Low Repair Capacity of DNA Double-Strand Breaks Induced by Laser-Driven Ultrashort Electron Beams in Cancer Cells

Laser-driven accelerators allow to generate ultrashort (from femto- to picoseconds) high peak dose-rate (up to tens of GGy/s) accelerated particle beams. However, the radiobiological effects of ultrashort pulsed irradiation are still poorly studied. The aim of this work was to compare the formation and elimination of γH2AX and 53BP1 foci (well known markers for DNA double-strand breaks (DSBs)) in Hela cells exposed to ultrashort pulsed electron beams generated by Advanced Research Electron Accelerator Laboratory (AREAL) accelerator (electron energy 3.6 MeV, pulse duration 450 fs, pulse repetition rates 2 or 20 Hz) and quasi-continuous radiation generated by Varian accelerator (electron energy 4 MeV) at doses of 250–1000 mGy. Additionally, a study on the dose–response relationships of changes in the number of residual γH2AX foci in HeLa and A549 cells 24 h after irradiation at doses of 500–10,000 mGy were performed. We found no statistically significant differences in γH2AX and 53BP1 foci yields at 1 h after exposure to 2 Hz ultrashort pulse vs. quasi-continuous radiations. In contrast, 20 Hz ultrashort pulse irradiation resulted in 1.27-fold higher foci yields as compared to the quasi-continuous one. After 24 h of pulse irradiation at doses of 500–10,000 mGy the number of residual γH2AX foci in Hela and A549 cells was 1.7–2.9 times higher compared to that of quasi-continuous irradiation. Overall, the obtained results suggest the slower repair rate for DSBs induced by ultrashort pulse irradiation in comparison to DSBs induced by quasi-continuous irradiation.


Introduction
In recent years, laser-generated particles' acceleration technologies have been actively developed. Laser-generated particle beams are characterized by ultrashort duration (from femtoseconds to

Dose-Responses for γH2AX and 53BP1 Foci
Comparative analysis of γH2AX and 53BP1 foci yields was performed at 1 h post-irradiation and 0.25-1.0 Gy doses, following cells the irradiation on Advanced Research Electron Accelerator Laboratory (AREAL) and Varian accelerators. Cell irradiation on the AREAL accelerator was carried out using different pulse repetition and dose-rates (2, 20 Hz and 1.2, 11.7 Gy/min, respectively), while irradiation on Varian machine was done at dose-rate of 5.6 Gy/min. Dose-dependent changes in γH2AX foci numbers are presented in Figure 1a. It has been shown that following irradiation using the AREAL accelerator at a 20 Hz repetition rate, a dose-dependent increase in foci numbers was described with a high fit (R 2 = 0.99, p < 0.001) by y = 3.57 + 22.29 × x linear equation, where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy). Lowering the repetition rate of irradiation up to 2 Hz resulted in y = 3.62 + 19.82 × x (R 2 = 0.99, p < 0.001) equation, where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy). In a linear equation (y = a + b × x), describing a dose-dependent relationship, the slope coefficient b reflects the increase in the effect per dose unit. The comparison of the slope coefficients of two different dose-response curves, obtained by different exposure types, reflects for how many fold one of them is more expressed compared to the other. After 2 Hz irradiation, the slope coefficient was found to be about 1.13 times lower with respect to 20 Hz, however, this difference was not statistically significant (z = 1.59, p = 0.056). For irradiation on the Varian accelerator, a dose-dependent relationship was found to be expressed by y = 2.64 + 17.60 × x (R 2 = 0.99, p < 0.001), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy). The slope coefficient was 1.27 times lower (z = 2.82, p = 0.002) as compared to irradiation on the AREAL accelerator with 20 Hz repetition rate.
1.27 times lower (z = 2.82, p = 0.002) as compared to irradiation on the AREAL accelerator with 20 Hz repetition rate. Dose-dependent changes in 53BP1 foci numbers are presented in Figure 1b. In overall, doseeffect relationships are similar to those revealed for γH2AX foci and were found to be described by the following equations: (1) Irradiation on the AREAL accelerator at a 20 Hz repetition rate-y = 4.81 + 20.82 × x (R 2 = 0.97, p = 0.002), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy); (2) Irradiation on the AREAL accelerator at 2 Hz-y = 4.19 + 18.19 × x (R 2 = 0.99, p < 0.001), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy); (3) Irradiation on the Varian accelerator-y = 3.62 + 16.35 × x (R 2 = 0.99, p < 0.001), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy). Dose-dependent changes in 53BP1 foci numbers are presented in Figure 1b. In overall, dose-effect relationships are similar to those revealed for γH2AX foci and were found to be described by the following equations: (1) Irradiation on the AREAL accelerator at a 20 Hz repetition rate-y = 4.81 + 20.82 × x (R 2 = 0.97, p = 0.002), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy); (2) Irradiation on the AREAL accelerator at 2 Hz-y = 4.19 + 18.19 × x (R 2 = 0.99, p < 0.001), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy); (3) Irradiation on the Varian accelerator-y = 3.62 + 16.35 × x (R 2 = 0.99, p < 0.001), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy).
No statistically significant differences between the linear slope coefficients were found after irradiation on the AREAL accelerator at two repetition rates (z = 1.14, p = 0.156). However, significant differences were observed between the irradiation effects on AREAL at 20 Hz and Varian accelerators, and particularly the linear slope coefficient differed by about 1.27 times (z = 1.96, p = 0.024).
Thus, the comparative analysis of dose-dependent changes in γH2AX and 53BP1 foci numbers after irradiation on AREAL and Varian accelerators revealed, that overall quantitative effects per dose unit are comparable. However, differences related to the changes of pulse repetition and associated dose rate used on AREAL accelerator were noted, and particularly, it was shown that decreasing the repetition rate from 20 to 2 Hz, respectively, resulted in a decrease in foci formation per dose unit by about 13-14%.

Post-Irradiation Changes of γH2AX and 53BP1 Foci Numbers
Then, we studied the post-irradiation changes in γH2AX and 53BP1 foci numbers in response to 1 Gy irradiation. The decrease in the time kinetics of γH2AX and 53BP foci numbers reflects the speed by which radiation-induced DSBs are repaired. The 1, 4 and 24 h after irradiation were selected as the time points.
After exposure with pulsed and quasi-continuous irradiations, a significant difference in the time course of γH2AX foci number changes were observed ( Figure 2). A statistically significant increase in γH2AX foci numbers at 4 h post-irradiation on the AREAL accelerator at a 20 Hz repetition rate was found compared to the irradiation on Varian. Moreover, 24 h post-irradiation, the statistically significant difference between the effects of pulsed and quasi-continuous irradiations was observed at both pulse rates (2 and 20 Hz). No statistically significant differences between the linear slope coefficients were found after irradiation on the AREAL accelerator at two repetition rates (z = 1.14, p = 0.156). However, significant differences were observed between the irradiation effects on AREAL at 20 Hz and Varian accelerators, and particularly the linear slope coefficient differed by about 1.27 times (z = 1.96, p = 0.024).
Thus, the comparative analysis of dose-dependent changes in γH2AX and 53BP1 foci numbers after irradiation on AREAL and Varian accelerators revealed, that overall quantitative effects per dose unit are comparable. However, differences related to the changes of pulse repetition and associated dose rate used on AREAL accelerator were noted, and particularly, it was shown that decreasing the repetition rate from 20 to 2 Hz, respectively, resulted in a decrease in foci formation per dose unit by about 13-14%.

Post-Irradiation Changes of γH2AX and 53BP1 Foci Numbers
Then, we studied the post-irradiation changes in γH2AX and 53BP1 foci numbers in response to 1 Gy irradiation. The decrease in the time kinetics of γH2AX and 53BP foci numbers reflects the speed by which radiation-induced DSBs are repaired. The 1, 4 and 24 h after irradiation were selected as the time points.
After exposure with pulsed and quasi-continuous irradiations, a significant difference in the time course of γH2AX foci number changes were observed ( Figure 2). A statistically significant increase in γH2AX foci numbers at 4 h post-irradiation on the AREAL accelerator at a 20 Hz repetition rate was found compared to the irradiation on Varian. Moreover, 24 h post-irradiation, the statistically significant difference between the effects of pulsed and quasi-continuous irradiations was observed at both pulse rates (2 and 20 Hz). Difference in the γH2AX foci numbers were observed also at 1 h after pulse and quasicontinuous irradiations, however, it did not reach statistical significance. To quantify for this trend, we calculated the relative changes in foci numbers with respect to the values at 1 h post-irradiation. It was shown that 4 h after exposure to pulsed irradiation, the relative foci number was 1.5-1.7 times higher as compared to the quasi-continuous one, while 24 h post-irradiation this value was 2.5-2.9 Difference in the γH2AX foci numbers were observed also at 1 h after pulse and quasi-continuous irradiations, however, it did not reach statistical significance. To quantify for this trend, we calculated the relative changes in foci numbers with respect to the values at 1 h post-irradiation. It was shown that 4 h after exposure to pulsed irradiation, the relative foci number was 1.5-1.7 times higher as compared to the quasi-continuous one, while 24 h post-irradiation this value was 2.5-2.9 times higher (Figure 2b). At the same time, no significant difference between the effects of pulsed irradiation at 2 and 20 Hz was noted.
Similar trends were observed also for 53BP1 foci (Figure 3). However, in this case, a statistically significant difference in residual foci numbers was found only 24 h post-irradiation, after exposure with pulsed and quasi-continuous irradiations. At 24 h post-irradiation, the relative foci number, induced by exposure to pulsed irradiation, was about 2.3-2.5 times higher than that induced by quasi-continuous irradiation. No difference was found between the effects of pulsed irradiation at 2 and 20 Hz. times higher (Figure 2b). At the same time, no significant difference between the effects of pulsed irradiation at 2 and 20 Hz was noted. Similar trends were observed also for 53BP1 foci (Figure 3). However, in this case, a statistically significant difference in residual foci numbers was found only 24 h post-irradiation, after exposure with pulsed and quasi-continuous irradiations. At 24 h post-irradiation, the relative foci number, induced by exposure to pulsed irradiation, was about 2.3-2.5 times higher than that induced by quasicontinuous irradiation. No difference was found between the effects of pulsed irradiation at 2 and 20 Hz.

Dose-Dependent Changes in γH2AX Residual Foci Numbers
Experiments for the dose-dependence studies of the number of residual γH2AX foci in HeLa and A549 cells at 24 h after pulse and quasi-continuous irradiations at doses of 500, 1000, 2000, 4000 and 10,000 mGy have been performed to confirm the phenomenon of the delayed elimination of DNA DSB repair proteins foci. It was shown that over the entire studied dose range, the residual γH2AX foci after exposure to the pulse irradiation was 1.7-2.9 times higher than that after quasi-continuous irradiation exposure (Figure 4). However, due to the high scatter of experimental data, the statistically significant difference was observed only after irradiation at doses of 1 and 2 Gy in the case of Hela cells (Figure 4a) and at doses of 2 and 4 Gy in the case of A549 cells (Figure 4b). It is noteworthy that after the quasi-continuous irradiation of dose-dependent relationships, changes in the residual foci number are described by linear equations for both HeLa and A549 cells (Figure 4), whereas after pulse irradiation, the linear dose-dependence of changes of residual foci number was observed only in A549 cells (y = 4.93 + 1.23 × x (R 2 = 0.91, p = 0.003), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy)). In the case of HeLa cells, the number of residual foci was changed insignificantly at the dose range of 1-10 Gy and the "saturation" effect was observed ( Figure  4a).

Dose-Dependent Changes in γH2AX Residual Foci Numbers
Experiments for the dose-dependence studies of the number of residual γH2AX foci in HeLa and A549 cells at 24 h after pulse and quasi-continuous irradiations at doses of 500, 1000, 2000, 4000 and 10,000 mGy have been performed to confirm the phenomenon of the delayed elimination of DNA DSB repair proteins foci. It was shown that over the entire studied dose range, the residual γH2AX foci after exposure to the pulse irradiation was 1.7-2.9 times higher than that after quasi-continuous irradiation exposure (Figure 4). However, due to the high scatter of experimental data, the statistically significant difference was observed only after irradiation at doses of 1 and 2 Gy in the case of Hela cells (Figure 4a) and at doses of 2 and 4 Gy in the case of A549 cells (Figure 4b). It is noteworthy that after the quasi-continuous irradiation of dose-dependent relationships, changes in the residual foci number are described by linear equations for both HeLa and A549 cells (Figure 4), whereas after pulse irradiation, the linear dose-dependence of changes of residual foci number was observed only in A549 cells (y = 4.93 + 1.23 × x (R 2 = 0.91, p = 0.003), where y is an average of foci numbers in a cell nucleus and x is an irradiation dose (Gy)). In the case of HeLa cells, the number of residual foci was changed insignificantly at the dose range of 1-10 Gy and the "saturation" effect was observed (Figure 4a)  The comparative analysis of the slope coefficients of dose-response changes in residual foci number has been performed in A549 cells. The linear slope coefficient was shown to be 1.7 times higher in the case of pulse irradiation compared to quasi-continuous irradiation (z = 2.5, p = 0.006).
Overall, the obtained results confirm the conclusion about the delayed nature of the elimination of pulse radiation-induced DSBs repair proteins foci in cancer cells compared to that after quasicontinuous irradiation. The comparative analysis of the slope coefficients of dose-response changes in residual foci number has been performed in A549 cells. The linear slope coefficient was shown to be 1.7 times higher in the case of pulse irradiation compared to quasi-continuous irradiation (z = 2.5, p = 0.006).
Overall, the obtained results confirm the conclusion about the delayed nature of the elimination of pulse radiation-induced DSBs repair proteins foci in cancer cells compared to that after quasi-continuous irradiation.

Discussion
In this study, it was shown that exposure with both ultrashort pulsed and quasi-continuous electron irradiation at 1 h post-irradiation led to the linear dose-dependent formation of γH2AX and 53BP1 foci, which are markers of DNA DSBs. This time-point was selected since it has been reported that the maximum number of γH2AX and 53BP1 foci is formed at 0.5-1 h post-irradiation [23,24]. For quasi-continuous electron irradiation, the quantitative yields of γH2AX and 53BP1 per Gy/cell are comparable to those reported in the literature for the exposures of cancer cells with low linear energy transfer (LET) radiation (17.6 and 16.4 foci/Gy/cell, respectively) [22,24]. After exposure with ultrashort pulsed irradiation at 2 and 20 Hz, the quantitative yield of foci was higher (about 1.1 and 1.3 times), though statistically significant changes were observed only at 20 Hz.
It is important to note that the quantitative yield of DNA DSBs marker proteins foci at certain time points after irradiation depends not only on irradiation-induced initial number of DSBs, but also on the complexity of these lesions. According to the two-lesion kinetic model of double-strand breaks rejoining, DNA DSBs can be divided into simple and complex ones [25]. The repair of simple, i.e., single DSBs, is carried out very fast (15 min repair half-time), while the repair of complex DSBs, i.e., complicated by other lesions, performed slowly (10 to 15 h repair half-time) [25]. Depending on the ratio of simple and complex DSBs, the kinetics of DNA repair and accordingly, the time course of changes in DSBs foci number will vary.
In overall, the post-irradiation kinetics of change in γH2AX and 53BP1 numbers upon exposure to quasi-continuous radiation was in accordance to those reported in the literature for low LET radiation [22,24]. At 4 h post-irradiation about 45% of the γH2AX and 40% of 53BP1 foci of those observed at 1 h still were found, while at 24 h these values were 7 and 6%, respectively. Quite a different picture was shown after the exposure to ultrashort pulse radiation. At 4 h post-irradiation, about 59-64% of γH2AX and 53-57% 53BP1 foci of those observed at 1 h post-irradiation were found. It should be pointed out that though foci numbers were slightly higher after irradiation at 20 Hz, statistically significant differences induced by 2 and 20 Hz frequencies were not observed.
In addition, the dose-dependent changes in the number of residual γH2AX foci have been studied in HeLa and A549 cells at 24 h post-irradiation at doses of 500, 1000, 2000, 4000 and 10,000 mGy. It was shown that the number of residual γH2AX foci after exposure to the pulse irradiation was 1.7-2.9 times higher compared to the quasi-continuous irradiation across the entire dose range studied.
In general, the obtained data indicated slower DSB repair rate induced by ultrashort pulsed irradiation, compared to the ones induced by quasi-continuous irradiation. The pulse duration of ultrashort irradiation is only 0.4 × 10 −12 s, however, a huge peak dose-rate of 1.6 × 10 10 Gy/s per pulse is achieved during the pulse. Apparently, it increases the possibility of complex difficulty repairable DSBs formation. Further detailed studies of the physicochemical mechanisms of biological effects induced by sub-picosecond pulse irradiation are needed.
Prior to irradiation, the cells were seeded at a density of 0.4 × 10 5 cells/mL in 2.5 mL of culture medium onto coverslips (Thermo Fisher Scientific, Waltham, MA, USA) placed inside 35 mm Petri dishes (Corning, New York, NY, USA) and incubated at 37 • C and 5% CO 2 for 20 h.

Ultrashort Beam Irradiation
Ultrashort beam irradiation was carried out using an electron beam generated by a laser-driven radiofrequency gun-based linear AREAL accelerator. The characteristics of the AREAL accelerator have been described previously [26]. The parameters of the AREAL laser-generated electron beam are presented in Table 1. The dosimetric measurements were performed with a Faraday cup (commercially available), estimating the integral dose over the pulse. Cells were irradiated at doses of 0.25-10 Gy (~140 electron pulses per 1 Gy) with a repetition rate of 2 or 20 Hz. A peak dose rate of 1.6 × 10 10 Gy/s was estimated from the electron pulse duration of 4.5 × 10 -13 s, based on the laser pulse length, acceleration process, and electron beam transport. The mean absorbed dose-rate of 1.17 ± 0.02 Gy/min (repetition rate 2 Hz sample mass of 4.2 g) and 11.70 ± 0.98 Gy/min (repetition rate 20 Hz sample mass of 4.2 g) was calculated over the period of irradiation and 1% charge fluctuation and 1% beam energy fluctuation was taken into account.

Quasi-Continuous Irradiation
Quasi-continuous irradiation was carried out using the Varian Trilogy (Varian Medical Systems, Palo Alto, CA, USA) electron linear accelerator. Characteristics: electron energy 4.0 MeV, dose-rate 5.6 Gy/min, electron beam size 250 × 250 mm. Dosimetry was performed by ionization method in a water phantom according to the International Atomic Energy Agency (IAEA) TRS-398 international protocol [27].

Statistical Analysis
Statistical and mathematical analyses of the data were conducted using the Statistica 8.0 software (StatSoft, Tulsa, OK, USA). The results are presented as the means of three independent experiments ± standard error. Statistical significance was tested using the Student t-test. The slope coefficients of dose-response curves were compared using the Z-test.