Novel Interactions of CAPS (Ca2+-dependent Activator Protein for Secretion) with the Three Neuronal SNARE Proteins Required for Vesicle Fusion*

CAPS (aka CADPS) is required for optimal vesicle exocytosis in neurons and endocrine cells where it functions to prime the exocytic machinery for Ca2+-triggered fusion. Fusion is mediated by trans complexes of the SNARE proteins VAMP-2, syntaxin-1, and SNAP-25 that bridge vesicle and plasma membrane. CAPS promotes SNARE complex formation on liposomes, but the SNARE binding properties of CAPS are unknown. The current work revealed that CAPS exhibits high affinity binding to syntaxin-1 and SNAP-25 and moderate affinity binding to VAMP-2. CAPS binding is specific for a subset of exocytic SNARE protein isoforms and requires membrane integration of the SNARE proteins. SNARE protein binding by CAPS is novel and mediated by interactions with the SNARE motifs in the three proteins. The C-terminal site for CAPS binding on syntaxin-1 does not overlap the Munc18-1 binding site and both proteins can co-reside on membrane-integrated syntaxin-1. As expected for a C-terminal binding site on syntaxin-1, CAPS stimulates SNARE-dependent liposome fusion with N-terminal truncated syntaxin-1 but exhibits impaired activity with C-terminal syntaxin-1 mutants. Overall the results suggest that SNARE complex formation promoted by CAPS may be mediated by direct interactions of CAPS with each of the three SNARE proteins required for vesicle exocytosis.

Peptide and neurotransmitter release from endocrine cells and neurons occurs by the regulated exocytic fusion of secretory vesicles with the plasma membrane (1). Membrane fusion is mediated by trans complexes of SNARE proteins that bridge the vesicle and plasma membrane to promote close membrane apposition and bilayer mixing (2). The SNARE complexes for endocrine and neuronal vesicle fusion consist of parallel bundles of four ␣-helical SNARE motifs with one helix contributed by the vesicle v-SNARE vesicle-associated membrane protein 2 (VAMP-2) 2 (aka synaptobrevin), one helix contributed by the plasma membrane t-SNARE syntaxin-1, and two helices contributed by the plasma membrane t-SNARE SNAP-25 (3). Based on a central layer residue (arginine or glutamine) and helix position in the SNARE bundle, these proteins are also classified as R-, Qa-, and QbQc-SNAREs, respectively (4). In vitro studies suggested a "zipper" model for the assembly of ternary SNARE complexes involving the pairing of syntaxin-1 (Qa) with SNAP-25 (QbQc) followed by the N-to C-terminal insertion of the R-SNARE VAMP-2 (5,6).
Vesicle exocytosis is a sequential multi-step process that involves the initial tethering of vesicles to the plasma membrane (7). Only a subset of membrane-attached vesicles rapidly fuse in response to Ca 2ϩ elevations and these are considered to be primed in steps that render the exocytic machinery fusioncompetent (8). Priming involves the disassembly of cis SNARE complexes and the assembly of loose trans SNARE complexes in advance of fusion triggering (9 -11). The in vivo pathway for trans SNARE complex assembly is unknown but may involve the zippering of VAMP-2 into syntaxin-1/SNAP-25 heterodimers as suggested from in vitro studies (12).
The assembly of SNARE complexes is catalyzed by essential accessory factors. At intracellular sites of membrane fusion in the secretory pathway, SNARE complex assembly is catalyzed by Sec-1/Munc18 proteins acting in concert with a heterogeneous class of tethering factors that directly bind SNARE proteins (13,14). For regulated vesicle exocytosis, Munc18-1 participates in promoting SNARE complex assembly (15) with other accessory proteins. Genetic and biochemical studies indicate that proteins in the Munc13/CAPS family function in the priming of vesicles in endocrine and neural cells (8). CAPS-1 is a multi-domain 1289 residue protein that reconstitutes regulated vesicle exocytosis in permeable neuroendocrine cells (16). Regions in the C-terminal half of CAPS exhibit homology to a corresponding C-terminal region of Munc13-1 that was reported to bind syntaxin-1 (17,18). CAPS contains a central pleckstrin homology (PH) domain that interacts with phosphatidylinositol 4,5-bisphosphate (PI(4,5)P 2 ) (19), a lipid that is required for priming in exocytosis (20,21). CAPS markedly accelerates SNARE-dependent liposome fusion in vitro when PI(4,5)P 2 is present in the syntaxin-1/SNAP-25-containing acceptor liposomes, which is dependent on the PH domain (22). The stimulation of liposome fusion by CAPS is preceded by the increased assembly of trans SNARE complexes on the liposomes (23).
A potential mechanism by which CAPS could promote SNARE complex formation would be through direct SNARE protein binding, however the SNARE binding properties of CAPS are unknown. The current work reveals that CAPS binds independently to each of the three SNARE proteins required for vesicle exocytosis. CAPS binds to the SNARE motifs in each of the SNARE proteins and exhibits specificity for a subset of exocytic SNARE protein isoforms. The functional significance of CAPS interactions with the SNARE (plus linker) domain of syntaxin-1 was shown in liposome fusion studies. We suggest that CAPS promotes the priming of vesicle exocytosis by driving trans SNARE complex formation through direct interactions with three SNARE motifs.
Proteoliposomes were also formed by the extrusion method (28). A lipid film containing 4.5 mol of DOPC:DOPS in an 85:15 mole ratio was resuspended in 300 l of reconstitution buffer. The mixture was vortexed for 5 min, subjected to 5 freeze-thaw cycles and extruded (MiniExtruder, Avanti Polar Lipids) for 33 passes through a 0.1 m Nuclepore tracketched polycarbonate membrane filter (Whatman) to produce 100 nm diameter proteoliposomes. The mixture was divided into 100-l aliquots, mixed with 200 l of SNARE proteins in elution buffer, incubated for 30 min at room temperature, and diluted with 300 l reconstitution buffer. The liposome mixture was dialyzed and separated by Accudenz gradient centrifugation as described above. Proteoliposomes made by extrusion had ϳ160 copies/liposome of syntaxin/SNAP-25 or ϳ360 copies/liposome of VAMP-2.
For cross-linking SNAP-25 to liposomes, extruded liposomes were made as described above with the following modifications. Lipid mixtures contained a DOPC:DOPS:MPB-PE (Avanti Polar Lipids) lipid mix with a 80:15:5 mole ratio. MPB-PE is a lipid moiety which contains a maleimido group that reacts with the thiol group of cysteines to form a stable thioether bond as described (29). MPB-PE was cross-linked to the native cysteines of SNAP-25-(1-206) and -(1-100) and a S115C mutation was used for SNAP-25-(101-206). 1 mM TCEP was substituted for 1 mM DTT in buffers used before crosslinking steps. SNAP-25 proteins were incubated with the liposome mixture for 1 h at room temperature and quenched with 1 mM ␤-mercaptoethanol for 10 min at room temperature. The syntaxin H3-(191-266)-Cys was cross-linked to liposomes in a similar manner.
Liposome Binding Assay-0.45 mM SNARE-containing liposomes were resuspended in reconstitution buffer without glycerol for a total volume of 75 l. CAPS protein was added at concentrations indicated in figure legends. After mixing, the reactions were incubated at room temperature for 30 min and then mixed with an equal volume of 80% (w/v) Accudenz and pipetted into the bottom of a Beckman Ultra-Clear TM centrifuge tube (5 ϫ 41 mm). The mixture was overlaid with 350 l of 30% (w/v) Accudenz, and reconstitution buffer without glycerol (20 l) was pipetted on top of the density gradient. The gradients were centrifuged at 45,000 rpm using a SW 50.1 rotor (Beckman) for 1-4 h at 4°C (maximum liposome recovery was achieved within 30 min of centrifugation). The top of the gradient 40 -60 l was collected and 10,000 DPM of the recovered liposomes were run on SDS-PAGE and analyzed by SYPRO Ruby (Invitrogen) or Western blotting. Western blots were probed with CAPS antibody, syntaxin-1 antibodies (HPC-1, Sigma-Aldrich; R31 from R. Jahn), VAMP-2 antibody (from R. Jahn), SNAP-25 antibody (polyclonal, Life Span Biosciences, Seattle, WA; monoclonal SMI-81, Covance, Madison, WI). Gels stained with SYPRO Ruby were analyzed on a Typhoon 9410 (Amersham Biosciences). For experiments displaying all fractions, 100 l of reconstitution buffer without glycerol was overlaid on top of the gradient and the gradients were divided into 8 or 10 equal samples of 60 or 75 l after centrifugation.
Lipid Mixing Fusion Assay-Lipid mixing fusion assays between fluorescent donor liposomes and non-fluorescent acceptor liposomes was performed as described (2). Lipid mixing was reported by the loss of FRET between fluorescent lipids (NBD-PE and Rh-PE) in VAMP-2 donor liposomes that occurs upon fusion with the non-fluorescent t-SNARE acceptor liposomes. The standard assay used 0.45 mM acceptor and 0.225 mM donor liposomes in a total volume of 75 l reconstitution buffer without glycerol. CAPS protein was added at a final concentration of 1 M. Negative controls were prepared for all conditions by substituting t-SNARE acceptor liposomes with protein-free (pf) liposomes to detect non-SNARE mediated lipidmixing. Reactions were assembled on ice and flick-mixed before addition to 96 well FluoroNunc plates (Nunc). Fusion was detected by measuring dequenching of NBD fluorescence (excitation, 465 nm; emission, 535 nm) every 90 s at 37°C in a Tecan Infinite F500 microplate spectrofluorometer (Tecan, Männedorf, Switzerland). After 2 h, fusion reactions were solubilized with 0.5% w/v dodecyl-D-maltoside and fluorescence readings were recorded for an additional 10 min. Lipid mixing results were quantified as previously described (22).
Surface Plasmon Resonance-Surface plasmon resonance was monitored with a BIACORE 2000 (Pharmacia Biosensor AB, Uppsala, Sweden) on a NTA sensor chip (Biacore AB, Uppsala, Sweden). The sensor chip has a carboxymethylated dex-tran matrix pre-immobilized with NTA (Qiagen). His-tagged CAPS was immobilized via Ni 2ϩ -NTA chelation, and proteins were examined for binding by flowing over the chip at a concentration of 20 M.

CAPS-1 Binds Membrane-integrated Syntaxin-1-CAPS-1
accelerates SNARE-dependent fusion between donor VAMP-2-containing and acceptor syntaxin-1/SNAP-25-containing liposomes (23). Fusion stimulated by CAPS is preceded by the formation of trans SNARE complexes that bridge the liposomes (23). These results suggested that CAPS may interact with one or more of the SNARE proteins but our previous studies failed to detect interactions of CAPS with truncated soluble SNARE proteins. It is increasingly apparent that full-length membraneintegrated SNARE proteins exhibit properties different from truncated soluble SNARE proteins (30,31). Thus, in the current work, we employed full-length membrane-integrated SNARE proteins. The SNARE proteins were reconstituted into PC/PS liposomes by dialysis or by direct detergent dilution methods (see "Experimental Procedures") at densities similar to those found in PC12 cells (23). There was stable incorporation of the SNARE proteins into liposomes by either method as indicated by their resistance to extraction by 0.1 M sodium bicarbonate, pH 11 buffer. CAPS binding to proteoliposomes was analyzed by separating liposome-bound from free CAPS by buoyant density flotation (Fig. 1A). CAPS bound to full-length syntaxin-1-containing liposomes whereas little binding to protein-free liposomes was detected at these liposome concentrations (Fig. 1B). Because the dissociation of bound CAPS was minimal during the liposome flotation, we were able to derive equilibrium binding constants. Binding isotherms indicated that CAPS bound to syntaxin-1 with a K d ϭ 220 nM and a B max ϭ 0.6 mol of CAPS per mol syntaxin-1 (Fig. 1C). Thus, CAPS exhibited high affinity interactions with membrane-integrated syntaxin-1.
Syntaxin-1 is comprised of several domains ( Fig. 2A) including an N-terminal triple helical domain (Habc), a helical SNARE domain (H3), and a C-terminal transmembrane domain (TMD). Munc13-1, which exhibits limited sequence homology to CAPS, was reported to bind the N-terminal domain of syntaxin-1 (17). By contrast, we found that CAPS interacted with C-terminal domains in syntaxin-1. CAPS bound to a membrane-integrated syntaxin-(183-288) protein that contained the H3, membrane-proximal linker and TMD domains but lacked N-terminal domains (Fig. 2B). We also determined whether the TMD participated in CAPS binding by cross-linking an H3-linker-containing protein, syntaxin-(191-266)-Cys lacking the TMD, to liposomes via a maleimide PE lipid anchor. CAPS was found to bind equally well to the cross-linked H3-linker protein and to the membrane-integrated H3-linker protein (Fig. 2C). These results revealed that the H3 plus membrane-proximal linker domain comprised the binding site on syntaxin-1 for CAPS.
To independently verify CAPS binding to the H3-linker domain, we conducted surface plasmon resonance (SPR) studies utilizing immobilized CAPS and soluble syntaxin-1 protein fragments. The results showed that CAPS interacted with soluble C-terminal syntaxin-(191-266) but not with soluble N-terminal syntaxin-(1-177) (Fig. 2D). CAPS also failed to interact with the soluble cytoplasmic domain syntaxin-(1-264) likely because it is in a "closed" conformation that occludes interactions with the H3 domain (32). The SPR results confirmed that CAPS interacts with the H3-linker domain of syntaxin-1. However, very high (20 M) syntaxin fragment concentrations were needed to detect binding and we were unable to obtain binding constants because of high backgrounds. Nonetheless, it appeared that CAPS exhibits very low affinity interactions with a soluble syntaxin H3-linker domain protein. This contrasts with the much higher affinity interaction of CAPS with membrane-integrated H3-linker and indicates that membrane is an important co-determinant for CAPS binding to syntaxin-1. PI(4,5)P 2 is essential for the CAPS promotion of trans SNARE complex formation and SNARE-dependent liposome fusion (22,23). However, CAPS binding to syntaxin-1 was not dependent on PI(4,5)P 2 inclusion in the liposomes.
To determine whether CAPS interactions with the H3-linker domain of syntaxin-1 are sufficient to promote fusion, we generated acceptor liposomes that contained a truncated syntaxin-(183-288) that lacks the N-terminal domain. This was reconstituted with SNAP-25 in PI(4,5)P 2 -containing liposomes to compare with liposomes containing full-length syntaxin plus SNAP-25 with PI(4,5)P 2 in a liposome fusion assay (Fig. 2E). The results showed that CAPS was able to drive fusion of VAMP-2-containing donor liposomes with syntaxin(183-288)/SNAP-25-containing acceptor liposomes (Fig. 2F). The studies of this section identify the H3-linker domain of syntaxin-1 as the binding site for CAPS and indicate that CAPS activity in SNARE-dependent liposome fusion does not require the N-terminal domain of syntaxin-1. Syntaxin Isoform Specificity of CAPS Binding-To assess the specificity of CAPS interactions with syntaxins, we prepared full-length syntaxins-1, -2, -3, -4, and -6 in PC/PS liposomes for binding studies. CAPS bound to syntaxins-1, -2, and -4 but not to syntaxins-3 and -6 ( Fig. 3A). Sequence comparisons (Fig. 3B) indicated that amino acids Tyr-243, Lys-260, and Lys-265 in syntaxin-1 were conserved in syntaxins-2 and -4 but not in syntaxins-3 or -6. Mutation of these residues in syntaxin-1 to their cognates in syntaxin-3 produced a chimeric protein (syntaxin-1 YKK/HQL) that exhibited impaired CAPS binding (Fig. 3C). To determine if CAPS binding to the H3-linker domain was also impaired by these mutations, we utilized SPR to test CAPS binding to mutant H3-linker domain proteins (Fig. 3D). CAPS binding to H3-linker domain proteins was strongly inhibited by the triple mutation of Tyr-243, Lys-260, and Lys-265 as well as by the double mutation at Lys-260 and Lys-265. Indeed, the single mutation at Lys-260 seemed to be responsible for the loss of binding (Fig. 3D). Inhibition of binding was also observed by mutation of Lys-252 to Ala, as is found in syntaxin-6, but not by mutation of more N-terminal residues in H3 (Fig. 3D). The results indicate that C-terminal sites in the H3 domain extending into the membrane-proximal linker region comprise part of the binding site for CAPS on syntaxin-1.
To determine whether reduced syntaxin-1 binding affected CAPS function, we prepared acceptor liposomes with full-length syntaxin-1(K260Q/K265L) plus SNAP-25 for testing in liposome fusion. Liposome fusion stimulated by CAPS was impaired in reactions containing the syntaxin-1(K260Q/K265L) mutant (Fig. 3E). That these mutations strongly decreased CAPS binding to syntaxin-1 (Figs. 3D) but did not fully impair CAPS-stimulated liposome fusion (Fig. 3E) may be attributed to CAPS interactions with SNAP-25 in the acceptor liposomes (see below). Overall, the results support the conclusion that CAPS interacts with a subset of exocytic syntaxin isoforms by binding to membrane-proximal regions of the H3-linker domain.
CAPS Binds SNAP-25-Binding studies with syntaxin-1-containing versus syntaxin-1/SNAP-25-containing liposomes showed increased CAPS binding to the latter (Fig. 4A). CAPS binding to membrane-integrated syntaxin-1/SNAP-25 was 2-fold greater than to membraneintegrated syntaxin-1 (Fig. 4B). This suggested that SNAP-25 might constitute a separate binding site for CAPS. To directly assess binding to SNAP-25 in the absence of syntaxin-1, we generated SNAP-25-containing liposomes by cross-linking the native cysteine residues of SNAP-25 to the maleimide lipid anchor MPB-PE. Liposome flotation studies revealed that CAPS bound to SNAP-25-containing liposomes to a greater extent than to protein-free liposomes (Fig. 4C). Binding isotherms, which were constructed after background subtraction of CAPS bound to protein-free liposomes, indicated an apparent K d of ϳ400 nM and a B max of 0.14 mol of CAPS per mol of SNAP-25 (Fig. 4D). Substoichiometric CAPS interactions with SNAP-25 may be due to the limited secondary structure of the helical domains of SNAP-25 in the absence of syntaxin-1 (33). We note that SPR studies with immobilized CAPS were unable to detect similar affinity interactions with soluble SNAP-25 (not shown). This indicates that membrane is an important co-determinant for CAPS binding to SNAP-25 as was observed for syntaxin-1 binding.

SNAP-25 is a QbQc-SNARE containing N-terminal (SN1)
and C-terminal (SN2) SNARE domains connected by a linker region (Fig. 5A). To determine which region of SNAP-25 bound CAPS, we generated N-terminal (SNAP-25-(1-100)) and C-terminal (SNAP-25-(101-206) S115C) proteins that we cross-linked to liposomes containing MPB-PE. CAPS bound to liposome-associated full-length SNAP-25 and to the N-terminal SNAP-25(1-100) fragment but exhibited very limited binding to the C-terminal SNAP-25-(101-206) fragment (Fig. 5B). Binding to the N-terminal SNAP-25-(1-100) fragment did not fully account for binding to full-length SNAP-25, but this may be due to reduced secondary structure. Given this uncertainty, we determined whether the C-terminal (SN2) helix of SNAP-25 participated in CAPS binding by cleaving full-length SNAP-25 with botulinum neurotoxin E protease, which eliminates a major portion of the SN2 SNARE domain. We found that CAPS binding to SNAP-25 was not altered by botulinum neurotoxin E cleavage (Fig. 5C). These results suggest that the SN1 region of SNAP-25 represents the main binding site for CAPS. Overall, the binding studies with Q-SNARE proteins indicate that CAPS independently interacts with the H3-linker helix of syntaxin-1 (Qa) and with the SN1 helix of SNAP-25 (Qb). These helices themselves interact in the formation of QaQbQc syntaxin-1/ SNAP-25 heterodimers (5). Because two binding sites for CAPS are preserved in heterodimers, the CAPS binding sites must reside on the faces of the H3 (Qa) and SN1 (Qb) helices that remain solvent-exposed in the heterodimer.
Competition for CAPS Binding to SNAREs-CAPS might utilize a single Q-SNARE-binding domain to interact with the H3 domain of syntaxin-1 or with the SN1 domain of SNAP-25.
Alternatively, CAPS may have distinct Qa binding and Qb binding domains that interact with syntaxin-1 or SNAP-25, respectively. We utilized competition binding studies to attempt to distinguish these alternatives. We first tested whether the soluble cytoplasmic domain of syntaxin-1-(1-264) competes with CAPS for binding to membrane-integrated syntaxin-1/ SNAP-25 heterodimeric SNARE complexes. As expected, soluble syntaxin-(1-264) competed poorly for binding (Fig. 6, A and B), which was consistent with the limited binding of soluble syntaxin-(1-264) to CAPS observed by SPR (Fig.  2D). By contrast, L165A/E166A syntaxin-(1-264), a mutationally "opened" form, was more effective in competition ( Fig. 6A and B) because its H3 domain was available to compete for CAPS binding. Consistent with this, the soluble syntaxin-(191-266) comprising the H3 plus linker domain was highly effective in its competition (Fig. 6, C  and D). In fact, complete inhibition of CAPS binding to membrane-integrated heterodimeric t-SNARE complexes was observed with the soluble H3-linker domain protein. That the soluble H3-linker protein fully competed CAPS binding to syntaxin-1/SNAP-25 complexes suggests that CAPS utilizes a single common binding domain to mediate interactions with Qaand Qb-SNARE motifs.
Additional competition binding studies with SNARE-binding proteins were conducted to probe the binding sites for CAPS on syntaxin-1 or syntaxin-1/SNAP-25 heterodimeric SNARE complexes. Munc18-1 exhibits multiple contact sites on soluble syntaxin-(1-265) (34) but interacts mainly with an N-terminal domain on membrane-integrated syntaxin-(1-288) (15). Munc18-1 failed to compete with CAPS for syntaxin-1 binding but instead enhanced binding about 2-fold (Fig. 6, E  and F). This result suggests that Munc18-1 may alter syntaxin-1 to promote increased access of CAPS to its C-terminal binding site. Moreover, CAPS and Munc18-1 were found to co-reside on syntaxin-1-containing liposomes (Fig. 6E), which is consistent with CAPS binding to C-terminal sites on syntaxin-1 that do not overlap with N-terminal sites for Munc18-1 binding.
Complexin-I interacts with heterodimeric syntaxin-1/ SNAP-25 complexes as well as with heterotrimeric SNARE complexes in which it occupies a C-terminal groove with contacts at syntaxin-1-(214 -232) and VAMP-2-(47-68) (35). We confirmed that complexin-I binds to heterodimeric SNARE complexes and found that it did not inhibit CAPS binding (Fig.  6, G and H). The lack of competition by complexin-I suggests that CAPS may occupy sites on syntaxin-1/SNAP-25 complexes that are more C-terminal than syntaxin-(214 -232). ␣-SNAP interacts with heterotrimeric SNARE complexes and with the H3 domain of syntaxin-1 (36). Although the binding sites for ␣-SNAP have not been precisely determined, it was proposed that ␣-SNAP binds to the C-terminal half of the helix bundle of heterotrimeric SNARE complexes (37). ␣-SNAP was found to very effectively inhibit CAPS binding to syntaxin-1/ SNAP-25 heterodimers (Fig. 6, G and H). This result supports the conclusion that CAPS binds C-terminal sites in t-SNARE heterodimers.
CAPS Also Binds VAMP-2-CAPS promotes the formation of trans SNARE complexes that bridge liposomes (23). This might result from CAPS interactions with Q-SNARE syntaxin-1/SNAP-25 heterodimers to promote favorable sites for the insertion of VAMP-2 to form trans complexes. Alternatively, CAPS might also interact with VAMP-2 to facilitate VAMP-2 binding to syntaxin-1/SNAP-25 heterodimers. We tested for possible CAPS-VAMP-2 interactions by conducting binding studies with VAMP-2-containing liposomes using buoyant density gradient flotation (Fig. 7A). At liposome concentrations and copy numbers similar to those used for syntaxin-1 or SNAP-25 binding, CAPS binding to VAMP-2 was limited. However, CAPS binding was evident as the concentration of VAMP-2-containing liposomes was increased and it exceeded the binding to protein-free liposomes (Fig. 7A). Binding isotherms constructed after background subtractions (Fig. 7, B and C) revealed that CAPS binding to membrane-integrated VAMP-2 was saturable and exhibited an apparent K d ϳ1.5 M (Fig. 7C). To determine the site on VAMP-2 that mediated CAPS binding, we prepared liposomes in which VAMP-2 was cleaved by botulinum neurotoxin D. These liposomes, which contained a membrane-inserted C-terminal VAMP-2-(60 -116) fragment, exhibited strongly reduced CAPS binding (Fig.  7D). This result indicates that CAPS binding requires part of the 59 residue N-terminal segment of VAMP-2.

DISCUSSION
CAPS is essential for optimal Ca 2ϩ -triggered vesicle exocytosis where it functions in priming reactions that precede triggered membrane fusion (19,38,39). Priming involves the assembly of trans SNARE complexes but the assembly pathway utilized and the accessory factors that catalyze it remain uncertain. CAPS promotes SNARE-dependent liposome fusion in vitro (22,23), which may represent priming reactions that proceed into fusion in the absence of downstream regulators such as synaptotagmin and complexin. CAPS stimulates the formation of SNARE complexes that bridge the liposomes (23) and the current work was undertaken to identify direct CAPS interactions with SNARE proteins. The major conclusion from this work is that CAPS binds to each of the SNARE proteins that are required for regulated vesicle exocytosis. The results suggest the possibility that CAPS may promote trans SNARE complex assembly through direct binding interactions with each of the three SNARE proteins. This possibility will need to be tested in future work.
General features of vesicle targeting to and fusion with acceptor membranes have been characterized at many trafficking steps in the secretory pathway. The core machinery for vesicle targeting and fusion consists of members from conserved families of Rab, Sec-1/Munc18, and SNARE proteins (40). At many membrane trafficking stations, the initial interaction of vesicles with target membranes is mediated by diverse stagespecific tethering factors or multi-subunit tethering complexes (14). These tethering complexes are frequently regulated by vesicle Rab proteins and they commonly interact with, or include, Sec-1/Munc18 family members. Some tethering factors that initially link vesicle and target membranes have been shown to directly interact with SNARE proteins and to catalyze trans SNARE complex formation. In ER-Golgi trafficking, Rab1-regulated p115 bridges membranes and promotes trans SNARE complex formation through direct interactions with the Q-SNARE syntaxin-5 and the R-SNARE GOS-28 (13). p115 also interacts with other SNARE proteins involved in ER-Golgi trafficking via its SNARE motif-like C-terminal coiled-coil (13).
For vesicle fusion in the regulated secretory pathway, stage-specific tethering complexes have not been identified. Mechanisms for the attachment of synaptic or densecore vesicles to the plasma membrane are poorly understood (7). However, CAPS and Munc13 proteins, which are both essential for regulated vesicle exocytosis at a stage following vesicle attachment, may be the functional homologues of the stage-specific tethering factors that prime membrane fusion. Consistent with this, C-terminal domains shared between CAPS and Munc13 proteins were reported to exhibit weak homologies to subunits of several tethering factors (41).
The SNARE protein binding properties of Munc13 proteins have not been fully determined. It was originally proposed that a C-terminal domain of Munc13-1 interacted with the N-terminal domain of syntaxin-1 (17), but this was not confirmed in biochemical studies (42). Instead, in direct binding studies, a C-terminal domain of Munc13-1 was found to interact with heterodimeric and heterotrimeric SNARE complexes but interactions with individual SNARE proteins was not evident (45,48). By contrast, the current work revealed direct interactions between CAPS and each of the individual SNARE proteins required for regulated vesicle exocytosis.
A novel and important feature of CAPS was revealed in finding that CAPS interacts with the SNARE motifs in the two Q-SNARE proteins. These comprised the H3 SNARE plus linker domain of syntaxin-1 and the SN1 SNARE domain of SNAP-25. High affinity CAPS binding required the membrane-integration or membrane-tethering of these SNARE proteins, whereas only very low affinity interactions occurred with soluble versions of the Q-SNARE proteins. In addition to the structural differences in membrane-integrated SNAREs responsible for this difference (31), it is likely that CAPS participates in forming a ternary CAPS-SNARE-membrane complex. The sites identified for CAPS binding in syntaxin-1 are membrane-proximal. The linker domain in syn- taxin-1 forms part of a continuous helix with the H3 domain (43), and residues required for CAPS binding are on the same face of the H3 domain (Lys-252) and linker (Lys-260) close to the membrane. Binding of the SN1 domain of SNAP-25 did not prevent CAPS interactions with syntaxin-1 but rather contributed an additional binding site for CAPS in heterodimeric Q-SNARE complexes (Fig. 4). Liposome fusion promoted by CAPS was previously shown to require SNAP-25 (23), and the current work showed that the C-terminal H3-linker domain of syntaxin-1 is sufficient for CAPS stimulation of fusion (Fig. 2). C-terminal mutations in syntaxin-1 impair CAPS function in fusion (Fig. 3), which is consistent with binding studies that show direct H3-linker binding by CAPS and strong binding competition by ␣-SNAP. These results for CAPS are different from but complementary to results for Munc18-1 that showed stimulation of SNARE-dependent liposome fusion requiring the N terminus of syntaxin-1 (15).
The finding that CAPS interacts with SNARE motifs in both syntaxin-1 and SNAP-25 suggests several potential novel mechanisms for how CAPS may participate in SNARE complex formation during priming. Because CAPS can oligomerize (16,44) and may utilize a common Qa-and Qb-binding site on each monomer (Fig. 6D), CAPS could function in several ways to regulate the Q-SNARE proteins syntaxin-1 and SNAP-25. CAPS might promote Q-SNARE heterodimer formation through Qa and Qb binding by each monomer in a CAPS oligomer. This is unlikely to be the key rate-limiting step promoted by CAPS because CAPS stimulates SNARE-dependent fusion with proteoliposomes that contain pre-formed Q-SNARE heterodimers (23). However, because the helices in Q-SNARE heterodimers may improperly align in an antiparallel manner, oligomeric CAPS could stabilize parallelaligned heterodimers as has been suggested for several accessory proteins including Munc13 (45). Alternatively, by interacting with the membrane-proximal regions of helices on syntaxin-1 and SNAP-25 in SNARE heterodimers, CAPS oligomers could function to cluster or organize Q-SNARE heterodimers at fusion sites for optimal insertion of vesicle VAMP-2 (22). This possibility could account for the observed ability of CAPS to promote liposome fusion at extremely low Q-SNARE densities (23). These proposed direct actions of CAPS on Q-SNAREs will need to be explicitly tested.
In accord with recent views that Munc18-1 regulates N-terminal domains of syntaxin-1 (15,46), Munc18-1 failed to compete with CAPS for binding to syntaxin-1 but it unexpectedly enhanced CAPS interactions with syntaxin-1 (Fig. 6F). This finding suggests that Munc18-1 may alter the conformation of syntaxin-1 to enable CAPS binding. Alternatively, CAPS and Munc18-1 might interact so that Munc18-1 binding to syntaxin-1 would provide an additional CAPS binding site. Nonetheless, that both Munc18-1 and CAPS were observed to co-reside on syntaxin-1-containing liposomes (Fig. 6E) is consistent with the assignments of N-terminal and C-terminal binding sites for Munc18-1 and CAPS, respectively. Because our liposome flotation studies of SNARE binding did not achieve complete stoichiometry, we cannot rule out possible binding of Munc18-1 and CAPS to different syntaxin-1-containing liposomes; however, it was clear that there is at least some overlap because Munc18-1 strongly affected CAPS binding (Fig. 6F). Overall, the results indicate that binding sites for Munc18-1 and CAPS on Q-SNARE heterodimers are complementary.
CAPS also interacts with the membrane-integrated R-SNARE VAMP-2 (Fig. 7). VAMP-2 is not completely structured in isolation (31) and it is unclear whether the reduced affinity and decreased stoichiometry for CAPS binding reflects the partly unfolded state for VAMP-2 on liposomes. CAPS interactions with VAMP-2, unlike those for syntaxin-1 C-terminal domain binding, were with N-terminal domains of VAMP-2 as indicated by the ability of botulinum neurotoxin D to abolish binding. A "zipper" model for heterotrimeric SNARE complex assembly proposes the N-to C-terminal insertion of Representative results for three experiments are shown. B, indicated concentrations of CAPS were incubated with VAMP-2-containing or protein-free liposomes and separated by gradient centrifugation followed by SDS-PAGE and SYPROா Ruby staining. C, CAPS bound to protein-free liposomes was subtracted from CAPS bound to VAMP-2 liposomes and the data (mean Ϯ S.E., n ϭ 3) were fit to a non-linear regression curve (B max ϭ 0.04, K d ϭ 1.49 M, R 2 ϭ 0.903). D, 1 M CAPS was incubated with protein-free or VAMP-2-containing liposomes that were untreated or treated with botulinum neurotoxin D. Bound fractions from gradients were immunoblotted for CAPS and VAMP-2. Representative results for three experiments are shown.
VAMP-2 into syntaxin-1/SNAP-25 heterodimers (5,6). This would imply an N-terminal nucleation sequence on VAMP-2 that is initially integrated into the SNARE helix bundle. Through its N-terminal VAMP-2 interactions, CAPS bound to heterodimeric Q-SNARE complexes could be involved in catalyzing the zippering of VAMP-2 into recipient complexes.
CAPS is the only protein essential for regulated vesicle exocytosis that has been shown to exhibit direct independent interactions with each of the SNARE proteins required for vesicle fusion. However, recent studies have demonstrated that Munc18-1, in addition to its interactions with Q-SNARE heterodimers ( Fig. 6 and Ref. 45), also interacts with VAMP-2 (47). The binding site for Munc18-1 on VAMP-2 is C-terminal whereas that for CAPS is N-terminal. Thus, studies on Munc18-1 and CAPS now indicate that their binding sites on Q-SNARE heterodimers and on the R-SNARE VAMP-2 are complementary. This suggests a possible mechanism for priming in which both factors bound to syntaxin-1/SNAP-25 heterodimers play a catalytic role in the insertion of VAMP-2 to form trans SNARE complexes. CAPS and Munc18-1 might act cooperatively with CAPS assisting in the initial nucleation of the N terminus of VAMP-2 into Q-SNARE recipient complexes followed by C-terminal "zippering" assisted by Munc18-1.
In summary, we describe for the first time the unique interactions of CAPS with each of the SNARE proteins required for regulated vesicle exocytosis. CAPS binding to Q-SNAREs is mediated by C-terminal, membrane-proximal interactions with two of three SNARE motifs in the heterodimeric complex. By contrast, CAPS binds N-terminal portions of the R-SNARE VAMP-2. These findings provide potential insight into mechanisms by which CAPS could promote trans SNARE complex formation for priming vesicle exocytosis.