Two GPX-like proteins from Lycopersicon esculentum and Helianthus annuus are antioxidant enzymes with phospholipid hydroperoxide glutathione peroxidase and thioredoxin peroxidase activities

Authors


  • Enzymes: catalase (EC 1.11.1.6); glutathione peroxidase (EC 1.11.1.9); glutathione reductase (EC 1.6.4.2); glutathione-S-transferase (EC 2.5.1.18); l-ascorbate peroxidase (EC 1.11.1.11); thioredoxin reductase (EC 1.6.4.5); superoxide dismutase (EC 1.15.1.1).

P. Roeckel-Drevet, UMR 547-PIAF INRA/ Université Blaise Pascal, 24 avenue des Landais, 63177 Aubière, France. Fax: + 33 4 73 40 79 16, Tel.: + 33 4 73 40 79 12, E-mail: Patricia.DREVET@univ-bpclermont.fr

Abstract

This study investigated the enzymatic function of two putative plant GPXs, GPXle1 from Lycopersicon esculentum and GPXha2 from Helianthus annuus, which show sequence identities with the mammalian phospholipid hydroperoxide glutathione peroxidase (PHGPX). Both purified recombinant proteins expressed in Escherichia coli show PHGPX activity by reducing alkyl, fatty acid and phospholipid hydroperoxides but not hydrogen peroxide in the presence of glutathione. Interestingly, both recombinant GPXle1 and GPXha2 proteins also reduce alkyl, fatty acid and phospholipid hydroperoxides as well as hydrogen peroxide using thioredoxin as reducing substrate. Moreover, thioredoxin peroxidase (TPX) activities were found to be higher than PHGPX activities in terms of efficiency and substrate affinities, as revealed by their respective Vmax and Km values. We therefore conclude that these two plant GPX-like proteins are antioxidant enzymes showing PHGPX and TPX activities.

Abbreviations
ROS

reactive oxygen species

GPX

glutathione peroxidase

GSH

glutathione

PHGPX

phospholipid hydroperoxide glutathione peroxidase

TPX

thioredoxin peroxidase.

In all aerobic organisms, reactive oxygen species (ROS) originating from the metabolism of oxygen constitute a threat to virtually any cell constituent. In plants, it has been shown that environmental stresses can cause an increase in ROS levels [1–4]. Despite their noxious effects on proteins, lipids and nucleic acids, which could ultimately lead to cell death, ROS, in a more controlled manner, can participate in early signaling pathways in responses to both biotic and abiotic stresses [5,6]. To cope with elevated levels of ROS, plants have evolved different enzymatic and nonenzymatic mechanisms. In the latter are found reducing molecules such as carotene, tocopherol, ascorbate, Fe2+, glutathione, while the antioxidant enzymatic equipment is composed of several enzymes including superoxide dismutase (SOD), ascorbate peroxidase (APX), catalase, glutathione reductase (GR), glutathione peroxidase (GPX), glutathione-S-transferase (GST) or thioredoxin peroxidase (TPX).

In mammals, the GPX family of proteins can be divided into five clades according to their amino-acid sequence, substrate specificity and subcellular localization; the cytosolic GPX (GPX1), the gastro-intestinal GPX (GPX2), the plasma GPX (GPX3), the phospholipid hydroperoxide GPX (GPX4) and selenoindependent epididymis GPX (GPX5) [7,8].

To date, cDNAs encoding proteins similar to animal GPX have been isolated from different plants and have been shown to be induced by biotic and abiotic stresses [9–11]. Plant GPX-like proteins exhibit the most identities to mammal selenium dependent GPX4. However, plant genes carry a codon for a cysteine residue instead of the opal codon UGA used for insertion of a selenocysteine in mammal GPXs. The selenocysteine residue is important for the catalytic activity of GPX as replacement of selenocysteine by cysteine greatly reduces the activity of the enzyme [12]. According to Eshdat et al. [13], this would result in a plant activity lower by three orders of magnitude when compared to the homologous animal GPX. However, replacement of the cysteine by a selenocysteine residue in the citrus GPX was not followed by a gain in activity comparable to that observed with selenium-dependent animal GPX [14]. Thus, the physiological role of plant GPXs is not yet clear. Furthermore, emerging reports on different living organisms display opposite results about the enzymatic functions of these GPX-like proteins [15–18]. These data prompted us to explore the enzymatic functions of these proteins in higher plants. In the present study, we characterized the expression in E. coli of two plant GPXs, GPXle1 and GPXha2, from Lycopersicon esculentum and Helianthus annuus, respectively. The purified recombinant proteins were obtained and used in enzymatic assays with various substrates in order to investigate their putative function.

Materials and methods

Plant materials and chemicals

Tomato (Lycopersicon esculentum Mill. cv. VFN8) and sunflower plants (Helianthus annuus Hybrid EL64, kindly provided by F. Vear, INRA, Clermont-Ferrand, France) were raised from seeds in moist vermiculite in a controlled environment room: 16 h daylight at 60 µmol·m−2·s−1, photosynthetically active radiation provided by 40-W white daylight tubes (Mazda LDL, TF 40), 23 ± 1 °C (day) and 19 ± 1 °C (night), 60 ± 10% relative humidity. At the cotyledon stage, tomato plants were transferred to a mineral solution [19], while sunflower plants were grown in pots. Glutathione, Saccharomyces cerevisiae glutathione reductase, β-NADPH, E. Coli thioredoxin, E. Coli thioredoxin reductase, Triton X-100 (peroxide free), t-butyl hydroperoxide, cumene hydroperoxide, hydrogen peroxide, linoleic acid, l-α-phosphatidylcholine dilinoleoyl and soybean lipoxidase (type IV) were purchased from Sigma (Saint Quentin Fallavier, France). Linoleic acid and l-α-phosphatidylcholine dilinoleoyl hydroperoxides were prepared using soybean lipoxidase as described previously [20]. Hydroperoxides formation was monitored by following the change in absorbance at 234 nm and their concentration calculated using an ε-value of 25 000 m−1·cm−1. The hydroperoxides were stored in ethanol at −20 °C.

Heterologous expression and purification of recombinant GPXle1 and GPXha2

Total RNA was extracted from Lycopersicon esculentum internodes and Helianthus annuus leaves according to the method of Hall [21]. Full-length cDNAs encoding GPXle1 (GenBank accession number y14762) and GPXha2 (GenBank accession number y14707) were amplified by reverse-transcription and PCR amplification as described by Drevet et al. [22] using total RNA as template. During amplification, the cDNAs were tagged with NdeI sites using appropriate primers. The sequence of the primers used in this study were 5′-GAATTCGACATATGGCTACGC-3′/5′-GCTCTCCCATATGGTCG-3′ and 5′-CGATAAGCA TATGGCTACGC-3′/5′-GAATACTCAACATATGCAT CC-3′ for each set of forward/reverse gpxle1 and gpxha2 primers, respectively. Amplified products were subsequently cloned into the NdeI linearized pET15b vector (Novagen, Fontenay-sous-bois, France) at the NdeI site to give in-frame fusion with a His6 tag, and transformed in E. coliBL21 (DE3) pLysS (Promega, Charbonnieres, France). For both clones, sequence fidelity and proper insertion were checked out by automated dye terminator sequence analysis using the CEQ 2000 sequencer (Beckman-Coulter, Roissy Charles De Gaulle, France). Clones were grown in ampicillin (100 mg·L−1)-supplemented Luria–Bertani media at 37 °C up to D600 = 0.6 and induced with 0.5 mm isopropyl thio-β-d-galactoside. Four hours after induction, cells were harvested by centrifugation (5000 g, 10 min, 4 °C) and resuspended in 0.05 m sodium phosphate, 0.3 m NaCl, 0.02 m imidazole at pH 7.5. The cells were then disrupted by sonication at 10 kHz for a total of 60 s with five intervals of 20 s each, and cell debris were sedimented by centrifugation (10 000 g, 30 min, 4 °C). The presence of the expected soluble recombinant protein was ascertained by SDS/PAGE. The His-tagged protein products of GPXle1 and GPXha2 were affinity purified from cell extracts on Ni2+-nitrilotriacetic acid matrix column according to the manufacturer's instructions (Qiagen, Courtaboeuf, France). Protein concentrations in the eluted fractions were determined using the Bradford assay [23] and fractions containing the protein peaks were assayed immediately for enzymatic activity. As a control, cultures of E. coliBL21 (DE3) pLysS transformed with pET15b vector alone were treated as indicated above in parallel experiments.

Enzymatic assays

Glutathione-dependent peroxidase activity was measured by monitoring NADPH oxidation with spectrophotometry at 340 nm [24]. A standard reaction mixture (1 mL), containing 100 mm Tris/HCl, pH 7.5, 5 mm EDTA, 0.2 mmβ-NADPH, 3 mm GSH, 0.1% (v/v) triton X-100, 1.4 U of glutathione reductase and 50–100 µg of recombinant protein, was incubated at 30 °C for 5 min. After 3 min of equilibration, the reaction was initiated by the addition of the peroxide substrate. The nonenzymatic activity due to auto-oxidation of GSH as well as the activity of any potentially co-purified E. coli proteins were also examined. Corrections were made to estimate the activity of recombinant proteins per se. Enzyme activities were calculated using an ε-value of 6220 m−1·cm−1. For measurement of thioredoxin-dependent peroxidase activity, GSH and glutathione reductase in the above-mentioned mixture were replaced with E. coli thioredoxin (4 µm) and E. coli thioredoxin reductase (0.3 U·mL−1), respectively. NADPH-dependent peroxidase activity was assayed in a similar fashion to glutathione-dependent peroxidase activity, except that GSH and glutathione reductase were not added to the reaction mixture.

Results

Heterologous expression of GPXle1 and GPXha2

E. coli BL21 (DE3) pLys cells transformed with the pET15b-derived expression plasmid efficiently produced GPXle1 or GPXha2, as indicated by the presence of a prominent band slightly greater than 20 kDa using SDS/PAGE [Fig. 1]. This apparent molecular mass was in agreement with the expected molecular mass (2181 Da from the His6 tag plus 18 847 Da from GPXle1 or 19 175 Da from GPXha2). The purification scheme using Ni-nitrilotriacetic acid affinity matrix yielded a product of apparent electrophoretic homogeneity (Fig. 1). No product was purified from extracts from E. coli transformed with pET15b vector alone (data not shown).

Figure 1.

Analysis by SDS/PAGE of the recombinant GPXle1 and GPXha2 proteins expressed in E. coli cells and purified by Ni-nitrilotriacetic acid affinity. Each crude extract (10 µg of protein) and purified recombinant enzyme (1 µg of protein) were analyzed by 15% SDS/PAGE. Lane 1, pET/GPXle1-transformed E. coli; lane 2, purified recombinant GPXle1; lane 3, pET/GPXha2-transformed E. coli; lane 4, purified recombinant GPXha2. Proteins were stained with Coomassie brilliant blue. Positions and sizes of molecular mass protein markers are shown on the left side of the panel.

Enzymatic properties of GPXle1 and GPXha2

The glutathione peroxidase activities of GPXle1 and GPXha2 towards several physiological and nonphysiological hydroperoxides were monitored in the presence of glutathione and glutathione reductase. Assays were carried out using purified recombinant proteins or, as negative controls, using either extracts from E. coli transformed with the pET15b vector alone that had been affinity purified in parallel or elution buffer alone. Such controls accounted for any nonenzymatic background due to auto-oxidation of GSH and also any E. coli peroxidase activity that might have co-purified with the recombinant proteins. We found no difference between NADPH oxidation in the presence of affinity purified extracts from the E. coli control or in the presence of the elution buffer alone. These data suggested that no E. coli peroxidase activity copurified with our recombinant proteins. Under these conditions, the apparent Km and Vmax values for a variety of substrates were calculated for GPXle1 and GPXha2 (Table 1). Both proteins exhibited a higher affinity towards phospholipid hydroperoxides and a weaker affinity towards t-butyl hydroperoxide, as indicated by their respective apparent Km values. There was no detectable activity with hydrogen peroxide.

Table 1.  Glutathione peroxidase activities of GPXle1 and GPXha2 towards different substrates. Glutathione peroxidase assays were performed as described in Experimental procedures with a fixed concentration of GSH (3 mm) using four or five different concentrations of peroxide. The data were analyzed by a Linewaever–Burk representation. Apparent maximum velocities (App. Vmax), apparent maximum Michaelis constant (App. Km) values (± SEM) and Vmax/Km ratios are shown as the average of three independent experiments. The Cit-sap protein values were taken from reference [32]. LA-OOH, linoleic acid hydroperoxide; PCdili-OOH, phosphatidylcholine dilinoleoyl hydroperoxide; t-butyl-OOH, ter-butyl hydroperoxide.
SubstrateGPXle1GPXha2Cit-Sap (citrus)
App. Vmax
(nmol·min−1·mg−1)
App. Km
m)
V max/KmApp. Vmax
(nmol·min−1·mg−1)
App. Km
m)
V max/KmApp. Vmax
(nmol·min−1·mg−1)
H2O200 0
t-Butyl-OOH37.7 ± 2.711280.29427.1 ± 1.4595.3 ± 2.560.28424
Cumene-OOH57.5 ± 1.27119.0 ± 1.393.0338.9 ± 0.6160.8 ± 1.370.64050
LA-OOH27.7 ± 0.04 39.3 ± 0.030.70542.4 ± 0.1182.7 ± 3.470.51644
PCdili-OOH19.0 ± 0.44 24.9 ± 0.820.76315.8 ± 0.4412.1 ± 0.551.3140

Considering the replacement of the selenocysteine, one of the catalytic residues known to be critical for animal GPX activity, and considering the sequence identities with PHGPX (GPX4) that was reported to have no specificity towards GSH [25], we have investigated the electron donor requirements of GPXle1 and GPXha2. Three alternative physiological reducing substrates, GSH, thioredoxin and NADPH, were tested. Peroxidase activities were assayed with a fixed t-butyl hydroperoxide concentration (100 µm) using four to five different reducing substrate concentrations. As carried out for GPX activity, control assays accounted for any nonenzymatic NADPH oxidation and also for any co-purified E. coli peroxidase activity. A thioredoxin-dependent peroxidase activity was found for both recombinant proteins in addition to the GPX activity. Double reciprocal plots of 1/activity against 1/[GSH](Fig. 2A) or 1/[thioredoxin] (Fig. 2B) were linear and reproducible in each case. Under these conditions, apparent Km and Vmax values were calculated (Table 2). Neither GPXle1 nor GPXha2 were able to reduce t-butyl hydroperoxide (Table 2) or others peroxides (data not shown) using NADPH as reducing substrate. Both plant enzymes showed higher affinity by three orders of magnitude towards E. coli thioredoxin than to GSH, as indicated by apparent Km values. Moreover, in reducing t-butyl hydroperoxide, apparent Vmax values revealed a thioredoxin-dependent peroxidase activity fivefold higher than glutathione-dependent peroxidase activity. For both proteins, the catalytic efficiencies (Vmax/Km) in the presence of thioredoxin are a lot higher than in the presence of glutathione [Table 2]. Thus, recombinant GPXle1 and GPXha2 presented a TPX activity, albeit a slight GPX activity. Substrate specificities of the TPX activity was further investigated using a fixed concentration of E. coli thioredoxin (4 µm) and four to five different substrate concentrations (Table 3). In agreement with the above data, whichever the tested substrate, TPX activity was found to be greater than the GPX activity in terms of efficiency and substrate affinity (Tables 1 and 3). Furthermore, both enzymes were able to reduce hydrogen peroxide, as well as linoleic acid, phosphatidylcholine dilinoleoyl and t-butyl hydroperoxides, using thioredoxin as reducing substrate whereas such an activity was not detected in the presence of GSH.

Figure 2.

Analysis of GPXle1- and GPXha2-catalyzed reduction of t-butyl hydroperoxide (100 µ m ) with different concentrations of GSH (A ) and thioredoxin (B ). The reciprocal apparent maximum velocities of GPXle1 (●) and GPXha2 (○) are plotted against the reciprocal GSH concentrations (1–10 mm) or E. coli thioredoxin concentrations (1–6 µm) as a Linewaever–Burk representation. Each value (± SEM) is representative of three experiments. GPX and TPX activities are expressed as nmol of NADPH oxidized per min per mg of protein, and GSH concentrations are expressed in mm whereas thioredoxin concentrations are expressed in µm.

Table 2.  Reducing substrate specificities of GPXle1 and GPXha2 in catalyzed reduction of t-butyl hydroperoxide (100 µ m ). Peroxidase assays were performed as described in experimental procedures with a fixed concentration of t-butyl hydroperoxide (100 µm) using four or five different reducing substrate concentrations. The reducing substrates tested are GSH (1–10 mm), NADPH (100–200 µm) and E. coli thioredoxin (1–6 µm). The data were analyzed by a Linewaever–Burk representation as illustrated in Fig. 2. Apparent maximum velocities (App. Vmax), apparent maximum Michaelis constant (App. Km) values (± SEM) and Vmax/Km ratios are shown as the average of three independent experiments.
SubstrateGPXle1GPXha2
App. Vmax
(nmol·min−1·mg−1)
App. Km
m)
V max/KmApp. Vmax
(nmol·min−1·mg−1)
App. Km
m)
V max/Km
Glutathione48.8 ± 4.569300 ± 2095.24 × 10−346.7 ± 3.894900 ± 1209.53 × 10−3
NADPH00
Thioredoxin (E. coli)263.2 ± 0.362.2 ± 0.30119.6243.9 ± 0.501.5 ± 0.06162.6
Table 3.  Thioredoxin-dependent peroxidase activities of GPXle1 and GPXha2 towards different substrates. Thioredoxin-dependent peroxidase assays were performed as described in Experimental procedures with a fixed concentration of E. coli thioredoxin (4 µm) using four different peroxide concentrations. The data were analyzed by a Linewaever–Burk representation. Apparent maximum velocities (App. Vmax), apparent maximum Michaelis constant (App. Km) values (± SEM) and Vmax/Km ratios are shown as the average of three independent experiments. LA-OOH, linoleic acid hydroperoxide; Pcdili-OOH, phosphatidylcholine dilinoleoyl hydroperoxide; t-butyl-OOH, tert-butyl hydroperoxide.
SubstrateGPXle1GPXha2
App. Vmax
(nmol·min−1·mg−1)
App. Km
m)
V max/KmApp. Vmax
(nmol·min−1·mg−1)
App. Km
m)
V max/Km
H2O2153.8 ± 1.7913.7 ± 0.0211.2147.1 ± 2.1213.9 ± 0.2010.6
t-Butyl-OOH147.2 ± 1.3416.6 ± 0.28 8.87161.3 ± 0.9614.1 ± 0.3511.4
LA-OOH147.1 ± 1.118.60 ± 0.5017.1169.5 ± 1.4616.2 ± 0.3510.5
PCdili-OOH108.7 ± 0.5614.4 ± 0.12 7.55126.7 ± 0.059.44 ± 0.3113.4

Discussion

An increasing number of proteins having at least two functions has been reported [26]. Among the GPX family, the animal GPX4 (PHGPX) has been reported to be both a structural protein and an active enzyme in sperm cells [27]. In addition, the animal selenium-independent and epididymis-restricted GPX (GPX5) was also recently suspected to bear dual-function [8,28].

This report shows that the two previously reported plant GPX-like proteins [10,11] display a thioredoxin-dependent peroxidase activity as well as a glutathione peroxidase activity. Based on identities in their primary sequences with animal GPXs, they were found to be more related to GPX4, the phospholipid hydroperoxide glutathione peroxidase [10,11]. This is also the case for other characterized plant GPXs [13]. The mammalian GPX4 differs from the other animal GPXs in that the protein is monomeric due to deletions in regions thought to mediate tetramerization [25]. The small size and hydrophobic surface of these proteins can explain that PHGPXs (GPX4) are unique in their activity towards hydroperoxides integrated in membranes [29], suggesting that they may play a significant role in protecting membranes from oxidative damage. The sequence similarities led us to suggest that GPX4-like plant GPXs could be involved in membrane protection. Indeed, in our experiments, GPXle1 and GPXha2 were found to display glutathione-dependent peroxidase activity towards organic peroxides such as phospholipid hydroperoxides, but not towards hydrogen peroxide, thus behaving as expected for a GPX4-like GPX. However, these in vitro activities remain low. This can be explained by the lack of the rare selenocysteine residue replaced by a cysteine in the catalytic site of plant GPXs [13]. To date, low activities [15,17,18], or no activity [16,30], were recorded for all seleno-independent GPXs that have been investigated. In addition, GPXle1 and GPXha2 exhibit a low affinity towards GSH and present apparent maximum velocities with glutathione concentrations which are far above evaluated physiological values estimated to range from 1 to 4.5 mm in the chloroplast [31]. Heterologous expressions of GPXle1 and GPXha2 in E. coli do not seem to affect their activity, because in vitro values were found to be similar to those obtained from a plant purified citrus GPX [32]. The low PHGPX activity of GPXle1 and GPXha2 recorded in vitro does not necessarily reflect the in vivo situation and does not rule out the possibility that these proteins are indeed involved in phospholipid hydroperoxides detoxification in the cell. In yeast, it has been reported that PHGPX deletion mutants were sensitive to induced lipid peroxidation, suggesting that this seleno-independent protein protects membranes from oxidative stress [17].

Our in vitro analysis of GPXle1 and GPXha2 enzymatic functions strongly suggests that these two GPXs can also function as thioredoxin peroxidases (TPX). Such a finding was recently reported for a previously characterized GPX from Plasmodium falciparum, which as a consequence has been reclassified as a TPX [18]. In addition, it has been very recently shown that a protein from chinese cabbage, which is highly homologous to PHGPX, functions also as a TPX [33]. Dual function for an antioxidant enzyme has also been recently reported for a human 1-cys peroxiredoxin, which exhibits glutathione peroxidase activity [34], and a bovine eye protein showing homologies to TPXs but acting as a seleno-independent GPX [35]. Our TPX assays rely on the use of exogenous thioredoxin and thioredoxin reductase from E. coli, instead of Lycopersicon esculentum and Helianthus annuus endogenous ones. This bacterial thioredoxin system has successfully been used with the plasmodium TPX protein [18]. As it was the case with the TPX from Plasmodium falciparum, one could expect that GPXle1 and GPXha2 react faster with endogenous thioredoxins from their respective plant species. However, thioredoxin systems are probably not markedly different among living organisms as proved by the fact that an E. coli thioredoxin has been shown to enhance recovery of human cells after oxidative stress [36]. Thus, it is likely that the TPX activities recorded in the present study reflect the activity in plants as well. Supporting further this dual GPX/TPX function, sequence alignments have shown that amino-acid residues necessary for GSH specificity are not conserved in plasma GPX (GPX3) and PHGPX (GPX4) groups (which include GPXle1 and GPXha2), suggesting that GSH is unlikely to be the sole physiological electron donor under all circumstances. For example, GPX3 can use thioredoxin as a reducing substrate [37], and GPX4 exhibits an alternate enzymatic thiol oxidase activity towards thiols contained in various proteins [38]. Nevertheless, GPXle1 and GPXha2 do not accept all reducing substrates, as indicated by the lack of activity when NADPH (Table 2) or NADH (data not shown) were used. This implies that GSH and thioredoxin affinities are somehow specific. Altogether, these data on plant and animal GPXs suggest a putative link existing between the glutathione-based antioxidant system and the thioredoxin-based one.

TPX activities monitored here can be considered physiological. Indeed, apparent Km values for thioredoxin are of a micromolar range, compatible with in vivo levels. An in vivo competition between GSH and thioredoxin for the plant GPXs cannot be ruled out, because of the uncertainties about the ratio between GSH and thioredoxin concentrations in many tissues and physiological circumstances. In line with these considerations, we can assume that the electron donor and therefore the enzymatic function of the proteins would depend on this ratio. Although there is no sequence homologies between our plant GPXs and classical TPXs, some similarities can be found with the PHCC-TPx from chinese cabbage [33]. In particular, there are several Cys residues that can be found at roughly equivalent positions in these proteins. Interestingly, Jung et al. [33] have put forward a putative role played by Cys residues in the dual GPX/TPX catalytic process (i.e. exchange of disulfide bonds) in the chinese cabbage PHCC-TPx. In any case, the efficient TPX activity of GPXle1 and GPXha2 do not exclude a PHGPX function but rather points to other unknown biological roles for plant PHGPXs. Another interesting trait of our results is that both GPXle1 and GPXha2 can reduce hydrogen peroxide in the presence of thioredoxin but not in the presence of GSH. Such data are in agreement with the literature, as classical TPXs [39] are known to metabolize hydrogen peroxide while plant GPX-like enzymes do not [29].

This report shows that in plants, GPXle1 and GPXha2 can behave in vitro both as a GPX or/and as a TPX, provided that the proper substrate and electron donor are available. Considering the various subcellular localizations of plant PHGPX-like proteins [10,11,40,41], the variations in the tissue and the subcellular concentrations of substrates and reducing substrates, dual catalytic activities for a given enzyme might constitute an economical way plant cells have evolved in order to cope with various physiological stresses or situations. Indeed, we have previously shown that both biotic and abiotic stresses were able to increase GPXha2 expression at the mRNA level [11].

Further in vivo investigations such as mutant analysis or modifications of expression in transgenic plants will be necessary to clarify this dual physiological role of plant PHGPXs.

Acknowledgements

S. H. is a recipient of a french pre-doctoral fellowship (Ministère de la Recherche et de l'Enseignment Supérieur). We thank G. Périot for technical assistance and Dr E. Maréchal (Laboratoire de Physiologie Cellulaire Végétale, CEA, Grenoble, France) for the gift of the pET15b vector.

Ancillary