Abstract
Intrinsic radiosensitivity is a major determinant of radiation response. Despite the extensive amount of radiobiological data available, variability among different studies makes it very difficult to produce high-quality radiosensitivity biomarkers or predictive models. Here, we characterize a panel of 27 human cell lines, including those derived from lung cancer, prostate cancer, and normal tissues. In addition, we used CRISPR-Cas9 to generate a panel of lines with known DNA repair defects. These cells were characterised by measuring a range of biological features, including the induction and repair of DNA double-strand breaks (DSBs), cell cycle distribution, ploidy, and clonogenic survival following X-ray irradiation. These results offer a robust dataset without inter-experimental variabilities for model development. In addition, we used these results to explore correlations between potential determinants of radiosensitivity. There was a wide variation in the intrinsic radiosensitivity of cell lines, with cell line Mean Inactivation Doses (MID) ranging from 1.3 to 3.4 Gy for cell lines, and as low as 0.65 Gy in Lig4-/- cells. Similar substantial variability was seen in the other parameters, including baseline DNA damage, plating efficiency, and ploidy. In the CRISPR-modified cell lines, residual DSBs were good predictors of cell survival (R2 = 0.78, p = 0.009), as were induced levels of DSBs (R2 = 0.61, p = 0.01). However, amongst the normal and cancerous cells, none of the measured parameters correlated strongly with MID (R2 < 0.45), and the only metrics with statistically significant associations are plating efficiency (R2 = 0.31, p = 0.01) and percentage of cell in S phase (R2 = 0.37, p = 0.005). While these data provide a valuable dataset for the modelling of radiobiological responses, the differences in the predictive power of residual DSBs between CRISPR-modified and other subgroups suggest that genetic alterations in other pathways, such as proliferation and metabolism, may have a greater impact on cellular radiation response. These pathways are often neglected in response modelling and should be considered in the future.