The γ-subunit of cyanobacterial and chloroplast ATP synthase, the rotary shaft of F1-ATPase, equips a specific insertion region that is only observed in photosynthetic organisms. This region plays a physiologically pivotal role in enzyme regulation, such as in ADP inhibition and redox response. Recently solved crystal structures of the γ-subunit of F1-ATPase from photosynthetic organisms revealed that the insertion region forms a β-hairpin structure, which is positioned along the central stalk. The structure–function relationship of this specific region was studied by constraining the expected conformational change in this region caused by the formation of a disulfide bond between Cys residues introduced on the central stalk and this β-hairpin structure. This fixation of the β-hairpin region in the α3β3γ complex affects both ADP inhibition and the binding of the ε-subunit to the complex, indicating the critical role that the β-hairpin region plays as a regulator of the enzyme. This role must be important for the maintenance of the intracellular ATP levels in photosynthetic organisms.
FoF1-ATP synthase (FoF1) is ubiquitous in energy-transducing membranes, such as chloroplast and cyanobacterial thylakoid membranes, mitochondrial inner membranes, and bacterial plasma membranes. FoF1 synthesizes ATP from ADP and inorganic phosphate by using the electrochemical proton gradient formed across these membranes via the photosynthetic or respiratory electron transfer reaction [1,2]. When FoF1 catalyzes ATP hydrolysis, protons are transported in the opposite direction across membranes from that for ATP synthesis conditions. FoF1 consists of the membrane-embedded portion Fo and the water-soluble portion F1. Fo works as a proton translocation device and is composed of a, b, and c subunits, with a stoichiometry of a1b2c10–15 [3–9], whereas F1 is the catalytic core for ATP synthesis and hydrolysis, and is composed of five different subunits designated α–ε, with a stoichiometry of α3β3γ1δ1ε1 . The minimum catalytically active complex for F1-ATPase is the α3β3γ complex [11–14], and the catalytic sites reside on each of the three β-subunits at the interface with the α-subunits . The rotary catalysis mechanism was first proposed by Boyer and co-workers based on detailed kinetic analyses of F1-ATPase . Following the crystal structure analysis of F1-ATPase, which clearly indicated that the γ-subunit is a rotary shaft within the α3β3 hexamer , the counterclockwise continuous rotation of the γ-subunit during ATP hydrolysis was observed by single-molecule observation techniques [17,18].
Since ATP is an essential physiological energy resource for all living cells, the cellular ATP level must be controlled in response to changes in the environment, and futile ATP hydrolysis by FoF1 must be prevented. The most common regulatory mechanism of F1-ATPase is ADP-induced inhibition (ADP inhibition), irrespective of the source of the enzyme. MgADP produced by ATP hydrolysis remains at the catalytic site, and strongly inhibits ATP hydrolysis but not ATP synthesis [19–23]. Another inhibitory mechanism, referred to as ε-inhibition, is attributed to the function of the intrinsic ε-subunit, which works as an inhibitor of ATP hydrolysis for bacterial and chloroplast enzymes. In the FoF1 complex, the ε-subunit is located at the bottom of the γ-subunit, and the C-terminal helix-turn-helix domain of the ε subunit of bacterial F1 shows a large conformational change from a retracted form to an extended form in response to external stimuli, such as a change in ATP level [24–27]. This large conformational change is supposed to be the cause of the ε-inhibition [28,29]. Although the ε-subunit of cyanobacteria and chloroplasts appears to inhibit ATP hydrolysis more strongly than the bacterial ε-subunit [30–32], the conformational change at the C-terminal helical region is unlikely to occur in the case of the cyanobacterial ε-subunit . Single-molecule observations of ε-inhibition on rotation of the γ-subunit in cyanobacterial and bacterial F1-ATPase showed that the ε-subunit of cyanobacterial F1 completely stops the rotation , whereas the ε-subunit of bacterial F1 decreases the average rotation speed and increases the pause duration . These inhibitions are important to avoid futile ATP hydrolysis and to ensure efficient ATP synthesis in vivo.
Because of the unique rotating shaft movement, the γ-subunit was expected to function in regulating ATP synthesis/hydrolysis. Cyanobacterial and chloroplast γ-subunits consist of an insertion region comprising 30–40 amino acids, in comparison with the γ-subunits of the bacterial and mitochondrial ATP synthases [36,37]. Although the insertion region of the chloroplast γ-subunit contains nine additional amino acids, including two redox-reactive cysteines [38–40], the insertion region of cyanobacteria lacks this 9-amino-acid portion. In our previous study, we determined the crystal structure of the γ–ε complex of F1-ATPase from the thermophilic cyanobacterium Thermosynechococcus elongatus BP-1 at a resolution of 1.98 Å . According to this structure, the insertion region forms a β-hairpin structure, which is positioned along the central stalk, and the top of the insertion region appears to interact with the DELSEED region of the β-subunit. This negatively charged region is considered to be important for the enzyme activity of FoF1 [42–44]. Furthermore, higher levels of B-factors in the insertion region of the γ-subunit indicate the considerable flexibility of this region . A similar structure is also observed in the γ-subunit of chloroplast FoF1 .
Based on these findings, we expected that the β-hairpin structure of the insertion region plays a regulatory role via certain movement in response to environmental stimuli, although the cyanobacterial γ-subunit does not contain the redox-regulation motif including two cysteines. We therefore prepared a mutant F1-ATPase of T. elongatus BP-1 to restrict the movement of this region by forming a disulfide bond between the introduced Cys residues on the β-hairpin structure and the central stalk of the γ-subunit. Our experiments restricting the β-hairpin structure clearly indicated that the insertion region is essential to control both ADP inhibition and ε-inhibition to prevent futile ATP hydrolysis.
Pyruvate kinase, lactate dehydrogenase, and NADH were purchased from Roche Diagnostics (Basel, Switzerland). Other chemicals were of the highest grade commercially available.
Escherichia coli strains used were DH5α for cloning and BL21(DE3) uncΔ702 [Tcr, ATPase mutant, BL21(DE3) uncΔ702, asnA::Tn10] for expression of the α3β3γ complex of T. elongatus BP-1. The latter strain was a kind gift from Dr. C. S. Harwood (University of Iowa).
Construction of expression plasmids for the α3β3γ complex containing mutant γ-subunit
The expression plasmid for the α3β3γ complex of T. elongatus BP-1, pTR19FR, in which a deca-histidine tag was fused to the N-terminal of the β-subunit, was originally constructed in a previous study . Using the plasmid as a template, expression plasmids for the complex containing the mutant γ-subunit were prepared. First, the γ-subunit-coding region, atpC, was amplified by PCR, and was cloned into the pGEM-T Easy Vector. Then, site-directed mutagenesis was performed by the overlap extension method to introduce the desired two Cys residues into the γ-subunit. The primers used for mutagenesis are shown in Supplemental Table S1. The vector containing mutated atpC was digested with EcoRI and NheI, and the mutated atpC fragment was fused to an EcoRI and NheI site of pTR19FR. Consequently, the expression plasmid for the α3β3γ containing the mutation at the insertion region of the γ-subunit was obtained.
Expression and purification of α3β3γ complex and ε-subunit
Expression and purification of the α3β3γ complex were performed as described previously  with some modifications. E. coli strain BL21 (DE3) unc 702, transformed with the desired plasmid, was cultured in 2xYT medium containing 100 µg/ml ampicillin and 0.2 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 37°C for 20 h. The expressed proteins were purified by Ni-affinity chromatography. The obtained protein complex was then subjected to size-exclusion column chromatography on a Superdex 200 Increase column 10/300 (GE Healthcare, Little Chalfont, U.K.), which was equilibrated with 50 mM HEPES–KOH, pH 8.0, 100 mM KCl, 0.1 mM MgCl2, and 0.1 mM ATP. The elution peak positioned at ∼11 ml at a flow rate of 0.5 ml/min was collected. The obtained complex protein was then flash-frozen in liquid nitrogen and stored at −80°C in 10% glycerol until use. Protein concentrations were determined using the Bradford method (Bio-Rad Protein Assay; Bio-Rad Laboratories, Inc., Hercules, U.S.A.) with bovine serum albumin as a standard.
The ε-subunit was expressed and purified as described elsewhere  with some modifications, and stored at −80°C in 10% glycerol until use.
Determination of the optimal oxidation and reduction conditions
Purified α3β3γ complex containing mutation in the γ-subunit was oxidized or reduced by various reagents (diamide, aldrithiol-2, and CuCl2 for oxidation, and DTT for reduction). The concentrations of the reagents used were 5, 50, and 500 µM for diamide and aldrithiol-2; 0.5, 5, and 50 mM for DTT; and 50 and 100 µM for CuCl2, for 30 and 60 min, at 25, 37, and 50°C. The reaction was then stopped by adding 5% (w/v, final concentration) trichloroacetic acid. After centrifugation, the supernatant was removed and the precipitate was washed away with 80 µl of acetone. The mixture was then centrifuged and air-dried after removing the supernatant. The precipitate was dissolved in AMS-labeling buffer (62.5 mM Tris–HCl, pH 6.8, 2% SDS, 7.5% glycerol, 0.01% BPB, and 2 mM AMS).
Measurement of ATP hydrolysis activity
ATP hydrolysis activities were measured as described previously  using an ATP-regenerating system (50 mM HEPES–KOH, pH 8.0, 100 mM KCl, 2 mM MgCl2, 2 mM phosphoenolpyruvate, 50 µg/ml pyruvate kinase/lactate dehydrogenase, 0.2 mM NADH, and 2 mM ATP) at 25°C. ATP hydrolysis was initiated by adding 2–20 µg of α3β3γ complexes to 1.2 ml of the assay buffer with or without LDAO (lauryldimethylamine-N-oxide) at a final concentration of 0.1% (v/v). The ATP hydrolysis rate after the addition of the enzyme was determined by monitoring the decrease in NADH absorption at 340 nm using a spectrophotometer V-550 (Jasco, Tokyo, Japan). The pH dependence of ATP hydrolysis activity was evaluated by the hydrazine sulfate reduction method . To determine the extent of ε-inhibition, the ε-subunit was added into the ATP hydrolysis assay mixture. The ATP hydrolysis activity, the activation ratio by LDAO, and the extent of inhibition was then determined by the steady-state slope (Supplementary Figures S1–S3) .
Results and discussion
Preparation of locked mutant γ-subunit to restrict the movement of the β-hairpin structure of the insertion region
Based on the structure in the γ–ε complex of F1-ATPase obtained from T. elongatus BP-1 , we prepared mutant γ-subunits whose β-hairpin structure of the insertion region can be locked to the central stalk of the γ-subunit via the formation of a disulfide bond between the introduced Cys residues. Val32 and Ala50 on the N-terminal α-helix, and Ala201, Leu208, and Thr210 on the β-strand of the insertion region were selected as candidate residues for Cys substitution (Figure 1). Accordingly, we prepared three mutants by combination of these Cys substitutions (V32C/L208C, V32C/T210C, and A50C/A201C). Site-directed mutagenesis was performed as described in ‘Experimental Procedures,' and all mutants were successfully expressed in E. coli and accordingly purified.
Position of Cys residues introduced for locking of the insertion region of the γ-subunit.
Determination of the optimal oxidation and reduction conditions
We examined the formation/cleavage conditions of the disulfide bond to lock and unlock the β-hairpin structure of the insertion region by changing the concentrations of oxidizing/reducing reagents, temperatures, and reaction periods. To evaluate the efficiency of disulfide bond formation, non-reducing SDS–PAGE analysis was performed. The locked γ-subunit showed a clear band shift on the gel as a consequence of the conformational constraints (Figure 2A,B). The oxidized and reduced γ-subunits were observed at 30 and 36 kDa, respectively. Similar band shifts caused by the intramolecular disulfide bond formation were also observed in our previous study . Consequently, V32C/L208V, V32C/T210C, and A50C/A201C mutants were mostly oxidized (Figure 2A) by 50 µM diamide, and reduced by 50 mM DTT. It was confirmed that the oxidation and reduction treatments did not affect the ATP hydrolysis activity (Supplementary Figure S4A), the activation ratio by LDAO (Supplementary Figure S4B,C), pH dependence of the activity (Supplementary Figure S5) and the ε-inhibition (Supplementary Figure S6) of the wild-type complex. We therefore concluded that the change in the activity caused by oxidation or reduction in the mutants was simply originating from the locking or unlocking the insertion region. We then investigated the effect of locking of the insertion region on these mutants.
Non-reducing SDS–PAGE and western blotting analysis of unlocked and locked mutants.
Change in ATP hydrolysis activity and the LDAO-activation ratio by unlocking and locking the β-hairpin structure of the insertion region
First, we examined the ATP hydrolysis activity and the activation ratio by LDAO of unlocked and locked mutants at pH 7.0 (Figure 3, white bars) and pH 8.0 (Figure 3, gray bars). LDAO is a nonionic detergent that releases F1-ATPase from the intrinsic regulatory mechanism, ADP inhibition. Therefore, the activation ratio by LDAO is considered to be a good indicator of the extent of ADP inhibition of the enzyme. The activity of A50C/A201C was lower and those of V32C/L208C and V32C/T210C was higher than that of WT, and there was no obvious relationship between the lock of the β-hairpin structure of the insertion region and the change in activity of the mutants. Currently, we cannot explain the reason of the change in ATPase activity caused by these cysteine substitutions. The activation ratio by LDAO of locked mutants of A50C/A201C and V32C/T210C was lower than that of unlocked ones though the activation ratio for V32C/L208C was not affected.
ATP hydrolysis activity and LDAO sensitivity at pH 8 and pH 7 of the locked and unlocked mutants.
The influence of pH on the insertion region must be critical because pH change in chloroplasts appears to be a significant physiological parameter in photosynthetic organisms. Indeed, a drastic pH change is known to occur during light–dark transition, which must markedly affect the activities of various enzymes involved in photosynthesis [49,50]. Although fluctuation of intracellular pH in cyanobacteria has not yet been reported, that in stroma of chloroplasts where F1-ATPase protrudes from thylakoid membranes has been observed between pH 7 (dark) and pH 8 (light) [49,50]. Therefore, assuming that the fluctuation of intracellular pH in cyanobacteria is similar to that of the stroma in chloroplasts, we examined the ATPase activity of the mutants at pH 7.0 and pH 8.0 (Figure 3). The activities of WT and the mutants at pH 7.0 were lower than those at pH 8.0, irrespective of the lock or unlock of the β-hairpin structure of the insertion region. In addition, the LDAO-activation ratio of WT at pH 7.0 was threefold or higher than that at pH 8.0. These results suggest that the decrease in the activities corresponding with lowering pH observed in the WT and most of the mutants was probably caused by ADP inhibition. In contrast, the activation ratio by LDAO of locked A50C/A201C was not pH dependent, although the unlocked one demonstrated a tendency similar to that of the WT, implying that the bottom of the insertion region, where the mutation A50C/A201C was introduced, may be responsible for regulation regarding the pH dependence of the extent of ADP inhibition.
pH dependence of ADP inhibition
To thoroughly characterize the effect of the bottom of the insertion region on the pH dependence of ADP inhibition, we investigated the enzyme activity by a colorimetric method, which quantifies the liberated phosphate. Consequently, similar pH dependence of the activity and the activation ratio by LDAO were observed for WT and the A50C/A201C mutant (Figure 4) as in the results from the coupling assay (Figure 3). Unlocked A50C/A201C showed the same pH dependence as WT, but locked A50C/A201C did not. These results also support the idea that the pH dependence of ADP inhibition is attributable to the conformation of the bottom of the insertion region. In addition, at pH 8.2, the activity of unlocked A50C/A201C was about threefold higher than that of locked A50C/A201C, while the activation ratio by LDAO of unlocked A50C/A201C was less than twice of that of locked A50C/A201C. Therefore, it is unlikely that the decrease in the activity at high pH, caused by locking the β-hairpin structure of the insertion region, results only from ADP inhibition. There are many negatively charged amino acids around the mutation positions (Supplementary Figure S7). These amino acids might contribute to the pH dependence of the activity and ADP inhibition in response to a pH change. According to the structure of chloroplast FoF1-ATP synthase , the mutation positions of A50C/A201C are near the Cys residues, which play roles in the redox regulation of chloroplast F1. Considering these findings together, the redox regulation observed in the chloroplast ATPase is supposed to be due to ADP inhibition and the flexibility of the bottom of the insertion region must provide the required conformational change for this regulation.
pH dependence of ATP hydrolysis activity and LDAO-activation ratios of WT and A50C/A201C.
Effect of locking the insertion region on ε-inhibition
ε-inhibition is conferred by the intrinsic function of the ε-subunit. We expected that the binding of the ε-subunit may affect the conformation around the insertion region because the ε-subunit is located adjacent to the γ-subunit in the structure and a β-strand of the γ-subunit forms a β-sheet structure with the ε-subunit . As shown in Figure 5, the extent of the ε-inhibition of locked mutants (open square), especially V32C/T210C, which is locked at the top of the insertion region, was weaker than that of unlocked mutants (open circle), indicating that the locking of the insertion region has a negative effect on ε-inhibition.
The effects of locking the insertion region on ε inhibition.
Relative conformational change between the insertion region and the central stalk of the γ-subunit caused by the ε-subunit
The locking of the β-hairpin structure of the insertion region had a negative effect on ε-inhibition (Figure 5). In other words, the locking may interfere with the conformational change of the γ-subunit caused by the binding of the ε-subunit. Therefore, we investigated the effect of ε-binding on the disulfide bond formation ability. The disulfide bond formation efficiency must reflect the change of the distance between two introduced Cys residues on the γ-subunit. The ε-subunit itself should not chemically affect the oxidation or reduction because the ε-subunit has no Cys residue. When oxidation or reduction treatment of α3β3γ was carried out in the absence of the ε-subunit, the locked γ-subunit was obtained at a rate of more than 80% by oxidation (Figure 6A and Supplementary Figure S8). In contrast, the locked γ-subunit was hardly obtained by oxidation, except for A50C/A201C, in the presence of the ε-subunit (Figure 6B and Supplementary Figure S8). The disulfide bond formation of V32C/T210C was strongly suppressed by the addition of the ε-subunit. These results indicate that binding of the ε-subunit to the α3β3γ complex induced the relative conformational change between the α-helix of the central stalk and the β-hairpin structure of the insertion region of the γ-subunit, especially at the top of the insertion region. A similar strong effect of the ε-subunit on the insertion region was also observed when the ε-inhibition was evaluated (Figure 5).
The effects of the binding of the ε subunit to α3β3γ on disulfide bond for locking the insertion region.
Considering these findings together, the binding of the ε-subunit induces the conformational change of the γ-subunit, which plays a significant role in ε-inhibition as well. Because all the previously reported structures of the ε-subunits of photosynthetic organisms show a retracted conformation [33,41,45] and are clearly different from the bacterial one, which can reach the DELSEED region of the β-subunit, the roles of the C-terminal α-helical regions of the chloroplast ε-subunit and the bacterial one must be very different from each other in the complex. The β-hairpin structure of the insertion region of the γ-subunit of photosynthetic organisms may therefore accomplish an alternative role of the C-terminal α-helical region of the bacterial ε-subunit in enzyme regulation.
K.A. and T.H. conceived the study, and K.A. and K.K. performed the experiments. K.W. and T.H. supervised the research. K.A., K.I., K.K., S.M., K.W., and T.H. discussed the data. K.A. and T.H. wrote the paper, and K.K. and K.W. commented on the manuscript.
This study was supported by JSPS KAKENHI (MEXT-KAKENHI) [Grant Number 16H06556 to T.H.] and by Dynamic Alliance for Open Innovation Bridging Human, Environment and Materials.
We thank the Biomaterials Analysis Division, Tokyo Institute of Technology for supporting DNA sequencing analysis.
The Authors declare that there are no competing interests associated with the manuscript.