VMA12 Encodes a Yeast Endoplasmic Reticulum Protein Required for Vacuolar H+-ATPase Assembly*

The Saccharomyces cerevisiae vacuolar membrane proton-translocating ATPase (V-ATPase) can be divided into a peripheral membrane complex (V1) containing at least eight polypeptides of 69, 60, 54, 42, 32, 27, 14, and 13 kDa, and an integral membrane complex (V0) containing at least five polypeptides of 100, 36, 23, 17, and 16 kDa. Other yeast genes have been identified that are required for V-ATPase assembly but whose protein products do not co-purify with the enzyme complex. One such gene,VMA12, encodes a 25-kDa protein (Vma12p) that is predicted to contain two membrane-spanning domains. Biochemical analysis has revealed that Vma12p behaves as an integral membrane protein with both the N and C termini oriented toward the cytosol, and this protein immunolocalizes to the endoplasmic reticulum (ER). In cells lacking Vma12p (vma12Δ), the 100-kDa subunit of the V0 complex (which contains six to eight putative membrane-spanning domains) was rapidly degraded (t½ ∼ 30 min). Protease protection assays revealed that the 100-kDa subunit was inserted/translocated correctly into the ER membrane of vma12Δ cells. These data indicate that Vma12p functions in the ER after the insertion of V0subunits into the ER membrane. We propose that Vma12p functions directly in the assembly of the V0 subunits into a complex in the ER, and that assembly is required for the stability of the V0 subunits and their transport as a complex out of this compartment.

Yeast cells lacking any one of the VMA genes exhibit a common set of phenotypes, including pH-sensitive growth and a loss of V-ATPase activity (14 -17). Cells lacking a V 1 subunit fail to assemble the remaining V 1 subunits onto the vacuolar membrane (18,19). In these vma mutant cells, unassembled V 1 subunits are stable and cytosolic, whereas V 0 subunits are present in the vacuolar membrane at normal levels (18 -19). In cells lacking a V 0 subunit, no V-ATPase subunits are found associated with the vacuolar membrane; V 1 subunits assemble into a stable complex in the cytosol (20 -21), whereas other V 0 subunits are present at reduced levels (18).
Vma12p is required for V-ATPase function but is not part of the V-ATPase complex (22). Yeast cells lacking Vma12p (vma12⌬ cells) are phenotypically indistinguishable from two other vma mutants, vma21⌬ and vma22⌬ (19,23). Vma21p and Vma22p are endoplasmic reticulum (ER) proteins required for V-ATPase function, yet these polypeptides are not part of the V-ATPase complex (24 -25). vma12⌬, vma21⌬, and vma22⌬ mutant cells behave just like cells lacking a V 0 subunit (11, 22, 24 -25). Vma12p, Vma21p, and Vma22p may function as an assembly complex at the level of the ER, but we presently cannot rule out an earlier function for these proteins, such as translocation or insertion of V 0 polypeptides into the ER membrane.
In the analysis reported here, Vma12p has been found to be an integral membrane protein that resides in the ER. Yeast cells lacking Vma12p (vma12⌬) rapidly degrade the 100-kDa V 0 subunit (Vph1p) at a rate that is independent of vacuolar proteases or secretory traffic exiting the ER. Vph1p is localized to the ER membrane in vma12⌬ cells, where it is translocated and inserted normally. These results indicate that Vma12p functions in assembly of the V 0 membrane complex after V 0 subunits are inserted into the ER membrane. Table I lists the strains of Saccharomyces cerevisiae used in this study. All strains were grown as described previously (11) and transformed using standard techniques (26). The wild type strain, SF838-1D, was transformed with XhoI-SpeI-digested pDJ4 and a Leu ϩ Vma Ϫ colony was selected to create DJY62 (vma12⌬::LEU2, pep4 -3). DJY62 was transformed with a linearized PEP4 fragment, and a Pep ϩ colony was selected (27) to create DJY63 (vma12⌬::LEU2, PEP4).

Strain Construction and Culture Conditions-
SF838-1D and SF838-9D were gene converted to VMA12::HA3 by loop-in of MunI-linearized pDJ19 and loop-out of VMA12 by passage over 5-FOA to create DJY69 and DJY58. These strains were mated to create DJY60, which was transformed with a galactose-inducible HO plasmid to effect mating type switching. A halo assay was performed (28) to identify a/a and ␣/␣ diploids (DJY71 and DJY72), which were mated to form the tetraploid VMA12::HA3 strain (DJY70). SF838-1D was gene converted to VMA12::HA1 by loop-in of MunI-linearized pDJ17 and loop-out of VMA12 by passage over 5-FOA to create DJY68.
Epitope tags were introduced into VMA12 by a two-step process. First, unique BglII sites were introduced into the VMA12 coding region of pDJ1 by making conservative base changes via site-directed mutagenesis (32) at base pair 644 and 386, creating pDJ5 and pDJ7. A 123-base pair BglII fragment encoding three 9-amino acid epitopes of the influenza virus hemagglutinin protein HA1 (33) was cloned from pBJ7122 into pDJ5 and pDJ7 to create pDJ10 (VMA12::HA1) and pDJ12 (VMA12::HA3). High copy and integrating vectors of the VMA12::HA alleles were constructed by cloning the 1.3-kilobase pair KpnI-SacI fragments from pDJ10 and pDJ12 into YEp352 and pRS306 (30), respectively, to create pDJ14 and pDJ16, and pDJ17 and pDJ19.
SDS-PAGE and Western Analysis-Protein extracts were generated and quantitated as described previously (11). Proteins were separated by SDS-PAGE and Western blots probed as described previously (34). Protein bands were visualized using either alkaline phosphate-conjugated (Promega) or horse radish peroxidase-conjugated (Amersham) secondary antibodies.
Vacuolar Membrane Preparation, V-ATPase Activity Assays, and Ex-traction of Vacuolar Membranes by Chaotropic Agents-Vacuolar membranes were prepared and V-ATPase activity assays performed as described previously (18). Vacuolar membranes were treated with either 0.5% Triton X-100, 100 mM Na 2 CO 3 , pH 11.5, or 0.5 M NaCl essentially as described by Bauerle et al. (11). Subcellular Fractionation-Cells were spheroplasted and lysed according to Feldheim et al. (37), except that instead of homogenization, cells were allowed to lyse osmotically on ice for 20 min before the removal of unlysed cells. Lysates were then subjected to subcellular fractionation as described in Horazdovsky and Emr (38). Briefly, lysates were separated into pellet and supernatant fractions after centrifugation at 13,000 ϫ g. The supernatant fraction was then subjected to centrifugation at 100,000 ϫ g to generate pellet and supernatant fractions.
Proteolysis of Microsomal Membrane Fractions-Cells were spheroplasted and lysed as described for subcellular fractionation (37). Microsomes were collected at 30,000 ϫ g and proteolyzed using either 0.1 mg/ml trypsin plus 0.1 mg/ml protease K as described previously (36) or 0.004 mg/ml subtilisin Carlsberg.
Immunoprecipitations--EXPRE 35 S 35 S label was purchased from NEN Life Science Products. Immunoprecipitation of radiolabeled Vph1p and CPY was performed as described previously (24). Briefly, after radiolabeling, spheroplasted cells were lysed in 0.6% SDS plus protease inhibitors and denatured for 5 min at 62°C (Vph1p) or at 100°C (CPY) before dilution and immunoprecipitation with polyclonal anti-Vph1p sera or monoclonal anti-CPY sera.

RESULTS
Fractionation of Vacuolar Membranes-A Kyte and Doolittle hydropathy plot of Vma12p predicts two membrane-spanning domains for this protein (22). To test this prediction, vacuolar membranes from wild type cells were treated on ice with chaotropic agents that solubilize membrane proteins to variable degrees. Treatment with 100 mM Na 2 CO 3 , pH 11.5, selectively solubilizes peripheral membrane proteins. NaCl (0.5 M) does not appreciably solubilize membrane proteins (40). By contrast, 0.5% Triton X-100 extensively solubilizes membrane proteins. After treatment with chaotropic agents, samples were separated into pellet and supernatant fractions by centrifugation at 100,000 ϫ g for 30 min. Pellet and supernatant fractions were then separated by SDS-PAGE and probed for the presence of marker proteins and Vma12p. Vph1p and Vma2p, an integral membrane and a peripheral membrane protein, respectively, fractionated as expected for the nature of their membrane association (Fig. 1). Vma12p fractionated with the membrane pellet when incubated in alkaline Na 2 CO 3 and was completely solubilized in Triton X-100, indicating that Vma12p behaves biochemically like an integral membrane protein.
Although low levels of Vma12p were found previously to fractionate with vacuolar membranes (22), it was unclear whether this resulted from the presence of contaminating membranes. To determine whether Vma12p is a vacuolar membrane protein, the extent of enrichment of Vma12p in vacuolar membranes was quantified (Fig. 2). Proteins from crude extracts and vacuolar membranes were analyzed by quantitative Western blotting (see "Experimental Procedures"). As expected, the V-ATPase subunit Vph1p was enriched 20 -40-fold in vacuolar membranes relative to crude extracts, a level of enrichment typical of other known vacuolar integral membrane proteins tested (data not shown; 39). Vma12p was only enriched 2-fold in the vacuolar membrane fraction compared with crude extract, suggesting that it is probably not a vacuolar membrane protein. The ER resident protein Dpm1p (41) also exhibited a 2-fold enrichment in the vacuolar fraction (Fig. 2). Thus, based on the low level of enrichment, Vma12p is likely to be present in membranes that contaminate the vacuolar membrane fraction.
Subcellular Localization of Vma12p-VMA12 cells were subjected to differential centrifugation to determine the subcellular location of Vma12p. The fractionation method of Horazdovsky and Emr (38) separates Golgi (P100), endosomal (P100), and cytosolic (S100) proteins from ER, vacuolar, and plasma membrane (P13) proteins. In this assay, the cytosolic protein phosphoglycerate kinase was found in the S100 fraction as expected (Fig. 3). The ER membrane protein Dpm1p was found exclusively in the P13 fraction as expected. Vma12p was also found exclusively in the P13 fraction, suggesting that, inasmuch as Vma12p was not enriched in vacuolar membranes, it resides either in the ER or plasma membrane.
Topology of Vma12p in the ER Membrane-It had been shown previously that Vma12p was required for the membrane association of the hydrophilic ER membrane protein Vma22p (25). In the absence of Vma12p, Vma22p is found in the cytosol, suggesting that a portion of Vma12p is exposed to the cytosol and responsible for Vma22p's membrane association. The Kyte and Doolittle hydropathy plot for Vma12p predicts that the N-terminal ϳ60% and the C-terminal ϳ20% are hydrophilic. To identify cytosolic portions of Vma12p with which Vma22p might interact, microsomes from yeast cells expressing either Vma12p-HA1 or Vma12p-HA3 (tagged in the C-and N-terminal domains, respectively) were subjected to partial proteolysis FIG. 1. Fractionation of vacuolar membranes. Vacuolar membranes were treated with 500 mM NaCl, 100 mM Na 2 CO 3 , or 0.5% Triton X-100 as described under "Experimental Procedures." Pellet (P) and supernatant (S) fractions were analyzed by SDS-PAGE, and blots were probed with anti-Vph1p (integral membrane protein), anti-Vma2p (peripheral membrane protein), and anti-Vma12p antibodies.

FIG. 2. Enrichment of Vma12p in vacuolar membranes.
Indicated amounts of crude extract and vacuolar membrane proteins were analyzed by SDS-PAGE and blots probed with anti-Vph1p (vacuolar integral membrane protein), anti-Vma12p, and anti-Dpm1p (endoplasmic reticular integral membrane protein) antibodies. Enrichment was determined by quantifying the amount of vacuolar membrane protein, compared with crude extract protein, required to give an equivalent signal intensity as determined by Western blot analysis.

FIG. 3. Subcellular fractionation of Vma12p.
Wild type cells (SNY28) were separated into lysate (L), P13 (endoplasmic reticulum, vacuolar, and plasma membranes), S100 (cytosol), and P100 (Golgi and endosomal membranes) fractions. Portions of each fraction representing equivalent amounts of starting material were analyzed by SDS-PAGE and blots probed with anti-phosphoglycerol kinase (PGK, cytosol), anti-Dpm1p (ER), or anti-Vma12p antibodies. (36). Digestion products were generated, separated by SDS-PAGE, and probed for marker proteins and Vma12p-HA. Kar2p, a soluble lumenal protein of the ER (44), was not digested under these conditions (Fig. 5A, lanes 1-4), demonstrating that the membranes remained intact throughout the incubation (37). Sec63p, an integral membrane protein of the ER, was cleaved on its cytosolic C-terminal tail to a slightly faster migrating species (Fig. 5A, lanes 5-8), indicating that proteins found in the microsomal membranes were both accessible to proteases and in their proper orientation (37). Vma12p-HA1 was degraded to a faster migrating species, which could only be detected for 15 s using an antibody against the HA epitope (Fig.  5B, lanes 1-4), demonstrating that the C terminus of Vma12p-HA1 was readily accessible to proteases and therefore located on the cytosolic face of the ER membrane. The N terminus of Vma12p-HA3 was also readily accessible to proteases (data not shown), indicating that it is also located on the cytosolic face of the membrane. Analysis with anti-Vma12p antibodies revealed that two Vma12p-HA1 proteolysis products could be detected after 10 min of treatment (Fig. 5B, lane 8). Neither product contained the HA epitope (compare lane 8 to lane 4), supporting the argument that the epitope had been removed from these Vma12p-HA1 proteolysis products (lane 8). Sensitivity of the HA tags to proteolysis supports the model (Fig. 5C) that the N and C termini of Vma12p are on the cytosolic face of the ER membrane.
Stability of Vph1p in vma12⌬ Cells-To investigate the function of Vma12p, we monitored the fate of the 100-kDa V 0 subunit (Vph1p) in vma12⌬ mutant cells. Hirata et al. (22) reported that Vph1p was present in crude extracts of vma12⌬ cells at 5-10-fold reduced levels relative to wild type cells. To investigate a potential role for Vma12p in the stability of Vph1p, a kinetic analysis was performed by immunoprecipitation of Vph1p from lysates of radiolabeled wild type and vma12⌬ cells. The half-life of Vph1p was calculated from data derived from an AMBIS™ quantitation of signal intensity present at each chase time point (Fig. 6). Vph1p is a long-lived protein in wild type cells with a half-life of Ͼ4.5 h (Fig. 6A,  lanes 1-4). In vma12⌬ cells, however, Vph1p exhibited a halflife of 25 min (Fig. 6A, lanes 5-12). Vph1p was not stabilized significantly in vma12⌬ cells lacking vacuolar proteases (pep4 -3) (compare either vma12⌬ plot to the WT (VMA12) plot, Fig. 6C), suggesting that Vph1p is degraded in a nonvacuolar compartment.
The results above suggest that Vph1p is turned over before reaching the vacuole. To investigate this possibility, sec12-4 mutant cells, which are incapable of budding ER-derived transport vesicles at the restrictive temperature (45), were used to determine if Vph1p turnover requires exit from the ER. Vph1p was not stabilized at the restrictive temperature (data not shown), suggesting that Vph1p is degraded in the ER of vma12⌬ cells.
To test whether the residual Vph1p in vma12⌬ mutant cells resided in the ER, Vph1p was immunolocalized by IIF in wild type and vma12⌬ cells. Vph1p was immunolocalized to vacuolar membranes in wild type cells and in yeast cells lacking a V 1 subunit (vma2⌬) (Fig. 7, left two columns), as reported previously (18,46). In strains lacking either a V 0 subunit (vma3⌬) or Vma12p (vma12⌬), however, Vph1p localized to the ER (Fig. 7, right two sets of columns), as revealed by staining that encircled the nuclear 4Ј,6Ј,-diamidino-2-phenylindole staining. These results indicate that Vph1p becomes a short-lived ER membrane protein when the V 0 complex cannot properly as- FIG. 4. Immunolocalization of Vma12p. A, tetraploid VMA12::HA3 (DJY70) cells were fixed, spheroplasted, and stained with monoclonal anti-HA epitope (12CA5) antibody. Cells were viewed using Nomarski microscopy for cell morphology and by epifluorescence microscopy for nuclear and Vma12p staining. Signal from the anti-HA antibody was amplified using a goat anti-mouse, mouse anti-goat, FITC-conjugated goat anti-mouse "sandwich." B, DJY70 cells were treated and stained as in A, but in addition were also stained with polyclonal anti-Eug1p antibodies to determine the staining pattern of a known ER protein.

FIG. 5. Partial proteolysis of Vma12p and a model of Vma12p
topology. Microsomes were prepared from DJY68 (VMA12::HA1) cells as described under "Experimental Procedures" and proteolyzed at 0°C for the times indicated with 0.1 mg/ml each trypsin and protease K. The reactions were stopped by trichloroacetic acid precipitation and analyzed by SDS-PAGE and immunoblotting. Lanes 1-4 in A were probed with anti-Kar2p antibody to determine if the microsomes remained intact throughout the incubation. Lanes 5-8 in A were probed with anti-Sec63p antibody to determine if the microsomes were substrates for proteolysis and if they were in their proper orientation. Lanes 1-4 in B were probed with anti-HA antibody to determine if the HA tag (at the C terminus of Vma12p-HA) was susceptible to proteolysis under these conditions. Similar results were obtained with an N-terminally tagged Vma12p (Vma12p-HA3). Lanes 5-8 in B were probed with anti-Vma12p antibody to determine if Vma12p-HA1 proteolysis products lacking the HA tag were present at later time points. C, model of Vma12p topology based on the results in Fig. 5B. Location of the HA tags shown as solid black triangles. semble due to the loss of either another V 0 subunit or Vma12p.
Translocation of Vph1p in vma12⌬ Cells-One model for Vma12p function consistent with the experimental results is that Vma12p functions as an assembly factor in the ER membrane for the V-ATPase V 0 complex. A second model is that Vma12p plays a specialized role in translocation/insertion of Vph1p into the ER membrane. To test the translocation model, sensitivity of Vph1p to exogenous proteases was monitored in microsomes from wild type and vma12⌬ yeast cells. Vph1p is a polytopic membrane protein (Fig. 8A) for which partially translocated or untranslocated protein would be expected to be differentially sensitive to exogenous proteases as compared with wild type protein. Microsomes from wild type and vma12⌬ cells were subjected to partial proteolysis as described previously, except subtilisin Carlsberg was used at 0.004 mg/ml. A profile of Vph1p proteolysis products was generated, and blots were probed using a polyclonal antibody raised against the Vph1p N-terminal hydophilic domain (24). Identical patterns consisting of ϳ75-, 55-, 20-, and 18-kDa Vph1p proteolytic fragments were generated from wild type and vma12⌬ cells (Fig. 8B). Identical patterns were also observed when trypsin and prote-ase K were used at 0.001 mg/ml each (data not shown). The microsomes remained intact throughout proteolysis, as demonstrated by the stability of Kar2p (Fig. 8C). Proteolysis of Vph1p from both wild type and vma12⌬ cells to an identical fragment pattern argues that Vma12p is not required for the translocation/insertion of Vph1p into the ER membrane.

DISCUSSION
In this work, we have characterized Vma12p in an effort to elucidate its function in yeast cells. Vma12p behaved biochemically like an integral membrane protein. A combination of differential centrifugation and indirect immunofluorescence revealed that Vma12p resides in the ER membrane. Protease protection analysis revealed that the N and C termini of Vma12p are located on the cytosolic side of the ER membrane.
Hirata et al. (22) reported that Vph1p was present at reduced levels in extracts from vma12⌬ cells. A kinetic analysis revealed that Vph1p was long-lived in wild type cells but was degraded rapidly in vma12⌬ cells. Degradation in the vma12⌬ cells was independent of the activity of vacuolar proteases, suggesting that Vph1p was degraded in a nonvacuolar location. , and vma12⌬ (DJY62) cells were fixed, spheroplasted, and stained with anti-Vph1p antibody. Cells were viewed by Nomarski optics for cell morphology and by epifluorescence microscopy for nuclear and Vph1p staining. For vma3⌬ and vma12⌬ cells, Vph1p signal was amplified using a goat anti-rabbit, rabbit anti-goat, FITC-conjugated goat anti-rabbit "sandwich." For WT and vma2⌬ cells, FITC-conjugated goat anti-rabbit secondary antibody was used.
A kinetic analysis was performed in vma12⌬ sec12-4 cells, for which secretory traffic exiting the ER is blocked at the restrictive temperature. Vph1p degradation was unaffected by the sec12-4 mutation, suggesting that Vph1p was degraded in the ER of vma12⌬ cells. Thus, Vma12p joins Vma21p and Vma22p as ER membrane proteins required for V-ATPase assembly (24,25). Based on these results, it appears that Vph1p does not exit the ER when the V 0 complex fails to assemble but instead is quickly degraded. In an assembly factor mutant or a mutant lacking a V 0 subunit (vma3⌬), the low steady-state level of Vph1p is localized to the ER. These results support a function for Vma12p in the assembly of the membrane sector of the V-ATPase in the yeast ER.
One hypothesis is that Vma12p functions in the translocation/membrane insertion of subunits of the V-ATPase membrane sector. An alternative hypothesis is that Vma12p functions as a molecular chaperone for subunits of the V-ATPase membrane sector. A growing body of evidence supports certain ER proteins functioning as chaperones for specific families of proteins. NinaA, a cyclophilin homologue in Drosophila, is proposed to encode a chaperone for transport and/or folding of specific rhodopsins in photoreceptor cells (47)(48)(49). SHR3 in yeast is proposed to encode a chaperone specific to amino acid permeases destined for the plasma membrane (50 -51). Our protease protection data revealed that Vph1p is equally sensitive to proteolysis in wild type and vma12⌬ cells. Based on these data, the translocation, insertion, and folding of Vph1p are normal in vma12⌬ cells and thus Vma12p does not function in these very early biosynthetic processes.
The model of Vph1p topology proposed in Fig. 8A is similar to the model recently proposed by Leng et al. (52), but differs on the topological assignment of Vph1p's N terminus and on the number of transmembrane domains. Whereas Leng et al. (52) proposed that the N terminus of Vph1p is lumenal, our data place the N-terminal domain of Vph1p in the cytosol. Leng et al. (52) also proposed that Vph1p contains seven membrane-spanning domains, placing the Vph1p C terminus on the cytosolic side of the membrane. Inasmuch as our preliminary results also place the Vph1p C terminus in the cytosol, 2 we propose that Vph1p has either six or eight transmembrane domains.
All of the data available for Vma12p indicate that this protein plays a critical role in the assembly of the V-ATPase in the ER, after the V 0 polypeptides have been inserted into the ER membrane. However, it is possible that Vma12p has additional roles in V-ATPase biosynthesis. For example, Vma12p could serve to load the assembled V-ATPase complex into ER-derived transport vesicles, as has been proposed for Shr3p (51). Vma12p might also escort the V-ATPase in ER-derived vesicles to the Golgi complex. Certain yeast ER proteins with the sequence -KKXX at their extreme C termini have been shown to exit the ER and be retrieved from an early Golgi compartment (24,(53)(54). Vma21p, for example, has a functional -KKXX retention signal at its C terminus (24). Vma12p has a weak ER retention motif, -KITL, at its extreme C terminus, but it appears that this sequence does not function in Vma12p ER retention, inasmuch as addition of the HA epitope to the C terminus did not compromise function for Vma12p-HA1. If the three known V-ATPase assembly factors form an assembly complex in the ER, then Vma21p could serve to retrieve this complex from post-ER compartments. Vma22p, a peripheral ER membrane protein, was found to associate with the ER membrane in a Vma12p-dependent manner (25), and we predict that Vma22p interacts with a cytosolic domain of Vma12p. We are currently investigating in detail the interactions between Vma12p, Vma21p, and Vma22p and other proteins in the ER membrane. B and C: microsomes were prepared from wild type (SF838-1D) and vma12⌬ (DJY62) cells as described under "Experimental Procedures" and proteolyzed at 0°C for the times indicated with 0.004 mg/ml subtilisin Carlsberg. The reactions were analyzed as in Fig. 5. B, Vph1p partial proteolysis products as detected by an antibody raised against the N-terminal hydrophilic domain of Vph1p. C, partial proteolysis products probed with anti-Kar2p antibody to determine if the microsomes remained intact throughout the incubation.