Deciphering key nano-bio interface descriptors to predict nanoparticle-induced lung fibrosis.

Abstract

Background: The advancement of nanotechnology underscores the imperative need for establishing in silico predictive models to assess safety, particularly in the context of chronic respiratory afflictions such as lung fibrosis, a pathogenic transformation that is irreversible. While the compilation of predictive descriptors is pivotal for in silico model development, key features specifically tailored for predicting lung fibrosis remain elusive. This study aimed to uncover the essential predictive descriptors governing nanoparticle-induced pulmonary fibrosis.

Methods: We conducted a comprehensive analysis of the trajectory of metal oxide nanoparticles (MeONPs) within pulmonary systems. Two biological media (simulated lung fluid and phagolysosomal simulated fluid) and two cell lines (macrophages and epithelial cells) were meticulously chosen to scrutinize MeONP behaviors. Their interactions with MeONPs, also referred to as nano-bio interactions, can lead to alterations in the properties of the MeONPs as well as specific cellular responses. Physicochemical properties of MeONPs were assessed in biological media. The impact of MeONPs on cell membranes, lysosomes, mitochondria, and cytoplasmic components was evaluated using fluorescent probes, colorimetric enzyme substrates, and ELISA. The fibrogenic potential of MeONPs in mouse lungs was assessed by examining collagen deposition and growth factor release. Random forest classification was employed for analyzing in chemico, in vitro and in vivo data to identify predictive descriptors.

Results: The nano-bio interactions induced diverse changes in the 4 characteristics of MeONPs and had variable effects on the 14 cellular functions, which were quantitatively evaluated in chemico and in vitro. Among these 18 quantitative features, seven features were found to play key roles in predicting the pro-fibrogenic potential of MeONPs. Notably, IL-1β was identified as the most important feature, contributing 27.8% to the model's prediction. Mitochondrial activity (specifically NADH levels) in macrophages followed closely with a contribution of 17.6%. The remaining five key features include TGF-β1 release and NADH levels in epithelial cells, dissolution in lysosomal simulated fluids, zeta potential, and the hydrodynamic size of MeONPs.

Conclusions: The pro-fibrogenic potential of MeONPs can be predicted by combination of key features at nano-bio interfaces, simulating their behavior and interactions within the lung environment. Among the 18 quantitative features, a combination of seven in chemico and in vitro descriptors could be leveraged to predict lung fibrosis in animals. Our findings offer crucial insights for developing in silico predictive models for nano-induced pulmonary fibrosis.