Photoaffinity Labeling of Wild-type and Mutant Forms of the Yeast V-ATPase A Subunit by 2-Azido-[32P]ADP*

Molecular modeling studies have previously suggested the possible presence of four aromatic residues (Phe452, Tyr532, Tyr535, and Phe538) near the adenine binding pocket of the catalytic site on the yeast V-ATPase A subunit (MacLeod, K. J., Vasilyeva, E., Baleja, J. D., and Forgac, M. (1998) J. Biol. Chem. 273, 150–156). To test the proximity of these aromatic residues to the adenine ring, the yeast V-ATPase containing wild-type and mutant forms of the A subunit was reacted with 2-azido-[32P]ADP, a photoaffinity analog that stably modifies tyrosine but not phenylalanine residues. Mutant forms of the A subunit were constructed in which the two endogenous tyrosine residues were replaced with phenylalanine and in which a single tyrosine was introduced at each of the four positions. Strong ATP-protectable labeling of the A subunit was observed for the wild-type and the mutant containing tyrosine at 532, significant ATP-protectable labeling was observed for the mutants containing tyrosine at positions 452 and 538, and only very weak labeling was observed for the mutants containing tyrosine at 535 or in which all four residues were phenylalanine. These results suggest that Tyr532 and possibly Phe452and Tyr538 are in close proximity to the adenine ring of ATP bound to the A subunit. In addition, the effects of mutations at Phe452, Tyr532, Tyr535, and Glu286 on dissociation of the peripheral V1 and integral V0 domains both in vivo and in vitro were examined. The results suggest that in vivodissociation requires catalytic activity while in vitrodissociation requires nucleotide binding to the catalytic site.

The V-ATPases are composed of two domains, a peripheral, cytoplasmic 570-kDa V 1 domain responsible for nucleotide binding and hydrolysis, and a membrane integral 260-kDa V 0 domain responsible for proton translocation across the membrane (1)(2)(3)(4)(5)(6). The V 1 domain contains eight different subunits (subunits A-H, with molecular masses 70 -14 kDa) while the V 0 domain contains five different subunits (subunits a, d, cЉ, c, and cЈ, with molecular masses from 100 to 17 kDa). The nucleotide binding subunits have been identified as the A and B subunits, which are believed to form a hexameric structure containing three copies of each subunit (11).
Previous studies have demonstrated that the catalytic sites reside on the A subunits and that a second class of sites, termed "noncatalytic" sites, reside on the B subunits, giving a total of six nucleotide binding sites per complex (12)(13)(14)(15). The A and B subunits of the V-ATPase are approximately 25% identical with the ␤ and ␣ subunits of the F-ATPase, respectively (16,17). The F-ATPases function in ATP synthesis in mitochondria, chloroplasts, and bacteria (18 -20). The x-ray crystal structures of F 1 indicate that the nucleotide binding sites are located at the interface of the ␤ and ␣ subunits, with the catalytic sites residing primarily on the ␤ subunits and the noncatalytic sites residing primarily on the ␣ subunits (21,22).
Although mutagenesis studies of the yeast V-ATPase have begun to identify residues important for catalytic activity (23)(24)(25)(26), the structure of the nucleotide binding sites on the V-ATPase has yet to be determined. One approach to characterizing the structure of nucleotide binding sites is the use of photoaffinity analogs. 2-Azido-[ 32 P]ATP has been used to label both the catalytic and noncatalytic nucleotide binding sites of the F-ATPase (27)(28)(29), as well as the nucleotide binding sites of the bovine coated vesicle V-ATPase (13). The reactive nitrene moiety generated upon photoactivation of 2-azido-ATP has been shown to have a strong preference for reaction with tyrosine residues (over, for example, phenylalanine residues). Thus, 2-azido-[ 32 P]ATP reacts with Tyr 368 at the catalytic site of the bovine mitochondrial F 1 ␤ subunit (27), but substitution of phenylalanine at the corresponding position in the Escherichia coli ␤ subunit prevents the formation of a stable adduct. 2 In addition, when 2-azido-[ 32 P]ATP is loaded at the noncatalytic sites of F 1 , modification of Tyr␤ 354 is observed (28,29). Replacement of this tyrosine residue with phenylalanine eliminates labeling of the ␤ subunit and shifts the label to the ␣ subunit (29). When Arg␣ 365 at the noncatalytic site is changed to tyrosine, significant labeling of the ␣ subunit is observed. This shift of labeling to ␣ is not observed on substitution of phenylalanine at position 365 (29). The x-ray crystal structures of F 1 have confirmed the proximity of these aromatic residues to ATP bound at the catalytic and noncatalytic sites (21,22). These results suggest that reaction with 2-azido-[ 32 P]ATP can be used to identify tyrosine residues in close proximity to bound nucleotides.
Molecular modeling studies using the x-ray coordinates of F 1 ; the sequence homology between the ␣, ␤, A, and B subunits; and energy minimization programs have predicted the presence of four aromatic residues (Phe 452 , Tyr 532 , Tyr 535 , and Phe 538 ) near the adenine binding pocket at the catalytic site on the yeast V-ATPase A subunit (26). Mutagenesis studies have indicated that alteration of these residues leads to significant changes in activity or affinity for ATP (26), but it is possible that these changes are the consequence of conformational changes rather than direct effects at the nucleotide binding site. To further test the proximity of these aromatic residues to the ATP-binding site, we have carried out photochemical labeling studies of the wild-type and mutant forms of the A subunit using 2-azido-[ 32 P]ADP. Mutants in which all four residues were phenylalanine or in which single tyrosine residues were introduced were tested.
We also wished to determine the role of the nucleotide binding sites in controlling dissociation of the V 1 and V 0 domains. It has previously been shown that nucleotide binding activates dissociation of the V 1 and V 0 domains in vitro (30 -32). In addition, reversible dissociation of V 1 and V 0 domains has been observed in vivo in yeast (33) and insects (34). We have therefore addressed the role of nucleotide binding and activity in regulating in vivo and in vitro dissociation of the V-ATPase complex. Yeast cells were grown in yeast extract-peptone-dextrose (YEPD) medium or in supplemented synthetic dextrose medium (SD).
All mutations were confirmed by DNA sequencing using the dideoxy method. Cassettes containing each mutation were subcloned into a wild-type pAlterI/VMA1 plasmid that had not been subjected to mutagenesis using HpaI and SalI. The full-length mutant VMA1 cDNAs were then subcloned, along with wild-type VMA1 for a positive control, into the yeast integration vector YIp5 (New England Biolabs) using BamHI and SalI sites.
Transformation-Wild-type VMA1 in YIp5 (WT-VMA1) as a positive control, or vma1 mutant YIp5 plasmids were linearized with ApaI to target the integration of the constructs to the URA3 locus. Yeast cells SF838 -5A␣ vma1⌬-8 were transformed with the linearized constructs using the lithium acetate method (36). The transformants were then selected on UraϪ plates as described previously (37). Chromosomal DNA was isolated from the transformed yeast cells and the VMA1 gene was amplified by a polymerase chain reaction. The presence of the site-directed mutations was confirmed by sequencing the polymerase chain reaction products. Growth phenotypes of the mutants were assessed on YEPD plates buffered with 50 mM KH 2 PO 4 and 50 mM succinic acid to either pH 5.5 or pH 7.5. Plates containing 50 mM CaCl 2 were buffered to pH 7.5 using 50 mM MES/MOPS (38).
Purification of the Yeast V-ATPase-Yeast integrated with either wild-type plasmid (WT-VMA1) or vma1 mutant plasmids were cultured overnight in 1 liter of YEPD pH 5.5 to log phase. Vacuoles were isolated from the different strains as described previously (39). For purification of the V-ATPase, vacuolar membranes were washed three times with 10 mM Tris-HCl, pH 7.5, and 1 mM EDTA, and solubilized in buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, 10% glycerol, 0.5 mM phenylmethylsulfonyl fluoride, and 2 mM dithiothreitol) with 0.5% Zwittergent 3-14, and the V-ATPase was isolated by glycerol density gradient sedimentation on 20 -50% glycerol gradients as described previously (40).
Labeling of the V-ATPase by 2-Azido-[ 32 P]ADP-V-ATPase purified on 20 -50% glycerol gradients was concentrated 10 -15-fold using Centricon 30 concentrators (Amicon) to a final concentration of 10 -20 g of protein/100 l. Protein was incubated with 120 M 2-azido-[ 32 P]ADP and 6 mM MgCl 2 , in the presence or absence of 5 mM ATP, for 20 min at 4°C in the dark followed by irradiation of the sample with a ultraviolet lamp (9UVGL-25 Mineralight 254/366) at short wavelength at a distance of 1 cm at 4°C for 20 min. Laemmli sample buffer was added to the samples followed by SDS-PAGE on a 10% acrylamide gel. The gel was incubated in 30% methanol and 7.5% acetic acid, dried, and autoradiography performed.
Metabolic Labeling and Immunoprecipitation of the V-ATPase-Yeast strains WT-VMA1 and vma1 mutants were grown in SD-methionine-free medium overnight, converted to spheroplasts, and metabolically labeled with Tran 35 S-label (50 Ci/5 ϫ 10 6 spheroplasts) for 60 min at 30°C. At the end of the incubation, unlabeled methionine and cysteine were added to a final concentration of 0.33 mg/ml each. For the chase, spheroplasts were pelleted and resuspended in YEP media containing 1.2 M sorbitol alone, or YEP media with 1.2 M sorbitol and 2% dextrose and incubated for 30 min at 30°C. Spheroplasts were pelleted, washed, and lysed in solubilization buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 10% glycerol) with 1% C 12 E 9 , and protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, 5 g/ml aprotinin, 1 g/ml pepstatin). The V-ATPase was cross-linked using dithiobissuccinimidylpropionate and immunoprecipitated using the monoclonal antibody 8B1-F3 against the yeast A subunit (Molecular Probes, Inc.), and protein A-Sepharose (Amersham Pharmacia Biotech). Samples were subjected to SDS-PAGE on a 12% acrylamide gel, fixed in 30% methanol and 7.5% acetic acid for 1 h, incubated in Enlightening solution (NEN Life Science Products) for 30 min, dried, and autoradiography performed.
In Vitro Dissociation of the V-ATPase-Vacuolar vesicles were washed three times with 10 mM Tris-HCl, pH 7.5, and 1 mM EDTA, and resuspended in buffer (10 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 10% glycerol). 30 -50 g of washed vacuoles were then incubated with or without 5 mM ATP and 6 mM MgCl 2 for 10 min on ice. Potassium iodide was added to the vacuoles to a final concentration 300 mM, and the vesicles were incubated on ice for 60 min. Membranes were pelleted at 100,000 ϫ g for 30 min, and supernatants were precipitated with 12% trichloroacetic acid for 30 min on ice. The precipitated proteins were solubilized in 50 mM Tris-HCl, pH 6.8, 8 M urea, 5% SDS, 1 mM EDTA, 5% ␤-mercaptoethanol at 70°C for 10 min, and subjected to SDS-PAGE using a 12% acrylamide gel. Dissociation of V 1 was determined by Western blot analysis using the monoclonal antibody 8B1-F3 against the yeast V-ATPase A subunit (Molecular Probes, Inc.). Following SDS-PAGE, samples were transferred to nitrocellulose, and probed with the monoclonal antibody 8B1-F3, followed by horseradish peroxidase-conjugated secondary antibody (Bio-Rad). Immunoblots were developed using a chemiluminescent detection method following the manufacturer's protocol from Kirkegaard and Perry Laboratories.
Other Procedures-ATPase activity was measured using a coupled spectrophotometric assay (39) in the absence or presence of 1 M concanamycin A, a specific inhibitor of the V-ATPase (43). SDS-PAGE was carried out as described by Laemmli (44), and silver staining was performed by the method of Oakley et al. (45). Protein concentrations were determined by Lowry assay (46), and determinations on purified enzyme included initial protein precipitation with 10% trichloroacetic acid.

RESULTS
Kinetic Analysis of vma1 Mutants-Site-directed mutations of the VMA1 cDNA encoding the yeast V-ATPase A subunit were constructed as described under "Experimental Procedures." The mutant vma1 cDNAs were subcloned into the yeast integration vector YIp5 and expressed in a vma1⌬ strain in which the VMA1 gene was deleted. The aromatic residues suggested from molecular modeling studies to be in proximity to the adenine binding pocket at the catalytic site on the A subunit are Phe 452 , Tyr 532 , Tyr 535 , and Phe 538 . The following nomenclature was employed to simplify discussion of tyrosine and phenylalanine mutants at these positions. ϪTyr refers to the double mutant in which the two endogenous tyrosine residues have been replaced by phenylalanine (Y532F/Y535F). Y452 refers to the triple mutant (F452Y/Y532F/Y535F) containing a tyrosine at position 452 but phenylalanines at the other three positions. Y535 refers to the single mutant (Y532F) containing a tyrosine at position 535 but phenylalanines at the other three positions. Y532 refers to the single mutant (Y535F) containing a tyrosine at position 532 but phenylalanines at the other three positions. Finally, Y538 refers to the triple mutant (Y532F/Y535F/F538Y) containing a tyrosine at position 538 but phenylalanines at the other three positions. All other mutations (i.e. Y532S, Y535S, etc.) refer to constructs containing only the indicated amino acid substitution.
Deletion of genes encoding subunits of the V-ATPase results in a conditional lethal phenotype (35,47,48), which is also observed for vma mutants possessing less than approximately 20% of wild-type V-ATPase activity (24). Strains bearing these mutations are able to grow on medium buffered to acidic pH (5.0 -5.5) but are unable to grow at neutral pH (7.5) or in neutral medium containing 50 mM CaCl 2 . Of the mutants tested, three showed altered growth at neutral pH. As previously reported, E286Q and F452A were both unable to grow at neutral pH (25,26), while the mutant Y535S showed slower growth at neutral pH. In addition, based on Western blot analysis of whole cell lysates and isolated vacuoles using the monoclonal antibody 8B1-F3 against the yeast V-ATPase A subunit, all of the vma1 mutants tested showed normal levels of protein expression and V-ATPase assembly (data not shown). Table I summarizes the kinetic analysis of the vma1 mutants. Vacuoles were isolated from the wild-type WT-VMA1 and vma1 mutant strains and then solubilized and the V-ATPases purified by glycerol density gradient sedimentation as described under "Experimental Procedures." ATPase activities of the purified enzymes were measured over a range of ATP concentrations from 25 M to 2.5 mM, while the MgCl 2 concentration was maintained at 1 mM above the ATP concentration. K m and V max values were calculated from double reciprocal plots of ATP concentration versus ATPase activity. The wildtype enzyme WT-VMA1 had a specific activity of 1.8 mol/ min/mg of protein at 1 mM ATP, a K m of 0.70 mM ATP and a V max of 2.5 mol/min/mg of protein. Removal of the two tyrosines at positions 532 and 535 (ϪTyr) resulted in a 5-fold decrease in K m for ATP and a 3-fold decrease in V max . Three mutants, Y452, Y532, and Y538, showed a similar decrease in K m for ATP. By contrast, Y535 showed kinetics similar to the wild-type enzyme, suggesting that a tyrosine at position 535 is important in maintaining the normal affinity for ATP and the normal rate of turnover of the enzyme.
As previously reported for the Y532S mutant (26), the K m for ATP of the Y535S mutant was estimated to be greater than 5 mM, suggesting that aromatic residues at positions 532 and 535 are required for high affinity ATP binding. The two additional mutants used in this study, E286Q (25) and F452A (26), are both inactive yet assembled enzymes. Glu 286 corresponds to the acidic residue in the F-ATPase ␤ subunit that has been suggested to function in catalysis (49,50), while Phe 452 corresponds to Tyr 345 of the mitochondrial F-ATPase ␤ subunit, which is photolabeled by 2-azido-[ 32 P]ATP (27).

Photolabeling of Wild-type and vma1 Mutants by 2-Azido-[ 32 P]ADP-To test the proximity of the four A subunit aromatic
residues to the adenine binding pocket at the catalytic site, 2-azido-[ 32 P]ADP was used to label the wild-type enzyme and vma1 mutants in which one of the four aromatic residues was a tyrosine, as described above. The double mutant in which the two endogenous tyrosine residues were replaced by phenylalanine (ϪTyr) was used as a negative control. 2-Azido-[ 32 P]ADP rather than 2-azido-[ 32 P]ATP was employed in these studies, since in preliminary experiments reaction with 2-azido-[ 32 P]ATP resulted in increased nonspecific labeling of other subunits. Fig. 1A shows the results of ultraviolet irradiation of the purified yeast V-ATPase in the presence of 120 M 2-azido-[ 32 P]ADP and 6 mM MgCl 2 in the presence and absence of 5 mM ATP, as described under "Experimental Procedures." As can be seen, 2-azido-[ 32 P]ADP shows strong labeling of the wild-type A subunit that is blocked in the presence of ATP. By contrast, the ϪTyr mutant shows greatly reduced labeling by 2-azido-[ 32 P]ADP, with little protection by ATP. Of the four mutants containing single tyrosine residues in the pocket, Y532 showed very strong ATP-protectable labeling, Y452 and Y538 showed somewhat less but still very significant labeling with substantial ATP protection, and Y535 showed only relatively weak labeling, with little protection by ATP. It should be noted that the limited amounts of purified V-ATPase that can be isolated from yeast vacuoles makes it infeasible to purify and identify the radiolabeled peptides, as had previously been done for the bovine V-ATPase A subunit (13). Nevertheless, the very low level of ATP-protectable labeling observed for the ϪTyr mutant (despite its high affinity for ATP) makes it likely that the radiolabel is attached to the introduced tyrosine residues in the Y452, Y532, and Y538 mutants. Labeling by 2-azido-[ 32 P]ADP was also used to assess the ability of other mutants to bind nucleotides. As shown in Fig.  1B, E286Q showed even stronger ATP-protectable labeling by 2-azido-[ 32 P]ADP than the wild-type enzyme. The Y535S mutant also showed good ATP-protectable labeling, whereas both Y532S and F452A showed considerably reduced labeling. These results indicate that the E286Q mutant is competent to bind nucleotides, whereas the F452A mutant binds nucleotides only poorly.
In Vitro Dissociation of the V-ATPase-It has previously been observed that ATP activates dissociation of V 1 from V 0 in the presence of potassium iodide (31,32). To determine whether V 1 dissociation is dependent upon ATP binding or catalysis, potassium iodide-induced dissociation in the absence and presence of 5 mM MgATP was compared by Western blot analysis of supernatants for the wild-type and several vma1 mutants. As can be seen in Fig. 2, the wild-type enzyme showed good ATP-activated dissociation of V 1 . Dissociation of the E286Q mutant was also ATP-activated, indicating that activity is not required for dissociation, although somewhat more dissociation occurred in the absence of nucleotide than for the wild type, suggesting a partial destabilization of the enzyme. By contrast, the F452A mutant showed no ATP-activated dissociation of V 1 . Because the F452A mutant bound nucleotides poorly (Fig. 1B), this suggests that nucleotide binding to the catalytic site is required for in vitro dissociation. In support of this, Y535S shows better ATP-activated dissociation of V 1 (Fig.   2) and better ATP-protectable labeling by 2-azido-[ 32 P]ADP than Y532S (Fig. 1B).
In Vivo Dissociation of the V-ATPase-Dissociation of the V 1 and V 0 domains has also been shown to occur in vivo in response to glucose deprivation (33), and to require enzyme activity (51). It is not known, however, whether nucleotide binding (without turnover) is sufficient to activate dissociation in vivo. To address this question, dissociation of the V 1 and V 0 domains in response to glucose deprivation was compared for the wild-type and four of the A subunit mutants. Dissociation of the V 1 and V 0 domains is reflected as a decrease in the amount of the V 0 subunits (particularly the 100-kDa a subunit) immunoprecipitated using an anti-V 1 antibody. As can be seen in Fig.  3, glucose deprivation results in partial dissociation of the V-ATPase from wild-type cells and, to a lesser extent, from the Y532S and Y535S mutants. By contrast, no appreciable dissociation is observed for either the E286Q or F452A mutants, both of which are catalytically inactive. Because the E286Q mutant is still able to bind nucleotides (Fig. 1B), these results suggest that nucleotide binding to the catalytic site is not sufficient to activate in vivo dissociation. DISCUSSION It has previously been shown that mutation of a number of aromatic residues postulated to be present at the catalytic site of the V-ATPases, including the A subunit residues Phe 452 , Tyr 532 , and Phe 538 , have significant effects on V-ATPase activity or affinity for ATP (26). Thus, substitution of alanine at position 452 completely eliminates ATP hydrolytic activity while substitution of serine at position 532 causes a greater than 7-fold increase in K m for ATP (26). In the present study it was shown that replacement of Tyr 535 by serine also causes a significant increase in K m for ATP. These results suggested the involvement of these residues in nucleotide binding at the catalytic site, but it is possible that the effects observed on activity and affinity are due to conformational changes induced by the mutations introduced.
To attempt to determine the proximity of these aromatic residues to the adenine ring of the bound nucleotide, photochemical labeling by 2-azido-[ 32 P]ADP was employed. Because 2-azido-adenine nucleotides show selectivity for reaction with tyrosine residues but not phenylalanine residues (29), we compared the reaction of mutant forms of the A subunit containing substitutions at these four aromatic residues. Significant labeling by 2-azido-[ 32 P]ADP was observed for tyrosines at positions 452, 532, and 538, with little labeling observed at position 535 or for the mutant containing four phenylalanine residues. The labeling observed for this latter mutant may be due to reaction  Table I for the definition of the mutant strains.
FIG. 2. In vitro dissociation of V 1 and V 0 following treatment with potassium iodide and MgATP. Vacuolar membranes (30 -40 g of protein) from the wild-type VMA1 and vma1 mutant strains indicated were incubated in the absence (lanes 1, 3, 5, 7, and 9) or presence (lanes 2, 4, 6, 8, and 10) of 5 mM ATP and 6 mM MgCl 2 for 10 min on ice. Potassium iodide was then added to the vacuoles at a final concentration of 300 mM, and the vesicles were then incubated on ice for 60 min. Membranes were pelleted at 100,000 ϫ g for 30 min, and the supernatants were precipitated by addition of 12% trichloroacetic acid and incubation for 30 min on ice. Following sedimentation at 10,000 ϫ g and removal of the supernatant, the precipitated proteins were solubilized and subjected to SDS-PAGE on a 12% acrylamide gel, as described under "Experimental Procedures." Proteins were then transferred to nitrocellulose and Western blot analysis performed using the monoclonal antibody 8B1-F3 against the yeast V-ATPase A subunit. of 2-azido-[ 32 P]ADP with other residues at the catalytic (or noncatalytic) sites. The absence of reaction of Tyr 535 with 2-azido-[ 32 P]ADP should not be taken as proof that this residue is not present at the catalytic site, since it is possible that the reactive nitrene is not oriented in such a way as to readily react with this residue. The stronger labeling observed for the Tyr 532 mutant than for the wild-type enzyme (which contains tyrosine residues at both positions 532 and 535) may be due to the higher affinity of the Y532 mutant for nucleotides.
The results described above suggest that at least residues Phe 452 , Tyr 532 , and Phe 538 are near the adenine ring of bound 2-azido-[ 32 P]ADP. In support of this, previous labeling studies of the bovine V-ATPase have revealed that the A subunit is modified by 2-azido-[ 32 P]ATP within a 12-kDa V8 fragment, which begins at residue 511 and which includes Tyr 525 and Tyr 528 (13). These residues correspond to Tyr 532 and Tyr 535 , respectively, in the yeast V-ATPase A subunit. In the mitochondrial F-ATPase ␤ subunit, the A subunit residues Phe 452 , Tyr 532 , and Tyr 538 correspond to Tyr 345 , Phe 418 , and Phe 424 , respectively (17,47,52). The x-ray crystal structures of F 1 show all three of these residues to be located near the adenine binding pocket of the ␤ subunit, with Tyr 345 the most nearly coplanar with the adenine ring (21,22). Tyr 345 is also the residue labeled by 2-azido-[ 32 P]ATP bound to the catalytic site of mitochondrial F 1 (27). Tyr 535 in the V-ATPase A subunit corresponds to Ala 421 in the mitochondrial ␤ subunit, but modeling studies have predicted that Tyr 535 may also be stacked with the adenine ring. Nevertheless, it is possible that the essential feature of these residues is their hydrophobic character rather than their aromaticity. Additional mutations substituting hydrophobic but nonaromatic residues at these positions may be able to resolve this question.
In addition to investigating the structure of the catalytic site of the V-ATPases, we also wished to determine the role of nucleotide binding to this site in controlling dissociation of the V 1 and V 0 domains, both in vivo and in vitro. Dissociation of the V-ATPase complex by chaotropic agents such as potassium iodide has been shown to be activated by ATP (30 -32), and the affinity of this site on the bovine-coated vesicle enzyme for ATP has been estimated to be approximately 200 nM (31). While this is higher in affinity than the kinetically measured K m for ATP (80 and 800 M for the bovine enzyme; Ref. 31), it is possible that this higher affinity represents ATP binding to a site associated with "uni-site" catalysis (for review, see Ref. 53). It is also unclear whether catalytic turnover of the enzyme is required for dissociation of V 1 and V 0 in vitro. In the present study, we observe that ATP is able to activate dissociation of V 1 and V 0 for a completely inactive mutant (E286Q), establishing that catalysis is not required for in vitro dissociation. On the other hand, the F452A mutant, which binds nucleotides only poorly, based upon labeling by 2-azido-[ 32 P]ADP, shows no ATP-activated dissociation of V 1 and V 0 at all. This suggests that nucleotide binding to the catalytic site (but not hydrolysis) is required to activate dissociation in vitro. This idea is supported by the observations with the Y532S and Y535S mutants.
In addition to their role in in vitro dissociation, we also wished to investigate the role of the catalytic nucleotide binding sites in in vivo dissociation of the V-ATPase complex. Reversible dissociation of the V-ATPase has been observed to occur in yeast in response to glucose deprivation (33), and has been suggested to play a role in regulating V-ATPase activity in both yeast and insect cells (33,34). Measurement of intracellular ATP levels in yeast indicate a 50% drop within 1 min of glucose deprivation, consistent with the observed onset of dissociation of V 1 and V 0 , but ATP levels recover to more than 80% of the initial level after 20 min of glucose deprivation, despite the fact that dissociation of V 1 and V 0 is maximal (51). These results argue that changes in the intracellular ATP concentration are probably not the primary signal responsible for controlling the assembly state of the V-ATPase, but do not rule out the possible involvement of the nucleotide binding sites on the V-ATPase in this process. Parra and Kane (51) have demonstrated that V-ATPase complexes inactivated by mutations in the V 0 domain show reduced dissociation in vivo in response to glucose deprivation, but the nucleotide binding properties of these enzymes have not been investigated. In the present study we observe that the catalytically inactive mutants E286Q and F452A show no dissociation of V 1 and V 0 in response to glucose deprivation, despite the fact that E286Q appears to be competent to bind nucleotides, based upon 2-azido-[ 32 P]ADP labeling. These results suggest that in vivo dissociation of V 1 and V 0 , unlike in vitro dissociation, is primarily dependent upon catalytic activity of the V-ATPase complex rather than on nucleotide binding. Interestingly, both the Y532S and Y535S mutants show in vivo dissociation in response to glucose deprivation. Because both of these mutants have a K m for ATP greater than 5 mM (as compared with a K m of 0.7 mM for the wild-type enzyme), it is likely that under intracellular conditions both of these enzymes will be turning over more slowly than the wildtype enzyme. Nevertheless, even under these conditions, in vivo dissociation is observed, suggesting that a very high rate of turnover may not be required for dissociation. It is possible that, in order for the enzyme to dissociate in vivo, it must pass through a conformational state which is only achieved during catalysis. The demonstration that the homologous F-ATPases undergo rotational motion during the catalytic cycle (54 -56) suggests that some intermediate state reached during rotation of the complex may be required for dissociation. FIG. 3. In vivo dissociation of the V-ATPase complex following glucose deprivation. The VMA1 wild-type strain and vma1 mutant strains indicated were grown in methionine-free medium overnight, converted to spheroplasts, and metabolically labeled with Tran 35 S-label (50 Ci/5 ϫ 10 6 spheroplasts) for 60 min at 30°C. Labeled spheroplasts were then chased for 30 min in YEP medium containing 2% glucose (lanes 1, 3, 5, 7, and 9) or YEP medium without glucose (lanes 2, 4, 6, 8,  and 10). The V-ATPase was then solubilized with detergent and immunoprecipitated using the monoclonal antibody 8B1-F3 against the A subunit as described under "Experimental Procedures." Samples were then subjected to SDS-PAGE on a 12% acrylamide gel and autoradiography was performed as described. Dissociation of V 1 and V 0 appears as a reduction in the intensity of the V 0 subunits (particularly the 100-kDa subunit) immunoprecipitated using the anti-V 1 subunit antibody.