ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2013

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Pharmacy 172

ATP-Binding-CassetteTransporters in Biliary Efflux andDrug-Induced Liver Injury

JENNY M. PEDERSEN

ISSN 1651-6192ISBN 978-91-554-8702-7urn:nbn:se:uu:diva-204200

Dissertation presented at Uppsala University to be publicly examined in B21, BMC,Husargatan 3, Uppsala, Sweden, Friday, September 6, 2013 at 09:15 for the degree of Doctorof Philosophy (Faculty of Pharmacy). The examination will be conducted in English.

AbstractPedersen, J. M. 2013. ATP-Binding-Cassette Transporters in Biliary Efflux and Drug-InducedLiver Injury. Acta Universitatis Upsaliensis. Digital Comprehensive Summaries of UppsalaDissertations from the Faculty of Pharmacy 172. 67 pp. Uppsala. ISBN 978-91-554-8702-7.

Membrane transport proteins are known to influence the absorption, distribution, metabolism,excretion and toxicity (ADMET) of drugs. At the onset of this thesis work, only a fewstructure-activity models, in general describing P-glycoprotein (Pgp/ABCB1) interactions, weredeveloped using small datasets with little structural diversity. In this thesis, drug-transportprotein interactions were explored using large, diverse datasets representing the chemicalspace of orally administered registered drugs. Focus was set on the ATP-binding cassette(ABC) transport proteins expressed in the canalicular membrane of human hepatocytes.Theinhibition of the ABC transport proteins multidrug-resistance associated protein 2 (MRP2/ABCC2) and bile salt export pump (BSEP/ABCB11) was experimentally investigated usingmembrane vesicles from cells overexpressing the investigated proteins and sandwich culturedhuman hepatocytes (SCHH). Several previously unknown inhibitors were identified for both ofthe proteins and predictive in silico models were developed. Furthermore, a clear associationbetween BSEP inhibition and clinically reported drug induced liver injuries (DILI) wasidentified. For the first time, an in silico model that described combined inhibition of Pgp, MRP2and breast cancer resistance protein (BCRP/ABCG2) was developed using a large, structurallydiverse dataset. Lipophilic weak bases were more often found to be general ABC inhibitors incomparison to other drugs. In early drug discovery, in silico models can be used as predictivefilters in the drug candidate selection process and membrane vesicles as a first experimentalscreening tool to investigate protein interactions.In summary, the present work has led to anincreased understanding of molecular properties important in ABC inhibition as well as thepotential influence of ABC proteins in adverse drug reactions. A number of previously unknownABC inhibitors were identified and predictive computational models were developed.

Keywords: ABC transport protein, Pgp, P-glycoprotein, ABCB1, BCRP, breast cancerresistance protein, ABCG2, MRP2, multidrug resistance-associated protein 2, ABCC2, BSEP,bile salt export pump, ABCB11, sandwich cultured human hepatocytes, SCHH, drug-inducedliver injury, DILI, multivariate data analysis, OPLS

Jenny M. Pedersen, Uppsala University, Department of Pharmacy, Box 580,SE-751 23 Uppsala, Sweden.

© Jenny M. Pedersen 2013

ISSN 1651-6192ISBN 978-91-554-8702-7urn:nbn:se:uu:diva-204200 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-204200)

The beautiful thing about today is that you get the choice

to make it better than yesterday.

Robin Sharma

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Pedersen, J. M., Matsson, P., Norinder, U., Bergström, C.

A. S., Hoogstraate, J. and Artursson, P. (2008) Prediction and Identification of Drug Interactions with the Human ATP-Binding Cassette Transporter Multidrug-Resistance Associated Protein 2 (MRP2; ABCC2). J Med Chem. (51), 3275-87.

II Matsson, P., Pedersen, J. M., Norinder, U., Bergström, C. A. S. and Artursson, P. (2009) Identification of Novel Spe-cific and General Inhibitors of the Three Major Human ATP-Binding Cassette Transporters P-gp, BCRP and MRP2 Among Registered Drugs. Pharm Res. (26), 1816-31.

III Pedersen, J. M., Svedberg, E., Lai, Y., Qiu, X., Norén, A., Palm, J., Hoogstraate, J., Artursson, P. Biliary Drug Excre-tion in Sandwich Cultured Human Hepatocytes: Predicted from Inverted Membrane Vesicles and Targeted Proteomics. Manuscript.

IV Pedersen, J. M., Bergström, C. A. S., Matsson, P., Norén, A., Hoogstraate, J., LeCluyse, E. and Artursson, P. Early Identification of Clinically Relevant Drug Interactions with the Human Bile Salt Export Pump (BSEP; ABCB11). Ac-cepted for publication in Toxicological Sciences.

Reprints were made with permission from the respective publishers.

Contents

List of Papers .................................................................................................. v

Abbreviations ................................................................................................. ix

Introduction ................................................................................................... 11 Membrane Transport of Drugs in Human ................................................ 12

Membrane Transport Proteins ............................................................. 14 Drug-Drug Interactions (DDI) ............................................................. 18

The Human Liver ..................................................................................... 19 Drug Disposition in the Liver .............................................................. 20 Cellular Models for Hepatobiliary Transport Investigations ............... 22 Drug-induced Liver Injury (DILI) ....................................................... 25

Computational Modeling of Transport Protein Interactions .................... 26

Aims of This Thesis ...................................................................................... 29

Methods ........................................................................................................ 30 Dataset Selection ...................................................................................... 30 Experimental Methods ............................................................................. 30

ABC Transport in Membrane Vesicles ................................................ 30 Inhibition of ABC Transport in Membrane Vesicles ........................... 31 Inhibition of Efflux in BCRP-Overexpressing Cells ........................... 32 Isolation of Primary Human Hepatocytes ............................................ 32 Culture of Human Hepatocytes ........................................................... 32 Biliary Excretion in SCHH .................................................................. 33 Identification of Compounds Causing DILI ........................................ 34 Pgp, BCRP and MRP2 Protein Expression ......................................... 35 Canalicular Efflux Predicted by the Maximal Transport Activity ....... 36

Computational Methods ........................................................................... 37 Generation of Molecular Descriptors .................................................. 37 Principal Component Analysis - PCA ................................................. 37 Partial Least Squares Projection to Latent Structures (PLS) ............... 37 Virtual Docking ................................................................................... 38

Statistical Analysis ................................................................................... 38

Results and Discussion ................................................................................. 39 Dataset Diversity ...................................................................................... 39

Identification of MRP2 and BSEP Inhibitors ........................................... 39 Overlapping Inhibition of the Canalicular ABC Proteins ........................ 42 Molecular Properties Important in ABC Inhibition .................................. 43 ABC Inhibition in Cellular and Vesicular Models ................................... 44 Prediction of Hepatobiliary Transport of E17G ....................................... 45 BSEP Inhibition and DILI ........................................................................ 46 Analyzing Bile Excretion in SCHH ......................................................... 49

Conclusions ................................................................................................... 50

Future Perspectives ....................................................................................... 51

Svensk populärvetenskaplig sammanfattning ............................................... 52

Acknowledgements ....................................................................................... 54

References ..................................................................................................... 56

Abbreviations

3D Three dimensional ABC ATP binding cassette ADMET Absorption, distribution, metabolism,

excretion and toxicity ADR Adverse drug reaction AMP Adenosine monophosphate ANOVA Analysis of variance AR Adverse reaction; FDA classification ATP Adenosine triphosphate BBB Blood brain barrier BCRP Breast cancer resistance protein BEI Biliary excretion index BIC Bile intracellular correlation BileAcc Bile accumulation in SCHH BileAcc, MTA Predicted bile accumulation in SCHH BSA Bovine serum albumin BSEP Bile salt export pump BW Boxed warnings; FDA classification CLBile Biliary clearance CLBile, MTA Predicted in vivo biliary clearance CycA Cyclosporine A CYP Cytochrome P450 enzyme DA Discriminant analysis DDI Drug-drug interactions DILI Drug induced liver injuries DMEM Dulbecco’s modified Eagle’s medi-

um DMSO Dimethyl sulfoxide E17G Estradiol 17-glucuronide E3S Estrone 3-sulfate EDTA Ethylenediaminetetraacetic acid EMA European Medicines Agency FDA The Food and Drug Administration HBSS Hank's Balanced Salt Solution HomProt Total protein amount in human liver

homogenate

ICAcc Intracellular accumulation in SCHH Km Concentration corresponding to half

maximal transport rate; Michaelis-Menten transport kinetics parameter

LogD7.4 Octanol water partition coefficient at pH 7.4

MDR Multidrug-resistance MRP Multidrug-resistance associated pro-

tein MSD Membrane spanning domain MTA Maximal transport activity NBD Nucleotide binding domain NMQ N-methyl quinidine OPLS Orthogonal partial least squares pro-

jection to latent structures PCA Principal component analysis Pgp P-glycoprotein PLS Partial least squares projection to

latent structures QSAR Quantitative structure-activity rela-

tionship SCHH Sandwich cultured human hepato-

cytes SLC Solute carrier transport protein TA Taurocholate TMD Transmembrane domain TotalAcc Summarized intracellular- and bile-

accumulation in SCHH Vmax Maximal transport rate; Michaelis-

Menten kinetics parameter WP Warnings and precautions; FDA clas-

sification

11

Introduction

New drugs are continuously being developed with the intention to prevent, cure, and treat diseases. During the last century, drugs developed on the ba-sis of important pharmacological discoveries, such as antibiotics, markedly increased human life expectancy [1]. Today, drug development is a lengthy, complex, and expensive effort. Even if exact quantifications are difficult, recent estimates (2012) state that the drug development process takes 10-15 years and costs more than one billion (~1,200,000,000) US dollars. The sta-tistics further reveal that only two of ten marketed drugs return revenues that reach or exceed their development cost [2]. The complexity of the process is well reflected in the interdisciplinary cooperation of scientists and clinical researchers required for the successful development of a new drug. Never-theless, the number of new drug applications approved by the American regulatory agency has increased 55% during the 1990s and the first decade of this century. Statistics from the Food and Drug Administration (FDA) show that the flow of drugs involving new molecular entities instead is de-creasing [3]. The latter may indeed reflect the complexity of drug develop-ment, including the constantly increasing challenge from surpassing the effi-cacy and safety level of already marketed drugs.

In the 1990s, studies revealed that poor pharmacokinetic properties were the main reason for compound attrition during drug development [4, 5]. Previ-ously, pharmacokinetics had played only a minor role in the drug develop-ment process. To ensure that substandard compounds were eliminated at an early stage from the development process, the industry began to implement rigorous strategies for detection of inadequate pharmacokinetic properties, which resulted in the development of new in silico and in vitro methods. This preventive approach has proven successful and in the year 2000 the attrition rate due to pharmacokinetic properties had dropped to less than ten percent [1]. However, new models able to predict toxicological responses are still warranted.

At the same time, studies of knock-out animal models and loss-of-function human genetic variants suggested that the bioavailability, disposition, and adverse reactions of drugs were more influenced by transport proteins than what was previously believed [6, 7]. During the last 20 years numerous stud-ies on drug-transporter interactions have provided important information on

12

how transport protein interactions influence the pharmacokinetic properties of drugs [8-11]. Transport proteins are now thought to work in concert with passive permeability and other proteins, such as drug metabolizing enzymes, in the absorption and elimination of drugs [9]. As a consequence of these research efforts, several regulatory agencies including the European Medi-cines Agency (EMA) and the American FDA have included drug-transporter interaction studies in their industry guidelines [12, 13]. In this thesis, drug-transporter interactions were examined to better understand the influence of transport proteins on biliary excretion, drug-drug interactions (DDI), and drug-induced liver injuries (DILI). In particular, this thesis focuses on the ATP-binding- cassette (ABC) transport proteins expressed in the canalicular membrane of human hepatocytes.

Membrane Transport of Drugs in Human Oral administration is generally the preferred administration route of drugs. In addition to low production costs, the ease of administration can improve patient compliance[14]. To achieve a systemic effect, all orally-administered drugs need to pass the intestinal epithelium to reach the circulatory system. Depending on their site of action, most drugs also have to overcome one or more additional membrane barrier, such as the plasma membranes (for intra-cellular effects) or the blood-brain barrier (BBB, for CNS effects).

The fundamental structure of the membrane barrier is a phospholipid bilayer with aqueous compartments on either side (Fig. 1). In cell barriers composed of adherent cells, two phospholipid bilayers and the aqueous intracellular compartment separate the two extracellular aqueous compartments on either side (Fig. 1B). In addition to the cells themselves, the integrity of a cell bar-rier is sustained by adherent- and tight junctions. The constituents of the phospholipid membranes and the cell junction integrity play a major role on the overall cell membrane permeability [15, 16]. Through adjusting the membrane composition, e.g., by protein expression, cholesterol load, and tight junction integrity, evolution has altered the membrane permeability to fit the purpose of a specific tissue. For instance, in the BBB that works to protect the brain from harmful compounds, the tight junction integrity is approximately 20 times higher compared to that of the jejunum [17]. The diverse membrane properties of individual tissues, e.g., transport protein expression, influence the relative importance of different transport routes, but in general, drug transport across a cell barrier depends on three distinct pathways, described below (Fig. 1).

13

Figure 1. The phospholipid bilayer of the cellular membrane contain hydrophilic head-groups (1) towards the outside aqueous compartments and hydrophobic carbon chains (2) in between. The cholesterol content (3) within the hydrophobic in-tramembranous compartment influences the membrane fluidity and passive diffusion (A). Transport proteins (B) in the cellular membranes can facilitate drug transport over cellular membranes in an ATP-dependent manner or by facilitated diffusion. Small hydrophilic compounds can pass cellular membranes through the paracellular transport route (C) between adjacent cells.

Passive Diffusion For many drugs, transcellular passive diffusion (Fig. 1A) is the major transport route in absorption and tissue distribution. Passive diffusion is a chemical movement from high to low concentration driven by the gain of system entropy. The extent of passive diffusion over a cellular membrane is determined by the molecular properties of the drug and the membrane com-position. Passive diffusion benefits from low molecular weight and high lipophilicity of the compound, as well as low cholesterol load of the mem-brane [18, 19]. Large molecules or particles can permeate a membrane barri-er through transcellular transport in membrane vesicles by endocytosis. However, this transport route has little influence on transport of small mole-cule drugs.

Paracellular Transport Tight junction integrity determines the size of the pores that allow paracellu-lar transport (Fig. 1C). Mainly due to the small fraction of the total mem-brane area comprised of paracellular pores, this route influences the mem-brane permeability of small hydrophilic drugs that are less able to permeate the phospholipid bilayer [20]. It has been shown that paracellular transport mostly influences the intestinal drug permeability of poorly absorbed drugs [21].

ATPADP

Passive diffusion

Facilitatedtransport

Activetransport

Paracellulartransport

1

2

3

4

A B C

14

Active Transport In addition to passive diffusion, membrane transport proteins (Fig. 1B) spanning the cellular membranes are important contributors to the membrane permeability of drugs [8, 10]. The transport capacity of the active transport is dependent on protein expression levels and transport efficiency and these parameters, in combination with the extent of passive permeability, deter-mine the influence of active transport on the membrane permeability.

Membrane Transport Proteins Membrane transport proteins can be regarded as gatekeepers for all cells and organelles in the body. They control the uptake and efflux of many com-pounds and are crucial in maintaining membrane potential or chemical gra-dients essential to many cell functions. To date, more than 400 membrane transport proteins have been cloned and many of them have been localized to tissues and organs throughout the human body [22]. The primary functions for most of these transport proteins include membrane transport of endoge-nous substrates such as sugars, amino acids, nucleotides or vitamins, or pro-tecting the body from dietary or environmental toxins. However, substrate specificities of the membrane transport proteins are not strictly limited to their physiological substrates, and about 20 of them are considered relevant to the absorption, distribution, metabolism, excretion and toxicity (ADMET) of drugs [23, 24].

Membrane transport proteins are important contributors in the membrane barrier function in e.g., the intestine [25, 26]. Membrane barriers are often comprised of polarized cells with diverse protein expression in the cellular domains facing the two sides of the membrane. The expression of drug transport proteins is generally highly restricted to one side, either the baso-lateral or apical domain, of the polarized cells [27, 28]. Transport proteins in the two membrane domains often work collaboratively to facilitate the sub-strate transport over the entire cell barrier [9, 29, 30]. Transport proteins differ in their two- and three-dimensional protein structures and they transport their substrates via different transport mechanisms. On the basis of these molecular and mechanistic differences, the transport proteins are orga-nized into protein superfamilies. The two main transport protein superfami-lies involved in drug transport are the solute carriers (SLC) and ABC pro-teins [31, 32].

Solute-Carrier-Transport (SLC) Proteins Most identified transport proteins are SLC transporters and, in total, more than 350 members of this family have been identified [31]. SLC transporters translocate their substrates across biological membranes through several

15

different mechanisms, including facilitated diffusion, ion coupling, and ion exchange. Depending on which driving force the SLC transport proteins use, they can be further classified into facilitated transporters and secondary ac-tive transporters. The facilitated transporters move their substrates with the concentration gradient and the transport is therefore not energy dependent. In contrast, secondary active transporters translocate their substrate using membrane gradients produced by other energy consuming transport proteins. This transport process is thus energy-dependent, even if the transport pro-teins themselves are unable to hydrolyze ATP to produce the energy for the transport process [33-35].

Figure 2. Predicted membrane topology of SLC transport proteins. The pore through the membrane consists of 12 TMDs. Glycosylation of the extramembranous loops regulates protein function and cellular trafficking.

Most SLC proteins are located in the plasma membrane where they translo-cate substrates from the extracellular to the intracellular compartment. How-ever, some of them are expressed in other cellular membranes such as the mitochondrial membrane [36]. Generally, members of the SLC protein fami-ly have a similar protein structure that consists of 12 putative membrane-spanning domains (Fig. 2) and their molecular mass is approximately be-tween 50 and 100 kDa [37, 38]. Several of the SLC transport proteins are ubiquitously expressed throughout the human body (e.g., SGLT/SLC5 and SNAT/SLC38) and they participate in essential transport processes including cellular uptake of glucose and amino acids [39, 40]. Other SLC proteins are specifically expressed in one or few tissues, such as the OATP1B1/SLCO1B1 that is involved in the hepatic uptake of bile acids and organic anions [41]. SLC transport proteins involved in drug transport are usually not expressed ubiquitously; rather, they are located in tissues im-portant to the ADMET properties of drugs, such as the intestine, liver and kidney [25, 26].

ATP-Binding-Cassette (ABC) Transport Proteins The ABC-transport protein family is one of the most widely expressed pro-tein families known. In humans, 48 proteins organized into seven subfami-

16

lies have been identified [42]. ABC proteins in the plasma membrane trans-locate their substrate from the intracellular to the extracellular compartment, but some of them, e.g., the phospholipid transporter ABCB4, only facilitate transport over one of the membrane leaflets [43]. The dislocation of the sub-strate is energy dependent and the substrate transport often occurs against steep concentration gradients. The ABC-transport proteins produce the ener-gy required for this transport through the binding, and subsequent hydrolysis of, ATP at their two nucleotide-binding domains (NBD). The NBDs consist of two highly conserved amino acid sequences (the Walker A and B motifs) that the ABC proteins share with other ATP binding proteins. However, the ABC proteins also consist of a third, highly-conserved amino acid sequence, located between the Walker A and B motifs. This third motif is unique to the ABC proteins, and often referred to as the ABC signature motif or the C motif [42]. High-resolution crystal structures of NDBs from the ABC pro-teins suggest that this motif participate in the recognition, binding, and hy-drolysis of ATP [44].

Figure 3. Structure and membrane topology of ABC transport proteins. (A) The high resolution crystal structure from mouse Pgp (Protein Data Bank ID 3G5U) show the two intracellular NBD and the alphahelices constituting the TMD. The grey area shows the approximate position of the cell membrane. (B) The predicted membrane topology of ‘full transporters’ including Pgp, MRP4 and MRP5. In these proteins the entire core unit of two MSD and two NBD are encoded for in a single polypeptide. (C) In addition to the core unit MRP1-3 and MRP6 have an additional MSD contain-ing five TMDs. (D) Half transporters such as the ABCG family, including BCRP, assemble homo- or heterodimers to form the functional core unit of the full trans-porters. Figure A altered from Aller et al. 2009 [45].

A high resolution crystal structure of a human ABC protein, the mitochon-drial ABCB10, was recently resolved for the first time [46]. Among the de-termined ABC crystal structures, mouse-Pgp (Protein Data Bank ID 3G5U; Fig. 3A) has the highest sequence homology to the human multidrug re-sistance-associated (MDR) proteins. For instance, human-Pgp has a se-quence homology of 87% with the mouse-Pgp, but only 37% sequence ho-

B

C

D

Extracellular

Intracellular

NBD1 NBD2

A

NBD1 NBD2

NBD1 NBD2

NBD1

17

mology to the human ABCB10 [45, 47]. Homology models used to predict human Pgp (or other MDR) ligand interactions are therefore often based on the crystal structure of mouse-Pgp. By general consensus, the core unit in functional ABC proteins consists of two membrane-spanning domains (MSDs) and two NBDs. The MSDs contain, in total, 12 transmembrane do-mains (TMDs) that are believed to constitute the channel that enables the transmembrane dislocation of the substrates [32, 42]. The entire core unit can be encoded in a single polypeptide chain, as is the case of the ‘full trans-porters’ (e.g., Pgp and Multidrug-resistance-associated protein 5/MRP5) (Fig. 3B) [48]. Alternatively, the core unit can assembles as a homo- or het-erodimer of peptides, each containing one MSD and one NBD. These trans-porters are often referred to as ‘half transporters’ and are exemplified by the ABCG family that includes the Breast-Cancer-Resistance Protein (BCRP; Fig. 3D) [49]. In addition to these four core functional units, MRP1-3 and MRP6 also comprise a third MSD of five transmembrane helices (Fig. 3C). The exact function of this domain has not been elucidated but it possibly aids in the membrane localization of the ABC protein [50, 51]. Despite these differences in domain organization, there is considerable overlap in the sub-strate specificities between several of the included transport proteins.

The ABC proteins accept a wide variety of substrates, including phospholip-ids, ions, peptides, steroids, polysaccharides, and bile acids [52, 53]. The physiological role of the ABC proteins becomes apparent in patients with low or no functional expression of individual ABC proteins. For instance, patients who lack MRP2 function develop chronic conjugated hyperbiliru-binemia due to low hepatobiliary excretion of bilirubin [52]. Other examples include the Tangier and Stargardt diseases, where ABC malfunctions lead to disturbed cholesterol balance and macular dystrophy, respectively [54, 55]. In addition to the large number of endogenous ABC substrates, many drugs and other xenobiotics also undergo ABC transport [24]. The role that ABC proteins play in MDR was first recognized in the context of cancer therapy, but now the ABC proteins are generally believed to influence pharmacoki-netic properties of drugs within most therapeutic classes [24, 56]. Three main subfamilies (ABCB, ABCC, and ABCG) are known to be involved in MDR [8]. Members of these families are distributed among all of the mem-brane barriers, such as the BBB, intestine and placenta. It is likely that these subfamilies, have evolved to aid in the excretion of cellular waste products and to protect the human body from xenobiotic exposure [25, 57, 58]. In this thesis, particular emphasis is placed on investigating the influence on the hepatobiliary excretion of drugs and endogenous compounds by four ABC transport proteins, the Pgp, BCRP, MRP2 and BSEP.

18

Drug-Drug Interactions (DDI)

To generate the intended effect, an administered drug needs to interact with its target site. This target site often consists of a protein expressed, for in-stance, in the plasma or nuclear membranes [59]. In addition to the target site, drugs also interact with other cellular components, including metabolic enzymes and transport proteins. These proteins are often of crucial im-portance to the drug bioavailability, efficacy, and safety [22, 60]. If a drug is co-administered with other drugs or some food, competitive protein interac-tions increase the risk of altered pharmacokinetic properties. Possibly, such interactions lead to clinically relevant DDIs where one or more of the admin-istered drugs lose effect or generate adverse drug reactions (ADR) [60, 61].

DDI mechanisms are generally related to inhibition or stimulation of pro-teins involved in the drug binding to the target site or to proteins influencing the metabolism and excretion of the drug. In the latter, reversible inhibition is responsible for the majority of all DDIs with clinical relevance [61]. As mentioned previously, transport proteins and metabolic enzymes work in concert to reduce the load of drugs in the body. Examples of DDI mecha-nisms including these proteins are:

1) Inhibition of hepatic SLC or metabolizing enzymes that decrease drug metabolism. This may lead to increased systemic drug concen-tration that in turn increases the risk of ADR.

2) Inhibition of metabolizing enzymes or intestinal-, BBB-, or kidney- ABC proteins can lead to increased bioavailability and systemic drug concentrations increasing the risk of ADR.

3) Inhibition of ABC transporters in the liver and kidney may cause increased intracellular or systemic levels of possibly toxic metabo-lites causing cellular damage or systemic adverse reactions. This can also lead to decreased drug excretion, increased systemic drug con-centration, and increased risk of ADR.

4) Inhibition of intestinal uptake transport proteins can reduce the bioa-vailability of an administered drug and therefore cause lack of effect.

5) Stimulation of hepatic transport proteins and metabolic enzymes can reduce systemic drug concentrations and therefore cause loss of drug efficacy.

19

The Human Liver After the skin, the liver is the largest organ in the human body and comprises of up to five percent of the adult body weight. It serves as the primary site of detoxification of natural and synthetic compounds in the systemic circulation and therefore plays an important role in human physiology. In addition to its impact on systemic detoxification, the liver synthesizes and secretes im-portant blood and bile main constituents, such as albumin, bile salts, and cholesterol. Other important functions include protein, steroid, and fat me-tabolism, vitamin and hormone synthesis, iron homeostasis, and sugar stor-age [62, 63]. The liver is a highly specialized tissue that comprises many different cell types, such as endothelial cells, macrophage Kupffer cells, and a small stem cell population called oval cells [64]. However, most of the functions attributed to the liver are performed by the parenchymal cell type, the hepatocytes. Hepatocytes compose approximately 70% of the hepatic cell population and are highly differentiated cells originating from fetal en-dothelial cells [63, 64]. They are highly polarized and depend on the mainte-nance of two distinct membrane domains to maintain normal hepatocyte functions [64, 65].

Figure 4. (A) The liver consists of 50, 000 to 100, 000 functional units called liver lobule [62]. The blood enters the liver at the portal triad (PT) and exits into the cen-tral vein (CV). (B) The PT comprises single branches of the portal vein (blue), he-patic artery (red) and bile duct (green). The blood from the portal vein and the hepat-ic artery mixes and travels through the microvasculature to a branch of the hepatic vein that continues to the CV. The bile flow is reversed and exits the liver at the PT. (C) The hepatic lobules are divided into three zones with different microenviron-ment and cell phenotypes. Zonal differences in hepatocyte phenotypes can be ob-served in e.g., expression of metabolic enzymes and transport proteins. For instance, a higher expression of the transport protein OATP1B1 can be observed in pericentral (zone III) hepatocytes, as shown in Figure 4C [66].

The liver is closely linked to the small intestine and processes the venous blood flow from the digestive tract containing all absorbed molecules such as nutrients and drugs. Most of the hepatic blood flow originates from the gastrointestinal tract and enters the liver from the portal vein. However, oxy-

CV PT

A B CCV

PT

PT

CV

Zone 1

Zone 2

Zone 3

20

genated blood also enters the liver from the hepatic artery. After passing through the sinusoidal microvasculature, the blood exits through the hepatic central vein. Histologically the liver consists of small functional units called hepatic lobules (Fig. 4A), with a centralized branch from the hepatic vein, connected to the central vein (CV), surrounded by several portal triads, each consisting of branches from the portal vein, hepatic artery, and bile ducts (Fig. 4B) [62]. Between the portal triad and the central vein, the liver is both structurally and functionally further divided into three zones with distinct differences in cell phenotype and maturation. There are also zonal variations in chemical concentrations e.g., of oxygen, nutrients, insulin, and glucagon [63]. The diverse microenvironments and cellular functions generate the zonal variations observed in many important hepatic functions. For example, hepatocytes around the central vein have a higher density of endoplasmic reticulum and express the highest levels of several metabolizing enzymes and uptake transporters involved in detoxification of e.g. drugs (Fig. 4C) [64, 66].

Drug Disposition in the Liver Only the free fraction of a drug is available for passive or active uptake into the hepatocytes. Nonetheless, the liver has a remarkable capacity to extract protein-bound drugs from the systemic circulation. For example, atorvastatin is well absorbed in the intestine and has plasma protein binding of 98%, yet, an absolute bioavailability of 14% [67]. Until approximately 20 years ago, it was generally assumed that the hepatobiliary clearance of drugs and drug metabolites was determined by passive diffusion, governed by the physico-chemical properties of molecules. However, it is now well established that transport proteins are important contributors to the hepatic uptake and efflux over the basolateral and apical membranes of the hepatocyte [24, 31, 68]. In this way and in cooperation with the metabolic enzymes, the transport pro-teins significantly influence the bile and systemic blood concentration of drugs and their metabolites [30, 69, 70].

Thanks to advances in molecular biology, all the major transport systems involved in the hepatobiliary transport of drugs and endogenous compounds have been cloned and characterized during the last 20 years [71, 72]. As a result we now know the identity of and some of the preferred substrates of the transport proteins involved in the basolateral uptake, basolateral efflux and canalicular efflux of drugs in the liver [23, 24].

Basolateral Uptake Basolateral uptake is the first step in hepatic clearance of compounds from the systemic circulation. The hepatic uptake is facilitated by SLC transport proteins abundantly expressed in the basolateral membrane of the hepato-

21

cytes [25]. Transport proteins with major contribution to the hepatic uptake of drugs have been identified in the SLC22A and SLCO superfamilies. They include the organic anion transporters, organic cation transporters, and the organic anion transporting polypeptides [73]. Members of these transport protein families support the uptake of a large number of substrate groups, including organic anions and cations, prostaglandins, and bile salts [74-76]. However, bile salt uptake is mainly performed by the Na+/taurocholate transporting polypeptide in the SLC10 superfamily [53].

Hepatic uptake transporters influence the pharmacokinetic properties of drugs by increasing the amount of substrate presented to the intracellular metabolic enzymes and efflux transporters. Therefore, they act as important determinants of the metabolism and biliary excretion of drugs that are de-pendent on facilitated hepatic uptake. Changes in the hepatic uptake process, such as transport inhibition, alter the drug metabolism and biliary excretion pattern and potentially decrease the overall hepatic drug clearance [30, 69, 70, 77].

Figure 5. Transport proteins involved in hepatic drug disposition. SLC transport proteins in the basolateral membrane facilitate uptake from the blood to the intracel-lular compartment. Inside the cell compounds can interact with metabolic enzymes or efflux transporters. Compounds can get effluxed back into the blood or get ex-creted to the bile by ABC transport proteins such as Pgp and BSEP or the two SLC proteins MATE1 and OST α/β.

Basolateral Efflux Basolateral efflux proteins transport endogenous and xenobiotic compounds and their respective metabolites from the intracellular to the basolateral compartment. ABC efflux transporters with basolateral membrane expres-sion include MRP3, MRP4 and MRP6 [24]. In addition, MRP1 and 5 have been identified to have low expression levels in healthy liver [25, 78]. Dur-ing e.g. cholestasis and primary biliary cirrhosis the expression levels of

OCT1OATP1B

1

OATP1B3

OATP2B1

OAT2OAT7

NTCP

MRP3MRP4

MRP6OSTMRP2

BCRP

Pgp BSEP

MATE1

Bile

Blood

22

MRP1, 4, and 5 increases, indicating that they may participate in the hepatic protective response mechanisms during some pathological episodes [79-83]. All the basolateral efflux transporters include anionic compounds and phase II conjugates among their substrates, thereby facilitating renal excretion [52]. Furthermore, MRP3 is co-expressed with several of the metabolizing cyto-chrome P450 (CYP) enzymes in the pericentral part (zone 3) of the liver lobule, further indicating that the basolateral ABCs may redirect metabolism excretion from bile to urine [84, 85]. Inhibition of the basolateral efflux transporters therefore potentially increases the risk of hepatic accumulation of metabolites.

Canalicular efflux ABC transport proteins in the canalicular membrane of hepatocytes include Pgp, BCRP, MRP2 and BSEP that mediate the biliary excretion of drugs, bile salts, and metabolites against a sometimes steep concentration gradient [8, 24, 25, 42, 52]. In addition, the SLC-transporter multidrug and toxin ex-trusion exchanger 1 has been identified to facilitate biliary excretion of or-ganic cations [86]. BSEP has a narrow substrate spectrum, primarily restrict-ed to monovalent, negatively charged bile acids and is therefore not believed to influence the biliary excretion of drugs [87]. However, drug inhibition of BSEP transport may be associated with increased risk of DILI [88, 89]. In contrast, Pgp, BCRP and MRP2 have broad substrate specificities and play important roles in several protective membrane barriers [26, 90-92]. In the liver they decrease the intracellular load of many endogenous and xenobiotic compounds or their metabolites. The broad substrate specificity of these transporters increases the risk for DDI at the transporter level [93-95].

Drugs, endogenous metabolites, bile salts, and cytokines, as well as several disease states or developmental stages, affect the expression levels of these transport proteins. Disorders with decreased function or protein expression, such as progressive intrahepatic cholestasis type II or Dubin-Johnson syn-drome, exemplify the importance of the canalicular-efflux transporters in human physiology [96, 97].

Cellular Models for Hepatobiliary Transport Investigations Several methods have been established to elucidate drugs interactions with transport proteins in cellular barriers throughout the body, including the hepatobiliary tract [98-104]. However, each method has strengths and limita-tions and, to date, no individual method can adequately predict the contribu-tion of individual hepatic transport proteins to the overall hepatic drug dispo-sition in humans in vivo.

23

Membrane Vesicles The simplest model comprises inverted membrane vesicles (Fig. 6A) pro-duced from cells overexpressing the transport protein of interest or from the canalicular membrane domain from isolated hepatocytes [99, 105]. This model is easy to use and allows automated high-throughput screening of transport inhibitors. Inverted membrane vesicles also allow the investigated compounds to interact directly with the transport protein without the first membrane passage that is required in cellular systems. This is an advantage in studies aimed to describe compound-protein interactions at a molecular level. However, this method has limited use in screening for transport sub-strates among compounds of high permeability. Due to the passive diffusion out of the vesicles such compounds fail to accumulate in the vesicles despite active transport of the investigated protein.

Figure 6. Models that can be used to study transport protein interactions. (A) A schematic figure of an inverted membrane vesicle from a membrane including an ABC transport protein. Because of the inverted configuration solutes are free to interact with the transport protein directly without first diffusing into the intracellu-lar compartment. (B) Some immortalized cell lines form high integrity cell mono-layers with tight junctions restricting the paracellular transport. Such cell lines, e.g., MDCK-II and Caco-2, can be used for bidirectional transport experiments. (C) Pri-mary hepatocytes express most or all proteins important in hepatic drug disposition. If cultured in 3D configuration with extracellular matrix they form canalicular net-works simulating the biliary compartment in the in vivo liver. The arrows in figure 6C indicate three such networks.

Immortalized Cell Lines ABC-overexpressing cell lines are frequently used to investigate specific efflux protein interactions [106-108]. Some cell lines, such as MDCK-II, benefit from their ability to establish high-integrity cell monolayers that con-sequently allow experimental setups that determine the vectorial transport of the test compounds (Fig. 6B). Because of the general absence of uptake transport proteins in these models, sufficient (logP > 0.5) passive permeabil-ity is a prerequisite for ABC-protein interaction studies in these models [103].

Immortalized hepatoma cell lines provide a reliable cell source with high yield. The HepG2 cell line is the most commonly used and best character-

A B C

Apical side

Basolateral side

24

ized of the hepatoma cell lines [109, 110]. Compared to primary hepato-cytes, HepG2 has a more stable phenotype that could benefit routine testing. In conformity with other available hepatoma cell lines, HepG2 generally expresses low levels of metabolic enzymes and transport proteins and often needs transfection or enzyme induction modifications to facilitate metabo-lism investigations [25, 111, 112]. The HepaRG hepatoma cell line consti-tutes the most promising substitute for primary hepatocytes [113]. When differentiated by treatment with dimethyl sulfoxide (DMSO), HepaRG cells express a variety of metabolic enzymes and functional receptor pathways important in drug metabolism [114-117]. In comparison to HepG2, the Hep-aRG cell line has a CYP expression profile and induction pattern that more resembles primary hepatocytes. However, both cell lines fail to report toxici-ty responses observed in primary hepatocytes [111].

Primary Hepatocytes Primary hepatocytes can be regarded as the most relevant experimental sys-tems for the evaluation of hepatic drug properties available today. These generally express all the transport proteins and metabolizing enzymes found in the human liver and are one of the in vitro models best mimicking in vivo liver [111, 118]. Primary hepatocytes enable investigations of most hepatic drug interactions such as transport protein interaction, phase I and II metabo-lism, DDI potential, and receptor response mechanisms [111, 119, 120]. They can be cultured in a sandwich configuration containing extracellular matrixes which allows a three dimensional (3D) culture with more in vivo-like cell-cell and cell-extracellular matrix interactions (Fig. 6C) [118, 121]. This configuration has been observed to maintain morphologically and func-tionally stable hepatocyte cultures for up to eight weeks [64]. However, not all protein expression remains stable during sandwich culture and one of the hurdles for researchers using isolated primary hepatocytes is to overcome the reduced activity of the metabolic enzymes and some basolateral uptake transporters during the first days of culture.

Stem Cell Derived Hepatic Cells Stem cells are undifferentiated cells with the ability to develop into many different cell types [122]. They also possess unlimited replication capacity [123]. The multipotent or pluripotent properties of adult and embryonic stem cells, respectively, provide a highly interesting model system for basic re-search, or as a potential hepatic cell source for drug discovery or toxicologi-cal research [123, 124]. Following intricate differentiation protocols, the differentiated hepatocyte-like cells exhibit several hepatocyte specific func-tions, including activity of metabolic enzymes [125-127]. Nevertheless, im-portant barriers including the often fetal phenotype of differentiated hepato-cytes still remain before differentiated stem cells become a reliable substitute for primary hepatocytes [98].

25

Drug-induced Liver Injury (DILI)

During the last decades, DILI has been the most frequent cause of safety-related drug marketing withdrawals in the US. Hepatotoxicity occurring in the late stages of drug discovery has restricted the use, or even prevented the approval, of many drugs [128].

DILI is commonly classified into intrinsic and idiosyncratic hepatotoxicity [129-131]. Intrinsic hepatotoxicity is regarded as dose-dependent and pre-dictable above a threshold dose, whereas idiosyncratic hepatotoxicity occurs without obvious dose-dependency and in an unpredictable fashion. Intrinsic hepatotoxic compounds are generally discovered and rejected in preclinical trials, while idiosyncratic hepatotoxicity may not be identified until post-marketing monitoring when the drug becomes available to a larger popula-tion [129]. Among drugs withdrawn from the market due to severe DILI, hepatic failure appears at frequencies less than 1 of 10 000 treated patients [128]. When compared to the few thousand individuals typically included in clinical trials for new drug applications, the risk of underestimating the oc-currence of severe DILI becomes apparent.

The initial of toxicity of a drug, or often its metabolites, can be caused through several mechanisms e.g., direct cell stress, targeted mitochondrial function, or triggered specific immune reactions [130, 132, 133]. Despite the different initial mechanisms, cell stress and triggered specific immune reac-tions commonly lead to mitochondrial damage and altered membrane poten-tial. This happens either through a direct pathway activated by severe cellu-lar stress or via an indirect pathway, initiated by mild cellular stress or acti-vated specific immune reactions [131]. Independently of the origin of the first insult, the mitochondria seem to play a critical role in the initiation of DILI. The extent of mitochondrial impairment possibly determines whether hepatocytes die by apoptosis or necrosis [134]. In most patients, this onset of hepatic injury is followed by adaptation that most often results in resolution of the injury, despite continued drug treatment. However, in some patients the stress and injury reach a critical threshold and death-signaling pathways become dominant [135]. The injury then exceeds the adaptive repair and regeneration capacity of the liver and contributes to a progressive liver inju-ry. Therefore, DILI can be seen as an active process involving recruitment of death-signaling pathways that mediate cell death rather than a passive pro-cess due to overwhelming biochemical injury [135].

26

Computational Modeling of Transport Protein Interactions Since the late 1980s computational (in silico) approaches to assess interac-tions between drug or drug-like molecules and transport proteins have be-come more widely used in early drug discovery [136]. Computational mod-els describing protein interactions generally follow one of two approaches. Protein based models describe the 3D protein structure and can be used for ligand docking [137, 138]. However, the lack of high-resolution crystal structures of human transport proteins restrict these models to be based on crystal structures of ABC proteins of other species- or protein family- origin [45, 139]. Ligand based models describe the molecular structure of the lig-and interacting with the protein. On the basis of the approach taken to de-scribe the ligand, the ligand-based models can be further divided into sub-groups including quantitative structure-activity models (QSAR) and pharma-cophore models [138, 140, 141]. Pharmacophore models use the spatial loca-tion of key molecular structure elements while QSAR uses numeric values of calculated molecular descriptors to describe important molecular features of protein-interacting compounds.

Developed computational models can be used to screen large, diverse data-bases of molecules to identify compounds likely to be substrates or inhibitors of transport proteins. Predictions can be performed from molecular struc-tures even before compound synthesis, and thereby aid in the selection of structures less likely to have undesirable transport-protein interactions. The dataset selection has a crucial impact on the possible fields of application of the final models. To allow for the development of predictive in silico models the dataset has to be well distributed through the chemical space that the final model aims to describe. Global models aim to describe an entire chemi-cal space, such as that of orally administered drugs. In contrast, models aimed to describe compound series or parts of a chemical space are generally referred to as local models. In this thesis, large, diverse datasets were used to develop global models of the chemical space of orally administered drugs. The methods included in the work are described in the following sections.

Principal Component Analysis (PCA) Given the large amount of variables that can be used to describe the variance within a dataset, exact knowledge of the information captured by each varia-ble is not always available. Therefore, the correlation, or overlap in infor-mation, between the included variables is unknown. If the variable correla-tion is unknown or if included variables are known to correlate with each other to some extent, principal component analysis (PCA) is a good tool to describe the variance within the dataset [142].

27

PCA can be used to reduce the number of variables needed to describe the variance in a dataset, and to identify outliers and trends. The method com-bines correlated variables into fewer artificial variables, called principal components. For example chemical descriptors such as molecular weight, number of carbons, and molecular surface area, are different properties relat-ed to molecular size. These three variables could therefore be combined into one principal component describing molecular size. Each of the variables influences the principal component to different extents, depending on the way that the molecular descriptors correlate to molecular size. Chemical diversity is a good example of when PCA is a suitable tool to describe the overall diversity in a dataset using few principal components, often describ-ing size, shape, lipophilicity or polarity.

In short, PCA is a way to extract information from many variables and com-bine them into in fewer principal components. PCA reduces the number of dimensions needed to describe the systemic variance and therefore can be used to display the variations in a dataset graphically.

Partial Least Squares Projection to Latent Structures (PLS) Partial least squares projection to latent structures (PLS) is a regression ex-tension of PCA based on the same theory that combines correlated variables into principal components. However, PLS models relate the components to a set of one or more response variables, such as experimental in vitro data. PLS models seek solutions that maximize the response variance explained by the included variables or components. Hence, these models describe the response variables using correlating, or co-varying, components [142, 143].

PLS can be performed in several different ways. If the response variable has a binary property, for instance a yes or no answer, discriminant analyses (DA) can be performed. In PLS-DA the model seeks components that distin-guish one group from the other. In orthogonal PLS (OPLS), the systematic variation in the dataset is divided into two groups: the predictive and the orthogonal variables. The predictive variables co-vary with the response variable while the orthogonal ones do not. As the name implies, the orthogo-nal components are perpendicular to the predictive component and therefore have an average zero variance in the predictive component [142, 143].

In this thesis, molecular properties important in ABC inhibition was identi-fied using OPLS-DA analyses of molecular descriptors and in vitro data as variables and response variables, respectively.

Ligand-Based Docking to Protein Models Ligand-based docking requires a crystal structure or homology model of the 3D structure of the investigated protein. In 2009, the first high-resolution

28

crystal structure of an eukaryotic ABC protein, mouse-Pgp, was determined [45]. On the basis of this crystal structure, homology models for human ABC proteins (previously restricted to bacterial ABC crystal structures), could be developed [95] and subsequently employed in ligand-docking mod-els [144, 145]. Protein-ligand docking is generally performed in three steps. First, the 3D protein structure is scanned for regions of tight atom packing that indicate possible ligand-binding sites (LBS). Next, energy-minimized 3D ligand configurations are determined. In the third and final step, the lig-ands are docked to the protein-binding site and scored according to favorable interactions with the protein-binding site (positive scores) and unfavorable protein-overlapping ones (negative scores).

29

Aims of This Thesis

The general objective of this thesis was to examine drug interactions with ABC transport proteins and investigate how such interactions influence the biliary excretion, DDI and DILI.

The specific aims were:

• To establish methods for the investigation of ABC-dependent transport,

using membrane vesicles (Paper I-IV) and sandwich cultured human hepatocytes (SCHH) (Paper III and IV).

• To identify MRP2 and BSEP transport inhibitors within diverse datasets distributed throughout the chemical space of orally-administered, regis-tered drugs (Papers I and IV).

• To investigate the Pgp-, MRP2-, and BCRP-inhibition in cellular and

vesicular in vitro model systems, and to identify inhibitors with specific-ity towards individual or multiple ABC proteins (Papers II and III).

• To develop predictive computational models describing the inhibition of

Pgp (Paper II), MRP2 (Paper I and II), BCRP (Paper II) and BSEP (Pa-per IV).

• To predict the maximal intrinsic biliary clearance and SCHH hepatobili-

ary transport, using quantified protein levels and in vitro transport kinet-ics (Paper III).

• To determine the role that BSEP inhibition plays in DILI, and to investi-

gate the extent to which BSEP transport inhibition can predict such ad-verse events (Paper IV).

30

Methods

Dataset Selection The studies in the thesis were focused on drug interactions with ABC transport proteins. Therefore, the datasets were chosen to be structurally diverse and well-distributed throughout the chemical space of orally-administered, registered drugs.

In papers I and IV, the aim was to describe the affinity pattern for MRP2 and BSEP inhibition, respectively. In addition to molecular diversity and drug-likeness, compounds in these studies were selected on the basis of previous knowledge of their affinities for ABC proteins. The aim of paper II was to compare the inhibition affinity pattern for Pgp, BCRP, and MRP2 in a di-verse dataset. Compounds were included on the basis of high-quality litera-ture data on Pgp inhibition in combination with in-house BCRP and MRP2 inhibition data. In paper III, the dataset was selected on the basis of the pre-vious results with the aim to further explore the general and more specific affinity patterns of Pgp, BCRP, and MRP2 inhibitors.

Experimental Methods ABC Transport in Membrane Vesicles

Inverted membrane vesicles from Sf9 cells overexpressing human MRP2 or BSEP were used in papers I, II and IV. Membrane vesicles prepared from HEK293 cells transiently transfected to overexpress Pgp, BCRP or MRP2 were used in paper III. Incubation times and substrate concentrations were selected to be within linear regions of transport kinetics. MRP2 transport was studied using estradiol 17-β-glucuronide (E17G) at 50 µM or 10 µM (Paper III). Both concentrations were in the linear transport range and. The higher concentration was used to minimize the occurrence of transport stim-ulation effects observed for E17G [146, 147]. BCRP transport in vesicles was investigated using 1 µM estrone 3-sulfate (E3S) or 10 µM E17G, both used within linear transport kinetics. For the same reason Pgp transport was investigated using 0.04 µM N-methyl quinidine (NMQ) and 10 µM E17G. In general, substrate concentrations three to ten times below Km were used.

31

The transport experiments were conducted using the rapid filtration tech-nique previously described [148]. In brief, membrane vesicles were quickly thawed from -85°C to 37°C and a volume corresponding to 10 – 20 µg total protein per well was pre-incubated with the substrate. Transport was initiated by the addition of ATP and creatine kinase or exchanged to AMP in control wells for the determination of passive permeability. Transport was stopped by the addition of ice-cold stop solution and after filtration and filter wash steps the vesicles were analyzed for substrate content through scintillation counting for radiolabeled substrate or HPLC-MS/MS for unlabeled sub-strates.

For scintillation analysis, the filters were completely dried before scintilla-tion cocktail was added after which the filters were analyzed for radioactivi-ty, using a TopCount NXT scintillation counter (PerkinElmer, Waltham, MA). LC-MS/MS quantifications were performed with Waters Xevo TQ MS with electrospray ionization coupled to Acquity HPLC or UPLC systems (Waters, Milford, MA). For HPLC-MS/MS analyses, the test compounds were extracted from the washed filter plates with acetonitrile and water (20:80) containing an internal standard. After centrifugation, the filtrates were injected and intravesicular compound levels were quantified. ATP-dependent transport was calculated by subtraction of passive permeability determined from AMP incubations and ATP transport kinetics was deter-mined using non-linear regressions (Prism, GraphPad, San Diego, CA).

Inhibition of ABC Transport in Membrane Vesicles The inhibition experiments were performed as normal transport assays (de-scribed above) but co-incubated with inhibitors at concentrations suitable for the purpose of the study. MRP2 (paper I and II) and BSEP (paper IV) inhibi-tion were screened at test compound concentrations of 80 µM and 50 µM, respectively. In IC50 determinations, an inhibitor concentration interval from 1 to 500 µM was generally used. However, due to the inhibition potency of ko143, a lower concentration interval (0.001 to 10 µM) was used for this compound. IC50 values were calculated from nonlinear regressions of the dose-response curves using Prism version 5 (GraphPad, San Diego, CA). Subsequent Ki values were calculated according to equation 1.

= (1)

where the IC50 denotes the inhibitor concentration that inhibits 50% of the substrate transport, [S] is the substrate concentration in the medium, and Km represents the concentration that corresponds to half the maximal substrate transport rate. In papers II and III, an inhibitor was regarded as specific for

32

one of the transport proteins if the IC50 or Ki was at least 10-fold lower com-pared to the other two transporters.

Inhibition of Efflux in BCRP-Overexpressing Cells In paper II, inhibition of BCRP-mediated efflux was investigated in Saos-2 cells, stably transfected to express human BCRP. To determine compounds inhibitory effects on BCRP transport, the cells were incubated with 1 µM mitoxantrone for 60 min, with or without the addition of 50 µM of the test compounds. Transport was terminated by the addition of ice-cold buffer, the cells were washed, trypsinated, and then analyzed by flow cytometry (Beckman Coulter FC500, CA) to determine intracellular mitoxantrone con-centrations. The BCRP inhibition obtained by 0.5 µM of the potent BCRP inhibitor ko143 was set to 100% and used to normalize the results of the test compounds.

Isolation of Primary Human Hepatocytes Liver tissues were obtained from human donors undergoing partial liver resections at the Department of Surgery, Uppsala University Hospital, Swe-den. Resected tissues from patients not suffering from primary hepatic can-cers or hepatitis infections were included. All donors gave informed consent, in accordance with the ethical approval from the Uppsala Regional Ethical Review Board (Ethical Approval no. 2009/028). Hepatocytes were isolated using the two-step liver digestion technique of LeCluyse and Alexandre (2010) [149]. Briefly, tissue perfusion and cell isolations were initiated with-in two hours after tissue resection. During this time the tissues were kept in ice-cold Dulbecco’s modified Eagle’s medium (DMEM). The tissues were flushed free from blood with ice-cold DMEM and cannulated to support two perfusion lines. The first perfusate, containing ethylenediaminetetraacetic acid (EDTA), bovine serum albumin (BSA), and ascorbic acid, was main-tained at 37°C and circulated for approximately 10 min. After terminating the first perfusion, the second perfusate including collagenase, protease, BSA, and ascorbic acid, was immediately started. When clear signs of tissue digestion were visible, the perfusion was stopped and the cells harvested. Three centrifugation steps, with the second one including 27% Percoll (GE-Healthcare, Sweden), were performed to enrich viable hepatocytes suitable for culture.

Culture of Human Hepatocytes Isolated primary hepatocytes with a viability > 85% were seeded in collagen-1 coated 24-well plates (BD Biosciences, Bedford, MA) at a density of 3.75 x 105 cells per well and maintained at 37°C and 5% CO2. To allow the

33

cells to attach to the plate, 500 µl DMEM supplemented with 5% (v/v) fetal bovine serum, 4 µg/ml insulin, 1 µM dexamethasone, 4 mM Lglutamine, 100 U/ml penicillin and 100 µg/ml streptomycin was added to each well. After two to three hours this medium was replaced with 500 µl culture me-dium (HMM (Lonza, Basel, Switzerland) supplemented with insulin 10 µg/ml, transferrin 5.5 µg/ml and selenium 5 ng/ml, 0.1 µM dexame-thasone, 100 U/ml penicillin and 100 µg/ml streptomycin). After overnight incubation, the cells were overlaid with 500 µl ice-cold 0.25 mg/ml Matrigel (BD Biosciences, Bedford, MA) in culture medium. To allow the formation of bile canaliculi, the cells were cultured for an additional four to six days during which time, the culture medium was exchanged every 24 hours.

Biliary Excretion in SCHH Biliary excretion in SCHH was determined using a method modified from that described by Liu et al 1999 [150]. In short, SCHH were rinsed twice with either standard Hank's Balanced Salt Solution (HBSS), to maintain the integrity of tight junctions and bile canaliculi, or with Ca2+ and Mg2+-free HBSS, to disrupt the tight junctions. Cultures were pre-incubated with standard or Ca2+ and Mg2+-free HBSS, including the investigated inhibitor (10 or 50 µM) in inhibition experiments. Substrate (1 µM E17G and TA in Paper III and IV, respectively) was added and the uptake and excretion into the bile canaliculi was studied. Transport was terminated after 10 min by removing the incubation medium and rinsing the cells with ice-cold standard or Ca2+ and Mg2+-free HBSS. Cells were lysed with 1 M NaOH and kept at 4°C overnight before LC-MS/MS analysis. Total protein content was deter-mined using the BCA Protein Assay Reagent Kit (Pierce Biotechnology, Rockford, IL), according to manufacturer’s instructions. Substrate uptake was converted to molar amounts by normalizing to measurements in sub-strate solutions of known concentrations and normalized to total protein con-tent.

Total accumulation (TOTALAcc; intracellular + bile from HBSS incubations) and intracellular accumulation (ICAcc; from Ca2+ and Mg2+-free HBSS incu-bations) were used to calculate biliary accumulation (BILEAcc) according to equation 2. = − (2)

Biliary excretion index (BEI; Eq.3) is a ratio describing the biliary accumu-lation of a compound in relation to its total accumulation [150]. BEI de-creases if a test compound inhibits the canalicular efflux. The BEI parameter includes the bile accumulation in both the numerator and denominator. Thus

34

a decrease in BEI need not be directly proportional to a decrease in canalicu-lar efflux caused by an added transporter inhibitor. BEI was calculated ac-cording to equation 3. = ∗ 100 (3)

Bile intracellular correlation (BIC) describes the biliary accumulation in relation to the intracellular accumulation. Thus, BIC compensates for varia-bility in intracellular substrate accumulation, which is of interest since it governs efflux kinetics. Because BIC is calculated using the accumulation in the adjacent intracellular and bile compartments, any decreases in BIC is directly proportional to decreases in the canalicular efflux. BIC was calculat-ed according to equation 4. = (4)

The in vitro biliary clearance (CLBile) was calculated according to equation 5. CLBile is dependent on the net substrate flux across the cell, i.e., on the net basolateral uptake, including basolateral efflux, and canalicular efflux. Therefore, CLBile decreases if a test compound inhibits the basolateral net uptake or the canalicular efflux. CLBile was calculated according to equation 5. = (5)

where AUC represents the dose calculated from substrate concentration in the medium and incubation time.

Identification of Compounds Causing DILI In paper IV, the influence of BSEP inhibition on DILI was investigated. The ADR classification system implemented in drug labels issued by the FDA was considered to be a suitable data source for indication of the drug’s po-tential to inflict DILI. The FDA drug labels represent a consensus of regula-tory and industry experts and they aim to include all known adverse effects of a drug, from in vitro results to post-marketing monitoring. The three ADR sections within the drug labels categorize ADRs by increasing severity, rang-ing from the least severe ‘Adverse Reactions’ (AR), through the intermediate ‘Warnings and Precautions’ (WP), to the most severe ‘Boxed Warnings’ (BW). Category inclusion criteria are regulated by US federal regulation 21CFR201.57 [151]. The AR section describes the overall adverse reaction

35

profile of a drug and includes those adverse events believed to have a causal relationship with the drug. As soon as reasonable evidence of a causal asso-ciation between an ADR and a drug is established, it should be included in the WP or BW section, depending on the severity of the ADR.

To assess the DILI potential of registered drugs, information on hepatic ADRs in FDA-approved drug labels was obtained from DailyMed (http://dailymed.nlm.nih.gov/) [152]. Briefly, FDA drug labels were re-viewed for hepatic adverse reactions by searching for keywords related to liver injury (see supporting information, Paper IV). A compound was re-garded to have the potential to inflict DILI if the keywords were identified in the BW, WP or AR sections within a drug label. If a keyword was men-tioned in several sections, the compound was classified in the most severe category (BW>WP>AR). Drugs where no hepatic ADRs could be identified were categorized as ‘Not mentioned’ (NM). The ADR sections in FDA drug labels describe ADRs in a hierarchical way on the basis of their severity. To assure that the ADR sections described DILI severity in the same hierar-chical way, without influence from other ADRs reported for the same drug, the severity of DILI reported in the different sections was evaluated using the method described by Chen et al. 2011 [153].

Pgp, BCRP and MRP2 Protein Expression Immunohistochemical data were determined as a part of the Human Protein Atlas project [66]. In brief, protein epitope signature tags (PrEST) were iden-tified and the corresponding recombinant protein was expressed to produce antibodies in New Zealand rabbits. Monospecific antibodies were purified through depletion of tag-specific antibodies after which affinity columns were loaded with the protein-specific PrEST. Quality assurance was con-firmed by sequence verification of the PrEST clone, size determination of the recombinant protein, and cross-reactivity with other identified PrESTs. Validations of the antibodies were investigated by matching the western blot staining pattern to those from reference antibodies [154, 155].

The protein expressions of Pgp, BCRP and MRP2 in liver tissues and the overexpressing membrane vesicles were quantified using targeted prote-omics. Membrane fractions from four snap-frozen human liver tissue sam-ples were retrieved using the ProteoExtract Native Membrane Protein Ex-traction Ki (Calbiochem, Temecula, CA) according to the manufacturer’s instructions. Peptides were generated through protein extraction and tryptic digestion of the membrane fractions and the membrane vesicles preparations expressing Pgp, BCRP and MRP2. The peptides, and subsequent proteins, were quantified by LC-MS/MS with the addition of stable isotope labeled internal standard peptides, as previously described [156-158].

36

Canalicular Efflux Predicted by the Maximal Transport Activity The maximal transport activity (MTA) is the estimated maximal in vivo transport rate predicted from the in vitro maximal transport rate (Vmax) and the in vivo and in vitro quantified protein levels [140]. MTA is directly pro-portional to the Vmax, in vivo assuming that all quantified protein contributes to the active substrate transport and that the compared cell types contain similar amounts of total membrane proteins. The maximal Pgp, BCRP and MRP2 dependent canalicular efflux of E17G was calculated according to equation 6. = . . . . ∗ V , (6)

where the Prot.Exp.in vivo and Prot.Exp.in vitro denotes the quantified protein levels of the transport proteins in human liver tissues (n = 4) and inverted membrane vesicles, respectively. Vmax, in vitro is the maximal substrate transport rate determined in the membrane vesicles.

The MTA was then used to predict the intrinsic in vivo biliary clearance (CLBile, MTA), according to equation 7. CL , = ∗ HomProt (7)

where [S] represents the in vivo hepatocellular substrate concentration, that was estimated to be << than Pgp, BCRP and MRP2 Km and was therefore disregarded. Km denotes the Michaelis-Menten constant determined in the membrane vesicles and HomProt the total protein amount in human liver homogenate, 88 ± 14 mg/g tissue.

Finally, the in vivo biliary clearance (CLBile, MTA) was scaled to SCHH biliary clearance (CLBile, SCHH) with the following assumptions: 133 x 106 hepato-cytes/g human liver tissue and 375 000 hepatocytes per well. The obtained values were subsequently used to predict the biliary accumulation (BileAcc,

MTA) of E17G in SCHH, according to equation 8. Bile , = CL , ∗ Inc. Time ∗ IC (8)

where Inc.Time and ICAcc denotes the SCHH incubation time (10 min) and the experimentally determined intracellular accumulation (12 µM), respec-tively.

37

Computational Methods

Generation of Molecular Descriptors The two-dimensional (2D) molecular structures of the compounds in the datasets were obtained using Sci Finder Scholar (CAS, OH) (Papers I and II) or PubChem [159] (Paper IV), and converted to 3D structures using Corina (Molecular Networks, Germany). The molecular descriptors used to develop the computational models were calculated from the 3D structures using Dragon X (Talete, Italy), ADMET Predictor (SimulationsPlus, CA), and MAREA [160].

Principal Component Analysis - PCA To evaluate and visualize the dataset selection, a PCA model describing the chemical space of orally-administered drugs was developed, using Simca-P (Umetrics, Sweden). Molecular descriptors calculated for 652 orally-administered drugs were used in the model development. Variables with subzero prediction values during the cross-validation were excluded. The resulting PCA model was then used as a background template on which the datasets were projected to evaluate structure diversity and drug-likeness.

Partial Least Squares Projection to Latent Structures (PLS) Throughout the computational studies, discriminant analysis using PLS re-gressions (PLS-DA) or extensions thereof (OPLS-DA) were used to describe and predict the inhibition of ABC transport proteins. All models were devel-oped using Simca-P (Umetrics, Sweden).

Dataset Preparation After generating the molecular descriptors and obtaining the experimental protein interaction results, the datasets were prepared for the model devel-opment in two steps. First, the datasets were randomly divided into a training set (for model development), and a test set (for the model validation). Then the differently sized groups of inhibitors and non-inhibitors were balanced to produce equally sized groups in the data set and hence, reduce the possibility of one of the groups weighting the model to be developed.

Model Development and Validation As a first step in model development, molecular descriptors were mean cen-tered and scaled to unity-of-variance. Skew descriptors and those with zero variance were excluded. The models were optimized by a variable selection procedure in which groups of molecular descriptors that did not contain in-

38

formation relevant to the response (i.e., noise) were removed in a stepwise manner. Descriptors were not included in the model if removing them result-ed in a statistically improved model, based on the classification accuracy for the training set. The statistical validity of the models was tested using a ran-dom permutation test (n = 100). All presented models collapsed to subzero cross-validated coefficients of determination (Q2) when the response varia-bles were permutated, demonstrating that they were indeed describing the response variables. More importantly, all models were challenged with test sets composed of one-third of the experimentally investigated compounds.

Virtual Docking Virtual docking of general ABC inhibitors to a homology model of the nu-cleotide-binding domain 1 (NBD1) was performed in Maestro version 8.5 (Schrödinger, CA). The human MRP1 (ABCC1) NBD1 (Protein Data Bank ID 2CBZ) was used as a docking template for the 3D molecular structures of the test compounds. The protein and ligands were prepared using the default settings of the Maestro protein and ligand preparation procedures, respec-tively. The ligands were docked using default docking parameters in Glide XP version 5.027 (Schrödinger, CA) and the alignments of ATP and docked ligand were visualized using Accelrys DS Visualizer version 2.0.1 (Accel-rys, CA), and QuoteMol version 0.4.1 [161].

Statistical Analysis Data are expressed as means with standard deviations or standard error of the mean, as indicated in the text. In the MRP2 and BSEP inhibition screens, compounds that significantly decreased substrate transport were identified by one-way ANOVA with Bonferroni’s post hoc test, as implemented in Prism 5 (GraphPad Software, Inc., La Jolla, CA). Differences in the propor-tions of DILI-inducing agents in each ADR class were determined using normality tests. The Marascuillo procedure was used to determine statistical differences of mild and severe DILI distribution in the different ADR sec-tions. In paper IV, analysis of SCHH accumulation data was conducted using two-way ANOVA with Bonferroni’s post hoc test.

39

Results and Discussion

Dataset Diversity The aim of this thesis was to describe the inhibition of ABC transport pro-teins involved in the biliary excretion by using large, diverse datasets evenly distributed through the chemical space of orally-administered drugs. The dataset selection process plays a crucial role for the predictivity of the final model and particular care must therefore be taken to assure good coverage of the chemical space that is to be modeled.

The final dataset used to investigate MRP2 inhibition (Paper I) included 191 compounds. The overlapping Pgp, BCRP and MRP2 inhibition (Paper II) was studied among 122 compounds, and 250 compounds were investigated for BSEP inhibition. The dataset used in Paper II (n = 122) was included in the MRP2 and BSEP datasets. The chemical diversity of the datasets was evaluated by superimposing them on the chemical space of orally adminis-tered, registered drugs. The datasets were proven suitable for the intended investigations as they covered the relevant chemical space.

Identification of MRP2 and BSEP Inhibitors In Papers I and IV, inverted membrane vesicles from Sf9 cells overexpress-ing human MRP2 and BSEP, respectively, were used to study the inhibition of the active transport of estradiol 17β-glucuronide (E17G) and taurocholate (TA), respectively.

Before investigating the transport inhibition, substrate transport in the vesi-cles was characterized with regard to affinity, time dependence and sensitivi-ty for potent MRP2 and BSEP inhibitors. For MRP2, ATP-dependent transport was linear for up to 30 min. The transport kinetics from non-linear regression showed an apparent Km of 94 ± 7 µM and a maximal transport rate of 680 ± 30 pmol/mg/min. The kinetics revealed an initial lower slope for the E17G transport rate, indicating that the substrate stimulates its own transport. The MRP2 dependent transport was inhibited by the established MRP inhibitor MK571, with an IC50 of 10 ± 1.5 µM (Fig. 7A-C). For BSEP, the results indicated linear transport for up to 15 min with Km and Vmax of

40

17.8 ± 5 µM and 286.2 ± 28.2 pmol/mg/min, respectively. The established BSEP inhibitor Cyclosporine A (CycA) had an IC50 of 4.6 ± 1.2 µM in this system (Fig. 7 D-F).

Figure 7. Kinetics determined for the MRP2 (Paper I) and BSEP (Paper IV) assays. Time dependent transport (A, D), transport kinetics (B, E) and concentration de-pendence of model inhibitors (C, MK571 for MRP2 and D, Cyc A for BSEP) was determined for MRP2 and BSEP model substrates E17G and TA, respectively.

Of the 191 compounds screened for inhibition of MRP2-mediated E17G transport, 42 reduced E17G transport by more than 50% compared to the controls (Fig. 8 A). A similar hit rate (21%) was observed for BSEP, where 53 compounds inhibited the BSEP-mediated TA transport by more than 50% (Fig. 8 B).

BSEP substrates are typically monovalent, negatively-charged bile acids. Of the few non-bile acid substrates identified to date, all carry a negative net charge at physiological pH [162, 163]. This is in agreement with the molecu-lar features of importance for BSEP inhibition as well, where positive charge is strongly associated with a lack of inhibition and only 6% of the inhibitors are positively charged (Fig. 9A). However, unlike substrate interactions with BSEP, which are clearly disfavored by the lack of a negative charge [87], the majority of the BSEP inhibitors (58%) were un-ionized. MRP2 also has a clear preference for negatively charged substrates and approximately 80% are anioic compounds. However, the inhibitor charge pattern differed signif-icantly from that of BSEP, and almost one-third of the MRP2 inhibitors car-ried a positive net charge at physiological pH (Fig. 9B). This highlights an important difference between MRP2 and BSEP inhibition affinity and may support the theory of an additional inhibitory binding site with a separate mechanism in MRP2.

A B C

D E F

0

100

200

0 5 10 15Time (min)

TA tr

ansp

ort

(pm

ol/m

g)

0 50 1000

100

200

300

TA Concentration (μM)

ATP

-dep

ende

nttra

nspo

rt(p

mol

/min

/mg)

Cyc A Concentration (μM)

BS

EP

trans

port

(% o

f con

trol)

0.01 1 1000

50

100

Time (min)

E17

G tr

ansp

ort

(nm

ol/m

g)

E17G Concentration (μM)A

TP-d

epen

dent

trans

port

(pm

ol/m

in/m

g)

MK571 Concentration (μM)

MR

P2

trans

port

(% o

f con

trol)

0

1

2

0 10 20 300

50

100

1 10 1000 100 200 300 4000

100

200

41

Figure 8. Inhibition of the ATP-dependent transport of MRP2 and BSEP. Inhibition of MRP2 mediated E17G transport was investigated for 191 compounds and 42 inhibitors were identified. In addition, 13 compounds were found to stimulate MRP2 transport. (B) BSEP inhibition was investigated for 250 compounds using TA as a model substrate. Fifty-three compounds was found to inhibit more than 50% of the transport.

Figure 9. The charge distribution among compounds observed to inhibit more than 50% of the MRP2 and BSEP mediated E17G and TA transport, respectively. In contrast to the shared preference for negatively charged substrates, BSEP (A) and MRP2 (B) show different inhibitor charge patterns.

-300 -200 -100 0 100

MRP2 Inhibition (%) (n = 191)

>50% inhibition42 inhibitors

>50% stimulation13 stimulators

-50 0 50 100BSEP Inhibition (%) (n = 250)

>50% inhibition53 inhibitors

37%

29%

34%

MRP2

34%

6%

58%

2%BSEP

anioncationun-ion zwitter

42

Overlapping Inhibition of the Canalicular ABC Proteins

Pgp, BCRP, and MRP2 are co-localized in several tissues or barriers im-portant in the protection of the body from foreign or toxic compounds. Over-lapping inhibition of the ABC transport proteins can therefore lead to serious adverse effects due to accumulation of toxic compounds or the introduction of compounds into tissues normally protected e.g., the brain. In Paper II, the inhibition overlap between Pgp, BCRP, and MRP2 was shown for the first time using a large diverse dataset (n = 122). In total, 66 compounds inhibited at least one of the transport proteins, and as many as 19 (29%) of these in-hibitors significantly influenced the substrate transport of all three proteins. Interestingly, among these overlapping ABC inhibitors, 73% of all MRP2 inhibitors identified in the dataset were found. The dataset explored in Paper II was expanded with BSEP inhibition data (Paper IV), for all but one com-pound (gefitinib). In the new dataset (n = 121), 45 of the 68 identified inhibi-tors influenced the transport of at least two of the ABC proteins (Fig. 10). It could be hypothesized that the totally overlapping ABC inhibitors (n = 11) interact with the proteins at, for instance, the ATP-binding sites. Further-more, the physicochemical properties of the general inhibitors show that they, in general, are more lipophilic and more often positively charged at physiological pH. These properties make them more likely to partition into the membrane and may therefore reflect other protein interaction mecha-nisms at the TMD.

Figure 10. Overlapping inhibition of MRP2, BCRP, Pgp and BSEP. A majority (n = 68) of the 121 compounds investigated for ABC transport inhibition was found to inhibit at least one of the four canalicular transport proteins. BCRP and Pgp inhi-bition was shared with MRP2 or BSEP for eight and nine compounds, respectively, but only one inhibitor was shared by only MRP2 and BSEP

MRP2 BSEPBCRP Pgp

3 311

11

7

8 9

1 1

10

11

1

10

53No ABC inhibition

43

With the exception of the totally overlapping inhibitors, few compounds shared BSEP and MRP2 inhibition (n = 3). At the same time both proteins showed a significant inhibitor overlap with Pgp or BCRP or both, and eleven and nine BSEP and MRP2 inhibitors, respectively, were shared with the other two ABC proteins only. This further supports the theory that BSEP and MRP2 inhibitor interaction clearly differ despite their mutual preference for negatively charged substrates.

Molecular Properties Important in ABC Inhibition One of the aims of the thesis work was to define molecular properties im-portant for the inhibition of ABC transport proteins. For this purpose, com-putational models that discriminated between inhibitors and non-inhibitors were developed. One of the benefits of the PLS/OPLS-DA models compared to other computational models, such as neural networks, is the transparency of the developed models. Such models include relatively few molecular de-scriptors and allow for the interpretation of how individual molecular de-scriptors influence the ABC inhibition. Despite the simplicity of the final models, they correctly classified the absolute majority of the inhibitors or non-inhibitors in the models describing MRP2-, general ABC-, and BSEP inhibition (Table 1).

Table 1.Predictions of MRP2-, general ABC-, and BSEP inhibition by the developed computational models.

Correctly Classified

Model Dataset Inhibitors Non-inhibitors

MRP2 Training set 86 % 86 %(Paper I) Test set 72 % 71 % General ABC Training set 79 % 83 %(Paper II) Test set 79 % 80 % BSEP Training set 84 % 91 %(Paper IV) Test set 82 % 94 %

Molecular properties predictive for general ABC inhibition, describing Pgp, BCRP, and MRP2 inhibition, were lipophilicity, aromaticity, and molecular size. These properties also dominate the model developed for MRP2 inhibi-tion. For BSEP, the single most important molecular property was lipophilic-ity, but other descriptors related to charge and hydrophobicity were also important. This indicates that even though the ABC transport proteins share many of the ligand interactions properties with each other, individual aspects must still be considered. Out of the four ABC transport proteins investigated herein, BSEP seems to have the most diverse inhibitor interactions. This is a

44

logical result given that the bile-acid transport protein BSEP lacks the wide substrate interaction profiles of the other three ABC proteins.

ABC Inhibition in Cellular and Vesicular Models Previous studies have indicated that the concentration range where various inhibitors display selectivity towards an ABC transport protein varies be-tween different model systems. In Paper III we investigated Pgp, BCRP, and MRP2 inhibition in HEK293 membrane vesicles and compared the results to those from investigations using other cellular or vesicular systems.

The results indicated that the inhibitor rank order was the same in cellular and vesicular systems (Fig. 11A-C). However, the cellular systems showed a general trend of higher IC50-values compared to those observed in the vesic-ular systems (Fig. 11D-F). This general decrease in inhibition potency was not observed between different vesicular systems or between different sub-strates in the HEK293 vesicles, where the IC50 instead varied individually for different inhibitors. These results indicate that the general increase in IC50-values for the cellular systems is dependent upon the system (cell versus vesicle) and not the substrate. A possible explanation for this system de-pendency is the initial membrane passage required before ABC protein in-teraction in cell based models decreases the ABC-interacting inhibitor con-centration and therefore the inhibition.

Figure 11. Inhibition rank order in (A) MDCKII-Pgp and HEK293-Pgp vesicles; (B) Saos2-BCRP and HEK293-BCRP vesicles; and (C) between Sf9 and HEK293 MRP2 vesicles. Inhibitioon potencies observed between the same model systems for Pgp (D), BCRP (E) and MRP2 (F).

45

Interestingly, several of the investigated compounds showed IC50-values below 10 µM in the HEK-vesicles. This is a concentration that may be well within the range of in vivo intracellular concentration, e.g., for compounds with SLC facilitated uptake. For such compounds, the risk of in vivo DDI and DILI is considerable.

Prediction of Hepatobiliary Transport of E17G In Paper III, a method [164] suggested for the prediction of individual transport protein contribution to the in vivo or in vitro membrane transport was evaluated experimentally for the first time. In this method maximal transport activity (MTA; Eq. 6) is used to predict the activity of individual transport proteins in target organs or in vitro model systems including all the investigated transport proteins, such as SCHH.

Figure 12. Protein expression and predicted MTA for MRP2, Pgp and BCRP. Im-munohistochemical staining of MRP2 (A), Pgp (B) and BCRP (C) showed clear canalicular staining of MRP2 and Pgp, but only weak staining for BCRP was ob-served. (D) BCRP, Pgp and MRP2 protein expression in human liver tissue (n = 4) quantified by targeted proteomics. (E) Biliary excretion of E17G in SCHH and pre-dicted by MTA on the basis of protein expression and in vitro transport kinetics.

The MTA predictions are based on the levels of protein expression (Fig. 12A-D) and the in vitro transport kinetics determined in transporter overex-pressing models, such as the membrane vesicles. The results presented in Paper III show, in line with literature data, that the biliary excretion of E17G was mainly dependent on MRP2 transport, and the MRP2 dependent biliary

Pro

tein

exp

ress

ion

(fmol

/μg

mem

. pro

t)

BCRP P-gp MRP20.0

0.5

1.0

Vesicles 143.9 39.0 60.2

A

D

B C BCRPPgpMRP2

E

0.1

1

10

100

SCHH MTA

BIL

Eac

c (p

mol

/mg)

BCRP

Pgp

MRP2

46

excretion of E17G was predicted to be 49 and 182 times higher than that mediated by Pgp and BCRP, respectively. More importantly, using this method we were able to predict the biliary excretion of E17G in SCHH. The E17G biliary excretion was predicted to 90 pmol/mg, in good agreement with the experimental results of 71.5 pmol/mg (Fig. 12E).

BSEP Inhibition and DILI Several drugs reported to cause DILI have been identified as BSEP inhibi-tors [89, 165, 166]. In Paper IV, the association of BSEP inhibition and DILI was investigated for the first time in a dataset with a DILI incidence compa-rable to that found among registered drugs.

Figure 13. DILI potential of drugs associated with weak or strong BSEP inhibition. The FDA drug-label sections, ‘boxed warnings’ (BW), ‘warning and precautions’ (WP), and ‘adverse reactions’ (AR), were used to classify DILI potential of the drugs (n = 182) in the dataset also screened for BSEP inhibition in membrane vesi-cles. Bars show the fraction of all compounds in the same BSEP inhibition class (strong/weak/non-inhibitor) that are associated with DILI of a certain severity. BSEP inhibition was shown to have a significant influence on the DILI potential. Normali-ty tests, using BSEP non-inhibitors as control, was used to test the significance of the frequency differences; illustrated by * if p < 0.05 and ** if p < 0.01.

The results gave further support to the hypothesis that BSEP inhibition is linked to DILI and a majority of the BSEP inhibitors increased the risk of more severe (BW and WP) DILI (Fig. 13). A significant proportion (18%) of the strong inhibitors were associated with potentially life-threatening, BW classified DILI, in comparison to 5% of BSEP non-inhibitors or weak inhibi-tors (Fig. 13). Interestingly, only the strong inhibitors increased the risk of BW-classified DILI, while the weaker inhibitors only increased the risk of

0

10

20

30

40

50

60

BW WP AR NM

% o

f com

poun

ds

Strong inhibitors Weak inhibitors Non-inhibitors

***

**

*

**

47

intermediate severity (WP) DILI (Fig. 13). Notably, nine compounds associ-ated with severe DILI had not previously been identified as BSEP inhibitors (amiodarone, atazanavir, celecoxib, clarithromycin, dipyridamole, erythro-mycin, ezetimibe, lovastatin, tipranavir), suggesting that this mechanism may contribute to their observed clinical toxicity.

In total, 38 out of the 61 drugs (62%) identified as inhibitors of BSEP-mediated transport increased the risk of the more severe BW- or WP-class DILI, while only 37 out of the 121 (31%) BSEP non-inhibitors had BW or WP-class DILI (Fig. 14). The analysis further showed that of all compounds with reasonable evidence of a causal relationship to DILI, 51% were associ-ated with BSEP inhibition.

Many cellular factors unassessed in BSEP-expressing membrane vesicles influence compounds ability to cause DILI. Therefore SCHH, a physiologi-cally more relevant model, was used to assess the correlation of BSEP inhi-bition with DILI. A subset of twelve model compounds, with different BSEP effects in the vesicle assay (inhibitors or non-inhibitors) and DILI potential (BW/WP or AR/NM), was selected for the SCHH studies. Three compounds were selected to represent each group.

Figure 14. . Correlation between DILI severity and BSEP inhibition. (A) Classifica-tion of the registered drugs in the dataset (n = 182) with respect to BSEP inhibition in inverted membrane vesicles and DILI severity. (B) Frequency of severe (BW/WP) DILI among BSEP inhibitors and non-inhibitors. (C) Frequency of BSEP inhibitors among drugs associated with severe DILI, compared with drugs with mild or no DILI. BSEP inhibition was shown to double the risk of DILI and every second (51%) compound associated with severe DILI inhibited BSEP mediated TA transport.

Our results on TA transport and accumulation in SCHH supported and ex-tended the observations obtained in the membrane vesicles. As expected from the vesicle results, the three BSEP inhibitors associated with severe

A Correlation DILIseverity vs BSEP inhibition

B DILI Association;BSEP inhibitors vs non-inhibitors

C BSEP inhibition;Drugs with severe

vs mild/no DILI

0

25

50

75

100

+ –

Sev

ere

DIL

I (%

of c

ompo

unds

)

0

25

50

75

100

+ –

BS

EP

inhi

bitio

n(%

of c

ompo

unds

)

BSEPinhibition

DILI(BW/WP)

No 37

BW/WP AR/NM

Yes

DILI

BSEP

Inhi

bitio

n

84

38 23drugs drugs

drugs drugs

62

3151

21

48

DILI (BW or WP) caused a major reduction in the canalicular efflux of TA in SCHH (Fig. 15A). Notably, this group of compounds had a significantly greater impact on the biliary TA excretion than any of the other groups in-vestigated. In contrast, the BSEP inhibitors with less severe or no reported DILI (AR or NM) only marginally affected the canalicular TA efflux in SCHH, despite showing similar degrees of inhibition in the vesicle assay (Fig.15B). These results support the utility of SCHH in distinguishing BSEP inhibitors associated with severe DILI from those with no or mild DILI. Importantly, compounds inhibiting BSEP but which have not been reported to cause severe DILI (Fig. 15B) still resulted in modest alterations of the canalicular efflux. Such compounds may contribute to an increased ADR incidence if co-administered with other drugs also interacting with transport-ers involved in canalicular efflux.

Figure 15. . The impact on TA transport in SCHH of 12 drugs with different BSEP inhibition and DILI potential. (A-D) Drug impact on TA disposition is shown for each of the four combinations of BSEP inhibition (yes or no) in inverted membrane vesicles and DILI classifications obtained from the FDA drug labels (severe or mild/no), as described in Figure 6D: (A) BSEP inhibitors associated with severe (BW/WP) DILI, (B) BSEP inhibitors associated with mild or no DILI (AR/NM), (C) BSEP non-inhibitors associated with BW/WP DILI, and (D) BSEP non-inhibitors associated with AR/ NM DILI. Each panel shows, from left to right, the effect of the compounds on BSEP transport in inverted membrane vesicles (dark blue for inhibi-tors and light blue for non-inhibitors), and on the canalicular efflux of TA transport in SCHH using three complementary parameters: BEI (white), BIC (gray), and CLbile (black), the latter of which describes the effect of the basolateral net TA uptake in addition to the canalicular efflux. Data are presented as the mean ± stand-ard error

AR/NM

= BSEP transport vesicles = BEI = BIC = CLbile

0

50

100

CyA Ritonavir Rosiglitazone

% o

f con

trol

0

50

100

Mifepristone Isradipine Budesonide

% o

f con

trol

0

50

100

Valproic acid Flutamide Zidovudine

% o

f con

trol

0

50

100

Omeprazole Cimetidine Haloperidol

% o

f con

trol

DILIBW/WP

Yes

NoB

SEP

inhi

bitio

n

A B

C D

/

49

Analyzing Bile Excretion in SCHH Different parameters for the analysis of transport in SCHH describe different aspects of the cellular flux in this model system; for instance, the total net transport from the medium to the canalicular compartment is described by CLBile. In this thesis, BIC (Eq. 4) is proposed as a parameter that facilitates analyses of the canalicular transport, or alterations thereof, in SCHH. BIC relates the biliary efflux process to the intracellular amount of substrate. Therefore, it will describe the canalicular efflux process both in stable and altered intracellular amounts of substrate. This is exemplified in Figure 16 A and B, where the intracellular substrate amounts are stable and increased, respectively. Furthermore, a parameter directly proportional to BIC, CLBile, int, was recently shown to significantly improve predictions of human in vivo biliary clearance [167].

Figure 16. BEI, BIC, CLbile and CLbile,int correlation with TA accumulations in SCHH in the presence (A) and absence (B) of basolateral efflux. If the intracellular concentration remains unchanged (A) despite canalicular efflux BIC, CLbile and CLbile,int correlates with the canalicular efflux inhibition. If the canalicular efflux inhibition result in increased intracellular substrate concentration BEI and CLbile correlates with the canalicular efflux inhibition. However, because of the increased intracellular concentration it could be argued that the canalicular efflux inhibition exceeds the decrease observed in the absolute amount in the bile, if the transport follows linear kinetics. Furthermore, BIC coincide with CLBile,int, recently shown to result in more accurate predictions of in vivo biliary clearance, indicating that BIC is a valid parameter to use in the interpretations of SCHH data [167].

0%

50%

100%

050100150Bile accumulation (pmol/mg)

ICacc →Totalacc↓

ICacc↑Totalacc→

B

A

Normal

Normal

Decreased bile efflux(with basolateral efflux)

0%

50%

100%

050100150Bile accumulation (pmol/mg)

BEI BIC CLbile CLbile int

Decreased bile efflux(no basolateral efflux)

Par

amet

er v

alue

(% o

f con

trol)

Par

amet

er v

alue

(% o

f con

trol)

BEI

BICCLbileCLbile int

BEI

BIC

CLbile

CLbile int

50

Conclusions

This thesis investigated the inhibition of the ABC transport proteins im-portant in the biliary excretion of drugs. Particular emphasis was placed on establishing stable in vitro techniques for investigations of drug-ABC protein interactions, and defining the molecular properties important to these inter-actions. Previously unknown inhibitors of MRP2 and BSEP were identified and the inhibition overlap between Pgp, BCRP and MRP2 was described for the first time using a large, diverse dataset. A majority of the inhibitors were found to affect more than one of the ABC proteins. Finally, BSEP inhibition was shown to have a pronounced influence on the occurrence of DILI, high-lighting the importance of BSEP inhibition determination in early drug dis-covery. The work in this thesis demonstrates the usefulness of simple in vitro and in silico methods to provide the basis for understanding the cooperative mechanisms influencing cellular drug disposition. The specific conclusions were:

• In vitro models for studies of ABC dependent transport was established. Membrane vesicle models can be used to examine individual drug-protein interactions while SCHH allow investigation of the hepatic drug disposition.

• Approximately every fifth drug in datasets representative of the chemical space of orally-administered drugs was found to inhibit MRP2 and BSEP transport, respectively.

• Compounds with a general inhibition of Pgp, BCRP, and MRP2 transport were, in general, highly lipophilic weak bases.

• ABC inhibition could be modeled with easily interpretable molecular descriptors. These models have the potential to work as predictive filters in early drug discovery to guide compound selection away from unwant-ed ABC interactions.

• The individual ABC protein contribution to the biliary excretion of E17G in SCHH was well predicted from transport kinetics determined in membrane vesicles and quantified protein expression levels.

• BSEP inhibition doubles the risk of severe DILI, and half of the drugs reported to inflict severe DILI were identified as BSEP inhibitors.

51

Future Perspectives

The research described in this thesis has expanded the knowledge of the molecular characteristics important in the inhibition of the major hepatobili-ary ABC efflux transporters BSEP, MRP2, Pgp, and BCRP.

In the human liver, the transport mechanism studied here acts in concert with SLC and other ABC transport proteins as well as intracellular metabolic phase I and II enzymes. A drug can be a substrate or inhibitor of several of these proteins and can thereby influence the protein interactions with other co-administered drugs or endogenous compounds, e.g., bile acids. The re-sulting change in hepatic clearance or metabolic profile may cause clinically relevant DDIs or altered disposition of endogenous compounds, potentially leading to ADRs. A major challenge for future investigations is therefore to clarify the interplay between the different transport and metabolic proteins, and to develop in vitro or in silico models capable of predicting this inter-play.

Furthermore, hepatotoxicity discovered in the late stages of drug develop-ment or even after marketing approval has restricted the use of many drugs during the last two decades. The large inter-individual differences in suscep-tibility to DILI and the low incidence of severe DILI make these events hard to predict. An expanded knowledge of how cellular mechanisms or intercel-lular events contribute to DILI could improve DILI predictions. To date, it has not yet been clarified how hepatic NPCs, such as the macrophage Kup-ffer cells or the endothelial cells, influence hepatotoxic events. Evidence suggests that these cells are key contributors to the initiation of hepatotoxici-ty. Compared to what is observed today in hepatocyte monocultures, hepato-cyte-NPC cocultures could have the potential to provide a more sensitive DILI model that may report a wider range of hepatic insults at clinically relevant drug concentrations. Therefore, to establish such cocultures is an important future challenge in hepatotoxicity research.

52

Svensk populärvetenskaplig sammanfattning

Huden är kroppens största barriär. Den består av cellager som förhindrar infektioner, skyddar oss mot kemiska ämnen i vår miljö och bibehåller vår vätske- och saltbalans. Även kroppens inre organ skyddas på ett liknande sätt av cellbarriärer. Ett läkemedels förmåga att passera sådana cellbarriärer är avgörande för att det ska kunna komma in i kroppen och utöva sin av-sedda effekt, men också för att det ska kunna ta sig ut ur kroppen utan att ansamlas i organ eller vävnader. Cellbarriärerna består av celler som är så tätt sammanlänkade att den enklaste vägen över barriären för många kemiska ämnen är genom cellen. Det är nu, när läkemedlet passerar cellens ytterhölje, eller membran, som de kommer i kontakt med transportproteiner. Dessa proteiner kan antingen hjälpa läkemedlet på vägen in i cellen eller förhindra dess passage över membranet.

I den här avhandlingen har arbetet fokuserat på hur läkemedel och andra kemiska ämnen påverkar fyra transportproteiners förmåga att utsöndra äm-nen till gallan. Tre av transportproteinerna (Pgp, BCRP och MRP2) transpor-terar väldigt många olika ämnen och läkemedel och är bl.a. inblandade i läkemedelsresistens vid cancerbehandlingar. Det fjärde proteinet (BSEP) finns bara i levern och är den viktigaste orsaken till att gallsalter utsöndras till gallan.

De experimentella metoderna som använts under avhandlingsarbetet inklu-derar försök i cellmembran som innehåller ett av de fyra transportproteinerna och i isolerade leverceller som innehåller de flesta eller alla proteiner som uttrycks i levern i kroppen. Levercellerna har odlats på ett sätt som gör att de bildar tredimensionella strukturer som liknar de som finns i levern, med en gallsida och en blodsida.

Resultaten visar att drygt vart femte läkemedel av de över 200 substanser som undersöktes för MRP2 och BSEP interaktion, minskar eller blockerar dessa transportproteiners utsöndring till gallan. Inhibition av dessa proteiner kan leda till ibland allvarliga biverkningar och leverskador. Detta är idag en av de vanligaste orsakerna till att läkemedel dras tillbaka från marknaden eller får inskränkt användningsområde. En fördjupad analys visade att häm-mad BSEP-transport kan fördubbla risken för läkemedelsrelaterade lever-skador, och bland läkemedel med ett redan känt samband med leverskador

53

var andelen som hämmar BSEP-transport kraftigt förhöjd. I andra försök bestämdes mängden transportprotein i levern. Genom att kombinera mäng-den protein med proteinernas transportförmåga var det möjligt att förutsäga hur mycket av ett ämne som utsöndrades till gallgången både i de experi-mentellt odlade levercellerna och i levern i kroppen.

Utöver de experimentella metoderna utvecklades även datamodeller med målet att kunna prediktera interaktioner med transportproteinerna från läke-medelssubstanserna struktur. De här modellerna kan användas i tidig läke-medelsframställning, redan innan ämnet syntetiserats kemiskt, för att styra utvecklingen av nya läkemedel bort från oönskade interaktioner med trans-portproteinerna. Enkla beskrivningsmodeller av molekyler visade sig vara tillräckligt för att utveckla modeller som korrekt förutsäger interaktioner mellan läkemedlet och proteinerna. Molekylstrukturer som ökar risken för interaktion med de fyra transportproteinerna kunde också identifieras.

Arbetet i den här avhandlingen har resulterat i nya experimentella metoder som kan användas för att studera transporten till gallan från levercellen samt nya datormodeller som snabbt kan förutsäga risk för interaktioner med transportproteiner. De modeller som utvecklats är viktiga för studier av lä-kemedelsbiverkningar och kommer att främja utvecklingen av nya, säkra läkemedel.

54

Acknowledgements

The studies were carried out at the Department of Pharmacy, Faculty of Pharmacy, Uppsala University, Sweden.

The financial support from the Swedish Research Council, the Knut and Alice Wallenberg Foundation, the Swedish Foundation for Strategic Re-search, and the Swedish Fund for Research without Animal Experiments is gratefully acknowledged. My participation in research conferences was made possible by travel grants from IF:s stiftelse, and Anna Maria Lundins stipendiefond.

I wish to express my sincere gratitude to all of you who, in one way or an-other, contributed to this thesis:

Mina handledare: professor Per Artursson, för att du guidat mig in i forsk-ningens underbara, och ibland smärtsamma, värld. Din kreativitet och entu-siasm gör min och gruppens forskning så mycket bättre. Men det är inte för inte som dr. House är en av dina favoritkaraktärer Docent Christel Berg-ström, för att du är en unik förebild inom forskningen. Jag kommer att sakna din input på forskningen och andra små livshändelser. Docent Janet Hoogs-traate, för ditt engagemang och för att du visat mig forskningsvärlden utan-för akademin.

Professor Martin Malmsten och professor Göran Alderborn för tillhandahål-landet av fina lokaler och en stimulerande, skönt mintgrön, arbetsmiljö.

Eva Nises Ahlgren, Ulla Wästberg-Galik, Harriet Pettersson, Christin Mag-nusson, Eva Lide Johansson, Birgitta Rylén och Lotta Wahlberg för all hjälp med praktiska saker och Joahn Gråsjö för statistiska bidrag.

My co-authors, Prof. Ulf Norinder, Dr. Pär Matsson, Elin Svedberg, Dr. Yurong Lai, Dr. Xi Qiu, Dr. Agneta Norén, Dr. Johan Palm and last but not least Prof. Edward LeCluyse, for improving the papers included in this the-sis.

Alla tidigare och nuvarande gruppmedlemmar, för att ni bidrar med rolig forskning och gör arbetslivet så roligt. Särskilt tack till Dr. Maria Karlgren för att du ger så bra svar på det mesta, Dr Ulrika Tehler för att ditt skånska blod gör att du kan äta lika mycket som jag och sist men inte minst Jonas,

55

Anna, André, Linda och Sara för att ni är de bästa lab-kompisarna. Jag säger bara: Tack tack

My students, Sandra Rieken, Christin Zengel and Henning Karlsson for all the good work you contributed with.

Alla tidigare och nuvarande doktorander på institutionen, för alla mer och mindre seriösa diskussioner. För roliga fester och allmänt umgänge. Särskilt tack till Dr. Peter Falkman, Dr. Pär Matsson och Dr Gustav Ahlin för att jag alltid fått vara liten och hänsynslös. Moahahaha!.

Mina rumskamrater: Dr Noel Dew, för alla underbara skratt och ibland ganska sjuka skämt Om jag någonsin hittar strumpstickan… Linda Pers-son, för att du alltid peppar och för att du stått ut med att dela rum med en lätt antisocial arbetsmyra.

Alla vänner utanför institutionen för alla fantastiska stunder och roliga min-nen ni gett mig. Hehe, det är bara att notera att middagarna de senaste åren numer är lätt fraktionerade och innehåller något mer barnvänliga rätter… John och Mette, för att ni helt enkelt är underbara.

Familjerna på Jan-Eriks väg och i Vedyxa, för att ni tagit emot mig med sådan värme, trots att Henke gifte sig med en skåning Ene och Lennart, för att ni alltid står med öppna armar och för att ni är så underbara svär- och farföräldrar.

Olof, för att du alltid hade hittat någon ny cool musik att spela när jag kom hem. Jag saknar dig.

Emma och Palle, för att vi alltid kan komma och våldgästa er och för att Oliver, Malte och Charlie alltid kan förgylla min dag. Inget kan slå Olivers legokreationer, du är helt klart bäst!

Fredrik och Helen, för roliga stunder och härliga skratt. Det är alltid roligt när vi testar hur många som får plats att sova i vårt lilla hus. Viktor, för att du är en så klok liten man, som tillsammans med Moa kan hitta på de mest tokiga busen.

Pappa och Ylva, tack för att ni alltid finns där och peppar till alla tokigheter som jag tar mig för, och inte minst för att ni hjälpt mig inse finessen med husbil

Mamma, utan dig hade jag inte varit den jag är idag. Tack för allt du gjort och gör för mig!

Henrik, för att du alltid finns vid min sida och för att du alltid ser livet från den ljusa sidan. Utan dig, Moa och Kalle hade världen saknat innehåll. Jag älskar er nu och alltid.

56

References

1. Kola, I. and J. Landis, Can the pharmaceutical industry reduce

attrition rates? Nat Rev Drug Discov, 2004. 3(8): p. 711-5. 2. Pharmaceutical Research and Manufacturers of America (PhRMA),

2013 Biopharmaceutical Research Industry Profile. http://phrma.org/sites/default/files/pdf/PhRMA%20Profile%202013.pdf, 2013.

3. Food and Drug Administration. Summary of NDA Approvals & Receipts, 1938 to the present. 2013; Available from: http://www.fda.gov/AboutFDA/WhatWeDo/History/ProductRegulation/SummaryofNDAApprovalsReceipts1938tothepresent/default.htm.

4. Kennedy, T., Managing the drug discovery/development interface. Drug Discovery Today, 1997. 2(10): p. 436-444.

5. Prentis, R.A., Y. Lis, and S.R. Walker, Pharmaceutical innovation by the seven UK-owned pharmaceutical companies (1964-1985). Br J Clin Pharmacol, 1988. 25(3): p. 387-96.

6. Schinkel, A.H., et al., Disruption of the mouse mdr1a P-glycoprotein gene leads to a deficiency in the blood-brain barrier and to increased sensitivity to drugs. Cell, 1994. 77(4): p. 491-502.

7. Strautnieks, S.S., et al., A gene encoding a liver-specific ABC transporter is mutated in progressive familial intrahepatic cholestasis. Nat Genet, 1998. 20(3): p. 233-8.

8. Choi, Y.H. and A.M. Yu, ABC Transporters in Multidrug Resistance and Pharmacokinetics, and Strategies for Drug Development. Curr Pharm Des, 2013.

9. Hendrikx, J.J., et al., P-glycoprotein and cytochrome P450 3A act together in restricting the oral bioavailability of paclitaxel. Int J Cancer, 2013. 132(10): p. 2439-47.

10. Shitara, Y., et al., Clinical significance of organic anion transporting polypeptides (OATPs) in drug disposition: their roles in hepatic clearance and intestinal absorption. Biopharm Drug Dispos, 2013. 34(1): p. 45-78.

11. Shu, Y., et al., Effect of genetic variation in the organic cation transporter 1, OCT1, on metformin pharmacokinetics. Clin Pharmacol Ther, 2008. 83(2): p. 273-80.

57

12. Food and Drug Administration, Drug Interaction Studies — Study Design, Data Analysis, Implications for Dosing, and Labeling Recommendations, U.S.D.o.H.a.H. Services, F.a.D. Administration, and C.f.D.E.a.R. (CDER), Editors. 2012.

13. European Medicines Agency (EMA), Guideline on the Investigation of Drug Interactions, C.f.H.M.P. (CHMP), Editor 2012.

14. Jin, J., et al., Factors affecting therapeutic compliance: A review from the patient's perspective. Ther Clin Risk Manag, 2008. 4(1): p. 269-86.

15. Lemmer, H.J. and J.H. Hamman, Paracellular drug absorption enhancement through tight junction modulation. Expert Opin Drug Deliv, 2013. 10(1): p. 103-14.

16. Orth, M. and S. Bellosta, Cholesterol: its regulation and role in central nervous system disorders. Cholesterol. 2012: p. 292598.

17. Anderson, J.M. and C.M. Van Itallie, Physiology and function of the tight junction. Cold Spring Harb Perspect Biol, 2009. 1(2): p. a002584.

18. Bjorkhem, I. and S. Meaney, Brain cholesterol: long secret life behind a barrier. Arterioscler Thromb Vasc Biol, 2004. 24(5): p. 806-15.

19. Fischer, H., R. Gottschlich, and A. Seelig, Blood-brain barrier permeation: molecular parameters governing passive diffusion. J Membr Biol, 1998. 165(3): p. 201-11.

20. Gunzel, D. and A.S. Yu, Claudins and the modulation of tight junction permeability. Physiol Rev, 2013. 93(2): p. 525-69.

21. Matsson, P., et al., Exploring the role of different drug transport routes in permeability screening. Journal of Medicinal Chemistry, 2005. 48(2): p. 604-613.

22. Lai, Y., et al., Impact of drug transporter pharmacogenomics on pharmacokinetic and pharmacodynamic variability - considerations for drug development. Expert Opin Drug Metab Toxicol, 2012. 8(6): p. 723-43.

23. Hillgren, K.M., et al., Emerging transporters of clinical importance: an update from the international transporter consortium. Clin Pharmacol Ther, 2013. 94(1): p. 52-63.

24. International Transporter, C., et al., Membrane transporters in drug development. Nat Rev Drug Discov, 2010. 9(3): p. 215-36.

25. Hilgendorf, C., et al., Expression of thirty-six drug transporter genes in human intestine, liver, kidney, and organotypic cell lines. Drug Metab Dispos, 2007. 35(8): p. 1333-40.

26. Englund, G., et al., Regional levels of drug transporters along the human intestinal tract: Co-expression of ABC and SLC transporters and comparison with Caco-2 cells. European Journal of Pharmaceutical Sciences, 2006. 29(3-4): p. 269-277.

27. Gao, B., et al., Differential expression of bile salt and organic anion transporters in developing rat liver. J Hepatol, 2004. 41(2): p. 201-8.

58

28. Maliepaard, M., et al., Subcellular localization and distribution of the breast cancer resistance protein transporter in normal human tissues. Cancer Res, 2001. 61(8): p. 3458-64.

29. Ming, X., B.M. Knight, and D.R. Thakker, Vectorial Transport of Fexofenadine across Caco-2 Cells: Involvement of Apical Uptake and Basolateral Efflux Transporters. Molecular Pharmaceutics, 2011. 8(5): p. 1677-1686.

30. Neve, E.P., et al., An Integrated in Vitro Model for Simultaneous Assessment of Drug Uptake, Metabolism, and Efflux. Mol Pharm, 2013.

31. Hediger, M.A., et al., The ABCs of solute carriers: physiological, pathological and therapeutic implications of human membrane transport proteinsIntroduction. Pflugers Arch, 2004. 447(5): p. 465-8.

32. Rees, D.C., E. Johnson, and O. Lewinson, ABC transporters: the power to change. Nat Rev Mol Cell Biol, 2009. 10(3): p. 218-27.

33. Schnetkamp, P.P., The SLC24 gene family of Na(+)/Ca(2)(+)-K(+) exchangers: from sight and smell to memory consolidation and skin pigmentation. Mol Aspects Med, 2013. 34(2-3): p. 455-64.

34. Koepsell, H., The SLC22 family with transporters of organic cations, anions and zwitterions. Mol Aspects Med, 2013. 34(2-3): p. 413-35.

35. Kanai, Y., et al., The SLC1 high-affinity glutamate and neutral amino acid transporter family. Mol Aspects Med, 2013. 34(2-3): p. 108-20.

36. Palmieri, F., The mitochondrial transporter family SLC25: identification, properties and physiopathology. Mol Aspects Med, 2013. 34(2-3): p. 465-84.

37. Pedersen, B.P., et al., Crystal structure of a eukaryotic phosphate transporter. Nature, 2013. 496(7446): p. 533-6.

38. Schlessinger, A., et al., Comparison of human solute carriers. Protein Sci, 2010. 19(3): p. 412-28.

39. Wright, E.M., Glucose transport families SLC5 and SLC50. Mol Aspects Med, 2013. 34(2-3): p. 183-96.

40. Schioth, H.B., et al., Evolutionary origin of amino acid transporter families SLC32, SLC36 and SLC38 and physiological, pathological and therapeutic aspects. Mol Aspects Med, 2013. 34(2-3): p. 571-85.

41. Hagenbuch, B. and B. Stieger, The SLCO (former SLC21) superfamily of transporters. Mol Aspects Med, 2013. 34(2-3): p. 396-412.

42. Locher, K.P., Review. Structure and mechanism of ATP-binding cassette transporters. Philos Trans R Soc Lond B Biol Sci, 2009. 364(1514): p. 239-45.

43. Groen, A., et al., Complementary functions of the flippase ATP8B1 and the floppase ABCB4 in maintaining canalicular membrane integrity. Gastroenterology, 2011. 141(5): p. 1927-37 e1-4.

59

44. Ramaen, O., et al., Structure of the human multidrug resistance protein 1 nucleotide binding domain 1 bound to Mg2+/ATP reveals a non-productive catalytic site. J Mol Biol, 2006. 359(4): p. 940-9.

45. Aller, S.G., et al., Structure of P-glycoprotein reveals a molecular basis for poly-specific drug binding. Science, 2009. 323(5922): p. 1718-22.

46. Shintre, C.A., et al., Structures of ABCB10, a human ATP-binding cassette transporter in apo- and nucleotide-bound states. Proc Natl Acad Sci U S A, 2013. 110(24): p. 9710-5.

47. Liesa, M., W. Qiu, and O.S. Shirihai, Mitochondrial ABC transporters function: the role of ABCB10 (ABC-me) as a novel player in cellular handling of reactive oxygen species. Biochim Biophys Acta, 2012. 1823(10): p. 1945-57.

48. Schinkel, A.H. and J.W. Jonker, Mammalian drug efflux transporters of the ATP binding cassette (ABC) family: an overview. Adv Drug Deliv Rev, 2003. 55(1): p. 3-29.

49. McDevitt, C.A., et al., Purification and structural analyses of ABCG2. Adv Drug Deliv Rev, 2009. 61(1): p. 57-65.

50. Bandler, P.E., et al., Identification of regions required for apical membrane localization of human multidrug resistance protein 2. Mol Pharmacol, 2008. 74(1): p. 9-19.

51. Westlake, C.J., S.P. Cole, and R.G. Deeley, Role of the NH2-terminal membrane spanning domain of multidrug resistance protein 1/ABCC1 in protein processing and trafficking. Mol Biol Cell, 2005. 16(5): p. 2483-92.

52. Keppler, D., Multidrug resistance proteins (MRPs, ABCCs): importance for pathophysiology and drug therapy. Handb Exp Pharmacol, 2011(201): p. 299-323.

53. Stieger, B., The role of the sodium-taurocholate cotransporting polypeptide (NTCP) and of the bile salt export pump (BSEP) in physiology and pathophysiology of bile formation. Handb Exp Pharmacol, 2011(201): p. 205-59.

54. Kaminski, W.E., A. Piehler, and J.J. Wenzel, ABC A-subfamily transporters: structure, function and disease. Biochim Biophys Acta, 2006. 1762(5): p. 510-24.

55. Nofer, J.R., et al., Apolipoprotein A-I activates Cdc42 signaling through the ABCA1 transporter. J Lipid Res, 2006. 47(4): p. 794-803.

56. Juliano, R.L. and V. Ling, A surface glycoprotein modulating drug permeability in Chinese hamster ovary cell mutants. Biochim Biophys Acta, 1976. 455(1): p. 152-62.

57. Silverton, L., M. Dean, and K. Moitra, Variation and evolution of the ABC transporter genes ABCB1, ABCC1, ABCG2, ABCG5 and ABCG8: implication for pharmacogenetics and disease. Drug Metabol Drug Interact, 2011. 26(4): p. 169-79.

60

58. Berggren, S., et al., Gene and protein expression of P-glycoprotein, MRP1, MRP2, and CYP3A4 in the small and large human intestine. Mol Pharm, 2007. 4(2): p. 252-7.

59. Overington, J.P., B. Al-Lazikani, and A.L. Hopkins, How many drug targets are there? Nat Rev Drug Discov, 2006. 5(12): p. 993-6.

60. Zanger, U.M. and M. Schwab, Cytochrome P450 enzymes in drug metabolism: regulation of gene expression, enzyme activities, and impact of genetic variation. Pharmacol Ther, 2013. 138(1): p. 103-41.

61. Lutz, J.D. and N. Isoherranen, In vitro-to-in vivo predictions of drug-drug interactions involving multiple reversible inhibitors. Expert Opin Drug Metab Toxicol, 2012. 8(4): p. 449-66.

62. Guyton, A.C., Hall, J. E., Unit XIII - Metabolism and Temperature Regulation, in Textbook of Medical Physiology, J.E. Hall, Editor. 2010, Saunders Elsevier. p. 809-843.

63. Suchy, F.J., Hepatobiliary Function in Medical Physiology - A Cellular and Molecular Approach, W.F. Boron, Boulpaep, E. L., Editor. 2012, Saunders Elsevier. p. 980-1007.

64. LeCluyse, E.L., et al., Organotypic liver culture models: meeting current challenges in toxicity testing. Crit Rev Toxicol, 2012. 42(6): p. 501-48.

65. Marion, T.L., et al., Endogenous bile acid disposition in rat and human sandwich-cultured hepatocytes. Toxicol Appl Pharmacol, 2012. 261(1): p. 1-9.

66. Uhlen, M., et al., Towards a knowledge-based Human Protein Atlas. Nat Biotechnol, 2010. 28(12): p. 1248-50.

67. Lennernas, H., Clinical pharmacokinetics of atorvastatin. Clin Pharmacokinet, 2003. 42(13): p. 1141-60.

68. Elferink, R.P., Understanding and controlling hepatobiliary function. Best Pract Res Clin Gastroenterol, 2002. 16(6): p. 1025-34.

69. Hua, W.J., W.X. Hua, and H.J. Fang, The role of OATP1B1 and BCRP in pharmacokinetics and DDI of novel statins. Cardiovasc Ther, 2012. 30(5): p. e234-41.

70. Ieiri, I., S. Higuchi, and Y. Sugiyama, Genetic polymorphisms of uptake (OATP1B1, 1B3) and efflux (MRP2, BCRP) transporters: implications for inter-individual differences in the pharmacokinetics and pharmacodynamics of statins and other clinically relevant drugs. Expert Opin Drug Metab Toxicol, 2009. 5(7): p. 703-29.

71. He, L., K. Vasiliou, and D.W. Nebert, Analysis and update of the human solute carrier (SLC) gene superfamily. Hum Genomics, 2009. 3(2): p. 195-206.

72. Vasiliou, V., K. Vasiliou, and D.W. Nebert, Human ATP-binding cassette (ABC) transporter family. Hum Genomics, 2009. 3(3): p. 281-90.

73. Hediger, M.A. Bioparadigms - SLC Table. 2013 [cited 2013 July 19]; Available from: http://www.bioparadigms.org/slc/intro.htm.

61

74. Burckhardt, G., Drug transport by Organic Anion Transporters (OATs). Pharmacol Ther, 2012. 136(1): p. 106-30.

75. Konig, J., Uptake transporters of the human OATP family: molecular characteristics, substrates, their role in drug-drug interactions, and functional consequences of polymorphisms. Handb Exp Pharmacol, 2011(201): p. 1-28.

76. Nies, A.T., et al., Organic cation transporters (OCTs, MATEs), in vitro and in vivo evidence for the importance in drug therapy. Handb Exp Pharmacol, 2011(201): p. 105-67.

77. Ahlin, G., et al., Genotype-dependent effects of inhibitors of the organic cation transporter, OCT1: predictions of metformin interactions. Pharmacogenomics J, 2010.

78. Konig, J., et al., Conjugate export pumps of the multidrug resistance protein (MRP) family: localization, substrate specificity, and MRP2-mediated drug resistance. Biochim Biophys Acta, 1999. 1461(2): p. 377-94.

79. Ros, J.E., et al., High expression of MDR1, MRP1, and MRP3 in the hepatic progenitor cell compartment and hepatocytes in severe human liver disease. J Pathol, 2003. 200(5): p. 553-60.

80. Donner, M.G., et al., Enhanced expression of basolateral multidrug resistance protein isoforms Mrp3 and Mrp5 in rat liver by LPS. Biol Chem, 2004. 385(3-4): p. 331-9.

81. Hirohashi, T., H. Suzuki, and Y. Sugiyama, Characterization of the transport properties of cloned rat multidrug resistance-associated protein 3 (MRP3). J Biol Chem, 1999. 274(21): p. 15181-5.

82. Donner, M.G. and D. Keppler, Up-regulation of basolateral multidrug resistance protein 3 (Mrp3) in cholestatic rat liver. Hepatology, 2001. 34(2): p. 351-9.

83. Keitel, V., et al., Expression and localization of hepatobiliary transport proteins in progressive familial intrahepatic cholestasis. Hepatology, 2005. 41(5): p. 1160-72.

84. Martin, H., et al., Induction of cytochrome P450 2B6 and 3A4 expression by phenobarbital and cyclophosphamide in cultured human liver slices. Pharm Res, 2003. 20(4): p. 557-68.

85. Soroka, C.J., et al., Cellular localization and up-regulation of multidrug resistance-associated protein 3 in hepatocytes and cholangiocytes during obstructive cholestasis in rat liver. Hepatology, 2001. 33(4): p. 783-91.

86. Otsuka, M., et al., A human transporter protein that mediates the final excretion step for toxic organic cations. Proc Natl Acad Sci U S A, 2005. 102(50): p. 17923-8.

87. Gerloff, T., et al., The sister of P-glycoprotein represents the canalicular bile salt export pump of mammalian liver. J Biol Chem, 1998. 273(16): p. 10046-50.

88. Dawson, S., et al., In vitro inhibition of the bile salt export pump correlates with risk of cholestatic drug-induced liver injury in humans. Drug Metab Dispos, 2012. 40(1): p. 130-8.

62

89. Morgan, R.E., et al., Interference with bile salt export pump function is a susceptibility factor for human liver injury in drug development. Toxicol Sci, 2010. 118(2): p. 485-500.

90. Borst, P. and R.O. Elferink, Mammalian ABC transporters in health and disease. Annu Rev Biochem, 2002. 71: p. 537-92.

91. Keppler, D., J. Konig, and M. Buchler, The canalicular multidrug resistance protein, cMRP/MRP2, a novel conjugate export pump expressed in the apical membrane of hepatocytes. Adv Enzyme Regul, 1997. 37: p. 321-33.

92. Leslie, E.M., R.G. Deeley, and S.P. Cole, Multidrug resistance proteins: role of P-glycoprotein, MRP1, MRP2, and BCRP (ABCG2) in tissue defense. Toxicol Appl Pharmacol, 2005. 204(3): p. 216-37.

93. Meier, Y., et al., Interindividual variability of canalicular ATP-binding-cassette (ABC)-transporter expression in human liver. Hepatology, 2006. 44(1): p. 62-74.

94. Miners, J.O., et al., In vitro-in vivo correlation for drugs and other compounds eliminated by glucuronidation in humans: pitfalls and promises. Biochem Pharmacol, 2006. 71(11): p. 1531-9.

95. Zhang, L., et al., Drug interactions evaluation: an integrated part of risk assessment of therapeutics. Toxicol Appl Pharmacol, 2010. 243(2): p. 134-45.

96. Davit-Spraul, A., et al., ATP8B1 and ABCB11 analysis in 62 children with normal gamma-glutamyl transferase progressive familial intrahepatic cholestasis (PFIC): phenotypic differences between PFIC1 and PFIC2 and natural history. Hepatology, 2010. 51(5): p. 1645-55.

97. Machida, I., et al., Mutational analysis of the MRP2 gene and long-term follow-up of Dubin-Johnson syndrome in Japan. J Gastroenterol, 2005. 40(4): p. 366-70.

98. Dash, A., et al., Hemodynamic flow improves rat hepatocyte morphology, function, and metabolic activity in vitro. Am J Physiol Cell Physiol, 2013. 304(11): p. C1053-63.

99. Glavinas, H., et al., Utilization of membrane vesicle preparations to study drug-ABC transporter interactions. Expert Opinion on Drug Metabolism & Toxicology, 2008. 4(6): p. 721-732.

100. Karlgren, M., et al., In vitro and in silico strategies to identify OATP1B1 inhibitors and predict clinical drug-drug interactions. Pharmaceutical research, 2012. 29(2): p. 411-26.

101. Karlsson, J., et al., Transport of celiprolol across human intestinal epithelial (Caco-2) cells: mediation of secretion by multiple transporters including P-glycoprotein. Br J Pharmacol, 1993. 110(3): p. 1009-16.

102. Lagas, J.S., M.L. Vlaming, and A.H. Schinkel, Pharmacokinetic assessment of multiple ATP-binding cassette transporters: the power of combination knockout mice. Mol Interv, 2009. 9(3): p. 136-45.

63

103. Matsson, P., et al., A global drug inhibition pattern for the human ATP-binding cassette transporter breast cancer resistance protein (ABCG2). J Pharmacol Exp Ther, 2007. 323(1): p. 19-30.

104. Sivertsson, L., et al., Hepatic differentiation and maturation of human embryonic stem cells cultured in a perfused three-dimensional bioreactor. Stem Cells Dev, 2013. 22(4): p. 581-94.

105. Niinuma, K., et al., Primary active transport of organic anions on bile canalicular membrane in humans. Am J Physiol, 1999. 276(5 Pt 1): p. G1153-64.

106. Di, L., et al., Development of a new permeability assay using low-efflux MDCKII cells. J Pharm Sci, 2011. 100(11): p. 4974-85.

107. Li, J., Y. Wang, and I.J. Hidalgo, Kinetic analysis of human and canine P-glycoprotein-mediated drug transport in MDR1-MDCK cell model: Approaches to reduce false-negative substrate classification. J Pharm Sci, 2013.

108. Neuhoff, S., et al., pH-dependent bidirectional transport of weakly basic drugs across Caco-2 monolayers: implications for drug-drug interactions. Pharm Res, 2003. 20(8): p. 1141-8.

109. Westerink, W.M. and W.G. Schoonen, Phase II enzyme levels in HepG2 cells and cryopreserved primary human hepatocytes and their induction in HepG2 cells. Toxicol In Vitro, 2007. 21(8): p. 1592-602.

110. Westerink, W.M. and W.G. Schoonen, Cytochrome P450 enzyme levels in HepG2 cells and cryopreserved primary human hepatocytes and their induction in HepG2 cells. Toxicol In Vitro, 2007. 21(8): p. 1581-91.

111. Gerets, H.H., et al., Characterization of primary human hepatocytes, HepG2 cells, and HepaRG cells at the mRNA level and CYP activity in response to inducers and their predictivity for the detection of human hepatotoxins. Cell Biol Toxicol, 2012. 28(2): p. 69-87.

112. Kullak-Ublick, G.A., U. Beuers, and G. Paumgartner, Molecular and functional characterization of bile acid transport in human hepatoblastoma HepG2 cells. Hepatology, 1996. 23(5): p. 1053-60.

113. Guillouzo, A., et al., The human hepatoma HepaRG cells: a highly differentiated model for studies of liver metabolism and toxicity of xenobiotics. Chem Biol Interact, 2007. 168(1): p. 66-73.

114. Antherieu, S., et al., Stable expression, activity, and inducibility of cytochromes P450 in differentiated HepaRG cells. Drug Metab Dispos, 2010. 38(3): p. 516-25.

115. Vee, M.L., et al., Polarized expression of drug transporters in differentiated human hepatoma HepaRG cells. Toxicol In Vitro, 2013.

116. Kanebratt, K.P. and T.B. Andersson, HepaRG cells as an in vitro model for evaluation of cytochrome P450 induction in humans. Drug Metab Dispos, 2008. 36(1): p. 137-45.

64

117. Kanebratt, K.P. and T.B. Andersson, Evaluation of HepaRG cells as an in vitro model for human drug metabolism studies. Drug Metab Dispos, 2008. 36(7): p. 1444-52.

118. Swift, B., N.D. Pfeifer, and K.L. Brouwer, Sandwich-cultured hepatocytes: an in vitro model to evaluate hepatobiliary transporter-based drug interactions and hepatotoxicity. Drug Metab Rev, 2010. 42(3): p. 446-71.

119. Klieber, S., et al., The use of human hepatocytes to investigate drug metabolism and CYP enzyme induction. Methods Mol Biol, 2010. 640: p. 295-308.

120. Konopnicki, C.M., et al., Evaluation of UGT protein interactions in human hepatocytes: Effect of siRNA down regulation of UGT1A9 and UGT2B7 on propofol glucuronidation in human hepatocytes. Arch Biochem Biophys, 2013. 535(2): p. 143-9.

121. Lecluyse, E.L., K.L. Audus, and J.H. Hochman, Formation of Extensive Canalicular Networks by Rat Hepatocytes Cultured in Collagen-Sandwich Configuration. American Journal of Physiology, 1994. 266(6): p. C1764-C1774.

122. Thomson, J.A., et al., Embryonic stem cell lines derived from human blastocysts. Science, 1998. 282(5391): p. 1145-7.

123. Blanpain, C., et al., Self-renewal, multipotency, and the existence of two cell populations within an epithelial stem cell niche. Cell, 2004. 118(5): p. 635-48.

124. Jensen, J., J. Hyllner, and P. Bjorquist, Human embryonic stem cell technologies and drug discovery. J Cell Physiol, 2009. 219(3): p. 513-9.

125. Brolen, G., et al., Hepatocyte-like cells derived from human embryonic stem cells specifically via definitive endoderm and a progenitor stage. J Biotechnol, 2010. 145(3): p. 284-94.

126. Sullivan, G.J., et al., Generation of functional human hepatic endoderm from human induced pluripotent stem cells. Hepatology, 2010. 51(1): p. 329-35.

127. Takayama, K., et al., Efficient generation of functional hepatocytes from human embryonic stem cells and induced pluripotent stem cells by HNF4alpha transduction. Mol Ther, 2012. 20(1): p. 127-37.

128. Food and Drug Administration, Drug-Induced Liver Injury: Premarketing Clinical Evaluation, C.f.D.E.a.R.C. U.S.D.o.H.a.H.S. Food and Drug Administration, Center for Biologics Evaluation and Research (CBER), Editor 2009.

129. Bleibel, W., et al., Drug-induced liver injury: review article. Dig Dis Sci, 2007. 52(10): p. 2463-71.

130. Lee, W.M., Drug-induced hepatotoxicity. N Engl J Med, 2003. 349(5): p. 474-85.

131. Russmann, S., G.A. Kullak-Ublick, and I. Grattagliano, Current concepts of mechanisms in drug-induced hepatotoxicity. Curr Med Chem, 2009. 16(23): p. 3041-53.

65

132. Pauli-Magnus, C. and P.J. Meier, Hepatobiliary transporters and drug-induced cholestasis. Hepatology, 2006. 44(4): p. 778-87.

133. Rashid, M., R. Goldin, and M. Wright, Drugs and the liver. Hosp Med, 2004. 65(8): p. 456-61.

134. Mackay, I.R., Immunological perspectives on chronic hepatitis: virus infection, autoimmunity and xenobiotics. Hepatogastroenterology, 1999. 46(30): p. 3021-33.

135. Russo, M.W., et al., Liver transplantation for acute liver failure from drug induced liver injury in the United States. Liver Transpl, 2004. 10(8): p. 1018-23.

136. Ekins, S., et al., Future directions for drug transporter modelling. Xenobiotica, 2007. 37(10-11): p. 1152-70.

137. Kairys, V., M.K. Gilson, and M.X. Fernandes, Using protein homology models for structure-based studies: approaches to model refinement. ScientificWorldJournal, 2006. 6: p. 1542-54.

138. Penzotti, J.E., et al., A computational ensemble pharmacophore model for identifying substrates of P-glycoprotein. J Med Chem, 2002. 45(9): p. 1737-40.

139. Yamaguchi, H., et al., Homology modeling and structural analysis of human P-glycoprotein. Bioinformation, 2012. 8(22): p. 1066-74.

140. Karlgren, M., et al., Specific and general inhibitors of the three hepatic organic anion transporters OATP1B1 (SLCO1B1), OATP1B3 (SLCO1B3) and OATP2B1 (SLCO2B1). Drug Metabolism Reviews, 2010. 42: p. 305-306.

141. Matsson, P., et al., Identification and Prediction of Inhibitors of the Three Major Human ATP-Binding Cassette Transporters Pgp, MRP2 and BCRP. Drug Metabolism Reviews, 2008. 40: p. 161-162.

142. Eriksson, L., et al., Megavariate analysis of environmental QSAR data. Part I--a basic framework founded on principal component analysis (PCA), partial least squares (PLS), and statistical molecular design (SMD). Mol Divers, 2006. 10(2): p. 169-86.

143. Abdi, H. and L.J. Williams, Partial least squares methods: partial least squares correlation and partial least square regression. Methods Mol Biol, 2013. 930: p. 549-79.

144. Wise, J.G., Catalytic transitions in the human MDR1 P-glycoprotein drug binding sites. Biochemistry, 2012. 51(25): p. 5125-41.

145. Bikadi, Z., et al., Predicting P-glycoprotein-mediated drug transport based on support vector machine and three-dimensional crystal structure of P-glycoprotein. PLoS One, 2011. 6(10): p. e25815.

146. Heredi-Szabo, K., et al., Potentiation of MRP2/Mrp2-mediated estradiol-17beta-glucuronide transport by drugs--a concise review. Chem Biodivers, 2009. 6(11): p. 1970-4.

147. Zelcer, N., et al., Evidence for two interacting ligand binding sites in human multidrug resistance protein 2 (ATP binding cassette C2). J Biol Chem, 2003. 278(26): p. 23538-44.

148. Ishikawa, T., et al., ATP-dependent primary active transport of cysteinyl leukotrienes across liver canalicular membrane. Role of

66

the ATP-dependent transport system for glutathione S-conjugates. J Biol Chem, 1990. 265(31): p. 19279-86.

149. Lecluyse, E.L. and E. Alexandre, Isolation and culture of primary hepatocytes from resected human liver tissue. Methods Mol Biol, 2010. 640: p. 57-82.

150. Liu, X., et al., Biliary excretion in primary rat hepatocytes cultured in a collagen-sandwich configuration. Am J Physiol, 1999. 277(1 Pt 1): p. G12-21.

151. US.Regulations, F., 21 CFR 201.57 - Specific Requirements on Content and Format of Labeling for Human Prescription Drug and Biological Products Described in § 201.56(B)(1). 2012.

152. DailyMed, 2012. http://dailymed.nlm.nih.gov/. 153. Chen, M., et al., FDA-approved drug labeling for the study of drug-

induced liver injury. Drug Discov Today, 2011. 16(15-16): p. 697-703.

154. Berglund, L., et al., A genecentric Human Protein Atlas for expression profiles based on antibodies. Mol Cell Proteomics, 2008. 7(10): p. 2019-27.

155. Nilsson, P., et al., Towards a human proteome atlas: high-throughput generation of mono-specific antibodies for tissue profiling. Proteomics, 2005. 5(17): p. 4327-37.

156. Deo, A.K., et al., Interindividual variability in hepatic expression of the multidrug resistance-associated protein 2 (MRP2/ABCC2): quantification by liquid chromatography/tandem mass spectrometry. Drug Metab Dispos, 2012. 40(5): p. 852-5.

157. Li, N., et al., Quantitative expression profile of hepatobiliary transporters in sandwich cultured rat and human hepatocytes. Mol Pharm, 2009. 6(4): p. 1180-9.

158. Zhang, Y., et al., Liquid chromatography/tandem mass spectrometry based targeted proteomics quantification of P-glycoprotein in various biological samples. Rapid Commun Mass Spectrom, 2011. 25(12): p. 1715-24.

159. NCBI, The National Center for Biotechnology Information, 2012. 160. Palm, K., et al., Effect of molecular charge on intestinal epithelial

drug transport: pH-dependent transport of cationic drugs. Journal of Pharmacology and Experimental Therapeutics, 1999. 291(2): p. 435-443.

161. Tarini, M., P. Cignoni, and C. Montani, Ambient occlusion and edge cueing to enhance real time molecular visualization. IEEE Trans Vis Comput Graph, 2006. 12(5): p. 1237-44.

162. Hirano, M., et al., Bile salt export pump (BSEP/ABCB11) can transport a nonbile acid substrate, pravastatin. J Pharmacol Exp Ther, 2005. 314(2): p. 876-82.

163. Matsushima, S., et al., Involvement of multiple efflux transporters in hepatic disposition of fexofenadine. Mol Pharmacol, 2008. 73(5): p. 1474-83.

67

164. Karlgren, M., et al., Classification of inhibitors of hepatic organic anion transporting polypeptides (OATPs): influence of protein expression on drug-drug interactions. J Med Chem, 2012. 55(10): p. 4740-63.

165. Kis, E., et al., Effect of membrane cholesterol on BSEP/Bsep activity: species specificity studies for substrates and inhibitors. Drug Metab Dispos, 2009. 37(9): p. 1878-86.

166. Stieger, B., et al., Drug- and estrogen-induced cholestasis through inhibition of the hepatocellular bile salt export pump (Bsep) of rat liver. Gastroenterology, 2000. 118(2): p. 422-30.

167. Nakakariya, M., et al., In vivo biliary clearance should be predicted by intrinsic biliary clearance in sandwich-cultured hepatocytes. Drug Metab Dispos, 2012. 40(3): p. 602-9.

Acta Universitatis UpsaliensisDigital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Pharmacy 172

Editor: The Dean of the Faculty of Pharmacy

A doctoral dissertation from the Faculty of Pharmacy, UppsalaUniversity, is usually a summary of a number of papers. A fewcopies of the complete dissertation are kept at major Swedishresearch libraries, while the summary alone is distributedinternationally through the series Digital ComprehensiveSummaries of Uppsala Dissertations from the Faculty ofPharmacy.

Distribution: publications.uu.seurn:nbn:se:uu:diva-204200

ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2013