Jasmonic acid (JA) and methyl jasmonate (MeJA), collectively termed jasmonates, are ubiquitous plant signalling compounds. Several types of stress conditions, such as wounding and pathogen infection, cause endogenous JA accumulation and the expression of jasmonate-responsive genes. Although jasmonates are important signalling components for the stress response in plants, the mechanism by which jasmonate signalling contributes to stress tolerance has not been clearly defined. A comprehensive analysis of jasmonate-regulated metabolic pathways in Arabidopsis was performed using cDNA macroarrays containing 13516 expressed sequence tags (ESTs) covering 8384 loci. The results showed that jasmonates activate the coordinated gene expression of factors involved in nine metabolic pathways belonging to two functionally related groups: (i) ascorbate and glutathione metabolic pathways, which are important in defence responses to oxidative stress, and (ii) biosynthesis of indole glucosinolate, which is a defence compound occurring in the Brassicaceae family. We confirmed that JA induces the accumulation of ascorbate, glutathione and cysteine and increases the activity of dehydroascorbate reductase, an enzyme in the ascorbate recycling pathway. These antioxidant metabolic pathways are known to be activated under oxidative stress conditions. Ozone (O3) exposure, a representative oxidative stress, is known to cause activation of antioxidant metabolism. We showed that O3 exposure caused the induction of several genes involved in antioxidant metabolism in the wild type. However, in jasmonate-deficient Arabidopsis 12-oxophytodienoate reductase 3 (opr3) mutants, the induction of antioxidant genes was abolished. Compared with the wild type, opr3 mutants were more sensitive to O3 exposure. These results suggest that the coordinated activation of the metabolic pathways mediated by jasmonates provides resistance to environmental stresses.
When plants undergo stress such as wounding or pathogen infection, certain secondary metabolites, including defence compounds, accumulate. Several secondary metabolites such as terpenoid indole alkaloids, indole glucosinolate, nicotine alkaloids, and polyamines are known to accumulate through the induction of biosynthetic genes by jasmonates (Brader et al., 2001; Goossens et al., 2003; Memelink et al., 2001). In the Madagascar periwinkle, Catharanthus roseus, several genes involved in terpenoid indole alkaloid biosynthesis have been cloned, and MeJA was found to induce the expression of all such genes tested (van der Fits and Memelink, 2000; Geerlings et al., 2000). MeJA also induces genes involved in primary metabolism, leading to the formation of tryptophan derivatives, which are terpenoid indole alkaloid precursors (van der Fits and Memelink, 2000). These results suggest that stress tolerance in plants requires jasmonate-mediated transcriptional activation of biosynthetic pathways for defence compounds.
Previously, we screened JRGs in Arabidopsis using cDNA macroarrays and showed that MeJA induces a set of JA biosynthetic genes, including lipoxygenase 2 (LOX2), allene oxide synthase (AOS), allene oxide cyclase (AOC) and 12-oxo-phytodienoic acid reductase 3 (OPR3) (Sasaki et al., 2001). Also, the tryptophan biosynthetic genes anthrenilate synthase α-chain1 (ASA1), anthranilate synthase β-chain1 (ASB1), tryptophan synthase α-chain1 (TSA1) and tryptophan synthase β-chain1 (TSB1) are induced by MeJA (Sasaki et al., 2001). These results indicate that jasmonates activate particular metabolic pathways via transcriptional activation of a set of metabolically related gene groups such as terpenoid indole alkaloid and indole glucosinolate. However, our current knowledge of jasmonate-responsive metabolic pathways is limited mostly to secondary metabolism, and thus the overall role of JRGs in stress tolerance remains largely unknown.
We performed a comprehensive analysis of jasmonate-regulated metabolic pathways using cDNA macroarrays containing 13516 expressed sequence tags (ESTs) covering 8384 loci. This approach identified nine metabolic pathways belonging to two functionally related groups: (i) ascorbate and glutathione metabolic pathways, which are important in defence responses to oxidative stress, and (ii) biosynthetic enzymes for indole glucosinolate, which is a defence compound in the Brassicaceae family. Although these metabolic pathways are involved in tolerance of oxidative stresses including pathogen infection and O3 exposure (May et al., 1998; Pignocchi and Foyer, 2003; Wittstock and Halkier, 2002), the mechanism by which these pathways are regulated is unknown. We propose that the coordinated activation of these metabolic pathways, mediated by jasmonates, provides stress tolerance in Arabidopsis.
Results and discussion
Identification of JRGs and classification by their metabolic functions
To obtain gene expression profiles in response to jasmonates, 10-day-old Arabidopsis seedlings were treated with 30 μm JA or MeJA. Total RNA was isolated at successive time points (0, 15, 30, 60, 180 and 360 min post-treatment), 33P-labelled by reverse transcription, and hybridized separately onto cDNA macroarray membranes containing 13516 ESTs.
We used three or four biologically independent RNA samples for each time point of JA or MeJA treatment and obtained a set of raw data (see Minimum information about a microarray experiment, MIAME). Procedures for normalization are described in the Experimental procedures section. We evaluated the correlation coefficient of the normalized gene expression values (NEetr; see Experimental procedures for definition) of an experiment versus that of other replicates at a time point (e.g. the correlation coefficient of NE values of the first experiment at time 0 versus that of the second experiment at time 0 was calculated). We calculated such correlation coefficients for all combinations at a time point. These correlation coefficients are shown in Supplementary Table S1. The averages of correlation coefficients among replicated experiments at each time point were more than 0.84, showing that the reproducibility of our cDNA macroarray data was high enough for further analysis. Then, NEetr values were combined by weighted average among the three or four replicate experiments (see Experimental procedures). The NEetr values were then transformed into data for the corresponding locus (NEt) as described in Obayashi et al. (2004). The NEt values are listed in Supplementary Table S2. When we calculated NEt, it was found that several ESTs, which corresponded to a locus, showed different expression profiles depending on experimental error such as background noise. Thus we calculated the maximum standard deviation (MaxSD) to evaluate the fluctuation of expression profile among the three or four replicate experiments (see Experimental procedures). In this step, we eliminated 278 loci with MaxSD > 0.301 (≈ base-10 logarithmic value of 2), which showed low reproducibility expression profiles of corresponding ESTs, for further analysis.
To screen 8106 loci for genes that responded to JA or MeJA, we calculated maximum-fold (MF) values, which indicate the maximum value of gene expression change (see Experimental procedures and Table S2). MF values for JA (MFJA) and MeJA treatments (MFMeJA) were marked on a scatter plot (Supplementary Figure S1). The correlation coefficient of JA-responsive loci, which had an MFJA of more than 2 or less than 0.5 versus MeJA-responsive loci, was 0.51, indicating a positive correlation between the effects of JA and MeJA on gene expression. To identify genes whose expression was affected by both JA and MeJA, avMF was calculated as the geometric mean of MFJA and MFMeJA (see Experimental procedures and Table S2). When we selected genes that showed an avMF of more than 2 or less than 0.5, 253 (149 genes induced and 104 repressed) of 8106 genes fulfilled this criterion. Because an avMF is an averaged value, it tends to be lower than MFJA or MFMeJA (Supplementary Table S2). However, this strict criterion is useful for elimination of pseudopositives. The MF values of these 253 genes are presented in Supplementary Figure S1; they are distributed around the diagonal line, thus confirming a positive correlation (r = 0.86) between JA and MeJA responses.
We defined the 253 JRGs in this paper (Supplementary Table S3) and categorized or characterized them in terms of known metabolic pathways using AraCyc, a database of Arabidopsis biochemical pathways (http://arabidopsis.org/tools/aracyc/). Forty-six of these 253 JRGs were mapped in AraCyc (Table 1). Thirty-eight genes listed in Table 1 were up-regulated by jasmonates. We classified them according to their function and they were assigned to nine metabolic pathways (Table 1). Consistent with our previous report (Sasaki et al., 2001), JA and tryptophan biosynthesis pathways were amongst these pathways. In addition, we observed the induction of genes involved in seven other pathways: serine biosynthesis, sulphur assimilation, cysteine biosynthesis, glutathione (GSH) biosynthesis, indole glucosinolate biosynthesis, ascorbate (AsA) biosynthesis, and AsA recycling. Eight genes were down-regulated by jasmonates. They were not involved in the nine metabolic pathways, and did not fall into any particular metabolic category. Thus we placed the eight down-regulated genes in a category ‘not classified’ (Table 1).
Table 1. Classification of 46 JRGs according to categories defined by AraCyc, a tool for visualizing Arabidopsis biochemical pathways. NE0 indicates basal expression level of each gene. NE0 = 0 indicates the median of all gene expression level. MF indicates maximum value of gene expression change after JA or MeJA treatment. avMF indicates the geometric mean of MFJA and MFMeJA
Then we focused on the nine metabolic pathways containing genes up-regulated by jasmonates. To visualize the individual gene expression profiles involved in these jasmonate-inducible metabolic pathways, we generated a metabolic map combined with line graphs of gene expression profiles (Figures 1 and 4). Because all genes in the metabolic pathways could not be identified in AraCyc, further information from the literature was also incorporated in Figures 1 and 4. To verify the induction by jasmonates of several genes involved in those metabolic pathways, we confirmed their expression profiles using Northern blot analyses. Consistent with the cDNA macroarray data, the expression of genes gamma-glutamyl-cysteine synthetase (GSH1; At4g23100), glutathione synthetase (GSH2; At5g27380), dehydroascorbate reductase (DHAR; At1g19570), monodehydroascorbate reductase (MDHAR; At3g09940), vitamin c-1 (VTC1; At2g39770), vitamin c-2 (VTC2; At4g26850) and VTC2-homologue (At5g55120) was increased by treatment with JA (Figure 2). In this experiment, we used biologically independent RNA from that used in the cDNA macroarray analysis. These results show that our cDNA macroarray conditions and criteria were sufficient to select JRGs.
Sulphur metabolic pathway
Among the 253 JRGs, phosphoserine aminotransferase (PSAT; At4g35630), ATP-sulphurylase 3 (APS3; At4g14680), adenosine phosphosulphate kinase-related gene 1 (AKN1; At2g14750), serine acetyltransferase 2;2 (Serat2;2; At3g13110), and cysteine synthase (CS; At3g59760) showed clear induction by jasmonates (Figure 1a and Supplementary Table S4). APS1 (At3g22890), AKN2 (At4g39940), 5’-adenylylsulphate reductase 1 (APR1; At4g04610) and sulphite reductase (SIR; At5g04590) had modest responses to jasmonates (avMF > 1.7; Figure 1a and Supplementary Table S4). All of these gene products are predicted to localize to chloroplasts (Supplementary Table S4), suggesting the co-regulation of sulphur assimilation and serine and cysteine biosynthesis pathways in chloroplasts. We also checked the organ-specific expression of these JRGs using cDNA macroarray data obtained from Obayashi et al. (2004). However, no obvious co-expression pattern was observed (Supplementary Table S4). GSH1 (At4g23100) and GSH2 (At5g27380) were also induced by jasmonates, as reported previously (Xiang and Oliver, 1998) (Figures 1a and 2).
As the jasmonate treatments (i.e. JA or MeJA treatment) induced the accumulation of mRNAs for the genes involved in most steps of the sulphur metabolic pathways, we examined whether jasmonates elevate the endogenous levels of thiol compounds. Total cysteine and GSH content clearly increased following 30 μm JA treatment (Figure 3a,b). We also performed the treatment with 200 μm JA. This treatment caused the accumulation of 3-fold more cysteine and 1.8-fold more GSH compared with controls after 72 h. The statistic significance of accumulation of these compounds was confirmed by Student's t-test (P < 0.05). These results demonstrate that JA activates sulphur assimilation and thiol compound biosynthetic pathways. Interestingly, several JA biosynthesis and sulphur assimilation genes are induced under sulphur-deficient conditions (Hirai et al., 2003; Nikiforova et al., 2003). Thus, it is possible that jasmonates function as signals for the induction of sulphur assimilation under these conditions.
Indole glucosinolate biosynthetic pathway
We also found that jasmonates induce genes involved in the indole glucosinolate biosynthetic pathway, some of which have previously been identified as MeJA-responsive genes (Brader et al., 2001). As shown in Table 1 and Supplementary Table S3, genes involved in indole glucosinolate biosynthesis were amongst the 253 JRGs. Indole glucosinolates are thought to be sulphur-containing defence compounds against generalist herbivores and probably also against pathogens, as they accumulate following pathogen infection (Brader et al., 2001; Wittstock and Halkier, 2002). To visualize the individual gene expression profiles involved in the indole glucosinolate biosynthetic pathway, we again generated a metabolic map combined with line graphs of gene expression profiles (Figure 4). In this pathway, indole-3-glycerol phosphate synthase (IGPS; At2g04400), tryptophan synthase alpha subunit (TSA1; At3g54640), tryptophan synthase beta subunit (TSB1; At5g54810), cytochrome P450 79B2 (CYP79B2; At4g39950), cytochrome P450 83B1 (CYP83B1; At4g31500), cystine lyase (Cslyase; At2g20610 and At4g23600) and UDP-glycosyltransferase (S-GT; At1g24100) were induced by jasmonates (Figure 4 and Supplementary Table S4). Consistent with our results, it has been reported that MeJA induces several genes encoding biosynthetic enzymes for the glucosinolate core structure, resulting in the accumulation of indole glucosinolates (Brader et al., 2001). Furthermore, our macroarray data revealed that serine, cysteine and 3′-phosphoadenosine 5′-phosphosulphate (PAPS) biosynthetic genes are simultaneously induced by jasmonates (Figure 1a). Serine is incorporated into the indole moiety during tryptophan biosynthesis. Cysteine and PAPS are also incorporated into the core structure of glucosinolates (Wittstock and Halkier, 2002) (Figure 4). These results reveal that jasmonates coordinately induce the biosynthesis of the indole glucosinolate core structure and secondary adducts.
AsA biosynthetic and recycling pathway
AsA, as well as GSH, is a key compound of the plant antioxidant system. A novel AsA biosynthetic pathway in plants has recently been proposed (Wheeler et al., 1998). However, the genes involved in this pathway have not yet been completely elucidated, and little is known about the control of AsA biosynthesis. We showed that three genes are induced by jasmonates, namely VTC1, which encodes GDP-mannose pyrophosphorylase (At2g39770), VTC2 (At4g26850), the mutation of which causes decreased levels of endogenous AsA (Conklin et al., 2000), and VTC2-homologue (At5g55120) (Figures 1b and 2, and Supplementary Table S4).
AsA is the major redox compound in plants. When reactive oxygen species (ROS) are generated in plant cells, AsA reduces ROS and, in turn, becomes oxidized. Oxidation of AsA produces the short-lived radical monodehydroascorbate (MDHA), which is converted to AsA by MDHAR or is disproportionated non-enzymatically to AsA and dehydroascorbate (DHA). DHA is recycled to AsA by DHAR using GSH as the reductant (Figure 1c). The activation of the AsA recycling pathway is important for increasing the cellular content of AsA (Chen et al., 2003). Thus, we also focused on the AsA recycling pathway. We found that jasmonates induce MDHAR (At3g09940) and DHAR (At1g19570) (Figures 1c and 2, and Supplementary Table S4). To clarify whether jasmonates activate these enzymes, we measured the activities of enzymes involved in AsA recycling after JA treatment. DHAR activity increased (1.8-fold greater than the control value) 24 h after treatment with 30 μm JA and remained elevated up to 48 h (1.6-fold greater than the control value; Figure 3c). The statistical significances of the increases in these enzymatic activities were confirmed by Student's t-test (P < 0.05). We also measured ascorbate peroxidase (APX) activity. Although the activity did not show a marked change (maximum change 1.3-fold greater than the control value), the slight increase was statistically significant (P < 0.05) after JA treatment (Figure 3d), implying that JA signalling contributes to activation of enzymes involved in scavenging for H2O2. MDHAR activity was not induced despite the clear induction of its mRNA by jasmonates (data not shown).
To confirm the positive effects of jasmonates on AsA biosynthetic and recycling pathways, we then measured the level of reduced AsA after JA treatment. The level of endogenous AsA increased after JA treatment (both 30 and 200 μm) and reached a level that was 2.8-fold higher than the control after 48 h (Figure 3e). The statistic significance of accumulation of reduced AsA was confirmed by Student's t-test (P < 0.05). The experiment was repeated using a photometric method (Sakaki et al., 1983), and similar results were obtained (data not shown). The above results demonstrate that JA simultaneously activates AsA and GSH metabolism (Figure 3). These compounds are part of the AsA–GSH cycle and provide crucial protection against oxidative damage (Noctor and Foyer, 1998). As in the case of genes involved in sulphur assimilation, genes involved in AsA biosynthesis and recycling did not show an obvious organ-specific expression profile. However, it should be noted that targetp (Available at http://www.cbs.dtu.dk/services/TargetP/; The Center for Biological Sequence Analysis, Technical University of Denmark, Lyngby, Denmark), a tool for predicting subcellular localization, indicated that jasmonate-responsive DHAR (At1g19570) and MDHAR (At3g09940) have no distinct transit peptide, suggesting that the DHAR and MDHAR localize outside the major organelles such as plastids and mitochondria (Supplementary Table S4). In contrast, chloroplast isoforms were not induced by jasmonates (Figure 1c and Supplementary Table S4). It is likely that jasmonates regulate outer organelle types of antioxidant enzymes. In this study, we measured the enzymatic activities of DHAR and MDHAR in crude extracts of whole seedlings. Although five homologues of MDHAR exist in the Arabidopsis genome, one of the five was jasmonate-responsive MDHAR. The subcellular localization of jasmonate-responsive MDHAR has not been elucidated. This might explain why an increase of MDHAR activity was not observed. It is important to clarify the localization of jasmonate-responsive antioxidant enzymes in future work. DHAR (At1g19570), which responded to jasmonates in this experiment (Figure 1c), is induced by chemical treatments such as norflurazon and antimycin A, which cause oxidative stress (Chew et al., 2003). Furthermore, the regulation of the redox state of antioxidants is a key process in several aspects of oxidative stress (Pignocchi and Foyer, 2003). These findings suggest that extraplastidic antioxidant enzymes are involved in scavenging of ROS under oxidative stresses, including O3 stress.
Jasmonate biosynthesis is required for O3 tolerance
Oxidative stress results from environmental factors promoting the formation of ROS which damage cells. Such factors causing oxidative stress include O3, drought, heat and cold, and wounding. Jasmonates accumulate after wounding and have a pivotal role in the wounding response (León et al., 2001; Reymond and Farmer, 1998). O3 exposure also increases JA levels in Arabidopsis (Kanna et al., 2003; Rao et al., 2000). Several reports suggest that JA signalling may be involved in the containment of O3-induced lesions (Kanna et al., 2003; Overmyer et al., 2000; Rao and Davis, 2001; Rao et al., 2000). Despite a number of reports supporting this hypothesis, the mechanism of O3 tolerance induced by jasmonates has not been clarified. Several reports suggest a relationship between O3 tolerance and JA biosynthesis. For example, biosynthetic mutants of jasmonates, such as fatty acid desaturase 3-2/7-2/8 (fad3-2/7-2/8) and jasmonic acid desaturase 1-1 (jar1-1), display an O3-sensitive phenotype (Overmyer et al., 2000; Rao et al., 2000). However, fad3-2/7-2/8 lacks not only precursors for JA biosynthesis (McConn and Browse, 1996) but also other downstream fatty acid-derived signalling compounds such as oxylipins (Farmer et al., 2003) (Figure 5a). JAR1 encodes the enzyme that conjugates isoleucine (Ile) to JA (Staswick and Tiryaki, 2004) (Figure 5a). Although jar1-1 shows moderate insensitivity to MeJA, it has normal male fertility, in contrast to other JA biosynthesis or signalling mutants (Staswick et al., 1992). Thus, JAR1 is not required for all jasmonate responses (Staswick et al., 2002). To verify the requirement for jasmonate signalling in O3 tolerance, we used opr3 mutants. OPR3 catalyses the middle step in JA biosynthesis (Figure 5a). OPR3 knockout mutants show a male sterile phenotype and do not accumulate JA, and thus the application of JA completely restores the phenotype (Stintzi and Browse, 2000; Stintzi et al., 2001). It has been reported that O3-induced lesion formation correlates with the amount of AsA (Pignocchi and Foyer, 2003). vtc1 plants, which have decreased levels of AsA, are sensitive to O3 (Conklin et al., 1996). In Populus leaves, the total GSH content increases after O3 fumigation (Sen Gupta et al., 1991); also, oxidative stress induces GSH in Arabidopsis suspension cultures (May and Leaver, 1993). Aono et al. (in press) reported that DHAR (At1g19570), which responded to jasmonates in the present study, is induced by O3 exposure. Furthermore, AsA levels in wild-type Arabidopsis were elevated during 6 h of O3 exposure (Aono et al., in press). We showed that jasmonates induced the expression of genes involved in these antioxidant metabolic pathways (Figures 1 and 2) and the accumulation of their metabolites (Figure 3). To verify the hypothesis that jasmonate signalling regulates antioxidant metabolism under O3 stress, we compared gene expression involved in antioxidant metabolism between wild type [Wassilewskija (WS)-2 background] and opr3 during O3 exposure [200 nl l−1 O3 under a photosynthetic photon flux density (PPFD) of 350 μmol photons m−2 s−1 of continuous light]. OPR3 was clearly induced by O3 exposure in WS-2 (Figure 5b), while the expression of OPR3 in the opr3 mutant was detected neither under normal conditions nor during O3 exposure (Figure 5b). GSH1 (At4g23100), DHAR (At1g19570), VTC1 (At2g39770), and VTC2-homologue (At5g55120) were also induced by 6 h O3 exposure in WS-2 (1.4-, 2.7-, 1.5- and 2.5-fold higher than at 0-h treatment, respectively); however, induction of these genes was abolished in opr3 [the expression levels of these genes induced by 6 h O3 exposure were only 1.0-fold (GSH1), 1.1-fold (DHAR), 1.1-fold (VTC1) and 1.3-fold (VTC2-homologue) the expression level at 0-h treatment, respectively (Figure 5b)]. These results show that these antioxidant metabolism genes require jasmonates to induce their expression under O3 stress. GSH2 (At5g27380), MDHAR (At3g09940) and VTC2 (At4g26850) were not induced by O3 exposure (1.0-, 0.9- and 0.9-fold the expression level for 0-h treatment in WS-2, respectively).
Our results suggest that defects of jasmonate signalling in opr3 disrupt activation of antioxidant metabolism (Figures 1–3 and 5b). To estimate the O3 sensitivity of opr3, 16-day-old WS-2 and opr3 seedlings were exposed to the same O3 exposure conditions as that of Northern hybridization. Three hours after the initiation of O3 exposure, the distal part of opr3 leaves began to wilt (data not shown). The wilting area of the leaves then spread inwards after 6 h of exposure (Figure 5c), and 84% of the leaves from O3-exposed seedlings were wilted. In contrast, in O3-exposed WS-2 seedlings, only 24% of leaves were wilted, and thus 76% of leaves were tolerant to O3 exposure (Figure 5c). After 6 h exposure the seedlings were transferred to fresh air for another 24 h. Finally, lesions formed in the leaves of opr3 (data not shown). Both WS-2 and opr3 in fresh air have few wilted leaves (0 and 5 %, respectively) (Figure 5c).
A clear spatial and quantitative correlation between ROS accumulation and formation of O3-induced lesions was found in several plant species (Overmyer et al., 2003). Thus, levels of H2O2, which is scavenged by the successive reactions of the AsA–GSH cycle, in O3-exposed WS-2 and opr3 leaves were visualized by 3, 3-diaminobenzidine (DAB) staining. Light brown coloration produced by DAB staining, indicating H2O2 accumulation, was observed only in and around the wilting region of opr3 leaves but not in WS-2 leaves; however, little difference in the coloration of the whole leaf between WS-2 and opr3 was detected (data not shown). Although the generation of H2O2 during O3 exposure has been reported in Arabidopsis (Rao et al., 2000), direct visualization of H2O2 generation in the opr3 mutant was difficult, probably because of the relatively low H2O2 concentration, compared with wound-induced accumulation, for example, under O3 stress. We then used ion leakage, which directly reflects the magnitude of ROS generation in the cell, as a quantitative indicator of plasma membrane damage. The damages of plasma membrane were involved in extent of cell death with visible lesions (Overmyer et al., 2000). In this study, lesions were formed in wilted leaves when the plants were exposed to O3 for 6 h and to fresh air for another 24 h (data not shown). Thus, the wilting phenotype results from ROS accumulation, which is involved in high levels of ion leakage. Ion leakage in O3-exposed opr3 leaves was 23-fold higher than that measured after fresh air exposure (Figure 5d). In contrast, wild-type leaves exposed to O3 showed the same level of ion leakage as those in fresh air (Figure 5d). These results show that jasmonate, which is generated after reduction of a cyclopentenone structure by OPR3, is required for O3 tolerance. The level of ion leakage in opr3 in fresh air was lower than that in the wild type (Figure 5d). It is difficult to interpret the difference in ion leakage level in fresh air between the wild type and opr3, because the basal level of ion leakage in opr3 is almost undetectable. Defects in jasmonate biosynthesis might affect the homeostasis of the plasma membrane.
The role of jasmonate-responsive metabolic pathways in stress responses
Here, we identified 253 JRGs using cDNA macroarrays. Forty-six of these were assigned to nine known metabolic pathways (Table 1), among which six are involved in AsA and GSH metabolism. These antioxidants are critical for stress tolerance. In addition to AsA and GSH, there are several other known antioxidants, such as ß-carotene, polyamine and tocopherol. Although arginine decarboxylases, which are involved in polyamine biosynthesis, are also induced by jasmonates (Table 1), we did not observe significant induction of other antioxidant metabolic genes (Supplementary Table S3 and MIAME). Thus, the AsA and GSH pathways are the major antioxidant metabolic pathways regulated by jasmonates at the transcriptional level, and their coordinated regulation is involved in O3 tolerance. Enzymes involved in extraplastidic AsA metabolism also play important roles in O3 tolerance (Pignocchi and Foyer, 2003). Induction of DHAR, VTC1 and VTC2-homologue by O3 exposure depends on jasmonate signalling (Figure 5b). APX is the most important enzyme in the AsA–GSH cycle for scavenging ROS (H2O2). It has been reported that the expression of cytosolic APX (At1g07890) is induced by O3 exposure (Kubo et al., 1995). Although jasmonate treatment did not induce the expression of cytosolic APX, APX activity after JA treatment was slightly increased (Figure 3d). This suggests that synergetic responses of both JA-dependent and JA-independent pathways are important in O3 tolerance.
Furthermore, GSH synthesis is activated by oxidative stress, and GSH is considered to be a mediator of ROS signalling in many species (May et al., 1998). O3 stress induces ROS production (the oxidative burst). An O3-induced oxidative burst and prolonged ROS accumulation result in the activation of cell death (Overmyer et al., 2003). The extent of O3-dependent lesion formation seems to be controlled by hormones such as salicylic acid and ethylene. In contrast, JA is involved in the containment of lesion propagation and is thought to be a negative regulator of cell death (Kanna et al., 2003; Overmyer et al., 2003; Rao and Davis, 2001). Thus, it is likely that the coordinated activation of antioxidant metabolism through jasmonate signalling is a crucial step for the containment of cell death (Figure 6), although the possibility of jasmonate-mediated activation of another mechanism that contains cell death under O3 stress, such as suppression of O3-induced ethylene emission (Kanna et al., 2003), cannot be excluded.
In addition to AsA and GSH metabolism, jasmonates also activated the biosynthesis of defence compounds such as indole glucosinolate (Figure 4 and Supplementary Table S4), which accumulates in pathogen-infected plants (Brader et al., 2001). The expression of CYP79B2 (At4g39950), which encodes a key enzyme of indole glucosinolate biosynthesis, is induced by wounding (Mikkelsen et al., 2000). These stresses promote the formation of ROS, resulting in oxidative stress. Several oxidative stresses are known to stimulate JA accumulation (Wasternack and Hause, 2002). Here, we also found that jasmonates simultaneously induced biosynthetic genes for serine, cysteine and PAPS (Figure 1a), compounds that are thought to be incorporated into the indole glucosinolate core structure (Wittstock and Halkier, 2002). Thus, it is likely that co-regulation of these metabolic pathways by jasmonates causes the marked accumulation of indole glucosinolate.
Myrosinase-related components were also shown to be regulated by jasmonates (Table 1). Intact glucosinolates are non-toxic; however, when plants are damaged, these compounds are hydrolysed and rearranged into toxic compounds such as isothiocyanates and nitriles by the myrosinase complex. This complex is composed of myrosinase, myrosinase-binding protein and myrosinase-associated protein (Rask et al., 2000). It is well known that homologues of the myrosinase-binding protein are induced by wounding or MeJA (Geshi and Brandt, 1998; Taipalensuu et al., 1997), and the same is true for myrosinase-associated protein (Taipalensuu et al., 1996). In this study, a number of homologues of these proteins were included in the cDNA macroarray, and two of three putative myrosinase-binding proteins and three of five putative myrosinase-associated proteins were induced by jasmonates (Supplementary Table S2). From these results we surmise that jasmonates strictly regulate the glucosinolate-myrosinase systems at the mRNA level.
Recently, Reymond et al. (2004) reported that DHAR (At1g19570) and MDHAR (At3g09940) are induced by pathogen infection, suggesting that AsA–GSH metabolism is also activated by jasmonate signalling during pathogen-induced stress. Pathogen infection causes an oxidative burst and cell death, and the molecular mechanism of this phenomenon is similar to that for O3-induced cell death (Overmyer et al., 2003). Thus, under oxidative stresses, jasmonate signalling presumably regulates defensive metabolism in cooperation with the antioxidant metabolism pathway via transcriptional activation (Figure 6).
Little is known about the mechanism(s) by which these metabolic pathways are regulated, and thus it will be informative to isolate mutants that lack the ability to activate these pathways. Elucidation of the regulatory components of this concerted response via mutant analyses will greatly enhance our understanding of cellular responses to environmental stresses in plants. Beyond the nine jasmonate-responsive metabolic pathways we identified, a number of jasmonate-responsive genes remain to be classified. The elucidation of these genes will complete our understanding of jasmonate-responsive metabolic pathways, thereby facilitating our ability to predict how plants will respond to a variety of biotic and abiotic stresses.
Seeds of Arabidopsis (accession Columbia) were germinated in Murashige–Skoog medium containing 1% sucrose. Seedlings were incubated on an orbital shaker under continuous light at 22°C. After 10 days, the plants were treated with 30 μm JA or MeJA. Identical growth conditions were employed for plants that were used for quantification of thiol compounds, reduced ascorbate and enzymatic activity.
cDNA macroarray design
The 13516 EST clones (Asamizu et al., 2000) were spotted onto nylon filters as described previously (Sasaki et al., 2001). EST clones were assigned to Arabidopsis loci using blast (Available at http://www.ncbi.nih.gov/; National Center for Biotechnology Information, Bethesda, MD, USA). Although these ESTs were clustered as non-redundant EST clones (Asamizu et al., 2000), 13516 ESTs were assigned to 8384 loci, and about 2000 loci had multi-EST clones.
Radiolabelling, hybridization, and image analysis of cDNA macroarrays
Total RNA was prepared from JA- or MeJA-treated 10-day-old Arabidopsis seedlings and further purified using the RNeasy Mini kit (Qiagen, Valencía, CA, USA) as described previously (Obayashi et al., 2004). Radiolabelling was performed as described by Matsumoto et al. (2004), with a slight modification. Using total RNA (10 μg per membrane) as a template, target DNA was labelled by reverse transcription in the presence of [α-33P]dATP and 0.5 μg of oligo-dT16−18 primer using the SuperScript First-Strand Synthesis System (Invitrogen, Valencía, CA, USA) according to the manufacturer's instructions. The labelled cDNA was denatured and used as ‘target’ DNA for hybridization. Hybridization with the labelled target was performed in the presence of 0.5 m Na2HPO4 (pH 7.2), 1 mm ethylenediaminetetraacetic acid (EDTA) and 7% sodium dodecyl sulphate (SDS) (Church and Gilbert, 1984) at 65 °C for 16 h. The membranes were then washed twice with 0.2× saline sodium citrate (SSC) containing 0.1% SDS at 65 °C and exposed to an imaging plate (Fuji Film, Tokyo, Japan) for 1–3 days under a shield box made of lead to reduce the effect of naturally occurring background radiation. This shield box improved the quality of the raw images, which were obtained using a high-resolution scanner (Storm, Amersham, NJ, USA). Signal intensity was quantified using Array Vision software (version 6.0; Amersham Biosciences).
Evaluation of data quality and data normalization
Quality evaluation and normalization of cDNA macroarray data were performed as described previously (Obayashi et al., 2004) with slight modifications in the data normalization procedures. Lowess normalization (Quackenbush, 2002) was used to reduce experimental bias after background subtraction and median centring.
where NEetr(BC) represents base-10 logarithm values of the normalized data corresponding to the spot intensity of an EST clone e, at a time point t, for repetition r.
We used 13 516 ESTs, six time points (0, 15, 30, 60, 180 and 360 min post-treatment with JA or MeJA), and three or four cDNA macroarray analyses using biologically independent RNAs. Data of varying quality were combined in these repetitions. To obtain the maximal potential information, NEetr values were combined by their weighted average at each time point. The Signal/Noise (S/N) value of each cDNA macroarray was used as the weight (Obayashi et al., 2004). Lowess normalization was achieved using the statistical language R (http://www.r-project.org/). The normalized data for ESTs were then transformed to data for the corresponding locus as described by Obayashi et al. (2004). The resultant value is NEt for a gene (locus).
Criterion for a locus with low reproducibility
We evaluated fluctuations among repetitive experiments to exclude pseudopositive expression profiles. First, we calculated NElevel for each EST in each experiment. NElevel,er is the average of the NEetr values at six time points (0, 15, 30, 60, 180 and 360 min).
where Nt represents the number of time points in the experiment. Each NEetr value was normalized by NElevel,er and the standard deviation of the resultant value among repetitive experiments was calculated as follows:
where Nr represents the number of repetitions. The maximum SDe in each EST was calculated as follows:
when multi-ESTs corresponded to one locus, we averaged the MaxSDe values. Finally, we defined this value for a locus as MaxSD. A total of 278 loci with MaxSD > 0.301 (≈base-10 logarithmic value of 2) were excluded from later analyses.
Selection of jasmonate-responsive genes
To screen genes that responded to JA or MeJA, we calculated the difference betweenNE at each time point (15, 30, 60, 180 and 360 min post-treatment) and the 0-min value (non-treated). We defined this value as relative expression (RE):
The value of maximum-fold (MF) was then calculated as follows:
Examples of log10(MF) values in cases (a) and (b) are shown in Supplementary Figure S2.
To compare the effects of JA and MeJA on gene expression, the MF values of JA treatment (MFJA) and MeJA treatment (MFMeJA) were calculated for each gene. MFJA and MFMeJA values were plotted on a scatter plot [MFJA and MFMeJA are the antilogarithms of log10(MFJA) and log10(MFMeJA); Supplementary Figure S1]. We selected genes having an MFJA or MFMeJA value of more than 2 or less than 0.5. The correlation coefficient of these MFJA values versus MFMeJA was 0.51, indicating a positive correlation between the effects of JA and MeJA on gene expression. To identify genes whose expression was affected by both JA and MeJA, avMF was calculated as the geometric mean of MFJA and MFMeJA:
when we selected genes that showed an avMF of more than 2 or less than 0.5. A total of 253 genes (149 genes induced and 104 repressed) fulfilled this criterion. The MF values of these genes are presented in Supplementary Figure S1; they are distributed around the diagonal line, thus confirming a positive correlation (r = 0.86) between JA- and MeJA-induced responses. We defined these 253 genes as JRGs in this paper.
Prediction of localization of gene products
We used the program targetp (Emanuelsson et al., 2000) to predict the subcellular localization of gene products and then estimated the location of the jasmonate-responsive metabolic pathways we identified. This program predicts subcellular localization of proteins based on their N-terminal amino acid sequence, which discriminates among proteins destined for mitochondria, chloroplasts, the secretory pathway and ‘other’ localizations.
Assignment of the 253 JRGs to biochemical pathways using AraCyc
To categorize the 253 JRGs according to their known function or annotation, we used AraCyc, a tool for visualizing Arabidopsis biochemical pathways (available at http://www.arabidopsis.org/tools/aracyc/). Further information on genes in jasmonate-responsive metabolic pathways was obtained from the literature and from ATTED, a database with utilities for array analysis (http://www.atted.bio.titech.ac.jp/).
Northern blot analyses
Five micrograms of total RNA was prepared from untreated or JA-treated liquid-cultured plants. The RNA was electrophoresed on 1.2% agarose/formaldehyde gel and blotted onto nylon membrane. Probes were prepared from plasmid DNAs of AV531196, AV559219, AV553681, AV550375, AV439931, AV440220, and AV519956 (accession numbers reported by Asamizu et al., 2000) for GSH1 (At4g23100), GSH2 (At5g27380), DHAR (At1g19570), MDHAR (At3g09940), VTC1 (At2g39770), VTC2 (At4g26850), and VTC2-homologue (At5g55120), respectively. Probes were labelled with [α-32P]dCTP. Hybridization was performed as described by Sasaki et al. (2001).
Three micrograms of total RNA was prepared from untreated or O3-exposed plants and used for Northern hybridization. The same radiolabelling and hybridization conditions were used for the RNAs.
Measurement of thiol compounds and ascorbate
Seedlings were homogenized with a mortar and pestle in 5 volumes of 0.01 m HCl (fresh weight basis). After centrifugation at 15000 g for 10 min at 4 °C, the supernatant was used for quantification. The cysteine and GSH contents were measured by high-performance liquid chromatography (HPLC) as described by Hirai et al. (2003), with slight modifications. The extract (20 μl) was reduced by treatment with 3 μl of 5 mm dithiothreitol (DTT) and 100 μl of 100 mm 2-(cyclohexylamino) ethanesulphonic acid (CHES), pH 9.3, for 20 min at 37°C. The labelling reaction was terminated by the addition of 30 μl of acetic acid, and the resulting solution was subjected to HPLC analysis.
For ascorbate measurement, seedlings were homogenized in 10 volumes of 5% metaphosphate (fresh weight basis) and, after centrifugation at 15000 g for 10 min, the reduced ascorbate content in the supernatant was determined using an RQflex plus reflectometer (Merck, Darmstadt, Germany) according to the manufacturer's instructions.
Measurement of dehydroascorbate reductase activity
Frozen seedlings (1–2 g) were homogenized with a mortar and pestle in 3 volumes of iced buffer (50 mM potassium phosphate, pH 6.5, containing 0.5 mM EDTA, and 10 mM β-mercaptoethanol). The homogenate was centrifuged at 8000 g for 20 min at 4°C, and the supernatant was used for enzyme assays. Protein concentration was determined by a dye-binding method using a protein assay (Bio-Rad Laboratories, Hercules, CA, USA) with γ-globulin as a standard.
Dehydroascorbate reductase activity was assayed at 25°C by measuring the increase in absorbance at 265 nm attributable to GSH-dependent production of AsA, based on an absorbance coefficient of 14000 mM−1 cm−1. The assay was performed as described by Hossain and Asada (1984) with a slight modification in the composition of the 1-ml reaction mixture, which contained 50 mM potassium phosphate, pH 6.5, 5 mM GSH, 0.5 mM DHA, 0.1 mM EDTA and crude extract.
Plant growth conditions and O3 treatment
Seeds of opr3 plants in a WS-2 background were generously provided by Dr John Browse (Washington State University, Pullman, WA, USA). WS-2 and opr3 plants were germinated on blocks of glass wool and were grown in a chamber at 22°C at a relative humidity of 50–60% under a photosynthetic photon flux density (PPFD) of 80 μmol photons m−2 sec−1 in 14 h light/10 h dark cycles. Plants were watered with a liquid fertilizer (Hyponex 5-10-5; Hyponex Japan, Osaka, Japan) diluted 2000-fold. Sixteen-day-old plants were exposed to a single dose of 200 nl l−1 O3 for 6 h in an O3 chamber as described by Matsuyama et al. (2002). The O3 chamber was maintained at 25°C at a relative humidity of 70% under a PPFD of 350 μmol photons m−2 sec−1 of continuous light. O3 was generated by an O3 generator (Sumitomo Seika Chemicals, Tokyo, Japan). For phenotype observation after O3 exposure, we used the above-ground parts of 20–60 plants for each experiment, and then counted the number of wilted leaves. Similar results were obtained from three identical experiments.
Measurement of ion leakage
Membrane damage during O3 or fresh air exposure was estimated by measuring ion leakage in two or three detached rosette leaves from 16-day-old seedlings. The leaves were shaken in an orbital shaker in 1 ml of distilled water for 1 h at 100 rpm, and then the electroconductivity of the water was measured using an ion conductivity meter (B-173; Horiba, Tokyo, Japan). Data are expressed as a percentage of total ions, which were determined after killing leaves by autoclaving.
We thank Dr J. Browse for kindly providing opr3 seeds. This research was supported by a project entitled ‘Development of Fundamental Technologies for Controlling the Process of Material Production of Plants’, based on funds provided by the Ministry of Ecology, Trade and Industry in Japan. YS-S was supported by a research fellowship of the Japan Society for the Promotion of Science for young scientists.