Selective interaction of complexin with the neuronal SNARE complex. Determination of the binding regions.

Complexins are evolutionarily conserved proteins that specifically bind to soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE) complexes and thus may regulate SNARE function. Using purified proteins, we have performed a detailed analysis of the structure of complexin and of its interaction with SNARE proteins. NMR spectroscopy revealed that isolated complexins have no tertiary structure but contain an unusual alpha-helical middle domain of approximately 58 amino acids that overlaps with the most highly conserved region of the molecules. Complexins form a stable stoichiometric complex with the central domain of the ternary SNARE complex, whereas no binding was observed to monomeric SNAREs. Using a combination of limited proteolysis, deletion mutagenesis, and NMR spectroscopy, we found that the helical middle region of complexin is responsible for binding to the SNARE complex. Binding was highly sensitive to substitution of syntaxin 1 or synaptobrevin 2 with other SNARE homologs but less sensitive to substitution of SNAP-25. In addition, a stretch of 12 amino acids in the middle of the SNARE motif of syntaxin 1A was able to confer binding activity to the non-binding relative syntaxin 4. Furthermore, disassembly of ternary complexes is not affected by complexins. We conclude that complexins are specific ligands of the neuronal core complex that bind with a central alpha-helical domain, probably to the middle of the surface groove formed by synaptobrevin and syntaxin. Complexins may regulate the function of ternary complexes and control membrane fusion through this interaction.


Introduction
Intracellular membrane fusion events are mediated by conserved sets of membranebound proteins referred to as SNAREs (acronym for soluble N-ethylmaleimide-sensitive factor attachment protein receptors) (1). SNAREs comprise a family of proteins that is distinguished by the presence of the SNARE motif, a homologous stretch of approximately 60 amino acids that is usually localized adjacent to the membrane anchor domains (2)(3)(4). Best characterized are SNARE proteins functioning in neuronal exocytosis. They include the synaptic vesicle protein synaptobrevin (also referred to as VAMP), and the synaptic plasma membrane proteins SNAP-25 and syntaxin 1. These proteins form a stable ternary complex, the core complex, that can be reversibly disassembled by the ATPase NSF (N-ethylmaleimide sensitive factor) and additional cofactors termed SNAPs (for soluble NSF attachment proteins) (5).
Studies performed largely on recombinant proteins in solution have revealed a detailed picture of SNARE complex assembly. Of the monomeric SNAREs, only syntaxin 1 is partially α-helical whereas both SNAP-25 and synaptobrevin are largely unstructured. Upon assembly, a dramatic increase in α-helical content is observed suggesting major conformational changes (6,7). Interactions in the complex are largely confined to the SNARE motifs. X-ray crystallography revealed that the assembled core complex consists of an elongated bundle of four intertwined α-helices, each representing a single SNARE motif. Two of the helices are contributed by SNAP-25, and one each by syntaxin 1 and synaptobrevin (8).
The core complex is resistant to proteolysis and to denaturation by the detergent SDS (unless heated) (9). Furthermore, it is unusually heat-stable and only denatures above 80°C (10).
These and other findings suggested that SNARE complex assembly is the essential step in initiating membrane fusion. According to this hypothesis, assembly of SNAREs localized on by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 5 the two opposing membranes form a tight connection between the membranes. In the resulting "trans"-complexes, the SNARE-motifs are partially assembled in a helical bundle, while the transmembrane domains are anchored in the still separated membranes. The energy released during assembly may, at least partially, overcome the barrier separating the membranes destined to fuse. During fusion, the SNAREs relax into a complex in which all transmembrane domains are aligned in parallel ("cis"-complex) and which then is re-energized by disassembly involving NSF and SNAPs (11).
In search for molecules regulating the function of neuronal SNAREs, several proteins have been identified that bind specifically to individual SNAREs or to SNARE core complexes (4). Among these are the complexins, two related proteins of approximately 15 kDa. Complexins, also named synaphins, were identified as polar and soluble proteins that are associated with the neuronal SNARE complex in membrane extracts (12)(13)(14). In vitro, complexins bind to assembled SNAREs and compete with α-SNAP for binding (12) suggesting that they are involved in the regulation of the SNARE assembly-disassembly cycle. Both complexins are colocalized with neuronal SNAREs in presynaptic nerve terminals although cytoplasmic pools are also present, particularly during early stages of neuronal maturation (15).
The precise function of the complexins remains to be elucidated. When the intracellular concentration of complexin was increased by microinjection or by overexpression, a moderate reduction of neurotransmitter release was observed. Injection of an anti-complexin antibody increased transmitter release suggesting that complexins may function as negative regulators (15,16). Disruption of the complexin II gene in mice resulted in viable mice with all parameters of synaptic transmission being normal except LTP (17).
However, disruption of both complexin genes causes massive neurological dysfunctions (N. by guest on March 24, 2020 9 conserved between rat syntaxins 1-3 suggesting that this discrepancy is probably due to a sequencing error.

Protein Expression and Purification
All recombinant proteins were expressed in E. coli BL21 (DE3) cells according to standard protocols with the exception of GST-complexin I for NMR studies. For this purpose cells were grown at 37 o C in M9 minimal media with 4 g/L glucose, 1 g/L NH 4 Cl, 1 mM MgSO 4 and 50 µg/ml ampicillin. Uniform 15 N and 13 C-labeling was achieved using 15 NH 4 Cl and 13 C 6glucose (Isotec) as the sole nitrogen and carbon sources, respectively. Cells were induced at 0.6 O.D. with 0.8 mM IPTG. His 6 -tagged proteins were affinity purified using Ni-NTA-Agarose (Qiagen) and GST-fusion proteins by Glutathion-Sepharose (Amersham/Pharmacia).
Usually the tags were cleaved by thrombin after elution from the beads. However, if used for precipitation in binding experiments (see below) tags remained attached to the proteins.
All SNARE complexes were purified using Mono-Q column after overnight assembly of the purified monomers (10,21). For technical reasons constructs for syntaxin 1A and SNAP-25A 10 derived from different expression vectors were used to assemble ternary complexes: syntaxin (120-206) (10). Minimal core complex with bound complexin was assembled in excess amounts of complexin II and purified by size-exclusion chromatography using a Superdex 200 HiLoad 16/60-column (Amersham/Pharmacia).
All protein concentrations were determined by the Bradford assay (27).

NMR Spectroscopy
All NMR spectra were acquired at 25ºC on a Varian Unity 500 spectrometer, except the 1 H-the F3, F2 and F1 dimensions, with the number of scans per FID and the total measurement time indicated in parenthesis, were: 1 H- 15

Binding of complexins to SNAREs and SNARE complexes
For the binding assays, purified recombinant proteins or preformed SNARE complexes were used that carried either a His 6 -tag, a GST-tag, or were unmodified. In some experiments, rat brain cytosol (obtained by subjecting a homogenate, generated in 4 mM Hepes-NaOH, pH 7.3, 320 mM sucrose, to centrifugation at 12,000 x g av (Fig. 4B) or 90,000 x g av (Fig. 9C)) was used as a source of native complexins. Unless indicated otherwise, binding experiments were carried out in incubation buffer (20 mM Tris, pH 7.4, 100 mM NaCl, 1 mM DTT, and 0.1% (v/v) Triton X-100). The incubation buffer was complemented with either 0.1% (w/v) BSA and 1 mM EDTA (GST-fusion proteins) or 20 mM imidazole (His 6 -fusion proteins).
Incubations were performed between 1-1.5 h at room temperature on a rotator. For immunoprecipitations, anti-synaptobrevin antibody (Cl 69.1 (31)) was then added for 1 h at 4°C. Complexes were isolated by affinity adsorption, using Glutathion-Sepharose beads, Ni-12 approximately 1 h. The beads were then washed 4-5 times in incubation buffer except that Triton X-100 and BSA were omitted and imidazole was lowered to 8 mM. Beads were analyzed by SDS-PAGE and Coomassie Blue staining. Where indicated, complexins were detected by immunoblotting using polyclonal antibody 942 (Synaptic Systems, Göttingen, Germany, 1:2000) against the C-terminus of complexin I and II (amino acid residues 122-134, rat sequence) and Enhanced Chemiluminescence (ECL, NEN) on a Fujifilm LAS-1000 system.

Binding to SNAREs reconstituted in proteoliposomes
Proteoliposomes with reconstituted full-length synaptobrevin or ternary complex with transmembrane regions were prepared as perviously described (20). Liposomes adjusted to 0.4 µM each of synaptobrevin 2 and ternary complex, respectively, were each incubated with 0.8 µM complexin I or II or both for 1 h at room temperature in standard buffer. Liposomes were then immunoisolated by incubating in monoclonal antibody Cl 69.1 (5 µl ascites) for 1 h at room temperature on a rotator, followed addition of Protein A-Sepharose and a further incubation for 1 h. After sedimentation and removal of the supernatant containing unbound material, the beads were washed 4 times with 0.5% (v/v) Triton X-100-containing standard buffer. Fifty percent of the sepharose-bound material were analyzed by immunoblot using the antibody 942.

Other Methods
For limited proteolysis, purified complexes (8 µM) were incubated in 1% (w/w) of proteinase K at 25°C in standard buffer. At the indicated times, proteolysis was stopped by adding sample buffer without SDS containing 2 mM PMSF and chilling on ice. All samples were analyzed by nondenaturing PAGE and Coomassie Blue staining. Multi-angle laser light scattering (MALLS) and circular dichroism (CD) spectroscopy were performed as described earlier (21).

Complexin contains a conserved α-helical middle region but lacks a tertiary structure
In vertebrates, two isoforms of complexin were described that differ only by a few amino acid substitutions and a very high degree of homology was found between distant vertebrate species. Since functionally important domains are usually more highly conserved than flanking regions, we performed database searches to identify invertebrate orthologs of complexins, and to compare them with the vertebrate sequences. As shown in Fig. 1, sequences probably representing complexins were found both in D. melanogaster (full length) and C. elegans (partial). Sequence alignments revealed a high degree of overall homology between vertebrates and invertebrates, which, however, is still significantly lower than among vertebrates. The highest degree of conservation was detected in a region at the middle of the sequence spanning approximately residues 34 to 77 (numbering according to the rat sequence).
The structure of purified recombinant rat complexin I in solution was analyzed using nuclear magnetic resonance (NMR) and circular dichroism (CD) spectroscopy. The 1 H-15 N heteronuclear single quantum correlation (HSQC) spectrum of complexin (Fig. 2) exhibits a poor dispersion of 1 H chemical shifts that suggests a lack of tertiary structure. In contrast, the CD spectrum of complexin contains a double minimum at 206 and 222 nm (data not shown), and is characteristic for a mixture of α-helical and random coil conformations. To locate the α-helical region, we used multidimensional NMR techniques. The severe overlap of the 1 H-NOESY-HSQC experiments allowed us to obtain nearly complete (97.1%) sequential assignments (Fig. 2).

The observation of numerous sequential NH/NH and medium range nuclear
Overhauser effects (NOEs) (Fig. 3A) confirmed that a substantial part of the complexin sequence forms an α-helical structure. The medium range NOEs are most abundant between residues 29-86, indicating that the α-helical content is highest in this region. This conclusion is corroborated and refined by plots of Hα and Cα conformational shifts versus the amino acid sequence of complexin ( Fig. 3B, solid diamonds). These plots show that a stable α-helix is formed from residues 29 to 64 while residues 65 to 86 contain a substantial but lower population of α-helix. The α-helical region overlaps with the conserved middle region identified by sequence comparison (Fig. 1). Other regions of the molecule may sample helical conformations only occasionally. Overall, the NMR results indicate that complexin is 35-40% α-helical.
A detailed analysis of all NOE data failed to reveal any long ranges NOEs, confirming the absence of tertiary structure. The presence of a stable α-helix in residues 29-64 is surprising since formation of stable secondary structure in the absence of tertiary interactions is unusual. However, the presence of an α-helix in this region agrees with theoretical predictions (Fig. 3B). Furthermore, the α-helix is not stabilized by oligomerization.
Complexin is monomeric when analyzed by size exclusion chromatography combined with multi-angle laser light scattering (MALLS, data not shown) and, furthermore, the 1 H-15 N HSQC spectrum does not change over a wide concentration range (0.01-2 mM).
The α-helical middle region of complexin stoichiometrically binds to the SNARE core complex Next we investigated the binding of complexin to individual SNAREs and SNARE complexes using purified recombinant proteins and protein fragments. Unless stated otherwise, all SNAREs were expressed in bacteria without their transmembrane region. First, the binding of syntaxin 1, SNAP-25, synaptobrevin, and of preassembled ternary SNARE complex to His 6 -tagged complexin II was investigated. Only the ternary complex associated with complexin in significant amounts under our experimental conditions (Fig. 4A). A similar result was obtained when GST-tagged SNAREs and GST-tagged ternary complex were incubated with rat brain cytosol as a source of native complexins (Fig. 4B). Both complexin I and complexin II bound to the ternary complex whereas no binding was observed to monomeric SNAREs. Together, these results agree with previous findings which showed that complexins preferentially bind to assembled SNARE complexes (12).
To confirm that complexin also binds to the ternary complex when it is incorporated into a membrane, purified synaptobrevin and ternary complex (including transmembrane regions) were reconstituted into proteoliposomes by detergent removal. Binding of both complexin I and complexin II was determined by immunoisolation of the proteoliposomes with the aid of an anti-synaptobrevin monoclonal antibody. As shown in Fig. 4C, liposomes containing the ternary complex bound each of the complexins while no binding was observed to synaptobrevin-containing liposomes. When equal amounts of both complexin isoforms were included in the assay, preferential binding of complexin I was observed.
Next, we measured the stoichiometry of complexin binding. Preliminary experiments revealed that the association between complexin and the ternary complex is sufficiently stable by guest on March 24, 2020 http://www.jbc.org/ Downloaded from to allow for its separation from both free ternary complex and the uncomplexed proteins by means of nondenaturing PAGE. Increasing concentrations of complexin II were incubated with a fixed concentration of purified ternary complex. As shown in Fig. 5A, a gradual shift from free ternary complex to complexin-bound ternary complex was observed until all free ternary complex disappeared. When more complexin was added, no further binding was observed, and unbound complexin accumulated. At equal concentrations of complexin and ternary complex, only bound complexin was detectable. We conclude that the stoichiometry between complexin and the ternary complex is 1:1.
As discussed in the Introduction, assembly of SNAREs into core complexes is confined to a region which encompasses the SNARE motifs. We have shown previously that a minimal core complex can be formed from recombinant fragments that has similar properties to the non-truncated complex with respect to assembly, disassembly, and stability (10). Since complexin binds to the ternary complex with strong preference over the individual proteins, we examined whether binding is confined to the minimal core complex. Minimal core complex was generated and purified as described previously (10), and complexin binding was analyzed as described above. As shown in Fig. 5B, complexin bound to this minimal complex with similar efficiency and with a 1:1-stoichiometry.
The stability of complexin binding to the minimal core prompted us to purify this complex for further structural investigations. After purification, the complex was subjected to time-dependent proteolysis using proteinase K, a protease with broad substrate specificity, followed by analysis with non-denaturing PAGE. As shown in Fig. 6, a gradual shift of the complex to higher mobilities was observed. Since the minimal core is largely resistant to proteolysis (Fig. 6, see also (10,33)), this shift is mostly due to truncation of complexin. SDS-PAGE revealed several complexin fragments that were further analyzed by matrix-assisted To further define the complexin binding region, we generated a series of deletion mutants of complexin II using the proteolysis experiment as a guide. As shown in Fig. 7, progressive deletion at both C-and N-terminal ends did not abolish binding, with the smallest binding fragment encompassing residues 41-97. In contrast, no binding was observed when the protein was "cut" in the middle, i.e. with the C-and N-terminal halves of complexin. In addition, no binding was observed when both fragments were added together (data not shown).
The data described so far suggest that the binding region of complexin corresponds approximately to the conserved α-helical region characterized above. To confirm and extend these findings, we used NMR spectroscopy. The changes in the 1 H- 15 (Fig. 8A). However, the residues corresponding to these cross-peaks are not clustered in any specific region but rather are spread throughout the complexin sequence. These results indicate that the observed interaction is not specific.
Next we acquired 1 H-15 N HSQC spectra of 15   Complexin binding is sensitive to exchange of syntaxin and synaptobrevin As discussed above, the core domain of the SNARE complex consists of a bundle of four twisted α-helices that are connected by leucine-zipper like layers of interacting hydrophobic amino acids in the center of the bundle. Mapping SNARE sequences on the crystal structure revealed that the interacting amino acids are more conserved than the residues exposed on the surface (3), explaining why SNARE complexes form promiscuously between different SNARE proteins (21,36). However, substitution of one SNARE for another will usually result in a profound alteration of the surface pattern which thus distinguishes different SNARE complexes from each other. Since complexin binds to the surface of the assembled core domain and most likely does not have access to its hydrophobic core, it is conceivable that complexin binding is more sensitive to SNARE substitution than SNARE complex formation.
To examine whether substitution of SNAREs affects binding of complexins, we generated mixed SNARE complexes in which one of the neuronal SNAREs was substituted by a different family member. All mixed complexes were purified, are stable, and can be disassembled by the ATPase NSF (21,36). As shown in Fig. 9A, replacement of synaptobrevin with the endosomal relative endobrevin abolished binding of complexins. In contrast, binding was largely preserved when SNAP-25 was replaced by its distant relative SNAP-29 (Fig. 9B). Since SNAP-25 and SNAP-29 are only 19.9% similar to each other (Clustal method with PAM250 residue weight table), and the conserved residues are mostly confined to the hydrophobic core of the SNARE complex, this result indicates that the SNAP-25 side chains contribute less to complexin binding. Finally, we replaced syntaxin 1 with syntaxins 2, 3, and 4. Complexins bound with equal efficiency to the syntaxin 3 containing complexes (Fig. 9C) which agrees with previous observations suggesting an association of 21 complexin with syntaxin 3, synaptobrevin and SNAP-25 in the retina (37). Replacement by syntaxin 2 reduced binding, with a preference for complexin II, whereas replacement by syntaxin 4 abolished binding.
In contrast to synaptobrevin/endobrevin and SNAP-25/SNAP-29, the SNARE motifs of syntaxins 1-4 display a relatively high degree of homology. To narrow down the sequence responsible for the observed differences in complexin binding, we compared syntaxins 1 and syntaxin 4 in order to identify the regions with the lowest degree of homology between the surface-exposed residues. The most conspicuous stretch includes residues 214-225 (numbering of syntaxin 1). To test whether this stretch is involved in complexin binding, we generated chimeric proteins by swapping these stretches of 12 residues between syntaxin 1 and syntaxin 4. The chimeras were then used to form ternary complexes with synaptobrevin and SNAP-25 that were purified and used in complexin binding assays. As shown in Fig. 10, the complex formed with the chimera of syntaxin 1 containing the stretch of syntaxin 4 bound significantly less complexins than the unmodified complex. Conversely, the syntaxin 4 chimera containing the stretch of syntaxin 1 exhibited partially restored complexin binding.
These findings show that this stretch plays a key role in defining the binding site for complexins although it is most probable that flanking regions of the molecule are also involved.

Complexin does not affect disassembly of the SNARE complex
In the final series of experiments, we investigated whether complexin has an influence on the disassembly of preformed SNARE complexes by NSF and α-SNAP. The minimal core of the neuronal SNARE complex was purified either with or without bound complexin II. For disassembly, purified α-SNAP and NSF were added in the presence of ATP. Since the by guest on March 24, 2020 http://www.jbc.org/ Downloaded from 22 minimal core complex is stable during SDS-PAGE, the appearance of the monomeric SNARE components is indicative of disassembly. As shown in Fig. 11A, disassembly was observed independently of whether complexin was bound or not. The minimal core complex was also disassembled when complexin was added in a 20fold excess over α-SNAP even when the incubation time was reduced to 5 min (Fig. 11B). Furthermore, no influence of complexin on disassembly was observed when the non-truncated versions of the SNARE-proteins were used (data not shown). We also performed preliminary experiments on isolated synaptic vesicles to explore whether complexin has an effect on disassembly of native SNARE complexes.
However, no changes were found in the presence of complexins (data not shown).

Discussion
Complexin is a small and hydrophilic protein that was identified by its ability to bind to the neuronal SNARE complex (12,13). In this study we have used a combination of approaches to characterize this interaction in detail. Our results show that complexin binds stoichiometrically to the ternary SNARE complex whereas it exhibits no binding to the individual SNAREs. Free complexin lacks a tertiary structure but contains a conserved αhelical domain at the center of its sequence which mediates binding. Furthermore, complexin is able to discriminate between different SNARE complexes. Substitution of individual SNAREs and site-directed mutagenesis suggest that complexin binds predominantly to the groove formed by syntaxin and synaptobrevin at the surface of the core complex.
The central α-helix of complexin was shown to be responsible for binding to the SNARE complex by means of three independent approaches, limited proteolysis, deletion mutagenesis, and NMR spectroscopy. Thus, the binding region of complexin corresponds to the structured part of the molecule. Furthermore, we were unable to detect signs for major conformational changes of either complexin or the core complex during the binding reaction.
These features differ from the assembly of SNARE motifs that are mostly unstructured as monomers and that undergo massive structural changes upon complex formation (6,7). The binding domain of complexin contains more than 60% charged residues. Since the surface of the neuronal core complex is also highly polar (8) it is likely that polar and electrostatic interactions are primarily responsible for binding.
Although the precise nature of complexin binding to the core complex remains to be established, several conclusions can be drawn from our results. First, the central region of the core complex is critical for complexin binding as shown by site-directed mutagenesis of by guest on March 24, 2020 http://www.jbc.org/ Downloaded from syntaxin 1 (Fig. 10). Second, the membrane-proximal region of the core complex does not appear to be involved because complexes formed from C-terminally truncated SNAREs (e.g. by clostridial neurotoxins (38)) bind with comparable efficiency (unpublished observations).
Third, syntaxin 1 and synaptobrevin are more important for complexin binding than SNAP-25. Since in the crystal structure of the neuronal SNARE complex the helices of syntaxin and synaptobrevin form one of the grooves extending through the length of the bundle it is possible that the helix of complexin aligns along this groove during binding.
One of our most interesting findings shows that complexin is capable of distinguishing between core complexes formed from different sets of SNAREs. Such complexes are predicted to differ much more in their surface patterns than in their core interactions (3).
Indeed, in vitro studies have shown that SNAREs form core complexes rather indiscriminately (21,36), and consequently one must look elsewhere for an explanation for the intracellular specificity of SNARE interactions and their regulation. Although no other relatives of complexin have been identified so far, it is possible that other related proteins exist which bind with comparable selectivity to different SNARE complexes.
How does complexin regulate the functioning of neuronal SNAREs? Although this question cannot be answered at present, some important features are beginning to emerge.
First, it is unlikely that complexin regulates pools of individual SNAREs since it does not bind with high affinity to any of the monomeric SNAREs, unlike, for instance, munc-18 (39).
Second, we need to consider whether complexin regulates disassembly of the SNARE complex by α-SNAP and NSF. In our hands, complexin was unable to inhibit or slow the disassembly of fully formed SNARE complexes with or without transmembrane regions, arguing against such a role. It should be borne in mind, however, that α-SNAP was shown earlier to displace bound complexin from the SNARE complex (12). Disassembly creates SNARE monomers that do not bind complexin, effectively abolishing the binding site of the protein. Thus, although less likely, we cannot discard the possiblility of a more subtle, kinetic effect of complexin on disassembly of core complexes.
Third, complexin may be involved in regulating the formation of SNARE complexes before membrane fusion. As outlined in the introduction, SNARE complexes are thought to initially form "trans"-complexes that connect the two membranes before fusion. These complexes are probably reversible and in dynamic equilibria between loose and tight states (40). After fusion, all SNAREs are aligned in parallel within the same membrane in the form of relaxed "cis"-complexes. Although complexin does bind to cis-complexes, it is attractive to speculate that complexin may assist in the formation of trans-complexes, subsequently guiding them through fusion until the relaxed cis-state is reached. Indeed, complexin may be able to "proof-read" SNARE complexes upon their initial formation. It may stabilize cognate neuronal complexes and ignore non-cognate complexes during their initial formation, thus contributing to the specificity of SNARE pairing (see above). Unfortunately, there are presently no reliable assays for measuring trans-complexes and therefore the influence of complexin on this reaction cannot be determined. Clearly, however, the assembly of ciscomplexes in vitro is not influenced by complexin because addition of excess amounts of complexin does not affect the kinetics or the extent of SNARE-assembly (unpublished observations).
Fourth, binding of complexin to trans-complexes could have a role in a phase directly preceding fusion, or in fusion itself. Several different mechanisms can be imagined. For instance, in a recent abstract it was suggested that complexins may cause higher order oligomers of SNARE complexes as might be expected to form around the fusion pore (41).
However, we were unable to observe any induced oligomerization when the complexes were analyzed by size exclusion chromatography followed by MALLS (unpublished observations).
More interesting possibilities include a prevention of fusion pore formation.

Complexins bind preferentially to the ternary SNARE complex
A. Purified recombinant SNAREs and ternary complex (4 µM) were incubated with His 6tagged complexin II (3 µM). Complexin was precipitated with Ni-NTA-agarose. Bound material was analyzed by SDS-PAGE and Coomassie Blue staining. In contrast to the standard protocol, the incubation buffer contained 1 mM PMSF, and DTT was omitted from the washing buffer.
B. Purified recombinant SNAREs (containing GST tags) and ternary complex (containing GST-syntaxin, 0.75 µM each) were incubated with rat brain cytosol (2 mg/ml, as a source of native complexins) followed by precipitation with Glutathion-Sepharose. Bound complexins were analyzed by SDS-PAGE and immunoblotting. As a control, 40 µg of cytosol were loaded.
C. Binding of recombinant complexins to proteoliposomes that were reconstituted with either synaptobrevin or ternary complex. Bound complexins were detected by immunoblot. Note that in the presence of equal amounts of both complexins preferential binding of complexin I is observed (right lane). Additional immunodetection with Cl 69.1 revealed that equal amounts of synaptobrevin and ternary complex were precipitated (data not shown).

Mapping of the binding domain of complexin by deletion mutagenesis
A series of truncated GST-complexin II-fusion proteins was immobilized to Glutathion-Sepharose beads (approx. 8 µM each) and incubated with purified minimal core complex (13 µM). After washing of the beads, bead-bound material was analyzed by SDS-PAGE and Coomassie Blue staining. All samples were boiled prior to separation to dissociate the SNARE components of the minimal core complex. Note that all complexin-fragments bound to the minimal core complex except of fragments 1-69 and 70-134. At the bottom the sequence of complexin I is shown summarizing the spectral changes observed upon addition of the "midi core complex" (spectra shown in B). Boxed residues with a black background correspond to cross-peaks that broaden beyond detection while underlined residues correspond to cross-peaks with moderate broadening or slight shifts. For some crosspeaks, spectral changes cannot be assessed due to severe overlap and the corresponding residues are indicated by open boxes. All experiments were performed at protein concentrations below 100 µM to minimize aggregation.  C. Binding of complexins (1.5 µM) to ternary complex (1.5 µM) containing either wildtype syntaxin 1A, syntaxin 1/4-, syntaxin 4/1-chimera, and syntaxin 4. Complexes were isolated by immunoprecipitation (see Methods) and analyzed for complexins I and II by SDS-PAGE and Coomassie Blue staining. Controls documented that equal amounts of ternary complex were recovered in each immunoprecipitate (data not shown). Incubation and washing steps were carried out in buffer as described for His 6 -and GST-fusion proteins, respectively. B. Disassembly is not affected when complexin is present in a 20fold excess over α-SNAP.
Purified minimal core complex (0.63 µM) was preincubated for 15 min at room temperature with 108 µM complexin II and disassembled as above by addition of 1.2 µM NSF, 5.4 µM α-SNAP, 4 mM MgCl 2 and 2.5 mM ATP for the indicated times. Disassembly was measured as in A except that the appearance of synaptobrevin and SNAP-25 (1-83) was monitored by immunoblotting using monoclonal antibodies Cl 69.1 and Cl 71.1 (43), respectively. by guest on March 24, 2020