Role of camalexin, indole glucosinolates, and side chain modification of glucosinolate-derived isothiocyanates in defense of Arabidopsis against Sclerotinia sclerotiorum

Authors


(fax +49 931 31 86182; e-mail henrik.stotz@uni-wuerzburg.de).

Summary

Plant secondary metabolites are known to facilitate interactions with a variety of beneficial and detrimental organisms, yet the contribution of specific metabolites to interactions with fungal pathogens is poorly understood. Here we show that, with respect to aliphatic glucosinolate-derived isothiocyanates, toxicity against the pathogenic ascomycete Sclerotinia sclerotiorum depends on side chain structure. Genes associated with the formation of the secondary metabolites camalexin and glucosinolate were induced in Arabidopsis thaliana leaves challenged with the necrotrophic pathogen S. sclerotiorum. Unlike S. sclerotiorum, the closely related ascomycete Botrytis cinerea was not identified to induce genes associated with aliphatic glucosinolate biosynthesis in pathogen-challenged leaves. Mutant plant lines deficient in camalexin, indole, or aliphatic glucosinolate biosynthesis were hypersusceptible to S. sclerotiorum, among them the myb28 mutant, which has a regulatory defect resulting in decreased production of long-chained aliphatic glucosinolates. The antimicrobial activity of aliphatic glucosinolate-derived isothiocyanates was dependent on side chain elongation and modification, with 8-methylsulfinyloctyl isothiocyanate being most toxic to S. sclerotiorum. This information is important for microbial associations with cruciferous host plants and for metabolic engineering of pathogen defenses in cruciferous plants that produce short-chained aliphatic glucosinolates.

Introduction

Natural plant products play an important role in shaping biotic interactions (Kliebenstein, 2004). Plant defenses depend on constitutive and induced production of antimicrobial phytoanticipins and phytoalexins, respectively (Hammerschmidt, 1999; Tierens et al., 2001; Pedras et al., 2007; Bednarek and Osbourn, 2009). For the model plant Arabidopsis thaliana, the phytoalexin camalexin (Schuhegger et al., 2006) is a significant component in defending against necrotrophic fungal pathogens (Thomma et al., 1999; Ferrari et al., 2003). Even though the antimicrobial activity of camalexin is susceptible to fungal detoxification mechanisms (Pedras and Ahiahonu, 2002; Kliebenstein et al., 2005), approximately half of the quantitative trait loci controlling resistance to the ascomycete Botrytis cinerea are associated with the accumulation of camalexin (Rowe and Kliebenstein, 2008). Glucosinolates (GSLs), commonly referred to as mustard oil glucosides, are the dominant source of phytoanticipins in the economically important Brassicaceae (Tierens et al., 2001; Halkier and Gershenzon, 2006). The biological activity of GSLs depends on the release of various toxic myrosinase-catalyzed hydrolytic products (Lambrix et al., 2001), including 4-methylsulfinylbutyl isothiocyanate (4MSOB-ITC), which inhibits the growth of various pathogens in vitro (Tierens et al., 2001). Further, the gsm1 mutant of A. thaliana is deficient in producing the 4MSOB-ITC precursor and is susceptible to excessive chlorosis upon infection by Fusarium oxysporum. However, because the gsm1 gene has not been cloned and the mutant is also depleted in another antimicrobial compound, the protective role of the isothiocyanate (ITC) is unclear. A fascinating new indole GSL catabolic pathway was recently shown to activate innate immunity, resulting in protection against fungal penetration (Bednarek et al., 2009; Clay et al., 2009). Still, the structural basis of GSL-derived antimicrobial defenses remains a challenge.

Necrotrophic pathogens cause extensive tissue damage and host cell death (Glazebrook, 2005). A good example for an aggressive necrotrophic pathogen with a broad host range is Sclerotinia sclerotiorum (Bolton et al., 2006). This ascomycete secretes oxalic acid to modulate the production of reactive oxygen species and to facilitate host cell death (Kim et al., 2008; Guo and Stotz, 2010). As a consequence, dying host cells may release myrosinase and GSLs, the mixture of which will generate toxic products (Lambrix et al., 2001). Among these toxic products, ITCs are formed by default, but generation of nitriles and epithionitrile requires the presence of specific proteins that interact with myrosinase to modulate its activity (Lambrix et al., 2001; Burow et al., 2009; Kissen and Bones, 2009).

The interaction between A. thaliana and S. sclerotiorum was recently established as a new necrotrophic pathosystem (Dickman and Mitra, 1992; Guo and Stotz, 2007). A complex network of hormonal and signaling pathways regulates defense responses against S. sclerotiorum (Perchepied et al., 2010). Positive regulators of this defense network include jasmonic acid, ethylene, abscisic acid, and nitric oxide (Guo and Stotz, 2007, 2010; Perchepied et al., 2010). Motivation for the present study was the strong susceptibility of the jasmonate receptor mutant coi1 to S. sclerotiorum infection (Guo and Stotz, 2007). Induction of the plant defensin PDF1.2 after S. sclerotiorum challenge was blocked in this mutant, suggesting interference with specific jasmonate-dependent defense responses.

Here we report a transcriptome analysis of A. thaliana infected with S. sclerotiorum, which enabled targeted examination of mutants with defects in camalexin and GSL biosynthesis. Specifically, the pad3 mutant defective in the last step of camalexin biosynthesis (Schuhegger et al., 2006), the cyp79b2/cyp79b3 double mutant defective in the first committed step of camalexin and indole GSL biosynthesis (Glawischnig et al., 2004), and mutants in three related R2R3 MYB transcription factors deficient in specific aliphatic GSLs were used (Hirai et al., 2007; Sonderby et al., 2007; Gigolashvili et al., 2008) to test the role of GSLs and camalexin in defense against S. sclerotiorum.

Results

Induced expression of aliphatic GSL biosynthetic genes in response to S. sclerotiorum infection

To explore host defense responses against necrotrophic pathogens, ATH1 microarrays were probed using cRNA derived from foliar mRNA of wild-type A. thaliana and coi1 mutant plants that were mock-inoculated or challenged with S. sclerotiorum. We identified 929 genes as differentially expressed (911 in wild-type and 450 in coi1 mutant leaves) 24 h after pathogen challenge (Table S1). Approximately one-third of these 929 genes belong to the gene ontology category of metabolic processes. Of these, genes involved in GSL, camalexin, tryptophan, and sulfur (S) metabolism were overrepresented. Infection with S. sclerotiorum clearly induced all four biosynthetic pathways (Figure 1). Genes acting early in the aliphatic GSL biosynthetic pathway, BCAT4, MAM1, and CYP83A1 in particular, were transiently induced (Table S1). Notably, genes contributing to each step in the camalexin and indole GSL biosynthetic pathways were induced. The majority of genes contributing to aliphatic GSL biosynthesis and to tryptophan and S metabolism were also induced. Pathogen-induced expression of the putative flavin monooxygenase (FMO) gene At1g12200 was confirmed by quantitative RT-PCR (Figure 2).

Figure 1.

 Activation of metabolic pathway genes in response to Sclerotinia sclerotiorum infection.
(a) Glucosinolate (GSL) and camalexin biosynthetic pathways. Genes in red were identified by anova with P < 0.001 and a minimum two-fold increase in expression 24 h post-inoculation. Selective expression analysis of aliphatic GSL biosynthetic genes resulted in identification of genes in blue (q < 0.05). Heat maps of gene expression ratios (pathogen- versus mock-inoculated) for 48 h post-inoculation time points of the following samples: coi1 mutant (left) and wild-type leaves challenged with S. sclerotiorum (middle), wild-type leaves challenged with Botrytis cinerea (GSE5684) (Edgar et al., 2002) (right); log of signal intensities (base 2) are shown. The color scale on the bottom right indicates the range of signal intensities. Broken arrows and gene names with question marks indicate uncertain steps in the pathway. IAOX, indole-3-acetaldoxime; IAN, indole-3-acetonitrile; DHC, dihydro camalexic acid; CAM, camalexin; 2-MD, 2-malate derivative; 3-MD, 3-malate derivative; GSL; glucosinolate; MTG, methylthioalkyl-GSL; MSG, methylsulfinylalkyl-GSL. CYP71A12 contributes to camalexin biosynthesis (Millet et al., 2010).
(b) Tryptophan biosynthetic pathway.
(c) Sulfur metabolism. PAPS, 3′-phosphoadenosine-5′-phosphosulfate; OAS, O-acetylserine.

Figure 2.

 Quantitative RT-PCR of the putative flavin monooxygenase gene At1g12200.
Plants were challenged with Sclerotinia sclerotiorum (fungus) or not (mock). Messenger RNA expression relative to 18S rRNA in unchallenged leaves (mock) of wild-type plants 24 h post-inoculation (hpi) was arbitrarily set to 1; all other values were calculated relative to it. Means and SEM of three independent biological replicates are shown. ΔΔCT values were used for statistical analysis (Yuan et al., 2006). Significant treatment differences between means are based on t-tests following F-tests; *P < 0.05; **P < 0.01.

COI1-dependent induction of individual camalexin and GSL biosynthetic genes varied (Figure 1a). Examples of genes that were induced in a COI1-dependent manner include CYP79B3 as well as the aliphatic GSL biosynthetic genes BCAT4, IPMDH encoding 3-isopropylmalate dehydrogenase, and the IPMI small subunit gene At3g58990 (Knill et al., 2009). Besides the induction of biosynthetic genes, a probe set for the nitrile-specifier proteins NSP1, NSP3, and NSP4 (Burow et al., 2009) responded positively to S. sclerotiorum infection in wild-type but negatively in coi1 mutant leaves (Table S1).

Surprisingly, the transcript levels of genes associated with aliphatic GSL biosynthesis that were elevated following S. sclerotiorum infection were not influenced by infection with the closely related pathogen B. cinerea (Figure 1a). In contrast, B. cinerea strongly induced camalexin biosynthetic genes.

Genotype- and infection-related changes in GSL and ITC content

To determine whether changes in metabolite levels parallel observed gene expression differences, foliar GSL profiles were analyzed. With the exception of 4-methylthiobutyl (4MTB)-GSL, the coi1 mutant produced significantly less aliphatic and indole GSLs than wild-type A. thaliana (Figure 3). Methylsulfinylalkyl GSL levels, in particular, differed greatly between coi1 mutant and wild-type plants. Accumulation of GSLs significantly increased from 24 to 48 h post-inoculation except for 4MTB-GSL (Table S2). Sclerotinia sclerotiorum infection had a relatively minor effect on GSL levels. Accumulation of 6-methylsulfinylhexyl (6MSOH), 7-methylsulfinylheptyl (7MSOH), and 1-methoxyindol-3-ylmethyl (1MOI3M)-GSLs significantly increased after pathogen inoculation. Conversely, total methylthioalkyl GSL levels decreased after 48 h of pathogen infection (Figure S1). A significant difference between coi1 mutant and wild-type plants in pathogen-dependent accumulation of 1MOI3M-GSL was also observed. This compound increased to wild-type levels in the coi1 mutant 24 h after pathogen inoculation but surpassed control levels only in wild-type leaves 48 h post-inoculation with S. sclerotiorum.

Figure 3.

 Glucosinolate (GSL) levels in coi1 mutant and wild-type leaves after inoculation with Sclerotinia sclerotiorum.
As control, leaves were mock-inoculated. Samples were taken 24 and 48 h after inoculation. I3M, indol-3-ylmethyl GSL; 1MOI3M, 1-methoxyindol-3-ylmethyl GSL; 8MTO, 8-methylthiooctyl GSL; 8MSOO, 8-methylsulfinyloctyl GSL; 4MTB, 4-methylthiobutyl GSL; 7MSOH, 7-methylsulfinylheptyl GSL; 5MSOP, 5-methylsulfinylpentyl GSL; 4MSOB, 4-methylsulfinylbutyl GSL; 7MTH, 7-methylthioheptyl; 6MTH, 6-methylthiohexyl; 5MTP, 5-methylthiopentyl; 6MSOH, 6-methylsulfinylhexyl; 3MSOP, 3-methylsulfinylpropyl. Minor and major GSLs are plotted separately; note the difference in scale of the ordinate. Means and SEM of three independent biological replicates are shown.

As GSLs are not biologically active per se, hydrolysis products were measured in aqueous extracts of mock-inoculated or S. sclerotiorum-challenged leaves from wild-type or coi1 mutant plants. The predominant catabolite in aqueous foliar extracts was 4MSOB-ITC. Consistent with the GSL data, the coi1 mutant generated significantly less 4MSOB-ITC than wild-type plants (Figure 4). Sclerotinia sclerotiorum infection increased the concentration of 4MSOB-ITC aqueous foliar extracts, but this elevation was not significant. Relative to 4MSOB-ITC, 4MTB-ITC and 3-methylsulfinylpropyl (3MSOP)-ITCs were measured at 49.4 ± 4.3% and 17.7 ± 3.1%, respectively. 5-Methylsulfinylpentyl (5MSOP)-nitrile, a catabolic product of 4MSOB-GSL, was detected at low levels without change in nitrile production attributable to S. sclerotiorum infection. Another low-abundance product was indole-3-acetonitrile.

Figure 4.

 Quantification of 4-methylsulfinylbutyl isothiocyanate (ITC) in aqueous foliar extracts from wild-type and coi1 mutant plants after mock inoculation or challenge with Sclerotinia sclerotiorum.
Samples were taken 48 h after inoculation. Means and SEM of five independent biological replicates are shown. Genotype had a significant effect on ITC concentration (anova; F1,16 = 5.990; P = 0.026), but the effect of treatment was not significant (anova; F1,16 = 3.968; P = 0.064).

Increased susceptibility to S. sclerotiorum of mutants with defects in camalexin or indole GSL biosynthesis

To test the significance of indole GSL and camalexin accumulation for defense against S. sclerotiorum, mutants defective in biosynthetic genes were analyzed. In contrast to the wild type, infected leaves of the pad3 mutant did not produce camalexin in response to S. sclerotiorum infection (Figure 5). In addition to the absence of camalexin, indole GSLs were not produced in infected and uninfected (systemic or mock-inoculated) leaves of the cyp79b2/cyp79b3 mutant (Table S3).

Figure 5.

 Accumulation of metabolites in wild-type plants and mutants with defects in camalexin and glucosinolate (GSL) biosynthesis.
Metabolite levels in leaves of wild-type and mutant plants 48 h after mock inoculation or after inoculation with Sclerotinia sclerotiorum. CAM, camalexin; I3M, indol-3-ylmethyl GSL; 1MOI3M, 1-methoxyindol-3-ylmethyl GSL; 8MTO, 8-methylthiooctyl GSL; 8MSOO, 8-methylsulfinyloctyl GSL; 4MTB, 4-methylthiobutyl GSL; 7MSOH, 7-methylsulfinylheptyl GSL; 5MSOP, 5-methylsulfinylpropyl GSL; 4MSOB, 4-methylsulfinylbutyl GSL; 7MTH, 7-methylthioheptyl; 6MTH, 6-methylthiohexyl; 5MTP, 5-methylthiopentyl; 6MSOH, 6-methylsulfinylhexyl; 3MSOP, 3-methylsulfinylpropyl; 3MTP, 3-methylthiopropyl. Groups of metabolites are separated by dotted lines. Means and SEM of three independent biological replicates are shown.

Based on the development of decay symptoms, a significant increase in susceptibility of the cyp79b2/cyp79b3 mutant relative to wild-type plants was observed 24, 48, 96, and 120 h after pathogen inoculation (Figure 6a). Susceptibility of the pad3 mutant was intermediate relative to the other two genotypes. In contrast to wild type and the pad3 mutant, entire leaves of the cyp79b2/cyp79b3 mutant decayed 48 h post-inoculation (Figure 6b). By 72 h post-inoculation, fungal spread through the stem occurred more frequently in the cyp79b2/cyp79b3 mutant than in wild-type and pad3 mutant plants (Figure 6b). Separate analysis of a larger number of plants ascertained that the susceptibility increase of the pad3 mutant relative to the wild type was also significant (Figure S2). It is therefore likely that both camalexin and indole GSLs contribute to metabolic defenses against S. sclerotiorum.

Figure 6.

 Effect of genetically altered camalexin and glucosinolate (GSL) production on susceptibility to Sclerotinia sclerotiorum.
(a) Quantitative analysis of decay symptoms. Lesion diameters of inoculated leaves were measured for the first 48 h after inoculation. Percentages of whole plant decay were scored thereafter. Means and SEM of separate plants are shown (n ≥ 8). Data from two independent experiments were pooled for analysis. Letters indicate significant genotypic differences (P < 0.05) for each time point according to Tukey’s honestly significant difference (HSD) test. No significant difference in genotypic variation for plant decay occurred 72 h post-inoculation. WT, wild type.
(b) Representative examples of decay symptoms after inoculation with the pathogen. Scale bars for the top and bottom rows of images are 0.5 and 1 cm, respectively.

Increased susceptibility to S. sclerotiorum of mutants with defects in elongated aliphatic GSL biosynthesis

To assess the contribution of aliphatic GSLs to defense against S. sclerotiorum, three related R2R3 MYB transcription factor mutants were examined. Of these, the myb76 mutant influenced aliphatic GSL accumulation the least (Figure 7). By comparison, the myb29 mutant had a slightly larger effect, significantly inhibiting the production of 3MSOP and 4MSOB-GSLs relative to the wild type (Table S4). The myb28 mutation had the most dramatic effect of all single mutants, significantly lowering the levels of three-carbon units (C3), C4, and C6–C8 methylsulfinylalkyl GSLs. In addition, the abundance of C6–C8 methylthioalkyl GSLs was reduced. These data confirm that MYB28 is a major regulator of aliphatic GSL biosynthesis (Gigolashvili et al., 2007; Hirai et al., 2007; Sonderby et al., 2007). The myb28/myb29 double mutant was deficient of all aliphatic GSLs (Table S5).

Figure 7.

 Accumulation of metabolites in wild-type plants and in mutants with defects in aliphatic glucosinolate (GSL) biosynthesis.
Metabolite accumulation in leaves of wild-type and mutant plants 48 h after mock inoculation or after inoculation with Sclerotinia sclerotiorum. CAM, camalexin; I3M, indol-3-ylmethyl GSL; 1MOI3M, 1-methoxyindol-3-ylmethyl GSL; 8MTO, 8-methylthiooctyl GSL; 8MSOO, 8-methylsulfinyloctyl GSL; 4MTB, 4-methylthiobutyl GSL; 7MSOH, 7-methylsulfinylheptyl GSL; 5MSOP, 5-methylsulfinylpropyl GSL; 4MSOB, 4-methylsulfinylbutyl GSL; 7MTH, 7-methylthioheptyl; 6MTH, 6-methylthiohexyl; 5MTP, 5-methylthiopentyl; 6MSOH, 6-methylsulfinylhexyl; 3MSOP, 3-methylsulfinylpropyl; 3MTP, 3-methylthiopropyl. Groups of metabolites are separated by dotted lines. Means and SEM of three independent biological replicates are shown.

Altered aliphatic GSL levels were strictly correlated with changes in susceptibility to S. sclerotiorum. Fungal decay symptoms developed similarly in the wild type and in myb76 and myb29 mutant plants (Figure 8a). The myb28 mutant, on the other hand, scored as significantly more susceptible than the wild type for the first 48 h after pathogen inoculation (Figure 8b). The myb28/myb29 double mutant was even more susceptible with an increase over the wild type in lesion diameter 48 h after pathogen challenge and elevated plant decay 96 and 120 h post-inoculation. Lesion expansion was associated with an increase in production of fungal DNA, clearly illustrating the dependence of symptom expression on microbial proliferation (Figure 8c). These data suggest that accumulation of aliphatic GSLs with six or more carbon units, in particular, is important for defense against S. sclerotiorum. As 4MSOB-GSL levels are also reduced to low levels in the myb28 mutant, a role of shorter chain GSLs cannot be ruled out.

Figure 8.

 Effect of genetically altered aliphatic glucosinolate (GSL) production on susceptibility to Sclerotinia sclerotiorum.
(a) Representative examples of decay symptoms 96 h post-inoculation. Note the rapid spread of disease symptoms through the center of the rosette in myb28 and myb28/myb29 mutant plants relative to other genotypes. Length of scale bar is 1 cm. WT, wild type.
(b) Quantitative analysis of decay symptoms. Lesion diameters of inoculated leaves were measured for the first 48 h after inoculation. Percentage of whole plant decay was scored thereafter. Means and SEM are shown (n ≥ 8). Letters indicate significant genotypic differences (P < 0.05) according to least-squares means. The effect of genotype was not significant 72 h post-inoculation.
(c) Amplification of fungal actin DNA from leaves that were infected for 24 or 48 h. No fungal DNA was amplified from uninfected leaves. The amount of total DNA in each PCR was estimated to be equal based on amplification of plant actin DNA.

Dependence of toxicity against S. sclerotiorum on ITC side chain modification

To evaluate whether plant defense chemicals directly interfere with fungal growth, in vitro plate cultures of S. sclerotiorum were exposed to camalexin and potentially toxic GSL catabolites (Halkier and Gershenzon, 2006). Artificial medium containing ≥ 125 μm camalexin inhibited fungal growth (Figure 9a). An amount of 0.5 μmol was sufficient for inhibition when this phytoalexin was applied to paper disks and placed to the periphery of growing fungal mycelia (data not shown).

Figure 9.

 Inhibition of fungal growth by camalexin and isothiocyanates (ITCs).
(a) Agar plugs containing mycelial tips of Sclerotinia sclerotiorum were applied to the center of wells containing potato dextrose agar and fungal growth was assessed 7 days later. Camalexin was added to the medium (Kliebenstein et al., 2005) at the indicated concentrations from a 10 mm stock solution in methanol (MeOH). As a control, fungal growth was tested in the presence of equivalent methanol concentrations. White bar = 1 cm. The shaded triangle illustrates confluent growth of the fungus at 31.25 and 62.5 μm camalexin but not at higher concentrations of the phytoalexin.
(b) Antimicrobial activities of ITCs with different side chain modifications and chain lengths. Paper disks were infused with 5 μmol of the indicated compound and placed on potato dextrose agar in the periphery of fungal mycelium growing from the center (Ishimoto et al., 2000). Inhibition of radial growth towards paper disks infused with the compounds was expressed relative to the solvent control. 8MTO, 8-methylthiooctyl ITC; 8MSOO, 8-methylsulfinyloctyl ITC; 7MSOH, 7-methylsulfinylheptyl ITC; 6MSOH, 6-methylsulfinylhexyl ITC; 4MSOB, 4-methylsulfinylbutyl ITC; 6MTH, 6-methylthiohexyl ITC. The experiment was duplicated with quantitatively similar results. Means and SEM are shown (n ≥ 3). The Mann–Whitney test was used for separation of means.

Besides camalexin, ITCs were tested because these products of myrosinase-catalyzed GSL breakdown reactions are known to influence biotic interactions (Ishimoto et al., 2000; Tierens et al., 2001; Halkier and Gershenzon, 2006). Paper disks containing 5-μmol methylsulfinylalkyl ITCs but not those containing methylthioalkyl ITCs inhibited fungal growth (Figure 9b). Antimicrobial activity was apparently a function of chain length with 8-methylsulfinyloctyl (8MSOO)-ITC being the most toxic compound. Growth-inhibitory effects of camalexin and 8MSOO-ITC were additive; 0.5 μmol phytoalexin and 5 μmol ITC inhibited growth by approximately 30 and 50%, respectively, whereas a combination of both resulted in 77% inhibition. Hyphal growth patterns differed in the vicinity of paper disks that were infused with either camalexin or 8MSOO-ITC (data not shown). Together, these data provide evidence for different modes of action and a two-pronged defense function of a phytoalexin and ITCs against a fungal pathogen.

Discussion

Pathogen-responsive regulation of GSL metabolism

Expression of biosynthetic genes leading to production of camalexin and GSLs was broadly activated in leaves of A. thaliana challenged with S. sclerotiorum (Figure 1). Among these genes, At1g12200 is a candidate gene closely related to FMOs that convert 4-methylthioalkyl to 4-methylsulfinylalkyl GSLs (Hansen et al., 2007; Li et al., 2008) and is pathogen-induced (Figure 2) independent of COI1 (Table S1). Simultaneous induction of methionine, tryptophan, camalexin, and GSL biosynthetic genes (Figure 1) provided evidence for coordinated regulation of primary and secondary metabolism (Brader et al., 2001; Hirai et al., 2005). In addition to the 47 probe sets for biosynthetic genes, one probe set for three NSPs responded positively to S. sclerotiorum infection in wild-type but negatively in coi1 mutant leaves (Table S1).

This modulation of gene expression was not paralleled by major changes in accumulation of GSLs or the pre-dominant 4MSOB-ITC in pathogen-challenged leaves (Figures 3 and 4). In contrast, the negative effect on expression of GSL biosynthetic genes and the reduction in GSL and ITC levels were correlated in the coi1 mutant. A reduced production of aliphatic and indole GSLs was previously noted in the coi1 mutant (Mewis et al., 2005). We also noticed that GSLs accumulated more strongly in systemic leaves than in local leaves challenged with S. sclerotiorum (Tables S3 and S5). Similar increases in GSLs were observed in systemic leaves of pathogen- and mock-inoculated plants (Table S5), suggesting that this response was related to plant manipulation rather than infection. Further research will be required to determine whether these metabolite changes are perhaps related to touch responses (Braam, 2005). Our data support published findings that camalexin locally accumulates to protect the host against microbial invaders, whereas GSLs operate at a distance (Kliebenstein et al., 2005).

In contrast to S. sclerotiorum, the closely related pathogen B. cinerea (Holst-Jensen et al., 1997) did not induce aliphatic GSL biosynthetic genes although camalexin biosynthetic genes were induced by both pathogens. This pathogen-specific difference in aliphatic GSL biosynthetic gene expression was based on analysis of signal ratios (treatment versus control) and time series (Figure S3). However, we cannot exclude an influence of developmental stage or differences in virulence on the results. Compared with S. sclerotiorum (Dickman and Mitra, 1992), B. cinerea is a relatively weak pathogen of A. thaliana (Figure 1). Specific differences in gene expression between these closely related ascomycetes raise the possibility that pathogen-specific elicitors trigger induction of aliphatic GSL biosynthetic genes.

Genetic evidence for GSL-based defense pathways

Mutant plants with defects in camalexin, indole, or aliphatic GSL production were hypersusceptible to S. sclerotiorum. Three biosynthetic pathways are therefore involved in defense responses against this pathogen. Relative to wild-type plants, disease symptoms spread more rapidly through the stems of cyp78b2/cyp78b3, myb28, and myb28/myb29 mutants (Figures 6 and 8). The speed of infection through the stem of bean plants was previously associated with differences in susceptibility to S. sclerotiorum (Chipps et al., 2005). Although the link between blocked camalexin and indole GSL biosynthesis (Figure 5) and increased susceptibility to S. sclerotiorum in the cyp79b2/cyp79b3 mutant is suggestive (Figure 6), it cannot necessarily be concluded that defense is a function of indole GSLs. The CYP79B pathway also contributes to auxin biosynthesis, but its function is restricted unlike other more prominent auxin biosynthetic pathways (Sugawara et al., 2009). Moreover, other indole compounds are generated via the CYP79B pathway (Böttcher et al., 2009; Truman et al., 2010) and GSL catabolism yields products that confine fungal infection either via their cytotoxicity or via activation of innate immune responses (Bednarek et al., 2009; Clay et al., 2009; Millet et al., 2010). The pad3 and cyp79b2/cyp79b3 mutants were previously shown to be hypersusceptible to B. cinerea, but the role of aliphatic GSLs was not investigated (Kliebenstein et al., 2005).

Susceptibility to S. sclerotiorum of the myb28 mutant with a deficiency in short- and long-chained aliphatic GSLs (Figure 7) was increased relative to the wild type (Figure 8), implicating a role for elongated aliphatic GSLs in pathogen defense. This result contrasts the specificity of GSL-dependent defense against the lepidopteran herbivore Mamestra brassicae in that larval weight gain and leaf damage were not significantly different between myb28 and myb29 mutants (Beekwilder et al., 2008). MYB28 was recently shown to alter metabolic pathways other than GSLs. A knockdown line of this transcription factor reduces adenosine 5′-phosphosulfate reductase activity (Yatusevich et al., 2010). However, the effect of altered MYB28 expression on sulfur metabolism is relatively minor compared to the more dramatic effect on GSL biosynthesis. Hence, the relatively large impact on transcription of genes involved in sulfate metabolism (Sonderby et al., 2007, 2010; Malitsky et al., 2008) is not reflected at the metabolite level (Yatusevich et al., 2010). It is therefore most likely that the observed changes in GSL levels are responsible for the differences in susceptibility to S. sclerotiorum although we cannot exclude the possibility that altered thiol levels affect pathogen responses.

Antifungal activity of GSL hydrolysis products

Methylsulfinylalkyl ITCs inhibited growth of the pathogen S. sclerotiorum in vitro. Growth inhibition was dependent on side chain modification in that methylthioalkyl ITCs were not toxic to the fungus (Figure 9b). Relative to shorter methylsulfinylalkyl GSLs, 8MSOO-ITC had the strongest antimicrobial activity in vitro. Its GSL precursor and camalexin were present at foliar concentrations of up to 638 and 83 ng−1 fresh weight (FW), respectively (Tables S3 and S4). The pathogen probably gets exposed to higher concentrations of camalexin because this phytoalexin was shown to accumulate in the immediate vicinity of an invading fungus (Kliebenstein et al., 2005). The biologically active concentration of camalexin in vitro (125 μm) is therefore close to the concentration the pathogen experiences in vivo. The toxicity of 8MSOO-ITC is approximately five to ten times less than that of camalexin, but concentrations more than seven times higher of the corresponding GSL than of camalexin are produced in leaves.

It was determined that 4MSOB-ITC reached a concentration of 700 ng−1 FW after aqueous extraction of pathogen-challenged leaves (Figure 4), a concentration similar to that of its GSL precursor (Figure 3). Indole-3-acetonitrile, another product of myrosinase-catalyzed GSL hydrolysis (de Vos et al., 2008), was reported not to be very effective in inhibiting the growth of S. sclerotiorum (Pedras et al., 2002), and we found that very low levels of this compound accumulated in S. sclerotiorum-infected leaves of A. thaliana ecotype Columbia (Col-0).

Evolutionary and ecological implications of GSL diversification

Myrosinase-catalyzed hydrolysis of GSLs results in the release of various metabolites, including ITCs, nitriles, epithionitriles, thiocyanates, ionic thiocyanate (SCN), and oxazolidine-2-thiones (Lambrix et al., 2001). The nature of the hydrolytic products depends on the GSL side chain and the reaction conditions. Indole GSLs are not known for producing ITCs. Instead, they produce mainly SCN and little indole-3-acetonitrile, along with other indole-derived products (Brown and Morra, 1997; Agerbirk et al., 2009). Here we used ecotype Col-0, which accumulates chain-elongated ITCs (Lambrix et al., 2001) and is less susceptible to S. sclerotiorum than Landsberg erecta (Figure S4), an ecotype that produces short nitriles (Lambrix et al., 2001). The importance of ITCs for the interaction with S. sclerotiorum was also supported by our observation that ITCs were far more abundant than nitriles when pathogen-challenged leaves were extracted under conditions that resulted in GSL hydrolysis.

Similar to our findings, 8MSOO-ITC was more toxic to various fungal root pathogens than 8-methylthiooctyl (8MTO)-ITC or 8-methylsulfonyloctyl ITC (Ishimoto et al., 2000). The compound 4MSOB-ITC has been implicated in protecting against F. oxysporum because less of the parent GSL accumulates in the gsm1 mutant than in the wild type (Tierens et al., 2001).

A new indole GSL catabolic pathway was recently shown to operate in living plant cells and protect against fungal pathogens (Bednarek et al., 2009; Clay et al., 2009). A mutant in an atypical myrosinase, pen2, is hypersusceptible to biotrophic powdery mildew pathogens, the soilborne ascomycete Plectosphaerella cucumerina, and oomycete pathogens (Lipka et al., 2005; Adie et al., 2007). Genes that control this GSL pathway, PEN2, CYP81F2, and MYB51, are induced in response to treatment with the pathogen-associated molecular pattern Flg22 (Clay et al., 2009) but not after inoculation with S. sclerotiorum (Table S1). Whereas powdery mildew pathogens penetrate epidermal cells, S. sclerotiorum merely breaches the cuticle, then grows between the cuticle and the epidermis and secretes cell wall hydrolases and oxalic acid to trigger host cell death (Lumsden and Dow, 1973; Lumsden and Wergin, 1980; Guimaraes and Stotz, 2004). This attack strategy is likely to result in indiscriminate hydrolysis of both indole and aliphatic GSLs.

Many Brassica species including Brassica napus do not produce chain-elongated GSLs (Li et al., 1999). Metabolic engineering of the aliphatic GSL biosynthetic pathway may therefore be an avenue for enhancing the resistance of crops to agronomically important necrotrophic pathogens, including S. sclerotiorum.

Experimental procedures

Biological materials and growth conditions

Seeds originated from the Arabidopsis Biological Resource Center except coi1-2 which was provided by Dr D.-X. Xie (Tsinghua University, Beijing, China). The cyp79b2/cyp79b3 and myb28/myb29 double mutants were generated locally by crossing single mutants (Sugawara et al., 2009). Wild-type (Col-0) and mutant plants were grown in soil mix (Sunshine SB40 or 1:2 vermiculite: potting soil) under short-day conditions with a 10-h light/14-h dark cycle at 22°C under fluorescent light at 250–300 μm−2 sec−1 (Guo and Stotz, 2007). Plants were fertilized every other week with a 1000-fold diluted fertilizer stock (Hyponex, http://www.hyponex.co.jp).

Plant inoculation

Sclerotinia sclerotiorum strain 1980 was grown on minimal medium (Guo and Stotz, 2007). A randomized complete block design was used to grow wild-type and mutant plants in 72-well flats. Rosette leaves of 4- to 6-week-old A. thaliana plants were inoculated by placing an agar plug (1.6 mm in diameter) containing actively growing hyphal tips on the adaxial leaf surface. Plants were scored without knowing their genotype. Lesion diameters of infected leaves were recorded with a digital caliper for the first 48 h after inoculation. Lesions that did not expand beyond the agar plug perimeter were inspected from the abaxial side. As lesions expanded beyond inoculated leaves 72 h post-inoculation, decay percentages were scored by determining the proportion of the discolored infected area relative to the entire rosette area using digital images (Guo and Stotz, 2007).

Microarray analysis

A randomized complete block design was used to grow wild-type and coi1 mutant plants in two 72-well flats. Two leaves per plant were mock-inoculated or challenged with S. sclerotiorum. Separate plants were harvested for the 24 h and 48 h post-inoculation time points and challenged leaves were frozen in liquid nitrogen. Total RNA was extracted using a phenol-LiCl method (Guo and Stotz, 2007). The RNA was quantified using a Nanodrop ND-1000 UV-Vis Spectrophotometer (http://www.nanodrop.com/). The purity of the RNA was checked using an Agilent Bioanalyzer 2100 (http://www.agilent.com/). Labeled target cRNA was prepared from 3 μg total RNA samples as described in the Affymetrix GeneChip® Expression Analysis Technical Manual (701021 Rev. 5; http://www.affymetrix.com/) protocols with an in vitro transcription (IVT) incubation time of 16 h. Ten micrograms of fragmented cRNA from each sample was hybridized for 16 h to an Affymetrix ATH1 Arabidopsis Genome Array. Washing, staining and scanning followed the GeneChip® Expression Wash, Stain and Scan Manual for Cartridge Arrays (P/N 702731, Affymetrix) protocols, using gcos software with an Affymetrix Fluidics Station 450 (wash protocol EukGE-WS2v5_450), and a GeneChip Scanner 3000 with autoloader. Image processing and data extraction were performed using gcos software version 1.4, and expression console v1.

Affymetrix ATH1 GeneChips were analyzed in triple biological replicates. Signal intensities were analyzed using microarray suite version 5 (MAS5; Affymetrix). Genes (16 561 of 22 746 probe sets) that were significantly expressed (detection P-value < 0.05) at least once in the assays were used for the subsequent analysis. Signal values in each chip were normalized by the median value and then transformed to a log2 scale. Differentially expressed genes were extracted by analysis of variance (anova) using the programs R and Perl. The anova conditions were wild type (WT) mock 24 h, WT inoculation 24 h, coi1 mock 24 h, and coi1 inoculation 24 h (P < 0.001 corresponding to false discovery rate (FDR) q-value 0.0037) (Storey and Tibshirani, 2003). Signal ratio (SR) values (average WT 1 day inoculation/average WT 1 day mock or average coi1 1 day inoculation/average coi1 1 day mock) were used to further filter differentially expressed genes (SR ≥ 2 or SR ≤ 0.5). Differentially expressed genes were analyzed by hierarchical cluster analysis using cluster 3 software (de Hoon et al., 2004). Cosine correlation was used as the similarity metric. The clustering method was centroid linkage. Results were visualized in java treeview (Saldanha, 2004). Data were deposited at EMBL-EBI ArrayExpress, accession E-MEXP-3122.

The Gene Expression Omnibus (http://www.ncbi.nlm.nih.gov/geo/) was used to access public microarray data on response to Botrytis cinerea infection in A. thaliana leaves (GSE5684) and to compare it with the microarray data on the foliar response to S. sclerotiorum infection reported here. We generated SR values for selective analysis of tryptophan, S, and GSL metabolism genes, Student’s t-test analysis was conducted (mock versus inoculated) in combination with FDR q-value analysis (Storey and Tibshirani, 2003).

Quantitative RT-PCR

Mock or pathogen-inoculated leaves (n > 5) were homogenized in liquid nitrogen and total RNA was extracted using the RNeasy Plant Mini kit (Qiagen, http://www.qiagen.com/). The QuantiTect Reverse Transcription kit (Qiagen) was used for cDNA synthesis and genomic DNA elimination. Primers 5′-CCCTTTCCGATGTTTGAGCTT-3′ and 5′-GAGATACTCGGCCGGACAAG-3′ were used in combination with the Taqman probe 5′-AAAGCAAGTGGGTCGCAGCGG-3′ labeled with FAM at the 5′-end and with MGB at the 3′-end. Two-step PCR was carried out as recommended (QuantiTect PCR Handbook; Qiagen) with 40 cycles of denaturation at 95°C for 15 sec and annealing/extension at 60°C for 60 sec following an initial denaturation of 15 min at 95°C. This program was followed by a melting curve analysis. Dilutions of genomic DNA were used for calibration using the Relative Standard Curve Method (Applied Biosystems, http://www.appliedbiosystems.com/). A VIC/MGB-labeled eukaryotic 18S endogenous control (Applied Biosystems) was used to determine relative mRNA abundance. Three biological replicates were analyzed for each data point.

Amplification of fungal and plant DNA

DNA was extracted from infected and uninfected leaves using a modification of the cetyltrimethylammonium bromide-based method (Clarke, 2009). DNA was quantified using a Nanodrop ND-1000 UV-Vis Spectrophotometer. Plant actin DNA was amplified with AtACT2/8 sense and AtACT2/8 antisense primers (An et al., 1996; Ellinger et al., 2010). Fungal actin DNA was amplified using forward 5′-TCTTGAGAGCGGTGGTATCC-3′ and reverse 5′-GATGATGGTGCAAGAGCAGT-3′ primers. Plant and fungal actin DNA were amplified for 35 cycles at annealing temperatures of 50 and 55°C, respectively, and extension for 30 sec at 72°C.

Analysis of GSL and GSL hydrolysis products

Challenged or unchallenged systemic leaves were harvested and frozen prior to analysis. The GSLs were analyzed by liquid chromatography–mass spectrometry using sinigrin as an internal standard for quantification (Sawada et al., 2009).

A Retsch MM301 Ball Mill (http://www.retsch.com/) was used for aqueous homogenization of leaves in screw cap vials containing zirconium oxide beads (0.4 mm diameter). Samples were incubated for approximately 5 min at 20°C and extracted with dichloromethane after the addition of 5 mm CaCl2 (Brown et al., 1994; Lambrix et al., 2001). The organic phase was dried with Na2SO4.

Samples were analyzed by gas chromatography–mass spectrometry (GC–MS) using a Trace GC Ultra (Thermo Scientific, http://www.thermoscientific.com/) with a Phenomenex ZB-5 column (30 m × 0.25 mm × 0.25 μm), splitless injection at 200°C, and a temperature program of 50°C for 5 min, a 12°C/min ramp to 96°C, and a 18°C min−1 ramp to 240°C with a 5-min final hold. For product identification, the column was coupled to a DSQ mass spectrometer (Thermo Scientific) with He as the carrier gas. Parameters for electron impact ionization were: interface temperature 240°C; electron energy 70 eV; source temperature 200°C. Peaks were identified using authentic standards and published retention times and mass spectra (Spencer and Daxenbichler, 1980).

Major ions were used for peak quantification. Analytical chemicals were purchased from Sigma Aldrich (http://www.sigmaaldrich.com/). Phenyl ITC and benzyl nitrile were used as internal standards. Sulforaphane was used to calculate the detector response factor by dividing the ratio of the slope of the phenyl ITC standard curve by the slope of the sulforaphane standard curve (Brown et al., 1994).

Nitrile products were analyzed at the Max-Planck-Institute of Chemical Ecology, Jena, Germany. We used GC-MS to screen for nitriles, followed by quantification using GC with a flame ionization detector (FID). Nitriles were also separated on LC-IonTap-MS for confirmation of results (Reichelt et al., 2002).

Antifungal assays

The ITCs were purchased from LKT Laboratories, Inc. (http://www.lktlabs.com/). Gas chromatography established purities for 4MSOB-ITC (99.02%) and -ITC (100%), while HPLC assured purities for 6-methylthiohexyl (6MTH)-ITC (99.5%), 7MSOH-ITC (98.64%), 8MTO-ITC (99.5%), and 8MSOO-ITC (97.1%). Compounds were dissolved in ethyl acetate to generate 250 mm stock solutions and stored as aliquots at −20°C. The concentration of the camalexin stock solution was 10 mm in methanol. Camalexin toxicity was studied by incorporating compounds into potato dextrose agar (PDA) (Kliebenstein et al., 2005). A small PDA plug containing actively growing mycelia was applied to the center of the toxin-containing PDA medium.

Camalexin and ITC toxicities were also tested with the help of toxin-infused paper disks (Advantec, http://www.advantec.co.jp) (Ishimoto et al., 2000). Volumes applied to disks were 50 μl in the case of camalexin or 20 μl for ITCs. Solvent was evaporated from discs in a vacuum prior to assays. Filters were placed in test tubes inside a desiccator and vacuum was drawn for 5 min using a pump (IWAKI ASP-13, Asahi Techno Glass, http://www.iwakipumps.jp). This period was kept short for the purpose of removing solvents. Discs were placed in the periphery of PDA plates with a PDA plug containing actively growing mycelia in the center.

Statistical analysis

Metabolite and phenotypic data were subjected to anova with or without logarithmic transformation (Guo and Stotz, 2007). Levene’s test was used to check for homogeneity of variances. Non-parametric tests were used when variances were not homogeneous. Plant decay percentages were arcsine-transformed prior to anova. Two-tailed tests were used with α < 0.05. Means were separated using Tukey’s honestly significant difference (HSD) test, least square means or the non-parametric Mann–Whitney test.

Acknowledgements

This work was funded by a Japan Society for the Promotion of Science fellowship (L08701) to HUS and the Core Research for Evolutional Science and Technology (CREST) of the Japan Science and Technology Agency (project name ‘Elucidation of amino acid metabolism in plants based on integrated omics analyses’) to MYH. We are grateful to Dr Kohki Yoshimoto (RIKEN Plant Science Center) for sharing purified camalexin originally obtained from Dr Jane Glazebrook (University of Minnesota). Ms Anne-Marie Girard (Oregon State University) assisted with microarray hybridization. Ms Chitose Takahashi (RIKEN Plant Science Center) helped with microarray data analysis. Ms Ayuko Kuwahara (RIKEN Plant Science Center, JST-CREST) and Ms Mutsumi Nagano (RIKEN Plant Science Center, JST-CREST) assisted in GSL analysis. Dr Makoto Tokuda (Kyushu University) helped with statistics. Dr Xiaomei Guo assisted with ecotype evaluations. Dr Shinjiro Yamaguchi and Dr Yukuse Jikumaru (RIKEN Plant Science Center) participated in discussions and provided experimental advice. We thank Dr Anthony Collins (Queens University) and Dr James Tokuhisa (Virginia Polytechnic Institute and State University) for helpful discussions and critical review of the manuscript. We appreciate Dr Michael Reichelt’s (Max-Planck-Institute for Chemical Ecology) advice and GC and LC analyses of ITCs and nitriles.

Ancillary