Identiﬁcation and Characterization of Resistance Loci to Wheat Leaf Rust and Stripe Rust in Afghan Landrace “KU3067”

Leaf rust and stripe rust are important wheat diseases worldwide causing signiﬁcant losses where susceptible varieties are grown. Resistant cultivars 12.9–20.5 and 5.4–12.5%, and QLr.cim-6BL are likely to be new. These QTL and their closely linked markers will be useful for ﬁne mapping and marker-assisted selection (MAS) in breeding for durable resistance to multiple rust diseases.


INTRODUCTION
Wheat leaf rust and stripe rust caused by obligate biotrophic fungus Puccinia triticina (Pt) and P. striiformis f. sp. tritici (Pst), respectively, are the most important foliar diseases of wheat worldwide. Leaf rust is the most commonly occurring disease and can cause yield losses up to 40% under favorable conditions (Knott, 1989). Stripe rust occurs in cool temperate regions and can cause yield losses ranging from 10 to 70% and up to 100% in highly susceptible cultivars (Chen, 2005). Although fungicides can effectively control wheat rusts, growing resistant cultivars is a more efficient, economic, environment-friendly, and long-term strategy to minimize losses.
Resistance to wheat rusts can be broadly classified either as race-specific or as race non-specific resistances (Johnson, 1988). Race-specific resistance (or) seedling resistance (or) allstage resistance is often characterized by a strong to moderate immune response usually associated with the hypersensitive response that fully curtails fungal infection and sporulation at all developmental stages if the pathogen possesses a corresponding avirulence gene (Flor, 1942). These resistance genes are usually effective against a single race or a few races of the pathogen. This kind of resistance may lose effectiveness when new virulent pathotypes arise through mechanisms of mutation or recombination. Race non-specific resistance on the other hand is under polygenic control and usually expressed susceptible response at the seedling stage and expressed quantitatively at post-seedling growth stages either as slow rusting or partial resistance, or adult plant resistance (APR). It is usually characterized by lower frequencies of infection, longer latency period, smaller uredinium, and less urediniospore production (Caldwell, 1968). The phenotypic effects of such genes are usually minor; however, several of such genes with additive effects can be combined leading to near levels of immunity . These genes usually are race non-specific and are effective against multiple races of the pathogen and compared with race-specific resistance, therefore, conditioning broader effectiveness and enhanced durability.
To date, 80 leaf rust resistance genes and 83 stripe rust resistance genes are officially cataloged in wheat (Li et al., 2020;Kumar et al., 2021) and most of the identified genes showed race-specific resistance. Only a few genes, for example, Lr34/Yr18/Pm38/Sr57 , Lr46/Yr29/Pm39/Sr58 , Lr67/Yr46/Pm46/Sr55 (Herrera-Foessel et al., 2011), andLr68 (Herrera-Foessel et al., 2012) are also known to confer pleiotropic effect on the resistance. In addition to the formally named genes, 249 leaf rust and 327 stripe rust resistance QTL have been reported on every chromosome (Wang and Chen, 2017;Pinto da Silva et al., 2018). Even though multiple genes have been characterized and cataloged only a handful of genes, conferring adequate levels of resistance to prevalent rust races have made an impact in rust resistance breeding as the majority of them are effective to specific races (or) do not provide adequate levels of resistance (or) are associated with negative linkage drag with yield and other traits (Bhavani et al., 2019). The advent of molecular markers and the availability of reference maps and completely annotated wheat genome has greatly facilitated rapid gene discovery and characterization (Clavijo et al., 2017;Rosewarne et al., 2013).
Molecular markers have been widely used for mapping APR genes for rust resistance through QTL analysis. Several highthroughput genotyping technologies, such as single nucleotide polymorphism (SNP) and genotyping-by-sequencing (GBS) have greatly facilitated the identification and characterization of genomic regions of complex traits such as rust resistance (Bhavani et al., 2021). Diversity arrays technology sequencing (DArT-Seq) developed by the Diversity Arrays Technology Pty Ltd (Canberra, Australia) is a new approach based on traditional DArT complexity reduction methods and Next Generation Sequencing (NGS) techniques (http://www.diversityarrays.com/ dart-application-dartseq).
Diversity arrays technology sequencing offers affordable genome profiling through the generation of high-density SNPs as well as PAV (presence/absence variation) markers. This technology can be better used in linkage map construction and accelerates high-resolution mapping and detailed genetic dissection of traits (Raman et al., 2014) and has been extensively used to study the diversity of wheat accessions in the CIMMYT gene bank (Sansaloni et al., 2020).
The wheat line "KU3067, " a landrace from Afghanistan, was stored in the National Bio-Resource Project of Japan in 1956 (Tanaka et al., 2008). It displays high levels of resistance to leaf rust and stripe rust in the Mexico's field conditions. This study aimed to determine the genetic basis of leaf rust, stripe rust resistance in KU3067, and identify molecular markers linked to these QTL that can be used in breeding.

Plant Materials and Pathotypes
The mapping population comprised of 148 BC1F 5 lines (backcross-inbred lines, BILs) was derived from the cross of KU3067 and Apav (Apav was used as the recurrent susceptible parent). KU3067 showed a high level of APR to both rusts in field trials, whereas Apav derived from the "Avocet-YrA/Pavon 76" mapping population was completely susceptible to the three rusts. Predominant Mexican Pst pathotype MEX14.191 and two Pt pathotypes (MBJ/SP and MCJ/SP) were used to test the BILs in the greenhouse and field. These Pt and Pst pathotypes were available at the greenhouse at the International Maize and Wheat Improvement Center (CIMMYT).

Field Experiments
Leaf rust field evaluations were carried out in CIMMYT ( . Field plots consisted of 0.7-m paired rows with approximately 60 plants on each line. Avocet nearisolines (NILs) carrying Yr24 and Yr26 were used as spreaders in leaf rust studies, whereas a mixture of Morocco, Avocet NIL carrying Yr31, and six lines possessing the Yr27 gene, derived from the cross Avocet/Attila, were used as stripe rust spreaders. The spreader cultivars were planted as hills in the middle of a 0.3m pathway on one side of each plot and planted perpendicular and adjacent to the test rows. Artificial inoculations on spreader rows and hills were carried out twice at weekly intervals by spraying aqueous suspensions of urediniospores of equal mixtures of Pt pathotypes MBJ/SP and MCJ/SP for leaf rust and Mex14.191 for stripe rust which were suspended in Soltrol 170 at a concentration of 2-3 mg/ml and dispensed onto the spreader rows at the tillering stage (Feekes growth stage 5; Large, 1954). The avirulence/virulence formulas of MBJ/SP and MCJ/SP were described in Herrera-Foessel et al. (2012) and of race Mex14.191 in Zhang et al. (2019). The host response to infection in adult plants was determined according to Roelfs et al. (1992). Disease severities were scored following the 0-100% visual ratings two or three times at weekly intervals according to the Modified Cobb Scale (Peterson et al., 1948). The area under the disease progress curve (AUDPC) was calculated using the method suggested by Bjarko and Line (1988).
where X i is the disease severity on assessment date i, T i is the number of days after inoculation on assessment date i, and n is the number of disease assessments. Maximum disease severity (MDS, %) data and AUDPC were used for QTL analysis.

Genetic and Statistical Analyses
Correlation analysis between MDS in different environments was conducted using bivariate two-tailed Pearson's correlation coefficients by IBM SPSS Statistics 21.0 (IBM Corp., Armonk, NY). The number of genes were estimated using chi-squared tests using the expected and observed segregation ratios. Based on disease severity and infection response, BILs were broadly classified into three phenotypic categorieshomozygous parental type resistant (HPTR), homozygous parental type susceptible (HPTS), and intermediate types (others)-following Singh and Rajaram (1992). Chi-squared tests were carried out to test the best fit for different gene segregation ratios.

Linkage Map Construction and QTL Detection
Genomic DNA of the parents and BILs were isolated from non-infected tissues by the CTAB method (Sharp et al., 1988). DNA concentration was measured using Thermo Scientific NanoDrop 8000. The 148 BILs and parents were genotyped with DarT-Seq in CIMMYT's Biotech Laboratory. The linkage map was constructed using the MAP function in IciMapping 4.1 (http://www.isbreeding.net/software/?type=detail&id=18, Li et al., 2007). QTL analysis was conducted with the ICIM-ADD function in BIP using the software QTL IciMapping 4.1 through 1,000 permutations at P = 0.01 (Li et al., 2007).
Stepwise regression analysis was used to detect the percentages of phenotypic variance explained (PVE, R 2 ) by individual QTL and additive effects at the LOD peaks. The linkage map was drawn using MapChart 2.3 (http://www.earthatlas. mapchart.com/, Voorrips, 2002). The sequences of all the markers were subjected to the BLAST against the Chinese Spring reference sequence (version 2.0 https://urgi.versailles.inra. fr/blast_iwgsc/blast.php, IWGSC, 2018) in order to determine physical positions.
Phenotypic distributions of stripe rust and leaf rust MDS were also compared between two groups of BILs that were classified based on the presence or absence of a QTL. Finally, all the BILs were grouped into different QTL genotypic classes by the flanking markers to check the additive effect, and disease severity was calculated by averaging rust scores within a QTL group across environments.

Seedling Response to Leaf Rust and Stripe Rust
In the seedling test, KU3067 and Apav displayed susceptible IT (3+ and 4) against both Pt races MBJ/SP and MCJ/SP based on a 0-4 scale. For the stripe rust, KU3067 expressed resistant IT 3 when tested with Pst Mex14.191, whereas Apav produced susceptible IT 7/8 based on the 0-9 scale. When tested on 148 BIL lines, 42 lines were found resistant, 101 lines were susceptible, and 6 lines showed segregation. Chi-squared analysis conformed to the expected frequency for a single stripe rust resistance gene in the seedling test against Pst Mex14.191 ( Table 1). The resistance gene was temporarily designated as YrKU.

Characterization of Leaf Rust and Stripe Rust Resistance in the Field
In the field trials, the MDS and IT reaction to leaf rust was 1MSS for KU3067 and 100S for Apav across all the seasons. Mean leaf rust severities on BILs ranged from 50.1 to 79.0% during all the leaf rust trials ( Table 2). The frequency distribution of BILs for leaf rust severity was continuous over the four environments ( Figure 1A), indicating the polygenic inheritance of APR to leaf rust in the population. Genetic analyses by Mendelian segregation analysis (Table 1) indicated the presence of 3-4 APR genes that confer resistance to the leaf rust in the population.
KU3067 and Apav displayed MDS and IT for stripe rust of 5R and 100S under field conditions. The mean MDS for all the BILs was 67.8% and 67.0% in YR2017 and YR2018, respectively. The stripe rust MDS scores for the 148 BILs across all environments also showed continuous distributions ( Figure 1B). Four genes were estimated to provide resistance to stripe rust using the Mendelian segregation analysis ( Table 1).
MDS scores for leaf rust across all the crop seasons were significantly correlated with correlation coefficients ranging from 0.73 to 0.88 (P < 0.001; Table 3). For stripe rust, the correlation coefficient for YR2017 and YR2018 was 0.84 (P < 0.001). The coefficients of the correlation between stripe rust and leaf rust disease scores ranged from 0.61 to 0.76 across experiments, indicating the presence of pleiotropic genes conferring resistance to both the rusts.

Linkage Map Construction and Mapping the Seedling Stripe Rust Resistance Gene YrKU
A total of 4,053 markers were used to construct the linkage map. The resulting linkage map comprised 40 linkage     Figure 2). All the resistance alleles were from KU3067. QLr.cim-1AS was detected in LR16B, LR16B-A (AUDPC for leaf rust at El Batán in 2016) and LR17B and explained 9.4-12.0% of the variation.  (Table 4).

Possible Pleiotropic Rust Resistance QTL
Among the QTL detected, two showed possible pleiotropic resistance for both rusts. The first one, QLr.cim-4DL/QYr.cim-4DL, proved to be Lr67/Yr46 based on the closely linked marker Tm4 (Moore et al., 2015) the marker was validated on the parent Ku3067, thus, confirming the presence of the gene Lr67.
Another QTL, QLr.cim-7BL/QYr.cim-7BL located on 7BL also showed pleiotropic resistance. QYr.cim-7BL overlapped the Yr seedling resistance gene YrKU and showed all-stage resistance to stripe rust. Overall, a total of six QTL for leaf rust and four QTL for stripe rust were identified in the population (Table 4; Figure 2). The total phenotypic variance explained by detected QTLs ranged from 55.1 to 83.9% across the environments for leaf rust and 44.4 to 71.6% for stripe rust, confirming their significant effect in reducing rust severity.

Average Effects of Two Potentially Pleiotropic Rusts QTL and Additive Effects of Leaf Rust and Stripe Rust QTL
Average effects of two potentially pleiotropic QTL were estimated in RIL carrying QTL and RIL lacking QTL were compared based on the closely linked flanking markers (Figure 3; Table 5). Lr67/Yr46 showed significantly large effect in reducing leaf rust and stripe rust by 37.9-66.9 and 35.1-48.6%, respectively, in four leaf rust trials and two stripe rust trials ( Table 5). Leaf rust severities in the presence of Lr67 ranged from 4.3-60.1% compared with 21.2-100% in its absence. For QLr.cim-7BL/QYr.cim-7BL, there was a significant reduction in mean leaf rust and stripe rust severity by 20.0-35.0 and 29.8-40.9%, respectively. The mean leaf rust and stripe rust severities of RIL carrying the 7BL QTL, ranged from 3.4-90.0 and 5.3-65.0% compared with 7.5-100 and 17.5-100% in its absence, respectively (Figure 3).
The BILs were divided into 24 and 15 genotypes based on the presence of six-leaf rust and four stripe rust resistance QTL, and the additive nature of identified QTLs is shown in Figure 4. More QTLs imparted higher rust resistance in the BILs. For leaf rust, one line contained five QTL (4D + 7B + 1A + 2A + 6B) with mean MDS 6.3%. The mean MDS of the -QTL group reached 83.3%. For stripe rust, four lines with all four YR QTL had a mean MDS of 7.3%. In total, 46 RIL's without any of the reported stripe rust QTL had an MDS of 86.6%.

DISCUSSION
Genetic analyses indicated that four genes/loci controlled the APR resistance to leaf and stripe rusts in KU3067. In this mapping study, six QTL for leaf rust resistance and four for stripe rust resistance were identified using ICIM. The result of stripe rust was consistent with the estimated gene number, while there is a slight discrepancy with the genetic analyses for leaf rust. The Mendelian genetic approach for estimating gene numbers is based on the theoretical assumptions of polygenetic quantitative inheritance, where each gene is considered to contribute equally to the phenotype. However, it is common that variable effects of different QTL on phenotypes occur in the wheat genotypes, which results in segregating distortion of various phenotypic classes. Therefore, it is often observed that the estimated number of genes varies from the number of QTL identified. Yang et al. (2013) and Lan et al. (2015) also reported the discrepancy on the population Sujata/Avocet-YrA and Chapio/Avocet RIL populations, respectively. The estimated gene number usually represents the minimum number of polygenic loci segregating in a population, as is also evident in this study.

Lr67/Yr46 on 4DL
In this study, the most consistent QTL across all the environments was Lr67/Yr46 on 4DL. Lr67/Yr46 with a large effect explained 10.9-56.4 and 12.1-29.8% of the phenotypic variation for leaf and stripe rusts, respectively. The gene also confers resistance to stem rust and powdery mildew (Herrera-Foessel et al., 2014). Lr67/Yr46 was originally transferred from PI250413, a Pakistani wheat accession, to Thatcher-derived line RL6077 (Dyck and Samborski, 1979). Earlier studies also confirm a larger phenotypic effect of Lr67/Yr46 in Indian wheat lines Sujata  and New Pusa 876 (Ponce-Molina et al., 2018). This study is the first to report the presence of Lr67/Yr46 in an Afghan landrace, which suggests a broad range of deployment of this gene in diverse wheat germplasm. This gene has been effective for over 60 years and remains a key gene in the development of lines with durable rust resistance in many breeding programs, including CIMMYT's program.

QLr.cim-7BL/QYr.cim-7BL
QLr.cim-7BL/QYr.cim-7BL overlapped the Yr seedling resistance gene YrKU, indicating that YrKU showed all-stage resistance in FIGURE 3 | Comparison of KU3067/Apav BILs for mean MDS of LR and YR in the presence or absence of resistance allele for Lr67/Yr46 and QLr.cim-7BL/QYr.cim-7BL in field trials. Different letters within each column following the mean indicate significant differences based on the T-test (P < 0.01).
the study. To date, six known Yr resistance genes Yr39 (Lin and Chen, 2007), Yr52 (Ren et al., 2012), Yr59 (Zhou et al., 2014), Yr67 (Li et al., 2009), YrZH84 (Li et al., 2006, and YrSuj  as well as three leaf rust resistance genes Lr14a, Lr14b (McIntosh et al., 1995), and Lr68 (   with the pleiotropic effect on APR to leaf rust is similar to our findings. Lan et al. (2015) also reported that YrSuj and Yr67 from Indian variety C591 might be same . The relationship among YrKU, Yr67, and YrSuj will be confirmed through the allelism test in the next step to confirm if it is a known gene or a new gene at that locus. For leaf rust, Lr14a and Lr14b, located on 7BL, are seedling resistance genes and seedling tests of the RIL population displayed infection type "4" or "3+, " indicating the two genes are ineffective against the Pt races that were tested, and hence, they are different from QLr.cim-7BL. Lr68 is a slow-rusting resistance gene flanked by markers cs7BLNLRR and Xgwm146 was mapped in CIMMYT line Arula (Herrera-Foessel et al., 2012). QLr.cim-7BL (729.4-753.8 Mb) mapped at a similar position with Lr68 (752.2-752.8 Mb); but the absence of Lr68 in KU3067 was confirmed by the linked molecular markers csGS and CAPS marker cs7BLNLRR . Hence QLr.cim-7BL may not be Lr68.

QLr.cim-1AS
The known seedling resistance gene Lr10 (Feuillet et al., 2003) and two QTL, QLr.cim-1AS from the Indian bread wheat cultivar Sujata  and QLr.cau-1AS from the American wheat cultivar Luke (Du et al., 2015), were mapped on 1AS. Lr10 expressed IT 3 and 4 to Pt pathotypes MBJ/SP and MCJ/SP in the seedling test indicating it is ineffective against the tested Pt. QLr.cim-1AS  and QLr.cau-1AS (Du et al., 2015) were mapped at the distal end of 1AS with physical positions of 1.7-8.5 and 9.0 Mb (Supplementary Table 2). The physical position of KU3067 QTL QLr.cim-1AS (1.1-46.1 Mb) overlapped the latter two QTL (Supplementary Table 2). QLr.cim-1AS from Sujata showed pleiotropic resistance to stripe rust; however, we did not detect any effect of QLr.cim-1AS on stripe rust in this study. The relationship of these three QTL needs further analysis in the next step.

Potential Application of QTL for Leaf Rust and Stripe Rust in Wheat Breeding
In this study, six QTL for leaf rust and four QTL for stripe rust were identified in the KU3067 RIL population. The leaf rust and stripe rust resistance of KU3067 had remained effective for more than 60 years. Despite their inferior agronomic traits (e.g., low grain yield, lodging), KU3067 and lines with multiple QTLs and combinations in the population can serve as a valuable source of resistance for common wheat improvement. The tightly linked SNP markers identified in this study can be converted to KASP assays (Semagn et al., 2014) for marker-assisted selection (MAS) and pyramiding of APR genes to improve leaf rust and stripe rust resistance in wheat-breeding programs.

DATA AVAILABILITY STATEMENT
The data used to present the findings reported in this study is available upon request through the corresponding author.

AUTHOR CONTRIBUTIONS
PZ, SB, and RPS contributed to the manuscript preparation, phenotyping, and QTL analysis. CL and JH-E contributed to population development. ZL contributed to experimental design and analysis. EL contributed to manuscript preparation and critical review. All authors contributed to the article and approved the submitted version.