Unisite Catalysis without Rotation of the γ-ε Domain in Escherichia coli F1-ATPase*

Unisite [γ-32P]ATP hydrolysis was studied in ECF1 from the mutant βE381C after generating a single disulfide bond between β and γ subunits to prevent the rotation of the γ/ε domain. The single β-γ cross-link was obtained by removal of the δ subunit from F1 and then treating with CuCl2 as described previously (Aggeler, R., Haughton, M. A., and Capaldi, R. A. (1996) J. Biol. Chem. 270, 9185–9191). The mutant enzyme, βE381C, had an increased overall rate of unisite hydrolysis of [γ-32P]ATP compared with the wild type ECF1 due to increases in the rate of ATP binding (k +1), Pi release (k +3), and ADP release (k +4). Release of bound substrate ([γ-32P]ATP) was also increased in the βE381C mutant. Cross-linking between Cys-381 and the intrinsic Cys-87 of γ caused a further increase in the rate of unisite catalysis, mainly by additional effects on nucleotide binding in the high affinity catalytic site (k +1 and k +4). In δ-subunit-free ECF1 from wild type or βE381C F1, addition of an excess of ATP accelerated unisite catalysis. After cross-linking, unisite catalysis of βE381C was not enhanced by the cold chase. The covalent linkage of γ to β increased the rate of unisite catalysis to that obtained by cold chase of ATP of the noncross-linked enzyme. It is concluded that the conversion of Glu-381 of β to Cys induces an activated conformation of the high affinity catalytic site with low affinity for substrate and products. This state is stabilized by cross-linking the Cys at β381 to Cys-87 of γ. We infer from the data that rotation of the γ/ε rotor in ECF1 is not linked to unisite hydrolysis of ATP at the high affinity catalytic site but to ATP binding to a second or third catalytic site on the enzyme.

F 1 F 0 -type ATPases are membranous protein complexes responsible for oxidative and photosynthetic ATP synthesis in eubacteria, mitochondria, and chloroplasts. The simplest form of this enzyme, as found in Escherichia coli (ECF 1 F 0 ), 1 contains eight different subunits. Five of these subunits (␣ 3 , ␤ 3 , ␥ 1 , ␦ 1 , ⑀ 1 ) are located outside the membrane in the catalytic F 1 portion. The other three subunits (a 1 , b 2 , c 9 -12 ) form a channel that transports protons through the membrane (for recent reviews, see Refs. 1 and 2). Each enzyme complex contains three catalytic sites located predominantly on ␤ subunits (3), which work cooperatively during ATP hydrolysis or synthesis (reviewed in Refs. 1 and 2). How catalytic site events are coupled to proton translocation is only now beginning to be understood.
It had been shown several years ago that the ␥ and ⑀ subunits undergo conformational changes in F 1 F 0 upon membrane energization (4 -6) and that they shift their positions between the ␣ and ␤ subunits of soluble F 1 in response to nucleotide binding (7)(8)(9). Recently, it has been established conclusively that these movements represent rotation of a domain formed by the ␥ and ⑀ subunits (10,11). Such rotation has been visualized directly during ATP hydrolysis in the soluble F 1 (12,13) and is inferred by cross-linking studies of the hydrolytic and synthetic reaction when catalyzed by the whole F 1 F 0 complex (11,14).
As yet, no correlation has been established between the rate of rotation of the ␥-⑀ domain and the kinetics of individual steps in the ATP hydrolysis or synthesis reactions. However, the hydrolytic activity of the enzyme is almost fully inhibited by cross-linking either ␥ or ⑀, or both subunits, to ␣ or ␤ subunits through engineered disulfide bonds (8 -10), thereby confirming the direct linkage between catalytic site events and rotation. The driving force for this rotation in the direction of ATP hydrolysis could be the binding energy of ATP and/or the free energy change associated with the ATP splitting reaction. However, complicating any analysis, ATP can bind in three catalytic sites that are characterized by high, medium, and low affinities for nucleotides (1)(2)(15)(16), a priori, binding in any of these sites could drive the rotation.
To some extent, the functioning in these sites can be differentiated because ATP hydrolysis in the high affinity catalytic site can be monitored by so-called "unisite" catalysis measurements (17). It is distinguished from nucleotide binding into low affinity sites which accelerates product release from the first site, as in cold chase experiments (18). Unisite catalysis is measured with substoichiometric amounts of [␥-32 P]ATP in relation to F 1 and is characterized by a highly exergonic ATP binding step (K 1 ), a reversible ATP hydrolysis/synthesis equilibrium that occurs with a negligible change in free energy (K 2 ), and a very slow product release step which is rate-limiting (k ϩ3 and k ϩ4 ) (see Equation 1 and review in Ref. 19). Therefore, if subunit rotation is coupled to unisite catalysis, ATP binding in the high affinity site would be the most probable driving force (20). The kinetic steps of the unisite catalytic cycle are as follows.
Here, we measure unisite catalysis in ECF 1 in which the ␥ subunit has been cross-linked to a ␤ subunit to prevent rota-* This work was supported by National Institutes of Health Grant HL24526. 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.

EXPERIMENTAL PROCEDURES
Materials-[␥-32 P]ATP (PB 218) was purchased from Amersham Pharmacia Biotech, and [␣-32 P]ATP was a gift from the laboratory of Dr. Peter von Hippel of this Institute of Molecular Biology. ATP was from Sigma. The mutant ␤E381C is described elsewhere (21). Wild type and mutant F 1 -ATPases were purified as described before (22). ECF 1 was depleted of ␦ subunit by gel filtration chromatography through Sephacryl S-300 in the presence of 0.5% LDAO. Enzyme (8 -10 mg) was dissolved in 1 ml of a buffer containing 50 mM Tris, 20% glycerol, 2 mM EDTA, 1 mM ATP, 1 mM DTT, and 40 mM EACA, pH 7.4 (4°C), along with 0.5% LDAO (buffer A). Samples were loaded onto a Sephacryl column (1.2 m x 1.0 cm), and protein was eluted with the same buffer. Samples lacking the ␦ subunit based on SDS-polyacrylamide electrophoresis were collected and concentrated to 2-3 mg ml in Amicon tubes and stored in liquid nitrogen.
CuCl 2 -induced Cross-linking of F 1 Preparations-Cross-linking was induced by CuCl 2 as described previously (21). Before starting the cross-linking reactions, enzyme dissolved in buffer A under liquid nitrogen was defrosted at room temperature and precipitated with 70% saturated ammonium sulfate at 4°C for 1-2 h. These suspensions were centrifuged at 16000 ϫ g for 6 min at 4°C, resuspended in 60 -100 l of 50 mM MOPS, 10% glycerol, and 0.1 mM EDTA, pH 7.0, and gel filtered through two consecutive Sephadex G-50 columns in the same buffer to remove nucleotides and ammonium sulfate. ECF 1 from the mutant ␤E381C was reacted with 200 M CuCl 2 for 2-4 h to maximize the cross-linking yield. The control enzyme was reacted under the same conditions but with 5 mM DTT present. Multisite hydrolytic activity was measured after cross-linking and before starting the unisite catalysis measurements. For the mutant ␤E381C, it was 15-20 mol/min/mg for the control (reduced with 5 mM DTT) and 0.3-0.4 mol/min/mg after cross-linking, i.e. multisite activity was inhibited 97-98%. The steadystate ATPase activity at 37°C of the wild type ECF 1 was 30 mol/min/ mg. These activities were not altered significantly by the removal of the ␦ subunit.
Time Courses of Unisite Hydrolytic Activities-Fully cross-linked and control ECF 1 from the ␤E381C mutant were diluted approximately 10-fold (to 1 M) for measurement of unisite ATPase activity in 50 mM Tris-SO 4 , 5 mM MgSO 4 , and 2 mM K 2 HPO 4 , pH 7.5. DTT (5 mM) was present in the samples of control enzyme.
Unisite reactions were started by adding 50 l of [␥-32 P]ATP to 50 l of F 1 during rapid mixing in a 13 ϫ 100 mm glass test tube. The [␥-32 P]ATP/F 1 ratio used was 0.1 or 0.2 without significant differences on the results obtained in both conditions. The final concentration of F 1 was 0.5 M and that of [␥-32 P]ATP was 0.1 or 0.05 M. Reactions were stopped during mixing at different reaction times with 0.4 ml of trichloroacetic acid (6% final concentration). In the case of cold chase experiments, 100 l of 10 mM MgATP was added to the F 1 ϩ [␥-32 P]ATP mixture at the times shown, and the reactions were stopped 2 s later with acid. In the control samples, 5 mM DTT was present in all reaction mixtures, except for the trichloroacetic acid solution, to avoid any crosslinking. To quantify the amount of [␥-32 P]ATP remaining, [ 32 P]P i was extracted from these aqueous samples as [ 32 P]P i -phosphomolybdate as described before (23). Reduction of this complex by DTT was avoided by two different methods, which gave exactly the same results. In one, the reaction was quenched directly with 0.5 ml of activated charcoal in 1 N HCl, and the adsorbed [␥-32 P]ATP was washed twice with 1 N HCl containing 20 mM P i . The bound [␥-32 P]ATP was eluted with ethanol/ ammonium hydroxide (24), and radioactivity was counted directly by liquid scintillation or Cerenkov counting. In the second method, control samples were treated with trichloroacetic acid as before, but then DTT was reacted with a 10-fold excess of N-ethylmaleimide for 5 min before the formation of the [ 32 P]P i -phosphomolybdate complex. Reaction with N-ethylmaleimide was preceded by the neutralization of each sample with NaOH or Tris to a pH of 7-7.5. Samples without DTT were processed in the same way.
Rate Constants for Unisite Catalysis-The rate of [␥-32 P]ATP binding was measured by a hexokinase trap (25). This was used instead of the cold chase method to avoid any inhibition of the cold chase response due to cross-linking and because of possible release of the bound [␥-32 P]ATP induced by the cold chase (26). In the hexokinase trap, the unisite reactions were started as described before, but at the times shown, 100 units of hexokinase were added in the presence of 10 mM glucose. The hexokinase reaction was allowed to proceed for 10 s, and the reactions were stopped with 1.3 N HCl. The remaining [␥-32 P]ATP was hydrolyzed by boiling for 12 min, and the [ 32 P]P i produced was extracted as described above. The radioactivity of the aqueous phase was taken as the amount of [ 32 P]glucose-6-phosphate formed, i.e. [␥-32 P]ATP that was not bound to F 1 . Controls showed that 98 -99% of the [␥-32 P]ATP was heat hydrolyzed, that 99 -100% of the [ 32 P]glucose-6-phosphate was heat-resistant, and that the efficiency of the hexokinase trap mixed with F 1 was 97-99%. The decay on the hexokinase accessible [␥-32 P]ATP against time was used to calculate the rate constant of [␥-32 P]ATP binding, supposing a second order association process according to Penefsky (27). The further measurement of the rate of [␥-32 P]ATP release (k Ϫ1 ) showed that the amount of [␥-32 P]ATP released in the first 10 s of reaction was negligible, allowing for a good estimation of k ϩ1 . On the other hand, k Ϫ1 was measured by a similar hexokinase trap according to the method of Penefsky (27), in which the [␥-32 P]ATP/F 1 mixture was aged for 30 s and the free 32 P was removed by gel filtration through Sephadex columns equilibrated with the standard unisite buffer containing 1 mg/ml bovine albumin (BSA). After saving aliquots for determination of [␥-32 P]ATP and [ 32 P]P i coeluted with F 1 , the efluent was diluted 10-fold in the same buffer containing 1 mg/ml BSA, 1 mg/ml hexokinase, and 20 mM glucose. At different times, aliquots were removed and the hexokinase reactions were stopped with 1.3 N HCl. The amount of [ 32 P]glucose-6-phosphate formation, measured as described before, was used to calculate k Ϫ1 according to Penefsky (27), i.e. multiplying by the respective values of (1 ϩ K 2 ) and correcting for rate of [ 32 P]P i release measured as described below.
To calculate the catalytic equilibrium constant (K 2 ), the [␥-32 P]ATP/ F 1 mixture made at ratios of 0.2 or 0.1, was filtered at different times through Sephadex columns equilibrated with buffer and BSA. The eluted protein was colected directly in 6% trichloroacetic acid, and the coeluted [␥-32 P]ATP and [ 32 P]P i were measured as described before (23). The rate of decay of the total bound 32 P was used to calculate the rate of [ 32 P]P i release. k ϩ3 was obtained by multiplying the P i off rate by the respective values of (1 ϩ 1/K 2 ), according to Penefsky (27) and Al-Shawi and Nakamoto (40) Using the values obtained for k ϩ1 , k Ϫ1 , K 2, and k ϩ3 , the respective forward (k ϩ2 ) and reverse (k -2 ) rate constants were obtained through a computer-assisted numerical integration of the unisite catalytic cycle (Equation 1). The software used was KINSIM (28) as described previously (29). Fitting of the k ϩ2 and k -2 rate constants to the experimental data was made with the FITSIM program (30). The values of k -3 and k -4 were not obtained. Therefore, the calculation of k ϩ2 and k -2 was made using only the values of k ϩ1 , k Ϫ1 , K 2 , and k ϩ3 . The validity of this procedure as has been used before (27) is based on the irreversibility of the 32 P i release step (with 2 mM P i present) and on the negligible amounts of ADP that are released in the first seconds of the reactions (values for k ϩ4 are in the range of 10 Ϫ3 , see Table I).
The rate of [␣-32 P]ADP release was measured by the dilution/gel filtration method of Cunningham and Cross (31) as modified by Al-Shawi and Senior (32). Unisite catalysis was started by mixing 2.5 l of 2 M F 1 with 2.5 l of 0.4 M [␣-32 P]ATP at 0 time. After 30 s, the samples were diluted 100-fold, and 120-l aliquots were filtered through Sephadex columns at different times. The 0 time points were filtered 30 s after the dilution step. The Sephadex columns were equilibrated with buffers containing 1 mg/ml of BSA to minimize the loss of F 1 during gel filtration. The decay in the radioactivity eluted with F 1 from the columns was used to calculate the rate of ADP release according to a first order exponential decay of the bound 32 P. This rate was multiplied by (1 ϩ 1/K 2 ) to correct for the equilibrium distribution of ATP and ADP as it was made for the calculation of k ϩ3 .
To test the accuracy of mixing, gel filtration, and curve fitting methods employed, some rate constants (k ϩ1 , k ϩ3 , K 2 , k ϩ2 , and k -2 ) were measured in parallel using the wild type enzyme (Ϯ ␦ subunit) treated in the same conditions used for the ␤E381C ECF 1 . The values of the rate constants obtained were in the range reported for the wild type ECF 1 (see Table I) Other Methods-Multisite ATPase activities were measured spectrophotometrically with the ATP regenerating system of Lötscher et al. (33) at 37°C. Gel electrophoresis was performed according to Laemmli (34) using SDS-polyacrylamide gradient gels of 10 -18%. Protein concentrations were measured using the BCA method.

RESULTS
ECF 1 from the mutant ␤E381C was used in the present study. The properties of this mutant have been described in detail elsewhere (21). Cross-linking in essentially full yield can be obtained between Cys-381 of ␤ and the intrinsic Cys-87 of the ␥ subunit. The effect of this cross-linking is to inhibit multisite ATPase activity as reported before (21), and confirmed here. In the present study, the effects of the crosslinking on unisite catalysis have been examined. The kinetics of ATP hydrolysis in a single high affinity catalytic site was measured using [␥-32 P]ATP, and the effect of a large excess of ATP as a "cold chase" on this kinetics was studied. This socalled cold chase examines the effect of ATP binding at a second and/or third catalytic site on the kinetics of [␥-32 P]ATP hydrolysis in the first, highest affinity site.
In ECF 1 , the promotion of unisite ATP hydrolysis by excess ATP is maximized in enzyme from which the ␦ subunit has been removed (35). This may have to do with non-physiological binding of the ␦ subunit in isolated ECF 1 (21). Recent studies have shown that the ␦ subunit is a two domain protein (41,42). In ECF 1 F 0 , the N-and C-terminal domains interact with each other (43,44), whereas in isolated ECF 1, the C-terminal domain can swing away to interact with the DELSEED region of a ␤ subunit (21). This is evident from the observed cross-linking from ␤E381C to the intrinsic Cys-140 of ␥ shown previously and confirmed here (Fig. 1).
A selective removal of the ␦ subunit was accomplished by gel filtration of ECF 1 in the presence of LDAO as described fully under "Experimental Procedures." Fig. 1 shows the separation of the subunits of intact ECF 1 and ␦-free enzyme from ␤E381C by SDS-polyacrylamide gel electrophoresis. Removal of ␦ was obtained without a significant loss of the ⑀ subunit, c.f. lanes 3 and 1. Fig. 1 also shows cross-linking of ECF 1 from the mutant ␤E381C induced by incubation with 200 M CuCl 2 (lanes 2 and  4). Removal of the ␦ subunit not only prevents non-physiological reactions of this subunit, but facilitates interpretation of experiments because the only cross-linked product observed in significant amounts is now between ␤-␥. Therefore, the fixing of these two subunits can be correlated directly with the observed effects.
In the presence or absence of the ␦ subunit, multisite or cooperative ATPase activity was inhibited by 97-98% in different experiments, in comparison with control samples treated with the equivalent levels of CuCl 2 , but with 5 mM DTT present. This inhibition was at roughly the same level as the yield of cross-linking.
The removal of the ␦ subunit had very little effect in the rate of acid quench-measured unisite hydrolysis of [␥-32 P]ATP in either wild type or mutant-(␤E381C) ECF 1 , as reported previously (35) and see Table I. Fig. 2 compares the rates of unisite catalysis of wild type ECF 1 and ␤E381C, both before and after cross-linking. The rates of unisite catalysis by wild type ECF 1 are comparable with those obtained by others under similar experimental conditions (35,40); see Table I for rate constants of catalysis. As evident in Fig. 1, the substitution of Cys for Glu-381 in the mutant ␤E381C significantly increased the rate of unisite ATP hydrolysis of ECF 1 , and this rate is increased more dramatically after cross-linking of the Cys at position 381 of ␤ with Cys-87 of the ␥ subunit.
Before making a detailed kinetic analysis of the unisite activity of the ␤E381C mutant, the possible contribution of multisite activity in the present experiments was examined by testing the effect of sodium azide (NaN 3 ). NaN 3 is an inhibitor which effectively blocks multisite ATPase activity of F 1 -AT-Pases without significant effect on unisite catalysis, presumably by affecting catalytic site cooperativity (36,37). Previous studies had suggested that cooperative or multisite catalysis could contribute even when ratios of [␥-32 P]ATP to F 1 of 0.3 mol/mol were used (38). Although cross-linking itself inhibits the multisite catalysis, this control is still relevant because small amounts of noncross-linked enzyme are present and could be providing all of the [␥-32 P]ATP hydrolysis being observed. However, as shown in Fig. 2, NaN 3 at concentrations that completely blocked multisite catalysis in noncross-linked enzyme (results not shown; Refs. 36 and 37) had no effect on the rates of unisite catalysis after cross-linking. Taken together, the above results confirm that the ATP hydrolysis being followed occurs in the so-called "high affinity" catalytic site in the cross-linked enzyme. Table I lists the rate and equilibrium constants that were measured. In relation to wild type ECF 1 , the rate of ATP binding (k ϩ1 ) is faster for the ␤E381C mutant both before and after cross-linking. Also, [␥-32 P]ATP release was faster in the mutant by a factor of 25 before cross-linking and more than 100-fold after cross-linking. The result is a 10-fold decrease in the affinity for ATP. As indicated in Table I, the mutant showed a more than 20-fold increase in k ϩ3 , the rate of P i release, which was not enhanced greatly by cross-linking. The cross-linking also increased the rate of ADP release from the mutant ECF 1 . Table I also lists K 2 measured for the different preparations. This equilibrium constant was not greatly affected in the mutant with or without cross-linking.
The effect of a cold chase of excess ATP on unisite catalysis for the wild type and mutant is shown in Fig. 3. There was the expected burst of [␥-32 P]ATP hydrolysis after addition of cold ATP in the wild type ECF 1 and in enzyme from the mutant without cross-linking. However, after cross-linking, the rate of unisite catalysis carried out by the ␤E381C mutant was by itself as fast as the accelerated unisite catalysis of the noncross-linked mutant. Unisite catalysis of the ␤E381C mutant was also much faster than the promoted and non-promoted unisite catalysis carried out by the wild type enzyme, which is limited by the rate of [␥-32 P]ATP binding (Fig. 3). DISCUSSION In the mutant ␤E381C, a cross-link can be obtained between ␤ and ␥ subunits in essentially full yield. Previous studies made in ECF 1 from the triple mutant ␤Y331W:␤E381C:⑀S108C (39) have established that binding of ATP to this cross-linked F 1 is similar to that of noncross-linked enzyme, i.e. there is one high affinity binding site (K d ϭ 90 -200 nM), a loose site (K d ϭ 2-7 M), and a weak binding site (K d ϭ 40 -50 M). The results here show that enzyme cross-linked between ␤ (at Cys-381) and  1 and 2) or without ␦ subunit (lanes 3 and 4), was crosslinked with 200 M CuCl 2 as described under "Experimental Procedures." The control enzymes (lanes 1 and 3) were incubated and loaded into the gel in the presence of 5 mM DTT. The multisite hydrolytic activity of these enzymes was diminished from 15 mol/min/mg to 0.3-0.4 mol/min/mg after cross-linking. 15 g of each protein were loaded per lane.
␥ (at Cys-87) retains a high affinity catalytic site with a K d for ATP in the nanomolar range (Table I). At the same time, cooperative multisite ATPase activity is essentially lost after cross-linking. Taken together, these results indicate that negative cooperativity of nucleotide binding and positive cooperativity of multisite ATPase activity need not be coupled to each other.
In novel labeling studies using laser-induced covalent incorporation of 2-azido-ATP, we have established a direct relationship between nucleotide binding affinity and the different interactions between ␥ and ⑀ subunits with the three ␤ subunits (15). The site with highest affinity is in that ␤ subunit which has neither of the smaller subunits crosslinked to it (␤ free ), the loose binding site is in that ␤ which cross-links with ⑀, and the weak site is in the ␤ which crosslinks to ␥ (15). This relationship can only be retained during enzyme turnover if nucleotide binding in at least one ␤ subunit, with the associated conformational changes that must occur, results in the movement of the ␥-⑀ subunit domain from one ␣/␤ pair to the next. A priori, the movements of ␥ and ⑀ could be due to binding in any of the three catalytic sites. The studies described here were directed toward asking whether nucleotide binding in the highest affinity catalytic site, bond cleavage in this site, or substrate binding in a second or third catalytic site, drives the rotation of ␥/⑀.
It was found that the mutation of Glu-381 of the DELSEED region of ␤ to a Cys (without any cross-linking) led to altered unisite catalysis by ECF 1 . This mutation increased the overall rate of unisite catalysis significantly by affecting both substrate and product binding. The most dramatic effect was on ATP release rate which was 30-fold faster than that for wild type enzyme. ATP binding (k ϩ1 ) was also increased, and there was a more than 20-fold increase in the rate of P i release. Fig.  4 compares the unisite rate constants for ␤E381C and those for two mutations in the ␥ subunit. The ␥T106C mutation also increased k ϩ1 while both this mutation and the change of Met-23 of ␥ to a Lys affected k ϩ3 . In fact, for the mutant ␥M23K, k ϩ3 was twice as fast again as for the DELSEED mutation. Structural studies show that Met-23 of ␥ interacts with the DELSEED region of one of the three ␤ subunits. This is a different ␤ subunit from that which forms the disulfide bond with Cys-87 of ␥ studied here. Clearly, the interaction of the DELSEED region of two different ␤ subunits with ␥ has a Standard deviations were omitted for simplicity, but these were no higher than 20%. Significant changes produced by the ␤E381C mutation are highlighted, and further increases induced by cross-linking are highlighted in italics.  The unisite catalysis of the ␤E381C F 1 (-␦) was measured at a [␥-32 P]ATP/F 1 ratio of 0.1. In the presence of 5 mM DTT, nonpromoted unisite catalysis (q-q) was stopped by the standard acid quench. Promoted unisite catalysis (E-E) was measured by adding a volume of 10 mM MgATP to ECF 1 undergoing unisite catalysis and allowing the reaction to proceed 2 s further during rapid mixing before adding the acid. The open circles indicate the time at which the reactions were stopped. The same procedure was used to measure the promoted (‚-‚) and non-promoted (OE-OE) unisite catalysis of the ␤E381C F 1 (-␦) previously cross-linked between ␤ and ␥ subunits. For comparison, the figure also shows the promoted (ϩ⅐⅐⅐⅐ϩ) and non-promoted (ϫ⅐⅐⅐⅐ϫ) unisite catalytic activity of wild type ECF 1 (-␦). For the wild type ECF 1 , promoted experimental data are plotted at the time of addition of the cold chase, but the reactions were stopped 1 min later, the same results were obtained if the cold chase reactions of the wild type ECF 1 were stopped after 2 or 5 s. Bound [␥-32 P]ATP (छ-छ) was also measured in parallel for the wild type enzyme by the hexokinase trap method. similar and direct effect on catalysis in the highest affinity site, as would be expected if ␥ subunit binding is determined and/or is a determinant of nucleotide binding affinities.
Cross-linking between ␤ and ␥ subunits of the mutant ␤E381C which inhibits multisite catalysis by 97%, was found to further increase the rate of unisite catalysis to the point that the reaction was essentially over within the first few seconds under the standard conditions used here. Curve fitting was used to analyze the kinetics of unisite catalysis as described under "Experimental Procedures." Good fits (S.E. Յ 20% in the constants fitted) were obtained for wild type enzyme and for mutant, giving values of 0.71 and 0.33 (wt) and 0.84 and 0.84 (␤E381C) respectively for k ϩ2 and k -2 . These values are similar to those reported earlier for wild type enzymes (35,40) but lower than those of the ␥T106 mutation (6,29), presumably due to the differences introduced by this mutation and/or its labeling with a fluorescent probe. The kinetics of unisite catalysis after cross-linking of ␤E381C was so fast that no satisfactory fit could be obtained (S.E. Ն 50% in k ϩ2 and k -2 ), therefore it remains unclear if cross-linking alters these two rate constants.
As shown in Fig. 3, the effect of covalently linking ␤Cys-381 to ␥Cys-87 is to increase the rate of unisite catalysis to that of non-cross-linked enzyme obtained with a cold chase of ATP. After cross-linking, the ATP cold chase did not increase the observed rate of unisite catalysis. The implication is that ECF 1 from the mutant ␤E381C is in a conformation facilitating the rapid release of nucleotide and P i that occurs on substrate binding in a second or third site. This conformation is then stabilized by cross-linking. A similar activation of the high affinity catalytic site can be obtained in Rhodospirillum rubrum F 1 -ATPase in the presence of Ca 2ϩ , but in this case, multisite ATPase activity is also enhanced (45). Further studies are needed to determine if these activated states are structurally similar.
The rapid rate of bond cleavage of ATP to ADP ϩ P i in enzyme in which the ␥ subunit is covalently linked to a ␤ subunit indicates that rotation of the ␥/⑀ subunit domain is not driven by, and does not require, nucleotide binding, or bond cleavage, in the high affinity catalytic site. Additional testing requires that methods such as those employed by Sabbert et al. (12) or Noji et al. (13) for demonstrating rotation of the ␥ subunit are adapted to unisite conditions. An absence of rotation of the ␥/⑀ subunits driven by ATP binding or bond cleavage in the high affinity catalytic site does not necessarily rule out that these reactions drive the coupling of ATP hydrolysis (or synthesis) to proton translocation. Conformational changes induced by ATP hydrolysis within the high affinity catalytic site could be transmitted to the c subunit ring and result in translocation of protons without rotation of the ␥/⑀/c subunit ring domain (presently, the favored model of energy transduction within the ATP synthase (10,44,46)). Instead, rotation could realign the molecule after each catalytic site-driven proton translocation event for the next turnover to proceed. In this connection, it is interesting that conformational changes have been observed by fluorescent probes attached at positions 8 and 106 of the ␥ subunit which correlate with unisite catalysis (6,29). Critical testing of the rotation of the c subunit ring as a function of catalytic site events remains an important prerequisite to getting a better understanding of the mechanism of the ATP synthase. FIG. 4. Comparison of rate constants of unisite catalysis of ␤E381C, ␥T106C, and ␥M23K mutants. This figure compares the rate constants of unisite catalysis affected by the ␤E381C mutation and by cross-linking, with those previously reported for the ␥T106C and ␥M23K mutations. The data for these other two mutants were taken from Refs. 6 and 40, respectively.