Italian durum wheat cultivar Ofanto is a spring cultivar that is resistant to leaf rust and powdery mildew diseases in Beijing, China. Its pedigree is Appulo/Valnova (De Vita et al., 2007). The susceptible durum wheat line Mo75 was provided by Prof. Xiao Chen (Institute of Crops, China Academy of Agricultural Science). We crossed Ofanto with Mo75 to generate F₁ seeds and its derived F₂ populations of 706 plants for genetic analysis. The pedigree of the elite common wheat cultivar Shi4185 is Zhi8.94/Baofeng7228//Shi84-7120. The common wheat line ND4503 was bred by China Agricultural University with the pedigree as ND3338/F390//Jingnong98-270. All the seeds of Ofanto, Mo75, Shi4185 and ND4503 are kept at China Agricultural University. We hypothesize that the lack of resistance in ND4503/Ofanto F₁ is due to the effect of recessive gene action in Ofanto or suppressor in ND4503. To investigate the genetic basis of the susceptible F₁ plants ND4503/Ofanto (6x/4x), we developed a three-way cross Shi4185/ND4503//Ofanto (6x/6x//4x) to create a mapping population that included 2537 plants (269 for genetic analysis, BSA analysis and primary mapping, 2268 for fine mapping). An F₁ was first obtained by crossing Shi4185 by ND4503, and the resulting F₁ was crossed with Ofanto. The seedling plants of the three-way Shi4185/ND4503//Ofanto population were used for leaf rust testing. The F₁ plants of the three-way cross (AABBDD X AABB) were pentaploid and sterile. Susceptible common wheat line Xuezao was used as a check in all experiments.

Urediniospores of Pt race PHT were originally provided by the Institute of Plant Protection, Chinese Academy of Agricultural Sciences, Beijing and were subsequently propagated on a susceptible genotype. PHT was avirulent on Ofanto and virulent on durum wheat line Mo75, common wheat cultivars/lines Shi4185, ND4503 and Xuezao. The urediniospores were propagated in the greenhouse on the susceptible plants. In all experiments, a susceptible common wheat line Xuezao was used as a check for successful inoculation.

The phenotypes of both parents and Ofanto/Mo75 F₂ and Shi4185/ND4503//Ofanto three-way progenies were evaluated at the seedling stage in the greenhouse. Wheat seeds were planted in 200-hole trays (10 × 20) at a density of one seed per hole in each tray and placed in a greenhouse at 15–20 °C. Approximately 15 days later, the seedlings with first leaves fully unfolded were inoculated; they were sprayed with a 1% aqueous solution of Tween-20® as surfactant followed by dusting with urediniospores and incubation in dark humidity chambers at 15 °C for 24 h before moving to a greenhouse maintained at 15–20 °C. Infection types (ITs) were evaluated 14 days after inoculation using a 0 to 4 scale (0 = hypersensitive flecks, 1 = small uredinia with necrosis, 2 = moderate size pustules with chlorosis, 3 = moderate-large size uredinia without necrosis or chlorosis, and 4 = large uredinia lacking necrosis or chlorosis) (McIntosh et al., 1995). ITs 0–2 were considered resistant, and the ITs 3–4 were considered susceptible. We tested 706 plants of Ofanto/Mo75 F₂ population and 2537 plants of three-way cross Shi4185/ND4503//Ofanto mapping population. We rechecked the results of phenotypes two more times after the first disease evaluation at two-day intervals. A chi-squared analyses was performed on segregation results to confirm the goodness of fit of observed and predicted ratios. The χ² analyses were executed in Microsoft Excel (version 2010) using the Bchitestˆ function to calculate χ² and p-values.

After disease evaluation, leaf tissues of segregating populations and parents were collected and kept at -80 °C. Leaves were ground into powder in liquid nitrogen and DNA was extracted by the CTAB method (Maroof et al., 1994). DNA samples were quantified using a NanoDrop One spectrophotometer instrument (Nanodrop Technologies) and diluted to a working concentration of 30 ng/µl.

Bulked Segregant Analysis (BSA) was performed using the KPS Wheat 90K/660K Chip according to the Affymetrix Axiom 2.0 Assay Manual Workflow protocol provided by Compass Biotech Co. (CBC, Beijing) to identify SNPs associated with leaf rust response (Guan et al., 2019). To make the BSA analysis of the resistance gene in Ofanto by KPS Wheat 660K Chip, genomic DNA extracted from 20 resistant and 20 susceptible plants from the Ofanto/Mo75 F₂ population were selected randomly and bulked in equal amounts to form resistance and susceptibility pools, respectively. For BSA analysis of the suppressor gene in ND4503, we selected 20 resistant and 20 susceptible plants from 269 plants of Shi4185/ND4503//Ofanto three-way population to make resistance and susceptibility pools, respectively. Markers polymorphic between the pools were then tested individually and confirmed across the mapping population.

The screened probes between pools were subjected to BLAST analysis to reveal their physical positions with respect to the CS reference genome sequence (IWGSC RefSeq v1.0). Next, the chromosomal segments enriched by these probes were analyzed. The workflow of genotype detection using the KPS Wheat 660K SNP array was similar to the 90K array described above.

To improve the efficiency of marker development in resistance gene mapping, we resequenced Ofanto and Mo75 by double-end sequencing using the Illumina HiSeq2500 sequencing platform at Novogene Bioinformatics Company Ltd., NBC, Beijing (Li et al., 2020). All high-quality sequence reads were aligned to the durum wheat Svevo reference genome using the Burrows-Wheeler Aligner 0.7.15 program with default parameters (Li and Durbin, 2009; Maccaferri et al., 2019). DNA libraries of Shi4185 and ND4503 were prepared in the same way and aligned with sequence reads of the CS reference genome (IWGSC RefSeq v1.0). The Shi4185 re-sequencing data are available under NCBI Sequence Read Archive accession PRJNA476679 (https://downloads-qcif.bioplatforms.com/bpa/wheat_cultivars/cultivars/). We submitted the re-sequencing data of Ofanto, Mo75, and ND4503 in Library ID SUB12497795 with accessions SAMN32108327, SAMN32108328 and SAMN32108330.

According to previously described methods, the insertion/deletion (InDel) variations between pairs of parents (i.e., Ofanto and Mo75, or Shi4185 and ND4503) in the target interval were chosen to design InDel markers (Chai et al., 2018). Based on IWGSC CS RefSeq v1.0, 24 polymorphic InDel markers were developed from polymorphisms between Shi4185 and ND4503 within the 56–149 Mb region of chromosome 2A. Primers were designed using primer3 v0.4.0 (http://bioinfo.ut.ee/primer3–0.4.0/) and CS IWGSC RefSeq v1.0 was employed to download sequences 200 bp upstream and downstream of the InDels. The 10 µL PCR system comprised 5 µL 2× TaqPCR StarMix, 1 µL of primers, 2 µL of 50–100 ng/µL DNA template, and 2 µL of H₂O. For polymorphism detection, PCR products were separated by 3% agarose gel electrophoresis with TAE buffer and 10% non-denaturing polyacrylamide gel electrophoresis (PAGE) (Marklund et al., 1995).

Kompetitive Allele-Specific PCR (KASP) markers were designed from SNPs (Ramirez-Gonzalez et al., 2015; Fang et al., 2020) using Polymarker (http://www.polymarker.info/). Three KASP markers were developed from SNPs between Shi4185 and ND4503 to narrow down the suppressor locus. The reaction mixtures subjected to the TouchDown program comprised 2 µL of 100–200 ng/µL DNA template, 2 µL of 2× KASP master mixture, and 0.1 µL of primer mixture, totaling a volume of 4.1 µL. Fluorescent signals from the PCR mixtures were detected on a Real–Time Quantitative PCR instrument (Bio-Rad Laboratories Inc., USA). Sequences of the InDel and KASP markers are listed in Table S1 .

Markers polymorphic between resistant and susceptible parents were used to genotype Ofanto/Mo75 F₂ plants and Shi4185/ND4503//Ofanto three-way plants. The leaf rust response data were used for linkage analysis in combination with PCR amplification results. Localization of markers and target gene was based on recombination between marker genotypes and disease phenotype. Genetic distances were calculated in centiMorgans (cM). JoinMap 4.1 was used to construct the initial linkage map. We used regression mapping algorithm (Stam, 1993; Van Ooijen and Voorrips, 2006) and Kosambi mapping function to calculate centiMorgans between markers (Kosambi, 1943). We chose the F₂ population setting in Joinmap 4.1 to analyze the genetic linkage of LrOft and BC₁ population setting for SuLrOft.

The DNA sequences and approximately 500-bp upstream and downstream annotated gene sequences within the mapped interval amplified from Ofanto, Shi4185, and ND4503 using TKS Gflex™ DNA polymerase (TAKARA, Dalian), along with the corresponding primer pairs, are provided in Table S1 . PCR products were separated in 1% agarose gels by electrophoresis and then sequenced at TsingKe Biological Technology Company, Beijing. Sequences were compared and analyzed using DNAMAN version 8.0 (Li et al., 2020). The SMART program (http://smart.embl-heidelberg.de/) was employed to predict the function of the candidate gene.

Ofanto displayed a resistant IT 0; against Pt race PHT, whereas Shi4185, ND4503, and Mo75 were susceptible with IT 4 ( Figure 1A ). F₁ seedlings from crosses Ofanto/Mo75 were resistant indicating that the leaf rust resistance in Ofanto was dominant. In the F₂ population derived from the Ofanto/Mo75 cross, 510 were resistant and 196 susceptible, fitting the ratio of 3:1 (χ² = 2.87, p > 0.05) ( Table 1 ; Figure 1B ). These results indicated that the leaf rust resistance in Ofanto was governed by a single dominant allele, provisionally designated as LrOft.

There were 8,904 SNPs between the resistant and susceptible pools constructed by Ofanto/Mo75 F₂ population; 795 were anchored on chromosome 6B, which was the highest number among the 14 chromosomes ( Figure 2A ). According to the Durum Wheat Svevo RefSeq v1.0, 214 SNPs were enriched in the 550–650 Mb region of chromosome 6BL ( Figure 2B ), suggesting the resistance gene was located on the long arm of chromosome 6B.

To map LrOft more precisely we designed five InDel markers based on sequence variation in the candidate region according to re-sequencing data from the parents. InDel markers 6B6847, 6B6891, 6B6941, 6B50L24, 6B50L25 were confirmed to be polymorphic between the parents and pools and were used to genotype the 211 F₂ plants from cross Ofanto/Mo75. A genetic linkage map of LrOft gene was constructed from the data, in which the LrOft locus was localized to a 2.5 cM genetic interval; 1.9 cM distal to 6B6941 and 0.6 cM proximal to 6B50L24 ( Figure 2C ).

We checked the plants in the three-way population by InDel marker 2AS50L14 and 2AS50L6 to make sure it’s true three-way hybrid. When we evaluated the leaf rust resistance of 269 plants of Shi4185/ND4503//Ofanto three-way population, we found that 132 were resistant and 137 susceptible, fitting the ratio of 1:1 (χ2 = 0.093, p > 0.05) ( Figure 1C ; Table 1 ), suggesting that ND4503 contained a genetic factor for suppression of LrOft.

We identified 2,183 SNPs with heterozygous vs. homozygous variations between the resistant and susceptible pools in the three-way population that were genotyped with the 90K SNP array. Chromosome 2A contained the highest number (393) of SNPs ( Figure 3A ), and 305 of them were clustered in the 50–150 Mb region ( Figure 3B ), suggesting the genetic factor suppressing LrOft was located on chromosome 2A. Since LrOft was on chromosome 6B, non-homologous to chromosome 2A, it was likely that the suppressing factor in ND4503 was a suppressor of LrOft. We named it SuLrOft. Suppression of LrOft was conferred by heterozygous SuLrOft in the cross of Shi4185/ND4503//Ofanto ( Table S2 ).

Two hundred and sixty-nine seedlings of the cross Shi4185/ND4503//Ofanto were genotyped using 19 InDel markers and a genetic map was constructed. The resultant genetic map of SuLrOft spanned 11.2 cM (2AS505-2AS10013) and the SuLrOft was delimited to a 2.6 cM interval flanked by markers 2AS50L14 and 2AS50L6. According to the CS reference genome (IWGSC RefSeq v1.0), this interval corresponded to an approximate 9 Mb physical region (2AS50L14 at 61288040 bp, and 2AS50L6 at 70278664 bp) ( Figure 3C ).

To fine-map the genomic interval surrounding the SuLrOft locus another five InDel markers (InDel29, InDel30, InDel31, InDel34, and InDel37) and three KASP markers (KASPmiss39, KASPmiss83, and KASPstop3) were developed based on re-sequencing data of Shi4185 and ND4503 corresponding to the 9 Mb interval of the CS IWGSC RefSeq v1.0. After confirming polymorphisms between the parents, these eight new markers and the two closest flanking markers 2AS50L14 and 2AS50L6 were used to genotype an additional 2,268 F₁ plants of cross Shi4185/ND4503//Ofanto; 67 recombinants between markers 2AS50L14 and 2AS50L6 were identified and 14 different recombinant genotypes were detected ( Table S3 ). The allelic state for LrOft is heterozygous in each plant of the three-way population, including the 14 recombinants. Based on the genotypic and phenotypic data for the recombinants, the SuLrOft locus was delimited to the interval InDel30–InDel31 ( Figure 4A ), corresponding to approximately 68.2 Kb in CS IWGSC RefSeq v1.1 (https://urgi.versailles.inra.fr/blast_iwgsc/). This interval contained three high-confidence and two low-confidence genes ( Figure 4B ).

The re-sequencing data for Shi4185 and ND4503 revealed no variation in the two low-confidence genes (TraesCS2A02G114800LC, TraesCS2A01G114900LC). Among the three high-confidence genes, missense variations were found in the CDS of TraesCS2A01G110800 and TraesCS2A01G110900, and there were two SNPs in 3’ UTR of TraesCS2A01G111000. TraesCS2A01G110800 was annotated as a putative disease resistance RPP13-like protein and TraesCS2A01G110900 was annotated as a putative disease resistance RGA4 protein. When we used primers Cx10900 to amplify and sequence the SNP-containing intron segment (2A-62361245) in TraesCS2A01G110900 in the eight recombinants between InDel30 and InDel31, we found seven recombinants ( Table S4 ) hence indicating that TraesCS2A01G110900 is not the SuLrOft candidate. We cloned and sequenced TraesCS2A01G110800 with primers Clone110800 in the eight recombinants ( Figure S1 , Table S1 ) and the results showed that SuLrOft co-segregated with TraesCS2A01G110800, suggesting TraesCS2A01G110800 as the most likely candidate gene ( Table S5 ).

The SMART program (http://smart.embl-heidelberg.de/) was employed to predict the function of TraesCS2A01G110800, which encoded NB-ARC, RPT1 and RPT2 domains in the CDS region ( Figure 4C ). There were many sequence variations between Shi4185 and ND4503in TraesCS2A01G110800, most occurring in the NB-ARC domain, resulting in variations in amino acid sequences, with only one SNP in the second intron ( Figure 4C ).