Xenobiotic, Bile Acid, and Cholesterol Transporters: Function and Regulation

1. Organic Anion Transporting Polypeptides.

Oatps/OATPs are members of the SLCO family (Table 1) and are responsible for the uptake of a wide range of substrates. Rat Oatp1a1 was the first member of the OATP family identified (Jacquemin et al., 1994) followed by rat Oatp2a1 (Kanai et al., 1995). OATP1A2 was the first human OATP to be cloned (Kullak-Ublick et al., 1995). The rapid and independent classification of multiple Oatp/OATP isoforms led to confusion regarding protein nomenclature. New nomenclature and classification of OATP isoforms according to evolutionary relationships and amino acid sequence identity were established in 2004 and approved by the HUGO Gene Nomenclature Committee (Hagenbuch and Meier, 2004). Oatps/OATPs with more than 40% amino acid sequence identity are members of the same family (OATP1, OATP2, OATP3…). Designation of isoforms to a particular subfamily (OATP1A, -1B, and -1C) requires more than 60% amino acid sequence identity. More than 15 rodent and 10 human isoforms have been described (Hagenbuch and Meier, 2004; Hagenbuch and Gui, 2008). In addition, an Oatp ortholog (Oatp1d1) was identified in skate liver and has been proposed to be an evolutionarily ancient precursor of mammalian OATP1B1/OATP1B3/Oatp1b2 responsible for uptake of cyclic peptides (Meier-Abt et al., 2007).

OATPs are integral membrane proteins predicted to contain 12 transmembrane helices with amino and carboxyl termini oriented to the cytoplasmic face (Noe et al., 1997; Hagenbuch et al., 2000; Mikkaichi et al., 2004). A large extracellular domain is thought to be located between transmembrane domains 9 and 10, with N-glycosylation sites present in extracellular loops 2 and 5 (Hagenbuch and Meier, 2003).

Expression of mouse Oatp isoforms varies among tissues (Fig. 1). Oatp1a1, -1a4, -1b2, and -2b1 are expressed in liver, whereas Oatp1a6, -3a1, and -4c1 are expressed in kidneys (Choudhuri et al., 2001; Cheng et al., 2005a). Levels of Oatp1a4 and -1c1 mRNA are high in the brain (Cheng et al., 2005a). Oatp1a5, -6b1, -6c1, and -6d1 mRNA are predominantly expressed in mouse testes. Within the testes, rat Oatp6b1 and -6c1 are expressed in Sertoli cells, spermatogonia, and Leydig cells (Suzuki et al., 2003). Oatp2a1, -4a1, and -5a1 are highest in the placenta (Cheng et al., 2005a).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

Tissue distribution of Oatp mRNA in mice and humans. Top, relative mRNA levels of transporters in mouse liver, kidneys, lung, stomach, duodenum, jejunum, ileum, large intestine, brain, gonads (testes and ovaries), and placenta are shown. Male (♂) mRNA is shown on the left, whereas female (♀) mRNA is shown on the right side of each box. References for mouse mRNA expression are included (Cheng et al., 2005a). Bottom, relative mRNA levels of transporters in human liver, kidneys, lung, heart, brain, testes, ovaries, placenta, and uterus are shown. Data for humans were obtained from GNF SymAtlas (http://symatlas.gnf.org/; now located at http://biogps.gnf.org). The GNF1H/MAS5 data set was accessed during September 2008.

The tissue distribution of the various OATP isoforms in humans also ranges from a single tissue to ubiquitous expression. For example, human OATP1B1 and OATP1B3 are primarily expressed in liver (Fig. 1) (Abe et al., 1999; Hsiang et al., 1999; König et al., 2000a,b). In contrast, OATP1C1, OATP2A1, OATP2B1, OATP3A1, and OATP4A1 mRNA can be detected in multiple tissues (Fig. 1) (Tamai et al., 2000a; Kullak-Ublick et al., 2001; Pizzagalli et al., 2002; Grube et al., 2006a).

There are similarities and differences in the tissue distribution of Oatp/OATP transporters between mice and humans (Fig. 1). For example, mouse Oatp1c1 and human OATP1C1 are highly expressed in brain (Pizzagalli et al., 2002; Cheng et al., 2005a). Likewise, both rat and mouse Oatp1a4 are most abundant in brain and liver (Noe et al., 1997; Cheng et al., 2005a). In contrast, Oatp4a1 is specifically expressed in mouse placenta, yet OATP4A1 is widely expressed in multiple human tissues (Tamai et al., 2000a; Cheng et al., 2005a). Therefore, differences in the tissue distribution of some Oatp/OATP isoforms are important when extrapolating from rodents to humans.

Cellular localization of Oatp/OATP isoforms can be apical or basolateral depending on the tissue and cell type (Table 2). In liver, Oatp/OATP isoforms are typically on the basolateral (also called sinusoidal) membrane of hepatocytes (Oatp1a1, -1a4, -1b2 and OATP1B1, -1B3, -2B1), although human OATP1A2 is localized to the apical surface of cholangiocytes (Bergwerk et al., 1996; Kakyo et al., 1999; Reichel et al., 1999; König et al., 2000a,b; Cattori et al., 2001; Lee et al., 2005a; Grube et al., 2006a). Within the liver, human OATP1B3 and rat Oatp1a4 are mostly confined to centrilobular hepatocytes, whereas human OATP1B1 is expressed uniformly throughout the lobule (Kakyo et al., 1999; Reichel et al., 1999; Ho et al., 2006b). Likewise, Oatps/OATPs are detected on both the apical and basolateral surfaces of the kidneys (apical OATP1A2 and Oatp1a1; basolateral Oatp4c1) and placenta (apical OATP4A1; basolateral OATP2B1) (Table 2) (Bergwerk et al., 1996; St-Pierre et al., 2002; Sato et al., 2003; Mikkaichi et al., 2004; Lee et al., 2005a; Grube et al., 2007).

Oatp/OATPs transport solutes with diverse characteristics. In general, Oatp/OATP substrates contain steroidal or peptide structural backbones and/or are anionic or cationic chemicals. Classes of pharmaceuticals transported by Oatp/OATPs include HMG-CoA reductase inhibitors (statins), angiotensin-converting enzyme inhibitors, angiotensin receptor II antagonists, and cardiac glycosides (Table 3). A number of endogenous chemicals, including thyroxine, steroid conjugates, bile acids, bilirubin, and prostaglandins are also substrates of Oatp/OATPs. It has long been recognized that chemicals secreted into bile are structurally larger than those excreted by the kidneys and may be due to selective extraction of bulky chemicals from the circulation by hepatic Oatp/OATPs. More recent evidence points to the influence of pH in the transport kinetics of Oatp/OATPs (Leuthold et al., 2009). Using Oatp/OATP-expressing oocytes and cultured cells, it was demonstrated that the transport activity of number of isoforms (with the exception of OATP1C1) is enhanced by low extracellular pH and that this flux is countered by bicarbonate efflux (Leuthold et al., 2009).

Although OATPs are typically considered uptake transporters, there are examples of bidirectional transport for various isoforms (Li et al., 2000; Mahagita et al., 2007). Transport of taurocholate and leukotriene C₄ by rat Oatp1a1 in oocytes is cis inhibited and trans stimulated by glutathione, suggesting that glutathione efflux provides a driving force for uptake (Li et al., 1998). Additional research demonstrates that Oatp1a4-mediated transport of taurocholate is bidirectional and stimulated by glutathione and its conjugates (Li et al., 2000). More recent research demonstrates that human OATP1B1 and -1B3 are similarly bidirectional facilitated diffusion transporters, but that glutathione is not a substrate or activator of their transport activity (Mahagita et al., 2007).

Human OATP1A2 transports and is inhibited by a large number of endogenous compounds as well as pharmaceuticals in in vitro systems (Table 3). Because of the promiscuity of this transporter, a number of drug-drug interactions have been proposed for OATP1A2. OATP1A2-mediated transport of fexofenadine is inhibited by antivirals, antifungals, antibiotics, and anticholesterol drugs (Cvetkovic et al., 1999). OATP1A2 can also transport the fluoroquinolone antibiotic levofloxacin and is inhibited by other quinolones (Maeda et al., 2007b). In addition to pharmaceutical inhibition, dietary constituents may also modulate drug transport by inhibiting OATP1A2 function. The grapefruit flavonoid naringin inhibits OATP1A2 uptake of fexofenadine in vitro, and thus grapefruit juice alters clinical fexofenadine pharmacokinetics (Dresser et al., 2002; Bailey et al., 2007; Glaeser et al., 2007). These data suggest that naringin probably interferes with the intestinal OATP1A2-mediated absorption of oral fexofenadine (Glaeser et al., 2007). Similar reports demonstrate in vitro inhibition of OATP2B1 transport by grapefruit and orange juices as well as other herbal extracts (Satoh et al., 2005; Fuchikami et al., 2006).

Rodent Oatp1a1 and -1a4 share similar substrates with OATP1A2, including unconjugated and conjugated bile acids, bromosulfophthalein, sulfated steroids, thyroid hormones, ouabain, β-lactam antibiotics, and fexofenadine (Cvetkovic et al., 1999; Reichel et al., 1999; Hagenbuch et al., 2000; Meng et al., 2002; van Montfoort et al., 2002; Nakakariya et al., 2008). Rat Oatp1a1 and Oatp1a4 can transport opioid peptides [d-Pen²,d-Pen⁵]-enkephalin and deltorphin II (Oatp1a1 only), which may be important in their transport across the blood-brain and blood-cerebrospinal fluid barriers (Kakyo et al., 1999; Gao et al., 2000). There are notable differences in transport by mouse and rat Oatp1a4; digoxin is a high-affinity substrate for rat Oatp1a4 but a low-affinity substrate for mouse Oatp1a4 (Noe et al., 1997; van Montfoort et al., 2002). Likewise, bromosulfophthalein is transported by mouse Oatp1a4 but not by the rat isoform (van Montfoort et al., 2002).

OATP1B1 and -1B3 are the primary OATP1B isoforms in human livers. Oatp1b2 is the rodent ortholog of OATP1B1 and -1B3. Because of the prominent expression of these transporters on the basolateral membrane of hepatocytes, they represent a critical mechanism for chemical uptake into liver. OATP1B1 and -1B3 exhibit overlapping and specific substrates (Table 3). Human OATP1B1 transports various statin drugs as well as thyroxine, taurocholate, and dehydroepiandrosterone sulfate (Hsiang et al., 1999). Both OATP1B1 and -1B3 can transport conjugated bilirubin; however, 1B1 appears to be more important for unconjugated bilirubin uptake (Cui et al., 2001). Using overexpressing oocytes, Oatp1b2 and OATP1B3 transport cholecystokinin, a gastrointestinal peptide that is released postprandially and stimulates gallbladder contraction, release of pancreatic enzymes, and intestinal motility (Ismair et al., 2001).

More recently, attention has been brought to the uptake mechanisms for hepatotoxic drugs, including bosentan and troglitazone. Bosentan and its active metabolite are substrates of OATP1B1 and -1B3 (Treiber et al., 2007). Likewise, OATP1B1 can transport and be inhibited by troglitazone sulfate (Nozawa et al., 2004b). It is hypothesized that inhibition of OATP1B1 by troglitazone sulfate may be a novel mechanism underlying idiosyncratic hepatotoxicity associated with this pharmaceutical (Nozawa et al., 2004b).

Clinical drug interactions may also occur at the level of OATP transporters. OATP1B1 transports pravastatin (Nakai et al., 2001). OATP1B1 transport is inhibited by fibric acid derivatives and may contribute to known drug-drug interactions, such as gemfibrozil-cerivastatin (Shitara et al., 2004; Yamazaki et al., 2005) and rifampin-atorvastatin (Lau et al., 2007). In contrast, rosuvastatin can be transported by a number of OATP isoforms, including OATP1B1, -1B3, -2B1, and -1A2 as well as rat Oatp1a1, -1a4, -1a5, and -1b2, probably reducing the chance of drug-drug interactions (Ho et al., 2006b). OATP1B1 also transports the active metabolite of the anticancer drug irinotecan (Nozawa et al., 2005). OATP1B1 and -1B3 transport the angiotensin-II blocker olmesartan (Nakagomi-Hagihara et al., 2006). Further work is necessary to better characterize clinical-relevant drug-drug interactions of these and other OATP1B1 and -1B3 substrates.

OATP1B1, -1B3, and rat Oatp1b2 participate in the uptake of rifampin (Tirona et al., 2003). Overexpression of OATP1B1 not only enhances rifampin transport but also its function as evidenced by enhancement of rifampin-stimulated pregnane X receptor gene transactivation (Tirona et al., 2003). Rifampicin can inhibit OATP1B1 and -1B3 transport (and be transported by them), whereas rifamycin SV can also inhibit OATP1A2 and -2B1 (Vavricka et al., 2002).

OATP2B1 is expressed in human placenta and, along with the breast cancer resistance protein, (BCRP; ABCG2) is probably responsible for transepithelial transport of sulfated steroids from the fetus to the mother during pregnancy (St-Pierre et al., 2002; Grube et al., 2007). Likewise, OATP2B1 is also expressed in ductal epithelial cells of the mammary gland (Pizzagalli et al., 2003). OATP2B1 prefers sulfate conjugates (estrone sulfate) rather than glucuronide conjugates (i.e., estradiol-17β-glucuronide) (Tamai et al., 2001b; Nozawa et al., 2004a). OATP2B1 can also transport dehydroepiandrosterone sulfate, the antihistamine fexofenadine, and the antidiabetic drug glibenclamide (Nozawa et al., 2004a; Satoh et al., 2005).

2. Organic Cation Transporters.

OCTs are polyspecific cationic transporters of the SLC22 family (SLC22A1–3) (Table 1). In 1994, Oct1 was the first member of the organic cation transporter family cloned from a rat kidney cDNA library (Gründemann et al., 1994). Human and mouse orthologs were soon cloned thereafter (Schweifer and Barlow, 1996; Gorboulev et al., 1997; Zhang et al., 1997). Subsequently, Oct2 and Oct3, two organic cation transporters with high homology to Oct1, were cloned and characterized in humans, rats, mice, and rabbits (Okuda et al., 1996; Gorboulev et al., 1997; Zhang et al., 1997; Kekuda et al., 1998; Urakami et al., 1998; Karbach et al., 2000). Oct3 is also called the extraneuronal monoamine transporter and participates in the uptake of extraneuronal monoamines in peripheral tissues and glia cells (also known as the uptake-2 system) (Gründemann et al., 1998; Wu et al., 1998). The membrane topology of OCT isoforms is predicted to be similar with 12 α-helical transmembrane domains with intracellular amino and carboxy termini (Burckhardt and Wolff, 2000). An extracellular loop between transmembrane domains 1 and 2 contains potential N-glycosylation sites (Burckhardt and Wolff, 2000). A large intracellular loop resides between transmembrane domains 6 and 7 and possesses predicted phosphorylation sites.

In mice, Oct1 mRNA expression is highest in kidneys and liver (Fig. 2) (Alnouti et al., 2006). Human OCT1 is primarily expressed in liver and to a lesser extent in other organs (Fig. 2) (Gorboulev et al., 1997; Zhang et al., 1997). Oct1/OCT1 proteins are localized to the basolateral membrane of centrilobular hepatocytes, proximal tubule cells, Sertoli cells, enterocytes, and in serotoninergic neurons of the small intestine (Table 4) (Meyer-Wentrup et al., 1998; Karbach et al., 2000; Muller et al., 2005; Maeda et al., 2007a). Prominent expression of OCT1 on the sinusoidal membrane of hepatocytes suggests that this transporter mediates the first step in hepatic excretion of cationic drugs.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

Tissue distribution of Oct, Octn, Oat, and Urat mRNA in mice and humans. Top, relative mRNA levels of transporters in mouse liver, kidneys, lung, stomach, duodenum, jejunum, ileum, large intestine, heart, brain, gonads (testes and ovaries), placenta, and uterus are shown. Male (♂) mRNA is shown on the left whereas female (♀) mRNA is shown on the right side of each box. References for mouse mRNA expression are included (Buist and Klaassen, 2004; Alnouti et al., 2006). Bottom, relative mRNA levels of transporters in human liver, kidneys, lung, heart, brain, testes, ovaries, placenta, and uterus are shown. Data for humans were obtained from GNF SymAtlas (http://symatlas.gnf.org/; now located at http://biogps.gnf.org). The GNF1H/MAS5 data set was accessed during September 2008.

Rodent Oct2 and human OCT2 mRNA are highest within the kidneys (Fig. 2) (Gorboulev et al., 1997; Slitt et al., 2002; Alnouti et al., 2006). Within renal proximal tubule cells, Oct2/OCT2 proteins are present on the basolateral membrane, which makes this transporter a key entry site for renally excreted cationic drugs (Table 4) (Karbach et al., 2000; Motohashi et al., 2002). Neurons of the human central nervous system have detectable OCT2 protein (Busch et al., 1998). Similar to Oct1, apical expression of Oct2 protein is seen in bovine olfactory mucosa and ciliated epithelial cells of rodent and human lungs (Lips et al., 2005; Kummer et al., 2006; Chemuturi and Donovan, 2007) (Table 4).

The tissue distribution of mouse Oct3 and human OCT3 is broader than Oct1/OCT1 and Oct2/OCT2. Oct3/OCT3 are expressed in many tissues with high levels in placenta, ovaries, and uterus (Fig. 2) (Kekuda et al., 1998; Wu et al., 1998, 2000b; Verhaagh et al., 1999; Slitt et al., 2002; Alnouti et al., 2006). Subcellular localization patterns for human OCT3 are cell-type specific. OCT3 protein is observed on basolateral (trophoblasts, renal tubule cells) and apical (enterocytes, Sertoli cells, ciliated lung epithelia) membranes (Table 4) (Lips et al., 2005; Muller et al., 2005; Sata et al., 2005; Kummer et al., 2006; Maeda et al., 2007a; Glube and Langguth, 2008) as well as in numerous regions of the rat brain (Vialou et al., 2004).

Oct/OCT transporters mediate the uptake of organic cations that are positively charged at physiological pH. OCTs are classified as uniporters and enhance cellular entry of chemicals by facilitated diffusion. OCT-mediated transport is electrogenic and independent from sodium (Koepsell and Endou, 2004). The primary driving force that determines the direction of translocation is the electrochemical gradient of the transported organic cation, typically an inside-negative membrane potential.

Substrates of Oct/OCT transporters have relatively low molecular weights and are hydrophilic organic cations with widely diverse molecular structures (Table 3). There is extensive overlap of substrate and inhibitor specificities among OCT1–3 from different species. Oct1/OCT1 orthologs from four species (rat, mouse, rabbit, and human) all transport tetraethylammonium. However, there are some differences in affinity and transport rates. In contrast to rabbit and human, rat and mouse Oct1 do not transport larger structural analogs (i.e., tetrapropylammonium and tetrabutylammonium) (Dresser et al., 2000).

Model compounds for Oct/OCT-mediated transport include tetraethylammonium, the neurotoxin 1-methyl-4-phenylpyridinium, and N¹-methyl-nicotinamide (Busch et al., 1996b; Gorboulev et al., 1997; Zhang et al., 1997; Kekuda et al., 1998; Urakami et al., 1998; Wu et al., 2000b). Pharmaceuticals have also been identified as Oct/OCT substrates and consist of the antidiabetic drug metformin (Kimura et al., 2005), the antiviral drugs acyclovir and zalcitabine (Takeda et al., 2002; Jung et al., 2008), the antineoplastic agent cisplatin (Ciarimboli et al., 2005b; Yokoo et al., 2007), the N-methyl-d-aspartate-receptor antagonist memantine (Busch et al., 1998), and the histamine H₂-receptor antagonist ranitidine (Bourdet et al., 2005). Biogenic amine neurotransmitters including dopamine, epinephrine, norepinephrine, and histamine are also transported by OCTs (especially OCT3 as the extraneuronal monoamine transporter) (Busch et al., 1996a, 1998; Amphoux et al., 2006). Substrates of the various isoforms are shown in Table 3.

3. Organic Cation/Carnitine Transporters.

Like OCT transporters, OCTNs are members of the SLC22 family. Although they can transport cationic chemicals, OCTNs are most notably known for their ability to influx carnitine (Table 4). OCTN1 was cloned from a human fetal liver cDNA library in 1997, and rat and mouse isoforms were subsequently isolated (Tamai et al., 1997, 2000b; Wu et al., 2000a). OCTN2 was cloned from a human kidney cDNA library (Tamai et al., 1998). Whereas OCTN1 protein is predicted to contain 11 transmembrane domains and one-nucleotide binding domain (Tamai et al., 1997), OCTN2 probably has 12 transmembrane domains (Tamai et al., 1998). Octn3 was first found in mice, and although OCTN3 protein has been detected in a human cell line, the human gene has not been described (Tamai et al., 2000b; Lamhonwah et al., 2003).

Mouse and rat Octn1 are most prominently expressed in kidneys, with detectable mRNA in small intestine, stomach, heart, etc. (Fig. 2) (Tamai et al., 2000b; Slitt et al., 2002; Alnouti et al., 2006). In situ hybridization localizes Octn1 transcript to rat brain, kidney (cortex and medulla), heart (myocardium and valves), and placenta (labyrinth zone) (Wu et al., 2000a). Human OCTN1 is expressed in kidneys, skeletal muscle, placenta, prostate, heart, fetal liver, eyes, and lungs (Fig. 2) (Tamai et al., 1997; Garrett et al., 2008). There is also prominent expression of human OCTN1 in spleen, bone marrow, and whole blood, with particularly high levels in CD14⁺ cells (Tokuhiro et al., 2003). Likewise, immunohistochemical findings demonstrate Octn1 in different regions of the mouse brain and on the apical membrane of mouse proximal tubule cells (Table 4) (Tamai et al., 2004; Lamhonwah et al., 2008). Although OCTN1 is typically localized to the plasma membrane, intracellular localization in mitochondria has been reported and may be responsible for carnitine accumulation in this organelle (Lamhonwah and Tein, 2006).

Mouse and rat Octn2 mRNA are primarily expressed in kidneys (Fig. 2) (Kido et al., 2001; Rodríguez et al., 2002; Slitt et al., 2002; Alnouti et al., 2006). Messenger RNA and/or protein staining also localize Octn2 to heart (myocardium, valves, and arterioles), epididymis, pancreas (α-cells), and brain (cortex, hippocampus, choroid plexus, cerebellum) (Table 4) (Wu et al., 1999; Rodríguez et al., 2002; Kai et al., 2005; Lamhonwah et al., 2008). Human OCTN2 is most notably detected in kidneys and placenta and to a lesser degree in other tissues (Fig. 2) (Tamai et al., 1998; Tokuhiro et al., 2003; Lahjouji et al., 2004; Garrett et al., 2008). Octn2 and OCTN2 proteins are present on the brush border membrane vesicles from kidneys (Tamai et al., 2001a), placental syncytiotrophoblasts (Grube et al., 2005), and small intestine enterocytes (Durán et al., 2005), as well as the basolateral surface of epididymal cells (Rodríguez et al., 2002) (Table 4).

Mouse testes and epididymal spermatozoa (middle piece of sperm tail) express the highest levels of Octn3 mRNA and/or protein (Fig. 2) (Tamai et al., 2000b; Alnouti et al., 2006; Kobayashi et al., 2007). Octn3 is also detected in mouse ovaries (Alnouti et al., 2006), along the basolateral membrane of rat enterocytes (Durán et al., 2005), and within multiple mouse brain regions (Lamhonwah et al., 2008). Although Octn3 and OCTN3 proteins are found on the plasma membrane of various cell types, localization of these proteins to the peroxisome has also been reported and may be important in supplying carnitine for peroxisomal lipid metabolism (Lamhonwah et al., 2005).

As implied by their name, Octn/OCTN proteins transport carnitine (Table 3). During the generation of metabolic energy, carnitine is required for the transport of fatty acids from the cytosol into the mitochondria during the breakdown of lipids. Carnitine transports long-chain acyl groups generated from fatty acids into the mitochondrial matrix, where they can be broken down through β-oxidation. Octn1/OCTN1 is an organic cation uniporter or H⁺/organic cation antiporter that can transport in both directions. Octn2/OCTN2 can act as organic cation uniporters or sodium-carnitine cotransporters (Tamai et al., 1998). Mouse Octn3 is the most selective transporter for carnitine, whereas Octn1 is the least (Tamai et al., 2000b). Mouse Octn1 and Octn2 transport carnitine in a sodium-dependent manner, and Octn3 transports carnitine in a sodium-independent manner (Tamai et al., 2000b).

Octn1/OCTN1 and Octn2/OCTN2 also transport organic cations (Table 3). Both OCTN1 and OCTN2 transport tetraethylammonium, verapamil, quinidine, ergothioneine, and pyrilamine (Tamai et al., 1997; Ohashi et al., 1999; Yabuuchi et al., 1999; Ganapathy et al., 2000; Grube et al., 2006b). OCTN2 also transports the antiseizure drug valproic acid, the antibiotic cephaloridine, and the diuretic spironolactone (Ohashi et al., 1999; Ganapathy et al., 2000; Grube et al., 2006b). In contrast to Octn1 and Octn2, Octn3 has little or no affinity for organic cation model compounds (i.e., tetraethylammonium) and seems to function only as a carnitine transporter (Tamai et al., 2000b).

4. Organic Anion Transporters.

OATs are members of the solute carrier family SLC22A along with OCTs and OCTNs (Table 1). OAT transporters have 12 predicted transmembrane domains arranged in two sets of six helical domains (Simonson et al., 1994; Hosoyamada et al., 1999; Race et al., 1999; Cha et al., 2000). OATs are thought to have two large loop structures between transmembrane domains 1 and 2 and domains 6 and 7 (Hosoyamada et al., 1999). The first loop is extracellular and contains glycosylation sites (Hosoyamada et al., 1999). Glycosylation at multiple sites often results in a range of molecular weights reported for OAT transporters. The second loop occurs intracellularly and contains phosphorylation sites (Hosoyamada et al., 1999). Based on hydropathy analysis, OAT1–3 probably contain cytoplasmic amino and carboxyl termini (Simonson et al., 1994; Hosoyamada et al., 1999). Individual OAT transporters are often linked as phylogenic pairs based upon closely related sequence alignment: OAT1 and OAT3; OAT4 and the urate transporter 1 (URAT1). For example, both OAT4 and URAT1 are found sequentially on chromosome 11q13.1.

Oat1 was first cloned from a rat kidney cDNA library in 1997 (Sekine et al., 1997; Sweet et al., 1997). During the same period, mouse Oat1 was cloned and called the novel kidney transporter (Lopez-Nieto et al., 1997). Human OAT1 was subsequently identified (Reid et al., 1998; Hosoyamada et al., 1999; Race et al., 1999). Rodent Oat1 and human OAT1 mRNA are highest in kidneys (Fig. 2) (Hosoyamada et al., 1999; Buist and Klaassen, 2004), and their proteins are abundantly expressed on the basolateral membranes of renal proximal tubules (Hosoyamada et al., 1999; Tojo et al., 1999). Specifically, OAT1 is strongly expressed on the basolateral membrane of proximal tubules in the S2 segment (Table 4) (Ljubojevic et al., 2004).

Oat2 was first identified in 1994 using a rat liver cDNA library and named the “novel liver-specific transporter” (Simonson et al., 1994). Oat2 was later recloned and renamed (Sekine et al., 1998; Sun et al., 2001b; Kobayashi et al., 2002b). Oat2/OAT2 show species differences in tissue distribution. Mouse Oat2 is found almost exclusively in kidneys (Fig. 2) (Kobayashi et al., 2002b; Buist and Klaassen, 2004). In contrast, rat Oat2 and human OAT2 are expressed primarily in liver with lower levels in kidneys (Fig. 2) (Sekine et al., 1998; Sun et al., 2001b). Furthermore, localization of Oat2/OAT2 proteins in kidney is species-dependent. Rodent Oat2 protein is expressed on the apical membrane of S3 proximal tubules (Table 4) (Kojima et al., 2002; Ljubojević et al., 2007), whereas human OAT2 protein is basolateral (Enomoto et al., 2002b). It is noteworthy that in liver, rat Oat2 protein traffics to the basolateral membrane of hepatocytes (Simonson et al., 1994).

Rat and mouse Oat3 and human OAT3 were identified simultaneously in 1999 (Brady et al., 1999; Kusuhara et al., 1999; Race et al., 1999). Mouse Oat3 was isolated from an animal model of osteosclerosis and termed reduced in osteosclerosis transporter (Brady et al., 1999). Expression of Oat3/OAT3 in mice and humans is confined primarily to the kidneys, where it is localized to the basolateral membrane of proximal tubule cells (Fig. 2, Table 4) (Cha et al., 2001; Kojima et al., 2002; Buist and Klaassen, 2004). Within the kidneys, rat Oat3 is observed in proximal tubule S1 and S2 segments as well as thick ascending limb, distal tubules, and collecting ducts (Ljubojevic et al., 2004). In brain, Oat3 mRNA is expressed in choroid plexus in rats (Choudhuri et al., 2003), and Oat3 protein localizes to the basolateral membrane of brain capillary endothelial cells in rodents (Kikuchi et al., 2003; Ohtsuki et al., 2004a). Mouse Oat3 protein is also expressed on the apical membrane of choroid plexus epithelial cells (Sweet et al., 2002) as well as in developing bone (Brady et al., 1999).

In 2000, OAT4 was identified and functionally characterized (Cha et al., 2000). OAT4 mRNA is expressed largely in kidneys and placenta (Fig. 2) (Cha et al., 2000). Within the kidneys, OAT4 is found on the apical membrane of renal proximal tubule cells (Table 4) (Babu et al., 2002a; Ekaratanawong et al., 2004). In contrast, OAT4 protein is expressed on the basolateral membrane of placental syncytiotrophoblasts (Ugele et al., 2008). No mouse or rat Oat4 ortholog has been identified.

Much less is known about OAT5–7. OAT5 was first identified in humans in 2001 (Sun et al., 2001b) and subsequently in mice (Youngblood and Sweet, 2004) and rats (Anzai et al., 2005). Mouse and rat Oat5 are primarily expressed in kidneys and localize to the apical membrane of proximal tubules in the outer medullary and juxtamedullary cortex in the S2 and S3 segments (Table 4) (Youngblood and Sweet, 2004; Anzai et al., 2005; Kwak et al., 2005). Oat6 has been described only in mice and is uniquely localized to the olfactory mucosa (Monte et al., 2004). OAT7 is the most recently described OAT and was cloned from a human liver cDNA library (Shin et al., 2007). OAT7 protein is localized to the basolateral membrane of human hepatocytes (Shin et al., 2007).

Like Oats, Urat1 is a member of the SLC22A family. The Urat1 transporter was first cloned from a mouse kidney cDNA library and named renal-specific transporter (Table 1) (Mori et al., 1997). The human URAT1 ortholog was later identified (Enomoto et al., 2002a). Urat1/URAT1 transporters are expressed predominantly in the kidneys along the apical border (Fig. 2, Table 4) (Enomoto et al., 2002a; Hosoyamada et al., 2004). Mouse Urat1 protein is also detected in brain capillaries and along the choroid plexus (Imaoka et al., 2004).

The function of Oat/OATs as organic anion exchangers (antiporters) is enabled by sodium and dicarboxylate gradients generated by the sodium-dicarboxylate cotransporter and the sodium-potassium ATPase. In the cases of OAT1 and -3, uptake of substrates across the basolateral membrane is coupled to an outwardly directed concentration gradient of dicarboxylates (i.e., α-ketoglutarate and glutarate) (Wolff et al., 1992; Sekine et al., 1997; Sweet et al., 1997; Bakhiya et al., 2003; Koepsell and Endou, 2004). The concentration gradient of the dicarboxylate provides the driving force for entry of organic anions against an opposing force (inside-negative membrane potential). Because the concentration gradient of the dicarboxylate is maintained by a sodium-potassium-ATPase pump, this mechanism of transport is typically referred to as tertiary transport (Srimaroeng et al., 2008). Coexpression of mouse Oat3 and a sodium-dicarboxylate transporter stimulates Oat3-mediated transport (Ohtsuki et al., 2004a). Rat Oat2 is not thought to be an organic anion-dicarboxylate exchanger (Sekine et al., 1998).

Oat1/OAT1 and Oat3/OAT3 display wide substrate selectivity including endogenous substrates (cyclic nucleotides, urate, indoxyl sulfate) and pharmaceuticals (antibiotics, nonsteroidal anti-inflammatory drugs, diuretics, anticancer drugs, uricosuric agents) (Table 3) (Sekine et al., 1997; Sweet et al., 1997; Apiwattanakul et al., 1999; Enomoto et al., 2003). p-Aminohippurate and estrone sulfate are the prototypical substrates of OAT1 and OAT3, respectively. Oat1/OAT1 transport antibiotics (penicillin G, cephaloridine, tetracycline) and antivirals (such as cidofovir, adefovir, zidovudine, acyclovir, etc.) (Cihlar et al., 1999; Jariyawat et al., 1999; Wada et al., 2000; Babu et al., 2002b). Mercapturic acids are N-acetyl-l-cysteine S-conjugates that are transported by Oat1 and thus eliminated by the kidneys. For example, rat Oat1 can transport S-(2,4-dinitrophenyl)-N-acetyl-l-cysteine (Pombrio et al., 2001). Oat1/OAT1 also transports the chelator 2,3-dimercapto-1-propanesulfonic acid and the mercury thiol conjugates of N-acetylcysteine, homocysteine, and cysteine, probably representing a mechanism for clearance of the environmental neurotoxin methyl mercury (Islinger et al., 2001; Pombrio et al., 2001; Koh et al., 2002; Zalups and Ahmad, 2005a,c).

Oat2/OAT2 mediates the sodium-independent uptake of not only p-aminohippurate but also endogenous (prostaglandins, glutarate) and other exogenous (methotrexate, valproic acid, allopurinol) chemicals (Sun et al., 2001b; Kobayashi et al., 2002b). Species-specific transport of salicylate has been noted: it is transported by rat Oat2, but not by mouse Oat2 (Sekine et al., 1998; Kobayashi et al., 2002b). OAT4 mediates sodium-independent transport of sulfate conjugates (estrone sulfate, indoxyl sulfate, dehydroepiandrosterone sulfate) (Cha et al., 2000; Babu et al., 2002a; Enomoto et al., 2003; Zhou et al., 2006). In general, Oat/OAT5–7 transport dehydroepiandrosterone sulfate and estrone sulfate (Anzai et al., 2005; Schnabolk et al., 2006; Shin et al., 2007). The mycotoxin ochratoxin A is also a substrate for mouse Oat5 (Youngblood and Sweet, 2004). Finally, human OAT7 transports butyrate (Shin et al., 2007).

Urat1/URAT1 seem to be urate-organic anion exchangers (Enomoto et al., 2002a) and are responsible for urate reabsorption in exchange for anions (Hosoyamada et al., 2004). Other organic anions transported by mouse Urat1 include ochratoxin, dehydroepiandrosterone sulfate, and benzylpenicillin (Imaoka et al., 2004).

5. Peptide Transporters.

PEPT1 and PEPT2 are members of the solute carrier family (SLC15A) that transports di- and tripeptides into cells (Table 1). Pept/PEPT1 and -2 were first identified as key peptide carriers in the small intestine and kidneys, respectively (Fei et al., 1994; Liu et al., 1995). PEPT transporters are predicted to have 12-α-helical transmembrane domains with a large extracellular loop between domains 9 and 10 and intracellular carboxyl and amino termini (Fei et al., 1994).

The tissue distribution of Pept/PEPT1 and -2 in mice and humans is shown in Fig. 3 (Saito et al., 1995; Rubio-Aliaga et al., 2000; Herrera-Ruiz et al., 2001; Lu and Klaassen, 2006). Pept/PEPT1 is most prominently expressed in the small intestine of rodents and humans, where it localizes to the apical membrane of enterocytes (Fig. 3, Table 5) (Fei et al., 1994; Ogihara et al., 1999; Shen et al., 1999; Terada et al., 2005; Lu and Klaassen, 2006). Both peptide transporters are detected in the kidneys with Pept1 along the brush border membrane of S1 proximal tubules in the rat and Pept2 expressed in the S2 and S3 segments (Shen et al., 1999). It is noteworthy that Pept1 mRNA is detected in rat but not mouse kidneys (Lu and Klaassen, 2006). In addition to plasma membrane localization, Pept1 is also found in lysosomes (Bockman et al., 1997; Zhou et al., 2000; Sun et al., 2001a). PEPT1 mRNA is also detected in human kidneys, lungs, colon, pancreas, and liver (Liang et al., 1995; Zhang et al., 2004a).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Tissue distribution of Pept, Cnt, Ent, and Mate mRNA in mice and humans. Top, relative mRNA levels of transporters in mouse liver, kidneys, lung, stomach, duodenum, jejunum, ileum, large intestine, heart, brain, gonads (testes and ovaries), placenta, and uterus are shown. Male (♂) mRNA is shown on the left whereas female (♀) mRNA is shown on the right side of each box. References for mouse mRNA expression are included (Lu et al., 2004; Lu and Klaassen, 2006; Lickteig et al., 2008). Bottom, relative mRNA levels of transporters in human liver, kidneys, lung, heart, brain, testes, ovaries, placenta, and uterus are shown. Data for humans were obtained from GNF SymAtlas (http://symatlas.gnf.org/; now located at http://biogps.gnf.org). The GNF1H/MAS5 data set was accessed during September 2008.

Pept2 mRNA is expressed primarily in mouse kidneys (Fig. 3) (Lu and Klaassen, 2006). Pept2 mRNA is expressed in specific cell types of the brain, including astrocytes, subependymal cells, and ependymal cells, and Pept2 protein is detected along the apical membrane of epithelial cells of the choroid plexus (Table 5) (Berger and Hediger, 1999; Shu et al., 2002). Pept2/PEPT2 mRNA and/or protein are also expressed in the enteric nervous system, colon, liver, pancreas, lungs, nasal mucosa, and mammary glands (Groneberg et al., 2001, 2002; Zhang et al., 2004a; Bahadduri et al., 2005; Ruhl et al., 2005; Lu and Klaassen, 2006; Quarcoo et al., 2009).

Pept/PEPT1 and -2 have broad substrate and inhibitor specificity, including di- and tripeptides but not amino acids or tetrapeptides (Table 3) (Daniel and Herget, 1997; Terada et al., 2000; Daniel and Kottra, 2004). Peptide transport by PEPT1 and -2 is coupled with the inward translocation of protons leading to electrogenic transport. Key structural features of PEPT1 and -2 substrates have been described elsewhere (Rubio-Aliaga and Daniel, 2008). Glycylsarcosine (Gly-Sar) is a prototypical substrate for Pept/PEPT1 and -2 transport (Liang et al., 1995; Liu et al., 1995). A number of pharmaceuticals are substrates of Pept/PEPTs, including β-lactam antibiotics (cefadroxil, cefixime, ceftibuten), the photosensitizing agent 5-aminolevulinic acid, and the investigational anticancer drug bestatin (Saito et al., 1995, 1996; Wenzel et al., 1996; Döring et al., 1998; Ocheltree et al., 2004a,b; Xiang et al., 2006; Hu et al., 2007).

6. Concentrative Nucleoside Transporters.

Nucleosides are glycosylamines consisting of a sugar moiety and a purine or pyrimidine base; they include cytidine, uridine, adenosine, guanosine, thymidine, and inosine. Nucleosides are precursors for nucleotides used in DNA and RNA synthesis and are necessary for cell growth. Uptake of nucleosides by hematopoietic and other cell types is a prerequisite for nucleotide synthesis by salvage pathways because these cells lack de novo synthetic ability. Furthermore, adenosine is an important signaling molecule for neurotransmission, platelet aggregation, and other physiological events. Nucleoside analogs have been developed as drugs to treat viral infections and cancers. Nucleoside uptake transporters have been classified according to their transport properties: concentrative (high-affinity sodium-dependent transport using a physiologic sodium gradient) (SLC28A) and equilibrative (low-affinity facilitated carrier transport) (SLC29A) (for review, see Pastor-Anglada et al., 2008; Young et al., 2008) (Table 1).

CNT1–2 transporters have been cloned from multiple species and are predicted to contain 13 transmembrane helices with cytoplasmic amino termini and extracellular carboxyl termini (Huang et al., 1994; Ritzel et al., 1997, 1998; Wang et al., 1997; Baldwin et al., 1999; Patel et al., 2000; Hamilton et al., 2001; Shin et al., 2003). CNT2 was originally named the sodium-dependent purine nucleoside transporter (Wang et al., 1997; Ritzel et al., 1998). In mice and rats, Cnt1 and Cnt2 mRNA are expressed primarily in all three segments of the small intestine as well as kidneys (Cnt1) (Fig. 3) (Huang et al., 1994; Che et al., 1995; Hamilton et al., 2001; Shin et al., 2003; Lu et al., 2004). In humans, CNT1 and CNT2 mRNA are high in liver and kidneys and CNT2 mRNA is also detected in heart, brain, placenta, skeletal muscle, small intestine, and pancreas (Fig. 3) (Wang et al., 1997; Ritzel et al., 1998; Shin et al., 2003; Damaraju et al., 2007; Govindarajan et al., 2007). Immunohistochemical studies localize rat Cnt1 protein to the apical surface of multiple cell types including cortical renal tubules, hepatocytes, and enterocytes (Table 5) (Hamilton et al., 2001; Duflot et al., 2002). Cnt3 mRNA is expressed in the uterus, testes, and ovaries of mice and the lungs of rats (Fig. 3) (Lu et al., 2004). CNT3 mRNA and/or protein are expressed in multiple tissues (Fig. 3) (Ritzel et al., 2001; Damaraju et al., 2007). Within the kidneys, CNT3 localizes to the apical surface of proximal tubules and thick ascending loops of Henle along with some intracellular staining (Table 5) (Damaraju et al., 2007).

Within the concentrative transporters, CNT1 transports pyrimidines (but also adenosine), CNT2 transports purines (but also uridine), and CNT3 transports both purines and pyrimidines (Table 3) (Huang et al., 1994; Fang et al., 1996; Ritzel et al., 1997, 1998, 2001; Wang et al., 1997; Schaner et al., 1999). The chemotherapeutic drug gemcitabine and the antiviral drugs stavudine, zalcitabine, and zidovudine are also substrates of CNT1 (Huang et al., 1994; Ritzel et al., 1997; Mackey et al., 1999; Graham et al., 2000; Cano-Soldado et al., 2004). Substrates of CNT2 include nucleoside analog drugs such as the hepatitis drug ribavirin (Patil et al., 1998). As the broadest nucleoside transporter, CNT3 substrates are more numerous (5-fluorouridine, zebularine, gemcitabine, cladribine, fludarabine, etc.) (Ritzel et al., 2001; Toan et al., 2003).

7. Equilibrative Nucleoside Transporters.

As low-affinity, facilitated carriers, ENTs transport chemicals down concentration gradients. Intracellular levels of nucleosides are typically low because they are converted to nucleotides. Although ENT transporters are most often considered uptake carriers, they can function bidirectionally. ENT1–4 transporters have been detected and cloned from different species (Table 1) (Griffiths et al., 1997; Yao et al., 1997; Crawford et al., 1998; Kiss et al., 2000; Baldwin et al., 2005; Zhou et al., 2007b). Hydropathy analysis suggests that Ent1–2 proteins are composed of 11 transmembrane domains with an internal amino terminus and extracellular carboxyl tail (Yao et al., 1997; Crawford et al., 1998).

The mRNA expression of Ent1 is primarily in mouse and rat liver, kidneys, lung, brain, and testes (Fig. 3) (Choi et al., 2000; Lu et al., 2004; Redzic et al., 2005). Human ENT1 expression is wide ranging, including liver, lungs, heart, ovaries, brain, kidneys, erythrocytes, fetal liver, and placenta (Fig. 3) (Griffiths et al., 1997; Anderson et al., 1999b; Pennycooke et al., 2001; Damaraju et al., 2007; Govindarajan et al., 2007). Within the kidneys, ENT1 staining is observed on the apical surface of proximal tubules and on both the apical and basal membranes of the thick ascending loops of Henle and collecting ducts (Table 5) (Damaraju et al., 2007). Similar apical localization of ENT1 is noted in human placental syncytiotrophoblasts (Govindarajan et al., 2007). Ent2 and Ent3 mRNA share similar tissue distributions with highest levels observed in kidneys, brain, gonads, and uteri of mice and rats (Fig. 3) (Anderson et al., 1999a; Lu et al., 2004; Redzic et al., 2005). The profile of ENT2 mRNA demonstrates high levels in skeletal muscle and heart and detectable amounts in other organs (Fig. 3) (Pennycooke et al., 2001). Likewise, ENT3 exhibits broad tissue distribution with prominent expression in placenta, lung, ovaries, spleen, and bone marrow (Baldwin et al., 2005). Mutations in ENT3 have been linked to H syndrome, which is characterized by skin, auditory, heart, and spleen abnormalities (Table 10) (Molho-Pessach et al., 2008). A fourth ENT isoform, ENT4, has also been cloned, although there is less information available about its tissue distribution and transport properties (Barnes et al., 2006; Xia et al., 2007; Zhou et al., 2007b).

Ent/ENT transport is sodium-independent and, like Cnt/CNT transporters, endogenous nucleosides as well as cancer and antiviral nucleoside analogs are common substrates (Table 3) (Griffiths et al., 1997; Mackey et al., 1999; Kiss et al., 2000; Baldwin et al., 2005; Damaraju et al., 2005; Nagai et al., 2007; Govindarajan et al., 2008). It is noteworthy that ENT1 is also expressed in the mitochondria, where it may be involved in the cellular toxicity of antiviral nucleoside drugs (Lai et al., 2004; Govindarajan et al., 2009).

8. Multidrug and Toxin Extrusion Transporters.

For a number of years, it was understood that organic cations entered the cells via OCT transporters; however, how they exited was not clear. Transporters were first identified in bacteria and were called NorM and YdhE. They were later named the multidrug and toxin extrusion (MATE) transporters (Morita et al., 1998; Brown et al., 1999; Otsuka et al., 2005; Terada and Inui, 2008). Although MATE transporters are Slc transporters, they function as efflux proteins. Hiasa and colleagues reported that there are three subgroups of mammalian MATE transporters: class I includes rodent Mate1 and human MATE1; class II includes human MATE2 (no rodent ortholog of this subgroup); class III includes mouse and rat Mate2 (Hiasa et al., 2007; Moriyama et al., 2008). In an attempt to avoid confusion, it has been proposed that mouse and rat Mate2 be renamed Mate3 (Terada and Inui, 2008). However, for the purpose of this review, we will continue to use the existing Mate2 designation. Another MATE transporter, multidrug and toxin extrusion 2-K (MATE2-K), shows 94% amino acid similarity with MATE2 (Masuda et al., 2006). MATE2-K was first reported to be a splice variant of MATE2; however, this is currently under reconsideration, because a second attempt to clone human MATE2 has not been successful (Masuda et al., 2006; Terada and Inui, 2008). Mate1 (SLC47A1) and Mate2-K (SLC47A2) have also been cloned from rabbits (Zhang et al., 2007a) (Table 1). Recent work suggests that MATE1 and MATE2-K are composed of 13 putative transmembrane domains with amino and carboxyl termini on the intracellular and extracellular faces of the plasma membrane, respectively (Masuda et al., 2006; Zhang et al., 2007a; Terada and Inui, 2008).

Mouse Mate1 mRNA is most abundant in kidneys and is detected at lower levels in the liver and heart (Fig. 3) (Otsuka et al., 2005; Lickteig et al., 2008). Human MATE1 is expressed in heart, liver, adrenal gland, testes, skeletal muscle, and kidneys (Otsuka et al., 2005; Masuda et al., 2006). Mouse Mate2 mRNA is strongly detected in testes, whereas human MATE2-K is expressed in kidneys (but not in the testes) (Fig. 3) (Otsuka et al., 2005; Masuda et al., 2006; Hiasa et al., 2007; Lickteig et al., 2008). Differences in the tissue distributions of mouse Mate2 and human MATE2-K may be due to their classification in class II and III subgroups, respectively (Hiasa et al., 2007).

MATE1 effluxes organic cations such as tetraethylammonium, 1-methyl-4-phenylpyridinium, oxaliplatin, and paraquat using a proton-coupled electroneutral exchange (Table 3) (Otsuka et al., 2005; Terada et al., 2006; Chen et al., 2007b; Yokoo et al., 2007). Rat Mate1 also transports the histamine H₂-receptor antagonist cimetidine, the antidiabetic drug metformin, and the antibiotic cephalexin (Ohta et al., 2006; Terada et al., 2006). Mouse Mate1 prefers N¹-methylnicotinamide and guanidine as substrates, whereas mouse Mate2 prefers tetraethylammonium (Hiasa et al., 2007). MATE2-K transports organic cations including tetraethylammonium, 1-methyl-4-phenylpyridinium, cimetidine, procainamide, and metformin (Table 3) (Masuda et al., 2006).

1. Multidrug Resistance Proteins.

ABC transporters contain ATP-binding domains that possess ATPase activity (hydrolysis of ATP to ADP) to provide energy for translocating substrates across membranes, most often against concentration gradients. Pgp was the first drug transporter described (Juliano and Ling, 1976). Pgp is encoded by multiple MDR genes, including MDR1 (ABCB1) and MDR3 (ABCB4) in humans, although Pgp most often indicates the ABCB1 gene product (Table 1) (van der Bliek et al., 1988; Dhir et al., 1990; Lincke et al., 1991). The rodent orthologs of MDR1 and MDR3 are Mdr1a/1b and Mdr2, respectively. To make things more confusing, the Mdr1a gene was also referred to as Mdr3 within the older literature (Devault and Gros, 1990; Dhir et al., 1990). MDR1, Mdr1a, and Mdr1b are drug transporters (Dhir et al., 1990), whereas MDR3 and Mdr2 translocate phospholipids such as phosphatidylcholine from the inner to the outer canalicular membrane (Schinkel et al., 1991; Ruetz and Gros, 1994; van Helvoort et al., 1996). The structural topology of Pgp consists of two distinct regions containing six putative transmembrane domains and one nucleotide binding domain (van der Bliek et al., 1988; Devault and Gros, 1990; Aller et al., 2009). The amino and carboxyl termini of Pgp are located intracellularly.

The tissue distribution of Mdr1a/1b and MDR1 is broad with their mRNA detected in many tissues (Fig. 4) (Chin et al., 1989; Melaine et al., 2002; Hitzl et al., 2004). Mdr1a mRNA is most prominent in the large intestine followed by the small intestine, kidneys, and brain (Fig. 4) (Cui et al., 2009c). Meanwhile, Mdr1b expression is highest in the kidneys, lungs, brain, ovaries, and placenta (Fig. 4) (Cui et al., 2009c). Within these tissues, MDR1/Mdr1a/1b proteins are detected on the apical/luminal surface (Table 6) (Schinkel et al., 1994; Lankas et al., 1998; Panwala et al., 1998; Rao et al., 1999; Miller et al., 2000; St-Pierre et al., 2000; Ushigome et al., 2003; Soontornmalai et al., 2006; Sun et al., 2006). Expression of MDR3 and Mdr2 is primarily restricted to the liver, where they localize to the canalicular membrane (Table 6) (Buschman et al., 1992; Smit et al., 1994; de Vree et al., 1998; Scheffer et al., 2000; Cui et al., 2009c). Although MDR3 and Mdr2 mRNA have been detected in additional tissues, functional protein expression has not been shown (Smit et al., 1994; Cui et al., 2009c).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

Tissue distribution of Mdr, Mrp, and Bcrp mRNA in mice and humans. Top, relative mRNA levels of transporters in mouse liver, kidneys, lung, stomach, duodenum, jejunum, ileum, large intestine, brain, gonads (testes and ovaries), and placenta are shown. Male (♂) mRNA is shown on the left whereas female (♀) mRNA is shown on the right side of each box. References for mouse mRNA expression are included (Maher et al., 2005b; Cui et al., 2009c). Bottom, relative mRNA levels of transporters in human liver, kidneys, lung, heart, brain, testes, ovaries, placenta, and uterus are shown. Data for humans were obtained from GNF SymAtlas (http://symatlas.gnf.org/; now located at http://biogps.gnf.org). The GNF1H/MAS5 data set was accessed during September 2008.

Overexpression of Mdr1a and -1b confers resistance to multiple drugs by enhancing cellular extrusion (Dhir et al., 1990; Raymond et al., 1990). Early work demonstrated that cells transfected with Mdr1b were resistant to colchicine and doxorubicin whereas cells overexpressing Mdr1a were resistant to actinomycin D (Table 7) (Devault and Gros, 1990; Tang-Wai et al., 1995). In addition, MDR1-transfected cells were resistant to vincristine, colchicine, daunorubicin, doxorubicin, and actinomycin D (Schinkel et al., 1991; Tang-Wai et al., 1995).

Mdr2 translocates a fluorescent phosphatidylcholine analog in overexpressing yeast cells (Ruetz and Gros, 1994). In addition, MDR3 overexpression in fibroblasts promotes the transfer of phosphatidylcholine from the inner to outer leaflet of the plasma membrane (Smith et al., 1994). Translocation from the inner to the outer leaflet of the canalicular membrane by MDR3 enhances the availability of phospholipids for extraction into the bile canaliculi by bile acids. Phospholipids form micelles with bile acids, thereby reducing the likelihood of injury to the biliary tree (Elferink et al., 1997).

2. Multidrug Resistance-Associated Proteins.

Mrp/MRP transporters constitute nine members of the ATP-binding cassette C subfamily (ABCC1–6, 10–12) (Table 1). Other transporters in the ABCC subfamily are the cystic fibrosis transmembrane conductance regulator (ABCC7) and two sulfonylurea receptor isoforms (ABCC8 and -9). The following nine Mrp/MRP isoforms have been cloned from various species: MRP1 (Cole et al., 1992), MRP2 (Ito et al., 1997; Keppler et al., 1997; Paulusma et al., 1997; Fritz et al., 2000), MRP3 (Kiuchi et al., 1998; Uchiumi et al., 1998), MRP4 (Kool et al., 1997; Lee et al., 1998; Chen and Klaassen, 2004), MRP5 (McAleer et al., 1999; Jedlitschky et al., 2000; Wijnholds et al., 2000b), MRP6 (Kool et al., 1999a), MRP7 (Hopper et al., 2001), MRP8, and MRP9 (Bera et al., 2001; Tammur et al., 2001; Shimizu et al., 2003). ABCC13 is likely to be a pseudogene encoding a truncated protein of fetal origin (Yabuuchi et al., 2002; Annilo and Dean, 2004).

MRP transporters contain consensus regions named the Walker A, Walker B, and Signature C motifs that are required for ATP binding. MRP1, -2, -3, -6, and -7 contain three membrane spanning domains with a total of 17 hydrophobic transmembrane regions. For these five MRP proteins, computational analysis suggests extracellular and intracellular amino and carboxyl termini, respectively. MRP4, -5, -8, and -9 are smaller proteins with only two domains that span the plasma membrane (12 total transmembrane regions). The amino and carboxyl termini are both predicted to be intracellular for MRP4, -5, -8, and -9. MRP proteins have two intracellular nucleotide binding domains. For details of MRP transporters, including historical highlights, the reader is referred to recent reviews (Jedlitschky et al., 2006; Nies et al., 2008; Toyoda et al., 2008; Zhou et al., 2008c).

Mouse Mrp1 mRNA is observed in testes, ovaries, brain, placenta, and stomach (Fig. 4) (Maher et al., 2005b). Human MRP1 mRNA and protein is most highly expressed in testes, lungs, heart, bladder, spleen, adrenal glands, placenta, kidneys, peripheral blood mononuclear cells, and skeletal muscle (Fig. 4) (Cole et al., 1992; Flens et al., 1996; Kool et al., 1997). Within the intestine, the highest MRP1 mRNA levels are found in the ascending and transverse colon (Zimmermann et al., 2005). Overexpression of MRP1 in MDCK cells traffics this protein to the basolateral membrane similar to its localization in the choroid plexus, bronchial epithelia, and intestinal crypt cells (Table 8) (Wright et al., 1998; Peng et al., 1999; Rao et al., 1999; Zhang et al., 2004d; Roberts et al., 2008). Likewise, MRP1 is expressed on the basolateral membrane of the amnion as well as the chorionic and decidua membranes, but on the apical membrane of placental syncytiotrophoblasts and brain capillary endothelial cells (St-Pierre et al., 2000; Pascolo et al., 2003; Zhang et al., 2004d; Aye et al., 2007).

Mouse Mrp2 mRNA is detected in the small intestine, liver, and kidneys (Fig. 4) (Maher et al., 2005b). The initial identification of Mrp2/MRP2 was associated with genetic disorders in rats and humans (Table 10) (Paulusma et al., 1996). Human MRP2 mRNA is most highly expressed in liver, followed by the duodenum and kidneys (Fig. 4) (Kool et al., 1997; Uchiumi et al., 1998). The expression of Mrp2/MRP2 decreases along the intestinal tract with lower detection in the colon compared with the duodenum and ileum (Mottino et al., 2000; Maher et al., 2005b; Zimmermann et al., 2005). It is noteworthy that little difference in Mrp2 mRNA expression was observed between the proximal and distal rat intestine, whereas protein levels declined from jejunum to ileum (Mottino et al., 2000). Overexpression of the ABCC2 gene in MDCK cells targets the MRP2 protein to the apical surface (Zhang et al., 2004d). Mrp2/MRP2 is expressed on the apical surface of hepatocytes, the amnion epithelial membrane, and proximal tubule cells (Fig. 4, Table 8) (Paulusma et al., 1996; Kinoshita et al., 1998; Scheffer et al., 2000, 2002a; van Aubel et al., 2002; Aye et al., 2007).

Mouse Mrp3 mRNA is expressed in the small and large intestine, liver, stomach, and retinal vascular endothelium (Fig. 4) (Maher et al., 2005b; Tachikawa et al., 2008). Levels of human MRP3 mRNA are highest in liver and are also detectable in duodenum, colon, pancreas, adrenal glands, kidneys, and lungs (Fig. 4) (Kool et al., 1997; Belinsky et al., 1998; Kiuchi et al., 1998; Uchiumi et al., 1998; König et al., 1999; Zimmermann et al., 2005). Typically, Mrp3/MRP3 protein localizes to the basolateral membrane of epithelial cells, including hepatocytes, cholangiocytes, distal convoluted tubules, gallbladder, pancreatic ductal cells, and enterocytes of the ileum and colon (Table 8) (Kool et al., 1999b; Scheffer et al., 2000, 2002b; Soroka et al., 2001; Rost et al., 2002; Zelcer et al., 2006). Two exceptions are localization to the lateral and apical membranes of the choroid plexus epithelial cells and placental syncytiotrophoblasts, respectively (St-Pierre et al., 2000; Soontornmalai et al., 2006).

Rodent Mrp4 mRNA is high in kidneys, prostate, and stomach (Fig. 4) (Chen and Klaassen, 2004; Maher et al., 2005b). Likewise, human MRP4 mRNA is most prominently expressed in the kidneys, followed by lungs, skeletal muscle, prostate, testes, ovaries, small intestine, bladder, platelets, and tonsil (Fig. 4) (Kool et al., 1997; Lee et al., 1998; Jedlitschky et al., 2004). Plasma membrane localization of MRP4 is dependent upon cell type (Table 8). For example, Mrp4/MRP4 localizes to the apical membrane in proximal tubule cells (van Aubel et al., 2002) and brain capillary endothelial cells (Roberts et al., 2008). In contrast, Mrp4/MRP4 protein is detected on the basolateral surface of hepatocytes (Assem et al., 2004), prostate glandular epithelial cells (Lee et al., 2000b; Rius et al., 2005), choroid plexus epithelia (Roberts et al., 2008), and visceral yolk sac epithelium (Aleksunes et al., 2008b).

Mouse Mrp5 and human MRP5 mRNA are widely expressed (Fig. 4) (Kool et al., 1997; Belinsky et al., 1998; McAleer et al., 1999; Dazert et al., 2003; Maher et al., 2005b). Overexpression of the ABCC5 gene in MDCK cells causes MRP5 protein localization to the basolateral membrane (Wijnholds et al., 2000b). In brain, MRP5 is expressed in astrocytes, pyramidal neurons, and along the blood-brain barrier (Nies et al., 2004). MRP5 is detected on the epithelial cells of the urethra and the urogenital tract (Nies et al., 2002b). Within the placenta, MRP5 is present on the basolateral membrane of syncytiotrophoblasts as well as near fetal blood vessels (Table 8) (Pascolo et al., 2003; Meyer Zu Schwabedissen et al., 2005). It is noteworthy that placental MRP5 mRNA decreases during human gestation (Meyer Zu Schwabedissen et al., 2005). In addition, the amniotic membrane from term pregnancies expresses MRP5 on its apical and basolateral epithelial cells (Aye et al., 2007).

MRP6 was first considered the anthracycline resistance-associated gene in resistant leukemia cell lines (Longhurst et al., 1996; O'Neill et al., 1998). Subsequent cloning demonstrated that the anthracycline resistance-associated gene is a partial protein product of ABCC6 (Belinsky and Kruh, 1999; Kool et al., 1999a). Despite expression of MRP6 in chemotherapy-resistant cell lines, it is not thought to play a role in conferring selective growth advantage to malignant cells (Kool et al., 1999a). The profiles of rodent Mrp6 and human MRP6 mRNA expression are largely similar (Fig. 4) (Kool et al., 1999a; Madon et al., 2000; Maher et al., 2005b, 2006b). In mice, rats, and humans, Mrp6/MRP6 mRNA is expressed in liver and kidneys, where their proteins are detected on the basolateral membranes of hepatocytes and proximal tubules (Table 6) (Kool et al., 1997; Belinsky and Kruh, 1999; Scheffer et al., 2002a; Maher et al., 2005b, 2006b). Likewise, overexpression of MRP6 in MDCK cells results in basolateral trafficking (Sinkó et al., 2003). MRP6 protein has also been found in enteroendocrine G cells of the stomach (Beck et al., 2005).

Mrp7/MRP7 is ubiquitously expressed (Fig. 4). Mouse Mrp7 is detected highly in testes, placenta, small intestine, kidneys, heart, and lungs (Kao et al., 2002; Maher et al., 2005b). Within the testes, Mrp7 is expressed in Sertoli cells (Augustine et al., 2005). Human MRP7 is expressed in skin, testes, stomach, spleen, colon, kidneys, brain, heart, and liver (Hopper et al., 2001). Human MRP8 is ubiquitously expressed in ovaries, heart, mammary glands, lungs, muscle, pancreas, testes, and intestine (Fig. 4) (Bera et al., 2001; Tammur et al., 2001). In the brain, MRP8 protein is located on axons in the white matter (Bortfeld et al., 2006). No mouse Mrp8 ortholog has been reported. In mice and rats, Mrp9 mRNA is only detected in testes (Fig. 4) (Maher et al., 2005b; Ono et al., 2007). Within the testes, mouse and boar sperm strongly express Mrp9 (Ono et al., 2007). Expression of human MRP9 is more ubiquitous than rodent counterparts (Tammur et al., 2001; Ono et al., 2007).

Although identification of MDR1 drew attention to the existence of efflux pumps in chemotherapy-resistant tumors, a number of cancers did not overexpress this gene. As a result, researchers proposed the likelihood of additional efflux pumps. MRP1 was first reported in 1992 (Cole et al., 1992) and subsequently linked to anticancer drug resistance (Barrand et al., 1994; Grant et al., 1994; Stride et al., 1997). Early studies demonstrated the ability of MRP1 to transport glutathione conjugates (including oxidized glutathione), prostaglandins, and leukotrienes (Table 7) (Leier et al., 1994, 1996; Jedlitschky et al., 1996; Pulaski et al., 1996; Zaman et al., 1996; Evers et al., 1997).

Functional transport analysis demonstrates the importance of Mrp2 in the apical excretion of substrates (such as β-lactam antibiotics, methotrexate, estradiol-17β-glucuronide) from the liver and gastrointestinal tract (Table 7) (Masuda et al., 1997; Gotoh et al., 2000; Morikawa et al., 2000; Kato et al., 2008). Overexpression of MRP2 in vitro confers resistance to a number of cytotoxic drugs, including etoposide, cisplatin, doxorubicin, and epirubicin (Cui et al., 1999; Kawabe et al., 1999). The excretion of glucuronide conjugates across the canalicular surface of hepatocytes is mediated by Mrp2/MRP2 and across the sinusoidal membrane by Mrp3/MRP3. Mrp3/MRP3 transports glucuronide conjugates in addition to chemotherapeutic drugs and bile acids (Hirohashi et al., 1999; Kool et al., 1999b; Zeng et al., 1999, 2000, 2001; Li et al., 2003). Similar substrate profiles have been observed between rat and human Mrp3/MRP3 proteins (Akita et al., 2002). Although both proteins can transport bile acids, rat Mrp3 has a higher affinity for bile acids such as taurine- and glycine-conjugated cholic acid compared with human MRP3 (Zelcer et al., 2003b). Instead, it is proposed that human MRP3 may be more important in bile acid handling during cholestasis (Zelcer et al., 2003b).

Initial studies of Mrp4/MRP4 function pointed to a role for this transporter in conferring resistance to nucleoside analog antiviral drugs such as 9-(2-phosphonylmethoxyethyl)adenine as well as the anticancer drugs 6-mercaptopurine, topotecan, and methotrexate (Table 7) (Schuetz et al., 1999; Lee et al., 2000b; Chen et al., 2001; Reid et al., 2003a; Tian et al., 2005; El-Sheikh et al., 2007). In addition, Mrp4/MRP4 transports endogenous molecules such as leukotrienes, prostaglandins, folate, bile acids, urate, and cyclic nucleotides (Chen et al., 2001, 2002; Lai and Tan, 2002; Reid et al., 2003b; Zelcer et al., 2003a; Jedlitschky et al., 2004; Van Aubel et al., 2005; Rius et al., 2006, 2008; Bataille et al., 2008; Lin et al., 2008). Mrp5/MRP5 transports endogenous (cAMP, cGMP, folate, hyaluronan) and exogenous chemicals [methotrexate, 6-mercaptopurine, 6-thioguanine, 9-(2-phosphonylmethoxyethyl)adenine, 5-fluorouracil] (McAleer et al., 1999; Jedlitschky et al., 2000; Wijnholds et al., 2000b; Wielinga et al., 2002, 2003, 2005; Reid et al., 2003a; Pratt et al., 2005; Schulz et al., 2007). To date, only a limited number of MRP6 substrates have been identified, including leukotriene C₄, etoposide, and the endothelin receptor antagonist BQ-123 (Belinsky et al., 2002; Iliás et al., 2002). Likewise, MRP7 transports leukotriene C₄ and estradiol-17β-glucuronide as well as a number of chemotherapeutic drugs (Chen et al., 2003; Hopper-Borge et al., 2004). Overexpression of MRP7 confers resistance to docetaxel, paclitaxel, and vincristine (Hopper-Borge et al., 2009). Consistent with these data, high levels of MRP7 correlate with the chemotherapeutic resistance of a number of cell lines (Naramoto et al., 2007; Oguri et al., 2008; Bessho et al., 2009). MRP8 transports cyclic nucleotides, sulfated steroids, antiviral drugs, and chemotherapeutic drugs (Guo et al., 2003; Chen et al., 2005b).

3. Breast Cancer Resistance Protein.

Despite the chemotherapeutic resistance conferred by MDR and MRP isoforms, resistant cancer cell lines lacking MDR/MRP over-expression suggested an additional ABC subfamily might be involved. A candidate gene named mitoxantrone resistance (MXR) was discovered in a resistant breast cancer cell line (Doyle et al., 1998; Miyake et al., 1999; Ross et al., 1999). At the same time, this transporter was also reported as the “ABC transporter highly expressed in placenta (ABCP)” (Allikmets et al., 1998). MXR/ABCP was later renamed the second member of the G subfamily of ABC transporters (ABCG2) or BCRP (Table 1). The BCRP protein is considered a “half-transporter” consisting of two domains: amino-terminal ATP-binding domain and carboxyl-terminal transmembrane domain (six transmembrane segments) (Wang et al., 2008a). To function, BCRP must form oligomers. Formation of a homodimer via extracellular loops between transmembrane helices 5 and 6 has been proposed (Henriksen et al., 2005a,b). Additional reports suggest that it is more likely that BCRP associates into higher order oligomers, specifically homotetramers (Xu et al., 2004).

Despite the original identification of BCRP in a breast cancer cell line, the expression of this transporter is quite variable among primary breast carcinomas, and there is no relationship between BCRP and the chemotherapeutic response to anthracyclines and/or survival of patients with breast cancer (Faneyte et al., 2002). Expression of BCRP in other tumor types is variable; detection is more frequent in adenocarcinomas of the digestive tract, endometrium, and lungs (Diestra et al., 2002). The relationship between BCRP expression and clinical outcomes in these other tumor types remains elusive.

Similar to other transporters (MDR and MRP) first associated with cancer cell resistance, BCRP is expressed not only in tumors but also in a number of organs associated with drug absorption, metabolism, and excretion. A similar distribution of mouse and rat Bcrp expression has been reported with high Bcrp mRNA in rodent kidneys, liver, small intestine, placenta, and testes (Fig. 4) (Tanaka et al., 2005). High expression of human BCRP mRNA is detected in the placenta as well as in the brain, liver, kidneys, small intestine, colon, prostate, spinal cord, adrenal gland, uterus, and testes (Fig. 4) (Doyle et al., 1998; Fetsch et al., 2006). Within the human gastrointestinal tract, BCRP mRNA is highest in the duodenum and decreases down to the rectum (Gutmann et al., 2005). Bcrp/BCRP is almost exclusively expressed on the apical surface of epithelial cells, including hepatocytes, proximal tubules, enterocytes, trophoblasts, yolk sac, and mammary glands as well as brain and retinal capillary endothelial cells (Table 8) (Maliepaard et al., 2001; Cooray et al., 2002; Jonker et al., 2002; Aronica et al., 2005; Tachikawa et al., 2005; Asashima et al., 2006; Fetsch et al., 2006; Pulido et al., 2006; Aleksunes et al., 2008b; Roberts et al., 2008). Expression of Bcrp/BCRP on lactating mammary glands in a number of species contributes to the excretion of chemicals into breast milk (Merino et al., 2005b; van Herwaarden et al., 2006, 2007; Pérez et al., 2009).

Because BCRP was first identified from chemotherapy-resistant cancer cells, early functional analysis focused upon anticancer drugs. Overexpression of Bcrp/BCRP reduces accumulation and confers resistance to mitoxantrone, daunorubicin, doxorubicin, topotecan, and rhodamine 123 (Doyle et al., 1998; Allen et al., 1999; Litman et al., 2000; Wang et al., 2003c). Bcrp/BCRP transports a wide range of substrates, including photosensitizers, antibiotics, antivirals, natural products, statins, and carcinogens (Table 7) (Merino et al., 2005a, 2006; Robey et al., 2005; Huang et al., 2006a; Ando et al., 2007; Enokizono et al., 2007; Pan et al., 2007; Myllynen et al., 2008). BCRP also transports the fluorescent dye Hoechst 33342, which is used to label stem cell populations (Kim et al., 2002b).

1. Sodium Taurocholate Cotransporting Polypeptide.

NTCP belongs to the SLC10A transporter family (Table 1). Ntcp was first cloned from rat liver (Hagenbuch et al., 1990). NTCP has seven putative transmembrane domains that are glycosylated with the carboxyl terminus oriented into the cytoplasm (Hagenbuch et al., 1991; Ananthanarayanan et al., 1994; Hagenbuch and Meier, 1994; Mareninova et al., 2005).

As part of the enterohepatic recirculation, Ntcp/NTCP is responsible for the basolateral uptake of bile acids from the portal blood into hepatocytes. Ntcp/NTCP transports bile acids such as taurocholate as well as conjugated di- and trihydroxy bile acids in a sodium-dependent manner (Hagenbuch et al., 1991; Boyer et al., 1994; Hagenbuch and Meier, 1994; Meier et al., 1997; Saeki et al., 2002). Although Ntcp was first identified in rat liver, this gene is expressed in livers of multiple species including human, mouse, rabbit, and guinea pig (Fig. 5) (Hagenbuch et al., 1991). Ntcp/NTCP proteins localize to the basolateral surface of hepatocytes in humans, rats, and mice (Table 9) (Ananthanarayanan et al., 1994; Stieger et al., 1994; Keitel et al., 2005; Aleksunes et al., 2006). It is noteworthy that Ntcp is also expressed in rat pancreas, where it traffics to the apical plasma membrane of acinar cells (Kim et al., 2002a). In addition to transporting bile acids and estrone sulfate, human NTCP (but not rat Ntcp) transports rosuvastatin (Craddock et al., 1998; Ho et al., 2006b).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 5.

Tissue distribution of Ntcp, Asbt, Bsep, Ost, Abca, and Abcg mRNA in mice and humans. Top, relative mRNA levels of transporters in mouse liver, kidneys, lung, stomach, duodenum, jejunum, ileum, large intestine, brain, gonads (testes and ovaries), and placenta are shown. Male (♂) mRNA is shown on the left, whereas female (♀) mRNA is shown on the right side of each box. References for mouse mRNA expression are included (Dieter et al., 2004; Cheng et al., 2007). Bottom, relative mRNA levels of transporters in human liver, kidneys, lung, heart, brain, testes, ovaries, placenta, and uterus are shown. Data for humans were obtained from GNF SymAtlas (http://symatlas.gnf.org/; now located at http://biogps.gnf.org). The GNF1H/MAS5 data set was accessed during September 2008.

2. Apical Sodium-Dependent Bile Acid Transporter.

It had been known for some time that there was active transport of bile acids across the apical membrane of ileal enterocytes, but it was not until 1994, when Asbt was cloned, that this process was better understood (Wong et al., 1994). ASBT is the second member of the SLC10A family and shares 35% amino acid identity to NTCP (Table 1) (Wong et al., 1994). Asbt was first cloned from a hamster ileal cDNA library and named the ileal sodium-dependent bile acid cotransporter (Wong et al., 1994). Similar to NTCP, hydropathy analysis predicts that the ASBT protein has seven putative transmembrane domains with an extracellular amino terminus and a cytoplasmic carboxyl terminus (Banerjee and Swaan, 2006).

Asbt-mediated uptake of bile acids represents the first step in bile acid reabsorption in intestine. Contrary to Ntcp/NTCP, which localizes to the basolateral membrane of hepatocytes, Asbt is found on the apical surface of cholangiocytes, where it participates in cholehepatic recirculation (Fig. 5, Table 9) (Alpini et al., 1997; Lazaridis et al., 1997). Shuttling or localization of ASBT to the apical surface is due to its cytoplasmic tail (Sun et al., 1998). In addition to ileal enterocytes and cholangiocytes, Asbt is also an apical protein in the kidneys (Christie et al., 1996).

Like Ntcp, Asbt/ASBT transports unconjugated and conjugated bile acids in a sodium-dependent manner (Wong et al., 1994; Craddock et al., 1998). The substrate specificity of Asbt is narrower than Ntcp (Craddock et al., 1998). Human ASBT prefers taurine- and glycine-conjugated bile acids, rather than the unconjugated forms (Craddock et al., 1998). In addition, the affinity of ASBT to dihydroxy bile acids is higher than that for trihydroxy bile acids (Craddock et al., 1998).

3. Bile Salt Export Pump.

Secretion of conjugated bile acids from hepatocytes into bile suggested the existence of an active transport mechanism across the canalicular membrane. In 1995, the sister of Pgp (SPGP, ABCB11) was cloned from a pig cDNA library (Childs et al., 1995) and later from additional species (Table 1) (Childs et al., 1998; Green et al., 2000; Lecureur et al., 2000; Byrne et al., 2002; Noe et al., 2002). SPGP was subsequently renamed Bsep (Gerloff et al., 1998). A 12-membrane-spanning domain protein containing putative glycosylation sites, nucleotide binding domains, and typical structures of ABC-transporters has been described for BSEP (Gerloff et al., 1998).

Bsep/BSEP is exclusively expressed in liver on the canalicular membrane of multiple species (Fig. 5) (Childs et al., 1995, 1998; Gerloff et al., 1998). As the primary canalicular bile acid transporter, Bsep/BSEP primarily transports conjugated bile acids (including taurochenodeoxycholate, taurocholate, tauroursodeoxycholate, glycochenodeoxycholate, and glycocholate) in an ATP-dependent manner (Gerloff et al., 1998; Green et al., 2000; Byrne et al., 2002). In contrast, Bsep does not transport cholic acid (Noe et al., 2002). Although BSEP primarily transports bile acids, it can also transport pharmaceuticals such as pravastatin (Hirano et al., 2005). A number of BSEP inhibitors have been identified (cyclosporine A, rifampicin, glibenclamide) (Byrne et al., 2002).

4. Organic Solute Transporters.

Bile acids are absorbed in the small intestine as part of the enterohepatic recirculation. Uptake of bile acids from the intestinal lumen is mediated by Asbt. Once inside the enterocyte, bile acids are translocated to the basolateral membrane by the intestinal bile acid binding protein and subsequently transported across the basolateral membrane by heterodimerized OSTα/β transporters. Ostα and -β were first identified in the liver of the marine skate (Wang et al., 2001d) and subsequently in human and mouse livers (Table 1) (Seward et al., 2003). OSTα is larger than OSTβ (Wang et al., 2001d). Whereas OSTα contains seven putative membrane-spanning domains, OSTβ contains only one (Seward et al., 2003).

Mouse Ostα and -β mRNA are highest in the ileum with detectable levels also in the kidneys, duodenum, jejunum, cecum, and the proximal colon (Fig. 5) (Ballatori et al., 2005; Dawson et al., 2005; Li et al., 2007b) Likewise, human OSTα and -β are expressed to varying levels in the testes, colon, liver, small intestine, adrenal glands, kidneys, and ovaries (Seward et al., 2003). Whereas OSTα is expressed in the human liver (hepatocytes and cholangiocytes), levels are very low in mouse liver and limited only to cholangiocytes (Ballatori et al., 2005). OSTα and -β localize to the basolateral membrane of ileal enterocytes, hepatocytes, cholangiocytes, and proximal renal tubules (Table 9) (Ballatori et al., 2005). Ostα is required for the delivery of Ostβ protein to the plasma membrane; in turn, OSTα and -β need to be coexpressed to function properly (Li et al., 2007b). Ostα and -β function as a heterodimer to transport not only bile acids but also estrone sulfate, digoxin, and prostaglandin E₂ (Wang et al., 2001d; Seward et al., 2003; Dawson et al., 2005).

5. ATP-Binding Cassette Transporter A1.

ABCA1 is a member of the ABC subfamily A and was cloned in 1999 (Langmann et al., 1999) at the same time that it was determined that defects in ABCA1 cause Tangier's disease, a disorder of impaired cholesterol transport (Bodzioch et al., 1999; Brooks-Wilson et al., 1999; Marcil et al., 1999; Remaley et al., 1999; Rust et al., 1999; Schippling et al., 2008) (Tables 1 and 10). As suggested by the clinical presentation of patients with Tangier's disease, ABCA1 effluxes cholesterol and apolipoprotein A1 in vitro (Neufeld et al., 2001; Wang et al., 2001b). The absence of ABCA1 results in premature atherosclerosis, splenomegaly, and hepatomegaly in patients with Tangier's disease.

Abca1/ABCA1 is highly expressed in placenta, uterus, liver, adrenal glands, small intestine, lungs, and heart (Fig. 5) (Langmann et al., 1999). Within these tissues, ABCA1 is often expressed on macrophages in addition to epithelium (Langmann et al., 1999; Schmitz et al., 1999). In the transfected polarized hepatocyte-like WIF-B cell line, ABCA1 immunostaining is observed along the basolateral surface (Neufeld et al., 2002). These findings suggest ABCA1 participates in the regulation of intracellular cholesterol accumulation in hepatocytes. In addition to cell surface expression, ABCA1 is observed in early and late endosomes, which may participate in protein trafficking as well as shuttling cholesterol to the cell surface for efflux (Neufeld et al., 2001).

6. ATP-Binding Cassette Subfamily G Members 5 and 8.

ABCG5 and ABCG8 are efflux transporters that work in concert as a heterodimer to prevent the absorption of plant sterols (Graf et al., 2003) (Table 1). Both isoforms are necessary for trafficking from the endoplasmic reticulum to the canalicular membrane and, in turn, the excretion of plant sterols and cholesterol into bile (Graf et al., 2003). ABCG5 is predicted to contain six putative transmembrane domains with cytosole-facing amino and carboxyl termini (Lee et al., 2001b). Mouse Abcg5 and -g8 mRNA are detected within the liver and small intestine (Fig. 5) (Lu et al., 2002). Within the mouse small intestine, Abcg5 and -g8 are similarly expressed among the three segments (Dieter et al., 2004). ABCG5 and -G8 mRNA are expressed in human liver, small intestine, and colon (Fig. 5) (Berge et al., 2000; Lee et al., 2001b). Abcg/ABCG5 and -g8 proteins are localized to the apical membrane of enterocytes, cholangiocytes, and hepatocytes (Table 9) (Graf et al., 2003; Klett et al., 2004a).

7. ATPase Copper-Transporting β Polypeptide.

ATP7B is an ATP-dependent copper efflux transporter that is primarily expressed in liver (Petrukhin et al., 1994) (Table 1). Messenger RNA expression of ATP7B is widely expressed in multiple tissues (Fig. 5), and mice lacking Atp7b exhibit copper accumulation in kidney, brain, placenta, and lactating mammary glands (Buiakova et al., 1999). ATP7B protein is localized on the apical membrane of placenta syncytiotrophoblasts (Table 9) (Hardman et al., 2004, 2007). Likewise, ATP7B is responsible for the canalicular excretion of copper into bile (Hernandez et al., 2008). ATP7B localization changes depending upon copper concentrations. At low concentrations of copper, ATP7B is present in the trans-Golgi network. As copper accumulates, ATP7B shifts toward the apical membrane of polarized cells (Roelofsen et al., 2000; Guo et al., 2005; Lutsenko et al., 2007; Weiss et al., 2008).

8. ATPase Class I Type 8B Member 1.

ATP8B1 is an ATP-dependent aminophospholipid transporter, also called the familial intrahepatic cholestasis 1 protein (Ujhazy et al., 2001) (Table 1). It is in the type 4 subfamily of P-type ATPases that are termed flippases (Paulusma and Oude Elferink, 2005). The ATP8B1 protein is predicted to contain 10 transmembrane domains (Paulusma and Oude Elferink, 2005). ATP8B1 mRNA is expressed predominantly in the small intestine, uterus, and pancreas, with moderate expression in the bladder, stomach, prostate, liver, and heart (Fig. 5) (Bull et al., 1998). ATP8B1 is expressed on the canalicular membrane of mouse, rat, and human hepatocytes, where phosphatidylserine and phosphatidylethanolamine are translocated from the outer to the inner leaflet bilayer (Table 9) (Eppens et al., 2001; Ujhazy et al., 2001). Because of ATP8B1 activity, the proportion of sphingomyelin and cholesterol in the outer leaflet is increased and is likely to enhance membrane resistance to bile acid toxicity. ATP8B1 is also expressed along the apical surface of rat enterocytes (Ujhazy et al., 2001).

There are a variety of transporters with similarities and differences in their tissue distribution and substrate affinities. In some cases, certain transporter isoforms are restricted to one or two tissues, whereas other isoforms are broadly expressed in multiple tissues. Additional insight into the function of transporters in various tissue and cell types will be addressed in the next section of this review article.

1. Basolateral Uptake Transporters in Liver.

Ntcp (Slc10a1) is a bile acid uptake transporter that localizes to the basolateral membrane of hepatocytes (Stieger et al., 1994; Keitel et al., 2005). Its function is to take up bile acids (especially taurine-conjugated) into hepatocytes using a sodium gradient (Hagenbuch et al., 1991; Boyer et al., 1994; Hagenbuch and Meier, 1994; Saeki et al., 2002). Down-regulation of rat Ntcp using antisense oligonucleotides almost completely abolishes taurocholate transport (95% reduction) in Xenopus laevis oocytes (Hagenbuch et al., 1996). Four nonsynonymous SLC10A1 (NTCP) polymorphisms exhibit reduced taurocholate, cholate, and estrone sulfate transport in vitro (Table 11) (Ho et al., 2004). Lower transport activity in only one of these variants could be explained by impaired cell surface expression (Ho et al., 2004). NTCP has a limited ability to transport pharmaceuticals (Ho et al., 2006b). In addition to SLC10A1, SLCO1B1 (OATP1B1) may also contribute to bile acid uptake into the liver. SLCO1B1 loss-of-function variants are associated with elevated serum bile acids in vivo (Xiang et al., 2009).

Basolateral uptake transporters are important determinants of liver injury induced by drugs or toxins. Rat Oatp1b2, human OATP1B1, and OATP1B3 transport the phalloidin analog demethylphalloin (Fehrenbach et al., 2003; Meier-Abt et al., 2004). Oatp1b2-null mice are resistant to hepatotoxicity caused by the mushroom toxin phalloidin and the blue-green algae toxin microcystin-LR (Lu et al., 2008). In both instances, resistance to toxicity is a result of reduced hepatic uptake of the toxins. Conversely, inhibition of NTCP-mediated bile acid uptake (as well as BSEP efflux) has also been proposed as a mechanism for hepatotoxicity induced by certain xenobiotics. It is noteworthy that some cholestatic chemicals (such as rifampicin, rifamycin SV, glibenclamide, and cyclosporin) inhibit the transport of taurocholate in NTCP- and BSEP-overexpressing polarized cells (Mita et al., 2006). Likewise, a reduced ability of the hepatotoxin bosentan to inhibit NTCP compared with rat Ntcp may explain differences in hepatotoxicity sensitivity between these two species (human > rat) by causing hepatocellular bile acid accumulation (Leslie et al., 2007).

Because of the high expression of Oatps/OATPs and the critical synthesis of cholesterol in the liver, the ability of OATP1B1 and OATP1B3 to transport anticholesterol drugs (including the statins) is an area of active research. Rifampicin and pravastatin are prototypical substrates of OATP1B1 and -1B3, respectively (Spears et al., 2005; Niemi et al., 2006b; Seithel et al., 2007). Hepatic uptake of both drugs is reduced in Oatp1b2-null mice with dramatic reductions in their liver-to-plasma ratios (Zaher et al., 2008). A conflicting report showed little difference in pravastatin pharmacokinetics between wild-type and Oatp1b2-null mice, but did observe reduced hepatic concentrations of another statin, lovastatin (Chen et al., 2008). In contrast, simvastatin uptake into liver is unchanged in Oatp1b2-null mice, demonstrating distinct differences in hepatic extraction by Oatp1b2 for this class of drugs (Chen et al., 2008). Altered drug disposition in Oatp1b2-null mice demonstrates the utility of this in vivo model for investigating the possible contributions of OATP1B1/1B3 to hepatic transport.

The pharmacokinetics of statins have been investigated in patients with SLCO1B1 (OATP1B1) gene SNPs N130D and V174A. In vitro studies demonstrate normal and reduced pravastatin (as well as estrone sulfate) uptake in N130D- and V174A-overexpressing cells, respectively (Table 12) (Kameyama et al., 2005). The combination of N130D and V174A SLCO1B1 SNPs leads to reduced pravastatin and pitavastatin clearance (Nishizato et al., 2003; Chung et al., 2005). The V174A SNP alone is sufficient to increase plasma concentrations of pravastatin (single dose), primarily in subjects of European American descent, suggesting delayed uptake of pravastatin into liver (Mwinyi et al., 2004; Ho et al., 2007). Furthermore, statins exhibit attenuated efficacy in lowering total cholesterol in patients with the V174A allele (Tachibana-Iimori et al., 2004). It is noteworthy that other SLCO1B1 SNPs have opposite effects. The N130D SNP reduces the area under the curve of pravastatin after a single dose in white subjects, and seems to accelerate OATP1B1-mediated uptake of pravastatin (Mwinyi et al., 2004). Likewise, the P155T variant is associated with greater reduction in low-density lipoprotein cholesterol levels by fluvastatin than in patients with the reference allele, suggesting a gain of function for certain SLCO1B1 alleles (Couvert et al., 2008).

The L543W SLCO1B1 SNP has been detected only in the Japanese population. Although this variant is rare, it has been associated with pravastatin-induced myopathy (Morimoto et al., 2004). More recently, an intronic SNP in SLCO1B1 was also identified as a strong risk factor in patients with statin-induced myopathy, using a genomewide screen (SEARCH Collaborative Group et al., 2008). In that study, more than 60% of myopathy cases in patients treated with a statin could be strongly associated with the variant SLCO1B1 allele (SEARCH Collaborative Group et al., 2008).

Ezetimibe is a cholesterol-lowering drug that inhibits intestinal absorption of cholesterol via the Niemann-Pick C1 like 1 protein. Ezetimibe undergoes glucuronidation and extensive enterohepatic circulation. Ezetimibe-glucuronide inhibits transport of sulfobromophthalein mediated by OATP1B1 and -2B1 (Oswald et al., 2008). Uptake of ezetimibe-glucuronide by OATP1B1 is reduced in cells transfected with the SLCO1B1 V174A variant compared with the wild-type transporter (Oswald et al., 2008). When evaluating ezetimibe single oral dose pharmacokinetics, subjects who are homozygous for the N130D allele exhibit lower bioavailability of ezetimibe, whereas subjects who are heterozygous for the V174A allele have reduced fecal excretion of ezetimibe (Oswald et al., 2008).

Whereas OATPs mediate organic anion uptake into liver, OCT1 is responsible for the influx of organic cations. Using Oct1-null mice, the roles for this transporter in hepatic uptake have been shown. Oct1-null mice have reduced hepatic accumulation and/or biliary excretion of organic cations, such as the model substrate tetraethylammonium, the neurotoxin 1-methyl-4-phenylpyridinium, the anticancer drug metaiodobenzylguanidine, and the antidiabetic drug metformin (Jonker et al., 2001; Wang et al., 2002a; Shu et al., 2007). In addition to pharmacokinetic implications, impaired intestinal absorption and hepatic uptake of metformin have pharmacodynamic consequences (Wang et al., 2002a). Oct1-null hepatocytes are resistant to the glucose-lowering effects of metformin after glucagon challenge (Shu et al., 2007). It is noteworthy that reduced hepatic metformin uptake in Oct1-null mice is associated with lower blood lactate levels compared with wild-type mice, demonstrating that the liver is central to metformin-induced lactic acidosis (Wang et al., 2003a).

Similar to Oct1 in mice, metformin is transported by human OCT1 and OCT2 (Kimura et al., 2005; Song et al., 2008). In vitro analysis of SLC22A1 (OCT1) variants has identified 1 deletion (Met420STOP) and six nonsynonymous polymorphisms (S14F, R61C, S189L, G220V, G401S, G465R) that exhibit reduced metformin uptake (Shu et al., 2003, 2007) (Table 13). These SLC22A1 variants lead to impaired metformin efficacy in lowering blood glucose after an oral glucose challenge (Shu et al., 2007) and increased renal clearance (Tzvetkov et al., 2009). An intronic variant is also associated with the glucose-lowering effect of metformin (Becker et al., 2009). In addition, a study of 24 responders and 9 nonresponders to metformin (as determined by glycosylated hemoglobin A_1c levels) demonstrated that the frequency of the SLC22A1 M408V allele is higher in nonresponders compared with responders (Shikata et al., 2007). Likewise, hepatic OCT1 mRNA levels are lower in livers of M408V carriers (Shikata et al., 2007). Patients with type 2 diabetes who carry the common variant M408V (allelic frequency higher than 10% in the general population) may have an insufficient therapeutic response to metformin therapy because of reduced uptake into liver, which is the major target for reducing circulating blood glucose.

2. Apical Efflux Transporters in Liver.

There are a large number of transporters on the apical surface of hepatocytes that are responsible for the biliary excretion of endobiotics and xenobiotics. Mutations and polymorphisms in canalicular transporters result in genetic disorders with distinct clinical phenotypes and/or marked alterations in chemical disposition.

The ability of Mdr1a/1b, also known as Pgp, to influence the disposition and hepatotoxicity of the environmental toxicant arsenic has been investigated in Mdr1a/1b-null mice (Liu et al., 2002). Administration of sodium arsenite to Mdr1a/1b-null mice yielded interesting findings, the null mice being more susceptible to hepatic injury and mortality than the wild-type mice (Liu et al., 2002). Enhanced susceptibility is probably due to elevated arsenic tissue concentrations in liver, kidneys, small intestine, and brain (Liu et al., 2002). Similar studies in Mdr1a/1b-null mice have been conducted for the mycotoxin fumonisin, but Pgp does not seem to be important for the disposition or toxicity of this chemical (Sharma et al., 2000).

The ability of Mdr2 to transport phospholipids as well as its localization in the canalicular membrane suggested a role for this transporter in protecting the biliary tree from bile acid toxicity by forming mixed phospholipid-bile acid micelles (Elferink et al., 1997). Results from Mdr2-null mice confirmed this hypothesis (Smit et al., 1993; Leveille-Webster and Arias, 1994). Livers from mice lacking Mdr2 exhibit focal hepatocyte necrosis, bile duct proliferation and inflammation, and elevated serum biomarkers of liver injury that are similar to nonsuppurative inflammatory cholangitis (Smit et al., 1993; Mauad et al., 1994). Pathology is more severe in female than male mice, which is thought to be due to the higher levels of hydrophobic bile acids in the bile of female Mdr2-null mice (van Nieuwerk et al., 1997). By 4 to 6 months of age, Mdr2-null mice develop preneoplastic nodules that progress to liver tumors (Mauad et al., 1994). Bile acid excretion into bile is similar in wild-type and Mdr2-null mice, whereas biliary excretion of phospholipids is absent in Mdr2-null mice (Smit et al., 1993; Oude Elferink et al., 1995). It is thought that bile duct proliferation contributes to the enhanced bile acid-independent bile flow in Mdr2-null mice (Oude Elferink et al., 1995; Elamiri et al., 2003). Mdr2-null mice are gaining utility as a rodent model of primary sclerosing cholangitis for identifying the interplay of phospholipids, sterols, and bile acids, as well as testing compounds as novel therapeutics (van Nieuwerk et al., 1997; Voshol et al., 1998; Elamiri et al., 2003; Fickert et al., 2006).

Cholestasis can be caused by genetic defects or as a secondary consequence of hepatobiliary obstruction or destruction. Progressive familial intrahepatic cholestasis (PFIC) represents a group of inherited, autosomal recessive disorders characterized by progressive liver disease with impaired bile flow but without irregularity of the hepatobiliary structure (Table 10). PFIC-III arises from mutations in the human ABCB4 (MDR3) gene, the human ortholog of mouse Mdr2 (de Vree et al., 1998). Patients with PFIC-III display elevated serum γ-glutamyltranspeptidase levels and marked bile duct proliferation. In addition, variants of ABCB4 are associated with the severe form of cholestasis of pregnancy, rare cases of juvenile cholesterol gallstones, and drug-induced hepatocellular and cholestatic injury (Lang et al., 2007; Wasmuth et al., 2007; Nakken et al., 2009; Bacq et al., 2009).

As the “sister” of Pgp, Bsep represents the primary bile acid exporter on hepatocyte canaliculi (Gerloff et al., 1998). Mutations in ABCB11 (BSEP) are responsible for PFIC-II in humans (Table 10) (Strautnieks et al., 1998; Jansen et al., 1999). A recent report identifies more than 10 mutations in the ABCB11 gene, although the functional relevance of these mutations has not been confirmed (Strautnieks et al., 2008). As expected from BSEP dysfunction, patients with PFIC-II present with high serum bile acid concentrations, normal serum γ-glutamyltranspeptidase activity and cholesterol, and low biliary bile acid concentrations. PFIC-II patients are at an increased risk of hepatobiliary malignancy (Knisely et al., 2006). Researchers have identified a number of mutations in ABCB11 that impair BSEP insertion into the apical membrane and therefore reduce taurocholate transport in vitro (Wang et al., 2002b). In addition, autoantibodies against BSEP have been implicated in recurrent graft failure after liver transplantation in a patient with PFIC-II (Keitel et al., 2009).

In an attempt to identify individuals with a genetic predisposition to drug-induced cholestasis or intrahepatic cholestasis of pregnancy, patients with acquired cholestasis have been genotyped for ABCB11 variants. Three highly conserved mutants/variants (V444A, D676Y, G855R) strongly associate with susceptibility to drug-induced cholestasis (Table 14) (Lang et al., 2007). Likewise, the V444A polymorphism is a risk factor for intrahepatic cholestasis of pregnancy in European patients as well as patients with contraceptive-induced cholestasis (Keitel et al., 2006; Dixon et al., 2008; Meier et al., 2008). More recent efforts have identified novel ABCB11 variants in the Japanese population (Kim et al., 2009).

In a surprising turn of events, Bsep-null mice exhibit a relatively mild cholestasis compared with humans lacking functional BSEP (Wang et al., 2001c). Bsep-null mice are viable and fertile but display growth retardation and lower liver weights compared with wild type (Wang et al., 2001c). Canaliculi from Bsep-null mice have dilated lumens, loss of microvilli, and retained biliary material (Wang et al., 2001c). Although the secretion of cholic acid is reduced in Bsep-null mice, total bile acid excretion is not abolished (∼30% of wild-type mice) (Wang et al., 2001c). Feeding a cholic acid-supplemented diet to Bsep-null mice does, however, precipitate a more pronounced PFIC-II-like phenotype (Wang et al., 2003b). A less severe phenotype in Bsep-null mice compared with patients with PFIC-II suggests that mice possess an alternate canalicular bile acid transport system or further hydroxylation of bile acids in the 6 position to increase their hydrophilicity and decrease their toxicity, all of which would compensate for the loss of Bsep. Subsequent research demonstrated the overexpression of Mdr1a and Mdr2 proteins in Bsep-null mice and led to the proposition that Mdr1a transports bile acids, albeit with a lower affinity (Lam et al., 2005). Indeed, in vitro Pgp overexpression is associated with taurocholate transport (Lam et al., 2005). Furthermore, triple-null mice lacking Bsep, Mdr1a, and Mdr1b exhibit a severe degree of cholestasis as evidenced by impaired bile formation, jaundice, and increased mortality (Wang et al., 2009). Mrp2 and Mrp3 proteins are also elevated in Bsep-null mice (to a lesser degree than Mdr1a) and may compensate for bile acid excretion.

Dubin-Johnson syndrome results from mutations in the ABCC2 (MRP2) gene (Kartenbeck et al., 1996;