Soybean (Glycine max L. Merrill) is an oil-producing legume crop in the Fabaceae family. Since the 1930s, soybean has been grown as an oilseed crop and contributed greatly to the recent United States economy (Wilson, 2008). It is the second most widely-grown crop in the United States next to corn (Zea mays), with a total area planted approximately 82.7 M acres in 2015 ($34.5 billion value). Soybean comprised 61% of the world oilseed production and is one of the major commercial crops worldwide (SoyStats, 2016). Aside from its oil content, soybean meal is a primary component of aquaculture and animal feeds (Wilson, 2008). In addition, the demand for soybean has increased as trends like biodiesel production is becoming popular in the market (Cao et al., 2005). With the increasing demand for soybean oil and other soybean products, it is imperative to increase its production. However, there are challenges that can affect the production of soybean including decreasing area for cultivation due to increasing population and climate change, abiotic stresses, and pest and disease problems (Hartman et al., 2011; Alexander et al., 2015; Iizumi and Ramankutty, 2015). To overcome these challenges in soybean production, and to meet the increasing global demands, soybean research communities are working to improve the genetic diversity that is available to plant breeders (Wilson, 2008; Patil et al., 2017).

Breeding of modern soybean varieties, which was focused on selection for modern traits such as higher yield, has created a genetic bottleneck for improvements in this crop (Valliyodan et al., 2016). These genetic bottlenecks resulted in loss of rare alleles and reduced the genetic diversity of soybean in the modern varieties (Hyten et al., 2006). One of the strategies that can be exploited to increase diversity in soybean genome is to introduce mutations. The reference soybean genome sequence (Schmutz et al., 2010) can be used to trace and identify mutations introduced to soybean. Plant mutation breeding involves harnessing the advantages of mutations to produce desirable variation for crop improvement (Pathirana, 2011). Gene mutagenesis changes nucleotide sequences, allowing production of novel alleles. There are four common mutagenesis methods (1) physical agents such as UV, X-ray radiation and fast neutron (FN), (2) chemical mutagens such as ethyl methane sulfonate (EMS), N-nitroso-N-methylurea (NMU), ethyl nitrosourea (ENU), 1,2:3,4-diepoxybutane (DEB), (3) biological agents such as T-DNA and transposons (Hancock et al., 2011), and (4) transgenic technologies such as CRISPR-Cas9, TALENs, gene knockdown using RNAi (Voytas and Gao, 2014; Lu et al., 2015).

Physical means of inducing mutagenesis include gamma radiation, X-rays, and neutrons. Fast neutron mutagenesis is conducted by treating the seeds with ionizing radiation and has been effective in plant mutation breeding. It creates large DNA fragment deletions resulting to loss of multiple candidate genes, serious damage to chromosomes, and large segmental duplications (Bolon et al., 2014; Oladosu et al., 2016). The deletions provide gene knock outs and phenotypic variations such as chlorophyll deficiency (Campbell et al., 2015), dwarfism (Hwang et al., 2014), hyper nodulation, and chimeras (Bolon et al., 2011).

Ethyl methane sulfonate is a chemical mutagen which is frequently used for seed mutation because it is effective and induces high frequency point mutations, some of which lead to a novel stop codon for different genes (Talebi et al., 2012; Chen et al., 2013). Aside from its effectiveness, EMS is also relatively easy to handle compared to other chemical mutagens such as nitroso compounds and can be detoxified via hydrolysis for disposal (Pathirana, 2011). Although several mutant populations have been developed earlier (Meksem et al., 2008; Bolon et al., 2011, 2014), the use of fast neutron and gamma radiation induces large-scale DNA deletion and inversion, which requires more study to narrow down the main gene involved in mutant phenotype. Point mutations produced by EMS can be detected using new technologies such as next generation sequencing and TILLING (Targeting Induced Local Lesion IN Genome). EMS mutagenized population can be analyzed using two approaches: (1) forward genetics, in which apparent phenotypes are characterized before the underlying gene is identified, and (2) reverse genetics, in which mutations in the genes of interest are detected first and later linked to a specific function or phenotype (Peters et al., 2003).

Soybean mutagenesis has been widely used to characterize loci controlling important functions, develop new varieties, discover new alleles, and screen for important agronomic traits (Cooper et al., 2008; Khan and Tyagi, 2013). For example, the disrupted version of FAD2-1A gene (omega-6 fatty acid desaturase; Glyma.03G144500) genes resulted in lower linoleic acid and higher oleic acid content, respectively (Meksem et al., 2008; Dierking and Bilyeu, 2009; Lakhssassi et al., 2017). Similarly, mutations in RS2 (raffinose synthase; Glyma.03G137900) genes lead to an increase in sucrose levels and a decrease in both raffinose and stachyose oligosaccharide levels (Dierking and Bilyeu, 2009). Aside from above mentioned traits, soybean mutants have also been screened for phenotypic variations such as altered plant architectures, root phenotypes, and seeds colors (Bolon et al., 2011, 2014; Tsuda et al., 2015). Bolon et al. (2011) revealed different morphological phenotypes in FN mutant soybean plants including yellow pigmentation, curled leaves, early pod, hyper nodulation and non-nodulation, short trichome, chimeric and a short-petiole mutant with crinkled leaf (Bolon et al., 2011). Similarly, Tsuda et al. (2015) conducted consecutive EMS mutation in two generations and revealed large physiological and morphological phenotypes, resulting in a population with mutation rate of one mutation per 74 kb (Tsuda et al., 2015).

Although several mutant populations have been developed previously, most of them are from early maturity soybeans such as Williams 82 (MG III) which was the first reference genome (Schmutz et al., 2010), and only a few mutant populations were derived from late maturity cultivars including an older cultivar Forrest (Cooper et al., 2008; Lightfoot, 2008; Meksem et al., 2008). Modern day cultivars with higher yields and with other suitable traits need to be used in generating mutant population. To develop improved cultivars, there is a need to generate mutant population from late maturity groups, which are more adapted to lower latitudes. We choose to mutagenize the newly released germplasm, “JTN-5203” which is high yielding and has resistance to various diseases including soybean cyst nematode (SCN) (Arelli et al., 2015).

The success of mutation breeding program depends first and foremost on the effectiveness and efficiency of the mutagen used (Arisha et al., 2015). However, different mutagens have different effects based on the concentrations and the materials being treated. It is imperative to optimize the concentration of the mutagen before treating the bulk materials to ensure high mutation frequency and at the same time obtain enough viable seeds. High concentrations of mutagen are detrimental to plants, however, higher concentrations can also give higher mutation frequency (Shah et al., 2015). If the concentration of mutagen applied is too low, the survival rate of treated plants is higher although there is reduction in frequency of mutation (Porch et al., 2009). According to Cooper et al. (2008), an increase in mutation frequency can dramatically decrease the recovery of viable seeds. In case of EMS, an increase in concentration can significantly decreases the seed germination rate (Talebi et al., 2012).

A total of 6,400 individual M₂ mutants were generated and DNA was collected for each plant. Several phenotypes were observed such as changes in leaf morphology, plant architecture, chlorophyll content, and germination rate. Leaf phenotypes such as narrow leaf, tetra-foliate, penta-foliate, and rough texture are important traits since they affect the leaf surface area. Leaf is the main site of photosynthesis, so the larger leaf surface area increases the photosynthesis rate as the leaf receives maximized sunlight for photosynthesis. Similar phenotypic changes have been observed with other mutagenesis programs. For example, the short-petiole phenotype was also observed in a common bean EMS mutant (Guner and Myers, 2001). The short-petiole phenotype is known to be controlled by single recessive gene lps3 and is considered an important trait in increasing soybean yield by improving planting density and canopy profile (Jun et al., 2009). Also, EMS mutagenesis of peppers by Arisha et al., 2015, resulted in mutants with changes in leaf color indicating lesser chlorophyll content.

The characterization of mutant phenotypes can be useful in identifying the genes responsible for controlling plant growth and development. Of interest to our group are the changes in seed composition such as increase in total oil and proteins. There are also promising mutants that exhibited elevated levels of oleic acid and lower levels of linoleic and linolenic acid. Of interest, mutant 295-2 showed both higher oleic acid and lower linoleic and linolenic acid. These type of phenotypic alterations have been observed in previous studies and been shown to be associated with the FAD2 gene in soybean (Meksem et al., 2008; Dierking and Bilyeu, 2009; Lakhssassi et al., 2017). We are also very interested in the mutants that demonstrate high yield potential with seven times higher single plant yield compared to the controls. Mutants showing interesting phenotypes such as additional lateral branching and dense pod set may also be able to contribute to higher yield potential. Maximizing yield is the first factor that farmers consider in selecting varieties to be planted. In evaluating yield potential of these mutants, it is will be important to evaluate the heritability of candidates and plant the mutants in multi-location yield trial that incorporate appropriate experimental design to determine if these mutations will consistently perform across different environments (Qin et al., 2015).

We have shown that the mutant library described in this study has been mutated sufficiently to produce observable phenotypic variations. Since the majority of the mutations present in the genome are recessive and are unlikely to produce obvious phenotypes, we anticipate that this population should be a valuable resource for functional genomics research in soybean. The use of PCR based strategy such as TILLING should allow identification of genes with interesting mutations. This can lead to an enhancement of the genetic variability that is useful in soybean breeding programs and in potential discovery of new alleles that may be valuable to the soybean industry. Identifying the underlying genes that control these interesting phenotypes will also be critical for understanding some of the functions of genes that are important for soybean improvement. For instance, a reverse genetics approach can be employed to identify novel alleles that are involved in biochemical pathways through finding gene homologs in the Arabidopsis genome. With TILLING, mutations in the genes will be mined from the mutant population, then individuals with the desired mutations will be subjected to phenotyping. Lines will eventually be selected and used for future breeding programs and markers can also be developed to aid in marker-assisted selection. In essence, trait discovery using this publicly available soybean mutant resource can be easily facilitated.

We have used EMS to produce DNA mutations resulting in a mutant population with increased phenotypic variation. From this population, we have detected mutants with high oleic acid, oil, sucrose, protein and low linoleic and low linolenic acid contents that may be suitable for use in soybean breeding programs. In addition, further study is needed to fully understand and analyze the genetic and molecular changes underlying the phenotypic variability observed. After seed increase and confirmation of the phenotype in the next generations, mutants with interesting agronomic and value-added traits will be made available to Soybase for the scientific community access and to serve as public genetic resource for research and breeding programs. Finally, the overall goal of this project is to utilize EMS mutagenesis to develop and improve soybean germplasm and use this population as a reverse genetic tool in functional characterization of 50,000 predicted genes in soybean. Climate change leads to dryer growing seasons and emerging new pest and diseases and therefore soybean improvement and increasing its genetic diversity is necessary. Soybean is used in various industries such as animal feed, biodiesel, oil, and human food manufacturing and this mutant population is a valuable resource in screening for valuable traits in such industries as well.

AT, ME, CA, ZY, VP, and PA helped in writing the manuscript. PA provided the JTN-5203 soybean germplasm. ME, CA, AB, EA, ZY, and AT helped in planning and conducting the experiments.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.