• Plants have evolved different defense components to counteract pathogen attacks. The resistance locus resistance to Leptosphaeria maculans 1 (RLM1) is a key factor for Arabidopsis thaliana resistance to L. maculans. The present work aimed to reveal downstream defense responses regulated by RLM1.
• Quantitative assessment of fungal colonization in the host was carried out using quantitative polymerase chain reaction (qPCR) and GUS expression analyses, to further characterize RLM1 resistance and the role of salicylic acid (SA), jasmonic acid (JA), and ethylene (ET) in disease development. Additional assessments of A. thaliana mutants were performed to expand our understanding of this pathosystem.
• Resistance responses such as lignification and the formation of vascular plugs were found to occur in an RLM1-dependent manner, in contrast to the RLM1-independent increase in reactive oxygen species at the stomata and hydathodes. Analyses of mutants defective in hormone signaling in the camalexin-free rlm1Lerpad3 background revealed a significant influence of JA and ET on symptom development and pathogen colonization.
• The overall results indicate that the defense responses of primary importance induced by RLM1 are all associated with physical barriers, and that responses of secondary importance involve complex cross-talk among SA, JA and ET. Our observations further suggest that ET positively affects fungal colonization.
Plants have evolved a range of sophisticated mechanisms to perceive pathogen attacks and to mediate appropriate defense responses. In contrast to mammals, which rely on highly specialized circulating defense cells, each individual plant cell has the ability to recognize nonself molecules and transmit signals emanating from infection sites (reviewed by Chisholm et al., 2006; Jones & Dangl, 2006). An important and well-characterized recognition mechanism is based on resistance (R) genes whose products confer recognition of cognate avirulence (Avr) proteins from different pathogens. The majority of R genes encode proteins containing both a nucleotide-binding (NB) and a leucine-rich repeat (LRR) domain with a variable amino terminus, among which the Toll interleukin 1 receptor (TIR) is the most common domain. Most NB-LRR proteins are localized intracellularly, and can recognize pathogen effectors indirectly by interacting with certain host factor proteins. However, new modes of recognition systems are still being found which demonstrate the high level of complexity of plant defense (Caplan et al., 2008).
Direct or indirect recognition of Avr proteins by R proteins, or of pathogen-associated molecular patterns (PAMPs) by pathogen recognition receptors (PRRs) such as plasma-membrane-resident proteins (Wang et al., 2008), triggers a variety of downstream responses. Among these, the hypersensitive response (HR), a rapid suicide mechanism in which plants sacrifice a few cells to prevent further tissue colonization of an invading pathogen, is one of the most common responses which often, but not always, follows recognition of Avr proteins. Upon the onset of the cell death process, several other events take place or are transmitted. One of the most rapid responses in this context is the oxidative burst that leads to cell death when elevated concentrations of highly reactive oxygen species (ROS) are present (Ma & Berkowitz, 2007). The ROS signaling network is complex and impacts various cellular events including perception of pathogen attacks (Kim et al., 2008). In contrast to interactions with biotrophs, interactions between plants and necrotrophic pathogens that kill host cells and acquire nutrients from dead tissues lack rapid recognizable induced host reactions such as the HR and recognition channeled via characterized R–Avr gene pairs.
Recent studies highlight the need to expand our understanding of plant–pathogen interactions, including new recognition, membrane trafficking, signaling and gene-regulating parameters, areas in which hemibiotrophic fungi and oomycetes have been proposed as models (Holub, 2006; Lamour et al., 2007). One emerging hemibiotrophic model fungus is Leptosphaeria maculans. The complex pathogenesis pathways and parasitic strategies of L. maculans include biotrophic/endophytic, necrotrophic and saprophytic growth (Howlett et al., 2001), providing unique opportunities to study several types of interactions in a single pathosystem. Despite its primarily necrotrophic nature, L. maculans displays a gene-for-gene relationship with both Brassica napus and Arabidopsis thaliana (Delourme et al., 2006; Staal et al., 2006), which is similar to that observed in many biotrophic pathosystems.
Some signaling components, such as salicylic acid (SA), are known to influence both R gene-dependent and basal resistances (Glazebrook, 2001), but the so-called basal resistances include a multitude of different components – ranging from PAMP-triggered immunity (PTI)-induced responses to various innate chemical and physical structures of the plant. The chemical structures of the plant phytoalexins in relation to the pathogen detoxification mechanisms are, for example, important factors in the differential colonization success of L. maculans on compatible Brassica hosts (Pedras & Ahiahonu, 2005). In the case of A. thaliana, the phytoalexin camalexin is induced by general stress responses in pathways partially overlapping both SA and jasmonic acid (JA) signaling, and thus its induction is not a pathogen-specific response (Kliebenstein, 2004). Several studies have confirmed that the induction of camalexin is triggered by successful pathogen colonization and that genotypes with functional R genes induce lower concentrations of camalexin, indicating that camalexin acts as a part of the defense when PTI and gene-for-gene resistances fail (Mert-Türk et al., 2003; Narusaka et al., 2004). We have previously demonstrated that camalexin induction plays a greater role when the R gene-dependent responses are impaired in the A. thaliana–L. maculans system (Staal et al., 2006; Kaliff et al., 2007). Unraveling independent levels of defense components is important to fully understand the host resistance mechanisms to hemibiotrophic fungal pathogens. The defense responses found to date in B. napus–L. maculans interactions are multiple, and include necrosis of guard cells, and induction of callose deposition, lignification, phytoalexins, pathogenesis-related (PR) proteins, and pectin in the xylem vessels (reviewed by Howlett et al., 2001). In this study, early defense responses in relation to the resistance to Leptosphaeria maculans 1 (RLM1) locus, and without the involvement of camalexin, were assessed in A. thaliana. In order to further dissect the impact of the hormonal signal transduction network of SA and JA/ethylene (ET), mutants impaired in those pathways were studied in an R gene- and camalexin-free background.
Materials and Methods
The quintuple plant material (coi1-16ein2NahG(gl1)rlm1Ler pad3(er)) was generated as follows: coronatine insensitive 1-16 (coi1-16) was crossed with the ethylene insensitive 2/salicylate hydroxylase over-expressor (ein2NahG) double mutant to obtain coi1-16ein2NahG(gl1), which subsequently was crossed with the rlm1Lerpad3(er) (25% Landsberg erecta (Ler-0) genotype) double mutant (Staal et al., 2006). The camalexin-deficient pad3-1 mutant was introduced to circumvent the impact of camalexin in the subsequent defense assessments. NahG and NahG-free progenies were identified by NahG-specific PCR with the primers NahG-F: TCCCGCAGATGTACTTAGGG and NahG-R: GTATAACTCGCCGGTTTCCA. The susceptible rlm1Lerrlm2Col line derived from the RLM1Colrlm2Col×rlm1Ler RLM2Ler cross (Staal et al., 2006) is henceforth denoted rlm1Ler. This genotype was included because all other mutants are in the Columbia (Col) (RLM1Colrlm2Col) background. Selection for informative genetic combinations in the F2 was performed on 10 µM 1-aminocyclopropane-1-carboxylic acid (ACC) for evaluation of the triple response, on 100 µM methyl jasmonate (MeJA) for evaluation of root inhibition and on 50 µg ml−1 kanamycin to select for NahG. The coi1-16 genotype was further confirmed by observation of male sterility at room temperature, and the plants were cut down and transferred to 16°C for propagation. A conditionally fertile coi1 allele was used in order to generate stable lines. Furthermore, effects produced by the mutations in erecta (er) deriving from Ler-0 and glabrous1 (gl1) from the coi1-16 ancestor, respectively, were identified visually and several independent lines polymorphic for these loci were obtained. Four genotypes with distinct disease phenotypes were selected and further backcrossed to rlm1Lerpad3(ER) (12.5% Ler-0 genotype). F3 plants were screened on selective media, and eight different combinations of the rlm1Lerpad3 background with or without ein2, NahG or coi1-16 were identified, with the following cleaved amplified polymorphic sequence (CAPS) markers: COI1-16, F: TTGCCAAAATCAGTCCCAAA and R: TTGGATAAGCGTACAATGGTCTT using the Hyp188III restriction enzyme; EIN2, F: CGCCATCTTTGTTTCAACAATCAGATCC and R: CCAGAGGAAAGAGAGTTGGATGTAAAGTACTCTACCGCT using BsrBI digestion (Resnick et al., 2006), and for NahG as described earlier in this paragraph. The resistant accessions Col-0 (RLM1Colrlm2Col) and Ler-0 (rlm1LerRLM2Ler) were used as controls. In addition, vascular-associated death 1 (vad1) (Lorrain et al., 2004), vad1-ein2-1, vad1-ein3-1 (Bouchez et al., 2007), rcd1 (Overmeyer et al., 2000), rar1-21, sgt1b-edm, rar1sgt1b (Holt et al., 2005), irregular xylem 4 (irx4) (Jones et al., 2001), lsd1 (Dietrich et al., 1997), pen2, pen2pad4, pen2sag101, pen2eds1, pad4sag101, pen2sag101pad4 (Lipka et al., 2005), snc, modifier of snc1 2 (mos2) (Zhang et al., 2005), flavin-dependent monooxygenase 1 (fmo1) (Bartsch et al., 2006), mos3 (Zhang & Li, 2005), pbs1, pbs3 (Warren et al., 1999), salk_021153 (pp5-1) and salk_104468 (pp5-2) mutants were evaluated for various resistance mechanisms, together with salk_014088 (rlm1ACol) and salk_110395 (rlm1BCol) which carry T-DNA insertions in the two genes (RLM1ACol and RLM1BCol) in the RLM1 locus required for Leptosphaeria maculans (Desm.) Ces. et de Not. resistance (Staal et al., 2006). Plants were grown as described in Bohman et al. (2004). For comparison, the susceptible Brassica napus (L.) cultivar Hanna (Sjödin & Glimelius, 1988), the resistant cultivar Surpass 400 (Li & Cowling, 2003), and the two nonhost species pea (Pisum sativum (L.); Oregon sugar pod) and barley (Hordeum vulgare (L.); ND B112) were used.
Fungal isolates and plant inoculations
The isolates Leroy and PHW1245 (Balesdent et al., 2005) were used for evaluations of L. maculans responses of mutants and genotypes. The responses to L. maculans were determined on Arabidopsis thaliana (L.) Heynh. rosette leaves, at the five to eight leaf stage, by inoculation of conidia using the puncture leaf test as described by Sjödin & Glimelius (1988). Standard inoculations of material harboring rlm1Ler were performed with 5 × 106 conidia ml−1 water and no initial period of 100% humidity for better resolution in the screening. Testing of RLM1 resistance was performed with 2 × 107 conidia ml−1 and a 3-d period of 100% humidity. Disease progression was determined with both qPCR assessment and visual screenings, where a plant was classified as ‘diseased’ when lesions corresponded to 3 or higher on the Delwiche & Williams (1979) disease severity scale. The scale is as follows: 0, no symptoms; 1, lesion diameter 0.5–1.5 mm; 3, dark necrotic lesions 1.5–3 mm; 5, lesions 3–5 mm, occasional sporulation; 7, gray-green tissue collapse, lesions 4–8 mm, sporulation; 9, rapid tissue collapse, accompanied by profuse sporulation in large lesions (> 5 mm). Brassica napus plants were grown and inoculated as described by Kaliff et al. (2007). Barley was spray-inoculated with inoculum until the first droplets fell off the leaves, with a concentration of 1 × 106 conidia ml−1 water. Pea plants were inoculated using the A. thaliana procedure. Both pea and barley were incubated at 100% humidity for the first 3 d after inoculation. All plant material was cultured as previously described (Kaliff et al., 2007). At least 20 plants per genotype were visually assessed, and experiments were repeated three times.
Construction of plasmids and fungal transformation
The plasmid pFG-GUS (Zhang et al., 2006) was digested with EcoRI and HindIII, and the fragment carrying the GUS gene under the control of the constitutive Aspergillus nidulans glyceraldehyde-3-phosphate dehydrogenase A (GPDA) promoter and tryptophan C (TrpC) terminator was introduced to the EcoRI/HindIII site of the binary vector pPZP121, generating pSO0 (Hajdukiewicz et al., 1994). The hygromycin B phosphotransferase (HPH) gene, under the control of the A. nidulans TrpC promoter, was obtained from HindIII-digested pLM1 (Zhang et al., 2006), and the fragment ligated with the HindIII linearized pSO0 vector, resulting in a binary vector harboring GPDA::GUS and TrpC::HPH (pSO1). The plasmid pSO1 was used for Agrobacterium tumefaciens-mediated transformation of L. maculans isolate PHW1245 (Gardiner & Howlett, 2004).
Quantitative real-time PCR
Quantitative measurements of L. maculans growth were performed using the following primers: shaggy-related protein kinase alpha 5′-CTTATCGGATTTCTCTATGTTTGGC-3′, 5′-GAGCTCCTGTTTATTTAACTTGTACATAC C-3′ for A. thaliana (Gachon & Saindrenan, 2004), and internal transcribed spacer (ITS) sequence for Leroy isolate 5′-GGTGTTGGGTGTTTGTTCCAC-3′, 5′-GGCTGCCAATTGTTTCAAGG-3′ for L. maculans. The PCR conditions were 2 min at 50°C and 10 min at 95°C, followed by 40 cycles of 15 s at 95°C, and 50 s at 64°C. To determine the quality of the product, a dissociation curve analysis was carried out after the program, with temperature increasing from 60 to 95°C. Genomic DNA was extracted as described in Brouwer et al. (2003) and amplified using 3 µM primers, 5 ng of template DNA per 25 µl of reaction and 12.5 µl of Power SYBR® Green PCR Master mix (Applied Biosystems, Foster City, CA, USA) using an ABI 7000 thermocycler (Applied Biosystems). All samples were amplified in technical triplicates. The cycle threshold (Ct) value was set automatically by the sds 1.2.3 software (Applied Biosystems). Primer specificity was confirmed by setting up a reaction with the A. thaliana template and L. maculans primers, or vice versa. Each genotype was assessed in three biological replicates of samples comprising four to six plants.
Histochemical staining and quantification of GUS activity
Plants were inoculated with the GUS-tagged isolate of L. maculans, and samples were harvested at 14 days post-inoculation (dpi). Three pools, each of which consisted of nine leaves harvested from three independent plants, were set up for each plant genotype, and the GUS activity of each pool was determined as described by Jeffersson (1987). The DyNA QuantTM 200 fluorometer (Hoefer Pharmacia Biotech Inc., San Fransisco, CA, USA) was used for the quantifications. Protein concentrations were determined as described by Bradford (1976).
Leaves were collected 2, 3, 4 and 5 dpi and stored in 95% ethanol until staining. Staining was performed using the Weisner reaction: 2 volumes 2% w/v phloroglucinol (Merck, Whitehouse Station, NJ, USA) in 95% ethanol and 1 volume concentrated HCl (Dean, 1997). Lignin appears red immediately and was photographed within 5–10 min post staining, before any fading of the staining had occurred, and was quantified using the APS assess software (L. Lamari, University of Manitoba, Winnipeg, Canada).
3,3-diaminobenzidine tetrahydrochloride (DAB) staining for ROS, guard cell necrosis and detection of vascular plugs
Arabidopsis thaliana pad3-1 and rlm1Lerpad3 plants were surface-inoculated with a solution containing Silwet L-77 and 1% sucrose (control) and one containing 107 conidia ml−1. Plants were spray-inoculated until a thin film of spore suspension covered all the leaves. Materials were kept at 65% humidity and collected for DAB staining (Thordal-Christensen et al., 1997) 16 h post-inoculation (hpi), 3 dpi and 6 dpi. Arabidopsis thaliana pad3-1 and rlm1Lerpad3 plants were collected 7 dpi, chlorophyll was removed in chloral hydrate (Chen & Howlett, 1996), and vascular plugs were quantified as dark matter using the APS assess software (L. Lamari). The observations were based on 9–10 independent leaves of each genotype combination and the experiment was repeated three times.
Establishment of quantitative assessments of L. maculans with real-time PCR
The amounts of L. maculans genomic DNA in planta were determined using qPCR through a course of fungal infection with the aim of establishing a quantitative disease scoring method in the A. thaliana–L. maculans pathosystem. Leptosphaeria maculans DNA was reduced from an initial level of 20 ± 7.5 pg ng−1 A. thaliana DNA at time-point zero to c. 6 pg ng−1 A. thaliana DNA during the first 6 dpi (Fig. 1a). At later time-points (> 10 dpi), a clear increase in fungal DNA in planta was observed only for the susceptible rlm1Lerpad3 genotype (Fig. 1a). A reduced level of L. maculans DNA was detected in the resistant genotype, as well as in the susceptible genotypes before the onset of colonization. A slight increase of fungal DNA was initially observed on B. napus independent of genotype, but at 10 dpi fungal growth declined in the resistant cultivar Surpass 400 and at 14 dpi a clear significant difference between the susceptible cultivar Hanna and Surpass 400 was seen (Fig. 1b). To determine whether these observations were attributable to residual DNA in dead fungal spores, RLM1Colrlm2Col and rlm1LerRLM2Ler A. thaliana leaves were washed extensively before DNA preparation. This treatment did not alter the level of L. maculans DNA in the material (Supporting Information Fig. S1a), and no difference between mock and fungal-inoculated materials was found after GUS staining and microscopic assessments. Two nonhost species, barley and pea, were also assessed for several days after inoculation with the fungus (Fig. S1b). A relatively late reduction of L. maculans DNA was observed, indicating a certain degree of survival of the fungus even on nonhost material. In summary, qPCR assessment of fungal DNA can reliably distinguish susceptible A. thaliana and B. napus genotypes from resistant genotypes at relatively late time-points, > 10 dpi.
Dose dependence of RLM1Col resistance
The RLM1Col locus harbors the two TIR-NB-LRR genes RLM1ACol (At1g64070) and RLM1BCol (At1g63880). A previous study demonstrated that a genomic clone of RLM1ACol is sufficient to complement rlm1Lercompletely (Staal et al., 2006). For a more thorough examination of the role of RLM1ACol and RLM1BCol, two T-DNA insertion mutants, rlm1ACol and rlm1BCol, were analyzed in the camalexin-deficient (pad3-1) background to avoid interference from camalexin in the analysis. Quantitative assessment with visual inspection and qPCR analysis showed that the rlm1AColpad3, rlm1BColpad3 and rlm1Lerpad3 genotypes were significantly more susceptible compared with the susceptible background (pad3-1) and resistant (Col-0) genotypes at 14 dpi (Fig. 2a,b), demonstrating that the resistance conferred by the RLM1Col locus is dependent on at least two TIR-NB-LRR genes (RLM1ACol and RLM1BCol) in a gene dose-dependent manner.
To confirm this R gene dose dependence, rlm1Lerpad3 was backcrossed with pad3-1 for thorough evaluations of RLM1Col/rlm1Ler (BC2) heterozygotes in a camalexin-free background. These BC2 plants were inoculated with a high concentration of spore suspension (2 × 107 conidia ml−1) and exposed to prolonged high humidity. At 12 dpi, (RLM1Col/rlm1Ler) pad3 plants showed necrotic lesions along the mid vein, and rlm1Lerpad3 showed clear susceptible phenotypes, whereas pad3-1 did not start showing lesions until 15 dpi (Fig. 3a). In the quantitative assessment of disease development, (RLM1Col/rlm1Ler) pad3 plants started showing symptoms, on average, 1.5 d earlier (P = 0.01) compared with pad3-1, while the rlm1Lerpad3 mutant developed lesions on average 6.2 d earlier (P = 7 × 10−10) than pad3-1. The quantitative assessments of the heterozygotes compared with the rlm1Lerpad3 and pad3-1 parents determined the degree of dominance of RLM1Col to 0.53 for visual screening under the given environmental conditions. Standard or mild inoculation conditions revealed no significant differences between (RLM1Col/rlm1Ler) pad3 heterozygotes and the pad3-1 parent (RLM1Col degree of dominance = 1).
Characterization of RLM1 resistance components
Mutants known to affect R gene-dependent resistance (Hammond-Kosack & Parker, 2003) were evaluated to further investigate the role of RLM1. The mutants of the R protein interacting proteins RAR1 (required for Mla-dependent resistance), SGT1b (a suppressor of the G2 allele of SKP1) and heat shock protein 90 (HSP90) were crossed to pad3-1 and characterized in terms of their disease development (Figs 2a, 3b). Visual analysis of disease progression indicated that the rar1pad3 and hsp90pad3 double mutants were more susceptible than plants of the pad3-1 background (Fig. 2a). Quantitative PCR analysis, however, indicated no increase in fungal colonization in the hsp90pad3 line compared with the pad3 single mutant, while fungal growth was apparently enhanced in the rar1pad3 genotype (Fig. 2b). Staining for callose revealed a decrease in callose deposition, a resistance response previously found to be linked to RLM1 (Staal et al., 2006), in the rar1 mutant compared with Col-0. The wild-type level of callose deposition appeared to be restored in the rar1sgt1b double mutant (not shown), supporting our argument regarding restoration of resistance in this genotype. The protein phosphatase PP5 has been shown to interact with NB-LRR proteins involved in innate immunity together with RAR1/Chp1 (CHORD-containing protein), HSP90 and the CS domain (Chp1/SGT1) in both animals and plants (Hahn, 2005; de la Fuente van Bentem et al., 2005). The T-DNA mutants pp5-1 and pp5-2, however, did not show any defect in RLM1-dependent resistance. Other R gene signaling components, such as the modifier of snc, mos2 and the AvrPphB susceptible pbs1 and pbs3 did not affect RLM1 function.
Analogously to other challenged necrotrophic fungal pathogens (Consonni et al., 2006), the AtMLO2 powdery mildew-resistant mutant pmr2-1 showed a moderately L. maculans-susceptible phenotype. The penetration mutant pen1, defective in syntaxin, was previously found to be moderately susceptible to L. maculans (Staal et al., 2006). The pen2 mutant, which is impaired in a glycosyl hydrolase, represents an alternative route for the loss of nonhost resistance to biotrophic powdery mildew (Lipka et al., 2005). The pen2 line also displayed a moderately susceptible (chlorotic) phenotype to L. maculans (Fig. 3c). Together, these results indicate that the A. thaliana–L. maculans pathosystem is different from the other biotrophic and necrotrophic systems assessed previously (Consonni et al., 2006), in terms of the roles of the pen-dependent and AtMLO2-dependent responses.
We previously found that the effects of RLM1, in contrast to all other reported TIR-NB-LRR R genes, are independent of those of enhanced disease susceptibility (EDS1) and phytoalexin deficient (PAD4) based on single mutant screenings (Bohman et al., 2004; Staal et al., 2006). EDS1 and PAD4 are physically interacting lipase-like proteins and contribute to the activation of SA synthesis in response to many, but not all, SA-inducing pathogens (Wiermer et al., 2005). They can also interact with senescence-associated gene 101 (SAG101) which, together with PEN2, plays important roles in host and nonhost resistance responses (Lipka et al., 2005). A re-evaluation using mutants comprising single, double and triple combinations of eds1, pad4, sag101 and pen2 was made to establish the independence of EDS1 and PAD4. The pen2sag101 double mutant responded as pen2 mutant (Fig. 3c), whereas mutants containing eds1 or pad4 failed to develop chlorosis after L. maculans inoculation (data not shown). The role of EDS1 and PAD4 in RLM1-dependent resistance does appear to be of minor importance, as no true susceptible phenotype could be confirmed for any of the mutant combinations. To shed more light on the question of whether an EDS1/PAD4-independent signaling pathway (Zhang et al., 2008) is present in this particular interaction, fmo1, agd2-like defense response protein 1 (ald1) and mos3 single mutants were assessed, but all showed an incompatible response (data not shown).
Some mutants with an accelerated cell death phenotype, such as acd1, lesion simulating disease 1 (lsd1), and vad1, are susceptible to L. maculans (Bohman et al., 2004; Fig. 3d), while the runaway cell death 1 (rcd1) mutant is not. The vad1ein2 and vad1ein3 double mutants did not show any increased susceptibility compared with vad1. These observations suggest that L. maculans does not necessarily benefit from host-induced cell death, even though it can adopt a necrotrophic mode of growth.
Analysis of RLM1-dependent resistance responses
To determine whether RLM1 contributes to similar resistance responses as reported for various Brassica species, histochemical assessments of pad3 and rlm1Lerpad3 genotypes after L. maculans challenge were made. No obvious difference in lignification response outside the immediate inoculation site could be detected between RLM1Col and rlm1Ler genotypes at 2, 3 or 4 dpi. At 5 dpi, a quantitative lignin staining analysis suggested a difference in lignification at the inoculation sites between pad3-1 and rlm1Lerpad3 (4.6 ± 1.4% and 2.7 ± 1.1% lignified stained area, respectively). The moderately susceptible phenotype of the irx4 mutant, which has 50% less lignin than the wild type, is another line of evidence regarding the role of lignification in A. thaliana resistance to L. maculans (Fig. 3c). Severe lignification was also found in the sgt1b mutant at 4 dpi, indicating involvement of RLM1 responses. Enhanced lignification was observed for the rar1sgt1b mutant compared with the rar1 mutant, suggesting that the restoration of resistance found in the rar1sgt1b double mutant may be attributable to an increase in lignification (Fig. 4a). No ROS response in guard cells could be seen at 16 h or 3 d post surface inoculation in RLM1 or rlm1Ler material. At 6 dpi, however, there was a clear difference between plants inoculated with L. maculans PHW1245 (Fig. 4b,c) and the mock-treated material (data not shown). In the L. maculans-inoculated material, DAB staining occasionally occurred at stomata and more often at hydatodes (Fig. 4d,e), suggesting that an HR-like response takes place to prevent pathogen entry via these natural openings. Chloral hydrate treatment of inoculated pad3 and rlm1Lerpad3 leaves revealed ‘dark matter’ in the vascular structures and intercellular space, similar to that described by Chen & Howlett (1996) as vascular plugs. The vascular plug was clearly more abundant (P < 0.05) and appeared to reduce the spread of L. maculans in pad3-1 compared with rlm1Lerpad3 plants, which showed 60 ± 25% less stained area compared with pad3-1. Responses surrounding the veins, however, appeared to display a similar degree of intercellular fungal growth in the two genotypes.
Quantitative and qualitative analyses revealed interactions among SA-, JA- and ET-induced components in L. maculans resistance
As found previously in proteomic analyses of L. maculans-resistant and -susceptible B. napus harboring Brassica carinata R genes (Subramanian et al., 2005), the L. maculans resistance genes were found not to be required for the induction of the SA- or JA/ET-responsive PR proteins. Similar results have also been obtained in the A. thaliana system (Bohman, 2001). Single mutants impaired in the SA, ET or JA signaling pathways were earlier found to be as resistant as the wild type, but both PR-1 and PDF1.2 marker genes were expressed upon fungal challenge (Bohman et al., 2004). We assumed that further evaluation of the roles of SA and JA/ET signaling would require elimination of the R gene- and camalexin-dependent defense layers. Because of the additive roles of both RLM1ACol and RLM1BCol relative to L. maculans gene-for-gene resistance, the rlm1Ler allele was chosen for further analysis of additional L. maculans defense mechanisms.
A cross between the susceptible double mutant rlm1Lerpad3-1 and the resistant triple mutant coi1-16ein2NahG was carried out. The er mutation in the rlm1Lerpad3-1 mutant and the gl1 mutation in the coi1-16ein2NahG triple mutant were used as markers for successful crosses in the F1 and selection in the F2. Four F3 lines from F2 progeny (rlm1Lerpad3-1 ×coi1-16ein2NahG) with distinct phenotypes were back-crossed with an rlm1Lerpad3-1 (ER, 12.5% Ler-0 genotype) line in order to isolate eight different combinations of mutant and wild-type alleles of COI1 and EIN2, and the presence or absence of the NahG transgene in the rlm1Lerpad3-1 (12.5% Ler-0) background. Like the parental F3 lines, the eight different combinations exhibited distinct categories of lesion phenotypes and susceptibility (Fig. 5a). A correlation between genotype and lesion phenotype was established 14 dpi. The rlm1Lerpad3-1 background with wild-type alleles of EIN2 and COI1 showed rather small areas of chlorosis around the inoculation site, and low levels of fungal growth were detected using GUS-tagged L. maculans (Fig. 5b). The single mutant of coi1-16 in the rlm1Lerpad3-1 background exhibited, in contrast, a severe disease phenotype with extensive necrosis and a reddish/purple color of the leaves. Consistent with its severe phenotype, fungal growth was markedly increased in this line compared with the rlm1Lerpad3-1 background.
Concerning ethylene, ein2-1 as a single mutation did not affect symptom development or fungal colonization in the rlm1Lerpad3 background. By contrast, the rlm1Lerpad3NahGcoi1-16 quadruple mutant with the EIN2 wild-type allele showed the severest necrosis among all the eight genotypes examined (Fig. 5a), showing more necrosis and a larger purple area than the rlm1Lerpad3coi1-16 triple mutant. Fungal growth was also enhanced in this quadruple line compared with rlm1Lerpad3coi1-16. Surprisingly, the disease phenotype of the rlm1Lerpad3ein2coi1-16 quadruple mutant was less severe compared with the rlm1Lerpad3coi1-16 mutant. When fungal growth was quantified, significantly reduced growth was observed for the rlm1Lerpad3ein2coi1-16 quadruple mutant compared with rlm1Lerpad3coi1-16, indicating that EIN2 function is required for enhanced fungal colonization and symptom development in the coi1-16 genotype. The rlm1Lerpad3ein2NahG quadruple mutant exhibited more severe necrosis and chlorosis at the inoculation site than those observed for rlm1Lerpad3ein2. Nevertheless, no clear difference in fungal growth was observed between these two mutants. Finally, the rlm1Lerpad3ein2NahGcoi1-16 quintuple mutant showed a subtle increase in chlorosis and necrosis compared with the rlm1Lerpad3ein2NahG genotype, and fungal growth was also greater, further demonstrating the role of JA in L. maculans resistance. The disease symptom in the rlm1Lerpad3ein2 NahGcoi1-16 mutant is apparently distinct from that in the rlm1Lerpad3NahGcoi1-16 mutant, and fungal colonization in the quintuple mutant was significantly reduced compared with the rlm1Lerpad3NahGcoi1-16 genotype, again indicating that EIN2 is essential for the dramatic disease phenotype conferred by the coi1-16 mutation.
Since 1979, visual disease assessment using the scale developed by Delwiche and Williams for B. napus challenged with L.maculans has represented the standard scoring method for this fungal pathogen. This scale has also been applied to A. thaliana (Bohman et al., 2004; Staal et al., 2006). To evaluate the reliability of the phenotype scoring, we attempted to determine fungal growth using the more unbiased method of real-time PCR. Our qPCR assessments showed a reduction of L. maculans DNA from 0 to 6 dpi followed by relatively constant levels until 14 dpi. There were significant amounts of fungal DNA present for a relatively long time in resistant and nonhost plants, which did not disappear after extensive washing. We interpret these results as our wound inoculation procedure provides entrance to the vascular system of resistant material and that L. maculans is able, to some extent, to establish as an endophyte in the vascular tissue but unable to infect resistant genotypes or nonhost species. Asymptomatic endophytic growth of a pathogen during a ‘latent phase’ of infection or in resistant hosts has also been reported in numerous other systems (reviewed by Schulz & Boyle, 2005). Vascular growth, primarily via xylem vessels, has previously been determined to be the main pathway for the spread of L. maculans throughout susceptible B. napus plants (Hammond et al., 1985). In our previous analyses of the growth of L. maculans in A. thaliana genotypes using green fluorescent protein (GFP) or tryphan blue staining, it was not possible to visualize growth within the vascular tissue (Bohman et al., 2004; Staal, 2006; Staal et al., 2006). The data show the reliability of the phenotypic disease scoring system employed. However, based on our comparisons, we can conclude that the quantitative analysis using qPCR to assess disease susceptibility is a sensitive method that generates extra valuable information in addition to that provided by the visual inspection of disease phenotypes.
The pad3hsp90.1 double mutant displayed clear disease phenotypes, but no enhancement of L. maculans growth could be detected. Similarly, no significant difference in fungal biomass was detected when rlm1Lerpad3 and rlm1Lerpad3NahG, or rlm1Lerpad3ein2 and rlm1Lerpad3ein2NahG were compared, despite a notable variation in symptom development being observed for these genotypes. These observations illustrate that phenotype development and fungal growth are not always synchronized, highlighting the importance of quantitative assessment of fungal growth in planta. One factor that is unclear in the Delwiche and Williams scale is the presence of chlorosis that usually precedes necrosis. It is generally believed that chlorosis is caused by secondary metabolites, for example toxins, produced especially by necrotrophic/hemibiotrophic pathogens, including L. maculans (Howlett et al., 2001; Howlett, 2006). Recent data have, however, shown that fungal toxin production can be regulated by the host plant (Bluhm & Woloshuk, 2005; Maggio-Hall et al., 2005). This raises the possibility that L.maculans can trigger the accumulation of various subsets of host proteins throughout disease progression. Exactly what takes place in the pad3hsp90.1 double mutant is unclear, but distinct chlorotic areas can be observed on average 2 d earlier than in pad3-1.
We have demonstrated that components of the defense response such as the production of callose, lignin and vascular plugs all are R gene-dependent responses in A. thaliana, as in the Brassica systems. Lignification is a common defense response in plant–pathogen interactions, and thus is likely to contribute to resistance (Vance et al., 1980). For example, in the A. thaliana–Pseudomonas syringae pathosystem, lignification has in some cases been found to be dependent on R–Avr interactions (Lee et al., 2001). Lignification in necrotic cell layers at the inoculation site in incompatible A. thaliana–L. maculans interactions has been observed previously (Chen & Séguin-Swartz, 1999). Our analyses indicate that lignification is an R gene-induced response in the A. thaliana–L. maculans interaction and that an efficient lignification response is required to maintain R gene-dependent resistance. It should, however, be noted that a pathogen-induced dirigent (lignan/lignin biosynthesis) gene (At1g64160) linked to the RLM1Col locus is closely related to the pea disease resistance response 206 (DRR206) protein which confers resistance to L. maculans when over-expressed in B. napus (Wang et al., 1999). The Ler allele of this DRR206-like gene linked to rlm1Ler may have a different activity compared with the Col-0 allele, which could lead to a false interpretation of an R gene-dependent lignification response. However, T-DNA insertion mutants in this DRR206-like gene did not cause any alterations of resistance (Staal et al., 2006). Further evidence that lignification is RLM1-dependent is the enhanced lignification response found in the sgt1b mutant genotypes (Fig. 4a).
Upon recognition of L. maculans attack, early PAMP responses take place as a first line of defense in A. thaliana, presumably in an AtMLO2-dependent manner. By analogy with other pathogens, it can be assumed that L. maculans secretes effector proteins to suppress the unspecific PAMP responses. Effector proteins recognized by an R protein (Avr proteins) will then trigger a second line of defense. The RLM1-dependent resistance blocks L. maculans entry both at natural openings and at wounds. However, recent analyses showed that A. thaliana has an R gene-independent pre-invasion defense against L. maculans (Elliott et al., 2008), in particular when long-distance transport via the vascular system is blocked. Local (intercellular) growth, however, is limited by an unspecific induction of the phytoalexin camalexin.
Gene-for-gene disease resistance is often perceived as an ‘arms race’ between the evolution of plant R genes and that of pathogen Avr virulence alleles, which is reflected in the positive selection seen on most plant R genes (Holub, 2001). RLM1 does not show any sign of positive selection, which may be a result of infrequent A. thaliana–L. maculans interactions in nature. Alternatively, the lack of positive selection could be a result of the molecular evolution of the RLM1Col and RLM2Ler loci. The TIR-NB-LRR genes at the RLM1Col locus show presence/absence polymorphism between accessions. R genes with this type of polymorphism do not show strong positive selection (Bakker et al., 2006; Shen et al., 2006; Staal et al., 2006). Another interesting observation is that several Brassica R genes (LepR1, LmR1, CLmR1, Rlm1, Rlm3, Rlm7 and Rlm9) are mapped to loci that correspond to a segment on A. thaliana chromosome 1 that harbors RLM1Col (Delourme et al., 2004; Mayerhofer et al., 2005; Parkin et al., 2005; Staal, 2006; Staal et al., 2006). A similar correspondence between multiple resistance loci in B. rapa and a single chromosomal position in A. thaliana has been observed for clubroot resistance (Suwabe et al., 2006).
The dose-dependent resistance response observed for RLM1 in heterozygote plants is consistent with observations for the B. nigra-derived L. maculans resistance gene Rlm6 in the B. napus background (Huang et al., 2006), supporting the hypothesis that L. maculans resistance is R gene dose-dependent. Another piece of evidence for R gene dose dependence is the susceptible phenotype of rar1 mutants (Staal et al., 2006), as RAR1 is responsible for R protein stability (Bieri et al., 2004). The quantitative requirement of an R gene is also analogous to the recessive or co-dominant nature of the recently identified B. juncea-derived TIR-NB-LRR candidate rjlm2 L. maculans R gene, which shows homology to the TNL-H subclass of A. thaliana R genes and thus is closely related to RLM1 (Saal & Struss, 2005; Staal et al., 2006).
ET, together with JA and SA, is associated with disease resistance to numerous necrotrophic fungi (Glazebrook, 2005). Our previous results demonstrated a complex role of ET in the A. thaliana–L. maculans pathosystem. The L. maculans-susceptible 1 (lms1) mutant displayed an ET-sensitive shoot but an ET-insensitive root response (Bohman et al., 2004), and the ET-sensitive accelerated cell death mutant responsive-to-antagonist 1 (ran1-1) displayed moderate L. maculans susceptibility. The initial screenings of pathogen response mutants also revealed occasional, moderate lesions (not enough for the mutants to be classified as susceptible) on ein3 and ein5/ain1 mutants (ain1, AAA insensitive 1). Further analyses on ein5 in the pad3-1 background revealed that the ein5pad3 mutant was more susceptible to L. maculans than the pad3-1 line (Kaliff, 2007). A similar complex interaction, where ET both was required for some resistance responses and caused enhanced susceptibility, has been seen in the A. thaliana–Verticillium longisporum system (Johansson et al., 2006). The present work demonstrated that functional host ET signaling is required for enhanced fungal colonization in planta and the development of severe symptoms in the rlm1Lerpad3coi1-16 background. Interestingly, in the Nicotiana benthamiana–Botrytis cinerea pathosystem, ET has been suggested to have dual roles affecting both the plant response and the fungal pathogen in their intimate interactions (Chaguéet al., 2006). Whether similar events occur in the A. thaliana–L. maculans pathosystem is presently unclear, but it is not unlikely, as some defense responses to B. cinerea and L. maculans are analogous to each other (Staal et al., 2008).
JA has a positive effect on L. maculans resistance, as evidenced by the markedly susceptible phenotype of the rlm1Lerpad3coi1-16 triple mutant. It appears, however, that EIN2 is necessary for the establishment of devastating symptoms and severe fungal colonization in rlm1Lerpad3coi1-16. The rlm1Lerpad3coi1-16ein2-1 quadruple mutant showed the second lowest fungal growth and much milder symptoms compared with the rlm1Lerpad3coi1-16 triple mutant. These observations suggest that the severe disease phenotype in rlm1Lerpad3coi1-16 is not attributable to the defect in JA-mediated defense responses per se. Under the hypothesis that ET positively affects fungal infection, JA might contribute to disease resistance through repression of ET-mediated enhancement of fungal colonization. While generating the set of hormone signaling-defective mutants presented here, we realized that the pen2-4 mutation is coupled to coi1-16 (Westphal et al., 2008). Thus, at present we cannot rule out the possibility that pen2-4 interferes with the disease phenotypes observed in our mutant collection.
We found that responses in A. thaliana resemble those described previously in the Brassica systems. Necrotic lesions on B. napus are usually only seen late in the growth season, indicating that L. maculans prefers an asymptomatic mode of growth until it is triggered by an external stimulus to enter necrotrophic ‘reproductive growth’ mode and form pycnidia/pycniodiospores (Howlett et al., 2001). We found additional genetic components in the A. thaliana background that influence resistance and characterized mechanisms of L. maculans containment in compatible interactions. Hormones such as SA, JA and ET all indirectly influence the growth of L. maculans, directing it into different progression strategies. In the case of JA, we can conclude that it has an antagonistic effect on fungal infection, as the rlm1Lerpad3-1coi1-16 mutant exhibits a clear disease phenotype and fungal growth. ET, however, has an inducing effect on fungal infection as wild-type alleles are necessary for full susceptibility, at least in certain genetic backgrounds. In the present analysis, the role of SA in A. thaliana–L. maculans interactions could not be determined clearly. The NahG transgene may cause additional effects not directly linked to SA-dependent responses in A. thaliana (Van Wees & Glazebrook, 2003). However, no disease phenotype was previously observed in NahG, NahGein2, SA-induction deficient (sid2; which is defective in isochorismate synthase 1 and consequently contains a reduced concentration of SA) or sid2ein2 mutants in the Col-0 background (Kaliff, 2007), suggesting a minor role for SA in this pathosystem. Our results indicate, furthermore, that defense responses induced by JA and ET in A. thaliana act as contrasting developmental triggers in L. maculans. Interestingly, the nuclear-localised bHLHzip-type transcription factor (AtMYC2) that differentially regulates two branches in the JA-signalling pathway causes a COI1-dependent repression of JA/ET-dependent PR proteins in favor of wound-induced transcripts (Lorenzo & Solano, 2005), which might explain the opposite roles of ET and JA in this system.
The switch from an endophytic to a pathogenic lifestyle has been suggested to be regulated by membrane domains, calcium or ROS production (Kogel et al., 2006). However, not much is known about the mechanisms underlying this mutual and finely-tuned balance, ranging from host entry and local colonization to system spread of the pathogen. A single mutation in Colletotrichum magna, for example, changed the fungus from a hemibiotroph in cucurbit plants to an asymptomatic endophyte with a wide host range (Kogel et al., 2006). The finding that Magnaporthe grisea, a leaf pathogen on rice (Oryza sativa), can switch to a root-infecting fungus is also interesting in this context (Sesma & Osbourn, 2004). Leptosphaeria maculans is an emerging model system for phytopathogenic fungi (Fitt et al., 2006), which offers excellent future opportunities for dissection of the complex hemibiotrophic host–pathogen interactions. We envisage that the imminent full-genome sequence of L. maculans together with forward and reverse genetics (Rouxel & Balesdent, 2005; Gout et al., 2006; Kuhn et al., 2006), in combination with host studies on A. thaliana will improve our fundamental understanding of the complex parasitic strategy of L.maculans, including biotrophic/endophytic and necrotrophic growth.
We would like to thank J. Glazebrook for the ein2NahG mutant and J. Turner for the coi1-16 conditionally fertile mutant. We are also grateful to D. Roby for vad1, vad1ein2-1 and vad1ein3-1, J. Kangasjärvi for rcd1, X. Li for mos2, R. Innes for pbs1 and pbs3, B. Holt for rar1-21, sgt1b-edm and rar1sgt1b, J. Dangl for lsd1, P. Schulze-Lefert for pen2/eds1/pad4/sag101 mutant combinations, H. Thordal-Christensen for fmo1, ald1 and mos3, J. Schmid (Massey University, New Zealand) for plasmids, B. Howlett for fungal isolates and fruitful discussions and R. Hopkins for language corrections. This work was supported by FGB, IMOP, the Nilsson-Ehle and Helge Ax:son Johnsons Foundations, and the Swedish Science Council (VR).