ATP-dependent Rotation of Mutant ATP Synthases Defective in Proton Transport*

During ATP hydrolysis, the γϵc10 complex (γ and ϵ subunits and a c subunit ring formed from 10 monomers) of F0F1 ATPase (ATP synthase) rotates relative to the α3β3δab2 complex, leading to proton transport through the interface between the a subunit and the c subunit ring. In this study, we replaced the two pertinent residues for proton transport, cAsp-61 and aArg-210 of the c and a subunits, respectively. The mutant enzymes exhibited lower ATPase activities than that of the wild type but exhibited ATP-dependent rotation in planar membranes, in which their original assemblies are maintained. The mutant enzymes were defective in proton transport, as shown previously. These results suggest that proton transport can be separated from rotation in ATP hydrolysis, although rotation ensures continuous proton transport by bringing the cAsp-61 and aArg-210 residues into the correct interacting positions.

ATP is synthesized by ATP synthase (F 0 F 1 ) coupled with an electrochemical proton gradient established by the respiratory chain (see Refs. 1-6 for reviews). F 0 F 1 reversibly hydrolyzes ATP to form a proton gradient. The Escherichia coli enzyme consists of catalytic sector F 1 and proton pathway F 0 , formed from five ␣ 3 ␤ 3 ␥␦⑀ and three ab 2 c 10 subunits, respectively, with unique stoichiometries. The basic subunit structure of bacterial F 0 F 1 , which is implicated in regulation, is similar to that of mitochondria, although the latter enzyme has additional subunits. Three ␣ and three ␤ subunits form an ␣ 3 ␤ 3 hexamer, the ␥ subunit being in the central space, as shown by the x-ray structure (7).
Early studies (8 -10) suggested a ring structure for c subunit oligomers (c-ring), with ab 2 located outside the ring. More recently, the crystal structure of mitochondrial F 1 with a c-ring formed from 10 monomers was reported (11). A series of combined genetic and biochemical studies suggested that the E. coli c-ring is formed from 10 monomers (12). Similar results were obtained recently for Bacillus c subunits (13). The c-ring structure may differ depending on the species, because the chloroplast (14) and Ilyobacter tartaricus (15) rings are formed from 14 and 11 monomers, respectively.
Three catalytic sites (each in a ␤ subunit) participate alter-nately in ATP synthesis or hydrolysis, as predicted from the binding change mechanism (3). The first reported crystal structure of mitochondrial F 1 (7) is consistent with this mechanism and suggests ␥ subunit rotation, which has been shown successively by studies involving electron microscopy (16), chemical cross-linking (17), and photo-bleaching of a probe (18). Furthermore, an actin filament connected to the ␥ subunit rotated continuously in the Bacillus ␣ 3 ␤ 3 ␥ complex (19). A ␥ rotation in E. coli F 1 has also been shown previously (20), prompting mutational studies on the mechanism (20,21).
Rotation of the ␥ subunit should be transmitted to F 0 for proton translocation during ATP hydrolysis, and reversibly, protons transported through F 0 should drive ␥ rotation during ATP synthesis. The rotor in the entire F 0 F 1 was proposed to comprise an assembly of the c-ring with the ␥ and ⑀ subunits (3,(22)(23)(24). Sambongi et al. (25) and later Pä nke et al. (26) showed that the c-ring rotates continuously together with the ␥ subunit in the purified F 0 F 1 , consistent with the tight structural connection between the ␥ subunit and the c-ring (11). Similarly, the ␣ 3 ␤ 3 complex rotated when F 0 F 1 was immobilized through the c-ring (27). Rotation of the subunit complex was observed in planar membranes without any chemical treatment (28). The c-ring rotation relative to the a subunit has also been demonstrated in membranes. These studies established that the ␥⑀c 10 complex rotates relative to the ␣ 3 ␤ 3 ␦ab 2 (29). A similar approach has been used to show subunit rotation in vacuole-type ATPases (30) found in organelles such as lysosomes and endosomes (31,32).
A model for the proton pathway in F 0 has been proposed based on biochemical and genetic results (33)(34)(35)(36)(37)(38)(39)(40). A proton binding site, the cAsp-61 residue (E. coli) located in the center of the hydrophobic helix of the c subunit, undergoes protonation or deprotonation by interacting with aArg-210 located in helix 4 of the a subunit (38 -40). Fillingame and co-workers (37,41) have shown that transmembrane helices 2, 4, and 5 of the a subunit form two half-water channels from the periplasm to cAsp-61 and from cAsp-61 to the cytoplasm.
The c-ring rotation may be tightly coupled to protonation and deprotonation of cAsp-61 and may exhibit a 2/10 step for continuous proton transport during ATP hydrolysis or synthesis. Thus, it became of interest to examine c-ring rotation when mutations are introduced at proton pathway residues including cAsp-61 and aArg-210. In this study, we examined the ATP hydrolysis-dependent rotation of mutant F 0 F 1 with these residues replaced.

EXPERIMENTAL PROCEDURES
Bacterial Strain, Growth Conditions, and Plasmids-The plasmid pBUR13D constructed from pBWU13 (27) carried modified c and ␤ subunit genes with a His 6 tag (six His residues) and a biotin-binding * This work was supported by grants-in-aid from the Ministry of Education, Science, and Culture of Japan, the Takeda Science Foundation, the Japan Foundation for Applied Enzymology, and the Noda Foundation. 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.
Membrane Preparation and Rotation Assay-Membranes were prepared from logarithmic phase cells disrupted with a French press (28). A flow cell was constructed from nitrocellulose-coated cover glasses and filled with buffer A (10 mM HEPES-NaOH, pH 7.8, 25 mM KCl, 5 mM MgCl 2 ) containing 0.8 M Ni 2ϩ -nitrilotriacetate-horseradish peroxidase conjugate. The cell was washed at 25°C with buffer A containing 10 mg/ml bovine serum albumin, membrane fragments (ϳ1-3 mg of protein/ml) were introduced, and the actin filaments were connected to the ␤ subunits through the biotin-binding domain (28). Immediately after ATP addition, a 1-mm 2 area was scanned under a Zeiss Axiovert 135 equipped with an intensified charge-coupled device camera, and rotating filaments were examined as described previously (28). Buffer A was included in all solutions used. When necessary, rotation was assayed by a person who did not know the properties of the membranes introduced into the flow cell by a second person.
Gel Electrophoresis and Immunoblotting-Wild-type or mutant membranes (10 g of protein) were subjected to 15% polyacrylamide gel electrophoresis in the presence of 0.1% sodium dodecyl sulfate followed by blotting onto nitrocellulose or polyvinylidene difluoride membranes (42). The biotin-binding domain connected to the ␤ subunit was detected with streptavidin-alkaline phosphatase conjugate and 5-bromo-4-chloro-3-indoxylphosphate/nitroblue tetrazolium. Protein bands were visualized (26), and the relative amounts of mutant subunits were estimated from the intensity of bands using the standard curves for the wild-type subunits. The blotting efficiencies were similar regardless of the mutations. A similar procedure was used to detect subunits in whole cells.
Other Procedures-ATPase activity, the formation of an electrochemical proton gradient, and the protein concentration were determined as described previously (27). Fluorescent biotinylated actin filaments were prepared by the published procedure (43). The chemicals used were of the highest grade commercially available.

Properties of Membranes with Mutations in cAsp-61 and
aArg-210 -The F 0 F 1 engineered for observing rotation had a His 6 tag and a biotin-binding domain connected to the c and ␤ subunits, respectively. We performed cD61N, cD61G, aR210K, aR210Q, or aR210A replacement for the engineered F 0 F 1 . E. coli cells with the mutant enzymes were unable to grow on succinate through oxidative phosphorylation and grew slower on glucose than the wild type (data not shown), exhibiting essentially the same characteristics as the corresponding original mutations in nonengineered F 0 F 1 (38,40).
The amounts of F 0 F 1 mutant membranes were estimated using streptavidin-alkaline phosphatase conjugate for the biotin-binding domain connected to the ␤ subunit (Fig. 1a). The amounts in the cD61N and cD61G mutants comprised 30 and 4% that of the wild type, respectively, and the amounts of F 0 F 1 in the aR210A, aR210K, and aR210Q mutants were 100, 100, and 50% that of the wild type, respectively. The decreases in the mutant enzymes were at least partly a result of their lower expression, because the relative amounts of mutant F 0 F 1 s in whole cells were similar to those in membranes (Fig. 1b).
ATPase Activities of Subunit a and c Mutants-Consistent with the lower amounts of F 0 F 1 , the mutant membrane ATPase activities were ϳ5-60% that of the wild type (Table I).The relative turnover rates of mutant membrane ATPases were estimated from the ATPase activities and amounts of F 0 F 1 . The rate for cD61N F 0 F 1 was 15% that of the wild type, whereas the rate for cD61G was 41% that of the wild type (Table I). The rates for subunit a mutants were also estimated, and the values for aR210K, aR210Q, and aR210A were 49%, 6%, and 21% that for the wild type, respectively. These results suggest that mutations in the F 0 sector affect ATPase activity of F 0 F 1 .
The ATPase activities of mutant F 1 sectors were increased ϳ1.4 -4.3-fold after solubilization. Thus, ϳ30 -80% of the activity was inhibited in the mutant membranes, whereas that of the wild type was not (Table I, F 1 ATPase inhibition in membranes). The cD61N and cD61G membranes did not show ATPdependent proton translocation, although they still exhibited FIG. 1. Amounts of F 0 F 1 s in mutant membranes. a, detection of the F 1 ␤ subunit in membranes. Mutant membranes (cD61N, cD61G, aR210K, aR210Q, and aR210A) (10 g of protein) were subjected to polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate and then subjected to immunoblotting with streptavidin-alkaline phosphatase conjugate to detect the biotin-binding domain connected to the ␤ subunit. b, detection of the F 1 ␤ subunit in whole cells. Wild type (WT) and mutant cells (60 g of protein) were treated with 2% sodium dodecyl sulfate and then subjected to polyacrylamide gel electrophoresis and immunoblotting.

TABLE I ATPase activities of mutant F 0 F 1 s in membranes
Membranes were prepared from logarithmic phase cells, and then their ATPase activities were assayed at 37°C and are shown with standard deviation. ATPase activity was also assayed after solubilization of F 1 from mutant or wild-type membranes with dilute buffer (49), and "F 1 ATPase inhibition in membranes" was estimated from the membrane and solubilized ATPase activities, "F 1 ATPase activity solubilized" is expressed as units/mg of protein in membrane before wash. After scanning the Western blot (Fig. 1a), the relative amounts of membrane F 0 F 1 were estimated using the wild-type ␤ subunit (with a biotin-binding domain) as a standard, and the values in the table are averages of the three preparations. "Relative turnover rate" was estimated from the levels of F 0 F 1 (Fig. 1) and membrane ATPase activities (above). Membranes were incubated at 25°C for 10 min in the presence of various concentrations of DCCD and then assayed for ATPase activity at 25°C. The inhibition with 50 M DCCD is shown (DCCD inhibition).

Mutation
Membrane proton transport dependent on respiration (Fig. 2). These results indicate that the mutant F 0 sectors could not transport protons but that their membranes were not leaky to protons. As expected, mutant membranes depleted of F 1 did not show proton permeability. The a subunit mutants (aR210K, aR210Q, and aR210A) gave similar results. These results confirmed that mutant F 0 sectors are defective in proton transport and that cAsp-61 and aArg-210 are pertinent residues.
Rotation of an Actin Filament Connected to Mutant F 0 F 1 -The ATPase activities of cAsp-61 and aArg-210 mutants prompted us to assay their rotation, because defects in proton translocation may affect rotation. Membranes were immobilized on a glass surface through a His 6 tag connected to the c-ring, and then an actin filament was attached to the ␤ subunit. The wild-type or mutant (cD61N, cD61G, aR210K,  aR210Q, or aR210A) filament rotated continuously upon ATP addition (Fig. 3). The mutant time courses were similar to that An actin filament connected to the ␤ subunit of a mutant membrane F 0 F 1 was examined for rotation. The time courses of the actin filament (ϳ0.7-2 m) rotation are shown: wild type, black; cD61G, blue; cD61N, purple; aR210K, green; aR210Q, red; and aR210A, yellow. Rotation was analyzed as described previously (20). of the wild type when ϳ0.7-2-m filaments were compared. The mutants generated essentially the same torque as the wild type (Fig. 4).
The effect of DCCD on rotating F 0 F 1 is hard to study because it does not instantly react with cAsp-61 (27)(28)(29). To overcome this difficulty, we incubated membranes with DCCD and then examined them for rotation (28). Numbers of rotating filaments were decreased by 80% when DCCD was included to wild-type membranes, confirming that rotation of wild-type F 0 F 1 is sensitive to DCCD (28). Similar numbers of rotating filaments were observed in cD61N, cD61G, aR210Q, and aR210A membranes, regardless of DCCD treatment. These results indicate that the mutant rotation was not sensitive to DCCD. The number of rotating filaments became ϳ20% lower in aR210K with DCCD. However, it was difficult to obtain the exact value expected from 35% ATPase inhibition in this mutant (Table I). The difficulty may be because of the large deviations for finding rotating filaments. DISCUSSION Rotational catalysis by ATP synthase F 0 F 1 and vacuolartype proton ATPase has been established (1-6). As a membrane-bound holoenzyme, the complex of ␥⑀c 10 of F 0 F 1 rotates relative to ␣ 3 ␤ 3 ␦ab 2 depending on ATP hydrolysis (28). The rotation should drive proton transport through protonation/ deprotonation of the cAsp-61 residue of the c-ring upon interaction with aArg-210 of the a subunit (4). An important question is how the ATPase activity, c-ring rotation, and proton transport are coupled. It has been shown that chemical or genetic modification of cAsp-61 inhibits proton transport (4). Consistent with its binding to the cAsp-61 residue, DCCD inhibits the ATPase activity and proton transport (27). ϳ30 -80% of the F 1 ATPase activity was inhibited when F 1 was bound to mutant F 0 in membranes, as estimated from the increase in ATPase activity after detachment from the membranes. Previous studies also showed that the ATPase activity of F 1 bound to cD61N or aR210Q F 0 was inhibited by 50% (39,44). As the mutations were not in the catalytic F 1 sector, its binding to the mutant F 0 caused an inhibition of ATP hydrolysis. Thus, it is tempting to assume that the lack of proton transport through F 0 lowered the ATPase activity of F 1 .
The aArg-210 mutations gave variable effects on DCCD sensitivity and F 1 ATPase inhibition. Significant DCCD inhibition was observed in aR210K, confirming previous results (40), whereas aR210Q and aR210A were essentially insensitive to the chemical. These results may suggest that the interaction between a and c affects the inhibition, because aArg-210 mutants have a DCCD binding site (cAsp-61).
To address the mechanism underlying the coupling between rotational catalysis and proton transport in a distinct part of the enzyme complex, we analyzed F 0 F 1 in planar membranes. Mutant F 0 F 1 with replacement of the cAsp-61 or aArg-210 residue showed ATP hydrolysis-dependent rotation, generating similar torque to the wild type. These results indicate that the rotation could be separated from the proton transport in ATP hydrolysis. Thus, during the interaction with aArg-210, possibly on a picosecond time scale, cAsp-61 could undergo protonation/deprotonation for proton transport (4,36), and the rotation, on a millisecond time scale, enables continuous transport.
For ATP synthesis, an electrochemical proton gradient should support protonation/deprotonation of cAsp-61, leading to c-ring rotation, which should be obligatorily coupled to proton transport in ATP synthesis, as evidenced by the cross-linking (45). Thus, it is reasonable to assume that the c-ring rotates through apparently different mechanisms in ATP synthesis and hydrolysis. Direct testing of this assumption will depend on the experimental systems for observing continuous rotation dependent on an electrochemical proton gradient. The method recently developed may contribute to such direction (46).
Gumbiowski et al. (47) observed that purified cD61N mutant F 0 F 1 could rotate upon ATP hydrolysis. The results of their experiments carried out with a detergent-solubilized enzyme are consistent with our results.
Early genetic studies showed that the wild-type c subunit gene carried by transducing phages or low copy number plasmids could not complement the c subunit mutants with cAsp-61 replacement (48), suggesting that a chimeric c-ring consisting of the mutant and wild-type c subunits was inactive as to proton translocation. More recently, the Bacillus c subunit decamer, which covalently connected nine wild-type and one mutant c with Gln at position Glu-56 (corresponding to E. coli cAsp-61), exhibited no ATP-dependent proton pumping activity 1 The abbreviation used is: DCCD, N,NЈ-dicyclohexylcarbodiimide.   (13). These results suggest that the glutamate residues of all c subunits forming the ring are functionally necessary. Our results could predict that the decamer can rotate depending on ATP hydrolysis, and the subunit a arginine residue can interact continuously with Glu-56 located in the nine c subunits of the decamer. Thus, it is not easy to understand why no proton transport was observed for the mutant decamer. The inability may suggest the presence of a covalently connected wild-type and mutant c in the Bacillus decamer altered c-ring structure for proton translocation.
In conclusion, the proton transport and c-ring rotation during ATP hydrolysis could be separated mechanically by replacing cAsp-61 or aArg-210. Thus, the c-ring rotation is not obligatorily coupled to proton transport, although the rotation should ensure uninterrupted proton transport.