Identification and Validation of a Common Stem Rust Resistance Locus in Two Bi-parental Populations

Races belonging to Ug99 lineage of stem rust fungus Puccinia graminis f. sp. tritici (Pgt) continue to pose a threat to wheat (Triticum aestivum L.) production in various African countries. Growing resistant varieties is the most economical and environmentally friendly control measure. Recombinant inbred line (RIL) populations from the crosses of susceptible parent ‘Cacuke’ with the resistant parents ‘Huhwa’ and ‘Yaye’ were phenotyped for resistance at the seedling stage to Pgt race TTKSK (Ug99) and in adult plants in field trials at Njoro, Kenya for two seasons in 2016. Using the Affymetrix Axiom breeders SNP array, two stem rust resistance genes, temporarily designated as SrH and SrY, were identified and mapped on chromosome arm 2BL through selective genotyping and bulked segregant analysis (BSA), respectively. Kompetitive allele specific polymorphism (KASP) markers and simple sequence repeat (SSR) markers were used to saturate chromosome arm 2BL in both RIL populations. SrH mapped between markers cim109 and cim114 at a distance of 0.9 cM proximal, and cim117 at 2.9 cM distal. SrY was flanked by markers cim109 and cim116 at 0.8 cM proximal, and IWB45932 at 1.9 cM distal. Two Ug99-effective stem rust resistance genes derived from bread wheat, Sr9h and Sr28, have been reported on chromosome arm 2BL. Infection types and map position in Huhwa and Yaye indicated that Sr28 was absent in both the parents. However, susceptible reactions produced by resistant lines from both populations against Sr9h-virulent race TTKSF+ confirmed the presence of a common resistance locus Sr9h in both lines. Test of allelism is required to establish genetic relationships between genes identified in present study and Sr9h. Marker cim117 linked to SrH was genotyped on set of wheat lines with Huhwa in the pedigree and is advised to be used for marker assisted selection for this gene, however, a combination of phenotypic and genotypic assays is desirable for both genes especially for selection of Sr9h in breeding programs.


INTRODUCTION
Stem rust, caused by Puccinia graminis f. sp. tritici (Pgt), is one of the most damaging fungal diseases of bread and durum wheat (Triticum aestivum L. and Triticum turgidum ssp. durum, (Desf.) Husnot., respectively) (Roelfs et al., 1992). It can cause complete yield losses under severe epidemics if susceptible cultivars are grown in rust hot-spot areas (Saari and Prescott, 1985;Dean et al., 2012). Global wheat production was threatened by stem rust when a highly virulent Pgt race known as Ug99 or TTKSK that combined virulence to Sr31 and various other commonly deployed resistance genes was detected in 1998 in Uganda (Pretorius et al., 2000;Jin et al., 2007Jin et al., , 2008. After its appearance, evaluations of international wheat germplasm and varieties in both field and greenhouse screenings revealed the predominance of wheat susceptibility to race TTKSK Singh et al., 2006Singh et al., , 2011. In subsequent years, new variants of Ug99 emerged that carry additional virulence to Sr24 Pretorius et al., 2010;Visser et al., 2011), Sr36 (Jin et al., 2009), and SrTmp (Newcomb et al., 2016) placing an even greater number of wheat varieties at risk. Races of the Ug99 race group have already spread over a wide geographical area including 13 countries in the East African highlands, Southern Africa, Yemen, Egypt and Iran, and there is a high chance of further spread into large wheat growing belts of Asia and beyond . Consequently, the new variants belonging to Ug99 race group and their geographical spread have further reduced the number of effective genes that can be used by breeding programs.
Observations on 'Huhwa' (H), 'Yaye' (Y) and their progenies in International Maize and Wheat Improvement Center (CIMMYT) germplasm indicated the presence of ASR to Pgt races belonging to the Ug99 group. Furthermore, several high yielding wheat lines derived from Huhwa and Yaye have been distributed in CIMMYT international nurseries. Despite the fact that Huhwa and Yaye carry stem rust resistance genes effective to Ug99 race group, their genetic and molecular characterization remains unclear. Therefore, the present study was undertaken to characterize stem rust resistance in these cultivars using recombinant inbred lines (RIL) populations, and to develop and validate SNP markers closely linked to these genes.

Field Evaluations
The two RIL populations and the parents were phenotyped for response to stem rust in field trials at the Njoro research station of Kenya Agricultural and Livestock Research Organization (KALRO) during two wheat growing seasons (2016-off and main seasons). About 5 g seed of each entry was planted in 0.7 m long paired-row flat beds, with 0.3 m spacing between them. Spreaders comprised of a mixture of stem rust susceptible cultivars 'Cacuke' and 'Robin, ' and six-Sr24 carrying lines (GIDs: 5391050, 5391052, 5391056, 5391057, 5391059, and 5391061) were planted as hill plots on one side of each test plot in the middle of the 0.3 m-wide pathways. Spreaders were also planted along the borders of the experimental field in 1 m plots. Spreaders were inoculated with a field collection of races TTKST and TTKTT belonging to the Ug99 race group Singh et al., 2015;Newcomb et al., 2016) by spraying a mixture of urediniospores suspended in water plus Tween 20 suspensions, and needle inoculations with the same suspension as described in Njau et al. (2013). Stem rust responses of each RIL and parents were recorded at postflowering stage when the susceptible parent displayed moderately susceptible to susceptible (MS to S) responses (Roelfs et al., 1992) with 70% disease severity (DS) following the Modified Cobb's Scale (Peterson et al., 1948). Responses were recorded a second time about 8 days later.

Molecular Marker Analysis
Leaf tissues were harvested from young leaves and genomic DNA was extracted using CIMMYT's laboratory protocols (Dreisigacker et al., 2015). The quality and quantity of DNA was assessed on 1% agarose gels and using a NanoDrop 8000 spectrophotometer (Thermo scientific). Selective genotyping was carried out for parents, 47 resistant and 47 susceptible RILs of C × H population, whereas bulk-segregant analysis (BSA; Michelmore et al., 1991) was performed on pooled DNA of 10 resistant and 10 susceptible RILs along with parents of the C × Y population. Genotyping was conducted using the 35K Affymetrix Axiom Breeders array (Allen et al., 2017) by outsourcing to TraitGenetics GmbH (Germany). The SNPs closely linked to stem rust resistance locus in each population were converted into Kompetitive allele-specific polymorphic (KASP) markers using the automated bioinformatics pipeline, PolyMarker (Ramirez-Gonzalez et al., 2015). KASP markers were then used to verify the targeted SNP polymorphism in the parents and polymorphic markers were applied on the respective mapping population. In addition, 3 KASP markers derived from the SNP markers IWB45296, IWB3891 and IWB45932 (Wang et al., 2014), linked to Sr9h were tested on the respective parents of both populations. Oligos were synthesized in 0.025 µM concentration from Sigma-Aldrich, United States. SNP markers were carried out in a final volume of 8 µl consisting of 3 µl of genomic DNA (30 ng/µl), 4 µl of 2x KASP mix [a mixture of FAM and HEX Fluorescence Resonance Energy Transfer (FRET) cassettes, ROX TM passive dye, an optimized buffer containing MgCl 2 , Taq polymerase and dNTPs], 0.11 µl primer mix containing two allele specific forward primers each with 12 and 30 µM of common primer and 0.89 µl of sterile water. PCR amplification was performed in a GeneAmp PCR System 9700 thermal cycler (Applied Biosystems) using the cycling conditions: 95 • C for 15 min for hot-start Taq DNA polymerase activation, followed by a touchdown profile of 9 cycles at 94 • C for 20 s and 61 • C for 1 min with a 0.6 • C reduction per cycle, followed by 38 cycles at 94 • C for 20 s and 55 • C for 1 min. The reactions were held at 12 • C for 3 min. Endpoint fluorescent images were visualized using the PHERAstar Plus (BMG LABTECH, Germany), and the data were analyzed using KlusterCaller TM software (LGC genomics).

Data Analyses and Genetic Mapping
Chi-squared analyses were performed to determine the goodnessof-fit of observed segregation with the expected genetic ratios for 1 gene segregation in F 4:5 RIL population using only the homozygous lines (HR: HS = 1:1). Genetic linkage maps for the stem rust resistance loci were constructed using genotypic data of molecular markers and seedling stem rust data using Inclusive Composite Interval Mapping (ICIM) software (Li et al., 2008). The Kosambi mapping function (Kosambi, 1943) was applied to convert recombination fractions into centimorgans (cM). Markers and respective resistance loci were grouped using a logarithm of odds (LOD) threshold of 10, the linkage maps were ordered and rippled using nearest-neighbor two-opt (nnTwoOpt) and sum of adjacent recombination (SAR) fraction algorithms, respectively.
Associations between markers on chromosome arm 2BL and quantitative resistance were revealed using the DS data recorded during 2016 off and main seasons using the ICIM software. A quantitative trait locus (QTL) was designated to be significant with a LOD value above a threshold calculated through 1000 permutation tests. Linkage map diagrams were prepared and aligned for a visual inspection of map order using MapChart 2.2 software (Voorrips, 2002).

Validation of Flanking Markers for Marker-Assisted Selection
Closely linked markers to stem rust resistance were used in validation studies by testing on a set of wheat genotypes carrying Huhwa and Yaye in their pedigree, respectively, and the efficacy of the reported markers for marker assisted selection (MAS). Sixty-seven and 44 lines with Huhwa and Yaye in their pedigree respectively, were chosen for validation of flanking markers. These lines were selected from the CIMMYT 9th and 10th Stem Rust Resistance Screening Nursery (SRRSN); the 26th High Rainfall Wheat Screening Nursery (HRWSN); 32nd, 33rd, and 34th Semi-Arid Wheat Screening Nursery (SAWSN); 47th, 48th, and 49th International Bread Wheat Screening Nursery (IBWSN) (Supplementary Tables 1, 2). Seedling IT's in response to races TTKSK, TTKSF+ and TTKTT (isolate 14KEN58-1, avirulent to Sr9h) were assessed at the USDA-ARS Cereal Disease Laboratory (CDL), St. Paul, MN following procedures described in Rouse et al. (2011b). Infection type data were recorded on the 0 to 4 scale of Stakman et al. (1962), where ITs 0, 1, and 2 (or combinations thereof) were considered resistant and ITs 3 and 4 were susceptible. For all wheat lines, stem rust infection response and severity data were also obtained from the international stem rust nurseries evaluated in 2013, 2014, and the 2015 off season in Kenya. The marker haplotype and stem rust phenotype in each line were compared to evaluate the efficiency of the markers in selecting resistance gene.

Greenhouse Evaluations
Huhwa and Yaye displayed resistant ITs 22+ and 2, respectively, in contrast to susceptible IT 3+ for the susceptible parent 'Cacuke' when tested at the seedling stage with Pgt race TTKSK ( Figure 1A). Although both populations were classified into three categories (C × H: 52 HR, 60 HS and 33 segregating;   Figure 1B). We refer to the single stem rust resistance genes segregating in the C × H and C × Y populations as "SrH" and "SrY," respectively.

Field Evaluations
Uniform disease development was observed in the field nurseries in both seasons in 2016. Terminal disease severity (TDS) and infection response for Cacuke was 70 S, whereas it ranged 10-20 MR for Huhwa and 5-15 RMR for Yaye over two seasons in the adult plant stage. A wide range of infection responses (MR to S) was observed among the RILs in both populations. TDS of RILs in C × H population ranged from 5 to 70% and 10 to 70% during 2016 off and main seasons, respectively (Figures 2A,B). In C × Y population, TDS varied from 10 to 100% and 10 to 80% during 2016 off and main seasons, respectively (Figures 2C,D). Only seedling stem rust data were used to classify lines carrying or not carrying the respective resistance locus in the two populations to avoid any chances of misclassification.

Effect of SrH and SrY on Stem Rust Disease Severity
To study the effect of SrH on disease severity, all RILs of the population C × H were reclassified into resistant (+SrH) and susceptible (−SrH) groups, based on the flanking markers cim109 and cim117. Among 112 RILs, two recombinants were observed between SrH and cim109 whereas 11 recombinants were identified between SrH and cim117. Based on the flanking markers, 100 non-recombinant RILs, (46 +SrH and 54 −SrH) were used for this analysis. In all cases, +SrH lines displayed MR or M infection responses in the field. Conversely, −SrH RILs displayed compatible infection responses (MS, MSS, or S). TDS of +SrH was 5 to 40% (mean = 17.2%) and 10 to 40% (mean = 19.3%) in off and main seasons in 2016, respectively, whereas it was 15-70% (mean = 38 and 38.4% during off and main seasons, respectively) in -SrH lines during both seasons (Figures 2A,B). The mean differences between the two groups were highly significant (t = 11.01, p < 0.0001 and t = 7.90, p < 0.0001 for 2016-off and 2016-main seasons, respectively) indicating that SrH confers significant effects on disease severity reduction in addition to conferring a resistant infection response. Similarly, RILs of C × Y were reclassified into resistant (+SrY) and susceptible (−SrY) classes, based on the flanking markers cim109 and IWB45932. Among 146 RILs, 18 recombinants were observed between SrY and cim109 whereas 11 between SrY and IWB45932. Nine recombinants were common between SrY, and markers cim109 and IWB45932. Based on flanking markers, 126 non-recombinant RILs, (69 +SrY and 57 −SrY) were used for this analysis. In all cases, +SrY RILs displayed RMR or MR infection responses in the field. In contrast, −SrY RILs displayed compatible responses (MS, MSS, or S). The disease severity of +SrY was 10-50% (mean = 31%) and 10-40% (mean = 17.2%) in off and main seasons in 2016, respectively, whereas it was 30-100% (mean = 66.8%) and 20-80% (mean = 39.4%) in −SrY lines (Figures 2C,D). The mean differences between the two groups were highly significant (t = 17.1, P < 0.0001 and t = 9.13, P < 0.0001 for 2016-off and 2016-main seasons, respectively) showing that SrY also confers significant effects on disease severity reduction in addition to conferring a resistant infection response.

Validation of Flanking Markers
Five markers (cim103, cim109, cim114, cim116, cim117) closely linked to SrH were used to genotype 67 CIMMYT advanced breeding lines derived from Huhwa (Supplementary Table 1). Based on the seedling data, 35 and 32 lines were postulated to carry and not-carry SrH, respectively. Marker cim117 amplified SrH-linked positive allele A in 32 lines postulated to carry SrH whereas 3 lines showed negative allele G. Positive validation of marker cim117 in 32 lines (out of 35) lines indicated 91.4% efficiency of this marker to identify lines with SrH. The seedling ITs of the SrH postulated lines ranged from 2-to 2 IT response. Out of 32 lines not postulated to carry SrH, 31 lines amplified negative allele G except one line with positive allele A. The other 4 markers were not associated with SrH in the advanced breeding lines.
We tested six markers cim103, cim109, cim114, cim116, IWB45932 and IWB45296 closely linked to SrY on a set of 44 wheat lines derived from Yaye (Supplementary Table 2). Marker IWB45932 amplified SrY-linked positive allele T in 14 lines postulated to carry SrY whereas one line showed negative allele C. Positive validation of marker IWB45932 in 14 of the 15 lines showed 93.3% efficiency of this marker to identify lines carrying SrY. When tested on 29 lines not postulated to carry SrY, marker IWB45932 amplified SrY-linked negative allele C in 13 lines whereas 14 lines carried SrY-linked positive allele T. Therefore, IWB45932 alone cannot be used to accurately predict the presence of SrY.

DISCUSSION
In this study, stem rust resistance genes SrH and SrY were identified in wheat lines Huhwa and Yaye, respectively, and mapped on chromosome arm 2BL using selective genotyping (SG) and bulked segregant analysis (BSA) methods, respectively. Both Huhwa and Yaye have been utilized as common parents in CIMMYT bread wheat improvement program. Therefore, it is expected that SrH and SrY are already widely present in recently developed wheat lines resistant to Pgt races in the Ug99 group.
Six characterized alleles at the Sr9 locus (Sr9a, Sr9b, Sr9d, Sr9e, Sr9f, and Sr9g) and Sr16 are not effective against Ug99 race, whereas Sr9h, Sr28, and Sr47 confer resistance Rouse et al., 2014). Three Sr9h-linked KASP markers (IWB45296, IWB3891, and IWB45932) showed no polymorphism between Cacuke and Huhwa. Although SSR marker gwm47 closely linked to Sr9h was polymorphic between Cacuke and Huhwa, the marker could not be integrated into a genetic linkage map with SrH due to lack of linkage. No variation shown by Sr9h-linked markers between parents and independent assortment of marker gwm47 with respect to SrH suggested that the stem rust resistance locus in Huhwa was unlikely to be Sr9h. The Sr28-linked marker wmc332 (Rouse et al., 2012) was mapped more than 30 cM distal to SrH. Moreover, IT differences between Sr28 and SrH, indicate potential unlikeness of SrH to Sr28. SrH displayed IT 22+ whereas Sr28 displayed IT;13-(as reported in Rouse et al., 2014).
In contrast, Sr9h-linked markers (IWB45296 and IWB45932) and two SSR markers (wmc175 and gwm47) were found polymorphic between Cacuke and Yaye. Upon genotyping the entire RIL population C × Y, markers IWB45932 and IWB45296 were mapped at 1.9 cM proximal and 2.6 cM distal to SrY, respectively. Association of Sr9h-linked markers with SrY indicated possible similarity of SrY to Sr9h. Sr28-linked marker wmc332 was monomorphic in the parents Cacuke and Yaye, therefore indicating that segregation at the stem rust resistance locus was possibly not associated with Sr28. An Aegilops speltoides introgression carried Sr47 which ruled out potential similarity of either SrY or SrH with Sr47 (Klindworth et al., 2012) since neither parent carry Ae. speltoides in their pedigree. Both SrH and SrY were mapped on chromosome arm 2BL. There is no common marker between the SrY and SrH maps that suggests that the position of the genes is different. Both genes are slightly distal to cim114. Advanced wheat lines carrying HUW234 and Yanac in their pedigree showed stem rust resistance indicating the most likely source of SrH is HUW234 and SrY is Yanac. Based upon pedigree information, Yanac could carry stem rust resistance gene Sr26, however, marker Sr26#43 did not amplify any product in resistant parent Yaye, indicating that the gene in Yaye is different from Sr26.
Susceptible reactions displayed by seedlings of resistant lines from both populations when tested with Sr9h virulent pathotype TTKSF+ confirmed a common resistance locus in both cultivars Huhwa and Yaye which is most likely Sr9h identified in wheat cultivar Gabo 56 (Rouse et al., 2014). Gene Sr9h is effective against Pgt race TTKSK whereas it is ineffective against race TTKSF+. Race TTKSF+, detected in South Africa and Zimbabwe , is virulent on Sr9h (Rouse et al., 2014). Similar response shown by Huhwa and Yaye indicated presence of Sr9h in both lines.
Association of KASP marker cim117 on a set of advanced wheat lines postulated to carry SrH indicates that this marker could be useful for marker-assisted selection of this gene in breeding. During marker validation process, we concluded that previously known Sr9h-linked markers can be used for identification of Sr9h or SrY together with phenotypic markers as the behavior of Sr9h linked markers vary with genetic backgrounds.
RILs carrying SrH showed disease severities of 5-40% and 10-40% during 2016 off and main seasons, respectively. However, these lines still displayed incompatible reactions to Pgt with infection responses of MR or M. In contrast, in the absence of SrH, disease severities of the RILs were 15-70% with corresponding infection responses of MS or S during 2016 off and main seasons. Similar, results were also found for SrY-carrying and SrY -lacking RILs. As reported by Basnet et al. (2015), disease severity per se is not useful for determining the presence or absence of race-specific stem rust resistance genes. Basnet et al. (2015) reported SrND643-carrying lines expressing 5-40% disease severities accompanied by R, MR or M infection responses. Similarly, Singh et al. (2011) observed that CIMMYT advanced lines with race-specific genes Sha7 and SrTmp displayed 1-30 and 5-60% disease severities accompanied by R or MR infection responses. Stem rust variation observed in the present study may be due to the presence of quantitative adult plant resistance genes that interact with SrH or SrY.

CONCLUSION
This study identified a common stem rust resistance locus located on chromosome arm 2BL in wheat cultivars Huhwa and Yaye. A robust SNP marker cim117 closely linked to SrH was identified and validated on advanced wheat lines carrying Huhwa in their pedigree. This marker can serve as a breeder friendly tool for marker assisted selection of SrH in breeding programs to generate new gene combinations in breeding Ug99 resistant cultivars.

KOMPETITIVE ALLELE-SPECIFIC POLYMORPHIC (KASP) MARKERS DATA SUBMISSION
We attempted to submit KASP markers generated in this study to European Variation Archive (EVA). EVA needs specific requirements including variant quality, filter, information, format etc. for each assay/variant that will only be available if we could have conducted axiom assay in our laboratory. As we outsourced genotyping using 35K axiom breeders' arrays, this specific information is not available to be included in the submission file to EVA. Markers were generated from publicly available contig sequences developed by International wheat Genome Sequencing Consortium (IWGSC) and available on http://www.cerealsdb.uk. net/cerealgenomics/CerealsDB/axiom_download.php. Markers developed in this study are PCR based markers representing SNPs of contig sequences from 35K axiom breeder's arrays. Therefore, we decided not to submit these markers to avoid any conflict of interest among wheat scientists.

AUTHOR CONTRIBUTIONS
MSR phenotyped the population, did analysis, and wrote the manuscript. MSR, SD, JN, and MS-S genotyped the population with SSR and SNP markers. MSR, RS, SB, and JH-E phenotyped the population at adult plant stage. MNR phenotyped the population at seedling stage.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fpls.2018.01788/ full#supplementary-material TABLE S1 | List of advanced wheat lines with Huhwa in their pedigree, their adult plant stem rust response, seedling infection types against Puccinia graminis f. sp. tritici (Pgt) races and haplotypes of SrH-linked markers.
TABLE S2 | List of advanced wheat lines with Yaye in their pedigree, their adult plant stem rust response, seedling infection types against Puccinia graminis f. sp. tritici (Pgt) races and haplotypes of SrY (Sr9h)-linked markers.