A new syntaxin family member implicated in targeting of intracellular transport vesicles.

Despite the central role vesicular trafficking occupies in protein targeting, the molecular coding of the trafficking signals and the mechanism of vesicle docking and fusion are just beginning to be understood. We report here the cloning and initial characterization of a new member of the syntaxin family of vesicular transport receptors. Syntaxin 6 is a 255-amino acid protein with two domains predicted to form coiled-coils, as well as a carboxyl-terminal membrane anchor. Syntaxin 6 is broadly expressed and localizes in the region of the Golgi apparatus. In vitro binding studies established that syntaxin 6 binds to α-soluble NSF attachment protein (α-SNAP). The sequence homology, topology, localization, and α-SNAP binding suggest that syntaxin 6 is involved in intracellular vesicle trafficking.

Despite the central role vesicular trafficking occupies in protein targeting, the molecular coding of the trafficking signals and the mechanism of vesicle docking and fusion are just beginning to be understood. We report here the cloning and initial characterization of a new member of the syntaxin family of vesicular transport receptors. Syntaxin 6 is a 255-amino acid protein with two domains predicted to form coiled-coils, as well as a carboxyl-terminal membrane anchor. Syntaxin 6 is broadly expressed and localizes in the region of the Golgi apparatus. In vitro binding studies established that syntaxin 6 binds to ␣-soluble NSF attachment protein (␣-SNAP). The sequence homology, topology, localization, and ␣-SNAP binding suggest that syntaxin 6 is involved in intracellular vesicle trafficking.
Eukaryotic cells contain highly specialized organelles that are defined by their specific protein complements. The mechanism by which cells are able to route proteins along particular pathways to these various organelles has been the focus of intense investigation. While it was theorized since the 1970's that transport vesicles mediated this process (1), it was not until recently that a convergence of genetic, biochemical, and cell biological approaches permitted significant insights into the molecular mechanisms underlying this process (2,3). The intersection of these approaches took place at the mammalian presynaptic nerve terminal, where a biochemical model for synaptic vesicle docking and fusion has been proposed (4). This model has been put forward as a paradigm for all intracellular vesicle trafficking (5).
Current models suggest that synaptic vesicle proteins (v-SNAREs) 1 interact specifically with plasma membrane-localized molecules (t-SNAREs), and that these complexes act as a scaffold for the assembly of the fusion apparatus (6,7). At the synapse a promenade of protein-protein interactions mediate the docking, priming, and fusion of synaptic vesicles (8). The first step in defining this mechanism was the identification of the proteins involved. Syntaxin 1a and SNAP-25 are the t-SNAREs present on the plasma membrane, while VAMP or synaptobrevin and synaptotagmin are the v-SNAREs on synaptic vesicles (4). A complex of these four membrane-bound proteins can be isolated in vivo and reconstituted in vitro (7,9,10). Concurrent with the dissociation of synaptotagmin, this complex binds the cytosolic protein ␣-SNAP, which was originally isolated through its requirement in a reconstituted Golgi transport vesicle fusion assay (11). The model predicts that the ATPase NSF then binds ␣-SNAP and, through the hydrolysis of ATP, primes the complex for the eventual calcium influx leading to vesicle fusion. Thus, it is the interaction of the t-SNAREs with their cognate v-SNAREs, which contributes to the specificity, and it is the binding of ␣-SNAP to the members of this complex that is essential to the transition from a docking state toward a fusion competent one.
The fundamental question that we begin to address in this study is: can the molecular machinery mediating vesicle fusion at the synapse serve as a model for all intracellular vesicle trafficking? The genetically tractable organism Saccharomyces cerevisiae has played a key role in helping to understand mammalian vesicle trafficking. For example, it has been shown that the yeast Golgi to plasma membrane constitutive secretory and mammalian synaptic transmission pathways are evolutionarily related (12). Syntaxin 1a, SNAP-25, VAMP, ␣-SNAP, and NSF all have yeast homologues. This molecular consistency appears to hold up in studies of intracellular trafficking pathways aside from Golgi to plasma membrane transport. Mammalian homologues of Bet1 and Sec22, yeast proteins involved in endoplasmic reticulum to Golgi trafficking, have been identified recently (13).
Not only do protein family members span species, but they also stretch across different intracellular trafficking pathways. The best example of this is the syntaxin family, whose members specifically localize to several organelles. For example, syntaxin 5 is present on the cis Golgi (14) and mediates endoplasmic reticulum to Golgi transport (15). Another member of the syntaxin family is yeast Pep12 (also termed VPS6). It was initially identified in a yeast genetic screen for proteins involved in vacuolar function (16) and is believed to mediate Golgi to vacuole transport (17).
In order to determine if the synapse can serve as an accurate model for all intracellular vesicle trafficking, we set out to identify other mammalian members of the syntaxin family present on distinct intracellular organelles. We characterized overlapping cDNAs encoding a 255-amino acid protein with two domains predicted to form coiled-coils and a carboxyl-terminal transmembrane domain. This protein shows homology to Pep12 and the mammalian syntaxin family members. An unexpected result was that this protein also shows significant identity to SNAP-25. Subcellular localization determined that this protein resides in the Golgi region. In vitro binding studies demonstrate its specific interaction with ␣-SNAP. Due to its structure, sequence homology, and protein-protein interactions we classify this protein as member of the syntaxin family. The data presented here are consistent with a role for syntaxin 6 in Golgi vesicle trafficking.

EXPERIMENTAL PROCEDURES
cDNA Cloning and Sequencing-The Human Genome Sciences, Inc. data base of expressed sequence tags (ESTs) was searched by comparing its translated contents to the yeast Pep12 primary sequence. One clone was found to be 21% identical to Pep12 over a 112-amino acid domain. This EST clone was initially discovered using established EST methods by Human Genome Sciences, Inc. (18,19). Custom-designed oligonucleotides were used in the polymerase chain reaction using a human brain cDNA library (Stratagene, La Jolla, CA) as the template. The expected product was recovered and used as a probe to screen a ZAP II rat brain library (Stratagene, La Jolla, CA). 6 ϫ 10 5 plaques were screened and resulting clones were converted to pBluescript KS and double strand sequenced using Sequenase (Amersham Corp.) and custom oligonucleotide primers.
RNA Distribution Analysis-32 P-Labeled probes were used to screen a multiple tissue Northern blot of poly(A) ϩ RNA (Clontech, Palo Alto, CA). Hybridization was performed at 42°C overnight in 5 ϫ SSPE 750 mM NaCl, 50 mM NaH 2 PO 4 -H 2 O, 5 mM EDTA, pH 7.4 and 50% formamide. Other hybridization and washing conditions were performed as described in the manufacturer's protocol. The washed blot was exposed to film overnight at Ϫ70°C.
Syntaxin 6 Constructs-Full-length syntaxin 6 without its carboxylterminal hydrophobic domain, as well as its deletion mutants, were prepared by polymerase chain reaction. The first mutant consisted of amino acids 1-166, the second mutant amino acids 166 -235, and the third one amino acids 26 -235. The DNA sequences of all deletion constructs were confirmed by dideoxy nucleotide sequencing, and the inferred amino acid sequences were identical to wild type. The constructs were subcloned into pGEX-KG (Qiagen, Chatsworth, CA), transformed into Escherichia coli, and induced by standard methods (6).
Antibody Production and Purification-Recombinant syntaxin 6 was expressed and purified as described above. This purified protein was used to immunize a rabbit (Josman laboratories, Napa Valley, CA). An affinity column was produced by binding recombinant syntaxin 6 to an equal volume slurry of Affi-Gel 10 and 15. The crude serum was purified by running it through this affinity column and eluting with 0.1 M acetic acid, pH 2.7. The eluted antibodies were then neutralized with 1 M Tris base.
Protein Distribution Analysis-Rat heart, brain, spleen, lung, liver, skeletal muscle, kidney, and testes were dissected, homogenized, and centrifuged at 3000 ϫ g for 5 min to yield post-nuclear supernatants. SDS-PAGE was performed on 10 g of protein from each tissue. These proteins were then transferred to nitrocellulose and subjected to antisyntaxin 6 (1/1500) Western blot analysis. ECL was performed as per the manufacturer's protocol (Amersham Corp.).
Membrane Extraction Studies-Rat brain post-nuclear supernatant was centrifuged at 20,000 ϫ g for 15 min to pellet crude membranes. This pellet was washed with homogenization buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM dithiothreitol) and resuspended in homogenization buffer to a final protein concentration of 2 mg/ml. This was aliquoted into five tubes, and an equal volume of each of the following extraction buffers was added: water only, 3 M NaCl, 0.4 M Na 2 CO 3 , pH 11.4, 4 M urea, and 4% Triton X-100. These solution were incubated at 4°C for 30 min and then centrifuged at 100,000 ϫ g for 20 min. The supernatant was removed, and the pellet was resuspended in an equal volume of homogenization buffer. Both the supernatant and pellet suspension were analyzed by SDS-PAGE and Western blot.
In Vitro Binding Assay-Fusion protein beads, soluble ␣-SNAP preparation, and binding assays were prepared as described (9,20)

A New Mammalian Syntaxin Family
Member-Genetic studies in yeast have identified a panel of genes coding for proteins involved in intracellular transport (21). These studies have advanced our understanding of mammalian vesicle transport in two significant ways. First, in areas where mammalian studies were further progressed, such as exocytosis, later yeast studies corroborated the role of several proteins implicated in mammalian vesicle transport (12). Second, in areas where yeast studies are more advanced, such as in endoplasmic reticulum to Golgi transport, they have allowed the isolation of several novel mammalian homologues (13). Pep12p is a yeast member of the syntaxin family and plays a role in Golgi to vacuole transport. We exploited the availability of large data bases of DNA sequence information to identify a human EST showing significant homology to Pep12. We used this EST to isolate overlapping clones containing the 255-amino acid open reading frame shown in Fig. 1. Several stop codons preceded the predicted starting methionine, which was surrounded by a consensus transcription initiation site (22). The open reading frame ended with a string of 20 hydrophobic amino acids immediately followed by a stop codon. This open reading frame predicts a protein of 29 kDa, which was confirmed by in vitro translation (data not shown).
When the GenBank™ nucleotide data base is translated and searched with the predicted amino acid sequence of syntaxin 6 using the TBLASTN algorithm (23), mammalian SNAP-25 received the highest relevant score (p ϭ 0.0024) and yeast pep12 received the second highest relevant score (p ϭ 0.024). When the amino acid sequences of these proteins were compared in a pairwise fashion using the bestfit algorithm (24), syntaxin 6 was found to be 25% identical and 56% similar to pep12 (Table  I). Syntaxin 6 was found to be 22% identical and 47% similar to SNAP-25.
Syntaxin 6 displays structural characteristics similar to the syntaxin family of vesicle transport proteins. Its 20-most carboxyl-terminal amino acids are hydrophobic and presumably act as a membrane anchor. Using the Coils algorithm (25,26), we determined that syntaxin 6 has two regions with high probabilities of participating in coiled-coils. The first is a short stretch from amino acids 47-71 (90% probability), and the second directly precedes the hydrophobic tail from amino acids 166 -225 (80% probability, see Fig. 5). The protein is also predicted to have a relatively low pI of 4.7. These characteristics are reminiscent of the syntaxin family.
Tissue Expression Patterns-To determine if syntaxin 6 performs a general vesicle trafficking function in all tissues or whether it is restricted to a subset of tissues, we performed Northern and Western blot analysis as shown in Fig. 2. A predominant RNA at 3.0 kb was found in all tissues examined, with relatively higher expression in brain, lung, and kidney. Transcripts of 5.9 and 8.2 kb were also present in most tissues, although in lower abundance. It is unclear whether these larger transcripts represent alternative splice products or distinct, but related, sequences. In addition to the 759 base pairs of coding sequence, our overlapping clones also contained 150 base pairs of 5Ј-untranslated sequence and 1.1 kb of 3Ј-untranslated sequence.
To corroborate the RNA distribution we performed Western blot analysis using affinity-purified antisera prepared by immunizing a rabbit with recombinant syntaxin 6. This antisera recognized a single band of 29 kDa in rat brain post-nuclear supernatant, which was not recognized by pre-immune sera. This band was eliminated by preincubating the antisera with soluble recombinant syntaxin 6 (data not shown). Western blot analysis confirmed the Northern blot analysis as shown in Fig.  2. Syntaxin 6 was found to be widely expressed, with relatively more protein in brain, lung, and kidney. This broad tissue expression would be expected of a protein involved in constitutive vesicle trafficking to a ubiquitous organelle, such as the Golgi or lysosome.
Membrane Association-The carboxyl-terminal 20 amino acids of syntaxin 6 are hydrophobic and likely to act as a membrane anchor. Consistent with this syntaxin 6 was found to reside in the high speed pellet fraction of rat brain post-nuclear supernatant. To determine if syntaxin 6 is indeed an integral membrane protein, we extracted rat brain membranes with a series of reagents (Fig. 3). Syntaxin 6 remained associated with the membrane pellet after extraction with NaCl, urea, and high pH; however, it was present in the supernatant following extraction with the nonionic detergent Triton X-100. These results are consistent with the prediction that syntaxin 6 is an integral membrane protein.
Subcellular Localization-To determine which membrane compartment syntaxin 6 associated with, immunofluorescence was used to visualize its location. Fig. 4 shows the subcellular localization of syntaxin 6 in FAO tissue culture cells. Syntaxin 6 was detected primarily in a juxtanuclear pattern, suggestive of the Golgi apparatus. The Golgi region was not labeled when the antibodies were preincubated with recombinant syntaxin 6 protein (data not shown). To confirm the Golgi region localization, we double-labeled cells with antibodies against the known Golgi markers mannosidase II (27) and rbet1 (13). Syntaxin 6 appeared to substantially colocalize with these two markers (Fig. 4).
Although the Golgi region-specific labeling was predominant, a subset of cells had an additional weak, broadly perinuclear, membranous pattern, which may represent a fraction of syntaxin 6 on vesicles shuttling to another intracellular organelle.
Syntaxin 6 Binds ␣-SNAP-We used an in vitro binding assay to characterize the interaction between syntaxin 6 and ␣-SNAP. Syntaxin 6 was expressed from the pGEX-KG vector in E. coli as a glutathione S-transferase fusion protein, with glutathione S-transferase constituting the amino terminus; the fusion protein was then bound to glutathione-agarose beads. These beads were incubated with increasing amounts of soluble histidine-tagged ␣-SNAP. Following three washes, the amount of bound ␣-SNAP was determined by SDS-PAGE and Western blot.
As shown in Fig. 5, this titration study demonstrated that ␣-SNAP binds syntaxin 6 with an EC 50 between 1 and 5 M. This affinity is consistent with syntaxin 6's homology to both the syntaxin family as well as SNAP-25: the estimated EC 50 for ␣-SNAP's binding to rat syntaxin 1a is 0.5-3 M, while the The percent amino acid identity and similarity between the syntaxin family members and SNAP-25 are displayed. Values above the diagonal represent percent amino acid identity. Values below the diagonal represent percent amino acid similarity with the following conservative changes: D-E, R-K, S-T, N-Q, and I-L-M-V. Optimal alignments and scores were determined with the Bestfit algorithm (gap creation penalty ϭ 3, gap extension penalty ϭ 0.1). Syntaxin 2. Syntaxin 6 is widely expressed.. The major RNA species recognized is 3.0 kb, with relatively weaker hybridization at 5.9 and 8.2 kb. The autoradiogram was exposed overnight. The recognized transcript is more abundant in brain, lung and kidney, yet present in all tissues after 4 days of exposure. Western blot of various rat tissue post-nuclear supernatants. Anti-syntaxin 6 rabbit affinity-purified antisera was used at a 1/1500 dilution and visualized by ECL. A protein migrating at ϳ31 kDa is more abundant in brain, spleen, lung, kidney, and testes, but is present in all tissue after a longer exposure.
EC 50 between ␣-SNAP and SNAP-25 has been reported to be as low as 0.5 M and as high as 15 M (9, 28). As a negative control, we used glutathione-agarose beads with only glutathione S-transferase bound to them; this controls for any background binding of ␣-SNAP to the beads themselves or the glutathione S-transferase portion of the fusion proteins. ␣-SNAP did not bind to these beads, even at 40 M (data not shown).
The binding of ␣-SNAP has been mapped to the carboxylterminal heptad repeat (H3) region of syntaxin 1a. To map the binding site of ␣-SNAP on syntaxin 6, we made three syntaxin 6 deletion constructs using the polymerase chain reaction. The first construct eliminated the carboxyl-terminal heptad repeat region; the second consisted of only this region; and the third deleted the 25 amino-terminal residues of the protein. We expressed these constructs as glutathione S-transferase fusion proteins and used them in the same in vitro binding protocol as above. Interestingly, as shown in Fig. 5, the carboxyl-terminal deletion mutant lacking the heptad repeat domain bound ␣-SNAP as well as the full-length protein. Furthermore, the heptad repeat domain alone was unable to bind ␣-SNAP. Finally, the deletion of the short amino-terminal stretch had no effect on ␣-SNAP binding.
The data show that while syntaxin 6 and ␣-SNAP bind specifically, the binding is not mediated by the predicted carboxyl-terminal coiled-coil, suggesting that the more aminoterminal heptad repeat is the site of interaction. While this domain is shorter than syntaxin 1a's H3 domain, it is sufficiently long to form coiled-coils and mediate protein-protein interactions (see "Discussion"). DISCUSSION The simplest conclusion that can be drawn from the data presented here is that syntaxin 6 is a member of the syntaxin family of vesicular transport receptors. Syntaxin 6 shows significant homology to syntaxin 3, Pep12, and SNAP-25. Syntaxin 6 has a carboxyl-terminal membrane anchor and behaves as an integral membrane protein. The predominant immunofluorescent pattern observed is juxtanuclear and colocalizes with Golgi markers. Syntaxin 6 binds to ␣-SNAP, which is believed to be a general mediator of vesicle fusion. Thus several lines of evidence implicate syntaxin 6 in vesicle trafficking.
It has been proposed that many of the protein-protein interactions involved in vesicle docking and fusion are mediated by coiled-coils. This idea has been motivated by studies coupling sequence analysis with deletion mapping experiments. In the case of syntaxin 1a, the carboxyl-terminal H3 helix, which is predicted to form coiled-coils, is necessary and sufficient for ␣-SNAP binding as well as 20 S complex formation. Similar studies have mapped binding sites on SNAP-25; syntaxin and ␣-SNAP bind to an amino-terminal heptad repeat region, while VAMP binds to a carboxyl-terminal one.
Syntaxin 6 has two regions with a high probability of forming coiled coils. The first one spans residues 47-71 (90% probabil-

FIG. 3. Syntaxin 6 behaves as an integral membrane protein.
Post-nuclear membrane pellet fractions were extracted with various disruptive agents, centrifuged at 100,000 ϫ g, and the resulting supernatants (S) and pellets (P) were analyzed by SDS-PAGE and antisyntaxin 6 (1/1500) Western blot analysis. All membranes were resuspended in homogenization buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1 mM dithiothreitol) and then an equal volume of the extraction solution was added to each sample to yield the following final concentrations of extracting agents: first and second lanes, only water added; third and fourth lanes, 1.5 M NaCl; fifth and sixth lanes, 0.2 M Na 2 CO 3 , pH 11.4, seventh and eighth lanes, 2 M urea; ninth and tenth lanes, 2% Triton X-100.  (25,26). The region from amino acids 47-71 has a 90% probability of forming coiled coils, and the region from amino acids 166 -225 has an 80% probability. The helical wheel representations above show the primarily hydrophobic residues at the a and d positions. B, binding of ␣-SNAP to each syntaxin 6 construct is shown. The syntaxin 6 fusion proteins were immobilized onto glutathione-agarose beads and incubated with increasing concentrations (1, 5, 10, 20, 40 M) of soluble His-tagged ␣-SNAP. The first row shows the binding curve to the full length syntaxin 6 without the hydrophobic tail, with an EC 50 between 1 and 5 M ␣-SNAP. Row 2 shows the binding to the carboxyl-terminal deletion; binding was unaffected. Row 3 shows no binding of ␣-SNAP to the predicted carboxyl-terminal coiled-coil forming region alone. Row 4 shows normal binding upon deleting the amino-terminal 25 amino acids. 40 M ␣-SNAP did not bind control glutathione S-transferase beads (data not shown). ity); the second one, comprising residues 166 -225 (80% probability), is reminiscent of syntaxin 1a's H3 helix in terms of its length and position before the transmembrane anchor. In this study we have shown that while ␣-SNAP specifically binds to syntaxin 6 with an EC 50 of 1-5 M, the carboxyl-terminal heptad repeat of syntaxin 6 is neither necessary nor sufficient for this binding. These data suggest that, unlike syntaxin 1a, the more amino-terminal heptad repeat region is involved in ␣-SNAP binding. While this domain is shorter than syntaxin 1a's H3 domain, this length is sufficient for forming coiled coils based on other studies (29). First, syntaxin 1a's H3 domain can be pared to 4.5 repeats and still bind SNAP-25 (9). Second, peptides of only three heptad repeats are able to participate in coiled coils (30,31). These lines of evidence argue that syntaxin 6's amino-terminal heptad repeat is capable of forming coiled coils and is responsible for binding ␣-SNAP. As noted above, SNAP-25 also utilizes an amino-terminal coiled-coil domain to interact with ␣-SNAP. The ability of a protein with syntaxinlike structural features to have SNAP-25-like binding properties supports the intriguing idea that the syntaxins and SNAP-25 are evolutionarily related (see below).
While syntaxin 6 is related to yeast Pep12, it is still unclear if it is the functional homologue or a related protein. There may be other uncharacterized yeast genes more similar to syntaxin 6. For example a 45,000-base pair yeast cosmid (a part of a yeast chromosome sequenced as part of the yeast genome mapping project, accession number gb U41528) contains a sequence which, when translated, is 23% identical to syntaxin 6 and has a carboxyl-terminal hydrophobic domain. Although Golgi region localization would not be predicted of a functional Pep12 homologue, it is not implausible; because subcellular localization within yeast is technically challenging, Pep12p has not been definitively localized. Syntaxin 6 is distinct from previously characterized proteins implicated in vesicle trafficking, due to its similarity to both the syntaxin and SNAP-25 families. It is striking that when the data base containing thousands of sequences is searched, the two most related protein sequences are proposed t-SNARE's involved in vesicle sorting. The existence of a single peptide sequence showing significant homology to two previously distinct families suggests that the syntaxin and SNAP-25 families are related, indicating they evolved from a common ancestor. It is possible that the transport step involving syntaxin 6 is not as tightly regulated as exocytosis. Thus syntaxin 6 may serve a function which requires both SNAP-25 and syntaxin 1a at the Golgi to plasma membrane step.
The genetic and biochemical studies of the presynaptic nerve terminal have revolutionized our understanding of synaptic vesicle docking and fusion and have allowed the creation of a testable hypothesis for the mechanism of all intracellular vesicle trafficking. The obligatory first step in testing this model is to identify proteins involved in other vesicle trafficking steps.
While other proteins, such as syntaxin 5, have been implicated in intracellular vesicle trafficking steps other than Golgi to plasma membrane, little is known about these transport steps in mammalian systems. Future studies will concentrate on defining the specific trafficking step syntaxin 6 participates in and determining the molecular mechanism underlying this step. Syntaxin 6 should be a valuable tool in answering fundamental questions regarding intracellular vesicle trafficking.