The upstream oxylipin profile of Arabidopsis thaliana: a tool to scan for oxidative stresses


  • Jean-Luc Montillet,

    1. CEA Cadarache, DSV-DEVM, Laboratoire de Radiobiologie Végétale, F-13108 Saint-Paul Lez Durance Cedex, France
    Search for more papers by this author
  • Jean-Luc Cacas,

    1. CEA Cadarache, DSV-DEVM, Laboratoire de Radiobiologie Végétale, F-13108 Saint-Paul Lez Durance Cedex, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Lionel Garnier,

    1. CEA Cadarache, DSV-DEVM, Laboratoire de Radiobiologie Végétale, F-13108 Saint-Paul Lez Durance Cedex, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Marie-Hélène Montané,

    1. CEA Cadarache, DSV-DEVM, Laboratoire de Radiobiologie Végétale, F-13108 Saint-Paul Lez Durance Cedex, France
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Thierry Douki,

    1. CEA Grenoble, DRFMC-SCIB, Laboratoire des Lésions des Acides Nucléiques, F-38054 Grenoble Cedex 9, France
    Search for more papers by this author
  • Jean-Jacques Bessoule,

    1. Laboratoire de Biogénèse membranaire, CNRS, UMR 5544, Université Victor Ségalan Bordeaux II, 146 rue Léo Saignat, 33076 Bordeaux Cedex, France
    Search for more papers by this author
  • Lidia Polkowska-Kowalczyk,

    1. Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Pawinskigo 5a str., 02-106 Warsaw, Poland
    Search for more papers by this author
  • Urszula Maciejewska,

    1. Institute of Biochemistry and Biophysics, Polish Academy of Sciences, Pawinskigo 5a str., 02-106 Warsaw, Poland
    Search for more papers by this author
  • Jean-Pierre Agnel,

    1. CEA Cadarache, DSV-DEVM, Laboratoire de Radiobiologie Végétale, F-13108 Saint-Paul Lez Durance Cedex, France
    Search for more papers by this author
  • Alexandra Vial,

    1. CEA Cadarache, DSV-DEVM, Laboratoire de Radiobiologie Végétale, F-13108 Saint-Paul Lez Durance Cedex, France
    Search for more papers by this author
  • Christian Triantaphylidès

    Corresponding author
    1. CEA Cadarache, DSV-DEVM, Laboratoire de Radiobiologie Végétale, F-13108 Saint-Paul Lez Durance Cedex, France
    Search for more papers by this author

(fax +33 4 42 25 46 56; e-mail


Various physiological imbalances lead to reactive oxygen species (ROS) overproduction and/or increases in lipoxygenase (LOX) activities, both events ending in lipid peroxidation of polyunsaturated fatty acids (PUFAs). Besides the quantification of such a process, the development of tools is necessary in order to allow the identification of the primary cause of its development and localization. A biochemical method assessing 9 LOX, 13 LOX and ROS-mediated peroxidation of membrane-bound and free PUFAs has been improved. The assay is based on the analysis of hydroxy fatty acids derived from PUFA hydroperoxides by both the straight and chiral phase high-performance liquid chromatography. Besides the upstream products of peroxidation of the 18:2 and 18:3 PUFAs, products coming from the 16:3 were characterized and their steady-state level quantified. Moreover, the observation that the relative amounts of the ROS-mediated peroxidation isomers of 18:3 were constant in leaves allowed us to circumvent the chiral analyses for the discrimination and quantification of 9 LOX, 13 LOX and ROS-mediated processes in routine experiments. The methodology has been successfully applied to decipher lipid peroxidation in Arabidopsis leaves submitted to biotic and abiotic stresses. We provide evidence of the relative timing of enzymatic and non-enzymatic lipid peroxidation processes. The 13 LOX pathway is activated early whatever the nature of the stress, leading to the peroxidation of chloroplast lipids. Under cadmium stress, the 9 LOX pathway added to the 13 LOX one. ROS-mediated peroxidation was mainly driven by light and always appeared as a late process.


The generic term of lipid peroxidation actually covers numerous distinct processes of fatty acid deoxygenation which generally occur in polyunsaturated fatty acids (PUFAs). In plant tissues, the most abundant PUFAs, linoleic (18:2) and linolenic (18:3) acids, are mainly found in galactolipids of the plastid membranes. In addition to this, in so-called ‘16:3 plants’ hexadecatrienoic acid exclusively present in chloroplast membranes is also susceptible to such oxidation (Joyard et al., 1998; Ohlrogge and Browse, 1995). Besides that fatty acids can be present either as free compounds released by lipid-hydrolysing enzymes from membranes or tightly packed in the latter as esters of glycerol, their oxidation can originate from the action of reactive oxygen species (ROS) which include oxygen-centred radicals (OCR) and singlet oxygen (1O2) or catalysed by enzymes such as lipoxygenases (LOXs) or α-dioxygenases (αDOXs), (Feussner and Wasternack, 2002; Gutteridge and Halliwell, 1990; Hamberg et al., 2003).

Two distinct types of reactions can contribute to ROS-mediated lipid peroxidation (Figure 1a). In the type I reaction, OCR can initiate hydrogen abstraction from the methylene of the pentadienyl group leading after isomerization and reaction with oxygen to conjugated diene PUFA hydroperoxides (Stratton and Liebler, 1997; see Figure 1a). OCR-mediated lipid peroxidation in cells is a membrane process. The molecular closeness of PUFAs favours the propagation of a radical chain reaction which can lead to disorganization of large membrane areas. The formation of the different hydroperoxide isomers which may arise from 18:2, 18:3 and 16:3 PUFAs is described in Figure 1(b). In the type II reaction, 1O2 generated during photo-oxidative processes adds to a double bond and leads, after rearrangement, to the production of both conjugated and non-conjugated diene PUFA hydroperoxydes (Stratton and Liebler, 1997, see Figure 1a). The formation of all different isomers depicted in Figure 1(b) is then to be expected. In this case, no propagation can be observed as far as no radical intermediate is generated. Besides these compounds, the reactivity of 1O2 can give double and even multiple dioxygenations leading to the formation of multi hydroperoxides and endoperoxides (Frankel and Neff, 1983). Indeed, ROS-mediated multiple peroxidation of PUFAs is at the origin of isoprostanes in plants (Thoma et al., 2003).

Figure 1.

Lipid peroxidation by reactive oxygen species (ROS) and by lipoxygenases (LOX) of the main polyunsaturated fatty acids (PUFAs) in plants.
(a) Mechanisms of lipid peroxidation at the pentadienyl group of PUFAs; oxygen-centred radicals (OCR) and LOXs give type I oxidative reactions; hydrogen abstraction occurs at the methylene of the pentadienyl group and leads to conjugated diene PUFA hydroperoxides; in the enzymatic process, the reaction is regioselective (see b) and can be enantioselective; in the type II reaction singlet oxygen (1O2) adds to a double bond leading to the production of both conjugated and non-conjugated diene PUFA hydroperoxides after rearrangement (Stratton and Liebler, 1997).
(b) Structures of the main PUFAs and positions of peroxidation; the position of peroxidation of the conjugated diene products analysed in the present work are mentioned by an arrow; the so-called 9 LOXs and 13 LOXs are acting at the corresponding positions of the 18:2 and 18:3 PUFAs; they were proposed to peroxidize the 16:3 at position 7 and 11, respectively, and such statement was assessed by mass spectrometry in this work (see Figure 2). Our results suggest that hydrogen abstraction by 9 LOXs and 13 LOXs occurs at the ω-7 position; the action of ROS (OCR and 1O2) leads to the accumulation of the same compounds and also to the conjugated compounds at the encircled positions via type I reactions, by hydrogen abstraction at the ω-4 position, or type II reactions; these latter compounds can be considered as specific of lipid peroxidation mediated by ROS (Berger et al., 2001; Rustérucci et al., 1999); in addition to this, the positions of peroxidation indicated by an arrow head (non-conjugated products) could be considered as being specific of 1O2 action.

Apart from a few exceptions (Feussner et al., 2001; Jung et al., 1985; Kuhn et al., 1990; Murray and Brash, 1988; Schewe et al., 1975), the peroxidation by LOX occurs on free PUFAs, released from membranes through the action of lipid-hydrolysing enzymes and according to the type I reaction. Moreover, the enzyme-catalysed lipid peroxidation is, in most cases, regioselective and enantiospecific (Feussner and Wasternack, 2002; see Figure 1b). Indeed, several isomers, such as the 12- and 16-hydroperoxides of 18:3 (Figure 1b), have never been observed as LOX products and are considered as ROS markers (Berger et al., 2001; Rustérucci et al., 1999). On the contrary, the abundance of the 9- or 13- stereoisomers (Figure 1b) can be considered to be a good enzymatic signature that nevertheless has yet to be confirmed by analysis of the enantiomeric ratio of the most abundant isomer (Berger et al., 2001; Jalloul et al., 2002; Rustérucci et al., 1999). The LOX-mediated reaction is one of the starting points to the important oxylipin metabolic pathway, being at the origin of many cell constituents and signalling molecules (Blee and Joyard, 1996; Feussner and Wasternack, 2002). For instance, jasmonate is involved in the signalling of both the wounding response and the defence against specific pathogens (Farmer et al., 1998). Besides LOXs, another group of fatty acid dioxygenases, the α-DOXs, produce non-conjugated fatty acid hydroperoxides by stereospecific oxygenation of the α-carbon of the PUFA chain providing a new class of oxylipins derived from the unstable 2(R)-hydroperoxy fatty acids. The latter metabolites play a role in the protection against cell death (Hamberg et al., 2003).

Both the physiological and environmental situations can influence the intensity and the nature of the lipid peroxidation process and under certain circumstances several types of peroxidation processes may coexist. An in-depth analysis of these phenomena is of utmost importance in order to understand their origin and to give an accurate picture of the oxidative stress nature which operates under a specific physiological situation. Numerous methods dedicated to the assessment of lipid peroxidation are available (Gutteridge and Halliwell, 1990) and include measurement of upstream compounds such as carbon- and OCR by electron spin resonance or primary peroxidation products. It is also possible to quantify secondary products resulting from the decomposition of lipid hydroperoxides and endoperoxides, either physically through spontaneous and thermal-induced photon emission (Havaux, 2003), or chemically, by using, for instance, the popular thiobarbituric acid-reactive substance (TBARS) assay (see for critical analysis, Hodges et al., 1999). Besides such global descriptions, additional information is necessary to discriminate between ROS-mediated processes and enzymatic reactions in order to provide further insights into the involved cellular mechanisms.

Several years ago we developed a simple method based on two principles: (i) Our first principle was applied at the extraction step. The complete chemical stabilization of fatty acid hydroperoxides into the corresponding alcohols was performed by reduction using sodium borohydride. (ii) The addition of a saponification step in order to evaluate free and esterified fatty acid peroxides together constituted our second principle (Degousée et al., 1995). Grinding of tissues in alkali solution of NaBH4 avoids endogenous transition metals (Fe2+ or Cu+) to degrade peroxides and enzymes to metabolize lipid hydroperoxides into various oxylipins. In these conditions, quantitative and qualitative information can be obtained. In addition to this, as the generated hydroxy fatty acids (HFAs) have the same stereo-chemical properties (positional, cistrans isomerisms and enantiomeric ratio) as the corresponding PUFA hydroperoxides, it is possible to go back to the mechanisms responsible for their formation. Consequently, the procedure allows us to evaluate the very early steps of the peroxidation process as a whole, one taking place in membranes (esterified lipids) and the other concerning free PUFAs. In the end, we developed a simple chromatographic method with which the different positional isomers of HFAs are separated by a high-performance liquid chromatography (HPLC) on a straight phase silica column. HFAs having conjugated dienes can then be specifically detected in a crude lipid extract and quantified by UV at 234 nm. with a fairly good detection limit of 0.1 nmol g−1 FW (Degousée et al., 1995).

In the present paper we first supplemented our upstream oxylipin profile by characterizing the hydroperoxides of 16:3, a PUFA of ‘prokaryotic’ origin and exclusively present in the chloroplastic membrane lipids of many plants including tobacco and Arabidopsis thaliana (Mongrand et al., 1998; Ohlrogge and Browse, 1995). Secondly, we investigated the lipid peroxidation process in depth in response to various biotic and abiotic stresses. The hypersensitive response (HR) in Arabidopsis has been chosen as a LOX-dependent model (Montillet et al., 2002), whereas ROS-mediated peroxidation has been sought in plants submitted to light stress under drastic photorespiratory conditions. Finally, as another condition leading to lipid peroxidation (Schüzendübel and Polle, 2002), cadmium toxicity has been studied as far as no data were available depicting the mechanisms involved in this process. During the course of our investigations we noticed that the ratio of the stereoisomers issued from the ROS-mediated peroxidation of 18:3 PUFA was fairly constant. Taking into account this observation, a simple method was developed discriminating between 13 LOX, 9 LOX and ROS-mediated lipid peroxidation of 18:2, 18:3 and 16:3 PUFAs.


Identification of hydroperoxide isomers of the trienoic fatty acids by mass spectrometry

Hexadecatrienoic acid (16:3) is a common PUFA found at high levels in the so-called ‘16:3-plants’ such as tobacco and A. thaliana where it represents about 10 and 14% of total fatty acids respectively (Mongrand et al., 1998; Rustérucci et al., 1999). In order to supplement the upstream peroxidation profile obtained with our ‘reduction saponification’ procedure, we first characterized HFA isomers of the latter PUFA. A mixture of PUFAs was prepared from a parsley leaf lipid extract (18:2/18:3/16:3 composition 7/56/37) and peroxidized either by using the commercial soya bean (13S) LOX or by an extract of wild potato leaf exhibiting a high (9S) LOX activity (1.7 nkat mg−1 of protein with a 9 specificity of 98% and an S enantioselectivity of 98%). The produced hydroperoxides were then reduced into the corresponding HFAs and separated by HPLC (Figure 2a). For both LOXs, quantification showed product distributions close to the initial PUFA composition, suggesting that the enzymes were not substrate specific. Based on regiospecificity of each LOX on both 18:2 and 18:3, the expected structures of the 16:3 products after reduction might be 11-HHTE for soya bean (13S) LOX and 7-HHTE for potato (9S) LOX (Figure 1b). In order to provide a definitive assignment to these structures, chromatographic peaks were collected and analysed by negative-ion electrospray tandem mass spectrometry. This powerful technique has allowed all the hydroxy stereo-isomers of eicosatetraenoic acid to be characterized(Murphy et al., 2001). Indeed, mass fragmentations occur on the carbon chain at the level of the hydroxy group providing unambiguous information about the isomer position. The mass fragmentations of the plant trienoic HFAs have been investigated in the present work. The mass spectra of the putative 11-HHTE (Figure 2c) and of 13-HOTE (Figure 2e) yielded the same fragmentations, with respect to their chain length (16:3 and 18:3 respectively), definitively assigning the structure of the former compound to 11-HHTE. The same comparison between the fragmentation spectrum of the putative 7-HHTE (Figure 2b) and 9-HOTE (Figure 2d) also confirmed the proposed isomer assignment. Chiral phase HPLC analysis of both 7-HHTE and 11-HHTE was also carried out. In the case of 7-HHTE, the two enantiomers could not be separated under our chromatographic conditions. On the contrary, the two enantiomers of 11-HHTE were separated (Figure 3b) and the (R)/(S) assignment (3/97) was attributed with reference to the (S) selectivity of soya bean LOX in the conversion of both 18:2 and 18:3. In addition to this, as the mass spectra of 12- and 16-HOTE, the ROS-mediated peroxidation markers of 18:3 had not yet been described, so we have prepared these compounds by action of tert-BuOOH plus Fe2+ on 18:3 as described previously (Degousée et al., 1995). After straight phase HPLC separation and the collection of HFA isomers, the fragmentation spectra obtained (Figure 2f,g) allowed us to assign the structure of the first eluting chromatographic peak to 12-HOTE and that of the second to 16-HOTE.

Figure 2.

Characterization of the products of lipoxygenase action on hexadecatrienoic acid.
(a) HPLC traces of hydroxy fatty acids (HFAs) coming from the products of soya bean (13S) LOX (lower trace) and potato (9S) LOX (upper trace) acting on a mixture of 18:2, 18:3 and 16:3 polyunsaturated fatty acids (PUFAs) (relative composition, 7/56/37 respectively). (b–g) Collision-induced fragmentation of [M-H] carboxylate anions derived from HFAs; mass spectrum of: 7-HHTE (b); 11-HHTE (c); 9-HOTE (d); 13-HOTE (e); 16-HOTE (f); 12-HOTE (g); the structure of each compound is given and the fragmentations provided according to the described decomposition of isomeric hydroxy eicosatetraenoic acid (Murphy et al., 2001).
The abbreviations for hydroxy PUFAs (HFAs) used throughout this work are: 15-HEDE, 15-hydroxy-11,13(Z,E) eicosadienoic acid, as internal reference; 7-HHTE, 7-hydroxy-8,10,13,(E,Z,Z) hexadecatrienoic acid; 11-HHTE, 11-hydroxy-7,9,13(Z,E,Z)-hexadecatrienoic acid; 9-HODE, 9-hydroxy-10,12(Z,E) octadecadienoic acid; t-9-HODE, 9-hydroxy-10,12(E,E) octadecadienoic acid; 13-HODE, 13-hydroxy-9,11(Z,E) octadecadienoic acid; t-13-HODE, 13-hydroxy-9,11(E,E) octadecadienoic acid; 9-HOTE, 9-hydroxy-10,12,15(E,Z,Z) octadecatrienoic acid; 12-HOTE, 12-hydroxy-9,13,15,(Z,E,Z)-octadecatrienoic acid; 13-HOTE, 13-hydroxy-9,11,15(Z,E,Z)-octadecatrienoic acid; 16-HOTE, 16-hydroxy-9,12,14,(Z,Z,E)-octadecatrienoic acid.

Figure 3.

Lipid peroxidation analysis in Arabidopsis leaves extracted with harpin.
(a) Compared HPLC chromatograms of hydroxy fatty acids from extracts obtained according to the ‘reduction-saponification’ procedure of Arabidopsis leaves infiltrated with water (lower trace) or with harpin (upper trace) and incubated for 46 h; in both cases the internal reference (15-HEDE) was at 40 nmol g−1 FW.
(b) Chiral HPLC of authentic samples of 16-HOTE (lower trace) and 11-HHTE (medium trace) and of the 16-HOTE +11-HHTE chromatographic peak (upper trace) collected from the upper trace described in (a); these results show that 11-HHTE is the main component from the two which accumulated upon elicitation. See Figure 2 legend for abbreviations.

Evidence for (13S) LOX-mediated lipid peroxidation of 18:2, 18:3 and 16:3 PUFAs during the hypersensitive reaction in Arabidopsis

Preliminary experiments demonstrated the accumulation of (13S) LOX metabolites of 18:2 and 18:3 in leaves of Arabidopsis plants challenged with avirulent pathogens or elicited with harpin (Montillet et al., 2002). The latter situation has been further investigated in the present work with the aim of completing the characterization of all the products of lipid peroxidation thereby providing additional insights into the induced oxylipin pathway. Necrotic symptoms started 1 day after elicitation and were fully developed after 2 days. Typical chromatograms of the HFAs obtained from mock- (MgCl2, 10 mm) and harpin- (10 μg ml−1) infiltrated leaves, 46 h after treatment, are shown in Figure 3(a). Increased levels of 13-HODE and 13-HOTE were observed in the elicited material, confirming the involvement of a 13 LOX in the process. Yet, compared with the level of 12-HOTE, a third chromatographic peak increased markedly, eluting at a retention time corresponding to 16-HOTE and to 11-HHTE. As 12- and 16-HOTE, both markers of ROS-mediated lipid peroxidation, were generally produced in biological samples at similar levels, this observation suggested that the chromatographic peak was mainly due to the presence of 11-HHTE. This hypothesis was checked and confirmed by chiral phase HPLC analysis of the chromatographic peak allowing the separation of the enantiomers of both 16-HOTE and 11-HHTE (Figure 3b). Whenever necessary (see below), the quantification of (16 R)-HOTE helped us to estimate the quantity of (16 S)-HOTE which contaminated the (11 R)-HHTE. The quantity of the latter compound was deduced by subtraction allowing us to properly estimate the proportion of 16-HOTE and 11-HHTE. Additionally, 13-HODE, 13-HOTE and 11-HHTE were all chiral (Table 1) demonstrating the involvement of a (13S) LOX activity in the process.

Table 1.  Enantioselectivity in lipid peroxidation induced by harpin in Arabidopsis leaves
(S) enantiomer (%)
Hydroxy PUFAsControlHarpin 46 h
  1. (S) enantiomer composition of the HFAs obtained by the ‘reduction-saponification’ procedure from Landsberg leaves infiltrated with harpin after 46 h of incubation and from control leaves. As a product of ROS-mediated lipid peroxidation, 16-HOTE (*) is supposed to be racemic; the enantiomer composition of 11-HHTE was then calculated by subtracting the contribution of (16 S)-HOTE (see Figure 3b). Results are expressed as mean and SD of two independent analyses. n.d., not determined.

9-HODE50 ± 150 ± 1
9-HOTE50 ± 552 ± 2
13-HODE81 ± 268 ± 2
13-HOTE80 ± 688 ± 1
11-HHTEn.d.91 ± 3
12-HOTEn.d.51 ± 3

LOX- versus ROS-mediated processes in response to strong photo-oxidative conditions

Acute photo-oxidative stresses are known to lead to ROS overproduction in higher plants (Asada, 1999) and the high light effect can be exaggerated by lowering the CO2 concentration in order to favour the photorespiratory conditions. These conditions may lead to ROS-mediated lipid peroxidation and ‘chaos’, a photo-oxidative stress-sensitive mutant (Hutin et al., 2003) and the corresponding wild type (WT) was chosen to investigate this process in Arabidopsis leaves. After 1 day under high light conditions at a normal CO2 concentration (360 p.p.m.), only the older leaves of both WT and ‘chaos’ mutant plants showed partial collapse of the limb by desiccation, starting around 24 h (not shown), and preceding bleaching which was more intense in ‘chaos’. With the increase in the duration of stress, the fully developed leaves of the WT plants were more resistant to light stress than the ‘chaos’ ones showing a typical intense production of anthocyanins (not shown) whereas the chaos plants hardly accumulated such light-protective compounds. Under a low concentration of CO2 (30 p.p.m.), the leaf symptoms of both WT and mutant plants enlarged and bleaching spread on the leaves, irrespective of their age. They were partly bleached and desiccated within 2 days and the process was more severe for the ‘chaos’ mutant (not shown). The effect of high light on lipid peroxidation levels of ‘chaos’ plants under both CO2 regimes is described by the typical HPLC chromatograms of Figure 4(a) which clearly illustrate the drastic peroxidation changes. Under 360 p.p.m. CO2, the predominance of 13 LOX metabolites is obvious (mainly 13-HOTE and 11-HHTE) whereas under low CO2 the level of peroxides is not only higher but the isomer distribution is also randomized, suggesting an intense ROS-mediated process. Kinetics of lipid peroxidation was further established (see Figure 4b) showing that under normal CO2 conditions, the 13 LOX metabolism was transiently and almost exclusively triggered. At a low CO2 concentration, although a 13 LOX was still operating, the peroxidation was enlarged by the presence, starting from 30 h of stress, of all the other isomers and especially 12-HOTE, a typical marker of ROS effects.

Figure 4.

Lipid peroxidation analysis in Arabidopsis leaves in response to photo-oxidative stress.
Arabidopsis plants were transferred to high light either at 360 p.p.m. CO2 (high CO2) or 30 p.p.m. CO2 (low CO2) for 30, 52 or 76 h.
(a) Typical straight phase HPLC chromatograms from ‘chaos’ plant leaf extracts after 52 h of high light either under 360 p.p.m. (lower trace) or 30 p.p.m. CO2 (upper trace) showing strong accumulation ROS-mediated lipid peroxidation products at 30 p.p.m. CO2.
(b) HFA distribution by HPLC analyses for the ‘chaos’ mutant; for each time point, the order of description of the HFA isomer level follows the retention time of the products in the straight phase HPLC analysis.
(c) Distribution of the HOTE isomers, after 52 h of stress under the indicated conditions and taking into account the chiral analysis from Table 2; the amount of 16-HOTE in the 16-HOTE + 11HHTE straight phase HPLC chromatographic peak was calculated from the chiral analyses, as in Figure 3(b); 9-HOTE was racemic and was then considered exclusively from ROS-dependent origin like 16- and 12-HOTE; results allowed to calculate the respective HOTE isomer ratio of the ROS-mediated process reported in Table 3.
(d) Time course of HOTE levels of (13S) LOX- versus ROS-mediated processes in WT and ‘chaos’ mutant; (13S) LOX-dependent lipid peroxidation was evaluated from the level of (13S)-HOTE, whereas ROS-mediated lipid peroxidation was established by adding the contribution of (13 R/S)-HOTE, 9-, 12- and 16-HOTE. The calculation of these levels starts from the experimental levels of 12-HOTE in individual samples to which were added the levels of the other ROS-isomers as predicted by the previously established ROS-isomer distribution (see Table 3). See Figure 2 legend for abbreviations. Mean and SD from two different analyses.

The analysis of HFA enantiomer distribution of the samples after 52 h of stress (Table 2) confirmed the results of Figure 4(b). As a matter of fact, 13-HODE, 13-HOTE and 11-HHTE were chiral indicating a (13S) LOX metabolism but they displayed a decreased chirality under stressful conditions (see the ‘chaos’ mutant under low CO2). At the opposite end, all the other isomers (12-, 16-, 9-HOTE and 9-HODE) were racemic showing a ROS-mediated origin. In an attempt to quantify both processes in one plant extract, we combined the data of the isomer and enantiomer distributions for the four isomers of HOTE, taken as reference compounds representing LOX versus ROS–mediated lipid peroxidation (see Figure 1). First, the level of 16-HOTE can be calculated from the chiral analyses, as described above. Secondly, assuming that a fully selective (13S) LOX is operating in Arabidopsis leaves (Table 2), the level of racemic 13-HOTE (50/50, R/S ratio), which would only result from a ROS-mediated lipid peroxidation can be calculated. The results reported in Figure 4(c) clearly show that the ratio of the four HOTE isomers is fairly constant within the experimental error (Table 3), not dependent on the stress condition, confirming that ROS are responsible for the process and suggesting that a 9 LOX is not operating under the present conditions. This makes it possible to discriminate and quantify both (13S) LOX- and ROS-mediated contributions in the lipid peroxidation of 18:3. As in Arabidopsis, the relative proportions of HOTE isomers of ROS-dependent peroxidation obtained from leaves from catalase antisense tobacco plants submitted to high light or with potato leaves (Table 3) were close to each other and constant, regardless of stress. The results show that the measurement of one of the exclusively ROS-representative HOTE (12- or 16-HOTE) enables us to quantify the amount of all four ROS-dependent HOTE isomers in a leaf extract.

Table 2.  Enantioselectivity in lipid peroxidation induced by photo-oxidative stress in Arabidopsis leaves
(S) enantiomer (%)
Hydroxy PUFAsWT 360 p.p.m. CO2‘Chaos’ 360 p.p.m. CO2WT 30 p.p.m. CO2‘Chaos’ 30 p.p.m. CO2
  1. (S) enantiomer composition of the HFAs obtained by the ‘reduction-saponification’ procedure from WT Landsberg and ‘chaos’ mutant plants transferred to high light (see Figure 4), under 360 or 30 p.p.m. CO2 for 52 h. The enantiomer composition of 11-HHTE was calculated as described in Table 1 assuming that (*) 16-HOTE is racemic. Results are expressed as mean and SD of two independent analyses.

9-HODE51 ± 150 ± 147 ± 249 ± 1
9-HOTE50 ± 150 ± 148 ± 148 ± 2
13-HODE81 ± 280 ± 563 ± 250 ± 1
13-HOTE94 ± 195 ± 183 ± 263 ± 9
11-HHTE92 ± 293 ± 488 ± 185 ± 6
12-HOTE52 ± 353 ± 253 ± 152 ± 3
Table 3.  Distribution of the products of ROS-mediated peroxidation of linolenic acid (18:3) in leaves of different plant species
Plant species (no. experiments)Isomer distribution (%)
  1. In Arabidopsis, the distribution of the HOTE isomers was calculated from control and stressed WT Landsberg and ‘chaos’ mutant plants taking into account straight and chiral phase analyses of the HFAs and assuming that: (i) 12- and 16-HOTE are racemic products specific to ROS-mediated peroxidation; (ii) the (13S) LOX is fully enantioselective and a 9-LOX is not operating. A similar approach was developed for unstressed leaves of wild potato (Solanum nigrum). In the case of tobacco, the results were obtained from the analysis of lipid peroxidation in bleached leaves of catalase 1 antisense plants transferred to high light conditions leading to an increase in photorespiratory H2O2 production (J.L. Montillet et al., in preparation). Results are mean and SD from n individual experiments.

A. thaliana (n = 8)32.2 ± 3.719.2 ± 2.222.8 ± 2.825.8 ± 4.8
S. nigrum (n = 5)23.0 ± 4.818.8 ± 0.930.3 ± 1.627.9 ± 2.8
N. tabacum (n = 9)21.4 ± 0.920.1 ± 1.830.8 ± 1.027.7 ± 2.6

Using this result, we calculated the respective contributions of (13S) LOX and ROS in peroxidation of 18:3 in Arabidopsis for all our kinetic data. The respective contribution of LOX- versus ROS-mediated peroxidation in samples is illustrated in Figure 4(d). First, a transient increase in (13S) LOX metabolism occurs under stress conditions. This observation was corroborated by the transient accumulation of 11-HHTE (not shown). Secondly, under normal CO2 conditions, ROS-mediated lipid peroxidation remained weak and constant in the WT leaves whereas it slightly increased with time in the ‘chaos’ mutant. Finally, under low CO2 conditions, the ROS-mediated process was, as expected, greater in both genetic backgrounds and increased with time and was considerably higher in the photo-oxidative stress-sensitive ‘chaos’ mutant.

Dissection of lipid peroxidation processes in response to cadmium: evidence for the induction of both 13 and 9 LOX metabolisms

Experimental evidence suggests that heavy metal toxicity in plants is at least partly due to the development of an oxidative stress (Schüzendübel and Polle, 2002). We investigated cadmium-induced leaf necrosis. Plants were submitted to a transient and acute intoxication by cadmium, then incubated either in the dark or under the usual day/night condition. In both cases, necrotic areas appeared on leaves within 1 day and were fully expanded within 2 days but developed earlier and were more severe when plants were kept in the light. Lipid peroxidation was evaluated as previously and the results are shown in Figure 5. A typical chromatogram of an extract from stressed leaves incubated 48 h in the dark is shown in Figure 5(a). It should be pointed out that, in contrast to 13- and 9-HOTE, the level of the ROS marker 12-HOTE is low, strongly suggesting that both 13 LOX and 9 LOX metabolisms are in operation. This hypothesis was corroborated by the presence of both 11- and 7-HHTE (Figure 5a) and by the results of the chiral analysis (Table 4) showing that the chirality of the 13-isomers remained high whereas the chirality of the 9-ones only slightly increased in response to the Cd treatment (58% of S enantiomer). These results suggested the involvement of a (9S) LOX yet displaying a low enantioselectivity. Furthermore, in the dark, the kinetics (Figure 5b) showed that the accumulation of 13-isomers occurred within 24 h and before the accumulation of the 9-ones. With the help of the previously established HOTE isomer distribution for the ROS-dependent 18:3 peroxidation (Table 3), we were able to unravel the respective contributions of ROS 13 and 9 LOX. The data shown in Figure 5(c) clearly indicate that in the dark, the ROS are weakly involved in contrast to both the 13 LOX, whose participation is as early as 24 h, and the 9 LOX that is involved afterwards. In the light, the peroxidation process was about twice as high than that observed in plants kept in the dark (Figure 5b). But the most striking difference was that ROS contribution continued to increase with time whereas LOX was transient (Figure 5c). The role of the two LOXs was equivalent at whatever point in time remaining major at 24 h (around 60% of total peroxidation) and becoming weak at 48 h (14%). This is consistent with the chiral experiments (Table 4), assuming (i) the involvement of a fully enantioselective (13S) LOX and (ii) the involvement of a (9S) LOX with a low enantioselectivity.

Figure 5.

Lipid peroxidation analysis in Arabidopsis leaves in response to cadmium stress.
Arabidopsis plants were submitted to a transient intoxication by cadmium and were then incubated either in the dark (‘Dark’) or under the usual day/night condition (‘Light’).
(a) Typical chromatograms from control (lower trace) or Cd-treated plants (upper trace) after 48 h in the dark depicting product accumulation of both 13 LOX and 9 LOX metabolism in response to Cd.
(b) Results of lipid peroxidation profiling at 24 and 48 h in control and Cd-treated leaves, either in the ‘Dark’ or in the ‘Light’ condition.
(c) Changes with time in the HOTE levels related to the (13S) LOX-, 9 LOX- and to the ROS-mediated processes in control and Cd-treated plants in the ‘Dark’ and in the ‘Light’ condition; the calculation of the levels was carried out as described in Figure 4 and in addition to this, 9 LOX-dependent lipid peroxidation was evaluated as the difference between the global level of 9-HOTE and the level of ROS-mediated production of 9-HOTE. See Figure 2 legend for abbreviations. Mean and SD from three analyses.

Table 4.  Enantioselectivity in lipid peroxidation induced by Cd stress in Arabidopsis leaves
Hydroxy PUFAs(S) enantiomer (%)
  1. (S) enantiomer composition of the HFAs obtained by the ‘reduction-saponification’ procedure from leaves of plants (ecotype Ws) treated with Cd after 48 h of incubation and from control leaves (see Figure 5 for conditions). The enantiomer composition of 11-HHTE is calculated as described in Table 1 assuming that (*)16-HOTE is racemic. Results are expressed as mean and SD of two independent analyses. n.d., not determined.

9-HODE47 ± 358 ± 744 ± 547 ± 1
9-HOTE47 ± 258 ± 242 ± 145 ± 2
13-HODE71 ± 374 ± 183 ± 650 ± 1
13-HOTE88 ± 189 ± 184 ± 766 ± 1
11-HHTE85 ± 294 ± 2n.d.n.d.
12-HOTE50 ± 251 ± 147 ± 251 ± 1


The proposed upstream oxylipin profile is based on the ‘reduction-saponification’ procedure leading to the extraction and analysis of HFAs by combining both straight and chiral phase HPLC analysis. Total lipid extraction was carried out according to Blight and Dyer (1959) and saponification led to the characterization of lipid peroxidation products from both free and membrane-bound fatty acids. Finally, UV detection enabled us to carry out the specific analysis of conjugated diene HFAs which may result from the action of LOXs or ROS. However, the method cannot assess the formation of PUFA endoperoxides yet, non-conjugated hydroperoxides produced by the action of 1O2 (Figure 1a) and of α hydroperoxides produced by α DOXs. Another point is the impossibility of discriminating PUFA hydroperoxides from HFAs or keto PUFAs (Vollenweider et al., 2000), which might be present in plant tissues. Nevertheless, as these two latter compounds are direct products of the PUFA hydroperoxide metabolism, the levels of HFAs that we measured can be considered to be a good estimation of an overall steady-state level of the upstream metabolites of the peroxidation pathway. The HFA levels described in this work are coherent with the levels of hydroperoxides quantified by a colorimetric test (20–300 nmol g−1 FW; see Griffiths et al., 2000). Thus, the present methodology which is easy to perform and highly effective in both the qualitative and quantitative description of lipid peroxidation, allows us to discriminate 9 LOX and 13 LOX metabolisms from ROS-mediated peroxidation of membranes and to reflect the early steps of the respective processes.

In this paper, we provide the characterization of products derived from enzymatic peroxidation of hexadecatrienoic acid (16:3) in vitro and in planta, as well. As assessed by mass spectrometry, the 13 LOXs act at position 11 of 16:3, whereas 9 LOXs at position 7. Considering that starting from the ω position, the structures of 18:3 and 16:3 PUFAs are identical, the enzymatic peroxidation with a given LOX occurs at the same position indicating that the pentadienyl group centred at ω-7 position is always involved in the reaction (see Figure 1b). Thus, as has been previously noticed (Berger et al., 2001; Rustérucci et al., 1999), the products arising from the hydrogen abstraction of the other pentadienyl group (centred at position ω-4) can be considered to be markers of ROS-mediated lipid peroxidation. Although the corresponding compounds coming from 16:3, that is 10-HHTE and 14-HHTE could theoretically be proposed for this use, in practice the 16:3 level turned out too low to allow us to characterize these compounds in the lipid extracts. We have used the 18:3 equivalent stereoisomers, 12- and 16-HOTE such as the ROS markers. As the HOTE isomer distribution due to ROS action was constant in planta (Table 3), it was possible to estimate the ROS-mediated process by measuring the 12-HOTE level and then to deduce the respective contributions of 9 and 13 LOX from the levels of 9- and 13-HOTE. Therefore, in routine experiments, a simple straight phase HPLC analysis of a lipid extract is sufficient to discern the differences among the 9 LOX, 13 LOX and ROS-mediated processes without carrying out complementary chiral analyses (see our results on Cd stress).

Previous work carried out on infected or elicited leaves of A. thaliana showed the rapid accumulation of (13S) LOX metabolite of 18:3 in leaves undergoing HR (Montillet et al., 2002). The characterization of the 13 LOX metabolites of 16:3, that is 11-HHTE in extracted leaves is a confirmation of the previous result. Interestingly, as 16:3 is a PUFA specific from chloroplasts, our observations highlight the role of chloroplastic lipids in the production of oxylipins during HR cell death in A. thaliana (Blee and Joyard, 1996). Six genes encoding for LOXs are present in the genome of this plant (Feussner and Wasternack, 2002) and four of them which code for the chloroplast with the required 13 specificity are potential candidates to drive that accumulation. Products of the (13S) LOX pathway are at the origin of jasmonates occurring in the signalling of wounding and pathogen responses (Farmer et al., 1998). Using the fad5 mutant of A. thaliana, a signalling pathway starting from the 16:3 was demonstrated and led to the production of jasmonates via dinor-oxo-phytodienoic acid (dnOPDA) (Weber et al., 1997). In this work we provide evidence of the production of the precursor of dnOPDA in extracted leaves, that is the 11-hydroperoxide of 16:3 postulated by Farmer et al. (1998). It is worth mentioning that ROS-mediated lipid peroxidation appears low during the development of the leaf HR, in comparison with the enzymatic reaction. A similar observation was made on cryptogein-extracted tobacco leaves (Rustérucci et al., 1999) but contrary to A. thaliana, none of the 16:3 metabolites were observed and the major products were the 9-specific metabolites of 18:2 and 18:3. Thus, these two ‘16:3’ plant models revealed two distinct strategies of peroxidation during HR. Several hypothesis are now being explored in order to understand the biological significance of this difference.

First, we investigated the photo-oxidative stress part of abiotic stress. In the ‘chaos’ mutant, the transport of nuclear-encoded proteins to the chloroplast is severely affected, leading to a photo-oxidative stress sensitivity, as previously assessed by the thermoluminescence measurement of lipid peroxidation (Hutin et al., 2003). Under the same conditions, that is high light and low temperature and 360 p.p.m. CO2, we did not observe important ROS-mediated lipid peroxidation by the HPLC assay (not shown). The most striking feature was the transient induction of the 13 LOX metabolism upon transfer of the WT or ‘chaos’ plants to high light. Such a result can be explained if we assume according to Asada (1999), that the fate of excess photon absorption leads, after CO2 assimilation, to energy dissipation via the water-water cycle, then to downregulation of PSII followed by photorespiration. The thermoluminescence signal could be related to lipid endoperoxide formation by singlet oxygen action (Havaux, 2003). Indeed, recent published results showed that in the conditional fluorescent (flu) mutant of Arabidopsis, 1O2 does not lead to the accumulation of ROS-mediated lipid peroxidation products but instead to an increase in the levels of 13 LOX metabolites (op den Camp et al., 2003). In fact, we observed by our HPLC assay a similar increase induced by high light and an intense ROS-mediated lipid peroxidation only when the ‘chaos’ mutant was submitted for 2 days to drastic photorespiratory conditions (low CO2). Under these conditions, lipid peroxidation resulted from H2O2 produced during photorespiration. Taken together, these results show that Arabidopsis plants are particularly resistant to photo-oxidative stress as strong photo-oxidative conditions (high light plus low CO2) are necessary to display a ROS-mediated lipid peroxidation of PUFAs.

We observed that the 13 LOX metabolism is also activated in response to cadmium stress leading to toxic effects. The corresponding products, that is 13-HOD(T)E and also 11-HHTE accumulated early. These results are consistent with the suggestion that jasmonate might be involved in the response of Arabidopsis to heavy metals (Xiang and Oliver, 1998). The accumulation of 9-specific metabolites suggests the involvement of a 9 LOX in the process. Surprisingly, although the LOX metabolites accumulate at high levels in the necrotic leaves in the dark, we did not observe any similar rise of ROS-specific ones, that is 12- and 16-HOTE. Thus, this upstream oxylipin profile strongly suggests that cadmium somehow activates 13 LOX followed by 9 LOX metabolisms in Arabidopsis and does not promote any significant ROS formation. Such a response might be involved in the execution of a programmed cell death. In the light, a ROS-mediated lipid peroxidation was superimposed and prevailed on a long term showing that light-driven production of ROS is taking quantitatively an important place.

Taken together, these results show that the methodology is powerful for the dissection of the early steps of the oxylipin route involving LOXs and ROS. We noticed an active 13 LOX metabolism in leaves of various plant material. In tobacco the main peroxidized PUFAs are 18:2 and 18:3 (Rustérucci et al., 1999). In Arabidopsis 16:3 is also involved in the reaction. The 9 LOX metabolism does not operate initially but can be activated later on by cadmium stress at least. Interestingly, the chiral analyses suggest that in planta the (13S) LOX activity is highly enantioselective (see our results on photo-oxidative stress; Table 2), whereas the (9S) LOX would have a much lower enantioselectivity (Table 4; see also Berger et al., 2001). Finally, in senescent Arabidopsis leaves, previous results showed that the ROS-mediated process progressively gives way to LOX metabolism (Berger et al., 2001). Contrary to previous results, our results show that most of the stress situations leading to leaf tissue necrosis involve a LOX metabolism. Tissue necrosis associated with ROS-mediated lipid peroxidation was observed specifically under drastic photorespiratory conditions or in necrotic leaves of cadmium-treated plants exposed to normal light. The latter results can be considered as the consequence of ROS overproduction.

In this work we have described various situations where 13 and 9 LOX metabolisms are operating and ROS-mediated lipid peroxidation can be observed in Arabidopsis. The chiral phase analysis always conformed with the straight phase HPLC analysis. Thus, we may propose to use such a lipid peroxidation profiling as a simple and valuable assay for any plant species. Last but not least, owing to its high sensitivity, this test can be successfully applied to non-photosynthetic tissues such as roots or cell suspensions in which PUFAs levels are low compared with leaves (unpublished results).

Experimental procedures

Chemicals and enzymes

The HFA chromatographic standards have been previously described (Rustérucci et al., 1999). Samples of 12-HOTE and 16-HOTE were prepared according to a previously described procedure, using tert butyl hydroperoxide in the presence of Fe++ to peroxidize 18:3 PUFA (Degousée et al., 1995), and compared with samples provided by I. Feussner (Göttingen, Germany). Their structure was assessed according to their mass spectra (see Figure 2f,g) and under our chromatographic conditions, 12-HOTE elutes before 16-HOTE. An enriched fraction containing 16:3 (16:3/18:3/18:2 composition 37/56/7) was prepared from parsley leaf lipid extract, by TLC and galactolipid hydrolysis, as described previously (Rustérucci et al., 1999). PUFAs and soya bean (13S) LOX were purchased from Fluka (Sigma-Aldrich Chimie SARL, France). Potato 9 LOX extract was prepared from wild potato leaves treated with an elicitor of Phytophthora infestans (Polkowska-Kowalczyk et al., 2004).

Plant growth and treatments

Arabidopsis thaliana plants were grown for 7 weeks at 130 μmol m−2 sec−1 light radiance, with a 8-h photoperiod, 22/18°C light/dark cycle and 55% relative humidity. Watering was carried out daily and every 3 days with a Coïc-Lesaint solution (Lesaint, 1982). For Cd treatment, plants (ecotype Ws) were grown similarly in sand thereby enabling us to gather them more easily their whole root structure intact. Plants were submitted to a transient and acute Cd intoxication by dipping the root system into a solution of 100 mm CdCl2 for 2 h. The roots were then thoroughly rinsed, transferred in water and the plants were then incubated either in the dark or under the usual day/night conditions. Harpin (10 μg ml−1) was syringe-infiltrated into the leaves of the Landsberg ecotype, as previously described (Montillet et al., 2002). For the photo-oxidative treatments, Landsberg ecotype and ‘chaos’ mutant plants (Hutin et al., 2003) were grown for 5 weeks as described above and then transferred at the beginning of the photoperiod for 3 days into atmosphere-controlled chambers set at either 30 or 360 p.p.m. CO2 at 22°C under high light intensity of 1400 μmol m−2 sec−1 and with a 14-h photoperiod.

Solanum nigrum plants were grown as described elsewhere (Polkowska-Kowalczyk et al., 2004). Nicotiana tabacum cv. Petit Havana (SR1) and catalase anti sense plants (Cat1AS) (Dat et al., 2003) were cultivated under low light (80 μmol m−2 sec−1, 14 h photoperiod) at 25°C and with a 70% relative humidity. Mature pre-flowering plants were transferred for 2 days at high light (350 μmol m−2 sec−1, 14 h photoperiod) and bleached/necrotic leaves were analysed for their lipid peroxidation content.

HPLC assay of conjugated hydroxy PUFA

Lipid peroxidation was assessed by HPLC analysis of HFAs recovered from plant tissue after NaBH4 reduction and saponification of total lipids. Arabidopsis leaves from six to eight plants were mixed and sampled (3 × 0.5–1 g), frozen in liquid N2 and stored at −20°C before extraction. Extraction was carried out according to the previously described procedure (Rustérucci et al., 1999), with the following modifications: the frozen leaves were homogenized at 0–4°C, in 5 ml 0.2 N NaOH, 5% (w/v) NaBH4 in the presence of the internal reference 15-HEDE (40 nmol g−1 FW). After addition of 10 ml of chloroform/methanol (50/50; v/v), the mixture was neutralized at 0–4°C with 12 N HClO4, with particular attention being paid to the H2 bubbling due to the degradation of NaBH4, adjusting pH to 4–4.5. The mixture was centrifuged (700 g × 5 min) and the organic layer withdrawn and recovered. The aqueous phase was extracted again with 5 ml chloroform and the two organic extracts pooled and vacuum evaporated to dryness. The lipid extract was dissolved in 2.5 ml ethanol and after addition of 2.5 ml 3.5 N NaOH, the mixture was submitted to hydrolysis in a water bath at 100°C for 15 min. The solution was then neutralized, at 0–4°C by 3.5 HClO4, to pH 4–4.5, transferred into a 10 ml tube and the balloon rinsed with 2 × 1 ml hexane/ethyl ether (70/30; v/v). The solvent was added to the aqueous phase and after thorough mixing, the organic phase was recovered by centrifugation (700 g × 5 min) and directly used for HPLC analysis.

An aliquot of the extract (50 μl) was submitted to straight phase HPLC (Waters, Millipore, St Quentin-Yvelines, France) using a Zorbax rx-SIL column (4.6 × 250 mm, 5 μm particle size, Hewlett Packard, Les Ullis, France), isocratic elution with 70/30/0.25 (v/v/v) hexane/diethyl ether/acetic acid at a flow rate of 1.5 ml min−1, and UV detection at 234 nm. Quantification was performed with reference to 15-HEDE. For the enantiomer composition analysis, chromatographic peaks were collected from straight phase HPLC and submitted to chiral phase HPLC, using a Chiralcel OD column (250 × 4.6 mm, 5 μm particle size, Diacel Chemical Industries, Interchim, France), with a mobile phase of 95/5/0.1 (v/v/v) hexane/2-propanol/acetic acid at a flow rate of 1 ml min−1 and UV detection at 234 nm.

Mass spectrometry analysis

Mass spectra were recorded on an API 3000 triple quadrupolar apparatus (Perkin-Elmer/Sciex, Thornhill, Canada) equipped with a turbo ionspray electrospray source. Samples were injected in solution in a [1:1] mixture of acetonitrile and 2 mm ammonium formate at a flow rate of 5 μl min−1 by using a Model ‘11’ syringe pump (Harvard Apparatus, Holliston, MA, USA). All analyses were performed on negative ions with the settings of the mass spectrometer optimized by injection of synthetic 13-HOTE. The spectra were first recorded in the single mass spectrometry mode (mass range: 50–550) to determine the mass of the pseudo-molecular ion. The latter ion was then analysed in the product ion scan mode. For this purpose, the pseudo-molecular ion was collected in the first quadrupole and fragmented in the collision cell. The resulting fragments were analysed in the third quadrupole (mass range: 50–300).


L.G. was fully supported by the CEA-programme ‘Toxicologie Nucléaire’. L. P.-K. was the recipient of a CNRS fellowship from the ‘Centre Franco-Polonais de Biotechnologie des Plantes’. We thank Mark Sullivan for reading the manuscript, Laurent Nussaume for ‘chaos’ mutant seeds, Frank Van Breusegem and Dirk Inzé for the use of data obtained from catalase antisense tobacco plants and the GRAP team (CEA Cadarache, DSV-DEVM) for technical support in plant culture and photo-oxidative stress experiments.