Modelling Genotoxic Effects of Metal Oxide Nanoparticles Using Qsar Approach

Abstract

We investigated the application of structure-activity relationship approaches to underpin structural properties that potentially control the genotoxic potential of 9 different metal oxide nanoparticles (CuO, ZnO, NiO, SiO2, TiO2, CeO2, Fe2O3, Fe3O4 and Co3O4). In particular, we compiled a pool of quantum-mechanical, experimental and periodic table-driven descriptors and explored their distinctive contribution to the measured activity (genotoxicity). We first employed a clustered heatmap and parallel coordinates plot for visual exploration of the clusters and outliers of the data and finding corresponding responsible physicochemical descriptors. We then investigated the strength (and direction) of the relationship among descriptors and between descriptors and genotoxicity using similarity metrics. By using orthogonal projections to latent structures (OPLS), we were able to quantify the relative contribution of each descriptor to the genotoxicity of metal oxide nanoparticles. Our results suggested that zeta potential, the ratio of core electrons to valence electrons, Fermi energy and electronegativity were significant predictors of genotoxicity. Such computer-assisted approaches hold considerable promise for maximizing the use of accumulated data in nanotoxicology, prioritizing nanoparticles for further testing and filling data gaps required for hazard assessment processes.