Active/inactive state transitions of the chloroplast F1 ATPase are induced by a slow binding and release of Mg2+. Relationship to catalysis and control of F1 ATPases.

Mg2+ is known to be a potent inhibitor of F1 ATPases from various sources. Such inhibition requires the presence of a tightly bound ADP at a catalytic site. Results with the spinach chloroplast F1 ATPase (CF1) show that the time delays of up to 1 min or more in the induction or the relief of the inhibition are best explained by a slow binding and slow release of Mg2+ rather than by slow enzyme conformational changes. CF1 is known to have multiple Mg2+ binding sites with Kd values in the micromolar range. The inhibitory Mg2+ and ADP can bind independently to CF1. When Mg2+ and ATP are added to the uninhibited enzyme, a relatively fast rate of hydrolysis attained soon after the addition is followed by a much slower steady-state rate. The inhibited steady-state rate results from a slowly attained equilibrium of binding of medium Mg2+. The Kd for the binding of the inhibitory Mg2+ is in the range of 1-8 microM, in the presence or absence of added ATP, as based on the extent of rate inhibition induced by Mg2+. Assessments from 18O exchange experiments show that the binding of Mg2+ is accompanied by a relatively rapid change to an enzyme form that is incapable of hydrolyzing MgATP. When ATP is added to the Mg2+- and ADP-inhibited enzyme, the resulting reactivation can be explained by MgATP binding to an alternate catalytic site which results in a displacement of the tightly bound ADP after a slow release of Mg2+. Both an increase in temperature (to 50 degrees C) and the presence of activating anions such as bicarbonate or sulfite reduce the extent of the Mg2+ inhibition markedly. The activating anions may bind to CF1 in place of Pi near the ADP. Whether the inhibitory Mg2+ binds at catalytic or noncatalytic nucleotide binding sites or at another location is not known. The Mg2(+)- and ADP-induced inhibition appears to be a general property of F1 ATPases, which show considerable differences in affinity for ADP, Mg2+, and Pi. These differences may reflect physiological control functions.

the membrane-bound F1 ATPase' from various sources (l-20). However, the factors underlying the Mg2+-induced inhibition and its relationship to the ADP-induced inhibition have not been elucidated clearly. Evidence suggests that both medium Mg2+ and an ADP bound at a catalytic site without accompanying Pi (16,18) are required for inhibition (7,9,11,16,17,20). Thus, it seems likely that only one type of inhibitory effect is involved. The inhibition of MgATP hydrolysis or the reactivation of the inhibited enzyme is unusual in that under conditions used frequently, the inhibition or reactivation requires many seconds or even some minutes to develop fully. For example, with ATP-heat-activated CF1 an initial rapid phase of MgATP hydrolysis during the first 30 s may be followed by a decrease of activity to a considerably slower but constant steady-state rate (20). When CF1 is exposed to both Mg2' and ADP or only to M$+, if a tightly bound ADP is present at a catalytic site, an inhibited enzyme form is obtained (11,18). Upon addition of ATP the enzyme is slowly reactivated, and a maximum steady-state rate is attained in about 2 min.
The transitions between inhibited and active states have generally been regarded as resulting from slow conformational changes following binding of Mg2+ and ADP. However, any conformational changes accompanying binding could be relatively rapid, and the slow transitions could result from a slow binding and release of Mg'+. To distinguish between these two possibilities and to obtain a better understanding of the role of Mg", we have explored several parameters of the M$'-induced inhibition, including experiments on its time and its Mg2' concentration dependence.
Other experiments using la0 exchange were designed to determine whether the Mg*+-and ADP-induced inhibition could be best explained by a progressive increase in inhibition of all enzyme molecules as the M$+ concentration is increased, or if the inhibition, as suggested (2,15,18), results from the interconversion of active and essentially inactive enzyme forms. We also report another approach for ascertaining whether tightly bound ADP at a catalytic site (17,18,20) is necessary for the Mg2+-induced inhibition and report on the effects of activating anions and temperature in relieving the inhibition. Overall, the results presented add some important new insights and confirm and extend some previous suggestions for the nature of the inhibition induced by both ADP and Mg*+.  ( earlier (3,37), is followed by a burst of activity and a transition to a slower steady-state rate (20). At this concentration of free Mg*+ the burst lasts about 60 s before a linear inhibited rate is achieved. As the free M$+ concentration is increased further the extent and the rate of the initial burst, the steadystate rate, and the time before the onset of the linear rate are all decreased. At the highest concentrations of free M$+ tested (450 PM), the burst is barely apparent, and a strongly inhibited steady-state rate is obrerved. If the free M$+ concentration is decreased below 10 pM by increasing further the total ATP concentration (>12 mM), activity is inhibited because of competition of free ATP for binding at the catalytic sites (3).
After the steady-state rate is reached, ATP hydrolysis continues linearly for many min in the presence of a pyruvate kinase trap to prevent the accumulation of ADP, which at higher concentrations (>70 pM) causes a weak competitive inhibition (38). The slower rates observed at the higher M$+ concentrations do not involve nonspecific electrostatic effects since CaATP hydrolysis does not exhibit a similar transition to an inhibited steady-state rate, and the slower steady-state rates do not result when the free Ca*+ concentration is increased (20). The strong activity dependence on the free Mg" concentration is consistent with a binding of Mg'+ to an inhibitory site on CFi. The final linear rate thus would reflect the attaining of an equilibrium between enzyme with bound Mg2+ at the inhibitory site and that without bound Mg'+. The slow transition to the inhibited rate at low Mg2+ concentrations suggests that M$+ binding might be relatively slow, a possibility supported further by other data in this paper. If the initial high rate of activity is assumed to correspond to 100% active CF1, then the steady-state rates can give an approximation of the percent of inactive enzyme at equilibrium in the presence of the different concentrations of free Mg'+. Estimates of Kd values for the inhibitory Mg" binding when the Mg2+ concentration was increased as reported in Fig. 1 were 1.1, 2.0, 3.6, 1.8, 5.6, 7.3, and 6.3 PM, respectively, corresponding to an average of 4.0 f 2.5 pM. The Kd, k,,, and Iz,fr rate constants were also derived from the data of Fig. 1 and other similar unreported data using a derivative-free nonlinear regression statistical software package as described under "Experimental Procedures" by fitting the observed data with the data predicted by a model in which the inhibition might result from one inhibitory Mg" binding to CF1. Analysis of the rate of approach to a linear steady state resulted in an estimate for the Kd of 8.4 -C 6.0 pM and K,, and k,ff rate constants of 4.5 f 3 X lo-" M-' s-' and 0.03 + 0.02 s-l, respectively.
Only One Catalytic Pathway Is Found with an Increased Inhibition of the Steady-state Rate by Increasing Mg" Concentrations-The effect of increasing the M$+ concentration could result from a progressive change in the reaction rate constants governing the catalytic pathway or from a conversion of active enzyme with one catalytic pathway to an inactive enzyme form. For the former, a progressive shift in characteristics of the reaction pathway would result. For the latter, only one catalytic pathway should be operative at all Mg2+ concentrations.
Measurement of the distribution of [180]P, isotopomers formed during the hydrolysis of [y-"O]ATP provides a sensitive probe of the reaction pathway for F, ATPases. When ATP is hydrolyzed, the bound Pi formed at the catalytic site can exchange oxygens with water before its release. The probability of the exchange occurring prior to release is expressed as the partition coefficient, P,. As the ATP concentration is lowered, the P, value increases from near 0 (little reversal of bound ATP hydrolysis prior to product release) to close to 1 (many reversals of bound ATP hydrolysis prior to product release). Statistical considerations are then used to find if the distribution of the ['"O]Pi isotopomers can be accounted for by the operation of a single catalytic pathway or whether more than one pathway is required (35,36).
To assess the effects of Mg2+ on P, values, the MgATP concentration was maintained constant at 10 pM and the free M$+ concentration increased until a nearly complete inhibition resulted. This MgATP concentration was chosen because it is in the range in which changes in P, values are readily detected. Ionic strength and pH of the incubation mixture were similar to those used to calculate the association constants for most of the complexes. The incubation mixture included at pH 8 50 mM glycylglycine, 100 mM KCl, and a pyruvate kinase-regenerating system. Fig. 2 compares the degree of inhibition of the steady-state rate and the changes in P, values as the free Mg2+ concentration is increased. The measurements were somewhat difficult to perform since at the low concentration of substrate used, P, values are very sensitive to MgATP concentration. In most "0 experiments, the Mg2+ concentration is in excess to maintain the MgATP concentration constant. However, in these experiments both Mg2+ and ATP concentrations were varied to allow experiments to be performed in the presence of both excess and very low values of Mg'+. What is noteworthy is that in ail trials only one major P, value was observed for each condition tested. The average P, value obtained was 0.56.

The M$+ Inhibition
Can Be Induced and Reversed in the Absence of Medium ATP or ADP-To determine whether MgATP hydrolysis might be a requirement to observe the inhibition, as has been noted for the action of inhibitory proteins from various F1 ATPases (39), the following experiments were performed. ATP-heat-activated CFI that contains an ADP at one of the catalytic sites and adenine nucleotides at two of the three noncatalytic sites (40) Fig. 3 shows the effect on ATP hydrolysis of prior exposure to 250 FM M$' for different periods of time. A control without the prior exposure to Mg2+ shows the usual kinetics when assayed with 5 mM ATP and 2 mM Mg2+, which provides a low free M$ concentration (20 PM). A 1-min exposure to 250 pM Mg" is enough to induce a pronounced lag that recovers slowly (about 1 min) before the same steady-state rate is observed as for the control. An increase in the time of exposure to Mg2+ to 5 min prior to assay increased the duration of the lag. Additional exposure for another 5 min gave no further increase. Thereafter, it appears that at 250 pM Mg2+ the binding site has been essentially saturated within 5 min. The results show that Mg" does not require either the presence of medium ATP or ADP or the occurrence of catalysis to be able to bind to ATP-heat-activated CF1 and to effect the changes required for the transition to the inactive enzyme. CF1 (1 mg/ml) was exposed to different concentrations of MF for different lengths of time. After the incubation period, assays were performed as described in Fig. 3. The rate of MgATP hydrolysis between 0 and 20 s from the start of the assay is compared with the rate of MgATP hydrolysis between 0 and 20 s of the enzyme that has not been exposed to Mg2'. This percent activity is plotted against the length of the Mg2+ incubation period. ysis during the burst phase as compared with a control without prior exposure to M$+. The activity during the burst phase was approximated from the slope of the time course between 0 and 20 s of catalysis in the presence of 2 mM MgATP. Between 0 and 20 s most of the M$'-inhibited enzyme has not had sufficient time to recover any significant activity, and the control has not yet reached its slower equilibrium rate. Therefore, the observed 0-20-s rate gives an approximation of the amount of inhibited enzyme present at the time of MgATP addition. Other experiments were performed to determine the Mg2+ concentration dependence of the inhibition after a lo-min exposure to Mg2+ (data not shown). The inhibition pattern as the Mg2+ concentration is increased shows a typical hyperbolic dependence.
One important conclusion from the data of Fig. 4 is that upon exposure to 250 PM Mg2+ the decrease in activity is rapid, resulting in more than a 90% inhibition at the first assay point (30 s). With 1 mM M$+ no activity was detected after 30 s of exposure. Increasing the Mg2+ concentration thus speeds up the transition to the inhibited enzyme. These results show that the M$'-induced inhibition does not result from stabilization of a slowly formed inactive enzyme conformation. The binding of Mg2+ causes or allows the transition to an inactive enzyme form within a few seconds or less. Relatively rapid conformational changes could accompany or follow the Mg2+ binding. A second important point from the data in Fig. 4 is that as the Mg2+ concentration is lowered, the Mg2+ binding appears to be approaching an equilibrium with time. The results may be explained by a relatively slow reversible binding of Mg*+ with a Kd of less than 10 pM.
To assess further the equilibrium of M$+ binding, an experiment was conducted to find if the same final equilibrium position was approached from both the fully active and the fully inactive enzyme forms. Fully active CF, and fully inactive CF1 were allowed to reach an equilibrium between active and inactive forms in the presence of 5 pM M$+. For the data presented in Fig. 5 either CF1 (30 rg/ml) was exposed to 5 pM M$+ for various time intervals prior to 2 mM MgATP hydrolysis, or CFi (300 pg/ml) was exposed to 50 pM M$+ for 20 min. This allowed nearly complete inactivation of the enzyme. The sample was then diluted to a final Mg2+ concentration of 5 pM and a final enzyme concentration of 30 pg/ ml. The time course of inactivation in the former case and the time course of reactivation in the latter case were monitored as described previously.
In both instances close to the same final equilibrium between active and inactive enzyme forms (ratio 60/40) was achieved, indicating a Kd of about 3 FM.
An Additional Assessment of the Requirement of Enzymebound ADP for the M$ '-induced Inhibition-The hydrolysis of MgGTP does not show the pronounced M$'-induced inhibition as does MgATP hydrolysis (20). Therefore, it was of interest to compare the degree of the Me-induced inhibition observed upon exposure of Mg2+ to either ADP-heat-activated CF, or GDP-heat-activated CFi. Heat activation with ADP results in a tightly bound ADP at a catalytic site and an ADP incorporated into one of the three noncatalytic sites. Heat activation in the presence of GDP prevents the ADP incorporation and likely results in the incorporation of a GDP into a catalytic site (40). The inhibition observed at two different Mg2+ concentrations (25 FM and 1 mM) was compared with either ADP-heat-activated CF, or GDP-heat-activated CFI. As shown in Fig. 6, when the tight binding of ADP at a catalytic site is prevented, the extent of M$+-induced inhibition is reduced drastically.
If 2 FM ADP is added to GDPheat-activated CF1 at any time during the Mg2+ incubation, the same degree of activity inhibition is attained as for the ADP-heat-activated enzyme by the time of the first time point (10 s). Therefore, ADP binding occurs quickly (<lo s) in comparison with the binding of Mg*+, which takes about 15 min to reach equilibrium at a concentration of 5 pM. The ADP binding to CFI could be inducing a conformational change that in itself does not inhibit enzyme activity but may be necessary to allow the transition to the inhibited form induced by Mg2+. Also the results show that M$+ is still able to bind to the enzyme in the absence of the bound ADP at a catalytic site. An inhibition is observed, however, only after the ADP has also bound to the enzyme. The Presence of Bicarbonate or an Increase in the Temperature of the Incubation Decreases the Inhibition of the Steadystate Rate-Either an increase in temperature (to 50 "C) or the presence of bicarbonate (60 mM) reduces drastically the Me-induced inhibition of the steady-state rate. Both temperature and bicarbonate exert their maximal activation when FIG. 5. Mg*+ binding to CFI is reversible and at 5 pM reaches an equilibrium within 15 min. CF, (30 rg/ml) was exposed to 5 pM Mg'+ for different lengths of time prior to activity assay (0). CF, (300 pg/ml) was exposed to 50 pM Mg*+ for 10 min prior to a lo-fold dilution to 30 rg/ml CF, and 5 pM Mg'+. The time course of reactivation was followed after the dilution step (W). Assays and the method to determine the percent activity remaining after the different Mg*+ incubation conditions were performed as in Fig. 4.

MgZ+ Inhibition of CF, ATPase
CFI is hydrolyzing ATP in the presence of excess Me. That bicarbonate increases the specific activity of CF, hydrolyzing MgATP in the presence of excess Mg2+ was first reported by Nelson et al. (41). Fig. 7 shows the effect of increasing the temperature under conditions in which there is a 1 mM excess of either Mg2+ or ATP, and the substrate concentration is maintained within the range of 30-40 pM MgATP. At higher temperatures the M$+-induced inhibition is reduced drastically. The fractional inhibition caused by 1 mM excess Mge decreases from approximately 80% at room temperature to 5% at 52 "C.
For the experiments of Fig. 7, the estimated MgATP concentration was 30 pM at room temperature. The concentration of MgATP increased slightly as the temperature was increased. A 1.4-fold increase in the K, for MgATP formation results from the increase in temperature from 25 to 37 "C (42). Even if the K, values doubled in going from 25 to 50 "C, the change in MgATP concentration would only be from about 30 pM to about 37 pM.

Other Consequences of Bicarbonate
Actiuation-A possibility that warranted further appraisal was whether bicarbonate decreased the affinity of ADP for its inhibitory site. To determine whether ADP binding to the enzyme might be affected by bicarbonate, the degree of inhibition of MgGTP hydrolysis caused by the addition of 2 pM ADP was compared in the presence or absence of bicarbonate. Addition of 2 pM ADP during MgGTP hydrolysis results in a 50% inhibition of the steady-state rate. The extent of the inhibition was nearly the same (5% less) in the presence of 60 mM bicarbonate (data not shown). Thus, the activation caused by bicarbonate does not result from a decreased binding of ADP to CFi during MgGTP hydrolysis. Similarly, when CFi was exposed to 50 PM Mp in the presence or absence of 60 mM bicarbonate prior to dilution and testing of MgATP hydrolysis, no change was observed in either the time course or the final degree of the inhibition (data not shown). Therefore, M$+ binding and release and any resultant enzyme conformational changes that occur in the absence of catalysis are not affected by the presence of bicarbonate.
In other experiments the effect of Pi on the bicarbonate activation was assessed (Fig. 8) in the presence of 3 mM Mg2+ and 3 mM ATP to have a high degree of the Mp-induced inhibition, and activity was monitored by the disappearance of NADH using a coupling assay (29). As the Pi concentration is increased from 0 to 80 mM in the absence of 60 mM bicarbonate, a small degree of activation (1.5 X) of MgATP hydrolysis is observed. On the other hand, the extent of bicarbonate activation decreases by about 3-fold as the Pi concentration is increased. This is consistent with Pi binding to the same site as the bicarbonate. However, the binding of Pi is much less effective than the binding of bicarbonate for activation of MgATP hydrolysis. Experiments performed with 30 mM sulfite, another activating anion, exhibited a similar inhibition of the sulfite activation in the presence of Pi. Bicarbonate and other activating anions do not appear to bind to the site to which MgATP binds since increasing the MgATP concentration from 0.5 to 3 mM while maintaining the free Mg2+ constant (1 mM) did not change the extent of anion activation or the inhibition by Pi of the bicarbonate activation (data not shown).

The Time Delays for Inhibition and Reactivation of CF, MgATP Hydrolysis
Can Be Explained by the Slow Binding and Slow Release of Inhibitory Mp-The results show that the time course of the inhibition of MgATP hydrolysis observed with ATP-heat-activated CFi is strongly dependent on the free Mg2+ concentration. The free Mg2+ concentration affects both the length of the initial burst of activity and the degree of inhibition attained during the subsequent linear steady-state rate. The M$+ causes inhibition by binding to an enzyme with a tightly bound ADP at a catalytic site. The Mg2+ binding promotes a transition to an enzyme form that is incapable of hydrolyzing MgATP. The slow transitions from an active to an inactive enzyme form when MgATP hydrolysis is initiated or from an inactive to an active enzyme form when the enzyme is exposed to Mg2+ prior to catalysis is due to the slow binding or slow release of Mg2* instead of slow conformational changes associated with a fast binding and release of M$'. The steady-state rate achieved at a given concentration of free Mp reflects the balance between Mg2 binding and release and the level of tightly bound ADP at a catalytic site. Once M$+ dissociates, the binding of MgATP i&f' Inhibition of CF, ATPase at an alternate catalytic site will promote ADP release.
The results as given in Figs. 1, 4, and 5 clearly suffice to show that the combination of the inhibitory M$+ with CF, is unusually slow. They suffice to give only a reasonable approximation of the values for Kd, and k,, and k,,, rate constants, principally because it is difficult to obtain accurate measurement of the uninhibited velocity. This is taken as the maximum velocity observed during the burst phase and is not a precise value. If the rate uninhibited by Mg2+ were higher than our estimated maximum, the values of Kd would be lower. The various estimates of Kd reported under "Results" fall in the range of l-8 pM. Low free Mg2+ concentrations thus suffice for maximal inhibition.
The data do provide convincing evidence that the binding of the inhibitory Mg2 is unusually slow (4.5 X 10e3 M-' s-l). In contrast, based on the rapid inhibition by 2 pM ADP of the GTP-heat-activated enzyme already exposed to Mg2+ as shown in Fig. 5, the k,, for ADP is of the order of 1 x lo6 M-' s-' or more than 200 times more rapid than the rate of Mg2+ binding. The slow M$+ binding could result if the Mg2+ must cross some type of barrier, e.g. conformational or charge, to reach the binding site.
In a previous investigation, The differences in Kd may result from differences in the preparation of the enzyme, such as loading of the noncatalytic sites, or may reflect differences in enzyme conformation of the static enzyme and an enzyme undergoing either MgATP or CaATP hydrolysis.
Only Fully Active or Essentially Inactive Enzyme Is Present as the Mp Concentration Is Increased-Previous research led to the suggestion that the inhibition by Mg2' resulted from the conversion of an active to an inactive enzyme state (2, 7). Our results using the "0 exchange methodology validate this previous suggestion. Hydrolysis of a low concentration of [r-"O]ATP shows the existence of only one catalytically active enzyme form, with kinetic characteristics that result in a single P, value of 0.56 at free Mg2+ concentrations varying from below 1 pM to 1 mM. No gradual change in the P, value was observed as Mg2+ concentrations were increased, ruling out the possibility that the Mg2+ affects all CF1 molecules similarly at a given Mg*+ concentration. Such equal effect on all CF, molecules could result if the association and dissociation of Mg2+ to one or more sites to cause inhibition were more rapid than the time for catalytic turnover.
The finding of only one P, value means that the Mg-and ADPinhibited enzyme did not have any detectable catalytic activity.
Additional Evaluation of the Requirement for Both Tightly Bound ADP and Mg2+ to Observe the Inhibition-As shown in Figs. 3 and 4, ATP-or ADP-heat-activated CF1 needs to be exposed only to M$+ in the absence of added medium ADP or ATP to induce the inhibition.
Heat activation with adenine nucleotides results in the tight binding of an ADP into a catalytic site (40). Prior exposure to Mg2+ has been noted to inhibit MgATP hydrolysis by rat liver mitochondrial particles (2) and by MFI (6). In these studies the endogenous nucleotide content was not determined, and the stringent requirement for a tightly bound ADP at a catalytic site was not reported. Other experiments with either MF1 or CF1 have detected a strict requirement for a tightly bound ADP to induce the inhibition of MgATP hydrolysis (15,17,20). Similarly, light activation of the membrane-bound CF1 ATPase has been correlated with the release of a tightly bound ADP (43,44). The deactivation that occurs once the thylakoid membranes are placed in the dark is accelerated by the presence of medium ADP (45,46) and is prevented if medium ADP is removed (1, 47). The presence of medium Mg2+ is necessary for the deactivation process (47,48). The results given in Fig. 6 confirm the mutual requirement for both Mg2+ and ADP to induce the inhibition and show the strong M$+ concentration dependence on the time required to induce the inhibition.
GDP-heat-activated CFI does not contain the tightly bound ADP at the catalytic site (40). GDP may be more loosely bound to the site or if bound, may be unable to induce the conformational changes required for Mg2+ to induce the inhibitory enzyme conformation. The presence of GDP in the medium decreases the Mg2+ inhibition of the light-activated membrane-bound CF, ATPase (46). The low level of inhibition during MgGTP hydrolysis may result with the catalytic site empty or loaded with GDP or may be due to ADP contamination of the GDP or of the enzyme. The rapid inhibition by ADP of the GDP-heat-activated enzyme that has been exposed to Mg2+ confirms that Mg2+ binds to the enzyme lacking the tightly bound ADP but is not capable by itself of inducing the change to the inactive enzyme form. Only when ADP is bound at a catalytic site does the transition to the inhibited state occur.
Location of the Inhibitory M$+ Binding Site-Where the inhibitory Mg2+ binds is not known, and because up to six M$+ binding sites have been reported for CF1 (49) the location is di!?icult to assess. Early studies to characterize the cation binding sites were performed by following Mn*+ binding to CF, by EPR (50, 51). Mn'+, like Mg"', induces the inhibition of ATP hydrolysis and has an even higher affinity than M$+. The EPR studies indicated the presence of one tight Mn*' binding site and five loose cation binding sites in the absence of exogenous nucleotide. In the presence of adenine nucleotide the cation binding pattern shifted to two tight cation binding sites and four loose binding sites. However, later studies with CF, from the same laboratory (49, 52) showed three cation binding sites with positive cooperativity and with a high affinity for M$+ plus three noninteracting cation binding sites with a lower affinity.
The three high affinity Mg2+ binding sites were suggested (52) to have higher k,, and krr rates than those reported in our studies. The slower binding and release of the inhibitory Mg2+ may not have been detected.
Differences observed in the cation binding characteristics have been suggested to reflect differences in the method of preparation and storage of the enzyme (49). Our enzyme preparation was more like that used in the earlier cation binding studies (50), particularly in that the enzyme was stored as an ammonium sulfate precipitate instead of being stored in glycerol at -80 "C (49).
All sites with high affinity for added Mn'+, and thus likely for added Mg2+, can also bind nucleotides (50,53). Reports from  favor a catalytic site location for high affinity Mg2+ binding sites, a view supported by Haddy et al. (53). However, the possibility that the inhibitory Mg2' binds to a noncatalytic site needs consideration. Noncatalytic site nucleotides on F1 ATPases are not readily replaced during ATP hydrolysis (54). One noncatalytic site on CR binds ATP tightly only in the presence of Mg2+ (40), but tilling this site does not inhibit catalysis. The function of noncatalytic binding sites is obscure although filling of the sites with ATP promotes GTP hydrolysis (40), and the sites may function similarly in ATP hydrolysis.
If the inhibitory M$+ binds to a catalytic site then the question arises as to whether the inhibitory Mg2+ is bound at M$+ Inhibition of CF, ATPase the same site as the inhibitory ADP. A possibility raised by Haddy et al. (53) is that with ADP already present at one catalytic site, the inhibitory M$+ binds at a second or third catalytic site in preference to MgATP binding and thus inhibits catalysis. This possibility is eliminated by the characteristics of the Mg2+ inhibition reported here. The Kd for the Mg2+ is sufficiently high and the time of M$+ release sufficiently rapid so that over minutes of exposure to excess MgATP the inhibition should be nearly completely reversed. This is not the case. The enzyme remains strongly inhibited when the MgATP to Mg*+ ratio is well over 1,000/l. The inhibitory Me, if it binds at a catalytic site, must combine to a catalytic site to which MgATP cannot bind, namely the site already occupied by ADP.
Combination of the inhibitory Mg2+ and ADP at the same site would mean that binding of only one Mg*+/enzyme is necessary for inhibition.
As mentioned briefly under "Results," the inhibition induced by Mg2+ shows a typical hyperbolic dependence, which is in harmony with the need for tight binding of only one Me/enzyme to induce inhibition. The evaluation of the Kd estimated from the data of Fig. 1 gave similar values over the wide range of free Mg2+ concentrations tested if binding of a single Mg2+ was considered necessary for the inhibition but markedly increasing values of Kd if the combination of two Mg2+/enzyme were required for the inhibition. Such results strongly favor a requirement of binding of only one Mg'+/enzyme for inhibition. Although only one bound M$+/enzyme may induce the inhibition, from previous Mg2+ binding studies mentioned above (49-52) it appears likely that other Mg2+ binding sites are also filled in the l-10 pM range.
On the Nature of the Mg2+-induced Inhibition-A reasonable possibility is that the inhibitory M$+ and ADP bind at the same catalytic site. If so, hydrolysis of MgATP should provide bound ADP, Mg", and Pi at the potential inhibitory site. An explanation is then needed as to why the onset of the inhibition is slow and why the extent of the inhibition depends on the medium Mg2+ concentration if bound ADP and Mg2 could already be present. A hypothesis that might account for these and other related observations is given in the following paragraphs.
The inhibited linear steady-state rate is much lower in comparison with the initial burst of activity (Fig. 1). Thus, during steady-state catalysis nearly all of the enzyme could have inhibitory ADP and M2+ at a catalytic site. When tightly bound ADP, Pi, and M$+ arise at the site, most of the time Pi could dissociate before ADP does, accompanied or preceded by dissociation of the Mg2' that was originally bound to ATP. After this event medium Mg*+ could rebind to enzyme with ADP remaining at a catalytic site in the absence of Pi. It would be this enzyme form that cannot hydrolyze MgATP. These suggestions are in accord with observations from others that Pi binding under certain conditions is able to prevent the Mg2+-induced inhibition (7,18,19). As long as Pi is tightly bound, the M$+ inhibition is reduced considerably. Binding of MgATP at a second catalytic site promotes the release or looser binding of Pi and probably Mp", as is evidenced by the ability of relatively low concentrations of MgATP to inhibit sharply the reversals of bound ATP formation at the catalytic site (23).
That the dissociation of inhibitory Mg2+ is necessary for subsequent promotion of the release of the tightly bound ADP by ATP for reactivation is supported by the data presented in Figs. 3 and 5. The time required to reach a steady-state rate of MgATP hydrolysis, after exposure of CF, to 250 PMfree Mg" for 10 min, as shown in Fig. 3, is only about 2-fold less than the time that is required to dissociate M$+ from CFi in the dilution experiment of Fig. 5. The relatively small time difference could result from a slight promotion of Mg2+ release by the binding of MgATP at a second site under the experimental conditions of Fig. 5. We suggest that when F, ATPases have ADP tightly bound at a catalytic site in the absence of Pi or an activating anion and have the inhibitory Mg2+ bound, the catalytic cooperativity essential for catalysis is blocked. Because an increase in the ATP concentration with an excess of medium M$+ present will not prevent or overcome the inhibition, it is probable that the binding of MgATP to additional catalytic sites is not prevented but that the conformational changes that cause release of ADP from the tight site are prevented or rendered ineffective.
Without bound Mg2+ present, or if activating anions are bound, the requisite changes occur more readily. This interpretation is in accord with the observation that at low substrate concentrations at which cooperativity does not occur, bicarbonate loses its ability to activate (55). Location of the Activating Anion Binding Site-Our data support the suggestions (19, 56) that activating anions and inhibitory anions combine in place of Pi at the site with inhibitory ADP present. This explains the ability of Pi to decrease the activation by bicarbonate (Fig. 8) and sulfite. P, protects MFi against azide inhibition (56). Earlier, Moyle and Mitchell (2) showed antagonistic effects of azide and activating anions with MF,. Neither the degree of bicarbonate activation nor the inhibition of bicarbonate activation which results from Pi addition is affected by MgATP concentration at a constant free Mg2+ concentration.
Therefore, the activating anions or Pi is not competing for MgATP binding sites.
The concentration of Pi required to inhibit activation by other anions is considerably less than that found necessary by Feldman and Sigman (57) for the formation of bound ATP from the tightly bound ADP at a catalytic site of CF,. We suggest that the release of tightly bound Pi formed by ATP hydrolysis is accompanied by a conformational change such that Pi can only bind loosely. Tight Pi binding competent for bound ATP synthesis by the isolated CF1 may be induced only by very high concentrations of Pi. How the activating anions promote reactivation is puzzling. We noted that bicarbonate did not prevent the inhibition induced by exposure to Mg2+ prior to the onset of MgATP hydrolysis. Reports in the literature on the effects of activating anions on binding affinities of MgATP, ADP, and Mg'+ are contradictory and include as examples both an increase (50) and a decrease in the affinity for M$+ (5), both an increase (2) and a decrease (58) in the affinity for ADP, and a decrease in the affinity for MgATP (59). No close interactions have been detected between the activating anion selenite and Mn2+ by x-ray absorption fine structure analysis (60), thus the activating anions may not be interacting directly with the bound M$+. CF1 that has been exposed to Mg2' to induce the inhibited enzyme form is much more rapidly reactivated after MgATP addition in the presence of bicarbonate. Thus, the anions either make the bound Mg2+ ineffective in blocking ADP release or promote the bound Mg*+ release that occurs when MgATP binds at another catalytic site. Similarly, we cannot offer a satisfactory explanation for the prominent decrease in the M$+-induced inhibition as the temperature is decreased. The participation of a conformational change involving multiple interactions could have a large temperature coefficient, but this is not diagnostic.
Differences in the Mg2+ and ADP Inhibition Observed for F1 ATPases from Various Sources-The Mg'+ and ADP inhibition appears to be a general phenomenon for all F1-type