Keywords

1 Introduction

A large number of clinically used drugs are administered orally, from which approximately 40% are cations or weak bases at physiological pH (Neuhoff et al. 2003). For absorption, distribution, metabolism, and elimination (ADME), they need to be taken up into and effluxed from various cell types in the body. Several families of membrane transporters have been recognized to play a role in the transport of organic cations across the plasma membrane. These include members of the solute carrier (SLC) family 22 (organic cation transporters, OCTs) and of the SLC family 47 (multidrug and toxin extrusion, MATEs) (Koepsell et al. 2007). The human SLC22 family can be divided into several subgroups according to substrates and transport mechanisms (Koepsell and Endou 2004) (Fig. 1). One subgroup comprises OCT1, OCT2, and OCT3, which translocate organic cations and weak bases in an electrogenic manner. Human MATE transporters have only recently been identified as proton/cation antiporters participating in the excretion of organic cations in the liver and kidney (Otsuka et al. 2005; Masuda et al. 2006). Alterations in the expression and function of these transporters may significantly contribute to drug pharmacokinetics and the interindividual variability of drug response. This review summarizes current knowledge about the molecular characteristics, tissue distribution, (drug) substrates and inhibitors, drug–drug interactions, and the fast-growing field of pharmacogenomics of human OCT and MATE transporters.

Fig. 1
figure 1_3

Phylogenetic tree of the 23 transporters of the human SLC22 family. Protein sequences were downloaded from the NCBI gene database and aligned with the ClustalX2 program (Larkin et al. 2007). The tree was drawn with the “drawtree” program of the PHYLIP3.67 program package (http://evolution.genetics.washington.edu/phylip.html). The distance along the branches is inversely correlated to the degree of sequence identity. For example, the amino acid sequence identity of OCT1 and OCT2 is 70% and that of OCT1 and URAT1 31%. Electrogenic cation transporters are marked by black boxes, transporters for organic cations and carnitine by gray boxes, and transporters for organic anions by white boxes. Transporters whose function is as yet unknown are unmarked

Full size image

2 Cloning and Molecular Characterization of OCT and MATE Transporters

A large number of physiological and biochemical studies had suggested the presence of different carrier systems mediating the transport of organic solutes in hepatocytes and renal proximal tubule cells (Giacomini et al. 1988; Boyer et al. 1992). However, molecular identification of these transporters succeeded not until molecular biology techniques became available in the late 1980s. The first member of the electrogenic OCT family was isolated from rat kidney by expression cloning (Gründemann et al. 1994). It took another 11 years until Otsuka et al. identified in 2005 human orthologs of the bacterial MATE family as proton/organic cation exchangers responsible for the electroneutral transport of organic cations into bile and urine.

2.1 OCT Transporters

The genes encoding human OCT1 (gene symbol: SLC22A1), OCT2 (SLC22A2), and OCT3 (SLC22A3) are located in a cluster on chromosome 6q26–q27 and have a common structure of 11 coding exons and 10 introns (Koehler et al. 1997; Gründemann et al. 1998; Hayer et al. 1999; Verhaagh et al. 1999; Gründemann and Schömig 2000). The amino acid sequence identity of OCT1 and OCT2 is 70%, and 50% for both OCT1/OCT3 and OCT2/OCT3. OCT orthologs have been cloned from other mammalian species as well (Koepsell et al. 2007) (Fig. 2).

Fig. 2
figure 2_3

OCT1 orthologs in different vertebrates. The phylogenetic tree on the left was constructed from OCT1/Oct1 protein sequences aligned using the ClustalX2 program (Larkin et al. 2007) and drawn with the “drawgram” program of the PHYLIP3.67 program package (http://evolution.genetics.washington.edu/phylip.html). The sequence comparison on the right shows the aligned sequences in the vicinity of amino acid arginine 61, which is highly conserved among species. A genetic variant was identified in human OCT1 that leads to a nonsynonymous exchange of arginine 61 to a cysteine (Kerb et al. 2002; Shu et al. 2003). OCT1-Cys61 shows a reduced in vitro transport function (Kerb et al. 2002; Shu et al. 2003, 2007), is associated with a significant decrease of hepatic OCT1 protein levels (Nies et al. 2009), and affects metformin pharmacokinetics in humans (Shu et al. 2008). For further details see Tables 4–12. The following protein sequences were used for alignments: human NP_003048; orangutan ENSPPYP00000019207; chimpanzee XP_527554; rhesus monkey ENSMMUP00000020546; dog XP_850971; mouse NP_033228; rat NP_036829; cow NP_001094568; pig NP_999154; elephant ENSLAFP00000009760; cat ENSFCAP00000002624; chicken XP_419621; rabbit ENSOCUP00000002189; bushbaby ENSOGAP00000004719; squirrel ENSSTOP00000008083; zebrafish ENSDARP00000048889. Accession numbers are either from the ENSEMBL genome server (http://www.ensembl.org; numbers starting with “ENS”) or from the “Protein” database at http://www.ncbi.nlm.nih.gov/entrez. Sequences from elephant, cat, rabbit, bushbaby, and squirrel are in part incomplete

Full size image

Based on sequence and hydropathy analyses, OCTs have a predicted topology comprising 12 transmembrane helices, an intracellular amino and carboxyl terminus, and a large glycosylated extracellular loop between the first two transmembrane helices (Fig. 3a). The large intracellular loop between transmembrane helix 6 and 7 carries several putative phosphorylation sites that are used for short-term modulation of OCT activity (Koepsell et al. 2007; Ciarimboli 2008). Employing detailed mutagenesis and modeling of the tertiary structure in analogy to the crystallized structure of lactose permease from Escherichia coli (Abramson et al. 2003), several amino acids in the 4th, 10th, and 11th transmembrane helix of rat Oct1 were identified that are involved in substrate and/or inhibitor binding (Gorboulev et al. 1999, 2005; Popp et al. 2005; Sturm et al. 2007; Volk et al. 2009). These amino acids are localized within the center of a large cleft that may exist in an outward- or inward-facing conformation. The cleft contains high- and low-affinity substrate and/or inhibitor binding sites (Popp et al. 2005; Gorbunov et al. 2008; Minuesa et al. 2009; Volk et al. 2009). Whereas the affinities of the low-affinity substrate binding sites are in the same range as the respective Michaelis-Menten constant values, the high-affinity binding sites may have a 10,000-fold higher affinity. The different substrate and inhibitor binding sites overlap and may exhibit competitive or allosteric interactions. Both the low- and high-affinity sites may be inhibitory (Minuesa et al. 2009). High-affinity binding sites may be also involved in transport since for inhibition of organic cation transport different IC50 values may be obtained when the uptake measurements were performed using different substrate concentrations far below the respective K m values (see e.g. Table 2: inhibition of OCT2-mediated MPP uptake by flecainide or quinidine). The existence of various substrate and inhibitor binding sites and the complex interactions between different sites explains why largely different IC50 values were obtained for individual transporters when different substrates were used for transport measurements (see, e.g., Tables 1–2: inhibition of OCT1-mediated TEA uptake versus MPP uptake by dopamine or histamine, or inhibition of OCT1-mediated TEA uptake vs. ASP uptake by quinidine). Many naturally occurring genetic variants of human OCT1, OCT2, and OCT3 exist that encode transporters with changed functions (Sect. 6).

Fig. 3
figure 3_3

Predicted membrane topology models of human OCT1 (a) and human MATE1 (b). Topology prediction was performed with the TMHMM algorithm (http://www.cbs.dtu.dk/services/TMHMM-2.0) and the model was drawn with TOPO2 (http://www.sacs.ucsf.edu/TOPO-run/wtopo.pl). (a) Tree-like structures indicate the location of putative N-glycosylation sites in OCT1. OCT2 and OCT3 have similar predicted secondary structures as OCT1. (b) There are no putative N-glycosylation sites in MATE1. Thirteen transmembrane segments are also predicted for human MATE2 and for most of the orthologs from other mammalian species (Terada and Inui 2008)

Full size image
Table 1 Physiological substrates and inhibitors of OCTs and MATEs
Full size table
Table 2 Clinically used drugs as substrates and inhibitors of OCTs and MATEs
Full size table
Table 3 Other selected xenobiotics as substrates and inhibitors of OCTs and MATEs
Full size table

2.2 MATE Transporters

The human MATE1 gene (SLC47A1) and the MATE2 gene (SLC47A2) are located in tandem on chromosome 17p11.2 and encode proteins of 570 and 602 amino acids, respectively (Otsuka et al. 2005). The amino acid sequence identity of MATE1 and MATE2 is 47.5%. Two additional human MATE2 isoforms have been cloned: MATE2-K (NM_001099646) coding for a 566-amino acid protein and MATE2-B encoding a truncated protein of 220 amino acids (Masuda et al. 2006). Of note, MATE2-K is currently the only isoform in the MATE2 subfamily, for which function has been demonstrated; MATE2-B is nonfunctional and MATE2 function has not been tested (Masuda et al. 2006; Tanihara et al. 2007). MATE orthologs have also been cloned from other mammalian species, including mouse (Otsuka et al. 2005; Kobara et al. 2008), rat (Terada et al. 2006; Ohta et al. 2006), and rabbit (Zhang et al. 2007).

The hydropathy analysis performed by Otsuka et al. (2005) suggested that MATE1 consists of 12 transmembrane helices. However, most of the current topology analysis programs predict 13 transmembrane helices with an extracellular location of the carboxyl terminus (Zhang et al. 2007; Terada and Inui 2008) (Fig. 3b). Immunocytochemical analyses using accessibility of an antibody to a carboxyl-terminal tag in nonpermeabilized cells proved the extracellular location of the carboxyl terminus of rabbit Mate1 (Zhang et al. 2007). Whether this holds true for other MATE orthologs awaits investigation. Several histidine, cysteine, and glutamate residues in different transmembrane helices of human MATE1 and MATE2-K are apparently involved in substrate binding and/or transport (Asaka et al. 2007; Matsumoto et al. 2008). As for the OCTs, naturally occurring genetic variants have been identified in human MATEs that lead to synthesis of functionally impaired transporters (Sect. 6).

3 Tissue Distribution and Subcellular Localization

By screening the abundance of human transcript sequences (“UniGene” database at http://www.ncbi.nlm.nih.gov) one can assess the approximate gene expression pattern for each OCT and MATE transporter. Northern blot and real-time quantitative PCR analyses have revealed the different mRNA expression profiles in more detail (Koepsell et al. 2007; Okabe et al. 2008). In addition to the mRNA expression profiles, knowledge of the protein expression profiles and the subcellular localization of each transporter in distinct cell types of a given tissue are of equal importance, and they have been analyzed to some extent as well. Although each cell is equipped with a number of different transporters, it is of particular interest to identify transporters in the absorptive and secretory cells of the small intestine, liver, and kidney, because these are the major organs of drug absorption, metabolism, and excretion. The combined action of electrogenic OCT uptake and MATE efflux transporters, which function as proton/cation antiporters, results in the transcellular movement of organic cations in the small intestine, liver, and kidney (Fig. 4).

Fig. 4
figure 4_3

Localization of OCT and MATE transporters in human hepatocytes (a) and proximal tubule epithelial cells in the kidney (b). The basolateral localization of OCT1 and OCT3 in hepatocytes and of OCT2 in proximal tubule epithelial cells together with the apical localization of MATE transporters results in the transcellular movement and, thereby, secretion of organic cations into bile and urine. MDR1 P-glycoprotein (ABCB1) is an ATP-dependent efflux pump for organic cations. In addition, OCTN1 (SLC22A4) and OCTN2 (SLC22A5) are present in the luminal membrane of proximal tubule cells, where they may exchange luminal carnitine plus sodium or luminal cations against intracellular cations. An apical OCT1 localization in proximal tubule cells was recently reported and was suggested to be involved in reabsorption of metformin from the urine

Full size image

Because OCTs and MATEs also transport cationic cytostatic drugs such as platinum drugs (see Sect. 4), transporter expression may affect intracellular levels of anticancer drugs and, thus, response to chemotherapy. Therefore, several studies have analyzed transporter expression profiles in cancer-derived cells as well as in normal tissue in comparison to cancerous tissue (Hayer-Zillgen et al. 2002; Zhang et al. 2006; Ballestero et al. 2006; Yokoo et al. 2008; Okabe et al. 2008). Only recently, OCT1 expression was identified as an important clinical determinant of the response to imatinib in chronic myeloid leukemia (Wang et al. 2008a) (see Sect. 6).

3.1 OCT1

Rat Oct1, the first cloned member of the SLC22A family, is strongly expressed in liver, kidney, and intestine (Gründemann et al. 1994). In humans, on the contrary, OCT1 mRNA is most prominently expressed in the liver (Gorboulev et al. 1997; Nishimura and Naito 2005; Jung et al. 2008; Nies et al. 2009). The OCT1 protein has been localized in the sinusoidal (basolateral) membrane of rat and human hepatocytes (Meyer-Wentrup et al. 1998; Nies et al. 2008), where it mediates the uptake of substrates from the blood and, thereby, mediates the first step in hepatic excretion of many cationic drugs (Fig. 4a). Other reported locations of human OCT1 include the lateral membrane of intestinal epithelial cells (Müller et al. 2005) and the luminal (apical) membrane of ciliated cells in the lung (Lips et al. 2005) and of tubule epithelial cells in the kidney (Tzvetkov et al. 2009).

3.2 OCT2

Human OCT2 mRNA is most strongly expressed in kidney (Gorboulev et al. 1997; Nishimura and Naito 2005; Jung et al. 2008), where the OCT2 protein has been localized in the basolateral membrane of proximal tubule epithelial cells (Motohashi et al. 2002; Nies et al. 2008). Analogous to OCT1 in hepatocytes, OCT2 plays an important role in the secretion of organic cations in the kidney by mediating the first step, that is, the uptake of organic cations across the basolateral membrane (Fig. 4b). OCT2 transcripts were also detected in several other human organs, including small intestine, lung, and different brain regions, and the inner ear (Gorboulev et al. 1997; Busch et al. 1998; Lips et al. 2005; Taubert et al. 2007; Ciarimboli et al. 2010). The human OCT2 protein has been localized in the luminal membrane of ciliated epithelial cells in the lung (Lips et al. 2005) and in pyramidal cells of the hippocampus (Busch et al. 1998).

3.3 OCT3

Human OCT3 was initially cloned from a kidney-derived cell line and termed extraneuronal monoamine transporter (EMT) because substrate specificity is similar to monoamine uptake measured in extraneuronal tissues, neuronal expression of OCT3 was not established, and it was not known that monoamines are also transported by OCT2 (Gründemann et al. 1998); for discussion see Koepsell et al. (2003). Unlike OCT1 and OCT2, OCT3 has a broad tissue distribution (Verhaagh et al. 1999; Nies et al. 2009) and transcripts have been detected, among others, in placenta, adrenal gland, liver, kidney, heart, lung, brain, and intestine (Koepsell et al. 2007). The human OCT3 protein was identified in basolateral membrane vesicles from placenta (Sata et al. 2005), in the plasma membrane of normal human astrocytes (Inazu et al. 2003), in the luminal membrane of bronchial and intestinal epithelial cells (Müller et al. 2005; Lips et al. 2005), and in the sinusoidal membrane of hepatocytes (Nies et al. 2009) (Fig. 4).

3.4 MATE1 and MATE2-K

Human MATE1 is strongly expressed in liver and kidney as well as in skeletal muscle, adrenal gland, and testis (Otsuka et al. 2005; Masuda et al. 2006). Immunolocalization analyses identified the MATE1 protein in the canalicular membrane of hepatocytes (Otsuka et al. 2005) and in the luminal membrane of tubular epithelial cells in the kidney (Otsuka et al. 2005; Masuda et al. 2006). Human MATE2-K is almost exclusively expressed in the kidney and is localized in the luminal membrane of proximal tubular epithelial cells (Masuda et al. 2006) (Fig. 4).

4 Functional Characterization of OCT and MATE Transporters

4.1 Common Functional Properties of OCTs

The functional characteristics of OCTs have been studied in detail using cRNA-injected Xenopus laevis oocytes or OCT-transfected mammalian cell lines. Several transport characteristics are shared by all OCTs irrespective of their subtype or the species. OCTs transport a broad range of organic cations with diverse molecular structures exhibiting K m values in the micro- to millimolar range (Tables 1–3). Typically, the relative molecular masses of the substrates are below 500 (Suhre et al. 2005; Schmitt and Koepsell 2005; Ahlin et al. 2008; Zolk et al. 2008). OCTs are electrogenic facilitative diffusion systems that translocate organic cations in both directions across the plasma membrane (Busch et al. 1996; Nagel et al. 1997; Kekuda et al. 1998; Budiman et al. 2000; Lips et al. 2005). Transport of organic cations by OCTs is driven by the electrochemical potential but not accelerated by gradients of sodium or protons. For rat Oct2, a nonselective cotranslocation of inorganic cations together with transported organic cation substrates has been observed under depolarized conditions (Schmitt et al. 2009). OCTs are inhibited by a large number of cations and uncharged compounds that are not transported themselves. Partial or total inhibition of transport activity may be achieved (Volk et al. 2009). Transport inhibition may be competitive, partial competitive, or noncompetitive. Importantly, the affinities of the inhibitors are also dependent on the transported substrate (Tables 1–3). For human OCTs, IC50 values between 10 pM and 24 mM have been determined. Transported substrates and inhibitors of OCTs are of endogenous origin, xenobiotics, and clinically used drugs.

4.2 Substrate and Inhibitor Specificities of Human OCTs

Human OCT1, OCT2, and OCT3 have largely overlapping but distinctly different substrate and inhibitor specificities (Tables 1–3). The substrates of human OCTs (hOCT) are typically organic cations with one positive charge or two positive charges (furamidine and paraquat) or weak bases that are positively charged at physiological pH (Tables 1–3). Noncharged compounds such as cimetidine at alkaline pH (Barendt and Wright 2002) may also be transported. Whether OCTs may be also able to transport organic anions remains to be clarified. Transport of prostaglandins by hOCT1 and hOCT2 has been reported by Kimura et al. (2002) but was not confirmed by Harlfinger et al. (2005).

Transported endogenous substrates of human OCTs include monoamine neurotransmitters, neuromodulators, and other compounds such as choline, creatinine, and guanidine. Among the >120 clinically used drugs that were shown to interact with human OCTs, about 20 were identified as transport substrates (Table 2). These include antineoplastic platinum compounds, the histamine H2 receptor antagonist cimetidine, the antiviral drugs acyclovir, ganciclovir, lamivudine, and zalcitabine, the antidiabetic drug metformin, and the antiarrhythmic drug quinidine. The neurotoxin 1-methyl-4-phenyl pyridinium (MPP), the antidiabetic drug metformin, and the antiviral drug lamivudine are transported with similar affinities by the three human OCT orthologs. The model cation TEA is transported with similar affinities by hOCT1 and hOCT2 but shows low-affinity interaction with hOCT3. At variance, epinephrine and norepinephrine are transported with similar affinity by hOCT2 and hOCT3, and only exhibit low-affinity interactions with hOCT1. Histamine is transported with higher affinity by hOCT3 compared to hOCT2 and is apparently not transported by hOCT1 (Koepsell et al. unpublished data).

Inhibitors of OCTs may have larger molecular weights compared to substrates. They may bind to the central substrate binding pockets of the OCTs or to more peripheral regions in the clefts. Two or more inhibitor molecules may bind at the same time. Transport of a specific substrate may be inhibited partially after inhibitor binding to a high-affinity site and total inhibition may be observed when the inhibitor has bound to the low-affinity site (Minuesa et al. 2009).

It may be difficult to distinguish whether a compound that inhibits an OCT transporter is translocated or not. The reasons are (1) that transport rates may be low, (2) that the expression of endogenous cation transporters may be different in transfected and nontransfected cell lines, and (3) that OCT inhibitors that inhibit control substrates may have different affinities for other substrates. It has to be kept in mind that a correlation between transporter expression and the effect of a drug that interacts with the transporter does not prove that the drug is transported because the transporter inhibition may block cellular uptake of an endogenous compound that may critically influence drug effects on cell functions.

Thomas et al. (2004) observed that compounds that inhibit OCTs decreased uptake of imatinib, a first-generation tyrosine kinase inhibitor, into a human T-cell lymphoblast-like cell line. Similarly, imatinib uptake into blood cells from patients with chronic-phase chronic myeloid leukemia (CML) was blocked by OCT inhibitors (White et al. 2006). When the CML cell line KCL22 was transfected with hOCT1, imatinib uptake was about 1.6-fold higher compared to uptake into control transfectants (Wang et al. 2008a). At variance, expression of hOCT1 in X. laevis oocytes or in human embryonic kidney cells did not lead to a significant increase of imatinib uptake (Hu et al. 2008 and Koepsell, Nies, et al. unpublished data). Independent from the conflicting transport data, it was demonstrated that OCT1 mRNA levels and OCT1 genotype are important clinical determinants of treatment response in CML patients (Wang et al. 2008a; Kim et al. 2009) (Sect. 6.3).

4.3 Drug–Drug Interactions Involving OCTs

Various clinically used drugs were identified as inhibitors of OCT-mediated transport by investigating their potency to inhibit in vitro uptake of transported cations (Table 2). When these inhibitory drugs are coprescribed with drugs that are transported by OCTs, drug pharmacokinetics may be altered. Several studies, therefore, investigated the ability of drugs to inhibit transport of the OCT drug substrates metformin or cimetidine in vitro. For example, OCT2-mediated cimetidine transport is inhibited by ranitidine (Tahara et al. 2005) and OCT2-mediated metformin transport by sodium channel blockers (Umehara et al. 2008), β-adrenergic receptor antagonists (Bachmakov et al. 2009), and cimetidine (Zolk et al. 2009). The oral antidiabetics repaglinide and rosiglitazone inhibit OCT1-mediated metformin transport (Bachmakov et al. 2008).

Clinical studies suggest that drug–drug interactions involving OCTs also occur in vivo and may mainly affect cationic drugs that are predominantly eliminated by renal secretion (Ayrton and Morgan 2008; Kindla et al. 2009). For example, cimetidine decreases the renal tubular secretion of ranitidine (van Crugten et al. 1986), procainamide (Lai et al. 1988), dofetilide (Abel et al. 2000), and varenicline (Feng et al. 2008). The inhibition of tubular secretion of metformin by cimetidine was first described more than 20 years ago (Somogyi et al. 1987), but only recently this drug–drug interaction was attributed to OCT2 (Wang et al. 2008b). Other in vivo drug–drug interactions were reported between lamivudine and trimethoprim and between cisplatin and cimetidine or imatinib. It was shown that renal lamivudine clearance was decreased after coadministration of trimethoprim (Moore et al. 1996) and that the concomitant administration of imatinib has a protective effect against cisplatin-induced nephrotoxicity and ototoxicity (Tanihara et al. 2009; Ciarimboli et al. 2010).

4.4 Common Functional Properties of MATEs

MATE transporters are electroneutral transporters that operate independently of a sodium gradient, but use an oppositely directed proton gradient as driving force; translocation of organic cations across the plasma membrane may occur in both directions (Otsuka et al. 2005; Tanihara et al. 2007). MATEs are apparently the functionally long known but searched for proton-driven cation efflux transporters of the canalicular hepatocyte membrane and the luminal membrane of proximal tubule epithelial cells, which have been functionally described for many years (Koepsell 1998; Otsuka et al. 2005).

4.5 Substrate and Inhibitor Specificities of MATEs

MATE1 and MATE2-K have similar substrate and inhibitor specificities, which overlap with those of OCTs (Tables 1–3). The OCT substrates MPP and TEA are also transported by the two MATE orthologs. Endogenous substrates include the organic cations creatinine, guanidine, thiamine, and also the organic anion estrone sulfate. About 30 clinically used drugs have been shown to interact with MATE transporters, and several were identified as transport substrates such as metformin, cimetidine, oxaliplatin, acyclovir, and fexofenadine (Table 2).

4.6 Drug–Drug Interactions Involving MATEs

Information of drug–drug interactions involving MATEs is currently limited. In vitro, cimetidine inhibits MATE1-mediated transport of fexofenadine (Matsushima et al. 2009) and metformin (Tsuda et al. 2009b). Thus, the clinical observation that metformin tubular secretion is inhibited by cimetidine (Somogyi et al. 1987) may not only be due to inhibition of OCT2-mediated metformin uptake (Wang et al. 2008b) but also to inhibition of MATE1-mediated luminal metformin efflux (Tsuda et al. 2009b).

5 Knockout Mouse Models

Knockout mouse models are valuable tools to identify the physiological and pharmacokinetic roles of transporters in vivo. For that purpose, Oct1 (Jonker et al. 2001; Shu et al. 2007), Oct2 (Jonker et al. 2003), Oct3 (Zwart et al. 2001; Wultsch et al. 2009), and Mate1 (Tsuda et al. 2009a) single-knockout mice and Oct1/Oct2 double-knockout mice (Jonker et al. 2003) have been generated. All strains are viable and fertile and show no apparent phenotypical abnormalities, indicating that none of the transporters is essential for obvious physiological functions in mice. However, the tissue distribution and disposition of endogenous or exogenous organic cations may differ significantly between wild-type mice and the knockout mouse strains. These knockout mouse models may be used for the prediction of pharmacokinetics in humans, especially in those carrying genetic variants that encode transporters with reduced function (Sect. 6).

5.1 Oct1 Knockout Mice

Intravenous injection of the model cation TEA into Oct1(−/−) mice resulted in a fourfold to sixfold reduced hepatic accumulation and in a twofold reduced direct intestinal excretion of TEA in comparison to wild-type mice (Jonker et al. 2001). On the other hand, urinary TEA excretion was increased, probably because lack of hepatic Oct1 leads to increased availability of TEA to the kidney. Similar to TEA, the levels of the anticancer drug meta-iodobenzylguanidine, the neurotoxin MPP (Jonker et al. 2001), and the antidiabetic drug metformin (Wang et al. 2002; Shu et al. 2007) were also lower in livers from Oct1(−/−) mice than in those from wild-type mice. The decreased hepatic metformin uptake resulted in a reduced effect on AMP-activated protein kinase phosphorylation and gluconeogenesis, and, in consequence, the glucose-lowering effect of metformin was completely abolished (Shu et al. 2007). Thus, mouse Oct1 – as well as human OCT1 (see Sect. 6) – is a major determinant of the pharmacodynamic responses to metformin. It is of interest that Oct1(−/−) mice do not develop metformin-induced lactic acidosis, which is a severe and rare adverse drug reaction of metformin treatment in humans (Wang et al. 2003).

5.2 Oct2 Single-Knockout and Oct1/Oct2 Double-Knockout Mice

In contrast to the absence of Oct1, the targeted disruption of the murine Oct2 gene had only little effect on the pharmacokinetics of intravenously injected TEA (Jonker et al. 2003). The hepatic and renal concentrations of TEA and the excretion of TEA in the urine and feces were similar in Oct2(−/−) and wild-type mice. Because Oct1 is expressed in mouse kidney in addition to Oct2 (Alnouti et al. 2006) and Oct1 and Oct2 have overlapping substrate specificities (Gründemann et al. 1999), renal Oct1 expression is apparently sufficient for secretion of most organic cations even in the absence of Oct2. In order to develop a mouse model for studying the renal secretion of organic cations, Oct1/Oct2 double-knockout mice have been generated (Jonker et al. 2003). Renal tubular secretion of TEA was completely abolished and TEA was only eliminated by glomerular filtration in these double-knockout mice, which resulted in significantly elevated TEA plasma levels compared to wild-type mice. Similarly, urinary excretion of cisplatin was significantly impaired in Oct1/Oct2(−/−) mice so that the animals were protected from severe cisplatin-induced renal tubular damage and from cisplatin-induced loss of hearing (Filipski et al. 2009; Ciarimboli et al. 2010).

5.3 Oct3 Knockout Mice

After cloning human OCT3 it was hypothesized that the functional defined corticosterone-sensitive extraneuronal transport activity for monoamine neurotransmitters is mainly mediated by OCT3 (Gründemann et al. 1998; Koepsell et al. 2003). Interestingly, steady-state norepinephrine and dopamine levels did not differ between several tissues from wild-type and Oct3(−/−) mice whereas differences in MPP accumulation were observed (Zwart et al. 2001). Intravenous injection of MPP into Oct3(−/−) mice resulted in significantly reduced MPP levels in heart, but not in small intestine, liver, kidney, brain, and placenta in comparison to tissues from wild-type mice. Moreover, fetuses from pregnant Oct3(−/−) mice had three times lower MPP levels. Because MPP is a substrate of murine Oct1, Oct2, and Oct3, these data suggest a prominent role of Oct3 in the heart and fetoplacental interface, whereas in other tissues the lack of Oct3 is apparently well compensated by the function of other Octs. Although Oct3(−/−) mice did not show overt phenotypical abnormalities, Oct3 is probably critically involved in central nervous function. Vialou et al. (2004) showed that Oct3 is implicated in the appropriate neural and behavioral responses to environmentally induced changes in osmolarity. Whether Oct3 also plays a role in the regulation of fear and anxiety is being discussed (Vialou et al. 2008; Wultsch et al. 2009). Of note, there is compensatory upregulation of Oct3 in the brain of mice that lack the neuronal serotonin transporter Slc6a4/Sert (Schmitt et al. 2003; Baganz et al. 2008).

5.4 Mate1 Knockout Mice

Pharmacokinetic characterization of Mate1(−/−) mice (Tsuda et al. 2009a) was carried out with metformin, a typical drug substrate of human MATE1 (Table 2). After intravenous injection, renal and hepatic metformin concentrations were markedly increased in the Mate1(−/−) mice compared to wild-type mice. In addition, plasma metformin levels were increased in Mate1(−/−) mice, whereas urinary metformin excretion was significantly decreased. These data indicate a crucial role of Mate1 in the renal clearance of metformin and probably other drugs as well.

6 Pharmacogenomics of OCT and MATE Transporters

It is well accepted that drug response to the same medication differs among individuals (Kerb 2006). Besides factors such as age, organ function, concomitant therapy, drug–drug interactions, and the nature of the disease, genetic factors have been recognized as important determinants of interindividual variability of drug response. Because OCTs and MATEs function as drug uptake and efflux transporters, respectively (Sect. 4), genetic variants in these transporters may account for interindividual variability of pharmacokinetics of many drugs (Ho and Kim 2005; Giacomini and Sugiyama 2006; Kerb 2006). At present, major research efforts are being taken to identify OCT and MATE variants, to analyze their potential functional consequences, and to determine their contribution to a patient’s response to pharmacotherapy.

6.1 Identification of Genetic Variants, Their Predicted Consequences, and Their Effects In Vitro

More than 1,100 and 450 single-nucleotide polymorphisms (SNPs) are currently listed for the OCT and MATE genes, respectively, in the NCBI-SNP database (National Center for Biotechnology Information, http://www.ncbi.nlm.nih.gov/SNP; build 130, January 2010). The Pharmacogenetics and Genomics Knowledge Base (PharmGKB, http://www.pharmgkb.org) is another public database comprising data and information related to all areas of pharmacogenetics including a large collection of DNA samples from ethnically diverse populations (Giacomini et al. 2007). Moreover, the International HapMap Consortium (http://www.hapmap.org) has generated a haplotype map of the human genome by identifying more than 3.1 million SNPs genotyped in 270 individuals from four geographically diverse populations (International HapMap Consortium et al. 2007). It is expected that the current next generation sequencing projects aiming at the complete sequencing of 1,000 human genomes (Kaiser 2008; Siva 2008) will identify more variants, especially those with low frequencies between 0.1% and 1% (Ionita-Laza et al. 2009).

Whereas most sequence variants are present in the introns, others are located in the 5′- and 3′-flanking regions and may lead to an altered expression level of the respective OCT or MATE transporter (Ogasawara et al. 2008; Nies et al. 2009; Hesselson et al. 2009; Ha Choi et al. 2009). Sequence variants within the exons (coding SNPs, cSNPs) may result in amino acid substitutions. These nonsynonymous or missense variants are of considerable interest because they may affect the transport function of the OCT and MATE transporters. A comprehensive list of the currently known cSNPs in the genes encoding human OCTs and MATEs are given in Table 4.

Table 4 Characteristics of nonsynonymous SLC22A1-A3/OCT1-3 and SLC47/MATE variants, their predicted functional consequences, and in vitro transport data
Full size table

PolyPhen (polymorphism phenotyping, http://genetics.bwh.harvard.edu/pph2; Ramensky et al. 2002) and SIFT (Sorting Intolerant from Tolerant, http://sift.jcvi.org/; Kumar et al. 2009) are two commonly used algorithms, with which the potential functional effects of single amino acid substitutions can be predicted in silico. Based on multiple sequence alignments and in part on information from known three-dimensional protein structures, the algorithms predict the probability that an amino acid substitution has an impact on protein structure and function. However, these in silico predictions cannot substitute for the experimental analysis of each amino acid variant to proof functional changes of the respective OCT or MATE transporter. For comparison, Table 4 lists the predicted functional consequences as well as in vitro transport data for many of the known nonsynonymous OCT and MATE variants. SIFT and PolyPhen predictions are similar for most variants though they differ for some (e.g., OCT1-Ser14Phe, OCT1-Leu23Val, OCT1-Pro341Leu, OCT2-Lys432Gln, MATE1-Leu125Phe). Moreover, amino acid substitutions that are predicted to be tolerated have no transport activity in vitro (e.g., OCT1-Gly220Val, MATE1-Val480Met). This shows limitations of the in silico predictions, which did not include recent structural analysis data (Popp et al. 2005; Volk et al. 2009). The differences may partly be due to the fact that several variants are not properly incorporated into the plasma membrane but are rather retained intracellularly (Shu et al. 2007; Kajiwara et al. 2009; Chen et al. 2009b). The observations that variants alter transport function in a substrate-dependent manner (e.g., OCT1-Ser189Leu, OCT1-Met420del) illustrate the difficulty to predict complex effects of mutagenesis on functions of polyspecific transporters.

6.2 Interethnic Variability

Geographic, ethnic, and racial differences in the frequency of genetic variants are well known and several examples in the field of ADME genes have been reported as a mechanistic basis for differences in drug response and/or drug toxicity (Schaeffeler et al. 2001; Klein et al. 2005).

Significant ethnic differences exist also in the frequency of allele and genotype distributions of SLC22A1, SLC22A2, and SLC47A1 variants as listed in Tables 5–8. For instance, whereas in European-Americans and Caucasians the allele frequency of the SLC22A1-Arg61Cys polymorphism is approximately 8%, in African-Americans as well as Asian-Americans, no variant subject including 260 individuals tested were identified. In contrast, for the SLC22A1-Pro341Leu variant a significant higher allele frequency was found in African-Americans and Asian-Americans (8% and 17%) than in Caucasians (up to 2%). Finally, the Met408Val polymorphism was detected with high-frequency distributions in all ethnic groups (Caucasians, Africans, Asians). Currently it is unclear whether these differences in allele frequencies between various ethnic groups are of any clinical importance and potentially may render individuals more susceptible to certain xenobiotics and/or environmental factors. For example, aflatoxin B1 is a substrate of OCT1 and it is well recognized that the incidence of hepatocellular carcinoma is significantly more frequent in Asians compared to Caucasians. One may assume that such a difference in disease frequency may be explained by functional relevant genetic variants of the uptake transporter OCT1 that are more common in Asian populations.

Table 5 Allele frequencies of SLC22A1 (OCT1) genetic variants in different ethnic populations
Full size table
Table 6 Allele frequencies of SLC22A2 (OCT2) genetic variants in different ethnic populations
Full size table
Table 7 Allele frequencies of SLC22A3(OCT) genetic variants in different ethnic populations
Full size table

6.3 Phenotype–Genotype Correlations

Currently data on tissue expression of OCTs and MATEs correlated to genetic variants are limited. The only polymorphism identified so far that affects OCT1 expression in human liver on mRNA and protein levels is Arg61Cys (Nies et al. 2009; Table 9) after correction for nongenetic factors (such as cholestasis) and additional SLC22A1 variants. Of note, a total of 36 variants in the SLC22A1 gene were tested including some SNPs, which showed reduced function in vitro (Table 4). It would be of major interest to analyze whether expression of OCT2, which is the predominant OCT uptake transporter in human kidney and involved in renal excretion of several drugs (e.g., metformin), is also influenced by genetic factors.

Table 8 Allele frequencies of SLC47/MATE genetic variants in different ethnic populations.
Full size table
Table 9 Phenotype-genotype correlations of SLC22A1(OCT1) in humans
Full size table
Table 10 Phenotype-genotype correlations of SLC22A2 (OCT2) in humans
Full size table
Table 11 Phenotype-genotype correlations of SLC22A3 (OCT3) in humans
Full size table
Table 12 Phenotype-genotype correlations of SLC47/MATE in humans
Full size table

Several studies addressed the association of OCT genotypes with pharmacokinetic parameters of OCT substrates in humans (Tables 9–12). These investigations were based on initial observations that some variants alter OCT function in in vitro cell experiments (Table 4). A key publication in this field showed that metformin AUC and C max are significantly higher in OCT1-variant healthy subjects as compared to individuals with OCT1 reference gene sequence (Shu et al. 2008). In addition and in line with Oct1 knockout mice experiments, OCT1 variant human subjects revealed poor response to metformin measured by the oral glucose tolerance test (Shu et al. 2007). These data suggested for the first time that OCT1 may be a promising candidate gene for better prediction of the response to the antidiabetic agent metformin. Although some studies including type 2 diabetes patients were subsequently performed, unfortunately the results are inconsistent (Tables 9–12). A second clinically highly relevant agent, which was related to OCT1 expression and activity (White et al. 2007; Wang et al. 2008a), is the tyrosine kinase inhibitor imatinib, a mainstay in treatment of patients with chronic myeloid leukemia (CML). Although one study suggests a significant contribution of the OCT1-Phe160Leu variant related to loss of response to imatinib or treatment failure (Kim et al. 2009), further confirmatory studies are still missing, which are mandatory to support such an association.

Regarding OCT2 variants, the Ala270Ser polymorphism was investigated in several pharmacokinetic metformin studies with discrepant results (Table 10). The study with the most representative number of subjects included (n = 103) did not show any association (Tzvetkov et al. 2009). Interestingly, the OCT2-Ala270Ser variant was also related to a significantly reduced cisplatin-induced nephrotoxicity in patients with solid tumors, which fits to the fact that cisplatin is indeed an OCT2 substrate and OCT2 is highly expressed in human kidney (Filipski et al. 2009).

Although the physiological role of OCTs and MATEs is not fully resolved, it is conceivable that membrane transporters determine intracellular concentration of potentially efficient and/or toxic agents and metabolites. In this context it is plausible to hypothesize that genotype-dependent OCT/MATE expression may also contribute to a certain disease susceptibility. Of interest, susceptibility for diseases was repeatedly related to OCT3 (Table 11), whereas convincing data for both, OCT1 and OCT2, are lacking. The SLC22A3 gene was identified as a potential risk factor for prostate cancer as well as coronary artery disease by genomewide association studies (GWA), including thousands of index cases and confirmed by independent control groups (Eeles et al. 2008; Tregouet et al. 2009).

Taken together, compared with other transport proteins the research on the impact of OCT and MATE variants is only at the beginning. Comprehensive genotype–phenotype correlation studies including different human tissues as well as clinical response data are required in the future.