The resistance of mammalian cells to DNA damages is influenced by their microenvironment. For instance, irradiated cancer cells surrounded by unirradiated neighbours sustain less DNA damage than unaccompanied cells exposed to the same dose of radiation. Meanwhile, the unirradiated bystanders in such co-cultures are known to develop genotoxicity under the influence of their irradiated neighbours. These Janus-like phenomena, known respectively as Radiation-Induced Rescue Effect (RIRE) and RadiationInduced Bystander Effect (RIBE) have profound implications on the outcome of cancer radiation therapy. Most previous studies in this field of research directly compared the level of DNA damages in irradiated cells with and without co-cultured bystanders. In these experimental models, RIRE and RIBE occurred simultaneously and may influence one another in a complicated entwining way. It is impossible to unequivocally establish whether RIRE occurs whenever irradiated cells are surrounded by any cells, or whether the stimulation of these bystander cells by signals from their irradiated neighbours, i.e., the occurrence of RIBE, is required for the generation of RIRE. As both irradiated cells and unirradiated cells released a myriad of factors into the shared medium, it was also difficult to characterise the functions of individual factors.

In this PhD project, I addressed these limitations by establishing a new experimental paradigm for studying RIRE. In this setup, X-ray- or UV-irradiated cells, termed “originators,” were first co-cultured with bystanders to induce RIBE on the latter. The stimulated bystanders were then cultured in isolation so that soluble factors released from these cells were collected in the conditioned medium. This medium was transferred to a new population of irradiated cells, termed “effectors,” where the impact of these factors on radiation damages was assessed. Effectors cultured in fresh medium and irradiated at the same dose were used to provide a baseline of radiation damages in the absence of RIRE. This experimental design separates the players of RIBE and RIRE and allows the independent assessment of the significance of each effect. Furthermore, the effectors were assessed at multiple cytotoxic endpoints, such as DNA lesions as determined by γH2AX and 53BP1 immunofluorescent staining, and apoptotic cell death cascade as determined by the levels of BAX and cytochrome C, the activities of caspases 3 and 7, and the cleavage of PARP1. This provides a holistic view on how different steps of the DNA repair and apoptotic cascade were modulated by RIRE.

Using these experimental designs, I examined the RIRE induced by and exerted on a wide range of human cancer cell-lines, including HeLa (cervical adenocarcinoma), MCF7 (mammary gland adenocarcinoma), HCT116 (colorectal carcinoma) and CNE-2 (nasopharyngeal carcinoma). A significant reduction of γH2AX and 53BP1 staining was observed in all tested cell-lines treated with bystander-conditioned media, confirming the existence of RIRE. Based on this starting ground, I investigated the relationship between the size of bystander population on the strength of RIRE. Irradiated effectors were exposed to conditioned media from 4x10⁴ and 2x10⁶ bystander cells, respectively, while the number of both originator and effector cells were kept unchanged (8x10⁴ ). My data indicated that the fifty-fold increase in bystander cells resulted in a dramatic increase in rescue effect, as judged by DNA damages, and reduction of apoptosis, as determined by BAX, cytochrome C, caspase-3 and 7 and PARP1 cleavage. ELISA analyses showed that the concentration of Tumour Necrosis Factor Alpha (TNF-α), a known transducer of RIRE, in the conditioned medium from 2x10⁶ bystander cells was 2.25-fold higher than that from 4x10⁴ bystander cells, suggesting that the observed effect of bystander population on RIRE may be mediated by this cytokine. Nuclear factor kappa-light-chain-enhancer of activated B (NF-κB) is one of the downstream factors of TNFα signalling, and previous studies have demonstrated an involvement of the NF-κB pathway in RIRE. As PARP1 has recently been identified as a co-activator of NF-κB-mediated DNA repair, I investigated the impact of RIRE on PARP1 in the effector cells. In consistent with the established role of PARP1 in DNA repair, X-ray-irradiation resulted in a significant upregulation of PARP1 in all tested cell-lines. The radiation-induced increase in PARP1 expression was suppressed by RIRE. This may be a reflection of the RIRE-associated reduction of DNA damages instead of compromised DNA repair activities, as irradiated effector cells treated with bystander conditioned medium exhibited a lower level of cell death despite a reduced PARP1 expression. As expected, the pharmacological inhibition of PARP1 functions in RIREstimulated effector cells led to a dramatic increase in DNA damages. Interestingly, in RIRE-stimulated effector cells, the inhibition of NF-κB functions downregulated PARP1 expression, and the inhibition of PARP1 functions downregulated NF-κB expression, and both inhibitors could abolish RIRE. These data confirmed the previous observation that NF-κB and PARP1 regulated each other in a positive feedback mechanism, and suggested that the TNF-α/NF-κB/PARP1 axis provided a rapid and sensitive adjustment of DNA repair responses in RIRE.

Next, I investigated the influence of bystander cell types on RIRE. In a tumour microenvironment, cancer cells intermingle with a variety of stromal and immune cells. Although sporadic studies have reported the use of non-cancer cells, such as normal fibroblasts, as bystander cells, the role of immune cells in RIRE remained unexplored. Here, I investigated the effect of macrophages on RIRE. To add an extra level of clarity, “naïve” bystanders co-cultured with mock-irradiated originators were used as a control. I observed that the conditioned medium of these naïve bystander cells generally provided a lower level of rescue on DNA damages in UV-treated effector cells than that of bystander cells pre-treated with irradiated cells. This indicates that at least a portion of the observed rescue effect required factors from irradiated originators and is possibly a specific consequence of RIBE. I named this effect “active RIRE.” My data show that the magnitude of active RIRE, relative to the total RIRE, was highly cell type-dependent. As bystander cells, MCF7 showed a pronounced active RIRE, whereas THP1-derived macrophages showed a strong overall RIRE but insignificant active RIRE. Interestingly, RIRE induced by macrophages could further be modulated by polarisation. RIRE from macrophages was abolished by M1 polarisation, while M2 macrophages and Tumour Associated Macrophages (TAM) demonstrated strong active RIRE. Furthermore, when a mixture of MCF7 cells and polarised macrophages were used as bystander cells, the magnitude of the resulting RIRE was dictated by the phenotype of the macrophage used: RIRE was suppressed by M1 macrophages but significantly enhanced by M2 and TAM. Hence, the RIRE of polarised macrophages in a mixed bystander population appeared to subsume that of other cell types. This study demonstrated a previously unappreciated role of the innate immune system in RIRE. Depending on their polarised phenotypes, macrophages in the tumour microenvironment can interfere with the effectiveness of radiotherapy by adjusting the magnitude of RIRE.

Taken together, results from this project unveil several novel and important aspects of RIRE. The observation of “active RIRE” suggests that signals from irradiated cells are required by the bystander cells to generate at least a portion of RIRE. By inducing protection towards surrounding cells against further damages, RIBE can be viewed as an adaptive response instead of just a collateral damage caused by distressed neighbours. This project also establishes that the strength of RIRE depends on the number of bystander cells, which is proportional to the paracrine concentration of TNF-α, a cytokine that triggers the build-up of NF-κB and PARP1 levels in positive feedback loop. There appears to be a window of optimal NF-κB and PARP1 concentrations for the strongest RIRE, as both the chemical inhibition of these factors and the pro-inflammatory stimulation from M1 macrophages could abolish RIRE. My data further show that the balance between pro- and anti-inflammatory forces within the tumour microenvironment, maintained at least in part by macrophages, plays a determining role in the magnitude of RIRE. By the therapeutic manipulation of these forces, e.g., by the control of M1 and M2 macrophage populations or by the pharmacological adjustment of NF-κB/PARP1 levels, it may be possible to attune the outcome of radiation on the life and death of cancer cells.