Quantum mechanics descriptors in a nano-QSAR model to predict metal oxide nanoparticles toxicity in human keratinous cells

The production of nanomaterials for biomedical research and applications increases exponentially. Interestingly, there is an increase in the use of nanoparticles in pharmaceutical sciences for diagnosis and treatment purposes, and therefore, nano-toxicity becomes one of the major role aspects in the future of pharmaceutical nanotechnology. This study focused on discerning and identifying the main variables that govern a group of metal oxide nanoparticles’ toxicity in human keratinous cells (HaCaT), combining computational simulation and semiempirical calculations with the available experimental data allowed revealing and explaining the nanoparticle toxicity for the corresponding cell line, through the development and validation of an interpretive nano-QSAR model with acceptable statistical quality by applying a multivariate linear regression with a coupled genetic algorithm. This function included only two descriptors, orthogonal to each other: the enthalpy of a standard formation of metal oxide nanocluster $$ {\Delta \mathrm{H}}_{\mathrm{f}}^{\mathrm{c}} $$ and the absolute value of Fermi energy from the cluster $$ {\upepsilon}_{\mathrm{Fermi}}^{\mathrm{c}} $$ .The values of statistical indices obtained for this model showed its quality and robustness, for example, R2 = 0.90; $$ {\mathrm{Q}}_{\mathrm{cv}}^2 $$ = 0.86 and F = 37.15. This study demonstrated the need to use quantum-mechanical descriptors to explain the toxicity of metal oxide nanoparticles, capable of characterizing the electronic state of nanostructures. Regularization methods based on LASSO and Ridge regression have been employed in the model selection and validation. Furthermore, we propose a mechanism for toxicological effects applicable to a relevant group of nanoparticles, as well as their generalization to other toxicity studies not available in the literature, with potential nanopharmaceutical applications.