Advertisement
Research ArticleArticle

In Vitro Inhibition of the Bile Salt Export Pump Correlates with Risk of Cholestatic Drug-Induced Liver Injury in Humans

Sarah Dawson, Simone Stahl, Nikki Paul, Jane Barber and J. Gerald Kenna
Drug Metabolism and Disposition January 2012, 40 (1) 130-138; DOI: https://doi.org/10.1124/dmd.111.040758

Abstract

Inhibition of the activity of the human bile salt export pump (BSEP: ABCB11) has been proposed to play a role in drug-induced liver injury (DILI). To enhance understanding of the relationship between BSEP inhibition and DILI, inhibition of human BSEP (hBSEP) and its rat ortholog (rBsep) by 85 pharmaceuticals was investigated in vitro. This was explored using assays that quantified inhibition of ATP-dependent [3H]taurocholate uptake into inverted plasma membrane vesicles from Sf21 insect cells, which expressed the proteins. Of the pharmaceuticals, 40 exhibited evidence of in vitro transporter inhibition and overall a close correlation was observed between potency values for inhibition of hBSEP and rBsep activity (r2 = 0.94), although 12 drugs exhibited >2-fold more potent inhibition of hBSEP than rBsep. The median potency of hBSEP inhibition was higher among drugs that caused cholestatic/mixed DILI than among drugs that caused hepatocellular or no DILI, as was the incidence of hBSEP inhibition with IC50 <300 μM. All drugs with hBSEP IC50 <300 μM had molecular weight >250, ClogP >1.5, and nonpolar surface area >180Å. A clear distinction was not evident between hBSEP IC50 or unbound plasma concentration (Cmax, u) of the drugs in humans and whether the drugs caused DILI. However, all 17 of the drugs with hBSEP IC50 <100 μM and Cmax, u >0.002 μM caused DILI. Overall, these data indicate that inhibition of hBSEP/rBsep correlates with the propensity of numerous pharmaceuticals to cause cholestatic DILI in humans and is associated with several of their physicochemical properties.

Introduction

ATP-dependent transporters expressed on the apical plasma membrane domain of hepatocytes are important mediators of canalicular bile flow. Impaired bile flow arising from genetically determined defects in transporters has been implicated in various inherited forms of cholestatic liver disease in humans. Progressive familial intrahepatic cholestasis type 2 (PFIC2) is characterized by early-onset cholestasis soon after birth and subsequent progressive degenerative liver injury and is fatal unless treated by liver transplantation (Strautnieks et al., 1998). PFIC2 is a consequence of mutations in the gene encoding the human bile salt export pump (BSEP, ABCB11; rodent ortholog Bsep, Abcb11; originally termed “sister of P-glycoprotein,” spgp), which mediates efflux of bile salts from hepatocytes into bile and is essential for normal bile formation and flow. In patients with PFIC2, mutations or single nucleotide polymorphisms (SNPs) in the BSEP gene result in either reduced levels of mRNA transcription or translation or reduced protein stability or activity (Strautnieks et al., 2008; Byrne et al., 2009; Ho et al., 2010), and liver injury is considered to be caused by intracellular accumulation of cytotoxic bile constituents (Perez and Briz, 2009). Less functionally severe ABCB11 gene mutations have been shown to result in benign recurrent intrahepatic cholestasis type 2, which is characterized by nonprogressive cholestasis (Noe et al., 2005). Defective expression of additional biliary transporters has been implicated in other forms of genetically inherited cholestatic liver injury. For example, mutations in multidrug resistance protein 3 (MDR3, ABCB4) result in impaired biliary excretion of phosphatidylcholine and cause PFIC3 (Gonzales et al., 2009). Furthermore, mutations in multidrug resistance-associated protein 2 (ABCC2), which mediates biliary excretion of numerous compounds including bilirubin glucuronide and many drug metabolites, cause hyperbilirubinemia and jaundice (Dubin-Johnson syndrome) (Kartenbeck et al., 1996).

Knockout mice have provided further insight into the complex interrelationships between expression of individual bile salt transporters, bile flow, and liver injury. Homozygous Bsep knockout mice were shown to develop severe cholestasis when fed a bile acid-enriched diet, whereas only mild cholestasis was observed when animals were fed a normal diet. This result has been attributed to adaptive changes in expression of other enzymes and transporters in spgp(−/−) mice, which enable them to cope with the lack of functional Bsep expression unless challenged with a high dietary bile acid load (Wang et al., 2003, 2009). Other transporters implicated in cholestatic liver injury via studies undertaken in knockout mice include Mdr2 (the rodent ortholog of human MDR3) (Fickert et al., 2004).

Liver injury is known to be caused by a large number of different drugs. Drug-induced liver injury (DILI) is of particular concern because it is a major cause of serious illness in humans, drug withdrawal postmarketing, regulatory failure, and attrition during drug discovery and clinical development (Abboud and Kaplowitz, 2007). At present, the underlying mechanisms remain ill defined. Many drugs cause either cholestatic or mixed hepatocellular/cholestatic liver injury and in a number of instances evidence of functional inhibition of Bsep activity by such drugs has been observed in vivo in preclinical species, resulting in impaired bile acid excretion and elevated plasma bile acid levels (Funk et al., 2001). Examples include cyclosporine, troglitazone, and bosentan (Böhme et al., 1994; Fattinger et al., 2001; Funk et al., 2001). In addition, numerous drugs that cause cholestatic or mixed DILI have been shown to inhibit BSEP and/or Bsep activity in vitro, using experimental systems including isolated hepatocytes cultured in sandwich configuration, purified liver canalicular membrane vesicle preparations, and membrane vesicles from cell lines transfected with BSEP/Bsep from various species (Stieger et al., 2000; Fattinger et al., 2001; Funk et al., 2001; Byrne et al., 2002; Horikawa et al., 2003; Hirano et al., 2006; Mano et al., 2007; Yabuuchi et al., 2008; Kis et al., 2009).

Although these findings raise the possibility that BSEP inhibition could play a role in development of DILI, very few studies have systematically compared the potency of BSEP/Bsep inhibition by drugs that cause DILI in humans with transporter inhibition by nonhepatotoxic drugs (Morgan et al., 2010). To obtain further insight into the possible relationship between in vitro BSEP inhibition and DILI in humans, we have evaluated inhibition of the human (hBSEP) and rat (rBsep) orthologs by 85 licensed hepatotoxic or nonhepatotoxic drugs, using membrane vesicles from Sf21 insect cells expressing the transporters. We have also considered the potential impact of drug dose and maximum unbound plasma concentration (Cmax, u) as additional DILI risk factors and have explored the relationship between hBSEP/rBsep inhibition and several physicochemical properties of the tested drugs.

Materials and Methods

Materials.

All chemicals were purchased from Sigma-Aldrich (St. Louis, MO) unless stated otherwise and were of the highest purity available. Flucloxacillin was provided by Ann K. Daly (University of Newcastle, Newcastle upon Tyne, UK). Complete protease inhibitor tablets were obtained from Roche (Basel, Switzerland). [3H]taurocholate (1 mCi/ml) was purchased from PerkinElmer Life and Analytical Sciences (Waltham, MA). All drugs (>95% purity) were synthesized by AstraZeneca or purchased from Sigma-Aldrich.

Expression of hBSEP and rBsep in Sf21 Insect Cells.

Human BSEP (ABCB11)-expressing baculoviruses were provided by Bruno Stieger (Stieger et al., 2000; Noé et al., 2002). Rat Bsep (Abcb11) cDNA was prepared from Sprague-Dawley rat liver as described by Jupp et al. (2006) and inserted into the pFastBac1 vector. The sequence of hBSEP and rBsep expressed in the vesicles was determined by dye terminator sequencing in two directions and analyzed using the phred/phrap/polyphred suite. The most common DNA sequence for ABCB11 is NM_003742, but with base substitutions at two SNPs, one synonymous (rs497692) and one nonsynonymous (rs2287622). The hBSEP DNA sequence expressed in the vesicles corresponds to NM_003742 with the following base substitutions: T78C, T81C, A501G, T1062C, T1331C (rs2287622), and A3084G (rs497692). This sequence represents the most common hBSEP protein variant expressing alanine at position 444 and matches NP_003733 with a substitution at A444V due to SNP rs2287622 at codon T1331C. The DNA sequence of rBsep matches NM_031760, and corresponds to the protein sequence NP_113948. To express hBSEP and rBsep in insect cells, recombinant baculoviruses were generated with the Bac-to-Bac baculovirus expression system (Invitrogen, Paisley, UK) according to the manufacturer's instructions. Insect Spodoptera frugiperda Sf21 cells (3 × 106/ml) were infected with the recombinant baculoviruses and cultured in SF900II insect serum-free medium (Invitrogen) at 27°C with gentle shaking. At 48 h after infection, cells were harvested by centrifugation and stored at −80°C until use.

Preparation of Plasma Membrane Vesicles from Sf21 Cells.

Plasma membrane vesicles were prepared from hBSEP- and rBsep-expressing Sf21 cells. Frozen cell pellets were thawed on ice, diluted 40-fold with hypotonic buffer (0.5 mM HEPES/Tris, pH 7.4, 0.1 mM EGTA, and complete protease inhibitor tablets) and then were sonicated to achieve approximately 80% cell lysis. After centrifugation at 1000g for 10 min (4°C), the supernatant was further centrifuged at 10,000g for 60 min (4°C). The resulting pellet containing the membrane fraction was collected and resuspended in SMS buffer (250 mM sucrose, 100 mM KNO3, and 5 mM HEPES/Tris, pH 7.4, and complete protease inhibitor tablets). After determination of the protein concentration by BCA (Thermo Fisher Scientific, Waltham, MA), the membrane suspension was stored at −80°C until used.

Determination of Transport of [3H]Taurocholate.

Vesicles (60 or 100 μg of protein/reaction for hBSEP and rBsep, respectively) were incubated with test compound or dimethyl sulfoxide and [3H]taurocholate [0.5 μM [3H]taurocholate, 5 mM ATP, 10 mM MgCl2, 7.5 mM HEPES/Tris, pH 7.4, 150 mM KNO3, 175 mM sucrose, and 12.5 mM Mg(NO3)2] for 5 min at 37°C. In control reactions ATP was replaced with 5 mM AMP. The final dimethyl sulfoxide concentration was 2% (v/v). At the end of the incubation period vesicles were transferred to GF/B filter plates (Whatman Plc, Kent, UK) containing ice-cold stop buffer (50 mM sucrose, 100 mM KCl, 5 mM HEPES/Tris, pH 7.4, 0.1 mM taurocholate, and 5 mM EDTA) to terminate the transport reaction. Under aspiration each well was washed four times with ice-cold stop buffer using a Millipore vacuum filtration system (Millipore Corporation, Billerica, MA). [3H]taurocholate incorporation into vesicles was measured by counting the radioactivity remaining on the filter using a MicroBeta JET (PerkinElmer Life and Analytical Sciences). All incubations were performed in triplicate in individual experiments, and all experiments were undertaken on a minimum of three separate occasions.

Data Analysis.

IC50 values were calculated with nonlinear regression for a sigmoidal dose-response using the four-parameter logistic equation in GraphPad Prism (version 4.03 for Windows; GraphPad Software Inc., San Diego, CA). Km and Vmax were derived from Michaelis-Menten analysis.

DILI Classification of Pharmaceuticals.

Eighty-five pharmaceuticals were tested for hBSEP and rBsep inhibition in vitro. Of these, 64 had been reported to cause symptomatic DILI in humans and were further categorized on the basis of literature reports of whether the pattern of liver injury observed was cholestatic or mixed cholestatic/hepatocellular DILI (“cholestatic/mixed DILI”) or hepatocellular DILI (“hepatocellular DILI”). In many instances, literature reports of patterns of liver injury caused by individual drugs in the human population exhibit marked variability among patients. For example, both diclofenac and troglitazone have been most commonly reported to cause hepatocellular DILI, whereas cholestatic/mixed DILI has been observed less frequently (Stricker, 1992). For simplicity, most of the present analysis was performed by grouping all drugs that have been reported to cause cholestatic or mixed cholestatic/hepatocellular DILI in humans together using the term cholestatic/mixed DILI, regardless of the relative frequency of such reports compared with reports of hepatocellular DILI. Any deviations from this classification are described in the text and figure legends. With reference to the different forms of cholestatic injury that have been observed, we refer to the hepatocanalicular/cholestatic hepatitis form as described by Stricker (1992) and Zimmerman (1999). We did not test any anabolic or contraceptive steroids known to cause bland/pure cholestasis in the current study. Drugs that caused DILI were also assigned to two different severity groups. Drugs that have been withdrawn from clinical use because of DILI or have been given DILI black box warnings were classified as “severe DILI,” whereas all other drugs that caused DILI were classified as “marked DILI”. Of the drugs, 21 have been used in humans without evidence of overt liver dysfunction (category “no DILI”). Several drugs in this category have been reported to cause elevations in serum transaminases in humans, suggesting mild but asymptomatic liver dysfunction. Data on DILI categorization of the drugs are summarized in Supplemental Table 1.

Physicochemical Properties.

Physicochemical property data were obtained from AstraZeneca databases, which incorporate standard packages for molecular weight (OEChem TK programming library), ClogP (CLOGP version 4.3 from Biobyte database), and nonpolar surface area (NPSA) (SELMA; an AstraZeneca in-house software package). For further information contact T. Olsson or V. Sherbukhin, Synthesis and Structure Administration, AstraZeneca R&D Mölndal (Mölndal, Sweden).

Results

Time Course and Kinetic Analysis of ATP-Dependent Taurocholate Transport in Membrane Vesicles Expressing hBSEP or rBsep.

To quantify active transport of the specific probe substrate [3H]taurocholate by hBSEP or rBsep expressed in Sf21 insect cells, substrate uptake into inverted plasma membrane vesicles was determined in the presence of ATP (which supports active transport) or AMP (which does not). Uptake in the presence of ATP was at least 5-fold higher than uptake in the presence of AMP (data not shown), and ATP-dependent transport was linear for up to 5 min for both hBSEP and rBsep (Fig. 1A). ATP-dependent [3H]taurocholate uptake was not observed in plasma membrane vesicles prepared from mock-transfected Sf21 cells (data not shown). hBSEP- and rBsep-mediated taurocholate transport was substrate concentration-dependent (Fig. 1B) and exhibited Michaelis-Menten kinetics (linear Eadie-Hofstee plots are shown in Fig. 1C). The Michaelis constants for taurocholate were as follows: hBSEP, Km = 8.9 ± 2.0 μM, Vmax = 74.6 ± 7.6 pmol · min−1 · mg protein−1; and rBsep, Km = 10.7 ± 1.2 μM, Vmax = 40.3 ± 2.2 pmol · min−1 · mg protein−1.

Fig. 1.
Fig. 1.

Time dependence and kinetics of ATP-dependent [3H]taurocholate uptake in hBSEP and rBsep expressing Sf21 plasma membrane vesicles. A, time course at 0.5 μM [3H]taurocholate. B, influence of [3H]taurocholate concentration. C, Eadie-Hofstee analysis. Results are the mean ± S.E.M. of three separate experiments. ■, hBSEP; □, rBsep.

Inhibition of hBSEP and rBsep Transport Activity in Vesicles by Pharmaceuticals.

The effects of three pharmaceuticals on hBSEP and rBsep activities are shown in Fig. 2, A and B, respectively: potent transport inhibition by cyclosporine, less potent inhibition by alpidem, and no detectable inhibition by streptomycin. IC50 values obtained for these compounds and for an additional 82 drugs, are summarized in Table 1. The pharmaceuticals were grouped into three categories, according to the pattern of DILI they have been reported to cause in humans. These were cholestatic or mixed cholestatic/hepatocellular DILI (n = 42), hepatocellular DILI (n = 22), or no DILI (n = 21) (Supplemental Table 1). Reproducibility of replicate experiments was high, demonstrated by the 95% confidence limits (Table 1). Overall, a close correlation was evident between inhibition of hBSEP and rBsep activity (r2 = 0.94) (Fig. 3). Of interest, 12 of the 85 compounds exhibited >2-fold more potent inhibition of hBSEP than of rBsep activity (seven cholestatic/mixed DILI, one hepatocellular DILI, and four no DILI). The most marked potency differences were observed with pioglitazone (hBSEP IC50 7-fold lower than rBsep IC50), rosiglitazone (6-fold), nefazodone (4-fold), and ketoconazole (5-fold), all of which caused cholestatic/mixed DILI.

Fig. 2.
Fig. 2.

Inhibition of vesicular hBSEP and rBsep transport activity by selected pharmaceuticals. Inhibition of hBSEP (A) and rBsep (B) activity by cyclosporine (○), alpidem (▴), and streptomycin (■) is shown. Results are the mean ± S.E.M. from three separate experiments.

View this table:
TABLE 1

Inhibition of human BSEP and rat Bsep mediated [3H]taurocholate transport by pharmaceuticals

Fig. 3.
Fig. 3.

Comparative potencies of inhibition of hBSEP and rBsep by 85 pharmaceuticals. Each point represents the geometric mean of the IC50 determined from a minimum of three separate experiments per compound. Red circles, cholestatic/mixed DILI; yellow triangles, hepatocellular DILI; green squares, no DILI. The solid line indicates a line of unity (slope = 1.0); dotted lines indicate 2-fold IC50 deviation from line of unity.

Relationship between DILI Category and hBSEP Inhibition.

Of the 42 tested drugs that cause cholestatic/mixed DILI, 26 (62%) inhibited hBSEP with IC50 values ranging from 0.3 μM (pioglitazone) to 598.6 μM (ibuprofen). Quantifiable IC50 values were obtained for 12 of 20 (60%) of the drugs reported to cause cholestatic DILI and 14 of 22 (64%) drugs reported to cause mixed cholestatic/hepatocellular DILI. Comparable median IC50 values were also observed for these two subcategories of drugs (34.4 μM versus 46.1 μM, respectively; analysis not shown). For subsequent analyses, drugs in the two subcategories were therefore grouped together and collectively termed cholestatic/mixed DILI.

The overall incidence of hBSEP inhibition by the cholestatic/mixed DILI drugs was markedly higher than the incidence of hBSEP inhibition in the groups of drugs that caused hepatocellular DILI (6 of 22 = 27%) or no DILI (8 of 21 = 38%) (Fig. 4). In addition, the median potencies of hBSEP inhibition by drugs that caused cholestatic/mixed DILI were markedly greater than potencies of hBSEP inhibition by drugs that caused hepatocellular or no DILI (Fig. 4; Table 2) (p < 0.05; cholestatic/mixed DILI compared with no DILI). Receiver operating characteristic curve analysis (data not shown) was undertaken to aid in assessment of the hBSEP IC50 value that provided the greatest discrimination between drugs causing cholestatic/mixed DILI and drugs causing hepatocellular DILI or no DILI. For hBSEP this was found to be 300 μM (dashed line on Figs. 4⇓–6). Of the 85 drugs tested, 33 caused hBSEP inhibition with IC50 values <300 μM and among the different DILI classes the incidences were cholestatic/mixed DILI, 24 of 42 = 57%, hepatocellular DILI, 4 of 22 = 18%, and no DILI, 5 of 21 = 24%.

Fig. 4.
Fig. 4.

Correlation between potency and incidence of hBSEP inhibition by 85 pharmaceuticals and DILI class. IC50 values are geometric means from at least three separate experiments per compound. Cholestatic (open red circles) and mixed cholestatic/hepatocellular (filled red circles) drugs have been displayed separately to demonstrate the similarity in potency and range of hBSEP IC50. Several nonoral drugs are indicated: salicylic acid (hepatocellular); clobetasol propionate, cromolyn, dopamine, isoproterenol, physostigmine, and streptomycin (no DILI). Praziquantel is also highlighted because this drug is not given via repeat daily dosing. The hBSEP IC50 dashed line (at 300 μM) indicates the value that gave optimal separation between drugs that caused cholestatic/mixed DILI and drugs that caused hepatocellular or no DILI, from receiver operating characteristic curve analysis.

View this table:
TABLE 2

Potency and incidence of human BSEP inhibition by 85 pharmaceuticals

Impact of Cmax, u, Dose, and Route of Drug Administration.

To explore the relationship between hBSEP inhibition, DILI class, and exposure, we considered the maximum unbound plasma concentrations (Cmax, u) and maximum daily doses in humans of the tested drugs. The Cmax, u value was calculated for each drug using the Cmax and plasma protein binding values available from published sources (Supplemental Table 2). Cmax, u values ranged from 0.0002 to 496.5 μM (Table 1). Similar ranges of Cmax, u values were evident for drugs in the different DILI categories (Fig. 5A), indicating that Cmax, u was not in itself a primary determinant of whether or not individual drugs caused DILI. The relationship between potency of hBSEP inhibition, Cmax, u, and DILI category is shown in Fig. 5B. Although no clear distinction was evident between hBSEP IC50, Cmax, u, and whether or not individual drugs caused DILI, it is notable that all of the 17 tested drugs with hBSEP IC50 <100 μM and Cmax, u >0.002 μM caused DILI and that 15 of these drugs (88%) were from the cholestatic/mixed DILI class (e.g., troglitazone). Furthermore, three of the five no DILI drugs that inhibited hBSEP with IC50 <300 μM had very low plasma exposure (Cmax, u ≤0.002 μM). These were the cholinesterase inhibitor donepezil, the topical steroid clobetasol propionate, and the anxiolytic buspirone (which has been reported to cause liver enzyme elevations in humans, indicating mild and asymptomatic liver dysfunction). The two remaining no DILI drugs that exhibited hBSEP IC50 <300 μM were the steroid dexamethasone and the antihelmintic praziquantel, which is administered at high dose (4200 mg) during a single day and thus sustained exposure in humans is not encountered. A similar pattern of results was obtained when the analysis was undertaken using maximum daily dose in place of Cmax, u (Supplemental Fig. 1).

Fig. 5.
Fig. 5.

Comparison between Cmax, u, DILI class, and potency of hBSEP inhibition for 85 pharmaceuticals. Each point represents an individual drug. A, Cmax, u in humans (micromolar concentration). B, comparison between potency of hBSEP inhibition and Cmax, u. Red circles, cholestatic/mixed DILI; yellow triangles, hepatocellular DILI; green squares, no DILI.

Lack of Relationship between hBSEP Inhibition and DILI Severity.

DILI caused by 18 of the 85 tested drugs resulted in drug withdrawal or in regulatory black box warnings. There was no significant difference in the incidence (10 of 18 = 55%) or median potency (IC50 28.2 μM) of hBSEP inhibition by these severe DILI drugs compared with the 47 other tested drugs that caused DILI (incidence 18 of 47 = 38%; median IC50 70.8 μM) (t test of severe versus marked DILI; not significant). Consideration of Cmax, u or drug dose did not affect the observed lack of correlation between potency of hBSEP inhibition and DILI severity category (data not shown). Data on DILI severity are summarized in Supplemental Table 1.

Relationships between hBSEP Inhibition Potency and Physicochemical Properties of Drugs.

Each of the 33 tested drugs that exhibited hBSEP IC50 <300 μM had molecular weight >250, ClogP >1.5, and NPSA >180 Å2 (Fig. 6; individual values are tabulated in Supplemental Table 3). Therefore, the prevalence of relatively potent hBSEP inhibition (i.e., IC50 <300 μM) was increased among drugs that exhibited these properties, compared with the prevalence of potent hBSEP inhibition among all the drugs tested (Table 3). However, it is notable that numerous drugs with molecular weight, ClogP, or NPSA values that exceeded these values exhibited no evidence of hBSEP inhibition. In addition, for drugs that exceeded these values only weak correlations were observed between the individual parameters and hBSEP IC50 values (molecular weight: r2 = 0.47, ClogP: r2 = 0.44, NPSA: r2 = 0.40; results not shown). A further increase in prevalence of potent hBSEP inhibition (0.77 versus 0.39) was evident for drugs with a combination of both molecular weight >250 and CLogP >1.5 (Table 3). No association could be identified between potency of hBSEP inhibition and the charge class of the drugs (acidic, neutral, basic, or zwitterionic) (analysis not shown; values are provided in Supplemental Table 3).

Fig. 6.
Fig. 6.

Relationship between hBSEP inhibition and physicochemical properties for 85 pharmaceuticals. A, molecular weight. B, ClogP. C, nonpolar surface area. Red circles, cholestatic/mixed DILI; yellow triangles, hepatocellular DILI; green squares, no DILI.

View this table:
TABLE 3

Relationship between human BSEP inhibition potency and physicochemical properties for 85 pharmaceuticals

Discussion

Testing of 85 pharmaceuticals for in vitro inhibition of hBSEP and its rat ortholog rBsep has revealed a broad range of IC50 values, ranging from 0.3 μM (pioglitazone) to >500 μM (ciprofibrate, cinchophen, probenecid, and ketotifen). Markedly higher incidences and potencies of transporter inhibition were evident among drugs that caused cholestatic/mixed DILI in humans compared with drugs that caused hepatocellular DILI or drugs that did not cause DILI. This finding provides key supportive evidence for the hypothesis that hBSEP inhibition in vivo contributes to the mechanism by which some drugs cause DILI in humans. It also suggests that in vitro evaluation of hBSEP inhibition could aid in selection of candidate drugs with reduced propensity to cause DILI.

Previous investigators have explored inhibition of both hBSEP and rBsep by pharmaceuticals, including a recent analysis of >200 drugs (Morgan et al., 2010). There are significant differences between the methods used in the present study to assess hBSEP and rBsep inhibition and those adopted by Morgan et al. (2010): most notably, the incubation temperatures (37°C versus 24°C), incubation times (5 min versus 15 min), and maximum tested drug concentrations (1 mM versus 133 μM). Nonetheless, there was good concordance between hBSEP IC50 values for the 11 drugs tested by both groups that exhibited potent transporter inhibition (analysis not shown). In addition, we have explored the possible impact on data interpretation of Cmax, u, dose, and physicochemical properties, compared incidences and potencies of hBSEP inhibition among drugs causing cholestatic/mixed versus hepatocellular DILI versus no DILI, and explored the possible relationship between hBSEP inhibition and DILI severity.

Of the tested drugs, 17 that exhibited hBSEP IC50 <100 μM and Cmax, u >0.002 μM have been reported to cause DILI in humans. This finding indicates likely interdependence between hBSEP inhibition potency and risk of DILI in humans, as does the finding that three of the five no DILI drugs with hBSEP inhibition IC50 <300 μM (donepezil, buspirone, and clobetasol propionate) exhibited low plasma exposure (Cmax, u ≤0.002 μM). This finding also suggests that consideration of dosing regimen, route of administration, and predicted Cmax, u has the potential to improve assessment of the DILI risk in humans posed by test compounds that inhibit hBSEP in vitro. Nonetheless, because two additional drugs (acitretin and rosiglitazone), which inhibited hBSEP and also had low plasma exposure, have been reported to cause cholestatic/mixed DILI, it is clear that the relationship between dose, Cmax, u, and human DILI risk posed by compounds that inhibit hBSEP is complex.

Our data analysis further suggests that there may be in vitro hBSEP and rBsep IC50 threshold values that can be used to distinguish between biologically relevant and irrelevant data. Although our studies have not addressed this issue directly, our data suggest that a useful operational threshold could be as high as 300 μM. This is higher than the Cmax, u values of all but a few of the tested drugs, even if adjustment is made for higher compound concentrations typically observed at the hepatic inlet compared with peripheral blood (Ito et al., 2002). It may be relevant that many drugs accumulate within hepatocytes at concentrations that are much higher than extracellular concentrations (Grime et al., 2008). In addition, whereas the calculated potency values were IC50 values, the fraction of transport inhibition required to cause functional alterations in vivo that contribute to DILI progression after long-term dosing are unknown and could be much lower (e.g., IC10 values). It is clear that further work will be required to address this important unresolved aspect.

A number of the drugs that caused DILI and that we have found to inhibit hBSEP in vitro have not been reported by previous researchers to inhibit the transporter. These are acitretin, alpidem, bezafibrate, dicloxacillin, flucloxacillin, glafenine, leflunomide, and ticlopidine. Therefore, we consider that our data provided novel insights into mechanisms by which these drugs may cause DILI, whereas the hBSEP inhibition we observed with three oxacillin antibiotics (cloxacillin, dicloxacillin, and flucloxacillin), all of which cause cholestatic DILI, indicates that the liability is a class effect for these drugs. Of interest, we also observed hBSEP inhibition by three drugs reported to cause severe hepatocellular DILI in humans: alpidem and glafenine, both of which have been withdrawn; and leflunomide, which was recently given a DILI black box warning. Perhaps for these and/or other drugs, the mechanism of DILI involves a combination of hBSEP inhibition and additional processes that remain to be elucidated.

We observed no association between hBSEP or rBsep inhibition by drugs that caused DILI and their DILI severity category, even when we took into account the Cmax, u exposure data. This observation is perhaps not surprising, because in vitro potency data take no account of differences in Cmax, u (which varied markedly among the tested drugs) or drug kinetics (which were not considered in the analysis). Furthermore, because hBSEP inhibition is one of numerous potential DILI mechanisms, evaluation of this liability in isolation should not be expected to provide a reliable overall assessment of the DILI liability of any individual molecule (Thompson et al., 2011). It may also be significant that the DILI severity categories used in our analysis primarily reflect regulatory outcome and not the relative DILI severities caused by the individual drugs or their incidence. For example, numerous studies have identified flucloxacillin as one of the drugs most commonly associated with DILI in numerous countries (but not in the United States, where it is not licensed), yet flucloxacillin has not been withdrawn because of its clear clinical value (Andrews and Daly, 2008). In addition, the marked DILI category contains drugs that differ considerably in DILI severities and incidence and includes both pioglitazone and rosiglitazone, which have been reported to cause cholestatic DILI (Floyd et al., 2009), albeit very infrequently.

Of the tested drugs, 79% exhibited similar potencies of inhibition of hBSEP and rBsep. This finding is not surprising because amino acid sequences of hBSEP and rBsep are highly conserved, with 80 to 85% identity (Noe et al., 2001), and similar kinetics of taurocholate transport were observed in the present study, which is consistent with data reported previously by others (Stieger et al., 2000; Byrne et al., 2002; Noé et al., 2002). In contrast, the potency of taurocholate transport inhibition was markedly greater (by up to 7-fold) for hBSEP than for rBsep for 12 drugs (14%) and for rBsep than for hBSEP for 6 drugs (7%). Species differences in inhibition potency of hBSEP have been observed previously for several of these drugs [e.g., troglitazone (Kis et al., 2009; Yabuuchi et al., 2008)]. At present, it is not clear whether they could involve interactions with the active site responsible for substrate transport or other regions of the molecules. These differences may help explain, at least in part, why rats and other preclinical species offer limited value for predicting whether candidate drugs may cause DILI in humans (Greaves et al., 2004).

It is also important to note that 16 of the 42 drugs that were reported to cause cholestatic/mixed DILI did not inhibit hBSEP in vitro (IC50 values >1000 μM) in the present study. hBSEP inhibition needs to be considered as one of numerous potential mechanisms that have been implicated in DILI, and liver injury caused by individual drugs in humans may involve multiple mechanisms acting synergistically in susceptible patients. Other contributory mechanisms include reactive metabolite formation, mitochondrial injury, and immune activation (Abboud and Kaplowitz, 2007; Greer et al., 2010; Thompson et al., 2011). Published data have implicated some or all of these mechanisms in DILI caused by a number of the tested drugs, some of which were found to inhibit hBSEP [e.g., flutamide (Fau et al., 1994) and troglitazone (Funk et al., 2001)] and some that were not [e.g., nitrofurantoin (Czaja, 2011) and valproate (Björnsson, 2008)]. Moreover, the present study was undertaken only with parent drugs and not their metabolites, which in some instances can exhibit more potent transporter inhibition [e.g., the sulfated metabolite of troglitazone (Funk et al., 2001)]. An absence of metabolic capability is a key limitation of the vesicle assay approach.

Finally, consideration of the physicochemical properties of the drugs revealed no apparent association between hBSEP inhibition and ion class (i.e., acid, base, or zwitterion), although it is notable that all drugs that inhibited hBSEP in vitro exceeded molecular weight of 250, ClogP of 1.5, and NPSA of 180Å. The practical utility of this finding for drug design is unclear, because many molecules designed currently by medicinal chemists exceed these values (Keserü and Makara, 2009). However, it could aid in prioritization of compounds for hBSEP inhibition testing and highlights the potential value to be gained from understanding the structural basis of hBSEP inhibition and the need to define quantitative structure-activity relationships. We anticipate that the data we have obtained will aid this endeavor.

Authorship Contributions

Participated in research design: Dawson, Stahl, Paul, Barber, and Kenna.

Conducted experiments: Dawson, Stahl, and Paul.

Performed data analysis: Dawson, Stahl, Paul, Barber, and Kenna.

Wrote or contributed to the writing of the manuscript: Dawson, Stahl, and Kenna.

Acknowledgments

We sincerely thank Graham Belfield, Eileen McCall, Manfred Ismair, Karen Jones, Tracy Mills, Steve Swallow, and Peter Webborn for their scientific and technical assistance and the University of Zurich (Bruno Stieger) for the supply of human BSEP baculovirus stocks.

Footnotes

  • Received May 29, 2011.
  • Accepted September 30, 2011.

References

  1. ↵
    1. Abboud G,
    2. Kaplowitz N
    (2007) Drug-induced liver injury. Drug Saf 30:277–294.
    OpenUrlCrossRefPubMed
  2. ↵
    1. Andrews E,
    2. Daly AK
    (2008) Flucloxacillin-induced liver injury. Toxicology 254:158–163.
    OpenUrlCrossRefPubMed
  3. ↵
    1. Björnsson E
    (2008) Hepatotoxicity associated with antiepileptic drugs. Acta Neurol Scand 118:281–290.
    OpenUrlCrossRefPubMed
  4. ↵
    1. Böhme M,
    2. Müller M,
    3. Leier I,
    4. Jedlitschky G,
    5. Keppler D
    (1994) Cholestasis caused by inhibition of the adenosine triphosphate-dependent bile salt transport in rat liver. Gastroenterology 107:255–265.
    OpenUrlPubMed
  5. ↵
    1. Byrne JA,
    2. Strautnieks SS,
    3. Mieli-Vergani G,
    4. Higgins CF,
    5. Linton KJ,
    6. Thompson RJ
    (2002) The human bile salt export pump: characterization of substrate specificity and identification of inhibitors. Gastroenterology 123:1649–1658.
    OpenUrlCrossRefPubMed
  6. ↵
    1. Byrne JA,
    2. Strautnieks SS,
    3. Ihrke G,
    4. Pagani F,
    5. Knisely AS,
    6. Linton KJ,
    7. Mieli-Vergani G,
    8. Thompson RJ
    (2009) Missense mutations and single nucleotide polymorphisms in ABCB11 impair bile salt export pump processing and function or disrupt pre-messenger RNA splicing. Hepatology 49:553–567.
    OpenUrlCrossRefPubMed
  7. ↵
    1. Czaja AJ
    (2011) Drug-induced autoimmune-like hepatitis. Dig Dis Sci 56:958–976.
    OpenUrlCrossRefPubMed
  8. ↵
    1. Fattinger K,
    2. Funk C,
    3. Pantze M,
    4. Weber C,
    5. Reichen J,
    6. Stieger B,
    7. Meier PJ
    (2001) The endothelin antagonist bosentan inhibits the canalicular bile salt export pump: a potential mechanism for hepatic adverse reactions. Clin Pharmacol Ther 69:223–231.
    OpenUrlCrossRefPubMed
  9. ↵
    1. Fau D,
    2. Eugene D,
    3. Berson A,
    4. Letteron P,
    5. Fromenty B,
    6. Fisch C,
    7. Pessayre D
    (1994) Toxicity of the antiandrogen flutamide in isolated rat hepatocytes. J Pharmacol Exp Ther 269:954–962.
    OpenUrlAbstract/FREE Full Text
  10. ↵
    1. Fickert P,
    2. Fuchsbichler A,
    3. Wagner M,
    4. Zollner G,
    5. Kaser A,
    6. Tilg H,
    7. Krause R,
    8. Lammert F,
    9. Langner C,
    10. Zatloukal K,
    11. et al
    . (2004) Regurgitation of bile acids from leaky bile ducts causes sclerosing cholangitis in Mdr2 (Abcb4) knockout mice. Gastroenterology 127:261–274.
    OpenUrlCrossRefPubMed
  11. ↵
    1. Floyd JS,
    2. Barbehenn E,
    3. Lurie P,
    4. Wolfe SM
    (2009) Case series of liver failure associated with rosiglitazone and pioglitazone. Pharmacoepidemiol Drug Saf 18:1238–1243.
    OpenUrlCrossRefPubMed
  12. ↵
    1. Funk C,
    2. Pantze M,
    3. Jehle L,
    4. Ponelle C,
    5. Scheuermann G,
    6. Lazendic M,
    7. Gasser R
    (2001) Troglitazone-induced intrahepatic cholestasis by an interference with the hepatobiliary export of bile acids in male and female rats. Correlation with the gender difference in troglitazone sulfate formation and the inhibition of the canalicular bile salt export pump (Bsep) by troglitazone and troglitazone sulfate. Toxicology 167:83–98.
    OpenUrlCrossRefPubMed
  13. ↵
    1. Gonzales E,
    2. Davit-Spraul A,
    3. Baussan C,
    4. Buffet C,
    5. Maurice M,
    6. Jacquemin E
    (2009) Liver diseases related to MDR3 (ABCB4) gene deficiency. Front Biosci 14:4242–4256.
    OpenUrlPubMed
  14. ↵
    1. Greaves P,
    2. Williams A,
    3. Eve M
    (2004) First dose of potential new medicines to humans: how animals help. Nat Rev Drug Discov 3:226–236.
    OpenUrlCrossRefPubMed
  15. ↵
    1. Greer ML,
    2. Barber J,
    3. Eakins J,
    4. Kenna JG
    (2010) Cell based approaches for evaluation of drug-induced liver injury. Toxicology 268:125–131.
    OpenUrlCrossRefPubMed
  16. ↵
    1. Grime K,
    2. Webborn PJ,
    3. Riley RJ
    (2008) Functional consequences of active hepatic uptake on cytochrome P450 inhibition in rat and human hepatocytes. Drug Metab Dispos 36:1670–1678.
    OpenUrlAbstract/FREE Full Text
  17. ↵
    1. Hirano H,
    2. Kurata A,
    3. Onishi Y,
    4. Sakurai A,
    5. Saito H,
    6. Nakagawa H,
    7. Nagakura M,
    8. Tarui S,
    9. Kanamori Y,
    10. Kitajima M,
    11. et al
    . (2006) High-speed screening and QSAR analysis of human ATP-binding cassette transporter ABCB11 (bile salt export pump) to predict drug-induced intrahepatic cholestasis. Mol Pharm 3:252–265.
    OpenUrlCrossRefPubMed
  18. ↵
    1. Ho RH,
    2. Leake BF,
    3. Kilkenny DM,
    4. Meyer Zu Schwabedissen HE,
    5. Glaeser H,
    6. Kroetz DL,
    7. Kim RB
    (2010) Polymorphic variants in the human bile salt export pump (BSEP; ABCB11): functional characterization and interindividual variability. Pharmacogenet Genomics 20:45–57.
    OpenUrlCrossRefPubMed
  19. ↵
    1. Horikawa M,
    2. Kato Y,
    3. Tyson CA,
    4. Sugiyama Y
    (2003) Potential cholestatic activity of various therapeutic agents assessed by bile canalicular membrane vesicles isolated from rats and humans. Drug Metab Pharmacokinet 18:16–22.
    OpenUrlCrossRefPubMed
  20. ↵
    1. Ito K,
    2. Chiba K,
    3. Horikawa M,
    4. Ishigami M,
    5. Mizuno N,
    6. Aoki J,
    7. Gotoh Y,
    8. Iwatsubo T,
    9. Kanamitsu S,
    10. Kato M,
    11. et al
    . (2002) Which concentration of the inhibitor should be used to predict in vivo drug interactions from in vitro data? AAPS PharmSci 4:E25.
    OpenUrlCrossRefPubMed
  21. ↵
    1. Jupp R,
    2. Dusanjh PK,
    3. Walding A,
    4. McHale M,
    5. Belfield GP,
    6. Delaney SJ
    (2006) Rapid simultaneous cloning of drug targets from multiple mammalian species. Altern Lab Anim 34:295–303.
    OpenUrlPubMed
  22. ↵
    1. Kartenbeck J,
    2. Leuschner U,
    3. Mayer R,
    4. Keppler D
    (1996) Absence of the canalicular isoform of the MRP gene-encoded conjugate export pump from the hepatocytes in Dubin-Johnson syndrome. Hepatology 23:1061–1066.
    OpenUrlPubMed
  23. ↵
    1. Keserü GM,
    2. Makara GM
    (2009) The influence of lead discovery strategies on the properties of drug candidates. Nat Rev Drug Discov 8:203–212.
    OpenUrlCrossRefPubMed
  24. ↵
    1. Kis E,
    2. Ioja E,
    3. Nagy T,
    4. Szente L,
    5. Herédi-Szabó K,
    6. Krajcsi P
    (2009) Effect of membrane cholesterol on BSEP/Bsep activity: species specificity studies for substrates and inhibitors. Drug Metab Dispos 37:1878–1886.
    OpenUrlAbstract/FREE Full Text
  25. ↵
    1. Mano Y,
    2. Usui T,
    3. Kamimura H
    (2007) Effects of bosentan, an endothelin receptor antagonist, on bile salt export pump and multidrug resistance-associated protein 2. Biopharm Drug Dispos 28:13–18.
    OpenUrlCrossRefPubMed
  26. ↵
    1. Morgan RE,
    2. Trauner M,
    3. van Staden CJ,
    4. Lee PH,
    5. Ramachandran B,
    6. Eschenberg M,
    7. Afshari CA,
    8. Qualls CW Jr.,
    9. Lightfoot-Dunn R,
    10. Hamadeh HK
    (2010) Interference with bile salt export pump function is a susceptibility factor for human liver injury in drug development. Toxicol Sci 118:485–500.
    OpenUrlAbstract/FREE Full Text
  27. ↵
    1. Noé J,
    2. Stieger B,
    3. Meier PJ
    (2002) Functional expression of the canalicular bile salt export pump of human liver. Gastroenterology 123:1659–1666.
    OpenUrlCrossRefPubMed
  28. ↵
    1. Noe J,
    2. Hagenbuch B,
    3. Meier PJ,
    4. St-Pierre MV
    (2001) Characterization of the mouse bile salt export pump overexpressed in the baculovirus system. Hepatology 33:1223–1231.
    OpenUrlCrossRefPubMed
  29. ↵
    1. Noe J,
    2. Kullak-Ublick GA,
    3. Jochum W,
    4. Stieger B,
    5. Kerb R,
    6. Haberl M,
    7. Müllhaupt B,
    8. Meier PJ,
    9. Pauli-Magnus C
    (2005) Impaired expression and function of the bile salt export pump due to three novel ABCB11 mutations in intrahepatic cholestasis. J Hepatol 43:536–543.
    OpenUrlCrossRefPubMed
  30. ↵
    1. Perez MJ,
    2. Briz O
    (2009) Bile-acid-induced cell injury and protection. World J Gastroenterol 15:1677–1689.
    OpenUrlCrossRefPubMed
  31. ↵
    1. Stieger B,
    2. Fattinger K,
    3. Madon J,
    4. Kullak-Ublick GA,
    5. Meier PJ
    (2000) Drug- and estrogen-induced cholestasis through inhibition of the hepatocellular bile salt export pump (Bsep) of rat liver. Gastroenterology 118:422–430.
    OpenUrlCrossRefPubMed
  32. ↵
    1. Strautnieks SS,
    2. Bull LN,
    3. Knisely AS,
    4. Kocoshis SA,
    5. Dahl N,
    6. Arnell H,
    7. Sokal E,
    8. Dahan K,
    9. Childs S,
    10. Ling V,
    11. et al
    . (1998) A gene encoding a liver-specific ABC transporter is mutated in progressive familial intrahepatic cholestasis. Nat Genet 20:233–238.
    OpenUrlCrossRefPubMed
  33. ↵
    1. Strautnieks SS,
    2. Byrne JA,
    3. Pawlikowska L,
    4. Cebecauerová D,
    5. Rayner A,
    6. Dutton L,
    7. Meier Y,
    8. Antoniou A,
    9. Stieger B,
    10. Arnell H,
    11. et al
    . (2008) Severe bile salt export pump deficiency: 82 different ABCB11 mutations in 109 families. Gastroenterology 134:1203–1214.
    OpenUrlCrossRefPubMed
  34. ↵
    1. Stricker BHCH
    (1992) Drug Induced Liver Injury, 2nd ed, Elsevier, Amsterdam.
  35. ↵
    1. Thompson RA,
    2. Isin EM,
    3. Li Y,
    4. Weaver R,
    5. Weidolf L,
    6. Wilson I,
    7. Claesson A,
    8. Page K,
    9. Dolgos H,
    10. Kenna JG
    (2011) Risk assessment and mitigation strategies for reactive metabolites in drug discovery and development. Chem Biol Interact 192:65–71.
    OpenUrlCrossRefPubMed
  36. ↵
    1. Wang R,
    2. Chen HL,
    3. Liu L,
    4. Sheps JA,
    5. Phillips MJ,
    6. Ling V
    (2009) Compensatory role of P-glycoproteins in knockout mice lacking the bile salt export pump. Hepatology 50:948–956.
    OpenUrlCrossRefPubMed
  37. ↵
    1. Wang R,
    2. Lam P,
    3. Liu L,
    4. Forrest D,
    5. Yousef IM,
    6. Mignault D,
    7. Phillips MJ,
    8. Ling V
    (2003) Severe cholestasis induced by cholic acid feeding in knockout mice of sister of P-glycoprotein. Hepatology 38:1489–1499.
    OpenUrlCrossRefPubMed
  38. ↵
    1. Yabuuchi H,
    2. Tanaka K,
    3. Maeda M,
    4. Takemura M,
    5. Oka M,
    6. Ohashi R,
    7. Tamai I
    (2008) Cloning of the dog bile salt export pump (BSEP; ABCB11) and functional comparison with the human and rat proteins. Biopharm Drug Dispos 29:441–448.
    OpenUrlCrossRefPubMed
  39. ↵
    1. Zimmerman HJ
    (1999) Hepatotoxicity: the Adverse Effects of Drugs and Other Chemicals on the Liver, 2nd ed, Lippincott Williams & Wilkins, Philadelphia.
Back to top

In this issue

Download PDF
Article Alerts
Email Article
Citation Tools

Research ArticleArticle

BSEP INHIBITION CORRELATES WITH RISK OF CHOLESTASIS

Sarah Dawson, Simone Stahl, Nikki Paul, Jane Barber and J. Gerald Kenna
Drug Metabolism and Disposition January 1, 2012, 40 (1) 130-138; DOI: https://doi.org/10.1124/dmd.111.040758
Twitter logo Facebook logo Mendeley logo

Jump to section

Advertisement