Substitution of βGlu201 in the α3β3γ Subcomplex of the F1-ATPase from the Thermophilic Bacillus PS3 Increases the Affinity of Catalytic Sites for Nucleotides

In the crystal structure of bovine mitochondrial F1-ATPase (MF1) (Abrahams, J. P., Leslie, A. G. W., Lutter, R., and Walker, J. E. (1994)Nature 370, 621–628), the side chain oxygen of βThr163 interacts directly with Mg2+coordinated to 5′-adenylyl β,γ-imidodiphosphate or ADP bound to catalytic sites of β subunits present in closed conformations. In the unliganded β subunit present in an open conformation, the hydroxyl of βThr163 is hydrogen-bonded to the carboxylate of βGlu199. Substitution of βGlu201(equivalent to βGlu199 in MF1) in the α3β3γ subcomplex of the F1-ATPase from the thermophilic Bacillus PS3 with cysteine or valine increases the propensity to entrap inhibitory MgADP in a catalytic site during hydrolysis of 50 μm ATP. These substitutions lower K m 3 (the Michaelis constant for trisite ATP hydrolysis) relative to that of the wild type by 25- and 10-fold, respectively. Fluorescence quenching of α3(βE201C/Y341W)3γ and α3(βY341W)3γ mutant subcomplexes showed that MgATP and MgADP bind to the third catalytic site of the double mutant with 8.4- and 4.4-fold higher affinity, respectively, than to the single mutant. These comparisons support the hypothesis that the hydrogen bond observed between the side chains of βThr163and βGlu199 in the unliganded catalytic site in the crystal structure of MF1 stabilizes the open conformation of the catalytic site during ATP hydrolysis.

The F 0 F 1 -ATP synthases found in energy-transducing membranes couple ATP synthesis and hydrolysis to proton or sodium ion electrochemical gradients. They are composed of F 0 , an integral membrane protein complex that mediates ion conduction, and F 1 , a peripheral membrane protein complex containing the catalytic sites. When separated from F 0 as a soluble complex, F 1 is an ATPase composed of five different subunits with ␣ 3 ␤ 3 ␥␦⑀ stoichiometry (1,2). F 1 contains six nucleotide-binding sites. Three of these participate directly in catalysis. The other three, called noncatalytic sites, do not have a well defined physiological role (2). However, it is clear that saturation of these sites with MgATP in MF 1 1 prevents entrapment of inhibitory MgADP when low concentrations of ATP are hydrolyzed (3). The 2.8-Å resolution crystal structure of bovine MF 1 determined by Abrahams et al. (4) shows that the catalytic sites are at ␣/␤ interfaces, with most residues participating directly in catalysis located in ␤ subunits. Noncatalytic sites are located at different ␣/␤ interfaces, with most residues contributing to the binding sites located in ␣ subunits. In the crystal structure, the three noncatalytic sites are homogeneously liganded with MgAMP-PNP. In contrast, catalytic sites are heterogeneously liganded. One (designated ␤ T ) contains MgAMP-PNP; another (designated ␤ D ) contains MgADP; and the third catalytic site (designated ␤ E ) is empty. The ␣ subunits contributing to ␤ T , ␤ D , and ␤ E are designated ␣ T , ␣ D , and ␣ E , respectively (4). In the crystal structure of the completely unliganded ␣ 3 ␤ 3 subcomplex of TF 1 determined at 3.2-Å resolution by Shirakihara et al. (5), the ␣ and ␤ subunits are arranged symmetrically. The common open conformation of ␤ subunits in the ␣ 3 ␤ 3 subcomplex is essentially identical to that of ␤ E in MF 1 , and the common conformation of the ␣ subunits is essentially identical to that of the liganded ␣ subunits in MF 1 .
Although it is generally assumed that the three catalytic sites in F 1 -ATPases exist in tight, loose, and open conformations, scrutiny of the crystal structure of bovine MF 1 reveals only two distinctly different conformations. These are the open, unliganded conformation of ␤ E and the nearly identical, closed conformations in ␤ T and ␤ D liganded with MgAMP-PNP and MgADP, respectively (6). In ␤ T and ␤ D , the hydroxyl oxygen of ␤Thr 163 is directly liganded with the Mg 2ϩ ion that interacts with anionic oxygens of phosphates in bound AMP-PNP or ADP, respectively. In contrast, in ␤ E , the hydroxyl of ␤Thr 163 is 2.9 Å from a carboxyl oxygen of ␤Glu 199 . These groups are 11.4 Å apart in ␤ T and ␤ D , suggesting that the open conformation of ␤ E might be stabilized by the hydrogen bond between the side chains of ␤Thr 163 and ␤Glu 199 . It is interesting that the carboxyl group of ␤Glu 199 in MF 1 or the equivalent residue in Escherichia coli F 1 and spinach chloroplast F 1 is specifically derivatized on inactivation of ATPase activity with dicyclohexylcarbodiimide. In contrast, when TF 1 is inactivated with dicyclohexylcarbodiimide (7-9), ␤Glu 190 , the equivalent of ␤Glu 188 in MF 1 , is derivatized (10).
The catalytic properties of site-directed mutants of the glutamates in TF 1 , E. coli F 1 , and Rhodospirillum rubrum F 1 that are equivalent to ␤Glu 199 in MF 1 also suggest a functional role for this side chain in ATP hydrolysis (11)(12)(13). Substitution of the equivalent residue with glutamine (11,12) or cysteine (12) in TF 1 or E. coli F 1 or with glutamine, lysine, or glycine in R. rubrum F 1 (13) significantly lowers ATP hydrolysis under saturating conditions, whereas ATPase activity is attenuated to a lesser extent when the equivalent residue is substituted with Asp (12). In the crystal structure of MF 1 , ␤Thr 163 is hydrogen-bonded to ␤Glu 199 in the open catalytic site, whereas it is liganded with Mg 2ϩ in the two closed, liganded catalytic sites as illustrated elsewhere (4,6). This suggests that the hydrogen bond between the side chains observed in the empty catalytic site of MF 1 might function to stabilize the open conformation of the catalytic site during catalysis. If this is indeed the case, the affinity of catalytic sites for MgADP and MgATP should increase on substituting ␤Glu 201 in the ␣ 3 ␤ 3 ␥ subcomplex of TF 1 with amino acids containing side chains incapable of hydrogen bonding. To test this possibility, the steady-state kinetics of ATP hydrolysis by mutant ␣ 3 ␤ 3 ␥ subcomplexes of TF 1 with the ␤E201C substitution with and without carboxymethylation and with the ␤E201V substitution have been examined in detail. In addition, the affinities of the ␤E201C/Y341W double mutant for nucleotides have been determined by assessing quenching of the introduced tryptophan by ATP and ADP in the presence of Mg 2ϩ .

EXPERIMENTAL PROCEDURES
Materials-Enzymes and biochemicals used in assays and buffer components were purchased from Sigma. Iodoacetic acid free of iodine was obtained from Sigma. LDAO was purchased from Calbiochem. Iodo[ 3 H]acetic acid was purchased from American Radiolabeled Chemicals. Primers for mutations were purchased from Life Technologies, Inc.
Generation of Mutant Subcomplexes-Plasmid pKK, which carries the genes for the ␣, ␤, and ␥ subunits of TF 1 , was used for mutagenesis and gene expression (14). Polymerase chain reaction was used to prepare mutant expression plasmids using the QuikChange TM site-directed mutagenesis kit (Stratagene). Wild-type pKK was used as the template for the ␤E201C and ␤E201V single mutants, and the ␤Y341W mutant plasmid pKK (15) was used as the template to generate the ␤E201C/Y341W double mutant. The primers 5Ј-GACTTGTACCATGT-GATGAAAGATTCCG-3Ј and 5Ј-GACTTGTACCATTGCATGAAAGAT-TCCG-3Ј (with the changed bases underlined) and their corresponding complementary primers were used in polymerase chain reaction to generate the ␤E201C and ␤E201V mutants, respectively. The plasmids were purified with the Wizard TM Plus Minipreps DNA purification system (Promega). The mutations were confirmed by sequence analysis. The resultant pKK mutant plasmids were expressed in JM103 (unc Ϫ ). Purification of wild-type ␣ 3 ␤ 3 ␥ and all mutant subcomplexes was performed as described (14). All enzymes were stored at 4°C as precipitates in 70% saturated ammonium sulfate.
Analytical Methods-Enzyme stock solutions were prepared by dissolving pelleted ammonium sulfate precipitates obtained by centrifugation in 50 mM Tris-Cl, pH 8.0, containing 1 mM CDTA. After 1 h at room temperature, dissolved enzymes were passed through 1-ml centrifuge columns of Sephadex G-50 equilibrated with 50 mM Tris-Cl, pH 8.0, containing 0.1 mM EDTA. Protein concentrations were determined by the method of Bradford (16) using Coomassie Blue (Pierce).
ATPase activity was determined spectrophotometrically using an ATP regeneration system at 30°C and pH 8.0 (3). Unless indicated otherwise, the Mg 2ϩ concentration in the assay medium was 1 mM in excess of the ATP concentration. The NADH used in the assays contained negligible ADP. Radioactivity was determined by liquid scintillation counting in Ecoscint (National Diagnostics, Inc. Fluorescence Titrations of the ␤Y341W and ␤E201C/Y341W Subcomplexes-Titrations of catalytic sites of the ␤Y341W and ␤E201C/Y341W mutant subcomplexes with nucleotides were determined directly by monitoring quenching of tryptophan fluorescence with a Spex Fluoromax-2 spectrofluorometer according to Weber et al. (17). Titrations were performed with 50 nM solutions of the ␤Y341W or ␤E201C/Y341W mutant enzyme in 50 mM Tris-Cl, pH 8.0, to which was added 50 nM to 1 mM ATP or ADP. In titrations with MgATP or MgADP, MgCl 2 was present in 1 mM excess over ATP or ADP. Fluorescence measurements were recorded 30 s after addition of nucleotide. During titrations with MgATP, fresh enzyme solutions were prepared for each concentration of nucleotide examined.

Comparison of Hydrolysis of 50 M ATP by the Wild-type and
Mutant Subcomplexes-It has been shown that the wild-type subcomplex hydrolyzes 50 M ATP in three kinetic phases (18). A burst rapidly decelerates to an intermediate, slow phase that slowly accelerates to a final rate that approaches the initial rate. Transition from the burst phase to the intermediate phase is caused by turnover-dependent entrapment of MgADP in a catalytic site, and transition from the intermediate phase to the final rate reflects slow binding of ATP to noncatalytic sites that promotes dissociation of the inhibitory MgADP from the affected catalytic site. Earlier studies have shown that only a single, rapid kinetic phase is observed when the wild-type subcomplex hydrolyzes 50 M ATP in the presence of low concentrations of LDAO, indicating little or no entrapment of inhibitory MgADP during turnover. Entrapment of inhibitory MgADP is augmented when the wild-type subcomplex is assayed in the presence of azide (15). To assess the propensity of mutant ␣ 3 ␤ 3 ␥ subcomplexes to entrap inhibitory MgADP during turnover, the characteristics of hydrolysis of 50 M ATP by the wild-type and mutant subcomplexes in the presence or absence of 0.06% LDAO or 1 mM NaN 3 were compared. The results of this comparison are illustrated in Fig. 1. The numbers associated with the traces in Fig. 1 are the approximate specific activities of the subcomplexes at the final 30-s interval recorded. Trace a in panel A-I illustrates the three kinetic phases exhibited when the wild-type ␣ 3 ␤ 3 ␥ subcomplex hydrolyzed 50 M ATP. In contrast, in the presence of LDAO, the rate of ATP hydrolysis by the wild-type subcomplex was nearly linear (trace c). LDAO stimulated the ATPase activity of the wild-type subcomplex ϳ3-fold. Trace b shows that hydrolysis of 50 M ATP in the presence of 1 mM NaN 3 proceeded with an initial burst that rapidly decelerated to a severely inhibited rate. It was previously shown that the ␣ 3 (␤Y341W) 3 ␥ subcomplex is much less sensitive than the wild type to turnover-dependent inhibition induced by azide during hydrolysis of 2 mM ATP (15). Comparison of traces b in panels A-I and A-II shows that hydrolysis of 50 M ATP by the ␣ 3 (␤Y341W) 3 ␥ subcomplex was less sensitive than that by the wild type to turnover-dependent inhibition by 1 mM NaN 3 . Comparison of traces c in these panels shows that LDAO had a much smaller stimulatory effect on the ␣ 3 (␤Y341W) 3 ␥ subcomplex than on the wild type. . Comparison of traces b in panels A-I, B-I, and B-II shows that the ␤E201C and carboxymethylated ␤E201C mutant subcomplexes were much more sensitive than the wild type to inhibition by azide and that carboxymethylation of the ␤E201C subcomplex slightly relieved inhibition by azide. Comparison of panels A-II and B-III illustrates that the ␣ 3 (␤E201C/Y341W) 3 ␥ double mutant was a much more sluggish enzyme in the presence and absence of LDAO and was much more sensitive than the ␣ 3 (␤Y341W) 3 ␥ subcomplex to turnover-dependent inhibition by azide.
Comparison of the Steady-state Kinetic Parameters of the Wild-type and Mutant ␣ 3 ␤ 3 ␥ Subcomplexes-Entrapment of inhibitory MgADP in a catalytic site when F 1 -ATPases hydrolyze ATP introduces an increase in slope in Lineweaver-Burk plots at high ATP concentration. This increased slope reflects ATP binding to noncatalytic sites rather than catalytic sites. Binding of ATP to noncatalytic sites promotes dissociation of inhibitory MgADP from the affected catalytic site, thus effectively increasing the concentration of active enzyme (3). The steady-state kinetic parameters obtained for the hydrolysis of 2-2000 M ATP by the wild-type and mutant enzyme subcomplexes are summarized in Table I. The K noncat values in Table  I were determined from extrapolation of the linear segment of the Lineweaver-Burk plots with highest slope. In cases where plots with only two distinct slopes were obtained, K m3 was estimated from the intercept on the negative abscissa of the segment of lower slope. When plots with three distinct slopes were observed, the segment of least slope was extrapolated to estimate K m2 , and the segment with intermediate slope was extrapolated to obtain K m3 . To obtain reliable K noncat and k cat values for the ␣ 3 (␤Y341W) 3 ␥ subcomplex, ATP hydrolysis was examined over the range of 2-5000 M ATP rather than that of 2-2000 M ATP examined with the other subcomplexes.
The Lineweaver-Burk plot for the wild-type subcomplex revealed a K m2 value for bisite ATP hydrolysis of 1.7 M and a K m3 value of 43 M for trisite ATP hydrolysis. The K noncat value for the wild-type enzyme is 180 M. The K noncat values varied from 170 to 420 M for the mutant subcomplexes. The very low rates of ATP hydrolysis observed when the ␣ 3 (␤E201C) 3 ␥ and ␣ 3 (␤E201V) 3 ␥ subcomplexes hydrolyzed low concentrations of ATP precluded estimation of K m2 values. However, following carboxymethylation with iodoacetate, the ATPase activity of the ␣ 3 (␤E201C) 3 ␥ subcomplex increased significantly, thus allowing estimation of a K m2 value of ϳ1 M.
The K m3 values obtained for the ␣ 3 (␤E201C) 3 ␥, ␣ 3 (␤E201V) 3 ␥, and ␣ 3 (␤E201C/Y341W) 3 ␥ mutants were 25-, 10-, and 13-fold lower than that of the wild type, respectively. The k cat values tabulated were determined from extrapolation of the linear segment of the Lineweaver-Burk plots with highest slope. They represent maximal rates of ATP hydrolysis when catalytic and noncatalytic sites are saturated with ATP.
Comparison of MgATP and MgADP Binding to the ␣ 3 (␤Y341W) 3 ␥ and ␣ 3 (␤E201C/Y341W) 3 ␥ Subcomplexes in the Presence and Absence of LDAO-By monitoring quenching of fluorescence of the introduced tryptophan in the ␤Y331W mutant of E. coli F 1 , Weber et al. (17) clearly established that MgATP binds to the three catalytic sites with widely different affinities. Moreover, they established that maximal velocity is attained when three catalytic sites are saturated with ATP. Similar results have been obtained with the ␣ 3 (␤Y341W) 3 ␥ subcomplex of TF 1 . Fig. 2 illustrates titration curves obtained from fluorescence measurements of 50 nM solutions of the ␣ 3 (␤Y341W) 3 ␥ and ␣ 3 (␤E201C/Y341W) 3 ␥ subcomplexes with 50 nM to 1 mM ATP and in the presence of 1 mM excess Mg 2ϩ with and without 0.06% LDAO. The curves illustrated were generated with the assumption that saturation of each catalytic site with MgATP quenched one-third of the total fluorescence of the introduced tryptophan. The titration curve for the ␣ 3 (␤Y341W) 3 ␥ subcomplex in the absence of LDAO illustrated in Fig. 2 differs from that previously published (15) in that the concentration of free MgATP rather than total MgATP is plotted on the abscissa. Fig. 2 shows that in the presence of excess Mg 2ϩ , but in the absence of LDAO, one catalytic site saturated at the lowest concentration of ATP added to the ␣ 3 (␤Y341W) 3 ␥ subcomplex. This corresponds to the high affinity catalytic site characterized in unisite experiments (19,20). Complex behavior was exhibited during titration of the second catalytic site with MgATP in the absence of LDAO. Under these conditions, a theoretical binding curve with a single K d that fit the experimental data in this region of the titration curve could not be generated by computer using the equation described by Weber et al. (17). However, when catalytic sites of the ␣ 3 (␤Y341W) 3 ␥ subcomplex were titrated with MgATP in the presence of LDAO, a theoretical binding curve was generated from which K d2 and K d3 values of 0.9 and 47 M, respectively, were estimated. Fig. 2 illustrates that two catalytic sites of the ␣ 3 (␤E201C/Y341W) 3 ␥ double mutant were nearly saturated at the lowest concentration of ATP added to it in the presence of Mg 2ϩ , but in the absence of LDAO. However, when titrated with MgATP in the presence of 0.06% LDAO, K d2 and K d3 values of 0.02 of 6.9 M, respectively, were estimated from a computer-generated theoretical binding curve. In all cases, the K d1 value was too low for reliable estimation. Fig. 3 compares titration of the ␣ 3 (␤Y341W) 3 ␥ and ␣ 3 (␤E201C/Y341W) 3 ␥ subcomplexes with MgADP in the presence and absence of LDAO. These titration curves are very similar to those shown in Fig. 2 for titrations with MgATP. Again, K d2 values could not be estimated by curve fitting when titrations were carried out in the absence of LDAO. The K d2 and K d3 values estimated from curve fitting for titrations in the presence of 0.06% LDAO are given in Table II.
LDAO Overcomes Inhibition of the ␣ 3 (␤Y341W) 3 ␥ Subcomplex Induced by Free Mg 2ϩ in the Assay Medium-ATP hydrolysis by F 1 -ATPases is inhibited by free Mg 2ϩ , which presumably provokes entrapment of inhibitory MgADP in a catalytic site (21,22). Fig. 4 compares the effects of LDAO on hydrolysis of 2 mM ATP by the wild-type and ␣ 3 (␤Y341W) 3 ␥ subcomplexes in the presence of increasing concentrations of Mg 2ϩ . In the absence of LDAO, the optimal ATP/Mg 2ϩ ratio for the wild type was ϳ2. At higher Mg 2ϩ concentrations, the ATPase activity of the wild-type subcomplex decreased. At 11 mM Mg 2ϩ , the ATPase activity was ϳ30% of that observed at the optimal Mg 2ϩ concentration of 1 mM. In the presence of 0.06% LDAO, the optimal ATP/Mg 2ϩ ratio for the wild-type subcomplex was close to 1:1 rather than 2:1. With LDAO present, the ATPase activity of the wild type in the presence of 11 mM Mg 2ϩ was 68% of that observed at the optimal Mg 2ϩ concentration of 2 mM.
The ␣ 3 (␤Y341W) 3 ␥ subcomplex responded differently than the wild-type subcomplex to increasing Mg 2ϩ concentrations in the assay medium in the presence or absence of LDAO. In the absence of LDAO, the optimal ATP/Mg 2ϩ ratio was close to 1:1 for the mutant rather than 2:1 observed with the wild type. At 11 mM Mg 2ϩ in the absence of LDAO, the ATPase activity of the a Cases in which a change in slope on Lineweaver-Burk plots was not exhibited in the presence of LDAO with the wild-type and ␣ 3 (␤E201C) 3 ␥ subcomplexes in the range of 2-50 M ATP.
b Cases in which, in the presence of LDAO, Lineweaver-Burk plots showed slight downward curvature at high ATP concentrations, but reliable K noncat values could not be obtained from extrapolations.
c Cases in which the K d2 value could not be detected in the concentration range of ATP examined. ␣ 3 (␤Y341W) 3 ␥ mutant was 72% of that observed at the optimal Mg 2ϩ concentration of 2 mM. In the presence of LDAO, the ATPase activity of the ␣ 3 (␤Y341W) 3 ␥ subcomplex was slightly stimulated when the Mg 2ϩ concentration was greater than the ATP concentration, and no inhibition was observed at 11 mM Mg 2ϩ . DISCUSSION It was previously reported that substitution of ␤Glu 201 in TF 1 (12) or its equivalent in E. coli F 1 (11) lowers ATPase activity compared with the wild type. However, this is the first demonstration that substitution of this residue with amino acids containing side chains that are incapable of hydrogen bonding increases the affinity of catalytic sites for nucleotides in the presence of Mg 2ϩ . When compared with the wild type, substitution of ␤Glu 201 in the ␣ 3 ␤ 3 ␥ subcomplex of TF 1 with valine or cysteine decreases k cat and K m3 , the Michaelis con-stant for trisite ATP hydrolysis. Consistent with the kinetic analyses, direct binding measurements obtained from nucleotide-induced quenching of fluorescence of introduced tryptophans in the ␣ 3 (␤Y341W) 3 ␥ and ␣ 3 (␤E201C/Y341W) 3 ␥ subcomplexes demonstrated that the ␤E201C substitution increases the affinities of the second and third catalytic sites for MgATP and MgADP. The K m3 values determined for the ␣ 3 (␤E201C) 3 ␥, ␣ 3 (␤E201V) 3 ␥, and ␣ 3 (␤E201C/Y341W) 3 ␥ mutant subcomplexes are 25-, 10-, and 13-fold lower than that of the wild type, respectively, whereas the K d3 value determined for binding of MgATP to the ␣ 3 (␤E201C/Y341W) 3 ␥ double mutant is only 8.4-fold lower than that for binding of MgATP to the ␣ 3 (␤Y341W) 3 ␥ subcomplex. This discrepancy may reflect that the K d values determined from fluorescence quenching measurements are true binding constants, whereas K m3 values are complex rate constants rather than true binding constants.
The overall findings of this study support the hypothesis under scrutiny that proposes that the hydrogen bond between the hydroxyl of ␤Thr 163 and the carboxylate of ␤Glu 199 present in ␤ E , but not in ␤ T or ␤ D , in the crystal structure of MF 1 (4) stabilizes the open conformation of the catalytic site during catalysis. The observation that carboxymethylation of ␣ 3 (␤E201C) 3 ␥ increases k cat and K m3 provides additional support for this hypothesis. Also consistent with this hypothesis are earlier reports demonstrating that the ␤T165S substitution in TF 1 increases K m3 and k cat (23,24). Replacement of threonine by serine in this position might decrease the affinity of MgATP bound to closed catalytic sites by interacting less strongly with Mg 2ϩ and/or by interacting more strongly with ␤Glu 201 in the open conformation of the catalytic site.
In experiments with R. rubrum ATP synthase, Nathanson and Gromet-Elhanan (13) found that ␤-less chromatophores reconstituted with recombinant ␤ subunits in which the equivalent of ␤Glu 201 in TF 1 was substituted with glutamine, gly-   3 ␥ subcomplexes in the presence or absence of LDAO. Samples (1 g each of the wild-type and ␣ 3 (␤Y341W) 3 ␥ complexes) were assayed with 2 mM ATP in the presence of the MgCl 2 concentrations indicated using the coupled assay system described under "Experimental Procedures." q, the wild type without LDAO; E, the wild type with 0.06% LDAO; f, the ␣ 3 (␤Y341W) 3 ␥ mutant without LDAO; Ⅺ, the ␣ 3 (␤Y341W) 3 ␥ mutant with LDAO. cine, or lysine had little or no MgATPase activity in the absence of sulfite. In contrast, reconstituted chromatophores containing the mutant ␤ subunits catalyzed substantial photophosphorylation. The low MgATPase observed probably reflects the high tendency of R. rubrum F 1 to entrap inhibitory MgADP (25,26). Nevertheless, ATP synthesis catalyzed by R. rubrum chromatophores containing the mutant ␤ subunits proceeds at 30 -60% of the rate of enzyme containing wild-type ␤. In contrast, ATP hydrolysis by the mutant ␣ 3 ␤ 3 ␥ subcomplexes of TF 1 containing the mutant ␤ subunits examined here proceeds at 5% or less of the rate exhibited by the wild type. This apparent anomaly can be reconciled by considering the models illustrated in Fig. 5. The models were developed for ATP hydrolysis and ATP synthesis. They account for the following: 1) demonstration that maximal rates of ATP hydrolysis and synthesis are attained when three catalytic sites are saturated (17,27); 2) demonstration that only two ␤ subunits can exist in the closed conformation simultaneously (28,29); 3) rotation of the ␥ subunit in opposite directions during hydrolysis and synthesis (30); and 4) microscopic reversibility.
The single round of ATP hydrolysis illustrated in Fig. 5A initiates as MgATP binds to ␤ E when the closed catalytic sites of ␤ D and ␤ T are liganded with MgATP. This promotes the simultaneous closing of ␤ E and opening of ␤ D . Formation of the transition state during ATP hydrolysis is postulated to occur as the catalytic site of ␤ D converts from the closed to the open conformation. In this spontaneous process, switching ␤ E from open to closed and ␤ D from closed to open forces the coiled-coil of the ␥ subunit to rotate 120°in the counterclockwise direction within the central cavity of the (␣␤) 3 hexamer. In the model for ATP synthesis illustrated in Fig. 5B, energy-dependent synthesis initiates as MgADP and P i bind together to the catalytic site of ␤ E when the closed catalytic sites of ␤ T and ␤ D are liganded with MgATP. Clockwise rotation of the ␥ subunit, propelled by ion translocation through F 0 , drives the simultaneous closing of ␤ E and opening of ␤ T . In this process, the coiled-coil of the ␥ subunit rotates through 120°as loosely bound MgADP and P i are converted to tightly bound MgATP at the catalytic site of ␤ E . The overall process illustrated in Fig. 5B is the microscopic reverse of the process illustrated for hydrolysis in Fig. 5A. The transition state for ATP synthesis forms as the catalytic site of ␤ E converts from the open to the closed conformation. ATP synthesized two rounds earlier dissociates as ␤ T opens simultaneously in the energy-driven process.
The finding that ␤-less R. rubrum chromatophores reconstituted with mutant ␤ subunits containing substitutions at the equivalent of ␤Glu 201 in TF 1 have substantial ATP synthase activity (13) is consistent with the models. During ATP synthesis, the energy-requiring step is closing the catalytic site of ␤ E when loosely bound MgADP and P i are converted to tightly bound MgATP (31)(32)(33). The results presented here suggest that substitution of ␤Glu 201 stabilizes the closed conformation of the catalytic site, thus favoring the energy-dependent closing of ␤ E depicted in Fig. 5B. However, by stabilizing the closed conformation of ␤ T , substitution of ␤Glu 201 with residues containing side chains incapable of hydrogen bonding would hinder dissociation of MgATP, thus accounting for the comparatively small decrease in rates of ATP synthesis observed with R. rubrum chromatophores containing the mutant ␤ subunits (13). During ATP hydrolysis by the ␣ 3 ␤ 3 ␥ subcomplex of TF 1 , substitution of ␤Glu 201 would also favor closing when ␤ E binds MgATP. However, in this case, by stabilizing the closed conformation of ␤ D , the substitutions will significantly attenuate formation of the transition state for hydrolysis, thus significantly reducing the overall rate.
All mutants containing ␤Glu 201 substitutions examined in this study have an increased propensity over the wild type to entrap inhibitory MgADP in a catalytic site during ATP hydrolysis. Kinetic analyses showed that the ␣ 3 (␤Y341W) 3 ␥ subcomplex has less tendency than the wild type to entrap inhibitory MgADP in a catalytic site during turnover. However, complex behavior was observed at low nucleotide concentrations when the tryptophan fluorescence of the ␣ 3 (␤Y341W) 3 ␥ and ␣ 3 (␤E201C/Y341W) 3 ␥ subcomplexes was titrated with MgATP or MgADP. In the absence of LDAO, it was not possible to estimate K d2 values from the titration data. However, when the mutant subcomplexes were titrated with MgATP or MgADP in the presence of LDAO, clearly distinguishable K d2 values could be estimated. This is consistent with observations indicating that LDAO promotes dissociation of inhibitory . Energy released during hydrolysis that promotes rotation of the ␥ subunit or rotational energy that drives closing of ␤ E containing MgADP and P i is shown by (ϳ).
MgADP during ATP hydrolysis. These are as follows. 1) In the presence of LDAO, the wild-type ␣ 3 ␤ 3 ␥ subcomplex hydrolyzes 50 M ATP linearly rather than in the three kinetic phases exhibited in the absence of LDAO illustrated in Fig. 1. 2) The wild-type ␣ 3 ␤ 3 ␥ subcomplex preloaded with MgADP in a single catalytic site hydrolyzes 2 mM ATP with a long lag when assayed in the absence of LDAO, whereas the lag disappears when the preloaded enzyme is assayed in the presence of LDAO (18). 3) [ 3 H]ADP preloaded onto a single catalytic site of the wild-type ␣ 3 ␤ 3 ␥ subcomplex dissociates very slowly when the enzyme is diluted into assay medium containing 40 M ATP, whereas when the assay medium contains LDAO, the preloaded [ 3 H]ADP dissociates rapidly (18).
The complex binding observed during titrations with MgATP and MgADP in the absence of LDAO is consistent with an equilibrium between F 1 MgADP and F 1 *MgADP that has been proposed to explain transient inhibition of F 1 -ATPases by MgADP during ATP hydrolysis (34). F 1 MgADP represents MgADP bound to F 1 in an active conformation, whereas F 1 *MgADP represents MgADP bound to F 1 in an inactive conformation.
Titrations of catalytic sites of the ␣ 3 (␤Y341W) 3 ␥ subcomplex with MgATP or MgADP illustrated in Figs. 2 and 3 differ from titrations of the subcomplex with MgATP or MgADP previously reported from this laboratory (15). In the titrations shown here, the number of sites occupied is plotted against free MgATP or MgADP concentration rather than total MgATP or MgADP concentration as previously reported in Ref. 15. Conversion of the data on the abscissas of the titration curves in Ref. 15 from total MgATP or MgADP to free MgATP or MgADP results in titration curves that closely resemble those illustrated in Figs. 2 and 3 of this study. In the previous study (15), titration curves were presented that suggested that binding of MgATP to the ␣ 3 (␤Y341W) 3 ␥ subcomplex was only slightly affected by the presence of 0.06% LDAO, whereas Fig. 2 clearly shows that LDAO shifts the titration curve for the subcomplex to lower affinity when it is titrated with MgATP. The curve shift illustrated in Fig. 2 is supported by the demonstration that LDAO overcomes entrapment of inhibitory MgADP during hydrolysis of 50 M ATP by the ␣ 3 (␤Y341W) 3 ␥ subcomplex (Fig. 1) and that LDAO overcomes inhibition induced by free Mg 2ϩ in the assay medium (Fig. 4).