JBC

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Pedersen, P. L.
Right arrow Articles by Lebowitz, M. S.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Pedersen, P. L.
Right arrow Articles by Lebowitz, M. S.
Social Bookmarking
Add to CiteULike   Add to Complore   Add to Connotea   Add to Del.icio.us   Add to Digg   Add to Reddit   Add to Technorati  
What's this?

Volume 270, Number 4, Issue of January 27, 1995 pp. 1775-1784
©1995 by The American Society for Biochemistry and Molecular Biology, Inc.
Rat Liver ATP Synthase
RELATIONSHIP OF THE UNIQUE SUBSTRUCTURE OF THE F(1) MOIETY TO ITS NUCLEOTIDE BINDING PROPERTIES, ENZYMATIC STATES, AND CRYSTALLINE FORM (*)

(Received for publication, August 18, 1994; and in revised form, October 14, 1994)

Peter L. Pedersen (§), ,  Joanne Hullihen,  Mario Bianchet ,  L. Mario Amzel,  Michael S. Lebowitz

From the Laboratory for Molecular and Cellular Bioenergetics, the Department of Biological Chemistry, and the Department of Biophysics and Physical Chemistry, Johns Hopkins University School of Medicine, Baltimore, Maryland 21205-2185

ABSTRACT
INTRODUCTION
EXPERIMENTAL PROCEDURES
RESULTS
DISCUSSION
FOOTNOTES
REFERENCES

ABSTRACT

The F(1) moiety of rat liver ATP synthase has a molecular mass of 370,000, exhibits the unique substructure alpha(3)beta(3), and fully restores ATP synthesis to F(1)-depleted membranes. Here we provide new information about rat liver F(1) as it relates to the relationship of its unique substructure to its nucleotide binding properties, enzymatic states, and crystalline form.

Seven types of experiments were performed in a comprehensive study. First, the capacity of F(1) to bind [^3H]ADP, the substrate for ATP synthesis and [P]AMP-PNP (5`-adenylyl-beta,-imidodiphosphate), a nonhydrolyzable ATP analog, was quantified. Second, double-label experiments were performed to establish whether ADP and AMP-PNP bind to the same or different sites. Third, total nucleotide binding was assessed by the luciferin-luciferase assay. Fourth, F(1) was subfractionated into an alpha and a beta fraction, both of which were subjected to nucleotide binding assays. Fifth, the nucleotide binding capacity of F(1) was quantified after undergoing ATP hydrolysis. Sixth, the intensity of the fluorescence probe pyrene maleimide bound at alpha subunits was monitored before and after F(1) experienced ATP hydrolysis. Finally, the catalytic activity and nucleotide content of F(1) obtained from crystals being used in x-ray crystallographic studies was determined.

The picture of rat liver F(1) that emerges is one of an enzyme molecule that 1) loads nucleotide readily at five sites; 2) requires for catalysis both the alpha and the beta fractions; 3) directs the reversible binding of ATP and ADP to different regions of the enzyme's substructure; 4) induces inhibition of ATP hydrolysis only after ADP fills at least five sites; and 5) exists in several distinct forms, one an active, symmetrical form, obtained in the presence of ATP and high P(i) and on which an x-ray map at 3.6 Å has been reported (Bianchet, M., Ysern, X., Hullihen, J., Pedersen, P. L., and Amzel, L. M.(1991) J. Biol. Chem. 266, 21197-21201).

These results are discussed within the context of a multistate model for rat liver F(1) and also discussed relative to those reported for bovine heart F(1), which has been crystallized with inhibitors in an asymmetrical form and has a propensity for binding nucleotides more tightly.

INTRODUCTION

Mitochondrial ATP synthases are comprised of two major units, one called F(0) and the other F(1) (for recent reviews, see (1, 2, 3, 4, 5, 6) and 57). The F(0) moiety spans the mitochondrial inner membrane and directs protons to the F(1) moiety, which binds ADP and P(i) and synthesizes ATP. The F(1) moiety of ATP synthases from animal cells have molecular masses near 370 kDa and contain five different subunit types in the unique stoichiometric ratio alphabeta. The presence of a single copy of the small subunits (, , and ) for three alphabeta pairs may impose asymmetry on the F(1) molecule(2, 57) . Alternatively, subunit asymmetry may be induced within an otherwise highly symmetrical F(1) molecule primarily by nucleotide binding, particularly if the small subunits are normally positioned in the center of the molecule closely in line with the 3-fold axis of the alpha(3)beta(3) unit(2, 57) . There is considerable evidence for asymmetry under noncatalytic conditions from cross-linking(7) , subunit dissociation(8) , and nucleotide binding studies(9, 10) . To account for the participation of all three alphabeta pairs during ATP synthesis, the position of either the small subunits (, , and ) or the large subunits (alpha, beta) are predicted to change relative to one another (i.e. rotate or flicker). This assumed dynamic behavior, for which there is now some cryoelectronmicroscopic evidence(11) , has given rise to a model frequently referred to as the ``rotating catalytic site'' model in which nucleotide binding changes are inferred(6) .

Despite the availability of a working model at the quaternary structural level, many questions remain. First, there is the question of the total number of reversible (exchangeable) nucleotide binding sites on F(1). Although commonly stated to be three(1, 2, 3, 4, 5, 6, 57) , most workers have not examined the simultaneous binding to F(1) of the substrate (ADP) and the product (ATP) of oxidative phosphorylation. Second, it remains controversial as to whether beta subunits contain two nucleotide binding sites (12, 13) or only a single nucleotide binding site(14, 15) . Third, the role of alpha subunits in contributing to both catalysis and reversible nucleotide binding to F(1) remains controversial, with the preferred view being that the alpha subunit optimizes the structure of beta for catalysis (16, 17) while contributing partially (18, 19) or not at all (20) to catalytic sites. Fourth, although a role for one or more of the small subunits (, , and ) in altering nucleotide binding is inferred from current models(1, 2, 3, 4, 5, 6, 57) , experimental evidence is lacking, and it remains possible that nucleotide binding may direct small subunit interactions. Finally, few workers have given serious consideration to the possibility that at least two distinct states of F(1), catalytically active and inhibited, with different structural features may be important, respectively, in the function and regulation of the enzyme. The latter possibility has taken on added interest in view of recent reports (21, 22) that rat liver and bovine heart F(1) exhibit differences in their crystalline forms.

The first x-ray map of an F(1) preparation, obtained on the rat liver enzyme at 3.6 Å(21) , revealed several important features of the molecule. First, subunits alpha and beta were shown to interdigitate in an alternating arrangement. Second, evidence for a nucleotide binding region on alpha subunits near alpha/beta interfaces was obtained. Finally, the small subunits, although not detected, were suggested to either reside in the center of the molecule or not to be ordered. An x-ray map of the bovine heart enzyme at 6.5-Å resolution (22) , although not distinguishing alpha and beta subunits, depicts a molecule with features very similar to those of the rat liver enzyme, but with some distinct asymmetrical features. These studies implicate two distinct forms of F(1) ATPases.

To help address some of the major questions described above, we have carried out a comprehensive study involving a variety of different approaches. The results obtained are discussed within the context of current views about the relationship of the unique substructure of F(1) to its nucleotide binding properties, catalytic functions, active and inhibited states, and crystalline form.

EXPERIMENTAL PROCEDURES

Materials

Rats (Sprague-Dawley, white males) were obtained from Charles River Breeding Laboratories. ATP, ADP, AMP-PNP, (^1)Tris, MgCl(2), EDTA, phosphoenolpyruvic acid, pyruvate kinase, lactic dehydrogenase, and bovine serum albumin were obtained from Sigma. Pyrene maleimide was obtained from Molecular Probes. Ammonium sulfate and potassium phosphate were from J.T. Baker Chemical Co. Tuberculin syringes were from Becton-Dickinson Co., and the Protein Pak 300 column was from Waters-Millipore Corp. Sephadex G-50 was purchased from Pharmacia Biotech Inc., and the luciferase/luciferin kit was from LKB Pharmacia Wallac. [alpha-P]AMP-PNP was from ICN, and [alpha-P]ATP, [-P]ATP, and [^3H]ADP were from DuPont NEN. All other reagents were of the highest purity commercially available.

Methods

Purification of Rat Liver F(1)-ATPase

The enzyme was purified by the procedure of Catterall and Pedersen (23) with the modification described by Pedersen et al.(24) . The purified enzyme, in 250 mM KP(i) and 5.0 mM EDTA, was divided into 100-µl aliquots and lyophilized to dryness and stored at -20 °C until use.

Crystallization of F(1)-ATPase

The purified enzyme, contained as lyophilized samples (100-200 µg) in Pyrex glass tubes, was redissolved in 100 µl of distilled water/tube to give a protein concentration of 1-2 mg/ml and a buffer concentration of 250 mM KP(i) and 5 mM EDTA, pH 7.5. The enzyme was then precipitated twice with ammonium sulfate, redissolving with 100 µl, 200 mM K(2)SO(4), and 10 mM Tris-Cl, pH 7.2, between the first and second precipitation. The final pellets in several tubes were dissolved in 200 mM KP(i) and 5 mM ATP, pH 7.4, and combined to give a final F(1) protein concentration of 5-10 mg/ml. Crystals obtained by ammonium sulfate precipitation as described previously (25) belong to the space group R32 with hexagonal cell dimensions of a = b = 146 Å; c = 368 Å. A 3.6-Å map determined by x-ray diffraction analysis using these crystals has been described(21) .

Assay For ATPase Activity

Two methods were employed. In the first, the spectrophotometric procedure was used in which ADP formed was coupled to the pyruvate kinase and lactic dehydrogenase reactions (23) . The reaction mixture contained the following in a volume of 1 ml at pH 7.5 and 25 °C: 4 mM ATP, 65 mM Tris-Cl, 4.8 mM MgCl(2), 2.5 mM KP(i), 0.40 mM NADH, 0.60 mM phosphoenolpyruvic acid, 5 mM KCN, 1 unit of lactic dehydrogenase, 1 unit of pyruvate kinase, and amounts of F(1) indicated in legends. In the second the chemiluminescent procedure was used. This method involved adding F(1) to the 1- ml reaction system described below for determining nucleotide binding by a chemiluminescent assay. The assay contained 0.15 µM ATP; after adding F(1), ATPase activity was monitored by following the decrease in the chemiluminescent signal.

Circular Dichroism Spectroscopy

CD spectra were obtained at 20.9 °C in a 0.10-mm path length cuvette (Suprasil) using an Aviv 60 HDS spectropolarimeter. F(1) was present at a concentration of 3.1 µM in a 40-µl system containing 250 mM KP(i), 5 mM EDTA, pH 7.5. The beta fraction was present at a concentration of 8.2 µM in a 40-µl system containing 50 mM Tris-Cl, pH 7.4, and 5 mM MgCl(2).

Gel Electrophoresis under ``Native'' Conditions

Electrophoresis was carried out at 25 °C in cylindrical gels (0.5 times 10 cm) using 7.5% polyacrylamide and a Tris glycine buffer, pH 8.3, as described by Williams and Reisfeld(26) . The conditions for stabilizing F(1) prior to and during electrophoresis are indicated in the appropriate figure legend.

Detection of F(1) on ``Native'' Gels after Electrophoresis Using an ATPase Activity Assay

ATPase activity was monitored by following the precipitation of Pb(3)(PO(4))(2)(27) . Gels were incubated at 25 °C in a solution containing 35 mM Tris-Cl, 270 mM glycine, 14 mM MgSO(4), 0.2% Pb(NO(3))(2), and 8 mM ATP.

Gel Electrophoresis in SDS-PAGE under Denaturing Conditions

Electrophoresis was carried out at 25 °C in a Bio-Rad Protein I apparatus in slab gels (14 cm wide, 12.7 cm in length, and 0.2 cm in thickness) using the buffer, fixing, and staining methods described by Weber and Osborn(28) .

HPLC Chromatography

Molecular sieve chomatography was carried out at 25 °C using a Waters Protein PAK 300 column. The eluting buffer was 100 mM Tris-Cl, pH 7.4. The appearance of protein in the eluate was monitored at 280 nm using a Waters 740 data module.

Nucleotide Binding: Column Centrifugation Assay

Binding assays were carried out at 25 °C by incubating F(1), or fractions derived therefrom, for 20 min in a final volume of 100 µl, containing concentrations of F(1), nucleotide, MgCl(2), and buffer indicated in legends to tables and figures. The entire reaction mixture was loaded onto a Sephadex G-50 ``fine'' column (1-cm^3 tuberculin syringe with a filter at the bottom), which had been preequilibrated with 50 mM Tris-Cl, pH 7.6, and precentrifuged for 1.5 min at 2,500 rpm in an IEC model HN-SII clinical centrifuge. Centrifugation of the reaction mixture was carried out for 1.5 min at 2500 rpm to separate nucleotide bound to F(1) from free nucleotide(10, 29) .

Nucleotide Binding: Chemiluminescent Assay

Assays for ATP were carried out in a 1-ml system containing 20 mM Hepes, 10 mM NaP(i), 5 mM MgCl(2), 1 mM EDTA, and 20 µl of an ATP monitoring agent (LKB Pharmacia Wallac) containing luciferin and luciferase. The final pH was 7.5 and the final temperature was 25 °C. The chemiluminescent signal was monitored on a LKB Wallac Luminometer 1250 interfaced with a LKB recorder. Assays for ADP were monitored as above in the presence of 0.6 mM phosphoenolpyruvic acid and 4 units of pyruvate kinase. Prior to all experiments, data were collected for a ``standard'' ATP or ADP curve. To remove bound nucleotides from F(1), samples were boiled for 2 min and centrifuged to remove denatured F(1). The supernatant was then added to the above assay mixtures for determination of ATP or ADP.

Nucleotide-induced Fluorescent Changes

Purified lyophilized F(1) (125 µg) was suspended in 100 µl of 250 mM KP(i) + 5 mM EDTA, pH 7.5, and precipitated twice with ammonium sulfate. The precipitate was redissolved in 100 µl of 50 mM Tris-Cl, pH 7.4, and, where indicated, nucleotide or nucleotide + MgCl(2) was added. Incubation was carried out for 20 min at 25 °C, and the entire reaction mixture subjected to column centrifugation exactly as described above. The resultant F(1) (50 µg in 40 µl), with and without bound nucleotide, was then added to a 1.5-ml quartz cuvette (designed for fluorescence) containing 1 ml of 50 mM Tris-Cl, pH 7.4. The fluorescence probe pyrene maleimide that specifically labels the F(1)-alpha subunit was added to give a final concentration of 1 µM and the fluorescence intensity monitored at an excitation wavelength of 339 nm and an emission wavelength of 378 nm using a Perkin-Elmer LS 50B spectrometer.

Determination of Protein

Protein was determined by the method of Lowry et al. (30) after first precipitating with 5% trichloroacetic acid. Amino acid composition data verified that this method accurately reflects the mass of a given amount of rat liver F(1).

RESULTS

Rat Liver F(1): General Properties and Prior Treatment

The rat liver F(1) preparation used in these studies when purified as described under ``Methods'' exhibits average specific activities, respectively, in Tris-Cl and Tris bicarbonate buffers at 25 °C of 24 and 55 µmol of ATP hydrolyzed/min/mg of protein(31) . The enzyme is fully competent in restoring ATP synthesis to F(1)-depleted inner membrane vesicles (32) and is free of the ATPase inhibitor peptide (33) known to enhance nucleotide binding to F(1)(34, 35) . Prior to all studies reported below, aliquots (150-250 µg) of purified, lyophilized F(1) (see ``Methods'') were dissolved at 25 °C in 100 µl of water and precipitated twice with ammonium sulfate to remove loosely bound nucleotide. Between precipitations F(1) was dissolved in 100 µl of 200 mM K(2)SO(4) + 10 mM Tris-Cl, pH 7.5, to preserve activity. Most experiments were carried out between 5 and 18 times on at least three different F(1) preparations.

Nucleotide Binding Capacity under Nonhydrolytic Conditions: Evidence for Five Sites with Distinct Sites for ADP and AMP-PNP

Although previous studies from this laboratory(10, 36) have shown that rat liver F(1) binds ADP reversibly at a single site (K(D) = 1 µM) and AMP-PNP at three sites (K(D) = 1 µM for 1 site and 25 µM for the other two), the total capacity for nucleotide binding has not been assessed, nor has the relationship between the ADP and AMP-PNP sites.

Table 1, part A, shows that purified F(1) precipitated twice with ammonium sulfate retains about 1 mol of ADP and 1 mol of ATP/mol of F(1). In part B, it is seen, consistent with our earlier findings(10, 36) , that addition of [^3H]ADP alone results in the net binding of 1 mol of ADP/mol of F(1) whereas the addition of [P]AMP-PNP alone results in the binding of 3 mol of AMP-PNP/mol of F(1). Significantly, however, the combined addition of the two differentially labeled nucleotides also results in the net binding of 1 mol of ADP/mol of F(1) and 3 mol of AMP-PNP/mol of F(1) emphasizing that the single detectable ADP site is separate and distinct from the three sites for AMP-PNP. Previously, we had shown that ADP does not alter the net binding of AMP-PNP, nor does AMP-PNP alter the net binding of ADP(10) . Additionally, the addition of MgCl(2) or CoCl(2) does not alters the stoichiometry of ADP or AMP-PNP binding(10, 36) .

Results presented in Table 1, part C, address the question of whether the single ADP site (part B) detected by radiolabeling is distinct from the endogenous ADP site (part A) detected by the chemiluminescent assay. If the two sites are identical, then prior incubation of F(1) with ADP followed by determination of total ADP by the chemiluminescent assay should reveal only 1 mol of ADP/mol of F(1). Conversely, if the two sites are distinct, 2 mol of ADP/mol of F(1) should be detected. As shown in part C the latter answer is obtained. As will be revealed in studies described below, these two ADP sites reside on different F(1) subunits. (In studies not presented here the single endogenous ATP site was shown to be readily reversible accounting for one of the three AMP-PNP sites.)

In summary (Table 1, part D), under nonhydrolytic conditions there are five readily detectable nucleotide binding sites on rat liver F(1), one reversible ADP site, three reversible AMP-PNP sites, and one site for the nonexchangeable binding of ADP. These nucleotide binding characteristics, together with those described earlier (10, 36) support the view that under the assay conditions employed, rat liver F(1) behaves as an asymmetrical molecule.

Relationship of F(1) Substructure to Catalytic Activity

Addition of MgCl(2) alone to rat liver F(1) has been shown to promote dissociation of the alpha and subunits from the intact complex with a simultaneous loss of ATPase activity(37) . Results presented in Fig. 1A show that incubation of F(1) with MgCl(2) for 20 h at room temperature instead of the 4 h previously used (37) results in a complete loss of ATPase activity as monitored by a spectrophotometric assay (see ``Methods''). Moreover, as shown in Fig. 1B the beta, , and subunits, which remain soluble, are separated completely from the alpha and subunits, which are insoluble and can be sedimented. (For simplicity, these two fractions are referred to below and throughout the paper as the beta fraction and the alpha fraction.)

Figure 1: A, loss of ATPase activity of F(1) in the presence of MgCl(2) accompanied by sedimentable protein. After precipitating F(1) (300 µg) twice with ammonium sulfate at 25 °C, the enzyme was dissolved in 200 µl of 50 mM Tris-Cl, pH 7.4. After dividing the sample into four 75-µl aliquots, 5 mM MgCl(2) was added and the samples allowed to remain at 25 °C for the times indicated. Samples were then subjected to centrifugation for 20 min at 12,000 rpm at 25 °C in a Sorvall RC 2B centrifuge, followed by protein determinations on the supernatant and pellet fractions. B, SDS-PAGE analysis of supernatant (S) and pellet (P) fractions obtained in A. SDS-PAGE was carried out in slab gels as described under ``Methods.'' The amount of protein loaded on the gels was, respectively, 7.5 µg (leftpanel) and 15 µg (rightpanel). Note: although the F(1) preparation used is over 95% pure, overloading of the gel and the photography to enhance visualization of the and subunits also enhances minor contaminants. In these long slab gels the small subunits, and , tend to diffuse. They are much more distinct when electrophoresis is carried out in a Bio-Rad Mini-Protean dual slab cell (see Fig. 4, inset). No evidence for heterogeneity in the F(1) preparations used (e.g. alpha(3)beta(3) and alpha(3)beta(3) forms) was observed when F(1) was subjected to electrophoresis in native gels at acrylamide concentrations ranging from 3 to 15%. (In subsequent legends to figures and tables the supernatant fraction is referred to as the beta fraction and the pellet fraction as the alpha fraction.)


Figure 4: Relative capacities of ADP and MgATP to induce changes in the fluorescence of the probe pyrene maleimide. F(1) (150 µg) was incubated at 25 °C for 20 min in a 0.1-ml system containing 50 mM Tris-Cl, pH 7.4, and either 5 mM ADP or 5 mM ATP + 5 mM MgCl(2). The incubation mixtures were then subjected to column centrifugation (see ``Methods''). These two different conditions result in the reversible binding of 1 mol of ADP/mol of F(1) (Table 1) and 2.5 mol of ADP/mol of F(1) (Table 3), respectively. The relative capacity of pyrene maleimide to bind to these two different nucleotide-containing preparations was then assessed fluorometrically exactly as described under ``Methods.'' Inset, SDS-PAGE of F(1) labeled with pyrene maleimide. Electrophoresis was carried out in a Bio-Rad Mini-Protean dual slab cell in 15% acrylamide according to the method of Laemmli(56) . The gel on the left was placed on a UV transmitter to visualize fluorescence and subsequently stained with Coomassie Blue to visualize protein. Pyrene maleimide labeling is not observed in the , , and subunits and is observed only in the larger band corresponding to the alpha and beta subunits. beta subunits contain no cysteine residues and are not labeled by pyrene maleimide.


Results presented in Fig. 2A show that the beta fraction retains significant secondary structure as revealed by circular dichoism spectroscopy, and similar to intact F(1), exhibits a high degree of alpha-helical character. However, no ATPase activity could be detected in the beta fraction even after electrophoresis under native conditions which resulted in retention of ATPase activity in control F(1). The method used to detect ATPase activity within the gel following electrophoresis (see ``Methods'') is based on the reaction of lead nitrate with the P(i) produced in the ATPase reaction to give a white precipitate of lead phosphate. The sensitivity of the assay is limited only by time, and even after several days no precipitate was observed in the gel resulting from electrophoresis of the beta fraction.

Figure 2: A, circular dichoism spectra of F and the beta fraction. Circular dichoism spectroscopy was carried out exactly as described under ``Methods.'' The percent of secondary structure was calculated from the program PROSEC(55) . B, native PAGE gels depicting F and the beta fraction when stained for protein and activity. PAGE was carried out under native condition in cylindrical gels and stained for protein and activity exactly as described under ``Methods.'' Both F(1) (25 µg) and the beta fraction (25 µg) were loaded in 50 mM Tris-Cl, pH 7.4, containing 50% glycerol. Gels containing F(1) also included 10% glycerol, 5.0 mM ATP, and 3 mM MgCl(2) to minimize its propensity to dissociate.


The finding that loss of ATPase activity of intact F(1) occurs simultaneously with loss of the alpha subunit (Fig. 1A), and the additional finding that the fraction containing the beta subunit exhibits no ATPase activity (Fig. 2B) while retaining significant structure (Fig. 2A), supports the view that the beta subunit alone is not a catalytically active unit(38) .

Relationship of F(1) Substructure to the Five Nucleotide Binding Sites Detected under Nonhydrolytic Conditions

To address this question, F(1) was incubated for 20 h at 25 °C in the presence of 5 mM MgCl(2) and then sedimented to remove the alpha fraction. Control F(1) and the beta fraction were then incubated in the presence of labeled ADP, AMP-PNP, or ADP + AMP-PNP exactly as described in Table 1and subjected to column centrifugation (see ``Methods''). Calculation of nucleotide binding was based on the original amount of F(1).

Results tabulated in Table 2show that when 5 mM nucleotide is in the binding assay, the four reversible nucleotide binding sites on F(1) (i.e. the one ADP site, and the three AMP-PNP sites) are accounted for within the beta fraction. Additionally, consistent with results obtained on intact F(1) (Table 1), ADP does not alter AMP-PNP binding nor does AMP-PNP alter ADP binding. However, it will be noted that at 1 mM nucleotide, where the single ADP site is still recovered in the beta fraction, only one of the three AMP-PNP sites can now be accounted for (see values in parentheses in Table 2). The K(d) for the one ADP site and the one AMP-PNP site remain near 1 µM (data not shown), values obtained for the intact enzyme(10, 36) . However, K(d) values for the other two AMP-PNP sites have increased over 4-fold, indicating either that they also reside on alpha subunits (i.e. at alpha/beta interfaces) or that they have been damaged in the preparation of the beta fraction.

Finally, Table 2summarizes results obtained on the alpha fraction, which is shown to retain the single endogenous, nonexchangeable ADP site characteristic of intact rat liver F(1) (Table 2). Previously, we have shown that the single tight Mg site characteristic of liver (10) and heart (39) F(1) preparations is also recovered in the alpha fraction(10) . Because of the insolubility of this fraction, it could not be tested for additional nucleotide binding by the column centrifugation method.

Of special interest is the finding that within the beta fraction the asymmetry in binding ADP is retained, implicating the tight association of either the or subunits (or both) with a single beta subunit. To investigate this possibility, the beta fraction was subjected to molecular sieve HPLC chomatography. One major peak followed by a trailing component well within the included volume was observed (Fig. 3A). Both the major peak and the trailing component were collected, concentrated, and subjected to SDS-PAGE (Fig. 3B). Significantly, the major peak which contained about two-thirds of the total starting protein migrated as a single species corresponding to the beta subunit. In contrast, the trailing component corresponding to about one-third of the total starting protein migrated as two species, one corresponding to the beta subunit and the other to the subunit. The subunit was not detected and may have bound irreversibly to the column. Nevertheless, these results indicate that one of the three beta subunits within the beta fraction may be associated with a single subunit, the other two beta subunits evidently migrating as a dimeric species prior to the beta unit (Fig. 3A).

Figure 3: A, elution profile of the beta fraction on a HPLC molecular sieve column. The beta fraction, 61.5 µg in 50 µl of 50 mM Tris-Cl, pH 7.4, containing 5.0 mM MgCl(2) was loaded on a Waters Protein Pak 300 molecular sieve HPLC column and eluted with 100 mM Tris-Cl buffer, pH 7.4. The fractions designated I and II were collected and concentrated in an Amicon-Centricon 10 device. B, SDS-PAGE of the total concentrates from fraction I and II relative to that of control F(1). SDS-PAGE was carried out as described under ``Methods.''


The above findings indicate that the beta fraction participates fully in the reversible binding of 1 mol of ADP and 1 mol of AMP-PNP/mol of F(1), and that the alpha fraction contains the one nonexchangeable ADP site. These findings further indicate that the asymmetry of nucleotide binding found in intact F(1) is preserved within the alpha and beta fractions.

Nucleotide Binding Properties of F(1) after Experiencing ATP Hydrolysis

In the studies described above, the nucleotide binding properties of F(1) were examined under nonhydrolytic conditions. These conditions consistently show that the rat liver enzyme binds reversibly no more than a single ADP molecule. It was of interest, therefore, to establish whether additional sites for binding ADP would become accessible during ATP hydrolysis. Two methods were employed, one involving the chemiluminescent assay (see ``Methods''), and a second less direct method in which MgATP was added as alpha-P-labeled ATP with and without EDTA.

Results presented in Table 3show that during ATP hydrolysis rat liver F(1) binds up to 2.5 mol of ADP/mol of F(1) or 1.5 mol more than the single mole that could be added prior to catalysis. Therefore, ATP hydrolysis induces F(1) to provide nearly two additional sites for binding ADP that were previously inaccessible to this nucleotide. Interestingly, the total nucleotide content following ATP hydrolysis still approaches 5 mol/mol of F(1) with only 1 mol being retained as ATP. The subunit distribution of the bound nucleotides could not be identified under these conditions as the MgATP-treated enzyme could not be subfractionated into alpha and beta fractions.

Nucleotide-induced Changes In the Fluorescence of the Probe Pyrene Maleimide

The above studies indicate that the form of F(1) that contains 1 reversible mol of ADP/mol of enzyme in the beta fraction (Table 2) may be structurally different from the form obtained after ATP hydrolysis where two additional ADP sites are loaded. To test this possibility, the enhancement of the fluorescent probe pyrene maleimide upon binding to specific sites on alpha subunits was monitored. This sulfhydryl-reactive reagent binds covalently to alpha-subunit cysteine residues without altering ATPase activity of rat liver F(1). In studies not presented here, rat liver F(1) has been shown to contain (per mole) 2 mol of cysteine located on alpha subunits accessible for reaction with N-ethylmaleimide. Subunits beta and do not contain cysteine residues and the and subunits, which do, are not labeled under these conditions (Fig. 4, inset).

Results presented in Fig. 4show pyrene maleimide, which alone is essentially nonfluorescent, induces a marked fluorescent enhancement upon binding to F(1) alpha-subunits. This enhancement is significantly decreased in the F(1) preparation that contains a single, reversible, mol of ADP/mol of F(1) but decreased much more in the F(1) preparation that has experienced ATP hydrolysis and bound ADP at two additional sites. Fig. 5presents the time course of the fluorescence response to the addition of pyrene maleimide. Here it is clear that over the time period monitored, the two different ADP-F(1) forms exhibit less fluorescence upon the addition of pyrene maleimide than control F(1), and that the form that has undergone ATP hydrolysis exhibits the least fluorescence. Significantly, results also presented in Fig. 5show that an F(1) form that has been prepared as described in Table 1in the presence of both ADP and AMP-PNP reduces the fluorescence of pyrene maleimide to almost the same extent as the form that has experienced ATP hydrolysis.

Figure 5: Time course of the reactivity of pyrene maleimide with F(1) samples pretreated with ADP, MgATP, or AMP-PNP + ADP. F(1) samples were treated exactly as described in Fig. 4with 5 mM nucleotide or, in the case of MgATP, with 5 mM ATP + 5 mM MgCl(2) and subjected to column centrifugation. The resultant F(1) samples containing bound nucleotide were then assessed fluorometrically for their interaction with pyrene maleimide exactly as described under ``Methods.''


These studies implicate three distinct forms of rat liver F(1), the starting preparation containing 2 mol of endogenously bound nucleotide/mol of enzyme (Table 1, panel A), the form containing an additional ADP bound at a reversible site within the beta fraction (Table 1, panel B; and Table 2), and the form that has either experienced ATP hydrolysis (Table 3) or been treated to load both ADP and AMP-PNP (Table 1, panel B).

Catalytic Activity and Nucleotide Content of Rat Liver F(1) Derived from Three-dimensional Crystals

Although rat liver F(1) has been crystallized and a 3.6-Å map has been obtained(21) , information about the catalytic activity and nucleotide content of the enzyme obtained from these crystals has not been documented. This information is of special interest both as it relates to the nucleotide binding properties of the enzyme in solution reported above and as it relates to the state of the enzyme on which a high resolution structure is now being obtained.

Rat liver F(1) is routinely crystallized from a solution containing 5 mM ATP and 200 mM KP(i), pH 7.5 (see ``Methods''). In this buffer the enzyme is maximally catalytically active when MgCl(2) is added(40) . Results summarized in Table 4, part A, show that, when crystals obtained by adding ammonium sulfate to F(1) in 5 mM ATP and 200 mM KP(i), pH 7.5, are redissolved after several months, there is full retention of catalytic activity both in Tris-Cl buffer and in the known activating buffer Tris bicarbonate(31) . The critical importance of ATP in maintaining rat liver F(1) in a stable form within the crystals was demonstrated in experiments where crystals were washed in a medium containing ammonium sulfate and P(i) but lacking ATP. Such crystals, which now contain only ADP (<2 mol/mol of F(1); Table 4, part B), yield upon redissolving an almost completely inactive enzyme (Table 4, part A). Although it is not possible to assess the exact stoichiometry (i.e. mol of ATP/mol of F(1)) in the crystals while in the presence of 5 mM ATP, it seems likely that at least 1 mol of ATP/mol of F(1) is present in the crystalline enzyme. Rat liver F(1) as isolated does contain 1 mol of ATP/mol of F(1) (Table 1), indicating that one tight site for binding ATP is present.

These results emphasize that rat liver F(1) crystals being used to obtain a high resolution structure maintain the enzyme in an ATP-dependent stable form that can be readily recovered with full retention of catalytic activity.

Relationship between the Nucleotide Content of Rat Liver F(1) and Its Catalytic Capacity

Experiments described here have referred to five different nucleotide containing states of the enzyme, which will be designated now as F(1)^A, F(1)^B, F(1)^C, F(1)^D and F(1)^F. F(1)^A is the enzyme as isolated after precipitation twice with ammonium sulfate. F(1)^B is the enzyme form that results from adding ADP to F(1)^A. F(1)^C is the form that results from adding ADP + AMP-PNP to F(1)^A, and F(1)^D is the form resulting from adding MgATP to F(1)^A and allowing ATP hydrolysis to occur. Finally, F(1)^E is a sixth form not discussed thus far which results by adding additional ADP to F(1)^D, and F(1)^F is the form resulting by incubating F(1)^A in crystallization buffer. In order to determine the extent to which nucleotide content alters catalytic activity, each of the six F(1) forms was prepared and immediately assayed for ATPase activity by the chemiluminescent procedure (see ``Methods''). This assay can be conducted at very low ATP concentrations and can readily detect inhibition due to ADP by monitoring initial rates after addition of F(1).

Results presented in Table 5show that when initial ATPase rates are compared to that of the fully active F(1)^A form, the second most active F(1) form is F(1)^F, the enzyme treated with crystallization buffer (5 mM ATP + 200 mM KP(I), pH 7.5). F(1)^B and F(1)^D also are highly active. In contrast, form F(1)^E is 91% inhibited and F(1)^C is almost 60% inhibited. It is of interest to note that F(1)^D, which has one site filled with ATP and nearly four sites filled with ADP, is highly active, and only when additional ADP is added directly to the assay to displace the bound ATP (or fill the sixth site with ADP) to give F(1)^E is there a dramatic inhibition. F(1)^C, which contains both ADP and AMP-PNP, is also not fully inhibited. However, by adding additional ADP to the assay F(1)^C, as F(1)^E, is now inhibited by over 90%.

These results emphasize that F(1)^F derived by incubating F(1)^A in crystallization buffer is almost fully active, and that a form of F(1) containing a 4/1 ADP/ATP ratio is almost fully active as well until additional ADP is added. In contrast, F(1)^C, which has five sites filled, 3 with AMP-PNP and 2 with ADP, is 60% inhibited.

DISCUSSION

Results described here represent the first comprehensive study of the nucleotide binding properties of intact rat liver F(1), and one of the most comprehensive studies of this nature on an F(1) preparation to date. Experiments were carried out with F(1) in the presence of ADP alone, AMP-PNP alone, ADP and AMP-PNP together, MgATP alone, and ATP and P(i) together. In addition, F(1) preparations resulting from these treatments have been assayed both for catalytic activity and for their capacity to interact with the fluorescent probe pyrene maleimide. Additionally, F(1) obtained from crystals currently being used to obtain a high resolution structure have been analyzed for both catalytic activity and nucleotide content. Finally, the stability of F(1) within the crystals has been determined. These studies are fundamental to our understanding both of how ATP synthases from mammalian tissues carry out ATP synthesis and of how these enzymes are regulated by product inhibition. In addition, these studies are fundamental in defining those enzymatic forms that an F(1) molecule can assume during its catalytic and regulatory modes and will permit us to identify those forms on which three-dimensional structures are being determined(21, 22) .

The experimental results obtained on rat liver F(1) are best discussed and interpreted within the framework of the scheme presented in Fig. 6. Here six different forms of the enzyme previously designated in Table 5as F(1)^A, F(1)^B, F(1)^C, F(1)^D, F(1)^E, and F(1)^F are presented. F(1)^A, an active form (Table 5), is the isolated enzyme after twice precipitating with ammonium sulfate. This form has 2 nucleotide binding sites filled (Table 1, part A), one with a nonexchangeable ADP recovered in the alpha fraction (Table 2, part B) and one with an ATP (Table 1, part A), exchangeable with AMP-PNP and accounted for in the beta fraction (Table 2, part A). The K(D) (1 µM) of this site in binding AMP-PNP (10) is in the same range as the K(i) (2.2 µM) of AMP-PNP in inhibiting F(1) ATPase activity(41) . In more conventional language, the very tight ADP site can be referred to as ``noncatalytic''and the exchangeable ATP site as ``catalytic.'' As ADP is retained tightly bound to the noncatalytic site even in the alpha fraction (Table 2, part B), its binding domain is predicted to lie predominantly within an alpha subunit, although at an alpha/beta interface consistent with our earlier x-ray crystallographic studies(21) . Similarly, as the AMP-PNP bound to the catalytic site is accounted for in the beta fraction without a loss of affinity (Table 2, part A), its domain is predicted to reside predominantly within a beta subunit.

Figure 6: Diagram depicting six different forms of rat liver F predicted from results of studies reported here. The data presented provide evidence for six distinct forms of F(1) designated as F(1)^A, F(1)^B, F(1)^C, F(1)^D, F(1)^E, and F(1)^F. F(1)^A represents the enzyme as isolated after twice precipitating with ammonium sulfate, F(1)^B the enzyme after adding ADP to F(1)^A, and F(1)^C after adding AMP-PNP to F(1)^B. F(1)^D is the form obtained after incubating F(1)^A with MgATP to induce ATP hydrolysis, and F(1)^E results by adding additional ADP to F(1)^D. F(1)^F results by adding ATP and high P(i) (200 mM) to F(1)^A. Rat liver F(1) has been crystallized under the latter conditions (see ``Methods''), which results in a catalytically active F(1) molecule ( Table 4and Table 5) with 3-fold symmetry. In contrast bovine heart F(1) has been crystallized in the presence of ADP and the inhibitors AMP-PNP and sodium azide (46) to give an enzyme with distinct asymmetrical features(22) . This form most closely resembles F(1)^C in the diagram. The small subunits , , and are not shown, although the subunit is believed to lie at the center of F(1) extending from the bottom to the top of the molecule(22) . (See ``Discussion'' for a more detailed description of the six F(1) forms depicted here. Additionally, please note that to depict the purine moiety of ADP or AMP-PNP as acting at an interface in some cases, it has been necessary to write these in reverse as PDA or PNP-PMA, respectively.)


F(1)^A is converted to F(1)^B, also an active form (Table 5), by adding ADP, which even at 5 mM fills only a single reversible site (Table 1, part B). The K(D) of this site is also near 1 µM(36) and is accounted for in the beta fraction without a loss of affinity (Table 2, part A). Moreover, ADP can be added to this site on F(1) without altering the binding of AMP-PNP or ATP bound at the catalytic ATP site (Table 1, parts B and C). Therefore, this site is predicted to reside on a second beta subunit conformationally distinct from the beta subunit containing ATP. This reversible ADP site is also predicted to be a catalytic site, now poised for ATP synthesis. Other data presented here (Fig. 3) indicate that a conformational change has occurred in the conversion of F(1)^A to F(1)^B as pyrene maleimide fluorescence is significantly reduced ( Fig. 4and Fig. 5).

F(1)^C, an inhibited form (Table 5), is obtained either by adding ADP + AMP-PNP to F(1)^A or only AMP-PNP to F(1)^B (Table 1, part B). The final enzyme that results has five sites filled, two with ADP and three with AMP-PNP. One site is filled with AMP-PNP by displacing the exchangeable ATP bound on a beta subunit and can be accounted for in the beta fraction (Table 2, A). The other two AMP-PNP sites cannot be accounted for by binding predominantly to beta subunits (Table 2) and are assumed to be noncatalytic and to reside predominantly on alpha subunits at alpha/beta interfaces. The conversion of F(1)^B to F(1)^C by addition of AMP-PNP induces a further decrease in the pyrene maleimide fluorescence, i.e. a greater decrease than that produced by adding ADP alone (Fig. 5). This indicates that F(1)^C is conformationally distinct from both F(1)^A and F(1)^B.

A second inhibited form, F(1)^E, is produced (Table 5) by incubating F(1)