Vear, a Novel Golgi-associated Protein with VHS and γ-Adaptin “Ear” Domains*

The molecular basis of the selectivity and the details of the vesicle formation in endocytic and secretory pathways are still poorly known and most probably involve as yet unidentified components. Here we describe the cloning, expression, and tissue and cell distribution of a novel protein of 67 kDa (called Vear) that bears homology to several endocytosis-associated proteins in that it has a VHS domain in its N terminus. It is also similar to γ-adaptin, the heavy subunit of AP-1, in having in its C terminus a typical “ear” domain. In immunofluorescence microscopy, Vear was seen in the Golgi complex as judged by a typical distribution pattern, a distinct colocalization with the Golgi marker γ-adaptin, and a sensitivity to treatment of cells with brefeldin A. In cell fractionation, Vear partitioned with the post-nuclear membrane fraction. In transfection experiments, hemagglutinin-tagged full-length Vear and truncated Vear lacking the VHS domain assembled on and caused compaction of the Golgi complex. Golgi association without compaction was seen with the ear domain of Vear, whereas the VHS domain alone showed a diffuse membrane- and vesicle-associated distribution. The Golgi association and the bipartite structure along with the differential targeting of its domains suggest that Vear is involved in heterotypic vesicle/suborganelle interactions associated with the Golgi complex. Tissue-specific function of Vear is suggested by its high level of expression in kidney, muscle, and heart.

Intracellular membrane trafficking, involving budding, transport, and fusion of transport vesicles, provides the mechanism for the movement of proteins between different membrane compartments of a cell (for a review, see Refs. 1 and 2). The specificity in the system is conferred by a selective recognition of the cargo, its specific sequestration in appropriate transport vesicles at the budding site, and discriminating docking at the target membrane. Instrumental in the initiating events of the vesicular trafficking are cytosolic proteins that assemble at the vesicle budding sites to form "coats." The best known coats involve heterotetrameric protein complexes called adaptor complexes 1 and 2 (AP-1 and AP-2), which mediate the recruitment of clathrin to the vesicle budding site and are associated with trafficking from the trans-Golgi network to the late endocytic pathway and from the plasma membrane to the early endosomes, respectively. Other thoroughly studied coats are formed by the "coatamer" proteins COPI and COPII, which are responsible for vesicle formation in the endoplasmic reticulum/Golgi and in the early secretory pathway, respectively. COP proteins are not associated with clathrin (3,4). AP-3, another adaptor complex originally thought to be associated with non-clathrin-dependent membrane trafficking in the trans-Golgi network, has also been shown to bind clathrin (5).
All the AP proteins have a similar heterotetrameric structure, being composed of two heavy chains (adaptins) and one medium () and one small () chain (6). The compositions of the AP proteins are as follows. AP-1: ␥and ␤ 1 -adaptins/ 1 / 1 ; AP-2: ␣and ␤ 2 -adaptins/ 2 / 2 ; and AP-3: ␦and ␤ 3 -adaptins/ 3 / 3 . In electron microscopy, AP proteins show an overall structure consisting of a bulky body that is formed by the N-terminal "head" domains of the two adaptins and by the smaller subunits embedded between them and of "ears" that protrude from the body and that correspond to the C termini of the adaptins (6 -8). There is also a considerable degree of sequence similarity between the various subunits and between the subunits in different species, suggesting conserved functional features. The ␤-adaptins are ϳ85% identical, whereas the others are Ͻ50% similar. All the adaptins have a similar domain structure: a large N-terminal head domain, a prolineand glycine-rich hinge region, and a C-terminal appendage or ear (7,9). ␣and ␥-adaptins have the least overall identity (only 25%). There are similarities also betweenand -subunits (10).
Based on morphological and functional studies, it is suggested that there are at least 12 different sorting events in vesicular membrane trafficking that putatively employ coat proteins (11). On the other hand, not as many coat proteins are known, calling for the presence of novel adaptor or coat proteins. Indeed, several novel isoforms or homologs of known adaptor/coat proteins (see Refs. 12 and 13) and membraneassociated protein complexes that bind to AP proteins (14,15) have recently been described in both mammals and yeast, suggesting that related but variable sets of subunits may be utilized to match the specificity requirements of the trafficking events. There is also an emerging group of proteins that are thought to be involved in the determination of the specificity of the vesicle fusion events. They are loosely called "tethering" proteins due to their extended shape and capacity to mediate interactions between different membrane compartments (16 -18). One of the best studied is Hrs, which contains a VHS and a FYVE domain, forms coiled-coil structures, and associates with endocytic vesicles (19).
In this study, we report the molecular cloning and initial characterization of a novel Golgi-associated protein that contains a VHS domain, which is characteristic of endocytosisassociated proteins (20,21), and a C-terminal end similar to the ear domain of ␥-adaptin. It shows a distinct membrane association and a Golgi-associated localization as determined by cell fractionation, immunofluorescence microscopy, and transfection experiments. Due to these features, we suggest that it plays a role in vesicular trafficking events in or in the immediate vicinity of the Golgi complex. Based on the sequence similarities (VHS domain and ear domain of ␥-adaptin), we call this protein "Vear."

EXPERIMENTAL PROCEDURES
General Methods-Standard solutions; buffers; and procedures for purification and precipitation of DNA, restriction enzyme digestion, and ligation reactions were as described by Sambrook et al. (22). Sequencing was done using an automated ABI PRISM 377XL DNA Sequencer (Perkin-Elmer). Synthetic oligonucleotides were obtained from Amersham Pharmacia Biotech. For sequence analysis and alignments, the Genetics Computer Group Wisconsin Package Program Suite 9.0 and CLUSTAL X (23) programs were used.
Data Base Searches and Sequence Analysis-BLAST searches (24) of EST 1 data bases were carried out on a World Wide Web server at NCBI. As a query, nucleotides 13-420 of the sequence encoding the VHS domain of EAST (GenBank TM accession number AJ224514) (25) were used. In reiterative runs, several matching and overlapping EST clones were found.
Northern Blot Analysis-The cDNA corresponding to nucleotides 501-950 of the coding region of Vear was amplified by polymerase chain reaction and used as a probe. The probe was labeled with [␣-32 P]dCTP (Amersham Pharmacia Biotech) using an Oligolabelling Kit (Amersham Pharmacia Biotech). Northern analysis of multiple human mRNAs was performed on a multiple-tissue Northern blot filter (CLONTECH) following the manufacturer's instructions. Total RNA from the cell lines was prepared using the SV Total RNA Isolation System (Promega). 20 g of total RNA was transferred to a Hybond-N nylon filter (Amersham Pharmacia Biotech). The blots were stripped and reprobed with the control ␤-actin cDNA as recommended by the manufacturer. Autoradiographic images were acquired using a Phos-phorImager (Molecular Dynamics, Inc.).
Cell Culture-COS-7 and MDBK cells were maintained in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) containing 10% heatinactivated fetal bovine serum (Hyclone Laboratories), 2 mM glutamine, and antibiotics (100 units/ml penicillin and 100 g/ml streptomycin). Culture medium for MDBK cells was also supplemented with amphotericin B (0.25 g/ml). MDCK and HeLa cells and human embryonic skin fibroblasts were maintained in Eagle's minimal essential medium with Earle's salts (Life Technologies, Inc.) containing 10% heat-inactivated fetal bovine serum, 2 mM glutamine, and antibiotics.
Antibodies-Antibodies to Vear were raised as follows. The fulllength sequence of Vear was amplified by polymerase chain reaction using Pfu polymerase (Stratagene). The product was subcloned into the BamHI/NotI cloning site of the pGEX-4T-1 expression vector (Amersham Pharmacia Biotech). Expression of the fusion protein in bacterial cells was induced with isopropyl-1-thio-␤-D-galactopyranoside (0.25 mM) at room temperature for 3-5 h. The cells were spun down, resuspended in PBS containing 1% Triton X-100, sonicated, and incubated on ice for 10 min. After centrifugation, supernatants were incubated with glutathione-Sepharose 4B beads (Amersham Pharmacia Biotech). The beads were washed several times in PBS, and the fusion protein was eluted with 20 mM reduced glutathione in 0.1 M Tris-HCl, pH 8.0. The purified fusion protein was used to immunize rabbits. The antisera were collected and affinity-purified on CNBr-activated Sepharose 4B beads (Amersham Pharmacia Biotech) coated with the fusion protein.
The monoclonal anti-␥-adaptin antibody was from Transduction Laboratories (A36120). The polyclonal anti-HA antibody was purchased from Santa Cruz Biotechnology (sc-805).
Transient Transfection-The cDNAs used for the transfection experiments were produced by polymerase chain reaction using Pfu polymerase (Stratagene). Full-length Vear (amino acids 1-612; Vear full ), the VHS domain of Vear (amino acids 29 -165; Vear VHS ), the C-terminal end of Vear (amino acids 170 -612; Vear C ), and the ear domain of Vear (amino acids 483-605; Vear ear ) were cloned into the BamHI/NotI cloning site of the pRK5 vector (a gift from Dr. J. Schlessinger). The HA epitope tag was added to the C terminus of the constructs by primer design. The authenticity of the constructs was confirmed by DNA sequencing. Transient transfections were done using Fugene 6 reagent (Roche Molecular Biochemicals) according to the manufacturer's instructions.
Cell Fractionation-COS-7 cells transfected with HA-Vear full in the pRK5 vector were washed twice in ice-cold PBS and pelleted by centrifugation at 1000 ϫ g for 5 min at 4°C. The pellets were then resuspended in ice-cold homogenization buffer (0.25 M sucrose and 10 mM Tris-HCl, pH 7.4, supplemented with protease inhibitors) and sonicated. The nuclei and cell debris were removed by centrifugation at 1000 ϫ g for 15 min at 4°C. The post-nuclear supernatant was then centrifuged for 60 min at 100,000 ϫ g to obtain cytosolic and total membrane fractions. The pelleted membranes were resuspended in Laemmli sample buffer. Equal amounts of cytosolic and membrane proteins were separated by SDS-polyacrylamide gel electrophoresis and subjected to Western blotting.
Immunofluorescence Microscopy-For immunofluorescence microscopy, the cells were grown on glass coverslips. They were washed in Hanks' salt solution and then fixed either with 4% paraformaldehyde in 100 mM Pipes, pH 6.8, 5 mM EGTA, 2 mM MgCl, and 0.2% Triton X-100 for 10 min or with Ϫ20°C methanol for 10 min. After washing in PBS, the cells were incubated with 10% fetal bovine serum in PBS/glycine for 30 min. They were then overlaid with the primary antibody for 30 min; washed; and incubated with the appropriate Texas Red-conjugated (Jackson ImmunoResearch Laboratories, Inc.), fluorescein isothiocyanate-conjugated (Caltag Laboratories), or Oregon Green-conjugated (Molecular Probes, Inc.) secondary antibody for another 30 min. The cells were viewed under an Olympus BH2 fluorescence microscope equipped with appropriate filters.
The effect of brefeldin A (BFA) on the distribution of Vear was studied by adding BFA (Alexis Corp.) to a culture medium of MDCK cells at a final concentration of 5 g/ml. After incubation for 2, 5, and 30 min at ϩ37°C, the cells were fixed with Ϫ20°C methanol for 10 min and handled for the immunofluorescence detection of Vear and ␥-adaptin as described above.

Molecular
Cloning and Sequence Analysis of Vear-EST data bases were utilized to identify novel proteins that could qualify as proteins involved in vesicular trafficking by virtue of their domain structure. We were especially interested in proteins carrying a VHS domain, a recently discovered domain present in several endocytic proteins (20). A search using the BLAST algorithm and the nucleotide sequence encoding the VHS domain of EAST (nucleotides 13-420) as a query yielded six matching and overlapping ESTs: AA671009, AA279107, AA235896, AA307157, AA070902, and AI017433. Aligning the sequences suggested the presence of an open reading frame encoding a protein of 613 amino acids with a calculated molecular mass of 67,149 Da.
EST clone AA070902 was purchased from the Human Genome Mapping Project Research Center (Hinxton, Cambridge, United Kingdom). Sequencing of it yielded a nucleotide sequence that was identical to that derived from the aligned sequences of AA671009, AA279107, AA235896, AA307157, AA070902 and AI017433. The sequence consists of a 5Ј-untranslated region of 21 nucleotides, an open reading frame of 1842 nucleotides, and a 3Ј-untranslated region of 134 nucleotides. The deduced amino acid sequence is shown in Fig. 1A. Based on the structural features of the protein, we call it Vear (see below). A schematic representation of the structural features and the domain structure of the protein is shown in Fig.  1B. Later on, we noticed that a human chromosome 16 bacte-rial artificial chromosome clone, CIT987SK-A-735G6 (Gen-Bank TM accession number AC002400), describes a putative protein sequence identical to that derived from the EST clones.
Comparison of the deduced amino acid sequence of Vear with the protein sequences in the data bases revealed the presence of a VHS domain in the N terminus and a high degree of similarity to the C-terminal ear domain of ␥-adaptin in the C terminus. Comparison of the N terminus of Vear with various VHS domains and comparison of the C terminus of Vear with the ear domain of ␥and ␥ 2 -adaptins, as generated by CLUSTAL X (23), are shown in Fig. 2. The VHS domain of Vear showed a high degree of similarity to the VHS domains of the following proteins ( Fig. 2A): Tom1 (CAA07361), a v-myb target gene of unknown function (27); Tom1-like protein (CAA08993); Tom1b (CAA69996); Hrs (D50050), a tyrosine kinase substrate localized on early endosomes (19); Hrs-2 (U87863), a SNAP-25interacting protein related to Hrs (28); STAM, a signal transducing adaptor protein that binds to Hrs (29,30); STAM2a (AAC63963); STAM2b (AAC63964); EAST (AJ224514), an epidermal growth factor receptor-and Eps15-associated protein (25); and a Caenorhabditis elegans sequence (CAA92026).
There was no significant overall sequence homology to any protein in the data banks, and a separate comparison of residues 170 -480 did not show any significant similarity to the proteins in the data banks. A distinct feature of this middle region is its high content of prolines. In the beginning of the linker region (amino acids 200 -250), there is an area with a predicted high ␣-helical content and a propensity to coiled-coil organization as predicted by the Coilscan program. Due to the sequence homologies, we call the protein Vear (for VHS and ear).
Tissue Distribution of Vear and Characterization of Anti-Vear Antibody-Northern blot analysis of human tissues revealed two major messages of 3.5 and 3.9 kb that were most abundant in kidney, skeletal muscle, and cardiac muscle (Fig.  3A). They were also seen at a lower level in placenta, lung, thymus, and spleen. In kidney, an abundant message of 5.2 kb was also seen. It was expressed at a lower level also in the other tissues mentioned above. A low level of expression of these species was seen in brain, colon, and small intestine. The nature of the multiple RNA species seen in some tissues is currently unknown, but could correspond to alternatively spliced forms of the Vear mRNA. We also carried out Northern hybridization of the total RNAs isolated from HeLa, COS-7, MDBK, and MDCK cells. A distinct band of 3.5 kb, corresponding to a similar-sized band in tissues, was seen in all cell lines. In HeLa and COS-7 cells, messages of 3.9 and 5.2 kb, respectively, were also seen (Fig. 3B). Western blot analysis of the cell solubilizates with anti-Vear antibody detected a single band of 75 kDa in COS-7, HeLa, MDBK, and MDCK cells (Fig. 4A, lanes 5-8). No bands were detected using a preimmune serum, attesting to the specificity of the reaction (lanes 1-4). The detection of a single band in all cell lines suggests a correspondence with the 3.5-kb mRNA species, which was also shared by all the cells tested (Fig. 3B). The results from Western blotting of the COS-7 cells transfected with the truncated parts of Vear with anti-Vear and anti-HA antibodies are shown in Fig. 4B. A distinct binding of anti-Vear antibody to the C-terminal part (lane 3), but not to the VHS or ear domain (lanes 1 and 2), of Vear is shown. The positive reaction with anti-HA antibody of all the HA-tagged proteins (lanes 4 -6) indicates that the peptides were appropriately expressed and transferred to the filter. These results show that the epitope recognized by anti-Vear antibody lies between amino acids 170 and 482, an area that shows no significant homology to any other protein in the data banks.
Immunofluorescence Microscopy-The subcellular distribution of Vear in MDCK, COS-7, HeLa, and human embryonic skin cells (Fig. 5, a, c, e, and g) was studied by immunofluorescence microscopy using anti-Vear antibody. In all cells tested, Vear was localized to a mostly tubular perinuclear structure closely resembling the Golgi complex both in localization and appearance. No staining was seen when preimmune serum was substituted for anti-Vear antibody. Moreover, the staining was completely inhibited by preincubating the antibodies with the recombinant glutathione S-transferase-Vear fusion protein (Fig. 5h, inset) but not with glutathione S-transferase alone (data not shown). This shows that anti-Vear antibody detects The residues are colored according to the following scheme. All glycines are brown, and prolines are yellow. Other coloring is by a recurring feature: hydrophobic residues (Ala, Cys, Phe, Ile, Leu, Met, and Trp) are light blue; tyrosine and histidine are dark blue; aspartate and glutamate are purple; arginine and lysine are orange; and asparagine, glutamine, serine, and threonine are green. More than 50% occurrence of a property results in coloring. A single fully conserved residue is indicated by an asterisk, strong conservation by a colon, and weak conservation by a dot. No coloring indicates poor conservation of a residue or a property. The conservation score for each column is indicated by the plots below the sequences. The accession numbers of the sequences are as follows: STAM2a, AF042273; STAM2b, AF042274; EAST, AJ224514; STAM, U43899; Hrs, D50050; Hrs-2, U87863; Tom1, AJ006972; Tom1b, Y08741; Tom1-like protein, AJ010071; Vear, AF165531; C. elegans protein, Z68014; ␥ 2 -adaptin, AB015318; and ␥-adaptin, O43747. The position of the first residue of the protein sequence is indicated at the beginning of each sequence. Numbers at the end of the sequences refer to the degree of identity and similarity to Vear as expressed in percentages. endogenous Vear, that the staining reaction is specific, and that Vear is probably associated with the Golgi in different types of cells.
To explore more closely the putative Golgi association of Vear, we compared the distribution of Vear with the localization of ␥-adaptin, a well established marker of the Golgi complex (33,34). Using double staining immunofluorescence microscopy for Vear (Fig. 5, a, c, e, and g) and for ␥-adaptin (Fig.  5, b, d, f, and h), an extensive codistribution was seen between Vear and ␥-adaptin in MDCK, COS-7, HeLa, and human embryonic skin cells. However, within the Golgi-like structure, the staining patterns were nonidentical; Vear was always seen in coarser, slightly tubular structures, whereas ␥-adaptin was seen associated with more delicate vesicular or dot-like structures.
BFA is a fungal metabolite that inhibits coat formation in Golgi membranes, thereby causing reversible apparent dissolution of the Golgi complex in many cells (35). Due to this property, treatment with BFA can be used to explore the pu-tative association of proteins with the Golgi apparatus and also their functional features (36 -38). ␥-Adaptin and ␤-COP are examples of Golgi proteins that are "sensitive" to BFA, i.e. they show an altered subcellular distribution upon BFA treatment of cells (34). Fig. 6 shows the distribution of Vear and ␥-adaptin in MDCK cells treated with BFA for 0, 5, and 30 min. No distinct changes were seen in cells treated for up to 2 min. In cells exposed to BFA for 5 min (Fig. 6c) or 30 min (Fig. 6e), on the other hand, a distinct disassembly of the tubular Vear-positive structure to numerous and more widely distributed vesicles was seen. A similar but not identical change was seen in the distribution of ␥-adaptin (Fig. 6, d and f). These results strongly suggest that Vear is associated with a BFA-sensitive subcompartment of the Golgi.
Cell Fractionation Studies-The subcellular distribution of Vear was also studied by subjecting Vear-transfected COS-7 cells to fractionation by ultracentrifugation, followed by Western blotting of the post-nuclear membrane and cytosolic fractions using anti-Vear antibody. Vear was seen to be present almost exclusively in the membrane fraction (Fig. 7); only a faint band corresponding to Vear was seen in the cytosolic fraction after a longer exposure (data not shown). This indicates that most of Vear is associated with cellular membranes and is well in line with the immunofluorescence microscopy observations that showed a Golgi-associated localization without any major cytosolic distribution.
Overexpression of Vear in Cultured Cells-To further investigate the localization and function of Vear in cultured cells, we tagged recombinant full-length Vear and its distinct domain regions with the HA epitope (HA-Vear) and transiently overexpressed these in cultured COS-7 cells (Fig. 8) and MDCK cells (data not shown). In most transfected cells, HA-Vear full was seen confined to the Golgi-like perinuclear structure. Double staining with anti-␥-adaptin antibodies showed a distinct codistribution verifying the Golgi-associated localization (Fig.   FIG. 3. Expression of Vear RNA in various human tissues and in different cell lines. Shown are the results from Northern hybridization analysis of the mRNA isolated from various human tissues (A) and total RNA isolated from HeLa, COS-7, MDBK, and MDCK cells (B). Hybridization for Vear RNA was carried out using a probe corresponding to nucleotides 501-950 of the Vear sequence (upper panels). Rehybridization was with a ␤-actin probe (lower panels). In B, 20 g of total RNA was loaded in each lane.  1 and 4), HA-Vear ear (lanes 2 and 5), or HA-Vear C (lanes 3 and 6). Immunoblotting was carried out using anti-Vear (lanes 1-3) and anti-HA (lanes 4 -6) antibodies. I.B., immunoblotting. 8, a and b). In most HA-Vear full -transfected cells, a distinctly dense, sharply delineated, "compacted" staining pattern was seen with anti-HA antibody. This was always associated with a correspondingly condensed ␥-adaptin distribution, which clearly differed from the more "non-compacted" distribution seen in nontransfected cells (Fig. 8b). A similar distribution was seen with HA-Vear C . In most cells, HA-Vear C showed a perinuclear Golgi-like distribution (Fig. 8c), which was similar to the compacted distribution of HA-Vear full and completely coincided with the distribution of ␥-adaptin (Fig. 8d). In nontransfected cells, a less compact and wider Golgi-associated distribution of ␥-adaptin was seen (Fig. 8, b and d, asterisks). In HA-Vear ear -transfected cells (Fig. 8e), a Golgi-associated staining was also seen. However, there was no distinct change in the morphology of the Golgi as judged by staining for ␥-adaptin in transfected and nontransfected cells (Fig. 8f). In marked contrast to the above, HA-Vear VHS showed a widespread membrane-and vesicle-associated distribution in the cytoplasm without any preferential localization to the Golgi complex (Fig.  8g). Its pattern was clearly different from that of the Golgi and was not associated with any change in the Golgi morphology as judged by staining with anti-␥-adaptin antibody (Fig. 8h). DISCUSSION In this study, we have identified and characterized Vear, a novel Golgi-associated protein with a unique domain structure. Vear is expressed at low levels in most human tissues and at a high level in kidney and skeletal and cardiac muscle. Both endogenous and exogenously expressed Vear proteins are confined to the Golgi complex, in which they colocalize with the Golgi marker ␥-adaptin.
The Golgi association of Vear was demonstrated by both immunofluorescence microscopy and cell fractionation studies. Moreover, HA-tagged Vear assembled on the Golgi upon cell transfection. No significant cytosolic pool of either endogenous or exogenously expressed Vear was identified, attesting to the predominantly membrane-associated localization. Comparison with ␥-adaptin showed a close codistribution in cultured cells, suggesting a Golgi-associated localization, which was further supported by a similar response of both Vear and ␥-adaptin to treatment with BFA. It is noteworthy, however, that, in closer  detail, the localizations of ␥-adaptin and Vear differ; instead of a finely vesicular arrangement along the Golgi, characteristic of ␥-adaptin, Vear was seen mostly associated with more elongated, tubule-like structures. This suggests a localization for Vear that is wider than just the trans-Golgi network, a Golgi subcompartment marked by the presence of ␥-adaptin (39). A more accurate assignment of Vear as to the substructure of the Golgi remains, however, to be determined.
The Golgi association of Vear was also supported by the distribution pattern of exogenously expressed Vear; it also showed a codistribution with ␥-adaptin. Intriguingly, however, in many transfected cells, overexpression of Vear was accompanied by an occurrence of a compacted Golgi. This differed from the distribution of endogenous Vear and ␥-adaptin in nontransfected cells in that it was denser and more sharply delineated. Moreover, in cells with a compacted Golgi, there were less scattered Vear-positive vesicles around the periphery of the Golgi. This suggests that the overexpression of Vear could cause the "compaction" either by preventing a "physiological" dispersal of the structure or by causing coalescence of the scattered vesicles on the Golgi structure.
Separate expression of the N-terminal VHS domain and the C-terminal end of Vear encompassing both the ␣-helical and proline-rich middle regions and the ear or the ear alone showed interesting differences that could be indicative of the function of Vear. Clearly, overexpression of the VHS domain led to a very widespread, mostly membrane-associated distribution that had no specific relation to or effect on the Golgi. The ear domain-containing constructs that were devoid of VHS, on the other hand, showed a Golgi-associated localization. Moreover, Vear C seemed to cause a compaction of the Golgi that was similar to that seen with full-length Vear. All in all, this strongly suggests that the VHS and ear domains of Vear serve different targeting functions; the VHS domain probably binds to a ligand that is widely present in cellular membranes, whereas the ear domain seems to bind specifically to a Golgiassociated ligand.
The VHS domain occurs almost invariably in proteins known to be involved in endocytosis or vesicular trafficking (21,40), but its precise function is not known. It is noteworthy, however, that when expressed alone, the VHS domain of EAST associates with the plasma membrane as judged by immunofluorescence microscopy (25). A function for targeting to vesicular membranes is also supported by its presence in endocytic vesicle-associated Hrs and by targeting to vesicles of Hrs severed of the FYVE domain, a known membrane targeting signal (19). Thus, it is quite likely that the VHS domain binds to a target that is widely present on cytoplasmic vesicles and membranes.
The presence in Vear of the ear domain is of special interest, first, because in adaptins, it has been suggested to have a targeting function (32,39), and, second, because it has not been found outside the realm of adaptins before. In adaptins, ear domains have a lower degree of similarity than, for example, the head domains, which have been shown to account, at least partially, for the specific membrane targeting and for the interaction with -subunits in ␣and ␥-adaptins (32,39). On the other hand, studies employing chimeric and mutant ␣and ␥-adaptins have shown that the ear domains also contribute to membrane binding possibly by stabilizing an association initiated by some other determinants (39). Clearly, this, together with the distinct localization of the exogenously expressed ear domain of Vear in the Golgi, strongly speaks in favor of it critically determining the Golgi-based localization of Vear.
The distinct difference seen in the Golgi organization of cells transfected with Vear C and Vear ear (compacted versus noncompacted Golgi) points to a potential importance of the middle portion of Vear. Interestingly, it encompasses a region with a propensity to coiled-coil arrangement and a proline-rich region. Supposedly, the presence of the coiled-coil region could lead either to homodimerization of Vear or to heterodimer formation with some other proteins. This could potentially lead to not only targeting of Vear, but also to bringing to a closer apposition the target organelles associated with the ends of Vear and/or of its putative dimerization partner. In this sense, Vear could be a novel member of a newly described group of tethering proteins that have been described in the vesicular transport systems such as the Golgi and that typically form bridges between different membrane elements (41,42). Further studies are needed to determine whether, analogously to the Golgiassociated members of this class of proteins, Vear could be involved in the docking of the vesicles to the membranes or in the stacking of the cisternae in the Golgi.