Background
Head and neck cancers are a heterogeneous group of tumours arising in multiple anatomic sub sites in the head and neck region, which are histologically mainly squamous cell carcinomas (SCCs) [
1]. The major risk factor for the SCCs arising in the oral cavity, oropharynx, hypopharynx, and larynx is habitual exposure to tobacco and/or alcohol consumption, while human papillomavirus (HPV) infection is associated with oropharyngeal tumours [
2]. Different etiology of head and neck SCC also reflects in different tumour biology. Namely, HPV-positive oropharyngeal tumours and cell lines are more sensitive to radiotherapy and chemotherapy treatment than HPV-negative [
3‐
5]. Most commonly, a combination of surgery and (chemo) radiotherapy or upfront chemoradiotherapy is used for the treatment of non-metastatic head and neck SCC, as they result in a better effect than surgery or radiotherapy alone [
6].
Despite the advances made in the management of head and neck SCC, a high locoregional recurrence or second primary cancer remain a serious problem in the previously treated area [
7,
8]. Management of the recurrent or second primary SCC is often limited to palliative radiotherapy or systemic therapy, with a high risk of normal tissue toxicity, impaired quality of life and poor outcome [
7].
The observed radioresistant tumour phenotype is a combination of numerous factors: disease clinical features, e.g. tumour stage and volume, microenvironment features, e.g. hypoxia, and intrinsic cellular features, e.g. enhanced DNA repair, modulation of cell cycle progression, free radical and reactive oxygen species (ROS) scavenging [
9]. These intrinsic radioprotective mechanisms give a survival advantage to radioresistant cells, leading to treatment failure.
To improve therapeutic efficacy, it is crucial to understand and elucidate the underlying mechanisms of intrinsic cellular radioresistance. Isogenic cell lines with different radiosensitivity can be established from parental cell lines by repeated exposure to radiation and/or carcinogens [
10]. Generally, these isogenic pairs include parental cells, and the established cells with a more radioresistant phenotype, while isogenic cells with a more radiosensitive phenotype are rarely commercially available [
9,
11].
The aims of this study were to establish radioresistant isogenic cells from parental head and neck SCC cells, by repeated exposure to radiation, and to identify the differences in the irradiation response of the newly established radioresistant cells in comparison to isogenic parental and radiosensitive cells. Furthermore, the radio- and chemosensitivity, cell proliferation, cell cycle, induction and resolution of DNA double-strand breaks (DSBs) in parental, radioresistant and radiosensitive isogenic cells were evaluated. Additionally, the DNA damage signalling gene expression in these isogenic cells before and after irradiation, with a specific focus on the early response to irradiation, was elucidated.
Discussion
In order to modulate cellular radioresistance and sensitivity, mechanisms responsible for these phenomena need to be identified. Isogenic cell lines with different radiosensitivity can serve as a good predictive model, as survival changes can be directly attributed to specific modifications in the intrinsic cellular features. Pairs of isogenic cell lines with different radiosensitivity have been established from different solid tumour cell lines, such as head and neck SCC [
11,
19‐
24], prostate carcinoma [
25,
26], non-small cell lung cancer [
27,
28]. We successfully established radioresistant FaDu-RR cells from parental FaDu cells by repeated exposure to irradiation, while radiosensitive 2A3 cells were derived from parental FaDu cells by stable transfection with HPV16 oncogenes E6 and E7 [
12]. Although variations in radiosensitivity of HPV-positive cells exist, they are generally more radiosensitive than the HPV-negative cells [
3,
4]. Radiosensitivity of 2A3 cells was confirmed in vivo in experimentally induced tumours in immunodeficient mice [
5]. To the best of our knowledge, this is the only study comparing both radioresistant and radiosensitive isogenic cells to parental cells, with a specific focus on gene expression in response to irradiation.
Radioresistant FaDu-RR cells were established by repeated exposure to daily 2 Gy doses from parental FaDu cells, making them 1.6-fold more radioresistant than the parental cells. The observed potentiation of radioresistance in FaDu-RR cells is comparable to the potentiation of radioresistance observed in other isogenic models of radioresistance [
23,
28,
29]. The induced radioresistance was either long-term, more than 3 years of passaging [
20,
21], or short-term, from 4 weeks to 18 weeks [
22,
25,
28‐
30], however, it could be restored after additional irradiation [
30,
31]. FaDu-RR displayed long-term radioresistance of more than 4 years. Protocols used in these studies varied significantly in specific parameters, such as dose/fraction (from 0.5 Gy to 10 Gy), total dose received (from 10 Gy up to more than 1600 Gy), overall treatment time (from 5 days up to 6 years), recovery periods (12 h, daily, weekly, fortnightly intervals) [
20,
26,
32,
33]. On the other hand, the recovery period was not always defined in time units, but in the percentage of culture confluence, ranging from 50 to 80% prior to the next fraction [
24,
34]. These parameters were determined experimentally, based on the radiosensitivity of the parental cells, and allowing selection of radioresistant clones with replicative potential. In our preliminary study, 10 Gy/fraction with 48 h recovery between fractions was too high for the recovery of radioresistant cells with replicative potential. This was supported by the observation that dose per fraction ranging from 2 to 6 Gy, was sufficient for a selection of radioresistant clones, while lower or higher doses were not [
33]. However, others have used lower or higher doses/fraction from 0.5 to 10 Gy combined with shorter recovery period for low dose/fraction, and longer for high dose/fraction to successfully establish radioresistant cells [
20,
32]. The observed survival advantages were also inconsistently reported in terms of increased surviving fraction at a specific dose of radiation [
11,
19,
22,
23,
25,
29,
32], the dose modifying factor at 2 Gy irradiation dose [
28], clonogenic potential [
27], mean inactivation dose ratio [
28,
35], the dose required to reduce cell survival to 10% [
33], and α/β ratio of LQ model [
23]. Variations in parameters reporting the survival advantage of radioresistant cells also limit direct comparison of the extent of induced radioresistance. Significant increase in radioresistance can be achieved with either 10 Gy or 1600 Gy of the total dose [
26,
32], however, the potentiation of radioresistance cannot be directly compared due to variations in parameters measuring the survival advantages in these cells. In addition to variabilities in protocols used for selection of radioresistant cells, a successful selection of radioresistant cells is dependent also on the cell type [
9].
Prolonged exposure to irradiation can select the more radioresistant cells within the cell population, such as cells expressing markers of cancer stem cells (CSC), commonly associated with increased radioresistance [
19]. On the other hand, radiation induced genetic instability can further select the more radioresistant cells [
9]. During the establishment of radioresistant cells in our study, we have first selected different subclones. By further irradiation, we were not able to select different subclones, but rather whole cell populations. The increased radioresistance observed in the established FaDu-RR cells is a combination of clonal selection of radioresistant cells within the existing cell population and radiation-induced genetic variability as confirmed by differential gene expression.
Changes in the cell cycle distribution can confer different radiosensitivity of cells due to variations in radiosensitivity of cell cycle phases - cells in S-phase being the most radioresistant. We did not observe any difference in cell cycle distribution and cell proliferation rate in non-irradiated isogenic cells. Similarly, no change in the cell cycle distribution or growth rate was observed in other isogenic models of radioresistance [
11,
22,
23,
27,
33,
36,
37]. However, 5 h after 5 Gy irradiation, we observed a decrease in G
1-phase cells and an increase in S-phase and G
2/M-phase cells with no significant difference between the studied isogenic cells. Similarly, no change in cell cycle distribution was observed in radioresistant and parental cells 4 h after 2 Gy irradiation [
25]. While the percent of G
2/M-phase parental and radioresistant cells remained unchanged 24 h after irradiation, the percent of radiosensitive cells in G
2/M phase significantly increased. Other reports show a similar level of G
2/M arrest in response to irradiation in other radiosensitive cells [
3,
4,
24]. Increase in G
2/M-phase cells was observed both in radiosensitive and radioresistant cells, but the disturbance of the cell cycle in response to irradiation was longer in radiosensitive cells [
38]. Based on these results, other mechanisms than alterations in cell cycle regulation were involved in the radioresistant phenotype of FaDu-RR cells.
Alterations in radiation-induced apoptosis can also contribute to altered cellular radiosensitivity. In our study, no apoptosis was observed in non-irradiated isogenic cells and none of the differentially expressed genes observed in the non-irradiated radioresistant cells were associated with the apoptotic processes. On the other hand, the proapoptotic
BBC3 gene was under-expressed in 2A3 radiosensitive cells. The reason for this could be a transfection of radiosensitive cells by HPV protein E6 [
12], which is known to interact with pro-apoptotic proteins to prevent apoptosis in HPV-positive cells [
39,
40]. In response to 5 Gy irradiation, fold-induction of apoptosis was the highest in parental cells, and only non-significantly increased in the radioresistant cells 72 h after irradiation. In addition, proapoptotic
BBC3 was under-expressed in radioresistant cells in response to irradiation. This indicates the involvement of anti-apoptotic regulatory mechanisms in radioresistant cells. Variability in the induction of apoptosis in response to radiation exists likely due to different radiation dose, time after irradiation, and the method used to detect specific hallmarks of apoptosis. Similarly, in the study by Wei QC et al., no induction of apoptosis or apoptotic morphological features were observed up to 48 h after 6 Gy irradiation in radioresistant and parental cells [
28]. Contrary, 24 h after 10 Gy irradiation less apoptotic cells were detected in the radioresistant population [
27].
Cell survival after irradiation largely depends on the balance between DNA damage signalling, induction, and repair. We evaluated DNA damage by immunofluorescence staining of γH2AX foci, surrogate markers of DNA DSBs [
41]. We observed no significant difference in endogenous levels of γH2AX foci in the studied isogenic cells. However, the range of the observed γH2AX foci levels was the highest in the 2A3 radiosensitive cells. More endogenous DNA damage was expected in the cells due to E6-mediated oxidative stress [
42]. On the other hand, some studies reported radioresistant cells to have less endogenous DNA damage, more effective DNA repair, and enhanced levels of ROS scavengers [
19,
23‐
25,
27,
28]
.
After irradiation, the induction of γH2AX foci was the most prominent in radiosensitive cells, while in radioresistant cells the induction of foci was delayed. In addition, at the peak of γH2AX foci expression, radioresistant cells displayed the lowest number of γH2AX foci/nucleus in contrast to the highest number of γH2AX foci/nucleus observed in radiosensitive cells. The highest percent of γH2AX-positive cells was observed in radiosensitive cells and the lowest in radioresistant cells. As expected, the resolution of γH2AX foci was the most prominent in radioresistant cells. Similarly to our results, less γH2AX foci/nucleus were observed in radioresistant cells in comparison to radiosensitive cells after irradiation [
43]. Loss of γH2AX foci was the most prominent in radioresistant cells, indicating a greater DNA DSB repair as another alteration contributing to radioresistant phenotype [
44].
DNA DSB is the most lethal lesion induced by irradiation, and if not repaired, leads to cell death. Biphasic kinetics of DNA DSB repair reflects the mechanistic differences of the two major repair pathways, homologous recombination (HR) and non-homologous end-joining (NHEJ) [
45]. NHEJ is a fast, highly efficient, error-prone process active throughout the cell cycle, but predominantly in G
1-phase cells. On the other hand, HR is a slow, accurate process active in S- and G
2-phase cells in the presence of sister chromatid. In our study, radiosensitive cells displayed initial fast repair, followed by a slow resolution of γH2AX foci, while the radioresistant cells exhibited fast γH2AX foci resolution in both phases of DSB repair. Similar to our study, faster disappearance of γH2AX was observed in other radioresistant cells compared to slower γH2AX foci resolution in radiosensitive cells [
38,
46]. We concluded that the higher ability to repair DSB observed in studied radioresistant cells, in comparison to both parental and radiosensitive cells by both NHEJ and HR, importantly contributes to the cellular radioresistance.
Incorrect rejoining of radiation-induced DNA breaks can lead to structural and numerical chromosomal alterations [
47]. As observed by flow cytometry isogenic FaDu, FaDu-RR, and 2A3 cells do not differ in chromosome number. On the other hand, alterations in chromosome structure can affect gene expression. A more detailed genomic analysis is needed to specifically identify radiation-induced mutations, loss and gain of genetic material, and chromosomal abberations. The observed under-expression and over-expression of DNA DSR genes could be either due to alterations of the genetic material, or due to changes in the translation of DNA.
We further focused on DNA DSR gene expression because alterations of these genes can importantly determine radiosensitivity [
48]. We observed alterations in the expression of studied genes in non-irradiated radioresistant and radiosensitive cells, compared with parental cells. We also observed over-expression of DNA DSR gene expression in parental and radioresistant cells, but not in radiosensitive cells 5 h after irradiation. Similarly, other studies found more DNA repair genes to be up-regulated in radioresistant cells 6 h after irradiation than in radiosensitive cells [
49]. Prolonged transcriptional activity of DNA damage response, cell cycle and apoptosis-related genes was observed in radiosensitive cells in comparison to radioresistant cells in response to irradiation [
38].
We identified 4 genes of interest,
FANCD2,
GADD45A,
H2AFX, and
XRCC2 to be over-expressed in radiosensitive cells, but under-expressed in radioresistant cells.
FANCD2 gene product is an important component of the Fanconi anemia pathway of DNA damage response and acts as a recruitment factor for other DNA repair proteins [
50].
FANCD2 accumulation is associated with HPV-activated Fanconi anemia pathway and is essential for the maintenance of viral episomes in HPV-infected cells [
51]. Over-expression of
FANCD2 in radiosensitive 2A3 cells could be associated with the HPV16 E6 and E7 transfection.
GADD45A gene product is involved in cell cycle regulation, DNA repair, and apoptosis [
52]. Over-expression of
GADD45A promotes transcriptional activity through global demethylation [
53,
54]. Based on these results, the detected over-expression of
GADD45A in radiosensitive cells could be associated with the increased transcriptional activity and differential gene expression observed in these cells in comparison to parental cells. On the other hand, silencing
GADD45A is associated with increased survival, reduced apoptosis, and reduced sensitivity to CDDP [
54,
55]. Under-expression of
GADD45A could contribute to CDDP cross-resistance in radioresistant FaDu-RR cells. These data suggest a radioprotective mechanism through under-expression of
GADD45A. However, contrary to these observations, under-expression of
GADD45A in melanoma cells increased sensitivity to CDDP and enhanced CDDP-induced DNA damage [
56]. In addition,
GADD45A gene and protein expression are increased in response to a variety of DNA damaging agents, including ionizing radiation [
52,
55,
57]. Similarly, we observed an increase in
GADD45A gene expression 5 h after irradiation in parental and radioresistant cells, but not in radiosensitive cells. It is possible that intrinsically higher levels of
GADD45A in radiosensitive cells limit any further increase in the expression of this gene in response to irradiation. The observed over-expression of genes in parental and radioresistant cells, but not in radiosensitive cells, could be associated with
GADD45A-mediated increase in transcriptional activity [
53,
54].
H2AFX gene encodes histone H2AX, which is rapidly phosphorylated to γH2AX in response to DNA DSBs induced by radiation or DNA damaging agents. γH2AX then interacts with MDC1, another over-expressed gene in radioresistant cells in response to irradiation, and recruits other DNA damage response-associated proteins to the site of the DSB [
16]. Silencing
H2AX induced activation of epithelial-mesenchymal transition factors, and promoted metastatic behavior [
58]. Over-expression of H2AFX observed in radiosensitive cells is in agreement with the observed higher range of γH2AX foci observed by immunofluorescence. On the other hand, over-expression of
H2AFX in radioresistant cells, but not parental or radiosensitive cells, in response to irradiation could be an indication of enhanced recognition of DSBs, and faster disappearance of γH2AX foci observed.
XRCC2 gene product is a paralogue of RAD51 protein, involved in DNA DSB repair via HR [
59]. Knockdown of
XRCC2 in colon carcinoma cells decreased cell proliferation, increased apoptosis and lead to cell cycle arrest induced by irradiation [
60]. XRCC2 is also involved in the regulation of replication fork progression to prevent DNA damage and genomic instability [
61]. Although
XRCC2 is under-expressed in radioresistant cells, its expression is up-regulated early in the response to 5 Gy irradiation, which could contribute to faster DSB repair by HR in radioresistant cells. No change in
XRCC2 expression in parental and radiosensitive cells was observed in response to irradiation which is in agreement with slower HR repair of DNA DSB.
Another Rad51 paralogue,
XRCC3 was the only under-expressed gene in radiosensitive cells in response to the irradiation, indicating on a possible mechanism of increased sensitivity to ionizing radiation. As shown by Cheng J et al, knockdown of
XRCC3 increased radiosensitivity in vitro and in vivo, while high expression of
XRCC3 correlated with radio- and chemoresistance [
62].
By direct comparison of gene expression in 5 Gy-irradiated isogenic cells, we identified another gene of interest,
XPA, possibly involved in the radioresistance mechanism.
XPA gene product plays a central role in the NER pathway, through which CDDP-DNA adducts are repaired [
63].
XPA was over-expressed in radioresistant cells, but under-expressed in radiosensitive cells in response to irradiation. This was also observed in radioresistant glioblastoma cells [
64]. The increased expression of
XPA, and consequently activated NER could be associated with cross-resistance to CDDP, observed in radioresistant cells. In a parallel study, we demonstrated a more efficient repair of CDDP and BLM-induced DNA damage in radioresistant cells compared to parental cells [
15]. Other mechanisms contributing to CDDP resistance include an impaired influx of CDDP, and activation of different multidrug mechanisms, including glutathione and multidrug-resistance associated protein (MRP) [
30,
65]. Alterations in sensitivity to specific cytotoxic agents, including increased resistance and sensitivity, were observed in many experimentally induced radioresistant cells [
21,
22,
25,
27,
28,
30]. Increased level of ROS scavengers associated with cellular radioresistance can also reduce sensitivity to platinum drugs through neutralization of platinum-induced oxidative stress [
24,
27,
66]. On the other hand, the isogenic radiosensitive cells were more sensitive to CDDP, Oxa, and BLM. Isogenic pairs of HPV-positive and HPV-negative cells have not been evaluated in terms of chemosensitivity and the available data on CDDP sensitivity of HPV-positive head and neck SCC is not conclusive. HPV-positive cells were found to be either CDDP-resistant [
67] or equally sensitive to CDDP compared to HPV-negative head and neck SCC [
68].
The observed differences in DNA damage signaling gene expression in isogenic cells with different radiosensitivity indicate a role of several DNA damage signalling genes in the mechanisms of radioresistance. However, this should be further confirmed by quantitative and functional expression of specific proteins which is one of the future perspectives of this study. Identification of deregulated proteins could be applied clinically in terms of either radioresistance-associated biomarkers or as potential targets for radiosensitization. These novel biomarkers related to either radiosensitivity or radioresistance would permit improved stratification of HNSCC patients based on the predicted response to irradiation [
69]. Radiosensitization of more radioresistant tumors could also lead to reduced total irradiation dose, fewer and/or milder adverse effects. Dose de-escalation is a feasible approach for HPV-positive oropharyngeal HNSCC, with high response rate and reduced toxicity [
5,
70,
71].