Quantitative and qualitative mutational impact of ionizing radiation on normal cells

Abstract

The comprehensive genomic impact of ionizing radiation (IR), a carcinogen, on healthy somatic cells remains unclear. Using large-scale whole-genome sequencing (WGS) of clones expanded from irradiated murine and human single cells, we revealed that IR induces a characteristic spectrum of short insertions or deletions (indels) and structural variations (SVs), including balanced inversions, translocations, composite SVs (deletion-insertion, deletion-inversion, and deletion-translocation composites), and complex genomic rearrangements (CGRs), including chromoplexy, chromothripsis, and SV by breakage-fusion-bridge cycles. Our findings suggest that 1 Gy IR exposure causes an average of 2.33 mutational events per Gb genome, comprising 2.15 indels, 0.17 SVs, and 0.01 CGRs, despite a high level of inter-cellular stochasticity. The mutational burden was dependent on total irradiation dose, regardless of dose rate or cell type. The findings were further validated in IR-induced secondary cancers and single cells without clonalization. Overall, our study highlights a comprehensive and clear picture of IR effects on normal mammalian genomes.

Keywords: carcinogen; clonalization; complex genomic rearrangement; ionizing radiation; irradiation dose; mutational signature; normal cell; single-base substitution; single-genome sequencing; structural variation.

Graphical abstract

Figure 1

Overview of the experimental design (A) Experimental design for detecting mutations induced by ionizing radiation (IR) in single cells using organoid culture technique. We collected cells irradiated in four different but complementary experimental settings, including (1) IR exposure on cultured cells (IR^vitro; top left), (2) IR exposure on living tissues (IR^vivo; bottom left), (3) post-radiotherapy (IR^post-RT; top right), and (4) IR-induced secondary cancers. The collected cells from the first three settings were clonally expanded into colonies using organoid techniques followed by WGS. IR-induced secondary malignancies were whole-genome sequenced without the clonalization step. For each experimental setting, we produced baseline (germline) genome sequences to sort out IR-associated somatic mutations from the colonies and cancer tissues. (B) Variant allele fraction (VAF) distribution of acquired single-base substitutions (SBSs) in non-irradiated (top) and irradiated (bottom) mouse colonies (pancreas). The black solid lines are Gaussian curves fitted to the distribution, and the gray solid line is a density curve. (C) Organoid-forming efficiencies of irradiated cells (bottom). Scale bar, 2.5 mm. Data are presented as mean ± SEM (the standard error of the mean) (top). n = 3. (D) Gene expression changes in cells exposed to 2 Gy irradiation for 24 h. Genes in the p53, cell-cycle, DNA replication, and double-strand breakage pair pathways are colored red, green, purple, and blue, respectively. See also Figures S1–S4 and S8.

Figure 2

Landscape of SBSs and indels (A) Summary of the acquired base substitution and indel mutations in the 135 colonies and 22 IR-induced secondary cancers. Stacked bar plots showing absolute number of SBS and indel mutations and the relative proportion of each SBS and indel signature. Full signatures delineated are shown in Figure S5. Annotation tracks include relevant information for each sample. The color codes in the figure legend are used for all subsequent figures. (B) Mutational spectrum of IR-associated indels delineated in mice (mID-A) (top). Additional mutational signatures (IR unrelated) are shown in Figure S5B. Expected mutational spectrum assuming random mutations is shown with y axis flipped (bottom). (C) Mutational spectrum of IR-associated indels delineated in humans (hID-A). Additional mutational signatures (IR unrelated) are shown in Figure S5B. Expected mutational spectrum assuming random mutations is shown with y axis flipped (bottom). (D) Pie chart showing differences between observed and expected indel spectra in mice (inner circles) and humans (outer circles). The expected indel spectrum (lighter color in each circle) was calculated based on the reference genome sequence. (E) Schematic representation of the formation of IR-induced DNA double-strand breaks (DSBs) and DNA repair processes for IRi-IDs. See also Figure S5.

Figure 3

Landscape of structural variations (SVs) (A) Summary of the acquired SVs in the 135 colonies. Each row indicates SV type, and each column represents the number of SVs observed in each sample. Colonies are classified into two groups: non-irradiated (n = 47) and irradiated (n = 88). SV types considered non-specific to the irradiated group are presented on the top. SV types specific to the irradiated organoids are presented on the bottom. Del-Inv, deletion-inversion composite; Del-Tra, deletion-translocation composite; Del-Ins, deletion-insertion composite. (B) Size distribution of microhomology in IR-induced SVs (IRi-SVs; purple). Compared to non-IRi-SVs (gray), breakpoints of IRi-SVs harbor shorter microhomology (p < 0.001, Kolmogorov-Smirnov test), which supports that non-homologous end joining is a predominant repair process for IR-induced DSBs. (C) Schematic representation of the consequences of IR-induced double DSBs in one chromosome (top) and in two chromosomes (bottom). (D) Schematic representation of the consequences of IR-induced triple DSBs in one chromosome (top) and in two chromosomes (bottom). (E) Frequency of Del-Ins and Del-Inv-and-Del-Tra events in non-irradiated organoids (n = 47), irradiated organoids (n = 88), primary lung cancer (n = 138; data from Lee et al.²⁹), and IR-induced secondary cancer (n = 22). ∗∗∗p < 0.001 for Del-Ins and ∗p < 0.05 for Del-Inv-and-Del-Tra, independent two-population proportions test. Data are presented as mean ± SEM. See also Figure S6.

Figure 4

Complex genomic rearrangements (A) Circos plots of four post-irradiated colonies exhibiting chromoplexy. (B) Schematic representation of the mechanism of chromoplexy observed in a human breast colony (hBR_50Gy_3). (C) Circos plot showing chromothripsis (red lines) on chromosome 15 in a 4 Gy irradiated organoid (PA_4Gy_7; top). The copy-number state in the catastrophic segment oscillates between three and one (bottom). (D) Schematic representation of a possible mechanism of chromothripsis observed in the 4 Gy irradiated organoid (PA_4Gy_7). (E) Circos plot showing chromothripsis (red lines) localized in chromosomes 12 and X found in a 2 Gy irradiated organoid (PA_2Gy_12; top) and patterns of rearrangements and copy-number states (bottom). (F) Schematic representation of the mechanism of chromothripsis observed in the 2 Gy irradiated organoid (PA_2Gy_12). (G) Circos plot showing a BFB-SV identified in an 8 Gy irradiated human organoid (FT_8Gy_2; left) and patterns of rearrangements and copy-number states (right). (H) Schematic representation of the mechanism of the BFB cycle observed in the 8 Gy irradiated human organoid (FT_8Gy_2). See also Figure S7.

Figure 5

Quantitative analysis of radiation-associated indels and SVs (A) Number of IR-induced indels per Gb with respect to radiation dose (Gy). Linear regression was used to estimate dose-response relationships. The data were divided into three subsets (experiments from IR^vitro, IR^vivo, and IR^post-RT). Boxplot shows median (midline; red), interquartile range (IQR) (box), and whiskers indicating minimum or maximum value within 1.5×IQR from the quartiles. (B) Number of IRi-SVs per Gb with respect to radiation dose (Gy). Linear regression was used to estimate dose-response relationships. The data were divided into three subsets as shown above. Boxplot whiskers extend up to 1.5×IQR from the quartiles. (C) Size distribution of breakpoint gap in IRi-SVs. IR-induced DSBs are usually accompanied by some nucleotide deletions. (D) Ratio of two means (μ_IRi-SV/μ_IRi-ID) with respect to irradiation type (IR^vitro, IR^vivo, and IR^post-RT). Marginal sums were compared between (1) IR^vitro vs. IR^vivo groups and (2) IR^vitro vs. IR^vivo and IR^post-RT groups using Fisher’s exact test. The data are shown as the ratio of means with an 83% confidence interval (CI) calculated by Fieller’s method; ∗p < 0.05, not significant (ns) > 0.1. (E) IRi-ID (middle), IRi-SV (bottom), and the ratio of the means (μ_IRi-SV/μ_IRi-ID; top) of in vivo, 8 Gy irradiated organoids with respect to tissue types. Only tissues with more than two data points were included. Fisher’s exact test on the sum of IRi-IDs and IRi-SVs shows a significant difference between the colon and liver; ∗p < 0.05. Error bars indicate 83% CI. (F) IRi-ID, IRi-SV, and the ratio (IRi-SV/IRi-ID) between high dose rate (one or four short exposures) and low dose rate (continuous exposure for 100 days) with a total irradiation of 8 Gy. The high rate involved irradiating a single 8 Gy or four 2 Gy doses, and the low rate involved approximately 0.08 Gy per day for 100 days; ns > 0.1. Boxplot whiskers extend up to 1.5×IQR from the quartiles. (G) Representative image of gamma-H2AX immunofluorescence staining (green) in 2 Gy irradiated mouse pancreas organoids following 3.2× tissue expansion using magnified analysis of proteome method (left). Number of gamma-H2AX foci per cell in control and 2 Gy irradiated samples was counted; ∗∗∗p < 0.001, two-sample t test. Boxplot whiskers extend up to 1.5×IQR from the quartiles. (H) Pie chart showing direct DSB count from the gamma-H2AX experiment (inner circle) and estimated DSB count from the sequencing data (outer circle). We estimated DSB counts from the sequencing data using the number of variants and their minimum number of DSBs necessary (for example, 1 DSB for an indel, 2 DSBs for an inversion, or 3 DSBs for composite SV). The number of variants was based on the coefficient of the linear model for the IR^vitro experiment (Figures 5A and 5B). The number of seamless indels was estimated from the proportion of the SV without a gap and the number of indels (Figure 5C). The counts were adjusted to account for inaccessible and repetitive genomic regions. (I) Enrichments of IRi-IDs and IRi-SVs in relation to genomic contexts. For an enrichment in the ATAC-seq (assay for transposase-accessible chromatin with sequencing) from the pancreas, SVs acquired from the pancreas colonies were used. Error bars indicate 95% CI. (J) Functional consequences of IRi-IDs (left) and IRi-SVs (right) detected from colonies. Error bars indicate 95% CI. (K) The difference of Myc expression in control and irradiated organoids (FT_8Gy_2). Data are presented as mean ± SEM. n = 3. TPM, transcripts per million.

Figure 6

Direct single-cell genome sequencing (A) Experimental design for detecting IR-induced mutations in single cells without clonal expansions, using WGS of the whole-genome-amplified single cells (MDA and PTA), Strand-seq, and duplex DNA sequencing (bottleneck sequencing and Concatenating Original Duplex for Error Correction) techniques. (B) SVs detected in the MDA experiments. (C) Mutational signature of IR-associated indels delineated from the PTA experiments. Full signatures are shown in Figure S9. (D) Number of indels observed in the PTA experiments, adjusted for allelic dropout rates. Boxplot whiskers extend up to 1.5×IQR from the quartiles. (E) Number of SVs observed in the PTA experiments, adjusted for allelic dropout rates. (F) Example of Strand-seq result of a 4 Gy irradiated single cell. (G) Number of SVs and sister-chromatid-exchange-like patterns observed in the Strand-seq experiments. See also Figures S9–S11.