N-Ethylmaleimide-sensitive factor acts at a prefusion ATP-dependent step in Ca2+-activated exocytosis.

An ATP-dependent activity of NSF (N-ethylmaleimide-sensitive factor) that rearranges soluble NSF attachment protein (SNAP) receptor (SNARE) protein complexes was proposed to be the driving force for membrane fusion. The Ca2+-activated fusion of secretory vesicles with the plasma membrane in permeable PC12 cells requires ATP; however, the ATP requirement is for a priming step that precedes the Ca2+-triggered fusion reaction. While phosphoinositide phosphorylation is a key reaction required for priming, additional ATP-dependent reactions are also necessary. Here we report that the NSF-catalyzed rearrangement of SNARE protein complexes occurs during ATP-dependent priming. NSF with α-SNAP (soluble NSF attachment protein) were required for ATP-dependent priming but not Ca2+-triggered fusion, indicating that NSF acts at an ATP-dependent prefusion step rather than at fusion itself. NSF-catalyzed activation of SNARE proteins may reorganize membranes to generate a vesicle-plasma membrane prefusion intermediate that is poised for conversion to full fusion by Ca2+-dependent mechanisms.

The regulated fusion of vesicles with the plasma membrane in neural and endocrine cells requires a core complex of proteins (synaptobrevin, syntaxin, and SNAP-25) that are specific substrates for clostridial neurotoxin proteases (1)(2)(3)(4). This complex is proposed to function in vesicle targeting, docking or fusion. Identification of these neuronal synaptic proteins (termed SNAREs) 1 as receptors for SNAP proteins that mediate the membrane association of NSF, a protein required for constitutive membrane fusion (1), suggested that NSF may be required for Ca 2ϩ -regulated neurosecretion (5). Genetic studies in Drosophila have established an essential role for NSF in neural function (6). Stimulatory effects of ␣-SNAP on neurotransmitter secretion from chromaffin cells and squid neurons have been reported (7,8). However, the precise stage in the regulated secretory pathway at which NSF acts has not been directly established. In vitro biochemical studies demonstrated that a 20 S complex of SNAREs, NSF, and ␣/␤-SNAP was disassembled by the ATP-dependent activity of NSF, and it was suggested that NSF-catalyzed SNARE protein rearrangements drive membrane fusion (9). However, previous studies with permeable PC12 and adrenal cells had shown that MgATP was required for a priming step that precedes the final fusion steps triggered by Ca 2ϩ (10,11). In the present studies, the execution point of NSF and ␣-SNAP was established as the ATP-dependent priming step that precedes Ca 2ϩ -activated fusion.
Immunoisolation of SNARE Protein Complexes-Permeable cells were incubated with rat brain cytosol for 30 min with or without MgATP; ATP-primed cells were further incubated in triggering reactions with Ca 2ϩ where indicated. Cells recovered by centrifugation were solubilized in lysis buffer (20 mM HEPES, pH 7.2, 100 mM KCl, 2 mM EDTA, 0.5 mM ATP, 1 mM dithiothreitol, 1% Triton X-100, and 0.4 mM phenylmethylsulfonyl fluoride) and lysates were clarified at 150,000 ϫ g for 90 min. Immunoprecipitations were conducted on supernatants with SNAP-25 (Sternberger Monoclonals Inc.) and syntaxin (Sigma, HPC-1) monoclonal antibodies conjugated to protein G-agarose beads with dimethylpimelimidate (Pierce). Immunoprecipitates were washed 5 times in lysis buffer to completely eliminate nonspecifically bound proteins and were resolved by SDS-polyacrylamide gel electrophoresis. Immunoblotting of proteins transferred to nitrocellulose was conducted with antibodies to SNAP-25 and syntaxin (commercial), NSF and SNAP (from T. Sollner and J. E. Rothman), synaptobrevin-2 (from M. Takahashi), and synaptotagmin I (from R. H. Scheller). SDS-resistant SNARE complexes were prepared and analyzed as described (16).

RESULTS AND DISCUSSION
Mechanical permeabilization of PC12 cells generates semiintact cells that are highly permeable, allowing access to the exocytotic apparatus (12). Most of the large dense core vesicles (LDCVs) in intact PC12 cells are close to the plasma membrane in a docked state (17). LDCVs remain docked in the permeable cells at the plasma membrane ( Fig. 1). Hence, permeable PC12 cells provide a means to analyze late post-docking steps in the LDCV exocytotic pathway. Ca 2ϩ -regulated NE secretion via exocytosis of LDCVs is reconstituted in the permeable cells by provision of MgATP and cytosolic proteins (12). A two-stage protocol identified an ATP-dependent priming step that precedes and is  required for Ca 2ϩ -activated NE secretion (10). Priming preincubations with MgATP and cytosolic proteins fulfill the ATP requirement for Ca 2ϩ -regulated exocytosis, whereas ATP␥S fails to substitute and reversibly inhibits priming by competition with ATP ( Fig. 2a, left). Incubation of extensively washed ATP-primed permeable cells with 10 M Ca 2ϩ and cytosolic proteins results in NE secretion in the absence (Ͻ10 M) of ATP (Fig. 2a, right). Inclusion of ATP␥S did not impair Ca 2ϩ -triggered release of NE from ATP-primed cells, confirming that the final steps of LDCV fusion do not require ATP. We sought to correlate functional stages of priming and triggering with biochemical changes in the permeable cells. The Ca 2ϩ -dependent release of ϳ60% of the stored NE from permeable cells under optimal incubation conditions (12) indicated that the majority of LDCVs, probably the docked LDCVs, participate in exocytotic fusion reactions. Immunoblotting showed that a full cellular complement of the vesicle SNARE (synaptobrevin) and the plasma membrane SNAREs (syntaxin and SNAP-25) was preserved following cell permeabilization. To assess the state of vesicle docking, the association of vesicle proteins (synaptotagmin, synaptobrevin) with plasma membrane proteins (syntaxin, SNAP-25; Ͻ1% on purified LDCVs) was determined by immunoprecipitation with syntaxin and SNAP-25 antibodies (Fig. 3a). Immune complexes isolated from detergent extracts of permeable cells prior to priming contained a significant percentage (30 -40% and 10 -20%, respectively) of the cellular synaptobrevin and synaptotagmin, consistent with the presence of docked vesicles prior to ATPdependent priming. Immunoprecipitates also contained ␣/␤-SNAP (Fig. 3a) and NSF (Fig. 3b). Hence, complexes prior to priming resembled a recently described "docking and fusion particle" (18). While SNARE-containing complexes can form readily in concentrated brain extracts following detergent extraction (16), the complexes detected in the dilute detergent extracts of PC12 cells do not represent a post-extraction artifact since they vary in composition with cell incubation conditions, are unaffected by the volume of lysis buffer used, and are observed upon rapid SDS quenching (see below).
When permeable cells were incubated under priming conditions with MgATP, significant changes in SNARE protein complexes were observed consisting of decreases in both synaptotagmin and synaptobrevin (Fig. 3, a and c). In contrast, the association of syntaxin with SNAP-25 was maintained. Extensive decreases of NSF (Fig. 3b) and ␣/␤-SNAP (Fig. 3a) in the complexes were also evident following ATP-dependent priming. These alterations in SNARE protein complexes during priming resembled previously described ATP-dependent reactions catalyzed by NSF in detergent extracts or with recombinant SNARE proteins (9,15,19,20). An alternative means for detecting SNARE protein complexes was also used, which relies on the reported SDS-resistance of ternary complexes formed by SNARE proteins (16). Analysis of SDS extracts of permeable PC12 cells (Fig. 4a) revealed ϳ230-, ϳ110-, and ϳ75-kDa SDSresistant complexes that contained significant percentages of the cellular SNARE proteins (24%, 22%, and 38% of the synaptobrevin, SNAP-25, and syntaxin, respectively). Decreased amounts of the SDS-resistant complexes were evident following ATP-dependent priming (Fig. 4, a and b), confirming that SNARE protein complexes disassemble during priming.
NSF in the permeable cells was responsible for catalyzing SNARE complex rearrangements during priming since the reaction was inhibited by NEM and ATP␥S (Fig. 4c), agents that block NSF activity (9,5,21), and inhibition by NEM was overcome by inclusion of recombinant NSF and ␣-SNAP (Fig.  4c). Neither protein alone was effective (not shown). Following ATP-dependent priming, a portion (40 -70%) of the NSF and SNAP proteins present in permeable cells was released to the soluble fraction of the incubations. This release of NSF and SNAP from permeable cells depended upon MgATP and was inhibited by ATP␥S and NEM (Fig. 4d), consistent with recycling of these proteins following ATP hydrolysis (5,9). No additional release of NSF and SNAP (Fig. 4d) was detected following Ca 2ϩ triggering of LDCV fusion in ATP-primed cells, suggesting that NSF action was restricted to priming. This was also indicated by the lack of significant further changes in SNARE protein complexes during Ca 2ϩ triggering (Fig. 3a). 25-30% of NSF and SNAP is retained by cells following permeabilization and washing. This appears to be adequate to sustain late stages of the LDCV exocytotic pathway, since addition of recombinant NSF and ␣-SNAP did not stimulate ATP-dependent, Ca 2ϩ -activated NE secretion (not shown). While secretion is NEM-sensitive, inhibition by NEM was not overcome by NSF (and/or SNAP) addition presumably because other NEM-sensitive proteins are required. The further release of NSF and SNAP from permeable cells during ATP-dependent priming (see Fig. 4d), however, suggested that it may be feasible to render these proteins rate-limiting for regulated NE secretion. Priming is reversed in incubations that lack ATP and multiple cycles of priming, and its reversal can be sustained in sequential incubations (10). Hence, the permeable cells were incubated under priming conditions with MgATP and washed to remove eluted NSF/SNAP. The permeable cells were then incubated without MgATP to reverse priming. Finally, these were incubated under various conditions to determine the requirements for optimal priming. Priming in these pretreated permeable cells was entirely MgATP-dependent as anticipated (Fig. 2b). However, priming was now also stimulated by a combination of NSF and ␣-SNAP, whereas neither protein alone was effective. Cytosol, which lacks active NSF (22), also stimulated priming as anticipated due to its content of PEP proteins (10,13,14). Maximal priming was observed by addition of both NSF/SNAP and cytosolic PEP proteins, indicating that both polyphosphoinositide synthesis (14) and NSF-catalyzed reactions contribute to the MgATP-dependent priming of exocytosis. The effects of NSF and ␣-SNAP on secretion were restricted to priming, and no stimulation by NSF and/or ␣-SNAP was evident in Ca 2ϩ -dependent triggering reactions (not shown).
The discovery that NSF and SNAP proteins interact with a core complex of neuronal synaptic SNARE proteins suggested a role for NSF in regulated exocytosis (5), but the precise point of action of NSF in the regulated secretory pathway has not previously been established. In vitro biochemical studies demonstrated that a 20 S complex of SNAREs, NSF, and ␣/␤-SNAP was disassembled by the ATP-dependent activity of NSF (21), and it was suggested that NSF-catalyzed protein rearrangements drive membrane fusion (9). This model implied that the fusion step in regulated exocytosis is ATP-dependent and follows a Ca 2ϩ -dependent step (9). In contrast, our studies support an alternative model (23,24) that the execution point of NSF is during the reversible ATP-dependent activation step (priming) that can follow docking and that precedes Ca 2ϩactivated fusion (Fig. 4e). While our results imply that NSF is not directly involved in the final stages that lead to full fusion in the regulated exocytotic pathway, ATP-dependent catalysis of SNARE protein rearrangements by NSF may generate a FIG. 3. SNARE complex disassembly occurs during MgATP-dependent priming. Panel a, detergent extracts from incubated permeable cells were immunoprecipitated (IP) with syntaxin or SNAP-25 antibodies and immune complexes were analyzed by immunoblotting for synaptotagmin, syntaxin, SNAP-25, ␣/␤-SNAP, and synaptobrevin. Thirty-min priming incubations with or without 2 mM MgATP or 3-min triggering incubations with Ca 2ϩ were conducted. Syntaxin immunoprecipitation was essentially quantitative and immunoprecipitates contained 30%, 10%, and 28% of the cellular synaptobrevin, synaptotagmin and SNAP-25, respectively. SNAP-25 immunoprecipitation was quantitative and immunoprecipitates contained 40%, 20%, and 19% of the cellular synaptobrevin, synaptotagmin, and syntaxin, respectively. Panel b, presence of NSF in SNAP-25 immune complexes from permeable PC12 cells incubated 30 min with or without 2 mM MgATP. Panel c, histograms show mean (with ranges for 5-7 determinations) recovery of indicated co-immunoprecipitated proteins from 30-min incubations of permeable cells with MgATP relative to incubations without MgATP (ϭ 100%). Data were normalized to the measured recovery of the immunoprecipitated protein (SNAP-25 or syntaxin) in each experiment. prefusion intermediate possibly similar to that suggested for viral fusion mechanisms (25). NSF could catalyze an ATP-dependent conformational change in SNARE proteins (9,20,26) that alters the organization of the cytoplasmic leaflets of vesicle and plasma membrane bilayers. In constitutive fusion events, this intermediate may spontaneously resolve to full fusion. In regulated exocytosis, this intermediate would be arrested and require Ca 2ϩ to allow progression to full fusion. Physical evidence for a prefusion intermediate should be sought.