Properties of ATP Tightly Bound to Catalytic Sites of Chloroplast ATP Synthase*

Under steady state photophosphorylating conditions, each ATP synthase complex from spinach thylakoids contains, at a catalytic site, about one tightly bound ATP molecule that is rapidly labeled from medium "'Pi. The level of this bound ["'PIATP is markedly reduced upon de-energization of the spinach thylakoids. The reduction is biphasic, a rapid phase in which the ["'PI ATPIsynthase complex drops about 2-fold within 10 s, followed by a slow phase, kobs = O.Ol/min. A decrease in the concentration of medium 32Pi to well below its apparent K , for photophosphorylation is required to decrease the amount of tightly bound ATPIsynthase found just after de-energization and before the rapid phase of bound ATP disappearance. The [32P]ATP that remains bound after the rapid phase appears to be mostly at a catalytic site as demonstrated by a contin- ued exchange of the oxygens of the bound ATP with water oxygens. This bound [32P]ATP does not ex- change with medium Pi and is not removed by the presence of unlabeled ATP. The levels of tightly bound ADP and ATP arising from medium ADP were measured by a novel method based on use of [/3-32P]ADP. After photophosphoryla- tion and within minutes after the rapid phase of bound ATP loss, the measured ratio of bound ADP to ATP was about 1.4 and the sum of bound ADP plus ATP was about llsynthase. This ratio is smaller than that found about 1 h after de-energization. Hence, while ATP bound at catalytic sites disappears, bound ADP appears. The results suggest that during and

Under steady state photophosphorylating conditions, each ATP synthase complex from spinach thylakoids contains, at a catalytic site, about one tightly bound ATP molecule that is rapidly labeled from medium "'Pi. The level of this bound ["'PIATP is markedly reduced upon de-energization of the spinach thylakoids. The reduction is biphasic, a rapid phase in which the ["'PI ATPIsynthase complex drops about 2-fold within 10 s, followed by a slow phase, kobs = O.Ol/min. A decrease in the concentration of medium 32Pi to well below its apparent K , for photophosphorylation is required to decrease the amount of tightly bound ATPIsynthase found just after de-energization and before the rapid phase of bound ATP disappearance. The [32P]ATP that remains bound after the rapid phase appears to be mostly at a catalytic site as demonstrated by a continued exchange of the oxygens of the bound ATP with water oxygens. This bound [32P]ATP does not exchange with medium Pi and is not removed by the presence of unlabeled ATP.
The levels of tightly bound ADP and ATP arising from medium ADP were measured by a novel method based on use of [/3-32P]ADP. After photophosphorylation and within minutes after the rapid phase of bound ATP loss, the measured ratio of bound ADP to ATP was about 1.4 and the sum of bound ADP plus ATP was about llsynthase. This ratio is smaller than that found about 1 h after de-energization. Hence, while ATP bound at catalytic sites disappears, bound ADP appears. The results suggest that during and after deenergization the bound ATP disappears from the catalytic site by hydrolysis to bound ADP and Pi with subsequent preferential release of Pi.
These and related observations can be accommodated by the binding change mechanism for ATP synthase with participation of alternating catalytic sites and are consistent with a deactivated state arising from occupancy of one catalytic site on the synthase complex by an inhibitory ADP without presence of Pi.
Tightly bound nucleotides associated with ATP synthases have been studied with chloroplast, mitochondrial, and bacterial membranes in various laboratories (see reviews, Refs. [1][2][3][4][5][6]. Present information is regarded by us as demonstrating * This work was supported by Department of Energy Contract DE-AT0370102. 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. j Present address, Plant Growth Laboratory, Wickson Hall, University of California at Davis Davis, CA 95616. To whom correspondence should be sent at Molecular Biology Institute, University of California, Los Angeles, CA 90024. presence of tightly bound ATP and ADP on both noncatalytic and catalytic sites of de-energized membranes and isolated ATPases. Such unusual retention of substrates at catalytic sites may be explained by alternating site participation in the binding change mechanism (7,8). In this mechanism, the ADP and P, that are loosely bound at a catalytic site are converted, in a step requiring an energy-driven conformational change, to tightly bound ADP and Pi capable of rapidly and reversibly forming bound ATP. In the same conformational transition, a transitorily tightly bound ATP at an alternate catalytic site is converted to loosely bound ATP.
The reverse of these binding changes occurs during ATP hydrolysis. When de-energization terminates net synthesis, some nucleotides appear to be retained at catalytic sites in tightly bound form.
Considerable clarification is needed as to the relationships between bound nucleotides found on isolated membranes and ATPases and the ATP and ADP transitorily tightly bound to catalytic sites on ATP synthase on energized membranes. In the studies reported here, we investigated the fate of the ATP, ADP, and Pi bound at catalytic sites when the protonmotive force is collapsed by de-energization. The experiments were designed to address some important questions about the nature of nucleotides bound at catalytic sites of the chloroplast ATP synthase. Does the ATP that remains on the catalytic site continue to undergo rapid, reversible hydrolysis? Could some ATP migrate from catalytic to noncatalytic sites without mixing with medium ATP, as suggested by Aflalo and Shavit (9) and Kozlov and Skulachev (lo)? During net photophosphorylation, do catalytic sites have a tightly bound ADP as well as ATP rapidly labeled from medium ADP present at catalytic sites, and, if so, how much tightly bound ADP is present? How much does the :"PI concentration need to be lowered before the amount of bound ATP rapidly labeled from medium Pi is appreciably reduced?
Closely related to the above questions is the possibility of control of ATPase and ATP synthase activities by bound nucleotides. Several laboratories have presented evidence that ADP binding can cause loss of ATPase activity of isolated ATPases from mitochondria (11)(12)(13), chloroplasts (14)(15)(16)(17), and bacteria (18). The ATPase synthase of de-energized membranes has been regarded as in a deactivated state that may correlate with a lack of ATPase activity (19). Our results also shed light on whether ADP binding without Pi at catalytic sites may control ATPase activity (16,20,21) and on the nature of catalytic site deactivation that accompanies deenergization.
ADP, and ATP were purchased from Sigma. Chloroplast thylakoids were prepared from market spinach as described by Vinkler et al. (22) with the modification of Rosen et al. (7). Chlorophyll concentration was measured by the method of Arnon (23), and ATPase concentration (1.3 nmol of CFl/mg of chlorophyll) was estimated as previously described (2). @-"P]ADP was prepared from 0.1 mCi of [ Y -~~P I A T P , incubated with 10 nmol of unlabeled ATP, 100 nmol of AMP, 5 pmol of MgClZ, and 10 pmol of Na/Tricine at pH 8, with 100 units of adenylate kinase for 1 min (final volume, 1 mlj. [/3-32P]ADP was separated from ATP and AMP on a column (22).
Time Course of P2P]ATP Disappearance-The standard assay mixture at pH 8 in 1 ml total volume contained 10 mM MgCl,, 25 mM NaC1, 20-2000 pM "PI (50,000 cpm/nmol), 100 pM ADP, 50 mM Na/ Tricine, 25 mM glucose, 33 pM phenazine methosulfate, 200-500 units of hexokinase, and chloroplast thylakoid membranes equivalent to about 50 fig of chlorophyll. These conditions gave typical steady state ATP synthesis rates of about 1200 pmol/h/mg chlorophyll. The mixture was illuminated for 20 s using a 300-watt slide projector; the duration of the illumination was controlled by an electronic shutter (22). When the illumination was terminated, 1 ml of 150 mM NH,CI in 50 mM Tris-C1, pH 8, (NN,Cl/Tris) was added. At various times during the course of the reaction, bound nucleotides were extracted from the thylakoid membranes by the addit.ion of 2 ml of 1 M perchloric acid containing 1-2 pmol of unlabeled ATP. [32P]ATP was isolated by passage through a charcoal column followed by a Dowex 1 column as in Ref. 22 except in experiments where more than IO6 cpm of "Pi were used per time point. In those cases, the perchloric acid quench contained 5 pmol of unlabeled Pi in addition to the carrier ATP. Also, the charcoal column was washed with 20 ml of a solution containing 0.025 M PPi, 0.1 M H3P04, and 0.3 M perchloric acid (22) after adsorption of ATP, and the Dowex column was washed with 25 ml of 60 mM HCl after ATP adsorption. In all experiments, [,'"P]ATP was quantitated by liquid scintillation counting in water using Cerenkov radiation.
Under the assay conditions, the total [32P]ATP present will consist mostly of bound ATP, but a low steady state level of medium [3'P] ATP would be expected. To assess the amount of medium [32P]ATP, repeated estimates were made of the amount of bound ATP rapidly formed from 32P, during illumination a t increasing concentrations of hexokinase, with perchloric acid quench after a few seconds in the "'P-labeled, bound ATP (7), showed that 85-90% of the [32P]ATP light. Extrapolation to infinite hexokinase, which gives a measure of present was bound. Bound ATP levels reported are corrected for this portion of free ATP. ATP e HzO Exchange Experiments-In the oxygen exchange experiments, assay conditions similar to those described above were used except that 2.2 mM P, or [IRO]Pi (including 32Pi a t 7500 cpm/ pmol), 200 p~ ADP, about 7000 units of hexokinase, and chloroplast thylakoid membranes equivalent to 50 mg of chlorophyll were used in a final volume of 480 ml. A 96-ml aliquot of the assay mixture with 10 mg of chlorophyll was placed in a wide beaker (21.5 cm) and was illuminated through a CuS04 filter for 20 s with two 650-watt Lentar photo lamps. This procedure was repeated 5 times. After illumination, 3.75 ml of 2 M NH,Cl in 50 mM Tris-CI, pH 8, were added to each aliquot, the solution incubated for 1 min, and 50 pmol of ATP/aliquot were added (2 mmol of EDTA were also added to the sample with ["O]P,j. The mixture was incubated for 5 min more and then centrifuged in 250-ml bottles at 10,000 rpm for 15 min. The pellets were homogenized in 5 ml of solution (50 mM Napricine, 10 mM MgC12, 25 mM glucose, 75 mM NH4C1, and 10 units of hexokinase/ml at pH The experiments with ["O]Pi were similar except the water of solution A contained 86% "0. The thylakoids were allowed to react 10 to 15 min with H"0H in solution A. Then, 0.35 ml of nearly 100% trichloroacetic acid was added and the mixture centrifuged a t 10,000 rpm for 5 min. The supernatant (pale yellow) was collected and neutralized with solid Tris base. The H"0H was recovered by lyophilization transfer and the residue brought to 200 ml, pH 8.5. The ATP was isolated with use of a 3-ml Dowex 1 column as described above.
The recovered ATP was neutralized with 50% NaOH, desalted over Sephadex G-10, and treated with glycerokinase to form Pi from the y-phosphoryl group of ATP (24,25). The Pi was derivatized to triethylphosphate with diazoethane for mass spectral analysis (26). In the experiment with ["O]P,, the following changes were made. The thylakoid pellet was suspended and incubated in solution A for 5 min, 5 ml of 1 M perchloric acid were added, and the mixture was centrifuged and the supernatant neutralized with 0.6 M KOH. ATP was eluted from the Dowex 1 column with 2 M triethylamine carbon-ate, pH 7.6, and rotoevaporated before treatment with glycerokinase.
Measurement of Tightly Bound Nucleotides under Various Conditions-The levels of ADP and ATP rapidly labeled from medium ADP and remaining on quenched thylakoid membranes were determined with [p-32P]ADP under the following assay conditions: 10 mM MgC12, 25 mM NaCI, 2 mM Pi, 5 pM [/3-"2]ADP (IO4 cpm/nmol), 50 mM Na/Tricine, 25 mM glucose, 33 WM phenazine methosulfate, about 100 units of hexokinase, and chloroplast thylakoid membranes equivalent to 50 pg/ml of chlorophyll, p H 8.0 (final volume, 1.5 ml). In experiment 1, chloroplasts in a 0.75-ml aliquot of the assay mixture without the radiolabel (solution 1 j were mixed with a 0.75-ml aliquot of the assay mixture with [/3-32P]ADP hut without hexokinase (solution 2), and illuminated for 23 s. In experiments 2 and 3, chloroplasts were illuminated in a 0.75-ml aliquot of solution 1 for 20 s, then combined with a 0.75-ml aliquot of solution 2 and illuminated for 3 s more. Experiment 3 was as experiment 2 but contained no Pi. When the illumination was terminated, the following were added in the order indicated 1.5 ml of the 2 M NH4C1/Tris solution, 0.01 ml of a solution containing 50 mM each of ADP and ATP, and 100 units of adenylate kinase. The mixture was incubated for 1 min to allow the enzyme-coupled reactions to remove nearly all medium ADP and ATP. The thylakoid membranes were filtered as in Ref. 7 except that they were washed with the NH4C1/Tris solution instead of EDTA, and both unlabeled ADP and ATP were present in the perchloric acid solution.
[32P]ADP and [3ZP]ATP were separated by anion exchange chromatography as described above and counted.
To check the catalytic competence of the [32P]ATP remaining bound to the synthase after de-energization and a period in the dark, photophosphorylation was reinitiated with addition of unlabeled ATP as a cold chase. Tightly bound [32P]ATP was generated in the light in 0.5 ml of the standard assay mixture with 100 JLM 32Pi, 400 units of hexokinase, and 40 pg of chlorophyll. The thylakoids were quenched with 20 pl of the 2 M NH4CI/Tris solution and incubated in the dark for 10 s. They were then diluted 10-fold with a similar assay mixture except with 500 units of hexokinase and without the "P radiolabel, mixed for 1 s, and reilluminated. When the light was extinguished, 190 pl of the 2 M NH4C1/Tris solution were added and samples incubated for 1 min. The nucleotides were extracted by adding 225 pl of concentrated perchloric acid. In one experiment, 50 min after the first NH4C1 quench, the thylakoids were centrifuged and washed a t 4 "C, recentrifuged at 7000 rpm, and the pellet resuspended in fresh wash solution, which contained 0.4 M sucrose, 5 mM MgCI2, 10 mM Na/Tricine, and 0.2 M choline, pH 8. Part of the suspension was reserved for the determination of the chlorophyll concentration and the remainder was illuminated a t room temperature in an unlabeled assay mixture for various times and quenched in 1 ml of the NHrC1-Tris solution. After 1 min, the bound nucleotides were extracted by adding 2 ml of 1 M perchloric acid, and the ["PI ATP was isolated using the procedure outlined above.

RESULTS
Time Course of Disappearance of p2P]ATP Tightly Bound to the Chloroplast ATP Synthase Complex upon De-energization-Rosen et al. (7) have shown that during net photophosphorylation with substrates near saturating concentrations, the ATP synthase complex contains approximately one tightly bound ATP molecule/enzyme complex that is rapidly labeled by "Pi and inaccessible to cleavage by hexokinase. They presented evidence that for short periods following "Pi addition to thylakoids under steady state photophosphorylating conditions the hexokinase-inaccessible [32P]ATP is mostly bound to catalytic sites. Furthermore, the authors showed that the amount of this tightly bound ATP falls from about 1 to about 0.5jsynthase when illumination is terminated and the thylakoids are quenched with the uncoupler, NH4CI.
To investigate further the process of the disappearance of the bound ["'PIATP upon de-energization, a time course of this event was followed. As shown in Fig. 1, the decay of labeled ATP is biphasic. Under the experimental conditions used, the initial level of bound ["PJATP during steady state photophosphorylation was about 0.58jATP synthase, and when the light was extinguished and NH4CI added, the level fell to about 0.27 within 10 s. This rapid drop was followed by a much slower decrease of koha = O.Ol/min. This biphasic by guest on March 24, 2020 http://www.jbc.org/ Downloaded from decay shows that two processes are occurring when the ATP synthase complex is de-energized. Aspects of both phases of bound ["PIATP decay are given in this report.
Similar results were obtained in the absence of NH,Cl, using only precautions to obtain essentially complete darkness for thylakoid de-energization. Thus, the behavior noted is not dependent upon presence of NH4C1.
Effect of Pi on Levels of Rapidly Labeled [32P]ATP on Energized and Deenergized Thylakoids-The results of Fig. 1 as compared to those of Rosen et al. (7) suggest that when thylakoids are illuminated a t subsaturating levels of Pi, the amount of bound [32P]ATP remaining in the dark after an NH&l quench is less than that remaining in the dark after an illumination at saturating levels of Pi. To characterize this phenomenon more carefully, the amount of bound ["'PIATP rapidly labeled from Pi/synthase complex was determined at different levels of Pi in the light and in the dark. With 50 pM or more Pi, about 1 ATP/synthase was rapidly labeled. A decrease of Pi to 2 p~, far below that required for halfmaximal velocity of net ATP formation, was required to drop the bound [32P]ATP level during steady state phosphorylation to about 0.3/synthase. However, the amount of ["PIATP remaining after de-energization decreased more than one-half by decreasing the medium Pi from 2 mM to 50 pM. Thus, results show that the level of bound ["PIATP remaining in the dark is more sensitive to the concentration of medium Pi than is the steady state level of bound ["PIATP rapidly labeled from '*Pi during photophosphorylation. Nonetheless, the biphasicity of the ATP decay occurs at all concentrations of Pi tested.
T o characterize better the effect of Pi concentration on the level of hexokinase-inaccessible ATP during photophosphorylation, measurements were made while medium P, was consumed. Results are shown in Fig. 2 Fig. 2, may allow an approximation of the rate of release of bound ATP from the catalytic site in the light under these conditions. In the binding change mechanism (7,8), catalysis is regarded as involving participation of alternating sites and catalytic cooperativity of product release. Normally, ATP at one catalytic site is released only after substrates bind to an alternate catalytic site. At concentrations of P, far below that required for half-maximal velocity, the turnover rate decreases to the point that ATP is being released from the enzyme before Pi binds to the alternate site. This results in a decrease in the concentration of bound ["PIATP. The rate of ["'PIATP decrease calculated from the slope of Fig. 2 at 10 p~ Pi may be used to give a rough estimate of the off constant for the release of tightly bound ATP from one catalytic site with only ADP but not P, bound at an alternate catalytic site. The off constant for this tightly bound ATP, calculated from the rate at 10 p~ Pi is 6.8/s, about 40-fold lower than the potential maximum turnover rate (250/s) under the conditions used.
ATP H'O Exchange of the Bound ("'PJATP on Deenergized Thylakoids- Fig. l shows that de-energized thylakoids retain a tightly bound ATP that was rapidly labeled with "'Pi during brief illumination. Under the experimental conditions used, most of this bound ATP originated at the active site of the synthase and presumably remains at the catalytic site after de-energization. Therefore, the possibility existed that this ATP might be undergoing the rapid, reversible hydrolysis to ADP and Pi characteristic of ATP formation catalyzed by the light-activated synthase. Measurements were made to test this possibility. Table I    was allowed to proceed for about 40 min in the dark. At high substrate concentrations, considerably less than 1 water oxygen appears in each ATP molecule made during rapid net photophosphorylation (27,28). Thus, the "0 enrichment of the bound ATP formed in the light is expected to be not much less than the enrichment of t,he PI used in the experiment, and the marked decrease in the enrichment of "0 we observed in the ATP could only occur in the dark.
Hackney et al. (28) showed that under low light intensity, intermediate ATP H,O exchange still occurs during ATP synthesis. Hence, the possibility existed that the exchange observed in experiment 1, Table I, had occurred only while the light intensity was decreasing and the NH4Cl was mixing with the thylakoid reaction mixture. Experiments 2 and 3 were conducted to test this possibility. In these experiments, H'"0H was added to the reaction mixture several min after the light was extinguished. Again, a slow exchange rate was observed at a rate approximately equal to that observed in which P, was labeled in experiment 1.
An important point from the data of Table I is that extensive exchange was observed when "0-labeled ATP was present at the time of de-energization or water H"OH added several minutes after de-energization. This gives convincing evidence that most of the ATP, at the time of and for a number of minutes after de-energization, is bound at catalytic sites.
The exchange of water oxygens with oxygens of bound ATP is not as complete as would be expected if the exchange were as rapid as during net photophosphorylation and if all the bound ATP were at catalytic sites. From this and other consideration discussed later in this paper, it appears clear that part of the bound ["'PIATP remaining after de-energization is not at catalytic sites. Further, if the exchange was continuing at a uniform rate up until the time of assay, then the actual exchange rate is much slower in the de-energized state. Each of the single time experiments reported in Table  I required considerable experimental investment, and at this stage it did not appear worthwhile to attempt to firmly establish the exchange rate.
Lack of Exchange between Medium P, and Bound (3"P]ATP on De-energized Thylakoids-Because the ["PIATP tightly bound to the synthase on de-energized thylakoids under conditions of our experiments undergoes a reversible hydrolysis, as shown by the oxygen exchange, the question arises as to whether a required product of hydrolysis, bound ["PIP,, is exchangeable with medium Pi. To test if such an exchange might occur, unlabeled Pi in the millimolar range was added 1 min after illumination of thylakoids in 20 p~ 32Pi. Even after 1 h, no loss of "'P from the bound ["'PIATP was detectable. The experiment was repeated under conditions giving a higher initial bound [:"P]ATP concentration. Thylakoids were illuminated at higher initial Pi concentrations, either 100 or 500 p~ "'Pi, and the concentration of unlabeled Pi present after quenching was 500 p M or 5 mM, respectively. An apparent slow exchange reaction occurred at 100 p~ Pi (hobs = O.OOZ/min), and at a faster rate a t 500 p~ Pi (kohS = O.OOG/min). However, we found that, when "'Pi at 100 p~ or higher concentration was used, some >'P, is incorporated into a product that is not ATP but that coelutes with ATP during the usual isolation procedure.' This side reaction complicates the interpretation of the experiments in which thylakoids are illuminated at relatively high concentrations of "'Pi and apparent wash out of label is measured. This side reaction did not greatly interfere with our measurements of the rate of bound ["'PIATP disappearance because the product isolated from ATP synthase at 2 mM Pi only 1 min after the NH4Cl quench was more than 85% ATP as tested by hexokinase cleavage. Therefore, any reaction to give the contaminating "P species is too slow to affect appreciably the accuracy of the early time points of the ATP decay curve at high Pi concentrations. And at low P, concentrations, this contaminating reaction does not interfere at any time points as indicated by a lack of effect of the addition of unlabeled Pi to the medium containing 20 PM "'Pi.
To assess more readily if the tightly bound P, formed from hydrolysis of ['"PIATP bound to thylakoids might be exchangeable with medium Pi, an inverse experiment was carried out i.e. illumination in 100 p~ unlabeled Pi, followed by a quench and incubation in NH4C1 and 500 p~ "Pi for 1 h.
Hexokinase was unable to cleave any of the radioactive compound isolated in the ATP fraction from this procedure (data not shown), demonstrating that radioactive ATP is not formed under these conditions. We thus conclude that the labeled ATP remaining bound does not undergo exchange with medium Pi upon de-energization.
Tests of Catalytic Competence of Tightly Bound ATP Remaining on Deenergized Thylakoids-The results of Aflalo and Shavit (9) suggest that some tightly bound ATP can shift from a catalytic to a noncatalytic site without equilibrating with medium ATP during net photophosphorylation or during or shortly after de-energization of thylakoids. Most of the bound [."'P]ATP remaining shortly after deenergization is at a catalytic site in view of the exchange of the y-phosphoryl oxygens with water. As another means of testing for binding to a catalytic site and for the change in binding with time, the following experiments were conducted. Thylakoids were illuminated in an assay mixture containing 100 p M "'Pi, and subsequently incubated in NH4C1 in the dark for 10 S. Then the thylakoids were diluted in fresh assay mixture without the radiolabel and reilluminated. Table 11 shows that time course of the disappearance of ["2P]ATP that was bound to de-energized thylakoids in the dark. After 2 S, about 80% of the bound ATP disappeared, and after 10 s only about 3% of the counts were left. The disappearance of ["'PIATP did not Also, some time lag is anticipated in attaining sufficient protonmotive force for the steady state phosphorylation rates. However, within 2 s, almost all of the ["PIATP bound at catalytic sites should have disappeared even under these conditions. The results in Table  I1 thus show that about 20% of the bound ["PIATP isolated shortly after an NH4C1 quench in the dark was not at a catalytic site. Similar results were obtained when the illumination period prior to de-energization was decreased from 15 to 5 s, hence the result is not due to building up of ATP in the medium that can exchange with nucleotides at the noncatalytic site. This decrease of catalytic ATP was not due to loss of photophosphorylation capacity as the thylakoids retained their original capacity. Also, incomplete chase of the ["'PIATP was observed with thylakoids that were washed by centrifugation to remove residual ammonia. Thus, the continued presence of a lower concentration of NH4Cl was not the cause of the slow rate of ATP release.
Additional experiments under conditions as used for Table  I1 showed that a smaller fraction of the remaining ["PIATP can be chased out in 2 s when the thylakoids are incubated for several minutes in the dark with NH&l before the second illumination. Presumably, this is due to disappearance of ATP from catalytic sites while the amount of ATP bound to noncatalytic sites remains constant. Thus, the change of binding from catalytic to noncatalytic sites appears to occur either during illumination or shortly after de-energization, but not during extended incubations in the dark.
Labeling of Chloroplast Thylakoids with ADP and ATP Rapidly Derived from Medium ADP-In most experiments (22,29) in which bound nucleotides are exchanged with medium nucleotides, thylakoids have been labeled with radioactive nucleotides in the light and washed to remove medium nucleotides. Such a procedure can take up to 1 h, and it was desirable to have a technique to quantitate immediately after illumination the levels of both bound ADP and ATP derived rapidly from medium components. For this reason, ATP was synthesized by the synthase for short periods with [@-"2P] ADP and P, as substrates, the reaction was quenched in the dark with NH4C1, and the mixture quantitated for [@-32P] ATP and [b-:"P]ADP bound tightly to the synthase. Both adenylate kinase and hexokinase were present in the NH4C1 quench. The coupled reaction of these enzymes caused medium [@-"'PIADP to be rapidly and almost completely converted to unlabeled AMP and ["P]glucose 6-phosphate. This method obviated the difficulties arising from the presence of relatively high levels of labeled nucleotides remaining in the medium when ["]ADP or [I4C]ADP was used.
Results obtained are given in Table 111. In experiment 1, thylakoid ATP synthase was incubated with only 5 ~L M [b-'"PIADP and Pi in the light for 23 s and then quenched. Under these conditions, the ratio of bound ADP to bound ATP is about 1.2 and the total amount of nucleotide bound per synthase molecule is about 1.1. In experiment 2, thylakoids were illuminated with unlabeled substrates for 20 s and then pulsed with [@-:"P]ADP for 3 s before the NH4C1 quench. Under the assay conditions used, the ADP concentration is only 2or 3-fold below its K,, so the photophosphorylation rate is fairly rapid. The short labeling time insures preferential labeling of nucleotides at the catalytic site during steady state photophosphorylation, while minimizing labeling of noncatalytic sites (7). The decrease in levels of ATP and ADP/ synthase molecule in experiment 2 when compared to experiment 1 is as expected for some noncatalytic site labeling in experiment 1. Also, the ratio of ADP to ATP is about 1.4, slightly increased, compared to that in experiment 1, indicative that noncatalytic sites become labeled more rapidly with ATP than ADP. In similar experiments, but with 100 PM Pi in the assay mixture, Vinkler et al. (22) showed that after extensive washing of thylakoids the ADP to ATP ratio was about 3, although, as in the present experiments, they showed that the amount of ADP plus ATP was about l/synthase. These results indicate that the decrease in ATP correlates with an increase in ADP. Therefore, hydrolysis of bound ATP to bound ADP and Pi and loss of that Pi probably occurs.
Reproducibility of the values for bound ADP present was only fair in repeated experiments; results varied by as much as 220%. The data do suffice to show the important point that both tightly bound ADP and ATP are formed from medium ADP with only a few turnovers of the synthase, and to give a fair estimate of the amounts of those transitorily tightly bound nucleotides at the catalytic site.
It is of interest that a considerable amount of bound ADP and even some bound ATP were formed in the absence of added P, (experiment 3, Table 111). The tight binding of ADP may reflect a quite rapid energy-dependent exchange (see Refs. 1 and 30) of ADP at a catalytic site with medium ADP. Whether or not this is accompanied by an equally fast or faster exchange of bound P, with a very low level medium P, present is uncertain from present data. A possible explanation for the presence of ["'PIATP is that it resulted from the synthase reaction due to endogenous P,, or from a reaction with a contaminating adenylate kinase that would catalyze the conversion of [p-"P]ADP to [~J-~~PP]ATP and AMP. The doubly labeled ATP would then be transferred to a site on the synthase. The first possibility is more likely since no labeled ATP was formed in the absence of light.
Inability of Medium A T P to Remove the Bound r2P]ATP-Chloroplast thylakoids have been shown to lose readily the light-induced capacity for ATPase in the dark when they do not have ATP available to maintain a protonmotive force (see Refs. 14, 30, and 31). Experiments were conducted to determine whether the site containing the bound ["']ATP rapidly labeled from "' Pi during photophosphorylation may participate in a slow ATPase after de-energization. Less than 20% of the ['"PIATP disappeared when unlabeled ATP (final concentration 100 PM) was added with an NH,CI quench. Also, when thylakoids were washed with 75 mM NH,Cl, then mixed with 4 mM unlabeled ATP and 10 mM Mg", no disappearance of ['"PIATP was observed. The catalytic site with bound ["2P]ATP in the de-energized thylakoids appears to be indeed tightly closed to medium ATP.

DISCUSSION
Our results show that de-energization of chloroplast thylakoid membranes causes a biphasic decrease in amount of a tightly bound ATP that was at the active site of the ATP synthase during net ATP formation. The data in this report together with those of Vinkler et al. (22) give evidence that this decrease in bound ATP is accompanied by a corresponding increase in bound ADP. These findings and other data to be discussed below are consistent with a rapid hydrolysis of part of the bound, catalytic ATP upon de-energization with release to the medium of the Pi and probably some of the ADP formed. The ATP remaining on the enzyme after the rapid decay phase appears to be largely at catalytic sites shown by the continued exchange of the y-phosphoryl oxygens of most of the ATP with water oxygens. But the catalytic site is in a deactivated state as shown by lack of exchange with medium Pi or ATP.
These and other data pertinent to how the biphasic disappearance of bound, catalytic ATP may occur can be usefully discussed as related to the binding change mechanism for ATP synthesis ( 7 , 8). Using rapid mixing-chase experiments, Smith and Boyer (32) showed that during steady state photophosphorylation there is about 1 ADP/synthase at catalytic sites. In accord with the binding change mechanism of ATP synthesis, the total ADP may represent the sum of two species of ADP, one of which is tightly bound at the site where reversible ATP formation occurs and the other loosely bound at another catalytic site of the same synthase complex. Similarly, for bound ATP, Rosen et al. ( 7 ) demonstrated the presence of about 1 bound ATP/synthase at catalytic sites during photophosphorylation, and the total bound ATP is the sum of that at the site where reversible hydrolysis is occurring and that at sites where interconversion no longer occurs. The presence of bound reactants at more than one catalytic site/ synthase molecule during net ATP formation seems reasonable since there are 2 or 3 catalytic subunits/synthase complex (see Refs. 6, 30, and 33). An explanation for events occurring during and after de-energization needs to consider different species of bound ADP and bound ATP that may be present at the catalytic sites. Also, the relatively facile medium ATP + P, exchange that occurs in the light (see Ref. 30) and the transient ATPase activity observed when illumination ceases need to be considered.
A sequence of events consistent with the above observations and with results reported here is as follows. When illumination is terminated and NH4C1 is added, a transient ATPase activity occurs; net catalysis is now in the reverse direction. The sites containing loosely bound ATP that was just about to be released are changed to tight binding sites where a reversible hydrolysis of the bound ATP occurs. And the tight sites where ATP + ADP + Pi interconversion was occurring in the light are converted, in the dark, to sites where only loosely bound ADP and Pi can be present. The ADP and Pi would then dissociate to the medium. At most, only a few such turnovers would be expected to occur because of the removal of the substrate ATP from the medium by hexokinase. The net result would be that some of the catalytic sites that contained bound ADP and ATP during synthesis would now be empty. The sites retaining bound ATP and ADP on the de-energized enzyme complexes would be the tight binding sites where reversible hydrolysis occurs. This expected continued turnover of the enzyme upon de-energization could account for why Aflalo and Shavit (9) observed decreases in bound ["'PIATP formed from 32Pi when unlabeled ATP is added with uncoupler.
The initial release of bound ATP and bound ADP and Pi upon de-energization would be expected to be rapid because of the relatively high potential ATPase activity. Thus, the rapid phase of disappearance of bound catalytic ATP when illumination ceases and NH4C1 is added can be accounted for.
The slow phase would reflect the properties of the remaining tightly bound ATP, ADP, and Pi. From the earlier results of Harris and Slater (29) showing that membrane bound CF, labeled with ATP in the light retains bound ADP and not Pi from the ATP when isolated, and the data of Smith and Boyer (32) and Vinkler et al. (22), it is apparent that the de-energized enzyme preferentially loses Pi and retains bound ADP. This release of Pi is not rapid because a continued oxygen exchange is observed. The slow phase of bound ATP disappearance we observe can thus be accounted for by the continued reversible hydrolysis of some of the bound ATP, and a slow preferential loss of the Pi formed to the medium. Our observation that a higher level of medium Pi can increase the amount of ['"PIATP bound to the catalytic site found shortly after de-energization may reflect reversal of the Pi release step that participated in the ATP + Pi exchange while some protonmotive force still remains. The above model invokes only two catalytic sites/synthase participating in the binding change mechanism. Whether two or three sites are present on each synthase complex is at present uncertain, but the relations to data reported here would be essentially the same even if three sites were operative. The stable end product when de-energization causes net ATP synthesis to cease is the synthase complex containing a tightly bound ADP, but not Pi, a t a catalytic site that had been in the conformation which catalyzed rapid reversal hydrolysis during net photophosphorylation.
It needs emphasis that in this model the principal change from a quiescent state to the state active for ne! phosphorylation is one where the E~A T P + =$?' ' equlllbrlum is not shifted more than a factor of 10 and is still close to unity; rates of both hydrolysis and synthesis are markedly increased. Two possible reasons for this change in properties appear to us to warrant consideration. One is that the decreased catalytic rates observed reflect changes in properties of the site due to the lack of occupancy of another catalytic site by medium ATP. Another is that there is a change in the properties of the synthase complex associated with the drop in protonmotive force. In either case, the quiescent state is not representative of a conformation that is present as part of the usual catalytic cycle.
Our data are in accord with those of Aflalo and Shavit (9) suggesting that ["'PIATP formed from "Pi can shift rapidly or preferentially to noncatalytic sites, either in the light or during de-energization. This shift could occur through a space not accessible to hexokinase as suggested by Aflalo and Shavit. Such behavior would also harmonize with evidence from Kozlov's laboratory (10) for some preferential access of catalytic ATP to noncatalytic sites. Evidence suggests that noncatalytic sites on a-subunits are close to catalytic sites on p-subunits (34), and such proximity or even surface migration could favor noncatalytic site capture of ATP once released from a catalytic site. Noncatalytic sites may become emptied during continued incubation with hexokinase and glucose present. The on constant for the tight binding at noncatalytic sites may be several orders of magnitude greater than that for hexokinase binding of ATP, and, in addition, some binding by hexokinase may be reversible before formation and release of ADP. Thus, even though present in relatively large amounts, hexokinase might compete poorly with avid, noncatalytic sites for ATP binding on the synthase complex.
Our demonstration that most [3'P]ATP remaining tightly bound to the synthase after de-energization can undergo oxygen exchange gives strong evidence that, at least initially, such an ATP is bound to catalytic sites. Further evidence for this comes from the relatively rapid disappearance of the bound ["'PIATP present in the light as medium "Pi is depleted (Fig. 2). The presence of tightly bound ATP at catalytic sites was not supported in a recent report of Aflalo and Shavit (9). Their experiments and earlier results (35, 36) on characteristics of tightly bound nucleotide exchanges upon energization did not provide a critical assessment of participation of a transitorily tightly bound ATP as a catalytic intermediate. The considerable evidence for such ATP as a catalytic intermediate is summarized elsewhere (7). The view of Shoshan and Selman (35) from their studies that, " . . . energy requiring conformation changes might be involved in the release of tightly bound nucleotide as a step in the catalytic sequence," is quite in accord with the earlier and present data supporting the binding change mechanism. An appropriate interpretation is that on the ATP synthase two types of tightly bound ATP may be present a t relatively high levels, that at the catalytic and that at the noncatalytic sites. Under some conditions, labeling of noncatalytic sites can be quite rapid (this paper and Refs. 1, 7, 9, 15, and 35-37). Further, rapid exchange of noncatalytic tight nucleotides with medium nucleotides is feasible. Even if dissociation constants were of the order of lo-', rebinding by a diffusion controlled process could result in exchange within a second or less. Experimental distinction of whether a label is at a catalytic or noncatalytic site can be difficult.
The tightly bound ADP that remains bound to the deenergized synthase likely represents that observed by Schlodder and Witt (37) which is released when the synthase is activated by electric pulses and the ADP that Smith and Boyer (32) observed to be rapidly released upon acid-base transition of thylakoid membranes. Also, it likely is that same type of ADP that accounts for the formation of bound ATP from medium Pi and isolated CF, as observed by Feldman and Sigman (38). Several laboratories have demonstrated that bound ADP inhibits the ATPase activity of both isolated and membrane bound CF, (14)(15)(16)(17)(34)(35)(36)39) as well as F1 ATPases (11)(12)(13)18). This inhibition may result from a tightly bound ADP present a t a catalytic site without accompanying Pi (21). That the tightly bound ADP is at catalytic sites has been sugested by results with thylakoid membranes (32, 35) CF, ATPase (39), submitochondrial particles (40), and heart F, ATPase (41)(42)(43).
Brief comment may be helpful on interactions between ADP and P, binding. Dunham and Selman (44) have observed that Pi can inhibit tight binding of ADP by light-activated spinach thylakoids. Binding of Pi a t a catalytic site may decrease ability of ADP to bind at the same or another catalytic site; their binding is anticooperative. The binding of Pi and formation of bound ATP from bound ADP obtained by Feldman and Sigman (38) requires much higher Pi concentrations and lower pH than that normally used for photophosphorylation assays.
Our results are pertinent to control of ATPase activity.
Light activation of ATPase and of Pi + ATP and Pi + HOH exchange activities has sometimes been considered to result from a special conformational change induced by protonmotive force. This remains feasible, but present data favor an alternate possibility, namely that inhibition of ATPase and other activities is an expression of occupancy of one catalytic site by ADP alone. The decay of ATPase activity after light is extinguished would result from formation of tightly bound ADP and Pi at a catalytic site followed by preferential loss of Pi. Similarly, the well documented inhibition of ATPase by ADP binding mentioned earlier may result from ADP binding at an unoccupied site. The retention of ATPase activity after illumination ceases and even in presence of NH,Cl, could reflect coordinated departure of both ADP and Pi by alternating site participation. A lone, tight ADP on isolated CFI, not activated by heat or other treatment, or on membrane bound CF,, may be only slowly displaced following ATP binding at an alternate site. Weak ATPase activity by turnover of single, nonalternating sites is possible, but when one catalytic site retains both Pi and ADP and the other only ADP, turnover of the former needs to be extremely slow to account for data presented here.
In conclusion, in our model when the light is extinguished, chloroplast thylakoids are transformed into a deactivated state where reversible ATP hydrolysis still occurs but net ATP hydrolysis and interchange of bound ATP with medium Pi or ATP is nearly completely blocked. Eventually, all of the ATP is hydrolyzed to enzyme bound ADP and Pi. When all of the Pi is released to the medium, the enzyme is in an inactive state with ADP but no Pi a t tight binding catalytic sites.