Substrate- and Inhibitor-induced Conformational Changes in the Yeast V-ATPase Provide Evidence for Communication between the Catalytic and Proton-translocating Sectors*

The vacuolar-type H+-ATPases (V-ATPases) are composed of two distinct sectors, a catalytic complex (V1) involved in ATP hydrolysis and a membrane-associated complex (V0) mediating proton translocation across a lipid bilayer. To date, little is known about the mechanism by which these two functions are coupled. We sought to examine the impact of nucleotide and cation binding on the structure of the core components of the catalytic complex and to determine whether conformational changes within the catalytic complex impact subunits of the membrane-associated complex. Nucleotide- and cation- induced changes in the catalytic core of the V-ATPase were investigated by monitoring changes in the rate and pattern of tryptic digests. ATP·Mg-induced changes were detected in both the catalytic (Vma1p or 69 kDa) and the regulatory subunits (Vma2p or 60 kDa) of the V1 sector. ATP alone increased the rate of trypsinization of the regulatory subunit, but did not have any effect on Vma1p. Surprisingly, ATP also had an impact on the 95-kDa subunit, a component of the V0 sector of the V-ATPase. Although the presence of divalent cations had no impact on the V1 sector, the rate of trypsinization of the 95-kDa subunit was greatly enhanced. The effect of divalent cations on the structure of the 95-kDa subunit was abrogated when trypsinization was performed in the absence of the catalytic sector. Addition of bafilomycin A1, a V-ATPase inhibitor that putatively binds to the 95-kDa subunit, increased the rate of trypsinization of the catalytic subunit. These data suggest that structural alterations within the V1 sector result in alterations within the V0 sector and vice versa. Clearly, a structural link must exist to couple the two sectors. The 95-kDa subunit is ideally suited to fulfill this role. Hydropathy analysis suggests a bipartite structure, with the NH2-terminal portion predicted to lie in an aqueous environment and the C-terminal portion predicted to contain 6 transmembrane segments. Tryptic digests of sealed vacuolar vesicles and immunofluorescence studies revealed that the large hydrophilic NH2-terminal domain of the 95-kDa subunit is localized toward the cytosol. This region therefore is ideally positioned to interact with components of the V1 complex, potentially functioning as the elusive link between the two sectors of the V-ATPase.

The vacuolar-type H ؉ -ATPases (V-ATPases) are composed of two distinct sectors, a catalytic complex (V 1 ) involved in ATP hydrolysis and a membrane-associated complex (V 0 ) mediating proton translocation across a lipid bilayer. To date, little is known about the mechanism by which these two functions are coupled. We sought to examine the impact of nucleotide and cation binding on the structure of the core components of the catalytic complex and to determine whether conformational changes within the catalytic complex impact subunits of the membrane-associated complex. Nucleotideand cation-induced changes in the catalytic core of the V-ATPase were investigated by monitoring changes in the rate and pattern of tryptic digests. ATP⅐Mg-induced changes were detected in both the catalytic (Vma1p or 69 kDa) and the regulatory subunits (Vma2p or 60 kDa) of the V 1 sector. ATP alone increased the rate of trypsinization of the regulatory subunit, but did not have any effect on Vma1p. Surprisingly, ATP also had an impact on the 95-kDa subunit, a component of the V 0 sector of the V-ATPase. Although the presence of divalent cations had no impact on the V 1 sector, the rate of trypsinization of the 95-kDa subunit was greatly enhanced. The effect of divalent cations on the structure of the 95-kDa subunit was abrogated when trypsinization was performed in the absence of the catalytic sector. Addition of bafilomycin A 1 , a V-ATPase inhibitor that putatively binds to the 95-kDa subunit, increased the rate of trypsinization of the catalytic subunit. These data suggest that structural alterations within the V 1 sector result in alterations within the V 0 sector and vice versa. Clearly, a structural link must exist to couple the two sectors. The 95-kDa subunit is ideally suited to fulfill this role. Hydropathy analysis suggests a bipartite structure, with the NH 2terminal portion predicted to lie in an aqueous environment and the C-terminal portion predicted to contain 6 transmembrane segments. Tryptic digests of sealed vacuolar vesicles and immunofluorescence studies revealed that the large hydrophilic NH 2 -terminal domain of the 95-kDa subunit is localized toward the cytosol. This region therefore is ideally positioned to interact with components of the V 1 complex, potentially functioning as the elusive link between the two sectors of the V-ATPase.
Vacuolar-type H ϩ -ATPases (V-ATPases) 1 are multimeric complexes that mediate the luminal acidification of various organelles (e.g. yeast and plant vacuoles, endosomes, lysosomes, phagosomes, and clathrin-coated vesicles) (for review, see Ref. 1). Organellar acidification is essential for a variety of cellular processes, including receptor-mediated endocytosis, processing and proteolysis of proteins, intracellular degradation of ingested pathogens and proton-coupled transport of small molecules. The V-ATPase complex contains two distinct structural sectors, a peripheral catalytic complex (V 1 ) mediating ATP hydrolysis and a membrane-associated complex (V 0 ) defining the proton channel proper (for review, see Ref. 1).
The soluble catalytic sector of the yeast V-ATPase consists of at least 8 subunits. The catalytic core is composed of a hexamer formed by alternating 69-kDa (Vma1p or A) and 60-kDa (Vma2p or B) subunits symmetrically arranged to generate a hollow core through the middle of the sphere. Vma1p and Vma2p both contain consensus sequences for ATP-binding domains; Vma1p has been ascribed a catalytic function, whereas Vma2p is thought to have a regulatory function. The hollow core formed by the ATP-binding subunits contains a central stalk, which is likely composed of the Vma4p, Vma7p, Vma8p, and Vma10p subunits (27,14,32, and 13 kDa, respectively) (2). The function of the V 1 subunits Vma13p (54 kDa) and Vma5p (42 kDa) is not known.
The membrane-bound V 0 complex is composed of at least 5 subunits. The proteolipid believed to form the proton channel (3) is encoded by three genes, VMA3 (17 kDa), VMA11 (17 kDa), and VMA16 (23 kDa) (4). The 100-kDa V 0 subunit is encoded in yeast by VPH1 (5) and STV1 (6). Vph1p and Stv1p are functional homologues, but they are targeted to different organelles within the cell, suggesting that this subunit is involved in the differential regulation of V-ATPase activity. The function of the 36-kDa V 0 subunit, Vma6p, is not known.
The V-type and the mitochondrial F-type ATPases share a common structural architecture. The F-type ATPases are also defined by a bipartite structure with a hydrophilic catalytic sector (F 1 ) and a hydrophobic proton-conducting component (F 0 ) (7). The crystal structure of the F 1 ATPase sector shows the catalytic (␤) and regulatory (␣) subunits arranged in an alternating hexagonal circular array, with the nucleotide binding sites localized to the ␤-␣ interface (8). The catalytic binding sites are located in the ␤-subunit with the noncatalytic sites restricted to the ␣-subunit. Because of the structural similarity between the ATPases, the catalytic and noncatalytic nucleotide binding sites in the V 1 sector are likely to be similarly arranged.
Several lines of evidence substantiate that the A subunit of the V-ATPases contains the catalytic binding sites. The A subunit contains all the consensus loops, including a glycine-rich loop region (9) that was identified as critical for ATP hydrolysis in the ␤-subunit of the F-ATPases. In addition, the inhibitors N-ethylmaleimide and 7-chloro-4-nitrobenz-2-oza-1,3-diazole label the A subunit by covalently modifying a critical cysteine residue in the ATP binding site (10,11). Site-directed mutagenesis of a single cysteine residue within the glycine-rich loop yields a fully active protein resistant to both inhibitors (12,13). Finally, photolabeling of the A subunit with 2-azido-[ 32 P]ATP correlates directly with inhibition of ATPase activity (14).
Current experimental evidence suggests that the nucleotide binding sites within the B subunit are noncatalytic. Although modification with photoaffinity analogues (3-0-(4-benzoyl)benzoyl-ATP (15) and 2-azido-[ 32 P]ATP (14) has demonstrated the presence of nucleotide binding sites within the B subunit, there is little evidence to suggest that these sites are catalytic in nature. The absence of a glycine-rich consensus sequence within the B subunit supports this conclusion.
The impact of nucleotide binding on the conformation of the V 1 catalytic and regulatory subunits remains unclear. The available data is contradictory; some evidence suggests that ATP binding to the catalytic subunit does not induce a conformational change, as assayed by alterations in sensitivity to proteolytic degradation (16). Conversely, conditions optimal for ATP hydrolysis promote the dissociation of the peripheral sector from the membrane, suggesting that catalysis induces a conformational change within V 1 that facilitates its release (17,18). Site-directed mutagenesis of the yeast catalytic subunit demonstrated that loss of activity impaired removal of the peripheral complex by KNO 3 (19). These data strongly suggest that the catalytic subunit alters its conformation in response to substrate binding and/or hydrolysis.
Dissociation of the V 1 V 0 complex results in a loss of proton translocation and ATP hydrolysis, with complex reassociation restoring both activities (20,21). The ability to reassociate into an active complex demonstrates that both sectors are functionally interdependent although remaining structurally independent in isolation. The structural independence of the two sectors has been demonstrated in vivo (22). Mutants lacking a single subunit of the enzyme yielded assembled cytosolic V 1 sectors in the absence of a V 0 subunit, and membrane-associated assembled V 0 sectors in the absence of a V 1 constituent (22). Despite their structural autonomy, physical coupling of the two sectors is essential for activity, suggesting that protein subunits of a given sector may be responsive to conformational changes within its functional counterparts. As physical interactions between the two sectors is an absolute requirement for function, a number of structural links must exist to allow for sector interaction. To date, no such link has been identified for V-ATPases. Here, we present evidence suggesting that Vph1p is part of the structural link.
The Vph1p subunit is unique to V-ATPases. Hydropathy analysis of Vph1p predicts two distinct structural domains, a hydrophilic NH 2 -terminal domain (ϳ50 kDa) and a hydrophobic C-terminal domain (ϳ50 kDa) that contains 6 -8 putative membrane-spanning regions. Through the use of limited proteolysis of intact vacuolar membranes and immunoblotting, as well as immunofluorescence, we have unambiguously localized the NH 2 terminus of Vph1p to the cytosol. The disposition of the NH 2 -terminal Vph1p domain toward the cytosolic space would permit physical interactions of Vph1p with protein sub-units of the V 1 sector. Because of its unique bipartite structure and its structural responsiveness to substrate binding, we propose that Vph1p is ideally suited to serve as a structural and functional link between the two sectors of the V-ATPase.
Purification of Vacuolar Membranes-Isolation of transport-competent vacuolar membrane vesicles through flotation of intact vacuoles on Ficoll gradients was performed as described in (5). Protein concentrations of vacuolar vesicles were measured by Lowry assays.
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 25 mM MES/ Tris, pH 7.8, 25 mM KCl were incubated with varying concentrations of trypsin at 37°C for 2 h. In addition, 5 mM of nucleotide (ATP, ADP, or ATP␥S) and 5 mM divalent cation (MgCl 2 or CaCl 2 ) were included during proteolysis as indicated for each experiment. The final reaction volume was 100 l. Trypsin proteolysis was inhibited by the addition of phenylmethylsulfonyl fluoride to a final concentration of 1 mM and tosyl-L-lysine chloromethyl ketone to a final concentration of 5 mM followed by incubation on ice for 15 min. Samples were denatured by the addition of 5ϫ SDS sample buffer, assayed by SDS-PAGE (24), and immunoblotting as described by Olmsted (25).
Proteolytic digests performed in the presence of ADP or ADP⅐Mg were treated as described above with a final concentration of 5 mM ADP and 5 mM MgCl 2 . Samples prepared in the presence of bafilomycin A 1 were preincubated for 10 min at room temperature. Bafilomycin A 1 was prepared in Me 2 SO and added to a final concentration of 1 M. Control samples were corrected for the addition of Me 2 SO.
Vacuolar Integrity-The integrity of the purified vacuolar membranes to exogenously added proteases was assayed by the ability of protease K to cleave pro-carboxypeptidase Y (CPY) to its mature form in the absence and presence of detergent. Intact vacuoles purified from BJ1991 were resuspended at 0.5 g/l in 10 mM MES-Tris, pH 6.9, and 30 g/ml protease K in the presence or absence of 0.5% (w/v) Triton X-100 for 0, 2, 10, and 20 min, respectively. Proteolytic digests were terminated at the designated times with the addition of phenylmethylsulfonyl fluoride to a final concentration of 22.2 mM followed immediately by protein precipitation with ice-cold 20% trichloroacetic acid. Protein pellets were washed in ice-cold acetone to remove residual acid and denatured for SDS-PAGE followed by immunoblotting with a monoclonal antibody to CPY, 10A5-B5 (Molecular Probes).
Raising of NH 2 Terminus Vph1p Antibody and Affinity Purification-The antigen for immunization of the rabbits was keyhole limpet hemocyanin-conjugated via a cysteine group to the nonadecapeptide (VSEL[E 3 G]ELG[K 3 L]VQFRDLNPDV) corresponding to residues 26 to 44 of the rat 116-kDa subunit, with bolded residues changed as indicated to correspond more closely with the yeast Vph1p sequence. Each rabbit was injected with 1 mg of peptide/carrier in Freund's complete adjuvant. At days 20, 38, 45, and 60 each rabbit was reinjected with 1 mg of labeled keyhole limpet hemocyanin in Freund's incomplete adjuvant. Rabbits were sacrificed at day 70. Serum was collected by incubating the rabbit's blood at 37°C until coagulation, centrifuging the sample, and then collecting the supernatant. Immunoglobulins were precipitated from raw sera by 50% (NH 4 ) 2 SO 4 , the pellet was resuspended to 50% of the original volume, and dialyzed overnight against 10 mM Tris-HCl, pH 7.3, 0.9% NaCl. Whole yeast homogenate prepared from a ⌬vph1 mutant strain was prepared as described (6), electrophoresed, and transferred to nitrocellulose. The blot was incu-bated with blocking buffer routinely used for immunoblotting (25). The dialysate and blot were incubated for 1 h to absorb any anti-yeast immunoglobulins with the obvious exception of anti-Vph1p antibodies. Anti-NH 2 terminus Vph1p antibodies were purified using the method described by Manolson et al. (5).
Antibodies-ATPase subunits were detected with one of the following antibodies: 1) a rabbit polyclonal antibody raised against the catalytic 69-kDa V-ATPase A subunit purified from Beta Vulgaris L. (described in Ref. 26); 2) a rabbit polyclonal antibody directed against the regulatory 60-kDa B subunit of murine V-ATPase (described in Ref. 27); 3) a monoclonal antibody directed against the 60-kDa Vma2p subunit (clone 13D11-B2, Molecular Probes); 4) a rabbit polyclonal antibody directed against full-length Vph1p (described in Ref. 5); 5) a rabbit polyclonal antibody directed against the NH 2 terminus of yeast Vph1p, as described above.
Immunofluorescence Microscopy-Slides for immunofluorescence were prepared exactly as described by Manolson et al. (6) and were viewed with a Nikon epifluorescence Diaphot-TMD microscope. Images were photographically recorded on Kodak TMAX film with resulting pictures subsequently digitized. The primary antibodies used were as follows: rabbit affinity-purified anti-Vph1p was diluted 1:50, and rabbit affinity-purified anti-NH 2 terminus Vph1p was diluted 1:25. The secondary antibody used was CY3 at a dilution of 1:1,000.
Assignment of Putative Transmembrane Segments for Vph1p-Currently, 9 primary amino acid sequences, deduced from cDNA, are available for the 95-kDa subunit of V-ATPase; yeast Vph1p (M89778), yeast Stv1p (V06465), rat Vpp1 (M58758), mouse TJ6 (M31226), Neurospora crassa (3929395), Caenorhabditis elegans VPP1 (211115), human (Z711460), human osteoclast (U45285), and bovine (L31770). Because multiple sequence-based secondary structure predictions are substantially more accurate (28), the above sequences were aligned and used to identify putative transmembrane segments. These sequences were submitted to the PHD mail server described by Rost et al. (29). The data obtained from the PHD algorithm, coupled with topological data presented in this paper, were utilized to generate a topological folding model of Vph1p.

Limited Proteolysis of Vacuolar H ϩ -ATPase Demonstrates That ATP⅐Mg Induces Conformational Changes in the Catalytic
Subunit, Vma1p-Physical coupling of the V 1 and V 0 subunits is a requirement for catalysis and proton translocation. Consequently, structural changes within one sector likely impact on the conformation of the other sector. Moreover, dissociation of V 1 from V 0 is enhanced in the presence of ATP⅐Mg, suggesting that substrate binding results in conformational changes within the V 1 sector, allowing for its release. These issues were addressed by a careful analysis of conformational changes induced within subunits of both sectors, as assayed by limited trypsin proteolysis.
Analysis of conformational changes induced within the catalytic and regulatory subunits of the V-ATPase by nucleotide, cofactor, and inhibitor binding was undertaken. Limited tryptic digests of purified vacuolar membranes were performed in the presence of nucleotides (ATP or ADP) and/or magnesium. Proteolytic products were resolved by SDS-PAGE and visualized by immunoblotting with antibodies directed against the catalytic subunit, Vma1p (Fig. 1). As previously noted (16), ATP had no protective effect on the catalytic subunit. Also, neither ADP nor ADP⅐Mg had a protective effect on Vma1p (Fig. 1, panels B and C). In contrast, the rate of trypsinization of Vma1p was retarded in the presence of ATP⅐Mg (Fig. 1, (30). Previous studies have reported that ATP did not alter the pattern or rate of trypsinization of the B subunit (16). Given that ATP⅐Mg renders the catalytic subunit resistant to trypsin proteolysis and that ATP labeling of the B subunit is cationdependent, experiments were undertaken to determine whether the presence of nucleotides and/or cations alter the proteolytic susceptibility of Vma2p.
Partially purified vacuolar membranes were treated with increasing concentrations of trypsin in the absence or presence of ATP. Proteolytic products were resolved by SDS-PAGE and assayed by immunoblotting (Fig. 2, panels A and B) with two different antibodies to identify the maximum number of trypsin-generated peptides. Vma2p was more sensitive than the catalytic subunit to proteolytic degradation under all conditions assayed, requiring the use of much lower trypsin concentrations (0.01 g/ml to 0.25 g/ml). Trypsin degradation of Vma2p was accelerated in the presence of ATP (Fig. 2, panels A and B, lanes 6 -9) and further enhanced with ATP⅐Mg ( Fig. 2, panels A and B, lanes 10 -13). Particularly striking was the generation of a 30-kDa peptide in samples treated with 0.25 g/ml trypsin in the presence of ATP⅐Mg (indicated by line). This peptide was not detected in samples treated with identical trypsin concentrations in the absence of nucleotide and cofactor. The 30-kDa peptide is not the result of a novel cleavage site unmasked by the presence of ATP⅐Mg, because an identical product was detected in samples treated with higher trypsin concentrations (Ͼ0.3 g/ml) in the absence of nucleotide and/or magnesium. This result demonstrates increased proteolytic susceptibility of Vma2p in the presence of ATP⅐Mg. The question remains as to whether these nucleotide-mediated conformational effects are due to direct binding to Vma2p or a consequence of nucleotide binding to Vma1p. Efforts to restrict nucleotide binding (via photolabeling with azido-ATP) to Vma1p to determine whether labeling at the catalytic site was sufficient to render the Vma2p more proteolytically sensitive, proved inconclusive.
Limited Proteolysis of the Vacuolar H ϩ -ATPase Reveals Distinct ATP and Mg 2ϩ -induced Conformational Changes in Vph1p, the 95-kDa Subunit of the V 0 Sector-Incubation of the V-ATPase complex with ATP⅐Mg leads to the dissociation of the V 1 sector from the membrane-bound V 0 sector (18). Preincubation of V-ATPases with N-ethylmaleimide at inhibiting concentrations prevents release of the V 1 sector (11). This observation, coupled with data presented here, suggests that ATP⅐Mg binding induces a dramatic conformational change within Vma1p and Vma2p. This structural change, once transmitted to the remaining complex, allows for dissociation of the two sectors. Thus binding of ATP⅐Mg to the catalytic and regulatory subunits may induce a conformational change(s) in additional subunits within the V-ATPase complex. Of particular interest are conformational changes induced within the V 0 complex, demonstrating a structural interdependence between the two sectors.
Our efforts focused on the 95-kDa subunit (Vph1p) for a number of reasons. Hydropathy analysis of Vph1p predicts a large hydrophilic amino-terminal domain and a hydrophobic C-terminal domain with 6 -8 putative transmembrane segments (5). Although the topology of Vph1p has not been elucidated, proteolytic digests and labeling with impermeant reagents have demonstrated that regions of the protein are cytosolically exposed (31,32). Because of its dual domain structure, the 95-kDa subunit is potentially available to interact with the V 1 and V 0 sectors (5). This hypothesis is supported by the finding that Vph1p is required for the proper assembly of V 0 and the functional association of the two sectors (33). Because Vph1p appears to be required for the structural coupling of V 1 and V 0 complexes, it may be responsive to conformational changes occurring within the peripheral sector. To determine whether this was the case, the following experiments were performed. Under routine purification conditions, the 95-kDa subunit is rapidly cleaved to a ϳ 80-kDa peptide product by endogenous vacuolar proteases. Isolation of vacuoles from pep4 -3 strains, in which proteinase A-dependent proteases are inactive, yields only the full-length 95-kDa protein (5). All vacuoles used for tryptic digests were purified from protease-deficient strains to avoid confounding effects due to endogenous proteases. Tryptic digests of vacuolar membranes in the presence of either ATP or Mg 2ϩ were performed, the trypsinized samples were resolved by SDS-PAGE, and assayed by immunoblotting with anti-Vph1p antibodies. Vph1p appears to undergo conformational changes in response to ATP and Mg 2ϩ . Samples digested in the presence of ATP were degraded less rapidly compared with controls (Fig. 3, ATP). Focusing on samples treated with 2.5 g/ml trypsin (Fig. 3, lanes 7, 13, and  19), it is clear that production of the smallest peptide product (indicated by a line) decreases with inclusion of ATP and, surprisingly, increases when digests are performed in the presence of magnesium. These results suggest that the conformation of Vph1p is responsive to both ATP and magnesium.
To date, no ATP or cation binding sites have been identified within the 95-kDa subunit (5). Ruthenium red binding has been successfully used to identify a number of cation binding proteins (34). No ruthenium red binding was detected for purified Vph1p (data not shown), confirming that Vph1p does not bind divalent cations. Thus, conformational changes induced by divalent cations or ATP must occur through their binding to other proteins within the V-ATPase complex. Clearly, the catalytic and regulatory subunits are likely candidates given that both bind nucleotides and divalent cations and undergo marked conformational perturbations upon binding. The data suggest that conformational changes induced by substrate binding to the V 1 sector are transmitted to Vph1p. In this case, removal of the peripheral sector should render Vph1p insensitive to either nucleotide or divalent cations and the rate of proteolysis should be unaffected by the presence of either compound.
To test this model, membranes were purified from a protease-deficient yeast strain lacking a functional copy of the gene encoding the 60-kDa regulatory subunit of V-ATPase (⌬vma2). In the absence of Vma2p, the remaining peripheral subunits fail to associate with the vacuolar membrane (23,35). V 0 complexes in the ⌬vma2 vacuolar membrane are functionally competent and reassemble with wild type V 1 subunits, as assayed by restoration of ATPase activity (36) demonstrating that the isolated V 0 complex adopts a native structure. Thus, membranes isolated from a ⌬vma2 strain are a suitable source for isolated, intact, and fully assembled V 0 sectors.
Although the isolated V 0 sector was more susceptible to trypsin proteolysis than the intact complex, no difference in the rate of degradation was noted in the presence of ATP or magnesium (Fig. 4). The increased susceptibility to proteolytic deg- radation of the isolated V 0 sector suggests that absence of V 1 renders the membrane sector more accessible to the protease. This enhanced proteolytic sensitivity is consistent with previous reports that noted an increased rate of proteolysis for V 0 sectors reconstituted in the absence of the peripheral sector (37). No novel proteolytic sites were detected, nor were tryptic sites lost in Vph1p. This suggests that the exposed regions of Vph1p do not undergo gross conformational changes in the absence of the V 1 complex. The observation that Vph1p has become conformationally insensitive to ATP and divalent cations suggests that the subunit itself does not bind to either of these compounds. Rather, it is the binding of ATP and divalent cations to the V 1 sector that transmits conformational changes to Vph1p.

Binding of Bafilomycin A 1 , an Inhibitor of V-ATPase Activity, Induces Conformational Changes in the Vma1p Subunit-
Bafilomycin A 1 , a potent microlide inhibitor of vacuolar-type ATPases, binds to the membrane-associated sector (V 0 ) of the complex (3,38,39). Specifically, it appears that the inhibitor binding site is contained within Vph1p. To determine whether bafilomycin A 1 binding has an effect on the structure of Vph1p, partially purified vacuolar membranes were trypsinized in the presence of inhibitor. In addition to bafilomycin A 1 , digests were performed in the presence and absence of ATP and/or magnesium. Vacuolar membranes were preincubated with bafilomycin A 1 for 10 min to ensure that inhibitor binding was not prevented by nucleotides or divalent cations. Samples were trypsinized, as described previously, and the generated peptides resolved by SDS-PAGE and assayed by immunoblotting. The presence of bafilomycin A 1 did not alter the rate of proteolytic degradation of Vph1p (data not shown). This result may indicate that inhibitor binding to Vph1p does not cause a structural change within Vph1p. Alternatively, inhibitor binding may occur within the hydrophobic region of Vph1p and thus any conformational changes would not be detected by the use of antibodies targeted to hydrophilic surface epitopes.
Bafilomycin A 1 alone or in the presence of nucleotide and cofactor rendered Vma1p more susceptible to proteolytic degradation (Fig. 5). This unexpected result indicates that binding of bafilomycin A 1 to Vph1p induces a conformational change in the catalytic subunit of the V-ATPase. Surprisingly, samples trypsinized in the presence of ATP and bafilomycin A 1 were most susceptible to proteolytic degradation (Fig. 5, lanes 5-7). Similar treatment in the presence of ATP⅐Mg was noted to minimize the effect of bafilomycin A 1 on the proteolytic susceptibility of Vma1p (Fig. 5, lanes 8 -10). These data suggest that a reciprocal structural relationship exists between the peripheral and membrane-associated sectors of the V-ATPase. As inhibitor binding to V 0 induces a conformational change in the catalytic subunit of V 1 , cation binding to the peripheral sector induces a structural alteration in the membrane-bound complex.
Localization of the NH 2 Terminus of Vph1p to the Cytosol-Results presented here indicate that Vph1p is responsive to conformational changes that originate within the peripheral sector (ATP⅐Mg binding to V 1 sector) and conversely, that structural changes occurring within the 95-kDa subunit (bafilomycin A 1 binding to Vph1p) can impact on the structure of the V 1 sector. Clearly, the 95-kDa subunit must establish extensive protein-protein interactions to mediate such a degree of intersector communication. Current topological models of Vph1p (40) localize the extensive hydrophilic amino terminus to the vacuolar lumen. If disposed toward the cytosol, this large region of Vph1p would be an ideal candidate to serve as the structural link between the membrane-associated and the peripheral sectors.
The cellular location of the NH 2 terminus of Vph1p was established by the following experiments. An antibody directed against the NH 2 terminus of yeast Vph1p was raised in rabbits and affinity purified as detailed under "Experimental Procedures." Intact purified vacuolar membranes were trypsinized and proteolytic products were assayed by immunoblotting. Proteolytic cleavage at low trypsin concentrations (0.1 g/ml) generated a single peptide product of approximately 90 kDa in size (Fig. 6, panel A), indicating the loss of either the N or C terminus of Vph1p. The failure of an antibody directed against the extreme NH 2 terminus of yeast Vph1p to recognize the 90-kDa product peptide demonstrates that this fragment does not contain this region of the protein (Fig. 6, panel A). This result suggests that the NH 2 terminus of Vph1p is disposed toward the cytosol.
Accurate interpretation of these results requires confirmation of the membrane integrity of the purified vacuoles. Access to the vacuolar lumen would allow for the proteolytic cleavage of an NH 2 terminus disposed toward the luminal space. To confirm the integrity of the vacuolar membrane, purified vacuoles, containing the luminally sequestered pro-form of carboxypeptidase Y (pro-CPY), were subjected to proteinase K digestion in the presence and absence of detergent (Triton X-100). Conversion of the pro-form of the enzyme to the mature form (CPY) was utilized as a marker for access by the protease to the vacuolar lumen. Samples were treated with (30 g/ml) proteinase K for 0, 2, 10, or 20 min. Pro-CPY was only converted to the mature enzyme form in the presence of detergent (Fig. 6, panel B). Incubation of the purified vacuolar membranes with proteinase K for up to 20 min failed to generate cleaved CPY in the absence of detergent. Clearly, purified vacuolar membranes are sealed and able to exclude proteases from the luminal interior.
The cellular location of the NH 2 terminus of Vph1p was confirmed by the use of immunofluorescence. A wild type yeast strain (BJ926) and a ⌬vma1 strain (MM108) containing only assembled V 0 complexes were used. The cells were fixed, spheroplasted, permeabilized, and mounted onto slides as described under "Experimental Procedures." As has been previously reported (41), vacuolar morphology of the wild type and mutant strains are indistinguishable under Nomarski optics (Fig. 6, panel C). The vacuolar membranes of the wild type and ⌬vma1 strain showed comparable staining with an antibody directed against full-length Vph1p, which indicates that there are similar levels of expression and localization between the two yeast strains of the 95-kDa subunit. Differences were seen when the membranes were stained with the antibody directed against the NH 2 terminus of Vph1p. Wild type vacuolar membranes showed minimal staining, whereas the ⌬vma1 strain showed levels of vacuolar staining comparable with that ob-tained with the antibody directed against the full-length protein. The presence of the V 1 subunit inhibits binding of the anti-NH 2 terminus Vph1p antibody. This indicates that the NH 2 terminus is localized to the cytosol and that it interacts with the V 1 to the extent that the peripheral sector can block antibody binding. The topology of the C terminus has not been resolved, yet sequence analysis predicts at least 6 -8 regions of sufficient length and hydrophobicity to function as membranespanning domains (5). To more accurately predict transmembrane segments, multiple sequence-based secondary structure predictions (as described in Ref. 29) were performed on the nine known protein sequences encoding the 95/116-kDa V 0 subunit. The data obtained from the PDH algorithm, coupled with the topological data presented in this paper, were used to generate the topological folding model presented in Fig. 7.

DISCUSSION
The vacuolar ATPase is composed of two structural sectors (V 1 and V 0 ) that perform distinct functions, ATP hydrolysis and proton translocation. The sectors are functionally coupled with complex dissociation rendering both inactive. As demonstrated by this work, this is a dynamic relationship with substrate or inhibitor binding to one sector inducing conformational alterations within both complexes. The study was initiated by examining the impact of substrate binding on subunits (Vma1p and Vma2p) of the catalytic sector.
Nucleotide-induced conformational changes within the catalytic and regulatory subunits of the V-ATPase were monitored by alterations in susceptibility to trypsin proteolysis. Our results concur with previously published data that ATP alone has no protective effect on the catalytic subunit (16). Similarly, the binding of ADP or ADP⅐Mg did not protect Vma1p from proteolytic degradation, suggesting that occupancy of the catalytic site by ADP⅐Mg does not dramatically alter the conformation of Vma1p. Conversely, ATP⅐Mg does induce a conformational change within the catalytic subunit as demonstrated by its The lower panel was probed with a polyclonal antibody directed against the extreme NH 2 terminus of yeast Vph1p. The full-length protein is indicated by a line at the left margin. Panel B, the integrity of the purified vacuolar membranes to exogenously added proteases was assayed by the ability of protease K to cleave pro-CPY (pCPY)to its mature form in the absence and presence of detergent. Pro-CPY and its mature form are normally sequestered in the vacuolar lumen. Intact vacuoles were treated with protease K (final concentration, 30 g/ml) in the presence or absence of 0.5% (w/v) Triton X-100 for 0, 2, 10, and 20 min, respectively. Following trichloroacetic acid precipitation, the protein pellets were denatured and resolved by SDS-PAGE. Both immunoblots shown were probed with a commercially available monoclonal antibody to CPY. The full-length pro-protein (66 kDa) is indicated by a line at the left margin, whereas the processed form of carboxypeptidase Y is indicated by an arrow. Panel C, wild type (WT) and ⌬vma1p, a Vma1p disruption mutant, were grown on complete medium as required. The cells were fixed, spheroplasted, permeabilized, and mounted onto slides as detailed under "Experimental Procedures." Cells were labeled with rabbit anti-fulllength Vph1p (␣-Vph1p) or rabbit anti-NH 2 terminus Vph1p (␣-NH 2 terminus Vph1p) followed by CY3 conjugated goat anti-rabbit IgG. Identical fields are shown as viewed under Nomarski optics (Nomarski) and CY3 fluorescence optics (Fluorescence).
increased resistance to proteolytic degradation. To determine whether ATP hydrolysis, rather than ATP binding, was required to induce the observed conformational change, purified vacuolar membranes were trypsinized in the presence of ATP␥S, a nonhydrolyzable ATP analogue. Trypsinization in the presence of ATP␥S generated a proteolytic pattern indistinguishable from that obtained with hydrolyzable nucleotide (data not shown) under all conditions assayed (with or without Mg 2ϩ ). Although the experimental conditions used allow for single site hydrolysis, it is clear that multisite catalysis is not required to induce a conformational change in the catalytic subunit.
The proteolytic susceptibility of Vma2p was increased in the presence of ATP and further enhanced in the presence of ATP⅐Mg. Clearly, substrate binding dramatically alters the conformation of the regulatory subunit. It remains unclear if this effect is due to ATP⅐Mg binding at the catalytic site, although this suggestion is indirectly supported by several lines of evidence. High resolution structural analysis of the F-ATPase peripheral sector suggests that Vma1p and Vma2p come together to form a catalytic site that resides primarily on Vma1p with some residues contributed by Vma2p (14,42). Because the catalytic site is essentially shared by the two subunits, it is likely that binding of ATP⅐Mg alters both conformations. This presumption is supported by studies of in vitro reconstitution of the core V 1 sector (subunits Vma1p, Vma2p, Vma4p, Vma7p, Vma8p, and Vma10p) showing that complex assembly is nucleated by the association of Vma1p and Vma2p with ATP⅐Mg stabilizing this critical interaction (2). These data suggest that binding of ATP⅐Mg likely induces a conformational change within Vma1p and/or Vma2p, providing a stable core for the formation of higher order complexes of the peripheral sector.
Compelling evidence indicates that ATP⅐Mg binding to the catalytic subunit induces a global conformational change within the peripheral sector, resulting in the dissociation of the soluble sector from the membrane-associated complex. It has been proposed that "chaotropic" anions mediate H ϩ -ATPase inhibition by oxidizing critical sulfhydryl residues, which in turn leads to the release of the peripheral sector. Indeed, oxi-dation of these sites in the Neurospora V-ATPase is associated with the rapid dissociation of the V 1 peripheral sector (43). In our hands, although chaotropic agents (KI) enhanced removal of the peripheral sector, ATP⅐Mg alone was sufficient to promote stripping of the V 1 sector (data not shown). This ATP⅐Mginduced peripheral sector dissociation is inhibited by N-ethylmaleimide modification of Vma1p. Clearly, substrate binding induces a global conformational change within the peripheral sector, allowing its release from the membrane-associated complex. Dissociation was neither inhibited or reduced in the presence of bafilomycin A 1 or DCCD, potent inhibitors of V-ATPase activity, suggesting that multisite hydrolysis is not required to release the soluble V 1 sector.
Of greater significance is the structural impact of substrate binding (ATP and/or divalent cations) on the conformation of components of the membrane-associated sector. Modification of the peripheral sector by ATP or divalent cations resulted in a conformational change within Vph1p, indicating that structural changes within the V 1 sector impact on V 0 components. There was a dramatic increase in the rate of Vph1p proteolysis in the presence of divalent cations (Mg 2ϩ , Ca 2ϩ , Mn 2ϩ ) supporting the conclusion that a degree of intersector communication exists within V-ATPases. Particularly intriguing is the importance of divalent cations in maintaining the quaternary structure of the peripheral domain. Reconstitution experiments involving the isolated subunits of the hydrophilic sector demonstrated that assembly of the core was enhanced in the presence of divalent cations (2). Mg 2ϩ , Ca 2ϩ , and Mn 2ϩ were found to be equally effective mirroring the pattern noted to increase the proteolytic susceptibility of Vph1p. These results suggest that cation binding induces a global conformational change that promotes formation of a stable peripheral complex. Moreover, this cation-induced conformational change is transmitted to the membrane-associated sector, as evidenced by the increased proteolytic susceptibility of Vph1p. The presence of cations did not enhance the rate of Vph1p proteolysis within isolated V 0 sectors, providing compelling evidence that structural changes within Vph1p are mediated via intersector communication versus direct cation binding to the membrane-associated complex.
Data presented in this paper suggest that intersector communication is bidirectional, with conformational changes initiated within the membrane-associated sector reflected in a structural perturbation within the peripheral sector and vice versa. The V-ATPase inhibitor, bafilomycin A 1 , is known to bind to the V 0 complex, with evidence suggesting that its binding site resides within Vph1p (3). Addition of bafilomycin A 1 accelerated the rate of proteolysis of the catalytic subunit indicating that inhibitor binding to the hydrophobic sector results in a conformational change within Vma1p. Chromaffin granule V-ATPases show simple Michaelis-Menten kinetics with three apparent K m values (38). Addition of bafilomycin A 1 impairs ATP hydrolysis and catalytic cooperativity, although hydrolysis at a single catalytic site remains resistant to the inhibitor (38). This suggests that binding of bafilomycin A 1 to the hydrophobic sector impacts on the structure and function of the hydrophilic sector, particularly the catalytic subunit. A similar transmission of an inhibitory signal from the hydrophobic to hydrophilic sector has been observed in the inhibition of F 0 F 1 ATPase by oligomycin and dicyclohexylcarbodiimide (44). Thus, intersector communication is reciprocal with conformational changes initiated within one sector reflected in compensatory structural changes within the opposing complex.
The specific structural link(s) that mediates this intersector communication remains unknown. Our work provides evidence that Vph1p might play a central role in coupling the two sectors of the V-ATPase with dynamic structural information readily Hatched TM segments were difficult to assign with certainty. In particular, TM7 and TM8 represent a long hydrophobic stretch (ϳ30 residues) that could exist as a single TM domain or as two shorter TM segments. transmitted between the V 0 and V 1 sectors. Because of its bipartite structure, Vph1p is an ideal candidate to serve as a functional and structural link between the catalytic and proton-translocating sectors of the V-ATPase. We have unambiguously localized the ϳ50-kDa hydrophilic NH 2 terminus of Vph1p to the cytosol, where it is available to interact with the peripheral sector. Because of its localization to the vacuolar membrane, the C-terminal domain of Vph1p is in close proximity to the proton-translocating core of the V-ATPase, thus allowing for catalytically induced conformational changes within the peripheral sector to impact on the functioning of the ion channel. This model is supported by previous work demonstrating that mutagenesis of key residues within the membrane-associated domain of Vph1p inhibits both proton-translocation and ATPase activity (40,45). This data indicates that the C terminus of Vph1p is essential for H ϩ translocation and that changes to this region impacts on the function of the catalytic sector. Thus, structural changes in the membrane domain of Vph1p are capable of transmitting an inhibitory signal to the catalytic sector, supporting that Vph1p mediates intersector communication.
Recent work on the structurally related F-type ATPases has revealed a possible structural link between the F 1 and F 0 sectors (8). The 2.8 Å resolution structure of the soluble F 1 complex shows that a central stalk connects the soluble and membrane-associated components of the F-type ATPase. The ␤and ␣-subunits are arranged in an alternating hexagonal array with the central pore of the structure occupied by two long ␣ helices contributed by the ␥-subunit. ATP hydrolysis drives the physical rotation of the central ␥ subunit (46,47) that, in turn, drives proton translocation. Efficient transduction of the torque exerted by the rotating stalk on F 0 requires the presence of a stator (48,49), which prevents loss of energy through the rotation of F 1 relative to F 0 . Subunits a, (b) 2 , and ␥ are likely candidates to fulfill this role in the F-type ATPases (50). To date, no corresponding subunits have been identified in the V-type ATPases, although a similar functional mechanism has been hypothesized. Because of its unique bipartite structure and its responsiveness to substrate binding, we propose that Vph1p is ideally suited to serve as the structural and functional link between the two sectors of the V-ATPase and suggest that it is part of the V-ATPase stator. Currently, work is underway to identify regions of interaction between the hydrophilic sector of Vph1p and components of the catalytic sector.