The H subunit (Vma13p) of the yeast V-ATPase inhibits the ATPase activity of cytosolic V1 complexes.

V-ATPases are composed of a peripheral complex containing the ATP-binding sites, the V(1) sector, attached to a membrane complex containing the proton pore, the V(o) sector. In vivo, free, inactive V(1) and V(o) sectors exist in dynamic equilibrium with fully assembled, active V(1) V(o) complexes, and this equilibrium can be perturbed by changes in carbon source. Free V(1) complexes were isolated from the cytosol of wild-type yeast cells and mutant strains lacking V(o) subunit c (Vma3p) or V(1) subunit H (Vma13p). V(1) complexes from wild-type or vma3Delta mutant cells were very similar, and contained all previously identified yeast V(1) subunits except subunit C (Vma5p). These V(1) complexes hydrolyzed CaATP but not MgATP, and CaATP hydrolysis rapidly decelerated with time. V(1) complexes from vma13Delta cells contained all V(1) subunits except C and H, and had markedly different catalytic properties. The initial rate of CaATP hydrolysis was maintained for much longer. The complexes also hydrolyzed MgATP, but showed a rapid deceleration in hydrolysis. These results indicate that the H subunit plays an important role in silencing unproductive ATP hydrolysis by cytosolic V(1) complexes, but suggest that other mechanisms, such as product inhibition, may also play a role in silencing in vivo.

V-ATPases 1 are highly conserved proton pumps distributed throughout the vacuolar network in all eukaryotic cells. V-ATPases maintain organelle acidification and affect cytosolic pH and ion balance, and their activity has been linked to a diverse array of cellular processes ranging from zymogen activation to protein sorting to viral membrane fusion events (1,2). V-ATPases are comprised of two structural domains, the V 1 domain, which consists of a complex of peripheral subunits containing the nucleotide-binding sites attached to the cytoplasmic face of membrane, and the V o domain, which is comprised of several integral membrane and tightly associated peripheral proteins that contain the proton pore (1,2). The yeast V-ATPase has at least eight V 1 subunits (designated A, B, C, D, E, F, G, and H) and five V o subunits (designated a, c, cЈ, cЉ, and d) (1,3,4). Genetic and biochemical approaches have converged to show that all of the subunits are required for function of V 1 V o complexes (3,4). Other eukaryotic V-ATPases have very similar subunit compositions.
Functional interdependence of V 1 and V o has been clearly established. Only fully assembled V 1 V o complexes can couple ATP hydrolysis to H ϩ translocation, and in vitro experiments indicate that upon V 1 dissociation, the V o domain does not conduct protons and the V 1 domain does not perform MgATP hydrolysis (5)(6)(7)(8). Nevertheless, many different cells have been shown to contain free V 1 and free V o sectors in addition to fully assembled V 1 V o complexes (9 -12). Independent experiments in yeast and Manduca sexta have indicated that the disassembled V 1 and V o sectors exist in a dynamic equilibrium with fully assembled complexes and that this equilibrium can be shifted in response to changes in extracellular conditions (10,13,14). Starvation appears to stimulate disassembly of V 1 from V o , but this disassembly is fully reversible upon refeeding (13)(14)(15). This reversible association between V 1 and V o is believed to regulate V-ATPase function in vivo: disassembly of V-ATPase complexes may conserve ATP when energy reserves are low and reassembly of the enzyme may provide the renewed proton pumping capacity necessary to prevent cytosolic acidification when active metabolism resumes (16,17).
A constitutively active free V 1 in the cytosol could quickly become lethal to the cell by hydrolyzing cytosolic reserves of ATP. Graf et al. (14) have isolated cytosolic V 1 complexes from M. sexta and shown that these complexes exhibit Ca 2ϩ -dependent ATP hydrolysis at nonphysiological Ca 2ϩ concentrations but hydrolyze MgATP only in the presence of methanol. The properties of V 1 complexes have also been examined by reconstitution of expressed subunits and biochemically isolated subcomplexes of the bovine clathrin-coated vesicle ATPase (18 -21). These studies have also revealed a shift from Mg 2ϩdependent to Ca 2ϩ -dependent ATPase activity in V 1 complexes detached from the membrane subunits and have suggested that CaATPase activity is a partial reaction characteristic of dissociated V 1 sectors that is functionally related to the MgATPase activity of the fully assembled proton pump (18).
In an attempt to gain more insight into the cellular mechanisms of V 1 -ATPase silencing, we have purified and characterized native yeast cytosolic V 1 complexes. Cytosolic V 1 were isolated from wild-type cells and from two vma mutant strains. We found that V 1 subunit C was not present in any of the isolated complexes. All of the isolated V 1 complexes hydrolyzed ATP in the presence Ca 2ϩ , but only V 1 complexes lacking subunit H had MgATPase activity. The current study also indicates that product inhibition of ATPase activity may occur in cytosolic V 1 complexes but cannot fully account for the inactivation of these complexes. In addition, structural changes within the V 1 complex itself, such as lost of an activator subunit (C subunit) and presence of at least one inhibitory subunit (H subunit) may be critical for silencing the MgATPase activity in vivo.

EXPERIMENTAL PROCEDURES
Materials and Strains-Zymolyase 100T was purchased from ICN. Concanamycin A was obtained from Wako Biochemicals. Prestained molecular mass markers (high range) were obtained from Life Technologies. ATP bioluminescence assay kit HS II and anti-Myc monoclonal antibody 9E10 were purchased from Roche Molecular Biochemicals. All other reagents were purchased from Sigma.
The wild-type yeast strain used in these experiments was SF838-1D (MAT␣ ade6 leu2-3,112 ura3-52 pep4-3 gal2 (22)). The vma3⌬ strain was congenic with the wild-type strain except for the vma3⌬::URA3 mutation (23). A congenic vma13⌬ strain was constructed by excising a BamHI-SacII fragment containing the vma13⌬::LEU2 allele from the deletion plasmid described below, and integrating into the VMA13 locus by a one-step gene disruption (24). Replacement of the wild-type allele by the deletion allele was confirmed by PCR of chromosomal DNA prepared from yeast cells.
VMA13 Plasmid Constructions-VMA13 in pRS315 was a gift from R. Hirata. VMA13 was tagged with the Myc epitope immediately following the methionine start codon by employing PCR and subcloning techniques. Two separate PCR reactions were performed using VMA13 as a template. Reaction A utilized primer 1: 5Ј-AGAAATAAGCTTTGT-TCCATTGTTCCTGAAATCGC, and primer 2: 5Ј-GACGAAGGAATTT-GAAAGAG, this reaction generated a 564-base product that encompassed 546 5Ј-untranslated region bases and 18 5Ј bases of the Myc epitope including the HindIII site that is found within the Myc sequence. Reaction B utilized primer 3: 5Ј-CAAAGCTTATTTCTGAAA-GACTTGGGAGCACGAAGATATT and primer 4: 5Ј-GATCACGCATA-ACC generating a 1912-base product which contained 27 bases of the Myc epitope including the HindIII site, 1433 bases of VMA13 open reading frame, and 452 bases of 3Ј-untranslated region. PCR products were ligated into pCR2.1 TM (Invitrogen). Orientation of products was determined based on restriction digests. Reaction A was subcloned into pRS316 using BamHI and HindIII sites, Reaction B product was subcloned into the newly constructed pRS316 vector containing the Reaction A product using KpnI and HindIII sites. Sequencing confirmed the presence of the Myc epitope within wild-type VMA13. N-Myc VMA13 is able to complement the growth phenotypes of a vma13⌬ strain, and vacuolar vesicles isolated from this strain possess 80% wild type ATPase activity. 2 VMA13 was cloned into the yeast shuttle vector pRS316 at the BamHI and NotI sites. A deletion plasmid was constructed by replacing the 1084-base pair BglII fragment within the ORF of VMA13 with a 2.2-kilobase Leu2 fragment.
Purification of Cytosolic V 1 Complexes-Cells were grown overnight to mid-log phase (3 A 600 /ml) in YEPD (1% yeast extract, 2% peptone, 2% glucose) medium adjusted to pH 5. 6000 A 600 units of cells (approximately 6 ϫ 10 10 cells) were harvested by centrifugation at 2500 ϫ g for 10 min and resuspended in 300 ml of 0.05 M Tris-HCl, pH 9.4, containing 10 mM dithiothreitol. Cells were rocked for 5 min at 30°C, pelleted by centrifugation for 5 min at 2200 ϫ g, and the pellet resuspended in 300 ml of 0.05 M Tris-HCl, pH 7.5, 1.2 M sorbitol, 2% glucose. Cells were converted to spheroplasts by adding 1500 units of zymolase 100T to the suspension and gently shaking at 30°C for 20 min. Spheroplasts were washed twice with 300 ml of YEPD medium containing 1.2 M sorbitol. In certain experiments, spheroplasts were briefly deprived of glucose by incubating 5 min at 30°C in 200 ml of YEP (1% yeast extract, 2% peptone) plus 1.2 M sorbitol. Otherwise, incubation was performed in 200 ml of YEPD plus 1.2 M sorbitol. Finally, spheroplasts were collected by centrifugation and lysed on ice in 15 ml of buffer A (0.05 M Tris-HCl, pH 7.5, 30 mM NaCl, 30 mM KCl, 0.3 mM EDTA) containing a protease inhibitor mixture (1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin, 5 g/ml aprotinin A, 1 g/ml leupeptin) in a Dounce homogenizer. Homogenate was centrifuged at 275,000 ϫ g for 1.25 h in a Ti-75 rotor and the supernatant (15-20 mg of protein/ml) precipitated with 50% ammonium sulfate. Precipitation was performed by dropwise addition of a cold saturated solution of ammonium sulfate, pH 7, in three steps of 0 -20, 20 -35, and 35-50% (v/v) with 15 min incubation between additions and constant stirring on ice. After the final addition, the mixture was incubated on ice for 30 min and the protein pelleted at 9,000 ϫ g for 11 min. The precipitated protein was resuspended in buffer A and desalted on a Centricon Plus-20 filter (100,000 dalton cutoff; Amicon), then filtered and applied to a Mono-Q2 column (Bio-Rad) equilibrated in buffer A containing 9.6 mM ␤-mercaptoethanol. The column was washed with 10 ml of equilibration buffer and the bound protein eluted with three sequential linear gradients: 5 ml of 0 -30% buffer B (0.05 M Tris-HCl, pH 7.5, 0.2 M NaCl, 0.2 M KCl, 0.3 mM EDTA, 9.6 mM ␤-mercaptoethanol) followed by a 20-ml isocratic flow of 70% buffer A, 30% buffer B, 6 ml of 30 -40% buffer B followed by a 20-ml isocratic flow in 60% buffer A, 40% buffer B, and 5 ml of 40 -100% buffer B followed by a 5-ml isocratic flow in 100% buffer B. 1-ml fractions were collected. Fractions were analyzed for the presence of V 1 subunits by Western blotting, and fractions containing V 1 subunits were immunoprecipitated under nondenaturing conditions with monoclonal antibodies 8B1 or 13D11 (against the 69-and 60-kDa V 1 subunits, respectively) to identify those containing V 1 complexes (11). Fractions containing V 1 complexes (fractions 43-54) eluted at 40% buffer B (0.1 M NaCl, 0.1 M KCl) and were pooled and concentrated on a Centricon Plus-20 filter (100,000 dalton cut-off). Pooled Q-2 fractions were applied to a Bio-Rad Sec 400 gel filtration column equilibrated with 30% buffer A, 70% buffer B. Purified V 1 complexes (0.05-0.5 mg) were collected in a single fraction. Chromatography was performed on a Bio-Rad BioLogic system.
Enzyme Assays-Hydrolysis of ATP quantitated colorimetrically as the phosphate released based on the Taussky and Schorr method (25). Briefly, reaction was started by addition of 1.5-15 g of purified V 1 to 500 l of ATPase assay medium (0.05 mM Tris-HCl, pH 6.8, containing 4 mM ATP or GTP and 1.6 mM CaCl 2 or MgCl 2 ). The final metalnucleotide complex concentration in the medium was calculated for each condition by the Bound and Determined computer program (26). Incubations were performed at 37°C for the indicated times (0.5-30 min). Reactions were stopped by addition of an equal volume of 10% (w/v) SDS. Phosphate released was determined by measuring the absorbance at 700 nm immediately after addition of 0.5 ml of Taussky and Schorr reagent (10% FeSO 4 , 1.2 N sulfuric acid, 1.2% ammonium molybdate). A blank containing only assay medium was measured for each reaction. A standard calibration curve for P i was used to calculate the micromoles of P i formed. Data were analyzed using the Sigma Plot curve-fitting application program.
The amount of ATP and ADP bound to the purified V 1 preparation was determined as follows. V 1 (20 -60 g) was precipitated by addition of perchloric acid to a final concentration of 0.44 M, and after 15 min on ice, the mixture was neutralized by addition of equal volume of ice-cold fresh 0.8 M potassium bicarbonate, then incubated for an additional 20 min on ice. Soluble nucleotides were recovered in the supernatant after centrifugation. ATP and ADP concentrations were measured using the luciferin-luciferase assay in an Autolunat LB953 luminometer. ADP was converted to ATP by addition of 8.3 mM phosphoenolpyruvate and 48 g of pyruvate kinase in buffer containing 50 mM HEPES-KOH, pH 7.5, 5 mM MgCl 2 , and 20 mM KCl. Protein concentrations were determined by Lowry assay (27).

RESULTS
Purification of Cytosolic V 1 Complexes-In order to better understand how cytosolic V 1 sectors are inactivated and possibly to gain insights into how V 1 dissociation is triggered, we isolated and characterized cytosolic V 1 sectors from yeast. Cytosolic V 1 complexes from wild-type yeast cells and vma3⌬ mutant cells, treated both with and without a brief glucose deprivation, were purified by a variation of the methods reported by Graf et al. (14). In wild-type cells, the population of V 1 complexes in the cytosol was increased by depriving the cells of glucose for 5 min. This treatment has been shown to trigger dissociation of approximately 75% of assembled V 1 V o complexes (13,15). vma3⌬ mutant cells lack the gene encoding the proteolipid subunit c of the V o sector (28) and provided an alternative source of V 1 . vma3⌬ mutants do not form stable V o sectors, but assemble stable V 1 complexes constitutively present in the cytosol (11,23). Wild-type and vma3⌬ cells were converted to spheroplasts and osmotically lysed. The purification procedure consisted of four fractionation steps and yielded ϳ0.05-0.1 and ϳ0.4 -0.5 mg of V 1 from 2 liters of log-phase culture of wild-type and vma3⌬ cells, respectively. Glucose deprivation improved the yield of V 1 sectors from wild-type cells, but did not significantly affect the yield of V 1 sectors from the vma3⌬ cells. Briefly, the purification consisted of isolation of a soluble fraction by high speed centrifugation, followed by protein precipitation with 50% ammonium sulfate and two sequential chromatographic columns: ion exchange on a Mono-Q column and gel filtration on a Biosilect Sec-400 column. Fig. 1A shows the protein elution profile from the ion exchange column, and Fig. 1B is a Western blot analysis of the elution pattern of A, B, and C V 1 subunits. The C subunit failed to bind to the column even at the lowest salt concentration and fractionated away from the rest of the cytosolic V 1 subunits. Because the A and B subunits were detected throughout the gradient, we assessed whether they were assembled with other subunits by nondenaturing immunoprecipitation of selected pooled fractions (not shown). Assembled V 1 complexes were present only in fractions eluted with 0.2 M salts (0.1 M NaCl, 0.1 M KCl); A and B subunits that eluted elsewhere from the column were either partially or fully dissociated from the remaining V 1 subunits. The assembled V 1 complexes eluted from the Mono-Q column (Fig. 1B, fractions 43-54) were concentrated and subjected to gel filtration chromatography. The elution profile for the gel filtration column is shown in Fig. 1C. The V 1 complexes eluted in a single peak (Fig. 1D) with an estimated molecular mass of 445 kDa.
Purified V 1 complexes from wild-type and vma3⌬ cells, either with or without glucose deprivation, showed a similar subunit composition. The presence of the 27-kDa E subunit in the V 1 complexes was confirmed by Western blotting (Fig. 1D). Silver staining of the peak eluted from gel filtration column (Fig. 2) showed additional bands of 32, 16, and 14 kDa that correspond in molecular mass to the previously identified D, G, and F subunits, respectively (4). These data indicate that the V 1 complexes obtained from both strains contained the A, B, D, E, F, and G V 1 subunits (Figs. 1D and 2). Both the E and G subunit had a somewhat smeared appearance in the V 1 preparations. In addition to the previously characterized V 1 subunits, bands of approximately 25 and 80 kDa and several high molecular mass bands were consistently present in the fractions containing the V 1 complexes. We have not yet determined whether these proteins are associated with cytosolic V 1 complexes.
We were particularly interested in determining whether the H subunit, encoded by the VMA13 gene in yeast (29), was associated with the cytosolic V 1 complexes. This protein has a molecular mass of 54 kDa and is often masked by the 60-kDa subunit, so the VMA13 gene was tagged with a Myc epitope to allow it to be clearly identified. The tagged protein was expressed in a vma13⌬ yeast strain and shown to fully complement the growth defects of the strain. Co-purification of subunit H with cytosolic V 1 complexes was confirmed using anti-Myc antibodies against V 1 complexes purified from vma13⌬ cells expressing the Myc-tagged VMA13 gene (Fig. 3). Therefore, the cytosolic V 1 sectors appear to contain all the previously characterized V 1 subunits except subunit C.
Enzymatic Activities of Cytosolic V 1 Complexes-The isolated V 1 domain of the M. sexta V-ATPase is not active as a MgATPase except in the presence of organic solvents (14). Similarly, the isolated yeast V 1 complexes did not hydrolyze ATP if the divalent cation supplied was Mg 2ϩ . In an attempt to activate the yeast V 1 ATPase activity, the purified V 1 was treated with 25% methanol, 30 mM octylglucoside, 5-50 mM sodium sulfite, 0.5% N,N-dimethyldodecylamine-N-oxide, and 5-10 mM dithiothreitol, treatments which had effectively activated the FIG. 1. Purification of cytosolic V 1 sectors. A and B, ion exchange chromatography of yeast cytosol. A, supernatant protein obtained after high speed centrifugation of a yeast cell lysate was precipitated with 50% ammonium sulfate, desalted, and applied to a Mono-Q2 ion exchange column as described under "Experimental Procedures." After an initial wash, proteins were eluted from the column with three sequential linear gradients. Protein concentration was monitored by measuring absorbance at 280 nm (A 280 ). A profile of the stepwise salt gradient used is superimposed on the protein elution profile and is described in more detail under "Experimental Procedures." B, the indicated fractions collected from the chromatogram shown in A were precipitated with 10% trichloroacetic acid, solubilized, and separated by SDS-PAGE, and blotted to nitrocellulose. The blot was probed with mouse monoclonal antibodies 7A2, which recognizes the C subunit, 13D11, which recognizes the B subunit, and 8B1, which recognizes the A subunit, followed by alkaline-phosphatase-conjugated goat anti-mouse antibodies (11). The fractions containing assembled V 1 complexes were identified by nondenaturing immunoprecipitation with monoclonal antibody 8B1 as described under "Experimental Procedures." C and D, isolation of cytosolic V 1 complexes by gel filtration. C, fractions 43-54 from the ion exchange chromatography column shown in A were pooled, concentrated, and loaded on a Bio-Rad Sec400 gel filtration column. Protein concentration in fractions eluted from the column was monitored by measuring the A 280 . D, V 1 subunits eluted from the gel filtration column in a single peak, centered at fraction 20. The indicated fractions were subjected to SDS-PAGE and immunoblotting. The A and B subunits were recognized with monoclonal antibodies 8B1 and 13D11 as described above, and the E subunit was recognized by polyclonal antiserum raised against the yeast E subunit (generously provided by Dr. Tom Stevens).

Vma13p Inhibits ATPase Activity of Cytosolic V 1 Complexes
MgATPase activity of the Manduca V 1 (14) or F 1 -ATPases from various sources (30 -33). None of these treatments elicited any MgATPase activity in the yeast V 1 complexes.
Cytosolic V 1 complexes from both wild-type and vma3⌬ cells did hydrolyze ATP in a Ca 2ϩ -dependent manner at nonphysiological (mM) Ca 2ϩ concentrations, however. CaATPase activity has been described in purified V 1 complexes from M. sexta (14), isolated chloroplast and Bacillus firmus F 1 complexes (30,31), and reconstituted mixtures of bovine V 1 subunits (18 -21). The enzymatic properties of complexes purified from glucose-deprived wild-type cells or vma3⌬ cells with or without glucose deprivation were very similar. Because vma3⌬ cells provided a more abundant source of cytosolic V 1 than wild-type cells, the kinetic analysis described below was performed on cytosolic V 1 complexes isolated from vma3⌬ mutant cells.
CaATP hydrolysis was first examined as a function of the time at a constant CaATP concentration (1.4 mM). When the incubation time was varied from 0.5 to 20 min, the plot of micromole of P i formed versus time had a hyperbolic shape showing a rapid initial rate that decayed until there was little further ATP hydrolysis after 3 min (Fig. 4A). The initial activity, detected at 1 min, was 1.7 mol of P i /min/mg. At a lower CaATP concentration (0.3 mM), it took longer for the activity to decay, but the activity was gone by 20 min. An apparent K m of 0.183 mM for CaATP, which is similar to the K m of yeast V 1 V o (0.210 mM; Ref. 34) was estimated from 1-min reactions performed at a larger range of concentrations. Based on this information, 1.4 mM CaATP should nearly saturate the enzyme, and the loss of activity over time seen in Fig. 4A cannot be attributed to substrate depletion.
The substrate specificity of the yeast V 1 complexes was examined. Ca 2ϩ -dependent hydrolysis of GTP was observed (Fig.  4B). Interestingly, CaGTP hydrolysis was linear for at least 20 min under conditions where ATP hydrolysis had ceased after 3 min. The V 1 complexes exhibited a specific activity for GTP hydrolysis of 0.47 mol/min/mg of protein. Once again the hydrolysis was Ca 2ϩ -dependent; Mg 2ϩ did not support any GTPase activity in the cytosolic V 1 complexes.
Loss of activity over time could be an indication of product inhibition of the ATPase activity. Product inhibition has been studied in considerable detail in F 1 -ATPases, and appears to be specific to ADP in many cases (35)(36)(37). Thus, although GTP is a substrate for F 1 , GDP is much less efficient in product inhibition (36). To further explore the possibility that the ATPase activity of the cytosolic V 1 complexes was inhibited by ADP, V 1 complexes were preincubated in the presence of CaADP before measurement of the CaATPase activity. Isolated V 1 complexes were preincubated with 1.1 mM CaADP for 1 min, then the V 1 -CaADP mixture was diluted 30-fold into assay medium and the CaATPase activity measured in the presence of 1.4 mM CaATP (Fig. 4A). The CaATPase activity of the V 1 complexes was not fully inhibited by CaADP preincubation. The initial ATPase activity was 51% that in the absence of ADP, and a decay in ATPase activity similar to that seen in the absence of ADP preincubation was observed over the next 3 min. We also determined whether the cytosolic V 1 preparation contained tightly bound nucleotides after isolation that might be involved in inhibition of either Mg 2ϩ -dependent or Ca 2ϩ -dependent ATPase activity. Only substoichiometric amounts of ADP (0.02 mol/mol V 1 ) and ATP (0.005 mol/mol V 1 ) were detected in the isolated cytosolic V 1 complexes. Pyrophosphate was shown to enhance MgATPase activity of F 1 -ATPases by removing tightly bound nucleotides (36). However, addition of 4.8 -9.4 mM PPi did not activate the ATPase activity of cytosolic yeast V 1 .
The sensitivity of the yeast cytosolic V 1 to a variety of inhibitors was examined. The CaATPase activity of the cytosolic V 1 was not affected by addition of the specific P-and F-type ATPases inhibitors sodium orthovanadate (1 mM) and sodium azide (10 mM), respectively. Concanamycin A, a specific V-type ATPase inhibitor believed to interact with the V o domain at the membrane (38) had no effect on the ATPase activity of purified V 1 . V-ATPases contain a set of three conserved cysteine residues that are essential for activity and render the enzyme sensitive to low concentrations of N-ethylmaleimide (39). Measuring N-ethylmaleimide sensitivity of the cytosolic V 1 sectors was difficult because the presence of reducing agent (␤-mercaptoethanol) appeared to be essential for purification of an active V 1 and maintenance of its activity. However, addition of N-ethylmaleimide in excess to the concentration of ␤-mercaptoethanol in the isolated V 1 preparation allowed us to estimate FIG. 3. Cytosolic V 1 complexes contain the H subunit. V 1 complexes were isolated from vma13⌬ mutant cells bearing a Myc epitopetagged VMA13 gene on a low copy plasmid as described in the legend to Fig. 1 and under "Experimental Procedures." The indicated fractions from the gel filtration column were separated by SDS-PAGE and subjected to immunoblotting. The blot was probed with monoclonal antibodies 8B1 against the A subunit and 9E10 against the myc epitope attached to the H subunit.

FIG. 4. CaATPase and CaGTPase activity of cytosolic V 1 complexes.
A, cytosolic V 1 complexes were isolated from vma3⌬ cells, and 11.2 g of the isolated complexes were incubated with 1.4 mM CaATP (open circles) in a 500-l total volume for the indicated times. P i release was monitored by colorimetric assay (25). CaATPase activity was also measured after a 1-min preincubation of the complexes with 1.1 mM CaADP followed by a 30-fold dilution into assay buffer (500 l final volume) containing 1.4 mM CaATP (closed circles). B, cytosolic V 1 complexes (11.2 g) were incubated with 1.4 mM CaGTP (closed circles) in a 500-l total volume for the indicated times (closed circles), and phosphate release was monitored as described above. CaATPase activity from A (open circles) is shown for comparison.

Vma13p Inhibits ATPase Activity of Cytosolic V 1 Complexes
an IC 50 of 0.5 mM for N-ethylmaleimide.
Purification and Characterization of Cytosolic V 1 Complexes Lacking the H Subunit-To address the function of the H subunit in cytosolic V 1 complexes, we isolated the V 1 complex from a vma13⌬ mutant strain. vma13⌬ cells assemble unstable and inactive V 1 V o complexes at the vacuolar membrane (29). vma13⌬ cells were briefly deprived of glucose (5 min in YEP) and the cytosolic V 1 complexes purified as described previously. As shown in Fig. 5, V 1 complexes isolated from vma13⌬ cells showed the same subunit composition as V 1 from wild-type and vma3⌬ cells with the exception of the loss of the H subunit, which runs just below the 60-kDa B subunit.
Overall CaATPase activity of V 1 complexes missing H subunit was higher than that of V 1 complexes from wild-type and vma3⌬ cells, primarily because the kinetics of ATP hydrolysis were linear for almost 20 min. The initial specific activity was 1.8 -4.0 mol of P i /min/mg in two different preparations (Fig.  6), only slightly higher than that seen in the V 1 complexes isolated from vma3⌬ cells. Linearity of the reaction over as much as 20 min suggested little or no product inhibition occurred during hydrolysis by the cytosolic V 1 complexes lacking subunit H. To better understand the lack of decay in the catalytic activity, we repeated the ADP preincubation experiment shown in Fig. 4A with the complexes from vma13⌬ cells. The cytosolic V 1 complex from vma13⌬ cells was preincubated with 1.1 mM CaADP and its ATPase activity measured as described above (Fig. 6B). After preincubation with ADP, the enzymes initial activity was 66% that without preincubation. V 1 complexes isolated from vma3⌬ cells retained 51% of the initial activity, so complexes from vma13⌬ cells were only slightly less sensitive to ADP inhibition. The kinetics of hydrolysis remained nearly linear over a 30-min period, however. These results indicate that in the absence of subunit H, the cytosolic V 1 complexes can still be inhibited by preincubation with ADP, but they still do not experience the decay of activity over time that may be an additional effect of product ADP in the complexes containing subunit H.
The cytosolic V 1 complexes lacking subunit H also exhibited some MgATPase activity, even in the absence of activating agents. Kinetics of MgATP hydrolysis revealed an initial specific activity of 1.2 mol of P i /min/mg measured at 1 min (Fig.  7). Activity drastically decreased after the first 5 min. An apparent specific activity of 0.265 mol of P i /min/mg was measured after 15 min in assay medium containing an initial concentration of 1.4 mM MgATP. Potential activators were added during a 15-min incubation, and their effects on the MgATPase activity are shown in Table I. Methanol (25% v/v) gave an almost 2-fold increase in the apparent specific activity, but octylglucoside and sodium sulfite proved to be somewhat inhibitory.

Subunit Composition and Enzymatic Activities of Cytosolic V 1 Complexes-
The yeast cytosolic V 1 complexes contain established V 1 subunits A, B, D, E, F, G, and H. Subunit C was the only V 1 subunit not associated with yeast cytosolic V 1 complexes; this subunit was present in a high speed supernatant, but fractionated away from the other V 1 subunits in ion exchange chromatography. Earlier immunoprecipitation experiments had indicated that the C subunit dissociated from both the V 1 and V o sectors during V-ATPase disassembly (13). The subunit composition of the yeast cytosolic V 1 complexes closely resembles that of the cytosolic V 1 -ATPase complexes from M. sexta (14,40). The M. sexta complexes appear to contain subunits A, B, D, E, F, and G, along with substoichiometric amounts of the C subunit (40). Subunit H has only recently been identified in M. sexta (17), so it is unclear whether this subunit is really not present in the insect cytosolic V 1 complexes, or is hidden by the B subunit, which generally runs very close to the H subunit on SDS-PAGE.
Both the insect and yeast cytosolic V 1 complexes are active as CaATPases, indicating that this activity does not require the presence of the C subunit. There were other striking similarities in the activities of the two enzyme preparations. Both showed a loss of CaATPase activity over time and a lower level of CaGTPase activity that did not decay over time. As noted by Graf et al. (14), these features are also shared by isolated F 1 sectors from chloroplasts and B. firmus (30,31). One difference between the insect and yeast V 1 preparations is the activation of MgATPase activity in the insect enzyme in the presence of 25% methanol; no MgATPase activity was observed for the wild-type yeast V 1 preparation, even in the presence of a wide variety of potential activating agents. A shift from Mg 2ϩ -dependent to Ca 2ϩ -dependent ATP hydrolysis has been observed in V 1 subunit reconstitution experiments of the bovine clathrin-coated vesicle V-ATPase, as well. These subunit reconstitution experiments had indicated an essential role for the C subunit in CaATP hydrolysis (19), however, and this appears to conflict with results from the native cytosolic V 1 preparations.
We had anticipated that cytosolic V 1 sectors isolated from wild-type yeast cells and vma3⌬ mutant cells before and after glucose deprivation might show differences in subunit composition that reflected their different histories and provided indications as to how glucose deprivation signals V 1 dissociation. This did not prove to be the case, at least at the level of analysis reported here. Both the subunit composition and basic enzymatic properties of the cytosolic V 1 sectors from different sources were very similar. It may still be that there are subtle differences in post-translational modifications of the subunits that we have not yet identified, and we plan to look at the different preparations in more detail in the future. It is also possible, however, that cytosolic V 1 sectors from wild-type cells before and after glucose deprivation have the same structure, and that different amounts of V 1 are present because the equilibrium of an ongoing dissociation and reassociation is shifted when glucose becomes limiting. Along the same lines, the cytosolic V 1 sectors that are formed in vma3⌬ cells but never attach to the membrane may be stable because they resemble the cytosolic wild-type V 1 sectors that are normally cycling on and off the membrane.
How Is Mg 2ϩ -dependent ATP Hydrolysis by Cytosolic V 1 Sectors Silenced?-Reversible disassembly of V-ATPases has been proposed to be a mechanism of down-regulating V-ATPase activity when growth conditions are unfavorable (16,17). Under- FIG. 5. Subunit composition of cytosolic V 1 complexes isolated from vma13⌬ cells. Cytosolic V 1 complexes were isolated from wildtype and vma13⌬ mutant cells as described in the legend to Fig. 1 and under "Experimental Procedures." The V 1 peak fraction after gel filtration chromatography was subjected to SDS-PAGE and stained with Coomassie Blue. The positions of known V 1 subunits are indicated; the identities of the A, B, and E subunits were confirmed by immunoblotting.
Vma13p Inhibits ATPase Activity of Cytosolic V 1 Complexes lying this proposal is the assumption, consistent with in vitro data (6,8), that the cytosolic V 1 sectors are inactive in ATP hydrolysis. As expected, native cytosolic V 1 complexes purified from wild-type yeast cells could not hydrolyze ATP when Mg 2ϩ was provided as the divalent cation, indicating that under physiological conditions, the yeast V 1 complexes are catalytically inactive. The results reported here suggest several potential reasons cytosolic V 1 sectors are not active in vivo.
Characterization of V 1 ATPase activity from a vma13⌬ mutant suggests that the H subunit may play an important role in inhibiting both Mg 2ϩ -and Ca 2ϩ -dependent ATP hydrolysis by cytosolic V 1 sectors. An inhibitory role for the H subunit was not expected from previous data. The yeast vma13⌬ mutant, which lacks the H subunit, assembles V 1 V o complexes in the membrane (29), but the complexes are unstable and inactive. Addition of sub-57-kDa dimer, which consists of two isoforms of the H subunit, to a V 1 complex reconstituted from bovine clathrin-coated vesicle subunits enhances CaATPase activity of the complexes, and sub-57-kDa dimer or either of the individual H subunit isoforms appears to be essential for MgATPase activity and proton pumping by the fully assembled bovine clathrincoated vesicle pump (21,41). Taken together, these data have suggested that the H subunit may act as an activator, not an inhibitor, of V-ATPase activity, but these experiments have focused predominantly on the intact V 1 V o complex, not isolated V 1 sectors. The experiments presented here suggest that the H subunit may play a role more similar to that of the ⑀ subunit of the E. coli F-ATPase, which inhibits the F 1 -ATPase when it is detached from the membrane (33), but may be critical for proper structural and function coupling of F 1 and F o (42,43).
Comparison of CaATP hydrolysis by cytosolic V 1 sectors with and without the H subunit (Fig. 6A) indicates that the H subunit may be particularly critical for the decay in ATP hydrolysis rate after the first few minutes of turnover. Cytosolic V 1 complexes from vma13⌬ cells showed only a slightly higher initial rate than those from vma3⌬ cells, but they were able to maintain this rate for at least 20 min, under conditions where complexes from vma3⌬ cells were almost completely inactive after less than 5 min. The higher activity of the complexes from vma13⌬ cells could not be attributed to loss of another V 1 subunit, because all of the subunits except subunit H appeared to be present in these complexes. It is even more intriguing that cytosolic V 1 complexes lacking the H subunit appear to exhibit some Mg 2ϩ -dependent ATP hydrolysis that could be further activated in the presence of methanol. As described above, cytosolic V 1 sectors from M. sexta, which may or may not contain an equivalent of the H subunit, exhibited methanolactivated MgATPase activity in the cytosolic V 1 sectors. However, it is notable that the M. sexta enzyme appeared to lose Ca 2ϩ -dependent activity under conditions where it gained Mg 2ϩdependent activity, but both Ca 2ϩ and Mg 2ϩ -dependent activities appear to be activated in the cytosolic V 1 complexes from vma13⌬ cells. This result suggests that the methanol does not FIG. 6. CaATPase activity of cytosolic V 1 complexes from vma13⌬ cells. A, cytosolic V 1 complexes were isolated from vma13⌬ cells, and 10.3 g of the isolated complexes were incubated with 1.4 mM CaATP in a 500-l total volume for the indicated times (closed circles). The activity from 11.2 g of V 1 complexes isolated from vma3⌬ cells and assayed under identical conditions is shown for comparison (open circles). P i release was monitored by colorimetric assay (25). B, CaATPase activity was also measured after a 1-min preincubation of the complexes with 1. Cytosolic V 1 complexes were isolated from vma13⌬ cells, and 1.5 g of the isolated complexes were incubated with 1.4 mM MgATP in a 500-l total volume for the indicated times. P i release was monitored by colorimetric assay (25). Vma13p Inhibits ATPase Activity of Cytosolic V 1 Complexes act on the M. sexta enzyme simply through release of the H subunit or a functional equivalent. The MgATPase activity of yeast vma13⌬ complexes showed a loss of activity with time similar to that seen for the CaATPase activity of cytosolic V 1 sectors from wild-type cells, indicating that cytosolic V 1 sectors are still prevented from exhibiting high levels of unproductive ATP hydrolysis in vma13⌬ cells in vivo. These data suggest that the H subunit is important in inactivating cytosolic V 1 -ATPase activity, but there are probably other silencing mechanisms that act in combination, as described below.
One of these other mechanisms may be inhibition by ADP. The data presented here suggest that ADP could play a rather complex role in inhibiting the activity of cytosolic V 1 complexes. The loss of CaATPase activity in the wild-type complexes or MgATPase activity in the vma13⌬ complexes over time could have at least two explanations. First, enzyme activity may destabilize the complex so that one or more subunits is lost, inactivating the enzyme. We cannot eliminate this possibility at present, but it should be possible to address it by careful determination of the subunit composition before and after catalysis. Alternatively, the loss of activity is suggestive of product inhibition, which could also be an effective means of minimizing unproductive ATP hydrolysis in vivo. Similar behavior of the M. sexta cytosolic V 1 has been attributed to product inhibition (14), and a more detailed analysis of the V 1 -ATPase of Thermus thermophilus (44) has clearly demonstrated that this enzyme can be inactivated during ATP hydrolysis by entrapping an inhibitory MgADP at the catalytic site. In both of these cases, the similarity to entrapping of MgADP by F 1 -ATPases has been noted, but there are also some important differences between the behavior of the yeast cytosolic V 1 complexes and F 1 -ATPases (32,(35)(36)(37). First, briefly preincubating the yeast cytosolic V 1 with 1 mM CaADP gave only a partial inhibition of the initial CaATPase activity and did not appear to accelerate its decay. The extent of inhibition due to CaADP preincubation was similar in cytosolic V 1 complexes with or without the H subunit, even though the complexes without the H subunit showed much less inactivation during hydrolysis. Second, a number of activating agents that are believed to act by stimulating release of MgADP entrapped at a catalytic site of F 1 -ATPases, for example, sulfite (32), do not have any effect on the Ca 2ϩ -dependent activity of the yeast cytosolic V 1 . Perhaps most importantly, entrapment of a tightly bound MgADP or MgATP that persists through our purification protocol cannot account for the lack of MgATPase activity in cytosolic V 1 complexes from wild-type cells because the complexes as isolated are almost completely devoid of ADP and ATP. These data indicate that there may be at least two inhibitory effects of ADP: one type of inhibition depends on the formation of ADP during turnover and does not occur in complexes lacking the H subunit and the second type can be seen after a brief preincubation with ADP and occurs in complexes with and without the H subunit. The switch from MgATPase activity to CaATPase activity in the cytosolic V 1 complexes cannot be easily accounted for by the tighter binding of an inhibitory MgADP, unless this binding is so rapid and so tight that it occurs before significant MgATP hydrolysis can be observed. Complex effects of ADP on the V-ATPase of bovine clathrin-coated vesicles have been reported previously (45). Further experiments will be necessary to characterize the mechanisms of ADP inhibition of the yeast cytosolic V 1 complexes and fully assess their physiological significance.
The data presented here suggest that the inhibitory H sub-unit and inhibition by product ADP may play important roles is silencing unproductive hydrolysis by cytosolic V 1 complexes in yeast, but it is important to emphasize that they do not exclude other mechanisms of silencing. We have demonstrated release of the C subunit from cytosolic V 1 sectors, but have not yet determined whether this release plays a functional role. We have not yet assessed whether there are post-translational modifications of any of the V 1 subunits when they are released from the membrane, but with the purification protocol developed here, we are poised to determine both whether there are reversible modifications and whether these modifications affect activity. Silencing cytosolic V 1 complexes in vivo is likely to be a synergistic effect rather than a simple event. Considering that inhibition of cytosolic V 1 complexes is vital, it would not be surprising if cells had more than one mechanism to lock the catalytic conformation of the complex and prevent futile ATP hydrolysis.