Sensitizing cholangiocarcinoma to chemotherapy by inhibition of the drug-export pump MRP3

Sensitizing cholangiocarcinoma to chemotherapy by inhibition of the drug-export pump MRP3 Maitane Asensio a,b,c , Oscar Briz a,b,c , Elisa Herraez a,b,c , Laura Perez-Silva a,b,c , Ricardo Espinosa-Escudero a , Diego Bueno-Sacristan b,d , Ana Peleteiro-Vigil a , Helen Hammer e , Oliver P¨ otz e,f , Onat Kadioglu g , Jesus M. Banales c,h,i , Maria L. Martinez-Chantar c,j , Matias A. Avila c,k , Rocio I.R. Macias a,b,c , Thomas Efferth g , Jose J.G. Marin a,b,c,*,1 , Elisa Lozano a,b,c,1 a Experimental Hepatology and Drug Targeting (HEVEPHARM), University of Salamanca, Salamanca, Spain b Institute for Biomedical Research of Salamanca (IBSAL), Salamanca, Spain c Centro de Investigaci´ on Biom´ edica en Red de Enfermedades Hep´ aticas y Digestivas (CIBEREHD), Carlos III National Institute of Health, Madrid, Spain d Service of Pathology, University Hospital of Salamanca, Salamanca, Spain e Signatope GmbH, Reutlingen, Germany f Natural and Medical Sciences Institute at the University of Tubingen (NMI), Reutlingen, Germany g Department of Pharmaceutical Biology, Institute of Pharmaceutical and Biomedical Sciences, Johannes Gutenberg University, Mainz, Germany h Department of Liver and Gastrointestinal Diseases, Biogipuzkoa Health Research Institute, Donostia University Hospital, University of the Basque Country (UPV/EHU), Ikerbasque, San Sebastian, Spain i Department of Biochemistry and Genetics, School of Sciences, University of Navarra, Pamplona, Spain j Liver Disease Laboratory, Center for Cooperative Research in Biosciences (CICbioGUNE), Basque Research and Technology Alliance (BRTA), Derio, Spain k Hepatology Laboratory, Solid Tumors Program, Center for Applied Medical Research (CIMA), University of Navarra, Pamplona, Spain ARTICLE INFO Keywords: ABC transporters Biliary tract cancer Chemosensitization Tyrosine kinase inhibitors Natural products ABSTRACT Aims: Drug export through ABC proteins hinders cancer response to chemotherapy. Here, we have evaluated the relevance of MRP3 (ABCC3) in cholangiocarcinoma (CCA) as a potential target to overcome drug resistance. Methods: Gene expression was analyzed in silico using the TCGA-CHOL database and experimentally (mRNA and protein) in resected CCA tumors. The effect of manipulating MRP3 function/expression was evaluated in vitro and in vivo. Results: High MRP3 expression at the plasma membrane of human CCA cells was found. MRP3 overexpression in HEK293T cells selectively impaired the cytotoxic effect of etoposide, cisplatin, SN-38, and mitoxantrone. Reduced MRP3 activity with shRNAs or pan-MRP blockers enhanced the sensitivity to these drugs. MRP3 interaction with natural and semisynthetic compounds (≈40,000) was evaluated by virtual drug screening and molecular docking. Two identified potential MRP3 inhibitors (EM-114, EM-188), and sorafenib impaired MRP3 transport activity and enhanced sensitivity of CCA cells to etoposide and cisplatin. The antitumor effect of cisplatin in the mouse xenograft model was enhanced by co-treatment with sorafenib, which was accompanied by a higher intratumor accumulation of cisplatin. Conclusions: Genetic and pharmacological MRP3 inhibition enhances the anti-CCA effect of several drugs, which constitutes a promising strategy to improve the response to chemotherapy in CCA patients. Abbreviations: ABC, ATP-binding cassette; AT, adjacent non-tumor; EGFP, enhanced green fluorescent protein; 5-FU, 5-fluorouracil; CCA, cholangiocarcinoma; CF, carboxyfluorescein; IF, immunofluorescence; LBE, ligand binding energy; MDR, multidrug resistance; MOC, mechanism of chemoresistance; MRP, multidrug resistance-associated protein; MTX, mitoxantrone; qPCR, quantitative PCR; RT, reverse transcription; TCGA, the Human Cancer Genome Atlas database; TKI, tyrosine kinase inhibitor; TXP, triple X proteomics. * Correspondence to: Department of Physiology and Pharmacology, University of Salamanca, Campus Miguel de Unamuno, E.D. Lab-231, Salamanca 37007, Spain E-mail address: [email protected] (J.J.G. Marin). 1 Both authors contributed equally as senior authors to this work. Contents lists available at ScienceDirect Biomedicine & Pharmacotherapy journal homepage: www.elsevier.com/locate/biopha https://doi.org/10.1016/j.biopha.2024.117533 Received 4 August 2024; Received in revised form 25 September 2024; Accepted 4 October 2024 Biomedicine & Pharmacotherapy 180 (2024) 117533 Available online 13 October 2024 0753-3322/© 2024 The Authors. Published by Elsevier Masson SAS. This is an open access article under the CC BY-NC license ( http://creativecommons.org/licenses/by-nc/4.0/ ). 1. Introduction Cholangiocarcinoma (CCA) is the second most frequent primary hepatobiliary cancer after hepatocellular carcinoma (HCC), comprising approximately 15 % of all primary liver tumors [1]. It derives from the malignant transformation of cholangiocytes, cells lining the biliary tree, and it can be classified according to the anatomic location of the tumor into three groups: i) intrahepatic CCA (iCCA), when the tumor arises above the second order bile ducts, within the liver parenchyma; ii) perihilar CCA (pCCA), if it grows in the right and/or left hepatic duct and their junction; and iii) distal CCA (dCCA), a subtype arising at the common bile duct. Previously, pCCA and dCCA were referred to as extrahepatic CCA (eCCA). Despite being a rare disease, incidence and mortality rates have been on a rising trend during the past few decades [2,3], which, together with its dismal prognosis, turns the CCA into a global health problem. Surgical resection is the only potentially curative treatment for CCA, especially if achieving surgical margins free of neoplasm (R0) [4,5]. In addition, the last clinical practice guidelines of the European Association for the Study of the Liver (EASL) and the International Liver Cancer Association (ILCA) for iCCA management recommends a 6-month course of adjuvant capecitabine or S1 as the standard of care after curative surgery [6]. Unfortunately, the development of CCA is usually asymptomatic, explaining why this type of cancer is often diagnosed in advanced stages of the disease, when surgical resection is not feasible. In these cases, gemcitabine and cisplatin plus humanized monoclonal antibodies able to block the programmed death protein 1 ligand (PD-L1), such as durvalumab and pembrolizumab, are currently recommended as the first-line systemic therapeutic option [6,7]. Second-line therapies include FOLFOX, i.e., the combination of oxaliplatin plus 5-fluorouracil (5-FU) or targeted therapies against druggable mutations such as ivosidenib (mutated IDH1) and pemigatinib and futibatinib (FGFR2 fusions) [6]. In addition, irinotecan plus 5-FU (FOLFIRI) is often used worldwide as a second-line option in patients with contraindications to FOLFOX or as a third-line regimen [6]. Despite the range of therapeutic options available, systemic chemotherapy remains ineffective, with a 5-year survival rate of <10 % [4]. The lack of response of CCA patients to chemotherapy may be attributed to the presence of different mechanisms of chemoresistance (MOCs) that work together in tumor cells to protect them from the cytotoxic activity of chemotherapy [8]. Moreover, these MOCs can be stimulated by pharmacological treatment, a phenomenon known as acquired resistance. One group of these MOCs is based on the ability of cancer cells to reduce intracellular drug concentrations, either by decreasing drug uptake (MOC-1a), for example, due to a lower expression of members of the solute carrier (SLC) superfamily of transporters; or by enhancing drug export (MOC-1b), mainly due to pumps belonging to the superfamily of ATP-binding cassette (ABC) proteins. Among the seven families of ABC transporters, only some members of the ABCB, ABCC, and ABCG families have been associated with cancer multidrug resistance (MDR) [9,10]. Nevertheless, their role in CCA remains still unclear. Thus, multidrug resistance protein 1 (MDR1 or P-glycoprotein, gene symbol ABCB1), and breast cancer resistance protein (BCRP, ABCG2), can transport a wide variety of antitumor drugs and are highly and moderately expressed in CCA, respectively [10]. Regarding the ABCC family, elevated levels of the multidrug resistance-associated protein 1 (MRP1, ABCC1) in iCCA have been associated with poorer overall survival [11]. The presence in the plasma membrane of CCA cells of another member of this family, MRP3 (ABCC3), has been reported [12,13]. Interestingly, MRP3 can transport several antitumor drugs, such as teniposide, etoposide [14,15], and methotrexate [16]. Consistently, MRP3 expression levels correlate with resistance to etoposide and anthracyclines in CCA cells [17]. In previous studies, we have demonstrated that SOX17 overexpression increases the response of CCA cells to 5-FU in vivo, probably because this transcription factor can negatively regulate MRP3 expression [18]. Besides, MRP3 has also been considered a target for the treatment of pancreatic adenocarcinoma [19], esophageal squamous cell carcinoma [20], and breast cancer [21]. However, the actual impact of MRP3 on the response of CCA to antitumor drugs is unknown. Therefore, this work aims to evaluate the relevance of MRP3 in the MDR phenotype of this tumor and its potential as a target for overcoming CCA chemoresistance. 2. Materials and methods 2.1. Human samples and in silico analysis Samples from surgically resected eCCA (n=15) and iCCA (n=35) tissue and paired adjacent non-tumor (AT) (n=9) liver tissue, were obtained from the biobank of the Salamanca University Hospital (Salamanca cohort). Patients had not received chemotherapy before tumor resection. Clinical and demographic data are summarized in Supplementary Table S1. Research protocols were approved by the Ethical Committee for Clinical Research of Salamanca University (March 30, 2020, reference PI 2020–02–42) and conducted according to the principles expressed in the Declaration of Helsinki. All patients signed written consents for using tissue samples for biomedical research. Moreover, an in silico study was carried out using transcriptomic data from CCA samples (iCCA, n=30; eCCA, n=6) and paired AT (n=9) publicly available at the TCGA-CHOL dataset included in the Human Cancer Genome Atlas database (TCGA). RNA-seq expression data (HiSeq Illumina platform) from the TCGA-CHOL cohort were mapped onto the GRCh38 version of the human genome. Values of FPKM (Fragments Per Kilobase of transcript per Million mapped reads) were used to compare gene expression between samples. 2.2. Chemicals 5(6)-Carboxyfluorescein diacetate (CF), 5-FU, 7-ethyl-10-hydroxycamptothecin (SN-38), apatinib, axitinib, bosutinib, cisdiamineplatinum(II) dichloride (cisplatin), erlotinib, etoposide, gefitinib, imatinib, lapatinib, mitoxantrone, MK571, nilotinib, pazopanib, and sunitinib were obtained from Sigma-Aldrich (Merck, Madrid, Spain). Gemcitabine and oxaliplatin were from Acros Organics (Thermo Fisher). Dasatinib was from Santa Cruz Biotechnology (Heidelberg, Germany). Brivanib, cabozantinib, cediranib, dovitinib, futibatinib, ivosidenib, lenvatinib, linifanib, pemigatinib, ponatinib, regorafenib, and tivozanib were from Selleckchem (Deltaclon, Madrid, Spain). Sorafenib was given by the Pharmacy Department of Salamanca University Hospital (Salamanca, Spain). The compounds (EM-XXX) selected from the docking analysis and used in the screening of potential MRP3 inhibitors were purchased in MolPort (Riga, Latvia). The purity of all these compounds was ≥98 %. All other chemicals were of analytical grade. 2.3. Cell cultures EGI-1 (ACC 385) and TFK-1 (ACC 344) cells derived from eCCA cases were obtained from the German Collection of Microorganisms and Cell Cultures (DSMZ, Braunschweig, Germany). The human embryonic kidney HEK293T cell line (ATCC CRL-3216) was purchased from the American Type Culture Collection (LGC Standards, Barcelona, Spain). The AGS cell line derived from human gastric adenocarcinoma (GAC) was obtained from the European Collection of Authenticated Cell Cultures (Sigma-Aldrich). EGI-1 and TFK-1 cells were cultured in DMEMlow glucose (Sigma-Aldrich) and RPMI-1640 medium (Gibco, Thermo Fisher, Madrid, Spain), respectively. HEK293T grew in DMEM-high glucose medium and AGS cells in F12 medium (Gibco, Thermo Fisher). All media were supplemented with 10 % heat-inactivated fetal bovine serum (FBS) and 1 % penicillin/streptomycin solution B (Gibco, Thermo Fisher). Glutamax® (HEK293T and AGS) and Non-Essential Amino Acid solution (EGI-1) (Gibco, Thermo Fisher) were also added to the media. The absence of contamination by mycoplasmas was M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 2 confirmed using the Mycoplasma Gel Form Kit (Biotools B&M Labs, Madrid). 2.4. Cloning procedures RNA isolated from EGI-1 cells was converted to cDNA using oligodT primers by RT. Then, the ABCC3 coding sequence was amplified into two separate amplicons (Amp1 and Amp2) by high-fidelity PCR carried out using AccuPrime Pfx DNA polymerase (Thermo Fisher) using specific primers (Supplementary Table S2). Both amplicons were ligated using BfuI and then cloned into the MluI and SpeI sites of the pWPI-V5 lentiviral vector. The resulting construct was confirmed by sequencing. The empty lentiviral vector pWPI-V5, containing the constitutive EF-1 α promoter and the V5 tag downstream of the cloned insert, was used as a control (MOCK group). The constructs were analyzed by restriction enzyme digestion and nucleotide sequencing to confirm the absence of mutations in the MRP3 coding sequence. To generate knock-out cells using CRISPR/Cas9, single guide RNAs (sgRNAs) with high efficiency (highest on-target score) and specificity (highest off-target score), targeting the selected exon of ABCC1, ABCC3, or ABCC4, which was shared by all variants, were designed using Design Custom gRNA software (Integrated DNA Technologies, IA, USA) (Supplementary Table S3). Complementary oligonucleotides were subjected to hybridization and cloned into a pGEM plasmid containing a cassette consisting of a U6 promoter, a scaffold, and terminator sequences. 2.5. Lentiviral particle production Recombinant lentiviruses were produced as previously published [18]. In brief, HEK293T cells were transfected using a standard polyethylenimine (PEI) protocol with the pWPI-derived transfer vectors, and the packaging plasmids psPAX2 and pMD2.G and viral particles were precipitated from the culture media by ultracentrifugation. Viral titers were determined by transduction of HEK293T cells with serial dilutions of the viral suspension, and subsequent analysis of EGFP-positive cells was carried out by flow cytometry (FACSCalibur, BD Biosciences, Madrid). Lentiviral particles carrying shRNAs against human MRP3 and the corresponding control lentivirus (MOCK) were purchased from Origene Technologies (Rockville, USA). 2.6. Generation of cell models with forced expression or silenced MRP3 Lentiviral vectors were added to target cells at a multiplicity of infection (MOI) of 25 infectious particles per cell in the presence of hexadimethrine bromide (Polybrene, Merck). Specifically, overexpression of MRP3 was achieved after transduction of HEK293T cells, whereas CCA (EGI-1) cells were targeted with shRNA lentivirus to reduce endogenous MRP3 expression levels. Transduced cells were maintained for at least four days before clone selection was carried out by the limiting dilution method (HEK-MRP3) or using 1 μ g/ml puromycin as a selection agent (EGI-1-shMRP3). MRP3 enhanced or silenced expression was verified by measuring mRNA and protein levels by RTqPCR and western blot, respectively. To generate the selected ABCCknockout model, AGS cells were co-transfected using Lipofectamine™ LTX transfection reagent (Invitrogen, Thermo Fisher) with the corresponding sgRNA-plasmid and a plasmid containing humanized S. pyogenes Cas9 (SpCas9) and puromycin resistance gene [22]. Transfected cells were selected after treatment with 1 μ g/ml puromycin for three days. MOCK cells were transfected with the same plasmids, but without containing sgRNA and were also treated with puromycin. The efficacy of gene silencing by adding insertions and deletions (indels) to the genome of cells was tested by a T7 endonuclease-based assay using the GeneArt™ Genomic Cleavage Detection kit (Invitrogen). Single-cell clones were obtained by the limited dilution method. Knock-out of each targeted MRP was verified by sequencing the edited exon after DNA extraction and measuring protein expression by western blot (data not shown). 2.7. Determination of mRNA levels Total RNA from cells and tissues was isolated using RNA mini-spin columns treated with RNase-free DNase I (GE Healthcare, Madrid). cDNA was synthesized by RT from total RNA using random hexamers and the avian myeloblastosis virus (AMV) retrotranscriptase included in the Cloned AMV First-strand cDNA Synthesis kit (Thermo Fisher). qPCR was performed using gene-specific primers spanning exon-exon junctions in the target mRNA (Supplementary Table S4) and AmpliTaq Gold™ DNA polymerase and SYBR Green I detection kit in a QuantStudio™ 3 Real-Time PCR System (Applied Biosystems, Thermo Fisher). The following thermal cycling conditions were used: single cycles at 50◦C for 2 min and at 95◦C for 10 min, followed by 40 cycles at 95◦C for 15 s and at 60◦C for 60 s. The mRNA abundance for each target gene in each sample was normalized to the mRNA content of GAPDH and HPRT1. Expression levels were calculated as 2 -ΔCt , where ΔCt was the difference of Ct in each sample between the target gene and the normalizer. 2.8. Western blot, immunofluorescence, and immunohistochemistry assays Western blot analyses of cell lysates were carried out in 7.5 % SDSPAGE, loading 30 μ g of protein per lane, and transferred onto nitrocellulose membranes (Bio-Rad, Madrid). Primary antibodies used to detect MRPs were from Alexis/Enzo Life Sciences (MRP1, clone MRPr1; MRP2, clone M2III-6), Sigma-Aldrich (MRP3), and Abcam (MRP4). Primary antibody (sc-32233) against GAPDH (clone 6C5) was from Santa Cruz Biotechnology (Santa Cruz, CA). The appropriate horseradish peroxidase-linked secondary antibodies were used (Invitrogen). An enhanced chemoluminescence detection system (Hybond ECL; GE Healthcare) was used to visualize the bands. Immunofluorescent (IF) staining was carried out in cells grown on coverslips, after fixation with ice-cold methanol and 1 h incubation with the corresponding primary antibodies (MRP1 and MRP2 from Enzo Life Science; Na + /K + -ATPase and MRP3, M3II-9, from Abcam; MRP4, from Novus Biologicals). The secondary antibodies Alexa Fluor-488 and Alexa Fluor-594 (Thermo Fisher) were diluted 1:2000. Nuclei were counterstained with Dapi. Confocal laser-scanning immunofluorescence microscopy (Leica TCS SP2) was performed. Immunohistochemistry (IHC) was performed at the Pathology Service of the Salamanca University Hospital using sections from formalinfixed paraffin-embedded (FFPE) blocks, with antigen retrieval at pH 6.0 and incubation for 40 min with a rabbit anti-MRP3 antibody (M0318, Sigma) diluted 1:50 in a Leica Biosystems BOND-III Fully Automated IHC and ISH Stainer. Slides were counter-stained with hematoxylin (Palex, Madrid), mounted with aqueous mounting medium, and scanned at ×200 magnification using the dotSlide virtual image system (Olympus, Tokyo, Japan) at the Compared Molecular Pathology Service of the Salamanca Cancer Research Center (CIC). Images of the sections were obtained using Olympus software (DotSlide, OlyVIA, Olympus). 2.9. Immunoaffinity LC-MS/MS Protein quantification in tumor samples of iCCA (n=8) and AT (n=7) was performed using a sensitive immunoaffinity technique termed triple X proteomics (TXP) [23], which is based on the reduction of sample complexity by trypsin proteolysis, followed by separation and detection of the peptides of interest using antibodies against these small epitopes (“TXP antibodies”) and liquid chromatography-mass spectrometry (LC-MS/MS) [24]. The previously published experimental procedure has been followed [25]. Peptide amounts were calculated from ratios of the integrated peaks of endogenous peptides and isotopically labeled M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 3 standards. Protein amounts were normalized and expressed as fmol per µg of extracted protein. 2.10. Cytostatic effect CCA (EGI-1) cells were seeded in 96-well plates (4500 cells/well) and incubated for 24 h before being treated with the desired drug for short (6 h) or long (72 h) incubation times, as required. 72 h after starting the exposure of the culture to the assayed drugs, cell viability was determined with the MTT test using thiazole blue tetrazolium bromide (Sigma-Aldrich). When evaluating the role of sorafenib as a chemosensitizer, both compounds were added in combination at different concentrations, starting from 0 until approximately the IC 50 of each drug. Cell viability was determined with the MTT test after 72 h, and data were analyzed with the online tool “Synergy Finder 3.0” and the zero-interaction potency (ZIP) model [26]. 2.11. Proliferation assays CCA (EGI-1) cells were cultured in 24-well plates (2 ×10 5 cells/well) coated with ibiTreat® polymer (Ibidi, Inycom, Madrid). After 24 h, the antitumor drug was added in the presence or absence of the potential inhibitor of MRP3 for a short (6 h) incubation time, after which the medium was replaced and cell proliferation was analyzed during 72 h using the HoloMonitor® Live Cell Imaging System (PHI, Boston, USA). If inhibitors were evaluated in combination with antitumor agents, cells were pre-incubated with the inhibitors for 2 h at 37◦C before adding the combination, and the chemosensitizer was maintained until the end of the experiment. 2.12. Transport assays To study the ability of different compounds to inhibit MRP3, cells were seeded in 6-well plates and incubated at 37◦C for 48 h. After trypsinization, cells in suspension were incubated for 15 min at 37◦C in 100 μ l of “transport medium” (96 mM NaCl, 5.3 mM KCl, 1.1 mM KH 2 PO 4 , 0.8 mM MgSO 4 , 1.8 mM CaCl 2 , 11 mM glucose, and 50 mM HEPES, pH 7.40) containing 1 μ M (HEK293T and EGI-1) or 0.5 μ M (AGS) of the fluorescent substrate CF with or without 10 μ M of MK571, used as a pan-MRP inhibitor, or the compounds that were tested as potential inhibitors. The loading period was stopped by dilution with 900 μ l of ice-cold “transport medium”, and the intracellular fluorescence was determined using a FACSCalibur™ flow cytometer (Becton-Dickinson). In cells transduced with vectors containing the EGFP reporter gene, the resulting fluorescence was corrected with that of these cells without CF incubation. For sorafenib transport assays, 1.5 ×10 5 cells were seeded in 12-well plates and incubated with 5 µM sorafenib at 37◦C during the indicated time. If evaluating the efflux activity of MRP3, the medium was replaced after a 30-min uptake period, and cells were incubated at 37◦C for the indicated time with sorafenib-free fresh medium. Sorafenib transport was stopped by rinsing cells four times with ice-cold “transport medium” and cells were then lysed in 250 µl of water with 5 µM prednisolone, used as an internal standard. Sorafenib content was determined by an adaptation [27] of a previously published method [28] using HPLC-MS/MS in a 6420 Triple Quad LC/MS (Agilent Technologies, Santa Clara, CA, USA). Briefly, separation was carried out by HPLC on a Zorbax C18 column (30 mm ×2.1 mm, 3.5 μ m) using 65:35 methanol/water, containing 5 mM ammonium acetate and 0.1 % formic acid, pH 4.6. The flow rate was 0.3 ml/min at 35◦C. Positive electrospray ionization was carried out with the following conditions: gas temperature 350◦C, gas flow 10 L/min, nebulizer 20 psi, and capillary voltage 2500 V. In multiple reaction monitoring mode (MRM), the specific m/z transition 465.1 m/z to 252.0 and 270.0 m/z was followed. Prednisolone was used as an internal standard (transition 361.0 m/z to 343.0 and 307.0 m/z). The results were corrected by protein content. Sorafenib levels were also measured in the medium after the efflux period. 2.13. Virtual drug screening and molecular docking At the beginning of this study, the human MRP3 protein had not been crystallized yet. For that reason, a homology model of the 3D structure of human MRP3 was developed using USCF-CHIMERA 1.14 software and the bovine MRP1 (PDB code: 6BHU) as a template. The library of natural and semi-synthetic compounds was obtained from ZINC database (https://zinc15.docking.org/ accessed on 07 May 2018). The docking studies were performed using PyRx [29] and AutoDock 4.2.6. software (The Scripps Research Institute, CA, USA). The grid box was placed around the reported drug bindingsite [30]. The center of the grid box was set at x=149.73, y=159.49, and z=152.91. Docking parameters were set to 250 runs and 25,000,000 energy evaluations for each cycle with at least four separate calculations. Selected compounds as potential MRP3 inhibitors met the following criteria: i) higher local-binding energy (LBE) values than the pan-MRP3 inhibitor, MK571; ii) to be commercially available; and iii) low price. 2.14. In vivo experiments Female nude mice (Swiss nu/nu) were purchased from Charles River Laboratories (Barcelona, Spain) and housed in sterile micro-isolator cages under controlled conditions of temperature (20◦C), humidity, and light/dark cycle (12 h/12 h), and fed on standard rodent chow (Panlab, Madrid) and water ad libitum. All animals received human care in accordance with the relevant guidelines and regulations, and all protocols were approved by the University of Salamanca Ethical Committee for Laboratory Animals and Junta de Castilla y Leon (November 9, 2023, reference 908). EGI-1 cells (n ≈10 6 ) were subcutaneously injected into the flanks of anesthetized 8-week-old nude mice. When the tumors reached the desired volume (≈12–20 mm 3 ), animals were randomly divided into four groups: i) Control, receiving vehicle alone (n=4); ii) SOR, receiving sorafenib i.p. (n=3); iii) CisPt, receiving cisplatin i.p. (n=3); and iv) SOR+CisPt, receiving both sorafenib and cisplatin i.p. (n=5). Animals were treated twice per week for 23 days. The doses of sorafenib (10 mg/ kg b.w.) and cisplatin (4.5 mg/kg b.w.) were selected based on previously published reports [25,31]. The tumors were measured twice per week with a sliding caliper, and their volume was calculated using the formula (length ×width 2 )/2. The animals were sacrificed at the end of the experiment, and tumor and blood samples were collected. Platinum content in tumors was measured by flameless atomic absorption spectrophotometry after mineralization of tumor samples as previously described [32]. The results were corrected by tumor weight. Serum levels of routine biochemical parameters were determined in a dry chemistry automated analyzer Spotchem EZ SP-4430 (Arkray Factory, A. Menarini Diagnostics, Badalona, Spain). 2.15. Statistical analyses Data management and statistical analysis have been done using Microsoft Office Excel (version 365) and GraphPad Prism 8. Unless otherwise indicated, data are shown as mean±SEM. If analyzing the differences between the two groups, paired or unpaired Student t-tests were used as appropriate. When comparing more than two mean values, one-way ANOVA followed by the Bonferroni method (parametric) or Kruskal-Wallis method (non-parametric) of multiple-range testing was used. Differences were considered significant if p<0.05. 3. Results 3.1. MRP3 is highly expressed in human CCA samples and cell lines To elucidate the relevance of MRP3 in the MDR phenotype of CCA, M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 4 we analyzed the mRNA levels of drug efflux pumps of the ABCC family MRP1–6 (ABCC1–6), of human CCA tumors included in the TCGA-CHOL database. Compared to the adjacent hepatic tissue (AT) formed mainly by hepatocytes, the expression of ABCC1, ABCC3, ABCC4, and ABCC5 was found elevated in iCCA and eCCA, whereas that of ABCC2 and ABCC6 was lower (Fig. 1A). The order of mRNA abundance was ABCC3>ABCC1>ABCC4≈ABCC5≈ABCC6>ABCC2. These results were validated by RT-qPCR in our experimental samples (Salamanca cohort, Fig. 1B). Despite being lower compared to AT, ABCC3 mRNA levels stood out in CCA. Of note, no differences were found between the anatomical subtypes of CCA in none of the cohorts. In addition, the measurement of protein abundance by immunoaffinity LC-MS/MS analysis of CCA samples confirmed the expression changes found at the mRNA level. Thus, protein levels of MRP1 and MRP3 were higher in iCCA compared with the paired AT, while MRP2 and MRP6 levels were markedly lower (Table 1). These results also confirmed that MRP3 is the most abundantly expressed MRP pump in CCA, which prompted us to carry out the immunohistochemical analysis of MRP3 in a reduced number of CCA samples (n=10), as a proof of concept. Low MRP3 expression was found in the control liver (Fig. 1C), confirming what had been previously reported [33]. As also expected, MRP3-positive staining at the basolateral membrane of hepatocytes and cholangiocytes was increased in a cholestatic liver, used here as a positive control (Fig. 1D). Interestingly, despite the existence of marked inter-individual heterogeneity, strong/moderate membranous staining of MRP3 protein in iCCA (Fig. 1E-F) and eCCA (Fig. 1G-H) cells was found in a high proportion of cases (8/10) and no staining was found in a few of them (2/10). The investigation of CCA-derived cell lines (EGI-1 and TFK-1) revealed that the MRP expression pattern was similar to that found in CCA tissue samples. Thus, these cells showed high levels of MRP3 (both mRNA and protein) located at the plasma membrane (Figure S1), being the most abundantly expressed MRP, followed by MRP1 and MRP2. 3.2. MRP3 overexpression confers resistance to selected antitumor drugs in vitro To study the ability of MRP3 to export antitumor drugs, a cell model with forced MRP3 expression was generated using HEK293T cells, which were used because they lack endogenous MRP3 expression. The ABCC3 coding sequence was cloned into a lentiviral vector used to transduce HEK293T cells, which were then subjected to monoclonal selection (Figure S2). High MRP3 expression in the four monoclonal sublines was confirmed by RT-qPCR and western blot. Moreover, MRP3 was correctly located at the plasma membrane of these cells (Figure S2C). Functional studies revealed that HEK-MRP3 cells showed a higher capacity to efflux CF (an MRP fluorescent substrate) than MOCK cells, as well as a decreased sensitivity to etoposide (an MRP3 substrate) (Figure S2D-F). However, MRP3 overexpression did not affect the proliferation of these cells in the absence of this drug (data not shown). Among the four monoclonal sublines, that with the highest expression of functional MRP3 was selected as the HEK-MRP3 model for further in vitro experiments. In previous studies, we have demonstrated that shortening the exposure time of the antitumor drug from 72 h to 6 h resulted in a better Fig. 1. Relative ABCC pump expression levels obtained from the TCGA database (A) of intrahepatic cholangiocarcinoma (iCCA, n=30), extrahepatic cholangiocarcinoma (eCCA, n=6) and adjacent non-tumor liver tissue (AT, n=9), and measured by RT-qPCR (B) in samples from the Salamanca cohort (iCCA, n=35; eCCA, n=15, AT, n=9). Boxes show the median and the whiskers indicate the maximum and minimum values; individual values are represented as circles. Data are expressed as FPKM (Fragments per Kilobase of Transcript per Million Mapped Reads) (A) or as percentage of the normalizer, GAPDH (B). *, p<0.05, compared with AT. Representative images of MRP3 staining by immunohistochemistry and hematoxylin counterstaining in control liver (C), liver with cholestasis (D), intrahepatic cholangiocarcinoma (iCCA) (E, F) and extrahepatic cholangiocarcinoma (eCCA) (G, H). Black arrows indicate bile ducts in AT (C, D). Table 1 MRP Protein abundance in intrahepatic cholangiocarcinoma. MRP protein Protein abundance (fmol/µg protein) AT iCCA MRP1 0.03±0.02 0.98±0.32 a MRP2 1.78±0.30 0.23±0.22 a MRP3 0.18±0.11 2.90±1.02 a MRP4 ND ND MRP5 0.55±0.30 0.62±0.46 MRP6 5.38±0.42 0.87±0.41 a Values from immunoaffinity LC-MS/MS analyses are shown as mean±SEM. AT, adjacent liver tissue (n=7); iCCA, intrahepatic cholangiocarcinoma (n=8); ND , not detected. a , p<0.05, as compared with AT. M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 5 condition to evaluate the contribution of the overexpressed transporter in total cell content and response to the drug, minimizing the entrance of the drug by unspecific mechanisms occurring during longer incubation times [18,34]. Therefore, using this “window exposure” approach HEK-MOCK and HEK-MRP3 cells were incubated with the main antineoplastic drugs used to treat CCA. Subsequently, cell sensitivity was measured 66 h later (Fig. 2A). Besides etoposide, included in this study as a typical MRP3 drug substrate, the classical first-line chemotherapeutic drugs for CCA cisplatin and gemcitabine, and also oxaliplatin, 5-FU, and the active metabolite of irinotecan, SN-38, used as second-line agents, were tested (Fig. 2B). In addition, targeted drugs approved for the treatment of CCA (pemigatinib and futibatinib) and HCC (sorafenib, regorafenib, cabozantinib) were also included in this study (Fig. 2C). The selection of the doses was based on 72 h sensitivity studies (Figure S3), choosing the concentration in the culture medium that reduced cell viability by 70–80 %, which, in our experience, is the appropriate concentration to move on to experiments with the 6 h “window exposure”. Under these experimental conditions, MRP3 overexpression significantly reduced cell sensitivity to etoposide, cisplatin, oxaliplatin, mitoxantrone, SN-38, and cabozantinib, whereas no modified effect was found for gemcitabine, 5-FU, and several TKIs (Fig. 2B-C). Fig. 2. Effect of MRP3 overexpression in the sensitivity to antitumor drugs. Schematic representation of the “window exposure” experimental procedure (A). HEK293T-MOCK (MOCK) and HEK293T-MRP3 (MRP3) cells were incubated with vehicle alone (Control) or a selected concentration of classic chemotherapeutics (B) and tyrosine kinase inhibitors (C) for 6 h and cell viability was determined by MTT assay 66 h later. Selected concentration of drugs: etoposide 5 µM (ETO), cisplatin 20 µM (CisPt), oxaliplatin 20 µM (OxaPt), gemcitabine 40 nM (GEM), 5-fluorouracil 10 µM (5-FU), mitoxantrone 200 nM (MTX), SN-38 200 nM, and 20 µM of cabozantinib (CAB), futibatinib (FUT), pazopanib (PAZ), pemigatinib (PEM) regorafenib (REG), and sorafenib (SOR). Values are mean±SEM of at least 5 experiments per triplicate. *, p<0.05, compared with MOCK. Effect of MRP3 silencing in the sensitivity of EGI-1 cells after a short-term exposure to antitumor drugs (A). EGI-1 cells were transduced with lentiviral vectors containing scrambled shRNA (shMOCK) or MRP3-targeting (shMRP3) shRNA and subjected to monoclonal selection. Cytostatic effect in EGI-1-shMOCK and -shMRP3 cells after incubation with vehicle (DMSO, Control) or the indicated drugs for 6 h (D). Selected concentrations: etoposide 20 µM (ETO), cisplatin 40 µM (CisPt), oxaliplatin 20 µM (OxaPt), 5-fluorouracil 25 µM (5-FU), SN-38 1 µM, mitoxantrone 0.5 µM (MTX) and cabozantinib 25 µM (CAB). Values are mean±SEM of at least 5 experiments per triplicate. *, p<0.05, compared with cells treated with shMOCK. MRP3-dependent sensitivity to the cytotoxic effect of anticancer drugs using “continuous exposure” experimental procedure (E). The effect of MRP3 inhibition with 10 µM MK571, a pan-MRP inhibitor, was evaluated in wild-type EGI-1 cells (F-H). The effect of MRP3 silencing was determined in shMOCK and shMRP3 EGI-1 cells (I-K). Cell viability was measured after 72 h of incubation with increasing concentrations of etoposide (F, I), cisplatin (G, J), and SN-38 (H, K). Values are mean±SEM of at least 4 experiments per triplicate. *, p<0.05, compared with cells in the absence of MK571 or treated with shMOCK. M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 6 3.3. Reduced MRP3 function enhances the response to antitumor drugs in vitro To further evaluate the role of MRP3 in drug sensitivity, we used EGI1 cells, which endogenously express high levels of this pump (Figure S1). Firstly, we used a “continuous exposure” approach to incubate EGI-1 cells with the pan-MRP inhibitor MK571 (Fig. 2E), known to inhibit several MRPs, including MRP3 [35,36]. Incubation of EGI-1 cells with 10 µM of MK571 induced mild toxicity (Figure S5A). However, the combination of MK571 with selected antitumor drugs resulted in a significantly increased sensitivity to etoposide (Fig. 2F), cisplatin (Fig. 2G), and SN-38 (Fig. 2H). To determine the specificity of MRP3-mediated effect on the response to antitumor drugs, MRP3 expression was reduced in EGI-1 cells. They were stably transduced with lentiviral vectors containing scrambled shRNA (EGI-1-shMOCK) or ABCC3-targeting shRNA (EGI-1shMRP3). Subsequently, the cells were subjected to monoclonal selection using puromycin. IF analysis revealed a complete loss of MRP3 protein in EGI-1-shMRP3 clones, which was accompanied by a significant increase in the sensitivity to etoposide when compared with shMOCK cells (Figure S4A). EGI-1-shMRP3 #2 was selected for further experiments. EGI-1 shMRP3 cells were significantly more sensitive to etoposide, cisplatin, and SN-38 after drug exposure for 72 h (Fig. 2I-K). This effect was more clearly seen in “window exposure” experiments (Fig. 2A) in which they also showed an increased sensitivity to mitoxantrone (Fig. 2D). However, the response to oxaliplatin, 5-FU, and cabozantinib was not modified by MRP3 silencing in this cell model (Fig. 2D and Supplementary Table S6). 3.4. Discovery of new MRP3 inhibitors Since these results highlighted the impact of MRP3 on the chemoresistance of several drugs and there are no specific inhibitors of MRP3 commercially available, we looked for new potential MRP3 inhibitors using two different approaches. On the one hand, a panel of natural and semisynthetic compounds (≈40,000) was screened in silico by molecular docking, and on the other hand, the potential usefulness of available TKIs as MRP3 inhibitors was studied. 3.4.1. Molecular docking and selection of potential chemosensitizers Among the top 100 ligands with the ligand binding energie (LBE) (Supplementary Table S7), eight compounds were chosen for functional evaluation of their ability to inhibit MRP3-mediated transport (Table 2). To ensure the entry of the potential inhibitors into the cells, these were incubated with the compounds for 2 h before the assay. In the absence of inhibitors, HEK-MRP3 cells accumulated significantly less CF than HEKMOCK cells (Fig. 3A-B). MK571-induced MRP inhibition resulted in an increased intracellular concentration of CF in both types of cells. Interestingly, six of the assayed compounds were able to enhance CF accumulation in HEK-MRP3 cells (EM-760>EM-114>EM-530>EM188≈EM-505≈EM-033), suggesting that they inhibit MRP3-mediated CF efflux. Besides, these compounds, except for EM-114, did not modify CF content in HEK-MOCK cells, which suggests that their effect could be due to specific MRP3 inhibition. EM-760 and EM-114 showed the highest MRP3-specific inhibitory index (Fig. 3C), which was of the same order as that of MK571. Nevertheless, EM-760, and EM-530 were discarded from further studies due to their high cellular toxicity (Table 2). To further confirm the ability of the selected compounds to reduce MRP3-mediated transport, we took advantage of another cell model previously developed in our group based on adenocarcinoma gastric (AGS) cells. This model permits analyzing selective MRP3mediated activity without the risk of artifacts associated with overexpression of this pump. In AGS cells all endogenous MRPs had been knocked out using CRISPR-Cas9 technology (AGS-MOCK) or all except MRP3 (AGS-MRP3). After a loading period, CF content was lower in AGS-MRP3 than in AGS-MOCK cells. This was significantly increased by MK571 and by EM-114 and EM-188 (Fig. 3D-F). EM-114 showed a less MRP3-specific profile, also affecting CF load in AGS-MOCK cells (Fig. 3D). However, EM-114-induced inhibition was more potent than that of EM-188 (Fig. 3D-F). The next step was the measurement of EM-114 and EM-188 concentration-dependent cytotoxicity in CCA cells (Figure S5B) to select the non-toxic range to perform functional studies. Under these circumstances, EM-114 significantly and EM-188, in a weaker way, enhanced the CF load of EGI-1 cells (Fig. 3G-H). To assess the chemosensitizing effect of these inhibitors, real-time cell growth was determined using holographic imaging microscopy in EGI-1 cells (Fig. 3I). Control and inhibitor-only groups showed similar proliferation rates (Fig. 3J). However, etoposide significantly reduced the growth rate, which was markedly enhanced by co-treatment with MK571. A similar but weaker effect was observed in the presence of EM-144 or EM-188 (Fig. 3I-J). On the contrary, the antiproliferative effect of cisplatin was not enhanced either by EM-114 or EM-188 (Figure S6). 3.4.2. Tyrosine kinase inhibitors as chemosensitizers through MRP3 inhibition Considering the modest effect of EM-114 and EM-188 as chemosensitizers observed in the in vitro models used here, we sought other potential MRP3 inhibitors. To this aim, we investigated a panel of 24 TKIs based on their ability to inhibit the activity of some ABC pumps. In EGI-1 cells, only sorafenib could significantly increase the intracellular accumulation of CF, whereas other TKIs, such as erlotinib, tivozanib, and nilotinib, induced only a slight increase (Fig. 4A). In HEK-MRP3 cells, MK571 and sorafenib also reduced MRP3-mediated CF efflux (Fig. 4B). In MOCK cells, the accumulation of CF was higher in the presence of these compounds. As a non-specific MRP inhibitor, MK571 induced a higher accumulation of CF than sorafenib in HEK-MOCK cells (Fig. 4B). These results suggest that sorafenib is a less potent inhibitor of Table 2 Potential human MRP3 inhibitors selected by molecular docking. Name ZINC ID MW (g/mol) LBE (kcal/mol) Cytotoxicity Structure PyRx AutoDock In silico In vitro EM−114 ZINC000245235114 628.98 −11.63 ±0.06 −12.84±0.55 ++ ++ Diterpenoid EM−661 ZINC000079216661 672.86 −11.37 ±0.23 −12.55±0.29 +NA Saponin EM−033 ZINC000005434033 562.55 −13.72 ±0.05 −10.12±0.29 + ++ Lignane EM−760 ZINC000253504760 780.95 −12.02 ±0.17 −10.34±0.49 ++++ ++++ Cardiac glycoside EM−505 ZINC000100392505 483.52 −11.12 ±0.09 −12.31±1.72 ++ + Alkaloid EM−188 ZINC000252587188 766.97 −11.20 ±0.00 −9.98±0.33 ++ + Saponin EM−530 ZINC000238809530 869.06 −10.87 ±0.05 −9.75±0.28 +++ +++ Saponin EM−119 ZINC000253501119 766.92 −11.20 ±0.26 −9.82±0.33 ++ NA Sterol glycoside Compounds are sorted according to the Ligand Binding Energy (LBE), which is expressed as mean ±SD of at least 7 runs of docking performed by Autodock Vina and Autodock Tools against the binding pocket of human MRP3. LBE for the pan-MRP inhibitor MK571 was −8.88±0.66. Cytotoxicity of the compounds was predicted using the ProTox-II software or determined in HEK-293T cells by MTT test in two independent experiments. Toxicity determined in in vitro experiments was considered Low/+(IC 50 >100 µM), Moderate/++ (IC 50 10–100 µM); High/+++ (IC 50 1–10 µM); or Very high/++++ (IC 50 <1µM). MW, molecular weight; NA, not assayed. M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 7 MRPs other than MRP3 expressed in HEK293T cells. In contrast, in HEKMRP3 cells, the ability of sorafenib to induce an increased accumulation of CF was slightly higher than that of MK571. To unravel whether the inhibitory effect of sorafenib was due to substrate competition, docking analysis in Pyrx program was performed against the ligand binding site of human MRP3. LBE for sorafenib (-9.85 ±0.07 kcal/mol) was similar to that of the well-known substrate of MRP3 estradiol 17β-D-glucuronide (-9.05±0.07 kcal/mol) and higher than that of MK571 (-7.50±0.01 kcal/mol), suggesting that this TKI might interact with the ligand binding pocket of MRP3 and perhaps be a substrate of the pump. To test this hypothesis, sorafenib uptake and efflux in HEK-MOCK and HEK-MRP3 cells were determined by HPLC-MS/MS (Fig. 4C-E). MRP3 overexpression did not affect intracellular sorafenib accumulation or efflux. Moreover, the amount of sorafenib in the culture medium after the efflux period was similar regardless the cells expressed or not MRP3 (Fig. 4C-E). These results suggest that sorafenib-induced MRP3 inhibition is not due to substrate competition. Finally, the ability of sorafenib to improve cell response to antitumor drugs was evaluated in CCA cells. EGI-1 cells were exposed to a range of concentrations from 0 to approximately the IC 50 of sorafenib and etoposide (Fig. 5A-B) or cisplatin (Fig. 5C-D). The combination of sorafenib with either antitumor drug achieved a significant reduction in cell viability in EGI-1 cells compared with etoposide or cisplatin alone at concentrations of the TKI ranging 0.6–5.0 µM. Overall, sorafenib showed a synergistic effect on cell viability when combined with etoposide or cisplatin, probably mediated by MRP3 inhibition. 3.5. MRP3 inhibition with sorafenib increases cisplatin antitumor effect in vivo Using a subcutaneous xenograft model in immunodeficient mice, the efficacy of sorafenib as a chemosensitizer in CCA was assessed in vivo. The animals were treated twice per week with sorafenib, cisplatin, or a combination of both drugs. Sorafenib did not affect serum biomarkers of liver and kidney function and did not worsen the known renal toxicity of cisplatin (Supplementary Table S8). The monotherapy using sorafenib or cisplatin alone induced a slight reduction in tumor growth compared with mice receiving only the vehicle (Control group) (Fig. 6A-B). However, the combination of sorafenib with cisplatin markedly decreased tumor volume compared with either drug alone (Fig. 6C). These findings were consistent with the higher accumulation of platinum in the tumors of mice receiving the combination of cisplatin with sorafenib (Fig. 6D). Fig. 3. Screening of potential inhibitors (EM-XXX) of the ability of MRP3 to transport carboxyfluorescein (CF). Two cell models were used: epithelial (HEK293T) cells either control (MOCK) or stably overexpressing MRP3 (MRP3) (A-C) and manipulated gastric adenocarcinoma (AGS) cells either control with the absence of all MRPs expression (AGS-MRP-KO; MOCK) or exclusively expressing endogenous MRP3 (MRP3) (D-F). Cells were incubated with 10 µM of the corresponding inhibitor at 37 ◦C for 2 h prior the transport assay. Afterward, cells were incubated with CF in the absence (Control) or presence of the inhibitors at 37◦C for a particular time, after which CF cell content was determined by flow cytometry. MRP3-Specific Inhibitory Index (C, F), CF content in the presence vs. in the absence of inhibitor in MRP3 cells. Values are mean±SEM of at least 5 experiments per duplicate. *, p<0.05, compared with Control; y, p<0.05, compared with MOCK. Effect of MRP3 inhibitors in the transport activity of this pump (G, H) and sensitivity to etoposide (I, J) in EGI-1 cells. These cells were incubated with 1 µM CF for 15 min in the absence (Control) or presence of MK571 10 µM or EM-XXX 50 µM, and cell content of CF was determined by flow cytometry to determine uptake (G) and the MRP3-specific Inhibitory Index (H). Representative recording of real-time proliferation of EGI-1 cells as monitored by holographic phase imaging after the exposure for 6 h to 5 µM etoposide (ETO) in the absence or presence of the indicated inhibitors, in this case, the inhibitor was incubated with the cells for 2 h before starting the assays (I). Images of 8 regions per well were captured every 60 min for 72 h and analyzed with the appropriate software. The proliferation rate was calculated as the inverse of doubling time (J). Values are mean±SEM of 5 separate experiments performed in duplicate. *, p<0.05, compared with Control; y, p<0.05, compared to ETO alone. M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 8 Fig. 4. Effect of tyrosine kinase inhibitors (TKIs) on the ability of MRP3 to transport carboxyfluorescein (CF) (A). EGI-1 cells were incubated with 1 µM of CF with or without 10 µM of the corresponding compound at 37 ◦C for 15 min, after which CF cell content was determined by flow cytometry. MK571 was used as a known panMRP inhibitor. Values are mean±SEM from data determined in 6 separate cultures. *, p<0.05, compared with Control. Evaluation of the interaction of sorafenib (SOR) with MRP3 transport activity (B-E). Epithelial (HEK293T) cells either control (MOCK) or stably overexpressing MRP3 (MRP3) were incubated with 10 µM MK571, a pan-MRP inhibitor, or 10 µM SOR for 2 h. Then, the cells were incubated with 1 µM CF in the absence or the presence of 10 µM MK571 or SOR for 15 min (uptake phase), and the intracellular content of CF was measured by flow cytometry (B). To determine SOR uptake (C) or efflux (D), cells were incubated with 5 µM SOR at 37◦C and cell drug content (uptake phase) was measured at min 15 and 30 (C), or their culture medium was replaced by SOR-free fresh media (efflux phase), and SOR cell content was determined at min 5, 15 and 30 min. SOR levels were measured in cell lysates (D) or culture media per well (E) by HPLC-MS/MS. Values are mean±SEM from data determined in 5 separate cultures. *, p<0.05, compared with Control. y, p<0.05, comparing MRP3 with MOCK. M. Asensio et al. Biomedicine & Pharmacotherapy 180 (2024) 117533 9