Dry seeds of A. thaliana ecotype Columbia (Col-0) were irradiated with heavy-ion beams as previously described (Kazama et al., 2008). In short, the seeds were irradiated with C-ion beams with LETs of 100 keV μm^-1 and 200 keV μm^-1, and Ar-ion beams with an LET of 290 keV μm^-1 at doses of 150 Gy, 75 Gy, and 50 Gy, respectively, using the E5 beam line in the RIKEN RI-beam factory. The penetration distances of C-ion beams with LETs of 100 keV μm^-1 and 200 keV μm^-1, and Ar-ion beams with an LET of 290 keV μm^-1 in water were calculated to be 1.0 mm, 0.12 mm, and 6.0 mm, respectively, using the SRIM-2013 code ‘The Stopping and Range of Ions in Matter (SRIM)’ (http://www.srim.org). All LET values were calculated behind seeds. Irradiation doses were determined as they showed around 90% survivals in the M1 generation, which are defined as the most effective doses in heavy-ion-beam mutagenesis (Kazama et al., 2008, 2011). More than 1,000 seeds were irradiated for each condition.

The irradiated M₁ seeds were surface-sterilized by dipping in 1% sodium hypochlorite for 10 min, washed five times with autoclaved Milli-Q water (1 mL each time), and incubated on Murashige and Skoog medium supplemented with 3% sucrose and 0.7% agar at 4°C in the dark for 4 d to induce vernalization. Subsequently, the seeds were incubated at 22°C under long-day conditions (16 h light, 8 h dark) with white light from fluorescent lamps at approximately 50-100 µM m^-2 s^-1 light intensity. Over 100 seedlings that developed true leaves were transplanted into plastic trays (13 × 9 cm²) that contained soil. Two seedlings were planted in each tray and grown at 22°C under long-day conditions in a growth chamber. The M₂ self-pollinated seeds were collected from each M₁ plant.

For each M₁ plant, forty M₂ plants were grown in the same way as M₁ plants. When the growing M₂ plants were transplanted into the plastic trays, their phenotypes were checked and the M₂ plants were divided into apparently normal plants and mutants; the former showed Col-0 like phenotypes and the latter showed visible mutant phenotypes. From each M₂ line, one normal plant and one mutant were randomly selected and transplanted into the plastic tray. Phenotypes of the mutants were checked again 30 days and 40 days after the cultivation started. For the lines showing no visible mutant phenotypes, two normal plants were randomly selected and transplanted. M₃ self-pollinated seed aliquots (40 grains each) derived from each M₂ plant were sown and their seedlings were grown in the same way as the M₁ plants ( Supplementary Figure S1 ). The M₃ lines were named according to the following rule: “irradiation type” (Ar-ion beams with an LET of 290 keV μm^-1: Ar50, C-ion beams with LETs of 100 keV μm^-1: C100, and 200 keV μm^-1: C200) - “line identification number in the M₂ generation” - “phenotype” (N means normal), for example, Ar50-01-N1. After confirmation of their phenotypes when they were in the M₂ generation, leaves of 40 M₃ plants were harvested and bulked to be used for DNA extraction, which were used for array CGH. At the same time, leaves of 15 individuals were harvested one by one to extract DNA from each individual, which were used for PCR and qPCR analyses.

TAGs were listed based on a previous study (Rizzon et al., 2006). The criteria “Low stringency” was adopted; if sequence similarity between two or more tandemly arrayed genes was 70% or more, they were listed as TAGs. Tandemly arrayed homologous genes having another gene between them were also included. Then, corresponding genes were extracted from the whole-genome sequence of A. thaliana (TAIR10, http://www.arabidopsis.org/). For each region having TAGs, 1-kbp upstream to 1-kbp downstream regions were tiled with oligonucleotides ( Supplementary Table S1 ). In total, 3,469 genes were mounted on the DNA array. The sequence of the oligonucleotide probes was initiated every 100 bp across the genome sequence, excluding repetitive sequences, and probe lengths ranged from 50 to 75 bp.

DNA was extracted from bulks of M₃ leaves by using a DNA extraction kit, Mag Extractor Plant Genome (Toyobo, Osaka, Japan), followed by the purification using High Pure PCR Cleanup Micro Kit (Roche Diagnostics GmbH, Mannheim, Germany). The DNA of an M₃ bulk and that of another M₃ bulk derived from different irradiation treatment was labelled with Cy5 and Cy3, respectively. Hybridization, washing, and scanning of the array CGH were conducted as previously described (Hirano et al., 2015). The positive signals of deletion were extracted as previously described (Hirano et al., 2015). In brief, for each spot on the array, signal values were calculated as log2 ratios of the Cy3-labeled sample (mutant) versus the Cy5-reference (wild type). Peak detection was performed using the Find Peaks feature in SignalMap software, version 1.9 (Roche NimbleGen Inc.) with the peak window size of 400 or 500 bp and the peak threshold of 35% instead of 10% in the previous study.

The candidate deletions detected as positive peaks in the array CGH were confirmed by PCR using genomic DNA from the 15 individual M₃ plants. Two primer sets were used for each candidate deletion ( Supplementary Table S2 ). The deletions which were confirmed by PCR were defined as homozygously inherited deletions ( Supplementary Table S2 ). For the deletions that were not confirmed by PCR, in which the same DNA fragment as in the wild-type plant was amplified, qPCR was performed on genomic DNA from the seven individual M₃ plants by using LightCycler and the Universal Probe Library detection format (Roche Diagnostics, Penzberg, Germany). Relative amplification ratios between the deleted and non-deleted regions were calculated by using the ΔΔCp method as previously described (Kazama et al., 2015). One primer set was used for each candidate deletion ( Supplementary Table S3 ). Deletions determined by qPCR in at least one of the seven siblings to have half the amount of DNA were defined as heterozygously inherited deletions. ( Supplementary Table S3 ).

Genomic DNA was extracted from the collected leaves of 10-15 M₃ plants using Extractor Plant Genome (Toyobo), followed by the purification using High Pure PCR Cleanup Micro Kit (Roche Diagnostics GmbH). The extracted DNA was sequenced using the HiSeq 4000 sequencing system (Illumina Inc., https://www.illumina.com) as described previously (Kazama et al., 2017). The read sequences obtained were analyzed by using AMAP as described previously (Ishii et al., 2016). The detected candidate mutations were visually confirmed by using the Integrative Genomics Viewer (IGV; Robinson et al., 2011). Visualization of genomic locations of the detected deletions, rearrangements, and essential genes were performed using Circos (Krzywinski et al., 2009).

Seeds of 90 M₃ lines that were confirmed to possess deletions in this study except for Ar50-76-pl1 and C200-11-N2 lines were used for metabolic profiling ( Supplementary Table S4 ). For each line, three replicates were prepared. Metabolic profiling by widely targeted metabolome analysis (single-grain-based) was conducted as previously described (Sawada et al., 2017). Metabolome data matrix with 128 metabolite ( Supplementary Table S5 ) intensities obtained by LC-ESI-QqQ-MS analysis (UPLC-TQS, Waters) was generated from the 270 samples: 90 mutant lines × three replicates derived from each individual plant. The missing values of signal intensities and the values less than 10 were set to 10. Metabolites with a signal-to-noise ratio (defined as the ratio of the averaged signal intensity to that of the extraction-solvent control) < 10 in all experimental groups were removed. In addition, metabolites that showed low signal-to-noise ratio in more than 30% of experimental groups were removed, leaving 75 metabolites. The intensities of the 75 metabolites were divided by those of the internal standards (80% methanol, 0.1% formic acid, 16.8 nmol L^-1 lidocaine, and 105 mol L^-1 10-camphorsulfonic), resulting in the metabolic profiles ( Supplementary Table S4 ). Further analyses were conducted using MetaboAnalyst 5.0 (Chong et al., 2018). In brief, missing values were replaced by 1/5 of minimum positive values of their corresponding intensities. Then, intensities were normalized by median. Univariate analysis was conducted by one-way Analysis of Variance (ANOVA) ( Supplementary Table S6 ). Hierarchical clustering was performed using ‘euclidean’ distance measure and ‘ward.D.’ clustering algorithm.

We merged the list of 510 embryo-defective (EMB) genes (Meinke, 2019) and 705 genes whose homologous deletions are lethal (Lloyd et al., 2015) into a list consisting of 811 essential genes ( Supplementary Table S7 ). We cited 1,765 genes that possess mutations (including 10 or more nonsynonymous SNPs, nonsense SNPs, frameshift small indels, full CDS deletions, or partial deletions) in ecotype Landsberg (Lu et al., 2012), i.e., genes that are most likely not essential for A. thaliana, and excluded 25 genes (AT1G24340, AT1G30610, AT1G62340, AT2G03870, AT2G15820, AT2G18510, AT2G24840, AT2G28880, AT2G32590, AT2G33160, AT3G05770, AT3G23110, AT3G24560, AT3G55400, AT4G04790, AT4G13750, AT4G16144, AT4G19490, AT4G21100, AT4G27010, AT4G28590, AT5G08080, AT5G24670, AT5G37630, and AT5G39750) included in the essential genes. We defined the consequent 1,740 genes as mutable genes.

The reference genome sequence was divided into 1-Mb bins. In each bin, the numbers of overlapping breakpoints of chromosomal rearrangements ( Supplementary Table S8 ) and essential genes ( Supplementary Table S7 ) were counted. Multiple breakpoints derived from the same mutant line were counted as one. The correlation coefficient was calculated using cor.test function implemented in R software environment (Ihaka and Gentleman, 1996).

To investigate the effect of high-LET beam on inducing the deletion of TAGs, the dry seeds of Arabidopsis thaliana Col-0 were irradiated by heavy-ion beams with three conditions: C-ion beams with LETs of 100 keV μm^-1 and 200 keV μm^-1, and Ar-ion beams with an LET of 290 keV μm^-1 (see Materials and Method). For each irradiation condition, 96 M₂ lines were grown ( Table 1 ), and 96 M₂-lines-derived 192 M₃ lines (two M₃ lines per one M₂ line) were used for the array CGH analysis. As a result, 266 putative deletion signals were detected ( Supplementary Table S2 ).

There are two types of deletions induced by heavy-ion irradiation; one is homozygously inherited deletion that shows mendelian inheritance and the other is deletion that is only inherited heterozygously because of homozygous lethality. To confirm these putative deletions that array CGH analysis sometimes false-positively detects, and to investigate inheritance pattern, PCR and qPCR were performed on the siblings of the M₃ lines. The deletions which were confirmed by PCR were defined as homozygously inherited deletions ( Supplementary Table S2 ). For the deletions not confirmed by PCR, qPCR was conducted to identify heterozygously inherited deletions ( Supplementary Table S3 ). Deletions in which at least one sibling showed the result “+” (fragments amplified but with one more amplification cycle) or “-” (fragment not amplified) were considered positive for heterozygous inheritance. Reverting to M₂ lines, 103 deletions were detected in 79 out of 288 lines by array CGH analysis ( Table 1 ) which included 163 and 125 M₂ lines with and without visible phenotypic abnormality, respectively. The number of induced deletions tended to increase of the value of LET, though it was not significant (Kruskal-Wallis test). Considering that the number of ion particles irradiated on seeds was highest in the irradiation with LET of 100 keV μm^-1, irradiation with higher LET is likely to induce deletions efficiently. One or more deletions were detected in 19% and 36% of the lines with and without visible phenotypic abnormality, respectively, resulting in no significant difference (chi-square test). This suggests that at least deletions on TAGs detected by array CGH analysis were induced regardless of phenotypic abnormality. We focused on the number of deleted TAGs detected by the array CGH analysis. There were three types of disruption on TAGs: loss of only a part of the TAGs (partial), deletion of one set of TAGs, and two or more sets of TAGs loss ( Table 2 ). Deletions covering only one TAGs were more likely to be homozygous while those covering two or more TAGs were more likely to be heterozygous (p < 0.01, chi-square test) ( Table 2 ). Deletions covering partial TAG were also more likely to be homozygous than those covering two or more TAGs (p < 0.01, chi-square test). The number of deleted TAGs tended to increase with increase of the value of LET, though only the difference between 100 and 290 keV μm^-1 was significant (p = 0.047, chi-square test).

We then asked how the deletions affect the accumulation of metabolites by conducting metabolic profiling of 126 metabolites on 90 out of 94 M₃ lines possessing deletions ( Supplementary Tables S4, S5 ). Seventy-five metabolites showed significantly different intensity between some combinations of the lines ( Supplementary Figure S2 , Supplementary Table S6 ). This suggests that metabolisms of irradiated line were affected by deletions regardless of phenotypic abnormality, or that mutations undetected by array CGH analysis affected the amounts of the metabolites.

To compare the sizes between homozygously and heterozygously inherited deletions in the 94 M₃ lines, both ends of probes showing deletion and both ends of deletions at base-pair levels were used to determine deletion size in the array-based detection ( Supplementary Table S9 ). Statistical analysis revealed that there was no significant difference in deletion size between the irradiation treatments with different LETs (p > 0.05; Kruskal-Wallis rank sum test; Figure 1A ). This tendency did not change when deletion sizes of homozygous and heterozygous deletions were compared separately ( Figures 1B, C ). However, at each irradiation condition, the mean size of heterozygously inherited deletions was significantly larger than that of homozygously inherited deletions (p < 0.05; Wilcoxon rank-sum test; Figures 1B, C ). This tendency may be attributed to their lengths; longer deletions are more likely to cover essential genes which make them homozygous lethal. This finding raises a question about how the homozygous and heterozygous deletions are distributed in the Arabidopsis genome.

We compared the distances between the closest pair of essential genes as theoretical maximum sizes of homozygous deletions and the sizes of observed homozygous or heterozygous deletions regardless of the LET value. The mean size of the homozygous deletions was significantly smaller than that of the distances between the closest pairs of essential genes ( Figure 2 , P < 0.01; Wilcoxon rank-sum test). This result strongly suggests that the mean size of homozygous deletions seemed to be influenced by the distribution of essential genes. Then we investigated the influence of essential genes on the distributions of homozygous and heterozygous deletions of the mutants. For the heterozygously inherited deletions that are relatively larger than others, genome resequencing was performed to confirm accurate sizes of the deletions as previously described ( Supplementary Table S9 ; Hirano et al., 2015; Kazama et al., 2017). We then collected 94 homozygous and 55 heterozygous deletions (> 100 bp) from 94 M₃ lines possessing deletions isolated in this study and 22 mutant lines previously reported ( Supplementary Table S9 ; Hirano et al., 2015; Kazama et al., 2017; Hase et al., 2020; Sanjaya et al., 2021). Next, we investigated the distribution of 1,740 mutable genes. In the 94 homozygous and 55 heterozygous deletions, 593 and 4,061 genes were overlapped, respectively. The mutable genes occupied 10% of the genes in homozygous (62 out of 593) and 6.6% of the genes in heterozygous (270 out of 4,061) deletions, respectively (p < 0.01, chi-square test) ( Figure 3 ). Second, we investigated the overlaps of the essential genes with the homozygous or heterozygous deletions ( Figure 3 ). No essential gene (0 out of 593) overlapped with homozygous deletions while 2.2% (90 out of 4,061) of the genes in the heterozygous deletions were essential genes (p < 0.01, chi-square test). This contrasting situation can be attributed to the nature of the essential genes that deletions including them cannot be inherited homozygously. These results suggest that the distribution of essential genes affects the upper limit of the deletion size ( Figure 2 ).

Through whole-genome mutation analysis, we and other groups have identified chromosomal rearrangements as well as deletions induced by heavy-ion-beam irradiation (Hirano et al., 2015; Kazama et al., 2017; Hase et al., 2020; Sanjaya et al., 2021). To investigate whether the locations of the breakpoints were influenced by essential genes, we compared the locations of the 532 breakpoints detected by NGS-based analysis in this study and the previous studies ( Supplementary Table S7 ; Hirano et al., 2015; Kazama et al., 2017; Hase et al., 2020; Sanjaya et al., 2021) and those of essential genes ( Figure 4 ). Almost all of the breakpoints did not overlap with the essential genes, except eight breakpoints: five were heterozygous deletions or chromosomal rearrangements with breakpoints overlapping the genes AT1G63160, AT4G03430, and AT4G27600, two were homozygous translocations with breakpoints in the 5′-UTR region of AT2G28880 gene, and the other one was a homozygous translocation whose breakpoint was in the coding region of AT4G27600 (NARA5 gene) which is essential for autotrophic photosynthetic growth (Ogawa et al., 2009). We supposed that the distribution of essential genes gave a bias to inheritance of chromosomal rearrangement. Indeed, the numbers of breakpoints and essential genes in every 1-Mb window of the Arabidopsis genome were found to be negatively correlated ( Supplementary Table S10 ) (correlation coefficient r = -0.32, p = 0.00036).

In the current study, we detected large-scale mutations in the Arabidopsis plants after heavy-ion irradiations. We performed NGS-based mutation analysis on the mutants that possessing heterozygously inherited deletions that were detected by array CGH analysis and were relatively larger than others. Some deletions were only detected through NGS-based analysis ( Supplementary Table S9 ). As the deletions were small (< 1 kbp), it is likely that they were not detected by the array CGH analysis, which can only detect deletions that overlap with probe regions. The breakpoints of chromosomal rearrangements, such as inversions, can also be detected through NGS-based analysis. Others were detected only by array CGH analysis, possibly due to resequencing mixed samples of 10-15 M₃ plants derived from the same M₂ plants, resulting in the deselection of infrequent mutations. Consequently, the combination of array CGH and NGS-based analyses has enabled comprehensive mutation detection.

We also performed metabolic profiling on 90 M₃ lines possessing deletions and suggested the influence of deletions on metabolome ( Supplementary Table S6 ). A heatmap of hierarchical clustering ( Supplementary Figure S2 ) showed that several metabolites were upregulated or downregulated specifically in some lines and clustered in the heatmap. For example, in the Ar50-44-as1 line that showed an abnormal shape of leaves ( Supplementary Figure S1 ; mutant ID 15 in Supplementary Figure S2 ), a metabolite cluster containing three metabolites (metabolite IDs 1128, 200005, and 200009) was upregulated. We set relevant keywords corresponding to these metabolites and searched the pathway map of A. thaliana in the KEGG PATHWAY database (Kanehisa et al., 2021) for pathways in which these metabolites are involved. These metabolites were commonly involved in the flavonoid biosynthesis pathway (ath00941) ( Supplementary Table S11 ). In the Ar50-44-as1 line, a deletion in chromosome 2 was also detected by microarray probes located from 8058304 bp to 8099777 bp. The size of this deletion could be expanded to the positions of adjacent probes leading to a maximum size possible (chr2:7920277-8245220) in which 92 genes were included ( Supplementary Table S12 ). We also searched the pathway map of A. thaliana in the KEGG PATHWAY database for pathways in which these genes are involved. The gene coding a peroxidase superfamily protein (AT2G18980) was involved in the phenylpropanoid biosynthesis (ath00940) that is connected to the flavonoid biosynthesis pathway (ath00941) via metabolites cinnamoyl-CoA and p-coumaroyl-CoA. It can be considered that the upregulation of the metabolite cluster was presumably due to the deletion of AT2G18980 gene though it is uncertain whether this metabolic change affected the phenotype. A series of combinational information of deletions and metabolite intensities of the mutant lines would help functional analysis of the deleted TAG in further studies.

In the current study, we detected multiple deletions in the Arabidopsis plants after irradiations with LET of 100, 200, and 290 keV μm^-1. In terms of TAG disruption, heavy-ion-beam irradiation with LETs of ≥100 keV µm^-1 was enough to disrupt TAGs of about two or three genes without affecting the essential genes. In addition to the frequency and size of deletions, the deletions and rearrangements detected in this study provided information for inheritable rearrangements that did not disrupt any essential genes. This could be a beneficial view in inducing deletions in the plant genome. Recently, mutation analysis has been effectively conducted by using whole genome sequencing and exome sequencing in several plant species (Li et al., 2017; Ichida et al., 2019; Li et al., 2019; Hase et al., 2020; Liu et al., 2023; Yang et al., 2019; Oono et al., 2020; Zheng et al., 2020, 2021; Sun et al., 2022; Ren et al., 2023; Wen et al., 2023; Xiong et al., 2023). By using whole-genome mutation analysis, we previously found a significant difference in mutation spectrum between 30 keV μm^-1 and 290 keV μm^-1 heavy-ion-beam irradiations (Hirano et al., 2015; Kazama et al., 2017; Morita et al., 2021). It was suggested that heavy-ion beams with higher values of LET tend to induce larger sizes of mutations. In this study, however, the lengths of deletions were not significantly different by LET in a range of 100-290 keV μm^-1 ( Figure 1A ), presumably affected by the distribution of essential genes. These results imply that the scale of induced mutation increases with the increasing value of LET but that the upper limit of the scale is subject to the distribution of essential genes. Indeed, the einkorn wheat mutants induced by heavy-ion beams showed deletions over 20 kbp or more, even those induced by carbon-ion irradiation with an LET of 50 keV µm^-1 (Shitsukawa et al., 2007; Nishiura et al., 2018; Hashimoto et al., 2021). In the 570 Mbp long Silene latifolia Y chromosome, C-ion irradiation with an LET of 30 keV µm^-1 induced large deletions covering more than a quarter of the Y chromosome (Kazama et al., 2016; Krasovec et al., 2019). These findings suggest that such large deletions can be induced by C-ion irradiation with the LETs of 30-50 keV µm^-1 in the species with long distances between essential genes. These findings were consistent with a classical image that any radiations can induce large deletions effectively. However, upper limits of heritable deletion sizes are largely dependent on the distribution of essential genes and only deletions that do not disrupt the essential genes can be inherited homozygously. Conversely, the ability of the heavy-ion beams with high LETs to induce large deletions with high efficiency in the Arabidopsis genome, in which essential genes are densely distributed, may be due to the ion beam causing localized damage on chromosomes.

We investigated the overlap between the essential genes and the 55 heterozygous deletions and revealed that 21 heterozygous deletions overlapped with essential genes ( Figure 3 ). In the remaining 34 heterozygous deletions, 32 deletions included 471 genes. It is possible that some of these genes are essential ( Supplementary Table S13 ), or that those deletions are linked with small mutations that disrupt the function of essential genes. It is also possible that a deletion including all genes in some TAG leads to lethality as if the TAG behaves as an essential gene. We compared the theoretical maximum sizes of homozygous deletions and the sizes of observed homozygous or heterozygous deletions ( Figure 2 ). This result not only strongly suggests that the mean size of heterozygous deletions also seemed to be influenced by the distribution of essential genes though not as much as homozygous deletions, but also implies that the essential genes of Arabidopsis might not be fully discovered yet. Moreover, deletions of some genes involved in gametogenesis are critical even if they are heterozygous, which would be a factor for the upper limit of sizes of both homozygous and heterozygous deletions. Naito et al. (2005) investigated the transmissibility of various sizes of deletions including GL1 gene induced in pollen cells (M₁) and proposed that compared to the small deletions (1-4 bp), large deletions (> 6 Mbp) are not transmitted to progeny (M₂) because the possibility that genes involved in gamete development or viability are overlapped with increasing size of deletions. Kitamura et al. (2022a) compared mutation spectrum between M₁ and M₂ generation after gamma-ray irradiation and revealed that chromosomal rearrangements tend not to be inherited to the next generation. Observations of the heritability of deletions in our study corroborates this hypothesis.

Our result implied that the locations of the essential genes are also influential on the locations of breakpoints ( Figure 4 ; Supplementary Table S10 ). As one of the exceptions of the relationship, we found a breakpoint in the coding region of NARA5 which is essential for autotrophic photosynthetic growth (Ogawa et al., 2009). The translocation was detected in the Ar-57-al1 mutant line that was grown on the medium containing sucrose from germination to young plantlet (Hirano et al., 2015). Therefore, Ar-57-al1 plants would be able to overcome the growing stage when NARA5 is essential. Except for this, no homozygous breakpoint was located in the coding region of essential genes. The irradiation experiments on the pollen grains of Cyrtanthus mackenii showed that C-ion irradiation with an LET of 22.5 keV µm^-1 is sufficient to induce rearrangements in the chromosomes of the male gametophyte and produces more pronounced biological effects on chromosomal bridge induction than Ar-ions with an LET of 280 keV µm^-1 during the first cell division (Hirano et al., 2013, 2021). Since almost all chromosomal bridges cannot be transmitted to the next generation, the C-ion irradiation with an LET of 22.5 keV µm^-1 cannot effectively produce inheritable rearrangements. On the other hand, inheritable chromosomal rearrangements were frequently produced with higher LET radiations (Kazama et al., 2017; Hirano et al., 2022). These results suggest that heavy-ion beams with higher LETs can effectively induce inheritable rearrangements because of localized damage on chromosomes.

Different mutation frequencies after heavy-ion irradiation were also observed between chromosomal locations (Kitamura et al., 2022b). In the current study, the percentage of intra-chromosomal rearrangements including large (≥ 100 bp) deletions in all chromosomal rearrangements was 72%. Ex-TAQing system that can induce chromosomal rearrangements randomly using restriction enzymes has been developed (Tanaka et al., 2020). The percentage of intra-chromosomal rearrangements in all chromosomal rearrangements induced by Ex-TAQing that were detected in the two diploid mutants was 82% (18 out of 22). Both heavy-ion beam and Ex-TAQing tend to induce more intra-chromosomal rearrangements than inter-chromosomal rearrangements (not significantly different by Z-test), which is probably because the heritability of inter-chromosomal rearrangement is less than that of intra-chromosomal rearrangement through cell division.

This study led to the conclusion that the distribution of essential genes affects the heritability of large deletions and chromosomal rearrangements. To create large deletions that are effective in destroying TAGs without deleting essential genes, more dense ionization, for example caused by high-LET heavy-ion beams, would be effective. Although less data of the effects of high-LET irradiations on phenotypic spectrum has been provided than low-LET irradiations or chemical mutagens, the effectiveness of the high-LET irradiations on the induction of large deletions and rearrangements may be a beneficial source for creating a new variety of mutants.