Mutational Analysis of the Nucleotide Binding Sites of the Yeast Vacuolar Proton-translocating ATPase*

To further define the structure of the nucleotide binding sites on the vacuolar proton-translocating ATPase (V-ATPase), the role of aromatic residues at the catalytic sites was probed using site-directed mutagenesis of the VMA1 gene that encodes the A subunit in yeast. Substitutions were made at three positions (Phe452, Tyr532, and Phe538) that correspond to residues observed in the crystal structure of the homologous β subunit of the bovine mitochondrial F-ATPase to be in proximity to the adenine ring of bound ATP. Although conservative substitutions at these positions had relatively little effect on V-ATPase activity, replacement with nonaromatic residues (such as alanine or serine) caused either a complete loss of activity (F452A) or a decrease in the affinity for ATP (Y532S and F538A). The F452A mutation also appeared to reduce stability of the V-ATPase complex. These results suggest that aromatic or hydrophobic residues at these positions are essential to maintain activity and/or high affinity binding to the catalytic sites of the V-ATPase. Site-directed mutations were also made at residues (Phe479and Arg483) that are postulated to be contributed by the A subunit to the noncatalytic nucleotide binding sites. Generally, substitutions at these positions led to decreases in activity ranging from 30 to 70% relative to wild type as well as modest decreases inK m for ATP. Interestingly, the R483E and R483Q mutants showed a time-dependent increase in ATPase activity following addition of ATP, suggesting that events at the noncatalytic sites may modulate the catalytic activity of the enzyme.

The vacuolar proton-translocating ATPases (or V-AT-Pases) 1 are a class of enzymes that couple the hydrolysis of ATP to the transmembrane movement of protons. This proton movement results in the generation of a pH gradient and the acidification of intracellular compartments (1)(2)(3)(4)(5)(6)(7)(8)(9) or, for certain specialized cells, the extracellular space (10 -12). The V-ATPases therefore play an essential role in many basic cellular functions which require acidification, including pro-tein processing and degradation, receptor-mediated endocytosis, intracellular membrane trafficking, and coupled transport (for review, see Refs. [1][2][3][4][5][6][7][8][9]. The V-ATPases are composed of two domains, a 500-kDa peripheral V 1 domain with the structure A 3 B 3 C 1 D 1 E 1 F 1 G 1 H 1 , which is responsible for ATP hydrolysis, and a 250-kDa V 0 domain with the structure a 1 d 1 cЉ 1 (c,cЈ) 6 , that is responsible for proton translocation (1,13). In Saccharomyces cerevisiae, the V-ATPase subunits are encoded by at least 14 genes, including VMA1, which encodes the 69-kDa A subunit (14,15), and VMA2, encoding the 60-kDa B subunit (16). Sequence homology indicates that the A and B subunits of the V-ATPases (14 -22) are related to the ␤ and ␣ subunits, respectively, of the F-type ATPases (F-ATPases), which function in mitochondria, chloroplasts, and bacteria to synthesize ATP (23)(24)(25)(26)(27)(28). The level of sequence identity between these proteins is only 20 -25%, however, and the A subunit contains a large (100-amino acid) insertion not present in ␤, suggesting that significant structural and functional differences may exist between these proteins.
From the x-ray crystal structure of the F 1 domain from bovine heart mitochondria it was confirmed that the nucleotide binding sites are at the interface between the ␤ and ␣ subunits, with the catalytic site primarily on ␤ and the noncatalytic site primarily on ␣ (29). Previous biochemical experiments on the V-ATPase suggest that the catalytic nucleotide binding site is located on the A subunit while the noncatalytic site resides on the B subunit. For the catalytic site, these data include ATP-protectable labeling of the A subunit by reagents such as N-ethylmaleimide and 7-chloro-4-nitrobenz-2-oxa-1,3-diazole, which correlates with inhibition of activity (for references, see Forgac (30)). In addition, modification of a single A subunit cysteine residue (Cys 254 of the bovine A subunit) by N-ethylmaleimide, cystine, or by disulfide bond formation with Cys 532 , leads to inactivation of the enzyme (31)(32)(33). Also, labeling of the A subunit by the photoaffinity ATP analog 2-[azido-32 P]ATP correlates well with inhibition by this reagent (34). Finally, the A subunit, unlike the B subunit, contains several consensus sequences that appear to be critical for ATP hydrolysis in the homologous ␤ subunit (35)(36)(37), including the glycine-rich loop.
There is limited information concerning the structure of the nucleotide binding sites on the V-ATPase, although mutagenesis studies in yeast have begun to identify residues important for V-ATPase activity (38 -40). To investigate residues involved in formation of the nucleotide binding sites, site-directed mutagenesis of the VMA1 gene in yeast was performed. Based on the crystal structure of the F 1 ␤ subunit and sequence alignment of the ␤ and A subunits, mutations were made in several residues proposed to form part of the adenine binding pocket at the catalytic site, including Phe 452 , Tyr 532 , and Phe 538 (corresponding to Tyr 345 , Phe 418 , and Phe 424 , respectively, of the bovine mitochondrial ␤ subunit) (29). In addition, mutations were constructed at two residues, Phe 479 and Arg 483 (corresponding to Tyr 368 and Arg 372 of the bovine F 1 ␤) (29), which are postulated to be contributed by the A subunit to the noncatalytic nucleotide binding sites. The effects of these mutations on activity, kinetics and assembly of the V-ATPase complex were assessed.

EXPERIMENTAL PROCEDURES
Materials and Strains-Zymolyase 100T was obtained from Sekagaku America, Inc. Concanamycin A was obtained from Fluka Chemical Corp. Tran 35 S-label was purchased from ICN Biomedicals. Leupeptin, aprotinin, and pepstatin were all purchased from Boehringer Mannheim. Yeast extract, dextrose, peptone, and yeast nitrogen base were purchased from Difco. Zwittergent 3-14 was purchased from Calbiochem-Novabiochem Corp. Molecular biology reagents were from Promega and New England Biolabs. ATP and most other chemicals were purchased from Sigma.
Mutagenesis-Mutagenesis was performed on the wild type VMA1 cDNA using the Altered Sites II in vitro mutagenesis system (Promega) following the manufacturer's protocol. The full-length VMA1 cDNA lacking the endonuclease spacer region (15) was cloned into pAlter-1 using BamHI and SalI sites. The mutagenesis oligonucleotides were as follows with the substitution sites underlined: All mutations were confirmed by DNA sequencing using the dideoxy method. Cassettes containing each mutation were subcloned using HpaI and SalI into a wild type pAlterI/VMA1 plasmid that had not been subjected to mutagenesis. 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, YIp5 vector alone (vma1⌬/YIp5) as a negative 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 (41). The transformants were then selected on Ura Ϫ plates as described previously (42). 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 7.5. Plates containing 50 mM CaCl 2 were buffered to pH 7.5 using 50 mM MES/MOPS (43).
Purification of the V-ATPase Enzyme-Yeast integrated with either wild type plasmid (WT-VMA1), vma1 mutant plasmids, or the vector YIp5 alone (vma1⌬/YIp5) as a negative control were cultured overnight in 1 liter of YEPD pH 5.5 to log phase. Vacuoles were isolated as described previously (44). 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 Zwittergent 3-14, and the V-ATPase was isolated by glycerol density gradient sedimentation on 20 -50% glycerol gradients as described previously (45).
Modeling of the Catalytic Nucleotide Binding Site on the V-ATPase A Subunit-A model for the V-ATPase was created from the 2.8 Å resolution structure of bovine heart mitochondrial F 1 -ATPase (29). The amino acid sequences were first aligned using the Genetics Computer Group (GCG) sequence analysis software package. Approximately 23% sequence identity was obtained after sequence alignment, with good agreement in predicted secondary structure (47). The positions of insertions and deletions created by the alignment procedure were examined on the model of F 1 -ATPase using InsightII (Biosym Technologies, San Diego, CA), and were found to be mostly in loops at the surface of the molecule. A few were located within secondary structure elements and the sequence alignment was adjusted to move these by a few residues to the end of the structural elements. The large insertion of about 100 amino acids present in the A subunit (but not ␤) was not modeled. The resultant structure now containing the amino acid sequence of V-ATPase was subjected to 200 steps of energy minimization using X-PLOR (48).
Other Procedures-ATPase activity was measured using a coupled spectrophotometric assay (44) in the absence or presence of 1 M concanamycin A, a specific V-ATPase inhibitor (49). ATPase assays performed on vacuoles or on purified enzyme contained 1 mM ATP, 2 mM MgCl 2 and were done at 37°C. SDS-PAGE was carried out as described by Laemmli (50). Protein concentrations were determined by Lowry et al. (51) assay, and determinations on purified enzyme included initial protein precipitation with 10% trichloroacetic acid.

RESULTS
Growth Phenotypes for vma1 Mutants-Site-directed mutations of the VMA1 cDNA encoding the yeast V-ATPase A subunit were constructed as described under "Experimental Procedures." Three of the residues mutated, Phe 452 , Tyr 532 , and Phe 538 , are located in the C-terminal domain of the A subunit and are candidates for contributing to the adenine binding pocket of the catalytic nucleotide binding site. Two additional A subunit residues, Phe 479 and Arg 483 , are postulated on the basis of sequence alignment and the crystal structure of F 1 to form part of the noncatalytic nucleotide binding site located on the B subunit. 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 deletion of genes encoding subunits of the V-ATPase results in a conditional lethal phenotype (14,15,52), which is also used to screen for mutants defective in vacuolar acidification. These strains 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 . Table I summarizes the growth phenotypes for these vma1 mutants. The wild type strain WT-VMA1 grew under each condition and the negative control vma1⌬/YIp5 strain exhibited the expected conditional lethal phenotype. F452A was unable to grow at pH 7.5 in the absence or presence of elevated extracellular calcium, exhibiting a growth phenotype similar to the vma1⌬ strain. F538W was unable to grow at pH 7.5 in the presence of high calcium. The remaining mutants, including those at the noncatalytic site, showed relatively normal growth at pH 7.5. It has previously been determined that 20% of V-ATPase activity is sufficient to rescue the growth phenotype (39). These results thus suggest that the F452A and F538W mutants are deficient in vacuolar acidification, but that the remainder of the mutant strains have a V-ATPase activity that is at least 20% of wild-type. It was next necessary to determine whether these mutations directly caused a loss of V-ATPase activity.
ATPase Activities in Purified Vacuoles from vma1 Mutants- Fig. 1 shows the V-ATPase activities of the wild type WT-VMA1, vma1⌬/YIp5, and vma1 mutant strains. Fig. 1A shows the relative ATPase activities for the strains containing mutations at the catalytic site, and Fig. 1B the activities for the strains with the proposed noncatalytic site mutations. V-ATPase activities were measured in isolated vacuoles using a coupled spectrophotometric assay at 1 mM ATP and 37°C as described under "Experimental Procedures." Activities are expressed relative to the WT-VMA1 strain, defined as 100%, which had a specific activity of 0.19 (Ϯ 0.04) mol of ATP/ min/mg protein. 2 The negative control, vma1⌬/YIp5 strain, had no measurable activity. F452Y, Y532F, F538Y, and F538A had nearly wild type levels of activity, whereas F452W and Y532S had approximately 50% and F538W had 20% of wild type levels. F452A had less than 5% of wild type activity and was comparable to the deletion strain. Mutations at the noncatalytic site all appear to reduce activity by 30 -70% as shown in Fig. 1B. These data are consistent with the growth phenotypes.
A Subunit Expression and V-ATPase Assembly in vma1 Mutants-It was next necessary to determine whether the observed decrease in activity was due to decreased protein levels or to decreased V-ATPase assembly. Fig. 2 shows the effects of the mutations on A subunit stability and the presence of the A subunit on the vacuolar membrane, which provides an initial measure of V-ATPase assembly. Fig. 2 is an immunoblot using the monoclonal antibody 8B1-F3 against the yeast A subunit to determine levels of the A subunit in either whole cell lysates (CELL) or isolated vacuoles (VAC). The top panel shows the immunoblots for the wild type WT-VMA1 and for vma1⌬/YIp5. The middle and bottom panels show the immunoblots for the strains containing mutations at the catalytic site, and the proposed noncatalytic site, respectively. Of the mutations tested, only R483E showed somewhat lower levels of A subunit 2 The specific activity reported is that for unwashed vacuoles at 1 mM ATP. Higher specific activities are obtained for washed vacuoles at saturating ATP concentrations.  1. Effect of vma1 mutations on concanamycin A-sensitive ATPase activity in isolated yeast vacuoles. ATPase activity was measured on aliquots of purified vacuoles (10 g of protein) as described under "Experimental Procedures." Activities are expressed relative to that of vacuoles isolated from the WT-VMA1 strain (defined as 100%), which had a specific activity of 0.19 mol of ATP/min/mg of protein at 1 mM ATP (see footnote 2). No measurable ATPase activity was observed in vacuoles isolated from the negative control, vma1⌬/YIp5, containing the vector alone. Each bar represents the average of two or three determinations made on two or three independent vacuolar membrane preparations. A shows the relative ATPase activities for the strains containing mutations at the catalytic site, and B the activities for the strains with mutations at the noncatalytic site.
in the whole cell lysate compared with the wild type. However, all of the mutated proteins, including R483E, showed normal levels of A subunit on the vacuolar membrane.
As a further test of assembly, the V-ATPase was immunoprecipitated from metabolically labeled cells. Cells were converted to spheroplasts and labeled with Tran 35 S-label for 60 min at 30°C followed by cell lysis, detergent solubilization, and immunoprecipitation using the monoclonal antibody 8B1-F3 as described under "Experimental Procedures." As can be seen in Fig. 3A, most of the mutants at the catalytic site showed normal assembly relative to wild type. An exception is the mutant F452A, which shows very little intact V-ATPase complex by immunoprecipitation. Because this mutant showed normal levels of A subunit on the vacuolar membrane (Fig. 2) and normal assembly following detergent solubilization and density gradient sedimentation (data not shown), it appears that this mutation has resulted in a destabilization of the complex such that it can no longer survive immunoprecipitation. Fig. 3B shows immunoprecipitation of the V-ATPase from the noncatalytic site mutants. As can be seen, none of the mutations dramatically reduces assembly, although F479W and R483E show somewhat lower labeling in the B subunit region. The possible significance of these results is discussed below.
Kinetic Analysis of vma1 Mutants-To determine if any of the mutations resulted in changes in K m ATP or V max for ATP hydrolysis, kinetic analysis on the purified enzymes was performed (Table II). Vacuoles were isolated from wild type WT-VMA1 and vma1 mutant strains and the V-ATPases were solubilized and 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 100 M to 2.5 mM ATP, while the MgCl 2 was maintained at 1 mM above the ATP concentration.  Table II, the K m and V max for F452Y and F538Y were not significantly different from wild type, whereas F452W, F538W, and Y532F had a decreased K m , suggesting an increase in substrate affinity. k cat /K m values for the former two mutations were reduced by about 2-fold, whereas for the Y532F mutation, a small increase in k cat /K m was observed. F538A had a K m value approximately 2 times that of WT-VMA1, suggesting a small decrease in ATP affinity.

FIG. 2. Effect of vma1 mutations on A subunit stability and A subunit levels in vacuolar membranes. Whole cell lysates (CELL)
and purified vacuoles (VAC) (5 g of protein) were prepared from WT-VMA1, vma1⌬/YIp5, and vma1 mutant strains, subjected to SDS-PAGE on a 12% acrylamide gel, transferred to nitrocellulose, and immunoblotted using the monoclonal antibody 8B1-F3 against the yeast A subunit, as described under "Experimental Procedures." The top panel shows the immunoblots for the wild type WT-VMA1 and for vma1⌬/ YIp5. The middle panel and bottom panel show the immunoblots for the strains containing mutations at the catalytic site, and the proposed noncatalytic site, respectively.

FIG. 3. Effect of vma1 mutations on V-ATPase assembly as assessed by immunoprecipitation from 35 S-labeled cells.
Yeast strains WT-VMA1, vma1⌬/YIp5, and vma1 mutant strains 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. The V-ATPase was solubilized using C 12 E 9 and immunoprecipitated using the monoclonal antibody 8B1-F3 against the yeast A subunit and protein A-Sepharose, and samples were subjected to SDS-PAGE on a 12% acrylamide gel and autoradiography performed as described under "Experimental Procedures." A shows the immunoprecipitations for the strains containing mutations at the catalytic site, and B the immunoprecipitations for the strains with mutations at the noncatalytic site.
Because the reaction rate for the Y532S mutant showed no sign of saturating, even at the highest ATP concentrations tested (2.5 mM), precise values for K m and V max could not be determined (data not shown). However, we estimate that the K m for ATP is greater than 5 mM, thus reflecting a significant decrease in affinity relative to the wild type. Some of the mutations proposed to be at the noncatalytic site also resulted in kinetic changes. F479W, F479A, and R483K all showed significant decreases in both K m and V max , although values of k cat /K m were either unchanged (F479A) or reduced by at most 2-fold. Interestingly, both R483Q and R483E exhibited a time-dependent increase in ATPase activity, as depicted for the R483Q mutant in Fig. 4. DISCUSSION Although mutagenesis has been employed in several studies to identify A subunit residues important for activity of the V-ATPases (39,40,53), there is currently no information concerning the residues that may participate in formation of the adenine binding pocket on the A subunit. In the F-ATPase ␤ subunit, the adenine ring appears to be in close contact with a hydrophobic surface formed by aromatic rings, in particular Tyr 345 of the bovine mitochondrial enzyme (29). To determine whether aromatic rings are important in this region of the A subunit, site-directed mutagenesis was performed on Phe 452 , Tyr 532 , and Phe 538 , located in the C-terminal domain of the A subunit. Conservative substitutions (such as Phe to Tyr and Tyr to Phe) as well as nonconservative substitutions (such as Phe to Ala or Tyr to Ser) were carried out. Substitution with a bulkier amino acid (Trp) was also performed.
From the data presented, it appears that an aromatic ring or a hydrophobic residue is important at each of these positions, particularly Phe 452 . Of the catalytic site mutations tested, none significantly reduced stability of the A subunit and only one (F452A) reduced stability of the V-ATPase complex as evidenced by the inability to immunoprecipitate the intact complex from metabolically labeled cells. With respect to their effects on activity of the V-ATPase complex, the conservative mutations F452Y, Y532F, and F538Y had little effect on enzyme activity whereas the F452W mutation reduced activity by 50 -70%, although this mutation also resulted in a lower value for K m , suggesting an increased affinity for ATP. Interestingly, the F452A mutant, which could be isolated as an intact complex from vacuoles, was completely inactive, suggesting that an aromatic or sufficiently large hydrophobic residue at this position is crucial. Phe 452 corresponds to Tyr 345 of the mitochondrial ␤ subunit, which appears from the crystal structure of F 1 to be in direct contact with the adenine ring (29). Tyr 345 of the mitochondrial ␤ subunit is labeled by the photoaffinity analog 2-[azido-32 P]ATP (54) and substitution of tryptophan for Tyr 331 (which corresponds to Tyr 345 in the Escherichia coli ␤ subunit) results in a reduction in both K m and V max by about 2-fold (55), whereas substitution with alanine reduces V max by 5-fold and increases K m by 9-fold (56). It should be noted that substitution of leucine for Tyr 331 in the E. coli F-ATPase results in a 7.5-fold increase in K m for ATP with little effect on V max (56), suggesting that a hydrophobic residue can partially substitute for the aromatic function at this position. Our data suggest that Phe 452 plays a comparably important role in nucleotide binding to the catalytic site on the V-ATPase A subunit. The inability to immunoprecipitate the intact V-ATPase from cells bearing the F452A mutation suggests that this substitution may also lead to a reduced stability of the V-ATPase complex.
Mutations at Tyr 532 and Phe 538 had rather different effects on activity. In particular, Y532S appeared to show a significantly reduced affinity for ATP, again consistent with a role of  Time course of ATP hydrolysis measured using a coupled spectrophotometric assay on the purified enzyme isolated from the wild type WT-VMA1 (a) and from the mutant R483Q (b) as described under "Experimental Procedures." The y axis shows absorbance at 340 nm, which descreases as NADH is converted to NAD in the coupled assay. ATPase activities were measured at 2.5 mM ATP and 3.5 mM MgCl 2 . this residue in ATP binding. This residue corresponds to Phe 418 of the bovine mitochondrial ␤ subunit which is situated near the end of the adenine binding pocket (29). Replacement of Phe 538 with alanine also significantly reduced the affinity for ATP (approximately 2-fold), consistent with its interaction with the adenine ring. The corresponding residue in the F 1 ␤ subunit, Phe 424 , appears to interact with the adenine ring from the opposite side of Tyr 345 (29). It should be noted that the effects of mutations in these ␤ subunit residues (Phe 418 and Phe 424 ) on nucleotide binding to the F-ATPases have not been reported. These results support a model in which aromatic or hydrophobic residues in the C-terminal domain of the A subunit contribute to nucleotide binding to the catalytic site of the V-ATPase, with their absence resulting in substantial changes in either nucleotide binding or activity. These results also point up the likely similarity in structure of the catalytic nucleotide binding sites of the V and F-ATPases.
Using the recently released coordinates of the bovine heart mitochondrial F 1 ATPase (29) and sequence alignment of the F-ATPase ␤ subunit and V-ATPase A subunit, we have carried out energy minimization to arrive at a tentative model of the catalytic nucleotide binding site on the V-ATPase (Fig. 5). Shown are the aromatic residues investigated in the current study (Phe 452 , Tyr 532 , and Phe 538 ) as well as A subunit residues previously characterized by site-directed mutagenesis (40), including Glu 286 , Lys 263 , Cys 261 , and Cys 539 . As for the ␤ subunit, both Glu 286 and Lys 263 are in close proximity to the triphosphates of bound ATP, consistent with their proposed role in catalysis (29). Cys 261 and Cys 539 are predicted to be approximately 13 Å apart, suggesting that some distortion of the nucleotide binding site may need to occur in order for these two residues to undergo disulfide bond formation (33). All three aromatic residues are predicted to be in relatively close proximity to the adenine binding pocket, with Phe 452 nearly stacked with the adenine ring. This is consistent with the mutagenesis data, where the largest effects on activity were observed with the F452A mutation.
We have previously observed that mutations in B subunit residues postulated to be contributed to the catalytic site have dramatic effects on activity (38), suggesting that, as with the F-ATPases, the catalytic sites of the V-ATPases are located at the interface between the two nucleotide binding subunits. We also wished to determine the effects of mutations in A subunit residues postulated to be contributed to the noncatalytic sites on the B subunit. These include Phe 479 (corresponding to Tyr 368 of the bovine F 1 ␤ subunit) and Arg 483 (corresponding to Arg 372 of F 1 ␤). This is particularly important as the function of the noncatalytic nucleotide binding sites remains uncertain.
Interestingly, all the mutations tested (even the conservative F479Y and R483K mutations) decreased enzyme activity in purified vacuoles by 30 -70% and reduced k cat /K m by at most 2-fold. These data are consistent with our previous findings that B subunit mutations at the noncatalytic site also resulted in modest (30 -60%) decreases in activity (38). These results suggest that changes in the noncatalytic site produce an enzyme that is not substantially altered in its catalytic efficiency.
Although several of the mutations (including F479W and R483E) showed slightly reduced assembly as assessed by immunoprecipitation, none of them completely abolished assembly. On the other hand, both of these mutations showed somewhat reduced labeling of the 50-kDa region of the immunoprecipitates, which might reflect partial loss of the VMA13 gene product. Vma13p (a 54-kDa V1 subunit) is required for activity but not assembly of the V-ATPase complex (57). Among the most interesting mutations identified were R483E and R483Q, both of which resulted in a time-dependent increase in enzyme activity. One possible explanation for this effect is the loss during purification of the mutant enzymes of an endogenous nucleotide (possibly at the noncatalytic site) required for activity. This endogenous nucleotide would then be restored on addition to the assay. Alternatively, these mutations may result in slowed release of an inhibitory nucleotide which, in the wild type, is released rapidly. Sulfite has been suggested to activate the yeast V-ATPase by promoting the release of inhibitory ADP (58). Nevertheless, the effects of these mutations suggest that changes in the noncatalytic nucleotide binding sites can modulate the catalytic activity of the V-ATPase.
We have provided preliminary evidence for the role of A subunit residues Phe 452 , Tyr 532 , and Phe 538 in formation of the catalytic nucleotide binding site of the V-ATPase, possibly through contribution to the adenine binding pocket. We have also demonstrated that mutations in residues on the A subunit, Phe 479 and Arg 483 , that are postulated to form part of the noncatalytic site on the B subunit, can have effects on V-ATPase activity. Further work will be required to more completely define the function of these sites in the control of activity.
Acknowledgment-We thank Dr. Patricia Kane, Department of Biochemistry and Molecular Biology, SUNY, Syracuse, for the generous gift of the yeast strain SF838-5A␣ vma1⌬-8, the plasmid pPK17-7, and for helpful discussions.
FIG. 5. Proposed model for the catalytic nucleotide binding site of the yeast V-ATPase. Shown is an energy minimized model of the structure of the catalytic nucleotide binding site on the yeast V-ATPase A subunit based upon the x-ray crystal structure of the bovine mitochondrial F-ATPase ␤ subunit (29) and sequence alignment of the A and ␤ subunits (see "Experimental Procedures"). Shown are the aromatic residues mutated in the current study (Phe 452 , Tyr 532 and Phe 538 ). Also shown are two cysteine residues (Cys 261 and Cys 539 ) previously shown to form an inhibitory disulfide bond (33) as well as two residues (Lys 263 and Glu 286 ) shown by previous mutagenesis studies (39,40) to be important for activity. The bound AMP-PNP molecule is shown in dark shading; the bound Mg 2ϩ is shown in the lightly shaded sphere, and the water molecule involved in hydrolysis is shown in the darkly shaded sphere.