Variability in Hepatic Expression of Organic Anion Transporter 7/SLC22A9, a Novel Pravastatin Uptake Transporter: Impact of Genetic and Regulatory Factors

ORIGINAL ARTICLE Variability in hepatic expression of organic anion transporter 7/SLC22A9, a novel pravastatin uptake transporter: impact of genetic and regulatory factors

A Emami Riedmaier1,2, O Burk1,2, BAC van Eijck1,2, E Schaeffeler1,2, K Klein1,2, S Fehr3, S Biskup3, S Müller1,2, S Winter1,2, UM Zanger1,2, M Schwab1,2,4 and AT Nies1,2

Human organic anion transporter 7 (OAT7, SLC22A9) is a hepatic transport protein poorly characterized so far. We therefore sought to identify novel OAT7 substrates and factors contributing to variable hepatic OAT7 expression. Using OAT7-expressing cells, pravastatin was identified as a substrate. Hepatic SLC22A9/OAT7 mRNA and protein expression varied 28-fold and 15-fold, respectively, in 126 Caucasian liver samples. Twenty-four variants in SLC22A9 were genotyped, including three rare missense variants (rs377211288, rs61742518, rs146027075), which occurred only heterozygously. No variant significantly affected hepatic SLC22A9/OAT7 expression. The three missense variants, however, showed functional consequences when expressed in vitro. Hepatic nuclear factor 4-alpha (HNF4α) emerged as a major transcriptional regulator of SLC22A9 by a series of in silico and in vitro analyses. In conclusion, pravastatin is the first identified OAT7 drug substrate. Substantial inter-individual variability in hepatic OAT7 expression, majorly driven by HNF4α, may contribute to pravastatin drug disposition and might affect response.

The Pharmacogenomics Journal (2016) 16, 341–351; doi:10.1038/tpj.2015.55; published online 4 August 2015

INTRODUCTION candidates. As inter-individual variability in hepatic expression is The solute carrier family 22 (SLC22) of human transporters consists typical for membrane transporter proteins, and may thus contribute 8,9 of 23 members that cluster together based on sequence to variation in drug response, we next comprehensively analyzed homology and substrate specificities into ‘organic cation trans- the impact of non-genetic and genetic factors on variable OAT7 porters’ (OCTs), ‘organic cation/zwitterion cotransporters’ (OCTNs) expression in 126 human livers of Caucasian origin. Next-generation and ‘organic anion transporters’ (OATs).1–5 A large number of exome sequencing of the SLC22A9 gene was performed to identify functional in vitro studies, as well as, pharmacogenetic studies novel variants. Finally, while non-genetic factors as well as genetic clearly indicate an important role of SLC22 transporter proteins variants were found to have only limited impact on variable hepatic (for example, SLC22A5/OCTN2, SLC22A12/URAT1) in absorption, OAT7 expression in this study, the hepatocyte nuclear factor distribution and elimination of a diverse group of endogenous 4-alpha (HNF4α) was elucidated as a regulatory factor in a series of compounds and of drugs.1–5 in silico and in vitro analyses. The organic anion transporter 7 (OAT7, encoded by SLC22A9)is so far poorly characterized. Since its initial discovery in 2007 as a novel liver-specific transporter localized to the sinusoidal hepato- MATERIALS AND METHODS cyte membrane,6 further functional characterization and studies on See Supplementary Information for detailed description of methods. the inter-individual variability of its hepatic expression have not 6 been undertaken. To date, no clinically relevant OAT7 substrates Cell culture and establishment of OAT7-transfected Flp-In HEK293 have been identified. Moreover, several well-established substrates cell lines and inhibitors for other OATs (for example, probenecid, para- Stable transfection of the Flp-In human embryonic kidney (HEK293) cell fl aminohippurate, nonsteroidal anti-in ammatory drugs, diuretics) line (Invitrogen Life Sciences, Carlsbad, CA, USA) with SLC22A9 was carried 6 failed to inhibit OAT7. Instead, OAT7 was found to preferentially out using the Metafectene Pro Kit (Biontex Laboratories GmbH, interact with sulfate conjugates of xenobiotics (for example, Martinsried, Germany) as described previously,10 with the exception that naphthyl sulfate, minoxidil sulfate) and steroid hormones.6 Because the pOG44 plasmid vector, provided with the Flp-In Complete System Kit, an orthologous murine gene corresponding to human SLC22A9 is was co-transfected with the pcDNA5/FRT/SLC22A9 plasmid, according to lacking,7 no Oat7 knockout mouse model is available that can be kit instructions. used to identify novel OAT7 substrates. Hence, OAT7-transfected cells are required for systematic in vitro analyses. Transport studies In this study, we therefore initially generated OAT7 transfectants Uptake measurements into stable transfected Flp-In HEK293 cells were and investigated an array of xenobiotics including drugs previously carried out as described previously.10 The uptake buffer consisted of identified as substrates for other OATs as potential OAT7 137 mM NaCl, 3 mM KCl, 8 mM Na2HPO4,1mM KH2PO4,1mM CaCl2 and

1Dr. Margarete Fischer-Bosch Institute of Clinical Pharmacology, Stuttgart, Germany; 2University of Tübingen, Tübingen, Germany; 3CeGaT GmbH, Tübingen, Germany and 4Department of Clinical Pharmacology, Institute of Experimental and Clinical Pharmacology and Toxicology, University Hospital Tübingen, Tübingen, Germany. Correspondence: Dr M Schwab or Dr AT Nies, Dr Margarete Fischer-Bosch Institute of Clinical Pharmacology, Auerbachstr. 112, Stuttgart D-70376, Germany. E-mail: [email protected] or [email protected] Received 9 January 2015; revised 26 May 2015; accepted 23 June 2015; published online 4 August 2015 OAT7, a novel pravastatin uptake transporter A Emami Riedmaier et al 342

Figure 1. Generation of an OAT7-expressing cell line and OAT7-mediated pravastatin transport. (a) Immunolocalization of OAT7 (green) in vector-transfected and OAT7-transfected Flp-In HEK293 cells through immunoﬂuorescence microscopy. Bar: 15 μm. (b) Time course of 100 nM estrone-3-sulfate (prototypic substrate) uptake in OAT7-transfected (●) and vector-transfected (Ο) cells. Experiments were repeated thrice and measurements were made in triplicates (± s.e.m.). (c–g) Time course of 1 μM statin uptake in OAT7-transfected (●) and vector-transfected (Ο) cells. Experiments were repeated twice and measurements were made in triplicates ( ± s.e.m.). Signiﬁcance in (c) was tested using the unpaired t-test comparing uptake in OAT7- vs uptake in vector-transfected cells for a given time.*Po0.05, **Po0.01, ***Po0.001. Symbol ▲ in (c) indicates OAT7-dependent uptake as calculated by subtracting uptake in vector-transfected from uptake in OAT7-transfected cells.

11 5.5 mM glucose, pH 7.4. Time-course studies were conducted at various Visceral and Transplantation Surgery at the University Medical Center time points and measurements for Km and Vmax determination were taken Charité, Humboldt University (Berlin, Germany) as described previously after 30-s incubation. (Supplementary Table 1).8 The study was approved by the ethics committees of the Charité, Humboldt University (Berlin, Germany) and Human liver samples the University of Tübingen (Germany) in accordance with the principles of Liver tissue and corresponding blood samples had been collected from the Declaration of Helsinki. Informed consent was obtained from each 126 patients undergoing liver surgery at the Department of General, patient in writing. High-quality total RNA was extracted from liver tissues,

3 Figure 2. Characterization of OAT7-mediated pravastatin transport. Uptake of (a) 270 nM [ H]pravastatin into vector-transfected (V) and OAT7- transfected cells (OAT7) after 30-s incubation and of (b)1μM estrone sulfate after 1 min incubation in the absence or presence of 2 mM estrone sulfate. Data are means ± s.e. from two experiments performed in triplicates. Significance was tested by one-way analysis of variance (ANOVA) with subsequent Bonferroni’s multiple comparisons test. (c) Concentration dependence of cellular pravastatin uptake, determined at an incubation time of 30 s. Pravastatin uptake is corrected for non-specific uptake in control vector-transfected cells. Measurements at all 3 concentrations were made in triplicates and repeated independently four to six times ( ± s.e.m.). (d) OAT7-dependent uptake of 390 nM [ H] pravastatin into OAT7-transfected cells after 30-s incubation in the presence of different cyclosporine A concentrations. Data are from triplicate determinations ± s.d. Significance was tested by one-way ANOVA with Dunnett’s multiple comparisons test using the value for 0 μM as control. *Po0.05, ***Po0.001. reverse-transcribed and used for mRNA quantification through TaqMan Statistical analysis 12 technology as described previously (Supplementary Table 2). Immuno- All statistical tests were two-tailed and statistical significance was defined blot analyses were performed to detect OAT7 protein in membrane as Po0.05. Univariate and multivariate statistical analyses were performed 8 fractions obtained from human liver tissue as described previously using a using the statistics software package R version 3.0.1 (ref. 14) to determine self-designed antibody (Supplementary Figure 1). the percentage of variability in SLC22A9/OAT7 expression explained by non-genetic, genetic and transcription covariates. SLC22A9 sequencing and genotyping Different strategies to identify variants in exonic and flanking intronic RESULTS regions, as well as in the 5’ and 3’ regions of SLC22A9 were used. Next- generation sequencing covered only exonic regions. Variants identified by Establishment of OAT7 transfectants and identification of novel next-generation sequencing were confirmed using Sanger sequencing. substrates Liver samples with particularly high or low OAT7 expression were To identify potential novel OAT7 substrates, stable OAT7- additionally screened for variants in the 5’ and 3’ regions as well as in transfected cells were generated and thoroughly characterized fl fi anking intronic regions through Sanger sequencing. Variants identi ed by according to the recommendations for transporter evaluation in Sanger sequencing were subsequently genotyped in all 126 DNA samples drug discovery and development.15 Using a newly designed by MALDI-TOF mass spectrometry, complemented by three previously ’ 13 polyclonal anti-OAT7 antibody (Supplementary Figure 1), OAT7 reported variants in the upstream 5 region of SLC22A9. Microarray fi technology covered TAG variants in the 5’,3’ and intronic regions protein was con rmed to be expressed in the plasma membrane (Supplementary Information and Supplementary Table 3). of OAT7- but not vector-transfected cells (Figure 1a). Additionally, transport studies with the probe substrate estrone-3-sulfate fi SLC22A9 promoter/reporter gene plasmids con rmed the function of OAT7 (Figure 1b). Next, we investigated whether OAT7 transports endogenous The SLC22A9 gene sequence encompassing the region from − 2831 to fi fi compounds and drugs that have been identi ed as substrates of +123, with respect to the transcriptional start site, was ampli ed by PCR fl from genomic DNA of HepG2 cells, using primers specified in other organic anion transporters, namely uric acid, 5- uorouracil, fi fl paclitaxel, sincalide, atorvastatin, pravastatin, simvastatin and Supplementary Table 4 and cloned into the re y luciferase reporter gene 4,16 vector pGL3-Basic (Promega, Madison, WI, USA). Deletions of the cloned fluvastatin (Figures 1c–g, Supplementary Figure 2). Because SLC22A9 sequence were generated by restriction enzyme digests as only pravastatin displayed a ∼ twofold higher accumulation in described in the Supplementary Information. OAT7- than in vector-transfected cells (Figure 1c), pravastatin was

© 2016 Macmillan Publishers Limited The Pharmacogenomics Journal (2016), 341 – 351 OAT7, a novel pravastatin uptake transporter A Emami Riedmaier et al 344 concluded to be a substrate according to the recent recommenda- linkage disequilibrium was detected (Supplementary Figure 4c). 15 tions for the evaluation of relevant drug transporters. At 270 nM Only the 14 variants with minor allele frequency of ⩾ 2% were pravastatin, a pharmacological relevant concentration reached at included in the univariate analyses (Supplementary Table 9). No the inlet to the liver (Supplementary Table 5), uptake was effect of these variants was found on SLC22A9 mRNA and OAT7 significantly greater in OAT7- than in vector-transfected cells protein expression after correction for multiple testing. (Po0.001, Figure 2a). Uptake of pravastatin and estrone-3-sulfate into OAT7 transfectants was reduced to levels of vector-transfected Transcriptional regulation of hepatic SLC22A9 expression cells by an excess concentration of estrone-3-sulfate (Po0.001, 6 As non-genetic and genetic factors could not explain the inter- Figures 2a and b), a known OAT7 inhibitor, indicating that individual variability of SLC22A9/OAT7 expression in univariate compound uptake is OAT7-mediated. Interestingly, estrone-3- analyses, as the next step, the impact of transcriptional regulation sulfate uptake into vector-transfected cells was also inhibited by on SLC22A9 expression was investigated. A series of promoter excess estrone-3-sulfate indicating the presence reporter gene constructs, representing unidirectional nested of an endogenous transporter. The kinetic parameters of deletions of 2.8 kb of the SLC22A9 5’ upstream region, was OAT7-mediated pravastatin uptake were Km 1.0 ± 0.3 mM and − 1 − 1 transfected into human hepatoma HepG2 cells. SLC22A9 promoter Vmax 2.3 ± 0.3 nmol mg min (Figure 2c). Notably, cyclosporine activity dropped significantly upon deletion beyond position A, which is a potent inhibitor of OATP1B1, OATP1B3 and − 116 (Po0.001, Figure 4a). The region up to position − 49 did not OATP2B1 (ref. 17) and clinically relevant drug, which has been 18,19 demonstrate any promoter activity exceeding the background shown to increase plasma levels of pravastatin, did not affect activity of empty vector pGL3-Basic. Similar results were obtained OAT7-mediated pravastatin uptake (Figure 2d). with human hepatoma Huh7 cells (data not shown). In silico analysis of the region between positions − 116 and − 49 Hepatic OAT7 expression using MATCH20 identified a putative direct repeat (DR) 1 HNF4α Comprehensive analysis of normalized cDNAs from 20 normal and binding site at − 62/ − 50 (Figure 4b), which harbors the recently 21 tumor human tissues indicated predominant expression of described HNF4α-specific binding motif (H4SBM). Gel shift SLC22A9 mRNA in the liver (Figure 3a). To investigate inter- analysis demonstrated in vitro HNF4α-binding to SLC22A9- individual variability in SLC22A9/OAT7 expression, SLC22A9 mRNA H4SBM, which was as strong as to the reference HNF4α-DR1 and OAT7 protein expression levels were studied in 126 selected motif from the APOC3 promoter. Binding was lost through human liver samples. SLC22A9 mRNA expression data were not mutation of SLC22A9-H4SBM. The specificity of binding was normally distributed and showed a 28-fold variability across liver confirmed by competition gel shift experiments using wild-type samples (Figure 3b, Supplementary Table 6). and mutated SLC22A9-H4SBM as competitors (Figure 4b). Immunoblot analysis was used to detect OAT7 in liver samples Moreover, transient transfection of a reporter gene with the (Figure 3c). Glycosylated OAT7 was detected as a band of mutated H4SBM motif, which showed complete loss of promoter approximately 80 kDa (Supplementary Figure 3) and following activity (Figure 4c), demonstrated its functional relevance. SLC22A9- deglycosylation, a band of approximately 60 kDa was detectable. H4SBM was mediating transcriptional activation by endogenous α α OAT7 protein expression was not normally distributed and HNF4 in HepG2 cells, because treatment with the HNF4 showed 15-fold variability across the liver samples (Figure 3d, antagonist BI6015 (ref. 22) resulted in the inhibition of promoter Supplementary Table 6). No relevant correlation between OAT7 activity only in constructs harboring an intact motif (Figure 4d). protein and transcript levels was observed (r = − 0.21). Similar results were obtained in co-transfection experiments with s α fi OAT7 expression was further investigated in cryosections of HNF4 -speci c siRNA (Supplementary Figure 5a). Co-transfection of α human liver from high and low expressers of OAT7 through HNF4 expression and SLC22A9 promoter constructs into HEK293 immunofluorescence microscopy using the anti-OAT7 antibody cells, not expressing the endogenous receptor, additionally fi α (Figure 3e). Intense staining of sinusoidal hepatocyte membrane con rmed that HNF4 mediates the transcriptional activation of was observed in OAT7 high protein expresser liver samples, the SLC22A9 promoter (Supplementary Figure 5b). Treatment of whereas only low levels of fluorescent signal were detected in HepG2 cells with BI6015 further demonstrated that mRNA α the OAT7 low-expresser samples. Canalicular membranes were expression of endogenous SLC22A9 is dependent on HNF4 ,asis the expression of APOC3,abona fide HNF4α target gene (Figure 4e). simultaneously stained with anti-dipeptidyl peptidase 4 (CD26) fi antibody, but did not show any co-staining with OAT7 signal. RPLP0 served as a negative control. These results were con rmed by transfection of HNF4α-specific siRNA (Supplementary Figure 5c). Furthermore, HNF4α and SLC22A9 mRNA expression were Impact of non-genetic factors on OAT7 expression significantly correlated (rs = 0.837; Po0.0001) in the 126 human The influence of non-genetic factors on SLC22A9/OAT7 expression liver samples (Figure 4f) while correlation between HNF4α and was investigated through univariate and multivariate analyses OAT7 protein was not relevant (rS = − 0.26). Furthermore, SLC22A9 (Supplementary Tables 7 and 8) with various clinical and mRNA levels also significantly (Po0.0001) correlated with mRNA demographic data described in Supplementary Table 1. Multivariate levels of drug transporters SLC22A1/OCT1 and SLCO2B1/OATP2B1 analyses indicated an association between OAT7 protein expression and of the drug-metabolizing enzyme CYP3A4 (Supplementary and regular alcohol consumption and reason for liver surgery. Figure 6), previously shown to be directly regulated by HNF4α.23–25 Altogether, these data indicate that HNF4α acts as a major direct Genetic variants in SLC22A9 and their impact on expression transcriptional regulator of hepatic SLC22A9 gene expression. Twenty genetic variants were identified in the SLC22A9 gene by the different sequencing and genotyping methods (Supplementary SLC22A9/OAT7 genotype–phenotype correlation analyses Figure 4a). Allelic frequencies in the 126 Caucasian samples are To comprehensively elucidate the associations between variants listed in Supplementary Table 3. All variants were in Hardy– with minor allele frequency ⩾ 2% in the SLC22A9 genomic region Weinberg equilibrium (P40.05). Six rare variants (minor allele and the expression of SLC22A9/OAT7, multivariate linear regres- frequency, o2%) were detected in a heterozygous manner. Four sion analyses considering different genetic models (co-/dominant, exonic variants were found only in a heterozygous manner: one recessive, additive) were performed (Supplementary Table 10). synonymous variant (rs17654567, p.T215T) and three missense Models were corrected for 10 non-genetic covariates and the two variants (rs377211288, p.R90C; rs61742518, p.T433M; rs146027075, regulatory factors HNF4α and HNF1α. HNF1α was additionally p.I479M; Supplementary Figure 4b). For some of the variants, high considered as a relevant regulatory factor as previously suggested

Figure 3. OAT7 expression and cellular localization in liver samples. (a) SLC22A9 mRNA expression was quantified in 20 different normal and tumor human tissues (cDNA array) by TaqMan real-time PCR technology. Results are presented as means ± s.e. (b) SLC22A9 mRNA expression levels in 126 human liver samples of Caucasian origin, determined through TaqMan technology. Results are presented as histogram (left y axis) in combination with a cumulative frequency plot (right y-axis). (c) Immunoblot analysis of 25 μg of 10 different liver samples. Protein lysate (7.5, 5.0, 2.5 and 1.25 μg) obtained from stable-transfected OAT7 Flp-In HEK293 cells were used to create a standard curve. OAT7 protein was detected at ~80 kDa. (d) OAT7 protein levels in 126 human liver samples were quantified through immunoblot analysis (top band). Results depicted as a histogram (left y axis) in combination with a cumulative frequency plot (right y axis). (e) OAT7 immunolocalization (green) in the sinusoidal membrane, but not canalicular membrane (stained red by CD26 antibody) of human hepatocytes in the liver sample of a high- expresser individual (left) and a low-expresser individual (right). Bar: 20 μm. by Klein et al.26 mRNA expression levels of HNF1α were quantified genetic factors (7%) and SLC22A9 variants (5%) displayed minor by TaqMan real-time quantitative PCR (Supplementary Information). contributions (Figure 5b). Of note, HNF1α showed only 19% HNF4α and HNF1α expression levels showed substantial inter- contribution to the overall SLC22A9 mRNA variability. Inter- individual variability (Supplementary Table 6). None of the genetic individual variability in OAT7 protein expression was only margin- models could explain inter-individual variability of SLC22A9/OAT7 ally explained by non-genetic factors (9%), SLC22A9 variants (7%) expression in human liver. Additionally, we calculated nine and regulatory factors (5%). haplotypes for all 126 samples (Figure 5a). Again no associations with a substantial effect size between haplotypes and SLC22A9/ Functional consequences of OAT7 missense variants OAT7 expression were detected (Figure 5a). Moreover, 10 HNF4α fi 27 We identi ed three SLC22A9 missense variants p.R90C, p.T433M variants, previously genotyped in our liverbank, showed no and p.I479M in our liver cohort (Supplementary Table 3). Because α association with hepatic HNF4 mRNA, SLC22A9 mRNA or OAT7 information on their functional consequences is lacking, we protein levels (Supplementary Table 11). stably expressed these variants in mammalian HEK293 cells, analyzed their expression and subcellular localization, and Contribution of non-genetic, genetic and regulatory factors to measured their transport activity (Figures 6 and 7). Remarkably, variable OAT7 expression we were unable to select a clone for rs61742518 (p.T433M) Inter-individual variability of hepatic mRNA expression of SLC22A9 expressing mRNA similar to OAT7-reference despite screening was primarily explained by HNF4α regulation (56%), while non- more than 45 different clones. In contrast, for p.R90C and p.I479M,

© 2016 Macmillan Publishers Limited The Pharmacogenomics Journal (2016), 341 – 351 OAT7, a novel pravastatin uptake transporter A Emami Riedmaier et al 346 clones with mRNA expression levels similar to OAT7-reference (rs61742518) compared with the reference mRNA (Figure 6b). were readily selected (Figure 6a). Using the Flp-In system, In agreement to reduced mRNA levels, protein levels of p.T433M the gene of interest is always integrated into the same genomic were also lower compared with OAT7-reference (Figure 6c). locus,28 suggesting that rs61742518 results in an unstable mRNA. Despite high mRNA levels in cells expressing variant This is supported by the prediction of the secondary mRNA rs377211288 (p.R90C) and rs146027075 (p.I749M), protein levels structure, indicating a different structure of the variant mRNA were reduced (Figure 6c). Immunoﬂuorescence analysis revealed

The Pharmacogenomics Journal (2016), 341 – 351 © 2016 Macmillan Publishers Limited OAT7, a novel pravastatin uptake transporter A Emami Riedmaier et al 347 that variant p.R90C was mostly retained intracellularly (Figure 6d). DISCUSSION In contrast, variants p.T433M and p.I479M were localized on the In this study, we performed a detailed functional characterization plasma membrane similar to OAT7-reference (Figure 6d). Trans- of OAT7 using newly established and thoroughly characterized port activity of p.R90C was almost completely abolished while p. OAT7-transfected cells. We identified pravastatin as the first drug T433M and p.I479M were able to transport estrone-3-sulfate and substrate of the liver-specific OAT7 transporter, which may pravastatin (Figures 7a and b). Again, the uptake of both contribute to pravastatin disposition under conditions when substrates was reduced to levels of vector-transfected cells in OATP1B1 and OATP1B3, the major pravastatin transporters,29 are the presence of excess estrone-3-sulfate. Considering the protein functionally impaired. Subsequent analyses using a human liver levels of the variants relative to OAT7-reference, p.T433M and p. tissue bank revealed substantial inter-individual variability in OAT7 I479M showed increased transport of both substrates (Figures 7c expression, which can be largely explained by HNF4α, whereas and d). SLC22A9 variants and non-genetic factors apparently have only a

Figure 5. Genotype–phenotype analyses and contribution of genetic, non-genetic and regulatory factors to variable OAT7 expression. (a) Alignment and frequencies of the nine most frequently identiﬁed SLC22A9 haplotypes in the 126 liver samples. The effect size depicts the differences of SLC22A9/OAT7 expression compared with the reference haplotype H1. Haplotypes with frequencies ⩾ 2% are presented. (b) Percentage of inter-individual variability in SLC22A9/OAT7 expression in the 126 liver samples apportioned to non-genetic factors (light gray), HNF4α (red), HNF1α (blue), genetic variants (green) and to the combination of all factors (black) was calculated using multivariate linear regression analyses and stepwise model selection.

Figure 4. Direct transcriptional regulation of SLC22A9 by HNF4α.(a) SLC22A9 promoter reporter gene constructs are shown schematically on the left. Numbers (in base pairs) refer to the transcriptional start site. Results (mean ± s.d.) of respective promoter activities, as determined in transfection experiments in HepG2 cells, are illustrated on the right and are presented in relation to the activity of empty reporter gene vector pGL3-Basic, which was designated as 1. Data were analyzed by one-way analysis of variance (ANOVA) and Dunnett’s multiple comparisons test with significant differences to the − 2831 construct indicated by asterisks. (b) The putative DR1 motif in SLC22A9 is shown on the right, with H4SBM underlined and bases mutated in mut22A9-H4SBM indicated in small letters. A representative gel shift assay using in vitro translated HNF4α bound to radiolabelled oligonucleotides (probes), containing the indicated motifs, is shown on the left. Competition assays were performed with increasing (5-, 25-, 125-fold) molar excess of unlabeled motifs as competitors. Complexes of HNF4α with oligonucleotides are marked by an arrow. (c) HepG2 cells were transfected with the indicated promoter reporter gene constructs, and promoter activities were presented and statistically analyzed as described above. Statistical significant differences to the activity of the − 116 construct are shown by asterisks. (d) Transfected HepG2 cells were treated with 10 μM BI6015 or vehicle only (0.1% DMSO) for 24 h. Promoter activities are shown as fold change by BI6015, with corresponding activities in vehicle-treated cells designated as 1. Data analysis was performed as described above, with differences significant to pGL3-Basic shown by asterisks. (e) mRNA expression changes by 24 h treatment of HepG2 cells with 10 μM BI6015. Expression in vehicle-treated cells was designated as 1. Statistically significant differences from this value, as analyzed by one-sample t- test, are indicated by asterisks. (f) Correlation analysis of HNF4α and SLC22A9 mRNA expression in the set of 126 livers. Expression levels were normalized to the corresponding expression of β-actin. The Spearman rank correlation coefficient (rs) and statistical significance (P) were calculated using GraphPad Prism. *Po0.05, **Po0.01, ***Po0.001.

Figure 6. Generation of cell lines expressing the three OAT7 missense variants identified in the human liver cohort and their functional characterization in vitro.(a) SLC22A9 mRNA levels in Flp-In HEK293 cell lines stably expressing OAT7 was determined through TaqMan real-time PCR. Data are from triplicate determinations ± s.d. Significance was tested by analysis of variance (ANOVA) with Dunnett’s multiple comparisons test using the value of ‘reference’ as control. ***Po0.001. (b) Prediction of secondary structures of the reference mRNA and the variant mRNAs using RNAfold.46 The optimal structures according to the minimal free energy (MFE) prediction are shown. Colors indicate base-pair probabilities ranging from 0 (blue) to 1 (red). Predicted MFE values are − 744.40 (reference), − 746.80 (R90C), − 745.60 (T433M) and − 745.70 kcal/mol (I479M). (c) Immunoblot analysis of 20 μg of protein lysate obtained from Flp-In HEK293 cells stably expressing OAT7 or the respective missense variants. OAT7 protein was detected at ~ 80 kDa. Semi-quantification showed that the expression of R90C, T433M and I479M was 0.18, 15.5 and 24.2%, respectively, relative to OAT7-reference. (d) Immunolocalization of OAT7 and the respective missense variant in stably transfected Flp-In HEK293 cells. Bar: 20 μm.

minor impact. Notably, only three missense variants were identified OATP1B1 and to a much lesser extent OATP1B3 have been by next-generation sequencing occurring at low allelic frequencies identified as the important statin uptake transporters in and only heterozygously. These three missense variants showed hepatocytes,29 but both transporters have been calculated to functional consequences when expressed in an in vitro cell system contribute, depending on the statin, only 20–80% to the hepatic suggesting a potential impact on drug pharmacokinetics in uptake clearance of statins, in the case of pravastatin, 69–83%.29,30 individuals carrying these variants. Our identification of OAT7 as a novel pravastatin transporter From the various statins tested, only pravastatin was trans- suggests that OAT7 may contribute to pravastatin disposition into ported. Comparing the physicochemical properties of the the hepatocytes as well. We hypothesize that OAT7-mediated investigated statins (Supplementary Figure 7), pravastatin has hepatic pravastatin uptake is ancillary when OATP1B1 and OATP1B3 the largest number of different three-dimensional conformers are active. However, when the function of OATPs is compromised indicating a higher flexibility than the other statins. This either via genetic variation and/or drug–drug interaction,31–33 OAT7 conformational flexibility may be required for a substrate to fit may contribute to hepatic pravastatin uptake. In a drug–drug into the substrate-binding site of OAT7. However, more studies are interaction study involving pravastatin and gemfibrozil, a maximum warranted in the future to identify additional OAT7 substrates to increase in pravastatin plasma levels was expected, given that define chemical properties and molecular features that render a gemfibrozil inhibits both OATP1B1 (ref. 34) and OAT3 (SLC22A8),35 compound an OAT7 substrate. which is involved in the renal excretion of pravastatin.36 However,

3 Figure 7. Characterization of pravastatin and estrone-3-sulfate transport by OAT7 variants. Uptake of (b) 270 nM [ H]pravastatin into vector- transfected cells (V) and transfected cells expressing OAT7-reference or the respective variant after 30-s incubation and of (a)1μM estrone-3- sulfate after 1-min incubation in the absence or presence of 2 mM estrone-3-sulfate. Data are means ± s.e. from two experiments performed in triplicates. Significance was tested by one-way analysis of variance (ANOVA) with subsequent Bonferroni’s multiple comparisons test. Values for vector and reference are from Figures 2a and b. (c and d) OAT7-dependent pravastatin and estrone-3-sulfate transport after correction for non-specific uptake in vector-transfected cells and correction for the relative expression of the respective variant. Significance was tested by ANOVA with Dunnett’s multiple comparisons test using the value of ‘reference’ as control. **Po0.01, ***Po0.001.

following combined inhibition of these transporters in a clinical of other important hepatic uptake transporters like OCT1 and OCT3 setting, only a mere twofold change was observed,37 indicating that as well as OATP1B1, OATP1B3 and OATP2B1 varies substantially.8,9 other transporters, such as OAT7, may be clinically important in the Genetic variants responsible for variable hepatic expression and uptake and disposition of pravastatin in case of OATP inhibition. function have been identified in transporter genes like SLC22A1 Furthermore, pharmacokinetic studies have shown that the and SLCO1B1.8,9 Moreover, simvastatin-induced rhabdomyolysis concomitant use of cyclosporine A and pravastatin leads to a 10- has been associated with OATP1B1 genetics41 resulting in fold increase in pravastatin levels, which is considered to be due to consequences for clinical practice.42 the inhibition of OATP1B1 via cyclosporine A as the major Of note, hepatic SLC22A9/OAT7 mRNA and protein expression pravastatin uptake transporter.18,19 However, pravastatin is still varied 28-fold and 15-fold, respectively, across the 126 human effective in reducing serum cholesterol levels in transplant patients liver samples. To elucidate underlying mechanisms for the receiving cyclosporine A19,38 suggesting an alternate pravastatin variability in hepatic OAT7 expression, we considered non-genetic, uptake mechanism in hepatocytes. Because all hepatically expressed genetic and regulatory factors. Whereas non-genetic factors OATPs (that is, OATP1B1, OATP1B3, OATP2B1) are inhibited by (for example, sex, age, pre-surgery medication; Supplementary cyclosporine A as suggested by in vitro studies17 while OAT7 is Table 1) and SLC22A9 variants (Supplementary Tables 9 and 10) apparently not inhibited (Figure 2d), we hypothesize that OAT7 may did not significantly contribute to the inter-individual variability, a contribute to pravastatin uptake into hepatocytes under these series of in silico, biochemical and cell biological analyses were conditions. Future extensive work is required to determine the carried out and led to the identification of HNF4α as a major contribution of OAT7 to overall hepatic pravastatin uptake. transcriptional regulator of SLC22A9. HNF4α is a member of the Given that the hepatocyte is the target cell for statins, inhibiting nuclear receptor superfamily of ligand-dependent transcription the rate-limiting step in cholesterol biosynthesis,39,40 and presum- factors43 and is majorly involved in the regulation of many liver- ing that OAT7 contributes to hepatic pravastatin uptake when specific genes including drug-metabolizing enzymes (for example, OATPs are functionally impaired, detailed knowledge about the CYP450 genes)25 and transporter genes (for example, MRP2, inter-individual variability in hepatic expression of OAT7 and OATP1B1, OCT1).23,24,44 Here, for the first time, HNF4α was shown potential underlying mechanisms explaining this variability are of to have a key role in the transactivation of the SLC22A9 promoter. relevance. It has been previously shown that, in fact, the expression Transcriptional regulation through HNF4α seems to be

© 2016 Macmillan Publishers Limited The Pharmacogenomics Journal (2016), 341 – 351 OAT7, a novel pravastatin uptake transporter A Emami Riedmaier et al 350 predominantly important for SLC22A9 mRNA variability, whereas 7 Jacobsson JA, Haitina T, Lindblom J, Fredriksson R. Identification of six putative OAT7 protein expression seems to be influenced by factors other human transporters with structural similarity to the drug transporter SLC22 family. than those tested in this study. It is plausible that epigenetic Genomics 2007; 90:595–609. regulators might contribute to the observed variable OAT7 8 Nies AT, Koepsell H, Winter S, Burk O, Klein K, Kerb R et al. Expression of organic cation transporters OCT1 (SLC22A1) and OCT3 (SLC22A3) is affected by genetic expression, which needs further investigations. factors and cholestasis in human liver. Hepatology 2009; 50: 1227–1240. Finally, our comprehensive work including next-generation 9 Nies AT, Niemi M, Burk O, Winter S, Zanger UM, Stieger B et al. Genetics is a major sequencing to detect genetic variants in SLC22A9 in the liver determinant of expression of the human hepatic uptake transporter OATP1B1, samples revealed the presence of only three rare missense but not of OATP1B3 and OATP2B1. Genome Med 2013; 5:1. variants, all of which occurred with low to very low allelic 10 Nies AT, Hofmann U, Resch C, Schaeffeler E, Rius M, Schwab M. Proton pump frequencies. This low frequency of missense variants is confirmed inhibitors inhibit metformin uptake by organic cation uptake transporters (OCTs). PLoS ONE 2011; 6: e22163. by the data from a large exome sequencing project involving 4300 11 Ekaratanawong S, Anzai N, Jutabha P, Miyazaki H, Noshiro R, Takeda M et al. 45 unrelated European Americans. Here, a total of 34 rare SLC22A9 Human organic anion transporter 4 is a renal apical organic anion/dicarboxylate variants (Supplementary Table 12 and Supplementary Figure 8) exchanger in the proximal tubules. J Pharmacol Sci 2004; 94: 297–304. were detected including the three missense variants p.R90C, p. 12 Wolbold R, Klein K, Burk O, Nussler AK, Neuhaus P, Eichelbaum M et al. Sex is a T433M and p.I479M identified in our cohort. Because information major determinant of CYP3A4 expression in human liver. Hepatology 2003; 38: – on functional consequences of these OAT7 missense variants is 978 988. 13 Hesselson SE, Matsson P, Shima JE, Fukushima H, Yee SW, Kobayashi Y et al. lacking, we generated cell lines expressing these missense Genetic variation in the proximal promoter of ABC and SLC superfamilies: liver variants. All three missense variants resulted in lower protein and kidney specific expression and promoter activity predict variation. PLoS ONE levels than OAT7-reference and in the case of p.R90C, even led to 2009; 4: e6942. completely abolished transport activity. We hypothesize that 14 R Core Team. R: A Language and Environment for Statistical Computing. http:// individuals homozygous for the variant p.R90C may have largely www.r-project.org 2014. reduced OAT7-mediated transport of estrone-3-sulfate and 15 Brouwer KL, Keppler D, Hoffmaster KA, Bow DA, Cheng Y, Lai Y et al. In vitro methods to support transporter evaluation in drug discovery and development. pravastatin into their hepatocytes. In contrast, p.T433M and p. Clin Pharmacol Ther 2013; 94:95–112. I479M show higher transport activities when considering their 16 Nies AT, Schwab M, Keppler D. Interplay of conjugating enzymes with OATP expression levels relative to OAT7-reference though reduced uptake transporters and ABCC/MRP efflux pumps in the elimination of drugs. protein levels were observed. Expert Opin Drug Metab Toxicol 2008; 4: 545–568. In summary, we identified liver-specific OAT7 as a novel 17 Letschert K, Faulstich H, Keller D, Keppler D. Molecular characterization and inhibition of amanitin uptake into human hepatocytes. Toxicol Sci 2006; 91: pravastatin uptake transporter whose basal expression in the liver – α 140 149. is determined by HNF4 . Substantial inter-individual variability in 18 Park JW, Siekmeier R, Merz M, Krell B, Harder S, Marz W et al. Pharmacokinetics of hepatic OAT7 expression may contribute to pravastatin uptake pravastatin in heart-transplant patients taking cyclosporin A. Int J Clin Pharmacol and disposition under conditions rendering OATPs functionally Ther 2002; 40:439–450. impaired. Potential consequences on drug response have to be 19 Hedman M, Neuvonen PJ, Neuvonen M, Holmberg C, Antikainen M. Pharmaco- elucidated in future clinical studies. kinetics and pharmacodynamics of pravastatin in pediatric and adolescent cardiac transplant recipients on a regimen of triple immunosuppression. Clin Pharmacol Ther 2004; 75:101–109. CONFLICT OF INTEREST 20 Kel AE, Gossling E, Reuter I, Cheremushkin E, Kel-Margoulis OV, Wingender E. MATCH: A tool for searching transcription factor binding sites in DNA sequences. fl The authors declare no con ict of interest. Nucleic Acids Res 2003; 31: 3576–3579. 21 Fang B, Mane-Padros D, Bolotin E, Jiang T, Sladek FM. Identification of a binding motif specific to HNF4 by comparative analysis of multiple nuclear receptors. ACKNOWLEDGMENTS Nucleic Acids Res 2012; 40: 5343–5356. We would like to acknowledge the expert technical assistance of Silvia Hübner, 22 Kiselyuk A, Lee SH, Farber-Katz S, Zhang M, Athavankar S, Cohen T et al. Monika Elbl, Heidi Köhler, Igor Liebermann, Karina Abuazi and Sabine Rekersbrink. We HNF4alpha antagonists discovered by a high-throughput screen for modulators would like to thank the NHLBI GO Exome Sequencing Project and its ongoing studies of the human insulin promoter. Chem Biol 2012; 19: 806–818. which produced and provided exome variant calls for comparison: the Lung GO 23 Saborowski M, Kullak-Ublick GA, Eloranta JJ. The human organic cation Sequencing Project (HL-102923), the WHI Sequencing Project (HL-102924), the Broad transporter-1 gene is transactivated by hepatocyte nuclear factor-4alpha. J 317 – GO Sequencing Project (HL-102925), the Seattle GO Sequencing Project (HL-102926) Pharmacol Exp Ther 2006; : 778 785. and the Heart GO Sequencing Project (HL-103010). This work was supported by 24 Knauer MJ, Girdwood AJ, Kim RB, Tirona RG. Transport function and transcriptional regulation of a liver-enriched Human Organic Anion Transporting Poly- grants from the Robert-Bosch Foundation, Stuttgart, Germany, the 7th FP EU Initial peptide 2B1 transcriptional start site variant. Mol Pharmacol 2013; 83: 1218–1228. Training Network program ‘FightingDrugFailure’ (PITN-GA-2009-238132) and the 25 Tegude H, Schnabel A, Zanger UM, Klein K, Eichelbaum M, Burk O. Molecular Federal Ministry for Education and Research (BMBF, Berlin, Germany) grant 0315755. mechanism of basal CYP3A4 regulation by hepatocyte nuclear factor 4alpha: evidence for direct regulation in the intestine. Drug Metab Dispos 2007; 35: – REFERENCES 946 954. 26 Klein K, Jüngst C, Mwinyi J, Stieger B, Krempler F, Patsch W et al. The human 1 Koepsell H. The SLC22 family with transporters of organic cations, anions and organic anion transporter genes OAT5 and OAT7 are transactivated by hepato- zwitterions. Mol Aspects Med 2013; 34:413–435. cyte nuclear factor-1alpha (HNF-1alpha). Mol Pharmacol 2010; 78: 1079–1087. 2 Burckhardt G. Drug transport by organic anion transporters (OATs). Pharmacol 27 Klein K, Winter S, Turpeinen M, Schwab M, Zanger UM. Pathway-targeted phar- Ther 2012; 136: 106–130. macogenomics of CYP1A2 in human liver. Front Pharmacol 2010; 1:129. 3 Nies AT, Koepsell H, Damme K, Schwab M. Organic cation transporters (OCTs, 28 Sauer B. Site-specific recombination: developments and applications. Curr Opin MATEs), in vitro and in vivo evidence for the importance in drug therapy. Handb Biotechnol 1994; 5: 521–527. Exp Pharmacol 2011; 201: 105–167. 29 Shitara Y, Maeda K, Ikejiri K, Yoshida K, Horie T, Sugiyama Y. Clinical significance of 4 Emami Riedmaier A, Nies AT, Schaeffeler E, Schwab M. Organic anion transporters organic anion transporting polypeptides (OATPs) in drug disposition: their roles in and their implications in pharmacotherapy. Pharmacol Rev 2012; 64:421–449. the hepatic clearance and intestinal absorption. Biopharm Drug Dispos 2013; 34: 5 Stocker SL, Emami Riedmaier A, Schwab M, Giacomini KM. OCT (SLC22A) and 45–78. OCTN family. In: Ishikawa T, Kim RB, König J (eds). Pharmacogenomics of Human 30 Kunze A, Huwyler J, Camenisch G, Poller B. Prediction of OATP1B1 and OATP1B3 Drug Transporters. Wiley, 2013, pp 171–208. mediated hepatic uptake of statins based on transporter protein expression and 6 Shin HJ, Anzai N, Enomoto A, He X, Kim DK, Endou H et al. Novel liver-specific activity data. Drug Metab Dispos 2014; 42:1514–1521. organic anion transporter OAT7 that operates the exchange of sulfate conjugates 31 SEARCH Collaborative Group. SLCO1B1 variants and statin-induced myopathy--a for short chain fatty acid butyrate. Hepatology 2007; 45: 1046–1055. genomewide study. N Engl J Med 2008; 359: 789–799.

The Pharmacogenomics Journal (2016), 341 – 351 © 2016 Macmillan Publishers Limited OAT7, a novel pravastatin uptake transporter A Emami Riedmaier et al 351 32 Niemi M, Pasanen MK, Neuvonen PJ. Organic anion transporting polypeptide 1B1: 39 Goldstein JL, Brown MS. Regulation of the mevalonate pathway. Nature 1990; 343: a genetically polymorphic transporter of major importance for hepatic 425–430. drug uptake. Pharmacol Rev 2011; 63:157–181. 40 Lennernas H, Fager G. Pharmacodynamics and pharmacokinetics of the HMG-CoA 33 König J, Müller F, Fromm MF. Transporters and drug-drug interactions: important reductase inhibitors. Similarities and differences. Clin Pharmacokinet 1997; 32: determinants of drug disposition and effects. Pharmacol Rev 2013; 65:944–966. 403–425. 34 Nakagomi-Hagihara R, Nakai D, Tokui T, Abe T, Ikeda T. Gemfibrozil and its glu- 41 Mosshammer D, Schaeffeler E, Schwab M, Mörike K. Mechanisms and curonide inhibit the hepatic uptake of pravastatin mediated by OATP1B1. Xeno- assessment of statin-related muscular adverse effects. Br J Clin Pharmacol 2014; biotica 2007; 37: 474–486. 78:454–466. 35 Nakagomi-Hagihara R, Nakai D, Tokui T. Inhibition of human organic anion 42 Wilke RA, Ramsey LB, Johnson SG, Maxwell WD, McLeod HL, Voora D et al. transporter 3 mediated pravastatin transport by gemfibrozil and the metabolites The clinical pharmacogenomics implementation consortium: CPIC guideline in humans. Xenobiotica 2007; 37:416–426. for SLCO1B1 and simvastatin-induced myopathy. Clin Pharmacol Ther 2012; 92: 36 Watanabe T, Kusuhara H, Watanabe T, Debori Y, Maeda K, Kondo T et al. Pre- 112–117. diction of the overall renal tubular secretion and hepatic clearance of anionic 43 Hwang-Verslues WW, Sladek FM. HNF4alpha--role in drug metabolism and drugs and a renal drug-drug interaction involving organic anion transporter 3 in potential drug target? Curr Opin Pharmacol 2010; 10:698–705. humans by in vitro uptake experiments. Drug Metab Dispos 2011; 39: 1031–1038. 44 Kamiyama Y, Matsubara T, Yoshinari K, Nagata K, Kamimura H, Yamazoe Y. Role of 37 Kyrklund C, Backman JT, Neuvonen M, Neuvonen PJ. Gemfibrozil increases plasma human hepatocyte nuclear factor 4alpha in the expression of drug-metabolizing pravastatin concentrations and reduces pravastatin renal clearance. Clin Phar- enzymes and transporters in human hepatocytes assessed by use of small macol Ther 2003; 73:538–544. interfering RNA. Drug Metab Pharmacokinet 2007; 22: 287–298. 38 Penson MG, Fricker FJ, Thompson JR, Harker K, Williams BJ, Kahler DA et al. Safety 45 NHLBI GO Exome Sequencing Project (ESP). Exome Variant Server 2014 p. http:// and efficacy of pravastatin therapy for the prevention of hyperlipidemia in evs.gs.washington.edu/EVS. pediatric and adolescent cardiac transplant recipients. J Heart Lung Transplant 46 Lorenz R, Bernhart SH, Honer Zu SC, Tafer H, Flamm C, Stadler PF et al. ViennaRNA 2001; 20: 611–618. Package 2.0. Algorithms Mol Biol 2011; 6:26.

Supplementary Information accompanies the paper on the The Pharmacogenomics Journal website (http://www.nature.com/tpj)