Role of the Cysteine-rich Domain of the t-SNARE Component, SYNDET, in Membrane Binding and Subcellular Localization*

Wild-type syndet is efficiently recruited at the plasma membrane in transfected AtT-20 cells. A deletion at the cysteine-rich domain abolishes palmitoylation, membrane binding, and plasma membrane distribution of syndet. Syndet, SNAP-25A, and SNAP-25B share four cysteine residues, of which three, Cys2, Cys4, and Cys5, are absolutely conserved in all three homologs. Mutations at any pair of cysteines within cysteines 2, 4, and 5 shift syndet from the cell surface into the cytoplasm. Thus, at least two cysteines within the conserved triplet are necessary for plasma membrane localization. Syndet C1S/C3S, with substitutions at the pair Cys1 and Cys3, distributes to the plasma membrane, a Golgi-like compartment, and the cytosol. We conclude that Cys1 and Cys3 are not absolutely necessary for membrane binding or plasma membrane localization. Our results show that the cysteine-rich domain of syndet plays a major role in its subcellular distribution.

SNAP-25 homologues belong to a small family of structurally similar proteins thought to function in neurotransmission and axonal growth (1)(2)(3)(4)(5). SNAP-25 is very abundant in neurons where it localizes to presynaptic membranes and synaptic vesicles (1,6). It has been shown that SNAP-25 and syntaxin-1, two proteins localized at the plasma membrane, form a complex with VAMP-2, an integral membrane protein of synaptic vesicles (3). This heterotrimeric complex is a receptor (SNARE) for SNAP and NSF, two soluble factors required for vesicular traffic (3). A current model predicts that SNARE receptors at an acceptor compartment (t-SNAREs) interact in a "lock and key" fashion with SNARE receptors at the surface of transport vesicles (v-SNAREs) (reviewed in Ref. 7). Recently, it has been proposed that SNARE components, by themselves, are sufficient to mediate fusion between two lipid bilayers (8) and pairing of specific v-SNAREs and t-SNAREs determines specificity of organelle traffic. In agreement with this hypothesis, there are multiple members of the VAMP and syntaxin family of proteins, each localized at a specific intracellular compartment (9). A homolog of SNAP-25, called SNAP-23, was identified from a human B lymphocyte library by yeast double-hybrid screening (10). SNAP-23 binds tightly to several syntaxins and VAMPs in vitro (10). Whether individual syntaxin, VAMP-2, and SNAP-25 homologs work in specific exocytotic events is still unknown.
We cloned syndet, which is 87% identical to human SNAP-23 from a mouse adipocyte library (11). Syndet and SNAP-23 are expressed in many tissues (11)(12)(13)(14). Syndet and syntaxin-4 are localized at the plasma membrane of adipocytes (11,12,14) and VAMP-2 is localized in glucose transporter-containing vesicles (14 -16). It has been proposed that the SNARE complex, syndet⅐syntaxin-4⅐VAMP-2 mediates insulin-dependent exocytosis of glucose transporter-containing vesicles to the plasma membrane of adipocytes (17). Thus, syndet in adipocytes, like SNAP-25 in neurons, seems to be involved in exocytosis of intracellular vesicles at the cell surface. SNAP-25 and syndet are integral membrane proteins, but do not have any predicted signal sequence or transmembrane domains (1,11). Post-translational modifications may provide these proteins with a hydrophobic anchor to membranes. It has been shown that SNAP-25 is palmitoylated through a thioester bond to some or all of its four cysteines clustered at a region near the center of the protein sequence (18,19). Deletion of this region abolishes palmitoylation of SNAP-25 and its binding to membranes (20). Also point mutations at single cysteines of SNAP-25 largely inhibit palmitoylation and membrane binding (19). This study was designed to investigate which domain of syndet is critical for its intracellular localization. In this paper we show that the number and configuration of cysteines at the cysteine-rich domain act as major determinants for the subcellular distribution of syndet.

EXPERIMENTAL PROCEDURES
Syndet Mutagenesis-The original template for polymerase chain reactions (PCR) 1 of syndet was pcDNA1-syndet plasmid (11). Mutations were encoded in the oligonucleotide primers synthesized by Life Technologies, Inc. (Gaithersburg, MD). The enzyme used for PCR was Pwo DNA Polymerase (3 units/reaction, Boehringer Mannheim Corp.) with its supplied reaction buffer. The primers were used at a final concentration of 2 M. The deoxynucleotide triphosphates (Promega, Madison, WI) were used at a final concentration of 200 M. The PCR was carried out on a "GeneAmp PCR System 2400" (Perkin-Elmer, Applied Biosystems Division, Foster City, CA). All restriction enzymes were purchased from Promega.
Delta-syndet mutant was also made from pcDNA1-syndet in two separate PCR reactions, using primers 1 and CGCATCTTACCCGGG-CTTGTTGAGTTCTGTTAAAGTCTT, which encodes an XmaI site, for the first reaction, and primers 2 and AGTGTTACGGTACCCGGGAAT-AGGACAAAGAACTTTGAG, which also encodes an XmaI site, for the second. The products of the first PCR reaction were digested with the restriction enzymes HindIII and XmaI and the products of the second PCR reaction with XmaI and XbaI. These two products were ligated into pCB7 vector to obtain delta-syndet-pCB7.
Syndet C2S/C4S/C5S-pCB7 vector was made using the technique of "overlap extension" PCR (22) with syndet-pCB7 as the template. Primers 1 and GGAAATACACAAACCAGAACACTTGTTGAG were used to make the 5Ј-half of syndet C2S/C4S/C5S and primers 2 and TTGTGTATTTCCCCGTCTAATAGGACAAAG to make the 3Ј-half. The two products were combined in a second PCR, digested with HindIII and XbaI, and subcloned into pCB7.
All other mutants were created using a combination of "long distance inverse" PCR (23) and "touchdown" PCR methods (24) with syndet-pCB7 as the template. The two primers were designed to encode the mutations and align "back to back" on the syndet-pCB7 backbone. The reverse primer was phosphorylated using T4 polynucleotide kinase (Promega) in the supplied reaction buffer along with the addition of ATP (2 mM). The PCR buffer was the same as above, except that it contained N,N-dimethylformamide at a final concentration of 1% (v/v). The PCR parameters were the following: denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 6 min. The annealing temperature was then reduced by 1°C per cycle for the next 11 cycles, and then kept at 44°C for the following 30 cycles. The PCR products were treated with DpnI, to digest the template DNA, purified by low-melt agarose gel electrophoresis, and ligated with T4 DNAligase. Primers TGCATCAGTCCTTGTAATAGGACA and GAGGCCT-GAACACTTGTTGAGTTC were used for making syndet C2S/C4S-pCB7; ATCTGCCCTTCCAATAGGACAAAGAAC and GCAGAGGCCACTAC-ACTTGTTGAGTTC for making syndet C2S/C5S-pCB7; GCAGAGGC-CACTACACTTGTTGAGTTC and CCTTCAAATAGGACAAAGAAC for making syndet C4S/C5S-pCB7; CTCTCCATCTGCCCTTGTAAT and CCCACAAGACTTGTTGAGTTCTGT for syndet C1S/C3S-pCB7; CTC-TGCATCTCCCCTTGTAATAGGACA and CCCACAACACTTGTT for syndet C4S-pCB7; CCCACAACACTTGTT and CTCTGCATCTGCCCT-TCTAATAGGACAAAGAAC for syndet C5S-pCB7; TCCTCACTTCTG-TATTTGCTG and TCCGCAAAGAAACTCATT for syndet delta-BoNT-/E; CTCTGCATCTGCCCTTGTAATAAGGCAACATGGGGA and CCC-ACAACACTTGTTCAGCATAGTGATGGT for syndet ␦␣1. The identity of all the mutants was confirmed by restriction analyses and sequencing.
Cell Culture and Transfection-AtT-20 cells were cultured and transfected as described (25).
Immunofluorescence and Confocal Microscopy-AtT-20 cells were grown on poly-D-lysine-coated glass coverslips and treated for immunofluorescence as described (11). As primary antibodies, anti-syndet affinity purified rabbit polyclonal antibodies at 1:80 dilution (11) iovert 100TV fluorescence microscope (Carl Zeiss, Thornwood, NY) with a Zeiss LSM410 laser scanning confocal attachment was used to obtain the images. The cells were excited with an Argon-Krypton laser using the standard wavelengths for rhodamine and fluorescein isothiocyanate. The images were collected as 1-m thick optical sections. They were processed using Zeiss LSM 3.95 software and Adobe Photoshop 3.0 (Adobe Systems).
Cell Fractionation and Western Blots-Protease inhibitors, leupeptin (6 g/ml) and aprotinin (1.5 g/ml), were purchased from Boehringer Mannheim, Phenylmethylsulfonyl fluoride (2 mM) was from Sigma. Cells were grown on 6-cm plates for at least 72 h and homogenized in 400 l of buffer A (10 mM Tris-Cl, pH 7.4, 150 mM NaCl, 2 mM EDTA, and protease inhibitors) by passing the cells through an insulin needle six times. The homogenates were mixed with an equal volume of SDS sample buffer, boiled, and loaded into the SDS-PAGE gel. In some experiments, a fraction of the homogenate (200 l) was spun at 400,000 ϫ g for 30 min. The supernatant fraction was recovered and the pellet was resuspended in 200 l of buffer A. Each fraction (40 l each) was mixed with an equal volume of SDS sample buffer, boiled, and loaded into the SDS-PAGE gel. Further steps including electrophoresis and Western blotting were carried out as described previously (11).
Metabolic Labeling and Immunoprecipitation of Syndet-Cells expressing syndet, delta-syndet, and syndet C1S/C3S were labeled with ([ 3 H]palmitate (0.5 mCi/ml, NEN Life Science Products Inc., Boston, MA) for 2 h, as described. (19). Cells grown in a 3.5-cm plate were washed twice with buffer A and lysed in 1.0 ml of buffer A with Nonidet P-40 (1% v/v, Sigma) along with protease inhibitors. The homogenate was centrifuged for 10 min at 5,000 ϫ g. Affinity purified syndet antibody (10 l) was added to the supernatants and samples were incubated at 4°C with gentle agitation for 1 h. Protein A-Sepharose beads (40 l dry beads, Boehringer Mannheim) were added to the samples, which were further incubated at 4 o C for 45 min while being agitated. After centrifugation, pelleted immunobeads were washed three times with lysis buffer A and mixed with SDS-PAGE sample buffer. Samples were boiled and analyzed by SDS-PAGE electrophoresis and fluorography as described (20). Preliminary experiments were run to determine that syndet, delta-syndet, and syndet C1S/C3S were immunoprecipitated to the same extent by affinity purified syndet antibodies. To visualize immunoprecipitated wild-type and mutated syndet by Western blot, immunobeads were mixed with sample buffer without reducing agents.

Targeting of Exogenous Syndet in Transfected AtT-20 Cells-
The endogenous level of syndet protein in AtT-20 cells is undetectable by Western blot and immunofluorescence (see below). We determined the distribution of exogenous syndet in stably transfected AtT-20 cells by immunofluorescence (Fig. 1). Syndet immunoreactivity was seen along the plasma membrane of the cell body, the neurite-like processes and the tips (Fig. 1B). Staining for SNAP-25, like syndet, was predominant along the entire plasma membrane (Fig. 1A). However, SNAP-25 antibodies also stained a region near the nucleus and the cytoplasm within the tip. We have shown elsewhere by immunoelectron microscopy that endogenous syndet in adult mouse kidney is found predominantly at the plasma membrane (11). At the ultrastructural level, we confirmed that exogenous syndet in AtT-20 cells was indeed localized at the plasma membrane. Fig. 2 shows that exogenous syndet (12-nm gold particles) and endogenous SNAP-25 (6-nm gold particles) are localized at the plasmalemma in agarose-embedded broken cells. There was no syndet or SNAP-25 labeling associated with the nuclear membrane, or mitochondria or endogenous murine leukemia virus particles. These results indicate that exogenous syndet expressed in AtT-20 cells is able to localize at the plasmalemma. We conclude that AtT-20 cells can be used as a model to study the role of various syndet domains in plasma membrane localization.
Domains of Syndet Involved in Plasma Membrane Localization-We prepared three deletion mutants of syndet to find domains responsible for targeting syndet to the plasma membrane (Fig. 3, upper panel). The deletion mutant delta-syndet lacks the syndet amino acid sequence from Cys 79 to Cys 85 . Pro 86 was left unchanged, and Cys 87 was mutated to a glycine. Thus, delta-syndet does not have any cysteines. Syndet ␦␣1 lacks the amino-terminal region which has a high propensity to form coiled-coil structures (11) and a short sequence after the cysteine-rich domain. The cysteine-rich domain was entirely preserved in the syndet ␦␣1 mutant. Syndet delta-BoNT/E is the expected proteolysis product of mouse syndet protein treated with the botulinum toxin E (BoNT/E) (26). Syndet delta-BoNT/E lacks approximately 40% of the carboxyl-terminal region with high probability to form coiled-coil structures (11). Western blots of homogenates from AtT-20 cells expressing wild type syndet (26 kDa), delta-syndet (25 kDa), syndet ␦␣1 (20 kDa), and syndet delta-BoNT/E (23 kDa) are shown in Fig. 3, A-C.
Deletion of the entire cysteine-rich domain of syndet (deltasyndet) shifted the distribution of syndet immunoreactivity from the plasma membrane to a diffuse intracellular localization (Fig. 4B). The localization of endogenous SNAP-25 was not changed by expression of delta-syndet (Fig. 4A). We conclude that the cysteine-rich domain of syndet is necessary for target- ing the protein to the cell surface. Syndet ␦␣1 and syndet delta-BoNT/E immunofluorescence staining accumulated at the plasma membrane of the cell body (Fig. 4, C and E), of the neurite-like processes (not clearly visible in the optical sections shown in Fig. 4) and of the tips (Fig. 4, C and D). Thus, the subcellular distribution of syndet ␦␣1 and syndet delta-BoNT/E was similar to the distribution of exogenous wild-type syndet (compare with Fig. 1). These experiments indicate that the entire predicted ␣-helix at the amino terminus of syndet and at least 40% of the helix at the carboxyl terminus are not required for targeting syndet to the plasma membrane. Unlike these regions, the cysteine-rich domain is necessary for plasma membrane targeting. transiently transfected AtT-20 cells was indistinguishable from that observed in the stable AtT-20 cell lines. Syndet C2S/C4S/ C5S was found in the cytoplasm (Fig. 6A), and had a distribution similar to delta-syndet (Fig. 4B). This experiment indicates that the conserved -Cys 2 -Cys 4 -Cys 5 -triplet is critical for syndet localization at the plasma membrane. Syndet C4S/C5S (Fig. 6B), syndet C2S/C4S (Fig. 6C), and syndet C2S/C5S (Fig.  6D) were also found in the cytoplasm and did not localize at the plasma membrane. Syndet proteins with single point mutations at C5S (Fig. 6E) or C4S (Fig. 6F) were localized at the plasma membrane. Thus, a mutation of any pair of cysteines within the conserved triplet is sufficient to prevent plasma membrane localization. Single substitutions do not alter syndet distribution. We conclude that at least 2 cysteines within the -Cys 2 -Cys 4 -Cys 5 -triplet are necessary for localization of syndet at the plasma membrane.

Requirement of Cysteines for Syndet Localization at the
Subcellular Localization of Syndet C1S/C3S-Syndet distribution in AtT-20 cell lines stably expressing the syndet C1S/ C3S protein was examined. In contrast to wild-type syndet (Fig. 1B), the mutant, syndet C1S/C3S, was found at an intracellular site as well as at the plasma membrane (Fig. 7B). Double staining with SNAP-25 antibodies indicate that endogenous SNAP-25 is predominantly at the plasma membrane (Fig. 7A). We determined the relative distribution of ACTH and syndet immunoreactivities in AtT-20 cells expressing wild-type syndet and syndet C1S/C3S (not shown). ACTH immunoreactivity is concentrated in a Golgi-like region near the nucleus and at the tips (25) where ACTH-containing granules accumulate (28). Wild-type syndet did not co-localize with ACTH immunoreactivity either at the cell body or at the tips. Syndet C1S/C3S co-localized with ACTH immunoreactivity in the Golgi-like compartment, but not at the tips of the processes. These results show that mutations at Cys 1 and Cys 3 : (a) shift a fraction of syndet from its plasma membrane localization to a Golgi-like intracellular compartment; (b) do not induce re-distribution of syndet to dense core granules.
Role of the Cysteine-rich Domain in Syndet Binding to Membranes-AtT-20 cells expressing wild-type syndet, delta-syndet, and syndet C1S/C3S were homogenized in a buffer containing either 150 mM NaCl or 100 mM Na 2 CO 3 at pH 11.5. Buffers containing sodium carbonate at pH 11.5 are known to dissociate peripheral proteins from membranes (29). The homogenates were then centrifuged at 400,000 ϫ g for 30 min to obtain a membrane-containing pellet and a cytosol-containing supernatant. Wild type SNAP-25 and syndet were mostly or entirely recovered in the pellet fraction (Fig. 8). We conclude that exogenous syndet in AtT-20 cells is an integral membrane component, like endogenous syndet in adipocytes. Overexpression of syndet did not affect the binding of endogenous SNAP-25 to membranes (Fig. 8). Delta-syndet was almost entirely recovered in the supernatant fraction, indicating that deletion of the cysteine-rich domain makes the protein soluble (Fig. 8). Approximately 50% of syndet C1S/C3S was tightly associated to membranes (Fig. 8). We conclude that Cys 1 and Cys 3 are important, but not absolutely necessary, for syndet binding to membranes.
Syndet Is Palmitoylated-To determine whether syndet is palmitoylated, we incubated transfected AtT-20 cells expressing wild-type syndet, delta-syndet, and syndet C1S/C3S with [ 3 H]palmitic acid. Palmitoylated proteins were analyzed by immunoprecipitation and SDS-PAGE fluorography (Fig. 9). Wild-type syndet and syndet C1S/C3S, but not delta-syndet, incorporated labeled palmitate. These experiments and the ones in Fig. 8 indicate that the cysteine-rich motif of syndet is palmitoylated and that incorporation of fatty acid is necessary for syndet binding to membranes.

DISCUSSION
Pairing of individual SNAREs before fusion may be at least one of the mechanisms by which cells maintain different organelles as unique compartments. In this light, it is essential that individual SNARE components be targeted to their correct destination. In this paper, we have identified the cysteine-rich region of the t-SNARE component, syndet, as critical for subcellular localization of syndet. We find that deletion of the cysteine-rich domain of syndet abolishes membrane binding. This result is in agreement with the finding that a mutant SNAP-25, lacking its cysteine-rich domain, is soluble (20). SNAP-25 is palmitoylated in the central nervous system and in transfected cells (18,19). Syndet, syndet C1S/C3S, but not the mutant lacking the cysteine-rich domain, are palmitoylated in the transfected AtT-20 cell model. Thus, it is likely that fatty acylation is necessary for the binding of syndet to membranes. Moreover, we find that the cysteine-rich domain of syndet is essential for targeting the protein to the plasma membrane. We conclude that the cysteine-rich domain of syndet has a dual role in membrane binding and plasma membrane targeting. Unlike the deletion of the cysteine-rich domain, deletions of 32 amino acids at the amino terminus domain or 25 amino acids at the carboxyl terminus domain of syndet do not have an appreciable effect on protein distribution.
Our data show that the distribution of syndet which has been mutated at three cysteines, syndet C2S/C4S/C5S, is shifted from the plasma membrane to a diffuse intracellular localization. We also find that syndet C4S/C5S, C2S/C4S, and C2S/C5S are unable to accumulate at the cell periphery and are found in the cytoplasm. Syndet C4S and syndet C5S are both localized at the plasma membrane. Our observations indicate that: 1) at least two cysteines within the -Cys 2 -Cys 4 -Cys 5 -triplet must be substituted to prevent syndet localization at the cell surface; 2) single and double mutations of cysteines within the -Cys 2 -Cys 4 -Cys 5 -triplet have the same effect on syndet distribution, irre- AtT-20 cells expressing wild-type syndet, delta-syndet, and syndet C1S/C3S were grown in 6-cm plates. Cells were homogenized in 400 l of the following buffers: 10 mM Tris-HCl, 1 mM EDTA, 150 mM NaCl, pH 7.4 (upper panel); 0.1 M sodium carbonate, pH 11.5, 1 mM EDTA (lower panel). Buffers contained protease inhibitors as indicated under "Experimental Procedures." Equal volumes of homogenates (H), membrane-containing pellets (P), and cytosol-containing supernatants (S) were prepared as described under "Experimental Procedures." Blots were probed with antibodies against syndet and SNAP-25. This experiment was repeated three times with similar results. spective of which residue is changed. We conclude that Cys 2 , Cys 4 , and Cys 5 are critical for plasma membrane localization and may have similar and somewhat overlapping roles.
Doubly lipid modified peptides are stably anchored to membranes (30). Since syndet and SNAP-25 have multiple cysteines, some cysteines may target these proteins to intermediate localizations along the biosynthetic route or to specialized domains at the plasma membrane. Indeed, it appears that SNAP-25 A, SNAP-25 B, and SNAP-23 accumulate in specific domains at the cell surface (5,(31)(32)(33). Modifications at cysteines may also modulate their binding to membranes (19) or the formation of SNARE complexes (34) .
Because syndet Cys 1 and Cys 3 are not conserved in SNAP-25 B and SNAP-25 A, it is reasonable to conclude that modifications at these residues may account at least in part for differences in the localization and function of these proteins. We have generated a mutant syndet protein in which both cysteines are substituted with serines. We find that, unlike the double mutants within -Cys 2 -Cys 4 -Cys 5 -triplet, a fraction of syndet C1S/C3S remains as an integral membrane component at steady state. Consistent with the concept that acylation of syndet is necessary for its binding to membranes, we find that syndet C1S/C3S is palmitoylated. We conclude that Cys 1 and Cys 3 , unlike Cys 2 , Cys 4 , and Cys 5 , are not absolutely required for membrane binding. We also find that mutations at Cys 1 and Cys 3 of syndet result in localization of the mutated protein to the plasma membrane, as well as a Golgi-like compartment and the cytosol. Interestingly, a mutation of a specific cysteine palmitoylation site in Src protein kinase p56 lck , can shift a fraction of the protein from the plasma membrane to a Golgi compartment (35). Our results support the concept that the cysteine-rich domain of syndet plays a major role in subcellular localization of the protein. We conclude that the distribution of syndet, like that of Src protein kinase p56 lck , may be influenced by the configuration of the cysteines.
SNAP-25 associates with the Golgi compartment at the cell body of the neurons and then undergoes fast axonal transport as an acylated protein (18,36). Thus, SNAP-25 reaches the plasma membrane by association with transport vesicles. We find that syndet C1S/C3S is found in the cytosol, the Golgi, and the plasma membrane. The distribution of mutated syndet may mimic the steps of syndet or SNAP-25 biosynthetic routes. If this is the case, then acylations within the -Cys 2 -Cys 4 -Cys 5triplet may be sufficient to bind syndet to a Golgi localization. Acylation(s) at Cys 1 and/or Cys 3 may shift wild-type syndet from the trans-Golgi network to the plasma membrane. In the absence of Cys 1 and Cys 3 transit to the plasma membrane is slow and mutated syndet C1S/C3S accumulates at the Golgi localization. Alternatively, it is also possible for mutated syndet C1S/C3S to localize both at the Golgi and at the plasma membrane directly, from the cytosol. In mast cells, SNAP-23 relocates from the plasma membrane to dense core granules in response to stimulation (37). Our results indicate that modifications at the cysteine-rich motif can induce redistribution of syndet to an intracellular compartment. However, in AtT-20 cells, syndet C1S/C3S is redistributed to an intracellular compartment different from dense core granules. More work is necessary to understand whether modifications at the cysteinerich motif play a role in changes of SNAP-23/syndet or SNAP-25 cell distribution under different conditions.
In conclusion, our paper shows that the cysteine-rich domain of syndet is a major determinant for the subcellular localization of the protein. We propose that modifications at the clustered cysteines may control syndet distribution and function in cells. FIG. 9. Syndet is palmitoylated at the cysteine-rich domain. Cells expressing syndet, delta-syndet, and syndet C1S/C3S were labeled with [ 3 H]palmitate for 2 h, as described (19). Wild-type and mutated syndet proteins were immunoprecipitated using the affinity purified anti-syndet antibody. Immunoprecipitates were analyzed by SDS-PAGE electrophoresis and fluorography (see "Experimental Procedures"). The gel was exposed for 20 days. This experiment was repeated three times with similar results.