Glutathione in plants: an integrated overview

Authors


  • GSH is used here to indicate the thiol (reduced) form of glutathione while GSSG denotes the disulphide form. The term ‘glutathione’ is used where no distinction is drawn or both forms may be concerned.

G. Noctor. e-mail: graham.noctor@u-psud.fr

ABSTRACT

Plants cannot survive without glutathione (γ-glutamylcysteinylglycine) or γ-glutamylcysteine-containing homologues. The reasons why this small molecule is indispensable are not fully understood, but it can be inferred that glutathione has functions in plant development that cannot be performed by other thiols or antioxidants. The known functions of glutathione include roles in biosynthetic pathways, detoxification, antioxidant biochemistry and redox homeostasis. Glutathione can interact in multiple ways with proteins through thiol-disulphide exchange and related processes. Its strategic position between oxidants such as reactive oxygen species and cellular reductants makes the glutathione system perfectly configured for signalling functions. Recent years have witnessed considerable progress in understanding glutathione synthesis, degradation and transport, particularly in relation to cellular redox homeostasis and related signalling under optimal and stress conditions. Here we outline the key recent advances and discuss how alterations in glutathione status, such as those observed during stress, may participate in signal transduction cascades. The discussion highlights some of the issues surrounding the regulation of glutathione contents, the control of glutathione redox potential, and how the functions of glutathione and other thiols are integrated to fine-tune photorespiratory and respiratory metabolism and to modulate phytohormone signalling pathways through appropriate modification of sensitive protein cysteine residues.

INTRODUCTION

Glutathione (or a functionally homologous thiol) is an essential metabolite with multiple functions in plants (Fig. 1). The fundamental and earliest recognized function of glutathione is in thiol-disulphide interactions, in which reduced glutathione (GSH) is continuously oxidized to a disulphide form (GSSG) that is recycled to GSH by NADPH-dependent glutathione reductase (GR). The central role of glutathione in defence metabolism in animals was established long ago, largely because selenium-dependent glutathione peroxidase (GPX) is a central pillar of animal antioxidant metabolism. Pharmacologically induced GSH deficiency in newborn mammals such as rats and guinea pigs leads to rapid multi-organ failure and death within a few days (Meister 1994). In plant cells, where reductive H2O2 metabolism has been linked to ascorbate since the 1970s, the absolutely irreplaceable role of glutathione was less apparent. However, glutathione depletion in Arabidopsis knockouts lacking the first enzyme of the committed pathway of GSH synthesis causes embryo lethality (Cairns et al. 2006). Thus, both plant and mammalian cells rely on at least some of the multifunctional properties of glutathione for their vigour and survival.

Figure 1.

General overview of some of the most important glutathione functions (synthesis, redox turnover, metabolism, signalling). Cys, cysteine; γ-EC, γ-glutamylcysteine; GS-conjugates, glutathione S-conjugates; GSNO, S-nitrosoglutathione; Glu, glutamate; Gly, glycine; RNS, reactive nitrogen species; ROS, reactive oxygen species.

Glutathione is the principal low-molecular-weight thiol in most cells. However, there are some intriguing variations in organisms such as halobacteria, in which GSH can be replaced by other sulphur compounds like γ-glutamylcysteine (γ-EC) and thiosulphate (Newton & Javor 1985). Similarly, in some parasitic protozoa, trypanothione [N1,N8-bis(glutathionyl)spermidine] can substitute for glutathione (Fairlamb et al. 1985). Some plant taxa contain glutathione homologues, in which the C-terminal residue is an amino acid other than glycine (Rennenberg 1980; Klapheck 1988; Klapheck et al. 1992; Meuwly, Thibault & Rauser 1993). These compounds include homoglutathione (γ-Glu-Cys-β-Ala), which is found alongside GSH in many legumes (MacNicol 1987; Klapheck 1988). Interestingly, gene duplication during evolution has resulted in the coexistence of different synthetases that produce GSH or homoglutathione (Frendo et al. 1999). Cereals produce another GSH variant (hydroxymethylGSH; γ-Glu-Cys-Ser) through reactions that remain to be fully elucidated but which likely involve modification of GSH rather than alternative synthesis pathways (Klapheck et al. 1992; Okumura, Koizumi & Sekiya 2003; Skipsey, Davis & Edwards 2005a). Disulphide forms of these homologues are reducible by GR (Klapheck 1988; Klapheck et al. 1992; Oven et al. 2001). Therefore, current information suggests that they do not require alternative reductive systems. Novel homologues may remain to be discovered (Skipsey et al. 2005a). For example, high-performance liquid chromatography (HPLC) analysis of thiols in poplar overexpressing a bacterial form of the second enzyme of glutathione synthesis revealed two novel peaks, in addition to GSH. The peaks were particularly abundant in conditions in which leaf glycine contents are depleted such as darkness (G. Noctor, A.C.M. Arisi, L. Jouanin, C.H. Foyer, unpublished data).

Like other thiols, glutathione can undergo numerous redox reactions. As well as GSSG, oxidized forms notably include the formation of ‘mixed disulphides’ with proteins and other thiol molecules. Other oxidized forms include thiyl radicals, sulphenic acid (SOH) and, possibly, sulphinic (SO2H) or sulphonic (SO3H) acids on the cysteine moiety of glutathione, similar to their formation on protein cysteine residues. Moreover, a wide array of glutathione conjugates can be formed with endogenous and xenobiotic electrophilic species (Wang & Ballatori 1998; Dixon & Edwards 2010) while interactions with the nitric oxide (NO) system via formation of S-nitrosoglutathione (GSNO) broaden the scope of glutathione as a reservoir of signalling potential (Lindermayr, Saalbach & Dürner 2005).

It has long been recognized that GSH is oxidized by reactive oxygen species (ROS) as part of the antioxidant barrier that prevents excessive oxidation of sensitive cellular components. Unlike the oxidized forms of many other primary and secondary metabolites that can also react with ROS, GSSG is rapidly recycled by the GRs in key organelles and the cytosol (Halliwell & Foyer 1978; Smith, Vierheller & Thorne 1989; Edwards, Rawsthorne & Mullineaux 1990; Jiménez et al. 1997; Chew, Whelan & Millar 2003; Kataya & Reumann 2010). A characteristic feature of glutathione is its high concentration in relation to other cellular thiols. In general, glutathione accumulates to millimolar concentrations, with tissue contents well in excess of free cysteine. A second key characteristic of the cellular glutathione pool is its high reduction state. In the absence of stress, tissues such as leaves typically maintain measurable GSH: GSSG ratios of at least 20:1 (e.g. Mhamdi et al. 2010a). It is important to note that this is an average value across tissues, and that ratios may be higher (e.g. cytosol) or lower (e.g. vacuole) in specific subcellular compartments (Meyer et al. 2007; Queval et al. 2011).

GLUTATHIONE ACCUMULATION, TURNOVER AND TRANSPORT

GSH biosynthetic pathway

As in animals, GSH is synthesized in plants from its constituent amino acids by two ATP-dependent steps (Rennenberg 1980; Meister 1988; Noctor et al. 2002a; Mullineaux & Rausch 2005). Each of the synthetic enzymes is encoded by a single gene (May & Leaver 1994; Ullman et al. 1996), and Arabidopsis knockout lines for either have lethal phenotypes. While knocking out expression of GSH1, encoding γ-EC synthetase (γ-ECS), causes lethality at the embryo stage (Cairns et al. 2006), knockouts for GSH2, encoding glutathione synthetase (GSH-S), show a seedling-lethal phenotype (Pasternak et al. 2008). Using forward genetics approaches, several mutants have been identified in which decreased GSH contents are caused by less severe mutations in the GSH1 gene. Of these, the rml1 (rootmeristemless1) mutant, which has less than 5% of wild-type glutathione contents, shows the most striking phenotype because it fails to develop a root apical meristem (Vernoux et al. 2000). In other mutants, in which glutathione is decreased to about 25 to 50% of wild-type contents, developmental phenotypes are weak or absent, but alterations in environmental responses are observed. In cad2, lower glutathione is associated with enhanced cadmium sensitivity, whereas rax1 was identified by modified APX2 expression and pad2 shows decreased camalexin contents and enhanced sensitivity to pathogens (Howden et al. 1995; Cobbett et al. 1998; Ball et al. 2004; Parisy et al. 2006). In addition to genetic approaches, a pharmacological tool that has frequently been used to deplete glutathione is buthionine sulphoximine (BSO), a specific inhibitor of γ-ECS (Griffith & Meister 1979).

The activity of γ-ECS is strongly associated with chloroplasts in wheat (Noctor et al. 2002a). Localization studies in Arabidopsis have demonstrated that the enzyme is restricted to plastids in this species (Wachter et al. 2005). The Arabidopsis GSH-S is found in both chloroplasts and cytosol. Of the two transcripts encoded by the GSH2 gene, the most abundant one is the shorter form, which is translated to produce a cytosolic GSH-S (Wachter et al. 2005). Thus, the first step of glutathione synthesis is plastidic while the second step is probably predominantly located in the cytosol.

Regulation of biosynthesis

Many factors affect the synthesis of glutathione, but the most important are considered to be γ-ECS activity and cysteine availability. Accordingly, constitutive increases in glutathione can be produced either by overexpression of the first enzyme of the committed glutathione synthesis pathway or of enzymes involved in cysteine synthesis (Strohm et al. 1995; Noctor et al. 1996, 1998; Creissen et al. 1999; Harms et al. 2000; Noji & Saito 2002; Wirtz & Hell 2007). Other factors that may affect GSH contents in certain conditions include glycine and ATP (Buwalda et al. 1990; Noctor et al. 1997; Ogawa et al. 2004). Increased γ-ECS activity may result from transcriptional or post-transcriptional changes (May et al. 1998), but so far relatively few conditions have been shown to cause marked induction of GSH1 or GSH2 transcripts. Both genes are induced by jasmonic acid (JA) and heavy metals (Xiang & Oliver 1998; Sung et al. 2009) and also respond to light and some stress conditions such as drought and certain pathogens. However, neither externally applied H2O2 nor intracellularly generated H2O2 leads to increased abundance of GSH1 or GSH2 transcripts in Arabidopsis, despite the well-described increases in glutathione in these conditions (Smith et al. 1984; May & Leaver 1993; Willekens et al. 1997; Sánchez-Fernández et al. 1998; Xiang & Oliver 1998; Queval et al. 2009). While some evidence has been presented that production of the γ-ECS protein is regulated at the level of translation (Xiang & Bertrand 2000), more attention has focused on post-translational redox controls. A partially purified enzyme preparation from tobacco was shown to be sensitive to inhibition by dithiols (Hell & Bergmann 1990), and similar effects were subsequently reported in other species (Noctor et al. 2002a; Jez, Cahoon & Chen 2004). Recently, it has been shown that the plant γ-ECS forms a homodimer linked by two disulphide bonds (Hothorn et al. 2006), one of which is involved in redox regulation (Hicks et al. 2007; Gromes et al. 2008). This is likely an important factor in the well-known up-regulation of glutathione synthesis in response to oxidative stress. Another mode of regulation that is likely to be important in glutathione homeostasis is feedback inhibition of γ-ECS by GSH. First reported for the animal enzyme (Richman & Meister 1975), this regulatory mechanism also occurs in plants (Hell & Bergmann 1990; Noctor et al. 2002a). Alleviation of feedback inhibition is likely to be an important mechanism driving accelerated rates of synthesis under conditions in which glutathione is being consumed (e.g. in the synthesis of phytochelatins). The mechanistic links between feedback inhibition and thiol/disulphide redox regulation of γ-ECS remain to be elucidated.

Integration of glutathione synthesis with sulphur assimilation

In accordance with the observation that enhanced cysteine supply favours glutathione accumulation, increases in GSH synthesis are associated with up-regulation of the cysteine synthesis pathway. For example, glutathione accumulation triggered by oxidative stress causes accumulation of transcripts encoding adenosine 5'-phosphosulphate reductase (APR) and serine acetyltransferase (SAT; Queval et al. 2009). While all three Arabidopsis APR genes encode plastidial enzymes, only one of the five SAT gene produces an isoform located in this compartment (Kawashima et al. 2005). Although the mitochondrial SAT makes the major contribution to cysteine synthesis under standard conditions (Haas et al. 2008; Watanabe et al. 2008), the chloroplast SAT was the most strongly induced during H2O2-triggered accumulation of glutathione (Queval et al. 2009). As well as up-regulation at the transcript level, ozone exposure activates at least one APR at the post-translational level (Bick et al. 2001). As in the case of post-translational activation of γ-ECS, the exact mechanisms regulating APR activation state remain unclear. One appealing possibility is that oxidation-triggered decreases in GSH:GSSG activate glutathione synthesis by increasing the chloroplast glutathione redox potential and allowing glutaredoxin (GRX)-mediated activation of both enzymes. Consistent with this model, accumulation of glutathione in catalase-deficient barley and Arabidopsis is associated with markedly increased chloroplast GSSG content (Smith et al. 1985; Queval et al. 2011). However, regulation could also be linked to thioredoxin (TRX) activity. Whatever the details of the underlying regulatory mechanisms, intracellular oxidative stress can drive glutathione accumulation to several-fold basal levels (Smith et al. 1984; Willekens et al. 1997). It has been estimated that a fivefold accumulation of glutathione in Arabidopsis cat2 mutants means that the amount of sulphur in the glutathione cysteine residue approaches that which is found in protein cysteine and methionine residues combined (Queval et al. 2009). As well as the changes in transcripts and extractable activities previously mentioned, H2O2-triggered glutathione accumulation in barley is accompanied by increased uptake of labelled sulphate (Smith et al. 1985). Accordingly, marked accumulation of glutathione achieved by transgenic enhancement using a bacterial enzyme with both γ-ECS and GSH-S activities was dependent on sufficient sulphur supply (Liedschulte et al. 2010).

Overexpression of glutathione biosynthesis

Regardless of the controls over synthesis, and the signalling roles discussed later, tissue glutathione contents can be markedly enriched in plants. Whereas overexpression of Escherichia coli GSH-S in poplar produced little effect on glutathione contents in optimal conditions (Foyer et al. 1995; Strohm et al. 1995), introduction of the E. coli γ-ECS caused a two- to fourfold increase in leaf glutathione, and this was observed whether the bacterial γ-ECS was targeted to the cytosol or the chloroplast (Noctor et al. 1996, 1998; Arisi et al. 1997). Expression of the same γ-ECS in the tobacco chloroplast also produced substantial increases in leaf glutathione (Creissen et al. 1999) whereas homologous overexpression of γ-ECS in Arabidopsis resulted in about twofold glutathione enrichment (Xiang et al. 2001). More recently, considerably greater increases in glutathione have been achieved by overexpression of a bifunctional γ-ECS/GSH-S from Streptococcus (Liedschulte et al. 2010).

There is still some debate about the impact of increasing glutathione contents in plants. Whereas increased glutathione triggered by chloroplastic overexpression of γ-ECS in tobacco was accompanied by oxidation and lesion formation (Creissen et al. 1999), studies with the same E. coli construct introduced into other species did not report marked phenotypic effects (Noctor et al. 1998; Zhu et al. 1999a). More recently, it has been shown that one of the chloroplast lines with multiple insertions shows symptoms of early leaf senescence (Herschbach et al. 2009). Another study of a single cytosolic overexpressor grown for 3 years in the field reported several effects, including lower biomass and photosynthesis (Ivanova et al. 2011). In terms of interpretation of phenotypes, a clear limitation of studies in poplar is the difficulty of genetic studies in this species. This necessitates analysis of several independent lines to be certain that the observed effects are linked to increases in glutathione. To date no marked deleterious effects have been reported in the tobacco overexpressors with very high glutathione (Liedschulte et al. 2010), though these lines are clearly interesting systems in which to analyse the impact of high glutathione concentrations on plant function. The reasons for the apparent discrepancy between the studies of Creissen et al. (1999) and Liedschulte et al. (2010) remain to be elucidated. An important factor could be differences in the introduced proteins. Unlike the E. coli γ-ECS, the Streptococcus protein has both GSH-S and γ-ECS activities.

Several studies have shown the benefits of elevating glutathione through overexpression of γ-ECS. These include enhanced resistance to heavy metals and certain herbicides (Zhu et al. 1999a; Gullner, Komives & Rennenberg 2001; Ivanova et al. 2011). Intriguingly, increased glutathione produced by overexpression of γ-ECS in the chloroplast was associated with higher leaf contents of several free amino acids, including tyrosine, leucine, isoleucine and valine (Noctor et al. 1998). This could bear some relation to redox regulation of the enzymes of amino acid metabolism in the chloroplast, several of which are potential TRX targets (Montrichard et al. 2009).

Although overexpression of GSH-S alone has less marked effects on tissue glutathione contents than boosting γ-ECS capacity (Foyer et al. 1995; Strohm et al. 1995; Noctor et al. 1998), the effects could be condition dependent if GSH-S becomes limiting when γ-EC supply is increased, for example, during exposure to cadmium (Zhu et al. 1999b). Similarly, expression of a soybean homoGSH-S in tobacco was successfully used to confer tolerance to the herbicide, fomesafen (Skipsey et al. 2005b). Interestingly, in this study, the homoGSH-S was expressed together with a homoGSH-preferring soybean GST (Skipsey et al. 2005b).

Turnover and degradation

Biochemical studies of glutathione degradation in tobacco conducted by the Rennenberg group (Rennenberg, Steinkamp & Kesselmeier 1981; Steinkamp & Rennenberg 1984, 1985; Steinkamp, Schweihofen & Rennenberg 1987) have been significantly extended over the last decade, notably by genetically based studies in Arabidopsis. Four different types of enzymes have been described that could initiate glutathione breakdown (Fig. 2). Some of these enzymes could use GSH, while others act preferentially on GSSG or other GS-conjugates. Firstly, glutathione or GS-conjugates could be degraded by carboxypeptidase activity (Steinkamp & Rennenberg 1985), which has been detected in barley vacuoles (Wolf, Dietz & Schröder 1996). A second type of enzyme that has been implicated in GS-conjugate breakdown is the cytosolic enzyme phytochelatin synthase (PCS; Blum et al. 2007, 2010). Because GS-conjugates are usually rapidly transported into the vacuole, the extent to which they accumulate in the cytosol remains unclear. However, some data obtained from work on Arabidopsis suggest that PCS may play some role in certain cell types or when the enzyme is activated by heavy metals (Grzam et al. 2006; Blum et al. 2007; Brazier-Hicks et al. 2008).

Figure 2.

Possible pathways of glutathione degradation. For simplicity, not all possible downstream reactions are shown, for example, metabolism of γ-EC by GGC. Disulphide forms of GSSG breakdown products are omitted for the same reason. As well as transpeptidation, GGT could also catalyse hydrolysis of GSH to Glu and Cys-Gly. aa, amino acid; Cpep, carboxypeptidase; GGC, γ-glutamyl cyclotransferase; GGT, γ-glutamyl transpeptidase; 5-OP, 5-oxoproline; 5-OPase, 5-oxoprolinase; PCS, phytochelatin synthase; X, S-conjugated compound.

The third type of enzyme, γ-glutamyl transpeptidase (GGT), has been the focus of several research groups in recent years. These enzymes act in the mammalian γ-glutamyl cycle (Meister 1988), and catalyse the hydrolysis or transpeptidation of GSH at the plasma membrane. The resulting γ-glutamyl amino acid derivatives are further processed by γ-glutamyl cyclotransferases (GGC) and 5-oxoprolinase (5-OPase) to produce free glutamate. In Arabidopsis, GGTs are encoded by at least three functional genes. Two of these (GGT1 and GGT2) encode apoplastic enzymes with activity against GSH, GSSG and GS-conjugates (Martin & Slovin 2000; Storozhenko et al. 2002). Extracellular GGT has also been described in barley and maize (Masi et al. 2007; Ferretti et al. 2009). Unlike their counterparts in animal cells, GGT1 and GGT2 are probably bound to the cell wall rather than to the plasmalemmma (Martin et al. 2007; Ohkama-Ohtsu et al. 2007a). Based on phenotypes of mutants and other studies, these enzymes may be important in countering oxidative stress or in salvaging excreted GSSG (Ohkama-Ohtsu et al. 2007a; Ferretti et al. 2009; Destro et al. 2011). Besides the extracellular GGTs, Arabidopsis has at least one vacuolar GGT, which is probably involved in the breakdown of GS-conjugates (Grzam et al. 2007; Martin et al. 2007; Ohkama-Ohtsu et al. 2007b). Along with PCS, GGTs may be important in metabolizing GS-conjugates that are formed during the synthesis of certain secondary metabolites (Ohkama-Ohtsu et al. 2011; Su et al. 2011).

In animals, γ-glutamyl peptides produced by GGT are further metabolized by GGC. An Arabidopsis gene (OXP1) has been identified that likely encodes 5-OPase. This enzyme catalyses the hydrolysis of 5-oxoproline, which is the product of GGC activity (Ohkama-Ohtsu et al. 2008). Based on 5-oxoproline accumulation in single oxp1 mutants, and in triple oxp1 ggt1 ggt4 mutants that are deficient in the major GGT activities as well as 5-OPase, it was proposed that the predominant pathway for GSH degradation is cytosolic and initiated by GGC (Fig. 2), and not vacuolar or extracellular GGT (Ohkama-Ohtsu et al. 2008). GGC is therefore a fourth type of enzyme potentially involved in initiating glutathione degradation, although both the rat and tobacco GGC have been reported to be unable to use GSH (Orlowski & Meister 1973; Steinkamp et al. 1987). The first gene encoding GGC was identified in humans recently, but no obviously homologous sequences exist in plants (Oakley et al. 2008). Yet other proteins may contribute to some extent to glutathione turnover, for example, GGP1, which has been implicated in the removal of the Glu residue from a GS-conjugate during glucosinolate synthesis (Geu-Flores et al. 2009). Because of this complexity, several questions remain on glutathione degradation in plants. These include cellular/tissue specificities, activities against the different forms of glutathione and, in some cases, the gene identities.

Despite the plethora of possible routes of glutathione catabolism, the extent of glutathione turnover and resynthesis remains unclear. Rates of glutathione catabolism in Arabidopsis leaves have been estimated to be as high as 30 nmol g−1 FW h−1 (Ohkama-Ohtsu et al. 2008). Extractable Arabidopsis leaf GSH-S activities can reach 10 nmol g−1 FW min−1, while the extractable activity of the rate-limiting enzyme, γ-ECS, is much lower, only about 0.5 nmol g−1 FW min−1 (Queval et al. 2009). This rate is therefore very close to the rates of turnover estimated by Ohkama-Ohtsu et al. (2008). However, this close correspondence could be coincidental. Extractable γ-ECS activities should theoretically be higher than true in vivo fluxes, because substrates, notably cysteine, may be less limiting and feedback inhibition by glutathione negligible in assays of desalted extracts. It should be noted, however, that redox regulation and the absence of other factors may make it difficult to recover true in vivo capacities of γ-ECS in in vitro assays. Indeed, assuming that accumulation of glutathione reflects neosynthesis and that there are no alternative routes awaiting description, our recent data strongly suggest that at least under some conditions, γ-ECS can work considerably faster in vivo than the measured extractable value. Glutathione oxidation in cat2 gr1 double mutants triggers accumulation of total glutathione to more than 10-fold basal levels within 4 d (Mhamdi et al. 2010a). Within the first 4 h after the beginning of excess H2O2 production (a consequence of the cat2 background), about 800 nmol glutathione g−1FW was newly accumulated in cat2 gr1, that is, a mean minimum synthesis rate of around 200 nmol g−1 FW h−1.

While the interplay between degradation and neosynthesis remains to be elucidated, typical leaf glutathione contents are 300 nmol g−1 FW. Thus, degradation rates of 30 nmol g−1 FW h−1 would mean that some glutathione pools must turn over and be resynthesized within a few hours. Another question concerns the possible role of glutathione degradation during oxidative stress. In this regard, it is interesting that GGT1 was among glutathione-associated genes that were significantly down-regulated in GSSG-accumulating cat2 gr1 mutants (Mhamdi et al. 2010a). By virtue of their activities in cleaving GS-conjugates, some enzymes implicated in glutathione degradation may also be important in biosynthesis pathways (e.g. Su et al. 2011; see below).

Compartmentation and transport

Glutathione is one of the major forms of organic sulphur translocated in the phloem (Herschbach & Rennenberg 1994, 1995; Bourgis et al. 1999; Mendoza-Cózatl et al. 2008), and must therefore move between cells, either apoplastically, symplastically or both. Glutathione can be detected in apoplastic extracts but at much lower levels than in whole tissue extracts (Vanacker, Carver & Foyer 1998). This is in line with immunolocalization studies that have detected weak or no labelling in the cell wall and apoplast (Zechmann et al. 2008). The apoplastic pool is also likely more oxidized than many intracellular pools. Current concepts suggest that both the apoplast and vacuole have low glutathione concentrations and GSH:GSSG ratios, and that most of the glutathione is concentrated in other compartments. While immunolocalization studies point to particularly high concentrations in the mitochondria (Zechmann et al. 2008), because of their greater volume the cytosol and chloroplasts have been estimated to account for about 50 and 30%, respectively, of total glutathione in Arabidopsis leaf mesophyll cells (Queval et al. 2011).

Uptake of both GSH and GSSG has been documented in both cells and protoplasts (Schneider, Martini & Rennenberg 1992; Jamaïet al. 1996). Transport across the plasmalemma could be accomplished by the oligopeptide transporter (OPT) family. Within this family, the Brassica uncea and rice genes, BjGT1 and OsGT1, are homologous to the yeast HGT1 transporter, which can transport GSH, GSSG and GS-conjugates, but also other small peptides (Bourbouloux et al. 2000; Bogs et al. 2003; Zhang et al. 2004). Of the nine annotated Arabidopsis OPT genes (Koh et al. 2002), OPT6 may play a role in long-distance transport (Cagnac et al. 2004). The OPT6 gene is highly expressed in the vasculature, where it may transport GS-conjugates and GS-cadmium complexes as well as GSH and GSSG (Cagnac et al. 2004). Another study reported that GSH, but not GSSG, was transported by OPT6 (Pike et al. 2009). Knockout opt6 mutants are aphenotypic, suggesting gene redundancy, and the transporter may be important in translocating other peptides as well as glutathione (Pike et al. 2009).

Several different types of transporter may be important in translocation of glutathione between subcellular compartments. Inner chloroplast envelope transporters have recently been described that likely act to link plastidic γ-ECS and cytosolic GSH-S via γ-EC export across the chloroplast envelope (Maughan et al. 2010). These transporters are encoded by three genes in Arabidopsis, called CLT1, CLT2 and CLT3. Further, the wild-type phenotype of gsh2 knockout mutants can be restored by transformation with a construct driving GSH-S expression exclusively in the cytosol (Pasternak et al. 2008). This observation provides further evidence that GSH can be imported from the cytosol into the plastid, consistent with radiolabelling studies of isolated wheat chloroplasts (Noctor et al. 2002a). Thus, current concepts suggest that γ-EC is synthesised exclusively in the chloroplast, then either converted to GSH in this compartment or transported to the cytosol, where part of any GSH formed can be transported into the chloroplast (Fig. 3).

Figure 3.

Compartmentation of glutathione synthesis, reduction and transport. γ-EC(S), γ-glutamylcysteine (synthetase); BAG, Bcl2-associated anathogene; CYT, cytosol. PER, peroxisome; GSH-S, glutathione synthetase; GR, glutathione reductase; GSSG, glutathione disulphide; CLT, chloroquinone resistance transporter-like transporter; MIT, mitochondrion; MRP, multidrug resistance-associated protein; OPT, oligopeptide transporter.

Tonoplast multidrug resistance-associated protein (MRP) transporters of the ATP-binding cassette (ABC) type may act to clear GS-conjugates or GSSG from the cytosol (Martinoia et al. 1993; Rea 1999; Foyer, Theodoulou & Delrot 2001). Indeed, certain Arabidopsis MRPs are competent in GSSG as well as GS-conjugate transport (Lu et al. 1998). Barley vacuoles can take up GSSG much more rapidly than GSH, and the removal of cytosolic GSSG may play a role in maintaining glutathione redox status in this compartment (Tommasini et al. 1993). Although glutathione concentrations in the vacuoles of unstressed plants have long been considered to be low or negligible (Rennenberg 1980; Zechmann et al. 2008), accumulation of GSSG in this compartment could be a physiologically important part of oxidative stress responses (Queval et al. 2011). Further, induction of specific MRPs occurs in response to oxidising agents (Sánchez-Fernández et al. 1998) and also accompanies GSSG accumulation in Arabidopsis (Mhamdi et al. 2010a).

Immunolocalization studies detect nuclear glutathione concentrations that are similar to those in the cytosol in cells in the undividing G0 state (Zechmann et al. 2008). Other recent studies suggest that nuclear/cytosol distribution of glutathione may be dynamic, with glutathione being recruited into the nucleus early in the cell cycle in both mammalian and plant cells (Markovic et al. 2007; Diaz-Vivancos et al. 2010a). A nuclear redox cycle within the cell cycle has been proposed, according to which GSH moves into the nucleus during the G1 phase: this in turn promotes oxidation of the cytosol and, consequently, enhanced glutathione accumulation (Diaz-Vivancos et al. 2010b). Oxidation of the cytosol at G1 is accompanied by enhanced levels of ROS and lowering of the oxidative defense shield (Diaz-Vivancos et al. 2010a). The accumulated GSH is divided between the daughter cells, in which the process begins again. These observations suggest the presence of proteins in plants that are able to alter the permeability of nuclear pores, facilitating GSH sequestration in the nucleus. Such proteins remain to be identified in plants. However, the anti-apoptotic factor Bcl-2 is thought to be a crucial component regulating GSH transport into the nucleus in mammalian tissues, as it is in mitochondria (Voehringer et al. 1998). Despite the lack of evidence for GSH synthesis in the mitochondria, high glutathione concentrations have been detected in this organelle (Zechmann et al. 2008). It is possible that pore-regulating proteins are involved in regulating mitochondrial and nuclear GSH concentrations (Fig. 3).

Like mitochondria, peroxisomes contain glutathione and GR (Jiménez et al. 1997) while apparently lacking the enzymes of glutathione synthesis. Immunolocalization studies suggest that the peroxisomal glutathione concentration is similar to the cytosolic concentration (Zechmann et al. 2008), and it has been estimated in leaf mesophyll cells to be around 3–4 mm (Queval et al. 2011). Presumably, this pool results from import across the single peroxisomal membrane, but transporters responsible for this activity remain to be characterized.

METABOLIC FUNCTIONS OF GLUTATHIONE

Regulation of sulphur assimilation

As a significant non-protein sink for reduced sulphur, glutathione contents are influenced by sulphur supply. Indeed, glutathione may be one of the S-containing compounds that link changes in sulphur nutrition to resistance to some pathogens, a phenomenon termed sulphur-induced resistance (SIR; Gullner et al. 1999; Bloem et al. 2007; Zechmann et al. 2007; Höller et al. 2010). Stresses that involve an oxidative component also cause up-regulation of the S assimilation pathway. For example, exposure to ozone increases cysteine and glutathione levels, effects linked to post-translational activation of APR1 (Bick et al. 2001). In catalase mutants, increased intracellular H2O2 triggers accumulation of glutathione and precursors, and this is accompanied by accumulation of transcripts for all three APRs (Queval et al. 2009).

Glutathione is an important form of translocated organic S and may thus act as an internal ‘barometer’ of plant S status (Kopriva & Rennenberg 2004). Glutathione regulates several steps involved in S assimilation. These include GSH inhibition of sulphate uptake and assimilation through effects on specific transporters and enzymes such as ATP sulphurylase 1 (APS1) and APR (Herschbach & Rennenberg 1994; Lappartient et al. 1999; Vauclare et al. 2002; Buchner et al. 2004). However, the role of glutathione in S assimilation is complex, because as well as these repressive effects, GSH is required by APR as reductant to produce sulphite (Leustek 2002). This occurs through a domain of the APR enzyme that allows it to act as a GRX, using GSH with an apparent KM value of about 1 mm (Bick et al. 1998). Further, thiol-disulphide status also acts to control certain APR isoforms post-translationally through activation in response to increased glutathione oxidation (Leustek 2002).

The influence of sulphur assimilation activity on glutathione contents also underlines the close relationship between glutathione and S status (Nikiforova et al. 2003). Arabidopsis mutants defective in certain sulphate transporters show decreased leaf glutathione accumulation (Maruyama-Nakashita et al. 2003). By contrast, expression of a bacterial APR in Arabidopsis enriches tissue cysteine and glutathione contents (Tsakraklides et al. 2002). Qualitatively similar effects are produced by overexpression of enzymes of the cysteine synthesis pathway (Harms et al. 2000; Noji & Saito 2002; Wirtz & Hell 2007).

Glutathione S-transferases

The classical reaction catalysed by GSTs is the formation of a covalent bond between the sulphur atom of glutathione and an electrophilic compound (Fig. 4). In plants, the activity of most GSTs depends on an active site serine, which stabilizes the GS-thiolate anion (Dixon & Edwards 2010), though in some GSTs such as the dehydroascorbate reductases (DHARs) this residue is replaced by a cysteine. This change confers the capacity for reversible disulphide bond formation with glutathione that is part of the DHAR catalytic mechanism (Dixon, Davis & Edwards 2002). Long considered as cytosolic enzymes, several GSTs may also localize at least partly to other compartments, including the chloroplast, peroxisome and nucleus (Thatcher et al. 2007; Dixon et al. 2009).

Figure 4.

Detoxification pathways involving glutathione S-conjugate formation and metabolism. The glyoxalase pathway shows methylglyoxal metabolism, but other oxo-aldehydes might also be metabolized through this route. FDH, formaldehyde dehydrogenase; FGH, S-formylglutathione hydrolase; GLYI, glyoxalase I; GLYII, glyoxalase II; GST, glutathione S-transferase; Me, heavy metal; PCS, phytochelatin synthase; X, electrophilic compound.

The GST family in plants is notable for its structural and functional diversity, and the biochemical and physiological functions of specific members remain to be elucidated. As well as or instead of catalysing conjugase reactions, some GSTs may have antioxidative functions. The DHAR type of GST is one example, while several subclasses of GST have peroxidase activity (Wagner et al. 2002; Dixon et al. 2009). GSTs are able to reduce organic peroxides though some specificity is observed between different peroxides (Wagner et al. 2002; Dixon et al. 2009; Dixon & Edwards 2010). Some GSTs are strongly inducible by H2O2 (Levine et al. 1994; Willekens et al. 1997; Wagner et al. 2002). For example, certain GST transcripts can be considered useful markers for increased intracellular availability of H2O2 (Vanderauwera et al. 2005; Queval et al. 2007, 2009; Chaouch et al. 2010), though several are also inducible by salicylic acid (SA; Sappl et al. 2009). The same or other GSTs may have functions in the detoxification of electrophilic xenobiotics, but also be important in biosynthetic or catabolic pathways. Recently described examples of biosynthetic pathways involving GSTs are the production of glucosinolates and camalexin, discussed further later. For a more detailed recent discussion of GSTs, we refer the reader to Dixon & Edwards (2010).

Glyoxalase and formaldehyde metabolism

Oxo-aldehydes such as glyoxal are reactive compounds that may interfere with sensitive cellular compounds. A common toxic oxo-aldehyde is methylglyoxal, which can be produced from triose phosphate as an intermediate in the triose-phosphate isomerase reaction (Maiti et al. 1997; Marasinghe et al. 2005). The glyoxalase system acts to convert these compounds to non-toxic hydroxyacids such as lactate, and consists of two enzymes acting in concert (Fig. 4). Glyoxalase I isomerizes the spontaneously generated GS-adduct while the hydrolytic reaction catalysed by glyoxalase II liberates the hydroxyacid and free GSH. In several plant species, overexpression of these enzymes increases tolerance to exogenous methylglyoxal and/or salt (Singla-Pareek, Reddy & Sopory 2003; Deb Roy et al. 2008; Singla-Pareek et al. 2008).

In addition to acting as a substrate for GSTs and glyoxalases, GSH may also act in detoxification reactions via formaldehyde dehydrogenase. Formaldehyde can enter plants through stomata or be produced by endogenous metabolism (Haslam et al. 2002). Two enzymes, formaldehyde dehydrogenase (FDH) and S-formylglutathione hydrolase (FGH), act to oxidize formaldehyde to formic acid, which may then be converted to CO2 or enter C1 metabolism (Fig. 4). Genes for both enzymes have been identified in Arabidopsis (Martínez et al. 1996; Haslam et al. 2002; Achkor et al. 2003). An important feature of the encoded FDH is that it is able to act as a GSNO reductase (GSNOR; Sakamoto, Ueda & Morikawa 2002; Díaz et al. 2003).

Phytochelatin synthesis

A conditionally important role of GSH is in the response to excessive levels of heavy metals (Fig. 4). Glutathione is the precursor of phytochelatins ([γ-Glu-Cys]nGly), compounds that are synthesized in response to cadmium and other heavy metals. Phytochelatins sequester the metal to form a complex that is then transported into the vacuole (Grill, Winnacker & Zenk 1987; Grill et al. 1989; Cobbett & Goldsbrough 2002; Rea, Vatamaniuk & Rigden 2004). These compounds are produced from glutathione or homologues by PCS, a cytosolic enzyme. In Arabidopsis, there are two genes encoding PCS, one of which (PCS1) is the gene affected by the cad1 mutation that exacerbates cadmium sensitivity (Howden & Cobbett 1992; Ha et al. 1999; Cazalé & Clemens 2001). As noted previously, it is possible that PCS has other biochemical roles, in addition to phytochelatin synthesis, for example, turnover of GS-conjugates (Rea et al. 2004; Blum et al. 2007, 2010; Clemens & Peršoh 2009).

The importance of sufficient amounts of GSH to support phytochelatin synthesis is evidenced by the identification of the cad2 mutant as affected in γ-ECS, causing a decrease in leaf GSH to about 20% wild-type levels (Howden et al. 1995; Cobbett et al. 1998). As well as a precursor role in phytochelatin synthesis, glutathione could be involved in heavy metal resistance via an antioxidant function: many heavy metals are considered to provoke perturbation of cellular redox homeostasis through several mechanisms, for example, displacement of redox-active metals from bound sites. Increased heavy metal tolerance has been observed in transgenic lines with enhanced glutathione synthesis (Zhu et al. 1999a,b; Lee et al. 2003), contrasting with effects of overexpressing PCS itself, which has generally yielded less clear-cut results (Peterson & Oliver 2006; Picault et al. 2006). While both shoots and roots may make a contribution to heavy metal detoxification, it is interesting that phytochelatin and GS-cadmium concentrations have been reported to be much higher in the phloem than in the xylem during exposure of B. napus to cadmium (Mendoza-Cózatl et al. 2008).

Glutathione metabolic reactions in defence against biotic stress

Glutathione has long been implicated in reactions linked to secondary metabolism and pathogen responses (Dron et al. 1988; Edwards, Blount & Dixon 1991). While at least some of the effects of glutathione during interactions with pathogens probably involve a signalling role, it now appears that others are linked to hormone and secondary metabolite synthesis. Crucial information had been generated by the analysis of glutathione-deficient mutants. The first identified glutathione-deficient Arabidopsis mutant, cad2, was reported to show unchanged resistance to virulent and avirulent strains of the oomycete Peronospora parasitica as well as to the bacterium Pseudomonas syringae (May et al. 1996b). However, a later study of cad2 and rax1-1 reported increased susceptibility to avirulent P. syringae (Ball et al. 2004). Mutants defective in the CLT glutathione chloroplast envelope transporters showed decreased expression of PR1 and also lower resistance to the oomycete Phytophthora brassicae (Maughan et al. 2010). Decreased PR1 expression linked to lower SA accumulation was also observed in gr1 mutants lacking cytosolic/peroxisomal GR (Mhamdi et al. 2010a).

The Arabidopsis pad2 mutant was first identified as deficient in camalexin, an indole phytoalexin that contains one S atom per molecule and whose thiazole ring is derived partly from cysteine (Glazebrook & Ausubel 1994). This line shows enhanced susceptibility to various bacterial, fungal and oomycete pathogens (Ferrari et al. 2003; Parisy et al. 2006). The affected gene in pad2 is GSH1 and the mutation results in GSH contents that are slightly lower than those in cad2 (Parisy et al. 2006). Of cad2, rax1 and pad2 mutants in GSH1, rax1 has the highest leaf glutathione contents, pad2 the lowest while cad2 is intermediate. The glutathione contents of these lines correlate inversely with their resistance to P. brassicae, though only pad2 was reported to be markedly affected in camalexin contents and its response to P. syringae (Parisy et al. 2006). Thus, there appears to be a certain level of glutathione required for the synthesis of pathogen defense-related molecules and disease resistance. This level could vary according to conditions and pathogen specificity. In the case of camalexin synthesis, recent data show that GSH is required as a precursor of the thiazole ring (Fig. 5a; Böttcher et al. 2009; Su et al. 2011).

Figure 5.

Possible roles of glutathione as an S atom donor in the production of camalexin (a) and glucosinolates (b). The pathways shown in (a) and (b) are adapted from those shown in Su et al. (2011) and Geu-Flores et al. (2009), respectively. Both schemes shown here specifically focus on steps involving glutathione. In both paths, other steps notably involve several different cytochromes P450 (CYPs), UDP-glucosyl transferases (UGT) or C-S lyase (CSL). The pathway in (b) may also involve partial metabolism of the GS-conjugate [as in (a)] prior to C-S lyase action. GGP, γ-glutamyl peptidase; Glc, glucose residue; PCS, phytochelatin synthase.

As well as effects on camalexin and resistance to microorganisms, pad2 shows decreased resistance to feeding of insect larvae, an effect that is linked to decreased accumulation of glucosinolates (Schlaeppi et al. 2008). This effect can be rescued by supplementation with GSH but not the general disulphide reductant, dithiothreitol (Schlaeppi et al. 2008). Using a dedicated microarray of more than 200 genes whose expression is associated with insect feeding, it was shown that unlike the JA signalling mutant, coi1, the pad2 mutant did not show significant differences in the expression of these genes, including glucosinolate synthesis genes, in response to feeding (Schlaeppi et al. 2008). This observation is consistent with a requirement for a certain level of GSH to support glucosinolate synthesis as a sulphur source, rather than as a signalling or regulatory molecule. Indeed, it has recently been reported that formation of the glucosinolate thioglucose moiety involves a GS-conjugate intermediate and that this compound is metabolized by a γ-glutamyl peptidase, GGP (Geu-Flores et al. 2009). Formation and metabolism of such GS-conjugates parallels stages in the camalexin synthesis pathway (Fig. 5), and the lack of GSH for the respective GST activities may explain both constitutive deficiency of camalexin and lower induction of glucosinolates in pad2 (Parisy et al. 2006; Schlaeppi et al. 2008). Both pathways shown in Fig. 5 involve formation of a GS-conjugate followed by further metabolism by GS-conjugate degrading enzymes. Studies of mutants implicated GGT1, GGT2 and, to a lesser extent, PCS1, in the control of Botrytis-induced camalexin accumulation (Su et al. 2011), whereas ggt4 mutants were found to accumulate a GS-conjugate of the JA synthesis precursor, 12-oxo-phytodienoic acid (OPDA), during incompatible interactions with P. syringae (Ohkama-Ohtsu et al. 2011).

Metabolism of ROS and ascorbate

Glutathione can react chemically with ROS and also with dehydroascorbate (DHA), the relatively stable oxidised form of ascorbate generated by dismutation of monodehydroascorbate (MDHA). In particular, there is a close relationship between increased availability of H2O2 and glutathione status. This is most evident from studies in which H2O2-metabolizing enzymes have been genetically or pharmacologically inhibited (Smith et al. 1984; May & Leaver 1993; Willekens et al. 1997; Noctor et al. 2002b; Rizhsky et al. 2002; Queval et al. 2007, 2009; Chaouch et al. 2010). Time-course analyses in conditional catalase mutants have shown that an initial conversion of GSH to GSSG, measurable within hours of exposure to the onset of H2O2 production, is followed by a several-fold induction of the total glutathione pool over the subsequent period of 3–4 d (Smith et al. 1984; Queval et al. 2009). At moderate rates of endogenous H2O2 production, this response involves a decrease in the whole-leaf GSH : GSSG ratio from above 20 to close to one (Mhamdi et al. 2010a). Such effects make glutathione status a useful marker for oxidative stress triggered by increased intracellular H2O2 production.

Several pathways may be involved in GSH-dependent H2O2 metabolism, while H2O2 may be metabolized by GSH-independent pathways (Fig. 6). The chemical reaction of GSH with H2O2 is slow, but three distinct types of peroxidases appear as the principal candidates to link peroxide reduction to GSH oxidation (Table 1). These are ascorbate peroxidase (APX), certain types of peroxiredoxin (PRX) and GSTs. Haem-based peroxidases are divided into two super-families, one of which (non-animal peroxidases) contains plant enzymes and is divided into three classes (Welinder 1992). While class II haem peroxidases are found in fungi and notably include secreted enzymes involved in lignin degradation, haem peroxidases in plants are found in classes I or III (Zámocky, Furtmüller & Obinger 2010). Class III enzymes are found only in plants. Also known as ‘guaiacol-type’ peroxidases, they are encoded by numerous genes and are located in the apoplast or vacuole. Their functions are not clearly established, but some are known to be involved in biosynthetic processes and/or in ROS production (Bindschedler et al. 2006; Cosio & Dunand 2009). In contrast, class I haem peroxidases include the intracellular antioxidative enzyme, APX. Like catalase, APX is relatively specific to H2O2 and does not metabolize other peroxides at high rates. It can participate in the ‘ascorbate-glutathione’ pathway in which H2O2 reduction is ultimately linked to NAD(P)H oxidation via ascorbate and glutathione pools. Alternatively APX activity could be coupled to NAD(P)H oxidation independently of glutathione via MDHAR activity (Fig. 6). While GSH can chemically reduce DHA at significant rates (though slower at pH 7 than at pH 8), an enzymatic link beween ascorbate and glutathione pools is provided by DHAR. Overexpression of this enzyme underlines the importance of GSH-dependent ascorbate pools in physiological processes like the regulation of stomatal opening (Chen et al. 2003), but the role of DHAR in the in vivo metabolism of peroxides still remains assumed rather than demonstrated.

Figure 6.

Glutathione function within major pathways for peroxide metabolism. APX, ascorbate peroxidase; CAT, catalase; DHA(R), dehydroascorbate (reductase); GPX, glutathione/thioredoxin peroxidase; GR, glutathione reductase; GRX, glutaredoxin; GST, glutathione S-transferase; MDHA(R), monodehydroascorbate (reductase); PRX, peroxiredoxin; ROOH, H2O2 or organic peroxide; TRX, thioredoxin. Ferredoxin-dependent pathways may be important in TRX and MDHA reduction in the chloroplast.

Table 1.  Simplified overview of peroxidases found in plants and their possible functional link to glutathione
EnzymeClassProsthetic or catalytic groupOxidantReductantFunctional link to glutathione
  1. Peroxiredoxin substrate specificities are based on information in Tripathi et al. (2009). Some PRX may also reduce peroxynitrite. See text for further explanation.

  2. APX, ascorbate peroxidase; GPX, glutathione peroxidase; GRX, glutaredoxin; GST, glutathione S-transferase; NTRC, NADPH-thioredoxin reductase C; POX, peroxidase; PRX, peroxiredoxin; ROOH, peroxide (H2O2 or organic); TRX, thioredoxin.

APXClass I haem peroxidaseHaemH2O2AscorbateAscorbate-glutathione cycle
Guaiacol-type POXClass III haem peroxidaseHaemH2O2Various?
2-Cys PRXPeroxiredoxinCysteinesROOHTRX, NTRC, cyclophilin?
1-Cys PRXPeroxiredoxinCysteinesROOHTRX?
PRX QPeroxiredoxinCysteinesROOHTRX?
Type II PRXPeroxiredoxinCysteinesROOHGSH/GRXGRX-dependent oxidation
GPXPeroxiredoxinCysteinesROOHTRX?
GSTGSTSerineROOHGSHPeroxidation

In contrast to haem-based enzymes, thiol peroxidases such as PRX are less specific to H2O2, and can also reduce other organic peroxides, though some may have preference for H2O2 or relatively small peroxides (Dietz 2003; Tripathi, Bhatt & Dietz 2009). There are four types of PRX described in plants, and three of these use either TRX or similar components such as NADPH-thioredoxin reductase (NTR) C (Dietz 2003, Pulido et al. 2010). One type, PRX II, can use GSH as a reductant via GRX action (Rouhier, Gelhaye & Jacquot 2002). Peroxiredoxins also include GPXs. Originally identified on the basis of their homology to animal GPX (Eshdat et al. 1997), and subsequently implicated in plant stress responses and signalling (Rodriguez Milla et al. 2003; Miao et al. 2006; Chang et al. 2009), these enzymes are now considered to act as TRX-dependent peroxiredoxins (Herbette et al. 2002; Iqbal et al. 2006; Navrot et al. 2006) and are, therefore, misleadingly named. Unlike the animal enzyme, where the reduction of peroxides involves formation of a mixed disulphide between a first GSH and the selenocysteine SeOH group, the GPXs found in plants generate a disulphide bridge from the initial sulphenic acid, with the disulphide intermediate being reduced back to the dithiol form by TRX. They are therefore unlikely to be involved in peroxide-mediated glutathione oxidation (Fig. 6). By contrast, peroxidation of GSH could be catalysed by GSTs (Wagner et al. 2002; Dixon et al. 2009). Other enzymes such as methionine sulphoxide reductase (MSR) may also contribute to ROS-triggered glutathione oxidation (Tarrago et al. 2009). Finally, although these studies, relying predominantly on in vitro analyses, have established the relative specificities in reducing substrates, it should be noted that the glutathione and TRX systems may be linked to some extent during in vivo ROS metabolism (Michelet et al. 2005; Reichheld et al. 2007; Marty et al. 2009).

In conclusion, GSH could be linked to H2O2 and/or peroxide reduction by at least two ascorbate-independent routes as well as the ascorbate-glutathione pathway (Fig. 6). Among these, only GSTs appear to act as direct GPXs (Table 1), all other enzymes requiring at least one additional protein to link peroxide reduction to GSH oxidation. The evidence from gene expression makes it clear that certain APX, GPX and GST genes are induced in response to oxidative stress (Willekens et al. 1997; Wagner et al. 2002; Levine et al. 1994; Sappl et al. 2009). Data from transcriptomics and enzyme and metabolite assays of plants deficient in catalase and/or GR suggest that the ascorbate-glutathione pathway and enzymes such as GSTs may act in concert to remove excess H2O2 and/or other peroxides (Mhamdi et al. 2010a).

Glutathione reductase

The GSSG produced by GSH oxidation is reduced by GR. While NTR can also reduce GSSG in a TRX-dependent manner (Marty et al. 2009), this enzyme is less efficient than GR, which is encoded by two genes in plants studied so far. Chloroplast and mitochondrial GR is encoded by GR2, while GR1 encodes a protein that is found in the cytosol and peroxisome (Creissen et al. 1995; Chew et al. 2003; Kataya & Reumann 2010). The dual targeting of these two genes is therefore sufficient to explain biochemical data on GR localization in all of these compartments (Edwards et al. 1990; Rasmusson & Møller 1990; Jiménez et al. 1997; Stevens, Creissen & Mullineaux 2000; Romero-Puertas et al. 2006). Despite the long-standing association of GR and glutathione with resistance to various stresses (Esterbauer & Grill 1978; Tausz, Sircelj & Grill 2004), overexpression of GR in several plant species has not in itself been reported to lead to marked increases in stress resistance (Foyer et al. 1991, 1995; Aono et al. 1993; Broadbent et al. 1995; Kornyeyev et al. 2005; Ding et al. 2009). However, increases in the reduction state of the ascorbate pool in plants overexpressing GR are consistent with efficient coupling of the reactions of the ascorbate-glutathione pathway (Foyer et al. 1995).

While Arabidopsis T-DNA mutants for the chloroplast/mitochondrial GR2 are embryo-lethal (Tzafrir et al. 2004), gr1 knockout mutants do not show phenotypic effects, despite a 30–60% reduction in extractable enzyme activity (Marty et al. 2009; Mhamdi et al. 2010a). Genetic analyses have shown that the aphenotypic nature of gr1 mutants results from partial replacement of GSSG regeneration by the cytosol-located NTR-TRX system (Marty et al. 2009), though this is not sufficient to prevent significant accumulation of GSSG in the mutant. Accumulation of GSSG is massively increased compared with the parent lines in cat2 gr1 double mutants deficient in both the major leaf catalase and GR1 (Mhamdi et al. 2010a). This effect is associated with a much exacerbated phenotype compared with cat2, showing that the NTR-TRX system cannot replace GR1 under conditions where H2O2 production is increased. Further, the gr1 mutation alters Arabidopsis responses to pathogens and expression of genes involved in defence hormone signalling, notably JA-associated genes (Mhamdi et al. 2010a). Although transcriptomic patterns documented in this study point to at most limited overlap between glutathione and TRX systems in oxidative stress functions, the observation of Marty et al. (2009) shows that interplay is possible at the biochemical level. For example, changes in glutathione and TRX redox states could act mutualistically to reinforce each other during plant interactions with pathogens. This issue, which is important to defining the frequently used but vague term ‘cellular redox state’ more clearly, remains an outstanding issue in understanding redox signalling in plants.

GLUTATHIONE IN PLANT DEVELOPMENT, GROWTH AND ENVIRONMENTAL RESPONSES

It has long been known that certain cell types, for example, the root quiescent centre and cells in organs such as seeds, maintain a highly oxidized intracellular state (Kranner & Grill 1996; Kranner et al. 2002, 2006). Auxin accumulation in the root stem cell niche is dependent on the oxidized status of the cells (Jiang & Feldman 2010). Treatment of Arabidopsis root tips with BSO led to disappearance of the auxin maximum in the root tips and altered expression of quiescent centre markers (Koprivova, Mugford & Kopriva 2010). The glutathione redox potential of such cell types is relatively high (i.e. positive or oxidizing). Even in cells where the global glutathione pool is highly reduced, compartments that lack GR or that are deficient in NADPH may contain low GSH:GSSG ratios (e.g. the vacuole or endoplasmic reticulum; Hwang, Sinskey & Lodish 1992; Enyedi, Várnai & Geiszt 2010). While an increase in the redox potential above the threshold-reducing value of more redox-sensitive compartments has been linked to growth arrest and/or death (Kranner et al. 2006), cell identity has a profound influence on the processes that govern cell fate (Jiang et al. 2006a,b) and associated responses to abiotic stress (Dinneny et al. 2008).

Plant growth and auxin

The analysis of the phenotypes of glutathione-deficient Arabidopsis mutants has demonstrated that GSH is required for plant development fulfilling critical functions in embryo and meristem development (Vernoux et al. 2000; Cairns et al. 2006; Reichheld et al. 2007; Frottin et al. 2009; Bashandy et al. 2010). The rml1 mutant, which has less than 5% of the glutathione present in the wild type, has a strong developmental phenotype that is characterized by a non-functional root meristem while the shoot meristem is largely unaffected (Vernoux et al. 2000). Crossing rml1 with ntra, ntrb double mutants produced an additive shoot meristemless phenotype (Reichheld et al. 2007). However, when the ntra ntrb double mutants were crossed with cad2 (which has about 30% of the glutathione present in the wild type), the resultant triple mutants developed normally at the rosette stage and underwent the floral transition but they produced almost naked flowering stems (Bashandy et al. 2010). The perturbation of the floral meristem in the ntra ntrb, cad2 triple mutants was linked to altered levels and transport of auxin, which plays an important role in the integration of meristem development (Bashandy et al. 2010).

Glutathione synthesis is also required for pollen germination and pollen tube growth (Zechmann, Koffler & Russell 2011). In this study, glutathione depletion was shown to result from disturbances in auxin metabolism and transport (Zechmann et al. 2011). The auxin-resistant axr1 and axr3 mutants were found to be less sensitive to BSO than the wild-type Arabidopsis plants and treatment of the tips of primary roots with BSO altered auxin transport (Koprivova et al. 2010). However, in these experiments, the effects of GSH on root growth could be partially reversed by dithiothreitol, suggesting that an as yet unidentified post-transcriptional redox mechanism is involved in the regulation of the expression of PIN proteins and hence auxin transport in roots (Koprivova et al. 2010). The stunted phenotype of mutants such as cat2 and derived lines, which accumulate GSSG when grown from seed in air, may be related to such thiol regulation of plant growth, as could the effects of GSH on the regeneration efficiency of somatic embryos (Belmonte et al. 2005).

Cell cycle regulation

Concepts of the regulation of mitosis in animals incorporate an intrinsic redox cycle in which transient oxidations serve to regulate progression through the cell cycle (Menon & Goswami 2007; Burhans & Heintz 2009). While very few studies have been conducted to establish whether similar processes operate in the control of the plant cell cycle, the recruitment of GSH into the nucleus in the G1 phase of the plant cell cycle has a profound effect on the redox state of the cytoplasm and the expression of redox-related genes (Pellny et al. 2009; Diaz-Vivancos et al. 2010a,b). A subsequent increase in the total cellular GSH pool above the level present at G1 is essential for the cells to progress from the G1 to the S phase of the cycle (Diaz-Vivancos et al. 2010a,b). As the movement of GSH into the nucleus can be visualized in the dividing cells of the developing lateral root meristem (Diaz-Vivancos et al. 2010a), it will be intriguing to see if this process underlies the post-transcriptional redox regulation of the PIN proteins and auxin transport in roots (Koprivova et al. 2010).

Biotic interactions, cell death and defence phytohormones

In addition to serving as a source of reduced S during synthesis of secondary metabolites, glutathione is involved in signalling processes. Exogenous GSH can mimic fungal elicitors in activating the expression of defence-related genes (Dron et al. 1988; Wingate, Lawton & Lamb 1988) including PATHOGENESIS-RELATED1 (PR1; Senda & Ogawa 2004; Gomez et al. 2004a). Moreover, accumulation of glutathione is triggered by pathogen infection (Edwards et al. 1991; May, Hammond-Kosack & Jones 1996a), and this can involve characteristic transient changes in glutathione redox state (Vanacker, Carver & Foyer 2000; Parisy et al. 2006). Similar changes have also been reported following exogenous application of the defence-related hormone salicylic acid (SA), or biologically active SA analogs (Mou, Fan & Dong 2003; Mateo et al. 2006; Koornneef et al. 2008). Glutathione perturbation in catalase-deficient cat2 is linked to hypersensitive response (HR)-like lesions as part of a wide spectrum of defence responses that are conditionally induced in this line (Chaouch et al. 2010). Glutathione is also required for the development of symbiotic N2-fixing nodules between legumes and rhizobia. Both GSH and the legume homologue, homoGSH, are found at high concentrations in N2-fixing nodules formed during symbiotic interactions with rhizobia: deficiency in these compounds inhibits nodule formation (Frendo et al. 2005; Pauly et al. 2006).

Thiol-disulphide status is clearly involved in the regulation of the SA-dependent NONEXPRESSOROFPATHOGENESISRELATEDGENES 1 (NPR1) pathway (Després et al. 2003; Mou et al. 2003; Rochon et al. 2006; Tada et al. 2008). Induction of PR gene expression by this pathway involves monomerization of the oligomeric cytosolic protein NPR1, which unmasks a nuclear localization signal motif that allows the protein to relocalize to the nucleus where it interacts with TGA transcription factors, themselves redox-sensitive (Després et al. 2003; Mou et al. 2003). The notion that this might be linked to glutathione redox state via GRX activity (Mou et al. 2003) has been superseded by a model based on activation by TRXh (Tada et al. 2008), some of which are known to be inducible by oxidative stress and pathogens (Laloi et al. 2004). If this model is correct, the well-documented ability of genetically or chemically increased GSH to induce PR gene expression could possibly be explained by redox interactions between glutathione and TRX.

Although activation of NPR1 involves a reductive change, the initial trigger for the events leading to PR gene expression involves oxidation. Excessive oxidation can also trigger HR, which is the most studied instance of environmentally induced cell death in plants. While redox changes involved in HR have been most closely associated with events triggered by apoplastic ROS production, oxidative stress of intracellular origin can also trigger such effects. Several observations are suggestive of a role for glutathione redox potential (or GSSG accumulation) in the regulation of genetically programmed cell suicide pathways. For example, increases in the total glutathione contents of leaves produced by ectopic γ-ECS overexpression in the tobacco chloroplast caused accumulation of GSSG and this was associated with lesion formation and enhanced expression of pathogenesis-related (PR) genes (Creissen et al. 1999). In addition, the HR-like lesions triggered by intracellular oxidative stress are associated with GSSG accumulation (Smith et al. 1984; Willekens et al. 1997; Chamnongpol et al. 1998; Chaouch et al. 2010). Alterations in pathogen responses have been described in mutants that are partly deficient in glutathione (Ball et al. 2004; Parisy et al. 2006). However, some of our own recent work in catalase-deficient cat2 and derived Arabidopsis lines suggest that there is unlikely to be a simple relationship between glutathione redox potential and engagement of cell death. Firstly, HR-like lesions in cat2 are under daylength control and this control does not correlate with the degree of glutathione oxidation (Queval et al. 2007). Secondly, daylength control is linked to SA synthesis: HR-like lesions can be prevented by blocking SA synthesis even though this effect is associated with a more oxidized glutathione status (Chaouch & Noctor 2010; Chaouch et al. 2010). Thirdly, in cat2 gr1 mutants, GSSG accumulates to a greater extent than in cat2 single mutants but this does not cause HR-like lesions (Mhamdi et al. 2010a). Despite these observations, however, we cannot exclude the possibility that values for whole-leaf glutathione status may not precisely reflect events in specific intracellular compartments (Queval et al. 2011). Even if this is the case, tissue glutathione status appears to be a poor marker for the engagement of cell suicide pathways. Nevertheless, analysis of H2O2 signalling in glutathione-deficient mutants demonstrates that some factor related to glutathione status is an important modulator of cell death triggered by oxidative stress (authors' unpublished results). Further, analyses of double cat2 atrboh Arabidopsis mutants, which lack both the major catalase and NADPH oxidase activities, point to a crucial role for AtrbohF in permitting accumulation of oxidized glutathione in response to intracellular H2O2. In double cat2 atrbohF (but not cat2 atrbohD) mutants, glutathione is much less perturbed than in the cat2 single mutant and this is associated with much lower accumulation of SA and related defence molecules (Chaouch, Queval & Noctor, unpublished results).

As well as possible roles in SA signalling, glutathione may modulate signalling through the JA pathway, which is involved in the regulation of development and in responses to necrotrophic pathogens and herbivores. JA induces GSH1, GSH2 and GR (Xiang & Oliver 1998), as well as other antioxidative genes (Sasaki-Sekimoto et al. 2005). In gr1 mutants lacking the cytosolic/peroxisomal GR, a suite of JA genes are repressed while introduction of the gr1 mutation into the cat2 background modulates H2O2-triggered expression of these and other JA-associated genes (Mhamdi et al. 2010a). Consistent with these interactions between JA and glutathione, prior wounding was shown to induce resistance to Botrytis cinerea, an effect that was partly or wholly suppressed in glutathione-deficient mutants (Chassot et al. 2008). This effect was associated, at least in part, with impaired induction of camalexin (Chassot et al. 2008). Among the components that could link JA signalling and glutathione are GRXs and GSTs. The SA-inducible protein, GRX480, represses up-regulation of certain JA-induced genes (Ndamukong et al. 2007). As well as acting in the activation of the SA pathway, NPR1 represses JA pathway (Spoel et al. 2003), and BSO treatment was shown to anatagonize this repressive effect (Koornneef et al. 2008). Together, these observations suggest that glutathione may act to repress JA signalling through SA-dependent induction of NPR1 and GRX480, both of which interact with TGA transcription factors. However, gene expression profiles and SA contents suggest that the more oxidized glutathione status in gr1 is associated with repression of both SA and JA pathways, and therefore that the roles of glutathione may be complex and not limited to regulating antagonism between the two hormones (Mhamdi et al. 2010a). Other possible effects of glutathione could occur through the regulation of JA synthesis or excess accumulation of intermediates like OPDA. Several GSTs are among early JA-induced genes or may catalyse formation of GS-oxylipin conjugates (Davoine et al. 2006; Yan et al. 2007; Mueller et al. 2008). Indeed, enhanced accumulation of a GS-OPDA conjugate in ggt4 mutants, which are expected to be unable to degrade vacuolar GS-conjugates, was recently reported. However, loss of GGT4 function did not affect contents of JA or derivatives such as the hormonally active JA-isoleucine conjugate compared with wild-type plants (Ohkama-Ohtsu et al. 2011). In addition to these interactions with glutathione, JA and wounding were also shown to decrease GSNOR transcripts (Díaz et al. 2003).

Light signalling

The abundance of glutathione or its homologues in leaves does not vary greatly over the day/night cycle. Similarly, in soybean, all leaves except those approaching senescence contain similar total amounts of glutathione/homoglutathione, as illustrated in Fig. 7. However, young poplar trees show more variation, with a gradient from young to old leaves (Arisi et al. 1997). The effect of stresses such as drought can also influence the relative abundance of glutathione/homoglutathione in different leaf ranks (Fig. 7). Shade conditions that involve a decrease in the red/far red ratio of the light environment favour a much lower leaf glutathione pool relative to conditions where the red/far red ratios are similar (Bartoli et al. 2009). While leaf glutathione contents are lower in plants grown under shade conditions, leaf GSH:GSSG ratios are relatively unaffected by light quality (Bartoli et al. 2009). Moreover, the adjustments in leaf glutathione pool in response to different red/far red ratios were very slow in comparison with the leaf ascorbate pool (Bartoli et al. 2009). Regardless of the relatively slow responses of the leaf glutathione pool to changes in light intensity and light quality, some potential links between daylength, light signalling and glutathione status have been reported (Becker et al. 2006; Queval et al. 2007). Relationships between glutathione and photoreceptor signalling have been suggested in studies on the arsenic-tolerant mutants, ars4 and ars5 (Sung et al. 2007). The ars4 mutation was identified as an allele of phytochrome A (phyA) and caused increased BSO resistance (Sung et al. 2007). The ars5 mutant, which is affected in a 26S proteasome component, had increased levels of glutathione when exposed to arsenic, accompanied by increased GSH1 and GSH2 transcripts (Sung et al. 2009)

Figure 7.

The effects of ontogeny and drought stress on total tissue contents of homoglutathione plus glutathione (a) and reduction state (b) in 5-week-old soybean (Glycine max cv Williams 82) plants. For the drought study, plants were deprived of water for 15 d until the soil water in the drought treatment had fallen to about half that measured under the optimal watering regime (76.97 ± 2.93). Data show the mean ± SE (n = 3) and are expressed in relative units. L, leaf; N, nodule; TF, trifoliate leaf.

The redox states of the plastoquinone and TRX pools are considered to be important components of the redox regulation model for the control of gene expression that adjusts energetic and metabolic demands to light-induced changes of the photosynthetic apparatus structure (Bräutigam, Dietzel & Pfannschmidt 2010). Glutathione has also been implicated in the signalling pathways that facilitate acclimation of chloroplast processes to high light (Ball et al. 2004). Exposure to photoinhibitory light can lead to increases in leaf H2O2 levels and to oxidation of the glutathione pool (Mateo et al. 2006; Muhlenbock et al. 2008). A light shift that favoured excitation of photosystem II (PSII) relative to photosystem I (PSI) slightly increased the amounts of glutathione in leaves, whereas glutathione levels were slightly decreased following a light shift that favoured excitation of PSI relative to PSII (Bräutigam et al. 2009, 2010). According to the model of Bräutigam et al. (2010), higher γ-ECS activities would be favoured by exposure to light wavelengths that predominantly drive PSII. However, this model remains to be substantiated.

GLUTATHIONE-LINKED SIGNALLING MECHANISMS

Redox regulation is inherent to all energy exchange processes. It is required to balance supply and demand between energy-producing and energy-utilizing processes, as these are driven by redox changes. Of the mechanistic controls that achieve this homeostasis, thiol-disulphide reactions are the best characterized and probably among the most important (Buchanan & Balmer 2005). Key thiol components in enzyme regulation or ROS metabolism include TRX, GRX, glutathione, GPX and PRX (Dietz 2003; Lemaire 2004; Meyer et al. 2008; Rouhier 2010). The roles of TRX in redox signalling are well established. By comparison, the study of glutathione-dependent signalling is still in its infancy, although it is interesting to note that interest in the roles of glutathione in cellular regulation dates back many years (Wolosiuk & Buchanan 1977). The following discussion first outlines factors that could link glutathione to modification of target protein activity and then analyses some of the issues surrounding the regulation and impact of altered glutathione status in cellular signalling.

Glutaredoxins

Glutaredoxins are also known as thiol transferases. The classical GRX reaction encompasses the reduction of a protein disulphide bond to two thiols with conversion of 2 GSH to GSSG. Some of these enzymes can also catalyse protein S-glutathionylation or de-glutathionylation. Land plants contain a large GRX family subdivided into three classes that can be distinguished on the basis of amino acid motifs in the active site (Lemaire 2004; Rouhier 2010). Most class I GRXs have a TRX-like active site consisting of two cysteines separated by two intervening amino acids. They can catalyse thiol-disulphide exchange but also other reactions such as regeneration of PRX and methionine sulphoxide reductase (MSR; Rouhier et al. 2002; Rouhier, Couturier & Jacquot 2006; Zaffagnini et al. 2008; Tarrago et al. 2009; Gao et al. 2010). Class II GRXs have only a single cysteine at the active site and are therefore sometimes called ‘monothiol’ GRX, though this term is potentially misleading as in at least some cases this cysteine may form a disulphide bridge with another cysteine located distally in the same protein or on another GRX of the same type (Gao et al. 2009a, 2010). Both class I and class II GRX play roles in assembly of iron-sulphur clusters (Rouhier et al. 2007; Bandyopadhay et al. 2008; Rouhier 2010). They can also catalyse protein de-glutathionylation, though some class II GRX do this in a dithiol- rather than GSH-dependent manner, and in vivo may depend on the TRX system for their regeneration (Zaffagnini et al. 2008; Gao et al. 2009a, 2010). While some of these GRX have been characterized biochemically, their in vivo roles remain unclear. Arabidopsis class II GRXs can interact with ion channels and have been implicated in responses to oxidative stress (Cheng & Hirschi 2003; Cheng et al. 2006; Guo et al. 2010; Sundaram & Rathinasabapathi 2010).

GRXs in the third class contain two adjacent cysteines at their active site and form the largest class in land plants. Reverse genetics approaches in Arabidopsis have revealed that three GRX forms are involved in plant development and phytohormone responses through interaction with TGA transcription factors. These are GRX480, implicated in SA-JA interactions, and ROXY1 and ROXY2, which function in petal and flower development (Ndamukong et al. 2007; Li et al. 2009). Overexpression and complementation studies point to biochemical redundancy between these proteins (Li et al. 2009; Wang et al. 2009). Based on these observations, the specificity of class III GRX function may be determined by regulation of their expression patterns (Ziemann, Bhave & Zachgo 2009). This notion receives some support from microarray analysis of GRX gene expression in mutants with perturbations in leaf glutathione redox state, in which specific members of class III were the only GRX that were found to respond (Mhamdi et al. 2010a). Among these four genes was GRX480, which showed expression patterns similar to many other JA-associated genes. In contrast, transcripts for ROXY1 or ROXY2 remained at wild-type abundance levels.

Protein S-glutathionylation

Protein S-glutathionylation involves the formation of a stable mixed disulphide bond between glutathione and a protein cysteine residue. This reaction could modify the conformation, stability or activity of the target protein. The existence of S-glutathionylated plant proteins has been known for some time (Butt & Ohlrogge 1991). However, the nature and role of this phenomenon has only been subject to thorough investigation relatively recently. Reversible S-glutathionylation is part of the catalytic cycle of glutathione-dependent enzymes such as GR, DHAR, MSRB1, as well as some GRXs and PRXs (Arscott, Veine & Williams 2000; Dixon et al. 2002; Tarrago et al. 2009; Rouhier 2010). Several techniques have been developed to identify other target proteins, allowing an inventory of potential target proteins to be established (Ito, Iwabuchi & Ogawa 2003; Dixon et al. 2005; Michelet et al. 2005, 2008; Zaffagnini et al. 2007; Holtgrefe et al. 2008; Gao et al. 2009a). Despite these advances, major gaps in our current knowledge remain, for example concerning quantification (i.e. the fraction of a given protein that is S-glutathionylated at a given time) and the in vivo significance of the modification. An additional uncertainty is the specificity of the modification, that is, whether the modified cysteine residue is glutathionylated in vivo, rather than nitrosylated or a target of TRX (or some combination of these modifications).

Most targets of glutathionylation identified to date are relatively abundant proteins. These notably include enzymes involved in primary metabolism such as TRXf, glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and glycine decarboxylase (GDC; Michelet et al. 2005; Zaffagnini et al. 2007; Palmieri et al. 2010). Glutathionylation of a TRXf cysteine found outside the active site lowers the protein's efficacy in activating chloroplast NADP-GAPDH, while another TRX-independent isoform of GAPDH is inhibited by glutathionylation (Zaffagnini et al. 2007). In the mitochondria, GDC has been shown to be inhibited by oxidative stress (Taylor, Day & Millar 2002), and is subject to thiol modifications, including S-glutathionylation (Palmieri et al. 2010). This enzyme is a major producer of NADH during conditions of active photorespiration. Mitochondrial redox homeostasis in these conditions may depend on appropriate activation of the alternative oxidase (Igamberdiev, Bykova & Gardeström 1997), which has been shown to be redox regulated through the TRX system (Vanlerberghe et al. 1995; Gelhaye et al. 2005). The glycolytic NAD-dependent GAPDH, a cytosolic enzyme found not only in the cytosol but also in the mitochondrion and nucleus (Giegéet al. 2003; Anderson, Ringenberg & Carol 2004), is sensitive to oxidation and has been identified as a glutathionylated protein (Dixon et al. 2005; Hancock et al. 2006; Holtgrefe et al. 2008).

Taken together, the current evidence suggests that glutathione and TRX systems work closely together to fine-tune photosynthetic and respiratory metabolism through appropriate modification of sensitive protein cysteine residues. The mechanisms of S-glutathionylation may involve GRX-catalysed thiol-disulphide exchange between protein thiol groups and GSSG, conversion of thiol groups to thiyl radicals or sulphenic acids, or be GSNO mediated (Dixon et al. 2005; Holtgrefe et al. 2008; Gao et al. 2009b; Palmieri et al. 2010). The reverse reaction, deglutathionylation, could be mediated by certain class I and class II GRX. Current evidence suggests that some class II GRX catalyse this reaction in a TRX reductase-dependent manner (Zaffagnini et al. 2008; Gao et al. 2009a, 2010). Despite the potential physiological importance of S-glutathionylation, and the proposed involvement of class I and class II GRX in either glutathionylation or de-glutathionylation, no GRX of either type responded at the transcript level in Arabidopsis mutants with increased H2O2 and GSSG (Mhamdi et al. 2010a).

Protein S-nitrosylation and GSNO reductase

GSNO can cause S-nitrosylation and S-glutathionylation of protein cysteine residues (Lindermayr et al. 2005, 2010; Romero-Puertas et al. 2008; Lindermayr & Durner 2009). Enzymes undergoing both modifications include GAPDH and GDC, with both effects leading to decreased activity (Lindermayr et al. 2006; Holtgrefe et al. 2008; Palmieri et al. 2010). Oxidative inhibition of GAPDH may act to regulate the oxidative pentose pathway relative to glycolysis in stress conditions (Holtgrefe et al. 2008). Inhibition of GDC by modifications of thiols may also have physiological significance, as photorespiratory metabolism involves high-flux pathways capable of modifying intracellular redox states (Foyer et al. 2009). S-nitrosylation of type II PRX has been proposed to play a role in ROS signalling (Romero-Puertas et al. 2007).

It was reported that S-nitrosylation acts in opposition to disulphide reduction by TRX in the regulation of NPR1. While monomerization of the NPR1 protein to the active form was TRX dependent, GSNO-dependent nitrosylation of NPR1 monomers was required for reoligomerization, possibly to inhibit depletion of the protein (Tada et al. 2008). However, redox regulation of NPR1 and the transcription factors with which NPR1 interacts is turning out to be complex. Four cysteine residues on TGA1 underwent S-nitrosylation and S-glutathionylation after GSNO treatment of the purified protein, while GSNO treatment of Arabidopsis protoplasts caused translocation of an NPR1-GFP fusion protein into the nucleus (Lindermayr et al. 2010). The importance of S-nitrosylation in regulating NPR1 function is in line with an important role for this modification in plant disease resistance. Production of the plant hormone ethylene, which regulates developmental responses like senescence as well as responses to certain pathogens, is known to be inhibited by NO. This effect may be mediated by inhibition of several enzymes involved in ethylene synthesis, including GSNO-triggered S-nitrosylation of a specific methionine adenosyltransferase (Lindermayr et al. 2006).

Inducible NO-producing enzymes in plants remain to be clearly described, and most attention has focused on the importance of a secondary reaction of nitrate reductase. However, studies in animals and plants have revealed that GSNOR is a key player in regulating S-nitrosylation. These enzymes catalyse the NADH-dependent reduction of GSNO to S-aminoglutathione, which then decomposes to free GSH and ammonia or other compounds (Sakamoto et al. 2002; Barroso et al. 2006; Díaz et al. 2003). Originally identified as a formaldehyde dehydrogenase, the Arabidopsis GSNOR appears to play a key role in biotic stress responses, and Atgsnor1 mutants show decreased resistance to virulent and avirulent pathogens (Feechan et al. 2005). This enzyme has also been implicated in the regulation of cell death and other functions (Lee et al. 2008; Chen et al. 2009). It is also interesting to note that carbon monoxide (CO) is now considered to be a gaseous signalling molecule in animals and plants. It is possible that glutathione is involved in CO signalling in Medicago sativa (Han et al. 2008).

Factors affecting the glutathione redox potential

Despite the advances described previously, it remains unclear to what extent glutathione-mediated changes in protein thiol-disulphide status are important in signalling. Regulation of thiol-disuphide status has traditionally been associated with TRX (Buchanan & Balmer 2005). However, genetic support for the physiological importance of glutathione-dependent disulphide reduction comes from the observations that the NADPH-TRX and glutathione systems play overlapping roles in development (Reichheld et al. 2007; Bashandy et al. 2010). A key factor governing interactions between glutathione and protein targets is redox potential, which depends on the relative rates of glutathione oxidation and its NADPH-dependent reduction. The actual glutathione redox potential is related to [GSH]2:GSSG. Thus, unlike many other redox couples (e.g. NADP+/NADPH), the glutathione redox potential depends on and can be influenced by absolute concentration as well as by changes in GSSG relative to GSH (Mullineaux & Rausch 2005; Meyer 2008). Even if the GSH:GSSG ratio remains unchanged, decreases in glutathione concentration alone will lead to an increase in redox potential, that is, the potential will become more positive and thus less reducing. This is in accordance with measurements of cytosolic GRX-dependent thiol/disulphide redox potential as detected by redox-sensitive roGFP in the glutathione-deficient cad2 mutant compared with the wild type (Meyer et al. 2007). Increases in the cytosolic but not plastidial redox potential in clt1 clt2 clt3 mutants, lacking the chloroplast envelope glutathione transporters, are also consistent with a depleted glutathione pool in the cytosol but not the chloroplast (Maughan et al. 2010).

Linking glutathione redox potential to target proteins

No components have yet been identified that directly link glutathione redox potential to modifications in target proteins. However, based on information from the much better studied TRX-dependent changes in protein thiol-disulphide status, we can infer that a change in glutathione redox potential of approximately 50 mV is likely to be biologically significant. Such changes are sufficient to alter the balance between oxidized and reduced forms of TRX-regulated proteins (Setterdahl et al. 2003). For example, an increase in TRX redox potential from −350 to −300 mV converts chloroplast glucose-6-phosphate dehydrogenase from almost completely inactive to active (Née et al. 2009). Indeed, in vitro redox titration of the thiol/disulphide-dependent roGFP in the presence of GRX at different glutathione concentrations revealed that interconversion of the oxidized and reduced forms of the sensor were associated with a calculated redox potential span of about 50 mV (Meyer et al. 2007). Despite this, calibrated quantification of in vivo changes in cytosolic redox potential using these probes has thus far revealed shifts of only about 20 mV in glutathione-deficient mutants compared with wild-type or in stress versus optimal conditions (Meyer et al. 2007; Jubany-Mari et al. 2010). Nevertheless, the physiological importance of such a shift cannot be discounted because systems that are able to sense glutathione redox potential with high sensitivity may await discovery. Furthermore, based on recent reports of interactions between cytysolic glutathione and TRX systems (Marty et al. 2009), it is possible that in some conditions changes in glutathione redox potential could influence the mechanisms that contribute to changes in the TRX redox potential and thus the biological activity of TRX targets.

Physiological and environmental factors that might cause shifts in glutathione redox potential in vivo

As previously discussed, the whole leaf and chloroplast GSH:GSSG ratios are not markedly influenced by variations in light intensity or light-dark transitions. However, oxidant production and oxidative stress can exert a strong influence over glutathione status. Oxidant production is governed by the rates of photosynthesis, photorespiration and respiration, as well as NADPH oxidases and other ROS-producing systems (Foyer & Noctor 2003). Presumably by changing the rate of ROS production through these processes or by affecting the rate of ROS removal, various stresses such as cold, drought, pollution and pathogens can modify whole leaf glutathione redox state (Sen Gupta, Alscher & McCune 1991; Vanacker et al. 2000; Bick et al. 2001; Gomez et al. 2004b). Indeed, the cytosolic glutathione redox potential has been shown to increase in response to wounding (Meyer et al. 2007). Numerous studies on plants deficient in H2O2-metabolizing enzymes have documented marked changes in GSH:GSSG ratios (Smith et al. 1984; Willekens et al. 1997; Rizhsky et al. 2002; Queval et al. 2007, 2009). These are most evident for catalase-deficient plants exposed to light under photorespiratory conditions, and are accompanied by several-fold changes in the total glutathione pool. It remains to be established to what extent these changes in whole tissue GSH:GSSG ratios involve changes in glutathione redox potential in specific compartments.

The importance of changes in absolute glutathione concentration compared with changes in GSH:GSSG

Accumulation of GSSG is often followed by an increase in the total glutathione pool size (Fig. 8). This response has been documented in plants exposed to ozone, is particularly evident in response to increased intracellular H2O2 and also occurs during pathogen responses. In such circumstances, changes in the total glutathione pool size occur predominantly through oxidant-induced changes in the expression and post-translational activities of enzymes involved in both cysteine and glutathione synthesis. This presumably acts as a homeostatic mechanism to offset what would otherwise be more severe increases in glutathione redox potential. If so, this response in itself suggests that the glutathione redox potential is important for at least some cellular functions. Further circumstantial evidence in favour of this notion comes from the observation that accumulation of GSSG is compartment specific, and relative increases are most marked in the vacuole (Queval et al. 2011). This is in line with the documented capacity of vacuoles to important GGSG more efficiently than GSH (Tommasini et al. 1993). Thus, mechanisms appear to have evolved to stabilize cytosolic glutathione redox potential during increased ROS production, and these include increases in total glutathione and accumulation of GSSG in compartments that may be less sensitive to redox perturbation (Fig. 8).

Figure 8.

Hypothetical scheme showing some glutathione-linked responses in oxidative stress signalling. Increased engagement of GSH in metabolism of H2O2 or other oxidants causes accumulation of GSSG and thus a more positive glutathione redox potential. Signalling initiated by this change includes activation of glutathione neosynthesis and transport, both of which act to offset increases in redox potential. Modified glutathione redox potential could contribute to signalling through components such as glutaredoxins or protein S-glutathionylation or, more indirectly, by oxidation of thioredoxins (TRX). Both thiol components may contribute to thiol-disulphide signalling, which is depicted as one part of a larger reactive oxygen species (ROS) signalling network.

Although studies of mutants have shown that relatively small changes in glutathione concentration can modulate gene expression as well as environmental and developmental processes in plants (Ball et al. 2004; Parisy et al. 2006; Bashandy et al. 2010), it is not yet established that such effects are caused by altered glutathione redox potential. A decrease in total glutathione from 2.5 mm to 50 µm would cause the redox potential to become 50 mV more positive (Meyer et al. 2007). This depletion is rather extreme, and corresponds to the situation in root tips in seedlings of the rml1 mutant, which contain less than 5% wild-type glutathione (Cairns et al. 2006; Meyer et al. 2007). Thus, while an increased glutathione redox potential could explain the observations reported in rml1 and ntra ntrb rml1 mutants (Vernoux et al. 2000; Reichheld et al. 2007), it is less clear whether it can account for effects observed in less severely affected backgrounds. If changes in glutathione redox potential are indeed an important part of glutathione-dependent oxidative signalling in wild-type plants, it seems that GSSG accumulation is likely to be a critical factor. Accumulation of GSSG could be relatively small (in absolute terms) if, as discussed further in the next section, the ‘resting state’ concentration in unstressed conditions is as low as some recent measurements imply.

As increased GSSG generally drives glutathione accumulation as part of what is assumed to be a homeostatic mechanism, decreased glutathione concentrations might be expected to cause a compensatory increase in GSH:GSSG. This should happen if the NADP(H) redox potential does not change and the two couples are in equilibrium. However, if GSSG concentrations are indeed in the nanomolar range, it is possible that kinetic limitations over GR (KMGSSG 10–50 µm; Smith et al. 1989; Edwards et al. 1990) become increasingly severe as glutathione decreases, thus explaining the failure to maintain the glutathione redox potential in equilibrium with NADP(H), even when the decrease in concentration is relatively modest.

Relationships between the glutathione and NADP(H) redox couples in signalling

The midpoint redox potential of NADP(H) at pH 7 is approximately −320 mV while that of glutathione is −230 to −240 mV. But what is the actual redox potential of glutathione in vivo? Many researchers assume that it is quite close to the midpoint potential and that differences between this value and the TRX redox potential, which for most TRX is quite similar to NADP(H), is one factor that potentially explains in vivo preference of protein targets for glutathione or TRX. However, as discussed previously, some roGFP studies suggest that the glutathione redox potential in wild-type Arabidopsis in the absence of stress is lower than −300 mV in the cytosol and in other compartments, that is, it approaches values similar to the TRX redox potential (Meyer et al. 2007; Schwarzländer et al. 2008; Jubany-Mari et al. 2010). Given that the cytosolic NADPH:NADP+ ratio is not far removed from one (Igamberdiev & Gardeström 2003), these redox potential values suggest that glutathione and NADP(H) are close to redox equilibrium in this compartment. This is a key unresolved issue in assessing the potential significance of redox signalling through glutathione-mediated disulphide bond formation. If sensitive thiol proteins are present, maintenance of a very low glutathione redox potential under optimal conditions could allow oxidative signalling to be initiated by protein disulphide bond triggered by relatively minor accumulation of GSSG. Glutathione redox potential values below −300 mV would require the GSH:GSSG ratio to be in the range of 105 to 106, whereas global ratios are much lower in plant tissues, even in the absence of stress, where they are typically about 20–30 (Queval & Noctor 2007; Mhamdi et al. 2010a). It is likely that, while they provide a useful indicator of redox status, particularly the presence of oxidative stress, global cellular or tissue glutathione data give only a relative measure of GSH:GSSG ratios in intracellular compartments (Queval et al. 2011).

There is little evidence as yet that ROS-triggered changes in glutathione are accompanied by marked changes in global tissue NADP(H) status (Mhamdi et al. 2010a,b). The extent to which cytosolic NADP pools turn over in optimal and stress conditions remains uncertain. Further, it is not clear how important the demand of GR is on the cytosolic NADPH pool, and whether there is any functional association between different NADPH-requiring and -generating enzymes. GR has a KM for NADPH below 10 µm (Smith et al. 1989; Edwards et al. 1990), while total cytosolic NADPH concentrations are above 100 µm (Igamberdiev & Gardeström 2003). Although a large proportion of pyridine nucleotide pools is bound to proteins at any one time, and thus available ‘free’ concentrations are lower than the total, it is generally considered that changes in NADPH are not likely to impact the glutathione redox potential through kinetic effects on GR. However, shifts in NADPH:NADP+ could perhaps impact the glutathione redox potential thermodynamically, as the reaction catalysed by GR is reversible. As previously discussed, kinetic limitations may occur through GSSG, whose resting concentration could be substantially below the GR KM value. Finally, it is interesting that the measurable capacity of GR in many plant systems is often somewhat lower than other enzymes of the ascorbate-glutathione pathway. No doubt this could reflect the presence of other pathways that can regenerate glutathione independent of ascorbate (Fig. 6). However, as we have emphasized in this review, glutathione may also be linked to ROS metabolism independently of ascorbate. Comparatively low GR activity may be a feature of plant redox systems that allows appropriate changes in GSSG to mediate signalling functions.

CONCLUSION AND PERSPECTIVES

A large body of literature is now available that illustrates the complex and integrated regulation of glutathione status by nutritional and environmental factors. Glutathione may be seen as a central or ‘hub’ molecule in cellular metabolism and redox signalling (Fig. 9). Even though many glutathione-dependent reactions are involved in oxidative stress-mediated cellular processes, the simple vision of glutathione as a ROS-scavenging antioxidant is inaccurate and misleading. For example, at least three types of mechanism can be defined that link glutathione to the regulation of biotic stress reactions: (1) redox signalling, mediated by components such as GRX or GSNO; (2) the synthesis of S-containing or secondary metabolites; and (3) GS-conjugate formation and metabolism to regulate the biological activity of metabolic intermediates or active products.

Figure 9.

Physiological importance of glutathione. The outer green circle shows examples of physiological processes affected by glutathione status or by glutathione-dependent components. The inner pink circle gives examples of major biochemical functions of glutathione. The darker pink spokes indicate components that are involved in these functions.

Key questions related to the antioxidant roles of glutathione concern the importance of different glutathione-dependent enzymes and the influence of changes in glutathione status within cell signalling pathways. The degree of interplay between TRX and glutathione systems in physiological conditions remains to be fully described. The emerging picture is that changes in the ‘general redox state’ (i.e. redox potential) of these two components work together with altered activity or abundance of specific redox signalling components (e.g. GSNOR, GRX). Through their coordinated operation, such effects can reinforce signalling through a given pathway. Alternatively, antagonistic changes could act to restrain activation of signalling pathways, a control that is important within a cellular network receiving multiple evironmental inputs. This concept receives support from the operation of multiple interdependent redox modifications reported for the bacterial H2O2 sensor, OxyR (Kim et al. 2002). A similar picture is emerging in plants, where the notion of relatively simple thiol-disuphide regulation of NPR1 (Mou et al. 2003) has given way to recognition that control is more complex (Tada et al. 2008; Lindermayr et al. 2010). By operating in conjunction with other regulatory mechanisms, such nuanced, flexible control integrates multiple inputs into a variable and appropriate response output. This is likely to be particularly important in plants, which must constantly deal with environmental changes of varying intensity and predictability. Thus, the challenge for the future is not only to characterize glutathione-linked redox modifications and to assess their interactions with other types of redox regulation, but also to evaluate the significance of these changes within the wider horizon of the cellular network.

ACKNOWLEDGMENTS

We thank Bernd Zechmann (University of Graz, Austria) for discussions on glutathione concentrations in different subcellular compartments and Stéphane Lemaire (Institut de Biologie Physico-Chimique, Paris, France) for discussions on glutaredoxins and protein glutathionylation reactions. The authors acknowledge funding from the French Agence National de la Recherche project ‘Vulnoz’ (Orsay) and the European Union Marie-Curie Initial Training Network ‘COSI’ project (Orsay, Leeds). Y.H. is a recipient of a Chinese Scholarship Council fellowship. B.M.G. thanks Subprograma Estancias de Movilidad posdoctoral en centros extranjeros (2009), Ministerio de Educación (Spain) for a fellowship.

Ancillary