Subunit Interactions in the Clathrin-coated Vesicle Vacuolar (H+)-ATPase Complex*

The vacuolar (H+)-ATPases (or V-ATPases) are structurally related to the F1F0ATP synthases of mitochondria, chloroplasts and bacteria, being composed of a peripheral (V1) and an integral (V0) domain. To further investigate the arrangement of subunits in the V-ATPase complex, covalent cross-linking has been carried out on the V-ATPase from clathrin-coated vesicles using three different cross-linking reagents. Cross-linked products were identified by molecular weight and by Western blot analysis using polyclonal antibodies raised against individual V-ATPase subunits. In the intact V1V0 complex, evidence for cross-linking of subunits C and E, D and F, as well as E and G by disuccinimidyl glutarate was obtained, while in the free V1 domain, cross-linking of subunits H and E was also observed. Subunits C and E as well as D and E could be cross-linked by 1-ethyl-3-(dimethylaminopropyl)carbodiimide, while subunits a and E could be cross-linked by 4-(N-maleimido)benzophenone. It was further demonstrated that it is possible to treat the V-ATPase with potassium iodide and MgATP in such a way that while subunits A, B, and H are nearly quantitatively removed, significant amounts of subunits C, D, E, and F remain attached to the membrane, suggesting that one or more of these latter subunits are in contact with the V0domain. In addition, treatment of the V-ATPase with cystine, which modifies Cys-254 of the catalytic A subunit, results in dissociation of subunit H, suggesting communication between the catalytic nucleotide binding site and subunit H. Finally, the stoichiometry of subunits F, G, and H were determined by quantitative amino acid analysis. Based on these and previous observations, a new structural model of the V-ATPase from clathrin-coated vesicles is proposed.

The V-ATPase is a heteroligomeric complex of molecular mass approximately 800 kDa composed of at least 13 different subunits arranged into two separate domains (1)(2)(3)(4)(5)(6). The peripheral V 1 19 (cЉ), and 17 (c, cЈ) kDa. Functional studies indicate that the V 1 domain is responsible for ATP hydrolysis while the V 0 domain carries out proton transport (1)(2)(3)(4)(5)(6). We have previously demonstrated that each V-ATPase complex contains three copies each of the A and B subunits, six copies of the 17-kDa subunits (c, cЈ), and single copies of the C, D, E, a, d, and cЉ subunits (10). The number of copies of subunits F, G, and H per complex has not been reliably determined.
The V-ATPases are known to be structurally and evolutionarily related to the ATP synthases (or F-ATPases) of mitochondria, chloroplasts and bacteria (11)(12)(13). Thus the nucleotide binding subunits of the V-ATPase (A and B) are homologous to the corresponding ␤ and ␣ subunits of F 1 (14,15), while the proteolipid subunits of the two complexes are also homologous (16,17). The structure of the peripheral F 1 domain of the mitochondrial F-ATPase has been determined by x-ray crystallography and shown to consist of a hexamer of alternating ␣ and ␤ subunits surrounding a central cavity containing the highly ␣-helical ␥ subunit (18,19). F 1 appears to be attached to the F 0 domain via both a central stalk (believed to include both the ␥ and ⑀ subunits) and a peripheral stalk (composed of the ␦ subunit and the soluble portions of subunit b) (20,21). The F 0 domain contains a ring of c subunits with the a and b subunits to one side (12,23).
While the V-ATPase complex is thought to resemble the F-ATPases, electron micrographs have revealed significant differences (24,25), and very little is known concerning the arrangement of subunits in the V-ATPase complex. In the current study, we have employed covalent cross-linking and Western blot analysis to identify subunit contacts within the V-ATPase. In addition, we have reevaluated the effects of treatment of the enzyme with cystine or potassium iodide on dissociation of specific subunits from the complex. Finally, we have determined the stoichiometry of subunits F, G, and H by quantitative amino acid analysis. These studies have allowed us to construct a new model for the structure of the V-ATPases. was from KPL Laboratories and prestained SDS-PAGE marker proteins were from Amersham Pharmacia Biotech.
Antibody Production and Purification-The bovine cDNAs encoding the full-length forms of subunits C, D, and E were cloned into the pET-21d(ϩ) expression vector and expressed in BL-21 cells. Expression of the His-tagged proteins was induced by incubation in 1 mM isopropyl-1-thio-␤-D-galactopyranoside, and disruption of cells was carried out using lysozyme and Triton X-100 as described previously (26). The recombinant proteins were recovered from inclusion bodies by sedimentation and solubilized with 8 M urea followed by dialysis against phosphate-buffered saline. The His-tagged proteins were then isolated by nickel-affinity chromatography and sent to Covance Inc. for production of polyclonal antibodies in rabbits. For subunits B, F, and H, the following synthetic peptides were prepared: subunit B, CPTSGPLAG-SREQAL; subunit F, CPSEKHPYDAAKDSILRR; subunit H, CHKSEK-FNRENPARLNEKN. The peptides were conjugated to keyhole limpet hemocyanin via the cysteine residues at the amino termini and were also sent to Covance, Inc. for production of polyclonal antibodies in rabbits. Antibodies were isolated from rabbit antisera using protein A-Sepharose affinity chromatography.
Preparation of Clathrin-coated Vesicles and Purification of the V-ATPase-Clathrin-coated vesicles were prepared from calf brain as described previously (27). Vesicles were stripped of their clathrin coat using 0.5 M Tris (pH 7.0), 2 mM EDTA and the V-ATPase was solubilized with C 12 E 9 and purified by glycerol density gradient centrifugation as described (27).
Preparation of the V 1 (-Subunit C) Subcomplex-Stripped vesicles (1.0 mg of protein/ml) were treated first with 0.2 M KI for 1 h at 4°C, followed by sedimentation for 1 h at 100,000 ϫ g and resuspension in 0.36 M KI and 2.5 mM MgATP, incubation for 1 h at 4°C, and sedimentation. The supernatant was dialyzed overnight versus two changes of 100 volumes of solubilization buffer (50 mM NaCl, 30 mM KCl, 20 mM HEPES (pH 7.0), 0.2 mM EGTA, 10% glycerol) to allow reassembly of the V 1 (-subunit C) subcomplex (referred to as the V 1 subcomplex). The V 1 subcomplex was then isolated by sedimentation for 16 h at 175,000 ϫ g on 15-30% glycerol density gradients as described previously (28).
Covalent Cross-linking of the Purified V-ATPase and V 1 Subcomplex-For DSG, the purified V-ATPase or V 1 subcomplex were concentrated to approximately 50 g of protein/ml using a Centricon-30 microconcentrator and incubated with 0.2 mM DSG for 30 min at 23°C. The reaction was stopped by addition of 50 mM ammonium acetate (pH 8.0). For EDC, the purified V-ATPase or V 1 subcomplex (50 g of protein/ml) was incubated with 5 mM EDC together with 5 mM NHS for 20 min at 23°C. The reaction was stopped by addition of 50 mM Tris-HCl (pH 6.8). For MBP, the reagent dissolved in dimethylformamide at 100 mM was added to the purified V-ATPase or V 1 subcomplex at a final concentration of 1 mM, followed by incubation in the dark for 30 min at 23°C. The unreacted MBP was quenched by addition of 10 mM dithiothreitol. The samples were dialyzed against two changes of 200 volumes of solubilization buffer for a total of 4 h, followed by irradiation with a long wavelength ultraviolet lamp for 20 min at 4°C.
Trypsin Treatment of the Purified V-ATPase-Where indicated, the purified V-ATPase was concentrated to 50 g of protein/ml and treated with 1 g of trypsin/ml for 2 h at 23°C. Proteolysis was stopped by addition of 5-fold (w/w) soybean trypsin inhibitor.
Electrophoresis and Immunoblotting-For analysis of cross-linked products, SDS-PAGE was performed using 8% or 15% acrylamide gels according to the method of Laemmli (29). After electrophoresis, the samples were electrophoretically transferred to nitrocellulose membranes for 16 h at 100 mA. The blots were then cut into strips, which were used in Western blot analysis as described previously (30). Protein A-Sepharose-purified antibodies were used at a dilution of 1:1000, and blots were developed using a chemiluminescent detection method from Kirkegaard and Perry Laboratories.
Determination of Subunit Stoichiometry-The stoichiometry of subunits F, G, and H and the 45-kDa glycoprotein were determined by quantitative amino acid analysis as described previously (10). The purified V-ATPase (10 -50 g of protein) was concentrated using a Centricon-30 microconcentrator and DEAE-Sepharose chromatography followed by dialysis against 0.02% C 12 E 9 and lyophilization. The lyophilized enzyme was then dissolved in 100 l of Laemmli sample buffer. For subunit H and the 45-kDa glycoprotein, samples were applied to a 10% polyacrylamide gel and separated by the method of Laemmli (29). For subunits F and G, samples were applied to a 16.5%/10% discontinuous acrylamide gel prepared and run according to the method of Schä gger and von Jagow (31). This Schä gger gel system was employed for subunits F and G because it provided greater resolution of lower molecular weight polypeptides. Following electrophoresis, proteins were electrophoretically transferred to Immobilon for 4.5 h at 380 mA, followed by staining with Coomassie Blue and destaining of the membrane as described previously (10). Quantitation of subunit H and the 45-kDa glycoprotein was done relative to subunit A, which we previously showed was present in three copies per complex, while quantitation of subunits F and G was done relative to the 40-kDa C subunit, which is present in one copy per complex (10). In all cases, the bands were excised and subjected to quantitative amino acid analysis using a Waters Pico-tag work station and high performance liquid chromatography. Transfer efficiency was determined by quantitation of the amount of residual protein remaining in the acrylamide gel for each subunit relative to known standards, and agreed well with the values obtained using 125 I-Bolton-Hunter-labeled protein described previously (10).

Covalent Cross-linking of the Coated Vesicle V-ATPase-To
facilitate identification of products following covalent crosslinking of the purified V-ATPase, polyclonal antibodies were prepared in rabbits against either full-length recombinant proteins expressed in Escherichia coli (subunits C, D, and E) or 15-19-amino acid synthetic peptides conjugated to keyhole limpet hemocyanin (subunits B, F, G, and H). Antibodies were purified by protein A-Sepharose chromatography and tested for specificity by Western blot analysis against the purified coated vesicle V-ATPase. As can be seen from Fig. 1, all antibodies showed specific recognition of the corresponding subunit by Western blot with the exception of the antisera raised against subunit G, which did not show significant reaction (data not shown). The antisera against subunit H recognized two bands of molecular mass 50 -57 kDa, consistent with previous reports on the existence of two isoforms of subunit H in the coated vesicle enzyme (34).
The first cross-linking reagent employed was DSG, a bifunctional amino-reactive reagent containing a 7.7-Å linker arm. Cross-linking by DSG was tested in both the intact V 1 V 0 complex and the free V 1 domain isolated by glycerol density gradient sedimentation following dissociation with potassium iodide and MgATP (28). As can be seen in Fig. 2, a cross-linked product of molecular mass 50 kDa showed cross-reaction with antibodies against both subunits D and F and was observed in both the intact V 1 V 0 complex and the free V 1 domain. The migration of this product agrees well with the predicted molecular mass of a complex containing one copy each of subunits D and F (48 kDa). While a band at 50 kDa could also be observed FIG. 1. Analysis of specificity of rabbit polyclonal antibodies raised against V-ATPase subunits by Western blot. The purified V-ATPase (2 g/lane) was applied to either a 15% (for subunit F) or 8% (the remaining subunits) polyacrylamide gel, followed by electrophoretic transfer to nitrocellulose. The nitrocellulose was cut into strips, and Western blotting was carried out using protein A-Sepharose purified antibodies (1:1000 dilution) prepared against the individual subunits indicated as described under "Experimental Procedures." The numbers shown represent the molecular mass of the individual subunits in kDa.
in the V 1 lane (and more faintly in the V 1 V 0 lane) with antibodies against subunit E, experiments described below suggest that this band does not represent cross-linking of subunits E and F. However, a 46-kDa cross-linked product that reacted with the anti-E subunit antibody but did not react with the anti-F subunit antibody was also observed in both V 1 V 0 and V 1 , and likely represents cross-linking of subunits E and G (predicted molecular mass 46 kDa).
In addition to the above complexes, cross-linking by DSG gave rise to a 90-kDa product in V 1 that reacted with antibodies against subunits H and E (predicted molecular mass 87-90 kDa) and a 75-kDa product in V 1 V 0 that was recognized by antibodies against subunits C and E (predicted molecular mass 73 kDa). Finally, a doublet around 70 kDa was observed in the V 1 lane with both the anti-H and anti-F antibodies (predicted molecular mass 68 -71 kDa). It should be noted that treatment with DSG appears to cause some internal cross-linking of subunit E, such that a second (or even a third) band appears with slightly lower mobility. This may account for the additional band observed at 50 kDa with the anti-E subunit antibody.
To further test the identity of the observed cross-linked products, the V 1 V 0 complex was first treated with trypsin prior to cross-linking with DSG. Under these conditions, subunits D, F, and H were largely digested. As can be seen in Fig. 3, the product at 75 kDa recognized by the antibodies against subunits C and E and the product at 46 kDa recognized by the anti-E subunit antibody were unaffected, confirming that these products do not include subunits D, F, or H.
The second cross-linking reagent employed was EDC, which, following treatment with NHS, results in cross-linking of amino and carboxyl groups. As can be seen in Fig. 4, treatment of the V 1 V 0 complex with EDC/NHS gave rise to a pair of products around 75 kDa that reacted with both the anti-C and anti-E subunit antibodies (similar to that observed for DSG) and a new product of molecular mass 67 kDa that was recognized by both the anti-D and anti-E subunit antibodies. The predicted molecular mass of a complex between subunits D and E is 67 kDa.
The final cross-linking reagent tested was MBP, which is a photoactivatible maleimide that has a linker arm of 10 Å (Fig.  5). Reaction with MBP in the dark followed by irradiation with ultraviolet light gave rise to a 135-kDa product that was recognized by the antibody against subunit E. The size of this complex is consistent with cross-linking of subunit E and subunit a of the V 0 domain. We have recently shown that subunit a contains a large soluble domain at the amino terminus, which is exposed on the cytoplasmic side of the membrane and is therefore situated to interact with the V 1 domain (35).
Dissociation of V 1 Subunits by Potassium Iodide and MgATP-We have previously demonstrated that treatment of clathrin-coated vesicles with potassium iodide and MgATP results in the release of the V 1 subunits from the membrane (36,37). Moreover, by Western blot analysis, the A subunit is released nearly quantitatively (38). An important question, how-FIG. 5. Cross-linking of the purified V-ATPase and V 1 subcomplex using MBP. The purified V-ATPase and V 1 subcomplex (50 g of protein/ml) were incubated with 1 mM MBP for 30 min at 23°C. The excess MBP was quenched by addition of 10 mM dithiothreitol, and the samples were then dialyzed against solubilization buffer for 4 h prior to irradiation with a long wavelength ultraviolet lamp for 20 min at 4°C. The samples were applied to an 8% acrylamide gel and subjected to SDS-PAGE according to the method of Laemmli (29). Proteins were then transferred to nitrocellulose and Western blotting using the antibodies against the indicated subunits carried out as described under "Experimental Procedures." A cross-linked product of molecular mass 135 kDa that cross-reacts with the antibody directed against subunit E was observed that has a size consistent with cross-linking of subunits a and E.
FIG. 2. Cross-linking of the purified V-ATPase and V 1 subcomplex using DSG. The purified V-ATPase and V 1 subcomplex (50 g of protein/ml) in solubilization buffer were incubated with 0.2 mM DSG for 30 min at 23°C. The reaction was stopped by addition of 50 mM ammonium acetate, and the cross-linked products were separated on an 8% acrylamide gel according to the method of Laemmli (29). The proteins were transferred to nitrocellulose and the nitrocellulose cut into strips and the strips subjected to Western blot analysis using protein A-Sepharose purified antibodies against the indicated subunits. Blots were developed using the chemiluminescence system described under "Experimental Procedures." The molecular mass of particular crosslinked products is indicated by the bars next to the individual lanes and was determined relative to standards of known molecular mass. The 90-kDa product in V 1 is observed to contain both subunits H and E, the 75-kDa product in V 1 V 0 contains subunits C and E, the 70-kDa product in V 1 contains subunits H and F, the 50-kDa product in both V 1 V 0 and V 1 contains subunits D and F and the 46-kDa product observed in both V 1 V 0 and V 1 contains, most likely, subunits E and G. FIG. 3. DSG cross-linking of trypsin-treated V-ATPase. The purified V-ATPase (50 g of protein/ml) was digested by incubation with trypsin (1 g/ml) for 2 h at 23°C, and proteolysis was then stopped by addition of 5-fold (w/w) soybean trypsin inhibitor. Cross-linking of the trypsin-treated V-ATPase by DSG, gel electrophoresis, and Western blot analysis were then carried out as described in the legend to Fig. 2. Both subunits D and F were extensively digested under these conditions, whereas subunits C and E as well as the 75-kDa (C/E) and 46-kDa (E/G) cross-linked products were unaffected.

FIG. 4. Cross-linking of the purified V-ATPase by EDC.
The purified V-ATPase (50 g of protein/ml) was incubated with 5 mM EDC and NHS for 20 min at 23°C. The reaction was stopped by addition of 50 mM Tris-HCl (pH 6.8), and the samples were applied to an 8% acrylamide gel and subjected to SDS-PAGE according to the method of Laemmli (29). Transfer to nitrocellulose and Western blotting using the indicated antibodies was carried out as described under "Experimental Procedures." Both a 75-kDa product containing subunits C and E and a 67-kDa product containing subunits D and E were observed. ever, is whether all of the V 1 subunits are equally well released from the membrane. To address this question, Western blot analysis was carried out on stripped coated vesicles either before or after treatment with potassium iodide and MgATP using the antibodies against subunits B, C, D, E, F, and H described above. As can be seen in Fig. 6, treatment with KI and MgATP results in nearly complete release of subunits B and H (as was previously observed for subunit A; Ref. 38), whereas significant amounts of subunits C, D, E, and F remain attached to the membrane. While it is likely that a larger fraction of subunits C, D, and E were removed at the lower protein concentration previously employed in the preparation of V 0 -containing membranes (28), these results indicate that it is possible to dissociate subunits A, B, and H while leaving behind substantial amounts of subunits C, D, E, and F.
Release of Subunit H by Treatment with Cystine-We have previously observed that treatment of the V-ATPase with cystine results in the release of the 50-kDa subunit of the AP2 adaptin complex that is associated with the enzyme and causes loss of V-ATPase activity (39). It has subsequently been reported that an active V-ATPase completely lacking the AP50 polypeptide could be isolated, but that removal of subunit H by urea treatment resulted in loss of ATPase activity (34). We therefore wished to determine whether cystine treatment of the V-ATPase also led to loss of subunit H, which on SDS-PAGE has a molecular mass similar to AP50. Stripped vesicles were treated with 1 mM cystine at 4°C as described previously and the V-ATPase then solubilized with C 12 E 9 , purified by density gradient sedimentation and subjected to SDS-PAGE on a 10% acrylamide gel and Western blot analysis using the anti-subunit H antiserum as described under "Experimental Procedures." As shown in Fig. 7, cystine treatment results in almost complete loss of the lower cross-reactive band as well as some loss of the upper cross-reactive band. Under the conditions employed, cystine treatment led to loss of approximately 93% of V-ATPase activity. Thus, cystine treatment of the V-ATPase leads to loss of at least one of the two isoforms of subunit H.
Stoichiometry of Subunits F, G, and H-We previously demonstrated using quantitative amino acid analysis that the V-ATPase complex contains three copies each of the A and B subunits, six copies of subunit c, and single copies of subunits C, D, E, a, d, and cЉ (10). At the time, subunits F, G, and H had not been identified. We therefore wished to determine the copy number of these additional V-ATPase subunits. To avoid inter-ference from the 50-kDa subunit of the AP2 adaptin complex, the V-ATPase was isolated from coated vesicles that had been stripped of their clathrin using 0.5 M Tris (pH 7.0), under which conditions the V-ATPase can be purified free of the AP50 polypeptide (data not shown). To measure the stoichiometry of subunit H, the purified V-ATPase was separated on a 10% acrylamide gel according to the method of Laemmli (29), while, for measurement of the copy number of subunits F and G, the purified enzyme was separated on a 16.5/10% Schä gger gel (31), which provides greater resolution of lower molecular mass polypeptides. Following gel electrophoresis, the proteins were transferred to Immobilon for quantitative amino acid analysis, with the efficiency of transfer measured as described under "Experimental Procedures." The copy number of subunit H is expressed relative to subunit A (which had previously been shown to be present in three copies per complex; Ref. 10), while subunits F and G are expressed relative to subunit C (which is present in a single copy per complex; Ref. 10). As can be seen from the data in Table I, there is one copy of subunit H as well as a single copy of subunit F per V-ATPase complex, while subunit G is present in two copies per complex. While the upper isoform of subunit H could not be adequately resolved from subunit B for its stoichiometry to be determined independently, the fact that equal amounts of these two isoforms were observed in Western blot analysis (Fig. 7) suggests that each V-ATPase complex contains one copy of each isoform of subunit H. Also shown are the data obtained for a 45-kDa glycoprotein, which copurifies with the V-ATPase and which had previously been cloned for the chromaffin granule V-ATPase (40). Unlike the other subunits, this polypeptide is present in only about 0.5 copies per complex, and thus appears to be present in only a subpopulation of the V-ATPases isolated from clathrin-coated vesicles. Table II summarizes the results of the cross-linking experiments described in this paper. As can be seen, both DSG and EDC were able to cross-link subunits C and E in the intact V 1 V 0 complex. This result is consistent with our previous observation that subunits C and E could be coimmunoprecipitated from a mixture of dissociated V 1 subunits using a monoclonal antibody directed against the E subunit (28), and extends this finding by demonstrating the proximity of these subunits in the intact complex. This contact is of particular interest since subunit C does not tightly associate with the V 1 complex, as evidenced by its absence from a V 1 complex reassembled in vitro (28) and by its absence from various V 1 subcomplexes observed in vivo (41,42). Thus subunit E may represent a principal contact of subunit C with the remainder of the V 1 domain.

DISCUSSION
Cross-linking by DSG also revealed contacts between subunit D and F as well as a likely complex between subunits E and G. The latter assignment is tentative since the presence of subunit G in the cross-linked product could not be demonstrated directly due to the unavailability of an anti-G subunit antibody. However, the molecular mass of the cross-linked product (46 kDa) together with the fact that it does not react with the anti-F subunit antibody makes it likely that it does reflect a contact between subunits E and G. These observations are consistent with previous findings that a subcomplex containing at least subunits E and G as well as a subcomplex containing at least subunits D and F could be isolated from a variety of yeast strains deleted in various vma genes (42,43), and demonstrate that these contacts are preserved in both the V 1 V 0 and V 1 complexes.
Information was also obtained in these studies about subunits in contact with subunit H. Subunit H can be cross-linked FIG. 6. Dissociation of V 1 subunits from stripped vesicles by treatment with potassium iodide and MgATP. Stripped vesicles (1 mg of protein/ml) that had been pretreated with 0.2 M KI for 1 h at 4°C and washed by sedimentation at 100,000 ϫ g were then treated with 0.36 M KI and 2.5 mM MgATP for 1 h at 4°C, followed by sedimentation for 1 h at 100,000 ϫ g. The vesicles were resuspended in solubilization buffer and dialyzed overnight against the same buffer. Control vesicles were subjected to the same treatment except in the absence of KI and MgATP during the second incubation. The membranes were solubilized in Laemmli sample buffer, and 10 g of protein was applied to each lane of a 15% acrylamide gel followed by SDS-PAGE according to the method of Laemmli (29). Following transfer to nitrocellulose, Western blotting using antibodies against the indicated subunits was carried out as described under "Experimental Procedures." It should be noted that the protein concentration employed during the dissociation (1 mg of protein/ml) was higher than that used previously in the preparation of V 0 (28) in order to determine whether it is possible to observe a differential release of subunits A, B, and H versus subunits C, D, E, and F. by DSG to both subunits E and F, although these products are observed only in the isolated V 1 domain, suggesting that these contacts may be shielded in the intact V-ATPase. Alternatively, the V 1 complex may distort relative to V 1 V 0 to allow some subunit contacts to occur that do not normally occur in the intact complex. Because V-ATPase complexes lacking subunit H can be isolated both in vitro (39,34) and in vivo in a yeast strain lacking the corresponding gene product (Vma13p) (44), it is likely that subunit H is located near the periphery of the complex. In fact, the VMA13 gene product is unique in being the only V-ATPase subunit whose absence does not disrupt assembly of the V-ATPase.
The final information obtained from the cross-linking studies presented concerns subunit E. Because EDC/NHS is a zerolength cross-linking reagent, the cross-linking of subunits D and E suggests that these subunits are, at least part of the time, in close contact with each other. A contact between subunit E and the 100-kDa subunit a of the V 0 domain is also suggested by the appearance of a product of the appropriate size on cross-linking of the intact complex (but not the V 1 domain) with MBP. While this assignment is again tentative pending the availability of an antibody that recognizes the a subunit by Western blot, it suggests that subunit E may play an important role in bridging the peripheral and integral domains of the protein.
We have suggested, based on mutagenesis studies (30,45), that the 100-kDa subunit corresponds to the V-ATPase homolog of the F-ATPase a subunit, which functions not only in proton translocation (12,46,47) but as part of the stator that is held rigid relative to the ␣ 3 ␤ 3 hexamer (21). Most recently we have demonstrated that the large amino-terminal domain of the 100-kDa subunit is oriented toward the cytoplasmic side of the membrane (35), making it a likely candidate for forming part of the stator structure in the V-ATPases.
We had previously suggested, based on cross-linking by the reversible cross-linking reagent DTSSP and analysis by twodimensional gel electrophoresis that subunits C, D, and E were in contact with subunit c of the V 0 domain (36). Because of the similarity in molecular mass of subunits c, F, and G, the results in the current study suggest the possibility that the products observed previously actually represented complexes between subunits C, D, or E and subunits F or G. Nevertheless, the observation that treatment of stripped vesicles with potassium iodide and MgATP can result in nearly complete removal of subunits B and H (Fig. 6) as well as subunit A (38) while leaving behind substantial amounts of subunits C, D, E, and F suggests that one or more of these latter subunits likely form part of the bridge between the V 1 and V 0 domains.
We had previously reported that treatment of the V-ATPase with cystine results in dissociation of AP50 (the 50-kDa subunit of the AP2 adaptin complex) and loss of activity (39). AP50 was identified by NH 2 -terminal sequence analysis, and the FIG. 7. Cystine treatment causes removal of subunit H from the V-ATPase complex. Stripped vesicles (0.1 mg of protein/ml) were dialyzed against 1 mM cystine, 20 mM HEPES (pH 7.5), 0.2 mM EGTA for 5 days at 4°C followed by sedimentation for 1 h at 100,000 ϫ g and resuspension of vesicles in solubilization buffer. Control vesicles were incubated under identical conditions but in the absence of cystine. The V-ATPase was then solubilized with C 12 E 9 and purified by density gradient sedimentation as described under "Experimental Procedures." Equal amounts of the purified V-ATPase (2 g of protein) from control and cystine-treated stripped vesicles were applied to an 8% acrylamide gel and subjected to SDS-PAGE as described by Laemmli (29). The proteins were transferred to nitrocellulose, and Western blotting was carried out using the antibody raised against subunit H as described under "Experimental Procedures."   45-kDa glycoprotein in the V-ATPase complex Stoichiometry of subunits H, F, and G and the 45-kDa glycoprotein were determined by quantitative amino acid analysis following SDS-PAGE and electrophoretic transfer to Immobilon as described under "Experimental Procedures." The stoichiometries of subunits H and GP45 were determined following separation on a 10% Laemmli gel (29) and are expressed relative to subunit A, which was previously shown to be present in three copies per complex (10). The stoichiometries of subunits F and G were determined following separation on a 16.5%/10% Schaegger gel (31), and are expressed relative to subunit C, which is present in one copy per complex (10). The data shown are for two independent trials. The blotting efficiency was determined as described under "Experimental Procedures" and used to calculate the total moles of amino acids in each band on the original gel. Moles of protein were calculated from corrected moles of amino acids based upon the number of amino acids in each polypeptide (which is known from the published amino acid sequences). The ratios shown in the last column are an average of the two independent determinations and are expressed relative to a value of 3.0 for subunit A (for subunit H and GP45) and 1.0 for subunit C (for subunits F and G). number of copies of the protein migrating at this position was measured using quantitative amino acid analysis (48). We have subsequently observed that only about 10 -15% of the protein at this position actually corresponds to AP50, the remainder corresponding to the lower molecular weight form of subunit H, which is not detected by NH 2 -terminal sequencing because it is blocked at the amino terminus. We therefore wished to determine whether treatment of the V-ATPase with cystine also resulted in the loss of subunit H. As shown in Fig. 7, cystine treatment caused nearly complete loss of the lower molecular weight form of subunit H with a partial reduction in the higher molecular weight form. Because this treatment results in nearly complete loss of activity, it appears that loss of either isoform of subunit H is sufficient to inactivate the enzyme. These two isoforms appear to be the product of alternative splicing of a single gene (34).
We have previously shown that cystine selectively modifies Cys-254 of the V-ATPase A subunit, leading to inactivation of the enzyme that can be reversed by treatment with dithiothreitol (33). It should be noted, however, that the activity of the V-ATPase depleted of subunit H by cystine treatment cannot be restored by dithiothreitol (39) and this loss of activity is therefore not a direct result of the active site modification but rather loss of subunit H. Nevertheless, the results clearly indicate that there is intramolecular communication between the catalytic site of the enzyme (located on subunit A; Ref. 33) and the contact region between subunit H and the remainder of the complex. Such conformational cross-talk may play an important role in regulation of assembly or activity of the V-ATPases.
Our previous measurement of the subunit stoichiometry of the V-ATPase complex revealed that each V-ATPase molecule contains three copies of subunit A, three copies of subunit B, six copies of subunit c, and single copies of subunits C, D, E, a, d, and cЉ (10). At the time, subunits F, G, and H had not been discovered and it was not known that the V-ATPase contained an additional proteolipid subunit of molecular mass 17 kDa (cЈ), as demonstrated first for the V-ATPase of yeast (17). In addition, it has been reported that the V-ATPase from chromaffin granules contains a 45-kDa integral membrane glycoprotein that has been cloned and sequenced (40). We have observed the presence of this polypeptide in the intact V-ATPase of clathrincoated vesicles (49), as well as in the isolated V 0 domain, but there has been no report of a corresponding vma gene in yeast, despite the completion of the genome sequencing. We therefore wished to determine the copy number of the more recently discovered subunits of the V-ATPase and to determine whether the 45-kDa glycoprotein was in fact present in stoichiometric amounts in our preparation. Because different polypeptides display very different staining efficiencies, relative staining of proteins is a very unreliable method for determining subunit stoichiometry. Instead, we employed the method we had previously used involving electrophoretic separation and transfer to Immobilon followed by quantitative amino acid analysis (10). As shown in Table I, subunit F and the lower molecular weight isoform of subunit H are present in single copies while subunit G is present in two copies per complex. Because Western blotting analysis suggests that the two isoforms of subunit H are present in equal amounts (Fig. 7), these results suggest that both isoforms of subunit H are present in one copy per complex. This result is consistent with the observation noted above that removal of just one of the subunit H isoforms is sufficient to lead to complete loss of activity, since if the two isoforms were present in different populations of V-ATPase, a less dramatic effect on activity would have been predicted. In addition, we find that the 45-kDa glycoprotein is present at a substoichiometric level, suggesting that it is only present in a subpopula-tion of the V-ATPases from coated vesicles. This is not entirely surprising, however, given that clathrin-coated vesicles are derived from both the plasma membrane and the Golgi, and that this 45-kDa polypeptide may be present in only one of these two populations. Fig. 8 shows a revised structural model of the V-ATPase, incorporating both previous findings and the data presented in the current paper. Electron microscopic analysis of the V-ATPase from Clostridium fervidus (25) and from clathrincoated vesicles 2 indicates that the peripheral V 1 domain is attached to the integral V 0 domain both by a central stalk and by a peripheral stalk, as has been observed for the F-ATPases (20). In the case of the F-ATPases, the central stalk is composed of the ␥ and ⑀ subunits while the peripheral stalk is composed of the ␦ subunit and the soluble domains of subunit b (21). Because the D subunit, like ␥ of F 1 , is predicted to have a very high ␣-helical content (50), we suggest that subunit D is the V-ATPase homolog to ␥, placing subunit D, and subunit F with which it interacts, in the central stalk. By contrast, subunits C and H can be readily dissociated, suggesting that they form part of a peripheral structure. Both of these subunits show interaction with subunit E, which can also be cross-linked with subunit G in V 1 and subunit a of the V 0 domain. We have therefore placed subunits C, E, G and H together with the amino-terminal soluble domain of subunit a in the peripheral stalk. We speculate that the cross-link observed between subunits D and E in the intact complex may represent a point of close contact between the peripheral and central stalks which may change during the course of catalysis. It will be important to test whether these or other subunit contacts are in fact altered during turnover of the enzyme. Such catalysis-dependent changes in subunit contacts have in fact been demonstrated for the F-ATPase complex using covalent cross-linking (22,51). The picture presented in Fig. 8 thus represents a working model of the structure of the V-ATPases, which can serve as the basis for further experimental analysis. and F, while the peripheral stalk includes subunits C, E, G, and H. The model shown includes contacts identified in the current paper between subunits C and E, subunits D and F, subunits E and G, subunits H and E, and subunits a and E. The model also reflects the stoichiometry of two copies of subunit G, one copy of each isoform of subunit H, and one copy of subunit F per V-ATPase complex. We suggest that the observed cross-linked product between subunits H and F may be a result of conformational distortion in V 1 and that the contact between subunits D and E may reflect a point of close approach between the peripheral and central stalks.