The Tether Connecting Cytosolic (N Terminus) and Membrane (C Terminus) Domains of Yeast V-ATPase Subunit a (Vph1) Is Required for Assembly of V0 Subunit d*

V-ATPases are molecular motors that reversibly disassemble in vivo. Anchored in the membrane is subunit a. Subunit a has a movable N terminus that switches positions during disassembly and reassembly. Deletions were made at residues securing the N terminus of subunit a (yeast isoform Vph1) to its membrane-bound C-terminal domain in order to understand the role of this conserved region for V-ATPase function. Shrinking of the tether made cells pH-sensitive (vma phenotype) because assembly of V0 subunit d was harmed. Subunit d did not co-immunoprecipitate with subunit a and the c-ring. Cells contained pools of V1 and V0(−d) that failed to form V1V0, and very low levels of V-ATPase subunits were found at the membrane. Although subunit d expression was stable and at wild-type levels, growth defects were rescued by exogenous VMA6 (subunit d). Stable V1V0 assembled after yeast cells were co-transformed with VMA6 and mutant VPH1. Tether-less V1V0 was delivered to the vacuole and active. It retained 63–71% of the wild-type activity and was responsive to glucose. Tether-less V1V0 disassembled and reassembled after brief glucose depletion and readdition. The N terminus retained binding to V1 subunits and the C terminus to phosphofructokinase. Thus, no major structural change was generated at the N and C termini of subunit a. We concluded that early steps of V0 assembly and trafficking were likely impaired by shorter tethers and rescued by VMA6.

fer protons from the cytosol to the extracellular milieu (4,5). In the kidney, plasma membrane V-ATPases of the intercalated cells are critical for regulation of the systemic acid-base balance (5,6). Mutations in human kidney V-ATPase cause distal-renal tubular acidosis (6). V-ATPases at the plasma membrane of osteoclasts are essential for bone resorption, and mutations result in osteopetrosis, a disease characterized by thickening of the bones (1,4,7). Complete loss of V-ATPase activity is lethal in eukaryotes other than fungi (3).
V-ATPases are multisubunit complexes that consist of two domains, V 1 (peripheral) and V 0 (membrane-bound) (1,2). Each of the subunits in the V-ATPase complex is critical for function and V 1 V 0 assembly (8). Deletion of a peripheral V 1 subunit leads to disruption of the entire V 1 domain in yeast. Loss of a V 0 subunit does not affect V 1 assembly but disrupts the entire V 0 domain, which also prevents V 1 from associating with the membrane. An exception is subunit a for which two functional isoforms (Vph1, Stv1) exist in yeast (9). Disruption of subunit a requires disruption of both genes (9).
Eight different subunits (A-H) compose the V 1 domain where ATP hydrolysis takes place at a catalytic hexamer A 3 B 3 (1). Six subunits (a, c, cЈ, cЈЈ, d, e) form V 0 , the membrane intrinsic domain that holds V 1 and forms the path for proton transport via a hydrophobic ring structure (c-ring). V 1 and V 0 subunits contribute to the formation of one central and three peripheral stalks that connect the c-ring and the catalytic hexamer A 3 B 3 (1). ATP hydrolysis drives rotation of the central stalk (connected to the c-ring) (10). Protons are transferred from the cytosol to each subunit of the c-ring and from the c-ring to the other side of the membrane passing through subunit a (11). As many protons, as subunits forming the c-ring, get transferred against a concentration gradient when hydrolysis of three ATP molecules powers 360°rotation.
V-ATPases are related to F-ATP (F 1 F 0 ATP) synthases. Both proteins work as molecular motors (10,(12)(13)(14). It is postulated that the asymmetry imposed by having a 3-fold symmetry in F 1 (and V 1 ) and an apparent 10-fold symmetry in the c-ring of F 0 (and V 0 ) requires energy to be transiently stored. The energy of coupling is thought to be stored in the peripheral (stator) and central (rotor) stalk structures of F 1 F 0 (15)(16)(17). Subunit a is the only peripheral stalk component of the V-ATPase complex that is secured in the membrane (18). It is key for maintaining structural stability when relative rotation of subunits occurs during catalysis. Thus, the tether of subunit a in V 0 could be functionally comparable with the tether of subunit b in F 0 (Escherichia coli), although subunits a (V 0 ) and b (F 0 ) do not share sequence homology.
Subunit a is a 95-kDa protein that consists of two domains that are structurally and functionally distinguishable. The hydrophilic N-terminal domain (ϳ45 kDa) is oriented toward the cytosolic side of the membrane and contains the information necessary to deliver V-ATPases to different compartments (19). The N terminus interacts with multiple V 1 subunits, including the catalytic subunit A (20) and peripheral stalkforming subunits H, C, E, and G of V 1 (18,21). It is through these interactions that the N-terminal domain serves as a stator, which prevents rotation of the A 3 B 3 hexamer during catalysis. The other half of subunit a, the C-terminal domain (ϳ50 kDa), is buried in the membrane by multiple transmembranespanning regions (9). The C-terminal domain interacts with the periphery of the c-ring (22) and contributes to the path for proton transport (11,19) by providing access to cytosolic protons and directing their exit to the luminal side of the membrane.
In contrast to its role as stator during catalysis, the N-terminal domain of subunit a is a movable element that switches positions when V 1 V 0 is regulated by disassembly and reassembly in vivo (1,2,23,24). Inactivation of V-ATPases by disassembly is a rapid response to glucose starvation in yeast (23,25). In the absence of glucose the V-ATPase complex dissociates into three parts: V 1 subunit C, V 1 (without subunit C), and V 0 (23). Disassembly is reversible, and the three components reassociate immediately after glucose addition, restoring ATP hydrolysis and proton transport. As V 1 V 0 disassembles and reassembles, the N-terminal domain of subunit a alternates between V 1 V 0 and V 0 (26,27). In V 1 V 0 it contributes to stabilizing the stator-forming V 1 subunits (1,18). In V 0 , its role has yet to be determined.
As its functional and regulatory roles emerge, it becomes clear that the cytosolic N terminus of V 0 subunit a is key for V 1 V 0 activity, assembly, and regulation. In this study deletions were made at amino acids that connect the N-terminal and C-terminal domains of subunit a Vph1. Shrinking of the tether that anchors subunit a to the membrane harmed assembly of subunit d into V 0 , making yeast cells sensitive to pH (vma growth phenotype). Growth defects were rescued by exogenous VMA6, the gene encoding subunit d. Remarkably, subunit d restored assembly and significant function of V-ATPase proton pumps that had up to 46 residues of the tether removed. Because V 1 V 0 containing tether-less vph1 assembled with peripheral V 1 subunits and with the glycolytic enzyme phosphofructokinase, we concluded that no major structural changes were generated at the N-and C-terminal domains. Early steps of V 0 assembly, and trafficking were likely impaired by shorter tethers and rescued by VMA6. The potential mechanisms by which overexpression of subunit d rescued subunit a deletions are discussed.

Material and Strains
Zymolase 100T was purchased from Seikagaku. Concanamycin A was purchased from Wako Biochemicals. Dithiobis[suc-cinimidyl propionate] was purchased from Pierce, Tran 35 S-label was from MP Biomedicals, and alkaline phosphataseconjugated secondary antibodies were from Promega. The QuikChange mutagenesis kit was from Stratagene. Prestained broad range molecular protein markers, nitro blue tetrazolium, and 5-bromo-4-chloro-3-indolyl phosphate were purchased from Bio-Rad. The antibodies 8B1, 13D11, and 10D7 were from Invitrogen. Ficoll and anti-HA antibody were from United States Biological. All other reagents were from Sigma.

Site-directed Mutagenesis
Mutagenesis was performed using the QuikChange site-directed mutagenesis kit from Stratagene according to manufacturer's protocol. The HA-VPH1 gene cloned in the CEN vector pRS316 (28) was used as template. The primers used for mutagenesis and their complementary oligonucleotides (not shown) are shown in Table 1. Mutations were confirmed by sequencing, and plasmids were used to transform the vph1⌬stv1⌬ strain by the lithium acetate method (29). Transformants were selected on SDϪUra pH 5, their growth phenotype was assessed on SDϪUra plates buffered to pH 5 (described above) and pH 7.5 (pH adjusted with 50 mM MOPS, 50 mM MES).

Immunoprecipitations
Yeast cell cultures (200 ml) were grown overnight up to 1-2 A 600 /ml, harvested, resuspended in 0.1 M Tris⅐HCl, pH 9.4, containing 10 mM dithiothreitol, and rocked at 30°C for 5 min. Cells were washed in 10 mM Tris⅐HCl, pH 7.5, containing 1.2 M sorbitol and converted to spheroplasts by treatment with zymolase (4 units/ml) at 30°C for 20 min. Spheroplasts were gently washed twice in 10 mM Tris⅐HCl, pH 7.5, 2% glucose, and 1.2 M sorbitol. Washed spheroplasts were incubated in YEPD plus 1.2 M sorbitol for 15 min, harvested, and lysed by the addition of The resulting supernatant was incubated with the indicated antibody overnight at 4°C in the presence of PBS containing 5 mg/ml bovine serum albumin and protease inhibitors. The monoclonal antibodies 8B1 (anti-subunit A) and 10D7 (antisubunit a) were used at a concentration of 5 g/ml each per immunoprecipitation and antibody anti-HA at 10 g/ml. A control (antibody alone) which did not have the cell lysate added was incubated in parallel under the same conditions. After a 100-l addition of the 40% suspension protein A-Sepharose, mixtures were incubated on ice for 1 h, and V-ATPase complexes were immunoprecipitated. Beads were spun down and washed twice with immunobuffer (10 mM Tris⅐HCl, pH 8.0, 0.1% Triton X-100 (v/v), 2 mM EDTA). Immunoprecipitated proteins were solubilized from the beads by incubation at 65-70°C for 30 min in cracking buffer (50 mM Tris-HCl, pH 6.8, 8 M urea, 5% SDS, 1 mM EDTA, 5% ␤-mercaptoethanol) at a concentration of 6 -10 A 600 of the original cell culture per l of cracking buffer. Immunoprecipitates were subjected to SDS-PAGE and analyzed by Western blots. Immunoprecipitation of biosynthetically labeled cells was as described before (25). Briefly, cells were grown to mid-log phase (0.5 A 600 /ml) in minimal medium lacking methionine or lacking methionine and the nutritional marker uracil. Cells were converted to spheroplasts by zymolase treatment and biosynthetically labeled with Tran 35 S-label. Labeled spheroplasts were lysed in PBS containing 1% C 12 E 9 and protease inhibitors in the presence of 1 mM dithiobis[succinimidyl propionate]. V-ATPase was immunoprecipitated under nondenaturing conditions (23,25) using antibody anti-HA at the concentration indicated above. Immunoprecipitates were separated by SDS-PAGE using 13% gels. Gels were dried, scanned in a Fuji Scanner 5100, and analyzed using the Multi Gauge and Sigma Plot software. When reversible disassembly was studied, excess unlabeled cysteine and methionine were added immediately after labeling. Chases were performed in the presence of 2% glucose (20 min in YEPD medium plus 1.2 M sorbitol), in the absence of glucose (10 min in YEP medium plus 1.2 M sorbitol), and after glucose readdition (10 min in YEP medium plus 1.2 M sorbitol and 2% glucose added) (25). V-ATPase was immunoprecipitated under nondenaturing conditions using antibody 8B1 to V 1 subunit A. Immunoprecipitated protein was analyzed as described above.

Other Methods
Whole Cell Lysis-Wild-type and mutant cells were grown overnight to mid-log phase in SDϪUra pH 5 medium, harvested, and lysed as described before (30).
Purification of Vacuolar Membranes-Vacuolar membranes were purified by flotation in one Ficoll gradient as described before (30,31). Vacuolar membranes were centrifuged, and pel-lets were solubilized in cracking buffer. Vacuolar protein was subjected to SDS-PAGE using 10% gels and transferred to nitrocellulose membranes as described before (30,31).
Western Blots-Protein was transferred from polyacrylamide gels to nitrocellulose membranes and probed with monoclonal antibodies anti-HA, 10D7, 8B1, and 13D11 and polyclonal antibodies against subunits d, D, and E (kind gifts from Dr. Tom Stevens, University of Oregon, Eugene, OR and Dr. Daniel Klionsky, University of Michigan, Ann Arbor, MI). IgG antimouse and anti-rabbit secondary antibodies were conjugated to alkaline phosphatase, and blots were developed as described before (30,31).
ATPase Activity-ATP hydrolysis was followed spectrophotometrically using an enzymatic assay coupled to oxidation of NADH (340 nm) (32). Each reaction contained 15-50 g of vacuolar membrane protein in the presence and absence of 1 M concanamycin A (33).
Protein Assays-Protein concentration was measured as described by Lowry (34).

The Tether of Vph1 Is Necessary for V-ATPase Function-The
N-terminal domain of subunit a is an important structural and regulatory component of V 1 V 0 (19 -21). Secondary structure prediction of the 409 residues forming the N terminus of subunit a showed primarily ␣-helical (49%) and coiled (39%) composition ( Fig. 1A) (35). Amino acids 362-407, which represent the tether that connects the N-terminal and C-terminal domains, are predicted to be mainly coiled. Although the N-terminal domain is poorly conserved, the residues at its tether exhibit 56% conservation (37% identity) between yeast and mouse isoforms. To address the role of its conserved tether, we deleted the entire region by site-directed mutagenesis (Fig. 1A). VPH1 containing two copies of the antigenic sequence HA immediately after residue Asn-185 (28) was used as the template. Because V-ATPase subunit a is encoded by two structural genes (VPH1 and STV1) which are functional homologues (9), complete deletion of the yeast subunit a requires disruption of both genes. We transformed vph1⌬stv1⌬ yeast cells, which lack subunit a, with the wild-type allele of subunit a (VPH1) and the truncation construct (VPH1-⌬362-407) separately, each cloned in the CEN vector pRS316. Therefore, the only isoform of subunit a present was HA-tagged Vph1 (wild type or mutant) expressed from pRS316 under control of its natural promoter.
Transformants were selected on SDϪUra plates adjusted to pH 5 because disruption of yeast V-ATPase function results in a distinctive vacuolar membrane ATPase (vma) phenotype that displays growth sensitivity to pH (36). Vma mutants grow at pH 5 but cannot grow at neutral pH. We assessed the effect of the mutation on V-ATPase function by comparing cell growth at pH 5 and 7.5. Vph1⌬stv1⌬ cells transformed with a wild-type allele were able to grow under both conditions (Fig. 1B). This was expected because the 2ϫHA sequence allows normal V-ATPase assembly and does not interfere with V-ATPase function (28). On the other hand, the mutant strain vph1-⌬362-407 showed vma growth phenotype as cells failed to grow on plates buffered to pH 7.5 but exhibited wild-type growth at pH 5 (Fig. 1B). We have shown that yeast vma mutants with only 25% of the wild-type V-ATPase activity can grow normally at neutral pH (30,31). Therefore, removal of residues 362-407 severely compromised V-ATPase function in vivo.
Because the mutation ⌬362-407 comprised the entire tether region, we generated shorter truncations to establish more precisely what residues within the tether were essential for V-ATPase function (Fig. 1A). Each of the shorter deletions (⌬362-377, ⌬372-384, ⌬378 -407, and ⌬398 -407) exhibited vma mutant growth phenotype (Fig. 1B), indicating that amino acids connecting the C-terminal and N-terminal domains of V 0 subunit a were essential for normal functioning of V-ATPase pumps in vivo.
Tether-less Vph1 Mutants Cannot Assemble V 1 V 0 -Whether the vma phenotype was caused by a defect on V-ATPase assembly was addressed by immunoprecipitating V-ATPase complexes under nondenaturing conditions using the monoclonal antibody 8B1 against subunit A of V 1 . Western blots were used to determine whether V 1 and V 0 subunits co-immunoprecipitated as a means of assessing for V 1 V 0 assembly. Immunoblots probed with antibodies to V 1 subunits A and B and anti-HA to V 0 subunit a revealed V 1 and V 0 subunits co-immunoprecipitated together from cells expressing the wild-type allele of subunit a Vph1 ( Fig. 2A). This result showed that the HA tag on subunit a did not prevent wild-type V 1 V 0 complexes from assembling. When cells expressed tether-less vph1, the V 0 subunit a did not coprecipitate with V 1 subunit A, indicating that V 1 V 0 was not assembled ( Fig. 2A). Instead, V 1 subunits A and B were co-immunoprecipitated, resembling vph1⌬stv1⌬ cells that lack V 0 subunit a (2, 28). Vph1⌬stv1⌬ cells have wild-type levels of V 1 subunits that assemble into V 1 complexes constitutively present in the cytosol because V 0 cannot be formed at the membrane (2, 28). Our results indicate that tether-less vph1 mutants assembled V 1 , but V 1 did not associate with V 0 . We cannot exclude the possibility that mutant V 1 V 0 was unstable and that V 1 detached from V 0 when we immunoprecipitated the complex.
Tether-less Vph1 Mutants Assembled V 0 (Ϫd)-We asked whether tether-less vph1 interfered with assembly of V 0 because V 1 V 0 formation and/or its stability can be jeopardized if V 0 failed to assemble properly at the membrane (8). We used the monoclonal antibody 10D7 to address this question. 10D7 recognizes a cryptic epitope at the N terminus of Vph1 subunit a that is exposed only when V 1 is not attached to V 0 (23,25). Antibody 10D7 can immunoprecipitate subunit a assembled in V 0 but not in V 1 V 0 . Western blots showed V 0 subunits a and d but not V 1 subunits A and B in immunoprecipitates from wild-type cells (Fig. 2B, left). Thus, 10D7 recognized the steady state pool of V 0 complexes that is present in wild-type cells (25). With the exception of vph1-⌬398 -407, which was recognized by 10D7, antibody 10D7 did not immunoprecipitate the other tether deletion mutants.
It is possible that the truncations introduced altered the epitope for 10D7, preventing binding of the antibody to subunit a. Because the HA sequences were retained in the tether-less vph1 constructs made (Fig. 1A), we used the anti-HA antibody to immunoprecipitate subunit a. The HA tag is not masked by V 1 , and we immunoprecipitated V 1 V 0 complexes from wildtype cells (Fig. 2B, right). Western blots detected the wild-type  (35). Only the first transmembrane-spanning helix of the C-terminal domain is shown. Tether deletions made in this study are indicated (‚). Shown are the predicted ␣-helix (gray cylinders), ␤-sheet (black arrow), and coiled (black connecting line) structures. Location of the two copies of the ⌯〈 sequence introduced after residue Asn-185 (28) are indicated by the box. B, growth phenotype of Vph1 tether deletion mutants. Vph1⌬stv1⌬ cells were transformed with the wild-type (WT) VPH1 gene or VPH1 carrying the indicated truncations at the tether. Wild-type and mutant VPH1 were expressed from the low copy plasmid pRS316. Serial dilutions (left to right: 10-, 10 2 -, 10 3 -and 10 4 -fold) were performed into sterile water, and aliquots of each dilution were transferred to SDϪUra plates either at pH 5 or pH 7.5. Cells were grown for 96 h at 30°C.
The Tether of V 0 Subunit a Vph1 JULY 17, 2009 • VOLUME 284 • NUMBER 29 subunit a together with V 0 subunit d and V 1 subunits A and B. Anti-HA also recognized mutant subunit a from each of the tether deletion strains, but the truncated subunit a did not co-immunoprecipitate V 1 subunits A and B, verifying that V 1 V 0 was not assembled.
The tether-truncated subunit a that was pulled down with anti-HA did not co-immunoprecipitate V 0 subunit d (Fig. 2B). The absence of subunit d was unexpected. When V 0 subunits are unable to associate with each other in the endoplasmic reticulum, subunits a and d are diverted for degradation (37). Subunit a was not degraded in the tether-less vph1 mutant strains because we immunoprecipitated subunit a from each of the truncation mutants (Fig. 2B). Despite that subunit d did not co-immunoprecipitate with mutant subunit a, subunit d was not degraded either. Wild-type levels of subunit d were present in these cells (Fig. 3A). Western blots of whole cell lysates also showed mutant subunit a, although the deletions made it fairly unstable. With the exception of vph1-⌬362-407, for which only traces were consistently detected, we found the level of subunit a to vary with different preps in the other tether-less mutants.
Next, we determined whether the proteolipid subunits forming the c-ring, which are the other major components of V 0 (38,39), co-immunoprecipitated with tether-less subunit a. We biosynthetically labeled cells with 35 S and used anti-HA to immunoprecipitate V-ATPase complexes. We included 35 S-radiolabeled V 1 V 0 and V 0 complexes in the same gel to identify the V-ATPase subunits (Fig. 3B, first two lanes). Subunit a was detected in both wild-type and tether truncation mutants, but it was not present in untransformed control vph1⌬stv1⌬ cells (Fig. 3B). Subunit c is the major proteolipid of the c-ring (38,39) and migrates at around 17 kDa. Subunit c was co-immunoprecipitated with subunit a from wild-type and mutant strains, but it was not present in the immunoprecipitate from vph1⌬stv1⌬ cells. The lower signal for subunit c detected in the mutants could be a result of the proximity of the deletions to the first transmembrane helix in the C-terminal domain of Vph1. We concluded that subunit a and the c-ring were assembled in cells expressing tether-less vph1. Subunit d, however, was not in the immunoprecipitate from truncation mutants. A faint 36-kDa band that corresponds to subunit d was detected only in the wild-type strain. Taken together, biosynthetic radiolabeling (Fig. 3B) and Western blot analyses (Fig. 2B) indicated that subunit d is not co-immunoprecipitated with tether-less vph1 mutant V 0 .
Subunit d Rescued Vph1 Tether-less Mutants-In an attempt to repair the assembly defects generated by truncations at the tether of subunit a, we introduced additional copies of VMA6, the gene that encodes subunit d. Vph1⌬stv1⌬ cells were co-transformed simultaneously with VPH1-⌬362-407 and VMA6. Each gene was expressed from the low copy vector pRS316 under control of its natural promoter. Transformants were selected on SDϪUra plates adjusted to pH 5.
Exogenous subunit d rescued the vma mutant phenotype of vph1-⌬362-407. Cells co-transformed with VPH1-⌬362-407 and VMA6 (Fig. 4A, left), but not with the plasmid pRS316 alone (Fig. 4B), were able to grow normally at pH 7.5. These results indicated that exogenous subunit d helped cells to overcome or alleviate the functional defects caused by truncation of the entire tether region. As expected, co-transformation with subunit d also rescued the vma mutant phenotype of cells FIGURE 2. Vph1 tether deletion mutants failed to assemble V 1 V 0 complexes. A, immunoprecipitation (IP) of V 1 subunit A. Cultures of vph1⌬stv1⌬ cells transformed with the wild-type (WT) VPH1 gene and VPH1 carrying the indicated truncations at the tether were converted to spheroplasts by treatment with zymolase. Spheroplasts were lysed in PBS containing protease inhibitors, 1% C 12 E 9 , and dithiobis[succinimidyl propionate]. Lysates were incubated with antibody 8B1 against V 1 subunit A, and V-ATPase subunits were immunoprecipitated after the addition of protein A-Sepharose as described under "Experimental Procedures." A control which did not have the cell lysate added was incubated in parallel under the same conditions (antibody alone (Ab)). Immunoprecipitates were solubilized by the addition cracking buffer, subjected to SDS-PAGE using 12% gels, and analyzed by Western blots. Immunoblots were proved with monoclonal antibodies to V 1 subunits A and B and V 0 subunit a (anti-HA). Alkaline phosphatase-conjugated goat anti-mouse secondary antibody was added to blots which were developed by the addition of the alkaline-phosphatase substrates nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate. Bands corresponding to V-ATPase subunits a (100 kDa), A (69 kDa), and B (60 kDa) and the antibody heavy chain (HC) are indicated. B, immunoprecipitation of V 0 subunit a. Vph1⌬stv1⌬ cells transformed with the wild-type VPH1 gene and VPH1 with truncations at the tether were converted to spheroplasts and lysed as described under "Experimental Procedures." V 0 subunit a was immunoprecipitated by using antibody 10D7 (left) and anti-HA (right). Immunoprecipitations were as described in the legend to A. Immunoprecipitates were analyzed by Western blots using antibodies to V 1 subunits A and B and V 0 subunits a (anti-HA) and d. carrying a shorter deletion at the tether such as vph1-⌬362-377 (Fig. 4A, right).
Western blots of whole cell lysates obtained from double transformant cells resembled whole cells lysates from single transformant tether-less vph1 mutants as the level of subunit a varied with different preps. Occasionally, we detected vph1-⌬362-407 in the rescued cells (Fig. 5A). This suggests that exogenous subunit d conferred some stability to the truncated subunit a. By using quantitative immunoblotting we compared the level of V 0 subunit d in the mutant cells rescued by VMA6 and wild-type cells. Subunit d was present at about 2-fold greater levels in tether-less mutants co-transformed with VMA6 than wild-type cells (Fig. 5B). The level of V 1 subunit B was comparable in the three strains. This modest overexpression of subunit d restored V 1 V 0 assembly in vph1-⌬362-407 and vph1-⌬362-377 mutants (Fig. 5C). The antibody anti-HA co-immunoprecipitated V 1 subunits A and B with the mutant subunit a.
That VMA6 complemented the growth phenotype of tetherless mutants (Fig. 4) suggests that total or partial V-ATPase activity was restored (30,31). The extent of V-ATPase activity restored was determined in vacuolar membranes. We measured concanamycin A-sensitive ATP hydrolysis in vacuolar vesicles purified by Ficoll gradient density centrifugation. Membranes from double transformant cells (vph1-⌬362-407ϩVma6 and vph1-⌬362-377ϩVma6) retained 63-71% of the wild-type V-ATPase activity ( Table 2). Although the level of V 0 subunit a was reduced compared with wild-type membranes, Western blots showed V 0 subunits a and d, and V 1 subunits A, B, D, and E, verifying that V 1 V 0 was assembled at the vacuole (Fig. 6). In the absence of exogenous subunit d, vph1-⌬362-407 and vph1-⌬362-377 membranes retained, respectively, 7 and 14% of the wild-type activity ( Table 2). Immunoblots of the inactive membranes had only residual levels of V 0 subunits (a and d) and V 1 subunits (A and B). The V 1 subunits E and D were not detected.
Tether-less V 1 V 0 Is Controlled by Glucose-An important regulatory mechanism that controls V-ATPase function is its reversible disassembly in response to glucose (2,(23)(24)(25). Because multiple and dynamic interactions between the N terminus of V 0 subunit a and several V-ATPase subunits take place as V 1 V 0 reversibly disassembles, we asked if deletion of its tether affected disassembly and reassembly.
Vph1-⌬362-407 and vph1-⌬362-377 cells, each co-transformed with subunit d, were biosynthetically labeled with 35 S. Labeled cells were chased in YEPD medium which contains glucose (condition for assembly), YEP medium which lacks glucose (condition for disassembly), and YEP medium followed by a second chase with glucose added (condition for reassembly) (25). After chases, V 1 V 0 was immunoprecipitated under nondenaturing conditions using the antibody to V 1 subunit A (8B1). We co-immunoprecipitated V 1 subunits A and B with V 0 subunit a from cells chased in the presence of glucose showing that V 1 V 0 was assembled (Fig. 7A). After glucose removal, the level of V 0 subunit a relative to V 1 subunits A and B decreased because V 1 V 0 complexes disassembled. Readdition of glucose after brief glucose depletion triggered reassembly of V 1 V 0 complexes as indicated by the greater level of V 0 subunit a associated with subunits A and B. These results showed that assembled V 1 V 0 containing vph1-⌬362-377 and vph1-⌬362-407 mutant subunit a reversibly disassembled in response to extracellular glucose.
Mutant V-ATPase Is Complexed with Phosphofructokinase-V-ATPase function is intimately linked to glycolysis (24,25), and binding of glycolytic enzymes to several V-ATPase subunits modulates V 1 V 0 activity and its assembly (40 -42). Subunit a of V 0 has been shown to interact with the glycolytic enzyme phosphofructokinase (40,42).
We asked if deletion of the tether prevented subunit a from binding to phosphofructokinase. Immunoblots of purified vacuolar membranes using antibodies to phosphofructokinase showed the glycolytic enzyme associated with mutant and wild-type membranes. Phosphofructokinase was present in membranes from vph1-⌬362-407ϩVma6 and vph1-⌬362-377ϩVma6 (Fig. 7B, left), which contained assembled mutant V 1 V 0 . That phosphofructokinase associated with V-ATPase pumps was shown by immunoprecipitating the V-ATPase complex with anti-HA and blotting against phosphofructokinase (Fig. 7B, right). Phosphofructokinase was coimmunoprecipitated with V 1 V 0 from cells carrying exogenous subunit d (vph1-⌬362-407ϩVma6 and vph1-⌬362-377ϩVma6). Phosphofructokinase was complexed also with A, whole cell lysates. Vph1⌬stv1⌬ cells transformed with the wild-type (WT) VPH1 gene or VPH1 carrying the indicated truncations were grown overnight to midlog phase in SDϪUra pH 5. A culture volume corresponding to Abs 600 nm ϭ 20 -30 from each strain was removed and lysed as described under "Experimental Procedures." Proteins in the whole cell lysate (1-2 A 600 of the original culture) were loaded onto a 12% polyacrylamide gel and separated by SDS-PAGE. Extracts were blotted with a polyclonal antibody against subunit d, and monoclonal antibodies against subunits a (HA), A, and B. B, immunoprecipitation of biosynthetically labeled V-ATPase. Vph1⌬stv1⌬ transformed with the wild-type-VPH1 gene and VPH1 carrying tether truncations were grown to mid-log phase in minimal medium lacking methionine and the nutritional marker uracil (SDϪMet-Ura pH 5). Cells were converted to spheroplasts and biosynthetically labeled with Tran 35 S-label as described under "Experimental Procedures." Labeled spheroplasts were lysed, and V-ATPase immunoprecipitated under nondenaturing conditions using an antibody against HA. Immunoprecipitates were separated by SDS-PAGE using 13% gels, scanned in a Fuji Scanner 5100, and analyzed using the Multi Gauge software. V 1 V 0 and V 0 complexes radiolabeled with 35 S were included in the same gel and used as reference for identifying V-ATPase subunits (left two lanes). JULY 17, 2009 • VOLUME 284 • NUMBER 29 JOURNAL OF BIOLOGICAL CHEMISTRY 19527 V 0 (Ϫd) from vph1-⌬362-407 and vph1-⌬362-377 cells. These results showed that deletion of amino acids forming the tether that connects the N-and C-terminal domains did not prevent binding of phosphofructokinase to Vph1.

DISCUSSION
Electron microscopy single particle reconstruction of V 1 V 0 shows the N terminus of subunit a filling part of a collar-like structure at the bottom of V 1 , bridging subunits C and H and stabilizing the three peripheral stalk-forming E/G dimers (18,43). After V 1 V 0 disassembly the N terminus of subunit a swings position and interacts with V 0 subunit d, a mass that lies on top of the c-ring structure in free V 0 (27). Thus, the hydrophilic N-terminal domain of subunit a alternates between V 1 peripheral stalk-forming subunits (E/G, C, H) and V 0 subunit d as the V 1 V 0 complex reversibly disassembles. Here, we presented the first evidence that the tether at the N terminus of subunit a is required for stable assembly of V 0 subunit d.
Direct interactions between the N terminus of subunit a and subunit d have been recently shown in the human isoforms a3 and d2 that are the isoforms selectively expressed in osteoclasts (44). We showed that deletion of amino acids connecting the N-terminal and C-terminal domains of subunit a yeast isoform Vph1 perturbed assembly of subunit d.
Mutant V 0 domains containing subunit a shortened by 10 -46 amino acids at its tether lacked subunit d. Subunit d did not co-immunoprecipitate with mutant subunit a, but subunit d was not degraded either. We detected apparent wildtype levels of endogenous subunit d in these cells (Fig. 3A). Subunit d was quite stable, even in vph1-⌬362-407, for which only traces of subunit a were consistently found. It has been shown that subunits d and a are targeted for degradation if V 0 does not assemble (37,45). Therefore, V 0 domains were formed in the mutant strains. That mutant subunit a was fairly stable, and co-precipitated proteolipid subunits of the c-ring (Fig. 3B) is consistent with the notion that V 0 did assemble or partially assemble.
Because improperly bound subunit d and its absence from V 0 would be sufficient to prevent V-ATPase assembly (45), tetherless mutants did not assemble V 1 V 0 complexes. Mutant cells contained V 1 and V 0 that failed to come together to form stable V 1 V 0 . Stable V 1 V 0 assembled only after exogenous subunit d was introduced in the cells by co-transforming vph1⌬stv1⌬ with subunit d and the tether-less subunit a mutant simultaneously. Expression of exogenous subunit d was from a low copy CEN plasmid and under control of its natural promoter so that we anticipated only a modest overexpression of the subunit. Quantitative immunoblotting of whole cell lysates confirmed our expectations. About 2-fold greater levels of subunit d were present in tether-less mutants co-transformed with VMA6 than in wild-type cells.
Modest overexpression of subunit d also rescued the vma phenotype of tether-less vph1 mutants, indicating that tetherless V 1 V 0 was functional. The assembly and functional defects associated with shorter tethers were specific to subunit d as the plasmid alone did not rescue the mutant phenotype. In addition, exogenous V 1 subunit C (Vma5), which is necessary for assembly of preexisting V 1 and V 0 (46), failed to rescue the vma phenotype of tether deletion mutants (not shown). We co-transformed vph1⌬stv1⌬ with VMA5 and either VPH1-⌬362-407 or VPH1-⌬362-377, but cells remained unable to grow at neutral pH. Furthermore, wild-type cells with exogenous subunit C developed vma mutant growth characteristics. There is evidence that overexpression of subunit C is harmful because it can stimulate uncoupled ATP hydrolysis by cytosolic V 1 (47). Like subunit d, we expressed subunit C from the CEN vector pRS316. Therefore, moderate overex-  VMA6 rescued vph1-tether deletions mutants. A, exogenous subunit d rescued growth of vph1-⌬362-407 and vph1-⌬362-377 at neutral pH. Vph1⌬stv1⌬ cells were co-transformed with the VMA6 gene plus either wild-type (WT) VPH1 gene or VPH1 carrying the indicated truncation. Each gene was expressed from the low copy plasmid pRS316 under control of their natural promoter. Transformants were selected in SDϪUra plates adjusted to pH 5. Cell growth was compared at pH 5 and 7.5 on SDϪUra plates for 96 h at 30°C. Controls included vph1⌬stv1⌬ transformed with wild-type VPH1 alone, VPH1-⌬362-407 alone, and VPH1-⌬362-377 alone. B, plasmid pRS316 alone cannot rescue pH-sensitive growth. Vph1⌬stv1⌬ cells were co-transformed with the plasmid pRS316 plus VPH1 carrying the indicated truncations at the tether. Controls included the wild-type VPH1 gene alone and the VMA6 gene alone, each in the plasmid pRS316. Transformants were selected in SDϪUra plates pH 5. Cell growth was compared on SDϪUra plates at pH 5 and 7.5 incubated for 96 h at 30°C.
pression of subunit C, but not subunit d, was harmful to wild-type V-ATPase function.
Once V 1 V 0 complexes were assembled, V-ATPase function was resilient to deletion of the region connecting the N terminus and C terminus of subunit a. Unexpectedly high levels of activity were retained. Vph1-⌬362-377 and vph1-⌬362-407 membranes retained 63-71% of the wild-type concanamycin A-sensitive ATPase activity. The level of activity correlated with the level of subunit a at the membrane, indicating that the complexes were not inhibited. Thus, significant binding of the subunit a N-terminal domain to peripheral stalk-forming subunits of V 1 was retained even after deleting 46 amino acids of its tether. This showed that the stator role of subunit a was preserved. It allowed ATP-driven rotation of the central stalk relative to a steady stator that secured the catalytic sites in the A 3 B 3 hexamer.
Our observations are consistent with recent studies showing that V 1 V 0 retained 80% of its ATP hydrolyzing activity after Vph1 N-and C-terminal domains were separated by cleavage at residue 376 of its tether (48). Those results suggested that residues at the C-end of the cleavage site (e.g. Gln-369) likely retained binding to V 1 stator subunits that contributed to supporting rotational catalysis. Our study, however, showed that such interactions were not essential after deleting the tether. Other interactions between V 1 stator-forming subunits and the N-terminal domain were retained and sufficient for its role as stator.
Tether-less V 1 V 0 was responsive to extracellular glucose. Deletion of the tether did not impair glucose-controlled V-ATPase disassembly and reassembly. Assembled V-ATPase complexes containing the mutant subunit a vph1-⌬362-377 and vph1-⌬362-407 disassembled and reassembled after brief glucose depletion and readdition (Fig. 7A).
Interactions of the V-ATPase complex with glycolytic enzymes are likely to intertwine glucose-dependent V-ATPase disassembly and reassembly with glycolysis (25,41,42). Phosphofructokinase, which modulates V-ATPase function through its binding to subunit a of the yeast and human pumps (40,42), was present in vacuolar membranes (Fig. 7B). Phosphofructokinase was present in mutant and wild-type vacuolar membranes alike and co-immunoprecipitated with mutant and wild-type V-ATPase complexes. Deletion of amino acids at the tether did not prevent phosphofructokinase binding to V 0 and V 1 V 0 . Unlike the glycolytic enzyme aldolase, which binds only assem- FIGURE 5. Exogenous subunit d allowed V 1 V 0 assembly of tether deletion mutants. A, whole cell lysates. Whole cell lysates were prepared from vph1⌬stv1⌬ cells co-transformed with VMA6 plus either wild-type (WT) VPH1 or VPH1 carrying tether deletions. Each gene was expressed from the low copy plasmid pRS316 under control of their corresponding natural promoter. Cells were grown overnight to mid-log phase in SDϪUra pH 5. Protein extracts were prepared as described in the legend to Fig. 3A, except that a 10% gel was used. Protein was transferred to a nitrocellulose membrane for detection of subunits a, A, B, and d by Western blots as described in the legend to Fig. 2. B, quantitative immunoblotting of V-ATPase subunits in whole cell lysates. Serial dilutions of whole cell lysates prepared from wildtype and tether-less vph1 mutant cells co-transformed with VMA6 were prepared in cracking buffer. Whole cell lysates were prepared as described above. The indicated amounts of A 600 in each well were separated in 10% SDS-PAGE and analyzed by Western blotting. C, immunoprecipitation of V 1 V 0 . Vph1⌬stv1⌬ co-transformed with VMA6 plus either wild-type VPH1 or VPH1 carrying the indicated truncations was grown in SDϪUra medium at pH 5. Cells were converted to spheroplasts, lysed, and immunoprecipitated with anti-HA antibody as described in the legend to Fig. 2B. Immunoprecipitated protein was separated by SDS-PAGE (10% gel) and transferred to a nitrocellulose membrane for blotting using primary antibodies to V-ATPase subunits a (HA), A, and B. The band corresponding to the antibody heavy chain (HC) is indicated. FIGURE 6. Tether-less mutant V 1 V 0 is delivered to the vacuoles of cells transformed with exogenous subunit d. Left, vacuolar vesicles from vph1⌬stv1⌬ cells co-transformed with VMA6 plus either wild-type VPH1 or mutant VPH1 were purified by Ficoll gradient density centrifugation. Vacuolar protein (25 g) solubilized in cracking buffer was loaded onto 10% polyacrylamide gels and separated by SDS-PAGE. Protein was transferred to nitrocellulose, and blots were probed with monoclonal antibodies against subunits a (HA), A, and B and polyclonal antibodies against subunits D, E, and d. Immunoblots were as described in the legend to Fig. 2. Right, Vacuolar membranes from either wild-type VPH1 or mutant VPH1 were purified by Ficoll gradient centrifugation. Vacuolar protein was solubilized and analyzed by immunoblotting using antibodies against V 0 subunits a and d and V 1 subunits A, B, D, and E. For loading control, membranes were probed with a monoclonal antibody against vacuolar alkaline phosphatase (ALK).

TABLE 2 Concanamycin A-sensitive V-ATPase specific activity in vacuolar membranes purified from subunit a (Vph1) tether deletion mutants
Reported is the activity sensitive to concanamycin A inhibition. Coupled ATPase assays were performed as described under "Experimental Procedures" in the presence and absence of concanamycin A. Data are expressed as the mean Ϯ S.E. for n separate vacuolar purifications. Relative mean values were used to express percentage activity.

Strain
Specific activity % Activity bled V 1 V 0 (41), phosphofructokinase did not discriminate between assembled and disassembled V-ATPase. Thus, binding of phosphofructokinase to V 0 was independent of the assembly state of the complex. Because phosphofructokinase binds residues at the C terminus of subunit a (40,42), we can infer that no major structural changes were generated at the C terminus when residues connecting the C-and N-terminal domains were deleted. Significant structural changes were not introduced at the N-terminal domain either because tether-less V-ATPases exhibited binding of V 1 subunits to V 0 (Fig. 6), hydrolysis of ATP (Table 2), and regulation by reversible disassembly (Fig. 7A).
The extent to which rescued cells retained V-ATPase function in tether-less mutants could be explained if residues at the tether form a loop or a flexible structure that can be shortened without causing major structural changes. Indeed, residues 362-407 are predicted to be primarily coiled (Fig. 1A). This sequence constitutes the most conserved region at the N-terminal domain of subunit a, showing 37% identity between mouse and yeast isoforms. The results from this study indicated that conserved residues at the tether are necessary for stable binding of subunit d in V 0 , and its deletion distressed V 1 V 0 assembly.
Subunit d is the only component of V 0 that is not an integral membrane protein. It dissociates in vitro from vacuolar membranes exposed to elevated pH, urea, and Na 2 CO 3 (45). Thus, dissociation of subunit d from tether-less V 0 can be envisioned. If shrunk tethers lowered subunit d affinity of binding for V 0 , loosely bound subunit d could have detached from V 0 . But once assembly of V 1 V 0 was stimulated by exogenous subunit d, its binding into V 0 was fairly stable. Tether-less V 1 V 0 was competent for reversible disassembly in response to glucose (Fig. 7A), suggesting that subunit d did not readily detach from V 0 when the V-ATPase disassembled after glucose depletion.
Without exogenous subunit d, only traces of V 1 and V 0 subunits were present at the vacuolar membrane. The fact that subunit d rescued V-ATPase assembly of tether-less vph1 mutants suggests that early steps of assembly and trafficking were likely affected by deletion of the tether. Residues at the tether could help recruit subunit d during assembly of V 1 V 0 either by binding subunit d directly or facilitating interactions of subunit d with other V-ATPase subunits and/or assembly factors. Alternatively, the tether region could influence the formation of the subunit d binding site.
Two major biosynthetic pathways are responsible for V 1 V 0 assembly in yeast. One pathway involves independent assembly of V 0 and V 1 domains followed by V 1 V 0 formation (37,49,50). The other pathway involves early association of V 1 and V 0 subunits and no preassembly of V 1 and V 0 subcomplexes (51). When one of these two pathways is blocked, the other maybe preferentially used (2,51). Subunit d appears to be the last subunit recruited during preassembly of V 0 in the endoplasmic reticulum (49,50). Thus, if subunit d was sequestered and V 0 formation stalled, tether deletion mutants could have accumulated partially assembled V 0 (Ϫd) complexes. One explanation of how subunit d rescued the assembly defect is that its overexpression could have shifted the equilibrium V 0 7 V 0 (Ϫd) to favor V 0 formation, allowing V 1 V 0 assembly.
The addition of exogenous subunit d could have stimulated also the concerted V 1 V 0 assembly pathway in which early associations between V 1 and V 0 subunits precede full assembly of either subcomplex. Because this pathway is characterized by fast kinetics of association between subunits A of V 1 and Vph1 of V 0 (51), its stimulation can help explain the increased amount of V 1 subunit A co-immunoprecipitated with anti-HA after wild-type cells were co-transformed with VMA6 (Fig. 5B).
Although F 1 F 0 synthesizes ATP and V 1 V 0 exclusively hydrolyzes ATP, the tether of V 0 subunit a could functionally resemble the tether of E. coli F 0 subunit b. Both tethers help to prevent rotation of the catalytic hexamer and could offer a mechanism for elastic energy storage during catalysis (15,16,52). If flexibility at the stator subunit secured in the membrane (F 0 subunit b and V 0 subunit a) is a requirement to resist the torque generated by rotation of the central stalk (15,52), other flexible elements could have stored elastic energy after deleting Vph1 residues 362-407. Another flexible element(s) can also be expected to aid in the major conformational changes of the N terminus when tether-less V-ATPase experienced reversible A, pulse-chased experiments followed by V 1 V 0 immunoprecipitation. Wildtype yeast cells were converted to spheroplasts and labeled with Tran 35 Slabel in SDϪMetϪUra pH 5. Excess unlabeled cysteine and methionine were added to the radiolabeled spheroplasts, which were immediately resuspended in chase medium containing 1.2 M sorbitol and chased at 30°C as follows: 20 min in YEPD (2% glucose), 10 min in YEP (no glucose), 10 min in YEP followed by an additional 10 min with 2% glucose added. A monoclonal antibody against subunit A of V 1 (8B1) was used for the immunoprecipitations. Immunoprecipitates were analyzed as described in the legend to Fig. 3B. V 1 subunits A and B and V 0 subunit a are indicated at the left. B, the glycolytic enzyme phosphofructokinase associates with V-ATPase. Left, vacuolar membranes were purified from wild-type (WT) and tether-less mutants (vph1-⌬362-407 and vph1-⌬362-377), each co-transformed with exogenous subunit d as described in the legend to Fig. 6. Shown is a Western blot of vacuolar membrane protein (25 g) proved with a polyclonal antibody against yeast phosphofructokinase (PFK). Right, subunit a was immunoprecipitated from pH-sensitive tether-less mutants (vph1-⌬362-407 and vph1-⌬362-377) and from mutant cells that have been rescued by exogenous subunit d (vph1-⌬362-407ϩVma6 and vph1-⌬362-377ϩVma6). Anti-HA was used for immunoprecipitations as described in the legend to Fig. 2. Immunoprecipitates (IP) were analyzed by Western blots by using a polyclonal antibody against yeast phosphofructokinase. The band corresponding to the antibody heavy chain (HC) is indicated.
disassembly. More than a third of the N terminus sequence of subunit a is predicted coiled, and a collection of studies have shown that many other structural elements at the peripheral stalks are needed for V-ATPase function (31, 47, 54 -56), arguing against complete redundancy.
Analogously to E. coli F 1 F 0 , yeast V 1 V 0 function was resistant to deletions at the tether. F 0 subunit b accommodates the deletion of up to 11 residues (53); V 0 subunit a accommodated deletion of up to 46 residues and still assembled functional V-ATPases in vivo when exogenous subunit d was provided. That V-ATPase supported much greater deletions is likely a reflection of the functional and structural features that distinguish V 1 V 0 from F 1 F 0 . F 1 F 0 has one peripheral stalk which protrudes from the membrane (52); V 1 V 0 has three peripheral stalks (E/G) 3 secured by the N terminus of subunit a, which protrudes from the membrane (18,43). In addition, only V 1 V 0 naturally disassembles and reassembles in vivo (1,2,(23)(24)(25). Therefore, a longer tether (more elasticity) may be a requirement in V 0 subunit a because the N-terminal domain holds three E/G dimers and resists the stress imposed by the detachment and reattachment of multiple subunits.
Although its function as stator in assembled V 1 V 0 is becoming clear (1,19), roles for the cytosolic N terminus of subunit a in free V 0 have yet to be established. The N terminus of subunit a Vph1 is not essential for blocking passive proton transport through the c-ring when V 1 is not attached (48). Removal of the N terminus (amino acids 1-376) of subunit a Vph1 from V 0 cannot dissipate proton gradients across yeast vacuolar membranes (48). It was not established in those studies whether subunit d was retained with V 0 at the vacuolar membrane after removal of the N terminus. Subunit d lies on the central cavity of the c-ring (27,(57)(58)(59), and subunit d itself could be responsible for blocking proton leakage via its interactions with the c-ring. We have shown that residues of subunit d predicted to lie in the central cavity of the c-ring are essential for proton transport in yeast (30). In mung bean, V 0 (Ϫd) domains reconstituted into liposomes exhibited passive proton transport which was blocked after subunit d addition (60). The tether deletion mutants of vph1 described in this study constitute an attractive source of V 0 (Ϫd) domains that can allow us directly to address the function of subunit d for passive proton transport in yeast.