Background

Quantitative disease resistance (QDR) is a complex phenomenon involving a plethora of molecular mechanisms [1, 2]. It is often a sustainable form of resistance relying on multiple genes and effects which cannot easily be overcome by the pathogen’s adaptation. Verticillium longisporum is a soil-borne vascular pathogen of recent evolutionary origin with a host range centred on crucifers [3]. QDR is the only form of resistance against this pathogen described so far. V. longisporum causes significant and increasing yield losses on oilseed rape [4–7]. The fungus enters the host via the root and, at the onset of flowering, spreads systemically inside the xylem, thereby colonizing the whole plant [5, 8, 9]. V. longisporum induces early senescence [6, 10] and colonizes senescent tissue to form microsclerotia that persist in the soil [11]. Since the disease is difficult to control, durable resistance in host plants is highly desirable.

Quantitative resistance against V. longisporum has been described for various accessions of Brassica species [12–15] (Konietzki and Diederichsen unpublished) and also for A. thaliana ecotypes [8, 10, 16, 17]. In some cases, QTL have been identified which control resistance traits, such as fresh weight [16], chlorosis [10, 16], systemic colonization, stunting, and axillary branching [8], or the area under the disease progression curve in Brassica species [15] (Konietzki and Diederichsen unpublished). However, the underlying genes and regulatory processes have rarely been identified. Secondary metabolism plays a role in resistance: A resistant line of B. napus produced more phenolic substances in the xylem of the hypocotyl upon infection than a susceptible line [18]. Indeed, QTL for contents of phenylpropanoid compounds co-localised with resistance QTL in B. napus [19], and it was shown that soluble phenylpropanoids played a role in A. thaliana defence against V. longisporum [20]. Rfo1 has been shown to mediate resistance against V. longisporum-induced fresh weight loss in A. thaliana [16], and encodes a wall-associated kinase-like (WAKL) -protein that conferred resistance against Fusarium oxysporum [21].

Recent research has revealed different processes that are involved in the host’s response to V. longisporum on the molecular level: Ethylene signalling plays either a protective or a deleterious role, depending on the signalling components involved [16, 22]. V. longisporum infection caused elevated levels of salicylic acid (SA) in the xylem of Brassica shoots [23]. Previous studies suggest that jasmonic acid (JA) signalling does not play a role in the host-pathogen interaction [10, 23], but the JA receptor COI1 promotes the disease in a JA-independent, yet unknown way [24]. Whereas an interplay of indole glucosinolates and camalexin has been shown to be involved in early defence against V. longisporum in A. thaliana roots [25], reactive oxygen species played a role in defence during the later stages of the disease [26]. Furthermore, the nuclear-localised ahl19 gene acted as a positive regulator of defence to V. longisporum and other Verticillium species in A. thaliana [27]. Several apoplastic enzymes were induced by V. longisporum in Brassica and possibly play a role in defence [28]. The host’s reaction to the pathogen involves trans-differentiation of bundle sheath cells into functional xylem elements under the control of the vascular-related NAC domain 7 transcription factor [29].

However, little is known about how these processes relate to QDR. It is unclear whether resistance QTL represent genes within regulatory systems that have already been shown to operate in the host-pathogen interaction, or whether they constitute new components, adding to the complexity of the pathosystem. Furthermore, the role of known defence signalling pathways in natural resistance to V. longisporum is poorly understood so far.

The present study aims at identifying genes and signalling pathways that account for differences in QDR against V. longisporum in A. thaliana. QTL for relevant resistance traits using a new (Bur×Ler) recombinant inbred line (RIL) population have been identified. It is shown that the Erecta gene corresponded to a strong QTL mediating stunting resistance and that a functional Erecta signalling pathway mediated resistance against V. longisporum-induced stunting. Evidence is provided that SA, abscisic acid (ABA) and JA contents responded to V. longisporum in a genotype-specific way, and that changes in ABA and JA content were controlled by the same QTL that also conferred resistance to systemic colonization by the fungus.

Methods

Material

A. thaliana ecotypes Bur-0, Col-0 and Ler-0 were originally obtained from the Arabidopsis Information Service (AIS) Frankfurt [30] and maintained in-house. All other A. thaliana genotypes were obtained from the Nottingham Arabidopsis Stock Centre (NASC). er-105, er-108, er-111, and er-118 were included as strong erecta mutants, whilst er-116 represented a weaker erecta mutant [31, 32]. agb1-1, which is defective in the β-subunit of the heteromeric G-protein, was included as a mutant of a signalling component acting downstream of Erecta [31]. The V. longisporum isolate ‘43’ (V43) [33] was used for inoculation experiments.

Generation of the (Bur×Ler) RIL population and near-isogenic lines (NILs)

A total of 189 RILs were created originating from an F1 between the ecotypes Bur-0 (♀) and Ler-0 (♂). A total of 189 F2 plants were propagated via single-seed descent to the F6 generation. All plants were grown in pots of 5 cm diameter in a greenhouse under long-day conditions (16 h light/8 h dark) at 20°C. A total of 94 F6 plants were genotyped and phenotyped in F7. Information about the RIL population will be submitted to the The Arabidopsis Information Resource (TAIR) database [34], and F8 seeds from bulked F7 offspring of the genotyped F6 plants will be made available through the A. thaliana stock centres Arabidopsis Biological Resource Center (ABRC) and NASC.

NILs were created by selfing RIL21, which was heterozygous for markers EH2-4 to nga361 on chromosome 2 and homozygous for all other marker loci investigated. NIL5 was selected from RIL21 offspring as homozygous for Bur alleles in the variable region, and NIL9 for Ler alleles.

Marker development and analysis

The RILs were genotyped with 73 markers that were polymorphic between the parental ecotypes Bur and Ler. Among them were 39 simple sequence repeat (SSR) markers, 21 sequence-characterized (SCAR) markers exploiting length polymorphisms between Bur and Ler, and 12 cleaved amplified polymorphic sequence (CAPS) markers developed on the basis of single nucleotide polymorphisms (SNPs; see Additional file 1). The erecta mutation was used as a morphological marker, however, a CAPS marker (BLC2-1) has also been developed to differentiate between the Bur and the Ler allele of Erecta. Fifty-one markers have been published before [8, 34, 35], and 22 new markers were designed to increase marker density especially in regions of major QTL (Additional file 1). Length polymorphisms between Bur and Ler were identified using the multiple SNP query tool (MSQT) database [36]. SNPs available through the seqviewer tool of TAIR [37, 38] were used for CAPS marker design. Webcutter 2.0 [39] was used to identify differential restriction sites at the sites of SNPs. Primers were designed on the basis of sequence information provided by TAIR.

Inoculation experiments

Inoculation experiments were performed in the greenhouse or in a growth chamber (Grobank, Mobylux, Germany) under long-day conditions (16 h light/8 h dark) on soil as described previously [8]. Thirty plants were grown in trays measuring 20 × 30 cm. Plants of the same treatment and genotype were arranged in batches of 15 (1/2 tray) or 30 plants (1 tray). Batch replicates were randomized. Plants flowered during the experiment and were grown until the first siliques turned yellow. Two experiments for phenotyping the RIL population were carried out with the same 94 lines (Table 1, experiments 1 and 2). Comparisons between wild type (WT) and erecta mutants were performed in two greenhouse experiments (Table 1, experiments 3 and 5), and phytohormone contents were analysed in three experiments (Table 1, experiments 4, 5 and 6). Regarding the analysis of phytohormone contents during development in Bur and Ler (Table 1, experiment 4), plants were harvested at three different developmental stages: Stage I (early-flowering stage) was analysed in Ler at 20 days post inoculation (dpi) and in Bur at 31 dpi. Stage II (mid- to late-flowering stage) was analysed at 27 dpi in Ler and at 38 dpi in Bur. Stage III (onset of silique maturity) was analysed at 31 dpi in Ler and at 48 dpi in Bur.

Table 1 Overview of inoculation experiments (exp.)
Full size table

Phenotypic analysis

Traits were recorded by different parameters (Table 1). Stunting resistance was measured as “performance height” (mean heightinoculated/mean heightcontrol × 100) and “performance fresh weight” (mean FWinoculated/mean FWcontrol × 100). The performance parameters compensate for differences in plant height, and fresh weight between erecta mutants and WT. The height was measured between the hypocotyl and the apex of the longest shoot at the beginning of silique maturation when shoots were out-grown. Systemic colonization was determined in apical segments of the main shoot at the onset of silique maturity. To determine the percentage of colonized plants per replicate, one segment per inoculated plant was placed on a malt agar plate as described previously [8], and the percentage of colonized segments was calculated from batches of 15-30 plants. In order to measure the fungal DNA via qPCR, 100 mg of shoot material was cut from segments used for plating and shock-frozen in liquid nitrogen. qPCR using Verticillium-specific primers was performed as described [40]. Chlorosis was determined by counting yellow and green rosette leaves. Three different parameters were used to quantify the trait: The number of yellow rosette leaves in inoculated plants was determined as the most direct measure of chlorosis. Furthermore, the percentage of yellow leaves relative to the total rosette leaf number was calculated to express the degree of chlorosis affecting the leaf rosette. As a third parameter, the difference in yellow leaves between inoculated and mock-inoculated plants of the same line is given, thus taking natural senescence into account. Chlorosis in both RIL experiments (experiments 1 and 2, Table 1) was assessed at 17 dpi when the control plants still showed little or no senescence-associated chlorosis, while the V. longisporum-inoculated plants were showing chlorotic leaves. In order to determine the time-course of chlorosis, mock-inoculated and inoculated Ler and Bur plants were assessed every 3-4 days after inoculation in a growth chamber experiment (experiment 7, Table 1). Cotyledons were not included.

Phytohormone quantification by HPLC-ESI-MS/MS

Approximately 6 cm from the upper half of the stalk were sampled from 15-30 plants of the same treatment and pooled in a single sample. Material was shock-frozen in liquid nitrogen directly after harvest, subsequently lyophilized and ground to a fine powder in a ball mill. An aliquot of 50-100 mg (dry weight) was placed in 2.0 ml tubes and used for phytohormone extraction according to a modified protocol [23]. One ml extraction solvent (20% acetone, 79% H20 and 1% CH3COOH, and 2 ng of the deuteriated internal standard D6-salicylic acid, Sigma-Aldrich Co.) was added to each sample. Extraction was performed for 45 min at 4°C on a rotary shaker (160 rpm). Subsequently, each sample was spiked with 1.0 ml diethyl ether (DEE), shortly vortexed, and shaken (160 rpm, 4°C) for an additional 30 min. The sample was centrifuged at 8500 g for 5 min. The upper DEE phase was transferred to a new 1.5 ml tube, and the remaining aqueous phase was re-extracted with 1.0 ml DEE exactly as in the previous step. Both DEE fractions were pooled and dried under vacuum at 30°C, re-dissolved in 200 μl HPLC (high-performance liquid chromatography) solvent (1:1 H2O:MeOH containing 7 mM CH3COOH), centrifuged once more at 8500 g for 5 min and transferred (180 μl) to an HPLC conical vial (200 μl internal volume, WICOM Germany).

Extracted samples (10 μl) were injected into the HPLC and eluted at 40°C at a flow rate of 0.2 ml/min-1. Chromatography was performed on a Kinetex® C18 (100 mm × 2.10 mm with 2.6 μm particle size) column equipped with a C18 guard column purchased from Phenomex Inc. (Aschaffenburg, Germany). The following programme was used for elution: 80% solvent A (water:acetonitrile 95:5) containing 7 mM CH3COOH and 20% solvent B (methanol containing 7 mM acetic acid) for 40 s; ramp to 98% B in 50 s; hold for 2 min 20 s followed by re-equilibration to 20% B.

Phytohormones were detected in multiple reaction monitoring mode (MRM) in a triple quadrupole mass spectrometer (LC12000) equipped with an electrospray interface using settings described in [41]. The following mass transitions (collision energy: CE) were used: SA 136.8/93.0 (CE 14.5 eV); d6-SA 140.9/97.0 (18.5 eV), JA 208.9/59.0 (9.5 eV), and ABA 262.8/153.0 (8.0 eV). A calibration curve of the ratio of peak areas of the unlabelled standard to the peak area of the deuterium-labelled standard was used for the quantification of SA. Other phytohormones were quantified with an external calibration curve obtained with pure standard.

Map construction and QTL analysis

Linkage groups were determined and allele frequencies were tested for segregation distortion with JoinMap [42]. Map construction and QTL analysis was performed with MapManager QTX 20b [43] using the Haldane mapping function. Simple interval mapping was performed scanning the genome in 1 cM-steps. MapManager QTX gives the LRS (likelihood ratio statistic) value to assess the probability of a false positive, where LRS = 4.6 × LOD (likelihood of odds) [44]. LRS significance threshold values for the 37% (suggestive), 95% (significant) and 99.9% (highly significant) genome-wide confidence levels were determined by permutation tests with 10,000 permutations. Confidence intervals for QTL were determined by bootstrap tests which calculate the QTL position for multiple resampled datasets of the original dataset. Epistatic interactions were searched using the “interaction” function of MapManager QTX 20b, testing pairs of markers for a possible interaction component. The confidence criterion for the total effect of a marker pair was set to p = 10-5. MapQTL® 6 [45] was also used for mapping to include covariates and cofactors in the QTL analyses. Cofactors were initially specified according to peak positions in interval mapping and selected by backwards elimination using the “Automatic cofactor selection” tool of MapQTL® 6. Maps were visualized using MapChart [46].

Statistics

All inoculation experiments were performed with batches of at least 15 plants per genotype and treatment. Most inoculation experiments were performed on 6 to 10 batch replicates. Some parameters, such as phytohormone contents, the performance parameters or the percentage of colonized shoot segments, were calculated on a batch basis, whereas some of the chlorosis parameters were calculated on a single-plant basis. Only one batch of 15-30 plants was inoculated per genotype to allow testing of 96 genotypes at the same time in the RIL inoculation experiments. The whole experiment was repeated to confirm the results. See Table 1 for an overview of single-plant and batch replicates. All statistical analyses, as indicated in the Results section, were performed with SPSS 20 [47].

Results

QTL controlling resistance traits against V. longisporum in the (Bur×Ler) RIL population

The parental ecotypes Bur and Ler differed in important resistance traits against V. longisporum (Figure 1). Ler was susceptible to systemic colonization by the pathogen, whereas Bur showed a high degree of resistance (Figure 1a). Bur, however, was more susceptible to V. longisporum-induced stunting than Ler. This phenotype was suppressed under high-light conditions in the greenhouse during summer (Figure 1c, d). V. longisporum promoted chlorosis in both Ler and Bur compared to mock-inoculated controls. Both natural and V. longisporum-induced chlorosis occurred much earlier in the early-flowering ecotype Ler than in the late-flowering ecotype Bur (Figure 1e).

Figure 1
figure 1

Phenotypic characterization of different A. thaliana genotypes. a,b) Systemic colonization of a) Ler, Bur and two NILs differing in the vec1 region (experiment 6; one-way ANOVA and post-hoc Tukey test, p < 0.05, n = 6-7), and b) two genotypes differing only in Erecta function (experiment 5, t-test, p < 0.05, n = 6). c,d) Resistance to V. longisporum-induced stunting expressed as performance height in c) an experiment performed in winter (experiment 1, t-test, p < 0.05, n = 6) and d) an experiment performed in spring (experiment 2, t-test, p < 0.05, n = 6). e) Proportion of chlorotic rosette leaves in Bur (squares) and Ler (circles) after V. longisporum-inoculation (closed symbols) and mock-inoculation (open symbols) during the course of a growth chamber experiment. Vertical bars denote standard deviations. Ecotypes and treatments differed significantly from 6 dpi on (p < 0.05; t-test; n = 41 for Ler mock, n = 38 for Ler V. longisporum, n = 60 for Bur mock, n = 58 for Bur V. longisporum).

Full size image

QTL controlling V. longisporum resistance traits were mapped in a new (Bur×Ler) RIL population. A genetic map was calculated and a physical map was produced using known positions of each marker on the Arabidopsis Genome Initiative (AGI) reference map [48] (Additional file 2). In the genetic map, chromosomes were supported as linkage groups with LOD scores ranging from 4 (chromosome 1) to 10 (chromosome 4). Genetic marker order on each chromosome was the same as for the physical map. The complete map size was 407.3 cM, with an average marker spacing of 5.7 cM and the largest distance between two markers being 17.6 cM. The proportion of heterozygous markers in F6 was 3.65%, which agreed well with a predicted value of 3.125%.

QTL were detected for all three resistance traits (Figure 2, Additional file 3). QTL controlling Verticillium c olonization (vec) were detected on chromosomes 2 and 4, QTL for st unting resistance (stre and r-stre) on chromosomes 1, 2 and 4, and QTL for resistance against V. longisporum-induced chl orosis (r-chl) also on chromosomes 1, 2 and 4. QTL were named according to the nomenclature used previously [8]. QTL exclusively discovered in the RIL population were prefixed with an “r-”, indicating their origin from RIL mapping. All parameters characterizing the traits showed either a normal or a multi-modal frequency distribution, both of which can occur in traits controlled by several QTL (Additional file 4).

Figure 2
figure 2

QTL for V. longisporum -related traits in the A. thaliana (Bur×L er ) RIL population. White bars represent chromosomes with marker positions indicated in cM. Coloured bars delimit the confidence intervals of QTL as follows: Light yellow: Systemic colonization, % colonized shoot segments, exp. 1. Dark yellow: Systemic colonization, % colonized shoot segments, exp. 2. Red: Stunting resistance, performance height, exp. 1. Dark green: Chlorosis, number of yellow leaves, exp. 1. Medium green: Chlorosis, % yellow leaves, exp. 1. Light green: Chlorosis, difference in yellow leaves between mock-inoculated and V. longisporum-inoculated plants, exp. 1. Black lines within the bars stand for LRS peak positions. Upward arrows indicate that the paternal ecotype (Ler) was the origin of the increasing allele; downward arrows stand for Bur as the source of the increasing allele. QTL from the most representative dataset are displayed in Figure 1. QTL names are written next to the bars. For reproducibility and further QTL information, see Additional file 3.

Full size image

All vec QTL controlling systemic colonization consisted of arrays containing several LRS peaks depending on the parameter and the inoculation experiment (Figure 2, Additional file 3). The map positions of the peaks were reproducible for both parameters (plating assay and qPCR) and in both inoculation experiments (Additional file 3). vec1 explained up to 19.3% of the trait variance, depending on the parameter and the experiment, vec2 up to 15% and vec3 up to 17.3% (Additional file 3). The vec1 QTL could be confirmed in NILs that differed in a ~ 3 megabase-segment on chromosome 2 comprising vec1. NIL9 with Ler alleles in the variable region was more susceptible to systemic colonization than NIL5 with Bur alleles in the respective region (Figure 1a). All vec alleles increasing the degree of colonization were of Ler origin. Thus, colonization resistance was entirely conferred by the Bur alleles of the respective QTL. The co-localisation of colonization and developmental QTL on chromosome 4 (Additional file 3) suggested an impact of development on systemic colonization, although all assessments were made at defined developmental stages. Development parameters were used as covariates in interval mapping or multiple QTL mapping (MQM) with MapQTL. Using the developmental parameters as a covariate, all vec QTL on chromosome 4 vanished, and vec1 on chromosome 2 remained as the only chromosomal region showing a significant effect on this resistance trait. Co-factor selection in the QTL region on chromosome 2 and application of the selected co-factors in MQM confined the QTL region controlling the degree of systemic colonization (both parameters) to a fragment between markers EH2-6, BLC2-2 and erecta, spanning approximately 1.6 cM.

QTL for stunting resistance could only be detected with the parameter “performance height” and not with “performance fresh weight”. Both QTL on chromosome 2 were detected in both inoculation experiments; the QTL on chromosomes 1 and 4 were only detected in experiment 1, which was performed in the winter. stre2 co-localised with marker BLC1-26, which is located in the Wakl9 gene, and explained up to 22.9% of the trait variance. stre1, explaining 16.8% variance, was also located in the vicinity of a Wakl gene (Wakl22/rfo1). r-stre1 near marker erecta on chromosome 2 explained approximately 30% of the trait variance in experiment 1 (Figure 2, Additional file 3). The stunting resistance alleles of the QTL on chromosome 2 were of Bur origin, whereas the stunting resistance QTL on chromosomes 1 and 4 were contributed by Ler. The fact that both parents contributed stunting resistance QTL explains the strong transgressive segregation of the trait (Additional file 4). As for the degree of systemic colonization, the development apparently had an impact on the degree of stunting. Including development traits as a covariate in mapping reduced the number of QTL controlling this trait by eliminating the QTL r-stre3 on chromosome 4, which had been detected in experiment 1.

Significant QTL for chlorosis parameters (r-chl) were only detected in experiment 1 (Figure 2, Additional file 3). r-chl1 on chromosome 4 was the strongest QTL explaining up to 29.3% of the trait variance. r-chl2 on chromosome 2 explained 11.7% and r-chl3 on chromosome 1 up to 14.8%. Surprisingly, all alleles conferring resistance to chlorosis were of Ler origin. Two QTL accelerating flowering time of Ler origin were found in the same region (dt1 and dt2, see Additional file 3). In general, a later onset of flowering is expected to correlate with a later onset of senescence. Furthermore, using development traits as a covariate did not affect r-chl1 on chromosome 4. Both facts are strong evidence that QTL have been mapped that were specific for V. longisporum-induced chlorosis and not QTL controlling development.

Epistatic interactions between marker loci could not be reproduced between the tests. Regarding the colonization data determined by the plating assay in experiment 1, a significant interaction between marker loci BLC2-7 on chromosome 2 and nga1111 on chromosome 4 was detected. Both loci lie within the confidence intervals of vec1 and vec2, respectively. In summary, it could be shown that QTL from both parents controlled different resistance traits against V. longisporum in the (Bur×Ler) RIL population.

V. longisporum-induced stunting was controlled by ERECTA signalling

The strong stunting resistance QTL r-stre1 co-localised with the morphological marker erecta. Erecta encodes for a receptor-like kinase with many functions in plant development and also in QDR [49]. The function of this protein in Ler is disrupted [32], while Bur contains a functional Erecta allele. Inoculation experiments were performed on erecta mutants and their respective WT ecotypes having a functional Erecta allele to investigate whether a functional Erecta pathway increases resistance. Ler was compared with the La-0 ecotype. Col-0 was compared with erecta mutants in the Col-0 background. The agb1-1 mutant in the Col-0 background that is defective in the β-subunit of heteromeric G-protein was included representing another component of the Erecta signalling pathway [31]. All erecta mutants and the agb1-1 mutant were significantly more stunted under V. longisporum challenge than their respective WT lines: The parameters “performance height” and “performance fresh weight” were decreased in all mutants compared to WT (Figure 3a, b). The weaker er-116 mutant did not perform better than the strong mutants. The developmental phenotype of er-116, however, was clearly attenuated: The mean height of the controls was 22.3 cm compared to 16.5 cm in the strong erecta mutants. A functional Erecta did not contribute significantly to colonization resistance in any of the two backgrounds (Figure 1b, Additional file 5), despite the strong colonization QTL vec1 that was localized near Erecta. Because one QTL for chlorosis (r-chl2) was localized in the Erecta region, it was investigated whether loss of Erecta function influences chlorosis. The level of chlorosis in Col-0 was higher than in the erecta mutants with Col-0 background (Figure 3c), suggesting a chlorosis-promoting effect of Erecta. However, the effect of Erecta was opposite in La-0/Ler: La-0 suffered less chlorosis than Ler (Figure 3c). Thus, a background-specific effect of Erecta on chlorosis must be assumed.

Figure 3
figure 3

Reaction of Erecta signalling mutants and corresponding WT lines to V. longisporum . La-0 is the corresponding WT for Ler, Col-0 for er-105, er-108, er-116, er-118, and agb1-1. Different WT/mutant combinations are separated by a vertical bar. Recorded parameters are a) performance height, b) performance fresh weight and c) % chlorotic leaves in inoculated plants. N = 10 for a,b), N between 91 and 260 for c). Significance of La-0/Ler differences was tested by t-tests; differences between lines with Col-0 background by one-way ANOVA and subsequent multiple comparisons (Tukey test). Means marked with different letters differed significantly at p < 0.05. Vertical bars denote standard deviations.

Full size image

It was demonstrated that Erecta underlay stunting resistance QTL r-stre1 and that a functional ERECTA signalling pathway mediated stunting resistance in A. thaliana.

SA and low light were associated with V. longisporum-induced stunting

SA content increased in A. thaliana stalks after infection with V. longisporum in both Bur and Ler (Figure 4). SA levels were strongly increased in Bur at early- to mid-flowering stages, while SA showed a mild but significant increase in Ler with the advance of flowering (Figure 4). SA hyperinduction was neither associated with vec1 nor with Erecta, as could be shown in NILs and erecta mutants (Additional files 6 and 7). Furthermore, Bur showed high susceptibility to V. longisporum-induced stunting in winter experiments (Figure 1c). This season-dependent susceptibility could be attributed to the Bur alleles of QTL stre1, stre2 and r-stre3, the first two co-localising with Wakl genes. Greenhouse experiments were always performed in long-day conditions with additional illumination at temperatures between 20 and 25°C. However, the intensity, dosage and spectral composition of ambient light were major factors differing between spring/summer and winter experiments: Quantum flux densities of artificial illumination alone ranged between 80 and 100 μmol*s-1*m-2, whereas additional natural daylight provided between 200 and 400 μmol*s-1*m-2.

Figure 4
figure 4

Contents of the phytohormone SA in L er and Bur, mock-inoculated and inoculated, at different developmental stages. I: early-flowering stage, II: mid-flowering stage, III: beginning of silique maturity. N = 8. Significance levels refer to differences of means between the ecotypes within one treatment (mock- and V. longisporum-inoculated, respectively) as determined by t-tests. Differences between mock- and V. longisporum-inoculated plants were always significant. V43 = V. longisporum isolate 43.

Full size image

It is hypothesized that SA induction and low light caused V. longisporum-induced stunting mediated by the seasonally influenced QTL stre1, stre2 and r-stre3 (see Discussion).

ABA and JA response to V. longisporum infection was controlled by vec1

Bur and Ler differed in their ABA and JA responses to V. longisporum: ABA contents were much higher in Ler than in Bur upon infection (Figures 5 and 6). Changes of JA levels were opposite in Ler and Bur at the maturity stage: the JA level in Ler stayed the same or increased, whilst the JA levels in Bur dropped in infected plants compared to mock-inoculated plants (Figures 5 and 6). Analysis of ABA and JA content in NILs at the maturity stage showed that this difference was associated with the genomic region containing vec1 and Erecta on chromosome 2 (Figure 6). NIL9, which was homozygous for Ler alleles in the variable region, showed strong ABA induction and no JA decrease – a reaction that was similar to the one in Ler. NIL5 with Bur alleles in the variable region showed a reaction of both phytohormones that resembled Bur: ABA increased less strongly and JA decreased in infected plants (Figure 6). The characteristic phytohormone response patterns corresponded to high systemic colonization in Ler and NIL9 and low colonization in Bur and NIL5 (Figure 1a). No effect of Erecta on ABA and JA levels in V. longisporum-infected plants was found when erecta/WT comparisons were made for ABA and JA contents (Additional file 7). ABA and JA signalling pathways are candidates for the control of systemic colonization by V. longisporum in A. thaliana.

Figure 5
figure 5

Contents of the phytohormones a) ABA and b) JA in L er and Bur, mock-inoculated and inoculated, at different developmental stages. Light blue: Ler control, dark blue: Ler inoculated, pink: Bur control, red: Bur inoculated. I: early-flowering stage, II: mid-flowering stage, III: beginning of silique maturity. N = 8. Significance levels refer to differences of means between the ecotypes within one treatment (mock- and V. longisporum-inoculated, respectively) as determined by t-tests. Differences between mock- and V. longisporum-inoculated plants were significant except for ABA in Bur at stage III, JA in Ler at stages I and II and JA in Bur at stage II (p < 0.05, t-test).

Full size image
Figure 6
figure 6

Contents of the phytohormones a) ABA and b) JA in Bur , L er and two near-isogenic lines. NIL9 contained Ler alleles in the variable region and NIL5 contained Bur alleles. a): Differences between mock- and V. longisporum-inoculated plants were significant (n = 6-7, p < 0.05). b) Asterisks refer to the significance level of differences between mock-inoculated and V. longisporum-treated plants within one genotype (t-test, n = 6). Mean contents of inoculated plants marked with different letters differed significantly at p < 0.05. Mean phytohormone contents of controls did not differ significantly (one-way ANOVA and post-hoc Tukey test). Vertical bars denote standard deviations. V43 = V. longisporum isolate 43.

Full size image

Discussion

Considerable progress has been made in understanding the molecular interactions of V. longisporum with its cruciferous hosts at a molecular level during recent years. Many genes and pathways have been shown to be involved, revealing an enormous complexity. The present study has been designed to further disentangle the complex network and to identify new components of genetic variation contributing to quantitative resistance to V. longisporum.

Our results corroborate the view that different resistance traits are controlled by different pathways and can be inherited independently. A plant showing a high degree of colonization resistance may still be susceptible to stunting. No single master mechanism controlled QDR in A. thaliana. Nevertheless, a higher resistance level can be achieved by combining a few well-defined QTL. Some components of the physiological basis of resistance were elucidated.

Phenotypic variation for colonization resistance and stunting resistance was assigned to specific QTL that have been partly detected before, but were mapped with much higher resolution in the RIL population. vec1, vec2 and vec3 were found to control systemic colonization by V. longisporum in an F2/F3 population originating from the same parents [8] and could be reproduced in the present study. RIL mapping revealed a high complexity of the vec QTL. Multiple, reproducible LRS peak positions suggest complex loci in which more than one gene contributed to the effect. The genes underlying the vec QTL are still unknown, but recent results strongly suggest that genes involved in the phenylpropanoid pathway underlie vec3 [20]: The genes cad5, cad8 and ugt84a3, encoding for cinnamyl alcohol dehydrogenases and an UDP-glycosyl-transferase, are all located on chromosome 4 close to vec3. Soluble phenylpropanoids have been shown to play a role in defence against V. longisporum [20]. An impact of development on resistance could be demonstrated. Systemic colonization of the upper parts of the plant starts with the onset of flowering in A. thaliana and Brassica spp. [8, 9]. Direct or indirect signals that induce or promote flowering possibly affect the development of V. longisporum as well, for example, by stimulating the formation of mobile conidia or by directing growth of fungal hyphae. The loss of the vec2 and vec3 QTL when using development as a covariate in QTL analyses indicates that the developmental differences on a physiological basis were still strong enough to cause the detection of these QTL, even though all plants had been assessed at the same phenotypic developmental stage. Developmental implications also played a role in stunting. The stunting resistance QTL on chromosome 4, r-stre3, similar to vec2 and vec3, disappeared when development was used as a covariate in MQM mapping.

The stunting resistance QTL on chromosomes 1 (stre1 and stre2) and 4 (r-stre3) depended on the season: They protected against stunting only during winter, which is consistent with prior observations for stre1 and stre2 [8]. Despite complementary lighting in the greenhouse, the development of A. thaliana was prolonged during winter (experiment 1) and developmental differences were more pronounced. Differences in light intensity and/or quality due to more natural illumination during summer are likely to cause these reactions. The strongest QTL on chromosome 1, stre2, co-localised with the Wakl9 gene, encoding a wall-associated kinase (WAK)-like protein [50]. A candidate for stre1 at the bottom of chromosome 1 is Rfo1 (Wakl22), encoding another WAK-like protein which has already been shown to play a role in V. longisporum resistance [16]. Since V. longisporum interacts with the host plant primarily in the apoplast [51], perception of pathogenesis-related molecules by cell-wall associated proteins may be crucial, and WAK proteins have been shown to mediate such perceptions [52]. Interestingly, several Wak genes have been shown to be induced by SA [53]. An SA induction of Wakl genes has not yet been shown, but it would correspond well with the early and strong SA induction detected in Bur. Increased SA levels have been observed to cause stunted growth also without disease [54]. The stunting effect of SA has been shown to be partially reverted by high-light conditions [55]. The combined effects of high levels of SA and lighting differences could explain the pronounced Verticillium-induced stunting mediated by the stre1 and stre2 alleles of Bur origin during winter. High levels of SA and its glucoside were also found in the xylem sap of V. longisporum-infected B. napus and were correlated to the degree of stunting and the amount of pathogen DNA [23]. Stunting is a common symptom in greenhouse or growth chamber experiments with V. longisporum occurring already at the rosette stage, but is never seen in infected field crops. Light intensity and quality is a major difference between field and greenhouse. This suggests that V. longisporum-induced stunting under experimental conditions also depends to a certain extent on the combination of high SA levels with artificial lighting.

The QTL controlling chlorosis seemed to be less affected by genes controlling development. The opposite has been reported in the literature [10]: A QTL delaying development and mediating resistance against Verticillium-induced chlorosis, Vet1, has been identified in A. thaliana ecotype C24. Interestingly, the r-chl1 QTL on chromosome 4 co-localised with Vet1. The close proximity of several genes controlling the transition to flowering in this region (fri, cry1/hy, det1, ted1) complicates an interpretation of these results.

In the present study, Erecta has been shown to underlie the stunting resistance QTL r-stre1 on chromosome 2, which explained a large part of the trait variation observed. Receptor-like kinases (RLKs) are often involved in controlling developmental processes or mediating disease resistance reactions [56]. The only Verticillium resistance gene identified so far, Ve1, also belongs to the RLK family [57, 58] and has been shown to recognize fungal effectors [59]. The leucine-rich repeat receptor-like Ser/Thr kinase ERECTA is an example of a signalling molecule controlling both developmental processes and QDR in A. thaliana [49]. erecta mutants show altered organ development resulting in compact growth [32, 60]. Cell proliferation [61] and stomatal patterning [62] were also shown to be controlled by Erecta. Additionally, several resistance traits are reported to be controlled by Erecta in different pathosystems: Growth of the bacterial pathogen Ralstonia solanacearum was inhibited and wilt symptoms were reduced in A. thaliana plants with a functional Erecta gene compared to erecta mutants [63]. Infection with the necrotrophic fungus Plectosphaerella cucumerina resulted in more chlorosis and necrosis in erecta mutants compared to the respective WT accessions [64]. Enhanced susceptibility of erecta mutants, leading to enlarged leaf lesions, has also been reported for infection with the oomycete Pythium irregulare [65]. These results illustrate the close interconnection between ERECTA, development and QDR. Hence, it is not always clear whether altered disease resistance is a consequence of the developmental changes or a direct effect of ERECTA signalling. Studying other signalling components of the ERECTA pathway can shed light on this question. Plants defective in Agb1 are morphologically distinct from erecta mutants [31], but defective in the same signalling pathway. In the present study, agb1-1 mutants were at least as susceptible to V. longisporum-induced stunting as the erecta mutants. This supports the view that a functional ERECTA pathway mitigates the stunting effect of V. longisporum infection independently of its effect on morphology.

Resistance to systemic colonization was not enhanced by a functional Erecta gene in La-0 compared to Ler. This is strong evidence that Erecta, although it is located in vec1, is not involved in mediating resistance to systemic colonization by V. longisporum. A tendency towards stronger colonization of erecta mutants in the moderately susceptible Col-0 background compared to WT was never significant, and is interpreted as an indirect effect of the reduced plant height caused by erecta. V. longisporum may reach the apex of a shorter shoot more easily. This view is corroborated by the fact that the long stalks of the agb1-1 mutant were very poorly colonized, thus behaving very differently from the erecta mutants, despite the fact that both AGB1 and ERECTA act in the same signalling pathway.

Altogether, Erecta has been proven to be a source of natural genetic variation in quantitative resistance not only to V. longisporum. Not much is known so far about the allelic variation of Erecta beyond the level of complete loss of function and its consequences for disease resistance in natural accessions. A comprehensive analysis of Erecta, its structural and regulatory variability and its homologues in cruciferous crop plants would, therefore, be desirable.

ABA and JA contents differed in a genotype-specific way that correlated with the rates of fungal colonization. It is not known whether V. longisporum produces ABA itself, which could explain the high levels in heavily colonized plants. In B. napus, however, V. longisporum infection did not increase ABA levels in xylem sap [23]. The sampling was not fully comparable in both studies, as different parts of the stalk were sampled and the samples in B. napus were taken at an earlier developmental stage. The results in B. napus make it more likely that different ABA levels depended on the host genotypes instead of fungal biomass.

The role of ABA signalling in V. longisporum resistance is complex. The ABA-deficient aba2-1 mutant was highly susceptible to V. longisporum-induced stunting, but other ABA signalling mutants were unaffected, indicating that a specific function of ABA2 was involved in the response observed that did not require ABA in general [16]. From the many functions of ABA in development and disease, different effects can be assumed in the context of Verticillium disease. ABA has been shown to influence plant defence reactions in various ways [66]. In most cases, ABA increased susceptibility to pathogens due to suppression of SA synthesis [67] and/or antagonism with jasmonate-ethylene signalling [68]. However, ABA can also stimulate JA biosynthesis and increase resistance [65]. ABA has important functions during seed development, such as a trigger for the acquisition of storage molecules during cell enlargement [69]. ABA plays a significant role in the induction of senescence [70]. A major effect of V. longisporum on B. napus in the field is the induction of premature ripening, which can be recognized by chlorotic stems and which is leading to reduced seed size. Increased levels of ABA might support the deviation of mobilised nutrients to foster the growth of fungal biomass. Furthermore, it is possible that clogged vessels after V. longisporum-infection induce increased ABA levels as a result of drought stress. However, it has recently been shown that V. longisporum infection can even increase drought tolerance of the host as a consequence of de-novo xylem formation [29]. Studying xylem trans-differentiation in susceptible and resistant genotypes could clarify the relevance of xylem formation for resistance.

JA contents decreased after infection with V. longisporum in Bur and the NIL that contained Bur alleles in the region of vec1. Plant-pathogenic Verticillium species are regarded as hemibiotrophs with a necrotrophic phase during the late stages of infection. Defence responses against necrotrophic pathogens are often induced by JA [71]. Accordingly, JA-deficient tomato plants have been shown to be more susceptible to Verticillium dahliae than WT plants [72]. In the present study, however, the colonization-resistant ecotype Bur was characterized by a decrease in JA levels after infection. Since JA is also involved in A. thaliana leaf senescence [73], V. longisporum possibly benefits from senescence processes induced by JA and ABA, which would suggest a stimulation of hormone production by the fungus. Alterations of senescence-like processes were also postulated to underlie increased resistance of the A. thaliana JA-receptor mutant coi1 against V. longisporum colonization [24]; however, this disease-promoting effect of Coi1 has been shown to be JA-independent. These results confirm that fine-tuning of a resistance reaction by cross-talk of phytohormone signalling pathways is highly individual for each pathosystem [74]. In addition, possible manipulations of the host’s hormone status by the pathogen for its own benefit should be considered.

Conclusions

Phytohormone signalling processes have been demonstrated to be subject to allelic variation and underlie QDR against V. longisporum in natural A. thaliana accessions. ERECTA, SA, ABA, and JA signalling has been shown to mediate an ecotype-specific response of A. thaliana to V. longisporum infection. The ecotype-specific differences for ABA and JA contents were mediated by the same genomic region on chromosome 2 that also controlled systemic colonization by V. longisporum. This region contains the major QTL controlling systemic colonization, vec1, and also Erecta. As colonization resistance was shown to be independent of Erecta, this region should contain other gene(s) that control the degree of fungal colonization. This type of resistance is likely to depend on ABA and JA signalling, as both NILs differed for these hormone contents in the same way as the parental lines that contributed the respective alleles in the polymorphic region. Ongoing studies on differential gene expression depending on vec1 should allow one to draw a more comprehensive picture of the defence reactions leading to colonization resistance. Furthermore, cloning of a gene that controls this reaction should be feasible by combining map-based confinement of the QTL region with expression analysis. It was demonstrated that different resistance traits were controlled independently on a physiological basis, but were still genetically linked. Understanding the biological basis of phenotypic variation in A. thaliana with respect to V. longisporum resistance will provide new approaches for implementing durable resistance in cruciferous crops.