Modeling uptake of nanoparticles in multiple human cells using structure-activity relationships and intercellular uptake correlations.

Biomedical applications of nanoparticles (NPs) are largely dependent on their cellular uptake potential that enables them to reach the specific targets in the body. Experimental determination of cellular uptake of diverse functionalized NPs in different human cell types is tedious, expensive and time intensive, hence compelling for alternative methods. We developed quantitative structure-activity relationship (QSAR) models for predicting uptake of functionalized NPs in multiple cell types in accordance with the OECD guidelines. The decision treeboost QSAR models precisely predicted uptake of 104 NPs in five different cell types yielding high R between experimental and model predicted values in the respective training (>0.966) and test (>0.914) sets. The cross-validation Q values ranged between 0.627 and 0.926. Low RMSE (<0.11) and MAE (<0.09) in test data emphasized for the usefulness of developed models for predicting new NPs, which outperformed the previous reports. Relevant structural features of NPs (modifier) that were responsible and influence the cellular permeability were identified. Here, we also attempted to develop intercellular uptake correlations based quantitative activity-activity relationship (QAAR) models for predicting cellular viability of NPs for all the cell types. The performances of all the 20 developed QAAR models were highly comparable with the QSAR models. The applicability domains of the developed models were defined using leverage method. The proposed QAAR models can be employed for extrapolating activity endpoints of NPs to either of the five cell types when the data for the other cell type are available. The developed models can be used as tools for screening new functionalized NPs for their cell-specific affinities prior to their biomedical applications.