Topological characterization of the c, c', and c" subunits of the vacuolar ATPase from the yeast Saccharomyces cerevisiae.

The vacuolar ATPase (V-ATPase) is a multisubunit enzyme that acidifies intracellular organelles in eukaryotes. Similar to the F-type ATP synthase (FATPase), the V-ATPase is composed of two subcomplexes, V(1) and V(0). Hydrolysis of ATP in the V(1) subcomplex is tightly coupled to proton translocation accomplished by the V(0) subcomplex, which is composed of five unique subunits (a, d, c, c', and c"). Three of the subunits, subunit c (Vma3p), c' (Vma11p), and c" (Vma16p), are small highly hydrophobic integral membrane proteins called "proteolipids" that share sequence similarity to the F-ATPase subunit c. Whereas subunit c from the F-ATPase spans the membrane bilayer twice, the V-ATPase proteolipids have been modeled to have at least four transmembrane-spanning helices. Limited proteolysis experiments with epitope-tagged copies of the proteolipids have revealed that the N and the C termini of c (Vma3p) and c' (Vma11p) were in the lumen of the vacuole. Limited proteolysis of epitope-tagged c" (Vma16p) indicated that the N terminus is located on the cytoplasmic face of the vacuole, whereas the C terminus is located within the vacuole. Furthermore, a chimeric fusion between Vma16p and Vma3p, Vma16-Vma3p, was found to assemble into a fully functional V-ATPase complex, further supporting the conclusion that the C terminus of Vma16p resides within the lumen of the vacuole. These results indicate that subunits c and c' have four transmembrane segments with their N and C termini in the lumen and that c" has five transmembrane segments, with the N terminus exposed to the cytosol and the C terminus lumenal.

The vacuolar H ϩ -ATPase (V-ATPase) 1 is a multisubunit enzyme responsible for the lumenal acidification of cellular organelles (1). Organelle acidification is essential for a variety of cellular processes such as receptor-mediated endocytosis, activation of proteases, and proton-coupled transport of small molecules and ions (2)(3)(4). Similar to the F-type ATP synthase, the V-ATPase is composed of two distinct sectors; a peripheral catalytic complex (V 1 ) and a membrane associated complex (V 0 ).
The peripheral catalytic V 1 subcomplex is a 560-kDa domain composed of eight subunits (A, B, C, D, E, F, G, and H). Its catalytic core is composed of a hexameric ring of alternating A (69 kDa) and B (57 kDa) subunits (5). Both the A and B subunits contain consensus sequences for ATP binding domains, where subunit A has been ascribed with the catalytic function (6), and subunit B is proposed to have a regulatory role (7). The role of the other V 1 subunits is not so clear. It is thought that the central stalk may be composed of D (32 kDa) and F (14 kDa) subunits (8,9). The stator region is thought to be composed of E (27 kDa), G (13 kDa), C (42 kDa), and H (54 kDa) subunits (9 -13). All subunits are required for a fully assembled complex with the exception of the H subunit; however, the H subunit is required for a functional complex (14). The membrane-associated V 0 subcomplex is composed of five subunits (a, c, cЈ, cЉ, and d). There is a ring of the proteolipid subunits Vma3p (16.5 kDa, subunit c), Vma11p (17 kDa, subunit cЈ), and Vma16p (23 kDa, subunit cЉ), and each proteolipid subunit contains a conserved acidic residue required for proton translocation (15). The proteolipid ring contains one copy each of the cЈ and cЉ subunits (16), with 4 -5 copies of subunit c (4,5). The a (100-kDa) subunit has two isoforms, Vph1p and Stv1p. The two isoforms specify to which organelles the V-ATPase complex is routed; the Vph1p containing V-ATPase complex localizes to the vacuole, whereas the Stv1p-containing complex remains in the Golgi/endosome membranes (17,18). The a subunit contains two domains, a hydrophilic N terminus domain believed to form part of the stator and a hydrophobic carboxyl domain, which also forms part of the proton pore (19 -21). The function of the d (36 kDa) subunit is currently unknown, but recently a x-ray crystal structure has been published, in which the authors propose that the d subunit may cap the top of the proteolipid ring (22). Like the F-type ATP synthase, the V-ATPase has been shown to work by a rotary mechanism. ATP is hydrolyzed by the A subunit causing a conformation change and, thus, rotating the central stalk composed of F and D (23,24). The stalk is fixed to the ring of c subunits and causes it to rotate as well (25). The c subunit ring rotates relative to the stator, thus pumping protons with a rotary mechanism. Unlike the F-type ATP synthase, the V-ATPase functions physiologically to pump protons across the lipid bilayer and not to synthesize ATP.
NMR and x-ray crystallography studies have shown that the F-type ATP synthase 8-kDa-subunit forms an ␣-helical hairpin with two transmembrane helices such that the N termini and the C termini oriented away from the F 1 subcomplex (26 -29). There is no high resolution structure of the c subunits from the V-ATPase. However, based upon the similarity to the F-type ATP synthase c subunit and hydropathy plots, Vma3p and Vma11p can be modeled to have four transmembrane helices and Vma16p to have five transmembrane helices (3). The V-ATPase subunits may have arisen from a gene duplication event of a common c subunit ancestor (30 -32). This would suggest that the orientation of the termini of Vma3p and Vma11p would lie within the lumen of the vacuole, whereas Vma16p would have either the N terminus or the C terminus in the lumen of the vacuole. Previous studies have suggested that Vma16p (cЉ) has either four (33) or five (34) transmembrane regions, and these studies came to different conclusions about the requirement for the first putative transmembrane helix of Vma16p.
In this paper we describe the molecular and biochemical approach that we used to address the topology of Vma3p, Vma11p, and Vma16p. We have attached epitope tags to the N and C termini of these three subunits and used protease protection assays to determine the orientation of the termini. Through immunoblot analysis of the limited proteolysis we have determined that the N and C termini of Vma3p and Vma11p are located within the lumen of the vacuole, whereas the N terminus of Vma16p is located in the cytoplasm, and the C terminus is in the lumen. In addition, we have also constructed chimeric fusions of Vma16p and Vma3p that further support our topological model of Vma16p.

EXPERIMENTAL PROCEDURES
Strains and Culture Conditions-The yeast strains and plasmids used in this study are listed in Tables I and II. All yeast strains were cultured in SD minimal media (0.67% yeast nitrogen base, 2% dextrose) supplemented with the appropriate amino acids or YEPD medium buffered to pH 5.0 using 50 mM succinate/phosphate. To test for a Vma Ϫ phenotype, saturated cultures were diluted to an A 600 ϭ 0.25. Serial dilutions of the diluted culture were applied to pH 7.5 YEPD plus 100 mM CaCl 2 media and allowed to grow overnight before observing results.
Plasmid Construction and Epitope Tagging-DNA sequence encoding a single c-Myc epitope was introduced immediately after the start codon of pLG84 to generate pAJ82 by PCR using outward facing oligonucleotides. PCR using outward-facing oligonucleotides was also used to delete base pairs 34 -114 from pAJ82 to generate pLG96. QuikChange mutagenesis was used to introduce an MluI site (A/ CGCGT) immediately after the start codon of VMA3 in plasmid pRHA316 (16). PCR was used to construct a 1ϫHA::VMA3 from pLG93 that was flanked by an MluI 5Ј site and a BglII 3Ј site. The MluI/BglIIdigested PCR fragment was then ligated into a MluI/BglII-digested pLG93 to create pAF271. QuikChange mutagenesis (Stratagene) was used to introduce a BglII site ((A/G)ATCT) immediately after the start codon in pRS316VMA11(pNUVA388) to generate pLG159. Complementary oligonucleotides were synthesized so that when annealed together they formed an oligonucleotide duplex encoding a single HA epitope with BglII ends. The oligonucleotide duplex was ligated into BglIIdigested pLG159 to generate pLG171. QuikChange mutagenesis was used to introduce an MluI site (A/CGCGT) immediately after the start codon in pLG93 to generate pLG101. A VMA16::3ϫHA fragment was generated from pLG84 by PCR using oligonucleotides that introduced MluI sites on the ends of the PCR fragment. MluI-digested pLG101 was ligated with MluI-digested VMA16::3ϫHA PCR fragment to generate pLG146. A BglII site was inserted immediately after the start codon in pRS316VMA16::c-myc (pLG33) to generate pLG94. A 3ϫHA BglII fragment from BJ7122 was ligated into digested pLG33 to form pLG94. A 3ϫHA::VMA16 PCR fragment was generated from pLG94 by PCR using oligonucleotides that introduce MluI sites on the ends of the fragment. QuikChange mutagenesis was used to also introduce a MluI site immediately before the stop codon of pRS316VMA3 (pRHA313) to generate pLG100. The 3ϫHA::VMA16 MluI-digested PCR fragment was ligated with digested pLG100 to create pLG145.
Proteolytic Protection Assays-Intact vacuolar vesicles were isolated using a previously described procedure with modifications (21,35). Briefly, yeast were grown to log phase of A 600 ϭ 4.0 in YEPD medium buffered to pH 5.0, cells were converted to spheroplast, and lysed with a loosely fitting Dounce homogenizer in Buffer A (10 mM Mes/Tris, pH 6.9, 0.1 mM MgCl 2 , 12% Ficoll). The lysate was centrifuged at 60,000 ϫ g for 45 min in a swinging bucket rotor. The floating white layer containing the vacuoles were removed with a weigh spatula and resuspended in 10 ml of modified buffer 88 (M88) (20 mM HEPES, pH 6.8, 150 mM potassium acetate, 1 mM magnesium acetate, 1 mM CaCl 2 , 250 mM sorbitol). The vacuoles were recovered by centrifugation at 13,000 ϫ g, washed twice in M88 buffer, and brought up to a final protein concentration of 4 mg/ml. Samples containing 1 mg of protein were treated at 4°C with trypsin (0.5 mg/ml) and proteinase K (0.5 mg/ml) in the absence or presence of 1% Triton X-100 for 10 min. The reactions were quenched by the addition of an equal volume of 40% trichloroacetic acid. The samples were pelleted at 13,000 ϫ g and washed with an equal volume of ice-cold acetone. The pellets were resuspended in Thorner buffer (8 M urea, 5% SDS, 40 mM Tris, pH 6.8, 0.1 mM EDTA, 0.4 mg/ml bromphenol blue, ␤-mercaptoethanol added fresh to 5%), and 10 g of protein was loaded per lane of a SDS-PAGE.
Protein Preparation, SDS-PAGE, and Immunoblot Analysis-Whole cell extracts and vacuolar membranes were prepared as described previously (35,36). SDS-PAGE analysis was performed. Proteins were transferred to 0.2-m nitrocellulose and probed with either an affinitypurified monoclonal anti-HA (Covance Research Products Inc.) used at 1:3000, the anti-Myc 9E10 described previously (37) used at 1:5, antibodies 10D7 or 7B1 used to probe for Vph1p at 1:500 (Molecular Probes), the monoclonal antibody 13D11 directed against Vma2p used at 1:1000 (Molecular Probes), or the monoclonal antibody 10E5 used to probe for CPY at 1:500 (Molecular Probes). Proteins were visualized using affinity-purified donkey anti-mouse secondary antibodies conjugated to horseradish peroxidase (1:20000) (Jackson ImmunoResearch Laboratories Inc.) and chemiluminescence (Amersham Biosciences) Fluorescence Microscopy-Quinacrine staining of live yeast cells was conducted as previously described (36) with the following modification. Concanavalin A TRITC (Molecular Probes) was added at a final concentration of 50 g/ml with the quinacrine to allow for visualization of the cell surface. Images were acquired on a Zeiss Axioplan 2 microscope and manipulated using AxioVision software.
Native Immunoprecipitation-Immunoprecipitations were carried out as previously described with the following exceptions (16). 1 mg of solubilized membranes was precleared with the addition of 50 l of 50% slurry of Protein G-Sepharose 4 Fast Flow (Amersham Biosciences). 5 l of affinity-purified goat anti-HA antibody (Novus Biologicals) was added to the pre-cleared supernatant and agitated at 4°C for 1 h. 50 l of 50% slurry of protein G-Sepharose 4 fast flow was added and incubated at 4°C for 1 h. The beads were collected with a brief centrifugation and washed 3 times with ice-cold phosphate-buffered saline plus 1% C 12 E 9 . The beads were resuspended in 200 l of Thorner buffer, and 20 l of the supernatant was loaded per well of SDS-PAGE and analyzed by immunoblot analysis.
V-ATPase Assays-ATPase activity was measured by a coupled spectrophotometric assay in the absence and presence of 1 M concanamycin A, as described previously (36).

RESULTS
Determining Topology of C Termini for Vma3p, Vma11p, and Vma16p-To determine whether the C termini of Vma3p, Vma11p, and Vma16p are located on the cytosolic face of the vacuole or in the lumen, the proteins were epitope-tagged with the HA epitope on their C termini. Previous reports demonstrated that strains expressing tagged copies of Vma3p, Vma11p, or Vma16p produced V-ATPase complexes indistinguishable from wild-type cells (15,16), indicating that the epitope-tagged forms of these proteins are fully functional.
To determine the topology of the proteins, the intact vesicles were isolated from strains expressing tagged Vma3p, Vma11p, or Vma16p and subjected to limited proteolysis. Vacuoles were incubated with a mixture of proteinase K and trypsin either in the presence or absence of 1% Triton X-100 at 0°C for 10 min (21). SDS-PAGE and immunoblot analysis were performed on the time points taken during the protease reaction. Two proteins, Vph1p and CPY, were monitored as controls. Vph1p confirms the accessibility of the proteases to proteins oriented on the cytoplasmic side, whereas CPY validates the intactness of the vacuoles. Using the 7B1 anti-Vph1p antibody, Jackson and Stevens (21) show that the N-terminal domain of Vph1p is located in the cytoplasm and is easily susceptible to proteolytic degradation without compromising the integrity of the membrane. CPY is translated with a propeptide domain, which is cleaved when CPY reaches the vacuole to yield the mature form of CPY (36). Because vacuoles were isolated from a yeast strain deficient in active vacuolar proteases due to a mutation in the PEP4 gene, only the proCPY form was present in the vacuole. The anti-Vph1p blot in Fig. 1a illustrates that the protein is degraded by protease in both the absence or presence of detergent, confirming the proteases were active. In contrast, the shift of the proCPY form to the mature CPY can only be seen in the presence of detergent, confirming the intactness of the vacuoles.
These results indicate that the isolated vacuoles were properly oriented and intact throughout the limited proteolysis assay. The samples were also probed using an antibody directed against the HA epitope (Fig. 1a). The epitopes present on Vma3p, Vma11p, and Vma16p were intact after 10-min of incubation with the proteases in the absence of detergent, indicating that they were not accessible to the proteases. The epitopes were only digested when the membranes were solubilized before the addition of the proteases. These results are consistent with the C termini of Vma3p, Vma11p, and Vma16p oriented toward the lumen of the vacuole.
The N Termini of Vma3p and Vma11p Are Lumenal-Hydropathy plots predict that Vma3p and Vma11p contain four transmembrane segments. Therefore, if this model is correct and the C termini are in the lumen of the vacuole, then the N termini should also be present within the lumen. To test this hypothesis, N-terminal epitope-tagged Vma3p and Vma11p were constructed. Yeast deletion strains of vma3⌬ or vma11⌬ expressing the corresponding plasmid-borne N-terminal tagged protein grew on pH 7.5-buffered media containing 100 mM CaCl 2 (data not shown), a standard test for V-ATPase function (3).
Vacuoles were isolated from cells expressing N-terminaltagged Vma3p or Vma11p and also subjected to protease digestion followed by immunoblot analysis. As with the C-terminal epitopes, Fig. 1b shows that the N-terminal HA epitope of both Vma3p and Vma11p were protected from proteolysis in the absence of detergent and degraded only with detergent present. These results support the hypothesis that there are four transmembrane regions and the N termini, and C termini of both Vma3p and Vma11p are orientated toward the lumen of the vacuole.
The N Terminus of Vma16p Is Cytosolic-Unlike Vma3p or Vma11p, Vma16p is predicted to form five putative transmembrane regions. Nishi et al. (33,38) suggest that the N termini and C termini of Vma16p are both cytosolic, implying that there would be only four transmembrane regions. However, other studies suggest that Vma16p has five transmembrane  regions (34,39). Our results indicate that the C terminus of Vma16p is oriented toward the lumen (Fig. 1a). If Vma16p forms five transmembrane regions, then one would predict the N terminus to be oriented toward the cytosol, whereas if only four transmembrane regions are present, then the N terminus of Vma16p should be located in the lumen of the vacuole. To further test these models a doubly epitope-tagged protein was constructed, subjected to protease treatment, and analyzed by immunoblot analysis. The c-Myc epitope was placed on the N terminus of the previously C-terminal HA-tagged Vma16p. Cells expressing a copy of the doubly epitope-tagged Vma16p were able to confer wild-type growth on pH 7.5 YEPD containing 100 mM CaCl 2 (Fig. 2c).
Vacuoles were isolated from cells expressing the doubly tagged Vma16p and treated with proteases in the presence or absence of detergent. The Vph1p and CPY controls indicated that the proteases were active, and the vacuoles were intact (Fig. 2a). Samples probed with an antibody recognizing the c-Myc tag illustrate that the Myc epitope is labile both in the absence and presence of detergent, indicating that Vma16p contains five transmembrane regions since the N terminus of Vma16p is on the cytosolic face of the vacuole (Fig. 2a). The C-terminal HA tag was protected from protease digestion un-less the membranes were solubilized with detergent as previously observed (Fig. 2a). The same results were observed for the N terminus and the C terminus of Vma16p if the epitope tags were transposed (N-terminal HA and C-terminal c-Myc; data not shown). These results support the model that Vma16p does have five transmembrane helices, since the C terminus of Vma16p is in the lumen of the vacuole, whereas the N terminus is cytosolic.
Previous reports disagree about whether the first transmembrane region is needed for function, with one study reporting a complete loss of function by deletion of amino acids 12-55 of Vma16p (34) and the other study reporting that the deletion of amino acids 12-41 of Vma16p only compromised V-ATPase activity by ϳ30% (33). Because the N terminus of Vma16p is oriented toward the cytosol, one would predict that a deletion of the first transmembrane region would now reposition the N terminus toward the vacuole lumen. To test this, Vma16p lacking only the hydrophobic residues of the first transmembrane region, Vma16p⌬12-38, was constructed from the fulllength doubly tagged Vma16p (Fig. 2b). The protein encoded by vma16⌬12-38 was able to fully complement a vma16⌬ strain (Fig. 2c). Vacuoles were isolated from vma16⌬ cells expressing doubly tagged Vma16p⌬12-38 and subjected to protease treatment to test for the orientation of the C terminus and N terminus of Vma16p lacking the first transmembrane region (Fig. 2b). As was found for the full-lengthVma16p, the C-terminal HA epitope tag of Vma16p⌬12-38 was protected from proteolysis in the absence of detergent (Fig. 2b). However, unlike the full-length protein, the c-Myc epitope tag on the N terminus of Vma16p⌬12-38 was also protected from proteolysis in the absence of detergent, indicating that it now resides within the lumen of the vacuole. These results provide compelling evidence that Vma16p has five transmembrane segments, with the N terminus facing the cytosol and the C terminus in the lumen.
The Fusion Protein Vma16-Vma3p Is Functional, Whereas Vma3-Vma16p Is Non-functional-Studies with the F-Type ATP synthase show that it is possible to construct fusions of up to four c subunits and still have a functional enzyme complex (40). The proposed topological models of the V-ATPase subunits, Vma3p and Vma16p, allow us to make predictions about the function of a fusion construct between Vma16p and Vma3p. Because Vma16p has been found to have a lumenal C terminus and Vma3p has a lumenal N terminus, we predict that a Vma16-Vma3p fusion construct should be functional (Fig. 3). However, we also predict that a Vma3-Vma16p fusion construct would not be functional in vma16⌬ yeast because it would flip the orientation of Vma16p and force the N terminus to be lumenal and the C terminus to be cytosolic.
To test this hypothesis two fusion constructs were made; that is, a Vma3-Vma16p dimer linked by a triple HA epitope tag and a Vma16-Vma3p dimer also linked by a triple HA epitope tag. The vma16⌬ cells expressing the Vma3-Vma16p dimer were unable to grow on pH 7.5 YEPD with 100 mM CaCl 2 , whereas strains expressing the Vma16-Vma3p dimer grew normally on pH 7.5 YEPD with 100 mM CaCl 2 (Fig. 4a). These results suggest that only the Vma16-Vma3p dimer was able to form a functional V-ATPase complex. The dimers were also tested in a vma3⌬ strain, and neither gene fusion complemented the Vma Ϫ phenotype of vma3⌬ cells(data not shown). Previous studies revealed that there is only one copy of Vma16p (16) and four to five copies of Vma3p per complex (5). Neither dimer would provide the correct proteolipid stoichiometry in vma3⌬ cells, because to have multiple copies of Vma3p the complex would also have to contain multiple copies of Vma16p. Therefore, expression of either hybrid protein in vma3⌬ cells would be expected to produce a non-functional V-ATPase, as we observed. However, in the vma16⌬ cells Vma3p would also be expressed from the genomic copy of VMA3. Therefore, the V-ATPase could assemble normally with one copy of Vma16p, provided by the Vma16-Vma3p dimer with the other copies of Vma3p provided by the genomic VMA3 gene, to allow for a proteolipid ring of normal stoichiometry.
Because yeast cells possessing as little as 20% of wild-type V-ATPase activity appear to grow as wild-type cells in pH 7.5 YEPD plus 100 mM CaCl 2 (41), a more sensitive method was used to access vacuolar acidification. A vital vacuolar-staining dye, quinacrine, was used to test cells expressing Vma16-Vma3p or Vma3-Vma16p. Quinacrine is a weakly basic fluorescent dye that accumulates in low pH compartments within the cell, such as the yeast vacuole (36). Cells with a functional V-ATPase, when observed under a microscope, will have fluorescent vacuoles, whereas cells without a functional V-ATPase present on the vacuole will have no fluorescence within the vacuole. Cells expressing the Vma16-Vma3p dimer contained fluorescent vacuoles, displayed as green disks within the yeast cell, similar to cells expressing wild-type Vma16p, whereas the Vma3-Vm16p dimer resembled the vma16⌬ (Fig. 4b). Cell surface staining, outlined in red, was accomplished by the addition of concanavalin A TRITC. The results from the growth on pH 7.5 YEPD plus 100 mM CaCl 2 and quinacrine staining indicate that expression of the Vma16-Vma3p in a vma16⌬ background results in a functional V-ATPase complex, whereas expression of the Vma3/Vma16p dimer does not yield a functional complex.
The expression of both fusion constructs was tested by immunoblot analysis from whole cell lysates using a HA antibody. Both dimers, Fig. 5a, had apparent molecular masses of 37 kDa, whereas the monomer Vma16p had a molecular mass of 23 kDa. The results from the whole cell lysates demonstrate that both dimers are expressed and apparently stable.
Previous studies have found that cells lacking one of the V 0 subunits results in rapid degradation of Vph1p and lower steady-state levels of this subunit (42); however, V 1 steadystate levels are unaffected by the deletion of a V 0 subunit (13). Antibodies directed against a V 1 subunit, Vma2p, and a V 0 subunit, Vph1p, were used to measure the levels of these proteins in cells expressing the hybrid proteins. As expected the levels of Vma2p (V 1 subunit) were normal in cells expressing either hybrid proteins (Fig. 5a). The immunoblot for Vph1p (Fig. 5a) reveals that Vph1p levels from vma16⌬ cells expressing Vma16-Vma3p more closely resemble the levels of Vph1p from the wild-type strain, consistent with Vma16-Vma3p assembling into a functional V-ATPase complex. In contrast, lysates from cells expressing Vma3-Vma16p exhibited dramatically reduced levels of Vph1p, suggesting that Vma3-Vma16p is unable to assemble into the V 0 subcomplex (Fig. 5a).
Previous studies have found that only fully assembled V 0 subcomplexes are able to exit the endoplasmic reticulum and be trafficked to the vacuole (21,42,43). To ascertain if the proteolipid dimers are transported to and present on the vacuole, vacuoles were isolated from cells expressing Vma16p, Vma16-Vma3p, or Vma3-Vma16p. Samples were probed with antibodies against the HA epitope to determine whether either dimer was present in enriched vacuole membranes. The lower panel of Fig. 5b reveals that Vma16p and Vma16-Vma3p, but not the Vma3-Vma16p dimer, were present in vacuole membranes. Although the Vma3-Vma16p dimer was expressed and rather stable (Fig. 5a), this dimer failed to assemble into a V 0 subcomplex and remained trapped in the endoplasmic reticulum (data not shown). These results are consistent with the hypothesis that the V 0 complex assembles normally with the Vma16-Vma3p dimer but not with the Vma3-Vma16p.
To further test for the presence of V 0 subunits on the vacuole, samples were also probed for Vph1p. The results were the same as those obtained from the HA blot (Fig. 5b); V 0 subunits only present on the vacuole membrane were from cells that ex- pressed either Vma16p or the Vma16-Vma3p dimer but not from cells with the Vma3-Vma16p dimer. To ascertain if the V 1 subcomplex was associated with the vacuoles, samples were also probed with antibodies specific for the V 1 subunit, Vma2p. The results show that Vma2p was only present in the lanes that correspond to vacuole membranes from cells with either Vma16p or the Vma16-Vma3p dimer. Together, these results indicate that the Vma16-Vma3p dimer, but not the Vma3-Vma16p dimer, is able to assemble into the V 0 subcomplex, which is then trafficked to the vacuole.
To verify that the Vma16-Vma3p dimer was present in the fully assembled complex, native immunoprecipitations were performed on isolated vacuoles with antibodies recognizing the HA tag. The samples were probed with antibodies against HA to determine whether full-length dimer was immunoprecipitated. The results show that the full-length Vma16-Vma3p dimer (37 kDa) was immunoprecipitated from vacuoles isolated from strains expressing Vma16-Vma3p (Fig. 5c). The immunoprecipitated samples were also probed for the presence of Vph1p and Vma2p. The results, shown in the upper two panels of Fig. 5c, reveal that both Vph1p and Vma2p were co-immunoprecipitated with the HA tag, present on Vma16p or the Vma16-Vma3p dimer. These results provide additional evidence that the Vma16-Vma3p dimer fusion is able to assemble into a functional V-ATPase complex that localized to the vacuole.
ATPase assays were also conducted on the isolated vacuole membranes from cells expressing one or the other proteolipid dimers to compare with wild-type vacuole membranes. Vacuole membranes from cells expressing the Vma16-Vma3p dimer had an activity of 0.21 Ϯ 0.02 mol of ATP/min/mg, whereas vacuole membranes from cells expressing Vma16p had an activity of 0.33 Ϯ 0.04 mol of ATP/min/mg. As expected, vacuole membranes from cells expressing the Vma3-Vma16p dimer failed to show any V-ATPase activity.
Together these results support the topologies outlined for Vma3p, Vma11p, and Vma16p in Fig. 3. The ability of the Vma16-Vma3p dimer to assemble and form a fully functional V-ATPase complex further supports the conclusion that Vma16p contains five transmembrane regions with its N terminus toward the cytosol and the C terminus orientated toward the lumen of the vacuole. DISCUSSION The global structure of the V-ATPase is modeled after the F-Type ATP synthase (3) with the V-ATPase divided into two subcomplexes, a hydrophilic V 1 and a V 0 subcomplex made up of four unique integral membrane proteins and one peripheral subunit. In the F-type ATP synthase, the proton-translocating proteolipid ring consists of 10 -12 copies of the c subunit (40,44), whereas the V-ATPase contains three unique c-like subunits, subunit c (Vma3p), cЈ (Vma11p), and cЉ (Vma16p). Powell et al. (16) found that there is one copy of the Vma11p and Vma16p and multiple copies of Vma3p per V-ATPase complex. Molecular modeling predicts that Vma3p and Vma11p form four transmembrane helices, and Vma16p is predicted to possess five transmembrane helices. Intermolecular and intramolecular cross-linking studies have addressed the packing arrangement of the four transmembrane helices of Vma3p, and the first helix appears to lie at the center of the c subunit ring (45)(46)(47)(48). Experiments using chemical cross-linkers have also established that the fourth helix of Vma11p is orientated to the periphery of the c subunit ring and interacts with Vph1p (49). The topology of Vma3p and Vma11p has been assumed to mimic the topology of the c subunit of the F-type ATP synthase, with both the C termini and N termini oriented away from the F 1 , yet little data exist to support this model.
Our results demonstrate that Vma3p and Vma11p each contain four transmembrane spanning regions with the N and C termini oriented toward the lumen of the vacuole. The findings are consistent with models in which Vma3p and Vma11p arose by gene duplication events in the evolutionary past of the FIG. 5. Detection of Vma16p-Vma3p and Vma3p-Vma16p dimers in whole cell lysates, isolated vacuoles, and native immunoprecipitations. a, whole cell extracts were isolated from strain LGY9(vma16⌬) containing either pLG84(VMA16), pLG146(VMA16-VMA3), pLG145(VMA3-VMA16), or pRS316(empty vector). 15 g of total protein were subjected to SDS-PAGE, transferred to nitrocellulose, and probed with anti-Vph1p (10D7), anti-Vma2p, or anti-HA antibodies. The apparent molecular mass of the Vma16p was 23 kDa, whereas the apparent molecular mass of both dimers was 37 kDa (indicated by the two arrows). b, vacuoles were isolated from the strains in a as described under "Experimental Procedures." 10 g of protein from the isolated vacuoles was loaded per lane of SDS-PAGE, transferred to nitrocellulose, and probed with anti-Vph1p (10D7), anti-Vma2p, or anti-HA antibodies. c, the HA-tagged Vma16p or dimer proteins were immunoprecipitated (IP) from solubilized vacuoles isolated in b using a polyclonal anti-HA antibody as described under "Experimental Procedures." Vma16p and Vma16-Vma3p were able to co-immunoprecipitate both Vph1p and Vma2p. The specific V-ATPase activity measured from vacuolar membranes isolated from cells expressing Vma16p was 0.33 Ϯ 0.04 mol of ATP/min/mg and from vacuolar membranes isolated from cells expressing the Vma16-Vma3p dimer was 0.21Ϯ.02 mol of ATP/min/mg. V-ATPase and that Vma3p and Vma11p can be modeled as two fused F-ATP synthase c subunits with their N and C termini orientated in the lumen of the vacuole (30 -32). Our results also provide strong evidence that the N terminus of Vma16p is on the cytoplasmic face of the vacuole, and the C terminus is in the lumen of the vacuole.
Before this study, there have been four studies that have addressed the topology of Vma16p, with each study coming to different conclusions. Nishi et al. (33,38) suggest that Vma16p contains only four transmembrane regions, with both the Nand C terminus on the cytoplasmic face of the vacuole. A global topological screen of 37 yeast proteins, which included Vma16p, contained data fitting the model with the C terminus in the lumen of the vacuole and the protein containing five transmembrane regions (39). Finally, Nelson and co-workers (50) used an HA antibody to inhibit V-ATPase activity of an isolated complex containing a N-terminally HA-tagged version of Vma16p, implying that the N terminus is on the cytoplasmic face of the vacuole. In light of these discrepancies, we sought an alternate way to further test our topology model.
Both models of the Vma16p topology agree that the N terminus is on the cytoplasmic face, and it is only the location of the C terminus that is in contention. We generated and tested two fusion proteins to test the orientation of Vma16p. By our model the Vma16-Vma3p chimeric protein should contain a properly oriented and functional Vma16p, whereas the Vma3-Vma16p hybrid would be non-functional because it would result in a mis-oriented Vma16p (Fig. 3). Because only the Vma16p-Vma3p was functional, the data best fit a model in which the C terminus of Vma16p is in lumen of the vacuole. Not only was the V-ATPase complex that contained the Vma16-Vma3p hybrid functional, but this V-ATPase had close to wildtype V-ATPase activity.
Even though the topologies of the three yeast V-ATPase proteolipids Vma3p, Vma11p, and Vma16p have now been established, a few fundamental questions still remain. Why does Vma16p have an extra transmembrane helix? Vma16p⌬12-38 and Vma16p⌬2-41, which lack the first transmembrane region, form functional and highly active V-ATPase complexes (33). Alignments of all the known Vma16p (cЉ)-like molecules do show an interesting trend. All known fungal and animal Vma16p molecules have the predicted five transmembrane regions, but plant cЉ proteins are predicted to only contain four transmembrane regions. The first transmembrane region in animal Vma16p polypeptides could merely be an evolutionary relic and, thus, have no significant function in the complex, but further testing would be needed to answer this question. As to placement of the first transmembrane region, future studies will need to be conducted to position it possibly in the interior of the c subunit ring or on the periphery of the ring.
The exact stoichiometry of Vma3p and arrangement of Vma3p, Vma11p, and Vma16p relative to one another within the c subunit ring still remain to be determined. The results from studies with the proteolipid dimers provide a compelling approach to answer these questions. If other proteolipid dimers prove to be functional, then it might be possible to construct a set of fusion proteins that would elucidate the V 0 proteolipid stoichiometry and arrangement.