Two specific domains of the γ subunit of chloroplast F_oF₁ provide redox regulation of the ATP synthesis through conformational changes

Significance

Among the F_oF₁-ATP synthase complexes of all organisms, chloroplast F_oF₁ (CF_oCF₁) is a unique enzyme with a redox regulation mechanism; however, the underlying mechanism of redox regulation of the adenosine triphosphate (ATP) synthesis reaction in CF_oCF₁ has not been fully elucidated. By taking advantage of the powerful genetics of Chlamydomonas reinhardtii as a model organism for photosynthesis, we conducted a comprehensive biochemical analysis of the CF_oCF₁ molecule. Here we identify structural determinants for the kinetics of the intracellular redox response and demonstrate that the redox regulation of ATP synthesis is accomplished by the cooperative interaction of two γ subunit domains of CF_oCF₁ that are unique to photosynthetic organisms.

Abstract

Chloroplast F_oF₁-ATP synthase (CF_oCF₁) converts proton motive force into chemical energy during photosynthesis. Although many studies have been done to elucidate the catalytic reaction and its regulatory mechanisms, biochemical analyses using the CF_oCF₁ complex have been limited because of various technical barriers, such as the difficulty in generating mutants and a low purification efficiency from spinach chloroplasts. By taking advantage of the powerful genetics available in the unicellular green alga Chlamydomonas reinhardtii, we analyzed the ATP synthesis reaction and its regulation in CF_oCF₁. The domains in the γ subunit involved in the redox regulation of CF_oCF₁ were mutated based on the reported structure. An in vivo analysis of strains harboring these mutations revealed the structural determinants of the redox response during the light/dark transitions. In addition, we established a half day purification method for the entire CF_oCF₁ complex from C. reinhardtii and subsequently examined ATP synthesis activity by the acid–base transition method. We found that truncation of the β-hairpin domain resulted in a loss of redox regulation of ATP synthesis (i.e., constitutively active state) despite retaining redox-sensitive Cys residues. In contrast, truncation of the redox loop domain containing the Cys residues resulted in a marked decrease in the activity. Based on this mutation analysis, we propose a model of redox regulation of the ATP synthesis reaction by the cooperative function of the β-hairpin and the redox loop domains specific to CF_oCF₁.

The F_oF₁-ATP synthase in chloroplasts and cyanobacteria produces adenosine triphosphate (ATP) from adenosine diphosphate (ADP) and inorganic phosphate. This reaction is driven by a proton electrochemical gradient (ΔμH⁺) across the thylakoid membranes formed by the photosynthetic electron transfer system (1, 2). This enzyme is a large complex (>500 kDa) consisting of a membrane-embedded F₀ component (abb’c_x; x is the variable number among spices) that functions as a proton channel and a membrane-peripheral F₁ component (α₃β₃γδε) that contains three catalytic sites. The ATP synthesis/hydrolysis reactions are catalyzed by the cooperation of these two parts (3). Under light conditions, the ΔμH⁺ formed by the photosynthetic electron transfer system induces rotation of the central stalk (γεc_x) of F_oF₁, and ATP is generated (1, 4). In contrast, under dark conditions, several inhibitory mechanisms, such as Mg-bound ADP (MgADP) inhibition and ε inhibition, inhibit enzyme activity and prevent the occurrence of a wasteful ATP hydrolysis reaction (5–9).

Chloroplast ATP synthase (CF_oCF₁) is equipped with a redox regulation mechanism. The redox regulation system has an important role in light-dependent signaling pathways in photosynthetic organisms, which exploits the change in redox states of critical Cysteine (Cys) residues of target enzyme molecules. For CF_oCF₁, the highly conserved Cys residue pair on the γ subunit exists as dithiols under reducing (photosynthetic) conditions. In contrast, under oxidizing (dark) conditions, the pair forms a disulfide bond (sequence alignment of the F₁-γ subunit is shown in Fig. 1A). The redox-sensitive Cys pair enables the control of CF_oCF₁ activity in response to intracellular redox conditions that fluctuate because of the unstable photosynthetic electron transfer chain activity associated with changing light intensity (10, 11). Therefore, this regulatory system must allow for a more rigorous regulation of CF_oCF₁ activity, resulting in the efficient maintenance of ATP production in response to fluctuating light conditions.

Fig. 1.

Generation of CF₁-γ mutant strains and the redox regulation/response. (A) Sequence alignment of the F₁-γ subunit from different organisms. The sequences were arbitrary selected from the UniProt knowledgebase. Sequences specific to photosynthetic organisms, such as plants, algae, and cyanobacteria, are marked in gray. The DDE motif is colored magenta. Redox-sensitive Cys residues are indicated by asterisks. (B) CF₁-γ mutants designed based on the spinach CF_oCF₁ structure (PDB ID: 6FKF). The β-hairpin and redox loop domains specific to photosynthetic organisms on the γ subunit (green) are colored orange. The right panel shows the amino acid sequences of the generated CF₁-γ mutants. The deleted regions are indicated by “–”.

The redox regulation of CF_oCF₁ has been extensively studied with respect to the ATP hydrolysis reaction catalyzed by the CF₁ component (i.e., F₁-ATPase). For example, F₁-ATPase purified from spinach (Spinacia oleracea) (12–14) and chimeric thermophilic bacterial F₁-ATPase containing the redox regulation portion of spinach F₁-ATPase (15, 16) were used to determine the relationship between ATP hydrolysis activity and redox regulation. Furthermore, we introduced the redox regulation region from spinach into the recombinant F₁-ATPase complex of cyanobacterium Thermosynechococcus elongatus BP-1 to prepare a more robust enzyme model (17). We concluded that the redox regulation of ATP hydrolysis is accomplished through inhibition caused by the occupation of the catalytic site by MgADP (i.e., MgADP inhibition), because the rotation of the γ subunit of chimeric F₁-ATPase is arrested at the MgADP inhibition step under oxidizing conditions. The affinity of the ε subunit, which induces potent inhibition of ATP hydrolysis activity, depends on the redox state of CF₁-γ in the case of spinach F₁-ATPase (18).

However, it remains unclear how redox regulation occurs within the CF_oCF₁ molecule because there are two main technical problems. First, the difficulty in gene modification of spinach hinders progress in the resulting functional analysis. Second, the purification process is complicated and time-consuming. Market spinach leaf is commonly used to purify CF_oCF₁ because of the ease of isolating chloroplasts (19); however, even with the improved protocol, it requires a relatively large-scale apparatus and many purification steps over several days (20).

In this study, we used the unicellular green alga Chlamydomonas reinhardtii as the experimental organism. C. reinhardtii is a model organism used to study photosynthesis and can be genetically modified. We generated strains harboring mutation(s) in CF₁-γ related to redox regulation. Using these strains, we identified the structure and motif determining the rate of light/dark-dependent redox responses in vivo. Moreover, we fused an 8×His-tag to the N-terminus of the CF₁-β subunit for affinity purification and established an experimental procedure to purify the entire CF_oCF₁ complex within half a day. Subsequently, we prepared progeny expressing both the His-tagged CF₁-β and mutated CF₁-γ by crossing with the generated strains {FUD50::8×His-tagged atpB [mating type (mt)⁺] and T1-54::mutated ATPC [mt⁻]}. Finally, we successfully obtained the entire CF_oCF₁ complex containing various mutations in CF₁-γ. Biochemical analyses using the mutated CF_oCF₁ complex revealed that the redox regulation of ATP synthesis is based on the cooperative interaction of two domains specific to the CF₁-γ.

Results

Generation of CF₁-γ Mutant Strains on the Redox Regulation/Response.

We first generated four CF₁-γ mutant strains to examine the redox regulation/response mechanisms of CF_oCF₁ based on the structural information for the specific region of photosynthetic organisms on the CF₁-γ (Fig. 1 A and B). The plasmids carrying wild-type (WT) or mutated ATPC genes encoding CF₁-γ were introduced into the C. reinhardtii T1-54 (ΔATPC) strain (21) (the vectors used are shown in SI Appendix, Fig. S1).

The Asp-Asp-Glu (DDE) motif is a cluster of negatively charged amino acids that is highly conserved in CF₁-γ from photosynthetic organisms (Fig. 1A). The DDE motif structurally positioned at the bottom of the CF₁-γ has been suggested to play key roles in intermolecular interaction with redox factors such as thioredoxin (Trx) (22) and intramolecular interaction with neighboring amino acids for redox regulation (23). In previous studies, the chimeric cyanobacterial F₁-ATPase was generated, in which the redox regulation region derived from spinach CF₁-γ was introduced at the appropriate position. Deleting the residues corresponding to the DDE motif caused the inverse regulation of ATP hydrolysis activity (24) and ATP hydrolysis-driven rotation of the γ subunit (25). To further determine the role of the DDE motif in the redox regulation of the entire CF_oCF₁ complex, the three negatively charged amino acids were replaced by the corresponding neutral amino acids Asn-Asn-Gln (NNQ), and we generated a DDE/NNQ mutant strain.

In addition to the DDE/NNQ mutant strain, we prepared three mutants named ΔRedox_loop, Δβ-hairpin, and ΔInsertion mutant strains as follows (Fig. 1B). The “insertion region” of CF₁-γ is specific to photosynthetic organisms (26) and consists of two structural domains: the redox loop domain and the β-hairpin domain. The redox loop domain is unique to the γ subunit in plants and algae but not cyanobacteria (27). Thus, the ΔRedox_loop mutant is expected to exhibit ATP synthesis activity similar to that of cyanobacterial F_oF₁. The β-hairpin domain is positioned along the central stalk of the F₁ part, and the top of this domain appears to interact with the “DELSEED” region of the catalytic subunit F₁-β. The DELSEED region plays a central role in torque transmission (28–30), and its interaction with the β-hairpin domain results in the suppression of ATP hydrolysis activity in cyanobacterial F₁-ATPase (31, 32). It also enables efficient ATP synthesis by cyanobacterial F₀F₁ (27). To elucidate the involvement of the β-hairpin domain in redox regulation in CF_oCF₁, we generated a Δβ-hairpin mutant (retaining the redox loop domain) as well as a ΔInsertion mutant (lacking both the redox loop and the β-hairpin domains).

Redox Response of the CF₁-γ Mutant Strains In Vivo.

We examined the change in the redox states of the CF₁-γ mutants in vivo caused by the light/dark transition. Dark-adapted C. reinhardtii cells were illuminated with white light (100 μmol photons m⁻² s⁻¹) for 10 min, then transferred into the dark for 20 min. The change in redox state of the CF₁-γ was determined by immunoblotting (SI Appendix, Fig. S2 and Fig. 2). The CF₁-γ of the WT, T1-54::WT ATPC (i.e., complemented T1-54 with the WT ATPC gene), was reduced by light irradiation over 5 min. After transferring to the dark, the WT CF₁-γ reverted to the oxidized form within 5 to 10 min (Fig. 2).

Fig. 2.

In vivo light/dark responses of the redox states in redox-sensitive CF₁-γ mutants. Dark-adapted C. reinhardtii were placed under medium light (100 μmol photons m⁻² s⁻¹) for the indicated period and subsequently transferred to the dark. The reduction level was calculated as the ratio of the reduced form to the total. Immunoblotting images are shown in SI Appendix, Fig. S2. Values are presented as the mean ± SD (n = 3, one measurement for each independently prepared sample).

The DDE/NNQ CF₁-γ showed the same reduction rate as WT CF₁-γ during the dark/light transition. Of note, the reduced form quickly returned to the oxidized form in the dark within 1 min (Fig. 2). This suggests that the DDE/NNQ mutation does not affect the generation of the reduced form but markedly accelerates the oxidation of the reduced form. The Δβ-hairpin CF₁-γ showed a rapid response of the redox state (Fig. 2), and both the reduction and oxidation of the Δβ-hairpin CF₁-γ occurred within 1 min. These results suggest that the β-hairpin domain is an important determinant for the kinetics of the redox response, and particularly, the charge of the DDE motif determines the oxidation efficiency.

Preparation of the Entire CF_oCF₁ Complex from C. reinhardtii.

We established a preparative method to obtain the entire CF_oCF₁ complex from C. reinhardtii. The coding region of the atpB gene encoding CF₁-β was modified to introduce an 8×His-tag (details are shown in SI Appendix, Fig. S3). The DNA was introduced into the chloroplasts of FUD50 cells, in which the chloroplast atpB gene is deleted (33). The entire CF_oCF₁ complex including the 8×His-tagged CF₁-β was purified by Ni-NTA affinity chromatography, followed by size exclusion chromatography (SEC) and confirmed by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (Fig. 3A).

Fig. 3.

Purification of the CF_oCF₁ complex from C. reinhardtii. (A) Reducing SDS-PAGE profile of the thermal denatured (95 °C, 5 min) CF_oCF₁ preparation obtained before and after SEC. 10 μg of protein was loaded in each lane. Parentheses indicate the gene name of the protein. (B) The ΔpH dependency of ATP synthesis. The initial velocity of synthesis (s⁻¹) was calculated from an exponential fit from 0 to 20 s after injection. The time courses are shown in SI Appendix, Fig. S9A. Values are presented as the mean ± SD [n = 3, one measurement for each independently prepared (pretreated) sample].

We confirmed the state of the protein complex and the subunit composition of the purified CF_oCF₁ using SEC, mass spectrometry, N-terminal sequence analysis, and immunoblotting (SI Appendix, Table S1 and Figs. S4 and S5). Next, we evaluated the ATP synthesis activity of the purified CF_oCF₁. The CF_oCF₁ was reconstituted into soybean liposomes, and ATP synthesis activity was measured using the acid–base transition method (Fig. 3B) (27, 34). After chemically reducing CF_oCF₁ with dithiothreitol (DTT), the ATP synthesis activity increased in proportion to ΔpH formed across the lipid bilayer of the proteoliposomes. This is a typical characteristic of ATP synthase as previously reported (35). In contrast, the oxidized CF_oCF₁ showed only slight activity (0.77 ± 0.13 s⁻¹) even when ΔpH was approximately 2.9. Based on these results, we confirmed that purified CF_oCF₁ contained all known subunits and retained redox-regulated ATP synthesis activity.

Reconstitution of the In Vitro System using Trx for Redox Regulation of CF_oCF₁.

The redox state of CF_oCF₁ in chloroplasts is primarily modulated by the chloroplast Trx in a light-dependent manner (13). Trx is a small protein (typically around 12 kDa) that transfers the reducing power generated by the electron transfer chain to various proteins in the chloroplasts. Five Trx isoforms (f1, f2, m, x, and y) are located in the chloroplast of C. reinhardtii. Of these, f1 (CrTrxf1) and f2 (CrTrxf2) may be involved in the redox regulation of CF_oCF₁, because CF_oCF₁ of S. oleracea was mainly reduced by f-type Trx in previous studies (13, 36). Therefore, we reconstituted the redox regulation system for CF_oCF₁ using recombinant CrTrxf1 and CrTrxf2 in vitro (Fig. 4A; purities and activities of these CrTrxs are shown in SI Appendix, Fig. S6). As shown in Fig. 4B, CF₁-γ was barely reduced in the absence of Trx but sufficiently reduced in the presence of CrTrxf1 or CrTrxf2.

Fig. 4.

Reconstruction of the CF_oCF₁ redox regulation system in vitro with recombinant C. reinhardtii Trx-f1 and f2. (A) Immunoblot analysis to detect the redox states of CF₁-γ in the reconstructed system. 30 μM Trx was preincubated in reconstitution buffer containing 3 mM DTT for 10 min at 25 °C. The reduction process was then initiated by adding 5 μM Trx and 0.5 mM DTT to 0.23 μM CF_oCF₁ proteoliposomes. (B) The reduction level of CF₁-γ was determined from the signal intensities shown in A. Values are presented as the mean ± SD [n = 3, one measurement for each independency prepared (pretreated) sample] and fitted to the pseudo-first-order kinetic model. The fitting curve for no addition of CrTrx could not be obtained because a lack of reduction.

Measurement of ATP Synthesis Activities of the CF₁-γ Mutants.

Finally, we evaluated the ATP synthesis activity of CF₁-γ mutants under reduced and oxidized conditions. Each CF₁-γ mutant was purified from a progeny (expressing both the 8×His-tagged CF₁-β and the mutated CF₁-γ, with no ATPC derived from the parental FUD50 strain) resulting from the cross-FUD50::8×His-tagged atpB and the T1-54::mutated ATPC (SI Appendix, Fig. S7). No significant differences were observed in subunit stoichiometry in these mutants, indicating that the substitution or deletion on the γ subunit did not affect the stability of the CF_oCF₁ complex (SI Appendix, Fig. S8). The purified complexes were subjected to an assay using an acid–base transition method promoted by the valinomycin-induced diffusion potential of K⁺, and each of the F_oF₁ preparations was reconstituted into soybean liposomes (34).

The ATP synthesis activity of the WT revealed a typical redox response, in which it was active in the reduced form and inactive in the oxidized form (Fig. 5). The DDE/NNQ mutant also exhibited a similar redox response in activity compared with the WT, suggesting that the DDE motif does not play a central role in the redox regulation of the ATP synthesis reaction. In contrast, the Δβ-hairpin mutant was constitutively active irrespective of redox changes. Moreover, the activity of the ΔRedox_loop mutant was as low as that of the oxidized WT; however, the ΔInsertion mutant lacking both the redox loop and the β-hairpin domains exhibited sufficient activity, although lower compared with that of the reduced WT form. These results indicate that only the redox loop, which is located at the bottom of the γ subunit, does not modulate activity, whereas the β-hairpin, which interacts with the DELSEED region of CF₁-β, induces the inactive form corresponding to the oxidized state of the enzyme.

Fig. 5.

ATP synthesis activities of redox-sensitive mutants (Left) and redox insensitive mutants (Right). Before acidification, the proteoliposomes of the CF_oCF₁ were incubated with 50 mM DTT at 30 °C for 2 h in reconstruction buffer (pH 8.0) to reduce CF₁-γ. The redox states of CF₁-γ and the time courses for ATP synthesis are shown in SI Appendix, Fig. S13. Values are presented as the mean ± SD [n = 4, one measurement for each independency prepared (pretreated) sample]. *** indicates P-value < 0.001; paired t test (Left) and unpaired t test (Right).

Discussion

The In Vivo Light/Dark Responses of the Redox States of CF_oCF₁.

The light/dark responses of the CF₁-γ mutant strains from Arabidopsis thaliana have been studied previously (22), and the negatively charged amino acids found uniquely in photosynthetic organisms were necessary for the redox response. In the present study, we demonstrated that CF₁-γ of the DDE/NNQ mutant and the Δβ-hairpin mutant strains exhibited typical redox responses; however, the kinetics of the redox responses were different compared with that of the WT (Fig. 2). There are two possibilities for the changes in kinetics. First, the mutations in CF₁-γ affect the interaction with reducing and oxidizing factors for CF₁-γ, such as the Trxs and Trx-like proteins. Trx-f, one of the Trx isoforms that reduces CF_oCF₁, contains an electropositive crown around the redox-sensitive Cys residues. The crown may contribute to the selective interaction with target enzymes from the crystal structure (37). Therefore, the intrinsic networks that occur with reducing and oxidizing factors may be altered in the DDE/NNQ mutant and the Δβ-hairpin mutant strains. Second, the redox responses of CF_oCF₁ mutants were modified to become Trx-independent. There may be other redox regulators of CF_oCF₁ besides the Trxs because CF_oCF₁ was reduced even in A. thaliana mutants with an impaired redox cascade (38). The cause of the rapid redox responses of the DDE/NNQ mutant and the Δβ-hairpin mutant strains (Fig. 2), which were significantly faster compared with that of the WT, may be attributed to enhanced unknown Trx-independent redox regulation mechanisms in chloroplasts.

Redox Regulation Mechanism on ATP Synthesis of CF_oCF₁.

By establishing an effective half day purification method for C. reinhardtii CF_oCF₁ and mutation constructs for the corresponding region on the enzyme complex, we obtained significant information regarding the redox regulation of the ATP synthesis reaction catalyzed by CF_oCF₁. The relationship between the redox-regulated ATP synthesis and mutations obtained from our mutant analyses shown in Fig. 5 is summarized in Fig. 6A. Based on previous biochemical and structural studies, the EDE (Glu-Asp-Glu) motif, corresponding to the DDE motif in C. reinhardtii CF_oCF₁, was considered to play an essential role in inducing inactivation by forming an electrostatic interaction network in the oxidized state (23, 24). However, the ATP synthesis activity of the DDE/NNQ mutant was clearly redox-regulated (Figs. 5 and 6A). Therefore, we conclude that the negative charges on the DDE motif are not essential to redox regulation itself. The ΔRedox_loop, Δβ-hairpin, and ΔInsertion mutants, in which the sequences of the γ subunit specific to photosynthetic organisms were truncated, provided insights into the mechanisms of redox regulation. Although we expected the ΔRedox_loop mutant to show similar ATP synthesis activity as the reduced form of the WT, because of its sequence similarity to cyanobacterial F_oF₁, the activity of the mutant was almost identical to that of the oxidized form of the WT (Figs. 5 and 6A). Based on the crystal structure and biochemical analysis, we hypothesize that γ-inhibition occurs in cyanobacterial F₁-ATPase caused by the interaction between the upper part of the β-hairpin domain and the DELSEED region of the β subunit (31). Therefore, the reason why the Δβ-hairpin mutant did not show redox regulation of the ATP synthesis reaction was due to the lack of direct interaction between the upper part of the β-hairpin domain and the DELSEED region. Furthermore, because the Δβ-hairpin and the ΔInsertion exhibited higher ATP synthesis activity compared with the ΔRedox_loop suggests that the β-hairpin is a negative regulator of the ATP synthesis activity of CF_oCF₁ (Figs. 5 and 6A). Because the Δβ-hairpin showed a similar extent of ATP synthesis activity compared with that of the reduced form of the WT, irrespective of their redox states, only conformational changes in the redox loop domain located at the bottom of the CF₁-γ are not able to regulate the activity. Therefore, we proposed a simple structural model for the redox regulation mechanism for the ATP synthesis activity of CF_oCF₁ (Fig. 6B). When the redox-responsive Cys pair is in an oxidized state, the redox loop does not structurally affect the β-hairpin because of a loss of flexibility, and the β-hairpin domain maintains rigidity. Consequently, the effect of steric hindrance of the β-hairpin with the CF₁-β is increased and inhibits rotation of the central stalk, which is required for ATP synthesis. In contrast, when the disulfide bond is reduced, the redox loop domain recovers its flexibility and prevents the β-hairpin from colliding with the CF₁-β, which facilitates central stalk rotation. Taken together, we propose that the redox regulation of CF_oCF₁ occurs through the cooperative interaction of the β-hairpin and the redox loop with the catalytic site in accordance with the redox state of the redox-responsive Cys pair.

Fig. 6.

Proposed model for the redox regulation mechanism on ATP synthesis of CF_oCF₁. (A) A summary of the relationships between redox-regulated ATP synthesis activities and mutations. Redox regulation properties of the mutants are summarized based on the results shown in Fig. 5. In the diagrams, the β-hairpin and the redox loop region are shown in orange on the γ subunit (green). (B) Proposed model for the redox regulation mechanism. In the oxidized form, the redox loop takes on a relatively tight conformation because of the disulfide bond between Cys233 and Cys239 on the redox loop. The tight conformation weakens the interaction between the redox loop and the β-hairpin. Consequently, the β-hairpin remains stuck in the cavity between the α and β subunit, like a stopper, and inhibits the rotation of the central stalk (γεc-ring). In a reduced form, the redox loop recovers flexibility to interact with the β-hairpin. The redox loop interacts to pull out the β-hairpin from the cavity and thus accelerates the central stalk, like a cooperative regulator.

Materials and Methods

ATP, ADP, and valinomycin were obtained from Oriental Yeast, Millipore, and Sigma Aldrich, respectively. N-Octyl-β-D-glucopyranoside and lauryl maltose neopentyl glycol (LMNG) were obtained from Anatrace. Other chemicals were of the highest commercially available grade.

C. reinhardtii Strains and Cell Culture.

The C. reinhardtii strains CC-124 [nit¹⁻(nitrate reductase), nit²⁻, agg¹⁻, and mt⁻], CC-125 (nit¹⁻, nit²⁻, and mt⁺), CC-1185 (atpB⁻ and mt⁺; FUD50 strain), and CC-3728 (cw, ATPC⁻, and mt⁻; T1-54 strain) were used. To eliminate the agg1 mutation, agg1⁺ progenies (mt⁺ and mt⁻) from the mating of CC-124 and CC-125 were used as WT strains (39). The T1-54 strain was used to express mutated CF₁-γ. The plasmids (SI Appendix, Fig. S1) encoding mutated CF₁-γ were introduced into the T1-54 strain by electroporation (NEPA21 Type II; NEPAGENE). Electroporation was performed as described (40), with slight modification (poring pulse: voltage 200.0 V, pulse length 5.0 ms). Homologous recombination of 8×His-tagged atpB was done on the chloroplast genome by introducing the plasmid (SI Appendix, Fig. S3) into the FUD50 strain using a particle gun (IDERA GIE-III; TANAKA) as described (41). CF_oCF₁, which harbors mutation(s) in CF₁-γ, was purified from a progeny (expressing both the 8×His-tagged CF₁-β and the mutated CF₁-γ, and having no ATPC derived from the parental FUD50 strain) resulting from the crossFUD50::8×His-tagged atpB and T1-54::mutated ATPC. The cells were grown in high salt minimal (HSM) medium (42) or in a Tris-acetate phosphate (TAP) medium (43) with aeration at 25 °C and a 12-h/12-h light/dark cycle (light: white, 50 μmol photons m⁻² s⁻¹).

Expression Vector for C. reinhardtii.

The plasmid coding 8×His-tagged atpB was prepared by inverse polymerase chain reaction (PCR) using the Prime STAR Mutagenesis Basal Kit (Takara Bio). First, the 4×His-tag sequence was inserted between the Ser⁴ and Ile⁵ codons on P-113 using 4×His_atpB_F and 4×His_atpB_R primers. The P-113 plasmid was obtained from the Chlamydomonas Resource Center as a template. Next, the 8×His_atpB_F and 8×His_atpB_R primers were used to insert four additional His residues to obtain the 8×His-tag sequence inserted into atpB. The plasmid for homologous recombination was prepared by restriction digestion of P-113, insertion of an 8×His-tag sequence with ClaI and EcoRI and cloning into P-120, which codes a wider chloroplast genome. Transformants were screened on an HSM medium plate. The 8×His-tag insertion was confirmed by DNA sequence and immunoblotting analyses (SI Appendix, Fig. S3). The primers used for plasmid preparation are listed in SI Appendix, Table S2.

Mutations in the ATPC gene encoding CF₁-γ were introduced by overlap extension PCR with the WT genome as a template. A genomic DNA fragment containing the ATPC gene was amplified by PCR with atpC_gDNA_F and atpC_gDNA_R primers. Next, the targeted mutation was introduced by overlap extension PCR. The primers used for these mutations are listed in SI Appendix, Table S2. The mutated ATPC gene was cloned into the pSI103 vector carrying the aphVIII gene (paromomycin resistant) (44) after digestion with EcoRI, using the hot fusion method (45). The targeted mutations were confirmed by DNA sequence analysis (SI Appendix, Fig. S1).

Determination of the Light/Dark Responses of the CF₁-γ Redox States In Vivo.

The redox states of CF₁-γ in vivo were determined by the methods using 4-acetamido-4′-maleimidylstilbene (AMS) (46) or maleimide-PEG₁₁-biotin (PEO) (17). C. reinhardtii cells were cultured in TAP medium to a logarithmic growth stage, adjusted to 2 × 10⁶ cells/mL with TAP medium, and dark-adapted for 5 h. C. reinhardtii cells were collected at the indicated times in the light (100 μmol photons m⁻² s⁻¹) and subsequent dark conditions and immediately mixed with 10% trichloroacetic acid. The mixtures were centrifuged to collect the proteins and washed twice with cold acetone. The resulting protein precipitates were dissolved in SDS-PAGE sample buffer containing 2 mM AMS for WT and Δβ-hairpin CF₁-γ, and 3 mM PEO for DDE/NNQ CF₁-γ to modify the free-Cys residues. Following thermal denaturation (95 °C, 5 min) and centrifugation, nonreducing SDS-PAGE and immunoblotting were done to determine the redox states of the CF₁-γ. Antibodies against CF₁-γ are described in the literature (47).

Purification of the CF_oCF₁ Complex from C. reinhardtii.

Two liters of C. reinhardtii cells cultured under aeration (25 °C, 50 μmol photons m⁻² s⁻¹) in TAP medium were harvested at late log phase by centrifugation, followed by flash-freezing in liquid nitrogen and stored at −80 °C until purification. The cells were resuspended in a thylakoid isolation buffer containing 50 mM HEPES-KOH (pH 7.5), 300 mM sucrose, 5 mM MgCl₂, and 10 mM NaCl were disrupted using a cold French Press at 300 kg cm⁻². The homogenate was centrifuged for 5 min at 300 × g at 4 °C to remove debris. The supernatant was then centrifuged at 125,000 × g at 4 °C for 40 min to precipitate the thylakoid membranes using a Hitachi P45AT rotor. The membranes were resuspended in thylakoid wash buffer containing 50 mM HEPES-KOH (pH 7.5), 300 mM sucrose, 10% glycerol, 10 mM NaCl, 2.5 mM MgCl₂, and 0.1 mM ADP and recentrifuged at 125,000 × g at 4 °C for 40 min. Thylakoid membranes were resuspended in a minimal volume (approximately 10 mL) of thylakoid wash buffer.

Thylakoid wash buffer and 10% LMNG were added at final concentrations of 1 mg/mL total chlorophyll and 1% LMNG, respectively, and ethylene diamine tetra-acetic acid free complete Protease Inhibitor Cocktail (Roche) was immediately added. The extraction mixture was gently stirred for 60 min in a cold room. The solubilized membrane proteins were then separated from the insolubilized membranes by centrifugation at 125,000 × g at 4 °C for 40 min and subjected to Ni-NTA affinity chromatography using 5 mL of resin (Ni Sepharose 6 Fast Flow; Cytiva) pre-equilibrated with a Ni-NTA wash buffer containing 20 mM potassium phosphate (KPi) (pH 7.5), 10% glycerol, 50 mM imidazole, 2.5 mM MgCl₂, 0.1 mM ADP, and 0.005% LMNG. The mixture of solubilized membrane proteins and the resin were gently stirred for 30 min in the cold room to capture His-tagged CF_oCF₁. After washing with Ni-NTA wash buffer, the proteins were eluted with a Ni-NTA elution buffer containing 20 mM KPi (pH 7.5), 10% glycerol, 200 mM imidazole, 2.5 mM MgCl₂, 0.1 mM ADP, and 0.005% LMNG until no protein was detectable by the Bradford assay (Bio-Rad). The eluted sample was concentrated using an Amicon Ultra-15 (MWCO: 100,000; Millipore) and centrifugally filtered (Ultrafree-MC, 0.22 μm; Millipore). The filtered sample was subjected to SEC (Superose 6 Increase 10/300 GL; Cytiva), in which the column was pre-equilibrated with 20 mM KPi (pH7.5), 10% glycerol, 2.5 mM MgCl₂, 0.1 mM ADP, and 0.005% LMNG at room temperature. The CF_oCF₁ containing fractions were collected as the final product, concentrated using an Amicon Ultra-4 (MWCO: 100,000; Millipore), flash frozen in liquid nitrogen, and stored at −80 °C until use. The concentration of the purified CF_oCF₁ was determined using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific) with bovine serum albumin as the standard.

Measurement of ATP Synthesis Activity.

Proteoliposomes were reconstituted by the method described (27) with slight modification. First, crude soybean L-α-phosphatidylcholine (type II-S; Sigma) was washed with acetone to remove K⁺ from the lipid. Next, the washed lipid was suspended in a reconstitution buffer (15 mM MES–tricine, 2 mM KOH, 5 mM NaCl, 2.5 mM MgCl₂, and 50 mM sucrose, with pH adjusted to 8.0 with NaOH). The aliquots were then sonicated to disperse lipids and centrifuged at 125,000 × g at 20 °C for 30 min. The sonication and centrifugation were repeated three times. Then, the resulting lipid was suspended at a final concentration of 32 mg/mL in the reconstitution buffer. Finally, the suspended lipid was flash-frozen in liquid nitrogen and stored at −80 °C until use. The preparation of CF_oCF₁ proteoliposomes was performed as follows. First, the suspended lipid was mixed with an equal volume of 4% n-octyl-β-D-glucoside in the reconstitution buffer. Next, 100 mg of Biobeads (SM-2; Bio-Rad) was added in plural times to 1 mL of the solution and gently stirred until the solution became cloudy. CF_oCF₁ was then added to the solution gently to be the final concentration of 0.15 mg/mL. Biobeads were then added to the mixture to remove excess detergent, and the mixture was incubated at 4°C overnight, followed by flash-frozen in liquid nitrogen, and stored at −80°C until use (27).

To measure ATP synthesis activity, the acid–base transition method was used with the valinomycin-induced diffusion potential of K⁺ as described (34) with slight modification. The CF_oCF₁/lipid ratio of proteoliposomes was set to 90 μg/9.6 mg to accurately measure ATP synthesis activity in both redox states. Before acidification, the proteoliposomes were reduced with 50 mM DTT for 2 h at 30°C in the reconstitution buffer. No oxidizing treatment was done because the purified CF_oCF₁ was in a completely oxidized form (SI Appendix, Figs. S9B and S10). Following the reaction, proteoliposomes were incubated in the acid buffer for 20 min for acidification. ATP calibration was performed by adding 5 µL of 20 µM or 10 μM ATP three times. The details of the conditions for measuring ATP synthesis activity are shown in SI Appendix, Figs. S11 and S12.

Construction of Expression Plasmids for C. reinhardtii Trxs.

Total RNA was isolated from C. reinhardtii and used as a template for RT-PCR of the Trx genes. Gene fragments encoding the mature protein region (predicted ChloroP server) of Trx-f1 (Cre01.g066552_4532) and Trx-f2 (Cre05.g243050_4532) were cloned into the pET-23c expression vector (Novagen). The Trx-f2 plasmid was modified with primers to enhance its expression in Escherichia coli. The primers used for the preparation of the Trxs are listed in SI Appendix, Table S2.

Expression and Purification of Recombinant C. reinhardtii Trx-f1 and f2.

The expression plasmids were transformed into E. coli strain BL21 (DE3). The transformed cells were cultured in a liquid broth medium at 37 °C. When the OD600 of the medium reached 0.3, the medium was cooled with flowing water and further cultured overnight at 21 °C. The cells were resuspended in a buffer containing 25 mM Tris-HCl (pH 8.0) and 0.5 mM DTT and disrupted using a cold French Press (1,200 kg cm⁻²). All subsequent steps were carried out at 4 °C. After centrifugation at 125,000 × g for 40 min, the supernatant was used to purify the protein of interest. For Trx-f1, the proteins were first subjected to anion chromatography (Toyopearl Q-600C AR; Tosoh) and eluted with a linear gradient of NaCl from 0 to 300 mM. The Trx-f1-containing fractions were dialyzed in a buffer containing 25 mM MES-NaOH (pH 6.1) and 0.5 mM DTT, followed by cation exchange chromatography (Toyopearl SP-650M; Tosoh). The proteins were applied to the column and eluted with a linear gradient of NaCl from 0 to 250 mM. The Trx-f1-containing fractions were collected as a final product. For Trx-f2, the proteins were first subjected to anion chromatography (Toyopearl DEAE-650M; Tosoh) and eluted with a linear gradient of NaCl from 0 to 200 mM. Ammonium sulfate was added to the fractions enriched in Trx-f2 to a final concentration of 30% saturation. The supernatant containing Trx-f2 was collected by centrifugation at 10,000 × g for 10 min and subjected to hydrophobic interaction chromatography (Toyopearl Butyl-650; Tosoh). The proteins were applied to the column and eluted with a linear gradient of ammonium sulfate from 20 to 0% saturation. The Trx-f2-containing fractions were collected as a final product and dialyzed in a buffer containing 25 mM Tris-HCl (pH 7.5). The resulting Trx-f1 and Trx-f2 preparations were concentrated using an Amicon Ultra-15 (MWCO: 10,000; Millipore) and then flash-frozen in liquid nitrogen and stored at −80 °C after the addition of glycerol to a final concentration of 20%. The concentrations of purified Trx-f1 and Trx-f2 were determined using the BCA protein assay. The redox responses of the recombinant Trxs were confirmed by the insulin reduction assay (SI Appendix, Fig. S6).

Reconstruction of the CF_oCF₁ Redox Regulation System.

Before reconstruction, each Trx (30 μM) was pre-reduced with DTT (3 mM) for 10 min at 25 °C. The reconstruction was initiated by adding 20 μL of the DTT and Trx mixture to 100 μL of CF_oCF₁ proteoliposome suspension (CF_oCF₁:Trx:DTT = 0.23 μM:5 μM:500 μM). Samples were collected at the indicated times and immediately mixed with 10% trichloroacetic acid. The redox state of CF₁-γ was determined by the method as described above. Kinetics data for the reduction of CF₁-γ were fitted to the pseudo-first-order kinetic model: Reduction level = (Maximum level of reduction) × (1−e^−k[Trx]t).

Supplementary Material

Appendix 01 (PDF)

Data, Materials, and Software Availability

All study data are included in the article and/or SI Appendix.