Winter wheat (Triticum aestivum L.) cultivar Jing411 was bred for high yield potential, cold tolerance, high tilling ability, and is frequently planted in the Northern winter wheat region in China. It has become the elite parent in wheat high-yield breeding processes. Seeds of Jing411 were used for ethyl methane sulphonate (EMS) treatment to induce random mutations.

Dry seeds (Supplementary Figure S1a) of Jing411 were initially soaked in tap water for 10 h until germinated and embryo buds began to emerge out of the seed coat (Supplementary Figure S1b). The germinated seeds were then incubated in 0.5, 1.0, or 1.5% EMS in 0.1 M sodium phosphate buffer (pH 7.0) for 4 h. Around 2000 seeds were used in each treatment. EMS treatment was performed on a shaker at 50 rpm in dark conditions at 20°C with one seed per milliliter EMS buffer. After incubation, seeds were washed in tap water for 4 h at room temperature and prepared for sowing.

Treated seeds were directly sowed with 5 cm spacing and grown into M₁ plants. All spikes were bagged to be used strictly in self-crosses. One single M₂ seed was harvested from each plant. All harvested M₂ seeds were sowed and grown into M₂ individual plants. Twenty M₂ seeds per row were planted in 19 rows with one row wild type Jing411 wheat. Each M₂ plant was harvested to develop the M₃ generation. Twenty M₃ seeds from individual M₂ plant were planted into one row and used to verify their phenotype variation.

Row length was 2 m long with 30 cm row-to-row spacing. All plants were grown in the experimental field, Institute of Crop Science, Chinese Academy of Agricultural Sciences, Beijing.

Mutations in growth habit, leaf morphology, heading date, and spike morphology were investigated in the M₂ generation. Heading date ±2 days compared to the wild type was considered a mutant. The plant height, tiller number, spike length, and TGW of all individuals were measured after harvesting and a mean value greater or less than two standard deviations of the wild type mean was classified as a mutant. The mutated phenotype was identified in the M₃ generation plants.

Mutation frequency was calculated as percentage of total number of mutants divided by corresponding population size.

Young leaves at seedling stage from M₂ individual plants were collected and genomic DNA was extracted using the DNA-quick Plant System kit (Tiangen Biotech, Beijing, China). The DNA concentration was normalized to 50 ng/μl according to Guo et al. (2017). The DNA was 2×-pooled for gene TILLING according to Till et al. (2006).

Point mutations of the target gene TaAGP.L-B1 (7725 bp; GenBank accession no: AJ563452) were screened using the TILLING approach following the protocol of Till et al. (2006). Specific primers were designed and used for PCR amplification of this region. Their sequences are: AGP-1F:ATCTGGATACTTGTACATCTGC, AGP-1R: GAAGACATAGACTCCCATAG; AGP-2F: CATTGCAGCCCCCAAAGTTC, AGP-2R: AAAGCGGCGTGGTTTCAGAC. Gene structure and location of the primers were shown in Supplementary Figure S2, lengths of the amplicons were 670 and 728 bp respectively, including exons and introns region. Potential mutations were further sequenced to verify nucleotide variation.

Each point mutation was analyzed online by the PARSESNP (Project Aligned Related Sequences and Evaluate SNPs¹) and SIFT (Sorting Intolerant from Tolerant²) programs. Mutations with PSSM > 10 or SIFT < 0.05 are predicted to have a severe effect in protein function (Ng and Henikoff, 2003; Taylor and Greene, 2003).

At 12, 18, and 24 days after flowering (DAF), seeds were sampled for gene expression analysis. Total RNA was extracted using Trizol (Invitrogen, Carlsbad, CA, United States), and cDNA was synthesized with 5×All In One MasterMix kit (With gDNase, ABM, Co.). Primers for RT-qPCR were designed using software Beacon Designer 7.5 (PREMIER, Palo Alto, CA, United States). The following primers were used: AGP-qF: GAACTACATGACTTTGGGTCTGAGA, AGP-qR: CATCATCATCGCATTCTTGAGCT. Amplification was performed with 0.5 μmol primer using SsoFast EvaGreen Supermix kit (Bio-Rad) on a CFX96 system. The program started from an initial denaturation at 95°C for 3 min, and followed by 40 cycles at 94°C for 30 s, annealing for 30 s, and 72°C for 10 s, and a melt curve stage. The actin gene (GenBank accession no: AAW78915) was used as an internal control. Three biological and technical replicates were used for each genotype. The relative expression level was calculated using ΔΔCt method (Livak and Schmittgen, 2001). Data were analyzed using a one-way ANOVA with P < 0.05 or 0.01 considered as significant.

Mature seeds were milled using a cyclone mill (CT410, FOSS Analytical, Co., Ltd) with a 0.5 mm flour screen. Starch content was measured using a Total Starch Assay Procedure (amyloglucosidase/α-amylase method, K-TSTA) from Megazyme (Megazyme International, Ltd, Wicklow, Ireland). Data were analyzed by one-way ANOVA and P < 0.05 or 0.01 was considered significant.

A total of 313, 489, and 417 individual M₂ plants and their M₃ lines were investigated at 0.5, 1.0, and 1.5% EMS treatment, respectively. Mutations were observed in the heading date, spike, plant architecture, and leaf and grain morphology. Phenotypic variations in the mutagenized wheat population can be observed in Figure 1. The mutation frequency of each morphological variation category grew as EMS concentration increased (Figure 2 and Supplementary Table S1).

Two primers specific to the gene TaAGP.L-B1 were designed and used for allelic variation screening in the M₂ population (n = 1218) treated with 0.5, 1.0, and 1.5% EMS (n = 312, 489, and 417, respectively), and 18 mutations were identified in 1.0 and 1.5% subpopulation with 72% transitions (C > T, G > A) and 22% transversions (C > A, T > C, A > C) (Table 1 and Supplementary Table S2), mutation density was 1/35 and 1/104 Kb respectively, none was screened in 0.5% subpopulation. Four missense mutations and one splice mutation were identified (Table 1). One of the missense mutations in line E3-1-3 resulted in a change from proline to leucine in the 378th amino acid. The amino acid site is located in a conserved region in wheat, maize, rice, and potato (Supplementary Figure S3). The PSSM and SIFT values also predicted a mutation in line E3-1-3 that might affect gene function. The other three missense mutations in lines E049-4, E422, and E051-6 had PSSM and SIFT values that predicted no significant alteration on function. Furthering analysis on gene expression and starch content were carried out to verify allelic effects.

The TaAGP.L-B1 expression patterns in developing seeds with missense mutations (lines E051-6, E422, and E3-1-3) were analyzed. Expression levels of the TaAGP.L-B1 gene in mutant E3-1-3 were significantly reduced (P < 0.01) compared to the wild type at 12, 18, and 24 DAF (Figure 3). The mutant line E051-6 showed significant change in gene expression at 18 DAF and 24 DAF compared to the wild type (P < 0.05). The E422 mutant line only showed a significant change in gene expression compared to the wild type at 12 DAF (P < 0.05).

Starch content in the wild type and mutant M₃ and M₄ seeds were measured. In the mutated line E3-1-3, starch content showed a significant decrease compared to the wild type, while no significant changes were observed in the other two mutants (Table 2).

Damaging effects in the M₁ population are closely related with chemical concentration, treatment time, temperature, seed physiological state, and genotype. Typical EMS treatments are carried out using dry seeds directly soaked in solutions at room temperature for 16–18 h in concentration ranging from 0.6 to 1.2% (Slade et al., 2005; Rakszegi et al., 2010; Chen et al., 2012). As germinated seed is more sensitive than dry seed (Lin et al., 1991), we pre-treated our seeds in tap water to achieve germination and only treated seeds for 4 h. Treated seedlings were half the size of the untreated wild type at the 1.0% EMS treatment (Supplementary Figure S1c), allowing treatment time to be remarkably shortened. Temperature is an important element in our treatment process to keep seeds in an appropriate physiological state. Here, we treated seeds at 20°C for 4 h with a 10 h pre-treatment of germinating. Higher temperatures would cause physiological and biochemical reactions to occur faster, in turn leading to severe damage effects. To avoid this, treatment times should be shortened. On the contrary, low temperatures result in minor damage effects, allowing for treatment times to be longer.

Gene mutation density in multi-ploid species is higher than in diploid species (Parry et al., 2009). It has been shown that high phenotypic mutation frequency can also be generated in a mutagenized durum wheat population (Bovina et al., 2014). In the present study, the single nucleotide mutation density in 1.0% subpopulation was similar with those in other hexaploid wheat (Slade et al., 2005; Uauy et al., 2009; Chen et al., 2012); while in 1.5% subpopulation was relatively lower. But phenotypic mutation frequencies were different, it ranged from 4 to 19% and increased linearly with increased EMS concentration. The relationship of nucleotide mutation density and phenotypic mutation frequency in mutagenized population need further analysis based on more target genes and larger population. However, mutation frequency varied among phenotypes and EMS treatments (Supplementary Table S1). As reported, EMS treatment resulted in more G/C to A/T transitions than other mutation types in rice, maize, soybean and wheat (Till et al., 2004, 2007; Slade et al., 2005; Cooper et al., 2008; Chen et al., 2012). In our population, 72% mutation type was G/C to A/T transitions, it reconfirmed that EMS induced changes are transitions especially G/C to A/T.

Cultivars arise from long-term evolution and artificial selection and their gene regulatory networks have been optimized during this process. However, mutation induction leads to sequence changes in one or more genes and can break the balance of this network. The majority of mutations result in negative effects on agronomic traits. The Jing411 winter wheat wild type used in this study is an elite cultivar with high yield potential. Its TGW is about 50 g, showing high quality wheat. In many mutants, the TGW was lower. Our result showed that the maximum mutation frequency was observed in seed morphology, followed by spike morphology.

Numerous phenotypic mutations have been identified in previous reports, but the specific mutated traits in each wild type cultivar were different (Rakszegi et al., 2010; Chen et al., 2012; Bovina et al., 2014; Dhaliwal et al., 2015). These mutant lines, including ours, could be used for further novel gene discovery. More plant height mutations have been observed in mutagenized populations of Indian and Jinmai 47 and this could be a result of both wild type heights being greater than other wheat cultivars (Chen et al., 2012; Dhaliwal et al., 2015). Here, we did not see much phenotypic variation in plant height in the mutagenized lines. Two of the semi-dwarfing mutants from our Jing 411 mutated resource carried the semi-dwarfing genes Rht-D1b, which is the mutation of the allele Rht-D1a (Xiong et al., in press). We deduced that mutation type and frequency could be closely related to the genes that are available in the wild type. More mutant resources derived from different wild type cultivars with far genetic distances could produce different phenotypic and genetic mutations that could ultimately be used to increase cultivar quality.

The large subunit gene of the AGPase enzyme is expressed during grain development (Smidansky et al., 2007; Tang et al., 2016), and the overexpression of this gene results in enhanced AGP activity and starch content (Li et al., 2011; Tuncel and Okita, 2013). The altered amino acid residue in mutant E3-1-3 was located in a conserved region shared with potato and rice. The specific residue interacts with amino acids of the small subunit (Tuncel et al., 2008; Dawar et al., 2013). We did not observe any other mutations in this region. In order to validate the effect of this novel mutated allele, we analyzed the change in AGPase expression at 12, 18, and 24 DAF, as these times are peak expression periods (Thorneycroft et al., 2003; Stamova et al., 2009; Zhang et al., 2016). We found that expression was significantly down-regulated in the mutant E3-1-3, and led to a significant decrease in starch content compared to the wild type. This is consistent with transgenic results (Tuncel and Okita, 2013). Our results confirmed that the mutant E3-1-3 carried a novel allele, which was non-lethal and might related with functional variation of AGPase large subunit gene, it could be used for further analysis of the AGPase large subunit gene effect on wheat growth and development.

Mutagenesis generates mutations randomly and usually produces more than one gene/allele mutation in an individual mutant line. These multi-genic mutations might demonstrate an effect on a phenotypic trait. To reduce mutation background, backcrossing is one of the most commonly used strategies (Konik-Rose et al., 2007; Sestili et al., 2015; Mizuno et al., 2016; Acevedo-Garcia et al., 2017). However, it may take several years for hexaploid wheat crossings and the development of triple mutant lines. Utilization of multiple independent mutants is another alternative strategy (Krasileva et al., 2017). In the present study, we compared expression and phenotypic results of line E3-1-3 with the other two mutant lines E051-6 and E422. The PSSM value of lines E051-6 and E422 were less than 10 and were predicated to have no severe effect. The gene expression patterns and starch contents in these two lines varied slightly, indicating that background mutations did not affect AGPase large subunit gene function. The starch content variation in the mutant E3-1-3 resulted from a novel allele in TaAGP.L-B1. Similarly, mutations on gene TaSSIVb-D have been identified in a mutagenized population, and by directly using two mutant lines, the gene function of TaSSIVb-D on transient starch synthesis in wheat leaves was elucidated (Guo et al., 2017). This suggests that using multiple individual mutant lines to illustrate gene characterization in multi-ploid species is an effective strategy for reverse genetics at least in major genes. Contribution of each homoeologous loci might be different (Yu et al., 2012; Yang et al., 2014), expression variations and phenotypic changes caused by gene mutation would be different either, those with higher genetic effects would be much easier to be analyzed. Genetic effects and their difference of the three homoeologous of TaAGP.L need to be verified by further experiments.