carboxylterminal to the transmembrane region (presumably its cytoplasmic domains) or the B repeat only. All five antibodies specifically and sensitively reacted with an M, = 65,000 protein in rat brain, although with different affinities, suggesting that they all recognized synaptotagmin (data not shown).
In a comparison of different species, the antibody against the amino-terminal peptide was specific for rat synaptotagmin and did not recognize bovine synaptotagmin (data not shown). The greatest sequence differences between human and rat synaptotagmin are found at the amino terminus (3), suggesting that sequence differences between bovine and rat proteins at the amino terminus may also be responsible for a lack of cross-speciescross-reactivity of the amino-terminal antibody. All other antibodies reacted with bovine synaptotagmin, but none reacted with Drosophila synaptotagmin (3).
Subcellular Localization of Synaptotagmin in Rat Brain and Bovine Adrenal Medulla-Using these antibodies, we examined whether synaptotagmin is highly enriched in synaptic vesicles or if it is shared with other organelles. For this purpose, rat brain homogenates were depleted of mitochondria and nuclei by low speed centrifugation (25,000 X gmaX for 20 min). Synaptic vesicles were immunoprecipitated from the resulting supernatant using beads coated with a monoclonal antibody against synaptophysin, a well characterized synaptic vesicle-specific protein in brain (4, 20, 21). Synaptotagmin was quantitatively co-immunoprecipitated with synaptophysin from the low speed supernatant by these antibody beads (Fig. 2). No synaptotagmin remained in the supernatant after synaptophysin-containing membranes had been removed, suggesting that in brain little synaptotagmin is localized in membranes other than synaptic vesicles.
In addition to antibody bead precipitation, the distribution of synaptotagmin on synaptic vesicles during the classical synaptic vesicle purification scheme with controlled poreglass chromatography as the last step (9, 10) was studied and compared with that of synaptophysin, synaptobrevin (5), and rab3A (6). Again we observed complete co-purification of all of these synaptic vesicle markers with synaptotagmin (data not shown). These results suggest that, like synaptophysin, synaptobrevin, and rab3, synaptotagmin is highly enriched on synaptic vesicles in brain.
Several proteins such as synaptophysin and rab3 are exclusively localized to synaptic vesicles in brain but are also expressed in the adrenal medulla (6, 21, 22). Here, synaptophysin, synaptobrevin, and rab3 are found on synaptic-like microvesicles that are distinct from the secretory granules in these cells (6,22). We therefore studied the subcellular localization of synaptotagmin in the bovine adrenal medulla using centrifugation and antibody bead precipitation. Total med- quantitatively remove synaptotagmin together with synaptophysin, a known synaptic vesicle-specific protein in brain cortex (20, 21) but precipitate less than 5% of the total protein, indicating a complete co-localization of these two proteins in rat brain cortex. The middle lane shows the supernatant from the antibody bead precipitation that has been depleted of synaptic vesicles by the antibody bead precipitation (10 pg of protein). ullary homogenate was subfractionated by low speed centrifugation (10,000 X gmaX for 20 min) into a crude chromaffin granule pellet and a supernatant containing microvesicles and plasma membranes. Chromaffin granules were then highly purified from the crude granule pellet using a sucrose step gradient (ll), whereas synaptophysin-containing microvesicles were purified from the low speed supernatant by antibody bead precipitation (Fig. 3). All fractions were analyzed by immunoblotting for the presence of synaptophysin, which is specific for microvesicles in chromaffin cells and absent from chromaffin granules, and for cytochrome b561, which is primarily localized to chromaffin granules but is also partially found on the synaptophysin-containing microvesicles (6,22). Fig. 3 shows that synaptotagmin co-localizes with cytochrome bs61 and is enriched to chromaffin granules that do not contain synaptophysin. These results indicate that in chromaffin cells synaptotagmin is sorted to both the synaptic like microvesicles and chromaffin granules. entution-The apparent molecular weight of synaptotagmin on SDS gels is M , = 65,000, whereas its sequence predicts a molecular weight of only M , = 47,500. Since a t least part of this discrepancy could be due to post-translational glycosylation, the sensitivity of synaptotagmin to cleavage by endoglycosidase F was investigated. After synaptic vesicles were digested with endoglycosidase F in the presence of detergent, a shift of approximately 3,000 in the apparent molecular weight of synaptotagmin was observed, suggesting that synaptotagmin is N-glycosylated (Fig. 4).
Protein kinase C is a cytoplasmic enzyme, suggesting that the domain of synaptotagmin containing the repeats homologous to protein kinase C may also be cytoplasmic. To test this hypothesis, the transmembrane orientation of synaptotagmin was investigated by partial proteolysis of synaptotagmin in intact synaptic vesicles. Synaptotagmin is cleaved by both trypsin and pronase a t a single site, suggesting that synaptotagmin has a hypersensitive proteolytic site that is  equally cleaved by these two proteases despite their different substrate specificities (Fig. 5). Using the antibody directed against the carboxyl-terminal domain of synaptotagmin, the same proteolytic product of approximately M , = 39,000 was detected over a 30-fold concentration range of both proteases.
The proteolytic fragment produced by mild trypsin or pronase treatment of synaptotagmin was soluble, whereas intact synaptotagmin was quantitatively pelleted with synaptic vesicles (data not shown). Higher protease concentrations led to a complete digestion of the carboxyl-terminal synaptotagmin epitope in intact synaptic vesicles. When proteolysis of synaptophysin was investigated under similar conditions, it was left intact except at the highest protease concentrations, and then only the cytoplasmic sequences were affected (4).
The release of a soluble carboxyl-terminal synaptotagmin fragment from intact synaptic vesicles after mild proteolysis suggests that the carboxyl-terminal sequences of synaptotagmin containing the homologies to protein kinase C are indeed cytoplasmic, and that a membrane-bound amino-terminal fragment should remain. To test this hypothesis, untreated and mildly proteolyzed synaptic vesicles were analyzed by immunoblotting using both the antibody against the carboxylterminal Cn-domains and an antibody raised against the 11 amino-terminal residues of synaptotagmin. As shown in Fig.  6, both antibodies recognized the same protein in intact synaptic vesicles. However, after partial proteolysis, the amino-terminal antipeptide antibody now labeled an M , = 28,000 band as opposed to the M , = 39,000 fragment detected by the antibody against the cytoplasmic epitope. Furthermore, when synaptic vesicles were subjected to endoglycosidase F digestion after proteolysis, only the smaller amino-terminal fragment changed in electrophoretic mobility (Fig. 6). These results suggest that synaptotagmin contains a small intravesicular amino-terminal and large cytoplasmic carboxyl-ter- The left blot was probed with the antibody against the recombinant protein in lane C, and the right blot with the antibody against a peptide corresponding to residues 100 to 120. Note that the latter antibody only weakly recognizes the proteolytic fragment and that the proteolytic fragment is approximately 5 kDa smaller than the recombinant protein, suggesting that its amino terminus is approximately localized between residues 90 and 110. minal sequence. The amino-terminal fragment is N-glycosy-late& inspection of the amino terminus of the synaptotagmin sequence reveals the presence of a single N-linked glycosylation site a t residue 24 (2).
To localize the hypersensitive proteolytic site in the synaptotagmin sequence, we studied which of the four antibodies that recognize epitopes carboxyl-terminal to the transmembrane region reacted with the cytoplasmic proteolytic fragment of synaptotagmin. In addition, its size was compared to that of recombinant synaptotagmin containing all sequences carboxyl-terminal to the transmembrane region (Fig. 7). The proteolytic fragment was found to be slightly smaller than the recombinant fragment whose amino terminus corresponds to residue 78 and to react with all antibodies against cytoplasmic sequences of synaptotagmin. The weakest response was obtained with the antibody against a synthetic peptide corresponding to residues 100 to 120, suggesting that the proteolysis affects this epitope. These results indicate that the hy-persensitive proteolytic site is localized between residues 90 and 100 of synaptotagmin.
Mapping of the 5' End of the Synaptotagmin Message-The discrepancy between the apparent molecular weight of synaptotagmin as determined by SDS-PAGE and its predicted size cannot be entirely accounted for by N-linked glycosylation since endoglycosidase F treatment of synaptotagmin causes only a small molecular weight shift (Figs. 4 and 6). There are several plausible explanations for the apparent high molecular weight of synaptotagmin. It may migrate a t a higher molecular weight on SDS-PAGE because it contains additional modifications such as 0-linked sugars, or it may have unusual secondary structures, or the predicted amino terminus of synaptotagmin maybe incorrect due to cloning artifacts. We therefore performed a series of experiments to rule out the possibility of cloning artifacts.
The finding that an antibody against the 11 amino-terminal residues of synaptotagmin reacts with the protein in synaptic vesicles demonstrates that these 11 amino acids are present in the mature protein. This suggests that a possible cloning artifact must be localized to the 5' untranslated region. In order to test if the long 5' untranslated region found by cDNA cloning is correct, blots of poly(A+)-enriched RNA from rat brain were hybridized with oligonucleotides from the 5' untranslated region and the coding region. Single messages of similar size (approximately 4.8 kilobases) were specifically hybridized with both probes (Fig. 8).
To investigate the validity of the cDNA cloning further, the 5' end of the rat synaptotagmin message was analyzed by primer extensions. Three different oligonucleotides were used that were complementary to either the 5' untranslated region or the 5' end of the coding region. With all three primers, specific extension products were observed with sizes as predicted by the cDNA sequence (Fig. 9). This result suggests that the 5' untranslated region of the cDNAs is correct and virtually full length and that no 5' coding region was missed due to a cloning artifact. Since the 5' untranslated region of synaptotagmin contains multiple stop codons in all three reading frames, it seems likely that the anomalously high apparent molecular weight of synaptotagmin is due to an intrinsic property of the protein.
Synaptotagmin Is Part of a High Molecular Weight Complex-Analysis of synaptotagmin by SDS-PAGE often resulted in the specific immunolabeling of a M, = 130,000 protein in addition to the monomeric M, = 65,000 protein.
Both the M, = 65,000 and 130,000 bands were similarly sensitive to endoglycosidase F digestion (Fig. 4). Both bands were also recognized by all five synaptotagmin antibodies (data not shown), suggesting that the M, = 130,000 band represents a synaptotagmin dimer that is partially resistant to SDS denaturation. To further investigate the possible association of synaptotagmin into dimers and high molecular weight complexes, synaptic vesicle proteins solubilized in CHAPS or in Zwittergent 3-10 were fractionated by sucrose density gradient centrifugation. The sucrose gradient fractions were then analyzed for synaptotagmin, synaptophysin, and the 116,000 subunit of the synaptic vesicle proton pump by immunoblotting (Fig. 10).
On sucrose density gradients, synaptotagmin migrates in a single peak as a high molecular weight complex of approximately M, = 220,000 in the presence of CHAPS, a nondenaturing detergent. Its molecular weight is higher than that of synaptophysin, which migrates as a broad peak comprised of dimers, trimers, and tetramers (23), and smaller than the proton pump complex (approximately M, = 530,000) (24). In Zwittergent 3-10, the proton pump is dissociated into its subunits and synaptotagmin co-migrates with the 116,000 subunit of the proton pump on the sucrose gradients. This suggests that Zwittergent 3-10 also partially dissociates the synaptotagmin complex but leaves synaptotagmin dimers intact (Fig. 10).
What is the mechanism of synaptotagmin dimerization? Analysis of the primary structure of synaptotagmin indicates the presence of a sequence that has a high potential of forming an amphipathic a-helix (shown in an a-helical wheel presentation in Fig. 11). Parallel strands of such an a-helix would not only be held together by their hydrophobic surface but would also form oppositely charged amino acid pairs along the helix, suggesting a mechanism by which stable dimers could be formed.

DISCUSSION
Synaptotagmin is an abundant synaptic vesicle membrane protein that contains two copies of an internal repeat homologous to protein kinase C (2). Synaptotagmin is highly conserved evolutionarily from Drosophila to humans, and it ap- pears to bind negatively charged phospholipids with a specificity dependent on both the acidic head group and hydrophobic diacyl backbone (2,3). These properties suggest a central role for synaptotagmin in synaptic vesicle function. Therefore, information on the biochemical properties of synaptotagmin in the synaptic vesicle membrane beyond its sequence is of great importance. In this study, we have raised a panel of polyclonal antibodies against different regions of synaptotagmin and used these to study the biochemical properties and subcellular localization of synaptotagmin.
Our results demonstrate that synaptotagmin is a glycoprotein that contains N-linked sugars. Synaptotagmin exhibits a single proteolytic hypersensitive site that was mapped between residues 90 and 110 based on the sizes of the fragments produced after proteolytic cleavage and on their reactivity with different antibodies. Cleavage at the hypersensitive site results in the release of a soluble M , = 39,000 fragment that corresponds to the carboxyl-terminal two-thirds of the protein and contains the protein kinase C-homologous repeats. A smaller amino-terminal fragment remains associated with the membrane. The amino-terminal proteolytic fragment is sensitive to endoglycosidase F digestion, whereas the carboxylterminal fragment is not.
Together, these results suggest the carboxyl-terminal twothirds of synaptotagmin including its protein kinase C-homologous repeats are cytoplasmic and that its 52 amino-terminal residues are intravesicular and N-glycosylated. This raises the question if synaptotagmin contains a cleavable signal sequence. The amino-terminal sequence of synaptotagmin is only remotely similar to a signal sequence (25) but exhibits sequence similarity to the amino terminus of the yeast a-1 mating factor precursor, a secreted eukaryotic protein that also lacks a cleavable signal sequence (26) (sequences are given in single letter code with the residue numbers shown on the left and right): * * * * * * * * Synaptotagmin: Mating factor:

A L A A P V N T T 25
Since mature synaptotagmin reacts with an antibody raised against the 11 amino-terminal residues of the amino acid sequence predicted from the cDNA clones, the amino terminus of synaptotagmin does not contain a cleavable signal sequence.
Prompted by the observation that synaptotagmin often appeared to migrate as a dimer on SDS-PAGE, sucrose density gradient centrifugations were performed in the presence of two different detergents to analyze the subunit structure of synaptotagmin. In the presence of CHAPS, a nondenaturing detergent, synaptotagmin migrated as a high molecular weight complex of approximately M , = 220,000. In Zwittergent 3-10, a detergent that leads to the dissociation of several multisubunit membrane proteins including the synaptic vesicle proton pump and the inositol 1,4,5-trisphosphate receptor (27), synaptotagmin migrated as a dimer at the same position as the monomeric 116,000 proton pump subunit (Fig. 10). These results suggest that synaptotagmin is part of a high molecular weight complex in the synaptic vesicle membrane whose basic component is a synaptotagmin dimer.
The primary structure of rat synaptotagmin contains a sequence with a strong potential of forming an amphipathic a-helix. This sequence could dimerize synaptotagmin via its hydrophobic face and form salt bridges between the two parallel helices (Fig. 11). Interestingly, the hypersensitive proteolytic site of synaptotagmin was mapped to the aminoterminal end of this amphipathic a-helix, compatible with the notion that the protein is particularly exposed to proteolytic attack at the point where its peptide backbone emerges from a leucine zipper-like dimer.
These biochemical data together with a sequence compari-son between different species lead to a five-domain model of synaptotagmin. The first domain consists of the intravesicular glycosylated 52 amino acids, the second domain of the transmembrane region, and the third domain is formed by the dimerizing amphipathic a-helix. The internal repeats homologous to protein kinase C constitute the fourth domain, and the fifth domain is made up of the last 40 amino acids following the internal repeats. All five domains are present in the rat, human, and Drosophila sequences of synaptotagmin, but only the last two are conserved evolutionarily, suggesting that they form the functional parts of the protein whereas the other domains may have a primarily structural role. Finally, the subcellular localization of synaptotagmin was investigated biochemically. Together with synaptophysin, synaptotagmin was found to be highly enriched in synaptic vesicles of brain cortex. In contrast, the distributions of synaptophysin and synaptotagmin partially segregate in the adrenal medulla. Here, synaptotagmin has the same distribution as cytochrome b561 and is found both on chromaffin granules and on synaptic-like microvesicles, whereas synaptophysin is exclusively localized to the latter (Figs. 2 and 3).
Several studies on the distribution of synaptophysin in the adrenal medulla have been reported with originally conflicting results (e.g. see Refs. 12, 22, and 28-30). However, more recently, a consensus seems to have evolved that most if not all synaptophysin in chromaffin cells is not on the chromaffin granules but on unidentified "synaptic-like microvesicles" in agreement with our results (e.g. see Refs. 6 and 31). Two studies have also previously investigated the distribution of both synaptophysin and synaptotagmin (p65) in the adrenal medulla (28, 29). In both studies, synaptotagmin was found to be on chromaffin granules in agreement with our findings, although synaptophysin in both studies was found to have more general distribution different from what is now recognized to be its restricted presence in synaptic-like microvesicles. The presence of synaptotagmin on chromaffin granules, which are similar to large dense core vesicles in brain, raises the possibility that this particular synaptic vesicle protein may have a general function as a docking or fusion protein in regulated secretion by neurons.