LUO Jiang-tao,ZHENG Jian-min,WAN Hong-shen,YANG Wu-yun,Ll Shi-zhao,PU Zong-jun

Crop Research Institute,Sichuan Academy of Agricultural Sciences/Key Laboratory of Biology and Genetic Breeding in Wheat(Southwest),Ministry of Agriculture and Rural Affairs,Chengdu 610066,P.R.China

Abstract Stripe rust,caused by Puccinia striiformis f.sp.tritici,is a serious disease in bread wheat (Triticum aestivum L.).Identification and use of adult plant resistance (APR)resources are important for stripe rust resistance breeding.Bread wheat line C33 is an exotic germplasm that has shown stable APR to stripe rust for more than 10 years in Sichuan Province of China.Here,183 recombinant inbred lines (RILs)derived from the cross between C33 and a susceptible line X440 were genotyped with diversity arrays technology (DArT)markers to identify resistance quantitative trait locus (QTL).Field trials were conducted in five years at Chengdu and Xindu of Sichuan Province,using maximum disease severity (MDS)as stripe rust reaction phenotypes.A total of four quantitative trait loci (QTLs)were detected,respectively designed as QYr.saas-3AS,QYr.saas-5AL,QYr.saas-5BL,and QYr.saas-7DS,explaining 4.14-15.21% of the phenotypic variances.QYr.saas-5BL andQYr.saas-7DS were contributed by C33.However,the level for stripe rust resistance contributed by them was not strong as C33,suggesting the presence of other unidentified QTLs in C33.QYr.saas-7DS corresponded to Yr18 andQYr.saas-5BL remains to be formally named.The RIL lines carrying combinations QYr.saas-5AL,QYr.saas-5BL,and QYr.saas-7DS showed comparability resistance with C33.The present study provides resources to pyramid diverse genes into locally adapted elite germplasm to improve the stripe rust resistance of bread wheat.

Keywords:APR,DArT,QTL mapping,Triticum aestivum L.

1.lntroduction

Stripe rust,caused by thePuccinia striiformisWestend.f.sp.triticiErikss.(Pst),is a serious and devastating wheat disease worldwide,resulting in significant yield losses(Wellingset al.2011).Although using fungicide treatment can control rust epidemics,their use increases production cost and causes environmental pollution.Use of resistance genes to breed resistant cultivars is an effective,economic,and environmental friendly method to control stripe rust.

To date,more than 70 formally named stripe rust resistance genes (Yr1-Yr79)and many quantitative trait loci(QTLs)have been described in wheat (Chen 2013; McIntoshet al.2013; Rosewarneet al.2013; Donget al.2017; Fenget al.2018).These resistance genes,according to the wheat growth stage of shown resistance,are generally categorized into all-stage resistance and adult plant resistance (APR)genes.Most of all-stage resistance genes are race specific and usually controlled by single genes.Because of their high levels of resistance,genes with all-stage resistance have been widely used in breeding of wheat,for example,Yr10,Yr17,Yr24/Yr26,and so on (Chen 2013).However,these race-specific genes can be rapidly overcome by new races.For example,Yr24/Yr26/YrCH42(on 1BS)had provided excellent resistance toPstrace CYR32 over the past decade and was widely used in wheat breeding programs (Wanet al.2004; Chenet al.2009; Zenget al.2014).However,the new race CYR34 (V26)has overcome its resistance in northwestern and southwestern China since 2008 (Zenget al.2014).

In contrast,APR or slow resistance gene is generally race non-specific and more durable (Renet al.2012; Chen 2013).“Fan 6” and its derivatives are a classical example for using APR.They exhibited high-temperature APR and once controlled of wheat stripe rust resistance more than 20 years in Sichuan Province,China.However,comparability to allstage resistance genes,the resistance of APR genes often showed the characteristics of quantitative character,a single QTL resistance is generally not strong enough to effectively control stripe rust.But there is an evidence to indicate that bread wheat with 4 to 5 APR QTLs could exhibit a high level of resistance (Luet al.2009; Singhet al.2011; Yanget al.2013).This generally requires the use of molecular markers to pyramid QTLs into a single wheat genotype.Several APR genes such asYr18,Yr29,Yr36,andYr46have been cloned (Fuet al.2009; Krattingeret al.2009;Chen and Kang 2017),providing important resources for durable resistance breeding.Identifying more APR genes to increase diversities of resistance genes is particularly important for durable control of stripe rust.

Stripe rust is frequently epidemic in Sichuan Province in southeast of China.The control of stripe rust in this region is thought to be very important for preventing the spread of stripe rust to other parts of China.The bread wheat line C33 (zidane89/3/peg`s`//hd2206/hork`s`)showed adult plant resistance to stripe rust in Iran (Karaj,Sari,and Moghan)(Malihipouret al.2012).After introduced from the International Center for Agricultural Research in the Dry Areas (ICARAD),our observations from 1997 to 2017 in Sichuan Province indicate it has stably shown APR to stripe rust,and susceptible in the seedling stage.However,the genetic basis of its resistance remains unknown.Here,we report the use of genome-wide linkage mapping to identify its stripe rust resistance QTLs.

2.Materials and methods

2.1.Plant materials

The stripe rust resistant line C33 was introduced from the ICARAD in 1997.A total of 183 recombinant inbred lines(RILs)were produced from the cross between C33 and a susceptible wheat line X440 by the single-seed descent method and used for QTL mapping.C33 stably performed a high level of APR to stripe rust during the past 20 years in fields under inoculated environments using mixed races prevalent in wheat production.X440 is highly susceptible to stripe rust in a range of environments.

2.2.Disease phenotyping

In order to map the resistance QTL,the RILs and two parental lines were evaluated for disease severity to stripe rust at Chengdu in 2013 and at Xindu County in 2014-2017.Field trials were conducted in randomized complete blocks with two replications.Each line was planted in a 1.5-m row spaced 30 cm apart.The highly susceptible cultivar Chuanyu 12 planted every 10 rows was used as a spreader.The stripe rust epidemic was initiated at the 3-leaf stage by inoculating plants with urediniospores mixtures,including currently predominant ChinesePstraces such as CYR32,CYR33,and CYR34.

Maximum disease severity (MDS)was scored as the infection percentage of leaves each line according to the modified Cobb Scale (Petersonet al.1948).The phenotype was recorded at the adult plant stage when the susceptible parent X440 displayed approximately 80% severity and repeated about a week later when it reached 90-100%(Petersonet al.1948; McIntoshet al.1995).

The stripe rust severity data collected in five consecutive years were subjected to analysis of variance.The genotype,environment,and genotype×environment interaction implanted in SAS were used to establish the levels of the stripe rust resistance across environments.

2.3.Linkage analysis and QTL identification

Genomic DNA was isolated from the 2-wk-old seedling leaves using the Zymo Research Plant DNA Kit (Zymo Research,USA).Diversity arrays technology (DArT)analysis was done by Triticarte Pty.Ltd.(Australia,http://www.triticarte.com.au).Calls shown heterozygote in RILs were set as missing values.Then the DArT marker data were filtered to contain ＜10% missing values and the segregation rations of markers were tested by Chi-square goodness-of-fit a 1:1 ratio at the significance level ofP＜0.01(Jighlyet al.2015).

Linkage and QTL mapping were performed using QTL Icimapping 3.3 (Menget al.2015).Filtered marker data were used to construct a linkage map.Markers were ordered according to the software,and map distances were calculated using the map function from QTL Icimapping 3.3,and linkage groups were declared on map distances and linkages between markers at a significance level ofP＜0.001.Linkage groups were assigned to chromosomes by referring to the partial known chromosome location markers.Unlinked markers were excluded in QTL analysis.

The inclusive composite interval mapping (ICIM)was used to detect the positions of QTL based on the MDS data in each environment (year).To determine the LOD threshold value above which a QTL is considered significant,1 000 permutations were performed (P＜0.05)(Doerge and Churchill 1996).The presence of a QTL was based on the genotype of the QTL-associated markers with the highest LOD values in the QTL interval.Stepwise regression was used to calculate the percentages of phenotypic variance explained (PVE)of individual QTL and the additive effect at the LOD peaks using IciMapping 3.3 (Menget al.2015).

3.Results

3.1.Phenotypic variations

The mean MDS was obviously lower in resistance parent C33 (0.0-15.0%)than the susceptible parent X440(40.0-85.0%)in the five environments from 2013 to 2017,indicating much higher resistance of C33 than X440(Table 1).The mean MDS of 183 RILs was 31% in those environments,ranging from 24.2 to 41.4%,with 2017 being the highest.The distribution frequency of MDS for RILs also varied among five environments (Appendix A)and all of them exhibiting a slightly skewness towards resistance(skewness values varied from 0.5 to 1.3; Table 1).The effects of genotypes,environments,and their interactions were highly significant (P＜0.001)(Table 2).

3.2.QTL analysis on stripe rust resistance

Out of 28 695 DArT and 7 032 SNP markers,7 608 (26.51%)and 776 (11.04%)were polymorphic between C33 and X440.These polymorphic markers were used to construct a genetic map.As a result,5 439 DArT and 623 SNP markers were mapped on the A (2 359),B (2 867),and D (836)genomes,covering genetic distances of 3 477.2,3 899.3,and 1 535.3 cM,respectively (Appendix B).The average distances between markers in A,B,and D genomes were 1.47,1.36,and 1.84 cM,respectively.

Four QTLs,for stripe rust resistance,were detected(Table 3 and Fig.1).They were located on chromosomes 3AS,5AL,5BL,and 7DS,designated asQYr.saas-3AS,QYr.saas-5AL,QYr.saas-5BL,andQYr.saas-7DS,respectively.

The resistance alleles atQYr.saas-5BLandQYr.saas-7DSloci were derived from the resistant parent C33.QYr.saas-7DSwas stably detectedacross four seasons and flanked by DArT markers 3956219 and 3956625 in a 8.4-cM interval,explaining 8.73,4.50,10.27,and 4.85% of the phenotypic variances in Chengdu 2013,Xindu 2015,Xindu 2016,and Xindu 2017,respectively.QYr.saas-5BL,flanked by DArT makers 1130553 and 1686211 in a 4.5-cM interval,was identified in two seasons,accounting for 4.55 and 15.21% of the phenotypic variances in Xindu 2015 and2016,respectively.

Table 1 Summary of the maximum disease severity (MDS)of recombinant inbred lines (RILs)and their parents

Table 2 Analysis of variance of the maximum disease severity (MDS)for stripe rust in the recombinant inbred lines (RILs)population derived from the C33/X440 cross

Table 3 Quantitative trait locus (QTL)for adult plant resistance (APR)to stripe rust detected in the recombinant inbred lines (RILs)population derived from the C33/X440 cross1)

The resistance allelesofQYr.saas-3ASandQYr.saas-5ALwere come from the susceptible parent X440.QYr.saas-3ASwasdetected in three seasons and flanked by DArT markers 4261525 and 1201921 within a 3.4-cM interval.It explained 6.59,6.72,and 4.14% of the phenotypic variances in Xindu 2014,2015,and 2016,respectively;QYr.saas-5ALwasdetected in two seasons and flanked by a SNP marker 3954825 and a DArT maker 7351706 in a 2.0-cM interval region.It explained 8.27 and 12.28% of the phenotypic variances in Xindu 2016 and 2017,respectively.

To obtain a more accurate individual and combined effects of these QTLs,the flanking markers for the three QTL regions with PVE more than 10% were used to select subsets of RILs in which these effects are controlled (e.g.,RILs possessing only the 5BL resistance were selected to represent the effect of the 5BL QTL,etc.).The results of this more controlled comparison are presented in Fig.2.The mean MDS of lines withQYr.saas-5BL(29.36%)was significantly lower than these of lines with the 5A (65.80%)and 7D (49.14%)resistance loci alone,but equal to the combination of 5A and 7D resistance alleles (24.86%).The mean MDS of lines with 5B resistance alleles in combination with either 5A (26.80%)or 7D (25.60%)resistance alleles were slightly lower than 5B resistance alleles alone(29.36%).Moreover,when the three resistance alleles were combined,the mean MDS (13.6%)was significantly reduced compared with both single resistance alleles and combinations of two among the three resistance loci (Fig.2).

4.Discussion

Here four resistance QTLs were identified,in which the resistance alleles of two (QYr.saas-5BLandQYr.saas-7DS)were from the resistant parent C33,whereas those of the other two (QYr.saas-5ALandQYr.saas-3AS)from susceptible parent X440.They may provide resources to pyramid diverse stripe rust resistance genes into locally adapted elite germplasm to improve the stripe rust resistance in wheat.

4.1.Comparison with previously reported genes/QTLs

Currently,Yr74is the only formally named resistance gene mapped on the long arm of chromosome 5B (Dracatoset al.2016),linked to the DArT marker 1091695 (Franckiet al.2009).However,QYr.saas-5BLis linked to the DArT marker 1130553,which is about 35 cM away from the DArT marker 1091695.According these linked markers'physical position,QYr.saas-5BL(at about 570 Mb on 5B)is far away from about 40 Mb toYr74(at about 531 Mb on 5B)(IWGSC 2018).More importantly,Yr74was a seedling resistance gene,suggesting thatQYr.saas-5BLandYr74are two different genes.Braianaet al.(2010)detected a resistance QTLQYr.sun-5Bnearby theQYr.saas-5BL.This QTL also was sensitive to different environments (Franckiet al.2009).Therefore,QYr.sun-5BandQYr.ica-5BLmay be the same or closely linked QTL.

Besides the geneYr18(Suenagaet al.2003),five APR QTLs also have been mapped on chromosome 7DS.However,all of them closely linked toYr18(Ramburanet al.2004;Luet al.2009; Barianaet al.2010; Zwartet al.2010;Yanget al.2013).In the present study,the QTLQYr.saas-7DSwas detected on 7DS,with high LOD scores in four environments.The DArT markers 3956219 and 995776 frankedQYr.saas-7DSto about 40 to 46 Mb region on Chinese Spring reference sequence v1.0,which is very close toYr18(about 47 Mb)(IWGSC 2018; Lillemoet al.2008),suggesting thatQYr.saas-7DSmay be same asYr18.

Fig.1 Chromosomal location of quantitative trait loci (QTLs)for stripe rust resistance.Only one represented marker was shown if there were more than two markers in the same position.Linkage 5A-1 was partly drawn because it is too long compared with other three linkage groups.

All-stage resistance gene,Yr76,is the only formally named resistance gene mapped on chromosome 3AS(Xianget al.2016).Two QTLs,associated with stripe rust resistance,were mapped on 3AS.Jighlyet al.(2015)reported that the DArT markerwpt-4868on chromosome 3AS was significantly associated with stripe rust resistance both at seedling and adult plant stages.Lillemoet al.(2008)also detected a QTL for stripe rust on 3AS linked to the SSR locusXbarc310.Based on the genetic maps,the DArT marker 120192 that was linked to our mapped QTLQYr.saas-3ASis close toXbarc310,but far away fromwpt-4868(Franckiet al.2009).

Fig.2 Average maximum disease severities (MDS)of lines carrying different combinations of quantitative trait loci (QTLs).Lines containing different QTL combinations (determined by flanking markers with underline in Table 3)were grouped together and the corresponding maximum disease severities were averaged over environments.Values are mean±SE.

The APR geneYr48and five QTLs for stripe rust reaction have been mappedon chromosome 5AL.Yr48was flanked by markersXwmc727andwpt-9800located on the distal region of 5AL (Loweet al.2011; Rosewarneet al.2012).Two QTLs,QYr.caas-5AL.2(Renet al.2012)and QTL-5AL (Rosewarneet al.2012)were mapped at a similar location toYr48.One QTLQYr.osu-5A(Fanget al.2011)was located in a region close to the centromere of chromosome 5A.Other two QTLs,QYrtb.pau-5A,flanked byXbarc151andXcfd12(Chhunejaet al.2008)andQYr.caas-5AL,flanked by markersXwmc410andXbarc261(Lanet al.2010)were mapped on the central region of chromosome 5AL (Fanget al.2011).The marker 7351706 closely linked toQYr.saas-5ALin this study was estimated 8 cM away from theQYr.caas-5ALlinked markerXwmc410.Therefore,we cannot determine whetherQYr.saas-5ALwas a different QTL.

4.2.Application in wheat breeding

The bread wheat line C33 is resistant to stripe rust and has a mean MDS of 7% (0-15%)in the five environments.This study detected two QTLs from this line,i.e.,QYr.saas-5BLandQYr.saas-7DS.In RILs,lines withQYr.saas-5BLexhibited average 29% MDS,which largely reduced the mean MDS about 30% compared with the null lines.It suggestedQYr.saas-5BLis valuable in wheat stripe rust resistance breeding.QYr.saas-7DS has a smaller effect on APR to stripe rust compared withQYr.saas-5BL,only with 9% reduction in disease severity compared with the null lines.However,it has good combinability withQYr.saas-5BL.This combinationled to 33% reduction of disease severities compared to lines without the two QTLs.However,the means MDS (25.6%)of these lines with both 5B and 7D resistance alleles was still observably higher than C33(mean 7%,varied from 0 to 15%),suggesting the presence of other unidentified QTLs.The failure of detection of other QTLs may be attributed to the unevenly distributing DArT markers used.

The bread wheat line X440 is susceptible to stripe rust.However,the resistance alleles of two QTLs (QYr.saas-5ALandQYr.saas-3AS)were detected from this line despite its small role on stripe rust resistance.The presence of these QTLsmay cause the field performance of RILs biased towards resistance.Meanwhile,the progenies combiningQYr.saas-5AL,QYr.saas-5BL,andQYr.saas-7DS showed a high resistance level that comparability to C33 (Fig.2).Thus,the small QTLs also can be used into wheat resistance breeding by combining stronger alleles.

5.Conclusion

In present study,two QTLs for APR to stripe rust in C33 were detected in C33/X440 RILs population and designated asQYr.saas-5BLandQYr.saas-7DS.Two QTLs (QYr.saas-3ASandQYr.saas-5AL)were also detected from the susceptible parent X440.These QTLs explained 4.14-15.21% of the phenotypic variances.QYr.saas-7DScorresponded toYr18andQYr.saas-5BLremains to be formally named.Although the level for stripe rust resistance contributed by single of them was not strong,combinationsQYr.saas-5AL,QYr.saas-5BL,andQYr.saas-7DSthree loci conferred the RIL lines had a comparability resistance with resistance parent C33.Thus,these QTLs and their closely linked markers are potentially useful for improving APR to stripe rust in wheat breeding.

Acknowledgements

The authors thanks to International Center for Agricultural Research in the Dry Areas,for providing the bread wheat line C33.This research was supported by the National Natural Science Foundation of China (31671683)and the Youth Foundation of Sichuan Academy of Agricultural Sciences,China (2016QNJJ-008).

Appendicesassociated with this paper can be available on http://www.ChinaAgriSci.com/V2/En/appendix.htm