Syntaxin 12, a Member of the Syntaxin Family Localized to the Endosome*

We have cloned a new member of the syntaxin family of proteins. The open reading frame encodes a polypeptide of 272 amino acids with potential coiled-coil domains and a C-terminal hydrophobic tail. Northern blot analysis showed that the transcript is fairly ubiquitous. A soluble recombinant form of the polypeptide without the hydrophobic region binds to a -SNAP (soluble N -eth-ylmaleimide-sensitive factor attachment protein) and syndet/SNAP-23 in vitro . Polyclonal antibody raised against the recombinant protein recognized a 39-kDa protein in the membrane fraction of cell lysates. Indirect immunofluorescence studies using the polyclonal antibody showed that the protein is localized to intracellular membrane structures. Selective permeabilization studies with digitonin and saponin indicate that the epitope(s) recognized by the antibody is expose to the cytoplasm, consistent with the predicted orientation characteristic of SNAP receptor molecules. Morpholog-ical alterations of the staining pattern of the protein with brefeldin A and wortmannin treatment indicate that the protein is localize to the endosome. The cDNA we have cloned apparently corresponded to three previously described expressed sequence tags named as syntaxins 12, 13, and 14, respectively. We therefore pro-pose to retain the name syntaxin 12 for this protein.

We have cloned a new member of the syntaxin family of proteins. The open reading frame encodes a polypeptide of 272 amino acids with potential coiled-coil domains and a C-terminal hydrophobic tail. Northern blot analysis showed that the transcript is fairly ubiquitous. A soluble recombinant form of the polypeptide without the hydrophobic region binds to ␣-SNAP (soluble N-ethylmaleimide-sensitive factor attachment protein) and syndet/SNAP-23 in vitro. Polyclonal antibody raised against the recombinant protein recognized a 39-kDa protein in the membrane fraction of cell lysates. Indirect immunofluorescence studies using the polyclonal antibody showed that the protein is localized to intracellular membrane structures. Selective permeabilization studies with digitonin and saponin indicate that the epitope(s) recognized by the antibody is expose to the cytoplasm, consistent with the predicted orientation characteristic of SNAP receptor molecules. Morphological alterations of the staining pattern of the protein with brefeldin A and wortmannin treatment indicate that the protein is localize to the endosome. The cDNA we have cloned apparently corresponded to three previously described expressed sequence tags named as syntaxins 12, 13, and 14, respectively. We therefore propose to retain the name syntaxin 12 for this protein.
A biochemical and biophysical understanding of vesicular transport at the molecular level has been facilitated by in vitro assays, which reconstitute transport processes in cell-free systems (1). Thus, molecular components required for both vesicle budding (2) and vesicle docking/fusion processes (3)(4) have been isolated. The N-ethylmaleimide-sensitive factor (NSF), 1 an ATPase whose activity regulates the formation and dissociation of fusion complexes, is the first cytosolic factor characterized as such. NSF works in conjunction with another soluble factor, the soluble NSF attachment protein (SNAP) in mediating vesicle docking (1). Membrane components that are respon-sible for determining the specificity of the docking and fusion of the right vesicles to the right membranes are eventually identified based on the ability to interact with SNAP (4). These are known as SNAP receptors (SNARE). SNAREs can be broadly divided into two classes. Those present on transport vesicles are the v-SNAREs, and those present on the target membranes are the t-SNAREs. Thus, the SNARE hypothesis, a working hypothesis proposed by Rothman and co-workers (3,4), holds that a transport vesicle chooses its target for fusion when v-SNAREs pair with the cognate t-SNAREs at the target membrane. Genetic dissections of the yeast secretory pathway and the biochemical characterization of molecules involved in synaptic vesicle docking and fusion have resulted in the isolation and/or molecular cloning of putative SNARE molecules that are structurally related (5)(6). However, isolation of SNAREs in the constitutive secretory and endocytotic pathway in mammalian cells remains difficult. This difficulty has been overcome in part by the availability of an expanding data base of expressed sequence tags (EST) and efficient data base search and sequence alignment programs (7).
The first member of the syntaxin family of proteins, syntaxin 1A, was first characterized as a neuronal-specific protein involved in the regulation of neurotransmitter release (8). Its localization to the plasma membrane and its interaction with the synaptic vesicle v-SNARE synaptobrevin point to its function as a t-SNARE. Subsequently, a family of syntaxin-related molecules that shares 23-84% amino acid identity has been identified (6,9). These syntaxins are more ubiquitous in their expressions in various tissues, which is indicative of their possible functions in other vesicular transport steps in the cell. These syntaxins also display a variety of cellular localizations within the secretory pathway. Whereas syntaxins 2, 3, and 4 are apparently cell surface proteins (6, 10 -11), syntaxin 5 and syntaxin 6 are localized to the Golgi region (6,9).
The transport from the early endosome to the late endosome and lysosome is a major route of intracellular membrane trafficking. However, little is known at the molecular level about the mechanisms regulating membrane interactions in the endocytic pathway beyond early endosomes. Recently, it has been shown that COP1 coat components participate in endosomal transport (12). Using an in vitro transport assay to study the biochemical properties of endosome docking and fusion events, Robinson et al. (13) have shown that NSF and SNAPs are required for several steps of endosomal membrane transport. Of the known SNAREs, only the v-SNARE cellubrevin has been shown to have an endosomal localization (14). Should the mechanism of vesicular transport in the endosomes not differ too drastically from its exocytic counterpart, one might expect to find members of the syntaxin family functioning as t-SNAREs in the endosomal membranes. In this report, we present such a molecule. * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF035632.
‡ The first two authors contributed equally to this work.

EXPERIMENTAL PROCEDURES
Materials-Cell lines were primarily from the American Type Culture Collection. Monoclonal antibody against trans-Golgi network 38 (TGN38) was kindly provided by Dr George Banting (University of Bristol, United Kingdom). Expressed sequence tag clones were generated by the Washington University MERCK EST project and were obtained from the IMAGE consortium. Syndet cDNA (15) was kindly provided by Dr. G. Baldini (Columbia University, New York). Syntaxin 1A cDNA (6) was kindly provided by Dr. R. Scheller (Stanford University, CA). mSEC13 cDNA (16) was kindly provided by Dr. Anand Swaroop (University of Michigan).
Methods-Data base searches were performed with the various BLAST algorithms available at the National Center for Biotechnology (NCBI) World Wide Web server. Library screening, cloning, and DNA sequencing were performed using standard methods as described (17). Northern blot analysis was performed using a rat multiple tissue Northern blot from CLONTECH.
In vitro translation of various constructs was performed using in vitro translation kits from Promega according to the manufacturer's protocol. For in vitro binding assays (18), the cytoplasmic domain of syntaxin 12 (amino acids 1-248) was generated by the polymerase chain reaction and cloned into the the plasmid pBSK (Stratagene).
[ 35 S]Methionine-labeled translation product was incubated with glutathione-Sepharose beads coated with either glutathione S-transferase (GST), GST-␣SNAP, or GST-syndet in binding buffer (20 mM Hepes, pH 7.5, 25 mM NaCl, 3% glycerol, 7 mM MgCl 2 , 1 mM CaCl 2 , and 1 mM EDTA) with 0.1% bovine serum albumin and 0.5% Nonidet P40 at 4°C for 3 h. The beads were washed twice with the complete incubation buffer, twice in buffer without bovine serum albumin, and twice in buffer without bovine serum albumin and Nonidet P-40. SDS sample buffer was then added, and the SDS eluates were analyzed by SDSpolyacrylamide gel electrophoresis.
The cytoplasmic domain of the protein is expressed either as hexahistidine-tagged or GST fusion proteins in bacteria. The fusion proteins were also used to immunize rabbits. Polyclonal antibodies were affinitypurified from serum harvested after several booster injections by the fusion proteins immobilized on nitrocellulose strips.
Cells were maintained in RPMI medium supplemented with 10% fetal bovine serum. Immunofluorescence microscopy was performed as described previously (19 -20). Cells plated on coverslips and subjected to various treatments were fixed with 3% paraformaldehyde followed by sequential incubation with the primary antibodies and fluorescein isothiocyanate or rhodamine-conjugated secondary antibodies. Fluorescence labeling was visualized using an Axiophot microscope (Carl Zeiss,

FIG. 2.
A, the alignment of the coiled-coil region of rat syntaxin 12 with rat syntaxins 3 and 4 (MegAlign program of DNASTAR). Identical residues are shaded. B, Northern blot analysis of syntaxin 12. H, heart; B, brain; S, spleen; L, lung; Li, liver; SM, skeletal muscle; K, kidney; T, testis; kb, kilobase pairs. C, schematic diagram showing how the translated sequence of the human ESTs AA167677, R29508, and T08774 match with the coding region of rat syntaxin 12 (syn12). The translated sequence of the human ESTs were based on BLAST search results returned by the server of the National Center for Biotechnology Information, using the TBLASTN program to search the dbest data base.

Molecular Cloning and Sequencing of a Novel Member of the
Syntaxin Family-Data base searches have allowed us to identified human ESTs (accession numbers R21569 and N99549) potentially coding for a syntaxin-like molecule. A complete cDNA was isolated from a rat brain cDNA library, and sequencing revealed a 272-amino acid open reading frame as shown in Fig. 1A. The predicted amino acid sequence has a stretch of 22 hydrophobic residues at the C terminus, as illustrated by a Kyte-Doolittle hydrophobicity plot (Fig. 1B). This primary structure is characteristic of a hydrophobic tail anchor. The polypeptide has several potential regions that may form coiledcoil structures, as revealed by the Coils version 2.1 program (Fig. 1C).
A data base search using the NCBI BLAST program revealed that the coding sequence has the highest homology with members of the syntaxin family, particularly in the coiled-coil region preceding the C-terminal transmembrane domain ( Fig. 2A). A multiple tissue Northern blot with the full-length cDNA showed that the transcript has a fairly ubiquitous expression, being more abundant in brain, lung, and kidney (Fig. 2B).
Based on ESTs identified by data base searches, a series of 10 novel syntaxins has been previously described (21). A search of the GenBank TM dbest data base revealed several human ESTs that, in view of their sequence homology, represent the human homologs of our rat cDNA. Among these, three human ESTs (R29508, AA167677, and T08774) have been listed as syntaxin 12, syntaxin 13, and syntaxin 14, respectively (21). Fig. 2C is a schematic diagram of how the translated sequence of these ESTs match with the coding region of the rat cDNA. As shown, the limited and inaccurate sequence information of these ESTs only allowed each of them to be matched with a portion of the coding region. There is a short overlap between AA167677 and R29508. T08774 matches to the C-terminal portion of the rat protein. Apparently, these assumed syntaxins 12, 13, and 14 represent the human homolog of the rat protein.
Based on the limited sequence data alone, these ESTs were assumed to represent individual syntaxin-like molecules. Our results therefore caution against the assignment of individual identity to an EST before full-length sequence information of each is available. To avoid any further confusion in the nomenclature of new members of the syntaxin family, we have retained the name syntaxin 12 for this protein.

Syntaxin 12 Is a SNAP Receptor Molecule
Localized to the Endosome-The predicted primary structure of syntaxin 12 and its homology to other syntaxins suggest that it is a SNARE molecule. We sought to confirm this by investigating if syntaxin 12 binds to ␣-SNAP in vitro. [ 35 S]Methionine-labeled translation product of the soluble cytoplasmic domain of both syntaxin 1A and syntaxin 12 was incubated with glutathione-Sepharose beads coated with either GST or GST-␣SNAP. As shown in Fig. 3, the binding of the syntaxin 1A cytodomain to GST-␣SNAP is significantly higher than to GST itself. Such is also the case for syntaxin 12. On the other hand, mSEC13 (20), a protein with multiple ␤-transducin or WD-40 repeats known to participate in protein-protein interactions, did not exhibit significant binding to GST-␣SNAP. The ability of syntaxin 12 to bind ␣SNAP is in good agreement with its putative function as a SNARE.
Syntaxin 1 is known to exist in complex with another neuronal-specific SNARE molecule, SNAP-25 (22)(23). A novel nonneuronal molecule that binds syntaxin and synaptobrevin and is homologous to SNAP-25 has since been cloned from human B lymphocyte (18), and based on its molecular size, is called SNAP-23. Recently, syndet, a ubiquitous mouse protein homologous to SNAP-25 has also been cloned (15). Based on the sequence similarity, syndet appear to be the mouse homolog of human SNAP-23. We sought to determine if syntaxin 12 could also interact in vitro with syndet/SNAP-23. As shown in Fig. 3, both syntaxin 1A and syntaxin 12 binds to syndet/SNAP-23 with high specificity, whereas mSEC13 again exhibits no binding. The ability of syntaxin 12 to bind SNAP-23 is highly suggestive of its functional similarity with other members of the syntaxin family.
That syntaxin 12 is indeed a membrane-anchored protein was confirmed by the fact that its full-length translated product could not be stripped off membranes by high salt (1 M KCl) or high pH (sodium bicarbonate, pH 11) treatment but could be solubilized with a detergent such as Triton X-100 (Fig. 4A).To further characterize the molecule, rabbit polyclonal antibodies were raised using bacterially expressed fusion protein. The affinity-purified antibody detected a ϳ39-kDa band by immunoblot analysis of the in vitro translated product of the fulllength cDNA and NRK cell lysates (Fig. 4B). Detection can be abolished by co-or preincubation of the antibody with excess amount of the fusion protein (not shown).
As a first step toward functional characterization of syntaxin 12, we performed indirect immunofluorescence microscopy to localized the protein in several cell lines. As shown in Fig. 5, the antibody-stained intracellular structures clustered at the perinuclear region of mouse (MEF), human (HeLa), and rat (NRK) cells. To confirm the membrane topology of syntaxin 12, cells were permeabilized with 20 g/ml digitonin, which would selectively permeabilized the plasma membrane while leaving the internal membranes intact, or 1 mg/ml saponin, which would permeabilize all membranes. Cells were then doublelabeled with syntaxin 12 polyclonal antibody and a monoclonal antibody against the lumenal domain of the Golgi protein mannosidase II. As shown in Fig. 5, digitonin-permeabilized cells were labeled for syntaxin 12 only and not mannosidase II. Permeabilization of the plasma membrane was indeed achieved because under the same conditions, cytoplasmic tubulin was also labeled (not shown). As expected, both syntaxin 12 and mannosidase II were labeled in saponin-permeabilized cells. Syntaxin 12 is thus membrane-anchored, with a large portion of its N terminus exposed in the cytoplasm, a topology characteristic of all syntaxins but not all SNAREs identified to date.
All syntaxins discovered to date were localized to either the Golgi region or the cell surface, indicating that they function along the exocytic pathway. As shown in Fig. 6, the antibody labeled perinuclear structures in NRK cells. The labeling pattern in NRK cells resembles that of the the TGN marker, TGN38 (24), double-labeled in the same cell although not completely colocalized. The fungal metabolite brefeldin A (BFA) has varying effects on the morphology of subcellular organelles and on the distribution of various markers on these organelles (25). The distribution of Golgi markers to the endoplasmic reticulum (25) and the collapse of TGN markers TGN38 (26) and furin proprotein convertase (27) and endosomal markers (28 -30) into the microtubule-organizing center upon BFA treatment had been extensively documented. The effect of BFA on the morphology of a particular protein is therefore often useful in determining its subcellular localization. Treatment of cells with 10 g/ml BFA resulted in the collapse of the structure into a compact structure characteristic of the microtubuleorganizing center, colocalizing well with that of TGN38. This result suggests that the perinuclear staining of syntaxin 12 is not that of the Golgi apparatus (which under this condition would have redistributed to the endoplasmic reticulum) but may be that of the TGN or the endosomes. BFA, however, also causes a fusion of the endosomes with the trans-Golgi network (28), and endosomal markers behave quite like TGN markers at the end point of the BFA effect.
To further determine the exact localization of the rat syntaxin and differentiate between the two possibilities, we treat cells with wortmannin, the phosphatidylinositol 3-kinase inhibitor (31). This drug has been shown to alter the morphology of endosomes but not the Golgi apparatus or the trans-Golgi network (32)(33). Although wortmannin treatment did not alter the perinuclear Golgi staining marked by the TGN38 monoclonal antibody, the perinuclear structure marked by the rat syntaxin antibody in the same cells were converted into swollen vacuoles, characteristic of wortmannin-induced changes to the stainings of endosomal markers such as the mannose 6-phosphate receptor (32). We observed this as well in another cell line, L2 (not shown). The above results strongly suggest that the rat syntaxin-like molecule is localized to a BFA and wortmannin-sensitive endosomal compartment. DISCUSSION We have therefore cloned a novel member of the syntaxin family with a unique subcellular localization. None of the syntaxins 1-6 published to date are associated with the endosome. Exogenous expression of syntaxin 1A does result in intracellular localization in Madin-Darby canine kidney cells, and exogenous expression of syntaxin 3 does have an intracellular component (10). These intracellular stainings were, however, shown not to be endosomal in nature but rather colocalized with a lysosomal marker (10). Another novel syntaxin, known as syntaxin 7, has recently been cloned in our laboratory (34) as well as by Wang et al. (35). Based on its homology to yeast and plant vacuolar syntaxins, Wang et al. proposed that syntaxin 7 may have a role in trafficking between the Golgi apparatus and the lysosomes. Our immunolocalization data, however, suggest that syntaxin 7 is localized to the endosomes (34). In view of the localizations of syntaxin 7 and syntaxin 12, the transport machinery of the endocytic pathway, like its exocytic counterpart, also utilizes members of the syntaxin family.
What may the function of syntaxin 12 be? There are several possibilities. Judging by its compact, perinuclear staining, syntaxin 12 may well reside in a late endosomal compartment. Indeed it does not colocalize with transiently internalized transferin (not shown). However, we could not rule out that small amounts of syntaxin 12 may reside in the early endosomes. If solely localize to a late endosomal compartment, syntaxin 12 may function to receive vesicles either from the TGN or the early endosome or participate in the recycling of surface receptors. Elucidation of its exact role in transport awaits experiments involving effective functional disruption either by the introduction of negative dominant mutants, inhibitory antibodies, or targeted disruption of the gene.
FIG. 6. Subcellular localization of rat syntaxin 12. Cells, either untreated (C) or treated with 10 g/ml BFA or 1 M wortmannin (wort) for 1 h were fixed with 4% paraformaldehyde and incubated with affinity-purified rabbit polyclonal antibody against rat syntaxin 12 (syn 12) and a monoclonal antibody against rat TGN38. This is followed by incubation with fluorescein isothiocyanate-labeled anti-rabbit IgG and TRITC-labeled antimouse IgG before being processed for indirect immunofluorescence and confocal microscopy. Bar, 10 m.