• biophysics
• DNA damage quantification
• FLASH radiotherapy
• fluorescence microscopy
• single-molecule localization microscopy
• super-resolution microscopy

FLASH radiotherapy is an innovative technique that achieves tumor control comparable to conventional radiotherapy while significantly reducing toxicity in healthy tissues. Despite extensive studies, the biophysical mechanisms underlying this so-called “sparing effect” remain unclear. My research stems from the hypothesis that this difference originates at the molecular level, with a primary role on how DNA structure responds to irradiation, one of the primary ways radiation-induced damage manifests.
Specifically, my thesis develops new quantitative and multiscale methods to investigate the spatial organization of the phosphorylated histone γH2AX, a key protein in DNA damage repair, after irradiation. To this end, I employed single-molecule localization microscopy (SMLM), particularly Stochastic Optical Reconstruction Microscopy (STORM), a super-resolution technique capable of achieving a lateral resolution of 20 nm, far beyond the diffraction limit of visible light (~300 nm). This breakthrough, along with other techniques which earned the 2014 Nobel Prize in Chemistry (Betzig, Hell, Moerner), enables the visualization of molecular-scale structures within cells.
While conventional approaches quantify DNA damage by analyzing γH2AX foci (i.e. spatial accumulations in the proximity of a damage site) in diffraction limited images (~500 nm), this method can overlook structural details. Notably, when comparing samples treated with the same total dose but different irradiation conditions (for example FLASH- and CONV-irradiated samples, which differ for the given dose rate), foci abundance alone often failed to distinguish between them. This motivated my investigation into molecular-scale metrics to quantify DNA damage more comprehensively.
Super-resolution imaging (20 nm resolution) allowed me to resolve the internal architecture of γH2AX foci. Previous studies (2017) identified a hierarchical organization: γH2AX first clusters into nanofoci (~100 nm), which then assemble into foci. I extracted spatial information from STORM data through clustering analysis, obtaining nanofoci information, then DBSCAN was employed to segment foci from nanofoci. This allowed me to define three non-redundant, multiscale metrics that describe γH2AX organization: the abundance of foci, the abundance of nanofoci, and the fraction of nanofoci that cluster into foci. These metrics exhibited a non-linear correlation and varied systematically across irradiated nuclei.
When visualized in the 3D space defined by these metrics, nuclei from different conditions aligned along a common trend. To derive a quantitative measure of DNA damage, I fitted the nuclei’s response to a simple empirical model, capturing the trend of these multiscale metrics. The model's curve length was parameterized, and the extracted parameter (B), defined as the arc-length of the curve from the origin point to the position of the median of each experimental group (i.e., cells from a fixed cell line irradiated with different doses or dose rates), served as a one-dimensional damage quantifier. This approach successfully distinguished between FLASH- and CONV-irradiated lung cancer cells at the same dose as well as highlighted the effect of spatially fractionated RT.
Specifically, I further applied this method to minibeam-irradiated samples, where dose deposition follows a spatial pattern. To ensure accurate sampling of regions corresponding to different dose levels, I developed a methodological pipeline for spatial navigation during the imaging phase. This allowed me to correlate the acquired data with the irradiation pattern, ensuring that measurements were taken at specific locations within the patterned irradiation. The results demonstrated a strong correlation between dose distribution and damage localization.
Beyond the specific applications presented in this thesis, the method developed here represents a versatile and quantitative tool for investigating the effects of ionizing radiation at the nanoscale. Future directions of this work could involve theoretical biophysical modeling of DNA structures, potentially elucidating whether the FLASH effect indeed arises from molecular-scale differences, thus providing deeper insight into its underlying mechanisms.