Double-stranded breaks (DSBs) in genomic DNA are potentially fatal for all cells, and damage-induced repair mechanisms are necessary in order to ensure cell survival (1). There are at least two fundamentally different repair mechanisms: Homologous recombination (HR) and non-homologous end-joining (NHEJ) (1). The activation of cell-cycle checkpoints by DNA damage leads to mitotic arrest until the damage is repaired. Once the damage is repaired, cells will re-enter mitosis regardless of whether the repair process introduced errors in the DNA sequences (2,3). DNA damage that is too severe to be successfully repaired instead triggers programmed cell death (4,5). Thus, DNA damage and the fidelity of the repair process are critical in determining whether cells survive with a normal phenotype, or become malignant with an abberant one.

Cancer stem cells (CSCs) are the major cause of resistance to therapy, disease recurrence and metastasis to other organs (6–8). As with the stem cells that underlie homeostasis in normal tissues, such as skin, intestine and the haematopoietic system (9), CSCs may be either dormant (dCSCs) or actively dividing (aCSCs). In hepatocellular carcinoma, CD13⁺ dCSCs residing in hypoxic areas of the tumour in the liver margin are capable of surviving chemotherapy or radiotherapy (10). The incidence of DSBs in CD13⁺ dCSCs is low following chemotherapy, most likely due to the scavenging of reactive oxygen species (ROS) by CD13 (aminopeptidase N) (10). This is in contrast to CD13⁻CD90^+/− dCSCs, which undergo ROS-induced cell death following chemotherapy (11). Significantly, chemotherapy results in a shift in the CSC population from aCSCs to dCSCs, with dormancy acting as a refuge for the survival of malignant cells. CD13 is critical in suppressing ROS-dependent cell death during the epithelial-mesenchymal transition of CSCs during metastasis (12). Exposure to the CD13-specific inhibitor, ubenimex, results in considerable eradication of malignant cells in vivo, indicating the likely benefit of the combination of conventional chemotherapy with the CD13-specific inhibitor.

In this study, we show that DSBs in dCSCs result predominantly in the activation of DNA repair via the NHEJ pathway, whereas DNA repair by HR is the predominant response in aCSCs. Based on these data, we propose that dCSCs should be the major target of therapy in order to completely eradicate the malignant cell population, although hibernation therapy (the induction to dormancy) may be a suitable rationale dependent on the medical condition of patients.

In hepatocellular carcinoma, CD13 is expressed in CSCs (10) and is involved in the scavenging of ROS via the glutathione reductase pathway. CD90 has been proposed as a marker of stem cells, and its expression is correlated with tumourigenicity in mouse in vivo tumour models and the clinical outcome in human patients (13). In hepatocellular carcinoma, the CD13⁺CD90⁻ cell population is predominantly dormant, CD13⁺CD90⁺ cells are predominantly in the S phase and CD13⁻CD90⁺ cells are in the G2/M phase (10). Following exposure to genotoxic insults, such as chemotherapy or radiation therapy, the proportions of these cell populations shift; CD13⁻ cells become less common, whereas the dormant CD13⁺ population increases. We were interested in the mechanism of DNA repair in the various CSC populations; therefore, we used fluorescence-activated cell sorting (FACS) to separate dormant CSCs (CD13⁺) from non-dormant CSCs (CD13⁻) and analysed the expression of proteins involved in the various DNA repair pathways. Rad51, Rad54, Rpa, Smc6, Xrcc2 and Xrcc3 are involved in HR, whereas Ku80, Ku70 and Prkdc are indicative of NHEJ. We found that the expression ratio of NHEJ-associated proteins to HR-associated proteins increased in dormant CSCs, indicating that DSBs are repaired predominantly through NHEJ in these cells. DNA repair by NHEJ is also the dominant mechanism in quiescent hematopoietic stem cells (16). Conversely, in non-dormant CSCs, HR-associated repair proteins increased relative to NHEJ-associated proteins (Figs. 1–3).

In proliferating CSCs, either in whole tissues or tissue culture, exposure to genotoxic insults, such as chemotherapy and radiation therapy, causes DNA damage during the S phase of the cell cycle, followed by apoptosis of tumour cells (4). In a fraction of the actively dividing tumour cells that survive therapy, the DNA damage is repaired by HR, consistent with the data shown in this study. DNA repair by HR is typically a high-fidelity, low-error mechanism. By contrast, dormant cells are more likely to utilise NHEJ for DNA repair. NHEJ may occur at any stage of the cell cycle but has only a limited requirement for sequence homology and, importantly, is more prone to error than repair by HR (1). Thus, NHEJ is associated with the misrepair of DSBs, and may result in chromosomal deletions, insertions or translocations and subsequent genomic instability (14). These genomic alterations may lead to the inactivation of tumour suppressor genes or activation of oncogenes, which lead to the development of primary and metastatic tumours. In our study, we found that DNA repair by NHEJ is characteristic of dormant CSCs. Due to the mutagenic effects of NHEJ, dormant CSCs that survive chemoradiation therapy may acquire new mutations and become more likely to cause tumour recurrence and/or metastasis (15). It is therefore critical to eradicate dCSCs in order to improve patient survival.

Quiescence is widely accepted to be an essential protective mechanism for stem cells that minimises the endogenous stress caused by cellular respiration and DNA replication in normal cells (16). However, it is also important in tumour evolution after therapy, since quiescent cells that survive therapy are likely to have acquired mutations during DNA repair by NHEJ (8). Thus, genotoxic therapies, such as chemoradiation therapy, that suppress genotoxic effects are a double-edged sword, removing primary tumour bulk, but also contributing to tumour recurrence after therapy.