Evidence that the NH2 terminus of vph1p, an integral subunit of the V0 sector of the yeast V-ATPase, interacts directly with the Vma1p and Vma13p subunits of the V1 sector.

The vacuolar-type H(+)-ATPase (V-ATPase) is composed of a peripherally bound (V(1)) and a membrane-associated (V(0)) complex. V(1) ATP hydrolysis is thought to rotate a central stalk, which in turn, is hypothesized to drive V(0) proton translocation. Transduction of torque exerted by the rotating stalk on V(0) requires a fixed structural link (stator) between the complexes to prevent energy loss through futile rotation of V(1) relative to V(0); this work sought to identify stator components. The 95-kDa V-ATPase subunit, Vph1p, has a cytosolic NH(2) terminus (Nt-Vph1p) and a membrane-associated COOH terminus. Two-hybrid assays demonstrated that Nt-Vph1p interacts with the catalytic V(1) subunit, Vma1p. Co-immunoprecipitation of Vma1p with Nt-Vph1p confirmed the interaction. Expression of Nt-Vph1p in a Deltavph1 mutant was necessary to recruit Vma13p to V(1). Vma13p bound to Nt-Vph1p in vitro demonstrating direct interaction. Limited trypsin digests cleaves both Nt-Vph1p and Vma13p. The same tryptic treatment results in a loss of proton translocation while not reducing bafilomycin A(1)-sensitive ATP hydrolysis. Trypsin cleaved Vph1p at arginine 53. Elimination of the tryptic cleavage site by substitution of arginine 53 to serine partially protected vacuolar acidification from trypsin digestion. These results suggest that Vph1p may function as a component of a fixed structural link, or stator, coupling V(1) ATP hydrolysis to V(0) proton translocation.

Vacuolar-type H ϩ -ATPases (V-ATPases) 1 are evolutionary conserved multimeric complexes that mediate the lumenal acidification of various organelles (yeast and plant vacuoles, endosomes, lysosomes, and clathrin-coated vesicles). Organel-lar acidification is essential for a variety of cellular processes such as receptor-mediated endocytosis, processing of proteins, intracellular degradation of ingested pathogens, and protoncoupled transport of small molecules (reviewed in Ref. 1).
V-ATPases are evolutionarily related to the mitochondrial F 1 F 0 H ϩ -ATPases (F-ATPases) (2). Thus, the 2.8-Å resolution of the catalytic complex of the F-ATPase (3) and experiments describing how intersubunit rotation leads to energy transduction in F-ATPases (4,5) have provided significant insight into the structure and function of V-ATPases. V-ATPases have a similar bipartite structure to F-ATPases, consisting of a membrane bound proton channel (V 0 ) and a peripherally bound catalytic core (V 1 ). Electron microscopy reveals the V 1 complex as a lollipop-like structure with a 6-nm long and 4-nm wide "central stalk" rising out of the membrane (6, 7) and a 12 ϫ 14-nm diameter ball symmetrically situated on top of the stalk (7,8). The ball is a hexamer composed of alternating 69-(Vma1p) and 60-kDa (Vma2p) subunits symmetrically arranged to form a hollow core through the middle of the sphere. Vma1p and Vma2p both contain consensus sequences for ATPbinding domains (9 -11); Vma1p has been ascribed a catalytic function (12) while Vma2p is thought to play a regulatory role (13). The hollow core formed by the ATP-binding subunits contains a central stalk, the composition of which is still debated. Xu et al. (14) propose that the central stalk is composed of just Vma7p (14 kDa) and Vma8p (32 kDa) while Tomashek et al. (15) hypothesize that the structure also includes Vma4p (27 kDa) and Vma10p (16 kDa). The membrane-bound V 0 complex contains six copies of the 17-kDa proteolipid, Vma3p (16), one copy of Vma6p (36 kDa), and at least one (17), but possibly up to three copies (18), of Vph1p (95 kDa). Vma11p (19) and Vma16p (20) are 17-and 23-kDa subunits with 56 and 35% amino acid identity to Vma3p, respectively. They are indispensable for V-ATPase activity, yet their role and stoichiometry in the V 0 complex is unknown (21).
In F-ATPases, ATP hydrolysis rotates a stalk structure composed of the ␥ subunit in a counterclockwise direction within the core of the catalytic hexamer (5). As energy coupling is thought to occur from the physical rotation of the stalk connecting the F 1 and F 0 complexes, a "stator"-like structure must exist to prevent the rotation of the catalytic hexamer with respect to the F 0 complex. A stator (defined as the stationary portion of a motor, dynamo, or turbine) would have to be anchored within the membrane and extend up to the catalytic hexamer without interfering with the rotation of the stalk. Structural evidence supports the presence of this second structural link between the two sectors. Electron microscopy of Escherichia coli F-ATPases revealed the presence of two stalks, a central shaft as previously discussed, and a second peripheral stalk that extends from the side of the F 1 complex down to the F 0 complex (22). This second physical link between the two complexes is ideally suited to serve as a stator. Cross-linking studies suggest that the peripheral stalk structure is likely composed of the a, (b) 2 , and ␦ subunits (23,24).
To date, the structure of the V-ATPase remains less well defined, however, the extensive homology between V-and F-ATPases suggests that the mechanical coupling of the V 1 and V 0 complexes likely occurs by a method analogous to that described for F-ATPase. Visualization of bacterial and bovine V-ATPases by electron microscopy supports both the presence of a central shaft linking V 1 to V 0 and a peripheral stalk (13 nm long and 6.5 nm off-center from the central stalk) that appears to originate from the V 0 complex and extend to the top of the V 1 complex (6,7,25). As there are no clear V-ATPase homologues for the F-ATPase a, b, or ␦ subunits, the components of the V-ATPase peripheral stalk are unknown. The function of the catalytic (Vma1p), regulatory (Vma2p), and proteolipid (Vma3p) subunits are reasonably established. The composition of the central stalk is suggested to be either Vma7p and Vma8p (14), or Vma7p and Vma8p together with Vma4p and Vma10p (15). Thus, four to six candidate subunits with unassigned function remain to form the second peripheral stalk or stator: Vma4p, Vma5p, Vma6p, Vma10p, Vma13p, and/or hydrophilic portions of Vph1p.
The 95-kDa Vph1p subunit is unique to vacuolar H ϩ -AT-Pases with no homologue found in any F-ATPase. Hydropathy analysis of Vph1p predicts two distinct structural domains; a hydrophilic NH 2 -terminal domain (ϳ45 kDa) and a hydrophobic COOH-terminal domain (ϳ50 kDa) containing up to 9 putative membrane-spanning regions (26). We have unambiguously localized the NH 2 terminus of Vph1p to the cytosol, making this domain an ideal candidate to contribute to the formation of a peripheral stalk (27). Here we demonstrate that the NH 2 terminus of Vph1p interacts with Vma1p and Vma13p, and present proteolytic evidence that suggests Vph1p is involved in coupling V 1 ATP hydrolysis to V 0 proton translocation.
pET-Nt-Vph1p (MM658)-PCR was performed on pVIP1-82 (28) with oligonucleotide MO33 (5Ј-caaggaaaaccatggcagag-3Ј: introduces NcoI restriction site at first methionine of Vph1p) and oligonucleotide MO74 (5Ј-gcgaattccttaattgatttctctgtactgagc-3Ј: introduces an EcoRI restriction site and a stop codon just before the first putative transmembrane region of Vph1p). The PCR product was cut with NcoI and EcoRI and cloned into the NcoI-EcoRI restriction sites of pET32b (Novagen, Madison, WI). Sequencing confirmed fidelity of the resulting plasmid.

Antibodies
ATPase subunits were detected or immunoprecipitated with one of the following antibodies. 1) A monoclonal antibody directed against Vma1p, the 69-kDa catalytic V-ATPase subunit (clone 8B1-F3, Molecular Probes). 2) A rabbit polyclonal antibody raised against the plant homologue of Vma1p, purified from Beta vulagris L. (described in Ref.

Two-hybrid Screening
The two-hybrid library, a kind gift from Dr. Stephen J. Elledge, was created from S. cerevisiae cDNA (size selected to be Ͼ600 base pairs) cloned into the XhoI site of Lambda ACT as described in Ref. 31. All components were obtained from CLONTECH and all assays were carried out as suggested by the manufacturer. The bait, pAS2-1-Nt-Vph1p, was created as described above. The pACT yeast cDNA library was introduced the yeast reporter strain, Y190, and 5 ϫ 10 6 transformants were selected for tryptophan, leucine, and histidine prototrophy. Isolated colonies (100) were subsequently tested for ␤-galactosidase activity.

␤-Galatosidase Assay
Transformed cells were streaked onto sterile Whatman No. 1 filters and grown on selective media plates. Following colony growth, the filters were assayed for ␤-galactosidase activity. The cells were permeabilized by a freeze-thaw cycle. Filters were floated on liquid nitrogen for 20 s and allowed to thaw at room temperature. Each filter was soaked with 2 ml of Z-buffer (CLONTECH Matchmaker TM protocol manual) containing 5-bromo-4-chloro-3-indoyl-␤-D-galactosidase. The filters were placed in a covered plastic container and incubated at 37°C for a maximum of 3 h.

Vma13p Binding Experiment
Ten g of Ni-NTA-bound pET-32b and pET-Nt-Vph1p (as estimated by Coomassie staining) were each incubated for 30 min with 2 l of induced supernatant containing Vma13p (described above) in a final volume of 75 l of binding buffer. The resin was then pelleted (45 ϫ g), the supernatant collected (unbound protein fraction), and the resin washed (5 ϫ 4 volumes binding buffer). The bound proteins were then eluted from the Ni-NTA resin by boiling the resin in SDS sample buffer.

Purification of Vacuolar Membranes
Isolation of vacuolar membranes through flotation of intact vacuoles on Ficoll gradients was performed as described in Ref. 33 with modifications detailed in Ref. 28. Protein concentrations of vacuolar vesicles were estimated as described in Ref. 34. Purified vacuolar membranes were subjected to controlled trypsin proteolysis in the following manner: 20 g of purified vacuolar membranes in 25 mM MES/Tris, pH 7.8, 25 mM KCl, 5 mM ATP were incubated with 0.1 g/ml trypsin at 37°C for 30 min in a final reaction volume of 100 l. Trypsin proteolysis was inhibited by the addition of PMSF to a final concentration of 1 mM and N ␣ -p-tosyl-L-lysine chloromethyl ketone (TLCK) to a final concentration of 5 mM followed by incubation on ice for 15 min. Samples were utilized for activity assays as detailed below. Remaining samples were denatured by the addition of 5 ϫ SDS sample buffer and assayed by SDS-PAGE (35) and immunoblotting as described (36).

Proteolysis of Purified Vacuolar Membranes
Purified vacuolar membranes were subjected to controlled trypsin proteolysis in the following manner: 20 g of purified vacuolar membranes in 5 mM ATP, 25 mM MES/Tris, pH 7.8, and 25 mM KCl were incubated with either 0.1 or 1.0 g/ml trypsin at 37°C for 30 min in a final reaction volume of 100 l. Trypsin proteolysis was inhibited as described above. Samples were denatured, assayed by SDS-PAGE, and immunoblotted as described above.

ATPase Activity and Proton Translocating Activity
ATPase assays and proton pumping activity were assayed on freshly prepared vacuolar membrane vesicles. Vesicles were trypzinised with 0.1 g/ml in 25 mM MES/Tris, pH 7.8, 25 mM KCl, 5 mM ATP for 15 or 30 min at 37°C. The final protein concentration was 0.2 g/ml. Control samples were similarly treated excluding the addition of proteases. Trypsin activity was inhibited by the addition of PMSF (1 mM final concentration) and TLCK (5 mM, final concentration). Samples were incubated for 10 min on ice and trypsin-treated and control samples were assayed for ATPase and proton pumping activity. Proton pumping activity was measured by tracking the ATP-dependent quenching of acridine orange using a Perkin-Elmer 650-40 fluorescence spectrophotometer with emission at 545 nm and excitation at 493 nm. Assays were performed in a 1-ml volume of 25 mM MES/Tris, pH 7.2, 5 mM ATP, 25 mM KCl, 5 mM acridine orange, and 5 mM MgCl 2 . Acidification rate was defined as the slope of the quench during the first 10 s immediately after the addition of MgCl 2 ; in all cases the first 10 s gave a linear response and was not limited by the formation of a proton-motive force. Bafilomycin A 1 -sensitive ATP hydrolysis of purified vacuolar membranes was assayed by measuring the production of inorganic phosphate utilizing the Ames assay as detailed in Ref. 28. Assays were performed in the presence of 0.5 mM azide and 0.05 mM vanadate to inhibit mitochondrial and plasmalemmal ATPase activity, respectively. Inhibitors were added to the assay mixture 10 min before initiating the reaction. Final assay mixtures contained 5 g of protein in a final volume of 250 l plus 1 l of 50 M bafilomycin A 1 or 1 l of dimethyl sulfoxide.

Preparation of Sample for NH 2 -terminal Sequencing
Large scale trypsin digests of vacuolar membranes were performed in order to purify sufficient quantities of the 80-kDa tryptic fragment for NH 2 -terminal sequencing. Briefly, 10 mg of purified vacuolar membranes were treated with 0.5 g/ml trypsin in 25 mM MES/Tris, pH 7.8, 25 mM KCl, 5 mM ATP, and 5 mM MgCl 2 at 37°C for 2 h. Trypsin proteolysis was inhibited by the addition of PMSF to a final concentration of 1 mM, TLCK to a final concentration of 5 mM, and incubation on ice for 15 min. Subsequently, peripheral membrane proteins were removed by incubation with 100 mM NaCO 2 , pH 11, for 20 min on ice. Vacuolar membranes were washed twice with phosphate-buffered sa-line, pH 7.4, and solubilized in an appropriate volume of 1 ϫ Laemmli sample buffer. Samples were resolved by standard SDS-PAGE, transferred to PVDF, and submitted for microsequencing analysis. Microsequencing was performed by Harvard Microsequencing Facilities.

RESULTS
The NH 2 Terminus of Vph1p Interacts with the COOH Terminus of Vma1p-The 95-kDa subunit (Vph1p) is unique to the V-ATPase, having no homologous counterpart in the F-ATPase complex. Vph1p is defined by two structurally distinct domains: an NH 2 -terminal hydrophilic domain (ϳ45 kDa) and a COOH-terminal hydrophobic domain containing up to 9 putative transmembrane segments (21). The NH 2 terminus of Vph1p has been unambiguously localized to the cytosol (27), where it is available to associate with cytosolic proteins. Current evidence suggests that the cytosolic domain of Vph1p interacts with the catalytic sector. The peripheral complex blocks binding of an anti-NH 2 terminus Vph1p antibody to vacuolar membranes containing an assembled V 0 complex (27,37). As well, substrate binding to the catalytic sector induces a conformational change within the 95-kDa subunit (27). A yeast two-hybrid system was utilized to identify cytosolic proteins that potentially interact with the NH 2 terminus of Vph1p.
The 45-kDa NH 2 -terminal domain of Vph1p (amino acid residues 1-406) was utilized as bait (pAS2-1-Nt-Vph1p) to screen 5 ϫ 10 6 prey clones, yielding 100 clones prototrophic for tryptophan, leucine, and histidine. One of these positive clones encoded the COOH-terminal portion (residues 332 to 617) of Vma1p, the catalytic 69-kDa V-ATPase subunit. Co-expression of the NH 2 terminus of the 95-kDa subunit (Nt-Vph1p) and the COOH terminus of Vma1p (Ct-Vma1p) resulted in activation of ␤-galactosidase transcription (Table I). Activation was independent of whether Nt-Vph1p or Ct-Vma1p were in either pACT or pAS2-1 vectors, but neither Nt-Vph1p nor Ct-Vma1p activated ␤-galactosidase transcription with the counter empty vector, confirming that protein-protein interaction between the V-ATPase subunits were required for ␤-galactosidase activation. The amino terminus of Stv1p, the functional homologue of Vph1p (38), was equally effective in activating ␤-galactosidase activity in the presence of Ct-Vma1p, further demonstrating the specificity of the interaction between the V-ATPase subunits.
The NH 2 Terminus of Vph1p Interacts with the Soluble V 1 Complex in Vivo-The interaction between Vph1p and Vma1p was confirmed in vivo by co-immunoprecipitation of the two proteins. Co-immunoprecipitation in a wild type strain would be uninformative, as antibodies to any V-ATPase subunit will immunoprecipitate the holoenzyme. Thus, Nt-Vph1p was expressed in a ⌬vph1 yeast mutant lacking the V 0 sector, yet expressing fully assembled and functional V 1 complexes (39). Yeast whole cell lysates were prepared from the ⌬vph1 mutant strain MM53 transformed with the supporting vector alone (MM53 ϩ pRS426) or with a vector expressing the NH 2 terminus of Vph1p (MM53 ϩ pRS426-Nt-Vph1p). As expected, immunoprecipitation with an ␣-Vma1p antibody recovered the catalytic subunit from the vector alone (V) and Nt-Vph1p (I) expressing lysates (Fig. 1, lanes 1 and 2). The 69-kDa doublet seen in lane 2 likely reflects post-translational modification of Vma1p; size heterogeneity of the V-ATPase catalytic subunit has been noted in several other species (30). Immunoprecipitation of Vma1p from lysates expressing Nt-Vph1p recovered both the catalytic subunit and the 55-kDa Nt-Vph1p (Fig. 1,  lane 2). This observation demonstrates that Nt-Vph1p interacts with the soluble V 1 domain in vivo, confirming the results obtained from the original two-hybrid screen. Binding of Nt-Vph1p likely occurs on the external surface of the catalytic hexamer given that assembly of the V 1 complex (catalytic core (34,40) and stalk (15)) is independent of Vph1p and that Nt-Vph1p binds to a fully assembled complex.
Incubation of the yeast lysate (MM53 ϩ pRS426-Nt-Vph1p) with ␣-Vph1p antibodies successfully immunoprecipitated the NH 2 terminus of Vph1p (Fig. 1, lane 3). Unfortunately, only trace levels of Vma1p were recovered under these experimental conditions. The majority of Nt-Vph1p precipitated likely represents free protein, with only a minimal component associated with the V 1 complex. The failure to immunoprecipitate bound Nt-Vph1p suggests that association of the truncated 95-kDa subunit to the catalytic complex occludes available epitopes. Immunofluorescence microscopy with ␣-Nt-Vph1p antibodies is only successful when the V 0 complex is dissociated from the catalytic sector (27,37), demonstrating that association of the V 1 complex to the membrane sector blocks access to Nt-Vph1p epitopes. Conversely, immunoprecipitation of the catalytic subunit would not be impeded by Nt-Vph1p binding if there is unequal stoichiometry between the two subunits. Even with Nt-Vph1p binding to one or two Vma1p subunits, the third Vma1p subunit would remain accessible to antibodies. The levels of immunoprecipitated Nt-Vph1p in lanes 2 and 3 do not reflect levels of bound versus unbound protein as immunoprecipitations were performed with an excess of cell lysate compared with antibodies.
Efforts to refine the essential domains required for association between Vph1p and Vma1p yielded inconclusive results. Co-expression of truncated domains of the interactive regions of Vph1p and Vma1p did not result in reproducible interactions. The failure to identify discreet regions required for Vma1p-Vph1p interaction may indicate that the complete domains identified are essential to establish stable protein-protein contacts. Alternatively, the inability to demonstrate inter-  action between truncated forms of these proteins may indicate that the interaction between Vma1p and Vph1p is weak and requires additional components of the V 1 complex to stabilize the association.
The NH 2 Terminus of Vph1p Is Required for Vma13p Association with the V 1 Sector-Given that Vph1p is part of the V 0 complex, and that Nt-Vph1p interacts directly with the catalytic subunit of the V 1 complex, it is reasonable to suggest that the 95-kDa subunit serves as a fixed structural link between the two V-ATPase sectors. If this is true, Vph1p is likely to interact with additional components of the V 1 complex. Of the remaining subunits of the catalytic sector, we focused on Vma13p for a number of reasons. Dissociation of the peripheral sector from the vacuolar membrane yields an inactive complex containing the majority of V 1 components (69, 60, 32, and 27 kDa) and free forms of Vma5p (42 kDa) and possibly Vma13p (54 kDa) (37,41). Sector re-association results in recruitment of Vma13p and Vma5p to the V 1 sector with concomitant restoration of V-ATPase activity (41). Given that Vph1p is essential for V 1 V 0 association and that Vma13p only binds to a fully assembled complex, we hypothesized that Nt-Vph1p may be required for recruiting the 54-kDa subunit to the catalytic complex.
To determine if this was the case, yeast cell lysates were prepared from a ⌬vph1 mutant strain transformed with a construct expressing Nt-Vph1p or the carrier vector alone (as described above). Immunoprecipitation with ␣-Vma1p sera was unable to co-precipitate Vma13p in the absence of Nt-Vph1p (Fig. 1, lane 1), suggesting that the 54-kDa subunit is not a component of the isolated V 1 complex. Immunoprecipitation of the catalytic subunit in the presence of Nt-Vph1p did result in the co-precipitation of Vma13p (Fig. 1, lane 2, bottom panel). Protein extracted from a ⌬vma13 mutant demonstrated that the 65-kDa polypeptide seen in lanes 1 and 2 of the blots probed with ␣-Vma13p sera is not Vma13p (data not shown). These results suggest that Nt-Vph1p is essential for the interaction of Vma13p with the catalytic sector, recruiting the 54-kDa subunit from its free cytosolic state to the fully assembled V-ATPase. If Nt-Vph1p interacts directly with Vma13p, one might expect co-immunoprecipitation of Nt-Vph1p and Vma13p, even in the absence of the V 1 complex. Surprisingly, immunoprecipitation with ␣-Vph1p sera was unable to co-precipitate Vma13p (Fig. 1, lane 3). One possibility to explain why a V 1 -free Vph1p-Vma13p interaction was not detected by immunoprecipitation could be that bound Vma13p occludes Vph1p epitopes. Alternatively, the NH 2 terminus of Vph1p may not interact with Vma13p alone, but could require additional components of the V 1 complex to mediate the interaction. A third possibility is that binding of Nt-Vph1p to the V 1 complex could induce a conformational change within the catalytic sector, allowing for Vma13p binding. Thus, the ability of the NH 2 terminus of Vph1p to recruit Vma13p to the catalytic sector may be an indirect effect rather than through direct binding. Either of the latter interpretations would be consistent with the failure of the original two-hybrid screen to select for a Vma13p clone. To distinguish between these possibilities, we asked whether Vma13p could interact directly with Vph1p in an in vitro assay.
The NH 2 Terminus of Vph1p Interacts with the COOH Terminus of Vma13p in Vitro-To determine if Nt-Vph1p recruits Vma13p to the V 1 complex by direct interaction or by inducing a conformational change within the catalytic sector, an in vitro Nt-Vph1p-Vma13p binding assay was designed. Briefly, the sequence encoding Nt-Vph1p was cloned into the pET32b expression plasmid and was used to produce histidine-tagged (His-tagged) Nt-Vph1p (pET32b-Nt-Vph1p, see "Experimental Procedures"). Bacterial lysates were also prepared from a strain expressing a truncated form (aa 160 -478) of Vma13p (Ct-Vma13p), a kind gift from Dr. Tom Stevens as described in Ref. 32. This truncated form of Vma13p fully complements a ⌬vma13 disruption indicating that removal of the 159 N 2 Hterminal region is not required for V-ATPase assembly or activity (32). Ni-NTA bound proteins produced from pET-32b-Nt-Vph1p or the pET-32b vector alone were incubated with equal amounts of bacterial lysates containing Ct-Vma13p. The Ni-NTA bound and unbound fractions were assayed by immunoblotting as described under "Experimental Procedures." Ct-Vma13p did not bind to the protein purified from the unmodified pET32b vector, and was completely restricted to the unbound fraction (Fig. 2, panels B and C, lanes 2 and 3). In the presence of Nt-Vph1p, Ct-Vma13p was retained in the bound fraction (Fig. 2, lane 4) demonstrating that Vma13p and Nt-Vph1p interact directly. Despite the overabundance of Nt-Vph1p, approximately 50% of the available Vma13p remained in the unbound fraction (Fig. 2, lane 5). This population of unbound Vma13p may indicate unequal stoichiometry of these subunits in vivo or that the bacterially expressed constructs are not assuming native conformations in solution. Dynamic lightscattering experiments with heterologously expressed Nt-Vph1p revealed that the protein is polydisperse (data not shown) indicating aggregation in solution. Aggregation of Nt-Vph1p could occlude Vma13p-binding sites.
Dissociation of ATPase and Proton Pumping Activity-Both in vivo (37,42) and in vitro (43)(44)(45)(46)(47), V 1 ATP hydrolytic activity is tightly coupled to V 0 proton translocation. V 1 V 0 dissociation renders the V 1 hydrolytically inactive (43)(44)(45)(46) and the V 0 proton impermeant (47), with both activities restored by complex reassociation (37,42). Chemical inhibition of the V 1 (by Nethylmaleimide) or the V 0 (by bafilomycin A 1 ) prevents both proton translocation and ATP hydrolysis, reiterating the functional dependence of the two sectors. Surprisingly, limited pro- teolysis of the V-ATPase in clathrin-coated vesicles partially dissociated these two activities (48). Trypzinization completely abolished proton translocation with only a 50% reduction in ATP hydrolysis leading to the conclusion that proteolytic cleavage of one or more of the mammalian homologues of Vph1p, Vma1p, Vma2p, Vma6p, and/or Vma4p was responsible for uncoupling of the two activities. To elucidate whether cleavage of one specific subunit was responsible for the dissociation of hydrolysis from proton pumping, intact vacuolar membranes were treated with 0.1 g/ml trypsin as described under "Experimental Procedures." These conditions are gentler than those previously employed, yielding a more discrete pattern of proteolytic cleavage. Only Vph1p, Vma2p, and Vma13p were cleaved under the conditions utilized (Fig. 3, panels 1, 3, and 4). The remaining subunits (Vma1p, Vma5p, Vma6p, and Vma4p) remain intact (Fig. 3, panels 2, 5, 6, and 7), with a 10-fold increase in the amount of trypsin not cleaving Vma1p, Vma4p, or Vma6p. The possibility that Vma7p, Vma8p, or Vma10p are cleaved and thus contribute to uncoupling cannot be excluded, as antibodies to these subunits were unavailable.
To determine if cleavage of Vph1p, Vma2p, and/or Vma13p subunits was sufficient to uncouple hydrolysis from proton translocation, vacuolar membranes were trypsinized under identical conditions described above and assayed for proton translocating and ATP hydrolytic activity. Trypsin treatment did not have a significant effect on ATP hydrolysis (Fig. 4,  panel A). The minimal reduction in ATPase activity with or without trypsin treatment is likely a consequence of thermal denaturation. In contrast, the same trypsin treatment resulted in a significant decrease in proton translocation. Trypsintreated samples showed a 22 and 67% reduction in proton pumping at 15 and 30 min, respectively, as compared with control (Fig. 4, panel B). These data suggest that proteolytic cleavage of Vph1p, Vma2p, or Vma13p, alone or in combination, results in the functional uncoupling of the two V-ATPase complexes.
Due to its unique bipartite structure and its interactions with components of the V 1 complex, Vph1p is ideally suited to serve as a structural link between the V 1 and V 0 complexes. To test whether the proteolytic cleavage of Vph1p was responsible for the dissociation of hydrolysis from proton translocation, the trypsin-sensitive site on Vph1p was identified and eliminated by site-directed mutagenesis. We have previously demonstrated that the limited trypsinization of intact vacuolar membranes results in the cleavage of 6 kDa from the NH 2 terminus of Vph1p (27). The remaining 90-kDa COOH-terminal fragment was isolated by SDS-PAGE, transferred to PVDF membrane, and subjected to NH 2 -terminal sequencing as described under "Experimental Procedures." The NH 2 -terminal 7-amino acid residues of the 90-kDa fragment are T 54 FVNEIR indicating that the preceding arginine (Arg 53 ) residue is the site of trypsin cleavage. Given that intact vacuoles were subjected to limited proteolysis, identification of arginine 53 as the trypsin cleavage site reconfirms that the hydrophilic NH 2 terminus of Vph1p is disposed toward the cytosolic space. To remove the trypsin cleavage site within the NH 2 terminus of Vph1p, arginine 53 was mutated to a serine residue as described under "Experimental Procedures" (MM635). The single copy CENbased plasmid pRS316-VPH1-R53S (MM636) was able to complement a ⌬vph1 mutation (MM53) and rendered Vph1p insensitive to 0.1 g/ml trypsin demonstrating that the R53S mutation eliminated the NH 2 -terminal proteolytic cleavage site while not interfering with the function of Vph1p (data not shown). In order to purify vacuoles from rich non-selective media, the VPH1 R53S mutation was integrated into the yeast genome as described under "Experimental Procedures" (MM638: ␣ ura3-52::URA3::VPH1-R53S ⌬vph1::LEU2); MM638 regained bafilomycin A 1 -sensitive ATPase activity and vacuolar acidification. Vacuolar membranes purified from MM638 and BJ926 (wild type control) were subjected to limited proteolysis for 15 and 30 min as described above. As previously noted, bafilomycin A 1 -sensitive ATPase activity was equally affected with or without proteolysis with 0.1 g/ml trypsin in vacuoles purified from either strain (Fig. 4, panel C). In contrast, the R53S mutation partially protects the MM638 vacuoles from loss of proton conductance when compared with wild type vacuoles (Fig. 4, panel D). That the R53S mutation only afforded limited protection was anticipated as 0.1 g/ml trypsin did not achieve 100% proteolysis of Vph1p (Fig. 3, panel 1), while proton translocation was effectively abolished (Fig. 4,  panel B). This suggests that cleavage of additional components of the V-ATPase also contribute to the loss of proton pumping activity. It is tempting to speculate that proteolysis of Vma13p may be involved in uncoupling ATP hydrolysis from proton translocation given that Vma13p was also cleaved by limited proteolysis (Fig. 3, panel 4), that it does not have an established Final trypsin concentrations were 0.1 or 1.0 g/ml, as indicated above the panels. All lanes contain 2 g of purified total vacuolar proteins. Panels 1-7, immunoblots were probed with the following antibodies: 1, 1:5000 ␣-Vph1p; 2, 1:5000 ␣-Vma1p; 3, 1:10,000 ␣-Vma2p; 4, 1:1000 ␣-Vma13p; 5, 1:500 ␣-Vma5p; 6, 1:1000 ␣-36 kDa; 7, 1:500 ␣-27 kDa. Size of the full-length protein for each of the V-ATPase subunits assayed is noted on the left side of the corresponding panel.
role in ATP hydrolysis, and that it interacts with Vph1p ( Figs.  1 and 2). However, it cannot be concluded that cleavage of Vma13p is responsible for the remaining 38% of proton-pump inhibition not eliminated by the R53S mutation without controls similar to those performed on Vph1p. Nevertheless, these results do show that the proteolytic cleavage of the NH 2 terminus of Vph1p is, in part, responsible for dissociating V 1 ATP hydrolysis from V 0 proton translocation.

DISCUSSION
The V-ATPases and F-ATPases share a common architecture with V-ATPase subunits having structural counterparts within F-ATPases. An exception to this generalization is the 95-kDa V 0 subunit which is unique to V-ATPases. This molecule is divided into two distinct structural domains, the NH 2 -terminal hydrophilic domain (ϳ45 kDa) and the COOH-terminal membrane-associated region. To date, the function of this subunit has not been elucidated. The initial observations that Vph1p isoforms were localized to specific organelles (38) and were differentially expressed in mammalian tissues (49 -51) suggested that Vph1p was involved in the targeting or regulation of single copy V-ATPase subunits in various cellular locations (52). Recent experimental evidence points to a more direct role in V-ATPase activity. Point mutations within the transmembrane domain of Vph1p do not impair V-ATPase assembly or targeting yet eliminate enzyme activity (53,54). Moreover, substrate binding to the catalytic domain induces conformational changes within Vph1p (27). Given the localization of the NH 2 terminus of Vph1p to the cytoplasm and the subunit's conformational responsiveness to substrate binding, it was proposed that this domain interacts directly with the V 1 sector (27). In vitro and in vivo binding studies demonstrate that the NH 2 terminus of Vph1p interacts with the catalytic subunit, Vma1p, and Vma13p (Table I and Figs. 1 and 2). Although Vma13p is a component of the soluble sector (55), recruitment of the 54-kDa subunit to V 1 requires the expression of Nt-Vph1p (Fig. 1, panel C).
As with Vph1p, the function of the 54-kDa V 1 subunit (Vma13p) has not been defined, although it is essential for V-ATPase activity (55). Genetic (55) or biochemical (56,57) removal of Vma13p from the holoenzyme activity results in complex inactivation. Although Vma13p is not required for the assembly of either complex, or V 1 /V 0 association, it stabilizes FIG. 4. Trypsinization of the yeast V-ATPase leads to the dissociation of ATPase catalytic activity and proton translocation. Panel A, bafilomycin A 1 -sensitive ATP hydrolysis of vacuolar membranes pretreated with trypsin (0.1 g/ml) as described for pumping assays. Hydrolytic activity was assayed for 30 min at 30°C in 25 mM MES-Tris, pH 6.9, 5 mM MgCl 2 , 5 mM Na 2 ATP, and 25 mM KCl with released inorganic phosphate determined by the Ames method. Bafilomycin-treated samples were preincubated with a final inhibitor concentration of 1 M for 10 min at room temperature prior to the assays. 6 g of total protein were utilized per assay (n ϭ 4). All assays were performed in the presence of 0.05 mM vanadate and 0.5 mM azide. Hatched bars correspond to control and shaded bars correspond to trypsinized samples. Panel B, V-ATPase-mediated acidification of purified intact vacuoles pretreated with trypsin. Intact vacuoles were pretreated with 0.1 g/ml trypsin for 0, 15, or 30 min in the presence of 5 mM ATP, 25 mM KCl, MES-Tris, pH 7.8. Control samples were prepared under identical conditions. Reactions were stopped by the addition of final concentrations of 1 mM PMSF and 5 mM TLCK. Luminal acidification was detected by following the rate of fluorescence quenching of acridine orange in buffer conditions identical to those described for the ATPase activity assays. 20 g of total protein was utilized per assay.
Hatched bars correspond to control and shaded bars correspond to trypsinized samples. Relative activities were calculated based on the maximal extent of acidification. Panels C and D, a plasmid coding for Vph1p lacking the trypsin-sensitive site (R53S) was expressed in a ⌬vph1p yeast strain. Vacuolar membranes purified from either this strain or a wild type strain were used to perform the experiments detailed in panels C and D. Panel C, bafilomycin-sensitive ATP hydrolysis of vacuolar membranes prepared from wild type and (R53S) yeast strains were pretreated with trypsin (0.1 g/ml) as described in panel A. White bars correspond to wild type vacuolar membranes and shaded bars correspond to (R53S) vacuolar membranes. Panel D, V-ATPase mediated acidification of vacuoles prepared from wild type and (R53S) yeast strains pretreated with trypsin (0.1 g/ml) was performed as described in panel B. White bars correspond to wild type vacuolar membranes and shaded bars correspond to (R53S) vacuolar membranes.
binding of the catalytic sector to the vacuole (55). Dissociation/ re-association and subsequent inactivation/re-activation of V-ATPase activity has also been observed for the bovine homologue of Vma13p, ␣SFD, in clathrin-coated V-ATPases (56,57). This dynamic process necessitates that Vma13p be peripherally located with respect to the catalytic core, allowing for its release without disruption of the Vma1p/Vma2p hexameric center. The observations that Vma13p interacts with the NH 2 terminus of Vph1p and that it stabilizes the association of the two V-ATPase sectors suggest that the 54-kDa subunit forms part of a fixed structural link between V 1 and V 0 .
The only other subunit with similar properties to Vma13p is the 42-kDa V 1 subunit, Vma5p. With the exception of Vma13p and Vma5p, disruption of an individual V-ATPase subunit prevents assembly of its parent complex (39). Vma13p and Vma5p readily dissociate from the holoezyme, inactivating the V-ATPase complex (32,55). To reproducibly co-immunoprecipitate the two subunits with the holoenzyme, the complex must be chemically cross-linked prior to solubilization (39). Furthermore, although both Vma13p and Vma5p are considered V 1 subunits, neither subunit is part of the catalytic core composed of Vma1p and Vma2p, or the central stalk, thought to be composed of Vma4p, Vma7p, Vma8p, and Vma10p (15,40,58). Taken together these data suggest that Vma13p and Vma5p bind at the periphery of a fully assembled V 1 complex. Moreover, data presented here suggest that, in the case of Vma13p, Vph1p is essential for the recruitment of the subunit to the V 1 sector.
The NH 2 terminus of Vph1p, through its interaction with Vma1p, also associates with the V 1 complex. Since V 1 and V 0 complexes can disassemble/re-assemble in vivo (37,42), wild type Vph1p must also readily dissociate/re-associate from the V 1 complex. Given that Vph1p, Vma13p, and Vma5p all readily dissociate/re-associate with assembled V 1 complexes, and that they are not part of the catalytic core or central stalk, it is reasonable to suggest that these three subunits are asymmetrically attached to the periphery of the V 1 complex. This has led us to hypothesize that Nt-Vph1p, Vma13p, and possibly Vma5p compose the peripheral stalk visualized by electron microscopy (6,25) and likely represents a fixed structural link between V 1 and V 0 . This asymmetric peripheral structure is analogous to the one identified in the closely related F-ATPases. In this complex, subunits a, (b) 2 , and ␦ form a stator which prevents loss of energy through the futile rotation of F 1 relative to F 0 . A V-ATPase stator would serve a similar role by preventing rotation of V 1 relative to V 0 , permitting the efficient transduction of torque exerted by the rotating stalk on V 0 . Given this functional model, proteolytic cleavage of the stator would likely result in uncoupling of ATP hydrolysis from proton translocation. Trypsin treatment of purified vacuoles results in cleavage of both Vph1p and Vma13p with the resultant loss of proton translocation and preservation of ATP hydrolysis (Figs. 3 and 4). Removal of the trypsin-sensitive site on Vph1p partially prevents this phenomenon, suggesting the 95-kDa subunit plays a key role in coupling the two V-ATPase complexes. Consistent with our model of Vma13p forming part of the stator, cleavage of Vma13p would also contribute to the dissociation of V-ATPase activity.
It is of note that the uncoupled ATP hydrolysis resulting from trypsin proteolysis remained sensitive to bafilomycin A 1 . This V-ATPase-specific inhibitor binds tightly but noncovalently to the hydrophobic V 0 complex, blocking passive proton conductance (16,59,60). Evidence suggests that the binding site is located on either the proteolipid Vma3p (61) or Vph1p (16), presumably on its hydrophobic COOH-terminal domain. Given that bafilomycin A 1 binds to the V 0 complex, one might expect trypsin-uncoupled V 1 ATP hydrolysis to be bafilomycin insensitive. Our results to the contrary indicate that the inhibitory effect of bafilomycin A 1 is not through disruption of the V 1 -Nt-Vph1p interaction but rather through another V 1 V 0 interaction. Models of V-ATPases suggest two separate links between the V 1 and V 0 complex: the peripheral stator (that we propose is partly composed of Nt-Vph1p) and the central rotor. Bafilomycin A 1 could inhibit ATP hydrolysis by preventing rotation of the central rotor. Indeed, one would expect a potent inhibitor to affect a dynamic rather than a static interaction FIG. 5. Model of V 1 and V 0 dissociation and reassociation. The NH 2 terminus of Vph1p (Nt-Vph1p) interacts with the catalytic subunit of the V 1 complex. This region of Vph1p is also required to recruit Vma13p to the V 1 complex. Dissociation of the V 1 complex releases both Vma13p and Vma5p into the cytosol with complex reassociation resulting in their recruitment. In mammalian V-ATPases, there are two isoforms of the Vma13p ortholog, ␣SFD and ␤SFD, each present in one copy per complex (14,57) and one copy of the Vma5p ortholog, subunit C, per complex (14). The question of whether there is one or more copies of Vph1p/complex is still under debate (17,18). We hypothesize the Vph1p is part of the peripheral stalk visualized by electron microscopy, and have thus drawn two subunits of Vph1p/complex to reflect the two peripheral stalks/ complex resolved by Boekema et al. (25).
within an enzyme complex. The fact that the uncoupled ATP hydrolysis remains bafilomycin-sensitive fits well with our model of the Nt-Vph1p-V 1 interaction composing the static link and the central rotor as the dynamic link between the V 1 and V 0 complexes.
The stoichiometry of Vph1p is still uncertain as quantitative amino acid analysis has calculated from one (17), to three (18) Vph1p subunits per complex. We show here that Nt-Vph1p interacts with Vma1p. As Vma1p is present in three copies/ complex, there are three binding sites available for Nt-Vph1p. Although there is nothing to suggest that all available binding sites are occupied, more than one Vph1p subunit per complex can be theoretically accommodated. We hypothesize here that Nt-Vph1p is part of the peripheral stalk structure visualized by electron microscopy (6,7). As recent electron microscopy by Boekema et al. (25) reveal that there are at least two, and possibly three peripheral stalks, we have drawn Fig. 5 to depict multiple copies of Vph1p.
Recent work (14) has identified a potential contact between Vph1p and the stalk subunit Vma4p. Based on our model, this interaction would be predicted to be transient, allowing Vma4p to rotate with respect to Vph1p.