Vma9p (Subunit e) Is an Integral Membrane V0 Subunit of the Yeast V-ATPase*

The Saccharomyces cerevisiae vacuolar proton-translocating ATPase (V-ATPase) is composed of 14 subunits distributed between a peripheral V1 subcomplex and an integral membrane V0 subcomplex. Genome-wide screens have led to the identification of the newest yeast V-ATPase subunit, Vma9p. Vma9p (subunit e) is a small hydrophobic protein that is conserved from fungi to animals. We demonstrate that disruption of yeast VMA9 results in the failure of V1 and V0 V-ATPase subunits to assemble onto the vacuole and in decreased levels of the subunit a isoforms Vph1p and Stv1p. We also show that Vma9p is an integral membrane protein, synthesized and inserted into the endoplasmic reticulum (ER), which then localizes to the limiting membrane of the vacuole. All V0 subunits and V-ATPase assembly factors are required for Vma9p to efficiently exit the ER. In the ER, Vma9p and the V0 subunits interact with the V-ATPase assembly factor Vma21p. Interestingly, the association of Vma9p with the V0-Vma21p assembly complex is disrupted with the loss of any single V0 subunit. Similarly, Vma9p is required for V0 subunits Vph1p and Vma6p to associate with the V0-Vma21p complex. In contrast, the proteolipids associate with Vma21p even in the absence of Vma9p. These results demonstrate that Vma9p is an integral membrane subunit of the yeast V-ATPase V0 subcomplex and suggest a model for the arrangement of polypeptides within the V0 subcomplex.

(vacuolar membrane ATPase) genes and two additional genes VPH1 and STV1 encode yeast V-ATPase subunits. Yeast that lacks any VMA gene display characteristic Vma Ϫ phenotypes as follows: nonacidified vacuoles and the failure to grow on media buffered to pH 7.5 or containing elevated CaCl 2 . The V-ATPase can be divided into two subcomplexes designated V 1 and V 0 . The V 1 subcomplex is composed of eight hydrophilic subunits ranging in size from 13 to 69 kDa and contains the ATP binding and hydrolysis domains. The V 0 subcomplex is hydrophobic, composed of six subunits ranging in size from 10 to 100 kDa, and is responsible for proton translocation. The V 0 subcomplex can be subdivided into two parts that move relative to each other, the stator and the proton-translocating ring. The V 0 portion of the stator is composed of the 100-kDa subunit a (Vph1p or Stv1p). The proton-translocating ring is composed of subunits Vma3p, Vma11p, and Vma16p (subunits c, cЈ, and cЉ respectively; see Ref. 14). Vma3p, Vma11p, and Vma16p have multiple transmembrane domains and are termed proteolipids because of their very hydrophobic nature (15). The 5th V 0 subunit Vma6p, subunit d, is a hydrophilic peripheral membrane protein. Cross-linking studies with the Thermus thermophilus V-ATPase suggest that Vma6p interacts with the V 0 subcomplex through the proteolipids (16).
In yeast there are three ER-localized proteins that function as V-ATPase assembly factors, which are required for the function of the V-ATPase but are not components of the final assembled complex (17). The V-ATPase assembly factors Vma12p, Vma21p, and Vma22p are found in the ER where they function to assemble the V 0 subcomplex (18). Vma12p is a transmembrane protein that anchors the peripheral membrane protein Vma22p in the ER (19). This Vma12p-Vma22p complex has been shown to interact with the V 0 subunit a early in V 0 biosynthesis. Vma21p is an integral membrane protein that has been found to associate with the V 0 subcomplex in the ER and may escort it to the Golgi body (20).
An additional small hydrophobic V-ATPase V 0 subunit (subunit e) has been identified in Manduca midgut (21) and bovine chromaffin granule (22) V-ATPase preparations. Recently such a polypeptide has been identified as a likely component of the yeast V-ATPase (23,24). Two different genome-wide yeast mutant screens identified the subunit e encoding gene VMA9 (previously CWH36), which when disrupted resulted in the full spectrum of Vma Ϫ phenotypes (23,25). Subunit e is conserved from yeast to humans, and Caenorhabditis elegans fus-1 (subunit e) mutants show cellular hyperfusion (8). Yeast Vma9p has been found to co-purify with the V-ATPase and is required for V-ATPase function (24).
We also identified the VMA9 gene in a yeast genome-wide screen for proteins involved in V-ATPase function (3). We report here that cells lacking Vma9p do not assemble a V 0 subcomplex, and that the subunit a isoform Vph1p is rapidly degraded. Biochemical and localization studies reveal that Vma9p is an integral membrane vacuolar protein that fails to be transported out of the ER in cells lacking any V 0 subunit or V-ATPase assembly factor. Vma9p associates with the V 0 -Vma21p assembly complex in the ER, and Vma9p is required for the association of two other V 0 subunits with the V 0 -Vma21p complex. These data indicate that Vma9p is a critical component of the V 0 subcomplex, and based on the results we propose a model for the position of Vma9p relative to the proteolipid ring.

EXPERIMENTAL PROCEDURES
Plasmid Construction-Enzymes utilized for DNA manipulations were purchased from New England Biolabs, Fermentas, and Stratagene; standard protocols for molecular biology with modifications were followed (26). The plasmids utilized in this study are shown in Table 1. Plasmid pLG42 was created by HindIII/SacI digestion of VMA21::HA from pKH28 and ligation into pRS315 digested with the same restriction enzymes. To generate pMAC200, genomic DNA was prepared from the yeast genome deletion collection diploid strain cwh36⌬::kan r (renamed vma9⌬::kan r ; Open Biosystems) as described previously (27,28). The genomic DNA was utilized in PCR with inward-facing oligonucleotides designed to bind ϳ500 bp 5Ј of the VMA9 start codon and 500 bp 3Ј of the stop codon. The 500-bp flanking vma9⌬::kan r PCR product was ligated into the pCR4 Blunt-TOPO vector (Invitrogen) generating pMAC200. The same method utilized to create pMAC200 was also used to generate pLG128 (vma3⌬::kan r ), pLG129 (vma11⌬::kan r ), and pLG130 (vma16⌬::kan r ) from the appropriate yeast genome deletion collection diploid strains.
Isolation of Vacuolar Membrane Vesicles-Vacuolar membrane vesicles were isolated as described previously, but without a second 8% Ficoll centrifugation (30,31). Yeast were grown to log phase in YEPD buffered to pH 5.0, and 4000 A 600 units of cells were resuspended in  Yeast was grown to mid-log phase in YEPD, pH 5.0, and cells were centrifuged at 4000 ϫ g. Centrifuged pellets were resuspended in Thorner buffer lacking ␤-mercaptoethanol to 75 A 600 units/ml, and 75 mg of glass beads were added to samples before vortexing. Samples were centrifuged at 4000 ϫ g; supernatants were saved, and protein concentrations were determined by a modified Lowry assay (32). Yeast whole cell lysates were then diluted to 2 mg/ml with Thorner buffer. For Western analysis, samples were loaded into 8 -20% gradient SDS-polyacrylamide gels, separated by electrophoresis, and transferred to 0.2 m of nitrocellulose. Depending on the experiment, blots were probed with the following primary antibodies: monoclonal anti-Vma2p 13D11 (Molecular Probes); monoclonal anti-Vph1p 10D7 (Molecular Probes); rabbit anti-MYC sc-789 (Santa Cruz Biotechnology); rabbit anti-Vma6p, as described previously (33); monoclonal anti-HA HA.11 (Covance); monoclonal anti-ALP 1D3 (Molecular Probes); monoclonal anti-Vma5p 7A2, as described previously (34); rabbit anti-Vma9p C6205 (see next description). New England Peptide Inc. generated the polyclonal anti-Vma9p by injection of synthetic peptide N-CQLH-PLVAPRRSDLRPEFAE-C into rabbits and subsequent recovery of antiserum. The Vma9p antibody was incubated with paraformaldehyde/ formaldehyde-fixed vma9⌬::kan r (MACY10) yeast cells prior to use in Western blot analysis and immunofluorescence microscopy. Secondary antibodies utilized were goat anti-mouse or anti-rabbit conjugated to horseradish peroxidase (Jackson ImmunoResearch). Proteins were visualized by chemiluminescence (Amersham Biosciences).
Extraction of Yeast Membranes with Chaotropic Agents-Yeast membranes were prepared in a similar manner as described previously (37). Strain SF838-1D␣ was grown to mid-log phase in YEPD at 30°C. Cells were treated with 10 mM dithiothreitol ϩ 50 mM Tris, pH 9.5, resuspended in spheroplast buffer (1.2 M sorbitol, 50 mM KPO 4 , pH 7.4, 1 mM MgCl 2 ) ϩ 250 g/ml Zymolyase 100T, and lysed in phosphate-buffered saline (PBS) ϩ protease inhibitors (0.5 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, and 1 g/ml pepstatin) with vigorous pipetting. Unbroken cells were separated from samples by centrifugation at 500 ϫ g. Samples were centrifuged at 13,000 ϫ g for 10 min to pellet membrane fractions; supernatants were removed, and membrane samples were resuspended in PBS. The protein concentration of the membrane samples was determine by the modified Lowry protocol (32). Membrane extractions were performed in a similar manner as outlined previously (33). Briefly, 100 g of yeast membrane protein was washed in TE (10 mM Tris, pH 7.4, 1 mM EDTA) and resuspended in 100 l of TE, 100 mM Na 2 CO 3 , or 0.5% Triton X-100. Membrane treatments were incubated on ice for 30 min and centrifuged at 100,000 ϫ g for 30 min. 100 l of supernatant were removed, and an equal volume of Thorner buffer was added. Pellet fractions were resuspended in 100 l of PBS and 100 l Thorner buffer. Equal volumes of pellet and supernatant samples were utilized for Western blot analysis as described above.
Vma21p Native Immunoprecipitations-Yeast cells bearing the appropriate VMA21 plasmids (see figure legends) were grown overnight in selective media at 30°C. Overnight cultures were used to inoculate YEPD, pH 5.0, media, and cells were grown for two divisions to mid-log phase. To generate immunoprecipitation samples, 20 A 600 units of cells were made into spheroplasts and lysed as described above (see under "Extraction of Yeast Membranes with Chaotropic Agents"). Lysed samples were solubilized for 30 min at 4°C with 1% C 12 E 9 (nondenaturing) and simultaneously pre-cleared with protein A-Sepharose or protein G-Sepharose. Samples were centrifuged at 13,000 ϫ g; supernatants were preserved, and monoclonal anti-HA prebound to protein A-Sepharose or goat anti-HA antibodies (for Vph1p final detection only) was added to the supernatants. Anti-HA protein A-Sepharosetreated supernatants were incubated for 3 h 4°C, centrifuged, and washed repeatedly with PBS, and finally resuspended in 100 l of Thorner buffer (denaturing). Goat anti-HA antibody-treated supernatants were incubated for 3 h at 4°C, incubated for 2 h after addition of protein

Vma9p Analysis
G-Sepharose, centrifuged and washed repeatedly with PBS, and resuspended in 100 l of Thorner buffer. To generate input samples, 4 -8 A 600 units of cells from the starting YEPD, pH 5.0, culture were prepared in the same manner as whole cell lysates (see above). However, 200 l of Thorner buffer was added before vortexing with glass beads. To visualize proteins, samples were subjected to Western blot analysis (see above).

RESULTS
Vma9p Is Required to Assemble the V-ATPase-To identify additional components required for yeast V-ATPase function, the yeast diploid deletion collection was screened for mutants that fail to grow on media buffered to pH 7.5 ϩ 100 mM CaCl 2 . One new mutant was identified, and this mutant carried a deletion of the CWH36 gene. 3 The CWH36 gene is now referred to as gene VMA9, because vma9⌬ yeast cells exhibit characteristic Vma Ϫ phenotypes: no growth at elevated pH, lack of vacuolar acidification in vivo, and a complete lack of V-ATPase activity in vacuole membranes (Refs. 23 and 24). 3 To better understand the role of Vma9p in V-ATPase function, we examined the presence of V-ATPase subunits on the vacuoles of wildtype and vma9⌬ cells. Blots comparing whole cell lysates and isolated vacuoles were probed with antibodies that recognize V 0 and V 1 subunits. The levels of Vma6p, Vma1p, and Vma5p in whole cell lysates were very similar between wild-type and vma9⌬ cells (Fig. 1). The decreased Vph1p levels of vma9⌬ cell extracts have been noted previously (23), although our results reveal a less severe effect. Both V-ATPase V 0 subunits (Vph1p and Vma6p) and V 1 subunits (Vma1p and Vma5p) were significantly depleted from the vacuoles of vma9⌬ yeast. In addition, the V 1 subunits Vma2p, Vma4p, Vma8p, Vma10p, and Vma13p were absent from the vacuoles of vma9⌬ cells (data not shown). As expected, the non-V-ATPase vacuolar membrane protein ALP was present in isolated vacuoles of both wild-type and vma9⌬ cells. These fractionation data are consistent with previous results (25) and indicate that V-ATPase subunits are not present on the vacuoles of vma9⌬ yeast.
Subunit a of the V-ATPase is present in yeast as two isoforms, Vph1p and Stv1p (39). As shown above, the Vph1p vacuolar isoform is expressed at lower levels in vma9⌬ strains as compared with wild type (Fig. 1). To further investigate the effects of Vma9p loss on the subunit a isoforms, Western analysis was performed on yeast whole cell extracts. The level of Vph1p in vma9⌬ cells was compared with yeast lacking a V 1 subunit (vma1⌬), V 0 subunit (vma3⌬), or V-ATPase assembly factor (vma21⌬). The lower Vph1p levels of vma9⌬ yeast were similar to levels observed in vma3⌬ and vma21⌬ strains ( Fig. 2A). Yeast lacking Vma1p had Vph1p levels that were indistinguishable from wild type. The V-ATPase of the late Golgi have Stv1p as their subunit a isoform (40). Like the Vph1p isoform, the level of Stv1p was reduced in vma9⌬, vma3⌬, and vma21⌬ crude cell extracts as compared with wild-type and vma1⌬ extracts (Fig. 2B). These results indicate that Vph1p and Stv1p require a competent V 0 subcomplex, including Vma9p, to be expressed at wild-type levels.
Reduced Vph1p and Stv1p levels in vma9⌬ cells could be due to decreased synthesis and/or stability of the subunit a isoforms. To investigate the stability of subunit a in vma9⌬ cells, Vph1p was immunoprecipitated from radiolabeled wild-type, vma9⌬, and vma3⌬ cells over a time course of 90 min. Immunoprecipitated Vph1p samples were separated on SDS-polyacrylamide gels and detected by autoradiography (Fig. 3A). In wild-type cells the levels of Vph1p remained essentially constant over the chase time, which is consistent with the previously reported Vph1p half-life of greater than 4 -8 h (Fig. 3B) (30,41). In cells lacking a V 0 subunit (vma3⌬), Vph1p was turned over quickly, and its half-life decreased to ϳ30 min. Vph1p was also degraded more rapidly in vma9⌬ cells, exhibiting a half-life of 45-60 min. We conclude from these data that Vph1p stability is decreased significantly in vma9⌬ cells, as was observed previously when cells lack other V 0 subunits (19).

Vma9p Is an Integral Membrane V 0 Subunit That Requires a Fully Assembled V 0 Subcomplex to Exit the ER-Yeast
Vma9p is predicted to be a 73-amino acid protein with two transmembrane domains (Fig. 4A). To study Vma9p a polyclonal peptide antibody was raised against the last 20 amino acids of the protein. Western blot analysis with the Vma9p antibody identified an ϳ10-kDa protein from wild-type cells that was absent in vma9⌬ cells (Fig. 4B). Western blots also revealed that Vma9p was present at wild-type levels in yeast strains that lack a V 1 subunit, V 0 subunit, or V-ATPase assembly factor. These results indicate that unlike Vph1p, Vma9p is stable in cells unable to assemble a V 0 subcomplex (20).

Vma9p Analysis
whether Vma9p is a peripheral membrane or integral membrane protein, membrane fractions from yeast were treated either with a gentle TE wash, the chaotropic agent Na 2 CO 3 , or the detergent Triton X-100. Na 2 CO 3 treatment extracts peripheral proteins from pelleted membranes but not integral membrane proteins that can only be solubilized with detergents (43). Vph1p and Vma6p were utilized as integral and peripheral membrane protein controls, respectively. The integral membrane protein Vph1p was not solubilized by the TE or Na 2 CO 3 treatments and remained with the pelleted membrane fraction (Fig. 4C). As expected, Vph1p fractionated with the supernatant only after solubilization with 0.5% Triton X-100. The peripheral membrane protein Vma6p was extracted from membranes with both Na 2 CO 3 and Triton X-100 treatments. Similar to Vph1p, Vma9p was extracted into the supernatant fraction only by Triton X-100 treatment. These results demonstrate that Vma9p is an integral membrane subunit of the V-ATPase.
Vma9p has been identified in isolated vacuoles by Western blot analysis (24). The Vma9p-specific antibody was utilized to localize Vma9p by immunofluorescence in wild-type and mutant yeast strains. As expected, Vma9p clearly labeled the limiting membrane of the vacuole in wild-type cells (Fig. 5). Wild-type vacuolar localization of Vma9p was also seen in vma1⌬ yeast, which lack V-ATPase function but have an assembled vacuolar V 0 subcomplex (34). In addition, yeast that lacked other V 1 subunits (vma2⌬, vma4⌬, vma5⌬, and vma8⌬) were observed to have vacuolar Vma9p localization (data not shown). Interestingly, Vma9p localization was perinuclear, encircling the DAPI-labeled nucleus in yeast cells lacking a V 0 subunit (vma3⌬) or V-ATPase assembly factor (vma21⌬). This perinuclear labeling indicates that Vma9p localizes to the ER in yeast that fails to assemble a V 0 subcomplex. Vma9p was also ER-localized in vma6⌬, vma11⌬, vma12⌬, vma16⌬, vma22⌬, and vph1⌬ stv1⌬ V 0 mutant strains (data not shown). These results demonstrate that ER exit of Vma9p requires a fully assembled V 0 subcomplex.
Vma9p Is Found in a V 0 -Vma21p Assembly Complex-The retention of Vma9p in the ER of yeast cells with defective V 0 assembly led us to investigate whether Vma9p associates with the V-ATPase assembly factor Vma21p during normal V 0 assembly. Previous work has shown that the ER-localized V 0 subunits form a complex that interacts directly with the assembly factor Vma21p (Fig. 6A) (20). To determine whether Vma9p associates with the V 0 -Vma21p complex as illustrated (Fig. 6A), an HA epitope-tagged version of Vma21p was immunoprecipitated from wild-type yeast lysates. Vma9p was observed to co-immunoprecipitate with Vma21p-HA from extracts of wild-type cells (Fig. 6B). Vma9p did not precipitate from controls with yeast cells that lack an HA epitope-tagged version of Vma21p. The Vma9p-Vma21p interaction was preserved in vma1⌬ yeast, which lack an assembled V 1 subcomplex yet still assemble the V 0 subcomplex (34). To better understand the relationship between Vma9p and the other V 0 subunits, the Vma9p-Vma21p-HA interaction was examined in a series of mutants lacking individual V 0 subunits. Vma21p-HA was immunoprecipitated from extracts of vma3⌬, vma6⌬, vma11⌬, vma12⌬, vma16⌬, and vph1⌬ stv1⌬ strains, and the precipitated protein fraction was probed for Vma9p. In all V 0 subunit mutant strains examined, the Vma9p-Vma21p-HA interaction was greatly reduced or abolished as compared with wild-type strains. The Vma9p-Vma21p-HA interaction was also disrupted in yeast lacking the assembly factor Vma12p. In summary, Vma9p interacts with a Vma21p assembly complex in the ER, and this interaction was abrogated with the loss of any single V 0 subunit.
To further examine the relationship between Vma9p and the other V 0 subunits, the V 0 -Vma21p interaction in vma9⌬ yeast was investigated. Vma21p-HA was immunoprecipitated from whole cell lysates of  (42). The C-terminal residues in black signify the peptide fragment utilized to generate the Vma9p antibody. B, whole cell lysates were prepared from WT (SF838-1D␣), vma1⌬ (LGY127), vma3⌬ (LGY128), vma21⌬ (LGY129), and vma9⌬ (LGY135) cells. 20 g of whole cell lysate protein was loaded into SDS-polyacrylamide gels to generate Western blots as described under "Experimental Procedures." Blots were probed with antibodies against Vma9p and the loading control Vma2p. C, membrane fractions from WT (SF838-1D␣) cells were treated with buffer only (TE), 100 mM Na 2 CO 3 , or 0.5% Triton X-100. After a 30-min 100,000 ϫ g centrifugation, pellet (P) and supernatant (S) samples were prepared and used to create Western blots. Blots were probed with antibodies against Vph1p (integral membrane protein), Vma6p (peripheral membrane protein), and Vma9p. Vph1p was immunoprecipitated from lysed cells, and samples were loaded into SDS-polyacrylamide gels and separated by electrophoresis, and the dried gel was exposed to a PhosphorImager cassette for detection. B, graphic depiction of Vph1p stability in WT, vma9⌬, and vma3⌬ strains. The 0-min chase time point in all three strains was designated as 100%, and the percent of Vph1p remaining was plotted against time.

Vma9p Analysis
wild-type and vma9⌬ cells, and these fractions were analyzed for the presence of Vma3p, Vma6p, Vma11p, and Vph1p. As expected, all V 0 subunits assayed were found to co-immunoprecipitate with Vma21p-HA from wild-type cells (Fig. 7). Except for Vph1p, loss of Vma9p did not affect the steady state levels of the other V 0 subunits. In yeast lacking Vma9p both the Vph1p-Vma21p-HA and Vma6p-Vma21p-HA interactions were significantly reduced or abolished (Fig. 7, A and B). However, the V 0 proteolipid subunits Vma3p and Vma11p interacted strongly with Vma21p-HA even in the absence of Vma9p (Fig. 7, B and C). These results indicate that the proteolipid proton-translocating ring (Vma3p, Vma11p, and we expect Vma16p) forms a complex with Vma21p in the ER independent of Vma9p, but surprisingly Vph1p and Vma6p association with this complex requires Vma9p.

DISCUSSION
Our results and the work of other groups demonstrate that Vma9p is an integral membrane protein of the V-ATPase V 0 subcomplex (this work and see Refs. 23 and 24). The genetic analysis of the VMA9 gene in yeast and C. elegans provides important in vivo evidence that Vma9p (subunit e) is required for V-ATPase function (8). The embryonic cellular hyperfusion observed in C. elegans fus-1 mutants is also observed when other V-ATPase subunits are eliminated by RNA interference, indicating that the hyperfusion phenotype observed in C. elegans reflects the loss of V-ATPase function and not a specific function of subunit e. The presence of a subunit e homolog associated with the purified bovine chromaffin granule V-ATPase complex provides solid evidence that subunit e is an important component of the mammalian V-ATPase complex as well (22)(23)(24). Plants, fungi, and animals are all predicted to express a small hydrophobic protein homologous to yeast subunit e (Vma9p), indicating that subunit e is likely to be an essential component of the V-ATPase complex in all eukaryotes (8,24,44).
Loss of Vma9p results in the elimination of in vivo acidification of yeast vacuoles (23), and the elimination of all V-ATPase activity associated with vacuolar membranes (24). Biochemical analysis reveals that Vma9p is an integral membrane protein that localizes to vacuolar membranes, and Vma9p co-purifies with the V-ATPase from isolated vacuoles (this work and see Ref. 24). The decreased half-life of Vph1p in vma9⌬ cells is consistent with our previous findings that Vph1p is rapidly degraded in yeast cells lacking a V 0 subcomplex (19,30,41). Taken together, the available data indicate that by all criteria Vma9p behaves biochemically and functionally as a subunit of the V 0 subcomplex (this work and see Ref. 25).
suggests that Vma9p must be incorporated into a competent V 0 subcomplex before ER to Golgi transport can occur. In support of this idea, the V 0 subunit Vph1p (subunit a) fails to exit the ER in cells lacking a proteolipid subunit or a V-ATPase assembly factor (19,30). As an ER quality control mechanism, V 0 subunit retention would ensure generation of only functional V 0 subcomplexes and economical use of individual V 0 subunits.
During the early steps of biosynthesis, Vma9p associates with the V 0 assembly complex bound to the V-ATPase assembly factor Vma21p in the ER (Fig. 6A). We found that all other V 0 subunits needed to be present for Vma9p to associate with this V 0 -Vma21p assembly complex. Interestingly, although Vph1p and Vma6p required the presence of Vma9p to associate with the V 0 -Vma21p complex, the proteolipids associated with Vma21p independent of Vma9p. Similarly, we previously reported that the proteolipids, Vma3p and Vma16p, associate with Vma21p independent of Vph1p and Vma6p (20). These results lead us to the model for the V 0 -Vma21p complex illustrated in Fig. 6A. Vma9p, like Vph1p, would be placed outside the proteolipid proton-translocation ring, which appears to directly interact with Vma21p through Vma11p (20).
What role does Vma9p (subunit e) play in the V 0 subcomplex of the V-ATPase? One clue comes from the fact that the proteolipid V 0 subunits (Vma3p, Vma11p, and Vma16p) all contain an essential glutamate residue in one of their transmembrane helices (14). These acidic residues are thought to serve as carriers for protons translocating through the membrane (14). There are no acidic residues in the predicted transmembrane regions of Vma9p. Thus, it is unlikely that Vma9p participates with the proteolipids in proton translocation. Another clue comes from the observation that the proteolipids form a complex with Vma21p in the ER independent of Vma9p, yet the association of Vma9p and Vph1p (20) with this Vma21p-V 0 complex is absolutely dependent on the presence of all three proteolipids. Taken together with the interdependence of Vma9p and Vph1p for their association with the V 0 -Vma21p assembly complex, these observations have led us to propose that Vma9p (subunit e) together with Vph1p (subunit a) constitute the stator that interfaces with the rotating proteolipid ring (Fig. 7D). Because of the Vma9p/Vph1p/Vma6p interdependence, we have positioned Vma9p with contacts to Vph1p and Vma6p, although the presence of such direct interactions has not yet been tested experimentally. Establishing the close contact interactions between Vma9p/subunit e and the other V 0 subunits remains an important goal for the future.