Characterization of the Functional Coupling of Bovine Brain Vacuolar-type H+-translocating ATPase

Vacuolar-type H+-translocating ATPases (V-ATPases or V-pumps) are complex proteins containing multiple subunits and are organized into two functional domains: a peripheral catalytic sector V1 and a membranous proton channel V0. The functional coupling of ATP hydrolysis activity to proton transport in V-pumps requires a regulatory component known as subunit H (SFD) as has been shown both in vivo and in vitro (Ho, M. N., Hirata, R., Umemoto, N., Ohya, Y., Takatsuki, A., Stevens, T. H., and Anraku, Y. (1993) J. Biol. Chem. 268, 18286–18292; Xie, X. S., Crider, B. P., Ma, Y. M., and Stone, D. K. (1994) J. Biol. Chem. 269, 25809–25815). Ca2+ is thought to uncouple V-pumps because it is found to support ATP hydrolysis but not proton transport, while Mg2+ supports both activities. The direct effect of phospholipids on the coupling of V-ATPases has not been reported, likely due to the fact that phospholipids are constituents of biological membranes. We now report that Ca2+-induced uncoupling of the bovine brain V-ATPase can be reversed by imposition of a favorable membrane potential. Furthermore we report a simple “membrane-free” assay system using the V0 proton channel-specific inhibitor bafilomycin as a probe to detect the coupling of V-ATPase under certain conditions. With this system, we have characterized the functional effect of subunit H, divalent cations, and phospholipids on bovine brain V-ATPase and have found that each of these three factors plays a critical role in the functional coupling of the V-pump.

The vacuolar or V-type ATPases constitute a broad class of electrogenic proton-translocating ATPases that are among the most widely distributed ATP-driven ion pumps in nature. Vtype proton ATPases are present in all eukaryotic cells, and the structure of these proton pumps is highly conserved from yeast to human (1)(2)(3)(4)(5)(6)(7)(8). Ion-transporting ATPases that are structurally similar to V-ATPases 1 have also been found in various bacteria (9 -11) where they transport either H ϩ or Na ϩ or function as an ATP synthase. Within eukaryotic cells, V-type proton pumps are distributed among intracellular organelles of both constitutive and regulated pathways. In addition, V-AT-Pases have been localized to the plasma membranes of certain specialized cells such as renal intercalated cells, osteoclasts, and macrophages.
Over the past 2 decades, the importance of V-type ATPases has become increasingly appreciated (12)(13)(14)(15)(16)(17)(18). The ubiquitous function of V-pumps within cells is to acidify intracellular organelles, which is essential for receptor-mediated endocytosis, lysosomal hydrolase activities, intracellular protein targeting, processing of hormones, and uptake and storage of neurotransmitters. In cells with certain specialized functions, V-pumps are localized in the plasma membrane and are instrumental in the regulation of intracellular pH, acid secretion, and bone reabsorption.
V-ATPases are complex proteins containing at least 13 different subunits and are organized into two distinct functional domains: a peripheral catalytic sector V 1 and a membranous proton channel V 0 , similar to the F 1 F 0 complex structure of mitochondrial and bacterial ATP synthases. V-pumps are regulated at various levels including transcription (19,20), translation (21,22), direct modulation of enzymatic activity by activators (23,24), and reversible sulfhydryl-disulfide bond interconversion (25). V-pump activity is also indirectly controlled through the activity a chloride channel (26,27). But a regulatory mechanism that is unique to V-pumps is the reversible dissociation and association of V 1 and V 0 in response to energy demand; this mechanism has been extensively studied in yeast and in tobacco hornworm (28,29). A novel complex, RAVE (regulator of the ATPase of vacuolar and endosomal membranes), has been shown to associate with the V 1 domain of the yeast V-ATPase and to promote glucose-triggered assembly of V-ATPase (30). In mammals, interaction between the V-pump and aldolase has been observed in both osteoclasts and kidney cells (31), implying a certain type of coupling between glucose metabolism and V-pump activity that could be similar to what has been characterized in yeast.
The physical association between V 1 and V 0 , however, is not sufficient for the functional coupling of V-pumps, which in addition requires subunit H (SFD). Subunit H is a key regulatory element of V-ATPase that is not required for the assembly of V 1 V 0 but is absolutely essential for the function of V-type proton pumps as has been shown both in vivo (32) and in vitro (33,34). It remains to be determined how subunit H regulates the functional coupling of V-ATPases, and it is equally unclear what role certain non-protein factors may play in this process, such as divalent cations and phospholipids.
Investigations of the role of divalent cations in the function of V-ATPase have been largely limited to comparisons of the rates * This work was supported by NIDDK, National Institutes of Health Grant 33627. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 214-648-7700; Fax: 214-648-7720; E-mail: xiao-song.xie@UTSouthwestern.edu. 1 The abbreviations used are: V-ATPase (V-pump), the vacuolar-type H ϩ -translocating ATPase; V 1 , the peripheral, catalytic sector of Vpumps; V 0 , the membranous proton channel domain of V-pumps; F- of ATP binding, ATP hydrolysis, and proton pumping catalyzed by various divalent cation-ATP complexes (35)(36)(37)(38). In general, most investigators have found that Mg 2ϩ and Mn 2ϩ support all three activities, whereas Ca 2ϩ only supports ATP binding and hydrolysis, not proton pumping. An exception to this generalization is a report indicating that Ca-ATP will support proton pumping in chromaffin granule ghosts (39), which other investigators have not observed for unknown reasons.
In contrast, a significant body of literature exists regarding the interactions of divalent cations with F-ATPase (40 -46). Multiple sites have been demonstrated; these include catalytic sites involved with cation-ATP interactions, a Mg 2ϩ specific non-catalytic site that is responsible for stabilization of the complex, and additional sites that have regulatory functions. In general, divalent cations have been proposed to regulate the functional coupling of ATP hydrolysis/synthesis and proton movement in F 1 F 0 systems. More specifically, Ca 2ϩ has been shown to uncouple ATP hydrolysis/synthesis from proton movement in F-ATPases (43).
Studies on ion transport require an assay system that is partitioned into two compartments by membranes, which contain phospholipids. This represents an inherent difficulty in determining whether phospholipids play a direct role in the functional coupling of V-pumps.
In this study, we demonstrate that Ca-ATP results in only a partial uncoupling of V-ATPase and that significant proton translocation in this setting is restored by a favorable membrane potential. In addition, we report a simple assay system using the V 0 proton channel-specific inhibitor bafilomycin A 1 as an indicator of functional coupling of ATP hydrolysis to proton translocation. With this assay system, a direct role of phospholipids in the functional coupling of V-ATPase is demonstrated, and the effects of subunit H and divalent cations are further characterized.

EXPERIMENTAL PROCEDURES
Materials-Acridine orange was obtained from Eastman Kodak Co., 5-carboxyfluorescein (5-CF) was from Molecular Probes, Zwittergent 3-16 was from Calbiochem, [␥-32 P]ATP was from Amersham Biosciences, and all phospholipids were from Avanti Polar lipids, Inc. The proton ionophore 1799 was the generous gift of Dr. Peter Heytler (DuPont). All other reagents were obtained from Sigma.
Preparations-The bovine brain clathrin-coated vesicle V-ATPase was purified to a specific activity of 14 -16 mol of P i ϫ mg Ϫ1 ϫ min Ϫ1 as described previously (47). The subunit H-free V-ATPase was prepared by removal of subunit H (SFD) from purified V-ATPase using Zwittergent 3-16 as reported previously (33) with a minor modification. In brief, 5 mg of purified V-ATPase was precipitated by ammonium sulfate at 45% saturation and centrifuged at 45,000 rpm with a Ti70 rotor for 20 min at 4°C. The precipitate was dissolved in 2 ml of a solution containing 1% Zwittergent 3-16, 20 mM Tris-Cl, pH 7.5, 0.5 mM EDTA, and 1 mM dithiothreitol and incubated on ice for 1 h. The resultant subunit H-free V-ATPase and dissociated subunit H were separated by glycerol density gradient centrifugation as described previously (33). Recombinant subunit H was prepared as reported earlier (34). The two isoforms of subunit H (SFD␣ and SFD␤) showed no detectable difference in their activation activity.
Reconstitution of V-pump into Proteoliposomes-The bovine brain V-pump was reconstituted into liposomes, which contain phosphatidylcholine, phosphatidylethanolamine (PE), phosphatidylserine (PS), and cholesterol at a weight ratio of 40:26.5:7.5:26, by the cholate dilution, freeze-thaw method as described previously (48) with some modifications. In brief, liposomes (200 g) were added to 1 g of V-ATPase and were well mixed. Glycerol, sodium cholate, KCl/NaCl, and MgCl 2 (or CaCl 2 ) were added to the protein-lipid mixture at final concentrations of 10% (v/v), 1%, 0.15 M, and 2.5 mM, respectively. The KCl/NaCl ratio varied to allow for designated interior [KCl] of proteoliposomes. For experiments with interior negative membrane potential, [KCl] in was 150 mM; for experiments with interior positive membrane potential, [KCl] decreased, but [KCl ϩ NaCl] was constant at 150 mM. The reconstitution mixture was incubated at room temperature for 1 h, frozen in liquid N 2 for 1 min, and then thawed at room temperature. For the proton pumping assays in which no membrane potential was applied, the mixture was directly diluted with 1.5 ml of proton pumping assay buffer (150 mM KCl, 20 mM Na-Tricine, pH 7.0, and a 3 mM concentration either of MgCl 2 or CaCl 2 ) in a spectrophotometer cuvette, which allows the formation of sealed ready-to-assay proteoliposomes. The volume of the reconstitution mixture was maintained at Ͻ30 l such that it would be diluted at least 50-fold when assayed. For experiments requiring the imposition of a membrane potential, the proteinlipid mixtures containing the designated [KCl] were diluted 60-fold in buffers with corresponding salt concentrations and centrifuged to sediment the sealed proteoliposomes. The proteoliposomes were then resuspended in a small volume of the dilution buffer and were stable for at least 2-3 days at 4°C.
Measurement of Proton Translocation-For experiments using the acridine orange quenching method (49), the assays were conducted in an SLM-Aminco DW2C dual wavelength spectrophotometer, and the activity was registered as ⌬A 492 Ϫ 540 . Generally 5-10 l of proteoliposomes were added to 1.5 ml of proton pumping assay buffer containing 20 mM Tricine, pH 7.0, 6.7 M acridine orange, a 3 mM concentration of either MgCl 2 or CaCl 2 , and varying concentrations of KCl from 0 to 150 mM to generate a predetermined membrane potential in the presence of valinomycin. The [KCl ϩ NaCl] in the assay solution was constant at 150 mM. Due to the contribution of KCl from the proteoliposome samples, the actual [KCl] out was at least 0.5 mM. The reaction was initiated by addition of 1.3 mM ATP (pH 7.0) and 1 g/ml valinomycin. For experiments where Mg 2ϩ or Ca 2ϩ was added to start the reaction, divalent cations were omitted initially from the assay buffer, which instead contained 0.2 mM EDTA to remove residual free divalent cations. For acidification assays using 5-CF, the experiments were carried out in a PerkinElmer Life Sciences LS-5B luminescence spectrometer at excitation and emission wavelengths of 490 and 520 nm, respectively. In a typical reaction, 5 l of proteoliposomes containing 0.5 mM 5-CF were diluted into 1.5 ml of assay solution containing various salt concentrations as described above. The reactions were initiated by the addition of valinomycin and ATP, and the results were normalized to control experiments conducted with Mg 2ϩ under equal internal and external K ϩ concentration.
Measurement of ATPase Activities-ATPase activity was measured as the liberation of 32 P i from [␥-32 P]ATP (47). For the soluble system, purified V-ATPase was first mixed with or without effectors such as subunit H or phospholipids. The reaction was started by addition of 200 l of ATPase assay solution A (30 mM KCl, 50 mM Tris-MES, pH 7.0, 3 mM MgCl 2 (or CaCl 2 ), and 3 mM [␥-32 P]ATP (400 cpm/nmol)) in the presence or absence of the inhibitor bafilomycin A 1 and continued for 15-30 min at 37°C. For the ATPase assay with proteoliposomes, the samples were first mixed with or without subunit H followed by addition of ATPase assay solution B to start the reaction. ATPase assay solution B is basically the same as the assay solution A except that it contained valinomycin and variable concentrations of KCl with a constant [KCl ϩ NaCl] at 150 mM. The assays with proteoliposomes were carried out at room temperature to reduce the nonspecific leak of ions through the membrane and in the presence or absence of the proton ionophore 1799 as designated in the legends to figures. The ATP hydrolysis reaction was terminated by adding 1.0 ml of 1.25 N perchloric acid, and the released 32 P i was extracted and counted in a Beckman scintillation counter as described previously (50). The results were expressed either as the actual amount of ATP hydrolyzed (nmol of P i /assay) or as specific activity (mol of P i /min/mg of protein).
Measurement of ATP-driven Membrane Potential-Changes in the absorbance of oxonol VI at ⌬A 625 Ϫ 587 (51) was used as a qualitative measure of ATP-driven membrane potential generated by V-ATPase. The assay was conducted in an SLM-Aminco DW2C dual wavelength spectrophotometer in a solution containing 200 mM sucrose, 50 mM NaCl, 20 mM Tris-MES, pH 7.0, and a 3 mM concentration of either CaCl 2 or MgCl 2 . The reaction was started by addition of 1.3 mM ATP and terminated by adding 1799 at 1 nM.

The Ca-ATPase Activity of Bovine Brain V-pump Is Coupled to Proton Translocation under an Advantageous Membrane
Potential-V-ATPase complex hydrolyzes ATP in the presence of either Mg 2ϩ or Ca 2ϩ , but proton pump activity is detected only in the presence of Mg 2ϩ and not in the presence of Ca 2ϩ . The dissociated V 1 domain, on the other hand, hydrolyzes ATP only in the presence of Ca 2ϩ (33, 52-54) with Mg 2ϩ being inhibitory under the same conditions. Based on these observa-tions, Ca 2ϩ is thought to uncouple V-pumps, as it does the F-type pumps, and the Ca-ATPase activity has been considered to reflect the enzymatic activity of V-ATPase in the uncoupled state.
This concept, however, was challenged by our observations that the Ca-ATPase activity of reconstituted bovine brain V-ATPase was stimulated by a proton ionophore 1799 (shown later in Fig. 3) and that a membrane potential was detected in the presence of Ca-ATP, albeit to a much lesser extent than that seen in the presence of Mg 2ϩ (Fig. 1A). Since the ionophore stimulation and the generation of membrane potential are both characteristics of a coupled ATPase activity, these data suggest that the Ca-ATPase activity of the intact V-pump is only partially uncoupled from proton movement and that the resulting acidification of proteoliposomes is too weak to detect under normal conditions. Our routine assay for proteoliposomes acidification was performed under conditions of [K ϩ ] out ϭ [K ϩ ] in ϭ 150 mM. We found that proton translocation in the presence of Ca-ATP was restored by decreasing [K ϩ ] out to 1.5 mM and therefore imposing a favorable membrane potential in the presence of the K ϩ ionophore valinomycin as described under "Experimental Procedures." As shown in Fig. 1B, ATP-driven proton translocation was only detected in the presence of Mg 2ϩ when [K ϩ ] out ϭ [K ϩ ] in (traces 1 and 2). However, when an interior negative electrical gradient was established (traces 3 and 4), Ca-ATPdriven proton transport was restored to a level comparable to that supported by Mg 2ϩ under usual conditions. To confirm that the activity is indeed supported by Ca 2ϩ , the reaction was initiated by adding the respective divalent cations. This is the first demonstration that the V-pump can be functionally coupled in the presence of Ca 2ϩ in an in vitro experiment using purified enzyme at an advantageous membrane potential. This reaction is distinct from the membrane potential-driven proton translocation of reconstituted V 0 proton channel (55) because it is ATP-dependent. We subsequently found that the weak acidification supported by Ca 2ϩ under normal conditions could be detected using a more sensitive fluorescent pH probe, 5-CF (see Fig. 2). Fig. 2 shows results from a series of experiments in which both proton translocation and ATPase activity of reconstituted proteoliposomes of bovine brain V-pump were measured in the presence of either Mg 2ϩ or Ca 2ϩ as a function of the imposed membrane potential. Both ATPase and proton pumping activities in the presence of Mg 2ϩ and at zero membrane potential were used as 100% activity, respectively. The membrane po-FIG. 1. Detection of membrane potential and acidification in the presence of Ca 2؉ or Mg 2؉ . Bovine brain V-ATPase was reconstituted into proteoliposomes as described under "Experimental Procedures." ATP-driven membrane potential (A) was measured as ⌬A 625 Ϫ 587 using oxonol VI as a membrane potential probe in an assay solution containing 200 mM sucrose, 50 mM NaCl, 20 mM Tris-MES, pH 7.0, and a 3 mM concentration of either CaCl 2 (trace 1) or MgCl 2 (trace 2), and the reaction was initiated by addition of 1.3 mM ATP. Because the reconstituted vesicles were sealed by dilution into the assay solution, the internal ion composition of the proteoliposomes should be basically the same as that of the assay solution. To measure proton translocation (B), the proteoliposomes loaded with 150 mM KCl were sealed first as described under "Experimental Procedures" prior to the assay. Acidification was measured using acridine orange quenching as ⌬A 492 Ϫ 540 in a solution containing 10 mM Na-Tricine, pH 7.0, 0.2 mM EDTA, and a 150 mM concentration of either KCl (traces 1 and 2) or NaCl (traces 3 and 4). ATP and valinomycin were added as indicated before addition of either CaCl 2 (traces 1 and 3) or MgCl 2 (traces 2 and 4) at 3 mM to start the reaction. Val, valinomycin.
FIG. 2. The proton translocation and ATPase activities of bovine brain V-pump as the function of imposing membrane potential. Bovine brain V-ATPase, 2 g of protein for each assay, was reconstituted into proteoliposomes and sealed as described under "Experimental Procedures." The membrane potential was set up by varying either the [KCl] out in the assay solution or the [KCl] in of the proteoliposomes as described under "Experimental Procedures." The [KCl ϩ NaCl] in any case was maintained at 150 mM. Both proton translocation and ATPase assays were carried out at room temperature in the presence of valinomycin and either Mg 2ϩ or Ca 2ϩ . ATPase activity was measured as liberation of 32 P i from [␥-32 P]ATP, and acidification was measured using 5-CF as the pH probe at excitation and emission wavelengths of 490 nm and 520 nm, respectively, as described under "Experimental Procedures." The activities were expressed using that in the presence of Mg 2ϩ and at zero membrane potential as 100% of ATPase and acidification activities, respectively. tential ⌬⌿ was established by varying the ratio of [K ϩ ] out to [K ϩ ] in in the presence of valinomycin. Since only a small number of ions moving across the membrane is needed to generate a substantial membrane potential, ⌬⌿ can be calculated using the formula E ϭ 58 mV ϫ log([K ϩ ] o /[K ϩ ] i ) by simplifying the Nernst equation for the specific conditions used in this study. As shown in Fig. 2, the rate of proton translocation parallels that of ATPase activity in the presence of Mg 2ϩ , which is an indication of tight coupling. In the presence of Ca 2ϩ , the proton pump activity does not parallel the ATPase activity but rather falls with a greater slope following the increase in the membrane potential, suggesting changes in the coupling efficiency. This result is analogous to the "slippage" phenomenon observed in some P-type ion-transporting ATPases working against a transmembrane ion gradient in which the coupling ratios of ion transport to ATP hydrolysis in those P-type pumps are reduced from theoretical values down to zero (56,57). While the mechanism underlying this slippage of V-pump in the presence of Ca 2ϩ remains unclear, this observation provides a convenient experimental tool to monitor the transition of V-ATPase from the uncoupled to coupled states since the Mg-ATPase activity of V-pump is difficult to measure for uncoupled enzyme, whereas the Ca-ATPase activity is present in both coupled and uncoupled states.
Effect of Phospholipids on the Coupling of V-ATPase in a Membrane-free System-The V-pump loses its Mg-ATPase activity when endogenous phospholipids are removed during the solubilization and purification of the enzyme, and the activity is restored when phospholipids are added back with the order of PS Ͼ phosphatidylcholine Ͼ phosphatidylethanolamine (see Fig. 4). However, a substantial level of Ca-ATPase activity remains in purified V-ATPase after the removal of endogenous phospholipids. Since the Ca-ATPase activity of V-pump is partially uncoupled, we compared its phospholipid dependence with that of the Mg-ATPase activity of the V-pump complex and the Ca-ATPase activity of dissociated V 1 . As shown in Table I, the Mg-ATPase activity of intact V-pump, which is tightly coupled to proton movement as shown in Fig. 2, is highly dependent on PS, whereas its Ca-ATPase activity is much less dependent. The Ca-ATPase activity of V 1 , which is obviously completely uncoupled, is not only independent of PS but is actually inhibited by PS. These findings raise a question: are phospholipids required for the enzymatic activity in a nonspecific manner, or do they play a direct and specific role in the coupling of V-pump?
Studies on ion transport often require assay systems with a partition between two compartments. The partition takes the form of either native or artificial membranes, both of which require phospholipids. To investigate the direct effect of phospholipids on the coupling of the V-pump we need a membranefree system. Since bafilomycin A 1 inhibits V-ATPase through its inhibition of the proton channel V 0 (55) and does not inhibit the Ca-ATPase activity of dissociated V 1 , 2 we explored the possibility of using bafilomycin sensitivity as an indicator to probe the coupling state of the V-ATPase under different conditions. Interpretation of these measurements is contingent on the assumption that bafilomycin inhibits only the ATPase activity that is coupled to proton movement through the V 0 proton channel, which needs to be confirmed. As discussed above, the Mg-ATPase activity of V-pump is tightly coupled to proton translocation in proteoliposomes, whereas the coupling of Ca-ATPase activity is conditional and can be dissected into coupled and uncoupled components by using a proton ionophore or by the imposition of a negative membrane potential. As shown in Fig. 3, bafilomycin A 1 is highly inhibitory to the Mg-ATPase activity but inhibits only the portion of Ca-ATPase activity that is stimulated either by the proton ionophore 1799 or by a favorable membrane potential. These results validate the feasibility of using bafilomycin sensitivity as an indicator for the coupling of ATPase activity to proton movement in V-pumps.
We next examined the bafilomycin sensitivity of both Mgand Ca-ATPase activities of the solubilized V-pump in the presence or absence of phospholipids (PLs). Each PL was prepared as a dilute aqueous solution in two steps. First, PL was solubilized by detergent (1% cholate) at a concentration of 5 mg of PL/ml, and the solution was sonicated until clear. The stock solution of phospholipids was then diluted 10-fold to 0.5 mg/ml with a buffer, by which the solubilized V-ATPase was prepared, for easy mixing during the assay. In general, 2 g of PL was used for each assay. As shown in Fig. 4, the Mg-ATPase activity of V-pump is almost undetectable in the absence of phospholipids but is activated dramatically by PS, phosphatidylcholine, or phosphatidylethanolamine. The PL-activated Mg-ATPase activity is highly sensitive to bafilomycin. In contrast, the Ca-ATPase activity of V-pump is easily detected without any addition of phospholipids but is resistant to bafilomycin. Phospholipids moderately stimulate the Ca-ATPase activity of the V-pump, and the PL-stimulated portion of Ca-ATPase activity is sensitive to bafilomycin. These data further demonstrate that phospholipids are not essential for the intrinsic ATPase activity of the enzyme but rather are required for the coupling of ATPase activity to proton movement through the bafilomycin-sensitive proton channel. The stimulation of ATPase activity by phospholipids occurred in the absence of a membrane structure and provided an explanation for the phospholipid dependence of V-pump Mg-ATPase activity in soluble systems 2 B. P. Crider and X.-S. Xie, unpublished observation.

TABLE I Phospholipid dependence of V-ATPase preparations
The ATPase activity of soluble V-ATPase preparations was measured by the liberation of 32 P i from ͓␥-32 P͔ATP as described under ''Experimental Procedures.'' The amounts of protein used were: intact V-ATPase, 0.5 g; V 1 preparation, 0.6 g. The enzyme preparations were incubated with PS, 2 g/designated assay, at room temperature for 5 min prior to the assay. The reaction was started by adding the assay solution and was carried out at 37°C for 15 min.  3. Effect of membrane potential and proton ionophore on the ATPase activities of reconstituted V-pump. The proteoliposomes of bovine brain V-ATPase were reconstituted, loaded with 150 mM KCl, sealed, and assayed for ATPase activity as described under "Experimental Procedures." The final concentrations of valinomycin, 1799, and bafilomycin A 1 were 1 g/ml, 1 g/ml, and 0.1 M, respectively. bafil., bafilomycin A 1 .
where proton movement is in a futile cycle. The direct role of phospholipids in the coupling of V-pump would be difficult to investigate using a membrane system, which is probably why the potential role of phospholipids in the functional coupling of V-pumps was not recognized previously.

Characterization of Subunit H as a Biochemical Switch for the Functional Coupling of V-ATPase-Subunit H (SFD) is a unique component of V-ATPase.
The V 1 V 0 complex can be assembled without it, but the resulting complex does not pump protons (32)(33)(34). Addition of subunit H to the subunit H-free enzyme simultaneously restores Mg-ATPase activity and proton pumping activity (33). We further investigated the coupling effect of subunit H together with that of PS by using bafilomycin sensitivity as the coupling indicator. In these experiments we used a subunit H-free V-ATPase preparation (33) and purified recombinant subunit H (34).
We first examined the effect of subunit H in a reconstituted proteoliposome system. As shown in Fig. 5, the Mg-ATPase activity of the V-pump in the absence of subunit H was minimal and was not responsive to bafilomycin or 1799. The Mg-ATPase activity was activated by subunit H, and the activated ATPase activity was highly responsive to stimulation by 1799 and to inhibition by bafilomycin, indicating functional coupling. The Ca-ATPase activity of the proteoliposomes in the absence of subunit H was high but not responsive to bafilomycin or 1799. Therefore the Ca-ATPase activity of the proteoliposomes in the absence of subunit H represents uncoupled ATPase activity. After addition of subunit H, the Ca-ATPase activity fell 40% and became partially responsive to both 1799 stimulation and bafilomycin inhibition, reflecting a transition from uncoupled to partially coupled activity.
We then measured the activities of the same V-ATPase preparation in the soluble system to analyze the effect of phospholipids. As shown in Fig. 6A, the ATPase activity of the subunit H-free enzyme was hardly detectable in the presence of Mg 2ϩ and was not stimulated by PS or activated by subunit H without the presence of PS. Upon the addition of both subunit H and PS, the Mg-ATPase activity was stimulated dramatically and was highly sensitive to bafilomycin. In the presence of Ca 2ϩ , the subunit H-free enzyme had high ATPase activity that was insensitive to bafilomycin, similar to the results obtained using the reconstituted system. The Ca-ATPase activity was substantially reduced by addition of PS (Fig. 6B, columns 1 and 3) but was re-elevated with both subunit H and PS present (Fig. 6B,  columns 3 and 7). The stimulation of Ca-ATPase activity by subunit H in the presence of phospholipids is consistent with our previous study (33) and is in fact due to the recovery of the activity that was inhibited by phospholipids. The recovered Ca-ATPase activity by addition of PS and subunit H, however, becomes partially bafilomycin sensitive. Thus, the functional coupling of the subunit H-free V-ATPase was "switched on" by subunit H in the presence of PS and would be complete in the presence of Mg 2ϩ but only partial in the presence of Ca 2ϩ . DISCUSSION The mechanism for coupling ATP hydrolysis to ion transport is a fundamental issue for ATP-driven ion pumps. It is of particular interest for V-ATPases because the coupling of these pumps is regulated as discussed in the Introduction. The coupling process of V-ATPase has two components: the physical association between V 1 and V 0 and the functional coupling of ATP hydrolysis to proton transport activities. The separation of these two steps has been demonstrated both in vivo and in vitro (32)(33)(34). The physical connection of the catalytic sector V 1 to the proton channel V 0 is the basic step essential for the formation of a coupled V-ATPase; however, functional coupling also requires the regulatory component subunit H (SFD) and two non-protein factors, phospholipids and divalent cations. As demonstrated in the present study, phospholipids are essential for the function of subunit H as a biochemical switch, and divalent cations determine the coupling efficiency of V-pump.
Studies on the functional coupling of V-pump have been hindered by the lack of a versatile and capable assay system. The assay of proton transport in a reconstituted system is convenient for measuring the overall function but cannot pro- vide a means for monitoring the transition of the coupling process or for detailing the effect of factors involved. In addition, the obligation of using phospholipids in reconstituted membrane systems makes it difficult to investigate the potential role of phospholipids in the functional coupling of V-pump. While ATP hydrolysis activity can be measured with or without the presence of a membrane system, it lacks a means to correlate ATP hydrolysis to proton movement. Furthermore Mg 2ϩ only supports a significant level of ATPase activity when the V-ATPase is coupled, and Ca 2ϩ was thought to uncouple the enzyme.
The observations that the coupling of V-ATPase can be restored by a favorable membrane potential in the presence of Ca 2ϩ and that bafilomycin inhibition can be used as an indicator for the functional coupling of V-ATPase have supplied additional tools to investigate the coupling of V-pumps in a system with or without a membrane structure. With this system, we have shown that phospholipids are not essential for the basic ATP hydrolysis activity of V-pump but rather are required for the functional coupling of this enzyme. This is demonstrated by the phospholipid-independent and bafilomycinresistant ATPase activity in the presence of Ca 2ϩ and the phospholipid dependence of the bafilomycin-sensitive ATPase activity in the presence of either Ca 2ϩ or Mg 2ϩ .
The Ca-ATP-driven proton pumping of V-ATPase in the presence of a favorable membrane potential and the membrane potential-driven proton translocation of V 0 proton channel (55) are distinct activities in nature because the former activity is an ATP-dependent process, while the latter is not. In addition, an acid activation is required for the V 0 proton channel to be opened in vitro (55), which will irreversibly inactivate both Caand Mg-ATPase activities (data not shown). One additional concern in this regard was the possibility of a Ca-ATP-gated proton channel, which in fact negated both factors in the argument: the ATP dependence of an active transport and the requirement of activation for the passive channel activity.
While the Ca-ATP gating mechanism is intriguing, it is unlikely. The observation of Ca-ATP-generated membrane potential (Fig. 1A) and the experiments with the more sensitive probe 5-CF (Fig. 2) evidently address this issue. First, these results together clearly showed that the Ca-ATP-driven proton transport indeed happens without an assisting membrane potential or even against an electrochemical potential. Second, if a Ca-ATP gating mechanism exists at the same time, it should have collapsed the Ca-ATP-induced membrane potential and should have prevented the proton influx at zero membrane potential or even generated a proton efflux under a positive membrane potential. It is apparently not the case.
The coupling efficiency of V-pump in the presence of Mg 2ϩ is constant over a wide range of applied membrane potential, while the coupling efficiency in the presence of Ca 2ϩ is directly affected by applied membrane potential. This variable coupling efficiency of V-pump in the presence of Ca 2ϩ is analogous to the slippage phenomenon observed in some P-type ion pumps (56,57) of which the coupling efficiency decreases following the increase of the respective ion gradient. The mechanism underlying this slippage of V-ATPase in the presence of Ca 2ϩ and the reason it does not occur in the presence of Mg 2ϩ remain unclear. However, studies on the isoforms of subunit a of the V 0 domain have shown a correlation between the isoform structure of this subunit and the coupling efficiency of V-ATPase (8).
In addition, a recent study has demonstrated that mutations within a certain region of subunit A alter the coupling efficiency of V-ATPase as well (58). Furthermore structure studies with electron microscopy have revealed the existence of a second stator (a peripheral stalk) in the connecting region between V 1 and V 0 (59), which indicates the involvement of more subunits in the coupling of V-ATPase than people originally thought. All these observations provide potential mechanisms by which the coupling of V-ATPase could be affected through the binding of a certain divalent cation to alter the structure of those regions.
The role of divalent cations in the coupling of V-pump is in addition to their function in the catalytic site interactions with ATP, indicating additional divalent cation binding site(s) as in the case of F-type pumps. This site(s) could either be specific for a certain cation, Mg 2ϩ or Ca 2ϩ , or have similar affinity to both cations but with different functional consequences. The former possibility is supported by the observation that Mg 2ϩ at micromolar concentrations inhibits the Ca-ATPase activity of dissociated V 1 (with [Ca 2ϩ ] at 3 mM) (52,53), suggesting the existence of a Mg 2ϩ -specific binding site(s) within the V 1 domain. Physiologically this property of V-ATPase provides a mechanism to prevent a futile cycle of ATP hydrolysis by free V 1 in cytosol. Biochemically the Ca-ATPase activity of V-pump offers investigators a more convenient tool to probe the transition of V-ATPase from an uncoupled to a coupled state since it is present in both situations and its coupling to proton movement is conditional.
The specific role of phospholipids, PS in particular, in the functional coupling of V-ATPase is in addition to their role as the constituents of biological membranes, although the mechanism underlying this reaction remains to be solved. It is possible that the binding site(s) for phospholipids with respect to their role in the coupling of V-ATPase is not located within or limited to the transmembrane domains. This is supported by the effect of PS on the ATPase activity of V 1 , which is not integrated into the membrane bilayer.
Subunit H (SFD) is a unique component of V-ATPase. The V 1 V 0 complex can be assembled without it, but the resulting complex does not pump protons (32,33). The effect of subunit H in the functional coupling of V-ATPase was further character- ized in a soluble system as well as reconstituted proteoliposomes. Addition of subunit H to the subunit H-free enzyme simultaneously restores Mg-ATPase activity and proton pumping activity (33,34). By measuring both Mg-and Ca-ATPase activities in their response to 1799 stimulation and bafilomycin inhibition, the results from this study clearly demonstrate the essential and exquisite role of subunit H in the transition of V-ATPase from an uncoupled to a coupled state. The assembly of V 1 V 0 complexes is a slow process in vitro, usually taking several hours to complete, but the activation and functional coupling of V-pump by subunit H is almost instantaneous. We originally reported that subunit H activation reaches a halfmaximum level within a few minutes (33). The preparation of subunit H used in that study contains a low concentration of Zwittergent 3-16. With a detergent-free and a highly purified recombinant subunit H preparation, the activation occurs so fast that we are no longer able to obtain a time course (data not shown). Thus it appears that subunit H functions as a biochemical switch of V-pumps, which turns functional coupling on instantly in the presence of certain phospholipids. We have no evidence that this process actually occurs in vivo as a means of regulation or how it is triggered under physiological conditions. However, it provides a biochemical technique and an in vitro system in which the functional coupling of V-ATPase can be turned on and off to help elucidate the coupling mechanism of V-pumps.
The uncoupled V 1 V 0 complex can only hydrolyze ATP in the presence of Ca 2ϩ , while the coupled complex can hydrolyze ATP in the presence of either Ca 2ϩ or Mg 2ϩ . However, it should be noted that Mg-ATPase activity is not necessarily always coupled to proton transport in every situation. For example, while V 1 under normal conditions has only Ca-ATPase activity (Mg 2ϩ is inhibitory), it has been shown that Mg 2ϩ inhibition can be partially reversed either by 25% methanol treatment or by removal of subunit H (53,54). Although neither of these two effects has been reported in a mammalian system, we did observe that bovine brain V 1 -ATPase activity reconstituted from a minimal number of individual V 1 subunits hydrolyzes ATP in the presence of either Ca 2ϩ or Mg 2ϩ (60). It is likely that the effect of Mg 2ϩ depends on the conformation of V-pumps and/or the presence of other subunits that are catalytically non-essential. Treatment with methanol or removal of certain components of V-ATPase may induce conformational changes that result in the expression of Mg-ATPase activity without coupling to proton transport. This would also explain the "leakiness" of the residual Mg-ATPase activity of the holoenzyme in the presence of bafilomycin.
Another potential difference between the V-ATPases from yeast and mammals, with respect to the function of subunit H, is the observation in a yeast system that the V 1 -ATPase isolated from vma13⌬ cells shows significant higher Ca-ATPase activity than that from vma3⌬ cells (54). The former is subunit H-deficient and the latter is not, suggesting that subunit H inhibits the Ca-ATPase activity of dissociated V 1 domain. While this may simply reflect an intrinsic difference between the V-ATPases of mammalian and non-mammalian origins, there could be other possible reasons. It is evident that the lower Ca-ATPase activity of V 1 from vma3⌬ cells is not due to a lower V max , as reflected in the initial activity, but is rather due to a decay of activity beyond 1 min. Therefore the effect of subunit H could be on the stability of V 1 or on its affinity to ADP as the reaction product. Furthermore the V 1 complex isolated from yeast cytosol lacks subunit C, whereas the bovine V 1 complex we used contains all V 1 subunits from A to G. Although subunit C is not essential for the catalysis, it stabi-lizes V-ATPase complex, enhances H ϩ pumping activity (61), and stimulates the Ca-ATPase activity of V 1 -ATPase when all other V 1 subunits are present (62). A recent mutational analysis on subunit C of yeast V-ATPase has shown that this subunit has significant impact on both the stability and catalysis activity of the enzyme at the opposite directions (63). Therefore it is possible that the absence of subunit C tilted the balance between the stability and catalysis of V-ATPase resulting in an alteration of the effect by subunit H.
The property of Mg-ATPase activity is consistent between V-pump preparations that are either isolated as an intact enzyme or reconstituted from the subunit H-free V-ATPase and recombinant subunit H. However, we noticed the existence of a certain difference between these two preparations in the coupling of Ca-ATPase activity. This difference is reflected in the bafilomycin sensitivity of Ca-ATPase activity in the respective preparations as well as its stimulation by 1799 and by PS in the proteoliposomes and in the soluble system, respectively. Basically the preparation of V-pump that is isolated as an intact enzyme shows much higher coupling efficiency of Ca-ATPase activity than that of the reassembled V-ATPase. It is possible that the V-pump preparation that was stripped of subunit H by 1% Zwittergent underwent a certain conformational change during the process that can be corrected by Mg 2ϩ but not by Ca 2ϩ . An alternative explanation is the loss of a minor and unidentified component of V-ATPase during subunit H removal.
It is noted that the stimulation of the Ca-ATPase activity by adding back subunit H in the presence of phospholipids is only observed in the soluble system but not in the reconstituted proteoliposomes (Fig. 5B). This difference is likely due to multiple factors. For example, an uncoupled ATPase activity can usually be fully expressed in a reconstituted system, but the coupled activity is restricted by membrane potential and chemical gradient. In addition, there is possibly a certain conformational change of V-ATPase when it is embedded into a membrane.
We were surprised that ATPase activity was slightly stimulated by bafilomycin in the absence of PS. The stimulation was more apparent for Ca-ATPase activity since it is highly active without PS but is consistent for both Ca-and Mg-ATPase activities. The significance of these observations is not clear at this time.
We have established a simple experimental system in which the functional coupling of V-ATPase can be effectively switched on and off, and the transition of this process can be used to study the effect of potential coupling-related factors either individually or together. We have shown previously that the V 1 domain of bovine brain V-pump can be reconstituted from individual V 1 subunits, which can subsequently be assembled with V 0 to form a functional complex (60). It will be interesting to combine this reconstitution procedure with the assay system of the present study to investigate the effect of other components of V-pump in the functional coupling of this enzyme. As aforementioned, certain variations in isoform structure affect the coupling efficiency of V-ATPases. An interesting example in this regard is the existence of different types of V-ATPases in lemon fruit with different H ϩ /ATP stoichiometry (64) that are affected by different factors, one by ⌬pH and the other by membrane potential. Experiments are underway to characterize the potential coupling effect of certain V 1 subunits, such as subunits C, D, and F that are known to be catalytically nonessential (60,61), and to identify the subunits that are responsible for interactions with subunit H and phospholipids or provide the specific binding site for divalent cations.