Genotype-Dependent Effects of Inhibitors of the Organic Cation Transporter, OCT1: Predictions of Metformin Interactions
The Pharmacogenomics Journal (2011) 11, 400–411 & 2011 Macmillan Publishers Limited. All rights reserved 1470-269X/11 www.nature.com/tpj ORIGINAL ARTICLE
Genotype-dependent effects of inhibitors of the organic cation transporter, OCT1: predictions of metformin interactions
G Ahlin1, L Chen2, L Lazorova1, Common genetic variants of the liver-specific human organic cation 2 2 transporter 1 (OCT1; SLC22A1) have reduced transport capacity for Y Chen , AG Ianculescu , substrates such as the antidiabetic drug metformin. The effect of the reduced 3 2 RL Davis , KM Giacomini and OCT1 function on drug interactions associated with OCT1 has not been P Artursson1 investigated and was, therefore, the focus of the study presented here. HEK293 cells expressing human OCT1-reference or the variants R61C, 1Pharmaceutical Screening and Informatics, V408M, M420del and G465R were first used to study the kinetics and Department of Pharmacy, Uppsala University, 2 inhibition pattern of the OCT1 substrate 4-(4-(dimethylamino)styryl)-N- Uppsala, Sweden; Department of þ 14 Biopharmaceutical Sciences, University of methylpyridinium (ASP ). In the second part OCT1-mediated C-metfor- California San Francisco, San Francisco, CA, USA min uptake was studied in the presence of drugs administered concomitantly and 3Center for Health Research Southeast, with metformin. Transport studies using ASP þ showed that the function Kaiser Permanente, Atlanta, GA, USA of the variants decreased in the following order: OCT1-referen- 4 44 Correspondence: ce ¼ V408M ¼ M420del R61C G465R. Variants M420del and R61C Dr P Artursson, Pharmaceutical Screening and were more sensitive to drug inhibition, with IC50 values up to 23 times lower Informatics, Department of Pharmacy, Uppsala than those of the OCT1-reference. Uptake studies using 14C-metformin were University, PO Box 580, SE-751 23 Uppsala, in qualitative agreement with those using ASP þ , with the exception that Sweden. E-mail: [email protected] a larger reduction in transport capacity was observed for M420del. Concomitantly administered drugs, such as verapamil and amitriptyline, revealed potential drug–drug interactions at clinical plasma concentrations of metformin for OCT1-M420del. The Pharmacogenomics Journal (2011) 11, 400–411; doi:10.1038/tpj.2010.54; published online 22 June 2010
Keywords: OCT1; inhibition; genetic variation; metformin; drug–drug interactions
Introduction
The human organic cation transporter 1, OCT1 (SLC22A1),1 which is an integral membrane protein highly expressed at the basolateral membrane of human hepatocytes,2,3 mainly interacts with organic cations.4 The OCT1 gene is highly polymorphic with a large number of nonsynonymous mutations leading to amino-acid changes/deletions that affect membrane localization, function and the transport capacity of the transporter.5,6 Thus, OCT1 uptake of model substrates such as 1-methyl-4-phenylpyridinium (MPP þ ) and tetraethylammo- nium is reduced in some common genetic variants of the SLC22A1 gene,5,7 as is the uptake of drug substrates such as the widely used antidiabetic drug Received 29 September 2009; revised 4 May 8,9 2010; accepted 5 May 2010; metformin. However, the effects of genetic variation on drug inhibition of published online 22 June 2010 OCT1 have not been investigated. Effect of inhibition on genetic variants of OCT1 G Ahlin et al 401
We recently performed a comprehensive in vitro study on The ClogP (octanol-water partition coefficient) and the drug inhibition of human wild-type OCT1 expressed in pKa values, used to calculate the net charge of the HEK293 (human embryonic kidney) cells. Using the model compounds at pH 7.4, were obtained using the ADMET substrate 4-(4-(dimethylamino)styryl)-N-methylpyridinium Predictor 2.3.0 software (SimulationsPlus, Lancaster, CA, (ASP þ ), we investigated 191 drugs for their inhibition USA). The molecular weight and polar surface area for the potential and identified dozens of new inhibitors.4 Most of compounds were generated using DragonX 5.4 (Talete, these were inhibiting OCT1 at concentrations above their Milano, Italy). unbound clinical plasma concentration, and are therefore unlikely to interfere with OCT1-mediated transport of drugs Cell culture such as metformin. We reasoned that, for genetic variants OCT1-mediated uptake and inhibition was investigated in with reduced function, such as OCT1 variants M420del and HEK293 cells stably transfected with the OCT1-reference R61C,5,6 much lower concentrations than those used to (the OCT1 sequence obtained from the HUGO project), inhibit the OCT1-reference would be required to inhibit R61C, V408M, M420del or G465R, with empty vector cells transporter function, increasing the risk for clinically used as negative control. The cells were cultured in 75 cm2 significant drug–drug interactions (DDI) with OCT1. cell culture flasks (Corning Inc., Corning, NY, USA) using To address this issue, we investigated the inhibitory Dulbecco’s modified Eagle’s medium (Gibco, Paisley, UK) pattern of some of the most important OCT1 variants containing 10% fetal bovine serum (Sigma-Aldrich), 100 identified in the human population using the model units per ml PEST (Gibco) and 72 mgml 1 Hygromycin B substrate ASP þ. In total, 24 compounds (substrates, inhibi- (Invitrogen, Carlsbad, CA, USA) at 37 1C in an atmosphere of tors and compounds not interacting with OCT1) were 95% air and 5% CO2 and were subcultured twice a week at studied in cells expressing the fully functional OCT1- 70–90% confluence. Passages between 4 and 30 were used reference (NM_003057) and four genetic variants of for the cell lines in this study. OCT1; R61C (rs12208357), V408M (rs628031), M420del For transport studies, the HEK293 cells were seeded onto (rs72552763) and G465R (rs34059508). We then mimicked poly-D-lysine-coated 96-well plates (Greiner, Frickenhausen, the clinical situation by studying the uptake of the type 2 Germany) at a density of 5 to 7 105 cells per well. The diabetic drug metformin in the presence and absence of experiments were carried out after 2–3 days when the cells concomitantly administered drugs, such as other antidia- had formed a confluent monolayer. betics, cardiovascular drugs and drugs used to treat neuro- pathic pain. The results showed that co-administered GFP-tagged OCT1 variants cationic drugs that are potent inhibitors of OCT1-mediated GFP-tagged clones were constructed for the OCT1-reference, transport inhibited metformin uptake significantly, with R61C, M420del and G465R. The EGFP gene was subcloned
IC50 values clearly lower than the clinically relevant drug from the pEGFP-C1 vector (BD Biosciences, Franklin Lakes, concentrations in M420del, supporting the hypothesis that NJ, USA) into the pcDNA5/FRT vector to generate the GFP- carriers of OCT1 variants with reduced function are more tagged clone, and the different variants were then con- susceptible to DDIs. structed by site-directed mutagenesis (Stratagene, La Jolla, CA, USA). The stable cells transfected with the GFP-contain- ing vectors described above were seeded at 3 105 cells per Materials and methods well on 12-mm poly-D-lysine-coated glass cover slips (BD Biosciences) in 24-well plates. Cells were stained using the Compounds Image-IT LIVE labeling kit (Invitrogen) and fixed in 4% All drugs used as inhibitors were obtained from Sigma- paraformaldehyde according to the manufacturer’s protocol. Aldrich (St Louis, MO, USA) and Tocris Bioscience (Ellisville, Coverslips were mounted in Vectashield antifade solution þ MO, USA). ASP was obtained from Molecular Probes (Vector Laboratories, Burlingame, CA, USA) on glass micro- 14 (Carlsbad, CA, USA). Radiolabeled C-metformin was scope slides and visualized by confocal microscopy using obtained from Moravek (Brea, CA, USA). a Zeiss 510 laser scanning microscope (Carl Zeiss, Oberkochen, Germany). Compound selection The inhibitors and noninteracting compounds included in Biotinylation of the cell surface and western blotting this study were selected from a large structurally diverse data Biotinylation of the HEK cell-surface proteins was performed set previously investigated to examine which of the with Pierce Cell Surface Protein Isolation Kit (Thermo compounds showed OCT1 inhibition.4 Further, the data Scientific, Rockford, IL, USA). HEK cells expressing OCT1 set was complemented with three OCT1 substrates.9,10 The reference and polymorphic variants were biotinylated for final data set used for ASP þ inhibition (n ¼ 24) was 30 min at 4 1C using 10 ml 490 mM sulfo-NHS-SS-biotin composed of OCT1 substrates, inhibitors and compounds solution in 1 phosphate-buffered saline. The biotinylation that did not interact with OCT1 (Table 1 ). The data set was reaction was terminated by adding Tris-HCl resulting in a structurally diverse, covering the structural oral drug space final concentration of 4.9 mM. The cells were collected by well, and including drugs from different therapeutical scraping off cells from four 4 10 cm2 culture plates, using groups (Table 1). 10 ml of phosphate-buffered saline containing 490 mM
The Pharmacogenomics Journal Effect of inhibition on genetic variants of OCT1 G Ahlin et al 402
Table 1 Inhibition of ASP+ uptake by OCT1 substrates, inhibitors and noninteracting compounds
Compound a Category Charge Inhibition of Inhibition of Inhibition of Inhibition of reference (%) b V408M (%) M420del (%) R61C (%)
Strong inhibitors Spironolactone Antihypertensive, antihypokalemic 0 89.4 (±2.01) 89.8 (±1.66) 97.9 (±0.52) 100 (±0.20) Verapamil Antihypertensive, antiarrhythmic, + 69.1 (±7.16) 76.4 (±5.77) 96.5 (±4.21) 100 (±1.07) antianginal Doxazosin Antihypertensive 0 77.5 (±2.86 84.5 (±2.89) 99.7 (±0.72) 100 (±0.52) Bucindolol Antihypertensive, congestive + 50.8 (±8.34) 61.2 (±3.77) 84.7 (±2.64) 98.2 (±1.34) heart failure Desipramine Antidepressant + 74.7 (±7.56) 68.6 (±4.88) 76.2 (±5.33) 92.6 (±1.83) Propafenone Antiarrhythmic + 76.6 (±3.19) 65.6 (±4.83) 82.5 (±4.86) 93.4 (±1.58) Ketoconazole Antifungals 0 56.7 (±5.33) 50.5 (±5.93) 83.5 (±23.3) 98 (±0.95) Chlorpromazine Antipsychotic, antiemetic, sedative + 50.2 (±5.85) 48 (±6.35) 57.4 (±8.77) 87.3 (±1.62)
Weak inhibitors Prazosin Antihypertensive + 16.2 (±9.85) 10.7 (±13.1) 37 (±8.73) 79.8 (±3.91) b-Estradiol Estrogen, antineoplastic 0 9.4 (±11.1) 12.4 (±10.2) 12.6 (±13.8) 49.4 (±9.81) Papaverine Vasodilator 0 15.4 (±16.1) 16.2 (±8.29) 39.6 (±9.37) 56.8 (±8.06) Tramadol Analgesic + 40.0 (±6.45) 45.1 (±6.25) 53.9 (±5.84) 57.7 (±11.5) Trimethoprim Antibacterial 0 27.0 (±13.8) 24.7 (±10.4) 29.0 (±9.86) 43.0 (±7.56) Quinidine Antiarrhythmic, antimalarial + 0(±15.4) 1.5 (±12.1) 20.3 (±11.3) 45.6 (±8.74) Clonidine Analgesic + 41.0 (±16.9) 43.8 (±12.0) 57.2 (±5.62) 73.6 (±4.96) Ondansetron Antiemetics 0 35.0 (±7.78) 37.0 (±9.79) 42.2 (±8.52) 56.4 (±7.31) Amiloride Antihypertensive, antihypokalemic Ampholyte 3.3 (±27.6) 0.0 (±36.6) 31.6 (±26.0) 53.2 (±23.2)
Non interacting Captopril Antihypertensive, vasodilator, + 0.0 (±18.3) 0.0 (±15.8) 0.0 (±16.4) 0.0 (±22.4) congestive heart failure Acetylsalicylic acid Analgesic, anti-inflammatory, + 0.0 (±11.8) 0.0 (±15.9) 0.0 (±14.8) 0.0 (±18.7) antipyretic, antirheumatic, antithrombotic Pravastatin Lipid-lowering drugs, antihyperlipidemic + 0.0 (±19.6) 0.0 (±9.70) 0.0 (±15.6) 0.0 (±17.9) Sulfasalazine Antirheumatic, inflammatory bowel disease + 1.5 (±24.3) 0.0 (±23.0) 29.2 (±12.7) 2.3 (±32.3)
Substrates N-1-Methyl-nicotinamide Xenobiotic, model cation, nicotinic + 0.0 (±16.0) 0.0 (±11.6) 33.1 (±25.1) 92.8 (±12.9) acid deficiency Metformin Antidiabetic, antihyperglycemic + 0.0 (±12.5) 0.0 (±11.8) 1.5 (±14.2) 11.9 (±35.5) Tetraethylammonium Xenobiotic, model cation + 0.0 (±23.4) 0.0 (±17.4) 1.7 (±23.7) 0.0 (±13.2) a The 24 compounds investigated for OCT1 inhibition at 50 mM in OCT1-reference, V408M, M420del and R61C. The previously identified OCT1 inhibitors were defined as strong inhibitors when they inhibited ASP+ uptake above 50%, whereas inhibitors showing below 50% inhibition were defined as weak inhibitors. bClassification of inhibition: dark gray, above 90%; gray, 75–90%; light gray, 50–75%; white, below 50%. oxidized glutathione, into 50 ml tubes. The cells were cence reagents (GE Healthcare, Piscataway, NJ, USA). centrifuged and the pellets were lysed on ice for 30 min, Primary antibodies were directed against sodium potassium using 1 ml of lysis buffer with protease inhibitor cocktails ATPase antibody (1:1000; Abcam, Cambridge, MA, USA), (Sigma, St Louis, MO, USA). The cell lysate was centrifuged and hOCT1 (1:500; Abcam). The quantification of western (16 100 g, 10 min) and protein concentration was deter- blot bands used the ImageJ software (http://rsb.info.nih. mined with BCA Protein Assay kit (Thermo Scientific). gov/ij/index.html). Each band from anti-hOCT1 was Clarified cell lysate (800 ml) was added to NeutrAvidin normalized to its loading control, anti-Na þ /K þ ATPase. Agarose slurry and unbound proteins were removed by washing buffers. SDS–PAGE Sample Buffer (200 ml; Bio-Rad, Fluorescence-based assay Hercules, CA, USA) containing 50 mM DTT was added to the Uptake into HEK293 cells was measured using the fluor- gel for 60 min at room temperature with end-over-end escent cation ASP þ as the model substrate for OCT1.4,11 The mixing on a rotator to elute the captured proteins. Columns assay was performed as described previously but at 20 min were centrifuged for 2 min at 1000 g and the samples were incubation time.4 The ASP þ uptake in empty vector cells assayed using western blot. Protein (50 ml) was added to the was subtracted from the ASP þ uptake from that in OCT1- 10% SDS–PAGE (Bio-Rad) and transferred to nitrocellulose transfected cells to assess the OCT1-specific uptake. All membranes. The membranes were blocked overnight at 4 1C ASP þ uptake studies were carried out at 37 1C on a Freedom with Tris-buffered saline with 0.05% Tween 20 in 5% nonfat EVO200 liquid handling workstation (Tecan, Ma¨nnedorf, milk. Immunoblotting was performed following standard Switzerland); the cellular ASP þ was measured with a procedures, and the signals were detected by chemilumines- Saphire2 plate reader (Tecan). The HEK293 cells transfected
The Pharmacogenomics Journal Effect of inhibition on genetic variants of OCT1 G Ahlin et al 403
with OCT1-reference and V408M, M420del and R61C, incubating with 100 ml2M HCl for 30 min at 37 1C. The cell respectively, showed an 8, 8, 7.5 and 6 times higher uptake solution (100 ml) was transferred to a new plate, supplemen- of ASP þ than that in the vector-transfected cells with 1 mM ted with 150 ml scintillation cocktail (PerkinElmer, Waltham, ASP þ (data not shown). Furthermore, the assays for the MA, USA) and analyzed using a Topcount NXT (PerkinEl- variants had Z-values between 0.47 and 0.70 (an excellent mer). The scintillation data were normalized to the protein assay, 14Z 40.5), which implies that our assay setup is content of each well using a BCA protein assay reagent kit suitable for screening purposes.12 Therefore, it was (Pierce Biotechnology, Rockford, IL, USA). concluded that the transfected HEK293 cells constitute a good model for studying OCT1 interactions. Determination of kinetic parameters of metformin The OCT1-mediated metformin uptake was calculated by Determination of kinetic parameters of ASP þ subtracting the metformin uptake levels in HEK293-empty The OCT1-mediated ASP þ uptake was calculated by vector cells from the uptake in OCT1-reference and in subtracting the ASP þ uptake levels in HEK293 empty vector variants V408M and M420del. The R61C was excluded from cells from the uptake in the OCT1-reference, R61C, V408M, this analysis as a result of the low uptake of metformin. M420del and G465R. Kinetic measurements were performed in OCT1-reference, Kinetic measurements were performed to obtain the V408M and M420del to obtain the apparent Km and Vmax apparent Km and Vmax parameters in the OCT1-reference, parameters. Metformin solutions containing 0.12, 0.24, R61C, V408M, M420del and G465R. The ASP þ stock 0.49, 0.98, 1.96, 3.92, 7.83 and 15.67 mM were prepared, solution was serially diluted with Hank’s balanced salt and the initial uptake rates, for 5 min, were plotted and solution buffer to 0.5, 1, 2, 5, 10, 20, 30, 40 and 50 mM. fitted to the Michaelis–Menten equation using GraphPad The initial uptake rates, for 5 min, were plotted and fitted to Prism version 4.02 (GraphPad). the Michaelis–Menten equation using GraphPad Prism version 4.02 (GraphPad, San Diego, CA, USA). Inhibition of metformin uptake We used the OCT1 substrate metformin at a concentration þ Inhibition of ASP uptake corresponding to Cmax—9.66 mM for the generation of Inhibition experiments were performed, initially at a single inhibition curves.13 Drugs co-administrated with metformin inhibitor concentration of 50 mM, to screen for compound- in patients with type 2 diabetes were identified using a induced inhibition of the 1 mM ASP þ uptake4 in OCT1- registry of 11 319 US patients treated with metformin. The reference, R61C, V408M and M420del. Drugs were dissolved drug combinations used in more than 1% of the patient in dimethyl sulfoxide (final concentration 0.5%; Sigma- population were investigated and IC50 values were deter- Aldrich) and diluted with Hank’s balanced salt solution mined. Inhibitor concentrations were chosen to cover the 14–18 buffer (with calcium and magnesium, 10 mM HEPES, and 1% reported Cmax of the compounds investigated. Further, fetal bovine serum, pH 7.4). To assess the effects of genetic a physiological level of human serum albumin (HSA, 4%) variation in the gene coding for OCT1, we studied the was included in all solutions to model the protein binding of concentration-dependent inhibition for eleven inhibitors the compounds investigated.19 (eight strong and three weak) for OCT1-reference, M420del The IC50 values of the compounds were determined by and R61C. The curves were generated using a 5-min-long fitting the data, using GraphPad Prism version 4.02 incubation time and a similar assay setup as for the single- (GraphPad). point measurements. The drugs were tested at 0.2, 0.5, 1, 2, Statistics 5, 10, 25, 50, 100, 200, 500 mM and the IC50 values for the compounds were assessed by fitting the data using Graph- In general, the experimental data were generated on two Pad Prism version 4.02 (GraphPad). independent occasions. t-Tests, using GraphPad Prism version 4.02 (GraphPad), were used to determine the Determination of cellular uptake of 14C-metformin significance levels for the data. The 14C-metformin uptake into HEK293 cells was measured to investigate the uptake of the therapeutically relevant Results substrate metformin in OCT1. When the seeded cells reached confluence, the growth medium was removed and Characteristics of selected variants the cells were washed twice with 200 ml371C Hank’s The genetic variants of OCT1 investigated in this study were balanced salt solution. The wells were incubated with chosen on the basis of their allele frequency and the 100 ml incubation solution (metformin and test compounds) function of the protein. The OCT1 sequence, as described and, after exactly 5 min, 200 ml ice-cold Hank’s balanced salt by the HUGO project, was defined as the OCT1 with normal solution was added to terminate the uptake. The cells were function and the function of polymorphic variants was washed additionally four times to eliminate extracellular compared with this OCT1-reference.20 Only genetic variants metformin, whereupon 50 ml trypsin solution was added for showing allele frequencies above 1% in a population were 30 min at 37 1C to detach the cells from the well surface. considered. For instance, the most common variants, OCT1- Thereafter, 100 ml2M NaOH was added at 37 1C for 2 h to V408M and M420del, have respective allele frequencies of lyse the cells. Neutralization of the sample was ensured by 68.2 and 10.5% in the US population. The corresponding
The Pharmacogenomics Journal Effect of inhibition on genetic variants of OCT1 G Ahlin et al 404
frequencies for the other two variants in this study, OCT1- ASP þ uptake by OCT1 variants in HEK293 cells R61C and G465R, are 3.1 and 1.6%, respectively, in the same The uptake of 1 mM ASP þ differed between the OCT1 population. The uptake of MPP þ and metformin by the variants. Variants V408M and M420del transported ASP þ most common variant V408M was comparable to the OCT1- at the same rate as the OCT1-reference, whereas the reference, whereas M420del showed normal uptake of transport by the R61C variant was significantly reduced MPP þ, but reduced uptake of metformin.5,6 The OCT1- and variant G465R was nonfunctional (Figure 2a). These R61C and G465R variants were included as they show a results were supported by kinetic parameters, obtained from strongly reduced uptake of both MPP þ and metformin.5,6 experiments on the concentration-dependent uptake of þ ASP (Figure 2b). Thus, the Vmax/Km was comparable for the OCT1-reference, variant V408M and M420del, whereas the transport efficiency of R61C was significantly reduced Subcellular localization of OCT1 variants (Figure 2b; Table 2). The G465R data did not fit the Images from GFP-tagged OCT1 proteins were used to Michaelis–Menten equation, which further confirms the determine the subcellular localization of the OCT1-refer- nonfunctionality of G465R, therefore it was excluded from ence, R61C, M420del and G465R (Figure 1a). The character- further studies. In addition, the similar ASP þ kinetics for the istics of the images for OCT1-reference, R61C and G465R OCT1-reference, V408M and M420del indicated that these corresponded to previously published images.5,6 Variant three variants had comparable access to the ASP þ substrate. R61C showed a diffuse intracellular localization and low expression at the cell membrane, whereas the nonfunctional G465R was only expressed in a subset of the cells at a low Characteristics of the OCT1 inhibitors level. Variant M420del showed clear membrane localization, The data set used to inhibit ASP þ transport by OCT1 (n ¼ 24) comparable to that of the OCT1-reference (Figure 1a). The was structurally diverse, covering the structural oral drug results were confirmed by cell-surface protein biotinylation space well, including compounds that did (substrates and followed by western blotting using OCT1-specific variants inhibitors) and did not interact with OCT1 (Figure 2d; (Figure 1b). The OCT1-reference and variants V408M and Table 1). All of the selected strong inhibitors had shown M420del had comparable cell-surface levels whereas the more than 50% inhibition of wild-type OCT1 in an earlier remaining two variants (R61C and G465R) had a signifi- study.4 The compounds were either cations or neutral at pH cantly lower expression on the cell surface. 7.4 and the molecular weight distribution ranged from
GFP only
OCT1-reference hOCT1
Na+/K+ ATPase
antibody
100 OCT1-M420del 80 60 40 20 (densitometry) OCT1-R61C 0 % of OCT1-reference
OCT1-G465R
Figure 1 (a) Confocal fluorescence images of HEK293 cells transfected with GFP only. HEK293 cells stable transfected with GFP-tagged OCT1-reference, M420del, R61C and G465R. The specific GFP staining is shown in the left images as green staining whereas plasma membrane was stained (red staining) using Alexa Fluor 594 conjugated to wheat germ agglutinin. The images to the right show the overlap (yellow staining) between GFP-specific staining and membrane staining. (b) Quantification of OCT1 polymorphisms on the plasma membrane by surface biotinylation. The OCT1-reference was used as the control and set as 100%. The density of each band was normalized to its loading control (Na þ /K þ ATPase).
The Pharmacogenomics Journal Effect of inhibition on genetic variants of OCT1 G Ahlin et al 405
Figure 2 (a) The ASP þ uptake (1 mM) in common OCT1 variants relative to uptake in OCT1-reference. A significant reduction in ASP þ uptake was seen for G465R and R61C, whereas V408M and M420del showed uptake similar to OCT1-reference. (b) Michaelis–Menten kinetic curves for the uptake of ASP þ in the investigated OCT1 variants. (c) The concentration-dependent effect of the OCT1 inhibitor verapamil on the uptake of the ASP þ .(d) The 24 compounds investigated for inhibition of OCT1 (black dots) projected in a principal component analysis of all oral drugs registered in Sweden (gray dots). *Po0.05, **Po0.01.
Table 2 The Michaelis–Menten kinetic parameters of ASP+ and metformin uptake in HEK293 cells transfected with OCT1- reference, V408M, M420del and R61C
ASP+ Metformin
a 1 c a 1 c Km (mM) Vmax (RFU min Vmax/Km Km (mM) Vmax (nmol min Vmax/Km per mg protein)b per mg protein)b
Reference 9.21 (±0.9) 23900 (±708) 2600 (±330) 5.45 (±0.29) 31.9 (±0.72) 5.86 (±0.78) V408M 8.81 (±1.8)NS 22200 (±1350)NS 2530 (±660)NS 6.36 (±0.49)* 34.2 (±1.18)* 5.38 (±0.127)NS M420del 11.06 (±1.4)NS 21000 (±850)* 1900 (±310)NS 5.19 (±0.16)NS 11.6 (±0.16)*** 2.23 (±0.23)*** R61C 5.64 (±0.9)** 6140 (±260)*** 1090 (±240)** — — —
Abbreviation: NS, not significant. aThe apparent substrate affinity. bMaximal transport rate. c Vmax/Km is a measurement of transport efficiency. *Po0.05, **Po0.01, ***Po0.001. Kinetic parameters for R61C for metformin were not investigated.
129.2 g mol 1 (metformin) to 531.5 g mol 1 (ketoconazole) Inhibition of ASP þ uptake with an average of 307.9 g mol 1. The data set showed a First, the inhibition of ASP þ uptake (1 mM) was compared for mean ClogP of 2.07, ranging from 2.59 (tetraethylammo- 24 compounds at a single concentration of 50 mM in the nium) to 5.28 (chlorpromazine). The polar surface area OCT1-reference, V408M-, M420del- and R61C-expressing ranged from 0 (tetraethylammonium) to 156.8 A˚ 2 (amilor- cells. All inhibitors that showed more than 50% inhibition ide) with a mean of 69.0 A˚ 2. for the OCT1-reference were defined as being strong
The Pharmacogenomics Journal Effect of inhibition on genetic variants of OCT1 G Ahlin et al 406
inhibitors, whereas the remaining inhibitors were said to be (IC50-OCT1-reference/IC50-OCT1 variant) were obtained weak (Table 1). The variants M420del, and in particular for prazosin, the inhibition ratios for spironolactone and R61C, were significantly more susceptible to inhibition than chlorpromazine were 6.8 times lower and 1.9-fold higher in the OCT1-reference and V408M (Table 1). The higher the R61C variant in comparison to the M420del variant susceptibility to inhibition in the R61C and M420del (Table 3). The variable inhibitor specificity for each variant variants was also evident for the weaker inhibitors, where was further shown by rank-order correlations of the IC50 it was manifested as a mean inhibition of 35.9 and 57.3% for values. For instance, the Spearman’s rank-order correlation
M420del and R61C, respectively, compared with 20.8 and between the IC50 values of the OCT1-reference and OCT1- 21.3% for the OCT1-reference and V408M, respectively M420del was 0.56 (data not shown). These results indicate (Table 1). The differences between the OCT1-reference and that caution should be exercised in generalizing results the R61C variant were significant (Po0.001) whereas the obtained with a single inhibitor. differences between the OCT1-reference and variants M420del and V408M were almost significant (P ¼ 0.0571) Metformin uptake by OCT1 variants and not significant, respectively. As expected, the noninter- Metformin uptake by OCT1-M420del and R61C was sub- acting compounds had no effect on the ASP þ transport by stantially decreased compared with the OCT1-reference any of the OCT1 variants (Table 1). Thus, it was not only (Figure 3a), which is in agreement with previous results.6 variant R61C, which showed reduced transport efficiency of In comparison with the ASP þ, the apparent affinity of ASP þ, but also the common genetic variant M420del, which metformin was about 600 times lower in the OCT1-reference transported ASP þ at a rate and to an extent comparable (Table 2). Further, in contrast to the results obtained with to that of the OCT1-reference, that was more sensitive to ASP þ, the transport of metformin was significantly reduced transport inhibition. in the M420del variant compared with the OCT1-reference In the next series of experiments, IC50 values were (Figure 3a). In agreement with previously reported data, the generated for 11, mainly strong, inhibitors. Because the uptake of metformin by variant V408M was comparable to variant V408M showed normal function and an inhibition that of the OCT1-reference (Figure 3a).6 The low transport pattern comparable to that of OCT1-reference, it was not capacity of OCT1-R61C (Figure 3a) resulted in inconclusive further investigated for ASP þ uptake. In concert with the kinetics and low resolution of the inhibition experiments single-point measurements, the calculated IC50 values were (data not shown) and, therefore, this rather rare variant was clearly lower for the M420del, and even more so for R61C excluded from further studies. Thus, in the following, the than for the OCT1-reference (Table 3). OCT1-R61C showed inhibition of the OCT1-reference was compared with that of an IC50 that was between 3 and 23 times lower for the V408M and M420del, the most common variants of OCT1 5 compounds than for the OCT1-reference, and the IC50 for in the US population. OCT1-M420del was between 1 and 14 times lower than for OCT1-reference, indicating that the potency of the inhibi- Drugs co-administered with metformin tors was different for different variants (Table 3). We used a large type 2 diabetes patient population For instance, although comparable inhibition ratios (n ¼ 11 319) to identify drugs administered concomitantly
Table 3 The Cmax,IC50 values and IC50 ratios for the OCT1-reference, R61C and M420del for eight strong and three weak inhibitors
a b Cmax (mM) IC50 (mM)IC50 ratio
Reference R61C M420del Ref/R61C Ref/M420del
Spironolactone 0.48 3.08 (±1.12) 0.36 (±1.59) 2.45 (±1.18) 8.54 1.26 Ketoconazole 12.2 7.40 (±1.21) 0.68 (±1.26) 2.70 (±1.09) 10.8 2.74 Verapamil 0.60 12.5 (±1.10) 2.38 (±1.11) 3.36 (±1.19) 5.27 3.73 Propafenone 2.19 14.1 (±1.18) 2.22 (±1.17) 10.3 (±1.24) 6.36 1.37 Doxazosin 0.04 14.8 (±1.22) 0.63 (±1.25) 4.27 (±1.15) 23.4 3.46 Ondansetron 0.36 36.2 (±1.40) 8.77 (±1.26) 23.8 (±1.26) 4.13 1.52 Desipramine 0.06 43.9 (±1.44) 14.7 (±1.22) 21.0 (±1.32) 2.99 2.09 Prazosin NA 50.9 (±1.37) 4.54 (±1.26) 4.81 (±1.21) 11.2 10.6 Chlorpromazine 0.03 52.3 (±1.23) 15.5 (±1.34) 8.20 (±1.15) 3.38 6.38 Quinidine 16.6 340 (±1.62) 20.5 (±1.28) 25.1 (±1.22) 16.6 13.6 Bucindolol NA 15.1 (±1.24) 1.18 (±1.21) 13.4 (±1.27) 12.8 1.13
Abbreviation: NA, not available. aThe maximal total plasma concentration. Obtained from Goodman and Gillman’s and Clarke’s isolation and identification of drugs.31,46 b The ratio between OCT1-reference IC50 and R61C/M420del. + IC50 values were derived from concentration-dependent inhibition curves of ASP uptake.
The Pharmacogenomics Journal Effect of inhibition on genetic variants of OCT1 G Ahlin et al 407
150 135% Inhibition of metformin uptake The inhibition of metformin uptake by each of the five 125 concomitantly administered drugs was first investigated at 100% clinically relevant-free drug concentrations obtained from the 100 literature.14–17,21 However, the addition of such small 75 amounts of the inhibitors to the cell culture medium resulted in poor reproducibility/mass balance, which, in our experi- 50 ence, is a result of nonspecific binding, for example, to the cell Percent uptake 25% culture plastics (data not shown). In an attempt to circumvent 25 9% such difficulties, we included a physiological concentration of 0 HSA and also, initially, acid a-glycoprotein in the cell culture ReferenceM420del V408M R61C medium, assuming that these additions would reflect the 45 physiological plasma protein binding adequately and improve Reference the mass balance.22 Preliminary experiments showed that 40 M420del addition of HSA with and without acid a-glycoprotein gave V408M 35 comparable results for the transport and inhibition, and 30 improved the reproducibility/mass balance to the same extent (data not shown). Addition of acid a-glycoprotein was, 25 therefore, not considered to be necessary. Thus, in the 20 following, the concentration-dependent inhibition studies were performed in the presence of HSA and over a range that 15 was appropriate for the clinical drug concentrations of the
nmol/mg protein/min 10 inhibitors. Metformin was used at its maximal plasma 13 5 concentration, 9.66 mM. All compounds inhibited metformin uptake in a concen- 0 tration-dependent manner that allowed the calculation of 0 2 4 6 8 10 12 14 16 þ IC50 values (Figure 3c; Table 4). As for ASP inhibition, the concentration (mM) ratio between IC50 values for the OCT1-reference and 50 M420del varied for the different drugs. Variant M420del showed an increased sensitivity to inhibition of metformin 40 transport compared with the OCT1-reference. The IC50 ratios (reference/M420del) for amitriptyline and verapamil 30 Reference were 1.5, and 6.8, respectively (Table 4). In contrast, the M420del sensitivity to inhibition of metformin transport in V408M 20 V408M was comparable to that of the OCT1-reference. The IC50 ratios (reference/V408M) for amitriptyline and verapamil were 1.5 and 1.0, respectively (Table 4).
pmol/mg protein/min 10 Thus, the IC50 value for verapamil in M420del was lower than the reported C for verapamil (Table 4). This suggests 0 max that co-administration of verapamil with metformin may -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 log concentration (uM) result in a reduced uptake of metformin in human hepatocytes. This conclusion can be made because the Figure 3 (a) The 14C-metformin uptake in OCT1-reference, R61C, clinical range of metformin concentrations is far below the M420del and V408M relative to uptake in OCT1-reference. A significant Km for OCT1 transport of this drug. All other drugs inhibited reduction in metformin uptake was seen for both R61C and M420del. metformin at concentrations above the clinical plasma (b) Michaelis–Menten kinetic curves for the uptake of 14C-metformin in OCT1-reference, V408M and M420del. (c) The concentration-depen- concentrations, suggesting that these have a limited poten- dent effect of the OCT1 inhibitor verapamil on the uptake of the tial to affect the metformin uptake. metformin in the OCT1-reference, V408M and M420del. Because portal vein concentrations at the liver directly after absorption may exceed maximal plasma concentrations with metformin. Five drugs administered with metformin in (Cmax) considerably, we calculated the maximal drug concen- more than 1% of the patients were identified (Table 4); the tration at the liver directly after intestinal uptake (Cmax, portal) type 2 diabetes drugs, glibenclamide (taken by 24.2% of the as suggested by Ito et al.23 according to equation 1. 11 319 patients) and pioglitazone (12.6%), the cardio- ka D Fa vascular drug verapamil (2.9%), the lipid-lowering drugs Cmax;portal ¼ Cmax þ ð1Þ simvastatin (16.8%) and amitriptyline (2.6%), which are Qh used to treat neuropathic pain in diabetic patients (Table 4). 15 These drugs were investigated for concentration-dependent Cmax, the dose (D), fraction absorbed and hepatic blood inhibition of metformin uptake by HEK293 cells. flow (Qh) were taken from the literature. The rate constant
The Pharmacogenomics Journal Effect of inhibition on genetic variants of OCT1 G Ahlin et al 408
Table 4 IC50 values and the IC50 ratios for the OCT1-reference, M420del and V408M
Frequency Reported Predicted Reference M420del V408M Ratio Ratio a b c d (%) Cmax (mM) Cmax, portal (mM) IC50 (mM) IC50 (mM) IC50 (mM) Ref/M420del Ref/V408M
Verapamil 4.9 0.60 15.05 0.62 (±1.07) 0.09 (±1.87) 0.63 (±1.25) 6.84 0.99 Amitriptyline 3.2 0.72 33.0 6.99 (±1.39) 4.70 (±1.56) 4.55 (±1.19) 1.49 1.54 Glibenclamide 37.0 0.73 1.41 199 (±1.47) 85.8 (±1.97) — 2.32 — Pioglitazone 21.3 4.49 12.3 185 (±1.46) 178 (±2.62) — 1.04 — Simvastatin 20.7 0.13 9.03 89.0 (±1.25) 26.5 (±1.87) — 3.36 — aThe treatment frequency of the drugs was derived from prescription data for 11 319 US patients treated with metformin, the type 2 diabetes drug. b The reported total Cmax. obtained from Goodman and Gillman’s and Clarke’s isolation and identification of drugs. cPredicted portal vein concentration as described by Ito et al. 23 d IC50 ratios between OCT1-reference and M420del/V408M. IC50 values were derived from concentration-dependent inhibition curves of metformin uptake.