You are on page 1of 40

DRUG DISPOSITION

Clin Pharmacokinet 2003; 42 (1): 59-98 0312-5963/03/0001-0059/$30.00/0 Adis International Limited. All rights reserved.

Role of P-Glycoprotein in Pharmacokinetics


Clinical Implications
Jiunn H. Lin and Masayo Yamazaki
Department of Drug Metabolism, Merck Research Laboratories, West Point, Pennsylvania, USA

Contents
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1. Structure and Mechanism of Drug-Transporting P-Glycoprotein . . . . . . . . . . 1.1 Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.2 ATP- and Substrate-Binding Sites . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Substrate Recognition . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2. Polymorphisms of P-Glycoprotein . . . . . . . . . . . . . . . . . . . . . . . . . . . 3. In Vitro/In Vivo Extrapolation and Species Differences . . . . . . . . . . . . . . . 4. Role of P-Glycoprotein in Pharmacokinetics . . . . . . . . . . . . . . . . . . . . . 4.1 Drug Absorption . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.1.1 Distribution of Intestinal P-Glycoprotein . . . . . . . . . . . . . . . . . . 4.1.2 Interindividual Variability of Intestinal P-Glycoprotein . . . . . . . . . . 4.1.3 Evidence of Intestinal P-Glycoprotein Involvement in Drug Absorption 4.1.4 Saturable Efflux Transport by Intestinal P-Glycoprotein . . . . . . . . . 4.2 Drug Distribution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Blood-Brain Barrier . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Placenta . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Drug Metabolism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Drug Excretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.1 Biliary Excretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4.2 Renal Excretion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5. P-Glycoprotein-Mediated Drug-Drug Interactions . . . . . . . . . . . . . . . . . . 5.1 P-Glycoprotein Inhibition Does Not Follow Simple Kinetics . . . . . . . . . . 5.2 Drug Interactions Caused by P-Glycoprotein Inhibition . . . . . . . . . . . . 5.3 P-Glycoprotein Induction is a Complex Process . . . . . . . . . . . . . . . . 5.4 Drug Interactions Caused by P-Glycoprotein Induction . . . . . . . . . . . . 6. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 61 61 62 63 64 66 68 69 70 70 71 72 75 75 79 80 81 82 83 84 84 86 88 90 91

Abstract

P-glycoprotein, the most extensively studied ATP-binding cassette (ABC) transporter, functions as a biological barrier by extruding toxins and xenobiotics out of cells. In vitro and in vivo studies have demonstrated that P-glycoprotein plays a significant role in drug absorption and disposition. Because of its localisation, P-glycoprotein appears to have a greater impact on limiting cellular uptake of drugs from blood circulation into brain and from intestinal lumen into epithelial cells than on enhancing the excretion of drugs out of hepatocytes and renal tubules into the adjacent luminal space. However, the relative contribution of intestinal

60

Lin & Yamazaki

P-glycoprotein to overall drug absorption is unlikely to be quantitatively important unless a very small oral dose is given, or the dissolution and diffusion rates of the drug are very slow. This is because P-glycoprotein transport activity becomes saturated by high concentrations of drug in the intestinal lumen. Because of its importance in pharmacokinetics, P-glycoprotein transport screening has been incorporated into the drug discovery process, aided by the availability of transgenic mdr knockout mice and in vitro cell systems. When applying in vitro and in vivo screening models to study P-glycoprotein function, there are two fundamental questions: (i) can in vitro data be accurately extrapolated to the in vivo situation; and (ii) can animal data be directly scaled up to humans? Current information from our laboratory suggests that in vivo P-glycoprotein activity for a given drug can be extrapolated reasonably well from in vitro data. On the other hand, there are significant species differences in P-glycoprotein transport activity between humans and animals, and the species differences appear to be substrate-dependent. Inhibition and induction of P-glycoprotein have been reported as the causes of drug-drug interactions. The potential risk of P-glycoprotein-mediated drug interactions may be greatly underestimated if only plasma concentration is monitored. From animal studies, it is clear that P-glycoprotein inhibition always has a much greater impact on tissue distribution, particularly with regard to the brain, than on plasma concentrations. Therefore, the potential risk of P-glycoproteinmediated drug interactions should be assessed carefully. Because of overlapping substrate specificity between cytochrome P450 (CYP) 3A4 and P-glycoprotein, and because of similarities in P-glycoprotein and CYP3A4 inhibitors and inducers, many drug interactions involve both P-glycoprotein and CYP3A4. Unless the relative contribution of P-glycoprotein and CYP3A4 to drug interactions can be quantitatively estimated, care should be taken when exploring the underlying mechanism of such interactions.

P-glycoprotein was first identified by Juliano and Ling as a surface phosphoglycoprotein expressed in drug-resistant Chinese hamster ovary cells.[1] This discovery led to the finding that Pglycoprotein is an energy-dependent efflux transporter driven by ATP hydrolysis. In humans, two members of the P-glycoprotein gene family (MDR1 and MDR3) exist, while three members of this family (mdr1a, mdr1b and mdr2) are found in mice.[2,3] The P-glycoprotein encoded by the human MDR1 and mouse mdr1a/1b genes functions as a drug efflux transporter, whereas human MDR3 P-glycoprotein and mouse mdr2 P-glycoprotein are believed to be functional in phospholipid transport.[4,5] However, the involvement of human MDR3 P-glycoprotein in drug transport has been
Adis International Limited. All rights reserved.

recently reported. An increased directional transport of digoxin, paclitaxel and vinblastine across polarised monolayers of MDR3-transfected cells has recently been reported by Smith et al.[6] These results suggest that MDR3 P-glycoprotein is also able to transport a range of drugs in addition to phospholipids. In addition to the expression in tumour cells, human P-glycoprotein is also highly expressed in normal tissues. This transporter is localised on the canalicular surface of hepatocytes in liver, the apical surface of epithelial cells of proximal tubules in kidneys, columnar epithelial cells of intestine, epithelial cells of placenta, and the luminal surface of capillary endothelial cells in brain in humans.[7,8] The anatomical localisation of P-glycoClin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

61

protein expression suggests that the efflux transporter can functionally protect the body against toxic xenobiotics by excreting these compounds into bile, urine and the intestinal lumen, and by preventing their accumulation in brain. Because of its localisation, it is believed that P-glycoprotein may play a significant role in the processes of absorption, distribution, metabolism and excretion of drugs in humans and animals. Indeed, the role of P-glycoprotein in drug absorption and disposition has been demonstrated in vivo using mdr1a and mdr1a/1b knockout mice.[9,10] A lack of either one of the two murine P-glycoproteins (mdr1a or mdr1b) results in significant changes in drug absorption and disposition. The purpose of this review is to briefly summarise the current knowledge regarding the structure and mechanism of drug-transporting P-glycoprotein and its role in drug absorption, distribution, metabolism and excretion. In addition, the potential for P-glycoprotein-mediated drug-drug interactions and its clinical implications will be discussed. 1. Structure and Mechanism of Drug-Transporting P-Glycoprotein
1.1 Structure

Since the identification of P-glycoprotein, major efforts have been made to elucidate the structure of the protein to gain better insight into the mechanism of its action. The cloning of genes and structure-function analysis of the protein by genetic and biochemical studies have contributed to a better understanding of the transporter. P-glycoprotein is composed of two homologous and symmetrical halves (cassettes), each of which contains six transmembrane domains that are separated by an intracellular flexible linker polypeptide loop with an ATP-binding motif.[2] Interestingly, the two halves of human P-glycoprotein are not identical, and of the amino acids aligned, only 43% are identical,[11] suggesting that the molecules of the two halves might have either evolved indepen Adis International Limited. All rights reserved.

dently or have undergone major intron movement after a duplication event.[2] Site-directed mutagenesis and antibody mapping studies suggest that the two cassettes of human P-glycoprotein interact cooperatively to form a single functional unit.[12,13] The direct evidence that supports the hypothesis of a single functional unit is from the study that co-expressed each cassette of human P-glycoprotein. The cDNA coding for human P-glycoprotein was divided in half and subcloned into separate plasmids in order to express each half as a separate polypeptide and to characterise its contribution to function. No drugstimulative ATPase activity was observed when Sf9 cells were transfected separately with cDNA coding for each cassette.[12] The hypothesis of the single functional unit is further supported by the mutation studies carried out by Takada et al.[14] In this study, the key lysine and cysteine residues in the Walker A motifs of ATP-binding domains were substituted by methionine and alanine, respectively. The results of this study clearly demonstrate that if one ATP-binding domain is not functional, there is no ATP hydrolysis even when ATP binds to the other ATP-binding domain. In addition to the ATP-binding domains, the intracellular flexible linker loops also play a key role in ATPase and transport activity. Deletion of the central core of the intracellular flexible linker region of human P-glycoprotein resulted in a protein without functional ATPase and transport activity.[15] Collectively, these data strongly suggest that the two cassettes of P-glycoprotein interact as a single transporter and that the flexible linker region is important for the proper interaction of the two cassettes. Several mechanistic models have been proposed to describe the mechanism for drug transport activity. The initial mechanistic model hypothesises that, similar to ion channel proteins, hydrophobic membrane-spanning regions and hydrophilic elements of P-glycoprotein form an aqueous transmembrane pore through which drugs are transported from the cytosol to the extracellular media.[16] However, another model suggests that
Clin Pharmacokinet 2003; 42 (1)

62

Lin & Yamazaki

P-glycoprotein might extrude drugs directly from the cell membrane even before they enter the cytoplasm. This second model is supported by a study of fura-2 acetoxymethyl ester in NIH-3T3 mouse fibroblasts.[17] Fura-2 acetoxymethyl ester, a fluorescent indicator, is hydrophobic and actively extruded by P-glycoprotein, whereas the hydrophilic free acid form of the indicator, to which the ester is rapidly hydrolysed once in the cellular cytoplasm, is not exported by P-glycoprotein. The intracellular trapping of Fura-2 free acid was remarkably reduced in P-glycoprotein-expressing NIH-3T3 mouse fibroblasts as compared with that in the control fibroblasts (no P-glycoprotein expression) by a factor of 10. Addition of verapamil did not alter the intracellular concentration of Fura-2 free acid in control cells, whereas it increased the amount of trapped Fura-2 free acid in the P-glycoprotein-expressing cells up to the level found in the control cells. From these results, the investigators concluded that hydrophobic molecules of Fura-2 acetoxymethyl ester interacted with the P-glycoprotein in the cell membrane before entering the cytosol, and hence cytoplasmic esterases had no chance to see Fura-2 acetoxymethyl ester. However, their conclusion is valid only if the rate of P-glycoprotein transport is the ratelimiting step in trapping of Fura-2 free acid, i.e. the rate of ester hydrolysis is much faster than the rate of transport. More convincing evidence that supports the concept of the interaction of substrates with Pglycoprotein in the lipid membrane of cells came from the study by Shapiro and Ling.[18] These investigators measured the kinetics of Hoechst 33 342 in P-glycoprotein-enriched plasma membrane vesicles from Chinese hamster ovary cells. Hoechst 33 342 is fluorescent only when bound to the membrane, but not when in the aqueous medium. Therefore, the movement of Hoechst 33 342 in and out of the membrane can be directly monitored by the fluorescence intensity. Using the fluorometric assay, the results revealed that the initial rate of transport was directly proportional to the amount of dye in the lipid phase, but not to the
Adis International Limited. All rights reserved.

concentration in the aqueous phase, suggesting that P-glycoprotein extruded Hoechst 33342 from the lipid membrane before it entered the cytosol. With some modifications of the second model, a recent and more favoured model proposes that P-glycoprotein intercepts lipophilic drugs as they move through the lipid membrane and flips the drugs from the inner leaflet to the outer leaflet and into the extracellular medium.[19] This model is consistent with the notion that the lipophilicity of a drug is an important determinant in its interaction with P-glycoprotein. Recently, Rosenberg et al.[20] used high-resolution electron microscopy and image analysis to obtain the first three-dimensional architecture of a P-glycoprotein purified from Chinese hamster ovary CHrB30 cells that retained the ability to bind substrates and hydrolyse ATP. When viewed from above the membrane plane, the P-glycoprotein is toroidal, with six-fold symmetry and a diameter of ~10nm. There is a large central pore of about 5nm in diameter, which is closed on the cytoplasmic surface of the plasma membrane forming an aqueous chamber within the membrane. Overall, the observed microscopic image appears to be consistent with the proposed models. Although in recent years there has been a great advancement in our understanding of the structure of P-glycoprotein through mutational and biochemical studies, the precise molecular mechanism of drug transport by P-glycoprotein is still not fully understood.
1.2 ATP- and Substrate-Binding Sites

There are two ATP-binding domains of P-glycoprotein, located in the cytosol side. Each ATPbinding domain contains three regions: Walker A and B and Signature C motifs.[2] The sequences of amino acid residues for the first ATP-binding domain are: from 427435 for Walker A motif, from 531542 for Walker B motif, and from 551556 for Signature C motif. The corresponding amino acid residues for the second ATP-binding domain are from 10701078, from 11761182 and from 11961201, respectively. Recently, Hung et al.[21]
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

63

reported that a highly conserved Lys residue within the Walker A motif of histidine permease, an ABC transporter, is directly involved with the binding of ATP, and a highly conserved Asp residue within the Walker B motif serves to bind the Mg2+ ion. Mutations in either one of these residues result in nonfunctional activity of histidine permease. These results of histidine permease suggest that the ATP-binding sites may also be restricted to the Walker A motifs of P-glycoprotein. It is clear now that ATP binding and subsequent hydrolysis are essential for drug transport.[22] Studies with photoactive analogues of ATP have shown that these analogues bind to the ATP-binding domains.[23] Based on the data from vanadate trapping studies, Senior and Gadsby[24] have proposed a so-called alternate ATP-binding site model, which explains that although both ATPbinding sites are capable of binding ATP, only one site participates in the catalysis at a given time, and conformation of this catalytic site precludes the other site from hydrolysing ATP. The stoichiometry of ATP hydrolysis to drug transport has been studied, and the data indicate that, depending on substrate, 0.63 molecules of ATP are hydrolysed for every molecule of drug transported out the cell.[25,26] The reason for the substrate-dependent ATP stoichiometry is still unknown. Unlike the ATP-binding sites that are restricted to the Walker A motifs of ATP-binding domains, many substrate-binding sites have been identified throughout the transmembrane domains (TM) of P-glycoprotein. Two substrate-binding sites were found in TM6 and TM12 by using photoaffinity probes.[27,28] After complete digestion of the Pglycoprotein with trypsin, two major photolabelled fragments (5 and 4kD) were mapped by immunological analysis. These two fragments are located within, or immediately next to, the last transmembrane domain of each cassette, TM6 and TM12 of P-glycoprotein, respectively. The 5kD fragment includes amino acid residues from 311 456, extending a few residues beyond the Walker A motif of the first ATP-binding site. In contrast, the 4kD fragment includes residues from 979
Adis International Limited. All rights reserved.

1048, but not including the Walker A motif of the second ATP-binding site. In addition to the regions of TM6 and TM12, a region that includes TM7 and TM8 was also reported to be photolabelled specifically by an analogue of paclitaxel.[29] Consistent with the studies of photoaffinity probes, studies of many mutant P-glycoprotein molecules suggest that the major drug-binding sites reside in or near TM6 and TM12.[22] However, mutational studies also suggest that amino acid substitutions that affect substrate specificity are scattered throughout P-glycoprotein, including TM1, TM4, TM6, TM10, TM11 and TM12.[30] Recent studies by Taguchi et al.[31,32] suggested that three amino acids (His61, Gly64 and Leu65) in TM1 are involved in the formation of a binding pocket that plays a key role in determining the suitable substrate sizes for P-glycoprotein. For example, substitution of His61 by an amino acid with a short side-chain increased resistance to vinblastine (large molecular size; molecular weight 811), whereas substitution of an amino acid with a long side chain increased resistance to colchicine (small molecular size; molecular weight 399).[31,32] In addition to the transmembrane domains, mutational analyses have suggested that the intracellular linker loops of P-glycoprotein are also important for substrate recognition.[33,34] The systematic mutagenesis of 20 Gly residues in the cytoplasmic loops revealed that Gly141 and Gly187 between TM2 and TM3, Gly288 between TM4 and TM5, and Gly812 and Gly830 between TM8 and TM9 are important in determining substrate specificity.[33] In summary, these data suggest that drug-binding sites are scattered throughout the P-glycoprotein molecule, including the transmembrane domains, intracellular loops and even the ATPbinding domains.
1.3 Substrate Recognition

One of the most intriguing aspects of P-glycoprotein is that a single integral membrane protein can recognise and transport so many drugs with a wide array of chemical structures, ranging from a
Clin Pharmacokinet 2003; 42 (1)

64

Lin & Yamazaki

molecular weight of 250 (cimetidine) to 1202 (cyclosporin).[35,36] Although most of the drugs transported by P-glycoprotein are basic or uncharged, there are many exceptions. The only common feature is that most of the P-glycoprotein substrates are hydrophobic in nature, suggesting that partitioning of the lipid membrane of cells is the first step for the interaction of a substrate with the active sites of P-glycoprotein. Various efforts have been made to establish the structure-activity relationship for P-glycoprotein substrates. Originally, it was believed that a basic nitrogen atom was a prerequisite for the interaction of substrate and P-glycoprotein. For example, studies of colchicine and its analogues suggested that the nitrogen atom of the acetamido group at the 7 position was essential for P-glycoprotein recognition.[37] However, compounds lacking a nitrogen atom, such as cortisol, aldosterone and dexamethasone, are recognised to be good substrates for P-glycoprotein.[38] Studies by Ecker et al.[39,40] have shown that both the lipophilicity and number of hydrogen bonds of compounds are probably the most important parameters in determining the affinity of compounds to P-glycoprotein. The higher the lipophilicity or the larger the number of hydrogen bonds, the better the substrates are for the Pglycoprotein transporter. Similarly, Seelig and Landwojtowicz[41] have also suggested that both lipophilicity and number of hydrogen bonds are important determinants for substrates and P-glycoprotein interaction. They concluded that partitioning of the lipid membrane is the rate-limiting step for the interaction of a substrate with P-glycoprotein. Furthermore, they suggested that dissociation rate of the P-glycoproteinsubstrate complex is controlled by the number of hydrogen bonds. Based on structural analysis of 100 P-glycoprotein substrates, Seelig[42] further proposed that in addition to lipophilicity and number of hydrogen bonds, some essential structural elements of substrates are required for an interaction with P-glycoprotein. The recognition elements of substrates are formed by two or three electron donor (hydrogenbonding acceptor) groups
Adis International Limited. All rights reserved.

with a fixed spatial separation: 2.5 0.3 or 4.6 0.6. Additionally, the surface area and amphiphilic characteristic of the substrate also appear to play a significant role in determining its P-glycoprotein activity.[43] Although a wealth of information on the relationship between physicochemical properties of substrates and P-glycoprotein activity has been generated in recent years, a clear structure-activity relationship for predicting P-glycoprotein substrates still cannot be established.[44] The lack of clear structure-activity relationship for substrate recognition is attributed mainly to the structural complexity of P-glycoprotein. 2. Polymorphisms of P-Glycoprotein Humans are not necessarily created equal in terms of biological make-up. Because of evolutionary and environmental factors, there is a remarkable degree of genetic variability built into the population. Like many cytochrome P450 (CYP) isoenzymes,[45] genetic polymorphisms of P-glycoprotein in animals and humans have been reported. Thus, the genetic polymorphisms of P-glycoprotein may also represent a major source of individual variability in the potential toxicity and pharmacokinetics of drugs. The genetic polymorphism of P-glycoprotein was first reported in CF-1 mice by Lankas and Umbenhauer at Merck in 1997.[46,47] A subpopulation of CF-1 mice, approximately 25%, was very sensitive to neurotoxicity following exposure to avermectin, an antiparasitic agent. The 50% lethal dose (LD50) values were 0.3 and 120 mg/kg for the sensitive and insensitive groups, respectively. Subsequently, it is now known that this avermectininduced neurotoxicity is the result of a deficiency in mdr1a P-glycoprotein that normally contributes to a functional blood-brain barrier (BBB). In normal (wild-type) CF-1 mice the abundant P-glycoprotein in the BBB pumps avermectin efficiently out of the brain, but in mdr1a-deficient mice, this protective function is absent, resulting in a more than 80-fold higher accumulation of avermectin in the brain. The P-glycoprotein-deficient CF-1 mice
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

65

are also at higher risk of birth defects caused by avermectin.[48] When female CF-1 mice were treated with avermectin during pregnancy, fetuses deficient in P-glycoprotein (/) were 100% susceptible to cleft palate, while their heterozygote litters (+/-) were less sensitive. The homozygous fetuses (+/+) with abundant P-glycoprotein were totally insensitive at the dose tested. The birth defect is attributed to fetal exposure to avermectin. The above cases of neurotoxicity and teratogenesis demonstrate the importance of P-glycoprotein in protecting the brain and fetus against toxic xenobiotics. The molecular basis of mdr1a deficiency in CF-1 mice was further studied at the RNA and DNA level, using reverse transcription-polymerase chain reaction (RT-PCR) and long PCR with oligonucleotides specific for mdr1a. Sequencing of the intron between exon 22 and 23 in P-glycoprotein-deficient CF-1 mice revealed an insertion of approximately 8.35kb of DNA at the exon 23 intron-exon junction. This insertion results in the aberrant splicing of the mRNA and loss of exon 23 during RNA processing.[49] A subpopulation of collie dogs is also known to be very sensitive to avermectin and it has been speculated that the avermectininduced neurotoxicity in the dogs is due to P-glycoprotein genetic polymorphisms.[50,51] Recently, RT-PCR studies revealed a deletion mutation of the MDR1 gene in avermectin-sensitive collie dogs.[52] The 4-bp deletion results in a frame shift, generating several stop codons that prematurely terminate P-glycoprotein synthesis. Genetic polymorphisms of human P-glycoprotein were first reported from in vitro studies with cancer cells.[53,54] However, the kinetic impact of polymorphism on P-glycoprotein function in vivo remained unclear until the recent report by Hoffmeyer et al.,[55] who identified a single nucleotide polymorphism (SNP) in exon 26 (C3435T) of MDR1. There was a significant correlation of the SNP and the functional activity of P-glycoprotein. The homozygous T-allele (mutant) is associated with more than 2-fold lower intestinal P-glycoprotein expression levels compared with homozygous
Adis International Limited. All rights reserved.

C-allele (wild type). Individuals carrying homozygous T-allele showed a lower duodenal P-glycoprotein level and consequently higher peak plasma concentrations (Cmax) of digoxin, a substrate of Pglycoprotein, probably through an increase in digoxin absorption as a result of decreased intestinal P-glycoprotein. This was the first example that indicated that P-glycoprotein polymorphism can directly affect drug absorption in humans. However, a recent study by Sakaeda et al.[56] suggests that there are no statistically significant differences in the Cmax and 24-hour area under the curve (AUC24) of digoxin between individuals carrying wild-type C-allele or homozygous mutant T-allele. Similarly, conflicting results of the effect of C3534T polymorphism on the absorption of fexofenadine have also been reported. Kim et al.[57] showed that individuals harbouring homozygous T-allele mutation tended to have lower plasma AUC24 of fexofenadine. In contrast, Drescher et al.[58] claimed that there were no significant differences between T/T and C/T genotypes. Additionally, kinetic studies in healthy subjects and renal transplant patients suggested that C3534T polymorphism had little effect on the absorption of cyclosporin.[59,60] Therefore, the impact of C3534T polymorphism on drug absorption is still not clear. A strong association between C3435T allele and G2677A/G2677T alleles was observed when MDR1 polymorphisms were investigated in 100 placentas from Japanese women.[61] Of 65 samples with a C3435T allele, 61 (93.8%) also had a mutant G2677T/G2677A allele. Interestingly, there appeared to be a correlation between the level of Pglycoprotein expression and G2677T/G2677A in exon 21, and between the P-glycoprotein level and C3435T in exon 26. The placental P-glycoprotein expression levels for wild-type, heterozygotes and homozygotes of G2677A/G2677T mutant allele were 2.44, 1.97, and 1.45 (arbitrary units), respectively, while the corresponding mean P-glycoprotein expression levels for the C/C, C/T and T/T genotypes at position 3435 were 2.11, 1.84, and 1.51 (arbitrary units). The frequency in exon 21 occurred in 58% of the sample as heterozygosity
Clin Pharmacokinet 2003; 42 (1)

66

Lin & Yamazaki

and 28% as homozygosity for the mutant allele, while the frequency in exon 26 occurred in 46% of the sample as heterozygosity and 19% as homozygosity for the mutant allele. Given the high frequency for the mutant allele G2677A/G2677T in this study from 100 placentas, it is quite puzzling why no G2677A/G2677T mutant allele was observed from a sample population of 188 in the aforementioned study by Hoffmeyer et al.,[55] even though a total of 15 different SNPs, including C3435T, were identified in their study. Recently, Hoffmeyers group detected both C3435T and G2677A/G2677T polymorphisms with high frequencies in a relatively larger sample population of 461.[62] Similar to the ethnic variation in the polymorphism of CYP,[63] interethnic differences in MDR1 polymorphisms are observed. Using a polymerase chain reaction-restriction fragment length polymorphism assay, 1280 subjects from 10 different ethnic groups were evaluated for the C3435T polymorphism in exon 26.[64] Marked differences in genotype and allele frequency were observed between the African and the Caucasian/Asian populations. The Ghanaian, Kenyan, African-American and Sudanese populations have frequencies of 83, 83, 84 and 73%, respectively, for the C/C (wild type) allele. In contrast, the British Caucasian, Portuguese, Southwest Asian, Chinese, Filipino and Saudi populations have lower frequencies of the C/C allele compared with the African groups, ranging from 3455%. Although the level of P-glycoprotein expression was not determined in this study, these results suggest that P-glycoprotein expression in African populations may be higher than that in the Caucasian/Asian populations. It is interesting to note that ivermectin, a potent anthelmintic agent used for the prevention and treatment of river blindness (onchocerciasis) in Africa, is a very safe drug, even though this drug causes neurotoxicity in animals with low P-glycoprotein expression. To date, of more than 20 million patients in Africa who had been treated with ivermectin, not a single case of neurotoxicity has been reported. The lack of neurotoxicity might be
Adis International Limited. All rights reserved.

attributed to the high P-glycoprotein expression in the African population. On the other hand, the high expression of P-glycoprotein might contribute to the high incidence of drug resistance to cancer treatment in individuals of African origin.[65] In summary, the MDR1 gene is highly polymorphic. To date, at least 16 SNPs have been identified in the MDR1 gene, and it is anticipated that more SNPs will be found in the future (table I). Although some variants lead to amino acid changes, most of the detected polymorphisms are intronic or silent. So far, only three SNPs (T129C in exon 1b, G2677A/G2677T in exon 21 and C3435T in exon 26) have been demonstrated to be associated with variation in P-glycoprotein expression. Undoubtedly, variation in P-glycoprotein expression resulting from MDR1 polymorphism is one of the major sources contributing to interindividual variability in drug absorption and disposition. 3. In Vitro/In Vivo Extrapolation and Species Differences As will be discussed in section 4, P-glycoprotein plays an important role in absorption, distribution, metabolism and excretion of many drugs. Because of the importance of P-glycoprotein in pharmacokinetics, many pharmaceutical companies have begun to incorporate P-glycoprotein drug transport screening into the drug discovery process. The availability of transgenic mdr knockout mice and in vitro cell systems has paved the way for studies of the role of P-glycoprotein in drug absorption and disposition. When applying in vitro and in vivo screening models to study P-glycoprotein function, there are two fundamental questions that industrial drug metabolism scientists must confront daily: (i) can in vitro data be accurately extrapolated to the in vivo situation, and (ii) can animal data be directly scaled to humans? To test whether in vitro P-glycoprotein activity of drugs can be extrapolated to the in vivo situation, we have conducted a study to measure in vitro and in vivo P-glycoprotein activity of ten model compounds.[67] The in vitro P-glycoprotein activity of these compounds was determined using mdr1aClin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

67

Table I. Single nucleotide polymorphisms (SNPs) in the MDR1 gene SNPa 5-flanking/-41 1a/145 1b/-129 2/1 2/61 5/25 5/35 5/307 6/+139 6/+145 11/1199 12/1236 12/+44 17/76 17/+137 21/2677 21/2677 21/2677 24/2956 24/2995 26/3220 26/3396 26/3435 28/4030 28/4036 a Exon/position. Mutation A/G C/G T/C G/A A/G G/T G/C T/C C/T C/T G/A C/T C/T T/A A/G G G/T G/A A/G G/A A/C C/T C/T G/C A/G 0.3 48.1 53.9 49.0 0 25.0 0.2 0 0 5.9 5.6 9.3 16.5 0.6 0.6 40.6 1.2 6.5 37.8 5.9 45.3 0.6 56.5 41.6 1.9 36.5 41.7 21.8 0 6.7 56.4 43.6 0 38.5 (T>C) 5.5 41.0 4.9 46.2 35.4 (T>C) 0 37.2 9.0 11.2 Mutant allele frequency (%) Hoffmeyer et al.[55] Ito et al.[66] 7.3 1.0 Cascorbi et al.[62] Tanabe et al.[61] 9.4 1.0 8.3 Mickley et al.[54]

transfected LLC-PK1 cells, and expressed as the ratio of basolateral-to-apical transport to apical-tobasolateral transport (B-to-A/A-to-B). On the other hand, the in vivo P-glycoprotein activity was determined using CF-1 mdr1a (+/+) and mdr1a (/) mice. Following intravenous administration, the drug concentration in brain and plasma was measured every 15 minutes up to 60 minutes. The ratio of brain AUC in mdr1a (/) mice to that in mdr1a (+/+) mice was used as an index of in vivo P-glycoprotein activity. There was a strong positive correlation (r2 = 0.93, p < 0.001) when the in vitro B-to-A/A-to-B ratio was plotted against the in vivo brain AUC ratio for these ten compounds.[67] A strong correlation was also observed when the brain concentration was normalised by plasma concentration. These results suggest that in
Adis International Limited. All rights reserved.

vivo P-glycoprotein activity of a given drug can be reasonably well extrapolated from in vitro data. In our laboratory, the in vitro and in vivo correlation was further evaluated with an additional 20 compounds, and a strong correlation between in vitro B-to-A/A-to-B ratio and the in vivo brain AUC ratio was observed again (unpublished data). The most obvious species differences in the drug-transporting P-glycoprotein between mice and humans is that there are two members of drugtransporting P-glycoprotein (mdr1a and mdr1b) for mice, and just one in humans (MDR1).[2] Although the kinetics and substrate specificities are generally similar between mouse mdr1a and mdr1b P-glycoprotein,[68-70] species differences in functional activity between human MDR1 P-glycoprotein and these two mouse P-glycoproteins have
Clin Pharmacokinet 2003; 42 (1)

68

Lin & Yamazaki

been reported by Tang-Wai et al.[71] Stably transfected cells were developed that expressed similar amounts of P-glycoprotein encoded by mdr1a, mdr1b and MDR1 genes. The three transfected cell lines were tested for their cellular resistance (cell survival) to dactinomycin, doxorubicin, colchicine and vinblastine. The 50% inhibitory concentrations (IC50 values) for all of the drugs were much lower (3- to 20-fold) in MDR1 transfected cells than in mdr1a cells.[71] Likewise, with the exception of dactinomycin, the IC50 values were lower in human MDR1 cells than in mdr1b cells. These results suggest that there are species differences in functional capacity between human MDR1 P-glycoprotein and murine P-glycoproteins. In our laboratory, more than 640 compounds have been evaluated for their P-glycoprotein activity in mouse mdr1a and human MDR1 transfected LLC-PK1 cells (L-mdr1a and L-MDR1, respectively). As shown in figure 1, there is a poor correlation between mdr1a and MDR1 transcellular transport activity as measured by B-to-A/A-to-B transport ratio (r2 = 0.44). Approximately 35% of the compounds exhibited substantial differences (>3-fold) between mdr1a and MDR1 gene transfected cells. The observed species difference in the P-glycoprotein transport was not due to the differences in the level of the P-glycoprotein expression. Western blotting data revealed that the P-glycoprotein level was similar in both mdr1a and MDR1transfected cells. Species differences in transport activity were also observed between human P-glycoprotein and P-glycoprotein of other animal species. We have recently compared the P-glycoprotein activity of marker P-glycoprotein substrates using cell lines expressing human, mouse, rat and canine P-glycoprotein. Again, significant species differences in P-glycoprotein activity were found among these animal species (unpublished data). Although these in vitro studies strongly suggest the possibility of species differences in P-glycoprotein activity, there are still no in vivo data to support the potential of species differences. As will be discussed in section 4, the mdr1a/1b knockout
Adis International Limited. All rights reserved.

120

B-to-A/A-to-B in L-MDR1 (human)

100

80

60

40

20

0 0 20 40 60 80 100 120 B-to-A/A-to-B in L-mdr1a (mouse)

Fig. 1. Correlation of transcellular transport ratios for 642 structurally diverse compounds in monolayers of LLC-PK1 cells transfected with mouse mdr1a or human MDR1 (L-mdr1a and L-MDR1, respectively). A-to-B = apical to basal; B-to-A = basal to apical.

mouse provides a very useful model for the study of the role of P-glycoprotein in pharmacokinetics of drugs. However, because of the potential species differences in P-glycoprotein activity, extrapolation from knockout mice to humans should be carried out with discretion. 4. Role of P-Glycoprotein in Pharmacokinetics Although the physiological function for P-glycoprotein is still not fully understood, the role of this efflux transporter in pharmacokinetics is becoming increasingly appreciated. In humans, Pglycoprotein is found on the apical surface of columnar epithelial cells of small and large intestines, the biliary canalicular membrane of hepatocytes, the apical surface of epithelial cells of the proximal tubules of kidney, the apical surface of epithelial cells of placenta and the apical surface of endothelial cells in blood capillaries of the brain.[7,8] Because of its strategic localisation, the P-glycoprotein transporter functionally can limit
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

69

cellular uptake of drugs from the blood circulation into the brain and placenta, and from the gastrointestinal lumen into the enterocyte. On the other hand, this transporter can also enhance the elimination of drugs from hepatocytes, renal tubules and intestinal epithelial cells into the adjacent luminal space. Therefore, it is very important to distinguish the localisation of P-glycoprotein in cells in relation to drug movement either uptake of drugs into cells or excretion of drugs out of cells. As will be discussed later, there is a tendency for P-glycoprotein to have a greater impact on drug uptake than on drug excretion. Perhaps the most important milestone in P-glycoprotein research was the development of mdrknockout mice. Since mdr knockout mice became available, our understanding of the role of P-glycoprotein in pharmacokinetics has increased exponentially. As discussed above, mice have two types of drug-transporting P-glycoprotein (mdr1a and mdr1b), which are expressed in a tissue-specific manner.[72] For example, only mdr1a P-glycoprotein is expressed in the brain and intestine of mice, while both mdr1a and mdr1b P-glycoprotein are expressed in the liver and kidney. Interestingly, mdr1a and mdr1b P-glycoprotein together appear to cover the same tissues as the single human MDR1 P-glycoprotein, suggesting that mdr1a and mdr1b together fulfil the same function as the single P-glycoprotein in humans. Because of tissuespecific expression, it is expected that genetic disruption of the mdr1a gene would have a greater impact on drug uptake into the brain and intestine than drug excretion from the liver and kidney. In addition, it should be noted that genetic disruption of one or both of the mdr genes might affect the expression and function of other transporter systems or even drug-metabolising enzyme systems in mice. For example, it is known that the mdr1b gene is upregulated in mdr1a knockout mice.[8] Recently, Schuetz et al.[73] have reported that both the protein expression and catalytic activity of CYP is significantly increased in the mdr1a, mdr1b and mdr1a/1b knockout mice housed in The Netherlands. However, for the ge Adis International Limited. All rights reserved.

netically identical mdr1a and mdr1a/1b knockout mice housed in the US, there were no significant changes in their protein expression and catalytic activity of CYP. Because of the possible existence of unrecognised factors that are associated with the genetic disruption, data derived from mdr1a single knockout mice or from mdr1a/1b double knockout mice have to be interpreted with caution.
4.1 Drug Absorption

There are many factors that influence the bioavailability of drugs, which can be broadly categorised as physicochemical and biological factors.[74,75] The former comprise the intrinsic properties of the drug, such as pKa, molecular size, lipophilicity and solubility, and the latter include gastric and intestinal transit time, lumen pH, membrane permeability, mucosa blood flow rate and first-pass metabolism. After oral administration, drug absorption occurs predominantly within the small intestine, because of its large surface area provided by epithelial folding and the villous structures of epithelial cells. During oral absorption, drugs can be transported by either the transcellular or paracellular pathway across the epithelial cells, or a combination of both. The relative contribution of the transcellular pathway to overall absorption is highly dependent on the lipophilicity of drugs. In an in vitro study with Caco-2 cells, the relative contribution of the transcellular pathway was determined to be 25, 45, 85 and 99% for chlorothiazide, furosemide, cimetidine and propranolol, respectively. The values correlated fairly well with the lipophilicity of the drugs, the logP values of which were 0.2, 0.08, 0.4 and 3.6, respectively.[76] Most orally administered drugs enter the systemic circulation by passive transcellular diffusion, because of their lipophilicity. Therefore, the intestinal absorption of a drug is often predicted on the basis of its lipophilicity. As noted earlier, the most common physicochemical property for P-glycoprotein substrates identified so far is that they are mostly lipophilic, which implies that lipophilic drugs are likely to be P-glycoprotein substrates. Therefore, absorption
Clin Pharmacokinet 2003; 42 (1)

70

Lin & Yamazaki

of drugs is further complicated by the existence of P-glycoprotein efflux transporter, which is highly expressed on the apical surface of epithelial cells. In the intestinal lumen, drugs that are P-glycoprotein substrates will be absorbed and cross the epithelial cell membrane by simple diffusion. Once inside the cells, a fraction of the drug molecules continues to diffuse along the concentration gradient into capillary blood. However, a portion of drug molecules will be removed by the efflux Pglycoprotein transporter out of cells back into the lumen and another part of the drug molecules is subject to intestinal metabolism. Consequently, the net amount of drug absorbed into the mesenteric blood circulation is the difference between the amount absorbed by the influx process and the summation of the amount extruded by efflux transport together with the amount metabolised by enzymes.
4.1.1 Distribution of Intestinal P-Glycoprotein

The distribution of intestinal P-glycoprotein is not uniform among cells along the epithelial villi. Immunohistological studies with human jejunum and colon using MRK16 antibody revealed that high levels of P-glycoprotein were only observed in the apical surface of columnar epithelial cells, but not in crypt cells.[7] Unlike hepatocytes, which regenerate only when untimely death occurs, intestinal epithelial cells have a programmed, limited life span. The villous epithelial cells are mature and nondividing, whereas the crypt cells continue to mature as they ascend toward the villus and are extruded at its tip. The time required for migration from the crypt base to the villous tip has been estimated to be 2 6 days.[77] Whether the programmed life span and rapid migration will have impact on the regulation of intestinal P-glycoprotein expression is an open question, and requires further investigation. The distribution of P-glycoprotein is also not uniform along the length of intestine. Fojo et al.[78] have measured the content of MDR1 mRNA expression over the total length of human gastrointestinal tract. The levels of mRNA appear to increase progressively from the stomach to the colon
Adis International Limited. All rights reserved.

with a low level in the stomach (5 arbitrary units), an intermediate level in the jejunum (20 arbitrary units) and a high level in the colon (30 arbitrary units). The uneven distribution of intestinal P-glycoprotein is expected to have a significant impact on the absorption of P-glycoprotein substrates. The influence of uneven distribution of P-glycoprotein in intestine was demonstrated in a clinical study with cyclosporin.[79] Cyclosporin was given to ten healthy volunteers at different parts of the gastrointestinal tract (stomach, jejunum and colon). The oral AUC of cyclosporin was in the rank order stomach > jejunum > colon. There was a negative correlation between MDR1 mRNA expression and oral AUC of cyclosporin. Uneven distribution of P-glycoprotein has also been observed in rats. Using the rat intestinal loop technique, vinblastine, a well-known P-glycoprotein substrate, was absorbed fairly well from ileal loops, ranging from 3060% of the dose in 30 minutes, whereas absorption of vinblastine from the jejunal loop was almost negligible, suggesting a high level of P-glycoprotein expression in rat jejunum.[80]
4.1.2 Interindividual Variability of Intestinal P-Glycoprotein

Although interindividual variability in drugmetabolising enzymes is well documented,[45] only a few papers deal with the issue of interindividual variability of P-glycoprotein. In a clinical study of 25 kidney transplant recipients, expression level of intestinal P-glycoprotein was measured using immunoblotting.[81] Biopsy specimens were obtained from the second portion of the duodenum of each patient. Because P-glycoprotein is expressed exclusively in mature epithelial cells in the villous tip of intestinal mucosa, differences in the number of total mature cells in individual biopsies might contribute to the variability. To correct this practical problem, the investigators used villin, a constitutively expressed protein in mature epithelial cells, as an internal standard. The results from this study indicate that there is a significant interindividual variability in the intestinal P-glycoprotein expression. More than 8-fold differences in the P-glycoprotein expression were observed in a small popuClin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

71

lation of 25 patients; the intestinal P-glycoprotein level of duodenal biopsies ranged from 31263 (arbitrary units). As discussed earlier, the MDR1 gene is highly polymorphic. At least 16 SNPs have been identified (table I). Although complete P-glycoprotein deficiency has not been reported for these polymorphisms, the SNP at 3435 in exon 26 does influence the expression level of intestinal P-glycoprotein. The mean values of intestinal P-glycoprotein expression for the C/C homozygotes (wild type, n = 6), C/T heterozygotes (n = 10), and T/T homozygotes (n = 5) at position 3435 were 1275, 956 and 627 (arbitrary units), respectively.[55] Interestingly, intraindividual variability in intestinal P-glycoprotein expression has also been reported. A profound intraindividual variability in intestinal P-glycoprotein expression (mRNA) was observed in a young patient during tacrolimus therapy after small bowel transplantation.[82] Both the mRNA expression and plasma concentration of tacrolimus were measured periodically during the immunosuppressant therapy. In a period of 120 days, there was a 4-fold variation in MDR1 mRNA expression level and a 2-fold variation in trough plasma concentration of tacrolimus. The variation in P-glycoprotein expression inversely related fairly well to the variation of tacrolimus concentrations after oral administration. Collectively, these results suggest that interindividual and intraindividual variability in intestinal P-glycoprotein expression may contribute to variability of oral absorption of drugs that are Pglycoprotein substrates.
4.1.3 Evidence of Intestinal P-Glycoprotein Involvement in Drug Absorption

Evidence of the involvement of intestinal Pglycoprotein in drug absorption was first demonstrated in vitro with Caco-2 cells in which P-glycoprotein was highly expressed. Using Caco-2 cells, the B-to-A transport of vinblastine and docetaxel was 10- and 20-fold, respectively, greater than the A-to-B transport, and the A-to-B transport was enhanced significantly in the presence of verapamil and MRK16 by blocking the
Adis International Limited. All rights reserved.

P-glycoprotein efflux function.[83,84] Caco-2 cells have also been used to study the intestinal transport of cyclosporin.[85] Similarly, the B-to-A transport of cyclosporin was much greater than the A-to-B transport. The A-to-B cyclosporin transport increased and the B-to-A transport decreased after treatment with the P-glycoprotein inhibitors progesterone and chlorpromazine. Furthermore, studies with Caco-2 cells revealed that active B-to-A transport of peptides was inhibited by verapamil, suggesting the involvement of P-glycoprotein in the absorption of peptides.[86] These results from in vitro studies clearly suggest that P-glycoprotein plays a significant role in drug absorption by limiting drug transport from intestinal lumen. Direct evidence for the role of intestinal P-glycoprotein in drug absorption was derived from in vivo studies with mdr1a (/) knockout mice. The oral absorption of paclitaxel was studied in mdr1a (/) and mdr1a(+/+) mice.[87] The plasma AUC of paclitaxel was 2- and 6-fold, respectively, higher in mdr1a (/) mice than mdr1a (+/+) mice after intravenous and oral drug administration. The increased AUC of paclitaxel after intravenous administration in mdr1a (/) mice reflected a decrease in elimination clearance, whereas the higher AUC after oral administration in mdr1a (/) mice resulted from a combination of a decrease in the elimination clearance and an increase in the extent of drug absorption from intestinal lumen. Based on the AUC values after intravenous and oral administration, the bioavailability of paclitaxel was calculated to be 11 and 35% for mdr1a (+/+) and mdr1a (/) mice, respectively. From this study, it is clear that intestinal P-glycoprotein does limit drug absorption by extruding drugs from epithelial cells back into the intestinal lumen. Further evidence for the involvement of intestinal P-glycoprotein in drug absorption in humans is provided by the clinical study by Hoffmeyer et al.,[55] who showed a negative correlation between duodenal P-glycoprotein expression and plasma level of digoxin. The efflux function of intestinal P-glycoprotein is further supported by the observations that a sigClin Pharmacokinet 2003; 42 (1)

72

Lin & Yamazaki

nificant amount of paclitaxel was excreted directly from blood circulation into intestinal lumen after intravenous administration. The effect of P-glycoprotein on intestinal excretion can be experimentally determined in mice after interruption of bile flow by gallbladder cannulation. A fraction of 11% of the dose was excreted into the intestinal lumen within 90 minutes after intravenous administration of paclitaxel to mdr1a (+/+) mice with a cannulated gallbladder, but only 2.5% of the dose in the intestinal lumen of mdr1a (/) mice.[87] Similarly, Pglycoprotein-mediated intestinal excretion in mice was also reported for digoxin. Approximately 16 and 2% of the digoxin dose, respectively, was excreted into the intestinal lumen of mdr1a (+/+) and mdr1a (/) mice with a cannulated gallbladder within 90 min after intravenous administration.[88] Although the phenomenon of intestinal excretion of drugs has been known for more than two decades,[89] involvement of P-glycoprotein in intestinal excretion has only been recognised recently. From a mechanistic point of view, intestinal excretion should also be considered as an additional pathway for the elimination clearance of P-glycoprotein substrates. Unlike the direct evidence obtained from mdr1a knockout mice, the involvement of MDR1 P-glycoprotein in drug absorption is difficult to prove directly in humans. Thus, the role of intestinal Pglycoprotein in drug absorption in humans is often derived indirectly from inhibition studies. In a clinical study, five patients received a safe oral dose of paclitaxel 60 mg/m2, and nine other patients received the same oral dose of paclitaxel combined with one single oral dose of cyclosporin 15 mg/kg.[90] The oral bioavailability of paclitaxel when given without cyclosporin was less than 5%, and increased to 50% when cyclosporin was coadministered. Similar results were found in another clinical study with docetaxel.[91] The bioavailability of docetaxel in cancer patients increased from 8% without cyclosporin to 88% in combination with cyclosporin. Because cyclosporin is known to be a potent P-glycoprotein inhibitor, these results suggest the involvement of
Adis International Limited. All rights reserved.

P-glycoprotein as an intestinal barrier in limiting the absorption of paclitaxel and docetaxel. Since cyclosporin is also an inhibitor of CYP3A4 and other CYP enzymes, it is likely that the observed increase in oral bioavailability of these drugs could also be partly attributed to reduced metabolism by the inhibition of CYP enzymes by cyclosporin. Moreover, cyclosporin is a potent inhibitor of other transporters, such as canalicular bile salt transporter and canalicular multispecific organic anion transporter (cMOAT). The inhibitory Ki values of cyclosporin for taurocholate (bile salt transporter), leukotriene C4 (cMOAT) and daunorubicin (P-glycoprotein) were 0.2, 3.4 and 1.5 mol/L, respectively.[92] Clearly, the increased bioavailability of paclitaxel and docetaxel caused by cyclosporin cannot be explained by P-glycoprotein inhibition alone. Therefore, the inhibition study can only provide a qualitative assessment as to whether P-glycoprotein is involved in drug absorption. Due to the lack of a specific P-glycoprotein inhibitor, it is difficult to estimate the quantitative contribution of P-glycoprotein to drug absorption by the inhibition approach. Another way by which evidence can be shown for intestinal P-glycoprotein involvement in drug absorption is to establish the correlation between the absorption profile (oral AUC) of drugs and the expression of intestinal P-glycoprotein (or mRNA). In a clinical study, cyclosporin was given by gavage to ten male volunteers at different parts of the gastrointestinal tract (stomach, jejunum/ileum and colon) in a crossover manner, with a washout period between the administrations of at least 7 days.[79] There was a strong negative correlation between the plasma AUC of cyclosporin after administration at different locations of the gastrointestinal tract and the local mRNA expression of intestinal P-glycoprotein. Similarly, based on the observation that a highly significant correlation exists between enterocyte P-glycoprotein content and cyclosporin absorption kinetics, Lown et al.[81] concluded that intestinal P-glycoprotein plays a significant role in the absorption of cyclosporin.
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

73

With several lines of supportive evidence, the conclusion from these two correlation studies that P-glycoprotein transport is involved in the absorption of cyclosporin appears valid and appropriate. However, it should be noted that a good correlation itself does not prove a causal relationship. Therefore, it is highly desirable to have other supportive in vitro and/or animal data when applying the correlation approach.
4.1.4 Saturable Efflux Transport by Intestinal P-Glycoprotein

Like that of drug-metabolising enzymes, the functional activity of P-glycoprotein is saturable. In a recent kinetic study involving Caco-2 cells,[93] the efflux of vinblastine and digoxin by P-glycoprotein have been demonstrated to be saturable; the Michaelis-Menten constants (Km values) for vinblastine and digoxin were 26 and 58 mol/L, respectively. A similar Km value for vinblastine (18 mol/L) was reported by other investigators using Caco-2 cells.[83] By using the Ussing chamber technique, the transport of digoxin was studied in human and rat intestinal tissues. Saturable transport for digoxin was also observed in both human and rat intestinal tissues. The Km values for digoxin were 81, 74, 51 and 59 mol/L, respectively, for rat jejunum, rat ileum, rat colon and human colon.[86] Interestingly, the Km values for digoxin derived from human colon tissue and Caco-2 cells are almost identical (59 vs 58 mol/L).[93] Using Caco-2 cells, transport of cyclosporin was shown to be saturable with a Km of 3.8 mol/L.[79] This Km value for cyclosporin was comparable to that found in other cell lines expressing human P-glycoprotein (8.4 mol/L).[94] Collectively, these results strongly suggest that the efflux function of intestinal P-glycoprotein may be saturated when drug concentrations in the intestinal lumen exceed the Km values after high oral doses. As discussed previously, for drugs that are Pglycoprotein substrates the net amount of drug passing through the intestinal epithelial cells is the difference between the amount absorbed by influx processes (passive diffusion and/or active uptake) and the summation of the amount extruded by
Adis International Limited. All rights reserved.

efflux transport together with the amount metabolised by enzymes. It is evident that the P-glycoprotein-mediated efflux and CYP-mediated metabolism are saturable processes. The saturable P-glycoprotein efflux may, at least in part, explain the observed dose-dependent absorption of talinolol (a P-glycoprotein substrate) in healthy volunteers.[95] For both enantiomers, the dosenormalised AUC increased with increasing doses after oral administration. The dose-normalised AUC of (S)-()-talinolol increased from 18 g h/L at a 12.5mg dose to 36 g h/L at a 200mg dose. Similar results were observed for (R)-(+)talinolol. Consistent with the in vivo observations, concentration-dependent permeability across Caco2 cell monolayers was observed when the concentration of talinolol was increased from 0.12 mmol/L. Similarly, saturable P-glycoproteinmediated efflux has also been reported for cyclosporin in rats.[96] The extent of cyclosporin absorption increased with increasing doses in rats; the bioavailability increased from 13% at an oral dose of 6 mg/kg to 25% at 18 mg/kg. These results indicate that P-glycoprotein-mediated efflux transport can be saturated when higher oral doses are given. There is a widespread misconception that the extent of oral absorption of a drug is always markedly limited by intestinal P-glycoprotein when the drug is a P-glycoprotein substrate. This is only true for a few P-glycoprotein substrate drugs that are given at low doses. Absorption of digoxin is a good example. Digoxin, a well-known P-glycoprotein substrate, which undergoes minimal metabolism, is given orally at a very low oral dose of 0.5 to 1mg. At these doses, the concentration of digoxin in intestinal lumen is estimated to be less than 10 mol/L, which is well below the Km value (58 mol/L) derived from Caco-2 cells or human colon.[93] Thus, P-glycoprotein plays a quantitatively significant role in the absorption of digoxin. The reported low and variable absorption of digoxin can most probably be attributed to the efflux transport of intestinal P-glycoprotein.
Clin Pharmacokinet 2003; 42 (1)

74

Lin & Yamazaki

However, the oral dose for most drugs is high (>50mg) and drugs in the intestinal lumen can easily reach the mmol/L concentration range. A detailed literature survey revealed that all of the reported Km values for P-glycoprotein drugs are relatively low, ranging from 4213 mol/L (table II). Given the low Km values for P-glycoprotein drugs, P-glycoprotein activity can readily be saturated when drugs are administered at high doses, and hence the role of intestinal P-glycoprotein in drug absorption becomes quantitatively less significant. Indinavir, an HIV protease inhibitor, is a Pglycoprotein substrate and is given orally at a dose of 800mg. At this high dose, the indinavir concentration in the intestinal lumen is expected to be greater than 1 mmol/L, which is much higher than the Km value for P-glycoprotein transport (140 mol/L) derived from Caco-2 cells.[97] Therefore, at high doses, the effect of intestinal P-glycoprotein on indinavir absorption becomes quantitatively less important. This can explain why indinavir has a reasonably good bioavailability (>60%) in patients, even though it is a good P-glycoprotein substrate.[98] Interestingly, in a recent literature survey, Chiou et al.[99] have concluded that

the in vivo oral absorption of 13 drugs is not significantly impeded by efflux transport, in spite of being good P-glycoprotein substrates. However, there are exceptions that intestinal Pglycoprotein still plays a significant role in absorption for some drugs, even when they are given at high doses. For example, the clinical oral dose is 200700mg for cyclosporin and 100200mg for paclitaxel, but clinical studies clearly indicate that P-glycoprotein does play a significant role in limiting their oral absorption.[79,82,87,101] This can be explained by the fact that both cyclosporin and paclitaxel have very poor water solubility, slow dissolution rate and large molecular weight (1202 for cyclosporin and 854 for paclitaxel). The poor water solubility and slow dissolution rate can result in low drug concentration in the intestinal lumen in relation to their Km value for P-glycoprotein transport, and the large molecular size can impede the rate of passive diffusion across the cell membranes. The notion of slow dissolution rate and/or slow membrane diffusion rate of cyclosporin in intestine is supported by the fact that peak concentrations of the drug occur slowly at 34 hours after administration to patients.[101]

Table II. Apparent values of the Michaelis-Menten constant (Km) for P-glycoprotein substrates Compound Cyclosporin Digoxin Material (flux evaluated) Caco-2 (net B-to-A) Caco-2 (net B-to-A) Stripped rat jejunum (net B-to-A) Stripped rat ileum (net B-to-A) Stripped rat colon (net B-to-A) Stripped human colon (net B-to-A) Etoposide Caco-2 (B-to-A) Stripped rat jejunum (B-to-A) Stripped rat colon (B-to-A) Indinavir Verapamil Caco-2a (net B-to-A) Stripped rat jejunum (B-to-A) Stripped rat ileum (B-to-A) Stripped rat colon (B-to-A) Vinblastine Caco-2 (net B-to-A) Caco-2 (net B-to-A) Stripped rat ileum (net B-to-A) Stripped rat colon (net B-to-A) a Treated with calcitriol. B-to-A = basal to apical. Apparent Km (mol/L) 3.8 58 81 74 51 59 213 94 119 140 31 29 4.4 19 27 48 ~100 References 79 93 93 93 93 93 100 100 100 97 100 100 100 83,84 93 93 93

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

75

In conclusion, it is clear that the effect of intestinal P-glycoprotein on drug absorption is unlikely to be quantitatively important unless a very small oral dose is given, or the dissolution and/or membrane diffusion rates of the drug are very slow.
4.2 Drug Distribution

certainly provide important insights into the mechanisms of drug distribution in the very near future.
4.2.1 Blood-Brain Barrier Lipophilicity and Brain Penetration

Drugs are often administered at a location distant from their intended site of action. To be effective, the drug must be absorbed and transported from the site of administration across several biomembranes to reach the target tissue and the site of action. Penetrating cell membranes is a complex process that is highly dependent on the nature of the membrane and the physicochemical properties of the drug. The physical and biochemical properties of membranes, such as lipid bilayer structure and dynamics, play an important role in drug penetration. In addition, the physicochemical properties of drugs, such as hydrophobicity, ionisation profile, molecular size and number of hydrogen bonds, also play a significant role in membrane penetration.[102] Although extensive efforts have been made to study the molecular mechanisms of the processes of drug absorption, metabolism and excretion, drug distribution has historically received much less attention than the other processes, in spite of its importance as a key factor in determining drug response. Because of this lack of attention, drug distribution has been regarded as a forgotten relative in clinical pharmacokinetics.[103] The lack of attention stems partly from a lack of useful experimental tools in studying drug distribution. However, the importance of drug distribution is becoming increasingly recognised. For example, Schinkel et al.[8,9] demonstrated large differences in drug distribution into the brain and other tissues between mdr1a (/) and mdr1a (+/+) mice. In addition, with recent advances in the molecular biology and biochemistry of transporter systems, several in vitro systems have now been developed for studying drug distribution.[104] These newly developed tools and experimental methodologies will
Adis International Limited. All rights reserved.

The brain is different from other organs of the body in many aspects. One of the most important features is that the brain is anatomically separated from the blood circulation by the BBB. All other organs are perfused by capillaries lined with endothelial cells that have small pores to allow for movement of drugs into the organ interstitial fluid from the circulation. However, the endothelial cells in brain capillary blood vessels are closely joined to each other, leaving no space between cells. Consequently, only lipophilic drugs can cross endothelial cells and enter the BBB by way of passive diffusion. A strong positive correlation between lipophilicity and brain penetration of drugs has been reported by many investigators.[105,106] In addition, factors other than lipophilicity may also play an important role in the transport of drugs across the BBB. For example, a negative correlation was found between the BBB permeability of lipophilic compounds (steroid hormones and peptides) and the total number of hydrogen bonds; the greater the total number of hydrogen bonds, the lower the permeability.[105,107] In addition, the molecular size of drugs is also an important determinant for brain penetration.[108] Although lipophilicity is an important factor in determining the BBB penetration of drugs, many lipophilic drugs have exhibited poor BBB penetration. In a rat study, Levin[108] reported a good correlation between the in vivo BBB permeability coefficient of 22 compounds and their lipophilicity. However, they found that vincristine and epipodophylotoxin displayed poor BBB permeability, despite relatively high lipophilicity (logP value of 2.8). Additionally, many other lipophilic compounds also exhibit poor BBB penetration. For example, a potent CCKB receptor antagonist candidate is a lipophilic compound with a logP value of 3.6, and it has poor brain penetration.[109] The poor BBB penetration of these drugs cannot be explained by the number of hydrogen bonds and
Clin Pharmacokinet 2003; 42 (1)

76

Lin & Yamazaki

molecular weight. Twenty years ago, these compounds were regarded as outlier compounds for brain penetration without knowing the exact cause. The possible efflux function of P-glycoprotein in the BBB was not connected with the observed poor BBB permeability of lipophilic drugs until the findings of P-glycoprotein in brain capillaries by Thiebaut et al.[110] and Cordon-Cardo et al.[111] Using monoclonal antibodies, they demonstrated that P-glycoprotein is highly expressed on the apical surface of the endothelial cells of the brain capillaries. With these findings, it is now clear that the observed poor BBB permeability of those lipophilic drugs is due mainly to the efflux function of P-glycoprotein.
Localisation of P-Glycoprotein in Brain

Localisation of P-glycoprotein in the brain is a widely disputed issue that has become the centre of much controversy in recent years. Beaulieu et al.[112] provided very convincing evidence for Pglycoprotein being predominantly localised in the luminal membrane of endothelial cells of rat brain capillaries facing blood circulation. Using a novel technique with cationic colloidal silica, the investigators were able to selectively isolate the luminal membrane of endothelial cells of rat brain capillaries. The isolation procedures resulted in a membrane preparation with a 9.9-fold enrichment of the endothelial membrane marker protein GLUT1 (glucose transporter 1) and a 17-fold enrichment of P-glycoprotein relative to isolated brain capillaries. Enrichment of GLUT1 and P-glycoprotein relative to whole brain membrane preparations was 280- and 500-fold, respectively. However, glial fibrillary acidic protein (GFAP), a specific marker for astrocytes, was enriched only 1.4-fold relative to brain capillary, suggesting a very minor contamination of astrocytes in the luminal membrane preparations. The co-enrichment of P-glycoprotein (500-fold) and GLUT1 (280-fold) in brain capillary luminal membranes compared with whole brain membrane preparations strongly suggests that P-glycoprotein is expressed predominantly in the luminal membrane of brain endothelial cells.
Adis International Limited. All rights reserved.

Consistent with Beaulieus results, Barrand et al.,[113] using a dual immunostaining approach in combination with confocal microscopy, concluded that endothelial marker C219 staining did not colocalise with astrocyte marker GFAP staining in rat brain microvessels. Similarly, Matsuoka et al.[114] have demonstrated that P-glycoprotein is localised in the brain capillaries of rats, using P-glycoprotein and GFAP double-immunolabelling technique. Furthermore, a recent study by Decleves et al.[115] showed that P-glycoprotein is expressed in both cultured rat endothelial cells and astrocytes. In this study, RT-PCR analysis showed that mdr1b mRNA was preferentially expressed in astrocytes, whereas both mdr1a and mdr1b mRNA were detected in endothelial cells. In this study, Western blotting analysis revealed much higher expression level of P-glycoprotein in endothelial cells compared with astrocytes. Collectively, these results consistently suggest that brain P-glycoprotein is predominantly expressed on the apical surface of endothelial cells of capillaries. Although most immunohistochemical studies indicate that P-glycoprotein is predominantly localised on the surface of endothelial cells facing the luminal side, Pardridge et al.[116] claim that Pglycoprotein is localised mainly to astrocyte foot processes. They conducted an immunochemical study with human brain microvessels using (a) the MRK16 antibody to human P-glycoprotein, (b) an antiserum to GFAP and (c) an antiserum to GLUT1. They observed that the P-glycoproteinspecific antibody MRK16 bound to microvessels with a similar, discontinuous staining pattern as an antiserum directed against GFAP. The apparently discontinuous and abluminal localisation of MRK16 staining in isolated brain capillaries and the similarity in immunostaining patterns by MRK16 and anti-GFAP antibodies led the investigators to conclude that P-glycoprotein is expressed in astrocyte foot processes, rather than in endothelial cells. The conclusion was further supported by their subsequent findings that staining of the endothelial membrane marker protein GLUT1 was continuous and showed only minimal overlap with
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

77

MRK16 staining. With their observations, Golden and Pardridge[117] have proposed a revised kinetic model of P-glycoprotein in the brain in contrast to the classic kinetic model. According to the revised kinetic model of Golden and Pardridge,[117] a deficiency (or inhibition) of P-glycoprotein would have no effect on BBB permeability, but would cause a decrease in the drug concentration in the interstitial fluid (ISF) of brain. On the other hand, according to the classic model, a deficiency (or inhibition) of P-glycoprotein would result in an increase in BBB permeability as well as an increase in drug concentration in the ISF of brain. Therefore, comparison of brain uptake of P-glycoprotein substrates in normal and P-glycoprotein-deficient mice can be used to address the question of whether the P-glycoprotein transporter is localised in the capillary endothelial cells or in the astrocytes. Using an intracerebral microdialysis technique, de Lange et al.[118] showed that a deficiency of P-glycoprotein in mice resulted in the same degree of increase in drug concentrations of rhodamine-123 (a P-glycoprotein substrate) in the brain as well as ISF. After an intravenous infusion of rhodamine-123, the total brain concentrations were about four times higher in the mdr1a (/) mice compared with wild-type mice. Similarly, the rhodamine-123 concentrations in ISF were also about four times higher in mdr1a (/) mice than in mdr1a (+/+) mice. These results appear to be consistent with the classic concept that in mice the P-glycoprotein transporter is expressed in the brain capillary endothelial cells. Recently, Lee et al.[119] have demonstrated that P-glycoprotein is expressed and functional in brain microglia. Using a continuous rat brain microglia cell line (MLS-9), immunocytochemistry studies revealed the location of P-glycoprotein along the nuclear envelope and plasma membrane of microglia. Furthermore, the accumulation of digoxin by microglia was significantly enhanced by valspodar (PSC-833), a potent P-glycoprotein inhibitor. Consistent with these findings, P-glycoprotein has also been detected in mixed glial cells, but not in primary cultured neurons, by other in Adis International Limited. All rights reserved.

vestigators.[114] The expression of P-glycoprotein in glial cells may partly explain the intriguing pharmacokinetics and pharmacodynamics of [Dpenicillamine2,5]enkephalin (DPDPE) in mdr1a (/) and mdr1a (+/+) mice demonstrated by Chen and Pollack.[120] Although the brain concentrations of DPDPE were 2- to 4-fold higher in mdr1a (/) mice than in mdr1a (+/+) mice after intravenous administration, the dose required to elicit comparable antinociception was more than 30-fold lower in mdr1a (/) mice compared with mdr1a (+/+) mice. Based on brain concentrations, the EC50 (concentration producing half-maximal antinociception effect) of DPDPE was 13 times lower in mdr1a (/) mice compared with wild-type mice (12 versus 160 ng/g). These results suggest either differences in DPDPE distribution within the brain or differences in the intrinsic activity of -opioid receptors between mdr1a (/) mice and mdr1a (+/+) mice. Subsequent pharmacokinetic and pharmacodynamic modelling suggested that the difference in antinociception between mdr1a (/) and mdr1a (+/+) mice was due to the distribution of DPDPE within the brain as well as between the blood and brain, but not due to differences in intrinsic response. These results are consistent with the notion that P-glycoprotein is expressed in other type of brain cells in addition to the endothelial cells of capillaries.
Evidence of P-Glycoprotein Involvement in Brain Uptake

The first experimental evidence that P-glycoprotein is involved in drug transport in the BBB came from Tsuji and coworkers.[121] Immunostaining with a P-glycoprotein antibody (MRK16) demonstrated an exclusively apical localisation of P-glycoprotein in the cultured bovine brain endothelial cells. Kinetic studies showed that efflux transport of vincristine from the bovine brain endothelial cells was inhibited by verapamil, resulting in a significant increase in intracellular drug concentration. Similar results were also reported by Tsuruo and colleagues[122] with mouse brain capillary endothelial cells. They showed by immunochemical studies that P-glycoprotein was
Clin Pharmacokinet 2003; 42 (1)

78

Lin & Yamazaki

localised on the apical surface of endothelial cells of mouse brain. The unidirectional transport of vincristine from basolateral side to apical side was demonstrated in the polarised monolayer of mouse endothelial cells. Similarly, efflux transport of cyclosporin was also observed in cultured endothelial cells of bovine and mouse brain capillaries.[123,124] Immunostaining with P-glycoprotein antibody also demonstrated an exclusively apical localisation of P-glycoprotein in human brain endothelial cells.[125] These in vitro studies provide evidence of functional involvement of P-glycoprotein in the BBB penetration of drugs. Although, in vitro studies have shown that ATP is essential for P-glycoprotein transport function,[126,127] the involvement of ATP in P-glycoprotein-mediated transport was first demonstrated in vivo by Tsuji and colleagues.[128] They demonstrated that the BBB permeability coefficient of doxorubicin increased from 14 l/min/g brain in control rats to 243 l/min/g brain in rats ATPdepleted by occlusion of vertebral and common carotid arteries. The BBB permeability coefficient of doxorubicin was determined by using in situ brain perfusion technique. Under the experimental conditions, the ATP content in the rats with transient brain ischaemia was only 3% of that in normal rats (0.04 versus 1.43 mol/g brain). Similarly, an increase in the BBB permeability coefficient of cyclosporin in ATP-depleted rats was also reported by Tsujis group.[129] These results from the ATPdepleted rat studies are consistent with the classic concept that the P-glycoprotein is expressed in the brain capillary endothelial cells, because the BBB permeability decreases when P-glycoprotein function is impaired. The knockout animal model of mdr1a(/) mice also provides a powerful tool for studying brain uptake of drugs. Oral administration of [3H]ivermectin in mdr1a (/) and mdr1a (+/+) mice resulted in 87-fold higher levels of radioactivity in the brain of mdr1a (/) mice as compared with wild-type mice, whereas the levels in liver, kidney, small intestine and plasma were increased by less than 4-fold.[9] In another study, when [3H]digoxin
Adis International Limited. All rights reserved.

and [3H]cyclosporin were given intravenously, markedly higher brain levels of radioactivity (17and 55-fold, respectively) were observed in mdr1a (/) mice than in mdr1a (+/+) mice.[130] Again, only a moderate increase in radioactivity levels (2to 3-fold) of these two drugs was observed for liver, kidney, small intestine and plasma of mdr1a (/) mice. The large differences in brain concentration were also observed between mdr1a/1b (/) double knockout and normal mice.[10] A 27-fold increase in the brain concentration of digoxin was observed in mdr1a/1b (/) double knockout mice compared with the wild-type mice, with only a 2.5fold increase in digoxin concentration in the liver, kidney and plasma of double knockout mice. It is puzzling why the most marked increase in drug concentration of P-glycoprotein substrates in mdr1a (/) knockout mice is always observed in the brain, while the increases in drug concentration in other tissues in which P-glycoprotein is also highly expressed, such as liver and kidney, are relatively modest. As mentioned earlier, the mdr gene in mice is expressed in a tissue-specific manner. Both mdr1a and mdr1b genes are expressed in the liver and kidney, while only mdr1a gene is expressed in the brain of mice.[131] At first glance, one might assume that the less profound increases in drug concentration in the liver and kidney in mdr1a (/) mice are due to the extra protective functions of mdr1b P-glycoprotein in these tissues. However, even in mdr1a/1b (/) double knockout mice, increases in drug concentrations in the liver and kidney are also much lower than that in the brain.[10] The marked increases in drug concentration in the brain of P-glycoprotein-deficient mice also cannot be explained by the loss of BBB integrity. Experiments performed using fluorescein and fluorescein-dextran-4000 as integrity markers showed that there were no differences in brain/plasma concentration ratio of these compounds between mdr1a (/) and wild-type mice, being 2.3 and 1.4%, respectively, indicating maintenance of BBB integrity in the absence of P-glycoprotein.[118] The underlying mechanism for the marked increases in brain concentration remains
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

79

unknown. Regardless of the underlying mechanisms, the data from the mdr1a and mdr1a/1bknockout mice studies suggest that the brain is more sensitive to changes in P-glycoprotein function than other tissues. Therefore, P-glycoprotein inhibitors should be used with caution to avoid potential neurotoxicity. Another important observation from these mdr1a and mdr1a/1b knockout mice studies is that for certain P-glycoprotein substrates, an appreciable amount of drug is still observed in the brain of wild-type mdr1a (+/+) mice. For example, the brain concentrations of tacrolimus were about 2.7fold higher than the corresponding plasma concentrations at the same time point in mdr1a(+/+) mice after intravenous administration.[132] Similarly, appreciable brain concentrations of cyclosporin, ranging from 3070% of the corresponding plasma concentration, were observed in mdr1a (+/+) mice.[130] Kinetically, the drug concentration in the brain is determined by the difference between the amount of drug transported by influx processes (passive diffusion and active uptake) and the amount of drug extruded by the P-glycoproteinmediated efflux process. A fraction of drug molecules can reach the brain tissue if the influx diffusion rate is greater than the P-glycoprotein efflux rate. Nonspecific binding to brain tissue may also be a contributing factor in determining drug concentration in the brain. Therefore, one should take the rate of influx diffusion, rate of P-glycoprotein efflux transport and nonspecific binding of compounds into consideration when predicting brain penetration.
4.2.2 Placenta

As noted earlier, Lankas et al.[48] have clearly demonstrated the protective role of placental Pglycoprotein in reducing fetal exposure to xenobiotics in CF-1 mice. The role of placental P-glycoprotein in protection of the fetus has been further evaluated using mdr1a/1b (/) double knockout mice.[133] Heterozygous mdr1a/1b (+/) female mice were mated with heterozygous male mice to produce fetuses of three genotypes: mdr1a/1b (/), mdr1a/1b (+/) and mdr1a/1b (+/+).
Adis International Limited. All rights reserved.

Following intravenous administration of digoxin, saquinavir and paclitaxel to pregnant heterozygous dams, fetal drug exposure was much higher in the mdr1a/1b (/) fetus than the wild-type mdr1a/1b (+/+) fetus. The ratio of drug concentration in the mdr1a/1b (/) fetus to that in the wildtype fetus was 2.5, 5 and 16, respectively, for digoxin, saquinavir and paclitaxel. On the other hand, the drug concentrations in the heterozygous mdr1a/1b (+/) fetus were similar to those in the wild-type fetus, suggesting that the P-glycoprotein level in the placenta of the heterozygous fetus is still sufficient to protect the fetus. In this study, the investigators further demonstrated that the P-glycoprotein inhibitors valspodar and FG-120918 were able to completely block the placental Pglycoprotein function. The fetal drug concentrations in the wild-type fetus were increased and were comparable to those in the mdr1a/1b (/) fetus after oral administration of the P-glycoprotein inhibitors to heterozygous mothers. As in the brain, these results clearly demonstrate that placenta is very sensitive to changes in P-glycoprotein function. Because of potential function blockade, it is recommended that P-glycoprotein inhibitors should not be used in women during pregnancy to avoid excessive fetal exposure to xenobiotics. P-glycoprotein is also highly expressed in human placenta. Using monoclonal antibody C219, immunostaining showed a high expression of Pglycoprotein in trophoblasts of human placenta.[134] Functional P-glycoprotein activity has been demonstrated in cultured human placenta choriocarcinoma epithelial cells (BeWo cells).[135] Western blotting studies with monoclonal antibody C219 or JSB-1 indicated that P-glycoprotein is highly expressed in BeWo cells. In the BeWo monolayer, the B-to-A transport of vinblastine, vincristine and digoxin was significantly greater than the A-to-B transport. Addition of cyclosporin resulted in an increase in the A-to-B transport of the drugs and a decrease in the B-to-A transport. These results suggest that the placental P-glycoprotein acts as an efflux transporter by removing xenobiotics from cells. Since P-glycoprotein is exClin Pharmacokinet 2003; 42 (1)

80

Lin & Yamazaki

pressed in human placental trophoblasts, it is likely that placental P-glycoprotein also protects fetuses from xenobiotics in humans as well. Genetic variation in the expression level of placental P-glycoprotein was studied by Western blotting in 100 placentas obtained from Japanese women.[61] In this study, nine SNPs were identified with an allelic frequency of 0.0050.42 (table I). Of these SNPs, G2677A (allelic frequency 0.18) and G2677T (0.39) in exon 21 were associated with an amino acid conversion from Ala to Thr and to Ser, respectively. Comparison of the MDR1genotyping and corresponding placental P-glycoprotein level revealed a correlation between the Pglycoprotein expression level and SNPs in exon 1b (T129C) and exon 21 (G2677A/G2677T). Individuals with heterozygous T129C (T/C) had significantly lower levels of P-glycoprotein than the wild-type (T/T) individuals (1.07 vs 1.99, arbitrary units). The expression levels of placental P-glycoprotein in homozygotes for wild-type allele, heterozygotes and homozygotes of mutant allele in exon 21 were 2.44, 1.97 and 1.45 (arbitrary units), respectively. Although the clinical implications of interindividual variability in placental P-glycoprotein remains to be investigated, high placental levels of P-glycoprotein may provide a better protection for the fetus against xenobiotics.
4.3 Drug Metabolism

It has been widely accepted that the liver is the major site of drug metabolism because of its size and high content of drug-metabolising enzymes. In addition to the liver, the small intestine and kidney may contribute significantly to overall metabolism in the body.[136] In humans, CYP3A4 is the principal enzyme involved in the hepatic and intestinal metabolism of drugs. Studies in rats by Debri et al.[137] suggest that the CYP3A enzymes in the liver are evenly distributed. The data of Watkins et al.,[138] who investigated the CYP3A content at ten different locations in a human liver, also indicate that hepatic CYP3A is homogeneously distributed. Unlike the liver, the distribution of CYP3A4 is not uniform along the length of the small intestine, or
Adis International Limited. All rights reserved.

along the villi within a cross-section of mucosa. Using a monoclonal antibody to CYP3A, the columnar absorptive epithelial cells of the villi exhibited the strongest immunoreactivity, whereas no immunostaining was detectable in the goblet and crypt cells.[139] It has also been shown that CYP3A4 expression varied along the length of small intestine: median values of 31, 23 and 17 pmol/mg microsomal protein were found in human duodenum, distal jejunum, and distal ileum.[140] In contrast to CYP3A4, the expression of P-glycoprotein appears to increase progressively along the length of intestine.[78] There is a striking overlap between CYP3A4 substrates and P-glycoprotein substrates. Because of overlapping substrate specificity, and because of coexpression of CYP3A enzymes and P-glycoprotein in the intestine, kidney and liver, it is conceivable that P-glycoprotein may play an important role in drug metabolism. However, the magnitude of the effect of P-glycoprotein on metabolism appears to be dependent on the spatial relationship between P-glycoprotein and CYP3A enzymes. In the liver and kidney, P-glycoprotein is localised on the luminal membrane of hepatic canaliculi facing the bile duct lumen or on the luminal brush-border membrane of renal proximal tubular cells facing the renal tubule lumen. This means that P-glycoprotein is localised at the exit site of hepatocytes and renal epithelial cells. Therefore, P-glycoprotein only sees drug molecules after cellular uptake, intracellular distribution and metabolism in both the liver and kidney. In contrast to the situation in the liver and kidney, P-glycoprotein is localised at the entrance site of epithelial cells of intestines. Drug molecules are exposed to P-glycoprotein prior to intracellular distribution and metabolism. A large fraction of drug molecules is extruded by intestinal P-glycoprotein from the inside of the epithelial cells back into the intestinal lumen after the drug molecules gain access across the luminal surface of the epithelial cells; however, a portion of the extruded drugs then can be reabsorbed into the epithelial cells. Through the repetitive processes of extrusion
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

81

and reabsorption, P-glycoprotein prolongs the intracellular residence time of drug molecules and increases the probability of exposure to drug-metabolising enzymes. Consequently, P-glycoprotein may enhance intestinal metabolism of drugs, whereas it has less of an effect on drug metabolism in liver and kidney. The effect of P-glycoprotein on CYP3A4mediated intestinal metabolism of indinavir, a substrate for both CYP3A4 and P-glycoprotein, has been carefully evaluated in our laboratory using calcitriol-treated Caco-2 cells expressing both CYP3A4 and P-glycoprotein.[97,141] The formation of the major metabolite (M6), expressed as the ratio of the amount of the metabolite formed to the amount of the parent drug transported across the monolayer, was more than 6-fold greater when the drug was applied at the apical side than when the drug was applied at the basolateral side. Similarly, the metabolism of cyclosporin in Caco-2 cells was higher from the apical side than from the basolateral side.[142] These results strongly suggest a role of P-glycoprotein in enhancement of CYP3A4mediated intestinal metabolism of drugs. The effect of P-glycoprotein on intestinal metabolism was also shown in vivo. The intestinal firstpass metabolism of indinavir increased from 6% in control rats compared with 34% in dexamethasone-treated rats.[143] Pretreatment of rats with dexamethasone (40 mg/kg orally for 3 days) resulted in a 2.5-fold increase in both CYP3A and P-glycoprotein levels in the intestine. The 6-fold increase in intestinal first-pass metabolism of indinavir cannot be explained by the 2.5-fold increase in intestinal CYP3A level alone. The increased intestinal first-pass metabolism is most probably due to a combination of increased intestinal CYP3A and P-glycoprotein levels, providing in vivo evidence that P-glycoprotein enhances the intestinal firstpass metabolism of indinavir. In this study, the effect of P-glycoprotein on hepatic metabolism was also investigated. Pretreatment of rats with dexamethasone also induced hepatic first-pass metabolism of indinavir, and the increased hepatic CYP3A enzyme activity alone
Adis International Limited. All rights reserved.

appeared to be able to explain the increased hepatic first-pass metabolism, suggesting that P-glycoprotein plays a less significant role in hepatic metabolism as compared with intestinal metabolism.[143] These results clearly suggest that P-glycoprotein may play an important role in intestinal metabolism of drugs. However, it should be reemphasised that the effect of P-glycoprotein on drug metabolism becomes quantitatively less significant when high doses are given.[144]
4.4 Drug Excretion

Drugs are generally eliminated from the body by metabolism and/or excretion. Both the liver and kidney play an important role in the excretion of unchanged drugs and their metabolites. In principle, biliary excretion and renal tubular excretion share certain characteristics. For biliary excretion, a drug must first traverse the sinusoidal (basolateral) membrane of the hepatocytes by passive diffusion and/or hepatic uptake transporters. The sinusoidal membrane of the hepatocyte contains a number of active transporters responsible for the uptake of cations, anions and endogenous substances into hepatocytes from the circulation.[145] Once in the hepatocytes, the drug molecules continue to diffuse and reach the canalicular membrane, where P-glycoprotein and other efflux transporter systems will pump the drug molecules into bile. Often, biotransformation occurs when the drug molecules are passing through the hepatocytes. Therefore, hepatic uptake, intracellular diffusion and metabolism, as well as other efflux transporter systems, have to be taken into consideration when biliary excretion of drugs is evaluated. Similarly, uptake of drugs across the basolateral membrane of renal epithelial cells is the first step in renal excretion, and biotransformation may occur. The basolateral membrane contains a number of active transporters responsible for drug uptake. The luminal brush-border membrane also contains numerous active transporters, including P-glycoprotein, which is responsible for the last step of
Clin Pharmacokinet 2003; 42 (1)

82

Lin & Yamazaki

excretion of P-glycoprotein substrates into the urine.[146]


4.4.1 Biliary Excretion

The involvement of P-glycoprotein in the biliary excretion of drugs was first suggested by immunohistochemical studies showing that P-glycoprotein is highly expressed on the canalicular membrane of hepatocytes.[6] Earlier experimental evidence of the potential involvement of P-glycoprotein in biliary excretion had come from in vitro studies with highly purified canalicular membrane vesicles and isolated perfused rat liver. Kamimoto et al.[127] demonstrated ATP-dependent transport of daunomycin in canalicular membrane vesicles. Furthermore, they showed that the transport of daunomycin was inhibited by verapamil, a potent P-glycoprotein inhibitor. These results suggest the involvement of P-glycoprotein in the biliary excretion of daunomycin. The involvement of P-glycoprotein in the biliary excretion of vincristine has also been demonstrated by using isolated perfused rat liver.[147] Vincristine is eliminated in rats mainly by biliary excretion; approximately 50% of the dose is excreted as unchanged drug into the bile. The biliary excretion of vincristine has been shown to be saturable and inhibited by verapamil. Similarly, studies in isolated perfused rat liver also suggest that P-glycoprotein plays a significant role in the biliary excretion of doxorubicin.[148] More direct evidence for the involvement of Pglycoprotein in biliary excretion of drugs has come from studies with mdr1a knockout mice. Digoxin is mainly excreted as unchanged drug in the bile and urine of mice; only a minor fraction of digoxin (<3%) is metabolised. The biliary clearance of digoxin is substantially greater in mdr1a (+/+) mice (2.3 ml/min/kg) than in mdr1a (/) mice (0.84 ml/min/kg).[149] Approximately 45% of the dose is excreted as unchanged digoxin in the bile of mdr1a (+/+) mice. These results clearly indicate that Pglycoprotein plays a significant role in the biliary excretion of digoxin in mice. Similarly, the role of P-glycoprotein in biliary excretion has also been reported for doxorubicin and vinblastine.[150] The excretion of unchanged doxorubicin and vinblas Adis International Limited. All rights reserved.

tine into bile is 3- to 5-fold greater in mdr1a (+/+) mice than in mdr1a (/) mice. However, the absolute biliary recovery of unchanged drugs in mdr1a (+/+) mice is only 5 and 13% of the dose for vinblastine and doxorubicin, respectively. The low biliary excretion of doxorubicin and vinblastine in mice is partly attributed to their high hepatic metabolism. From these two studies, it is important to note that not all P-glycoprotein substrates are subject to significant biliary excretion. Because both mdr1a and mdr1b genes are expressed on the canalicular membrane, biliary excretion of drugs was further investigated in a double knockout mouse model in which both mdr1a and mdr1b genes were disrupted. It was found that the complete absence of both mdr1a and mdr1b P-glycoprotein at the canalicular membrane results in a greater decrease in biliary excretion of a number of basic drugs compared with mdr1a (/) single knockout and wild-type mice.[151,152] Vecuronium, a neuromuscular blocking agent, is eliminated predominantly by biliary excretion in mice, and the biliary excretion of vecuronium is profoundly reduced in mdr1a/1b(/) double knockout mice compared with mdr1a (/) single knockout and wild-type mice. The biliary clearance of vecuronium in mdr1a/1b (/) double knockout mice was found to be about six times lower than in wild-type mice, whereas the biliary clearance in mdr1a (/) mice was only 2.5 times smaller than in the wildtype mice.[149,150] Interestingly, appreciable residual biliary excretion of vecuronium was still observed in mdr1a/1b (/) double knockout mice, suggesting that additional carrier systems are involved in the biliary excretion of cationic drugs. Collectively, the results from both in vitro and in vivo studies clearly demonstrate that P-glycoprotein plays a significant role in biliary excretion of P-glycoprotein substrates. However, additional factors, such as hepatic uptake, intracellular distribution and metabolism have to be taken into consideration when quantitatively assessing the role of P-glycoprotein in biliary excretion.
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

83

4.4.2 Renal Excretion

Renal excretion of drugs usually involves three processes: glomerular filtration, renal tubular secretion, and reabsorption from the renal tubular lumen. The relationship between renal clearance (CLR) and these processes can be expressed as equation 1:
CLR = fu GFR + CLS CLRA

Rearrangement of equation 1 yields equation 2:


CLR/(fu GFR) = 1 + (CLS CLRA)/(fu GFR)

where GFR, CLS, and CLRA are glomerular filtration rate, secretion clearance and reabsorption clearance, respectively, and fu is the unbound fraction of drug in plasma. GFR is a passive process by which only unbound drugs can be filtered, while tubular secretion and reabsorption often involve active transporters. Non-filtered drugs must first cross the basolateral membrane and then the apical membrane of epithelial cells of the renal tubule, either by passive diffusion or carrier-mediated processes. Immunohistochemical studies[6] reveal that Pglycoprotein is localised at the apical brush-border membrane of the proximal renal tubule, which is the major site of renal secretion. The finding of the localisation of renal P-glycoprotein has led to recognition of the importance of this transporter in tubular secretion of drugs. In vitro systems have proven to be very useful tools for the study of the P-glycoprotein role in tubular secretion. Human MDR1 gene-transfected Madin-Darby canine kidney (MDCK) and porcine kidney (LLC-PK1) epithelial cell lines are the two most widely used models for the study of renal P-glycoprotein function. The B-to-A transepithelial transport of digoxin across LLC-PK1 monolayers expressing human Pglycoprotein is much greater than the A-to-B transport by a factor of 7. Addition of cyclosporin results in a marked decrease in the B-to-A transport and an increase in the A-to-B transport.[153] Similarly, LLC-PK1 cells expressing human P-glycoprotein exhibit greater B-to-A transport of vinblastine than A-to-B transport.[154] Transepithelial transport of vinblastine has also been observed in
Adis International Limited. All rights reserved.

MDCK monolayers expressing human P-glycoprotein.[155] The B-to-A transport of vinblastine has been observed to be about six times higher than A-to-B transport. These results suggest that the Pglycoprotein functions as an efflux transporter at the apical membrane of epithelial cells of the renal tubule. The isolated perfused kidney technique has also been used to investigate the role of P-glycoprotein in tubular secretion. As shown in equation 2, there are three processes involved in renal excretion. Although quantification of each process of renal excretion is difficult due to practical limitations, the CLR/(fu GFR) ratio can be used as a simple way to assess the relative contribution of each process to overall renal excretion. When the CLR/(fu GFR) ratio of a drug is greater than unity, this means that tubular secretion of the drug occurs. Conversely, when the ratio is less than unity, reabsorption of the drug from the tubular lumen occurs. Using the CLR/(fu GFR) ratio approach, Hori et al.[156] showed that digoxin was actively secreted in the isolated perfused rat kidney with a CLR/(fu GFR) ratio of 2.5. The P-glycoprotein inhibitors quinidine and verapamil inhibited tubular secretion and decreased the ratio. In the presence of quinidine and verapamil at a concentration of 8 mol/L, the CLR/(fu GFR) ratio became unity, suggesting minimal reabsorption process of digoxin. Similar results were also observed for digoxin when the drug was studied in dog isolated perfused kidney using the single-pass multiple indicator dilution method.[157] These results provide further evidence that digoxin is actively secreted into the renal tubular lumen by P-glycoprotein. Although transgenic mice have also been used to study the role of P-glycoprotein in renal excretion, the results from these mouse studies have remained controversial. Renal clearance of digoxin was compared in mdr1a (+/+) and mdr1a (/) mice following intravenous administration. As expected, the renal clearance of digoxin in mdr1a (+/+) mice was three times greater than that in mdr1a (/) mice, at 3.15 and 0.99 ml/min/kg, respectively.[149] Because doxorubicin is a P-glycoClin Pharmacokinet 2003; 42 (1)

84

Lin & Yamazaki

protein substrate, it is expected that the absence of P-glycoprotein in renal tubule will result in a decrease in renal excretion of doxorubicin. However, in contrast with the expectation, the renal excretion of unchanged doxorubicin was found to be higher in mdr1a (/) mice (15% of the dose) than in mdr1a(+/+) mice (10%) after intravenous administration.[158] Likewise, the renal excretion of three basic compounds (tributylmethylammonium, azidoprocainamide and vecuronium) in mdr1a/1b (/) double knockout mice was significantly higher than that in mdr1a/1b (+/+) mice.[152] Since these three basic compounds are P-glycoprotein substrates, lower renal excretion in mdr1a/1b (+/+) mice was not expected. The reason for the conflicting results is presently not known. One possible explanation is that other transporter systems may be involved in the renal excretion of doxorubicin and these basic compounds, and disruption of mdr1a and mdr1b genes may increase the expression of other transporter systems. 5. P-Glycoprotein-Mediated Drug-Drug Interactions Inhibition and induction of CYP enzymes, particularly CYP3A4, are probably the most common causes for documented drug interactions.[159,160] Several prominent drugs have been withdrawn from the market because of serious adverse effects as a result of CYP-mediated interactions.[161,162] Therefore, CYP-mediated drug interactions have always been a major concern for clinicians and patients. Like CYP-mediated drug interactions, Pglycoprotein-mediated drug interactions may be anticipated when P-glycoprotein substrates and P-glycoprotein inhibitors (or inducers) are coadministered. Inhibition and induction of P-glycoprotein in animals and humans have been reported, although less frequently than for CYP enzymes, and their pharmacokinetic consequences are similar to those observed for inhibition and induction of CYP enzymes.
Adis International Limited. All rights reserved.

5.1 P-Glycoprotein Inhibition Does Not Follow Simple Kinetics

As discussed in section 1, P-glycoprotein has more than one drug-binding site, even though the exact number of binding sites is not yet known. In addition, two ATP-binding domains are also involved in the P-glycoprotein function of drug transport. All of the drug-binding sites and ATPbinding domains interact cooperatively as a functional unit. Thus, inhibition of P-glycoprotein transport of a drug by other drugs could potentially result from either competition for drug-binding sites or from blockage of the ATP hydrolysis process. For example, verapamil inhibits the transport function in a competitive manner without interrupting the cyclic activity (ATP hydrolysis) of P-glycoprotein, vanadate interacts with the ATPbinding domains of P-glycoprotein without interacting with the substrate-binding sites, and cyclosporin inhibits transport function by interfering with both substrate recognition and ATP hydrolysis.[163-166] These results clearly indicate that multiple mechanisms are responsible for P-glycoprotein inhibition. Because of the complexity, it is difficult to assess the mechanism and type of P-glycoprotein inhibition when P-glycoprotein substrate drugs and P-glycoprotein inhibitors are given simultaneously. As shown in the following examples, the pattern of P-glycoprotein inhibition appears to be substrate-dependent. For instance, the P-glycoproteinmediated transepithelial transport of digoxin in LLC-PK1 cells was inhibited by cyclosporin, but digoxin surprisingly did not inhibit the P-glycoprotein-mediated transepithelial transport of cyclosporin.[153] Furthermore, in vitro studies with multidrug-resistant P388 leukaemia cells reveal that verapamil competitively inhibits P-glycoproteinmediated daunomycin uptake, whereas vinblastine shows noncompetitive inhibition with daunomycin.[167] On the other hand, P-glycoprotein-mediated transepithelial transport of vinblastine in Caco-2 cells was inhibited by verapamil in a competitive manner, and by dideoxyforskolin in a noncompetitive manner.[83] Similarly, the P-glycoproClin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

85

tein-mediated uptake of azidopine into plasma membrane vesicles obtained from P-glycoproteinexpressing multidrug-resistant cells was noncompetitively inhibited by vinblastine or by cyclosporin, whereas vinblastine uptake into plasma membrane vesicles was competitively inhibited by cyclosporin.[164,168] P-glycoprotein-mediated interactions have also been investigated by measuring the displacement of reversible binding to P-glycoprotein using membrane vesicles in the absence of ATP. The displacement of morphine binding by verapamil was only partial, although vinblastine binding was completely displaced by verapamil.[169] The ATPase activity assay has also been used as a convenient tool for determining the type of interaction between P-glycoprotein substrates. A systematic analysis of inhibition patterns of verapamilinduced P-glycoprotein ATPase activity revealed that noncompetitive inhibition of verapamilstimulated ATPase activity was found with vanadate, whereas competitive inhibition was found with cyclosporin.[170] Similarly, in another ATPase activity study, nicardipine displayed competitive interaction with vinblastine but noncompetitive interaction with verapamil.[171] Although the competition of two substrates for the same P-glycoprotein usually results in an inhibitory effect on the P-glycoprotein-mediated transport of the substrates, activation of P-glycoprotein-mediated efflux transport has been reported in some cases. The P-glycoprotein-mediated doxorubicin efflux out of multidrug-resistant HCT-15 colon cells was significantly increased by some flavonoids.[172] Similarly, rhodamine 123 and Hoechst 33 342 stimulated the rate of P-glycoprotein-mediated transport of each other in P-glycoprotein-enriched plasma membrane vesicles isolated from Chinese hamster ovary CHrB30 cells.[173] Furthermore, the transport of Hoechst 33 342 was stimulated by daunorubicin and doxorubicin, but the transport of rhodamine 123 was inhibited by daunorubicin and doxorubicin.[173] Stimulation was also observed in ATPase activity studies. For example, verapamil-induced ATPase activity was
Adis International Limited. All rights reserved.

stimulated by progesterone, diltiazem, amitriptyline and propranolol.[170] Collectively, these results suggest that two substrates are able to bind simultaneously to P-glycoprotein at different sites that may interact allosterically. It is of interest to note that activation is not a phenomenon limited to P-glycoprotein interaction; activation of CYP3A4 has also been reported.[174] Similarly, the concept of multiple binding sites has also been proposed to explain many unusual enzyme kinetics observed for CYP3A4.[174] The interaction between substrates and inhibitors of CYP3A4 does not always follow simple enzyme kinetics. Wang et al.[175] have demonstrated that in vitro interaction patterns of CYP 3A4 substrates are substrate-dependent. Mutual inhibition, partial inhibition and activation have been observed in the interactions of testosterone with terfenadine, testosterone with midazolam and terfenadine with midazolam, respectively. To test the hypothesis of multiple binding sites, Shou et al.[176] have successfully described the enzyme kinetics of CYP3A4-mediated metabolism of diazepam and its derivatives with a kinetic model which consists of two substrate-binding sites (apoprotein) and one catalytic site (prosthetic haem). In summary, the interaction between P-glycoprotein substrates does not always follow simple kinetics. The pattern of P-glycoprotein interaction can be classified into at least three major categories: competitive inhibition, noncompetitive inhibition and co-operative stimulation. Competitive inhibition suggests that two substrates act on the same sites of P-glycoprotein and that only one or the other can bind at any one time. Noncompetitive inhibition means that two substrates are able to bind simultaneously to P-glycoprotein molecule at distinct sites that are functionally independent. The situation can be even more complicated if allosteric effects are involved in interaction between substrate and inhibitor. The complexity of the molecular mechanism for P-glycoprotein inhibition prevents our ability to predict the potential of Pglycoprotein-mediated drug-drug interactions, either quantitatively or qualitatively.
Clin Pharmacokinet 2003; 42 (1)

86

Lin & Yamazaki

5.2 Drug Interactions Caused by P-Glycoprotein Inhibition

Because of overlapping substrate specificities and inhibitors between CYP3A4 and P-glycoprotein, many drug interactions may involve both CYP 3A4 and P-glycoprotein. Therefore, it is important to distinguish CYP3A4-mediated inhibition from P-glycoprotein-mediated inhibition in order to make appropriate interpretation of drug interaction data. For this reason, Wandel et al.[177] have proposed to use the ratio of IC50 for CYP3A4 to IC50 for P-glycoprotein as an index of the relative selectivity of a drug for P-glycoprotein-mediated inhibition versus CYP3A4-mediated inhibition. If the ratio is much greater than unity, it means that the relative contribution by P-glycoproteinmediated inhibition is quantitatively more significant. Relative selectivity is best exemplified by the following study by Choo et al.[178] The IC50 values for ketoconazole to inhibit digoxin transport and nifedipine metabolism were 1.2 and 0.15 mol/L, respectively, and the corresponding values for LY-335979 were 0.024 and 5 mol/L.[178] In mdr1a (+/+) mice, pretreatment with LY335979 (25 mg/kg intravenously) resulted in a 15fold increase in brain concentrations of nelfinavir, but had little effect on plasma concentrations. On the other hand, coadministration of ketoconazole (50 mg/kg intravenously) caused an 8.5-fold increase in brain concentrations of nelfinavir and a 3.5-fold increase in plasma concentrations. From the ratio of IC50 values (208 for LY-335979 and 0.13 for ketoconazole), it is clear that the increased brain concentrations of nelfinavir by LY-335979 are caused mainly by P-glycoprotein inhibition. In contrast, the inhibitory effect of ketoconazole on the brain and plasma concentrations of nelfinavir is attributed mainly to CYP3A-mediated inhibition, and to a lesser extent to P-glycoprotein inhibition. As shown in the above study, it is clear that coadministration of a P-glycoprotein inhibitor causes a much greater increase in drug concentration in brain than in plasma. Extensive inhibition
Adis International Limited. All rights reserved.

of BBB P-glycoprotein function was also observed when valspodar or GF-120918 (potent P-glycoprotein inhibitors) were coadministered with other drugs in mice. Pretreatment with GF-120918 (250 mg/kg/day for 4 days, orally) led to a 13-fold increase in brain concentrations, but only a modest increase (2-fold) in plasma concentrations, of amprenavir in mdr1a (+/+) mice.[179] In another study, a 10-fold increase in brain concentrations of digoxin was also observed when valspodar 50 mg/kg was given orally to mdr1a (+/+) mice.[180] Similarly, pretreatment with intravenous valspodar 25 mg/kg resulted in an 80-fold increase in brain concentrations of nelfinavir in mdr1a (+/+) mice.[178] These experimental findings present a major problem that may confound attempts to use P-glycoprotein modulators in the clinical setting. Although the use of effective P-glycoprotein modulators (P-glycoprotein inhibitors), such as LY335979, valspodar and FG-120918, may improve the treatment of cancers, these P-glycoprotein modulators also inhibit P-glycoprotein function in normal cells, resulting in increased toxicity. This is particularly true for the brain; the profound increase of drug concentration in the brain by these P-glycoprotein modulators increases the risk of neurotoxicity. The increased neurotoxicity caused by P-glycoprotein modulators is best exemplified by a clinical study with loperamide, an antidiarrhoeal agent.[181] Loperamide 16mg was administered to healthy male volunteers with or without coadministration of quinidine 600mg. Loperamide produced no respiratory depression when administered alone, but respiratory depression occurred when loperamide was given with quinidine. Since loperamide is a P-glycoprotein substrate, under normal conditions the brain penetration of this drug is limited as a result of P-glycoprotein extrusion. However, in the presence of quinidine (a potent P-glycoprotein inhibitor), the deliver of loperamide to the brain increases, resulting in serious neurotoxicity. Inhibition of hepatic and intestinal P-glycoprotein has also been reported. The inhibitory effect
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

87

of valspodar on hepatic and intestinal P-glycoprotein was compared in wild-type mice and mdr1a/1b (/) double knockout mice, using digoxin as a model compound.[180] As expected, oral pretreatment with valspodar 50 mg/kg led to a marked decrease in both intestinal and biliary excretion of digoxin in wild-type mice. However, appreciable residual biliary excretion of digoxin (14% of an intravenous dose) was still observed in mdr1a/1b (/) double knockout mice, and pretreatment with valspodar caused an unexpected decrease in biliary excretion of digoxin (<5%) in these animals. These results suggested that other unidentified transport systems, rather than P-glycoprotein, may also be involved in the biliary excretion of digoxin in mice, and that these unidentified transporters can be inhibited by valspodar. An important lesson to be learned from these studies is that the underlying mechanisms for drug interactions can be very complicated, and care should be exercised to make appropriate interpretations. P-glycoprotein inhibition as a cause of drug interaction has also been reported in humans. Perhaps the most compelling clinical evidence of P-glycoprotein-mediated drug interactions in humans is the interaction of digoxin with other cardiac drugs, such as verapamil and quinidine.[182-185] A daily dosage of verapamil 160mg caused a 40% increase in digoxin plasma concentration, whereas a daily dosage of 240mg caused a 6080% increase, suggesting dose-dependent P-glycoprotein inhibition. Because digoxin is exclusively eliminated in humans by renal excretion as unchanged drug,[186] it is highly likely that the observed drug interaction between digoxin and verapamil (or quinidine) is due to inhibition of P-glycoprotein activity, resulting in increased absorption and decreased elimination of digoxin. Consistent with the clinical data, increased absorption of digoxin by verapamil and quinidine has been shown in rats,[187,188] and decreased renal excretion of digoxin by verapamil and quinidine has been demonstrated in in vitro studies using isolated perfused rat and dog kidney.[156,157]
Adis International Limited. All rights reserved.

An important issue related to P-glycoprotein inhibition is the concept of P-glycoprotein modulation in cancer research. Resistance of tumour cells to chemotherapeutic agents is a major problem in the treatment of human cancers. Tumour cells that exhibit multidrug resistance are often associated with overexpression of P-glycoprotein. In 1981, Tsuruo et al.[189] reported that verapamil, a potent P-glycoprotein inhibitor, is able to restore the in vitro sensitivity to vincristine in multidrug-resistant cell lines by inhibiting P-glycoprotein-mediated drug transport. This finding implies that clinical drug resistance can be circumvented through the concomitant administration of P-glycoprotein inhibitors and anticancer drugs. The idea of reversing P-glycoprotein-mediated drug resistance has led to intensive efforts to develop potent and specific P-glycoprotein modulators, such as valspodar, FG-120918 and LY-335979. As expected, Pglycoprotein-mediated interactions were observed when anticancer drugs were given with these potent P-glycoprotein modulators in cancer patients. For example, both valspodar and FG-120918 caused an increase in systemic exposure to doxorubicin associated with a decrease in doxorubicin clearance in patients.[190,191] However, the pharmacodynamic results from clinical modulation studies are less than encouraging.[192] One of the main reasons for the clinical disappointment is that the coadministration of P-glycoprotein modulators with anticancer drugs fails to improve the toxicity profiles of the chemotherapeutic agents. Although the P-glycoprotein modulators might completely inhibit P-glycoprotein function in tumour cells and restore drug sensitivity, the modulators could also inhibit the P-glycoprotein protective function of normal cells, leading to cytotoxicity. As shown in the mouse studies discussed above, treatment with P-glycoprotein modulators resulted in a much greater increase in drug concentration in tissues than in plasma. Therefore, the lack of improvement of toxicity profiles might result from the increased tissue distribution of drugs caused by P-glycoprotein modulators. The digoxin intoxication by quinidine and verapamil
Clin Pharmacokinet 2003; 42 (1)

88

Lin & Yamazaki

can also be explained by the increased drug concentration in tissues. During combination therapy with digoxin and verapamil (or quinidine), both verapamil and quinidine enhanced digoxin toxicity even when digoxin plasma concentrations were in the therapeutic range observed in the absence of the P-glycoprotein inhibitors.[184,193,194] Unless a specific P-glycoprotein modulator that selectively inhibits tumour cell P-glycoprotein can be developed, manipulation of the balance between tumour cell kill and toxicity to normal tissue will be very difficult to achieve.
5.3 P-Glycoprotein Induction is a Complex Process

Like some of the CYP isoenzymes, the expression of P-glycoprotein is inducible. Because CYP enzymes and P-glycoprotein function together to protect the body from accumulation of toxins and xenobiotics, it is not surprising, from an evolutionary point of view, that both defensive systems possess adaptive manoeuvres to reinforce their protective role. Regulation of P-glycoprotein expression in response to inducers has been extensively studied in vitro using cell lines derived from animals and humans. Fardel et al.[195] reported that 3-methylcholanthrene strongly induced functional P-glycoprotein levels in a dose-dependent manner through an increased expression of the mdr gene in rat liver epithelial cells. In another study, mdr RNA levels were found to increase substantially in rat and mouse cell lines following acute exposure to cytotoxic drugs, such as doxorubicin, daunomycin and mitoxantrone.[196] Interestingly, although these cytotoxic drugs induced mdr RNA in all rat and mouse cell lines, they had no effects in human cell lines. These results suggest species differences in inductive response to P-glycoprotein inducers. Species differences are well known for the induction of CYP enzymes,[159] and recently it has been demonstrated that the species differences in CYP3A induction are due mainly to sequence differences in the ligand-binding domain of pregnane X receptor (PXR).[197] As will be discussed below,
Adis International Limited. All rights reserved.

P-glycoprotein is also likely to be regulated by PXR, and therefore it is possible that the observed species differences in CYP3A induction and P-glycoprotein induction have a similar molecular basis. Dexamethasone has been shown to induce the mdr1a and mdr1b P-glycoproteins in a mouse cell line.[198] Similarly, the inductive effect of dexamethasone on P-glycoprotein expression in cultured rat hepatocytes has been reported.[199] Consistent with in vitro observations, pretreatment of rats with dexamethasone (40 mg/kg/day orally for 3 days) resulted in significant increases in both intestinal and hepatic P-glycoprotein expression by approximately 2- to 3-fold.[143] In another study, a 5-fold increase of rat hepatic P-glycoprotein level was reported after intraperitoneal administration of dexamethasone at 100 mg/kg/day for 4 days.[200] These results suggest that the induction of P-glycoprotein expression is a dose-dependent process. Similarly, dose-dependent induction of P-glycoprotein expression was observed in rats receiving cyclosporin orally at 2, 10 or 30 mg/kg/day, or subcutaneously at 1, 5 or 15 mg/kg/day, for 28 days.[201] Western blot analysis showed that cyclosporin induced renal P-glycoprotein in a dosedependent manner. In another study, the induction of P-glycoprotein expression by cyclosporin was demonstrated to be not only dose-dependent, but also time-dependent. When cyclosporin was given subcutaneously at 10 mg/kg/day for 5, 10 and 15 days, the increase of the P-glycoprotein levels in liver, intestine, kidney and lungs was maximal after 10 days of treatment in most tissues of rats.[202] Interestingly, the inductive response to P-glycoprotein inducers also appears to be tissuedependent. After administration of cyclosporin 10 mg/kg/day for 15 days, an increase in the expression of P-glycoprotein was detected in many, but not all, tissues of rats.[202] In kidneys, intestine, stomach, liver and lungs, the amount of P-glycoprotein detected was increased by 256, 239, 161, 144 and 69% compared with control groups, respectively. In all of these tissues, the increase was maximal after 10 days of treatment. In heart, testis, and spleen, the increase in P-glycoprotein expresClin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

89

sion was 82, 74 and 50% compared with control groups, respectively, and the induction was maximal after 5 days of treatment in heart and after 15 days of treatment in testis and spleen. However, no change in the level of P-glycoprotein expression was found in the brain during treatment. The lack of P-glycoprotein induction in the brain could be due to the high basal level of P-glycoprotein in the brain capillaries. The high level of P-glycoprotein in this tissue could cause a rapid extrusion of cyclosporin from the capillary endothelial cells, leaving no opportunity for this compound to induce P-glycoprotein. Similarly, tissue differences in P-glycoprotein induction were also observed after administration of cycloheximide to rats.[203] Cycloheximide caused an 8-fold increase in mRNA of mdr1a P-glycoprotein in lung, but only a 1.5fold increase was observed for small intestine. Collectively, these results clearly point to a tissuedependent inductive response to P-glycoprotein inducers. However, the mechanisms underlying tissue differences in the inductive response require further investigation. Recently, Durr et al.[204] have reported that St Johns Wort induces P-glycoprotein and CYP3A enzymes in rats and humans. Administration of St Johns Wort extract to rats at an oral dose of 1000 mg/kg/day for 14 days resulted in a 3.8-fold increase of intestinal P-glycoprotein and a 2.5-fold increase of hepatic CYP3A2 expression level. Oral administration of St Johns Wort extract to eight healthy male volunteers for 14 days at a dose of 300mg three times daily resulted in a 1.4- and 1.5fold increase of duodenal P-glycoprotein and CYP3A4, respectively. These results suggest the coordinated regulation of P-glycoprotein and CYP3A enzymes in both rats and humans. The inducing effect of St Johns Wort on intestinal P-glycoprotein was considerably more pronounced in rats than in humans. These differences most probably reflect a dose-dependent effect, because the rats received a dose of St Johns Wort that was approximately 50 times higher per kilogram bodyweight than that received by the volunteers. An alternative explanation could be that species dif Adis International Limited. All rights reserved.

ferences exist in the susceptibility to P-glycoprotein induction. Similarly, co-induction of P-glycoprotein and CYP3A enzymes by dexamethasone was also observed in rats.[143] Consistent with in vivo observations, coordinated regulation between CYP3A and P-glycoprotein has been reported in vitro in animal and human cell lines. In a cell line derived from human colon adenocarcinoma LS 180/WT and its doxorubicinresistant subline (LS 180/AD 50), both P-glycoprotein and CYP3A4 were induced after treatment with many known inducers, including phenobarbital, rifampicin (rifampin), clotrimazole and reserpine.[205] Similarly, dexamethasone, a potent CYP3A4 inducer, has been shown to induce Pglycoprotein in human hepatoma.[199,206] Collectively, these in vitro and in vivo observations have led to the speculation that regulation of CYP3A4 and MDR1 gene expression is coordinated through a similar mechanism. The possible coordinate regulation of P-glycoprotein and CYP3A4 gene expression was further hypothesised by Wacher et al.[207] based on the proximity of the chromosomal loci of these two genes. The human MDR gene has been mapped to chromosome locus 7q21.1,[208] whereas the CYP3A4 gene is located at 7q22.1.[209] Recent in vitro studies have demonstrated that the PXR (also known as SXR) plays a central role in regulating CYP3A4 transcription.[210] A diverse array of drugs, including rifampicin, phenobarbital, clotrimazole and hyperforin (one of the constituents of St Johns Wort) are known to activate the nuclear receptor PXR and induce CYP3A4 expression.[211-213] It has been speculated that induction of P-glycoprotein is also attributed to the same PXR activation. However, direct evidence was only made available recently, when it was shown that the PXR activates the expression of the MDR1 gene.[214,215] Using pharmacological and genetic approaches, Synold et al.[214] showed that the PXR is activated by paclitaxel and rifampicin, and this activation results in an increase in MDR1 mRNA and P-glycoprotein expression in primary human hepatocytes and LS 180 colon cancer cells. Paclitaxel and rifampicin also induce CYP3A4 and
Clin Pharmacokinet 2003; 42 (1)

90

Lin & Yamazaki

CYP2C8 mRNA and enzyme protein expressions in primary human hepatocytes through PXRmediated transcriptional effects. These findings strongly suggest that PXR plays an important role in regulating drug-metabolising enzymes and the P-glycoprotein transporter. The molecular mechanism of PXR-mediated Pglycoprotein induction by rifampicin has been studied in detail by Geick et al.[215] using the human colon carcinoma cell line LS174T. By using DNA binding assay and transfections, it was shown that the induction of MDR1 is mediated by a DR4 nuclear response element at about 8 kilobase pairs of the MDR1 upstream region to which PXR binds. The above two studies provide direct evidence that a similar mechanism is responsible for induction of CYP3A4 and MDR1 by xenobiotics. Interestingly, in addition to the liver and intestine, the nuclear receptor PXR and P-glycoprotein are co-expressed in a number of other tissues, including the kidney and placenta.[211,216] The question as to whether the tissue-dependent differences in P-glycoprotein induction are attributed to the tissue differences in the expression level of PXR, or to different mechanisms in different tissues, remains to be explored. Although it is evident that the nuclear receptor PXR plays a central role in P-glycoprotein induction, the inductive effects of a single drug may be mediated by multiple mechanisms. Recently, Lee[203] reported that the induction of rat mdr1a and mdr1b mRNA transcripts by cycloheximide is mediated by different mechanisms. The induction of mdr1a P-glycoprotein is transcriptionally regulated, whereas the induction of mdr1b P-glycoprotein is post-transcriptionally regulated. The post-transcription is probably due to mRNA stabilisation, although it is currently not known how the mRNA is stabilised. Clearly, the regulatory processes of P-glycoprotein induction are very complex and every inducer has its own pattern of induction. Therefore, a detailed knowledge of the inductive processes for each inducer is required in order to understand its implications and consequences.
Adis International Limited. All rights reserved.

5.4 Drug Interactions Caused by P-Glycoprotein Induction

The most compelling evidence to date for P-glycoprotein induction as a cause of drug interactions was provided by Greiner et al.[217] in a clinical study comparing the pharmacokinetics of digoxin before and during coadministration of rifampicin 600 mg/day for 10 days in eight healthy volunteers. The plasma Cmax and AUC of digoxin decreased from 5.4 g/L and 55 g h/L before rifampicin pretreatment to 2.6 g/L and 38 g h/L, respectively, during rifampicin pretreatment when the volunteers received a single oral dose of digoxin 1mg. However, pretreatment with rifampicin had little effect on the AUC and renal clearance of digoxin after intravenous administration. Duodenal biopsies were obtained from each volunteer before and after administration of rifampicin. Rifampicin treatment increased intestinal P-glycoprotein content 3.5-fold, which correlated inversely with the oral AUC of digoxin. Since digoxin is eliminated exclusively by renal excretion, and not by metabolism, and since digoxin is given orally at a very low dose, these results strongly suggest that the digoxin-rifampicin interaction mainly occurs at the level of the intestine through P-glycoproteinmediated induction. This means that the decreased plasma concentration of digoxin during rifampicin treatment is caused by reduced bioavailability of digoxin as a result of induction of intestinal Pglycoprotein. Similarly, administration of St Johns Wort extract (three doses of 300mg per day for 14 days) resulted in an 18% decrease of plasma AUC after a single digoxin dose of 0.5mg in healthy volunteers.[204] Treatment with the extract resulted in a 40% increase in the expression of duodenal P-glycoprotein. The decreased plasma AUC correlated reasonably well with the increased intestinal P-glycoprotein expression. Pretreatment with rifampicin also significantly decreased systemic exposure to fexofenadine in healthy volunteers.[218] Twenty volunteers received a 60mg oral dose of fexofenadine before and after treatment with oral rifampicin 600mg for 6 days. The Cmax and AUC of fexofenadine decreased by
Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

91

2- and 3-fold, respectively, in volunteers after rifampicin treatment. With the assumption that fexofenadine is predominantly eliminated by biliary excretion, but not by metabolism, the investigators concluded that the decreased plasma profiles are the result of a reduced bioavailability caused by induction of intestinal P-glycoprotein. However, the assumption of lack of metabolism may not be valid. From a human mass balance study, it was shown that approximately 80 and 11% of an oral dose of [14C]fexofenadine were recovered in the faeces and urine, respectively.[219] MDL-4,829, a metabolite of fexofenadine, accounted for about 20% of the radioactivity in urine. However, it is unknown whether the faecal component represents unabsorbed drug or the result of biliary excretion, and information on the metabolites in faeces is not available.[219] If based on the urinary recovery, metabolism of fexofenadine is not insignificant. Therefore, it is possible that the observed interaction between fexofenadine and rifampicin is due to a combination of both CYPmediated and P-glycoprotein-mediated induction. Furthermore, it has recently been demonstrated that organic anion transporting polypeptide (OATP) is involved in the hepatic uptake of fexofenadine.[220] It is possible that rifampicin is able to induce OATP. Thus, the involvement of OATP in the hepatobiliary excretion of fexofenadine may further complicate the interpretation of the observed interaction between fexofenadine and rifampicin. The observed interaction between cyclosporin and rifampicin in healthy volunteers is also probably due to a combination of both CYP3A4 and P-glycoprotein induction.[221] The pharmacokinetics of cyclosporin were studied in six healthy volunteers after administration of cyclosporin orally (10 mg/kg) and intravenously (3 mg/kg) with and without rifampicin pretreatment (600 mg/day for 11 days). Blood clearance of cyclosporin increased from 0.3 L/h/kg (5 ml/min/kg) before rifampicin treatment to 0.42 L/h/kg (7 ml/min/kg) during rifampicin treatment, and bioavailability decreased from 27% without rifampicin to 10% with rifam Adis International Limited. All rights reserved.

picin. Rifampicin not only increased elimination clearance of cyclosporin, but also decreased its bioavailability to a greater extent than would have been predicted by the increased clearance. Since cyclosporin is a substrate for both CYP3A4 and P-glycoprotein, and since rifampicin can induce both CYP3A4 and P-glycoprotein, the increased clearance and decreased bioavailability of cyclosporin during rifampicin treatment is most probably due to a combination of CYP3A4 and P-glycoprotein induction. Because of overlapping substrate specificity between CYP3A4 and P-glycoprotein, and because of the similarities in the inhibitors and inducers between these two proteins, many drug interactions may involve both P-glycoprotein and CYP3A4. Unless the relative contribution of Pglycoprotein and CYP3A4 to overall drug interactions can be quantitatively differentiated, care should be taken in interpreting data for drug interactions, particularly in terms of the underlying mechanisms. Although attempts have been made to quantify the relative contribution of CYP3A4 and P-glycoprotein to overall interaction,[222] there is still no simple way by which the relative contribution of these two systems can be quantified because of the complexity of the interplay involved between intestinal and hepatic CYP3A4 and P-glycoprotein. 6. Conclusions Although the physiological function of P-glycoprotein is still not fully understood, the role of this efflux transporter in drug absorption and disposition is becoming increasingly defined. P-glycoprotein is highly expressed in various tissues, and the anatomical localisation of P-glycoprotein in relation to the sequences of drug movement (cellular uptake, intracellular distribution, metabolism and excretion) is a very important factor in determining P-glycoprotein function. From transgenic mouse studies, it appears that P-glycoprotein has a greater impact on limiting cellular uptake of drugs from the blood circulation into the brain and placenta, and from intestinal lumen into epithelial
Clin Pharmacokinet 2003; 42 (1)

92

Lin & Yamazaki

cells, than on enhancing the excretion of drugs in the liver and kidney into the adjacent luminal space. Although the transgenic mdr1a (/) and mdr1a/1b (/) mice provide a powerful tool for studying drug absorption and disposition, the interpretation of the data derived from the transgenic mouse studies are not always straightforward, and therefore should be viewed carefully. Inhibition and induction of P-glycoprotein have been reported as the cause of drug-drug interactions. Although the magnitude of the reported Pglycoprotein-mediated drug interactions, based on plasma AUC, appears to be quantitatively less important compared with CYP-mediated drug interactions, the potential risk of P-glycoproteinmediated drug interactions might be underestimated if only the plasma concentration is monitored. From animal studies, it is clear that P-glycoprotein inhibition has a much greater impact on tissue distribution, particularly with regard to the brain and placenta, than on plasma concentration of drugs. Therefore, one should carefully assess the potential risk of P-glycoprotein-mediated drug interactions when potent P-glycoprotein inhibitors and Pglycoprotein substrates are administered together. Another important lesson to be learned from the literature is that pharmacokinetic drug interactions often involve multiple mechanisms. Because of overlapping substrate specificity between CYP3A4 and P-glycoprotein, and because of similarities in P-glycoprotein and CYP3A4 inhibitors and inducers, many drug interactions involve both Pglycoprotein and CYP3A4. Unless the relative contribution of P-glycoprotein and CYP3A4 to overall drug interactions can be quantitatively estimated, care should be taken when exploring the underlying mechanism of drug interactions. Acknowledgements
The authors have provided no information on sources of funding or on conflicts of interest directly relevant to the content of this review.

References
1. Juliano RL, Ling V. A surface glycoprotein modulating drug permeability in Chinese hamster ovary cell mutants. Biochim Biophys Acta 1976; 455: 152-62

2. Gottesman MM, Pastan I. Biochemistry of multidrug resistance mediated by the multidrug transporter. Annu Rev Biochem 1993; 62: 385-427 3. Schinkel AH. The physiological function of drug-transporting P-glycoproteins. Cancer Biol 1997; 8: 161-70 4. van Helvoort A, Smith AJ, Sprong H, et al. MDR1 P-glycoprotein is a lipid translocase of broad specificity, while MDR3 P-glycoprotein specifically translocates phosphatidylcholine. Cell 1996; 87: 507-17 5. Ruetz S, Gros P. Phosphatidylcholine translocase: a physiological role for the mdr 2 gene. Cell 1994; 77: 1071-81 6. Smith AJ, van Helvoort A, van Meer G, et al. MDR3 P-glycoprotein, a phosphatidylcholine translocase, transports several cytotoxic drugs and directly interacts with drugs as judged by interference with nucleotide trapping. J Biol Chem 2000; 275: 23530-9 7. Thiebaut F, Tsuruo T, Hamada H, et al. Cellular localization of the multidrug resistance gene product P-glycoprotein in normal human tissues. Proc Natl Acad Sci U S A 1987; 84: 7735-8 8. Cordon-Cardo C, OBrien JP, Boccia J, et al. Expression of the multidrug resistance gene product (P-glycoprotein) in human normal and tumor tissues. J Histochem Cytochem 1990; 38: 1277-87 9. Schinkel AH, Smit JJM, van Tellingen O, et al. Disruption of the mouse mdr 1a P-glycoprotein gene leads to a deficiency in the blood-brain barrier and to increased sensitivity to drugs. Cell 1994; 77: 491-502 10. Schinkel AH, Mayer U, Wagenaar E, et al. Normal viability and altered pharmacokinetics in mice lacking mdr1-type (drugtransporting) P-glycoproteins. Proc Natl Acad Sci U S A 1997; 94: 4028-33 11. Chen C-J, Chin JE, Ueda K, et al. Internal duplication and homology with bacterial transport proteins in mdr1 (P-glycoprotein) gene from multidrug-resistant human cells. Cell 1986; 47: 381-9 12. Loo TW, Clarke DM. Reconstitution of drug-stimulated ATPase activity following co-expression of each half of human P-glycoprotein as separate polypeptides. J Biol Chem 1994; 269: 7750-5 13. Muller M, Bakos E, Welker E, et al. Altered drug-stimulated ATPase activity in mutants of human multidrug resistance protein. J Biol Chem 1996; 271: 1877-83 14. Takada Y, Yamada K, Taguchi Y, et al. Non-equivalent cooperation between the two nucleotide binding folds of P-glycoprotein. Biochim Biophys Acta 1998; 1373: 131-6 15. Hrycyna CA, Arian LE, Germann UA, et al. Structural flexibility of the linker region of human P-glycoprotein permits ATP hydrolysis and drug transport. Biochemistry 1998; 37: 13660-73 16. Ford JM, Hait WN. Pharmacology of drugs that alter multidrug resistance in cancer. Pharmacol Rev 1990; 42: 155-99 17. Homolya L, Hollo Z, Germann UA, et al. Fluorescent cellular indicators are extruded by the multidrug resistance protein. J Biol Chem 1993; 268: 21493-6 18. Shapiro AB, Ling V. Extraction of Hoechst 33342 from the cytoplasmic leaflet of the plasma membrane by P-glycoprotein. Eur J Biochem 1997; 250: 122-9 19. Higgins CF, Gottesman MM. Is the multidrug transporter a flippase? Trends Biochem Sci 1992; 17: 18-21 20. Rosenberg MF, Callaghan R, Ford RC, et al. Structure of multidrug resistance P-glycoprotein to 2.5 nm resolution determined by electron microscopy and image analysis. J Biol Chem 1997; 272: 10685-94

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

93

21. Hung LW, Wang IX, Nikaido K, et al. Crystal structure of the ATP-binding subunit of an ABC transporter. Nature 1998; 396: 703-7 22. Ambudkar SV, Dey S, Hrycyna CA, et al. Biochemical, cellular, and pharmacological aspects of the multidrug transporter. Annu Rev Pharmacol Toxicol 1999; 39: 361-98 23. Schurr E, Raymond M, Bell JC, et al. Characterization of the multidrug resistance protein expressed in cell clones stably transfected with the mouse mdr cDNA. Cancer Res 1989; 49: 2729-34 24. Senior AE, Gadsby DC. ATP hydrolysis cycles and mechanism in P-glycoprotein and CFTR. Semin Cancer Biol 1997; 8: 143-50 25. Shapiro AB, Ling V. Stoichiometry of coupling of rhodamine 123 transport to ATP hydrolysis by P-glycoprotein. Eur J Biochem 1998; 254: 189-93 26. Ambudkar SV, Cardarelli CO, Pashinsky I, et al. Relationship between the turnover number for vinblastine transport and for vinblastine-stimulated ATP hydrolysis by human P-glycoprotein. J Biol Chem 1997; 272: 21160-6 27. Greenberger LM. Major photoaffinity drug labeling sites for iodoaryl azidoprazosin in P-glycoprotein are within or immediately C-terminal to transmembrane domains 6 and 12. J Biol Chem 1993; 268: 11417-25 28. Bruggemann EP, Germann UA, Gottesman MM, et al. Two different regions of phosphoglycoprotein are photoaffinitylabeled by azidopine. J Biol Chem 1989; 264: 15483-8 29. Wu Q, Bounaud P, Kudul S, et al. Identification of the domains of photoincorporation of the 3- and 7-benzophenone analogues of taxol in the carboxyl-terminal half of murine mdr1b P-glycoprotein. Biochemistry 1998; 37: 11272-9 30. Ueda K, Taguchi Y, Morishima M. How does P-gp recognize its substrates? Semin Cancer Biol 1997; 8: 151-9 31. Taguchi Y, Kino K, Morishima M, et al. Alteration of substrate specificity by mutations at the His61 position in predicted transmembrane domain 1 of human MDR1/P-glycoprotein. Biochemistry 1997; 36: 8883-9 32. Taguchi Y, Morishima M, Komano T, et al. Amino acid substitutions in the first transmembrane domain (TM1) of P-glycoprotein alter substrate specificity. FEBS Lett 1997; 413: 142-6 33. Loo TW, Clarke DM. Functional consequences of glycine mutations in the predicted cytoplasmic loops of P-glycoprotein. J Biol Chem 1994; 269: 7243-8 34. Currier SJ, Kane SE, Willingham MC, et al. Identification of residues in the first cytoplasmic loop of P-glycoprotein involved in the function of chimeric human MDR1-MDR2 transporters. J Biol Chem 1992; 267: 25153-9 35. Pan B-F, Dutt A, Nelson JA. Enhanced transepithelial flux of cimetidine by Madin-Darby canine kidney cells overexpressing human P-glycoprotein. J Pharmacol Exp Ther 1994; 270: 1-7 36. Wu C-Y, Benet LZ, Hebert MF, et al. Differentiation of absorption and first-pass gut and hepatic metabolism in humans: studies with cyclosporine. Clin Pharmacol Ther 1995; 58: 492-7 37. Tang-Wai DF, Brossi A, Arnold LD, et al. The nitrogen of the acetamido group of colchicine modulates P-glycoprotein-mediated multidrug resistance. Biochemistry 1993; 32: 6470-6 38. Ueda K, Okamura N, Hirai M, et al. Human P-glycoprotein transports cortisol, aldosterone, and dexamethasone, but not progesterone. J Biol Chem 1992; 267: 24248-52

39. Chiba P, Holzer W, Landau M, et al. Substituted 4-acylpyrazoles and 4-acylpyrazolones: synthesis and multidrug resistance-modulating activity. J Med Chem 1998; 41: 4001-11 40. Ecker G, Huber M, Schmid D, et al. The importance of a nitrogen atom in modulators of multidrug resistance. Mol Pharmacol 1999; 56: 791-6 41. Seelig A, Landwojtowicz E. Structure-activity relationship of P-glycoprotein substrates and modifiers. Eur J Pharm Sci 2000; 12: 31-40 42. Seelig A. How does P-glycoprotein recognize its substrates? Int J Clin Pharmacol Ther 1998; 36: 50-4 43. Osterberg T, Norinder U. Theoretical calculation and prediction of P-glycoprotein-interacting drugs using MolSurf parametrization and PLS statistics. Eur J Pharm Sci 2000; 10: 295-303 44. Chiba P, Ecker G, Schmid D, et al. Structural requirements for activity of propafenone-type modulators in P-glycoproteinmediated multidrug resistance. Mol Pharmacol 1996; 49: 1122-30 45. Lin JH, Lu AYH. Interindividual variability in inhibition and induction of cytochrome P450 enzymes. Annu Rev Pharmacol Toxicol 2001; 41: 535-67 46. Lankas GR, Cartwright ME, Umbenhauer DR. P-glycoprotein deficiency in a subpopulation of CF-1 mice enhances avermectin-induced neurotoxicity. Toxicol Appl Pharmacol 1997; 143: 357-65 47. Umbenhauer DR, Lankas GR, Pippert TR, et al. Identification of a P-glycoprotein-deficient subpopulation in CF-1 mouse strain using a restriction fragment length polymorphism. Toxicol Appl Pharmacol 1997; 146: 88-94 48. Lankas GR, Wise LD, Cartwright ME, et al. Placenta P-glycoprotein deficiency enhances susceptibility to chemically induced birth defects in mice. Reprod Toxicol 1998; 12: 457-63 49. Pippert TR, Umbenhauer DR. The subpopulation of CF-1 mice deficient in P-glycoprotein contains a murine retroviral insertion in the mdr1a gene. J Biochem Mol Toxicol 2001; 15: 83-9 50. Pulliam JD, Seward RL, Henry RT, et al. Investigating ivermectin toxicity in Collies. Vet Med 1985; 80: 33-40 51. Paul AJ, Tranquilli WJ, Seward RL, et al. Clinical observations in Collies given ivermectin orally. Am J Vet Res 1987; 48: 684-5 52. Mealey KL, Bentjen SA, Gay JM, et al. Ivermectin sensitivity in collies is associated with a deletion mutation of the mdr1 gene. Pharmacogenetics 2001; 11: 727-33 53. Kioka N, Tsubota J, Kakehi Y, et al. P-glycoprotein gene (MDR1) cDNA from human adrenal: normal P-glycoprotein carries Gly185 with an altered pattern of multidrug resistance. Biochem Biophys Res Commun 1989; 162: 224-31 54. Mickley LA, Lee J-S, Weng Z, et al. Genetic polymorphism in MDR-1: a tool for examining allelic expression in normal cells, unselected and drug-selected cell lines, and human tumors. Blood 1998; 91: 1749-56 55. Hoffmeyer S, Burk O, von Richter O, et al. Functional polymorphisms of the human multidrug resistance gene: multiple sequence variations and correlation of one allele with P-glycoprotein expression and activity in vivo. Proc Natl Acad Sci U S A 2000; 97: 3473-8 56. Sakaeda T, Nakamura T, Horinouchi M, et al. MDR1 genotyperelated pharmacokinetics of digoxin after single oral administration in healthy Japanese subjects. Pharm Res 2001; 18: 1400-4 57. Kim RB, Leake BF, Choo EF, et al. Identification of functionally variant MDR1 alleles among European Americans and African Americans. Clin Pharmacol Ther 2001; 70: 189-99

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

94

Lin & Yamazaki

58. Drescher S, Schaeffeler E, Hitzl M, et al. MDR1 gene polymorphisms and disposition of the P-glycoprotein substrate fexofenadine. Br J Clin Pharmacol 2002; 53: 526-34 59. von Ahsen N, Fichter M, Grupp C, et al. No influence of the MDR-1 C3435T polymorphism or a CYP3A4 promoter polymorphism (CYP3A4-V allele) on dose-adjusted cyclosporin A trough concentrations or rejection incidence in stable renal transplant recipients. Clin Chem 2001; 47: 1048-52 60. Min DI, Ellingrod V. C3435T mutation in exon 26 of the human MDR1 gene and cyclosporine pharmacokinetics in healthy subjects. Ther Drug Monit 2002; 24: 400-4 61. Tanabe M, Ieiri I, Nagata N, et al. Expression of P-glycoprotein in human placenta: relation to genetic polymorphism of the multidrug resistance (MDR)-1 gene. J Pharmacol Exp Ther 2001; 297: 1137-43 62. Cascorbi I, Gerloff T, Johne A, et al. Frequency of single nucleotide polymorphisms in the P-glycoprotein drug transporter MDR1 gene in white subjects. Clin Pharmacol Ther 2001; 69: 169-74 63. Kalow W, Bertilsson L. Interethnic factors affecting drug response. Adv Drug Res 1994; 25: 1-53 64. Ameyaw M-M, Regateiro F, Li T, et al. MDR1 pharmacogenetics: frequency of the C3435T mutation in exon 26 is significantly influenced by ethnicity. Pharmacogenetics 2001; 11: 217-21 65. Elmore JG, Moceri VM, Carter D, et al. Breast carcinoma tumor characteristics in black and white women. Cancer 1998; 83: 2509-15 66. Ito S, Ieiri I, Tanabe M, et al. Polymorphism of the ABC transporter genes, MDR1, MRP1, and MRP2/cMOAT, in healthy Japanese subjects. Pharmacogenetics 2001; 11: 175-84 67. Yamazaki M, Neway WE, Ohe T, et al. In vitro substrate identification studies for P-glycoprotein mediated transport: species difference and predictability of in vitro results. J Pharmacol Exp Ther 2001; 296: 723-35 68. Gruol DJ, Vo QD, Zee MC. Profound differences in the transport of steroids by two mouse P-glycoproteins. Biochem Pharmacol 1999; 58: 1191-9 69. Devault A, Gros P. Two members of mouse mdr gene family confer multidrug resistance with overlapping but distinct drug specificities. Mol Cell Biol 1990; 10: 1652-63 70. Taylor JC, Ferry DR, Higgins CF, et al. The equilibrium and kinetic drug binding properties of the mouse P-gp1a and Pgp1b P-glycoproteins are similar. Br J Cancer 1999; 81: 783-9 71. Tang-Wai DF, Kajiji S, DiCapua F, et al. Human (MDR1) and mouse (mdr1, mdr3) P-glycoproteins can be distinguished by their respective drug resistance profiles and sensitivity to modulators. Biochemistry 1995; 34: 32-9 72. Croop JM, Raymond M, Haber D, et al. The three mouse multidrug resistance (mdr) genes are expressed in a tissue specific manner in normal mouse tissues. Mol Cell Biol 1989; 9: 1346-50 73. Schuetz EG, Umbenhauer DR, Yasuda K, et al. Altered expression of hepatic cytochromes P-450 in mice deficient in one or more mdr1 genes. Mol Pharm 2000; 57: 188-97 74. Aungst BJ. Novel formulation strategies for improving oral bioavailability of drug with poor membrane permeation or presystemic metabolism. J Pharm Sci 1993; 82: 979-87 75. Ho NF, Park JY, Ni PF, et al. Advancing quantitative and mechanistic approaches in interfacing gastrointestinal drug absorption studies in animals and humans. In: Crouthamel W, Sarapu AC, editors. Animal models for oral drug delivery in man: in situ and in vivo approaches. Washington, DC: Amer-

76.

77. 78.

79.

80.

81.

82.

83.

84.

85.

86.

87.

88.

89. 90.

91.

92.

93.

ican Pharmaceutics Association, Academy of Pharmaceutical Sciences, 1983: 27-106 Pade V, Stavchansky S. Estimation of the relative contribution of the transcellular and paracellular pathway to the transport of passively absorbed drugs in the Caco-2 model. Pharm Res 1997; 14: 1210-5 Creamer B. The turnover of the epithelium of small intestine. Br Med Bull 1967; 23: 226-30 Fojo AT, Ueda K, Slamon DJ, et al. Expression of a multidrug resistance gene in human tumors and tissues. Proc Natl Acad Sci U S A 1987; 84: 265-9 Fricker G, Drewe J, Huwyler J, et al. Relevance of P-glycoprotein for the enteral absorption of cyclosporine A: in vitro-in vivo correlation. Br J Pharmacol 1996; 118: 1841-7 Nakayama A, Saitoh H, Oda M, et al. Region-dependent disappearance of vinblastine in rat small intestine and characterization of its P-glycoprotein-mediated efflux system. Eur J Pharm Sci 2000; 11: 317-24 Lown KS, Mayo RR, Leichtman AB, et al. Role of intestinal P-glycoprotein (mdr1) in interpatient variation in the oral bioavailability of cyclosporine. Clin Pharmacol Ther 1997; 62: 248-60 Masuda S, Uemoto S, Hashida T, et al. Effect of intestinal Pglycoprotein on daily tacrolimus trough level in a living-donor small bowel recipient. Clin Pharmacol Ther 2000; 68: 98-103 Hunter J, Jepson MA, Tsuruo T, et al. Functional expression of P-glycoprotein in apical membranes of human intestinal Caco-2 cell layers: kinetics of vinblastine secretion and interaction with modulators. J Biol Chem 1993; 268: 14991-7 Hunter J, Hirst BH, Simmons NL. Drug absorption limited by P-glycoprotein-mediated secretory drug transport in human intestinal epithelial Caco-2 cells. Pharm Res 1993; 10: 743-9 Augustijins PF, Bradshaw TP, Gan LSL, et al. Evidence for a polarized efflux system in Caco-2 cells capable of modulating cyclosporine A transport. Biochem Biophys Res Commun 1993; 197: 360-5 Burton PS, Conradi RA, Hilgers AR, et al. Evidence for a polarized efflux system for peptides in the apical membrane of Caco-2 cells. Biochem Biophys Res Commun 1993; 190: 760-6 Sparreboom A, van Asperen J, Mayer U, et al. Limited oral bioavailability and active epithelial excretion of paclitaxel (Taxol) caused by P-glycoprotein in the intestine. Proc Natl Acad Sci U S A 1997; 94: 2031-5 Mayer U, Wagnaar E, Beijnen JH, et al. Substantial excretion of digoxin via the intestinal mucosa and prevention of longterm digoxin accumulation in the brain by the mdr1a P-glycoprotein. Br J Pharmacol 1996; 119: 1038-44 Israili ZH, Dayton PG. Enhancement of xenobiotic elimination: role of intestinal excretion. Drug Metab Rev 1984; 15: 1123-59 Meerum Terwgot JM, Malingre MM, Beijnen JH, et al. Co-administration of cyclosporin A enables oral therapy with paclitaxel. Clin Cancer Res 1999; 5: 3379-84 Malingre MM, Richel DJ, Beijinen JH, et al. Coadministration of cyclosporine strongly enhances the oral bioavailability of docetaxel. J Clin Oncol 2001; 19: 1160-6 Bohme M, Buchler M, Muller M, et al. Differential inhibition by cyclosporines of primary-active ATP-dependent transporters in hepatocyte canalicular membrane. FEBS Lett 1993; 333: 193-6 Stephens RH, ONeill CA, Warhurst A, et al. Kinetic profiling of P-glycoprotein-mediated drug efflux in rat and human intestinal epithelia. J Pharmacol Exp Ther 2001; 296: 584-91

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

95

94. Saeki T, Ueda K, Tanigawara Y, et al. Human P-glycoprotein transports cyclosporine A and FK506. J Biol Chem 1993; 268: 6077-80 95. Wetterich U, Sphn-Langguth H, Mutschler E, et al. Evidence for intestinal secretion as an additional clearance pathway of talinolol enantiomers: concentration- and dose-dependent absorption in vitro and in vivo. Pharm Res 1996; 13: 514-22 96. Ueda CT, Lemaire M, Gsell G, et al. Apparent dose dependent oral absorption of cyclosporine A in rats. Biopharm Drug Dispos 1984; 5: 141-51 97. Hochman JH, Chiba M, Nishime J, et al. Influence of P-glycoprotein on the transport and metabolism of indinavir in Caco2 cells expressing cytochrome P450 3A4. J Pharmacol Exp Ther 2000; 292: 310-8 98. Lin JH. Role of pharmacokinetics in the discovery and development of indinavir. Adv Drug Deliv Rev 1999; 39: 33-49 99. Chiou WL, Chung SM, Wu TC, et al. A comprehensive account on the role of efflux transporters in the gastrointestinal absorption of 13 commonly used substrate drugs in humans. Int J Clin Pharmacol Ther 2001; 39: 93-101 100. Makhey VD, Guo A, Norris DA, et al. Characterization of the regional intestinal kinetics of drug efflux in rat and human intestine and in Caco-2 cells. Pharm Res 1998; 15: 1160-7 101. Handschumacher RE. Immunosuppressive agents. In: Gilman AG, Palmer T, Nies AS, editors. Goodman and Gilmans the pharmacological basis of therapeutics. 8th ed. New York (NY): McGraw-Hill Inc, 1990: 1264-76 102. Mouritsen OG, Jorgensen K, Honger T. Permeability of lipid bilayers near the phase transition. In: Disalvo EA, Simon SA, editors. Permeability and stability of lipid bilayers. Boca Raton (FL): CRC Press, 1995: 137-60 103. Eichler H-G, Muller M. Drug distribution: the forgotten relative in clinical pharmacokinetics. Clin Pharmacokinet 1998; 34: 95-9 104. Kim RB. Transporters and drug disposition. Curr Opin Drug Discov Devel 2000; 3: 94-101 105. Pardridge WM. Transport of protein-bound hormones into tissue in vivo. Endocr Rev 1981; 2: 103-23 106. Rapoport SI. Transport in cells and tissues. In: Rapport SI, editor. Blood-brain barrier in physiology and medicine. New York (NY): Raven Press, 1976: 17-42 107. Chikhale EG, Ng K-Y, Burton PS, et al. Hydrogen bonding potential as a determinant of the in vitro and in situ bloodbrain barrier permeability of peptides. Pharm Res 1994; 11: 412-9 108. Levin VA. Relationship of octanol/water partition coefficient and molecular weight to rat brain capillary permeability. J Med Chem 1980; 23: 682-4 109. Lin TH, Lin JH. Effects of protein binding and experimental disease states on brain uptake of benzodiazepines in rats. J Pharmacol Exp Ther 1990; 253: 45-50 110. Thiebaut F, Tsuruo T, Hamada H, et al. Immunohistochemical localization in normal tissues of different epitopes in the multidrug transport protein P170: evidence for localization in brain capillaries and crossreactivity of one antibody with a muscle protein. J Histochem Cytochem 1989; 37: 159-64 111. Cordon-Cardo C, OBrien JP, Casals D, et al. Multidrug-resistance gene (P-glycoprotein) is expressed by endothelial cells at blood-brain barrier sites. Proc Natl Acad Sci U S A 1989; 86: 695-8 112. Beaulieu E, Demeule M, Ghitescu L, et al. P-glycoprotein is strongly expressed in the luminal membranes of the endothelium of blood vessels in the brain. Biochem J 1997; 326: 539-44

113. Barrand MA, Bennett GC, Taylor CJ, et al. Immunohistochemical localization in rat brain microvessels of transporters involved in solute and water movements across the blood-brain barrier [abstract]. IVth International Conference: Cerebral Vascular Biology, Blood-Brain Barrier; 2001 Apr 1-5; Cambridge, UK 114. Matsuoka Y, Okazaki M, Kitamura Y, et al. Developmental expression of P-glycoprotein (multidrug resistance gene product) in the rat brain. J Neurobiol 1999; 39: 383-92 115. Decleves X, Regina A, Laplanche J-L, et al. Functional expression of P-glycoprotein and multidrug resistance-associated protein (Mrp1) in primary cultures of rat astrocytes. J Neurosci Res 2000; 60: 594-601 116. Pardridge WM, Golden PL, Kang Y-S, et al. Brain microvascular and astrocyte localization of P-glycoprotein. J Neurochem 1997; 68: 1278-85 117. Golden PL, Pardridge WM. Brain microvascular P-glycoprotein and a revised model of multidrug resistance in brain. Cell Mol Neurobiol 2000; 20: 165-81 118. de Lange ECM, de Bock G, Schinkel AH, et al. BBB transport and P-glycoprotein functionality using mdr1a (/) and wildtype mice: total brain versus microdialysis concentration profiles of rhodamine-123. Pharm Res 1998; 15: 1657-65 119. Lee G, Schlichter L, Bendayan M, et al. Functional expression of P-glycoprotein in rat brain microglia. J Pharmacol Exp Ther 2001; 299: 204-12 120. Chen C, Pollack GM. Altered disposition and antinociception of [D-penicillamine2,5]enkephalin in mdr1a-gene-deficient mice. J Pharmacol Exp Ther 1998; 287: 545-52 121. Tsuji A, Terasaki T, Takabatake Y, et al. P-glycoprotein as the drug efflux pump in the primary cultured bovine brain capillary endothelial cells. Life Sci 1992; 51: 1427-37 122. Tatsuta T, Naito M, Oh-hara T, et al. Functional involvement of P-glycoprotein in blood-brain barrier. J Biol Chem 1992; 267: 20383-91 123. Tsuji A, Tamai I, Sakata A, et al. Restricted transport of cyclosporine A across the blood-brain barrier by a multidrug transporter, P-glycoprotein. Biochem Pharmacol 1993; 46: 1096-9 124. Shirai A, Naito M, Tatsuta T, et al. Transport of cyclosporin A across the brain capillary endothelial cell monolayer by Pglycoprotein. Biochim Biophys Acta 1994; 1222: 400-4 125. Biegel D, Spencer DD, Pachter JS, et al. Isolation and culture of human brain microvessel endothelial cells for the study of blood-brain barrier properties in vitro. Brain Res 1995; 692: 183-9 126. Hsing S, Gatmaitan Z, Arias IM. The function of Gp170, the multidrug-resistance gene product, in the brush border of rat intestinal mucosa. Gastroenterology 1992; 102: 879-85 127. Kamimoto Y, Gatmaitan Z, Hsu J, et al. The function of Gp170, the multidrug resistance gene product, in rat liver canalicular membrane vesicles. J Biochem Chem 1989; 264: 11693-8 128. Ohnishi T, Tamai I, Sakanaka K, et al. In vivo and in vitro evidence for ATP-dependency of P-glycoprotein-mediated efflux of doxorubicin at the blood-brain barrier. Biochem Pharmacol 1995; 49: 1541-4 129. Sakata A, Tamai I, Kawazu K, et al. In vivo evidence for ATPdependent and P-glycoprotein-mediated transport of cyclosporin A at the blood-brain barrier. Biochem Pharmacol 1994; 48: 1989-92 130. Schinkel AH, Wagenaar E, van Deemter L, et al. Absence of the mdr1a P-glycoprotein in mice affects tissue distribution and pharmacokinetics of dexamethasone, digoxin, and cyclosporin A. J Clin Invest 1995; 96: 1698-705

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

96

Lin & Yamazaki

131. Schinkel AH, Mol CAAM, Wagenaar E, et al. Multidrug resistance and the role of P-glycoprotein knockout mice. Eur J Cancer 1995; 31A: 1295-8 132. Yokogawa K, Takahashi M, Tamai I, et al. P-glycoproteindependent disposition kinetics of tacrolimus: studies in mdr1a knockout mice. Pharm Res 1999; 16: 1213-8 133. Smit JW, Huisman MT, van Tellingen O, et al. Absence or pharmacological blocking of placental P-glycoprotein profoundly increases fetal drug exposure. J Clin Invest 1999; 104: 1441-7 134. Nakamura Y, Ikeda S, Furukawa T, et al. Function of P-glycoprotein expressed in placenta and mole. Biochem Biophys Res Commun 1997; 235: 849-53 135. Ushigome F, Takanaga H, Matsuo H, et al. Human placental transport of vinblastine, vincristine, digoxin and progesterone: contribution of P-glycoprotein. Eur J Pharmacol 2000; 408: 1-10 136. Krishna DR, Klotz U. Extrahepatic metabolism of drugs in humans. Clin Pharmacokinet 1994; 26: 144-60 137. Debri K, Boobis AR, Davis DS, et al. Distribution and induction of CYP3A1 and CYP3A2 in rat liver and extrahepatic tissues. Biochem Pharmacol 1995; 50: 2047-56 138. Watkins PB, Murray SA, Thomas PE, et al. Distribution of cytochromes P-450, cytochrome b5, and NADPH-cytochrome P-450 reductase in an entire human liver. Biochem Pharmacol 1990; 39: 471-6 139. Murray GI, Barnes TS, Sewell HF, et al. The immunochemical localisation and distribution of cytochrome P-450 in normal hepatic and extrahepatic tissues with a monoclonal antibody to human cytochrome P-450. Br J Clin Pharmacol 1988; 25: 465-75 140. Thummel KE, Kunze KL, Shen DD. Enzyme-catalyzed processes of first-pass hepatic and intestinal drug extraction. Adv Drug Deliv Rev 1997; 27: 99-127 141. Hochman JH, Chiba M, Yamazaki M, et al. P-glycoprotein-mediated efflux of indinavir metabolites in Caco-2 cells expressing cytochrome P450 3A4. J Pharmacol Exp Ther 2001; 298: 323-30 142. Gan L-SL, Moseley MA, Khosla B, et al. CYP3A-like cytochrome P450-mediated metabolism and polarized efflux of cyclosporin A in Caco-2 cells. Drug Metab Dispos 1996; 24: 344-9 143. Lin JH, Chiba M, Chen I-W, et al. Effect of dexamethasone on the intestinal first-pass metabolism of indinavir in rats: evidence of cytochrome P-450 3A and P-glycoprotein induction. Drug Metab Dispos 1999; 27: 1187-93 144. Lin JH, Chiba M, Baillie TA. Is the role of the small intestine in first-pass metabolism overemphasized? Pharmacol Rev 1999; 51: 135-57 145. Meijer DKF, Smit JW, Muller M. Hepatobiliary elimination of cationic drugs: the role of P-glycoproteins and other ATP-dependent transporters. Adv Drug Deliv Rev 1997; 25: 159-200 146. Koepsell H, Gorboulev V, Arndt P. Molecular pharmacology of organic cation transporters in kidney. J Membr Biol 1999; 167: 103-17 147. Watanabe T, Miyauchi S, Sawada Y, et al. Kinetic analysis of hepatobiliary transport of vincristine in perfused rat liver: possible roles of P-glycoprotein in biliary excretion of vincristine. J Hepatol 1992; 16: 77-88 148. Ballet F, Vrignaud P, Robert J, et al. Hepatic extraction, metabolism and biliary excretion of doxorubicin in the isolated perfused rat liver. Cancer Chemother Pharmacol 1987; 19: 240-5

149. Kawahara M, Sakata A, Miyashita T, et al. Physiologically based pharmacokinetics of digoxin in mdr1a knockout mice. J Pharm Sci 1999; 88: 1281-7 150. van Asperen J, van Tellingen O, Beijnen JH. The role of mdr1a P-glycoprotein in the biliary and intestinal secretion of doxorubicin and vinblastine in mice. Drug Metab Dispos 2000; 28: 264-7 151. Smit JW, Schinkel AH, Muller M, et al. Contribution of the murine mdr1a P-glycoprotein to hepatobiliary and intestinal elimination of cationic drugs as measured in mice with an mdr1a gene disruption. Hepatology 1998; 27: 1056-63 152. Smit JW, Schinkel AH, Weert B, et al. Hepatobiliary and intestinal clearance of amphiphilic cationic drugs in mice in which both mdr1a and mdr1b genes have been disrupted. Br J Pharmacol 1998; 124: 416-24 153. Okamura N, Hirai M, Tanigawara Y, et al. Digoxin-cyclosporin A interaction: modulation of the multidrug transporter P-glycoprotein in the kidney. J Pharmacol Exp Ther 1993; 266: 1614-9 154. Tanigawara Y, Okamura N, Hirai M, et al. Transport of digoxin by human P-glycoprotein expressed in a porcine kidney epithelial cell line (LLC-PK1). J Pharmacol Exp Ther 1992; 263: 840-5 155. Horio M, Chin K-V, Currier SJ, et al. Transepithelial transport of drugs by the multidrug transporter in cultured MadinDarby canine kidney cell epithelia. J Biol Chem 1989; 264: 14880-4 156. Hori R, Okamura N, Aiba T, et al. Role of P-glycoprotein in renal tubular secretion of digoxin in the isolated perfused rat kidney. J Pharmacol Exp Ther 1993; 266: 1620-5 157. De Lannoy IAM, Koren G, Klein J, et al. Cyclosporin and quinidine inhibition of renal digoxin excretion: evidence for luminal secretion of digoxin. Am J Physiol 1992; 263: F613-22 158. van Asperen J, van Tellinge O, Tijssen F, et al. Increased accumulation of doxorubicin and doxorubicinol in cardiac tissue of mice lacking mdr1a P-glycoprotein. Br J Cancer 1999; 79: 108-13 159. Lin JH, Lu AYH. Inhibition and induction of cytochrome P450 and the clinical implications. Clin Pharmacokinet 1998; 35: 361-90 160. Ronis MJJ, Ingelman-Sundberg M. Induction of human drugmetabolizing enzymes: mechanism and implications. In: Woolf TF, editor. Handbook of drug metabolism. New York (NY): Marcel Dekker Inc, 1999: 239-62 161. Lazarou J, Pomeraanz BH, Corey PN. Incidence of adverse drug reactions in hospitalized patients: a meta-analysis of prospective studies. JAMA 1998; 279: 1200-5 162. Duchateau AMJA. Posicor: veni, vidi, foetsie. Pharm Weekbl 1998; 133: 1294-5 163. Ford JM. Experimental reversal of P-glycoprotein-mediated multidrug resistance by pharmacological chemosensitisers. Eur J Cancer 1996; 32A: 991-1001 164. Tamai I, Safa AR. Azidopine noncompetitively interacts with vinblastine and cyclosporin A binding to P-glycoprotein in multidrug resistant cells. J Biol Chem 1991; 266: 16796-800 165. Ramachandra M, Ambudkar SV, Chen D, et al. Human P-glycoprotein exhibits reduced affinity for substrates during a catalytic transition state. Biochemistry 1998; 37: 5010-9 166. Senior AE, Al-Shawi MK, Urbatsch IL. The catalytic cycle of P-glycoprotein. FEBS Lett 1995; 377: 285-9 167. Ayesh S, Shao Y-M, Stein WD. Co-operative, competitive and non-competitive interactions between modulators of P-glycoprotein. Biochim Biophys Acta 1996; 1316: 8-18

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

P-Glycoprotein

97

168. Tamai I, Safa AR. Competitive interaction of cyclosporins with the vinca alkaloid-binding site of P-glycoprotein in multidrug resistant cells. J Biol Chem 1990; 265: 16509-13 169. Callaghan R, Riordan JR. Synthetic and nature opiates interact with P-glycoprotein in multidrug resistant cells. J Biol Chem 1993; 268: 16059-64 170. Litman T, Zeuthen T, Skovsgaard T, et al. Competitive, noncompetitive and cooperative interactions between substrates of P-glycoprotein as measured by its ATPase activity. Biochim Biophys Acta 1997; 1361: 169-76 171. Pascaud C, Garrigos M, Orlowski S. Multidrug resistance transporter P-glycoprotein has distinct but interacting binding sites for cytotoxic drugs and reversing agents. Biochem J 1998; 333: 351-8 172. Critchfield JW, Welsh CJ, Phang JM, et al. Modulation of adriamycin accumulation and efflux by flavonoids in HCT-15 colon cells. Biochem Pharmacol 1994; 48: 1437-45 173. Shapiro AB, Ling V. Positively cooperative sites for drug transport by P-glycoprotein with distinct drug specificities. Eur J Biochem 1997; 250: 130-7 174. Houston JB, Kenworthy KE. In vitro-in vivo scaling of CYP kinetic data not consistent with the classical MichaelisMenten model. Drug Metab Dispos 2000; 28: 246-54 175. Wang RW, Newton DJ, Liu N, et al. Human cytochrome P450 3A4; in vitro drug-drug interaction patterns are substratedependent. Drug Metab Dispos 2000; 28: 360-6 176. Shou M, Mei Q, Ettore MW, et al. Sigmoidal kinetic model for two co-operative substrate-binding sites in a cytochrome P450 3A4 active site: an example of the metabolism of diazepam and its derivatives. Biochem J 1999; 340: 845-53 177. Wandel C, Kim RB, Kajiji S, et al. P-glycoprotein and cytochrome P-450 3A inhibition: dissociation of inhibitory potencies. Cancer Res 1999; 59: 3944-8 178. Choo EF, Leake B, Wandel C, et al. Pharmacological inhibition of P-glycoprotein transport enhances the distribution of HIV1 protease inhibitors into brain and testes. Drug Metab Dispos 2000; 28: 655-60 179. Polli JW, Jarrett JL, Studenberg SD, et al. Role of P-glycoprotein on the CNS disposition of amprenavir (141W94), an HIV protease inhibitor. Pharm Res 1999; 16: 1206-12 180. Mayer U, Wagenaar E, Dorobek B, et al. Full blockade of intestinal P-glycoprotein and extensive inhibition of bloodbrain barrier P-glycoprotein by oral treatment of mice with PSC833. J Clin Invest 1997; 100: 2430-6 181. Sadeque AJ, Wandel C, He H, et al. Increased drug delivery to the brain by P-glycoprotein inhibition. Clin Pharmacol Ther 2000; 68: 231-7 182. Verschraagen M, Koks CHW, Schellens JHM, et al. P-glycoprotein system as a determinant of drug interactions: the case of digoxin-verapamil. Pharmacol Res 1999; 40: 301-6 183. Bussey HI. The influence of quinidine and other agents on digitalis glycosides. Am Heart J 1982; 104: 289-302 184. Mordel A, Halkin H, Zulty L, et al. Quinidine enhances digitalis toxicity at therapeutic serum digoxin levels. Clin Pharmacol Ther 1993; 53: 457-62 185. Pedersen KE. Digoxin interaction: the influence of quinidine and verapamil on pharmacokinetics and receptor binding of digitalis glycosides. Acta Med Scand 1985; 697: 11-40 186. Hinderling PH, Hartmann D. Pharmacokinetics of digoxin and main metabolites/derivatives in healthy humans. Ther Drug Monit 1991; 13: 381-401 187. Sababi M, Borga O, Hultkvist-Bengtsson U. The role of P-glycoprotein in limiting intestinal regional absorption of digoxin in rats. Eur J Pharm Sci 2001; 14: 21-7

188. Su FG, Huang JD. Inhibition of the intestinal digoxin absorption and exsorption by quinidine. Drug Metab Dispos 1996; 24: 142-7 189. Tsuruo T, Lida H, Tsukagoshi S, et al. Overcoming of vincristine resistance in P388 leukemia in vivo and in vitro through enhanced cytotoxicity of vincristine and vinblastine by verapamil. Cancer Res 1981; 41: 1967-72 190. Advani R, Fisher GA, Lum BL, et al. A phase I trial of doxorubicin, paclitaxel, and valspodar (PSC833), a modulator of multidrug resistance. Clin Cancer Res 2001; 7: 1221-9 191. Sparreboom A, St Planting A, Jewell RC, et al. Clinical pharmacokinetics of doxorubicin in combination with GF120918, a potent inhibitor of MDR1 P-glycoprotein. Anticancer Drugs 1999; 10: 719-28 192. van Zuylen L, Nooter K, Sparreboom A, et al. Development of multidrug resistance convertors: sense or nonsense? Invest New Drugs 2000; 18: 205-20 193. Johannessen A, Rendtorff C, Poulsen S. Digoxin intoxication induced by verapamil in an uremic patient. Clin Nephrol 1985; 24: 158-9 194. Klein HO, Lang R, Weiss E, et al. The influence of verapamil on serum digoxin concentration. Circulation 1982; 65: 998-1003 195. Fardel O, Lecureur V, Corlu A, et al. P-glycoprotein induction in rat liver epithelial cells in response to acute 3methycholanthrene treatment. Biochem Pharmacol 1996; 51: 1427-36 196. Chin KV, Chauhan SS, Pastan I, et al. Regulation of mdr RNA levels in response to cytotoxic drugs in rodent cells. Cell Growth Differ 1990; 1: 361-5 197. LeCluyse EL. Pregnane X receptor: molecular basis for species differences in CYP3A induction by xenobiotics. Chem Biol Interact 2001; 134: 283-9 198. Fardel O, Lecureur V, Guillouzo A. Regulation by dexamethasone of P-glycoprotein expression in cultured rat hepatocytes. FEBS Lett 1993; 327: 189-93 199. Zhao JY, Ikeguchi M, Eckersberg T, et al. Modulation of multidrug resistance gene expression by dexamethasone in cultured hepatoma cells. Endocrinology 1993; 133: 521-8 200. Salphati L, Benet LZ. Modulation of P-glycoprotein expression by cytochrome P450 3A inducers in male and female rat livers. Biochem Pharmacol 1998; 55: 387-95 201. Liu J, Brunner LJ. Chronic cyclosporine administration induces renal P-glycoprotein in rats. Eur J Pharmacol 2001; 418: 127-32 202. Jette L, Beaulieu E, Leclerc J-M, et al. Cyclosporin A treatment induces overexpression of P-glycoprotein in the kidney and other tissues. Am J Physiol 1996; 270: F756-65 203. Lee CH. Induction of P-glycoprotein mRNA transcripts by cycloheximide in animal tissues: evidence that class I Pgp is transcriptionally regulated whereas class II Pgp is post-transcriptionally regulated. Mol Cell Biochem 2001; 216: 103-10 204. Durr D, Stieger B, Kullak-Ublick GA, et al. St. Johns Wort induces intestinal P-glycoprotein/MDR1 and intestinal and hepatic CYP3A4. Clin Pharmacol Ther 2000; 68: 598-604 205. Schuetz EG, Beck WT, Schuetz JD. Modulators and substrates of P-glycoprotein and cytochrome P450 3A coordinately upregulated these proteins in human colon carcinoma cells. Mol Pharmacol 1996; 49: 311-8 206. Pichard L, Fabre I, Daujat M, et al. Effect of corticosteroids on the expression of cytochromes P450 and on cyclosporin A oxidase activity in primary cultures human hepatocytes. Mol Pharmacol 1992; 41: 1047-55

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

98

Lin & Yamazaki

207. Wacher VJ, Wu C-Y, Benet LZ. Overlapping substrate specificities and tissue distribution of cytochrome P450 3A and P-glycoprotein: implications for drug delivery and activity in cancer chemotherapy. Mol Carcinog 1995; 13: 129-34 208. Callen DF, Baker E, Simmers RN, et al. Localization of the human multidrug resistance gene, MDR1, to 7q21.1. Hum Genet 1987; 77: 142-4 209. Inoue K, Inazawa J, Nakagawa H, et al. Assignment of the human cytochrome P450 nifedipine oxidase gene (CYP3A4) to chromosome 7 at band q22.1 by fluorescence in situ hybridization. Jpn J Hum Genet 1992; 37: 133-8 210. Quattrochi LC, Guzelian PS. CYP3A regulation: from pharmacology to nuclear receptors. Drug Metab Dispos 2001; 29: 615-22 211. Jones SA, Moore LB, Shenk JL, et al. The pregnane X receptor: a promiscuous xenobiotic receptor that has diverged during evolution. Mol Endocrinol 2000; 14: 27-39 212. Blumberg B, Sabbagh W Jr, Juguilon H, et al. SXR, a novel steroid and xenobioticsensing nuclear receptor. Genes Dev 1998; 12: 3195-205 213. Moore LB, Goodwin B, Jones SA, et al. St. Johns Wort induces hepatic drug metabolism through activation of the pregnane X receptor. Proc Natl Acad Sci U S A 2000; 97: 7500-2 214. Synold TW, Dussault I, Forman BM. The orphan nuclear receptor SXR coordinately regulates drug metabolism and efflux. Nature Med 2001; 7: 584-90 215. Geick A, Eichelbaum M, Burk O. Nuclear receptor response elements mediate induction of intestinal MDR1 by rifampin. J Biol Chem 2001; 276: 14581-7

216. Masuyama H, Hiramatsu Y, Mizutani Y, et al. The expression of pregnane X receptor and its target gene, cytochrome P450 3A1 in prenatal mouse. Mol Cell Endocrinol 2001; 172: 47-56 217. Greiner B, Eichelbaum M, Fritz P, et al. The role of intestinal P-glycoprotein in the interaction of digoxin and rifampin. J Clin Invest 1999; 104: 147-53 218. Hamman MA, Bruce MA, Haehner-Daniels BD, et al. The effect of rifampin administration on the disposition of fexofenadine. Clin Pharmacol Ther 2001; 69: 114-21 219. Lippert C, Ling J, Brown P, et al. Mass balance and pharmacokinetics of MDL16,455A in healthy male volunteers [abstract]. Pharm Res 1999; 12: S390 220. Cvetkovic M, Leake B, Fromm MF, et al. OATP and P-glycoprotein transporters mediate the cellular uptake and excretion of fexofenadine. Drug Metab Dispos 1999; 27: 866-71 221. Hebert MF, Roberts JP, Prueksaritanont T, et al. Bioavailability of cyclosporine with concomitant rifampin administration is markedly less than predicted by hepatic enzyme induction. Clin Pharmacol Ther 1992; 52: 453-7 222. Wacher VJ, Silverman JA, Zhang Y, et al. Role of P-glycoprotein and cytochrome P450 3A in limiting oral absorption of peptides and peptidomimetics. J Pharm Sci 1998; 87: 1322-30

Correspondence and offprints: Dr Jiunn H. Lin, Drug Metabolism, Merck Research Laboratories, West Point, WP75A203, PA 19486, USA. E-mail: jiunn_lin@merck.com

Adis International Limited. All rights reserved.

Clin Pharmacokinet 2003; 42 (1)

You might also like