Seventy-one durum wheat (Triticum durum Desf.) cultivars and advanced breeding lines with 2 universal controls, 24 emmer wheat (Triticum dicoccum L.) genotypes and 11 wild emmer (Triticum dicoccoides L.) genotypes (Table 1) were used as genetic materials in this study. Canadian durum wheat cultivar “Commander” and advanced line “Dt 812” kindly provided by Dr Y. Ruan from Agriculture and Agri-Food Canada were used as negative and positive control, respectively. Ten emmer wheat and all wild emmer genotypes were obtained from Turkish Seed Gene Bank (Ankara, Turkey). Other emmer wheat lines used in the study were developed via selection breeding in a project funded by TUBITAK (The Scientific and Technological Research Council of Turkey, Project No: 214O401).

Seeds of all genotypes were sown on trays and then leaf samples were collected from plants for DNA extraction at 2-3 leaf stage. DNA was extracted according to cetyl trimethylammonium bromide (CTAB) method (18). The extracted DNA samples were loaded on agarose gel (Biomax, Thomas Scientific, Swedesboro, NJ, USA) with a DNA standard in order to determine the quality and concentration of the DNA and then were stored in sterile distilled water at -20 °C until use. To amplify Cdu1 gene alleles by PCR, the co-dominant CAPS marker usw47, which was derived from an expressed sequence tag (EST) XBF474090 co-segregating with Cdu1 (16), was used.

PCR was carried out as follows: the total volume of the reaction mixture was 15 μL containing 100 ng genomic DNA, 1× PCR buffer (Sigma Aldrich, Merck, St. Louis, MO, USA), 1.5 mM MgCI₂ (Sigma Aldrich, Merck), 0.2 mM of dNTP mix (Thermo Fisher Scientific, Waltham, MA, USA) 0.4 μM of usw47 forward primer (5’-GCTAGGACTTGATTCATTGAT-3’), 0.4 μM of usw47 reverse primer (5’-AGTGATCTAAACGTTCTTATA-3’), 1.25 U Taq DNA polymerase (Sigma Aldrich, Merck). Amplification was performed in a thermocycler (MyGenie^TM 96; Bioneer, Daejon, Korea) under the following conditions: 94 °C initial denaturation for 5 min, 30 cycles at 94 °C for 30 s, annealing temperature 55 °C for 30 s, 72 °C for 1 min, and then a final extension of 10 min at 72 °C.

The PCR products were digested by Hpy188I (New England Biolabs, Ipswich, MA, USA) restriction enzyme after amplification. The total volume of the reaction mixture for enzymatic digestion was 15 μL containing 4 μL PCR product, 0.25 μL Hpy188I gene from Helicobacter pylori, 1× NEBuffer 4 (New England Biolabs) and 9.25 μL distilled water. Enzymatic digestion was performed in a thermo-shaker (Biosan, Riga, Latvia) under the following conditions: 37 °C for 1 h, 65 °C for 20 min and holding at 10 °C for 5 min and then the products were loaded in 2% agarose gel and visualized under UV light after staining with ethidium bromide (Sigma Aldrich, Merck).

After molecular analysis, a small set of commonly cultivated 14 genotypes (Ege-88, Amanos-97, Sarıçanak 98, Şölen 2002, Turabi, Svevo, Zenit, Fırat-93, Fuatbey 2000, GAP, Gediz 75, Tüten 2002, Diyarbakır and Levante) was grown in pots in three replicates. The soil was mixed with acidic peat, in 1:1 ratio to increase the Cd uptake by plants, and then each pot was filled in with 2 kg of the mixture. A volume of 10 mL of CdCI₂·H₂O (Merck, Darmstadt, Germany) was added to each pot with automatic pipette to achieve final Cd mass fraction of 8 mg/kg. At physiologically ripening stage based on Zadoks growth scale (Z 98), grain and stem parts were sampled for each genotype and the samples were dried at 70 °C to constant mass. Dried plant samples of 0.5 g each were digested with 10 mL HNO₃/HClO₄ acid (4:1; Merck) mixture on a hotplate. The samples were then heated until a clear solution was obtained. The same procedure was repeated several times. The samples were filtered and diluted to 100 mL using distilled water, and then Cd mass fraction of the combusted samples with other elements such as P, Mg, Ca, K, Zn, Cu, Fe and Mn was measured by inductively coupled plasma-optical emission spectrometer (ICP-OES) (Optima, PerkinElmer Inc., Waltham, MA, USA). Additionally, soil in each pot was analyzed to determine total Cd accumulation from the soil in the biomass of each genotype at the end of the pot experiment.

Basic statistical parameters such as mean and standard error of mean were determined. Analysis of variance (ANOVA) was performed with least significant difference (LSD) test at the 95% confidence level using SAS statistical software (19). Additionally, correlation and principal component analyses (PCA) were performed to determine relationships among the elements by XLSTAT statistical software (20).

Fig. 1 shows the results of PCR analysis of alleles associated with the accumulation of Cd obtained from usw47 marker. According to banding patterns of usw47, there are three possible alleles: for low Cd content, high Cd content and heterogeneous. Fig. S1 and Fig. S2 show all gel visualizations obtained from molecular analysis. Genotyping results of 108 tested tetraploid wheats are shown in Table 2.

Based on the molecular analysis, 21 (52.5%) out of 40 durum wheat cultivars had alleles associated with high and 12 (47.5%) with low Cd content, and 19 (36.4%) out of 33 advanced breeding lines had alleles associated with high and 21 (63.6%) with low Cd content (Table 2 and Fig. S1). Additionally, only 1 (4%) of the 24 emmer wheat genotypes had alleles associated with high Cd content, and 7 (63.6%) of 11 wild emmer genotypes with low and 4 (36.4%) with high Cd content (Table 2 and Fig. S2). Similar results were obtained by Zimmerl et al. (17), who reported 166 (53%) of 314 tetraploid wheat genotypes associated with low Cd content and usw47 marker can be successfully used to determine low Cd accumulators in tetraploid wheat accessions. Moreover, Vergine et al. (21) also genetically characterized tetraploid genotypes by a sequence-characterized amplified region (SCAR) marker, ScOPC20 in terms of Cd accumulation. However, this marker allows to display two different alleles: one associated with low Cd content (band absent) and another with high Cd content (band present), therefore, heterogeneous state cannot be detected.

Fig. 2 shows Cd mass fractions in the grain, stem and underground parts of fourteen durum wheat cultivars. The results in Fig. 2a demonstrate that Cd addition (8 mg/kg) to the soil mixture clearly increases Cd accumulation in the grain. The cultivar Diyarbakır was the highest accumulator of Cd in the grain in both control and samples with added Cd (0.38 and 0.91 mg/kg respectively, Fig. 2a), while the control sample of Turabi cultivar and the sample of Amanos-97 cultivar grown in the soil with added Cd had the lowest Cd content in grain (0.1 and 0.12 mg/kg respectively, Table S1). All cultivars used in pot experiment except Amanos-97 and Sarıçanak 98 accumulated more Cd in grains after the addition of Cd to the soil (Table S1). In addition to cadmium, phosphorus, potassium, calcium, magnesium, iron, zinc, copper and manganese mass fractions were also determined (Table S1). A two-way ANOVA showed a significant difference (p<0.01) in the mass fractions of these elements among cultivars. Moreover, taking into account cultivar × treatment interaction, a significant difference was found in the mass fractions of all elements at the p=0.01 except for phosphorus (p<0.05). As expected, phenotypic data obtained from pot experiment for Cd accumulation in the grain were similar to molecular data. Low mass fraction of Cd was found in cultivars Ege-88, Amanos-97, Sarıçanak 98, Sölen 2002 and Turabi, which have the allele associated with low Cd content (Fig. 2a). Zimmerl et al. (17) and Perrier et al. (22) reported that varieties with the allele associated with high Cd content had 2.4-fold more Cd in the grain than the varieties with the allele associated with low Cd content.

In addition to grain, stem Cd mass fractions were determined (Fig. 2b). The addition of Cd (8 mg/kg) to the soil mixture increased the stem Cd mass fraction in almost all cultivars used in pot experiment. Moreover, among control samples, Gap cultivar had the highest stem Cd mass fraction, while Tüten 2002 cultivar had the lowest (Fig. 2b and Table S2). Durum wheat cultivars with low Cd mass fraction in their stems can be beneficial feed sources especially for small ruminants that graze wheat stems and leaves after grain harvest under rainfed conditions in Turkey. Svevo cultivar also had the highest stem Cd mass fraction in addition to its high grain Cd accumulation (Fig. 2b). The results of a two-way ANOVA show that there were significant differences (p<0.01) in mass fractions of all elements determined in stem for cultivar and cultivar × treatment interaction (Table S2). On the other hand, Cd mass fractions in underground parts (roots and stubble) of cultivars grown in the soil with the addition of 8 mg/kg Cd were also determined and the results showed that most of the added Cd was accumulated by the plants (Fig. 2c). While Amanos-97 cultivar had the lowest Cd mass fraction in each organ in general, Ege-88 cultivar had the highest Cd mass fraction in the underground parts in particular (Fig 2c). Considering Cd distribution in plant organs, most of the Cd was found in the underground parts, in which cultivars with the alleles associated with low Cd content had 4.13 mg/kg i.e. 67% total Cd (Fig. 3a), whereas those containing alleles associated with high Cd content had 3.75 mg/kg, i.e. 60% of the total Cd (Fig. 3b).

In order to understand the relationships among elements, correlation analysis was performed (Table 3). There was a negative correlation between the grain Cd and Cu (r=-0.76, p<0.01) and Mn (r=-0.56, p<0.01) in the control samples. Grain Cd was also negatively correlated with Mg (r=-0.55, p<0.01) in the grain samples grown in soil with added Cd (Table 3). An opposite finding was reported by Perrier et al. (22) that grain Cd was positively correlated with Mn (r=0.61, p<0.01) and Mg (r= 0.38, p<0.05). In addition to these, there was a non-significant correlation between the grain Cd and Cu (22). Liu et al. (23) also studied correlations between Cd and mineral nutrients in parts of roots and leaves in rice and they reported that Cd²⁺ was generally correlated with Fe³⁺, Mn²⁺, Cu²⁺ and Mg²⁺. Jalil et al. (24) conducted a similar study of durum wheat with different Cd mass fraction added to nutrient solution and they reported that for all of them, the mass fractions of Mn, Zn, Cu and Fe were not affected significantly but Cd additions to the solution depressed the uptake of Zn and Mn. A similar negative interaction between Cd and Mn was also found in this study.

Additionally, principal component analysis (PCA) was performed to determine the relationships between genotypes and plant organs (Fig. 4). PCA showed that the first two components (PC1 and PC2) accounted for 96.31% of the total variance. PC1 explained 70.18% variance, while PC2 elucidated 26.13% of the total variance (Fig. 4). Moreover, contribution of each plant organ to the PC1 and PC2 shows that Cd in the underground parts of plant (43.51) was major contributor to PC1, whereas Cd in the grain (79.72) mainly contributed to PC2 (Table 4). Vergine et al. (21) similarly performed PCA for determination of Cd mass fraction in durum wheat and they reported that roots and kernels contributed to PC1 and grains mostly contributed to PC2. As a result of biplot visualization, different groups were revealed for each plant organ such as underground part (shoots and roots), stem and grain (Fig. 4). Each circle represents different group of Cd mass fractions in each plant organ in the bi-plot graph. The Diyarbakır and Levante cultivars, marked with yellow color, accumulate the highest mass fractions of Cd in the grain. The second group marked with green consists of Svevo and Tüten 2002 cultivars, which had the highest mass fractions of Cd in the stem. The third group marked with blue color comprises Amanos-97 and Sarıçanak 98 cultivars, which accumulate high mass fractions of Cd in the root. All other cultivars, Fırat-93, GAP, Fuatbey-2000, Zenit, Gediz 75, Turabi and Ege-88, had lower Cd mass fractions in all plant organs (Fig. 4). Svevo cultivars accumulated the highest mass fraction of Cd in both grain and stem, while Amanos-97 cultivar had the lowest mass fraction of Cd in the stem and grain. At this point, difference in the root to grain translocation of Cd among durum wheat genotypes is very important to develop new cultivars that can be grown in Cd-contaminated soils. If this translocation is weak or root sequestrates the Cd efficiently, grain Cd content will be low (25, 26). In addition to this, partitioning of Cd among plant organs is the second important strategy for low Cd mass fraction in the grain. Perrier et al. (22) highlighted that growing long-stemmed cultivars may have advantages since lower mass fractions of Cd are moved to plant organs such as stem, leaves, bracts, rachis and grains. Arduini et al. (27) found that partitioning to shoots and grains with increasing Cd supply was markedly higher in Svevo cultivar. They also reported that high Cd content in grains of Svevo cultivar may be related to the high allocation of biomass in roots during vegetative growth stage coupled with high post-heading dry matter accumulation and root to grain re-mobilization. Higher accumulation of elements from the soil in the plant is a desired trait to obtain higher yield and quality; therefore, breeding studies have focused on improving yield components to increase crop yield (28). Due to these concerns, modern wheat varieties tend to accumulate elements such as Cd in the grain (21, 22, 28). However, since high Cd content in the grain is not a desirable trait, cultivars with alleles associated with low Cd content and high yield should be given first priority in durum wheat breeding.