The a subunit ala-217 --> arg substitution affects catalytic activity of F(1)F(0) ATP synthase

A large number of mutations affecting the F(0) sector of Escherichia coli F(1)F(0) ATP synthase have been constructed and characterized. A subset of the missense mutations resulted in fully assembled enzyme complexes blocked in proton translocation and displaying marked decreases in ATP hydrolysis activity. The catalytic activities of one such mutant enzyme, a(ala-217-->arg), have been determined using both multisite and unisite catalysis conditions. As expected, the V(max) of the a(ala-217-->arg) enzyme was reduced under conditions of saturating substrate concentration. However, the F(0) sector amino acid substitution did not affect nucleotide occupancy of the noncatalytic sites. Moreover, the microscopic rate constants measured using unisite methods yielded no significant differences between the intact wild type F(1)F(0) ATP synthase and the a(ala-217-->arg) mutant enzyme. In general, the values for unisite activities in both preparations were very similar to numbers reported in the literature for E. coli F(1)-ATPase. The results suggest that the a(ala-217-->arg) substitution resulted in a defect in catalytic cooperativity and most likely altered the enzyme by inhibiting the rotational mechanism of F(1)F(0) ATP synthase.     

Kinetic characterization of the unisite catalytic pathway of seven beta-subunit mutant F1-ATPases from Escherichia coli

We have studied the kinetics of "unisite" ATP hydrolysis and synthesis in seven mutant Escherichia coli F1-ATPase enzymes. The seven mutations are distributed over a 105-residue segment of the catalytic nucleotide-binding domain in beta-subunit and are: G142S, K155Q, K155E, E181Q, E192Q, M209I, and R246C. We report forward and reverse rate constants and equilibrium constants in all seven mutant enzymes for the four steps of unisite kinetics, namely (i) ATP binding/release, (ii) ATP hydrolysis/synthesis, (iii) Pi release/binding, and (iv) ADP release/binding. The seven mutant enzymes displayed a wide range of deviations from normal in both rate and equilibrium constants, with no discernible common pattern. Notably, steep reductions in Kd ATP were seen in some cases, the value of Kd Pi was high, and K2 (ATP hydrolysis/synthesis) was relatively unaffected. Significantly, when the data from the seven mutations were combined with previous data from two other E. coli F1-beta-subunit mutations (D242N, D242V), normal E. coli F1, soluble and membranous mitochondrial F1, it was found that linear free energy relationships obtained for both ATP binding/release (log k+1 versus log K1) and ADP binding/release (log k-4 versus log K-4). Two conclusions follow. 1) The seven mutations studied here cause subtle changes in interactions between the catalytic nucleotide-binding domain and substrate ATP or product ADP. 2) The mitochondrial, normal E. coli, and nine total beta-subunit mutant enzymes represent a continuum in which subtle structural differences in the catalytic site resulted in changes in binding energy; therefore insights into the nature of energy coupling during ATP hydrolysis and synthesis by F1-ATPase may be ascertained by detailed studies of this group of enzymes.     

Complete kinetic and thermodynamic characterization of the unisite catalytic pathway of Escherichia coli F1-ATPase. Comparison with mitochondrial F1-ATPase and application to the study of mutant enzymes

A complete analysis is presented of the component rate constants of the "unisite" reaction pathway in normal Escherichia coli F1-ATPase. Gibbs free energy profiles of the unisite reaction pathway were constructed for both normal E. coli F1 and bovine-heart mitochondrial F1, and comparison indicated that E. coli F1 is an ancestral form of the mitochondrial enzyme. Similar kinetic and thermodynamic analyses of the unisite reaction pathway were done for mutant beta-Asn-242 and beta-Val-242 E. coli F1-ATPases. Both mutations affected unisite binding and hydrolysis of MgATP but had little effect on release of products or binding of MgADP. It was apparent that a primary effect of the mutations was on the interaction between the catalytic nucleotide-binding domain and the substrate MgATP. The catalytic transition state [F1-ATP]++ was the most destabilized step in the reaction sequence. Measurements of delta delta G[F1.ATP]++ and linear free energy plots for the catalytic step were consistent with the view that, in normal enzyme, residue beta-Asp-242 accepts an H-bond from the transition-state substrate in order to facilitate catalysis. Both mutations impaired positive catalytic cooperativity. This was caused by energetic destabilization of the catalytic transition state and was an indirect effect, not a direct effect on signal transmission per se between catalytic nucleotide-binding domains on beta-subunits. Therefore, impairment of unisite catalysis and of positive catalytic cooperativity appeared to be linked. This may provide a unifying explanation as to why a series of other, widely separated mis-sense mutations within the catalytic nucleotide-binding domain on F1-beta-subunit, which have been reported to affect unisite catalysis, also impair positive catalytic cooperativity. Linear free energy plots for the ATP-binding step of unisite catalysis demonstrated that beta-Asn-242 and beta-Val-242 mutant enzymes did not suffer any gross disruptive change in structure of the catalytic nucleotide-binding domain, reinforcing the view that impairment of catalysis was due to a localized effect. Such analyses confirmed that six other F1-beta-subunit mutants, previously generated and characterized in this laboratory and thought to have inhibitory side-chain substitutions in the catalytic nucleotide-binding domain, are also devoid of gross structural disruption.     

The properties of hybrid F1-ATPase enzymes suggest that a cyclical catalytic mechanism involving three catalytic sites occurs

Maximal rates of ATP hydrolysis catalyzed by F1-ATPase enzymes are known to involve strong positive catalytic site cooperativity. There are three potential catalytic nucleotide-binding sites on F1. Two important and unanswered questions are (i) whether all three potential catalytic sites must interact cooperatively to yield maximal rates of ATP hydrolysis and (ii) whether a cyclical three-site mechanism operates as suggested by several authors. We have studied these two questions here by measuring the ATPase activities of hybrid enzymes containing normal beta-, gamma-, delta-, and epsilon-subunits together with different combinations of mutant and normal alpha-subunits. The mutant alpha-subunits were derived from uncA401, uncA447, and uncA453 mutant E. coli F1-ATPase, in which positive cooperativity between catalytic sites is strongly attenuated by defined mis-sense mutations. Our data show that three normal catalytic sites are required to interact in order to achieve maximal ATPase rates and suggest that a cyclical mechanism does operate. Hybrid enzyme containing one-third mutant alpha-subunit and two-thirds normal alpha-subunits had substantial but submaximal activity, showing that cooperativity between three sites in a noncyclical fashion, or between pairs of sites, can achieve effective catalysis.     

F1-ATPase with cysteine instead of serine at residue 373 of the alpha subunit.

Escherichia coli strain AN718 contains the alpha S373F mutation in F1F0-ATP synthase which blocks ATP synthesis (oxidative phosphorylation) and steady-state F1-ATPase activity. The revertant strain AN718SS2 containing the mutation alpha C373 was isolated and shown to confer a phenotype of higher growth yield than that of the wild type in liquid medium containing limiting glucose, succinate, or LB. Purified F1 from strain AN718SS2 was found to have 30% of wild-type steady-state ATPase activity and 60% of wild-type oxidative phosphorylation activity. Azide sensitivity of ATPase activity and ADP-induced enhancement of bound aurovertin fluorescence, both of which are lost in alpha S373F mutant F1, were regained in alpha C373 F1. N-Ethylmaleimide (NEM) inactivated alpha C373 F1 steady-state ATPase potently but had no effect on unisite ATPase. Complete inactivation of alpha C373 F1 steady-state ATPase corresponded to incorporation of one NEM per F1 (mol/mol), in just one of the three alpha subunits. NEM-inactivated enzyme showed azide-insensitive residual ATPase activity and loss of ADP-induced enhancement of bound aurovertin fluorescence. The data confirm the view that placement at residue alpha 373 of a bulky amino acid side-chain (phenylalanyl or NEM-derivatized cysteinyl) blocks positive catalytic cooperativity in F1. The fact that NEM inhibits steady-state ATPase when only one alpha subunit of three is reacted suggests a cyclical catalytic mechanism.     

Tightly bound nucleotides affect phosphate binding to mitochondrial F1-ATPase

The interaction of inorganic phosphate with native and nucleotide-depleted F1-ATPase was studied. F1-ATPase depleted of tightly bound nucleotides loses the ability to bind inorganic phosphate. The addition of ATP, ADP, GTP and GDP but not AMP, restores the phosphate binding. The nucleotides affecting the phosphate binding to F1-ATPase are located at the catalytic (exchangeable) site of the enzyme. The phosphate is thought to bind to the same catalytic site where the nucleotide is already bound. It is thought that ADP is the first substrate to bind to F1-ATPase in the ATP synthesis reaction.     

Proton-powered subunit rotation in single membrane-bound F0F1-ATP synthase

Synthesis of ATP from ADP and phosphate, catalyzed by F(0)F(1)-ATP synthases, is the most abundant physiological reaction in almost any cell. F(0)F(1)-ATP synthases are membrane-bound enzymes that use the energy derived from an electrochemical proton gradient for ATP formation. We incorporated double-labeled F(0)F(1)-ATP synthases from Escherichia coli into liposomes and measured single-molecule fluorescence resonance energy transfer (FRET) during ATP synthesis and hydrolysis. The gamma subunit rotates stepwise during proton transport-powered ATP synthesis, showing three distinct distances to the b subunits in repeating sequences. The average durations of these steps correspond to catalytic turnover times upon ATP synthesis as well as ATP hydrolysis. The direction of rotation during ATP synthesis is opposite to that of ATP hydrolysis.     

Directed mutations of the strongly conserved lysine 155 in the catalytic nucleotide-binding domain of beta-subunit of F1-ATPase from Escherichia coli

The amino acid sequence -Gly-X-X-X-X-Gly-Lys- occurs in many, diverse, nucleotide-binding proteins, and there is evidence that it forms a flexible loop which interacts with one or other of the phosphate groups of bound nucleotide. This sequence occurs as -Gly-Gly-Ala-Gly-Val-Gly-Lys- in the beta-subunit of the enzyme F1-ATPase, where it is thought to form part of the catalytic nucleotide-binding domain. Mutants of Escherichia coli were generated in which residue beta-lysine 155, at the end of the above sequence, was replaced by glutamine or glutamate. Properties of the soluble purified F1-ATPase from each mutant were studied. The results showed: 1) replacement of lysine 155 by Gln or Glu decreased the steady-state rate of ATP hydrolysis by 80 and 66%, respectively. 2) Characteristics of ATP hydrolysis at a single site were not markedly changed in the mutant enzymes, implying that lysine 155 is not directly involved in bond cleavage during ATP hydrolysis or bond formation during ATP synthesis. 3) The binding affinity for MgATP was weakened considerably in the mutants (Lys much much greater than Gln greater than Glu), whereas the binding affinity for MgADP was affected only mildly (Lys = Gln greater than Glu), suggesting that lysine 155 interacts with the gamma-phosphate of ATP bound at a single high affinity catalytic site. 4) The major determinant of inhibition of steady-state ATPase turnover rate in the mutant enzymes was an attenuation of positive catalytic cooperativity. 5) The data are consistent with the idea that during multisite catalysis residue 155 of beta-subunit undergoes conformational movement which changes substrate and product binding affinities.     

Significant quantities of endogenous GDP and ADP are present on catalytic sites of the F1-ATPase isolated from M. lysodeikticus in the absence of added nucleotides.

The F1-ATPase from Micrococcus lysodeikticus is isolated in the absence of exogenous nucleotides. After removing loosely bound nucleotides from the isolated enzyme by gel permeation chromatography, analysis for tightly bound nucleotides revealed in 14 experiments 0.4 +/- 0.1 mol ADP, 0.5 +/- 0.2 mol GDP, and 0.8 +/- 0.2 mol ATP per mol of F1. Incubation of the isolated enzyme with Mg2+ or Ca2+ did not alter the endogenous nucleotide composition of the enzyme, indicating that endogenous ATP is not bound to a catalytic site. Incubation of the enzyme with P(i) decreased the amount of tightly bound ADP and GDP but did not effect the ATP content. Hydrolysis of MgATP in the presence of sulfite raised the tightly bound ADP and lowered tightly bound GDP on the enzyme. In the reciprocal experiment, hydrolysis of MgGTP in the presence of sulfite raised tightly bound GDP and lowered tightly bound ADP. Turnover did not affect the content of tightly bound ATP on the enzyme. These results suggest that endogenous ADP and GDP are bound to exchangeable catalytic sites, whereas endogenous ATP is bound to noncatalytic sites which do not exchange. The presence of endogenous GDP on catalytic sites of isolated F1 suggests that the F0F1-ATP synthase of M. lysodeikticus might synthesize both GTP and ATP under physiological conditions. In support of this hypothesis, we have found that plasma membrane vesicles derived from M. lysodeikticus synthesize [32P]GTP from [32P]P(i) using malate as electron donor for oxidative phosphorylation.     

Changes in the adenine nucleotide content of beef-heart mitochondrial F1 ATPase during ATP synthesis in dimethyl sulfoxide.

Beef-heart mitochondrial F1 ATPase can be induced to synthesize ATP from ADP and inorganic phosphate in 30% Me2SO. We have analyzed the adenine nucleotide content of the F1 ATPase during the time-course of ATP synthesis, in the absence of added medium nucleotide, and in the absence and presence of 10 mM inorganic phosphate. The enzyme used in these investigations was either pretreated or not pretreated with ATP to produce F1 with a defined nucleotide content and catalytic or noncatalytic nucleotide-binding site occupancy. We show that the mechanism of ATP synthesis in Me2SO involves (i) an initial rapid loss of bound nucleotide(s), this process being strongly influenced by inorganic phosphate; (ii) a rebinding of lost nucleotide; and (iii) synthesis of ATP from bound ADP and inorganic phosphate.     

Studies on the mechanism of oxidative phosphorylation. Catalytic site cooperativity in ATP synthesis

Oxidative phosphorylation catalyzed by bovine heart submitochondrial particles appears to exhibit negative cooperativity with respect to [ADP] and positive cooperativity in catalysis. Eadie-Hofstee plots (v/[S]versus v) of the kinetics of oxidative phosphorylation at the variable ADP concentration range of 1-1200 microM were curvilinear and could be analyzed for two apparent KmADP values differing by one order of magnitude, and two apparent Vmax values. The KmADP values with either NADH or succinate as the respiratory substrate were in the ranges of 10 and 100 microM, and the Vmax values in nmol of ATP formed X min-1 (mg of protein)-1 were, respectively, 500 and 1840 when NADH was the oxidizable substrate, and 550 and 100 when succinate was the energy source. Site-site cooperativity of the ATP synthase, which is a central feature of current theories for the mechanism of oxidative phosphorylation, has been well-documented for ATP hydrolysis by isolated F1-ATPase, but never before demonstrated for mitochondrial ATP synthesis.     

Energy-dependent transformation of the catalytic activities of the mitochondrial F0 x F1-ATP synthase

The ADP(Mg2+)-deactivated, azide-trapped F0 x F1-ATPase of coupled submitochondrial particles is capable of ATP synthesis being incapable of ATP hydrolysis and ATP-dependent delta muH+ generation [FEBS Lett. (1995) 366, 29-32]. This puzzling phenomenon was studied further. No ATPase activity of the submitochondrial particles catalyzing succinate-supported oxidative phosphorylation in the presence of azide was observed when ATP was added to the assay mixture after an uncoupler. Rapid ATP hydrolysis was detected in the same system when ATP followed by an uncoupler was added. Less than 5% of the original ATPase activity was seen when the reaction (assayed with ATP-regenerating system) was initiated by the addition of ATP to the azide-trapped coupled particles oxidizing succinate either in the presence or in the absence of the uncoupler. High ATP hydrolytic activity was revealed when the reaction was started by the simultaneous addition of the ATP plus uncoupler to the particles generating delta muH+. The energy-dependent conversion of the enzyme into latent uncoupler-activated ATPase was prevented by free ADP (Ki approximately 20 microM) and was greatly enhanced after multiple turnovers in oxidative phosphorylation. The results suggest that the catalytic properties of F0 x F1 are delta muH+-dependent which is in accord with our hypothesis on different conformational states of the enzyme participating in ATP synthesis or hydrolysis.     

Energy-dependent transformation of F0.F1-ATPase in Paracoccus denitrificans plasma membranes

F(0).F(1)-ATP synthase in tightly coupled inside-out vesicles derived from Paracoccus denitrificans catalyzes rapid respiration-supported ATP synthesis, whereas their ATPase activity is very low. In the present study, the conditions required to reveal the Deltamu(H+)-generating ATP hydrolase activity of the bacterial enzyme have been elucidated. Energization of the membranes by respiration results in strong activation of the venturicidin-sensitive ATP hydrolysis, which is coupled with generation of Deltam?(H+). Partial uncoupling stimulates the proton-translocating ATP hydrolysis, whereas complete uncoupling results in inhibition of the ATPase activity. The presence of inorganic phosphate is indispensable for the steady-state turnover of the Deltam?(H+)-activated ATPase. The collapse of Deltam?(H+) brings about rapid deactivation of the enzyme, which has been subjected to pre-energization. The rate and extent of the deactivation depend on protein concentration, i.e. the more vesicles are present in the assay mixture, the higher the rate and extent of the deactivation is seen. Sulfite and the ADP-trapping system protect ATPase against the Deltam?(H+) collapse-induced deactivation, whereas phosphate delays the rate of deactivation. A low concentration of ADP (<1 microm) increases the rate of deactivation. Taken together, the results suggest that latent proton-translocating ATPase in P. denitrificans is kinetically equivalent to the previously characterized ADP(Mg2+)-inhibited, azide-trapped bovine heart mitochondrial F(0).F(1)-ATPase (Galkin, M. A., and Vinogradov, A. D. (1999) FEBS Lett. 448, 123-126). A Deltam?(H+)-sensitive mechanism operates in P. denitrificans that prevents physiologically wasteful consumption of ATP by F(0).F(1)-ATPase (synthase) complex when the latter is unable to maintain certain value of Deltam?(H+).     

Catalytic site forms and controls in ATP synthase catalysis

A suggested minimal scheme for substrate binding by and interconversion of three forms of the catalytic sites of the ATP synthase is presented. Each binding change, that drives simultaneous interchange of the three catalytic site forms, requires a 120 degrees rotation of the gamma with respect to the beta subunits. The binding of substrate(s) at two catalytic sites is regarded as sufficing for near maximal catalytic rates to be attained. Although three sites do not need to be filled for rapid catalysis, during rapid bisite catalysis some enzyme may be transiently present with three sites filled. Forms with preferential binding for ADP and P(i) or for ATP are considered to arise from the transition state and participate in other steps of the catalysis. Intermediate forms and steps that may be involved are evaluated. Experimental evidence for energy-dependent steps and for control of coupling to proton translocation and transition state forms are reviewed. Impact of relevant past data on present understanding of catalytic events is considered. In synthesis a key step is suggested in which proton translocation begins to deform an open site so as to increase the affinity for ADP and P(i), that then bind and pass through the transition state, and yield tightly bound ATP in one binding change. ADP binding appears to be a key parameter controlling rotation during synthesis. In hydrolysis ATP binding to a loose site likely precedes any proton translocation, with proton movement occurring as the tight site form develops. Aspects needing further study are noted. Characteristics of the related MgADP inhibition of the F(1) ATPases that have undermined many observations are summarized, and relations of three-site filling to catalysis are assessed.     

The Na(+)-translocating F(1)F(0) ATP synthase of Propionigenium modestum: mechanochemical insights into the F(0) motor that drives ATP synthesis

The ATP synthase of Propionigenium modestum encloses a rotary motor involved in the production of ATP from ADP and inorganic phosphate utilizing the free energy of an electrochemical Na(+) ion gradient. This enzyme clearly belongs to the family of F(1)F(0) ATP synthases and uses exclusively Na(+) ions as the physiological coupling ion. The motor domain, F(0), comprises subunit a and the b subunit dimer which are part of the stator and the subunit c oligomer acting as part of the rotor. During ATP synthesis, Na(+) translocation through F(0) proceeds from the periplasm via the stator channel (subunit a) onto a Na(+) binding site of the rotor (subunit c). Upon rotation of the subunit c oligomer versus subunit a, the occupied rotor site leaves the interface with the stator and the Na(+) ion can freely dissociate into the cytoplasm. Recent experiments demonstrate that the membrane potential is crucial for ATP synthesis under physiological conditions. These findings support the view that voltage generates torque in F(0), which drives the rotation of the gamma subunit thus liberating tightly bound ATP from the catalytic sites in F(1). We suggest a mechanochemical model for the transduction of transmembrane Na(+)-motive force into rotary torque by the F(0) motor that can account quantitatively for the experimental data.     

F0F1-ATPase/synthase is geared to the synthesis mode by conformational rearrangement of epsilon subunit in response to proton motive force and ADP/ATP balance

The epsilon subunit in F0F1-ATPase/synthase undergoes drastic conformational rearrangement, which involves the transition of two C-terminal helices between a hairpin "down"-state and an extended "up"-state, and the enzyme with the up-fixed epsilon cannot catalyze ATP hydrolysis but can catalyze ATP synthesis (Tsunoda, S. P., Rodgers, A. J. W., Aggeler, R., Wilce, M. C. J., Yoshida, M., and Capaldi, R. A. (2001) Proc. Natl. Acad. Sci. U. S. A. 98, 6560-6564). Here, using cross-linking between introduced cysteine residues as a probe, we have investigated the causes of the transition. Our findings are as follows. (i) In the up-state, the two helices of epsilon are fully extended to insert the C terminus into a deeper position in the central cavity of F1 than was thought previously. (ii) Without a nucleotide, epsilon is in the up-state. ATP induces the transition to the down-state, and ADP counteracts the action of ATP. (iii) Conversely, the enzyme with the down-state epsilon can bind an ATP analogue, 2',3'-O-(2,4,6-trinitrophenyl)-ATP, much faster than the enzyme with the up-state epsilon. (iv) Proton motive force stabilizes the up-state. Thus, responding to the increase of proton motive force and ADP, F0F1-ATPase/synthase would transform the epsilon subunit into the up-state conformation and change gear to the mode for ATP synthesis.     

Synthesis and hydrolysis of ATP and the phosphate-ATP exchange reaction in soluble mitochondrial F1 in the presence of dimethylsulfoxide.

In medium containing 40% dimethylsulfoxide, soluble F1 catalyzes the hydrolysis of ATP introduced at concentrations lower than that of the enzyme [Al-Shawi, M.K. & Senior, A.E. (1992), Biochemistry 31, 886-891]. At this concentration of dimethylsulfoxide, soluble F1 also catalyzes the spontaneous synthesis of a tightly bound ATP to a level of approximately 0.15 mol per mol F1 [Gómez-Puyou, A., Tuena de Gómez-Puyou, M. & de Meis, L. (1986) Eur. J. Biochem. 159, 133-140]. The mechanisms that allow soluble F1 to carry out these apparently opposing reactions were studied. The rate of hydrolysis of ATP bound to F1 under uni-site conditions and that of synthesis of ATP were markedly similar, indicating that the two ATP molecules lie in equivalent high affinity catalytic sites. The number of enzyme molecules that have ATP at the high affinity catalytic site under conditions of synthesis or uni-site hydrolysis is less than the total number of enzyme molecules. Therefore, it was hypothesized that when the enzyme was treated with dimethylsulfoxide, a fraction of the F1 population carried out synthesis and another hydrolysis. Indeed, measurements of the two reactions under identical conditions showed that different fractions of the F1 population carried out simultaneously synthesis and hydrolysis of ATP. The reactions continued until an equilibrium level between F1.ADP + Pi <--> F1.ATP was established. At equilibrium, about 15% of the enzyme population was in the form F1.ATP. The DeltaG degrees of the reaction with 0.54 microM F1, 2 mM Pi and 10 mM Mg2+ at pH 6.8 was -2.7 kcal.mol-1 in favor of F1.ATP. The DeltaG degrees of the reaction did not exhibit important variations with Pi concentration; thus, the reaction was in thermodynamic equilibrium. In contrast, DeltaG degrees became significantly less negative as the concentration of dimethylsulfoxide was decreased. In water, the reaction was far to the left. The equilibrium constant of the reaction diminished linearly with an increase in water activity. The effect of solvent is fully reversible. In comparison to other enzymes, F1 seems unique in that solvent controls the equilibrium that exists within an enzyme population. This results from the effect of solvent on the partition of Pi between the catalytic site and the medium, and the large energetic barrier that prevents release of ATP from the catalytic site. In the presence of dimethylsulfoxide and Pi, ATP is continuously hydrolyzed and synthesized with formation and uptake of Pi from the medium. This process is essentially an exchange reaction analogous to the phosphate-ATP exchange reaction that is catalyzed by the ATP synthase in coupled energy transducing membranes.     

Regulation of the F0F1-ATP synthase: the conformation of subunit epsilon might be determined by directionality of subunit gamma rotation

F(0)F(1)-ATP synthase couples ATP synthesis/hydrolysis with transmembrane proton transport. The catalytic mechanism involves rotation of the gamma epsilon c(approximately 10)-subunits complex relative to the rest of the enzyme. In the absence of protonmotive force the enzyme is inactivated by the tight binding of MgADP. Subunit epsilon also modulates the activity: its conformation can change from a contracted to extended form with C-terminus stretched towards F(1). The latter form inhibits ATP hydrolysis (but not synthesis). We propose that the directionality of the coiled-coil subunit gamma rotation determines whether subunit epsilon is in contracted or extended form. Block of rotation by MgADP presumably induces the extended conformation of subunit epsilon. This conformation might serve as a safety lock, stabilizing the ADP-inhibited state upon de-energization and preventing spontaneous re-activation and wasteful ATP hydrolysis. The hypothesis merges the known regulatory effects of ADP, protonmotive force and conformational changes of subunit epsilon into a consistent picture.