Genome-wide QTL mapping for stripe rust resistance in spring wheat line PI 660122 using the Wheat 15K SNP array

Introduction Stripe rust is a global disease of wheat. Identification of new resistance genes is key to developing and growing resistant varieties for control of the disease. Wheat line PI 660122 has exhibited a high level of stripe rust resistance for over a decade. However, the genetics of stripe rust resistance in this line has not been studied. A set of 239 recombinant inbred lines (RILs) was developed from a cross between PI 660122 and an elite Chinese cultivar Zhengmai 9023. Methods The RIL population was phenotyped for stripe rust response in three field environments and genotyped with the Wheat 15K single-nucleotide polymorphism (SNP) array. Results A total of nine quantitative trait loci (QTLs) for stripe rust resistance were mapped to chromosomes 1B (one QTL), 2B (one QTL), 4B (two QTLs), 4D (two QTLs), 6A (one QTL), 6D (one QTL), and 7D (one QTL), of which seven QTLs were stable and designated as QYrPI660122.swust-4BS, QYrPI660122.swust-4BL, QYrPI660122.swust-4DS, QYrPI660122.swust-4DL, QYrZM9023.swust-6AS, QYrZM9023.swust-6DS, and QYrPI660122.swust-7DS. QYrPI660122.swust-4DS was a major all-stage resistance QTL explaining the highest percentage (10.67%–20.97%) of the total phenotypic variation and was mapped to a 12.15-cM interval flanked by SNP markers AX-110046962 and AX-111093894 on chromosome 4DS. Discussion The QTL and their linked SNP markers in this study can be used in wheat breeding to improve resistance to stripe rust. In addition, 26 lines were selected based on stripe rust resistance and agronomic traits in the field for further selection and release of new cultivars.


Introduction
Wheat stripe rust, caused by Puccinia striiformis Westend.f. sp.tritici Erikss.(Pst), is one of the most destructive diseases in the world (Chen, 2005;Milus et al., 2009).Losses from stripe rust typically range from 10% to 70% in commercial production environments, depending on the cultivar, prevailing climatic conditions, and inoculum pressure (Bariana et al., 2016;Zhou et al., 2022a).However, the disease can cause a 100% loss of yield for susceptible varieties (Bariana et al., 2016).Since 1950, the disease has occurred on an annual average of 4 million hectares in China.In particular, the five major outbreaks of wheat stripe rust in 1950,1964,1990,2002, and 2017 all occurred on over 5.5 million hectares, resulting in a loss of 13.8 million tons of yield (Li and Zeng, 2000;Wan et al., 2004;Huang et al., 2018).Stripe rust can be controlled by resistant cultivars, fungicides, and some cultural practices.Compared to other approaches, planting resistant cultivars has been proven to be the most effective, easy-to-use, economical, and environmentally friendly way to control disease (Line, 2002;Chen, 2005).
Depending on phenotypic performance at different growth stages, wheat rust resistance can be classified into two types: allstage resistance (ASR) and adult plant resistance (APR), sometimes also known as high-temperature adult plant (HTAP) resistance (Qayoum and Line, 1985;Chen, 2005;Lin and Chen, 2007;Rosewarne et al., 2013).All-stage resistance, also known as seedling resistance, can be detected at the seedling stage but is expressed at all growth stages.Such resistance is often race-specific and, thus, easily overcome by virulent races (Line, 2002;Chen, 2005;Chen, 2013).Due to race specificity, ASR often fails within 3-5 years of deployment (Jambuthenne et al., 2022).In contrast, HTAP resistance becomes more effective as plants grow older and the weather becomes warmer.It is usually non-race-specific, quantitatively inherited, and more likely to be durable.However, HTAP resistance is mostly incomplete, and the level is influenced by plant growth stage, temperature, and disease pressure (Chen, 2013).
The development of molecular markers, especially singlenucleotide polymorphism (SNP) markers, has revolutionized QTL analysis.A SNP is caused by a single-nucleotide mutation due to the insertion, deletion, and replacement of a single base segment in the genome.SNPs exist in the entire genomes of biological individuals and are the most abundant.SNP markers are now widely used in genetic analysis and breeding (Ma et al., 2019).Recent advances in sequencing technology have led to the availability of many SNP arrays in wheat (Rasheed et al., 2017).High-throughput genotyping techniques, including Wheat 9K (Cavanagh et al., 2013), 15K (Soleimani et al., 2020), 90K (Wang et al., 2014;Wu et al., 2018), 660K (Cui et al., 2017), and 820K SNP (Winfield et al., 2016) arrays, are now available.Among these SNP arrays, the 15K array is generally adequate and cost-effective for mapping traits of interest (Soleimani et al., 2020).
PI 660122, a spring wheat germplasm, was developed by the Wheat Health, Genetics, and Quality Research Unit of the US Department of Agriculture, Agricultural Research Service (USDA-ARS), and Washington State University and deposited in the USDA-ARS National Small Grains Collections (NSGC).In previous studies, the germplasm showed a high level of resistance in field tests over multiple years (Wang et al., 2012;Zhou et al., 2015b).At the seedling stage, it was resistant to US races PST-43 and PST-127 and Chinese races CYR29, CYR31, CYR32, and CYR33 and moderately resistant to US races PST-100 and PST-114 and Chinese race PST-HY8 of Pst (Wang et al., 2012;Zhou et al., 2015b).A comparison of greenhouse and field tests indicated that PI 660122 had effective ASR and possible HTAP resistance.The objectives of the present study were to further characterize the stripe rust resistance in PI 660122, map QTL for ASR and APR, and identify the QTL by comparing their chromosomal locations with previously reported stripe rust resistance QTL.

Plant materials
To map the QTL for stripe rust resistance in PI 660122, we developed a mapping population from a cross between Zhengmai (ZM9023, as the female parent) and PI 660122 (as the male parent).PI 660122 was developed from cross Avocet S/PI 610755 (Wang et al., 2012).Avocet S (AvS), an Australian spring wheat selection, is highly susceptible to most Pst races in China and many other countries and has been used as a susceptible control in stripe rust tests.PI 610755 is a Mexico spring wheat variety, selected from the cross Altar 84/Aegilops tauschii (191)//Opata M85.ZM9023, a spring wheat cultivar developed by the Wheat Research Institute of Henan Academy of Agricultural Sciences, is moderately or highly susceptible to the currently predominant Pst races in China (Xue et al., 2014).We developed a total of 239 F 5 and F 6 recombinant inbred lines (RILs) from the ZM9023 × PI 660122 cross, using the single-seed descent method.

Greenhouse tests
Seedling tests were conducted in a greenhouse to evaluate the stripe rust responses of PI 660122 and Zhengmai 9023.For each genotype, 10-12 seeds were seeded in a 9 cm × 9 cm × 9 cm plot.At the one-leaf stage, seedlings were uniformly inoculated with fresh urediniospores of a Pst race mixed with talc at a ratio of 1:50.Three Chinese Pst races, CYR31, CYR32, and CYR34, were used in the seedling tests.Inoculated seedlings were kept in a dew chamber in the dark at 8°C and above 100% relative humidity for 24 h.The seedlings were then moved to a growth chamber at 16°C with a daily 16-h light for stripe rust development.The infection type (IT) data were recorded 18 days to 21 days after inoculation using the 0-9 scale (Line and Qayoum, 1992).Seedlings of AvS were included as the susceptibility check in each race test.Later, 15 RILs selected for each containing only one QTL were also tested together with the parents with the three races at the seedling stage in the greenhouse under the same conditions.

Field tests
The F 5 and F 6 RILs and their parents were tested for stripe rust responses to stripe rust in the experimental fields in Mianyang (MY; 31°33′N, 104°55′E) in 2021 (21) and both MY and Guangyuan (GY; 22:32°14′N, 106°17′E) in the Sichuan Province in 2022 (22).The field tests were conducted with one replicate at 21MY and 22GY and two replicates (completely randomized block design) at 22MY based on the available seed quantity.Each plot consisted of a single row, 1.0 m in length and with 25 cm between rows.Approximately 20 to 30 seeds were sown in each row.AvS was planted in a row every 20 rows as a susceptible check and spore spreader for increasing stripe rust pressure and uniformity in the nursery.To increase the Pst inoculum, AvS was also planted around the nursery.MY and GY are ideal regions for stripe rust, as Pst can over-winter and over-summer, and the nursery was naturally infected without artificial inoculation (Zhou et al., 2019).
The stripe rust IT of each parent or RIL was rated on a scale of 0-9 (Line and Qayoum, 1992).Disease severity (DS) was scored using a modified scale as previously described (Lin and Chen, 2007).Both IT and DS data were collected twice in each season.The first record was taken when susceptible AvS showed approximately 80% severity, and the second was approximately a week later (Nsabiyera et al., 2018).Agronomic traits such as plant height (PH), spike length (SL), productive tiller number (PTN), kernels per spike (KPS), and thousand-grain weight (TGW) were determined to select RILs.PH was measured from the ground to the top of the spike excluding awn after the milking stage; KPS, SL, and PTN of each plant were counted at maturity; TGW was measured after harvest.

DNA extraction and genotyping
Fresh young leaves of PI 660122, ZM9023, and 239 F 5 RILs were harvested from the experimental field in January 2021.DNA from the fresh leaves was extracted using a modified cetyltrimethyl ammonium bromide (CTAB) method (Li et al., 2013).DNA was dissolved in ddH 2 O (100 mL), and DNA quality and concentration were determined by spectrophotometry (NanoDrop ND-1000, Thermo Scientific, Wilmington, DE, USA) after the DNA.DNA stock solutions were diluted with sterilized ddH 2 O to different concentrations according to individual experimental requirements for molecular analyses.

Statistical analysis, genetic map construction, and QTL mapping
Analysis of variance (ANOVA) and analysis of Pearson's correlation coefficients were performed to analyze the stripe rust phenotypic data using the "AOV" tool in the QTL Ici Mapping V4.2 software (Wang, 2009;Meng et al., 2015).The same software was also used to analyze the genotypic data.After the genotypic data were scanned for missing and undetected data, redundant markers were deleted using the "Bin" function.Genetic maps were constructed using the Kosambi mapping function (Kosambi, 2016).QTL mapping was performed using the genetic maps and the IT and DS data based on inclusive composite interval mapping (ICIM) with preset parameters Step = 1 cM, value p for input variables (PIN) = 0.0001, and logarithm of odds (LOD) = 2.5.A QTL was identified when the logarithm of odds (LOD) score was greater than 2.5.To determine the additive effects of QTL, the effects of QTL combinations were demonstrated by plotting box plots for mean IT and mean DS of RILs sharing the same number of beneficial alleles.

Stripe rust responses of the parents and RILs
In the greenhouse seedling tests, PI 660122 was highly resistant (IT of 2) to the tested three Chinese Pst races, whereas Zhengmai 9023 was highly susceptible (IT of 8-9) similar to the susceptible check AvS (Figure 1A).In the field tests under natural Pst infection, the final adult plant IT of PI 660122 was 2 across the two years and two locations, and its DS ranged from 5% to 10%, (Figure 1B).In contrast, Zhengmai 9023 was moderately resistant (IT of 5-6) with DS of 40%-50%.For comparison, AvS had IT of 9 and DS of 100%.
The RIL population had ITs ranging from 0 to 9 and DS from 0 to 90% across the years and locations (Figure 2).The IT and DS data from both sites and from both 2021 and 2022 at MY were each highly correlated (r = 0.76-0.81,p < 0.001 for IT; r = 0.61-0.75,p < 0.001 for DS) (Table 1).The ANOVA results showed significant variations (p < 0.001) among RILs, environments, and line × environment interactions for both IT and DS.The stripe rust phenotypes were influenced more by the environment than by the interaction of line and environment.The broad-sense heritability (h 2 ) was estimated at 0.92 using the IT data and 0.86 based on the DS data across the two sites (Table 2).

Genetic linkage map construction
A total of 5,432 SNPs in the 15K SNP array showed homozygous polymorphisms between the two parents.After the redundant markers were filtered out, 4,102 SNPs with known chromosome locations were obtained and used as inputs in the linkage analysis using QTL Ici Mapping V4.2.The 4,102 SNPs covered a total map length of 7,937.6 cM, with the genetic length ranging from 135.6 cM for chromosome 1B to 635.5 cM for chromosome 5A (Table 3).The number of markers per chromosome ranged from 60 for chromosome 6A to 322 for chromosome 2A, with an average of 189 SNPs.The mean distance between adjacent SNP markers ranged from 0.5 cM for chromosome 1B to 7.3 cM for chromosome 2D, with an overall mean of 1.9 cM.Genomes A, B, and D included 1,374 (33.50%), 1,672 (40.76%), and 1,056 (25.74%)SNPs covering lengths of 2,630.1 cM, 2,144.1 cM, and 3,163.4cM with mean marker distance of 1.91 cM, 1.28 cM, and 3.00 cM, respectively.The map was used to identify significant associations between SNPs and stripe rust resistance.

QTL combinations
To determine the effects of the QTL in various combinations for Pst resistance, the 239 RILs were grouped into different genotypic groups based on the presence of markers closely Clearly, RILs carrying any number of QTL had lower mean DS than those without any of the QTL.Lines without any QTL had a mean IT of 6.6 and a mean DS of 46.91%.In comparison, when 0, 1, 2, 3, 4, 5, 6, and more than 6 QTLs were combined, the lines with one QTL had mean IT of 5.9 and mean DS of 36.04%,those with two QTLs had mean IT of 5.0 and mean DS of 31.81%,those with three QTLs had mean IT of 4.0 and mean DS of 23.25%, those with four QTLs had mean IT of 3.7 and mean DS of 19.52%, those with five QTLs had mean IT of 3.1 and mean DS of 14.76%, those with six QTLs had mean IT of 2.7 and mean DS of 10.45%, and those with seven or more QTL had mean IT of 2.4 and mean DS of 10.58%, close to the resistance level of PI 660122 (Figure 5).

Selection of breeding lines
Various agronomic traits, including PH, PTN, SL, KPS, and TGW, of the parents and the 239 RILs were assessed in 2021 and 2022 in Mianyang and 2022 in Guangyuan.The mean PH values of PI 660122 and ZM9023 were 90.3 cm and 79.3 cm, respectively, and the RILs were mainly distributed in the range of 81-110 cm.The mean PTN values of PI 660122 and ZM9023 were 5 and 4, respectively, and the mean PTN values of RILs were between 4 and 10.The mean SL values of PI 660122 and ZM9023 were 9.8 cm and 8.4 cm, respectively, and the mean SL values of RILs were between 7.3 cm and 11.3 cm.The mean KPS values of PI 660122 and ZM9023 were 48 and 44, respectively, and the mean KPS values of RILs were between 33 and 58.The mean TGW values of PI 660122 and ZM9023 were 48 g and 44 g, respectively, and the mean TGW values of RILs were between 30.6 g and 58.4 g.In order to select RILs with desirable agronomic traits, the following criteria were used: PH between 80 cm and 100 cm, PTN 5 or more, SL greater than 9 cm, KPS not less than 45, and TGW over 42 g, with stripe rust IT of 1-3 and DS < 20%.Based on these criteria, 26 lines were selected.The QTLs detected by their highly associated SNP markers in the selected lines are listed in Table 6.These lines had at least two QTLs, and three lines (F 6 -61, F 6 -78, and F 6 -86) had as many as seven QTLs.According to Table 7, the DS is negatively correlated with SL, PTN, KPS, and TGW, indicating that with the increase of DS, the SL, PTN, KPS, and TGW will decrease.
The physical map position of Yr5/Yr7/YrSP was 685.265-685.27Mb.Yr43, an ASR gene, was flanked by Xwgp110 and Xwgp103, but its physical map position is unknown.Yr44, an ASR gene, was derived from spring wheat cultivar Zak and flanked by XpWB5/ N1R1 and Xwgp100, but its physical map position is unknown.
Yr53, an ASR gene, was derived from PI 480148 and flanked by Xwmc441 and XLRRrev/NLRRrev350.The physical map position of Xwmc441 was 598,064,318-598,064,477 bp.Yr72, an ASR gene, was derived from AUS27507 and flanked by Xsun481 and IWB12294.
Baimangmai (Hu et al., 2022).It was flanked by markers Xgpw7272 and Xwmc652.The physical map position of YrBm was 611.1-621.1 Mb.QYrPI660122.swust-4BLoverlapped with Yr50, but further studies are needed to confirm the relationship between QYrPI660122.swust-4BLand Yr50 and determine the relationships with other QTLs on chromosome 4BL.QYrPI660122.swust-4DS,an ASR QTL, was derived from PI 660122, and it was flanked by SNP markers AX-110046962 and AX-111093894 and corresponds to the region from 1,702,954 bp to 9,555,772 bp of the CS 4DS chromosome.Yr28 has been mapped to the short arm of chromosome 4D (Singh et al., 2000).Yr28 is a major ASR gene conferring stripe rust resistance from Ae. tauschii and located between SSR markers Xbcd265 and Xmwg634.Yr28 has been cloned and characterized, which encoded a typical nucleotide oligomerization domain-like receptor (NLR) (Zhang et al., 2019).The gene was further mapped between Xsdauw92 and Xsdauw96, approximately 0.13-cM interval, and its physical map position of Yr28 was 1.820-1.826Mb.Based on the physical map position, the resistance type, and the Mexican wheat genotype PI 610755 that has Ae.tauschii in the pedigree as stripe rust resistance donor of PI 660122 (Wang et al., 2012) et al., 2020), have been also reported on chromosome 4DL.QYr.ucw-4DL was linked with the IWA2395, and its physical map position was 497.65 Mb (Cobo et al., 2018).QYr.hbaas-4DL is linked to SNP marker IWB44356 (Jia et al., 2020), with the physical map position of approximately 477.9 Mb.Based on the different physical map positions of QYrPI660122.swust-4DLfrom those of Yr46, QYr.ucw-4DL, and QYr.hbaas-4DL, QYrPI660122.swust-4DL is likely a new QTL for stripe rust resistance.
QYrZM9023.swust-6DS was also derived from Zhengmai 9023 and flanked by SNP markers AX-11475193 and AX-109317417 corresponding to the 35,630,857-44,498,347-bp region of the CS chromosome 6DS.Few genes or QTL for stripe rust resistance have been reported on 6DS.Yr77 is an APR gene flanked by Xbrac54 and Xcfd188, and the physical map position of Xcfd188 is within the 238,118,118,personal communication).QYr.ucw-6D is linked with IWA167 (Maccaferri et al., 2015), which is at the physical map position of 73.2 Mb.QYr.ufs-6D is flanked by Xgwm325 and Xbarc175 (Agenbag et al., 2012), and its physical map position is within the 79.96-411.88-Mbregion.QYR7 is flanked by Xbcd1510 and XksuD27 (Boukhatem et al., 2002), and its physical map position is approximately 12 Mb.Based on the physical positions, QYrZM9023.swust-6DS is likely different from these genes or QTLs for stripe rust resistance genes previously mapped on chromosome 6DS.

Conclusions
In the present study, we mapped nine QTLs conferring different types and levels of resistance to stripe rust.Among these QTLs, QYrZM9023.swust-1BLwas identified as Yr29, QYrPI660122.swust-4DSas Yr28, and QYrPI660122.swust-7DSas Yr18, while QYrPI660122.swust-4BS,QYrPI660122.swust-4BL,and QYrZM9023.swust-6DSshould be new.We demonstrated that combinations of different QTLs increased the levels of resistance.Furthermore, we selected lines from the RIL population with high adequate resistance to stripe rust combined with desirable agronomic traits, and these lines can be used in further evaluation for releasing commercial cultivars.The resistant lines and molecular markers for resistance QTL should be useful in developing wheat cultivars with high levels and durable resistance to stripe rust.
FIGURE 1 Stripe rust response of resistant parent PI 660122, susceptible parent Zhengmai 9023 (ZM9023), and susceptible check AvS with Chinese race CYR34 of Puccinia striiformis f sp.tritici at the seedling stage (A) and stripe rust reactions on flag leaves of ZM9023 and PI 660122 (B).
FIGURE 5 Effects of individual QTL and their combinations on stripe rust scores illustrated by mean infection type (IT) (A) and disease severity (DS) (B) scores of recombinant inbred lines from Zhengmai 9023 × PI 660122 (ZM9023 in three environments, 2021 Mianyang (21MY), 2022 Mianyang (22MY), and 2022 Guangyuan (22GY).Box plots indicate the infection type (IT) and disease severity (DS) associated with the identified QTL and their combination.

TABLE 1
Correlation coefficients (r) of infection type (IT) and disease severity (DS) of the recombinant inbred lines Zhengmai 9023 × PI 660122 tested in different environments.

TABLE 2
Analysis of variance and estimate of broad-sense heritability (h 2 ) of infection type (IT) and disease severity (DS) in the recombinant inbred line (RIL) population of Zhengmai 9023 × PI 660122 tested at Mianyang in 2021 and 2022 and Guangyuan in 2022.
a df, degree of freedom.b "***" denotes the significance level of p < 0.001.

TABLE 3
Summary of chromosome assignment, number of SNPs, map length, and marker density of the genetic maps of the Zhengmai 9023 × PI 660122 recombinant inbred population.

TABLE 4
Summary of nine stripe rust resistance QTLs identified based on mean disease severity (DS) and infection type (IT) of 239 RILs from Zhengmai 9023 × PI 660122 cross-tested in Mianyang 2021-2022 and Guangyuan 2022.
a LOD, logarithm of odds score.bAdd, additive effect of resistance allele.cPVE, percentages of the phenotypic variance explained by individual QTL.QTLs, quantitative trait loci; RILs, recombinant inbred lines.

TABLE 5
Numbers of recombinant inbred lines from the Zhengmai 9023 × PI 660122 cross having only one stripe rust resistance and their infection types (ITs) at the seedling stage and mean IT and disease severity (DS) at the adult-plant stage in the fields of 2021 (21) and 2022 (22) at Mianyang (MY) and/or Guangyuan (GY).

TABLE 6
Mean stripe rust response, agronomic traits, and presence (+) and absence (−) of resistant QTLs detected with SNP markers in Zhengmai (ZM) 9023, PI 661022, and selected recombinant inbred lines a .

TABLE 7
Correlation coefficients (r) of important traits of the recombinant inbred lines Zhengmai 9023 × PI 660122.