The Activity of the ATP Synthase from Escherichia coli Is Regulated by the Transmembrane Proton Motive Force*

The ATP synthase from Escherichia coli was reconstituted into liposomes from phosphatidylcholine/phosphatidic acid. The proteoliposomes were energized by an acid-base transition and a K+/valinomycin diffusion potential, and one second after energization, the electrochemical proton gradient was dissipated by uncouplers, and the ATP hydrolysis measurement was started. In the presence of ADP and Pi, the initial rate of ATP hydrolysis was up to 9-fold higher with pre-energized proteoliposomes than with proteoliposomes that had not seen an electrochemical proton gradient. After dissipating the electrochemical proton gradient, the high rate of ATP hydrolysis decayed to the rate without pre-energization within about 15 s. During this decay the enzyme carried out approximately 100 turnovers. In the absence of ADP and Pi, the rate of ATP hydrolysis was already high and could not be significantly increased by pre-energization. It is concluded that ATP hydrolysis is inhibited when ADP and Pi are bound to the enzyme and that a high Δμ̃H+ is required to release ADP and Pi and to convert the enzyme into a high activity state. This high activity state is metastable and decays slowly when Δμ̃H+ is abolished. Thus, the proton motive force does not only supply energy for ATP synthesis but also regulates the fraction of active enzymes.

Membrane-bound F 0 F 1 -ATPases catalyze ATP synthesis in bacteria, chloroplasts, and mitochondria. The enzymes consist of two major parts: the membrane embedded hydrophobic F 0 part involved in proton translocation across the membrane and the hydrophilic F 1 part containing the nucleotide-and phosphate-binding sites (for review see Refs. [1][2][3]. In the recent years, major progress has been made toward an understanding of the enzymatic mechanism; the structure of the hydrophilic F 1 part has been solved to 2.8 Å resolution (4), and an ATP induced rotation of the ␥ subunit within the ␣ 3 ␤ 3 barrel has been deduced (5,6) and visualized (7). Most recently, the ATPinduced rotation of the c-ring in the F 0 part has also been shown (Refs. 8 and 9, but see also Ref. 10).
The electron transport during respiration or photosynthesis generates a transmembrane electrochemical potential difference of protons, ⌬ H ϩ (energization of the membrane), which F 0 F 1 -ATPases use as driving force for ATP synthesis (11). In the reverse direction, the enzymes hydrolyze ATP and generate a transmembrane ⌬ H ϩ. Whereas in chloroplasts and mitochondria this direction is inhibited by special regulations (for review see Refs. 12 and 13), ATP hydrolysis has an important physiological function in bacteria. The proton motive force, ⌬p, is given by the following equation.
(⌬pH being the transmembrane pH difference and ⌬ the transmembrane electric potential difference). According to the chemiosmotic theory (11), the two components of ⌬p are thermodynamically equivalent, i.e. a change of ⌬pH by one unit will result in the same change of the ATP/ADP equilibrium as a shift of ⌬ by 59 mV (at 23°C). For thylakoid membranes, chromatophores, and some bacterial F 0 F 1 -ATPases, it appears that ⌬pH and ⌬ are also kinetically equivalent, i.e. the rate of ATP synthesis is changed by the same factor when either ⌬ is changed by 59 mV or ⌬pH is changed by one unit (14 -17).
In chloroplasts, photosynthetic bacteria, and mitochondria, it has been shown that ⌬ H ϩ is not only the free energy input for ATP synthesis but also strongly regulates the activity of the F 0 F 1 -ATPases, i.e. ⌬ H ϩ induces a metastable high activity state of the enzyme (18 -23). This phenomenon is called ⌬ H ϩ activation, or only activation and occurs also under physiological conditions (24). This physiological regulation of the activity has to be distinguished from treatments (with lauryl dimethylamine oxide (LDAO), trypsin, or heat, for example) that increase the rate of ATP hydrolysis by partial denaturation of the enzyme and are sometimes also called activation. The fraction of active enzymes increases with increasing ⌬ H ϩ and the two components, ⌬pH and ⌬, appear to be equivalent in this respect in chloroplasts (14). In contrast, in the photosynthetic bacterium Rhodobacter capsulatus, ⌬ is more effective than ⌬pH (23), and there are indications that in mitochondria this might also be the case (21,25). The stability of the metastable active state is influenced by a variety of parameters, like ADP, Mg 2ϩ , and P i concentration, and in chloroplasts F 0 F 1 -ATP synthase also by the redox state of the enzyme (14, 23, 26 -28).
Recently, we reconstituted the ATP synthase from Escherichia coli, EF 0 F 1 , into liposomes and measured ATP synthesis after energization by an acid-base transition. (29,30). Surprisingly, the data gave hints that the E. coli enzyme too might be activated by ⌬ H ϩ. Therefore, we decided to investigate this phenomenon in more detail. Activation by ⌬ H ϩ is detected most easily by measuring the increase of the uncoupled rate of ATP hydrolysis after the enzyme has been exposed to a ⌬ H ϩ * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (31), and with this reconstituted system we were able to carry out these experiments.

EXPERIMENTAL PROCEDURES
Purification and Reconstitution of EF 0 F 1 into Liposomes-E. coli strain DK8 (⌬unc) transformed with plasmid pBWU13 (32) was a gift from Prof. Futai (Osaka, Japan). The bacteria were grown in a minimal medium and EF 0 F 1 was isolated as described (29,32). The enzyme was obtained in a solution containing 10 mM Mes 1 and 10 mM Tricine/NaOH, pH 7.0, 500 M MgCl 2 , 5 mM thioglycerol, 10 g/liter octylglucoside, and 300 g/liter sucrose with a protein concentration between 1 and 2 g/liter. EF 0 F 1 was rapidly frozen and stored in liquid nitrogen. The protein concentration was determined with Amido Black (33), and a molecular mass of 530 kDa was used.
Liposomes from phosphatidylcholine and phosphatidic acid were prepared by dialysis (30), and EF 0 F 1 was reconstituted into these preformed liposomes using Triton X-100 (29,34). The proteoliposomes were obtained in a buffer at pH 8.0 containing 13 mM Tricine, 8 mM succinate, 9 mM NaOH, 0.5 mM KCl, and 4 mM MgCl 2 with a lipid concentration of approximately 8 g/liter and a EF 0 F 1 concentration of 30 nM. The proteoliposomes were concentrated by centrifugation at 265,000 ϫ g for 2 h at 18°C. The pellet was resuspended in a minimal volume of buffer LP (20 mM succinate, 20 mM Tricine, titrated with NaOH to pH 8.0) to get a final concentration of approximately 1 M EF 0 F 1 . KCl was added to a final concentration of 0.5 mM. These concentrated proteoliposomes were kept at room temperature (23°C) for up to 3 days. For energization at lower ⌬ or in the absence of ⌬, KCl was added to final concentrations of 8 and 100 mM, respectively, and the proteoliposomes were incubated at room temperature for 4 h in the presence of 50 M valinomycin prior to measurements.
ATP Hydrolysis in an ATP Regenerating System-The continuous measurement of ATP hydrolysis was carried out in a spectrophotometer at 23°C by detecting the decrease of NADH in an enzymatic coupled assay (35). The reaction medium contained 100 mM Tris-HCl, pH 8.0, 100 mM KCl, 4 mM MgCl 2 , 2.5 mM phosphoenolpyruvate, 18 units/ml pyruvate kinase, 16 units/ml lactate dehydrogenase, and 0.2 mM NADH. For measuring the uncoupled rate of ATP hydrolysis, the reaction medium contained in addition 1 M valinomycin and 1 M nigericin.
ATP Hydrolysis after Energization-Proteoliposomes were energized by an acid-base transition and an additional K ϩ /valinomycin diffusion potential in the presence of 100 M ATP and different concentrations of P i and ADP at room temperature (23°C). Proteoliposomes (10 l) were mixed with 20 l of acidic medium (Ϯ⌬) and incubated for 2 min. Thereafter, ⌬ H ϩ was generated by adding 50 l basic medium (Ϯ⌬) (energization). After 1 s, ⌬ H ϩ was dissipated by adding 10 l of uncoupling medium (Ϯ⌬) containing nigericin to a final concentration of 1 M (uncoupling). The time of uncoupling and addition of [␥-32 P]ATP is the start of 32 P release, and this is t ϭ 0 for all of the following experiments. The uncoupling medium (Ϯ⌬) contained [␥-32 P]ATP (Amersham Pharmacia Biotech) to a specific activity of 40 kBq/ml. The reaction was stopped at different times by adding 40 l of perchloric acid to a final concentration of 0.9 M. The points at t ϭ 0 were obtained by adding perchloric acid to proteoliposomes after the acid base jump and then adding, after about 5 s, uncoupling medium (Ϯ⌬). The released 32 P i was separated from [␥-32 P]ATP by organic solvent extraction of a molybdate-P i complex, and the radioactivity was measured by liquid scintillation counting in a Tri-Carb 2100TR (Canberra Packard). The 32 P i found at t ϭ 0 amounted to about 4% of the total and was subtracted from all data.
Compositions of the buffers were as follows: acidic medium (ϩ⌬) contained 20 mM succinate, 0.5 mM KOH, 2.5 mM MgCl 2 , either 0 or 15 mM NaH 2 PO 4 , 0 Ϯ 150 M ADP, 80 M valinomycin (freshly added from a 10 mM ethanolic solution) titrated to the desired pH values with NaOH (from 4.0 to 8.3); basic medium (ϩ⌬) contained 200 mM Tricine, 100 mM KOH, 2.5 mM MgCl 2 , either 0 or 10 mM NaH 2 PO 4 , 150 M ATP, and ADP concentrations between 0 -100 M titrated to the desired pH values with NaOH (from 8.0 to 8.6); and uncoupling medium (ϩ⌬) was like basic medium (ϩ⌬), except that the ATP concentration was 100 M, and it contained 9 M nigericin (freshly added from a 1 mM ethanolic solution) and [␥-32 P]ATP (360 kBq/ml). For energization with a lower diffusion potential, composition of buffers was as above except that the basic medium and uncoupling medium contained 8 mM KOH instead of 100 mM. For energization in the absence of diffusion potentials, the following media were used: acidic medium (Ϫ⌬) was like acidic medium (ϩ⌬), except that 100 mM KOH/KCl were present instead of 0.5 mM KOH; basic medium (Ϫ⌬) was like basic medium (ϩ⌬); and uncoupling medium (Ϫ⌬) was like uncoupling medium (ϩ⌬).
After 2 min of incubation of the proteoliposomes in acidic medium (Ϯ⌬), the internal pH (pH in ) was assumed to be equal to the pH measured after mixing two parts of the acidic medium with one part of the proteoliposomes. This time was enough to reach the maximal ATP synthesis rate (30). The external pH (pH out ) was the pH value that resulted after mixing three parts of the acidified proteoliposomes with five parts of the basic medium (Ϯ⌬).
Measurement of ATP Synthesis-The rate of ATP synthesis catalyzed by EF 0 F 1 was measured at 23°C as described earlier (30). Proteoliposomes were energized by an acid-base transition and an additional K ϩ /valinomycin diffusion potential. Proteoliposomes (15 l) were mixed with acidic medium S (60 l), and after 2 min of incubation 50 l of the acidified suspension were injected into 950 l of basic medium S. The ATP concentration was monitored continuously with luciferin/luciferase (Merlin) in a luminometer (LKB 1250). The following media were used in the acid-base transition: acidic medium S contained 20 mM succinate, 0.6 mM KOH, 2.5 mM MgCl 2 , 10 mM NaH 2 PO 4 , 0.1 mM ADP, 20 M valinomycin (freshly added) titrated to pH 4.5 with NaOH; and basic medium S contained 200 mM Tricine, 130 mM KOH, 2.5 mM MgCl 2 , 10 mM NaH 2 PO 4 , 0.1 mM ADP, titrated to pH 8.8 with NaOH. After mixing, pH in was 4.7, and pH out was 8.8.

RESULTS
EF 0 F 1 was reconstituted into liposomes at a concentration of 30 nM. The proteoliposomes were concentrated by centrifugation to about 1 M EF 0 F 1 and stored at room temperature for up to 3 days. The rate of ATP synthesis was measured before and after centrifugation and on every following day. Centrifugation caused about 45% loss of ATP synthesis activity, i.e. from 65 to 35 s Ϫ1 . For comparison with our previous ATP synthesis data, all the following enzyme concentrations were corrected by this factor. A correction factor was also applied to compensate for the 10% loss of ATP synthesis activity observed after 3 days of storage.
ATP hydrolysis catalyzed by proteoliposomes after centrifugation was measured with an ATP regenerating system in the absence and presence of uncouplers (nigericin and valinomycin). At 4 mM ATP, the steady state rate in the absence of uncouplers was 40 s Ϫ1 ; it increased after addition of valinomycin to 80 s Ϫ1 and after addition of nigericin to 210 s Ϫ1 . The 5-fold increase of the rate indicates that in the absence of uncouplers, proton transport-coupled ATP hydrolysis generates ⌬ H ϩ across the membrane, and its back pressure limits the rate. When the proton permeability of the membrane was increased by the uncouplers nigericin and valinomycin, this back pressure was released, and the rate was increased. The stimulation by a factor 5 indicates a good coupling between ATP hydrolysis and proton transport, i.e. a good efficiency of reconstitution of the enzyme into the liposomes. The same experiment was carried out with 100 M ATP. In this case, the coupled rate was 15 s Ϫ1 and was also increased 5-fold after uncoupling.
Activation of ATP Hydrolysis by ⌬ H ϩ -To investigate whether reconstituted EF 0 F 1 can be activated by ⌬ H ϩ, the experiments were carried out as follows. EF 0 F 1 proteoliposomes were incubated for 2 min in acidic medium, and ⌬ H ϩ was generated by addition of the basic medium (⌬/⌬pH jump). One second after the ⌬/⌬pH jump, the ⌬ H ϩ was collapsed by addition of an uncoupler (nigericin), and, simultaneously, ATP hydrolysis was started by adding [␥-32 P]ATP. ATP hydrolysis was stopped after various times between 1 and 20 s by addition of perchloric acid and the released P i was measured. In the following, the rate of ATP hydrolysis measured in this way is called pre-energized ATP hydrolysis.
For comparison ATP hydrolysis was measured under identical conditions but without pre-energization. In this case, after incubation in the acidic medium, the uncoupler was added simultaneously with the basic medium, so that no ⌬ H ϩ was generated. After one second, [␥-32 P]ATP was added, and ATP hydrolysis was assayed as described above. In the following, the rate of ATP hydrolysis measured under these conditions, i.e. without pre-energization, is called the basal rate. Both the basal and the pre-energized rates of ATP hydrolysis are assayed under the same experimental conditions (composition of medium, substrates, uncouplers, inhibitors etc.) except for the enzyme that has been exposed to ⌬ H ϩ or left untreated. Notice that both the pre-energized and the basal rates of ATP hydrolysis were measured under uncoupled conditions.
In a first set of experiments the effect of pre-energization on ATP hydrolysis was measured in the presence of 10 mM P i for ADP concentrations between 20 and 100 M (Fig. 1, A-C) and in the absence of ADP and P i (Fig. 1D). Throughout these experiments, ⌬ H ϩ was the same during pre-energization, with pH in ϭ 4.9, pH out ϭ 8.3 (⌬pH ϭ 3.4), and a [K ϩ ] out /[K ϩ ] in ratio of 100 mM/0.5 mM, equivalent to a Nernst potential of 134 mV. The data for the pre-energized and basal rates are shown in filled and open circles respectively, and the rates of ATP hydrolysis are given by the sloped lines.
In the presence of 100 M ADP (Fig. 1A), the rate of ATP hydrolysis was increased from 3 to 26 s Ϫ1 by pre-energization, i.e. an activation factor of about 9 was obtained. The high rate was observed only transiently and decayed to a level close to the basal one within about 15 s. When the ADP concentration in the reaction medium was lowered to 50 M (Fig. 1B) or to 20 M (Fig.  1C), the pre-energized rates were not affected. In contrast, the basal rates obtained were higher for lower ADP concentrations, and this resulted in a lower activation factor (6.4 and 3.6 for 50 and 20 M ADP, respectively). In the absence of added ADP and P i (Fig. 1D), the pre-energized and basal rates were nearly the same and were about two times higher compared with the preenergized rates in the presence of ADP and P i . Under pre-energized conditions, the kinetics of ATP hydrolysis show a small lag (Fig. 1, A-C). The most likely explanation for this lag is a delay in reaching a complete dissipation of ⌬ H ϩ under these experimental conditions. In such cases the rates in Fig. 1 refer to the highest slope of the curves.
These results can be summarized as follows: (i) In the presence of ADP and phosphate, the rate of ATP hydrolysis is low but it can be increased by pre-energization. (ii) Once ⌬ H ϩ is dissipated the high activity decays back to the basal activity within 10 to 20 s. (iii) In the absence of ADP and P i the enzyme activity is already high and cannot be further increased by pre-energization.
Throughout these experiments, ATP was always present during pre-energization, being present in the basic medium. In some experiments, ATP was omitted from the basic medium, and, correspondingly, the ATP concentration in the uncoupling medium was increased to reach the same final concentration of 100 M ATP in the reaction medium. The same basal rates and pre-energized rates were obtained compared with the measurements shown in Fig. 1A. This indicates that the presence of ATP during pre-energization is not required for activation.
ADP and P i Inhibition of the Basal ATP Hydrolysis Rate-In further experiments, we measured the effects of increasing concentrations of ADP and P i on the basal rate of ATP. Fig. 2 shows the results obtained in the presence of either 10 mM P i or 100 M ADP alone and in the presence and absence of both. The two substrates alone inhibited the ATP hydrolysis rate, to 23 and 42% of the maximal rate, respectively. The inhibition was FIG. 1. ATP hydrolysis with and without pre-energization. EF 0 F 1 proteoliposomes were energized by an acidbase transition. A ⌬pH of 3.4 units (pH in ϭ 4.9, pH out ϭ 8.3) and a ⌬ of about 134 mV ([K ϩ ] out ϭ 100 mM, [K ϩ ] in ϭ 0.5 mM in the presence of valinomycin) were generated. After 1 s, nigericin and [␥-32 P]ATP were added to dissipate ⌬ H ϩ and to start the release of 32 P i . q, pre-energized rate. In each panel, the left arrow indicates the acid-base transition (addition of basic medium), and the right arrow represents the addition of nigericin and [␥-32 P]ATP (addition of uncoupling medium). EF 0 F 1 and ATP concentrations were 54 nM and 100 M, respectively. The reaction was stopped at various times by adding perchloric acid, and samples were analyzed for 32 P i content. As controls, the same experiments were carried out except that nigericin was present during the acidbase transition, i.e. no ⌬ H ϩ was generated (basal rate, E). The solid lines represent least square fits of the data. The numbers at the slopes give the rates of ATP hydrolysis in mol ATP/mol EF 0 F 1 /s, i.e. in s Ϫ1 . In all cases, the maximum slope is indicated. ADP and P i concentrations were as follows: A, 100 M ADP, 10 mM P i ; B, 50 M ADP, 10 mM P i ; C, 20 M ADP, 10 mM P i ; D, no added ADP and P i . much more pronounced when both substrates were present together, leaving 2.5% of the maximal rate. Note that the experimental conditions for the measurements shown in Fig. 2 are similar to those of Fig. 1 (A and D, open circles). The only difference is that in this case acidic medium, basic medium, and uncoupling medium were first mixed, and then the proteoliposomes were added. Because the rates are the same within experimental error, it appears that the inhibitory effect of ADP and P i was not caused by preincubation in the acidic medium.
In Fig. 3A the basal rate of ATP hydrolysis is shown as a function of ADP concentration in the presence of 10 mM P i . Because 0.5% ADP was found as a contaminant in the commercial ATP, 0.5 M was the ADP concentration when no ADP was added to the reaction medium. The half-maximal inhibition is observed at 10 M ADP. ATP hydrolysis was also measured as a function of P i concentration in the presence of 100 M ADP. These data are shown in Fig. 3B, indicating a half-maximal inhibition at 470 M P i . We conclude that the basal rate observed in Fig. 1A could not be significantly decreased by a further increase in ADP or P i concentration.
ATP Hydrolysis after Exposure to Various ⌬ H ϩ Values-In additional experiments we varied the ⌬ H ϩ values generated in the pre-energization step. First we checked whether one of the two components of ⌬ H ϩ, ⌬ or ⌬pH alone, was responsible for the activation phenomenon. ATP hydrolysis was measured after pre-energization with either ⌬ or ⌬pH alone. The experiments were carried out in the same way as those presented in Fig. 1B, i.e. in the presence of 50 M ADP and 10 mM P i . For ⌬pH ϭ 0, pH in and pH out were both 8.3, and the [K ϩ ] out /[K ϩ ] in ratio was 100 mM/0.5 mM, equivalent to a Nernst potential of 134 mV. For ⌬ ϭ 0, [K ϩ ] out and [K ϩ ] in were both 100 mM with pH in ϭ 4.9, pH out ϭ 8. 3 (⌬pH ϭ 3.4). In both cases no signifi-cant activation could be seen, indicating that neither ⌬ nor ⌬pH alone was responsible for the activation seen in Fig. 1.
We also checked whether ⌬pH and ⌬ have the same effect on the activation by varying the relative contribution of ⌬ and ⌬pH at the same level of ⌬ H ϩ(⌬p Ϸ 240 mV). Fig. 4 shows the effect of pre-energization on ATP hydrolysis when, at constant ⌬ H ϩ, the relative contribution of ⌬pH and ⌬ were varied. It can be seen that no activation is observed for ⌬ ϭ 0; however, at the same ⌬ H ϩ, the pre-energized rate was increased by a factor 2 when ⌬ was increased to about 80 mV. Therefore, ⌬ appears to be more effective than ⌬pH in driving the activation step.

DISCUSSION
Most investigations on the catalytic mechanism of the ATP synthase from E. coli have been carried out with the EF 1 part (for review see Ref. 1). In this work, we have investigated ATP hydrolysis catalyzed by EF 0 F 1 reconstituted into liposomes. First, we have shown that the steady state rate of ATP hydrolysis was limited by the back pressure of ⌬ H ϩ, indicating that the proteoliposomes were well coupled. Correspondingly, these proteoliposomes catalyzed high rates of ATP synthesis.
A strong inhibition of the rate of ATP hydrolysis by ADP (Fig.  3A) and P i (Fig. 3B) was observed. Because all measurements reported in this work were carried out in the presence of 2.5 mM Mg 2ϩ , ADP was present almost completely as MgADP, which suggests that it is the binding of this complex to EF 0 F 1 that inhibits ATP hydrolysis. For EF 1 , the occurrence of an inhibitory MgADP catalytic site complex has been reported by Hyndman et al. (36), similar to what was found for soluble portions of the chloroplast (37) and mitochondrial F 1 F 0 ATP synthase (38 -40). These authors have proposed that this inhibitory complex could be responsible for EF 1 occurring as an heterogeneous mixture of active and inactive enzyme, as initially suggested by kinetic studies (41).
Binding of P i to the E. coli enzyme has not been seen in competition studies with ATP using mutagenized tryptophanes as reporters of nucleotide binding to catalytic and non catalytic sites (42,43). On the other hand, relatively low concentrations of P i (2 mM) were shown to inhibit uni-and multi-site ATP hydrolysis in EF 1 (41) and P i binding to EF 1 (K D ϭ 50 M) has been shown to change the rate of trypsin cleavage of the ⑀ subunit (44). Possibly, the binding of P i is sensitive to the particular pattern of nucleotide occupancy of the enzyme. Because a small amount of ADP, as a contaminant of ATP, was present in our experiments, it is also possible that inhibition by P i occurs only when ADP is present. The data of Mendel-Hartvig and Capaldi (44), considered in the light of the present work, do suggest that the ⑀ subunit could be considered to play a main role in the ⌬ H ϩ-induced inhibited-to-active transition. It is also interesting that, within the complexity of P i effects on the activity of various ATP synthases to be found in the literature, there are reports of an ADP-requiring P i inhibition of the mitochondrial F 1 (45) and of the chloroplasts F 0 F 1 (27,46).
We then investigated the effect of pre-energization on ATP hydrolysis. The rate of ATP hydrolysis was significantly increased compared with the rate without energization (Fig. 1A). Remarkably, this activation effect was only evident in the presence of inhibitory ADP and P i . In their absence, the enzyme was able to catalyze high rates of ATP hydrolysis that could not be increased significantly by a ⌬ H ϩ (Fig. 1D).
A simple way to explain these observations is that membrane-bound EF 0 F 1 exists in at least two different states: an inhibited state, E i , with bound ADP and P i , and an active state, E a . Energization of the proteoliposomes by ⌬ H ϩ induces the transition from E i to E a . The high rates of ATP hydrolysis are detected only after dissipation of ⌬ H ϩ by uncoupling, because the back pressure of ⌬ H ϩ inhibits the ATP hydrolysis reaction. The active state is metastable and decays after dissipation of ⌬ H ϩ to the inhibited state within about 15 s. During this time, the enzyme carries out about 100 turnovers. This can be summarized by the following scheme.
According to this scheme, the E a form is able to catalyze high rates of both ATP hydrolysis and synthesis, the direction being determined by the phosphate potential and membrane energization. It is not yet clear what causes the release of ADP and P i and whether the conformational change results in an increase of the dissociation constants of bound ADP and P i or a ⌬ H ϩdriven ATP synthesis. In chloroplasts it has been shown that, after energization, ADP is released from the enzyme without previous phosphorylation (19,47,48).
At 100 M ATP, 100 M ADP and 10 mM P i were enough to inhibit the enzyme almost to completion (about 2% of the maximal activity was left; Figs. 1 and 2). One might ask whether the maximum ⌬ H ϩ applied in this work (⌬pH ϭ 3.4; ⌬ Ϸ 134 mV; Fig. 1) was enough to activate all enzymes. Comparison of D and A of Fig. 1 shows that the highest hydrolysis rate was 64 s Ϫ1 (ϩ⌬ H ϩ, absence of ADP and P i ), whereas the hydrolysis rate observed after ⌬ Hϩ activation in the presence of the highest ADP/P i concentration was 26 s Ϫ1 , i.e. only 40%. This lower rate might be due to two different effects: (i) the applied ⌬ H ϩ activates all enzymes; in this case the lower activity can be attributed to competitive product inhibition of the ATP hydrolysis; and (ii) the applied ⌬ H ϩ activates only 40% of the enzymes, and, correspondingly, the rate is lower.
It is striking that the activation of EF 0 F 1 requires ⌬ H ϩ values that are higher than the thermodynamic threshold for ATP synthesis under these conditions. Taking ⌬G 0Ј P ϭ 35 kJ mol Ϫ1 (49) and [ATP] ϭ 100 M, [ADP] ϭ 50 M, [P i ] ϭ 10 mM, the thermodynamic threshold of ⌬ H ϩ for ATP synthesis is 12 kJ mol Ϫ1 (⌬p ϭ 125 mV) for n H ϩ ϭ 4 or 16 kJ mol Ϫ1 (⌬p ϭ 166 mV) for n Hϩ ϭ 3. Fig. 4 shows that no activation took place up to ⌬pH ϭ 4.1. On the other hand, an increase of ATP hydrolysis by a factor of 2 was seen at the same ⌬ H ϩ but in the presence of a ⌬ (⌬pH ϭ 2.9, ⌬ Ϸ 70 mV). It is interesting to compare these data with those of Fischer and Grä ber (30), who found that no ATP synthesis could be detected at ⌬pH ϭ 4.1 but that FIG. 4. ATP hydrolysis with and without pre-energization under different ⌬˜ H ؉ conditions. EF 0 F 1 proteoliposomes were energized by an acid-base transition. After 1 s, nigericin and [␥-32 P]ATP were added to dissipate ⌬ Hϩ and to start the release of 32 P i . The left arrow indicates the acid-base transition (addition of basic medium), and the right arrow represents the addition of nigericin and [␥-32 P]ATP (addition of uncoupling medium). EF 0 F 1 and ATP concentrations were 54 nM and 100 M, respectively. The reaction was stopped at various times by adding perchloric acid, and samples were analyzed for 32 P i content. Solid circles indicate data after pre-energization, and open circles indicate data without pre-energization. The solid lines represent least square fits of the data, and the numbers at the slope give the rate of ATP hydrolysis in s Ϫ1 . A, pH in ϭ 4.7, pH out ϭ 8.8, ⌬ Ϸ 0, i.e. ⌬p Ϸ 240 mV. B, pH in ϭ 5.9, pH out ϭ 8. about 20% of the maximal rate could be observed at ⌬pH ϭ 2.9, ⌬ Ϸ 70 mV. Both sets of data could fit together if it is assumed that the ATP synthesis reaction was limited by the activation process and if the activation process is more sensitive to ⌬ than to ⌬pH. This last point awaits the support of more experimental data. For the moment, though, we consider it as very likely also in the light of data previously obtained in the photosynthetic bacterium R. capsulatus, indicating a much higher efficiency of ⌬ relative to ⌬pH in driving the activation process (23). It is even possible that in E. coli the presence of a ⌬ is obligatory for activation.
Activation phenomena have first been found in the chloroplast ATP synthase (18,19,50,51). ATP synthesis/hydrolysis in the light/dark cycles of plants has been shown to be regulated, in addition to a redox reaction of a disulfide bridge in CF 0 F 1 , by the ⌬ H ϩ (24,31,52,53). Similar activation phenomena have been reported for the photosynthetic bacteria R. rubrum (22) and R. capsulatus (23). Activation of the mitochondrial ATP synthase has been shown to involve an inhibitor protein that dissociates from the enzyme in an energy-dependent manner leading to an activated enzyme (Refs. 21, 54, and 55; for a recent review see Ref. 12). More recently, activation by ⌬ H ϩ has been observed also in the absence of the inhibitor protein (56). In this work we have provided evidence that, similar to F 0 F 1 -ATPases from chloroplasts, mitochondria, and photosynthetic bacteria, EF 0 F 1 is strongly regulated by ⌬ H ϩ. Possibly, this type of regulation is a common feature of all F 0 F 1 -ATPases.