Mutual antagonism of ethylene and jasmonic acid regulates ozone-induced spreading cell death in Arabidopsis


For correspondence (fax +358 9191 59552; e-mail

Current address: Department of Plant Physiology, Umeå Plant Science Centre, Umeå University, SE-90187 Umeå, Sweden.

Current address: Department of Biology, Coker Hall 108, CB#3280, University of North Carolina at Chapel Hill, Chapel Hill, NC 27599-3280, USA.

Current address: Department of Biology, University of Joensuu, PO Box 111, FIN-80101 Joensuu, Finland.

The first two authors contributed equally to this study.


Ethylene (ET) and jasmonic acid (JA) have opposite effects on ozone (O3)-induced spreading cell death; ET stimulates, and is required for the spreading cell death, whereas JA protects tissues. We studied the underlying molecular mechanisms with the O3-sensitive, JA-insensitive jasmonate resistant 1 (jar1), and the O3-tolerant, ET-insensitive ethylene insensitive 2 (ein2) mutants. Blocking ET perception pharmacologically with norbornadiene (NBD) in jar1, or ET signaling genetically in the jar1 ein2 double mutant prevented the spread of cell death. This suggests that EIN2 function is epistatic to JAR1, and that the JAR1-dependent JA pathway halts oxidative cell death by directly inhibiting ET signaling. JAR1-dependent suppression of the ET pathway was apparent also as increased EIN2-dependent gene expression and ET hypersensitivity of jar1. Physiological experiments suggested that the target of JA is upstream of Constitutive Triple Response 1 (CTR1), but downstream of ET biosynthesis. Gene expression analysis of 1-aminocyclopropane-1-carboxylic acid (ACC)-treated and O3-exposed ein2 and jar1 revealed reciprocal antagonism: the EIN2-mediated suppression of the JA pathway. The results imply that the O3-induced spreading cell death is stimulated by early, rapid accumulation of ET, which can suppress the protecting function of JA thereby allowing cell death to proceed. Extended spreading cell death induces late accumulation of JA, which inhibits the propagation of cell death through inhibition of the ET pathway.


Cell death is a common response in plants to several biotic and abiotic stresses, such as avirulent pathogens and ozone (O3) exposure (Morel and Dangl, 1997; Rao et al., 2000a). The plant hormones ethylene (ET) and jasmonic acid (JA) are implicated in the regulation of cell death as they modulate plant responses to a wide array of stress factors. ET and JA are involved in plant defense against wounding, insect feeding, and necrotrophic pathogens (Devoto and Turner, 2003; Thomma et al., 2001; Wang et al., 2002). The activity of these hormones is mediated through rapid changes in their concentrations and concomitant induction of the corresponding signaling pathways.

Several components of the ET biosynthetic and signaling cascades have been identified in Arabidopsis (Wang et al., 2002). The ET-overproducing mutants ethylene overproducer 1 (eto1)–eto3 are affected in the regulation of ET biosynthesis. ETO2 encodes one of the ET biosynthetic enzymes, 1-aminocyclopropane-1-carboxylate synthase 5 (ACS5). ETO3 encodes ACS9, and the function of ETO1 is to decrease the activity of ACS5 by decreasing the half-life of the protein (Chae et al., 2003). The Arabidopsis mutants ethylene resistant 1 (etr1), ethylene insensitive 2 (ein2), and ein3 are ET insensitive, whereas constitutive triple response 1 (ctr1) shows a constitutive ET response. ETR1 encodes one of the five ET receptors, CTR1 is an Raf-like protein kinase, EIN2 a membrane-bound metal transporter-like protein, and EIN3 a transcriptional activator.

Arabidopsis mutants have been instrumental also in the elucidation of the JA-signaling cascade. In addition to JA-biosynthetic mutants such as the fatty acid desaturase (fad)3fad7 fad8 triple mutant and 12-oxophytodienoate reductase (opr)3, several JA-insensitive mutants, such as coronatine insensitive (coi)1 and jasmonate resistant 1 (jar1), have been identified (Devoto and Turner, 2003). The fact that coi1 is sterile, but jar1 is fertile suggests that JAR1 is not involved in all JA-mediated responses. coi1 is also more insensitive to JA than jar1 (Ellis and Turner, 2002). Both COI1 and JAR1 have been identified. COI1 is an F-box protein and a component of an Skp1, Cullin, F-box (SCF) complex that is involved in the Constitutive Photomorphogenesis (COP)9-signalosome-mediated protein degradation (Devoto et al., 2002; Feng et al., 2003). JAR1 belongs to an acyl adenylate-forming firefly luciferase superfamily with adenylation activity towards JA, suggesting that adenylation of JA is crucial at least for certain aspects of JA function (Staswick et al., 2002).

ET is required together with JA for defense against necrotrophic pathogens and for systemic resistance induced by root-colonizing bacteria (Thomma et al., 2001). However, ET signaling can also function independent of JA, or even inhibit JA-dependent responses (Brader et al., 2001; Ellis and Turner, 2001; Thomma et al., 2001). This complexity is also observed at the level of transcriptional regulation. ET and JA induce expression of several defense-related genes in a synergistic fashion (Lorenzo et al., 2003; Norman-Setterblad et al., 2000; Penninckx et al., 1998), and it has been suggested that ET and JA stimulate each other's biosynthesis in wound responses (Laudert and Weiler, 1998; O'Donnell et al., 1996). However, ET can also inhibit JA-mediated gene expression (Ellis and Turner, 2001; Rojo et al., 1999; Shoji et al., 2000; Winz and Baldwin, 2001). The diversity of these observations may reflect the fact that interactions between the hormonal pathways depend on the nature of the external challenge or on the particular plant-specific defense response.

We have studied the interactions between the ET and JA pathways in response to exposure of Arabidopsis plants to O3. O3 degrades in the apoplast to various reactive oxygen species (ROS), which in turn induce active, endogenous ROS production in planta followed by programmed cell death similar to the pathogen-induced hypersensitive response (HR; Overmyer et al., 2000, 2003; Pellinen et al., 1999, 2002; Rao and Davis, 1999; Rao et al., 2000a; Schraudner et al., 1998; Wohlgemuth et al., 2002). We have previously shown that ET enhances and JA suppresses spreading of O3-induced cell death (Overmyer et al., 2000, 2003). Accordingly, defects in the ET and JA pathways, as observed in the ET-insensitive ein2 and the JA-insensitive jar1 mutants of Arabidopsis, result in opposite O3-induced cell death phenotypes, ein2 displaying reduced and jar1 enhanced cell death (Overmyer et al., 2000; Rao et al., 2000b). In this study, we investigated interactions between the JA and ET pathways by studying regulation of the spreading cell death present in the jar1 mutant. We show with double mutant analysis that JA protects cells from O3-induced spreading cell death by directly modulating ET signaling. Analysis of pathway-specific gene expression revealed further that, similarly, ET can suppress the JA pathway. It is predicted that both the temporal and spatial kinetics of ET and JA accumulation fine-tune the hormonal interactions and the spreading cell death induced by O3.


JAR1-dependent processes protect plants from oxidative cell death by suppressing the cell-death-promoting function of ET

To analyze the relationship of JA and ET signaling in the regulation of cell death in Arabidopsis, we analyzed the O3 sensitivity of the jar1 ein2 double mutant. Measurement of relative ion leakage as an indicator of the extent of cell death showed that ein2 and Columbia (Col)-0 were tolerant to O3, while jar1 developed spreading lesions in response to O3 concentrations as low as 250 nl l−1 (Figure 1a). The double mutant jar1 ein2 had lower O3-induced ion leakage than jar1 (Figure 1a). These results were consistent with the visible damage in the O3-exposed plants (Figure 1b); 300 nl l−1 O3-induced extensive HR-like lesions in jar1, occasionally small lesions in jar1 ein2, while no lesions were visible in Col-0 or ein2. Superoxide accumulated extensively in jar1 (Figure 1c), suggesting that ROS-induced lesion propagation is superoxide dependent, and similar to that of the lesions stimulating disease resistance (lsd)1 and radical-induced cell death (rcd)1 mutants and the Cape Verdi Islands (Cvi-0) ecotype (Jabs et al., 1996; Overmyer et al., 2000; Rao and Davis, 1999). Altogether, ein2 reduced, but did not completely block the O3-induced increase in ion leakage of jar1. This was further investigated by blocking ET signaling with the ET-receptor antagonist, norbornadiene (NBD). Application of NBD to jar1 after a 4-h O3 exposure (300 nl l−1) completely blocked spreading cell death (Figure 1d). Therefore, the cell-death-promoting function of ET seems to be epistatic to the function of JAR1 during oxidative stress.

Figure 1.

Suppression of ET-mediated spreading cell death by JAR1.

(a) O3-induced ion leakage (% of total ion leakage ± SE) was measured in whole rosettes of Arabidopsis Col-0, ein2, jar1, and jar1 ein2 at 7 h after the onset of a 6-h exposure to 250 and 300 nl l−1 of O3.

(b) O3-induced phenotype in 3-week-old Col-0, ein2, jar1, and jar1 ein2 photographed 24 h after the onset of a 6-h exposure to 300 nl l−1 of O3. O3-induced HR-like lesions are marked with arrows.

(c) Accumulation of superoxide in Col-0, ein2, jar1, and jar1 ein2 was detected by NBT staining of leaves grown in clean air (−O3) and at 7 h after the onset of a 6-h exposure to 300 nl l−1 of O3 (+O3). The presence of purple formazan precipitate indicates the location and extent of superoxide accumulation.

(d) Reduction of O3-induced ion leakage (±SE) in jar1 by ET-action blocker NBD 4 h after the onset of a 4-h exposure (indicated by the black bar) to 300 nl l−1 of O3.

ET and JA pathways interact directly during oxidative cell death

Direct evidence for the function of ET and JA in the regulation of spreading cell death was obtained by infiltrating detached leaves of Arabidopsis (Col-0) with the superoxide-generating system Xanthine/Xanthine oxidase (X/XO; Jabs et al., 1996; Overmyer et al., 2000) together with various combinations of 1-aminocyclopropane-1-carboxylic acid (ACC; the precursor of ET) and methyl jasmonate (MeJA). We have previously shown that X/XO and O3 act in a similar fashion in inducing spreading cell death in ecotypes and mutants varying in their O3 sensitivity (Overmyer et al., 2000). Application of MeJA afforded significant protection against X/XO-induced cell death, whereas the addition of ACC with X/XO enhanced cell death in the O3-tolerant Col-0 (Figure 2). ACC and MeJA had no effect on cell death in the absence of X/XO. When ACC and MeJA were applied simultaneously, the ACC-mediated increase in cell death was compromised (Figure 2), supporting the argument that JA suppresses the ET-mediated stimulation of the oxidative cell death also in a genotype (Col-0) that has low sensitivity to oxidative cell death.

Figure 2.

Modification of oxidative cell death by exogenous ACC and MeJA in vitro.

Arabidopsis Col-0 leaves were infiltrated in the superoxide-generating system X/XO with the infiltration buffer plus 50 µm ACC, 1.4 µm MeJA, or combinations of these compounds, as indicated. Cell death was quantified as relative ion leakage (% of total ions ± SE) after 18 h. Values labeled with same letters do not differ significantly (P > 0.05) according to Tukey's Honestly Significant Difference (HSD) test. The experiment was repeated three times with similar results.

JAR1 suppresses ET-mediated gene expression

The JAR1-mediated suppression of the ET pathway was further elucidated at molecular level by expression analysis of pathway-specific marker genes. A custom-made cDNA array was constructed to analyze expression of genes related to the function of ET and JA (for details of the complete data set, gene list, and validation, see Table S1; Figure S1). Samples were collected both from control and from O3-exposed Col-0, ein2, and jar1 at 2 and 8 h after the onset of the exposure. The results are shown as fold changes in response to O3 exposure (Figure 3a,b) and as fold differences to Col-0 after hierarchical cluster analysis (Figure 3c–e). ET-regulated genes were defined by their low expression levels in ein2 by cluster analysis (CL1; Figure 3c–e). Within 2 h after the initiation of O3 exposure, several genes in CL1 (Lesions Stimulating Disease Resistance (LSD)1, Glutathione S-Transferase (GST)1, GST2, Mitogen activated Protein Kinase (MPK)3, GST6, Blue Copper Binding protein (BCB), Mouse Apoptotic protein 3 (MA3)-3, Tryptophan synthase (TRP)A, EIN4, TRPB) were hyperactivated in jar1 (Figure 3d,e), supporting the hypothesis that JAR1 (or the JA pathway) suppresses at least a part of the ET responses.

Figure 3.

Gene expression in the O3-exposed ein2 and jar1.

Expression of the ET and JA-pathway specific genes was analyzed with a custom cDNA array in 3-week-old seedlings of Col-0, ein2, and jar1 collected at 2 and 8 h from clean air or after the onset of a 6-h exposure to 250 nl l−1 O3. The two left panels (a,b) show log2 ratios between gene expression in the O3-treated and the clean air-grown seedlings after 2 h (a) and 8 h (b). The panels (c–e) show log2 ratios between gene expression in each genotype and the corresponding Col-0 in clean air and after 2 and 8 h of O3 exposure. For a complete list of the genes, see Table S1. Red labeling indicates transcriptional activation and green labeling transcriptional repression, according to the key shown at the bottom. The data in panels (c–e) was analyzed in a single run by hierarchical clustering. CL1 and CL2 denote clusters of genes that are related to the function of ET (CL1) and JA (CL2) pathways.

O3exposure triggers transient increases in hormone levels

Defects in the JAR1-mediated suppression of ET signaling are expected to result in either increased ET biosynthesis or increased sensitivity to ET. We therefore analyzed the O3-induced accumulation of ET and JA in Col-0, ein2, jar1, and jar1 ein2 in detail. Three-week-old plants were exposed to a 6-h O3 pulse of 300 nl l−1, which induced extensive HR-like spreading lesions in jar1 3–4 h after the onset of O3 treatment, occasional lesions in jar1 ein2, and no lesions in Col-0 or ein2 (see Figure 1). In response to O3, jar1 had lower ET evolution than the wt Col-0 (Figure 4a), demonstrating that the JAR1-dependent JA pathway does not suppress ET biosynthesis. In accordance with previous observations (Guzmán and Ecker, 1990), ein2 and jar1 ein2 had high ET evolution (Figure 4a), presumably reflecting dysfunction in the feedback inhibition of ET biosynthesis.

Figure 4.

Accumulation of ET and JA in response to an O3 exposure of 300 nl l−1.

(a) ET evolution was measured in whole rosettes that were incubated in glass vials for 3 h. One-microliter samples were analyzed by GC. The results represent mean ± SE. The analysis was repeated several times with similar results.

(b) Concentration of JA was analyzed in whole rosettes by GC–MS using [1,2-13C]-JA as an internal standard. The analysis was repeated two times with similar results for the different genotypes. The results represent mean ± SE.

Accumulation of JA was apparent within 5 h after the onset of O3 exposure only in jar1 and jar1 ein2 (Figure 4b). This could reflect either the lack of feedback inhibition of JA biosynthesis through JAR1 or substrate limitation where substrate for JA biosynthesis is released only from the damaged membranes in the dying cells. No increase in JA accumulation was evident in Col-0 and ein2, which did not have visible lesions (Figure 4b). The accumulation kinetics of the JA-precursor 12-oxo-phytodienoic acid (OPDA) was similar to that of JA, while conjugated OPDA levels increased prior to the increase in JA (not shown).

JA suppresses the ET pathway downstream of ET biosynthesis

As the JAR1-mediated suppression of the ET pathway is not through regulation of ET biosynthesis, it could affect ET signaling. This was studied in detail by examining the effect of direct JA application on the triple response of 3-day-old etiolated seedlings using inhibition of hypocotyl elongation as an index of ET action. We used the Arabidopsis mutants eto1 and ctr1, both of which display constitutive activation of the ET pathway in the absence of external ET and therefore inhibited hypocotyl elongation (Wang et al., 2002). In untreated controls, the average hypocotyl length was 4.5 mm in eto1, 5.3 mm in ctr1, 13.4 mm in jar1, and 13.2 mm in Col-0. In Col-0, also exogenous MeJA had an inhibitory effect on hypocotyl elongation (Figure 5a), demonstrating that both ET and JA inhibit hypocotyl elongation. The suppression of hypocotyl elongation was, however, compromised in eto1 where increasing concentrations of exogenous MeJA, instead of decreasing, actually increased the hypocotyl length (Figure 5a) without affecting ET biosynthesis (Figure 5a inset). In contrast to eto1, ctr1 reacted like Col-0 control, and the effect of constitutive ET-signaling was not relieved but MeJA further decreased hypocotyl length in ctr1 (Figure 5a). These results suggest that JA antagonizes the ET-signaling pathway at a point upstream of CTR1, but downstream of ETO1.

Figure 5.

Molecular characterization of the JA action.

Hypocotyl elongation was assessed as an index of ET action in Col-0, eto1, ctr1, and jar1(a) and in jar1 and Col-0 (b) in response to logarithmic series of concentrations of MeJA (a) and ACC (b). Hypocotyl length of 3-day-old etiolated seedlings was analyzed from seedlings grown in vitro, and expressed as a percentage of the control (±SE). Inset boxes show ET evolution (pl seedling−1 h−1 ± SE) in response to exogenous MeJA in eto1 (a) and in response to exogenous ACC in Col-0 and jar1 (b). The experiments were repeated four times (a) or two times (b) with similar results.

The fact that JA mediates suppression of ET signaling downstream of ET biosynthesis, suggests that jar1 is hypersensitive to ET. This is also supported by the extensive induction of ET dependent cell death by O3 in jar1 (Figure 4), which occurs in spite of only a modest O3-induced increase in ET accumulation. We analyzed the ET sensitivity of etiolated jar1 and Col-0 seedlings by assaying ACC-induced inhibition of hypocotyl elongation as an index of ET action. Figure 5(b) shows that jar1 is indeed hypersensitive to ET, as exogenous application of the ET precursor ACC inhibited hypocotyl elongation in jar1 more and at lower concentrations than in Col-0. Again, this response was not because of differences in ET evolution between the ACC-treated Col-0 and jar1 (Figure 5b inset). In addition, the JA-biosynthetic mutant fad3-2 fad7-2 fad8 exhibited similar ET hypersensitivity (data not shown).

ET suppresses JAR1-dependent JA-regulated gene expression

Expression analysis of JA pathway-specific genes revealed the presence of additional interactions between ET and JA. JA-specific marker genes in CL2 were identified because of their low expression level in jar1 (Figure 3c–e). Expression of most of the CL2 genes was not significantly altered in response to O3 (Figure 3a,b), but there were differences between the genotypes. Eight hours after the onset of the exposure, expression of several of the CL2 genes (Allene Oxide Synthase (AOS), Lipoxygenase 2 (LOX2), Iron Superoxide Dismutase 1 (FeSOD1), β-fructosidase 1 (BFS1), Vegetative Storage Protein (VSP)) was slightly higher in ein2, suggesting that EIN2-dependent ET signaling suppresses JA-regulated gene expression during oxidative stress (Figure 3e). These results were also verified in expression analysis of the JA-specific marker gene VSP1 and LOX2 in vitro. We chose to analyze the expression of VSP1 and LOX2 as they were specifically induced by MeJA and not by other hormones such as ET (Table S1).

First, we tested the reliability of the experimental system by analyzing the expression of basic chitinase (CHIB) that is known to respond to ET and JA in a synergistic manner (Norman-Setterblad et al., 2000); ACC had a slight inductive effect on the expression of CHIB, but as expected, the effect was multiplied when applied together with MeJA (Figure 6a,c). Exogenous MeJA strongly stimulated VSP1 expression in all genotypes studied (Figure 6e). ACC slightly but consistently reduced the expression of VSP1 in a concentration-dependent manner in all genotypes except in ein2 (Figure 6d). The results were similar for the second JA-regulated marker gene LOX2 (Figure 6g,h). The antagonistic effect of ET on the JA pathway was confirmed by simultaneous application of MeJA and ACC (Figure 6f,i); ACC significantly decreased the induction of VSP1 and LOX2 by 1 µm MeJA in a concentration-dependent manner, but in ein2, the decrease was less pronounced. Also greenhouse-grown Col-0 plants demonstrated the same result; exogenous ET drastically reduced the MeJA-induced expression of VSP1 and LOX2 (not shown). Taken together, these results suggest that ET suppresses the JA pathway. Therefore, the interactions between some branches of the JA and ET pathways are mutually antagonistic.

Figure 6.

Expression of CHIB, VSP1, and LOX2 in response to exogenous ACC and MeJA. Expression of the CHIB, and the JA-pathway marker genes VSP1 and LOX2 was analyzed in response to 48-h treatment of ACC and/or MeJA in 2-week-old seedlings of Col-0, ein2, and jar1 grown in vitro. The results are expressed as percentage of the Col-0 control on a log2 scale.


Mutually antagonistic action of ET and JA

We have earlier shown that ET stimulates and JA inhibits spreading of oxidative cell death (Overmyer et al., 2000; Vahala et al., 2003). Here, we show with double mutant analysis a novel finding; the cell death inhibition of the JAR1-dependent JA pathway acts through a direct suppression of ET signaling. At O3 concentrations up to 300 nl l−1, blockage of ET signaling because of the ein2 mutation in the jar1 ein2 double mutant or inhibition of ET perception in jar1 with NBD prevented the spreading cell death typical to jar1 (Figure 1). At higher concentrations, O3 tolerance of the jar1 ein2 double mutant was more similar to jar1 (Tuominen H, Keinänen M, Overmyer K, Kangasjärvi J, unpublished results), which suggests that either there is residual ET sensitivity in jar1 ein2 that allows spreading of cell death in the case of extensive oxidative pressure, or that the increased cell death in the double mutant was a result of increased cell death initiation by the ROS formed directly from O3. We have also observed that the concentration of salicylic acid (SA) is significantly higher in the O3-exposed jar1 ein2 double mutant than in the jar1 or ein2 single mutants (Tuominen H, Keinänen M, Overmyer K, Kangasjärvi J, unpublished results). As SA promotes cell death and is involved in the lesion initiation (Overmyer et al., 2003; Rao and Davis, 1999; Van Camp et al., 1998), the elevated SA level might be responsible for the increased cell death by increasing the lesion initiations in the jar1 ein2 double mutant at extremely high O3 concentrations. The stomatal conductance and behavior (and thus also O3 influx) were similar in all the strains used in these experiments (data not shown); thus, the differences in the magnitude of cell death in the mutants reflect the role, contribution, and interactions of the hormonal signaling cascades in the initiation and propagation of cell death by apoplastic ROS.

The JAR1-mediated inhibition of ET signaling was visible at the molecular level as exaggerated ET-mediated responses (gene expression, suppression of hypocotyl elongation) in the jar1 mutant where JA signaling is deficient (Figures 3d,e and 5b). This antagonism is not typical only to the JAR1-dependent part of the JA pathway as direct application of MeJA suppressed ET-mediated oxidative cell death (Figure 2) and the ET-mediated suppression of hypocotyl elongation (Figure 5a). In addition, a second JA-insensitive mutant, coi1, as well as the JA-deficient fad3-2 fad7-2 fad8, displayed similar hypersensitivity and physiognomy in response to O3 as jar1 (Tuominen H, Keinänen M, Overmyer K, Kangasjärvi J, unpublished results).

Suppression of JA-dependent gene expression by ET (Figures 3e and 6) supports the presence of mutually antagonistic interactions between ET and JA. The antagonistic effect of ET seems to act at the level of JA signaling rather than JA biosynthesis as JA accumulation was not affected in ein2. Although both suppression of JA-related genes by ET (Shoji et al., 2000; Winz and Baldwin, 2001) and upregulation of the JA-responsive genes in the ET-insensitive genotypes (Ellis and Turner, 2001; Norman-Setterblad et al., 2000; Rojo et al., 1999) have been reported earlier, this interaction has largely been overlooked in models explaining interactions between the signaling pathways.

ET and JA act in parallel or synergistically in the regulation of defense gene expression, which is believed to result in resistance against necrotrophic pathogens, such as Botrytis cinerea and Erwinia carotovora (Norman-Setterblad et al., 2000; Penninckx et al., 1998; Thomma et al., 2001). Our results on the mutually antagonistic interaction between JA and ET in the regulation of spreading cell death are in contrast to these results. However, it is widely known that hormone interactions can alternate with respect to the physiological process involved. For instance, ET and abscisic acid have been shown to affect root growth in an additive manner but seed germination in an antagonistic manner (Gazzarrini and McCourt, 2003). Therefore, the observations that ET and JA function in an antagonistic manner in the regulation of the spreading cell death does not exclude synergistic or parallel function of ET and JA in defense responses, such as defense-gene induction or initiation of cell death. In fact, this kind of dual interaction between JA and ET seems to operate during the necrotrophic pathogen attack (Thomma et al., 2001). Synergistic function of the JA and ET pathways is required to allow maintenance of a high defense profile. On the other hand, JA functions to repress cell death as a defense against the pathogen, while ET has a cell-death-promoting effect (Berrocal-Lobo et al., 2002; Govrin and Levine, 2000; Thomma et al., 2001). The dual interactions between JA and ET signaling are not limited only to defense responses. We observed that JA suppressed ET-mediated hypocotyl elongation, and the same was demonstrated in apical hook formation (Ellis and Turner, 2002). This indicates that the JA-mediated suppression of the ET pathway is implicated, not only in the process of spreading cell death but also in two different developmental processes. In conclusion, ET and JA pathways can act both in a synergistic and an antagonistic fashion depending on the nature of the stimuli.

JA suppresses ET signaling downstream of ETO1 but upstream of CTR1

Several potential mechanisms could be responsible for the mutual hormone antagonism observed. The effect of exogenous MeJA on the two constitutive ET response mutants eto1 and ctr1 suggests that the target of the JA-mediated antagonism resides downstream of ETO1, but upstream of CTR1 (Figure 5a), which implies that JA affects ET signaling at the receptor level. ET receptors are constitutively active negative regulators of the downstream pathway, and upregulation of their synthesis leads to decreased ET sensitivity (Bleecker and Kende, 2000; Ciardi and Klee, 2001; Wang et al., 2002). Accordingly, published microarray data show that at least one Arabidopsis ET-receptor (ERS2) is JA inducible (Schenk et al., 2000). This suggests that, in addition to desensitization of ET perception by ET-induced synthesis of new receptor protein, also JA can affect ET sensitivity through increased ET receptor protein accumulation.

JA-mediated suppression of the ET pathway has also previously been shown by Ellis and Turner (2002) in connection to the ET-mediated hypocotyl hook formation, where the results illustrate that exogenous MeJA suppresses the hook formation in ctr1. This suggests that JA suppresses the ET pathway downstream of CTR1 when the hypocotyl hook formation is used as a marker for ET action. However, the results provided by Ellis and Turner (2002) also show that, in agreement with our results, MeJA did not suppress the ET-mediated inhibition of hypocotyl elongation in ctr1. This suggests that the various ET-mediated responses are regulated differentially in etiolated seedlings, or that there are multiple entry points for JA in the ET signaling, which is also supported by the observation that a downstream component of the ET pathway, EIN2, seems to transduce the synergistic activation of defense gene expression by ET and JA (Alonso et al., 1999).

The molecular mechanism for the ET-mediated suppression of the JA pathway is not known. It is possible that ET and JA signal pathways share common regulatory factors and they compete for binding to the same target promoters. For example, a common cis-element has been found in the promoter of an ETHYLENE RESPONSIVE ELEMENT BINDING PROTEIN/APETALA 2 (EREBP/AP2)-family transcription factor that relays both ET and JA signaling (Menke et al., 1999). In addition, Mitogen activated protein (MAP) kinase cascades have been implicated in both ET and JA signaling (Ouaked et al., 2003; Zhang and Klessig, 2001), and it is possible that these cascades impinge on each other or share common elements resulting in the observed antagonism between the pathways. It is also possible that the target for the antagonistic function is the same for both ET and JA, i.e. possibly the ET receptor(s), and the balance between the two signaling molecules determines the outcome of the spreading cell death.

The temporal and spatial regulation of the ET and JA pathways during oxidative cell death

The function of ET and JA signaling has both temporal and spatial components in the regulation of oxidative cell death, and thus it is possible that the antagonistic and synergistic functions are also separated in place and/or time. O3 induces ET evolution within 1 h, and accumulation of JA within 5 h after the onset of the exposure (Figure 4). The rapid O3-induced increase in ET evolution suggests that ET functions early and it has been shown that the early ROS accumulation responsible for the spreading of cell death is ET dependent (Moeder et al., 2002; Overmyer et al., 2000, 2003). We can also conclude that ET alone is not enough to promote spreading of cell death, but additional stimuli such as extensive radical production in the apoplast are required (Figure 2). Hence, O3-tolerant genotypes, such as the Col-0 wild type (wt) where O3 does not induce extensive ET biosynthesis, do not develop spreading lesions in response to O3 exposure. However, if ET is provided by external application (as in Figure 2), or if an otherwise O3-tolerant genotype (Col-0) contains a mutation that triggers extensive ET synthesis after stimulation by O3 (eto1 or eto2; Rao et al., 2002; Tuominen H, Keinänen M, Overmyer K, Kangasjärvi J, unpublished results), ET-dependent lesion propagation takes place. It is noteworthy that ET is not required in the initial cell death (Bent et al., 1992; Overmyer et al., 2000), but cell death is initiated even in the ET-insensitive genotypes, for example in ein2 at 400 nl l−1 (Tuominen H, Keinänen M, Overmyer K, Kangasjärvi J, unpublished results).

Because of the antagonistic function of ET on the JA pathway, ET suppresses JA signaling in the beginning phase and thereby allows cell death to proceed during the propagation phase of lesion formation (Figure 7). Both ROS accumulation (Overmyer et al., 2000; Wohlgemuth et al., 2002) and ET synthesis (Dat et al., 2003; Moeder et al., 2002) are located in the lesion borders. The level of JA increases only quite late during the lesion formation, and our data suggest that JA is primarily formed in the wounds, i.e. the borders of the lesions, as JA accumulation was observed only in plants that were damaged during the O3 exposure. It is also known that precursors of JA are released from disrupted membranes (Laudert and Weiler, 1998). From this follows that the capacity to suppress the spreading cell death is induced by the cell death process itself. If there is no cell death, as in the Col-0 wt, there is no increase in the JA level as there is no need to restrict spreading cell death. In the genotypes displaying spreading cell death, the late accumulation of JA suppresses the death-promoting ET pathway resulting in containment of the O3-induced lesions (Figure 7). Thus, both ET and JA syntheses seem to be locally restricted, and it is possible that the antagonistic action of ET and JA might occur only in the borders of the spreading lesions. JA-insensitive mutants cannot limit the spreading cell death, which results in rampant collapse of leaf tissue in response to oxidative stress. Taken together, the data support the function of the JAR1-dependent JA pathway in protection of plants from O3-induced spreading cell death by suppressing the cell-death promoting ET pathway.

Figure 7.

The regulation of O3-induced spreading cell death by mutually antagonistic interactions between ET and JA. The horizontal lines depict the antagonistic interactions that were observed between the signaling pathways. Vertical arrows depict increases in hormone levels that were observed during the propagation and containment phase of the O3-induced cell death process. The width and the height of the arrows reflect the strength of the interaction or the increase in the hormone concentration, respectively.

Experimental procedures

Plant material

Seeds of Arabidopsis thaliana (Col-0, ein2-1, eto1-1, ctr1, and jar1, all in Col-0 background) were obtained from the Arabidopsis Biological Resource Center (ABRC; Ohio State University, Columbus, OH, USA). The triple mutant fad3-2 fad7-2 fad8 was obtained from John Browse, Institute of Biological Chemistry, Washington State University.

The jar1 ein2 double mutant was created by selecting F2 individuals from the cross between jar1 and ein2 on plates containing 20 µm ACC by screening for the lack of the triple response (Guzmán and Ecker, 1990). Jasmonate-insensitive F3 lines were selected on 10 µm MeJA plates (Aldrich, Germany) by screening for the lack of inhibition of root elongation by MeJA as described by Overmyer et al. (2000). Homozygous jar1 ein2 lines were verified by backcrosses.

O3exposure, treatments, and analyses

Plants were grown in controlled growth chambers for 21 days (12-h day/12-h night, temperature 21/18°C, relative humidity 70/90%, photon flux density 250–300 µmol m−2 sec−1), and exposed to 300 nl l−1 of O3 for 6 h according to Overmyer et al. (2000) unless otherwise stated. Treatment with the 30 µl l−1 ET antagonist NBD (Sigma-Aldrich, St Louis, MO, USA) was performed in a desiccator jar. O3-induced cell death was quantified by measuring ion leakage with a conductivity meter (Mettler Toledo, Switzerland) in detached rosettes after 1-h incubation in pure water. Ion leakage was expressed as a percentage of total ions, quantified after killing the leaves by boiling. O3-induced accumulation of superoxide was detected with nitroblue tetrazolium (NBT; Boehringer Mannheim, Germany) as described by Jabs et al. (1996) and Overmyer et al. (2000). Ten leaves from five plants per genotype were stained in each of the two replicate experiments.

In vitro treatments

The effect of exogenous ACC or MeJA on hypocotyl length was analyzed by sowing sterilized Arabidopsis seeds on plates containing 0.5× MS with a logarithmic series of ACC or MeJA concentrations. Seedlings were grown vertically in the dark at 22°C for 3 days. Hypocotyl length was measured from digital photographs of the seedlings with the nih image program ( ET evolution was analyzed in response to both ACC and MeJA by growing seedlings for 3 days in airtight vials containing media with ACC or MeJA.

The superoxide-generating system X/XO (0.5 mm/0.05 U ml−1; Sigma-Aldrich) was applied to detached leaves as described by Overmyer et al. (2000). MeJA (1.4 µm) or ACC (50 µm) was applied as indicated in the section under Results. Cell death was quantified by measuring ion leakage after 18 h.

Quantification of ET emission and JA

Ethylene emission was quantified by placing a single rosette in a 14-ml airtight glass vial with 1 ml of water. After a 3-h incubation, a 1-ml sample was analyzed by a flame ionization gas chromatograph (Varian 3700, Varian Inc., Walnut Creek, CA, USA) equipped with a porapak Q column (80–100 mesh, 1 m × 3.2 mm). Column, injector, and detector temperatures were 40, 150, and 200°C, respectively.

JA was extracted and quantified with [1,2-13C]-JA as an internal standard as described by Baldwin et al. (1997) with the modifications described by Vahala et al. (2003). Samples were analyzed by GC–MS (Agilent 6890, 5973, Avondale, PA, USA) using an Rtx-5MS column (30 m × 0.25 mm i.d., 0.25 µm; Restek Corp., Bellefonte, PA, USA) with a helium flow rate of 1 ml min−1. The injector temperature was 250°C, and the temperature program was 1.5 min at 55°C, 10°C min−1 to 200°C, 20°C min−1 to 300°C, and 7 min at 300°C. The methyl ester of JA was detected in single-ion-monitoring mode (m/z 224 and 226).

Gene expression analysis

Expression of 126 defense-related genes (see Table S1) was studied by a custom-made cDNA array analysis using cDNA clones and expressed sequence tag (EST) clones from the ABRC (Columbus). All clones were re-sequenced. Seventy-five nanograms of each PCR-amplified sample was blotted onto Hybond N+ membranes with a 96-well dot blot device. Seventy-five nanograms of oligo-dT21, and pSPORT and pBS plasmids were used as negative controls. The two constitutively expressed genes ACT2 (H36835) and ACT8 (H36854) were applied to the membrane four times. Hybridization and detection were carried out according to Overmyer et al. (2000) except that 33P-dCTP was used for probe labeling. The results were normalized by reference to the mean hybridization signals for ACT2 and ACT8. Genes with expression levels below a numerical value of 0.001 in any of the samples were excluded from this analysis. Hybridizations were performed at least two times, and the results represent the mean of the duplicate signals. In addition, critical results were verified by comparison to results obtained from a second experiment.

Gene expression data were analyzed by hierarchical clustering of log2-transformed ratios ( according to Eisen et al. (1998). cDNA array results were verified by RNA gel blot analysis (Figure S1) of four randomly chosen genes according to Carpenter and Simon (1998). Expression of each of ACS6 (U79524), ARABIDOPSIS THALIANA ETHYLENE RESPONSE FACTOR 1 (ATERF1) (T42394), CHIB (AA067518), and AOS (N65720) was normalized against the expression of 18S RNA.

Expression of CHIB and the JA pathway-specific marker genes VSP1, and LOX2 was analyzed by RNA dot blot hybridization in Col-0, ein2, and jar1. Plants were grown for 14 days (12-h day, 70% RH, 22°C) in vitro on microtiter plates containing 1 ml of 0.5× MS medium (Sigma-Aldrich) supplemented with 2% sugar and 0.7% Bacto agar (pH 5.7). Whole rosettes were collected 48 h after application of ACC, MeJA, or appropriate control. Five micrograms of total RNA was applied with a 96-well dot blot device onto a Hybond N+ (Amersham Biosciences, Piscataway, NJ, USA) membrane and hybridized sequentially with a 32P-labeled gene-specific probe and an 18S-specific probe according to Church and Gilbert (1984). The gene-specific probes were prepared from the clones AA067518 (CHIB), AF043343 (VSP1), and T22547 (LOX2). Hybridization was detected and quantified with phosphorimager, and normalized against the expression of 18S RNA.


We are grateful to Ms Pinja Pulkkinen and Mr Mika Korva for technical help, Prof. Ian Baldwin (Max-Planck-Institute for Chemical Ecology, Jena, Germany) for supplying the JA standard, Prof. John Browse (Washington State University, Pullman, WA, USA) for providing the fad3-2 fad7-2 fad8 mutant, and Prof. Olevi Kull (Tartu University, Estonia) for the use of the AP4 porometer. This work was supported financially by the Scientific Council of Research of Environment and Natural Resources in Finland (Grants 43671, 52336, and 37995, and postdoctoral fellowships 41615 to H.T. and 48640 to M.K.), and by the Finnish Centre of Excellence Programme (2000–2005).

Supplementary Material

The following material is available from

Figure S1. Validation of the custom-made cDNA array.

(a–d) Expression of four randomly selected genes was analyzed by two different methods: the cDNA array analysis and the RNA gel blot analysis in 22 different samples. The data points corresponding to the different samples have been arbitrarily connected with a line to emphasize the differences between the different methods. Y-scales were adjusted arbitrarily for every plot as they were different for the two methods because of different normalization procedures (based on expression of 18S RNA for the RNA gel blot analysis, and expression of ACT2 and ACT8 for the cDNA array analysis). The comparison reflects variation between the two methods resulting from both the hybridization and the normalization procedures. The genes are: (a) ACS6, (b) ATERF1, (c) CHIB, and (d) AOS.
(e) Linear regression curve is shown for every data value obtained from the cDNA array analysis and exceeding the numerical value of 0.001 (n = 1495) plotted against its duplicate value. The duplicate values were obtained from two separate hybridization events and from two different membranes. The figure shows the threefold cut off limits (y = 3x and y = x/3) as calculated by SigmaPlot (SPSS Inc., AC Gorinchem, the Netherlands).

Table S1  The list of the cDNA clones used in the custom-made cDNA array and the raw data