Novel sources of variation in grain Zinc (Zn) concentration in bread wheat germplasm derived from Watkins landraces

A diverse panel of 245 wheat genotypes, derived from crosses between landraces from the Watkins collection representing global diversity in the early 20th century and the modern wheat cultivar Paragon, was grown at two field sites in the UK in 2015–16 and the concentrations of zinc and iron determined in wholegrain using inductively coupled plasma-mass spectrometry (ICP-MS). Zinc concentrations in wholegrain varied from 24–49 mg kg-1 and were correlated with iron concentration (r = 0.64) and grain protein content (r = 0.14). However, the correlation with yield was low (r = -0.16) indicating little yield dilution. A sub-set of 24 wheat lines were selected from 245 wheat genotypes and characterised for Zn and Fe concentrations in wholegrain and white flour over two sites and years. White flours from 24 selected lines contained 8–15 mg kg-1 of zinc, which was positively correlated with the wholegrain Zn concentration (r = 0.79, averaged across sites and years). This demonstrates the potential to exploit the diversity in landraces to increase the concentration of Zn in wholegrain and flour of modern high yielding bread wheat cultivars.


Introduction
Zinc (Zn) and iron (Fe) are essential micronutrients for human health. Inadequate dietary intakes of Zn and Fe are linked to diarrhoea, stunting (low height for age in childrem) and anaemia, and to increased mortality [1][2][3]. Approximately, 20% of the global population has insufficient Zn in their diets, rising to 50% in parts of South Asia and sub-Saharan countries where diets are based on cereals crops including wheat [4]. In the UK, 11% children between 4 and 10 years of age have Zn intakes below than lower reference nutrient intake value and Fe deficiency has increased up to 9% in adolescent girls and 5% in women. After red meat and processed meats, bread wheat and products are the main sources of dietary Zn and Fe intake in the UK population [5,6].
Bread wheat (Triticum aestivum L.) is an important cereal crop globally and provides up to 60% of the total energy intake in some countries (FAOSTAT, 2013; http://www.faostst.fao.org) and is also a major source of daily dietary Zn and Fe [7]. Modern  whole-grain Zn concentration less than the target of 40 mg kg -1 considered appropriate for human diets under field conditions in S. Asia [8]. For example, grain Zn concentration ranged from 25-35 mg kg -1 in a panel of 36 widely adapted Indian wheat genotypes under field conditions [9]. Grain Zn concertation in wheat can be increased by agronomic and genetic biofortification. Agronomic biofortification involves the use of Zn fertilisers, which can increase both crop yield and grain Zn concentration in wheat. However, it also increases the cost and may require to changes to current farming practices because Zn applied to soils at normal agronomic rates is unlikely to have large effects on grain Zn concentration [8,10,11]. Agronomical biofortified wheat flour using foliar applications was provided to children between 4 to 6 years of age for six months and showed 17% and 40% reduction in incidences of pneumonia and vomiting, respectively [12].
Genetic Zn biofortification of wheat is likely to be more cost-effective in the longer term than the use of fertilisers, but requires the existence of genetic variation in wheat germplasm. Wild wheat species and landraces have shown variation for grain Zn concentration in wheat and may therefore be used to achieve the breeding target of 40 mg kg -1 grain Zn in wheat [8,11,13]. The 'HarvestPlus' programme has been the primary driver of genetic biofortification in wheat. It has developed and deployed high-Zn wheat varieties in South Asia using synthetic wheat lines derived from wild wheat relatives, Aegilops tauschii (D genome donor of wheat), Triticum spelta and wild T. dicoccon. HarvestPlus has released locally adapted high-Zn wheat varieties in India and Pakistan with 30-40% higher Zn in grain than a notional baseline of 25 mg kg -1 grain Zn [14]. Cakmak et al. [15] reported grain Zn concentration ranging from 30-98 mg kg -1 in 518 accessions of T. dicoccoides grown in the field, while Guzman et al. [16] screened 93 advance lines, which were developed by the crossing of Mexican landraces, Triticum dicoccoides and Ae. tauschii with durum and wheat genotypes, grown under field condition in Mexico and reported grain Zn concentration from 27-53 mg kg -1 . Similarly, Srinivasa et al. [17] reported grain Zn concentration from 25-60 mg kg -1 in 185 recombinant inbred lines derived from crossing between T. spelta and T. aestivum grown under field conditions.
Most studies of wild species, landraces and their derivatives report Zn concentration in whole grain not in wheat white flour (endosperm). Whereas the wheat flour (atta) used to prepare unleavened flatbreads (chapattis or rotis) in South Asia comprises about 95% of the whole grain, white wheat flour is commonly used to make leavened bread (yeast fermented) and other products in Europe, the USA and Oceania countries (Australia and New Zealand) [18]. This corresponds to the starchy endosperm tissue and is produced by milling to remove the aleurone, outer layers and bran. Because Zn accumulates in the aleurone layer and embryo [19][20][21], white flour may contain <10 mg kg -1 Zn which is not sufficient for human diet [20]. Furthermore, the bioavailability of the Zn in the aleurone and embryo is reduced by the presence of phytic acid, the major storage form of phosphorus which is present in the same tissues [22]. Phytate forms insoluble complexes with Zn and inhibits its absorption in monogastric animals [23]. Therefore, screening of wider germplasm for variation in Zn concentration in white flour should allow the development of wheat varieties with more bioavailable Zn in white flour.
The objectives of this study were to characterise a diverse panel of wheat genotypes, derived from landraces, and the UK cultivar Paragon, grown on two sites in the UK for variation in wholegrain Zn concentration. Based on these analyses a sub-set of genotypes were selected and analysed for Zn in both wholegrain and white flours using material grown in two years and on two sites. This identified lines which are suitable for exploitation in breeding to produce cultivars with high bioavailable Zn.

Plant material
A panel of 245 wheat genotypes, derived from crosses between landraces from the Watkins collection and the UK Spring wheat cultivar Paragon, were grown at Bunny Farm site of the University of Nottingham in 2015-16 and 2016-17 and, at the Rothamsted Research farm in 2015-16 and 2017-18. The landraces were collected from 34 countries by A.E. Watkins in the early 1930s [24]. Marker analysis allowed a core set of 119 lines to be identified and 85 of these crossed with Paragon to develop F1 progeny, followed by four to six rounds of selfing (F4: F6) to develop populations of recombinant inbred lines [24,25]. Each population comprises 94 lines, and 12 or 13 lines from 19 populations were selected for the present project to constitute a panel of 245 wheat genotypes (S1 Table). These lines were selected based on their performance (including yields) and adaptation to the UK (including flowering time).

Experimental sites and management
The details of field experiments conducted at the University of Nottingham and Rothamsted Research, sites are described in Table 1.
The plots were arranged in a randomised factorial block design with three replications at both the sites. The plot size was 1 m 2 . To achieve a target population size of 200 plants m -2 , seed rates were selected on 1000 grain weight (TGW) basis and,~350 seeds m -2 were used per plot. Seeds were planted using a power drill after field preparation.
Nitrogen fertiliser at Nottingham was applied at the rate of 175 kg ha -1 as Nitram (34.5% N), one-third of total N was applied at sowing time and the remaining was applied at growth stage 31, onset of stem extension (GS31) in 2015-16 and 2016-17. At Rothamsted nitrogen fertilisers were applied in two doses; first dose at the rate of 222 kg ha -1 as double top (27% N) and 2 nd dose @ of 101 kg ha -1 as ammonium nitrate (34.5% N) in 2015-16 while in 2017-18 only a single dose of ammonium nitrate was applied at the rate of 370 kg ha -1 . All the plots were managed and protected from pest, diseases, and weeds by using herbicides and pesticides to ensure to get the maximum yield. Crops were harvested at maturity by a small combine harvester.

Grain sampling and digestion of 245 wheat genotypes
Ears were harvested from an area of 0.5-meter length from central rows of each plot, dried at 80˚C for 48 hours in the oven and then threshed mechanically. Approximately 10 grains of all 245 wheat genotypes in three replicates from Nottingham (n = 735) and in one replicate from Rothamsted site (n = 245) in 2015-16 were weighed and the grain samples were soaked in this solution, 3 mL 70% Trace Analysis Grade (TAG) HNO 3 and 2 mL hydrogen peroxide (H 2 O 2 ), overnight at room temperature. Grain was digested using a microwave system as described previously [9,26]. Briefly, this comprised a Multiwave 3000 platform with a 48-vessel MF50 rotor (Anton Paar Gmbh, Graz, Austria); digestion vessels were perfluoroalkoxy (PFA) tubes in polyethylethylketone (PEEK) pressure jackets (Anton Paar GmbH). Grains were digested in 3 mL 70% Trace Analysis Grade (TAG) HNO 3 and 2 mL hydrogen peroxide (H 2 O 2 ).
For total nitrogen analysis, wholegrain wheat subsamples were digested using the Kjeldahl method [27]. Briefly, the Kjeldahl system comprised a heating platform with a 42-vessel digestion unit, a scrubber unit with 20% sodium hydroxide neutraliser and a recirculating water pump (VELP, town, Germany). Wholegrain wheat subsampales (~0.1 g) were digested in 8 mL 95% sulphuric acid (Fisher Scientific UK Ltd, Loughborough, UK). The Kjeldahl settings were as follows: temp = 390˚C, water pump pressure = 1.5 bar, time = 120 minutes. Three operational blanks were included in each digestion run and, duplicate samples of certified reference material (CRM: Wheat flour SRM 1567b, NIST, Gaithersburg, MD, USA) were used in every third run. After digestion, each tube was made up to a final volume of 50 mL by adding 11 mL de-ionsed water and transferred to falcon tubes. The nitrogen concentration of wholegrain wheat subsamples were determined by using a microtiter plate and the colorimetric salicylic acid method [28]. Nitrogen % was converted into grain protein content (GPC%) by multiplying the grain N% with conversion factor 5.7 (N% x 5.7), according to the AACCI Methods 46-30 [29].

Milling and digestion of a sub-set 24 wheat genotypes
Based on the concentrations of grain Zn and Fe in the wholegrain samples of a panel of 245 wheat genotypes grown at Nottingham and Rothamsted in 2015-16, a sub-set of 24 genotypes was selected to capture the diversity present in the panel. Grain samples of these genotypes were resampled from three replicate plots per line from Nottingham in 2015-16, and from freshly harvested plots in 2016-17, and resampled from three replicate plots per line from Rothamsted in 2015-16, and from freshly harvested plots in 2017-18. Samples were dried at 40˚C overnight before milling, 6 g subsamples of grain were then milled at 30 rpm for 30 seconds using a Laboratory Flour Mill AQC 806 (Agromatic AG, Laupen, Switzerland) to obtain wheat white flour. Wholegrain samples (n = 129) of these sub-sets of genotypes, resampled from the Nottingham and Rothamsted 2015-16 material, were digested in a similar manner as the 245 wheat genotypes, with a soaking step. The subsample (~0.300 g DW) of wheat white flour samples (n = 129) were digested without soaking step. The wholegrain (n = 114) and white flour (n = 114) samples of these sub-sets of genotypes from Nottingham in 2016-17 and from Rothamsted in 2017-18 were digested using a Multiwave Pro platform with 41-vessels, HVT56 rotor system where microwave settings were: power = 1500 W, temperature = 175˚C, and time = 40 minutes. Wholegrain and white flour samples were digested in 6 mL 70% TAG HNO3 solution. The wholegrain samples were soaked in this solution for one hour at room temperature.
Two operational blanks were included in each digestion run. Duplicate samples of certified reference material (CRM: Wheat flour SRM 1567b, NIST, Gaithersburg, MD, USA) were included approximately every fourth digestion run. A laboratory reference material (LRM), Paragon was used for each digestion run. In 2015-16, following digestion, each tube was made up to a final volume of 15 mL by adding 11 mL Milli-Q water, then transferred to a 25 mL universal tube (Sarstedt Ltd., Nümbrecht, Germany) and stored at room temperature. Samples were further diluted 1:5 with Milli-Q water into 13 ml tubes (Sarstedt Ltd.) prior to analysis. In 2016-17, each tube was made up to a final volume of 24 mL by adding 18 mL Milli-Q water, then transferred to a 25 mL universal tube and stored at room temperature. Samples were further diluted 1:10 with Milli-Q water into 13 ml tubes prior to analysis.

Wholegrain and white flour Zn and Fe analysis
Wholegrain and white flour concentration of Zn and Fe were determined by ICP-MS (Thermo Fisher Scientific iCAPQ, Thermo Fisher Scientific, Bremen, Germany). The other elements also measured were Ag, Al, As, B, Ba, Be, Ca, Cd, Cr, Co, Cs, Cu, K, Li, Mg, Mn, Mo, Na, Ni, P, Pb, Rb, S, Se, Sr, Ti, Tl, U, and V, but these are not reported in this study. The other operation details and setting of ICP-MS were described in Khokhar et al. [9].

Statistical analysis
Variance components associated with Zn and Fe concentrations in wholegrain, GYD, TGW and GPC were calculated. Analyses of Variance (ANOVA) and Least Significant Difference (LSD) tests were used to test for differences in mean wholegrain elemental concentration between sites and wheat genotypes. Differences between sites and genotypes for wholegrain and white flour Zn and Fe concentrations traits were considered significant at P<0.001. Pearson correlation coefficients were calculated to study the relationships for Zn and Fe concentration and yield in the wholegrain and white flour. All analyses were conducted using GenStat 17 th Edition (VSN International Ltd, Hemel Hempstead, UK).

Variation in wholegrain Zn, Fe concentration grain yield, TGW and GPC of 245 Watkins derived wheat genotypes
The mean wholegrain Zn concentration across all plots (n = 980) was 33.6 mg kg -1 and varied from 15.6 to 60.1 mg kg -1 in 2015-16 ( Table 2). The mean wholegrain Fe concentration was 34.3 mg kg -1 and varied from 19.4 to 71.2 mg kg -1 , across all plots. The mean GYD was 5.2 t ha -1 and varied from 0.31 to 9.76 t ha -1 , across all plots. The mean TGW was 34.4 g and varied from 22.2 to 47.7 g, across all plots. The mean grain protein content (GPC) was 15.4% and varied from 5.62 to 27.3%, across all plots ( Table 2). Primary data of 245 lines are provided in S2 Table. Genotype accounted for 30% and 38% of the variation in wholegrain Zn and Fe concentrations, respectively (Table 3). Site accounted for a greater proportion of the variation in wholegrain Zn concentration (24%) than wholegrain Fe concentration (3.7%). The G � E interaction term was associated with 8% of the variation in wholegrain Zn and Fe concentrations. The residual term (R) was associated with 38% and 50% of the variation in wholegrain Zn and Fe concentration, respectively; this could be due to plot to plot variation and technical/measurement errors (Table 3). Genotype and site accounted for 23% and 56% of the variation in GYD, respectively. Genotype accounted for a greaer proportion of the variation in TGW (63%) than GPC (28%). Similarly, site accounted for a greater proportion of the variation in TGW (19%) than GPC (8%). The residual factor accounted for a greater proportion of the variation in GPC (52%) than in GYD (14%) and TGW (14%).

Variation in wholegrain Zn, Fe concentrations, grain yield, TGW and GPC between sites
Grain yield and wholegrain Zn concentration were lesser at Nottingham than at Rothamsted (P<0.001) (Fig 1).  and a non-significant negative weak correlation at Rothamsted (r = -0.12; P = 0.065). The TGW showed positive significant correlation with GYD at Nottingham (r = 0.14; P<0.05) and at Rothamsted (r = 0.27; P<0.001), but not correlated with GPC at both, Nottingham and Rothamsted sites. The grain protein content (GPC) showed significant negative correlation with GYD at Nottingham (r = -0.14; P<0.05) and at Rothamsted (r = -0.28; P<0.001).

Variation in wholegrain Zn, Fe concentrations, GYD, TGW and GPC between 245 wheat genotypes
The    (Table 4). The primary data of 24 wheat lines are provided in S3 Table. There Twenty four genotypes showed greater variation in white flour for Zn and Fe concentrations than that in wholegrain at both the Nottingham and Rothamsted sites (Table 4). In particular, the mean Zn concentration of the 24 genotypes, averaged across years, the Zn concentration of high-Zn genotypes was 64% higher in wholegrain and 68% higher in white flour compared to Zn concentration in wholegrain and white flour of the low-Zn genotypes when grown at Nottingham. Similarly, the mean Zn concentration of high-Zn genotypes was 42% higher in wholegrain and 44% higher in white flour compared to Zn concentration in wholegrain and white flour samples of low-Zn genotypes when grown at Rothamsted.

Relationships between Zn and Fe concentrations in wholegrain and white flour of a sub-set of 24 genotypes
The mean concentration of Fe in the 24 genotypes, averaged across years, the Fe concentration of high-Fe genotypes was 66% higher in wholegrain samples and 77% higher in white flour compared to Fe concentration in wholegrain and white flour of low-Fe genotypes when grown at Nottingham. Similarly, the Fe concentration of the high-Fe genotypes was 34% higher in wholegrain and 49% higher in white flour samples when compared to Fe concentration in wholegrain and white flour samples of low-Fe genotypes grown at Rothamsted (Table 4). Table 5 shows the concentration of minerals in the wholegrain and white flour of the four groups of genotypes (called groups 1-4) between sites and years. Groups 1, 2 and 4 showed significant differences in Zn concentrations in wholegrain and white flour between sites. Group 3 showed significant differences only in wholegrain Zn concentration between sites not in white flour Zn concentration. There were no significant differences in the Zn concentration in wholegrain and white flour of four different groups between the years (Table 5).
There were no significant differences in the Fe concentrations in wholegrain and white flour of groups 1, 2 and 3 between sites, while group 4 showed significant differences in wholegrain Fe concentration between sites, not in white flour Fe concentration. However, there were significant differences in the Fe concentration of wholegrain and white flour of all groups between the years except group 3 (Table 5).

Discussion
Despite the presence of high Zn and Fe concentrations in the grain of wheat landraces, only a few studies have exploited the potential of wheat landraces to increase the Zn and Fe levels of  [20]. There was a negative correlation between wholegrain Zn concentration and grain yield and grain protein content. Cakmak et al. and Oury et al. [20,35] reported lower grain Zn concentration and GPC in high yielding genotypes which may have been due to yield dilution. It is therefore important to continue to monitor grain Zn concentration in the wider context of yield and yield component traits in breeding programmes, especially given that grain yields in the region are relatively low. A sub-set of 24 genotypes showed strong relationships between the concentrations of Zn and Fe in wholegrain between the years at each site and between the sites. The subset, therefore, performed consistently for wholegrain Zn and Fe concentration between years and sites, indicating the stability of these traits under different growing conditions. However, the concentrations of Zn in wholegrain samples were higher at Rothamsted than at Nottingham in both the years. Grain Zn and Fe concentration are complex traits controlled by many genes and it is therefore not surprising that they are affected by soil and environmental factors. Khokhar et al [9] and Velu et al. [36] also reported significant effects of the environment on Table 5. One way analysis of variance (ANOVA) to study the performance of four groups with different Zn and Fe level combinations in wholegrain and white flour between the sites and years. Each group consist of 6 genotypes and data are 3 replicate plots per genotypes at each site and year. variation in grain Zn and Fe concentrations. Bread wheat is an important dietary source of Zn and Fe globally. Although most wheat products consumed in South Asia are made from high extraction flours (approx. 95% wholegrain), most wheat-based foods consumed in other parts of the world are made from white flour, which corresponds to the starchy endosperm and accounts for about 75-85% of wholegrain [4,18]. A further aim of the project was therefore to determine the variation in Zn and Fe concentration in wholegrain and white flour of a sub-set of 24 wheat genotypes selected to vary in their levels of both minerals. The contents of Zn in white flours of these genotypes were 56 to 72% lower than in wholegrain, with the concentrations of Fe showing similar differences. This is consistent with both minerals being concentrated in the embryo and aleurone layer of the grain [19,37,38], which are removed when milling to produce white flour. Similar differences in the mineral contents of wholegrain and white flours have been reported by Zhang et al. [39], Szira et al. [40] and Eagling et al. [41]. We also showed correlations between the concentrations of Zn in white flour and wholegrain samples, but not between the contents of Fe. This contrasts with Eagling et al. [41] who showed correlations between the contents of Fe in wholegrain and white flour of 6 commercial European genotypes. We have no explanation for this difference.

Levels Groups
In conclusion, the present study shows that there is substantial genetic diversity for wholegrain Zn and Fe concentration in a panel of 245 wheat genotypes derived from Watkins landraces and the UK elite cultivar Paragon. This should be available by exploitation by plant breeders, allowing the development of new biofortified cultivars with increased Zn and Fe concentrations in wholegrain and white flour, high grain yields, good agronomic performance, and good processing quality.