The Formation or the Reduction of a Disulfide Bridge on the γ Subunit of Chloroplast ATP Synthase Affects the Inhibitory Effect of the ε Subunit*

We have studied the change of the catalytic activity of chimeric complexes that were formed by chloroplast coupling factor 1 (CF1) -γ, α and β subunits of thermophilic bacterial F1 after formation or reduction of the disulfide bridge of different γ subunits modified by oligonucleotide-directed mutagenesis techniques. For this purpose, three mutant γ subunits were produced: γΔ194–230, here 37 amino acids from Pro-194 to Ile-230 are deleted, γC199A, Cys-199 is changed to Ala, and γΔ200–204, amino acids from Asp-200 to Lys-204 are deleted. All of the chimeric subunit complexes produced from each of these mutant CF1-γ subunits and α and β subunits from thermophilic bacterial F1 lost the sensitivity against thiol reagents when compared with the complex containing wild-type CF1-γ. The pH optimum (pH 8.5–9.0) and the concentration of methanol to stimulate ATPase activities were not affected by these mutations. These indicate that the introduction of the mutations did not change the main features of ATPase activity of the chimeric complex. However, the interaction between γ subunit and ε subunit was strongly influenced by the type of γ subunit itself. Although the ATPase activity of the chimeric complex that contained γΔ200–204 or γC199A was inhibited by the addition of recombinant ε subunit from CF1 similarly to complexes containing the reduced wild-type γ subunit, the recombinant ε subunit did not inhibit the ATPase of the complex, which contained the oxidized form of γ subunit. Therefore the affinity of the ε subunit to the γ subunit may be dependent on the state of the γ subunit or the ε subunit may bind to the oxidized form of γ subunit in a mode that does not inhibit the activity. The ATPase activity of the complex that contains γΔ194–230 was not efficiently inhibited by ε subunit. These results show that the formation or reduction of the disulfide bond on the γ subunit may induce a conformational change in the region that directly affects the interaction of this subunit with the adjacent ε subunit.

F 0 F 1 -ATP synthase synthesizes ATP from ADP and P i at the expense of a proton-motive force (1)(2)(3). The enzymes consist of the membrane-embedded sector F 0 responsible for proton translocation , and the extrinsic catalytic part F 1 . The architecture of F 1 is very similar in various kinds of cells or organelles. The F 1 part is composed of five different subunits designated as ␣, ␤, ␥, ␦, and ⑀, and their molecular stoichiometry is 3:3:1:1:1. Nucleotide binding sites reside on each of the ␣ and ␤ subunits, i.e. there are altogether 6 nucleotide binding sites per F 1 . The catalytic sites are located at the interfaces between ␣ and ␤ subunits. The high resolution x-ray structure of bovine heart mitochondrial F 1 confirmed that most of the amino acid residues that form this site are provided from the ␤ subunit (4). The ␣ and ␤ subunits, which have a similar three-dimensional structure, alternate in a hexagonal arrangement around a central cavity containing the ␥ subunit as already expected from previous electron microscopic studies (5). The crystal structure of an ␣ 3 ␤ 3 complex from the thermophilic Bacillus PS3 was completely symmetric (6), but the incorporation of the ␥ subunit into this complex induced a functional asymmetry among the three catalytic sites (7).
Rotation of the ␥ subunit related with catalysis was suggested from kinetic results (8), the exchange of a disulfide bridge formed between ␥ and ␤ subunits (9), and polarization anisotropy relaxation measurement of the fluorophore-labeled ␥ subunit of chloroplast F 1 (CF 1 ) 1 (10). Recently, Noji et al. (11) directly observed that this ␥ subunit rotates in the central cavity formed by ␣ and ␤ subunits like a motor axis during the ATP hydrolysis reaction. This experiment clearly shows that the interaction between the part of coiled-coil of the ␥ subunit and the inner surface of the central cavity formed by ␣ and ␤ subunits is not tight, and the interaction between the ␥ subunit and one of ␣ or ␤ subunit can alternate step by step in one direction according to the catalytic reaction occurring sequentially at each of three catalytic sites.
The CF 0 CF 1 -ATP synthase of chloroplasts is regulated by the proton gradient, which activates the enzyme and by the reduction or the oxidation of a disulfide bridge in the ␥ subunit, which modulates activity. The latter regulation is known as thiol modulation (12). The structural basis for the thiol modulation is assigned to the sequence motif of 9 amino acids comprising two cysteines in the ␥ subunit (13). In vitro reduction can be achieved by dithiothreitol (DTT) or other dithiols, but the natural reductant is a reduced thioredoxin f (14), which was reduced by the photosynthetic electron transport chain via ferredoxin. Recently, Ross et al. (15) have succeeded in constructing a mutant ␥ subunit with one or two cysteines substituted by serine in the green algae Chlamydomonas reinhardtii. By these substitutions, CF 1 became a DTT-insensitive enzyme. On the other hand, Gabrys et al. (16) selected a couple of mutants with chloroplast ATPase redox responses that were different from that of the wild-type plant by screening the Arabidopsis grown from the seeds that were previously treated with mutagen. They selected some mutant plants that might contain mutations within the ␥ subunit of CF 1 .
The ⑀ subunit of CF 1 is known as an intrinsic inhibitor protein. Recently, Cruz et al. (17,18) produced recombinant mutant ⑀ subunits and tested their effects on the ATPase activity of ⑀-deficient CF 1 (CF 1 (Ϫ⑀)). They determined that the most important part of the ⑀ subunit as an inhibitor was the NH 2 -terminal region. Similar experiments were reported for the ⑀ subunit of F 1 from Escherichia coli (EF 1 ) (19). These studies suggested that about 15 amino acid residues from the NH 2 terminus were necessary for the inhibition. Wilkens et al. (20) reported the three-dimensional structure of the isolated ⑀ subunit from EF 1 solved in solution by NMR spectroscopy. The structure shows that the NH 2 -terminal 90 amino acids form so-called ␤-sandwich structure with two five-stranded ␤ sheets. The C-terminal domain, on the other hand, is formed by two ␣-helices. Recently, Uhlin et al. (21) solved the crystal structure of this subunit at 2.3 Å resolution. They confirmed the ␤-sandwich structure reported by Wilkens et al. (20) and found that the C-terminal two helices arranged in an anti-parallel coiledcoil structure. From the three-dimensional structure and crosslinking experiments employing cysteine mutants of ␥ and ⑀ subunits, it was concluded that the contact region of ⑀ subunit to ␥ subunit is at one side of the ␤-sandwich structure (20,22).
The results of Capaldi and co-workers (20,22) show that about 40 amino acids from the NH 2 terminus of the ⑀ subunit, which form one-half of the ␤-sandwich structure, represent the region where the subunit is in direct contact with the ␥ subunit. Recently Schulenberg et al. (23) also reported that the ⑀ subunit of CF 1 can contact the ␥ subunit at the similar region. The characteristics of the interaction between ␥ subunit and ⑀ subunit of CF 1 were investigated directly (24,25) and indirectly (26). Andralojc and Harris (24) reported that the affinity of the ⑀ subunit to CF 1 (Ϫ⑀) was decreased when the ␥ subunit of the complex was in a reduced state. From the inhibition of Ca 2ϩ -ATPase activity, they estimated a K d of 60 nM for the reduced CF 1 (Ϫ⑀) and 0.14 nM for the oxidized one. Duhe and Selman (25) also reported a stimulation of dissociation of the ⑀ subunit from CF 1 of Chlamydomonas in the presence of DTT. They suggested that the dissociation of the ⑀ subunit is an obligatory process in the DTT-induced unmasking of ATPase activity of soluble CF 1 . Soteropoulos et al. (26) reported an activation of CF 1 -ATPase by dilution of the enzyme. For the reduced CF 1 , a half-maximal activation was obtained at a much lower dilution than with oxidized CF 1 , and they estimated that the affinity for the ⑀ subunit would be decreased about 20-fold by the reduction of CF 1 .
Recently, we reported on the reconstitution of a chimeric complex from recombinant ␣ and ␤ subunits from F 1 of the thermophilic Bacillus PS3 (TF 1 ) and the recombinant ␥ subunit (␥ c ) from spinach CF 1 (27). The complex had substantial ATPase activity and this activity was affected by the disulfide/ dithiol state of the two regulatory cysteine residues on the ␥ subunits. That means that ␥ c imposed redox control on the chimeric complex. Furthermore, the activity of the chimeric complex was suppressed by the addition of recombinant ⑀ subunit from CF 1 (⑀ c ), but not by the addition of the ⑀ subunit from TF 1 . These results suggest that the regulatory functions of ␥ and ⑀ subunits of CF 1 may be linked to each other.
Here, we prepared three modified ␥ subunits of CF 1 by oligonucleotide-directed mutagenesis. We investigated the effects of the mutations on the enzyme activity and its regulation in chimeric complexes formed by these ␥ subunits, ⑀ c , and ␣ 3 ␤ 3 from TF 1 . We found the region around the disulfide bridge of the ␥ subunit to be important for the regulatory interaction between the ␥ subunit and ⑀ subunit.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonucleases were obtained from Toyobo Inc., Tokyo, Japan. The Bradford protein assay system was from Bio-Rad. Urea was purchased from Nacalai Tesque, Kyoto, Japan. DTT was from Sigma. Other chemicals were the highest grade commercially available.
Construction of the Plasmids for the Mutant ␥ c and Their Expression-Recombinant plasmids carrying the gene for the ␥ subunit from spinach plastids (atpC) was previously constructed (27). Oligonucleotide-directed mutagenesis was carried out as described by Kunkel et al. (28). The oligonucleotide used to create the ␥ c with the additional amino acid stretch of the ␥ subunit of CF 1 (from Pro-194 to Ile-230) deleted (␥ ⌬194 -230 ) was: 5Ј-ATCCACACCCTACTCCCCTTAAGAAAAACCGAA-ACACCAGCATTTT-3Ј. The one used to create the ␥ c with Cys-199 changed to Ala (␥ C199A ) was: 5Ј-ACCCTACTCCCCTTAAGACCCAAAG-GAGAAATTGCGGACATCAATGGAAAA-3Ј. To create ␥ c with the amino acid sequence from Asp-200 to Lys-204 deleted (␥ ⌬200 -204 ), the following oligonucleotide was used: 5Ј-ACCCTACTCCCCTTAAGACC-CAAA-GGAGAAATTTGCTGTGTCGACGCAGCAGAA-3Ј. Each of the genes was transferred to the expression vector pET23d (Novagen) and was transformed into the expression host E. coli strain BL21(DE3). Each of the ␥ c proteins was expressed yielding inclusion bodies and further purified by the methods described previously (27).
Expression and Purification of the Recombinant ⑀ c -The plasmid pSocB149 (28), which contains the gene for the subunit ⑀ of CF 1 from spinach (⑀ c ) was a generous gift from Dr. Whitfeld, R. P., Australia. The expression plasmid for ⑀ c was constructed according to the method described by Hisabori et al. (27) and transformed into E. coli strain BL21(DE3). ⑀ c was over-expressed by the method used for ␥ c . The ⑀ c inclusion bodies were first dissolved by the addition of 8 M urea, 40 mM Tris-Cl (pH 8.0), 0.4 mM DTT, and 0.8 mM EDTA and further purified by the method described previously (27).
Reconstitution of the Chimeric Subunit Complex-Reconstitution of the chimeric subunit complex was formed by the same method described previously (27). Briefly, each of the isolated ␥ c subunits was mixed with the ␣ and ␤ subunits from TF 1 in a ratio of 1:1:1 (w/v), and a solution containing 8 M urea, 1 mM EDTA, 0.5 mM DTT, and 50 mM Tris-Cl (pH 8.0) was added to yield a final urea concentration of 4 M. The solution was then dialyzed against 50 mM Tris-HCl (pH 8.0), 200 mM NaCl, 0.4 mM MgCl 2 , and 0.4 mM ATP at 20°C for 3 h. After the dialysis, the unsolved ␥ c subunits were removed from the solution by centrifugation and provided for the measurement of ATPase activity.
Activation and Deactivation of the Complex and the Measurement of ATPase Activity-To activate or deactivate the ␣ 3 ␤ 3 ␥ c complexes by the formation or reduction of disulfide bridge on the ␥ subunit, formed complexes were incubated in the presence of 2 mM DTT or 50 M CuCl 2 for 2 h at 30°C. Then 10 l of the complex solution (normally containing 2-3 g of protein) was added to 90 l of the reaction mixture containing 50 mM Tricine-KOH (pH 8.0), 2 mM ATP, and 2 mM MgCl 2 to initiate the reaction. The reaction was continued for 5-10 min, and then terminated by the addition of 100 l of ice-cold 2.4% (v/v) perchloric acid. The amounts of liberated P i were measured by a colorimetric method. Since the ATPase activity was measured by the mixture of the chimeric complex and the monomer proteins, which were not incorporated into the complex, the specific activity of the ATPase was calculated based on the amounts of ␣ plus ␤ subunits that were used for the formation of the complex.

RESULTS AND DISCUSSION
Responses of the Chimeric Complexes Containing Mutant ␥ c to Oxidation/Reduction-In comparison to the ␥ subunit from thermophilic bacteria, the ␥ subunit of CF 1 contains a stretch of an additional 37 amino acids comprising the two cysteines (Cys-199 and Cys-205), which are responsible for the thiol modulation (Fig. 1A). To characterize the differences of the properties between bacterial ␥ and chloroplast ␥, we designed three mutations of the ␥ subunit of CF 1 . These were ␥ ⌬194 -230 , which is lacking the CF 1 -␥ specific additional amino acid stretch (see Fig. 1B), ␥ ⌬200 -204 , in which the space between the two regulatory cysteines is shortened, and ␥ C199A , which cannot form the disulfide bridge involved in thiol modulation. The complexes containing those mutant ␥ c subunits were not supposed to show any responses against oxidation/reduction. As shown in Fig. 2, the ATPase activity of these chimeric complexes was not altered by the incubation with DTT or with CuCl 2 , although the ATPase activity of the complex containing wild-type ␥ c was remarkably stimulated by the incubation with DTT and suppressed by the incubation with CuCl 2 as reported previously (27).
For further characterization of the mutant ␥ subunits, some properties of the chimeric complexes were investigated. First, we measured the pH dependence of the ATPase activity of the complexes. We could not find any change of pH dependence of ATPase activity; all four kinds of complexes showed the same pH profiles (data not shown), indicating that the property of the catalytic sites did not change by the mutation of ␥ c subunit. Also the optimal pH (pH 8.5-9.0) was the same as that found for the authentic TF 1 (32) or CF 1 (33).
Methanol Activation of the ATPase Activity of the Chimeric Subunit Complexes-Methanol stimulation of ATPase activity is one of the unique features of isolated (33,34) and membranebound CF 1 (35). In a previous report, we found that the incorporation of the ␥ subunit from CF 1 into the bacterial ␣ 3 ␤ 3 hexagon conferred the property of methanol stimulation of the ATPase to the complex, whereas the ␣ 3 ␤ 3 ␥ complex formed by the bacterial subunits only is insensitive against methanol (27). As shown in Fig. 3, the ATPase activities of all four chimeric complexes were stimulated by the addition of methanol and optimal concentrations of methanol were 25% (v/v). The most remarkable difference between the ␥ subunit from thermophilic bacteria and the ␥ subunit of CF 1 is the occurrence of an intercalated amino acid stretch of more than 30 amino acids in CF 1 -␥ (Fig. 1A). One may expect that the region that senses methanol would be located in this additional stretch. However, when the mutant ␥ subunit, which is lacking in this region, was used to form the chimeric complex, the methanol sensitivity was the same as for the complex that contained the wild-type ␥ subunit. Therefore, the methanol sensitivity of CF 1 -␥ subunit must be attributed to some other portion of the protein.
Inhibitory Effects of the ⑀ Subunit-The ⑀ subunit of F 1 is an intrinsic ATPase inhibitor. However, this subunit may also be involved in H ϩ coupling of the ATPase (36). Recently, Cruz et al. (17,18) expressed the mutant ⑀ subunit of CF 1 in E. coli and investigated their inhibitory effects on the CF 1 (Ϫ⑀). Similar deletion experiments were carried out for EF 1 by Jounouchi et al. (19). They reported that the deletion of the NH 2 -terminal 16 amino acids is strongly affecting the coupling between ATP hydrolysis and H ϩ translocation, but F 1 with an ⑀ subunit lacking the 15 amino-terminal residues could bind to F 0 in a functionally competent manner. However, Cruz et al. (17) found that the deletions of 6 amino acids from the C terminus or the deletions of 11 amino acids from the NH 2 terminus decreased the inhibitory effect of this subunit on the ATPase of CF 1 .
The Mg 2ϩ -ATPase activity of the oxidized form of CF 1 is quite low or almost zero. However, the chimeric complex displayed remarkable Mg 2ϩ -ATPase activity even in its oxidized form. Therefore it was possible to investigate the interaction between the ⑀ and ␥ subunits under the reduced and oxidized The chimeric complex was reconstituted from 500 g of ␥ c , ␣ subunit, and ␤ subunit according to the method described under "Experimental Procedures." Then each of the complexes was incubated with 2 mM DTT (t) or 50 M CuCl 2 (Ⅵ) for 2 h at 30°C, and their ATPase activities were measured. The ATPase activity (mol P i released/mg ␣ ϩ ␤/min) of each of the complexes that were not treated with DTT or CuCl 2 (ٗ) was 0.672 with wild-type ␥, 0.254 with ␥ ⌬194 -230 , 0.557 with ␥ ⌬200 -204 , and 0.534 with ␥ C199A , and set as 100%.
conditions by measuring the inhibition of the ATPase. Surprisingly, the ATPase activity was less inhibited by the ⑀ subunit in the chimeric complex with the oxidized ␥ subunit than with the reduced ␥ subunit (Fig. 4A) (27). The addition of methanol reduced the inhibitory effect of ⑀ c for both the reduced and oxidized state complexes, indicating the stimulation effect of methanol on CF 1 -ATPase can be partially attributed to the release of the ⑀ subunit from the enzyme. A change of the responses of CF 1 against the ⑀ subunit under the reduced or oxidized condition were already reported by Andralojc and Harris (24), Duhe and Selman (25), and Soteropoulos et al. (26). Andralojc and Harris (24) investigated the inhibition of Ca 2ϩ -ATPase activity. By adding the various amounts of isolated ⑀ subunit to CF 1 (Ϫ⑀), they concluded that the oxidized CF 1 (Ϫ⑀) has a higher affinity for the isolated ⑀ subunit than the reduced enzyme. Soteropoulos et al. (26) diluted CF 1 solution to nanomolar concentration and found a difference of the activation ratio by oxidation/reduction. Activation by dilution occurred for the reduced enzyme at higher than for the reduced enzyme. From their results, they concluded that the affinity of the ⑀ subunit to the reduced CF 1 is about 20-fold lower than to the oxidized one.
It is difficult to explain why the apparent affinity of the ⑀ subunit to the chimeric complex is lower when the ␥ subunit is in the oxidized state. Possibly the origin of the ␣ 3 ␤ 3 hexamer influences the interaction, too. On the other hand, we cannot be sure that the ␥ c really has the same conformation in the chimeric complex as the authentic CF 1 .
Relation between the Conformation of the ␥ Subunit and the Effect of the ⑀ Subunit-If the additional amino acid stretch observed only on CF 1 is responsible for the interaction between the ␥ and ⑀ subunits, a mutation of this segment might affect the inhibition of ATPase by the ⑀ subunit. The sensitivity of the complexes that contain ␥ C199A or ␥ ⌬200 -204 against the ⑀ subunit was the same as that of the complex with the reduced form of the ␥ subunit (Fig. 4B), although we expected that the conformation of ␥ ⌬200 -204 was similar to the oxidized-form of the ␥ subunit.
On the other hand, the complex that contains ␥ ⌬194 -230 was not inhibited by the addition of the ⑀ subunit. From the report of Capaldi and co-workers (20,22), the contact region between the ⑀ and ␥ subunits is located around 40 amino acids from the NH 2 terminus of the ⑀ subunit. The chemical cross-linking experiments carried out by using cross-linker-labeled ⑀ subunit and CF 1 (Ϫ⑀) gave the same conclusion (23). The contact region on the ␥ subunit is very close to the position where the additional amino acid stretch is intercalated in CF 1 -␥ (13) (see Fig.  1A). Hence, the structure of this additional amino acid stretch might be very important not only for the redox regulation but also for the interaction between the ␥ and ⑀ subunits.
Ross et al. (37) reported that mutation of the spacer region between the two regulatory cysteines (GEICD(K or A)VDGK-(D)CVDAA) diminished redox regulation of CF 1 from Chlamydomonas. They used C. reinhardtii strain atpC1, which lacks the gene for the ␥ subunit, and complemented the photophosphorylation activity with mutated ␥ subunit genes. Thylakoid vesicles prepared from the mutant strain containing ␥ D199K/K203D or ␥ D199A did not show a remarkable change of the photophosphorylation in the presence or the absence of DTT. Accordingly the authors concluded that the spacer region be- A, the chimeric complex was formed the same way as described in the legend for Fig. 2. 100 g/ml of the reduced (closed symbols) or the oxidized (open symbols) complex was incubated with the indicated amounts of ⑀ c for 1 h at room temperature. The ATPase activity of the complex was measured for 5 to 10 min in the absence (circle) or presence (triangle) of 20% methanol. B, the chimeric complexes with three kinds of mutant ␥ subunit were formed the same way as described in the legend for Fig. 2. 100 g/ml of each of the complexes containing ␥ ⌬194 -230 (q), ␥ ⌬200 -204 (OE), and ␥ C199A (E) was incubated with the indicated amounts of ⑀ c for 1 h at room temperature, and the ATPase activity of the complex was measured. tween the disulfide bridge is also involved in redox regulation of CF 1 . This result corresponds with our result concerning ␥ ⌬200 -204 . Although ␥ ⌬200 -204 still contains the two cysteines, the complex with this subunit did not show any difference in activity by incubation with DTT or CuCl 2 , respectively (Fig. 2). Apart from this all other properties of ATPase of this complex were the same as those of the wild type. This complex appeared to have an affinity for ⑀ subunit because the ATPase of this complex was strongly inhibited by the addition of ⑀ c subunit (Fig. 4B).
Hence, our results strongly suggest that the change in the conformation that occurred at the lower part of the ␥ subunit is drastic enough to affect the binding of the ⑀ subunit to this subunit, although the point that is different from the results of Ross et al. (37) as mentioned above should be further investigated.