Glutathione deficiency of the Arabidopsis mutant pad2-1 affects oxidative stress-related events, defense gene expression, and the hypersensitive response.

The Arabidopsis (Arabidopsis thaliana) phytoalexin-deficient mutant pad2-1 displays enhanced susceptibility to a broad range of pathogens and herbivorous insects that correlates with deficiencies in the production of camalexin, indole glucosinolates, and salicylic acid (SA). The pad2-1 mutation is localized in the GLUTAMATE-CYSTEINE LIGASE (GCL) gene encoding the first enzyme of glutathione biosynthesis. While pad2-1 glutathione deficiency is not caused by a decrease in GCL transcripts, analysis of GCL protein level revealed that pad2-1 plants contained only 48% of the wild-type protein amount. In contrast to the wild type, the oxidized form of GCL was dominant in pad2-1, suggesting a distinct redox environment. This finding was corroborated by the expression of GRX1-roGFP2, showing that the cytosolic glutathione redox potential was significantly less negative in pad2-1. Analysis of oxidative stress-related gene expression showed a higher transcript accumulation in pad2-1 of GLUTATHIONE REDUCTASE, GLUTATHIONE-S-TRANSFERASE, and RESPIRATORY BURST OXIDASE HOMOLOG D in response to the oomycete Phytophthora brassicae. Interestingly, oligogalacturonide elicitation in pad2-1 revealed a lower plasma membrane depolarization that was found to act upstream of an impaired hydrogen peroxide production. This impaired hydrogen peroxide production was also observed during pathogen infection and correlated with a reduced hypersensitive response in pad2-1. In addition, a lack of pathogen-triggered expression of the ISOCHORISMATE SYNTHASE1 gene, coding for the SA-biosynthetic enzyme isochorismate synthase, was identified as the cause of the SA deficiency in pad2-1. Together, our results indicate that the pad2-1 mutation is related to a decrease in GCL protein and that the resulting glutathione deficiency negatively affects important processes of disease resistance.

Interactions between plants and microbes are the result of a finely tuned coevolution. To counter microbial attacks, plants have developed perception systems that activate various defense mechanisms. Two main defense pathways are described in plant innate immunity (Boller and Felix, 2009). The first, named pathogen-associated molecular pattern (PAMP)triggered immunity, is based on the recognition of pathogen-, microbe-, or damage-associated molecular patterns by pattern recognition receptors . The second and more specialized effector-triggered immunity is activated when plant disease resistance gene products detect the presence of pathogen effectors.
The perception of an invader's molecular tag is followed by several changes in plants (Tsuda and Katagiri, 2010). At the cellular level, many signaling events are rapidly detected, such as ion fluxes (Ca 2+ , K + , NO 3 2 , Cl 2 ) and enhanced production of reactive oxygen species (ROS), mainly catalyzed by plasma membrane NADPH oxidases, encoded by Respiratory Burst Oxidase Homolog (Rboh) genes (Simon-Plas et al., 2002;Torres et al., 2002Torres et al., , 2006. This oxidative burst contributes, together with nitric oxide (NO) production, redox state changes, and mitogen-activated protein kinase cascade activation, to a transcriptional reprogramming leading to the activation of defense responses (Asai et al., 2002;Wendehenne et al., 2002;Torres et al., 2006;Besson-Bard et al., 2008;Foyer et al., 2009). Changes in hormonal balance, particularly salicylic acid (SA), jasmonic acid (JA), and ethylene (ET), have been shown to be key events in the activation and fine-tuning of plant immunity upon pathogen attack. Finally, the production of antimicrobial compounds such as phytoalexins (Hammerschmidt, 1999), pathogenesis-related (PR) proteins (van Loon et al., 2006), and cell wall strengthening arise from this defense activation to restrict pathogen growth.
In Arabidopsis (Arabidopsis thaliana), it is now well accepted that SA signaling is generally important for immunity against biotrophs, whereas JA/ET signaling is important for immunity against necrotrophs (Glazebrook, 2005). However, it has been shown that disease resistance of Arabidopsis against the oomycete pathogen Phytophthora brassicae is SA, JA, and ET independent but needs the indole glucosinolate/ camalexin pathways (Roetschi et al., 2001;Schlaeppi et al., 2010). In the case of biotrophic pathogens, like P. brassicae, a form of plant programmed cell death, named the hypersensitive response (HR), can be detected during the incompatible interaction that may restrict pathogen growth (Roetschi et al., 2001;Hofius et al., 2007). On the whole, this complex defense network allows plants to resist most of their invaders.
Many studies have shown that pad2-1 displays a pleiotropic phenotype. Concerning defense responses, the production of camalexin, the main phytoalexin of Arabidopsis, is strongly affected in pad2-1 in response to P. syringae pv maculicola (approximately 15% of the wild type; Glazebrook and Ausubel, 1994), B. cinerea (approximately 25% of the wild type; Ferrari et al., 2003), A. brassicicola (approximately 35% of the wild type; van Wees et al., 2003), or P. brassicae (approximately 40% of the wild type; Roetschi et al., 2001;Parisy et al., 2007). This camalexin deficiency is directly due to glutathione depletion, since glutathione is required for glutathione S-transferase (GSTF6)-catalyzed synthesis of glutathione-indole-3-acetonitrile, a precursor of camalexin biosynthesis (Su et al., 2011). Besides, the impaired camalexin biosynthesis and resistance to P. brassicae in pad2-1 can be restored using either glutathione feeding or transformation with wild-type GCL cDNA (Parisy et al., 2007).
Glucosinolates are another class of sulfur-containing compounds of Brassicaceae whose degradation products are well known to be toxic against insects and pathogens (Halkier and Gershenzon, 2006). Interestingly, Schlaeppi and colleagues recently showed that pad2-1 is impaired in the production of indole glucosinolates in response to insects (approximately 55% of the wild type; Schlaeppi et al., 2008) or P. brassicae infection (approximately 40% of the wild type; Schlaeppi et al., 2010). Moreover, stress-induced SA accumulation and PR-1 expression are also very low in pad2-1 in response to P. brassicae (Roetschi et al., 2001). The SA deficiency of pad2-1 is possibly linked to the requirement of glutathione to modulate the redox state needed for the oligomer/monomer transition of Nonexpressor of Pathogenesis-Related1 (NPR1) genes preceding NPR1 translocation into the nucleus to activate PR-1 gene expression (Després et al., 2003;Mou et al., 2003;Pieterse and Van Loon, 2004). However, it remains unexplained how glutathione modulates SA acting upstream of NPR1.
Many studies reported that the total glutathione concentration, as well as the ratio of reduced to oxidized forms of glutathione, affect cellular redox homeostasis during plant development or environmental stress Mullineaux and Rausch, 2005;Foyer and Noctor, 2011). These cellular redox changes influence the genome-wide expression profile (Ball et al., 2004) and protein activity. For example, GCL is redox activated by the formation of two disulfide bonds (Jez et al., 2004;Hothorn et al., 2006;Hicks et al., 2007). Despite the apparent impact of glutathione depletion on defense signaling events and responses Impaired Defense Signaling in pad2-1 such as HR, these underlying mechanisms have not yet been studied using a genetic approach.
The aim of this study was to identify the molecular origin of the glutathione depletion in pad2-1 and to investigate the putative links between glutathione, the cellular redox potential, ROS production, defense gene expression, and HR. For this purpose and to better understand the disease susceptibility of pad2-1, we compared early signaling events occurring in wildtype Columbia-0 (Col-0) and pad2-1 cells treated with oligogalacturonides (OG), a damage-associated molecular pattern that elicits typical defense responses, and during plant infection with P. brassicae. Our results indicate that the glutathione deficiency in pad2-1 is related to a decreased GCL protein content. They also highlight a role of glutathione in modulating the cellular redox environment, membrane depolarization, ROS and NO production, expression of key defense-related genes, and HR development. These data highlight the crucial role of glutathione in signaling processes underlying essential defense responses in plants.

RESULTS
The GCL Protein Amount Is Lower in pad2-1 and Its Redox State Is More Oxidized The pad2-1 mutation S298N of GCL causes a low glutathione content as compared with the wild type (approximately 20% of the wild type; Parisy et al., 2007). Since the glutathione deficiency of pad2-1 is not related to a reduced expression of GCL or GS genes, which encode the two enzymes involved in glutathione synthesis (Parisy et al., 2007), we analyzed if the S298N mutation affects the GCL protein amount. While the cad2-1, rax1-1, and rml1-1 mutations have been proposed to directly affect amino acid residues involved in the binding of one of the GCL substrates (Glu, Cys, or ATP; Hothorn et al., 2006), nothing was known about the pad2-1 mutation. Using PyMOL software (PyMOL Molecular Graphics System; Schrödinger), in silico prediction indicated that the S298N exchange in pad2-1 did not affect the ternary structure of GCL, contrary to cad2-1 and rax1-1 mutations, which displayed a minor structure modification in one b-sheet (data not shown). Moreover Ser-298 has not been described to be involved in the catalytic site of GCL (involving 12 residues in Arabidopsis:  or in its thiol-based redox regulation sites (Hothorn et al., 2006;Hicks et al., 2007;Gromes et al., 2008). To test the hypothesis that the pad2-1 mutation could affect GCL protein folding or stability, we first compared the GCL protein level between pad2-1 and wild-type plants by western-blot analysis using an antibody raised against purified recombinant AtGCL (Hicks et al., 2007). As GCL occurs in oxidized and reduced forms, tissue extracts were reduced prior to western blotting. Interestingly, pad2-1 leaf tissue con-tained only 48% of the wild-type amount of GCL. However, in the allelic cad2-1 and rax1-1 mutants, the GCL levels were almost the same as in the wild type (Fig. 1A). These results show that pad2-1 is the only one to display a very low GCL protein level, which could explain its low glutathione content. GCL activity is regulated by cellular redox state (Jez et al., 2004;Hothorn et al., 2006;Hicks et al., 2007). In Arabidopsis, the oxidation of the inactive reduced form switches the protein into an active form via the formation of intramolecular disulfide bonds (Hicks et al., 2007). Therefore, we investigated the oxidized and reduced forms of GCL after oxidative stress (Fig.  1B). For this purpose, leaf discs were treated with 5 mM hydrogen peroxide (H 2 O 2 ) and total proteins were extracted in reducing (+dithiothreitol [DTT]) or nonreducing (2DTT) conditions and examined by western blotting. Under reducing conditions, only one band was detected, with a molecular mass of 50 kD in both wild-type and pad2-1 plants (Fig. 1). In nonreducing conditions, two bands were detected in wild-type plants, corresponding to the oxidized and reduced forms of GCL, with apparent molecular masses of 56 and 50 kD, respectively. In accordance with Hicks et al. (2007), the reduced form was found to be dominant in wild-type plants. Oxidative stress induced by H 2 O 2 shifted GCL to the oxidized active form. Under nonreducing conditions, both the reduced and oxidized forms of GCL were detected at a very low level in pad2-1. However, in contrast to wild-type plants, the oxidized form of GCL was already dominant at 0 min, and no obvious change of the redox state of the enzyme was observed within 60 min after treatment with H 2 O 2 . Thus, the redox regulation of the GCL enzyme is impaired in pad2-1 plants. These data also suggest that pad2-1 might display a distinct redox environment, known to be essentially maintained by the glutathione pool (Foyer and Noctor, 2011).

The Glutathione Redox Potential Is Less Reducing in pad2-1 Than in the Wild Type
To investigate whether the cytosolic glutathione redox potential (E GSH ) differs between pad2-1 and the wild type, E GSH was measured with the glutathione-specific redox-sensitive GFP (roGFP) variant GRX1-roGFP2. It was previously demonstrated that this fusion sensor of the human Glutaredoxin1 (GRX1) and roGFP2 equilibrates with glutathione and thus reports the local E GSH (Gutscher et al., 2008).
For probe calibration, wild-type and pad2-1 tissues were treated with 10 mM H 2 O 2 or 10 mM DTT to fully oxidize or reduce the GRX1-roGFP2 protein (Fig. 2B). After 10 mM H 2 O 2 treatment, the GRX1-roGFP2 ratio of the wild type (4.52 6 1.41) and pad2-1 (4.24 6 1.46) increased to almost the same value. Similarly, the ratio of the DTT-treated tissues was comparable in the wild type (0.46 6 0.06) and pad2-1 (0.61 6 0.13). These results showed that the ratiometric behavior of the fluorescent probe, including its dynamic range, was the same in the mutant and the wild-type plants. Based on these calibration values, the ratio values measured in leaves under nonstress conditions were converted to E GSH according to Meyer et al. (2007). While E GSH was 2312 6 10 mV in the wild type, it was only 2275 6 5 mV in pad2-1. Together, these data indicated that the glutathione redox buffer in pad2-1 has less reducing power than that in the wild type.
Expression of Oxidative Stress-Related Genes Is Higher in pad2-1 during P. brassicae Infection Since E GSH is less reducing in pad2-1 compared with the wild type, we evaluated whether pad2-1 perceived a stronger oxidative stress during pathogen infection. Thus, we chose to analyze the expression of oxidative stress-related genes by real-time quantitative PCR at 24 h post inoculation (hpi) with P. brassicae (Fig. 3).
NADPH-dependent Glutathione Reductase1 (GR1) is involved in the glutathione-ascorbate cycle by catalyzing the regeneration of oxidized to reduced glutathione under oxidative stress (Foyer and Noctor, 2011). Indeed, the gr1 mutant of Arabidopsis showed a low capacity to maintain a reduced glutathione pool after H 2 O 2 treatment and displayed a similar E GSH to pad2-1 (Marty et al., 2009). During infection, P. brassicae triggered a significant expression of GR1 transcripts that was more pronounced in pad2-1 than in the wild type (Fig. 3). No significant difference was found between unchallenged pad2-1 and wild-type plants.
Among plant GST genes, which are induced by a variety of stimuli, AtGSTF6 (also called AtGST1) was reported to be induced upon infection or in response to H 2 O 2 treatment (Marrs, 1996;Nutricati et al., 2006;Belhaj et al., 2009). In this study, we show that the transcript level of GSTF6 was significantly up-regulated Figure 2. Comparison of the glutathione redox potential of Col-0 and pad2-1 epidermal cells expressing GRX1-roGFP2 in the cytosol. A, Images were taken by confocal laser scanning microscopy. GRX1-roGFP2 was excited with 405-and 488-nm wavelengths. The color scale of the fluorescence ratio represents the redox state of roGFP2 oscillating between an oxidized (ox) state (red) and a reduced (red) state (blue). B, Quantitative ratio values for images taken on Col-0 and pad2-1 mutant tissue. Water-treated (control), fully oxidized (10 mM H 2 O 2 ), or fully reduced (10 mM DTT) samples for Col-0 and the pad2-1 mutant were compared (mean 6 SE; n = 6). Results are representative of three independent experiments. Impaired Defense Signaling in pad2-1 in response to P. brassicae both in pad2-1 and the wild type ( Fig. 3). However, the increase in transcript accumulation in pad2-1 was more than 2-fold higher than in the wild type.
In addition to GR1 and GSTF6 genes, we chose to analyze the expression of the RbohD gene, because Rboh genes encode NADPH oxidase enzymes that catalyze the production of the ROS O 2 .2 (Levine et al., 1994;Torres et al., 2002;Sagi and Fluhr, 2006). Moreover, RbohD has been shown to be involved in the enhanced production of H 2 O 2 detected after OG elicitation (Galletti et al., 2008) or in response to P. brassicae (Belhaj et al., 2009). Figure 3 shows that the transcript level of the RbohD gene was 2.5 times higher in pad2-1 compared with wild-type plants in response to P. brassicae.

pad2-1 Is Impaired in H 2 O 2 and NO Production during Biotic Stresses
As the expression of the oxidative stress-related marker genes GR1, GSTF6, and RbohD was significantly higher in pad2-1 than in the wild type, we checked H 2 O 2 production in response to two biotic stress conditions: elicitation by OG and inoculation with P. brassicae (Fig. 4).
H 2 O 2 production was first determined in cell suspensions using the chemiluminescence of luminol. In wild-type cells, OG treatment elicited a transient increase in H 2 O 2 accumulation, which peaked at 10 min and decreased slowly until 120 min (Fig. 4A). In pad2-1 cells, the transient increase in H 2 O 2 production was delayed and represented only 30% of the maximal production detected in wild-type cells. This low content of H 2 O 2 in pad2-1 was due to an impairment of its production rather than to an increase in its degradation, as H 2 O 2 half-life and global peroxidase or catalase activities were similar in the two genotypes (Supplemental Fig. S1).
The unexpected loss of OG-induced H 2 O 2 accumulation was also observed in pad2-1 leaves using a modified diaminobenzidine (DAB) staining method ( Fig. 4B). In presence of peroxidases, which are ubiquitous in plant tissue, H 2 O 2 reacts with DAB to produce brown precipitates (Thordal-Christensen et al., 1997). In order to allow a higher sensitivity of the approach, an excess of horseradish peroxidase was added to the DAB solution prior to infiltration into Figure 3. Expression of oxidative stress-related genes in response to P. brassicae at 24 hpi. Plant leaves were infected by agar plug inoculation of 8-d-old mycelium. Mycelium-free plugs were used for mock treatment. The transcript accumulation of genes encoding GR1, GSTF6, and RbohD was analyzed by real-time qPCR. After normalization with UBQ10, results were expressed as fold change in transcript level compared with the time zero sample (mean 6 SE; n = 3) from three independent biological experiments. Hash marks indicate statistically significant differences between mock and infected samples (P , 0.05), and asterisks indicate statistically significant differences between infected Col-0 and infected pad2-1 samples (P , 0.05), using unpaired heteroscedastic Student's t test. production in response to OG or P. brassicae in Col-0 and pad2-1. A, H 2 O 2 was measured using the chemiluminescence of luminol after OG treatment (0.5 g L 21 ) in Col-0 and pad2-1 cell suspensions. Data are expressed as percentage of the maximum H 2 O 2 production in OG-treated Col-0 cells. Values are means 6 SE of three independent experiments. Asterisks indicate statistically significant differences between the wild type and pad2-1, using Student's t test (P , 0.01). B, In planta H 2 O 2 detection. Leaves were treated with 2.5 g L 21 OG during 20 min or inoculated for 1 d with P. brassicae agar plugs before DAB staining. H 2 O 2 production was detected as brown precipitates. Results are representative of three independent experiments.
leaves. This modification allowed improved visualization of OG-induced H 2 O 2 accumulation. No precipitate was detected in control leaves of wild-type and pad2-1 plants. While a strong brown precipitate was observed in wild-type leaves treated during 20 min with OG, H 2 O 2 production was only weakly detected in pad2-1 leaves under the same conditions. Similar results were obtained with leaves infected with P. brassicae for 1 d. Wild-type leaves showed an intense brown precipitate at the infection site, whereas the response was much weaker in pad2-1 leaves (Fig. 4B). H 2 O 2 accumulation in response to OG treatment or P. brassicae infection was completely abolished in the rbohD mutant (data not shown; Belhaj et al., 2009), demonstrating that the H 2 O 2 accumulation in response to P. brassicae or OG treatment depends on RbohD.
Another second messenger rapidly produced after elicitation is NO (Foissner et al., 2000;Besson-Bard et al., 2008). Using the NO-sensitive probe 4,5-diaminofluorescein diacetate (DAF-2DA), intracellular NO production was monitored during 12 h in leaf discs infiltrated with OG (8-h data shown in Supplemental  Fig. S2). In water-infiltrated leaf discs, a basal fluorescence was observed in the wild type and pad2-1, probably reflecting constitutive NO production and/ or the wounding response caused by the preparation of leaf discs. After OG treatment, an increase in fluorescence was observed in the wild type and pad2-1. However, this increase was significantly less pronounced in pad2-1 at 8 h after treatment. The use of the specific NO scavenger carboxy-PTIO confirmed the NO specificity of the probe by suppressing the fluorescence triggered by OG in wild-type and pad2-1 cells.
The Lower Plasma Membrane Depolarization in pad2-1 Cells Acts Upstream of ROS Production Ion fluxes have been described to act upstream of the oxidative burst, known to be involved in HR and notably observed during the resistance of wild-type Arabidopsis to P. brassicae (Heath, 2000;Delledonne et al., 2001;Roetschi et al., 2001;Wendehenne et al., 2002;Greenberg and Yao, 2004). Having established that the production of H 2 O 2 was altered in pad2-1, we checked whether a perturbation of ion fluxes at the plasma membrane level could be the source of this impairment in pad2-1. To test the hypothesis, wild-type and pad2-1 cell suspensions were treated with OG and changes in the plasma membrane potential were monitored by using the voltage-sensitive fluorescent probe bis-(1,3-dibutylbarbituric acid)-trimethine oxonol [Di-BAC 4 (3); Fig. 5; Lamotte et al., 2006;Konrad and Hedrich, 2008].
In wild-type and pad2-1 cells responding to OG treatment, a transient increase in fluorescence, reflecting a plasma membrane depolarization, occurred within the first 5 min and then peaked at 10 min. Within 30 min, DiBAC 4 (3) fluorescence returned to the basal level (Fig. 5). Interestingly, in pad2-1 cells, the maximal fluorescence at 10 min was two times lower than in wild-type cells. This result shows that pad2-1 is partly impaired in OG-dependent plasma membrane depolarization. Addition of the anionic channel inhibitor niflumic acid to wild-type cells 10 min before treatment totally inhibited the increase in DiBAC 4 (3) fluorescence, indicating the involvement of anion channels in this process (Supplemental Fig. S3A). Moreover, the total inhibition of OG-dependent H 2 O 2 production by niflumic acid treatment provided evidence that ROS production occurs downstream of anion channel activation and plasma membrane depolarization (Supplemental Fig. S3B).
These results indicate that pad2-1 is impaired in plasma membrane depolarization and H 2 O 2 and NO production during elicitation.
pad2-1 Is Impaired in HR Development in Response to P. brassicae In many plant-pathogen interactions, one of the mechanisms stopping the pathogen invasion into leaf tissue is the HR. Since pad2-1 is impaired in key defense-related signaling events in response to biotic stresses and because it is susceptible to many pathogens, we compared HR formation in the wild type and pad2-1 after inoculation with P. brassicae (Fig. 6). Plants were inoculated by spraying a zoospore suspension, and the formation of HR and pathogen growth were analyzed by trypan blue staining at 1 d post inoculation (dpi). No significant difference in zoospore germination and pathogen penetration efficiency was found between the wild type and pad2-1 (data not shown). However, in wild-type plants, HR occurred in 86% of attempted infections, whereas this percentage Figure 5. Plasma membrane potential changes in response to OG treatment in Col-0 and pad2-1 cell suspensions. Cells were incubated with the voltage-sensitive fluorescent probe DiBAC 4 (3) (10 mM) in the dark before OG treatment (0.5 g L 21 ). Plasma membrane depolarization was monitored by following the variation of DiBAC 4 (3) fluorescence and was expressed as a relative percentage: [fluorescence (OG)fluorescence (water)]/fluorescence (water). The chart shows mean values 6 SE out of five independent experiments. Asterisks indicate statistically significant differences between the wild type and pad2-1, using Student's t test (P , 0.01).
Impaired Defense Signaling in pad2-1 was reduced to 22% in pad2-1 (Fig. 6A). The reduced rate of HR formation correlated with the enhanced rate of pathogen growth. Encysted zoospores producing hyphae were seen only in 14% of the observations in the wild type compared with 78% in pad2-1 (Fig. 6A). The different responses of the two genotypes to P. brassicae were also illustrated at 7 dpi. The wild type showed HR at both microscopic and macroscopic levels (Fig. 6, B and C), whereas in pad2-1, the pathogen developed extensively, as demonstrated by the massive occurrence of oospores (Fig. 6B) and the wilting of leaves (Fig. 6C).
Expression of SA-Related Genes Is Impaired during P. brassicae Infection of pad2-1 One important phenotypic aspect observed in pad2-1 is the impairment of SA accumulation in response to P. brassicae (Roetschi et al., 2001). In order to find the putative origin of this SA deregulation, we first analyzed the expression of the ISOCHORISMATE SYNTHASE1 (ICS1) gene, which encodes an important enzyme involved in SA biosynthesis (Nawrath and Métraux, 1999;Wildermuth et al., 2001). ICS1 transcript accumulation was 6-fold increased in the wild type in response to P. brassicae, but no increase in transcript level was detected in pad2-1 (Fig. 7). In parallel, the regulation of PR-1 and NPR1 gene expression was checked at 24 hpi with P. brassicae (Fig. 7). In accordance with the low ICS1 transcript levels and the previously reported reduction of SA accumulation in pad2-1, transcript accumulation of the SA marker gene PR-1 was three times lower in pad2-1 compared with the wild type. NPR1 gene expression was not induced in either genotype.

Transcript Levels of Genes Involved in Camalexin and Indole Glucosinolate Biosynthesis Are Similar in pad2-1 and the Wild Type
Together with camalexin, glucosinolates have recently been shown to play a major role in the resistance to P. brassicae (Schlaeppi et al., 2010). The pad2-1 mutant displayed a reduced accumulation of camalexin and glucosinolates (Roetschi et al., 2001;Parisy et al., 2007;Schlaeppi et al., 2010). To understand whether this defense pathway was down-regulated at the transcriptional level, gene expression was analyzed for CYP79B2 and CYP79B3, which encode two enzymes involved in the conversion of Trp to indole-3-aldoxime, a common precursor of camalexin and indole glucosinolates (Zhao et al., 2002;Bednarek et al., 2009). Furthermore, the expression of the indole glucosinolate-specific CYP81F2 gene and the camalexin-specific PAD3 gene (CYP71B15) were analyzed (Böttcher et al., 2009;Pfalz et al., 2009). All four genes were similarly up-regulated in response to P. brassicae in pad2-1 and wild-type plants (Supplemental Fig. S4). Together, these results indicate that the impairment in the indole glucosinolate/camalexin pathway, revealed by the lower metabolite content in pad2-1, is not due to the different transcript accumulation of the analyzed biosynthetic genes.

DISCUSSION
The Glutathione Depletion in pad2-1 Is Related to a Lower Amount of GCL Protein GCL is described as the major control point of glutathione synthesis in plants, and the amount of glutathione seems to depend on the regulation of GCL at both transcriptional and posttranslational levels Jez et al., 2004;Hicks et al., 2007). Under normal physiological conditions, GCL and GS transcript levels are similar in pad2-1 and wild-type plants (Parisy et al., 2007). Moreover, during infection with P. brassicae, GCL and GS transcript accumulation is higher in pad2-1 as compared with the wild type, Figure 6. HR and P. brassicae development in Col-0 and pad2-1 plants. Leaves were inoculated by spraying a zoospore suspension of P. brassicae (10 5 zoospores mL 21 ) and harvested at the time indicated below. HR and pathogen structures were observed with lactophenol trypan blue staining. A, Percentages of the microscopic observations of three biological replicates at 1 dpi. Asterisks indicate statistically significant differences between the wild type and pad2-1, using a x 2 test (P , 0.01). B, Microscopic observation of HR (arrow) and pathogen development (dotted arrow; oospore) at 7 dpi. C, HR and water-soaked disease symptoms at 7 dpi with P. brassicae agar plugs in Col-0 and pad2-1 plants, respectively.

Dubreuil-Maurizi et al.
while the glutathione level remains lower in the mutant (approximately 20% of the wild-type amount; Parisy et al., 2007).
To establish a functional relationship between glutathione and GCL in pad2-1, we compared the amount and posttranslational regulation of the GCL enzyme in pad2-1 and wild-type plants. GCL activity is modulated by redox control through the formation of two intramolecular disulfide bonds (C186-C406 and C349-C364), which enable a switch of the inactive reduced form to the active oxidized one under oxidative conditions (Jez et al., 2004;Hothorn et al., 2006;Hicks et al., 2007). Under reducing conditions, our results showed that the total amount of GCL was drastically lower in pad2-1 compared with the wild type or the two allelic mutants cad2-1 and rax1-1 (Fig. 1A). Studying the crystal structure of GCL, Hothorn et al. (2006) showed that cad2-1 and rax1-1 mutations are located in regions critical for binding of the substrates Glu and/ or Cys, thus explaining the glutathione depletion in cad2-1 and rax1-1. However, all the critical residues involved in the catalytic activity or the redox regulation of GCL are not affected by the pad2-1 mutation S298N. Our results indicate that the glutathione deficiency of pad2-1 is likely caused by a drastic decrease in GCL protein content. Nevertheless, we cannot fully exclude that the proximity of the pad2-1 mutation to Arg-300 might affect Glu binding of the GCL enzyme (Hothorn et al., 2006), leading to a change in enzyme activity and thus a decrease in glutathione content.

pad2-1 Plants Sense a Permanent Oxidative Stress
Glutathione is known to play a central role in antioxidant processes as an efficient scavenger of ROS through its cysteinyl thiol group (Noctor, 2006), although its direct role is still disputed (Winterbourn, 2008). The balance between the reduced and oxidized forms of glutathione is involved in cellular redox homeostasis and influences, directly or indirectly, the regulation of many cellular processes at the gene and/ or protein levels (Cobbett et al., 1998;Mou et al., 2003;Ball et al., 2004;Jez et al., 2004). To check whether the susceptibility of pad2-1 to pathogens could be related to a deregulation of the glutathione-dependent control of the cellular redox state, we analyzed, on the one hand, the reduced and oxidized forms of GCL and, on the other hand, the E GSH with the GRX1-roGFP2 fluorescent probe in vivo (Meyer et al., 2007;Gutscher et al., 2008;Schwarzländer et al., 2008).
Our study confirms that the oxidative stress triggered by H 2 O 2 switches the reduced to the oxidized form of the GCL enzyme in the wild type (Fig. 1B), as demonstrated previously by Hicks et al. (2007). The more oxidized GCL protein observed in pad2-1 (Fig.  1B) is most probably due to a less reducing E GSH caused by the diminished glutathione levels in the mutant, as revealed by the GRX1-roGFP2 measurements.
In the absence of any treatment, GRX1-roGFP2 was more oxidized in pad2-1 than in the wild type, indicating higher E GSH in pad2-1 (E GSH of approximately 2275 mV compared with 2312 mV in the wild type). The gr1 mutant of Arabidopsis showed a low capacity to maintain a reduced glutathione pool after H 2 O 2 treatment and displayed a similar E GSH to pad2-1 (2270 mV; Marty et al., 2009). Similarly, in the glutathione-deficient cad2-1 mutant, containing 30% of wild-type glutathione, an increased fraction of GRX1-roGFP2 is found in the oxidized state (E GSH of approximately 2300 mV; Meyer et al., 2007). The gr1 and cad2-1 mutants are susceptible to P. syringae pv tomato and pv maculicola, respectively (Parisy et al., 2007;Mhamdi et al., 2010). Moreover, cad2-1 is moderately susceptible to P. brassicae, but not to such an extent as pad2-1 (Parisy et al., 2007). All the abovementioned mutants, having higher E GSH , are more sensitive to pathogen attack (Meyer et al., 2007;Parisy et al., 2007;this work). These data collectively indicate that a shift in the E GSH in the cytosol is sufficient to reduce plant disease resistance, as recently shown by Maughan et al. (2010). Modulation of the cellular redox state is known to be one of the components acting on the regulation of gene expression through Figure 7. Expression of SA-related genes in response to P. brassicae at 24 hpi. Plant leaves were agar plug inoculated with 8-dold mycelium. Mycelium-free plugs were used for mock treatment. The transcript accumulation of ICS1, PR-1, and NPR1 was analyzed by real-time qPCR. After normalization with UBQ10, results were expressed as fold change in transcript level compared with the time zero sample (mean 6 SE; n = 3) from three independent biological experiments. Hash marks indicate statistically significant differences between mock and infected samples (P , 0.05), and asterisks indicate statistically significant differences between infected Col-0 and infected pad2-1 samples (P , 0.05), using unpaired heteroscedastic Student's t test.
Impaired Defense Signaling in pad2-1 oxidation or a reduction of transcription factors (Apel and Hirt, 2004;Mittler et al., 2004). Interestingly, during P. brassicae infection, the expression of marker genes of oxidative stress, such as GR1, GSTF6, and RbohD, was significantly higher in pad2-1, suggesting the perception of a more oxidized environment during biotic stresses (Fig. 3). However, our results show that H 2 O 2 production was drastically decreased in pad2-1 after OG elicitation or P. brassicae infection ( Fig.  4; Supplemental Fig. S2). As there was no difference in the ability to degrade H 2 O 2 between pad2-1 and the wild type (Supplemental Fig. S1), this striking discrepancy highlights a central role of glutathione in the regulation of ROS production. As RbohD has been shown to be the main source of ROS detected after OG treatment (Galletti et al., 2008) or P. brassicae inoculation (Belhaj et al., 2009), these results indicate that RbohD gene expression is not sufficient to trigger enhanced NADPH oxidase activity but that posttranslational modifications and upstream signaling events, like plasma membrane depolarization, are also needed (Gauthier et al., 2007, Jeworutzki et al., 2010. As NADPH oxidase activity is directly regulated by cytosolic Ca 2+ partly released from internal pools such as the endoplasmic reticulum Vandelle et al., 2006), it is interesting that S-nitrosoglutathione influences cardiac reticulum Ca 2+ release by activating the ryanodine receptor via S-nitrosylation (Sun et al., 2008). As pad2-1 produces less NO and glutathione, one hypothesis would suggest that cytosolic Ca 2+ variations might be impaired in pad2-1, thus explaining its impaired H 2 O 2 production.

pad2-1 Is Impaired in Early Signaling Events Critical for HR Induction
Ion fluxes across membranes were shown to act upstream of PAMP-triggered early signaling events, including changes in plasma membrane potential, as well as H 2 O 2 and NO synthesis (Garcia-Brugger et al., 2006;Gauthier et al., 2007;Jeworutzki et al., 2010). Our data show a strong decrease in plasma membrane depolarization in pad2-1 cells responding to OG elicitation (Fig. 5). In the wild type, the use of the anionic channel inhibitor niflumic acid has shown that channel activity is involved in plasma membrane depolarization and acts upstream of ROS production during OG elicitation (Supplemental Fig. S3). In accordance with its lower plasma membrane depolarization, pad2-1 also displayed reduced NO and H 2 O 2 production in response to elicitor treatment or P. brassicae infection. These data are consistent with those of Belhaj et al. (2009), who found that the Arabidopsis resistance to Phytophthora1 mutant, which is highly susceptible to P. brassicae, is also impaired in H 2 O 2 production. More generally, the impairment of early signaling events observed in pad2-1 might explain, at least partly, the strong reduction of HR in response to P. brassicae (Fig. 6). Indeed, plasma membrane depolarization and ROS and NO production have been reported to mediate pathogen-or PAMP-triggered defense gene expression and cell death (Wendehenne et al., 2002;Zago et al., 2006;Zaninotto et al., 2006;Hofius et al., 2007;Mur et al., 2008). In mammals, recent studies indicate that some channels could be regulated via S-glutathionylation . Thus, glutathione depletion could disturb ion fluxes across the plasma membrane, impairing its normal depolarization.
SA-Dependent Defense Responses, But Not the Indole Glucosinolate/Camalexin Pathway, Are Blocked at the Transcriptional Level in pad2-1 Glutathione has been shown to be important for inducing the SA-dependent defense pathway, notably by modifying the redox state of NPR1 (Després et al., 2003;Mou et al., 2003). Moreover, glutathione accumulation has been shown to coincide with SA production and the concomitant activation of the PR-1 marker gene (Koornneef et al., 2008). Previous experiments have also demonstrated that pad2-1 is impaired in the synthesis of SA and in PR-1 gene expression (Roetschi et al., 2001;Parisy et al., 2007). Our results show that, in contrast to the wild type, PR-1 gene expression remains very low during P. brassicae infection on pad2-1 plants (Fig. 7). As NPR1 gene expression was not affected in pad2-1 (Fig.  7), it provides further evidence that redox regulation of the NPR1 protein is mandatory to trigger PR-1 gene expression (Després et al., 2003;Mou et al., 2003;Pieterse and Van Loon, 2004). Using the glutathionedeficient cad2-1 and rax1-1 mutants, Ball et al. (2004) also showed that the content of glutathione influences the expression of other defense-related genes. However, redox regulation of NPR1 cannot fully explain how glutathione can modulate the level of SA. Because the ICS1 enzyme has been shown to participate in the biosynthesis of SA (Wildermuth et al., 2001), ICS1 expression was followed in response to P. brassicae. Interestingly, we found that the normal up-regulation of ICS1 gene expression after pathogen infection is totally abolished in pad2-1 (Fig. 7), thus providing a possible explanation for the low SA levels observed in this mutant. As the two transcription factors SAR Deficient1 (SARD1) and CBP60g have been shown to be key regulators for ICS1 expression induction via binding to its promoter (Zhang et al., 2010), further investigation will be needed to fully understand exactly how glutathione depletion and redox potential modifications affect SA biosynthesis.
Production of the secondary metabolites camalexin and indole glucosinolates is strongly affected in pad2-1 in response to pathogens or insects (Glazebrook and Ausubel, 1994;Roetschi et al., 2001;Ferrari et al., 2003;van Wees et al., 2003;Parisy et al., 2007;Schlaeppi et al., 2008Schlaeppi et al., , 2010. Here, we show that these deficiencies are not caused by the down-regulation of genes encoding enzymes of the biosynthetic pathway, because CYP79B2, CYP79B3, CYP81F2, and CYP71B15 are normally expressed in pad2-1 in response to P. brassicae (Supplemental Fig. S4). The camalexin and indole glucosinolate deficiencies are more likely to be caused by the lack of glutathione as a biosynthetic substrate. Glutathione was recently identified as the sulfur donor of camalexin and indole glucosinolate biosynthesis (Geu-Flores et al., 2009;Su et al., 2011). Furthermore, we cannot exclude that many hemecontaining enzymes, such as cytochrome P450, could also be disturbed in their catalytic process by the distinct redox environment existing in pad2-1 (Fig. 2). Indeed, it is interesting that the iron (III) protoporphyrin IX, important for P450 oxidoreductase activity, is maintained in the catalytic site by a thiol from a cysteinyl residue. This heme-Cys bond could probably be modified by the glutathione depletion, thus disturbing the normal electron transfer flow. In mammals, recent results have defined how thiol/disulfide redox switches control heme binding to regulate the activities of different enzymes (Ragsdale and Yi, 2011).

CONCLUSION
In summary, we report that the glutathione depletion in pad2-1 is related to an alteration of GCL content associated with redox state modifications that disturb biological responses under normal and stress conditions. Our results highlight that glutathione modulates early signaling events (plasma membrane depolarization and ROS and NO production), governs oxidative stress-related processes (redox state and GR1, GSTF6, and RbohD gene expression), defense gene expression (ICS1 and PR-1), and programmed cell death (HR) that normally enhance disease resistance. These deficiencies, in addition to the previously described deficiency in the accumulation of indole-derived secondary metabolites, are likely to contribute to the enhanced susceptibility of pad2-1 to pathogens and pests.

Cell Culture Conditions and Treatment
Arabidopsis (Arabidopsis thaliana) ecotype Col-0 wild-type and pad2-1 cell suspensions were cultivated in Gamborg B5 medium (Duchefa;Gamborg et al., 1968) with 30 g L 21 Suc and 0.2 mg L 21 1-naphthalene acetic acid (Duchefa) on a rotary shaker (120 rpm) at 24°C under continuous light (25 mmol m 22 s 21 ). Cell suspension was routinely subcultured once per week. For elicitor treatment, cells were collected 1 d after subculturing, washed with M10 buffer (10 mM MES, 175 mM mannitol, 0.5 mM K 2 SO 4 , and 0.5 mM CaCl 2 , pH 6.2), and resuspended at 0.1 g fresh weight of cells mL 21 M10 buffer. After 1 h of equilibration (125 rpm at 24°C), cells were treated with water or 0.5 g L 21 OG (Goëmar Laboratories), and measurements of plasma membrane depolarization or H 2 O 2 production were performed.

Plant Growth Conditions and Treatment
Wild-type and pad2-1, cad2-1, and rax1-1 mutant Arabidopsis plants were grown in Jiffy-7 peat pellets (Jiffy) in a controlled growth chamber under a 10/ 14-h day/night cycle at 20°C/18°C (70% relative humidity) with a light intensity of 175 mmol m 22 s 21 provided by fluorescent tubes.
The transformation of wild-type and pad2-1 plants with GRX1-roGFP2 was done by floral dip, according to Clough and Bent (1998).
For elicitation, leaf discs (7 mm diameter) from 8-week-old plants were treated with water or 2.5 g L 21 OG in 50 mM Tris-HCl (pH 7.5) under vacuum conditions for 2 min and then collected at the indicated times. For NO production measurement, the NO scavenger carboxy-PTIO (0.5 mM; Sigma-Aldrich) and the fluorescent probe DAF-2DA (20 mM; Sigma-Aldrich) were added in the Tris buffer at the same time as OG. Precisely eight leaf discs from eight plants were used for each treatment. For GCL redox state analysis, 12 discs were cut and treated from three plants per kinetic point. For H 2 O 2 detection, leaves were cut and infiltrated with 20 mM MES buffer (pH 6.2) containing 2.5 g L 21 OG, 10 units mL 21 peroxidase, and 1 g L 21 DAB (Sigma-Aldrich).

Pathogen Growth and Plant Inoculation
Phytophthora brassicae isolate D was routinely grown on 20% V8 agar medium (Campbell Foods) supplemented with 3.5 g L 21 CaCO 3 (pH 5) in a controlled chamber in the dark at 19°C.
For H 2 O 2 detection and gene expression analysis, 4-week-old plants were inoculated by putting agar plugs of 8-d-old mycelium upside down on the leaf surface. Mycelium-free plugs were used for mock treatments. Twelve leaves from three plants per time point were used, and susceptibility was checked at 7 dpi by estimating symptom development as described previously (Parisy et al., 2007). For HR and susceptibility assays, plants were inoculated by spraying a suspension of zoospores at a concentration of 10 5 zoospores mL 21 . Inoculated plants were put under the same conditions as pathogen growth conditions at 100% relative humidity with a dark period for the first 14 h.

Measurement of Plasma Membrane Depolarization
Plasma membrane depolarization was monitored using the voltage-sensitive fluorescent probe DiBAC 4 (3) (Sigma-Aldrich; Konrad and Hedrich, 2008). After 1 h of equilibration, cells were incubated for 30 min with 10 mM DiBAC 4 (3) in M10 buffer in the dark. Then, 0.5 mL of cells was transferred onto 24-well plates (Costar) and treated with water or OG. DiBAC 4 (3) fluorescence was recorded at 1-min intervals using a fluorimeter (Fluoroskan Ascent Fluorometer [Labsystems]; l excitation = 485 nm and l emission = 535 nm) and expressed as relative fluorescence units. Cells were maintained under shaking during the course of the experiment.

H 2 O 2 Production Measurement and Leaf Tissue Localization
In cell suspensions, H 2 O 2 production was determined using the chemiluminescence of luminol. An aliquot of cells (250 mL) was added to 300 mL of H50 buffer (50 mM HEPES, 175 mM mannitol, 0.5 mM K 2 SO 4 , and 5 mM CaCl 2 , pH 8.5) and 50 mL of 0.3 mM luminol. Chemiluminescence was measured with a luminometer (Lumat LB9507; Berthold). The content of H 2 O 2 (nmol g 21 fresh weight of cells) was calculated using a calibration based on the addition of H 2 O 2 to the aliquot of cells.
In plants, a modified procedure described by Thordal-Christensen et al. (1997) was applied. Leaves were vacuum infiltrated with 20 mM MES buffer (pH 6.2) containing 2.5 g L 21 OG, 1 g L 21 DAB, and 10 units mL 21 horseradish peroxidase, then washed with MES buffer. The same solution without OG was used as a control. Infiltrated leaves were incubated in the dark at room temperature for 1 to 8 h. Then, leaves were destained with a solution of glycerol:lactic acid:ethanol (1:1:3, v/v/v) at 85°C for 5 min. H 2 O 2 production was visualized macroscopically as brown precipitates in plant tissues.
Measurement of H 2 O 2 Half-Life, Total Peroxidase Activity, and Total Catalase Activity H 2 O 2 half-life and catalase activity were determined spectrophotometrically by measuring the absorbance of H 2 O 2 at 240 nm, and peroxidase activity was determined by measuring the absorbance of oxidized guaiacol at 470 nm (for details, see Supplemental Protocol S1).

NO Production Measurement
Intracellular NO accumulation was determined using the specific fluorophore DAF-2DA (Sigma-Aldrich) as described by Besson-Bard et al. (2009). Briefly, after infiltration and treatment with fluorophore, leaf discs were incubated for 1 h in the dark, washed three times with 50 mM Tris buffer, pH 7.5, to wash off excessive fluorophore, and transferred onto 96-well plates (Costar) containing 100 mL of the treatment solution. NO production was measured using a fluorimeter (Fluoroskan Ascent Fluorometer [Labsystems]; l ex = 485 nm and l em = 510 nm). Fluorescence was expressed as relative fluorescence units.

Protein Extraction and Western-Blot Analyses
Totals proteins were extracted from 12 leaf discs in 350 mL of extraction buffer (50 mM HEPES, 10 mM EGTA, 10 mM EDTA, 1 mM Na 3 VO 4 , 50 mM b-glycerolphosphate, 10 mM NaF, 5 mg L 21 leupeptin, 5 mg L 21 antipain, and 1 mM phenylmethylsulfonyl fluoride, pH 7.5) with or without 0.1 M DTT. After centrifugation (15 min, 10,000g), 15 mg of soluble proteins was treated with Laemmli buffer (4 min at 100°C) and subjected to 12% SDS-PAGE before transfer to nitrocellulose membranes (Hybond ECL; Amersham Biosciences) for western-blot analysis. GCL protein (At4g23100) was detected with an antibody raised against recombinant AtGCL (antibody provided by J. Jez, Donald Danforth Plant Science Center). Primary antibody detection was performed as described for the LumiGlo detection kit (Cell Signaling Technology) with a horseradish peroxidase-coupled secondary antibody (Sigma-Aldrich).

RNA Extraction and Real-Time Quantitative PCR Analysis
Total RNA was extracted using the total RNA isolation kit (Promega), and 1 mg of total RNA was reverse transcribed using the SuperScript III reverse transcriptase kit (Invitrogen). Real-time quantitative (q)PCR was performed using 2 mL of 80-fold diluted cDNA, ABsolute qPCR SYBR Green ROX mix (containing Taq polymerase, deoxyribonucleotide triphosphate, and SYBR Green dye; ABgene), and 500 nM primers (primer sequences are shown in Supplemental Table S1) in a 5-mL volume. Triplicate quantitative assays per biological experiment were performed by using the LightCycler480 detection system (Roche). The activation factor of gene expression was determined with the comparative cycle threshold (Ct) method (Livak and Schmittgen, 2001): 2 2DDCt ; with DDCt = DCt(t = 24 h) 2 DCt(t = 0) and DCt = Ct(target gene) -Ct (reference gene) for both inoculated and mock samples. UBQ10 was used as the reference gene (Supplemental Table S1).

HR and Pathogen Development
The infected leaves were harvested at the appropriate time points and stained with an alcoholic lactophenol trypan blue solution (Keogh et al., 1980) by boiling for 1 min and keeping them overnight in the staining solution. Leaves were clarified in 70% (w/v) chloral hydrate solution in water. HR and pathogen structures were examined with a Leitz DM RB microscope (Leica).

Confocal Laser Scanning Microscopy Imaging and Ratiometric Analysis
Confocal laser scanning microscopy imaging and ratiometric analysis in wild-type and pad2-1 plants expressing GRX1-roGFP2 were carried out as described by Schwarzländer et al. (2008). Briefly, the fluorescence of GRX1-roGFP2 was analyzed in leaf tissue of 7-d-old in vitro-grown seedlings. Images were taken with a 253 lens, in multitrack mode, with line switching and averaging of two frames. The probe was excited with a 405-nm laser in track 1 and with a 488-nm laser in track 2. GRX1-roGFP2 fluorescence was collected with a band-pass filter of 505 to 530 nm.
Ratiometric image analysis was done using a custom Matlab analysis suite as described by Schwarzländer et al. (2008).

Supplemental Data
The following materials are available in the online version of this article.
Supplemental Figure S1. Measurement of H 2 O 2 half-life and total peroxidase and catalase activities in Col-0 and pad2-1 plants.
Supplemental Figure S2. Intracellular NO production at 8 h after treatment with OG in Col-0 and pad2-1 plants.
Supplemental Figure S3. Effect of the anionic channel inhibitor niflumic acid on OG-induced changes of plasma membrane potential and H 2 O 2 production in Col-0 cell suspensions.
Supplemental Figure S4. Expression of genes involved in camalexin and glucosinolate biosynthesis in response to P. brassicae at 24 hpi.
Supplemental Table S1. Primers used for qPCR.
Supplemental Protocol S1. Measurement of H 2 O 2 half-life, total peroxidase activity, and total catalase activity.