Cell survival of NSCLC cell lines after X-ray treatment in normoxia and hypoxia

The survival ability of H460, A549 and Calu-1 cells under normoxia (21% O2) and hypoxia (0.1% O2) following X-ray exposure was investigated. Cell survival fraction (SF) curves were fitted according to the linear-quadratic model (SF = exp[−(αD + βD2)]) [20]. In normoxia (Fig. 1A, B), the plots clearly indicated that H460 cells had the lowest cell survival ability after IR for all the doses, in comparison to the higher SF for A549 and Calu-1 cells (Additional file 2: Table 1). However, in hypoxia, the differences among the SF profiles were abolished and the cells displayed a more radio-resistant response compared to their normoxic counterparts (Fig. 1A, B and Additional file 2: Table S1). Interestingly, the hypoxic samples exhibited a general increase in the α values and a decrease in the β values (Additional file 2: Table S2), implicating a more linear curvature with loss of the quadratic component compared with the normoxic profiles. The radio-protective effects of diminished oxygen levels were further evaluated using the OER factor which was determined by calculating the ratio of doses in normoxia and hypoxia at 10% of cell survival (D10) (Additional file 2: Table S3). H460 cells had the highest OER (2.18 ± 0.179) compared to A549 (1.78 ± 0.699) and Calu-1 (1.88 ± 0.386), indicating that the removal of oxygen had a more profound effect on enhancing the RR of the more radio-sensitive oxygenated cell line. Therefore, the influence of hypoxia in cell RR and the survival of NSCLC cells was dependent on the O2 percentage at the time of irradiation.

Fig. 1

Radiosensitivity of H460, Calu-1 and A549 cells. A Clonogenic assays showing varying radioresistance at 21% O2 (upper plot) and 0.1% O2 (lower plot). Cell lines were irradiated with doses ranging from 2 to 8 Gy in normoxia and 4–16 Gy in hypoxia. B The respective survival curves separated for each cell line. Each data point is presented as mean ± SD of three separate experiments performed in quintuplicate. The survival fractions were normalized to the respective 0 Gy samples. ***p < 0.001 as compared to 0 Gy normoxia; +++p < 0.001 as compared to 0 Gy hypoxia

DNA DSB detection and cell repair ability after IR in normoxia and hypoxia

In order to understand how the oxygen level influenced cell repair capability, the initial (30 min) and residual (24 h) IR-induced γ–H2AX foci formation, which suggests accumulation of DNA damage, (Fig. 2A), was investigated. The remaining foci at 24 h can provide an indication of the repair ability of each cell line.

Fig. 2

Effects of X-rays on DNA damage, gene expression, cell cycle and viability in H460, Calu-1 and A549 cells. A Representative confocal immunofluorescent images of γ–H2AX foci at 0 Gy and 8 Gy, 30 min after IR. B Detection of γ–H2AX foci in normoxia and hypoxia at 30 min and 24 h post-IR at different doses, including the respective non-irradiated controls (0 Gy). C Repair capacity determined by the ratio of foci count per cell (FPC) at 30 min and 24 h in normoxia and hypoxia. The data in A and C represent mean FPC ± SEM (n = 3); samples were analyzed at a confocal microscope with a sample size of > 100 cells per replicate and compared to their respective 0 Gy group (one-way ANOVA). D Stacked bar chart showing the cell cycle distribution (in %) of normoxic and hypoxic NucPE1-positive nuclei after 24 h from X-ray exposure. E Relative mRNA expression levels of WEE1 in normoxic and hypoxic samples 24 h post-IR. The data are presented as mean of the ΔΔCt values ± SEM (n = 2, each in triplicate). All samples were compared to the 0 Gy normoxic sample represented as the 0 on the x-axis (one-way ANOVA). F Flow cytometric quantification of dead cells 48 h post-IR by staining with propidium iodide (PI). The data in D and E represent the mean ± SEM (n = 3) and were statistically compared to their respective 0 Gy group (one-way ANOVA). *p < 0.05, **p < 0.01, ***p < 0.001

At 30 min post-IR in normoxia, a dose-dependent increase for all cell lines was observed in comparison with the respective non-irradiated (0 Gy) samples, with H460 and Calu-1 cells having a higher number of Foci Per Cell (FPC) (Fig. 2B). In hypoxia, the number of FPC was lower for all cell lines at 30 min post-IR, even with the OER factor applied to the doses (Fig. 2B). 24 h after X-ray exposure, the normoxic samples had repaired a considerable amount of foci in a dose-dependent manner, although for the highest doses the DNA DSBs remained significantly higher than the non-irradiated samples (Fig. 2B). This suggested that normoxic cells were able to activate the repair programs although not sufficiently for complete recovery of DNA DSBs at high doses (8 Gy). In hypoxia, the residual damage at 24 h was reduced and the dose dependency remained consistent for all cell lines (Fig. 2B). The calculated repair capacity (i.e. ratio of 30 min FPC and 24 h FPC) (Fig. 2C) reduced with increasing dose for all cell lines in both oxygen conditions, although there was a significantly lower repair ability seen in hypoxia compared with normoxia.

Cell cycle and death post-IR in normoxia and hypoxia

Ionizing radiation induces DNA damage and consequently forces cells into cycle arrest, ultimately resulting either in cell death, if reparation is not possible, or in continued proliferation with increased genome instability. Therefore, the cell cycle profile was investigated 24 h post-IR. At 30 min, no significant changes were detectable in the cell cycle distribution most likely because of the reduced time after exposure (data not shown).

At 24 h post-IR in normoxia (Fig. 2D, and Additional file 2: Table S4), nuclei were mainly accumulated in the G0-G1 phase compared to non-irradiated samples, but at a high dose of 8 Gy a block was also visible in the G2 phase in H460 and Calu-1 cells. At 0.1% O2, cells resulted mainly arrested in the G2 phase at the high doses. Overall, the activation of cell cycle blocks showed intercellular variability and intra-cellular dose-dependency.

The mRNA expression of WEE1 (Fig. 2E), whose main role is preventing cells from entering mitosis too early by activating a block in the G2 or S phases [21], appeared downregulated in normoxia for H460 and A549 cells post-IR, while no significant changes were detectable in Calu-1 cells. WEE1 expression in hypoxia was instead predominantly increased (except in some cases at high doses) across cell lines which was in accordance with the hypoxic cell cycle profiles. This indicated a potential WEE1 involvement in favoring the G2 block, although the O2 reduction appeared to be the main factor influencing WEE1 expression compared to IR.

Moreover, the possibility that cells could undergo cell death after cycle arrests was tested 24 and 48 h after X-ray exposure by measuring the amount of dead cells and counting cells. These viability results (Fig. 2F and Additional file 3: Fig. S1 and Additional file 4: Fig. S2) showed no statistically significant changes in the dead or dying cell pool up to 48 h in both oxygen conditions (except for the H460 cells at the highest dose). This suggested that either the cells were arrested, yet viable at these time-points, or recovered from blocking, but in both cases radiation treatment did not result in cell death up to 48 h from IR.

DNA damage repair gene expression in normoxia versus hypoxia 24 h post-IR

Based on these previous observations, the mRNA expression of some genes involved in the DDR sensory machinery (RAD50), NHEJ (XRCC5, XRCC6, DNA-PKcs, DCLRE1C and LIGASE4) and HR (RAD51, RAD52, BRCA1 and BRCA2) repair pathways was analyzed 24 h post-IR (Fig. 3). Data revealed that in all cell lines, the fluctuations of XRCC5 and XRCC6 (also known as KU80 and KU70, respectively), DNA-PKcs, DCLRE1C (also known as ARTEMIS) and LIGASE4 (LIG4) expressions in normoxia were not statistically significant, except LIG4 at the highest dose in H460 and A549 cells. At 0.1% O2, the NHEJ-related genes were downregulated for H460 and A549 cells with the exception of LIG4 that showed a trend to increase in H460 cells. In Calu-1 cells instead, DCLRE1C and LIG4 were both up-regulated. On the contrary, RAD51 and RAD52 showed similar trends among the cell lines in both O2 conditions in, albeit not always significant, with a more pronounced downregulation in hypoxia for RAD51, which was evident also for hypoxic Calu-1. BRCA1 and BRCA2 resulted both downregulated in all cell lines. Therefore, in severe hypoxic conditions, the IR doses played different roles depending on the cell lines and the two repair pathways were in general less expressed, with the exception of Calu-1 cells in which some upregulation was still detectable at 24 h from IR.

Fig. 3

Effects of X-rays and O2 levels on the expression of DNA-damage response genes. Relative mRNA expression levels of RAD50, XRCC5, XRCC6, DNA-PKcs, RAD51, RAD52, BRCA1 and BRCA2 in normoxic and hypoxic samples 24 h post-IR. The data are presented as mean of the ΔΔCt values ± SEM (n = 2, each in triplicate). All samples were compared to the 0 Gy normoxic sample represented as the 0 on the x-axis (one-way ANOVA)

Nuclear and cytosolic H2O2 levels post-IR in normoxia and hypoxia

To qualitatively estimate the nuH2O2 and cyH2O2 production after IR and its modulation in normoxia and hypoxia, NucPE1 and PY1-ME were used. Confocal images (Fig. 4A, B) confirmed the preferential localization of NucPE1 and PY1-ME in the nuclei and cytosol, respectively, of all cell lines. Moreover, to reduce the possibility of cytosolic background fluorescence or artifacts, the NucPE1 signal was measured only on stained nuclei extracted after IR.

Fig. 4

Effects of the combined hypoxia and X-ray treatments on H2O2 production and the antioxidant responses. A and B Representative confocal images of nuH2O2 and cyH2O2 detection in normoxic non-irradiated NSCLC cells by staining with NucPE1 and PY1-ME, respectively. Nuclei (blue) were counterstained with Hoechst 33,342. C and D Relative estimation of nuH2O2 and cyH2O2 content at 30 min and 24 h post-IR in normoxia and hypoxia. All samples were normalized and statistically compared to their respective normoxic or hypoxic control (0 Gy). Error bars are represented as SEM (n = 3) (one-way ANOVA). E Correlation between normalized (to 0 Gy) DSBs per cell and nuH2O2 levels for each dose in normoxia and hypoxia measured at 30 min post-IR. F Relative mRNA expression levels of intracellular NFE2L2 and CAT genes evaluated using RT-qPCR at 24 h following IR treatment in normoxia and hypoxia. The data are presented as mean of the ΔΔCt values ± SEM (n = 2, each in triplicate) (one-way ANOVA). *p < 0.05, **p < 0.01, ***p < 0.001 as compared to 0 Gy normoxia; +p < 0.05, ++p < 0.01, +++p < 0.001 as compared to 0 Gy hypoxia

The H2O2 measurements at 30 min post-IR aimed at investigating the initial IR-induced H2O2 levels and potentially correlating nuH2O2 with the DSB damage. Further, the H2O2 content analyzed 24 h post-IR aimed at evaluating cell ability to counteract the oxidative stress induced by IR treatments.

Results obtained from the normoxic samples at 30 min indicated a fast rise in nuH2O2 production with increasing doses for A549 and Calu-1, whereas for H460 cells similar increases were produced for all doses with a clear plateau of the profile (Fig. 4C and Additional file 5: Fig. 3A). At 24 h post-IR, nuH2O2 levels returned to basal values in H460 cells, whereas they remained higher than the controls in A549 and Calu-1 cells for 4 and 8 Gy. By reducing oxygen availability, the effects of radiation after 30 min on nuH2O2 were less pronounced (Fig. 4C and Additional file 5: Fig. 3A) and no dose dependency was observed for all cell lines. At 24 h, changes in the nuH2O2 production were no longer detectable, except for the highest doses in Calu-1 cells (Additional file 2: Table 5). With the exception of Calu-1 cells at 30 min, all the hypoxic cells showed a minor content of nuH2O2 compared to the normoxic samples, which was more evident at 24 h post-IR (Additional file 5: Fig. 3A).

Interestingly, by plotting the DSB data against the nuH2O2 data obtained 30 min post-IR (Fig. 4D), a linear dependence of the DSBs produced with nuH2O2 was present in normoxia. For the hypoxic cells instead, no correlation was observed (Fig. 4D).

The cyH2O2 levels of normoxic cells were seemingly less responsive to IR compared to the nuH2O2 levels (Fig. 4E and Supplementary Fig. 3B). In fact, at both time-points, only normoxic H460 cells appeared to increase their cyH2O2 content. On the contrary, in hypoxic cells, the cyH2O2 levels were subjected to a decrease after IR at both time-points, except for A549 cells at 30 min (Additional file 2: Table 6). Moreover, also the cyH2O2 levels were reduced compared to the normoxic cells, albeit H460 cells at 0 Gy showed a higher content (Additional file 5: Fig. 3B). This apparent discrepancy at the moment remains to be better investigated.

In general, results potentially suggested a lower oxidative stress in hypoxia and a more readily active and sustained antioxidant capacity in the cytoplasm compared to the nucleus.

Based on these findings, the relative mRNA expression levels of the antioxidant response master regulator, NFE2L2 (also known as NRF2), were analyzed 24 h post-IR (Fig. 4F). At 21% O2, in H460 and A549 cells a dose-increase was mainly observed which resulted significant only for the highest dose. In normoxic Calu-1 cells, no modulation of NFE2L2 gene expression was detectable at any dose. However, the reduction of O2 markedly induced the overexpression of NFE2L2 in all cell lines independently on the IR doses. The analysis of mRNA levels of CAT, which is induced by NFE2L2 and is directly involved in H2O2 removal, showed IR-induced stimulation in normoxia for all cell lines (Fig. 4F). In hypoxia, CAT mRNA was upregulated in H460 and Calu-1 cells, with a drop at 16 Gy for the latter cell line. In A549 cells, a significant change in CAT mRNA expression was observed only at 16 Gy, even though NFE2L2 was upregulated, suggesting that NFE2L2 was not directly influencing CAT gene expression in this cell line. However, a mild tendency towards an increased expression was observed following irradiation in hypoxia.

These findings suggested that IR triggered some antioxidant responses which were noticeably influenced by the O2 level.

Measurements of glutathione levels after X-rays in normoxia and hypoxia

In order to investigate cellular ability to scavenge H2O2, measurements of intracellular reduced and oxidized glutathione levels were also performed. Glutathione, in fact, is the most abundant low-molecular-mass antioxidant serving as an essential cofactor for the reduction of H2O2 to H2O catalyzed by glutathione peroxidases (GPX). The ratio of GSH/GSSG is high under normal conditions and decreases upon pro-oxidant stresses providing an indication of the cellular redox environment (Fig. 5A).

Fig. 5

Effects of the IR-induced oxidative stress on the glutathione system in normoxia and hypoxia. A Schematic representation of some key components of the glutathione system involved in H2O2 detoxification. B Ratio of GSH/GSSG levels in normoxic and hypoxic samples at 30 min and 24 h post-IR in H460, A549 and Calu-1 cells. C Oxidized (GSSG) and reduced (GSH = total glutathione—GSSG) glutathione levels at 24 h post-IR following incubation in normoxia and hypoxia. In B and C the data are showed as mean ± SD (n = 3, in duplicate) and were statistically compared to their respective 0 Gy (one-way ANOVA). D Relative mRNA expression levels of GPX1, GPX4, GSR, GLRX genes at 24 h post-IR. Data are presented as the ΔΔCt mean ± SEM (n = 2, each in triplicate) (one-way ANOVA). *p < 0.05, **p < 0.01, ***p < 0.001 as compared to 0 Gy normoxia; +p < 0.05, ++p < 0.01, +++p < 0.001 as compared to 0 Gy hypoxia

In normoxia, changes in the GSH/GSSG ratio at 30 min were observed only in H460 cells with a slight reduction at 2 and 8 Gy and such a profile was maintained, although slightly decreased, up to 24 h (Fig. 5B). A similar trend at the later time-point was observed for normoxic A549, suggesting a condition of mild oxidative stress in both these cell lines. However, the strongest difference in the GSH/GSSG ratios was observed in hypoxic samples compared to their normoxic counterparts (Fig. 5B), which were maintained higher also after the X-ray exposure at both time-points, with only a decreased trend in Calu-1 cells. In this latter cell line, at 24 h, the GSH/GSSG reduction was dose-dependent and associated with a GSH depletion (Fig. 5C). Instead, in the other two cell lines, the GSSG hypoxic levels were maintained at a lower level than in normoxia (Fig. 5C). This suggested a stronger ability to cope with the IR-induced intracellular oxidative stress in comparison to Calu-1 cells.

Altogether, this evidence indicated that the main role in modulating the glutathione ratio was played by the oxygen levels and the radiation treatment exhibited only cell specific effects.

In order to assess whether the observed effects were correlated with a modulation of the glutathione system genes, the mRNA expression of GPX1 and 4, Glutathione-disulfide Reductase (GSR), and Glutaredoxin-1 (GLRX) was evaluated (Fig. 5D). The results showed an upregulation of GPX1 mRNA post-IR in normoxic samples to a different extent in all cell lines. Under hypoxia, cell-type specific behaviors were observed with downregulations in H460 and A549 cells, and upregulations in Calu-1 cells. When irradiated in hypoxia, all cell lines responded with an increased gene expression compared to their respective untreated hypoxic controls. However, in comparison with the 0 Gy normoxic samples, in H460 cells GPX1 expression was upregulated already at 4 Gy, in A549 cells only the highest dose led to an upregulation, and for Calu-1 cells GPX1 was always overexpressed. Instead, GPX4 mRNA levels were upregulated in all cell lines, although not always significantly. Also, GSR mRNA levels were not statistically modified in normoxia and were observed downregulated in hypoxia. In contrast, GLRX mRNA levels were always upregulated in all cell lines, with a slight dose-dependency in normoxia and a larger effect following O2 reduction for H460 and Calu-1 cells compared to A549 cells. Therefore, in hypoxia, GLRX rather than GSR could mostly participate in maintaining the low levels of GSSG observed in the present study.