Altering glucosinolate profiles modulates disease resistance in plants


  • Günter Brader,

    1. Division of Genetics, Department of Biological and Environmental Sciences, Faculty of Biosciences, Viikki Biocenter, University of Helsinki, PO Box 56, FIN-00014 Helsinki, Finland, and
    Search for more papers by this author
  • Michael Dalgaard Mikkelsen,

    1. Plant Biochemistry Laboratory, Department of Plant Biology, Center for Molecular Plant Physiology (PlaCe), The Royal Veterinary and Agricultural University, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Copenhagen, Denmark
    Search for more papers by this author
  • Barbara Ann Halkier,

    1. Plant Biochemistry Laboratory, Department of Plant Biology, Center for Molecular Plant Physiology (PlaCe), The Royal Veterinary and Agricultural University, Thorvaldsensvej 40, DK-1871 Frederiksberg C, Copenhagen, Denmark
    Search for more papers by this author
  • E. Tapio Palva

    Corresponding author
    1. Division of Genetics, Department of Biological and Environmental Sciences, Faculty of Biosciences, Viikki Biocenter, University of Helsinki, PO Box 56, FIN-00014 Helsinki, Finland, and
    Search for more papers by this author

(fax +358 9 19159076; e-mail


Plant diseases are major contributing factors for crop loss in agriculture. Here, we show that Arabidopsis plants with high levels of novel glucosinolates (GSs) as a result of the introduction of single CYP79 genes exhibit altered disease resistance. Arabidopsis expressing CYP79D2 from cassava accumulated aliphatic isopropyl and methylpropyl GS, and showed enhanced resistance against the bacterial soft-rot pathogen Erwinia carotovora, whereas Arabidopsis expressing the sorghum CYP79A1 or over-expressing the endogenous CYP79A2 accumulated p-hydroxybenzyl or benzyl GS, respectively, and showed increased resistance towards the bacterial pathogen Pseudomonas syringae. In addition to the direct toxic effects of GS breakdown products, increased accumulation of aromatic GSs was shown to stimulate salicylic acid-mediated defenses while suppressing jasmonate-dependent defenses, as manifested in enhanced susceptibility to the fungus Alternaria brassicicola. Arabidopsis with modified GS profiles provide important tools for evaluating the biological effects of individual GSs and thereby show potential as biotechnological tools for the generation of plants with tailor-made disease resistance.


More than 50 000 plant natural products have been identified, of which many play key roles in the defense against herbivores and potential pathogens (Osbourn et al., 2003). One category of compounds implicated in plant defense consists of the amino acid-derived glucosinolates (GSs), which are characteristic natural products of the Brassicaceae family including Brassica crops and the model plant Arabidopsis. In human nutrition, several GS degradation products have been shown to possess potent anti-carcinogenic properties (Johnson, 2002). Recently, major progress has been made in the identification of genes responsible for biosynthesis of the core GS structure (Kliebenstein et al., 2005; Mikkelsen et al., 2002, 2004; Wittstock and Halkier, 2002). The first committed step is the conversion of precursor amino acids to their corresponding aldoximes by substrate-specific cytochromes P450 belonging to the CYP79 family. The post-aldoxime enzymes, although specific for the GS pathway, have low substrate specificity for the amino acid side chain and efficiently convert even exogenous aldoximes to the corresponding GSs (Halkier, 1999; Mikkelsen et al., 2002; Wittstock and Halkier, 2002). Consequently, in planta accumulation of specific GSs can be achieved by over-expressing CYP79 genes. For example, over-expressing the endogenous, phenylalanine-specific CYP79A2 in Arabidopsis (subsequently referred to as CYP79A2 plants) leads to accumulation of high levels of benzyl-GS (BGS), which is only present in minute amounts in seeds of wild-type plants (Wittstock and Halkier, 2000). The over-expression of CYP79A2 does not alter the profile of the endogenous GSs in Arabidopsis leaves (Mikkelsen et al., 2002; Wittstock and Halkier, 2000). Introduction of exogenous CYP79s with no functional homologues in Arabidopsis, e.g. from the evolutionarily related cyanogenic glucoside pathway, has resulted in accumulation of novel GSs (for review see Mikkelsen et al., 2002). In Arabidopsis, constitutive expression of the cassava CYP79D2 (CYP79D2 plants) leads to accumulation of valine-derived isopropyl-GS (IPGS) and isoleucine-derived 1-methylpropyl-GS (MPGS) (Mikkelsen and Halkier, 2003). Similarly, constitutive expression of the Sorghum bicolor CYP79A1 (CYP79A1 plants) leads to accumulation of tyrosine-derived p-hydroxybenzyl-GS (p-OHBGS) (Bak et al., 1999). Neither IPGS, MPGS nor p-OHBGS are natural constituents of Arabidopsis ecotype Columbia-0 (Col-0), nor does the accumulation of these affect the composition of the endogenous GSs (Bak et al., 1999: Mikkelsen and Halkier, 2003).

Several studies have suggested that GSs or rather their degradation products are involved in plant defense against insects and pathogens. Upon tissue disruption, e.g. as a consequence of wounding or pathogen attack, the GSs, which are stored in the vacuole, are hydrolyzed by endogenous β-thioglucosidases (myrosinases) (Rask et al., 2000) to primarily nitriles and isothiocyanates (ITCs). The ITCs in particular play important roles as repellents against insects (Agrawal and Kurashige, 2003; Halkier, 1999; Rask et al., 2000), and as volatile attractants to specialized insects (Agrawal and Kurashige, 2003; Nielsen et al., 2001; Ratzka et al., 2002). Less documentation exists on the role of GSs in plant–pathogen interactions. In vitro studies have demonstrated that ITCs can inhibit growth of fungal and bacterial pathogens (Brader et al., 2001; Manici et al., 1997; Mithen et al., 1986; Tierens et al., 2001). Earlier in vivo studies of disease resistance of Brassica napus cultivars bred for altered levels of GSs were not conclusive with respect to the role of the individual GSs (Giamoustaris and Mithen, 1997; Li et al., 1999; Wretblad and Dixelius, 2000).

It is of major interest to agriculture to increase the disease resistance of crop plants to improve yield and minimize the use of pesticides. A possible strategy is to increase the accumulation of specific natural plant defense compounds, such as GSs. Here, we explore the potential of engineering specific GS profiles to improve plant disease resistance. We assess the disease resistance of transgenic CYP79A2, CYP79A1 and CYP79D2 Arabidopsis plants accumulating high levels of BGS, p-OHBGS or IPGS and MPGS, respectively (Bak et al., 1999; Mikkelsen and Halkier, 2003; Petersen et al., 2001; Wittstock and Halkier, 2000), towards the three plant pathogens Alternaria brassicicola, Pseudomonas syringae and Erwinia carotovora which employ distinct pathogenesis strategies. We demonstrate how engineering of specific GS profiles alters disease resistance against specific pathogens either by direct toxic effects or by modulating plant defense signaling.


In vitro toxicity of GSs and their breakdown products

To explore the defense potential of the introduced novel GSs, we analyzed the toxicity of the corresponding GS breakdown products in vitro. In Arabidopsis Col-0, the genetic background of the transgenic lines used in this study, GSs are predominantly converted to ITCs upon degradation (Lambrix et al., 2002). Indeed, the transgenic CYP79A1, CYP79A2 and CYP79D2 plants contained the expected large amounts of the ITCs p-hydroxybenzyl isothiocyanate (p-OHBITC), benzyl isothiocyanate (BITC) and isopropyl isothiocyanate (IPITC) as major degradation products after wounding, but also 10–27% of the corresponding nitriles (Figure 1). Accordingly, we assessed the in vitro inhibitory effects of p-OHBGS, BGS, IPGS, their corresponding ITCs and nitriles as well as a mixture of the GSs together with myrosinase to simulate the effect of tissue disruption in planta using the three test pathogens E. carotovora subsp. carotovora SCC1, P. syringae pv. tomato DC3000 and A. brassicicola. Intact GSs, nitriles and p-hydroxybenzyl alcohol as a further degradation product of the relatively unstable p-OHBITC did not significantly inhibit growth of any of the test pathogens at concentrations up to 1000 μm. The major degradation products of BGS and IPGS, i.e. BITC and IPITC, were potent inhibitors of pathogen growth, whereas p-OHBGS in the presence of myrosinase had no growth inhibitory effect on the test pathogens at 1000 μm (Table 1). BITC and IPITC were the most potent inhibitors of P. syringae growth, with 50% growth inhibition (IC50) at 55 μm (95% fiducial limits from 40–95 μm) and 402 μm (318–523 μm), respectively. In comparison, IC50 values for E. carotovora subsp. carotovora were 224 μm (164–343 μm) and 711 μm (537–915 μm), and those for A. brassicicola 444 μm (407–499 μm) and approximately 1000 μm for BITC and IPITC, respectively. Although the IC50 values were relatively high, a comparison with in vivo concentrations of GSs in the transgenic plants (Table 1) suggests that CYP79A2 plants produce sufficient levels of BGS to inhibit the growth of all three test pathogens in planta, while CYP79D2 plants potentially accumulate sufficient levels of IPGS and MPGS to inhibit growth of the two bacterial pathogens.

Figure 1.

 Analysis of GS degradation products in wounded leaves.
Leaves of CYP79A1, CYP79A2, CYP79D2, vector control and wild-type plants were crushed in water and extracted after 5 min with dichloromethane. The lipophilic phase containing the GS degradation products isothiocyanates (white), alcohols (black) and nitriles (gray) were analyzed by GC–MS as described in Experimental procedures. The degradation products derived from the introduced GSs (Int) of transgenic plants or the combined degradation products of the naturally occurring major aliphatic GSs 4-methylthiobutyl-GS and 4-methylsulfinyl-butyl-GS (Nat) are indicated, and correspond to the average of three to four replicates ± standard error.

Table 1.   Novel GSs in transgenic Arabidopsis and in vitro antimicrobial activity of the GSs and their degradation products towards three pathogens
GSPlantIn plantaIC50 (FL)
Erwinia carotovora subsp. carotovoraPseudomonas syringaeAlternaria brassicicola
GS + myrITCGS + myrITCGS + myrITC
  1. In planta: average GS concentration (nmol g−1 fresh weight) in 4-week-old rosette leaves ±standard error (n = 10). These GSs are not detectable in the leaves of wild-type plants. IC50: in vitro concentration (μm) required for 50% growth inhibition evaluated after 15 h (E. carotovora subsp. carotovora) or 18 h (P. syringae) at 28°C or after 48 h at 22°C (A. brassicicola) with 95% fiducial limits (FL). GS + myr: GS plus 1 unit ml−1 myrosinase. ITC: major ITC degradation product of the indicated GS. ND: not determined.

inline imageCYP79A13400 (±401)>1000ND>1000ND>1000ND
inline imageCYP79A2790 (±92)248 (202–365)224 (164–343)67 (46–117)55 (40–95)556 (40–749)444 (407–499)
inline imageCYP79D2640 (±162)ND711 (537–915)ND402 (318–523)ND∼1000

Arabidopsis with specifically altered GS profiles display increased resistance to bacterial pathogens

The potential of novel GS profiles to enhance resistance to bacterial pathogens in planta was tested in bioassays. CYP79A1, CYP79A2, CYP79D2, vector control and wild-type plants were infected with, respectively, the aggressive strain SCC1 of the necrotroph E. carotovora subsp. carotovora and the virulent strain DC3000 of the hemi-biotroph P. syringae. A clear increase in resistance to E. carotovora subsp. carotovora, typically causing complete maceration of plant tissues upon infection, was observed in CYP79D2 plants. While 63 and 66% of the wild-type and vector control plants died after infection, the mortality rate of CYP79D2 plants was 30% (Figure 2a). This enhanced resistance was already manifested within the first 24 h post-inoculation (p.i.), with a differential spreading of the maceration in CYP79D2 and wild-type plants (Figure 2b). In CYP79D2 plants, the maceration usually stopped and dried out, while it continued in the wild-type. At 48 h p.i., the infection in wild-type plants had often spread to systemic leaves, stems and buds and subsequently led to plant death, whereas CYP79D2 plants developed normally (Figure 2a,c). Neither CYP79A1 nor CYP79A2 plants displayed enhanced resistance to E. carotovora subsp. carotovora (Figure 2a).

Figure 2.

 Measurement of bacterial disease resistance in Arabidopsis plants with engineered GS profiles.
Three to four-week-old CYP79A1, CYP79A2, CYP79D2, vector control and wild-type plants were infected with 104 cells of Erwinia carotovora subsp. carotovora SCC1 by pipetting 5 μl bacterial suspension on leaves wounded with a pipette tip (a–c) or infiltrated with 5–10 × 105 CFU ml−1Pseudomonas syringae DC3000 (d,e) and kept at >95% humidity after infection.
(a) Percentage of dead plants 5 days after treatments. The data represent the mean of three experiments of 28 plants each, ±standard error. Asterisks indicate a significant difference (P < 0.05) between CYP79D2 and wild-type using a non-parametric Mann–Whitney test.
(b) Spreading of maceration. Lesion diameters were measured 24 h p.i. The data represent the means of 16 leaves ± standard error.
(c) Spreading of disease 48 h p.i. Spreading of infection to the bud is marked with red arrows; sustained local infection is indicated with white arrows.
(d) Growth of P. syringae in infected wild-type (bsl00067), vector control (•), CYP79A1 (bsl00066), CYP79A2 (bsl00063) and CYP79D2 (bsl00001) plants. Each point represents the mean (±standard error) of 16 replicates. Significant different values in each experiment (P < 0.05) were calculated by one-way anova with Tukey's HSD test and are indicated by asterisks.
(e) Disease symptoms in vector controls and CYP79A2 plants 3 days p.i.

Although the CYP79D2 plants displayed increased resistance to E. carotovora subsp. carotovora, infection by P. syringae was not inhibited significantly (Figure 2d). However, CYP79A1 and CYP79A2 plants showed increased resistance to P. syringae. A 10-fold reduction in bacterial growth was observed in CYP79A2 plants accumulating BGS compared to wild-type, and the plants showed less severe disease symptoms (Figure 2d,e). Growth of P. syringae was slightly retarded on CYP79A1 plants accumulating p-OHBGS (Figure 2d). The results demonstrate that expression of specific CYP79 genes confers enhanced resistance to specific bacterial pathogens.

Enhanced susceptibility to A. brassicicola in plants with novel GS profiles

To investigate whether other groups of pathogens were affected by altered GS profiles, we employed the necrotrophic fungal pathogen A. brassicicola. The fungus causes significant cellular disruption during infection, and therefore an efficient degradation of GSs to the toxic ITCs is expected. Surprisingly, CYP79A1 and CYP79A2 plants showed enhanced susceptibility to A. brassicicola compared to wild-type. In contrast, CYP79D2 plants were as resistant as wild-type and vector control plants, which displayed only very limited disease development (Figure 3a). Disease symptoms were observed as early as 3–5 days p.i. in CYP79A1 plants. After 6–8 days, both CYP79A1 and CYP79A2 plants displayed severely enhanced disease symptoms (Figure 3a). The increased symptoms were accompanied by enhanced fungal growth, which was quantified as the relative amount of fungal DNA by quantitative PCR analysis (Figure 3b). The relative amount of fungal DNA compared to plant DNA was almost 1000 times higher in CYP79A1 plants and about 10 times higher in CYP79A2 plants than in the wild-type control (Figure 3b). The results do not correlate with the in vitro toxicity data (Table 1) and it is therefore unlikely that they can be attributed to direct toxic effects of the GS degradation products.

Figure 3.

Alternaria brassicicola infection of leaves of Arabidopsis plants with engineered GS profiles.
Three to four-week-old CYP79A1, CYP79A2, CYP79D2, vector control and wild-type plants were inoculated with 5 μl A. brassicicola spore suspensions (5 × 105 spores ml−1).
(a) Symptoms 7 days p.i.
(b) Quantification of fungal growth 7 days p.i. by quantitative PCR. Bars represent the average amount of fungal DNA in relation to plant DNA and are the means of six replicates ± standard error. Significantly different values indicated by different letters (P < 0.05) were calculated by one-way anova with Tukey's HSD test.

Altered GS profiles affect plant defense signaling

The apparent discrepancy between in vitro and in planta data suggested that other plant defense mechanisms were compromised in plants with modified GS profiles. Resistance to A. brassicicola requires induction of jasmonate (JA)-dependent defenses, and Arabidopsis mutants deficient in JA sensing, such as coi1, exhibit increased susceptibility to A. brassicicola (Thomma et al., 2001). To determine whether the increased susceptibility to A. brassicicola in the CYP79A1 and CYP79A2 plants was caused by defective JA-mediated defenses, we assessed the expression of PDF1.2, a marker gene for the JA signaling pathway (Kunkel and Brooks, 2002; Penninckx et al., 1996). Comparison of methyl jasmonate (MeJA)-induced PDF1.2 expression in CYP79A1, CYP79A2 and vector control plants by quantitative RT-PCR showed that induction of PDF1.2 was reduced to 27 and 66% of wild-type levels in CYP79A1 and CYP79A2 plants, respectively (Figure 4).

Figure 4.

 Analysis of PDF1.2 expression in MeJA-treated CYP79A1 and CYP79A2 plants.
Local leaf samples were collected from CYP79A1, CYP79A2 and vector control plants at 0 h (0) and 24 h after treatment with MeJA (MJ). Total RNA was extracted and analyzed by quantitative RT-PCR for PDF1.2 expression. The data were normalized to Actin2 expression. Values correspond to the average of three replicates ± standard error.

The reduced ability of the CYP79A1 and CYP79A2 plants to induce JA-mediated defenses indicates that this signal pathway is suppressed by accumulation of the novel GSs. We hypothesized that the degradation products of the introduced aromatic BGS or p-OHBGS could act as signaling compounds and thereby modify the wild-type defense responses. The aromatic signaling molecule salicylic acid (SA) is a known inducer of systemic acquired resistance (SAR) (Glazebrook, 2001; Kunkel and Brooks, 2002; Uknes et al., 1992), but it is also capable of repressing the JA-mediated defense against A. brassicicola (Kariola et al., 2005). Quantitative RT-PCR was used to investigate the signaling potential of the BGS breakdown product BITC by measuring the transcript levels of the SAR markers PR1 and PR2 in wild-type plants following treatment with SA and BITC. Both SA and BITC induced PR1 and PR2 transcript accumulation (Figure 5), although the expression level of PR1 and PR2 was lower following BITC treatment. This treatment did not cause visible damage on the application sites and did not induce increased accumulation of SA, which indicates that the observed effect is not mediated by SA (data not shown). These data are in agreement with the hypothesis that BITC mimics a signaling molecule in the SA defense pathway. If aromatic ITCs act as signaling compounds, wounding of CYP79A1 and CYP79A2 plants should result in enhanced expression of SA-responsive genes. Indeed, quantitative RT-PCR analyses showed an increase in wound-induced expression of PR1 and PR2 in leaves of CYP79A1 and CYP79A2 plants compared to vector control plants (Figure 6a). In addition, even the basal expression of PR2 and PR1 is increased in the CYP79A1 plants. CYP79A2 plants do not display increased basal levels of either PR1 or PR2, but both genes show enhanced induction upon wounding compared to wild-type (Figure 6a). SA levels were increased in both CYP79A1 and CYP79A2 plants and correlated with the elevated levels of PR1 and PR2 (Figure 6b). The results show that enhanced susceptibility to A. brassicicola in CYP79A1 and CYP79A2 plants is accompanied by elevated levels of aromatic GSs, elevated levels of SA, increased expression of the SA-responsive genes PR1 and PR2, and reduced expression of the JA-responsive marker, PDF1.2. Similarly, the data offer an explanation for the increased resistance to P. syringae, which is known to require activation of SA-dependent defenses.

Figure 5.

 Effect of BITC on PR1 and PR2 expression.
Three to four-week-old leaves were collected from wild-type Arabidopsis 24 h after treatment with 0.05% Tween-20 plus 0.5% ethanol (control), SA (2 mm) or BITC (2 mm) or untreated (0). Total RNA was extracted and analyzed by RT-PCR for PR1 and PR2 expression. The data were normalized to Actin2 expression. Values correspond to the average of three replicates ± standard error.

Figure 6.

PR1 and PR2 expression and SA levels in plants containing aromatic GSs.
Three to four-week-old leaves of CYP79A1, CYP79A2 or vector control plants were analyzed at 0 h (0) and 24 h after wounding (w).
(a) Total RNA was extracted and analyzed by quantitative RT-PCR for PR1 and PR2 expression. The data were normalized to Actin2 expression. Values correspond to the average of three replicates ± standard error. (b) Free SA content in wounded and control leaves. The values represent the average of three replicates ± standard error. Significant differences (P < 0.05) were calculated by one-way anova with Tukey's HSD test and are indicated by different letters.


The present study demonstrates the functional role of specific GSs as defense compounds in planta, and shows that engineering of novel GS profiles by expression of specific CYP79 genes can alter disease resistance. The changes in disease resistance depended on both the pathogen and the GS profile of the host. Production of the aliphatic IPGS and MPGS resulted in enhanced resistance to the bacterial necrotroph E. carotovora subsp. carotovora, whereas the aromatic GSs had no significant effect. In contrast, enhanced resistance to the bacterial hemi-biotroph P. syringae was provided by aromatic GSs, in particular BGS, but not by the aliphatic GSs.

What is the mechanism of bacterial disease resistance and the observed specificity of this resistance? Our data suggest that the enhanced resistance is partly a consequence of the direct toxic effects of GS breakdown products. This is suggested by in vitro toxicity tests showing that growth of P. syringae was strongly inhibited by BITC or by BGS plus myrosinase, and moderately inhibited by IPITC (Table 1). Although P. syringae only caused limited tissue damage at early stages of infection (Tierens et al., 2001), the high toxicity of BITC suggests that levels sufficient to reduce pathogen growth in planta were reached. In contrast, the quantity of the novel aliphatic GSs in CYP79D2 plants was not sufficient to generate increased resistance to P. syringae (Figure 2d,e). This may be explained by the higher IC50 values of the aliphatic GS breakdown products compared to those of the aromatic GS breakdown products, and suggests that the increased resistance of CYP79A2 plants to P. syringae may be to a certain extent caused by direct toxic effects of the GS breakdown products.

Resistance to E. carotovora subsp. carotovora was increased in the CYP79D2 plants accumulating aliphatic IPGS and MPGS, whereas CYP79A1 and CYP79A2 plants accumulating aromatic GSs exhibited wild-type levels of resistance (Figure 2a–c). This is unexpected based on the in vitro analysis that suggests a higher toxicity of BITC compared to IPITC. Possibly, the high volatility of the IPGS and MPGS breakdown products results in a rapid decrease in the actual concentrations of IPITC in vitro (IPITC has a boiling point of 28–30°C). This hypothesis is supported by results showing that IPITC, in contrast to BITC, cannot be detected in the in vitro test medium at 15 h p.i. (data not shown). This suggests that the actual IC50 value of IPITC is significantly lower than determined by the in vitro assays. Upon E. carotovora subsp. carotovora infection, the maceration process progresses very rapidly and is completed within 24 h in infected leaves (Kariola et al., 2003) (Figure 2b,c). Accordingly, the difference in disease resistance between CYP79D2 and CYP79A2 plants may reflect the more efficient disturbance of the pathogen progression by a continuous release of volatile IPITC and 1-methylpropyl ITC than by release of the less diffusible BITC. The difference between E. carotovora subsp. carotovora and P. syringae might be explained by the relatively higher sensitivity of E. carotovora subsp. carotovora towards IPITC (Table 1). Alternatively, IPITC could negatively influence the activation of virulence genes of E. carotovora subsp. carotovora required for prolonged maceration and successful infection in planta.

Apart from possible direct toxic effects of the GS breakdown products, our results indicate that the GSs and their breakdown products further affect the observed resistance phenotypes by modulation of plant defense signaling. This is evidenced by results with both P. syringae and A. brassicicola. Resistance to the hemi-biotrophic pathogen P. syringae, which in wild-type plants is triggered through the SA-dependent defense pathway, was increased both in the CYP79A1 and CYP79A2 plants (Figure 2d,e). Unlike the degradation products of BGS (CYP79A2 plants), the degradation products of p-OHBGS (CYP79A1 plants) did not show in vitro activity against P. syringae (Table 1). Yet CYP79A1 plants exhibited enhanced resistance to P. syringae. This resistance was accompanied by strong induction of the SAR marker genes PR1 and PR2 (Figure 6a). A weaker induction of SAR markers was observed in CYP79A2 plants (Figure 6a). This suggests that the observed higher resistance in CYP79A2 plants is due to a combination of the direct toxicity of BITC and activation of SAR.

Activation of the SA-mediated defense by aromatic GSs is remarkable. Our results demonstrate that SA induction of PR1 and PR2 expression can be partly mimicked by BITC (Figure 5). The lack of SA accumulation after external BITC treatment suggests that the signal acts downstream of SA. However, elevated levels of SA in wounded CYP79A1 and CYP79A2 plants indicate the presence of further increased PR gene induction via SA after wounding and subsequent release of breakdown products of aromatic GSs. Alternatively, the expression of CYP79A1 and CYP79A2 may perturb a negative feedback regulation of SA biosynthesis, as both pathways share biosynthetic intermediates. A possible mode of action of ITCs involves disturbance of redox homeostasis of the cell, as BITC has been shown to induce oxidative stress and cell death in animal cells and to stimulate anti-oxidative enzymes (Miyoshi et al., 2004; Nakamura et al., 2002; Zhang et al., 2005). Although BITC treatment does not cause any visible damage in Arabidopsis leaves, it is possible that application of BITC causes microscopic tissue damage, which in turn perturbs oxidative signaling in leaves. Modification of the redox balance in plant cells can lead to an activation of PR gene expression, possibly due to NPR1 modification (Pieterse and Van Loon, 2004). Moreover, as the p-OHBITC generated after degradation of p-OHBGS is unstable (Buskov et al., 2000), the actual signal and inducer of SA accumulation in this case could be a further metabolized aromatic product.

Interestingly, plant resistance to the necrotrophic fungus A. brassicicola could not be improved by altering the GS profiles, although in vitro assays showed clear growth inhibition of the fungus by BITC (Table 1 and Figure 3). None of the plants with novel GS profiles showed an enhanced resistance towards this pathogen, and, unexpectedly, introduction of aromatic GSs led to increased susceptibility. This indicates that, in this case, the signaling potential of the breakdown products is more important for resistance than the direct toxic effects. A negative correlation between GS content in Brassica napus and resistance to strains of A. brassicicola and A. brassicae was identified previously (Giamoustaris and Mithen, 1997). We did not observe a negative correlation between resistance to A. brassicicola and increased levels of the aliphatic IPGS and MPGS, but only to increased levels of aromatic GS. This is possibly because aromatic GSs alter signaling of defense pathways. Our results indicate that JA signaling is compromised in both CYP79A1 and CYP79A2 plants (Figure 4). The requirement for a functional JA signaling pathway for A. brassicicola resistance in Arabidopsis (Thomma et al., 1998, 2001) may explain the increased susceptibility of CYP79A1 and CYP79A2 plants to this pathogen. A possible scenario is that the tissue damage caused by A. brassicicola infection releases aromatic breakdown products in the CYP79A1 and CYP79A2 plants, triggering SA-mediated defenses (Figure 6), which will subsequently antagonize JA sensitivity and the JA signaling pathway (Glazebrook et al., 2003; Kariola et al., 2005; Li et al., 2004). This hypothesis is supported by the close correlation of the increased susceptibility to A. brassicicola with an enhanced induction of SA-responsive genes and repression of JA-responsive PDF1.2 in the CYP79A1 and CYP79A2 plants (Figures 3, 4 and 6). However, we cannot rule out the possibility that A. brassicicola strains adapted for growth on Brassicaceae (our strain was originally isolated from B. oleracea) may have additionally evolved mechanisms to take advantage of GS production in host plants, and the aromatic GSs or their breakdown products could promote growth of the fungus in planta.

In conclusion, we have demonstrated that generation of plants enriched for certain GSs alters disease resistance towards specific pathogens. We suggest that the increased resistance is partly due to direct toxic effects of the GS breakdown products, but also that these products can interfere with plant defense signaling so that increased accumulation of specific defense compounds (e.g. aromatic GSs) may unexpectedly result in increased susceptibility towards certain pathogens. This influence on signaling pathways illustrates the complexity of the plant defense system and might be of even greater importance in natural environments, where co-infection by wounding and (hemi-)biotrophic pathogens may occur. Introducing or eliminating specific CYP79s may enhance resistance to certain pathogens while retaining wild-type levels of resistance to other pathogens and herbivores. The results of this study highlight the importance of transgenic plants with altered GS profiles for evaluating the biological effect of individual GSs, and their potential as biotechnological tools for generation of cruciferous crop plants such as oilseed rape and Brassica vegetables with custom-designed defense strategies. Ultimately, when all genes in the GS pathway have been identified, it may be possible to utilize this approach in any plant species.

Experimental procedures

Plant material

Seeds of Arabidopsis lines were grown on a 1:1 mixture of vermiculite and peat (Finnpeat, B2 Kekkilö Oyj, Tuusula, Finland), and cultivated under a 12 h light period at 22°C with a photon flux density of 100–120 μE m−2 sec−1. Generation and characterization of the CYP79A1 (35S::CYP79A1; Bak et al., 1999), CYP79A2 (35S::CYP79A2; Wittstock and Halkier, 2000) and CYP79D2 (35S::CYP79D2; Mikkelsen and Halkier, 2003) plants have been described previously.

Chemical treatments and plant inoculations

Salicylic acid (SA; 2 mm, Sigma-Aldrich, St Louis, MO, USA), benzyl isothiocyanate (BITC; 2 mm; Sigma-Aldrich), methyl jasmonate (MeJA; 100 μm; Sigma-Aldrich) in 0.05% v/v Tween-20 and 0.5% v/v ethanol were applied as five 5 μl droplets per leaf after making a small wound with the pipette tip. Erwinia carotovora subsp. carotovora strain SCC1 (Pirhonen and Palva, 1988) was cultured overnight at 28°C in LB medium. Bacteria were harvested by centrifugation (4000 g), resuspended in 0.9% w/v NaCl, diluted to 1–3 × 106 CFU ml−1, and applied as 5 μl droplets on leaves on plants kept at >95% relative humidity after infection. Pseudomonas syringae pv. tomato DC3000 was cultivated in King's B medium, the cells were pelleted, resuspended in 10 mm MgCl2 and infiltrated into the lower surface of leaves with a syringe without a needle at a concentration of 5–10 × 105 CFU ml−1. Assessing bacterial growth was performed by plating dilution series of leaf disks ground in 10 mm MgCl2 on King's B plates containing 25 μm ml−1 rifampicin as described by Weigel and Glazebrook (2002). Alternaria brassicicola (strain 567.77; Centraalbureau voor Schimmelcultures, Utrecht, The Netherlands) was maintained on potato/carrot extract agar, and spore suspensions (5–10 × 105 conidial spores ml−1 in potato dextrose broth) were applied on leaves as 5 μl drops after making a small wound with a pipette tip. After inoculation, plants were kept at >95% relative humidity. To assess fungal growth, ten replicates of single treated and control leaves (treated with potato dextrose broth only) were frozen in liquid nitrogen on the 7th day after infection. Relative fungal biomass was determined by quantitative PCR using an ABI Prism 7000 Sequence Detection System (Applied Biosystems, Foster City, CA, USA) as described previously (Kariola et al., 2005).

Assays for antimicrobial activity

The concentrations of the test compounds resulting in 50% inhibition of microbial growth (IC50) were determined essentially as described previously (Brader et al., 2001). Briefly, the test compounds were dissolved in 50 mm phosphate buffer, pH 7, containing 0.5% ethanol and 0.1% Tween-20. E. carotovora subsp. carotovora SCC1 and P. syringae pv. tomato DC3000 cultures suspended in M9 minimal medium containing 0.2% sucrose and 0.1% casamino acids (Elbing and Brent, 2002) at 104 CFU ml−1, with or without myrosinase (1 unit ml−1), were added to serial dilutions of the test compounds in microtiter plates sealed with parafilm. Growth inhibition was evaluated after 15 h (SCC1) or 18 h (DC3000) of incubation at 28°C, and the IC50 values and the error range (95% fiducial limits) were calculated by probit analysis. For A. brassicicola, the test compounds were inoculated with spores (104 ml−1 in potato dextrose broth). Fungal growth was monitored microscopically (Tierens et al., 2001), and 50% reduction of fungal growth (IC50) was calculated after 48 h incubation at 28°C in the dark. IPITC, isobutylnitrile, BITC, benzylnitrile, p-hydroxybenzylnitrile, p-hydroxybenzyl alcohol and myrosinase (thioglucosidase, EC from white mustard) were purchased from Sigma-Aldrich, BGS (glucotropaeolin) from Merck (Darmstadt, Germany) and p-OHBG from Bioraf (Åkirkeby, Denmark).

Quantification of SA, GS and GS degradation products

SA was extracted and quantified with 13C1-SA as an internal standard using the protocol described by Baldwin et al. (1997) with a Trace-DSQ gas chromatograph–mass spectrometer (GC–MS) from Thermo (Waltham, MA, USA). GSs were determined as described previously (Brader et al., 2001). GS degradation products were extracted from 100 mg leaves as described by Lambrix et al. (2002), with 100 ng 3-hydroxy-4-methoxybenzylalcohol (Sigma-Aldrich) and isobutyl isothiocyanate (Lancaster Synthesis, Heysham, Lancashire, UK) as internal standards. Aliquots (1 μl) of samples dissolved in 25 μl dichloromethane were analyzed by GC–MS with a ZB-5 column (30 m × 0.25 mm × 0.25 μm film; Phenomenex, Torrance, CA, USA), splitless injection at 200°C, and a temperature program of 35°C for 3 min, a 12°C min−1 ramp to 96°C, a 18°C min−1 ramp to 300°C and a final 6 min hold at 300°C, an ion source temperature of 250°C and 70 eV electron energy for ionization. For analysis of hydroxy derivatives, the samples were dried, silylated with 10 μl N-methyl-N-(trimethylsilyl)-trifluoroacetamide (MSTFA, Sigma-Aldrich) at 37°C for 30 min, diluted with pyridine to 25 μl and analyzed by GC–MS as described above. Nitriles, alcohols and isothiocyanates were identified by comparison with authentic standards and comparison with mass spectra from literature (Spencer and Daxenbichler, 1980) and a mass spectral library (Wiley7, Wiley STM Databases, Chichester, West Sussex, UK). A standard for silylated p-hydroxybenzyl isothiocyanate was prepared from p-OHBG by 30 min incubation with myrosinase in 50 mm phosphate buffer, pH 7, followed by extraction with dichloromethane and silylation.

RNA analyses with real-time PCR

Isolation of total RNA was performed as described previously (Kingston, 1997). A 1 μg aliquot of total RNA treated with DNaseI (Amersham Biosciences, Piscataway, NJ, USA) was incubated for 30 min at 48°C in a 50 μl reaction with 1 μl MultiScribe reverse transcriptase (50 U μl−1; Roche, Branchburg, New Jersey, USA), 1 μl RNAse inhibitor (20 U μl−1; Roche) and 2.5 μl oligo dT primer (50 μm), 5 μl TaqMan RT buffer (10×; Roche), 11 μl MgCl2 (25 mm) and 10 μl dNTP (2.5 mm). PR1, PR2, PDF1.2 and Actin2 expression was quantified by RT-PCR using the primers 5′-AAGGGTTCACAACCAGGCAC-3′ and 5′-CACTGCATGGGACCTACGC-3′ (PR1), 5′-GCTCTCCGTGGCTCTGACAT-3′ and 5′-TCTTGAACCCACTTGTCGGC-3′ (PR2), 5′-TCTTTGCTGCTTTCGACG-3′ and 5′-AAACCCCTGACCATGTCCC-3′ (PDF1.2) and 5′-CTCCCGCTATGTATGTCGCC-3′ and 5′-CAGAATCCAGCACAATACCGGT-3′ (Actin2) using 1 μl of the reverse transcription reaction as a template.


This study was supported by the Academy of Finland (projects 79776 and 202886; Finnish Centre of Excellence Programme 2000–2005).