Syndet is a novel SNAP-25 related protein expressed in many tissues

SNAP-25 is a synaptosomal associated protein expressed in neurons and neuroendocrine cells (Oyler et al., 1989). In nerve terminals, SNAP-25 and syntaxin 1 are thought to form a plasma membrane-bound receptor (t-SNARE) for two cytoplasmic components, the N-ethylmaleimide sensitive fusion protein (NSF), and the soluble NSF attachment proteins (SNAPs) (Clary et al., 1990; Malhotra et al., 1988; Sollner et al., 1993a). SNAPs and NSF are also bound by the vesicle SNAP-receptors (v-SNAREs): VAMP-1 (vesicle associated membrane protein-1) and VAMP-2, which are integral membrane proteins on the surface of synaptic vesicles (Trimble et al., 1988, 1990). According to the current concept, synaptic vesicles dock to the presynaptic membrane via a protein bridge formed by v-SNARE, t-SNARE, SNAPs and NSF (Bennett and Scheller, 1994; Rothman and Warren, 1994). Fusion of vesicles and release of neurotransmitter would occur after ATP hydrolysis by NSF (Sollner et al., 1993a). The SNARE complex seems to be essential for neurotransmitter release, as indicated by the finding that botulinum toxins and tetanus toxins, which block synaptic vesicle release, specifically cleave VAMP-2, SNAP-25 and syntaxin 1 (Blasi et al., 1993a,b; Schiavo et al., 1992). Since NSF and SNAPs are required for intracellular traffic (Rothman and Warren, 1994), docking and fusion of vesicles may occur by a similar mechanism at any point in the pathway from the endoplasmic reticulum to the plasma membrane. Specific sets of SNARE on the donor and acceptor compartment would have to pair correctly before NSF hydrolyzes ATP and fusion occurs. This hypothesis implies that there is a family of related proteins which act as specific v-SNARE and t-SNARE receptors. In accordance with this hypothesis, closely related syntaxin homologues have been identified. Syntaxin 2, 3, 4 and 5 are expressed in many tissues (Bennett et al., 1993). Syntaxin 1-4 are localized at the plasma membrane where they may constitute specific receptors for different kinds of exocytotic vesicles of both the constitutive and regulated pathways. The v-SNARE family of proteins includes VAMP-1 and VAMP-2, which are abundant in brain, and cellubrevin (McMahon et al., 1993) which is expressed in many tissues. VAMP-2 interacts with syntaxin 1 and 4, but not with syntaxin 2 and 3 indicating that the specificity of membrane fusion may be determined by the correct pairing of v-SNAREs and t-SNAREs (Calakos et al., 1994). Moreover, SNAP-25 stabilizes specifically the complex formed by syntaxin 1, or 4, and VAMP-2 (Pevsner et al., 1994). However, cellubrevin, a v-SNARE associated with the endosomal compartment (Galli et al., 1994; McMahon et al., 1993), also interacts with the neuron-specific t-SNARE syntaxin-1/SNAP-25, suggesting that pairing of SNAREs is not the only factor conferring specificity for constitutive and regulated vesicle fusion (Chilcote et al., 1995). To understand how exocytotic and recycling vesicles fuse at specific sites on the plasma membrane, it is necessary to identify all the components of the SNARE complexes functioning at the cell surface (De Camilli, 1993). SNAP-25 expression is restricted to neurons and neuroendocrine cells in which the protein is necessary for synaptic transmission and hormone release (Oyler et al., 1989). Here we characterize a new member of the SNAP-25 family which is expressed at the plasma membrane of non-neuronal cells. This protein may be an essential component of the machinery for docking/fusion of secretory or endo-exocytotic vesicles with the plasma membrane.

The amino acid sequences ⁵¹DEQGEQL and ¹⁴³ENEMDEN of human SNAP25A are conserved in several species with some modifications: residues ⁵¹E and ¹⁴⁴N may be substituted with aspartic acid and ¹⁴⁷D residue may be substituted with glutamic acid. To find novel sequences related to neuronal SNAP-25, and expressed also in nonneuronal cells, single stranded DNA derived from an adipocyte 3T3-L1 cDNA library was used as a template (Baldini et al., 1992) and amplified by PCR with the following primers: primer 1, 5’-GAY-GANCARGGNGARCARYT-3’ (sense; Y= C or T; R= A or G; N= A,C,G or T corresponding to the sequence DE(D)QGEQL and primer 2, 5‘-RTTYTCNTCCATYTCRTYYTC-3‘ (antisense) corresponding to the sequence ED(N)EME(D)EN. DNA from the 3T3-L1 adipocyte library (0.1 µg) and 1 µM of sense and antisense primers in 100 µl of the reaction mix (GeneAmpPCR core reagents, Perkin Elmer Cetus) containing 1 mM MgCl₂ were subjected to 35 cycles of 94°C for 1 minute, 46°C for 40 seconds and 72°C for 20 seconds. A PCR reaction product of the expected size of approximately 300 base pairs was labeled with digoxigenin-11-dUTP (The Genius System, Boehringer Mannheim) according to the manufacturer’s instruction. After screening 250,000 colonies from a 3T3-L1 adipocyte cDNA library with the labeled PCR product we isolated one positive clone, designated as pcDNA1-syndet. Syndet cDNA was excised from pcDNA1 vector with the restriction enzymes HindIII and XhoI and subcloned in the same sites of plasmid pBluescript SK+. The insert was subjected to automated sequencing on an Applied Biosystem 373-A sequencer. Sequence analysis was performed with the DNA star package. The sequence data are available from GenBank, accession number U73143. Homology searches were performed at NCB using the Blast network service, and alignments were performed with the Megalign program from DNAstar using the Clustal algorithm.

Antibodies against two syndet peptides were generated in rabbits by Research Genetics (Huntsville, AL). Peptide 1 (amino acids 1-14): MDNLSPEEVQLRAH and peptide 2 (amino acids 115-128): VSKQPSRITNGQPQ were synthesized with the MAP technique on a poly-lysine carrier core (MAP-peptides) and directly injected into rabbits. Peptide 1 induced a much stronger immune response, this antiserum was designated as NVG-19609 and used for all subsequent experiments. For affinity purification of antibodies the peptides were coupled to CNBr-Sepharose (Pharmacia). Affinity purification was performed according to the manufacturer’s instructions. The monoclonal antibodies Cl42.2 against Rab3A were provided by Dr R. Jahn (Yale University) (Matteoli et al., 1991). Polyclonal antibodies against lamins A and C were provided by Dr Worman (Columbia University) (Cance et al., 1992). Glut4 antibodies were previously characterized (Baldini et al., 1991).

3T3-L1 cells (Martelli et al., 1995) were grown in DME containing 10% calf serum. Cells were differentiated into adipocytes by an established protocol (Frost and Lane, 1985). Cos-7 cells were grown in DME containing 10% fetal calf serum. Cos-7 cells were transfected with the plasmid pcDNA1-syndet which contains the syndet coding sequence downstream of the cytomegalovirus promoter of pcDNA1 plasmid. Cell transfection was done with Lipofectin (Gibco BRL) according to the manufacturer’s instructions. Twenty-four hours after transfection, Cos-7 cells were trypsinized; a fraction of Cos-7 cells was then plated on poly-lysine coated coverslips for immunofluorescence.

Cells were washed twice in ice-cold PBS, and 0.25 ml of buffer B (50 mM Tris-HCl, 1 mM EDTA, 150 mM NaCl, 1% (v,v) Triton X-100, 0.1% (w,v) SDS, 300 µg/ml PMSF, 1.5 µg/ml leupeptin and 6 µg/ml aprotinin), was added to 6 cm cell culture dishes. Cells were scraped and homogenized with a 2 ml Teflon pestle tissue homogenizer. Homogenates were kept on ice for thirty minutes. Samples were then centrifuged at 4°C in an Eppendorf centrifuge for 10 minutes at maximal speed. Pellets were discarded and the supernatants, containing detergent extracts of cells, were analyzed for protein concentration with the BCA Protein Reagent (Pierce) and kept at-70°C. Tissues were homogenized in buffer B and detergent extracts of tissues were obtained with the procedure described above.

The same immunofluorescence procedure was used for Cos-7 cells and 3T3-L1 adipocytes. Cells were grown on poly-lysine coated coverslips. All procedures were carried out at room temperature unless otherwise indicated. We used the following protocol: (1) cells were washed twice in ice-cold buffer M2 (150 mM NaCl, 20 mM Hepes buffer, 0.70 mM CaCl₂, 5 mM KCl and 1 mM MgCl₂) and fixed for 30 minutes in M2 buffer containing 3.7% formaldehyde; (2) the samples were washed 3 times for 5 minutes with M2 buffer, subsequently samples were incubated at 37°C for 30 minutes in M2 buffer with the addition of 0.1 mg/ml saponin and 1 mg/ml ovalbumin (M3 buffer). Cells were then incubated at 37°C for 1 hour with M3 buffer containing affinity purified syndet antibody NVG-19609. Control samples were incubated in parallel with the omission of the primary antibody; (3) cells were washed 4 times for 5 minutes in M3 buffer and incubated at 37°C for 1 hour with FITC-conjugated donkey antirabbit IgG (Jackson laboratories) in M3 buffer; (4) samples were washed four times for 5 minutes in M3 buffer and once for 5 minutes in M2 buffer, treated with methanol for 5 minutes at-20°C and with acetone for 5 minutes at-20°C and mounted with SlowFade (Molecular Probes). In some experiments, 3T3-L1 adipocytes were double-labeled with FITC-conjugated wheat germ agglutinin and syndet antibodies. Adipocytes were washed twice with ice-cold DME and incubated at 4°C for 1 hour in DME containing 5 µg/ml FITC-conjugated wheat germ agglutinin (Molecular Probes). The cells were washed and fixed in M2 buffer containing 0.37% formaldehyde for 16 hours at 4°C. Cells were then processed for labeling with syndet antibodies as described above, with the exception that the secondary antibody was Cy3-cyanine-conjugated donkey anti-rabbit IgG (Jackson laboratories). To control the specificity of syndet immunofluorescence, 0.1 ml of the antibody preparation was pre-incubated for 1 hour at room temperature with 0.1 ml of Sepharose beads coupled to MAP-peptide 1. The beads were pelleted and the supernatant was used for adipocyte staining at the same dilution as syndet antibodies. Fluorescence micrographs were taken with a ×100 oil immersion objective on an Olympus Vanox-T fluorescence microscope using rhodamine (for Cy3-cyanine) or fluorescein conditions. Confocal microscopy was carried out using a Bio-Rad MRC-600 microscope.

Cell fractionation of 3T3-L1 adipocytes was done essentially as described before (Baldini et al., 1991; Simpson et al., 1983). Briefly, 3T3-L1 adipocytes were washed 3 times in DME medium 16 hours before the experiment. Immediately prior to insulin incubation, adipocytes were again washed once with DME and incubated with 100 nM porcine insulin (Sigma) in DME or DME alone for 30 minutes at 37°C. All procedures were then carried out at 4°C. Cells were washed twice with an ice-cold solution containing 20 mM Tris-HCl/1 mM EDTA, 225 mM sucrose (TES buffer). Adipocytes were scraped from the culture plates in TES buffer containing 300 µg/ml PMSF, 1.5 µg/ml leupeptin and 6 µg/ml aprotinin and homogenized in the same buffer. The homogenates were centrifuged at 21,000 RPM (16,000 g) in a TLA 100.2 rotor of the Beckman TL Optima centrifuge (Beckman Instruments) for 9 minutes to obtain a pellet (P0) and a supernatant fraction (S1). P0 was resuspended in TES solution containing protein inhibitors, layered onto a 38.3% sucrose cushion and centrifuged at 36,000 rpm (81,0000 g) for 20 minutes in a TLS-55 rotor to obtain a plasma membrane fraction (P1) floating above the sucrose cushion and a pellet containing mitochondria and nuclei (P2) which sediments at the bottom of the sucrose cushion. The 16,000 g supernatant fraction was spun at 37,000 rpm (48,000 g) for 9 minutes in a TLA 100.2 rotor to obtain a microsomal pellet (P3) and a supernatant fraction (S2). S2 was further centrifuged at 95,000 rpm in the TLA 100.2 rotor to obtain a soluble fraction (S3) and the microsomal fraction, P4.

Mice were anesthetized with pentobarbital (60 mg/kg body weight). Fixative (4% paraformaldehyde in 0.1 M phosphate buffer, pH 7.3, PBS) was perfused (100 ml via the ascending aorta) and the organs were rapidly removed. Blocks of tissue were postfixed for 2 hours in the same fixative and embedded in 8% gelatin. Sections (40 µm) were cut on the Vibratome (Oxford Instruments) and treated for the demonstration of sites of syndet immunoreactivity as follows: sections were washed in PBS, treated for 20 minutes with 0.5% H₂O₂ to quench endogenous peroxidase activity, again washed in PBS then treated for 30 minutes with 3% normal goat serum to eliminate non-specific background staining. Sections were then incubated for 1 week at 4°C with affinity purified syndet antibody 1:20 diluted in PBS with 0.05% saponin. Sites of antigen antibody complexes were visualized using the avidin-biotin horseradish peroxidase method (ABC kit, Vectastain) with 3,3‘-diaminobenzidine tetrahydrochloride (DAB) as the chromogen. We have found that this protocol for fixation and immunocytochemistry allows for the retention of antigenicity and at the same time preserves the ultrastructural integrity of cell membranes (Witkin et al., 1995). In control experiments, there was no reaction product in tissue sections treated with the above protocol except for omission of the primary antibody. For ultrastructural examination, tissue sections were osmicated for 1 hour with 2% OsO₄ in 0.9% saline containing 1.5% postassium ferricyanide. Tissues were dehydrated and flat embedded in tEpon. Thin sections (70 nm) were examined without counterstain and photographed on a JEOL 1200EX electron microscope.

Extraction of 3T3-L1 adipocytes with sodium carbonate was performed as described before (Baldini et al., 1994). SDS-PAGE and immunoblotting were performed as described before (Baldini et al., 1991; Laemmli, 1970).

Primer 1 and Primer 2, encoding two amino acid regions conserved in several SNAP-25 isoforms (Risinger et al., 1993), were used to amplify SNAP-25 related sequences from a 3T3-L1 adipocyte cDNA library. A PCR product of approximately 300 bp was amplified, subcloned and used to transform bacteria. The cDNA was isolated from 5 independent colonies and sequenced. The nucleotide sequences of all 5 cDNA inserts were identical, indicating that the PCR reaction product consisted mainly of one cDNA species. A 3T3-L1 adipocyte library was screened with a digoxygenin-labeled PCR reaction product and a positive clone was isolated and sequenced. The first ATG of the single open reading frame has the surrounding sequence TCACCATGG which corresponds closely to the consensus translation initiation site (Kozak, 1991) (not shown). The open reading frame of 633 nucleotides encodes a protein designated syndet, of 210 amino acids with predicted molecular mass of 23 kDa and isoelectric point of 4.78. The analysis of the amino acid sequence by PSORT program (Lupas et al., 1991) predicts that syndet, as SNAP-25, contains neither a transmembrane domain nor a signal sequence. The name syndet is derived from the Greek syndetes (an agent that binds separate parts together).

Fig. 1 shows the alignment of the predicted amino acid sequence of syndet with several members of the SNAP-25 family. The overall percentage similarity of syndet to SNAP-25 isoforms ranges from 48.8 (syndet versus Drosophila SNAP-25), to 57.8 (syndet versus human or mouse SNAP-25). The similarity of other SNAP-25 isoforms with respect to each other ranges from 55.9% to 100%. Sequences between residues 18-91 and 133-210 have 79% and 67% amino acid identity with the corresponding sequences of the mouse SNAP-25 isoform. Two regions of syndet are divergent from other members of the SNAP-25 family. One is at the amino terminus of syndet (between residues 1 and 17), the other (between residues 92 and 132), is downstream of the conserved cysteine-rich domain. Syndet has a cysteine-rich region (amino acid residues 79-87) which is conserved in all known SNAP-25 related sequences excluding Sec9, the yeast cognate of SNAP-25 (Brennwald et al., 1994) (Sec9 sequence is not shown in Fig. 1). Unlike human, mouse or chicken SNAP-25 isoforms, all 5 cysteine residues present in the Torpedo synapse isoform of SNAP-25 are conserved in syndet.

Sequence analysis using the COILS program (Lupas et al., 1991; Parry, 1982) revealed that syndet, like SNAP-25 (Chapman et al., 1994), has regions containing sets of multiple heptad repeats characteristic of α-helices that form coiled-coil structures (Cohen and Parry, 1990). Fig. 2 shows the propensity of syndet and SNAP-25 to form coiled-coil structures. Five sets of heptad repeats reside between amino acid residues 45 and 79 of syndet and correspond to a conserved domain in the amino-terminal half of the molecule. Nine sets of heptad repeats are in the conserved carboxyl-half between amino acid residues 148 and 210 of syndet. The two predicted amphipathic α-helices of syndet are separated by a gap of 69 amino acid residues which includes the cysteine-rich domain and a sequence which is highly divergent from SNAP-25. Unlike syndet, SNAP-25 has an additional amphipathic α-helix at the amino-terminal end.Syndet has a broad tissue distribution

In Cos-7 cells transfected with plasmid pcDNA1-syndet, syndet appeared as a protein band of approximately 26 kDa (Fig. 3, lanes 2 and 4), a value close to its predicted molecular mass of 23 kDa. Low levels of endogenous syndet may be present in untransfected Cos-7 cells (lanes 1 and 3). Endogenous syndet in adult mouse tissues had the same SDS-PAGE gel mobility as does syndet overexpressed in Cos-7 cells (Fig. 4, compare lane 1 with other lanes). The tissues with the highest expression of syndet were kidney, testis, liver and spleen (lanes 3, 7, 9, 10), an intermediate level of expression was observed in fat and lung tissues (lower panel, lanes 5 and 11). Syndet levels were low or undetectable in pituitary, brain, skeletal muscle, pancreas and heart tissues.

Association of syndet with membranes was tested in 3T3-L1 adipocytes. Cell homogenates were prepared in carbonate at pH 11.5 and centrifuged at 321,000 g. The membrane pellet and the supernatant were then analyzed for the presence of syndet. Fig. 5 shows that syndet is found only in the membrane pellet indicating that it behaves as an integral membrane protein (Fujiki et al., 1982). As a control, Rab3A, a protein known to cycle between a membrane-bound and a cytosolic pool (Fischer von Mollard et al., 1991), was recovered both in the carbonate soluble fraction and in the pellet.

To identify the membrane compartments where syndet is present, its subcellular distribution was determined by differential centrifugation of 3T3-L1 homogenates and western blot analysis of the fractions obtained. The scheme of the membrane fractionation is presented in Fig. 6. We used Glut4 antibodies to label different membrane fractions derived from insulin-treated and untreated cells. Glut-4 is a glycosylated protein which appears as multiple protein bands centered at 43 kDa (Baldini et al., 1991). The bands below and above Glut-4 in P2 fraction are due to nonspecific staining since they are detected by the pre-immunne serum. Glut4 immunoreactivity was enriched in microsomal fractions P3 and P4. Conversely, Glut4 was low or undetectable in S3 and in P2 fractions which contain mitochondria and nuclei, as indicated by the specific staining with antibodies against nuclear lamin A and C. P1 fraction contains adipocyte plasma membrane as indicated by the increased amounts of Glut4 upon stimulation of cells (Baldini et al., 1991; Simpson et al., 1983). There is a concomitant decrease in Glut-4 in the P3 fraction. Syndet immunoreactivity in basal and insulin-stimulated cells was enriched in P1, suggesting that syndet is primarily localized at the plasma membrane of 3T3-L1 adipocytes. Additional syndet immunoreactivity co-sedimented with membranes containing the translocatable pool of Glut4 (fraction P3) indicating that a smaller fraction of adipocyte syndet may be localized in intracellular organelles and/or that fraction P3 contains some plasma membranes. No syndet immunoreactivity is found in the cytoplasm fraction (S3) or in fraction P4. Data from three separate experiments indicate that syndet is not translocated to the plasma membrane in response to insulin.

Fig. 7 shows 3T3-L1 adipocytes immunostained with affinity-purified syndet antibody. Syndet immunofluorescence was predominant at the periphery of adipocytes suggesting a plasma membrane localization. The staining at the periphery of cells had a patchy appearance possibly due to accumulation of membranes at the confocal plane or to the clustering of syndet moieties at specialized locations in the plasma membrane. When the image was focused nearer to the nucleus, additional syndet staining appeared as intracellular patches of immunofluorescence. These regions may represent invaginations of the plasma membrane tangential to the plane of the confocal image. Another possibility is that syndet may partially localize to intracellular organelles.

To investigate if there is an intracellular fraction of syndet, we double labeled adipocytes with wheat germ agglutinin (Fig. 8A) and syndet affinity purified antibodies (B). Fluorescein-coupled wheat germ agglutinin was added to the incubation medium of intact 3T3-L1 adipocytes at 4°C and stained the outer surface of 3T3-L1 adipocytes. Syndet immunoreactivity was distributed along the cell profile indicating co-localization at the plasma membrane. In addition, syndet antibodies labeleled a Golgi-like structure in the vicinity of the nucleus. The staining both at the plasma membrane and at the perinuclear location was undetectable in experiments where antibodies were pre-incubated with peptide-coupled Sepharose beads (D). These data show that there is an intracellular pool of syndet. Bright puncta of fluorescence which were present at variable levels both in samples treated with syndet antibodies and in the controls are likely to be part of the backgound staining.

To localize syndet at an ultrastructural level, immunoelectronmicroscopy studies were performed in kidney, a tissue with high expression of syndet. The micrograph shown in Fig. 9A shows the basolateral side of proximal tubule cells lying on the basal lamina. Syndet immunoreactivity is evident along the plasma membranes either at cellular infoldings or in regions of contact between adjacent cells (arrows). At higher magnification of two other fields (Fig. 9B and C) shows that syndet immunoreactivity is clearly associated with plasma membranes. The immunoreactivity is visible in patches along the plasma membrane. These patches seemed to occur on both sides of adjacent plasma membranes suggesting that labeling occurs in regions which are accessible to the antibody. Another possibility is that syndet resides in specialized locations of the plasma membrane. Syndet staining was never observed in regions of contact of cell membranes with the basal lamina, possibly because of an exclusion of syndet from these contact sites. No syndet immunoreactivity was visible in the nuclear and mitochondrial membranes or in cellular tubular-vesicular structures. The control experiment in which labeling was done with only the secondary antibody showed no specific staining (Fig. 9D).

To search for a SNAP-25 homologue expressed in nonneuronal tissue, we screened an adipocyte library with two degenerate primers conserved in all known SNAP-25 isoforms. With this approach, we identified the cDNA for a novel protein designated as syndet. Syndet and SNAP-25 are similar because: (1) there is a high percentage of amino acid identity in two regions at the amino- and carboxyl-terminal halves of both proteins; (2) both structures indicate a high probability of forming amphipathic α-helices and intermolecular coiled coils; (3) syndet contains a cysteine-rich domain which is conserved in all members of the SNAP-25 family; (4) there is an absence of predicted signal sequences or transmembrane domains; (5) syndet, as well as SNAP-25, behaves as an intrinsic membrane protein.

SNAP-25, syntaxin and VAMP contain sets of heptad repeats predicted to form amphipathic α-helices (Calakos et al., 1994; Chapman et al., 1994; Dascher et al., 1991; Inoue et al., 1992) likely to assemble as intermolecular coiled coil structures (Cohen and Parry, 1990; Lupas et al., 1991). SNAP-25 binds via these domains to syntaxin and VAMP (Calakos et al., 1994; Chapman et al., 1994) to constitute a SNARE complex necessary for synaptic vesicle fusion. Syndet contains sets of heptad repeats likely to arrange in coiled coil structures in two separate domains of the protein. Both domains reside in regions of high identity between syndet and SNAP-25. The structural similarity of syndet and SNAP-25 suggests that syndet interacts with members of the syntaxin and VAMP/cellubrevin families of proteins to form a novel type of SNARE.

Despite the structural similarities between SNAP-25 and syndet, the peptide sequences of SNAP-25 recognized by botulinum neurotoxin serotypes A and E are not conserved in syndet. Botulinum neurotoxin serotype A hydrolyzes SNAP-25 at the Gln197-Arg198 peptide bond (Binz et al., 1994; Blasi et al., 1993a; Schiavo et al., 1993). However, the SNAP-25 Gln197 residue is replaced in syndet by the Thr202 residue. Since the positively charged Gln residue is required for peptide hydrolysis (Montecucco and Schiavo, 1994), it is likely that botulinum neurotoxin serotypes A will not cleave the protein. The SNAP-25 peptide recognized by botulinum neurotoxin serotypes E is Asp179-Arg180-Ile181-Met182 (Binz et al., 1994; Schiavo et al., 1993; Montecucco and Schiavo, 1994). This sequence is replaced in syndet by Gln184-Lys185-Πe186-Thrш. Botulinum neurotoxin serotype E recognizes a positively charged residue at the P1 site and a hydrophobic residue at the P1‘ site (Montecucco and Schiavo, 1994). This requirement is fulfilled by syndet Lys185-Ile186 peptide. However, it is unclear if substitutions at amino acid residues adjacent to the peptide will allow hydrolysis by botulinum neurotoxin serotype E.

Syndet is bound to membranes because: (1) the protein is associated with membrane-containing fractions after subcellular fractionation of 3T3-L1 adipocytes; and (2) syndet is recovered in the membrane pellet after carbonate extraction at pH 11.5 and thus behaves as an intrinsic membrane component (Fujiki et al., 1982). Syndet and SNAP-25 lack signal sequence and transmembrane domains. Therefore syndet is, like SNAP-25, a cytosolic protein bound to membranes. SNAP-25 is associated with the lipid bilayer via palmitoylation at one or more of the cysteines in the cysteine-rich domain (Hess et al., 1992). It is likely that a similar thioacylation within the conserved cysteine-rich domain of syndet binds the protein to membranes. Differences in subcellular localization of neuronal SNAP-25a and SNAP-25b have been attributed to modifications in the conserved cysteine quartet (Bark et al., 1995). Unlike the other SNAP-25 isoforms, syndet maintains the arrangement of all 5 cysteine residues found in Torpedo SNAP-25. The importance of this domain for syndet subcellular localization remains to be established.

The SNARE hypothesis predicts that the basic machinery involved in docking/fusion of synaptic vesicles operates in constitutive traffic as well as in other specialized secretory pathways (Bennett and Scheller, 1993). SNAP-25 isoforms are localized in brain, ganglia or neuroendocrine cells, indicating a specific role in synaptic vesicle and dense core granule docking/fusion events (Oyler et al., 1989; Risinger et al., 1993; Sadoul et al., 1995; Sollner et al., 1993a,b). Syndet and the recently identified SNAP-23 (Ravichandran et al., 1996) are expressed in many tissues, differently from all the other SNAP-25 isoforms known so far. Mouse syndet and human SNAP-23 amino acid sequences are 86% identical, unlike mouse and human SNAP-25 which are 100% identical. Thus, syndet and SNAP-23 may be products of closely related genes, rather than species-specific isoforms of the same protein. The identification of SNAP-25 related proteins expressed in non-neuronal tissues provides evidence that all the components of the neuronal SNARE complex are expressed in many cell types.

We used four different approaches to determine that syndet is localized at the plasma membrane with the following results: (1) the majority of syndet immunoreactivity is recovered in a plasma membrane fraction of 3T3-adipocytes; (2) the immunofluorescence distribution of endogenous syndet in adipocytes, determined by confocal microscopy, shows a peripheral staining consistent with plasma membrane localization; (3) syndet immunoreactivity co-localizes at the cell surface with wheat germ agglutinin staining; (4) immunoelectronmicroscopy of kidney proximal tubule cells shows syndet immunoreactivity at the plasma membrane. Therefore we conclude that syndet is localized at the plasma membrane. We also find that, in 3T3-L1 adipocytes, a fraction of syndet protein has an intracellular, Golgi-like distribution. Both SNAP-25 and Syntaxin-4 have an intracellular localization in organelles involved in synaptic vesicles recycling (Walch-Solimena et al., 1995). The intracellular fraction of syndet may constitute a pool of protein recycling to the plasma membrane. Another possibility is that newly synthetized syndet may associate to intracellular organelles before being delivered to the plasma membrane.

SNAP-25 binds to syntaxin 1 and 4, but not to syntaxin 2 and 3 and potentiates binding of VAMP-2 with syntaxin 1A and to syntaxin 4 (Chapman et al., 1994; Pevsner et al., 1994). These experiments indicate that combinatorial pairing of specific syntaxin isoforms with SNAP-25 form heterogeneous receptors at the plasma membranes. The structure and localization of syndet suggests that the protein associates with members of the syntaxin family to form novel plasma membrane t-SNAREs. This possibility supports the hypothesis that multiple vesicle receptors function at the cell surface and that correct pairing of t- and v-SNAREs confers specificity for vesicle fusion. However, the finding that cellubrevin is associated with synaptic-like microvesicles in neuroendocrine cells indicates that v-SNAREs are shared by the constitutive and regulated pathways (Chilcote et al., 1995). How is it possible to reconcile the existence of multiple vesicle receptors at the plasma membrane with the apparent restricted repertoire and lack of specificity of v-SNAREs? A possible answer to this question is that some of the v-SNAREs for exocytotic and recycling vesicles remain to be identified. Interestingly, complete cleavage of cellubrevin by tetanus toxin inhibits only partially the recycling of transferrin receptor-containing vesicles (Galli et al., 1994). These data indicate that a v-SNARE other than cellubrevin may be also involved in recycling of vesicles to the plasma membrane. The identification of syndet may offer tools to find new members of the VAMP/cellubrevin family of proteins.

Syndet, as well as some members of the syntaxin isoforms (Bennett et al., 1993), is differentially expressed in tissues. These findings suggest that regulated expression of t-SNARE components changes the abundance and type of vesicle receptors at the cell surface of different cells. The number and subunit composition of t-SNAREs at the cell surface may regulate the type of vesicles allowed to fuse with the entire plasma membrane or with specific plasma membrane domains in polarized cells. Therefore, the structure and localization of specific t-SNAREs may play important roles in determination of cell morphology and polarity. The identification of syndet opens new avenues for investigating the localization and function of SNAREs at the cell surface.

The authors thank Dr Reinhard Jahn for generously providing the antibodies against Rab3A, Dr Michael G. Smith for help with the confocal microscope, Dr Timothy E. McGraw, Dr Harvey F. Lodish and Dr Bernard Thorens for critical comments on the manuscript, Dr Monica Visintin for advice on ancient Greek and Dr Phillip Leopold for advice on immunoelectronmicroscopy. P.E.S. was funded by a fellowship from the Swiss National Science Foundation.