Seeds of the 15 Arabidopsis thaliana accessions were obtained from Arabidopsis Biological Resource Center (ABRC) (Supplementary Table S1). Seeds were surface sterilized, cold stratified for 3 days and were grown on MS medium supplemented with 1.5% agar. Seedlings were then transferred to six-inch pots containing Soilrite mix (Soilrite: Perlite: Vermiculite mixed in the ratio of 4:1:1) pre-soaked with nutrient solution. The plants were grown at 22°C and 60% humidity, under 10-h/14-h light/dark cycle for 5 weeks before challenging them with the pathogen.

The isolate of Alternaria brassicae (J3) utilized in the current study was collected from the experimental field station in the village of Jaunty (Delhi), purified, and maintained as single spore cultures for further use. For in vitro culture and maintenance of A. brassicae isolates, radish root sucrose agar (RRSA) medium was used (Thakur and Kolte, 1985). Fungal cultures were subcultured every fortnight and maintained at 22°C under 12-h/12-h light/dark cycle. To maintain the virulence of the strain, the pathogen was periodically given a passage through B. juncea var. Varuna (its natural host) and the reisolated fungus used for pathogenicity assays.

Spores from 15-day old cultures were used for infections. Cultures were flooded with distilled water to collect spores, and the resulting spore suspension was filtered through two layers of muslin cloth. The filtered spore suspension was adjusted to a spore concentration of 3–5 × 10³ spores mL^-1 before inoculation. Inoculations were done on 5-week old Arabidopsis plants by drop inoculation method (Xu and Ko, 1998). A total of 8–10 similarly aged leaves per plant were marked for infection. Four droplets of 5 μL spore suspension were applied onto the adaxial surface of the true leaves. Inoculated plants were kept at 100% humidity at 20 ± 1°C under a 10-h/14-h light/dark cycle. Infected plants were evaluated for symptoms 7 days post infection (dpi).

The disease was discerned by the formation of necrotic lesions that were encircled by chlorotic halos. Different evaluation criteria were used for the RIL and F₂ populations. The RIL population was evaluated using a DI score where the number of necrotic spots developed was taken into account (McKinney, 1923). Since the F₂ populations were scored on the basis of the disease reaction from single F₂ individuals, the plants were scored not only for the number of necrotic spots developed but also the size of the spot. The cumulative score was thus used to model the disease spread more finely. Additionally, the cumulative DI (CDI) data was normalized to the susceptible parent.

The RIL population was quantitatively scored on the basis of DI calculated from the number of necrotic spots using the formula:

where,

DI = Disease Index

K = Total no. of leaves (usually 8)

ns = No. of necrotic spots/leaf

ns_max = Maximum no. of necrotic spots (usually 4).

The F₂ populations were scored on the basis of CDI calculated from the number of necrotic spots and the size of each spot using the formula:

where,

CDI = Cumulative Disease Index

K = total no. of leaves

ns = total no. of necrotic spots/leaf

ss = size of the necrotic spot

ns_max = maximum no. of necrotic spots (usually 4)

ss_max = maximum size of the necrotic spot.

DNA isolation from Arabidopsis leaves was carried out using a rapid extraction protocol (Edwards et al., 1991). Briefly, leaves (<100 mg) were crushed in 500 μL of Extraction buffer [200 mM Tris-HCl, 250 mM NaCl, 25 mM EDTA (pH 8.0), 0.5% SDS] and centrifuged at 13,000 rpm for 5 min to remove the debris. An equal amount of isopropanol was added to the supernatant to precipitate the DNA and centrifuged again at 13,000 rpm for 10 min. The pellet was washed with 70% ethanol, air-dried, and resuspended in 50 μL of 2 mM Tris (pH 8.0). Genomic DNA extraction for SNP genotyping was carried out using MDI (Advanced Microdevices, Pvt. Ltd, India) plant genomic DNA miniprep kit according to the manufacturer’s instructions.

The Cvi-0 (♀) × Gre-0 (♂) RIL (CvG) population, consisting of 125 individuals, was generated by single-seed descent method from a single F₁ hybrid. The population was advanced up to F₇ generation before use.

Two F₂ populations were generated using CIBC-5 and Ei-2 as the female parent and Zdr-1 as the common male parent. Mature single siliques containing F₁ seeds were harvested separately, and seeds from a single silique per cross were grown. DNA from each plant was isolated, and heterozygosity of the F₁ plants was assessed using five polymorphic markers from across the genome. The F₁ plants were allowed to self-pollinate, and the F₂ seeds were collected and used for phenotyping. A total of 220 F₂ individuals from the CIBC-5 × Zdr-1 (CZ) cross and 180 individuals from the Ei-2 × Zdr-1 (EZ) cross were phenotyped.

A linkage map for the CvG population was constructed using the genotypic data of 125 RILs for 202 markers (75 SSR, 86 AFLP, and 41 IP markers) using the JoinMap program (Version 4.1) (Supplementary Figure S1; Supplementary Table S2). SNP markers were used for constructing the linkage maps in the CZ and EZ populations. Polymorphic SNPs between the two parents were selected using the Multiple SNP Query Tool (MSQT) (Warthmann et al., 2007). SNP genotyping was carried out using KBioscience KASPar assay. The primer design, assay development, and genotyping were undertaken by KBiosciences Pvt. Ltd (Hoddesdon, UK). A total of 152 SNP markers (evenly distributed across the genome) were used for genotyping each of the F₂ populations and linkage maps for both the populations were constructed using the JoinMap program (Version 4.1) (Supplementary Figures S2 and S3; Supplementary Tables S3 and S4).

The DI and normalized CDI (NCDI) were used as the trait values for QTL mapping in the RIL and F₂ populations, respectively. QTLs were detected with QTL Cartographer 2.5 by composite interval mapping (Wang et al., 2012). Genome scan for QTLs was carried out at a walk speed of 2 cM. Statistical significance of QTL for each trait was assessed by performing 1000 permutation tests with a significance level of 0.05 (Churchill and Doerge, 1994; Doerge and Churchill, 1996). A minimum LOD value of 2.5 for the RIL and 3.1 for the F₂ populations was used for the identification of putative QTLs based on the permutations tests. The additive effects and contribution of each QTL to phenotypic variance (R²) were estimated from the composite interval mapping results. QTLs were named as Resistance to Alternaria brassicae [population name] [chromosome number-1].

Phenotypic variation in resistance to A. brassicae was tested in 15 accessions of Arabidopsis belonging to different geographical origins (Supplementary Table S1). The accessions showed considerable variation in their response to A. brassicae (Figure 1A). The accessions displayed variation in the frequency of lesion formation (presence/absence of necrotic lesions) as well as the size of the necrotic lesions formed in response to pathogen challenge. The resistant accessions displayed pinhead-sized necrotic lesions at very low frequencies whereas the highly susceptible accessions had lesions expanding beyond the droplet size at almost all the inoculated spots. The expanding lesions in the highly susceptible accessions occasionally coalesced to form a single large lesion. Typical chlorotic halos encircling the necrotic lesions were also observed in the susceptible accessions. The normalized DI (NDI) derived from the incidence of lesion formation, varied from 0.03 (CIBC-5) to 1.07 (Zdr-1). The variation in NDI was largely genetically determined as indicated by the broad sense heritability score (H² = 0.90). Three different mapping populations were developed from the highly resistant and susceptible accessions to dissect the genetic basis of resistance to A. brassicae.

The contrasting disease reactions of the parental accessions are given in Table 1 (Figure 1B). The phenotype of the F₁ progeny from each of the three crosses was intermediate to that of the respective parental accessions, however, they were closer to the resistant parent than the susceptible one (Table 1). The DI in the CvG population ranged from 0 to 0.671 of which 17 lines showed no infection. The frequency distribution based on the DI showed nearly a bimodal distribution (Figure 2). The CDI in the CZ population varied continuously and was normally distributed with a positive skew while the CDI in the EZ population had nearly a bimodal distribution (Figure 2). Few transgressive segregants were observed in all the mapping populations. Heritability in the three populations showed that a significant proportion of the phenotypic variation (∼50%) could be explained by genetic factors (Table 1).

In the CvG population, four QTLs were identified. The major QTL RtAbe_CvG2-1 was located on the long arm of chromosome 2 at around 35.5 cM in the CvG genetic linkage map (Table 2). The confidence interval based on a 2 LOD score drop from the peak placed RtAbe_CvG2-1 between 27.9 and 48.1 cM covering 12 markers. The major QTL explained up to 42% of the phenotypic variance. Other minor QTLs cumulatively explained up to 19% of the variance. The Gre alleles of RtAbe_CvG2-1 and RtAbe_CvG1-1 increased the DI while those at RtAbe_CvG4-1 and RtAbe_CvG3-1 decreased the DI.

In the CZ population, three QTLs were detected. The major QTL RtAbe_CZ5-1 located on the long arm of chromosome 5 was positioned around 57.9 cM in CZ linkage map (Table 2). The RtAbe_CZ5-1 QTL spanned four markers between 53.4 and 63.3 cM. RtAbe_CZ5-1 explained approximately 11% of the phenotypic variance while RtAbe_CZ3-1 and RtAbe_CZ2-1 explained approximately 8% and 6% of the phenotypic variance respectively. The Zdr-1 allele at RtAbe_CZ2-1 and RtAbe_CZ5-1 increased the CDI whereas the Zdr-1 allele at RtAb_CZ3-1 decreased it. The presence of QTLs with opposing additive effects (RtAb_CZ3-1 vs. RtAbe_CZ2-1 and RtAbe_CZ5-1) may thus be responsible for the transgressive segregants observed in the population.

In the EZ population, a single major QTL was detected. The QTL RtAbe_EZ2-1 located on chromosome 2 with the maximum LOD score positioned around 37.6 cM, covered four markers between 36.6 and 39.7 cM (Table 2). The QTL explained nearly 57% of the phenotypic variance. The Zdr-1 allele at this QTL increased the CDI. This locus also displayed a moderate dominance effect (D = 0.1140), wherein the Ei-2 allele was partially dominant over the Zdr-1 allele.

A total of eight QTLs were identified in the three populations out of which three were major QTLs (Table 2; Figure 3). Since, the physical positions of most of the markers (except AFLP markers) are known, comparisons can be made across the populations even though different set of markers have been used for different populations. The QTLs identified on chromosome 2 in the three populations had overlapping physical intervals and thus could be said to be a common QTL though it had varying effect sizes in different populations. Thus, six QTLs were identified in the study, of which one is a common QTL and other five are population-specific.

Since the markers used in the construction of the genetic maps correspond to physical positions in the Arabidopsis genome, all the annotated genes in the QTL intervals could be easily identified. The RtAbe_CvG2-1 QTL from the CvG population was flanked by the markers in the genes At2g18950 and At2g34100. This 6.2 Mb interval contained 1609 protein-coding genes. The RtAbe_CZ5-1 QTL from the CZ population was flanked by markers in the genes At5g42970 and At5g48400. A total of 601 protein-coding genes were found in this 1.8 Mb interval. The markers in the genes At2g19900 and At2g28950 flank the RtAbe_EZ2-1 QTL from the EZ population, and this 1.6 Mb region contained 958 protein-coding genes. Also, the minor QTL RtAbe_CZ2-1 from the CZ population approximately lies in the same physical interval flanked by the genes At2g24090 and At2g26880.

Subsequently, these QTL intervals were examined for genes with known functions in disease resistance. A GO term enrichment analysis of the genes in each QTL interval was carried out. A list of genes with the GO term – response to stress and other related GO terms was compiled. The RtAbe_CvG2-1 QTL consisted of 55 probable candidate genes that were annotated with known functions in disease resistance pathways. The RtAbe_EZ2-1 QTL consisted of 33 such genes, which overlapped with the genes in the RtAbe_CvG2-1 QTL. Some of the possible candidate genes/gene families with a role in host defense in this region include WRKYs, DEFLs (Defensin-like), EIN3, Glycosyl hydrolases, PR (pathogenesis-related) genes, lipid acyl hydrolase and ERFs (ethylene response factors).

The RtAbe_CZ5-1 QTL on chromosome 5 was significantly enriched for GO terms such as defense response and response to stress (Supplementary Table S5). A total of 27 disease resistance genes were located in this region out of which 24 were TIR-NBS-LRR class, and three were CC-NBS-LRR class. Other possible candidate genes in this region that are involved in defense signaling pathways include BIR1 (BAK1-interacting receptor-like kinase), MYC3 (JAZ-interacting transcription factor), ERFs (Ethylene response factors), BRG (Botrytis-induced Related Gene), RBOHD (Respiratory Burst Oxidative Homolog D) and NPR3 (Non-expressor of PR3). Additionally, this interval contained nearly 14 defensin and putative defensin-like genes that are known to have antimicrobial /antifungal activity against many pathogenic fungi (Figure 4).