Understanding Drug-Induced Cholestasis: Mechanistic exploration of hepatic bile acid disposition

Drug-induced liver injury (DILI) is the most frequent reason for drug withdrawal from the market (Xu et al., 2004; Atienzar et al., 2016). Hence, DILI poses a major risk for patient health and pharmaceutical industry. Nowadays, pharmaceutical industry is still confronted with the poor prediction of DILI in non-clinical safety environment during drug development. The current non-clinical in vivo animal tests are not able to reliably detect DILI, as several key pathways involved in DILI differ between animals and human. In addition, taking the 3R principle (replacement, refinement, reduction) into account, non-clinical safety assessment is eagerly searching to substitute the conventional in vivo animal tests and implement new and reliable in vitro tools. These in vitro methods are pivotal for the accurate prediction and screening of compounds that can cause DILI. Unfortunately, the quest for the development of new in vitro tools is hampered with the limited mechanistic knowledge concerning DILI. Therefore, it is crucial to enhance our understanding regarding the toxicity pathways and thus to address the challenge to improve our apprehension about DILI.

Drug-induced cholestasis (DIC) accounts for about half of the DILI cases. DIC implies a disturbance of bile acid homeostasis, which eventually results in toxicity. Therefore, it is important that the pathways, responsible for bile acid homeostasis in vivo, are mimicked appropriately in vitro. In fact, over the past decades, the inhibition of the Bile Salt Export Pump (BSEP) by drugs served as the prime mechanism of DIC. As bile acid homeostasis is a complex interplay of several (other) hepatic transporters and metabolizing enzymes, there is growing interest in expanding our mechanistic knowledge concerning other causes of DIC, apart from BSEP inhibition. Furthermore, an improved insight in the different aspects of bile acid homeostasis will expand our horizon on the development of more advanced in vitro and in silico models. 

The overall goal of the current thesis research aimed to shed more light on the mechanistic understanding of interferences with bile acid homeostasis, caused by drugs. Therefore, sandwich-cultured rat or human hepatocytes (SCRH or SCHH) were applied to mimic the bile acid disposition pathways in vitro. Sandwich-cultured hepatocytes preserve all hepatocyte-specific functions, together with the maintenance of hepatic transporters and metabolic enzymes. In that respect, an hepatocyte-based in vitro DIC assay, that was previously developed by our research team, was optimized and validated using an extended set of test compounds (Aim 1). This in vitro DIC assay is based on the co-incubation of a cholestatic compound with a physiological bile acid mixture, followed by the measurement of urea production to determine the cell functionality. In a next phase, this in vitro DIC assay was used as a starting point to improve our knowledge about the role of (other) hepatic bile acid transporters in DIC (Aim 2). As such, this thesis research provided an answer to following specific research questions:

Can cryopreserved hepatocytes inter-changeably be applied in the in vitro DIC assay, next to freshly-isolated hepatocytes? (Chapter 3; part of Aim 1).

Is the in vitro DIC assay able to reliably predict cholestatic compounds in vitro and translate it to risk for DIC in vivo? (Chapter 4; part of Aim 1).

Is there a link between domperidone-indiuced modulation of OATP1B1-mediated uptake transport in hepatocytes and domperidone-induced alteration of bile acid homeostasis? (Chapter 5 & 6; part of Aim 2).

What are the predominant bile acid disposition pathways affected by bosentan, when evaluated in SCHH with chenodeoxycholic acid (CDCA) as a prototypic bile acid?  (Chapter 7; part of Aim 2).

The first study demonstrated that hepatocytes before or after cryopreservation can inter-changeably be applied in the in vitro DIC assay. Using SCRH from freshly-isolated and cryopreserved sources, we demonstrated the robustness of the model. Cyclosporin A and troglitazone were used as model compounds for cholestasis. Comparable Drug-Induced Cholestasis Index (DICI) values were obtained, irrespective of the cryopreservation history. The DICI reflects the ratio of the urea produced in the conditions where the model compounds were co-incubated with bile acids and the urea produced in the conditions where the model compounds were incubated alone (without bile acids challenge). A DICI value lower than 0.80 indicates that a compound at a particular concentration is able to disturb bile acid homeostasis in the SCRH. As an important outcome of this study, we proposed to normalize the urea production for the confluence rate of the cultures as SCRH from cryopreserved hepatocytes yielded a lower attachment efficiency, compared to SCRH from freshly-isolated hepatocytes.

To answer the second research question, 14 test compounds (set of hepatotoxic, cholestatic and non-hepatotoxic compounds) of the EU-EFPIA IMI project MIP-DILI consortium were applied to investigate the performance of the in vitro DIC model and thus to expand on validation of the model. Several batches of human hepatocytes in sandwich-culture were qualified for DIC assessment by verifying the bile acid-dependent increase in susceptibility to the toxic effects of cyclosporin A. To account for the high-interindividual variability between donors, “Cyclosporin A DICI-positive” human batches were introduced to exclude human hepatocyte donors which did not reveal a disturbance in bile acid homeostasis in presence of cyclosporin A. The 14 test compounds were classified based on the safety margin (SM)-concept. Using the SM, which is the ratio of the lowest compound concentration with a DICI ≤ 0.80 to the Cmax,total, we were able to determine the cholestatic risk of each compound in vivo. We unambiguously classified all 14 compounds for their risk of DIC in vivo, based on in vitro predicted SM values and in vivo incidence reports as provided in literature.

The next two studies focused on the mechanistic understanding of the drug-induced disturbances of bile acid homeostasis. In a first study, the role of hepatic uptake as mediated by the Organic Anion Transporting Polypeptide 1B1 (OATP1B1) was explored with respect to bile acid homeostasis. We hypothesized whether domperidone, for which modulation of OATP1B1-mediated transport has been shown in vitro, would be able to modulate the uptake of bile acids and as such, disturb bile acid homeostasis. Initially, we further explored the stimulatory effect of domperidone on OATP1B1-mediated uptake of two fluorescent probe substrates (sodium fluorescein and cGamF) using different hepatic in vitro models namely, OATP1B1-transfected cell lines, rat and human hepatocytes in suspension. In addition, the role of OATP was investigated in the hepatic uptake of domperidone in rat and human hepatocytes. Overall, it was shown that the stimulatory capacity of domperidone was substrate- and isoform- dependent. In particular, domperidone was able to stimulate sodium fluorescein uptake in OATP1B1-transfected cell lines and human hepatocytes, while it resulted in inhibition of sodium fluorescein uptake in rat. Also, our findings indicated a possible involvement of the Organic Cation Transporter (OCT) in the domperidone uptake in rat and human hepatocytes. Secondly, the in vitro DIC assay was applied to investigate the effect of domperidone on bile acid homeostasis. Again a species-dependent effect could be determined, showing that domperidone clearly disturbed bile acid homeostasis in human, but not in rat hepatocytes. Next to decreases in DICI values by domperidone in SCHH, domperidone was able to decrease the levels of glycine conjugated bile acids, while increases were observed in the amount of unconjugated bile acids in a concentration-dependent manner. The Bile Acid Disturbance Index (BADI) was introduced to comprehensively express the capability of a compound to disturb bile acid homeostasis in vitro. The BADI represents a ratio of the relative difference in the accumulation of unconjugated bile acids in the culture medium and the relative difference in the accumulation of intracellular conjugated bile acids. Both domperidone and the cholestatic reference drug cyclosporin A were able to increase the BADI in a similar fashion while also decreases in DICI values were seen for both compounds. Furthermore, it was observed that exposure to domperidone induced a decrease in the levels of endogenous bile acids. The study concluded that domperidone and cyclosporin A were able to alter bile acid disposition by mechanisms that are not fully elucidated yet. However, the modulation of bile acid uptake transport in human hepatocytes could potentially serve as a potential mechanism playing part in DIC. Moreover, the reliable measurement of endogenous or exogenously added bile acids in the culture medium and cells serves as an interesting and sensitive approach to investigate alterations in bile acid homeostasis in vitro. In that respect, our study confirms that a disturbance in bile acid homeostasis is a key event of DIC, but this does not necessarily imply an increased intracellular accumulation of bile acids in particular.

In the final study of this thesis project, we hypothesized that mechanisms, other than BSEP inhibition, could potentially be involved in bosentan-induced cholestasis. Therefore, we studied the effects of clinically relevant concentrations of bosentan on the disposition of endogenous bile acids, but also on the disposition of externally added CDCA in SCHH. We showed that bosentan was able to inhibit the uptake of CDCA, inhibit the conversion of CDCA to GCDCA and alter both sinusoidal and canalicular efflux clearances, as concluded from the disturbances caused by bosentan on the CDCA disposition. In addition, inhibition of the bile acid synthesis by bosentan was observed by a decreased amount of endogenous bile acids in presence of bosentan in SCHH. In conclusion, we gained insights in the overall influence of bosentan on bile acid disposition in SCHH. Therefore, these results may provide a mechanistic perspective on the interference of bosentan with bile acid disposition in human hepatocytes.

In summary, this doctoral research aided in the development and improvement of new and currently existing in vitro tools for a better prediction of DIC in vivo. The in vitro DIC assay enables early identification of drug candidates (or their metabolites) with a potential to cause DIC in vivo and supports adequate decision-making on the DIC assessment. Furthermore, the developed in vitro tool served as a starting point for gaining more insights in mechanisms involved in drug-induced alterations of bile acid homeostasis, other than BSEP inhibition, by proposing the impact of uptake transporters in DIC. A disturbance of bile acid homeostasis, but not necessarily an intracellular accumulation of bile acids is playing part in bile acid-mediated toxicity, as shown by measurement of bile acids in medium, cell and bile compartments of SCH. Therefore, bile acid measurements serve as valuable endpoints to mechanistically determine pathways, altered by a cholestatic compound, in bile acid disposition. Although many fundamental questions in this field remain to be answered, the obtained results will be useful in the further development of other, more advanced in vitro models to predict DIC and emphasizes the importance to sharpen our knowledge concerning the key pathways playing roles in the onset of DIC.