DNA DSB Repair Dynamics following Irradiation with Laser-Driven Protons at Ultra-High Dose Rates

Protontherapy has emerged as more effective in the treatment of certain tumors than photon based therapies. However, significant capital and operational costs make protontherapy less accessible. This has stimulated interest in alternative proton delivery approaches, and in this context the use of laser-based technologies for the generation of ultra-high dose rate ion beams has been proposed as a prospective route. A better understanding of the radiobiological effects at ultra-high dose-rates is important for any future clinical adoption of this technology. In this study, we irradiated human skin fibroblasts-AG01522B cells with laser-accelerated protons at a dose rate of 109 Gy/s, generated using the Gemini laser system at the Rutherford Appleton Laboratory, UK. We studied DNA double strand break (DSB) repair kinetics using the p53 binding protein-1(53BP1) foci formation assay and observed a close similarity in the 53BP1 foci repair kinetics in the cells irradiated with 225 kVp X-rays and ultra- high dose rate protons for the initial time points. At the microdosimetric scale, foci per cell per track values showed a good correlation between the laser and cyclotron-accelerated protons indicating similarity in the DNA DSB induction and repair, independent of the time duration over which the dose was delivered.


DNA DSB damage repair kinetics induced by laser-accelerated protons.
The effect of laser-accelerated protons on the DNA DSB damage and repair was quantified by using the p53 binding protein-1 (53BP1) foci formation assay at 0.5, 1, 2, 6 and 24 hours after irradiation. The irradiation set up used in our study is shown in Fig. 1. The cells were grown on 3 μm Mylar mounted on bespoke stainless steel dishes and held perpendicular to the dispersion plane of the laser-accelerated proton beam. Beam characterization on a shot-to-shot basis was carried out via routine EBT2 Gafchromic film densitometry as shown in Fig. 2. After irradiation and immunofluorescence staining (see methods section) the cells were scored for 53BP1 foci quantification as shown in Fig. 3, in the form of Box-Whisker plots. The Box-Whisker plots show the range of foci per cell obtained at each time point. The dividing line in each box shows the median and the error bars indicate 5 and 95 percentiles of the 53BP1 foci per cell for each time point. The outliers are indicated above and below the error bars.
Foci induced by the laser-accelerated protons were compared to 225 kVp X-rays induced foci and the results are shown in Fig. 4, where the experimental data were fitted by means of a biphasic (two phase) exponential decay equation, as used in Grosser et al. 20 : where Y(t) is the number of foci at time t; A F and A S represent the initial induced foci related to the fast and slow kinetics, respectively; k F and k S are constants accounting for the fast and slow rate of repair, respectively; and B indicates background foci which are left unrepaired. The mean foci/nucleus/Gy measured at 0.5 hours for 225 kVp X-rays and 10 MeV protons were 24 ± 1 and 26 ± 2 respectively. At the 24 hours time-point, the majority of the DNA DSB were repaired as shown by the disappearance of 53BP1 foci leading to an overall reduction in average foci number close to the background level of 1.8 ± 0.5 and 2.9 ± 0.5 respectively for X-rays and protons, while the values for controls were 1.4 ± 0.8 and 1.5 ± 0.2 respectively for X-rays and proton groups.
By fitting our data with the above model we obtained slight (non-significant) differences between the repair rates of proton-induced foci and X-ray-induced foci. The best-fit values for the constant parameters of the Design of the dish where cells were grown as monolayers on 3 µm thin Mylar. Before irradiation the dish was mounted with another piece of mylar, to prevent the drying of monolayers and the gap between the two mylar pieces was filled with cell culture medium. During irradiation the medium was withdrawn with a motorized syringe system and refilled after irradiation of cells.  www.nature.com/scientificreports www.nature.com/scientificreports/ function were: for the fast repair kinetic, A F equal to 25.9 ± 3.4 and 39.1 ± 6.1, and k F equal to 1.08 ± 0.03 and 2.25 ± 0.45, respectively for protons and X-rays; for the slow repair kinetic, A S equal to 9.8 ± 1.0 and 10.6 ± 0.8, and k S equal to 0.080 ± 0.007 and 0.14 ± 0.03, respectively for the laser accelerated protons and X-rays. The foci background B term is assumed to be equal to the number of foci of the controls. The fast and slow half-lives were 8.7 ± 0.8 and 0.64 ± 0.02 hours for laser-accelerated protons, and 4.9 ± 1.1 and 0.31 ± 0.06 hours for X-rays. There is an observable difference in the slow and fast repair half-lives of the laser accelerated protons and X-rays induced foci, however it is uncertain whether this simply results from the higher RBE of protons compared to the X-rays or the higher dose-rate has an effect. To fully understand the implications of ultrahigh dose rate upon the DNA DSB foci kinetics, a detailed comparison with RF-accelerated protons at similar energy and LET is further warranted. 53BP1 foci persistence measurements. In order to obtain an insight on the complexity of DSB foci repair we further calculated the percentage of residual DNA DSB damage remaining as shown in Fig. 4b, where this is defined as the percentage of the number of radiation-induced 53BP1 foci at the time (t) with respect to the maximum number of radiation-induced foci (obtained at 0.5 hours), for the same radiation type. The foci repair kinetics in terms of percentage of foci remaining for 225 kVp X-rays and 10 MeV protons followed a two-phase exponential decay fitting, as shown in Fig. 4a and b, and as demonstrated in earlier papers for low LET radiation of 2 keV/µm and 4.59 keV/µm, respectively 21,22 . Sub-population radiosensitivity analysis. Additionally, the heterogeneity in the radiation response of the cells scored for foci was evaluated through sub-population radiosensitivity analysis as shown in Fig. 5. At the 6 hours time point in the X-ray irradiated group, we measured around 79% of cells showing 5-9 foci compared to the proton irradiated cells where only 50% of cells show 5-9 foci and 44% cells showed greater yields of 10-14 foci per cell. The repair of the damage (assumed following foci disappearance) could be observed for both protons and X-rays at 24 hours, where about 50% of cells showed 0-4 foci and over 45% of cells showed 5-9 foci. However in the X-ray irradiated group only 1% of cells showed 10-14 foci per cell.
Relative Foci Induction comparisons. For treatment planning optimization, a RBE value of 1.1 is typically assigned to the protons, although several investigators have shown variations in the proton RBE. While we have not calculated a cell killing RBE in this manuscript, nonetheless we compare the biological effectiveness in terms of foci induction, henceforth referred to as Relative Foci Induction or RFI, defined as the ratio of average foci per cell per Gy induced by protons to the same dose of 225 kVp X-rays in the cells. This is plotted in Fig. 6a, where a dashed line indicates the baseline of 1.1 constant value (considering a fixed effectiveness value of protons) and variations observed in the calculated RFI over time.
Size of 53BP1 foci. We measured the size of the 53BP1 foci to gain insights on the local accumulation of 53BP1 protein in the DNA DSB domains upon damage induced by the laser-accelerated protons or the conventional 225 kVp X-rays at the 0.5, 2, 6 and 24 hours time-points post exposure. The foci sizes were analyzed using the Analyze Particles plugin in the ImageJ software. At least 100 foci for each data point were evaluated and the results are shown in Fig. 6b. We found a time dependent increase in the foci size from 0.5 to 24 hours, with a statistical significance of p = 0.0005 and p < 0.0001 for 225 kVp X-rays and 10 MeV protons, respectively. Overall the size of the foci induced by the laser-accelerated protons and X-rays showed a close similarity indicating the www.nature.com/scientificreports www.nature.com/scientificreports/ low LET nature of the laser-accelerated protons. The increased foci size observed at 24 hours, is mainly due to the presence of some of the unrepaired DSB repair foci still persisting at 24 hrs although the frequency of such cells is very low at this time point. Figure 7 shows the comparison of Laser accelerated protons (LAP) induced 53BP1 foci per cell per track values with cyclotron-accelerated protons (CAP) induced foci per cell per track at 30 minutes and 24 hours after irradiation. We could only compare the data at these two time points due to the availability of cyclotron accelerated protons data for similar energy at these time points only. Our results show a close correlation between the LAP and CAP induced foci per track at both time points. Further, the comparison of the ratio of foci per cell per track at 30 minutes and 24 hours in case of LAP (as shown in Fig. 7c), also matched closely to the ratio of CAP induced foci, with non-significant variations.

Discussion
This work is broadly motivated by the prospective development of novel approaches to cancer therapy, and by the need to develop a basic understanding of the pre-clinical radiobiology in normal human cells of ion therapy at ultra high dose rates. We have used normal human skin fibroblast cell line AG01522B, a well characterized, radio-biologically relevant model system, to study DNA DSB repair dynamics following exposure to  www.nature.com/scientificreports www.nature.com/scientificreports/ laser-accelerated protons at dose rates >10 9 Gy/s. As the dosimetry of laser-accelerated protons is complex, slight variations in the energy and dose can make biological observations error prone, thus requiring careful consideration of all the confounding factors in dosimetry.
The dose measured by radiochromic film does not solely represent the actual dose deposited in cells and requires two corrections to adjust for the measurement limitations. The first correction takes into account the degradation of energy as the ion beam penetrates the medium: due to the active layer of the RCF being located at different positions to that of the cell monolayer, ions are required to pass through further layers of varying thickness/density before reaching the active film layer. The second correction is due to the variation of the dose response of the RCF films with ion energy and LET, as reported by Kirby et al. 23 . Protons delivered in the single ultra-short pulses had variable energy spectrum and for this study we used 10 MeV protons as the flux of protons at this energy was relevant to deliver dose close to 1 Gy which could be compared to the previous data we obtained for cyclotron accelerated protons as well as X-rays.
Protons, along with high LET particles, are well-reported for inducing clustered DNA lesions 24 which can be visualized using persistent γ-H2AX or 53BP1 foci 25,26 . The foci repair kinetics curves display both the fast and slow components of the repair which predominantly describe the nature of the DSB lesion complexity. These ionizing-radiation-induced 53BP1 foci are not only mere indicators of DSB, but they are also reported as the local DNA repair centers where the damaged chromatin is processed for repair 25 . Based on the complexity of the breaks, the foci may disappear fast or persist for longer times such as upto 72 hours post ions exposure as reported by Grosser et al. 20 In our experiment we observed non-significant variations in the 53BP1 foci induction and repair upto 1 hour post-irradiation with X-rays and protons, in line with previous studies 17,27 . At 24 hours post-irradiation with laser accelerated protons a slight increase in 53BP1 foci with respect to X-rays was observed which is however non-significant.
We used asynchronous cells where the cells across the distribution may not be in the same phase of cell cycle. The radiation response of an asynchronous cell culture may be heterogeneous and averaging the DNA DSB foci number may obscure any cell to cell variations as suggested by Gruel et al. 28 . For this reason, we measured the sub-population radiosensitivity or foci per cell distribution, as shown in  29 . It should be noted that cell inactivation RBE and relative foci induction, may not be directly related to each other, as clonogenic cell death is a complex physiological process involving the multiple processes in a cell which lead to cell death. However, despite the existence of variations between the relative foci induction (RFI) and cell killing RBE, the former can still be used as a surrogate of relative effectiveness. In our study some differences in RFI were noticed for the initial time points and at 24 hours post-irradiation.
Furthermore, the amount of the residual foci remaining at 24 hours and the size of the foci showed no statistically significant difference between 225 kVp X-rays and 10 MeV protons. The foci size appeared similar between www.nature.com/scientificreports www.nature.com/scientificreports/ X-rays and laser-accelerated protons at 24 hours, with the size of foci increasing significantly with respect to 30 minutes size for both the X-rays and protons. This can be understood on the basis that the majority of the DSBs at earlier time-points consisted of both indirect and direct damage, while over time most of the indirect damages are repaired and the lesions formed by the direct interaction of the protons or X-rays persist longer. For foci size scoring at 24 hours, these persistent foci were the only ones contributing to the observation shown in Fig. 6b. This is supportive of the work performed by Costes et al. 30 , who measured changes in foci size over a 24 hour period post-irradiation and found an increase in size with time for high LET irradiation. Ibanez et al., also observed an increase in foci size for up to 6 hours for both lithium and protons at the Bragg peak 26 . In the experimental results presented here, although time dependent variations in the foci size were observed, these were not statistically significant which could be attributed to the similar LET values of X-rays (~2 keV/µm) and 10 MeV protons (~4.6 keV/µm).
Ion tracks are the main biophysical parameters to model radiation quality effects and predict normal tissue complication probabilities in treatment planning algorithms [31][32][33][34] . Track structure leads to clustering of DNA damage events comprising of single strand breaks, base damage, double strand breaks etc. within a few base pairs of DNA as clustered DNA damage 25,35 . Increasing LET induces more repair-refractory clustered damage further increasing the RBE of a particular radiation type 36 . Foci per track calculations are used to model the DNA DSB damage response and here we have used this approach to cross validate the dose of laser-accelerated protons delivered to the cells given that laser-accelerated proton dosimetry is still a developing area. Using conventionally accelerated proton beams we plotted the average foci per track as a function of LET for foci induction at 30 minutes and 24 hours and found a linear relationship between the LET and foci per track.
As shown in Fig. 7, the foci per track values induced by the laser-accelerated protons showed a close correlation to the data obtained with conventional cyclotron-accelerated protons for the initial or residual DSBs. The ratio of the foci per track at 30 minutes to 24 hours showed small differences between the laser-accelerated protons and conventionally accelerated protons possibly indicating the impact of ultra-high dose rate delivery on the repair of the DNA DSBs. Similar observations for the DNA DSB repair process with laser and conventionally accelerated protons were also reported, using fractionated dose delivery, by Raschke et al. 12 , who however did not comment on the foci per track values or the ratio of the foci per track for the initial and residual DSBs.

Conclusion
AG01522B cells were irradiated with laser-accelerated protons in single pulses at ultra-high dose rates of the order of 10 9 Gy/s. The induction and distribution of radiation-induced foci was measured over a 24 hour period giving a preliminary RFI for protons of 2.9 ± 0.5 at 24 hours with X-rays used as reference radiation. The residual component remaining at 24 hours and the size of the foci showed non-significant variations between 225 kVp X-rays and 10 MeV protons. Foci per track per cell analysis revealed a close correlation between the foci induced by laser-accelerated protons and cyclotron accelerated protons, broadly supporting the findings reported in previous work.

Methods
Cell culture and handling. AG01522B cells were maintained in α-modified Minimum Essential Medium (MEM) (Sigma Merck,) supplemented with 20% Fetal Bovine Serum (FBS) and 1% penicillin-streptomycin (Gibco, Life Technologies Carlsbad, CA, USA). All cells were incubated in 5% CO 2 with 95% humidity at 37 °C. For this experiment 80-90% confluent T175 flasks were completely filled with warm low serum (2.5%) medium, then sealed and packed in polystyrene foam containers and shipped to the Gemini laser system at the Rutherford Appleton Laboratory, STFC, United Kingdom. Upon arrival at the facility, low serum medium was replaced with regular full growth medium and the flasks were incubated for at least 24 hours to allow cells to recover from any stress induced during transportation. set up and Beam characterization. Cells were irradiated at the Rutherford Appleton Laboratory, Science and Technology Facilities Council, United Kingdom using the Gemini laser system which is able to deliver ~12 J in a single shot at a pulse duration and central wavelength of 45 femtoseconds and 0.8 µm, respectively. The laser pulse was reflected off a double plasma mirror to enhance the temporal contrast of the laser. The total throughput of the double plasma mirror was ~50%, reducing the laser pulse energy on target to about 6 J. An f/2 off-axis parabola was used to focus the linearly polarized laser pulse onto a 25 nm amorphous carbon target at normal incidence, yielding an intensity of ~6 × 10 20 Wcm −2 . The Ions (protons and carbon) were emitted from the rear surface of the target and spatially selected with a rectangular aperture slit (W × H = 900 µm × 400 µm) located 37 mm behind the target. A dipole magnet (100 mm long, with magnetic field strength of 0.90 T) placed behind the aperture, was used both to deflect the protons away from background radiation (X-rays, electrons) and disperse the particles at the cell plane according to their energy. The dispersed protons and carbon ions exited the vacuum chamber through a 50 µm kapton window positioned 50 mm away from the dipole output and irradiated the cells. The laser driven ion beam was characterized using a Thomson Parabola Spectrometer 37,38 , which was obtained inserting two parallel electric plates between the dipole and the kapton window. After the beam was characterized, and protons spectra were known, the electric plates were removed. A 5.5 µm thick aluminium foil was used in front of the cell dish to filter out the scattered light avoiding RCF film overexposure. Low energy carbon ions coming from the target were filtered out from the cell irradiation region of interest by the several layers of materials interposed in front of the cells monolayer plane.
Dosimetry and Irradiation procedure. Cells were plated 24 hours before irradiation at a density of 3 × 10 5 per dish on 3 µm thin Mylar mounted on a customized cell dish (as shown in Fig. 1) pre-sterilized using 1 kGy dose of X-rays. Fresh medium was added to the cell monolayer and the cell dish was sealed with another 3 µm www.nature.com/scientificreports www.nature.com/scientificreports/ thin Mylar sheet held by a stainless steel ring, to prevent the cells from drying during the transit and irradiation procedure. For the irradiation, the cell dish was placed vertically and in air after the kapton window, with the cells at a distance of 24 mm from the kapton. The cell dish was mounted vertically on a holder which allowed to place the cell plane always at the same location with approximately ~100 µm accuracy. EBT2 radiochromic film was placed immediately behind and in contact with the 3 µm Mylar on which the cells were attached (i.e. cell plane) to measure the dose deposited in the cells. The EBT2 film was calibrated with the MC40 Cyclotron at the University of Birmingham using a monochromatic proton beam of 29 MeV.
The energy broadband TNSA beam resulted in a large region of the cells exposed to a wide range of proton energies dispersed vertically at the plane of the cells. The mean energy of protons considered for cell irradiation analysis was around 10 MeV and the corresponding LET value was about 4.6 keV/µm. Due to the short range of penetration in water of 10 MeV protons (i.e. ~1.2 mm) the culture media was removed (before each irradiation) from the dish using an automated and sterilized pump system at a slow flow rate of 0.2 ml/s, leaving only a thin film of medium on the cells 9 . Furthermore, the vertical dimension of the aperture (i.e. along the proton energy dispersion axis) used for proton energy selection and the size of the analyzed cell region of 10 mm × 1 mm (H × V) centered at 10 MeV proton energy give a proton energy range of 10 ± 1.1 MeV, that corresponds to a LET spread of 4.6 ± 0.4 keV/µm. Due to this energy spread and some inhomogeneity in the proton spatial distribution, the dose value to which cells are exposed to was obtained with an uncertainty of ≈15%.
The AG01522 cells were irradiated in a single shot with protons, delivering doses ranging from approximately 1 to 2 Gy. The time-of-flight (ToF) of protons from the target to the cell plane, the spread of proton energy along the dispersion axis due to the dipole and the size of the aperture all contribute to determining the ion pulse duration at the cell plane; in our system, 10 MeV protons delivered around 1 Gy dose in a single pulse of nanosecond duration corresponding to a dose rate of 10 9 Gy/s. Following irradiation, fresh cell culture media was added back to the cell dish and the cell dish was placed back into the incubator until time of fixation. Reference X-rays induced DSB damage kinetics was obtained by irradiating similar passage cells with 225 kVp X-rays using an XRAD225 X-ray cabinet (Precision X-ray Inc. N. Branford, CT, USA) fitted with a 2 mm copper filter, at the Public Health England (Chilton, Oxford) radiation facility. 53BP1 foci formation data used in Fig. 7. Data for the cyclotron accelerated protons at similar energies to the laser accelerated protons was taken from our previously published paper 17 where the proton beam of 60 MeV was generated at the Douglas Cyclotron of the Clatterbridge Centre for Oncology, Wirral, Liverpool, UK and was degraded using PMMA beam degraders to obtain 10 MeV energy. 53BP1 foci formation assay. 53BP1 foci formation assay was carried out following the method described in 17 with slight modifications. Briefly, after irradiation and incubation for the stipulated time intervals, cells were washed in cold phosphate buffered saline (PBS) and fixed in a 1:1 solution of methanol and acetone (Sigma Aldrich, St Louis, MO, USA) at ~4 °C for 10 minutes. Fixed samples were stored in PBS at 4 °C until stained. For staining, cells were washed with cold PBS and permeabilized in chilled methanol, washed then blocked 10% goat serum and 0.2% triton X-100 in PBS, for 1 hour at room temperature. The cells were then probed with 53BP1 primary antibody (Novus Biologicals, Littleton, CO, USA) at a dilution of 1:1000 in blocking buffer for 1 hour at 37 °C. Subsequently the cells were washed, probed with secondary Alexa Flour ® 488 conjugated secondary antibody at a dilution of 1:1000 and counterstained with prolong gold anti-fade reagent containing DAPI, after curing overnight the Mylar was incised from each dish and mounted on regular glass slides, sealed with nail paint and stored in −20 °C until scored.
Image acquisition and Foci quantification. Images were acquired using a Carl Zeiss Axiovert 200 M Fluorescence Microscope (Carl Zeiss, Germany) with a 63X magnification oil immersion objective (numerical aperture of 1.4). The samples were imaged in 122 µm × 139 µm steps along the region exposed to 10 MeV protons and the number of 53BP1 foci were scored manually using ImageJ 39 inside the field encompassing the 10 MeV energy. For each data point at least 50-100 cells were scored randomly in two independent sets of images for each time point. 53BP1 foci were scored in the nucleus of each cell in the field of view and the results were expressed as average foci per cell per Gy with error bars representing the standard error of the mean of two independent replicates.