Energy-dependent Dissociation of ATP from High Affinity Catalytic Sites of Beef Heart Mitochondrial Adenosine Triphosphatase*

Incubation of [-p3’P]ATP with a molar excess of the membrane-bound form of mitochondrial ATPase (F,) results in binding of the bulk of the radioactive nucleo- tide in high affinity catalytic sites (K, = lo1’ “l). Subsequent initiation of respiration by addition of suc- cinate or NADH is accompanied by a profound decrease in the affinity for ATP. About one-third of the bound radioactive ATP appears to dissociate, that is, the [y-32P]ATP becomes accessible to hexokinase. The NADH-stimulated dissociation of [y3’P]ATP is en- ergy-dependent since the stimulation is inhibited by uncouplers of oxidative phosphorylation and is pre- vented by respiratory chain inhibitors. The rate of the energy-dependent dissociation of ATP that occurs in the presence of NADH, ADP, and Pi is commensurate with the measured initial rate of ATP synthesis in NADH-supported oxidative phosphorylation catalyzed by the same submitochondrial particles. Thus, the rate of dissociation of ATP from the high affinity catalytic site of submitochondrial particles meets the criterion of kinetic competency under the conditions of oxidative phosphorylation. These experiments provide evidence in support of the argument that energy conserved during the oxidation of substrates by the respiratory chain can be utilized to reduce the very tight binding of product ATP in high affinity catalytic sites and to promote dissociation of the nucleotide. p~ gluc0se-6-~'P units hexokinase to 1 ml of containing 0.1 nmol of converted 98% of radioactive ATP to gluco~e-6-~'P. If 1 mg or more of submitochondrial particles were added before the hexokinase, only about 3-4% of the radioactive ATP was available to the hexokinase. If 200 units of hexokinase were added to 1 mg of submitochondrial particles before adding 0.1 nmol of 95% or more of the added converted gluco~e-6-~'P.

the catalytic step that is near unity (1,4). These observations support a model for ATP synthesis in oxidative phosphorylation which proposes that ATP forms from ADP and Pi bound in high affinity catalytic sites with little or no change in free energy and that the major requirement for energy in oxidative phosphorylation is the release of product ATP from high affinity catalytic sites (1-5). In terms of this model, the coupling mechanism linking the energy store to ATP synthesis in oxidative phosphorylation can be considered a device that lowers the affinity of the high affinity catalytic sites for ATP and facilitates dissociation (5). It was proposed that such a device could be expressed via changes in the state of *This research was supported in part by Research Grant GM 21731 from the National Institutes of Health, United States Public Health Service. The costs of publication of this article were defrayed in*part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ionization of the glutamate carboxyl group of subunit c (the carboxyl group of beef heart Fo (the membrane-embedded portion of the ATPase complex) that reacts with dicyclohexylcarbodiimide (6, 7)) or by changes in the state of ionization of other charged amino acid residues in Fo. Conformational changes in Fo, resulting from changes in the state of ionization of charged groups of the subunit proteins, could, when transmitted to F1, bring about changes in the conformation of F1 that reversibly alter the catalytic sites and thus the desired decrease in affinity (3). If the high affinity catalytic sites of membrane-bound F1 participate in ATP synthesis in oxidative phosphorylation, it may expected that these sites will exhibit energy-dependent changes in affinity for bound ATP. This paper examines the fate of [-p3'P]ATP bound in high affinity catalytic sites of submitochondrial particles. It is shown, during respiration initiated by NADH or NADH plus ADP, that an energy-dependent dissociation of about one-third of the bound radioactive ATP takes place.

EXPERIMENTAL PROCEDURES
Material~-'~P; (enzyme grade) was purchased from ICN and used without further purification. [y-"P]ATP was prepared as described (8) and stored at -20 "C in small volumes. The specific radioactivity of most preparations was about lo7 cpm/nmol. The preparations of radioactive ATP were used until the amount of free "Pi present reached about 5% of the total radioactivity. Polyethyleneimine-impregnated cellulose sheets (CEL PEI-UV) were purchased from Brinkmann Instruments. Myxothiazol was purchased from Boehringer Mannheim and antimycin, crystalline yeast hexokinase, and NADH were purchased from Sigma. Hexokinase was dialyzed before use to remove ammonium sulfate. FCCP' was a gift from Dr. P. G. Heytler, E. I. du Pont de Nemours & Co., Wilmington, DE. S-13 was a gift from Dr. P. C. Hamm, Monsanto, St. Louis, MO.
Methods-Submitochondrial particles prepared in the presence of ATP and magnesium were prepared as described (9) and activated by washing with buffered solutions of KC1 (3). The specific activity of the resulting submitochondrial particles ranged between 6 and 10 units/mg. The steady state rate of respiration catalyzed by the activated submitochondrial particles was 0.25 pA of O/min/mg of protein with succinate as substrate and 0.4-0.5 pA of O/min/mg of protein with NADH. The P/O ratio with succinate as substrate was 0.2-0.3. When NADH was incubated with 1 mg of activated submitochondrial particles in 1 ml of reaction mixture, at least 60 s of respiration were possible before anaerobiosis occurred. The rate of ATP synthesis during the initial 3 s following the addition of 1 mM NADH, 0.1 mM ADP, and 10 mM Pi to activated submitochondrial particles was 1.2 nmol of ATP formed per nmol of membrane-bound Fl/s. When the formation of 32Pi was measured in ATPase experiments, reactions were stopped by adding 50 p1 of a solution containing 60% perchloric acid and 5 mM ATP. The deproteinized reaction mixture was extracted with isobutyl alcohol/benzene (10) and the amount of "Pi formed was determined by counting the organic phase. The net 'The abbreviations used are: FCCP, carbonyl cyanide ptrifluoromethoxyphenylhydrazone; S-13, 5-chloro-3-tert-buty1-2'chloro-4'-nitrosalicylanilide; TNP-ATP and TNP-ADP, the 2',3'-0-(2,4,6-trinitrophenyl) derivatives of ATP and ADP, respectively; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid. amount of 3zPi formed was calculated by subtracting any 3zPi present in controls to which perchloric acid was added before the submitochondrial particles. The amount of [y-32P]ATP remaining was determined by counting the aqueous phase.
When the amount of glu~ose-6-~'P was determined in reaction by adding 1.0 ml of 2 N HCl that also contained 0.1 pmol of glucose-mixtures containing glucose and hexokinase, reactions were stopped 6-P. Samples of the deproteinized reaction mixtures were heated for 7 min in a boiling water bath to hydrolyze any [Y-~'P]ATP, esterified 32P was separated from 32Pi by extraction with isobutyl alcohol/ benzene (lo), and the amount of gluc0se-6-~'P was determined by counting the aqueous phase. Controls, minus hexokinase and particles, were routinely carried through the extraction procedures in order to obtain a correction factor for radioactivity appearing in the aqueous phase that was not gluco~e-6-~'P. This number was less than 0.5% of the total radioactivity in the reaction mixture. Radioactivity was quantitated by adding 3.5 ml of the organic phase or 1 ml of the aqueous phase to 10 ml of a Triton/toluenebased scintillation mixture (11). Alternatively, Liquiscint (National Diagnostics) was used in the same proportions. Radioactivity was counted in a Beckman LS-355 liquid scintillation counter.
Adenine nucleotides (ATP, ADP, and AMP) were separated by chromatography on plastic sheets layered with polythelenemine-impregnated cellulose (Brinkmann Instruments). The sheets were washed before use by ascending chromatography with water and airdried. The samples containing radioactive and carrier nucleotides were applied to the sheets, and the chromatograms were developed with a solvent consisting of 2 N formic acid and 0.5 N LiC1. The sheets were dried, and the resolved nucleotides were identified under ultraviolet light and scraped into scintillation vials. The samples were extracted for 45 min with 0.5 ml of 1 N HCl, 5 ml of scintillant channels for 3H and 32P. (Liquiscint) was added, and radioactivity was determined in separate The amounts of hexokinase used in these experiments (200 units/ ml) were sufficient to convert 0.1 p~ [Y-~'P]ATP to gluc0se-6-~'P in 200 ms. Addition of 200 units of hexokinase to 1 ml of buffer containing 0.1 nmol of [Y-~'P]ATP converted 98% of the radioactive ATP to gluco~e-6-~'P. If 1 mg or more of submitochondrial particles were added before the hexokinase, only about 3-4% of the radioactive ATP was available to the hexokinase. If 200 units of hexokinase were added to 1 mg of submitochondrial particles before adding 0.1 nmol of [y3'P]ATP, 95% or more of the added [y3'P]ATP was converted to gluco~e-6-~'P.

RESULTS
Incubation of [y-32P]ATP with activated submitochondrial particles (specific activity = 6 units/mg or higher) results in binding of the radioactive ATP in high affinity catalytic sites and the establishment of an equilibrium in the catalytic sites between bound substrate and bound hydrolysis products (4). Since the equilibrium constant is approximately 0.5 (4), it is expected, as found in Table I, that a short-term incubation of radioactive substrate and a molar excess of membrane-bound F1 result in 20-40% hydrolysis of the bound substrate. Although the net rate of ATP hydrolysis during single site catalysis is limited by the rate at which products dissociate from the catalytic site (1, 4), a considerable increase in rate is observed when an oxidizable substrate such as succinate or NADH is made available to the submitochondrial particles ( Table I). The increased hydrolysis is energy-dependent, since the stimulation is reduced by rotenone (Experiment 3) or myxothiazol plus antimycin (Experiment 2) as well as the uncouplers FCCP and S-13 (Experiments 1, 3, and 4). It is noteworthy that only the reduced form of the pyridine nucleotide is effective in stimulating hydrolysis (Experiment 5 ) and that ethanol, the solvent for the various inhibitors, was without effect on the extent of hydrolysis (Experiment 2).  MgSO4, 0.092 p M [y3'P]ATP (specific activity = 1-5 X lo6 cpm/ nmol), 4 mM Pi, and 1 mg/ml activated submitochondrial particles (the specific activity was 6.1 units/mg for Experiments 3-5 and 10 units/mg for Experiments 1 and 2). Where indicated, the reaction mixture also contained 10 mM sodium succinate, 1 mM NADH, 1 mM NAD, 1.8 p M FCCP, 8 pM myxothiazol, 10 p M antimycin, 0.4 pg of rotenone, or 4 pl of ethanol. All inhibitors were dissolved in ethanol and added in 4-pl volumes to reaction mixtures. In Experiment 1, the reaction mixture was incubated with succinate but without [T-~'P] ATP for 7 min at room temperature in order to permit development of a high steady state rate of respiration. [y-32P]ATP was added and the incubation was continued for 20 s before addition of 50 p1 of a stopping solution containing 60% perchloric acid and 5 mM ATP. In Experiments 2 and 4, the submitochondrial particles were incubated, with or without inhibitors, for 1 min at room temperature. [Y-~'P] ATP was added and, after an additional 5 s to allow the radioactive substrate to bind to the particles, NADH or NAD was added as indicated. For Experiment 3, the submitochondrial particles were incubated with rotenone for 10 min before addition of [y-32P]ATP. Respiration was allowed to continue for 20 s (Experiment 4) or for 40 s (Experiments 2, 3, and 5) before addition of 50 pl of stopping solution. Tubes that did not receive an oxidizable substrate were incubated with [y3'P]ATP for the same time periods as the experimental tubes. 32Pi and [Y-~'P]ATP were separated and counted as described under "Experimental Procedures." Per cent hydrolysis is calculated as the per cent of the total [Y-~'P]ATP added to reaction mixtures. 32P]ATP bound in high affinity catalytic sites is strongly dependent on the presence of Pi and in fact the stimulation is increased with increasing Pi concentration. In the absence of NADH, hydrolysis was not influenced by Pi up to 8 mM (Fig. 2). Addition of NADH to a reaction mixture containing no added Pi was accompanied by a small but detectable increase in hydrolysis. However, at higher concentrations of Pi, a considerable enhancement in the hydrolysis of bound radioactive ATP occurs and the levels observed, which reach   Table I.
In order to rule out the possibility that the NADH-stimulated hydrolysis of [ Y -~~P ] A T P might represent an energydependent exchange reaction between the radioactive y-phosphoryl of ATP bound in the high affinity catalytic site and Pi in the medium, an experiment was carried out with [3H,y-32P]ATP. The results are shown in Table 11. It  Further investigation of the NADH-dependent hydrolysis of [-p3'P]ATP indicated that the mechanism of the stimulation included dissociation and rebinding of radioactive ATP, since the presence of hexokinase in the reaction mixture effectively prevented any stimulation. It may be seen in Fig.  3A that, while the NADH-stimulated hydrolysis (curue I ) was similar to that observed in Fig. 1, in the presence of hexokinase the rate of hydrolysis was similar to that observed in the absence of NADH (compare curue 2 of Fig. 3A and curue A of Fig. 1). In the absence of glucose, hexokinase failed to inhibit ATP hydrolysis (not shown). Additional evidence supporting the conclusion that the mechanism of the stimulation of hydrolysis by NADH includes dissociation and rebinding of [ T -~~P I A T P is seen in Fig. 3B. In this experiment, a constant amount (0.092 nmol) of [ T -~~P I A T P was added to increasing amounts of activated submitochondrial particles. At low concentrations of submitochondrial particles, a marked NADHdependent stimulation of hydrolysis is observed. However, as the concentration of submitochondrial particles increases, and thus the availability of empty catalytic sites, the extent of the stimulation by NADH is reduced to a point approximating the per cent hydrolysis observed in the absence of NADH. It should be noted that at 6 mg of submitochondrial particles, respiration would continue for about 10 s before exhaustion of the oxygen in the reaction mixture (see "Experimental Procedures") and, in the absence of other considerations, should have caused a 50% increase in hydrolysis (Fig. 3A).
Direct evidence for the participation of NADH in an energy-dependent dissociation of [y-32P]ATP from high affinity catalytic sites is shown in Table 111. In Experiment 1, line a, it may be seen that addition of 0.087 nmol of [y-32P]ATP to 1 mg of submitochondrial particles results in the binding of 97% of the added nucleotide, since only about 3% is subsequently available to hexokinase. Addition of 1 mM NADH (line b) results in the dissociation of about 27% of the bound nucleotide in an energy-dependent fashion, since in the presence of either FCCP or s-13 the extent of dissociation is considerably reduced. Inhibitors of respiration also reduce the extent of NADH-dependent dissociation of [y3'P]ATP (Table 111, Experiment 2). Whereas 30% of the bound [y-"P] ATP dissociates following addition of NADH (line b) if the submitochondrial particles are pretreated with myxothiazol and antimycin (line c), only 12% of the bound nucleotide dissociates. As in other experiments with water-insoluble inhibitors described in this paper, ethanol alone (line d) was without effect on NADH-dependent dissociation of [r-"P] ATP. Table 111, Experiment 3, also shows that the NADHdependent dissociation of [T-~'P]ATP is not influenced by Pi.
The slightly increased total binding of [ Y -~~P I A T P in the presence of Pi (compare line b with line a) is routinely observed in these experiments. Table I11 also indicates that approximately one-third of the bound radioactive nucleotide dissociates following the addition of NADH. In Table IV it may be seen that the extent of the dissociation (about one-third) is independent of the concentration of either [y3'P]ATP or submitochondrial particles.
The fate of [y3'P]ATP, bound in high affinity catalytic sites of activated submitochondrial particles following the addition of NADH and ADP, is shown in Fig. 4. Several points are immediately apparent. First, of 0.089 p~ [ Y -~~P ] ATP incubated with 1 mg/ml submitochondrial particles, 0.085 p~ is bound in high affinity catalytic sites. Immediately after mixing with NADH and ADP, bound [ Y -~~P ] A T P dis- 32P]ATP was 1.3 X lo6 cpm/nmol. The particles were incubated with all additions except for radioactive ATP and NADH for 1 min at room temperature (26 "C). After ATP addition, the incubation was continued for 45 s (Experiments 1 and 3). For Experiments 2 and 4, the particles were incubated with radioactive ATP for 5 s, NADH was added, and incubations were continued for 40 s. Reactions were stopped by adding 50 p1 of 60% perchloric acid followed by 25 pl of a solution containing 1 mM each nonradioactive ATP, ADP, and AMP. Experiments 3 and 4 received, in addition, 5 p1 of 0.2 M Napi. Denatured protein was removed by centrifugation. The supernatants were cooled on ice and 90 p1 of 6 N KOH were added. The neutralized extracts were cooled on ice for 10 min and centrifuged to sediment the precipitates. Aliquots of the supernatants (0.5 ml) were withdrawn and extracted with isobutyl alcohol/benzene as described under "Experimental Procedures" for analysis of 32Pi formed. Individual amounts of the radioactive forms of ATP, ADP, and AMP present in the neutralized extracts were determined by chromatographing 10-p1 aliquots on polyethyleneimine-impregnated cellulose sheets as described under "Experimental Procedures." The values for 3H-nucleotides shown were corrected for 70% recovery from the chromatograms. The values for [3H]ADP and [3H]AMP also were corrected for the small amounts of these compounds that were present in the original stock solutions: 0.002 and 0.001 nmol, respectively. Per cent hydrolysis was calculated as the ratio of [3H]ADP to total 3H-nucleotides present or as the ratio of 32Pi formed to the total [y-32P]ATP added. Both ratios were multiplied by 100.  Thus, initiation of oxidative phosphorylation by adding NADH plus ADP to a suspension of submitochondrial particles containing [y3'P]ATP bound in high affinity catalytic sites is followed by dissociation and hydrolysis of 90% of the bound nucleotide at a rate commensurate with the turnover of F, in oxidative phosphorylation.

DISCUSSION
The most striking observation made in this paper is the energy-dependent dissociation of ATP bound in high affinity catalytic sites of submitochondrial particles. Dissociation requires an oxidizable substrate such as succinate or NADH, is dependent upon respiration, and is inhibited by uncouplers of oxidative phosphorylation. Of equal importance is the fact that the rate of the energy-dependent dissociation is at least as fast as the initial rate of ATP synthesis in oxidative phosphorylation. Thus, two criteria of kinetic competence suggest that the high affinity catalytic sites are normal catalytic sites on the enzyme. First, the maximum rate of hydrolysis of [y-32P]ATP bound in high affinity catalytic sites of the soluble (2, 12) or membrane-bound (4) forms is the same as the normally observed turnover number (600-700) for the enzyme. Second, as shown in Fig. 4 S-13, 4 mM Pi, 4 p1 of ethanol, and 200 units of crystalline yeast hexokinase. FCCP and S-13 were dissolved in ethanol and added to reaction mixtures in volumes of 4 p1. The final volume was 1.0 ml. The specific activities of the submitochondrial particles were 6.1 units/mg (Experiments 1  and 3) and 9 units/mg (Experiment 2). The submitochondrial particles were incubated at room temperature for 1 min with all additions indicated except [y-32P]ATP, NADH, and hexokinase. Experiment 1, line a, [y3'P]ATP was added and the reaction continued for 25 s before addition of 1.0 ml of 2 N HC1; lines b-d, incubation with [y-32P]ATP was continued for 5 s to form the enzyme-substrate complex, hexokinase was added, and an additional 5 s was allowed before addition of NADH. The reaction was allowed to continue for an additional 15 s before it was terminated with 1.0 ml of 2 N HC1. Experiment 2, lines a and b, the submitochondrial particles were incubated with [Y-~'P]ATP for 5 s, hexokinase was added, and the incubation was continued for 40 s before addition of 1.0 ml of 2 N HCl; lines c and d, NADH was added immediately after the hexokinase, and incubation was continued for 40 s and stopped with 1 ml of 2 N HC1. Experiment 3, the submitochondrial particles were incubated with all additions above but Pi, NADH, and hexokinase for 1 min. Where indicated, Pi was included in the incubations. [y-32P] ATP was added, followed 5 s later by hexokinase. Lines a and b, the incubations were continued for 20 s; lines c and d, NADH was added 5 s after hexokinase and incubations were continued for 15 s. Reactions were stopped with 1 ml of 2 N HC1. The deproteinized reaction mixtures were treated as described under "Experimental Procedures" to determine the amounts of free [y-32P]ATP. The amount of glucose-6-32P found in the aqueous phase is equated with "free" [y-32P]ATP. Per cent [y-32P]ATP dissociated is calculated as the nanomoles of gluc0se-6-~'P formed divided by the nanomoles of [y-32P]ATP added, mdtidied bv 100. high affinity catalytic sites of submitochondrial particles dissociates at rates expected for the turnover of the membranebound enzyme in oxidative phosphorylation.

, [y3'P]ATP bound in
These observations are directly relevant to the molecular mechanism of ATP synthesis in oxidative phosphorylation. The observations support a model €or ATP synthesis which proposes that ATP is formed from bound ADP and Pi, with little or no change in free energy and that the major requirement for energy in oxidative phosphorylation is for the dissociation of product ATP from catalytic sites (1, 4, 5 ) . Since the K, for ATP binding in high affinity catalytic sites is 10l2 M-' (l), release of product ATP formed in the same sites  The reaction mixture contained 20 mM Tris-SO4, pH 8, 3 mM MgSO,, 4 mM Pi, 1 mM NADH, [y-32P]ATP, and activated submitochondrial particles in the amounts shown. The final volume was 1.0 ml. The particles were incubated for 1 min at room temperature in the reaction mixture minus NADH, [y-32P]ATP, and hexokinase. [y-32P]ATP was added and incubated for 5 s, followed by hexokinase and an additional 5 s of incubation, and, finally, NADH was added and incubated for 40 s. The total incubation time for all tubes was the same. The reaction was stopped by adding 1 ml of 2 N HC1, and samples of the deproteinized reaction mixtures were treated as described under "Experimental Procedures" in order to separate 32Pi from esterified 32P. The per cent [y-32P]ATP free is equated with the amount of gl~cose-6-~'P found in the aqueous phase and is calculated relative to the total [y-32P]ATP added. would require a decrease in binding affinity of many orders of magnitude. It is clear from experiments described in this paper that energy released during oxidation of substrates by the respiratory chain is capable of causing such a decrease in binding affinity. The outline of a coupling device that might link the energy store to reversible alterations in the affinity of the catalytic site for ATP was presented recently. Treatment of submitochondrial particles with dicyclohexylcarbodiimide under conditions such that only the single carboxyl group of subunit c in Fo should be altered was followed by impaired binding of ATP in high affinity catalytic sites 13). Since the glutamate carboxyl group is approximately 20 A from a presumed catalytic site on F, (13), it was proposed, in keeping with previous suggestions (6, 7), that chemical modification caused a conformational change in FO which, transmitted to F1, caused a conformational change in the ATPase. The resulting reduced affinity of the catalytic sites for ATP was considered to arise from these changes in F, (3). The possibility is thus raised that, during oxidative phosphorylation, a change in the state of ionization of charged amino acid residues in Fo, in response to electrochemical proton gradients, also leads to a conformational change in FO that is transmitted to F1 and gives rise to reversible changes in affinity for product ATP (3).
There is precedent for the suggestion that catalytic sites on F1 can undergo very large changes in binding affinity. Beef heart F1 binds TNP-ATP and TNP-ADP with equal high affinity. The affinity constant is 1 or more orders of magnitude greater than lo9 M-' (12). Since TNP-ATP is a substrate for F1 and product TNP-ADP must be released, it is clear that during turnover large, reversible shifts in binding affinity take place.
The energy-dependent stimulation of the hydrolysis of [y-32P]ATP bound in the high affinity catalytic site is of some interest. The stimulation requires oxidation of substrates such as succinate or NADH and is inhibited by uncouplers ( Table  I) The experiment was carried out in a rapid mixing apparatus, using three syringes and two mixers (1). Each syringe contained 50 mM NaHepes, 10 mM Napi, 2 mM MgSO,, and 50 mM glucose. The pH was 8. In addition, syringe 1 contained 2 mg/ ml activated submitochondrial particles, syringe 2 contained 0.178 p M [y3'P]ATP (specific activity = 3.4 X lo7 cpm/nmol), and syringe 3 contained 3 mM NADH, 0.3 mM NaADP, and 3.9 mg/ml hexokinase. Equal volumes (86 pl) from syringes 1 and 2 were mixed via mixer I in the first push on the syringe plungers, and the mixed reactants were allowed to age in hose A for 5 s in order to form the enzymesubstrate complex. A second push on the plungers mixed most of the contents of hose A, via mixer 11, with a half-volume from syringe 3. (Thus, the components in syringe 3 were diluted 3-fold.) The mixed reactants were allowed to incubate in hose B for periods of time indicated on the abscissa and then injected through the nozzle into the receiving vessel. The latter contained 1 ml of a solution consisting of 1.7 N HCl, 1 mM ATP, and 0.5 mM glucose-6-P. The final volume after collecting the sample was 1.7 ml. The amount of [y-32P]ATP free in solution after the 5-s incubation in hose A was determined by disconnecting mixer I1 and injecting the aged reaction mixture through the nozzle directly into 1 ml of a solution that contained 50 mM NaHepes, 10 mM Napi, pH 8, 3 mM MgS04, 50 mM glucose, and 1.3 mg of hexokinase. After 5 s of incubation, the reaction was stopped by adding 1.2 ml of a solution that contained 2 N HC1, 1 mM ATP, and 0.5 mM glucose-6-P. Equal volumes of deproteinized reaction mixtures (0.5 ml) were removed. One was heated for 7 min in a boiling water bath and then extracted with isobutyl alcoholfienzene. The second was directly extracted with isobutyl alcoholfienzene. The amounts of gl~cose-6-~'P and [Y-~'P]ATP present were determined as described under "Experimental Procedures." All measurements of 32Pi were corrected for the small amount (4%) of 32Pi present in controls incubated without submitochondrial particles. Measurements of gl~cose-6-~'P were corrected for radioactivity (usually less than 0.5% of the total added) found in the aqueous phase in the absence of hexokinase. The amount of glu~ose-6-~'P formed is set equal to "free" [Y-~'P]ATP. The data are calculated to show the nanomoles of 32Pi or glu~ose-6-~'P present in 1 ml of mixed reactants in hose A. The amount of 32Pi present at zero s of incubation of the enzyme-substrate complex with NADH and ADP (shown on the ordinate) is that expected for the equilibrium distribution of substrate and products in the catalytic site (3). stimulation of hydrolysis is observed (Fig. 3). The order of addition of hexokinase is important. If there is a delay of more than 3 s between additions of NADH and hexokinase, very little gl~cose-6-~'P is formed? Rapid rebinding of dissociated [T-~'P]ATP is expected from the bimolecular rate constant describing ATP binding in either the high affinity catalytic site or additional catalytic sites on the membranebound enzyme (4). Rapid rebinding is independent of the presence or absence of Pi2 and serves to explain why large amounts of hexokinase are required to demonstrate the energy-dependent dissociation of ATP. NADH-stimulated hy-' H. S. Penefsky, unpublished experiments. drolysis also is not observed in the presence of an excess of submitochondrial particles (conditions favoring single site catalysis for the rebound ATP) (Fig. 3).
While the NADH-stimulated hydrolysis requires Pi (Fig.  2), there was no effect of Pi on the NADH-dependent dissociation of ATP (Table 111). The requirement for Pi cannot be explained in terms of a Pi-ATP exchange reaction that would cause 32Pi to appear in the medium, since this is ruled out by the experiment of Table 11. In addition, the rate of the 32Pi-ATP exchange reaction continues to increase even at 40 mM Pi (14), whereas maximum NADH-dependent hydrolysis occurred at about 4 mM Pi (Fig. 2).
The NADH-dependent stimulation of hydrolysis should be distinguished from the stimulated hydrolysis observed under the conditions of oxidative phosphorylation, that is, in the presence of NADH, ADP, Pi, and hexokinase (Fig. 4). In the latter experiment, hexokinase prevented rebinding of dissociated ATP. The observed stimulation of hydrolysis appears to be due to ADP alone, since ADP serves as a promoter molecule in the acceleration of the hydrolysis of [Y-~'P]ATP bound in high affinity catalytic sites (4). ADP alone, in the absence of NADH, also gives rise to a small, but reproducible dissociation of bound [y3'P]ATP. Approximately 4% of the total ATP bound dissociates following addition of 0.1 mM ADP. ' The finding that only about one-third of the ATP bound in high affinity catalytic sites participates in an energy-dependent dissociation could be suggestive of heterogeneity in binding sites. However, there appears to be little heterogeneity in binding sites when cold chase experiments are carried out with submitochondrial particles (4). Alternatively, the extent of the observed dissociation might be influenced by the presence of nonfunctional phosphorylating assemblies which nevertheless bind ATP in high affinity catalytic sites. The fact that the extent of dissociation was the same for a number of different preparations of activated submitochondrial particles with differing specific ATPase activities (4.6-10 units/ mg) argues against such an interpretation. The proposal of Boyer and co-workers (15) of alternation between three catalytic sites in the mechanism of action of F1 could help to explain partial dissociation. Initially, [y-32P]ATP would bind randomly in high affinity catalytic sites on the unenergized membrane. Subsequent generation of an electrochemical proton gradient via oxidation of NADH would impose an asymmetry on the bound enzyme such that one of the three sites becomes poised for dissociation. As a result, the affinity for ATP in one of three sites is reduced, and one-third of the bound radioactive nucleotide dissociates. In the presence of NADH and Pi, a second turnover, which should cause additional rapid dissociation of [y3'P]ATP, apparently does not occur. In the presence of NADH, ADP, and Pi, a second turnover would be without effect on the extent of dissociation because 90% of the bound radioactive ATP is disposed of within 3 s. These aspects of the problem are currently under investigation.