Slow binding of ATP to noncatalytic nucleotide binding sites which accelerates catalysis is responsible for apparent negative cooperativity exhibited by the bovine mitochondrial F1-ATPase.

The bovine heart mitochondrial F1-ATPase depleted of nucleotides (nd-MF1) hydrolyzes 50 microM ATP in three kinetic phases at 30 degrees C. An initial "burst" rapidly transforms into an intermediate, slower rate, which slowly accelerates to the final, steady-state rate. The intermediate phase disappears progressively as the concentration of ATP in the assay medium is increased and is absent at 2 mM. Activation in the intermediate phase is lost when nd-MF1 is inactivated by 5'-p-fluorosulfonylbenzoyladenosine, which modifies three noncatalytic sites. Correlation of [3H]ATP binding to nd-MF1, after treatment either with 50 microM Mg[3H]ATP plus a regenerating system or 10 mM free [3H]ATP, with stimulation of the intermediate phase suggests that this phase is abolished when at least two noncatalytic sites are filled with ATP. Prior incubation of nd-MF1 with MgPPi stimulates hydrolysis of 30 microM to 2 mM ATP and abolishes the intermediate phase. Following incubation with Mg[32P]PPi, 3.3 mol of [32P]PPi/mol of enzyme are bound, 1 and 0.5 mol of which are released by cold chases with MgATP and MgITP, respectively. Since the cold chases diminish activation only slightly, the stimulatory effect is not caused by PPi binding to catalytic sites. A Lineweaver-Burk plot of initial rates of the intermediate phase for hydrolysis of 30 microM to 2 mM ATP by nd-MF1 is biphasic, extrapolating to apparent Km values of 120 and 440 microM. The latter value is the same as the apparent Kd determined from dependence of the rate of activation of the intermediate phase on ATP concentration in the assay medium. After prior incubation of nd-MF1 with MgPPi or free ATP, Lineweaver-Burk plots are linear with the highest Km disappearing. Thus, this Km reflects rate acceleration when ATP binds to noncatalytic sites. From these results it is concluded that slow binding of ATP to noncatalytic sites during hydrolysis of low concentrations of substrate, which accelerates catalysis, is responsible for apparent negative cooperativity exhibited by MF1.

From the amino acid sequences of its component subunits, the molecular weight of the bovine heart enzyme is 371,000 (1). The F,-ATPases contain six nucleotide binding sites (2)(3)(4). For the beef heart enzyme, three of these sites bind and exchange nucleotides readily during catalysis. Thus, each is a potential catalytic site. In contrast, the other three sites, once filled with adenine nucleotides, exchange very slowly during turnover and are therefore called noncatalytic (2,5).
Chemical modification studies with 2-azido-ATP have provided evidence that Tyr-345 of the 0-subunit is located at catalytic sites of the beef heart enzyme, whereas Tyr-368 is located at noncatalytic sites (6). Selective modification of the catalytic or the noncatalytic site has also been achieved using two other nucleotide analogues, FSBI' and FSBA. Maximal inactivation by the former reagent is accompanied by modification of Tyr-345 in a single 0 subunit (7), whereas maximal inactivation by the latter is accompanied by mutually exclusive modification of Tyr-368 or His-427 in all three p subunits That noncatalytic sites are involved in control of catalysis has been suggested on the basis of several experimental criteria with different F1-ATPases. Steady-state kinetic analyses of F1-ATPases from different sources have revealed apparent negative cooperativity when ATP hydrolysis is measured over a wide range of substrate concentration. Two (9-15) or three (16,17) K,,, values have been deduced from steady-state kinetics, depending on the conditions of analysis. The participation of noncatalytic sites in the control of catalysis has also been suggested from the catalytic properties of mutant enzymes with defective ATPase activity (14, 18) and the characteristics of ADP-induced hysteretic inhibition (19)(20)(21). Furthermore, it has been shown recently that the ATPase activity of the CF1-ATPase from spinach chloroplasts depends on the binding of ATP to noncatalytic sites (22,23). We report here that, at 30 "C, saturation of noncatalytic sites with ATP stimulates ATPase activity of the mitochondrial F1-ATPase depleted of endogenous nucleotides (nd-MF1) to an optimal steady-state level. EXPERIMENTAL PROCEDURES ously described (24) modification of the method of Knowles and MF, was prepared from bovine heart mitochondria with a previ-Penefsky (25) and was depleted of nucleotides by gel permeation chromatography in the presence of 50% glycerol (v/v) as described by Garrett and Penefsky (26). After chromatography, protein frac- The abbreviations used are: FSBI, 5'-p-fluorosulfonylbenzoylinosine; FSBA, 5'-p-fluorosulfonylbenzoyladenosine; S-Ns-FSBA, 5'p-fluorosulfonylbenzoyl-8-azidoadenosine; MF1, bovine heart mitochondrial F,-ATPase; nd-MF1, MF1-ATPase depleted of nucleotides; CF,, F1-ATPase from spinach chloroplasts,; TF,, F1-ATPase from the thermophilic bacterium PS3. t ions with an A2XO/A260 ratio of >1.95 were stored at room temperature in 100 mM Tris-SO.,, pH 8.0, containing 4 mM EDTA and 50% glycerol (v/v). When submitted to nucleotide analysis by HPLC as previously described (21), the nd-MF1 used in this study contained about 0.2 mol of ATP/mol of enzyme and essentially no bound ADP. The nd-MF, hydrolyzed 100-120 wmol of ATP/mg of protein a t 30 "C. ATPase activity was determined spectrophotometrically at 30 "C with an assay medium containing 50 mM Hepes/KOH, pH 8.0,30 mM KC1,4 mM phosphoenolpyruvate, 0.4 mM NADH, 20 pg/ml lactate dehydrogenase, 40 pg/ml pyruvate kinase (both enzymes obtained from Boehringer Mannheim as solutions in 50% glycerol), and MgC12 in 1 mM excess over the concentration of ATP specified in the figure legends. ATP concentrations were determined by absorbance at 259 nm ( e = 15,400 M" cm"). Protein concentrations were determined by the bicinchoninic acid method described by Smith et al. (27). The radioactive ["HIATP (45 Ci/mmol, Amersham Corp.) and ["PIPP, (3 Ci/ mmol, Du Pont-New England Nuclear) were diluted with the nonradioactive reagents to the desired specific radioactivity.
Binding of ["]ATP to nd-MF1 was performed using 3-ml centrifuge columns filled with 2.5 ml of wet Sephadex G-50 (28) under conditions specified in the figure legends. Binding of ["'PIPP: to nd-MF, was assessed using 3-ml centrifuge columns filled with 1.5 ml of wet Sephadex G-50 according to the procedure of Issartel et al. (29).
T h e purity of ["'PIPP, solutions was checked by thin layer chromatography on silica gel (29). After radioautography, the radioactive areas were removed from the thin layer sheets and were submitted to liquid scintillation counting. The samples of ["PIPPi used contained less than 3% contaminating ["'PIP,.
Biochemicals used in the assays and buffer components were purchased from Sigma. Bicinchoninic acid was purchased from Pierce Chemical Co. FSBA or FSBI were synthesized as described elsewhere (30). Radioactivity was detected with a T m Analytic 6895 liquid scintillation counter using Liquiscint from National Diagnostics.

A n Activation Phase Is Exhibited When nd-MF1 Hydrolyzes
Low, but Not High Concentrations of ATP-Hydrolysis of 50 P M ATP by nd-MF1 proceeds in three distinct phases as shown in Fig. 1 (trace a). In the first two phases, an initial "burst," lasting less than 20 s, rapidly transforms to a slow, interme- 1. Illustration of the three kinetic phases observed during hydrolysis of 50 HM ATP by nd-MF1. A solution of 0.7 mg/ ml nd-MF1 was prepared in 50 mM Tris-S04, pH 8.0, containing 0.4 mM EDTA and 10% (w/v) glycerol. The traces shown in the figure were recorded when the following amounts of this solution were assayed with 50 p~ ATP using the regeneration system described under "Experimental Procedures," which contained 21 pg/ml pyruvate kinase and 11 pg/ml lactate dehydrogenase: traces a and b, 3 pl; trace c, 2 @I; trace d, 1 p l . The assay mixture that produced trace b also contained 10 mM Na,SO,. diate phase of hydrolysis. This phenomenon has been described in detail by several laboratories (31-35). The slow intermediate phase gradually accelerates to a final steadystate rate. The latter transition is an activation process which has not been previously described for MF1. Whereas the initial burst is not affected, 10 mM Na2S04 in the assay medium nearly completely abolishes activation of the intermediate phase as shown by trace b. Sulfate introduced with pyruvate kinase and lactate dehydrogenase is sufficient to affect the intermediate phase. Therefore, these enzymes were added as solutions in glycerol to provide strict exclusion of sulfate from the assay medium. As shown by traces a, c, and d, which are recordings of assays initiated by addition of 2.1, 1.4, and 0.7 p g of nd-MF1, respectively, the magnitude of the burst phase decreases with decreasing quantities of enzyme in the assay medium and is barely discernible when 0.7 pg of enzyme are assayed. In contrast, activation of the intermediate phase remains prominent when less than 1 pg/ml enzyme is assayed. In order to eliminate interference from the initial burst, the remainder of the enzyme assays reported in this study were carried out with less than 1 pg of nd-MFl/ml. Recordings of hydrolysis were initiated 20 s after addition of enzyme to the spectrophotometer cell. Therefore, "initial rate" in these experiments refers to the rate in the 20-60-s interval after initiating hydrolysis by addition of enzyme. the assay medium is increased, the intermediate phase becomes progressively shorter and is entirely absent when assays are conducted at 2 mM ATP as shown by Fig. 2B (trace a). Prior incubation of nd-MF1 with either 10 mM ATP (Fig. 2 A , trace b) or 50 PM ATP in the presence of Mg2+ and an ATP regenerating system (not shown) also eliminates the intermediate phase. These prior incubations have no effect on the linear, initial rate of hydrolysis of 2 mM ATP (Fig. 2B, trace  b). Prior incubation of nd-MF1 with 2 mM PPI in the presence of Mg2+ not only abolishes the intermediate phase observed during hydrolysis of 50 PM ATP, but also stimulates ATPase activity during assay at low and high ATP concentrations as shown by comparison of traces c with traces a in Fig. 2 (A and  B ) . Addition of 20 mM bicarbonate to the assay medium does not eliminate the intermediate phase observed during hydrolysis of 50 PM ATP but does stimulate the final steady-state rate and the rate of hydrolysis of 2 mM ATP (Fig. 2, panels A and B, traces d).
Modification of Noncatalytic Sites with FSBA Abolishes the Lag Phase Observed during Hydrolysis of 50 or 100 P M A T P by nd-MF1-Recordings for hydrolysis of 50 PM ATP are progressively transformed from curved to linear as nd-MF1 is inactivated with FSBA. Trace e of Fig. 2A shows a recording of the hydrolysis of 50 PM ATP catalyzed by nd-MF1 which had been inactivated by 81% with FSBA. In contrast, the lag phase remains after inactivation of nd-MF1 with FSBI. Trace f of Fig. 2A illustrates hydrolysis of 50 PM ATP by nd-MF1 inactivated by 55% with FSBI. The lag phase remains even after inactivation of nd-MF1 with FSBI reaches 95% (not shown). Recordings of activity remained linear when residual ATPase activity was determined with 2 mM ATP during inactivation of nd-MF1 with either FSBA or FSBI as illustrated by Fig. 2B (traces e and f ).
Recordings for hydrolysis of 50 PM ATP by samples of nd-MFl undergoing inactivation by FSBA become more linear as noncatalytic sites are increasingly modified. This suggests that the progressive increase in activity observed during the lag phase exhibited by unmodified enzyme is caused by slow binding of ATP to noncatalytic sites. Therefore, it was of interest to compare the rate of inactivation of nd-MF1 by FSBA, determined by monitoring the initial, slow rate, on the one hand, and the final, fast rate, on the other, during hydrolysis of low concentrations of ATP. As illustrated in Fig.  3A, the initial rate of hydrolysis of 100 PM ATP is much less sensitive to inactivation by FSBA than is the final steadystate rate. The half-times observed when inactivation is measured by monitoring the initial, slow rate and final, fast rate of ATP hydrolysis are, respectively, 42 and 15 min. When inactivation by FSBA is monitored by assaying samples with 2 mM ATP, the rate of inactivation is close to that determined by monitoring the final steady-state rate during hydrolysis of 100 PM ATP (t1/2 of about 19 min.). After prior incubation of nd-MF1 with 10 mM ATP, the rate of inactivation of the enzyme by FSBA is the same when residual activity is monitored by assay with either 100 PM or 2 mM ATP as illustrated in Fig. 3B. It should be noted that the FSBA concentration is higher in this experiment in order to compensate for the protective effect of ATP.
Also shown in Fig. 3B, prior incubation of nd-MF1 with 2 mM PPi plus M$+ leads to the same rate of inactivation of the enzyme when residual activity is monitored by assaying samples at 100 PM or 2 mM ATP. Comparison of the rate inactivation of nd-MF1 monitored with 2 mM ATP in the assay medium in the presence (   Fig. 4B. The results obtained after the cold chase indicate that maximal stimulation occurs when two noncatalytic sites are saturated with ["]ATP. Fig. 5A shows that prior incubation of nd-MF, with 50 MM ATP plus 1 mM M$+ in the presence of an ATP regenerating system leads to a time-dependent disappearance of the lag phase when the activity is subsequently monitored by assay with 50 p~ ATP. Measurements of the binding of [3H]ATP to the enzyme under these conditions (Fig. 5 B ) show that   (Fig. 44, lower curue). The high concentration of ATP used ATP PIUS a regenerating system. Relative activation is defined in the in the cold chase is necessary to for the high legend of Fig. 4 is loaded with higher concentrations of [3HlATP remove unbound ["HIATP with ( 0 ) or without (0) a 20-s cold chase reflects filling of noncatalytic sites. Correlation of the disapby addition of ATP and MgC1, to final concentrations of 2 mM each, about 4.5 mol of nucleotide/mol of enzyme are bound during the time required to abolish the lag phase (upper curve). However, after a cold chase with 2 mM ATP plus 2 mM Mg*+ t o unload catalytic sites, about 3 mol of nucleotide remain bound to the enzyme (Fig. 5B, lower curve). Approximately 1.5 mol of nucleotide/mol of MF, is removed in the cold chase, irrespective of the time of initial incubation of enzyme with ["HIATP, suggesting that only noncatalytic sites are involved in eliminating the lag phase. Fig. 5A (inset) shows that activation of nd-MF, by ATP plus Mg2+ in the presence of the regeneration system occurs with saturation of the second and third noncatalytic sites with ATP. Taking into account differences in experimental conditions, this observation is not necessarily inconsistent with the results illustrated in Fig. 4, which show that prior incubation of nd-MF, with free ATP eliminates the lag phase with apparent filling of the first two noncatalytic sites. When loading of noncatalytic sites is accomplished with free ATP, the high affinity noncatalytic site, which is only seen in the presence of Mg2+, might fill rapidly on introducing the enzyme to the assay medium.
Correlation of the Binding of Pyrophosphate to nd-MF, with Stimulation of ATPase Activity-As shown in Table I, incubation of nd-MF1 with [32P]PPi a t concentrations that stimulate the hydrolytic activity %fold (see Fig. 2) is accompanied by binding of about 3 mol of [32P]PPi/mol of enzyme as detected by gel permeation chromatography on centrifuge columns of Sephadex (2-50. When a cold chase with ATP or I T P preceded gel permeation chromatography, the amount of ['"P]PPi bound decreases by -1 and -0.5 mol/mol of enzyme, respectively. Interestingly, the enzyme remains nearly completely activated after a cold chase with ATP or ITP, indicating that at least part, if not all of the binding sites for PPi that participate in the activation process are not catalytic sites. Apparent Negative Cooperatiuity Exhibited during A T P Hydrolysis by nd-MF, Is Associated with A T P Binding to Noncatalytic Sites-Owing to the prominent lag in the intermediate phase observed when nd-MF1 hydrolyzes low concentrations ATP, the steady-state parameters obtained from kinetic analyses are critically dependent on whether the initial or final rates of this phase are analyzed. As illustrated in Fig. 6A (upper curve), the Lineweaver-Burk plot of initial rates of the intermediate phase exhibits apparent negative cooperativity with the following kinetic parameters obtained from extrapolated intercepts: K,, = 120 pM; Vmaxl = 284 s-' (43 units/ mg); K,, = 440 pM; Vmax2 = 726 s-l (110 units/mg). Apparent negative cooperativity is also exhibited on a Lineweaver-Burk plot constructed from the final steady-state rate data (Fig.  6A, lower curve). However, in this case, owing to the slight curvature throughout the plot, K, and V,,, values cannot be estimated with certainty.  Prior incubation of nd-MF, with 10 mM ATP or 2 mM PPi plus Mg2+ abolishes apparent negative cooperativity. The kinetic parameters estimated from Lineweaver-Burk plots constructed from rate measurements obtained following incubation with ATP or PPi (Fig. 5B) are, respectively: K , = 108 pM, and V,,, = 759 s-' (115 units/mg); K, = 120 p~, and V, , , = 1353 s" (205 units/mg). When assayed at low ATP concentrations in the presence of 20 mM bicarbonate, a lag phase remains as shown in Fig. 2 (trace d). Lineweaver-Burk plots of initial rates obtained in the presence of bicarbonate exhibit apparent negative cooperativity (not shown). In contrast, a Lineweaver-Burk plot of the final steady-state rates obtained in the presence of bicarbonate (Fig. 6B)

A T P Binding to
Semilogarithmic plots of the time-dependent activation observed during hydrolysis of 30-300 PM ATP are linear, from which first order rate constants, k', for the activation process are obtained as shown in Fig. 7A. As illustrated in Fig. 7B, a double-reciprocal plot of the activation constants, k', against ATP concentration is linear. Extrapolation of the line in Fig.  7B to the abscissa and ordinate intercepts reveals a maximal activation constant of 6.6 X lo-' sK1 and an apparent Kd of 430 PM for the activation process. The latter value is the same as K,,,, estimated by extrapolation of the steep phase of the upper Lineweaver-Burk plot shown in Fig. 6A. The pseudobimolecular rate constant calculated from the ratio, k'/Kd, is 1.6 X 10' M" S-'.

DISCUSSION
The results presented clearly show that nd-MF, hydrolyzes low concentrations of ATP in three kinetic phases. An initial 1 " " " " " i burst phase decelerates rapidly to a slow intermediate phase, which, in turn, gradually accelerates to a final steady-state rate. Binding of ATP to two noncatalytic sites during prior treatment of nd-MF, with high free ATP concentrations or low MgATP concentrations in the presence of a regenerating system correlates well with loss of the lag in the intermediate phase when the treated samples are subsequently assayed at low ATP concentrations. Correlation of binding with activation when preloading is performed with MgATP plus a regenerating system suggests that activation is caused by slow binding of MgATP to the second and third noncatalytic sites, whereas rapid binding of MgATP to the first site has no effect on the transition of the intermediate phase to the final steadystate rate. Activation of the intermediate phase appears to occur when only two noncatalytic sites are loaded by prior incubation of nd-MF1 with free ATP. However, it is probable that MgATP binds rapidly to the open noncatalytic site when enzyme preloaded in this manner is introduced into the assay medium.
From the preloading experiments, it appears that a major component of the final steady-state rate depends on binding of ATP to noncatalytic sites, whereas the initial rate of the intermediate phase is a basal rate which is observed in the absence of apparent regulatory site to catalytic site cooperativity. This contention is supported by the observation that the initial rate of inactivation of nd-MF1 by FSBA, which modifies noncatalytic sites, is considerably faster when loss of activity is monitored by assaying the final steady-state rate than when monitored by assaying the initial rate of the intermediate phase.
Comparison of ATP binding to noncatalytic sites when nd-MF, is incubated with high concentrations of free ATP or low concentrations of MgATP in the presence of a regenerating system suggests heterogeneity of noncatalytic sites. One noncatalytic site binds MgATP rapidly, but when filled alone, does not accelerate rate in the intermediate phase. Whether this site must be filled to observe the initial burst of ATPase activity and/or subsequent activation of rate in the intermediate phase when ATP binds to the second and third noncatalytic sites remains an open question. What relationship this site has to the noncatalytic site that binds MgADP with low affinity described by Kironde and Cross (37) is not obvious.
The initial burst observed during hydrolysis of low concentrations of ATP was first reported for native MF1 containing endogenous nucleotides and for submitochondrial particles by Vasilyeva et al. (31). Subsequently Drobinskaya et al. (32) reported that a lag phase replaces the initial burst phase after treating MFl containing endogenous nucleotides with Mg2+ prior to assay with 100 PM ATP. The rate of the lag phase induced by treatment with Mg2+ slowly accelerat.es to the rate of the slow phase observed in the absence of treatment with M$+ (the intermediate phase described in this study). They also found that prior treatment of nd-MF1 with M$+did not alter the initial burst phase, but reported that treatment of nd-MF1 with an equivalent amount of ADP in the presence of M C also leads to replacement of the initial burst phase with a lag phase which slowly accelerates to a rate equivalent to the slow phase observed in the absence of pretreatment. From these observations, Drobinskaya et al. (32) proposed that deceleration in the initial burst phase in the absence of pretreatment is caused by abortive binding of MgADP formed during catalysis to a catalytic site. That this might be the case is supported by recent affinity labeling studies with 2-N3-ADP carried out independently by Milgrom and Boyer (34) using nd-MF1 and Chernyak and Cross (35) using MF, with two or three noncatalytic sites filled with adenine nucleotides.
Both laboratories reported that prior incubation of MF, with stoichiometric 2-N3-[32P]ADP in the presence of M$+elicits inhibition similar to that caused by MgADP binding. Subsequent irradiation of the resulting complexes is accompanied by derivatization of Tyr-/3345, which is present in the catalytic site. Therefore, the transition from the initial burst phase to the intermediate phase described in this and other studies (31)(32)(33)(34)(35) appears to be associated with retention of MgADP a t a catalytic site.
The slow rate acceleration observed in this study during transition of the intermediate phase to the final steady-state rate when nd-MF, hydrolyzes of 50 or 100 PM ATP was not noted in the studies cited above (31)(32)(33)(34)(35). In all but one of these studies, the rate of hydrolysis of 1-100 p~ ATP was recorded for only short time intervals after initiating catalysis, thus precluding observation of the activation phase. The presence of 1 mM sulfate in the assay medium used by Chernyak and Cross ( 3 5 ) may have precluded appearance of the activation phase in the longer time interval examined in their study. As shown in this study, sulfate inhibits the transition from the intermediate phase to the faster, final steady-state phase during hydrolysis of 50 PM ATP by nd-MF,, presumably by interfering with binding of ATP to noncatalytic sites.
Milgrom et al. (22) have reported direct evidence that sulfate lowers the affinity of noncatalytic sites of CFl for ATP. It is interesting that the CF, preparation which they examined contained about 2 mol of ADP/mol, one bound to a catalytic site and the other to a noncatalytic site. This enzyme preparation exhibits a lag when hydrolyzing 20 FM ATP. The lag is significantly extended in the presence of 5 mM sulfate. After prior incubation of the enzyme with ATP plus a regeneration system to load noncatalytic sites, the CF, preparation hydrolyzes 20 PM ATP without a lag in the presence or absence of sulfate. An initial burst phase is not observed when the CFl preparation hydrolyzes low concentrations of ATP, possibly because inhibitory ADP is bound initially at a catalytic site. The TF1-ATPase, which is isolated free of nucleotides (38), hydrolyzes ATP without exhibiting an initial burst. Yoshida and Allison (39) showed that the lag is extended after binding MgADP to a single catalytic site. Yohda et al. (14) reported that a mutant form of the cy subunit of TF, with the D261N substitution has a 10-fold diminished affinity for ADP. Whereas the Lineweaver-Burk plot for ATP hydrolysis catalyzed by the reconstituted a &~ complex of TF, containing wild-type cy subunit is biphasic and hydrolyses ATP with a pronounced lag, the Lineweaver-Burk plot for the reconstituted complex containing the mutant aDZ6lN subunit is linear and the mutant complex hydrolyzes ATP without a lag. These observations suggest that the apparent negative cooperativity exhibited by the cy&y complex of TFl and the lag observed in ATPase activity catalyzed by the complex are associated with binding of ATP to noncatalytic sites.
The observation that prior incubation of nd-MF, with PPi abolishes the intermediate phase and accelerates the final steady-state rate when the enzyme hydrolyzes 50 p~ ATP is consistent with the results of Kalashnikova et al. (33). They reported that the initial rate of hydrolysis of 100 PM ATP is unaffected by the presence of 2 mM PPi in the assay medium, but deceleration of the initial burst phase is markedly decreased resulting in a stimulated final rate. It was also shown that the inactive complex formed on incubating nd-MF1 with equimolar ADP in the presence of M$+ rapidly reactivates when treated with 1 mM PPi. From the observation that PPi increases the affinity of Pi, presumably bound to a catalytic site, Kalashnikova et al. (33) concluded that PPi binds to noncatalytic sites, In contrast, Peinnequin et al. (40) recently reported that PPi binds to three sites on MF,, irrespective of occupancy of catalytic and noncatalytic sites with ligands, but that the total number of sites occupied with nucleotides and PPI never exceeds 6 sites/mol of MF,. From these results, they suggested that the binding sites for PPi are unique and interact with both catalytic and noncatalytic sites. The observation reported here that prior incubation of nd-MF, with PPi in the presence of Mg2f mimicks prior incubation of the enzyme with ATP or MgATP in that the intermediate phase disappears, suggests that PPi binds to noncatalytic sites. However, the observation that PPI accelerates inactivation of the enzyme by FSBA contradicts this suggestion, unless simultaneous binding of PPI and FSBA to noncatalytic sites accelerates derivatization.
It is interesting that the Given these observations, we propose that the apparent negative cooperativity observed when the initial rate of the intermediate phase is plotted according to the method of Lineweaver and Burk is caused by slow binding of ATP to noncatalytic sites of nd-MF1. At sufficiently low substrate concentrations, very little ATP is bound to noncatalytic sites during the interval of the initial rate measurement. However, a t higher concentrations of ATP in the assay medium, the noncatalytic sites begin to saturate within the time frame of the rate measurement leading to stimulation of the rate of ATP hydrolysis. Therefore, the apparent K,,, of 440 y~ observed on Lineweaver-Burk plots for hydrolysis of ATP by nd-MF1 does not reflect binding of ATP to catalytic sites, but rather represents binding of ATP to noncatalytic sites which accelerates catalysis, presumably by promoting release of inhibitory MgADP from a catalytic site. Supporting this contention is the observation that Lineweaver-Burk plots are linear when ATP hydrolysis is catalyzed by nd-MF1 after loading noncatalytic sites with ATP. Lineweaver-Burk plots are also linear after treating nd-MF1 with PPi plus M$+, which, according to Kalashnikova et al. (33), promotes dissociation of inhibitory MgADP from a catalytic site. The K,,, values of 108 and 120 y M observed with enzyme activated with ATP and M$+ plus PPi, respectively, are equivalent to the lower K , of 120 PM observed for the hydrolysis of ATP by nd-MF1 in the absence of pretreatment.
The proposal that slow saturation of noncatalytic sites with ATP promotes dissociation of inhibitory MgADP from the affected catalytic site during hydrolysis of low concentrations of substrate by nd-MF1 is consistent with the results of a study reported by Milgrom and Murataliev (41) on interaction of MgADP with nd-MF1. After loading a catalytic site of nd-MF1 with Mg [14C]ADP, addition of ATP promoted dissociation of the complex with an apparent pseudo-bimolecular constant of 1.1 x IO2 M-' s-'. This is very close to the apparent pseudo-bimolecular rate constant of 1.6 X 10' M" s-l determined in this study from the dependence of the rate of activation of the intermediate phase during catalysis on ATP concentration. Table I1 compares the two K , and associated V, , , values determined in this study with the multiple K,,, and V, , , values previously reported by others for the hydrolysis of ATP by In addition, the concentration of MgSO, was 1 mM in excess over the concentration of ATP.  Table I1 for hydrolysis of 3-2000 p~ ATP by MF, containing endogenous nucleotides which was stored in buffer containing 10 mM sulfate and 2 mM EDTA. After removing sulfate and EDTA by dialysis, the same enzyme preparation exhibited two K,,, values of 10 and 145 p~. Since the time intervals of the rate measurements in the studies of others summarized in Table I1 were not reported, it is not possible to comment further on the differences or similarities among the K , values.
The multiple K , and associated Vmax values determined for MF, have been variously interpreted in terms of the number of catalytic sites participating in catalysis at given concentrations of ATP. For instance, Cross et al. (15) have suggested that hydrolysis of MgATP at a single site of MF1 occurs with such high affinity and slow product release that it cannot be detected by steady-state measurements. They also proposed that the K, values of 30 and 150 p~ determined from steadystate measurements represent catalysis with two (bisite) and three (trisite) catalytic sites filled with substrate. An opposing interpretation was put forward by Harris (42), who observed that MF, exhibits a single K, of about 1 pM in the presence of 1 mM azide, which is equivalent to the low K, value reported by Gresser et al. (12), Wong et al. (16), and Roveri and Calcaterra (17). Harris (42) suggested that the K,,, of about 1 p~ represents ATP hydrolysis a t a single catalytic site and that all three catalytic sites are operative during ATP hydrolysis in the physiological range. The results of this study are not easily accommodated by either interpretation. As described in detail earlier, it is our view that the K , of 150 p~ and associated V,,, of 600 s-l determined by Cross et al. (15), represent augmentation of ATP hydrolysis when ATP binds to noncatalytic sites.
If the VmaxZ value of 726 s" reported in this study represents cooperative hydrolysis of ATP with three catalytic sites operating maximally, then the Vmaxl value of 284 s" should represent ATP hydrolysis with two catalytic sites operating cooperatively and the third occupied with inhibitory MgADP. If this is the case, then the K, value of about 1 p~ should represent bisite catalysis.
It has been suggested that prior dissociation of product Pi from a catalytic site is responsible for the transition in rate that is observed when the initial burst phase decelerates to the intermediate phase (32,35). If this is indeed the case, then the binding of ATP to noncatalytic sites must be necessary to ensure optimal dissociation of MgADP from catalytic sites, thus preventing entrapment of inhibitory MgADP at a catalytic site. Vasilyeva et al. (33) reported that MF1 containing endogenous nucleotides exhibits a burst phase which decelerates to a slower phase equivalent to the intermediate phase described in this study. Given that MFl is usually isolated with 2 mol of adenine nucleotides bound to noncatalytic sites and exhibits apparent negative cooperativity (9, 43), it appears that binding of ATP to the third noncatalytic site to be filled promotes dissociation of inhibitory MgADP from a catalytic site in this case as well.