Kristof De Vos
KU Leuven
Objectives: Drug-induced liver injury (DILI) is a relatively rare hepatic condition, yet the most frequent cause of acute liver failure in North America and Europe. Drug-induced cholestasis (DiCho) is responsible for about 50% of all DILI cases. One way to pursue a better mechanistic understanding of DiCho lies in the use of Adverse Outcome Pathways (AOPs). Altered bile acid (BA) disposition and mitochondrial impairment represent key events of the AOP for DiCho, yet this approach lacks the quantitative dimension. Therefore, the aim of this work was to gain mechanistic understanding of bile acid disposition and mitochondrial functions at the hepatocyte level, via a quantitative in vitro-in silico integrated approach.
Methods: In vitro experiments were performed on primary human hepatocytes to generate longitudinal datasets of (i) chenodeoxycholic acid (CDCA) and lithocholic acid (LCA) disposition, (ii) mitochondrial respiration, and (iii) abundances of several BA-relevant proteins. The evaluation of the BA disposition encompassed LC-MS-based quantification of BA species (CDCA, LCA, as well as amidated-and/or sulphated-conjugates) in samples corresponding to the intracellular, extracellular, and/or bile canalicular compartments of the hepatocyte cultures. The mitochondrial respiration rate was monitored with Seahorse XF analyzer in absence and presence of bosentan, while absolute protein abundances (fmol/µg protein) of the relevant protein targets were quantified with targeted proteomics.
Results: The in vitro datasets were utilized to establish quantitative (ODE-based) kinetics models. One training dataset and two independent validation datasets were used for the CDCA disposition model, while only 1 dataset was available for the LCA disposition model. BA (basolateral and canalicular) transport across the different compartments and BA metabolism (glycine-amidation and sulfation) within the (intracellular) compartment were included. Most of the kinetic processes were parameterized as first-order clearances. However, some of these clearances in the CDCA model were subsequently replaced by a time-dependent clearance, by making it dependent on protein abundance and activity. This approach was required to cover homeostasis of hepatic GCDCA, and it improved the model prediction accuracy based on the validation datasets. Furthermore, several diagnostic plots, like goodness-of-fit plots and parameter likelihood profiles, confirmed that the current mechanisms and equations in the models adequately described the kinetics of the BA disposition, as well as the mitochondrial respiration rates. Additive model RSEs ranged between 9% and 22%, while individual parameter estimation RSEs ranged between 4% and 48%.
Conclusions: The parameter estimates for BA disposition obtained in these primary hepatocytes under biorelevant conditions could serve as baseline values for comparison against estimates when perturbed conditions (for instance in the presence of a cholestatic compound) are used as model input. This will allow to identify the (sometimes subtle) upregulation and/or downregulation of the disposition pathways when hepatocytes are being exposed to xenobiotics. Moreover, simulations can be used to identify associations between the pathways involved in BA disposition, mitochondrial functioning, and protein abundance and activity.
References:
[1] Text for reference 1.
[2] Text for reference 2, etc etc
Reference: PAGE 32 (2024) Abstr 11071 [www.page-meeting.org/?abstract=11071]
Poster: Drug/Disease Modelling - Other Topics