The amino-terminal domain of the E subunit of vacuolar H(+)-ATPase (V-ATPase) interacts with the H subunit and is required for V-ATPase function.

Vacuolar H(+)-ATPases (V-ATPases) are highly conserved proton pumps that couple hydrolysis of cytosolic ATP to proton transport out of the cytosol. Although it is generally believed that V-ATPases transport protons by a rotary catalytic mechanism analogous to that used by F(1)F(0)-ATPases, the structure and subunit composition of the central or peripheral stalk of the multisubunit complex are not well understood. We searched for proteins that bind to the E subunit of V-ATPase using the yeast two-hybrid assay and identified the H subunit as an interacting partner. Physical association between the E and H subunits of V-ATPase was confirmed in vitro by precipitation assays. Deletion mapping analysis revealed that a 78-amino acid fragment at the amino terminus of the E subunit was sufficient for binding to the H subunit. Expression of the amino-terminal fragments of the E subunits from human and yeast as dominant-negative mutants resulted in dramatic decreases in bafilomycin A(1)-sensitive ATP hydrolysis and proton transport activities of V-ATPase. Our data demonstrate the physiological significance of the interaction between the E and H subunits of V-ATPase and extend previous studies on the arrangement of subunits on the peripheral stalk of V-ATPase.

Vacuolar H ϩ -ATPases (V-ATPases) 1 energize and acidify intracellular compartments of the vacuolar system of eukaryotic cells. They are essential for the normal function of secretory vesicles, the trans-Golgi network, endosomes, lysosomes, the yeast vacuole, and other intracellular membrane compartments (1,2). In some specialized cells such as the intercalated cells of the kidney and the osteoclasts, V-ATPases reside at high levels on the plasma membrane, where they are responsible for transepithelial or cellular proton transport required for normal acid-base homeostasis and bone remodeling (2). Despite their wide range of physiological functions, V-ATPases share a highly conserved structure and common enzymatic properties that couple hydrolysis of cytosolic ATP to proton transport out of the cytosol (3). They contain two macrodomains or sectors: V 1 , a catalytic domain composed of peripheral membrane proteins, and V 0 , a transmembrane domain composed of intrinsic membrane proteins that transmits protons through the lipid bilayer (4). The V 1 domain attaches to the V 0 domain at the cytoplasmic face of the membrane. In Saccharomyces cerevisiae, the V 1 domain is composed of eight distinct polypeptide chains, and the V 0 domain contains five (4).
V-ATPases are evolutionarily related and structurally similar to F 1 F 0 -ATPases (F-ATPases) of bacteria, chloroplasts, and mitochondria (3). F-ATPases also have two sectors: F 1 , a peripherally attached complex composed of a catalytic head and a stalk, and F 0 , composed of intrinsic membrane subunits and a stator arm. F-ATPases have a rotary catalytic mechanism (5)(6)(7). The proton electrochemical gradient across the membrane drives translocation of protons through a pathway composed of the a and c subunits in the F 0 sector, thought to drive rotation of a "wheel" of c subunits and consequently rotation of the attached F 1 stalk. Rotation of the F 1 stalk produces conformational changes in the catalytic head of F 1 , driving ATP synthesis. A stator arm, composed of long cytosolic extensions of the F 0 b subunit attached to the F 1 ␦ subunit, holds the ␣ 3 ␤ 3 catalytic head of F 1 in place against the rotation of the central stalk.
Although the structure of V-ATPase remains less well defined, it is generally believed that the mechanical coupling of the V 1 and V 0 complexes occurs by a mechanism analogous to that of F-ATPase. Electron microscopy analysis confirms the presence of a central shaft and a peripheral stalk in bacterial and bovine V-ATPases (8,9). The A and B subunits of V-ATPase are arrayed as a hexagon around a central stalk. The A subunit shares homology with the ␤ subunit of F-ATPase and is the site of ATP hydrolysis. The B subunit, a homolog of the F 1 F 0 ␣ subunit, may have a regulatory role in ATP hydrolysis. There are two isoforms of the B subunit that are encoded by different genes and have unique amino-and carboxyl-terminal sequences. The proteolipid subunits of V-ATPase are also homologous to the corresponding subunits of F-ATPase that form the proton pore. Other subunits of V-ATPase have limited or no amino acid sequence similarity to any subunits of F-ATPase. Although they are believed to function structurally in connecting the V 1 and V 0 domains either on the central or peripheral stalk, the subunit composition, structure, and subunit interactions of the V 1 stalk and the putative stator arm of V-ATPases are not well understood.
Recent studies have begun to reveal some of the subunit interactions of the V-ATPase stalk. Using yeast two-hybrid assay and co-immunoprecipitation, Landolt-Marticorena et al. (10) showed that the a subunit of V-ATPase interacts with the H and A subunits. Arata et al. (11) showed that the B and E subunits of V-ATPase are in close proximity to each other by introducing unique cysteine residues into a cysteine-less form of the B subunit of V-ATPase and cross-linking purified vacuolar membranes. Their findings suggest that the E subunit forms part of the peripheral stalk of V-ATPase.
In this study, we report the site of interaction and physiological significance of the interaction between the V-ATPase E and H subunits. A 78-amino acid (aa) domain from the amino terminus of the E subunit of V-ATPase is required for binding to the H subunit and for maintaining proper V-ATPase function. Detection of the interaction between the E and H subunits of V-ATPase suggests that both subunits form part of the peripheral stalk of V-ATPase that anchors on V 1 via binding to the B subunit and on V 0 via binding to the a subunit. Yeast Two-hybrid Assays-The 1.2-kb NcoI-SmaI cDNA fragment containing the entire coding region of the human E subunit of V-ATPase was isolated from a Bluescript SK recombinant plasmid and cloned in-frame into the multiple cloning site on the yeast expression vector pAS2-1 (Clontech, Palo Alto, CA). PJ69 yeast cells harboring three reporter genes (12) were prepared by treatment with lithium acetate (LiAc). 0.1 g of DNA was added to 100 l of competent cells, followed by addition of 600 l of sterile polyethylene glycol/LiAc solution (0.1 M LiAc, 10 mM Tris (pH 7.5), and 1 mM EDTA). The transformation mixture was incubated at 30°C for 30 min with shaking, treated with 10% Me 2 SO, heat-shocked at 42°C for 15 min, and spread onto 100-mm plates of minimal synthetic medium lacking tryptophan. Colonies of yeast cells were allowed to grow for 4 days and harvested for subsequent transformation with a human kidney cDNA expression library in the vector pACT2 (Clontech). After selection for growth on Ade Ϫ /Leu Ϫ / Trp Ϫ triple-dropout plates, positive clones were streaked on His Ϫ /Leu Ϫ / Trp Ϫ dropout plates in the presence of 1 mM 3-aminotriazole and assayed for ␤-galactosidase activity.

Materials
Processing of the positive clones was performed as described (13). The selected two-hybrid colonies were grown overnight in yeast/peptone/dextrose medium to stationary phase in polypropylene tubes. 1 ml of the overnight culture was transferred to a microcentrifuge tube and centrifuged for 30 s at room temperature. The supernatants were removed, and the cell pellets were resuspended in 200 l of breaking buffer (2% Triton X-100, 1% SDS, 100 mM NaCl, 10 mM Tris (pH 8), and 1 mM EDTA). 0.3 g of glass beads and 200 l of phenol/chloroform/ isoamyl alcohol (25:24:1) were added to the resuspended cells. The samples were vortexed for 2 min and centrifuged for 5 min at room temperature. The aqueous upper phase was carefully collected, treated with 10 units of RNase A for 10 min at 37°C, and applied to a QIAGEN mini-preparation spin column. After centrifugation for 1 min, the columns were washed twice with 70% ethanol, and DNA was eluted from the columns with 10 mM Tris (pH 8) and 1 mM EDTA. Recombinant plasmids were recovered either by electroporation in DH5␣ bacteria (Invitrogen) or by high-fidelity PCR amplification (Roche Molecular Biochemicals) using primers 5Ј-CTATTCGATGATGAAGATACCCCAC-CAAACCCAAAAAAAGAG-3Ј and 5Ј-GTGAACTTGCGGGGTTTTTCA-GTATCTACGAT-3Ј flanking the cloning site on the yeast expression vector and the Expand Long enzyme system, which generates mutations at a rate of only one nucleotide in every 10 kb of DNA. DNA sequencing was carried out by incubation with BigDye terminators and read by an automatic DNA sequencer (PE Applied Biosystems, Foster City, CA). The derived DNA sequences were analyzed by searching the GenBank TM /EBI Data Bank for homology to known DNA and peptide sequences.
Glutathione S-Transferase (GST) Precipitation Assays-The entire coding region of the human V-ATPase E subunit cDNA was amplified by high-fidelity PCR as described above using the forward primer 5Ј-AATTGGATCCATGGCTCTCAGCGATGCTGACGTGC-3Ј and the reverse primer 5Ј-GAGAGAATTCGTCCAAAAACTTCCTGTTGGCATTT-GC-3Ј. The PCR-amplified product was purified with a QIAEX II kit (QIAGEN Inc.), digested with the restriction enzymes BamHI and EcoRI (Invitrogen) using the manufacturer's buffer solution, purified again with the QIAEX II kit, and cloned in-frame into the GST expression vector pGEX-2TK (Amersham Biosciences). Clones containing the fusion construct were identified by DNA mini-preparation and restriction enzyme digestion and verified by DNA sequencing as described (14). Bacterial colonies were inoculated in LB medium containing ampicillin and grown for 15 h at 37°C in a shaking incubator. The cultures were diluted 1:10 and grown for 4 h; isopropyl-␤-D-thiogalactopyranoside was added to a final concentration of 0.2 mM; and incubation was continued for 1 h. Bacteria were pelleted by centrifugation at 10,000 ϫ g for 2 min and resuspended in ice-cold phosphate-buffered saline. Cell lysis was carried out by sonication (Sonic Dismembrater, Fisher) for 2 ϫ 30 s. Triton X-100 was added to a final concentration of 1% to minimize aggregation of the fusion protein with bacterial proteins. Samples were centrifuged at 10,000 ϫ g for 5 min, and the supernatants were collected, mixed with a 50% slurry of glutathione-agarose beads (Sigma) in phosphate-buffered saline, and incubated for 5 min at room temperature. The beads were collected by centrifugation, washed with ice-cold phosphate-buffered saline, and stored at 4°C in the presence of bovine serum albumin and protease inhibitors.
The H subunit of V-ATPase from a human kidney cDNA expression library was amplified by high-fidelity PCR using the forward primer 5Ј-GCATGAATTCATGACCAAAATGGATATCCG-3Ј and the reverse primer 5Ј-ACACAAGCTTCAGGCTTAGCTTCGGGCG-3Ј. After restriction enzyme digestion with EcoRI and HindIII, the cDNA fragment was cloned in-frame into the bacterial expression vector pET33b (Novagen, Madison, WI). Recombinant DNA clones were identified by restriction enzyme digestion and DNA sequencing. Purified plasmid DNA was used to produce a 35 S-labeled H subunit by in vitro translation using T7 RNA polymerase, [ 35 S]methionine, and the TNT coupled reticulocyte lysate system (Promega, Madison, WI). The 35 S-labeled H subunit was incubated with 2 g of GST-E subunit fusion protein bound to 30 l of glutathione beads (Sigma) in Nonidet P-40 buffer (150 mM NaCl, 1% Nonidet P-40, and 50 mM Tris (pH 8)) at 4°C for 3 h with shaking. The beads were recovered by centrifugation, washed five times with Nonidet P-40 buffer, resuspended in Laemmli sample loading buffer (2% SDS, 10% glycerol, 100 mM dithiothreitol, 60 mM Tris (pH 6.8), and 0.001% bromphenol blue), and boiled for 3 min. After centrifugation, the supernatants were collected and applied to SDS-polyacrylamide gels for autoradiography analysis.
Deletion Mapping Analysis-Deletion mutants of the E and H subunits of V-ATPase were made by PCR using the high-fidelity Long Expand enzyme system and oligonucleotide primers as follows: wildtype Ea, 5Ј-AATTCCATGGTGGCTCTCAGCGATGCTGAC-3Ј; wildtype Eb, 5Ј-ATATGGATCCGTCCAAAAACTTCCTGTTGGC-3Ј; E-Na, H-Cb, 5Ј-ACACGAATTCAGGCTTAGCTTCGGGCG-3Ј. Amplified cDNA fragments were digested with restriction enzymes; purified; and cloned into the pAS2-1 and pGEX-2TK expression vectors, respectively. The sequences of all mutants were confirmed by DNA sequencing. Yeast two-hybrid assay and GST precipitation assay were carried out as described above.
Immunoprecipitation-Yeast cells were grown in supplemented minimal medium lacking methionine, harvested by centrifugation at 1000 ϫ g for 5 min, resuspended in pretreatment buffer (0.1 M Tris-HCl (pH 9.0) and 10 mM dithiothreitol), and incubated for 5 min. Spheroplasts were generated by treatment with zymolyase 100T for 20 min in buffer containing 1% glucose, 1 M sorbitol, 50 mM K 2 HPO 4 , and 16 mM citric acid (pH 5.8); labeled with 35 [S]methionine for 60 min; lysed in solubilization buffer (10 mM Tris-HCl (pH 7.5), 10% glycerol, 1 mM EDTA, 2 mM dithiothreitol, and 1% C 12 E 9 ) for 15 min; and incubated with anti-B subunit antibody 13D11 for 60 min. After another 60-min incubation with protein A-agarose beads, the immunoprecipitates were collected by centrifugation and washed three times and incubated for 5 min at 70°C in 50 l of cracking buffer (50 mM Tris-HCl (pH 7.0), 8 M urea, 5% SDS, and 5% ␤-mercaptoethanol) for SDS-PAGE and autoradiography analysis. Quantitation of the V 0 and V 1 subunits was performed using a densitometer.
Purification of Vacuolar Membranes-Vacuolar membrane vesicles were prepared from mid-log phase yeast cells for examination of ATPase activity and proton transport. Cells were treated with zymolyase (15), and the spheroplasts were resuspended in Buffer A (12% Ficoll 400, 0.1 mM MgCl 2 , and 10 mM Mes/Tris (pH 6.9)); homogenized in a loosely fitting Dounce homogenizer with 15-20 strokes; and centrifuged in a swinging bucket rotor at 4500 ϫ g for 10 min to remove unlysed cells, mitochondria, and nuclei. The supernatants were transferred to clean centrifuge tubes, and Buffer A were layered on top. The tubes were centrifuged at 52,000 ϫ g. The white layer on top of the tubes containing the vacuoles was collected and resuspended in Buffer A with a homogenizer. Buffer B (8% Ficoll 400, 0.1 mM MgCl 2 , and 10 mM Mes/Tris (pH 6.9)) was layered on top. After recentrifugation under the same conditions, vacuoles free from contaminating lipid granules and other membranous organelles were recovered from the top of the tubes. Purified vacuolar membrane vesicles were prepared for experimental studies by dilution in Buffer C (25 mM KCl, 5 mM MgCl 2 , and 10 mM Mes/Tris (pH 6.9)).
ATPase and Proton-translocating Activities-Bafilomycin A 1 -sensitive ATP hydrolysis of vacuolar membranes was assayed by measuring the production of inorganic phosphate using a modified Ames assay as described by Taussky and Shorr (16). Briefly, vacuolar membrane vesicles in ATPase buffer (150 mM NaCl, 2 mM MgCl 2 , 1 mM vanadate, and 1 mM azide (pH 6.75)) were preincubated for 15 min at room temperature in the presence and absence of bafilomycin A 1 . The reaction was initiated by addition of ATP at a final concentration of 3 mM and stopped by addition of trichloroacetic acid. The samples were extracted with chloroform to remove lipid and detergent. After centrifugation, the upper aqueous phase was transferred to clean test tubes and incubated with buffers containing ascorbic acid and ammonium molybdate. The concentration of inorganic phosphate was measured by a spectrophotometer at 700 nm and converted to rate of ATP hydrolysis.
Proton transport activity was measured by ATP-dependent quenching of acridine orange using a PerkinElmer Life Sciences fluorescence spectrometer with excitation at 493 nm and emission at 545 nm (17). Purified vacuolar membrane vesicles were resuspended in 25 mM Tris (pH 7.2), 25 mM KCl, 5 mM acridine orange, and 5 mM MgCl 2 in the presence and absence of bafilomycin A 1 . ATP was added at a final concentration of 5 mM to initiate transport.

Detection of the Interaction between the E and H Subunits of V-ATPase-
We previously cloned the gene coding for the E subunit of V-ATPase (18) and made monoclonal antibodies against its protein product (19). To search for protein binding partners of the E subunit, we performed yeast two-hybrid assays using the E subunit as the bait, selecting from a human kidney cDNA expression library. One of the first cDNA clones identified as interacting with the E subunit was the H subunit of V-ATPase (13). This finding is consistent with a previous report indicating that the E subunit could be cross-linked to the H subunit of V-ATPase from bovine clathrin-coated vesicles (20).
In a yeast two-hybrid assay, true positive colonies express selectable reporter genes only when the bait fusion protein interacts with the target fusion protein. False-positive colonies carry plasmids that do not encode hybrid proteins interacting directly with the bait protein and express reporter genes by other mechanisms (21). The use of three different reporter genes (ADE, HIS3, and lacZ) under the control of different promoters eliminates many false-positive colonies, particularly from proteins that do not bind to the bait fusion protein, but instead interact directly with promoter sequences flanking the Gal4-binding site or with proteins bound to the flanking sequences (12). Nevertheless, putative true positive clones should be independently tested for binding to the bait protein.
To confirm the interaction between the E and H subunits of V-ATPase, we performed precipitation assays with GST-E subunit fusion proteins. The coding sequence for the E subunit was ligated in-frame to the coding sequence for GST in a bacterial FIG. 1. Physical association between the E and H subunits of V-ATPase. The E subunit of V-ATPase was fused in-frame with the GST gene. After induction, the GST-E subunit fusion protein and the GST native protein were purified using glutathione beads and incubated with the 35 S-labeled H subunit, respectively. After washing, proteins bound to the beads were recovered and analyzed by SDS-PAGE and autoradiography. Physical interaction was detected between the GST-E subunit fusion protein and the H subunit. No binding was observed between the H subunit and the GST native protein.

FIG. 2. A 78-aa domain in the E subunit is required for interaction with the H subunit.
A, deletion mapping analysis of the binding domain in the E subunit using the two-hybrid assay. Overlapping subfragments of the E subunit were fused in-frame with the Gal4 DNA-binding domain. The recombinant plasmids were cotransformed with the H subunit fused in-frame with the Gal4 activation domain. Protein-protein interactions were examined by selection of transformed cells for growth on dropout plates. Growth (ϩ) or no growth (Ϫ) associated with expression of the cDNA fragments is shown on the right. B, confirmation of binding by the 78-aa domain of the E subunit to the H subunit. The 78-aa amino-terminal domain of the E subunit was fused in-frame with the GST protein. The resulting fusion protein was purified using glutathione beads along with the GST native protein. The protein-bound glutathione beads were incubated with the 35 S-labeled H subunit. The H subunit was detected in association with the fusion protein containing the 78-aa binding domain, but not with the GST native protein.
expression vector as described under "Experimental Procedures." The resulting GST-E subunit fusion protein was purified from bacteria with glutathione-agarose beads and incubated with the H subunit protein labeled in vitro with [ 35 S]methionine. After extensive washing, the glutathione beads were boiled in sample buffer, and the supernatant was analyzed by SDS-PAGE and autoradiography. Binding of the labeled H subunit to the GST-E subunit fusion protein was evident in the precipitation assay (Fig. 1, first lane). In contrast, no binding to GST was detected (Fig. 1, second lane).
A 78-aa Amino-terminal Fragment of the E Subunit Interacts with the H Subunit-To identify domains in the E and H subunits required for the interaction, we carried out deletion mapping analysis. Two overlapping cDNA subfragments were amplified by high-fidelity PCR from the coding region of the E subunit, designated E-N (aa 1-148) and E-C (aa 132-226) ( Fig.  2A), and fused in-frame with the Gal4 DNA-binding domain in the two-hybrid vector pAS2-1. The resulting fusion plasmids were cotransformed into PJ69 cells with the pACT2 vector containing the H subunit and selected for cell growth on dropout plates. Colonies were observed from cells harboring E-N, indicating that the interaction domain in the E subunit required for interaction with the H subunit resides in the amino terminus of the E subunit from aa 1 to 148. In contrast, no growth of yeast cells was detected on dropout plates when transformed with E-C. Subsequent deletion mapping analysis of the amino terminus of the E subunit revealed a region of 78 aa (aa 1-78), E-N1, as the binding domain required for interaction with the H subunit ( Fig. 2A). As indicated by ␤-galactosidase activity, the interaction between E-N1 and the H subunit generated target gene activation at a level similar to that observed with the full-length E and H subunits. Additional deletions in the 78-aa region, E-N11 and E-N12, failed to bind to the H subunit ( Fig. 2A). Two overlapping deletion mutants covering the entire coding region of the H subunit were also prepared. Neither H subunit fragment showed interaction with the E subunit, suggesting that the entire H subunit might be required to fold to a particular configuration for interaction with the E subunit.
We used GST precipitation assays to examine the H subunitbinding domain at the amino terminus of the E subunit in greater detail. A construct was prepared expressing a chimeric protein consisting of the 78-aa fragment at the amino terminus of the E subunit in tandem with the GST protein. The fusion protein was purified using glutathione beads and incubated with the H subunit protein labeled with [ 35 S]methionine. As shown in Fig. 2B, the H subunit bound directly to the fusion protein containing the 78-aa amino-terminal fragment of the E subunit.
The Amino-terminal Domain of the E Subunit Is Required for ATP Hydrolysis and Proton-translocating Activities-To investigate the significance of the interaction between the V-ATPase E and H subunits, we examined the physiological effects of expression of the amino-terminal domain of the E subunit in yeast. The 78-aa fragment at the amino terminus of the human E subunit and the corresponding fragment of the yeast E subunit were separately cloned into the yeast expression vector YC2/CT and used to transform PJ69 cells. Sequence comparison analysis revealed a high degree of homology in this region of the E subunit between human and yeast (Fig. 3). Yeast vacuoles were isolated from the transformed cells as described under "Experimental Procedures." Bafilomycin A 1 -sensitive ATPase and proton transport activities were measured on pre- FIG. 4. Bafilomycin A 1 -sensitive ATP hydrolysis and proton transport of V-ATPase. A, vacuolar membrane vesicles were isolated from yeast cells harboring the empty vector (Wild-type), the full-length human E subunit (Wild-type HE), the 78-aa amino-terminal domain of the human E subunit (HE78), the full-length yeast E subunit (Wild-type YE), and the corresponding 83-aa domain of the yeast E subunit (YE83). After preincubation in the presence and absence of bafilomycin A 1 , ATP was added as substrate to start the reaction. Hydrolytic activity was assayed for 30 min. The reaction was stopped by addition of trichloroacetic acid. The samples were extracted with chloroform to remove lipid and detergent. After centrifugation, the upper aqueous phase was transferred to clean test tubes and incubated with buffers containing ascorbic acid and ammonium molybdate. The concentration of inorganic phosphate was measured by a spectrophotometer at 700 nm and converted to rate of ATP hydrolysis. B, proton-pumping activity was measured by tracking the ATP-dependent quenching of acridine orange using a fluorescence spectrometer with excitation at 493 nm and emission at 545 nm (17). Purified vacuolar membrane vesicles were resuspended in 25 mM Tris (pH 7.2), 25 mM KCl, 5 mM acridine orange, and 5 mM MgCl 2 in the presence and absence of bafilomycin A 1 . ATP was added at a final concentration of 5 mM to initiate transport. Proton transport rate was determined as the slope of the quench that gave a linear response during the first 10 s immediately after addition of ATP.

FIG. 3. Alignment of the amino-terminal domain of the E subunit of V-ATPase.
The 78-aa amino-terminal domain of the human E subunit is aligned with its counterpart from the yeast E subunit (Vma4). Amino acid residues that are conserved are shown as a consensus at the bottom. Residues that represent conservative substitutions are indicated by a plus sign. Note that 50% overall homology was observed between the human and yeast E subunits. pared vacuolar membrane vesicles. As shown in Fig. 4, a marked decrease in both ATPase activity and proton translocation was observed in cells expressing the human or yeast E subunit amino-terminal fragment. As expected, expression of the wild-type yeast E subunit (Vma4) using the same vector resulted in no noticeable changes in ATPase activity and proton transport. The reduction in ATPase activity correlated well with decreases in proton transport (Fig. 4), suggesting no disruption of coupling between the V 1 and V 0 domains.
A Truncated Form of the E Subunit Is Unable to Reconstitute V-ATPase Function-Although expression of the amino-terminal domain of the E subunit resulted in dramatic decreases in ATP hydrolysis and proton-pumping activities, significant levels of residual V-ATPase activities remained in yeast cells harboring the dominant-negative mutants (Fig. 4). The residual V-ATPase activities could be due to insufficient levels of expression of the dominant-negative mutants. Alternatively, disruption of the interaction between the E and H subunits of V-ATPase might lead to only partial ablation of V-ATPase function, and other protein-protein interactions might contribute to the residual V-ATPase activities. To distinguish these two possibilities, a truncated form of the E subunit of yeast V-ATPase lacking the amino-terminal domain was cloned into the YC2/CT expression vector and introduced by transformation into a mutant yeast strain in which the E subunit had been deleted by gene knockout experiments. Background levels of ATP hydrolysis were observed in the mutant yeast cells expressing the truncated form of the E subunit. This finding suggests that the residual V-ATPase activities detected in the yeast cells expressing the dominant-negative mutants represent partial disruption of the interaction between the E and H subunits (Fig. 5).

Expression of the Amino-terminal Domain of the E Subunit Results in a Phenotype Analogous to That Displayed by H Subunit Deletion Mutants-Deletion of the E subunit of V-
ATPase disrupts V-ATPase assembly and results in decreased ATPase and proton transport activities (22). In contrast, H subunit deletion mutants show normal V-ATPase assembly, but complete ablation of ATPase and proton transport activities (23). It has been suggested that deletion of the H subunit leads to a conformational change in the V-ATPase holoenzyme that renders the proton pump nonfunctional. To examine the mechanism of inhibition by the E subunit amino-terminal domain, we carried out immunoprecipitation analysis using the anti-B subunit antibody. As shown in Fig. 6A, the cellular content and assembly of V-ATPase were normal in cells ex-pressing the amino-terminal domain of the E subunit. These results indicate that the cells expressing the amino-terminal fragment display a phenotype similar to that observed in H subunit deletion mutants, which also display profound inhibition of ATPase and proton transport activities, but with normal V-ATPase assembly (23).
Although no obvious effect on the subunit composition of V-ATPase was observed (Fig. 6A), V-ATPase may not function normally in cells expressing the truncated E subunit. As indicated by decreased V-ATPase activities (Fig. 4), overexpression of the truncated E subunit may block the binding site for the full-length E subunit on the H subunit and disrupt the normal interaction between the E and H subunits. Incorporation of the truncated E subunit on V-ATPase (Fig. 6B) may lead to a conformational change in the V-ATPase holoenzyme, resulting in decreased ATP hydrolysis and proton-translocating activities (Fig. 4). Because the H subunit also binds the a subunit (10), blocking the binding site for the full-length E subunit is not expected to displace the H subunit and to cause global changes in V-ATPase subunit composition. Consistent with this notion, the tagged H subunit was detected in V-ATPase immunoprecipitated using anti-B subunit antibody 13D11 (Fig. 6C). No major changes in the assembled H subunit were observed in the presence and absence of the N-terminal fragment of the E subunit (Fig. 6C). DISCUSSION V-ATPases and F-ATPases share the common structural elements of a peripheral catalytic head attached by a stalk domain to an intrinsic membrane domain for proton translocation. In each, the catalytic head is composed of a hexagon of three heterodimers of homologous subunits, and proteolipid subunits compose an essential part of the proton translocation domain. On the basis of these observations, it is believed that V-ATPases operate by a rotary mechanism analogous to that of F-ATPases. According to the rotary mechanism, ATP hydrolysis in the catalytic head drives rotation of a central stalk that is tightly linked to a ring of proteolipid subunits, each of which has a proton-binding acidic residue in the hydrophobic core. Rotation of the proteolipid ring is thought to induce conformational changes in individual proteolipid subunits as they pass across another membrane subunit, driving proton transport across the membrane.
Subunits of V-ATPase other than those in the catalytic head and the proteolipids have little or no sequence similarity to subunits of F-ATPases, suggesting that their function may be FIG. 5. Reconstitution of V-ATPase function by the wild-type E subunit but not by truncated E subunits. The amino-terminal domain (aa 1-83) and the remaining carboxyl-terminal fragment (aa 84 -233) of the yeast E subunit were cloned into the expression vector YC2/CT. The recombinant plasmid DNA was transformed into an E subunit deletion mutant. Transformation of the wild-type yeast E subunit and an empty expression vector was also carried out as positive and negative controls, respectively. After selection for growth on uracil dropout plates, colonies were picked, cultured overnight, and harvested by centrifugation. Vacuolar membrane vesicles were prepared as described under "Experimental Procedures." Bafilomycin A 1 -sensitive ATPase activity was measured using a modified Ames assay (16). Mutant yeast cells expressing the truncated E subunits showed background levels of ATPase activity, indicating failure in restoration of V-ATPase function. distinct. One approach for determining the function of these V-ATPase subunits is to identify their arrangement in V-ATPase through studies of their interaction with other subunits. Although the exact location of the E subunit in the V-ATPase complex is not known, cross-linking studies suggest that the E subunit is in close proximity to the B, D, and H subunits of V-ATPase (11,20).
We used the yeast two-hybrid assay to search for binding partners for the E subunit. Among the clones we identified was a cDNA clone encoding the H subunit of V-ATPase. Because the yeast two-hybrid assay sometimes generates false-positive clones, we carried out a series of experiments to verify independently the E and H subunit interaction and to demonstrate the physiological significance of the interaction. First, the labeled H subunit protein bound to a GST-E subunit fusion protein, but not to GST alone (Fig. 1). Second, deletion mapping analysis showed that the 78-aa fragment at the amino terminus of the E subunit was responsible for binding to the H subunit (Fig. 2). Third, when the 78-aa fragment was expressed in yeast cells, dramatic decreases in ATP hydrolysis and proton transport activities were observed in purified vacuolar membrane vesicles, indicating disruption of V-ATPase function (Fig.  4). Taken together, our results support the notion that the interaction between the E and H subunits is physiologically relevant to V-ATPase structure and function.
The identification of interaction between the E and H subunits of V-ATPase extends previous studies on the arrangement of subunits on the peripheral stalk of V-ATPase. The a subunit (Vph1p) of V-ATPase has no homologous counterpart in F-ATPase and is an integral part of the V 0 domain. The cytosolic amino terminus of the a subunit has been shown to bind to the H subunit (10). Further deletion mapping analysis showed that a truncated form of the H subunit (aa 160 -478) with a deletion of the 159 aa at the amino terminus retains binding to the a subunit (10). It has been proposed that the interaction between the a and H subunits represents the point of contact between V 1 and V 0 (10). By introducing unique cysteine residues into a cysteine-less form of the B subunit of V-ATPase and performing cross-linking on purified vacuolar membranes, Arata et al. (11) showed that the B subunit is in close proximity to the E subunit on the outer surface of the V 1 domain, indicating that the E subunit likely forms part of the peripheral stalk of V-ATPase. The E subunit can be crosslinked with cysteine residues introduced near the top of the B subunit and in a region closer to the membrane, but not within the A 3 B 3 catalytic hexamer.
Although it has yet to be determined whether the B and E subunits interact directly, the data from this and previous studies (10,11) indicate that the peripheral stalk of V-ATPase is composed of the a, E, and H subunits. The discovery of interaction between the E and H subunits of V-ATPase in this study provides a critical missing link that suggests a plausible structure for the peripheral stalk of V-ATPase. From the catalytic head, the B subunit provides an anchor on the peripheral stalk probably through interaction with the E subunit. In turn, the E subunit binds to the H subunit, which anchors on the V 0 domain through interaction with the a subunit. It is unlikely, however, that the V-ATPase peripheral stalk subunits are arrayed linearly. For example, the a subunit of V-ATPase, a component of the peripheral stalk, not only interacts with the H subunit, but also directly interacts with the A subunit of the catalytic head (10). Further work will be needed to determine whether additional subunits of V-ATPase reside on the peripheral stalk and to identify the interacting surfaces on the various subunits.
The E subunit physically associates not only with the H subunit, but also with regulatory protein molecules that are not V-ATPase subunits. The glycolytic enzyme aldolase interacts in vitro and in vivo with the E subunit of V-ATPase (13). Upon immunofluorescent staining, aldolase co-localizes extensively with V-ATPase both in the kidney proximal tubule and in the osteoclast (13). In osteoclasts, cells that are essential for bone remodeling, aldolase undergoes a subcellular redistribution from cytoplasmic vesicles to the ruffled membrane together with V-ATPase following osteoclast activation (13). In yeast, deletion of the aldolase gene results in a profound reduction in cellular V-ATPase content and complete disassembly of V 1 from V 0 (13). Our recent data show that normal levels of V-ATPase content and assembly can be restored by aldolase The immunoprecipitates were analyzed by SDS-PAGE and autoradiography. The V 1 subunits (A, B, and E) and the V 0 subunits (d and c/cЈ) were readily detected. Note that the a subunit (Vph1p) of V 0 of 95 kDa appeared to be cleaved by protease(s), and a degraded product of the a subunit formed a doublet with the A subunit. B, assembly of the 78-aa domain of the human E subunit and the corresponding 83-aa domain of the yeast E subunit into the V-ATPase holoenzyme. V-ATPase was purified by immunoprecipitation using anti-B subunit monoclonal antibody 13D11, denatured, subjected to SDS-PAGE, transferred to nitrocellulose membranes, and probed with antibody against the His tag. C, detection of the tagged H subunit in V-ATPase immunoprecipitates. The yeast V-ATPase H subunit was cloned into the expression vector YES3/CT and transformed into yeast cells harboring the N-terminal domain of the E subunit. Immunoprecipitation was carried out using antibody 13D11. The immunoprecipitates were probed by Western blotting using antibody against the V5 tag. complementation. 2 Taken together, these data indicate a direct interaction between glycolysis, an ATP-and proton-generating pathway, and the E subunit of V-ATPase, an ATP-hydrolyzing and hydroxyl-generating proton pump.
The glycolytic pathway and the RAVE complex have recently been shown to play a role in the expression and assembly of V-ATPase (13,24). Aldolase-and RAV1-deficient cells have a similar phenotype, failing to form intact V-ATPase in response to glucose. It remains to be determined whether the glycolytic pathway and the RAVE complex act coordinately in glucoseinduced assembly and disassembly of V-ATPase. Several other associated proteins interacting with V-ATPase have been identified, including the Na ϩ /H ϩ exchanger regulatory factor (25), the platelet-derived growth factor receptor (26), ␣ 2 ␤ 1 integrin (27), the human immunodeficiency virus protein Nef (28), and the human Papillomavirus protein E5 (26).
In summary, we have demonstrated that the amino-terminal domain of the V-ATPase E subunit interacts directly with the H subunit. Disruption of the E-H subunit interaction results in a marked decrease in both ATP hydrolysis and proton transport activities of V-ATPase. Our findings are consistent with the assignment of the E and H subunits to the peripheral stalk of V-ATPase, which is required for maintaining proper V-ATPase structure and function. The role of the E-H subunit interaction in controlling V-ATPase activity in response to proteins interacting with V-ATPase remains to be determined. Elucidation of the physiological significance of these protein-protein interactions is likely to further advance our understanding of this fundamental proton pump.