Structural asymmetry of F1-ATPase caused by the gamma subunit generates a high affinity nucleotide binding site.

The α3β3γ and α3β3 complexes of F1-ATPase from a thermophilic Bacillus PS3 were compared in terms of interaction with trinitrophenyl analogs of ATP and ADP (TNP-ATP and TNP-ADP) that differed from ATP and ADP and did not destabilize the α3β3 complex. The results of equilibrium dialysis show that the α3β3γ complex has a high affinity nucleotide binding site and several low affinity sites, whereas the α3β3 complex has only low affinity sites. This is also supported from analysis of spectral change induced by TNP-ADP, which in addition indicates that this high affinity site is located on the β subunit. Single-site hydrolysis of substoichiometric amounts of TNP-ATP by the α3β3γ complex is accelerated by the chase addition of excess ATP, whereas that by the α3β3 complex is not. We further examined the complexes containing mutant β subunits (Y341L, Y341A, and Y341C). Surprisingly, in spite of very weak affinity of the isolated mutant β subunits to nucleotides (Odaka, M., Kaibara, C., Amano, T., Matsui, T., Muneyuki, E., Ogasawara, K., Yutani, K., and Yoshida, M.(1994) J. Biochem. (Tokyo) 115, 789-796), a high affinity TNP-ADP binding site is generated on the β subunit in the mutant α3β3γ complexes where single-site TNP-ATP hydrolysis can occur. ATP concentrations required for the chase acceleration of the mutant complexes are higher than that of the wild-type complex. The mutant α3β3 complexes, on the contrary, catalyze single-site hydrolysis of TNP-ATP rather slowly, and there is no chase acceleration. Thus, the γ subunit is responsible for the generation of a high affinity nucleotide binding site on the β subunit in F1-ATPase where cooperative catalysis can proceed.

The ␣ 3 ␤ 3 ␥ and ␣ 3 ␤ 3 complexes of F 1 -ATPase from a thermophilic Bacillus PS3 were compared in terms of interaction with trinitrophenyl analogs of ATP and ADP (TNP-ATP and TNP-ADP) that differed from ATP and ADP and did not destabilize the ␣ 3 ␤ 3 complex. The results of equilibrium dialysis show that the ␣ 3 ␤ 3 ␥ complex has a high affinity nucleotide binding site and several low affinity sites, whereas the ␣ 3 ␤ 3 complex has only low affinity sites. This is also supported from analysis of spectral change induced by TNP-ADP, which in addition indicates that this high affinity site is located on the ␤ subunit. Single-site hydrolysis of substoichiometric amounts of TNP-ATP by the ␣ 3 ␤ 3 ␥ complex is accelerated by the chase addition of excess ATP, whereas that by the ␣ 3 ␤ 3 complex is not. We further examined the complexes containing mutant ␤ subunits (Y341L, Y341A, and Y341C). Surprisingly, in spite of very weak affinity of the isolated mutant ␤ subunits to nucleotides (Odaka, M., Kaibara, C., Amano, T., Matsui, T., Muneyuki, E., Ogasawara, K., Yutani, K., and Yoshida, M. (1994) J. Biochem. (Tokyo) 115, 789 -796), a high affinity TNP-ADP binding site is generated on the ␤ subunit in the mutant ␣ 3 ␤ 3 ␥ complexes where single-site TNP-ATP hydrolysis can occur. ATP concentrations required for the chase acceleration of the mutant complexes are higher than that of the wild-type complex. The mutant ␣ 3 ␤ 3 complexes, on the contrary, catalyze single-site hydrolysis of TNP-ATP rather slowly, and there is no chase acceleration. Thus, the ␥ subunit is responsible for the generation of a high affinity nucleotide binding site on the ␤ subunit in F 1 -ATPase where cooperative catalysis can proceed.
F 1 is a hydrophilic portion of H ϩ -ATP synthase and can catalyze hydrolysis of ATP. Isolated F 1 -ATPase is comprised of five different subunits in a stoichiometry ␣ 3 ␤ 3 ␥␦⑀. Each of isolated ␣ and ␤ subunits can bind one nucleotide, but neither of them has ATPase activity (Yoshida et al., 1977a;Dunn and Futai, 1980;Ohta et al., 1980;Issartel and Vignais, 1984;Hisabori et al., 1986;Rao et al., 1988;Bar-Zvi et al., 1992). The crystal structure of bovine mitochondrial F 1 -ATPase (MF 1 ) 1 (Abrahams et al., 1994) revealed that the ␣ and ␤ subunits have a similar fold alternating in a hexagonal arrangement around a central cavity containing the ␥ subunit as expected from previous electron microscopic studies (Gogol et al., 1989;Fujiyama et al., 1990), and catalytic nucleotide binding sites reside mostly on ␤ subunits whereas noncatalytic nucleotide binding sites are mostly on ␣ subunits. The ␥ subunit has at least three stretches of ␣ helices, and two of them form a coiled coil structure that penetrates the central cavity of the ␣ 3 ␤ 3 structure.
Penefsky and his colleagues showed that bovine MF 1 has a single high affinity ATP binding site Cross et al., 1982). ATP added at substoichiometric amount binds rapidly to this site and is hydrolyzed slowly (single-site or "uni-site" catalysis). This slow hydrolysis is greatly accelerated by the addition of excess ATP (chase acceleration). However, in the case of F 1 -ATPase from a thermophilic Bacillus PS3 (TF 1 ), the single-site hydrolysis of ATP is much faster than that of MF 1 , and only a very poor chase acceleration is observed . Nonetheless, steady state ATP hydrolysis by TF 1 does not obey simple Michaelis-Menten type kinetics but shows apparent cooperativity (Wong et al., 1984;Yokoyama et al., 1989). The attenuated incorporation of water oxygen into the product P i during hydrolysis of increasing concentrations of ATP by TF 1 also provided a support for cooperative nature of the catalysis (Kasho et al., 1989). We found that TF 1 hydrolyzed a substoichiometric amount of trinitrophenyl adenosine triphosphate (TNP-ATP) slowly, and this slow hydrolysis was accelerated by the chase addition of excess ATP (Hisabori et al., 1992). These observations have lead us to the conclusion that TF 1 , as well as F 1 from other sources, has a functional asymmetry among the three catalytic sites, although TF 1 molecules, different from F 1 from other sources, do not contain any endogenously bound nucleotide. Moreover, from the analysis of the nucleotide binding of the TF 1 ⅐ADP 1:1 complex, we suggested the functional asymmetry among the three catalytic sites (Hisabori et al., 1994).
TF 1 is unique in its ability to reconstitute from each isolated subunit. Complexes with various combinations of subunits, such as ␣ 3 ␤ 3 ␥ and ␣ 3 ␤ 3 ␦ (Yoshida et al., 1977a;Yokoyama et al., 1989) and ␣ 3 ␤ 3 (Miwa and Yoshida, 1989;Kagawa et al., 1989) have been characterized. Most recently, the minimum catalytic unit, whose most probable subunit composition was ␣ 1 ␤ 1 , was reconstituted on the solid support (Saika and Yoshida, 1995). As expected, this ␣ 1 ␤ 1 complex shows simple Michaelis-Menten type kinetics when it hydrolyzes ATP. Therefore a critical question has arisen: is the kinetics of the ␣ 3 ␤ 3 complex cooperative or not? Because the structure of this complex should be a perfect 3-fold symmetry, all three catalytic * This work was supported by the Hayashi Memorial Foundation for Female Natural Scientists (to C. K.) and by Grants-in-Aid for Scientific Research on Priority Areas 04266103 and 05266103 from the Ministry of Education, Science, Sports, and Culture of Japan (to M.Y.). 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.: 81-45-924-5233; Fax: 81-45-924-5277; E-mail: myoshida@res.titech.ac.jp. 1 The abbreviations used are: MF 1 , F 1 -ATPase from mitochondrial inner membrane; TF 1 , F 1 -ATPase from thermophilic Bacillus strain PS3; TNP-ATP and TNP-ADP, the 2Ј,3Ј-O-(2,4,6-trinitrophenyl) deriv-atives of ATP and ADP, respectively; PIPES, piperazine-N,NЈ-bis(2ethanesulfonic acid); Tricine, N-[2-hydroxy-1,1-bis(hydroxymethyl)ethyl]glycine; HPLC, high pressure liquid chromatography. sites, as well as three noncatalytic sites, should be equivalent. If catalysis by this complex obeys noncooperative kinetics, then one can conclude that kinetic cooperativity observed for F 1 -ATPase might be due to structural asymmetry caused by the presence of single-copy subunits. If it is cooperative, this means that structural asymmetry introduced by binding of the first substrate is responsible for the kinetic complication of F 1 -ATPase. However, answering the above question by experiment has turned out to be not as easy as it seemed at first. We had previously reported that the ␣ 3 ␤ 3 complex exhibited cooperative kinetics (Miwa and Yoshida, 1989), but this observation could be interpreted in a different way because Harada et al. found that the complex tended to dissociate into ␣ 1 ␤ 1 complexes in the presence of ATP (Ͼ6 M) or ADP (Ͼ30 M) (Harada et al., 1991). Here, we report that TNP-ATP (or TNP-ADP) does not destabilize the ␣ 3 ␤ 3 complex, and this provides an opportunity to examine the kinetic properties of the ␣ 3 ␤ 3 complex. Comparison of the characteristics of the ␣ 3 ␤ 3 ␥ complex and those of the ␣ 3 ␤ 3 complex indicates that the structural asymmetry of F 1 -ATPase introduced by the ␥ subunit to the ␣ 3 ␤ 3 structure is responsible for generation of a high affinity nucleotide binding site where single-site hydrolysis can occur. In addition, the complexes containing mutant ␤ subunits, whose Tyr-341 was replaced by Leu, Ala, or Cys, were also compared. Although the affinity of isolated mutant ␤ subunits to nucleotide are greatly diminished (Odaka et al., 1994), the binding of substoichiometric TNP-ATP (or TNP-ADP) to the mutant ␤ subunits returned to normal when they were assembled into the ␣ 3 ␤ 3 ␥ structure.

EXPERIMENTAL PROCEDURES
Preparation of the Subunit Complexes-The ␣ and ␤ subunits of TF 1 were expressed in Escherichia coli strain DK8 (bglR, thi-1, rel-1, HfrPO1, ⌬(uncB-uncC) ilv::Tn10) (Ohta et al., 1988) and purified as described previously (Ohtsubo et al., 1987). To isolate the ␣ 3 ␤ 3 complex, each subunit was precipitated by ammonium sulfate, dissolved in the minimum volume of 50 mM Tris-SO 4 buffer (pH 8.0), and mixed. After incubation at 30°C for 30 min, the solution was loaded on a gel filtration HPLC column (TSK G3000SWXL) equilibrated with 10 mM PIPES and 0.2 M Na 2 SO 4 (pH 7.0) (PIPES-Na 2 SO 4 buffer), and fractions containing pure ␣ 3 ␤ 3 complex were collected (Kaibara et al., 1993). The wild-type and mutant ␣ 3 ␤ 3 ␥ complexes were over-expressed as the complex in E. coli strain JM103⌬uncB-D and purified as described previously (Matsui and Yoshida, 1995). Expression vectors for mutant complexes were made from that of the wild-type complex by exchanging a SmaI-MluI fragment obtained from the vectors to express mutant ␤ subunits (Odaka et al., 1990). The purified ␣ 3 ␤ 3 ␥ complexes contained no or very little bound nucleotide (Ͻ0.1 mol/mol). The concentrations of proteins of the wild-type and mutant complexes were determined by measuring absorbance at 280 nm. The factor 0.45 at 280 nm as 1 mg/ml (Yoshida et al., 1977a) was used. The effect of replacement of Tyr-341 by other residues on the above factor was small and neglected.
Stability of the ␣ 3 ␤ 3 Complex-10 l of the ␣ 3 ␤ 3 complex (3.2 M) were preincubated for 1 min at 25°C and subjected to gel filtration HPLC (TSK G3000SWXL). The solutions used for preincubation, equilibration, and elution of the HPLC column were all the same: PIPES-Na 2 SO 4 buffer containing indicted concentrations of nucleotide and MgSO 4 . The flow rate was 0.5 ml/min, and elution was monitored by measuring the absorbance at 280 nm. The maximum concentration of TNP-ATP (or TNP-ADP) tested was 50 M due to high base-line absorbance.
Measurement of the Nucleotide Binding-The binding capacity of nucleotides on the subunit complexes were measured directly by equilibrium dialysis and indirectly by difference absorption spectrum induced by the interaction between the complex and nucleotides. Equilibrium dialysis was carried out as described by Hisabori and Sakurai (1984). Sample cells contained 0.02-0.5 M ␣ 3 ␤ 3 ␥ complex in 20 mM Tricine-NaOH (pH 8.0) or 0.36 -0.81 M ␣ 3 ␤ 3 complex in the PIPES-Na 2 SO 4 buffer. The opposite side of the cells contained various concentrations of TNP-ADP in 20 mM Tricine-NaOH (pH 8.0), 200 mM NaCl, and 2 mM MgCl 2 (␣ 3 ␤ 3 ␥ complex) or PIPES-Na 2 SO 4 buffer containing 2 mM MgCl 2 (␣ 3 ␤ 3 complex). After dialysis, solutions in both sides of the cell were treated with perchloric acid, and the amount of TNP-ADP present in the supernatant fraction was measured. The amounts of TNP-ADP bound on the complex were determined as the difference between the amounts of TNP-ADP in the cell containing protein and those in the opposite side. Difference absorption spectrum were measured at room temperature with a double-beam spectrophotometer model UV-2200 (Shimadzu Co., Kyoto, Japan) using a double-sector cuvette according to the method described by Hisabori et al. (1986). Difference spectra were measured 5 min after mixing.
TNP-ATP Hydrolysis under the Single-site Catalysis Condition-Single-site catalysis was measured as follows at 25°C according to the method described by Hisabori et al. (1992). 50 l of 1 M subunit complex were added to 50 l of 0.3 M TNP-ATP to initiate the reaction. At the indicated time, 5 l of ice-cold 24% perchloric acid were added to terminate the reaction (acid quench). To measure the ATP chase, 50 l of 10 mM (␣ 3 ␤ 3 ␥ complex) or 3 M (␣ 3 ␤ 3 complex) of ATP were added at the indicated time instead of perchloric acid. After 5 s, the reaction was terminated by addition of 5 l of ice-cold 24% perchloric acid. Quantitative analysis of TNP-ATP and TNP-ADP were carried out with HPLC according to the method described by Hisabori et al. (1992).

Stability of ␣ 3 ␤ 3 Complexes in TNP-AT(D)P-
The ␣ 3 ␤ 3 complex of TF 1 dissociates into ␣ 1 ␤ 1 complexes in the presence of ATP (Ͼ6 M) and ADP (Ͼ30 M) (Harada et al., 1991). However, TNP-ATP did not induce dissociation of the ␣ 3 ␤ 3 complex. As shown in Fig. 1A, when the ␣ 3 ␤ 3 complex was incubated with TNP-ATP and analyzed on a gel filtration HPLC column equilibrated with the buffer containing the same concentration of TNP-ATP, the retention times of the protein peaks remained unchanged, suggesting that no dissociation took place. TNP-ADP also did not induce dissociation (Fig. 1B). When the ␣ 3 ␤ 3 complex was preincubated with 20 M TNP-ATP at first and then applied to a HPLC column equilibrated with 10 M ATP, most of the ␣ 3 ␤ 3 complex remained intact (Fig. 1C). However, complete dissociation was observed at 20 M ATP for the complex preincubated with 20 M TNP-ATP. Thus, different from ATP, TNP-ATP (or TNP-ADP) is a "safe" nucleotide that does not induce destabilization of the ␣ 3 ␤ 3 complex. Taking this FIG. 1. Stability of the ␣ 3 ␤ 3 complex in the presence of TNP-ATP (or TNP-ADP). A, the ␣ 3 ␤ 3 complex was preincubated for 1 min in the solution containing the indicated concentrations of TNP-ATP and subjected to gel filtration HPLC. The column was equilibrated and eluted with the same solution used for preincubation. The elution was monitored by measuring the absorbance at 280 nm. B, the same as A except that TNP-ADP was used instead of TNP-ATP. C, the same as A except that the ␣ 3 ␤ 3 complex was preincubated with 10 M TNP-ATP and applied to a gel filtration column equilibrated with the solution containing the indicated concentrations of ATP. All nucleotide solutions contained MgSO 4 at concentrations equal to those of nucleotides. advantage, following experiments were carried out.
Binding of TNP-ADP to the ␣ 3 ␤ 3 ␥ and ␣ 3 ␤ 3 Complexes-It is clear from the results of equilibrium dialysis that the ␣ 3 ␤ 3 ␥ complex has at least two classes of binding sites for TNP-ADP with different affinity; a high affinity binding site with a K d value in the low nM range and several low affinity sites with K d values in the M range (Fig. 2, closed circles). On the contrary, the ␣ 3 ␤ 3 complex has only low affinity binding sites with K d in the M range, and no high affinity binding site was observed (Fig. 2, open circles). Therefore, the ␥ subunit in the ␣ 3 ␤ 3 ␥ complex is responsible for generating a single high affinity nucleotide binding site.
Subunit Location of the High Affinity Nucleotide Binding Site-It has been known that the difference absorption spectra induced by the interaction between TNP-ATP (or TNP-ADP) and the isolated ␣ or ␤ subunit are characterized with a trough at 450 nm and a peak at 510 nm (␣ subunit) or a trough at 395 nm and a peak at around 420 nm (␤ subunit), respectively (Hisabori et al., 1992). We can determine the subunit location of the TNP-ADP binding site of the complex by this means. When TNP-ADP at a 0.25 molar ratio was added to ␣ 3 ␤ 3 ␥ complex, the shape of the induced difference spectrum was almost the same as the one observed for the isolated ␤ subunit, indicating that the binding of TNP-ADP occurred to the site on one of three ␤ subunits in the complex (Fig. 3, top traces and the trace that is second from the top). For the ␣ 3 ␤ 3 complex, the shape of the difference spectrum was not the one typical for the ␤ subunit, and the magnitude of the spectrum was far smaller than that of the ␣ 3 ␤ 3 ␥ complex under the same condition (data not shown). This small magnitude of the spectrum agrees with a weak affinity of the ␣ 3 ␤ 3 complex to TNP-ADP indicated from equilibrium dialysis (Fig. 2). When we measured a series of the difference spectra induced by each step-wise addition of TNP-ADP to the ␣ 3 ␤ 3 ␥ complex, the shapes of the spectra induced by each addition were similar to the one observed for the isolated ␤ subunit until the concentration of added TNP-ADP reached a 1.25:1 molar ratio to the complex (Fig. 3). When the molar ratio exceeds 1.5:1, the shape of spectra induced by each addition of TNP-ADP becomes similar to the one observed for the isolated ␣ subunit, a peak at 510 nm. These results clearly show that TNP-ADP binds at first to a high affinity site on the ␤ subunit, and after this site is filled it starts to bind to a site on the ␣ subunit.
Single-Site Hydrolysis of TNP-ATP by the ␣ 3 ␤ 3 ␥ and ␣ 3 ␤ 3 Complexes-The ␣ 3 ␤ 3 ␥ complex hydrolyzes TNP-ATP, which is added in a substoichiometric molar ratio to the complex (Fig. 4A, closed circles). When ATP was chase-added to the reaction mixture on the progress of this single-site reaction, the hydrolysis of TNP-ATP was strongly accelerated. Even 1 M ATP (Fig. 4A, open circles) is as effective as 3.3 mM ATP (Fig.  4A, open squares) for inducing acceleration of TNP-ATP hydrol-ysis. As observed for MF 1 (Grubmeyer and Penefsky, 1981), rapid binding of TNP-ATP is followed by slow hydrolysis, which then is accelerated by the occupation or hydrolysis of ATP at the second nucleotide binding site.
The ␣ 3 ␤ 3 complex also catalyzes TNP-ATP hydrolysis under the single-site hydrolysis condition (Fig. 4B, closed circles), but this hydrolysis is not accelerated by the chase incubation for 5 s with the addition of 1 M ATP (Fig. 4B, open circles). The lack of the chase acceleration at 1 M ATP for the ␣ 3 ␤ 3 complex cannot be attributed to the dissociation of the complex. As described above, nucleotide-induced destabilization of the ␣ 3 ␤ 3 complex did not occur under this condition. In addition, Sato et al. (1995) reported that the dissociation of the ␣ 3 ␤ 3 complex into the ␣ 1 ␤ 1 complexes induced by ATP proceeds fairly slowly; at 1 mM ATP, only very little dissociation was observed in 20 s, and it took about 100 s for the complete dissociation. We measured the amounts of ATP and ADP in the reaction tubes of the experiment of Fig. 4B after termination of the chase reaction and confirmed that added ATP was almost completely hydrolyzed in 5 s (data not shown). Thus, lack of chase acceleration for the complex may not be due to inability of ATP at 1 M to interact with the complex, even though a possibility remains that ATP preferentially binds and is hydrolyzed by the complexes without bound TNP-ATP. To be sure, we measured the effect of chase addition of 1 mM ATP, and as expected no acceleration was observed (data not shown).
Single-Site Hydrolysis of TNP-ATP by the ␣ 3 ␤ 3 ␥ and ␣ 3 ␤ 3 Complexes Containing Mutant ␤ Subunits-Tyr-341 of the TF 1 -␤ subunit is located at a catalytic site and involved in binding of ATP (or ADP) (Bullough and Allison, 1986;Xue et al., 1987;Jault et al., 1994;Abrahams et al., 1994). Replacement of this residue by other residues such as Leu, Ala, and Cys resulted in a drastic decrease of binding affinity of the isolated ␤ subunit to ATP, and K m values for ATP of the mutant ␣ 3 ␤ 3 ␥ complexes increased in a parallel manner (Table I) (Odaka et al., 1994). To know the relation between the nucleotide binding site on the ␤ subunit and a single high affinity site of the ␣ 3 ␤ 3 ␥ complex, hydrolysis of TNP-ATP by the ␣ 3 ␤ 3 ␥ complexes containing mutant ␤ subunits (Y341L, FIG. 2. Binding of TNP-ADP to ␣ 3 ␤ 3 ␥ (q) and ␣ 3 ␤ 3 (E) complexes measured by equilibrium dialysis. Details of the experiments are described under "Experimental Procedures." FIG. 3. Transition of difference absorption spectra of the ␣ 3 ␤ 3 ␥ complex induced by each of the step-wise additions of TNP-ADP. Into 1 ml of 2 M ␣ 3 ␤ 3 ␥ complex solution in 20 mM Tricine-NaOH (pH 8.0) and 2 mM of MgCl 2 , 5 l of 100 M of TNP-ADP were added repeatedly in a step-wise manner, and the difference spectra induced by each addition were recorded. The molar ratios of TNP-ADP to the complex before and after each addition and the scale of induced difference spectra are shown in the figure. The difference spectra observed for the isolated ␤ and ␣ subunits are shown at the top and bottom, respectively, as references. The magnitudes of these two spectra were changed arbitrarily. Details of the experiments are described under "Experimental Procedures." Y341A, or Y341C) were examined. Surprisingly, ␣ 3 ␤ 3 ␥ complexes containing these mutant ␤ subunits exhibited almost the same kinetics in single-site hydrolysis of TNP-ATP and in chase acceleration by 3.3 mM ATP as those of the wild-type complex (Fig. 4, C, E, and G, closed circles and open squares).
However, when 1 M ATP was chase-added, the ␣ 3 ␤(Y341L) 3 ␥ and ␣ 3 ␤(Y341A) 3 ␥ complexes exhibited smaller extent of acceleration, and the ␣ 3 ␤(Y341C) 3 ␥ complex did not show chase acceleration (Fig. 4, C, E, and G, open circles). ATP concentrations necessary for half-maximal chase acceleration were measured for each of the mutant complexes, and it was shown that higher concentrations of ATP, compared with the wild-type complex, were required for acceleration of the mutant complexes (Table I). For the wild-type ␣ 3 ␤ 3 ␥ complex, less than 1 M ATP was sufficient to induce half-maximal acceleration of single-site TNP-ATP hydrolysis. However, nearly 40 M of ATP was required for the ␣ 3 ␤(Y341C) 3 ␥ complex. Chase-added ATP should bind to the second (or third) nucleotide binding sites of the complex whose first site is already occupied by TNP-ATP. Although the result of the ␣ 3 ␤(Y341A) 3 ␥ complex is somehow not exactly parallel with the nucleotide binding affinity of the isolated ␤ subunit, the affinity of the second (or third) site to ATP seem to reflect intrinsic nucleotide binding affinity measured for the isolated ␤ subunit.
Different from the ␣ 3 ␤ 3 ␥ complex, single-site hydrolysis of TNP-ATP by mutant ␣ 3 ␤ 3 complexes proceeded more slowly than that by wild-type ␣ 3 ␤ 3 complex (Fig. 4, D, F, and H, closed  circles). The hydrolysis was not accelerated by the chase addition of 1 M ATP (Fig. 4, D, F, and H, open circles). The rate of a Affinity of the isolated ␤ subunit to ATP expressed as dissociation constants. The values were adapted from a reference (Odaka et al., 1994). b K m values of ATPase activities of the ␣ 3 ␤ 3 ␥ complexes containing listed ␤ subunits. The values were adapted from a reference (Odaka et al., 1994).
c The concentration of chase-added ATP necessary for 50% of the maximum acceleration of the single-site TNP-ATP hydrolysis in 5 s by the ␣ 3 ␤ 3 ␥ complexes containing listed ␤ subunits. d The time required for hydrolysis of 50% of the TNP-ATP by the ␣ 3 ␤ 3 complex under the single-site hydrolysis condition. The values were estimated from Fig. 4. single-site hydrolysis of TNP-ATP by each of mutant ␣ 3 ␤ 3 complexes shows a parallel relation with the nucleotide binding affinity of each isolated mutant ␤ subunit to ATP (Table I).
When the same experiments of difference spectra induced by the step-wise addition of TNP-ADP as shown for the wild-type complex (Fig. 3) were repeated for the ␣ 3 ␤ 3 ␥ complexes containing mutant ␤ subunits, very similar transition of the spectra with increasing TNP-ADP were observed, that is, ␤ subunit-type spectra (molar ratio Ͻ ϳ1.5) at first and then ␣ subunit-type spectra (molar ratio Ͼ ϳ1.5) (data not shown). Therefore, in agreement with the results of single-site catalysis, a high affinity site is generated on one of the mutant ␤ subunits when the mutant ␤ subunits are assembled into the ␣ 3 ␤ 3 ␥ complex in spite of the fact that the isolated mutant ␤ subunits have only very weak affinities to nucleotides (Odaka et al., 1994). For the mutant ␣ 3 ␤ 3 complexes, the magnitude of difference spectra were very small, and clear transition as described above was not observed (data not shown).

DISCUSSION
The Role of ␥ Subunit in Generation of a High Affinity Nucleotide Binding Site on the ␤ Subunit-It has been known that the affinity of the isolated TF 1 ␤ subunit to TNP-ADP (apparent K d ϭ 4.5 M) greatly increases when it is assembled in the form of TF 1 (K d ϭ 2.2 nM), and only a single ␤ subunit out of three ␤ subunits in a TF 1 molecule bears this high affinity site (Hisabori et al., 1992). Here, using TNP-ATP (or TNP-ADP) as safe nucleotides that do not induce destabilization of the ␣ 3 ␤ 3 complex (Fig. 1), we compared nucleotide binding characteristics and kinetics of the ␣ 3 ␤ 3 and ␣ 3 ␤ 3 ␥ complexes. The main differences between these two complexes are: 1) a single high affinity nucleotide binding site exists on the ␣ 3 ␤ 3 ␥ complex but not on the ␣ 3 ␤ 3 complex (Fig. 2) and 2) an acceleration of the single-site TNP-ATP hydrolysis by the chase addition of ATP is observed for the ␣ 3 ␤ 3 ␥ complex but not for the ␣ 3 ␤ 3 complex (Fig. 4). Thus the ␥ subunit is responsible for both generation of a single high affinity catalytic site and communications among catalytic sites in the ␣ 3 ␤ 3 ␥ complex. This conclusion may not be restricted in TF 1 . Gao et al. (1995) reported recently that F 1 -ATPase from chloroplast thylakoid membrane lacking ␦ and ⑀ subunits, designated as CF 1 (Ϫ␦⑀), whose subunit composition should be ␣ 3 ␤ 3 ␥, retained about four nucleotides (2 ADP and 2 ATP) after passage through two centrifuge gel filtration columns, whereas the ␣␤ complex, whose subunit composition should be ␣ 3 ␤ 3 , retained about one ATP. Although some ambiguity remained because of the retention of one ATP on the ␣␤ complex, the results suggested that binding of the ␥ subunit to ␣ 3 ␤ 3 structure induces a three-dimensional conformation that is necessary for high affinity asymmetric nucleotide binding.
The importance of the ␥ subunit in coupling of ATP hydrolysis and proton translocation was pointed out a long time ago (Yoshida et al., 1977b), and recent studies using cross-linking and fluorescent probes have provided evidence for movement or large conformational change in the ␥ subunit during catalysis (Turina and Capaldi, 1994;Aggeler et al., 1995). Because the ␣ 3 ␤ 3 complex should have a structure with a perfect 3-fold symmetry 2 and it contains essentially no endogenously bound nucleotide, structural asymmetry of the ␣ 3 ␤ 3 ␥ complex is caused solely by the ␥ subunit. Crystal structure of bovine MF 1 has shown that each ␤ subunit has a different contact with a centrally located ␥ subunit and is in a different state in terms of nucleotide occupancy: one occupied by AMPPNP, another by ADP, and the third by none (empty) (Abrahams et al., 1994). Our results indicate that the heterogeneous state of the catalytic sites and their mutual communication are dependent on structural asymmetry brought by the ␥ subunit.
Generation of a High Affinity Site in the ␣ 3 ␤ 3 ␥ Complexes Containing Mutant ␤ Subunits-The isolated ␤ subunit can bind ATP with a K d value of 15 M (Hisabori et al., 1986), but it loses the affinity to nucleotides at great extent by replacing Tyr-341 with Leu, Ala, or Cys (Table I) (Odaka et al., 1994). Corresponding to this reduced affinity, the ␣ 3 ␤ 3 complexes containing the mutant ␤ subunits appear to have reduced binding affinities to TNP-ATP compared with the wild-type ␣ 3 ␤ 3 complexes. However, when these mutant ␤ subunits are incorporated into the ␣ 3 ␤ 3 ␥ complex, they display high affinity binding to TNP-ATP, which was demonstrated by single-site hydrolysis ( Fig. 4). High affinity to TNP-ADP was also demonstrated for the mutant ␣ 3 ␤ 3 ␥ complexes by titration of difference spectra, which is almost indistinguishable from that of the wild-type complex. Thus, a single high affinity site is generated by the aid of the ␥ subunit even if the intrinsic affinity of the mutant ␤ subunit to nucleotides is severely impaired. Some residues other than Tyr-␤341 should contribute to generate the high affinity site under the influence of the ␥ subunit. However, a mention should be added that the affinity of the second nucleotide binding site of the mutant ␣ 3 ␤ 3 ␥ complexes to ATP seems to be weaker than that of the wildtype complex (Table I). It may be also the case for the third site, the catalysis on which should constitute a dominant part of the overall V max at high ATP concentrations, and then K m values of the steady state ATP hydrolysis by the mutant ␣ 3 ␤ 3 ␥ complexes become very large (Table I) (Odaka et al., 1994). The results of both the wild-type and the mutant complexes provide strong evidence for the major contribution of the ␥ subunit in generation of the single high affinity nucleotide binding site in F 1 -ATPase.