The cytoplasmic domain of Vamp4 and Vamp5 is responsible for their correct subcellular targeting: the N-terminal extenSion of VAMP4 contains a dominant autonomous targeting signal for the trans-Golgi network.

SNAREs represent a superfamily of proteins responsible for the last stage of docking and subsequent fusion in diverse intracellular membrane transport events. The Vamp subfamily of SNAREs contains 7 members (Vamp1, Vamp2, Vamp3/cellubrevin, Vamp4, Vamp5, Vamp7/Ti-Vamp, and Vamp8/endobrevin) that are distributed in various post-Golgi structures. Vamp4 and Vamp5 are distributed predominantly in the trans-Golgi network (TGN) and the plasma membrane, respectively. When C-terminally tagged with enhanced green fluorescent protein, the majority of Vamp4 and Vamp5 is correctly targeted to the TGN and plasma membrane, respectively. Swapping the N-terminal cytoplasmic region and the C-terminal membrane anchor domain between Vamp4 and Vamp5 demonstrates that the N-terminal cytoplasmic region of these two SNAREs contains the correct subcellular targeting information. As compared with Vamp5, Vamp4 contains an N-terminal extension of 51 residues. Appending this 51-residue N-terminal extension onto the N terminus of Vamp5 results in targeting of the chimeric protein to the TGN, suggesting that this N-terminal extension of Vamp4 contains a dominant and autonomous targeting signal for the TGN. Analysis of deletion mutants of this N-terminal region suggests that this TGN-targeting signal is encompassed within a smaller region consisting of a di-Leu motif followed by two acidic clusters. The essential role of the di-Leu motif and the second acidic cluster was then established by site-directed mutagenesis.

Intracellular traffic between different membrane compartments involves diverse membrane-enclosed intermediates in the form of small transport vesicles and/or larger containers/ carriers (1)(2)(3)(4)(5). These transport intermediates are generated at a given donor compartment by the concerted action of membrane proteins as well as cytosolic coat proteins and then delivered to the vicinity of a specific acceptor compartment. After faithful tethering of the intermediates with the target compartment, a process catalyzed by the action of Rab small GTPases and their effectors (6 -7), the final short range docking of the intermediates with the acceptor membrane is mediated by the interaction of vesicle-associated SNAREs 1 with target SNAREs to form trans-SNARE complexes that may mediate the subsequent physical fusion between the intermediates and the target compartment (8 -10).
Combined efforts using approaches of genetics, biochemistry, molecular biology, and bioinformatics have led to the identification of about 35 distinct members of the SNARE superfamily characterized by the presence of a common structural domain referred to as the SNARE motif (11). Based on whether the residue in the center of the SNARE motif is an Arg or a Gln and the relatedness of SNARE sequences, SNAREs have been roughly classified into four families: the R-SNAREs, Qa-SNAREs, Qb-SNAREs, and Q-syntaxins (10,12). For SNAP-23, SNAP-25, and SNAP29/GS32, both Qa-SNARE and Qb-SNARE motifs coexist in the same protein (10). For a given transport event, it is generally believed that each family contributes one SNARE motif so that the four SNARE motifs will form a parallel, highly twisted, four helices bundle that catalyzes the fusion (8). The emerging theme is that the combinatorial use of these SNAREs will give rise to a wide spectrum of trans-SNARE complexes that mediate various transport events (13)(14). A given SNARE may participate in several different transport processes by incorporating into different SNARE complexes. The availability of a specific set of SNAREs is thus important in membrane traffic. The function of SNAREs is regulated primarily through targeting/sorting of specific SNAREs to defined subcellular compartments and/or transport intermediates (8 -10). Additional regulations may be mediated by interacting with sequestering/inhibitory factors (15) and/or phosphorylation (16). The physiological relevance of SNARE targeting/trafficking is reflected by several studies (17)(18)(19)(20) showing the importance of SNARE targeting motifs in various cellular functions of SNARE. The demonstration that a spectrum of transport machinery proteins participates in SNARE trafficking further sustains the importance of SNARE targeting/trafficking in cellular physiology (21)(22)(23). For example, the trafficking signals of Vamp2/synaptobrevin2 are well defined (24 -25), and only wild-type Vamp2, but not mutants defective in proper trafficking, could support regulated secretion when the endogenous protein was inactivated (17)(18). Similarly, only wild-type but not endocytosis-defective Vamp2 could support axonal polarization of developing neurons (19). Snc1p is a yeast Vamp involved in exocytic transport from the late Golgi to the plasma membrane as well as in endocytosis (26 -27). The trafficking of Snc1p is crucial for its function and involves many transport machinery proteins (20 -23, 26 -27). These observations highlight the importance to dissect the targeting signals of SNAREs in our ultimate goal to understand the cellular, physiological, and pathological aspects of this superfamily of proteins.
The seven members of Vamp subfamily are R-type SNAREs anchored to the membrane by their C-terminal transmembrane tails, and they are distributed in various post-Golgi compartments (8). Vamp1 and -2 are primarily found in synaptic vesicles of neurons and secretory granules of endocrine and exocrine cells and function as vesicle-associated SNAREs of regulated secretion (28 -29). Vamp3/cellubrevin is ubiquitously expressed and is primarily found in the sorting and recycling endosomes (29). Associated with the TGN is Vamp4 (30), whereas Vamp5 is primarily distributed on the cell surface of differentiated myotubes in culture and in transfected cells (31). Vamp7/Ti-Vamp has been reported to act both in the lysosome and the apical surface of epithelial cells (29), whereas Vamp8/ endobrevin is detected primarily in early as well as late endosomes (32)(33). The majority of Vamp4 and Vamp5 seems to be distributed at the most proximal (TGN) and most distal (plasma membrane) parts of the post-Golgi structures, respectively, and they thus represent excellent choices for dissecting their targeting signals (30 -31).
In our present study, evidence is presented to suggest that the N-terminal cytoplasmic domain of Vamp4 and Vamp5 is responsible for their subcellular targeting, whereas the transmembrane domain of these two SNAREs plays only a passive role. Furthermore, the N-terminal extension of Vamp4 is shown to contain an autonomous and dominant targeting motif, likely based on a di-Leu motif followed by an acidic cluster, for the TGN. Interestingly, data base searches reveal no equivalent Vamp4 in lower eukaryotes although this N-terminal extension is highly conserved among Vamp4 of several species (human, rat, mouse, zebrafish, and fugu) of vertebrates.

Materials
The expression vector pEGFP-N1 was purchased from Clontech. Normal rat kidney (NRK) cells were obtained form American Type Culture Collection (Manassas, VA). Fluorescein isothiocyanate-conjugated goat anti-mouse IgG and fluorescein isothiocyanate-or rhodamine-conjugated goat anti-rabbit IgG were ordered from Roche Applied Science. Brefeldin A (BFA) was from Epicentre Technologies (Madison, WI). LipofectAMINE was obtained from Invitrogen.
V5V4-EGFP-By using mouse Vamp5 cDNA as the template, oligonucleotides F (see above) and G (5Ј-CATTTGCCTTCGTAAAGTCTTG-GTTGTCTTG-3Ј) were used to retrieve the coding region of the Nterminal cytoplasmic domain of Vamp5. Oligonucleotides H (5Ј-CAAGACTTTACGAAGGCAAATGTGGTGG-3Ј) and B (see above) were used to retrieve the coding region for the C-terminal transmembrane domain of mouse Vamp4 using Vamp4 cDNA as the template. The two PCR products from these two PCRs were mixed and then used for a third PCR using oligonucleotides F and B. The final PCR fragment was gel-purified, cut with EcoRI and BamHI, and inserted into the corresponding sites of pEGFP-N1. V5V4-EGFP consists of N-terminal 58 residues of Vamp5 followed by the C-terminal 34 residues of Vamp4 and EGFP.
V4V5-EGFP-Oligonucleotides A and C (5Ј-CTGCTGGGCAAGCT-GTTTGGATCTGTTG-3Ј) were used to retrieve the coding region for the N-terminal cytoplasmic domain of Vamp4 by PCR using mouse Vamp4 cDNA as the template. The coding region for the C-terminal transmembrane domain of Vamp5 was recovered by PCR using oligonucleotides D (5Ј-CAAACAGCTTGCCCAGCAGAAGCGCTGG-3Ј) and E with mouse Vamp5 cDNA as the template. The PRC products of these two PCRs were then mixed and then used for a third PRC using oligonucleotides A and E. The final PCR product was gel-purified, digested with EcoRI and BamHI, and then inserted into the corresponding sites of pEGFP-N1. V4V5-EGFP consists of N-terminal 105 residues of Vamp4 followed by the C-terminal 46 residues of Vamp5 and EGFP.
V4nV5-EGFP-Oligonucleotides A and J (5Ј-CTGAATTCCCAT-CATTTCTAGGTCCAAATCTTGG-3Ј) were used to retrieve the coding region for the N-terminal 51-residue extension of Vamp4 by PCR using Vamp4 cDNA as the template. The retrieved fragment was cut with EcoRI and ligated into the EcoRI site in construct for expressing human Vamp5-EGFP. Colonies with correct orientation of the inserted fragment were identified by PCR with oligonucleotides A and 3 (right orientation gives rise to a 524-bp fragment).
V4 (30)V5-EGFP-To generate this deletion mutant in the background of V4nV5-EGFP, PCR with oligonucleotides Q (5Ј-CTCAAG-CTTCCACCATGAGGAGAAATCTTTTGGAAGAT-3Ј) and 3 was performed using V4nV5-EGFP as the template. The PCR fragment (about 463 bp) was gel-purified, cut with HindIII and BamHI, and inserted into the corresponding sites of pEGFP-N1.
V4 (19)V5-EGFP-To generate this deletion mutant in the background of V4nV5-EGFP, PCR with oligonucleotides R (5Ј-CTCAAGCT-TCCACCATGGAAGAGGACTTTTTTCTACGG-3Ј) and 3 was performed using V4nV5-EGFP as the template. The PCR fragment (about 430 bp) was gel-purified, digested with HindIII and BamHI, and then inserted into the corresponding sites of pEGFP-N1.
V4nV5-EGFP/LL-AA-With V4nV5-EGFP cDNA as the template, two separate PCR products using oligonucleotides A and N1 (5Ј-TGA ATC ATC TTC CGC AGC ATT TCT CCT CTC-3Ј) as well as N2 (5Ј-GAG AGG AGA AAT GCT GCG GAA GAT GAT TCA-3Ј) and 3 were generated. The two PCR fragments were mixed and then used for a PCR employing oligonucleotides A and 3. The final PCR fragment was gel-purified, digested with EcoRI and BamHI, and then inserted into pEGFP-N1.
V4nV5-EGFP/EDD-3A-With cDNA for V4nV5-EGFP as the template, two separate PCR products using oligonucleotides A and N3 (5Ј-TTC TTC ATC TGA TGC AGC TGC CAA AAG ATT-TCT-3Ј) as well as N4 (5Ј-AGA AAT CTT TTG GCA GCT GCA TCA GAT GAA GAA-3Ј) and 3 were generated. The two PCR fragments were mixed and then used for a PCR employing oligonucleotides A and 3. The final PCR fragment was gel-purified, digested with EcoRI and BamHI, and then inserted into pEGFP-N1.
V4nV5-EGFP/S-A-With cDNA for V4nV5-EGFP as the template, two separate PCR products using oligonucleotides A and N5 (5Ј-CTC TTC TTC ATC TGC ATC ATC TTC CAA-3Ј) as well as oligonucleotides N6 (5Ј-TTG GAA GAT GAT GCA GAT GAA GAA GAG-3Ј) and 3 were generated. The two PCR fragments were mixed and then used for a PCR employing oligonucleotides A and 3. The final PCR fragment was gel-purified, digested with EcoRI and BamHI, and then inserted into pEGFP-N1.
V4nV5-EGFP/5A-With cDNA for V4nV5-EGFP as the template, two separate PCR products using oligonucleotides A and N7 (5Ј-TAG AAA AAA GGC CGC CGC TGC AGC TGA ATC ATC-3Ј) as well as oligonucleotides N8 (5Ј-GAT GAT TCA GCT GCA GCG GCG GCC TTT TTT CTA-3Ј) and 3 were generated. The two PCR fragments were mixed and then used for a PCR employing oligonucleotides A and 3. The final PCR fragment was gel-purified, digested with EcoRI and BamHI, and then inserted into pEGFP-N1.
V4nV5-EGFP/FF-AA-With cDNA for V4nV5-EGFP as the template, two separate PCR products using oligonucleotides A and N9 (5Ј-TGG TCC CCG TAG TGC AGC GTC CTC TTC TTC-3Ј) as well as oligonucleotides N10 (5Ј-GAA GAA GAG GAC GCT GCA CTA CGG GGA CCA-3Ј) and 3 were generated. The two PCR fragments were mixed and then used for a PCR employing oligonucleotides A and 3. The final PCR fragment was gel-purified, digested with EcoRI and BamHI, and then inserted into pEGFP-N1.

Transfection
All the above constructs were transfected into NRK cells using a LipofectAMINE-based system, and this was carried out according to the protocol provided by the manufacturer (31). Stable transfectants were selected by culturing the cells in media with G418 (1000 g/ml), and the pooled transfectants were used for analysis.

Immunofluorescence Microscopy
This was performed as described previously (31). For the treatment of cells with BFA, cells were incubated in culture medium supplemented with BFA at 10 g/ml for 1 h at 37°C. Cells grown on coverslips were washed twice in PBSCM (phosphate-buffered saline supplemented with 1 mM CaCl 2 and 1 mM MgCl 2 ) and then fixed in PBSCM containing 3% paraformaldehyde for 30 min at room temperature. Fixed cells were then permeabilized with 0.1% saponin in PBSCM and immunolabeled with polyclonal antibodies against Golgi SNARE GS15 (14, 34) followed by rhodamine-conjugated goat anti-rabbit IgG. Confocal microscopy was performed with Zeiss AxioplanII microscope (Thornwood, NY) equipped with a Bio-Rad MRC1024 confocal scanning laser.

Selective Surface Biotinylation and Analysis
This was performed as described previously (35). Cells were grown on a 6-well plate until 80 -90% confluency. The expression of EGFP-tagged proteins was induced overnight with 2 mM sodium butyrate. Cells were washed four times with PBSCM (5-10 min each) at 4°C. 1 ml of PBSCM containing s-NHS-biotin (0.5 mg/ml diluted from 200 mg/ml stock in Me 2 SO) was added to each well. The biotinylation was performed twice on ice (15-20 min each) and stopped by washing four times with PBSCM containing 50 mM NH 4 Cl (5-10 min each). After washing twice with PBSCM (5 min each), biotinylated cells were scraped off the plate and then extracted in 1 ml of lysis buffer (25 mM Tris-HCl, pH 7.5, 250 mM NaCl, 5 mM EDTA, 1% Triton X-100, 1% bovine serum albumin, 10% fetal bovine serum, 1 mM phenylmethylsulfonyl fluoride) at 4°C with agitation. The extract was spun down at 14,000 rpm for 10 min at 4°C to clear off cell debris. The supernatant was then incubated with streptavidin-agarose (70 l of 50% slurry) at 4°C for 2 h. After washing once with lysis buffer, three times with buffer A (25 mM Tris-HCl, pH 7.5, 500 mM NaCl, 0.5% Triton X-100, and 1 mM phenylmethylsulfonyl fluoride), and three times with buffer B (10 mM Tris-HCl, pH 7.5, 150 mM NaCl), the beads were boiled for 5 min in 60 l of 2ϫ SDS sample buffer and analyzed by SDS-PAGE. Proteins resolved by SDS-PAGE were then transferred to a nitrocellulose membrane. The filters were then analyzed by immunoblotting to detect the respective proteins using a chemiluminescence detection kit (Pierce).

Vamp4 and Vamp5 Tagged with EGFP at the C Terminus
Are Faithfully Targeted-To have an easy system to follow the targeting of various chimeric and mutant versions of Vamp4 and/or Vamp5, we first tested whether attaching EGFP to the C terminus of Vamp4 and Vamp5 is permissive for their correct subcellular delivery. The entire coding region of Vamp4 (30) and Vamp5 (31) was each inserted in-frame with EGFP in pEGFP-N1 vector, resulting in constructs for expressing Vamp4-EGFP and Vamp5-EGFP, respectively (the upper panel of Fig. 1). NRK cells were transfected with each of these two constructs. In order to achieve expression of these exogenous proteins at levels that are well received by the cells, stably transfected cells were selected, pooled, and expanded. This procedure will ensure that their expression levels were within a certain kind of "physiological range" (i.e. no cytotoxic effects). This is important as studies of SNARE targeting by analyzing transiently transfected cells sometimes could be misleading as compared with conclusions drawn from analyzing endogenous proteins, whereas exogenous proteins in stably transfected cells are normally faithfully targeted (36 -37). The stably transfected cells were then analyzed by indirect immunofluorescence FIG. 1. Vamp5 and Vamp4 tagged with EGFP at the C terminus are correctly targeted to the plasma membrane and TGN, respectively. The entire coding region of Vamp5 and Vamp4 was inserted into pEGFP-N1 vector so that Vamp5 and Vamp4 will be expressed as C-terminally EGFP-tagged proteins (upper panel for the illustration). NE, Nterminal extension; TM, transmembrane. Stably transfected cells expressing mouse Vamp5-EGFP (panels a-c) and mouse Vamp4-EGFP (panels d-i) were then analyzed to view the Vamp5-EGFP and Vamp4-EGFP by their EGFP signal (panels a, d, and g) and endogenous GS15 by antibody labeling (panels b, e, and h). The merged images are also shown (panels c, f, and i). For panels g-i, the cells were first incubated with 10 g/ml BFA for 60 min before being fixed for analysis. Similar targeting of human Vamp5-EGFP to the plasma membrane was also observed (data not show). Bar, 10 m. microscopy to view the EGFP signal of Vamp4-EGFP or Vamp5-EGFP as well as endogenous GS15, a Golgi SNARE (14,34), as revealed by antibody labeling. As shown, Vamp5-EGFP was mainly seen on the plasma membrane as the most intense labeling was found in the cell-cell boundary (Fig. 1, panel a), a characteristic of proteins associated with the plasma membrane. This result is consistent with the known enrichment in the plasma membrane of endogenous Vamp5 in differentiated myotubes and N-terminally Myc-tagged Vamp5 in transfected cells (31). Similar to endogenous Vamp5 and Myc-tagged Vamp5, some Vamp5-EGFP was also seen in the Golgi region, likely representing those molecules in transit in this compartment. The majority of Vamp4-EGFP was seen in the perinuclear structures that were marked by GS15 (Fig. 1, panels d-f), suggesting that the majority of Vamp4-EGPF was targeted to the Golgi apparatus. In order to define the subcompartment of the Golgi apparatus where Vamp4-EGFP is present, we took advantage of the fact that proteins in the cis-Golgi, Golgi stack, and TGN respond differently to the treatment of fungal metabolite BFA (38). When cells were treated with BFA, Vamp4-EGFP (Fig. 1, panel g) was re-distributed into a compact structure near the microtubular organizing center (MTOC), a characteristic response of proteins in the TGN, including endogenous Vamp4 (30,38). Consistent with its enrichment in the Golgi stack (14,34), the majority of GS15 (Fig. 1, panel h) was redistributed into an endoplasmic reticulum-like labeling. These results suggest that, similar to endogenous Vamp4, Vamp4-EGFP was faithfully targeted to the TGN. Therefore, Vamp4 and Vamp5 C-terminally tagged with EGFP are correctly targeted to their subcellular destinations in the stable transfectants.
The Cytoplasmic Domain of Vamp4 and Vamp5 Is Responsible for Their Correct Subcellular Localization-Similar to Vamp1, Vamp2, Vamp3/cellubrevin, and Vamp8/endobrevin, Vamp5 consists essentially of the cytoplasmic SNARE domain followed by the C-terminal hydrophobic tail region (28,31), whereas Vamp4 and Vamp7 have an additional N-terminal extension of about 51 and 113 residues, respectively (30,39), as compared with the other five Vamps. We next investigated whether the targeting information of Vamp4 and Vamp5 resides in the cytoplasmic domain, the tail anchor region, or both. The cytoplasmic domain and tail anchor region of Vamp4 and Vamp5 were swapped in the context of C-terminally EGFPtagged versions, resulting in the creation of V5V4-EGFP and V4V5-EGFP, respectively (Fig. 2, upper panel). These two swapping constructs were transfected into NRK cells, and the resulting stable transfectants were analyzed. As shown, V5V4-EGFP, which consists of the N-terminal 58-residue cytoplasmic domain of Vamp5 followed by the C-terminal 34-residue transmembrane tail anchor of Vamp4, was targeted to the plasma  , e, and h). The merged images are also shown (panels c, f, and i). Cells in panels g-i were first incubated with 10 g/ml BFA for 60 min before being fixed for analysis. Bar, 10 m. membrane (Fig. 2, panel a), in much the same way as did Vamp5-EGFP. This result suggests that the cytoplasmic SNARE domain of Vamp5 but not its tail anchor is responsible for its correct targeting to the plasma membrane. Similarly, V4V5-EGFP, which consists of the N-terminal 105-residue cytoplasmic domain (the N-terminal extension plus the SNARE region) of Vamp4 followed by the C-terminal 46-residue transmembrane tail anchor of Vamp5, was predominantly targeted to the peri-nuclear Golgi apparatus (Fig. 2, panel d) marked by GS15 (panel e). Furthermore, BFA treatment resulted in the redistribution of V4V5-EGFP into a compact structure near the MTOC (Fig. 2, panel g), characteristic of TGN proteins, whereas GS15 was mainly redistributed into an endoplasmic reticulum-like labeling (panel h). These results suggest that the majority of V4V5-EGFP was targeted to the TGN and that the cytoplasmic domain but not the tail anchor of Vamp4 is responsible for TGN targeting.

The N-terminal Extension of Vamp4 Contains a Dominant and Autonomous Targeting Signal for the TGN-Because
Vamp4 contains an N-terminal extension of 51 residues relative to Vamp5 (30 -31), we were interested in the possibility that this unique region may harbor a targeting motif for the TGN. We have therefore appended this N-terminal extension of Vamp4 directly onto the N terminus (replacing the initiation Met) of Vamp5, resulting in a construct for expressing V4nV5-EGFP (Fig. 3, upper panel). Stable transfectants were selected after transfecting NRK cells and then analyzed. As shown, V4nV5-EGFP (Fig. 3, panel a) was colocalized with GS15 (panel b) in the Golgi apparatus. Upon treatment with BFA, V4nV5-EGFP (Fig. 3, panel d) behaved like endogenous Vamp4 and Vamp4-EGFP in that it was redistributed into a compact structure near MTOC. Under the same conditions, the majority of GS15 responded by re-distributing into an endoplasmic reticulum-like labeling (Fig. 3, panel e). These results establish that this N-terminal 51-residue extension of Vamp4 contains a dominant and autonomous signal for targeting to the TGN.
In order to support the conclusions drawn from the immu-nofluorescence analysis, we have employed an independent and more quantitative approach to assess the relative distribution of various chimeric proteins on the plasma membrane and internal compartments (in this case, the Golgi apparatus). Stable transfectants expressing Vamp5-EGFP, V5V4-EGFP, and V4nV5-EGPF were each surface-biotinylated with membraneimpermeable s-NHS-biotin at 4°C so that only proteins exposed on the outside of the plasma membrane will be accessible to biotinylation, whereas proteins in the intracellular compartments such as the Golgi apparatus and cytosol will not be biotinylated (35). Because the EGFP portion of these chimeric proteins is located either in the lumen of intracellular compartments or exposed on the outside of the plasma membrane, the extent of biotinylation of various chimeric proteins will be a reliable assessment of the amount exposed on the plasma membrane. In these experiments, ␣ 5 integrin was used as a positive control for surface-exposed protein (40), whereas peripheral Golgi and cytosolic ADP-ribosylation factors detected by monoclonal antibody 1D9 (41) served as a negative control. As shown in Fig. 4, the extent of surface biotinylation of Vamp5-EGFP (lanes 1 and 2, upper panel) was comparable with that of integrin (lanes 1 and 2, middle panel), whereas ADP-ribosylation factors were not biotinylated (lanes 1 and 2, lower panel). This observation establishes that, like ␣ 5 integrin, Vamp5-EGFP behaves like a typical plasma membrane protein. Vamp5-EGFP is therefore primarily targeted to the surface in the entire population of stably transfected cells. Consistent with results derived from immunofluorescence studies, V5V4-EGFP was readily accessible to surface biotinylation to a similar extent (Fig. 4,  lanes 3 and 4) and is therefore primarily associated with the plasma membrane. In marked contrast, V4nV5-EGFP was essentially not biotinylated under conditions that ␣ 5 integrin was efficiently biotinylated (Fig. 4, lanes 5 and 6). This result is consistent with the primary association of V4nV5-EGFP with the TGN as revealed by immunofluorescence microscopy.  For panels d-f, the cells were first incubated with 10 g/ml BFA for 60 min before being fixed for analysis. Bar, 10 m.
Acidic Clusters-Visual examination of the 51-residue sequence of Vamp4 N-terminal extension revealed the presence of a di-Leu motif followed by two clusters of acidic residues (upper panel of Fig. 5). In order to assess whether this smaller region plays a critical role in TGN targeting, two deletion mutants were created in the context of V4nV5-EGFP. V4(30)V5-EGFP has a deletion of the N-terminal 21 residues (replaced by a new initiation Met) of this 51-residue extension, whereas the di-Leu motif and the acidic clusters remain intact. V4 (19)V5-EGFP has a deletion of N-terminal 32 residues (replaced by a new initiation Met) so that the di-Leu motif, the proximal acidic cluster, and two residues of the distal acidic cluster are deleted. Constructs for expressing V4(30)V5-EGFP and V4(19)V5-EGFP were stably transfected into NRK cells and then analyzed. As shown, V4(30)V5-EGFP (Fig. 5, panel a) colocalized with GS15 (panel b) in the Golgi apparatus, whereas V4(19)V5-EGFP (panel d) was predominantly delivered to the plasma membrane. BFA treatment also shifted the majority of V4(30)V5-EGFP into a compact structure near MTOC (data not shown). These results suggest that the major determinants of this autonomous TGN targeting signal lie in the small region containing the di-Leu motif and the downstream acidic clusters.
An Essential Role of the Di-Leu Motif and the Distal Acidic Cluster in TGN Targeting-To identify the key residues in the N-terminal extension of Vamp4 that are crucial for this TGN targeting capability, we have created five different mutants via site-directed mutagenesis in the context of V4nV5-EGFP (Fig.  6). V4nV5-EGFP/LL-AA had the di-Leu motif (residues [25][26] replaced by two Ala residues. The first acidic cluster EDD (residues 27-29) was substituted by three Ala residues in V4nV5-EGFP/EDD-3A. In V4nV5-EGFP/S-A, the Ser-30 residue (between the two acidic clusters) was changed to an Ala residue. The distal acidic cluster (DEEED, residues 31-35) was replaced by a stretch of five Ala residues in V4nV5-EGFP/A5, whereas the two Phe residues at positions 36 -37 were replaced by Ala residues in V4nV5-EGFP/FF-AA. Constructs for expressing these mutants were each transfected into NRK cells.
Stably transfected cells were selected, pooled, and expanded. The subcellular distribution of these various mutant forms of V4nV5-EGFP was investigated by immunofluorescence microscopy, as results shown above clearly establish that this is a reliable method for the general assessment of their locations. As shown in Fig. 6, V4nV5-EGFP/EDD-3A (panel b), V4nV5-EGFP/S-A (panel c), and V4nV5-EGFP/FF-AA (panel e) were primarily detected in the Golgi apparatus, suggesting that the first acidic cluster downstream of the di-Leu motif, the Ser residue between the two acidic clusters, and the two Phe residues following the second acidic cluster are not essential for Golgi targeting. In marked contrast, the majority of V4nV5-EGFP/LL-AA was detected on the plasma membrane (Fig. 6,  panel a), clearly establishing a role of this di-Leu motif in mediating Golgi targeting. Similarly, V4nV5-EGFP/A5 was no longer confined to the Golgi apparatus, and it was also distributed as a diffuse pattern broadly present in the cell (Fig. 6,  panel d), suggesting a role of the second acidic cluster in mediating efficient Golgi targeting. As the overall distributions of V4nV5-EGFP/LL-AA and V4nV5-EGFP/A5 are different, it seems that the di-Leu motif and the distal acidic cluster may play overlapping but distinct roles in a concerted action to mediate Golgi targeting.
The N-terminal Extension of Vamp4 Is Highly Conserved in Vertebrates and Vamp4 Seems to be Absent in Lower Eukaryotes-Using the N-terminal 60 residues or the entire sequence of mammalian VAMP4 to search the completed genome and annotated proteome of fly (Drosophila melanogaster), worm (Caenorhabditis elegans), plants (Arabidopsis thaliana and Oryza sativa), and yeasts (Saccharomyces cerevisiae and Schizosaccharomyces pombe) has failed to identify any structural counterpart in these six organisms. However, highly conserved sequences were identified in zebrafish and fugu genomes. Alignment of the N-terminal 54-residue sequences of human, rat, mouse, zebrafish, and fugu Vamp4 revealed that the first 47 N-terminal residues (inclusive of the di-Leu motif and the acidic clusters) are either identical (42 of 47 shown in red with a yellow background) or conserved (5 of 47, indicated by arrows) (Fig. 7). The failure to detect any structural orthologues of Vamp4 in any of the six "completely" sequenced and well annotated genomes indicates that Vamp4 could be a unique member of the Vamp subfamily that is unique to vertebrates. In marked contrast, orthologues of Vamp7, the other Vamp member with an N-terminal extension, were readily identified in many lower eukaryotes.

DISCUSSION
As SNAREs are generally believed to play a key role in the final stage of docking and subsequent fusion events in intracellular traffic of all known eukaryotes (ranging from yeast to fly to human) and the primary regulation of their function is at the level of their subcellular targeting (8 -10, 28 -29), we have chosen two members (Vamp4 and Vamp5) of the Vamp subfamily of SNAREs to address the structural domain and/or motifs responsible for their subcellular targeting. The basis for choosing these two Vamps is that they are clearly segregated into distinct post-Golgi structures (TGN for Vamp4 and plasma membrane for Vamp5) that can be easily resolved by indirect immunofluorescence microscopy. Vamp5 consists of a SNARE domain followed by a tail anchor region, whereas Vamp4 also contains an additional N-terminal 51-residue extension (30 -31). The tagging of EGFP at the C terminus seems to be permissive for their respective subcellular delivery. In order to ensure that we were looking at the targeting of these SNAREs at expression levels that are within some kind of physiological range, the subcellular distribution of chimeric and mutants of Vamp4 and Vamp5 was examined in stably transfected and pooled cells as high levels of expression in transiently transfected cells can result in mis-targeting (36 -37).
After establishing that the C-terminal EGFP tagged Vamp4 and Vamp5 were faithfully targeted, it was then clearly shown that the N-terminal cytoplasmic domain of Vamp4 and Vamp5, but not the C-terminal transmembrane tail anchor, contains the structural information for correct subcellular targeting. Although we have performed site-direct mutagenesis to change several Vamp5-specific residues into the corresponding residues of Vamp4 (such as C9V/A13V; T17I/L21Q) in the cytoplasmic domain of Vamp5, we have been unable so far to identify any specific residues that are important for targeting Vamp5 to the plasma membrane (data not shown), suggesting that a higher order of structural information may be involved in its preferential delivery to the plasma membrane or that its lack of a specific signal could be responsible for its default localization at the plasma membrane. More studies will be needed to sort out these possibilities. Because Vamp4 contains a 51-residue N-terminal extension, we have therefore tested the possibility that this N-terminal extension may contain an autonomous targeting signal. This turns out to be true as appending this N-terminal extension onto the N terminus of Vamp5 resulted in the chimeric protein (V4nV5-EGFP) being delivered to the TGN. Visual examination of the amino acid sequence of this N-terminal extension revealed the presence of a di-Leu motif followed by two acidic clusters separated by a Ser residue. The importance of these motifs was first shown by N-terminal truncation studies of V4nV5-EGFP. Deleting the N-terminal 21residue sequence, leaving these motifs intact, had no major effect on TGN targeting as revealed by the predominant TGN localization of V4(30)V5-EGFP. However, when the di-Leu motif, the first acidic cluster, as well as two residues of the second acidic cluster were deleted, the resulting protein, V4(19)V5-EGFP, lost its property to be delivered to the TGN. It was rather re-routed to the plasma membrane. These results suggest that the key structural elements of this dominant TGN targeting signal are likely to be the di-Leu motif and its downstream acidic clusters. The importance of each of these structural features was then directly investigated by site-directed mutagenesis. It was then firmly established that the di-Leu motif and the second acidic cluster are crucial for efficient Golgi targeting, whereas the first acidic cluster, the Ser residue between the two acidic clusters, and the di-Phe motif are not essential for Golgi targeting. Complementary to our studies showing the sufficiency of this TGN targeting signal, mutation of the di-Leu motif within Vamp4 background also affected TGN targeting of Vamp4 (42), suggesting that this signal is also necessary for TGN targeting of Vamp4. However, the importance of other structural features in Golgi targeting was not assessed in this earlier study (42), whereas our study clearly establishes a role of the second acidic cluster in mediating efficient Golgi targeting. Because mutation of the di-Leu and the second acidic cluster had different effects on the Golgi targeting, these two structural features may work together to mediate the Golgi targeting. Future studies will be needed to understand how these two structural motifs bring about the final capability for Golgi targeting.
Signals based on di-Leu and/or acidic residues have been shown for other proteins, including sortilin, low density lipoprotein receptor-related protein 3 (LRP3), cation-independent mannose 6-phosphate receptor (CI-M6PR), cation-dependent mannose 6-phosphate receptor (CD-M6PR), ␤-site APP-cleaving enzyme 2 (BACE2), p55 tumor necrosis factor receptor 1 (TNFR1p55), furin, and PC6B (43)(44)(45)(46)(47)(48). In these other proteins, the acidic cluster is usually found upstream of the di-Leu motif, which is usually near the C terminus. In Vamp4, the acidic cluster is rather downstream of the di-Leu motif. These structural motifs of Vamp4 are not near the end of the polypeptide. These differences may either reflect different mechanisms of action and/or could partially be explained by their different membrane topologies. As type I integral membrane proteins, the signals in sortilin, LRP3, CI-M6PR, CD-M6PR, BACE2, and TNFR1p55 are located in the very C-terminal cytoplasmic domain in such a way that the acidic cluster is more proximal than the di-Leu motif relative to the lipid bilayer. As a tail-anchored protein for Vamp4, the more C-terminal acidic cluster is actually more proximal than the di-Leu motif relative to the lipid bilayer. In this consideration, the geometry of Vamp4 signal relative to the lipid bilayer is similar to those type I proteins except that the polarity of the polypeptide backbone is reversed.
Although the underlying mechanism for the action of this Vamp4 targeting signal remains to be explored, one possible model could be proposed based on our knowledge about Vamp4 and known mechanisms of action of di-Leu and acidic cluster signals of other proteins. The Golgi targeting of sortilin, furin, and TGN38 involves internalization from the plasma membrane followed by selective delivery from either the early (such as TGN38) or late endosome (such as furin) to the TGN (49). Vamp4 may follow a similar pathway(s) to achieve a steady- FIG. 5. A di-Leu motif and following acidic clusters are key determinants for TGN targeting. Visual examination of the 51-residue sequence of the N-terminal extension of Vamp4 revealed the presence of a di-Leu motif (shown in red) followed by two acidic clusters (acidic residues shown in blue) separated by a Ser residue. As illustrated, two deletion mutants were created: V4(30)V5-EGFP has a truncation of the N-terminal residues 2-21, leaving the di-Leu and acidic clusters intact; whereas V4(19)V5-EGFP has a longer truncation of the N-terminal residues 2-31 with the di-Leu motif and most residues of the acidic clusters being deleted. NE, N-terminal extension; TM, transmembrane. After transfection, stable transfectants expressing V4(30)V5-EGFP (panels a-c) or V4(19)V5-EGFP (panel d) were then analyzed to reveal the EGFP signal (panels a and d) and endogenous GS15 (panel b). The merged image is also shown (panel c). Bar, 10 m. state accumulation in the TGN. For example, the di-Leu motif may be first exploited for efficient endocytosis from the plasma membrane. Once in the endosome, the distal acidic cluster may act alone or in conjunction with the di-Leu motif to mediate specific transport to the TGN. This scenario could explain the observed accumulation of mutant V4nV5-EGFP/LL-AA in the plasma membrane, as this mutant is expected to have defects in internalization from the plasma membrane according to this scenario. The accumulation of V4nV5-EGFP/A5 mutant in a diffuse structure could potentially be explained by a defect in efficient concentration in the endosomes and/or transport to the TGN. As the di-Leu motif is still intact in this mutant, endocytosis from the plasma membrane may still take place so that accumulation of this mutant in the plasma membrane was not observed. The acidic TGN signal of furin and the acidic cluster di-Leu-based signal of CI-M6PR are shown to be regulated by Ser phosphorylation mediated by casein kinase II (or related kinases) in that phosphorylation of Ser residue(s) within the acidic cluster enhances the interaction of the signal with PACS-1 and GGA1-3, respectively (44,47). Interaction of FIG. 6. An essential role of the di-Leu motif and the second acidic cluster in mediating efficient Golgi targeting. The di-Leu motif, the first acidic cluster, the Ser residue between the two acidic clusters, the second acidic cluster, and the di-Phe motif in the context of V4nV5-EGFP were mutated into Ala residues in V4nV5-EGFP/LL-AA (a), V4nV5-EGFP/EDD-3A (b), V4nV5-EGFP/S-A (c), V4nV5-EGFP/5A (d), and V4nV5-EGFP/ FF-AA (e), respectively. Constructs for expressing these mutants were transfected into NRK cells. Stably transfected cells were selected, pooled, and expanded. The cells were then analyzed for the subcellular distribution of these EGFP-tagged proteins. Bar, 10 m.
FIG. 7. The amino acid sequence of Vamp4 N-terminal extension is highly conserved among various species of vertebrates. The N-terminal 54-residue sequences of Vamp4 proteins from human, rat, mouse, zebrafish (zebra), and fugu were aligned. The identical residues are shown in red with a yellow background, and arrows indicate the conserved residues. A red bar above the human sequence indicates the region containing the di-Leu motif and the acidic clusters, whereas the Ser residue between the two acidic clusters is indicated by *. phosphorylated furin with PACS-1 is implicated in its delivery from the endosome to the TGN (47). Dephosphorylation by protein phosphatase 2A reduces the affinity of furin acidic signal with PACS-1 (47). Because there is a Ser residue between the two acidic clusters of Vamp4, and its flanking sequence fits with the consensus ((S/T)XX(D/E)) site for phosphorylation by casein kinase II, it is possible that the dynamic accumulation of Vamp 4 in the TGN could potentially be regulated by phosphorylation and dephosphorylation of this Ser residue. We have therefore directly examined this possibility by replacing this Ser with an Ala residue. Because the resulting mutant V4nV5-EGFP/S-A was primarily detected in the Golgi apparatus, it seems that phosphorylation of this residue is not essential for Golgi accumulation, although a role of this residue in other aspects of trafficking could not be formally excluded at the moment. The di-Leu motif has been shown to be important for Vamp4 to interact with the AP-1 adaptor complex that is operational between the endosome and TGN (42). Taken together, these observations suggest that the di-Leu motif and the second acidic cluster may mediate/regulate interaction with AP-1 and/or other cytosolic factors to govern the selective trafficking from the endosome to the TGN. The detection of some Vamp4 in the early endosome (50) is in agreement with this possibility. Although an earlier study suggests that Vamp4 does not cycle to the plasma membrane (51), our preliminary studies indicate that some Vamp4 could cycle from the plasma membrane back to the TGN (data not shown). Future studies will provide additional insight into this aspect.
The importance of this TGN targeting signal is also evident from the high degree of amino acid sequence identity between the N-terminal extensions of Vamp4 from several species (human, rat, mouse, zebrafish, and fugu) of vertebrates. The residues in the di-Leu motif and acidic clusters are identical in all five species with only one acidic residue (Asp) in mammalian Vamp4 being replaced by a conserved Glu in zebrafish and fugu. Despite the fact that the N-terminal extension of Vamp4 is highly conserved in vertebrate species, extensive data base searches of "completed" genomes and predicted proteomes of six lower eukaryotes (including fly, worm, two plants, and two yeasts) have failed to identify an orthologue in any of these lower eukaryotes. Although it remains to be verified by independent approaches, it raises an intriguing possibility that Vamp4 could be a SNARE unique to the vertebrate, and Vamp4 may likely participate in a transport event(s) that is either unique to the vertebrate or that is developed to a higher level that needs a specific SNARE. Further studies along these lines will shed more light not only on Vamp4 but also on vertebrate physiology. Unlike Vamp4, orthologues of Vamp7 (which also contains an N-terminal extension) were readily identified in many lower eukaryotes.