Many cancers have defects in their DNA-repair pathways that can be targeted by agents and ionising radiation (IR) to control malignant growth. IR induces double-strand breaks in the DNA that cancer cells harbouring DNA-repair defects cannot resolve, leading to cell cycle arrest and cell death. Sensitivity to IR can be affected by several factors, such as differentiation, proliferation, tumour microenvironment and genetic alterations. Extensive studies on IR-mediated DNA damage and repair responses have gone some way to improve our understanding of the molecular mechanisms, however intra-tumoral heterogeneity has been identified as a potential complicating factor for efficacy. Potential biomarkers for IR sensitivity in the transcriptome have also been extensively analysed using next-generation sequencing (NGS). Many of these studies use RNA from the bulk of the tumour and only offer an average response of differential expression across large numbers of cells.
The emergence of single-cell transcriptomic technology has enabled the analysis of heterogeneous populations of tumour cells. In this study, the authors used barcoded Smart‑seq2 to analyse 3’ mRNA libraries from 334 single cells of the MDA‑MB‑231 breast cancer cell line assessing response to IR and the role of ataxia telangiectasia mutated (ATM).
Eighty-eight single cells were collected from the MDA‑MB‑231 breast cancer cell line with or without IR treatment and analysed by Smart‑seq. 6,937 genes were detected in the control group and 6,488 in the irradiated group.
There were 142 significant differentially expressed genes between the Control and IR treatments, with 71 upregulated and 71 downregulated in IR compared to control.
For the upregulated genes, GO term analysis demonstrated association with the G1/S phase of the cell cycle and ribosomes. Downregulated genes were enriched for antibiotic metabolism and daunorubicin/doxorubicin metabolic processes. These GO terms were confirmed using STRING analysis.
T‑SNE clustering analysis revealed four subpopulations of cells, one of which, cluster 3, appears to be mostly present in irradiated samples.
Markers for clusters 1 and 2 analysed for GO terms suggested they were involved in antibiotic metabolism and daunorubicin/doxorubicin metabolic processes and M-phase of the cell cycle. Cluster 4 had the highest number of genes associated with the Ribosome GO term.
Cell cycle analysis of cluster 3 demonstrated that most of the cells were in G1 or S phase. The induction of this population suggests that IR may block cell cycle progression. Further cell cycle analysis of clusters 1 and 3 demonstrated that a higher proportion of cells in cluster 1 were in G2/M after IR treatment and the authors conclude that cells in cluster 3 are derived from cluster 1 control cells.
In the differential expression clusters, cluster 1 had the fewest changes between control and IR treatment and could be a subpopulation of cells that are resistant to IR. Using a previously published set of 147 genes associated with radiation resistance, each cell was assigned a radiation resistance score. Cluster 1 had the highest resistance score, confirming the possibility that this is a resistant subpopulation and may survive the IR induced death.
Three candidate proteins were identified that potentially mediated radiation sensitivity. MCM3, MCM4 and SLBP. MCM3 and MCM4 are subunits of the pre-replication complex, and SLBP has been shown to bind to the stem-loop structure in replication-dependent histone mRNAs. Knock-down experiment on these proteins all resulted in significantly lower survival rates when treated with IR.
The authors performed a similar single-cell transcriptomic analysis with ATM knock-down in the MDA‑MB‑231 cell line. ATM‑KD gene expression was compared to control, with 162 genes upregulated and 248 downregulated. GO term analysis demonstrated enrichment of the Ribosome term in upregulated genes whilst downregulated genes had enrichment of antibiotic and daunorubicin/doxorubicin metabolic processes.
IR treatment of ATM‑KD cells resulted in far fewer differentially expressed genes being detected, suggesting that ATM knock-down severely attenuated the IR response.
T-SNE analysis of ATM‑KD and ATM‑KD IR treated resulted in only three clusters being detected, unlike the four detected using WT cells. Analysis of marker genes indicated that cluster 4 from the wild type irradiation experiment was not present in the ATM‑KD experiment. This cluster in WT cells was the one with the most differentially expressed genes and had the most robust response to IR.
In WT cells, cluster 3 cells were mostly due to IR treatment, but in the ATM‑KD experiment cells were present from ATM‑KD and ATM‑KD IR, suggesting that ATM KD may induce G1/S arrest.
Unlike WT cells, where most of the cells in cluster 3 were IR responders from cluster 1, in the ATM‑KD experiment, most of the cells in cluster 3 came from cluster 2. This suggests that IR responsive cells in cluster 1 were dependent upon ATM for G1/S cell cycle arrest.
In this study, the authors used single-cell transcriptomic analysis to determine the effects of ionising radiation in a breast cancer cell line. Despite cell lines being relatively homogenous populations of cells, the authors demonstrate subpopulations of cells that may be resistant to IR treatment. Understanding how heterogeneous populations of tumour cells in cancer patients react to IR treatment is critical, given its role as a major curative therapy. Biomarkers of radiation sensitivity could be invaluable. This study identified three genes, MCM3, MCM4 and SLBP for further investigation as biomarkers of radiation sensitivity.
Gao et al. A heterogeneous cellular response to ionizing radiation revealed by single-cell transcriptome sequencing. Am J Cancer Res. 2021; 11(2):513-529