Frequent generation of mutants with coincidental changes in multiple traits via ion-beam irradiation in soybean

Shun Mikuriya, Megumi Kasai, Kenta Nakashima, Natasia, Yoshihiro Hase, Tetsuya Yamada, Jun Abe and Akira Kanazawa* Research Faculty of Agriculture, Hokkaido University, Kita 9, Nishi 9, Kita-ku, Sapporo, Hokkaido 060-8589, Japan Takasaki Advanced Radiation Research Institute, National Institutes for Quantum and Radiological Science and Technology, 1233 Watanuki-machi, Takasaki, Gunma 370-1292, Japan


INTRODUCTION
Mutagenesis has been used to engineer novel traits in a wide range of plants.In soybean (Glycine max), early studies showed that irradiation with X-rays or thermal neutrons can result in genetic variation for traits such as yield, plant height, flowering time, seed size (Rawlings et al., 1958), and oil and protein contents in seeds (Williams and Hanway, 1961).Subsequently, plants have been produced with desirable traits in terms of seed components (Takagi et al., 1990;Fehr et al., 1991;Hajika et al., 1991;Kitamura, 1991;Takahashi et al., 1994;Hayashi et al., 1998), nodule formation (Carroll et al., 1985) or herbicide tolerance (Sebastian et al., 1989) via mutagenesis using X-ray or gamma-ray irradiation or chemicals.Recently, mutants have been generated using fast neutrons (Bolon et al., 2011) or ethyl methanesulfonate (Tsuda et al., 2015) in combination with a high-throughput next-generation sequencing approach, indicating the feasibility of these new methods to obtain mutants.
Soybean is a paleotetraploid plant, and its genome comprises a large number of duplicated genes: nearly 75% of the genes are present in multiple copies as a consequence of genome duplications that occurred approximately 59 and 13 million years ago (Schmutz et al., 2010;Cannon and Shoemaker, 2012).In addition, genes are sometimes duplicated in tandem in the soybean genome, and some of them have been proved to be functionally redundant (e.g., Yoshino et al., 2002;Kong et al., 2010).Such features can potentially be a constraint on the production of novel mutants through conventional methods of mutagenesis.
Gaining increasing attention as a mutagen to produce novel mutants, ion-beam irradiation has proved to be suitable for producing mutants with a high frequency (Shikazono et al., 2003) and a broad spectrum of traits (Shikazono et al., 1998) in various plant species (Tanaka, 2009).The high linear energy transfer (LET) of ion-beam radiation causes a profound change in the genome of an organism such as deletion, inversion or translocation of a large DNA segment (Shikazono et al., 2005;Tanaka et al., 2010;Hirano et al., 2012).
Because of these characteristics, we first used ion-beam irradiation to mutagenize soybean in the years 2005 to 2010.By monitoring the generation of plants with a visibly altered trait in the M 2 generation and optimizing the method and dosage of irradiation, we demonstrated for the first time that ion-beam irradiation can induce mutations in soybean (Arase et al., 2011).Among the population of M 2 plants, we found mutants with chlorophyll-deficient phenotypes including pale-green-leaf mutants, albino mutants and variegated-leaf mutants.Subsequently, in addition to expanding the mutant population by the same approach, we have been maintaining some of the progeny of plants with these altered phenotypes through self-fertilization and have fixed the traits.In the present study, we analyzed changes in the chlorophyll content during the development of these plants.We also characterized them in terms of flowering time and seed components, agronomic traits that are important for the production of soybean and the use of its products, respectively.For seed components, we analyzed the level of isoflavones, major secondary metabolites in soybean seeds that are beneficial to human health (Jung et al., 2000 and references therein), to characterize mutants in terms of changes in secondary metabolism.We report frequent coincidental changes in chlorophyll deficiency and flowering time as a consequence of ion-beam irradiation in soybean and discuss a possible mechanism(s) to account for this phenomenon.

MATERIALS AND METHODS
Plant materials and mutagenesis Populations of plants harboring mutants of the soybean cultivar 'Nourin No. 2' were previously obtained through ion-beam irradiation (Arase et al., 2011).As well as these mutant populations, additional mutant populations of this cultivar and another soybean cultivar, 'Kariyutaka', were produced, and plant lines derived from these populations were used in this study.Dried soybean seeds were exposed to 320-MeV carbon ions (LET 76 keV μm − 1 ) generated by an AVF cyclotron (Japan Atomic Energy Agency, Takasaki, Japan) as described previously (Arase et al., 2011).Taking into account the limited penetration distance of ion beams, we placed soybean seeds with the hilum side facing the irradiation source in order to ensure irradiation of the meristem tissues at an equal dose.The dose of irradiation was optimized on the basis of the effects of irradiation at doses ranging up to 20 or 30 Gy on plant growth and seed yield (Arase et al., 2011).The irradiation doses for mutagenesis were 2.5, 5.0 or 7.5 Gy for Nourin No. 2, and 5.0, 10 or 15 Gy for Kariyutaka.Some mutants were produced via recurrent irradiation.Chlorophyll-deficient mutants were screened on the basis of visible phenotypic changes in the M 1 or M 2 generation in the field.For both cultivars, progeny (M 3 to M 6 generations) of plants that had visibly altered phenotypes were obtained by self-fertilization.These progeny plants comprised those that maintained visible chlorophyll deficiency and those that did not, the latter of which were used for analyses as potential mutants.Mutant lines are listed in Supplementary Fig. S1.
Field experiments Field-grown plants were used for analyses of chlorophyll content and flowering time.Field experiments were conducted in 2014 on a Haplic Gray Lowland soil at the experimental field of Hokkaido University, Sapporo, Japan (43°07′N, 141°35′E).Seeds were sown in paper pots, and after growing for one week in a greenhouse, seedlings were transplanted into soil on 6 June 2014.Rows were 80 cm apart and plants in the same row were 25 cm apart.Three to six plants per mutant line were grown.Fertilizer was applied at a rate of 12 kg ha − 1 N, 39 kg ha − 1 P 2 O 5 and 30 kg ha − 1 K 2 O.
Analysis of chlorophyll content Chlorophyll content in the leaf tissues of field-grown plants was analyzed using the SPAD-502Plus chlorophyll meter (Konica Minolta) according to the manufacturer's instruction.Three to five individual plants per line were used for the analysis unless otherwise noted.Chlorophyll content in one of the leaflets of each trifoliolate leaf (1-7 or 8) was measured three times as soon as it had fully expanded to obtain three replicates.

Analysis of flowering time
The time from sowing to the first flower and the duration of flowering, namely, from the date that plants displayed the first flower until the date that plants no longer developed new flowers, were monitored.Data for the days from seeding to first flowering were obtained from three to five plants per plant line unless otherwise noted and used for comparison between different lines.
Analysis of isoflavone content Isoflavones were extracted from soybean seeds and analyzed by high-performance liquid chromatography (HPLC) according to a method described previously (Nagamatsu et al., 2007).
Statistical analyses Differences between mean values for the soybean lines were compared using analysis of variance and the Tukey-Kramer multiple comparison test.

RESULTS
Production and maintenance of mutant lines We previously demonstrated the efficacy of ion-beam irradia-tion as a means of mutagenesis in the soybean cultivar Nourin No. 2 (Arase et al., 2011).In addition to extending the mutant populations of Nourin No. 2, we used the same procedure to produce new mutant populations using another cultivar, Kariyutaka, an early-flowering plant, to examine whether a different spectrum of mutation could be generated when plants having different characteristics were used.Plants with a visibly altered phenotype were screened from the M 2 population of these plants.Plants of subsequent generations were grown from seeds that were randomly chosen from those produced via selffertilization.We refer to progeny from mutagenized plants of Nourin No. 2 as "N2-derived lines", and those of Kariyutaka as "KY-derived lines".These plants were used for subsequent analyses.Content was measured on one of the leaflets of the trifoliolate leaves (for three replicates) for each of 3-5 individual plants per line using the SPAD chlorophyll meter, except for N2-1 (only one individual was measured).Means that differed significantly from the mean of wild-type plants are indicated by * (P < 0.05) or ** (P < 0.01).Lanes 1-21 indicate line N2-1 to line N2-21, respectively.Lane 22, N2 B-1; lane 23, N2 B-2; lane 24, N2 C-1; lane 25, N2 C-2; lane 26, N2 D-1; lane 27, N2 D-2; lane 28, N2 M1a-1; lane 29, N2 M1a-3; lane 30, N2 M1a-4; lane 31, N2 M1a-1 EE; lane 32, N2 wild type (WT).Plants of the M 5 generation were used for N2-1 to N2-21, N2 C-2, N2 M1a-1 to N2 M1a-4 and N2 M1a-1 EE; plants of the M 6 generation were used for N2 B-1, N2 B-2, N2 C-1, N2 D-1 and N2 D-2.Mutant lines of Nourin No. 2 have been named on the basis of populations used for screening: B, C and D in line names refer to plant populations used for our initial screening (Arase et al., 2011); "M1a" refers to the progeny of a particular mutant plant.
Changes in chlorophyll content Analysis of the SPAD values for chlorophyll content in each trifoliolate leaf indicated significant differences in chlorophyll content between plant lines with regard to all the analyzed leaves, namely, the 1st to 8th trifoliolate leaves of N2-derived lines (F = 10.5-41.7,P < 0.001) and the 1st to 7th trifoliolate leaves of KY-derived lines (F = 11.8-38.2,P < 0.001).
Figures 1 and 2 show the chlorophyll content in the leaves of N2-derived lines and KY-derived lines, respectively.None of the 31 N2-derived lines had a higher level of chlorophyll than the wild-type plants, especially with regard to the 1st to 5th leaves (for the 1st, 3rd and 5th leaves, see Fig. 1).The chlorophyll content was reduced the most in the 1st leaf: the greatest reduction was detected in the 1st leaf of N2 D-2, which had only 23% of the chlorophyll level of the wild type.A multiple comparison test indicated that 26 lines had a significantly lower level of chlorophyll in one or more of the 1st, 3rd and 5th leaves (Fig. 1).Similarly, a lower chlorophyll content was found for some of the 10 KY-derived lines analyzed (Fig. 2): KY-1, KY-3, KY-6, KY-7, KY-8 and KY-10 lines had a significantly lower level of chlorophyll in one or more of the 1st, 3rd and 5th leaves.
We examined changes in the chlorophyll content during development until almost all leaves had developed in plant lines that had a significant decrease in the chlorophyll content in the 1st, 3rd or 5th leaf (Figs. 3 and  4).This analysis indicated the presence of variation in the developmental changes in chlorophyll content.For example, the chlorophyll content of the 1st leaf was higher than that of the 3rd, 4th or 5th leaf in the N2 wild type.This pattern was also observed in most of the N2-derived lines but not in N2-2, N2 B-2, N2 C-1, N2 D-2, N2 M1a-1, -3, -4 or -1 EE (Fig. 3).In KY-derived lines, the pattern of changes was somewhat similar to the wild type, which had a sharp increase in chlorophyll from the 4th to 5th leaves (as did KY-1, KY-3, KY-6 and KY-8); slightly more moderate changes (KY-7 and KY-10) were also measured (Fig. 4).We also analyzed changes in the   Ion beam-induced mutations in soybean chlorophyll content from the 1st to 4th leaves of plants in the subsequent two years and generations (Supplementary Fig. S2).The pattern of developmental changes in the wild-type plants was different in different years, which indicates that the observed patterns of changes in chlorophyll content are under the influence of an environmental factor(s).Meanwhile, some of the mutant lines that had a profoundly reduced level of chlorophyll also had different patterns in different years, which were also different from those of the wild-type plants.Taken together, although such developmental changes can thus be influenced by an environmental factor(s), these results indicate that both the level and pattern of developmental changes in chlorophyll content were altered by mutagenesis.
Changes in flowering time When the time from sowing to the first flower and the period of flowering were monitored, the number of days from seeding to the first flowering differed significantly among the N2-derived lines (F = 183.8,P < 0.001) and among the KY-derived lines (F = 25.6,P < 0.001).Among the N2-derived lines, Figure 5 N2 some flowered significantly earlier than the wild type (N2 M1a-1, N2 M1a-3 and N2 M1a-1 EE) and others flowered significantly later (N2-11, N2-15, N2-16, N2 B-1, N2 B-2, N2 C-1, N2 C-2, N2 D-2 and N2 M1a-4) (Fig. 5A).In the KY-derived lines, only lines that flowered later than wild-type plants (KY-3, KY-7, KY-8 and KY-10) occurred; none flowered earlier (Fig. 5B).The period of flowering was shorter, especially in lines that had late flowering for both N2-derived and KY-derived lines (Fig. 5).content in relation to flowering time in the N2-derived lines (Fig. 6) and KY-derived lines (Fig. 7), lines with an altered flowering time also had altered chlorophyll content.More specifically, lines that flowered earlier than wild-type plants and those that flowered later both had a lower level of chlorophyll in multiple leaves than wildtype plants.Such coincidental phenotypic changes were observed in 12/31 of N2-derived lines and 4/10 of KYderived lines.

Changes in seed isoflavones
In seeds produced from plants of several N2-derived lines that had reduced lev-els of chlorophyll in the leaf tissues, we also examined levels of isoflavones, the major flavonoid compounds in soybean seeds.Changes in the levels of glucosides (daidzin and genistin) and malonylglucosides (malonyldaidzin and malonylgenistin) of isoflavones in the seeds produced in five N2-derived lines were analyzed by HPLC (Fig. 8).The levels of these compounds, in particular those of genistin and malonylgenistin, were lower in most of these N2-derived lines than in wild-type plants, which indicated that the observed chlorophyll deficiency in the leaf tissues could be accompanied by a decrease in isoflavones in seeds.The relative content profiles appeared very similar between the four isoflavone compounds, which implies that a process or a factor that is common to the synthesis and/or accumulation of all of these compounds was suppressed as a consequence of mutagenesis, e.g., due to reduction of a substrate for isoflavone synthesis.radiation and bring about a large deletion and/or rearrangement of a DNA segment (Shikazono et al., 2005;Tanaka et al., 2010;Hirano et al., 2012).A genome-wide analysis of structural changes induced by ion-beam irradiation in Arabidopsis revealed that deletions as large as 1.2 Mbp could occur (Hirano et al., 2015).Such a large deletion can involve the loss of multiple linked genes and, therefore, may explain some of the coincidental changes observed in this study.
A second possibility is that mutations in genes that affect one of these traits indirectly affected the other trait.Transcriptome analyses in Arabidopsis have shown that the expression of genes involved in various metabolic pathways (e.g., carbon and nitrogen assimilation, carbohydrate metabolism, flavonoid biosynthesis, ascorbate glutathione cycle) is affected in albino or pale green mutants (Satou et al., 2014).This observation is consistent with our finding that the isoflavone content in seeds often decreased in soybean plants that had reduced chlorophyll content in the leaf tissues (Fig. 8).Likewise, metabolic changes associated with the chlorophyll deficiency might affect flowering time, although not all plants with chlorophyll deficiency had altered flowering time.Intriguingly, all lines that had a reduced level of isoflavone had a reduced level of chlorophyll in all of the 1st, 3rd and 5th leaves.Similarly, all the lines that had an altered flowering time except for N2-11 and N2-15 had a reduced level of chlorophyll in all of the 1st, 3rd and 5th leaves, and N2-11 and N2-15 had a reduced level of chlorophyll in the 1st and 3rd leaves.On the other hand, lines that had a reduced level of chlorophyll only in the 1st leaf did not have altered flowering time.It seems probable that changes in multiple traits are associated with chlorophyll deficiency that persisted throughout plant development rather than that occurred temporarily or conditionally.Additionally, it is also possible that a gene(s) that affects both traits might have been mutated.An example in Arabidopsis relevant to this possibility is that a mutation in the LHY gene, which encodes a transcription factor, alters both flowering time and chlorophyll content along with circadian rhythms (Schaffer et al., 1998).Similarly, the EF8 gene in rice delays flowering and negatively regulates chlorophyll biogenesis (Feng et al., 2014).Resequencing analysis of the genomes of mutant lines may reveal which one or more of these mechanisms is relevant to the observation.

DISCUSSION
Here we found that multiple traits could be affected in the same progeny plants derived from ion-beamirradiated plants.The observed differences in the extent of decrease and/or the pattern of developmental changes of chlorophyll content suggest that the set of ion-beammutagenized progeny lines harbors various mutations affecting chlorophyll level.We also found that while the pattern of developmental changes of chlorophyll content could be influenced by an environmental factor(s), mutant lines had a developmental pattern of chlorophyll accumulation that differed from that of wild type.Genetic changes that resulted in an altered developmental pattern of chlorophyll accumulation may reflect an altered response to environmental conditions.At least 20 loci that are responsible for chlorophyll deficiency have been identified in soybean (Palmer et al., 1990(Palmer et al., , 2000;;Zou et al., 2003;Kato and Palmer, 2004;Zhang et al., 2011 and references therein), some of which might be mutated in these lines.Reports on phenotypic changes associated with mutations at these loci that affect chlorophyll content are limited (Zhang et al., 2011); hence, whether a mutation at a single locus can have pleiotropic effects remains largely unexplored.
Coincidental changes in chlorophyll content and flowering time: possible mechanisms Particularly intriguing in this study is the finding that chlorophyll content and flowering time were frequently altered coincidentally (12/31 of N2-derived lines; 4/10 of KY-derived lines).Several patterns of mutations could lead to the generation of this phenomenon.First, it is possible that mutations were induced in genes that affect chlorophyll content and genes that affect flowering time in the same plant.An observation that is consistent with this notion comes from the generational difference in the appearance of phenotypic changes in line N2 M1a-1 EE.The first phenotypic change in this line was, unusually, generated in the M 1 generation: the plant showed chlorophyll deficiency.We found that all plants of the next generation that were analyzed (7/7) had chlorophyll deficiency and two of them had early flowering in addition to chlorophyll deficiency.This observation suggests that at least two genes, one that affects chlorophyll content and another that affects flowering time, were mutated in this line.We found segregation of a visible chlorophylldeficient phenotype in N2-17 and N2-21 in the M 3 and M 4 generations, respectively, but not in other plants (for segregation in the M 3 generation, see Supplementary Fig. S1).There was no noticeable segregation in flowering time in the generation after screening except for N2 M1a-1 EE plants.These observations suggest that mutated loci that mainly account for the altered phenotypes have been fixed.Ion beams comprise high-LET 2002) in rice and 1-3% (Kazama et al., 2008) to 8.4% (Shikazono et al., 2003) in Arabidopsis.The frequency of chlorophyll-deficient mutants has long been used as an indicator of the frequency of mutation in various plants including soybean (Carroll et al., 1986).However, we are not aware of any reports that particularly focused on frequent production of mutants with coincidental changes in chlorophyll content and flowering time through mutagenesis.Assuming that such a phenomenon occurs at a higher rate in soybean, a factor potentially relevant to the phenomenon is the characteristics of the soybean genome, in particular its duplicated composition.Mutation of a limited number of copies of duplicated genes may result in a unique phenotype when the duplicated genes confer an additive effect on a trait.Exemplifying this possibility in soybean is the generation of an intermediate phenotype caused by dysfunction of one of the multiple phytochrome A genes via an insertion of a retrotransposon (Liu et al., 2008).A recent transcriptome analysis indicated that approximately 50% of duplicated genes in the soybean genome have undergone subfunctionalization in terms of gene expression (Roulin et al., 2013).The presence of a large amount of subfunctionalized duplicated genes potentially contributes to the complexity of phenotypic changes brought about by mutagenesis.
Despite technological progress such as transposonmediated mutagenesis (Mathieu et al., 2009;Hancock et al., 2011) or genome engineering (Curtin et al., 2011;Jacobs et al., 2015) in soybean, the present observations indicate that a novel spectrum of mutations can be explored by ion-beam irradiation.Thus, our study highlights the plasticity of the soybean genome, which is manifested when combined with the highly mutagenic activity of ion-beam irradiation, and yields new insights into the relationship between induction of mutation and genome organization.

Fig. 4 .
Fig. 4. Mean ( ± SE) chlorophyll content in trifoliolate leaves 1 to 7 in plants of KY-derived lines.Content was significantly reduced in one or more of the 1st, 3rd and 5th trifoliolate leaves compared with wild-type plants as assessed by a Tukey-Kramer test.The seven bars (from light to dark) for each plant line correspond to trifoliolates 1 to 7. Three to five individual plants per line were each measured three times except for the 1st leaf of KY-8 (only one individual was measured).N.A., not analyzed (KY-3, 1st trifoliolate); WT, wild-type plants.

Fig. 5 .
Fig. 5. Variation in period of flowering in N2-derived and KYderived lines.Shaded bars indicate the period of flowering, from the date of the first flower until the date that plants no longer had a flower.Wild-type data are highlighted as filled bars.Numbers above the line at the top of each panel indicate days from sowing.A, N2-derived lines; B, KY-derived lines.The number of days from seeding to the first flowering was statistically analyzed: means that differed significantly from the mean of wild-type plants are indicated by * (P < 0.05) or ** (P < 0.01).