A F₆ population comprising 151 RIL, derived from a cross between YM4 and YZ1 was developed by a single seed descent approach at the Institute of Crop Sciences, Chinese Academy of Agricultural Sciences (CAAS). These RIL populations were grown at Wanfu Experimental Base of Lixiahe Institute of Agricultural Sciences (Yangzhou, Jiangsu) in 2019-2020 (2020YZ) and 2020-2021 (2021YZ). Field experiments were conducted in a completely randomized design with two replications. For each RIL and cultivar, an average of fifty seeds per row were sown in 180-cm double rows separated 25 cm apart. Fertilization and field management should adhere to local agricultural practices, ensure timely pest and disease control, and manually harvest each variety at maturity according to the designated plot. In 2021, wheat grains were harvested at their respective maturity stages. After harvest, the healthy seeds were measured for thousand grain weight using the Wanshen SC-G seed detector. Afterward, the wheat seeds underwent three rounds of washing with deionized water, followed by drying, baking in a constant temperature box until reaching a constant weight, and grinding through a sample sieve for subsequent analysis and testing. One hundred and forty-nine wheat advanced lines developed by the Lixiahe Institute of Agricultural Sciences were grown in 2022 using the same protocol as RIL ( Supplementary Table 1 ).

The analysis of micronutrients was conducted by Nanjing Yizhiyuan Testing Technology Co., Ltd. The determination of Mn, Fe, Cu, and Zn involves weighing a 0.2 g sample (with an accuracy of 0.0001 g) and placing it in a PTFE tank. Then, 5 mL of concentrated nitric acid and 2 mL of hydrogen peroxide were added into the tank. The mixture is placed on a graphene electric heating plate and heated at 150°C for digestion. Nitric acid is added midway until the digestion solution becomes transparent. The temperature is then set to 170°C to evaporate excess acid until approximately 1 mL of solution remains. After cooling with pure water up to a volume of 50 mL, the solution is filtered for analysis using a Thermo Fisher double channel inductively coupled plasma emission spectrometer (ICP-AES) to determine the content of Mn, Fe, Cu, and Zn. Two replicates were set up for each experiment and were conducted simultaneously. The mixed standard solution was injected into ICP-AES, and the signal response values of the element and the internal standard element were determined. The standard curve was drawn with the concentration of the element as the abscissa and the ratio of the response signal value of the element to the selected internal standard element as the ordinate. The blank solution and the sample solution were injected into ICP-AES, and the signal response values of the element and the internal standard element were determined. The concentration of the element in the digestion solution was obtained according to the standard curve. The content of the element to be examined in the sample:

Where: X-element content in the sample (mg/kg); P-element mass concentration in the sample solution (mg/L); P₀-element mass concentration in the blank solution of the sample (mg/L); V-constant volume of sample digestive liquid (mL); f-dilution factor of the sample; m-weighing mass of the sample (g).

Determination of Se: weigh 0.5 g sample (accurate to 0.0001 g), place in a PTFE tank, add 5 mL concentrated nitric acid, 2 mL hydrogen peroxide, place on the graphene electric heating plate, 150°C heating digestion, midway add nitric acid until the solution is transparent, set 170°C to chase acid to the solution remaining about 1 mL, cooling with pure water to 50 mL, filter, atomic fluorescence spectrometer (AFS-933) on the machine determination of Se content. The blank test was also conducted. The mixed standard solution was injected into the atomic fluorescence spectrometer to determine the signal response values of Se and internal standard elements. The concentration of Se was plotted on the abscissa, while the ratio of the response signal value of Se to the selected internal standard element served as the ordinate for drawing a standard curve. Subsequently, both the blank solution and sample solution were injected into the atomic fluorescence spectrometer to determine their respective signal response values for Se and internal standard elements. Finally, based on the obtained standard curve, the concentration of Se in the digestion solution was determined. The content of Se in the sample is calculated by the formula above.

The initial data were processed and statistically analyzed using SPSS 22.0 and Microsoft Excel 2019, primarily including descriptive statistics and t-tests. The analysis of variance (ANOVA) was estimated using IciMapping v4.1, and the broad-sense heritability ( h b 2 ) for micronutrients was estimated using the formula h b 2 = σ 2 G / ( σ 2 G + σ 2 G × E / E + σ 2 e / E R ) , where E and R represent the number of environments and replicates, respectively, σ 2 G is the genotypic effect, σ 2 G × E is the effect of genotype by the environment, and σ 2 e is the residual error (Nyquist and Baker, 1991).

The genomic DNA of the tested materials were extracted by the CTAB method (Stacey and Isaac, 1994). The tested parents and RIL populations were detected for 55 K SNP markers by using the Bead Array technology of the Illumina SNP Genotyping technology test platform (Beijing BOA Biotechnology Co., Ltd.), and the corresponding KASP markers including Rht-B1, Rht-D1 and TaGW2-6A genes that control important traits of wheat were integrated into the polymorphic markers (Rasheed et al., 2016). After filtering and screening all genotypes of polymorphic markers, a total of 7974 polymorphic SNP markers were obtained. After removing the redundancy, 1546 SNPs were obtained. Finally, 1440 SNPs were used to construct a genetic linkage map covering 21 chromosomes of wheat with a length of 3574.10 cM. The inclusive composite interval mapping (ICIM) was employed to identify QTL significantly associated with grain micronutrient content in wheat, with a LOD threshold set at 2.5 (Li et al., 2021). Genetic maps for each chromosome were drawn using MapChart 2.3 software (Hu et al., 2023b). QTL exhibiting a genetic position peak within 10 cM on the same chromosome were considered identical, and the nomenclature of QTL was followed by Hu et al. (2023a). To determine the novel loci identified in this study, we compared the flanking markers sequences of these loci with those of previously reported loci in the EnsemblPlants database (http://plants.ensembl.org/).

KASP markers were developed using the flanking SNP of the target QTL according to the method of Xu et al. (2019) (PolyMarker, http://polymarker.tgac.ac.uk/). Specific sequences that can bind to FAM fluorescence were added to the tail of the F1 primer, and specific sequences that can bind to HEX fluorescence were added to the tail of the F2 primer, synthesized by Beijing Jiacheng Biotechnology Co., Ltd. A total of one hundred and forty-nine wheat advanced lines were used for KASP marker validation. The PCR reaction was performed on an ABI Veriti 384 PCR instrument (Thermo Fisher). The PCR amplification products were scanned, and fluorescence values were obtained using the Omega F SNP typing detector (LGC Genomics Ltd, KBS-0024-002). Kluster CallerTM (KBioscience) software was used for genotyping analysis.

Genes and their annotations within the mapping intervals were extracted according to International Wheat Genome Sequencing Consortium Reference Sequence v2.1 (IWGSC RefSeq v2.1) (http://202.194.139.32/jbrowse-1.12.3-release/?data=Chinese_Spring2.1&loc) (Hu et al., 2022). To further define the biological process and molecular function of the candidate genes, the description and GO enrichment were also analyzed using Triticeae-GeneTribe (http://wheat.cau.edu.cn/TGT/) (Zhao et al., 2023).

The Fe, Cu, Zn, and Se content of YM4 were significantly higher than those of YZ1, and the Mn content of YZ1 was significantly higher than that of YM4 ( Table 1 ). In the RIL population, Mn content ranged from 2.11 to 25.63 mg/kg, Fe content from 10.97 to 76.34 mg/kg, Cu content from 0.01 to 4.82 mg/kg, Zn content from 16.02 to 67.82 mg/kg, and Se content from 0.01 to 0.15 mg/kg based on the mean datasets of two trials ( Table 1 ). A pattern of continuous distribution for each micronutrient was observed in each environment and the mean datasets in the RIL population. Heritability of Mn, Fe, Cu, Zn, and Se content in RIL population were 0.85, 0.87, 0.93, 0.69, and 0.89, respectively ( Table 1 ).

In the YM4/YZ1 RIL population, Pearson correlation analysis was conducted for all traits based on the mean datasets across different environments. The results indicated significant positive correlations were observed between Mn and Fe (r = 0.370, P < 0.01), Mn and Cu (r = 0.501, P < 0.01), Mn and Zn (r = 0.481, P < 0.01), and Mn and TGW (r = 0.280, P < 0.01) ( Table 2 ). Additionally, Fe showed significant positive correlations with Cu (r = 0.426, P < 0.01), Zn (r = 0.415, P < 0.01), and TGW (r = 0.235, P < 0.01) ( Table 2 ). Cu was found to be significantly and positively correlated with Zn (r = 0.409, P < 0.01) and TGW (r = 0.206, P < 0.01) ( Table 2 ). Furthermore, Zn exhibited significant positive correlations with Se (r = 0.170, P < 0.05) and TGW (r = 0.310, P < 0.01) ( Table 2 ).

Among all the detected QTL for micronutrients, QMn.yaas-1B, QCu.yaas-1A, QCu.yaas-7A, QZn.yaas-7D, and QSe.yaas-2D could be detected in two environments and mean values, while QMn.yaas-4D, QFe.yaas-2D, and QZn.yaas-4D could be detected in only one environment ( Table 3 ). Two QTL for Mn were identified on chromosomes 1B and 4D, and the positive allele of QMn.yaas-1B was derived from YZ1, explaining 10.16% to 18.04% of phenotypic variances, while the positive allele of QMn.yaas-4D was derived from YM4, explaining 6.02% of phenotypic variances ( Table 3 ). One QTL QFe.yaas-2D for Fe was identified on chromosome 2D, and the positive allele of it is from YM4, explaining 8.04% of the phenotypic variances ( Table 3 ). Two QTL for Cu were identified on chromosomes 1A and 7A, and the positive allele of QCu.yaas-1A is from YZ1, explaining 15.87%–19.92% of the phenotypic variances, while the positive allele of QCu.yaas-7A is from YM4, explaining 5.77%–10.59% of the phenotypic variance ( Table 3 ). Two QTL for Zn were detected on chromosomes 4D and 7D, and the positive allele of QZn.yaas-4D is from YM4 with 12.29% of phenotypic variances, while the positive allele of QZn.yaas-7D is from YZ1, explaining 7.66%–11.26% of phenotypic variances ( Table 3 ). One QTL QSe.yaas-2D for Se was identified on chromosome 2D, and the positive allele of it is from YM4, explaining 11.52% to 20.11% of phenotypic variances ( Table 3 , Figure 1 ).

The flanking linkage markers AX109009142 (1A), AX109372911 (1B), AX111684175 (2D), AX110572006 (4D), AX109110307 (7A), and AX111640147 (7D) for QCu.yaas-1A, QMn.yaas-1B, QFe/Se.yaas-2D, QMn/Zn.yaas-4D, QCu.yaas-7A, and QZn.yaas-7D were used to evaluate their effects on TGW in the YM4/YZ1 population. For QCu.yaas-1A and QMn.yaas-1B, the YZ1 homozygous allele increased TGW by 3.52% (P<0.05) and 3.45% (P<0.05) relative to the YM4 homozygous allele, respectively ( Figures 2A, B ). For QFe/Se.yaas-2D, the YM4 homozygous allele decreased TGW by 6.45% (P<0.001) relative to the YZ1 homozygous allele ( Figure 2C ). For QMn/Zn.yaas-4D, the YM4 homozygous allele increased TGW by 7.51% (P<0.001) relative to the YZ1 homozygous allele ( Figure 2D ). QCu.yaas-7A and QZn.yaas-7D did not significantly affect TGW ( Figures 2E, F ).

The SNP markers AX109009142 (1A), AX109372911 (1B), AX111684175 (2D), AX110572006 (4D), AX109110307 (7A), and AX111640147 (7D), which linked with the target QTL in the current study were finally converted into KASP markers and designated as KASP.1A, KASP.1B, KASP.2D, KASP.4D, KASP.7A and KASP.7D ( Table 4 ). The KASP markers were used to screen one hundred and forty-nine wheat advanced lines planted in 2022 in the Yangzhou yield evaluation nursery to verify the target QTL effect on wheat grain micronutrient content. For QCu.yaas-1A, the lines carrying the YZ1 allele increased Cu content by 44.90% relative to those carrying the YM4 allele (P<0.001) ( Table 5 , Figure 3A ). For QMn.yaas-1B, the lines carrying the YZ1 allele increased Mn content by 41.72% relative to those carrying the YM4 allele (P<0.001) ( Table 5 , Figure 3B ). For QFe.yaas-2D, the lines carrying the YM4 allele increased Fe content by 29.02% relative to those carrying the YZ1 allele (P<0.001) ( Table 5 , Figure 3C ). For QSe.yaas-2D, the lines carrying the YM4 allele increased Se by 40% relative to the lines carrying the YZ1 allele (P<0.001) ( Table 5 , Figure 3D ). For QMn.yaas-4D, the lines carrying the YM4 allele increased Mn content by 19.35%relative to those carrying the YZ1 allele (P<0.05) ( Table 5 , Figure 3E ). For QZn.yaas-4D, the lines carrying the YM4 allele increased Zn content by 24.00% relative to those carrying the YZ1 allele (P<0.001) ( Table 5 , Figure 3F ). For QCu.yaas-7A, the lines carrying the YM4 allele increased Cu content by 28.86% relative to those carrying the YZ1 allele (P<0.001) ( Table 5 , Figure 3G ). For QZn.yaas-7D, the lines carrying the YZ1 allele increased Zn content by 24.12% relative to those carrying the YM4 allele (P<0.001) ( Table 5 , Figure 3H ).

Among the one hundred and forty-nine wheat advanced lines, eight lines pyramiding six positive alleles at all loci, with the Mn content ranging from 5.72 mg/kg to 21.28 mg/kg; the Fe content ranging from 24.87 mg/kg to 69.28 mg/kg; the Cu content ranging from 3.46 mg/kg to 9.15 mg/kg; the Zn content ranging from 37.52 mg/kg to 116.29 mg/kg; the Se content ranging from 0.02 mg/kg to 0.12 mg/kg ( Table 6 ). The average micronutrient values of these eight lines were 13.24 mg/kg for Mn, 41.63 mg/kg for Fe, 6.49 mg/kg for Cu, 68.57 mg/kg for Zn, and 0.07 mg/kg for Se. Seventeen lines harbored five positive alleles, with average Mn of 10.99 mg/kg, Fe of 39.70 mg/kg, Cu of 5.30 mg/kg, Zn of 68.59 mg/kg, and Se of 0.06 mg/kg. These twenty-five lines, along with the KASP markers linked with the micronutrient content QTL, will be used in wheat breeding for quality.

A total of 60 annotated high-confidence genes (TraesCS2D03G0955600 to TraesCS2D03G0967800) were identified in the target interval of the QFe/Se.yaas-2D ( Supplementary Table 2 ). These genes predominantly include were mainly acyclic sesquiterpene synthase, probable cation transporter HKT7, Protein FORGETTER 1, and UPF0481 protein At3g47200, etc. ( Supplementary Table 2 ). A total of 51 annotated high-confidence genes (TraesCS4D03G0057000 to TraesCS4D03G0065300) were identified in the genomic region of QMn/Zn.yaas-4D, mainly encoding Adenylosuccinate synthetase 2, Senescence-specific cysteine protease SAG39, Zinc finger CCCH domain-containing protein 24 and Zinc finger protein 593 homolog, etc. ( Supplementary Table 3 ). GO enrichment analysis indicated that these candidate genes of QFe/Se.yaas-2D are primarily involved in positive regulation of cellular response to heat, histone binding, and regulation of gene expression, epigenetic ( Supplementary Table 4 , Supplementary Figure 1 ). Several candidate genes of QMn/Zn.yaas-4D involved in biological processes of cell development, ethylene responsive, and senescence-associated vacuole ( Supplementary Table 4 , Supplementary Figure 2 ).

Rare studies focused on the genetic base of micronutrients in wheat grains. Graham, (1987, 2001) and Schlegel et al. (1991) identified a Cu content QTL on chromosome 5B. Pei (2016) identified the associated loci related to Se content in wheat on chromosomes 3D and 5A. QTL related to Fe content were located on chromosomes 2A and 7A, while QTL related to Se content in wheat grains was located on chromosome 7A (Tiwari et al., 2009). Tong et al. (2022) detected SNP significantly associated with Zn content in wheat grains on chromosomes 1A, 2A, 3A, 3B, 5A, 5D, 6A, 6B, 6D, 7A, 7B, and 7D through GWAS, and also identified SNP significantly associated with Fe content on chromosomes 1A, 1B, 5A, 5B, 7A, 7B and 7D. The above loci were not in the same physical intervals as the detected QTL in this study. Previous studies have shown that there may be a close relationship between different elements, and grain micronutrients have pleiotropic effects on other agronomic traits (Pei, 2016). The QFe.yaas-2D and QZn.yaas-2D identified in this study were found to be located within the same physical interval that exhibited a significant correlation with Fusarium head blight (FHB) resistance in our previous study (Hu et al., 2023b). Previous studies also indicated a relationship between Se content in wheat grains and FHB resistance (Mao et al., 2020a, b). QMn. yaas-4D and QZn. yaas-4D co-localize within the same physical region as well as in the wheat dwarf gene Rht-D1. Previous studies have demonstrated that Rht-D1 locus regulated Zn and Mn accumulation in maize grains (Zhou et al., 2010). It is probable that the accumulation of Mn and Zn in grains has a similar genetic basis, while the accumulation of these two micronutrients in wheat grains may be related to the signal transduction of gibberellin.

Hu et al. (2023a) located QTgw.yas-2DL (529.23-537 Mb) and QTgw.yas-4DS (16.58-19.19 Mb) on chromosomes 2D and 4D, respectively, which are within the same confidence interval as the QFe/Se.yaas-2D and QMn/Zn.yaas-4D identified in this study. The interaction between the wheat gene network and external environmental factors results in a positive correlation between micronutrients such as manganese and iron and the thousand grain weight phenotype. The pleiotropic effects of QTL on thousand grain weight indicate that the increasing allele of QFe/Se.yaas-2D has a negative impact on TGW, while the increasing allele of QMn/Zn.yaas-4D has a positive effect on TGW. This suggested that iron or selenium may have an opposite effect on TGW compared to manganese and zinc. Therefore, further fine-mapping of QFe/Se.yaas-2D, and QMn/Zn.yaas-4D would promote us to analyze the interaction mechanism between micronutrient content and TGW. KASP markers for QCu.yaas-1A, QMn.yaas-1B, QMn/Zn.yaas-4D could be used in wheat breeding for improving micronutrients and TGW.

For one hundred and forty-nine wheat advanced lines, twenty-five lines were found to carry five or more positive alleles at the detected QTL in the current study. The average values for these 25 lines were 11.71 mg/kg for Mn, 40.32 mg/kg for Fe, 5.68 mg/kg for Cu, 68.58 mg/kg for Zn, and 0.06 mg/kg for Se, which increased Mn, Fe, Cu, Zn, and Se by 37.83% (P<0.001), 35.80% (P<0.001), 45.38% (P<0.001), 23.42% (P<0.001), and 3.95% (P<0.001) relative to the average values of micronutrient content the for the one hundred and forty-nine wheat advanced lines. Therefore, these 25 lines can be used as excellent resources for wheat quality breeding.

The high-confidence genes within the physical interval of QFe/Se.yaas-2D were primarily involved in the synthesis of acyclic sesquiterpene synthase, probable cation transporter HKT7, and UPF0481 protein At3g47200. Acyclic sesquiterpene synthase was associated with the synthesis of gibberellins, which were essential for promoting plant nutrition and reproductive growth, as well as regulating physiological processes and stress responses (Toyomasu et al., 2000). The probable cation transporter HKT7 played a crucial role in regulating ion balance in plants, ensuring their normal growth and development under salt stress conditions (Huang et al., 2006). UPF0481 protein At3g47200 was linked to plant stress tolerance and growth development (https://www.uniprot.org/uniprotkb/A0A1D1Y378/entry). The high-confidence genes in the physical interval of QMn/Zn.yaas-4D were mainly related to Adenylosuccinate synthetase 2, Senescence-specific cysteine protease SAG39, and Zinc finger protein 593 homolog. Adenylosuccinate synthetase 2 synthesizes adenylosuccinate in plant chloroplasts, an important intermediate in the purine nucleotide biosynthesis pathway. This enzyme played a crucial role in regulating nucleotide synthesis, cell proliferation, and responses to environmental stress (Zhu et al., 2023). Senescence-specific cysteine protease SAG39 is a protease specifically expressed during plant senescence and played a key role in plant aging and stress responses (Wang et al., 2022). Zinc-finger proteins were characterized by a structural domain that coordinates zinc ions and binds nucleic acids, participating in the regulation of growth, development, and stress adaptation in plants (Han et al., 2021). The results indicated that the aforementioned genes in the QFe/Se.yaas-2D and QMn/Zn.yaas-4D intervals may be related to stress tolerance, growth development, and zinc synthesis in plants.