Impact of key residues within chloroplast thioredoxin-f on recognition for reduction and oxidation of target proteins

Thioredoxin (Trx) is a redox-responsive protein that modulates the activities of its target proteins mostly by reducing their disulfide bonds. In chloroplasts, five Trx isoforms (Trx-f, Trx-m, Trx-x, Trx-y, and Trx-z) regulate various photosynthesis-related enzymes with distinct target selectivity. To elucidate the determinants of the target selectivity of each Trx isoform, here we investigated the residues responsible for target recognition by Trx-f, the most well-studied chloroplast-resident Trx. As reported previously, we found that positively-charged residues on the Trx-f surface are involved in the interactions with its targets. Moreover, several residues that are specifically conserved in Trx-f (e.g. Cys-126 and Thr-158) were also involved in interactions with target proteins. The validity of these residues was examined by the molecular dynamics simulation. In addition, we validated the impact of these key residues on target protein reduction by studying (i) Trx-m variants into which we introduced the key residues for Trx-f and (ii) Trx-like proteins, named atypical Cys His-rich Trx 1 (ACHT1) and ACHT2a, that also contain these key residues. These artificial or natural protein variants could reduce Trx-f–specific targets, indicating that the key residues for Trx-f are critical for Trx-f–specific target recognition. Furthermore, we demonstrate that ACHT1 and ACHT2a efficiently oxidize some Trx-f–specific targets, suggesting that its target selectivity also contributes to the oxidative regulation process. Our results reveal the key residues for Trx-f–specific target recognition and uncover ACHT1 and ACHT2a as oxidation factors of their target proteins, providing critical insight into redox regulation of photosynthesis.

Thioredoxin (Trx) 2 is a ubiquitous protein that modulates the activities of various enzymes mostly by reducing disulfide bonds on the target protein molecules. Trx possesses the conserved active-site motif WCGPC, and it also carries a common structure named Trx-fold (1)(2)(3). In chloroplasts, Trx modulates the activities of various enzymes involved in photosynthesis-related metabolism (4). Because this process is promoted by the reducing power produced by the lightdriven photosynthetic electron transport system, the redoxbased regulation system enables plants to adequately adapt to fluctuating light conditions. Two chloroplast Trxs named Trx-f and Trx-m were initially identified as activators of fructose-1,6-bisphosphatase (FBPase) and NADP-malate dehydrogenase in chloroplasts, respectively (5)(6)(7). Several enzymes, including Calvin-Benson cycle enzymes (glyceraldehyde-3-phosphate dehydrogenase, sedoheptulose-1,7-bisphosphatase, and phosphoribulokinase (4,8)), glucose-6-phosphate dehydrogenase (9), and chloroplast ATP synthase ␥-subunit (10), were then identified as targets of Trx-mediated reduction. In 2000, the whole-genome sequences of the green plant Arabidopsis thaliana were revealed (11), and five chloroplast Trx isoforms (f-, m-, x-, y-, and z-type) were identified in the genome based on their sequence similarities. In addition, the plant genome data enabled us to study unknown Trxtargeted proteins via various proteomic approaches such as two-dimensional SDS-PAGE analysis, following chemical modification of free thiols, and affinity chromatography using the immobilized monocysteine mutant Trx (12)(13)(14)(15). These approaches explosively increased the number of Trxtargeted proteins.
Chloroplast Trxs specifically recognize their targets. For example, Trx-f appeared to preferentially reduce proteins involved in the Calvin-Benson cycle such as FBPase and CP12, whereas other Trx isoforms in chloroplasts cannot efficiently reduce these proteins (6 -8, 16 -19). However, the determinant of the target selectivity of Trx-f and the differences among the determinants of Trx isoforms are not known very well.
The midpoint redox potentials (E m ) of the redox-responsive cysteine pairs on Trx and its target molecules are some of the factors determining the direction of the reactions (20). In addition, a relationship between the E m values of chloroplast Trxs and the rates of electron transfer from their reductase to Trxs was also reported (21). However, the target selectivity of each chloroplast Trx isoform is apparently irrelevant to its E m (16,18,21,22). For example, Trx-f and Trx-m have similar E m values, but only Trx-f can efficiently reduce FBPase (18). These Trx-dependent reduction processes were found to be rate-limiting steps in the light-dependent activation of chloroplast enzymes (23). Thus, target selectivity might determine the activation rate of each enzyme and enable the sophisticated regulation of various chloroplast functions.
To elucidate the determinant of the target selectivity of Trx-f, several mutational and kinetic analyses have been conducted (24 -28). These previous studies were mostly based on only the differences among Trx-f and -m from spinach and Trx from Escherichia coli. In addition, other Trxs (Trx-x, Trx-y, and Trx-z) were not considered at that time because they were not identified before the A. thaliana whole-genome analysis (16,29,30). Thus, the determinants of the target selectivity of Trx isoforms have not been completely identified. To address this question, we studied the residues specifically conserved in Trx-f via comparisons of the amino acid sequences of chloroplast Trxs from various plants. Some of the conserved residues should be candidate residues that are involved in Trx-f-specific target recognition. We then adopted a recently-developed sensor protein that can detect the reduction of a targeted protein of Trx-f as a fluorescent signal. This Förster resonance energy transfer (FRET)-based sensor protein, which is named "change in redox state of thioredoxin 1 (CROST1)" (31), is useful for studying the efficiency of reducing power transfer from Trx-f to the target. Using this new sensor protein and various site-directed mutants of Trxs, we could identify the key residues as the determinant of the target selectivity of Trx-f.
We recently identified Trx-like 2 (TrxL2) as a critical protein for the oxidation process of Trx-targeted proteins in chloroplasts (32). A similar result was reported previously (33) regarding atypical Cys His-rich Trx 4 (ACHT4), which is also a Trxlike protein. In this study, we found that ACHT1 and ACHT2 also possess the key residues, which were identified on Trx-f, based on the sequence comparison. We then demonstrated the function of these residues during the target oxidation process.

Specific residues conserved in each type of chloroplast Trxs
Ten chloroplast Trx isoforms, which were designated as Trx-f1, Trx-f2, Trx-m1, Trx-m2, Trx-m3, Trx-m4, Trx-x, Trx-y1, Trx-y2, and Trx-z, were found in A. thaliana (19). The orthologs of these Trx genes were also identified in other plant genomes via sequence similarity-based surveys (34,35). To explore the residues potentially involved in the target selectivity of each type of Trxs, we first conducted an amino acid sequence comparison of Trxs from several plants that are well-annotated in the UniprotKB database. We also included well-studied Chlamydomonas Trxs, although the Trx system in Chlamy-domonas is somewhat different from those of vascular plants (8,35). Consequently, 14 residues were identified as candidate residues responsible for the target selectivity of Trx-f (Fig. 1, A-C, highlighted in magenta). Notably, most of the specific residues for each type of Trxs were located on the surface of the molecule, which presumably can interact with target proteins (see residues highlighted in magenta or black and the Interaction interface line in Fig. 1A, as well as B-D).

Decrease of the target reduction activity via mutation of Trx-fdistinctive residues
We then prepared various Trx-f1 mutants harboring mutations of Trx-f-distinctive residues (Fig. 1B). Five amino acid residues of Trx-f1 (Met-94, Thr-96, Asn-127, Leu-137, and Asn-151) were replaced by their counterparts in the other types of Trxs because they are mostly conserved as other amino acid residues in the other types of Trxs (Fig. 1A). Seven hydrophilic residues of Trx-f1 (Gln-97, Asp-118, Arg-131, Lys-135, Glu-156, Thr-158, and Lys-161) were substituted with Ala, and one hydrophobic residue (Val-141) was replaced by Ser. Cys-126 was substituted with both Ser and Ala. Because Val-142 is located at the cis-ProϪ1 position (Fig. 1A, asterisk) and the amino acid at this position is reported to be important for the redox property of Trx-family proteins (20), we also introduced a mutation in this residue (see "Discussion" for details). However, all residues at this position of chloroplast Trxs have a similar hydrophobic property. We therefore replaced Val-142 with Met because Met is found at this position in TrxL2 (35), which could not efficiently reduce Trx-f-specific targets in our previous study (32). Trx-f1 mutants expressed in E. coli were then purified to homogeneity, although the mutants for Met-94, Asp-118, and Leu-137 were not obtained as soluble proteins.
To confirm the activity of the purified Trx-f1 mutants as reductants, we evaluated their E m values and insulin-reducing activities (36) in the presence of reduced dithiothreitol (DTT). The E m value for the wildtype (WT) Trx-f1 is reported to be Ϫ321 mV at pH 7.5 (18). Indeed, half of WT Trx-f1 was reduced in the buffer with the redox potential Ϫ321 mV ( Fig. 2A). In this redox buffer, the ratio of the reduced fractions of most of Trx-f1 mutants was around 0.5, indicating that their E m values are similar to that of WT Trx-f1. So far, the E m values of A. thaliana chloroplast Trx isoforms are reported to be within the range from Ϫ335 to Ϫ276 mV at pH 7.5 (18,21). As shown in Fig. 2A, the ratio of the reduced fractions of all Trx-f1 mutants was higher than 0.5 at redox potential Ϫ335 mV and lower than 0.5 at redox potential Ϫ276 mV. These results clearly demonstrate that E m values of Trx-f1 mutants are in the range of those of typical chloroplast Trxs. Insulin-reducing activity measurement shows that all examined mutants retained more than 85% of the activity of WT Trx-f1 (Fig. 2, B and C). These results indicate that the residues of Trx-f1 examined in this study are not strongly involved in the redox properties of Trx, namely the reductase activity.
Subsequently, we examined the reduction activities of these mutants using the typical f-type Trx-specific target enzyme FBPase. To confirm that the residues are important not only for FBPase recognition but also for other Trx-f-specific targets, we

Determinant of the target selectivity of chloroplast Trx-f
used the artificial Trx-f-specific target protein CROST1 as well. CROST1 is a FRET-based sensor protein constructed using cyan fluorescent protein, yellow fluorescent protein, and the N-terminal region of CP12, a redox-sensitive regulator protein of the Calvin-Benson cycle. The reduction of CROST1 is easily assessed via its fluorescent signal, and the redox status of Trx-f was successfully monitored in vivo using this sensor protein (31). The reduction activities of Trx mutants were assessed by analyzing the rate constants of target protein reduction (see "Experimental procedures"). Consequently, most of the mutations in Trx-f1 significantly decreased their reduction activities, particularly for FBPase reduction (Fig. 3). One exception was the mutant N151G, which exhibited a similar ability to reduce target proteins as WT. Based on these results, we concluded that the Trx-f-distinctive residues shown in Fig. 1, A and B, excluding Met-94, Asp-118, Leu-137, and Asn-151 are the key residues for the target selectivity of Trx-f.

Interaction between Trx-f and FBPase in a dynamic structural model
To evaluate the validity of these key residues for target recognition, the interaction between Trx-f and FBPase was In the alignment result, the type-specific residues, active-site residues, and residues conserved among Trx-x, Trx-y, and Trx-z are highlighted in magenta or black, orange, and gray, respectively. The residue numbers of A. thaliana Trx-f1 (counted from the translational start site amino acid methionine) are shown above the figure. Each letter and an asterisk below the alignment result denote conserved residues in typical Trxs (2) and the cis-ProϪ1 residue (20), respectively. The secondary structure of SoTrx-f (PDB code 1F9M) is shown on the Secondary structure line. On the Interaction interface line, the residues of Trxs that are predicted to interact with target proteins (upper, blue) or Trx reductases (bottom, green) are shown. Full names of the organisms are described in Table S1. B-D, model structure of A. thaliana Trx-f1. Trx-f-specific residues (magenta) and Cys residues in the active site (orange) (B and C), residues involved in the interaction with targets (blue) (D) are shown in a ribbon or molecular surface model.

Determinant of the target selectivity of chloroplast Trx-f
examined in a structural model of their complex. To explore a feasible complex structure, we generated a large number of protein/protein complex models by docking, and we selected reasonable models based on Cys residue distances followed by molecular dynamics (MD) simulations. Consequently, we could obtain a dynamic structural model of a Trx-f/FBPase complex (Movies S1-S3) and a representative structural model of the complex (Fig. 4A). Based on this structural model, minimum pairwise heavy atom distance between each of the Trx-f key residues and FBPase (d min ) was determined (Fig. 4B). Assuming that residues whose d min values are less than 4 Å interact with FBPase, Thr-96, Arg-131, Val-141, and Val-142 almost constantly interact with FBPase, whereas d min values of the other Trx-f-distinctive residues were found to oscillate significantly ( Fig. 4B and Movies S1-S3). According to these results, most of the key residues for Trx-f are supposed to interact with a target protein concomitantly with continuous conformational changes.

Reduction of Trx-f-specific target proteins by ACHT1 and ACHT2a
Based on the amino acid sequence comparison of chloroplast Trxs and Trx-like proteins in A. thaliana, we found that ACHT1 and ACHT2a partially maintain the key residues identified on Trx-f, whereas ACHT4a, another isoform of ACHT, lacks several key residues that are maintained in ACHT1 and ACHT2a (Fig. 5D). The reduction activities of these ACHTs against CROST1 and FBPase were then examined. As shown in Fig. 5, B and E, CROST1 and FBPase were sufficiently reduced by both ACHT1 and ACHT2a, whereas ACHT4a could not reduce either protein. These results clearly demonstrate that these key residues contribute to the interaction with Trx-fspecific target proteins.

Efficient oxidation of Trx-f target proteins by ACHT1 and ACHT2a
Although ACHT1 and ACHT2a could reduce Trx-f-specific target proteins, their efficiencies were lower than that of Trx-f1, and rate constants of Trx-f1, ACHT1, and ACHT2a for CROST1 reduction were calculated to be 23.8, 5.16, and 6.22 M Ϫ1 min Ϫ1 , respectively (Fig. 5, B and E). To clarify the differ-

Determinant of the target selectivity of chloroplast Trx-f
ence of the reduction efficiencies of these proteins, we determined the E m values of ACHTs. This E m determination was performed under the same conditions described in the previous reports (18,21,32) to compare the obtained values with the values of other Trxs in these reports. As shown in Fig. 6A, values of Ϫ252 and Ϫ247 mV were obtained for ACHT1 and ACHT2a at pH 7.5, respectively. The E m value of ACHT4a was not deter-mined because of the multiple band-shift patterns (Fig. 6B).
The obtained E m values of ACHT1 and ACHT2a were markedly higher than those of typical Trxs and moderate between those of TrxL2.1 and TrxL2.2 ( Fig. 6C) (18,21,32), which are two isoforms of TrxL2. These results raised the possibility that ACHT1 and ACHT2a act as oxidation factors similarly as TrxL2 proteins (32).

Determinant of the target selectivity of chloroplast Trx-f
We therefore examined whether ACHT proteins could oxidize CROST1 and FBPase. Consequently, both ACHT1 and ACHT2a oxidized CROST1 more efficiently than Trx-f1 (Fig. 7A). Oxidation of CROST1 was more obvious after 30 s in the presence of 0.1 M ACHT1 and ACHT2a than in the presence of 1 M Trx-f1 as indicated in Fig. 7A. This implies that oxidation of CROST1 by ACHT1 and ACHT2a occurred at least 10-fold faster than that by Trx-f1. In addition, partial re-reduction of the oxidized CROST1 was observed. Because the E m value of CROST1 was Ϫ266 mV (31), which was not significantly different from those of ACHT1 and ACHT2a, these proteins might have remained in a redox equilibrium state in the reaction mixture. FBPase was also efficiently oxidized by ACHT1 and ACHT2a concomitant with their reduction (Fig. 7, B and C). When the oxidation of FBPase was examined with Trx-f1, ϳ60% of the enzyme was oxidized, but the time required to reach the equilibrium state was less than 30 s, as observed for ACHT1 and ACHT2a. ACHT4a was also able to oxidize these targets, although the oxidation rate was extremely slow.
If ACHT1 and ACHT2a can serve as oxidation factors, then they should drain electrons to a certain acceptor similarly as TrxL2 (32). We therefore examined the capacity of ACHT1 and ACHT2a to reduce 2-Cys peroxiredoxin A (2-Cys PrxA) and PrxQ. Efficient H 2 O 2 reduction was observed when ACHT1 and ACHT2a were added to 2-Cys PrxA in the presence of DTT (Fig. 8). Conversely, these reductants could not work with PrxQ, suggesting that ACHT proteins transfer reducing power from their targets to H 2 O 2 via 2-Cys Prx in chloroplasts.

On the reduction process
Previous studies have largely expanded knowledge on the target selectivity of chloroplast Trxs (16,18,29). The determinants of the target selectivity of Trx have also been studied (24 -28). However, they have been hardly clarified to date because of the variety of the important residues they suggested. In addition, some of them are not conserved in plant Trx-f. For example, Geck et al. (25) identified several amino acid residues involved in target recognition by spinach Trx-f. One of the critical residues identified in their study was Lys-58 (corresponding to Ala-111 in A. thaliana Trx-f1). However, this residue was not conserved in other Trx-f proteins and was even found in several Trx-m isoforms (Fig. 1A), implying that this residue should not be a common target recognition residue in Trx-f. To reveal the general characteristics of Trx-f, we first performed a sequence comparison of Trx proteins from various plants and identified the residues that are specifically conserved in each type of Trx isoforms (Fig. 1A). We then found that most of the specific residues are located in the putative interaction interface between Trx and target proteins (Fig. 1). Although the co-crystal structures of chloroplast Trxs with target proteins have not been reported, the Trx-target interaction interface can be predicted using the co-crystal or NMR structures of Trx and target proteins from other origins. The interaction interface between Trx and Trx reductase (TrxR) was also predicted, and it appeared to mostly overlap with the Trx-target interaction interface (Fig. 1A, blue bars and green bars). These results suggest that residues specifically conserved in each type of Trx can be involved in both target and TrxR recognition, although we previously demonstrated that the rate of chloroplast Trx reduction by TrxR correlates with E m (21).

Determinant of the target selectivity of chloroplast Trx-f
The specific residues in each type of Trx isoforms were observed even at the positions in which residues should be universally conserved among Trx isoforms (letters shown below the alignment in Fig. 1A) (2). In particular, the cis-ProϪ1 residue (shown in Fig. 1A with an asterisk), which is reported to be Ile in many Trx isoforms (20), is Val in Trx-f; Ile in Trx-m; Leu in Trx-x, Trx-y, Trx-z, and ACHT; and Met in TrxL2 (Figs. 1A,  5D, and 9, A and B). Other Trx-like proteins also have specific cis-ProϪ1 residues, e.g. the Trx domain of NADPH-Trx reductase C (NTRC) has Thr at this position (Fig. 9C, highlighted in  magenta). The residue at this cis-ProϪ1 position is reported to greatly affect the redox properties of the Trx-family proteins (20). We therefore included this residue in the candidates of the key residue for Trx-f. The information on the specific residues in each type of Trx isoforms identified in this study might pro-vide insights into the target selectivity of various isoforms, such as the Trx-y-specific activation of monodehydroascorbate reductase (37).
Interestingly, several residues of Trx-z that are predicted to be involved in target recognition are not located in the interaction interface (see Type-z and Interaction interface lines in Fig.  1A). This might be a unique feature of Trx-z because this isoform is suggested to serve as one of the components of plastidencoded RNA polymerase complex in addition to functioning as a redox mediator (30, 38 -41).
In this study, we assessed the impact of the Trx-fdistinctive residues on the interaction with Trx-f-specific targets via three strategies: (i) evaluation of the target reduction activity of Trx-f1 mutants containing a mutation at each Trx-f-distinctive residue; (ii) evaluation of the target reduc-  Trx-m2WT  Trx-m2f135-143  Trx-m2f152-153  Trx-m2f167-172  Trx-m2f135-172  ACHT1  ACHT2a ACHT4a Trx  3). Each symbol indicates a significant difference (*, p Ͻ 0.05; †, p Ͻ 0.01; ‡, p Ͻ 0.001; Welch's t test) between the designated value and that of Trx-m2 WT . C, FBPase reduction activities of Trx-m2 variants. After 30 min of incubation of 1 M oxidized (Ox) FBPase with 2 M Trx in the presence of 0.5 mM reduced (Red) form of DTT, the fraction of reduced FBPase was determined via a thiol group modification-based method followed by band intensity measurement. D, amino acid sequence comparison of A. thaliana Trx-f1, Trx-f2, Trx-m2, and ACHT proteins. The key residues for Trx-f and the active-site residues are highlighted in black and orange, respectively. Residue numbers of A. thaliana Trx-f1 (counted from the translational start site amino acid methionine) are shown above the figure. The phylogenetic tree is shown with bootstrap values (%). E, FBPase reduction activities of ACHT proteins. FBPase reduction by ACHT proteins was monitored as described in C.

Determinant of the target selectivity of chloroplast Trx-f
tion activity of Trx-m2 mutants containing some of the Trxf-distinctive residues; and (iii) evaluation of the target reduction activities of ACHT1 and ACHT2a, two Trx-like proteins that possess the Trx-f-distinctive residues. The results obtained from the strategy i indicate that most of the Trx-f-distinctive residues were crucial for the reduction of Trx-f-specific targets, whereas these residues did not greatly affect the reduction capacity of Trx itself (Figs. 2 and 3). In this experiment, mutation at Asn-151 did not alter the reduction activities of Trx-f. This result is consistent with the fact that Asn-151 is found at the opposite surface from the position of the active site in the molecule (Fig. 1B). The negatively-charged molecular surface around the regulatory Cys residues of targets was reported to be preferentially recognized by the compatible positively-charged surface of Trx-f (24 -26, 28, 42, 43). These observations are consistent with our finding that the key residues for Trx-f contain positively charged residues compared with the other types of Trxs. In addition to these positively-charged residues, we demonstrated that most of the residues that are specifically conserved in Trx-f are involved in target recognition. The results of MD simulations clearly show these interactions ( Fig. 4 and Movies S1-S3). Although a rigid-body docking simulation of Trx-f and FBPase has already been conducted (43), their interaction should be confirmed in a dynamic structure because regulatory Cys residues of target proteins are mostly located on the flexible loops (42). Indeed, our results certainly show that the key residues for Trx-f interact with FBPase along with conformational changes (Movies S1-S3). Hence, such a dynamic structure or solution structure might be necessary to explore residues in both Trx and target that involved in their interaction. Although residues Asn-127 and Glu-156 could not make contact with the target protein molecule in our model within the calculation period, the distance between these residues and the target protein molecule was also continuously changing (Fig. 4B), implying that these residues may interact with the target protein during longer time-scale simulations.
The key residues for Trx-f are almost completely conserved among f-type Trxs in vascular plants (Fig. 10). This fact indicates that the target recognition mechanism of Trx-f is broadly conserved among vascular plants. Moreover, key residues for Trx-f are also conserved in moss and green algae Trx-f proteins, excluding one of the Trx-f isoforms of green algae (Fig. 10), whereas their mature protein sequence identity against A. thaliana Trx-f1 ranged from 40 to 67%. Trx-f from Physcomitrella patens lacks two of the key residues for Trx-f. This fact is in line with the previous result that P. patens Trx-f could reduce CROST1, but the efficiency was lower than that of A. thaliana Trx-f1 (31). We also compared amino acid sequences of Trx-f and NTRC, and we found that NTRC proteins have only a few residues of the key residues identified on Trx-f (Fig. 9C). Supportive of this alignment, NTRC has no reduction activity or very low reduction activity against Trx-fspecific targets in vitro (44), whereas NTRC is suggested to be (directly and/or indirectly) involved in reduction of these targets in vivo (19).
From the results of experiments ii and iii, we could validate the significance of the key residues on Trx-f for target recognition (Fig. 5). Together with the result of experiment i, our study revealed a hierarchy of the impact of the residues involved in the selectivity. In these key residues, the portion including ␣-helix 3 appeared to be most important for determining selectivity (see Trx-m2 f135-143 mutant in Fig. 5, B and  C). It must be noted that even the combination of all of the feasible mutations (f135-172) could not increase the reduction activity of Trx-m2 to the same level as that of Trx-f1 (Fig. 5, B and C). Residues of Trx-f1, such as Met-94 and Thr-96, could have a certain role in this process, although we could not prepare mutants at these positions because of the insolubility of recombinant proteins. Another candidate residue involved in the interaction is Asn-130. We did not include this residue as a potential key residue for Trx-f because a previous study revealed that the ambiguous assignment of this residue is Trx-f-specific depending on the alignment (24). Nevertheless, the apparent capacities of Trx-m2 mutants, ACHT1 and ACHT2a, to reduce Trx-f-  (n ϭ 3). B, E m of ACHT4a was not determined because multiple band-shift patterns appeared, and the redox states could not be assigned. C, E m values of typical Trx, TrxL2, and ACHT proteins at pH 7.5. The E m values of typical Trx and TrxL2 proteins were determined in our previous research (18,21,32), and those of ACHT proteins determined in this study are summarized. These E m values were determined by using the same method.

Determinant of the target selectivity of chloroplast Trx-f
specific targets indicate that the key residues identified on Trx-f are the major determinants of their target selectivity.

On the oxidation process
As mentioned previously, one of the determinants of the rate and direction of redox reaction is the E m values of the electron donor and acceptor, although their physical interaction is certainly a prerequisite for the reaction. Because ACHT1 and ACHT2a could efficiently perform the dithiol-disulfide exchange reaction with Trx-f-specific targets (Fig. 5, B and E), we determined their E m values, and they were remarkably high similar to those of TrxL2 proteins (Fig. 6). Because the amino     Figure 9. Conservation of the key residues for Trx-f in TrxL2 and NTRC molecules. A, amino acid sequence comparison of Trx-f1 and TrxL2 proteins. The residue numbers of Trx-f1 and TrxL2.1 counted from the translational start site amino acid methionine are indicated above (black) and below (sky blue) the alignment result, respectively. B, structure comparison of Trx-f1 (light gray) and TrxL2.1 (sky blue). A and B, residues of TrxL2.1 corresponding to the key residues for Trx-f are highlighted in green. C, amino acid sequence comparison of Trx-f1 and Trx domains of plant NTRCs. The residue numbers of A. thaliana NTRC counted from the translational start site amino acid methionine are shown above the figure. Residues highlighted in magenta are the cis-ProϪ1 residues of NTRC. Full names of the organisms are described in Table S1. A-C, residues highlighted in orange and black are the active site and the key residues for Trx-f, respectively.  Table S1.

Determinant of the target selectivity of chloroplast Trx-f
acid residues around active-site cysteines and the cis-ProϪ1 residue influence the E m value of Trx (20,(45)(46)(47), these residues might increase the E m values of ACHT1 and ACHT2a as well. We therefore examined whether ACHT1 and ACHT2a can also oxidize Trx-f-specific targets. As shown in Fig. 7, they could oxidize both CROST1 and FBPase more efficiently than Trx-f1. The efficiency was also higher than that of ACHT4a (Fig. 7) and TrxL2 proteins (refer our previous results shown in Refs. 31,32), probably because only the part of the key residues for Trx-f are conserved in ACHT4a and TrxL2 compared with the sequences of ACHT1 and ACHT2a (Figs. 5D, and 9, A and B). Because ACHT1 and ACHT2a lack several key residues, it is possible that introduction of these residues to ACHT1 or ACHT2a increases their reduction and oxidation activities.
As discussed previously, ACHT1 and ACHT2a can receive reducing power from their targets. We therefore investigated the acceptor of the reducing power and found that they can efficiently transfer reducing power to 2-Cys PrxA and finally to H 2 O 2 (Fig. 8), similarly as TrxL2 proteins (32,48). The interaction between ACHT proteins and 2-Cys Prx in vivo and in vitro has already been reported (33, 48 -50), suggesting that 2-Cys Prx is also an efficient electron acceptor for ACHT proteins, although these previous studies did not present direct evidence of target oxidation by ACHT1 or ACHT2a. In addition, the most recent studies indicated that 2-Cys Prx is the major acceptor of the reducing power of Trx-targeted proteins in chloroplasts (32,(51)(52)(53). Although the major reductant of 2-Cys Prx is NTRC, a part of oxidizing power via 2-Cys Prx might be used for the oxidation of redox-regulated proteins via oxidation factors in vivo. Our results in this study, together with these previous findings, strongly suggest that ACHT1 and ACHT2a are efficient oxidation factors at least for Trx-f-specific target proteins, and they ultimately transfer reducing power to H 2 O 2 via 2-Cys Prx (Fig. 11).
In this study, we identified the key residues for Trx-f-specific target recognition. In addition, we successfully uncovered the roles of ACHT1 and ACHT2a as oxidation factors. Our results imply the existence of target selectivity in the oxidative regulation of photosynthesis, and this selectivity can be inferred from the information on the target selectivity of typical Trxs and amino acid sequences of Trx-like proteins.

In silico analyses
The amino acid sequences presented in this study were obtained from the UniProtKB database (https://www.uniprot. org/) 3 and the Arabidopsis Information Resource database. Their accession numbers are listed in Table S1.
Multiple sequence alignment was performed using the Clustal Omega program (https://www.ebi.ac.uk/) 3 with default parameters. Using the result of the multiple sequence alignment and MEGA7 software (54), phylogenetic relationships were inferred via the neighbor-joining method with 500 bootstrap replicates.
Models of the three-dimensional structures of AtTrx-f1 and TrxL2.1 were built using the SWISS-MODEL program (https:// swissmodel.expasy.org/) (56) with their amino acid sequences and the structures of SoTrx-f (PDB code 1F9M) and poplar TrxL2.1 (PDB code 5NYK) as templates. The ribbon structure and molecular surface of the models were visualized using UCSF Chimera software (57).

Modeling
To construct an atomic model of the complex structure of Trx-f and FBPase, first we performed a protein-protein docking calculation between spinach Trx-f (1F9M.pdb (58)) and pea FBPase (1D9Q.pdb (59)) using rigid-body docking ZDOCK 3.0.2 (60,61). The dense mode (grid search of the rotational space at 6°increments) was used, and the docking procedure was repeated 10 times to generate 540,000 complex models. Among them, optimum models were chosen whose distances between the Cys residues of Trx-f (Cys-46 and Cys-49) and those of FBPase (Cys-153 and Cys-173) are the closest. Three models were selected based on the sulfur-sulfur distances of the Cys residues in the two proteins. To conduct structure refinement of these models, 500-ns MD simulation was performed for each model. MD simulations were performed using Gromacs 2018.1 (62,63). The system was brought to thermodynamic equilibrium at 300 K and 1 atm using the Nosé-Hoover thermostat and the Parrinello-Rahman barostat. The equations of motion were integrated with a time step of 2 fs. Long-range Coulomb energy was evaluated using the particle mesh Ewald (PME) method. The ff14SB force field (64) and SPC/Eb model (65) were used for protein and water models, respectively. The best model was selected based on the sulfur-sulfur distances obtained from MD simulations. Employing a representing structure of the spinach Trx-f/pea FBPase complex as a template, we next constructed an atomic model structure of the A. thaliana Trx-f1/FBPase complex. The structure used for the homology modeling was the representative structure of the dominant cluster obtained by clustering the MD trajectory, and 3 Please note that the JBC is not responsible for the long-term archiving and maintenance of this site or any other third party hosted site.

Determinant of the target selectivity of chloroplast Trx-f
the homology modeling was conducted with the SWISS-MODEL server (56). Finally, we performed 300-ns MD simulations of the obtained model structure in explicit water to refine the structure and analyzed inter-molecular interactions.

Preparation of expression plasmids
All proteins used in this study, excluding CROST1, were derived from A. thaliana. Total RNA was extracted from frozen Arabidopsis leaves using the TRIzol reagent (Invitrogen) according to the manufacturer's instructions. cDNA was synthesized via RT-PCR with extracted RNA and was used for expression plasmid construction.
Mature protein-coding regions of ACHT1, ACHT2a, and ACHT4a described in previous research (48) were cloned into the pET-23c vector (Novagen) to express C-terminally Histagged recombinant proteins. Expression plasmids for Trx-f1, Trx-m2, PrxQ, 2-Cys PrxA, and CROST1 were constructed previously (18,31,44). For FBPase expression, the previously constructed plasmid (18) was modified to enhance its expression in E. coli via site-directed mutagenesis in accordance with the 31C-FO H method described by Boël et al. (66). To express Trx-f1 mutants and Trx-m2 mutants, mutations were introduced into each expression plasmid via site-directed mutagenesis. Site-directed mutagenesis was performed using a Prime-STAR mutagenesis basal kit (Takara) according to the manufacturer's instructions. The primers used for cloning and site-directed mutagenesis are shown in Table S2.

Protein expression and purification
E. coli strain BL21(DE3) pLysS (for Trxs and FBPase) or strain BL21(DE3) (for ACHTs, CROST1, and Prxs) was transformed with each expression plasmid. Transformed cells were cultured at 37°C, and expression was subsequently induced by adding isopropyl ␤-D-thiogalactopyranoside (final concentration, 0.5 mM) followed by further culture for 16 h at 21°C. Collected cells were disrupted and centrifuged (125,000 ϫ g for 40 min) to obtain the soluble fraction containing the protein of interest for subsequent purification. CROST1 was purified as described previously (31). Trxs and ACHTs were purified via a combination of nickel-nitrilotriacetic acid affinity chromatography with nickel-nitrilotriacetic acid-agarose (Qiagen), cation-exchange chromatography (CEC) with TOYOPEARL SP-650 M (Tosoh) (this step was only for ACHTs), and sizeexclusion chromatography (SEC) with Superdex 75 Increase 10/300 GL (GE Healthcare). The other proteins were purified via anion-exchange chromatography with TOYOPEARL DEAE-650 M (Tosoh) (for FBPase and 2-Cys PrxA) or CEC with TOYOPEARL SP-650 M (for PrxQ), followed by hydrophobic-interaction chromatography (HIC) with TOYOPEARL Butyl-650 M (Tosoh). Trxs, and ACHTs were treated with 2 mM reduced DTT for 15 min before SEC. FBPase and Prxs were treated with 1 mM diamide and 0.5 mM H 2 O 2 , respectively, for 1 h after HIC. Finally, medium containing each protein was replaced by buffer containing 25 mM Tris-HCl (pH 7.5) and 150 mM NaCl (for Trxs); 25 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 50 M EDTA (for ACHTs); or 25 mM Tris-HCl (pH 7.5) (for the others) during SEC; alternatively, dialysis was performed after HIC. All purification steps were performed at 4°C. The concentrations of the purified proteins were determined using a BCA protein assay kit (Pierce).

Insulin reduction assay
The capacity of Trx to reduce disulfide bonds was determined via an insulin reduction assay according to previous studies with minor modifications (18,36). The reaction was initiated by adding reduced DTT (final concentration, 0.5 mM) into a reaction mixture containing 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1 mM EDTA, 230 M bovine insulin (Sigma), and 5 M Trx. Turbidity, which resulted from reduced insulin, was monitored at 650 nm at 25°C. Insulin reduction activity (⌬OD 650 nm min Ϫ1 ) was determined via linear regression of the data, for which OD 650 nm ranged from 0.2 to 0.8.

Target reduction and oxidation assays
For the target reduction and oxidation assays, medium containing 1 mM EDTA (only for CROST1), 50 mM Tris-HCl (pH 7.5), and 50 mM NaCl was used, and the reaction was performed at 25°C. The redox states of CROST1 and FBPase were monitored via the emission of fluorescence at 480 and 530 nm, and the thiol group modification-based method, as described previously (18,31). For the target oxidation assay, targets were reduced before the assay as described previously (31,32). For the FBPase oxidation assay, 0.6 M Trx-f1 or ACHT was preincubated in medium containing 25 mM Tris-HCl (pH 7.5), 50 mM oxidized DTT, and 1 M reduced DTT for 3 h at 25°C, and DTT was then removed via SEC. The concentrations of proteins and DTT used for each reaction are shown in each figure legend. Target reduction activities were calculated by fitting the data of the time-dependent reduction assay to Equation 1 built in accordance with a previous study (31), where k (M Ϫ1 min Ϫ1 ) is the rate constant used to determine the activity of Trx, and t (min) is time after the reaction was initiated.

Determination of E m
E m values of Trx-f1 mutants are roughly estimated as described under "Results." E m values of ACHT proteins were determined essentially as described previously (18). ACHT (0.6 M) was incubated for 3 h at 25°C in medium containing 25 mM Tris-HCl (pH 7.5), 50 mM oxidized DTT, and 0.0005-5 mM reduced DTT. After the incubation, the ratio of oxidized and reduced form at each redox potential was determined via the thiol group modification-based method. E m values of ACHT proteins were calculated by fitting the obtained data to the Nernst equation. The redox potential of each reaction mixture was calculated using Ϫ357 mV as E m of DTT at pH 7.5.

H 2 O 2 detoxification assay
To monitor the reduction of Prxs by Trx-f1 or ACHT, 2 M 2-Cys PrxA or PrxQ was incubated with 1 M Trx or ACHT in medium containing 50 mM Tris-HCl (pH 7.5), 50 mM NaCl, 0.5 mM reduced DTT, and 100 M H 2 O 2 at 25°C. At each indicated time, the H 2 O 2 concentration in the reaction mixture was

Determinant of the target selectivity of chloroplast Trx-f
determined via ferrous oxidation in the xylenol orange assay (67).