High-content imaging of human hepatic spheroids for studying compounds with cholestatic liability

Drug-induced cholestasis (DIC) constitutes a subgroup of drug-induced liver injury (DILI) and represents a major factor for drug failure during the premarketing and post-marketing phases1. DIC can be the result of multiple triggering factors or molecular initiating events with bile salt export pump (BSEP) inhibition being implicated as a major mechanism. Significant gaps in the mechanistic understanding of DIC still exist and its preclinical prediction is mainly limited to assessing the compound’s potential to inhibit the BSEP2. Therefore, earlier identification of drug candidates with cholestatic signatures remains a challenge. This study aimed to explore the suitability of a 3D liver spheroid model of HepaRG cells to predict DIC in early drug development using high-content imaging (HCI). Nine annotated tool compounds divided into three categories (cholestatic DILI, non-cholestatic DILI, and non-DILI) were tested. HepaRG spheroids were cultured in 384-well ultra-low attachment plates. 8 days after seeding, the spheroids were exposed to nine half-log spaced concentrations per compound. Spheroid viability was assessed with an ATP assay after 24 h, 72 h, and 7 days of exposure to testing compounds. For HCI, 24 h exposure was selected as an appropriate time point, and the analysis relied on the most important key events described in the cholestasis adverse outcome pathway (AOP), including the assessment of oxidative stress, mitochondrial dysfunction, endoplasmic reticulum stress, BSEP, and tight junctions. Time was an important factor in the cytotoxicity of non-DILI compounds, while time- and concentration-dependent response effects were observed for cholestatic-DILI and non-cholestatic DILI compounds. HCI analysis was performed on maximum intensity projection of small Z-stacks, indicating that, the mechanisms behind the toxicity of the cholestatic DILI compounds tend to induce mitochondrial and endoplasmic reticulum stress followed by altering BSEP distribution. Follow-up studies with a larger sample number are required to further explore the potential of HCI to detect compounds
with cholestatic liability. Multiparametric HCI analysis and hierarchical clustering analysis of the key events induced by several compounds that cause DILI might be promising tools for mechanistically evaluating cholestatic effects.