Cell Culture for Chromosome Aberration Test

In the experiments, the A_L cell line was used, which has been established in our laboratory at the Technische Universität München (TUM) to investigate structural or functional CA after exposure to ionizing or non-ionizing radiation. This cell line was originally derived from the Chinese hamster ovary-K1 (CHO-K1) wild-type cell line, a subclone from an adult Chinese hamster (Cricetulus griseus). The A_L cells contain a standard set of CHO-K1 chromosomes plus a single human chromosome 11. Our previous chromosome analysis showed a modal number of 22 chromosomes (85% of analyzed cells); 15% of the analyzed cells had a loss or gain of chromosomes (18). The cells were grown as monolayer cultures in RPMI-1640 medium (Sigma Aldrich), supplemented with 16% fetal calf serum and antibiotics (100 units penicillin and 100 µg streptomycin per ml culture medium; Sigma Aldrich). The cell monolayers were maintained at 37°C in a humidified atmosphere of 5% CO₂ in air. Under these growth conditions, the A_L cells had a doubling time of 18 h. Special irradiation containers were used for all types of irradiations to fit the needs of the particle irradiation of cells at SNAKE. These cell containers were constructed in such a way that a Mylar foil (6 µm thick) was stretched and clamped between two steel plates. The design of the irradiation containers has been described previously (19). Approximately 24 h before exposure, cells were seeded on mylar foil pre-treated with Cell-TAK (Corning). Just before irradiation, the monolayer of cells was washed to remove any unattached cells.

Cell Culture for Clonogenic Cell Survival

The cell survival experiments were performed with the original CHO-K1 cell line, which is well established for colony forming assay (5). Cells were cultivated in RPMI-1640 medium (Sigma Aldrich) supplemented with 10% fetal calf serum (Sigma Aldrich), 100 units of Penicillin and 100 µg of Streptomycin per ml culture medium, 2 mM L-Glutamine and 1 mM Sodium pyruvate (Sigma Aldrich). The cells were grown as a monolayer culture in an incubator at 37°C, 5% CO₂ and 95% humidity. Under these culture conditions the CHO-K1 cells had doubling time of ∼12 h. Special irradiation containers were used for the particle irradiation of cells at SNAKE, as described above. For cell survival experiments, it is essential that all seeded cells are exposed, since unirradiated cells would alter the experimental outcome. For this purpose, the cell cultivation setup was modified and adapted for the application in microbeam irradiation experiments as described in detail by Greubel et al. (20).

Briefly, approximately 4 h before irradiation cells were seeded on mylar foil pre-coated with Cell-TAK (Corning) in an area at the center defined by an O-ring (4-mm diameter) (5). The total cell growth area was about 15 mm², which ensured a sufficient number of cells (∼10⁴ cells) after irradiation and reseeding to analyze clonogenic cell survival. Directly before irradiation, the O-ring and its support were removed and cells were washed to get rid of all unattached cells.

Cell Irradiation with 20 MeV Protons

Ion irradiations were performed at the Scanning Ion Microprobe SNAKE installed at the Munich 14 MV tandem accelerator (21). Recent developments at SNAKE allowed focusing of particles to a minimum spot size of about 0.5 × 1.0 µm² with a particle rate of 200–500 kHz for protons (20). The protons were focused and transported in vacuum to the beam exit nozzle, which was covered with 7.5-µm thick Kapton foil. After passing the Kapton foil, the ions traversed a 6-µm thin Mylar foil before interacting with the cells. Behind the cell sample, a detector, as described by Hauptner et al. (16), was placed to count the ions. With this setup and an additional fast high-voltage beam switch, it is possible to irradiate with a known number of ions impinging a single spot. Using electrostatic beam scanning, single spots in a square field of 500 × 500 µm² were irradiated with a certain number of ions. The cell container was mounted on the XY stage of an inverted optical microscope (Zeiss Axiovert 200 M) (17) and by mechanically moving the cell sample, the square fields were stitched together to cover the entire area of cell growth. For a quasi-random proton irradiation, the microbeam was scanned over an area of 500 × 500 µm² with a scan rate much higher than the count rate, and protons were counted until the desired number of particles, and therefore the desired dose, was reached.

Before irradiation, about half of the culture medium was removed and the irradiation containers were tightly closed by clamping a second 6 µm Mylar foil. The exposure took place at room temperature (about 22°C). During irradiation, the container with the cell monolayer was positioned vertically in front of the beam exit nozzle with cells facing the microbeam. In this position, cells were not covered by culture medium, but the medium in the cell container prevented cells from drying out by ensuring a saturated atmosphere (16, 22).

Typically, the total time required for sample preparation and exposure was less than 25 min, including about 15 min in the upright position. Under these conditions, there were no differences in the number of observed chromosome aberrations in non-irradiated cells and cell samples remaining in the incubator (19). For cell survival, a sham-irradiated sample was used as control (where the sample was placed in the SNAKE setup for 15 min without switching on the beam) to exclude uncertainties in the data coming from the treatment in upright position.

Two independent experiments for chromosome aberrations (experiments I and II) and three for cell survival (experiments III, IV and V) were performed to demonstrate reproducibility as well as inter-test variability. Each irradiation was done using 20 MeV protons (LET = 2.6 keV/µm) not targeted to single cells but in a regular pattern of 5.4 × 5.4 µm spot distance for matrix irradiation using the same average dose of 1.7 Gy, which leads to 117 protons per point. Given the size of a CHO cell nucleus of 70 µm², the irradiation with a regular pattern of this size results in an average 2.4 ± 0.8 irradiated spots per nucleus (20). For the AL cells, a size of 117 µm² gives an average of 4.0 ± 0.2 hits per nucleus. A schematic image shows how the hits are distributed over the cells in   Supplementary Fig. S1 (10.1667_RADE-23-00217.1.S1.pdf)² ( https://doi.org/10.1667/RADE-23-00217.1.S1). A histogram showing the fraction of cells with various doses are shown in   Supplementary Fig. S2 (10.1667_RADE-23-00217.1.S1.pdf) ( https://doi.org/10.1667/RADE-23-00217.1.S2). Homogeneous dose was achieved by applying on average 4 ions per µm² in the quasi-homogeneous scan mode.

Spot irradiation was performed with three different spot sizes. The smallest beam spot was achieved by the best possible focusing, which varies for different beam times and also locally on the radiation field, due to external influences such as chromatic aberrations of the lens, beam position and electromagnetic interfering fields, to a full width at half maximum of 1.0 × 0.2 µm for colony forming and 0.78 × 0.01 µm for dicentric chromosomes, with the same dimensions in the x and y direction. The sample was placed in contact with the beam exit window to ensure irradiation with this small beam spot. The other two spot sizes were achieved by exploiting the scattering of the beam in the Kapton window. By moving the sample away from the beam exit at a certain distance, beam sizes of 1.80 × 0.15 µm and 2.80 × 0.18 µm for cell survival and 1.60 × 0.02 µm and 2.70 × 0.03 µm for CA were achieved. The difference in the spot sizes for cell survival and CA originate from the different minimal beam sizes, as the distance from the sample to the beam exit was kept the same. Beam spot size was measured as explained by Greubel et al. (20) using fluorescent nuclear track detectors (FNTDs). Briefly, 500 × 500 µm² fields with 5.4 × 5.4 µm spot distance were irradiated using the three different beam spot sizes inducing fluorescent color centers inside the FNTD crystals. After irradiation, the FNTDs were imaged using a Leica TCS SP8 3X confocal microscope with excitation at 635 nm and detection of fluorescent light between 645–740 nm and a pixel size of 50 × 50 nm² with a pixel dwell time of 24 µs and a resolution of approximately 200–250 nm. A Leica HCX PL APO100x/1.4 Oil objective with 100x magnification and numerical aperture of 1.4 was used. The spot size is determined by deconvolving the measured spots with the single ion response as described in detail by Greubel et al. (20). For size determination of the whole beam, nine regions from the field each containing 5 × 5 spots are analyzed, one from each corner, the center of the field and one in the middle of each side of the field. This is done to ensure coverage of all possible beam variations in x- and y-direction while keeping the effort practical. All 225 spots were then fitted using a two-dimensional Gaussian distribution with the full width at half-maxima representing the x- and y- beam spot size, where the y beam spot is ∼25% larger than the x beam spot. From this data, the mean beam size in each direction is achieved from these two values, and the mean spot diameter is calculated. The provided error represents the standard deviation.

The reference X-ray irradiation was performed in the same cell dishes, with 200 kV X rays for cell survival (5) and 70 kV X rays for CA (14), as described before.

After Irradiation Culture and Chromosome Analysis

The cell culture conditions have been described in detail previously (12). The cells were trypsinized and reseeded in 4 ml RPMI-1640 medium supplemented with 20% fetal calf serum and antibiotics (penicillin/streptomycin) and incubated at 37°C in a humidified atmosphere of 5% CO₂ in air. Colcemid (Life Technologies Darmstadt, Germany) (0.1 µg/ml) was present in the cultures during the last 4 h of the total incubation time of 20 h. Chromosome preparation and Giemsa staining were carried out according to our standardized laboratory procedure originally described for human lymphocytes (23). All object slides were coded to ensure blinded evaluation. Only complete cells containing the modal number of 22 chromosomes were analyzed for dicentrics. Tricentric chromosomes were counted as two dicentrics, and similarly chromosomes with N > 3 centromeres, if present, were counted as N-1 dicentrics. Since the measured dicentric yields are generally used as the most characteristic and quantitatively reproducible biological endpoint of radiation-induced chromosome-type aberrations, only the data for dicentrics were used in the quantitative analysis, although the yields of centric rings and acentric fragments were also determined.

Survival Assay

Immediately after irradiation, the cells were disassociated from the mylar foil by trypsinization, and cell concentration from each sample was determined using the Fuchs-Rosenthal cell counting chamber (NanoEnTek). Cell counting was performed four times for each cell sample and the cell suspension was diluted in a dose-specific manner and reseeded into 12-well plates (Corning). In the case of the unirradiated (sham) samples, 100 cells were plated per well, while for the irradiated cell samples, 200 cells were reseeded per well. For each sample, cells were reseeded into wells of two to three 12-well plates, depending on the determined cell concentration. The plates were incubated under standard culture conditions until the cells had formed sufficiently large colonies, which consist of at least 50 cells (24). After five days, the medium was removed and cells were rinsed once with PBS. The colonies were fixed with methanol and stained with 0.1% crystal violet. The counting of stained colonies was performed by an automatic counting bioreader (BIO-SYS GmbH). In each experiment 5 replicates were used, and the focused data consist of 3 independent experiments, whereas for the homogeneous irradiation, five experiments were used.

Statistical Analysis of the Data

Dicentric Data Analysis. To test if the values for different spot sizes differ significantly, a statistical analysis is performed. A null hypothesis that the observations of the mean frequencies of dicentrics (±SE) determined in the independently performed experiments are not different was tested according to the z-test statistics. The intercellular distribution of dicentrics was tested for deviation from a Poisson distribution: A test quantity u has been used, which is related to the numbers of cells analyzed (N) and dicentrics scored (n) and the dispersion ratio (variance σ² over mean y = n/N) as described by Schmid et al. (25).

For a Poisson distribution, σ² = y holds and hence u = 0. Values of u more than 1.96 indicate over-dispersion at the 5% level of significance.

Cell Survival Data Analysis. Cell survival data from multiple irradiated samples obtained from three beam times were available. The average effect, i.e., average of the negative logarithms of the survival, was determined from multiple irradiated samples within each beam time. Uncertainties of the effects were calculated as the square root of the sum of effect variances for irradiated and control samples, the latter of which must be included as the control samples are used to calculate the plating efficiency. Thereby, retrieved effects from different beam times were pooled and average values for surviving fractions were calculated (see Table 1). Associated uncertfigainties for surviving fractions were determined by Gaussian error propagation as the square root of the uncertainty square sum. This pooling of data is justified as the spot sizes and all other conditions were comparable in the independent experiments carried out in the different beam times. Hence. error bars and given uncertainty ranges in Table 1 are based on standard deviations and Gaussian error propagation, thus representing approximately 68% confidence intervals.

TABLE 1

Surviving Fractions (±SE) and RBE Values (±SE) for Cell Inactivation in CHO-K1 Cells after Exposure to 1.7 Gy of 20 MeV Protons either in Quasi Homogeneous Exposure Condition (P_hom) or in a Focused Mode Following a 5.4 × 5.4 µm² matrix (P_foc) at Different Beam Spot Sizes

RBE Calculation. In the following calculations, the enhanced effect of the focused radiation is used. The RBE is calculated by identifying the dose of reference radiation necessary for achieving the same biological effect as the 1.7 Gy focused radiation. From this, the RBE is calculated as RBE = D_ref/D_focused, therefore the RBE for each focusing is determined for another level of biological effect.

To calculate the influence of the different spot size on the RBE for dicentrics, our previously established dose-response curve of dicentrics in A_L cells exposed to 70 kV X rays (14) was used. For this reference radiation quality, a weighted least-squares approximation was applied to fit the dicentric data with the linear-quadratic function. The resulting linear-quadratic coefficients (±SE) were α = (0.010 ± 0.003) Gy^–1 and β = (0.0048 ± 0.0012) Gy^–2. The dose-response curve of 200 kV X-ray irradiation of CHO-K1 cells established in a previous study (5) was used as a reference to determine the RBE for cell survival after random and focused proton irradiation. The linear quadratic coefficients α = (0.156 ± 0.045) Gy^–1 and β = (0.0235 ± 0.0055) Gy^–2 and covariance Cov(α, β) = 0.00025 Gy^–3 were evaluated from the fit to the data points.

The reference radiation doses that caused effects equivalent to 1.7 Gy for random and focused proton irradiation of different beam spot sizes were determined by inverting the fitted dose-response curve with –ln(S), giving the effect for dicentrics induction or cell survival:

Associated uncertainties for the doses, ΔD were calculated according to Gaussian error propagation:

Biophysical Effect Modeling

PARTRAC Simulation. PARTRAC (26–30) is a suite of Monte Carlo codes for biophysical modeling of radiation effects at subcellular and cellular levels. Radiation track structures, i.e., energy deposition patterns with single-event resolution, are generated by simulating individual interactions of photons, electrons, protons or ions with the traversed medium, using cross section data based on the work from Dingfelder et al. (31). Production of reactive species, their diffusion and mutual reactions are followed. Initial radiation-induced damage to DNA is scored, considering both direct energy depositions and indirect effects via attacks of reactive species. Multi-scale chromatin models are implemented that range from DNA in atomic resolution over its binding to nucleosomes, formation of chromatin fibers and loops, to chromatin domains and chromosome territories within human cell nuclei. The repair of DNA DSB via the non-homologous end-joining pathway (NHEJ) is modeled, explicitly accounting for both the kinetics of enzymatic processes and the mobility of individual broken ends. Chromosome aberrations and in particular dicentrics that result from misrejoining events are scored.

In this work, irradiation of A_L cells with 20 MeV protons was simulated with PARTRAC under diverse focusing modes as used in the experiments. In quasi-homogeneous irradiation, single protons per point were started in a matrix mode with a small grid size of 0.5 µm; the actual tracks were allowed to deviate from the grid points according to a Gaussian distribution with a comparably large spot size (full width at half-maximum) of 1.41 µm. For the three degrees of focusing, the spot size amounted to 2.70, 1.60 or 0.78 µm, respectively, always with 117 protons per spot and a grid size of 5.4 µm. The chromatin structure of A_L cells was approximated using two copies of chromosomes 1–11 from human lymphocytes (15). Standard PARTRAC modules and parameters were used to simulate particle tracks, reactive species, and induction of initial DNA damage as described previously (27). The model parameters were first optimized for this project as previously published (28). From there, further optimizations were performed (15), which were then used to predict the outcome in this study. Here, we used the NHEJ model (without hypothetical repair centers) to simulate the DNA repair. For each irradiation scenario, 500 runs were simulated, each corresponding to matrix irradiation of a single cell. The induction of dicentrics within 24 h postirradiation was analyzed.

LEM Simulation. The LEM is a simulation tool, which predicts the effect of ionizing radiation based on two properties: the amorphous track structure and the photon dose response curve for the experimental endpoint. The amorphous track structure is the local dose deposited by an ion as a continuous function. reflecting the expected local energy concentration of a large ensemble of particle tracks. The photon dose response is characterized by the α and β values coming from linear-quadratic fitting and a threshold dose, which marks the transition to a linear regime and is set to D_t = 4 Gy + 1.1 α/β. A detailed description of the model approach can be found in the published literature (5, 32). The model was adapted to be able to predict the effect of single Gaussian shaped beam spots in cell nuclei (5) by considering the geometrical properties of the beam and the cell nuclei. The LEM simulations were performed separately for 33 beam sizes from 0.12 µm to 3.96 µm in 0.12 µm steps, assuming a round beam shape.

Dicentric Data

The detailed experimental results for dicentrics and their intercellular distribution in A_L cells are presented in Tables 1 and 2 together with the corresponding background frequencies. The spontaneous frequency in unirradiated cells of 0.0007 ± 0.0003 per cell determined in 4,500 cells (Table 2) is not statistically different from the value of 0.0006 ± 0.0003 dicentrics per cell obtained in 5,312 cells reported earlier (19). No significant differences have been determined between the yields of dicentrics, which were observed in any exposure mode between experiments I and II [P values (one-sided T-test) from 0.30 to 0.90] listed in Table 2. Therefore, the yields of dicentrics from the replicates within the different experiments were pooled. The resulting mean values of dicentrics are plotted in Fig. 1A. The highest degree of focusing to a 0.78 µm beam spot increases the induction of dicentrics almost twofold compared to the quasi-homogeneous case. These enhancement factors decrease with decreasing degree of focusing. The data can also be compared with the data collected in a previous study (14). There, an elliptical spot of a size of 0.5 µm × 1.0 µm was applied, and a dicentrics induction of 0.083 ± 0.008 dicentrics/metaphase was observed. These data are well corresponding with the measurements performed in this study.

TABLE 2

Frequency (±SE) of Chromosome Aberrations and the Intercellular Distribution of Dicentrics in A_L Cells after Exposure to 1.7 Gy of 20 MeV Protons either in Quasi-Homogeneous Exposure Condition (P_hom) or in a Focused Mode Following a 5.4 × 5.4 µm² Matrix (P_foc) At Different Step Numbers

FIG. 1

Panel A: Comparison of PARTRAC simulation (triangles) and experiment (squares) for dicentrics induction at different spot sizes. Mean values representing measuring data from two proton beam times with three replicates per spot size and beam time are shown. For further comparison, data from Schmid et al. (14) are added, which nicely fit the actual experiment. One has to note that in this case an elliptical beam was used, and the spot diameter was calculated as the mean of the x (0.5 µm) and y width (1.0 µm). Error bars for the data points represent the standard errors (SE); Panel B: Relationship between RBE for the induction of dicentric chromosomes and spot size.

The RBE values were calculated using the X-ray doses that produced the same response as 1.7 Gy of the presently applied radiation qualities as explained in the Materials and Methods section. The resulting mean RBE values of dicentrics in experiments I and II are plotted in Fig. 1B. The RBE values are significantly higher when 117 protons were focused to a 0.78 µm spot within a 5.4 × 5.4 µm² matrix compared to homogeneously applied protons (RBE_0.78 = 1.96 ± 0.16 vs. RBE_hom = 1.30 ± 0.16). By doubling the spot size to 1.6 µm, the RBE decreased to RBE_1.6 = 1.52 ± 0.16. By further increasing the spot size to 2.7 µm, the RBE was no longer different (RBE_2.7 = 1.36 ± 0.14) from the homogeneous radiation, yet above unity.

The measured data of dicentric induction were compared to the dicentric yields predicted by PARTRAC simulation (Fig. 1A). The simulation was performed based on the parameter values detailed in a previous study (15) and without further optimization and without knowledge of the present biological results, i.e., only-spot diameters as measured in the experiments and number of particles per spot were known. For homogeneous irradiation, PARTRAC predicts a slightly lower number of dicentrics per metaphase than measured. For the focused geometry, PARTRAC is able to predict the trend that dicentric induction is increasing with increasing degree of focusing, i.e., with decreasing spot size. Between the smallest spot (0.78 µm) and the largest spot (2.7 µm), a reduction in the predicted frequency of dicentrics induction from 0.10 to 0.064 corresponds to a 1.6-fold decrease. Compared to homogeneous proton distribution, this corresponds to a 2.6-fold increase for the smallest spot size. In total, the dicentrics induction for the focused irradiation geometry is systematically overestimated between 20% and 30% by simulation compared to measurement, and slightly underestimated for homogeneous irradiation, while individual data points coincide within the uncertainties except for one.

Cell Survival Data

To calculate the influence of the different spot sizes on the RBE for cell survival, our previously established dose-response curve of cell survival in CHO-K1 cells exposed to 200 kV X rays was used, and the corresponding survival data of two independent experiments are shown elsewhere (5). The cell survival rates after microbeam exposure from three different beam times were pooled and average values for surviving fractions were calculated (Table 1) and plotted in Fig. 2A. The value from exposure of cells to a mean dose of 1.7 Gy of homogeneously applied protons resulted in 78% ± 5% cell survival. The highest degree of proton focusing to a 1.0-µm beam spot reduces the surviving fraction to 54% ± 6%. By increasing the beam spot size to 1.8 µm and 2.8 µm, the surviving fraction was increased to 60% ± 7% and 70% ± 6%, respectively. The data can be compared to data from a previous study (5) with the same setup, where an elliptical beam was used, with width in x-direction of 0.56 ± 0.05 µm and y-direction of 1.1 ± 0.06 µm. For comparison the beam size of 0.81 ± 0.1 µm, was calculated as the mean of the width in x and y direction. Here a surviving fraction of 39% ± 6%, was measured, which shows that for even smaller focusing, cell killing is further enhanced.

FIG. 2

Panel A: Comparison of LEM simulation (triangles) and experiment (squares) for cell survival at different spot sizes. The mean value of homogeneous proton irradiation was obtained from five independent experiments, whereas the mean values of focused proton irradiations were obtained from three independent experiments. In each experiment each spot size contained five replicates. For comparison, the data point of Friedrich et al. (5) for protons and single carbon ions (55 MeV, LET = 338 keV/µm) was added. Error bars for the data points represent the standard errors (SE); Panel B: Relationship between RBE for cell inactivation and spot size.

The RBE values for cell inactivation were calculated using the X-ray doses that produced the same response as 1.7 Gy of the presently applied radiation qualities and the necessary X-ray doses have been determined by inverting the fitted dose-response curve. The resulting mean RBE values for cell survival are plotted in Fig. 2B.

The RBE for homogeneous proton radiation was 0.79 ± 0.2, which was not expected due to the common paradigm of an RBE of close to 1 for proton radiation, but is in agreement with previous results (5). The RBE was enhanced to 1.65 ± 0.25 after focusing protons to the smallest beam spot of 1-µm diameter. The RBE for cell inactivation using wider beam spots of 1.8 µm and 2.8 µm decreased to 1.42 ± 0.30 and 1.05 ± 0.31, respectively. The increased RBE can now be compared to the RBE of single carbon ion hits (RBE = 8.31 ± 0.21). It is notable that the increase of RBE by focusing to 1µm diameter covers approximately 20% of the increased RBE covered by the single ion hits with the same LET.

The measured cell survival data were compared to simulation using the LEM model (Fig. 2A). Notably, the simulation was performed without any specific parameter optimization and without prior knowledge of the biological data, using only the specific information that the spot should be considered round and 117 protons per spot are used. Therefore a true prediction of the results was performed using the model as it was optimized based on the data from Friedrich et al. (5). The spot size was varied between 0.12 µm and 3.96 µm in 0.12 µm steps. For spot sizes below 0.5 µm, the predicted surviving fraction was almost constant at _∼40%. After that, there was a steep increase with increasing spot size up to a 1 µm spot size, and the curve flattened towards even larger spot diameters. At 2-µm spot size, a slight increase of survival fraction was still visible but surviving fractions were already comparable to the results for the simulation of homogeneous irradiation. The simulation data agreed reasonably with the measured surviving fractions and the general trend of the shape and magnitude of the increase. However, for homogeneous irradiation, the LEM simulation tends to underestimate cell survival and for the smallest beam spot size the LEM overestimates the survival fraction compared to the experiments. One has to note, however, that this latter data point originates from a previous study, in which irradiation was performed with an elliptical spot (axes of 0.56 µm and 1.05 µm) rather than a round spot (5).