•Signalling between reactive oxygen species (ROS) and salicylic acid (SA)-dependent programmed cell death (PCD) and defence responses is complex and much remains to be discovered. Recent reports have implicated myo-inositol (MI) in defence responses, but the relationships between MI, ROS and SA remain to be elucidated.
•This question was investigated in catalase-deficient Arabidopsis thaliana plants (cat2), in which a peroxisomal H2O2 trigger induces SA-dependent lesion formation and a wide range of pathogen responses.
•GC-MS analysis revealed that leaf MI contents were markedly decreased in cat2 independently of SA accumulation. Supplying MI to cat2 blocked lesion formation, SA accumulation and associated defence responses in a manner that closely mimicked the effect of genetically blocking SA synthesis through isochorismate synthase 1 (ICS1). The effects of MI were linked to repression of ICS1 transcripts but not decreased oxidative stress or signalling, and caused loss of resistance to virulent bacteria. The antagonistic effects of MI on lesion formation and resistance could be partly restored by supplying SA.
•Our findings demonstrate a role for MI in cell death triggered by peroxisomal H2O2, and suggest that the tissue content of this compound is a key factor determining whether oxidative stress induces or opposes defence responses.
Programmed cell death (PCD) is necessary for plant life, and occurs during growth and developmental processes, as well as in stress responses. For example, PCD occurs during seed development, seed germination, senescence and after exposure to extreme temperatures or ozone (Souter & Lindsey, 2000; Young & Gallie, 2000; Gunawardena et al., 2004; Vacca et al., 2004; Ahlfors et al., 2008). However, the most studied PCD is cell death that is part of the hypersensitive response (HR) occurring during incompatible plant–pathogen interactions (van Doorn & Woltering, 2005; Vlot et al., 2009). HR is associated with the generation of reactive oxygen species (ROS) at the sites of entry of avirulent pathogens and the induction of defence responses in both infected and distant tissues (Shirasu et al., 1997). It is widely assumed that the primary origin of ROS during HR is apoplastic and that they are most likely to be generated by NADPH oxidases and cell wall peroxidases (Bolwell et al., 2002; Torres et al., 2006). However, intracellular compartments may also be important potential sources of ROS during PCD and disease resistance processes. High amounts of ROS can be produced as by-products of plant metabolism in mitochondria, chloroplasts and peroxisomes (Foyer & Noctor, 2003). Thus, increased ROS availability may be caused by enhanced production or decreased removal, or both. For example, treatment of tobacco suspension cells with salicylic acid (SA), H2O2 and a specific mitochondrial function inhibitor leads to intracellular ROS accumulation and induction of genes that are induced during PCD (Maxwell et al., 2002), while work on catalase-deficient lines shows that H2O2 produced in the peroxisomes can also trigger responses such as lesion formation and induction of pathogenesis-related (PR) genes (Takahashi et al., 1997; Chamnongpol et al., 1998).
An early defence response that characterizes the onset of HR is SA accumulation (Lamb & Dixon, 1997). At least some effects of SA action are clearly downstream of ROS (Bi et al., 1995), but ROS and SA could also act synergistically in a signal amplification loop (Shirasu et al., 1997; Overmyer et al., 2003). Despite the extensive work on the roles of ROS and SA in hypersensitive cell death, the relationship between them is complicated, and many factors determining this relationship remain to be identified (Vlot et al., 2009).
A powerful approach to unravelling the pathways of PCD, particularly the HR, is the identification of mutant plants with misregulated cell death. Such plants are called lesion mimic mutants (LMMs) because they present spontaneous HR-like lesions in the absence of pathogens (Alvarez, 2000). Since the discovery of the lsd1 mutant (Dietrich et al., 1994), many other LMMs have been identified in Arabidopsis. Some but not all show phenotypes that are SA-dependent (Lorrain et al., 2003). In many cases, the roles of ROS in LMMs remain to be demonstrated. A recently identified LMM is mips1, which is deficient in the enzyme catalysing the limiting step of myo-inositol (MI) synthesis (Meng et al., 2009; Donahue et al., 2010). This provides a first indication of a role for this compound in PCD control. MI is a precursor for numerous compounds, including several with possible antioxidant activity (Loewus & Murthy, 2000; Nishizawa et al., 2008). It may also be involved in ascorbate synthesis (Torabinejad et al., 2009). However, the relationship between SA, MI, and ROS signalling remains to be established.
Recently, we reported that the Arabidopsis cat2 mutant, which is deficient in the major leaf catalase isoform (Queval et al., 2007), shows conditional activation of a wide range of pathogen responses and presents the physiological characteristics of an initiation-type LMM (Chaouch et al., 2010). The initial trigger of effects in cat2 is H2O2 produced in the peroxisomes, which causes characteristic and predictable adjustments in cell redox state that are measurable as increases in oxidized glutathione and induction of oxidative stress marker genes. Construction of a double cat2 sid2 mutant and complementation experiments showed that cell death and associated defence responses were completely dependent on SA in cat2 (Chaouch et al., 2010). These findings establish cat2 as a useful model system in which to investigate factors that modify oxidative stress sensitivity and/or SA-dependent responses unambiguously triggered by H2O2. Thus, we have been investigating metabolites that could affect cell death and pathogen responses in the cat2 mutant. In the present rapid report, we show that peroxisomal-sourced H2O2 only induces SA-dependent lesion formation and disease resistance responses if MI is decreased. Our findings therefore situate MI action downstream of ROS in pathogen-associated responses, and also provide evidence that MI does not act primarily through effects on antioxidant status.
Materials and Methods
Plant material and sampling
All mutants were in the Arabidopsis thaliana Columbia (Col-0) background. The cat2 line was cat2-2 (Queval et al., 2007), now renamed cat2-1. The sid2 (salicylic acid induction deficient2) mutation in isochorismate synthase 1 abolishes pathogen-induced SA accumulation and SA-dependent responses (Wildermuth et al., 2001). This mutation was introduced into the cat2 background and double homozygotes characterized as described by Chaouch et al. (2010). For all experiments, plants were grown in a controlled-environment growth room at an irradiance of 200 μmol quanta m−2 s−1 at leaf level, 20/18°C, and 65% humidity, and given nutrient solution twice per wk. For results presented in Fig. 1, plants were grown from seed in a day/night regime of 16/8 h (LD) and under atmospheric concentrations of CO2 (400 μl l−1, i.e. air). For all other results, plants were initially grown in a day/night regime of 8/16 h (SD) and under a high CO2 concentration (3000 μl l−1) for 3 wk then transferred to air in LD for 8 d or the time indicated. Complementation experiments began by treating plants by a single spraying on the day after transfer to LD in air, and the treatment was repeated each day until sampling. Rosettes were treated with 1 mM SA or 11 mM MI, or both. Samples from at least three different plants were taken, rapidly frozen in liquid nitrogen, and stored at −80°C until analysis. Unless otherwise stated, data are means ± SE of three to four different samples. All experiments were repeated two to three times with similar results.
Pathogen resistance tests
For resistance tests, the virulent Pseudomonas syringae pv. tomato DC3000 strain was used in a medium titre of 106 cfu ml−1. One week after transfer to LD, whole leaves of plants were infiltrated using a 1 ml syringe without a needle and leaf discs were taken for analysis at the same time point on the subsequent day. Five to six samples were made by pooling two leaf discs (0.5 cm2 each) harvested from different inoculated leaves of different plants. Bacterial growth was assessed by homogenizing leaf discs in 400 μl of water, plating appropriate dilutions on solid King B medium containing rifampicin and kanamycin, and quantifying colony numbers after 3 d. Trypan blue staining was performed as in Chaouch et al. (2010).
RNA was extracted using the kit NucleoSpin® RNA plant (Macherey-Nagel, Hoerdt, France) and reverse-transcribed with the SuperScript™ III First-Strand Synthesis System (Invitrogen). qPCR was performed according to Queval et al. (2007) using ACTIN2 transcripts as a control. Primer sequences were as in Chaouch et al. (2010).
Oxidized and reduced forms of glutathione and ascorbate were measured by plate-reader assay, as described in Queval & Noctor (2007). Camalexin and free and total forms of SA and scopoletin were measured by high-performance liquid chromatography (HPLC), according to Chaouch et al. (2010). MI was measured using GC-MS as in Noctor et al. (2007).
H2O2-induced cell death is associated with decreased MI
When grown in air and in LD conditions, the cat2 mutation leads to perturbation of cellular redox status, which in turn activates signalling events that induce spontaneous lesion formation (Fig. 1a). As shown in Fig. 1, this effect is associated with accumulation of both free and total forms of SA (Fig. 1c,d). Metabolite profiling of cat2 by GC-MS to identify oxidative stress-sensitive compounds revealed a dramatic decrease in MI compared with Col-0 (Fig. 1b,e). Introduction of the sid2 mutation into the cat2 background prevented SA accumulation (Fig. 1c,d) and formation of lesions (Fig. 1a). The sid2-dependent reversion of cell death was not accompanied by changes in MI, which remained at the low concentrations observed in cat2 (Fig. 1b,e).
Effects of MI on H2O2-induced lesion formation and cell redox state
Because of the cat2-dependent decreases in MI, we examined the effect on the cat2 phenotype of treating plants with exogenous MI. For this, we developed an experimental protocol that takes advantage of the conditional photorespiratory nature of the cat2 mutant (Queval et al., 2007). Plants were initially grown at high CO2, where this mutation is silent because photorespiratory H2O2 production is very slow. Accordingly, Col-0 and cat2 were phenotypically identical after 3 wk growth at high CO2 (Supporting Information, Fig. S1). They were then transferred to conditions permissive for the development of cell death in cat2 (i.e. LD and atmospheric CO2 concentration) and treated with water, MI, or MI and SA together. In water-treated plants, lesions were clearly visible at 6 d after transfer and were extensive after 8 d (Fig. S1). In contrast, lesions did not develop on plants receiving MI treatment, even after 8 d (Fig. 2a). Preliminary experiments showed that the inhibitory effect of MI on lesion formation was already apparent at a concentration of 1 mM and became saturated at 10 mM (data not shown). Other studies on leaf discs have used a treatment concentration of 5 mM MI (Endres & Tenhaken, 2009), and a concentration of 11 mM was used here to ensure full effects. When MI treatment at this concentration was performed together with treatment with 1 mM SA, lesion formation in cat2 was partly restored. None of the treatments affected the phenotype of Col-0 plants (Fig. 2a). Likewise, neither Col-0 nor cat2 phenotypes were affected by treatment with SA alone (Chaouch et al., 2010).
Several factors make quantification of H2O2 in plants problematic (Queval et al., 2008). Our previous analyses have shown that global increases in tissue H2O2 are not detectable in cat2 growing under the conditions used in this report (Chaouch et al., 2010; Mhamdi et al., 2010). Therefore, to assess whether the inhibition of cell death in cat2 by MI was mediated via a purely or predominantly antioxidant effect, we measured leaf contents of glutathione and ascorbate, two biochemical markers of redox state, as well as two oxidative stress marker transcripts (Davletova et al., 2005; Vanderauwera et al., 2005). This analysis showed that the glutathione pool increased and was more oxidized in both water- and MI-treated cat2 plants. In fact, MI treatment increased rather than decreased glutathione oxidation (Fig. 2b). The ascorbate pool was not significantly affected by either the cat2 mutation or MI treatment (Fig. 2b). Compared with Col-0, GSTU24 and APX1 were induced in cat2 in both conditions, further supporting the conclusion that MI treatment abolished cell death without decreasing H2O2-triggered oxidative stress in cat2 (Fig. 2c). Control experiments showed that treating cat2 with sorbitol did not prevent lesion formation (data not shown).
MI mimics the sid2 mutation-induced repression of SA and SA-dependent cell death and defence responses
Since lesion formation in cat2 was SA-dependent and prevented by MI (Fig. 2a), we measured the effects of MI treatment on SA contents. Free and total SA were dramatically reduced in cat2 following MI treatment (Fig. 3a,b). Thus, MI decreased SA in cat2 in a similar way to the sid2 mutation (Fig. 1c,d). Moreover, MI treatment decreased ICS1 transcript levels (Fig. 3f), and also drastically down-regulated transcripts of the SA marker gene, PR1 (Fig. 3g). The pathogen defence-related molecules scopoletin and camalexin were also quantified. Induction of free and total scopoletin (Fig. 3d) in cat2 was prevented by MI treatment. The same effect was observed for camalexin (Fig. 3e) and for the transcripts of PAD3 (Fig. 3h), which encodes a cytochrome P450 involved in camalexin synthesis (Glawischnig, 2007; Nafisi et al., 2007). Thus, MI not only prevented H2O2-triggered activation of defence pathways, but actually decreased these to below the basal values observed in Col-0.
Mutations in one of the three genes for myo-inositol phosphate synthase (MIPS1) conditionally trigger lesions (Meng et al., 2009). However, quantification of MIPS1 transcripts showed that they were not affected by the cat2 mutation or by MI treatment (Fig. 3i). This indicates that the decrease in MI caused by oxidative stress is not linked to repression of MIPS1 function, at least at the level of transcript abundance.
MI abolishes resistance of cat2 to P. syringae DC3000
Because cat2 develops HR-like responses (Figs 1, 3), which confer resistance to a broad spectrum of pathogens, we tested the effect of the different treatments on the response of Col-0, cat2 and cat2 sid2 to the virulent bacterium P. syringae DC3000. This analysis confirmed that cat2 was more resistant than Col-0 and that sid2 mutation caused loss of that resistance (Chaouch et al., 2010; Fig. 4a). Like the sid2 mutation, MI treatment rendered cat2 plants as sensitive as water-treated cat2 sid2 but had no effect on either Col-0 or cat2 sid2 (Fig. 4b). Treating plants with a solution containing MI and SA rescued resistance lost in plants treated with MI alone and restored resistance to the values observed in water-treated Col-0, although bacterial growth remained somewhat higher than in water-treated cat2 (Fig. 4c). When Col-0 and cat2 sid2 were treated with SA only, resistance was restored to that observed in cat2 (Fig. 4d).
Numerous LMMs have been identified in Arabidopsis. The cat2 mutant is distinguished from many of these by the fact that effects are initiated by enhanced intracellular H2O2 availability. Although the H2O2 trigger in cat2 is peroxisomal rather than plasma membrane/apoplast-located, cell death and associated pathogen responses are completely dependent on SA accumulation through the isochorismate pathway (Chaouch et al., 2010). Nontargeted metabolite profiling of cat2 revealed that H2O2-triggered SA accumulation was associated with marked decreases in MI (Fig. 1). This provided a first indication that MI acts downstream of H2O2, and led us to examine the relationship between these two components and SA by analysing the effects of chemical complementation and the sid2 mutation on pathogen-associated components and resistance.
Treatment with MI abolishes SA-dependent cell death without decreasing oxidative stress
The sid2 mutation abolishes cell death and pathogen responses in cat2 without decreasing the severity of intra-cellular redox perturbation (Chaouch et al., 2010). The present study shows that lesion formation in cat2 can also be blocked by treatment with MI. This compound and the signalling pathways dependent on its phosphorylated derivatives (phosphoinositides) are important for many plant developmental and physiological processes (Murphy et al., 2008). Indeed, MI signalling is involved in responses to abscisic acid (Burnette et al., 2003; Lee et al., 2007) and abiotic stress (Takahashi et al., 2001). Recent studies of the MI synthesis pathway also point to roles in cell death (Murphy et al., 2008; Meng et al., 2009; Donahue et al., 2010). Lesions in the mips1 mutant are dependent on the isochorismate pathway (Meng et al., 2009), but relationships with ROS as well as with SA were not investigated. To establish whether MI inhibited the cat2 cell death phenotype by stopping oxidative stress or by acting downstream of it, we used two measures of intracellular redox state (glutathione and two H2O2-responsive marker genes). Analysis of a wide range of redox markers in cat2 has shown that these measurements provide a more sensitive indication of oxidative stress intensity than estimates of H2O2 using several techniques (Queval et al., 2008; Chaouch et al., 2010; Mhamdi et al., 2010). The responses of these markers to MI treatment of cat2 suggest that abrogation of cell death was accompanied by increased, rather than decreased, redox perturbation (Fig. 2b,c). Thus, although recent studies have reported that some functions of MI are linked to antioxidant pools (Nishizawa et al., 2008; Torabinejad et al., 2009), the present analysis suggests that this compound inhibits cell death and pathogen responses through mechanisms other than decreases in general cellular oxidative stress.
MI acts upstream of SA as a key switch that links oxidative stress to pathogen responses
Although MI did not act to decrease oxidative stress in cat2, it drastically dampened SA contents. The down-regulation of SA may partly occur at the transcriptional level, as evidenced by decreased ICS1 transcript abundance. Since ICS1 expression and SA contents in Col-0 were not affected by MI, the inhibition of SA accumulation below basal amounts requires the concomitant presence of oxidative stress. Given that SA accumulation in cat2 requires decreased MI, it appears that the latter compound acts as a switch that determines whether oxidative stress causes increases or decreases in SA. This would then be the cause of the antagonistic responses of SA-dependent factors (scopoletin, camalexin, PR1, PAD3, disease resistance) in cat2 in the absence or presence of added MI (Figs 3, 4).
The annulment of H2O2-induced SA accumulation and SA-dependent responses by MI, in the presence of continuing redox perturbation, suggests that MI is a key link between oxidative stress and SA. Indeed, simultaneously treating plants with SA and MI restored lesion formation and disease resistance in cat2. Interestingly, however, this restoration was only partial. This further underlines the requirement for decreased MI in order to trigger the full range of H2O2-triggered pathogen responses, but perhaps indicates that MI down-regulates some cell death-related signalling events that are upstream or independent of SA.
In conclusion, the present observations show, first, that redox perturbation induced by increased H2O2 availability acts to decrease MI contents. They therefore suggest that recent observations in mips1 mutants have considerable relevance to ROS-dependent cell death, and situate the primary action of MI downstream of H2O2 (Fig. 5). Like the sid2 mutation, MI treatment reverts cell death in cat2 without decreasing oxidative stress responses. The effects of MI on ICS1 expression, SA contents and SA-dependent responses show that this compound acts antagonistically to SA to govern the outcome of increased intracellular H2O2 (Fig. 5). When oxidative stress occurs in the absence of increases of SA (cat2 sid2) or decreases in MI (cat2 + MI), at least some pathogen defence responses are decreased below basal Col-0 values (Figs 3, 4a,b). This suggests that some components of intracellular oxidative stress may act to antagonize activation of pathogen responses (Fig. 5, dotted line). Although decreased MI is necessary to allow peroxisomal H2O2 to induce SA and pathogen responses, it does not appear to be sufficient. This is evidenced by decreased MI contents in cat2 sid2, in which oxidative stress no longer triggers lesions (Fig. 1). An additional light signal may be required to link H2O2-triggered decreases in MI to activation of SA synthesis, cell death and pathogen responses. This notion is consistent with the conditional nature of the mips1 mutant (Meng et al., 2009). More research is required to elucidate these interactions further.
We thank Patrick Saindrenan and Shengchun Li (IBP, Orsay) for providing HPLC facilities and help with RNA extractions, respectively. This work was supported by a Centre National de la Recherche Scientifique PhD fellowship (S.C.) and a French Agence Nationale de la Recherche-GENOPLANTE (no. GNP0508G) grant to G.N.