Isolation of vacuolar membrane H(+)-ATPase-deficient yeast mutants; the VMA5 and VMA4 genes are essential for assembly and activity of the vacuolar H(+)-ATPase.

The vacuolar membrane H(+)-ATPase of the yeast Saccharomyces cerevisiae is a multisubunit enzyme complex composed of an integral membrane V0 sector, and a peripherally associated V1 sector. Deletion of one of several structural genes for vacuolar H(+)-ATPase subunits was previously demonstrated to prevent proper assembly of the remaining V1 subunits onto the vacuolar membrane (Kane, P.M., Kuehn, M.C., Howald-Stevenson, I., and Stevens, T.H. (1992) J. Biol. Chem. 267, 447-454). A genetic screen was designed to identify new genes whose products were essential for the synthesis, assembly, and/or function of the yeast vacuolar H(+)-ATPase. Mutants were identified based on phenotypes associated with vacuolar membrane H(+)-ATPase loss of function (vma), including an inability to grow on media buffered at neutral pH. Representatives in five complementation groups were identified, including four novel mutant vma5, vma21, vma22, and vma23, all of which were defective in vacuolar ATPase enzyme activity. We report here the characterization of two genes, VMA4 and VMA5, that encode peripheral subunits of the vacuolar H(+)-ATPase. We determined that VMA5 encodes the 42-kDa subunit of the vacuolar H(+)-ATPase. The VMA4 gene, originally described by Foury (Foury, F. (1990) J. Biol. Chem. 265, 18554-18560), was determined to encode the 27-kDa subunit of the purified yeast vacuolar H(+)-ATPase. Characterization of the vma5 and vma4 mutants revealed that the 42- and 27-kDa subunits are essential for the assembly of the peripheral membrane portion of the H(+)-ATPase onto the vacuolar membrane.

zyme that is responsible for the acidification of the yeast vacuole. The yeast enzyme has been reported to be composed of at least eight polypeptides, including a loo-, 69-, 60-, 42-, 36-, 32-, 27-, and 17-kDa species (Kane et al., 1989). Several of these polypeptides, including the 69-, 60-, and 42-kDa species appear to associate peripherally with the enzyme on the cytoplasmic face of the vacuolar membrane. These subunits have been proposed to comprise the catalytic ATP hydrolyzing domain of the enzyme, or VI sector (Kane et al., 1992). Other polypeptides of this enzyme, including the 100and 17-kDa species, behave as hydrophobic polypeptides that are integrally associated with the vacuolar membrane. These hydrophobic subunits have been proposed to compose the proton pumping portion of the enzyme, or Vo sector (Kane et al., 1992).
Three genetic screens have successfully identified mutants deficient in the vacuolar H+-ATPase, and these include calcium-sensitive strains (cls), vacuolar pH mutants (uph), and trifluoperazine-resistant mutants (tfp) (Ohya et al., 1986;Preston et al., 1989;Shih et al., 1988Shih et al., , 1990. Although these genetic screens have dissimilar primary selections, the mutants identified by these screens that are devoid of vacuolar membrane H+-ATPase activity all exhibit the u n a mutant phenotypes described above. The molecular genetic characterization of the urna mutants and several VMA genes, as well as the characterization of the VMA gene products, have significantly enhanced the understanding of the vacuolar H+-221 Characterization of uma Mutants ATPase complex. However, it is not yet known how many subunits actually compose the final form of this enzyme. In addition, it is likely that many additional gene products play a role in assembly and/or function of this multisubunit enzyme in the yeast vacuole. We designed a genetic screen to identify mutant strains that exhibited uma phenotypes in order to elucidate the subunit composition of the yeast vacuolar H'-ATPase. Using this genetic screen, we hoped to identify mutations in genes that were specifically required for vacuolar membrane H+-ATPase function and/or assembly of this multisubunit enzyme complex. This paper describes the isolation and characterization of mutant strains identified by the vma mutant screen. We report the cloning of the gene, VMA5, that complements one of these mutations, as well as the subsequent characterization of the VMAS gene product. In addition, the VMA4 gene product, which was originally described by Foury (1990), is more fully characterized in this paper.

EXPERIMENTAL PROCEDURES
Materials-Enzymes for recombinant DNA methods were purchased from New England Biolabs, Bethesda Research Laboratories, Promega Biotec, or Boehringer Mannheim. All other chemicals were described by Kane et al. (1992).
VMAS Cloning, Subcloning, and Gene Disruption-The VMA5 gene was originally isolated on a yeast genomic DNA-containing YEp24 plasmid pSELl (Lillie and Brown, 1992). A partial sequence of this plasmid insert showed an open reading frame that had similarity to the cDNA encoding the 39-kDa subunit C of the chromaffin granule H+-ATPase (Lillie and Brown, 1992). Our laboratory received pSELl as a gift from Sue Lillie and Susan Brown (University of Michigan School of Medicine, Ann Arbor, MI). The VMAS disruption construct has a 2.2-kb' HpaI-HpaI LEU2 fragment replacing the 1050-base pair EcoRI to Sal1 portion of the VMA5 gene (Fig. 2). The vma5A::LEU2 disruption construct, pMH12, was linearized with MluI and SpeI generating a 3.2-kb fragment that was used to integrate into the wild type VMA5 locus of haploid yeast cells and Leu+ prototrophs selected (Rothstein, 1983). The structure of the disruption allele was confirmed by Southern blot analysis (Strauss, 1987). Genomic DNA was prepared from wild type and uma5A yeast strains (Nasmyth and Reed, 1980), and was digested with EcoRV. Southern blots were probed with the HindIII-EcoRI 300-base pair fragment of the 5'-end of the VMA5 gene. The probe hybridized to a -4.5-kb fragment for the wild type VMA5 gene, or a -1.2-kb fragment for the disrupted gene (LEU2 contains a single EcoRV site). The VMA5 1.7-kb complementating region (EcoRV-SpeI) was cloned into pRS315 (Sikorski and Hieter, 1989) generating pMH14. pMH14 was transformed into SF838-1Da uma5A::LEUZ yeast strains and Ura' prototrophs were selected. All yeast strains were grown in YEPD (1% yeast extract, 2% Bacto-peptone, 2% dextrose) or SD (0.67% yeast nitrogen base, 2% dextrose) with the appropriate supplements. All YEPD media was buffered with 50 mM phosphate, 50 mM succinate to either pH 5.0 or 7.5 (Yamashiro et al., 1990). Tetrad analysis was performed as described previously (Sherman et al., 1982). All manipulations of DNA were carried out as described by Sambrook et al. (1989). The Escherichia coli strain used for all manipulations was MC1061: F-hdR, The abbreviations used are: kb, kilobase; PAGE, polyacrylamide gel electrophoresis. hsdM+, arad139, (araABOIC-leu7679), (lac)X74, galU, galK, rpsL (Casadaban and Cohen, 1980). Sequence Determinations-The chain termination method of Sanger et al. (1977) was used to determine the DNA sequence of the VMAS gene. The VMA5 gene was sequenced on both strands over the entire MluI to SpeI fragment of pSEL1. Double stranded DNA sequencing using Taq DNA polymerase was done according to the nested deletions (Henikoff, 1987) in a Bluescript pKS+ plasmid manufacturer's recommendations (Promega) on ExoIIIISl nuclease system (Stratagene). Nested deletions were generated in both directions using ExoIII nuclease digestions (Promega Biotec).
The amino acid sequences of the 42-and 27-kDa vacuolar H+-ATPase subunits were determined first by isolating the subunits of the vacuolar H'-ATPase (Uchida et al., 1985). The peak H+-ATPase fractions from eight glycerol gradients (>40 mg starting vacuolar vesicle protein) were collected and precipitated with 10% trichloroacetic acid on ice for 30 min. Precipitated protein was pelleted at 10,000 X g, and the remaining supernatant was aspirated off. Protein pellets were resuspended in sample buffer (8 M urea, 5% SDS) and neutralized with Tris base (final concentration 10 mM). Samples were heated at 37 "C for 1-2 h. Proteins were separated by SDS-PAGE and electrotransferred onto nitrocellulose membrane (Millipore) (350 mA for 5 h at 4 "C). Proteins were visualized with Ponceau S (Sigma) and the strip of nitrocellulose containing the relevant protein band was excised, washed, digested with trypsin, and tryptic peptides separated by high performance liquid chromatography according to the method of Aebersold et al. (1987). Peptide sequences were obtained with an Applied Biosystems Model 470A equipped with an Applied Biosystems Model 120A phenylthiohydantoin analyzer.
Protein Preparation, Antibodies, and Western Blotting-Whole cell and vacuolar vesicle protein extracts were prepared as previously described (Kane et al., 1992). Proteins were separated on 10% polyacrylamide gels and electrotransferred onto nitrocellulose at 150 mA overnight at room temperature. Western blots were probed as previously described (Kane et al., 1992). Primary antibodies were usually applied to the blots at a concentration of 1/1000 in the presence of 0.1% Tween 20 and 3.0% nonfat milk. Proteins were visualized using an alkaline phosphatase assay (Promega). E. coli strains containing pKHlOO were used to produce Vma4p antigen as previously described (Roberts et al., 1989). pKHlOO contains the HindIII-AccI (codons 22-233) fragment of the VMA4 gene cloned into the SmaI site of the pEXP3 expression vector. The antigen was injected into New Zealand White rabbits as described (Vaitukaitis, 1981) and the resulting anti-Vma4p antiserum affinity purified (Stevens et al., 1982) using the Vma4p antigen produced from the pKHlOO plasmid. Monoclonal antibodies specific for the H+-ATPase subunits were made against vacuolar membranes (Kane et al., 1989(Kane et al., , 1992. The 100-kDapolypeptide was probed with monoclonal antibody 7B1, the 69-kDa subunit with llE6, the 60-kDa subunit with 13Dl1, and the 42-kDa subunit with 7A2. formed using monoclonal antibodies to the 100-(10D7) or BO-kDa Microscopy-Indirect immunofluorescence microscopy was per-(13Dll) subunit (Kane et al., 1992;Roberts et al., 1990). Analysis of vacuolar uptake of quinacrine was performed as previously described (Weisman et al., 1987;Roberts et al., 1990). Microscopy was performed using a Zeiss Axioplan Photomicroscope equipped for NOmarski optics and epifluorescence with a X 100 oil-immersion lens. Photography was done with Kodak TMax-400 film and developer. An exposure time of 8 s was used for all of the fluorescence micrographs.

RESULTS
Screen for vma Mutants-We were interested in designing a genetic screen that would identify gene products essential for the activity of the vacuolar H+-ATPase. The selection criteria for this genetic screen were based on phenotypes (Vma-) that have been described for cells disrupted for one of the vacuolar membrane ATPase subunit genes (Manolson et al., 1992;Nelson and Nelson, 1990;Yamashiro et al., 1990;Ohya et al., 1991;Foury, 1990). urna mutant strains are unable to grow on media buffered at neutral pH, and are petite (Yamashiro et al., 1990;Ohya et al., 1991). In addition, a red pigment that is an intermediate of adenine biosynthesis fails to accumulate in the vacuoles of vma ade2 yeast cells (Foury, 1990). Together, these phenotypes identify vacuolar H'-ATP- ase-deficient mutants in an ade2 background as those that form white colonies that fail to grow when replica plated onto neutral buffered media. The urna genetic screen was initiated by mutagenizing SEY6211a haploid cells with UV light. Mutagenized cells were plated on YEPD media buffered to pH 5.0. Colonies were grown at 30 "C and the red pigment allowed to accumulate. Colonies that were white were picked and tested for growth on YEPD media buffered to pH 7.5. Colonies that were unable t o grow on pH 7.5 buffered YEPD plates, but grew relatively well on media buffered to pH 5.0, were tested for quinacrine accumulation in the vacuole. Mutants that did not accumulate quinacrine, and in addition had all of the above mutant phenotypes, were designated vma mutants.
Of the 146,000 colonies originally screened, 18 exhibited the Vma-phenotypes. Each of the mutants was crossed to strains containing tfpl, vma2, uma3, uma4, umall, vmal2 and umal3 disruption alleles. Thirteen of these original mutants failed to complement a strain containing the tfpl disruption allele. TFPl encodes the 69-kDa vacuolar H+-ATPase subunit (Hirata et al., 1990;Kane et al., 1990). The other five mutants complemented all of the uma disruption strains. These mutants were backcrossed to SEY6211a and the resulting heterozygotes sporulated and dissected. The Vma-phenotype segregated 2:2 in all tetrads examined (data not shown). Vmabackcrosses of each of the five mutants were crossed against each other, and analysis revealed that the five mutants defined four new complementation groups ( Table I).
Characterization of the vma Mutants-Whole cell protein extracts were prepared from representative strains of each of the newly isolated urna complementation groups. These extracts were analyzed using monoclonal antibodies against the loo-, 69-, and 60-kDa subunits of the vacuolar H+-ATPase in order to determine whether the synthesis or stability of these polypeptides was perturbed in these mutant strains (Kane et al., 1989;. In addition, we analyzed the whole cell protein extracts with monoclonal antibodies directed against the 42-kDa polypeptide, which was predicted to be a subunit of the vacuolar H+-ATPase based on biochemical data (Kane et al., 1992). The results of this analysis are shown in Fig. 1. Whole cell extracts prepared from one of these mutants, uma5, appeared to be devoid of the 42-kDa putative subunit as determined by this analysis (Fig. 1, lane 2). The synthesis of the remaining subunits that were monitored in these mutant cells appeared not to be altered as compared to wild type strains. uma21, uma22, and uma23 cell extracts contained wild type levels of all subunits monitored (Fig. 1, lanes 3-5). Interestingly, mutants in two complementation groups, umu21 and uma22, showed decreased levels of the 100-kDa polypeptide in protein extracts from whole cells.' The full characterization of the uma21 and uma22 mutants will be reported el~ewhere.~ Together, these results suggested that the synthe- sis and/or stability of the vacuolar H+-ATPase subunits was altered in representatives of three of the four newly characterized vma complementation groups. Mutants from the urna complementation group that failed to complement the tfpl disruption strain were devoid of the 69-kDa subunit, but these cells contained wild type levels of the loo-, 60-, and 42-kDa polypeptides as measured by Western analysis of whole cell protein extracts (data not shown). This is consistent with previous descriptions of tfpl strains (Kane et al., 1992).
Vacuolar vesicles were isolated from each of the newly characterized vma mutants. Vacuolar membranes from vma5, vma21, and uma22 strains all showed 4 % of wild type vacuolar ATPase activity (Table I). However, vacuoles prepared from the uma23 mutant retained 15-20% of the wild type levels of vacuolar ATPase activity. The residual ATPase activity detected in uma23 vacuolar membranes was inhibited by the addition of 2 PM bafilomycin AI, a potent and specific inhibitor of the vacuolar H+-ATPase (Bowman et al., 1988). These results show that the activity of the vacuolar membrane H+-ATPase was altered in all of the newly identified urna strains.
We have identified four novel vma complementation groups (uma5, vma21, vma22, and uma23) using a mutant screen to identify gene products essential for the activity of the vacuolar H+-ATPase in yeast. Members of each group had substantially decreased levels of vacuolar H+-ATPase activity as compared to wild type strains. In addition, three of the four u n a complementation groups were altered in the levels of known or putative vacuolar H+-ATPase subunits. These newly identified urna mutants appear to have mutations in genes whose products are directly or indirectly responsible for the normal activity of the vacuolar membrane H+-ATPase in yeast.
Cloning of the VMA5 Gene-We were interested in determining whether the genes identified by the vma genetic screen actually encode subunits of the vacuolar H+-ATPase. It was also possible that these gene products indirectly lead to loss of vacuolar H+-ATPase activity. In order to begin to resolve these issues, the gene that complemented the uma5 mutant phenotypes was cloned and the encoded protein characterized. SEY6211a vma5 was transformed with a plasmid (pSEL1) containing a portion of the yeast genome in the YEp24 multicopy vector. This insert has been partially sequenced by Lillie and Brown (1992) (identified as a gene adjacent to a multicopy suppressor of a my02 mutation) and exhibits amino acid similarity to the cDNA encoding the 41-kDa subunit C of the bovine chromaffin granule vacuolar H+-ATPase (Lillie and Brown, 1992; Nelson et al., 1990). vma5 mutant colonies transformed with pSELl were able to grow on YEPD media buffered at pH 7.5, suggesting that the uma mutant phenotype was fully complemented. In addition, these transformed colonies were red, indicating that they were accumulating the adenine biosynthetic intermediate in their vacuoles (Foury, 1990). We identified the minimum region required for complementation (Fig. 2) by showing that this region complemented the uma5 mutant phenotypes when carried o n a centromere-containing (CEN) single copy plasmid (pMH14).
The VMAS Gene Encodes the 42-kDa Subunit of the Vacuolar H+-ATPase-The DNA sequence was determined for both strands of the complementing region, and an open reading frame of 373 amino acids was identified (Fig. 3). We designated this gene VMA5. The predicted molecular mass of the putative VMA5 polypeptide is 42,283 daltons. The sequence predicts a hydrophilic polypeptide that lacks both a signal sequence and membrane-spanning domain. Data bank analysis showed that the VMAS gene has 37.5% amino acid identity with subunit C of the bovine chromaffin granule vacuolar H+-ATPase (Nelson et al., 1990). The sequence encoded by the VMA5 gene was found to be identical to the recently reported sequence of subunit C of yeast vacuolar H+-ATPase (Beltran et al., 1992). In addition, the predicted Vma5p sequence contained internal peptides that were derived from the 42-kDa polypeptide that copurified with the vacuolar H+-ATPase enzyme (Fig. 3). Therefore, it was concluded that the VMA5 gene encodes the 42-kDa subunit of Predicted amino acid sequence of the VMAS gene as determined by DNA sequence (plain type) as it aligns with internal peptide sequences derived from the 42-kDa subunit of the vacuolar H'-ATPase (bold type). Sequences were determined as described under "Experimental Procedures." the yeast vacuolar H+-ATPase. Disruption of the VMA5 Gene-To determine whether the phenotype of the vma5 mutant strain coincided with loss of the VMA5 gene function, the chromosomal VMAS allele was replaced by the vma5A::LEU2 disruption allele (Fig. 2). The vma5A::LEU2 gene was transformed into SEY6211a, SF838-lDa, and SF838-5Aa strains and Leu+ prototrophs selected. Disruption of the VMAS gene in SF838-1Da was confirmed by Southern blot analysis (data not shown).
The vma5A::LEU2 disruption strains had phenotypes that were indistinguishable from the originally isolated uma5 mutant. Particularly, vacuolar vesicles derived from the uma5A::LEU2 mutant strains contained <1% of the wild type vacuolar H+-ATPase activity (data not shown). Furthermore, Western analysis of whole cell protein extracts revealed that the 42-kDa subunit was absent from the strains containing the vma5 disruption allele (Fig. 4). The mutant phenotypes resulting from the vma5 gene disruption were fully complemented by the VMA5 gene carried on a CENplasmid, pMH14. In addition, SF838-1Da vdA::LEU2 cells transformed with plasmid pMH14 contained wild type levels of the 42-kDa polypeptide as determined by Western analysis of whole cell protein extracts (Fig. 4). The vma5A::LEU2 strain was crossed with the vma5 mutant strain isolated in our vma mutant screen, the diploid sporulated, and tetrads dissected. The Vma-phenotype segregated 4: O in all 13 tetrads examined, indicating close linkage between the alleles and suggesting that the cloned gene was indeed VMA5. Tetrad progeny exhibited 2:2 segregation for all nutritional markers (data not shown). Together, these results indicate that a loss of the 42-kDa vacuolar H+-ATPase subunit is responsible for the mutant phenotypes exhibited by the vma5 strain that was identified by the uma mutant screen.
The VMA4 Gene Encodes the 27-kDa Vacuolar H+-ATPase Subunit-The VMA4 gene was recently cloned and predicted to encode a subunit of the yeast vacuolar H+-ATPase (Foury, 1990). The sequence predicted that Vma4p is a hydrophilic polypeptide of 26.6 kDa. The internal peptide sequence LLSEEALPAIR, which was derived from the 27-kDa polypeptide that copurified with the vacuolar H+-ATPase (Kane et al., 1989), is identical to a region (aa209-aa219) of Vma4p. A polyclonal antibody was generated against VMA4 protein expressed in E. coli. Immunoblots of whole cell protein extracts from wild type and vma4A::URAS cells were probed using this Vma4p antibody. These immunoblots verified that Vma4p migrates at an apparent molecular mass of 27 kDa and that this protein was absent from the extracts derived from vma4A::URAS strains, verifying that the antibody was specific for the VMA4 gene product (Fig. 5). We therefore concluded that the VMA4 gene encodes the 27-kDa subunit u Vma5p   FIG. 4. The VMA5 gene on a CEN plasmid complements the uma5::LEU2 mutation. Whole cell protein extracts were prepared from SF838-1Da wild type, umaSA::LEU2, or uma5A::LEU2 transformed with the VMA5 gene on a CEN containing plasmid (pMH14).
Protein extract equivalent to 2 X lo7 cells was loaded for each sample, and the polypeptides separated by SDS-PAGE on a 10% acrylamide gel. Immunoblots of the extracts were probed with monoclonal antibody specific for the 42-kDa vacuolar H+-ATPase subunit.  5. urna5A and uma4A cells contain wild type levels of the other vacuolar H+-ATPase subunits. Whole cell extracts were prepared from SF838-1Dn wild t.ype, ffplA::LEU2 (tfp1.l). umaFjA::LEU2 (umaijA), or uma4A:: URA3 (uma4.l). Protein extract equivalent to 5 X 10' cells was loaded to detect the 100-kDa polypeptide, 2 X 10' cells to detect the 69and 42-kDa polypeptides, and 2 X 10" cells loaded to detect 60-and 27-kDa polypeptides. Proteins were separated and immunoblotted as described in the legend to Fig. 1. of the yeast vacuolar H'-ATPase.
Synthesis and Assembly of the Vacuolar H+-ATPase in uma5A and vma4A Mutants-Yeast strains carrying a null allele of either the VMA5 or VMA4 gene were devoid of vacuolar H+-ATPase activity (see above; Foury (1990)). We were interested in determining the levels of the remaining vacuolar H'-ATPase subunits in the VMA5 and VMA4 disruption strains. In addition, we were interested in determining the degree of assembly of the vacuolar H'-ATPase in these mutant strains. In order to determine the level of synthesis and stability of vacuolar H'-ATPase subunits in uma5A and vma4A strains, whole cell protein extracts were prepared from SF838-1Da wild type cells and from SF838-1Da strains disrupted in the 42-kDa subunit gene (uma5A), the 27-kDa subunit gene (uma4A), or the 69-kDa subunit gene (tfpld). Immunoblots of these whole cell extracts were probed using antibodies that recognize the loo-, 69-, 60-, 42-, and 27-kDa vacuolar H' -ATPase subunits. As expected, these results showed that the 69-kDa subunit was absent from the tfplA strain (Hirata et al., 1990;Kane et al., 1992) (Fig. 5). In addition, the 42-kDa subunit was absent from uma5A cells and the 27-kDa subunit was absent from uma4A cells. However, disruption of any of these subunit genes appeared to have no effect on the synthesis or stability of the remaining vacuolar H'-ATPase subunits in whole cell protein extracts (Fig. 5).
Vacuolar vesicles were prepared from tfpl A, vma5A, and uma4A mutant strains to determine whether or not the H+-ATPase subunits synthesized were assembled onto the vacuolar membrane. The resulting immunoblots are shown in Fig.  6. Vacuolar membranes prepared from wild type cells contained all of the vacuolar H'-ATPase subunits, characteristic of the fully assembled enzyme. However, vacuolar membranes prepared from the mutants were largely deficient in the 69-, 60-, 42-, or 27-kDa polypeptides, whereas a significant portion of the 100-kDa subunit (and/or the 75-kDa breakdown product) (Kane et al., 1992) was still present on the vacuolar membrane in all three mutant strains. The results obtained for the tfplA mutants are consistent with those previously reported (Kane et al., 1992). Together, these data show that the enzyme is not assembled on the vacuolar membrane in the tfpl A, uma51, or vma41 strains. Indirect immunofluorescence microscopy was used in a second procedure to analyze the extent of H'-ATPase assembly onto the vacuolar membrane in tfplA, vma5A, and uma4A mutant cells. The localization of peripheral membrane vacuolar H'-ATPase subunits, represented by the 60-kDa subunit, was analyzed in these mutants by indirect immunofluorescence microscopy. These results (Fig. 7a) show that the 60-kDa subunit was present on the vacuolar membrane in wild type cells. The location of the vacuole was visualized by Nomarski optics and appeared as a large round depression in the yeast cell (Roberts et al., 1990). In uma5A and vma4A mutant cells, however, the 60-kDa polypeptide appeared to be distributed diffusely throughout the cytoplasm. In addition to diffuse c-ytoplasmic staining, a small portion (-10% based on western analysis) of the 60-kDa subunit remained associated with the vacuolar membrane in vma5A strains. Together, these results demonstrated that the peripheral membrane 60-kDa H+-ATPase subunit was largely dissociated from the vacuolar membrane in tpfl A, uma5A, and uma4A mutant cells. The localization of the integral membrane subunits of the vacuolar H+-ATPase, represented by the 100-kDa subunit, was also analyzed using indirect immunofluorescence microscopy (Fig. 7b). Monoclonal antibodies against the 100-kDa subunit recognize an epitope that is masked in cells containing the wild type enzyme (Kane et al., 1992). However, in the tfplA, uma5A, and vma4A mutant cells, the 100-kDa epitope was unmasked. In these mutant cells, the 100-kDa subunit was clearly visible on the vacuolar membrane. This result demonstrated that the 100-kDa subunit remained associated with the vacuolar membrane and was stable in tfplA, uma5A, and uma4A mutant cells. The results obtained for the tfplA mutants are consistent with those previously reported (Kane et al., 1992). These data clearly show that the 100-kDa integral membrane H'-ATPase subunit is stable in the vacuolar membrane in yeast cells lacking either the 27-or 42-kDa peripheral membrane vacuolar H+-ATPase subunits.

DISCUSSION
Deletion of any one of the yeast vacuolar H+-ATPase structural genes disrupts both the function and composition of the enzyme complex (Yamashiro et al., 1990). Yeast cells that are devoid of vacuolar membrane H+-ATPase (vma) activity possess distinct phenotypes including the inability to grow on media buffered a t neutral pH. A genetic screen was designed based on several of these uma phenotypes in order t o fully elucidate the subunit composition of the vacuolar H+-ATPase, and to identify nonsubunit proteins required for vacuolar H+-ATPase function. This screen resulted in the identification of four novel uma mutants, uma5, uma21, uma22, and uma23, whose wild type gene products are essential for normal activity of the vacuolar H+-ATPase. It is clear that we have not saturated the screening process as we did not isolate mutations in the VMAP, VMA3, or VMA4 genes, although we did isolate mutations in TFPl 13 times. This result may reflect differential sensitivity of urna genes to ultraviolet radiation. Nevertheless, subsequent screenings for uma mutants will be necessary to fully elucidate the subunit composition of this multisubunit enzyme complex.
Our results show that the protein encoded by VMAB is the 42-kDa subunit of the vacuolar H+-ATPase originally described by Kane et al. (1989). The VMA5 gene was deleted (uma5A) from the yeast genome and the phenotypes were analyzed. The phenotypes of the uma5A strains were indistinguishable from those exhibited by the originally isolated uma5 mutant. The 373 amino acid VMAS gene product is predicted to be a hydrophilic polypeptide and does not contain a signal sequence or a membrane spanning domain. While this work was in progress, Beltran et al. (1992) reported the sequence of "subunit C" of the yeast vacuolar H+-ATPase. The VMAS gene is identical to the subunit C gene.
The VMA4 gene, which was originally cloned by Foury (1990), was determined to encode the 27-kDa subunit of the yeast vacuolar H+-ATPase that was originally described by Kane et al. (1989). Yeast strains containing disruption alleles of the VMA4 gene displayed Vma-mutant phenotypes. As expected, uma4A strains completely lacked the 27-kDa vacuolar H+-ATPase subunit.
We were interested in determining whether the biosynthesis, localization, and assembly of the remaining vacuolar H+-ATPase subunits was altered in cells lacking either the 42-(uma5A) or 27-kDa (vma4A) polypeptides. vma5A and uma4A cells contained wild type levels of all the remaining vacuolar H'-ATPase subunits in the whole cell protein extracts. However, vacuolar vesicles prepared from either mutant strain were largely devoid (~1 0 % of the total) of any of the peripherally associated subunits (collectively referred to as the VI sector of the enzyme) of the H+-ATPase complex. The VI portion of the enzyme is predicted to be composed of hydrophilic subunits that associate peripherally with the cytoplasmic face of the vacuolar membrane, and is responsible for the catalytic activity of the enzyme. VI was proposed to consist of a t least three subunits the 69-, 60-, and 42-kDa polypeptides (Kane et al., 1992). We have demonstrated that none of the components of the VI portion of the enzyme is properly assembled onto the vacuolar membrane in uma5A and uma4A strains, In addition, vacuolar vesicles from uma5A and tfplA cells were largely devoid of the 27-kDa polypeptide. We therefore conclude that the 27-kDa subunit is also a component of the V, sector of the yeast vacuolar H+-ATPase.
The 100-and 17-kDa subunits are components of the integral membrane VO portion of the vacuolar enzyme. The integral membrane vacuolar H+-ATPase subunits that constitute the VO sector were stably assembled into the vacuolar membrane in uma5A and uma4A mutant strains. Therefore, it appeared that although the V I subunits of the vacuolar H+-ATPase were not properly assembled onto the vacuolar membrane in either the uma5A or the uma4A mutant strains, the VO subunits of the enzyme (represented by the 100-kDa subunit) were properly targeted to and stable in the vacuolar membrane.
This paper describes a genetic screen that was designed to identify novel genes whose products are essential for the synthesis, assembly, and/or function of the vacuolar membrane H+-ATPase. We identified four new yeast genes, vma5, uma21, uma22, and uma23, that are essential for vacuolar H+-ATPase activity. We subsequently characterized VMAS and showed that the gene product is essential for the activity of the vacuolar H+-ATPase, and for the assembly of this multisubunit complex onto the vacuolar membrane. In addition, Vma4p, which was originally cloned by Foury (1990), is also shown in this paper to be essential for the assembly and activity of the vacuolar H'-ATPase. Biochemical and immunolocalization data support the assignment of Vma4p as a component of the VI sector of the yeast vacuolar H+-ATPase.
The characterization of these VMA genes enhances our understanding of the yeast vacuolar H'-ATPase enzyme complex. However, it is not clear how many genes are required for the function of the vacuolar H+-ATPase in yeast. To date, 12 yeast genes have been described that are essential for vacuolar H+-ATPase function, VPHl (Manolson et al., 1992), VMAf (TFPl) (Shih et al., 1988;Hirata et al., 1990;Kane et al., 1990), VMA2 (VAT2) (Nelson and Nelson, 1990;Yamashiro et al., 1990), VMA3 (Umemoto et al., 1990), VMA4 (Foury, 1990), VMAS (Beltran et al., 1992; this work), VMAll (Shih et al., 1990;Umemoto et al., 1991), VMAl2 and VMA13 (Ohya et al., 1991), VMA21, VMA22, and VMA23 (this work). The identification of additional uma complementation groups will be necessary to fully elucidate the subunit composition of the vacuolar H+-ATPase and to determine the assembly process of this multisubunit enzyme en route to the vacuolar membrane. strains and for stimulating discussions throughout the course of this work. Also, we thank Patty Kane for many thoughtful conversations.