Feng Huang,Zhaoyan Chen, Dejie Du, Panfeng Guan,Lingling Chai,Weilong Guo, Zhaorong Hu, Mingming Xin, Huiru Peng, Yingyin Yao,Zhongfu Ni,*

aKey Laboratory of Crop Heterosis and Utilization (MOE)/State Key Laboratory for Agrobiotechnology/Key Laboratory of Crop Genetic Improvement,China Agricultural University,Beijing 100193,China

bNational Plant Gene Research Center,Beijing 100193,China

cZhoukou Academy of Agriculture Sciences,Zhoukou 466001,Henan,China

Keywords:GWAS QTL mapping STARP marker Triticum aestivum

A B S T R A C T Root hairs are fast growing, ephemeral tubular extensions of the root epidermis that aid nutrient and water uptake. The aim of the present study was to identify QTL for root hair length(RHL)using 227 F8 recombinant inbred lines(RILs)derived from a cross of Zhou 8425B(Z8425B)and Chinese Spring(CS),and to develop convenient molecular markers for markerassisted breeding in wheat. Analysis of variance of root hair length showed significant differences(P<0.01)among RILs.The genetic map for QTL analysis consisted of 3389 unique SNP markers. Using composite interval mapping, four major QTL (LOD > 2.5) for RHL were identified on chromosomes 1B (2), 2D and 6D and four putative QTL (2 ≤LOD ≤2.5) were detected on chromosomes 1A, 3A, 6B, and 7B, explaining 3.32%–6.52% of the phenotypic variance.The positive alleles for increased RHL of QTL on chromosomes 2D,6B and 6D(QRhl.cau-2D,qRhl.cau-6B,and QRhl.cau-6D)were contributed by Z8425B,and CS contributed positive QTL alleles on chromosomes 1A(qRhl.cau-1A),1B(QRhl.cau-1B.1 and QRhl.cau-1B.2),3A(qRhl.cau-3A) and 7B (qRhl.cau-7B). STARP markers were developed for QRhl.cau-1B.1, QRhl.cau-2D, QRhl.cau-6D,and qRhl.cau-7B.Haplotype and association analysis indicated that the positive allele of QRhl.cau-6D had been strongly selected in Chinese wheat breeding programs.Collectively,the identified QTL for root hair length are likely to be useful for marker-assisted selection.

1. Introduction

Root hairs are fast growing, ephemeral tubular extensions of the plant root epidermis, and play a key role in water and mineral nutrient uptake [1–7]; particularly for phosphorus acquisition [8,9]. Genotypes with longer root hairs were reported to enhance the uptake of phosphorus in barley(Hordeum vulgare L.), wheat (Triticum aestivum L.) and soybean(Glycine max L.Merr.),and preserved economically stable grain yield even in low phosphorus conditions [10–12]. A genetic study in wheat found some QTL controlling root hair length were co-localized with QTL for yield components [3]. Thus, a sound understanding of the molecular basis of root hair length could be important for sustainable, low-input agricultural ecosystems [2,3].

Root hair length in a variety of species is a complex quantitative trait that is both genetically determined and environmentally responsive [3,13–16]. QTL analysis provides an approach for understanding genetic mechanisms of complex quantitative traits and for development of molecular markers closely linked to the QTL for application in crop breeding [17]. Many QTL for root hair length were identified using segregating populations in common bean and maize[8,14].For example,Yan et al.[14]examined root hair and acid exudation traits in relation to phosphorus uptake in a common bean recombinant inbred line (RIL) population and identified five QTL for both root hair length and acid exudation. Zhu et al. [8] examined root hair length in a 169 maize RIL population under phosphorus-limited conditions and identified three QTL for RHL under high fertility,and one QTL for root hair length under low phosphorus.

Horn et al.[3]conducted a QTL analysis of root hair length in two DH wheat populations and identified four QTL on chromosomes 1A, 2A, 6A, and 2BL. Other workers showed that rhizosheath size is strongly correlated with root hair length and hence could be used as an indicator of root hair length [12,18–21]. A previous study identified six major QTL for rhizosheath size on wheat chromosomes 2B, 4D, 5A, 5B,6A,and 7A[18].Liu et al.[12]analyzed rhizosheath size using aneuploid lines of Chinese Spring(CS)and chromosomes 1A,1D and 5A appeared to harbor genes controlling root hair length. James et al. [19] identified five major QTL for rhizosheath size on chromosomes 1D, 3A, 3B, 6A, and 7B using an F6RIL population grown in acid soils. The QTL on chromosome 1D accounted for 34% of the genotypic variation[19].Our previous studies revealed that the orthologs of the Arabidopsis root hair genes ROOT HAIR DEFECTIVE SIXLIKE 2 (AtRSL2) and HAIR DEFECTIVE SIX-LIKE 4 (AtRSL4)contributed to variation in root hair length. Consistent with a crucial function of root hairs particularly in nutrient-poor environments, overexpression of TaRSL2 and TaRSL4 led to increased shoot fresh biomass under nutrient-poor conditions[22,23].

Here, we performed a QTL analysis of root hair length using 227 recombinant inbred lines(RILs)derived from a cross of Zhou8425B (Z8425B) and CS. A haplotype distribution analysis using STARP markers was performed to determine the frequencies of CS and Z8425B alleles in diverse panels of wheat accessions.

2. Materials and methods

2.1. Plant materials

The 227 F8RILs and parents were kindly provided by Professor Xianchun Xia,Institute of Crop Sciences,Chinese Academy of Agricultural Sciences(CAAS).Z8425B is an elite Chinese wheat germplasm bred by the Zhoukou Academy of Agricultural Science and CS is a well-known Chinese landrace[24,25].The RIL population was grown at Zhoukou under ideal conditions of fertility and plant protection during the 2015–2016 cropping season to ensure high quality seeds.Eighty-five derivatives of Z8425B kindly provided by Dr.Yonggui Xiao,Institute of Crop Sciences, CAAS; and 224 accessions from the Chinese wheat mini-core collection, 86 Chinese wheat varieties and 216 accessions from worldwide wheat core collections assembled by French workers were used to determine the frequencies of specific haplotypes [26].

2.2. Grain germination

Seeds with uniform size for each line were equally spaced on wet germination paper, which was then rolled up and placed in a water box. The boxes were placed in a greenhouse maintained at 24°C/16°C and relative humidity of 75%with a 16 h light/8 h darkness photoperiod for 48 h.

2.3. Growth method

Seedlings showing uniform germination were grown for 48 h on a 1.5% agar medium containing in a tissue culture room at a relative humidity of 75% and 16 h light (24 °C)/8 h darkness (16 °C) photoperiod. There were three biological replications of the parents and RILs and measurements were made on at least five seedlings from each replicate.

2.4. Imaging and root hair measurement

Seedling root hairs at 48 h after germination were visualized and recorded digitally with a fluorescence stereo microscope(SZX16; Olympus, Tokyo). The images were directly used to measure root hair length with Photoshop software. The longest root hairs within the mature zone were measured three times in each image.

2.5. Statistical analysis

SPSS software (v17.0) (SPSS, Chicago, USA) was used for analysis of variance (ANOVA) and descriptive statistics.Narrow-sense heritability was estimated as h2b=σ2g/(σ2g+σ-2e), where σ2gwas the genetic variance, σ2g= (MSgenotype–MSerror)/r, σ2ewas error variance, and r was the number of replications[3].

2.6. QTL analysis

A genetic map for the Z8425B/CS RIL population including 3389 unique SNP markers kindly provided by Professor Zhonghu He,Institute of Crop Sciences,CAAS.QTL analysis of RHL was based on mean root hair lengths of the three replications. The analysis was performed using composite interval mapping (CIM) in QTL Cartographer version 2.5[27]. Parameters were set as follows: model 6 with forward and backward regression, five markers as co-factors,window size of 10 cM, and walk speed of 1 cM [28–30].Empirical threshold LOD scores for CIM were calculated at P ≤0.05 [28–30]. QTL with LOD > 2.5 were regarded as major QTL and those with 2 ≤LOD ≤2.5 were regarded as putative QTL; they were designated “Q” and “q”, respectively [31].Confidence intervals of each QTL were estimated based on LOD ± 2 [28–30]. QTL were named according to McIntosh et al. [32].

2.7. STARP markers development

We converted the SNP markers tightly linked to corresponding QTL into STARP (Semi-thermal asymmetric reverse PCR)markers [30,33]. Each primer contained two AMAS (asymmetrically modified allele-specific) primers (STARP-F1and STARP-F2) and a corresponding reverse primer (STARP-R).STARP-F1was designed to amplify one of the parental alleles with a 10 bp insertion(ACGACTGCTG)at the 5′terminus,and STARP-F2was designed to amplify the other parental allele.The principle of nucleotide substitution followed the description by Long et al. [33]. The STARP primers were designed using the primer design tool Primer3 (http://bioinfo.ut.ee/primer3-0.4.0/)(Table S1).

The three primers were mixed in a ratio of 1:1:2 (STARPF1: STARP-F2: STARP-R) and diluted. A 10 μL PCR system including 5 μL 2× Taq PCR Star Mix,2 μL mixed primer, 2 μL DNA template(50–100 ng μL?1)and 1 μL H2O.PCR for marker assays were performed by following the procedures and conditions:94°C initial denaturation for 5 min,followed by 12 cycles of a 3-step touchdown PCR protocol starting at 94°C for 30 s, then annealing at 68 °C for 30 s with the annealing temperature being decreased by 1 °C per cycle and extension at 72 °C for 30 s. Touchdown PCR was followed by 30 cycles of 3 steps (94 °C for 30 s, annealing at 57°C for 30 s and extension at 72°C for 30 s),finally,holding at 72 °C for 10 min. The amplified PCR products were separated by 8%PAGE.

3. Results

3.1. Variation in RHL

The average RHL of CS and Z8425B were 2.00 mm (±0.01) and 1.95 mm(±0.029),respectively.ANOVA showed no significant difference in RHL between the parents (Fig. 1-a). However,there was wide variation among the RILs, ranging from 1.63–2.41 mm (Fig. 1-a, -b). The frequency distribution of RHL among RILs showed a normal distribution (Fig. 1-b) with considerable transgressive segregation. ANOVA revealed significant differences (P < 0.01) among RILs (Table 1). Narrow sense heritability was 0.95,indicating that the population was suitable for QTL analysis.

3.2. QTL analysis

Four major QTL (LOD > 2.5) controlling RHL were distributed on chromosomes 1B(2 QTL),2D,and 6D(Table 2,Fig.2-a).The adjacent QTL on chromosome 1B were designated QRhl.cau-1B.1 and QRhl.cau-1B.2. QRhl.cau-1B.1 had a LOD score of 3.07 and explained 5.02% of the phenotypic variance, with an additive increasing effect of 0.04 mm.The closest marker was IAAV4702. QRhl.cau-1B.2 explained 4.56% of the phenotypic variation, with a LOD score of 2.77. RAC875_c45678_1083 was the closest marker.Chinese Spring contributed the increasing alleles for both QTL. QRhl.cau-2D, had a LOD score of 3.96, an additive effect of ?0.04 mm, and explained 6.52% of the phenotypic variance, with the increasing effect allele from Z8425B. The closest marker was Ku_c9369_364. QRhl.cau-6D had a LOD score of 2.90, explained 4.99% of the phenotypic variance,and had an additive effect of ?0.03 mm.The closest marker was wsnp_BE445201D_Ta_1_1. The positive allele for increased RHL of QRhl.cau-6D was from Z8425B.

Four putative QTL (2 ≤ LOD ≤ 2.5) were located on chromosomes 1A, 3A, 6B, and 7B (Table 2, Fig. S1). qRhl.cau-1A and qRhl.cau-3A explained 3.68% and 3.57% of the phenotypic variance, respectively. qRhl.cau-6B, explained 3.32% of the RHL variance;the closest marker was BobWhite_c23416_168.qRhl.cau-7B with closest marker BobWhite_c39053_78 and LOD score of 2.09 had an additive increasing effect of 0.031 mm, explaining 3.41%of the phenotypic variation.The increasing allele of qRhl.cau-6B was from Z8425B;the other increasing alleles were from CS.

Fig.1–Phenotypic analysis of parents and RILs.(a)Root hairs of parents and RILs with the longest and shortest root hairs.Bar=1 mm.(b) Frequency distribution of RIL population for root hair length(RHL).CS,Chinese Spring;Z8425B,Zhou 8425B.

Source df SS MS F-value P-value Blocks 2 0.0068 0.0034 3.02 0.07 Genotypes 226 16.09 0.0712 56.88 <0.01 Error 452 0.5658 0.0013

3.3. User-friendly markers developed for major and putative QTL

To develop user-friendly markers of the QTL controlling RHL,we attempted to transfer SNP markers tightly linked to each QTL into STARP markers. Two, three, three and one SNP markers were converted into STARP markers for QRhl.cau-1B.1,QRhl.cau-2D, QRhl.cau-6D, and qRhl.cau-7B, respectively (Table S1).

To test the efficiency of the STARP markers, STARP-1B.1,STARP-2D.1, STARP-6D.1, and STARP-7B were selected to genotype the RILs (Fig. 2-b). The results of single marker analysis showed that all STARP markers were significantly(P<0.05)associated with RHL(Table S2).In addition,for QRhl.cau-1B.1 and qRhl.cau-7B, the average RHL of RILs with CS alleles was longer (P < 0.05) than that of RILs with Z8425B alleles (Fig. 3). On the contrary, RILs with Z8425B alleles had longer average RHL(P<0.05)than RILs with CS alleles for QRhl.cau-2D and QRhl.cau-6D(Fig.3).Thus,the STARP markers likely represented their corresponding QTL.

3.4. Haplotype analysis of QTL using STARP markers

To determine the potential implications of our findings on wheat breeding, the haplotype frequencies of 85 Z8425B derivatives were analyzed using the four STARP markers(Fig. 4-a, Table S3). The increasing alleles of QRhl.cau-1B.1(STARP-1B.1-CS) and qRhl.cau-7B (STARP-7B-CS) were in 10.59% and 21.17% of lines, respectively (Fig. 4-a). The frequency of increasing alleles for QRhl.cau-2D (STARP-2D.1-Z8425B) and QRhl.cau-6D (STARP-6D.1-Z8425B) were 25.88%and 75.29%, respectively (Fig. 4-a). Moreover, RHL data for a total of 60 Z8425B derivatives were used for association analysis of the four loci (Table S3). Derivatives with the Z8425B allele of QRhl.cau-6D had longer RHL (5.79%) than derivatives with the CS allele, but no significant difference was observed for the other three loci (Fig. 4-b). Thus, we decided to explore the frequency distributions of CS and Z8425B alleles for QRhl.cau-6D in three diverse wheat panels with the STARP marker (Fig. 4-c, Tables S4, S5, S6). The frequency of the increasing allele from Z8425B for QRhl.cau-6D in the 86 Chinese common wheat variety panel (70.93%)was much higher than that in Chinese mini-core(35.71%)and worldwide core (47.69%) collections (Fig. 4-c, Tables S4, S5,S6).

4. Discussion

4.1. Comparisons with previous reports

Root hairs play a key role in providing plants with water and mineral nutrients[1–3].Genetic studies have shown that root hair length in wheat is controlled by polygenes or QTL[3,12,18,19]. Mapping chromosomal locations that affect root hair length could help in exploiting the natural variation in breeding improved cultivars [3]. At least 18 QTL related to wheat root architecture (e.g. RHL, root angle, root number)have been identified on chromosomes 1A, 1D, 2A, 2B, 3A, 3B,4D, 5A, 5B, 6A, 7A, and 7B [3,34]. In this study, we used an F8RIL population derived from cross Z8425B × CS in a QTL analysis of root hair length.Four major QTL on chromosomes 1B(2), 2D and 6D and four putative QTL on chromosomes 1A,3A, 6B, and 7B were identified (Table 2, Fig. 2-a, Fig. S1). All eight QTL were different from QTL associated with RHL in the previous studies,suggesting that all were novel and therefore worthy of further investigation.

A recent report identified QTL controlling RHL on chromosomes 2A and 6A that were co-localized with previously described QTL for yield components,indicating that selection for root hair length might be beneficial in breeding [3]. Using the same RIL population, Gao et al. [24,35] performed QTL mapping for yield-related traits and physiological traits(ground cover,normalized difference in vegetation index,and canopy temperature depression). A total of 86 QTL for yieldrelated traits and 24 QTL for physiological traits were identified. Interestingly, QRhl.cau-2D was consistently correlated with a QTL associated with thousand kernel weight. QRhl.cau-1B.1 and/or QRhl.cau-1B.2 identified in the current study co-localized with QTL controlling spike number m?2and spike length reported by Gao et al.[24],indicating that these QTL might have pleiotropic effects on controlling RHL and yield-related traits.

Table 2–QTL for RHL in the RIL population of Z8425B/Chinese Spring.

Fig.2– QTL mapping of RHL in the Zhou8425B/Chinese Spring RIL population and PCR products of STARP markers in the RIL population.(a) Chromosomal locations of major QTL for root hair length(RHL).Pink horizontal bars represent QTL with increasing effect from CS and gray horizontal bars represent QTL with increasing effect from Z8425B.SNP markers converted into STARP markers are highlighted in red.(b) PCR products of STARP-1B.1,STARP-2D.1,and STARP-6D.1 in individual RILs.A and B, alleles from CS and Z8425B,respectively.

Fig.3– Comparisons of RHL between lines with CS and Zhou8425B(Z8425B)alleles in the RIL population for QRhl.cau-1B.1,QRhl.cau-2D,QRhl.cau-6D,and qRhl.cau-7B using STARP markers.* and***,significantly different at P< 0.05 and P <0.001,respectively(ANOVA analysis).

Fig.4– Haplotype and association analysis of QTL by using STARP markers.(a) Frequency distributions of CS and Z8425B alleles of STARP markers linked to four QTL in derivatives of Zhou8425B (n =85).(b) Comparisons of RHL between accessions with CS and Z8425B alleles in derivatives of Zhou8425B for the four QTL loci(n=60).**,significantly different at P<0.01;ns,no significant difference(ANOVA analysis).(c)Frequency distributions of CS and Z8425B alleles of QRhl.cau-6D in three wheat panels.Panel 1,Chinese wheat varieties(n =86); Panel 2,Chinese mini-core collection(n= 224);Panel 3,worldwide core collections(n=216). CS,Chinese Spring alleles;Z8425B,Zhou8425B alleles;NULL,data not available.

4.2. Implications of QTL for RHL in wheat breeding

Although there was no significant difference for RHL between the parental lines in the present study, considerable transgressive segregation was observed in the RIL population,ranging from 1.63 to 2.41 mm. Consistent with this, our QTL analysis demonstrated that the root hair length was controlled by polygenes,and both parents contributed increasing alleles.Alleles of three QTL(QRhl.cau-2D,qRhl.cau-6B,and QRhl.cau-6D) increasing RHL were provided by Z8425B, and alleles increasing RHL of QRhl.cau-1B.1, QRhl.cau-1B.2, qRhl.cau-1A,qRhl.cau-3A, and qRhl.cau-7B were from CS (Table 2). RIL139 with the longest RHL carried all seven alleles for increased RHL. By contrast, RIL241 with the shortest RHL carried all seven alleles for decreased RHL. Collectively, these data indicated that the new cultivars with long root hairs can be developed by pyramiding the increasing alleles of these QTL using marker-assisted selection.

To develop markers for marker assisted selection we converted SNP markers in four QTL to STARP markers. Single marker analysis confirmed that the alleles for increased RHL from both parents were significantly associated with higher RHL. Furthermore, an association analysis of Z8425B derivatives showed that accessions with the Z8425B QRhl.cau-6D allele had higher RHL(5.79%)than those with the CS allele.We then investigated the haplotype distribution of the two alleles in three diverse wheat panels using the STARP marker. The Z8425B allele for increased RHL occurred at high frequency(70.93%) in the Chinese variety panel relative to the other panels suggesting that the increasing allele (Z8425B allele) of QRhl.cau-6D for RHL had been selected by breeders.

Previous studies showed that identification of crop cultivars with more and longer root hairs substantially increased root surface area and could be useful for increasing soil nutrient uptake[10].Z8425B is an elite wheat germplasm characterized by semi-dwarf plant height, large spikes, high thousand grain weight and multiple disease resistances [24,25]. The ability of root hairs to take up water and nutrients is likely associated with surface area,which is likely related to root hair length and number. Considering the difficulty in measuring numbers of root hairs,our study mainly focused on root hair length,but we also noted that the root hair number of Z8425B was higher than that of CS (Fig. 1-a). Thus, we compared the numbers of root hairs in 85 Z8425B derivatives with the naked eye.Among them,only two elite cultivars(Zhengmai 103 and Luomai 22)showed many more root hairs than CS (Fig. S2). It seems that this superior trait in Z8425B has not been strongly selected in breeding,thus offering potential for wheat breeding.

Declaration of competing interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This work was supported by the National Key Research and Development Program of China (2017YFD0101004) and the National Natural Science Foundation of China(31991214).We thank Professor Xianchun Xia and Dr. Yonggui Xiao (both from the Institute of Crop Sciences, CAAS, Beijing) for providing seeds of the Z8425B/CS RIL population and the Z8425B derivatives,respectively.

Author contributions

Zhongfu Ni conceived the project; Feng Huang measured the RHL of RIL population; Feng Huang, Panfeng Guan and Dejie Du conducted the QTL analysis; Feng Huang, Zhaoyan Chen and Lingling Chai developed the STARP markers to genotype the diverse panels of wheat accessions; Weilong Guo,Zhaorong Hu, Mingming Xin, Huiru Peng and Yingyin Yao helped to revise the manuscript; Feng Huang, Zhaoyan Chen and Zhongfu Ni analyzed experimental results and wrote the manuscript.

Appendix A. Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2020.02.007.