Zinc Finger Nuclease Mediated Gene Knockout Results in...

9
1521-009X/43/2/199207$25.00 http://dx.doi.org/10.1124/dmd.114.057216 DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 43:199207, February 2015 Copyright ª 2014 by The American Society for Pharmacology and Experimental Therapeutics Zinc Finger NucleaseMediated Gene Knockout Results in Loss of Transport Activity for P-Glycoprotein, BCRP, and MRP2 in Caco-2 Cells Kathleen E. Sampson, 1 Amanda Brinker, 2 Jennifer Pratt, Neetu Venkatraman, 3 Yongling Xiao, Jim Blasberg, Toni Steiner, Maureen Bourner, and David C. Thompson Sigma-Aldrich Corporation, St. Louis, Missouri Received February 5, 2014; accepted November 10, 2014 ABSTRACT Membrane transporters P-glycoprotein [P-gp; multidrug resistance 1 (MDR1)], multidrug resistanceassociated protein (MRP) 2, and breast cancer resistance protein (BCRP) affect drug absorption and disposition and can also mediate drug-drug interactions leading to safety/toxicity concerns in the clinic. Challenges arise with interpreting cell-based transporter assays when substrates or inhibitors affect more than one actively expressed transporter and when endogenous or residual transporter activity remains following overexpression or knockdown of a given transporter. The objective of this study was to selectively knock out three drug efflux transporter genes (MDR1, MRP2, and BCRP), both individually as well as in combination, in a subclone of Caco-2 cells (C2BBe1) using zinc finger nuclease technology. The wild-type parent and knockout cell lines were tested for transporter function in Transwell bidirectional assays using probe substrates at 5 or 10 mM for 2 hours at 37°C. P-gp substrates digoxin and erythromycin, BCRP substrates estrone 3-sulfate and nitrofuran- toin, and MRP2 substrate 5-(and-6)-carboxy-29,79-dichlorofluorescein each showed a loss of asymmetric transport in the MDR1, BCRP, and MRP2 knockout cell lines, respectively. Furthermore, transporter interactions were deduced for cimetidine, ranitidine, fexofenadine, and colchicine. Compared with the knockout cell lines, standard transporter inhibitors showed substrate-specific variation in reducing the efflux ratios of the test compounds. These data confirm the generation of a panel of stable Caco-2 cell lines with single or double knockout of human efflux transporter genes and a complete loss of specific transport activity. These cell lines may prove useful in clarifying complex drug-transporter interactions without some of the limitations of current chemical or genetic knockdown approaches. Introduction Membrane drug transporters play an important role in the distribution of endogenous molecules and xenobiotics throughout the body and are implicated in detoxification mechanisms as well as multidrug resistance. Members of the ATP-binding cassette (ABC) efflux transporter family, such as P-glycoprotein [P-gp; multidrug resistance 1 (MDR1); ABCB1], multidrug resistanceassociated protein (MRP) 2 (ABCC2), and breast cancer resistance protein (BCRP; ABCG2), actively efflux a wide variety of small-molecule substrates out of the cell to protect cells and organs against harmful drugs or toxins (Litman et al., 2001; Shitara et al., 2006). ABC transporters have a protective role in blocking intestinal absorption (Takano et al., 2006; Oude Elferink and de Waart, 2007) and enhancing excretion of endogenous and xenobiotic compounds from the hepatic canalicular membrane and the kidney proximal tubules (Köck and Brouwer, 2012; Masereeuw and Russel, 2012). They play a role in clinical drug resistance to multiple chemotherapeutic agents (Szakacs et al., 2006; Veringa et al., 2013) and in drug-drug interactions that may alter systemic exposure and lead to clinical adverse events (Lin, 2007; Marquez and Van Bambeke, 2011; Müller and Fromm, 2011; DeGorter et al., 2012). Guidelines have recently been published by the US Food and Drug Administration (http://www.fda.gov/downloads/Drugs/ GuidanceComplianceRegulatoryInformation/Guidances/UCM292362.pdf) and European Medicines Agency (http://www.ema.europa.eu/docs/en_GB/ document_library/Scientific_guideline/2012/07/WC500129606.pdf) on screening new chemical entities for interactions with clinically relevant transporters. In vitro evaluation of specific transporter interactions can employ a variety of tools, including transporter-expressing cell lines, membrane vesicles, and tissues, along with a panel of substrates and inhibitors as control probes. The standard assay format for ABC transporter function measures the transcellular permeability of a test article through a monolayer of cells grown on a permeable filter, and comparison of the absorptive versus secretory flux. Caco-2 cells are derived from a human intestinal adenocarcinoma and are widely used as a model of intestinal absorption and transporter activity (Elsby et al., 2008). These cells differentiate in culture to an intestinal phenotype with a well defined apical brush border, are able to form tight junctions, and express the ABC efflux transporters P-gp, BCRP, and MRP2, as well as other uptake and efflux transporters normally expressed in human intestinal enterocytes (Hilgendorf et al., 2007). The cells express the transporters in a polarized fashion, enabling the vectorial transport of substrates, and are considered the gold standard for efflux transporter screening. 1 Current affiliation: Covance, Madison, Wisconsin. 2 Current affiliation: University of MissouriKansas City, Kansas City, Missouri. 3 Current affiliation: Confluence Life Sciences, St. Louis, Missouri. dx.doi.org/10.1124/dmd.114.057216. ABBREVIATIONS: A, apical; ABC, ATP-binding cassette; B, basolateral; BCRP, breast cancer resistance protein; CDCF, 5-(and-6)-carboxy-29,79- dichlorofluorescein; CDCFDA, 5-(and-6)-carboxy-29,79-dichlorofluorescein diacetate; KO, knockout; LC-MS/MS, liquid chromatographytandem mass spectrometry; MDR1, multidrug resistance 1; MRP, multidrug resistanceassociated protein; PCR, polymerase chain reaction; P-gp, P-glycoprotein; ZFN, zinc finger nuclease. 199 at ASPET Journals on August 28, 2019 dmd.aspetjournals.org Downloaded from

Transcript of Zinc Finger Nuclease Mediated Gene Knockout Results in...

Page 1: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

1521-009X/43/2/199–207$25.00 http://dx.doi.org/10.1124/dmd.114.057216DRUG METABOLISM AND DISPOSITION Drug Metab Dispos 43:199–207, February 2015Copyright ª 2014 by The American Society for Pharmacology and Experimental Therapeutics

Zinc Finger Nuclease–Mediated Gene Knockout Results inLoss of Transport Activity for P-Glycoprotein, BCRP, and

MRP2 in Caco-2 Cells

Kathleen E. Sampson,1 Amanda Brinker,2 Jennifer Pratt, Neetu Venkatraman,3 Yongling Xiao,Jim Blasberg, Toni Steiner, Maureen Bourner, and David C. Thompson

Sigma-Aldrich Corporation, St. Louis, Missouri

Received February 5, 2014; accepted November 10, 2014

ABSTRACT

Membrane transporters P-glycoprotein [P-gp; multidrug resistance1 (MDR1)], multidrug resistance–associated protein (MRP) 2, andbreast cancer resistance protein (BCRP) affect drug absorption anddisposition and can also mediate drug-drug interactions leading tosafety/toxicity concerns in the clinic. Challenges arise with interpretingcell-based transporter assays when substrates or inhibitors affectmore than one actively expressed transporter and when endogenousor residual transporter activity remains following overexpression orknockdown of a given transporter. The objective of this study was toselectively knock out three drug efflux transporter genes (MDR1,MRP2, and BCRP), both individually as well as in combination, ina subclone of Caco-2 cells (C2BBe1) using zinc finger nucleasetechnology. The wild-type parent and knockout cell lines were testedfor transporter function in Transwell bidirectional assays using probe

substrates at 5 or 10 mM for 2 hours at 37�C. P-gp substrates digoxinand erythromycin, BCRP substrates estrone 3-sulfate and nitrofuran-toin, and MRP2 substrate 5-(and-6)-carboxy-29,79-dichlorofluoresceineach showed a loss of asymmetric transport in the MDR1, BCRP, andMRP2 knockout cell lines, respectively. Furthermore, transporterinteractions were deduced for cimetidine, ranitidine, fexofenadine,and colchicine. Compared with the knockout cell lines, standardtransporter inhibitors showed substrate-specific variation in reducingthe efflux ratios of the test compounds. These data confirm thegeneration of a panel of stable Caco-2 cell lines with single or doubleknockout of human efflux transporter genes and a complete loss ofspecific transport activity. These cell lines may prove useful inclarifying complex drug-transporter interactions without some of thelimitations of current chemical or genetic knockdown approaches.

Introduction

Membrane drug transporters play an important role in the distributionof endogenous molecules and xenobiotics throughout the body and areimplicated in detoxification mechanisms as well as multidrug resistance.Members of the ATP-binding cassette (ABC) efflux transporter family,such as P-glycoprotein [P-gp; multidrug resistance 1 (MDR1); ABCB1],multidrug resistance–associated protein (MRP) 2 (ABCC2), and breastcancer resistance protein (BCRP; ABCG2), actively efflux a wide varietyof small-molecule substrates out of the cell to protect cells and organsagainst harmful drugs or toxins (Litman et al., 2001; Shitara et al., 2006).ABC transporters have a protective role in blocking intestinal absorption(Takano et al., 2006; Oude Elferink and de Waart, 2007) and enhancingexcretion of endogenous and xenobiotic compounds from the hepaticcanalicular membrane and the kidney proximal tubules (Köck andBrouwer, 2012; Masereeuw and Russel, 2012). They play a role inclinical drug resistance to multiple chemotherapeutic agents (Szakacset al., 2006; Veringa et al., 2013) and in drug-drug interactions that mayalter systemic exposure and lead to clinical adverse events (Lin, 2007;

Marquez and Van Bambeke, 2011; Müller and Fromm, 2011; DeGorteret al., 2012).Guidelines have recently been published by the US Food and

Drug Administration (http://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/UCM292362.pdf)and European Medicines Agency (http://www.ema.europa.eu/docs/en_GB/document_library/Scientific_guideline/2012/07/WC500129606.pdf)on screening new chemical entities for interactions with clinically relevanttransporters. In vitro evaluation of specific transporter interactions can employa variety of tools, including transporter-expressing cell lines, membranevesicles, and tissues, along with a panel of substrates and inhibitors as controlprobes. The standard assay format for ABC transporter function measuresthe transcellular permeability of a test article through a monolayer of cellsgrown on a permeable filter, and comparison of the absorptive versus secretoryflux. Caco-2 cells are derived from a human intestinal adenocarcinomaand are widely used as a model of intestinal absorption and transporteractivity (Elsby et al., 2008). These cells differentiate in culture to anintestinal phenotype with a well defined apical brush border, are able toform tight junctions, and express the ABC efflux transporters P-gp,BCRP, and MRP2, as well as other uptake and efflux transportersnormally expressed in human intestinal enterocytes (Hilgendorfet al., 2007). The cells express the transporters in a polarized fashion,enabling the vectorial transport of substrates, and are considered thegold standard for efflux transporter screening.

1Current affiliation: Covance, Madison, Wisconsin.2Current affiliation: University of Missouri–Kansas City, Kansas City, Missouri.3Current affiliation: Confluence Life Sciences, St. Louis, Missouri.dx.doi.org/10.1124/dmd.114.057216.

ABBREVIATIONS: A, apical; ABC, ATP-binding cassette; B, basolateral; BCRP, breast cancer resistance protein; CDCF, 5-(and-6)-carboxy-29,79-dichlorofluorescein; CDCFDA, 5-(and-6)-carboxy-29,79-dichlorofluorescein diacetate; KO, knockout; LC-MS/MS, liquid chromatography–tandemmass spectrometry; MDR1, multidrug resistance 1; MRP, multidrug resistance–associated protein; PCR, polymerase chain reaction; P-gp,P-glycoprotein; ZFN, zinc finger nuclease.

199

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 2: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

Transporters recognize and interact with a broad range of com-pounds based on their physicochemical characteristics (Didziapetriset al., 2003; Zhou et al., 2008), with overlapping substrate recognitionbetween transporters. Transporter-“specific” inhibitors are used to helpdefine interactions, but may also show overlapping interactions betweentransporters (Matsson et al., 2009). In addition, substrates may interactwith different binding sites per transporter, necessitating the use ofmultiple inhibitors with different binding site–specific affinities for eachtransporter (Giri et al., 2009). This lack of specificity can cause mis-interpretation in biologic systems with multiple transporters or en-dogenous transporters in transfected cell lines (Goh et al., 2002;Wang et al., 2008; Mease et al., 2012). Thus, there exists a needfor human testing systems that allow unambiguous identificationof specific substrate interaction without dependence on chemicalinhibition.Targeted suppression of gene expression by RNA interference

techniques has been explored in several laboratories using Caco-2cells (Celius et al., 2004; Watanabe et al., 2005; Zhang et al., 2009;Darnell et al., 2010; Graber-Maier et al., 2010). Transfection of shorthairpin RNA vectors and the resultant downregulation of transportersoffers an advantage over reliance on inhibitors to elucidate specificdrug-transporter interactions. However, not all RNA oligos are able toknock down the targeted mRNA efficiently, and they may invoke off-target effects on similar mRNAs. Most importantly, substantialresidual activity may remain in a cell line in spite of reduced mRNA andprotein levels (Darnell et al., 2010; Wang et al., 2014).Zinc finger nuclease (ZFN) technology involves transfection of

highly specific gene-targeting reagents linked to DNA cleavageenzymes, allowing exquisite specificity and total gene knockout (KO)in a stable cell line while minimizing off-target effects (Santiago et al.,2008). We report here the generation and characterization of a panel ofKO cell lines targeting MDR1, BCRP, and MRP2 transporters in theCaco-2 subclone C2BBe1 cell line using ZFNs. The resultant panel ofsingle or double KO cells shows disruption of gene sequence as wellas complete loss of transporter function in bidirectional transportassays. These transporter KO cell lines provide a powerful new toolfor elucidating transporter interactions.

Materials and Methods

Unless otherwise indicated, all cell culture media, biochemical reagents, andchemicals were obtained from Sigma-Aldrich (St. Louis, MO). CostarTranswell HTS 24-well plates were purchased from Sigma-Aldrich. 5-(and-6)-Carboxy-29,79-dichlorofluorescein (CDCF) and 5-(and-6)-carboxy-29,79-dichlorofluorescein diacetate (CDCFDA) were purchased from Life Technologies(Carlsbad, CA). Primers specific for MDR1, BCRP, MRP2–4 transporter genes,and the endogenous control human glyceraldehyde-3-phosphate dehydroge-nase gene were purchased as Taqman Gene Expression Assays from LifeTechnologies. For Western blotting experiments, a rabbit monoclonal antibodyto MDR1 and a mouse monoclonal antibody to b-actin were purchased fromAbcam Inc. (Cambridge, MA), while rabbit polyclonal antibody to BCRP andrabbit monoclonal antibody to MRP2 were obtained from Cell SignalingTechnology, Inc. (Danvers, MA). Secondary antibodies for b-actin (donkeyanti-mouse) and transporters (donkey anti-rabbit) were obtained from JacksonImmunoResearch Laboratories, Inc. (West Grove, PA).

Cell Culture. The C2BBe1 (Caco-2 brush border–expressing) cell line,a subclone of Caco-2 cells, was obtained from American Type CultureCollection (Manassas, VA). The original tissue donor was a 72-year-old malewith colorectal adenocarcinoma. The C2BBe1 cell line had been cloned fromthe Caco-2 cell line (ATCC HTB-37) by limiting dilution and was selectedon the basis of morphologic homogeneity and exclusive apical villin local-ization (Peterson and Mooseker, 1992). Cells were maintained in high-glucoseDulbecco’s modified Eagle’s medium with 10% heat-inactivated fetal bovineserum, 1% (v/v) minimum Eagle’s medium nonessential amino acids, 2 mM

L-glutamine, 1 mM sodium pyruvate, 100 units/ml penicillin, and 100 mg/mlstreptomycin. Cells were cultured in humidified incubators at 37�C in 5%CO2. Culture medium was refreshed at 2- to 3-day intervals. Cells werepassaged upon reaching confluence, at least once per week, using 0.25%trypsin-EDTA.

ZFN-Mediated DNA Modification and Subclone Selection. KO cell lineswere generated using CompoZr Custom Zinc Finger Nuclease (Sigma-Aldrich)kit components as previously described (Pratt et al., 2012). Briefly, 2 mg ofeach ZFN forward and reverse mRNA primer for each transporter gene, alongwith 4 mg of gene-specific mammalian single-strand annealing reporterplasmids containing two complementary portions of the green fluorescentprotein, were nucleofected into C2BBe1 cells using the Amaxa Cell LineNucleofector Kit T for Caco-2 cells (Lonza, Basel, Switzerland) as per themanufacturer’s directions. Nucleofected cells were immediately placed in 20%fetal bovine serum growth medium and cultured in six-well plates at 30�C for2 days to increase efficiency of nucleofection, then were moved to 37�C.Medium was refreshed once cells had attached. After growing to .70%confluence (1–2 weeks), cells were trypsinized and stained with 1 mg/ml ofpropidium iodide before flow cytometry sorting based on green fluorescentprotein–positive (indicating successful ZFN cutting) and propidium iodide–negative (indicating viable cells) sort gates. Cells were single cell sorted into96-well plates using the FACSAria III (BD Biosciences, San Jose, CA) andwere cultured for several weeks to form substantial colonies before testing formutations. Genomic DNA was obtained using QuickExtract DNA ExtractionSolution (Epicentre Biotechnologies, Madison, WI) and scaled up usingpolymerase chain reaction (PCR) amplification with the ZFN Cel-1 primersspecific for each target gene. PCR product was run on the 96-capillary 3730xlDNA Analyzer using Peak Scanner software v1.0 (Life Technologies). Clonesshowing non–wild-type and out-of-frame mutations were selected for subcloning.DNA was amplified and subcloned into competent Escherichia coli cells usingthe TOPO TA Cloning Kit (Life Technologies) according to the manufacturer’sdirections. DNA was isolated from colonies using GenElute MammalianGenomic DNA Miniprep kit (Sigma-Aldrich) and sequenced to confirm genedisruption through base-pair deletion and/or insertion and identification ofspecific homozygous KO clones.

mRNA Expression Analysis. Cells at subconfluent densities were trypsinizedfrom T75 flasks and centrifuged at 800 rpm for 5 minutes. Medium wasaspirated, and cell pellets were stored at 280�C until use. RNA from each ofthe cell pellets was isolated using the RNeasy Protect Mini Kit (Qiagen,Valencia, CA). To remove genomic DNA, on-column DNase digestion wasperformed using the On-Column DNase 1 digestion set (Sigma-Aldrich)according to instructions.

Reverse-transcription PCR reactions were set up using the Taqman RNA-to-Ct 1Step kit (Life Technologies) including individual transporter gene primersor endogenous control glyceraldehyde-3-phosphate dehydrogenase primers(Taqman Gene Expression Assays; Life Technologies) and 100 ng of RNA.PCR cycling conditions consisted of 48�C for 30 minutes, 95�C for 10 minutes,followed by 40 cycles of a denaturing step at 95�C for 15 seconds and anannealing/extension step with fluorescence monitoring at 60�C for 1 minute.The relative expression changes were calculated as described by Livak andSchmittgen (2001).

Immunoblot Protein Expression Analysis. Confluent T75 flasks of Caco-2cells were lysed with 2 ml of 1� LDS sample buffer (Life Technologies)containing protease inhibitor cocktail. Cells were scraped from the flask, ho-mogenized using Qiashredder columns (Qiagen), and stored at 280�C. Thawedlysates were denatured by heating for 10 minutes at 65�C, then loaded (20 ml perlane) onto NuPage Novex 4–12% Bis-Tris gels (Bio-Rad Laboratories, Hercules,CA) and run at 200 V in 4-morpholinepropanesulfonic acid SDS buffer. Gelswere transferred to polyvinylidene fluoride membranes for 15 minutes using theTrans-Blot Turbo system (Bio-Rad). Membranes were blocked with Blottocontaining 0.05% Tween (Blotto+T) for 2–3 hours while shaking at roomtemperature. Membranes were placed in Blotto+T containing primary antibodies(diluted 1:250, or 1:1000 for b-actin) and incubated at 4�C overnight whileshaking. Following multiple 10-minute washes in Tris-buffered saline with 0.05%Tween, membranes were placed in Blotto+T containing anti-rabbit or anti-mousesecondary antibodies (diluted 1:10,000) and incubated for 1 hour while shaking.Following multiple 10-minute washes in Tris-buffered saline with 0.05%Tween, proteins were visualized using Super Signal West Dura detection

200 Sampson et al.

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 3: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

reagent (Thermo Scientific, Rockford, IL) and imaged on a ChemiDoc imagerusing Image Laboratory software v4.0 (Bio-Rad).

Bidirectional Transport Assay. C2BBe1 wild-type and KO cell lines wereplated at 4 � 104 cells/well onto Costar HTS-Transwell 24-well permeablesupport plates (0.4-mM pore size, 0.33-cm2 polyethylene terephthalate filter).Cells were cultured for 20–22 days to obtain differentiated monolayers withtight junctions and polarized transporter expression. On day of study, cellmonolayers were rinsed twice and then preincubated for 30–60 minutes withtransport buffer (Hanks’ balanced salt solution with 25 mM D-glucose and10 mMHEPES; pH 7.4) in both chambers at 37�C. Test articles were diluted from10 mM dimethylsulfoxide stocks to 5 or 10 mM in transport buffer and placedin either apical (A) or basolateral (B) donor chambers in triplicate, while freshtransport buffer alone was placed in receiver chambers. Additional transportstudies were conducted in the presence of inhibitors MK571 [5-(3-(2-(7-chloroquinolin-2-yl)ethenyl)phenyl)-8-dimethylcarbamyl-4,6-dithiaoctanoicacid], Ko143 [(3S,6S,12aS)-1,2,3,4,6,7,12,12a-octahydro-9-methoxy-6-(2-methylpropyl)-1,4-dioxopyrazino[19,29:1,6]pyrido[3,4-b]indole-3-propanoic acid1,1-dimethylethyl ester], or verapamil, which were added to both chambers duringtransport assay at indicated concentrations. Plates were incubated at 37�C for2 hours, at which time aliquots were removed from donor and receiver chambers forquantitation. Analyte concentrations were determined by liquid chromatography–tandem mass spectrometry (LC-MS/MS).

For the fluorescent MRP2 substrate CDCF, plates were treated with 10 mMCDCFDA and incubated at 37�C for 2 hours as above. Receiver and donorchamber samples were transferred to black-walled 96-well plates. A triplicatestandard curve of 2-fold serial dilutions from 10 mMCDCF was generated froma 10 mM dimethylsulfoxide stock. Plates were quantified by fluorescence at485-nm emission, 538-nm excitation on a SpectraMax Gemini XS plate readerusing SOFTmax Pro software v 3.1.2 (Molecular Devices, Sunnyvale, CA), bothimmediately postassay and after 24 hours at room temperature for maximalhydrolysis of dosing and donor solutions to fluorescent product.

At the conclusion of the transport assay, residual buffer was aspirated fromall wells. Fresh transport buffer was added to basolateral chambers, and Luciferyellow (dilithium salt) at 0.1 mg/ml in Hanks’ balanced salt solution was addedto apical chambers. The plates were incubated at 37�C for 1 hour. Samples weretransferred from the basolateral chambers to black-walled 96-well plates andquantified by fluorescence at 485-nm emission, 538-nm excitation as describedabove. Lucifer yellow permeability A to B was calculated; those wells ex-hibiting permeability .2 � 1026 cm/s were eliminated from assay results.

LC-MS/MS Analysis. Concentration of test articles in samples wasanalyzed by LC-MS/MS using an API-4000 Q Trap mass spectrometer witha Turbo V atmospheric pressure electrospray ionization source (AB SCIEX,Framingham, MA). Samples (40 ml) were injected onto a Fortis C8 column(2.1 � 50 mm, 5 mm) and eluted by a mobile phase gradient optimized for eachtest article [mobile phase A: 4 mM ammonium formate; mobile phase B: 4 mMammonium formate in 90% (v/v) acetonitrile]. Flow rate was 0.5 ml/min. Usingpositive or negative ionization mode, analytes were quantitated using multiplereaction monitoring specific for each analyte and internal standard (tolbuta-mide) parent-product ion pairs. Peak areas of analyte and internal standard andresulting ratios were quantified using Analyst 1.5 (AB SCIEX).

Calculations. The apparent permeability (Papp, in centimeters per second)was determined for both A to B and B to A directions by the followingcalculation:

Papp ¼ 1A*CDð0Þ*

dMr

dt

in which A is the area of filter membrane, CD(0) is the initial concentration ofthe test drug, dMr is the amount of transported drug, and dt is time elapsed. Theefflux ratio (ER) was calculated from

ER ¼ ðPapp;  B  to AÞ=ðPapp; A  to  BÞ

An efflux ratio of $2 suggests an active transport process, identifying thecompound as an apical efflux transporter substrate.

Statistics. Unless otherwise noted, all transport assays were carried out intriplicate and repeated on at least three separate days. The data are presented asmean 6 S.D. Statistical significance was determined using one-way analysis ofvariance calculations.

Results

ZFN-Mediated Disruption of Genomic DNA Sequence. Follow-ing nucleofection with ZFN pairs and single cell sorting, C2BBe1clones exhibiting mutations in all four alleles were initially identifiedby fragment analysis. These clones were further expanded for genomicDNA sequencing within the ZFN target area. Small insertions and/ordeletions (indels) were confirmed within each allele in the tetraploidcells for each single and double KO clone. Genotype analysis of thesingle KO clones is shown in Table 1. Each MDR1 and BCRP KOclone contained out-of-frame indels resulting in the generation ofa premature stop codon. For the MDR1 KO clone, each allelecontained a unique modification, while for BCRP KO, only twomodified sequences were observed among the four alleles. The MRP2KO clone contained two separate in-frame deletions (29, 236); how-ever, these deletions overlapped a splice site and thus were still effectiveat disruption of translation into a functional MRP2 protein. The MDR1KO, BCRP KO, and MDR1/BCRP KO clones were all generatedindependently; the MRP2 KO was used to generate the MDR1/MRP2 KO and the MRP2/BCRP KO clones in a second round ofZFN nucleofections.Protein Expression Analysis. Western blots were run to confirm

the absence of the targeted transporter protein in each of the single KOclones (Fig. 1). For each transporter, the protein was detected in thewild-type (parental) cells but was completely absent in the appropriateKO cell line. Equivalent loading of protein samples per lane wasconfirmed by measuring b-actin staining.

TABLE 1

Genotype analysis of single KOs

Sequences show the ZFN binding sites for each of the three target genes; lower-case letters designate the ZFN cut site. Underlined basesrepresent deletions in each allele, while bold highlighted bases represent insertions.

ZFN Binding Site Indels

MDR1 geneAllele 1 GTCCTGTTCTTGGACtgtcaGCTGCTGTCTGGGCAAAG 22Allele 2 GTCCTGTTCTTGGACtgtcaGCTGCTGTCTGGGCAAAG 24Allele 3 GTCCTGTTCTTGGACtgtcaGCTGCTGTCTGGGCAAAG 25Allele 4 GTCCTGTTCTTGGACtgtcAaGCTGCTGTCTGGGCAAAG 29, +1

BCRP geneAlleles 1 and 2 TACACCACCTCCTTCTGTcatcaACTCAGATGGGT 24Alleles 3 and 4 TACACCACCTCCTTCTGTcGTCATatcaACTCAGATGGGT +5

MRP2 geneAlleles 1 and 2 GTCTCCCTAGTCCATGATggcagtGAAGAAGAAGACGATGAC 29Alleles 3 and 4 GTCTCCCTAGTCCATGATggcagtGAAGAAGAAGACGATGAC 236

P-gp, BCRP, and MRP2 Knockout in Caco-2 Cells 201

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 4: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

mRNA Expression Analysis. To determine whether substantivechanges in expression of related ABC transporters occurs when singleor double transporter genes are knocked out, mRNA expression levelsof MDR1, BCRP, MRP2, MRP3, and MRP4 were measured in eachof the single and double KO cell lines. In Fig. 2A, relative mRNAexpression levels of these five transporters are shown for each of thesingle KO cell lines compared with wild type (normalized to 1). Asexpected, the mRNA level of each target gene was reduced in itsrespective KO line, most likely due to decreased stability of themutated transcript. Only small changes in expression levels weredetected for any of the other transporters. Maximum changes detectedwere a 2-fold increase in MRP3 mRNA expression in BCRP KO cellsand a 2-fold decrease in MRP3 mRNA expression in MDR1 KO cells.Similar modest compensatory changes were observed in the doubleKO cells (Fig. 2B). Here, a 2.5-fold increase in expression of MRP3mRNA was noted in the MDR1/MRP2 KO cells.Cell Line Characterization. Each of the KO cell lines exhibited

morphology and growth characteristics that were similar to theparental C2BBe1 cells, with the exception of a slight lag in growthrate for the MRP2 KO cell lines. The cellular phenotype for each KOcell line (loss of activity toward a model substrate) was stable out to atleast 40 passages post generation of the master cell bank (data notshown); KO cells were not tested past 40 passages. All cell lines tooka typical 21-day growth period to fully differentiate and form tightjunctions on Transwell plates. Passive permeability data for two markercompounds, atenolol (low permeability, ,1 � 1026 cm/s) and met-oprolol (high permeability, .15 � 1026 cm/s), were used to comparethe passive permeability of wild-type cells with each KO cell line, andalso to serve as a quality control when running test compounds inthese assays. Passive permeabilities of both atenolol and metoprololwere similar in wild-type and all KO cell lines (data not shown). As anadditional control used in all assays, Lucifer yellow A to B permeability

was checked as a paracellular permeability marker postassay to ensurethat tight junctions remained intact.Bidirectional Transport Activity Using Probe Substrates. The

transport of probe substrates for each targeted transporter wasexamined in the full panel of KO cell lines generated, and resultswere compared with those achieved in the wild-type C2BBe1cells.The A to B and B to A permeability values and the resultant efflux ratiosare shown in Tables 2–4. Efflux ratios for digoxin and erythromycin werereduced to near unity in the MDR1 single and double KO cell lines(Table 2). This was a reflection of both an increase in permeability in theA to B (absorptive) direction and a decrease in the B to A (secretory)direction. The use of transporter-specific inhibitors in the parentalC2BBe1 cell line was compared with results obtained using KO cells.The P-gp inhibitor verapamil (100 mM) successfully inhibited digoxinand erythromycin efflux in wild-type cells to the same extent seen in theKOs. Surprisingly, digoxin permeability rates in the A to B direction weredecreased somewhat in the BCRP, MRP2, and MRP2/BCRP KO celllines (from 0.99 to;0.25 � 1026 cm/s), while the B to A rates remainednearly unchanged (12.5 to 18.2 � 1026 cm/s). This resulted in higherefflux ratios for digoxin in these cell lines compared with wild type.Nitrofurantoin and estrone 3-sulfate were used as model substrates

to test for loss of function in the BCRP KO cell lines. The efflux ratiosfor both compounds were reduced to near unity in the BCRP singleand double KO cell lines (Table 3). As a comparator, the BCRPinhibitor Ko143 (1 mM) reduced the efflux ratios for estrone sulfate

Fig. 1. Western blots of P-gp, BCRP, and MRP2 protein expression in wild-type (WT)and KO cell lines. Whole-cell lysates from Caco-2 or KO cells were analyzed fortransporter expression using 1:250 dilutions of primary antibodies followed bychemiluminescent detection. b-Actin expression was used to confirm consistentprotein quantities per lane.

Fig. 2. Relative mRNA expression of efflux transporters in wild-type (WT) and KOcell lines. mRNA was isolated from each cell line and subjected to reverse-transcription PCR for 40 cycles using transporter-specific primers. (A) RelativemRNA levels in single KO cell lines. (B) Relative mRNA levels in double KO celllines. Expression was calibrated to WT levels = 1. Data represent mean 6 S.D. from$2 RNA isolations per cell line.

202 Sampson et al.

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 5: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

and nitrofurantoin to between 2 and 3 in the wild-type cells, suggestingthat Ko143 is less effective in blocking BCRP function than the ZFN-mediated gene KO.For the MRP2 KO cell line, the nonfluorescent compound CDCFDA

was used as the probe substrate. CDCFDA passively diffuses into cells,where it is hydrolyzed by intracellular esterases to the fluorescentproduct CDCF, which is then rapidly excreted by MRP2 (Siissalo et al.,2009). The CDCF efflux ratio was reduced to ;2 in the MRP2 singleand double KO cell lines (Table 4). In contrast, the MRP2 inhibitorMK571 (25 mM) only partially inhibited CDCF transport, with an effluxratio of .8 in wild-type cells.Additional Test Compounds. To further probe the utility of these cell

lines, several additional compounds (ranitidine, cimetidine, fexofenadine,and colchicine) were tested for transporter interactions. Ranitidine wasidentified as a substrate of P-gp only, based on loss of polarized transportin all MDR1 KO lines (Fig. 3). In contrast, the efflux ratio for cimetidinewas only partially reduced in both the MDR1 and BCRP single KO lines(Fig. 4). However, in the MDR1/BCRP double KO cells, the efflux ratiowas fully reduced to unity, thus identifying cimetidine as a substratefor both P-gp and BCRP. The P-gp and BCRP inhibitors verapamiland Ko143 gave a similar pattern of results. Verapamil fully inhibitedranitidine efflux, whereas Ko143 had no effect (Fig. 3), confirming therole of P-gp as sole transporter. For cimetidine, a combination of eitherKO + inhibitor or both inhibitors, resulting in a cumulative loss ofboth P-gp and BCRP activity, was required to reduce the efflux tonear unity (Fig. 4).

Fexofenadine showed a slightly more complex picture of transporterinteractions. In wild-type cells, the efflux ratio for fexofenadine was5.11 (Fig. 5). This was largely inhibited by verapamil (efflux ratio =2.08) but not at all by MK571. In the KO panel, all cell lines lackingP-gp (i.e., MDR1, MDR1/BCRP, and MDR1/MRP2 KOs) showeda complete reduction in efflux, demonstrating that fexofenadine isa substrate for P-gp. In contrast, the BCRP, MRP2, and MRP2/BCRPKO cell lines did not show any inhibition of fexofenadine effluxexcept in the presence of verapamil. Similar to the parental cell line,addition of verapamil only partially reduced the efflux ratio in the KOcell lines. Notably, with each of the three MDR1 KO cell lines, theefflux ratio was reduced to ,1 (0.51–0.72), suggesting a net shifttoward basolateral efflux, or absorptive transport, in the absence ofP-gp. The addition of MK571 slightly increased the efflux ratio offexofenadine in these KO cell lines (0.93–1.52), suggesting inhibitionof the basolateral transport.The colchicine efflux ratio in wild-type cells was 12.8; this efflux

was completely inhibited in the KO cell lines lacking P-gp, but not inthe other KO cell lines, clearly demonstrating that colchicine is asubstrate for P-gp (Fig. 6). Similar to results seen with fexofenadine,the efflux ratio of colchicine in the MDR1 KO cells was slightly ,1(efflux ratios between 0.88 and 0.96) and was also slightly increasedin the presence of MK571 (1.0–1.4), again suggesting that a basolateralMRP is involved. In the cell lines expressing P-gp, both verapamil andMK571 partially affected colchicine transport when used alone and wereable to fully inhibit colchicine efflux only when used in combination

TABLE 2

Known P-gp substrates demonstrating functional consequence of MDR1 gene KO

The permeability rates and efflux ratios of digoxin (5 mM) and erythromycin (5 mM) were assessed in the complete panel of wild-type (C2BBe1) and KO cell lines.Verapamil (100 mM) was used as a P-gp inhibitor in the wild-type cells. Values represent mean 6 S.D.; n = 3 replicates in $3 assays in each experimental group.

Cell Line

Digoxin Erythromycin

PermeabilityEfflux Ratio

PermeabilityEfflux Ratio

A to B B to A A to B B to A

� 1026 cm/s � 1026 cm/s

C2BBe1 0.998 6 0.514 17.7 6 8.3 17.7 6 0.9 0.413 6 0.160 6.95 6 1.63 16.8 6 1.4C2BBe1 + verapamil 3.85 6 1.89 3.72 6 0.83 0.965 6 0.148 0.638 6 0.168 1.11 6 0.17 1.73 6 0.09MDR1 KO 3.29 6 1.00 4.59 6 1.46 1.40 6 0.37 1.82 6 0.81 1.89 6 0.56 1.04 6 0.41BCRP KO 0.251 6 0.087 12.5 6 2.6 49.6 6 4.5 0.322 6 0.112 8.25 6 2.54 25.6 6 15.2MRP2 KO 0.254 6 0.054 16.9 6 3.7 66.4 6 6.5 0.527 6 0.649 8.92 6 1.58 16.9 6 4.2MDR1/BCRP KO 3.41 6 1.33 4.66 6 1.29 1.37 6 0.16 0.758 6 0.135 0.951 6 0.120 1.25 6 0.05MDR1/MRP2 KO 3.57 6 1.32 5.16 6 1.86 1.44 6 0.20 0.870 6 0.344 0.776 6 .0255 0.892 6 0.032MRP2/BCRP KO 0.206 6 0.093 18.2 6 9.0 88.7 6 9.3 0.170 6 0.044 8.55 6 0.90 50.2 6 3.4

TABLE 3

Known BCRP substrates demonstrating functional consequence of BCRP gene KO

The permeability rates and efflux ratios of estrone 3-sulfate (5 mM) and nitrofurantoin (5 mM) were assessed in the complete panel of wild-type (C2BBe1) and KOcell lines. Ko143 (1 mM) was used as a BCRP inhibitor in the wild-type cells. Values represent mean6 S.D.; n = 3 replicates in$3 assays in each experimental group.

Cell Line

Estrone 3-Sulfate Nitrofurantoin

PermeabilityEfflux Ratio

PermeabilityEfflux Ratio

A to B B to A A to B B to A

� 1026 cm/s � 1026 cm/s

C2BBe1 0.550 6 0.227 12.5 6 2.3 22.7 6 10.9 1.17 6 0.25 15.4 6 1.1 13.2 6 1.6C2BBe1 + Ko143 1.62 6 0.47 3.96 6 1.22 2.44 6 0.20 3.53 6 0.80 10.4 6 1.29 2.93 6 0.23MDR1 KO 0.738 6 0.415 8.56 6 0.74 11.6 6 2.4 2.34 6 0.21 18.9 6 4.8 8.09 6 0.53BCRP KO 1.35 6 0.19 2.38 6 0.81 1.76 6 0.28 4.14 6 1.04 6.98 6 2.37 1.68 6 0.12MRP2 KO 0.280 6 0.202 21.4 6 10.6 76.2 6 10.4 0.874 6 0.018 18.1 6 2.6 20.7 6 1.3MDR1/BCRP KO 1.85 6 0.23 1.70 6 0.37 0.920 6 0.48 5.49 6 1.06 11.8 6 1.2 2.15 6 0.13MDR1/MRP2 KO 0.345 6 0.048 11.6 6 7.9 33.5 6 9.9 1.98 6 0.64 18.0 6 2.0 9.06 6 0.98MRP2/BCRP KO 1.42 6 0.19 2.46 6 0.37 1.73 6 0.03 8.06 6 0.92 9.23 6 1.75 1.15 6 0.07

P-gp, BCRP, and MRP2 Knockout in Caco-2 Cells 203

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 6: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

(efflux ratios between 1.4 and 2.1). However, the MRP or other target ofMK571 is currently unknown.Effect of MK571 on Caco-2 Cell Permeability. To further

investigate the role of MK571 in colchicine transport, we testedcolchicine in the wild-type and single KO cell lines with MK571 at 0,10, 25, and 100 mM. At 100 mM MK571, the colchicine efflux ratioin wild-type cells was reduced from .15 to ;1.2, suggesting totalinhibition of active transport. However, the permeability of colchicinehad increased significantly in both directions (data not shown), and theLucifer yellow data from the postassay integrity control indicated a12- to 16-fold-increased permeability in wells that had been exposed to100 mMMK571. Additional assays were run using 10–100 mMMK571under the same experimental conditions but in the absence of any othercompound. We observed that Lucifer yellow A to B permeabilityincreased significantly in C2BBe1 cells with concentrations of MK571of$50 mM, to.10-fold higher at 100 mM (Fig. 7); similar results wereseen in all KO cell lines.

Discussion

Clinically relevant drug-drug interactions have been associated withtransporter inhibition, including efflux transporters (Giacomini et al.,2010). In addition, interaction with efflux transporters has been linkedto poor bioavailability and/or altered rates of clearance (Misaka et al.,2013). Although several cell-based and membrane model systemsexist for studying transporter interactions, these are typically de-pendent on the use of transporter-specific substrates and inhibitors, orthey target a single transporter overexpressed in a nonhuman cellsystem that contains endogenous transporters and lacks the fullcomplement of human transporters. To address some of the limitationsof these current systems, we generated single and double KO cell linesfor the ABC family efflux transporters P-gp, BCRP, and MRP2 inhuman intestinal C2BBe1 cells using ZFN gene editing technology.The KOs were confirmed by genetic analysis, Western blotting, andfunctional assays using model substrates. The KO cell lines appearedsimilar to the wild type in terms of growth rates and morphology,differentiation, formation of tight junctions, passive permeability ofmodel compounds, and stability of phenotype.A key concern in all KO models, whether in vivo or in vitro, is the

potential for adaptation or compensation for the loss of the target geneby changes in the expression of related genes. To address this concern,we compared mRNA expression levels of the three efflux transportersas well as MRP3 and MRP4 in parental and all KO cell lines. Our datasuggest little if any impact on the expression level of these trans-porters; however, only a few genes were examined in the present invitro study, and the possibility of compensation at the protein ex-pression level cannot be definitively ruled out. Comparative analyseshave been carried out in rat models in which P-gp, BCRP, or MRP2have been knocked out using ZFN technology (Chu et al., 2012;Huang et al., 2012; Zamek-Gliszczynski et al., 2013). Zamek-Gliszczynski et al. (2013) reported that expression analyses of a setof 112 genes relevant to absorption, distribution, metabolism, andelimination in liver, kidney, intestine, and brain tissues of the three KOrat lines demonstrated only modest compensatory changes and didnot preclude their general application to study transporter-mediatedpharmacokinetics.

TABLE 4

Known MRP2 substrate demonstrating functional consequence of MRP2 gene KO

The permeability rates and efflux ratios of CDCF (10 mM, added as CDCFDA) were assessedin the complete panel of wild-type (C2BBe1) and KO cell lines. MK571 (25 mM) was used as anMRP2 inhibitor in the wild-type cells. Values represent mean 6 S.D.; n = 3 replicates in$3 assays in each experimental group.

Cell Line

CDCF

PermeabilityEfflux Ratio

A to B B to A

� 1026 cm/s

C2BBe1 0.357 6 0.117 11.5 6 2.77 32.3 6 4.47C2BBe1 + MK571 0.428 6 0.136 3.59 6 1.50 8.38 6 1.26MDR1 KO 0.573 6 0.121 13.9 6 2.04 24.3 6 2.12BCRP KO 0.545 6 0.051 14.0 6 1.80 25.7 6 0.48MRP2 KO 2.04 6 0.63 4.12 6 1.52 2.03 6 0.37MDR1/BCRP KO 0.778 6 0.486 11.3 6 1.34 14.5 6 1.51MDR1/MRP2 KO 2.76 6 0.36 6.46 6 0.49 2.34 6 0.07MRP2/BCRP KO 2.46 6 0.83 4.01 6 1.75 1.63 6 0.41

Fig. 3. Efflux ratios for ranitidine in wild-type and KO cell lines. Experimentswere carried out with ranitidine (5 mM) in Transwell plates for 2 hours at 37�C inthe presence or absence of the P-gp inhibitor verapamil (100 mM) or the BCRPinhibitor Ko143 (1 mM). Values represent mean 6 S.D.; n = 3 replicates in $3assays.

Fig. 4. Efflux ratios for cimetidine in wild-type and KO cell lines. Experiments werecarried out with cimetidine (5 mM) in Transwell plates for 2 hours at 37�C in thepresence or absence of the P-gp inhibitor verapamil (100 mM) and/or the BCRPinhibitor Ko143 (1 mM). Values represent mean6 S.D.; n = 3 replicates in$3 assays.

204 Sampson et al.

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 7: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

Bidirectional transport studies were carried out in all KO cell linesusing well characterized substrates specific for each individual trans-porter and comparing results with the wild-type cells. Compounds usedincluded digoxin and erythromycin for P-gp, estrone sulfate andnitrofurantoin for BCRP, and CDCF for MRP2. Each KO cell lineshowed an appropriate reduction of efflux ratio with the representativesubstrate. Inhibitory effects were consistent between the single anddouble KO cell lines in which the same transporter was absent.Several additional compounds, including some with known cross-

over between multiple transporters, were tested in the KO cell lines.Cimetidine has been reported to be a substrate for both P-gp andBCRP (Pavek et al., 2005; Taur and Rodriguez-Proteau, 2008), whileranitidine is transported primarily by P-gp (Collett et al., 1999;Bourdet and Thakker, 2006). Our results confirmed that cimetidinewas a substrate of both P-gp and BCRP, as the efflux ratio wasreduced to unity only in the MDR1/BCRP KO cell line, while

ranitidine was identified as a substrate for P-gp alone. For bothcimetidine and ranitidine, the inhibitors verapamil and Ko143 wereequally effective at inhibiting the efflux ratio as the KO cells.The H1 antagonist fexofenadine has been described as a substrate

for P-gp (Cvetkovic et al., 1999; Drescher et al., 2002), although thepossible involvement of multiple efflux transporters in its hepaticdisposition, including MRP2, has been suggested (Matsushima et al.,2008; Tian et al., 2008). In the present studies, fexofenadine wasobserved as a substrate for P-gp, as reduced efflux was clearly ob-served in the P-gp KO cell line whereas no reduction of efflux wasobserved in either the BCRP or MRP2 single KO cell lines. The KO celllines provided a clearer assessment of interactions with transporters thandid the use of chemical inhibitors, because verapamil was only able toreduce the efflux ratio for fexofenadine to 2.08, 4.20, and 2.74 in thewild-type, BCRP KO, and MRP2 KO cell lines, respectively.Interestingly, the efflux ratio for fexofenadine in the MDR1 KO

cells was significantly ,1 (0.51), suggesting absorptive transport.This was also observed in the two double KO cell lines that lackedfunctional P-gp. The potential involvement of a basolateral effluxtransporter in the MRP family is supported by the observation that theaddition of MK571 increased the efflux ratio in each of the cell lineslacking P-gp (up to 1.52 in the case of MDR1 KO cells). These datasupport the conclusions drawn by Ming et al. (2011) that fexofenadineapical efflux in Caco-2 cells is predominantly mediated by P-gp,whereas basolateral efflux is predominantly mediated by MRP3. Basedon data using MK571 and a P-gp/BCRP inhibitor (GW120918), Minget al. (2011) further suggested that MRP2 makes a small contribution tothe apical efflux of fexofenadine, although our data using the MRP2 KOcell lines do not support the involvement of MRP2.Similar results were found for the microtubule polymerization

inhibitor colchicine. We observed that colchicine was a substrate forP-gp, based on reduction in efflux ratio in the P-gp KO cell lines andlack of effect in the other KO cell lines. Similarly to fexofenadine, theefflux ratios in cell lines lacking P-gp were below unity, but wereslightly increased by addition of MK571, suggesting that a basolateralMRP may interact with colchicine in the absence of P-gp. Colchicinehas been reported as a substrate for both P-gp and MRP2 in Caco-2cells and rodent intestine (Dahan et al., 2009); however, our data donot support colchicine interaction with MRP2. Reasons for thisdiscrepancy in results may include the lack of MK571 specificitywithin the MRP family as well as a negative impact on the Caco-2 cellmonolayer at higher concentrations, and point to the challenges inusing chemical inhibitors versus gene KO technology.

Fig. 5. Efflux ratios for fexofenadine in wild-type and KO cell lines. Experimentswere carried out with fexofenadine (5 mM) in Transwell plates for 2 hours at 37�C inthe presence or absence of the P-gp inhibitor verapamil (100 mM) and/or the MRPinhibitor MK571 (25 mM). Values represent mean 6 S.D.; n = 3 replicates in$3 assays.

Fig. 6. Efflux ratios for colchicine in wild-type and KO cell lines. Experiments werecarried out with colchicine (5 mM) in Transwell plates for 2 hours at 37�C in thepresence or absence of the P-gp inhibitor verapamil (100 mM) and/or the MRP inhibitorMK571 (25 mM). Values represent mean 6 S.D.; n = 3 replicates in $3 assays.

Fig. 7. Effect of MK571 on Lucifer yellow permeability in C2BBe1 wild-type cells.Experiments were carried out with MK571 in both chambers at concentrations of10–100 mM in Transwell plates for 2 hours at 37�C, followed by Lucifer yellowassay for 1 hour at 37�C. Values represent Lucifer yellow permeability mean6 S.D.;n = 9–10 in each experimental group. Statistical significance was determined by one-way analysis of variance. ***P , 0.001 versus untreated control.

P-gp, BCRP, and MRP2 Knockout in Caco-2 Cells 205

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 8: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

In the present experiments, inhibitors gave equivalent results inwild-type cells compared with the KO cells (full inhibition of efflux)for ranitidine and cimetidine, but not for fexofenadine or colchicine,suggesting substrate dependence as another obstacle when usingchemical inhibition in transport interpretations. The KO cell linesrepresent complete inhibition of the targeted transporter and are anefficient alternative to the suggested use of multiple substrates andinhibitors to ensure coverage of multiple binding sites, substratespecificity, and affinity that may occur when characterizing transporterinteractions in vitro (Brouwer et al., 2013).The problem of inadequate specificity for inhibitors used in trans-

porter assays has been well documented. For example, Matsson et al.(2009) reported that each of the inhibitors used in the current study(verapamil, Ko143, and MK571) have varying degrees of overlap withother efflux transporters at higher concentrations, but were chosen forthese studies due to their common use and commercial availability.Although nonselectivity can be partially addressed by carefullychoosing the inhibitor concentration, it is difficult to accurately assessthe intracellular concentration of inhibitor at the transporter site. Themost promiscuous of the three, MK571, has IC50s of 10, 26, and 50 mMfor MRP2, P-gp, and BCRP, respectively (Matsson et al., 2009). Inaddition, our results and the work of others suggest that MK571inhibits not only MRP2 but other MRPs, although potencies have notbeen established. Virtual docking experiments have shown that MK571binds to the ATP catalytic site, which may contribute to its relativelynonspecific inhibition profile (Matsson et al., 2009). Because MK571is often used at 50 mM or higher concentrations in the literature, off-target effects should be anticipated. Furthermore, we found that MK571negatively affects passive permeability within the cell monolayer whenused at 50 mM or higher, further complicating the interpretation of trans-porter experiments with this inhibitor.In comparison with other formats for studying drug-transporter

interactions, KO cell lines provide a new and complementary ap-proach to determine the profile of efflux transporters with whicha given compound may interact. These cell lines were generated froma human parental line (Caco-2) extensively used for transporter studiesfor .2 decades. This offers the advantage of the presence of the fullcomplement of other relevant human transporters in the same cellsystem, while avoiding potential contributions from nonhuman trans-porters (Kuteykin-Teplyakov et al., 2010). In addition, the double KOcell lines can be used to confirm effects seen in the single KO cells or tostudy the remaining apical efflux transporter in relative isolation.In summary, we have generated stable MDR1, BCRP, and MRP2

single and double KO Caco-2 cell lines using ZFN technology. TheseKO cell lines show complete loss of transporter function using specificsubstrates in the bidirectional transport assay format and are useful inidentifying specific drug-transporter interactions by comparison of trans-port between the wild-type and KO lines. These cell lines represent avaluable tool for application in the assessment of drug-transporterinteractions without dependence on chemical inhibitors with poorlydefined specificities or RNA knockdown systems with residual activities.

Acknowledgments

The authors thank Kelly Keys and Gene Pegg for technical support, TimBrayman and Michael Mitchell for valuable scientific input, and Cole Meyerfor assistance with the tables and figures.

Authorship ContributionsParticipated in research design: Sampson, Bourner, Thompson.Conducted experiments: Sampson, Brinker, Venkatraman, Pratt, Xiao,

Steiner, Blasberg.Contributed new reagents or analytic tools: Brinker, Venkatraman, Pratt.

Performed data analysis: Sampson, Xiao.Wrote or contributed to the writing of the manuscript: Sampson, Thompson.

References

Bourdet DL and Thakker DR (2006) Saturable absorptive transport of the hydrophilic organiccation ranitidine in Caco-2 cells: role of pH-dependent organic cation uptake system andP-glycoprotein. Pharm Res 23:1165–1177.

Brouwer KL, Keppler D, Hoffmaster KA, Bow DA, Cheng Y, Lai Y, Palm JE, Stieger B,and Evers R; International Transporter Consortium (2013) In vitro methods to supporttransporter evaluation in drug discovery and development. Clin Pharmacol Ther 94:95–112.

Celius T, Garberg P, and Lundgren B (2004) Stable suppression of MDR1 gene expression andfunction by RNAi in Caco-2 cells. Biochem Biophys Res Commun 324:365–371.

Chu X, Zhang Z, Yabut J, Horwitz S, Levorse J, Li XQ, Zhu L, Lederman H, Ortiga R,and Strauss J, et al. (2012) Characterization of multidrug resistance 1a/P-glycoprotein knockoutrats generated by zinc finger nucleases. Mol Pharmacol 81:220–227.

Collett A, Higgs NB, Sims E, Rowland M, and Warhurst G (1999) Modulation of the perme-ability of H2 receptor antagonists cimetidine and ranitidine by P-glycoprotein in rat intestineand the human colonic cell line Caco-2. J Pharmacol Exp Ther 288:171–178.

Cvetkovic M, Leake B, Fromm MF, Wilkinson GR, and Kim RB (1999) OATP andP-glycoprotein transporters mediate the cellular uptake and excretion of fexofenadine. DrugMetab Dispos 27:866–871.

Dahan A, Sabit H, and Amidon GL (2009) Multiple efflux pumps are involved in the trans-epithelial transport of colchicine: combined effect of p-glycoprotein and multidrug resistance-associated protein 2 leads to decreased intestinal absorption throughout the entire small intestine.Drug Metab Dispos 37:2028–2036.

Darnell M, Karlsson JE, Owen A, Hidalgo IJ, Li J, Zhang W, and Andersson TB (2010) In-vestigation of the involvement of P-glycoprotein and multidrug resistance-associated protein 2in the efflux of ximelagatran and its metabolites by using short hairpin RNA knockdown inCaco-2 cells. Drug Metab Dispos 38:491–497.

DeGorter MK, Xia CQ, Yang JJ, and Kim RB (2012) Drug transporters in drug efficacy andtoxicity. Annu Rev Pharmacol Toxicol 52:249–273.

Didziapetris R, Japertas P, Avdeef A, and Petrauskas A (2003) Classification analysis ofP-glycoprotein substrate specificity. J Drug Target 11:391–406.

Drescher S, Schaeffeler E, Hitzl M, Hofmann U, Schwab M, Brinkmann U, Eichelbaum M,and Fromm MF (2002) MDR1 gene polymorphisms and disposition of the P-glycoproteinsubstrate fexofenadine. Br J Clin Pharmacol 53:526–534.

Elsby R, Surry DD, Smith VN, and Gray AJ (2008) Validation and application of Caco-2 assaysfor the in vitro evaluation of development candidate drugs as substrates or inhibitors ofP-glycoprotein to support regulatory submissions. Xenobiotica 38:1140–1164.

Giacomini KM, Huang SM, Tweedie DJ, Benet LZ, Brouwer KL, Chu X, Dahlin A, Evers R,Fischer V, and Hillgren KM, et al.; International Transporter Consortium (2010) Membranetransporters in drug development. Nat Rev Drug Discov 9:215–236.

Giri N, Agarwal S, Shaik N, Pan G, Chen Y, and Elmquist WF (2009) Substrate-dependent breastcancer resistance protein (Bcrp1/Abcg2)-mediated interactions: consideration of multiplebinding sites in in vitro assay design. Drug Metab Dispos 37:560–570.

Goh LB, Spears KJ, Yao D, Ayrton A, Morgan P, Roland Wolf C, and Friedberg T (2002)Endogenous drug transporters in in vitro and in vivo models for the prediction of drug dis-position in man. Biochem Pharmacol 64:1569–1578.

Graber-Maier A, Gutmann H, and Drewe J (2010) A new intestinal cell culture model to discriminatethe relative contribution of P-gp and BCRP on transport of substrates such as imatinib.Mol Pharm7:1618–1628.

Hilgendorf C, Ahlin G, Seithel A, Artursson P, Ungell AL, and Karlsson J (2007) Expression ofthirty-six drug transporter genes in human intestine, liver, kidney, and organotypic cell lines.Drug Metab Dispos 35:1333–1340.

Huang L, Be X, Tchaparian EH, Colletti AE, Roberts J, Langley M, Ling Y, Wong BK, and Jin L(2012) Deletion of Abcg2 has differential effects on excretion and pharmacokinetics of probesubstrates in rats. J Pharmacol Exp Ther 343:316–324.

Köck K and Brouwer KL (2012) A perspective on efflux transport proteins in the liver. ClinPharmacol Ther 92:599–612.

Kuteykin-Teplyakov K, Luna-Tortós C, Ambroziak K, and Löscher W (2010) Differences in theexpression of endogenous efflux transporters in MDR1-transfected versus wildtype cell linesaffect P-glycoprotein mediated drug transport. Br J Pharmacol 160:1453–1463.

Lin JH (2007) Transporter-mediated drug interactions: clinical implications and in vitro assess-ment. Expert Opin Drug Metab Toxicol 3:81–92.

Litman T, Druley TE, Stein WD, and Bates SE (2001) From MDR to MXR: new understandingof multidrug resistance systems, their properties and clinical significance. Cell Mol Life Sci58:931–959.

Livak KJ and Schmittgen TD (2001) Analysis of relative gene expression data using real-timequantitative PCR and the 2(-Delta Delta C(T)) method. Methods 25:402–408.

Marquez B and Van Bambeke F (2011) ABC multidrug transporters: target for modulation ofdrug pharmacokinetics and drug-drug interactions. Curr Drug Targets 12:600–620.

Masereeuw R and Russel FG (2012) Regulatory pathways for ATP-binding cassette transportproteins in kidney proximal tubules. AAPS J 14:883–894.

Matsson P, Pedersen JM, Norinder U, Bergström CA, and Artursson P (2009) Identification ofnovel specific and general inhibitors of the three major human ATP-binding cassette trans-porters P-gp, BCRP and MRP2 among registered drugs. Pharm Res 26:1816–1831.

Matsushima S, Maeda K, Hayashi H, Debori Y, Schinkel AH, Schuetz JD, Kusuhara H,and Sugiyama Y (2008) Involvement of multiple efflux transporters in hepatic disposition offexofenadine. Mol Pharmacol 73:1474–1483.

Mease K, Sane R, Podila L, and Taub ME (2012) Differential selectivity of efflux transporterinhibitors in Caco-2 and MDCK-MDR1 monolayers: a strategy to assess the interaction ofa new chemical entity with P-gp, BCRP, and MRP2. J Pharm Sci 101:1888–1897.

Ming X, Knight BM, and Thakker DR (2011) Vectorial transport of fexofenadine across Caco-2cells: involvement of apical uptake and basolateral efflux transporters. Mol Pharm8:1677–1686.

Misaka S, Müller F, and Fromm MF (2013) Clinical relevance of drug efflux pumps in the gut.Curr Opin Pharmacol 13:847–852 DOI: 10.1016/j.coph.2013.08.010.

206 Sampson et al.

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from

Page 9: Zinc Finger Nuclease Mediated Gene Knockout Results in ...dmd.aspetjournals.org/content/dmd/43/2/199.full.pdf · 96-well plates using the FACSAria III (BD Biosciences, San Jose, CA)

Müller F and Fromm MF (2011) Transporter-mediated drug-drug interactions. Pharmacoge-nomics 12:1017–1037.

Oude Elferink RP and de Waart R (2007) Transporters in the intestine limiting drug and toxinabsorption. J Physiol Biochem 63:75–81.

Pavek P, Merino G, Wagenaar E, Bolscher E, Novotna M, Jonker JW, and Schinkel AH (2005)Human breast cancer resistance protein: interactions with steroid drugs, hormones, the dietarycarcinogen 2-amino-1-methyl-6-phenylimidazo(4,5-b)pyridine, and transport of cimetidine.J Pharmacol Exp Ther 312:144–152.

Peterson MD and Mooseker MS (1992) Characterization of the enterocyte-like brush bordercytoskeleton of the C2BBe clones of the human intestinal cell line, Caco-2. J Cell Sci 102:581–600.

Pratt J, Venkatraman N, Brinker A, Xiao Y, Blasberg J, Thompson DC, and Bourner M (2012)Use of zinc finger nuclease technology to knock out efflux transporters in C2BBe1 cells. CurrProtoc Toxicol Chapter 23:Unit 23.2 DOI: 10.1002/0471140856.tx2302s52.

Santiago Y, Chan E, Liu P-Q, Orlando S, Zhang L, Urnov FD, Holmes MC, Guschin D, Waite A,and Miller JC, et al. (2008) Targeted gene knockout in mammalian cells by using engineeredzinc-finger nucleases. Proc Natl Acad Sci USA 105:5809–5814.

Shitara Y, Horie T, and Sugiyama Y (2006) Transporters as a determinant of drug clearance andtissue distribution. Eur J Pharm Sci 27:425–446.

Siissalo S, Hannukainen J, Kolehmainen J, Hirvonen J, and Kaukonen AM (2009) A Caco-2 cellbased screening method for compounds interacting with MRP2 efflux protein. Eur J PharmBiopharm 71:332–338.

Szakács G, Paterson JK, Ludwig JA, Booth-Genthe C, and Gottesman MM (2006) Targetingmultidrug resistance in cancer. Nat Rev Drug Discov 5:219–234.

Takano M, Yumoto R, and Murakami T (2006) Expression and function of efflux drug trans-porters in the intestine. Pharmacol Ther 109:137–161.

Taur JS and Rodriguez-Proteau R (2008) Effects of dietary flavonoids on the transport ofcimetidine via P-glycoprotein and cationic transporters in Caco-2 and LLC-PK1 cell models.Xenobiotica 38:1536–1550.

Tian X, Zamek-Gliszczynski MJ, Li J, Bridges AS, Nezasa K, Patel NJ, Raub TJ, and BrouwerKL (2008) Multidrug resistance-associated protein 2 is primarily responsible for the biliaryexcretion of fexofenadine in mice. Drug Metab Dispos 36:61–64.

Veringa SJ, Biesmans D, van Vuurden DG, Jansen MH, Wedekind LE, Horsman I, Wesseling P,Vandertop WP, Noske DP, and Kaspers GJ, et al. (2013) In vitro drug response and effluxtransporters associated with drug resistance in pediatric high grade glioma and diffuse intrinsicpontine glioma. PLoS One 8:e61512.

Wang L, Zhao Z, Meyer MB, Saha S, Yu M, Guo A, Wisinski KB, Huang W, Cai W, and PikeJW, et al. (2014) CARM1 methylates chromatin remodeling factor BAF155 to enhance tumorprogression and metastasis. Cancer Cell 25:21–36.

Wang Q, Strab R, Kardos P, Ferguson C, Li J, Owen A, and Hidalgo IJ (2008) Application andlimitation of inhibitors in drug-transporter interactions studies. Int J Pharm 356:12–18.

Watanabe T, Onuki R, Yamashita S, Taira K, and Sugiyama Y (2005) Construction of a functionaltransporter analysis system using MDR1 knockdown Caco-2 cells. Pharm Res 22:1287–1293.

Zamek-Gliszczynski MJ, Goldstein KM, Paulman A, Baker TK, and Ryan TP (2013) Minor com-pensatory changes in SAGEMdr1a (P-gp), Bcrp, and Mrp2 knockout rats do not detract from theirutility in the study of transporter-mediated pharmacokinetics. Drug Metab Dispos 41:1174–1178.

Zhang W, Li J, Allen SM, Weiskircher EA, Huang Y, George RA, Fong RG, Owen A,and Hidalgo IJ (2009) Silencing the breast cancer resistance protein expression and function incaco-2 cells using lentiviral vector-based short hairpin RNA. Drug Metab Dispos 37:737–744.

Zhou SF, Wang LL, Di YM, Xue CC, Duan W, Li CG, and Li Y (2008) Substrates and inhibitorsof human multidrug resistance associated proteins and the implications in drug development.Curr Med Chem 15:1981–2039.

Address correspondence to: Dr. David C. Thompson, Sigma-Aldrich Corporation,2909 Laclede Avenue, St. Louis, MO 63103. E-mail: [email protected]

P-gp, BCRP, and MRP2 Knockout in Caco-2 Cells 207

at ASPE

T Journals on A

ugust 28, 2019dm

d.aspetjournals.orgD

ownloaded from