Vma22p is a novel endoplasmic reticulum-associated protein required for assembly of the yeast vacuolar H(+)-ATPase complex.

The Saccharomyces cerevisiae vacuolar H-ATPase (V-ATPase) is a multi-subunit complex that can be structurally and functionally divided into peripheral (V) and integral membrane (V) sectors. The vma22-1 mutation was isolated in a screen for mutants defective in V-ATPase function. vma22Δ cells contain no V-ATPase activity due to a failure to assemble the enzyme complex; V subunits accumulate in the cytosol, and the V 100-kDa subunit is rapidly degraded. Turnover of the 100-kDa integral membrane protein was found to occur in the endoplasmic reticulum (ER) of vma22Δ cells. The product of the VMA22 gene, Vma22p, is a 21-kDa hydrophilic protein that is not a subunit of the V-ATPase but rather is associated with ER membranes. The association of Vma22p with ER membranes was perturbed by mutations in VMA12, a gene that encodes an ER membrane protein (Vma12p) that is also required for V-ATPase assembly. These results indicate that Vma22p, along with Vma21p and Vma12p, form a set of ER proteins required for V-ATPase assembly.

The Saccharomyces cerevisiae vacuolar H ؉ -ATPase (V-ATPase) is a multi-subunit complex that can be structurally and functionally divided into peripheral (V 1 ) and integral membrane (V 0 ) sectors. The vma22-1 mutation was isolated in a screen for mutants defective in V-ATPase function. vma22⌬ cells contain no V-ATPase activity due to a failure to assemble the enzyme complex; V 1 subunits accumulate in the cytosol, and the V 0 100-kDa subunit is rapidly degraded. Turnover of the 100-kDa integral membrane protein was found to occur in the endoplasmic reticulum (ER) of vma22⌬ cells. The product of the VMA22 gene, Vma22p, is a 21-kDa hydrophilic protein that is not a subunit of the V-ATPase but rather is associated with ER membranes. The association of Vma22p with ER membranes was perturbed by mutations in VMA12, a gene that encodes an ER membrane protein (Vma12p) that is also required for V-ATPase assembly. These results indicate that Vma22p, along with Vma21p and Vma12p, form a set of ER proteins required for V-ATPase assembly.
The acidification of many organelles including the Golgi, vacuoles, endosomes, and clathrin-coated vesicles is generated via a vacuolar-type proton-translocating ATPase or V-ATPase. 1 The electrochemical gradient generated by V-ATPases is used in such cellular processes as protein sorting, receptor-mediated endocytosis, and zymogen activation (Mellman et al., 1986).
The yeast V-ATPase is similar in structure and function to V-ATPases from other fungi, plants, and animals (Uchida et al., 1985;Kane et al., 1989). All are multi-subunit complexes that, in a manner similar to the well characterized F 1 F 0 -ATPase, can be divided structurally and functionally into V 1 and V 0 sectors. The yeast V-ATPase V 1 sector consists of subunits peripherally associated with the membrane and constitutes the catalytic and regulatory functions of the enzyme complex. The V 0 portion of the enzyme contains integral membrane subunits and forms a pore through which protons are translocated into an organelle .
Disruption of genes (vma, vacuolar membrane ATPase) encoding V-ATPase subunits (with the exception of VPH1) gives rise to a characteristic set of phenotypes (Vma Ϫ ) associated with the loss of enzyme function. These include slow growth, an inability to grow on media buffered to a neutral pH, sensitivity to high Ca 2ϩ , and a Pet Ϫ phenotype. In addition, vma ade2 double mutant colonies are white instead of the red color typical of ade2 colonies. These phenotypes have been used to design screens for mutations in genes encoding structural subunits of the V-ATPase complex as well as factors required for the assembly/stability but which are not subunits of the final enzyme complex (Ho et al., 1993a;Ohya et al., 1991).
Studies of the assembly of the V-ATPase in cells mutant in either V 0 or V 1 subunit genes have indicated that assembly requires the coordination of integral membrane subunits and peripheral subunits. Disruption of a gene encoding any single V 1 peripheral subunit of the V-ATPase complex leads to a failure of the V 1 sector to assemble on the vacuolar membrane and the stable accumulation of remaining V 1 subunits in the cytosol. The V 0 sector of the enzyme, however, is transported to and stable in the vacuolar membrane in the absence of an assembled V 1 sector (Doherty and Kane, 1993;Ho et al., 1993a). The absence of a V 0 sector subunit leads to the destabilization/ degradation of the remaining V 0 polypeptides as well as a failure of V 1 subunits to associate with the vacuolar membrane Bauerle et al., 1993).
Recent studies have characterized the VMA12 and VMA21 genes, which are involved in V-ATPase assembly but whose products (Vma12p and Vma21p) are not subunits of the final vacuolar enzyme complex (Hirata et al., 1993;Hill and Stevens, 1994). Vma21p is a membrane protein of the endoplasmic reticulum (ER), and mutations in VMA21 lead to a rapid degradation of the 100-kDa V 0 polypeptide. A model has been proposed in which assembly of the V-ATPase complex is initiated in the ER, and failure to do so results in the rapid turnover of the 100-kDa integral membrane polypeptide within the ER .
In this work, we describe the characterization of the VMA22 gene and its product, Vma22p. Similar to Vma21p, Vma22p is a non-subunit assembly factor that is associated with the ER. Loss of Vma22p function leads to a rapid and specifically ERbased degradation of the 100-kDa V 0 polypeptide and failure of the V-ATPase complex to assemble.

EXPERIMENTAL PROCEDURES
Strains, Media, and Microbiological Techniques-Yeast strains used in this study are listed in Table I. Strains SNY28 and SF838 -1D are isogenic except at the PEP4 locus. SNY28, KHY13, and KHY38 are isogenic except at the VMA22 locus. SF838 -1D, KHY34, and KHY39 are isogenic except at the VMA22 locus. SNY28 and DJY63 are isogenic except at the VMA12 locus. Yeast strain MBY10 -7A was a gift from A. Nakano. Yeast extract peptone dextrose buffered to pH 5.0 or 7.5 was prepared as described by Yamashiro et al. (1990). Low adenine containing synthetic dextrose media used for cloning VMA22 was prepared as described by Cooper et al. (1993). Other standard yeast media and genetic manipulations were performed as described by Sherman et al. (1986).
Plasmid Construction and DNA Sequencing-Plasmids used in this study are listed in Table II. Plasmids derived from pKH2201 used in complementation testing were constructed as follows. Plasmid pKH2201 was digested with EcoRI, and the 3.1-kb fragment generated was inserted into EcoRI-digested pRS316 (Sikorski and Hieter, 1989) to create pKH2203. The 4.7-kb fragment also generated from the above digest was inserted into EcoRI-digested pRS316 to create pKH2204. pKH2204 was then digested with BamHI, a 3.05-kb fragment excised, and the plasmid religated to create pKH2206, which contained a 1.65-kb yeast DNA insert. The 3.05-kb fragment was inserted into BamHI-digested pRS316 to create pKH2207. pKH2207 was digested with SpeI and religated to create pKH2208. pKH2208 contained a 1.9-kb yeast DNA insert. pKH2207 was also digested with KpnI and religated to create pKH2209, leaving a 2.45-kb insert. Finally, pKH2208 was digested with SnaBI and SalI, and the plasmid religated to create pKH2211 and leave a 1-kb yeast DNA insert.
The structure of the vma22⌬ disruption in KHY38 and KHY39 was confirmed by PCR amplification using oligonucleotide AUX1 as described above and oligonucleotide STHN2 (5Ј-ATCACATGGTGTTT-TATGGG-3Ј) to amplify a 1.4-kb DNA fragment specific for the vma22⌬::URA3 disruption. The vma22⌬ disruption in KHY13 and KHY34 cells was also confirmed by PCR amplification. Oligonucleotides STHN2 as above and oligonucleotide LEU5OUT (5Ј-CAAGTATTGT-GATGCAAGC-3Ј) corresponding to sequences 3Ј to the LEU2 open reading frame were used to amplify a 380-bp DNA fragment specific for the vma22⌬::LEU2 disruption.
Sequencing of VMA22-The nucleotide sequence of VMA22 and flanking sequences contained in pKH2211 were generated from nested deletions made using ExoIII nuclease (Erase-A-Base, Promega). pKH2211 was digested with either SacI and XhoI or with KpnI and XhoI, deletions generated as outlined by the manufacturers, and the DNA sequence determined by the method of Sanger et al. (1977). Data base searches were performed using the BLAST service of the National Center for Biotechnology Information (Altschul et al., 1990). Sequence analysis was performed using the GCG analysis software available from the University of Wisconsin.
Antibodies, Protein Preparation, and Western Blotting-To generate rabbit polyclonal antibodies against Vma22p, the following DNA constructs were prepared. A BglII site was introduced via in vitro mutagenesis (Kunkel et al., 1987) near the N terminus of VMA22 (in pKH2212) (base pairs 43 and 44 of the VMA22 open reading frame) changing bp GC 3 AT and thus amino acid Glu-15 3 Leu to create pKH2214. pKH2214 was then digested with BglII, and the ends were blunted with Klenow enzyme. The plasmid was digested with XbaI, and the 760-bp fragment inserted into SmaI and XbaI cut pEXP2-S (Roberts et al., 1989) to create pKH2218 (Fig. 1A). Induction of Escherichia coli carrying plasmid pKH2218 with isopropyl-␤-D-thiogalactopyranoside produced a 21-kDa fusion protein that was purified and used to inject New Zealand White rabbits for antibody elicitation (Vaitukaitis, 1981). Antiserum recognizing Vma22p was affinity purified against protein produced from pKH2224 using the method of Roberts et al. (1991). Plasmid pKH2224 was created by inserting the BglII (blunted)-XbaI fragment described above into XmnI-and XbaI-digested pMAL-c2 (New England Biolabs). Upon induction with isopropyl-␤-D-thiogalactopyranoside, E. coli cells expressing pKH2224 produced a ϳ63-kDa Vma22p-MBP fusion protein.
Whole cell protein extracts and vacuolar membrane fractions were prepared as described . Vacuolar membrane fractions were extracted with chloroform/methanol as described by Ho et al. (1993b). Monoclonal antibodies recognizing the 100-(10D7, 7B1), 69-(8B1), 60-(13D11), and 42-kDa (7A2) V-ATPase subunits were used as described by . Monoclonal antibodies specific for the 100-, 69-, and 60-kDa V-ATPase subunits and Dpm1p were from Molecular Probes Inc. Polyclonal antibodies directed against the 100-, 54-, and 27-kDa V-ATPase subunits were purified and used as described by Hill and Stevens (1994). Polyclonal antibody directed against phosphoglycerol kinase was used as described by Nothwehr et al. (1995). Monoclonal antibodies specific for the ER membrane protein, Dpm1p, were used as described by Piper et al. (1994). Secondary antibodies conjugated to horseradish peroxidase were purchased from Amersham. Immunoblots were detected using chemiluminescent detection (DuPont NEN).
Immunoprecipitation-35 S-Express label was purchased from Du-Pont NEN. Immunoprecipitation of Vph1p was performed as described . Diploids were sporulated and dissected, and tetrad progeny was tested for temperature and pH sensitivity. KHY22 and KHY20 are spores from a cross between MBY10 -7A and KHY4. . Immunoprecipitation of alkaline phosphatase was performed as described by Nothwehr et al. (1995). Immunoprecipitation of CPY was as previously described (Raymond et al., 1992) omitting the separation of samples into intracellular and extracellular fractions. Temperature-sensitive strains were grown at 22°C, harvested, and resuspended in synthetic dextrose medium lacking methionine and pre-incubated for 15 min at 36°C before pulse-chase immunoprecipitation.
Fluorescence Microscopy-Monoclonal anti-Pma1p antibodies were provided by Dr. John Teem and were used at a 1:10 dilution. Affinitypurified anti-Vma22p antibodies were adsorbed against KHY38 (vma22⌬) cells as described by Roberts et al. (1991). DAPI staining, slide preparation, and photography were carried out as described by Roberts et al. (1989). Quinacrine staining of cells was carried out as described by Roberts et al. (1991).
Subcellular Fractionation-Strains SNY28, KHY4, and DJY63 were spheroplasted, lysed, and subjected to differential centrifugation essentially as described by Horazdovsky and Emr (1993), except that cells were not radiolabeled before fractionation.

RESULTS
Cloning the VMA22 Gene-The vma22-1 allele (KHY28) was isolated in a screen for mutants that were pH-sensitive and formed white colonies in the SEY6211 (ade2) strain, which normally forms red colonies (Ho et al., 1993a). KHY28 cells also displayed other phenotypes that are characteristic of vma mutants such as slow growth and a failure to grow on non-fermentable carbon sources (Pet Ϫ ) or on media containing 100 mM CaCl 2 (Cls Ϫ ). KHY28 cells also failed to accumulate the fluorescent weak base quinacrine in their vacuoles, indicating the organelle was not acidified.
VMA22 was cloned by complementation of the pH and color defects of vma22-1 cells. Strain KHY28 was transformed with a plasmid library (Rose et al., 1987). 1 of 2500 Ura ϩ transformants was red, able to grown on media buffered to pH 7.5, and complementation was plasmid dependent. Plasmid pKH2201 was recovered and found to contain a yeast genomic DNA insert of 8.5 kb. Various subclones of pKH2201 (pKH2203, pKH2204, pKH2206, pKH2207, pKH2208, pKH2209, and pKH2211) demonstrated that a 1-kb SnaBI-SpeI fragment (pKH2211, Fig. 1A) was sufficient to fully complement vma22-1.
The nucleotide sequence of the 1-kb complementing fragment in pKH2211 was determined for both strands (Genbank accession number U24501). One long open reading frame (VMA22) of 543 bp was found predicting a protein (Vma22p) of 181 amino acids with a molecular mass of 21 kDa (Fig. 1B). Data base searches showed that VMA22 was identical to the sequence designated YHR060w in the CPR2-ERG7 intergenic region of chromosome VIII recently identified by the yeast genome project. No significant homology to other sequences in the data base was found. Rabbit antiserum generated using an E. coli expressed Vma22p fusion protein (codons 16 -181 of Vma22p) was used to detect Vma22p in yeast cells. Anti-Vma22p antibodies recognized a single band of 21 kDa in wild-type cells, and this 21-kDa protein was absent in vma22⌬ strains (KHY38 and KHY13) (Fig. 1C).
Phenotypes of vma22⌬ Cells-vma22⌬ cells displayed phenotypes that were characteristic of other vma strains and identical to the vma22-1 mutant. vma22⌬ cells (KHY38) contained no vacuolar ATPase activity. Vacuolar membranes purified from wild-type cells were assayed for ATPase activity and were found to contain a V-ATPase specific activity of 7.0 units versus 0.034 units measured in vacuolar membranes from KHY38 cells (specific activity units were calculated as mol of ATP hydrolyzed/min/mg protein; Roberts et al. (1991)).
To address the basis of the lack of V-ATPase activity seen in vma22⌬ cells, we examined the levels of various subunits of the enzyme complex. Western blot analyses of the V-ATPase subunits in wild-type (SNY28) and vma22⌬ (KHY38) cells are shown in Fig. 2A. Immunoblots of whole cell protein extracts indicated that the level of the 100-kDa V-ATPase subunit (Vph1p) was substantially lowered in vma22⌬ cells as compared to wild-type cells. However, the steady state levels of other subunits (69, 60, 54, 42, 36, and 27 kDa) appeared comparable between wild-type and vma22⌬ cells.
Western blot analyses of vacuolar membrane fractions prepared from wild-type and vma22⌬ cells indicated that although V 1 subunits appeared to be produced at wild-type levels in vma22⌬ cells, they failed to associate with the vacuolar membrane ( Fig. 2A). In addition, indirect immunofluorescence of the 60-kDa subunit of the V 1 sector in vma22⌬ cells showed a cytosolic pattern instead of the vacuolar membrane-staining pattern seen in wild-type cells (data not shown). Immunoblots also indicated that while a low level of the 100-kDa V-ATPase subunit was present in vma22⌬ cells, no 100-kDa polypeptide was present on the vacuolar membrane ( Fig. 2A). Chloroform/ methanol treatment of vacuolar membranes has been shown to extract the 17-kDa proteolipid (Vma3p/Vma11p) component of the V 0 sector of the V-ATPase, which can then be detected by SDS-PAGE and silver staining (Ho et al., 1993b). The 17-kDa Vma22p expression construct in pEXP2-S pKH2224 Vma22p expression construct in pMAL-c2 proteolipid was substantially reduced in vma22⌬ cells in contrast with the proteolipid extracted from wild-type vacuolar membranes (Fig. 2B). In summary, immunoblot results indicate that Vph1p (100 kDa) and Vma3p/Vma11p (17 kDa), constituents of the V 0 integral membrane sector of the V-ATPase enzyme, are destabilized in vma22⌬ cells. Subunits that comprise the V 1 sector of the enzyme (69, 60, 54, 42, 36, and 27 kDa) appeared to be present at wild-type levels and were stable in vma22⌬ cells but were unable to associate with the vacuolar membrane. These results fully accounted for the lack of V-ATPase activity seen in vma22⌬ cells, since this enzyme complex was unassembled. The low level of the 100-kDa subunit (Vph1p) seen in immunoblots from vma22⌬ cells led us to investigate further the nature of this defect. We performed a kinetic analysis of the synthesis and turnover of Vph1p in wild-type and vma22⌬ cells. We also investigated the effect of the PEP4 gene status on the turnover and stability of Vph1p. Pep4p is a critical protease in the activation cascade of vacuolar hydrolases (Ammerer et al., 1986;Woolford et al., 1986). Any effect of a mutation in the PEP4 gene (pep4 -3) on the half-life of Vph1p would implicate vacuolar proteases in the turnover of this protein.
Wild-type and vma22⌬ cells were radiolabeled with [ 35 S]methionine and chased by the addition of non-radiolabeled methionine. At various chase times, samples were immunoprecipitated with anti-Vph1p antibodies and analyzed by SDS-PAGE and fluorography. In wild-type cells (SNY28), Vph1p was very stable with a half-life of ϳ500 min (Fig. 3, lanes 1-4). The The region of pKH2211 used to generate anti-Vma22p antibodies is also shown. Restriction endonucleases are B*, BglII (* indicates this site was introduced by in vitro mutagenesis); S, StyI; Sn SnaBI; Sp, SpeI; X, XmnI. B, nucleotide sequence of VMA22 and predicted amino acid sequence of Vma22p. The nucleotide sequence of the SnaBI-SpeI fragment contained in pKH2211 is shown (Genbank accession no. U24501). The initiation and termination codons of VMA22 are in bold. Vma22p amino acid residues are given in the single letter code and are numbered in bold. C, immunoblot analysis to detect Vma22p. Whole cell protein extracts were prepared from SNY28 (wild type), KHY38 (vma22⌬::URA3), and KHY13 (vma22⌬::LEU2) cells and 15 g of protein loaded onto SDS-PAGE gels (15% acrylamide). The resulting immunoblots were probed with a 1 ⁄250 dilution of affinity-purified anti-Vma22p antiserum and visualized by chemiluminescent detection. ORF, open reading frame; WT, wild type.

FIG. 2. Assembly of the V-ATPase in wild-type (WT) and vma22⌬ cells.
A, whole cell protein extracts were prepared from wildtype (SNY28) and vma22⌬ (KHY38) cells. Samples (20 g) were loaded onto SDS-PAGE gels, and the resulting immunoblots were probed to detect the 100-, 69-, 60-, 54-, 42-, 36-, and 27-kDa V-ATPase subunits. Whole cell protein immunoblots were compared with those prepared from 4 g of vacuolar membrane extracts from both wild-type and vma22⌬ cells. B, vacuolar membrane fractions from wild-type and vma22⌬ cells were treated with chloroform/methanol (see "Experimental Procedures") and loaded onto SDS-PAGE gels; the V-ATPase 17-kDa proteolipid was detected by silver staining. stability of Vph1p was unaffected by introduction of the pep4 -3 mutation (SF838 -1D, half-life ϳ460 min; Fig. 3, lanes 9 -12), indicating that vacuolar hydrolases have little effect on the turnover of Vph1p in wild-type cells. In vma22⌬ cells (KHY13), Vph1p was rapidly degraded with a half-life of only 36 min (Fig.  3, lanes 5-8), and introduction of the pep4 -3 mutation into vma22⌬ cells (KHY34) did little to stabilize the protein (halflife 40 min; Fig. 3, lanes 13-16). These results suggested that the deceased level of Vph1p seen in vma22⌬ cells was due to its rapid turnover and that this degradation likely occurred outside of the vacuole.
To establish where in the secretory pathway Vph1p was being degraded in vma22⌬ cells, we made use of a temperaturesensitive sec mutation. At the non-permissive temperature (36°C), sec12-4 cells are blocked in vesicle budding from the ER (Nakano et al., 1988), and in a sec12-4 vma22⌬ double mutant newly synthesized Vph1p should remain in the ER. If Vph1p is being degraded in a post-ER compartment in vma22⌬ cells, the protein should be stabilized in sec12-4 vma22⌬ cells shifted to 36°C. Alternatively, if Vph1p is being degraded in the ER of vma22⌬ cells, then blocking membrane traffic out of the ER will not influence its half-life.
Strains carrying sec12-4 (MBY10 -7A), vma22⌬ (KHY13), or both sec12-4 and vma22⌬ (KHY9) alleles were grown at the permissive temperature (22°C), shifted to 36°C for 15 min prior to addition of radiolabel, and subjected to immunoprecipitation of Vph1p as previously described. Arrest of protein exit from the ER was monitored by immunoprecipitation of CPY. CPY is found in the ER as the p1CPY precursor (67 kDa), is modified to form the p2CPY precursor in the Golgi (69 kDa), and processed to the mature enzyme (mCPY) in the vacuole (Stevens et al., 1982). Accumulation of p1CPY served to confirm that protein traffic out of the ER was blocked in sec12-4 and sec12-4 vma22⌬ cells (Fig. 4C, lanes 1-6). Additional controls also ensured that degradation of Vph1p was not merely the result of higher temperature or strain background differences (Fig. 4B).
The half-life of Vph1p in sec12-4 cells (Ͼ400 min, Table III) was similar to that found in wild-type cells (half-life Ͼ400 min; Table III). In contrast, vma22⌬ cells showed a rapid turnover of Vph1p at the non-permissive temperature (Fig. 4A, lanes  6 -10), with a half-life of ϳ23 min (Table III). The rapid degradation of Vph1p seen in vma22⌬ cells was not stabilized by  vma21⌬ sec12-4 20 a The radioactivity of Vph1p samples visualized by immunoprecipitation were quantified using an AMBIS radioanalytic imaging system (AMBIS Systems, San Diego) and then plotted against time to estimate the half-life of Vph1p. introduction of the sec12-4 temperature-sensitive allele (Fig.  4A, lanes 11-15), with the half-life of Vph1p in these double mutant cells still only ϳ30 min. Together, these data clearly indicate that the degradation of Vph1p in vma22⌬ cells is occurring in the ER compartment of the secretory pathway.
We had previously determined that Vph1p was rapidly degraded in another vma mutant, vma21 . We have also found that the degradation of Vph1p in vma21⌬ cells  occurs in the ER. KHY22 cells (vma21⌬, half-life ϳ28 min) and KHY21 cells (vma21⌬ sec12-4, half-life ϳ20 min) showed approximately the same rate of degradation of Vph1p at the restrictive temperature ( Fig. 4B and Table III).
Effects of vma22⌬ on Other Membrane Proteins-To investigate whether the vma22⌬ mutation had any effect on other membrane proteins transiting the secretory pathway, we examined the subcellular localization of the multiple membranespanning plasma membrane ATPase, Pma1p (Serrano et al., 1986), in wild-type and vma22⌬ cells by indirect immunofluorescence. The immunolocalization of Pma1p was indistinguishable between wild-type (SNY28) and vma22⌬ cells (KHY38) (Fig. 5A) and indicated that Vma22p was not required for the intracellular targeting of this protein. We also examined the processing of the vacuolar membrane protein, alkaline phosphatase, which transits the secretory pathway as a precursor before delivery and activation to the mature enzyme in the vacuole (Klionsky and Emr, 1989). Pulse-chase immunoprecipitation of alkaline phosphatase from wild-type (SNY28) and vma22⌬ (KHY38) cells showed that this vacuolar membrane protein was correctly translocated into the secretory pathway, targeted to the vacuole, and matured normally in vma22⌬ cells (Fig. 5B).
Subcellular Localization of Vma22p-The amino acid sequence of Vma22p predicted a hydrophilic protein containing neither an ER signal sequence nor a transmembrane domain, suggesting that Vma22p might be a cytosolic protein. To determine the subcellular localization of Vma22p, we immunolocalized the protein by indirect immunofluorescent staining with affinity-purified anti-Vma22p antibodies. Wild-type yeast cells showed a weak ER staining pattern for Vma22p. However, wild-type yeast cells producing approximately twice the normal level of Vma22p (due to the presence of a low copy, centromerebased plasmid carrying VMA22) (pKH2211) exhibited a perinuclear staining pattern, as well as some cytosolic staining (Fig.  6). The immunofluorescent staining for Vma22p was specific since vma22⌬ cells showed a very faint, nonspecific, cytosolic staining pattern (data not shown). The perinuclear staining pattern seen for Vma22p is typical of ER proteins and has been observed for ER proteins such as Vma21p , Kar2p (yeast BiP) (Rose et al., 1989), andEug1p (Tachibana and. Interestingly, yeast cells overproducing Vma22p (due to the presence of a multicopy plasmid carrying the VMA22 gene) exhibited a bright cytosolic as well as an ER staining pattern, suggesting that the association of Vma22p with the ER is saturable (data not shown).
Association of Vma22p with the ER membrane might result from an interaction between Vma22p and an ER membrane protein. Two intriguing candidate proteins are Vma12p and Vma21p. Yeast cells lacking either of these two proteins (vma21  or vma12 2 ) have been found to rapidly degrade Vph1p. Vma21p  and Vma12p (Hirata et al., 1993) 2 are ER membrane proteins necessary for V-ATPase assembly and activity. We investigated whether the subcellular fractionation profile of Vma22p could be altered by mutations in either the VMA21 or VMA12 gene. SNY28 (wild-type), KHY4 (vma21⌬), and DJY63 (vma12⌬) cells were spheroplasted and lysed, and the lysates were subjected to differential centrifugation to generate lysate, P13, S100, and P100 fractions. Horazdovsky and Emr (1993) have demonstrated that fractions generated by this procedure separate soluble cytosolic proteins (S100) from Golgi (P100) and plasma membrane/ER/vacuolar (P13) proteins. Lysate, P13, S100, and P100 fractions were separated by SDS-PAGE, and immunoblots were probed to detect phosphoglycerol kinase, a soluble cytosolic protein, Dpm1p, an ER transmembrane protein (Orlean et al., 1988), and Vma22p. In wild-type cells, Vma22p was detected in the P13 membrane fraction (Fig. 7), consistent with an ER localization as shown by indirect immunofluorescence experiments. While the P13 fraction contains both ER and vacuolar membranes, additional experiments with purified vacuolar membranes revealed that Vma22p is not associated with this organelle (data not shown). As expected, phosphoglycerol kinase and Dpm1p fractionated with the cytosol and P13 membranes, respectively, in these preparations. The fractionation profile of Vma22p was unaltered in vma21⌬ cells (data not shown). However, when fractions prepared from vma12⌬ cells were probed with anti-Vma22p antibodies, Vma22p was now detected in the S100 fraction with little detectable signal in the P13 membrane fraction. The fraction-2 D. D. Jackson and T. H. Stevens, unpublished results.

FIG. 5. Localization of Pma1p and processing of alkaline phosphatase in wild-type and vma22⌬ cells.
A, Wild-type (SNY28) and vma22⌬ (KHY38) cells were fixed, spheroplasted, and stained with anti-Pma1p antibodies. Cells were viewed by Nomarski optics to observe cell morphology and epifluorescence microscopy using filter sets specific for DAPI to observe nuclear staining and fluorescein to observe anti-Pma1p staining. B, Wild-type (SNY28) and vma22⌬ (KHY38) cells were radiolabeled with [ 35 S]methionine for 10 min and chased for 60 min before alkaline phosphatase was immunoprecipitated and analyzed by SDS-PAGE and fluorography. pALP refers to the precursor form of alkaline phosphatase, and mALP refers to the mature vacuolar form of alkaline phosphatase. ation of phosphoglycerol kinase and Dpm1p in vma12⌬ cells was identical to that seen in wild-type cells. These results suggest that the association of Vma22p with the ER membrane is mediated through an interaction with another ER membrane protein, Vma12p.

DISCUSSION
In this work, we report the isolation and characterization of the yeast VMA22 gene, whose product Vma22p is required for the expression of a functional V-ATPase enzyme. Yeast cells lacking Vma22p display phenotypes that are characteristic of other vma mutants including pH and Ca 2ϩ -sensitive growth. Vacuolar membranes from vma22⌬ cells contain no V-ATPase activity, and correspondingly their vacuoles are not acidified.
The specific defect in vma22⌬ cells appears to be at the level of assembly of the V-ATPase complex. Peripherally associated V 1 subunits were stable in vma22⌬ cells but were unable to associate with the vacuolar membrane and instead accumulated in the cytosol. V 0 subunits appeared to be destabilized in vma22⌬ cells, and the 100-kDa subunit, Vph1p, was not found on the vacuolar membrane. The small amount of proteolipid found in chloroform/methanol extracts from vma22⌬ vacuolar membranes could reflect a stable but unassembled vacuolar portion of the V 0 sector or merely represent the contamination of vacuolar membrane preparations with membranes from other organelles.
The low level of Vph1p in vma22⌬ cells was the result of destabilization of the protein. In kinetic analyses, Vph1p was rapidly degraded in vma22⌬ cells with a half-life of ϳ40 min, approximately 8% of that observed in wild-type cells. Degradation of Vph1p was independent of the vacuolar hydrolase, Pep4p, indicating that the increased turnover of Vph1p seen in vma22⌬ cells was likely occurring in an organelle other than the vacuole.
To define the subcellular organelle in which Vph1p was degraded in vma22⌬ cells, we performed our kinetic analyses in sec12-4 vma22⌬ double mutant cells, which accumulate the ER compartment at the non-permissive temperature. The continued degradation of Vph1p in these cells at the permissive or non-permissive temperature showed clearly that degradation of Vph1p occurred in the ER. We also confirmed that the rapid turnover of Vph1p seen in vma21⌬ cells  also occurred specifically in the ER.
The fate of the proteolipid component of the V-ATPase V 0 sector in vma22⌬ cells remains uncertain due to a lack of antibodies to perform a similar kinetic analysis as done here for Vph1p. At present, we are constructing epitope-tagged versions of Vma3p and Vma11p to address these questions and also determine the location of the small amount of proteolipid present in vma22⌬ cells.
Other membrane proteins of the secretory pathway appeared to be unaffected by the vma22⌬ mutation. Pma1p was present at the plasma membrane of both wild-type and vma22⌬ cells, as revealed by immunofluorescence experiments. Also, the vacuolar membrane protein, alkaline phosphatase, was transported to and matured in the vacuole similarly in wild-type and vma22⌬ cells. These results suggest that the vma22⌬ mutation may be specific in its effects on the assembly of the V-ATPase complex.
Although vma22⌬ cells displayed phenotypes typical of other vma mutations, including those in subunit structural genes, Vma22p was found not to be a subunit of the V-ATPase complex. Vma22p was de-enriched in vacuolar membrane fractions and did not co-purify with the V-ATPase. The immunofluorescent labeling pattern of Vma22p showed both an ER and some cytosolic localization. The ER membrane localization of Vma22p is similar to that for Vma21p  and is suggestive of an ER-based set of proteins required for V-ATPase expression. Although Vma22p is predicted to be a hydrophilic protein, both indirect immunofluorescence and FIG. 6. Immunolocalization of Vma22p. Wild-type cells (SNY28) carrying pKH2211 were fixed, spheroplasted, and stained with anti-Vma22p antibodies. Cells were viewed with Nomarski optics to observe cell morphology and epifluorescence microscopy using filter sets specific for DAPI to observe nuclear staining and fluorescein to observe anti-Vma22p staining.
FIG. 7. Membrane association and subcellular fractionation of Vma22p. Wild-type (SNY28) and vma12⌬ (DJY63) cells were fractionated into lysate, P13 (vacuolar, ER, and plasma membrane), S100 (cytosolic), and P100 (Golgi and endosomes) fractions. Equal proportions of each fraction were loaded on SDS-PAGE gels, and the resulting immunoblots were probed to detect Vma22p, the cytosolic protein phosphoglycerol kinase (PGK), and the ER membrane protein, Dpm1p. subcellular fractionation experiments indicate that Vma22p is associated with the ER membrane.
The effects of the vma22⌬ mutation on the V-ATPase complex described here are similar to those reported for vma21⌬  and vma12⌬ (Hirata et al., 1993) 2 mutants. The 100-kDa polypeptide is rapidly degraded in vma21⌬ cells, and in this work we have shown that, as in vma22⌬ cells, Vph1p turnover occurs in the ER. Like Vma22p, Vma21p is also an ER protein, and its ER localization is determined by a C-terminal di-lysine motif . Vma12p, a 25-kDa membrane protein, also resides in the ER. 3 In vma12⌬ cells, Vma22p did not fractionate with membranes as in wild-type cells but rather with soluble cytosolic proteins such as phosphoglycerol kinase. It therefore appears likely that ER membrane association of Vma22p is mediated through an interaction with Vma12p. Although we did not find any perturbation of Vma22p localization in vma21⌬ cells, it will be necessary to extend our analysis of potential interactions between all combinations of Vma21p, Vma12p, and Vma22p. Our results suggest that this set of proteins is necessary for the assembly of the V 0 sector of the V-ATPase complex in the ER. In the future, we will attempt to establish whether Vma21p/ Vma22p/Vma12p exist as a protein complex and also if there are other proteins associated with this complex. Preliminary cross-linking experiments indicate that at least Vma22p and Vma12p physically interact. 4 Because we have a large number of other vma mutants to screen for phenotypes similar to those of vma21/vma22/vma12, we may yet identify additional proteins required for V-ATPase assembly.
What could the function be of Vma22p/Vma12p/Vma21p in V-ATPase assembly? We have not ruled out a role for these proteins in the correct insertion of the 100-kDa integral membrane protein into the ER. As yet we have no data relating to the topology of the V 0 sector polypeptides in either wild-type or vma mutant cells. While it is too early to be confident, we favor an alternative model in which the Vma22/Vma21/Vma12 proteins act as chaperones, allowing the proper assembly of the V 0 subunits after their insertion into the ER membrane.
The requirement for additional ER proteins for folding, insertion into the ER, or oligomerization has been observed for membrane proteins other that the 100-kDa V-ATPase subunit. Shr3p, an ER protein, is required for the exit of amino acid permeases from the ER in yeast (Ljungdahl et al., 1992). Shr3p is proposed to act either in a chaperone-like role in the folding of permeases or possibly as a permease-specific component of the ER translocation machinery. In Drosophila, mutations in the cyclophilin homologue ninaA result in a failure to transport two related opsin proteins from the ER of photoreceptor cells (Colley et al., 1991). Finally, the influenza virus HA membrane protein requires BiP for trimerization (Gething et al., 1986), and this oligomerization is required for exit from the ER (Copeland et al., 1988).
In summary, we have identified a protein, Vma22p, which is required for the ER assembly of the V 0 sector of the V-ATPase complex. Failure to assemble the V 0 sector leads to rapid degradation of the 100-kDa polypeptide in the ER and failure to assemble the V-ATPase enzyme complex. Vma22p, along with Vma21p and Vma12p, may form an ER assembly complex specific for the V-ATPase enzyme.