QTL Mapping for Grain Zinc and Iron Concentrations in Bread Wheat

Deficiency of micronutrient elements, such as zinc (Zn) and iron (Fe), is called “hidden hunger,” and bio-fortification is the most effective way to overcome the problem. In this study, a high-density Affymetrix 50K single-nucleotide polymorphism (SNP) array was used to map quantitative trait loci (QTL) for grain Zn (GZn) and grain Fe (GFe) concentrations in 254 recombinant inbred lines (RILs) from a cross Jingdong 8/Bainong AK58 in nine environments. There was a wide range of variation in GZn and GFe concentrations among the RILs, with the largest effect contributed by the line × environment interaction, followed by line and environmental effects. The broad sense heritabilities of GZn and GFe were 0.36 ± 0.03 and 0.39 ± 0.03, respectively. Seven QTL for GZn on chromosomes 1DS, 2AS, 3BS, 4DS, 6AS, 6DL, and 7BL accounted for 2.2–25.1% of the phenotypic variances, and four QTL for GFe on chromosomes 3BL, 4DS, 6AS, and 7BL explained 2.3–30.4% of the phenotypic variances. QTL on chromosomes 4DS, 6AS, and 7BL might have pleiotropic effects on both GZn and GFe that were validated on a germplasm panel. Closely linked SNP markers were converted to high-throughput KASP markers, providing valuable tools for selection of improved Zn and Fe bio-fortification in breeding.


INTRODUCTION
Wheat provides the starch, protein, and mineral nutrition needs for 35-40% of the world population (1). Mineral nutrition is crucial for a healthy diet. Over 17% of people suffer from malnutrition worldwide due to lack of mineral nutrition and more than 100,000 children under the age of five die from zinc (Zn) deficiency annually (2)(3)(4). The CIMMYT Harvest-Plus program initiated in the early 21st century aimed to address the "hidden hunger" issue by increasing micronutrient concentrations in staple food grains by plant breeding (5). Zn and Fe deficiency were identified as major causes of malnutrition, especially in underdeveloped regions where cereal grains make up most of the food (6).
Zn is a crucial cofactor in many enzymes and regulatory proteins, such as carbonic anhydrase, alkaline phosphatase, and DNA polymerase enzyme synthesis (7). Zn deficiency, first reported in 1961, affects the immune system, taste perception, site, and sexual function (4). Fe deficiency in humans most commonly leads to nutritional anemia in women and children (8). Therefore, it is very important to improve the nutritional quality of wheat by enhancing the Zn (GZn) and Fe (GFe) concentrations in grain (9,10).
Cultivar Jingdong 8, with high yield and resistance to stripe rust, leaf rust, and powdery mildew, was released in the early 1990s in the China Northern Winter Wheat Region. It was used widely as a parent in breeding and was verified to have high GZn and GFe levels across environments (13). Bainong AK58, a high yielding cultivar in the Southern Yellow-Huai Valley Winter Wheat Region, has wide adaptability and good resistance to stripe rust, powdery mildew, and lodging, but has lower GZn and GFe. The main goals of the present study were to (1) identify QTL for GZn and GFe in the Jingdong 8/Bainong AK58 RIL population using inclusive composite interval mapping, and (2) develop and validate breeder-friendly markers for markerassisted selection (MAS) for Zn and Fe biofortification in wheat breeding programs.

Plant Materials
Two hundred fifty-four F 6 RILs developed from Jingdong 8/Bainong AK58 cross were used for QTL mapping of GZn and GFe concentrations. A germplasm panel, including 145 cultivars/lines with a wide range of variation in GZn and GFe from the Chinese wheat germplasm bank (13), were used for validation of QTL for GZn and GFe identified in the RIL population.

Field Trials and Phenotyping
The field trials were conducted at the wheat breeding station of the Institute of Crop Sciences (ICS, CAAS) located at Gaoyi (37 •  Grain samples were hand-harvested and cleaned to avoid potential contamination of mineral elements. Micronutrient analysis of grain samples collected from the 2016-2017 cropping season was performed at the Institute of Quality Standards and Testing Technology for Agro-products of CAAS using inductively coupled plasma atomic emission spectrometry (ICP-AES, OPTIMA 3300 DV) after samples were digested in a microwave system with HNO 3 -H 2 O 2 solution (20). For grain samples from the 2017-2018 and 2018-2019 cropping seasons and the germplasm panel, a "bench-top, " nondestructive, energydispersive X-ray fluorescence spectrometry (EDXRF) instrument (model X-Supreme 8000, Oxford Instruments plc, Chengdu) was used to measure GZn and GFe, following the standard method for high-throughput screening of micronutrients in whole wheat grain (21).

Statistical Analysis
Analysis of variance (ANOVA) was performed by PROC MIXED with method type3 and all effects were treated as fixed in SAS 9.4 software (SAS Institute, Cary, NC). Variance and covariance components for genotype and genotype by environment interaction effects were estimated using PROC MIXED, assuming all effects as random. A similar model was also performed by PROC MIXED with genotype effect as fixed, while environment, replication nested in environment, and interactions involving environment as random, to estimate best linear unbiased estimate (BLUE). Broad-sense heritabilities (H b 2 ) on the basis of BLUE value were estimated using the following equation and standard errors were calculated following Holland et al. (22): where σ 2 g represents the variance of genotypes, σ 2 ge and σ 2 ε represent the variances of genotype × environment interaction and error, and e and r represent environments and number of replicates per environment, respectively. Phenotypic and genotypic correlations and their standard errors were estimated after Becker (23). Student's t test was performed by PROC TTEST.

SNP Genotyping and QTL Analysis
Genomic DNA extracted from fresh seedling leaves of RILs and parents by CTAB method (24) were used for genotyping by the wheat 50K SNP Array. The wheat 50K SNP Array was developed in collaboration by CAAS and Capital-Bio, Beijing, China (https://www.capitalbiotech.com/). Linkage analysis was performed with JoinMap v4.0 using the regression mapping algorithm (25). QTL analysis was performed by inclusive composite interval mapping with the ICIM-ADD function using QTL IciMapping v4.1 (http://www.isbreeding.net). Phenotypic values of RILs averaged from two replicates in each environment and BLUE value across nine environments were used for analyses. QTL detection was done using a logarithm of odds (LOD) threshold of 2.5. Pleiotropic QTL were analyzed using the module JZmapqtl of multi-trait composite interval mapping (MCIM) in Windows QTL Cartographer v2.5 (26). QTL pyramids were plotted using ggplot2 in R (27). Physical maps for the positional comparisons of GZn and GFe QTL with previous reports were exhibited using MapChart v2.3 (28).

Conversion of SNPs to KASP Markers
Kompetitive Allele Specific PCR (KASP) markers were developed from SNPs tightly linked with the targeted QTL, each including two competitive allele-specific forward primers and one common reverse primer. Each forward primer incorporated an additional tail sequence that corresponds to only one of the two universal fluorescence resonance energy transfers. Primers were designed from information in the PolyMarker website (http://polymaker. tgac.ac.uk/). PCR procedures and conditions followed Chandra et al. (29). Gel-free fluorescence signal scanning and allele separation were conducted by microplate reader (Multiscan Spectrum BioTek, Synegy/H1) with Klustercaller 2.24.0.11 software (LGC, Hoddesdon, UK) (30).

Phenotypic Evaluation
ANOVA showed that GZn and GFe were significantly influenced by lines, environments, and line by environment interaction effects, with line by environment interaction effects contributing the highest variation, followed by line and environment effects ( Table 1). The broad-sense heritabilities of GZn and GFe were 0.36 ± 0.03 and 0.39 ± 0.03, respectively. Jingdong 8 accumulated significantly higher GZn and GFe than Bainong AK58. Wide-ranging continuous variation among the RILs suggests polygenic inheritance (Table 2, Figure 1). Significant and positive correlations of GZn (r = 0.25-0.67, P < 0.01) and GFe (r = 0.26-0.70, P < 0.01) were observed across the nine environments (Table 3). Additionally, positive phenotypic and genotypic correlations between GZn and GFe (r = 0.78 ± 0.01 and 0.81 ± 0.03, P < 0.001) (Figure 2), indicated that GZn and GFe were, to some degree, simultaneously accumulated.

Linkage Map Construction
Among 54,680 SNP markers in the 50K SNP array, 20,060 were polymorphic after removal of markers that were monomorphic, absent in more than 20% of assays, and minor allele frequency was <30%. A high-density linkage map spanning 3423 cM and including all 21 chromosomes was constructed using 3328 representative SNP markers of each bin. The average chromosome length was 163 cM, ranging from 116.72 cM (1B) to 237.40 cM (5A) (Supplementary Table 2).  Table 3, and Figure 3), with all superior alleles coming from Jingdong 8. Among these QTL, three were identified for both GZn and GFe at the same or overlapping location on chromosomes 4DS, 6AS, and 7BL. Three chromosomal intervals were detected using MCIM including 4DS, 6AS, and 7BL, corresponding to co-localized QTL for GZn and GFe by ICIM-ADD ( Table 5). Two intervals on chromosomes 4DS and 6AS were detected in most environments for GZn and GFe, while the one on chromosome 7BL was found in most environments for GZn but only one environment for GFe.

QTL Pyramids and Validation
It indicated that superior alleles of pleiotropic QTL on 4DS, 6AS, and 7BL were all from Jingdong 8. Accumulation effect of the three co-localization QTL for GZn and GFe was calculated based on the closely linked markers. The average concentration of GZn increased from 37.79 to 44.43 mg/kg and that of GFe increased from 41.02 to 50.37 mg/kg, with lines containing zero to three favorable alleles (Supplementary Figure 1).
Flanking SNPs closely linked to the QTL on chromosomes 4DS and 7BL and a SNP near QTL region of 6AS were converted to KASP markers and validated in the germplasm panel (Tables 6,  7). Cultivars with the same superior allele as Jingdong 8 had significantly higher GZn and GFe than those with the inferior allele from Bainong AK58 for all QTL, except for QFe.caas-6AS. The difference between the superior and inferior allele of the QTL on chromosomes 4DS, 6AS, and 7BL was 1.7, 2.8, and 3.5 mg/kg for GZn and 1.4, 1.0, and 4.7 mg/kg for GFe, respectively ( Table 7).

Comparisons With Previous Reports
In this study, QTL for GZn and GFe were mapped on chromosomes 1D, 2A, 3B, 4D, 6A, 6D, and 7B, and on chromosomes 3B, 4D, 6A, and 7B, respectively. Previously identified QTL are summarized in Supplementary Table 1 and partly shown in Figure 3. In addition to consensus maps, the IWGSC RefSeq v1.0 Chinese Spring reference sequence (31) was used for comparisons of QTL identified in different studies.

QZn.caas-1DS
QZn        (39). A gene-specific KASP marker K-AX-86170701 was identified for Rht2 (40), and lines with allele from Bainong AK58 had significantly lower GZn and GFe than that with allele from Jingdong 8 (Supplementary Figure 2). Therefore, it was possible that the lower concentrations of Zn and Fe in Bainong AK58 was associated with the Rht2 allele.

Pleiotropic Effects of QTL
The co-localization QTL for GZn and GFe on chromosomes 4DS, 6AS, and 7BL might be pleiotropic QTL based on the same or overlapping region detected using MCIM, in agreement with the significant positive phenotypic and genotypic correlations (r = 0.78 ± 0.01 and 0.81 ± 0.03, P < 0.01) between GZn and GFe. Gorafi et al. (32) identified a significant and positive phenotypic correlation between GZn and GFe (r = 0.78) and a pleiotropic QTL on chromosome 5D; significant and positive correlations between GZn and GFe were also found in other studies (11,42). It has been reported that some transporters, chelators, and genes regulated GZn and GFe simultaneously in a high frequency (10,43). These findings indicated that Zn and Fe could be improved simultaneously in breeding programs targeting mineral biofortification.

Potential Implication in Wheat Breeding
MAS has been applied in crop breeding for more than two decades. Therefore, it would be effective for traits that were controlled by low numbers of major QTL (37). The phenotypic analysis on GZn and GFe was time-consuming and laborious, indicating that identification of molecular markers linked to GZn and GFe would be of interest for improvement of nutritional quality in wheat. In the present study, pleiotropic QTL on chromosomes 4D, 6A, and 7B were detected in multiple environments. SNP markers linked to some of these QTL were converted to KASP markers, and the QTL were verified in a germplasm panel, indicating potential application in wheat breeding programs.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article and/or Supplementary Material, further inquiries can be directed to the corresponding authors.

AUTHOR CONTRIBUTIONS
YoZ and ZH conceived the idea. YW and XXu conducted the experiments, analyzed the data, and prepared the manuscript. YeZ, YL, ZP, YT, and DX contributed to mapping population development, phenotyping and statistical analysis. XXi and YH assisted in writing and revising the manuscript. All authors contributed to the article and approved the submitted version.

FUNDING
This study was supported by the Agricultural Science and Technology Innovation Program of CAAS (CAAS-XTCX20190025-2, CAAS-ZDRW202002, and CAAS-S2021ZD04).