Towards a Functional Dissection of Thioredoxin Networks in Plant Cells


  • Toru Hisabori,

    Corresponding author
    1. Chemical Resources Laboratory, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan
    2. ATP System Project, Exploratory Research for Advanced Technology (ERATO), Japan Science and Technology Agency  (JST), Nagatsuta-cho, Midori-ku, Yokohama, Japan
    Search for more papers by this author
  • Ken Motohashi,

    1. Chemical Resources Laboratory, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan
    2. ATP System Project, Exploratory Research for Advanced Technology (ERATO), Japan Science and Technology Agency  (JST), Nagatsuta-cho, Midori-ku, Yokohama, Japan
    Search for more papers by this author
  • Naomi Hosoya-Matsuda,

    1. Chemical Resources Laboratory, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan
    Search for more papers by this author
  • Hanayo Ueoka-Nakanishi,

    1. Chemical Resources Laboratory, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan
    2. Laboratory of Cell Dynamics, Graduate School of Bioagricultural Sciences, Nagoya University, Nagoya, Japan
    Search for more papers by this author
  • Patrick G. N. Romano

    1. Chemical Resources Laboratory, Tokyo Institute of Technology, Nagatsuta, Midori-ku, Yokohama, Japan
    Search for more papers by this author

  • This invited paper is part of the Symposium-in-Print: Photobiology in Asia.

*email: (Toru Hisabori)


Thioredoxins are a ubiquitous family of redox equivalent mediators, long considered to possess a limited number of target enzymes. Recent progress in proteomic research has allowed the identification of a wide variety of candidate proteins with which this small protein may interact in vivo. Moreover, the activity of thioredoxin itself has been recently found to be subject to regulation by posttranslational modifications, adding an additional level of complexity to the function of this intriguing enzyme family. The current review charts the technical progress made in the continuing discovery of the numerous and diverse roles played by these proteins in the regulation of redox networks in plant cells.


Thioredoxin (Trx) was first identified as the source of reducing equivalents to ribonucleotide reductase in Escherichia coli (1). It is a small ubiquitous protein which catalyses the dithiol–disulfide exchange reaction. Proteins involved in thiol oxidation and reduction are critical in maintaining redox homeostasis within the intracellular environment; Trx is a key player in the sensing and transfer of reducing equivalents in chloroplasts, mitochondria and cytosol. Trx and its homologs constitute a big family, and they are defined by their conserved active site sequence (classically Trp-Cys-Gly-Pro-Cys-[Lys or Arg]), although numerous variants of this sequence exist especially in plants. Formation of a disulfide bond between the two active site cysteine residues lies at the heart of the disulfide–dithiol exchange reaction, allowing Trx to reduce disulfide bonds on a wide range of target proteins. As the regulatory mechanism commonly consists of oxidation and reduction of dithiols in the protein molecule, Trx-target enzymes are called thiol enzymes and the regulatory system is referred to as thiol modulation.

Although, by the mid-1980s the basic concepts of thiol modulation had been established based on extensive research on both Calvin cycle enzymes and other redox-sensitive enzymes, it has taken more than 20 years to begin to unravel the diverse array of functions of Trx in plant cells. Located in the chloroplast of higher plants, f-type Trx (Trx-f) and m-type Trx (Trx-m) are of significant interest and have been the subject of intense study. Within this photosynthetic organelle, both f- and m-type Trxs are known to regulate a number of Calvin cycle enzymes such as glyceraldehyde 3-phosphate dehydrogenase (GAPDH), fructose 1,6-bis phosphatase (FBPase), sedoheptulose 1,7-bis phosphatase (SBPase) and phosphoribulokinase (PRK) (2), as well as key metabolic enzymes such as malate dehydrogenase (MDH), glucose 6-phosphate dehydrogenase (G6PDH) (3) and ATP synthase (4). While the above regulatory interactions are known in detail, for example between Trx-f and FBPase, the specific function and target of each Trx isoform are yet to be determined very well.

Why do plants possess so many trx isoforms?

The completion of the genome sequence of Arabidopsis thaliana (5) revealed the presence of a number of genes encoding Trx and related proteins in plant cells (6,7). Two evolutionarily independent genes encoding the Trx-m and Trx-f isoforms are present in the nuclear genome, and their protein products are targeted to the chloroplast stroma. Based on the phylogenetic analysis of Trx isoforms, A. thaliana Trx-m groups with the prokaryotic Trxs and it is also closely related to the cyanobacterial Trxs (Fig. 1). The cyanobacterium Synechocystis sp. PCC 6803 possesses five Trx and Trx-related genes (slr0623, sll1057, slr1139, slr0233 and sll1980) (8,9), and Anabaena sp. PCC 7120 has eight genes (all1839, all1866, all2341, all2367, alr0052, alr0570, alr2205 and alr3955) (10). The products of the slr0623, all1866 and alr0052 genes are most closely related to the plant Trx-m1, Trx-m2 and Trx-m4 isoforms (Fig. 1; Prokaryotic subgroup M). Although genomic analysis of A. thaliana has revealed two additional chloroplast Trx isoforms named Trx-x (11) and Trx-y (12,13), they are yet to be identified by biochemical analysis. Phylogenetically distinct from m-type Trxs, the Trx-x protein is an ortholog of the products of the slr1139 and all2341 genes, which are Trxs found in Synechocystis PCC6803 and Anabena PCC7120 respectively (Fig. 1; Prokaryotic subgroup X). Trx-y also diverged from the m-type Trxs, and shows a strong similarity to the slr0233 and alrl1893 gene products of these cyanobacteria (Fig. 1; Prokaryotic subgroup Y). The recombinant Trx-y obtained from Chlamydomonas reinhardtiishowed a remarkable difference in the specificity of the targets, although the in vivo target proteins are not identified yet (12). No ortholog of the sll1057 and alr3955 gene products (for which we suggest the novel designation “subgroup Z”) exists in A. thaliana. The similarities observed among chloroplast and cyanobacterial Trxs are unsurprising given that cyanobacteria are known to be the evolutionary progenitors of the higher plant chloroplast.

Figure 1.

 Unrooted phylogenic tree of thioredoxin (Trx) proteins in Arabidopsis thaliana. Phylogenic relationships were analyzed with the DIALIGN multiple sequence alignment algorithm (52) and the tree was written using the software PHYLIP ver. 3.6 by Felsenstein J. Trx isoforms from Synechocystis sp. PCC 6803 and Anabena sp. PCC 7120 are written with their gene names (lower case) as reference. The gene names starting with “s” are from Synechocystis PCC 6803 and those with “a” from Anabena PCC 7120. The accession numbers of the sequence of Trx from A. thaliana are M1, NP_849585; M2, Q9SEU8; M3, AT2g15570; M4, Q9SEU6; x, Q8LD49; y1, At1g76760; y2, At1g43560; H1, P29448; H2, Q38879; H3, At5g42980; H4, Q39239; H5, Q39241; H7, At1g59730; H8, AT1g69880; H9, At3g08710; H10, AT3g56420; F1, Q9XFH8; F2, Q9XFH9; O1, At2g35010; O2, NP_564371; HCF164, AT4g37200; and CDSP32, CAC39419.

The bacterial Trxs, including those of photosynthetic bacteria, can be categorized into subgroup M. The mammalian mitochondrial Trx occurs on a separate branch from the large prokaryotic group (data not shown) and is phylogenetically distinct from Trxs found in photosynthetic bacteria, organisms believed to be the evolutionary origin of mitochondria. The plant h-type (“heterotroph” type) Trxs form another large group, and the chloroplast f-type isoform appears to have branched off from h-type Trxs at an early stage in the evolution of Trx proteins. Interestingly, the recently identified mitochondrial Trx-o isoform also groups with the eukaryotic Trx branch. Trx-o1 was located in mitochondria but its physiological function has not yet been determined (14).

While the phylogenetic analysis presented elegantly demonstrates the evolution of plant cell organelles through endosymbiosis of their prokaryotic ancestors, the reason for the presence of multiple Trx isoforms in both the chloroplast and cytosol is still unclear. This multiplicity differs considerably from mammalian cells. In order to address this conundrum, the specific location and function of each plant Trx isoform must be determined.

Development of effective methodologies for comprehensive screening of plant trx-target proteins

As stated above, our knowledge of the Trx-target proteins or thiol enzymes in plant cells has remained limited for a long time. Following its discovery as the source of reducing equivalents to ribonucleotide reductase, Trx was often used in conjunction with the reductant dithiothreitol (DTT) during analysis of newly discovered enzymes, to avoid potential inactivation of the enzyme of interest caused by unexpected oxidation. Trx was thus pigeonholed as a source of reducing equivalents and its potential to interact with multiple targets was somewhat overlooked. Thus, until recently, research in the Trx field has made limited progress.

To identify the Trx-target proteins, the knowledge of the shared properties of the target proteins must be ascertained. However, what determines whether a protein will be targeted by Trx? No sequence similarity has been observed in proximity to the Trx targeted cysteines in the reported thiol enzymes. In addition, based on the reported structures of the thiol enzymes GAPDH, FBPase, MDH and G6PDH, no similarity exists in the exposed target disulfide bond constituting the Trx docking site, suggesting that Trx recognizes the surface target disulfide bond with fairly low selectivity. These findings suggest that the chloroplast must contain numerous as yet unidentified Trx targets.

In the last 5 years, a number of experimental strategies have been developed to isolate and identify Trx-target proteins. In one of the first attempted screens of Trx-target proteins, Yano et al. (2001) carried out 2D gel electrophoresis analysis of peanut seed total proteins by specifically labeling Trx-target proteins with the thiol modifier monobromobimane (15). In their study, the proteins reduced by Trx occurred as specific fluorescence labeled spots on a 2D gel. This technique allowed them to identify three Trx-targeted allergen proteins in addition to several previously unknown proteins. More recently, we have developed a novel affinity chromatography-based method for the identification of Trx-target proteins in plant chloroplasts. In this technique, resin-immobilized mutant Trx, containing a single cysteine to serine substitution within its active site, is used as a “bait” to “trap” potential Trx-target proteins (16). By applying stromal protein extract to the resin-immobilized mutant Trx, we obtained several target protein candidates, including the previously determined chloroplast GAPDH and SBPase enzymes, as well as Rubisco activase, which was recently confirmed as a thiol enzyme. Following the development of our pioneering screening strategy, both the above techniques have been extensively applied to help identify Trx-target proteins in various organelles from a wide range of organisms (17).

Discovery of trx-target proteins in chloroplasts using the proteomic approach: a cautionary note

The invaluable impact of the proteomic approach on Trx research is apparent from Figs. 2 and 3; before the development of this technique only five enzymes were known as thiol enzymes involved in the Calvin cycle. However, with the widespread adoption of this novel methodology, at least five additional potential Calvin cycle proteins, as well as carbonic anhydrase, have been identified as potential Trx targets in A. thaliana and C. reinhardtii. While the proteomic approach has proved to be an excellent tool for the discovery of Trx-target protein candidates (see Fig. 2), the methodical confirmation of potential interactions between Trx and the newly identified target enzymes is critical for the consolidation of any initial data obtained. For example, following its identification as a potential Trx target by the proteomic approach, we synthesized recombinant ribose 5-phosphate isomerase (IX in Fig. 3) in order to investigate its redox sensitivity. To date, we have detected neither a redox regulation of the activity of this enzyme by Trx nor any Trx-dependent reduction of a disulfide bond (H. Ueoka-Nakanishi and T. Hisabori, unpublished results). The results found are in agreement with homology modeling of this protein, which shows that the two proposed conserved cysteine residues are located at opposite ends of an α-helix, making the formation of a disulfide bond between them highly unlikely. One possibility is that ribose 5-phosphate isomerase may form a supercomplex with other Calvin cycle enzymes. There are other explanations such as Trx could be able to reduce oxidized forms of the Cys residues or may be glutathione adducts.

Figure 2.

Thioredoxin (Trx) networks in chloroplasts and cytosol. In the chloroplasts, the enzymes and pathways whose redox regulation has already been determined are shown in white letters. Other proteins and pathways listed are based on information obtained from proteomic analyses of Trx counterpart proteins (16,22,49). Dotted arrows show the flow of electrons.

Figure 3.

 Redox regulation of Calvin cycle-related enzymes. Known thiol enzymes are shown with Roman numerals in closed circles and recently suggested candidates are in open circles. “C” plus numbers are short for the carbohydrate in the Calvin cycle. The listed enzymes are I, glyceraldehyde 3-phosphate dehydrogenase (53); II, fructose 1,6-bis phosphatase (54); III, sedoheptulose 1,7-bis phosphatase (55); IV, phosphoribulokinase (56); V, Rubisco activase (26); VI, triose phosphate isomerase (18,22); VII and VIII, transketolase (18,22); IX, ribose 5-phosphate isomerase (18); X, ribulose 5-phosphate epimerase (18); XI, carbonic anhydrase (18); and XII, CP12 (18). As we failed to confirm the redox regulation of ribose 5-phosphate isomerase (IX), “?” was appended (see text).

Our findings underline the need for a step-by-step confirmation procedure, critical in determining whether the enzymatic activity of any potential Trx-target proteins is actually regulated by Trx. As another case in point, CP12 has been reported as a redox-dependent regulator of GAPDH and PRK through the formation of a supercomplex with these proteins (Fig. 3; XII) and a mixed disulfide with Trx (18). As CP12 contains four cysteines (19), which can form two intramolecular disulfide bonds, it is possible that this protein may be targeted by Trx, although direct interaction between the disulfide bonds on CP12 and chloroplast Trxs is yet to be determined.

Target protein determinants for recognition by trx

The catalytic mechanism of Trx is extensively studied and the significance of the conserved buried Asp residue in plant Trx in addition to Trp in the active site sequence motif is suggested based on the structure analysis of the h-type Trx (20,21). In contrast, the molecular recognition mechanism of target proteins by Trx has long attracted significant attention in this field. Although clearly an excellent method for identifying novel target proteins, results obtained from Trx-affinity chromatography have shed limited light on this matter. Surface charges surrounding the reactive cysteines of the Trx active site vary considerably between Trx-f and Trx-m. However, Trx-affinity chromatography shows no substantial differences in the specificities of Trx isoforms for target protein candidates (16,22). Unfortunately, the Trx-affinity chromatography does not exclusively trap target proteins on the immobilized Trx through the formation of mixed-disulfide intermediates. In fact, one drawback of this method is that the protein pool used for affinity chromatography needs to be incubated with the immobilized Trx mutants for significant time in order for sufficient interaction to be established. Because of the extended incubation time, most of the candidate proteins will be trapped on the immobilized Trx, irrespective of their affinity; once a target protein forms a mixed-disulfide intermediate with the immobilized Trx mutant, the bond cannot be rereduced during this incubation period under the experimental conditions used.

To this end, biochemical analysis of the reduction process is required to determine the specificity of Trx isoforms. An insight into this specificity has been obtained from detailed investigations into the kinetics of activation by reduction of the γ subunit of chloroplast ATP synthase, in the presence of Trx-f or Trx-m (23). In this case, Trx-f preferentially acts as a reducing equivalent transducer for ATP synthase, and acetyl-CoA carboxylase shows a similar tendency (24,25). Reductive activation of Rubisco activase is preferentially promoted by Trx-f (26), and its activity is unaffected by incubation with Trx-m and DTT, even though this enzyme was captured by Trx-affinity chromatography (16). In contrast, Trx-m is more efficient than Trx-f both for the activation of spinach PRK (27) and for the reduction of G6PDH, one of the two chloroplast enzymes inactivated by reduction (28). The lumenal protein FKBP13 was also recently found as a redox-sensitive enzyme whose activity is inactivated by reduction (29).

In order to determine the specificity of f- and m-type Trxs for recognition of their target proteins, Geck et al. investigated the role of critical amino acids (determined by sequence comparison between Trx-f, Trx-m and E. coli Trx) close to the active site residues by site-directed mutagenesis (30). They found several amino acid residues around the active site cysteines, Lys-58, Asn-74, Gln-75 and Asn-77 of Trx-f are important for its function. In contrast, by studying the electrostatic interactions between Trx and target enzymes, Wolosiuk and coworkers concluded that the charged residues around the nucleophilic cysteine of the active site are important for the interaction (31,32). Related work by Bunik et al. (1999) on the interaction between various Trx isoforms and 2-oxoacid dehydrogenase suggests that the polarization of regions adjacent to the Trx active site is a critical determinant of the efficiency of the interplay between Trx and this 2-oxoacid dehydrogenase (33). These reports indicate that the major determinant of the specificity of Trx is not attributed to specific amino acid residues around the active site cysteines, but that the surface charge of Trx itself is critical to distinguish its interacting counterparts. 3D structural determination of the cocrystals of Trx and their target proteins may be instrumental in providing additional information required to resolve this question.

Heterologous complementation of Trx function in yeast has also been used to help understand the determinant of the specificity of Trx to its target proteins (34). This study clearly showed a distinct affinity of specific Trx-m isoforms as in vivo electron donors for the reduction of peroxide molecules. Comparison of the prokaryotic-type chloroplast Trxs (Trx-m1 to m4 and Trx-x) shows that Trx-m3 cannot confer tolerance to peroxides when expressed in yeast, which is unsurprising given the evolutionary divergence of this isoform from other m-type Trxs (Fig. 1). The functional divergence of Trx-m3 from other Trx-m isoforms has also been shown by in vitro experiments on the activation of chloroplast MDH (11).

Trx-like proteins in chloroplasts: hcf164 and cdsp32

Biochemical studies have confirmed the presence of two Trx-like proteins in A. thaliana. Although they show little homology to other Trx isoforms (Fig. 1), the HCF164 protein groups with a Trx-like protein (the gene products of sll1980 and alr0570) found in cyanobacteria. The HCF164 protein was originally identified in an A. thaliana screening for mutants that displayed a high chlorophyll fluorescence (hcf) phenotype, where it was found to be involved in the biogenesis of the cytochrome b6f complex of thylakoid membranes (35). HCF164 is a thylakoid membrane anchored protein, with the Trx-like domain facing the lumenal side. hcf164 mutant plants show a strong suppression in the accumulation of cytochrome b6f complex in the thylakoid membranes.

Recently, a novel protein containing a CXXC motif homologous to that of the prokaryotic thiol disulfide transporter CCDA has been identified in A. thaliana (36). CCDA was originally found as a protein required for bacterial cytochrome c biosynthesis, and designated from its cytochrome c defective (Ccd) mutant phenotype (37). Deletion of the CCDA protein in A. thaliana also results in impaired accumulation of cytochrome b6f complex, and the authors suggest that CCDA may function as an electron shuttle from the stroma to the lumen, to supply reducing equivalents to HCF164. These findings raise the question of whether cytochrome b6f biogenesis is the sole consumer of the reducing equivalents in the lumen, although the protein which is reduced by HCF164, and probably assists the assembly of this complex, is yet to be identified. Recent evidence showing that a lumenal immunophilin called AtFKBPI3, also shown to be required for regulating cytochrome b6f accumulation (57), is a redox-active enzyme (29) suggests the existence of multiple potential targets of redox regulation in this chloroplast subcompartment. We recently succeeded to show the reduction of HCF164 by stroma-side Trx (58). In addition, we found that the reduced form HCF164 can interact with couple of membrane proteins including cytochrome b6f complex via the formation of the mixed-disulfide intermediates (58).

Originally identified in Solanum tuberosum as a 32 kDa chloroplastic drought-induced stress protein (38) CDSP32 is a Trx-like protein located in the chloroplast stroma, which accumulates under drought and oxidative stress conditions (39). Consisting of a duplicate copy of the Trx-fold and the C-terminal redox active motif 160HCGPC165V, the Trx activity of CDSP32 has been confirmed in vitro, and a single cysteine mutant has been used for affinity chromatography (40). By using a total protein extract from Arabidopsis leaf tissue, the chloroplast 2-Cys type peroxiredoxin BAS1 was captured, suggesting strong affinity of CDSP32 to BAS1. The physiological significance of CDSP32 and BAS1 as antioxidative stress proteins has been extensively corroborated in vivo (40,41). Although the protein which supplies reducing equivalents to CDSP32 is yet to be identified, the CDSP32–BAS1 system is likely to be important for protecting plants from oxidative damage. Moreover, the specificity of this interaction may be instrumental for our understanding of the target recognition mechanisms of the Trx protein family.

Trx is not just a mediator for the dithiol–disulfide exchange reaction

Trx has long been considered the key mediator as a reducing equivalent transducer for the dithiol–disulfide exchange reaction. As such, the function of Trx has been predominantly studied in the context of its interaction with ferredoxin-Trx reductase or nicotinamide adenine dinucleotide phosphate (NADPH)-Trx reductase, which use reducing equivalents derived from photosynthetic electron transport and cytosolic NADPH, respectively. However, recent progress in mass spectrometry has allowed the determination of small molecular modifications made to proteins, leading to the discovery that Trx itself may be subject to different types of posttranslational modification. Glutathionylation of Trx was first observed in human Trx (42), and later also noted in the poplar Trx-h isoform Trx-h2 (43). Most recently Michelet et al. reported the glutathionylation of chloroplast Trx-f (44). Glutathionylation occurs at an extra cysteine residue surrounded by positively charged amino acids located away from the active site, causing a decrease in the reductive efficiency of Trx towards its target proteins. The finding that glutathionylation of human Trx occurs following oxidative stress (a condition in which cellular GSH levels are high) suggests that this modification may constitute an important regulatory mechanism of Trx function. Following Trx-f glutathionylation, efficiency of the reduction/activation of the chloroplast thiol enzymes MDH and GAPDH is also partly affected. Although the cellular conditions required to trigger Trx-f glutathionylation remain unknown, this kind of modification may be an important regulator of Trx reduction efficiency in vivo.

Nitrosylation of Trx at the extra cysteine position described above has been found to be a second type of Trx modification, although it has only been documented for human Trx (45). Nitrosylation of Trx results in the activation of the apoptosis signal-regulating kinase 1, which in turn triggers apoptosis. In contrast, Mitchell and Marletta recently reported that the extra cysteine (Cys73) of Trx is required for the transnitrosylation reaction with caspase-3, the cysteine proteinase and key enzyme required for apoptosis (46). In particular, they found that S-nitrosylation of the caspase-3 cysteine residue is mediated by the extra cysteine of Trx, which in turn is nitrosylated by S-nitrosoglutathione, with the S-nitrosylation of caspase-3 being the main cause of inhibition of apoptosis by nitric oxide.

In contrast, little is known about the role of S-nitrosylation in the control of plant protein function. A recent proteomic screen of S-nitrosylated proteins of A. thaliana allowed the identification of 63 proteins from cell culture and 52 proteins from leaves as potential S-nitrosylated proteins (47). Proteins identified included stress-related proteins, redox-related proteins, proteins related to protein synthesis, cytoskeletal proteins, metabolic enzymes and components of the photosynthetic apparatus. Interestingly, many of the proteins identified were found to overlap with Trx-target proteins previously reported from the plant chloroplast and cytosol (16,22,48–51). Lindermayr et al. also found Trx-f to be S-nitrosylated and suggest that nitrosylation occurs at the active site cysteine residues (47). However, it is likely that, given the similarity in crystal structure between plant and mammalian Trx, the target cysteine may be the extra cysteine located at the similar position which is nitrosylated in the mammalian Trx.

Although there is still no concrete evidence regarding the regulation of Trx function in plants via these posttranslational modifications, they may constitute a cross-talk point of the redox regulation system and other signal transduction systems in the plant cell.

Concluding remarks

While redox-mediated responses form the basis of important regulatory responses to oxidative stress and fluctuations in environmental conditions (e.g. the dark–light cycle), the precise roles of Trx in the maintenance of redox homeostasis within the cell, as a modulator of other redox-sensitive enzymes and as a source of reducing equivalents for enzymes such as peroxiredoxin are yet to be determined. The recent identification of novel interacting protein partners and posttranslational modifications of Trx underlines the importance of this protein family as a key player in metabolic regulation. However, the reason for the abundance of Trx isoforms in plants and the nature of their interacting partners are still unclear. The Trx system neatly exemplifies how static organisms such as plants are able to efficiently maintain redox homeostasis under changing environmental conditions.

Acknowledgements— This work was supported by a Grant-in-Aid for Scientific Research (17370015 and 17GS0316 to T.H.) from the Ministry of Education, Culture, Sports, Science and Technology, Japan and by the Japan Society for the Promotion of Science. The work is partially supported by ATP System Project, ERATO, JST.