Vanadate, a Transition State Inhibitor of Chloroplast CFl-ATPase*

The activity of CF1-ATPase was inhibited by vana- date in an allosteric manner with respect to CaATP as substrate. The cooperative interaction was enhanced by preincubation of the enzyme in the presence of ADP and Caa+ ions and of free divalent metal ions during assay of the activity. The strongest cooperative interaction with a Hill coefficient of 5.3 f 0.1 was found when the reaction was stopped after 30 s, before steady state was reached. Under these conditions, the concen- tration of an exchangeable ADP, tightly bound to one of the active sites on the enzyme, was shown to be the highest. A K, of 12.4 f 1.2 I.~M for vanadate inhibition was determined under these conditions. Direct meas- urements with the aid of ‘lV NMR indicated that vanadate binds to CFl in the presence of Ca2+ and ADP in a positive cooperative manner with a Hill coefficient of 2.3 f 0.2 and an average & of 0.3 f 0.04 nM. It was suggested that a formation of pentacovalent van- adyl-ADP at the active site caused the inhibition. Van-adyl-ADP was suggested to be a strong inhibitor, being an analogue of a pentacovalent phosphoryl-ADP, which is proposed to be the transition state intermedi- ate of CFl. The proton-translocating reversible ATPase contains sections: proteins situated the membrane and serving as a proton channel, to which coupling factor 1 (CF,)’ is attached The catalytic sector (CF,)

the active site included a high sequence homology of the , f 3 subunits from various organisms, a conserved region in the vicinity of the nucleotide binding sequence (3), and affinity labeling studies with ADP and ATP analogues (4). CF, was suggested to have a total of six nucleotide binding sites (4,5).
Three of the sites were assigned to the p subunits and three to the a at the interface of the /3 subunits. Based on specificity, capacity to synthesize (6) or to hydrolyze (7) ATP, and affinity labeling studies (8), the catalytic sites were suggested to be on the /3 subunits while noncatalytic sites were assigned to the a subunits. The mechanism of the ATP synthase is probably described best by the "alternating site" hypothesis (9). ATP is formed from tightly bound ADP and Pi or hydrolyzed from bound ATP, alternately, at the three catalytic sites on the F,. The release of the product is induced by a confor-*This work was supported by Grant 706/92 from The Israel Academy of Sciences. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The abbreviations used are: CFI, coupling factor 1 of chloroplast H*-ATPase; Tricine, N-tris(hydroxymethy1)methylglycine; MOPS, 3-(N-morpholino)propanesulfonic acid. mational change caused either by the electrochemical potential of protons across the membrane during synthesis or by cooperative interaction caused by binding of ATP to other sites during hydrolysis. The sequence of reactions is coordinated through cooperative interaction among the three active sites of the enzyme. Interactions among the subunits were exemplified by the ability of ADP, bound to a single site, to quench the activated state of CF1 (10) and to modulate ATPase activity in the isolated CFl (11). H2180-ATP exchange was also modulated by cooperative binding of ATP (9). Support for this hypothesis came from the finding of ATP hydrolysis with an equilibrium constant of approximately 1 (12) and a concomitant catalysis of H2180-ATP exchange which occurred when the nucleotides were tightly bound to a single site on the enzyme.
Catalysis in the F-type ATPase was shown to proceed through direct in-line trasfer of phosphoric residue between ADP and water and not through the formation of a covalently bound phosphate intermediate (13). However, other ion pumps such as the P-type Ca2'-ATPase, the Na+/K'-ATPase, or the plasma membrane H+-ATPase (14) form a covalently bound phosphate intermediate. In these enzymes, a pentacovalent phosphorus was assumed to be a transition state intermediate since they are readily inhibited by the Pi analogue vanadate (15). Myosin and dynein ATPases, as well as ribonuclease, phosphoglucomutase, alkaline phosphatase, aryl sulfatase, and phosphoglucomutase belong to another group of enzymes which are inhibited by a vanadyl adduct of substrate or product (16). The formation of a trigonal-bipyramidal coordination geometry about vanadate was shown by neutron diffraction of the crystalline uridine-vanadate-ribonuclease complex (17).
Direct measurements by x-ray absorption of the metal at the three cooperative interacting sites revealed the existence of the ternary complex of enzyme, Mn2+, and ATP at these sites in CFl (18,19). Based on the structure of the metal-ATP complex in the active sites and on the known chemistry of ATP hydrolysis in other types of ATPases, it might be expected that the catalysis in F-type ATPases also proceeded through a pentacovalent phosphate intermediate. The configuration of the oxygens about the phosphate indicates an inversion at the y-phosphate of ATP during hydrolysis. Such an inversion could indicate that the hydrolysis proceeded through a formation of pentacovalent intermediate. Vanadate is a weak competitive inhibitor of oxidative phosphrylation (20), but it failed to inhibit the activity of mitochondrial F-ATPase (21). In preliminary work we have shown (22) that vanadate inhibited the activity of CF1-ATPase with halfmaximal effect at 0.5 mM. In this work it was shown that in the presence of CaADP micromolar concentrations of vanadate strongly inhibited the activity. The results were interpreted to suggest the involvement of a pentacovalent phosphorus as a transition state intermediate in reactions catalyzed by CF1-ATPase.

EXPERIMENTAL PROCEDURES
ATPase Activity-CF, was isolated from chloroplasts prepared from lettuce (var. romaine) leaves as earlier described (23). Protein concentration was determined as in Ref. 24. Stored CF, was passed through a centrifuged Sephadex G-50 column equilibrated with 50 mM sodium Tricine, pH 8, and 1 mM EDTA. Latent CF1 was heatactivated at 64 "C in the presence of 5 mM dithiothreitol and 40 mM ATP as described (25). Following activation, the enzyme was passed through a centrifuged Sephadex G-50 column as above. ATPase activity was assayed in a medium containing 50 mM sodium Tricine, pH 8, 2-25 pg/ml heat-activated CF, and CaATP, and free Ca2+ as indicated. The reaction time was varied from 30 s to 10 min, as indicated. Dissociation constants of 0.17 and 1.51 mM for CaATP and CaADP (26), respectively, were used for calculations of the concentrations of the complexes and of free ATP, ADP, and Ca2+ ions. The released Pi was determined by the spectrophotometric method (27). 61 V NMR Measurements-NMR measurements were performed in a Brucker 41 M360 spectrometer, equipped with a multinuclear probe at 96.43 MHz. Pulse width of 50", sweep widths of 20 KHz, and acquisition times of 0.13 s, with a recycle delay of 0.05 s, were used. A line-broadening of 50 Hz was applied to all spectra before Fourier transforming to the frequency domains with the use of a 4K data set zero-field to 8K. For NMR measurements, stored CFl was passed through a centrifuged Sephadex G-50 column equilibrated with 50 mM Tris-C1, pH 8, as above. CF1 was activated with 50 mM dithiothreitol and 1 mM ATP at 25 "C for 1 h. This method is more efficient than heating when high concentrations of the enzyme are activated. Following activation, the enzyme was passed twice through a centrifugal column. The enzyme was concentrated by ultrafiltration to 20-25 MM. A sample of 2 ml of activated CF1 was titrated with a 10 mM solution of Na3V04, pH > 12 (Alfa Products). A standard curve of vanadate was measured in 50 mM Tris-C1, pH 8, and was used for binding analysis.
Calculation of Vanadate Composition-Aqueous solutions of vanadate anions contain several oligomers including monomers, dimers, and tetramers (28). 61V NMR measurements were used to determine the concentration of various oligomeres in 50 mM Tris-Mops, pH 6 to pH 9.5, at 0.1 to 2 mM vanadate (Figs. 4 and 6). Association constants of 0.07 mM" and 0.037 mM-3 were calculated for the formation of dimeric and tetrameric vanadate from monomeric vanadate at pH 8 as in Ref. 28. The composition of vanadate in the solutions calculated by these constants was in agreement with the experimental data obtained from 'lV NMR measurements (Fig. 4, inset).
Data Analysis-The kinetic parameters of ATPase activity of CFl were determined using the Dixon (Eq. l), Hill (Eq. 2), and Michaelis-Menten (Eq. 3) equations. Binding data were plotted according to the Scatchard (Eq. 4) equation. The binding constants for the interacting binding sites were analyzed by a semiempirical method as previously described (29). Data were fit to the equations by the "KaleidaGraph 2.1," a graphic software system by Synergy Software, Reading, PA. Theoretical binding and ATPase activity plots were also drawn by the "KaleidaGraph" graphic software.

RESULTS
Effect of Free Ca2+ Ion on Inhibition by Vanadate-In preliminary work we have shown (22) that vanadate, an analogue of phosphate, showed an allosteric inhibition of ATPase activity of CF,. Although half-maximal inhibition was obtained at 0.5 mM vanadate, direct binding measurements by 51V NMR spectroscopy indicated that in the presence of ADP and a divalent cation, vanadate binds to the enzyme at a much lower dissociation constant. It was also shown that the inhibition was enhanced in the presence of a free divalent cation at concentrations higher than 0.1 mM. Initially, the effect of free divalent cation on the inhibition by vanadate was explored. The activity was assayed in the presence of Ca2+ ions which are analogues of the physiological Mg2+ ions, because free Ca2+ ions are much less inhibitory than free Mg2+ (1). The less inhibitory ions were preferred in this study in order to minimize interference with the inhibition by vanadate. Yet ATPase activity was shown to be equally sensitive to vanadate in the presence of each of these cations preincubation of the enzyme in the presence of Ca2+ and with 0.5 mM free Ca2+ ions during the assay of the activity. In the absence of ADP or vanadate, the activity as a function of CaATP concentration gave a hyperbolic line (Fig. 2). However, both ADP and vanadate inhibited the activity in a sigmoidal manner with Hill coefficients of 1.6 f 0.2 and 2.1 f 0.1, respectively (Table I). Much stronger inhibition was obtained in the presence of both vanadate and ADP. The inhibition was allosteric with CaATP as substrate with a Hill coefficient of 5.3 f 0.1 (Table I), indicating strong subunit interaction. Although the effects of the two inhibitors on V,,, were additive, a strong synergistic effect was seen at low substrate concentrations. Heat-activated CF1 was preincubated at 37 "C for 5 min, and the reaction was started by the addition of ATP. ATPase activity was measured for 30 s as described under "Experimental Procedures.'' ATPase activity at CaATP concentrations of 0.04 to 0.8 mM was expanded (inset).

TABLE I The effect of the length of the reaction time of ATPase activity on the
Hill coefficient Data were analyzed using the Michaelis-Menten or Hill equations as described under "Experimental Procedures.'' The Hill coefficient for data fit to the Michaelis-Menten equation is equal to 1. CaATP concentration was varied from 0.03 mM to 4 mM with 1 mM free ca2+ ions. Both ADP and vanadate concentrations were 0.5 mM. Heatactivated CFl was preincubated at 37 "C for 5 min. The reaction was started by addition of ATP. V, , is given in micromoles X mg" X rnin". ATPase activity was measured for the time indicated as described under "Experimental Procedures. These results were interpreted to suggest that the enzyme sensitivity to inhibition by vanadate greatly increased by binding of divalent cations and ADP. Divalent ions and ADP were shown to increase the pre-steady state lag of ATPase activity (31). The possibility that the sigmoidal response was a result of this lag was therefore evaluated. The rate of presteady state activity was low and was accelerated on addition of CaATP at a rate constant of 0.05 s-l until steady state rate was reached when CFl was preincubated with divalent cations (31). The rate constant for the acceleration was also shown to increase as a function of CaATP concentration. The inhibition by ADP and divalent cations was therefore expected to be stronger at low substrate concentrations resulting in a sigmoidal dependence on substrate concentration. A more pronounced sigmoidity was also expected when the reaction was stopped before steady state was reached. The Hill coefficient values of ATPase activity measured in the presence of ADP, vanadate, or vanadate and ADP were 1.3 f 0.1, 1.0, and 1.5 f 0.1, respectively, when the reaction time was 10 min (Table I). However, the Hill coefficients measured for 30 s of reaction were 1.6 f 0.2, 2.1 f 0.1, and 5.3 f 0.1 for similar effector combinations, respectively. A pronounced increase in cooperative interaction was observed at pre-steady state activity.
The pre-steady state acceleration of ATPase activity was also shown to be slower following preincubation of the enzyme with divalent cations (31). Assuming that the sigmoidal response was due to the pre-steady lag in the activity, preincubation of the enzyme in the presence of divalent cations should also increase the sigmoidal effect of ADP and vanadate. Indeed, in the presence of ADP or vanadate or ADP and vanadate, Hill coefficient values of 1.7 f 0.2, 1.30 f 0.3, and 2.5 f 0.6 were obtained, respectively, when the enzyme was preincubated in the presence of Ca2+ ions and the reaction started with the addition of ATP, compared to Hill coefficient values of 1.1 f 0.2, 1, and 1.8 f 0.2, respectively obtained when the reaction was started with the addition of CF1 without preincubation ( Table 11). Heat-activated CF, was preincubated at 37 "C for 5 min when indicated. The reaction was started by the addition of ATP after preincubation or by CF1 without preincubation as indicated. V, , is given in micromoles X mg" X rnin". ATPase activity was measured for 30 s as described under "Experimental Procedures."  1 f 0.2 11.4 1.6 1.7 f 0.2 3.5 f 2.0  1.0 f 0.0 9.0 f 0.3 1.3 f 0.3 6.1 k 2.1   ADP and vanadate 1.8 f 0.2 8.2 f 0.7 2.5 f 0.6 4.3  The Dixon plot used for determination of the Ki gave a straight line (data not shown) indicating that CaADP is directly involved in the inhibition of the activity. Even very low concentrations of vanadate greatly enhanced the inhibition of the activity by CaADP (Fig. 3). However, since the solution of vanadate contains monovanadate and its polymers, it was important to verify the active species in the mixture. The concentrations of the various polymers of vanadate were determined by 61V NMR of the solutions used in the measurement of ATPase activity. Up to 1 mM total vanadate, there was an almost linear increase in monovanadate which is the major species in the solution. Significant concentrations of di-and tetravanadate were detected only above 0.5 mM total vanadate (Fig. 3, inset). Linear relations were obtained in a Dixon plot of the change in the K i for CaADP as a function of monovanadate concentration (Fig. 4) (Fig. 4). The nonlinearity indicates no direct relation between the change in the concentrations of the dimer and the tetramer and the inhibition of the activity.  Fig. 3. The composition of vanadate solutions from 0.003 rnM to 0.2 mM was determined and calculated by NMR measurments as described in Fig. 3 (inset).
inhibition was determined at increasing CaADP concentrations. There was a decrease in the K, for  The Effect of pH on the Inhibition of ATPase Actioity by Vanadate-The relative concentrations of vanadate and its various polymers depend, among other factors, on the concentration of total vanadate, on the ionic strength, and on the pH of the solution. As the pH was increased from 6 to 9.5, the relative concentration of di-and tetravanadate decreased from about 30% to less than 1% while monovanadate increased accordingly (Fig. 5, inset). The inhibition of ATPase activity by vanadate increased with the elevation of the pH in correlation with the increase in the relative concentration of the monvanadate (Fig. 5). An increase in inhibition with the increase of the one of the species of vanadate supports the suggestion that monovanadate rather than the polyvanadates was involved in the inhibition of ATPase activity. (see inset). The binding to the interacting sites was obtained following subtraction of the contribution of the noninteracting sites to the total binding as described under "Experimental Procedures." "V NMR spectroscopy. Presentation of the data according the Scatchard plot gave an isotherm which curved upward to a maximum and then descended nonlinearly (Fig. 6, inset). The isotherm was assumed to represent two types of binding sites: one type of interacting sites with a positive cooperativity which gave a downward concave curve and a second type of noninteracting sites which gave a descending isotherm. Using a semiempirical analysis (29), the contribution of noninteracting sites was subtracted from the total yielding a plot of the interacting site (Fig. 6). The data were fitted to three interacting sites with a Hill coefficient of 2.23 f 0.2 and an average Kd of 0.3 -t 0.03 nM. Additionally, approximately 30 weakly noninteracting binding sites having an aproximate Kd of 6 mM was calculated. There was a 5-fold increase in the Kd of binding of vanadate to activated CFl without the addition of Ca2+ and ADP. The fact that the Kd for binding was lower than the K, of inhibition supports the suggestion that the Ki was a result of the steady state concentration of the enzyme in transition state during the hydrolysis of ATP.

DISCUSSION
Transition State Conformation-Based on preliminary evidence, we have suggested (22) that an inhibition of CF1 by vanadate was caused by a formation of a pentacovalent vanadyl ADP adduct at the active site. Such a complex was suggested to be structurally similar to the transition state of ATP during hydrolysis and synthesis. A transition state analogue would be expected to bind to the enzyme more strongly than the substrate. Yet, vanadate was a weak inhibitor, having Ki similar to the K,,, for phosphate (22). The rather high K, could be understood as vanadate was shown to behave as a reversible rather than a dead end inhibitor. In such a case, it was reasonable to assume that the Ki might depend on the concentration of the enzyme which was in transition state conformation. In this state, nucleotides were shown to bind tightly to the active site of the enzyme (12). Some experimental evidence indicated that each of the three active sites alternately assumed this conformation during the catalytic cycle. Yet, before ATP was added to the activated CFl, one tightly bound exchangeable molecule of ADP was found on the enzyme. It was possible that this ADP was bound to the subunit which assumed a transition state conformation and therefore was tightly bound. Nucleotide binding site 1 (7), where the exchangeable ADP was bound, was shown to be located on the p subunit (8) and to catalyze hydrolysis of bound ATP (7) and synthesis of ATP from bound ADP (6,32). Therefore, this was suggested to be an active site of the enzyme. If this were the case, vanadate should bind most strongly to the exchangeable bound ADP. The inhibition of the ATPase activity by vanadate was therefore measured under experimental conditions known to increase the concentration of the bound ADP by slowing down its dissociation from the enzyme. Positive cooperative binding of divalent cations to three sites on the enzyme (30) slowed down the release of the exchangeable ADP caused by the onset of ATPase activity on addition of substrate (33,34). The release of ADP occurred concomitantly and at the same rate constant as the pre-steady state acceleration of ATPase activity (30). It was assumed that following preincubation of the enzyme with divalent cations and if the reaction was ended before steady state was reached, the concentration of the enzymebound ADP would be high and that the inhibition by vanadate would be the strongest. Since the rate of acceleration of the activity depended also on the concentration of the substrate it was expected that a sigmoidal activity curve will be obtained as a function of ATP Concentration. Indeed, in this work, a sigmoidal response with the Hill coefficient of 5.3 & 0.1 in the presence of vanadate and ADP was observed. The high value for Hill coefficient indicated strong cooperative interaction among vanadate binding sites on the enzyme.
Correhtion between Sigmoidity and the Concentration of the  site (37, 38).

The K, for Vanadate Inhibition-A marked decrease in the
Ki for vanadate inhibition was observed under conditions which favored the increase in the concentration of the enzyme-bound ADP. This was achieved under the experimental conditions which induced a high Hill coefficient with respect to CaATP. Thus, preincubation of the enzyme in the presence of divalent metal ions, assay for a short time in the presence of CaADP, low substrate concentration, and free Ca2+ ions gave the lowest Ki for vanadate inhibition. It is suggested that under such conditions the concentration of the transition state of the enzyme was increased and favored the formation of a pentacovalent vanadyl-ADP by the enzyme-bound ADP and vanadate. This analogue of the transition state intermediate had a low dissociation constant and was only slowly released from the enzyme on addition of substrate, causing a strong inhibition. Yet the Kd obtained from direct binding measurements of vanadate to three interacting sites on the enzyme was much lower than K , for inhibition, thus indicating that under steady state conditions only part of the enzyme molecules assumed transition state conformation. It can be argued further that the fact that vanadate, which tends to form a pentacovalent intermediate, was a most effective inhibitor when bound to the enzyme-bound ADP supports the suggestion that the ADP was bound tightly because the subunit assumed a transition state conformation.
Monovanadate Rather Than Di-or Tetravanadate Caused the Inhibition-Three modes of inhibition could be observed in enzymes which were inhibited by vanadate. Monovanadate inhibited by directly binding to the enzyme, by binding as an adduct substrate or a product, or by binding as one of the polymeric forms of vanadate (16). It was therefore important to verify which was the mode of vanadate inhibition in CFI. The use of 'lV NMR spectroscopy enabled the determination of the concentration of the monomeric and the various polymeric forms at a given experimental condition. Although almost complete inhibition of ATPase activity was observed at concentrations ranging from 0-0.1 mM total vanadate, there was a lag in the increase of the relative concentrations of diand tetravanadate. If divanadate or tetravanadate were the inhibitors, a lag in the inhibition as a function of the total vanadate concentration would be expected. Yet, no such lag was observed. The linearity of the Dixon plot of the change in the Ki for inhibition by CaADP as a function of monovanadate concentration compared to deviation from linearity in the case of diand tetravanadate also indicated that the former was the inhibitory species. Changes in the medium pH dramatically altered the relative concentrations of di-and tetravanadate, but showed no correlation between the extent of inhibition and the increase in the concentration of these forms of vanadate. There was, however, a good correlation between monovanadate concentration and the extent of inhibition of ATPase activity.