Androstano-arylpyrimidines: Novel small molecule inhibitors of MDR1 for sensitizing multidrug-resistant breast cancer cells

Apart the numerous physiological functions of MDR1, it is widely known for its role in granting multidrug resistance to cancer cells. This ATP-driven transmembrane protein exports a wide range of chemotherapeutic agents from cancer cells, thereby deterring drugs to reach effective intracellular concentrations. Thus, inhibition of MDR1 expression or function would be a viable option to enhance the accumulation of cytotoxic agents in cancer cells which in turn could improve significantly the success rate of chemotherapy. Although, several pharmacological inhibitors have been designed and tested in the past, due to their unsuccessful translation to clinical application, there is still ongoing research to find suitable compounds to manipulate MDR1 function and potentially overturn multidrug resistance. In the present study, we demonstrate that novel DHT-derived A-ring-fused arylpyrimidinone derivatives, based on their acetylation status, can inhibit MDR1 efflux activity in MDR1 overexpressing multidrug-resistant breast adenocarcinoma cells. Strikingly, all derivatives carrying an acetoxy group on the sterane D -ring were highly potent in hindering Rhodamine 123 export via MDR1, however deacetylated molecules were not capable to exert a similar effect on multidrug resistant cancer cells. The possible molecular and cellular mechanisms underlying the efflux pump inhibiting function of acetylated derivatives were dissected using the most potent MDR1 inhibitor, compound 10g and its deacetylated counterpart ( 11g ). Importantly, molecule 10g was able to sensitize drug resistant cells to doxorubicin-induced apoptosis, further verifying the highly advantageous nature of efflux pump inhibition upon chemotherapy. Our experiments also revealed that neither mitochondrial da- mage, nor MDR1 gene regulation could lay behind the MDR1 inhibitory function of compound 10g . Molecular docking studies were carried out to analyze the interactions of 10g and 11g with MDR1, however no significant differences in their binding properties were observed. Nevertheless, our results indicate that the ER stress in-ducing potential of molecule 10g might be the fundamental mechanism behind its inhibitory action on MDR1 . With additional studies, our work can yield a structural platform for a new generation of small molecule MDR1 inhibitors to sensitize drug resistant cancer cells and at the same time it elucidates the exemplary involvement of endoplasmic reticulum stress in the molecular events to defeat multidrug resistance.

Apart from the numerous physiological functions of MDR1, it is widely known for its role in granting multidrug resistance to cancer cells. This ATP-driven transmembrane protein exports a wide range of chemotherapeutic agents from cancer cells, thereby deterring drugs to reach effective intracellular concentrations. Thus, inhibition of MDR1 expression or function would be a viable option to enhance the accumulation of cytotoxic agents in cancer cells which in turn could improve significantly the success rate of chemotherapy. Although, several pharmacological inhibitors have been designed and tested in the past, due to their unsuccessful translation to clinical application, there is still ongoing research to find suitable compounds to manipulate MDR1 function and potentially overturn multidrug resistance.
In the present study, we demonstrate that novel DHT-derived A-ring-fused arylpyrimidinone derivatives, based on their acetylation status, can inhibit MDR1 efflux activity in MDR1 overexpressing multidrug-resistant breast adenocarcinoma cells. Strikingly, all derivatives carrying an acetoxy group on the sterane D-ring were highly potent in hindering Rhodamine 123 export via MDR1, however deacetylated molecules were not capable to exert a similar effect on multidrug resistant cancer cells. The possible molecular and cellular mechanisms underlying the efflux pump inhibiting function of acetylated derivatives were dissected using the most potent MDR1 inhibitor, compound 10g and its deacetylated counterpart (11g). Importantly, molecule 10g was able to sensitize drug resistant cells to doxorubicin-induced apoptosis, further verifying the highly advantageous nature of efflux pump inhibition upon chemotherapy. Our experiments also revealed that neither mitochondrial damage, nor MDR1 gene regulation could lay behind the MDR1 inhibitory function of compound 10g. Molecular docking studies were carried out to analyze the interactions of 10g and 11g with MDR1, however no significant differences in their binding properties were observed. Nevertheless, our results indicate that the ER stress inducing potential of molecule 10g might be the fundamental mechanism behind its inhibitory action on MDR1. With additional studies, our work can yield a structural platform for a new generation of small molecule MDR1 inhibitors to sensitize drug resistant cancer cells and at the same time it elucidates the exemplary involvement of endoplasmic reticulum stress in the molecular events to defeat multidrug resistance.

Introduction
Although a considerable amount of scientific effort has been dedicated to understand the cellular and molecular events that drive cancer recurrence, invasion and metastasis, and important advancements have been achieved in identifying potential pharmacological targets, nevertheless, adequately competent therapeutic approaches to defeat malignant cells are still wanted. The first line treatment for most cancer types is usually, small molecular drug-based chemotherapy, where the development of drug resistance is, unfortunately, a likely outcome. Decreased sensitivity to several cytotoxic drugs and broad cross-resistance to a large number of structurally dissimilar antineoplastic agents emerge quickly, advancing the evolution of multidrug-resistant (MDR) cancer phenotypes. MDR, either intrinsic or acquired, displays a substantial hurdle to effective cancer therapy leading to poor patient survival (Luqmani, 2005). Although the cellular and molecular features of MDR involve modification of signaling pathways, endurance of oxidative stress and increased apoptotic threshold, the major component of the cancer cells' strategy to reduce cellular accumulation and thereby evading the toxic effects of chemotherapy drugs is the overexpression of various efflux pumps. Such an ATP-dependent transmembrane drug transporter protein is P-glycoprotein, also known as MDR1 or ABCB1, which is encoded by the multidrug resistance gene 1 (MDR1). Via such ABC transporters -membrane-residing pumps containing ATP binding motives -cytotoxic agents can be readily expelled from the cytoplasm immediately after their uptake, at almost the same speed as they entered tumor cells, leaving no time to exert their anticancer activities (Gillet and Gottesman, 2010;Szakács et al., 2014Szakács et al., , 2004. As a consequence, a broad spectrum of approaches was tested to suppress the expression and the function of ABC transporters, in particular, those of MDR1, in order to potentially overturn MDR (Callaghan et al., 2014;Chen and Tiwari, 2011;Choi and Yu, 2014). Along this line, based on the protein structure and on the conformational changes associated with MDR1 efflux activity, a multitude of inhibitors have been designed and developed to hinder drug transport across MDR1; sadly, translation of their application to the clinical practice have failed owing to sideeffects related to their pharmacological properties (i.e. cardiotoxicity elicited by verapamil treatment) (Coley, 2010;Ekins et al., 2002;Kathawala et al., 2015;Robert and Jarry, 2003). Despite the somewhat disappointing results of the clinical trials performed with some of the selected MDR1 inhibitors, there is still intensive ongoing research to find the ultimate compound/s with the unique capability to manipulate specifically MDR1 activity without exerting unpredictable toxicities, thereby opening avenues to overcome MDR cancer.
Among the numerous candidates designed for MDR1 inhibition, several natural and synthetic steroids have already been examined (De Ravel et al., 2015). In synthetic studies, physiologically relevant steroids have been utilized as precursors, where structure optimization was aimed to establish or enhance the inhibiting performance of the molecule/s on MDR1 without triggering undesired interactions. Modulation of the substitution profile of natural sex-hormones, glucocorticoids or other steroids enables a huge versatility of chemical modifications leading to a substantial transformation of the physicochemical, biological and pharmacokinetic properties of the parent compound. In fact, progesterone and its synthetic derivatives have been thoroughly screened and it was found that the fairly weak activity of progesterone on MDR1 could be enhanced by introducing 7α-thiophenyl groups, 11α-benzoate, and carbamate substituents or via modifications at the 17β-acetyl side chain (Ichikawa-Haraguchi et al., 1993;Leonessa et al., 2002). MDR1 inhibiting properties were observed also for steroidal anti-estrogens, for some synthetic glucocorticoid derivatives, as well as for modified primary and secondary conjugated and unconjugated bile acids (Cooray et al., 2006;Lo et al., 2008;Rocheblave et al., 2016). Based on structure−activity studies, modeling and docking investigations and fluorescence-based techniques, it seems plausible that interactions between MDR1 and its steroid modulators are centered near the ATP-binding domain of the transporter, suggesting that the influence on efflux activity exhibited by these compounds is associated with the ATP-driven conformational change required for drug transport (Dayan et al., 1997;Li et al., 2005;Mares-Samano et al., 2009a).
Modifications of natural steroids by heterocycles either connected to or condensed with one of the steroid rings have gathered grounds in recent developments. Among various modifications, the introduction of pyrimidines is exceptionally relevant as the resultant hybrid derivatives with high stability, tunable composition and multifunctionality, manifest an extensive range of biological properties, such as antiviral, antimicrobial, antioxidant, as well as anti-cancer activities (Gore and Rajput, 2013;Kaur et al., 2014;Mohana Roopan and Sompalle, 2016). In view of the biological relevance of steroidal heterocycles, novel arylpyrimidine-fused androstanes were recently synthesized via threecomponent, modified Biginelli-type reactions under microwave irradiation and the cytotoxicity of the compounds was screened in vitro on two prostate cancer (PC-3 and DU 145), on MCF-7 breast cancer and on non-cancerous lung fibroblast (MRC-5) cell lines (Baji et al., 2017). Remarkable association between structure and biological activity were noted since acetylated, A-ring-fused 4′-arylpyrimidin-2′-one derivatives of dihydrotestosterone (DHT) 17-acetate exhibited significant cytotoxicity, whereas deacetylated analogs failed to induce cancer cell death. We concluded that arylpyrimidines can be utilized in future drug design and synthetic approaches as structural scaffolds of androstanes, motivating the rational design and pharmacological investigation of additional derivatives and the highly advantageous cytotoxic features of acetylated androstanes with the arylpyrimidine modifications could be further exploited to kill different types of cancer cells.
Thus, our next aim was to test the remarkable anticancer performance of acetylated DHT-arylpyrimidinone derivatives on cancer cells with multidrug-resistant phenotype and to examine the impact of these compounds on the efflux activity and on the cellular responses of MDR cancer cells. Therefore, the present study was undertaken with the aim to examine the MDR1 inhibiting properties of a series of semi-synthetic, acetylated as well as deacetylated A-ring-modified heterocyclic DHT derivatives -selected from the compound library synthesized and investigated previously (Baji et al., 2017) -in order to reveal structure−activity criteria relevant for attenuating drug efflux via MDR1. Since we could identify several compounds exhibiting prominent MDR1 inhibiting capacity, using one suitable compound the mechanism behind its action on multidrug-resistant breast cancer cells has been investigated in details. Since attenuation of MDR1 activity by the synthetized androstane derivatives can be the result of transcriptional and translational suppression of MDR1, or of mitochondrial dysfunction leading to decreased ATP generation, apoptosis or endoplasmic reticulum stress, we tested these cellular and molecular events in multidrug-resistant MCF-7/KCR cells, following treatments with the most promising DHT derivative.

Synthesis of androstano-pyrimidinones
The synthesis of arylpyrimidinone-modified DHT derivatives was reported previously (Baji et al., 2017). Briefly, DHT-acetate (1 mmol), para-substituted benzaldehyde (2 mmol) and urea (1 mmol) were dissolved in acetic acid (10 mL) and concentrated H 2 SO 4 (2 drops) was added. The mixture was irradiated in a closed vessel at 110°C for 10 min using a CEM Corporation Focused Microwave System, Model Discover SP. After completion of the reaction, the mixture was poured into saturated NaHCO 3 solution (10 mL) and extracted with dichloromethane (2 × 10 mL). The combined organic layers were washed with water (10 mL), dried over anhydrous Na 2 SO 4 and evaporated in vacuo. The residue was then dissolved in acetone (20 mL) and 1 mL Jones reagent (1.8 M) was added. The mixture was poured into water after 30 min of stirring at room temperature, and extracted with dichloromethane (2 × 10 mL). The combined organic layers were dried over anhydrous Na 2 SO 4 and evaporated in vacuo. The crude product was purified by column chromatography using MeOH/CH 2 Cl 2 = 5:95 as eluent to give 10b−g (Yields: 60−69%). For deacetylation, compound 10b-g (0.25 mmol) was dissolved in methanol (10 mL), and KOH (1 mmol) was added. The solution was stirred at room temperature for 8 h, and then diluted with water. After adding NH 4 Cl, the mixture was cooled and the resulting precipitate was filtered off, washed with water and dried to give 11b−g (Yield: 85−95%). The original compound numbers used in Baji et al. (Baji et al., 2017) were retained in this manuscript for clarity and comparability.

Cell culture
The MCF-7 human breast adenocarcinoma cell line was obtained from ATCC. The drug-resistant MCF-7/KCR cell line was developed from MCF-7 under selection pressure using doxorubicin from 10 nM to 1 µM concentration (Kars et al., 2006). Cell lines were maintained in RPMI-1640 medium (LONZA) supplemented with 10% FBS, 2 mM glutamine and penicillin-streptomycin solution at 37°C, 5% CO 2 and 95% humidity. MCF-7/KCR cells were cultured in media with 1 µM doxorubicin for 1 week then in culture medium without doxorubicin for 1 week to maintain the drug-resistant phenotype. Before experiments, MCF-7/KCR cells were grown in a doxorubicin-free medium.

Cell viability assay
Cells were seeded at 10 4 /well density in 96-well plates. On the following day, cells were treated with either 20 µM of compound 10g or 11g for 24 h or with verapamil in 40 µM concentration for 2 h. To detect the sensitizing capability of compound 10g on multidrug-resistant MCF-7/KCR cells to doxorubicin-induced toxicity, cells were treated with 20 µM of compound 10g in combination with doxorubicin in 10; 20; 40; 50 and 60 µM concentration. Following treatments cells were washed and incubated with RPMI-1640 medium containing 0.5 mg/mL MTT reagent (Sigma-Aldrich). Formazan crystals were solubilized in DMSO and absorbance was measured at 570 nm using a Synergy HTX microplate reader (BIOTEK®). Measurements were repeated three times using 4 independent biological replicates. Absorbance values of the untreated control samples were considered as 100% viability.

Apoptosis detection
Cells were seeded at 2 × 10 6 cells/well density in 6-well plates. On the following day, cells were treated either with compound 10g or 11g in 20 µM concentration or with doxorubicin (5 µM) for 24 h. Combinational treatments using doxorubicin (5 µM) + verapamil (4 µM) and doxorubicin (5 µM) + 10g (20 µM) as well as doxorubicin (5 µM) + 11g (20 µM) were also performed. Cells were collected and stained with AnnexinV-FITC/propidium iodide (PI) (Life Technologies) according to the manufacturer's recommendation. Since fluorescent properties of PI are similar to doxorubicin, we did not show PI filters in the dot plots during analysis. Fluorescence intensities of at least 10,000 cells/sample were measured by FACSCalibur™ and data were analyzed by FlowJo V10 software. Fluorescence of Annexin V-FITC and forward scatter (FSC) were represented. Experiments were repeated three times using at least three replicates.

Molecular docking assay
Homology model of the human MDR1 was obtained from the SWISS-MODEL Repository (P08183) (Bienert et al., 2016). The threedimensional structure of 10g and 11g were built manually and energy minimized using Chem3D Pro-ver 12.0. Molecular dockings were performed with AutoDock ver. 4.2.6 ( Morris et al., 2010). To cover the putative ATP binding site of the receptor the grid volume was set up to 80 × 80 × 80 Å with 0.375 Å spacing and centered to the nucleotidebinding domain 2. Rotatable bonds of 10g and 11g as well as receptor amino acid side chains in contact with ATP were kept flexible (Kadioglu et al., 2016;Mares-Sámano et al., 2009b). Blind docking was carried out using the Lamarckian Genetic Algorithm, and 1000 dockings were done. The resultant docked complexes were ranked according to the corresponding binding free energies and inspected with Auto-Dock Tools ver. 1.5.6 graphical interface. The lowest energy complex in which 10g and 11g formed interactions with amino acids in the putative ATP binding site was selected and images were generated using Pymol ver. 2.3.4 (PyMOL Molecular Graphics System, Schrödinger, LLC.).

PharmMapper target prediction
PharmMapper is a web server that predicts therapeutic drug targets for small molecules provided as query. 2D structures of 10g and 11g were drawn using PubChem sketcher ("PubChem Sketcher V2.4,"). The downloaded structures were given as an input for PharmMapper available at http://www.lilab-ecust.cn/pharmmapper/. For human targets parametric value was set to 2241, maximum number of matched targets was set to 100 and all other parameters were set to default values.

JC-1 staining
To measure changes in the mitochondrial membrane potential after treatments, JC-1 staining was performed. Cells were seeded onto coverslips placed into 24-well plates (10 5 cells/well). On the following day, cells were treated with compound 10g or 11g in 20 µM concentration or with carbachol in 500 µM concentration for 24 h. Carbachol was used as a positive control, as it induces mitochondrial damage by raising cytoplasmic calcium levels. Since JC-1 is a MDR1 substrate, therefore, before JC-1 loading, 40 µM of verapamil was added to the samples for 1 h. Then cells were washed and incubated with RPMI-1640 medium containing 10 µg/mL JC-1 (Life Technologies) for 15 min. Coverslips were inversely mounted in Fluoromount™ (ThermoFisher) on glass slides and JC-1 fluorescence was visualized by OLYMPUS BX51 microscope equipped with an Olympus DP70 camera using the same exposition time for all samples. Image analysis was performed by ImageJ software. We have to note that pretreatment of each sample with verapamil was required to improve intracellular retention of JC-1, a MDR1 substrate, however, in those samples which were treated with compound 10g inevitably more JC-1 was retained since these cells received two MDR1 inhibitors (verapamil and compound 10g) (fig S1). As a consequence, we could not apply the usual analysis and representation of JC-1 staining (aggregate/monomer ratio), but instead, the average number of mitochondria/cell was counted in case of 100 cells in each sample and expressed as a measure of mitochondrial damage. Experiments were repeated three times using three independent biological replicates.

Reverse transcription and real-time PCR
Cells were seeded at 2 × 10 6 cells/well density in 6-well plates. On the next day, cells were treated either with compound 10g or 11g in 20 µM concentration or with dithiothreitol (DTT, 2 mM) for 24 h. Total cellular RNA was prepared using RNeasy® Mini Kit (QIAGEN) according to the manufacturer's recommendation. 45 ng/µL RNA was reverse transcribed (TaqMan® Reverse Transcription kit, Applied Biosystems). PCR reactions were performed on PicoReal™ Real-time PCR (Thermo Scientific) using SYBR Green qPCR Master Mix (Thermo Scientific) with an input of 100 ng cDNA. Each primer (Table 1) was used at 200 nM concentration. Relative transcript levels were determined by the ΔΔCt analysis using GAPDH as the reference gene. Experiments were repeated three times using three biological replicates.

Compound 10g inhibits MDR1 efflux activity in multidrug-resistant breast cancer cells
Based on structure−cytotoxicity relationships verified previously on numerous cancer cell lines (Baji et al., 2017), compounds 10b, 10c, 10d, 10f and 10g and their deacetylated counterparts 11b, 11c, 11d, 11f and 11g (chemical structures are shown in Fig 1 and supplementary Fig S2) were selected from the compound library of synthetic heterocyclic androstane derivatives for the present investigation (the original compound numbers used in Baji et al. (Baji et al., 2017) were retained for clarity and comparability). Multidrug-resistant MCF-7/KCR cells were treated with these substances at 20 µM concentration for 24 h. Drug-resistant phenotype of MCF-7/KCR cells was confirmed previously by detecting significant MDR1 expression and high efflux activity in these cells (Gopisetty et al., 2019) and the applied concentration of the test compounds and the treatment time were established according to preliminary experiments (unpublished results). Rhodamine 123 drug efflux assay was performed to examine the potential inhibitory effects of the test compounds on MDR1 activity. The fluorescent dye Rhodamine 123 (RH123) is a substrate of MDR1, thus it can be readily expelled from drug-resistant cancer cells, whereas intracellular retention of RH123 demonstrates inhibition of MDR1. As a positive control, cells were treated with the known MDR1 inhibitor verapamil. The obtained mean intracellular RH123 fluorescence values in MCF-7/KCR cells upon treatments with the test compounds are represented in Table S1.
As expected, verapamil treatment resulted in the highest mean intracellular RH123 fluorescence (Fig 2A and Table S1). Compared to the fluorescence of cells receiving verapamil treatment (set as 100%) the lowest intracellular fluorescence was found after treating MCF-7/KCR cells with compounds 11b (4.6%), 11d (4.6%), 11f (6.4%), 11g (8.7%) and 11c (9%), whereas the fluorescence of cells exposed to most acetylated derivatives -apart from compound 10d (9.6%) -was significantly higher i.e. 10f (24.6%), 10b (30.2%), 10c (42.6%) and 10g (44.4%) (see Table S1). All the acetylated derivatives (10b, 10c, 10d, 10f, and 10g) manifested significant inhibitory potential on MDR1 efflux activity, demonstrated by significantly higher intracellular RH123 fluorescence values, compared to the deacetylated counterparts (11b, 11c, 11d, 11f, and 11g) (Fig 2A, Fig S3 and Table S1), which clearly indicates a strong structure−function relationship manifesting in multidrug-resistant MCF-7/KCR cells. Among the tested compounds, treatments with compound 10g resulted in the highest intracellular RH123 accumulation, i.e. the strongest MDR1 inhibition (Fig 2A and Table S1). Therefore, compound 10g was selected and applied in subsequent experiments to uncover the precise mechanisms underlying the inhibition of MDR1 efflux activity. To expose the possible differences in molecular actions between the acetylated compound (10g) and its deacetylated counterpart, compound 11g (the fluorescence intensity of cells treated with 11g was only 8.7% compared to the fluorescence of positive control) was also included in successive investigations. First, it was validated that the concentration of 20 µM of compounds 10g and 11g -applied in the RH123 efflux assay -would not cause significant cytotoxicity on MCF-7/KCR cells. For this, MTT assays were performed after incubating the cells for 24 h with compounds 10g and 11g. Results show that compared to untreated control the viable percentage of MCF-7/KCR cells treated with 20 µM of 10g or 11g was 77.8 ± 22.8 and 97.5 ± 19.3, respectively. This indicates that neither of these two compounds is able to induce serious cytotoxicity of drug resistant cells at 20 µM concentration (Fig 2B).
DHT is a parent compound of the selected synthesized derivatives, including 10g and 11g. To examine the possibility of the involvement of hormonal effects exhibited by 10g (or by 11g) underlying MDR1 inhibition, DHT was tested for its inhibitory activity on MDR1-mediated efflux in multidrug-resistant breast cancer cells. To achieve this, MCF-7/KCR cells were treated with 20 µM DHT for 24 h then Rhodamine 123 assay was performed. Our results indicate that DHT treatment does not lead to the inhibition of MDR1 activity (Fig 2C).

10g treatment sensitizes multidrug-resistant breast cancer cells to doxorubicin-induced killing
As the inhibition of MDR1 efflux leads to the intracellular retention of chemotherapy drugs, it prolongs the time for the drug to exert its anticancer activity within cancer cells, thus it improves the therapeutic efficacy. Therefore, treatment with 10g, a MDR1 inhibitor, might sensitize multidrug-resistant MCF-7/KCR cells to doxorubicin-induced cytotoxicity. To test this hypothesis, MCF-7/KCR cells were treated with 20 µM of 10g and with doxorubicin in 10, 20, 40, 50 and 60 µM concentrations for 24 h and compared the results with those obtained on cells receiving only doxorubicin treatment at identical concentrations as indicated above. We observed that treatment with 10g significantly augmented the cytotoxicity provoked by doxorubicin in MCF-7/KCR cells (Fig 3A). The IC 50 values calculated from viability curves were 55.96 ± 0.1 µM for doxorubicin and 35.26 ± 0.1 µM for 10g+doxorubicin treatments, respectively ( Fig S4). Next, we wanted to check Table 1 Sequence of primers used in qPCR experiments.

Primer
Forward Reverse M.K. Gopisetty, et al. European Journal of Pharmaceutical Sciences 156 (2021) 105587 whether the improved toxic effects of doxorubicin in the presence of compound 10 g are the result of enhanced apoptosis induction. MCF-7/ KCR cells were again treated with either 20 µM of 10 g or 5 µM doxorubicin or with the combination of 20 µM 10 g and 5 µM doxorubicin for 24 h. Cells receiving only DMSO were considered as negative control and those treated with 4 µM verapamil for 24 h were set as positive control. The percentage of apoptotic cells (Q2+Q3) was 0.97% in control samples, and it was 0.59% in doxorubicin-treated samples. When cells were exposed to compound 10 g 1.86% of cells, by doxorubicin+verapamil-treated cells 26.55%, and finally, 43.22% of 10 g+doxorubicin-exposed cells were undergoing apoptosis (Fig 3B). In line with the viability results, Annexin V/PI apoptosis assay clearly showed that treatments with compound 10 g+doxorubicin resulted in a significantly higher number of apoptotic cells compared to cells receiving only doxorubicin treatment ( Fig 3B). Surprisingly, the drug sensitizing effect of 10 g seems to be stronger than that of verapamil, a well-known MDR1 inhibitor. As expected, treatments with 20 µM of 11 g alone or in combination with 5 µM doxorubicin did not influence  MCF-7/KCR sensitivity to doxorubicin-induced apoptosis (Fig S5). These results verify the MDR1 inhibiting feature of molecule 10 g and indicate that in fact, there is a link between MDR1 inhibition and improved doxorubicin sensitivity upon 10 g treatment.

In silico approaches predicted different binding targets of 10g and 11g but similar binding to MDR1
Molecular docking study was carried out to predict if compound 10g or 11g is able to establish interactions with MDR1. In the lowest energy complex, wherein ligands were bound to the nucleotide-binding domain 2 of MDR1 (Fig 4A), the predicted binding free energy was −6,53 kcal/mol for 10g, and −5.20 kcal/mol for 11g, respectively. The acetoxy group of 10g formed a hydrogen bond with Tyr 1044 while the methyl group on the phenyl ring was predicted to interact with the side chain of Arg 1047 . The pyrimidine ring attached with Arg 262 and Asp 805 side chains, while Cys 1074 was connected to the terminal cyclopentane ring (Fig 4B). Similar binding pose was revealed in case of the deacetylated counterpart. The acetoxy substituting hydroxyl group faced to Tyr 1044 and Arg 1047 residues, respectively. The main difference in the formed interactions of 11g that the methyl group on the phenyl ring formed interaction with the side chain of Glu 686 . Additionally, the sterane ring interacted with Asp 805 and Pro 807 , respectively ( Fig 4B). Since in docking studies we did not see significant differences in predicted binding properties of 10g and 11g to MDR1, we proceeded for PharmMapper analysis to find possible drug targets of 10g and 11g (Wang et al., 2017(Wang et al., , 2016. Unlike docking studies which rely on simple free energy based fit scores, PharmMapper encompasses computational tools that use ligand-pharmacophore fit scores. Thereby, predicts in silico druggable targets of a query compound from annotated pharmacophore model database through reversed pharmacophore matching thus improves prediction reliability. Among the top 10 potential proteins predicted to be druggable targets of 10g and 11g in humans, (Table 2) carbonic anhydrase 2, caspase 3, cholinesterase and peptidylprolyl cis-trans isomerase were identified as typical 10g targets, whereas caspase 7, PTP1B, KIF11 and aldo-keto reductase C3 were found as potential 11g targets. However, here it has to be emphasized that both molecular docking studies and PharmMapper analysis are in silico tools to predict with fair confidence the molecular targets of a given compound, but are not by themselves sufficient to either support the supposed molecular mechanism or accurately prove the exact reallife targets of the compounds.

Treatment with 10g does not influence either transcriptionally or translationally the expression of MDR1 gene
After we did not detect significant difference in the predicted binding properties of 10g and 11g to MDR1, although the MDR1-inhibiting property of compound 10g but not of 11g was identified, we aimed to delineate the possible molecular mechanism/s behind the 10g-induced cellular phenomena. Inhibition of MDR1 expression either transcriptionally and (or) translationally upon 10g treatment can be a possible reason for the modulation of MDR1 efflux activity. It was also previously suggested that proto-oncogenes CYCD and c-MYC are able to regulate MDR1 expression (He et al., 2000;Wang et al., 2012). Therefore, relative mRNA levels of MDR1, CYCD and c-MYC in MCF-7/KCR cells exposed to 20 µM of 10g for 24 h were examined by quantitative real-time PCR. Our results indicate that treatments with 10g did not change the transcript levels of either MDR1, CYCD or c-MYC suggesting that compound 10g does not influence the mRNA expression of MDR1 (Fig 5A). In a control experiment, the effect of compound 11g on the expression levels of these genes was also investigated and no changes were observed compared to untreated control cells (Fig S6). Then we analyzed MDR1 protein levels of 10g-exposed multidrug-resistant cells. Western blots revealed no difference in MDR1 protein expression between control and 10g-treated cells (Fig 5B). Our results imply that compound 10g is not triggering the transcriptional and translational activity of MDR1 gene, therefore the observed inhibition of MDR1-related efflux activity upon 10g treatment is not associated with the regulation of its expression.

Treatments with 11g but not with 10g induce mitochondrial damage
MDR1 efflux is an energy-dependent process, therefore, cellular energy status can have a direct impact on MDR1 activity. Hence, we hypothesized that if compound 10g damages mitochondria this would attenuate the ATP-driven MDR1-mediated membrane transport. To test this hypothesis, MCF-7/KCR cells were treated with 20 µM of 10g, and for comparison also with compound 11g for 24 h. As a positive control, cells were exposed to 500 µM carbachol for 24 h, since carbachol induces mitochondrial damage by raising cytoplasmic calcium levels. After the treatments, JC-1 staining was performed and the stained cells were observed under fluorescence microscope. The average number of mitochondria/cell was counted and expressed as a measure of with the nucleotide-binding domain 2 of MDR1. 10g is yellow while 11g is magenta coloured. Receptor amino acid side chains with interactions with the ligands are labeled and pink-coloured.

Table 2
PharmMapper generated top ten druggable targets of 10g and 11g.

Rank
10g targets 11g targets 1 Serine/threonine-protein kinase Pim-1 Integrin alpha-L 2 Carbonic anhydrase 2 Serine/threonine-protein kinase Pim-1 3 Integrin alpha-L Aldose reductase 4 Aldo-keto reductase C2 Aldo-keto reductase C2 5 Bone morphogenetic protein 2 Bone morphogenetic protein 2 6 Caspase-3 Caspase-7 7 Aldose reductase PTP1B 8 Cholinesterase Mitogen-activated protein kinase 1 9 Peptidyl-prolyl cis-trans isomerase Kinesin-like protein KIF11 10 Mitogen-activated protein kinase 1 Aldo-keto reductase C3 M.K. Gopisetty, et al. European Journal of Pharmaceutical Sciences 156 (2021) 105587 mitochondrial damage, where lower values indicate significant damage to this organelle. Results indicate that 10g treatment did not induce mitochondrial damage, whereas compound 11g triggered significant reduction in the number of functional mitochondria (Fig 6). If the underlying cause of the observed inhibition of MDR1 efflux activity was the damage to mitochondria, then compound 11g should have manifested a more efficient efflux inhibition compared to 10g. Since this was not the case, our results rule out the direct link between 10g-induced MDR1 inhibition and mitochondrial damage.

Compound 10g induces endoplasmic reticulum stress in multidrugresistant cancer cells
Cellular distribution of MDR1 can be disturbed due to endoplasmic reticulum (ER) stress (Gopisetty et al., 2019). Moreover, our in silico experiments revealed that one among the predicted druggable targets of compound 10g is peptidyl-prolyl cis-trans isomerase (Table 2), which protein is strongly associated with protein folding in the ER and its inhibition leads to ER stress (Kim et al., 2008). Therefore, we supposed that if compound 10g induces ER stress in multidrug-resistant MCF-7/  KCR cells then this could explain the attenuated MDR1 efflux activity upon treatments with 10g. To address this issue, MCF-7/KCR cells were treated with 20 µM of 10g, or for comparison with 11g for 24 h and transcriptional and translational activation of ER stress response genes BIP and CHOP were measured using quantitative real-time PCR and western blot respectively. Cells exposed to 4 mM of dithiothreitol (DTT) for 24 h served as positive control for the induction of ER stress. Our results revealed that 10g but not 11g induces significant increase in the expression levels of the ER stress response genes BIP and CHOP compared to untreated control (Fig 7A, B), which implies the involvement of this cellular stress condition in the complex molecular mechanism behind MDR1 inhibition.

Discussion
The success of cancer chemotherapy is largely limited by inherent or attained multidrug resistance of cancer cells. Often, upon exposure to therapeutic drugs, cancer cells undergo an evolution by rewiring their molecular mechanisms resulting in cancer cell phenotypes which are capable to evade the toxic effects of therapeutic drugs. Among the various mechanisms involved in this process, the upregulation of MDR1 is predominant since it ensures cancer cell resistance to many structurally and functionally dissimilar drugs. Fueled by ATP hydrolysis, MDR1 pumps drugs out of cancer cells keeping the intracellular amount of these agents below the required concentration to elicit cytotoxicity. Therefore, inhibition of MDR1 has always been regarded as a promising approach to improve cancer chemotherapy also in clinical settings. Despite the fact that numerous MDR1 inhibitors were tested and have failed in clinical trials due to questionable safety and thus were dismissed for further therapeutic applications, the quest for potent MDR1 modulators is still on (Chung et al., 2016).
In the present work we examined a series of pyrimidinones fused to the A-ring of DHT or DHT-acetate (Baji et al., 2017) and compared their MDR1 inhibiting properties in MCF-7/KCR cells that overexpress MDR1. Here we show a remarkable structure-related discriminative behavior manifested by these androstane-derivatives on MDR1 activity, since all the 17-acetylated derivatives 10b, 10c, 10d, 10f, 10g exhibited a marked MDR1 inhibitory potential in contrast to their 17-OH pairs 11b, 11c, 11d, 11f, 11g (Fig S3, Fig 2A). As compound 10g showed the highest capacity in hindering the MDR1 efflux activity (table S1) from all the tested derivatives 10g and its deacetylated pair 11g were selected for subsequent experiments to unravel the possible mechanisms behind the 10g-induced reduction in MDR1 function. Based on previous cytotoxicity tests 10g or 11g were applied in 20 µM concentration in further experiments as this concentration did not cause any cytotoxicity on the examined cell lines (Fig 2B). The fact that DHT, a parent compound with a similar structural platform, did not show MDR1 inhibitory activity ( Fig 2C) indicates a modification-dependent gain-of-function of the acetoxy group carrying by the A-ring fused arylpyrimidinone-androstanes.
The drug resistance displayed by MCF-7/KCR signifies that such cancer cells can evade the toxic effects of unusually high concentrations of doxorubicin (Gopisetty et al., 2019). Treatment with 10g improved cytotoxicity of doxorubicin ( Fig 3A) indicated by decreased IC 50 values for this drug (Fig S4) in cells co-treated with 10g, compared to cells treated with doxorubicin alone. Furthermore, we proved that 10g treatment sensitizes drug-resistant MCF-7/KCR cells to doxorubicininduced apoptosis (Fig 3B). Surprisingly, the doxorubicin-sensitizing effect of 10g exceeded that of the well-known, first generation MDR1 inhibitor and our positive control verapamil (Fig 3B). Molecular docking for 10g and 11g performed to predict their binding to MDR1 and other druggable targets, revealed no significant differences in their binding properties with MDR1 (Fig 4A, B). Nevertheless, other in silico approaches (such as PharmMapper) revealed different predicted binding targets for these molecules, among them peptidyl-prolyl cistrans isomerase as a potential target of compound 10g, which is a protein involved in physiological ER functions (table 2).
Down-regulation of MDR1 expression would undoubtedly impact the quantity of drugs expelled by MDR1 efflux; therefore, we analyzed the transcript and the protein levels of MDR1 as well as the relative mRNA amount of well-known MDR1 gene expression regulators c-MYC and CYCD following treatment of MCF-7/KCR cells with compound Fig. 7. 10g treatment induces endoplasmic reticulum stress in MCF-7/KCR cells. A. Treatment with compound 10g leads to elevated relative mRNA (A) and protein (B) levels of the important ER stress response genes BIP and CHOP, as representative blots indicate. Values represent mean ± SD calculated from three independent experiments. (****P < 0.0001, Fisher's LSD test).
10g. The results imply that these two latter genes were not transcriptionally elevated (Fig 5A) upon 10g treatment and obeying these data MDR1 was also not down-or upregulated either transcriptionally ( Fig 5A) or translationally (Fig 5B).
Drug efflux through MDR1 is an energy-dependent process, which consumes energy from ATP hydrolysis; therefore, mitochondria -the major generators of cellular ATP -are indirectly linked to ABC transporter activity. Mitochondrial damage can lead to a reduced MDR1 efflux due to insufficient cellular ATP. A direct correlation between MDR1 inhibitory activity and cellular ATP diminishing capacity was verified in a recent report on the novel compound RY10-4 (Xue et al., 2014). To test the possibility of mitochondrial dysfunction, MCF-7/KCR cells were treated with either compound 10g or 11g or as a positive control with carbachol, and JC-1 staining was performed to measure mitochondrial health. We observed that treatment with molecule 11g induced significant damage to mitochondria (Fig 6), whereas compound 10g did not induce similar mitochondrial impairment. These findings rule out the possibility of mitochondrial damage being the underlying factor of the observed MDR1 inhibition following treatment with 10g.
Endoplasmic reticulum is the major site of protein homeostasis. Perturbations in its function lead to ER stress and deregulation of the protein folding machinery. Since MDR1 is a glycoprotein, it requires proper handling by the ER to be capable to attain its functional conformation before reaching the plasma membrane. Endoplasmic reticulum stress perturbs the proper folding of this transport protein, which leads to a decreased number of functional MDR1 in the plasma membrane resulting in diminished efflux activity. Since a direct connection between endoplasmic reticulum stress and inhibition of MDR1 activity has already been demonstrated (Gopisetty et al., 2019), therefore we hypothesized that compound 10g might induce endoplasmic reticulum stress as well, ultimately precipitating to a decreased efflux function. To test this hypothesis, the relative changes in the transcript and protein levels of ER stress markers BIP and CHOP were assessed in MCF-7/KCR cells treated either with compound 10g or 11g or with the ER stress inducer compound DTT. Our results strongly indicate that 10g treatment induces significant endoplasmic reticulum stress, as both the mRNA and the protein levels of CHOP and BIP were increased upon 10g exposures (Fig 7). We did not observe significant mRNA expression of these ER stress markers in multidrug-resistant cancer cells treated with molecule 11g (Fig 7); nevertheless, the protein expression of CHOP was somewhat elevated following 11g treatments ( Fig 7B).
CHOP is activated during ER stress; however, ER stress is not the only source of CHOP activation. Cellular stresses other than ER stress through p38 MAPK also activate CHOP (Sano and Reed, 2013). Absence of transcriptional and translational elevation of BIP suggests that 11ginduced CHOP activation might not be the result of ER stress. Moreover, stress induced CHOP activation is mostly post-transcriptional (Bi et al., 2005), which explains the possibility for observed elevation in protein but not in mRNA levels of CHOP in our results (Fig 7A, B). These findings demonstrate the reasonable implication of this cellular stress condition in the complex molecular mechanism leading to MDR1 inhibition. However, further studies are required to ascertain this connection.
The present study proved that novel A-ring-fused arylpyrimidine androstane derivatives exhibit MDR1-inhibiting potential in multidrugresistant adenocarcinoma cells and we unraveled some of the possible molecular events underlying this feature. Furthermore, we also proved that the MDR1 inhibitory activity of these A-ring-fused arylpyrimidinones is strongly dependent on their acetylation status. With further studies, our present work can open a structural platform for the design and synthesis of new generation MDR1 inhibitors that can attenuate drug resistance and sensitize resistant cancer cells to clinically applied chemotherapy drugs. Apart from oncotherapeutic approaches, these molecules could also be utilized in the treatment of other pathological conditions like epilepsy (Summers et al., 2004).

Author contributions
All authors read and approved the final manuscript.

Declaration of Competing Interest
None.