Exposure to galactic cosmic radiation compromises DNA repair and increases the potential for oncogenic chromosomal rearrangement in bronchial epithelial cells

Participants in deep space missions face protracted exposure to galactic cosmic radiation (GCR). In this setting, lung cancer is a significant component of the overall risk of radiation-exposure induced death. Here we investigate persistent effects of GCR exposure on DNA repair capacity in lung-derived epithelial cells, using an enzyme-stimulated chromosomal rearrangement as an endpoint. Replicate cell cultures were irradiated with energetic 48Ti ions (a GCR component) or reference γ-rays. After a six-day recovery, they were challenged by expression of a Cas9/sgRNA pair that creates double-strand breaks simultaneously in the EML4 and ALK loci, misjoining of which creates an EML4-ALK fusion oncogene. Misjoining was significantly elevated in 48Ti-irradiated populations, relative to the baseline rate in mock-irradiated controls. The effect was not seen in γ-ray irradiated populations exposed to equal or higher radiation doses. Sequence analysis of the EML4-ALK joints from 48Ti-irradiated cultures showed that they were far more likely to contain deletions, sometimes flanked by short microhomologies, than equivalent samples from mock-irradiated cultures, consistent with a shift toward error-prone alternative nonhomologous end joining repair. Results suggest a potential mechanism by which a persistent physiological effect of GCR exposure may increase lung cancer risk.


Results
Assay for EML4-ALK rearrangement. Choi and Myerson described an approach using Cas9/sgRNA technology for induction of oncogenic rearrangements affecting endogenous genes 36 . A pair of Cas9/sgRNA nucleases are used to make separate incisions in the linked, endogenous ALK and EML4 loci. Misrepair, with inversion of an approximately 12 Mbp intervening sequence, creates an EML4-ALK fusion (Fig. 1A). The fusion protein is expressed at high levels and can be detected by immunofluorescence or flow cytometry using anti-ALK antibodies. In addition, the novel EML4-ALK junction can be detected by PCR using primers flanking the novel junctions. The original studies were performed in transformed, SV40 T-antigen expressing, HEK-293T cells. Fig. S1 shows replication of these results. Consistent with the original report, EML4-ALK joining was dependent on expression of both Cas9 and sgRNAs. In the HEK-293T model, the inversion event occurred with a frequency of up to 10.5% (Fig. S1).
For the main study of radiation effects on DSB repair in human bronchial epithelial cells, we adopted a sensitive and specific assay for qPCR detection of the novel EML4-ALK junctions using Taqman methodology. In a reconstruction experiment, the Taqman assay proved to be linear across a three order of magnitude range in input DNA concentration (Fig. 1B). Irradiation experiments used a human bronchial epithelial cell line, HBEC3-KT F25F, that was immortalized and transformed to anchorage-independent growth, but remains non-tumorigenic, as a model for precancerous lung cells. Replicate cultures were exposed to energetic 48 Ti ions, allowed to recover for 7 d, and challenged by expression of the Cas9/sgRNA pair. After a further 80 h, genomic DNA was harvested for analysis.
Representative raw data from Taqman qPCR analyses of a single-copy RNase P gene standard and of EML4-ALK junctions are shown in Fig. 1D,E, respectively. For the single-copy standard, amplification curves were tightly clustered among replicates, reflecting good technical reproducibility for DNA isolation and amplification. For the EML4-ALK junctions, amplification curves showed a low, but detectible, occurrence in mock-irradiated cultures and a higher, albeit variable, frequency in post-irradiated cultures. The variability was typical and perhaps reflects the stochastic nature of non-targeted effects (see Discussion).
Comparison of the number of cycles required to reach the same normalized fluorescence reporter values (ΔRn) indicates that the EML4-ALK junctions require 6-12 additional cycles of amplification compared to the single-copy standard. Assuming equal efficiencies for the different primer-probe sets, the frequency of Scientific REPORTs | (2018) 8:11038 | DOI:10.1038/s41598-018-29350-5 EML4-ALK fusion alleles in these cell populations thus ranged from about 0.025% (1/2 12 ) in the mock-irradiated groups to 1.5% (1/2 6 ) in the most responsive irradiated group.
Quantitative analysis of dose and radiation quality effects. We performed a large-scale experiment to explore the frequency of misrepair in cell populations with different irradiation histories. The inhibitory effect of HZE particle radiation on c-NHEJ is thought to reflect levels of complex DNA damage products 32 , which are in turn a function of radiation dose and LET values, with higher LETs resulting in more complex damage at a given dose. We thus performed irradiations at three 48 Ti doses (0.2, 0.5, and 1.0 Gy). By modulating the 48 Ti beam energy, we were also able to perform irradiations at two LET values (108 and 200 keV/μm). Particle fluences and estimated mean hits per cell nucleus are shown in Table S1. Cells were allowed to recover for 7 d and challenged by expression of Cas9/sgRNAs. After 80 h, DNA was extracted and analysed as in Fig. 1. At these doses and energies, there was little or no effect of irradiation on cell viability at 48 h post-irradiation ( Fig. S2A and S2B). Moreover, cell counts at the time of sub-culturing indicated that populations had undergone approximately one doubling (Fig. S2C), suggesting that the majority of the cells in the challenged population were progeny of the directly irradiated cells.
Results showed a marked increase in the frequency of Cas9/sgRNA-stimulated inversions with increasing radiation dose ( Fig. 2A). As there was no significant difference in results, values obtained with the 108 and 200 keV/μm ion beams at each dose were pooled for statistical analysis. The increase in inversion frequency in the 1.0 Gy group, relative to the mock-irradiated control group, was slightly more than 3 ΔCt units, corresponding to a difference in allele frequency of 9-fold, and was statistically significant at the p < 0.05 level. The 0.5 Gy group showed a trend toward increased in frequency of rearranged alleles, although the differences with the mock-irradiated group was not statistical significant. We also investigated the effect of reference 137 Cs γ-rays, a form of low-LET radiation (0.67 keV/μm). There was no effect at doses of up to 3.0 Gy (Fig. 2B).  We performed an independent, smaller-scale experiment during a separate beam line campaign. Results are shown in Fig. S3. Here, duplicate cultures were irradiated at two 48 Ti doses (0.3 and 1.0 Gy) at a single LET value (108 keV/μm). There was a significant increase in Cas9/sgRNA-stimulated inversion frequency at the 1.0 Gy dose, relative to mock-irradiated control cultures (Fig. S3). The ability to replicate the key finding in two beam line campaigns, approximately six months apart, gives confidence in the robustness of the results.

Misrepair junction characterization. To gain further insight into the underlying mechanisms associated
with DSB misjoining, we analysed the DNA sequences of the novel EML4-ALK junctions. Genomic DNA from irradiated or mock-irradiated control cell populations was extracted, and the junctional sequences were amplified by nested PCR. The mixed reaction products were subjected to Sanger sequencing and analysed at the population level using the Tracking of Indels by Decomposition (TIDE) method 39 . Examples of electropherograms representing an exact EML4-ALK joint, versus the mixed products obtained by PCR amplification of a genomic DNA sample, are shown in the top and bottom panels of Fig. 3A, respectively. For the mixed products, the sequence is unambiguous in the primer-proximal region (left side) but becomes ambiguous 5 nt prior to the junction site, indicating the presence heterogeneous insertions or deletions in the genomic DNA population.
The TIDE algorithm analyses the electropherograms to identify the major indel species and estimate their frequency. In the mock-irradiated cultures, >90% of Cas9/sgRNA-stimulated EML4-ALK joints arose from exact joining of the blunt end cleavage products. By contrast, in the irradiated cultures, a significant fraction of the junctions harboured deletions (Fig. 3B). Two recurrent deletions (∆12 and ∆20) dominated the indel spectrum for cells exposed to 48 Ti ions at an LET of 108 keV/μm, and another two (∆15 and ∆17) dominated the spectrum for cells exposed to 48   by a resection-dependent repair mechanism. Figure 4B shows sequences of individual clones. Mutants ∆20 and ∆17 harboured unidirectional deletions extending into the ALK gene (Fig. 4B). Mutants ∆12 and ∆15 harboured bidirectional deletions extending into both the EML4 and ALK genes. The bidirectional deletions were flanked in both cases by 1-2 bp of microhomology (denoted by boxes), a hallmark of resection-mediated joining. The figure also illustrates the proposed mechanism of deletion formation, which involves limited resection to expose short 3′ tails, apposition or hybridization of DNA ends, and trimming of 3′ flap sequences. We noted that some deletions were present exclusively in the 108 keV/μm populations and others exclusively in the 200 keV/μm populations. At this time, we do not have a hypothesis to explain why the pattern of recurrent deletions should be LET dependent, and we cannot rule out the possibility that it occurred by chance.
Some of the sequenced regions (11/86) contained point mutations, primarily AT > GC transitions. There was no significant association between the frequency of point mutations and the treatment group or the concurrent presence of deletions. There was also no significant enrichment proximal to the sequence junctions. Flanking point mutations are not typical products of Cas9/sgRNA mutagenesis, and we cannot exclude the possibility that the ones seen here are PCR artefacts.

Discussion
Here we show that a prior history of energetic heavy ion exposure compromised the fidelity of DSB repair in a human lung-derived cell line. Specifically, the frequency of incorrect joining of two enzymatically-induced DSBs was significantly higher in cell populations that were previously exposed to 48 Ti ions, a GCR component. In addition, sequencing of the novel junctions arising in the post-irradiated cells revealed the presence of recurrent, 12-20 nt deletions. In the most severely affected population (1.0 Gy, 200 keV/μm), almost 70% of the novel joints harboured deletions, far more than the 5% in the mock-irradiated control population.
Two different repair pathways, c-NHEJ and alt-NHEJ, contribute to rejoining of radiation-induced DSBs in mammalian cells. Although alt-NHEJ was initially considered to be the main source of misjoining to create chromosomal rearrangements, more recent work has shown that c-NHEJ can also promote DNA misjoining, particularly under conditions of ATM kinase deficiency 40,41 . However, even in the setting of misjoining, c-NHEJ typically generates indels of only a few bp at sites of rejoining, whereas alt-NHEJ generates larger deletions. Although  Fig. 2A. Red bars denote recurrent deletions; grey bars denote exact joining. Only peaks scored as significant are shown. Contribution of each species to the total is indicated (%). Percentages sum to less than 100 because nonsignificant peaks are not scored or shown.
further studies with mutant cells will be needed to be certain, the prevalence of >10 nt deletions in the irradiated samples is suggestive of alt-NHEJ than c-NHEJ mediated joining.
Alt-NHEJ was initially described as a backup pathway in cells or cell extracts that were deficient for c-NHEJ, which is the normal default pathway for end-joining in mammalian cells 42,43 . Several factors influence the regulation of alt-NHEJ. One of these is position in the cell cycle, with alt-NHEJ up-regulated in G2 phase 44 . We did not measure cell cycle distribution in the present study, although in a previous study using a different cell line we found only a slight overrepresentation of G2/M cells at 7d post-irradiation (18.00 ± 1.39% for 1.0 Gy vs. 15.53 ± 0.23% for non-irradiated controls) 33 , which would be too small to account for observed effects on repair pathway utilization. Interestingly, low-LET radiation stimulates alt-NHEJ in a transient, plasmid-based assay by stimulating phosphorylation of XRCC1 45 . Our attempts to detect changes in XRCC1 phosphorylation at 6 d post-irradiation have been unsuccessful. We note also a report that although XRCC1 reinforces the role of DNA ligase III in chromosomal translocations mediated by alt-NHEJ, it is not essential 46 .
An alternative explanation for an increased prevalence of alt-NHEJ-mediated joining is inhibition of c-NHEJ. A recent study showed that induction of even a small number of DNA fragments significantly affects repair pathway utilization 14 . An earlier study suggested that small DNA fragments, generated as direct products of high-LET radiation, bind to and inhibit the function of Ku protein 32 . If these were sufficiently persistent, they could influence results of a challenge initiated at 7 d post-irradiation. We are intrigued by a recent report that exposure to high-LET radiation generates small DNA fragments by a secondary mechanism involving resection by the MRE11 repair nuclease 47 . These engage the cytoplasmic cGAS receptor and trigger a STING-dependent pro-inflammatory response. We note that Ku protein, a key factor in the c-NHEJ pathway, is exquisitely sensitive No individual clone had more than one point mutation, and no mutation was recovered more than once in a given deletion context. to oxidative stress [48][49][50] . Inflammation and consequent changes in cellular redox status could thus provide an additional molecular basis for inhibition of c-NHEJ in post-irradiated cells.
In prior work using reporter lines, we demonstrated that co-culturing with HZE particle-irradiated cells increases misjoining of I-SceI induced DSBs in non-irradiated bystander cells 35 . Attempts to extend this finding to the epithelial cell model used here were unsuccessful. It could be that the increased use of error-prone pathways is a cell-autonomous effect in the HBEC3-KT F25F model, or alternatively that the magnitude of the effect in bystander cells was below the threshold for detection.
We noted considerable variability in the response of replicate cultures to the same doses of HZE particle radiation. While we cannot rule out the possibility of technical issues, the levels of the RNase P internal standard detected by qPCR were highly reproducible, and it was only the levels of EML4-ALK joints that were variable. We speculate that the variability might be biological in origin. One hypothesis is that there is a small, and thus variable, number of cells in each flask that drive the observed phenotype. Perhaps, these correspond to cells with persistent unrepaired damage. The presence of low-abundance driver cells is a subject of ongoing interest.
In space, organisms are subjected to a mixed GCR field. The fluence of low-LET 1 H or 4 He ions is far greater than that of heavier species 5 . There is considerable interest in how characteristic damages inflicted by different ion species interact to modify overall risk. The design of the present experiments, where cells were exposed to heavy ion irradiation to create complex DSBs, then later challenged by induction of simple DSBs, emulates a situation where traversal of a cell by a high-LET particle is followed by one or more encounters with low-LET ions, which create simple DSBs distributed throughout the genome. The observation that HZE particle exposure affects capacity for accurate repair of subsequent DSBs provides insight into how persistent physiological effects of heavy ion radiation might amplify cancer risk in the space environment.
An important caveat for beam line studies is that they are conducted using doses and dose rates that are higher than anticipated human exposures for an exploration-class mission 51 . It may be that different biological response mechanisms dominate in different dose regimes. The mechanisms considered here may not dominate or occur in space. The development of more sensitive biological models, and of ground-based facilities for low dose, mixed field, and protracted exposures to GCR components, remain as important research challenges.

Cell culture. HEK-293FT cells (Life Technologies, Carlsbad, CA) were grown in Dulbecco's Modified Eagle
Medium with 10% foetal bovine serum, 2 mM L-glutamine, 100 U/ml penicillin and 100 μg/ml streptomycin. HBEC3-KT F25F cells, which were derived from the parental HBEC3-KT line 38 by selection for anchorage-independent growth following exposure to 0.25 Gy of 56 Fe particle radiation, were grown in Keratinocyte-SFM medium supplemented with 50 µg/mL bovine pituitary extract and 5 ng/mL epidermal growth factor (Gibco) at 37 °C in 5% CO 2 .

Sequence analysis of novel EML4-ALK junctions.
Novel EML4-ALK fusion junctions were examined using a two-round nested PCR approach. Genomic DNA (1 µg) was amplified in the first round PCR using primers P9 d(CCTTCAGGCTACTCTTGTTAGTT) and P10 d(TCACTGATGGAGGAGGTCTT), in 25 μl reactions with LA Taq polymerase (Clontech) and Mg ++ -containing reaction buffer. Amplification was with cycling parameters of 94 °C for 1 min, then 27 cycles of 98 °C for 10 s, 55 °C for 1 min, 72 °C for 1 min, then 10 min at 72 °C. 1.5 µl of amplification product was add to the second round PCR using primers P11 d(TGACCATGCA-CAGGGAAATAA) and P12 d(GTTAGTCTGGTTCCTCCAAGAAG), in 25 μl reactions with LA Taq polymerase (Clontech) and Mg ++ -containing reaction buffer. Amplification was with cycling parameters of 94 °C for 1 min, then 30 cycles of 98 °C for 10 s, 52.5 °C for 1 min, 72 °C for 1 min, then 10 min at 72 °C.
Total PCR amplicons from each sample were sequenced using primer P11 d(TGACCATGCACAGG-GAAATAA). The raw trace files were then analysed using the TIDE (Tracking of Indels by Decomposition) web tool 39 . Individual EML4-ALK fusion junction sequences were detected by cloning single PCR amplicon into a pCR 2.1 vector, using an Original TA cloning Kit (Invitrogen).