The R-SNARE motif of tomosyn forms SNARE core complexes with syntaxin 1 and SNAP-25 and down-regulates exocytosis

Tomosyn is a 130 kDa syntaxin binding protein that contains a large N-terminal domain with WD40 repeats and a C-terminal domain homologous to R-SNAREs. Here we show that tomosyn forms genuine SNARE core complexes with the SNAREs syntaxin 1 and SNAP-25. In vitro studies with recombinant proteins revealed that complex formation proceeds from unstructured monomers to a stable four-helical bundle. The assembled complex displayed features typical for SNARE core complexes including a profound hysteresis upon unfolding-refolding transitions. No stable complexes were formed between the SNARE-motif of tomosyn and either syntaxin or SNAP-25 alone. Furthermore, both native tomosyn and its isolated C-terminal domain competed with synaptobrevin for binding to endogenous syntaxin and SNAP-25 on inside-out sheets of plasma membranes. Tomosyn-SNARE complexes were effectively disassembled by the ATPase NSF together with its cofactor α -SNAP. Moreover, the C-terminal domain of tomosyn was as effective as the cytoplasmic portion of synaptobrevin in inhibiting evoked exocytosis in a cell-free preparation derived from PC12 cells. Similarly, overexpression of tomosyn in PC12 cells resulted in a massive reduction of exocytosis but the release parameters of individual exocytotic events remained unchanged. We conclude that tomosyn is a soluble SNARE that directly competes with synaptobrevin in the formation of SNARE complexes and thus may function in down-regulating exocytosis.


Introduction
Neurotransmitter release is mediated by regulated exocytosis of synaptic vesicles.
Exocytosis is carried out by membrane fusion machines involving evolutionary conserved proteins that are common to all fusion events of the secretory pathway in eukaryotic cells. Among the conserved proteins are the SNAREs, small membrane proteins characterized by homologous stretches of 60-70 amino acids referred to as SNARE motifs (1). The SNAREs involved in neuronal exocytosis include the vesicle protein synaptobrevin (also termed VAMP), and the plasma membrane associated proteins syntaxin 1 and SNAP-25 (acronym for synaptosome associated protein of 25 kDa). These proteins undergo an assembly-disassembly cycle that involves the SNARE motifs and that is associated with large structural and energetic changes (2,3).
It is currently believed that assembly leads to a tight connection between the vesicle and the plasma membrane that initiates the opening of the fusion pore (for review see 4,5).
The components of the basic fusion apparatus are acted upon by arrays of control proteins that are responsible for the specific features of each fusion reaction. For the SNARE proteins involved in neuronal exocytosis many candidate regulator proteins have been identified, but their molecular mechanisms are only beginning to be understood (5). Considering the central role of SNARE assembly and disassembly for exocytosis it is thus necessary to understand in detail which of such regulatory proteins interact with the SNARE conformational cycle and how exactly the cycle is influenced.
Tomosyn is a candidate regulatory protein that was first identified as a binding protein for syntaxin 1 (6). Tomosyn is a 130 kDa protein that lacks a membrane anchor and in which two domains are distinguished: a small C-terminal region with homology to the R-SNARE subfamily of SNAREs, and a large N-terminal region that contains WD40 repeats and shares similarities with the Drosophila tumor suppressor lethal (2) giant larvae (l(2)gl) (7), its recently characterized by guest on March 23, 2020 http://www.jbc.org/ Downloaded from mammalian homologue Mgl1 (8), and the yeast proteins Sro7p and Sro77p (9). Three splicing variants of tomosyn are known, two of which are predominantly expressed in brain whereas one is ubiquitously distributed (10).
Tomosyn has been reported to bind to the SNARE motif of syntaxin (11). In tissue extracts, tomosyn was found to be associated with syntaxin, SNAP-25, and synaptotagmin but not with synaptobrevin (6). Binding of synaptobrevin and tomosyn appear to be mutually exclusive leading to the suggestion that tomosyn may regulate synaptobrevin binding to the other SNAREs.
Furthermore, it was suggested that tomosyn displaces Munc18 from syntaxin (6). Munc18 is a conserved protein essential for exocytosis that forms tight complexes with syntaxin 1 in vitro.
Munc18 binding maintains syntaxin in an inactive state and inhibits formation of SNARE complexes (12,13). Hence, it has been proposed that tomosyn may activate syntaxin, allowing it to interact with synaptobrevin in preparation for exocytosis (6). On the other hand, overexpression of tomosyn in PC12 cells resulted in a slight but significant inhibition of exocytosis (6,13) suggesting that the role of tomosyn cannot be confined to an activator of membrane fusion.
Interestingly, the binding region of tomosyn could be assigned to the C-terminal region containing the R-SNARE motif raising the possibility that tomosyn may substitute as an R-SNARE in the formation of SNARE complexes (10,11). SNARE complexes are represented by bundles of four α-helices in which each helix is contributed by a single SNARE motif (14,15). All helices are aligned in parallel, with the transmembrane domains extending at one end. The α-helices are connected by layers of interacting amino acid side chains that are mostly hydrophobic and thus characteristic for coiled-coils. In the center, an unusual polar layer was observed consisting of three glutamines contributed by syntaxin and the two SNARE motifs of SNAP-25, and one arginine contributed by synaptobrevin (14). These residues are among the most highly conserved of the SNARE protein superfamily and led to the classification of SNAREs into Q-and R-SNAREs, respectively (16). If tomosyn substitutes as an R-SNARE for synaptobrevin in 5 core complexes, it would be expected to interfere with synaptobrevin binding before fusion, thus acting as an inhibitor rather than an activator of SNARE function. However, it is still unknown whether tomosyn forms SNARE core complexes, whether it can interact with membrane-bound SNAREs, or whether it interfers with the SNARE association-dissociation cycle.
In the present study, we show that the C-terminal domain of tomosyn functions as a genuine R-SNARE motif. Tomosyn SNARE complexes exhibit features typical for SNARE core complexes, and they are sensitive to disassembly by the ATPase NSF. The R-SNARE motif of tomosyn directly competes with synaptobrevin during complex formation with the plasma membrane-resident Q-SNAREs syntaxin 1 and SNAP-25. Membrane binding is exclusively mediated by the SNARE motif of tomosyn that is also responsible for the inhibitory action of the protein on exocytosis. Neither binding nor NSF-mediated dissociation appears to be affected by the large N-terminal domain of tomosyn, leaving the function of this domain still unexplained.

Materials
The expression constructs of synaptobrevin 2 (1-96) S28C and of a chimera of human neuropeptide Y with EGFP (NPY-EGFP) were described previously (17,18). NSF and α-SNAP were expressed and purified as described (19). Mouse monoclonal antibodies were used for the other antibodies are obtainable from Synaptic Systems (Göttingen, Germany). A rabbit polyclonal antibody raised against syntaxin 1A (R31) was used for the detection of syntaxin 1A (18). An antibody to tomosyn was raised in rabbits against a C-terminal fragment (tomosyn: residues 985-1051) that was expressed as a GST fusion protein in Escherichia coli using pGEX-KG (Amersham Pharmacia). This antibody was purified on affinity chromatography of Sepharose 4B (Amersham Pharmacia) coupled with GST-tomosyn (985-1051).

Molecular cloning
The cDNA corresponding to the open reading frame of rat m-tomosyn (1-1116) was amplified by polymerase chain reaction (PCR) using the PfuTurbo DNA polymerase (Stratagene) from rat cerebellum cDNA libraries (Stratagene). The fragment was subcloned into pBluescript vector (Stratagene) and then sequenced to confirm the identity with the published sequence.

Purification of recombinant proteins
The tomosyn fragment (residues 985-1051) and the tomosyn-SNARE-motif (residues 1051-1116) were expressed in E. coli as N-terminal GST-tagged proteins and purified by glutathione-Sepharose (Amersham Pharmacia Biotech). The GST-tagged tomosyn-SNARE-motif (1051-1116) was digested by thrombin to remove the tag and was further purified on a Mono-S column using an Äkta FPLC-system (Amersham Biotech). thrombin, all proteins were further purified using a Mono-Q column on an Äkta system (Amersham Biotech) essentially as described (20,21). All proteins were 95% pure as determinated by SDS-PAGE and Coomassie Blue staining. incubated for 30min at 37°C, and centrifuged again as above. 9

Competition assay
The single cysteine at position 28 of synaptobrevin 2 (1-96) S28C was labeled with a 10-fold molar excess of Alexa594 C5 maleimide (Molecular Probes Inc., Eugene, OR) according to the manufacture's instruction. The labeled protein was separated by gel filtration from the dye. Rat brain cytosol (1mg) was incubated for 30 min at 4°C with anti-tomosyn antibody or control IgG (100 µg each) in KGlu buffer (20 mM HEPES pH 7.2, 120 mM K-glutamate, 20 mM K-acetate, 2 mM EGTA) and added 30 µl Protein A-Sepharose (50% slurry) to deplete endogenous tomosyn in the cytosol. After incubation for 1 h at 4°C with gentle rotation, supernatant fraction was collected by centrifugation (immuno-depleted cytosol fraction).
PC12 cells were grown on poly-L-lysine-coated coverslips and disrupted as previously described (23)  Imaging of the membrane sheets and quantification of fluorescence was performed as previously described (24).

In vitro exocytosis assay
For the in vitro exocytosis assay, PC12 cells expressing the secretory granule marker neuropeptide Y-EGFP (18,22) were used. At 40-50 h after transfection, membrane sheets were generated as described above and pre-incubated for 10 min at room temperature in KGlu buffer with 80 µM of the inhibitors as indicated (synaptobrevin (1-96) 2 S28C, tomosyn (1051-1116) or by guest on March 23, 2020 http://www.jbc.org/ Downloaded from lysozyme). After pre-incubation, membrane sheets with at least 15 brightly fluorescent granules were stimulated using KGlu buffer containing 0.5 mg/ml rat brain cytosol, 0.5 µM [Ca 2+ ] free buffered with 10 mM EGTA (see 25) and 80 µM of the corresponding inhibitors. For controls, membrane sheets were incubated with K Glu buffer containing rat brain cytosol and 2 mM EGTA.
The membrane sheets were imaged every min for 30 min. The focal position of the objective was controlled throughout all experiments using a low voltage piezo translator drive and a linear variable transformer displacement sensor/controller (Physik Instrumente, Waldbronn, Germany).
Finally, the membrane sheets were visualized using TMA-DPH (Molecular Probes Inc., Eugene, OR). Imaging and image analysis was performed as previously described (25).

Carbon fiber amperometry
Carbon fiber amperometry was used essentially as described (26). Amperometric signals from PC12 cells were recorded with an EPC-7 amplifier (Heka Electronics) with a holding potential of the carbon fiber set to +800mV. The tip of the fiber was brought into direct contact with the plasma membrane to minimize temporal distortion of the release signal and diffusional were filtered with 3kHz and 10kHz, digitized at 25kHz and stored on a personal computer. For data analysis, the AutesW software (NPI Electronics, Tamm, Germany) was used as described in (26). Signals were again digitally filtered at 3kHz (overall band width 2.1kHz). Additionally, a routine for analysis of foot events preceding the main amperometric spike was included (see Table  PC12 cells ((28); passage number 27 to 33) were cultured as described previously (29). Other methods.
SDS-PAGE and immunoblotting was carried out using standard procedures.
Electroblotting was performed on Immobilon-P membranes (Millipore, Bedford, MA). As secondary antibodies, we used HRP-conjugated anti-rat IgG or anti-mouse IgG, and the blots were developed by enhanced chemiluminescence using SuperSignal West (PIERCE) according to the manufacture's instructions. Multi-angle laser light scattering measurements and CD spectroscopy were performed as described (3,21).

Results
The C-terminal domain of tomosyn forms SNARE core complexes SNARE core complexes display several highly unusual properties including progression from unstructured monomers to helical bundles upon assembly, extraordinary heat stability, and profound hystereses upon unfolding-refolding cycles (3). To examine whether the C-terminal domain of tomosyn can function as a genuine R-SNARE in the formation of core complexes, purified recombinant SNARE motifs of tomosyn, syntaxin 1, and SNAP-25 were combined and the features of the resulting complexes were studied.
First, we used circular dichroism (CD) spectroscopy to examine whether binding of tomosyn to syntaxin 1 and SNAP-25 involves a transition from unstructured monomers into a helical bundle. As shown in Fig. 1A, a large increase in α-helical content was observable when the SNARE motif of tomosyn was mixed with SNAP-25 and syntaxin. No secondary structure was detectable in the CD-spectrum of the isolated SNARE motifs (Fig. 1A). Furthermore, no structural change was observed when the SNARE motif of tomosyn was combined with SNAP-25 or with its isolated SNARE motifs, and only a minor increase was observed upon mixing with the H3-domain of syntaxin 1 (Fig. 1B). For further characterization, tomosyn-containing SNARE complexes were purified by ion exchange chromatography. When analyzed by size-exclusion chromatography and multi-angle laser light scattering, the purified complex migrated largely as a single peak with a molecular mass of approximately 45,000 dalton (data not shown), as expected for a complex with a 1:1:1:1 stoichiometry.
We then tested whether the tomosyn-SNARE complex is heat stable, and if so, whether heat denaturation and re-naturation would display hysteresis characteristic for SNARE core complexes (3). Fig.1 C shows that the tomosyn SNARE complex has an unfolding transition temperature of over 80°C whereas re-folding requires temperatures below 60°C. Together, these properties are indistinguishable from SNARE complexes containing synaptobrevin (3) documenting that the R-SNARE motif of tomosyn is fully functional in substituting for synaptobrevin as R-SNARE. Tomosyn was immunoprecipitated from detergent extracts of PC12 cells and tested for association of SNARE-proteins ( Fig. 2A). Both syntaxin 1 and SNAP-25 coprecipitated with tomosyn showing that native tomosyn interacts with these SNAREs in intact cells. Furthermore, a minor amount of synaptobrevin was found in the immunoprecipitate which may be due to interactions involving the transmembrane domains rather than the SNARE motifs (30).
Conversely, tomosyn coprecipitated with syntaxin when a monoclonal antibody specific for the N-terminal domain of syntaxin was used (Fig. 2B).
To estimate the contribution of the N-terminal domain of tomosyn to the formation of SNARE-complexes, we transfected PC12 cells with expression vectors coding for myc-tagged versions of full-length tomosyn, the N-terminal domain (residues 1-1030), and the C -terminal region carrying the R-SNARE motif (residues 1031-1116). Immunoprecipitations were then performed from detergent extracts of PC12 cells as above using an anti-myc antibody. As shown in Next we investigated whether the R-SNARE motif of tomosyn substitutes for synaptobrevin when SNARE complexes are formed in intact membranes. To address this issue, we took advantage of our recent finding that the recombinant R-SNARE motif of synaptobrevin forms core-complexes with endogenous membrane-resident SNAP-25 and syntaxin 1 when added to freshly prepared inverted sheets of plasma membranes (24). PC12 cells were grown on coverslips and subjected to brief sonication to disrupt the cells, resulting in the generation of inside-out membrane sheets. The membrane sheets were incubated for 30 min at 37 °C with recombinant synaptobrevin (lacking the transmembrane domain) that was labeled with the fluorescent dye Alexa 594. As shown in Fig. 3, synaptobrevin bound to syntaxin clusters in these membrane sheets resulting, as shown previously (24), in the formation of genuine cis-core complexes. As expected for a specific and saturable interaction, binding of labeled synaptobrevin was progressively reduced when increasing amounts of unlabeled protein were added (Fig. 3G). When synaptobrevin was replaced with the SNARE motif of tomosyn in these experiments, competition with the binding of labeled synaptobrevin was of comparable efficacy. These data show that the proteins bind to the same site with similar affinities.
We next investigated whether full-length tomosyn is also capable of competing with synaptobrevin binding in this assay. Unfortunately, we have so far been unable to prepare recombinant full-length tomosyn for binding studies. We therefore took an indirect approach and used whole brain cytosol as source for endogenous tomosyn. To account for effects by factors other than tomosyn, part of the cytosol was depleted of endogenous tomosyn by immunoprecipitation resulting in near-quantitative removal of the endogenous protein ( Fig. 3I) with no detectable changes in the overall protein pattern (not shown). Addition of cytosol to the binding reaction that was pre-treated with a control IgG resulted in efficient competition of synaptobrevin binding (Fig. 3H). Equal concentrations of immunodepleted cytosol were significantly less efficient although some inhibitory activity remained. Addition of the recombinant SNARE motif of tomosyn to the reaction was as effective in competing with synaptobrevin as in the absence of cytosol. Thus, native tomosyn forms SNARE complexes with syntaxin and SNAP-25, and there appears to be no interference by cytosolic factors with the R-SNARE activity of the protein.
The data described so far suggests that the large N-terminal domain does not interfere with the ability of the C-terminal SNARE motif to enter SNARE complexes. However, it remains to be established whether the same applies to the "reverse"-reaction, i.e. whether SNARE-complexes containing native tomosyn can be disassembled by the ATPase NSF.
Membrane fractions, prepared from PC12 cells, were pretreated with N-ethylmaleimide to block endogenous NSF and then incubated with ATP either in the absence or the presence of recombinant NSF and α-SNAP. NSF-treatment resulted in a release of tomosyn from the membrane that was dependent on the presence of Mg-ATP (Fig. 4). Similarly, tomosyn could be removed from the plasma membrane of digitonin-permeabilized PC12 cells upon incubation with ATP-NSF (data not shown). We conclude that neuronal SNARE complexes containing tomosyn are effectively disassembled by NSF.

The SNARE motif of tomosyn is responsible for downregulating exocytosis in PC12 cells
So far we have established that the R-SNARE motif of tomosyn forms SNARE complexes with syntaxin and SNAP-25 that can be disassembled by NSF and that are similar to core complexes containing synaptobrevin. It has previously been shown that overexpression of tomosyn causes partial inhibition of exocytosis in PC12 cells (6), and it has been proposed that the tomosyn may mediate its inhibition by functionally substituting an R-SNARE motif.
To investigate whether the R-SNARE motif of tomosyn is indeed responsible for inhibiting exocytosis in PC12 cells, we took advantage of a recently developed cell-free system for by guest on March 23, 2020 http://www.jbc.org/ Downloaded from exocytosis that is based on the preparation of membrane sheets similar to those described above.
We have shown previously that such membrane sheets contain bound secretory granules that retain their ability to undergo Ca 2+ -dependent exocytosis for some time after cell disruption (23). PC12 cells whose secretory granules were fluorescently labeled by expression of neuropeptide Y-EGFP (18) were grown on coverslips and sonicated to produce plasma membrane sheets containing labeled secretory granules (Fig. 5). Exocytosis was monitored by changes in the fluorescence intensity as described (23,25). After a pre-incubation period of 10 min in the presence of cytosol and ATP, calcium was added, resulting in a calcium-dependent exocytic activity of approximately 21% of all granules during the subsequent 30 min (Fig. 5D). When recombinant synaptobrevin (1-96) was added, exocytosis was largely inhibited. A very similar inhibition was observed when the C-terminal part of tomosyn was added.
In the final set of experiments, we addressed whether tomosyn overexpression has an impact on the exocytotic fusion event itself. It has previously been reported that tomosyn overexpression in PC12 cells leads to a slight reduction in the release of co-transfected human growth hormone (6). To gain more insight into the underlying mechanism, we analyzed the effect of tomosyn overexpression on the frequency and release parameters of dopamine release from single secretory vesicles. PC12 cells were co-transfected with full-length tomosyn and EGFP in order to identify transfected cells and exocytosis was monitored by carbon fiber amperometry.
Upon depolarizing stimulation with high K + -containing solution control cells respond with an initial phase of high exocytotic activity (2.1 ± 0.8 events/sec) that slowly declines in order to proceed at a more constant rate (1.1 ± 0.4 events/sec) as long as the stimulus is applied (Fig. 6).
Tomosyn-expressing cells showed no biphasic response but exocytosed vesicles with an average rate of about 0.25 ± 0.1 events/sec. The lack of an initial phase with higher exocytotic activity suggests that the number of rapidly-releasing granules is strongly diminished upon overexpression of tomosyn. In contrast, no significant changes were observed in the analysis of the single spike parameters including the event's quantal size, amplitude, rise-time and halfwidth and in the parameters of the amperometric foot signal that represents the opening kinetics of the exocytotic fusion pore (Table 1). Taken together, the data suggest that tomosyn governs the readiness of the secretory organelles for exocytosis rather than the molecular mechanism of membrane merger.

Discussion
In the present study we have shown that the C-terminal portion of tomosyn acts as a fully functional R-SNARE motif with similar properties to the R-SNARE motif of synaptobrevin. It mediates membrane binding by forming SNARE complexes with syntaxin 1 and SNAP-25 that are sensitive to disassembly by NSF. Furthermore, it mimics the inhibitory action of synaptobrevins' SNARE motif on exocytosis (31,32). In contrast to synaptobrevin, tomosyn does not contain a membrane anchor domain but has a large N-terminal domain that does not appear to interfere with the SNARE activity of the C-terminus.
As outlined in the introduction, tomosyn is related to the Drosophila tumor suppressor lethal (2) giant larvae (l(2)gl) (7), its mammalian homologue Mgl1 (8) and the yeast proteins Sro7p and Sro77p (9). The homology is confined to the N-terminal domain containing WD40 repeats, protein binding domains with a conserved structure of about 40 residues, each containing a central Trp-Asp motif. WD40 domains are involved in specific protein-protein interactions, best characterized in G protein mediated signal transduction (33). In contrast to tomosyn, none of the hitherto known relatives possesses a SNARE motif (based on unpublished sequence comparisons, see also discussion in (8)). Intriguingly, however, several of the other family members have not only been demonstrated to play a role in membrane traffic of polarized cells but also to directly interact with intracellular SNAREs. In yeast, cells deficient in SRO7 and SRO77 exhibit a defect in vesicle transport from the Golgi apparatus to the plasma membrane, and the protein Sro7p binds to the yeast SNAP-25 homologue Sec9p (9). Furthermore, Mgl1 appears to play a role in polarized membrane trafficking as it becomes associated with the lateral membrane when MDCK cells establish polarity, and membrane binding of Mgl1 seems to be mediated by binding to syntaxin 4 (8). These findings contrast with our observation that the N-terminal domain of tomosyn does not interact with membranes. Furthermore, Yokoyama et al. (10) were unable to detect syntaxin binding with this domain using a blot overlay assay, corroborating the findings reported here. It is tempting to speculate that the acquisition of the C-terminal SNARE motif is due to a gene fusion event during evolution and that SNARE binding of the N-terminal domain subsequently degenerated. However, we cannot exclude that the N-terminal domain of tomosyn (particularly the WD40 repeats) interacts with other intracellular proteins, either as an independent entity or as a regulator of the C-terminal SNARE motif. In the latter case, the SNARE motif would be constitutively "on" in the intact molecule since the presence of the N-terminal domain did not have any measurable effect on the SNARE-activity of the C-terminal domain,.
Our data show that the SNARE-motif of tomosyn does indeed function as an R-SNARE in SNARE assembly and disassembly reactions. The complex formed with syntaxin and SNAP-25 is remarkably similar to the ternary SNARE complex formed with synaptobrevin, strongly suggesting that it is represented by a four-helix bundle characteristic for all SNAREs. The similarities include 1:1:1:1 stoichiometry of the participating SNARE motifs, unusual heat resistance, strong hysteresis during folding-unfolding transitions, and major conformational changes during assembly that proceeds from unstructured monomers to a highly helical complex ((3), see (5) for review). Furthermore, the recombinant SNARE motif of tomosyn binds to membrane-bound syntaxin 1 and SNAP-25 in a manner indistinguishable from the cytoplasmic part of synaptobrevin. Synaptobrevin was previously shown to form cis-core complexes under these conditions (24).
The fact that tomosyn functions as a SNARE in the formation of SNARE core complexes is not easily reconcilable with the idea that the tomosyn-SNARE complex is an intermediate in an activation sequence during which it removes Munc18 from syntaxin and "hands" it over to its cognate SNARE partner synaptobrevin (see e.g. 10). Rather, tomosyn-containing SNARE complexes require the chaperone ATPase NSF for disassembly that acts upon all partially and fully assembled SNARE complexes. Insofar, tomosyn acts as a competitive inhibitor of synaptobrevin We have recently established that the plasma membrane pools of syntaxin and SNAP-25 are essentially uncomplexed and active under steady-state conditions (24). This indicates that NSF maintains all SNAREs in a predominantly disassembled state. Thus, the membrane pool of tomosyn is probably relatively small although we cannot exclude that disassembly of tomosyn-SNARE complexes is slower than that of genuine SNARE complexes, or down-regulated by its N-terminal domain. It is conceivable that tomosyn becomes locked at the membrane as soon as NSF activity is reduced which may result in an overestimate of the membrane-bound pool in some of our experiments. That overexpression of tomosyn reduces the rate of evoked exocytosis does not necessarily imply that a reduction of tomosyn would lead to an increase in exocytosis. In any case we can exclude that tomosyn affects the exocytotic fusion event itself since no changes were observed in the kinetics of single vesicle exocytosis.
It remains to be established to which extent the function of tomosyn is confined to regulated exocytosis. Three splicing variants of tomosyn, all containing the R-SNARE motif, were described, one of which is ubiquitously expressed (10). The SNARE motif of tomosyn was identified in yeast two-hybrid screens searching for binding partners of syntaxin 4 and SNAP-23 although binding to syntaxin 4 was not observed in pull-down assays. Both syntaxin 4 and SNAP-23 are thought to be involved in fusion events at the plasma membrane suggesting that tomosyns may be specific regulators of exocytosis in many different cell types.   H, Competition of synaptobrevin binding using tomosyn-containing cytosol that was either immuno-depleted with control IgG or with anti-tomosyn antibodies. Treatment with anti-tomosyn antibodies led to near quantitative removal of tomosyn (immunoblot analysis shown in I). As control, all cytosol fractions were heat-inactivated prior to the experiment. Tomosyn is released from membranes by NSF and α-SNAP Immunoblots of PC12 cell membranes and the corresponding supernatants that were incubated with NSF and α-SNAP. Incubation was carried out for 30 min at 37°C with either an ATP-regenerating system (ATP) or an ATP-regenerating system supplemented with EDTA to inactivate ATP-cleavage due to chelation of Mg 2+ . Where indicated, purified recombinant NSF and α-SNAP (each at 6 µM final concentration) were added. The membrane fractions were pretreated with 4 mM NEM on ice for 30 min prior to membrane isolation to inactivate endogenous NSF (see Methods). After the incubation, membranes were sedimented by high-speed centrifugation, and equal proportions of the pellet (P) and supernatant fractions (S) were analyzed by SDS-PAGE and immunoblotting. The C-terminal region of tomosyn (residues 1051-1116) inhibits Ca 2+ -dependent exocytosis in a semi-intact system derived from PC12 cells.

Figure legends
PC12 cells were transfected with NPY-EGFP as a granule content marker, followed by cell disruption, leading to the generation of inverted lawns of plasma membranes (see A for visualization of phospholipid membranes with TMA-DPH) containing bound secretory granules (B and C). For the triggering and monitoring of exocytosis, ATP, cytosol and 500 nM free calcium were added and membrane sheets were imaged every minute for 30 minutes (see B and C for the first and last image of a sequence). Exocytic activity of individual granules resulted in abrupt changes in fluorescence intensity in the sequence leading either to dimming (due to loss of GFP, compare dashed circles in B and C) or brightening (due to pH-dequenching of GFP, compare dotted circles in B and C) of individual granules (25). The percentage of active granules was related to all GFP-labeled granules present at the beginning (for details see 25). For each condition, the background exocytic activity in the absence of calcium was measured separately and subtracted. For each experiment 6-14 (mean = 10) membrane sheets were analysed. Values are presented as means ± S.E M. The concentration of synaptobrevin and tomosyn fragments was 80 µM.  Table 1 Properties of individual amperometric signals in control cells and cells overexpressing tomosyn full-length protein. Amperometric events were characterized with respect to charge (fC) reflecting the area under spike, maximum amplitude (pA), 50 -90 % rise time and halfwidth (representing the duration of the spike at the 50% value of its maximum amplitude). Only events with an amplitude larger than 3 pA were included. Foot signals were analysed with respect to charge (fC), current amplitude (pA) and duration (µs) . The start of a foot signal was defined as the time point when the current amplitude exceeds two times the average baseline noise standard deviation, its end is indicated by the inflection point in the current signal marking the transition between the slow and the fast increase of the current amplitude during the foot and spike phase, respectively.
Values are given as mean ± s.e.m..