Reactive electrophile species activate defense gene expression in Arabidopsis


* For correspondence (fax +41 21 692 4195; e-mail


Compounds containing α,β-unsaturated carbonyl groups are increasingly implicated as potent regulators of gene expression; some are powerful cytotoxins known to accumulate at the site of lesion formation in host–pathogen interactions. We used a robust measurement of photosynthetic efficiency to quantify the toxicity of a variety of lipid derivatives in Arabidopsis leaves. Small α,β-unsaturated carbonyl compounds (e.g. acrolein and methyl vinyl ketone) were highly active and proved to be potent stimulators of expression of the pathogenesis-related gene HEL (PR4). These small volatile electrophiles were far more active than larger alkenal homologs like 2(E)-hexenal, and activated HEL expression in a manner independent of salicylate, ethylene, and jasmonate production/perception. Electrophile treatment massively increased the levels of unesterified cyclopentenone jasmonates, which themselves are electrophiles. Patterns of gene expression in response to electrophile treatment and in response to avirulent bacteria were compared, which revealed strikingly similar transcript profiles. The results broaden the range of known biologic effects of reactive electrophile species to include the activation of a pathogenesis-related gene (HEL) and genes involved in metabolism. Electrophiles can act as mediators of both genetic and biochemical effects on core defense signal transduction.


Successful resistance responses in wild-type plants typically result in powerful defense gene expression combined with lesion development at the site of attack. Although cell death and defense gene expression are genetically dissociable (Greenberg et al., 2000; Yu et al., 1998), host cell damage or death and powerful defense responses typically go hand in hand. At the molecular level, other responses typify the host–pathogen interface. Reactive oxygen species (ROS), for example, accumulate as a hallmark feature of resistance responses (Alvarez et al., 1998). Also, highly characteristic of a broad range of pathosystems is the production of oxylipins and other, smaller lipid-derived compounds (Howe and Schilmiller, 2002). Lipid peroxidation associated with both the enzymatic and the non-enzymatic production of these types of molecules in diseased and stressed tissues is highly documented (e.g. Deighton et al., 1999; Imbusch and Mueller, 2000; Jalloul et al., 2002; May et al., 1996; Rustéricci et al., 1999; Vollenweider et al., 2000). Stress-related lipid oxygenation, whether enzymatic or non-enzymatic, produces a remarkable diversity of compounds covering a wide range of molecular masses. However, features common to many of these compounds are the presence of carbonyl groups in highly conserved molecular contexts. Simplest are the alkanals, for example hexanal, which is often released by wounded or infected leaves (e.g. Croft et al., 1993). Another large and interesting subgroup of lipid oxidation products contains α,β-unsaturated carbonyl groups. These can be 2(E)-alkenals like acrolein or 2(E)-hexenal in which the double bond is conjugated to an aldehyde, or molecules like fatty acid ketodienes, ketotrienes, and oxo-phytodienoic acid (OPDA) in which the carbonyl group is present as a ketone, conjugated to an α,β-double bond. Proximity to an unadulterated double bond increases the electrophilic properties of carbonyl groups, enhancing their chemical reactivity. The molecules are prone to several characteristic chemical reactions including Michael addition whereby nucleophilic atoms (S, N, etc.) react with the β-carbon in the double bond. The accumulation of these sorts of molecules, including malondialdehyde, 2(E)-hexenal, 2(E)-nonenal, oxo-phytodienoic acid (OPDA), fatty acid ketodienes, etc. (Figure 1), in diseased and wounded plant tissues is well known. Other very small electrophiles are widely distributed in nature. Acrolein, the smallest α,β-unsaturated carbonyl compound, is a natural product of lipid oxidation in animal tissues and cells (Uchida et al., 1998).

Figure 1.

Examples of naturally occurring α,β-unsaturated carbonyl compounds in plants.

Several oxylipins and small lipid-derived compounds, most of which are strongly implicated in stress responses and pathogenesis, are shown. Electrophilic regions comprising α,β-unsaturated carbonyl groups are encircled with dashed lines. Acrolein and MVK are the smallest members having the carbonyl group as an aldehyde or a ketone, respectively.

What are the regulatory activities of electrophiles in diseased plant tissues? Many aldehydes like 2(E)-hexenal are antimicrobial and have been directly implicated in defense responses (e.g. Croft et al., 1993). However, a second potentially important feature of the molecules is their ability to alter gene expression. Several years ago, xenobiotics containing α,β-unsaturated carbonyl groups were discovered to be active as inducers of phase II enzymes like glutathione-S-transferases (GSTs) in animal cells (Prestera et al., 1993). More recently, naturally occurring compounds including malonaldehyde and fatty acid ketodienes and ketotrienes were shown to powerfully activate GST1 gene expression in plants via their electrophilic properties (Vollenweider et al., 2000). Other stress-related genes are clearly activated by 2(E)-alkenals (Bate and Rothstein, 1998), and this activation also correlates with electrophilic properties. Additionally, lipid-derived cyclopentenones have been strongly implicated in the regulation of gene expression in animals (Straus and Glass, 2001). In plants, studies on the cyclopentenone jasmonate signal OPDA have concentrated on host–pathogen interactions and defense responses against insects. Endogenous OPDA was shown to be essential for resistance against an insect and a fungal pathogen (Stintzi et al., 2001), and in the same study, the ability of exogenous OPDA to stimulate gene expression via its electrophilic properties was demonstrated. Finally, extensive and careful studies have shown that germination and haustorial development of some parasitic plants is initiated by host-derived quinones, carbonyl-group-containing phenolic compounds, which are electrophilic in character (Keyes et al., 2001).

In the present work, we have continued exploring the biologic activities of α,β-unsaturated carbonyl-group-containing molecules whose production at the host–pathogen interface typifies defense responses (Vollenweider et al., 2000). Two potentially related aspects of their activity have been identified – their damaging effects on plant cells and their ability to stimulate defense-related gene expression (Stintzi et al., 2001; Vollenweider et al., 2000). So far the types of genes in plants shown to be activated by electrophilic carbonyl compounds include those involved in metabolism, such as chalcone synthase (CHS), dihydroflavonol reductase (DHS), allene oxide synthase (AOS; Bate and Rothstein, 1998), glutathione-S-transferase1 (GST1; Vollenweider et al., 2000), and oxo-phytodienoic acid reductase1 (OPR1), and defense genes such as elicitor-induced3 (ELI3; encoding an aromatic alcohol:NADP+ dehydrogenase), PR4 (HEL), and myrosinase-binding protein (MBP; Stintzi et al., 2001 with Supplementary Material at the PNAS Website).

In this study, we investigated electrophilic effects on the expression of pathogenesis-related genes. Using a quantitative assay, the structural features of oxylipins and other lipid-derived compounds were rigorously investigated in relation to their damaging effects on photosynthetic efficiency. Once identified, damaging molecules were tested for their ability to activate HEL gene expression and interact at both the genetic and the biochemical levels with defense-related signaling networks. As pathogenesis-related (PR) genes are of particular interest in immune responses, it is important to know whether signal pathways controlling the expression of such genes are in any way affected by the types of α,β-unsaturated carbonyl-group-containing molecules that typify pathogenesis or are otherwise found in the plant's environment. We found that the defense gene activation correlated with toxicity, demonstrating the utility of assaying for cell damage. The activation of the HEL gene by an electrophile was found to be largely insensitive to several well-characterized mutations in signal pathways. Comparison of gene expression in response to a bacterial pathogen with that induced by electrophile treatment revealed a remarkable similarity in transcript profiles. Together, the results suggest that plant immune responses can be activated or modulated by reactive electrophile species and, in particular, small α,β-unsaturated carbonyl-group-containing molecules.


Cytotoxicity of lipid derivatives

To quantify the damaging effects of natural lipid derivatives before visible symptom development, we measured the ratio Fv/Fm, a robust measure of photosystem II fluorescence (Björkman and Demming, 1987; Krause and Weis, 1991). This widely used assay allowed a rapid, non-invasive, and quantitative assessment of the damaging effects of lipids on leaf cells. Conditions for infiltration of compounds or for their use as volatiles were standardized to reduce variability of results (see Experimental procedures). Infiltrations were carried out at identical times with respect to the light regime of plants. As a positive control for damaging effects, we first monitored the relative toxicity of linolenic acid 13(S)-hydroperoxide (20 nmol, Table 1(a)). As expected (Rustéricci et al., 1999), this compound dramatically reduced Fv/Fm whereas the negative control linolenic acid had relatively little effect on this ratio. A larger systematic survey of the effects of oxylipins on PSII fluorescence was then undertaken. Arabidopsis leaves were infiltrated with freshly prepared suspensions of each compound (20–200 nmol in 10 µl water), and the Fv/Fm ratio was measured 15 h after infiltration. Table 1(a) shows the results of infiltrating leaves with linolenic acid, its 13-hydroperoxide, and a variety of oxylipins including epoxides, OPDA, jasmonic acid (JA), and a variety of alkanals and alkenals. Reduction of the carboxyl group of the latter compound to an aldehyde (linolenic aldehyde) did not significantly increase its harmful effect even when large amounts (100 nmol) were infiltrated. Fatty acid epoxides also had little effect on PSII under the conditions tested. Traumatin (12-oxo-10(E)-dodecenoic acid) was weakly active in reducing photosynthetic efficiency when relatively high amounts (40 nmol) were infiltrated into Arabidopsis leaves. The linoleic acid-derived ketodiene 13-oxo-octadeca-9(Z),11(E)-dienoic acid (13-KODE) slightly reduced the photosynthetic efficiency of Arabidopsis leaves; however, the 13-ketotriene 13-KOTrE was more active. 4-hydroxy-2(E)-nonenal was found to be relatively damaging to PSII. As this alkenal could be unstable at the acidic pH of the plant cell wall, it was infiltrated into the leaves in separate experiments in either water or a buffer. The relevant control showed that this buffer did not affect PSII fluorescence. The cyclopentanone JA was then compared to its cyclopentenone precursor OPDA. Both compounds (20 nmol) diminished PSII fluorescence, but OPDA had a more dramatic effect on PSII, causing greatly reduced fluorescence. A second series of experiments were then carried out with volatile electrophiles under controlled conditions in 1 l sealed jars. Table 1(b) reports the results of such a survey. At the highest concentrations tested (200 µmol l−1), all compounds, except hexane, led to strong decreases in Fv/Fm. At lower levels (20 µmol l−1), both 2(E)-nonenal and 2(E),6(Z)-nonadienal caused a significant reduction of PSII fluorescence whereas hexanal and 2(E)-hexenal were without significant effect. Direct comparison of the damaging effects of the 2(E)-alkenal series, acrolein, 2(E)-hexenal, and 2(E)-nonenal revealed the following activity series: acrolein > 2(E)-nonenal > 2(E)-hexenal.

Table 1.  Electrophile treatment and decay in photosystem II fluorescence in Arabidopsis plants (a)
Infilrated compoundAmount (nmol)Fv/FmSEM (±)Number (n)
No treatment0.8560.00212
Linolenic acid200.8140.0074
13(S)-hydroperoxylinolenic acid200.4140.0384
rac 9,10-EpOME200.8210.0108
rac 12,13-EpOME200.8300.0798
12-Oxo-phytodienoic acid200.2970.05510
Jasmonic acid200.7750.02815
9-Oxo-nonanoic acid200.8240.00612
4-Hydroxy-2(E)-nonenal (HNE)200.6390.04012
HNE in buffera200.6500.0874
Linolenic aldehyde1000.8030.02221
Table (b). 
Volatile compoundConcentration (µmol l−1)Fv/FmSEM (±)Number (n)
  1. (a) Leaves were infiltrated with suspensions of fatty acids and oxylipins in water.

  2. (b) Plants were exposed to volatile alkanals and alkenals in 1 l glass chambers. All measurements were conducted at 15 h after infiltration of molecules or exposure to volatiles except for propionaldehyde, acrolein, 2-butanone, and methyl vinyl ketone; for these substances, Fv/Fm measurements were taken 3 h after exposure. Results are given as ±SEM for the number of measurements (n) as indicated.

  3. Abbreviations: 9,10-EpOME, 9,10-epoxy-12(Z)-octadecenoic acid; 12,13-EpOME, 12,13-epoxy-9(Z)-octadecenoic acid; 9-KODE, 9-oxo-octadeca-10(E),12(Z)-dienoic acid; 13-KODE, 13-oxo-octadeca-9(Z),11(E)-dienoic acid; 13-KOTrE, 13-oxo-octadeca-9(Z),11(E),15(Z)-trienoic acid; and traumatin, 12-oxo-10(E)-dodecenoic acid. rac = racemic.

  4. a The buffer was 10 mm potassium phosphate, pH 6.3.

No treatment0.8470.00324
Methyl vinyl ketone100.5770.01910

Cytotoxic electrophiles activate HEL gene expression

The data in Table 1 provide information on the structural features of natural lipid-derived molecules, which make them powerfully damaging to plant leaves. This effect may be correlated with the ability to induce gene expression (Vollenweider et al., 2000). We compared the activities of the 2(E)-alkenal series of acrolein, 2(E)-hexenal, and 2(E)-nonenal on HEL expression 3 h after exposure to the compounds. Only acrolein activated HEL gene expression, although in some experiments 2(E)-nonenal was weakly active (Figure 2a). In order to eliminate the possibility that only aldehyde-containing molecules activated HEL gene expression, we then tested an α,β-unsaturated ketone, methyl vinyl ketone (MVK), along with the closely related control compound 2-butanone. MVK induced HEL gene expression under the conditions tested with 2-butanone being inactive (Figure 2b).

Figure 2.

Natural electrophile species capable of inducing HEL gene expression.

(a) Relative activities of the 3-, 6- and 9-carbon 2(E)-alkenal series. Plants were either not exposed to compounds (negative control), or treated with acrolein (C3), 2(E)-hexenal (C6), or 2(E)-nonenal (C9). In each case, compounds were used as solvent-free volatiles (1 µmol l−1 air in 1 l glass chambers). Leaves were harvested after 3 h exposure to compounds for RNA-gel blot analysis and probed with a HEL cDNA.

(b) HEL transcript accumulates in response to methyl vinyl ketone. Plants were exposed to either no compound (–) or 2 µmol l−1 air volume of 2-butanone or methyl vinyl ketone for 3 h prior to RNA extraction. The gene Lhb1B2 was used as a loading control.

Electrophile interaction with defense signal pathways at the genetic and biochemical levels

Three signaling mutants, affected in either the jasmonate pathway (coi1-1; Feys et al., 1994), the ethylene pathway (ein2-1; Guzman and Ecker, 1990) or the salicylate pathway (sid2-1; Nawrath and Métraux, 1999; Wildermuth et al., 2001), were tested for their response to MVK (Figure 3). HEL expression was MVK inducible in wild-type plants and was similarly inducible in all three mutants. Basal HEL transcript levels varied slightly from experiment to experiment, and the lower-than-wild-type level of HEL transcript in sid2 plants (Figure 3) was not always observed. MVK treatment repressed the loading control gene Lhb1B2 (Figure 3) in the ein2-1 mutant. In parallel to these experiments on gene expression, the effects of electrophiles on photosynthetic efficiency were investigated in the three mutants, ein2-1, coi1-1, and sid2-1. The mutations did not significantly affect the damage response to MVK (measured as Fv/Fm) with respect to the wild-type plants (not shown).

Figure 3.

‘Autonomous’ effects of an electrophile on defense gene expression.

Wild type (WT) or coi1-1, ein2-1, and sid2-1 mutants were exposed to methyl vinyl ketone (2 µmol l−1 in a 15.5 l air volume) for 3 h prior to extraction of RNA and analysis by RNA gel blot. The loading control gene was Lhb1B2; the blot was probed with this gene and with a HEL cDNA.

In order to complement and broaden the experiments on gene expression, we investigated the effects of electrophiles at the biochemical level, measuring effects of electrophile treatment on members of the jasmonate family of regulators. Acrolein, a small electrophile capable of stimulating HEL expression (Figure 2) was employed in these studies. Plants were exposed to acrolein vapor (10 µmol l−1 air volume; these high levels resulted in strong effects on jasmonate pools but lower levels also affected jasmonate levels) for 1 h and then placed in fresh air under constant light for a further 3 h prior to oxylipin analysis. Levels of jasmonate family members (JA, OPDA, and dinorOPDA) were then quantified using a standard assay (Weber et al., 1997). JA levels increased from 0.25 ± 0.02 nmol g−1 fresh weight (FWT) in the control (propanal-treated) plants to 0.38 ± 0.02 nmol g−1 FWT in acrolein-treated plants. Acrolein treatment powerfully increased levels of both OPDA and dinorOPDA to 4.19 ± 0.11 and 3.47 ± 0.05 nmol g−1 FWT, respectively, when compared to the control levels of 1.13 ± 0.33 nmol g−1 FWT for OPDA and 0.41 ± 0.11 nmol g−1 FWT for dinorOPDA. Similar observations were made with methyl vinyl ketone treatments, which also affected the levels of cyclopentenone jasmonates (not shown).

Gene expression in response to bacterial pathogenesis and electrophile treatment

To further investigate the hypothesis that the generation of electrophilic α,β-unsaturated carbonyl compounds has relevance in pathogenesis (Vollenweider et al., 2000), we tested whether treatment of plants with electrophiles would upregulate pathogen-inducible genes. Arabidopsis plants were exposed to either MVK or identical amounts of the control compound 2-butanone (1 µmol l−1 air space, 3 h exposure), and samples analyzed on a rigorously constructed cDNA microarray containing stress- and defense-related genes (Reymond et al., 2000). As expected, exposure to MVK led to increased transcript levels of not only HEL and GST1, but also many other genes (Table 2). Parallel experiments with ethyl vinyl ketone gave similar results (not shown). We also examined gene expression during bacterial infection of Arabidopsis plants infiltrated with either virulent or avirulent strains of Pseudomonas syringae pv. tomato DC3000 (Pst). Before initiating these experiments, the progression of disease was followed by measuring both bacterial growth and photosynthetic efficiency. Significant differences in bacterial growth began at around 18 h post-infiltration (Figure 5a) whereas reduction in Fv/Fm occurred earlier in the incompatible interaction than in the compatible interaction (Figure 5b). We chose a time point at which bacterial populations were similar but significant reduction of Fv/Fm was measurable in Pst avrRpm1-infiltrated leaves for gene expression studies. Nine hours after infiltration with bacteria, leaves were extracted and mRNA was used as a template for fluorescent-labeled cDNA synthesis. Comparison of both compatible and incompatible interactions showed that the transcript levels of several genes were slightly higher in plants infiltrated with Pst avrRpm1 than in plants infiltrated with virulent Pst (Table 2). However, the overall pattern of gene expression observed in response to MVK treatment was remarkably similar to that observed in response to Pst. All genes induced by MVK were also upregulated by Pst avrRpm1 and, to a lesser extent, by Pst, the unique exception being ROF1 which was only MVK inducible. Conversely, only the genes HIN1, PR1, and PR3, as well as the genes involved in aromatic amino acid metabolism like COMT, CCOAMT, and CM1, were induced by Pseudomonas.

Table 2.  Transcript levels in response to Pseudomonas syringae infection and methyl vinyl ketone treatment
MVKPstPst avr
  1. Plants were exposed to methyl vinyl ketone (MVK) or to the control compound 2-butanone (1 μmol l−1 air volume). After 3 h, leaves were removed and mRNA was extracted, labeled, and hybridized to cDNA microarrays. The results show MVK-induced changes in gene expression (as fold-induction) relative to the 2-butanone controls. Similarly, plants were infiltrated with Pst or with PstavrRpm1 (2.5 ×105 cfu in 10 μl MgCl2, 10 mm). In this case, control plants were infiltrated with 10 μl MgCl2, 10 mm. Leaves were harvested 9 h later for cDNA microarray analyses. Each column indicates the average of three biologically independent experiments. The complete gene list can be accessed at Listed by functional groups are genes that were upregulated by a minimum of twofold in at least one treatment. Ratios above 2.0 (99.9% confidence) are highlighted in red, and ratios between 1.5 (99.0% confidence) and 2.0 are highlighted in pink. Genes printed on the array are listed in Table S1 and full experimental results are given in Table S2.

HELHevein-like proteinPR protein3.456.076.95
ELI9Hydroxyproline-rich glycoproteinPR protein1.522.863.28
HIN1Harpin-induced proteinPR protein1.3013.7212.22
PDF1.2DefensinPR protein1.534.013.11
PGIPPolygalacturonase-inhibiting proteinPR protein2.521.822.23
PIN2Proteinase-inhibiting proteinPR protein0.971.692.53
PR1Antifungal proteinPR protein1.413.709.99
PR2GlucanasePR protein1.861.653.52
PR3B1ChitinasePR protein1.141.082.29
GST1Glutathione-S-transferaseOxidative stress5.752.963.79
GST5Glutathione-S-transferaseOxidative stress2.211.892.38
GST8Glutathione-S-transferaseOxidative stress4.752.863.14
GPX2Glutathione peroxidaseOxidative stress1.733.012.92
APX1Ascorbate peroxidaseOxidative stress2.191.422.09
SODCUSuperoxide dismutaseOxidative stress1.972.182.84
MFP2Multifunctional proteinBeta-oxidation2.111.632.55
AOSAllene-oxide synthaseJasmonate synthesis1.851.672.48
HPLHydroperoxide lyaseLipid metabolism1.781.772.36
EHEpoxide hydrolaseLipid metabolism2.521.752.81
PIOXAlpha-oxygenaseLipid metabolism2.471.892.97
ELI3oxidoreductaseAromatic metabolism3.442.123.90
COMTO-methyl transferaseAromatic metabolism1.123.035.08
CCOAMTO-methyl transferaseAromatic metabolism0.885.313.85
CM1Chorismate mutaseAromatic metabolism1.151.982.82
ASA1Anthranilate synthase (alpha)Tryptophan synthesis2.732.783.22
ASBAnthranilate synthase (beta)Tryptophan synthesis2.813.513.18
TSATryptophan synthase (alpha)Tryptophan synthesis3.113.664.09
TSBTryptophan synthase (beta)Tryptophan synthesis2.383.114.04
CYP79B2Cytochrome P450Indole glucosinolate synthesis1.484.945.35
CYP83B1Cytochrome P450Indole glucosinolate synthesis2.242.312.95
JR3IAA-Ala hydrolaseAuxin synthesis3.984.645.01
NIT2NitrilaseAuxin synthesis2.511.52.30
TCH1CalmodulinCalcium sensing1.912.032.09
TCH3Calmodulin-like proteinCalcium sensing2.192.951.87
ACO1ACC oxidaseEthylene synthesis3.432.263.09
ERD1ClpA/B proteaseStress response2.642.392.61
OPR1OPDA reductaseStress response5.271.231.71
ROF1RotamaseStress response2.801.031.28
OZI1UnknownStress response1.832.762.31
RNS2RNaseNucleic acid metabolism2.171.742.28
ERECTAErecta receptor kinaseFlower development1.901.672.3
ADL4Dynamin-like proteinCytoskeleton/transport2.611.772.15
CYP71B6Cytochrome P450Unknown/metabolism2.873.182.86
RAB1CSmall GTP-binding proteinSignaling2.161.622.36
Figure 5.

Disease progress for Pst and PstavrRpm1 in Arabidopsis leaves.

(a) Growth of bacteria in the leaves.

(b) Photosynthetic efficiency in Arabidopsis leaves infiltrated with bacteria. Plants were infiltrated with bacteria (2.5 × 105 cfu in 10 µl MgCl2, 10 mm) and were maintained at saturating humidity, i.e. under the same conditions as for plants used for all microarray analyses shown in Table 2.


Electrophiles as cytotoxins

The production of oxylipins and other oxygenated lipid derivatives is highly characteristic of the responses to severe insult, and plant tissues produce a large range of these molecules during pathogenesis. To date, many studies have focused on the activities of octadecanoids (Howe and Schilmiller, 2002). However, at one extreme of the spectrum of oxygenated lipid derivatives are small and highly reactive compounds, such as acrolein, methyl vinyl ketone, and ethyl vinyl ketone. We undertook experiments with these smaller lipid derivatives in order to explore their effects on cell damage and on gene expression. Understanding exactly how the chemical context in the vicinity of the α,β-unsaturated carbonyl group of electrophilic lipid derivatives affects their cytotoxic properties and their biologic activities is a prerequisite for assessing the impact of electrophile generation during pathogenesis. As a quantitative means of assessing cytotoxicity, we measured Fv/Fm. The results (Table 1) give a strong indication that the molecular context of the α,β-unsaturated carbonyl group is highly important, and indicate how this context determines the activity. Aldehydes, such as propanal, hexanal, or linoleic aldehyde, have very little effect on PSII. However, the α,β-unsaturated carbonyl group in small molecules, such as acrolein and MVK, was powerfully damaging to PSII.

The results in Table 1 can be compared with the structure/function studies of the toxicity of fatty acids and their derivatives in vertebrate cells. Increasing numbers of carbon atoms in fatty acid derivatives increase their toxicity to human fibroblasts (Kaneto et al., 1987). In the present study with plant cells, we have observed that this is clearly not the case with the 2(E)-alkenal series. The same study (Kaneto et al., 1987) found that increasing numbers of double bonds in molecules increased their toxicity. This relationship appears to hold true in the present study also. For example, the fatty acid ketodiene 13-KODE (20 nmol) was found to be only slightly toxic to PSII under the conditions tested but the ketotriene 13-KOTrE was reproducibly more active (Table 1). This correlates well with the observation that 13-KOTrE was more active than 13-KODE in inducing GST1 gene expression in Arabidopsis (Vollenweider et al., 2000). Concerning the effect of jasmonates, the fact that OPDA (an α,β-unsaturated carbonyl compound) and JA (which does not contain an α,β-unsaturated carbonyl group and which is far less electrophilic) both decrease photosynthetic efficiency is fully consistent with the literature. JA initiates a senescence program involving the destruction of chloroplast components (Reinbothe et al., 1993). In the present study, OPDA was more damaging to PSII, a property which correlates well with the presence of an electrophilic α,β-unsaturated carbonyl feature in this molecule (Stintzi et al., 2001). It is also of note that the ‘wound hormone’ traumatin (12-oxo-10(E)-dodecenoic acid), a compound which might be involved in wound healing (Zimmerman and Coudron, 1979), was weakly active in damaging PSII. The electrophilic nature of traumatin could perhaps help to explain some of its reported effects on plant cells.

The following activity series for the damaging effects of α,β-unsaturated carbonyl-group-containing molecules to PSII can be defined starting with the most active compounds: acrolein > MVK > 2(E),6(Z)-nonadienal > 2(E)-nonenal > 2(E)-hexenal > OPDA = 13-KOTrE > 13-KODE > traumatin. The ranking is tentative because it is not possible to directly compare the activity of volatile and non-volatile compounds. Additionally, diffusion of the molecules into the cells, as well as their in vivo half-lives, might have a powerful influence on their activities. Hydrophobicity alone is clearly not enough to explain the effects of the compounds. Acrolein, for example, is a very active inducer of HEL gene expression but 2(E)-nonenal, a more hydrophobic molecule, is not (Figure 2). Another factor that is expected to influence the activity of electrophiles is reduction potential. In a nutshell, our results indicate that two antagonistic properties help to determine the toxicity/biologic activity of the α,β-unsaturated carbonyl compounds we investigated: the smaller the molecule the more active, the more double bonds the more active. Thus, several factors are likely to govern the activity of a particular molecule and further studies will be needed to better define structure/activity rules. The results of the toxicity assay draw particular attention to small (C3 or C4) α,β-unsaturated carbonyl-group-containing molecules among which are acrolein and MVK. These naturally occurring molecules are potent not only in damaging cells but also in selectively stimulating gene expression in a manner strikingly paralleled by activation of gene expression by the pathogen Pst. Importantly, these compounds, including the most active compound acrolein (2-propenal; Hayase et al., 1984), have been detected in plants.

Electrophiles activate HEL gene expression

The accumulation of host-produced electrophiles is typical of pathogenesis and, in particular, incompatible interactions (Bate and Rothstein, 1998; Croft et al., 1993; Deighton et al., 1999; Farmer, 2001). The compounds elicit two of the phenomena most characteristic of many diseases, i.e. cell damage and defense gene expression. Many defense genes in plants are under the control of low-molecular-mass regulators, such as salicylic acid, members of the jasmonic acid family, and ethylene (Feys and Parker, 2000; Glazebrook, 2001). α,β-unsaturated carbonyl compounds can also powerfully affect defense-related gene expression. The expression of some genes is activated by exogenous OPDA via its electrophilic properties (Stintzi et al., 2001). One immune response gene that was observed to be upregulated by OPDA was HEL (Stintzi et al., 2001 and Supplementary Material), and this observation provided a starting point for studying the effects of electrophiles on this well-studied gene.

The regulatory effects of acrolein in plants have received little attention, but the molecule, which is an in vivo marker of oxidative damage to lipids (Uchida et al., 1998), has potent regulatory activities in animal cells influencing the activity of key, stress-related transcription factors (Kehrer and Biswal, 2000). In the present study, we found that small volatiles like acrolein were potent inducers of HEL expression. 2(E)-hexenal was previously characterized as an inducer of stress-related gene expression (Bate and Rothstein, 1998), but under the conditions of our experiments, it was weak or inactive as an HEL regulator. To further investigate the effect of the overall size of electrophiles and the molecular context of the α,β-unsaturated carbonyl group, i.e. whether it could be internal as opposed to presented as an aldehyde, we tested the ketone methyl vinyl ketone (MVK, Figure 2b). This compound is highly damaging (Table 1b) to PSII whereas a related control, 2-butanone, was not. HEL activation showed similar structure dependency. The data in Figure 2 indicate that the α,β-unsaturated carbonyl group can be terminal or internal in the parent molecule without losing activity in inducing HEL expression. Interestingly, levels of transcript for the loading control gene Lhb1B2 (which encodes a chlorophyll-binding protein) were diminished in some electrophile treatments (Figure 3). This is not surprising given the damaging nature of electrophiles such as MVK, but it might also hint at specific effects on the chloroplast and warrants future investigation.

Interaction of electrophiles with defense signal transduction

HEL can be activated by salicylic acid, ethylene, and jasmonic acid (Norman-Setterblad et al., 2000; Potter et al., 1993; Reymond et al., 2000; Thomma et al., 1998; van Wees et al., 1999) and is thus a good reporter gene with which to investigate the interaction of electrophiles with these well-established signals. In order to investigate the effects on HEL gene expression, three signaling mutants, ein2-1, coi1-1 and sid2-1, were treated with MVK. MVK powerfully activated HEL expression in all three mutant backgrounds, indicating that, for this gene, an input pathway to gene activation involving electrophiles can be envisaged. The results are highly consistent with a previous report concerning HEL regulation. Much, but not all, HEL (PR4) activation during fungal pathogenesis was abolished in plants deficient in jasmonate perception (Thomma et al., 1998), meaning that there may be other input pathways to the activation of the gene. Our data suggest that electrophiles may be the candidates to provide such input. Additionally, there are strong indications of resistance phenomena and defense gene activation that are fully independent of jasmonate, salicylate, and ethylene pathways (e.g. Roetschi et al., 2001; Thara et al., 1999).

The above results showed the ability of electrophiles to affect HEL gene expression, but could these compounds also affect the levels of small signal molecules? To test this, the effect of acrolein on biochemical components of the JA signaling pathway was quantified. Levels of free JA increased in response to acrolein treatment, but even larger increases in the levels of unesterified cyclopentenone jasmonates (OPDA and dinor OPDA) were measured (Figure 4). Interestingly, both OPDA and dinorOPDA are themselves electrophiles. It is thus clear that exogenous electrophiles (e.g. acrolein) can influence endogenous electrophile levels. Increased levels of cyclopentenone jasmonates could most likely come from the electrophile-stimulated lipase activity on esterified cyclopentenone pools (Stelmach et al., 2001) or from de novo synthesis. Increases in allene oxide synthase transcript level after treatment with 2(E)-hexenal have been observed (Bate and Rothstein, 1998). Alternatively, acrolein could cause general damage to cell membranes, leading to de-esterification of acyl groups. Whatever the mechanism, the results warrant a further investigation into what appears to be a previously unrecognized regulatory input controlling cyclopentenone levels.

Figure 4.

Elevation of the levels of jasmonates after exposure to acrolein.

Plants were enclosed with solvent-free volatiles (10 µmol l−1 air volume) for 1 h and then placed in fresh air under constant light for a further 3 h prior to oxylipin analysis. Values are given as ±SEM (n = 3); FW = fresh weight.

The molecular mechanism(s) by which the α,β-unsaturated carbonyl compounds activate HEL gene expression needs to be addressed. The present results do not allow us to rigorously determine whether cytotoxicity is causally linked to the induction of HEL expression by electrophiles. It is not possible to exclude the possibility that very low doses of these compounds could cause the formation of minute and difficult-to-detect lesions similar to those observed in some pathosystems (Alvarez et al., 1998). Whether or not the regulation of gene expression is linked to cell damage, both effects of electrophiles are highly relevant to the diseased state: host cell collapse and defense gene expression being typical in many pathosystems.

There are several molecular mechanisms by which electrophiles can exert their effects on the cells. First, the ability of these compounds to stimulate phase II gene expression in animal cells correlates tightly with their reactivity towards thiol groups (Dinkova-Kostova et al., 2001). Electrophile treatment may thus cause depletion of cellular reductants, and this could then activate a redox-based signal pathway, perhaps by altering cellular reduction potentials. We favor this explanation because such a broad range of electrophiles activate plant genes. Secondly, a covalent reaction with key signaling proteins has been observed in vertebrate cells, for example, in the case of cyclopentenone prostaglandins (reviewed in Straus and Glass, 2001). This can lead to post-translational modifications in protein activity. Yet other possibilities will have to be considered equally in elucidating the route(s) to gene activation. For example, in vivo intermolecular reactions between electrophiles or reaction of electrophiles with other small molecules could potentially generate an array of new ‘foreign’ molecular forms some of which might be toxic and/or have biologic activities.

Parallels between Pst and electrophile-inducible gene expression

Electrophile production is not simply the production of toxins. Clearly, the compounds have powerful and selective biologic activities which have been investigated herein at the level of defense-related gene expression. Treatment of plants with methyl vinyl ketone (MVK) induced the expression of a surprisingly wide range of genes involved in defense (HEL, PGIP), oxidative stress (GST1, GST5, GST8, and APX1), β-oxidation (ACX1, ACX2, KAT2/PED1, and MFP2), aromatic amino acid metabolism/tryptophan synthesis (e.g. ASA1, ASB, TSA, TSB, etc.), calcium signaling (TCH1 and TCH3), and various other signaling/stress responses including ethylene synthesis (ACO1). These observations show that gene activation is not restricted to genes like GST1, which are involved, primarily, in detoxification processes. Remarkably, all of these genes were upregulated significantly in response to bacterial infection, and several were more highly upregulated in the PstavrRpm1 interaction than during infection with virulent Pst (Table 2). Furthermore, only one MVK-inducible gene that was also not induced during Pst/PstavrRpm1 infection was identified: ROF1. These observations are striking; transcript profiles for MVK treatment and bacterial pathogenesis (particularly the incompatible response to PstavrRpm1) are highly similar. It is well known that biologically active electrophile species like 2(E)-hexenal are generated at the host–pathogen interface and produced in greater quantity in incompatible than in compatible interactions (e.g. Croft et al., 1993). Small volatile electrophiles such as ethyl vinyl ketone (penten-3-one) are released by wounded Arabidopsis leaves (van Poecke et al., 2001), and several electrophilic oxylipins have already been shown to accumulate in Pst-infected Arabidopsis, again in greater quantity in the incompatible interaction (Vollenweider et al., 2000). Particularly interesting among genes commonly upregulated by MVK and Pseudomonas are those involved in lipid metabolism and β-oxidation. KAT2 (PED1), for example, is a gene encoding a thiolase with demonstrated importance in β-oxidation (Germain et al., 2001; Hayashi et al., 1998). The several tryptophan pathway genes, which are upregulated by MVK and Pst (ASA1, ASB, TSA, TSB), also indicate that treatment with an electrophile might stimulate the synthesis of low-molecular-mass defense-related metabolites, such as alkaloids. Similarly, the effects of MVK treatment on genes known to be involved in calcium sensing, i.e. TCH1 and TCH3, are noteworthy.

MVK treatment cannot fully mimic the pathogen response at the level of gene expression. It is noteworthy that MVK treatment failed to upregulate several key genes including PR1, PR3, and HIN1 and genes such as COMT, CCOAMT, and CM1, which are involved in defense and aromatic amino acid metabolism, respectively. MVK treatment clearly cannot mimic the effect of Pseudomonas on defense genes like PR1, which is regulated, in large part, by the salicylic acid pathway (Glazebrook, 2001). This corroborates the observation that MVK can induce HEL expression in the salicylic acid mutant sid2-1 (Figure 3). Finally, it is noteworthy that the avirulent pathogen PstavrRpm1 is responsible for the rapid reduction of Fv/Fm in Arabidopsis leaves. Defense signaling in many natural pathosystems takes place in the context of host cell damage and not in entirely healthy tissues such as are often employed when plants are treated with signal compounds like JA under simplified experimental conditions. Together, the strong parallels in MVK- and pathogen-inducible gene expression lend weight to the hypothesis that electrophilic compounds containing α,β-unsaturated carbonyl groups could, along with other signals including ROS, play central regulatory roles in host–pathogen interactions (Farmer, 2001; Vollenweider et al., 2000). Based on the results, we postulate that one or more electrophiles are produced in Arabidopsis in response to pathogens like Pst. One possibility is that acrolein itself or closely related molecules like MVK and EVK accumulate and affect defense-related gene expression. Finally, our results provide a basis with which to test this and related hypotheses by identifying a variety of electrophile-sensitive reporter genes.

Reactive electrophile species

Early events in pathogen perception or response to environmental stresses include the production of reactive oxygen species (ROS; Alvarez et al., 1998) and nitric oxide (Delledonne et al., 2001). These molecular species are important in the disease process, although their exact roles are still under investigation (Torres et al., 2002). Increasing evidence shows that molecules now characterized as electrophiles accumulate during pathogenesis or in response to other stresses (e.g. Bate and Rothstein, 1998; Croft et al., 1993; Howe and Schilmiller, 2002; Vollenweider et al., 2000). Many of the compounds considered in the present study are likely to be highly reactive, unstable in vivo, and of potential importance as regulators. In analogy to ROS, the term RES (reactive electrophile species) has been used (Farmer, 2001) to indicate this fact. The term is not restricted to α,β-unsaturated carbonyl-group-containing molecules because other chemical configurations can confer electrophilic properties to the molecules. Along with ROS and reactive nitrogen species, RES may play central roles in defense gene regulation and might contribute to events leading to host (and pathogen) cell damage, a process typical of responses to avirulent pathogens (May et al., 1996; Shirasu and Schultze-Lefert, 2000). Diverse enzymatically and non-enzymatically generated RES may play core roles in the control of gene expression at the host–pathogen interface. A future challenge will be to identify the relative contribution of the many potential molecules involved.

Experimental procedures

Plants, bacteria and fluorescence measurements

Arabidopsis thaliana ecotype Columbia-0 was cultivated under 150 µmol m−2 sec−1 light (23°C, 70% relative humidity) for 9 h and darkness (18°C, 65% relative humidity) for 15 h unless otherwise stated. For infiltration into plants, plastic insulin syringes (1 ml) were used to gently infiltrate 10 µl solutions or suspensions (made by brief sonication in water) of freshly prepared fatty acids and oxylipins into the abaxial side of living Arabidopsis leaves. No solvents, detergents, or other additives were employed in any of the infiltrations. During the experiments, leaves were never detached from the plants. In similiar experiments, plants were exposed to volatiles either in 1 l glass jars or in 15.5 l hermetic Plexiglass boxes. For fluorescence measurements, the ratio Fv/Fm (Björkman and Demming, 1987; Krause and Weis, 1991) was monitored using a plant efficiency analyzer (PEA MK2, Hansatech Intruments Ltd., King's Lynn, Norkolk, UK). The excitation wavelength used was 650 nm. All treatments shown in Table 1 were initiated at the identical time of day and identical light regime. For infiltration experiments in Table 1(a), plants were infiltrated at the beginning of the light cycle and Fv/Fm measured 15 h later. For volatile experiments, plants were either treated under the same conditions or, as indicated, harvested after 3 h in constant light (120 µmol m−2 sec−1) in which Fv/Fm was measured. The effects of dark adaptation (Krause and Weis, 1991) were considered not large enough to influence the interpretation of results because standard errors within an infiltration experiment often exceeded the difference in light- and dark-adapted control plants. Pseudomonas syringae pv. tomato DC3000 (Pst) and Pst with the avirulence gene avrRpm1 (PstavrRpm1) were employed (Debener et al., 1991; Bisgrove et al., 1994). Leaf tissue was extracted 9 h after infiltration of 2.5 × 105 colony-forming units in 10 µl of 10 mm MgCl2 (Vollenweider et al., 2000). Bacterial growth and decrease in Fv/Fm are shown in Figure 5.

RNA detection and microarrays

RNA detection experiments involving RNA gel blotting were conducted as described previously (Vollenweider et al., 2000). The ESTs used as probes in RNA gel blot experiments were: for HEL, GenBank accession U01880; for Lhb1B2, accession R89981. Microarray experiments were performed as described (Reymond et al., 2000); full data are posted on the Web at To estimate the confidence of cut-off values of transcript induction or repression, we performed a negative control hybridization using two independent untreated control samples. This experiment was repeated thrice on a 12K Arabidopsis microarray and showed that the probability of a ratio being greater than 1.5 by chance was less than 0.01, and that of a ratio greater than 2 less than 0.001.

Oxylipin quantification and chemicals

Jasmonates were quantified according to Weber et al. (1997). Racemic JA, racemic OPDA, fatty acid hydroperoxides, fatty acid hydroxides, epoxides, and KODEs were from Cayman Chemical Co. (Ann Arbor, MI). Linolenic acid 13-ketotriene was synthesized as described (Vollenweider et al., 2000). Traumatin was a gift from H. Gardner and B. Vick. Linolenic aldehyde, (9(Z),12(Z),15(Z)-octadecatrienal), was synthesized from linolenic acid according to Fehrentz and Castro (1983) and purified by liquid chromatography. Electron ionization mass spectrometry (70 eV electron potential) of the underivatized molecule yielded M+ = 262 (4%); m/z = 233 (2%), 206 (7%), 135 (7%), 108 (26%), and 79 (100%). Mass spectrometry of the underivatized molecule yielded M+ = 140 (0.5%); m/z = 111 (8%), 69 (64%), and 55 (100%). All other compounds were purchased form Aldrich Chemical Company (Buchs, Switzerland).


H. Gardner and B.Vick generously supplied traumatin. We thank R. Strasser for guidance on fluorescence measurements; B. Humair and H. Weber for help in performing jasmonate measurements and B. Künstner for expert care of plants; B. Proebsting and H. Weber for advice on the manuscript; and J-J. Pernet for generous help in preparing figures. This work was supported by the Swiss National Science Foundation (NCCR Plant Survival) and the Etat de Vaud.

Supplementary Material

The following material is available from

Table S1  List of genes printed on the cDNA microarray

Table S2  Treatment with methyl vinyl ketone and Pseudomonas syringae pv. tomato DC3000