A 29-Kilodalton Golgi SolubleN-Ethylmaleimide-sensitive Factor Attachment Protein Receptor (Vti1-rp2) Implicated in Protein Trafficking in the Secretory Pathway*

Expressed sequence tags coding for a potential SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptor) were revealed during data base searches. The deduced amino acid sequence of the complete coding region predicts a 217-residue protein with a COOH-terminal hydrophobic membrane anchor. Affinity-purified antibodies raised against the cytoplasmic region of this protein specifically detect a 29-kilodalton integral membrane protein enriched in the Golgi membrane. Indirect immunofluorescence microscopy reveals that this protein is mainly associated with the Golgi apparatus. When detergent extracts of the Golgi membrane are incubated with immobilized glutathione S-transferase α soluble N-ethylmaleimide-sensitive factor attachment protein (GST-α-SNAP), this protein was specifically retained. This protein has been independently identified and termed Vti1-rp2, and it is homologous to Vti1p, a yeast Golgi SNARE. We further show that Vti1-rp2 can be qualitatively coimmunoprecipitated with Golgi syntaxin 5 and syntaxin 6, suggesting that Vti1-rp2 exists in at least two distinct Golgi SNARE complexes. In cells microinjected with antibodies against Vti1-rp2, transport of the envelope protein (G-protein) of vesicular stomatitis virus from the endoplasmic reticulum to the plasma membrane was specifically arrested at the Golgi apparatus, providing further evidence for functional importance of Vti1-rp2 in protein trafficking in the secretory pathway.

Participation of NSF 1 and soluble NSF attachment proteins (SNAP) in diverse transport events in the secretory and endocytotic pathways is in conjunction with a superfamily of membrane proteins termed SNAP receptors (SNAREs) (1)(2)(3)(4)(5). The SNARE hypothesis suggests that vesicles derived from a donor compartment harbor a set of vesicle-associated SNAREs (v-SNAREs) that will interact specifically with those associated with those on the target acceptor membrane (t-SNAREs) (6 -11). This v-/t-SNARE pairing is a key event in the docking and fusion of the vesicle with its specific target membrane (6 -11). Vesicle-associated membrane proteins (VAMPs) or synaptobrevins are v-SNAREs associated with the synaptic vesicles, whereas syntaxin 1 and SNAP-25 (synaptosome-associated protein of 25 kDa) are t-SNAREs associated with the presynaptic membrane. The specific interaction of VAMPs/synaptobrevins with the syntaxin 1-SNAP-25 complex plays a fundamental role in the docking/fusion of synaptic vesicles with the presynaptic membrane (9 -11).
Because of the central role of SNAREs in diverse vesicular transport steps, molecular identification, biochemical characterization, and subcellular localization of novel SNAREs constitute fundamentally important aspects of study in the field of vesicular transport. The Golgi apparatus plays a major role in the secretory pathway (1)(2)(3)(4)12). Currently, five distinct SNAREs have been shown to be associated with the Golgi apparatus in mammalian cells. These include syntaxin 5 (13)(14)(15), GS15 (16), GS27 (also termed membrin) (17)(18), GS28 (also named GOS-28) (19 -20), and syntaxin 6 (21)(22). Syntaxin 5 and GS28 have both been shown to be involved in the endoplasmic reticulum (ER) to Golgi transport. GS28 has also been implicated in transport from the cis-to the medial-Golgi (15, 19 -20). GS27 was shown to be involved in transport from the cis/medial-to trans-Golgi/trans Golgi network (18). The functional aspects of GS15 and syntaxin 6 remain to be established (16,(21)(22). In this report, we describe the molecular, biochemical, and cell biological characterizations of Vti1-rp2, a novel 29-kDa SNARE associated with the Golgi apparatus. Vti1-rp2 is structurally homologous to Vti1p, a recently described yeast Golgi SNARE (23). We further show that Vti1-rp2 exists in distinct syntaxin 5-and syntaxin 6-containing SNARE complexes and is functionally important for protein trafficking in the secretory pathway.

EXPERIMENTAL PROCEDURES
Materials-Mouse EST clones (accession numbers AA016379 and W13616) were generated by the Washington University-Merck expressed sequence tag (EST) project and made available by IMAGE consortium via Research Genetics Inc. (Huntsville, Alabama). The mouse mRNA multiple tissues Northern blot was obtained from CLON-TECH (Palo Alto, CA. Mouse monoclonal antibody against Golgi mannosidase II was from Babco (Berkeley, CA). Fluorescein isothiocyanateconjugated goat anti-mouse IgG and rhodamine-conjugated goat antirabbit IgG were purchased from Boehringer Mannheim. Brefeldin A was from Epicentre Technologies.
cDNA Cloning and Sequencing-Mouse EST clones were fully sequenced by the dideoxy chain termination method using a kit from U. S. Biochemical Corp. The complete coding region was assembled using the DNA Strider 1 program.
Northern Blot Analysis-A mouse multiple tissue blot of poly(A) ϩ mRNA was probed with the insert of the EST clone AA016379 followed by actin probe as described previously (24).
Expression of Recombinant Proteins in Bacteria-GST fusion pro-* This work was funded by the Institute of Molecular and Cell Biology, National University of Singapore (to W. H.). 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) AF035823.
Differential Extraction of Golgi Membranes and Immunoblot Analysis-These were performed as described previously (16,30).
In Vitro GST-␣-SNAP Binding Assay and Immunoprecipitation-This was performed as described previously (16,26).
Micoinjection and in Vivo Transport of Vesicular Stomatitis Virus G-protein-Vero cells grown on coverslips were infected with the vesicular stomatitis virus ts045 strain at 31°C for 45 min and then shifted to 40°C for 1 h. Cells were then transferred to 4°C in Dulbecco's modified Eagle's medium without fetal bovine serum, and microinjection was performed under a Zeiss Axiophot microscopy using an Eppendorf micromanipulation system. Cells were transferred back to 40°C immediately after injection and incubated at 40°C for 2 h to accumulate the G-protein in the ER. Transport of G-protein was performed by incubating cells at 31°C for 45 min in the presence of cycloheximide (to prevent new synthesis of G-protein). Cells were then fixed and processed for indirect immunofluorescence double-labeling to detect microinjected antibodies and the G-protein.

Vti1-rp2, a Mammalian Protein Homologous to Yeast Vti1p-
Searching the EST data bases using the amino acid sequence of a novel Golgi SNARE characterized in the lab 2 led to the identification of mouse EST clones (accession numbers AA016379 and W13616) that encode a putative SNARE. The EST clone W13616 was fully sequenced, and the nucleotide and the deduced amino acid sequences are shown in Fig. 1A. This protein was independently identified in three other laboratories and has been referred to as Vti1-rp2 (31), Vti1b (32), and Vti1a (33), respectively. To avoid further confusion in nomenclature, we have adopted the name Vti1-rp2 for this protein. Vti1-rp2 is a protein of 217 residues. Although the predicted molecular weight of Vti1-rp2 is 24,971 daltons, its apparent size as revealed by SDS-polyacrylamide gel electrophoresis is 29 kDa (see below). There exists a 22-residue carboxyl-terminal hydrophobic region that may function as a membrane anchor, a characteristic of the majority of known SNAREs (6 -11, 16). Preceding the carboxyl-terminal hydrophobic tail are four regions (residues 32-62, 69 -93, 112-134, 146 -179) that have the potential to form coiled-coil structures as predicted by the COILS 2.1 program (Fig. 1B). A search of motifs with the ScanProsite program revealed the existence of an ATP/GTP binding motif A (P-loop) in the sequence (residues 170 -177, ADANLGKS). Vti1-rp2 is homologous to Vti1p, a yeast Golgi SNARE implicated in at least two trafficking events (23). The homology between Vti1-rp2 and Vti1p occurs throughout the entire polypeptide with an overall amino acid identity of 28% and similarity of 45% (Fig. 1C). Since Vti1p does not contain the consensus ATP/GTP binding site motif A, it is not clear if the ATP/GTP binding site motif A observed in Vti1-rp2 is functionally important.
Vti1-rp2 mRNA Is Widely Expressed-To examine whether Vti1-rp2 is involved in a general cellular process or its function is restricted to certain tissues, Northern blot analysis was performed to examine the levels of Vti1-rp2 mRNA in various mouse tissues ( Fig. 2A). A major mRNA species of about 2.6 kilobases was detected at varying levels in all the tissues examined, consistent with the notion that Vti1-rp2 may participate in an event common to all the cell types. Interestingly, the Vti1-rp2 mRNA detected in the testis has a smaller size (about 1.5 kilobases). The basis and significance for this different mRNA size of Vti1-rp2 in the testis is currently unknown. Vti1-rp2 Is a 29-kDa Integral Membrane Protein Associated with the Golgi Apparatus-The cytoplasmic domain (residues 1-185) of Vti1-rp2 was expressed as a recombinant fusion protein with GST (GST-Vti1-rp2). The purified GST-Vti1-rp2 was used to raise polyclonal antibodies against Vti1-rp2 in rabbits. Although the predicted size for Vti1-rp2 is 249710, its apparent size revealed by SDS-polyacrylamide gel electrophoresis is about 29 kDa (Fig. 3A), because it migrates (lane 1) in between the 30-kDa marker (the marker lane) and GS28 (lane 2) (a Golgi SNARE with an apparent size of 28 kDa) (20,30). Another mammalian protein homologous to Vti1p has also been identified and has been referred to as Vtil-rp1 (31), Vti1 (32), and Vti1b (33), respectively. To avoid further confusion, we have adopted the name Vti1-rp1 for the other mammalian homolog of yeast Vti1p. Since Vti1-rp2 displays significant amino acid sequence identity (about 30%) with Vti1-rp1, it is essential to establish that our affinity-purified antibodies do not cross-react with Vti1-rp1. As shown (Fig. 3A), the detection of the 29-kDa protein in immunoblot was selectively abolished by preincubation of antibodies with recombinant cytoplasmic domain of Vti1-rp2 (lane 4) but not with the cytoplasmic domain of Vti1-rp1 (lane 3), establishing that our antibodies are specific for Vti1-rp2. Immunoblot analysis revealed that Vti1-rp2 is enriched in Golgi membranes (Fig. 3B, lanes 4 -6) as compared with total membranes, and microsomal membranes. The enrichment of Vti1-rp2 in Golgi membranes is comparable with that of Golgi ␣2,6-sialyltransferase (lanes 1-3). When Golgi-enriched membranes were subjected to different extraction conditions (Fig. 3C), Vti1-rp2 was effectively extracted by detergents but not by phosphate-buffered saline, 2.5 M urea, 0.1 M sodium bicarbonate (pH 12), or 2 M KCl. Vti1-rp2 is thus an integral membrane protein enriched in the Golgi fractions.
Indirect immunofluorescence microscopy was used to examine the exact subcellular localization of Vti1-rp2 (Fig. 4). Affinity-purified antibodies against Vti1-rp2 specifically labeled perinuclear structures (Fig. 4A, panel a) characteristic of the Golgi apparatus (34), and this labeling colocalized well with that of Golgi mannosidase II (panel b) (35). When the Golgi apparatus was fragmented by nocodazole treatment (panels c-d), Vti1-rp2 and mannosidase II were colocalized well in the fragmented Golgi apparatus. Similar to mannosidase II and other Golgi proteins (36), Vti1-rp2 was redistributed to the ER when cells were treated with brefeldin A (e-f). These results firmly establish that Vti1-rp2 is an integral membrane protein associated preferentially with the Golgi apparatus.
Vti1-rp2 Is a Novel Golgi SNARE-To investigate whether Vti1-rp2 indeed functions as a novel SNARE of the Golgi apparatus, we examined the potential interaction of Vti1-rp2 with ␣-SNAP. As shown in Fig. 5A (upper panel), Vti1-rp2 in the Golgi detergent extract was specifically retained by immobilized GST-␣-SNAP in a dose-dependent manner. Under identical conditions, Vti1-rp2 was not retained by immobilized GST or several other control GST fusion proteins (data not shown). Furthermore, other Golgi proteins, including ␣2,6-sialyltransferase, were not retained by immobilized GST-␣-SNAP (Fig.  5A, lower panel). The interaction of Vti1-rp2 with ␣-SNAP was further investigated (Fig. 5B). Proteins in the Golgi extract were incubated with GST (lanes 1 and 4), GST-␣-SNAP ( lanes  2 and 5), and GST-␥-SNAP (lanes 3 and 6); after extensive washing, the beads (lanes 1-3) and 1/10 of the supernatants (lane 4 -6) were analyzed by immunoblot to detect Vti1-rp2 (upper row) as well as GS28 (lower row), which serves as a positive control. Vti1-rp2 was retained by GST-␣-SNAP as efficiently as GS28 (lanes 2 and 5). Neither Vti1-rp2 nor GS28 was retained by GST (lanes 1 and 4). To lesser extents, Vti1-rp2 and GS28 was significantly retained by GST-␥-SNAP. These results establish that interaction of Vti1-rp2 with GST-␣-SNAP is specific and occurs with efficiencies comparable with that of known Golgi SNAREs such as GS28. Furthermore, interaction of Vti1-rp2 with immobilized GST-␣-SNAP could be abolished by NSF in conditions that promote dissociation of SNARE complexes (Fig. 5C). When Golgi extract was incubated with immobilized GST-␣-SNAP in the presence of increasing amounts of NSF in conditions (assembly buffer) that promote formation of SNARE complexes (lanes 1-6), comparable amounts of Vti1-rp2 were retained. However, retention of Vti1-rp2 by immobilized GST-␣-SNAP was readily abolished by NSF in conditions (lane 7-12) that promote ATP hydrolysis by NSF and disassembly of SNARE complexes. These results not only further confirmed that the interaction of Vti1-rp2 with ␣-SNAP is specific but also revealed that the interaction of Vti1-rp2 with ␣-SNAP is in the context of Vti1-rp2-containing SNARE complexes.

FIG. 4. Vti1-rp2 is associated preferentially with the Golgi apparatus.
Control normal rat kidney cells (a and b), normal rat kidney cells treated with 10 g/ml nocodazole for 1 h at 37°C (c and d), and normal rat kidney cells treated with 10 g/ml brefeldin A for 1 h at 37°C (e and f) were double-labeled with rabbit polyclonal antibodies against Vti1-rp2 (a, c, and e) and a monoclonal antibody against Golgi mannosidase II ( Man II) (b, d, and f). Bar, 10 M. by antibodies against syntaxin 6. In contrast, syntaxin 5 was not coimmunoprecipitated by anti-syntaxin 6 antibodies ( lanes  5 and 7). These results suggest that significant amounts of Vti1-rp2 exist in at least two distinct SNARE complexes, one containing syntaxin 5 and the other containing syntaxin 6. This conclusion was further substantiated by our observation that significant amounts of syntaxin 5 and syntaxin 6 were coimmunoprecipitated by antibodies against Vti1-rp2 (data not shown).
A Role of Vti1-rp2 in Protein Transport in the Secretory Pathway-The association of Vti1-rp2 with the Golgi apparatus and its establishment as a SNARE suggest that it may participate in protein trafficking in the secretory pathway. To investigate this, Vero cells grown on coverslips were first infected with vesicular stomatitis virus ts045 and then microinjected with affinity-purified antibodies against Vti1-rp2. Transport of G-protein along the secretory pathway was monitored by indirect immunofluorescence microscopy. Since microinjection of antibodies against EAGE epitope of ␤-COP was shown previously to inhibit G-protein transport (29), cells microinjected with ␤-COP antibodies serve as the positive control. We have shown recently that syntaxin 7 is in the endosomal compartment (28), and syntaxin 7 is thus not expected to function in the secretory pathway. Cells microinjected with syntaxin 7 antibodies thus serve as a negative control. As shown in Fig. 7, in cells microinjected with antibodies against Vti1-rp2 (C, arrows), surface labeling of G-protein was dramatically reduced, resulting in accumulation of G-protein in perinuclear structures characteristic of the Golgi apparatus (D, arrows). This inhibitory effect is comparable with that seen in cells microinjected with antibodies against ␤-COP (A-B, arrows). In marked contrast, transport of G-protein to the cell surface was unaffected (E and F, arrows) in cells microinjected with syntaxin 7 antibodies. These results suggest that transport of G-protein from the ER to the plasma membrane is specifically inhibited in cells microinjected with antibodies against Vti1-rp2, and the site of inhibition seems to be at the level of the Golgi apparatus because G-protein was seen to accumulate in structures characteristic of the Golgi apparatus and the arrested G-protein colocalized well with markers of the Golgi apparatus such as 12-(N-methyl-N-(7-nitrobenz-2-oxa-1,3-diazol-4-yl))-ceramide and binding sites for lectin lens culinaris agglutinin (data not shown). Although more detailed future experiments are needed to address the mechanistic aspects of Vti1-rp2 involvement in protein transport, these results clearly revealed a role of Vti1-rp2 in protein transport in the secretory pathway. DISCUSSION We have identified a novel 29-kDa mammalian protein (Vti1-rp2) that has characteristics of a SNARE based on the presence of a COOH-terminal hydrophobic membrane anchor and several regions that can potentially form coiled-coil structures (6 -11, 16). Three observations establish that Vti1-rp2 is indeed a SNARE. First, Vti1-rp2 in Golgi detergent extract can interact with immobilized GST-␣-SNAP in a specific and dose-dependent manner. Interaction of Vti1-rp2 with immobilized GST-␣-SNAP occurs with efficiencies comparable with that of known Golgi SNAREs such as GS28 (20). The second line of evidence is that association of Vti1-rp2 with GST-␣-SNAP could be abolished by NSF, specifically under conditions that promote NSF ATPase activity and dissociation of SNARE complexes. This suggests that interaction of Vti1-rp2 with immobilized ␣-SNAP occurs through Vti1-rp2-containing SNARE complex(es) in the Golgi extract. The demonstration of existence of Vti1-rp2 in at least two distinct SNARE complexes (one containing syntaxin 5 and the other containing syntaxin 6)  1 and 4), GST-␣-SNAP (lanes 2 and 5), or GST-␥-SNAP (lanes 3 and 6) immobilized onto the glutathione beads. After extensive washing, the beads (lanes 1-3) and 1/10 of the supernatants (lanes 4 -6) were analyzed by immunoblot to detect Vti1-rp2 (upper row) or GS28 (lower row). C, interaction of Vti1-rp2 with immobilized GST-␣-SNAP could be abolished by NSF in conditions that promote disassembly of SNARE complexes. 200 g of Golgi extract was incubated with 2 g of immobilized GST-␣-SNAP in the presence of indicated amounts of NSF in either assembly (lanes 1-6) or disassembly buffer (lanes 7-12). The amounts of Vti1-rp2 bound onto the beads were then determined by immunoblot. FIG. 6. Vti1-rp2 exists in distinct syntaxin 5-and syntaxin 6-containing Golgi SNARE complexes. The Golgi detergent extract was immunoprecipitated (IP) with antibodies against syntaxin 5 ( lanes  1 and 3), antibodies against syntaxin 6 ( lanes 5 and 7), or control rabbit IgG (lanes 2, 4, 6, and 8). The immunoprecipitates (lanes 1, 2, 5, and 6) and 1/10 of supernatants (lanes 3, 4, 7, and 8) were analyzed by immunoblot to detect the indicated proteins. Please note that antibodies against syntaxin 5 detect two bands as has been described previously (16 -17). P, beads; S, unbound fraction.
provides the third line of evidence to support that Vti1-rp2 is a novel SNARE. The subcellular localization of Vti1-rp2 was established by two independent results. First, Vti1-rp2 is highly enriched in a membrane fraction that is also enriched for the Golgi apparatus. Furthermore, Vti1-rp2 colocalized well with the Golgi marker mannosidase II in both control as well as nocodazole-fragmented Golgi apparatus. Like mannosidase II, Vti1-rp2 could be redistributed into ER-like structures by brefeldin A. It is thus firmly established that Vti1-rp2 is a novel SNARE of the Golgi apparatus.
Data base searches with Vti1-rp2 sequence revealed that Vti1-rp2 is most homologous to Vti1p. Vti1p is a recently identified v-SNARE of the yeast Golgi and has been implicated in two independent vesicular transport events (23). By interacting with the early Golgi t-SNARE Sed5p (the yeast counterpart of syntaxin 5) (37-38), Vti1p has been suggested to function as a v-SNARE for vesicles involved in retrograde intra-Golgi transport. Furthermore, Vti1p has also been shown to be involved in transport from the late Golgi to the vacuole (equivalent to the mammalian lysosome) by interacting with Pep12p (39), a syntaxin-like t-SNARE of the pre-vacuolar compartment (equiva-lent to the mammalian late endosome). Whether Vti1-rp2 represents the mammalian counterpart of Vti1p remains to be further investigated, although another mammalian protein (Vti1-rp1) homologous to Vti1p could functionally substitute for the yeast Vti1p (32). Besides its sequence homology with Vti1p, another property of Vti1-rp2 that is similar to Vti1p is that a significant amount of Vti1-rp2 exists in a syntaxin 5-containing Golgi SNARE complex. In addition, a significant amount of Vti1-rp2 was also shown to be present in a syntaxin 6-containing SNARE complex. Since coimmunoprecipitation of syntaxin 5 and syntaxin 6 was not observed, these results suggest that Vti1-rp2 exists in distinct syntaxin 5-and syntaxin 6-containing SNARE complexes. Although the functional aspects remain to be established, syntaxin 6 has been shown recently to be enriched in the trans-Golgi network (22).
The presence of Vti1-rp2 in at least two distinct Golgi SNARE complexes indicates that it may function as a SNARE for at least two types of vesicle-mediated transport events. One will dock and fuse with cis-Golgi by interaction with syntaxin 5, whereas the other will dock and fuse with trans-Golgi network via interaction with syntaxin 6. The interaction of Vti1-rp2 with at least two syntaxin-like t-SNAREs is consistent with a recent study showing that yeast Vti1p could interact with at least five distinct syntaxin-like t-SNAREs (40). Since Vti1p participate in two distinct transport events (one associated with the secretory pathway and the other in the endosomal pathway), the existence of two distinct mammalian proteins homologous to Vti1p indicates that the two equivalent transport events in mammalian cells may be mediated by two distinct proteins. The preferential association of Vti1-rp2 with the Golgi apparatus indicates that Vti1-rp2 may participate in a transport event in the secretory pathway. Consistent with this, we have shown that microinjection of antibodies against Vti1-rp2 specifically inhibited transport of G-protein to the cell surface at the level of Golgi apparatus. The extents of inhibition of G-protein transport seen in cells microinjected with Vti1-rp2 antibodies are comparable with those seen in cells microinjected with antibodies against ␤-coat protein. Serving as a negative control, G-protein transport to the plasma membrane was unaffected in cells microinjected with antibodies against endosomal syntaxin 7. Vti1-rp2 thus plays a role in protein transport in the secretory pathway, and the role of yeast Vti1p in the secretory pathway is most likely mediated by Vti1-rp2 in mammalian cells. Furthermore, our preliminary studies with Vti1-rp1 suggests that it is preferentially associated with the trans-Golgi network and/or the endosomal compartment, 3 indicating that the endosomal role of yeast Vti1p is most likely mediated by Vti1-rp1 in mammalian cells.