Reconstitution of vacuolar-type rotary H+-ATPase/synthase from Thermus thermophilus.

Vacuolar-type rotary H(+)-ATPase/synthase (V(o)V(1)) from Thermus thermophilus, composed of nine subunits, A, B, D, F, C, E, G, I, and L, has been reconstituted from individually isolated V(1) (A(3)B(3)D(1)F(1)) and V(o) (C(1)E(2)G(2)I(1)L(12)) subcomplexes in vitro. A(3)B(3)D and A(3)B(3) also reconstituted with V(o), resulting in a holoenzyme-like complexes. However, A(3)B(3)D-V(o) and A(3)B(3)-V(o) did not show ATP synthesis and dicyclohexylcarbodiimide-sensitive ATPase activity. The reconstitution process was monitored in real time by fluorescence resonance energy transfer (FRET) between an acceptor dye attached to subunit F or D in V(1) or A(3)B(3)D and a donor dye attached to subunit C in V(o). The estimated dissociation constants K(d) for V(o)V(1) and A(3)B(3)D-V(o) were ∼0.3 and ∼1 nm at 25 °C, respectively. These results suggest that the A(3)B(3) domain tightly associated with the two EG peripheral stalks of V(o), even in the absence of the central shaft subunits. In addition, F subunit is essential for coupling of ATP hydrolysis and proton translocation and has a key role in the stability of whole complex. However, the contribution of the F subunit to the association of A(3)B(3) with V(o) is much lower than that of the EG peripheral stalks.

Introduction

Vacuolar-type ATPases (VoV1) are members of the rotary ATPase/ATP synthase superfamily, which catalyze the exchange between energy generated by proton translocation across a membrane and energy generated by ATP hydrolysis/synthesis (1–4). They are widely distributed in eukaryotic cells and bacteria (5, 6). Most prokaryotic VoV1 (also referred to as A-ATPase or AoA1 (1, 2)) produce ATP using the energy stored in a transmembrane electrochemical proton gradient (3, 7), whereas the VoV1 of some anaerobic bacteria, such as Enterococcus hirae, function as a sodium pump (8).

Thermus thermophilus VoV1 is capable of both ATP-driven proton translocation and proton-driven ATP synthesis in vitro and functions as an ATP synthase in vivo (3). The subunit structure of this VoV1 is simpler than the eukaryotic counterpart, being composed of nine subunits, A, B, D, F, C, E, G, I, and L. Each subunit shows a significant sequence similarity to its eukaryotic counterpart (supplemental Table 1). Several lines of evidence had previously suggested that the D, F, C, and L subunits form a central rotor with the I, E, and G subunits, constituting a stator apparatus together with the A3B3-hexamer (2, 9, 10) (Fig. 1). The recent cryo-EM map finally confirmed this subunit arrangement for T. thermophilus VoV1 (11).

FIGURE 1.

Schematic representation of Subunits in V1 and in Vo are shown in white and gray, respectively.

The ATPase-active V1 domain is composed of four subunits with a stoichiometry of A3B3D1F1 (12). The central rotor of V1 is composed of two different subunits, D and F, with subunit F functioning as an activator of ATPase activity (13). In contrast, the equivalent subunit in F1-ATPase, subunit ϵ, functions as an endogenous regulator of ATPase activity. However, the precise function of subunit F in the holoenzyme remains as yet unknown.

VoV1 and F-type ATPases (FoF1) are evolutionary related and share the rotary mechanism coupling ATP synthesis/hydrolysis and proton translocation across the membrane (1, 2). However, these two types of ATPase exhibit significant differences. The reversible association/dissociation of the catalytic and membrane-associated subcomplexes is unique to VoV1 and thought to be key for regulation of activity (14). Glucose deprivation has been shown to cause rapid dissociation of yeast VoV1 into free V1 and Vo, a process that is both reversible and independent of de novo protein synthesis. For eukaryotic VoV1, two groups had reported reconstitution of VoV1 in vitro (15, 16). However, the dynamics of the reconstitution of VoV1 have not been reported.

In addition, significant differences are observed between the overall features of the two ATPases, particularly in the stalk region. The central stalk is considerably longer in VoV1 than in FoF1 (17). Subunit C (eukaryotic d subunit) is located at the interface between V1 and the proteolipid ring, and this subunit is a major contributor to the extra length of the stalk region (18). This fact indicates the central shaft composed of subunits D and F does not contact the proteolipid ring directly. VoV1 also has a more complex peripheral stalk structure than FoF1. The stator structure of bacterial FoF1 consists of a single peripheral stalk formed by subunit b, whereas electron microscopic images of VoV1 suggest that V1 is connected with Vo by two or three peripheral stalks (11, 17, 19). The complex structure of the VoV1 stalk seems to be relevant for a comparatively more rigid association of V1 with Vo.

In this study, we show in vitro reconstitution of T. thermophilus VoV1 from isolated V1 and Vo. The reconstitution in real time was measured by fluorescence resonance energy transfer (FRET) analysis using labeled V1 and Vo, and thermodynamic parameters for the reconstitution were calculated. In addition, A3B3 and A3B3D subcomplexes also associated with Vo, suggesting that the peripheral stalks are mainly responsible for connecting V1 to Vo.

EXPERIMENTAL PROCEDURES

Isolation of Vo

Wild-type or mutant VoV1 (C-S105C/C-C268S/C-C323S) T. thermophilus strains incorporating a His8 tag on the N terminus of subunit A were generated by the integration vector system (20). The modified T. thermophilus strains were cultured as described previously (9). The cells (200 g) harvested at log phase growth were suspended in 400 ml of 50 mm Tris-Cl (pH 8.0), containing 5 mm MgCl2, and disrupted by sonication. The membranes were precipitated by centrifugation at 100,000 × g for 20 min and washed with the same buffer twice. The washed membranes were suspended in 20 mm imidazole sodium (pH 8.0), 0.1 m NaCl, and 10% Triton X-100 (w/v), and the suspension was sonicated. Cell debris and insoluble material were removed by centrifugation at 100,000 × g for 60 min, and the supernatant was applied onto a nickel-nitrilotriacetic acid superflow column (Qiagen, 3 × 5 cm) equilibrated with 20 mm imidazole sodium (pH 8.0), 0.1 m NaCl, 0.1% Triton X-100. The column was washed with 200 ml of the same buffer. The protein was eluted with a linear imidazole gradient (20–100 mm). The fractions containing the VoV1 were applied to a RESOURCE Q column (6 ml, GE healthcare) equilibrated with 20 mm Tris-Cl (pH 8.0), 0.1 mm EDTA, and 0.05% n-dodecyl-β-d-maltoside (Sigma). The proteins were eluted with a linear NaCl gradient (0–0.5 m). Each fraction containing Vo was combined and concentrated and then subjected to FPLC with a Superdex HR-200 column (GE healthcare) equilibrated with MOPDM buffer (20 mm MOPS, pH 7.0, 100 mm NaCl, 0.05% n-dodecyl-β-d-maltoside). The proteins were eluted with the same buffer. The mutated Vo (C-S105C/C-C268S/C-C323S) was used for the FRET experiments. The Vo fractions were combined and used immediately.

Isolation of V1 (A3B3DF), A3B3D, and A3B3

Escherichia coli strain BL21-CodonPlus-RP (Stratagene) was used for expression of V1 (A3B3DF), A3B3D, and A3B3. These recombinant subcomplexes were isolated as described previously (13). The expressed cells were suspended in 20 mm imidazole/HCl (pH 8.0) containing 0.3 m NaCl and disrupted by sonication. After removal of heat labile proteins derived from the host cells by heat treatment at 65 °C for 30 min, the solution was applied to an Ni2+ affinity column (Qiagen, 3 × 5 cm), which was then washed thoroughly and eluted with 0.5 m imidazole/HCl (pH 8.0) containing 0.3 m NaCl. The buffer was exchanged to 20 mm Tris/HCl (pH 8.0) containing 1 mm EDTA by ultrafiltration (Vivaspin, Vivascience), and the solution was applied to a RESOURCE Q column. The fractions containing subcomplexes were concentrated, and contaminating proteins were removed on a Superdex HR-200 column equilibrated with MOPDM buffer. The mutant V1 (A-His8/ΔCys, A-C255A/A-S232A/A-T235S/F-S54C), mutant A3B3 (A-His8/ΔCys, A-C255A/A-S232A/A-T235S), and mutant A3B3D (A-His8/ΔCys, A-C255A/A-S232A/A-T235S, A-His8/ΔCys, A-C255A/A-S232A/A-T235S, D/E48C) were used for either reconstitution or FRET experiments.

Reconstitution of VoV1, A3B3D-Vo, and A3B3-Vo

The purity of each subcomplex was confirmed by SDS- and AES-PAGE, a nondenaturing PAGE suitable for analysis of membrane protein complexes. V1, A3B3D, or A3B3 (each >1 mg/ml) in MOPDM buffer was mixed with 1 mg/ml Vo solution at an equal volume ratio. For reconstitution of VoV1, a range of different concentrations of Vo was added into V1 solution, and then the mixtures were incubated at 25 °C for 1 h. The mixtures were incubated for 1 h at 25 °C and then applied onto the Superdex HR-200 column equilibrated with the same buffer. The reconstituted complexes were collected and used for further analysis immediately.

FRET Analysis

The purified V1 (A-His8/ΔCys, A-C255A/A-S232A/A-T235S/F-S54C) or A3B3D (A-His8/ΔCys, A-C255A/A-S232A/A-T235S, D/E48C) was immediately labeled with a 2 m excess of Cy3TM-maleimide (GE healthcare, used as a donor molecule) in MOPDM buffer. Following a 60-min incubation at 25 °C, proteins were separated from unbound reagent with a PD-10 column (GE Healthcare). The mutated Vo (C-S105C/C-C268S/C-C323S) was labeled with Cy5TM-maleimide (GE Healthcare, used as an acceptor molecule) by the same method described above. The specific labeling of subunit F in V1 or subunit C in Vo was checked by measurement of each subunit fluorescence. FRET signals as a result of reconstitution of VoV1 were monitored with a fluorometer using an excitation wavelength of 532 nm and an emission wavelength of 570 nm (FP-3000, Hitachi). Typically, a quartz cuvette was filled with 1.2 ml of MOPDM buffer containing 5 nm labeled V1 or A3B3D and incubated at 25 °C until the fluorescence intensity reached a constant level. For measurement of binding kinetics, 10 μl of Vo-C105C-Cy5 at the indicated final concentration was added into the cuvette.

Other Assays

Protein concentrations of V1 were determined from UV absorbance calibrated by quantitative amino acid analysis; 1 mg/ml gives an optical density of 0.88 at 280 nm. Protein concentrations of Vo and VoV1 were determined by BCA protein assay, and V1 was used as protein standard. ATPase activity was measured at 25 °C with an enzyme-coupled ATP regenerating system. The ATPase assay solution contained 50 mm Tris-HCl (pH 8.0), 100 mm KCl, 2 mm MgCl2, 4 mm Mg-ATP, 2 mm phosphoenolpyruvate, 100 μg/ml lactate dehydrogenase, 100 μg/ml pyruvate kinase, 0.2 mm NADH, and 0.05% n-dodecyl-β-d-maltoside. Polyacrylamide gel electrophoresis in the presence of SDS or AES was carried out as described previously (9). The proteins were stained with Coomassie Brilliant Blue. For measurement of ATP synthesis activity, the ATPases were reconstituted into liposome with bacteriorhodopsin, and light-induced ATP synthesis activities were measured as described previously (21).

RESULTS

Reconstitution of VoV1 from Isolated Vo and V1

The Vo and V1 of T. thermophilus were isolated from membranes of T. thermophilus expressing His8-tagged VoV1 and further purified by gel permeation column chromatography. SDS-PAGE and AES-PAGE confirmed the purity of the subcomplexes (Figs. 2 and 3). Reconstitution of VoV1 was confirmed with AES-PAGE (Figs. 2 and 3A). With increasing concentrations of Vo, there is a steady decrease in the levels of free V1 and a concomitant increase in the levels of VoV1. At a concentration of 16 μg or above of Vo, all the V1 is reconstituted into VoV1 (Fig. 2). This result clearly indicates that the isolated Vo and V1 assemble into VoV1 with a very low dissociation constant (K).

FIGURE 2.

AES-PAGE analysis of V Reconstitution was performed by the addition of 20 μl of MOPDM buffer containing increasing amounts of Vo into 20 μl of 2 μm V1 solution. Following incubation, half of each test condition was loaded onto the gel. Proteins were stained by Coomassie Brilliant Blue. The amount of added Vo was, 0, 2, 4, 6, 8, 10, 12, 14, 16, 18, and 20 μg (lanes 1–11). Lane 12, VoV1 from T. thermophilus; lane 13, V1; lane 14, Vo. The lower band corresponds to free V1, and the upper band corresponds to the VoV1 complex.

FIGURE 3.

AES- and SDS-PAGE analysis of the reconstituted complexes. A and B, the isolated reconstituted complexes were subjected to AES (A) or SDS-PAGE (B). The proteins were stained by Coomassie Brilliant Blue. Lane 1, the reconstituted VoV1; lane 2, the reconstituted A3B3D-Vo; lane 3, the reconstituted A3B3-Vo; lane 4, the VoV1 from T. thermophilus; lane 5, the Vo from T. thermophilus; lane 6, V1 (A3B3DF); lane 7, A3B3D; lane 8, A3B3.

Reconstitution of A3B3D-Vo and A3B3-Vo Complex

The ability of the A3B3D and A3B3 subcomplexes to reconstitute with Vo was also assessed to determine whether the central stalk is necessary for reconstitution. The purity and subunit stoichiometry of the A3B3D and A3B3 subcomplexes were confirmed by SDS-PAGE and AES-PAGE (Fig. 3, A and B, lanes 7 and 8). When an excess of A3B3D was incubated with Vo, a new gel permeation peak was observed with the same retention time as VoV1 (data not shown). SDS-PAGE analysis revealed that the peak was composed of subunits A, B, C, D, I, E, G, and L (Fig. 3B). The A3B3D-Vo reconstituted complex migrated as a single band on AES-PAGE with the same mobility as VoV1. Gel permeation HPLC and AES-PAGE (Fig. 3) also showed that the A3B3 was also able to form a stable complex with Vo. SDS-PAGE analysis confirmed that the complex did not include the D and F subunits (Fig. 3B). The results clearly indicate that the central shaft subunits D and F are not essential for reconstitution of a stable complex between the A3B3 domain and Vo.

Properties of the Reconstituted Complexes

Both the ATP hydrolysis and the synthesis activity of the reconstituted complexes were investigated. The reconstituted VoV1 exhibited simple Michaelis-Menten kinetics with a K value of 420 ± 80 μm and a kcat of about 16 ± 2.0 s−1. These values are in the same range as those obtained for the wild-type enzyme (3). Fig. 4A shows the sensitivity of the reconstituted complex to inactivation by DCCD, a specific inhibitor that modifies a critical carboxylate in the proteolipid subunit. An ability to be inhibited by DCCD is an indication that an FoF1 complex is intact; if proton translocation and ATP hydrolysis are uncoupled due to damage of the functional connection between Fo and F1, ATPase activity is no longer sensitive to DCCD inhibition (22). The ATPase activity of reconstituted VoV1 was inhibited by DCCD and almost completely lost 30 min after the addition of the inhibitor (Fig. 4A). The reconstituted VoV1 was also capable of ATP synthesis activity; vesicles containing bacteriorhodopsin and the VoV1 synthesized ATP following illumination. The ATP synthesis activity of reconstituted VoV1 was comparable with that of wild VoV1 (Fig. 4B). These results indicate that the reconstituted VoV1 is functional.

FIGURE 4.

Enzymatic properties of the reconstituted complexes. A, DCCD sensitivity of the reconstituted complexes. Each reconstituted complex was incubated with 50 μm DCCD for 30 min in MOPDM buffer, and ATPase activity was measured as described under “Experimental Procedures.” In control experiments, the reconstituted complexes were treated in the same way with the exception that DCCD was not included. B, ATP synthesis activity of the reconstituted complexes in proteoliposomes. Open circles, VoV1; closed squares, the reconstituted VoV1; open squares, the A3B3D-Vo; gray circles, the A3B3-Vo. C, effect of ATP on the reconstituted complexes. The reconstituted complexes were incubated for 1 h at 25 °C in the absence (left) or presence (right) of 2 mm ATP, submitted to AES-PAGE analysis. Lane 1, the VoV1; lane 2, A3B3D-Vo; lane 3, A3B3-Vo; lane 4, A subunit; lane 5, B subunit; lane 6, Vo.

The turnover rate of A3B3D-Vo was ∼20 s−1 at the same range of activity of the A3B3D complex reported previously (13). ATP hydrolysis by A3B3-Vo proceeded in two distinct phases: an initial rapid phase and a slow steady state phase (∼10 s−1, Fig. 4A). In contrast to the V1 complex, the A3B3 subcomplex showed an initial rapid phase, and ATPase activity was gradually lost due to ATP-induced disassembly of the A3B3 subcomplex (23). The ATP hydrolysis profile of A3B3-Vo was mostly identical to that of the A3B3 subcomplex, suggesting that the association of Vo and A3B3 does not change the catalytic properties of A3B3 domain and that A3B3-Vo gradually disassembles during ATP hydrolysis. In fact, both reconstituted VoV1 and reconstituted A3B3D-Vo were resistant to the ATP-induced disassembly; however, the reconstituted A3B3-Vo disassembled into Vo and monomeric A and B subunits following the addition of ATP (Fig. 4C). This result indicates that the association of Vo and A3B3 does not prevent ATP-induced disassembly of the A3B3. Unlike the reconstituted VoV1, both the A3B3D-Vo and the A3B3-Vo did not show sensitivity to DCCD (Fig. 3A). In addition, no ATP synthesis activity was detected by reconstituted proteoliposome containing either A3B3D-Vo or A3B3-Vo (Fig. 4B). These results clearly indicate that the F subunit is essential for coupling ATP synthesis with proton translocation across the membrane.

Real-time Monitoring of Reconstitution by FRET

FRET is an excellent method for detecting protein association (24). To measure the reconstitution of VoV1 or A3B3D-Vo in real time, V1, A3B3D, and Vo were labeled with different fluorescent dyes. The subunit F in V1 (F-S54C) and the subunit D in A3B3D (D-E48C) were labeled with the maleimide derivative Cy3. For labeling of the subunit C in Vo, a cysteine was introduced at position 105, which was then labeled with Cy5 maleimide. The degree of labeling was ∼80% for V1, ∼95% for A3B3D, and 80% for Vo as determined by UV-visible spectroscopy of the labeled complexes. The specificity of each labeling was confirmed by SDS-PAGE (supplemental Fig. 1). Reconstitutions were carried out in a quartz cuvette containing 1.2 ml of the V1-F54C-Cy3 solution. Upon the addition of Vo-C105C-Cy5 to the V1-F54C-Cy3, the emission intensity at 570 nm decreased, and an alternative peak at 650 nm was detectable (Fig. 5A). The magnitude of the peak at 650 nm initially depended on the amount of the added Vo-C105C-Cy5. Fluorescence at 570 nm decreased sharply upon the addition of the Vo-C105C-Cy5 (Fig. 5B). In comparison, the addition of excess nonlabeled Vo into V1-F54C-Cy3 resulted in no decrease in fluorescence at 570 nm. These results clearly indicate that the decrease in fluorescence following the addition of Vo-C105C-Cy5 is due to reconstitution of VoV1. The rate of decrease in fluorescence at 570 nm and the associated increase in fluorescence at 650 nm following the addition of the acceptor is higher at 25 °C than at lower temperatures. Indeed there was no detectable decrease in fluorescence at 570 nm observed at 4 °C (Fig. 5C, black line). These results indicate that the association of V1 and Vo is temperature-dependent.

FIGURE 5.

Probing reconstitution of V A, fluorescence spectra of the donor at different concentrations of acceptor. Fluorescence spectra of V1-F54C-Cy3 were recorded (final concentration of 23 nm, excitation at 532 nm). Different concentrations of Vo-C105C-Cy5 (0, 5, 10, 15, 20, 25 nm final concentration) were added. The test samples were incubated for 10 min, and then the fluorescence spectra of each mixture were measured. A. U., arbitrary units. B, time course of the donor fluorescence. The black line shows the fluorescence obtained when 10 μl of 4 μm of Vo-C105C-Cy5 was added into a cuvette containing 23 nm V1-F54C-Cy3 at the time indicated by the black arrow. The red line shows the fluorescence obtained when 10 μl of 5 μm Vo was added into a cuvette containing 23 nm V1-F54C-Cy3 at the time indicated by the black arrow, and then a further 10 μl of 4 μm of Vo-C105C-Cy5 was added at the time indicated by the red arrow. The green line shows the fluorescence obtained when 10 μl of MOPDM buffer was added at the time indicated by the black arrow. C, temperature effect on the time course of donor fluorescence. 10 μl of 4 μm of Vo-C105C-Cy5 was added into a cuvette containing 23 nm V1-F54C-Cy3 at the indicated temperature.

To estimate the dissociation constant, K, for VoV1 into V1 and Vo, a range of concentrations of Vo-C105C-Cy5 (0.1–1.2 nm final concentration) was modified to a single concentration of V1-F54C-Cy3. The fluorescence intensity at 570 nm decreased with increasing concentration of Vo-C105C-Cy5 (Fig. 6A). The time-dependent changes were analyzed using a simple sequential model for the dissociation and association of the complexes, and the apparent rate constants were calculated by nonlinear regression fitting (25, 26) and plotted against the concentration of added Vo-C105C-Cy5 (Fig. 6C). The K was estimated to be 0.26 ± 0.23 nm (n = 7). The fluorescence of Cy3 in A3B3D-D48C-Cy3 was also decreased by the addition of Vo-C105C-Cy5 (Fig. 6B). The dissociation constant for the A3B3D-Vo was estimated to be 1.1 ± 0.54 nm (n = 4). The results suggest that the F subunit is not essential for association of A3B3 domain to Vo, but contributes to the stability of the complex.

FIGURE 6.

Typical time course of donor fluorescence in presence of V A and B, different concentrations of Vo-C105C-Cy5 as indicated in panels C and D were added into a cuvette with 1.2 ml of MOPDM buffer containing 80 pm V1-F54C-Cy3 (A) or 130 pm A3B3D-D48C-Cy3 (B), and fluorescence changes of the donor were measured. A. U., arbitrary units. C and D, apparent rate constants (kapp = kon[ATP] − koff) were determined by fitting the fluorescence decrease after the addition of Vo-C105C-Cy5 for V1-F54C-Cy3 (C) or A3B3D-D48C-Cy3 (D) with a single exponential equation (25, 26), plotted against ATP concentrations ([ATP]). From a linear fit to the plot, kon and koff were calculated as a slope and an intercept, respectively (25).

DISCUSSION

In this study, we demonstrated the reconstitution of VoV1 of T. thermophilus in vitro by mixing individually isolated Vo and V1 subcomplexes. The reconstitution was complete at equimolar concentrations of V1 and Vo, indicating that both the isolated V1 and the isolated Vo retain full reconstitution ability. The dissociation/association of VoV1 in prokaryotic cells has not been reported yet; however, our results suggest that the prokaryotic VoV1 is assembled by association of a cytosolic V1 with a membrane-embedded Vo in vivo. In addition, we have demonstrated real-time reconstitution of VoV1 by FRET. The K for VoV1 is estimated as ∼0.3 nm at 25 °C, giving a Gibbs free energy (ΔG°) of binding of V1 to Vo as ∼54 kJ/mol with ΔG° = −RTlnK. Because the rate of reconstitution increases at higher temperature (Fig. 5C), a much lower K is predicted for VoV1 in T. thermophilus cells, whose optimum growth temperature is 60–80 °C. The low K indicates that the equilibrium between association/dissociation of VoV1 might be biased toward association of V1 and Vo in the cells.

Surprisingly, A3B3D and A3B3 also associated with Vo, producing a holoenzyme-like complex lacking the F subunit or the central shaft DF subcomplex, respectively. Electron microscopic studies of VoV1 have indicated that the V1 domain was connected to the Vo domain with two (11) or three peripheral stalks (27), whereas the stator structure of bacterial FoF1 consists of a single peripheral stalk formed by the b subunit (28). The results presented in this study suggest that the A3B3 domain is tightly associated with the two EG peripheral stalks of Vo, even in the absence of the central shaft subunits. The crystal structure of the F1c10 of yeast (PDB ID: 3ZRY) indicated that the α3β3 domain binds tightly to the proteolipid ring through the central shaft subunit γ and ϵ (29). In contrast, T. thermophilus VoV1 has a subunit C located in the central stalk, which caps one end of the subunit-L ring with the internal cavity of subunit C open toward the V1 side to accommodate the shaft composed of the D and F subunits (18). The socket-like structure of the C subunit seems to be favorable for the reversible dissociation/association of the central shaft, but unfavorable for tight binding of the V1 central shaft with Vo. It is likely that the instability of the VoV1 holoenzyme due to the detachable central stalk might be compensated for by the binding of the two EG peripheral stalks to V1.

The reconstituted complex without subunit F hydrolyzed ATP, but did not show ATP synthesis activity or DCCD-sensitive ATPase activity. This indicates that ATP synthesis and proton translocation in the A3B3D-Vo were uncoupled due to complete loss of the functional connection between A3B3D and Vo. That is, intramolecular uncoupling of A3B3D-Vo should have occurred. In this case, the binding energy of the D subunit to the C subunit in the central stalk is thought to be much lower than the observed ΔG° for ATP synthesis, ∼40 kJ/mol at 25 °C (30). The estimated K for A3B3D-Vo is ∼1 nm, giving a ΔG° of binding of A3B3D to Vo as ∼50 kJ/mol, slightly lower than that of VoV1 (∼54 kJ/mol). This suggests that the F subunit reinforces the stability of whole complex, but the contribution of the F subunit to association of A3B3DF (V1) with Vo is lower than that of two EG peripheral stalks.