Protein corona formation is critical for the design of ideal and safe nanoparticles (NPs) for nanomedicine, biosensing, organ targeting, and other applications, but methods to quantitatively predict the formation of the protein corona, especially for functional compositions, remain unavailable. The traditional linear regression model performs poorly for the protein corona, as measured by R2 (less than 0.40). Here, the performance with R2 over 0.75 in the prediction of the protein corona was achieved by integrating a machine learning model and meta-analysis. NPs without modification and surface modification were identified as the two most important factors determining protein corona formation. According to experimental verification, the functional protein compositions (e.g., immune proteins, complement proteins, and apolipoproteins) in complex coronas were precisely predicted with good R2 (most over 0.80). Moreover, the method successfully predicted the cellular recognition (e.g., cellular uptake by macrophages and cytokine release) mediated by functional corona proteins. This workflow provides a method to accurately and quantitatively predict the functional composition of the protein corona that determines cellular recognition and nanotoxicity to guide the synthesis and applications of a wide range of NPs by overcoming limitations and uncertainty.

The development of machine learning provides solutions for predicting the complicated immune responses and pharmacokinetics of nanoparticles (NPs) in vivo. However, highly heterogeneous data in NP studies remain challenging because of the low interpretability of machine learning. Here, we propose a tree-based random forest feature importance and feature interaction network analysis framework (TBRFA) and accurately predict the pulmonary immune responses and lung burden of NPs, with the correlation coefficient of all training sets >0.9 and half of the test sets >0.75. This framework overcomes the feature importance bias brought by small datasets through a multiway importance analysis. TBRFA also builds feature interaction networks, boosts model interpretability, and reveals hidden interactional factors (e.g., various NP properties and exposure conditions). TBRFA provides guidance for the design and application of ideal NPs and discovers the feature interaction networks that contribute to complex systems with small-size data in various fields.

No abstract

Although increasing attention has been paid to the nanotoxicity of graphene oxide quantum dots (GOQDs) due to their broad range of applications, the persistence and recoverability associated with GOQDs had been widely ignored. Interestingly, stress-response hormesis for algal growth was observed for Chlorella vulgaris as a single-celled model organism. Few physiological parameters, such as algal density, plasmolysis, and levels of reactive oxygen species, exhibited facile recovery. In contrast, the effects on chlorophyll a levels, permeability, and starch grain accumulation exhibited persistent toxicity. In the exposure stage, the downregulation of genes related to unsaturated fatty acid biosynthesis, carotenoid biosynthesis, phenylpropanoid biosynthesis, and binding contributed to toxic effects on photosynthesis. In the recovery stage, downregulation of genes related to the cis-Golgi network, photosystem I, photosynthetic membrane, and thylakoid was linked to the persistence of toxic effects on photosynthesis. The upregulated galactose metabolism and downregulated aminoacyl-tRNA biosynthesis also indicated toxicity persistence in the recovery stage. The downregulation and upregulation of phenylalanine metabolism in the exposure and recovery stages, respectively, reflected the tolerance of the algae to GOQDs. The present study highlights the importance of studying nanotoxicity by elucidation of stress and recovery patterns with metabolomics and transcriptomics.

Elucidation of the relationships between nanoparticle properties and ecotoxicity is a fundamental issue for environmental applications and risk assessment of nanoparticles. However, effective strategies to connect the various properties of nanoparticles with their ecotoxicity remain largely unavailable. Herein, an untargeted metabolic pathway analysis was employed to investigate the environmental risk posed by 10 typical nanoparticles (AgNPs, CuNPs, FeNPs, ZnONPs, SiO 2 NPs, TiO 2 NPs, GO, GOQDs, SWCNTs, and C 60 ) to rice (a staple food for half of the world's population). Downregulation of carbohydrate metabolism and upregulation of amino acid metabolism were the two dominant metabolic effects induced by all tested nanoparticles. Partial leastsquares regression analysis indicated that a zerovalent metal and high specific surface area positively contributed to the downregulation of carbohydrate metabolism, indicating strong abiotic stress. In contrast, the carbon type, the presence of a spherical or sheet shape, and the absence of oxygen functional groups in the nanoparticles positively contributed to the upregulation of amino acid metabolism, indicating adaptation to abiotic stress. Moreover, network relationships among five properties of nanoparticles were established for these metabolic pathways. The results of the present study will aid in the understanding and prediction of environmental risks and in the design of environmentally friendly nanoparticles.