Novel vacuolar H+-ATPase complexes resulting from overproduction of Vma5p and Vma13p.

The vacuolar H(+)-ATPase (V-ATPase) is a multisubunit complex composed of two sectors: V(1), a peripheral membrane sector responsible for ATP hydrolysis, and V(0), an integral membrane sector that forms a proton pore. Vma5p and Vma13p are V(1) sector subunits that have been implicated in the structural and functional coupling of the V-ATPase. Cells overexpressing Vma5p and Vma13p demonstrate a classic Vma(-) growth phenotype. Closer biochemical examination of Vma13p-overproducing strains revealed a functionally uncoupled V-ATPase in vacuolar vesicles. The ATP hydrolysis rate was 72% of the wild-type rate; but there was no proton translocation, and two V(1) subunits (Vma4p and Vma8p) were present at lower levels. Vma5p overproduction moderately affected both V-ATPase activity and proton translocation without affecting enzyme assembly. High level overexpression of Vma5p and Vma13p was lethal even in wild-type cells. In the absence of an intact V(0) sector, overproduction of Vma5p and Vma13p had a more detrimental effect on growth than their deletion. Overproduced Vma5p associated with cytosolic V(1) complexes; this association may cause the lethality.

The vacuolar H ϩ -ATPase (V-ATPase) 1 is an electrogenic proton pump found throughout the endomembrane system (1,2). The V-ATPase harnesses the energy derived from ATP hydrolysis to pump protons across membranes, creating an electrochemical gradient. V-ATPases share both sequence and structural similarities with F-type ATP synthases, and preliminary experiments indicate a fundamental similarity in their mechanisms (3). Few specific details about how V-ATPases couple ATP hydrolysis and proton transport are available, however, because the sequence similarities with F-ATPases lie primarily in the ATP-binding and proton pore subunits, not in the "stalk" subunits believed to be predominantly responsible for structural and functional coupling (4).
The Saccharomyces cerevisiae V-ATPase has been extensively studied and has proven to be an excellent model for all eukaryotic V-ATPases. To date, 13 subunits have been identified and assigned to either the peripheral cytoplasmic V 1 sector or the membrane-associated V 0 sector; however, a limited number of subunits have assigned function. The V 1 sector subunit Vma1p contains the catalytic nucleotide-binding sites, whereas Vma2p possesses regulatory nucleotide-binding sites (2,3). The V 0 sector proteolipid subunits Vma3p, Vma11p, and Vma16p are thought to form the proton pore, probably in combination with the large integral membrane subunit Vph1p (5)(6)(7). The functions of the remaining subunits are poorly understood.
Vma1p and Vma2p are present in three copies each per complex and are believed to reside within the bulky head domain seen in low resolution electron microscopic studies of the V-ATPase (8). The head domain appears to be attached to the membrane via two stalk regions. The composition of the stalk regions is unknown, but it is hypothesized that they may contain most of the subunits of undefined function and that these subunits could be involved in coupling the V-ATPase both structurally and functionally.
The V-ATPase is a tightly coupled enzyme that only exhibits activity when the enzyme is fully assembled with all subunits at the membrane. When cells are deprived of glucose, V 1 sectors disassemble from the membrane (9,10). Disassembly results in the silencing of both sectors: V 1 is inactive as an Mg-ATPase, and the V 0 proton pore is closed (11)(12)(13)(14). The free V 1 and V 0 sectors can be rapidly reassembled to form functional V-ATPase complexes upon glucose re-addition. This is thought to be a form of regulation of the V-ATPase that may assure that cytosolic ATP stores are conserved in times of starvation (15,16).
Results from a variety of investigations have led to the hypothesis that the C and H subunits of the V 1 sector (Vma5p and Vma13p, respectively) are instrumental in structural and functional coupling of the V-ATPase. When VMA5 is deleted from the yeast genome, both a fully assembled V 0 subcomplex and a core V 1 subcomplex are formed, whereas in most other V 1 subunit deletion strains, the V 1 core complex is fully or partially disassembled (17,18). Glucose deprivation studies have shown that Vma5p is lost from both sectors during disassembly, suggesting that Vma5p may be directly involved in signaling V 1 release from the membrane (9,14). When VMA13 is deleted, the V-ATPase does assemble, but is inactive, indicating that Vma13p is an activator or a structural stabilizer (19). Vma13p is found in cytosolic V 1 complexes formed in the absence of an intact V 0 sector or released from the membrane by glucose deprivation (14). V 1 sectors from a vma13⌬ strain contain all of the other V 1 subunits, but unlike cytosolic V 1 complexes from wild-type cells, are active as Mg-ATPases (14). This suggests a more complex role for Vma13p as a regulator of the V-ATPase, in which it silences cytosolic V 1 , but activates assembled V 1 V 0 complexes. If Vma5p and Vma13p are involved in tightly coupling and regulating the V-ATPase, the levels of subunit within a cell could be important in V-ATPase function and regulation. In this study, we have addressed the effect of overexpression of Vma5p and Vma13p on V-ATPase function.

EXPERIMENTAL PROCEDURES
Materials and Strains-Molecular biology reagents were purchased from New England Biolabs Inc. LA-Taq was obtained from Panvera. Zymolyase 100T and Tran 35 S-label were from ICN. Dithiobis(succinimidyl propionate) was purchased from Pierce, and concanamycin A was purchased from Wako Bioproducts. N-Octyl ␤-D-glucopyranoside was purchased from Calbiochem. Precast 16% Tris/Tricine gels and a Mono-Q1 ion-exchange column were purchased from Bio-Rad. Monoclonal antibody 9E10 (anti-Myc epitope) was purchased from Santa Cruz Biotechnology. Alkaline phosphatase-conjugated goat anti-mouse and goat anti-rabbit antibodies were from Promega. Oligonucleotides were purchased from MWG Biotech. All other chemicals were purchased from Sigma.
Strains were transformed overnight as described by Elble (25), and transformants were selected on supplemented minimal medium lacking uracil (SDϪUra medium). All experiments were performed with fresh transformants to assure comparable levels of subunit overexpression. Over time, plasmid copy number in the transformants appeared to decrease, resulting in low expression levels, probably because of selection against cells overexpressing VMA5 and VMA13 in culture.
Biochemical Characterization-Quinacrine staining was performed as described by Roberts et al. (26). Cells were grown overnight in SDϪUra medium to mid-log phase. Cells were then harvested and incubated at room temperature for 5 min in YEPD medium (yeast extract, peptone, and 2% dextrose) containing 200 M quinacrine, followed by a wash and resuspension in 50 mM Na 2 HPO 4 (pH 7.6) containing 2% glucose. Cells were visualized within 10 min of resuspension using a Zeiss Axioskop fluorescence microscope. Images were captured with a Spot II camera and analyzed using Adobe Photoshop Version 5.0.
Isolation of vacuolar vesicles was performed as described by Roberts et al. (26) with the following modifications. Freshly transformed cells were grown overnight in SDϪUra medium to ensure that they maintained the 2 plasmids and then harvested at a density of 1.8 A 600 or less. For strains carrying the 2 plasmids, 2 liters of overnight cultures were combined. Cells were converted to spheroplasts and washed as described above, and then the spheroplasts were incubated at 30°C for 10 min in YEPD medium containing 1.2 M sorbitol to assure V-ATPase assembly. Spheroplasts were lysed in 50 ml of lysis buffer, followed by vacuolar flotation as described (26). Protein concentrations were determined by the method of Lowry et al. (27). Vacuolar vesicles were assayed for ATPase activity in the presence and absence of 100 nM concanamycin A (28) using the coupled enzyme assay described by Lotscher et al. (29). Proton pumping assays were performed on vacuolar vesicles (5 g of total vesicle protein) suspended in proton translocation buffer (25 mM Mes/Tris (pH 7.2), 5 mM MgCl 2 , 25 mM KCl, and 2 M quinacrine). Vesicles were placed in a Fluorolog-3-21 spectrofluorometer with stirring, and pumping was initiated by the addition of 2.5 mM MgATP. Quinacrine fluorescence (420 nm excitation and 490 nm emission) was monitored, and data were analyzed using Microsoft Excel.
Whole cell lysates were prepared as previously described (30). Ly-sates and vacuolar vesicles were suspended in cracking buffer (50 mM Tris-HCl (pH 6.8), 8 M urea, 5% SDS, and 5% ␤-mercaptoethanol) and subjected to SDS-PAGE, followed by immunoblotting. A titration was performed to determine a linear range of vacuolar vesicle load for each subunit antibody. Vph1p, Vma1p, Vma2p, and Vma5p, were detected using monoclonal antibodies 10D7, 8B1, 13D11, and 7A2, respectively (30). The Myc-tagged Vma13p subunit was detected using monoclonal antibody 9E10. Polyclonal antisera against the Vma4p and Vma8p V 1 subunits were a generous gift from Tom Stevens. Binding of the antibodies was detected using alkaline phosphatase-conjugated goat antimouse or goat anti-rabbit secondary antibodies, followed by colorimetric development.
Immunoprecipitations from isolated vacuolar vesicles were executed as described previously (12). Immunoprecipitations were analyzed by SDS-PAGE, followed by silver staining (31). Biosynthetic labeling and nondenaturing immunoprecipitations from whole spheroplasts were performed as described (9) and analyzed on a Molecular Dynamics Storm PhosphorImager.
Growth of cells carrying different plasmids was monitored by two different methods. To obtain doubling times of liquid cultures, fresh transformants were diluted into fresh SD-Ura medium and shaken at 30°C, and the density of the culture was monitored with time by measuring the absorbance at 600 nm. Alternatively, freshly transformed strains were grown overnight in SD-Ura medium to logarithmic phase, and then cells were diluted in water to 1.0 A 600 /ml. Serial dilutions were made in water, and equal volumes of cells were transferred to plates for growth at 30°C.
Cytosolic V 1 sectors were isolated from strains as described by Parra et al. (14) with the following modifications. The vma11⌬ strain was transformed with the 2-VMA5 plasmid and grown overnight in SD-Ura medium. 2 liters of cells at a density of 1.5 A 600 /ml were combined. 2-ml fractions were collected from the Mono-Q1 column and precipitated overnight on ice using 10% trichloroacetic acid. Precipitated fractions were analyzed for the presence of several V 1 subunits by immunoblotting as described above.

RESULTS
Overproduction of Vma5p and Vma13p-To examine the effects of their overexpression, VMA5 and VMA13 were each subcloned into a multicopy (2) plasmid and transformed into the corresponding vma5⌬ and vma13⌬ strains, which lack the endogenous copies of the genes. Whole cell lysates were prepared from cells containing the genes expressed from multicopy plasmids and from the same mutants carrying the genes on a low copy (CEN) plasmid. Vma13p has been tagged at the N terminus with the Myc epitope (14). The resulting N-Myc-VMA13 fully complements the growth phenotype of the deletion strain and yields a V-ATPase with properties comparable to those of the wild-type enzyme (see below). This epitopetagged version of VMA13 was used in all experiments. Western blot analysis of whole cell lysates prepared from the various strains showed that cells containing multicopy plasmid 2-VMA5 or 2 -VMA13 displayed a 2-3-fold steady-state level of overexpression of the two subunits ( Fig. 1).
To determine whether the overproduction of Vma5p or Vma13p had an effect on the V-ATPase, the cells were screened for a Vma Ϫ phenotype, characterized by sensitivity to elevated pH and Ca 2ϩ levels (32,33). The deletion strains carrying 2-VMA5 or 2-VMA13 exhibited a mutant growth phenotype, whereas those transformed with CEN-VMA5 or CEN-VMA13 did not (Fig. 2). This indicates that a 2-3-fold overproduction of Vma5p or Vma13p compromises V-ATPase activity.
How Is V-ATPase Activity Affected by Overproduced Vma5p and Vma13p?-The presence of a Vma Ϫ phenotype generally suggests that the cell possesses Ͻ25% of the wild-type V-ATPase activity (34). We first examined the ability of the V-ATPase to acidify the vacuoles in these strains in vivo. Using the lysosomotropic dye quinacrine, we were able to qualitatively examine the acidification of vacuoles in the presence of excess Vma5p or Vma13p. Cells were stained with quinacrine and then visualized using fluorescence microscopy and Nomarski optics (26). Wild-type cells exhibited bright quinacrine staining that colocalized with the vacuole (Fig. 3A). Under the same conditions, vma5⌬ cells and cells carrying 2-VMA13 displayed no quinacrine staining (Fig. 3, B and D), indicating that the vacuoles were not acidified. Cells overexpressing Vma5p did show some quinacrine staining localized to the vacuole, but the levels were generally less than in wild-type cells (Fig. 3C). These results suggest that, in the presence of excess Vma5p, it is still possible to establish a proton gradient across the membrane, but the gradient may not be as strong as in wild-type cells. Cells overexpressing Vma13p did not appear to establish a proton gradient sufficient for any vacuolar quinacrine accumulation. It was notable that we found many cases in which the bud was brightly stained compared with the mother cell in cells overexpressing Vma5p (Fig. 3C), but we currently have no explanation for this phenomenon.
To directly measure ATPase activity, vacuolar membranes were isolated from various strains, and concanamycin A-sensitive ATP hydrolysis was measured as described under "Experimental Procedures." Overproduction of Vma5p or Vma13p moderately reduced the ATP hydrolysis activity of the V-ATPase, but not as much as expected from the growth phenotype. vma5⌬ cells carrying VMA5 on a 2 plasmid had 77% as much ATPase activity as the same cells carrying VMA5 on a CEN plasmid (Table I), even though their growth properties ( Fig. 2) were dramatically different. Surprisingly, the V-ATPase retained up to 72% of its ATP hydrolysis activity in isolated vacuoles from the strain containing overproduced Vma13p, even though there was no quinacrine uptake in vivo (Fig. 3). A closer look at proton translocation by the V-ATPase was needed.
To address whether the proton translocation is specifically affected, vacuolar vesicles were assayed for quinacrine quenching. Isolated vesicles were suspended in transport buffer containing quinacrine, and quenching was monitored after the addition of saturating MgATP (6). Fig. 4 compares the pumping activities of wild-type and vma13⌬-, 2-VMA5-, and 2-VMA13-containing strains in a single experiment. Vacuolar vesicles isolated from the wild-type strain exhibited an MgATP-activated decrease in quinacrine fluorescence. Vacuolar vesicles from vma13⌬ and Vma13p-overproducing strains showed very little MgATP-activated quinacrine quenching. Vesicles from cells overproducing Vma5p exhibited an intermediate level of MgATP-activated quenching. To confirm that the decrease in quinacrine fluorescence was directly related to the V-ATPase, we tested the concanamycin A sensitivity of pumping. Concanamycin A (100 nM) completely abolished the proton pumping of both the wild-type and 2-VMA5 vacuolar vesicles (data not shown). To obtain a more quantitative assessment of proton pumping, we compared the final extent of quenching obtained for two separate vacuole purifications from these strains. The final extent of quinacrine quenching was 54 Ϯ 5.1% of the wild-type level in vesicles from cells overexpressing VMA5, 3.5 Ϯ 0.7% of the wild-type level in vesicles from vma13⌬ cells, and 5.9 Ϯ 4.6% of the wild-type level in vesicles from cells overexpressing VMA13. These results mirror the qualitative analysis of vacuolar acidification in Fig. 3. Proton pumping in cells overexpressing Vma5p was reduced, but was

TABLE I V-ATPase activities in isolated vacuolar membranes
Vacuolar membranes were isolated, and ATPase activity was assayed as described under "Experimental Procedures." Activities shown are sensitive to 100 nM concanamycin A and represent means Ϯ S.E. of n separate vacuole isolations.

Strain
Specific activity enough to create a smaller proton gradient across the vacuolar membrane. It is somewhat surprising that this level of V-ATPase activity and proton pumping was insufficient to support growth of the cells at elevated pH and Ca 2ϩ levels. The V-ATPase from cells containing overproduced Vma13p was capable of ATP hydrolysis, but was almost completely incapable of establishing a proton gradient. To our knowledge, this is the first time that such extensive functional uncoupling of a V-ATPase has been observed.
Overproduction of Vma5p and Vma13p Has a Structural Effect on the V-ATPase-The loss of the proton gradient in Vma13p-overproducing cells could arise either from a defect in the V-ATPase itself or from binding of Vma13p to another vacuolar protein that results in leakiness of the vacuolar membrane. Therefore, we determined whether the V-ATPase complex is structurally intact in cells overexpressing VMA5 and VMA13. The V-ATPase in wild-type vacuolar vesicles is stable to low salt washes, and several EDTA washes are incorporated into purification of the V-ATPase (36). In contrast, vacuolar vesicles isolated from cells containing 2-VMA5 or 2-VMA13 lost a substantial amount of ATPase activity after two washes in 1 mM EDTA. On average, specific ATP hydrolysis activity decreased by 47% in both 2-VMA5-and 2-VMA13-containing strains, whereas wild-type vesicles gained ϳ4% specific activity after washing. This indicates that the V-ATPase in cells containing overproduced Vma5p and Vma13p is structurally altered.
To determine whether the subunit composition of the V-ATPase is affected in the 2-VMA5-and 2-VMA13-containing strains, we took several approaches. First, the levels of various V 1 and V 0 subunits present at the vacuolar membrane were analyzed by Western blot analysis using antibodies against the Vph1p V 0 subunit and the Vma1p, Vma2p, Vma5p, Vma8p, and Vma4p V 1 subunits of the V-ATPase. A set of titrations of vacuolar vesicles from vma13⌬ cells transformed with the CEN-VMA13 vector was performed to allow quantitative comparison of Western blot signals (Fig. 5A). Identical titrations were observed for vacuolar vesicles from wild-type cells (data not shown) and vma5⌬ cells carrying a CEN-VMA5 plasmid. Vacuoles from cells overproducing Vma13p contained only ϳ50% of the wild-type level of a predicted V 1 sector stalk subunit (Vma4p) in this experiment even though the other V 1 and V 0 subunits tested appeared to be at wild-type levels. We found that Vma4p was always at reduced levels in vacuolar vesicles from cells overproducing VMA13, and the levels could be as low as 25% of the wild-type level (data not shown). Vma8p was also reduced in some vesicle preparations from this strain, but the effect was more variable (see below). Levels of V 1 and V 0 subunits in vacuolar vesicles isolated from the Vma5poverproducing strain closely resembled those from the strains carrying the low copy plasmids. Significantly, isolated vacuolar  1, 2,  and 3, respectively. B, immunoprecipitation of V-ATPase complexes from isolated vacuolar membranes. Vacuolar vesicles were isolated as described for A and then solubilized with 2% N-octyl ␤-D-glucopyranoside. Solubilized vesicles were incubated on ice with monoclonal antibody 8B1 directed against Vma1p, which is capable of co-immunoprecipitating the entire V-ATPase, followed by protein A-Sepharose. Immunoprecipitated complexes were separated by SDS-PAGE on a 16.5% Tris/Tricine-acrylamide gel, followed by silver staining. A mock immunoprecipitation (containing antibody without solubilized vesicles) was conducted in parallel. vesicles from the strains overproducing Vma5p and Vma13p did not contain elevated levels of these subunits.
To determine the assembly state of the V-ATPase from the different strains by an independent method, we performed immunoprecipitations from a separate isolation of vacuolar vesicles. Using monoclonal antibodies capable of co-immunoprecipitating the entire V-ATPase complex, we confirmed that overexpression of Vma13p resulted in lower levels of the V 1 sector stalk subunit Vma4p and also observed lower levels of Vma8p (Fig. 5B). Co-immunoprecipitation confirmed that strains overproducing Vma5p had an apparently wild-type subunit composition.
Finally, we purified the V-ATPase from isolated vacuolar vesicles by glycerol gradient fractionation (36). We found that V-ATPase complexes purified from wild-type, 2-VMA5-containing, and 2-VMA13-containing strains all peaked at a similar glycerol density (data not shown). This result was anticipated because the V-ATPase from Vma5p-overproducing strains resembled the wild-type enzyme composition. The V-ATPase from Vma13p-overproducing strains had reduced levels of some smaller stalk subunits, but this may not generate a sufficient change in size or structure in the enzyme to have a major effect on its fractionation pattern.
Taken together, these data indicate that the V-ATPase from cells overproducing VMA13 is both functionally and structurally defective. It is notable, however, that the structural defects cannot be accounted for by the coordinate loss of the V 1 subunits from the membrane that is seen both in most V 1 subunit deletion mutants and in certain V 1 point mutants. Instead, certain stalk subunits appear to be specifically lost, and this could help to account for the functional uncoupling of ATP hydrolysis and proton transport. In contrast, the structural defects of the V-ATPase from Vma5p-overproducing cells appear to be much more subtle; the overall stability of the enzyme is reduced, but the subunit composition does not appear to be altered.
Growth Defects of Strains Overexpressing Vma5p and Vma13p-If overexpression of Vma5p and Vma13p simply destabilizes the V-ATPase, then the effects of overexpression of these subunits should never be worse than the effects of their deletion. In performing the experiments above, however, we noticed that the growth defects of overproducing strains could be quite pronounced. Quantitative analysis of doubling times for strains containing the 2-VMA5 and 2-VMA13 plasmids in glucose-containing medium indicated that their growth defects were comparable to those of vma deletion cells (see below), but the level of overexpression indicated in Fig. 1 (2-3-fold) was also quite modest. We therefore addressed whether high level expression of VMA5 and VMA13 is toxic to cells by an alternate method. Unfortunately, the most commonly used methods for inducible expression of yeast genes require either growth of cells on galactose, which can cause disassembly of the V-ATPase complex, or the addition of heavy metals, which can be toxic to cells with impaired V-ATPase activity. Therefore, we decided to overexpress VMA5 and VMA13 by cloning the genes into a system in which very high plasmid copy numbers can be induced by alterations in the growth medium. Both genes were subcloned into a leu2-d-based vector (37). This vector contains dual selectable markers, a defective LEU2 allele, and a wildtype URA3 gene. Transformants can be selected and maintained by growth on medium lacking uracil, but complementation of a leucine auxotrophy requires high levels of amplification of the leu2-d-containing plasmid. The leu2-d plasmid was used to assess effects of "inducible" high level expression of both Vma5p and Vma13p on the cell. Wild-type cells, which have a single chromosomal copy of VMA5 and VMA13, were transformed with an empty leu2-d vector or the same vector containing VMA5 or VMA13. When cells were grown on minimal medium lacking uracil, on which there is little overexpression of VMA5 and VMA13, all of the transformants were able to grow (Fig. 6A). When cells are plated onto minimal medium lacking leucine, the plasmid must be amplified at high levels to support growth, and the VMA5 and VMA13 genes would also be amplified. Cells bearing the leu2-d plasmid carrying VMA5 and VMA13 were not viable on medium lacking leucine, even at low pH (pH 5.7) (Fig. 6B). This suggests that highly overexpressed Vma5p and Vma13p compromise the growth of yeast cells beyond the Vma Ϫ growth defects characteristic of vma mutations and that overexpressed VMA5 and VMA13 may exhibit a "gain of function" that is deleterious to the cell.
Gain-of-function mutations in the V-ATPase have not been described previously, but there are at least two obvious ways that alterations in V-ATPase structure could generate novel and potentially damaging activities. The peripheral V 1 sector of the enzyme and the membrane-bound V 0 sector exhibit considerable structural independence. Free V 1 or V 0 sectors are synthesized and stable in strains that fail to assemble one of the sectors (18), and free V 1 and V 0 sectors exist in a dynamic equilibrium with fully assembled V-ATPase complexes in wildtype cells (16,38). One important characteristic of the free V 1 and V 0 sectors is that their activities appear to be silenced; free V 1 sectors are not active as MgATPases, and free V 0 sectors do not appear to be open proton pores. If the MgATPase activity of free V 1 sectors were activated or the V 0 proton pore were opened in the presence of overproduced Vma5p and Vma13p, then a new and potentially deleterious function of the V-ATPase subcomplex might be revealed. To test this possibility, we examined whether overexpression of Vma5p and Vma13p would generate an additional growth defect in vma mutants that lack either an intact V 1 or V 0 sector, but assemble the other sector. These mutants have already lost all V-ATPase activity and thus exhibit a full Vma Ϫ phenotype. Therefore, any additional growth defect would suggest an additional defect arising from Vma5p or Vma13p overproduction.
Yeast vma2⌬ mutants lack the B subunit of the V 1 sector, but assemble intact V 0 sectors at the vacuole (12). vma11⌬ mutants FIG. 6. High level overexpression of Vma5p and Vma13p is lethal to yeast cells. VMA5 and VMA13 were subcloned into the leu2-d-based vector pCKR201-1. Wild-type strain W303a was transformed with the empty vector (empty) or with constructs carrying the VMA5 (ϩVMA5) or VMA13 (ϩVMA13) subunit gene. Transformants were grown overnight in SD-Ura medium, conditions that maintain the plasmid at low copy number, and then serially diluted in water to the indicated densities (expressed as A 600 /ml) and plated onto SD-Ura (A) or SD-Leu (B) medium. Growth on the SD-Leu plates requires amplification of plasmid copy number (37). Both plates were incubated at 30°C for 72 h. SD-Ura and SD-Leu plates have a pH of ϳ5.7.
lack the cЈ subunit, one of the three essential proteolipids of the V 0 sector, but assemble intact cytoplasmic V 1 sectors (7,17). Table II describes the doubling times of a variety of mutant and wild-type strains containing either the empty vector or the 2 vectors containing VMA5 and VMA13 described above. The wild-type strain exhibited little difference in its doubling time in minimal medium (pH 5.7) whether it contained the empty plasmid or the VMA5 and VMA13 multicopy plasmids. Similarly, the vma2⌬ strain grew more slowly than the wild-type strain under all conditions, as reported previously for vma mutant strains (21), but there was no major change in doubling time in the presence of the 2 plasmids. In contrast, growth of the vma11⌬ strain was not significantly affected by the empty 2 plasmid, but was 39 -48% slower in the presence of either the VMA5-or VMA13-containing plasmid. This suggests once again that the overproduced subunits are capable of affecting cell growth by a mechanism beyond disruption of V-ATPase structure, but these data also indicate that this new function can be exhibited in the presence of free V 1 (but not V 0 ) sectors alone.
We wanted to further characterize the V 1 sectors within vma11⌬ cells in the presence of overproduced Vma5p and Vma13p. Free V 1 sectors in vma11⌬ cells normally did not contain Vma5p (Fig. 7, second lane). However, Vma5p could be co-immunoprecipitated with cytosolic V 1 sectors in vma11⌬ cells carrying overproduced Vma5p (Fig. 7, third  lane). This association could potentially affect the function or silencing of the free V 1 complexes. Overproduction of Vma13p did not cause Vma5p to associate with free V 1 sectors in the vma11⌬ strain (Fig. 7, fourth lane); these complexes appear to have a wild-type subunit composition. However, it is notable that the free V 1 sectors contained the same levels of Vma4p and Vma8p because these subunits were present in lower levels in assembled V-ATPase complexes from cells overexpressing VMA13 (Fig. 5B).
To further characterize the V 1 sectors in cells overexpressing Vma5p, we purified cytosolic V 1 complexes from vma11⌬ cells overexpressing Vma5p. We had previously found that Vma5p did not copurify with assembled V 1 complexes; Vma5p did not bind to a Mono Q1 ion-exchange column and was detected only in the first few flow-through fractions (14). Surprisingly, the ion-exchange profile of vma11⌬ transformed with 2-VMA5 showed that Vma5p was present in fractions throughout the purification (Fig. 8). Vma5p copurified with the V 1 subunits Vma1p, Vma2p, and Vma4p. We determined by nondenaturing immunoprecipitations that Vma5p was assembled with these V 1 complexes (data not shown). We assayed MgATPase activity in all of the fractions containing the assembled V 1 complexes, including fractions 20 -22, which were shown to contain fully assembled V 1 complexes in the previous studies, but did not observe any MgATP hydrolysis (data not shown).

DISCUSSION
Vma5p and Vma13p have been implicated in the structural and functional coupling of the V-ATPase. Based on this, we had hypothesized that overexpression of one or both of these two subunits might inactivate the V-ATPase by dissociating the fully assembled complex into free V 1 and V 0 sectors. Although we did observe a Vma Ϫ phenotype characteristic of inactivation of the V-ATPase, the structural basis of this phenotype is less straightforward. This study does provide new insights into the functional roles of these two subunits, however. Overproduction of Vma13p results in an assembled V 1 V 0 complex that is capable of ATP hydrolysis, but not proton transport. This is the first reported example of a fully uncoupled V-ATPase complex.
Overexpression of Vma5p appears to destabilize fully assembled V-ATPase complexes by an unknown mechanism, but more significantly, drives association of Vma5p with cytosolic  vma2⌬ (W303 vma2⌬), and vma11⌬ (YPH500 vma11⌬) cells were transformed with the empty 2 vector YEp352 or with the same vector containing VMA5 or VMA13. Log-phase cultures of all strains were diluted into fresh SD ϪUra medium; growth of the strains was monitored by periodically measuring the absorbance of the culture at 600 nm; and the doubling times during log-phase growth of the cultures were determined from growth curves.

Strain
Doubling time Overproduced Vma5p is associated with cytosolic V 1 . Wild-type cells (wt; SF838-5A␣), vma11⌬ cells, vma11⌬ cells transformed with 2-VMA5, or vma11⌬ cells transformed with 2-VMA13 were deprived of methionine during overnight growth. Cells were harvested and converted to spheroplasts. Spheroplasts were incubated with Tran 35 S-label for 60 min and then lysed. Proteins were solubilized under nondenaturing conditions in the presence of the cross-linker dithiobis(succinimidyl) propionate (9). Fully or partially assembled V-ATPase complexes were immunoprecipitated with monoclonal antibody 8B1, which recognizes Vma1p, the catalytic subunit of the V 1 sector. Immunoprecipitated complexes were subjected to SDS-PAGE and then analyzed using a PhosphorImager. The positions of known V-ATPase subunits are shown on the right; subunits belonging to the V 0 sector are italicized. Vma5p shows variable association with the immunoprecipitated complexes and is indicated by *. The positions of molecular mass standards are shown on the left (arrows) and correspond to the following molecular masses (from top to bottom): 200, 96, 69, 43, 29, and 18.4 kDa. V 1 complexes. We hypothesize that this association may be responsible, at least in part, for the fact that overexpressed Vma5p has even more damaging effects on cells than does a total loss of V-ATPase activity. It is significant that these structural and functional phenotypes of overproduction have not been seen for deletion or point mutations in any yeast V-ATPase subunit and thus are completely novel. We have not determined whether the effects of Vma5p or Vma13p overexpression arise during biosynthesis of the V-ATPase, through some type of exchange or sequestering of subunits in the cytosol or by another mechanism; but we describe the implications of these phenotypes below.
Overexpression of Vma13p Functionally Uncouples ATP Hydrolysis and Proton Translocation-Vma13p is a member of the peripheral V 1 sector, but one of its distinctive characteristics is that it does not appear to be essential for association of V 1 subunits with V 0 sectors in isolated vacuoles (19). Vacuoles from vma13⌬ cells lack all V-ATPase activity, but do assemble other V 1 subunits at the membrane; deletion of other V 1 subunits appears to prevent association of the remaining peripheral subunits with the membrane (19). Cytosolic V 1 sectors isolated from vma13⌬ cells also contain all of the other V 1 subunits, but do exhibit some MgATPase activity, suggesting they have lost some of the "silencing" of MgATPase activity that usually accompanies release of V 1 from the membrane (14). Based on this behavior of vma13⌬ mutants, Vma13p is believed to act as an activator of V 1 V 0 complexes and as an inhibitor of free V 1 complexes. It has been hypothesized that it achieves this dual role by stabilizing a "coupled" conformation of V 1 V 0 that is competent for ATP-driven proton pumping and then undergoing a conformational change when V 1 is released from the membrane that prevents catalysis by free V 1 (14,39).
Interpretation of the data from Vma13p overproduction in this context would suggest that, in the presence of extra copies of Vma13p, an intermediate conformation is present in which Vma13p has moved out of its V 1 inhibitory conformation to allow MgATP hydrolysis to occur, but V 1 and V 0 are not attached in a manner compatible with ATP-driven proton pumping. From the data presented here and the information available about Vma13p binding and function, we cannot distinguish whether the V-ATPase is functionally uncoupled because the V 0 is an open proton pore or because conformational changes generated by ATP hydrolysis cannot be relayed to the proton pore, perhaps because of a structural defect in one of the stalks. Partial loss of Vma4p and Vma8p, two putative stalk subunits, would be consistent with either functional defect; further experiments may allow us to distinguish between these two possibilities. It is notable that a milder uncoupling defect was also observed for several vma8⌬ mutants (40). How does overproduction of Vma13p generate the uncoupled V 1 V 0 complexes? If Vma13p has binding sites on both the V 1 and V 0 sectors, as suggested by its role in activating coupled ATP hydrolysis and proton transport and by cross-linking (39) and two-hybrid (41) experiments, excess Vma13p might allow binding of individual Vma13p molecules to V 1 and V 0 . Isolated vacuoles do not contain excess Vma13p (Fig. 5A), but this does not eliminate the possibility that low affinity binding of a second Vma13p is responsible for generating the final complex we observed. The data showing normal levels of Vma4p and Vma8p in cytosolic V 1 complexes (Fig. 7) argue against an alternative model in which excess Vma13p sequesters these subunits in the cytosol, preventing their assembly with V 1 . It is notable that mammalian V-ATPases contain two isoforms of subunit H (Vma13p) (42,43), and there is evidence indicating that both isoforms may be present at nonequivalent sites in a single V-ATPase complex (39). There is no evidence of a second Vma13p isoform in yeast, but it is possible that overexpression allows inappropriate binding of a second Vma13p to a site similar to that occupied by the second isoform in mammalian cells and thus alters the properties of the yeast enzyme.  Fig. 7 were lysed, and cytosolic V 1 sectors were isolated and characterized as described (14). A, shown is the purification profile of cytosolic proteins applied to a Mono-Q1 ion-exchange column. Cells were converted to spheroplasts, and cytoplasmic proteins were obtained by centrifugation and then fractionated by ammonium sulfate precipitation and desalted. The desalted sample was applied to a Mono-Q1 column equilibrated with buffer A (0.05 mM Tris-HCl (pH 7.5), 30 mM NaCl, 30 mM KCl, 0.3 mM EDTA, and 9.6 mM ␤-mercaptoethanol). Bound proteins were eluted with three stepwise linear gradients of 0 -30% buffer B (0.05 mM Tris-HCl (pH 7.5), 0.2 M NaCl, 0.2 M KCl, 0.3 mM EDTA, and 9.6 mM ␤-mercaptoethanol), 30 -40% buffer B, and 40 -100% buffer B as indicated. Protein peaks were monitored by measuring absorbance at 280 nm. B and C, 2-ml fractions were collected from the ion-exchange column, and proteins were precipitated with 10% trichloroacetic acid and then solubilized, separated by SDS-PAGE, and transferred to nitrocellulose. Immunoblots were probed with antibodies against Vma1p, Vma2p, and Vma4p (B) or Vma5p (C) as described in the legend to Fig. 5. Fractions 20 -26 were previously determined to contain fully assembled cytosolic V 1 sectors by nondenaturing immunoprecipitation (14).
Several other features of Vma13p must also be considered in interpreting these data, however. Vma13p is unique among the V-ATPase subunits in its ability to interact with other, non-V-ATPase complexes. Surprisingly, human VMA13 was shown to bind the human immunodeficiency virus-1 NEF protein and more recently was demonstrated to play a critical role in facilitating NEF internalization (44). In yeast, Vma13p was recently demonstrated to interact with the Golgi-localized ectoapyrase Ynd1p and to act as an inhibitor of its activity (45). Although Vma13p appeared to act in the context of fully assembled V-ATPase complexes (whether active or inactive) in its regulation of Ynd1p, overexpressed Vma13p generated even lower levels of Ynd1p-dependent ADPase activity than VMA13 expressed at wild-type levels, probably due both to down-regulation of Ynd1p levels and further inhibition of the enzyme by excess Vma13p. It is possible that the full spectrum of Vma13pbinding partners and cellular functions is not yet known, and phenotypes of VMA13 overexpression could reflect its activities in these other complexes as well. This could help to account for the gain of function that we observed when VMA13 was overexpressed at modest levels in vma11⌬ cells or at high levels in wild-type cells. We do not believe that alternative Vma13p functions account for the uncoupled V-ATPase and proton pumping activities in isolated vacuoles, however, because the V-ATPase complexes show specific structural defects. It is also significant that modest levels of VMA13 overexpression in an otherwise wild-type background (Fig. 1) yield a characteristic Vma Ϫ phenotype, suggesting that a specific V-ATPase defect predominates under these conditions.
Overexpression of Vma5p Generates a Novel Cytosolic V 1 Complex-Although both Vma13p and Vma5p have been viewed as structurally and functionally bridging V 1 and V 0 complexes, there are both similarities and differences in the phenotypes of vma13⌬ and vma5⌬ strains. Both deletions exhibit the full range of Vma Ϫ growth defects, and both allow more assembly of both V 1 and V 0 sectors than any other V 1 subunit deletion strain (18,19,22). Isolated vacuoles from vma5⌬ strains contain very low levels of V 1 sector subunits, however (22). Vma5p is also lost from both sectors when V 1 and V 0 sectors dissociate during glucose depletion and is reassembled with both sectors upon glucose re-addition (9). These data are consistent with a critical role for Vma5p in attachment of V 1 and V 0 and possible involvement in signaling their reversible disassembly in vivo. It has also been proposed that release of Vma5p from V 1 could play a role, in cooperation with Vma13p inhibition, in the silencing of MgATP hydrolysis in cytosolic V 1 sectors (14).
The data shown here demonstrate that excess Vma5p can affect both fully assembled V-ATPase complexes and free V 1 complexes. V-ATPase complexes in isolated vacuoles are destabilized by the presence of excess Vma5p, but the specific structural defect is subtle because the complexes have an apparently normal subunit composition. This type of subtle defect in V-ATPase structure is also characteristic of a number of VMA5 point mutations 2 and may indicate that Vma5p is responsible not only for the crude attachment of V 1 sectors to V 0 , but also for more subtle features underlying "correct" attachment of the two complexes. Significantly, there is no evidence of uncoupling of V 1 and V 0 activities in isolated vacuoles; ATP hydrolysis and proton pumping are decreased by roughly the same amount in vacuoles from VMA5-overexpressing strains (Table I and Fig.  4). The effects of VMA5 overexpression on cytosolic V 1 sectors are somewhat clearer. Vma5p has never been isolated with cytosolic V 1 sectors from glucose-deprived cells or yeast cells containing only cytosolic V 1 sectors because of deletion of a V 0 subunit (14). However, it does coprecipitate with V 1 from vma11⌬ cells overexpressing VMA5. In this case, excess VMA5 may allow binding to a low affinity site on V 1 that cannot be bound in the presence of normal Vma5p levels. Although we did not observe Mg-ATPase activity in our assays of the ion-exchange fractions, these fractions contained lower levels of subunit protein than we assayed in the past (14). It remains an intriguing possibility that Vma5p bound to cytosolic V 1 complexes activates some level of futile ATP hydrolysis and that this accounts for the deleterious gain of function observed when Vma5p is overproduced at high levels or in the presence of high levels of cytosolic V 1 (in a vma11⌬ strain). Further experiments will be necessary to test this possibility. Vma5p has not yet been shown to participate in other complexes as Vma13p does, so it is more difficult to attribute overexpression phenotypes to non-V-ATPase-associated activities.
Physiological Relevance of V 1 Subunit Overexpression Studies-Although we have focused on the biochemical significance of the complexes formed in the presence of VMA5 and VMA13 overexpression, these results do have physiological relevance, particularly for VMA5. The human VMA5 gene was the only V-ATPase subunit gene identified among those genes whose mRNA levels change in response to cytomegalovirus infection; VMA5 mRNA levels were 8-fold higher 24 h after viral infection (46). VMA5 mRNA levels were 8.9-fold higher after yeast cells were treated for 10 min with 0.4 M NaCl; but in this case, several other V-ATPase subunit genes were coordinately upregulated (35). Protein levels were not determined in either of these studies, so it remains to be determined whether and to what level Vma5p is overproduced; but the studies reported here suggest that the functional consequences of even 2-3-fold overproduction of Vma5p could be very significant.