Characterization of the catalytic and noncatalytic ADP binding sites of the F1-ATPase from the thermophilic bacterium, PS3.

Two classes of ADP binding sites at 20 degrees C have been characterized in the F1-ATPase from the thermophilic bacterium, PS3 (TF1). One class is comprised of three sites which saturate with [3H]ADP in less than 10 s with a Kd of 10 microM which, once filled, exchange rapidly with medium ADP. The binding of ADP to these sites is dependent on Mg2+. [3H]ADP bound to these sites is removed by repeated gel filtrations on centrifuge columns equilibrated with ADP free medium. The other class is comprised of a single site which saturates with [3H]ADP in 30 min with a Kd of 30 microM. [3H]ADP bound to this site does not exchange with medium ADP nor does it dissociate on gel filtration through centrifuge columns equilibrated with ADP free medium. Binding of [3H]ADP to this site is weaker in the presence of Mg2+ where the Kd for ADP is about 100 microM. [3H]ADP dissociated from this site when ATP plus Mg2+ was added to the complex while it remained bound in the presence of ATP alone or in the presence of ADP, Pi, or ADP plus Pi with or without added Mg2+. Significant amounts of ADP in the 1:1 TF1.ADP complex were converted to ATP in the presence of Pi, Mg2+, and 50% dimethyl sulfoxide. Enzyme-bound ATP synthesis was abolished by chemical modification of a specific glutamic acid residue by dicyclohexylcarbodiimide, but not by modification of a specific tyrosine residue with 7-chloro-4-nitrobenzofurazan. Difference circular dichroism spectra revealed that the three Mg2+ -dependent, high affinity ADP binding sites that were not stable to gel filtration were on the alpha subunits and that the single ADP binding site that was stable to gel filtration was on one of the three beta subunits. It has also been demonstrated that enzyme-bound ATP is formed when the TF0.F1 complex containing bound ADP was incubated with Pi, Mg2+, and 50% dimethyl sulfoxide.

Two classes of ADP binding sites at 20 "C have been characterized in the F1-ATPase from the thermophilic bacterium, PS3 (TF1). One class is comprised of three sites which saturate with [3H]ADP in less than 10 s with a Kd of 10 WM which, once filled, exchange rapidly with medium ADP. The binding of ADP to these sites is dependent on Mgz+. r3H]ADP bound to these sites is removed by repeated gel filtrations on centrifuge columns equilibrated with ADP free medium. The other class is comprised of a single site which saturates with [3H]ADP dissociated from this site when ATP plus M g + was added to the complex while it remained bound in the presence of ATP alone or in the presence of ADP, Pi, or ADP plus Pi with or without added Mg+. Significant amounts of ADP in the 1:l TFx*ADP complex were converted to ATP in the presence of Pi, Mg2+, and 50% dimethyl sulfoxide. Enzyme-bound ATP synthesis was abolished by chemical modification of a specific glutamic acid residue by dicyclohexylcarbodiimide, but not by modification of a specific tyrosine residue with 7-chloro-4-nitrobenzofurazan. Difference circular dichroism spectra revealed that the three Mg2+-dependent, high affinity ADP binding sites that were not stable to gel filtration were on the a subunits and that the single ADP binding site that was stable to gel filtration was on one of the three / 3 subunits. It has also been demonstrated that enzyme-bound ATP is formed when the TFo*F1 complex containing bound ADP was incubated with Pi, Mg2+, and 50% dimethyl sulfoxide.
The energy transducing ATP synthase complexes of bacterial plasma membranes, chloroplast thylakoid membranes, the mitochondrial inner membrane, and the chromatophore membranes of photosynthetic bacteria catalyze ATP synthesis driven by electron transport processes (1-4). The enzyme is composed of an integral membrane protein sector, Fo, which mediates H+ conduction, and a peripheral protein sector, F1, which contains the catalytic site for ATP synthesis. Fl is * This work was supported by Grant 5858D117 from the Ministry of Culture and Education of Japan, by United States Public Health Service Grant GM-16974, and by National Science Foundation Grant DMB-8417723. 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. composed of five different polypeptide chains designated a-c in order of decreasing Mr. The molecular weight of F, from several sources is about 380,000 and the subunit stoichiometry is a3p3y8c (5-8). The results obtained from chemical modification studies indicate that the active sites of F1 are located on the ,8 subunits alone (6,9) or at the interface of a! and p subunits (10, 11).
The isolated a and p subunits of bacterial Fl-ATPases can bind ATP or ADP (12)(13)(14). Excluding CF1,l for which only three adenine nucleotide binding sites have been described (E), it is generally accepted that here are six adenine nucleotide binding sites in the Fl-ATPases (16). With the exception of TF, which, when isolated, is free of nucleotides, the F1-ATPases isolated from various other sources contain 1-4 mol of endogenous adenine nucleotides (15,16). These ATPases will bind an additional 2-4 mol of adenine nucleotides/mol of enzyme when incubated with exogenous nucleotides. It has been generally accepted that the bovine mitochondrial F1-ATPase contains three tight adenine nucleotide binding sites which are not exchangeable with medium adenine nucleotides, and that there are three loose adenine nucleotide binding sites which are exchangeable (17, 18). The former are noncatalytic sites, the function of which is uncertain, and the latter sites are catalytic. As is implied in the work of Penefsky and his colleagues, the catalytic sites, owing to cooperative interactions, show a wide range of affinities for adenine nucleotides. Cross and Nalin (18) have shown that the noncatalytic sites also exhibit a range of affinities for adenine nucleotides (18). Therefore, tightness or looseness of binding are not useful parameters for distinguishing catalytic sites from noncatalytic sites.

5714
complex which was slowly inactivated by DCCD. We report here a more thorough characterization of the TF,. ADP complexes formed in the presence and absence of M$+ that includes identification of the subunits with which the different complexes are associated and identification of the ADP binding site that is involved in enzyme-bound ATP synthesis in the presence of 50% dimethyl sulfoxide (21).

EXPERIMENTAL PROCEDURES
M~terials-[2,8-~H]ADP was obtained from New England Nuclear. ADP and ATP were purchased from Kyowa Hakko. Dimethyl sulfoxide and other chemical reagents were the products of Wako Chemicals, Tokyo. Commercial ADP was purified on Dowex 1-X4 resin to remove contaminating ATP as follows. Dowex 1-X4 in a 3-ml column was washed with 20 ml of 1 M HC1 and then washed with water until pH of the effluent was that of the water used as wash. After the column was equilibrated with 60 mM HC1, 1.7 ml of 0.3 M ADP in 60 mM HCl, pH 1.5, was applied on the column. ADP did not bind and was eluted from the column with 60 mM HCl. The combined, collected fractions containing ADP were neutralized with NaOH. After determination of the ADP concentration, the solution was divided into 100-pl aliquots which were stored at -80 "C until used. A fresh aliquot of the ADP solution was used for each experiment. This procedure decreased the amount of contaminating ATP from 0.6 to 0.02%. Pyruvate kinase (from rabbit skeletal muscle, specific activity 500 units/mg) and lactate dehydrogenase (from rabbit skeletal muscle, specific activity 1200 units/mg) were purchased from Sigma. The growth, harvesting, and lysis of the thermophilic bacterium, PS3, and the purification of TF, were carried out as described previously (22).
To 12 mg of TF, dissolved in 3.4 ml of 50 mM MES-NaOH, pH 5.6, Preparation of Inactive Derivatives of TF,: DCCD Inactivationwere added 68 pl of 10 mM DCCD (Aldrich) in ethanol. After 40 min at 23 "C a second 68-p1 dose of 10 mM DCCD was added. The reaction mixture was then incubated an additional 50 min at which time 1.7 g of solid ammonium sulfate was dissolved. The resulting protein precipitate was removed by centrifugation and was dissolved in a minimal volume of distilled water.
Quinacrine Mustard Inactivation-To 33 mg of TFI dissolved in 15 ml of 50 mM triethanolamine sulfate, pH 7.3, containing 10 p M CDTA were added 150 p1 of 50 mM quinacrine mustard (Sigma) in ethanol. The reaction mixture was incubated for 30 min at 23 "C at which time 150 pl of 0.5 M dithiothreitol were added to quench residual reagent. Solid ammonium sulfate (15 g) was dissolved in the solution and the resulting protein precipitate was collected by centrifugation and was then dissolved in a minimal volume of water.
MES-NaOH, pH 5.6, were added 52 pl of 10 mM EEDQ (Aldrich EEDQ Inactiuation-To 9 mg of TF, dissolved in 2.6 ml of 50 mM Gold Label). The mixture was incubated at 23 "C. Additional 52-pl doses of 10 mM EEDQ were added 40 and 80 min after initiating the inactivation. After 140 min, 1.3 g of solid ammonium sulfate were added to precipitate the enzyme. The precipitated protein, collected by centrifugation, was dissolved in a minimal volume of distilled water.
The 0-Nbf Derivative of TF1-To 22 mg of TFl dissolved in 6.3 ml of 50 mM Tris sulfate, pH 8.0, were added 126 p1 of 10 mM Nbf-C1 (Sigma) in ethanol. After 50-min incubation at 23 "C, 3 g of solid ammonium sulfate were added to precipitate the protein. The precipitate was collected by centrifugation and dissolved in a minimum volume of distilled water.
The N-Nbf Derivative of TF1-To 9.3 mg of lyophilized 0-Nbf-TF1 dissolved in 0.93 ml of 10 mM triethanolamine sulfate, pH 7.3, were added 20 p1 of ammonium hydroxide to raise pH to 9.9. The O-+N migration of the Nbf group was monitored spectrophotometrically and by loss of reactivation of the enzyme promoted by dithiothreitol (23). The migration was complete after 5 h. No reactivation due to hydrolysis of the 0-Nbf group was observed.
Preparation of the TFl.ADP and TF, .ADP.Mg Complexes-TF, at concentrations of 1-10 p~ was incubated for 3 h at 20 "C with 500 p M ADP or [3H]ADP in 50 mM triethanolamine sulfate, pH 7.3, containing 10 yM CDTA. The solution was subjected to centrifugeelution on a I-ml syringe column (24) of Sephadex G-50 equilibrated with the same buffer. The effluent was incubated for 12 h at 20 "C at which time it was reapplied to a 1-ml centrifuge column equilibrated with 50 mM triethanolamine sulfate, pH 7.3, containing 10 nM CDTA.
The content of ADP ranged from 0.8 to 1.1 mol/mol of TF1. The TF, . ADP. Mg complex was prepared by incubating the TF1 .ADP complex with 1 mM MgSO, in 50 mM triethanolamine sulfate, pH 7.3, for 10 min at 20 "C. The mixture was then subjected to centrifugeelution on a 1-ml syringe column equilibrated with 50 mM triethanolamine sulfate, pH 7.3, containing 10 p~ CDTA. Membrane Filter Assay for Bound PHIADP-Samples containing solutions of TF1 in the presence of [3H]ADP were applied to 0.45-pm Millipore nitrocellulose membrane filters under suction. The filters were washed with 2 ml of 50 mM triethanolamine sulfate, pH 7.3, containing 1 mM MgS0,. The washed filters were dried at 45 "C and were then counted in scintillation vials which contained 10 ml of Aquasol-2 universal counting mixture from New England Nuclear.
Circular Dichroism Measurements-Circular dichroism spectra were measured at 20 "C with a JASCO J-50 circular dichroism spectrometer equipped with a computerized data processor using a quartz cuvette with a 3-mm light path. The difference spectra obtained were the average of four or eight scans from 300 to 250 nm and were generated with the computerized data processor attached to the instrument.
HPLC Analysis of Adenine Nucleotides-Analyses of adenine nucleotides were performed by HPLC on a TSK DEAE-2SW column, with a slight modification of a previously published procedure (21). To 50-100-pl sample solutions were added 2 p1 of 70% perchloric acid. The acidified solutions were placed in an ice bath for 30 min and then denatured protein was removed by centrifugation. The supernatants were neutralized with 5 pl of 5 M K&03 and were placed in an ice bath for 30 min. The precipitated Kclo, was then removed by centrifugation. The supernatant was injected onto an HPLC column that was equilibrated and eluted with 0.4 M sodium phosphate, pH 6.0, at 0.8 ml/min at room temperature. Adenine nucleotides were detected by absorbance at 254 nm. Concentrations were determined by automatic integration of the eluting peaks which were compared with those of standard solutions of known ATP and ADP which were subjected to the same analysis.
Other Analytical Methods-ATPase activity was assayed at 20 "C by monitoring inorganic phosphate release as described earlier (22). In experiments illustrated in Fig. 5, ATPase activity was monitored at 20 "C by a coupled enzyme assay. Reaction mixtures contained 50 mM Tris-SO4, pH 8.0,5 mM MgS04, 10 mM KCl, 2 mM phosphoenolpyruvate, 0.32 mM NADH, 10 pg/ml pyruvate kinase, and 10 yg/ml lactate dehydrogenase. Reactions were initiated by the addition of TF, or the TF, . ADP complexes to 1 ml of the assay mixture and the rate of decrease of absorbance at 340 nm was monitored. Protein was determined with the dye binding method of Bradford (25). When perchloric acid precipitation preceded protein assays (as was used for the data presented in Table 111) to remove residual triethanolamine which interferes with the Lowry protein assay, the precipitate was resuspended in 200 pl of 5% perchloric acid and centrifuged. The precipitate was then dissolved in 100 pl of 1% sodium dodecyl sulfate solution and protein was determined by the Lowry method (26). The molecular weight of TF, was taken to be 380,000 (7).

ADP Binding
Sites of the TFl-ATPase 5717 bound, did not exchange with medium ADP. However, a small amount of [3H]ADP (0.14 mol/mol TF1) bound rapidly, which in turn was rapidly dissociated when nonradioactive ADP was added to the mixture as shown in Fig. 2 4  We have previously shown that when mixtures of TFI and ADP were incubated for 2 h in the presence or absence of M e and then were subjected to repeated gel filtration on centrifuge columns equilibrated in the absence of ADP, a 1:l TF1 .ADP complex or a 1:l:l TFI .ADP.Mg2+ complex were isolated. These complexes, especially the 1:l TF,.ADP complex were stable to many further manipulations (20). The rates of formation of these complexes, which are stable to gel filtration on centrifuge columns, are illustrated in Fig. 3A.
The binding of ADP to form the complexes was slow, exhibiting a second order rate constant of 2 and 4 s-l M-' in the presence or absence of M$+, respectively. From the data of Fig. 3B, Kd values of about 100 and 30 ~L M were estimated for complexes formed in the presence and absence of Mg2+, respectively. Under these conditions M$+ depresses ADP binding. Table I summarizes the parameters obtained from assessment of ADP binding to TF, by the membrane filter method and by gel filtration on centrifuge columns. Four ADP binding sites were apparent in these analyses. Three of them bound ADP with relatively high affinity, exchanged rapidly with medium ADP, and depended on the presence of Mg2+. ADP binding to the fourth site was of relatively low affinity, was nonexchangeable, and was slightly inhibited by M P . It is clear that the single, nonexchangeable ADP binding site characterized by the membrane filter assay is the same as the binding site which is stable to multiple gel filtrations on centrifuge columns. The 1:l:l TF1.
[3H]ADP.M$+ complex was in fact obtained by repeated gel filtrations of the same mixture that was used to determine nonexchangeable binding of [3H]ADP.
Subunit Localization of the ADP Binding Sites by Circular Dichroism-Characteristic changes in circular dichroism occur when the isolated (Y or p subunits bind adenine nucleotides as described previously (12). Based on changes in circular dichroism, it has been possible to identify the subunits which contain the three high affinity, exchangeable ADP binding sites and the single low affinity, nonexchangeable ADP binding site. Fig. 4 shows the circular dichroism difference spectra obtained for the 1:l TF, -ADP complex and for the complexes formed between ADP and the isolated (Y or p subunits in the presence and absence of M$+. By comparison of Fig. 4b with 4c it is clear that the 1:l stable TFI. ADP complex results in a difference spectrum which is characteristic of ADP binding to the , B subunit. It is interesting that the addition of ADP to TF, in the absence of M$+ gave a / 3 type circular dichroism difference spectrum (Fig. 4d) which indicates that ADP binds only to the p subunit in the absence of Mg2+. The magnitude of the difference spectrum of this mixture was 3 times that of the complex of the isolated subunit with ADP (Fig. 4b) and    (Fig.  4c). Provided that the circular dichroism changes associated with ADP binding to the isolated /3 subunits represents the whole population of the p subunits and that the magnitude of the circular dichroism change for the isolated subunit is the same as that for the 6 subunit in the intact enzyme, these The reactions were initiated for the addition of 16 pg of TF2 or 16 pg of the TFI. ADP complex isolated as described under "Experimental Procedures." data suggest that each subunit binds ADP when TF1 is incubated with 250 ~L M ADP, but that ADP dissociates from two of them during the membrane filter assay or during gel filtration on centrifuge columns.
When TF1 was incubated with M$+ and low concentrations of ADP, the circular dichroism spectrum characteristic of the a subunit developed as shown in Fig. 4g. A difference spectrum between TF, in the presence of ADP and M$+ and TF, in the presence of ADP alone is shown in Fig. 4h. This shows that in the presence of M e and ADP, TF1 develops a circular dichroism spectrum characteristic of the complex of the a subunit with Mg2+ plus ADP. The magnitude of the spectra shown in Fig. 4, g and h per mol of TF1 is greater than three times on a molar basis than the spectrum generated by the interaction of ADP with the a subunit shown in Fig. 3e. This might reflect that the isolated cr subunit might have been partially denatured during preparation.
From these results it is concluded that the M$+-dependent, exchangeable sites are located on each of the three a subunits and that the single, nonexchangeable binding site which is stable to membrane filtration and repeated gel filtration on centrifuge columns is located on one of the p subunits.
Properties of the 1:l TF, .ADP Complex-Once isolated, the 1:1 TFl.ADP complex is stable in the absence of medium ADP for at least three days at 4 "C as described earlier (20). Mg+ affected the stability of the complex, it appears that enzyme turnover is required to release ADP from the 1:l complex. AMP-PNP is a strong, competitive inhibitor of the hydrolytic reaction catalyzed by TF1; presumably by OCCUpying at least one of the catalytic sites of the enzyme.

Bound [3H]ADP in this complex
The 1:l TF, .ADP complex has the same V,,, and K, values as free TF1 for the steady state hydrolysis of ATP. A short lag time precedes the attainment of the steady state Vmax for TF, as shown in Fig. 5. This lag time increased when the 1:l TF, .ADP complex was assayed, again as shown in Fig. 5. The length of the lag is dependent on the ATP concentration in the assay mixtures as is also illustrated by comparison of the rates observed with 1:l mM ATP or 0.27 mM ATP as substrate shown in Fig. 5.
Synthesis of Bound ATP from ADP Bound by TF1 in the Presence of 50% Dimethyl Sulfoxide-It has been demonstrated that TF, will synthesize enzyme-bound ATP from enzyme-bound ADP in the presence of Pi, M$+, and 50% M. Yoshida, unpublished results. dimethyl sulfoxide (21). Although it was not determined directly, the 1:3:3 TF, .ADP. Mg2+ complex was probably present under the conditions used in the earlier experiments. Table I1 shows the results of experiments in which enzymebound ATP was synthesized on the 1:l TF,-ADP complex and on the 1:l:l TF, .ADP. Mg2+ complex. It is clear from the data presented in Table I1 that significant amounts of enzyme-bound ATP were formed from both complexes under appropriate conditions. Thus, the ADP binding site on a single p subunit which is stable to gel filtration and membrane filtration is the site where enzyme-bound ADP is converted to bound ATP. It is interesting that the formation of enzymebound ATP from the 1:l:l TFl . ADP -Mg2+ complex does not depend on medium M e . Therefore, the components of this complex are directly involved in enzyme-bound ATP formation. These results strongly suggest that ADP and M$+ in these complexes are bound to one of the catalytic sites of TF1, which in the presence of bound Pi and appropriate other conditions can be converted to a TF1. ATP. M P complex.
To test the possibility that the other class of ADP binding sites which are dependent on Mg+ might also form enzymebound ATP in the presence of 50% dimethyl sulfoxide, these sites were filled at low ADP concentration (1.12 p M ) in the presence of Mg2+ where the single, stable site remained unoccupied. Under these conditions, ADP bound only to the a! subunits as determined by a difference circular dichroism spectrum. When this complex was incubated with Pi, M$+, and 50% dimethyl sulfoxide, an insignificant amount of ATP was synthesized. Therefore, ADP bound to the a! subunits cannot be converted to ATP in the presence of Mg2+, Pi, and 50% dimethyl sulfoxide.  Table 111. Of the ATPase inactivators tested, only treatment with DCCD produced a nearly complete loss in the capacity of the enzyme to promote enzyme-bound ATP formation. Different carboxyl groups are modified when TF1 and MF, are inactivated by DCCD (19, 28). Thus, the observation that TF, modified by DCCD does not carry out enzyme-

Mg complexes
The TF, .ADP and the TF1. ADP. Mg complexes were prepared as described under "Experimental Procedures," except that the last centrifuge column was equilibrated with 100 mM MES-NaOH and 40 mM sodium phosphate, pH 6.4. Enzyme-bound ATP synthesis was carried out in 60-pl reaction mixtures which contained 50% Me2S0, 50 mM MES-NaOH, and 20 mM sodium phosphate, pH 6.3, 48 g of the complexes, and 5 mM MgS04. The reaction mixtures were incubated for 1 h at 23 "C at which time 50 pl were removed and placed in a microfuge tube. Then 2 pl of 70% perchloric acid were added to precipitate the protein. The protein was removed by centrifugation. After neutralization with KzC03 and removal of KC104, the supernatants were analyzed for adenine nucleotides by HPLC and the precipitated protein was assayed as described under "Experimental Procedures."  Lyophilized preparations of native and inactivated TFI, prepared as described under "Experimental Procedures," 0.2-0.4 mg, were dissolved in 0.2 ml of 50 mM triethanolamine sulfate, pH 7.3, which contained 10 p~ CDTA and 2 mM ADP. After incubation for 1 h at 23 "C, 100 pl of these solutions were applied to 1-ml centrifuge columns of Sephadex G-50 which were equilibrated with 50 mM triethanolamine sulfate, pH 7.3, which contained 10 p~ CDTA. The effluents were incubated for 12 h at 23 "C and were then applied to a second centrifuge column equilibrated with 100 mM MES and 40 mM sodium phosphate, pH 6.3. Samples were removed to determine ATP and ADP content and protein concentration. Enzyme-bound ATP synthesis was carried out with 0.1-0.2 mg of TFl or the inactivated enzymes in 50 pi of 50% Me2S0, 4 mM MgSO,, 50 mM MES, 20 mM sodium phosphate, pH 6.3. In one experiment, 2 mM NaN3 was added. The reaction mixtures were incubated for 1 h at 23 "C at which time they were subjected to gel filtration on 1-ml centrifuge columns of Sephadex G-50 which were equilibrated with 50 mM triethanolamine sulfate, pH 7.3, containing 10 p~ CDTA. The ATP content of TF, before the addition of M e 8 0 was less than 0.02 mol/mol of enzvme.

Inhibitor
ATPase treatment treatment ADP ADP ATP TFo. F1. When the precipitated enzyme was repeatedly resuspended in ADP-free buffer and recentrifuged, the bound ATP decreased to 1.6 mol per mol of TFo.Fl. When this isolated complex was subsequently incubated with Pi, Mg2+, and 50% dimethyl sulfoxide and then subjected to centrifugation, a significant amount of enzyme-bound ATP was formed as shown in Table IV. The enzyme-bound ATP formed in these experiments cannot be attributed to free TF,. If TF, had dissociated from the complex, it would have remained in the supernatant after centrifugation.

DISCUSSION
The results presented show that two classes of,binding sites for ADP exist in TF,. One class is comprised of three sites which bind ADP with high affinity ( K d = 10 PM) only in the presence of M$+. ADP bound to these sites dissociates when the enzyme is subjected to gel filtration. The ADP bound to these sites also exchanges rapidly with medium ADP. The other class is comprised of a single site that binds ADP with low affinity in the presence ( K d = 100 p~) and the absence (Kd = 30 p~) of M$+. ADP bound to this site is stable to gel filtration and does not exchange with medium ADP in the absence of enzyme turnover. For convenience, during the

MezSO-dependent A T P synthesis by the TFo.Fl complex
The TFo.F1 complex, 1.86 mg, was incubated for 1 h at 20 "C in 200 p1 of a solution which contained 2 mM ADP, 2 mM MgS04, and 50 mM triethanolamine sulfate, pH 7.3. The solution was diluted with 8 ml of 50 mM triethanolamine sulfate, pH 7.3, containing 1 mM MgS04, and then centrifuged at 140,000 X g for 30 min. The pellet was homogenized with 8 ml of 50 mM triethanolamine sulfate, pH 7.3, containing 1 mM MgSO4 and the suspension was centrifuged under the conditions described above. The washed pellet was homogenized in 1 ml of triethanolamine sulfate, pH 7.3, containing 1 mM MgSO,. The adenine nucleotide and protein concentrations were determined with 150 pl and the remainder at the sample was incubated for 12 h at 20 "C, at which time it was diluted with 8 ml of 50 mM triethanolamine sulfate, pH 7.3, containing 1 mM MgSO, and then centrifuged. Part of the pellet was suspended in 50 mM triethanolamine sulfate, pH 7.3, containing 10 p~ CDTA to assay for adenine nucleotides and protein. The rest of the pellet was suspended in 200 p1 of 20 mM sodium phosphate, pH 6.3,2 mM MgS04, and 50% Me2S0. This mixture was incubated for 1 h at 23 "C, at which time it was diluted with 8 ml of 50 mM triethanolamine sulfate, pH 7.3, which contained 1 mM MgS04. The diluted sample was centrifuged and the pellet was suspended in 100 pl of 50 mM triethanolamine sulfate, pH 7.3, containing 1 mM MgS04 and the protein concentration and adenine nucleotide content were determined as described under "Experimental Procedures." remainder of this discussion the former sites will be called the high affinity ADP binding sites and the latter site will be called the low affinity ADP binding site.
It is also clear from the circular dichroism difference spectra presented that the high affinity ADP binding sites are on the three a subunits and the low affinity ADP binding site resides on a single p subunit. It was also shown that the ADP bound to a subunit in the 1:l TFl. ADP complex is converted to enzyme-bound ATP in the presence of Pi, M$+, and 50% dimethyl sulfoxide at pH 6.3. The ADP in the 1:l:l TF1. ADP.Mg'+ complex, which is also bound to a / 3 subunit, is converted to enzyme-bound ATP in the presence of Pi and 50% dimethyl sulfoxide in a reaction which does not require additional M$+. These results provide the least ambiguous evidence collected to date which shows that the site or sites for ATP synthesis do indeed reside in the p subunits.
Based on the observation that both the a and p subunits are labeled when the Fl-ATPases are inactivated with photoaffinity analogs of ATP containing nitrene (10) or carbene (11) generators esterified to the ribose moiety, it has been suggested that the catalytic site resides at the interface of a and p subunits. Although not ruled out, this contention is not supported by the observations presented here. The circular dichroism difference spectrum observed when ADP binds to the low affinity site of the intact enzyme is remarkably similar to that observed when ADP binds to the isolated p subunit.
This suggests that the circular dichroism difference spectrum observed for the 1:l TF1 ADP complex arises from interactions of bound ADP with the , R subunit only.
The observation that only one of the three @ subunits forms stable 1:l TFl-ADP and 1:l:l TF1. ADP -Mg'+ complexes reflects an asymmetry of nucleotide binding to p subunits. Whether the asymmetry is induced on the binding of ADP to a single, low affinity site, or whether it pre-exists is not known. Since the isolated 1:1 TF,. 13H]ADP complex did not dissociate when it was incubated with nonradioactive ADP and Mg'+ followed by repeated gel filtrations indicates that the asymmetry is retained when all of the ADP binding sites are occupied. Asymmetry of nucleotide binding appears to be characteristic of the F1-ATPases (18,(29)(30)(31)(32)(33)(34)(35). Penefsky and his colleagues (31) have described a single, high affinity catalytic site in MF, which binds Mg . ATP-' with a dissociation constant of 10-l' M. Cross and Nalin (18) have shown that AMP-PNP binds to a single, exchangeable site, presumably in a / 3 subunit, with a dissociation constant of 18 nM. Gautheron's laboratory has shown that the "hysteretic" inhibition, which develops slowly when ADP binds to MF1 in the presence of M$+, is associated with a single binding site which also appears to reside in a / 3 subunit (35). When the 1:l TF1. ADP complex is assayed with an ATP regenerating system, a slowly developing hysteretic inhibition which is characteristic of MF, is not seen but, instead, a transient inhibition or lag is observed which disappears as the assay proceeds. A similar, but much shorter lag is observed when native TF, is assayed with an ATP regenerating system. ADP bound to the low affinity site of TF1 differs from ADP bound to the hysteretic site of MF, in another respect. ADP bound to the hysteretic site of MF, does not dissociate during enzyme tunover (33), while the ADP bound to the low affinity site dissociates when Mg2+ and ATP are added to the enzyme under conditions where turnover occurs.
The noncatalytic ADP binding sites in the a subunits of TF, differ considerably from the noncatalytic nucleotide binding sites described for MF, (17, 18). While the ADP bound to the high affinity binding sites of TF, is removed by gel filtration and exchanges with medium ADP, the nucleotides bound to the noncatalytic binding sites of MF,, which presumably also reside in the a subunits, are not removed by gel filtration and do not exchange with medium nucleotides in the presence or absence of enzyme turnover (17, 18).
Recently Khananshvili and Gromet-Elhanan (36,37) have reported that 2 mol of ADP or ATP bind per mol of the isolated subunit of the R-ATPase from Rhodospirillum rubrum. One site binds ADP with high affinity (Kd = 6.7 pM) in the presence and absence of M$+, while the second site binds ADP with low affinity (Kd = 80 pM) which is absolutely dependent on the presence of M$+. It has been suggested that the low affinity binding site is the catalytic site. The Kd of ADP for the low affinity binding site on the isolated p subunit of F1 from R. rubrum is similar to that observed in this study for the binding of ADP to both the isolated p subunit of TF, and to the catalytic site of intact TF1. However, while ADP binding to the low affinity site of R. rubrum Fl is absolutely dependent on the presence of Mg'+, the stable binding of ADP to either the isolated p subunit of TF1 or to a single catalytic site in the intact enzyme does not depend on M F . Therefore, it is possible that the high affinity binding site on the isolated @ subunit of R. rubrum Fl, which does not depend on the presence of M$+, is equivalent to the low affinity ADP binding site on TF1 which we have shown to be a catalytic site which resides on a p subunit. It would be interesting to compare the circular dichroism spectra generated when ADP binds to the high and low affinity sites of the isolated R. rubrum ( 3 subunit with the difference spectra shown here when ADP binds to the isolated @ subunit of TF,. It is interesting that, of the inactivators of the TF1-ATPase tested, DCCD is the only one which also abolishes the capacity of the soluble enzyme to synthesize enzyme-bound ADP from bound ADP in the presence of medium Pi and dimethyl sulfoxide (27). It has been shown that @-Glu-188 (using the residue numbers of the @ subunit of MF1) is modified when DCCD inactivates the ATPase activity of TF, (19) and that DCCD reacts with P-Glu-199 when it inactivates the MFl-ATPase (28). These observations suggest that P-Glu-188, at least in TFI, plays a role in both the hydrolysis and synthesis of ATP, while @-Glu-199 in MF, does not. The fact that modifications of TFl at the residue equivalent to @-Tyr-311 of MF1 with Nbf-C1 (38,39), at the residue equivalent to @-Lys-162 by migration of the Nbf-group (40,41), or by EEDQ or quinacrine mustard at unknown, essential residues, inactivate ATPase activity without altering enzyme-bound ATP synthesis is curious. Sakamoto and Tonomura (27) have observed that modification of MFl with DCCD or Nbf-C1 also abolishes ATPase activity without severely affecting the capacity of the enzyme to synthesize enzyme-bound ATP from bound ADP in the presence of medium Pi and dimethyl sulfoxide. On the basis of different observations, reports have appeared from other laboratories which have proposed that the catalytic site or sites of Fl might exist in two interconvertible conformations, one geared for ATP hydrolysis and the other geared for ATP synthesis (35,42,43). It is possible that the reagents which have been observed to abolish the hydrolytic reaction catalyzed by the Fl-ATPases without seriously affecting their capacity for enzyme-bound ATP synthesis do so by locking the catalytic sites in a synthetic mode.
The observation that enzyme-bound ATP synthesis is promoted by the intact TFo.Fl complex in the presence of dimethyl sulfoxide shows that this reaction is not just an artifact associated with the soluble enzyme. Feldman and Sigman (44, 45) have reached the same conclusion from experiments with isolated CFl and thylakoid preparations in which enzymebound ATP synthesis was demonstrated by adding high concentrations of Pi to either CFl or the CFo.Fl complex, each containing tightly bound ADP.