Abstract
The ATP-binding cassette (ABC) transporters constitute a large family of membrane proteins, which transport a variety of compounds through the membrane against a concentration gradient at the cost of ATP hydrolysis. Substrates of the ABC transporters include lipids, bile acids, xenobiotics, and peptides for antigen presentation. As they transport exogenous and endogenous compounds, they reduce the body load of potentially harmful substances. One by-product of such protective function is that they also eliminate various useful drugs from the body, causing drug resistance. This review is a brief summary of the structure, function, and expression of the important drug resistance–conferring members belonging to three subfamilies of the human ABC family; these are ABCB1 (MDR1/P-glycoprotein of subfamily ABCB), subfamily ABCC (MRPs), and ABCG2 (BCRP of subfamily ABCG), which are expressed in various organs. In the text, the transporter symbol that carries the subfamily name (such as ABCB1, ABCC1, etc.) is used interchangeably with the corresponding original names, such as MDR1/P-glycoprotein, MRP1, etc., respectively. Both nomenclatures are maintained in the text because both are still used in the transporter literature. This helps readers relate various names that they encounter in the literature. It now appears that P-glycoprotein, MRP1, MRP2, and BCRP can explain the phenomenon of multidrug resistance in all cell lines analyzed thus far. Also discussed are the gene structure, regulation of expression, and various polymorphisms in these genes. Because genetic polymorphism is thought to underlie interindividual differences, including their response to drugs and other xenobiotics, the importance of polymorphism in these genes is also discussed.
Knowledge of the absorption, distribution, metabolism, and excretion (ADME) profiles of foreign molecules in a physiological system is integral to the disciplines of pharmacology and toxicology. For the last 40 years or so, significant progress has been made in understanding the metabolism (M) aspect of ADME. The major groups of phase I and phase II drug- and xenobiotic-metabolizing enzymes have been cloned, their tissue-specific and developmental stage–specific expression patterns have been studied in various species, and their catalytic properties, inducers, and substrates have been identified (Guengerich 1994). Such knowledge has had an enormous impact on drug discovery as well as basic mechanistic toxicology studies. In contrast, studies on the other three aspects of ADME, that is, absorption (A), distribution (D), and excretion (E), have lagged behind that on metabolism. These three phenomena were traditionally interpreted by classical physiological explanations of solute movement through cell membrane. Advances in molecular biological techniques have now changed that state of understanding.
During the last decade, extensive membrane physiology and molecular biology studies on membrane transporters have been undertaken. Many transporters have been cloned, their expression patterns in tissues have been studied, a number of their substrates have been identified, and significant progress has been made in understanding the mechanism of transport of various xenobiotics as well as endobiotics in and out of cells. Understanding drug and xenobiotic transport process is important not only from the scientific standpoint of gaining insight into an important cellular function, but also from the perspective of drug development and toxicology. An understanding of the substrate specificity, tissue expression, and transport kinetics of various transporters can lead to insight comparable to that gained from studies of drug-metabolizing enzymes, which has significantly benefited the drug discovery and development processes. Such knowledge will help understand transport-mediated drug-drug interactions as well as toxicity. For example, in humans quinidine (antiarrhythmic drug) can reduce biliary clearance of digoxin (cardiac glycoside) by about 40% because it inhibits the multidrug transporter (MDR1) that transports digoxin (Horio, Gottsman, and Pastan 1988). Similarly, bosentan (endothelin receptor antagonist) inhibits the bile acid transporter BSEP, resulting in reduced transport of bile acids and consequent liver damage (Fattinger et al. 2001). Inhibition of BSEP has been implicated as one mechanism of cholestatic liver injury. Understanding the conformation of the substrate binding site of transporters may also help in the development of effective inhibitors, which, in turn, may help reverse the drug resistance phenotypes.
Genome sequencing has helped identify numerous transporters that were previously unidentified. According to an initial estimate, transporters constitute about 2% of the total genes in the human genome (International Human Genome Sequencing Consortium [IHGSC] 2001; Venter et al. 2001). Because the human genome consists of approximately 30,000 to 40,000 genes, 2% translates to roughly 700 transporter genes. All these putative transporters have not been identified or characterized yet.
Membrane transporters have been classified into two major families by the HUGO Gene Nomenclature Committee (http://www.gene.ucl.ac.uk/nomenclature/). These are the ATP-binding cassette (ABC) transporter family and the solute carrier (SLC) transporter family.
The following discussion will be limited to human transporters of the ATP-binding cassette (ABC) transporter family, which have been relatively well characterized for their ability to transport xenobiotics (drugs, toxins) and endobiotics. The discussion of ABC transporters will focus mainly on MDR1/P-glycoprotein, MRPs, and BCRP.1
The expression MDR stands for multiple drug resistance. The gene is designated MDR1, and the protein encoded by it is designated P-glycoprotein (P-gp). Resistance to multiple anticancer agents is a major impediment to the successful treatment of many forms of malignant disease. In tumor cell lines, multidrug resistance was originally attributed to the overexpression of the 170-kDa MDR1/P-gp. However, the isolation of a second distantly related ABC protein, the 190-kDa multidrug resistance protein 1 (MRP1) facilitated the discovery of eight more MRPs that can all potentially contribute to drug resistance.
The expression MRP stands for multidrug resistance protein. The genes are designated ABCC/MRP(italicized) with a numerical suffix, such as ABCC1(or MRP1), ABCC2 (or MRP2), whereas the protein is designated without italicizing, such as ABCC1 (or MRP1), ABCC2 (or MRP2).
The expression BCRP stands for breast cancer resistance protein. It is a recently discovered clinically relevant ∼70-kDa efflux transporter of the ABC family that also confers drug resistance. It is about half the size of P-gp.
Found in the cells of all living organisms, ABC membrane proteins use the energy of ATP hydrolysis to unidirectionally transport a wide range of molecules across the cell membrane. P-gp, MRPs, and BCRP are responsible for the efflux of many xenobiotics and clinically important drugs, thereby causing drug resistance phenotype. It now appears that these transporters can explain the phenomenon of multidrug resistance in all cell lines analyzed thus far.
CLASSIFICATION OF ATP-BINDING CASSETTE (ABC) TRANSPORTERS
The term ABC transporter was introduced in 1992 by Christopher Higgins (Higgins 1992). The designation ABC was based on the highly conserved ATP-binding cassette, the most characteristic feature of this family. In the human ABC transporter family, 49 genes have been identified so far and classified into seven subfamilies. The subfamily designations are as follows, with the number of members in each subfamily indicated in parenthesis: ABCA (12), ABCB (11), ABCC (13), ABCD (4), ABCE (1), ABCF (3), and ABCG (5). Members of the ABCB subfamily can be referred to as the “MDR-ABC transporters” and members of ABCC subfamily can be referred to as the “MRP-ABC transporters.” Table 1 shows some characteristics of the ABC transporters discussed in the text. For a detailed classification of human ABC transporter genes, the reader is referred to the following Web sites: http://www.gene.ucl.ac.uk/nomenclature/genefamily/abc.html; http://nutrigene.4t.com/humanabc.htm.
SUBFAMILY ABCB (MDR-ABC TRANSPORTERS)
Nomenclature and General Characteristics
The human ABCB subfamily (MDR-ABC transporters) consists of 11 members, ABCB1 through ABCB11. One pseudogene ABCB10P has also been reported. Among these, ABCB1 (MDR1), ABCB4 (MDR2/3), and ABCB11 (BSEP) are the largest proteins, consisting of about 1280 (both ABCB1 and ABCB4) and 1320 (ABCB11) amino acids. The proteins contain 12-transmembrane segments (α-helices) and are localized in the apical part of the plasma membrane. The ABCB1 (MDR1) protein with a molecular weight (MW) of 170-kDa, is also referred to as P-glycoprotein (P-gp) and the gene is designated as MDR1. Human ABCB1 (MDR1) is found in the epithelia of many tissues, such as intestine, liver, kidney, blood-brain barrier, testis, placenta, and lung; ABCB4 (MDR2/3) and ABCB11 (BSEP) are found mainly in hepatocytes. In contrast, ABCB2 (TAP1) and ABCB3 (TAP2) are located in endoplasmic reticulum, and ABCB6, ABCB7, and ABCB10 are located in mitochondria. The size of these transporters is somewhat different, varying between ∼650 and ∼850 amino acids, depending on the specific transporter. Accordingly, these proteins are predicted to contain six to eight transmembrane segments. ABCB2 (TAP1) and ABCB3 (TAP2) form heterodimers and are involved in antigen presentation (hence the name TAP or
ABCB11, also known as sister of P-glycoprotein (SPgp), is the bile acid export pump (BSEP) at the hepatocyte canalicular (apical) membrane. It is able to transport primary and secondary bile acids, such as taurocholate, glycocholate, cholate, taurochenodeoxycholate, tauroursodeoxycholate, etc., with high affinity (Gerloff et al. 1998). In contrast, sulfated bile salts, such as taurolithocholate sulfate, are not transported by BSEP (Akita et al. 2001). Recently, Hayashi et al. (2005b) studied the transport properties of human BSEP (hBSEP) and rat Bsep (rBsep) using membrane vesicles from HEK293 cells infected with recombinant adenoviruses, containing hBSEP or rBsep cDNA. They reported that the biliary bile salts of human differ from those of rat in containing a greater proportion of glycine conjugates and taurolithocholate 3-sulfate (TLC-S), and interstingly TLC-S was significantly transported by human BSEP but hardly transported by rat Bsep. By mediating ATP-dependent bile acid secretion across the canalicular membrane of hepatocytes, BSEP limits intracellular bile acid concentrations and also maintains the enterohepatic circulation of bile acids. In humans, mutation in the ABCB11/BSEP gene results in type 2 progressive familial intrahepatic cholestasis (PFIC2), a disease that is characterized by interruption in hepatic bile acid secretion (Strautnieks et al. 1998). However, not all mutations in BSEP result in PFIC2 (Hayashi et al. 2005a). Recent evidence also indicates that multidrug resistance protein 3 (MDR3, ABCB4) might be a more important risk factor than BSEP in the development of intrahepatic cholestasis of pregnancy (ICP) (Pauli-Magnus et al. 2004).
Humans have two MDR genes (MDR1 and MDR3, the latter is also called MDR2/3) whereas rodents have three Mdr genes—Mdr1a, Mdr1b, and Mdr2. The P-gp encoded by human MDR1 and mouse Mdr1a/1b functions as drug efflux transporter, whereas human MDR3 and mouse Mdr2 encode phospholipidtranslocating P-type ATPases (flippases).
General Structural Features of MDR-ABC Transporter
The general structural features of ABC transporters have been elucidated based on the study of its most well-characterized member, the P-gp. A typical MDR-ABC transport protein is composed of two halves that share a high degree of sequence similarity. Each half consists of one hydrophobic transmembrane domain (TMD) and one hydrophilic nucleotide-binding domain (NBD), also called nucleotide-binding fold (NBF), which is located at the cytoplasmic face of the membrane. Thus, the entire protein contains four distinct domains: two highly hydrophobic TMDs and two hydrophilic NBDs. Each NBD is associated with one TMD; NBD1 being associated with TMD1 and NBD2 with TMD2. Each TMD typically contains six-transmembrane segments. Thus, the protein contains 12 transmembrane segments (α-helices) and are localized in the apical part of the plasma membrane (Figure 1A ). Computer-assisted algorithms show that the two TMDs of P-gp fold uniquely and are joined together by a ∼75–amino acid linker region (Rao and Nuti 2003).
Each NBD contains three characteristic motifs called Walker A, Walker B, and an ABC signature (C) motif, located just upstream of the Walker B site. Figure 1B shows these motifs and above each motif is the respective amino acid sequence found in human MDR1/P-gp. The Walker A and Walker B sites are separated by a 90–120-amino acid-long spacer region and they extensively hydrogen bond with the bound ATP (Hyde et al. 1990; Higgins and Linton 2004). The ABC signature motif of NBD1 remains positioned opposite to the Walker A sequence of NBD2 in a so-called sandwich configuration. As a result, the ATPs exist in a sandwiched position between these adjacent Walker A and ABC signature motifs. The Ser residue in each ABC signature motif appears to be necessary for such interaction between the Walker A and ABC signature sequences and it probably interacts with the γ-phosphate of ATP (Leslie, Deeley, and Cole 2005). Sequence homology between the two halves of P-gp indicates that this protein originated by duplication of a primordial gene (Chen et al. 1990).
Three-Dimensional/Crystal Structure of MDR-ABC Transporter
The three-dimensional structure of a complete MDR-ABC transporter was first reported by Rosenberg et al. (1997). It was a 25-Å resolution structure determined by electron diffraction. The source of the P-gp protein was Chinese hamster ovary (CHO) cells. Using x-ray crystallography, Chang and Roth (2001) determined the first crystal structure of a complete MDR-ABC transporter at 4.5-Å resolution. The structure is that of the MsbA transporter from Escherichia coli (Eco-MsbA). Eco-MsbA is an ABC transporter that functions as a lipid flippase, transporting lipid molecules from the inner layer of the cell membrane to the outer layer. It is homologous to two human proteins implicated in multidrug resistance, one of which (MDR3) is itself a lipid flippase. Eco-MsbA is also closely related to LmrA from Lactococcus lactis. LmrA functions as a multidrug transporter. Eco-MsbA and LmrA are both half-size ABC transporters functioning as homodimers (Borst and Elferink 2002). Most recently, Rosenberg et al. (2005) determined the three-dimensional structure of Chinese hamster P-gp at a resolution of ∼8 Å obtained from two-dimensional crystals of P-gp trapped in the nucleotide-bound state. Reyes and Chang (2005) also described the 4.2-Å x-ray crystal structure of MsbA in complex with transition state mimic adenosine diphosphate (ADP), vanadate, and lipopolysaccharide (LPS).
The Rosenberg structure (25-Å resolution) of 1997 depicts that P-gp has a large central chamber within the membrane, and it is closed at the cytoplasmic side of the membrane. There is an opening from this chamber to the lipid phase. In the Chang-Roth crystal structure of Eco-MsbA (4.5-Å resolution), two polypeptide chains pack together to form a homodimer that forms a cone-like transmembrane domain, with the top of the cone pointing out of the cell. There is a central chamber that allows free access of substrates from the cytoplasmic side, while excluding molecules from the outer leaflet. The substrate to be transported binds to the substrate-binding cavity within the chamber, the chamber closes, and the substrate undergoes flip-flop. Flipping of the substrate triggers structural rearrangement of the chamber and the substrate is expelled.
The most recent report of Chinese hamster P-gp by Rosenberg et al. (2005) raises some interesting issues regarding P-gp structure. This study shows that in cross-section, the P-gp molecule has an overall clover-leaf shape. Conformational change of the protein opens out a central cavity in the TMD region, and such conformational changes mediate transport of substrates. According to the authors, data from the current study are in agreement with their earlier findings on Chinese hamster P-gp. However, their present study also reveals that the packing of the two halves of the molecule in Chinese hamster P-gp, with respect to each other, is very different from the tapering conical arrangement of Eco-MsbA homodimer reported by Chang and Roth. It remains to be seen whether ABC transporters will display a wide range of folds for the TMDs or whether some structural consensus will emerge.
Previous structures of MsbA in the absence of nucleotide and substrate had revealed a cytoplasmic accessible chamber. In the Reyes and Chang (2005) structure, a “flipping” of the accessibility of this chamber to the extracellular side of the membrane was suggested as a result of conformational change and movement of the TMDs on substrate binding. Lipopolysaccharide was found to bind the membrane-exposed sides of the protein at the dimer interface comprised of transmembrane segments TM1, TM5, and TM6 from one monomer and TM2 from the other monomer. The authors concluded that the structure suggests a model of substrate “flipping” where the sugar head groups of the lipopolysaccharide molecules are sequestered and then “flipped” in the internal chamber while the hydrophobic tails of the lipid are dragged through the bilayer.
Genomic Organization and Single-Nucleotide Polymorphisms (SNPs) of MDR1 Gene
Genomic Organization
Organization of human MDR1 gene was reported by Chen et al. (1990). The MDR1 gene contains 29 exons numbered −1 to 28. The numbering of exons reflects the fact that the MDR1 gene can be transcribed from two different promoters, an upstream promoter and a downstream promoter, the latter being preferentially expressed in most cell lines. The upstream promoter is found at the beginning of exon −1, and the downstream promoter is located within exon 1. Because the downstream promoter is located within exon 1, two parts of exon 1 that flank the downstream promoter have been further named exon 1a and exon 1b. The portion of exon 1 located 5′ to the downstream promoter is designated exon 1a, and the other portion of exon 1 located 3′ to the downstream promoter is designated exon 1b. The ATG translation initiation codon is located within exon 2. Thus the protein-coding sequence of MDR1 gene comprises 27 exons, 14 of which encode the first half and 13 encode the second half of the protein. There are 28 introns, 26 of which interrupt the protein-coding sequence.
The human MDR1 gene does not have a TATA box in the promoter (unlike the murine homolog) but instead contains an initiator element (Inr) defined by the consensus Py-Py-A(+1)-N-(T/A)-Py-Py. In the absence of a TATA box, initiator elements direct basal transcription and also ensure accurate transcriptional initiation. Transient transfection studies reveal that the sequence between −6 and +11 bp is sufficient for proper initiation of transcription (Van Groenigen, Valentijn, and Bass 1993). The promoter of human MDR1 gene has numerous transcription regulatory elements, such as GC-box, Y-box (inverted CCAAT element) that interacts with the trimeric transcription factor NF-Y, p53 element, AP-1 element, CAAT element, C/EBP element, HSE (heat shock element), MEF1 (MDR1 promoter-enhancing factor 1), and others. Most of these cis-acting response elements are within the first 300-bp upstream (–300) with respect to the transcription start site (Labialle et al. 2002). There is a steroid and xenobiotic response element (SXRE), also called pregnane X receptor response element (PXRE), located in the enhancer of the MDR1 gene at about −8 kb. The SXRE/PXRE contains a cluster of DR4 motifs (
Single-Nucleotide Polymorphisms (SNPs)
In a recent review of human MDR1 gene polymorphism, Marzolini et al. (2004) described 29 SNPs that had been reported in the literature. Recently, Allabi et al. (2005) reported 3 more novel SNPs, bringing the total number of reported MDR1 SNPs to 32. Of the 29 SNPs described by Marzolini et al., 19 are located in the coding region and 11 of them are nonsynonymous, that is, they result in amino acid changes in the protein. Two synonymous SNPs (C1236T in exon 12 and C3435T in exon 26) and a nonsynonymous SNP (G2677T, Ala893Ser) in exon 21 were found to be in linkage disequilibrium, thereby forming a SNP haplotype. This haplotype occurred in 62% of European Americans and 13% of African Americans (Kim et al. 2001). Thus, the SNP haplotype C1236T-G2677T-C3435T (i.e., T1236-T2677-T3435) is frequent in European Americans whereas the SNP haplotype C1236C-G2677G-C3435C (i.e., C1236-G2677-C3435) is frequent in Africans (Kim et al. 2001; Ozawa et al. 2004).
The synonymous SNP in exon 26 (C3435T) was first shown to be associated with decreased expression of P-gp in the duodenum of subjects with the T allele (variant) compared to those with the C allele (wild-type) (Hoffmeyer et al. 2000). However, subsequent studies showed that the decreased expression may be due to the nonsynonymous SNP in exon 21 (G2677T/A) (Kim et al. 2001). Interestingly, the SNPs in exon 21 at position 2677 can result in two distinct amino acid changes, namely, Ala893Ser (G2677T) and Ala893Thr (G2677A). Marked differences in the allele frequency of the exon 26 (C3435T) SNP were observed between the African and the Caucasian-Asian populations. The allelic frequency of C allele (wild-type) is high in African populations (∼80%) compared to that in Caucasian-Asian populations (∼ 45–55%) (Ameyaw et al. 2001). Because the C allele was initially found to be associated with a higher expression of P-gp, it was hypothesized that the observed higher frequency of the CC genotype in Africans may have resulted from a selective advantage offered by this genotype against gastrointestinal tract infections (Marzolini et al. 2004).
Recently, Allabi et al. (2005) sequenced exons 2, 7, 10, 11, 12, 14, 17, 21, 26, and the surrounding intronic regions of the MDR1 gene using genomic DNA from 111 Beninese subjects to examine various SNPs. They identified 12 SNPs in MDR1 gene, 3 of which were novel: IVS9-57delA, IVS9-8T→ A, 1662G→C (exon 14). The most common SNP was IVS14+38A→G, with a frequency of 49%. The authors concluded that ABCB1 (MDR1) exonic SNPs are less frequently observed in Beninese population (West Africans) than in African Americans.
MDR1 Substrates and Transport Mechanism
Table 2 shows a partial list of drugs transported by P-gp. For a comprehensive list of P-gp substrates and inhibitors, the reader is referred to the reviews by Dantzig, de Alwis, and Burgess (2003), Dietrich, Geier, and Oude Elferink (2003), Fromm (2004), and Marzolini et al. (2004). Substrates transported by P-gp include metabolic products, lipids and sterols, and drugs and other xenobiotics. A variety of pharmacologically distinct drugs used in cancer chemotherapy, hypertension, allergy, infection, immunosuppression, neurology, and inflammation are P-gp substrates. The precise transport mechanism of P-gp for drugs and xenobiotic is unknown. However, there are at least three models worth mentioning: the classical model, the hydrophobic vacuum cleaner model, and the flippase model (Ramakrishnan, 2003). Figure 2A shows a diagrammatic representation of the hydrophobic vacuum cleaner model and Figure 2B shows the flippase model.
The classical model depicts that the two TMDs of P-gp are organized into a pore-forming arrangement so that P-gp acts by actively expelling drugs from the cytoplasm to the extracellular location (Borst and Schinkel 1997).
According to the hydrophobic vacuum cleaner model, P-gp can pump substrates into the extracellular medium either from the intracellular compartment or when the substrates are still in the lipid bilayer of the plasma membrane, by recognizing them as foreign to the membrane. Thus, drugs and xenobiotics can be detected and expelled as they enter the lipid bilayer of the plasma membrane in the manner of a hydrophobic vacuum cleaner (Higgins and Gottesman 1992; Gottesman and Pastan 1993).
According to the flippase model, the substrate-binding site of P-gp is located close to the cytoplasmic face of the molecule and is accessible from the inner face of the plasma membrane. The hydrophobic portion of a substrate molecule is oriented toward the hydrophobic core of the membrane, and the charged portion toward the polar cytosolic face of the membrane. Thus, the model depicts that the substrate diffuses laterally until encountering and binding to a site on the MDR1 protein that is in the inner leaflet of the lipid bilayer. The protein then flips the substrate into the outer leaflet using energy obtained from ATP hydrolysis by the ATPase activity of MDR1 (Higgins and Gottesman 1992). Once in the exoplasmic face, the substrate diffuses out into the aqueous phase on the outside of the cell. Support for the flippase model of transport by MDR1 comes from MDR3, a homologous protein with phospholipid flippase activity. The flippase model is the currently favored mechanism of P-gp action (Ramakrishnan 2003). The Chang and Roth model of Pg-P depicts that the central chamber of Pg-P allows free access of substrates from the cytoplasmic side and excludes molecules from the outer leaflet. The substrate binds to the substrate-binding cavity within the chamber. Flipping of the substrate triggers structural rearrangement of the chamber and the substrate is expelled. Most biological membranes possess an asymmetric bilayer distribution of phospholipids. Endogenous enzymes expend energy to maintain that arrangement by promoting the rate of phospholipid translocation.
Such phospholipid translocation from one membrane layer to the other (such as, inner membrane to outer membrane) is called flip-flop (Figure 2C ). The translocation (or flip-flop) of phospholipids in an abiotic bilayer membrane is a very slow process with a half-life of hours to days. Flip-flop rates in these artificial systems are strongly dependent on the composition of the polar head group and less dependent on the length of the acyl chains. In biological plasma membranes, the flip-flop rates are thought to be modulated by specific membrane-bound enzymes. Flip-flop is also known to occur in the membranes of the endoplasmic reticulum and mitochondria (Boon and Smith 2002). However, no specific biogenic membrane flippases have been identified or purified so far (Menon, Watkins, and Hrafnsdottir 2000). Recently, Contreras et al. (2005) demonstrated that increase in ceramides concentration in membrane preparations promotes transbilayer (flip-flop) lipid motion in membranes. Ceramides or N-acyl sphingosines constitute the basic structure of sphingolipids such as sphingomyelin. In the plasma membrane, ceramides are released from the hydrolysis of sphingomyelin by the action of neutral sphingomyelinase (Huwiler et al. 2000). Contreras et al. studied the phospholipid flip-flop in erythrocyte membranes by using pyrene-labeled phospholipid analogues. They found that addition of egg ceramide to the membrane or enzymatic generation of ceramide by adding spingomyelinase induced transbilayer redistribution (flip-flop) of phospholipid analogues. Such flip-flop was further influenced by the bilayer lipid composition. The authors concluded that when one membrane leaflet becomes enriched in ceramides, they diffuse toward the other leaflet. This is counterbalanced by lipid movement in the opposite direction, so that net mass transfer between monolayers is avoided.
So far, at least four drug-binding/transport sites have been identified in transmembrane segments TM1, TM5/6, and TM11/12 (Loo and Clarke 1997, 1999; Ueda, Taguchi, and Morishima 1997; Martin et al. 2000). Although the structure-activity relationship for P-gp substrates is yet to be clearly defined, the majority of substrates are amphipathic hydrophobic organic cations with a MW ranging between 300 and 2000 Da, and they contain at least two planar rings. Substrates can be cationic, anionic, or uncharged (Gottesman and Pastan 1993; Ueda, Taguchi, and Morishima 1997). The lipophilicity and the number of hydrogen bonds appear to be directly correlated to the affinity of the compound for P-gp (Ecker et al. 1999). The transmembrane amino acid sequence of P-gp shows a high degree of hydrogen bond donor side chains, and these are thought to interact with the hydrogen bond acceptor sites (electron donor sites) of the substrates (Seelig 1998). Based on some experimental evidence, it has been hypothesized that P-gp binds substrates through an induced-fit mechanism, where the size and shape of the substrate changes packing of the TM segments. Each substrate would cause specific shifts in the different TM segments responsible for its binding (substrate-induced fit), allowing common residues to be involved in the binding of diverse substrates. Direct evidence for this hypothesis has been provided by cysteine-scanning mutagenesis and oxidative cross-linking methodology, in which different substrates promoted cross-linking between different TM segments (Loo, Bartlett, and Clarke 2003).
MDR1 Expression and its Regulation
MDR1 gene in humans is expressed in small intestine, liver, kidney as well as in the brain (Figures 3 to 6). In blood-brain barrier, P-gp is known to be expressed apically in capillary endothelial cells (Figure 6A ) so that it transports various substrates back into the blood and prevents their entry into the brain. However, Rao et al. (1999) showed that P-gp is expressed apically in choroid plexus epithelium (Figure 6B ). As a result, it transports substrates back into the cerebrospinal fluid (CSF), preventing their entry from the CSF into the blood. In general, P-gp is located in blood-tissue barriers so that it prevents the entry of xenobiotics in the tissue, thereby providing protection to the tissue from potential toxic insults. P-gp protein is localized on the apical side of the cell, that is, on the canalicular membrane of hepatocytes in liver, on the brush-border membrane of enterocytes in the intestine, on the brush-border membrane of the proximal tubular cells in the kidney, on the brush-border membrane of the choroid plexus epithelium, and on the luminal side of the blood-brain barrier (BBB) endothelium. Such localizations of P-gp suggest that this transporter plays a significant role in the bioavailability, distribution, and excretion of drugs and other xenobiotics. Functional inactivation of mouse Mdr1a resulted in the marked sensitivity of these mice to the neurotoxic effects of ivermectin, an antiparasitic agent, which is a substrate for P-gp (Lankas, Cartwright, and Umbenhauer 1997). The absence of functional P-gp at the blood-brain barrier in those mice resulted in more than 80-fold higher brain accumulation of ivermectin, resulting in neurotoxicity. Similarly, absence or inhibition of P-gp in the placenta of pregnant mice has been shown to cause rapid penetration into the fetus of P-gp substrates administered to the pregnant mother (Borst and Elferink 2002). Study of the Mdr-null mice models has shown that Mdr1a is the major drug-transporting P-gp and the only form expressed in brain capillaries, intestinal tract, and placenta, whereas Mdr1b is expressed only in the liver and the kidney. Therefore the two P-gp isoforms in mice appear to fulfill the same function as the single MDR1-encoded P-gp in humans.
P-gp plays an important role in the defense of the body against xenobiotics. In the gut mucosa, it prevents the entry of toxins into the body; in the blood-brain barrier, placental trophoblasts, testis, and bone marrow, it provides protection of vital body parts; in the gut, liver, and kidney, it helps to eliminate toxins from the body. The downside of P-gp’s ability to pump xenobiotics out is that it interferes with the delivery of drugs to target tissues. The development of potent P-gp inhibitors with low toxicity has opened up new ways to overcome this undesirable interference of P-gp with medical treatment (Borst and Elferink 2002).
Many drug substrates of P-gp are also substrates of drug-metabolizing enzymes, particularly, cytochrome P450 3A4 (CYP3A4). Because both MDR1 and cytochrome P4503A4 genes contain the PXRE enhancer sequence, many xenobiotics that activate PXR can induce both MDR1 and CYP3A4 (Geick, Eichelbaum, and Burk 2001; Choudhuri and Valerio 2005). Such coordinate regulation results in transport-biotransformation coupling of drugs and other xenobiotics. The broad and overlapping substrate specificities of CYP3A4 and MDR1, and their coordinate regulation and expression in organs such as the liver and intestine, led to the hypothesis that these two proteins evolved to protect the host organism from exposure to environmental or dietary toxins. Interestingly, both genes are also located on the same chromosome in close proximity, 7q22.1 (CYP3A4) and 7q21.1 (MDR1). However, the extent of their substrate overlap is not complete because there are substrates for CYP3A4 (i.e., midazolam) that are not transported by P-gp. Conversely, digoxin, fexofenadine, and talinolol are P-gp substrates that do not interact to a significant extent with CYP enzymes including CYP3A4 (Marzolini et al. 2004).
MDR1 gene expression can be induced by a variety of stimuli, such as ultraviolet (UV), sodium butyrate, retinoic acid, phorbol esters, many chemotherapeutics, etc. Scotto and colleagues have shown that all these divergent stimuli converge on a region of MDR1 promoter that they have termed “enhancesome,” which spans across the GC-rich elements, and the CCAAT box, and therefore includes binding sites for transcription factors like Sp1, NF-Y, etc. (Scotto 2003).
The Effect of Polymorphisms on MDR1 Expression and Function
The effect of SNPs on MDR1 expression is at the moment debatable because the initial observation of Hoffmeyer et al. (2000) was not corroborated by a number of subsequent studies. The study by Hoffmeyer et al. reported a two-fold reduction in P-gp expression in duodenum of healthy Caucasian subjects with the 3435T allelic variant. Consistent with this finding, increased digoxin plasma concentrations were reported in those subjects, suggesting greater intestinal drug absorption due to low intestinal P-gp levels. But many recent findings are not consistent with this earlier finding, such as a report of no change in digoxin clearance between Caucasian subjects carrying the 3435T allele or the wild-type 3435C allele (Gerloff et al. 2002) and report of elevated expression of MDR1 mRNA in Japanese subjects carrying the 3435T allele (Nakamura et al. 2002) or Caucasian subjects carrying the 3435T allele (Siegmund et al. 2002).
Recently, Putnam et al. (2005) investigated the effect of rifampin-mediated induction of P-gp on the clearance of dicloxacillin and its metabolite, and also the effect of C3435T SNP on the induction of P-gp in healthy human volunteers. The authors found that rifampin treatment significantly increased the clearance and the mean absorption time of the P-gp substrate, dicloxacillin, but they did not observe a correlation between expression of the MDR1 C3435T variant and the pharmacokinetics of dicloxacillin. Similar conflicting data have been reported for other P-gp drug substrates, such as fexofenadine, cyclosporine, and tacrolimus. Conflicting data have also been reported for the effect of exon 21 SNP (G2677T/A) (Marzolini et al. 2004).
Recently, SNP haplotype analysis of the MDR1 gene in a Japanese population revealed lower renal clearance of irinotecan, an anticancer agent, and its active metabolite SN-38, among patients bearing a specific haplotype C1236T-G2677T-C3435T (Ozawa et al. 2004). Saitoh et al. (2005) examined the impact of SNP C3435T on nelfinavir pharmacokinetics and the response to highly active antiretroviral therapy (HAART) in 71 human immunodeficiency virus (HIV)-1–infected children. The frequencies of C/C, C/T and T/T genotypes were 44% (n = 31), 46% (n = 33), and 10% (n = 7), respectively. Children with the C/T genotypes had higher 8-h postdose concentration and lower clearance rate for nelfinavir, compared to those with the C/C or T/T genotypes. No compensatory polymorphisms were observed between MDR1 and CYP genotypes.
Therefore, studies have thus far failed to demonstrate a definitive, reproducible correlation between the frequently observed SNPs in MDR1 gene and alterations in P-gp transporter function that can result in altered clearance of drugs/xenobiotics.
SUBFAMILY ABCC (MRP-ABC TRANSPORTERS)
Nomenclature and General Structural Features
The human ABCC subfamily (MRP-ABC transporters) consists of 12 members, ABCC1 through ABCC12. Of these, ABCC 1–6 (MRP1–6), ABCC10 (MRP7), ABCC11 (MRP8), and ABCC12 (MRP9) are transporters (Dean and Allikmets, 2001; Kruh and Belinsky, 2003; Human ATP-Binding Cassette Transporters [URL: http://nutrigene.4t.com/humanabc.htm]); the other three members are the cystic fibrosis transmembrane conductance regulator (CFTR/ABCC7), which is an ion channel, and the two sulfonylurea receptors, ABCC8 (SUR1) and ABCC9 (SUR2) (Dean and Allikmets 2001; Kruh and Belinsky 2003). Recently a truncated member (ABCC13) has also been reported (Yabuuchi et al. 2002). Of these, ABCC1 (MRP1), ABCC2 (MRP2/cMOAT), ABCC3 (MRP3), ABCC6 (MRP6), and ABCC7 (CFTR) are the larger MRPs containing greater than 1500 amino acids (except the CFTR, which contains 1480 amino acids). MRP2 with a MW of 190 kDa and 1545 amino acids is the largest MRP. MRPs have three hydrophobic transmembrane domains (TMDs) which are also referred to as membrane-spanning domains (MSDs). The domains are designated TMD0 (MSD0) through TMD2 (MSD2) from N-terminal to C-terminal. TMD0 has 5 transmembrane segments whereas TMD1 and TMD2 each has 6 transmembrane segments, making a total of 17 transmembrane segments. Unlike P-gp, which has an intracellular N-terminal end, in the larger MRPs with 17 transmembrane segments, the N-terminal end is extracellular (Figure 7A ). There are two NBDs at the cytoplasmic face of the membrane, NBD1 is associated with TMD1 and NBD2 with TMD2 (Borst and Elferink 2002; Kruh and Belinsky 2003; Chan, Lowes, and Hirst 2004). In contrast, some other MRP-ABC transporters, such as ABCC4 (MRP4), ABCC5 (MRP5), ABCC8 (SUR1), and ABCC9 (SUR2) contain two TMDs (TMD1 and TMD2) encompassing 12 transmembrane segments along with two NBDs. In these transporters, the N-terminal end is intracellular like that in P-gp (Figure 7B ). An important feature of the ABCC subfamily members is that they lack 13 amino acids between the Walker A and Walker B motifs in NBD1; these are present in NBD2 as well as in both NBDs of most other eukaryotic ABC proteins (Hipfner, Deeley, and Cole 1999). Figure 7C shows the amino acid sequence of Walker A, Walker B, and the ABC Signature (C) motifs of human MRP1 and MRP2.
Genomic Organization and Single-Nucleotide Polymorphisms (SNPs) of MRP1 and MRP2 Genes
Genomic Organization
Grant et al. (1997) analyzed the exon-intron organization of the human MRP gene, which is MRP1. Human MRP1 gene contains 31 exons and 30 introns. Exon 1 contains the 5′-UTR and first 48 bp of the coding sequence; exon 31 contains a 109-bp coding sequence and the rest is 3′-UTR. All 30 introns follow the GT … AG rule. Each of the NBDs is encoded by three exons; exons 16 to 18 for NBD1 and exons 28 to 30 for NBD2. The proximal promoter of human, mouse, and rat MRP1 gene is GC-rich (>70%). The first 100 nucleotides in human, mouse, and rat promoters show significant sequence conservation; they contain several overlapping GC-boxes that can bind transcription factors Sp1 and Sp3. Beyond the first 100 nucleotides the rodent and human sequences become quite different. However, the rat and mouse flanking regions retain more than 70% sequence identity over the 3.5 kb of sequence compared. In all three (human, rat, and mouse) MRP1 gene promoters, there is no TATA or CAAT element (Kurz, Cole, and Deeley 2001; Muredda et al. 2003). As a comparison, human MRP3 gene promoter (GenBank accession number AF481958) is also a TATA-less promoter and is less GC-rich than the MRP1 promoter. Luciferase assays with a series of truncated 5′-flanking regions indicate that the −23/–127 region is important for MRP3 expression. The binding of Sp1 also appears to be involved in MRP3 transcription (Takada, Suzuki, and Sugiyama 2000).
Human MRP2/cMOAT gene contains 32 exons and 31 introns. Exon 1 contains the 5′-UTR and the first 33 bp of the coding sequence, whereas exon 32 contains a 130-bp coding sequence and the rest of it is 3′-UTR. All 31 introns follow the GT … AG rule. Each nucleotide binding domain is encoded by 3 exons; exons 16 to 18 for NBD1 and exons 29 to 31 for NBD2. (Toh et al. 1999; Tsujii et al. 1999).
Single-Nucleotide Polymorphisms
In order to identify single-nucleotide polymorphisms (SNPs) within the MRP1 gene with potentially important functional significance, Wang et al. (2005) examined 13 SNPs across the MRP1 genomic region in ∼480 individuals from five different populations, namely, Chinese, Malays, Indians, European Americans, and African Americans. SNP frequencies in the African-American population were significantly different from that in other non-African populations at most SNP loci. The authors found relatively high haplotype diversity in MRP1 gene and weak linkage disequilibrium of various haplotypes. Interestingly, SNP G3140C in exon 23 (Cys1047Ser; in cytoplasmic loop 6) occurs at a relatively high frequency (4.5%) in African-American population; it has not been found in any Caucasian individual screened so far. Another SNP G2012T in exon 16 (Gly671Val) was found to have no functional consequences, whereas exon 10 SNP G1299T (Arg433Ser; in the cytosolic interface of transmembrane segment 8) resulted in significant decrease in the transport ability of many organic anions but increased doxorubicin resistance (Conseil, Deeley, and Cole 2005). For a detailed list of MRP1 SNPs, the reader is referred to a recent review by Conseil, Deeley, and Cole (2005).
SNP analysis of MRP2 was performed in 48 healthy Japanese subjects by Ito et al. (2001e). Six types of SNPs were identified, of which C24T (promoter), a missense mutation G1249A (Val417Ile), and a silent mutation C3972 (Ile1324) were frequently observed with an allelic frequency of 18.8%, 12.5%, and 21.9%, respectively. Other SNPs identified with low frequency (∼1%) were C2302T (Arg768Trp), C2366T (Ser789Phe), and G4348A (Ala1450Thr), all being missense mutations. Of these, the G4348A (Ala1450Thr) is located in the ABC signature motif, and hence may impact MRP2 function.
Suzuki and Sugiyama (2002) performed additional SNP analysis using 72 cell lines established from surgically dissected tumors from 72 Japanese individuals, and confirmed the preponderance of the three above-mentioned frequent SNPs. They found a total of 4 SNPs in the promoter region and 23 SNPs in exons. However, functional analysis of these SNPs remains to be performed.
Hirouchi et al. (2004) investigated the effect of SNPs on MRP2 function by studying the expression and cellular localization of the wild-type and three reported frequent SNP variants of MRP2 protein, in LLC-PK1 cells. They found that the most frequently observed amino acid substitution (Val417Ile) as well as the two less frequently observed amino acid substitutions (Ser789Phe and Ala1450Thr) may not affect drug disposition function of MRP2. However, mutant MRP2 protein with two less frequently observed amino acid substitutions (Ser789Phe and Ala1450Thr) showed significantly lower expression levels compared to the wild-type MRP2. Thus, the lower overall transport level of these two mutant forms of MRP2 protein may be a function of the lower expression level rather than lower transport ability.
Mutations of MRP2 Gene in Dubin-Johnson Syndrome (DJS)
Because Dubin-Johnson syndrome (DJS) in humans is caused by the loss of MRP2 function, the role of various mutations in the MRP2 gene in the development of DJS has been investigated by a number of laboratories. Paulusma et al. (1996) first reported that TR− rats, an animal model of DJS, are defective in cMOAT (MRP2) organic anion transporter and found that a single nucleotide (nucleotide 1179) deletion in the gene resulted in reduced mRNA abundance and absence of the protein. Using immunostaining, Kartenbeck et al. (1996) reported the absence of the MRP/cMOAT protein in the liver of a DJS patient. Subsequently, Paulusma et al. (1997) cloned MRP2 cDNA from a DJS patient and found that a C → T mutation at codon 1066 created a stop codon (TGA) from a CGA codon, which encodes arginine, thereby creating a truncated protein.
Morimasa Wada and coworkers (Wada et al. 1998; Toh et al. 1999) studied a total of 12 Japanese DJS patients (Toh et al. 1999). They analyzed the cMOAT cDNA sequence in these patients and compared them to the gene. They found four major types of mutations; two of these were missense mutations and the other two were deletion mutations. The two missence mutations were C2302T transition in exon 18, resulting in amino acid replacement Arg768Trp in the active transport family signature motif; and A4145G transition in exon 29, resulting in amino acid substitution Gln1382Arg in the position within the ABC signature motif at the C-terminal end. This latter missence mutation (Gln1382Arg) is also found in the cystic fibrosis transmembrane conductance regulator gene (CFTR) in patients with cystic fibrosis (Dörk et al. 1994), suggesting that this mutation could affect the function of MRP2/cMOAT. The two deletion mutations were a 168-nucleotide deletion from nucleotides 2272 to 2439 (2272del168) in exon 18, and a 147-nucleotide deletion from nucleotides 1669 to 1815 (1669del147) in exon 13. Both of these are at the exon-intron boundary and they involve a T → A (GT → GA) transition in the consensus splice donor site (Wada et al. 1998; Toh et al. 1999).
Tsujii et al. (1999) identified other mutations in the MRP2 gene from two other DJS patients. In one patient, the nonsense mutation was in codon 1066 in exon 23 (CGA to TGA; Arg → STOP; in cytoplasmic loop 6). This is a known mutation in DJS patients and it leads to premature termination of MRP2 protein synthesis (Paulusma et al. 1997). In another patient there was a 6-nucleotide deletion affecting codons 1392 to 1394 in exon 30. This deletion results in the loss of two amino acids (Arg and Met) that are located between the Walker A motif and the ABC signature motif in the NBD2 of the MRP2 protein.
Additional missense mutations Ile1173Phe and Arg1150His have been found in Iranian Jewish and Moroccan Jewish DJS patients, respectively. By examining the probenecid-sensitive efflux of carboxyfluorescein in MRP2 cDNA-transfected HEK293 cells, it was found that the function of these two mutants was almost completely abolished (Mor-Cohen et al. 2001).
Shoda et al. (2003) analyzed mutations of the MRP2 gene of a 39-year-old Japanese woman with Dubin-Johnson syndrome. They detected two novel mutations, C298T and C3928T, and two SNPs, C3972T and C24T (SNPs are mutations that occur in at least 1% of the population). One of the two mutations (C298T) is in the transmembrane domain. The other mutation (C3928T) and the SNP (C3972T) are concentrated in the second ATP-binding cassette.
Studies on the SNPs in other MRPs, such as MRP3, MRP4, and MRP5, have been done on a limited scale; their population distributions are yet to be widely studied. Thus far, no significant functional consequences have been found between the SNPs studied and transport ability of these transporters. Thus, more detailed work is needed to characterize these mutations and investigate their functional consequences. For a description of the SNPs in MRP3, MRP4, and MRP5 genes, the reader is referred to a recent review by Conseil, Deeley, and Cole (2005).
Substrates, Expression, and Regulation of Different MRPs
Table 3 compares the K M for transport of some MRP substrates (mostly endobiotics), and Table 4 provides a partial list of substrates for MRP1 and MRP2, two of the best studied MRPs. For a comprehensive list of various MRP substrates, the reader is referred to Konig et al. (1999), Leslie, Deeley, and Cole (2001), Suzuki and Sugiyama (2002), Dietrich et al. (2003). Transport ability of MRPs has been studied using both in vivo and in vitro experiments. Most of the in vitro evidence has been acquired through transport assays using inside-out membrane vesicles prepared from cells expressing elevated levels of the transporter. In vivo studies of the pharmacological and toxicological significance of MRP1 and MRP2 have been aided by the generation of Mrp1 −/− knockout mice and the availability of natural Mrp2-deficient rats, such as TR−, Groningen-Yellow (GY), and Eisai hyperbilirubinemic (EHBR) rats (Leslie, Deeley, and Cole 2001). Figures 3 to 6 show the subcellular localization of various MRP proteins in intestine, liver, kidney, blood-brain barrier, and choroid plexus, respectively.
MRP1
Substrates. The substrate selectivity of MRP1 is quite broad. MRP1 is an important transporter of a variety of organic anion conjugates, such as glutathione conjugates (leukotriene C4 or LTC4; 2,4-dinitrophenyl-S-glutathione or DNP-SG), glucuronides (bilirubin glucuronide; estradiol-17-β-glucuronide or E217β G), as well as glutathione disulfide (GSSG), unconjugated anionic drugs, and dyes. It also transports amphipathic neutral/basic drugs and oxyanions. The oxyanions arsenite and antimonite and the unconjugated drugs are cotransported with GSH.
The transport of many substrates of MRP1 that can occur without GSH, is enhanced in the presence of GSH, such as estrone-3 sulfate, etoposide glucuronide, etc. (Hipfner, Deeley, and Cole 1999; Leslie, Deeley, and Cole 2001; Kruh and Belinsky 2003). Because free glutathione is required for MRP1-mediated efflux, MRP1-conferred resistance is subject to inhibition by buthione sulfoximine (BSO), an agent that blocks the synthesis of glutathione by inhibiting γ-glutamylcysteine synthase (Schneider et al. 1995; Versantvoort et al. 1995; Zaman et al. 1995).
MRP1 is the resistance factor for anthracyclines (such as doxorubicin and daunorubicin), epipodophyllotoxins (such as teniposide and etoposide), vinca alkaloids (such as vincristine and vinblastine), and camptothecins (such as topotecan) (Kruh and Belinsky 2003). In addition to being the resistance factor for many anticancer drugs, MRP1 is normally involved in glutathione homeostasis and in inflammatory processes involving LTC4. In vitro studies indicated that MRP1 is the efflux pump for LTC4 from mast cells (Bartosz et al. 1998).
Of all the MRPs, MRP1 has the highest affinity for LTC4. However, recently, Rigato et al. (2004) demonstrated that unconjugated bilirubin (UCB) has a 10-fold higher affinity for MRP1 than LTC4, and UCB can significantly inhibit LTC4 transport by MRP1 (mean K M = 0.01 μM for UCB and 0.1 μM for LTC4). The authors suggested that MRP1 probably plays an important role in protecting cells from bilirubin toxicity (mean V max for UCB transport by MRP1 = 100 pmol/mg protein/min). A reverse scenario of GSH depeletion by MRP1 resulting in cellular oxidative stress has also been suggested (Leslie, Deeley, and Cole 2003). Reduced glutathione (GSH) is transported by MRP1 with very low affinity (K M =>5 mM), and certain MRP1 substrates are transported in association with it. However, GSH transport is markedly increased by many bioflavonoids, such as apigenin, genistein, quercetin, of which apigenin was the most effective. In the presence of apigenin, the apparent K M and V max of MRP1 for GSH were 116 μM and 666 pmol/mg protein/min, respectively. Depletion of cellular GSH store, mediated by bioflavones, may result in oxidative stress, with toxicological consequences (Leslie, Deeley, and Cole 2005).
Localization, Expression, and Regulation. MRP1 is expressed in most tissues and is localized to the basolateral membrane of epithelial cells. In murine kidney, Mrp1 is expressed at the basolateral membrane of cells of Henle’s loop and the cortical collecting duct, but is not expressed in proximal tubular cells (Van Aubel et al. 2000). In contrast, in humans, MRP1 is expressed in basolateral membranes of proximal tubular cells (Schaub et al. 1999; Cha et al. 2001). Thus it pumps its substrates into the interstitial space rather than excreting them into the bile, urine, or gut (Borst et al. 1999, 2000). MRP1 may help in tubular resorption of organic anions back into the capillary blood.
MRP1 has a protective role in some tissues (Borst et al. 1999). Its levels are relatively high in lung, testes, and kidney and it is also expressed in several blood-organ barrier contributing to protected sites (Cole et al. 1992; Borst et al. 1999, 2000). Choudhuri et al. (2003) showed that Mrp1 mRNA is very highly expressed constitutively in the choroid plexus of rats, much more than in kidney. Such strategic location of Mrp1 is consistent with the reports that Mrps transport xenobiotics from cerebrospinal fluid (CSF) to blood. Using a combination of Mrp1/Mdr1a/Mdr1b triple-knockout mice, along with Mdr1a/Mdr1b double-knockout mice, Wijnholds et al. (2000a) observed a 10-fold increase in etoposide concentrations in CSF in the triple-knockout mice compared with the double-knockout mice, after intravenous (i.v.) administration of the drug. This strongly suggests a role of Mrp1 in etoposide transport from CSF to blood. In other words, Mrp1 functions as a barrier to drug penetration into CSF. MRP1 is known to be expressed in many cancers and it probably plays an important role in cancers that are treated with natural product drugs (Kruh et al. 1995).
Muredda et al. (2003) showed that human MRP1 promoter binds only Sp1 but not Sp3 whereas rodent Mrp1 promoter binds both Sp1 and Sp3. Transient transfection experiments demonstrated that the conserved GC-boxes of all three promoters are the major determinants of basal activity. However, the human promoter retains substantial activity even in the absence of the conserved GC-boxes. Wild-type p53 protein could suppress the expression of both human and rodent MRP1 promoters in a dose-dependent fashion, probably by interacting with Sp1 and neutralizing its promotional effect (Muredda et al. 2003; Scotto 2003). Loss of p53 expression is correlated with increased MRP1 expression in colorectal cancer (Fukushima et al. 1999). Manohar et al. (2004) showed that human MRP1 gene expression can also be upregulated by the MYCN oncogene in childhood cancer neuroblastoma (NB). MYCN is a member of the MYC family and it dimerizes with MAX. Such MYCN-induced increase in MRP1 mRNA and protein levels is accompanied by increased drug resistance.
Structural Features Important for Transport. Deeley, Cole, and colleagues extensively investigated the effects of replacement of various amino acid residues on MRP1 expression and function (Zhang et al. 2003; Situ et al., 2004; Koike et al. 2004). Substitution of 18 proline residues in both nonmembrane and transmembrane regions of TMD1 and TMD2 showed that all MRP1-Pro mutants except Pro1113Ala were expressed in human embryonic kidney cells at levels comparable with wild-type MRP1. Proline mutations in or proximal to the transmembrane (TM) helices resulted in the reduction of transport ability. In contrast, mutation of Pro1150 in the cytoplasmic loop 7 (CL7) linking transmembrane helices 15 and 16 resulted in a substantial increase in 17β-estradiol-17-β-(
Situ et al. (2004) investigated the effect of replacement of six ionizable amino acid residues at the cytoplasmic interface of transmembrane helices 13 to 17 on MRP1 function and expression. Opposite charge substitutions of Arg1202 and Glu1204 in TM helix 16 reduced protein expression by >80%. In contrast, same charge or neutral charge substitutions had no effect. Organic anion transport by Glu1204Leu substitution was substantially reduced although substrate binding by the transport-compromised Glu1204Leu mutant remained intact. In contrast, replacement with same charge residue (Glu1204Asp) did not have any effect, except for GSH. In these mutatnts, GSH transport activity was less than 25% of that of the wild type. The authors concluded that the Glu1204 serves a dual role in membrane expression of MRP1 and a step in its catalytic cycle subsequent to initial substrate binding.
Zhang et al. (2003) showed that mutations of Asp1084 that eliminate the carboxylate side chain, markedly decreased resistance to all drugs tested, as well as transport of both E217βG and LTC4, despite the fact that LTC4 binding was unaffected. The authors concluded that Asp1084 may be critical for interaction between the cytoplasmic loop connecting TM13 and TM14 and a region of NBD2 between the conserved Walker A and ABC signature motifs.
MRP2
Substrates. MRP2 was originally cloned from rat liver as cMrp (Büchler et al. 1996) and from human liver as the canalicular multispecific organic anion transporter (cMOAT) (Paulusma et al. 1996). Studies in hepatocytes have shown that MRP2 is involved in exporting a variety of both conjugated and unconjugated anionic compounds into bile (Figure 8A ). The substrate specificity of MRP2 is similar to that of MRP1, and it includes glutathione conjugates such as LTC4 and 2,4-dinitrophenyl-S-glutathione, bilirubin glucuronides, and a number of drugs and conjugated drug metabolites (Jedlitschky et al. 1997; Kawabe et al. 1999; Madon et al. 1997).
In vitro studies using different cell lines, such as MDCKII, Caco-2, HepG2, etc., stably transfected with MRP2 showed that the drug resistance profile of MRP2 is similar to that of MRP1 for the anthracyclines, epipodophyllotoxins, vinca alkaloids, and camptothecins (Koike et al. 1997; Cui et al. 1999; Kawabe et al. 1999). Additionally, MRP2 confers resistance against cisplatin, which forms toxic glutathione conjugates in the cell (Ishikawa and Ali-Osman 1993). Overexpression of MRP2 is associated with cisplatin-resistance in cell lines (Taniguchi et al. 1996). Recently, using quantitative reverse transcriptase–polymerase chain reaction (RT-PCR), Hinoshita et al. (2000) also reached the same conclusion. They analyzed data on 45 patients with colorectal cancer. They found MRP2 mRNA expression was low in normal colorectal mucosa and specifically increased in cancerous regions compared with noncancerous regions. Of the anticancer agents prescribed for patients with colorectal cancers, including doxorubicin, mitomycin C, cisplatin, 5-fluorouracil, etoposide, and a camptothecin derivative, MRP2 mRNA expression was significantly associated with resistance to cisplatin. The authors concluded that MRP2 may be important for resistance to cisplatin treatment in colorectal cancer. MRP2 also transports various constituents in food, such as dietary flavonoids quercetin 4′-β-glucoside (Walgren et al. 2000) and sulphate conjugates of the tea flavonoid epicatechin (Vaidyanathan and Walle, 2001).
Both MRP2 and BSEP are involved in bile acid secretion. Whereas BSEP transports a broad range of nonsulfated bile acids, MRP2 transports only sulfated bile acids such as taurochenodeoxycholate-3-sulfate and taurolithocholate-3-sulfate, but not monoanionic bile acids such as taurocholate (Akita et al. 2001).
Localization, Expression, and Regulation. MRP2 is expressed mainly in liver, intestine, and kidney tubules and is localized to the apical membranes of these tissues (Büchler et al. 1996; Kruh and Belinsky 2003; Chan, Lowes, and Hirst 2004). In spite of the similarity with MRP1 in the range of substrates transported, MRP2 serves distinct functions in the body because of differential expression and subcellular polarity. MRP2 appears to play an important role in detoxification by secreting metabolites into bile. The unique importance of MRP2 is evidenced by patients with Dubin-Johnson syndrome (Paulusma et al. 1997 in Chan, Lowes, and Hirst 2004) and in EHBR (Esai hyperbilirubinaemic rat) and TR−/GY rat strains congenitally lacking Mrp2 (Paulusma and Oude Elferink 1997; Büchler et al. 1996; Paulusma et al. 1996; Yamazaki et al. 1995).
Expression of MRP2 can also cause toxicity by concentrating toxic metabolites in bile, causing damage to the cholangiocytes. MRP2 can also increase the excretion of GSH, thereby increasing the potential for cellular oxidative stress in liver. For example, Dietrich et al. (2001a) showed that MRP2 can transport compounds that form reversible complex with GSH, such as alpha-naphthylisothiocyanate (ANIT) and arsenite, into bile. The complexes dissociate in the bile, releasing GSH and freeing the toxin. GSH is excreted in bile, and the released toxin reenters the cell. Upon reentry into the cell, the toxin again complexes with GSH, transported by MRP2, and the cycle is repeated. Such recycling of toxins results in a net disproportionate efflux of GSH by MRP2, which, in turn, may have toxicological consequences. Intestinal MRP2 plays the role of gatekeeper in the gut. For example, the dietary carcinogen 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) is an abundant heterocyclic amine formed during the frying and cooking of meat. PhIP is mutagenic and carcinogenic in rodents and has also been implicated in intestinal and breast cancer in humans. Studies comparing PhIP absorption and excretion in wild-type and TR−(Mrp2-deficient) rats show that Mrp2 prevents the absorption of PhIP in the gut. Mrp2 is also capable of biliary excretion of the parent compound and all its major metabolites. Transport of PhIP by both rat Mrp2 and human MRP2 is GSH-dependent (Dietrich et al. 2001b).
Characterization of the 5′-flanking region of the rat Mrp2 gene promoter revealed binding sites for pregnane X receptor (PXR), constitutive androstane receptor (CAR), and farnesoid X receptor (FXR) (Tanaka et al. 1999; Kast et al. 2002). Kast et al. (2002) found that all three nuclear receptors could activate rat Mrp2 gene expression via a novel response element (ER-8) in the proximal promoter. The ER-8 is a 26-bp sequence (–401 to −376, relative to the translation start site) that contains two copies of the hexad 5′-AGTTCA-3′ organized as an everted repeat (ER) with 8-bp spacer (hence ER-8). MRP2 mRNA and protein expression in human duodenal enterocytes is induced by the antibiotic rifampin which also induces CYP3A4 and P-gp (Fromm et al. 2000). Johnson and Klaassen (2002) investigated the effects of treatment with ligands for various nuclear receptors, such as aryl hydrocarbon receptor (AhR), CAR, PXR, and peroxisome proliferator activated receptor α (PPARα) (known inducers for CYP1A, CYP2B, CYP3A, and CYP4A, respectively) on rat hepatic Mrp2 mRNA and protein expression. Whereas PXR ligands (including dexamethasone and spironolactone) increased, and PPAR α inducers (including clofibric acid and perfluorodecanoic acid) reduced hepatic Mrp2 protein expression, none of these compounds affected Mrp2 mRNA expression, suggesting that regulation of rat Mrp2 protein expression by such compounds is mediated, at least in part, by post-transcriptional mechanisms. This conclusion is further supported by a recent publication by Jones et al. (2005). They analyzed Mrp2 synthesis, expression, and degradation in control female, 19- and 20-day pregnant, and PCN-treated rats, using in vivo metabolic-labeling studies with [35S]cysteine/methionine or [14C]NaHCO3, polysomal distribution analyses, and RNase protection assays (RPA). The authors concluded that differences in the degradation of Mrp2 protein cannot explain the post-transcriptional regulation of Mrp2 in control, pregnant, and PCN-treated rats. Rather, the observed difference in protein synthesis suggests an intrinsic role for the translational regulation of rat Mrp2 protein. These findings are also consistent with earlier studies showing discrepancies between the levels of intestinal Mrp2 mRNA and protein expression in rat (Gotoh et al. 2000; Mottino et al. 2000).
Structural Features Important for Transport. By replacing each of 13 basic residues with alanine around transmembrane regions (TMs) 6 to 17 of human MRP2 using site-directed mutagenesis, Ryu et al. (2000) demonstrated that TM6, TM9, TM16, and TM17 of human MRP2 are involved in transport activity. These TM segments in MRP2 correspond to functionally equivalent TM1, TM11, and TM12 segments of P-gp. In rat Mrp2, however, TM 11, TM14, and TM16 are important for transport activity (Ito et al., 2001a, 2001b, 2001c). Recently, Ito et al. (2001d) showed that in rat Mrp2, substitution of Arg at position 586 with Leu and Ile, and substitution of Arg at position 1096 with Lys, Leu, and Met, resulted in the acquisition of taurocholate transport activity, while retaining transport activity for glutathione and glucuronide conjugates. The authors concluded that the presence of the cationic charge at Arg586 and Arg1096 in rat Mrp2 prevents the transport of taurocholate.
MRP3
Substrates. MRP3 is also an organic anion transporter and basolateral in position like MRP1 (Kool et al. 1999b). Human MRP3 is most closely related to MRP1 with 58% amino acid identity (Belinsky and Kruh 1999). MRP3 shares considerable substrate specificity with MRP1, MRP2, and BSEP, albeit with significantly lower affinity for substrates that are also transported by MRP1 and MRP2 (Hirohashi et al. 2000; Zeng et al. 2000).
MRP3 can not transport GSH. This explains why overexpression of MRP3 does not confer resistance to most anticancer drugs that are cotransported with GSH by MRP1 and MRP2. However, MRP3 can confer resistance to etoposide, teniposide, vincristine and methotrexate (MTX), although less effectively than MRP1 and MRP2 (Kool et al. 1999b; Zeng et al. 1999).
Human MRP3 can transport glutathione and glucuronide conjugates (Kruh and Belinsky 2003) but rat Mrp3 does not appear to transport glutathione conjugates (Hirohashi, Suzuki, and Sugiyama 1999; Zeng et al. 2000). MRP3 transports both unconjugated and conjugated bile acids. This includes monoanionic bile acids, such as taurocholate and glycocholate, which constitute a significant component of the bile acid pool in humans and rodents (Hirohashi, Suzuki, and Sugiyama 2000; Zeng et al., 2000). The ability of MRP3 to transport nonsulfated bile acids is a feature that is not shared by MRP1 or MRP2. MRP3 also transports bile acid conjugates, such as taurolithocholate-3-sulfate and taurochenodeoxycholate-3-sulfate (Hirohashi, Suzuki, and Sugiyama 2000; Zeng et al. 2000). Because MRP3 transports glucuronides, such as estradiol-17β-glucuronide, MRP3 may contribute to the efflux transport of toxic/carcinogenic agents detoxified by glucuronidation.
Localization, Expression, and Regulation. MRP3 is basolateral in position like MRP1 and is predominantly expressed in liver, kidney, and gut, particularly in ileum (Chan, Lowes, and Hirst 2004). Choudhuri et al. (2003) showed that the constitutive expression level of Mrp3 mRNA in rat choroid plexus is twice that of liver.
MRP3 proximal promoter contains several GC-rich elements (–21/–86) that interact with Sp1, and deletion of this region decreases MRP3 transcription (Takada, Suzuki, and Sugiyama 2000). The expression of MRP3 is upregulated in liver by cholestasis. Inokuchi et al. (2001) studied the transcriptional regulation of human MRP3 in human colon cancer cell line Caco-2. Using nuclear run-on assays they showed that human MRP3 expression is upregulated through de novo transcriptional increase following treatment with a number of bile acids, such as taurocholic acid, taurolithocholic acid, chenodeoxycholic acid (CDCA), and taurochenodeoxycholic acid (TCDCA). Transient transfection experiments with a series of 5′-deleted MRP3 promoter constructs driving luciferase reporter expression revealed the existence of putative bile acid–responsive region (–229/–138), whose expression could be upregulated by bile acids, such as CDCA. The α-1-fetoprotein transcription factor (FTF), also called liver receptor homolog-1 (LRH-1), was found to mediate bile acid-mediated transcriptional increase of human MRP3, through two FTF-response elements located in the bile acid–responsive region (Inokuchi et al. 2001).
Taken together, the fact that MRP3 is a basolateral efflux pump (it can transport bile acids and glucuronide conjugates), can be induced by bile acids, and is induced during cholestasis strongly suggest that MRP3 in liver plays a back-up role in secondary protective mechanism when normal excretory functions are compromised (Chan, Lowes, and Hirst 2004). In situations such as cholestasis, induced MRP3 can extrude bile acids and glucuronide conjugates including those of toxicants into sinusoidal blood, thereby preventing their accumulation in hepatocytes (Figure 8B ). However, as discussed below, it has been claimed that MRP4 may also play important role in such extrusion of bile acids during cholestasis.
Using tissues isolated from rats treated with known enzyme inducers that are ligands for various nuclear transcription factors (AhR, CAR, PXR, PPARα), Cherrington et al. (2002) showed that Mrp3 mRNA expression in liver, but not kidney or large intestine, was significantly increased by CAR activators, such as phenobarbital (PB).
Structural Features Important for Transport. Using site-directed mutagenesis, Ito et al. (2001d) showed that in rat Mrp3, substitution of Leu1084 with Lys (L1084K), but not with Val or Met, resulted in the loss of transport activity for taurocholate and glucuronide conjugates. Thus, the presence of neutral amino acids at this position of rat Mrp3 appears to be required for the transport of substrates because rat Mrp3 transports taurocholate.
MRP4, MRP5
Substrates. These are shorter MRPs that lack the TMD0. They are both organic anion pumps, but unlike MRP1-3 they transport cyclic and acyclic nucleoside monophosphates as well as nucleoside monophosphate analogs (Reid et al. 2003a). MRP4 is an organic anion transporter and it transports E217β G, methotrexate, reduced folate, cAMP, cGMP, and bile acids (Chen et al. 2001, 2002; Lai and Tan, 2002; Van Aubel et al. 2002; Zelcer et al. 2003). Rius et al. (2003, 2005a) showed that human MRP4 can mediate ATP-dependent cotransport of GSH and taurocholate. The K M values were 2.7 mM for GSH in the presence of 5 μM taurocholate. The rate of GSH transport in the presence of 5 μM taurocholate was seven-fold higher than that in controls. Human MRP4, in the presence of physiological concentrations of GSH, also has a high affinity for the taurine and glycine conjugates of the common natural bile acids as well as the unconjugated bile acid cholate. Four GSH co-substrates are taurochenodeoxycholate (K M = 3.6 μM), glycochenodeoxycholate (K M = 5.9 μM), tauroursodeoxycholate (K M = 7.8 μM), and glycoursodeoxycholate (K M = 12.5 μM). The K M values of the bile acids for MRP4 were in a similar range as those reported for the canalicular bile salt export pump BSEP. The authors concluded that under physiological conditions the sinusoidal MRP4 may compete with canalicular BSEP for bile acids, thereby playing a key role in determining the hepatocyte concentration of bile acids. The authors also suggested that in cholestatic conditions MRP4 may become a key pathway for the efflux of bile acids from hepatocytes into blood, a role ascribed to MRP3 as well. A recent study from the same group (Rius et al. 2005b) demonstrated that MRP4 and prostanoid synthesizing enzymes are coexpressed in human seminal vesicles, ureter, prostate, bladder, and corpus cavernosum. Transport studies showed that MRP4 is an efflux pump for prostaglandin E2 (PGE2) (K M = 3.5 μM), thromboxane B2 (TXB2) (K M = 9.9 μM), and prostaglandin F2 α (PGF2 α )(K M = 12.6 μM).
In hepatocytes, MRP4 is located basolaterally but in renal tubular cells, it is located apically (Van Aubel et al. 2005). Such differential location of a transporter is ideal if it has to transport a substrate from hepatocyte to blood in the liver, but from renal cells to the tubular lumen in the kidney. Using two sublines of HEK293 cells, one overexpressiong human MRP4 (HEK293/4.63), and the other stably expressing an MRP4-specific RNA interference (RNAi) construct (HEK293/RNAi-53) to knock-down MRP4 expression, Van Aubel et al. (2005) demonstrated that human MRP4 transports uric acid. The authors suggest that MRP4 in hepatocytes mediates hepatic export of urate into the circulation because of its basolateral localization, whereas MRP4 in renal tubular cells mediates excretion of urate into the tubular lumen because of its apical localization.
Because MRP4 can also transport bile acids, it remains to be demonstrated whether the membrane localization of MRP4 in liver and kidney may also play a role in the clearance of bile acids during cholestasis in the same manner suggested for uric acid.
Zelcer et al. (2003) showed that MRP4 also transports dehydroepiandrosterone 3-sulfate (DHEAS) with a higher affinity than other MRPs that have been shown to transport it (Table 3). The authors concluded that since the circulating level of DHEAS is among the highest steroid levels in human blood, a major function of MRP4 may be the transport of DHEAS and/or structurally related steroids.
Recently, human MRP4 was shown to confer resistance to camptothecins (topotecan, irinotecan, and SN-38) (Tian et al. 2005; 2006).
Both MRP4 and MRP5 confer resistance to the antiviral agent 9-(2-phosphonomethoxyethyl) adenine (PMEA), and they transport PMEA in an unmodified form. MRP4 also mediates substantial resistance against other acyclic nucleoside phosphonates, but MRP5 does not. Cells overexpressing MRP4 can also confer resistance against AZT, the nucleotide analog HIV drug (Borst and Elferink 2002). MRP5 mediates efflux of 2′, 3′-didehydro-2′, 3′-dideoxythymidine 5′-monophosphate (d4TMP), a pyrimidine-based antiviral, 2′, 3′-dideoxynucleoside and its phosphoramidate derivative alaninyl-d4TMP, and the anticancer purine analogs 6-mercaptopurine and thioguanine (Wijnholds et al. 2000b).
Methylation of thionucleotides can switch substrate affinity between MRP4 and MRP5. MRP4 appears to prefer methylated thionucleotides, but MRP5 prefers unmethylated thionucleotides (Borst and Elferink 2002). However, the affinity of MRP4 and MRP5 for nucleotide-based substrates is low (Reid et al. 2003a).
Localization, Expression, and Regulation. MRP4 is expressed on the apical membrane of human renal proximal tubular cells (Van Aubel et al., 2002.) but on the basolateral membrane of the tubuloacinar cells of the prostate (Lee, Klein-Szanto, and Kruh 2000). Human, mouse, and rat MRP4 is also expressed basolaterally in hepatocytes (Rius et al. 2003; Van Aubel et al. 2005). MRP5 is expressed in colon, liver, kidney, and skeletal muscle (Kool et al. 1997; McAleer et al. 1999), but its membrane localization is currently not clear. Nevertheless, human MRP5 is targeted to the basolateral membrane when stably transfected in MDCKII cells (Wijnholds et al. 2000b). Choudhuri et al. (2003) reported that the constitutive expression levels of Mrp4 and Mrp5 mRNA in rat choroid plexus is very high, significantly higher than that of liver, kidney and ileum.
MRP6 to MRP9
MRP6 is highly expressed in kidney and liver and with low expression levels in other tissues, such as duodenum, colon, brain, and salivary glands (Kool et al. 1999a; Zhang et al. 2000; Scheffer et al. 2002). Initial experiments utilizing overexpression of MRP6 in Chinese hamster ovary (CHO) cells indicate that MRP6 can transport many glutathione-conjugate substrates that are also transported by MRP1 to MRP3, such as etoposide, LTC4, and DNP-SG but not E217β G (Belinsky et al. 2002). These data suggest a role of MRP6 in drug transport in tissues where it is expressed; however, the importance of MRP6 in conferring drug resistance is yet to be determined. The correlation between MRP6 deficiency and the occurrence of the genetic disorder pseudoxanthoma elasticum, which is a rare autosomally inherited connective tissue disease, was a surprising discovery, but the cause-and-effect relationship is yet to be settled (Kruh and Belinsky 2003). Mrp6 in mice and rats is located on the basolateral membrane of hepatocytes and renal tubular cells (Borst and Elferink 2002).
Relatively little is known about the recently discovered members MRP7 to MRP9. MRP7 is expressed at low levels in liver, kidney, colon, spleen, stomach, and testis (Hopper et al. 2001). MRP7 appears to confer very high level of resistance against one class of anticancer drug, the taxanes (Hopper-Borge et al. 2004). This is an interesting finding because the only other ABC transporter that confers resistance against taxanes is P-gp. Chen et al. (2003a) showed that human MRP7 is able to transport Estradiol 17-β-
SUBFAMILY ABCG MEMBER 2 (ABCG2)
Nomenclature and General Structural Features
ABCG2 (also called BCRP/ABCP/MXR) is a half-size ABC transporter containing six transmembrane segments and one NBD at the N-terminal end, which is at the cytoplasmic side (Figure 1C ). Doyle et al. (1998) identified this protein in a subline of human breast carcinoma MCF-7 cells selected for resistance to the anthracycline doxorubicin in the presence of verapamil, an inhibitor of P-gp. The resultant multidrug-resistant subline was designated MCF-7/AdrVp. These cells were characterized by high mitoxantrone resistance and lower resistance to anthracyclines and camptothecins. Such resistance was found to be due to overexpression of this protein, which Doyle et al. named as breast cancer resistance protein (BCRP). BCRP was also independently cloned by two other groups and was named as ABCP (for human placenta-specific ATP-binding cassette gene) (Allikmets et al. 1998) and MXR (for mitoxantrone resistance protein) (Miyake et al. 1999). MXR was cloned from mitoxantrone-resistant S1-M1-80 human colon carcinoma cells.
The BCRP mRNA codes for a putative 655-amino acid- long protein (Allikmets et al. 1998; Doyle et al. 1998). Under denaturing conditions, the apparent MW of BCRP protein from a number of mitoxantrone-resistant human cancer cell lines was reported to be 72 kDa (Litman et al. 2002). Ishikawa et al. (2003) expressed human BCRP in Sf9 insect cell line and found an apparent MW of 65 kDa under denaturaing conditions, but the MW shifted to 130 kDa under nondenaturing conditions. This suggests that BCRP possibly functions as a homodimer.
Genomic Organization and Single-Nucleotide Polymorphisms (SNPs) of ABCG2 Gene
Genomic Organization
Genomic organization of human ABCG2 gene and its promoter characterization was described by Bailey-Dell et al. (2001). The gene spans more than 66 kb. It consists of 16 exons ranging from 60 to 532 bp, and 15 introns ranging from 587 bp to over 18 kbp. All splice junctions conform to the GT … AG rule. The translation start site is in exon 2. Within the first 500 nucleotides (up to the −500 nucleotide position) the promoter has a high GC content and it contains multiple Sp1 sites. There is no canonical TATA and CCAAT box. The promoter also contains AP-1 and AP-2 sites. Determination of transcriptional activity using partial promoter constructs in a luciferase reporter gene assay revealed that the region between nucleotides −1285 and −628 (with respect to the transcription start site) contains positive regulatory elements, whereas the region between nucleotides −628 and −312 contains negative regulatory elements. Thus, the authors designated the region 312 bp upstream of the transcriptional start site as the basal promoter.
Single-Nucleotide Polymorphisms
Many single-nucleotide polymorphisms (SNPs) and at least one indel (insertion-deletion) polymorphism in ABCG2 gene and their linkage disequilibrium analysis have been reported in the literature, but these studies mostly lack functional analysis.
Recently, Kondo et al. (2004) reported the effect of single nucleotide polymorphisms (SNPs) in ABCG2 gene on its localization, expression level, and transport activity of the BCRP protein. The cellular localization was identified using the wild-type and seven different SNP variants of BCRP protein (Val12Met, Gln141Lys, Ala149Pro, Arg163Lys, Gln166Glu, Pro269Ser, and Ser441Asn), following their expression in LLC-PK1 cells. Their expression levels and transport activities were determined using membrane vesicles from HEK293 cells expressing the wild-type and these same SNP variants. The authors concluded that Gln141Lys variant of the BCRP protein may be associated with a lower expression level, and Ser441Asn variant may lower both the expression level and cellular localization. It is possible that subjects with these polymorphisms may have lower expression of BCRP protein and, consequently, a reduced ability to export its substrates.
The effect of Gln141Lys variant of the BCRP protein function was also studied by other investigators, but the results are not consistent. Using DNA Pyrosequencing, de Jong et al. (2004) performed ABCG2 genotyping in American Caucasians, African Americans, Africans, and Han Chinese, as well as in European Caucasian patients treated with irinotecan. The purpose of the study was to evaluate the ethnic distribution and potential functional consequence of the C421A SNP on ABCG2 gene function (that results in the Gln141Lys amino acid change in BCRP protein). Significant differences in the frequency of this SNP were observed between these populations; the variant allele being most common in the Han Chinese population with a frequency of 0.34 (34%), but with a very low frequency (<1%) in the sub-Sahara African population. No significant changes in irinote-can pharmacokinetics were observed in relation to the ABCG2 C421A genotype, i.e., the Gln141Lys variant of the BCRP protein. The authors concluded that the contribution of this genetic variant may have been obscured by a functional role of other polymorphic proteins. This conclusion was not corroborated by a recently published study by Morisaki et al. (2005), which is described below.
SNP analyses of the ABCG2 gene by Morisaki et al. (2005) revealed three nonsynonymous SNPs that resulted in amino acid substitution of the BCRP protein; these were Val12Met, Gln141Lys, and Asp620Asn. When human embryonic kidney cells (HEK-293) were stably transfected with wild-type or variants of ABCG2/BCRP, it was found that cells transfected with wild-type Arg482 ABCG2 showed IC(50) values up to 1.2-fold to 5-fold higher than cells expressing comparable levels of Gln141Lys variant of ABCG2, suggesting that the Gln141Lys SNP affects drug transport, such as that of mitoxantrone, topotecan, SN-38, or diflomotecan. This suggests that the Gln141Lys SNP affects the transport efficiency of ABCG2 that may result in altered pharmacokinetics or drug-resistance profiles in clinical oncology.
The prevalence of C421A SNP in ABCG2 gene (resulting in Gln141Lys variant of the BCRP protein) has been recently confirmed by Kobayashi et al. (2005). Using polymerase chain reaction (PCR) and single-strand conformation polymorphism (SSCP) analysis followed by DNA sequencing, Kobayashi et al. (2005) studied the polymorphism in ABCG2 gene in Japanese, Caucasians, and African Americans. The authors found two SNP variants in ABCG2 gene with a relatively high frequency, G34A (18%) and C421A (35.5%). On the basis of haplotype analysis for the three major SNP variants (i.e., G34A, C376T, and C421A), four haplotypes were identified: G-C-C, G-C-A, A-C-C, and G-T-C, with their corresponding allelic frequencies being 46%, 35%, 18%, and 1%. They also found significantly higher frequencies of G34A and C421A in Japanese subjects. From these results, the authors concluded that genetic frequencies of BCRP variants appeared to be dependent on ethnicity.
In a study using nine BCRP-overexpressing sublines and two parental cell lines, Honjo et al. (2001) reported that in two lines a mutation at amino acid position 482 of BCRP was associated with its altered transport function. The wild-type version has an Arg at position 482 (Arg482) that is associated with a lack of rhodamine 123 transport ability, low resistance to anthracyclines, and high resistance to mitoxantrone and topotecan. Replacement of Arg482 by Gly (Arg482Gly) or Thr (Arg482Thr) was found to be associated with rhodamine transport ability and higher doxorubicin resistance. However, during SNP analysis of ABCG2 gene from 90 ethnically diverse individuals, the same group (Honjo et al. 2002) found no such SNPs at amino acid 482 (Arg482Gly or Arg482Thr) that they had previously reported in two cell lines. In fact, the authors concluded that ABCG2 is well conserved and that the described amino acid polymorphisms seem unlikely to alter transporter stability or function. Some other SNPs, including functionally relevant SNPs in the ABCG2 gene were also reported by other authors.
ABCG2/BCRP Substrates, Localization, and Expression
Table 5 shows a partial list of substrates transported by BCRP. For a more comprehensive list of BCRP substrates, the reader is referred to Krishnamurthy and Schuetz (2005) and Mao and Unadkat (2005). Like P-gp, BCRP does not require glutathione (GSH) for the transport of electroneutral amphipathic drugs.
In polarized cells (LLC-PK1 and MDCK-II), the murine Abcg2 is targeted to the apical membrane (Jonker et al. 2000). Northern analysis demonstrated high levels of expression of BCRP in the S1-M1-80 cells and in the human breast cancer subline, MCF-7 AdVp3000. Interestingly, the gene was found to be amplified 10 to 12-fold in the MCF-7 AdVp3000 cells, but not in the S1-M1-80 cells (Miyake et al. 1999). Using commercial human MTN (multiple tissue Northern) blot, human BCRP mRNA was shown to be expressed in many tissues, with the highest levels in the placenta (Doyle et al. 1998). Immunohistochemical studies by Maliepaard et al. (2001) revealed that human BCRP protein is expressed in multiple tissues including the apical membranes of placental syncytiotrophoblasts, small intestine, colon epithelium, liver canalicular membrane, breast ducts and lobules, capillary endothelium, placenta, blood-brain barrier, and lungs (Leslie, Deeley, and Cole 2005; Figures 3, 4, and 6). Such localizations strongly suggest that ABCG2/BCRP is involved in the efflux of intruding xenobiotics including its substrate drugs. BCRP is strongly induced in the mammary gland of mice, cows, and humans during lactation and it actively secretes clinically and toxicologically important substrates such as the dietary carcinogen PhIP, the anticancer drug topotecan, antiulcerative cimetidine, antibiotic nitrofurantoin, as well as the fluoroquinolone antibiotics (ciprofloxacin, ofloxacin, and nor-floxacin). In apparent contradiction with the detoxifying role of BCRP in mothers, this contamination of milk exposes suckling infants and dairy consumers to xenotoxins (Jonker et al. 2005; Merino et al. 2005, 2006; van Herwaarden and Schinkel 2006).
Wild-type BCRP has an Arg at position 482. In MCF-7/AdrVp3000 cell line, this has been mutated to Thr (R482T) and in S1-M1-80 cell line this has been mutated to Gly (R482G). Wild-type BCRP does not transport anthracyclines but the R482T mutant BCRP confers higher anthracycline resistance. In contrast, the antifolate drug methotrexate appears to be a substrate of wild-type BCRP only, which is a high-capacity but low-affinity transporter of methotrexate, with a K M value of ∼1 mM. Mutant BCRP is also able to efflux rhodamine 123, but wild-type BCRP does not. These data suggest that amino acid at position 482 is crucial for substrate selectivity of BCRP (Honjo et al. 2001; Chen et al. 2003b; Mao and Unadkat 2005).
Using branch DNA (bDNA) technique, Tanaka et al. (2005) compared the tissue-specific expressions of Abcg2 (Bcrp) mRNA in mice and rats. They found that rat Bcrp mRNA levels were high in intestine and male kidney, and intermediate in testes, whereas mouse Bcrp expression was maximum in kidney, followed by liver, ileum, and testes. Male-predominant expression of Bcrp mRNA was observed in rat kidney and mouse liver. Such gender-dependent expression was found to be influenced by the sex hormones. Gonadectomy and hypophysectomy experiments suggest that male-predominant expression of Bcrp mRNA in rat kidney was due to the suppressive effect of estradiol, and male-predominant expression of Bcrp mRNA in mouse liver was due to the inductive effect of testosterone. This is the first study demonstrating the possible effect of sex hormones on gender-predominant expression of Abcg2 in animal models.
TRANSPORTERS AS THERAPEUTIC TARGETS
The tissue distribution and cellular localization of these transporters have the potential to influence the toxicity of, and resistance to various drugs and environmental toxins. Although this efflux/extrusion activity seems to have evolved to protect the organism, this is also responsible for various patterns of drug resistance including resistance to many anticancer drugs. Thus, induction of expression of these transporters through oral administration of inducers, such as P-gp inducer rifampin, is associated with a corresponding decrease in AUC of the substrate drugs and increase in drug resistance (Greiner et al. 1999; Tsuji 2002). A decrease in AUC (area under the curve, or area under the plasma concentration curve) means lower drug concentration in plasma, because most of the drug can not get past P-gp and is transported out by it.
Use of appropriate inhibitors of P-gp (also called MDR modulators) has been a rational approach in decreasing drug extrusion and increasing the therapeutic effect, sometimes even with lower amounts of the drug. There are now three generations of P-gp inhibitors. The first generation inhibitors were actually developed for other indications such as the calcium channel blocker, verapamil, and the immunosuppressive agent, cyclosporin A (CsA). They are less potent, not selective, and have undesirable side effects. Second generation P-gp inhibitors were initially developed to reduce adverse side effects that were seen with the first generation of inhibitors. Some of these second generation inhibitors are also more potent inhibitors of P-gp than the first generation inhibitors. For example, PSC833 (valspodar) is a close structural analog of CsA that lacks the immunosuppressive effect and is also 10-times more potent than CsA in terms of its P-gp inhibiting ability. The
However, both first and second generation compounds inhibit two or more ABC transporters. For example, PSC833 inhibits P-gp, MRP2 and BSEP. Both VX-710 (biricodar) and MS209 inhibit P-gp and MRP1, whereas GF120918 (elacridar) inhibits both P-gp and BCRP. Thus, third-generation P-gp inhibitors were developed to be more potent and more specific. For example, LY335979 (zosuquidar·3HCl), XR9576 (tariquidar), and OC144-093 are about 10-fold or more potent than the others, as they inhibit P-gp in the low nanomolar range (30 to 100 nM), but do not inhibit MRP1. LY335979 also does not inhibit MRP2, MRP3, or BCRP at concentrations of 5 μM or greater (Dantzig et al. 2003). It has to be remembered, however, that the use of P-gp inhibitors has thus far not fully revolutionized cancer therapy and reverse multidrug resistance. This is partially because of unwanted accumulation of the drug in brain and other tissues, and consequent adverse side effects. In this context, the emerging siRNA technology appears be a promising additional tool for initial in vitro screening to find out whether a compound is substrate for P-gp or BCRP. Peng et al. (2004) showed that in the human leukemia cell line k562/A02, si-MDR1 significantly inhibited MDR1 mRNA as well as protein expression. Depletion of MDR1 expression correlated well with the increased retention of and sensitivity to daunorubicin by the cells. One base-pair mutated control (si-MDR1-Mut) did not show these effects.
DISCUSSION
The body burden of drugs and other xenobiotics is driven by a balance between their absorption and elimination. P-gp and BCRP affect the oral bioavailability of many drugs, as well as their accumulation in brain, fetus, testis, etc., and their disposition and elimination from the body. It now appears that P-gp, MRP1, MRP2, and BCRP can explain the phenomenon of multidrug resistance in all cell lines analyzed thus far (Borst and Elferink 2002). In addition, MRP1 to MRP4 and MRP6 to MRP8 are all lipophilic anion pumps that are also able to confer resistance to anticancer drugs. In general, these efflux pumps are not only responsible for multidrug resistance, but are also responsible for reducing the cellular burden of many other drug conjugates (such as acetaminophen glucuronide), various dietary and plant-derived compounds, pesticides, inorganic compounds, as well as endogenous substances (bile acids, hormones, etc.) (Chandra and Brouwer 2004; Leslie et al. 2005).
Studies conducted thus far show that there is a good cross-species concordance in the overall transport ability of the efflux transporters discussed here. Nevertheless, there are differences and many of them have been discussed in the text; for example, human MRP3 can transport glucuronide conjugates but rat Mrp3 does not. Likewise, taurolithocholate 3-sulfate (TLC-S) is significantly transported by human BSEP, but is hardly transported by rat Bsep. It has been shown that changing one crucial amino acid may alter substrate specificity of a transporter, such as the acquisition of anthracycline transport ability by changing the 482nd amino acid in BCRP from Arg to Thr. Because the amino acid sequences and the tissue expression of orthologous transporters are not identical, it is reasonable to contemplate that although the functions of various transporter classes have been largely conserved across different mammalian species, specific transporter functions evolved to adapt to and meet species-specific cellular metabolic needs. Therefore, specific studies with human transporters are still necessary, since the ultimate goal of transporter research is to translate the knowledge to facilitate drug development and therapy and improve the quality of human life.
A new area that needs detailed exploration is the natural polymorphisms in transporters that affect drug disposition or response to drugs and other xenobiotics. From the current state of our knowledge on transporter polymorphism-function relationship, it is clear that much remains to be learnt. Single-nucleotide polymorphisms (SNPs) found in our genes are thought to play a major role in the observed interindividual differences in their ability to transport, metabolize, and respond to various drugs, other xenobiotics, as well as nutrients. A SNP (“snip”) is a point mutation that adds variation to one’s genome. For a variation to be considered a SNP, it must occur in at least 1% of the population. SNPs generally occur about every 1000 base pairs in the genome, but may also occur more frequently in certain regions. The importance of SNPs in the etiology of disease is demonstrated by SNPs in apolipoprotein E (ApoE) gene and Alzheimer’s disease. Two SNPs in ApoE gene results in three possible alleles of the gene: E2, E3, and E4. Each allele differs by one DNA base, and the protein product of each gene differs by one amino acid. Individuals inheriting the E4 allele have a greater chance of getting Alzheimer’s disease whereas those inheriting the E2 allele are less likely to get the disease. Apparently, the change of one amino acid in the E4 protein alters its structure and function enough to make disease development more likely. In order to discover and systematically study SNPs, The SNP Consortium (TSC; URL: http://snp.cshl.org/) was established in 1999 as a collaboration of several companies and institutions to produce a public resource of SNPs in the human genome. To date, the consortium has discovered and characterized nearly 1.8 million SNPs (Choudhuri 2006).
At present, drug-drug and drug-xenobiotic interactions that occur at the level of transporters are probably underestimated and not properly characterized; but it may result in altered systemic concentrations of xeno- and endobiotics, leading to either increased toxicity of xenobiotics or decreased therapeutic efficacy of drugs. Despite many reports of SNPs in MDR1, MRP1, MRP2, and BCRP genes, their true impact on the expressions and functions of these transporters remains to be clearly defined. In order to achieve the ultimate goal, that is, personalized medicine, a detailed knowledge of transporter polymorphism-function relationship will be as important a tool as the knowledge of drug metabolizing enzyme polymorphism-function relationship. Another important area that needs more studies is the regulation of hepatic transport processes under pathological conditions. Overall, the recent explosion in transporter research will undoubtedly enhance our current understanding of hepatobiliary drug transport by characterizing the molecular basis of membrane translocation, elucidating the substrate binding site(s) of these proteins, and defining the mechanisms by which molecules inhibit and induce hepatic transport proteins. Application of this knowledge in the drug development process represents a new and exciting aspect of this discipline (Chandra and Brouwer 2004).
Footnotes
Figures and Tables
The opinions expressed in this article are the authors’ personal opinions and do not necessarily reflect those of FDA, DHHS, or the Federal Government.
This work was funded in parts by grants from the National Institute of Health (ES009716, ES009649, and ES013714).
1
Abbreviations: ABC, ATP-binding cassette; BCRP, breast cancer resistance protein; BSEP, bile salt export pump; cMOAT, canalicular multispecific organic anion transporter, commonly known as MRP2; MDR, multiple drug resistance; MRP, multidrug resistance protein; NBD, nucleotide binding domain; NBF, nucleotide binding fold; P-gp, p-glycoprotein; SPgp, sister of P-glycoprotein, commonly known as BSEP; SNP, single-nucleotide polymorphism; TAP, transporter associated with antigen presentation; TMD, transmembrane domain.