Identification of herbicide resistance loci using a genome-wide association study and linkage mapping in Chinese common wheat

Chaonan Shi,Yueting Zheng,Junyou Geng,Chunyi Liu,He Pei,Yan Ren,Zhongdong Dong,Lei Zhao, Ning Zhang, Feng Chen

National Key Laboratory of Wheat and Maize Crop Science/Agronomy College, Henan Agricultural University, Zhengzhou 450046, Henan,China

ABSTRACT Carfentrazone-ethyl and tribenuron-methyl, the two widely used herbicides for weed control in field crops,frequently cause phytotoxicity to wheat seedlings in the field.In this study, a total of 697 wheat accessions containing three panels were scanned using wheat 90 K and 660 K SNP arrays to identify important herbicide resistance loci. Genome-wide association study(GWAS)revealed 329 significant single-nucleotide polymorphisms(SNPs)with phenotypic variance explained (PVE) of 11.3% to 27.6%. Among these SNPs, 15 were detected in multiple environments and they were mainly distributed on chromosomes 1B,2D, 5B, 5D, 6D, and 7D. Further analysis indicated that gHR-5B (467-587 Mb), gHR-7D(46-52 Mb), and gHR-1B (517-580 Mb) were important herbicide resistance loci in wheat.Linkage mapping in a bi-parental population detected one QTL(qHR-1B)with PVE of 7.44% to 8.28%.This is reliable locus because its physical position(554-566 Mb)overlapped with gHR-1B by GWAS in the genome of Chinese Spring.This study provided some herbicide-resistant germplasm and important genetic loci for identifying genes of common wheat.

1. Introduction

Wheat is one of the most important food crops globally.However, wheat plants frequently suffer injuries from the external environment in their developmental stage, which results in yield losses[1,2].Multiple abiotic and biotic stresses,such as heat,cold,diseases,insect pests,and weeds,seriously affect plant growth and development. Previous studies showed that four herbicides (2,4-D butyl esters, Piaoma,Shima and Juxing) had serious effects on grain filling, and wheat yield decreased by 113.8 to 167.1 kg ha−1in different cultivars [3,4]. In recent years, the weed populations in farmland have increased, and the manual removal of weeds is becoming difficult due to many young adult villagers leaving farming communities to join urban populations.Therefore, herbicide application is becoming increasingly popular in wheat cultivation to eliminate the hazards associated with weeds[5].

In agricultural environments, weeds are controlled by the regular application of herbicides, but some weeds have evolved resistance due to long-term herbicide use [6,7]. In China,farmers frequently use chemical herbicides in conventional agricultural production[8].There are many reports and warnings about the adverse effects of herbicides on ecological,environmental, and human health (through the process of food production). The excessive use of herbicides has caused serious herbicidal pollution, and wheat cultivars in the field also suffer from different degrees of herbicide damage[9].

As a wide spectrum and effective herbicide, carfentrazoneethyl and tribenuron-methyl were commonly applied to wheat cultivars from the Yellow and Huai River valleys of China in recent years.Carfentrazone-ethyl is a selective herbicide applied to kill specific weeds in wheat fields.In the process of chlorophyll biosynthesis under light conditions, carfentrazone-ethyl causes the accumulation of toxic intermediates by inhibiting protoporphyrin peroxidase.This damages the cell membranes of weeds,leading to rapid drying and death of leaves [10]. Carfentrazoneethyl has been shown to disrupt the soil balance by reducing soil microbial diversity and abundance, soil enzyme activity, biochemical activity,and spring wheat yields[9].On the other hand,tribenuron-methyl is a type of acetolactate synthase (ALS)inhibitor with a high activity, strong selectivity, and wide herbicide spectrum. It is safe for mammals and is widely used in wheat fields to control broadleaf weeds[11].By inhibiting the action of its only target,ALS,tribenuron-methyl can prevent the biosynthesis of some essential branched-chain amino acids in plants. Long-term use of tribenuron-methyl can easily lead to herbicide resistance,and the application of excessive dosages can even lead to herbicide pollution and wheat phytotoxicity[12-14].Aminobenzene sulfonyl as one kind of the tribenuron-methyl can cause slow root growth, delay the returning green stage,inhibit the growth of the plant, decrease the grain number per ear,and cause the death of the whole plant in winter rape[15].

Carfentrazone-ethyl and tribenuron-methyl are effective herbicides against broadleaf weeds, especially against resistant weeds that have evolved in wheat fields due to the long-term use of sulfonylurea herbicides. Both herbicides are widely used in wheat fields in the Yellow and Huai River valleys of China [16].However, when carfentrazone-ethyl is mixed with tribenuronmethyl, it affects the uniform distribution of water and causes burning spots on wheat leaves[17].Resistance to carfentrazoneethyl and tribenuron-methyl varies among different crops and even among different cultivars of the same crop. Herbicide accidents have severely affected agricultural production,threatened food security, and cause serious economic losses. In the 2009 cropping season, large areas of wheat in Taizhou and Zhenjiang of Jiangsu province in China were suffering from different degrees of herbicide damage after the application of a carfentrazone-ethyl and Maiji mixture (wettable powder of 15% clodinafop-propargyl) due to the high concentration of carfentrazone-ethyl.Some wheat seedlings in the first or second leaf stage died due to serious leaf damage [18]. These consequences can be effectively mitigated by using herbicide-resistant cultivars in wheat production.Therefore,it is very important to identify herbicide-resistant wheat cultivars and resistance genes.

To date, many herbicide-resistant genes have been reported in different weeds, including Lolium multiflorum [19],

Papaver rhoeas [20], Descurainia sophia L. [21], and Myosoton aquaticum L. [22]. Many of the herbicide-resistant genes were involved in the action of cytochrome P450, ATP-binding cassette transporters, glutathione S-transferases, and glycosyltransferases[23,24].In rice,a herbicide-resistant gene HIS1(4-hydroxyphenylpyruvate dioxygenase [HPPD] inhibitorsensitive 1) was identified to confer resistance to benzobicyclon (BBC) and other beta-triketone herbicides.HIS1 encodes an Fe(II)/2-oxoglutarate-dependent oxygenase that detoxifies beta-triketone herbicides by catalyzing their hydroxylation. The application of the HIS1 gene in crop breeding will support the diversity of herbicide selection and the control of weeds in farmland[25].Therefore,the screening of herbicide resistance loci in crops is very urgent.

Few studies have been reported on the identification of herbicide resistance loci in wheat cultivars. Therefore, it is highly important to screen for herbicide-resistant wheat germplasm and to identify their important genetic loci for application in wheat breeding. Genome-wide associations studies (GWAS) and quantitative trait loci (QTL) mapping have been successfully used for identifying important loci associated with different traits in many crops [26-29]. The aim of this study was to screen some common wheat germplasm with strong herbicide resistance and to identify the important herbicide-resistant genetic loci using GWAS and QTL mapping. This information could be used for breeding herbicide-resistant wheat cultivars in wheat breeding programs.

2. Materials and methods

2.1.Plant materials

A total of 697 Chinese wheat accessions were divided into three panels. The first panel (Panel_I) consisted of 163 historical and current Chinese wheat cultivars as described by Chen et al. [30] and Sun et al. [31]. The second panel(Panel_II) was composed of 243 wheat accessions in the Yellow and Huai River valleys after 2014 [32]. The third panel(Panel_III) was the offspring of Yanzhan 1 containing 291 advanced introgression lines.Panel_I and Panel_II were grown with two and three replicates, respectively, in Suiping(33.22°N, 114.03°E) in the 2016/2017 cropping season and in Yuanyang (35.11°N, 113.95°E) in the 2017/2018 cropping season. Panel_III were planted in Suiping in the 2016/2017 cropping season with three replicates. The bi-parental population UP (UC1110/PI610750) containing 187 recombinant inbred lines (RILs, F10generation) was kindly provided by Jorge Dubcovsky from the University of California,Davis.This population was planted in Suiping in 2017/2018 cropping season with two replicates.

2.2. Herbicide treatment and phenotypic evaluation of herbicide resistance

A total of 26 g of 10% tribenuron-methyl wettable powder and 10 g of 10% carfentrazone-ethyl wettable powder were added to 15 L of water and were fully blended in a bucket. At the stage when the wheat plants began to turn green (sheaths start elongating and getting erect), the 22.5 mL m−2of herbicide reagent was evenly sprayed on the wheat seedlings.The phenotype of wheat herbicide resistance was classified after spraying for 5 to 7 days.

The phenotypic evaluation used to obtain a herbicide resistance index (HR) was based on the number and area of wheat leaf burns in the field. Phenotypic evaluation criteria(0-4)were used to evaluate the resistance of the cultivars(Fig.1-A-C), where 0: No apparent burn spots; 1: A few burn spots appeared on the leaves of the plant, but the plant displayed normal growth; 2: Less than half of the plant leaves showed obvious burn spots, and the plant growth was slightly affected;3:More than half of the plant leaves showed obvious burn spots, and plant growth was seriously affected;4: Plant leaves of the whole wheat line showed large areas of burn spots and withered.

2.3. Genotyping and genome wide association study

Panel_I was genotyped using a wheat 90 K SNP assay [33].Panel_II and Panel_III were genotyped using a wheat 660 K SNP assay [30]. SNP markers with a minor allele frequency(MAF) ≤0.05 and missing data ＞20% were removed using PLINK software for quality control[34].

The population structure of the three panel populations were evaluated using STRUCTURE software 2.3.4 with unlinked markers (r2= 0) [31,35]. Principle component analysis (PCA) was performed by the GAPIT procedure using R software to assess the population structure [31,36].The GWAS was performed using R software through the GAPIT software package based on the mixed linear model(PCA + K) [36]. The thresholds of the significant P-values were set to be 1.0e-3 for Panel_I, and 1.0e-4 for Panel_II and Panel_III. Haploview 4.2 was used to analyze haplotypes[37].

2.4.QTL mapping

The F10RIL population UP was genotyped using 1494 markers consisting of three types of markers(1228 DArT,251 SSR,and 15 ESTs)[38].Inclusive composite interval mapping(ICIM)was used to detect the QTL based on HR in each environment.LOD scores for declaring significant QTL were calculated from 1000 permutations at the P=0.05 level.QTL IciMapping V4.1(http://www.isbreeding.net/) was used to construct the genetic linkage map, and the threshold of LOD score for declaring the presence of the QTL was set as 2.5[39,40].

3. Results

3.1. Phenotypic distribution of herbicide resistance in Chinese wheat

In all surveyed wheat accessions, a majority of cultivars displayed herbicide resistance at a level of 1-2, but some cultivars were classified as level 2-4 (Fig. 1-D). The HR values of Panel_I, Panel_II, and Panel_III showed ranges of 0-2.75,0.33-2.83, and 0-3.33, with an average HR of 1.12, 1.30, and 1.36,respectively.In Panel_I,which was composed of Chinese landraces and historical cultivars, 34%, 56%, and 10% of the cultivars showed an HR of 0 to 1,1 to 2,and 2 to 3,respectively.In Panel_II,which was composed of current popular cultivars,21%,69%,and 9% of the cultivars showed an HR of 0 to 1,1 to 2,and 2 to 3, respectively. In Panel_III, which was composed of introgression lines, 31%, 37%, 24%, and 7% of cultivars showed an HR of 0 to 1, 1 to 2, 2 to 3, and 3 to 4, respectively. The results indicated that wheat accessions with an HR of 1 to 2 were predominant in Chinese landraces, and historical and current cultivars (Table 1).

Some cultivars showed strong herbicide resistance, such as Yunong 035, Jimai 20, Yumai 8, Aikang 58, and Xinong 979(Table 1). These could be used in Chinese wheat breeding programs. However, some widely planted cultivars displayed very weak herbicide resistance, and their HR reached 4 in certain environments. Therefore, precaution should be taken in applying carfentrazone-ethyl to these cultivars, which include Zhengmai 004, Yumai 9, Yunong 9901, Yunong 1325,and Yanzhan 4110, to avoid irreversible phytotoxicity. Further analysis indicated that the broad-sense heritability of herbicide resistance index in wheat cultivars of Panel_I, Panel_II and Panel_III were 74.8%, 67.1% and 78.5%, respectively.

3.2. Genome-wide association study

The population structures of Panel_I, Panel_II and Panel_III were shown in Sun et al. [31], Yang et al. [32] and Fig. S1. After quality control, a total of 41,561 SNPs in Panel_I, 395,782 SNPs in Panel_II, and 350,883 SNPs in Panel_III were used for the GWAS. The results showed that a total of 329 SNPs were significant and were distributed on 20 chromosomes with PVE of 11.3% to 27.6% (Table S1).

In Panel_I, 79 SNPs with PVE of 17.5% to 24.4% were significant on chromosomes 1B, 2A, 2B, 2D, 3A, 5B, and 5D (Fig.2-A). Among these, 12 were identified on chromosome 5B (471 to 473 Mb; Table S1) with averaged PVE of 21.31% (Table 2,Table S1). In Panel_II, 110 SNPs were significant and were mainly distributed on chromosomes 1B, 2D, 3A, 5B, 6B, 7A, 7B,and 7D (Fig. 2-B). Among these, 19 were found from 494 to 587 Mb on chromosome 5B with PVE of 11.7% to 14.5% (Table 2,Table S1). In Panel_III, 140 SNPs were significant and were distributed on chromosomes 1B, 2D, 3A, 5B, 6D, and 7D (Fig. 2-C). Among these, 20 SNPs with PVE of 22.4% to 23.9% were identified from 467 to 552 Mb on chromosome 5B. These results indicated that there was an important genetic locus at the interval between 467 and 587 Mb on chromosome 5B that controls herbicide resistance (Table 2, Table S1). Additionally,eight significant SNPs were detected in the range from 518 to 580 Mb on chromosome 1B with PVE of 18.7% to 26.1%, and six significant SNPs were detected from 46.2 to 52.4 Mb on chromosome 7D with PVE of 22.8% to 24.5% (Table 2, Table S1).

3.3. Haplotype analysis of the genetic loci related to herbicide resistance

Further analysis indicated that the 467-587 Mb interval on chromosome 5B has an important herbicide-resistant genetic locus (gHR-5B) in the three panels. This region contained 51 significant SNPs with PVE of 11.6% to 23.9%. Haplotype analysis in Panel_I showed two blocks in the gHR-5B interval;seven SNPs were clustered in a 626 kb block(PI_Hap_5B1a:AGGGCTTTTTAGAC; PI_Hap_5B1b: GGAGCCCTCTGGCC) and four SNPs were clustered in a 438 bp block(PI_Hap_5B2a: AAAATTAA; PI_Hap_5B2b: AGGGCCAC;Fig. 3-A1, Table 3). Comparison of HR values indicated that PI_Hap_5B1a (0.47 to 1.28) and PI_Hap_5B2a (0.86 to 1.32)showed significantly stronger herbicide resistance than PI_Hap_5B1b (0.96 to 1.47) and PI_Hap_5B2b (1.07 to 1.48;Table 3). In Panel_II, haplotype analysis also identified two blocks in the gHR-5B interval;17 SNPs were in a 1.042 Mb block(PII_Hap_5B1a: CCAACCTTCCCCAAGGCCAACC GGCCGGCCGGGG; PII_Hap_5B1b: TTGGTTCCTTTTGGCCTTGGCCTTCCGGCCGGCC) and 13 SNPs were in a 2.254 Mb block (PII_Hap_5B2a: CCGGCCCCGGCCCCAAAATTCCCCGG;PII_Hap_5B2b:TTCCTTTTTTTTAATTGGCCTTTTAA; Fig. 3-B1, Table 3). Comparison of HR values indicated that PII_Hap_5B1a (0.84-1.34) and PII_Hap_5B2a(0.9-1.49)showed significantly higher herbicide resistance than PII_Hap_5B1b (1.11 to 1.85) and PII_Hap_5B2b(1.06-1.83;Table 3).

Pyramiding analysis indicated that wheat cultivars with a combination of PI_Hap_5B1a and PI_Hap_5B2a showed a significantly lower HR (0.84) than cultivars with a combination of PI_Hap_5B1b and PI_Hap_5B2b (1.29) in Panel_I(Fig. 3-A2, A3). Cultivars with PII_Hap_5B1a and PII_Hap_5B2a showed a significantly lower HR (1.08) than those with PII_Hap_5B1b and PII_Hap_5B2b (1.94) in Panel_II(Fig. 3-B2, B3).

Another herbicide resistance locus in the interval of 461 to 524 Mb on chromosome 7D(gHR-7D)contained six significant SNPs (AX-111656163, AX-110539368, AX-108907541, AX-111533232, AX-109831221, and AX-110426671) with PVE of 22.8% to 24.5%. Haplotype analysis showed that there was a 6.221 Mb block showing two main haplotypes (PIII_Hap_7Da:CCAAAAGGTTTT and PIII_Hap_7Db:TTGGGGAACCCC).Cultivars with PIII_Hap_7Da showed higher herbicide resistance (HR: 0.56-0.96) than PIII_Hap_7Db (HR: 1.28-1.68; Fig. 3-C2,C3).

Fig.2- The typical Manhattan and Q-Q plots of the three panels of wheat accessions.A for Panel_I,B for Panel_II,and C for Panel_III.

3.4. QTL mapping in bi-parent population

The parents UC1110 and PI610750 of the RIL population UP had HR values of 3 and 1,respectively.QTL mapping in the UP population showed that one QTL on chromosome 1B for HR(qHR-1B) was detected between wPt-8279 and cfd48-1B with PVE of 7.44% to 8.28% in the two replicates(Fig.4-A,B).Wheat lines with qHR-1B_AA showed a significantly stronger HR(1.29-1.42)than those with qHR-1B_BB(1.73 to 1.83)in the two environments(Fig.4-C).The physical position analysis(Fig.4-B, Table 2) in the ‘Chinese Spring’ wheat genome indicated that qHR-1B was mapped to the 554 to 566 Mb region on chromosome 1B, and fell into the same region as those detected by GWAS (517-580 Mb) in Panel_III. This indicated that qHR-1B was an important genetic locus that controls herbicide resistance in common wheat.

Table 2-Summary of important herbicide resistance loci identified by genome-wide association study (GWAS) and quantitative trait loci(QTL)mapping.

4. Discussion

Fig.3-Haplotype analysis of single-nucleotide polymorphisms(SNPs)of gHR-5B and gHR-7D that were significant in multiple environments.(A1)Haplotype analysis in Panel_I.(A2)Pyramiding analysis of PI_Hap_5B1 and PI_Hap_5B2.(A3)Herbicide resistance comparison of combined PI_Hap_5B1 and PI_Hap_5B2.(B1)Haplotype analysis in Panel_II.(B2)Pyramiding analysis of PII_Hap_5B1 and PII_Hap_5B2.(B3)Herbicide resistance comparison of combined PII_Hap_5B1 and PII_Hap_5B2.(C1)Haplotype analysis in Panel_III. (C2,C3)Herbicide resistance of two alleles at PIII_Hap_7D locus.

Table 3-Herbicide resistance comparison of different haplotypes of gHR-5B and gHR-7D in three panels.

As one of the most important staple food crops all over the world, the stability of wheat production is crucial to national food security.In large-scale agricultural production processes,effective control of the weeds in the field could contribute to stable crop yields. However, the use of herbicides in recent years has become increasingly widespread, which has provided a favorable environment for the evolution and selection of herbicide-resistant weeds [41-44]. As a result, the field of crop variety improvement and molecular breeding requires new combinations of herbicides and herbicide resistance genes [45]. In this study, we identified some herbicideresistant wheat germplasm in a large number of wheat accessions from the Yellow and Huai River valleys and they could be directly or indirectly used in Chinese wheat breeding programs to avoid herbicide accident.We also identified three important genetic loci regulating herbicide resistance on chromosome 1B, 5B and 7D, using GWAS and linkage mapping, which could provide valuable information for improving wheat herbicide resistance by marker-assisted breeding in the future.

Fig.4-Quantitative trait loci(QTL)mapping(A)and superior allele analysis(B,C)for herbicide resistance on chromosome 1B,which was identified in the UC1110/PI610750(UP)population.PVE,phenotypic variance explained;ADD,additive effect;Ev1 and Ev2,environments 1 and 2 for UP at Suiping in 2017.

A total of 697 wheat accessions composed of historical cultivars, current popular cultivars, and introgression lines were used to identify phytotoxicity in common wheat from the Yellow and Huai River valleys and to screen for herbicideresistant germplasm. The results showed that wheat accessions with an HR value of level 1 to 2 were predominant and that some popular cultivars displayed serious phytotoxicity levels of 3 to 4 (HR). This suggests that the improvement of herbicide resistance is necessary in Chinese common wheat cultivars.Additionally,comparison of the HR values of wheat cultivars from different periods indicated that current wheat cultivars showed a weaker herbicide resistance than historical cultivars. Furthermore, our study also screened some cultivars,such as Aikang 58,Jiangshuang 2,Yunong 035,Yangmai 12, and Jimai 20, that were strongly resistant to herbicides.These cultivars should be utilized as backbone parents in herbicide resistance breeding programs. Some cultivars, such as Zhengmai 004, Yumai 9, Yunong 9901, Yunong 1325, and Yanzhan 4110,that are widely planted in the Yellow and Huai River valleys were highly susceptible to herbicides. More care should be taken when spraying these cultivars with carfentrazone-ethyl.

Research on the resistance mechanisms of weeds to ALSinhibitor herbicides has been mainly focused on the following aspects: the metabolism and detoxification processes of weeds in response to herbicides, absorption and conduction of herbicides, and genetic mutations of herbicide targets in weeds [46]. The substitution of amino acids at eight known sites in the ALS sequence can lead to different degrees of resistance in weeds to ALS-inhibitor herbicides (http://weedscience.org/summary/country.aspx). Most reports have aimed to identify the sites that confer resistance to carfentrazone-ethyl and tribenuron-methyl in different weeds [19-22]. However, few researchers have carried out investigations into herbicide resistance genes in crops.

Maeda et al. [25] recently discovered a herbicide-resistant gene HIS1, which has been successfully used for herbicide resistance in rice breeding. In maize, a regulatory gene that was sensitive to the herbicide, nicosulfuron, was co-located on maize chromosome 7 by two research groups [47,48]. A wheat QTL, tsl (tribenuron-methyl susceptible lethality) on chromosome arm 1DS, was detected in the F2:3population from the cross of BN044371 (tribenuron-methyl sensitive mutants found in wheat breeding) and BN044370(tribenuron-methyl insensitive parent) [49]. This locus was linked to Xgdm33, Xcfd58, and swes98 with the genetic distances 11.8 cM, 12.2 cM, and 12.6 cM, respectively [49]. In this study,we identified three important herbicide resistance loci gHR-5B, gHR-1B/qHR-1B, and gHR-7D. These loci could be considered as important genetic regulatory sites for identifying herbicide-resistant genes in common wheat and for developing molecular markers for marker-assisted selection in wheat breeding programs. With the rapid development of molecular biology in recent years, molecular markers associated with wheat trait loci not only accelerate the elucidation of related genes, but can also facilitate deployment of different genes into widely used cultivars, directionally improving some traits of cultivars[18,50-52].This may greatly improve the efficiency of breeding herbicide-resistant wheat in the future.

5. Conclusions

The herbicide resistance of 697 wheat cultivars from the Yellow and Huai River valleys of China was investigated and some herbicide-resistant wheat germplasm was identified in this study. Through the GWAS and QTL analysis, a total of 329 significant SNPs with PVE of 11.3% to 27.6% were identified,which amounted to three important genetic loci related to herbicide resistance. The gHR-5B locus was co-located in all three wheat populations studied, and gHR-7D was identified in multiple environments by GWAS. qHR-1B/gHR-1B was identified by both GWAS and QTL mapping. The important herbicide-resistant germplasm and herbicide resistance loci identified from this study may be used to accelerate gene mapping, to provide effective help for map-based cloning of herbicide resistance genes, and to reveal the underlying regulatory mechanism of herbicide resistance.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2020.02.004.

Declaration of competing interest

The authors declare that they have no conflict of interest.

Acknowledgments

This work was supported by the National Key Research and Development Program of China (2016YFD0101802), the National Natural Science Foundation of China (3181101544),Henan Major Science and Technology Projects(181100110200), and Ten-Thousand Talents Plan (Z04295) of China.

Author contributions

Feng Chen designed the study. Chaonan Shi, Yueting Zheng,Junyou Geng, Chunyi Liu, and He Pei phenotyped herbicide resistance in different environments. Chaonan Shi, Feng Chen, Zhongdong Dong, and Ning Zhang performed GWAS analysis.Chaonan Shi,Yan Ren,and Lei Zhao performed QTL mapping.Chaonan Shi and Feng Chen wrote the manuscript.All authors read and approved the final manuscript.