Thiol-based redox regulation is crucial for adjusting chloroplast functions under fluctuating light environments. We recently discovered that the thioredoxin-like2 (TrxL2)/2-Cys peroxiredoxin (2CP) redox cascade supports oxidative thiol modulation by using hydrogen peroxide (H2O2) as an oxidizing force. This system plays a key role in switching chloroplast metabolism (e.g. Calvin–Benson cycle) during light to dark transitions; however, information on its function is still limited. In this study, we report a novel protein-activation mechanism based on the TrxL2/2CP redox cascade. Glucose-6-phosphate dehydrogenase (G6PDH) catalyzes the first step of the oxidative pentose phosphate pathway (OPPP). Biochemical studies, including redox state determination and measurement of enzyme activity, suggested that the TrxL2/2CP pathway is involved in the oxidative activation of G6PDH. It is thus likely that the TrxL2/2CP redox cascade shifts chloroplast metabolism to night mode by playing a dual role, namely, down-regulation of the Calvin–Benson cycle and up-regulation of OPPP. G6PDH was also directly oxidized and activated by H2O2, particularly when H2O2 concentration was elevated. Therefore, G6PDH is thought to be finely tuned by H2O2 levels in both direct and indirect manners.
Thiol-based redox regulation is a posttranslational mechanism for controlling enzyme activity by switching the oxidation/reduction states of Cys residues (e.g. formation/cleavage of disulfide bonds). In this regulatory system, the small ubiquitous protein thioredoxin (Trx) plays a pivotal role in transferring reducing power to redox-regulated target proteins [1,2]. Trx contains the highly conserved amino acid sequence WCGPC at its active site. By utilizing two Cys residues in this motif, Trx performs a dithiol-disulfide exchange reaction with its targets and thus modulates their activity.
Although Trx-dependent redox regulation is preserved in all kingdoms of life, its regulatory system in plant chloroplasts is functionally distinct from that of other organisms; it is coupled to the light [3,4]. Upon illumination, a photosynthetic electron transport chain in the thylakoid membrane converts light energy to reducing power, a part of which is used for redox regulation. Photosynthetically reduced ferredoxin (Fd) provides reducing power to Trx via Fd-Trx reductase (FTR) . In turn, a reduced form of Trx transfers reducing power to its target proteins. Trx-targeted proteins mediate several key reactions of photosynthesis, including ATP synthesis and the Calvin–Benson cycle; they are activated upon reduction [6,7]. It is therefore recognized that the FTR/Trx redox cascade allows reductive activation of photosynthetic processes in concert with light perception [3,4].
Another feature of the redox-regulatory system in chloroplasts is the occurrence of multiple Trx subtypes, divided into f-, m-, x-, y-, and z-types [8,9]. They have different midpoint redox potentials and protein surface charges, which possibly confers functional diversity to each of the Trx subtypes (e.g. distinct target selectivity) [10–12]. In addition, several Trx-related proteins are present in chloroplasts, including the NADPH-Trx reductase C (NTRC) and the chloroplastic drought-induced stress protein of 32 kDa (CDSP32); they also exhibit functional properties that differ from those of the Trx subtypes [13,14]. Consequently, it is now documented that all of these proteins are highly organized in a complicated redox network, ensuring flexible regulation of chloroplast biology [15–17]. However, its whole picture remains undescribed yet.
One of the elusive issues is a mechanism by which redox-regulated proteins are re-oxidized upon the interruption of light exposure. Such a response has been observed in planta [18,19], but its underlying molecular basis has long been a matter of debate. In a recent study, we finally identified the Trx-like2 (TrxL2) protein as the oxidation factor for redox-regulated proteins . TrxL2 possesses a Trx-like WCRKC motif in the molecule. Owing to its high redox potential, TrxL2 efficiently oxidizes certain redox-regulated proteins, including Rubisco activase (RCA), fructose 1,6-bisphosphatase (FBPase), and sedoheptulose 1,7-bisphosphatase (SBPase). Subsequently, TrxL2 transfers reducing power to 2-Cys peroxiredoxin (2CP) and ultimately to hydrogen peroxide (H2O2). Discovery of the TrxL2/2CP redox cascade represents a breakthrough in the research field of chloroplast redox regulation, but further studies are needed to expand our knowledge regarding this topic . In particular, the chloroplast proteins under the control of TrxL2/2CP pathway remain largely uncharacterized.
Using a biochemical procedure, here we investigated the involvement of the TrxL2/2CP pathway in the redox regulation of glucose-6-phosphate dehydrogenase (G6PDH) in chloroplasts. G6PDH catalyzes the first committed step of the oxidative pentose phosphate pathway (OPPP), a major pathway for supplying NADPH in the dark or in non-photosynthetic plastids . G6PDH is one of the redox-regulated proteins, but its regulatory manner is strikingly different from that of other proteins. Namely, G6PDH is deactivated in the light upon reduction via the FTR/Trx pathway [23,24]. In contrast, G6PDH is oxidized and activated in the dark , but the details concerning the responsible regulatory factor remain undetermined. This study provides in vitro evidence that the TrxL2/2CP redox cascade supports oxidative activation of G6PDH. The direct involvement of H2O2 in G6PDH activation is also shown.
Preparation of expression plasmids
Total RNA was isolated from Arabidopsis thaliana as described previously  and used as a template for RT-PCR. The gene fragment encoding the predicted mature protein region of G6PDH1 (At5g35790; Phe51–Ala576) was amplified with the following oligonucleotide primers: 5′-CATGCCATGGCGTTCTTCGCCGAGAAACATTC-3′ (NcoI) and 5′-CCGCTCGAGAGCTTCTCCAAGATCTCCCC-3′ (XhoI). The restriction sites for the enzymes in parentheses are underlined. The amplified DNA was ligated into the NcoI and XhoI sites of the pET-23d expression vector (Novagen). The sequence of the G6PDH1 expression plasmid was confirmed to be correct by DNA sequencing (3730xl DNA Analyzer; Applied Biosystems). The G6PDH1 plasmid was designed to express the C-terminal His-tagged fusion protein. The other expression plasmids (Trx-f1, At3g02730; Trx-m1, At1g03680; TrxL2.1, At5g06690; TrxL2.2, At5g04260; 2-Cys peroxiredoxin A (2CPA), At3g11630; RCA, At2g39730) were prepared in our previous studies [11,12,20,27].
Point mutations in G6PDH1 (Cys65 to Ser; Cys249 to Ser) were introduced using the PrimeSTAR Mutagenesis Basal Kit (Takara) according to the manufacturer's instructions. The following oligonucleotide primers were used for site-directed mutagenesis: for Cys65 to Ser, 5′-AATGGGTCTGCCACTAATTTCGCTTCT-3′ (forward) and 5′-AGTGGCAGACCCATTTGAAGTATCAAG-3′ (reverse); for Cys249 to Ser, 5′-ACTAGATCTCTGAAACAGTATCTTACA-3′ (forward) and 5′-TTTCAGAGATCTAGTTAACTCTCCGGA-3′ (reverse). The mutated codons are underlined.
Protein expression and purification
G6PDH1 expression plasmid was transformed into the Escherichia coli strain BL21(DE3). Transformed cells were cultured at 37°C until OD600 reached 0.5–0.7. Expression was induced by the addition of 0.5 mM isopropyl-1-thio-β-d-galactopyranoside, followed by the overnight culture at 21°C. Cells were disrupted by sonication. After centrifugation (125 000×g for 40 min), the resulting supernatant was used to purify the protein. His-tagged G6PDH1 protein was purified using a Ni-nitrilotriacetic acid affinity column as described previously . Other recombinant proteins were prepared as described previously [11,12,20,27]. The protein concentration was determined with a BCA protein assay (Pierce).
Redox reaction in vitro
Proteins were reacted at 25°C in buffer containing 50 mM Tris–HCl (pH 7.5) and 50 mM NaCl. Concentrations of proteins and reducing/oxidizing agents are described in each figure.
Determination of protein redox state
The protein redox states were determined by discriminating the thiol status with the use of thiol-modifying reagents (maleimide-PEG11-biotin (MW 922.09) for G6PDH1 and 4-acetamido-4′-maleimidylstilbene-2,2′-disulfonate (MW 536.44) for RCA). These reagents change protein mobility on SDS–PAGE, allowing the determination of protein redox states with an observable band shift. Following the in vitro reaction, proteins were precipitated with 10% (w/v) trichloroacetic acid and then washed with ice-cold acetone. Precipitated proteins were labeled with the reagents described above for 1 h at room temperature. Proteins were subjected to non-reducing SDS–PAGE and stained with Coomassie Brilliant Blue R-250.
G6PDH1 activity was measured at 25°C in reaction buffer containing 50 mM Tris–HCl (pH 7.5), 50 mM NaCl, 5 mM MgCl2, 1 mM NADP+, the indicated concentration of glucose-6-phosphate (G6P), and 0.2 µM G6PDH1. Activity was monitored as an increase in absorbance at 340 nm due to NADP+ reduction. A molar extinction coefficient for NADPH of 6.2 mM−1 cm−1 was used for calculating the amounts of catalyzed NADP+.
H2O2-detoxifying activity of the Trx/2CP and TrxL2/2CP pathways
The H2O2-detoxifying activity of 2CP was measured as described recently  with slight modifications. The 2CPA (3 µM) was incubated with 0.2 mM dithiothreitol (DTT), 1.5 µM Trx (Trx-f1 or Trx-m1) or TrxL2 (TrxL2.1 or TrxL2.2), and 100 µM H2O2 in reaction buffer containing 50 mM Tris–HCl (pH 7.5) and 50 mM NaCl for the indicated time at room temperature. The H2O2 concentration was then determined by ferrous oxidation of xylenol orange assay.
Size-exclusion chromatography was performed using a Superdex 200 10/300 GL column (GE Healthcare). The G6PDH1 protein was eluted in 25 mM Tris–HCl (pH 7.5) and 150 mM NaCl at 0.5 ml/min. A Gel Filtration Calibration Kit (GE Healthcare) was used to calibrate the elution volume with the molecular mass.
Modeling of the G6PDH1 three-dimensional structure
The three-dimensional structure model of G6PDH1 was constructed using the Swiss-Model software . The crystal structure of Homo sapiens G6PDH (PDB code: 5VG5) was used as a template.
Statistical analyses were performed with the Tukey–Kramer multiple comparison test using SPSS 12.0J software (SPSS Inc.).
Results and discussion
Arabidopsis G6PDH1C65S/C249S mutant is a suitable model for studying the redox regulation of plastid G6PDH
Plant G6PDH has several isoforms targeted to plastids or cytosol, each of which shows a tissue-specific expression pattern [29–33]. In A. thaliana, there are four plastidic (G6PDH1-4) and two cytosolic (G6PDH5 and 6) isoforms (Supplementary Figure S1). Only the plastidic isoforms are redox-regulated and commonly have redox-active Cys residues (Cys149 and Cys157; numbers for G6PDH1) that form an intramolecular disulfide bond under oxidizing conditions . Of these, G6PDH1 was shown to exert high enzyme activity with substantial expression in photosynthetic tissues . In contrast, the enzyme activities of G6PDH2-4 were shown to be relatively low. We therefore focused on the Arabidopsis G6PDH1 isoform in this study.
G6PDH1 has additional Cys residues that are not found in the other G6PDH isoforms (Cys65 and Cys249; Supplementary Figure S1). We constructed two types of G6PDH1 as recombinant proteins, the wild type (G6PDH1WT) and its variant in which Cys65 and Cys249 are substituted to Ser (G6PDH1C65S/C249S). G6PDH1WT formed oligomers under non-reducing conditions, whereas G6PDH1C65S/C249S did not (Supplementary Figure S2). This result indicates that Cys65 and/or Cys249 are involved in G6PDH1 intermolecular disulfide bond formation. A three-dimensional structural model of G6PDH1 showed that Cys249 is located on the protein surface (Supplementary Figure S3), further supporting this idea. Importantly, Cys65 and Cys249 in Arabidopsis G6PDH1 are not conserved in plastid G6PDH in plants (Supplementary Figure S4), suggesting that the oligomer formation observed in G6PDH1WT is a specific case to the Arabidopsis plastid G6PDH1. The presence of Cys65 and Cys249 also hampered the determination of redox state using a thiol-modifying reagent, which was a key method used in this study (Supplementary Figure S5). In contrast, the redox state of G6PDH1C65S/C249S could be clearly discriminated (Supplementary Figure S5). For these reasons, we used G6PDH1C65S/C249S in the following experiments as a representative of plastid G6PDH.
Redox state and enzyme activity are tightly coupled in G6PDH
We assessed G6PDH1 redox regulation using two different methods: direct determination of redox state and measurement of enzyme activity. The former method indicated that G6PDH1 was present in an oxidized form under control conditions (Figure 1A). When incubated with Trx-m1 and the reducing agent DTT, G6PDH1 was converted to the reduced form. This reduction response was accompanied with a drastic loss of G6PDH activity (Figure 1B). These results underpin the close link between the redox state and enzyme activity in G6PDH1. These also agree with the suggestion that cleavage of the disulfide bond between Cys149 and Cys157 leads to G6PDH1 deactivation through lowering substrate accessibility to the active site and modifying cofactor-binding properties .
Relationship between redox state and enzyme activity in G6PDH1.
TrxL2 is inefficient in G6PDH reduction
We recently showed that TrxL2 can oxidize RCA, FBPase, and SBPase, but fails to reduce these proteins . We thus examined whether the inability of TrxL2 to reduce proteins also applies to G6PDH1 (Figure 2). In this experiment, DTT was added to the reaction at 0.5 mM, which could not reduce G6PDH1 (2 µM) directly. It was reported that, of the five Trx subtypes, Trx-f and Trx-m efficiently deactivate G6PDH in Arabidopsis  and poplar . In accord with these data, G6PDH1 was efficiently reduced by Trx-f1 or Trx-m1 even at low concentrations (0.1 µM), but not by the two isoforms of TrxL2 (TrxL2.1 and TrxL2.2). Higher concentrations of TrxL2.2 (0.5 or 1 µM) could only partially reduce G6PDH1. These results indicate that TrxL2 may physically interact with G6PDH1, but it cannot serve as an efficient reducer for G6PDH1.
Trx and TrxL2 selectivity for G6PDH1 reduction.
G6PDH is directly oxidized and activated by H2O2
We then studied the mechanisms of G6PDH1 oxidative activation. For this purpose, G6PDH1 was initially reduced and deactivated by Trx-m1 and DTT (as shown in Figure 1). Subsequently, G6PDH1 was subjected to size-exclusion chromatography to remove Trx-m1 and DTT (Supplementary Figure S6). The resulting protein elution profiles indicated that G6PDH1 is present as a high-ordered oligomer and its apparent mass is further elevated upon reduction. This is due to the fact that disulfide bond formation results in a more compact conformation of G6PDH1 without affecting its quaternary structure .
Early studies gave implications for the H2O2-dependent stimulation of G6PDH activity [38,39]. In contrast, a later study suggested that direct activation of G6PDH by H2O2 does not occur . Given these inconsistencies, we investigated whether or not H2O2 is directly involved in G6PDH1 oxidative activation. Our results indicated that a reduced form of G6PDH1 was oxidized by H2O2 in a dose-dependent manner (Figure 3A). Furthermore, G6PDH1 activity was restored by H2O2 incubation (Figure 3B). These results strongly suggest that G6PDH can be oxidized and activated by H2O2 in a direct way. For comparison, we tested the possibility of H2O2-dependent oxidation of RCA. RCA could not be efficiently oxidized by H2O2 even at a concentration of 100 µM. Instead, complete reconstitution of the TrxL2/2CP pathway was required for RCA oxidation (Figure 3C). As demonstrated recently , direct oxidation by 100 µM H2O2 was also marginal for other redox-regulated proteins, including FBPase and SBPase. It therefore seems likely that H2O2 sensitivity is a unique redox-regulatory characteristic of G6PDH, at least to a certain extent. However, it should be noted that H2O2 potentially affects the chloroplast proteome with different modifications (e.g. sulfenylation) [40,41].
H2O2-dependent oxidation and activation of G6PDH1.
G6PDH is efficiently oxidized and activated by the TrxL2/2CP redox cascade
Next, we investigated whether the TrxL2/2CP redox cascade is involved in G6PDH oxidative activation. An inactive reduced form of G6PDH1 was incubated with the reconstituted TrxL2/2CP pathway. H2O2 was added to the reaction at 25 or 100 µM. Under conditions with 25 µM H2O2, two isoforms of TrxL2 promoted G6PDH1 oxidation with different efficiencies (Figure 4A). In particular, TrxL2.2 drastically enhanced the rate of G6PDH1 oxidation, which could be related to the higher midpoint redox potential of TrxL2.2 (−245 mV at pH 7.5) versus TrxL2.1 (−258 mV at pH 7.5) . Accordingly, TrxL2.2 largely reactivated G6PDH1 (Figure 4B). Under conditions with 100 µM H2O2, G6PDH1 was efficiently oxidized and activated within 15 min even without TrxL2 (Figure 4C,D), because this H2O2 concentration was effective enough for direct oxidation and activation of G6PDH1 (Figure 3). However, TrxL2 significantly accelerated G6PDH1 oxidation and activation under these conditions, albeit to a lesser extent. These results suggest that the TrxL2/2CP redox cascade acts as an oxidative activator of G6PDH with changing its contribution relying on H2O2 levels.
The involvement of TrxL2/2CP redox cascade in G6PDH1 oxidative activation.
Other potential factors involved in G6PDH oxidative activation
It is possible that other factors also mediate G6PDH oxidative activation. For example, the atypical Cys His-rich Trx (ACHT) has the potential to oxidize some redox-regulated proteins in chloroplasts [42,43]. Similar to that of TrxL2, the midpoint redox potential of ACHT is much less negative than that of Trx , supporting the function of ACHT as a protein-oxidation factor. The interaction of ACHT with G6PDH has been entirely uncharacterized so far, which warrants further investigation.
The involvement of Trx in G6PDH oxidative activation was also suggested; Née et al. demonstrated that G6PDH is activated by Trx (especially Trx-f and Trx-m) in the presence of a chemical oxidant (i.e. the oxidized form of DTT (DTTox)) . Their observation appears irrelevant in a physiological context, because DTTox was used at too high concentration (10 mM) in those experiments, possibly conferring an unusually strong oxidizing force on Trx. Recent studies have implicated that 2CP functions as a physiological Trx oxidase, enabling Trx to mediate protein oxidation [45,46]. This proposal is partially in agreement with our model in terms of the participation of 2CP in the protein-oxidation pathway . However, it remains unclear whether the Trx/2CP pathway is an efficient oxidative regulator of G6PDH (and other redox-regulated proteins) due to the following aspects. First, the midpoint redox potential of Trx (Trx-f, −351 mV at pH 7.9; Trx-m, −358 mV at pH 7.9)  is lower than that of G6PDH (−329.9 mV at pH 7.9) . Second, the 2CP-reducing efficiency of Trx is low compared with that of TrxL2 . We verified this issue again by monitoring the H2O2 detoxification of the reconstituted Trx/2CP and TrxL2/2CP pathways (Supplementary Figure S7). The results provided further evidence for much better reactivity of TrxL2 with 2CP. Finally, the in vivo redox state of Trx immediately reverts to a dark-adapted basal level following interruption of light exposure . This result indicates that the redox-relay reaction via Trx absolutely depends on light and the resulting photosynthetic electron transport; in other words, it does not occur during light to dark transitions. We therefore need further studies to conclude the function of Trx as a protein-oxidation factor.
It is known that the activity of chloroplast G6PDH shows dynamic light/dark modulation [23–25]. The present study provides biochemical insights into how G6PDH is activated in the dark. In combination with recent data , we can consider a dual role of the TrxL2/2CP redox cascade that deactivates the Calvin–Benson cycle and activates OPPP (Supplementary Figure S8). This appears to be beneficial for readily switching the chloroplast metabolism to a non-photosynthetic mode at night. G6PDH is also directly oxidized and activated by H2O2 especially under elevated H2O2 conditions. As a next major challenge, it is of great importance to understand to what extent each of two pathways (the TrxL2/2CP pathway and direct pathway to H2O2) works in vivo. Although argued historically [47–49], there is little consensus concerning the H2O2 accumulation level in chloroplasts, which seems to be key to solve this issue. In addition, we need to characterize the redox dynamics of G6PDH and its relationship with H2O2 fluctuations under day/night cycle.
2-Cys peroxiredoxin A
atypical Cys His-rich thioredoxin
chloroplastic drought-induced stress protein of 32 kDa
NADPH-thioredoxin reductase C
oxidative pentose phosphate pathway
K.Y. and T.H. designed the research; K.Y., E.U., and S.H. performed the research; K.Y., S.H., and T.H. discussed the data; K.Y. and T.H. wrote the paper. S.H. commented on the paper.
This study was supported by the Japan Society for the Promotion of Science (JSPS) KAKENHI Grant [16H06556] (to K.Y. and T.H.), the Sumitomo Foundation [180881; to K.Y.], and the Yoshinori Ohsumi Fund for Fundamental Research (to K.Y.).
We thank Ichiro Terashima, Chikahiro Miyake, and Masaru Kono for discussions; the Biomaterial Analysis Division, Tokyo Institute of Technology for supporting DNA sequencing analysis. This study was partly supported by Dynamic Alliance for Open Innovation Bridging Human, Environment and Materials.
The Authors declare that there are no competing interests associated with the manuscript.