Assembled F1-(αβ) and Hybrid F1-α3β3γ-ATPases fromRhodospirillum rubrum α, Wild Type or Mutant β, and Chloroplast γ Subunits

Refolding together the expressed α and β subunits of the Rhodospirillum rubrumF1(RF1)-ATPase led to assembly of only α1β1 dimers, showing a stable low MgATPase activity. When incubated in the presence of AlCl3, NaF and either MgAD(T)P or CaAD(T)P, all dimers associated into closed α3β3 hexamers, which also gained a low CaATPase activity. Both hexamer ATPase activities exhibited identical rates and properties to the open dimer MgATPase. These results indicate that: a) the hexamer, as the dimer, has no catalytic cooperativity; b) aluminium fluoride does not inhibit their MgATPase activity; and c) it does enable the assembly of RrF1-α3β3 hexamers by stabilizing their noncatalytic α/β interfaces. Refolding of the RrF1-α and β subunits together with the spinach chloroplast F1 (CF1)-γ enabled a simple one-step assembly of two different hybrid RrF1-α3β3/CF1γ complexes, containing either wild type RrF1-β or the catalytic site mutant RrF1β-T159S. They exhibited over 100-fold higher CaATPase and MgATPase activities than the stabilized hexamers and showed very different catalytic properties. The hybrid wild type MgATPase activity was, as that of RrF1 and CF1 and unlike its higher CaATPase activity, regulated by excess free Mg2+ ions, stimulated by sulfite, and inhibited by azide. The hybrid mutant had on the other hand a low CaATPase but an exceptionally high MgATPase activity, which was much less sensitive to the specific MgATPase effectors. All these very different ATPase activities were regulated by thiol modulation of the hybrid unique CF1-γ disulfide bond. These hybrid complexes can provide information on the as yet unknown factors that couple ATP binding and hydrolysis to both thiol modulation and rotational motion of their CF1-γ subunit.

The F 0 F 1 -ATP synthases catalyze the synthesis of ATP from ADP and P i at the expense of a transmembrane electrochemical proton gradient generated by the respiratory or photosynthetic electron transport chains. Its membrane-bound F 0 sector functions as the proton pathway and has, in bacterial and photosynthetic cells, a subunit composition of a 1 b 2 c 9 -12 and a 1 b 1 bЈ 1 c 9 -12 , respectively. The catalytic F 1 sector, which functions as a soluble ATPase, has in all cells a subunit composition of ␣ 3 ␤ 3 ␥ 1 ␦ 1 ⑀ 1 and six nucleotide-binding sites located on the ␣ and ␤ subunits. Its three catalytic ␤ sites show high negative cooperativity in substrate binding and strong positive cooperativity in catalysis (1)(2)(3)(4)(5). The minimal F 1 subcomplex, which resembles the whole F 1 -ATPase in its catalytic properties, is the F 1 -␣ 3 ␤ 3 ␥-ATPase. Such subcomplexes were reconstituted from either native (6,7) or recombinant (8,9) individual subunits of the respiratory TF 1 -1 and EcF 1 -ATPases. Similar highly active pure photosynthetic subcomplexes have been assembled up to now only by incubating an isolated native CF 1 (␣␤) complex with a native (10) or recombinant (11) ␥ C subunit.
A high resolution x-ray structure of the bovine heart mitochondrial MF 1 (12) demonstrated the alternating arrangement of the ␣ and ␤ subunits in a closed hexamer, with the resolved N-and C-terminal ␣-helices of the ␥ subunit embedded in its central cavity and all six nucleotide-binding sites residing at alternating ␣/␤ interfaces. The three catalytic sites, located mainly on the ␤ subunits, appeared in three different conformational states that, in association with the unique resolved part of the ␥ subunit, imposed an asymmetric structure on the ␣ 3 ␤ 3 hexamer (12). This MF 1 -␣ 3 ␤ 3 ␥ structure is compatible with the binding change mechanism (3,13), which proposed that ATP synthesis involves transitions between different but interacting catalytic sites, via energy-dependent affinity changes in substrate binding and product release. Such transitions were first suggested to occur via movement or rotation of a cluster of the catalytic ␣ 3 ␤ 3 subunits around a core of the single copy ␥ or ⑀ subunits (14). The reversible proton-translocating ATP hydrolysis was later proposed to generate rotation of the F 1 -␥ subunit which, when transmitted to F 0 , could result in pumping of protons back across the membrane (15), possibly via coupled rotation of the F 1 -␥ subunit with the F 0 -c subunits (3,16). MgATPase-induced rotation of ␥ within immobilized ␣ 3 ␤ 3 hexamers was observed in genetically engineered respiratory TF 1 -and EcF 1 -␣ 3 ␤ 3 ␥ complexes (17,18). But its further coupling with F 0 -c rotation, which has been tested in some recent reports, did not yield clear results (19 -21). So there is at present no direct correlation between ␥ rotation and protoncoupled ATP synthesis and hydrolysis.
Full elucidation of the detailed mechanism of action of the F 0 F 1 -ATP synthase will depend on the identification of the specific domains that participate in its proton-coupled ATP synthesis and hydrolysis as well as in the regulation of these reversible activities and their possible correlation with the ␥ or ␥-c rotation. Tight regulation of ATP hydrolysis is especially important in photosynthetic cells, where it prevents the depletion of essential cellular ATP pools in the dark (1,4,5). Plant chloroplasts and bacterial chromatophores have a number of such regulatory pathways, which operate in their membranebound F 0 F 1 -as well as the solubilized F 1 -ATPases. Both chloroplasts (22) and chromatophores (23) show a high sensitivity to inhibition by excess free Mg 2ϩ ions, which results in a drastic decrease of their MgATPase activities at Mg/ATP ratios above 0.5. A unique chloroplast regulatory system, termed thiol modulation, involves reduction-oxidation of a disulfide bond (24) formed between Cys 199 and Cys 205 in a region of its ␥ C subunit that does not appear in any other F 1 -␥ subunits (25). But there are as yet no assembled CF 1 -␣ 3 ␤ 3 ␥ or similar photosynthetic complexes that can be engineered for studies aimed at elucidating the molecular mechanism of their regulatory systems and the possible ATPase-induced rotation of their ␥ subunit.
In this investigation, we have used our earlier developed procedure for refolding and assembly of the ␣ R and ␤ R subunits of the photosynthetic bacterium Rhodospirillum rubrum into ␣ R 1 ␤ R 1 dimers (26) to follow their further association into ␣ R 3 ␤ R 3 hexamers, and with the recombinant chloroplast ␥ C (11), into hybrid ␣ R 3 ␤ R 3 ␥ C -ATPases. Two types of such highly active hybrids were assembled, containing WT ␤ R or the catalytic site mutant ␤ R -T159S (27), and both retained the specific ␥ C redox regulation. These hybrid complexes provide the first photosynthetic F 1 assemblies that can be genetically engineered for probing rotational catalysis. Both hybrids show, besides a very high MgATPase also a CaATPase activity, that in the hybrid WT ␣ R 3 ␤ R 3 ␥ C is much higher than the MgATPase activity but does not respond to any tested MgATPase effector. This hybrid WT complex thus supplies an additional interesting system for assaying rotation, which has been induced up to now (17)(18)(19)(20)(21) only by MgATP hydrolysis.

EXPERIMENTAL PROCEDURES
Materials-The recombinant ␣ R (26,28), WT ␤ R (29), and the mutant ␤ R -T159S (27) subunits were expressed in insoluble inclusion bodies and solubilized by urea, as described by Du and Gromet-Elhanan (26). Recombinant ␥ C was expressed and refolded as described by Sokolov et al. (11). RrF 1 was prepared from R. rubrum chromatophores according to Weiss et al. (30). All other reagents were of the highest purity available.
Assembly and Isolation of ␣ R 1 ␤ R 1 Dimers-The dimers were assembled by refolding the urea-solubilized ␣ R and ␤ R together according to the procedure developed for their optimal refolding into functional monomers (26). The refolded mixture was concentrated to about 1 mg/ml by Centriprep-10 (Amicon), precipitated with 60% saturated (NH 4 ) 2 SO 4 , and resuspended in TGN buffer containing 50 mM Tricine-NaOH, pH 8.0, 20% glycerol, and 50 mM NaCl. The remaining insoluble aggregates were removed by centrifugation, and the refolded mixture was loaded on the size-exclusion HPLC Superdex-200 column (Amersham Pharmacia Biotech) and eluted with 100 mM NaP i , pH 7.0, containing 10% glycerol at a flow rate of 0.5 ml/min. The pooled ␣ 1 ␤ 1 dimer peak was concentrated, transferred to TGN buffer by elution-centrifugation through Sephadex G-50 columns, and stored at Ϫ80°C.
Assembly of ␣ R 3 ␤ R 3 Hexamers-The isolated ␣ R 1 ␤ R 1 dimers could be fully converted into the ␣ R 3 ␤ R 3 hexamers only when incubated, at Ͼ1 mg of protein/ml for 1 h at 22°C, in TGN buffer containing also 10 mM NaF, and 0.5 mM AlCl 3 , in the presence of either 1 mM of CaCl 2 or MgCl 2 and 1 mM ATP or ADP (see Fig. 1D). Their activities could therefore be assayed directly by diluting samples assembled in the presence of each cation and ADP into the same cation-ATPase assay mixtures.
Assembly and Isolation of the Hybrid WT and Mutant ␣ R 3 ␤ R 3 ␥ C Complexes-The hybrid WT ␣ R 3 ␤ R 3 ␥ C was assembled by two procedures as follows: 1) incubation of the isolated ␣ R 1 ␤ R 1 dimers with refolded ␥ C for 1 h at 22°C in TGN buffer in the presence of 1 mM MgCl 2 or CaCl 2 and 1 mM ADP which in this mixture, as in the hexamer assembly mixture, is as effective as ATP. Each incubated sample could therefore be assayed directly for its ATPase activity as well as size-exclusion HPLC. 2) Refolding the urea-solubilized recombinant ␣ R and ␤ R each at 50 g/ml, together with urea-solubilized ␥ C at 20 g/ml, according to the procedure developed above for assembly of ␣ R 1 ␤ R 1 dimers. The assembled hybrid complex was isolated by size-exclusion HPLC as described for the dimers, except that the hybrids were eluted with buffer containing 50 mM Tricine-NaOH, pH 8.0, 50 mM NaCl, and 10% glycerol. The peak containing the hybrid complex was pooled, concentrated, exchanged into TGN buffer as described for the dimers, and stored at Ϫ80°C. The hybrid mutant ␣ R 3 (␤ R -T159S) 3 ␥ C complex was assembled and isolated as described for the hybrid WT.
Assays of ATPase Activities-The activities of RrF 1 and both hybrid ␣ R 3 ␤ R 3 ␥ C WT and mutant complexes were measured with 4 -20 g of protein for 5 min at 35°C in 0.5 ml of an assay mixture containing 50 mM Tricine-NaOH, pH 8.0, 50 mM NaCl, 4 mM ATP, and 2 mM of either MgCl 2 or MnCl 2 or 4 mM CaCl 2 . The activities of the isolated ␣ R 1 ␤ R 1 dimers were measured with 30 g of protein for 30 min under the conditions described above. To compare the ATPase activities of the ␣ R 3 ␤ R 3 hexamers with those of RrF 1 , the hexamers were first assembled by incubating the dimers as described above, and RrF 1 was incubated under identical conditions. The ATPase activities of this RrF 1 and the freshly assembled ␣ R 3 ␤ R 3 hexamers were measured by diluting each incubation mixture into the relevant assay mixtures, which also contained 10 mM NaF and 0.5 mM AlCl 3 . All ATPase activity assays were started by adding the protein complexes and stopped by adding 50 l of 2 M trichloroacetic acid, and the released P i was measured as described by Taussky and Shorr (31). The effect of reduction or oxidation on the ATPase activities of RrF 1 and the hybrid complexes was tested by their preincubation for 1 h at 35°C in TGN buffer containing either 10 mM DTT or 100 M CuCl 2 , followed by dilution into the relevant assay mixtures.
Other Procedures-SDS-PAGE was carried out on the Novex Pre-Cast 10 -20% Tris glycine gradient gels. The protein bands were visualized by staining with Coomassie Brilliant Blue R-250. Protein concentrations were determined by the Bradford method (32) or according to Lowry et al. (33), using bovine serum albumin as a standard.

RESULTS
Stepwise Assembly of RrF 1 -␣ R 1 ␤ R 1 Dimers and ␣ R 3 ␤ R 3 Hexamers-Active dimers, but no hexamers, have been assembled by incubation of the isolated ␣ R and ␤ R monomers for 5 min at 35°C in the presence of MgATP (26). These dimers showed a maximal MgATPase rate of 0.14 units/mg of protein, which remained linear for at least 1 h at 35°C. In search for conditions that might enable the assembly of ␣ R 3 ␤ R 3 hexamers, the urea-solubilized ␣ R and ␤ R subunits (26) were refolded together as described under "Experimental Procedures." They did indeed assemble directly, but again only into ␣ R 1 ␤ R 1 dimers with no indication for the appearance of any larger complexes (Fig.  1A). This simple one-step refolding/assembly procedure enabled the isolation of large amounts of the pooled concentrated dimer peak, which remained very stable in TGN buffer at protein concentrations above 1 mg/ml, even when incubated for 1 h at 22°C (Fig. 1B). However, when diluted to the 10 -20-fold lower protein concentrations used for ATPase activity assays, they remained stable only in presence of either MgADP or MgATP or even CaAD(T)P (not shown).
The closed ␣ 3 ␤ 3 hexamer, resolved in the x-ray crystallographic structure of bovine mitochondrial MF 1 (12), has all six F 1 nucleotide-binding sites arranged at alternating catalytic and noncatalytic ␣/␤ interfaces. The isolated ␣ R 1 ␤ R 1 dimers could therefore have either one of these interfaces. But the fact that their MgATPase activity (26) is similar to that of CF 1 -␣ 3 ␤ 3 (10,34) indicates that these dimers contain the catalytic nucleotide-binding site at their ␣/␤ interface. Their inability to associate into an ␣ R 3 ␤ R 3 hexamer therefore seems to reflect a very specific lower stability of the noncatalytic RrF 1 ␣/␤ interfaces. Indeed from the R. rubrum chromatophore-bound RrF 0 F 1 , only native dimers have been isolated (35). From chloroplast, on the other hand, only unstable ␣ C 3 ␤ C 3 hexamers could be obtained, and they readily dissociated into mixtures of their respective ␣ and ␤ monomers (10,36).
A search for compounds or conditions that can stabilize the noncatalytic RrF 1 ␣/␤ interfaces and enable the association of the ␣ R 1 ␤ R 1 dimers into hexamers has yielded the results demonstrated in Fig. 1, C and D. Incubation of the concentrated stable dimers (Fig. 1B) with NaF and AlCl 3 that form aluminum fluoride (AlF x ), 2 a transition state analog of F 1 nucleotidebinding sites (37)(38)(39)(40), resulted in their partial conversion into the closed ␣ R 3 ␤ R 3 hexamers (Fig. 1C). Full association of these dimers into hexamers was obtained by their incubation with both NaF and AlCl 3 in the presence of either Mg 2ϩ or Ca 2ϩ and either ADP or ATP at a cation/nucleotide ratio of 1 (Fig. 1D). AlF x was reported earlier to inhibit various F 1 -MgATPases but only after their very specific stepwise preincubation first with low ADP and high MgCl 2 concentrations, then with NaF, and finally with AlCl 3 . The structure of bovine MF 1 , fully inhibited by this procedure, has recently been resolved (41). It shows that both aluminum trifluoride and Mg 2ϩ -ADP were bound, in a quasi-irreversible manner, to the catalytic nucleotide-binding site of the MF 1 -␤ DP subunit. However, under our very different one-step incubation conditions, which associated practically all the ␣ R 1 ␤ R 1 dimers into ␣ R 3 ␤ R 3 hexamers (compare Fig. 1, B and D), the RrF 1 -MgATPase activity was hardly affected by AlF x (see Fig. 5). Furthermore, these ␣ R 3 ␤ R 3 hexamers remained fully stable only in the presence of AlF x and Mg-or CaAD(T)P at cation/AD(T)P ratios of 1 (Fig. 1, E and F). So under our association conditions AlF x does not bind irreversibly to the catalytic ␣/␤ interface of the dimers and does not inhibit their MgATPase activity. It rather seems to bind in a reversible manner to the open noncatalytic nucleotide-binding site on the dimer ␣ R subunit and facilitate its association with the ␤ R of another dimer, leading to their assembly into the closed hexameric structure.
Assembly of Hybrid WT and Mutant of AlF x resulted in their assembly into a stable, highly active ␣ 3 ␤ 3 ␥ complex (Fig. 2). A recombinant CF 1 -␥ C subunit was used for these studies since there is as yet no available native or recombinant ␥ R . This ␥ C was found to assemble with the native unstable CF 1 -␣ C 3 ␤ C 3 into a stable highly active (11), and its incubation with the isolated ␣ R 1 ␤ R 1 dimers resulted in their assembly into a hybrid ␣ R 3 ␤ R 3 ␥ C (Fig. 2, A and B). This assembly could also be followed by the dramatic 100-fold increase (Fig. 3) of the low 0.1-0.15 units/mg MgATPase activity of the ␣ R 1 ␤ R 1 dimers (26). But unlike the fast assembly of the ␣ R and ␤ R monomers into dimers, which reached their maximal MgATPase activity after a 5-min incubation at 22°C (26), this assembly was slow, requiring about 60 min for completion. A very similar time dependence was reported for the assembly of a CF 1 -(␣␤␥)complex from isolated native CF 1 -(␣␤) and a CF 1 -␥ C (10). The increase in activity during the ␣ R 3 ␤ R 3 ␥ C assembly was also fully dependent on the amount of ␥ C , saturating at a molar FIG. 1. Refolding of the RrF 1 -␣ R and ␤ R into ␣ R 1 ␤ R 1 dimers and their further association into stable ␣ R 3 ␤ R 3 hexamers in presence of aluminum fluoride. A, the urea-solubilized ␣ R and ␤ R were folded together, and 15 g of the concentrated product were loaded on a Superdex-200 column and eluted as described under "Experimental Procedures"; B-D, 15 g of the pooled ␣ R 1 ␤ R 1 dimer peak, concentrated to 1.5 mg of protein/ml as described under "Experimental Procedures," were incubated in TGN buffer for 1 h at 22°C with the following additions: B, none; C, 10 mM NaF and 0.5 mM AlCl 3 ; D, as C plus 1 mM MgAD(T)P or CaAD(T)P; E and F were incubated as in D, diluted by 10-fold into TGN buffer, and further incubated for 1 h at 22°C without (E) or with (F) 10 mM NaF, 0.5 mM AlCl 3 , and 1 mM MgAD(T)P or CaAD(T)P. All incubated samples were loaded on the column and eluted as described for A.

FIG. 2.
Size-exclusion HPLC demonstrates the assembly of a hybrid ␣ R 3 ␤ R 3 ␥ C complex by two different procedures. In the first procedure (A and B), the isolated ␣ R 1 ␤ R 1 (A) was incubated at 9.2 g of protein with 5 g of refolded ␥ C for 1 h at 22°C in 100 l of TGN buffer containing 1 mM MgADP (B). In the second procedure (C and D), the individually expressed urea-solubilized ␣ R , ␤ R , and ␥ C subunits were refolded together as described under "Experimental Procedures" (C). The pooled ␣ R 3 ␤ R 3 ␥ C peak of C was concentrated and re-run in D. Each sample was loaded on the column and eluted as described in Fig. 1. Inset, SDS-PAGE profile of the isolated RrF 1 and the pooled ␣ R 3 ␤ R 3 ␥ C peak. ratio of 2 ␥ C /␣ 1 ␤ 1 (Fig. 3, inset). The relatively high amount of ␥ C required for obtaining this saturated activity was due to the tendency of both native and refolded ␥ C to aggregate, since they remained soluble only when stored in a buffer containing 0.3 M LiCl at pH 9.5 (10,11). So when ␥ C was diluted into the incubation buffer with the ␣ R 1 ␤ R 1 , it partially precipitated out during their slow assembly into the hybrid ␣ R 3 ␤ R 3 ␥ C complex. The time-and ␥ C -dependent increase in the MgATPase activity during the assembly is therefore presented in units/mg ␣␤ (Fig.  3).
The hybrid ␣ R 3 ␤ R 3 ␥ C was also assembled by refolding together all three urea-solubilized subunits under the conditions developed for refolding the ␣ R 1 ␤ R 1 dimers (see Fig. 1A). This much simpler one-step refolding procedure, which resulted in direct assembly of the ␣ R 3 ␤ R 3 ␥ C complex (Fig. 2C), enabled the isolation of large amounts of a pure, fully stable hybrid WT complex ( Fig. 2D and inset). It was also used for the assembly of a hybrid mutant complex containing the ␤ R -T159S catalytic site mutant (27). This mutant ␤ R subunit was shown to bind in the presence of small amounts of monomeric ␣ R into a ␤-less chromatophore membrane-bound RrF 0 F 1 , which also lacked about 20% of its ␣ subunit and lost all ATP synthesis and hydrolysis activities. The reconstituted chromatophores regained all their Mg 2ϩ -dependent but none of the Ca 2ϩ -dependent activities (27).
Modulation of the Hybrid WT and Mutant F 1 -␣ R 3 ␤ R 3 ␥ C -ATPase Activities by Their ␥ C Oxidation/Reduction-A unique feature of the chloroplast CF 1 -ATPase activity is its high regulation by the reduction/oxidation of the disulfide bond formed between Cys 199 and Cys 205 in its ␥ C subunit (24). The region containing these cysteine residues is completely missing from respiratory F 1 -␥ subunits, as well as from the ␥ subunit of cyanobacteria and purple photosynthetic bacteria, including the RrF 1 -␥ (25). All ATPase activities of RrF 1 showed indeed no response to either reduction by DTT or oxidation by CuCl 2 (Fig.  4A). However, both ␥ C -containing hybrid WT and mutant F 1 - R 3 ␥ C complexes showed clear redox regulation of their Mg 2ϩ -, Mn 2ϩ -, as well as Ca 2ϩ -dependent ATPase activities, which were about 2-fold higher in the reduced as compared with the oxidized states (Fig. 4, B and C). The MgATPase activity of a hybrid ␣ 3 ␤ 3 ␥ complex containing TF 1 -␣ and ␤ subunits and the CF 1 -␥ C was also shown to be regulated by the ␥ C redox state (42). These results indicate that the specific thiol modulation (24) of ␥ C can be transferred to various hybrid ␣ 3 ␤ 3 ␥ complexes.
The Catalytic Properties of the Isolated RrF 1 Dimers, Hexamers, and Both Hybrid WT and Mutant ␣ R 3 ␤ R 3 ␥ C Complexes-An additional and more general tight regulation of the MgATPase activity of both chloroplasts and chromatophores is their sensitivity to inhibition by excess free Mg 2ϩ ions (22,23). Both isolated hybrid complexes showed the same optimal dependence on MgCl 2 concentrations as the native RrF 1 , reaching maximal levels at a Mg/ATP ratio of 0.5 ( Fig. 5 and inset). However, their much higher MgATPase activities showed a lower sensitivity than RrF 1 to inhibition by excess free Mg 2ϩ ions, being only 50% inhibited as compared with the RrF 1 80% at an Mg/ATP ratio of 2.5. On the other hand, the similar, very low MgATPase activities of the dimers and the AlF x -stabilized hexamers were not subject to any regulation by excess free Mg 2ϩ ions. They showed only a simple saturation curve, with no inhibition even at a Mg/ATP ratio of 2.5 (Fig. 5, inset). The MgATPase activity of RrF 1 (ϩAlF x ), which was incubated and assayed under the conditions used for assembly of the ␣ R 3 ␤ R 3 hexamers in presence of AlF x , was only slightly lower than that of native RrF 1 and retained its full pattern of inhibition by excess free Mg 2ϩ ions (Fig. 5, inset). These results indicate that the completely different response of both dimers and hexamers, as compared with RrF 1 , to increasing MgCl 2 concentrations is due to absence of the ␥ subunit and not to the presence of AlF x .
FIG. 3. Time-dependent assembly of a highly active hybrid ␣ R 3 ␤ R 3 ␥ C -ATPase complex. 4.6 g of the isolated ␣ R 1 ␤ R 1 dimers and 2.5 g of a refolded ␥ C were incubated at 22°C in 50 l of TGN buffer with 1 mM MgADP. The MgATPase activity of separate samples was assayed at the indicated intervals with the first point representing a sample mixed directly into the ATPase assay mixture containing 4 mM ATP, 2 mM MgCl 2 , and 50 mM sulfite. The ATPase activity was measured for 5 min at 35°C as described under "Experimental Procedures." Inset, dependence of the increase in the hybrid ATPase activity on the amount of ␥ C . Samples containing 9.2 g of the isolated dimers were incubated for 1 h at 22°C with increasing amounts of ␥ C and assayed for their activity as described above. The properties of the CaATPase activities of the RrF 1 and both hybrid WT and mutant complexes were very different from their MgATPase activities (compare Figs. 5 and 6). They were dependent on the presence of CaCl 2 , reaching saturation at a Ca/ATP ratio of 1.0, but showed no inhibition even at a ratio of 5.0. They were thus rather similar to the MgATPase activities of the dimers and hexamers (see Fig. 5

, inset).
A detailed comparison of the Ca 2ϩ -and Mg 2ϩ -dependent ATPase activities of RrF 1 and all assembled complexes (Table  I) demonstrated a number of additional most interesting differences as follows: 1) Between the dimers and hexamers. The dimers have practically no CaATPase activity, although their MgATPase is similar in its rates and properties to that of the hexamers (Fig. 5 and Table I). These results illuminate clear differences in the structure of catalytic nucleotide-binding sites occupied by Ca 2ϩ versus Mg 2ϩ , since CaATP can bind to these sites and enable the appearance of CaATPase activity only in the closed hexamer. On the other hand, the identical and very low dimer and hexamer MgATPase activities, which in the dimers cannot have any catalytic cooperativity, suggest its absence also in the hexamers. Indeed, both dimer and hexamer MgATPases do not respond to the usual MgATPase effectors of CF 1 and RrF 1 . They are not stimulated by sulfite nor inhibited by azide (Table I).
2) Between both dimers and hexamers and the RrF 1 (ϮAlF x ) as well as the hybrid WT complex. The ␥-containing complexes show much higher Ca-and MgATPase activities as well as clear differences between the functional properties of these two ATPase activities. Their MgATPases are tightly regulated by excess free Mg 2ϩ ions (Fig. 5), highly stimulated by sulfite and methanol (not shown) and inhibited by azide (Table I). But their 3-10-fold higher CaATPase activities are not regulated (Fig. 6) and, as both dimer and hexamer MgATPase activities, do not respond to any tested MgATPase effectors (Table I).
3) Between the RrF 1 (ϩAlF x ) Ca 2ϩ -and Mg 2ϩ -dependent ATPase activities. Surprisingly, although AlF x does not affect the RrF 1 -MgATPase ( Fig. 5 and Table I), it inhibits by 70% the 10-fold higher CaATPase (Table I). In light of these results the similar Mg-and CaATPase activities in the ␣ 3 R ␤ R 3 hexamers might be misleading, because their low CaATPase is the AlF xinhibited activity. There is no possible way to confirm this suggestion in the RrF 1 hexamers since they can assemble only in presence of AlF x (Fig. 1). But in the TF 1 -␣ 3 ␤ 3 hexamers, which were assembled without AlF x , the CaATPase activity is 5-fold higher than their MgATPase, although the whole native TF 1 has a 10-fold lower Ca-than MgATPase activity (43). The specific inhibition of the RrF 1 -CaATPase by AlF x , which has not been tested on any other F 1 -CaATPase activity, provides an additional clear difference of functional properties of the RrF 1 -CaATPase and MgATPase activities. 4) Between the hybrid WT and mutant complexes. Both RrF 1 and the hybrid WT showed a 3-10-fold higher CaATPase than MgATPase activities. On the other hand, the hybrid mutant showed a much higher MgATPase activity than either RrF 1 or the hybrid WT complex but a lower CaATPase (Figs. 5 and 6). Its CaATPase, as all other CaATPase activities, was not regulated and did not respond to any tested MgATPase effector. But the very high MgATPase activity of this mutant was also much less responsive to these effectors (Table I). DISCUSSION In this study large amounts of highly active hybrid WT and mutant photosynthetic F 1 -␣ 3 ␤ 3 ␥ complexes were assembled by refolding the recombinant R. rubrum RrF 1 -␣ R and WT ␤ R (26) or mutant ␤ R -T159S subunits (27) together with the spinach CF 1 -␥ C (11). All ATPase activities of both isolated hybrid complexes showed, unlike those of RrF 1 , the specific thiol modulation (24) of their unique ␥ C disulfide bond. Also all ATPase activities of the hybrid WT ␣ R 3 ␤ R 3 ␥ C , which were between 9and 30-fold higher than those of the ⑀-containing RrF 1 (Table  I), retained the catalytic properties of both RrF 1 and CF 1 -ATPases. This includes the specific regulation of the photosyn-  (Table I). This also includes a much higher Ca 2ϩ -than Mg 2ϩ -dependent ATPase activity that is not inhibited by increasing Ca 2ϩ concentrations nor by azide and is not stimulated by sulfite.
This hybrid WT ␣ R 3 ␤ R 3 ␥ C complex provides a most suitable candidate for studies aimed at elucidating the molecular mechanism involved in the ␥ C thiol modulation of its high, ϳ40 units/mg, Mg 2ϩ -and Ca 2ϩ -dependent ATPase activities. Two other hybrid ␣ 3 ␤ 3 ␥ subcomplexes, exhibiting the regulatory thiol modulation of ␥ C , were constructed with the TF 1 -␣ and ␤ subunits. In the first report (42) the TF 1 subunits were mixed with a recombinant ␥ C , but the isolated hybrid showed at least a 10-fold lower MgATPase activity than the hybrid WT ␣ R 3 ␤ R 3 ␥ C complex. In a more recent report (44) a mutant TF 1 -␣ 3 ␤ 3 ␥ complex was constructed by replacing 111 amino acid residues from the central region of the TF 1 -␥ with 148 residues of the homologous region from spinach ␥ C , including the regulatory stretch with Cys 199 and Cys 205 . This mutant complex was expressed and purified in large amounts and responded to the ␥ C thiol modulation, but even its DTT-reduced MgATPase activity reached at the most 5 units/mg. Furthermore, no CaATPase activity has been reported in this mutant TF 1 -␣ 3 ␤ 3 ␥ complex, probably because in the native TF 1 , unlike in RrF 1 (Table I) and CF 1 (30), the CaATPase activity is 10-fold lower than its MgATPase activity (43).
The hybrid WT ␣ 3 R ␤ 3 R 3 ␥ C complex provides also a promising system for following the possible CaATPase as well as MgAT-Pase-induced ␥ C rotation, which has not been measured as yet in any photosynthetic F 1 complex. The CaATPase activity has not been tested as an inducer of any F 1 -␥ rotation. It is, however, a most important candidate for such assays because, although it reaches in both RrF 1 and the hybrid WT complexes even higher rates than those of the MgATPase, it has very different catalytic properties. CaATPase, unlike the MgAT-Pase, appears only in the AlF x -stabilized, closed ␣ R 3 ␤ R 3 hexameric structure (Table I). These results suggest that the Ca 2ϩ binding affinity to the RrF 1 catalytic nucleotide-binding sites is lower than that of Mg 2ϩ , which induces a similar MgATPase activity in the open dimers as in the closed hexamers. A lower binding affinity of Ca 2ϩ could lead to its lower catalytic cooperativity. Indeed, the RrF 1 -CaATPase, unlike its MgATPase, is not inhibited at all by azide (Table I). Furthermore, the similar low MgATPase activity of the RrF 1 -␣ R 1 ␤ R 1 dimers and ␣ R 3 ␤ R 3 hexamers, which in the open dimers has certainly no catalytic cooperativity, is also fully resistant to inhibition by azide (Table I). Azide, which was used as the inhibitor of ␥ rotation inside the ␣ 3 ␤ 3 hexamer cavity (17,18), has recently been suggested to block the signal transmission between catalytic sites, which leads to positive catalytic cooperativity in all F 1 -MgATPases (45). An additional, most important difference between Ca 2ϩ -and Mg 2ϩ -or Mn 2ϩ -dependent ATPase activities was recorded in the membrane-bound RrF 0 F 1 where the CaATPase was not coupled to proton translocation and Ca 2ϩ did not enable any ATP synthesis (23). So in the hybrid WT ␣ R 3 ␤ R 3 ␥ C the catalytic nucleotide-binding site in presence of Ca 2ϩ has no clear catalytic cooperativity and is decoupled from any proton translocation. In a recent study on ␥ rotation in a genetically engineered EcF 1 -␣ 3 ␤ 3 ␥ containing an uncoupled mutation of ␥Met 23 to Lys (18), the mutant ␥ was found to rotate rather similarly to the WT ␥. This unexpected capacity of the uncoupled ␥ subunit to rotate was explained by suggesting that its defective coupling might be after ␥ rotation, possibly in the interactions between the F 1 and F 0 sectors (46). This explanation cannot hold for a catalytic site bound CaATP that cannot induce any proton translocation. The highly active CaATPase of the hybrid WT ␣ R 3 ␤ R 3 ␥ C is therefore a very interesting target for comparing the capacity of its Ca 2ϩ versus Mg 2ϩ -occupied catalytic nucleotide-binding sites to induce catalytic cooperativity and/or rotational catalysis.
Another interesting target for such studies is the hybrid mutant ␣ R 3 (␤ R -T159S) 3 ␥ C . The ␤ R -Thr 159 is equivalent to the MF 1 ␤-Thr 163 , which was identified as a ligand to Mg 2ϩ in the catalytic nucleotide-binding sites of the bovine heart crystal structure (12). This fully conserved residue has been mutated to serine in several respiratory F 1 -ATPases (47)(48)(49) as well as in Chlamydomonas reinhardtii CF 1 (50). All of them showed, as our hybrid mutant ␣ R 3 (␤ R -T159S)␥ C (Table I), a very high MgATPase activity, with a much lower sensitivity to stimulation by sulfite or methanol and to inhibition by azide (Table I). But our hybrid mutant also has a 3-4-fold lower CaATPase activity that has not been followed in the other mutants. Since the hydroxyl group of serine is less nucleophilic than that of threonine, it would lower the bond energy between serine and both divalent cations and decrease their binding affinity. In the hybrid mutant, the Mg 2ϩ -occupied binding site would thus become rather similar to the lower affinity Ca 2ϩ -occupied catalytic site of our hybrid WT complex, which shows very high rates but lower or no catalytic cooperativity. But the mutant Ca 2ϩoccupied catalytic sites would reach an even lower affinity that does already drastically reduce its overall activity (Table I).
In the membrane-bound RrF 1 this ␤ R -T159S mutant was, however, as effective as the WT ␤ R in restoring the proton-translocating Mg 2ϩ -dependent ATP synthesis and hydrolysis. But it could not restore any membrane-bound CaATPase activity (27). These results demonstrate that there must be two sets of clear differences in I Ca 2ϩ -and Mg 2ϩ -dependent ATPase activities of RrF 1 and the assembled RrF 1 -(␣␤) dimers, hexamers, and hybrid (␣␤␥) complexes containing the chloroplast CF 1 ␥ Assembly and/or isolation of all complexes is described under "Experimental Procedures." The ␣ R 3 ␤ R 3 hexamers were assembled and assayed in the presence AlCl 3 and NaF, which form AlF x , and the RrF 1 (ϩAlF x ) was incubated and assayed under identical conditions. The ATPase activities were measured as described under "Experimental Procedures" with 4 mM ATP and 2 mM MgCl 2 or 4 mM CaCl 2 , either with no additions (None) or with 50 mM sulfite or 2 mM azide. the geometry of either the WT or mutant ␤ R catalytic sites when occupied by Ca 2ϩ as compared with Mg 2ϩ . One is operating in the soluble state and a different one in the membrane-bound state, whose ␤ R -T159S-containing RrF 0 F 1 shows the maximal difference between the fully operative proton-coupled Mg 2ϩ -occupied sites and the complete absence of any active Ca 2ϩ -occupied catalytic sites. The similar catalytic properties of the highly active hybrid mutant MgATPase and the hybrid WT CaATPase make them very promising tools for obtaining information on the as yet unknown factors that couple ATP binding and hydrolysis to rotational motion of the ␥ subunit of the catalytic F 1 -ATPase. Since azide does not inhibit the CaATPase activity, another inhibitor will be required for such comparative studies. A very suitable candidate is the specific CF 1 effector tentoxin, which at low concentrations inhibits but at high concentrations stimulates both CF 1 Ca-and MgATPase activities (30). RrF 1 is, however, completely resistant to tentoxin. We have therefore assembled another set of hybrids composed of WT and mutated ␣ R together with ␤ C and ␥ C . 3 They provide the possibility of assaying both ␥ C thiol modulation and rotational catalysis in the presence of inhibitory as well as stimulating tentoxin concentrations.