Predicting in vivo phospholipidosis-inducing potential of drugs by a combined high content screening and in silico modelling approach

• Published in: Toxicology in vitro 2015-04, Vol.29 (3), p.621-630
• Main Authors: Bauch, Caroline, Bevan, Samantha, Woodhouse, Heather, Dilworth, Clive, Walker, Paul
• Format: Article
• Language: English
• Subjects: Algorithms; Animals; Combined screening strategy; Computer Simulation; Drug-Related Side Effects and Adverse Reactions; Hep G2 Cells; High content screening; High-Throughput Screening Assays; Humans; In silico models; In vitro; Lipidoses - chemically induced; Lysosomal trapping; Lysosomes - metabolism; Models, Biological; Pharmaceutical Preparations - chemistry; Phospholipidosis; Phospholipids - metabolism; Rats; Reproducibility of Results
• ISSN: 0887-2333; 1879-3177
• DOI: 10.1016/j.tiv.2015.01.014

•Two high content screening assays: phospholipidosis and lysosomal trapping.•Validation of two cellular models: HepG2 and H9c2 cells.•Applicability domains are required for compound classes like aminoglycosides.•In vitro and in silico methods to predict phospholipidosis in vivo are compared.•Formulation of a screening strategy to better predict phospholipidosis in vivo. Drug induced phospholipidosis (PLD) is an adverse side effect which can affect registration of new drug entities. Phospholipids can accumulate in lysosomes, organelles essential in cellular biogenesis and if compromised can lead to cellular toxicity. Drug accumulation in lysosomes (lysosomotropism) is a known mechanism leading to PLD, however phospholipidosis can also occur indirectly by altering synthesis and processing of phospholipids. Drug induced PLD can be measured in vitro using High Content Screening (HCS) approaches, by either determining accumulation of phospholipids conjugated to dyes in cells or by determining accumulation of drugs within lysosomes, by competitive loss of lysosomal dye uptake. In this study we validate two in vitro assays using HepG2 and H9c2 cells in conjunction with in silico models based on physico-chemical properties using 56 compounds (28 phospholipidogenic, 25 non-phospholipidogenic and three kidney specific). Using HCS to determine PLD and lysosomal trapping in HepG2 cells in combination with in silico modelling increase the overall prediction of PLD in vivo with a sensitivity of 96%, specificity of 92% and overall accuracy of 94%. The findings of this study demonstrate the applicability of in vitro and in silico approaches to understand the mechanism underlying PLD and the utility of these approaches as a screening strategy in the pharmaceutical industry to select drug candidates with a low in vivo PLD liability.