Estimation of efficient dose for heavy-ion beam mutagenesis by whole-genome mutational analysis in Arabidopsis thaliana

Heavy-ion beams are used as an effective mutagen that induces localized mutations due to its higher linear energy transfer (LET). The observation of the occurrence ratio of albino mutants as the mutation frequency in the M2 generation after carbon-ion beam irradiation (30 keV/μm) at doses of 50–500 Gy showed that the mutation frequency increased with increasing dose between 50–350 Gy and plateaued at 400 Gy.1) However, the measurement of mutation frequency by visual screening of mutants from a large number of irradiated plants is time-consuming. In this study, we tried to estimate an efficient dose for heavy-ion beam mutagenesis by performing whole-genome mutational analysis on relatively small numbers of irradiated plants. Dry seeds of Arabidopsis thaliana (the Col-0 strain) were irradiated with C (135 MeV/nucleon) ions, at doses of 0–350 Gy. The LET of C ion beams was controlled to 30 keV/μm. M2 seeds were harvested from selfpollinated M1 plants, and ten M3 plants were harvested from one self-pollinated M2 plant. Genomic DNA was extracted from the mixture of leaves of the ten M3 plants. Five DNA pools were sequenced for each dose using HiSeq X-Ten sequencing systems (Illumina Inc.). The read sequences obtained were input into the mutational analysis pipeline AMAP, as described previously, with some modifications.2) In short, after mapping the read sequences to the reference genome sequence (TAIR10) with BWA (BWA-MEM) software, the mutation candidates were detected with GATK (HaplotypeCaller), PINDEL, and BREAKDANCER software. Then, the AMAP filtered out the false-positives by using its own algorithm. In spite of the process of filtering out false-positives, confirmation of the mutation candidates in one DNA pool (350-Gy irradiation) by using Integrative Genomics Viewer (IGV) software revealed that 81% of the candidates were still false-positives (data not shown). We configured additional criteria for the determination of false-positives. Among the mutation candidates output from GATK, those with the ‘QUAL’ value (calculated by GATK) less than 150 or ‘MAPPING_QUAL’ value (calculated by BWA) less than 60 were treated as false-positives. Among the mutation candidates output from PINDEL, those with the ratio of the number of read sequences supporting the mutation to that of mapped read sequences less than 0.2 or with the number of reads supporting the mutation less than 5 were treated as false-positives. Among the mutation candidates output from BREAKDANCER, those with the ratio of the number of read sequences supporting the mutation to that of mapped read sequences less than 0.15 were treated as false-positives. Also, among the mutation candidates