Molecular mapping of the grain iron and zinc concentration, protein content and thousand kernel weight in wheat (Triticum aestivum L.)

Genomic regions responsible for accumulation of grain iron concentration (Fe), grain zinc concentration (Zn), grain protein content (PC) and thousand kernel weight (TKW) were investigated in 286 recombinant inbred lines (RILs) derived from a cross between an old Indian wheat variety WH542 and a synthetic derivative (Triticum dicoccon PI94624/Aegilops squarrosa [409]//BCN). RILs were grown in six environments and evaluated for Fe, Zn, PC, and TKW. The population showed the continuous distribution for all the four traits, that for pooled Fe and PC was near normal, whereas, for pooled Zn, RILs exhibited positively skewed distribution. A genetic map spanning 2155.3cM was constructed using microsatellite markers covering the 21 chromosomes and used for QTL analysis. 16 quantitative trait loci (QTL) were identified in this study. Four QTLs (QGFe.iari-2A, QGFe.iari-5A, QGFe.iari-7A and QGFe.iari-7B) for Fe, five QTLs (QGZn.iari-2A, QGZn.iari-4A, QGZn.iari-5A, QGZn.iari-7A and QGZn.iari-7B) for Zn, two QTLs (QGpc.iari-2A and QGpc.iari-3A) for PC, and five QTLs (QTkw.iari-1A, QTkw.iari-2A, QTkw.iari-2B, QTkw.iari-5B and QTkw.iari-7A) for TKW were identified. The QTLs together explained 20.0%, 32.0%, 24.1% and 32.3% phenotypic variation, respectively, for Fe, Zn, PC and TKW. QGpc.iari-2A was consistently expressed in all the six environments, whereas, QGFe.iari-7B and QGZn.iari-2A were identified in two environments each apart from pooled mean. QTkw.iari-2A and QTkw.iari-7A, respectively, were identified in four and three environments apart from pooled mean. A common region in the interval of Xgwm359-Xwmc407 on chromosome 2A was associated with Fe, Zn, and PC. One more QTL for TKW was identified on chromosome 2A but in a different chromosomal region (Xgwm382-Xgwm359). Two more regions on 5A (Xgwm126-Xgwm595) and 7A (Xbarc49-Xwmc525) were found to be associated with both Fe and Zn. A QTL for TKW was identified (Xwmc525-Xbarc222) in a different chromosomal region on the same chromosome (7A). This reflects at least a partly common genetic basis for the four traits. It is concluded that fine mapping of the regions of the three chromosomes of A genome involved in determining the accumulation of Fe, Zn, PC, and TKW in this mapping population may be rewarding.

Introduction the breeding process, identifying loci involved in grain accumulation trait per se can also be an effective approach. For this, an extensive exploration of plant genetic resources for identifying those with higher mineral accumulation in seeds is indispensable; critical also is the understanding of the effect of environment on traits expression and so also its interaction with genotype [16][17]. This would aid in identifying loci or genomic regions (QTLs) harbouring genes for the accumulation of high Fe and Zn levels in the wheat grain and would allow breeders more efficiently to develop biofortified cultivars by using closely linked molecular markers to screen and select the most favorable genotypes. This indirect marker-aided selection (MAS) is more important for difficult to breed traits like grain Fe and Zn and PC, which exhibit high genotype-environment interaction [18][19][20][21][22][23][24].
The type and the size of a mapping population and the frequency and distribution of molecular markers on framework linkage map are considered to be important determinants for identification of a QTL. However, in most studies on identification of QTLs for Fe [28][29][30][31][32][33][34][35], Zn [28][29][30][32][33][34][35][36][37], PC [28,[32][33][38][39], and TKW [40][41][42][43][44][45] the size of mapping population was limited. The present study was aimed to identify the genomic region(s) associated with grain Fe, Zn, PC, and TKW in a set of the 286 RILs. The third location, Pusa Bihar at ICAR-IARI, Regional Station, Research Farm, Samastipur, 25˚14'N, 87˚2'E, 62.5 m AMSL belonged to the NEPZ. The crop was timely sown under irrigated conditions in first fortnight of November at all the locations in both the years. The genotypes were planted in a randomized complete block design (RCBD) with two replications per entry and two rows (5 m length) per replication with a plant-to-plant spacing of 10 cm and row-to-row spacing of 25 cm. Standard agronomic practices were followed for raising the crop.

Micronutrient analysis
The mapping population consisted of 286 RILs and two parents. The entire set was grown in two field replications. After harvest of 25-30 spikes from each replicate, the spikes of a replicate were bulk threshed in a clean cloth bag by beating with a wooden stick and the grain was separated from husk in a plastic chaaj. Care was taken at every step to avoid dust and metal contamination and sub sampling was done; finally three samples were drawn. Grain iron and zinc in each of these samples was estimated and the averaged value was considered the value of one replication. The grain samples for Fe and Zn were measured using Energy Dispersive X-ray Fluorescence (ED-XRF). The XRF machine model X-Supreme 8000 (M/s Oxford Inc, USA) was used for the purpose. This machine was calibrated using the glass beads based values. The glass beads contain standards with different concentrations of different minerals (which are earlier measured with ICP-MS by Dr. James Stangoulis; Technical coordinator, HarvestPlus at Flinders University, Australia). The glass beads were provided under the HarvestPlus Programme to AMS associated as a NARS partner. The same method has been used by others recently [21,28,30]. Thus, 3456 data points were generated for iron and zinc.

PC and TKW
The PC in the grain samples as determined by the Kjeldahl method using the Autokjeltech system 3100 from Foss, Tecator, USA. 20 gm grain sample was ground using cyclotech mill (Foss Tecator, Sweden). 10 ml concentrated H 2 SO 4 and 4.5 g of catalyst mixture was added to the 0.5 g of grain sample. The samples were digested at 450˚C for 30-45 minutes till contents were clear. After cooling, 70 ml distilled water was added to the digestion tubes, the digested contents were distilled with the distillation system. The PC was obtained by multiplying with a factor of 5.7 and the values were expressed at 12% moisture basis. The moisture content of each of the samples was also simultaneously obtained from Single Kernel Characterization System (Perten, Australia). Each entry was evaluated twice. To record TKW, reading was set at 1000 grains in the Numigral grain Counter and the weight of the grains was recorded with an electronic balance.
Linkage map construction. A total of 714 SSR markers were used for parental polymorphism survey [47][48]. These were chosen such as to cover all chromosome arms of bread wheat genome. The bands generated by SSR primers were scored by giving code 'A' for those RIL populations showing the same band as that of WH542 and 'B' for those showing the same band size as of synthetic derivative in the case of polymorphic SSRs. Allelic segregation at each of the marker loci was analyzed for segregation distortion from the expected 1:1 ratio in the RIL population using the goodness of fit (χ 2 ) test. The markers which exhibited 1:1 segregation pattern were only further used for linkage map construction. Linkage analysis was performed using MapMaker/EXP 3.0 [49-50]. Markers were first grouped using a cut-off recombination value of 0.39 and threshold LOD score of 3.0. Then, the Order command was used to determine a linear order. The Try and Ripple commands were used to add markers to the map and re-test the final marker order. The Kosambi mapping function was used to convert recombination frequencies into map distances.

Data analysis and QTL mapping
Pearson correlation coefficient was calculated for all the six environments to study the association of Fe, Zn, PC, and TKW in the grains of RILs population. Student's t-test was used to test the significance of the correlation coefficient. The QTL analysis was performed using adjusted means of the phenotypic trait value and genotyping data using QTL cartographer 2.1 software employing the full-QTL model. The adjusted mean trait phenotypic values were used as input data to detect QTLs. The composite interval mapping method [51-52] was used for identification of QTLs. The test window size was set to 5cM with a walk speed of 2cM and a cut-off probability of 0.05 for deciding the significance of the QTL. LOD threshold for declaring a QTL significant was determined by permutation tests using 1,000 reiterations.

Phenotypic variation and association studies
The Mean values of Fe, Zn, PC, and TKW are presented in Table 1. Individual mean values of RILs for Fe, Zn, PC, and TKW tested in six environments are given in supplementary tables as S1 Table, S2 Table, S3 Table and S4 Table, respectively. The parents were contrasting for all the four traits. Fe, Zn, PC and TKW, averaged over the replications, years and locations were 31.68mg/kg, 35.07mg/kg, 13.77%, and 28.76gm respectively, for WH542 (henceforth called the 'low parent'or P1), whereas, the averages of Fe, Zn, PC, and TKW were 45.37mg/kg, 44.39mg/ kg, 17.46%, and 34.47gm, respectively, for the synthetic derivative (henceforth called the 'high parent' or P2). A wide range of variation was observed among the 286 RILs. The population showed the continuous distribution for Fe, Zn, PC and TKW (Figs 1-3), that for pooled Fe and PC was near normal, whereas, for pooled Zn, RILs exhibited positively skewed distribution. Transgressive segregants appeared for all the four traits and which surpasses the parents and check variety (HD2967). Pearson's correlation coefficient (r 2 ) of Fe, Zn, PC and TKW in the RIL population was determined and are presented in Table 2. The associations were highly significant and positive in most of the environments.

Molecular mapping
A linkage map was constructed using 136 polymorphic SSR data of RILs. A length of 2155.3cM was covered across the 21 chromosomes. Phenotypic data obtained from all the six environments for Fe, Zn, PC, and TKW were analyzed together with genotypic data for all mapped loci assigned to the genetic maps using QTL Cartographer, version 2.1. For each of the QTLs, the closest/nearest marker was considered as the one on the linkage map closest to the peak of the QTL. A total of 16 QTLs located on chromosomes 1A, 2A, 2B, 3A, 4A, 5A, 5B, 7A, and 7B were identified in different environments (Fig 4 and Table 3). Four QTLs were identified for Fe with LOD scores ranging between 2.5 and 4.1, explaining 2.3-6.8% phenotypic variation. The four QTLs together explained 20.0% phenotypic variation for Fe, when the largest contribution of a QTL in any environment was considered. The identified QTLs for Fe were designated as QGFe.iari-2A, QGFe.iari-7B, QGFe.iari-7A and QGFe.iari-5A. QGFe.iari-2A was mapped in Pusa Bihar 2012-13 environment and the nearest marker was Xgwm359. QGFe.iari-7B, which was identified in two environments viz., Pusa Bihar 2012-13 and 2013-14, as well as in pooled mean and the nearest linked marker was Xgwm577. QGFe.iari-5A was mapped in pooled mean of Fe, the nearest marker identified was Xbarc144. QGFe.iari-7A was mapped in Pusa Bihar 2013-14 environment and the closely linked marker and flanking markers, respectively, were Xcfa2019 and Xbarc49-Xwmc525.
Five QTLs were identified for Zn in this study and these were designated as QGZn.iari-2A, QGZn.iari-4A, QGZn.iari-5A, QGZn.iari-7A and QGZn.iari-7B. The QTLs explained 3.2-14.4% phenotypic variation with LOD scores ranging between 2.4 and 13.5 and QTLs together explained 32.0% phenotypic variation for Zn. Xgwm249 was the nearest marker linked to QGZn.iari-2A in two environments viz., ICAR-IARI 2012-13 and 2013-14. QGZn.iari-2A was also mapped for pooled mean and the marker nearest to the QTL was Xgwm359. QGZn.iari-4A was identified only in Pusa Bihar 2012-13 environment and the nearest marker was Xbarc78. In the GBPUA&T 2012-13 environment, two QTLs were identified (QGZn.iari-5A and QGZn.iari-7A) and the closely linked markers were, respectively, Xbarc144 and Xcfa2019.       GBPUA&T 2012-13, Pusa Bihar 2012-13 and 2013-14 apart from pooled mean. The nearest linked marker and flanking markers, respectively, were Xbarc222 and Xwmc525-Xbarc222. QTkw.iari-1A was mapped in GBPUA&T 2012-13 environment and the closely linked marker and flanking markers, respectively, were Xwmc9 and Xbarc209-Xwmc9. Xwmc551 was the nearest linked marker for QTkw.iari-2B that was identified in GBPUA&T 2013-14 environment within the marker interval of Xbarc266-Xwmc551. In the Pusa Bihar 2012-13 environment, one QTL was identified (QTkw.iari-5B) and the closely linked marker and marker interval, respectively, were Xwmc746 and Xgwm499-Xwmc746. A low TKW can result from shriveled grains where the bran fraction may be higher. Since bran is richer in micronutrients as compared to the starchy endosperm, it is important to know if the parent that contributed favorable alleles for high for grain micronutrients was also the parent for contributing favorable alleles for TKW.
All the QTLs identified in the present study had favorable alleles (allele that increases Fe, Zn, PC and TKW) contributed by the high parent with the exception of QGZn.iari-4A and QTkw.iari-5B, where favorable allele was contributed by the low parent WH542 (Table 3).
A common region between Xgwm359-Xwmc407 on chromosome 2A was associated with the increase of Fe, Zn and PC and the identified QTLs were co-located. Some of the abovementioned studies have also identified the common region(s) associated with both Fe and Zn and among Fe, Zn and PC [26][27]. QTLs for all the studied traits (Fe, Zn, Protein, and TKW) on chromosome 2A and for three traits (Fe, Zn, and TKW) on chromosome 7A were identified. Highest phenotypic variation associated with the QTL as expressed by R 2 was also reported in 2A. Therefore, common regions on chromosome 2A and 7A harbours the QTLs for all the studied traits and could be prominent candidate chromosome regions for traits improvement. Though some of the chromosomes and the genomes are common among different studies, the latter differ with respect to the genomic regions harbouring the QTLs. Apart from soil variations in terms of nutrient distribution and their availability in the soil prompting the plants to use different mechanisms for the uptake of nutrients, the level of contamination that may result in sampling and actual experimentation may also cause shifts in the chromosomal locations of the QTLs [61].
While twelve QTLs for the four traits were mapped on A genome, only four were detected on B genome and none on the D genome in this study. The same order of the frequency of involvement of genomes was observed in earlier studies also.

Conclusion
The study with 286 RILs developed by single seed descent from the cross between a cultivated bread wheat variety and a synthetic derivative has shown that Fe, Zn, PC, and TKW are quantitatively inherited traits and their expression is greatly influenced by environment. The strong positive association among Fe, Zn, PC, and TKW suggest that improving one of the traits would allow the others to be improved simultaneously. Four, five, two and five QTLs were identified, respectively, for Fe, Zn, PC, and TKW. The QTLs identified in the present investigation represent novel genomic regions associated with Fe, Zn, PC, and TKW. The presence of a common locus on chromosome 2A associated with Fe, Zn and PC, and one QTL identified at different chromosomal region for TKW on the same 2A chromosome, as well as of the common loci on chromosome 5A and 7A associated with both Fe and Zn, along with one QTL identified at different chromosomal region for TKW on the same 7A chromosome imply that at least three chromosomes of A genome are important in determining accumulation of Fe, Zn, PC and TKW and, therefore, fine mapping of the involved regions in these chromosomes may be rewarding.
Supporting information S1