Computer-aided nanotoxicology: risk assessment of metal oxide nanoparticles via nano-QSAR

Abstract

Since the application of nanomaterials has expanded as a therapeutic methodology, the cytotoxicity accompanying nanomaterials during the therapeutic process has attracted significant attention. Thus, it is necessary to investigate the relationship between the structure and cytotoxicity of different nanomaterials. To avoid case-by-case testing, intelligent strategies combining testing methods with non-testing predictive modelling should be developed. The quantitative structure–activity relationship (QSAR) is a promising tool in understanding the properties that affect the potency of metal oxide (MeOx) nanoparticles (NPs) and in predicting toxic responses. In this work, a combined experimental and computational study was performed to estimate the acute cytotoxicity and develop predictive models for MeOx NPs. Improved SMILES-based optimal descriptors were applied to describe the nanostructure characteristics of MeOx NPs. Based on the experimental test, four nano-QSAR models were established for predicting the median lethal concentration (LC₅₀) of MeOx NPs to human lung adenocarcinoma (A549) cells by considering the influence of their particle size and zeta potential on cytotoxicity. The models showed satisfactory predictivity and robustness. The predominant nanostructure characteristics and the mechanism responsible for the cytotoxicity of MeOx NPs to A549 cells were identified through model interpretation with subsequent experiments on the reactive oxygen species (ROS) of MeOx NPs. The proposed models can reliably predict and assess the acute cytotoxicity of novel NPs solely from their nanostructures, and provide guidance for prioritising the design, synthesis, and manufacture of safer and green nanomaterials with expected properties.