Calcium can disrupt the SNARE protein complex on sea urchin egg secretory vesicles without irreversibly blocking fusion.

The homotypic fusion of sea urchin egg cortical vesicles (CV) is a system in which to correlate the biochemistry and physiology of membrane fusion. Homologues of vesicle-associated membrane protein (VAMP), syntaxin, and SNAP-25 were identified in CV membranes. A VAMP and syntaxin immunoreactive band at a higher apparent molecular mass (approximately 70 kDa) was detected; extraction and analysis confirmed that the band contained VAMP, SNAP-25, and syntaxin. This complex was also identified by immunoprecipitation and by sucrose gradient analysis. VAMP in the complex was insensitive to proteolysis by tetanus toxin. All criteria identify the SNARE complex as that described in other secretory systems. Complexes exist pre-formed on individual CV membranes and form between contacting CV. Most notably, CV SNARE complexes are disrupted in response to [Ca2+]free that trigger maximal fusion. N-Ethylmaleimide, which blocks fusion at or before the Ca2+-triggering step, blocks complex disruption by Ca2+. However, disruption is not blocked by lysophosphatidylcholine, which transiently arrests a late stage of fusion. Since removal of lysophosphatidylcholine from Ca2+-treated CV is known to allow fusion, complex disruption occurs independently from the membrane fusion step. As Ca2+ disrupts rather than stabilizes the complex, the presumably coiled-coil SNARE interactions are not needed at the time of fusion. These findings rule out models of fusion in which SNARE complex formation goes to completion ("zippers-up") after Ca2+ binding removes a "fusion-clamp."

Membrane fusion is the fundamental cellular process by which exocytotic secretion, enveloped virus entry, intracellular trafficking, and fertilization occur. Recently, conceptual advances have been made in how we think of the proteins involved in intracellular membrane trafficking and exocytosis (1)(2)(3), due in part to the discovery of a series of interactions between homologous proteins known to be required for membrane trafficking in vitro, yeast secretion in vivo, and synaptic transmission at the neuromuscular junction. These interactions are thought to contribute to the formation of a protein complex that is postulated to mediate the targeting, docking, and subsequent fusion of membranes (2). Since clostridial tox-ins cause a block of synaptic transmission, the finding that these same toxins specifically cleave the membrane proteins of this complex (termed "SNAREs") 1 supports the notion that the complex formed by SNARE components is essential in the pathway of Ca 2ϩ -triggered exocytosis (4). A cytosolic protein, the N-ethylmaleimide-sensitive factor (NSF), catalyzes the ATP-dependent disassembly of SNARE complexes assembled in vitro (5), suggesting that the disruption of the SNARE complex is actually causal in bilayer fusion (2). Alternate hypotheses propose that NSF and SNAPs have a chaperone-like function, altering the conformation of the SNAREs so as to activate them for docking, fusion, or both (6 -10). Most recently, it has been suggested that inter-membrane, coiled-coil SNARE interactions provide the energy for membrane contact prior to fusion (11)(12)(13).
Given that most systems studied are capable of many rounds of membrane fusion and retrieval, it is not clear at which stage in this cycle the SNARE complex functions. One model system that is not complicated by such recycling is the exocytosis of sea urchin egg cortical vesicles (CV). The major permanent block to polyspermy in many eggs is the synchronous exocytosis of thousands of CV, which line the plasma membrane, triggered by an increase in intracellular Ca 2ϩ caused by the fertilizing sperm. Isolated sheets of the plasma membrane with docked CV (called cell surface complexes or cortices) are fully functional for exocytosis (14,15). This preparation provides easy access to the intracellular compartment, as the cytoplasmic surface of the egg membrane faces the bathing buffer solution. The CV become localized and attached to the plasma membrane during maturation of the unfertilized egg; thus, all of the components needed for Ca 2ϩ sensing and bilayer fusion are pre-assembled and in place. In fact, only an increase to micromolar free Ca 2ϩ concentration is required to trigger exocytosis (14,16); by definition, both priming and docking have already occurred as ATP is not required for fusion (17)(18)(19). The CV are tightly bound to the plasma membrane. This has been demonstrated in ultrastructural studies and explains why the complete fusion machinery of the isolated planar cortex remains intact despite shear forces that remove cytoplasm (20). The lack of complications such as reserve vesicle mobilization (21) and vesicle recycling (22) makes it ideally suited for studying the final steps in a single round of exocytosis.
The CV can be removed from the plasma membrane. These isolated CV can be used to reconstitute CV-plasma membrane fusion (19,23) but will also fuse, with comparable Ca 2ϩ sensitivities, to other CV (24) and to liposomes (25). CV-CV fusion is one of the simplest cases of membrane fusion. In particular, the fact that the isolated CV can fuse with liposomes indicates that CV have sufficient protein machinery for Ca 2ϩ sensing and fusion.
Homologues of rat VAMP2 vesicle-associated membrane protein and syntaxin 1A genes have been cloned from Strongylocentrotus purpuratus ovary and testis cDNA libraries, and the presence of the corresponding proteins in the egg and in sperm has been demonstrated (26,27). Moreover, there have been two reports showing involvement of SNARE components in this sea urchin exocytosis in vivo and in vitro (28,29). Here, we have characterized the distribution and interactions of SNARE proteins on isolated CV to determine if these proteins could be elements of the molecular machinery for Ca 2ϩ -dependent CV-CV fusion. We demonstrate that homologues of syntaxin, VAMP, and SNAP-25 exist on CV and form the heterotrimeric complexes first described in other cell types. We have found that, in the absence of cytosolic factors, SNARE complex disruption in the native membrane is triggered by free Ca 2ϩ concentrations that trigger maximal membrane fusion. This is contrary to the hypotheses in which stable SNARE complexes are needed for fusion, since calcium does not stabilize them in causing fusion.

MATERIALS AND METHODS
Reagents-ATP, dithiothreitol (DTT), and protease inhibitors were purchased from Boehringer Mannheim, and bovine serum albumin was from ICN (Costa Mesa, CA). Polyvinylidene difluoride membranes were obtained from Millipore (Bedford, MA). Horseradish peroxidase-conjugated sheep anti-mouse and donkey anti-rabbit IgGs and enhanced chemiluminescence (ECL) reagents were from Amersham International (Amersham, UK). Tetanus toxin was purchased from List Biological Laboratories (Campbell, CA). N-Ethylmaleimide (NEM) and high quality calcium chloride were from Fluka (Ronkonkoma, NY). Lauroyllysophosphatidylcholine (LPC) was supplied by Avanti Polar Lipids (Alabaster, AL). All other reagents were of analytical grade and were purchased from Sigma. Antibodies were prepared as described (VAMP,Ref. 30) or were the kind gifts of Drs. Reinhart Jahn (SNAP-25, clone 71.1) and Thomas C. Sü dhof (syntaxin, I378).
Obtaining and Handling Eggs-Sea urchins of the species S. purpuratus were purchased from Marinus (Long Beach, CA) and maintained in aquaria in artificial sea water (435 mM NaCl, 40 mM MgCl 2 , 15 mM MgSO 4 , 11 mM CaCl 2 , 10 mM KCl, 10 mM HEPES, 1 mM EDTA, pH 8.0) at a temperature of 10°C. Eggs were obtained by injecting 0.5 M KCl into the intracoelomic cavity, and the jelly coat was removed by passing through 90-m nylon mesh. Eggs were maintained on ice until use.
Preparation of Sea Urchin Egg Cortical Granules-CV were prepared by a variation of the method of Crabb and Jackson (23). Cell surface complexes were prepared by homogenization in IM buffer (IM: 220 mM potassium glutamate, 500 mM glycine, 10 mM NaCl, 5 mM MgCl 2 , 5 mM EGTA, 1 mM benzamidine HCl, 2.5 mM ATP, 2 mM DTT, 2 g/ml aprotinin, 2 g/ml pepstatin, 2 g/ml leupeptin, pH 6.7) essentially as described previously (31) and were transferred to a high pH granule isolation buffer consisting of 450 mM KCl, 5 mM EGTA, 50 mM NH 4 Cl, 1 mM benzamidine HCl (pH 9.1) on ice. After 1 h many vesicles had detached from the plasma membrane. CV were purified from the plasma membrane debris by centrifugation at 700 ϫ g for 5 min at 4°C. The supernatant containing the CV was retained and this centrifugation was repeated. Finally, the CV were collected by centrifugation at 2000 ϫ g for 5 min at 4°C. This final CV pellet consisted exclusively of vesicles approximately 1 m in diameter.
CV-CV Fusion Assay-Fusion was measured at room temperature essentially as described previously (24). CV were suspended in IM to give an A 405 of 0.2 -0.3 units, measured using a ThermoMax microtiter plate reader (Molecular Devices, Menlo Park, CA). Aliquots of a suspension of CV (100 l) were dispensed into 96-well, flat-bottom microtiter plates (Costar, Cambridge, MA) and CV-CV contact initiated by centrifugation (1000 ϫ g, 10 min). The turbidity (A 405 ) of the resulting sheets of CV was measured. Fusion was triggered by addition of an equal volume of Ca 2ϩ -IM stock, designed to give the desired free Ca 2ϩ concentration, and the plates were centrifuged again. A final turbidity measurement was made and the extent of fusion calculated as ⌬A/ A initial , corrected for a background determined by lysing the CV with distilled water. Final free Ca 2ϩ concentrations were verified in mock samples using a Ca 2ϩ -sensitive electrode (World Precision Instruments, Sarasota, FL). All data were normalized so that the lowest free Ca 2ϩ concentration used (ϳ100 nM) gave 0% fusion; as there was no significant difference in the extent of fusion triggered by 150 -500 M free Ca 2ϩ , the mean fusion determined at these free Ca 2ϩ concentrations in each experiment was set to 100%.
Preparation of Membrane Protein Samples-Membrane proteins from CV were prepared by phase separation in Triton X-114 (32). Briefly, CV were dissolved in Triton X-114 buffer (20 mM PIPES, pH 6.8, 5 mM EGTA, 5 mM EDTA, 5 mM BAPTA, 2 mM DTT, 2 mM benzamidine, 2 g/ml aprotinin, 2 g/ml pepstatin, 2 g/ml leupeptin and 2.5% Triton X-114) for 10 min on ice. The sample was centrifuged at 1750 ϫ g for 5 min at 4°C to remove particulate material; no insoluble fraction was evident with CV but was found when cortices were extracted. After this, the solution was warmed to 30°C to induce phase partitioning and the two phases were separated by centrifugation at 1750 ϫ g for 10 min at room temperature. Protein was precipitated from the lower phase (containing the Triton X-114) by mixing with nine volumes of acetone/ ethanol (1/1) overnight at Ϫ30°C. The protein was concentrated by centrifugation at 1750 ϫ g for 5 min at 4°C and dissolved in sample buffer (50 mM Tris-HCl, pH 6.8, 1.5% SDS, 10 mM DTT, 2 mM EDTA, 11% sucrose, and 0.01% bromphenol blue), with or without heating for 3 min in boiling water. Total protein was measured by the bicinchoninic acid assay (Pierce) with bovine serum albumin as standard.
When CV were treated with tetanus toxin, 50 g of tetanus holotoxin was incubated with 10 mM dithiothreitol for 30 min at 37°C just prior to use. Isolated CV in 1 ml of IM buffer were incubated in the absence or presence of 300 nM toxin (8 g of tetanus toxin light chain equivalent) at 30°C for 16 h with constant rotation to prevent settling. Membrane proteins were then extracted from CV with Triton X-114 as described above. When the protein sample was prepared from CV treated with calcium buffer to induce fusion, CV were resuspended in IM to give an absorbance of 0.2-0.3 at 405 nm. Aliquots (3 ml) were dispensed into six-well flat-bottomed plates (Costar, Cambridge, MA). To make contacting, layered granules, these plates containing the suspended CV were centrifuged (1000 ϫ g for 5 min). After discarding 1.5 ml of supernatant, 1.5 ml of Ca 2ϩ -containing IM buffer to make final free Ca 2ϩ concentrations of 100 -450 M, or an equal volume of IM buffer (control), was added to the wells to trigger fusion, followed by centrifugation at 1000 ϫ g for 10 min. Proteins were extracted with Triton X-114 as described above. Pretreatments with and without NEM (5 mM) or LPC (100 M) were carried out on CV in suspension at room temperature for 30 min or 5 min, respectively. Before Ca 2ϩ treatment, all samples of CV suspension were examined microscopically to ensure that over 98% of CV remained as single vesicles of 1 m diameter. Parallel fusion assays were always carried out when the effects of Ca 2ϩ , tetanus toxin, or inhibitors of fusion were being studied for their effects on protein complexes.
Electrophoresis and Western Blotting-SDS-PAGE was performed in 12% polyacrylamide gels. Following electrophoresis, proteins were transferred to polyvinylidene difluoride membranes in a transfer buffer containing 25 mM Tris, 192 mM glycine, and 20% methanol. Blots were blocked for 30 min at room temperature in a solution consisting of 140 mM NaCl, 10 mM NaPO 4 , 0.05% Tween 20, 50 mg/ml bovine serum albumin, pH 7.4. Blots were incubated 90 min at room temperature with primary antibodies diluted in a solution consisting of 140 mM NaCl, 10 mM NaPO 4 , 0.05% Tween 20, 20 mg/ml bovine serum albumin, pH 7.4. Immunoreactive bands were detected using secondary antibodies conjugated to horseradish peroxidase, followed by ECL. Quantitation of immunoblot signals was done by densitometric scanning of films and analysis with ImageQuant software (Molecular Dynamics, Sunnyvale, CA).
SNARE Complex "Dissection"-A small strip corresponding to the location of the 70-kDa VAMP immunoreactive band was cut from a preparative 12% polyacrylamide SDS gel following electrophoresis of an unboiled sample of Triton X-114-extracted CV membrane protein. This piece of gel was finely minced and the protein eluted by grinding with a micropestle in the presence of SDS sample buffer. After removal of gel fragments by centrifugation, protein was concentrated by precipitation with chloroform-methanol, dissolved in SDS-sample buffer once more, boiled, and analyzed by SDS-PAGE and immunoblotting.
Immunoprecipitation-Isolated CV were dissolved in ice-cold solubilization buffer (100 mM KCl, 20 mM Tris-HCl, 8 mM EDTA, 2 mM DTT, 0.12 mM AEBSF, 10 g/ml aprotinin, and 10 g/ml leupeptin) contain-ing 1% (v/v) Triton X-100. 10 l of syntaxin 1A antiserum or VAMP-2 antiserum was covalently coupled to protein G-agarose (100 l of packed beads) using an Immunopure ® protein G IgG orientation kit (Pierce). The beads were washed with the buffer described above, and incubated with solubilized CV samples for 5 h at 4°C, followed by washing again with the same buffer. Bound antigens were eluted with 50 mM triethanolamine (pH 11.5). The immunoprecipitates were analyzed by immunoblotting using anti-syntaxin 1A and anti-VAMP-2 antibodies.

SNARE Proteins
Are Present on Isolated CV-Samples of extracted CV membranes were analyzed for the presence of SNARE proteins by SDS-PAGE and immunoblotting by standard procedures. CV contained proteins that cross-reacted with polyclonal antibodies to mammalian VAMP2 at an apparent molecular mass of ϳ19 kDa, with antibodies to syntaxin 1A at ϳ40 kDa, and with a monoclonal antibody to SNAP-25 at ϳ35 kDa ( Fig. 1). Samples of rat brain membrane proteins (5) yielded comparable results, although the mammalian VAMP ran at a slightly higher apparent molecular mass (data not shown).
Identification of a High Molecular Weight, SDS-resistant Complex of VAMP, Syntaxin, and SNAP-25 on CV Membranes-The feature of the SNARE hypothesis that accounts for the specificity of membrane interactions is the formation of a complex between pairs of t-and v-SNAREs. It is known that mammalian SNAREs can form a SDS-stable complex that is disrupted only at high temperature (33,34). To visualize the sea urchin SNARE complex, we prepared a CV membrane protein sample dissolved in SDS sample buffer without boiling. In these samples, syntaxin and VAMP immunoreactivities now also appeared in a form with lower electrophoretic mobility ( Fig. 2; about 70 kDa), in addition to the monomeric forms previously detected in boiled samples. Immunoblotting with the anti-syntaxin antibody always resulted in a broader and somewhat more diffuse band at 70 kDa than did blotting with the VAMP antibody. After boiling, the 70-kDa band was not detected by either antibody, as shown in Fig. 1. Additional studies showed that when cut and eluted from the gel, boiled, and subsequently analyzed by SDS-PAGE, the constituents of the 70-kDa band migrated as monomeric VAMP, syntaxin, and SNAP-25 (Fig. 2). Only one band of anti-VAMP-reactive protein at ϳ19 kDa was detected in this material derived from the complex band (data not shown). SDS-PAGE of solubilized CV protein at several different gel concentrations (7-15%) still yielded only a single band at 70 kDa, further indicating that this was a single complex (data not shown). Only a small fraction of the total amount of these proteins present in the CV is detected as a SDS-resistant complex.
To confirm that CV membrane proteins which cross-reacted with VAMP2, syntaxin 1A, and SNAP-25 antibodies interacted specifically to form a complex, like known t-and v-SNARES, CV membrane proteins were solubilized in 1% Triton X-100 and immunoprecipitated with anti-syntaxin 1A antibody or with anti-VAMP2 antibody. Each immunoprecipitate was analyzed by immunoblotting in duplicate lanes using anti-VAMP2, anti-syntaxin 1A, and anti-SNAP-25 antibodies. A protein recognized by the VAMP2 antibody was immunoprecipitated together with syntaxin using the syntaxin 1A antibody, while syntaxin co-immunoprecipitated with VAMP using the VAMP2 antibody (Fig. 3). SNAP-25 was also detected in both immunoprecipitates (Fig. 3). Thus a stable complex containing the three putative sea urchin SNAREs was recovered from Triton X-100 extracts of CV membranes. Immunoprecipitation using antibodies to a non-SNARE protein did not result in any detectable recovery of syntaxin and only a faint band corresponding to VAMP (data not shown). We also confirmed the presence of SNARE protein complexes using sucrose gradient centrifugation according to the procedures of Schulz et al. (27). The results of these experiments were also fully consistent with the presence of a heterotrimeric SNARE protein complex on the CV membrane (data not shown). Thus, by direct SDS-PAGE (Fig. 2), as well as co-immunoprecipitation studies (Fig.  3) and sucrose gradient analysis, a fraction of the total VAMP in CV was found in a complex together with both syntaxin and SNAP-25.
Effect of Tetanus Toxin on the SNARE Complex-Another feature of the SNARE complex is the resistance of its constituent proteins to cleavage by clostridial proteases (33,34). Tetanus toxin is a zinc endopeptidase that specifically recognizes and cleaves VAMP (35). It has been reported that exocytosis is inhibited in neuronal, neuroendocrine, and other cells treated with this toxin (28, 36 -43). To determine if the SNARE complex exists in a preparation of contacting CV membranes and is thus a candidate for involvement in Ca 2ϩ -triggered CV fusion, or whether it is only formed in detergent solution after extraction, we treated intact CV with tetanus toxin. Isolated CV were incubated at 30°C for 16 h in the absence or presence of 300 nM reduced tetanus toxin. After incubation, microscopic examination showed the CV to still exist singly in suspension. CV membrane proteins were extracted with Triton X-114, dissolved in sample buffer without boiling and analyzed by SDS-PAGE and immunoblotting with anti-VAMP2 antibody. Most of the sea urchin VAMP monomer was cleaved by tetanus toxin, whereas VAMP in the 70-kDa complex was insensitive to cleavage (Fig. 4A). However, tetanus toxin treatment did not inhibit FIG. 1. CV contain proteins that cross-react with antibodies to rat brain VAMP2, syntaxin 1A, and SNAP-25. Proteins from CV membranes were extracted with Triton X-114, then dissolved in SDS sample buffer, boiled, and analyzed by SDS-PAGE on 12% gels followed by immunoblotting (50 g of total protein/lane). The blot was probed with polyclonal antibodies to VAMP2 and syntaxin 1A and a monoclonal antibody to SNAP-25, as indicated. Immunoreactive bands were detected by ECL .   FIG. 2. High molecular mass immunoreactive bands, present in unboiled protein samples, contain VAMP, syntaxin, and SNAP-25. CV proteins extracted with Triton X-114 were dissolved in SDS sample buffer at 37°C and analyzed by SDS-PAGE on 12% gels followed by immunoblotting. Immunoblots were probed with antibodies to VAMP and syntaxin 1A as indicated. In parallel samples, CV proteins were separated by SDS-PAGE on 12% preparative gels. Protein in the 70-kDa region was recovered and analyzed once more by SDS-PAGE as described under "Materials and Methods." Resulting immunoblots were probed with antibodies to VAMP2, syntaxin 1A, and SNAP-25, as indicated.
CV-CV fusion (Fig. 4B) or CV-liposome fusion. 2 Thus the bulk of the VAMP monomer on CV is not needed for Ca 2ϩ -triggered fusion. It is possible that the remaining VAMP in the SNARE complex or as monomer was sufficient for fusion. Attempts to further lower the VAMP concentration on CV failed to reduce fusion, but also failed to remove all the VAMP (see "Discussion"). Since the other cross-reacting bands at Ͻ14 kDa are not components of the complex (Fig. 2), they are unlikely to be tetanus toxin-resistant VAMPs.
SNARE Complex Disruption under Conditions That Induce CV-CV Fusion-Might disruption of the SNARE complex accompany fusion if the complex is linking two membranes early in the fusion process, as suggested by the SNARE hypothesis (2)? We tested the effect of Ca 2ϩ , which triggers CV-CV fusion, on SNARE complex stability. Sufficient CaCl 2 was added to contacting, layered CV to yield a final free Ca 2ϩ concentration of 450 M, a concentration known to cause maximal fusion in this system (Ref. 24; see also "Materials and Methods"). Proteins were subsequently analyzed (without boiling) by SDS-PAGE, and immunoblots were probed with antibodies to VAMP-2. After this Ca 2ϩ treatment, the SNARE complex was no longer detected (Fig. 5). All the VAMP was now in the monomeric form. To check for losses of proteins from CV, proteins were also extracted from the supernatant after the addi-tion of Ca 2ϩ buffer, and from the aqueous phase after detergent extraction, but the SNARE proteins were not detected (data not shown). This complete Ca 2ϩ -triggered disruption of SNARE complexes was seen at all free Ca 2ϩ concentrations tested (ϳ90 -500 M; see also Figs. 6 and 7); parallel CV-CV fusion assays confirmed a maximal extent of fusion under these conditions (data not shown).
Attempts were made to mimic this disruptive effect of Ca 2ϩ on SNARE complexes in the native membrane by treating detergent-extracted protein complexes with Ca 2ϩ . Neither SDS-nor Triton X-114-solubilized SNARE complexes were disrupted by the addition of Ca 2ϩ (data not shown).
The SNARE Complex Exists on Single CV Membranes and Forms between Contacting CV-To test for the formation of SNARE complexes between interacting CV, we used a Ca 2ϩ pretreatment to disrupt the already existing complexes on individual CV. Non-contacting, isolated CV in suspension were treated with or without [Ca 2ϩ ] free sufficient to cause maximal fusion of contacting CV; samples were subsequently diluted in buffer to reduce the free Ca 2ϩ concentration, and finally sufficient CaCl 2 was added to the controls so that both controls and To test for the presence of a SNARE protein complex, CV membrane proteins were solubilized in 1% Triton X-100 prior to immunoprecipitation with anti-VAMP or anti-syntaxin antibodies. The immunoprecipitates were analyzed by immunoblotting using anti-VAMP, anti-syntaxin, and anti-SNAP-25 antibodies. As indicated by PAGE (Fig. 2), the SDS-resistant complex contains all three SNARE proteins.

FIG. 4. Tetanus toxin cleaves CV VAMP monomers but not VAMP in the high M r complex.
A, isolated CV in 1 ml of IM buffer were incubated for 16 h at 30°C in the absence (Ϫ) or presence (ϩ) of 300 nM tetanus toxin (Tetx). After incubation, CV membrane proteins were extracted, dissolved in SDS sample buffer without boiling, and analyzed by SDS-PAGE on 12% gels followed by immunoblotting. Immunoblots were probed with anti-VAMP antibody. Panel is representative of four experiments. B, fusion assays of CV (open triangles) or tetanus toxin-treated CV (solid triangles), carried out in parallel with the protein analysis (A) using a portion of each preparation.
FIG. 5. High free Ca 2؉ results in loss of the SNARE complex. CV suspended in IM buffer at sufficient density to give an A 405 of 0.2-0.3 were centrifuged to yield contacting, layered vesicles (see "Materials and Methods"). Samples were then treated with either IM buffer alone or Ca 2ϩ /IM stock to yield 450 M free Ca 2ϩ . Proteins were subsequently extracted with Triton X-114, dissolved in SDS sample buffer without boiling, and analyzed by SDS-PAGE on 12% gels followed by immunoblotting. Immunoblots were probed with anti-VAMP antibody. Densitometric scanning of the Western blot film indicates no 70-kDa complex (0 Ϯ 5% versus control, representative of eight experiments).
FIG. 6. Pre-formed SNARE complexes exist on CV and also form between contacting CV. SNARE complexes exist on isolated CV and are disrupted upon exposure to Ca 2ϩ ; complexes form again between contacting CV. Suspensions of monodisperse CV were treated with either IM buffer alone (lanes 1 and 3) or Ca 2ϩ -IM stock to yield 250 M free Ca 2ϩ (lanes 2 and 4); there was no fusion of CV under these conditions. These samples were then diluted 12-fold with buffer and extracted directly (lanes 1 and 2), or the CV were centrifuged into contact (lanes 3 and 4) prior to extraction with Triton X-114. Extracted proteins were dissolved in SDS sample buffer without boiling and analyzed by SDS-PAGE on 12% gels followed by immunoblotting. Immunoblots were probed with anti-VAMP antibody. Figure is representative of two experiments. Ca 2ϩ -pretreated samples had the same final Ca 2ϩ concentration. Proteins were extracted from the CV in suspension and analyzed by immunoblotting. As shown in Fig. 6 (lane 1), preexisting SNARE complexes were detected in control, non-contacting CV extracted from suspension. Ca 2ϩ -pretreated CV no longer had SNARE complexes (lane 2). Thus, SNARE complexes on single CV membranes also respond to Ca 2ϩ by disrupting. To test if SNARE proteins could form complexes between interacting CV, control and Ca 2ϩ -pretreated CV were centrifuged on separate plates to make contacting layers, and proteins were then extracted as before and analyzed by immunoblotting. When Ca 2ϩ -pretreated CV, devoid of complex, were diluted with IM buffer to chelate the added Ca 2ϩ and then centrifuged into contact, SNARE complexes reappeared (Fig. 6,  lane 4). As complexes do not reform on single CV in suspension over the same period of time (Fig. 6, lane 2), this suggests that SNARE proteins on opposed CV membranes can interact to form intermembrane complexes as proposed by the SNARE hypothesis (2).
Calcium Disrupts SNARE Complexes at a Late, Lipid-dependent Stage of Fusion prior to Bilayer Merger-To determine the stage during fusion at which complex disruption occurs, we used two inhibitors of CV-CV fusion known to act at different points in the fusion pathway. Prior treatment of CV with 5 mM NEM, which blocks fusion early at the stage of calcium sensing or initial conformational change (Refs. 24 and 31; confirmed, data not shown), prevented the Ca 2ϩ -triggered loss of the 70-kDa complex (Fig. 7, lane 3; 100% of control, lane 1). Treatment with lysophosphatidylcholine (LPC), which reversibly blocks a late, lipid-dependent step during bilayer merger, reversibly blocks CV-CV fusion (44). In contrast to the effect of NEM, the SNARE complex was no longer detectable in immunoblots of CV treated with LPC and then Ca 2ϩ (Fig. 7, lane 4; 0% of control, lane 1). CV-CV fusion assays were carried out in parallel and confirmed the reported inhibitory effects of LPC (44). Thus LPC cannot prevent the Ca 2ϩ -triggered loss of the 70-kDa complex, even though removal of LPC does not prevent subsequent fusion due to calcium (44). DISCUSSION In summary, an SDS-and tetanus toxin-resistant complex of syntaxin, VAMP, and SNAP-25 was found both on and between secretory vesicles of the sea urchin egg. This complex was disrupted by micromolar concentrations of Ca 2ϩ when on membranes, but not when the complex was in detergent solution. When bilayer fusion of contacting vesicles was reversibly inhibited with lysophosphatidylcholine, complex disruption was still caused by Ca 2ϩ , even though vesicles remained capable of rapid and complete fusion in response to Ca 2ϩ . These results suggest that 1) the SNARE complex of sea urchin secretory vesicles is stable, like those in other examples of Ca 2ϩ -triggered exocytosis and 2) SNARE complex disruption, not stabilization, is caused by the trigger for bilayer fusion: Ca 2ϩ . It is thus unlikely that the coiled-coil stability of the SNARE complex is required for bilayer fusion per se.
Properties of This SNARE Complex-The membrane proteins SNAP-25, syntaxin, and VAMP (synaptobrevin) have been implicated as central elements of an exocytotic membrane fusion complex in both neuronal (35,45) and non-neuronal cells (28,37,39,46). The presence of VAMP, syntaxin, and SNAP-25 homologues in membranes of the sea urchin egg secretory organelle, as reported here (Fig. 1) and suggested by other recent work (26,29), indicates that the molecular mechanisms controlling exocytosis are similar in neuronal cells and sea urchin eggs. Sea urchin cytosol contains a NSF homologue, NSF binding sites are present in the isolated exocytotic apparatus, and recombinant mammalian NSF and ␣-SNAP plus a Triton X-100 extract of CV form a 20 S complex (47). The core of the 20 S complex is the heterotrimeric, SDS-resistant complex formed by the SNAREs in the absence of NSF or ␣-SNAP (4,48). As shown in Figs. 2 and 3, sea urchin homologues of VAMP, syntaxin, and SNAP-25 also form this SDS-resistant complex. Whether this exists as the 70-kDa complex on CV membranes in vivo or whether it is part of a larger complex that is disrupted in the process of sample preparation is unknown. If the SNARE complex is important for fusion, then it must already be in a fusogenic conformation, since cytosolic factors and ATP are not required for Ca 2ϩ -triggered fusion (17)(18)(19).
In the SNARE hypothesis (2), syntaxin and SNAP-25 are postulated to be target membrane constituents, t-SNAREs, and are expected to be found only in the plasma membrane, not in secretory granules. As shown in Fig. 1, isolated CV contain syntaxin and SNAP-25 homologues. However, in unfertilized eggs, sea urchin CV are already docked. Therefore, we cannot exclude the possibility that when CV are isolated they may dissociate from the plasma membrane in such a way that the entire fusion complex, even parts originally derived from the plasma membrane, remains attached. Whether the syntaxin and SNAP-25 detected in isolated CV originate from the plasma membrane or from the CV membrane itself remains unknown. However, it has been reported that syntaxin is associated with chromaffin granules (49), purified synaptic vesicles (50,51), and internal membranes of non-neuronal cells (52), indicating that syntaxin is not exclusively localized to the plasma membrane. Furthermore, the presence of heterotrimeric complexes has also been seen on isolated synaptic vesicles (53) and yeast vacuoles (10).
The ability of tetanus toxin to proteolytically cleave the main CV VAMP band (Fig. 3) reinforced the identification of the sea urchin cross-reactive material as a VAMP homologue. Indeed, the only sea urchin VAMP isoform so far identified using molecular cloning contains a site likely to be a substrate for tetanus toxin (26). The failure of tetanus toxin to inhibit CV-CV fusion might be explained by the fact that CV VAMP complexed with syntaxin and SNAP-25 is resistant to cleavage by tetanus toxin, as has been shown in mammalian systems (54). This indicates that the complex is already formed on single isolated CV, substantiating the direct measurement on isolated vesicles (Fig. 6). Bi et al. (28) reported that tetanus toxin injection into the unfertilized sea urchin egg prevents exocytosis of CV that had been transiently dissociated from the plasma membrane by treatment with stachyose. This seems inconsistent with our data since tetanus toxin did not inhibit fusion between CV, which requires contact between vesicles and some form of docking. Although the reason for this difference is not clear, one possible explanation may be the procedure used to undock CV from the plasma membrane, such that undocking CV with stachyose in vivo disrupts the SNARE complex, while isolating CV with high pH buffer in vitro leaves complexes intact. CV undocked in vivo would then have only free VAMP monomer, the substrate for tetanus toxin. An alternative possibility is that the cleavage of CV VAMP monomer in vivo may trigger a degradation pathway for the CV that includes an inactivation of fusogenicity independent of toxin activity, perhaps involving cytosolic factors that are absent from our in vitro studies. It is also possible that tetanus toxin has activity in addition to the proteolysis of VAMP.
Another recent report demonstrated partial inhibition of exocytosis after tetanus toxin treatment of egg cortices from another species of urchin, Lytechinus pictus (29), despite apparently complete cleavage of VAMP by the toxin. This imperfect correlation between cleavage of VAMP and blockade of exocytosis has also been reported in other, non-neuronal systems (37,40,41) and has not been definitively explained. The discrepancy with our data, showing no inhibition of CV-CV fusion by toxin treatment, may have a simple explanation; VAMP is not absolutely required for membrane fusion, but has a priming or docking function that has already been carried out in fully primed and docked systems such as the sea urchin egg cortex. Partial inhibition of exocytosis in L. pictus cortices may therefore reflect transient undocking during toxin treatment due to incubation at 37°C, an unphysiologically high temperature for the sea urchin. Redocking may require VAMP. This interpretation is consistent both with the time-dependent decay in the extent of exocytosis, even in the absence of toxin in that study (29), and with the results after undocking of CV from the plasma membrane (28). In the isolated CV-CV fusion system, where docking is at least initiated by centrifugation of the vesicles into contact, the fraction of VAMP that is in the SNARE complex and protected from toxin cleavage (carried out herein at 30°C) might be sufficient for docking and fusion. We have tried to test this using a combination of Ca 2ϩ pretreatments to disrupt the SNARE complexes (as in Fig. 6) and incubation with tetanus toxin (as in Fig. 4), but have never succeeded in removing more than ϳ95% of the total VAMP; quantitative analyses yield an estimate of ϳ3500 copies of VAMP per CV with VAMP in excess over syntaxin and SNAP-25 (data not shown). However, even these treatments had no effect on Ca 2ϩ -triggered CV-CV fusion. Thus, VAMP, crucial for some step in exocytosis, may not be part of the bilayer fusion apparatus per se. Several other examples of VAMP-independent membrane fusion have been described. VAMP cleavage by tetanus toxin in permeabilized pancreatic B-cells strongly inhibited Ca 2ϩ -stimulated, but not GTP␥Sstimulated insulin release (41). v-SNARES are not required for yeast homotypic vacuole fusion in vitro as t-SNARE vesicles can fuse to each other in the absence of v-SNARE (7). Likewise, neurons of embryonic Drosophila expressing tetanus toxin still show spontaneous release of neurotransmitter (55).
The recent identification of novel VAMP isoforms in mammalian cells raises the possibility that such isoforms might also exist on CV. While the possible existence of another VAMP isoform on the CV membrane, which is not cleaved by tetanus toxin, cannot now be excluded, results to date suggest the presence of only one isoform of each of the SNARE proteins on a given secretory organelle (52,56,57). PCR-based cloning strategies in S. purpuratus ovary have yielded only one VAMP isoform and one syntaxin in this tissue (26), while various strategies have led to the isolation of a single SNAP-25 isoform from both ovary 3 and testis 4 of S. purpuratus. SNARE Complex Disruption by Calcium-We have shown that the SNARE protein complex is not found on CV under conditions of maximal Ca 2ϩ -induced fusion. The Ca 2ϩ -induced loss of complex may coincide with protein conformational changes or with a disruption of the complex that might occur during the fusion process. We detected SNARE complexes from contacting layers of CV (Figs. 2 and 4 -7) and by immunoprecipitation of CV protein extracts (Fig. 3). However, as in synaptic vesicles (53) and yeast vacuoles (10), SNARE complexes already exist on single CV membranes (Fig. 6). These complexes are also Ca 2ϩ -sensitive, even in the absence of membrane-membrane contact, undergoing complete disruption at [Ca 2ϩ ] free capable of causing maximal CV-CV fusion. This is not consistent with the original SNARE hypothesis (2). Consistent with most hypotheses is the observation that SNARE complexes form between CV brought into contact in the presence of low [Ca 2ϩ ] free (Fig. 6). This indicates that SNARE proteins on single CV membranes form intermembrane complexes with complementary SNARE proteins on opposed membranes without the contemporary intervention of cytoplasmic factors such as NSF. These results are also consistent with reports showing that homotypic membrane fusion in yeast can use interactions between v-SNAREs on one membrane and t-SNAREs on the apposing membrane (7) but does not require NSF at the time of docking nor subsequently (8). However, unlike the yeast vacuole system, in which the ATP-dependent action of NSF and ␣-SNAP induces a metastable, fusion-competent state which lasts less than 90 min (8), isolated CV can be maintained as separate granules for Ͼ16 h in a primed, fusion-competent state (see Fig. 4B).
Docked and Primed Pre-fusion States Exist Downstream of Complex Disruption-This is the first report of a Ca 2ϩ -triggered disruption of the SNARE complex. Since the isolated CV system preserves both the biochemical constituents and the functional Ca 2ϩ sensitivity of the intact egg, we believe that this reflects a biological process. We note that strong interactions among membrane proteins may be needed for efficient membrane-membrane contact prior to fusion but could inhibit fusion pore enlargement if they persist (58). Thus, disruption of the SNARE complex, to effect its removal from the interface between two apposed membranes, may be one of the sites of Ca 2ϩ action during triggered membrane fusion. To date, most characterizations of protein-protein interactions have used SNARE complexes reconstituted in detergent or recombinant proteins lacking their membrane-spanning domains, and one recent study shows disassembly of the naturally occurring SDS-resistant complex in synaptic vesicles by NSF and ␣-SNAP (53). The present study has focused on disruption of the SNARE complex in native membranes by micromolar [Ca 2ϩ ] free , in the complete absence of cytosolic factors. Since detergent-solubilized complexes are not disrupted by Ca 2ϩ , we speculate that, in the native membrane, the component proteins of the SNARE complex may assume conformations that create a Ca 2ϩ binding site and/or that one or more other Ca 2ϩbinding factors (such as other proteins or lipids) are associated with the core complex but are lost during complex isolation.
The block of Ca 2ϩ -triggered CV-CV fusion, but not of complex disruption, by LPC (Fig. 7) indicates that complex disassembly must occur before fusion, consistent with predictions of the original SNARE hypothesis (2) but not subsequent modifica-tions (11)(12)(13). The SNARE complex can form between vesicles that are docked and primed to fuse, yet disruption of the complex occurs prior to fusion, in the LPC-arrested stage. These CV are still docked and primed, since vesicles exposed to Ca 2ϩ during the LPC-arrested stage are known to fuse upon subsequent washout of LPC and addition of Ca 2ϩ (59,60). If zippering-up of SNARE protein coiled-coil domains is the driving force for bilayer fusion, as has recently been suggested (11)(12)(13), this force must be stored in the membrane after Ca 2ϩ treatment in the LPC-arrested stage in a novel type of prefusion complex. This is unlike the pre-fusion complex in HAmediated fusion, in which HA activated by low pH during the LPC-arrested stage is still needed to obtain fusion after removal of LPC (61). Furthermore, it follows that Ca 2ϩ does not remove a putative "fusion clamp" that blocks further intermembrane coiled-coil domain interactions between SNARE proteins since that suggestion predicts that Ca 2ϩ would stabilize the SNARE complex, and the fact is that it disrupts the SNARE complex.
In conclusion, the combination of physiological and biochemical approaches has allowed us to narrow the field of hypotheses on the role of SNARE proteins in fusion. It is the reduced nature of the CV system that will now allow us to critically examine the role of the SNARE complex in the molecular mechanism for membrane fusion.