The use of 8-azido-ATP and 8-azido-ADP as photoaffinity labels of the ATP synthase in submitochondrial particles: evidence for a mechanism of ATP hydrolysis involving two independent catalytic sites?

ATP synthase and the actions of inhibitors utilized to study its roles in human health, disease, and other scientific areas

ATP synthase, a double-motor enzyme, plays various roles in the cell, participating not only in ATP synthesis but in ATP hydrolysis-dependent processes and in the regulation of a proton gradient across some membrane-dependent systems. Recent studies of ATP synthase as a potential molecular target for the treatment of some human diseases have displayed promising results, and this enzyme is now emerging as an attractive molecular target for the development of new therapies for a variety of diseases. Significantly, ATP synthase, because of its complex structure, is inhibited by a number of different inhibitors and provides diverse possibilities in the development of new ATP synthase-directed agents. In this review, we classify over 250 natural and synthetic inhibitors of ATP synthase reported to date and present their inhibitory sites and their known or proposed modes of action. The rich source of ATP synthase inhibitors and their known or purported sites of action presented in this review should provide valuable insights into their applications as potential scaffolds for new therapeutics for human and animal diseases as well as for the discovery of new pesticides and herbicides to help protect the world's food supply. Finally, as ATP synthase is now known to consist of two unique nanomotors involved in making ATP from ADP and P(i), the information provided in this review may greatly assist those investigators entering the emerging field of nanotechnology.

Cloning and characterization of the subunits comprising the catalytic core of the Trypanosoma brucei mitochondrial ATP synthase

The Trypanosoma brucei mitochondrial F(1)-ATPase has been previously isolated and characterized. It is composed of five subunits of molecular weights 55000, 42000, 32000, 22000, and 17000 [1]. We have identified the alpha and beta subunits of the T. brucei F(1)-ATPase by N-terminal sequence determination together with analysis of cDNA and genomic clones. The genes for both subunits are homologous to the same subunits from other organisms. They contain the Walker A and B boxes of homology and a putative mitochondrial import sequence. The isolated T. brucei alpha subunit is unusually small at 42 kDa. The alpha cDNA clone encodes a protein of predicted size 59 kDa with a mitochondrial import presequence at the N-terminus. The predicted size was confirmed by expression of a 59 kDa protein from the cDNA clone in vitro. These results suggest that the alpha subunit may have an unusually large mitochondrial presequence of 159 amino acids. In contrast, the estimated size of the native beta subunit (55 kDa) correlates well with the size predicted from the cDNA clone, 57 kDa, from which a 21 amino acid presequence has been removed in vivo. The size of the beta subunit was confirmed by expression in an in vitro and an Escherichia coli expression system. The purified recombinant beta subunit, like the native F(1)-ATPase, can be labeled by the photoaffinity nucleotide analogue 8-azido ATP. Binding of the 8-azido ATP probe is best competed by the natural substrate ATP, and is significantly reduced by pretreatment with the inhibitor 7-chloro-4-nitrobenzo-2-oxa-1,3-diazide as has been shown with beta subunits of other organisms. The differential binding of this photoaffinity analogue was used to resolve the identities of the alpha and beta subunits of the ATP synthase from T. brucei. These results are in contrast to results previously obtained for a related trypanosomatid Crithidia fasciculata.

One of the non-exchangeable nucleotides of the mitochondrial F1-ATPase is bound at a beta-subunit: evidence for a non-rotatory two-site catalytic mechanism

In active MF1, one of the two non-exchangeable tightly bound adenine nucleotides is an ATP, while the other is an ADP. The respective sites are called the T-site and the D-site. The activity of the enzyme correlates linearly with the amount of bound ATP, ADP at the T-site being inhibitory. When MF1 is stored at room temperature in 50% glycerol and 100 mM Tris-HCl (pH 7.3) after slow passage through a Sephadex column, the tightly bound ATP is slowly dephosphorylated to ADP which is subsequently released, without effect on activity. When enzyme with about one residual ADP left (at the D-site) was incubated at pH 7.3, after dilution of the glycerol, with 400 &mgr;M [14C]ATP under varying conditions, the amount of tightly bound nucleotide triphosphate again correlated well with activity, the residual ADP being bound at the D-site. Optimal results were obtained when the incubation was performed in the presence of a regenerating system. Binding of 2-azido-ATP instead of ATP to the T-site as a triphosphate, as indicated by the specific activity of the enzyme, appeared to be optimal when the binding was performed at pH 6.4 in the absence of Mg2+ and with high concentrations of the nucleotide. Under such conditions, 3 mol 2-azido-AXP per mol F1 remained tightly bound after ammonium sulfate precipitation and column centrifugation, in addition to about one residual ADP at the D-site. After a 2-min period of turnover with ATP/Mg2+ as substrate two mol 2-azido-AXP were left on the enzyme, of which one was bound at a beta-site. These results show that one of the non-catalytic nucleotide binding sites that contain tightly bound nucleotides, is a beta-site, in conflict with the requirements for a rotatory tri-site mechanism for ATP hydrolysis. This beta-site can further be identified with the T-site. The validity of these conclusions for F1 from other sources and for catalysis by membrane-bound enzyme is discussed.

Characterization of the respiratory chain from cultured Crithidia fasciculata

Mitochondrial mRNAs encoding subunits of respiratory-chain complexes in kinetoplastids are post-transcriptionally edited by uridine insertion and deletion. In order to identify the proteins encoded by these mRNAs, we have analyzed respiratory-chain complexes from cultured cells of Crithidia fasciculata with the aid of 2D polyacrylamide gel electrophoresis (PAGE). The subunit composition of F0F1-ATPase (complex V), identified on the basis of its activity as an oligomycin-sensitive ATPase, is similar to that of bovine mitochondrial F0F1-ATPase. Amino acid sequence analysis, combined with binding studies using dicyclohexyldiimide and azido ATP allowed the identification of two F0 subunits (b and c) and all of the F1 subunits. The F0 b subunit has a low degree of similarity to subunit b from other organisms. The F1 alpha subunit is extremely small making the beta subunit the largest F1 subunit. Other respiratory-chain complexes were also analyzed. Interestingly, an NADH: ubiquinone oxidoreductase (complex I) appeared to be absent, as judged by electron paramagnetic resonance (EPR), enzyme activity and 2D PAGE analysis. Cytochrome c oxidase (complex IV) displayed a subunit pattern identical to that reported for the purified enzyme, whereas cytochrome c reductase (complex III) appeared to contain two extra subunits. A putative complex II was also identified. The amino acid sequences of the subunits of these complexes also show a very low degree of similarity (if any) to the corresponding sequences in other organisms. Remarkably, peptide sequences derived from mitochondrially encoded subunits were not found in spite of the fact that sequences were obtained of virtually all subunits of complex III, IV and V.

Purification and characterization of cytochrome c oxidase from the insect trypanosomatid Crithidia fasciculata

Cytochrome c oxidase was purified from the mitochondrial lysate of the insect trypanosomatid Crithidia fasciculata with the aid of a methyl hydrophobic interaction column in a rapid one-step procedure. The purified complex displayed all characteristics expected from a eukaryotic cytochrome c oxidase: the presence of CuA in electron paramagnetic resonance analysis, a characteristic 605 nm peak in reduced-minus-oxidized optical spectroscopy, and the capacity to efficiently oxidize homologous, but not heterologous, cytochrome c. Two-dimensional PAGE showed that C. fasciculata cytochrome c oxidase consists of at least 10 different subunits. N-terminal sequences were obtained from the six smallest subunits of the complex, one of them showing significant similarity to Neurospora crassa cytochrome c oxidase subunit V. The N-terminus of each of the four largest subunits was found to be blocked.

Identification and partial purification of (Ca2+ or Mg2+)-ATPase in renal brush-border membranes

The protein responsible for the (Ca2+ or Mg2+)-ATPase activity in brush-border membranes from pig kidney tubular cells was characterized to distinguish this enzyme from the N-ethylmaleimide-sensitive Mg(2+)-ATPase, also present in renal brush borders. Both enzymes are clearly different in their pH optimum and their sensitivity to divalent cations, nucleoside 5'-triphosphates and inhibitors. Solubilization of the (Ca2+ or Mg2+)-ATPase from brush-border membrane vesicles was accomplished with Nonidet P-40 or dodecylmaltoside. However, simultaneous inactivation of the enzyme was inevitable. A tenfold enrichment of the ATPase activity was obtained by chromatofocusing of Nonidet-P-40-solubilized brush borders. A similar degree of purification was achieved by ion-exchange chromatography of dodecylmaltoside-solubilized preparations. From the SDS/polyacrylamide gels of partially purified (Ca2+ or Mg2+)-ATPase, a few protein bands could still be tentatively identified as responsible for the enzyme activity. Labeling of solubilized brush-border preparations with several radioactive ATP analogues also revealed that a protein band of molecular mass 90 kDa is the most probable candidate for the catalytic peptide of the (Ca2+ or Mg2+)-ATPase. Finally, immunoprecipitation as well as semi-dry blotting with antibodies generated against partially purified enzyme preparations, confirmed that a 90-kDa component is a reasonable candidate for the (Ca2+ or Mg2+)-ATPase in renal brush-border membranes.

Beef heart mitochondrial F1-ATPase: inhibition by azidoadenyl-5'-yl imidodiphosphates and cooperative binding of substrate

Two ATP analogs, 2- and 8-azidoadenyl-5'-yl imidodiphosphate, were synthesized, purified and utilized as inhibitors of soluble beef heart mitochondrial F1-ATPase under non-photolytical conditions. In the range of 5 microM to 3 mM ATP, the initial rates of ATP hydrolysis in the presence and absence of the inhibiting ATP analogs can be adequately described by two pairs of Km and Vmax values (3 microM, 8.5 mumol ATP/min per mg; 255 microM, 42.0 mumol ATP/min per mg). With increasing inhibitor concentrations, the apparent Km,2 increases as in competitive inhibition, while Vmax,1 decreases as in non-competitive inhibition. The Ki values derived for both types of inhibition are similar, but strongly different for 2- and 8-azido-AMP-PNP (4 microM and 460 microM, respectively). The decrease of the high-affinity Vmax is compensated by an increase in low-affinity catalysis, resulting in a constant sum of maximal velocities. These data can be described by a model where two sites interact with negative cooperativity in binding of substrate.

Characteristics of the non-exchangeable nucleotide binding sites of mitochondrial F1 revealed by dissociation and reconstitution with 2-azido-ATP

The dissociation of mitochondrial F1-ATPase with 3 M LiCl at 0 degrees C, followed by reconstitution, has been analysed. FPLC over a gel filtration column in the dissociation buffer revealed the presence of two protein moieties, an alpha 3 gamma delta epsilon complex and single beta-subunits. When the dissociation and chromatography is performed at pH 6.2, the former protein moiety still contains some adenine nucleotides. Reconstitution of the dissociated complex is not possible any more after FPLC, probably due to the loss of residual adenine nucleotides. After a single column centrifugation step one nucleotide per F1 still remains bound. For reconstitution, additional ATP, or a suitable analog, is required. 2-Azido-ATP, but not 8-azido-ATP or ITP, can replace ATP during the reconstitution. F1, reconstituted in the presence of 2-azido-ATP, contains three tightly bound nucleotides, similar to freshly isolated F1, of which in this case one is an adenine nucleotide and two are azido-adenine nucleotides. One of the latter can be rapidly exchanged and is bound to a catalytic site. Covalent binding (at a beta-subunit) of the other tightly bound 2-azido-ATP by ultraviolet illumination does not result in inhibition of the enzyme. Digestion of F1 with trypsin, followed by HPLC, showed that the label is not bound to the fragment containing Tyr-368, nor to the fragment containing Tyr-345. This result was confirmed by CNBr digestion, followed by SDS-urea PAGE. We conclude that during dissociation of F1 one tightly bound nucleotide (ADP) remains bound at an alpha/beta interface site and that for reconstitution binding of ATP to a (non-catalytic) beta-site is required. The conformation of this site differs from that of the two catalytic beta-sites.

Interaction of Mg2+ with F0.F1 mitochondrial ATPase as related to its slow active/inactive transition

Bovine heart submitochondrial particles incubated with a low concentration of ADP in the presence of Mg2+ and passed through a Sephadex column equilibrated with EDTA exhibit sensitivity of their initial ATPase activity to preincubation with Mg2+. By using particles thus prepared, several characteristics of a Mg(2+)-specific inhibitory site on F0.F1 ATPase were studied. The inhibition was shown to be both time- and Mg(2+)-concentration-dependent, with an equilibrium constant (at infinite time) of 2 x 10(-6) M (25 degrees C, pH 7.5). The dependence of the pseudo-first-order rate constant for the inhibition process on Mg2+ concentration suggests the presence of a single Mg(2+)-binding site with K8 = 1.1 x 10(-4) M. The data obtained are consistent with a two-step mechanism of Mg(2+)-F0.F1 interaction which results in a loss of the ATPase activity; it includes rapid pH-dependent binding of Mg2+ at the site with K8 = 1.1 x 10(-4) M, followed by a slow interconversion of the Mg(2+)-F1 complex into inactive ATPase (kin. = 0.65 min-1, kact. = 0.01 min-1). The Mg(2+)-inhibited ATPase is very slowly (t1/2 approximately 90 min) re-activated in the presence of EDTA. The rate of EDTA-induced re-activation is pH-independent and can be dramatically increased by added ATP, Pi and sulphite. The dissociation constants for free ATP and P1 (5 x 10(-7) M and 1 x 10(-3) M respectively) and the maximal activation rates were determined by measuring the hyperbolic dependencies of the EDTA-induced re-activation of Mg(2+)-de-activated ATPase on the concentrations of the accelerating ligands. Taken together, the data obtained show two functionally detectable free nucleotide-specific binding sites, one site for Pi and one Mg(2+)-specific ATPase-inhibitory site on the F0.F1 mitochondrial ATP synthase complex.

1H-NMR studies on nucleotide binding to the catalytic sites of bovine mitochondrial F1-ATPase

The conformation of adenine nucleotides bound to bovine mitochondrial F1-ATPase was investigated using transfer nuclear Overhauser enhancement measurements. It is shown that all nucleotides investigated adopt a predominantly anti conformation when bound to the catalytic sites. Furthermore, the experiment suggests that 8-azido-ADP and 8-azido-ATP, which are predominantly in the syn conformation in solution, are in the anti conformation when bound to F1 catalytic sites.

Kinetics of the release of the mitochondrial inhibitor protein. Correlation with synthesis and hydrolysis of ATP

(1) The kinetics of the release of the mitochondrial inhibitor protein (IF1) is studied in bovine heart submitochondrial vesicles supplemented with 125I-labelled IF1, using a method for rapidly 'freezing' the state of F1-IF1 interaction. It is shown that generation of a protonmotive force leads to release of IF1 from F1 into solution, following an exponential process. (2) In one set of experiments the rate of IF1 release, in IF1 supplemented vesicles generating a protonmotive force, is correlated with the induction of ATP hydrolytic capacity. It is found that, even under different metabolic states (phosphorylating and non-phosphorylating conditions), both processes follow the same time-course (half-time of around 40 s) and that there is a direct correlation between induced ATPase capacity and IF1 released. This finding rules out the possibility of a non-inhibitory binding site for IF1 on the membrane. (3) In a second set of experiments, also using IF1 supplemented vesicles, the induction of the ATP hydrolytic capacity after energisation is correlated with the induction of the ATP synthetic capacity. Initial rates of both processes are monitored using firefly luciferase, keeping the assay systems as similar as possible. It is shown that the induction of each capacity follows an exponential time-course, with a half-time of around 40 s. This is in good agreement with the half-times obtained for the induction of ATP hydrolytic capacity and the rate of IF1 release, using the quench-stop method. (4) If the induction of ATP hydrolytic and synthetic capacities is followed in untreated submitochondrial vesicles, i.e., vesicles not supplemented with IF1, the extent and time-course of the change in both hydrolytic and synthetic capacities remain correlated, but the half-time of the transient falls to around 10 s. It is suggested that the length of the transient, observed in IF1 supplemented vesicles, results from partial loss of coupling during repeated centrifugations. (5) These results demonstrate that energy-dependent release of IF1 from F1 into solution results in a concomitant increase in both ATP synthetic and hydrolytic capacities of the ATP synthase complex, and that the time-course of this process is sensitive to the degree of coupling of the vesicles.

Binding and hydrolysis of 2-azido-ATP and 8-azido-ATP by isolated mitochondrial F1: characterisation of high-affinity binding sites

The kinetic parameters for the hydrolysis by F1 of the photoreactive nucleotide analogue 2-azido-ATP were determined (Vmax, 105 U/mg F1; Km, 250 microM, in the presence of 1.0 mM SO2-3). In the absence of an activating anion, a non-linear relationship in a Lineweaver-Burk plot was found for the hydrolysis of 2-azido-ATP. The 2-azido-analogues of ATP and ADP proved to be good photoaffinity labels causing notable inactivation of the F1-ATPase activity upon irradiation at 360 nm. This inhibition was also used to demonstrate high-affinity binding of these analogues to a catalytic binding site on the F1. High-affinity binding proved to be an Mg2+-requiring process, occurring with both 2-azido-ATP and 2-azido-ADP but hardly or not occurring with 8-azido-AT(D)P. Covalent binding of 2-nitreno-ATP upon irradiation of F1 containing tightly bound [beta-32P]2-azido-ATP results in a proportional inhibition of ATPase activity, extrapolating to 0.92 mol of covalently bound label per mol of F1 needed for the complete inactivation of the enzyme. When the F1 was irradiated in the presence of excess [beta-32P]2-azido-AT(D)P, 3-4 mol of label were bound when the enzyme was fully inactivated. In all cases, all or most of the radioactivity was found on the beta subunits.

Demonstration of two exchangeable non-catalytic and two cooperative catalytic sites in isolated bovine heart mitochondrial F1, using the photoaffinity labels [2-3H]8-azido-ATP and [2-3H]8-azido-ADP

The photoreactive nucleotides [2-3H]8-azido-ATP and [2-3H]8-azido-ADP could be used to label the nucleotide binding sites on isolated mitochondrial F1-ATPase to a maximum of 4 mol of nucleotide per mol F1, also when the F1 was depleted of tightly bound nucleotides. At a photolabel concentration of 300-1000 microM, label was found on both alpha and beta subunits in a typically 1:3 ratio, independent of the total amount bound. Under these conditions the covalent binding of two nucleotides is needed for full inactivation (Wagenvoord, R.J., Van der Kraan, I. and Kemp, A. (1977) Biochim. Biophys. Acta 460, 17-24). At lower concentrations of [2-3H]8-azido-ATP (20 microM), it was found that covalent binding of only 1 mol of nucleotide per mole F1 was required for complete inactivation to take place indicating catalytic site cooperativity in the mechanism of ATP hydrolysis. Under those conditions, radioactivity was only found on the beta subunits, which would indicate that the catalytic site is located on a beta subunit and that a second site is located on the alpha/beta interface. It is found that four out of the six nucleotide binding sites are exchangeable and can be labelled with 8-azido-AT(D)P, i.e., two catalytic sites and two non-catalytic sites.

Steady-state kinetics of F1-ATPase. Mechanism of anion activation

The kinetic behaviour of the ATPase activity of beef heart F1 depends largely on the exposure of the enzyme to some anionic ligands such as sulphate and/or EDTA. F1 prepared in the presence of such anions exhibited a triphasic kinetic pattern whereas F1 from which those anions were removed by dialysis exhibited only two Km values for ATP. Conversely to what has been previously reported, bicarbonate did not linearize F1-ATPase kinetics. Moreover, anion activation cannot be simply explained by promotion of ADP release but mainly by an increase in affinity of the third catalytic site for ATP.

Inhibition of energy-transducing reactions by 8-nitreno-ATP covalently bound to bovine heart submitochondrial particles: direct interaction between ATPase and redox enzymes

The photoaffinity label 8-azido-ATP has been used to study the effect of inhibition of ATP synthase on ATP-driven reverse electron transfer from succinate to NAD+ ('reversal'), succinate- and NADH-driven ATP synthesis and ATP-Pi exchange. In reversal, where ATPase functions as primary proton pump, inactivation by covalently bound nitreno-ATP results in an inhibition that is proportional to the inactivation of ATP hydrolysis, or, consequently, with the concentration of inactivated ATP synthases. Up to 60% inactivation of the reversal rate does not lead to a decrease in delta mu H+. Inhibition of ATP synthase as secondary proton pump results in case of NADH-driven ATP synthesis in a proportional inhibition, but with succinate as substrate ATP synthesis is less than proportionally inhibited, compared with inactivation of ATP hydrolysis. Inhibition of one of the primary pumps of NADH-driven ATP synthesis, the NADH:Q oxidoreductase, with rotenone also resulted in an inhibition of the rate of ATP synthesis proportional to that of the NADH oxidation. ATP-Pi exchange is much more affected than ATP hydrolysis by photoinactivation with 8-azido-ATP. Contrary to reversal and NADH-driven ATP synthesis the rate of ATP-Pi exchange does not depend linearly, but quadratically on the concentration of active ATP synthases. The observed proportional relationships between inhibition of the primary or secondary pump and the inhibition of the overall energy-transfer reactions do not support the existence of a pool intermediate in energy-transduction reactions. However, the results are consistent with a direct transfer of energy from redox enzymes to ATP synthase and vice versa.