Abstract

Combating hidden hunger through molecular breeding of nutritionally enriched crops requires a better understanding of micronutrient accumulation. We studied natural variation in grain micronutrient accumulation in barley (Hordeum vulgare L.) and searched for candidate genes by assessing marker-trait associations (MTAs) and by analyzing transcriptional differences between low and high zinc (Zn) accumulating cultivars during grain filling. A collection of 180 barley lines was grown in three different environments. Our results show a pronounced variation in Zn accumulation, which was under strong genotype influence across different environments. Genome-wide association mapping revealed 13 shared MTAs. Across three environments, the most significantly associated marker was on chromosome 2H at 82.8 cM and in close vicinity to two yellow stripe like (YSL) genes. A subset of two pairs of lines with contrasting Zn accumulation was chosen for detailed analysis. Whole ears and flag leaves were analyzed 15 days after pollination to detect transcriptional differences associated with elevated Zn concentrations in the grain. A putative α-amylase/trypsin inhibitor CMb precursor was decidedly higher expressed in high Zn cultivars in whole ears in all comparisons. Additionally, a gene similar to barley metal tolerance protein 5 (MTP5) was found to be a potential candidate gene.

Introduction

Hidden hunger, synonymously used for micronutrient malnutrition, represents a major global health burden. Next to iron (Fe), zinc (Zn) is one of the most important undersupplied micronutrients [13]. Children are particularly affected by Zn deficiency and suffer from developmental defects, mental retardation and an increased susceptibility to infectious diseases [3,4]. Monotonous diets based on cereals can cause micronutrient deficiency, which is thus especially prevalent in low- and middle-income countries [3]. There, ∼70% of the consumed food is based on seeds, which are often lacking sufficient amounts of bioavailable micronutrients [5]. Elevated atmospheric levels of carbon dioxide (CO2) are predicted to cause a decrease in grain Zn and Fe concentrations, which would further aggravate the problem [6,7]. Hence, besides food supplementation, approaches to enrich micronutrients in edible parts of plants are pursued to tackle the problems of micronutrient deficiency [810]. This includes biofortification through foliar application of micronutrient fertilizers and breeding approaches [1113]. For the latter, the understanding of molecular mechanisms is essential in order to enable genetic engineering and directed, conventional breeding, e.g. the Selection with Markers and Advanced Reproductive Technologies (SMART) breeding. Thus, there is a strong interest to elucidate mechanisms of micronutrient mobilization and transport into and within the plant, and of processes related to nutrient storage within the edible parts of the plant [14,15].

A model organism for studying these mechanisms in cereals is barley (Hordeum vulgare L.). Barley is an economically important crop, is diploid, self-fertile and diverse germplasm collections are available [1619]. The ICARDA (International Center for Agricultural Research in the Dry Areas) [20] germplasm panel represents the world's barley diversity and was genotyped using Diversity Arrays Technology (DArT™) markers [21,22]. In addition, the barley genome was first published by the International Barley Sequencing Consortium in 2012 and, recently, followed by a high-quality reference genome assembly [18,23]. The genetic information can be linked to phenotypic traits in a genome-wide association study (GWAS) mapping approach to identify candidate genetic loci possibly contributing to variation in, e.g. Zn accumulation. A basic requirement for the identification of quantitative trait loci (QTLs) is the variation of the desired trait. Zn accumulation showed considerable variation in barley grains between the genotypes of a mapping panel [24,25]. Furthermore, transcriptomic approaches may elucidate differences leading to higher Zn concentrations and can help to identify mechanisms involved in micronutrient accumulation in the grain [26,27]. For example in barley, an analysis of transcripts of cv. ‘Golden Promise’ 20 days after pollination (dap) derived a potential roadmap of Zn trafficking in different tissues [28].

In this study, we aimed to elucidate the molecular mechanisms leading to higher Zn accumulation in barley grains by detecting QTLs and transcriptional differences of low and high Zn accumulating cultivars. The first objective was to perform a phenotypic characterization of Zn accumulation in barley grains and second, to identify potential candidate genes. Grains of 180 lines of the ICARDA germplasm collection grown in three different environments were analyzed for their Zn concentrations. To understand the relationship and overall transport or storage mechanisms for Zn, the concentrations of Fe, manganese (Mn), copper (Cu), carbon (C [%]) and total protein were measured in parallel. Because the micronutrients Fe, Mn and Cu at least partially share distribution pathways with Zn, correlations can help to build hypotheses about mechanisms involved in micronutrient accumulation in the grain. Carbon or total protein contents can indicate major ligand environments of Zn. To identify significant marker-trait associations (MTAs) and genomic regions responsible for micronutrient variations, all nutrients were analyzed by GWAS using 703 DArT markers. Third, to identify general mechanisms leading to high Zn accumulation, we selected two lines each that accumulated consistently lower or higher Zn concentrations in grains across the different environments. Here, we analyzed first the spatial distribution of Zn in different grain tissues to identify potential differences of low and high Zn accumulating lines. Then, the four lines were cultivated and a comparative transcript analysis conducted on whole ears and flag leaves in order to identify putative mechanisms leading to high Zn accumulation.

Materials and methods

Plant material and growth in different environments

Barley (H. vulgare L.) grains of the ICARDA germplasm panel were supplied by the Federal ex situ Gene Bank at IPK Gatersleben (Germany) [21,22].

To determine variation of Zn accumulation in grains, 180 barley lines (Supplementary Table S1) were cultivated at three different sites [Folientunnel (FT), Selkeweg (SE), Kirschweg (KW)] at IPK. First site: five replicates of each line were multiplied in a randomized block design in the foil tunnel (FT). In March 2007, grains were sown in pots using standard culture medium (70% compost soil, 20% white peat, 10% sand) with 8.6 mg Zn kg−1, 43 mg Mn kg−1, 1.4 mg Cu kg−1 and a pH value of 6.9. In July 2007, grains were harvested and partly used for field experiments in 2008 and partly stored at 20°C and 50% relative humidity for further usage. Second and third site: field trials were conducted at SE and at KW. At SE soil quality was classified as clayish loam (toniger Lehm) with 4.3 mg Zn kg−1, 12 mg Mn kg−1, 1.2 mg Cu kg−1 at a pH of 7.6. At KW, soil was determined as humic loam with 16.6 mg Zn kg−1, 145 mg Mn kg−1, 5.9 mg Cu kg−1, 4.0% organic matter at a pH of 7.3. Detailed results of soil analysis were provided by AGROLAB Oberdorla (www.agrolab.de) and are shown in Supplementary Table S2. In February 2008, at both sites, SE and KW, 100 grains for each line were sown in field plots of 1 m × 1 m, harvested in July 2008 and stored at 20°C and 50% relative humidity for further usage. Micronutrient concentrations were analyzed in grains of 150 (KW), 161 (SE) or 162 (FT) lines. The missing lines failed to establish during cultivation.

Plant material and growth conditions for microarray analysis and grain dissection

Grains of the selected lines 140, 154, 143 and 156 were heat treated at 43°C for 20 min to minimize the risk of infection and to prevent spraying pesticides during cultivation, which could affect gene expression. In three different cultivation rounds, plants were grown in pots in the glasshouse or in a growth chamber under long day conditions (16h:8 h, light:dark) at 100–150 µmol s−1 m−2.

The soil was a mixture of soil type GS90L (Einheitserde Werkverband e.V.), pricking soil (Ökohum GmbH) and vermiculite (Deutsche Vermiculite Dämmstoff GmbH) in a ratio of 3:3:1. Before use, the moistened soil was heat pasteurized for 1.5 h at 80°C using a Sterilo1K (Harter Elektrotechnik), to reduce the risk of soil-borne diseases. Soil total Zn concentration was 89.7 mg kg−1 dry weight (DW), which is representative for European topsoils according to the FOREGS Geochemical Atlas of Europe (http://weppi.gtk.fi/publ/foregsatlas/). For all lines, three (growth chamber) to five (glasshouse) plants were cultivated in 3 l (growth chamber) or 4 l (glasshouse) pots and pooled for analysis. Plants were fertilized once a week with 200 ml 0.2% Wuxal Super (Wilhelm Haug GmbH & Co. KG) per pot from BBCH 35 on until ripening (BBCH 89). Only healthy plants were sampled. For grain dissection, whole, mature grains were prepared as described by Detterbeck et al. [24]. In brief, grains were soaked in bidistilled water at 4°C for 4 h before dissecting husk (including bran, epidermal cell layers and testa), endosperm (including parts of the aleurone layer) and embryo tissues under a stereomicroscope.

For microarray analysis, whole ears and flag leaves were cut 15 dap. The sampling was always done at the same time of the day to eliminate the influence of a circadian rhythm or differences in light conditions. The developmental stage can be determined exactly between 3 and 10 dap and was evaluated for grains positioned at the lower third of the ear on the basis of caryopsis development [29]. Ears were marked at early developmental stages and harvested 15 dap. Excluding awns and poor grains, whole ears and flag leaves including leaf blade and sheath were harvested. After harvest, the samples were immediately frozen in liquid nitrogen. Whole ears and flag leaves were separately ground in liquid nitrogen using porcelain pestles and mortars and stored as powder at −80°C for further analysis.

Determination of metal concentration

Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES) was used to analyze total metal concentration in grains and in the soil. Oven-dried grains were wet-digested in a microwave (START 1500; MLS GmbH) using 2 ml bidistilled water, 2 ml 30% H2O2 (w/w) and 4 ml 65% HNO3 (w/w) at 180°C for 12 min. Dissected endosperm tissue was digested in the same way. Dissected husk and embryo tissues were microwave digested with only 1 ml bidistilled water, 1 ml 30% H2O2 (w/w) and 2 ml 65% HNO3 (w/w) in smaller vessels.

Total metal concentration of soils used for plant growth was determined by adding 2.25 ml conc. HCl (37%) and 0.75 ml conc. HNO3 (65%) to 0.25 g of dried soil. Formed gasses were allowed to dissipate for 30 min and, afterwards, vessels were closed and microwave digested (START 1500; MLS GmbH) with ramping 15 min to 160°C and holding this temperature for 15 min. The final volume was adjusted to 10 ml using bidistilled water.

Metal concentrations were measured using an iCAP 6500 (Thermo Scientific) at wavelengths of 213.8 nm (Zn), 238.2 nm (Fe), 257.6 nm (Mn) and 324.7 nm (Cu).

Determination of total protein and C [%] content

Carbon and nitrogen concentrations in dried grain tissue (∼40 mg) were analyzed by Dumas combustion using a Vario Macro elemental analyzer (Elementar Analysensysteme GmbH, Hanau, Germany). Prior to weighing into tin capsules and analysis, all samples were oven dried at 60°C for a minimum of 2 h. Data quality was evaluated by the analysis of standard reference materials (acetanilide and B2166, National Institute of Standards and Technology, Gaithersburg, MD, U.S.A.). Grain protein concentration was calculated as grain N concentration multiplied by 5.4 [30].

Population structure and GWAS

DArT™ markers and their chromosomal positions were obtained from Triticarte Pty Ltd, Yarralumla, ACT, Australia, and comprised 703 markers with average distances of 1.6 cM and a minor allele frequency of 16.6%.

The population structure was analyzed by the STRUCTURE 2.3.4 software [31] using the full marker set. By applying the admixture model, a burn-in of 5000 iterations and a 5000 Markov Chain Monte Carlo duration, K-values were tested in the range from one to 10 using 10 replicates each. The likely number of sub-populations (Q) present was estimated by a phylogenetic tree based on the DArT marker and unweighted pair group method with arithmetic average (UPGMA) using TASSEL 2.1 software [32]. A phylogenetic tree was drawn using The Tree Drawing Tool ‘FigTree’ version 1.4.3 (http://tree.bio.ed.ac.uk/).

For association analysis, phenotypic data were analyzed using REML (Residual Maximum Likelihood) implemented in GenStat 18 software [33]. The best linear unbiased estimates (BLUEs) of all the genotypes were derived by assuming fixed genotypic and random environment effects and taking into account the genotype × environment (G×E) variance. The broad-sense heritability (H2) parameter was calculated using the following equation: 
formula
Thereby, σ2G is the variance of genotypes (G); σ2GE is the variance component of the interaction G × environment (E), σ2RGE is the variance component of the interaction replicate (R) × G × E, σ2RGET is the variance component of the interaction R × G × E × treatment (T) and r is the number of replicates.

Analysis of variance (ANOVA) was applied to genotypes, environments, treatments and the least significance differences at P ≤ 0.05 (LSD5%) used for discrimination (VSN International Ltd, U.K.).

A mixed linear model was used to calculate single MTAs between 703 genetically mapped DArT™ markers and the estimated phenotypic data (BLUEs). The population structure was corrected by the kinship and associations were regarded as significant when threshold exceeded −log(PM) ≥ 2.5. Non-DArT markers within 0.5 and 2.5 cM on either side of MTA loci were in range of the significant LD at P ≤ 0.001 and at P ≤ 0.05, respectively, and were used to identify putative functions of the associated markers available at the IPK barley BLAST server, Gatersleben (http://webblast.ipk-gatersleben.de/barley_ibsc/).

RNA extraction and cDNA synthesis

High RNA quality was achieved by a combination of TRIzol Reagent (Thermo Fisher Scientific Life Technologies GmbH) extraction and RNeasy Mini Kit (Qiagen N.V.; see Supplementary Figure S1), which prevented the clogging of the silica membrane with starchy endosperm in the columns provided in the RNeasy Mini Kit. One hundred milligram powder from whole ears of flag leaves was mixed with 1 ml TRIzol for 3 min and allowed to set at room temperature for 5 min. To remove remaining plant material including starch, samples were centrifuged at 12 000×g and 4°C for 10 min. The supernatant was mixed with 0.2 ml chloroform and 1 ml TRIzol reagent for 15 s. Again, the mixture was centrifuged at 12 000×g and 4°C for 10 min. From this step, the RNeasy Mini protocol described for plants and fungi was applied and the final supernatant was mixed with 1 ml RLC buffer and inverted. Afterwards 0.5 volume ethanol was added to the aqueous phase. Quality and integrity of RNA were checked by agarose gel and NanoPhotometer (Implen GmbH) measurements. RNA samples not meeting purity greater than 1.8 (260/280) or 1.6 (260/230) were subjected to another cleanup step by using RNeasy Mini Kit (Qiagen) RNA Cleanup protocol.

For cDNA synthesis, 500 ng RNA were treated with DNase-I (Life Technologies GmbH) and transcribed with PrimeScript RT MasterMix (Perfect Realtime) (Takara Bio Inc.) following the manufacturer's protocol.

Microarray analysis

The aRNA was amplified following the manufacturer's instruction of the GeneChip 3′ IVT Express Kit (Affymetrix Inc.) with 100 ng of RNA and recommended incubation times for microarray hybridization to GeneChip Barley Genome Arrays (Affymetrix Inc.). aRNA purification was performed with RNeasy Mini Kit (Qiagen) following RNA Cleanup protocol. Arrays were hybridized at the Zentrum für Medizinische Grundlagenforschung, Medizinische Fakultät Martin-Luther-Universität Halle-Wittenberg, Universitätsklinikum Halle. Hybridization was done according to the manufacturer's protocol at 45°C over night. Arrays were stained with biotin and streptavidin using the GeneChip™ Fluidics Station 450 (Thermo Fisher Scientific) and finally scanned with the GeneChip™ Scanner 3000 7G (Thermo Fisher Scientific).

Data analysis was performed with RobiNA software (Version 1.2.4_build656) in basic mode except GcRMA was used for normalization. Clustering and PCA of single experiments were conducted in basic mode. P-value correction was performed by Bonferroni Hochberg and multiple testing with nested.F. The P-value cutoff was set at 0.05 with a minimal log fold change of one. Heatmaps were created with R package ‘gplots’ [34] and dendrograms were performed with function ‘hclust’.

Database search for metal-related genes

Four different strategies were combined to assemble the list of potential metal homeostasis genes. First, a key word search was conducted on ensembl (http://plants.ensembl.org/Hordeum_vulgare/Info/Index) with terms based on known metal-related genes (list of keywords see Supplementary Word file S1). Then, sequences of detected MLOC genes were blasted against Affymetrix target sequences (NetAffx BLAST, https://www.affymetrix.com/analysis/netaffx/blast.affx [35]). Contig hits with an e-value ≤ 1.00 × 10−10 were selected as related to genes connected to the specific key words. Afterwards, known Arabidopsis thaliana metal genes (based on a manually curated list of MapMan transition metal homeostasis genes of A. thaliana’ in [36]) were blasted (BlastP) against exemplar sequences and Contig hits with an e-value ≤ 1.00 × 10−10 and assigned as genes related to metal homeostasis genes found in A. thaliana.

Also, a Gene Ontology (GO) search (http://www.ebi.ac.uk/QuickGO/GSearch?q=metal&what=GO&limit=1000) based on the key word ‘metal’ was conducted to identify genes already assigned to metal homeostasis on the barley gene chip. Contigs assigned to bins related to metal homeostasis or metal-related processes were included according to Supplementary Word file S1. After collecting Contigs potentially related to metal homeostasis, all lists were combined and duplicates of Contigs were removed.

Real-time RT-PCR

Transcript quantification was performed on extracted RNA from whole ears of four different lines (140, 154, 143 and 156). To determine a potential change in transcript abundance during grain development, RNA was extracted at three different developmental stages (7, 15 and 27 dap). Similarity of cDNA sequences of all barley lines was analyzed and checked for the absence of polymorphisms in primer binding sites to ensure equal amplification efficiency. Primer sequences were designed based on the exemplar sequence of Contig25867 provided by Affymetrix Inc. (HvMTP-fw: GACTTCGAGTTCACACACCG and HvMTP-rev: ACTAGTAGTACTTCCCAAGCATCC. EF1-α (TC146566) was chosen as reference gene [37]. Real-time RT-PCR reactions were performed in 96-well plates (Bio-Rad iCycler, MyiQ real-time PCR detection system) using SYBR Green (Eurogentec). Cycle threshold (CT) values were calculated as described previously [38].

Statistical analysis

To show variation in micronutrient concentrations, box plots displaying the median, the upper and lower quartile as well as outliers were generated. For comparing the relationship of micronutrients in barley grains and of different field sites, micronutrient concentrations were subjected to regression analysis using SigmaPlot (vers. 13.0). The Pearson product moment correlation coefficients were computed and squared in order to specify the coefficients of determination (r2) with a significance level of P ≤ 0.05. For the calculation of genotypic or environmental effects (G×E) on different nutrients and their contribution to the total variance, a two-way ANOVA was conducted with R vers. 3.0.1 [39] using package ‘multcomp’ [40]. Testing for significance (P ≤ 0.05) was performed with ANOVA and Tukey's HSD posthoc test with R vers. 3.0.1 [39] using packages ‘multcomp’ [40] and ‘agricolae’ [41].

Results

Assembly of the barley collection

A subset of 180 lines of the spring barley ICARDA (International Center for Agricultural Research in the Dry Areas) germplasm panel showing higher genetic diversity was selected for analysis. It comprised two- and six-row lines, including 117 landraces and 63 cultivars/breeding lines from 30 different countries (Supplementary Table S1). Based on the DArT™ marker results, the assembly clusters into four Q-groups representing predominantly two-row landraces and breeding lines from ICARDA, North Africa and East Asia (Q1), six-row landraces from North Africa (Q2), two-row cultivars from North Africa, Europe and Australia (Q3) and six-row landraces from Asia (Q4) (Supplementary Figures S2 and S3). The cluster structure corresponds to the phylogenetic tree and respective clusters were found for a different subset of the ICARDA collection [21].

Natural variation in grain element concentrations across different environments

A profound variation in grain micronutrient concentrations was found in lines grown in the three different environments FT, SE and KW. The median Zn concentration (Figure 1a) ranged from 45.9 mg kg−1 DW (SE) to 56.5 mg kg−1 DW (FT). In environment FT, the widest range in Zn concentrations was found with 24.8 mg kg−1 DW to 91.1 mg kg−1 DW, i.e. a 3.7-fold variation. In the other environments, Zn showed a 2.7- (SE) to 3.5- (KW) fold variation. Median values for Fe, Mn and Cu varied from 48.9 to 59.9 mg kg−1 DW, 16.0 to 27.0 mg kg−1 DW and 5.7 to 7.6 mg kg−1 DW, respectively (Supplementary Figure S4). Median C [%] and total protein contents ranged between 42.6 to 45.0% and 15.9 to 17.1%, respectively. Overall three environments and lines, median Zn concentrations for landraces and breeding lines or cultivars were comparable with 50.4 mg kg−1 DW for landraces (n = 309) and 50.8 mg kg−1 DW for cultivars (n = 164). For the other nutrients Fe, Cu, Mn, C [%] and total protein [%], concentrations were slightly higher in cultivars compared with landraces (Supplementary Table S3).

Variation in zinc (Zn) concentration within a selection of barley lines from the ICARDA collection.

Figure 1.
Variation in zinc (Zn) concentration within a selection of barley lines from the ICARDA collection.

(a) Median grain Zn concentrations of barley lines grown in three different environments (FT, SE, KW). Box plots display the median, the upper and lower quartile as well as extremes. The minimum number of accessions analyzed per sample group was 150. (b) 3D scatter plot of Zn concentrations for lines grown in the three environments. Lines chosen for further analysis are indicated in blue (high Zn accumulating lines) and yellow (low Zn accumulating lines). n = 130. DW, Dry weight.

Figure 1.
Variation in zinc (Zn) concentration within a selection of barley lines from the ICARDA collection.

(a) Median grain Zn concentrations of barley lines grown in three different environments (FT, SE, KW). Box plots display the median, the upper and lower quartile as well as extremes. The minimum number of accessions analyzed per sample group was 150. (b) 3D scatter plot of Zn concentrations for lines grown in the three environments. Lines chosen for further analysis are indicated in blue (high Zn accumulating lines) and yellow (low Zn accumulating lines). n = 130. DW, Dry weight.

The genotype × environment analyses revealed a strong genotype effect for total protein (77.3%, Table 1), Fe (64.5%), Zn (58.1%) and Cu (46.3%) concentrations. In contrast, Mn (50.8%) concentrations and C [%] (78.7%) were more strongly affected by environmental factors. In agreement, was the highest for Fe (0.77) and Zn concentrations (0.73), and the lowest for C [%] (0.6) across the three growth environments FT, SE and KW.

Table 1
ANOVA table for comparisons of genotype and environment effects on grain zinc (Zn), iron (Fe), manganese (Mn), copper (Cu) concentrations, C [%] and total protein [%]
Nutrient Source Df Sum Sq Mean Sq F-value P-value Contribution to variation [%] 
Zn Genotype 129 38 891.95 301.49 4.07 ≤0.001 58.1 
Environment 8985.38 4492.69 60.66 ≤0.001 13.4 
Residuals 258 19 109.3 74.07    
Fe Genotype 129 65 839.23 510.38 4.94 ≤0.001 64.5 
Environment 9515.06 4757.53 46.03 ≤0.001 9.3 
Residuals 258 26 667.83 103.36    
Mn Genotype 129 5190.82 40.24 3.59 ≤0.001 31.6 
Environment 8346.91 4173.45 372.46 ≤0.001 50.8 
Residuals 258 2890.91 11.21    
Cu Genotype 129 592.64 4.59 3.11 ≤0.001 46.3 
Environment 306.92 153.46 103.9 ≤0.001 24.0 
Residuals 258 381.08 1.48    
Genotype 145 66.15 0.46 1.96 ≤0.001 14.1 
Environment 370.13 370.13 1591.02 ≤0.001 78.7 
Residuals 145 33.73 0.23    
Total protein Genotype 145 967.8 6.68 5.48 ≤0.001 77.3 
Environment 106.89 106.89 87.72 ≤0.001 8.5 
Residuals 145 176.68 1.22  ≤0.001  
Nutrient Source Df Sum Sq Mean Sq F-value P-value Contribution to variation [%] 
Zn Genotype 129 38 891.95 301.49 4.07 ≤0.001 58.1 
Environment 8985.38 4492.69 60.66 ≤0.001 13.4 
Residuals 258 19 109.3 74.07    
Fe Genotype 129 65 839.23 510.38 4.94 ≤0.001 64.5 
Environment 9515.06 4757.53 46.03 ≤0.001 9.3 
Residuals 258 26 667.83 103.36    
Mn Genotype 129 5190.82 40.24 3.59 ≤0.001 31.6 
Environment 8346.91 4173.45 372.46 ≤0.001 50.8 
Residuals 258 2890.91 11.21    
Cu Genotype 129 592.64 4.59 3.11 ≤0.001 46.3 
Environment 306.92 153.46 103.9 ≤0.001 24.0 
Residuals 258 381.08 1.48    
Genotype 145 66.15 0.46 1.96 ≤0.001 14.1 
Environment 370.13 370.13 1591.02 ≤0.001 78.7 
Residuals 145 33.73 0.23    
Total protein Genotype 145 967.8 6.68 5.48 ≤0.001 77.3 
Environment 106.89 106.89 87.72 ≤0.001 8.5 
Residuals 145 176.68 1.22  ≤0.001  

A set of different barley genotypes were grown in three (Zn, Fe, Mn, Cu, n = 130) or two (C, total protein, n = 146) different environments for the analysis.

The interrelationship between different grain nutrients can provide important information on distribution or storage mechanisms (Figure 2a,b and Supplementary Table S4). Zn was found to be significantly correlated with Fe at r2 = 0.58 (P ≤ 0.001). Regarding correlations of total protein or C [%] and Zn or Fe, correlations were very weak (Figure 2c,d). For total protein, coefficients of determination were 0.12 (P ≤ 0.001) and 0.11 (P ≤ 0.001) for Zn and Fe, respectively. There was a clear clustering of C [%] values for different sites, and only weak correlation [for Fe, coefficient of determination of 0.02 (P ≤ 0.01)] or no correlation (Zn) with micronutrients (Figure 2d and Supplementary Figure S5). The correlation between Mn, Cu and total protein was strongly dependent on the environment and coefficients of determination were both higher in SE (Mn: 0.24, P ≤ 0.001 and Cu: 0.15, P ≤ 0.001; Supplementary Figure S5, Pearson Correlation Coefficients Supplementary Tables S5 and S6). In summary, correlations of Zn and Fe concentrations are stronger and more stable over different environments than correlations between micronutrients and total protein or C [%].

Correlation of micronutrients, total protein and carbon (C [%]) within the ICARDA collection.

Figure 2.
Correlation of micronutrients, total protein and carbon (C [%]) within the ICARDA collection.

(a) Regression plots of zinc (Zn) vs iron (Fe), manganese (Mn) and Fe vs Mn (b) and regression plots of Cu (copper) vs Zn, Fe and Mn. Regression plots include data from all three environments. (c) Regression plots of total protein vs Zn and Fe (d) and of C [%] vs Zn and Fe. Regression plots include data from two environments (SE, KW). r2: coefficient of determination. All displayed coefficients of determination were significant (n = 473 (a,b), n = 275 (c,d)).

Figure 2.
Correlation of micronutrients, total protein and carbon (C [%]) within the ICARDA collection.

(a) Regression plots of zinc (Zn) vs iron (Fe), manganese (Mn) and Fe vs Mn (b) and regression plots of Cu (copper) vs Zn, Fe and Mn. Regression plots include data from all three environments. (c) Regression plots of total protein vs Zn and Fe (d) and of C [%] vs Zn and Fe. Regression plots include data from two environments (SE, KW). r2: coefficient of determination. All displayed coefficients of determination were significant (n = 473 (a,b), n = 275 (c,d)).

MTAs for micro- and macronutrients varied between different environments

Overall, 76 MTAs were found for the six traits in the three environments and dispersed over 6 of the 7 barley chromosomes (Supplementary Table S7 and Figure S6). From grains grown in all environments and BLUEs, 13 MTAs were identified for Zn, 15 MTAs for Cu, 6 MTAs for Fe, 9 MTAs for Mn, 13 MTAs for C [%] and 20 MTAs for protein content. Most MTAs (35) were found on chromosome 2H and were associated with 18 markers. The MTAs found for the different environments FT, SE, KW and BLUEs numbered 10, 20, 25 and 21, respectively. Out of those, 17 markers were significantly associated with more than one trait in the different environments.

Most MTAs involving Zn were found on 2H and the most significantly associated markers were 2H│bPb9754 showing an explained phenotypic variation (R2) of 0.07 in FT, 2H│bPb9199 (R2 of 0.08 in KW), 2H│bPb1566 (R2 of 0.03 in SE) and 2H│bPb9754 (R2 of 0.06 for the BLUEs) (Figure 3). Within ±0.5 cM of significant MTAs, candidate genes connected to Zn homeostasis were not detected (Supplementary Table S8). In close vicinity (±2.5 cM) of the MTA at 2H│bPb9754 (82.77 cM), we identified two yellow stripe like (YSL) genes. One gene is homologous to AtYSL2 (MLOC_40066.1, position 80.89 cM) and the other one is annotated as YSL9 (MLOC_61170.4, position 80.95 cM) in barley (Supplementary Table S9). Both genes, MLOC_40066.1 and MLOC_61170.4, show expression in roots and shoots of seedlings and during grain development (Supplementary Figure S7).

Significant MTAs for zinc (Zn) concentration in barley grains grown under different environmental conditions.

Figure 3.
Significant MTAs for zinc (Zn) concentration in barley grains grown under different environmental conditions.

ICARDA lines were grown in three environments (FT, KW, SE) and data on Zn content in grains were used for GWAS. BLUEs were calculated across the three environments. All significant MTAs at −log(PM) > 2.5 were mapped on chromosomes 1H, 2H and 6H. Associated DArT markers and marker positions (in cM) are given.

Figure 3.
Significant MTAs for zinc (Zn) concentration in barley grains grown under different environmental conditions.

ICARDA lines were grown in three environments (FT, KW, SE) and data on Zn content in grains were used for GWAS. BLUEs were calculated across the three environments. All significant MTAs at −log(PM) > 2.5 were mapped on chromosomes 1H, 2H and 6H. Associated DArT markers and marker positions (in cM) are given.

Over all micronutrients, the marker 2H│bPb4040 revealed the highest significant association [−log(PM) = 3.8 for Cu], whereas for the macronutrients, the marker 2H│bPb2501 at 47 cM achieved the highest significance level [−log(PM) = 5.6] for C [%] of grains grown in KW. The most significant association for protein content was found at 102.4 cM [2H│bPb6194].

In total, 11 pleiotropic loci, i.e. marker loci associated with more than one phenotypic trait, were identified for micronutrients (Supplementary Table S8). On chromosome 1H at 72.9 cM, the traits Cu, Mn and Zn were associated with bPb4909. On chromosome 2H between 82.1 and 82.8, Cu and Zn were linked to the same markers bPb4040 and bPb9754 and on chromosome 6H, MTAs for Fe and Zn were associated with bPb8836.

Cultivation under controlled conditions and grain dissection

Based on the broad variation of grain Zn concentrations between 26.4 and 83.9 mg kg−1 DW in the different environments (Figure 1b and Supplementary Figure S8), four lines were selected for further analysis. All lines were six-row barleys and originated from Pakistan (140, 143) or Tunisia (154, 156). Lines 140 and 143 were grouped in subgroup Q4, line 154 in Q3 and line 156 in Q2.

Selected lines were grown under controlled conditions in three different cultivation rounds and ICP-OES measurements on grains confirmed that the selected lines were either low or high Zn accumulating lines (140: 39.4 ± 2.0 mg kg−1 DW, 154: 38.4 ± 0.6 mg kg−1 DW, 143: 61.2 ± 4.1 mg kg−1 DW, 156: 65.3 ± 7.9 mg kg−1 DW; Figure 4a). Fe concentrations showed significant differences between lines 154 and 156 (Fe: 140: 37.7 ± 4.6 mg kg−1 DW, 154: 36.0 ± 0.4 mg kg−1 DW, 143: 44.1 ± 2.5 mg kg−1 DW, 156: 56.0 ± 10.2 mg kg−1 DW; Figure 4b). Mn and Cu concentrations in grains of the selected lines were comparable (Figure 4c,d).

Zinc (Zn), iron (Fe), copper (Cu) and manganese (Mn) concentration of two contrasting pairs of barley lines in mature grains.

Figure 4.
Zinc (Zn), iron (Fe), copper (Cu) and manganese (Mn) concentration of two contrasting pairs of barley lines in mature grains.

(a) Grain Zn, (b) Fe, (c) Mn and (d) Cu concentrations of lines 140, 154, 143 and 156 for three different cultivation rounds under controlled conditions in the glasshouse or growth chamber (n = 3, 3–5 replicates per cultivation round combined). Values are means + SD. Statistical differences were calculated using ANOVA/Tukey's HSD post hoc test and are displayed as letters (P ≤ 0.05).

Figure 4.
Zinc (Zn), iron (Fe), copper (Cu) and manganese (Mn) concentration of two contrasting pairs of barley lines in mature grains.

(a) Grain Zn, (b) Fe, (c) Mn and (d) Cu concentrations of lines 140, 154, 143 and 156 for three different cultivation rounds under controlled conditions in the glasshouse or growth chamber (n = 3, 3–5 replicates per cultivation round combined). Values are means + SD. Statistical differences were calculated using ANOVA/Tukey's HSD post hoc test and are displayed as letters (P ≤ 0.05).

To detect potential hotspots of Zn accumulation in barley grains, whole grains of the selected lines were dissected into embryo, endosperm and husk (Figure 5). Lines having high Zn concentrations in grains had also elevated Zn concentrations in all separated tissues. Lines showing low Zn concentrations in grains differed significantly from the lines accumulating high Zn concentrations in the endosperm tissue (Figure 5a). The highest Zn concentrations were observed in embryo tissues ranging between 129.0 ± 7.9 mg kg−1 DW (154) and 211.8 ± 26.6 mg kg−1 DW (143). Based on the total Zn concentration in the grains, the embryo tissue contributed a fraction of 0.10 ± 0.02 (154) to 0.13 ± 0.01 (143) while the endosperm comprised a fraction between 0.69 ± 0.02 (156) to 0.77 ± 0.04 (140) (Figure 5b). Similar patterns were shown for Fe and Cu, whereas for Mn, embryo and husk contributed, taken together, as much to total Mn as the endosperm (Supplementary Figure S9).

Zinc (Zn) concentration in different mature seed tissues of two contrasting pairs of barley lines.

Figure 5.
Zinc (Zn) concentration in different mature seed tissues of two contrasting pairs of barley lines.

(a) Grain Zn concentrations of embryo, endosperm and husk in four different barley lines (140, 154, 143, 156). (b) Fractions of total Zn contributed by different tissues (embryo, endosperm and husk). Fractions were calculated by multiplying the relative weight of each tissue and the corresponding Zn concentration measured for each tissue and referred to the total Zn concentration. Plants were cultivated in three different cultivations in the glasshouse or growth chamber (n = 3, 3–5 replicates per cultivation round combined). For each sample, five grains per line and cultivation round were pooled. Values are means + SD. Statistical differences between different lines were calculated for each tissue using ANOVA/Tukey's HSD post hoc test and are displayed as letters (P ≤ 0.05).

Figure 5.
Zinc (Zn) concentration in different mature seed tissues of two contrasting pairs of barley lines.

(a) Grain Zn concentrations of embryo, endosperm and husk in four different barley lines (140, 154, 143, 156). (b) Fractions of total Zn contributed by different tissues (embryo, endosperm and husk). Fractions were calculated by multiplying the relative weight of each tissue and the corresponding Zn concentration measured for each tissue and referred to the total Zn concentration. Plants were cultivated in three different cultivations in the glasshouse or growth chamber (n = 3, 3–5 replicates per cultivation round combined). For each sample, five grains per line and cultivation round were pooled. Values are means + SD. Statistical differences between different lines were calculated for each tissue using ANOVA/Tukey's HSD post hoc test and are displayed as letters (P ≤ 0.05).

For transcriptome analysis, whole ears and flag leaves were sampled 15 dap and micronutrient concentrations were measured. Zn concentrations of whole ears were mostly consistent with the results for mature grains in the different lines (Figure 6a). In flag leaves, line 154 showed lower Zn concentrations than the other lines (154: 18.6 ± 1.3 mg kg−1 DW, 140: 30.1 ± 10 mg kg−1 DW, 143: 40.4 ± 17.5 mg kg−1 DW, 156: 50.7 ± 17.9 mg kg−1 DW) but this difference was also not significant. For Fe, Mn and Cu, there was a similar pattern between lines in whole ears 15 dap compared with mature grains, but again no significant differences could be detected at 15 dap. Fe and Cu concentrations were comparable between the selected lines in flag leaves at 15 dap, while Mn concentrations were highly variable between the lines. Zn concentrations in whole ears and flag leaves were significantly correlated (r2: 0.38, P = 0.03, n = 12), whereas Zn concentrations of flag leaves and mature grains were not correlated (r2 = 0.17, P = 0.18, n = 12).

Zinc (Zn), iron (Fe), copper (Cu) and manganese (Mn) concentrations of two contrasting pairs of barley lines in whole ears and flag leaves 15 dap.

Figure 6.
Zinc (Zn), iron (Fe), copper (Cu) and manganese (Mn) concentrations of two contrasting pairs of barley lines in whole ears and flag leaves 15 dap.

(a,c,e,g) Metal concentrations in whole ears (b,d,f,h) and flag leaves of lines 140, 154, 143 and 156. Plants were cultivated in three different cultivations in the glasshouse or growth chamber (n = 3, 3–5 replicates per cultivation round combined). Values are means + SD. Calculation of statistical differences with ANOVA/Tukey's HSD post hoc test (P ≤ 0.05) did not reveal any significant differences.

Figure 6.
Zinc (Zn), iron (Fe), copper (Cu) and manganese (Mn) concentrations of two contrasting pairs of barley lines in whole ears and flag leaves 15 dap.

(a,c,e,g) Metal concentrations in whole ears (b,d,f,h) and flag leaves of lines 140, 154, 143 and 156. Plants were cultivated in three different cultivations in the glasshouse or growth chamber (n = 3, 3–5 replicates per cultivation round combined). Values are means + SD. Calculation of statistical differences with ANOVA/Tukey's HSD post hoc test (P ≤ 0.05) did not reveal any significant differences.

Transcriptomic differences in low and high Zn accumulating barley lines

Transcriptome data for independent biological replicates clustered depending on the plant organ sampled (Figure 7). Lines 140 and 143 clustered for whole ears and were more similar to each other than lines 154 and 156, which is consistent with results obtained in the population structure analysis (Supplementary Figure S2). Flag leaf transcriptome data clustered depending on cultivation round. Within each cultivation round, lines 140/143 and 154/156 clustered together. In general, PCA of flag leaf transcriptome data indicated a higher variability and a higher influence of external growth factors than found for the transcriptomes of developing ears.

Clustering of whole ear and flag leaf transcriptomic data.

Figure 7.
Clustering of whole ear and flag leaf transcriptomic data.

(a) Cluster dendrogram of transcripts from whole ears (b) and flag leaves. Hclust, method: ‘pearson’. (c) Principal component analysis of transcript signals from whole ears (d) and flag leaves. n = 3. Normalization method: GcRMA, P-value correction: Bonferroni Hochberg, analysis with RobiNA, multiple testing: nested.F.

Figure 7.
Clustering of whole ear and flag leaf transcriptomic data.

(a) Cluster dendrogram of transcripts from whole ears (b) and flag leaves. Hclust, method: ‘pearson’. (c) Principal component analysis of transcript signals from whole ears (d) and flag leaves. n = 3. Normalization method: GcRMA, P-value correction: Bonferroni Hochberg, analysis with RobiNA, multiple testing: nested.F.

The main objective of the transcriptome analyses was to identify genes possibly associated with the differences in grain Zn. Therefore, we searched for transcripts that showed consistent differences in abundance between low Zn cultivars and high Zn cultivars across the four selected cultivars with contrasting grain Zn. Following comparisons of all possible pairs of low vs high Zn cultivars, i.e. 140 vs 143 or 156 and 154 vs 143 or 156 (Figure 8), we identified differentially expressed genes (DEGs) shared in all single comparisons. In whole ears, 7 genes showed higher and 13 genes lower transcript abundance in high Zn lines compared with low Zn lines. Five and 11 genes with higher and lower transcript abundance, respectively, were found in flag leaves (for full list of genes see Supplementary Table S10). The highest difference between transcript levels within one comparison was found for an α-amylase/trypsin inhibitor CMb precursor (HB09A04w_s_at) with +8.50-fold higher expression and for an transmembrane protein (similar to AT5g02160 [A. thaliana], Contig7943_at) with −7.18-fold lower expression in high compared with low Zn accumulating lines (Figure 9). Three genes showed higher transcript abundance in both analyzed tissues in high Zn lines compared with low Zn lines. Through a blastx search, two of the corresponding contigs could be linked to a putative cytochrome P450 and a ribosomal protein S30 homolog, respectively. Conversely, six genes showed lower expression levels in high Zn lines in both tissues. Nineteen out of 26 genes were not assigned to a specific functional category (bin). Blastx searches conducted using exemplar sequences from Affymetrix revealed known conserved domains in nine sequences (Supplementary Table S11), adding information to three predicted proteins. This included an exonuclease-endonuclease-phosphatase (EEP) (Contig15482_at), an Ist1 (Contig5646_at) and a DUF247 (Contig10162_s_at) domain superfamily.

Differences in transcript abundance between high and low zinc (Zn) lines.

Figure 8.
Differences in transcript abundance between high and low zinc (Zn) lines.

(a,b) Number of genes showing higher or lower transcript abundance in each possible comparison of high Zn (143 and 156) vs low Zn (140 and 154) barley lines 15 dap in whole ears and (c,d) flag leaves. Circles are representing comparisons of line 143 vs 140 in red, 143 vs 154 in green, 156 vs 140 in blue and 156 vs 154 in yellow. Genes shared between comparisons of different lines are given within circles. Dashes below the corresponding numbers represent the number of single comparisons included (max. 4). Statistical parameters of transcript analysis: P-value cutoff: 0.05, log fold change min = 1.

Figure 8.
Differences in transcript abundance between high and low zinc (Zn) lines.

(a,b) Number of genes showing higher or lower transcript abundance in each possible comparison of high Zn (143 and 156) vs low Zn (140 and 154) barley lines 15 dap in whole ears and (c,d) flag leaves. Circles are representing comparisons of line 143 vs 140 in red, 143 vs 154 in green, 156 vs 140 in blue and 156 vs 154 in yellow. Genes shared between comparisons of different lines are given within circles. Dashes below the corresponding numbers represent the number of single comparisons included (max. 4). Statistical parameters of transcript analysis: P-value cutoff: 0.05, log fold change min = 1.

Heatmap analysis showing differences in transcript abundance in whole ears and flag leaves shared in all possible comparisons of high vs low zinc (Zn) lines.

Figure 9.
Heatmap analysis showing differences in transcript abundance in whole ears and flag leaves shared in all possible comparisons of high vs low zinc (Zn) lines.

Colored visualization and hierarchical clustering of differences in transcript abundance of each comparison of high (143 or 156) vs low (140 or 154) Zn accumulating barley lines 15 dap in whole ears and flag leaves. Color key displays variation in fold-changes from lower (yellow) to higher (blue) transcript abundance in high Zn lines compared with low Zn lines. A histogram is shown in light blue. Dendrogram and heatmap were generated with hclust (gplots) in R. Statistical parameters of transcript analysis: P-value cutoff: 0.05, log fold change min = 1.

Figure 9.
Heatmap analysis showing differences in transcript abundance in whole ears and flag leaves shared in all possible comparisons of high vs low zinc (Zn) lines.

Colored visualization and hierarchical clustering of differences in transcript abundance of each comparison of high (143 or 156) vs low (140 or 154) Zn accumulating barley lines 15 dap in whole ears and flag leaves. Color key displays variation in fold-changes from lower (yellow) to higher (blue) transcript abundance in high Zn lines compared with low Zn lines. A histogram is shown in light blue. Dendrogram and heatmap were generated with hclust (gplots) in R. Statistical parameters of transcript analysis: P-value cutoff: 0.05, log fold change min = 1.

Database search for metal-related genes

Many of the genes found to show differences in transcript abundance between low and high Zn lines have not yet been assigned a function. Following the roadmap proposed for Zn loading of barley grains [28], we searched for genes that are potentially involved in metal homeostasis and connected those to contigs available on the barley gene chip.

The different searches for putative metal homeostasis genes resulted in 1447 genes defined by bins, 1037 by GO terms, 231 by A. thaliana blast and 88 genes by ensembl search. Three hundred and eighty-three of these genes could be identified as or connected to already known metal homeostasis genes (Supplementary Excel file S1). These included barley YSL genes (HvYSL2, HvYSL3, HvYSL6 and HvYSL7). In whole ears and flag leaves none of the putative metal homeostasis genes were differentially expressed in all possible comparisons of high Zn lines with low Zn lines (143 vs 140 or 154 and 156 vs 140 or 154, respectively). We then lowered the stringency of our analysis and included genes that were differentially expressed at least in one comparison between related cultivars with contrasting grain Zn, i.e. when comparing 156 with 154 or 143 with 140. These pairs are more closely related to each other than to the other analyzed cultivars according to the clustering of transcriptome data and assignment to Q-groups. While in flag leaves, none of the defined metal-related genes was differentially expressed in any of the single comparisons, Contig25867_at showed significantly lower transcript abundance in whole ears of both pairwise comparisons of the more related lines (143–140: −1.97-fold and 156–154: −2.04-fold). Blast searches based on the exemplar sequence of the contig revealed high similarity to barley Metal tolerance protein 5 (MTP5: HORVU1Hr1G071930, Expect = 7 × 10−66, Identities = 208/252 (82%)). Within the query sequence, a dimerization domain of a zinc transporter was detected (Expect = 1.79 × 10−04). Results of quantitative real-time RT-PCR confirmed data of the microarray analysis with higher relative transcript abundance in low Zn lines compared with high Zn lines within pairs 143–140 and 156–154 (Figure 10). In addition, analysis of different developmental stages showed that relative transcript abundance of the MTP5 gene was in all lines higher at 27 dap than at the other two analyzed developmental stages. At 7 dap, no significant difference in relative transcript abundance within each pair was detected, while differences were more pronounced at 27 dap, with significant differences in both pairs. At 27 dap, relative transcript abundance in low Zn lines was 2.2-fold (154/156) to 2.7-fold (140/143) higher compared with high Zn lines.

Transcript levels of the MTP5 candidate gene at three developmental stages in whole ears of low and high Zn accumulating lines.

Figure 10.
Transcript levels of the MTP5 candidate gene at three developmental stages in whole ears of low and high Zn accumulating lines.

MTP5 transcript abundance in whole ears at 7, 15 and 27 dap in low (140, 154) and high (143, 156) Zn lines. Transcript level was determined relative to EF1α. Values are means + SD of three independent experiments (three to five plants of each line were pooled for each data point). Statistical differences were calculated using ANOVA/Tukey's HSD post hoc test on single comparisons of pairs 140/143 and 154/156 and are displayed as asterisks (** P ≤ 0.01).

Figure 10.
Transcript levels of the MTP5 candidate gene at three developmental stages in whole ears of low and high Zn accumulating lines.

MTP5 transcript abundance in whole ears at 7, 15 and 27 dap in low (140, 154) and high (143, 156) Zn lines. Transcript level was determined relative to EF1α. Values are means + SD of three independent experiments (three to five plants of each line were pooled for each data point). Statistical differences were calculated using ANOVA/Tukey's HSD post hoc test on single comparisons of pairs 140/143 and 154/156 and are displayed as asterisks (** P ≤ 0.01).

Discussion

Micronutrient deficiencies affect large fractions of the global population [4244]. An estimated 2 billion people worldwide are threatened by insufficient Zn supply [45]. Most of them rely on diets that lack diversity and are dominated by cereal grains, which are notoriously poor in bioavailable Zn and also Fe, another undersupplied essential metal [1,46]. The ongoing rise in atmospheric CO2 levels is predicted to further aggravate this problem [6,7]. Thus, there is an urgent need to breed for cereals with higher density of bioavailable micronutrients. This has increasingly been realized and drives large-scale programs such as HarvestPlus (www.harvestplus.org). However, modern breeding tools can only be applied when the underlying molecular mechanisms of micronutrient grain loading are known. In crop plants, mechanisms of grain Zn loading are still not well understood. In fact, there are no success stories yet reporting the use of suitable markers to breed for micronutrient-enriched cereals [47]. While several genes and proteins involved in the uptake, translocation and accumulation of Zn, Fe and other transition metals are known, the factors explaining natural variation in micronutrient density remain unidentified.

The majority of the genes implicated in metal homeostasis of cereals encode metal transporters including members of the Metal Tolerance Protein (MTP) and heavy metal ATPase (HMA) families or enzymes involved in the synthesis of metal chelating molecules [14,46,48,49]. Arguably most advanced is the knowledge for rice. However, rice is predominantly grown in flooded paddy fields, i.e. in conditions with fundamentally altered phytoavailability especially of Fe in the soil. Thus, cereals cultivated in aerated soil need to be studied in detail as well [50]. Barley represents a suitable model system. It is adapted to a wide range of environmental conditions, genetically tractable and shows a wide genetic variation since it has been in cultivation since ∼8000 BC [51]. Furthermore, barley seeds are important in Northern Africa and Asia [52]. Several studies have already demonstrated a considerable degree of genetically controlled variation in grain micronutrient concentration, i.e. in barley [24,25,53], wheat [54,55] and rice [56,57]. Thus, it appears feasible to identify factors associated with differences in grain Zn concentration.

We selected a representative subset of the ICARDA germplasm for which marker data are available and grew this collection in three different environments to determine the degree of variation in grain micronutrient concentration [16,58]. The median Zn concentration (Figure 1a) was similar to other barley collections [24,25] while the variation of 3.7-fold was slightly higher than reported for barley germplasm collections grown under controlled conditions [24]. We found slightly higher median Fe concentrations and with 9.6-fold a higher degree of variation than other studies on different barley panels [53,59]. Importantly, Fe and Zn concentrations varied little across different environments, resulting in comparatively high heritability (H2 = 0.77 and 0.73, respectively) and a genotype contribution of 64.5% (Fe) and 58.1% (Zn) to the total phenotype (Table 1). A pronounced heritability for Zn had previously been shown in barley landraces from Ethiopia and Eritrea (H2 = 0.65) and in sorghum (H2 = >0.60 for both Zn and Fe) [25,60]. Tight genetic control of micronutrient concentration can be interpreted from an evolutionary perspective as arising from the need to provide an adequate supply of nutrients to the progeny via the seed reserves.

Correlations between different macro- and micronutrients may indicate potential overlaps in grain loading pathways. Therefore, we analyzed correlations of Zn with Fe, Mn and Cu as well as with C [%] and total protein (Figure 2). About 60% of the variation in Fe concentration could be explained by the Zn concentration in the grain. Comparable findings were reported for barley [24], wheat [54,55,6164], rice [65], sorghum [60], pearl millet [66] and black gram [67]. These results suggest that Zn and Fe are at least partially sharing metal transporters and/or low molecular mass chelators.

The only molecularly understood QTL explaining differences in cereal micronutrient concentration is the Gpc-B1 locus in wheat [68]. Allelic variation at this locus, which encodes a transcription factor, affects protein, Fe, and Zn concentration of the grain. On the other hand, there are indications that in wheat grain more Zn than Fe is bound to proteins [69,70]. Therefore, it is also important to investigate how Zn concentration is correlated with other grain components. Our results suggest that both, Zn and Fe, were correlated with total protein rather than with C [%]. This appears to be similar in wheat, while in rice, a slight, albeit non-significant, negative correlation between Zn or Fe and protein content was found [71]. However, correlations were highly affected by different environments in our experiments indicating that other binding environments than storage proteins or C [%] play important roles.

Based on the observed variation in grain Zn we followed two strategies to find candidate genes possibly contributing to differences between barley accessions. First, the available marker data for the analyzed ICARDA collection were used to map loci associated with grain Zn. Second, pairs of contrasting lines were selected for in-depth analysis including transcript analysis of ears and flag leaves.

Zn accumulation is a quantitative inherited trait and was affected by seven loci on chromosomes 1H, 2H and 6H for grains grown across three different environments (Figure 3). By using 54 introgression lines, Reuscher et al. demonstrated that markers in comparable regions such as on chromosome 1H (64.8–90.9 cM) and 2H (34.3–66.8 cM) were also associated with differences in Zn concentrations [50]. Furthermore, they found a candidate gene on chromosome 2H that showed highest sequence similarity to the A. thaliana ZIP1 and the rice OsZIP3 gene, members of the Zinc-regulated transporter Iron-regulated Protein (ZIP) family proposed to be involved in Zn2+ uptake [50]. Based on the latest barley genome sequence [23], this gene could be the one now annotated as Zinc transporter gene 8 (HORVU2Hr1G025400). It is located near the centromeric region between 725 227 365 and 725 232 183 bp. The closest marker associated with Zn concentration found in our analysis was bPb4040 at 82.1 cM, which may correspond to a physical position of 663 877 529, indicating that a gene other than the Zinc transporter gene 8 affects the Zn concentration in the ICARDA panel. Furthermore, by using 336 cultivars of a barley breeding program from ICARDA among others, three QTL regions were identified on chromosomes 2H, 3H and 5H [72]. Again, on chromosome 2H at 87.3 cM, the locus was in close vicinity to the MTAs for Zn concentration in the FT environment and for BLUEs located at 2H│bPb9754 (82.77 cM). Within a range of ±2.5 cM of this most relevant marker on chromosome 2H, two genes were detected that belong to the yellow stripe like (YSL) transporter family. A few YSL proteins have been shown or hypothesized to transport metals that are bound to nicotianamine (NA) and phytosiderophores (PS) and several family members have been implicated in Fe and Zn transport into grains [7375]. HvYSL9 is expressed throughout the grain with exception of the embryo [28]. Tauris and colleagues suggested that YSL transporters could play a role in transferring Zn–NA complexes from the endosperm cavity into the apoplast of transfer cells and between testa and aleurone [28]. Increased transport capability from maternal to filial tissues could explain the overall higher Zn concentrations in different grain tissues we identified in the high Zn accumulating lines. AtYSL3 was found to be important for loading of metals into the seeds [74]. In rice, YSL9 was found to play an important role in Fe(II)–NA and Fe(III)–deoxymugineic translocation from endosperm to embryo in the seeds, especially at peripheral and inner parts of the endosperm near the embryo [76]. We observed elevated Zn in the endosperm near the embryo when barley grains were analyzed by µPIXE [24]. Thus, the locus on chromosome 2H could potentially play an important role in Zn accumulation and the YSL transporters are interesting candidates for further investigation of Zn grain accumulation differences in barley. Future studies will have to show whether polymorphisms and/or expression differences for the YSL genes exist between barley accessions and whether they can be directly associated with grain Zn. Transcriptome analyses of the very limited set of four barley lines did not reveal significant differences.

Several observations indicate that breeding for high yield has selected against micronutrient density, especially in bread wheat cultivars [54,77]. This and the dilution effect of larger endosperm explain why the nutritional value of bred crops often seems to be lower than of landraces [78]. In our study, however, Zn concentrations varied to the same extent in landraces and barley cultivars with highly comparable median Zn concentrations (Supplementary Table S3). Landraces are commonly understood as locally adapted varieties that evolved because of natural and non-targeted human selection. Due to their specific allele combinations, landraces may reveal variations in morphological, quality and yield characteristics and can harbor important genes for plant breeding [79]. For example, many barley landraces originate from Zn-deficient soils [50] and could therefore carry alleles supporting higher Zn efficiency. We chose four landraces, two pairs of genetically related lines with contrasting grain Zn across the different environments, for detailed analysis under controlled conditions (Figure 4).

In the context of biofortification, the distribution of Zn within the grain is relevant because only parts of the grain are consumed. Therefore, we dissected mature barley grains into husk, embryo and endosperm tissues (Figure 5). Cultivars that show high Zn accumulation have higher Zn concentrations in all grain tissues, including the endosperm, i.e. the part that remains after milling. This result is consistent with a previous study that employed metal imaging via EXAFS [24] and was found for wheat as well [70,80,81]. It suggests that Zn trafficking from maternal to filial tissues including the loading zones of the grain is the major factor causing accumulation differences. Additionally, high Zn lines could have more Zn storage capability across the different grain tissues [64,82]. To date no genes have been directly linked to variation in grain micronutrient loading or sink strength. Hence, transgenic approaches aim to analyze mechanisms of micronutrient accumulation in the grain by expressing transporters such as heavy metal ATPase 2 (HMA2) [9,83,84]. Functional analysis of TaHMA2 showed an implication in Zn and cadmium (Cd) transport with decreased Zn concentrations in the grains in TaHMA2 over-expressing rice and wheat [84]. Examples like this elucidate the potential of transgenic approaches in gaining mechanistic insights into micronutrient loading of the grain.

Laser dissection-assisted transcript analysis of barley grain tissues had suggested a roadmap of Zn trafficking [28]. To determine whether genes on or beyond this roadmap can be linked to differences in grain Zn, whole ears and flag leaves were harvested 15 dap and micronutrient concentrations were determined (Figure 6). At 15 dap, all grain tissues are differentiated and the grain filling process takes place [85,86]. In whole ears, the patterns of low or high Zn lines were the same at 15 dap and maturity. Interestingly, we measured slightly lower micronutrient concentrations at 15 dap compared with mature grains. These results are in contrast with observations made in wheat, where Zn concentrations decreased by 45% until 27 dap and remained stable until maturity [87]. These results indicate stringent control over micronutrient levels in low or high Zn lines during grain development.

There are two possible pathways for Zn translocation towards the reproductive tissues: Zn is either directly mobilized from the soil and transported to the developing grain or previously stored Zn is remobilized from the flag leaves [9,50,8892]. Therefore, flag leaves were included in the analysis. However, in flag leaves, there was no obvious correlation connecting flag leaf micronutrient concentration to grain micronutrient concentration at maturity. Also, the analysis of microarray data grouped flag leaf samples based on cultivation round, not genotype (Figure 7). In contrast, for whole ears, transcriptome data of the single lines clustered together, indicating a stronger influence of environmental factors on gene expression in flag leaves relative to whole ears.

For transcriptome analysis, we first searched for genes differentially expressed in all four possible comparisons between high and low Zn lines. This stringent approach resulted in a relatively small number of DEGs shared in both groups (Figure 8). Within these DEGs, 19 out of 26 were not assigned yet and could not be connected to any function or annotation. As no genes encoding metal transporters or enzymes involved in the synthesis of Zn ligands were present in our list, we performed further database searches to link more genes of the barley gene chip to metal homeostasis genes. These included well-known families such as HMA, YSL, ZIP, CDF (cation diffusion facilitator), Nramp (natural resistance-associated protein), VIT (vacuolar iron transporter), CAX (cation exchanger), NAS (nicotianamine synthase), ZIF (zinc-induced facilitator), ZIFL (zinc-induced facilitator-like), MT (metallothionein), NAAT (nicotianamine aminotransferase), PME (pectin methylesterase) or PCS (phytochelatin synthase). A total of 383 genes could be assigned a putative function in metal transport, trafficking or storage based on similarity to already known metal homeostasis-related genes (Supplementary Excel file S1/Word File S1). None of these genes was significantly higher or lower expressed in both high Zn lines compared with both low Zn lines. When searching for genes differentially expressed in single related pairs of contrasting lines, a transcript similar to barley metal tolerance protein 5 (HvMTP5) was found to be significantly less abundant in high Zn lines compared with low Zn lines. Blast searches based on the exemplar sequence implicated a function in Zn transport. CDFs, or in plants also known as MTPs, are membrane bound transporters involved in Zn efflux from cytoplasm to cellular compartments, e.g. the vacuoles [93]. HvMTP1 was previously identified to transport Zn and cobalt (Co) and to be expressed in phloem and aleurone cells in the grain [28,94]. Recently, transgenic barley plants expressing HvMTP1 under control of an endosperm-specific promotor were shown to contain higher Zn concentrations in the grain compared with the reference line cv. ‘Golden Promise’, possibly because sink strength of the endosperm was enhanced. In contrast, we would have to postulate that MTP5 is involved in the trapping of Zn within cells along the loading pathway in maternal tissue and thereby reduce the availability of Zn for transfer to filial tissues. However, at this stage we do not have detailed information on where exactly MTP5 is expressed. Also, we do not know if expression is responsive to Zn supply. To our knowledge, this study contributes the first transcriptome data exploring natural variation in barley grain Zn accumulation. Our GWAS and transcriptome analyses indicate a potential role of YSL and MTP transporters. Now, more in depth studies such as targeted knockouts of promising transporter genes are needed to investigate the specific involvement of the annotated genes in Zn accumulation.

Abbreviations

     
  • ANOVA

    analysis of variance

  •  
  • BLUEs

    best linear unbiased estimates

  •  
  • CT

    cycle threshold

  •  
  • dap

    days after pollination

  •  
  • DArT™

    Diversity Arrays Technology

  •  
  • DW

    dry weight

  •  
  • FT

    foil tunnel

  •  
  • GWAS

    genome-wide association study

  •  
  • GxE

    genotype x environment

  •  
  • H2

    heritability

  •  
  • ICARDA

    International Center for Agricultural Research in the Dry Areas

  •  
  • KW

    Kirschweg

  •  
  • LSD

    least significant difference

  •  
  • MTAs

    marker-trait associations

  •  
  • QTLs

    quantitative trait loci

  •  
  • REML

    residual maximum likelihood

  •  
  • SE

    Selkebreite

  •  
  • SMART

    Selection with Markers and Advanced Reproductive Technologies

  •  
  • UPGMA

    unweighted pair group method with arithmetic average

  •  
  • YSL

    yellow stripe like

Author Contribution

A.D., S.R., M.W., D.P.P. and V.C. performed experiments; A.D., M.N., A.B., J.K.S. and S.C. planned experiments and analyzed data. A.D., M.N. and S.C. wrote the paper. All authors approved the final version.

Funding

This work was in part financially supported by the European Union through its Sixth Framework Program for RTD (contract no FOOD-CT-2006-016253). Funding was also supplied from the Independent Research Fund Denmark—Natural Sciences [grant 4002-00181]; The molecular Fe and Zn speciation in cereals: Unravelling the chemistry controlling human bioavailability of essential trace elements).

Acknowledgements

We thank Angelika Mustroph, Department of Biology, University of Bayreuth, for the help with blastp searches regarding A. thaliana metal homeostasis genes and with microarray data analysis. We are grateful to Stefan Holzheu, Bayreuth Center for Ecology and Environmental Research, University of Bayreuth, for his advice on statistical analysis, and to Christiane Meinen, Pia Schuster and Silke Matros for their expert technical assistance.

Competing Interests

The authors declare that there are no competing interests associated with the manuscript.

References

References
1
Sandstead
,
H.
(
1991
)
Zinc deficiency. A public health problem?
Am. J. Dis. Child.
145
,
853
859
2
Wessells
,
K.R.
and
Brown
,
K.H.
(
2012
)
Estimating the global prevalence of zinc deficiency. Results based on zinc availability in national food supplies and the prevalence of stunting
.
PLoS ONE
7
,
e50568
3
WHO
. (
2009
)
Global Health Risks: Mortality and Burden of Disease Attributable to Selected Major Risks. Mortality and Burden of Disease Attributable to Selected Major Risks
,
World Health Organization
,
Geneva, Switzerland
4
Singh
,
M.
(
2004
)
Role of micronutrients for physical growth and mental development
.
Ind. J. Pediatr.
71
,
59
62
5
Bewley
,
J.D.
and
Black
,
M.
(
2013
)
Seeds. Physiology of Development and Germination
,
Springer
,
New York, U.S.A.
6
Myers
,
S.S.
,
Zanobetti
,
A.
,
Kloog
,
I.
,
Huybers
,
P.
,
Leakey
,
A.D.B.
,
Bloom
,
A.J.
, et al.  (
2014
)
Increasing CO2 threatens human nutrition
.
Nature
510
,
139
142
7
Dietterich
,
L.H.
,
Zanobetti
,
A.
,
Kloog
,
I.
,
Huybers
,
P.
,
Leakey
,
A.D.B.
,
Bloom
,
A.J.
, et al.  (
2015
)
Impacts of elevated atmospheric CO2 on nutrient content of important food crops
.
Sci. Data
2
,
150036
8
Shahzad
,
Z.
,
Rouached
,
H.
and
Rakha
,
A.
(
2014
)
Combating mineral malnutrition through iron and zinc biofortification of cereals
.
Compr. Rev. Food Sci. Food Saf.
13
,
329
346
9
Palmgren
,
M.G.
,
Clemens
,
S.
,
Williams
,
L.E.
,
Krämer
,
U.
,
Borg
,
S.
,
Schjørring
,
J.K.
et al.  (
2008
)
Zinc biofortification of cereals: problems and solutions
.
Trends Plant Sci.
13
,
464
473
10
Murgia
,
I.
,
Arosio
,
P.
,
Tarantino
,
D.
and
Soave
,
C.
(
2012
)
Biofortification for combating ‘hidden hunger’ for iron
.
Trends Plant Sci.
17
,
47
55
11
Welch
,
R.M.
(
2002
)
Breeding strategies for biofortified staple plant foods to reduce micronutrient malnutrition globally
.
J. Nutr.
132
,
495S
499S
12
Darnton-Hill
,
I.
and
Nalubola
,
R.
(
2002
)
Fortification strategies to meet micronutrient needs: successes and failures
.
Proc. Nutr. Soc.
61
,
231
241
13
Gómez-Galera
,
S.
,
Rojas
,
E.
,
Sudhakar
,
D.
,
Zhu
,
C.
,
Pelacho
,
A.M.
,
Capell
,
T.
et al.  (
2010
)
Critical evaluation of strategies for mineral fortification of staple food crops
.
Transgenic Res.
19
,
165
180
14
Waters
,
B.M.
and
Sankaran
,
R.P.
(
2011
)
Moving micronutrients from the soil to the seeds. Genes and physiological processes from a biofortification perspective
.
Plant Sci.
180
,
562
574
15
Schroeder
,
J.I.
,
Delhaize
,
E.
,
Frommer
,
W.B.
,
Guerinot
,
M.L.
,
Harrison
,
M.J.
,
Herrera-Estrella
,
L.
, et al.  (
2013
)
Using membrane transporters to improve crops for sustainable food production
.
Nature
497
,
60
66
16
Bothmer
,
R.V.
(
2003
)
Diversity in Barley. (Hordeum vulgare)
,
Elsevier
,
Amsterdam, Boston
17
Sreenivasulu
,
N.
,
Graner
,
A.
and
Wobus
,
U.
(
2008
)
Barley genomics. An overview
.
Int. J. Plant Genomics
2008
,
486258
18
International Barley Genome Sequencing
,
Mayer
,
K.F.
,
Waugh
,
R.
,
Brown
,
J.W.S.
,
Schulman
,
A.
,
Langridge
,
P.
,
Platzer
,
M.
, et al.  (
2012
)
A physical, genetic and functional sequence assembly of the barley genome
.
Nature
491
,
711
716
19
Ullrich
,
S.E.
(
2011
)
Barley, Production, Improvement, and Uses
,
Wiley-Blackwell
,
Chichester, West Sussex, UK
;
Ames, Iowa
20
Jilal
,
A.
,
Grando
,
S.
,
Henry
,
R.J.
,
Lee
,
L.S.
,
Rice
,
N.
,
Hill
,
H.
et al.  (
2008
)
Genetic diversity of ICARDA's worldwide barley landrace collection
.
Genet. Resour. Crop Evol.
55
,
1221
1230
21
Nagel
,
M.
,
Kranner
,
I.
,
Neumann
,
K.
,
Rolletschek
,
H.
,
Seal
,
C.E.
,
Colville
,
L.
et al.  (
2015
)
Genome-wide association mapping and biochemical markers reveal that seed ageing and longevity are intricately affected by genetic background and developmental and environmental conditions in barley
.
Plant Cell Environ.
38
,
1011
1022
22
Varshney
,
R.K.
,
Paulo
,
M.J.
,
Grando
,
S.
,
van Eeuwijk
,
F.A.
,
Keizer
,
L.C.P.
,
Guo
,
P.
et al.  (
2012
)
Genome wide association analyses for drought tolerance related traits in barley (Hordeum vulgare L.)
.
Field Crops Res.
126
,
171
180
23
Mascher
,
M.
,
Gundlach
,
H.
,
Himmelbach
,
A.
,
Beier
,
S.
,
Twardziok
,
S.O.
,
Wicker
,
T.
, et al.  (
2017
)
A chromosome conformation capture ordered sequence of the barley genome
.
Nature
544
,
427
24
Detterbeck
,
A.
,
Pongrac
,
P.
,
Rensch
,
S.
,
Reuscher
,
S.
,
Pecovnik
,
M.
,
Vavpetic
,
P.
et al.  (
2016
)
Spatially resolved analysis of variation in barley (Hordeum vulgare) grain micronutrient accumulation
.
New Phytol.
211
,
1241
1254
25
Mamo
,
B.E.
,
Barber
,
B.L.
and
Steffenson
,
B.J.
(
2014
)
Genome-wide association mapping of zinc and iron concentration in barley landraces from Ethiopia and Eritrea
.
J. Cereal Sci.
60
,
497
506
26
Cao
,
F.
,
Chen
,
F.
,
Sun
,
H.
,
Zhang
,
G.
,
Chen
,
Z.-H.
and
Wu
,
F.
(
2014
)
Genome-wide transcriptome and functional analysis of two contrasting genotypes reveals key genes for cadmium tolerance in barley
.
BMC Genomics
15
,
611
27
Nanda
,
A.K.
,
Pujol
,
V.
and
Wissuwa
,
M.
(
2017
)
Patterns of stress response and tolerance based on transcriptome profiling of rice crown tissue under zinc deficiency
.
J. Exp. Bot.
68
,
1715
1729
28
Tauris
,
B.
,
Borg
,
S.
,
Gregersen
,
P.L.
and
Holm
,
P.B.
(
2009
)
A roadmap for zinc trafficking in the developing barley grain based on laser capture microdissection and gene expression profiling
.
J. Exp. Bot.
60
,
1333
1347
29
Gubatz
,
S.
,
Dercksen
,
V.J.
,
Brüss
,
C.
,
Weschke
,
W.
and
Wobus
,
U.
(
2007
)
Analysis of barley (Hordeum vulgare) grain development using three-dimensional digital models
.
Plant J.
52
,
779
790
30
Mariotti
,
F.
,
Tomé
,
D.
and
Mirand
,
P.P.
(
2008
)
Converting nitrogen into protein–beyond 6.25 and Jones’ factors
.
Crit. Rev. Food Sci. Nutr.
48
,
177
184
31
Pritchard
,
J.K.
,
Stephens
,
M.
and
Donnelly
,
P.
(
2000
)
Inference of population structure using multilocus genotype data
.
Genetics
155
,
945
959
PMID:
[PubMed]
32
Bradbury
,
P.J.
,
Zhang
,
Z.
,
Kroon
,
D.E.
,
Casstevens
,
T.M.
,
Ramdoss
,
Y.
and
Buckler
,
E.S.
(
2007
)
TASSEL: software for association mapping of complex traits in diverse samples
.
Bioinformatics (Oxford, England)
23
,
2633
2635
33
VSN International
. (
2013
)
GenStat for Windows 17th Edition
,
VSN International
,
Hemel Hempstead, UK
34
Warnes
,
G.R.
,
Bolker
,
B.
,
Bonebakker
,
L.
,
Gentleman
,
R.
,
Huber
,
W.
,
Liaw
,
A.
et al.  (
2016
)
gplots: Various R Programming Tools for Plotting Data. R package version 3.0.1
35
Liu
,
G.
(
2003
)
Netaffx. Affymetrix probesets and annotations
.
Nucleic Acids Res.
31
,
82
86
36
Suryawanshi
,
V.
,
Talke
,
I.N.
,
Weber
,
M.
,
Eils
,
R.
,
Brors
,
B.
,
Clemens
,
S.
et al.  (
2016
)
Between-species differences in gene copy number are enriched among functions critical for adaptive evolution in Arabidopsis halleri
.
BMC Genomics
17
,
1034
37
Faccioli
,
P.
,
Ciceri
,
G.P.
,
Provero
,
P.
,
Stanca
,
A.M.
,
Morcia
,
C.
and
Terzi
,
V.
(
2007
)
A combined strategy of ‘in silico’ transcriptome analysis and web search engine optimization allows an agile identification of reference genes suitable for normalization in gene expression studies
.
Plant Mol. Biol.
63
,
679
688
38
Deinlein
,
U.
,
Weber
,
M.
,
Schmidt
,
H.
,
Rensch
,
S.
,
Trampczynska
,
A.
,
Hansen
,
T.H.
, et al.  (
2012
)
Elevated nicotianamine levels in Arabidopsis halleri roots play a key role in zinc hyperaccumulation
.
Plant Cell Online
24
,
708
723
39
R Core Team
R: A Language and Environment for Statistical Computing
,
R Foundation for Statistical Computing
,
Vienna, Austria
40
Hothorn
,
T.
,
Bretz
,
F.
and
Westfall
,
P.
(
2008
)
Simultaneous inference in general parametric models
.
Biom. J.
50
,
346
363
41
Mendiburu
,
F.
(
2013
)
agricolae. Statistical Procedures for Agricultural Research. R package version 1.1-6
42
WHO The world health report
. (
2002
)
Reducing Risks, Promoting Healthy Life
43
Muthayya
,
S.
,
Rah
,
J.H.
,
Sugimoto
,
J.D.
,
Roos
,
F.F.
,
Kraemer
,
K.
and
Black
,
R.E.
(
2013
)
The global hidden hunger indices and maps. An advocacy tool for action
.
PLoS ONE
8
,
e67860
44
Ruel-Bergeron
,
J.C.
,
Stevens
,
G.A.
,
Sugimoto
,
J.D.
,
Roos
,
F.F.
,
Ezzati
,
M.
,
Black
,
R.E.
et al.  (
2015
)
Global update and trends of hidden hunger, 1995–2011. The Hidden Hunger Index
.
PLoS ONE
10
,
e0143497
45
Tulchinsky
,
T.H.
(
2010
)
Micronutrient deficiency conditions. Global health issues
.
Public Health Rev.
32
,
243
255
46
Clemens
,
S.
(
2014
)
Zn and Fe biofortification. The right chemical environment for human bioavailability
.
Plant Sci.
225
,
52
57
47
Garcia-Oliveira
,
A.L.
,
Chander
,
S.
,
Ortiz
,
R.
,
Menkir
,
A.
and
Gedil
,
M.
(
2018
)
Genetic basis and breeding perspectives of grain iron and zinc enrichment in cereals
.
Front. Plant Sci.
9
,
937
48
Olsen
,
L.I.
and
Palmgren
,
M.G.
(
2014
)
Many rivers to cross: the journey of zinc from soil to seed
.
Front. Plant Sci.
5
,
30
49
Xu
,
Y.
,
An
,
D.
,
Li
,
H.
and
Xu
,
H.
(
2011
)
Review: Breeding wheat for enhanced micronutrients
.
Can. J. Plant Sci.
91
,
231
237
50
Reuscher
,
S.
,
Kolter
,
A.
,
Hoffmann
,
A.
,
Pillen
,
K.
,
Krämer
,
U.
and
Perovic
,
D.
(
2016
)
Quantitative trait loci and inter-organ partitioning for essential metal and toxic analogue accumulation in barley
.
PLoS ONE
11
,
e0153392
51
Zohary
,
D.
,
Hopf
,
M.
and
Weiss
,
E.
(
2012
)
Domestication of Plants in the old World. The Origin and Spread of Domesticated Plants in South-West Asia, Europe and the Mediterranean Basin
,
Oxford University Press
,
Oxford
52
Grando
,
S.
and
Macpherson
,
H.G.
(
2005
)
Food Barley: Importance, Uses and Local Knowledge
.
Proceedings of the International Workshop on Food Barley Improvement, 14–17 January 2002
,
Hammamet, Tunisia, Aleppo, Syria
53
Ma
,
J.F.
,
Higashitani
,
A.
,
Sato
,
K.
and
Takeda
,
K.
(
2004
)
Genotypic variation in Fe concentration of barley grain
.
Soil Sci. Plant Nutr.
50
,
1115
1117
54
Zhao
,
F.-J.
,
Su
,
Y.H.
,
Dunham
,
S.J.
,
Rakszegi
,
M.
,
Bedo
,
Z.
,
McGrath
,
S.P.
et al.  (
2009
)
Variation in mineral micronutrient concentrations in grain of wheat lines of diverse origin
.
J. Cereal Sci.
49
,
290
295
55
Chatzav
,
M.
,
Peleg
,
Z.
,
Ozturk
,
L.
,
Yazici
,
A.
,
Fahima
,
T.
,
Cakmak
,
I.
et al.  (
2010
)
Genetic diversity for grain nutrients in wild emmer wheat. Potential for wheat improvement
.
Ann. Bot.
105
,
1211
1220
56
Pinson
,
S.
,
Tarpley
,
L.
,
Yan
,
W.
,
Yeater
,
K.
,
Lahner
,
B.
,
Yakubova
,
E.
et al.  (
2015
)
Worldwide genetic diversity for mineral element concentrations in rice grain
.
Crop Sci.
55
,
294
57
Shi
,
J.
,
Li
,
L.
and
Pan
,
G.
(
2009
)
Variation of grain Cd and Zn concentrations of 110 hybrid rice cultivars grown in a low-Cd paddy soil
.
J. Environ. Sci.
21
,
168
172
58
Dawson
,
I.K.
,
Russell
,
J.
,
Powell
,
W.
,
Steffenson
,
B.
,
Thomas
,
W.T.B.
and
Waugh
,
R.
(
2015
)
Barley. A translational model for adaptation to climate change
.
New Phytol.
206
,
913
931
59
White
,
P.J.
and
Broadley
,
M.R.
(
2009
)
Biofortification of crops with seven mineral elements often lacking in human diets–iron, zinc, copper, calcium, magnesium, selenium and iodine
.
New Phytol.
182
,
49
84
60
Phuke
,
R.M.
,
Anuradha
,
K.
,
Radhika
,
K.
,
Jabeen
,
F.
,
Anuradha
,
G.
,
Ramesh
,
T.
, et al.  (
2017
)
Genetic variability, genotype × environment interaction, correlation, and GGE biplot analysis for grain iron and zinc concentration and other agronomic traits in RIL population of sorghum (Sorghum bicolor L. Moench)
.
Front. Plant Sci.
8
,
712
61
Cakmak
,
I.
,
Turun
,
A.
,
Millet
,
E.
,
Feldman
,
M.
,
Fahima
,
T.
,
Korol
,
A.
et al.  (
2004
)
Triticum dicoccoides. An important genetic resource for increasing zinc and iron concentration in modern cultivated wheat
.
Soil Sci. Plant Nutr.
50
,
1047
1054
62
Morgounov
,
A.
,
Gómez-Becerra
,
H.F.
,
Abugalieva
,
A.
,
Dzhunusova
,
M.
,
Yessimbekova
,
M.
,
Muminjanov
,
H.
et al.  (
2007
)
Iron and zinc grain density in common wheat grown in Central Asia
.
Euphytica
155
,
193
203
63
Peleg
,
Z.
,
Saranga
,
Y.
,
Yazici
,
A.
,
Fahima
,
T.
,
Ozturk
,
L.
and
Cakmak
,
I.
(
2008
)
Grain zinc, iron and protein concentrations and zinc-efficiency in wild emmer wheat under contrasting irrigation regimes
.
Plant Soil
306
,
57
67
64
Guttieri
,
M.J.
,
Baenziger
,
P.S.
,
Frels
,
K.
,
Carver
,
B.
,
Arnall
,
B.
and
Waters
,
B.M.
(
2015
)
Variation for grain mineral concentration in a diversity panel of current and historical great plains hard winter wheat germplasm
.
Crop Sci.
55
,
1035
65
Jiang
,
S.L.
,
Wu
,
J.G.
,
Feng
,
Y.
,
Yang
,
X.E.
and
Shi
,
C.H.
(
2007
)
Correlation analysis of mineral element contents and quality traits in milled rice (Oryza sativa L.)
.
J. Agric. Food Chem.
55
,
9608
9613
66
Bashir
,
E.M.A.
,
Ali
,
A.M.
,
Ali
,
A.M.
,
Melchinger
,
A.E.
,
Parzies
,
H.K.
and
Haussmann
,
B.I.G.
(
2014
)
Characterization of Sudanese pearl millet germplasm for agro-morphological traits and grain nutritional values
.
Plant Genet. Resour.
12
,
35
47
67
Singh
,
J.
,
Kanaujia
,
R.
,
Srivastava
,
A.K.
,
Dixit
,
G.P.
and
Singh
,
N.P.
(
2017
)
Genetic variability for iron and zinc as well as antinutrients affecting bioavailability in black gram (Vigna mungo (L.) Hepper)
.
J. Food Sci. Technol.
54
,
1035
1042
68
Uauy
,
C.
,
Distelfeld
,
A.
,
Fahima
,
T.
,
Blechl
,
A.
and
Dubcovsky
,
J.
(
2006
)
A NAC Gene regulating senescence improves grain protein, zinc, and iron content in wheat
.
Science
314
,
1298
1301
69
Persson
,
D.P.
,
Hansen
,
T.H.
,
Laursen
,
K.H.
,
Schjørring
,
J.K.
and
Husted
,
S.
(
2009
)
Simultaneous iron, zinc, sulfur and phosphorus speciation analysis of barley grain tissues using SEC-ICP-MS and IP-ICP-MS
.
Metallomics
1
,
418
70
Eagling
,
T.
,
Neal
,
A.L.
,
McGrath
,
S.P.
,
Fairweather-Tait
,
S.
,
Shewry
,
P.R.
and
Zhao
,
F.-J.
(
2014
)
Distribution and speciation of iron and zinc in grain of two wheat genotypes
.
J. Agric. Food Chem.
62
,
708
716
71
Anandan
,
A.
,
Rajiv
,
G.
,
Eswaran
,
R.
and
Prakash
,
M.
(
2011
)
Genotypic variation and relationships between quality traits and trace elements in traditional and improved rice (Oryza sativa L.) genotypes
.
J. Food Sci.
76
,
H122
H130
72
Gyawali
,
S.
,
Otte
,
M.L.
,
Chao
,
S.
,
Jilal
,
A.
,
Jacob
,
D.L.
,
Amezrou
,
R.
et al.  (
2017
)
Genome wide association studies (GWAS) of element contents in grain with a special focus on zinc and iron in a world collection of barley (Hordeum vulgare L.)
.
J. Cereal Sc.
77
,
266
274
73
Conte
,
S.S.
and
Walker
,
E.L.
(
2012
)
Genetic and biochemical approaches for studying the yellow stripe-like transporter family in plants
.
Curr. Topics Membr.
69
,
295
322
74
Chu
,
H.-H.
,
Chiecko
,
J.
,
Punshon
,
T.
,
Lanzirotti
,
A.
,
Lahner
,
B.
,
Salt
,
D.E.
et al.  (
2010
)
Successful reproduction requires the function of Arabidopsis yellow stripe-like1 and yellow stripe-like3 metal-nicotianamine transporters in both vegetative and reproductive structures
.
Plant Physiol.
154
,
197
210
75
Waters
,
B.M.
,
Chu
,
H.-H.
,
Didonato
,
R.J.
,
Roberts
,
L.A.
,
Eisley
,
R.B.
,
Lahner
,
B.
et al.  (
2006
)
Mutations in Arabidopsis yellow stripe-like1 and yellow stripe-like3 reveal their roles in metal ion homeostasis and loading of metal ions in seeds
.
Plant Physiol.
141
,
1446
1458
76
Senoura
,
T.
,
Sakashita
,
E.
,
Kobayashi
,
T.
,
Takahashi
,
M.
,
Aung
,
M.S.
,
Masuda
,
H.
et al.  (
2017
)
The iron-chelate transporter OsYSL9 plays a role in iron distribution in developing rice grains
.
Plant Mol. Biol.
95
,
375
387
77
Fan
,
M.-S.
,
Zhao
,
F.-J.
,
Fairweather-Tait
,
S.J.
,
Poulton
,
P.R.
,
Dunham
,
S.J.
and
McGrath
,
S.P.
(
2008
)
Evidence of decreasing mineral density in wheat grain over the last 160 years
.
J. Trace Elem. Med. Biol.
22
,
315
324
78
Hebelstrup
,
K.H.
(
2017
)
Differences in nutritional quality between wild and domesticated forms of barley and emmer wheat
.
Plant Sci.
256
,
1
4
79
Pinheiro de Carvalho
,
M.A.A.
,
Bebeli
,
P.J.
,
Bettencourt
,
E.
,
Costa
,
G.
,
Dias
,
S.
,
Dos Santos
,
T.M.M.
et al.  (
2013
)
Cereal landraces genetic resources in worldwide GeneBanks. A review
.
Agron. Sustain. Dev.
33
,
177
203
80
Liu
,
D.
,
Liu
,
Y.
,
Zhang
,
W.
,
Chen
,
X.
and
Zou
,
C.
(
2017
)
Agronomic approach of zinc biofortification can increase zinc bioavailability in wheat flour and thereby reduce zinc deficiency in humans
.
Nutrients
9
,
E465
81
Zhang
,
Y.
,
Shi
,
R.
,
Rezaul
,
K.M.
,
Zhang
,
F.
and
Zou
,
C.
(
2010
)
Iron and zinc concentrations in grain and flour of winter wheat as affected by foliar application
.
J. Agric. Food Chem.
58
,
12268
12274
82
Xue
,
Y.-F.
,
Eagling
,
T.
,
He
,
J.
,
Zou
,
C.-Q.
,
McGrath
,
S.P.
,
Shewry
,
P.R.
et al.  (
2014
)
Effects of nitrogen on the distribution and chemical speciation of iron and zinc in pearling fractions of wheat grain
.
J. Agric. Food Chem.
62
,
4738
4746
83
Menguer
,
P.K.
,
Vincent
,
T.
,
Miller
,
A.J.
,
Brown
,
J.K.M.
,
Vincze
,
E.
,
Borg
,
S.
et al.  (
2018
)
Improving zinc accumulation in cereal endosperm using HvMTP1, a transition metal transporter
.
Plant Biotechnol. J.
16
,
63
71
84
Tan
,
J.
,
Wang
,
J.
,
Chai
,
T.
,
Zhang
,
Y.
,
Feng
,
S.
,
Li
,
Y.
et al.  (
2013
)
Functional analyses of TaHMA2, a P(1B)-type ATPase in wheat
.
Plant Biotechnol. J.
11
,
420
431
85
Pielot
,
R.
,
Kohl
,
S.
,
Manz
,
B.
,
Rutten
,
T.
,
Weier
,
D.
,
Tarkowská
,
D.
, et al.  (
2015
)
Hormone-mediated growth dynamics of the barley pericarp as revealed by magnetic resonance imaging and transcript profiling
.
J. Exp. Bot.
66
,
6927
6943
86
Sreenivasulu
,
N.
,
Borisjuk
,
L.
,
Junker
,
B.H.
,
Mock
,
H.-P.
,
Rolletschek
,
H.
,
Seiffert
,
U.
et al.  (
2010
)
Barley grain development: toward an integrative view
.
Int. Rev. Cell Mol. Biol.
281
,
49
89
87
Ozturk
,
L.
,
Yazici
,
M.A.
,
Yucel
,
C.
,
Torun
,
A.
,
Cekic
,
C.
,
Bagci
,
A.
et al.  (
2006
)
Concentration and localization of zinc during seed development and germination in wheat
.
Physiol. Plant
128
,
144
152
88
Impa
,
S.M.
,
Morete
,
M.J.
,
Ismail
,
A.M.
,
Schulin
,
R.
and
Johnson-Beebout
,
S.E.
(
2013
)
Zn uptake, translocation and grain Zn loading in rice (Oryza sativa L.) genotypes selected for Zn deficiency tolerance and high grain Zn
.
J. Exp. Bot.
64
,
2739
2751
89
Waters
,
B.M.
,
Uauy
,
C.
,
Dubcovsky
,
J.
and
Grusak
,
M.A.
(
2009
)
Wheat (Triticum aestivum) NAM proteins regulate the translocation of iron, zinc, and nitrogen compounds from vegetative tissues to grain
.
J. Exp. Bot.
60
,
4263
4274
90
Hussain
,
S.
,
Rengel
,
Z.
,
Mohammadi
,
S.A.
,
Ebadi-Segherloo
,
A.
and
Maqsood
,
M.A.
(
2016
)
Mapping QTL associated with remobilization of zinc from vegetative tissues into grains of barley (Hordeum vulgare)
.
Plant Soil
399
,
193
208
91
Sperotto
,
R.A.
(
2013
)
Zn/Fe remobilization from vegetative tissues to rice seeds. Should I stay or should I go? Ask Zn/Fe supply!
.
Front. Plant Sci.
4
,
464
92
Maillard
,
A.
,
Diquélou
,
S.
,
Billard
,
V.
,
Laîné
,
P.
,
Garnica
,
M.
,
Prudent
,
M.
et al.  (
2015
)
Leaf mineral nutrient remobilization during leaf senescence and modulation by nutrient deficiency
.
Front. Plant Sci.
6
,
317
93
Ricachenevsky
,
F.K.
,
Menguer
,
P.K.
,
Sperotto
,
R.A.
,
Williams
,
L.E.
and
Fett
,
J.P.
(
2013
)
Roles of plant metal tolerance proteins (MTP) in metal storage and potential use in biofortification strategies
.
Front. Plant Sci.
4
,
144
94
Podar
,
D.
,
Scherer
,
J.
,
Noordally
,
Z.
,
Herzyk
,
P.
,
Nies
,
D.
and
Sanders
,
D.
(
2012
)
Metal selectivity determinants in a family of transition metal transporters
.
J. Biol. Chem.
287
,
3185
3196