Monte Carlo Simulation of Proton Radiation Induced DNA Double Strand Breaks Bench Marked on AFM Measured DNA Fragment Size Distributions
Abstract
Purpose
To simulate, using Geant4-DNA, proton radiation induced plasmid DNA fragmentation in the plateau as well as Bragg peak regions in a cell-free, aqueous environment of different free radical scavenger concentrations. Atomic Force Microscopy (AFM) measured DNA fragment size distributions are used to benchmark and validate the modeling.
Methods
pUC19 plasmid DNA was irradiated with 110 MeV proton beams in the Bragg peak and plateau regions at doses of 1 and 3 kGy in buffers containing 2 and 10 mM HEPES. AFM images of the irradiated DNA samples were used for fragmentation and DSB yields measurement. Corresponding Geant4-DNA simulations of 1700 plasmid in 10×10×10 µm³ volume were performed with full physical and chemical stage modeling using scavenger-dependent hydroxyl radical reaction rates. Simulation results were compared with AFM measurements.
Results
MC simulations showed increased fragmentation with dose and reduced fragmentation with HEPES, with significantly higher fragmentation at the Bragg peak than the plateau, but quantitative discrepancies with experiments remained at both dose levels. At 1 kGy, experimental fragmentation was 10-16% higher than predicted, with both DSB/DNA (0.70-0.98 vs. 0.19-0.35) and DSB/broken-DNA (2.46-2.60 vs. 1.13-1.27) significantly exceeding simulated values. At 3 kGy (2 mM-HEPES), AFM revealed 85-91% fragmented plasmids, while simulations predicted only 58-62%. Here, experimental DSB/DNA and DSB/broken-DNA were higher than simulations (2.11±0.06 vs. 0.93±0.05 and 2.46±0.06 vs. 1.61±0.03). Furthermore, length-resolved distributions indicated that simulations overrepresented very short (0-50 nm) and near-full-length (900-950 nm) fragments, whereas intermediate-length fragments were more prevalent in the experimental data.
Conclusion
While Geant4-DNA captures qualitative damage trends, quantitative gaps in DSB yield and fragmentation persist at high doses. These differences reflect sensitivity to endpoint definitions, damage clustering thresholds, and experimental effects such as lesion-to-break conversion and fragment loss. Consistent endpoint definitions and detection modeling are essential for model improvement and quantitative validation of Geant4-DNA MC simulations.