Adult plant stem rust resistance in durum wheat Glossy Huguenot: mapping, marker development and validation

Adult plant stem rust resistance locus, QSrGH.cs-2AL, was identified in durum wheat Glossy Huguenot and mendelised as Sr63. Markers closely linked with Sr63 were developed. An F3 population from a Glossy Huguenot (GH)/Bansi cross used in a previous Australian study was advanced to F6 for molecular mapping of adult plant stem rust resistance. Maturity differences among F6 lines confounded assessments of stem rust response. GH was crossed with a stem rust susceptible F6 recombinant inbred line (RIL), GHB14 (M14), with similar maturity and an F6:7 population was developed through single seed descent method. F7 and F8 RILs were tested along with the parents at different locations. The F6 individual plants and both parents were genotyped using the 90 K single nucleotide polymorphism (SNP) wheat array. Stem rust resistance QTL on the long arms of chromosomes 1B (QSrGH.cs-1BL) and 2A (QSrGH.cs-2AL) were detected. QSrGH.cs-1BL and QSrGH.cs-2AL were both contributed by GH and explained 22% and 18% adult plant stem rust response variation, respectively, among GH/M14 RIL population. RILs carrying combinations of these QTL reduced more than 14% stem rust severity compared to those that possessed QSrGH.cs-1BL and QSrGH.cs-2AL individually. QSrGH.cs1BL was demonstrated to be the same as Sr58/Lr46/Yr29/Pm39 through marker genotyping. Lines lacking QSrGH.cs-1BL were used to Mendelise QSrGH.cs-2AL. Based on genomic locations of previously catalogued stem rust resistance genes and the QSrGH.cs-2AL map, it appeared to represent a new APR locus and was permanently named Sr63. SNP markers associated with Sr63 were converted to kompetetive allele-specific PCR (KASP) assays and were validated on a set of durum cultivars.


Introduction
Stem rust, caused by Puccinia graminis f. sp. tritici (Pgt), is one of the most devastating fungal diseases affecting wheat production worldwide. The detection of Pgt isolate Ug99 (Pretorious et al. 2000) and its variants with additional virulence led to a global research program (https:// www. globa lrust. org/). Subsequent appearance of highly virulent non-Ug99 races in Ethiopia (Olivera et al. 2012(Olivera et al. , 2015 and Germany (Olivera et al. 2017), and an outbreak of Pgt race TTTTF in Italy (http:// rustt racker. cimmyt. org/?p= 7083; Bhattacharya 2017) on both common wheat and durum highlighted Communicated by Lingrang Kong.
1 3 the need for continuous discovery of new resistance genes and monitoring of variation among Pgt populations.
Of the 62 stem rust resistance genes that have been described; most belong to the race-specific all-stage resistance (ASR) category. These genes generally confer high to moderate levels of resistance to avirulent races but are often short-lived due to the emergence of new races with virulence for commercially deployed resistance genes. A few adult plant (APR) stem rust resistance genes include Sr2 (McFadden 1930), Sr55 (Herrera-Foessel et al. 2014), Sr56 (Bansal et al. 2014), Sr57, and Sr58 . APR genes typically reduce the growth rate of the pathogen (Herrera-Foessel et al. 2012) and may be race non-specific. APR genes are also known to interact with ASR genes to enhance their expression (Ayliffe et al. 2013; Bariana et al. 2001;Lagudah 2011).
Many stem rust resistance genes in hexaploid wheat were introgressed from wild relatives or progenitors including Sr2, Sr22 and Sr35 (Fedak 2015). Among the tetraploid species of wheat, T. turgidum ssp. durum, is the main species grown commercially. Durum, the 10th most important crop in the world with an annual production of approximately 36 million metric tons, accounts for 5% of global wheat production (Bassi and Sanchez-Garcia 2017). Although the Mediterranean basin accounts for ~ 75% of global durum production, durum is grown to a lesser extent worldwide. For example, in Sub-Saharan Africa, Ethiopia is the largest wheat producer, with nearly 40% of its wheat-growing area under durum production (Bergh et al. 2012) and accounting for 90% of durum production in that region (Sall et al. 2019). Despite its economic importance and value as a source of stem rust resistance genes in common wheat, far less information is available on stem rust resistance in tetraploid wheat than in common wheat.
Many commercial durum cultivars carry stem rust resistance genes Sr8b, Sr9e and Sr13a-d allelic variants either singly or in different combinations (Toor et al. 2013;Letta et al. 2013;Gill et al 2021), with only Sr13a-c being broadly effective against the Ug99 lineage. The adult plant stem rust resistance gene Sr2, which was introgressed into hexaploid wheat from tetraploid Yaroslav emmer, has remained effective for over a century (Ellis et al. 2014; https:// globa lrust. org/ gene/ sr2), but it is not known to be present in any durum cultivar. A relatively few studies have described new stem rust resistance genes including APR in durum genotypes. Herrera-Foessel et al. (2011) reported the presence of Sr58/Lr46/Yr29 in durum cultivar Quetru and it was later detected in a large number of durum cultivars (Li et al. 2020). Haile et al. (2012) identified minor QTL against Ug99. Toor et al. (2013) screened 73 tetraploid landraces from the Watkins Collection and identified lines with varying levels of APR. Using association mapping on 183 elite durum accessions, Letta et al. (2013) identified 13 QTL located in a chromosomal regions where no stem rust resistance gene was reported previously; however, further characterization is required to determine the effectiveness and distribution of those QTLs.
According to Hare (1997), Glossy Huguenot is a waxless selection believed to be derived from the Algerian durum variety Huguenot, an awnless selection from Medeah. Both Huguenot and Medeah were cultivated in Western Australia between 1890 and 1922 (Valder 1900). Hare (1997) crossed Glossy Huguenot (GH, Aus2499) with the stem rust susceptible durum genotype Bansi and based on analysis of F 3 lines concluded that stem rust resistance in Glossy Huguenot was conferred by a single dominant APR gene. The possibility of a second gene was not ruled out. With the objective to map the adult plant stem rust resistance in Glossy Huguenot, we developed a recombinant inbred line (RIL) F 6:7 population and conducted QTL analysis using the 90 k single nucleotide polymorphism (SNP) wheat array and identified resistance-linked molecular markers.

Plant materials
Initially, F 3 population from a cross of adult plant resistant parent GH and susceptible parent Bansi were enhanced to F 5:6 RIL generation and the population was named GHB. This population was screened at the University of Sydney Plant Breeding Institute, Cobbitty in 2003-2010. The segregation for maturity and stripe rust among this population affected stem rust assessments. Hence, a set of 210 F 6:7 RILs from a cross between GH and a stem rust susceptible but stripe rust-resistant RIL GHB14 (M14) was developed through the single seed descent method. M14 was similar in maturity to GH. F 7 and F 8 generations were tested for adult plant stem rust response variation.
A set of 20 Australian durum cultivars and 155 lines from 39th International Durum Screening Nursery (IDSN) from CIMMYT were used for marker validation. Parents Glossy Huguenot and M14 were included as controls.

Genetic map construction and statistical analysis
DNA from F 6 generation was isolated according to Lagudah et al. (1991). A set of 192 RILs and parental lines were genotyped using the Infinium 90 K wheat SNP array described in Wang et al. (2014), following the manufacturer's instructions (Illumina Ltd, USA). A high-density linkage map was constructed from 7811 markers using ASMap (Taylor and Butler 2017), an interface to MSTmap (Wu et al. 2008) for the R environment (R Core Team 2020, https:// www.r-proje ct. org/). Genetic distances were estimated using the Kosambi function (Kosambi 1943). Linkage groups were formed using the criterion of p > 1e−6 for each pairwise marker association. Linkage groups were assigned to the 14 chromosomes using marker location in a durum consensus map that contained most of the markers (Maccaferri et al. 2015). Rust severity data from trials were analysed with a mixed linear model using ASReml-R (Butler et al. 2009), treating years as a fixed effect and genotypes and replicates as random effects. The QTL analysis was performed using multivariate whole genome average interval mapping (Broman et al. 2003;Verbyla and Cullis 2012).
In addition, we also used MapManager version QTXb20 (Manly et al. 2001) for linkage map construction. The Kosambi mapping function (Kosambi 1943) was used to convert recombination fractions into centiMorgans (cM). The redundant markers were excluded using the command "hide redundant loci", option and the phenotypic data was imported into MapManager. QTL cartographer (Wang et al. 2012) was used for composite interval mapping (CIM) with 1000 permutations. Graphical representation of the linkage map was generated using MapChart (Voorrips 2002).

Molecular marker analysis and marker validation
Simple sequence repeat (SSR) markers previously assigned to different wheat chromosomes were obtained from the GrainGenes website (https:// wheat. pw. usda. gov/ GG3/). Kompetitive allele-specific PCR (KASP) assays for traitlinked SNPs were obtained from PolyMarker (http:// www. polym arker. info/ desig ned_ prime rs). All KASP markers were tested on the entire RIL population. Sequences of some critical markers are provided in Table S1. PCR reactions were performed in 96-well plates in a total volume of 8 μl, comprising of 25-30 ng of genomic DNA, 4 μl of KASP master mix (LGC Biosearch Technologies, Hoddesdon, UK), and 0.11 μl of assay mix containing 12 mM of each allele-specific forward primer, and 36 mM of the reverse primer. The cycling conditions were: 15 min at 95 °C, 10 touchdown cycles of 20 s at 94 °C, 60 s at 65 °C (dropping 0.8 °C per cycle), 25 cycles of 20 s at 94 °C, and finally, 60 s at 57 °C. Fluorescence detection of PCR products was done using Bio-Rad CFX96 manager v 3.1 software. If genotyping clusters were not formed after initial amplification, 5 to 8 additional amplification cycles were performed, and the samples read again. SNPLr46G22, linked to the Lr46/Sr58/Yr29 APR locus (Lagudah unpublished) was used as a KASP marker to genotype the mapping population based on Kandiah et al. (2020). KASP_32429 was amplified using allele-specific primers 1. AGT GAA TGC ATT GCT CAT AAA AAT T, 2. AGT GAA TGC ATT GCT CAT AAA AAT G and a common primer CAT TAA CCA GTA ACC ACC AAAGG. Allele primers 1 and 2 were tagged with FAM and HEX, respectively.

Assessment of QTL interaction
Mean stem rust severities across sites/years for RIL carrying GH and M14 alleles for peak markers were compared to assess QTL interaction and to determine their individual mean contributions towards reductions in stem rust severity. Least significant differences (LSD) were calculated to compare mean disease severities of RILs carrying different combinations of QTL using R (https:// www.r-proje ct. org/).

Rust responses
Glossy Huguenot was susceptible (IT 3+) to Australian Pgt races 21-0 and 98-1,2,(3),(5),6, this confirmed the results published by Hare (1997), who also used Pgt 21-0 and additional races in his study and confirmed APR in Glossy Huguenot. Phenotyping at the seedling stage with the Mexican Pgt race RTR also showed IT 33+ .
GH exhibited 10 to 20% stem rust severities across all field trials, whereas the responses of the susceptible parent M14 ranged from 50 to 100% at the adult plant stage. The stem rust responses of RILs ranged from 10 to 80% in 2014 and 2019 trials, whereas 20 to 100% disease severities were noted in 2015. Figure 1 shows the distribution of disease severity during three seasons. The high inoculum pressure in 2015 compared to the other seasons resulted in higher disease severities across the whole population.

QTL mapping
The map of the GH/M14 RIL population (192 lines) comprised 23 linkage groups, each of which mapped with high 1 3 confidence to one of the 14 durum chromosomes. A highquality map was generated based on 7812 SNPs distributed across all 23 linkage groups. There were 1293 recombination bins, resulting in an average of six markers per bin covering a total map length of 1937.2 cM.
Significantly positive correlations among the rust severity data among the three trials were observed (Table 1). One QTL each was detected on the long arms of chromosomes 1B and 2A in each growing season and these were named QSrGH.cs-1BL and QSrGH.cs-2AL, respectively. Both QTL were contributed by GH. The positions of SNP markers associated with both QTL are shown in Fig. 2 and the QTL intervals, the likelihood of odds (LOD) scores and phenotypic variation explained (PVA) are presented in Table 2.
Several SSR markers previously mapped to chromosomes 1B and 2A were used to define the target regions. Only gpw1077-1B among nine SSR markers was polymorphic between the parents and was used for mapping (Fig. 3). Markers gwm265, gwm382, gwm312, and cfa2164 from chromosome 2A were polymorphic between the parents and all were used in mapping, only gwm312 mapped close to the QTL and is included in the map (Fig. 3)

Mendelisation of QSrGH.cs-2AL
QSrGH.cs-1BL was mapped to the chromosome 1B region reported to carry the pleiotropic gene Lr46/Yr29/Sr58/ Pm39. This QTL was therefore considered to be either Sr58 or a closely linked gene. Genotyping with Sr58/Lr46/ Yr29/Pm39 (hereafter referred to as Sr58)-closely linked marker csLV46G22 predicted its presence in GH. The GH/ M14 RILs that carried the Sr58-linked csLV46G22 allele were removed and the remaining RIL population was used to Mendelise QSrGH.cs-2AL. Thirty-seven RILs were scored resistant (mean stem rust severity 52.1%) and 36 were susceptible (stem rust severity 64.5%). The phenotypic data were converted into genotypes; 'A' for GH and 'B' for M14 and incorporated into the GH/M14 map. QSrGH.cs-2AL was flanked by markers IWB32429 (2.7 cM) and IWB4881 (3.0 cM) (Fig. 3). This locus was named Sr63. Table S2 shows genotyping of critical RILs with KASP markers.

Development of KASP assays and validation
The SNP markers linked to QSrGH.cs-2AL (Figs. 3, 4) were converted to KASP assays. Only IWB32429 was reliably scored as a diagnostic KASP (KASP_32429). As expected, the KASP marker co-segregated with the respective SNP from which it was developed. KASP_32429 was also used for genotyping a panel of durum cultivars. No cultivar carried the Sr63-linked KASP_32429 allele indicating the absence of QSrGH.cs-2AL (Table 4). Of the 155 durums from CIMMYT, 2 showed the Sr63 diagnostic SNP (Table S3). This marker will be useful for marker-assisted introgression of this APR allele. Marker csLV46G22 diagnostic for Sr58/Lr46/Yr29 has been described previously (Cobo et al. 2019;Kandiah et al. 2020).   Fig. 5).

Discussion
Although Pgt race TTKSK (aka Ug99) detected in Uganda in 1998 (Pretorius et al. 2000) posed a major threat to wheat production globally, durum cultivars were not significantly affected by that race or its derivatives, and presumably due to the widespread presence of resistance gene Sr13 and its alleles (H.S. Bariana unpublished results). However, other Pgt races virulent for these genes were reported (Singh et al. . Race analysis following a severe stem rust outbreak in Ethiopia in 2013-14 indicated the presence of TTKSK (Ug99), JRCQC and TKTTF (Olivera et al. 2015). Only 5% of durum cultivars had effective resistance against the non-Ug99 races JRCQC and TKTTF (Letta et al. 2013;Olivera et al. 2015). The Pgt race involved in a stem rust outbreak in Sicily in 2016 that affected both common wheat and durum wheat was typed as TTTTF, which is virulent on Sr13 (http:// www. globa lrust. org/ races/ ttttf; Bhattacharya 2017). Although there is now evidence for allelic variation following the cloning of Sr13 and different sequence variants (haplotypes) respond differently to the Pgt races (Zhang et al. 2017), breeding material is highly unlikely to contain all haplotypes. We identified two consistent stem rust resistance QTL, QSrGH.cs-1BL and QSrGH.cs-2AL, explaining high levels of variation in stem rust response in Glossy Huguenot. Hare (1997) concluded that a single APR gene was present in Glossy Huguenot, but the wide variation in maturity in his materials would have been a confounding factor in determining reliable differences among his materials. QSrGH.cs-1BL was mapped in the same region as the pleiotropic APR gene Sr58. Molecular isolation of this locus will shed further light on this. Marker analysis with the linked marker, csLV46G22, also suggested the presence of Sr58 in GH. However, a higher level of resistance conferred by Sr58 remains a query to be resolved, but it is well known that disease resistance genes confer higher levels of protection at lower ploidy levels. A high level of expression by this chromosome 1BL locus has been observed for stripe rust resistance (Pakeerathan et al. 2019). Although, the ASR gene Sr21 is also located on the long arm of chromosome 2A and the current knowledge indicates that distinctive ASR and APR genes are unlikely to be allelic. Moreover, we located QSrGH.cs-2AL (Table 3) in the interval 683-696 Mb in CS-Refv2.1 (Zhu et al. 2021), whereas Sr21 was located at 713 Mb (Chen et al. 2018). Since there is no previously reported APR gene in this region, SrGH is named Sr63.
Although the most widely used APR gene Sr2 in common wheat originates from the tetraploid Yaroslav emmer, no known durum cultivar is known to carry this gene. Li et al. (2020) suggested the presence of Sr58 in several durum cultivars. Tests of tetraploid landraces in the 'Watkins Collection' showed varying degrees of APR to stem rust (Toor et al. 2013). Genome-wide association mapping of seedling and adult plant resistance in a panel of 280 durum wheats (Megerssa et al. 2021) identified regions on chromosomes 5B, 6A and 7A for field resistance. No QTL was identified on the long arms of chromosomes 1B or 2A. Test in Ethiopia on durum mapping populations indicated the involvement of chromosome IBL ( Bariana and Bansal unpublished results).
High levels of resistance to Ug99, has been achieved by combining APR genes in hexaploid wheat (Pretorius et al. 2017). Comparison of disease severities of RILs carrying  QSrGH.cs-1BL QSrGH.cs-2AL

No QTL
Sr58 and Sr63 in combination with those carrying these genes individually demonstrated positive interaction causing reduced stem rust severity and it is even evident from responses of GH in different environments. All lines of a panel of durum accessions in this study were positive when tested with a marker that is linked with Sr58. Sr63 linked allele was detected only in two out of 175 lines indicating that it is likely to be rare in durum cultivars and this gene can be introgressed into breeding populations using the KASP marker developed in this study.