Physicochemical and cell-based approach for early screening of phospholipidosis-inducing potential

Some of the principal requisites of toxicity screening methods in drug discovery are their ease to perform and high throughput, as well as the possibility to predict the occurrence of clinical events. Phospholipidosis is one of the toxicities often induced by potential drugs. Several physicochemical methods for the prediction of phospholipidosis have been reported. The purpose of the present study was to examine the predictability of methods based on lipophilicity and charge parameters. We employed a test set of 33 compounds including 11 in-house compounds. The phospholipidosis-inducing potential (PLIP) of the test set compounds was determined by the fluorescence-labeled lipid accumulation assay using isolated rat hepatocytes. This assay was verified by transmission electron microscopy (EM). The usefulness of the ClogP - most basic pK(a) (pK(a) -MB) plot to the PLIP of compounds was examined. This plot was unable to predict the PLIP of zwitterions. In order to improve its predictability, the net charge of a given molecule (NC) was introduced instead of pK(a) - MB, since the NC corresponds directly to the ionization state of compounds in the organelles. Compounds with high ClogP (> 1) and high NC (1< or =NC< or =2) tended to be positive. This finding was also confirmed using 30 additional validation set compounds obtained from the literature. The ClogP - NC plot differentiated positive and negative compounds with more than 98% accuracy (62/63), indicating its usefulness in drug discovery.